diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -28090,3 +28090,2009 @@ Use FP16 precision: False
 01/29/2022 17:10:29 - INFO - codeparrot_training - Step 21998: {'lr': 0.00031473637056993837, 'samples': 4223808, 'steps': 21998, 'loss/train': 1.3815385401248932}
 01/29/2022 17:10:33 - INFO - codeparrot_training - Step 21999: {'lr': 0.0003147205660614055, 'samples': 4224000, 'steps': 21999, 'loss/train': 1.090144783258438}
 01/29/2022 17:10:33 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+01/29/2022 17:11:06 - WARNING - huggingface_hub.repository - Several commits (11) will be pushed upstream.
+01/29/2022 17:11:06 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+01/29/2022 17:12:10 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/neo-code-py
+   5707af2..7f31785  expert-dust-2 -> expert-dust-2
+
+01/29/2022 17:12:15 - INFO - codeparrot_training - Step 22000: {'lr': 0.00031470476127563017, 'samples': 4224192, 'steps': 22000, 'loss/train': 0.7294651418924332}
+01/29/2022 17:12:21 - INFO - codeparrot_training - Step 22001: {'lr': 0.00031468895621268036, 'samples': 4224384, 'steps': 22001, 'loss/train': 2.3257076740264893}
+01/29/2022 17:12:25 - INFO - codeparrot_training - Step 22002: {'lr': 0.00031467315087262344, 'samples': 4224576, 'steps': 22002, 'loss/train': 2.015672028064728}
+01/29/2022 17:12:29 - INFO - codeparrot_training - Step 22003: {'lr': 0.0003146573452555274, 'samples': 4224768, 'steps': 22003, 'loss/train': 1.4181785881519318}
+01/29/2022 17:12:34 - INFO - codeparrot_training - Step 22004: {'lr': 0.0003146415393614597, 'samples': 4224960, 'steps': 22004, 'loss/train': 3.6781225204467773}
+01/29/2022 17:12:38 - INFO - codeparrot_training - Step 22005: {'lr': 0.00031462573319048827, 'samples': 4225152, 'steps': 22005, 'loss/train': 1.7055628895759583}
+01/29/2022 17:12:43 - INFO - codeparrot_training - Step 22006: {'lr': 0.0003146099267426806, 'samples': 4225344, 'steps': 22006, 'loss/train': 1.408981740474701}
+01/29/2022 17:12:47 - INFO - codeparrot_training - Step 22007: {'lr': 0.00031459412001810474, 'samples': 4225536, 'steps': 22007, 'loss/train': 1.6812592148780823}
+01/29/2022 17:12:52 - INFO - codeparrot_training - Step 22008: {'lr': 0.000314578313016828, 'samples': 4225728, 'steps': 22008, 'loss/train': 1.4872789978981018}
+01/29/2022 17:12:56 - INFO - codeparrot_training - Step 22009: {'lr': 0.00031456250573891825, 'samples': 4225920, 'steps': 22009, 'loss/train': 1.6449995040893555}
+01/29/2022 17:13:00 - INFO - codeparrot_training - Step 22010: {'lr': 0.0003145466981844434, 'samples': 4226112, 'steps': 22010, 'loss/train': 1.891547977924347}
+01/29/2022 17:13:07 - INFO - codeparrot_training - Step 22011: {'lr': 0.00031453089035347084, 'samples': 4226304, 'steps': 22011, 'loss/train': 2.0247682929039}
+01/29/2022 17:13:12 - INFO - codeparrot_training - Step 22012: {'lr': 0.0003145150822460685, 'samples': 4226496, 'steps': 22012, 'loss/train': 1.6697564721107483}
+01/29/2022 17:13:16 - INFO - codeparrot_training - Step 22013: {'lr': 0.00031449927386230397, 'samples': 4226688, 'steps': 22013, 'loss/train': 1.3253786265850067}
+01/29/2022 17:13:20 - INFO - codeparrot_training - Step 22014: {'lr': 0.0003144834652022451, 'samples': 4226880, 'steps': 22014, 'loss/train': 1.430372267961502}
+01/29/2022 17:13:24 - INFO - codeparrot_training - Step 22015: {'lr': 0.0003144676562659595, 'samples': 4227072, 'steps': 22015, 'loss/train': 2.2930973768234253}
+01/29/2022 17:13:30 - INFO - codeparrot_training - Step 22016: {'lr': 0.000314451847053515, 'samples': 4227264, 'steps': 22016, 'loss/train': 1.6549941301345825}
+01/29/2022 17:13:34 - INFO - codeparrot_training - Step 22017: {'lr': 0.0003144360375649792, 'samples': 4227456, 'steps': 22017, 'loss/train': 0.5592041909694672}
+01/29/2022 17:13:38 - INFO - codeparrot_training - Step 22018: {'lr': 0.0003144202278004199, 'samples': 4227648, 'steps': 22018, 'loss/train': 1.3562363684177399}
+01/29/2022 17:13:42 - INFO - codeparrot_training - Step 22019: {'lr': 0.0003144044177599047, 'samples': 4227840, 'steps': 22019, 'loss/train': 0.9630638659000397}
+01/29/2022 17:13:47 - INFO - codeparrot_training - Step 22020: {'lr': 0.00031438860744350156, 'samples': 4228032, 'steps': 22020, 'loss/train': 2.045553982257843}
+01/29/2022 17:13:52 - INFO - codeparrot_training - Step 22021: {'lr': 0.000314372796851278, 'samples': 4228224, 'steps': 22021, 'loss/train': 1.2689038217067719}
+01/29/2022 17:13:56 - INFO - codeparrot_training - Step 22022: {'lr': 0.00031435698598330185, 'samples': 4228416, 'steps': 22022, 'loss/train': 2.3658615946769714}
+01/29/2022 17:14:01 - INFO - codeparrot_training - Step 22023: {'lr': 0.0003143411748396408, 'samples': 4228608, 'steps': 22023, 'loss/train': 1.384789913892746}
+01/29/2022 17:14:05 - INFO - codeparrot_training - Step 22024: {'lr': 0.00031432536342036255, 'samples': 4228800, 'steps': 22024, 'loss/train': 1.129883050918579}
+01/29/2022 17:14:09 - INFO - codeparrot_training - Step 22025: {'lr': 0.00031430955172553497, 'samples': 4228992, 'steps': 22025, 'loss/train': 1.4963155388832092}
+01/29/2022 17:14:14 - INFO - codeparrot_training - Step 22026: {'lr': 0.00031429373975522555, 'samples': 4229184, 'steps': 22026, 'loss/train': 1.7065992951393127}
+01/29/2022 17:14:18 - INFO - codeparrot_training - Step 22027: {'lr': 0.00031427792750950227, 'samples': 4229376, 'steps': 22027, 'loss/train': 1.350190132856369}
+01/29/2022 17:14:23 - INFO - codeparrot_training - Step 22028: {'lr': 0.0003142621149884327, 'samples': 4229568, 'steps': 22028, 'loss/train': 1.9066177010536194}
+01/29/2022 17:14:27 - INFO - codeparrot_training - Step 22029: {'lr': 0.00031424630219208474, 'samples': 4229760, 'steps': 22029, 'loss/train': 1.6766796112060547}
+01/29/2022 17:14:31 - INFO - codeparrot_training - Step 22030: {'lr': 0.00031423048912052585, 'samples': 4229952, 'steps': 22030, 'loss/train': 1.5325748920440674}
+01/29/2022 17:14:38 - INFO - codeparrot_training - Step 22031: {'lr': 0.0003142146757738241, 'samples': 4230144, 'steps': 22031, 'loss/train': 1.5229666829109192}
+01/29/2022 17:14:42 - INFO - codeparrot_training - Step 22032: {'lr': 0.000314198862152047, 'samples': 4230336, 'steps': 22032, 'loss/train': 1.5144955515861511}
+01/29/2022 17:14:47 - INFO - codeparrot_training - Step 22033: {'lr': 0.00031418304825526236, 'samples': 4230528, 'steps': 22033, 'loss/train': 1.776663064956665}
+01/29/2022 17:14:51 - INFO - codeparrot_training - Step 22034: {'lr': 0.0003141672340835379, 'samples': 4230720, 'steps': 22034, 'loss/train': 1.6651784777641296}
+01/29/2022 17:14:55 - INFO - codeparrot_training - Step 22035: {'lr': 0.0003141514196369414, 'samples': 4230912, 'steps': 22035, 'loss/train': 1.576603353023529}
+01/29/2022 17:15:00 - INFO - codeparrot_training - Step 22036: {'lr': 0.0003141356049155406, 'samples': 4231104, 'steps': 22036, 'loss/train': 1.7657893896102905}
+01/29/2022 17:15:05 - INFO - codeparrot_training - Step 22037: {'lr': 0.00031411978991940324, 'samples': 4231296, 'steps': 22037, 'loss/train': 1.5722812414169312}
+01/29/2022 17:15:09 - INFO - codeparrot_training - Step 22038: {'lr': 0.000314103974648597, 'samples': 4231488, 'steps': 22038, 'loss/train': 1.6582667827606201}
+01/29/2022 17:15:13 - INFO - codeparrot_training - Step 22039: {'lr': 0.0003140881591031898, 'samples': 4231680, 'steps': 22039, 'loss/train': 1.839832842350006}
+01/29/2022 17:15:17 - INFO - codeparrot_training - Step 22040: {'lr': 0.0003140723432832492, 'samples': 4231872, 'steps': 22040, 'loss/train': 1.0712084770202637}
+01/29/2022 17:15:25 - INFO - codeparrot_training - Step 22041: {'lr': 0.00031405652718884304, 'samples': 4232064, 'steps': 22041, 'loss/train': 1.984972894191742}
+01/29/2022 17:15:29 - INFO - codeparrot_training - Step 22042: {'lr': 0.00031404071082003903, 'samples': 4232256, 'steps': 22042, 'loss/train': 2.6697086691856384}
+01/29/2022 17:15:33 - INFO - codeparrot_training - Step 22043: {'lr': 0.000314024894176905, 'samples': 4232448, 'steps': 22043, 'loss/train': 0.5624170750379562}
+01/29/2022 17:15:38 - INFO - codeparrot_training - Step 22044: {'lr': 0.00031400907725950865, 'samples': 4232640, 'steps': 22044, 'loss/train': 1.0981973111629486}
+01/29/2022 17:15:43 - INFO - codeparrot_training - Step 22045: {'lr': 0.00031399326006791765, 'samples': 4232832, 'steps': 22045, 'loss/train': 2.2021772861480713}
+01/29/2022 17:15:47 - INFO - codeparrot_training - Step 22046: {'lr': 0.00031397744260219996, 'samples': 4233024, 'steps': 22046, 'loss/train': 0.7804614007472992}
+01/29/2022 17:15:52 - INFO - codeparrot_training - Step 22047: {'lr': 0.00031396162486242317, 'samples': 4233216, 'steps': 22047, 'loss/train': 2.0868544578552246}
+01/29/2022 17:15:56 - INFO - codeparrot_training - Step 22048: {'lr': 0.0003139458068486551, 'samples': 4233408, 'steps': 22048, 'loss/train': 1.5211027264595032}
+01/29/2022 17:16:00 - INFO - codeparrot_training - Step 22049: {'lr': 0.0003139299885609635, 'samples': 4233600, 'steps': 22049, 'loss/train': 1.6194189190864563}
+01/29/2022 17:16:04 - INFO - codeparrot_training - Step 22050: {'lr': 0.00031391416999941606, 'samples': 4233792, 'steps': 22050, 'loss/train': 2.1286940574645996}
+01/29/2022 17:16:10 - INFO - codeparrot_training - Step 22051: {'lr': 0.00031389835116408073, 'samples': 4233984, 'steps': 22051, 'loss/train': 1.686644732952118}
+01/29/2022 17:16:14 - INFO - codeparrot_training - Step 22052: {'lr': 0.000313882532055025, 'samples': 4234176, 'steps': 22052, 'loss/train': 1.4268637597560883}
+01/29/2022 17:16:18 - INFO - codeparrot_training - Step 22053: {'lr': 0.000313866712672317, 'samples': 4234368, 'steps': 22053, 'loss/train': 1.9818872809410095}
+01/29/2022 17:16:23 - INFO - codeparrot_training - Step 22054: {'lr': 0.0003138508930160241, 'samples': 4234560, 'steps': 22054, 'loss/train': 1.6147372126579285}
+01/29/2022 17:16:27 - INFO - codeparrot_training - Step 22055: {'lr': 0.0003138350730862144, 'samples': 4234752, 'steps': 22055, 'loss/train': 0.8677935898303986}
+01/29/2022 17:16:34 - INFO - codeparrot_training - Step 22056: {'lr': 0.00031381925288295536, 'samples': 4234944, 'steps': 22056, 'loss/train': 1.8375222086906433}
+01/29/2022 17:16:38 - INFO - codeparrot_training - Step 22057: {'lr': 0.000313803432406315, 'samples': 4235136, 'steps': 22057, 'loss/train': 1.5995088815689087}
+01/29/2022 17:16:43 - INFO - codeparrot_training - Step 22058: {'lr': 0.000313787611656361, 'samples': 4235328, 'steps': 22058, 'loss/train': 1.803549349308014}
+01/29/2022 17:16:47 - INFO - codeparrot_training - Step 22059: {'lr': 0.00031377179063316106, 'samples': 4235520, 'steps': 22059, 'loss/train': 1.7444196939468384}
+01/29/2022 17:16:51 - INFO - codeparrot_training - Step 22060: {'lr': 0.0003137559693367831, 'samples': 4235712, 'steps': 22060, 'loss/train': 0.8833642601966858}
+01/29/2022 17:16:57 - INFO - codeparrot_training - Step 22061: {'lr': 0.0003137401477672947, 'samples': 4235904, 'steps': 22061, 'loss/train': 1.0041523575782776}
+01/29/2022 17:17:01 - INFO - codeparrot_training - Step 22062: {'lr': 0.00031372432592476383, 'samples': 4236096, 'steps': 22062, 'loss/train': 1.14071524143219}
+01/29/2022 17:17:05 - INFO - codeparrot_training - Step 22063: {'lr': 0.0003137085038092582, 'samples': 4236288, 'steps': 22063, 'loss/train': 1.8395987749099731}
+01/29/2022 17:17:10 - INFO - codeparrot_training - Step 22064: {'lr': 0.00031369268142084555, 'samples': 4236480, 'steps': 22064, 'loss/train': 1.5687727332115173}
+01/29/2022 17:17:14 - INFO - codeparrot_training - Step 22065: {'lr': 0.00031367685875959375, 'samples': 4236672, 'steps': 22065, 'loss/train': 1.7396534085273743}
+01/29/2022 17:17:19 - INFO - codeparrot_training - Step 22066: {'lr': 0.0003136610358255704, 'samples': 4236864, 'steps': 22066, 'loss/train': 1.9377068281173706}
+01/29/2022 17:17:24 - INFO - codeparrot_training - Step 22067: {'lr': 0.00031364521261884336, 'samples': 4237056, 'steps': 22067, 'loss/train': 1.7466498613357544}
+01/29/2022 17:17:28 - INFO - codeparrot_training - Step 22068: {'lr': 0.00031362938913948046, 'samples': 4237248, 'steps': 22068, 'loss/train': 2.048322916030884}
+01/29/2022 17:17:32 - INFO - codeparrot_training - Step 22069: {'lr': 0.0003136135653875495, 'samples': 4237440, 'steps': 22069, 'loss/train': 1.6285280585289001}
+01/29/2022 17:17:36 - INFO - codeparrot_training - Step 22070: {'lr': 0.00031359774136311823, 'samples': 4237632, 'steps': 22070, 'loss/train': 1.958620011806488}
+01/29/2022 17:17:43 - INFO - codeparrot_training - Step 22071: {'lr': 0.0003135819170662545, 'samples': 4237824, 'steps': 22071, 'loss/train': 2.1949211955070496}
+01/29/2022 17:17:48 - INFO - codeparrot_training - Step 22072: {'lr': 0.00031356609249702587, 'samples': 4238016, 'steps': 22072, 'loss/train': 1.0873979330062866}
+01/29/2022 17:17:52 - INFO - codeparrot_training - Step 22073: {'lr': 0.0003135502676555004, 'samples': 4238208, 'steps': 22073, 'loss/train': 1.5199854969978333}
+01/29/2022 17:17:56 - INFO - codeparrot_training - Step 22074: {'lr': 0.0003135344425417457, 'samples': 4238400, 'steps': 22074, 'loss/train': 1.9332205653190613}
+01/29/2022 17:18:00 - INFO - codeparrot_training - Step 22075: {'lr': 0.0003135186171558297, 'samples': 4238592, 'steps': 22075, 'loss/train': 2.001096189022064}
+01/29/2022 17:18:06 - INFO - codeparrot_training - Step 22076: {'lr': 0.00031350279149782004, 'samples': 4238784, 'steps': 22076, 'loss/train': 1.598426878452301}
+01/29/2022 17:18:10 - INFO - codeparrot_training - Step 22077: {'lr': 0.0003134869655677846, 'samples': 4238976, 'steps': 22077, 'loss/train': 2.054503619670868}
+01/29/2022 17:18:14 - INFO - codeparrot_training - Step 22078: {'lr': 0.00031347113936579116, 'samples': 4239168, 'steps': 22078, 'loss/train': 1.9067557454109192}
+01/29/2022 17:18:18 - INFO - codeparrot_training - Step 22079: {'lr': 0.00031345531289190756, 'samples': 4239360, 'steps': 22079, 'loss/train': 1.8355541825294495}
+01/29/2022 17:18:23 - INFO - codeparrot_training - Step 22080: {'lr': 0.0003134394861462014, 'samples': 4239552, 'steps': 22080, 'loss/train': 1.8083798289299011}
+01/29/2022 17:18:28 - INFO - codeparrot_training - Step 22081: {'lr': 0.0003134236591287407, 'samples': 4239744, 'steps': 22081, 'loss/train': 1.2462304830551147}
+01/29/2022 17:18:32 - INFO - codeparrot_training - Step 22082: {'lr': 0.0003134078318395933, 'samples': 4239936, 'steps': 22082, 'loss/train': 2.2743841409683228}
+01/29/2022 17:18:36 - INFO - codeparrot_training - Step 22083: {'lr': 0.00031339200427882676, 'samples': 4240128, 'steps': 22083, 'loss/train': 1.7213464379310608}
+01/29/2022 17:18:41 - INFO - codeparrot_training - Step 22084: {'lr': 0.00031337617644650907, 'samples': 4240320, 'steps': 22084, 'loss/train': 2.180558681488037}
+01/29/2022 17:18:45 - INFO - codeparrot_training - Step 22085: {'lr': 0.00031336034834270786, 'samples': 4240512, 'steps': 22085, 'loss/train': 1.6368547081947327}
+01/29/2022 17:18:52 - INFO - codeparrot_training - Step 22086: {'lr': 0.00031334451996749117, 'samples': 4240704, 'steps': 22086, 'loss/train': 2.209353983402252}
+01/29/2022 17:18:56 - INFO - codeparrot_training - Step 22087: {'lr': 0.00031332869132092654, 'samples': 4240896, 'steps': 22087, 'loss/train': 1.6674118638038635}
+01/29/2022 17:19:00 - INFO - codeparrot_training - Step 22088: {'lr': 0.00031331286240308205, 'samples': 4241088, 'steps': 22088, 'loss/train': 2.7394291162490845}
+01/29/2022 17:19:05 - INFO - codeparrot_training - Step 22089: {'lr': 0.00031329703321402526, 'samples': 4241280, 'steps': 22089, 'loss/train': 2.1045597195625305}
+01/29/2022 17:19:09 - INFO - codeparrot_training - Step 22090: {'lr': 0.00031328120375382414, 'samples': 4241472, 'steps': 22090, 'loss/train': 2.014697313308716}
+01/29/2022 17:19:14 - INFO - codeparrot_training - Step 22091: {'lr': 0.0003132653740225464, 'samples': 4241664, 'steps': 22091, 'loss/train': 1.3775839805603027}
+01/29/2022 17:19:19 - INFO - codeparrot_training - Step 22092: {'lr': 0.0003132495440202599, 'samples': 4241856, 'steps': 22092, 'loss/train': 1.9096525311470032}
+01/29/2022 17:19:23 - INFO - codeparrot_training - Step 22093: {'lr': 0.0003132337137470324, 'samples': 4242048, 'steps': 22093, 'loss/train': 1.6905410885810852}
+01/29/2022 17:19:27 - INFO - codeparrot_training - Step 22094: {'lr': 0.00031321788320293176, 'samples': 4242240, 'steps': 22094, 'loss/train': 1.1693287789821625}
+01/29/2022 17:19:32 - INFO - codeparrot_training - Step 22095: {'lr': 0.00031320205238802583, 'samples': 4242432, 'steps': 22095, 'loss/train': 1.082742154598236}
+01/29/2022 17:19:37 - INFO - codeparrot_training - Step 22096: {'lr': 0.0003131862213023823, 'samples': 4242624, 'steps': 22096, 'loss/train': 2.118521511554718}
+01/29/2022 17:19:41 - INFO - codeparrot_training - Step 22097: {'lr': 0.0003131703899460692, 'samples': 4242816, 'steps': 22097, 'loss/train': 2.0646414756774902}
+01/29/2022 17:19:46 - INFO - codeparrot_training - Step 22098: {'lr': 0.0003131545583191541, 'samples': 4243008, 'steps': 22098, 'loss/train': 6.935418605804443}
+01/29/2022 17:19:50 - INFO - codeparrot_training - Step 22099: {'lr': 0.00031313872642170493, 'samples': 4243200, 'steps': 22099, 'loss/train': 2.029514729976654}
+01/29/2022 17:19:54 - INFO - codeparrot_training - Step 22100: {'lr': 0.0003131228942537895, 'samples': 4243392, 'steps': 22100, 'loss/train': 0.9553154110908508}
+01/29/2022 17:19:59 - INFO - codeparrot_training - Step 22101: {'lr': 0.00031310706181547567, 'samples': 4243584, 'steps': 22101, 'loss/train': 1.1325762569904327}
+01/29/2022 17:20:06 - INFO - codeparrot_training - Step 22102: {'lr': 0.0003130912291068312, 'samples': 4243776, 'steps': 22102, 'loss/train': 1.2766943871974945}
+01/29/2022 17:20:10 - INFO - codeparrot_training - Step 22103: {'lr': 0.000313075396127924, 'samples': 4243968, 'steps': 22103, 'loss/train': 2.0860885977745056}
+01/29/2022 17:20:14 - INFO - codeparrot_training - Step 22104: {'lr': 0.0003130595628788217, 'samples': 4244160, 'steps': 22104, 'loss/train': 1.6747584342956543}
+01/29/2022 17:20:19 - INFO - codeparrot_training - Step 22105: {'lr': 0.00031304372935959235, 'samples': 4244352, 'steps': 22105, 'loss/train': 1.2529475390911102}
+01/29/2022 17:20:23 - INFO - codeparrot_training - Step 22106: {'lr': 0.00031302789557030364, 'samples': 4244544, 'steps': 22106, 'loss/train': 0.4839460104703903}
+01/29/2022 17:20:28 - INFO - codeparrot_training - Step 22107: {'lr': 0.00031301206151102353, 'samples': 4244736, 'steps': 22107, 'loss/train': 1.918887734413147}
+01/29/2022 17:20:32 - INFO - codeparrot_training - Step 22108: {'lr': 0.00031299622718181964, 'samples': 4244928, 'steps': 22108, 'loss/train': 2.3766363859176636}
+01/29/2022 17:20:37 - INFO - codeparrot_training - Step 22109: {'lr': 0.00031298039258276, 'samples': 4245120, 'steps': 22109, 'loss/train': 1.6896631121635437}
+01/29/2022 17:20:41 - INFO - codeparrot_training - Step 22110: {'lr': 0.0003129645577139123, 'samples': 4245312, 'steps': 22110, 'loss/train': 2.2402930855751038}
+01/29/2022 17:20:45 - INFO - codeparrot_training - Step 22111: {'lr': 0.00031294872257534446, 'samples': 4245504, 'steps': 22111, 'loss/train': 0.9991984069347382}
+01/29/2022 17:20:50 - INFO - codeparrot_training - Step 22112: {'lr': 0.0003129328871671243, 'samples': 4245696, 'steps': 22112, 'loss/train': 1.2491111755371094}
+01/29/2022 17:20:55 - INFO - codeparrot_training - Step 22113: {'lr': 0.0003129170514893196, 'samples': 4245888, 'steps': 22113, 'loss/train': 1.5176453590393066}
+01/29/2022 17:20:59 - INFO - codeparrot_training - Step 22114: {'lr': 0.0003129012155419983, 'samples': 4246080, 'steps': 22114, 'loss/train': 1.8644778728485107}
+01/29/2022 17:21:03 - INFO - codeparrot_training - Step 22115: {'lr': 0.00031288537932522807, 'samples': 4246272, 'steps': 22115, 'loss/train': 1.7288707494735718}
+01/29/2022 17:21:10 - INFO - codeparrot_training - Step 22116: {'lr': 0.0003128695428390769, 'samples': 4246464, 'steps': 22116, 'loss/train': 1.2865969240665436}
+01/29/2022 17:21:15 - INFO - codeparrot_training - Step 22117: {'lr': 0.0003128537060836125, 'samples': 4246656, 'steps': 22117, 'loss/train': 1.9253059029579163}
+01/29/2022 17:21:19 - INFO - codeparrot_training - Step 22118: {'lr': 0.0003128378690589028, 'samples': 4246848, 'steps': 22118, 'loss/train': 1.8014668822288513}
+01/29/2022 17:21:23 - INFO - codeparrot_training - Step 22119: {'lr': 0.00031282203176501573, 'samples': 4247040, 'steps': 22119, 'loss/train': 1.5278865694999695}
+01/29/2022 17:21:27 - INFO - codeparrot_training - Step 22120: {'lr': 0.0003128061942020189, 'samples': 4247232, 'steps': 22120, 'loss/train': 1.342042475938797}
+01/29/2022 17:21:33 - INFO - codeparrot_training - Step 22121: {'lr': 0.00031279035636998037, 'samples': 4247424, 'steps': 22121, 'loss/train': 1.453108012676239}
+01/29/2022 17:21:37 - INFO - codeparrot_training - Step 22122: {'lr': 0.0003127745182689678, 'samples': 4247616, 'steps': 22122, 'loss/train': 1.8244303464889526}
+01/29/2022 17:21:41 - INFO - codeparrot_training - Step 22123: {'lr': 0.00031275867989904923, 'samples': 4247808, 'steps': 22123, 'loss/train': 1.6131058931350708}
+01/29/2022 17:21:45 - INFO - codeparrot_training - Step 22124: {'lr': 0.0003127428412602923, 'samples': 4248000, 'steps': 22124, 'loss/train': 1.2411067485809326}
+01/29/2022 17:21:50 - INFO - codeparrot_training - Step 22125: {'lr': 0.00031272700235276507, 'samples': 4248192, 'steps': 22125, 'loss/train': 2.2602809071540833}
+01/29/2022 17:21:57 - INFO - codeparrot_training - Step 22126: {'lr': 0.00031271116317653513, 'samples': 4248384, 'steps': 22126, 'loss/train': 1.3730115294456482}
+01/29/2022 17:22:01 - INFO - codeparrot_training - Step 22127: {'lr': 0.00031269532373167063, 'samples': 4248576, 'steps': 22127, 'loss/train': 1.6034016609191895}
+01/29/2022 17:22:05 - INFO - codeparrot_training - Step 22128: {'lr': 0.0003126794840182392, 'samples': 4248768, 'steps': 22128, 'loss/train': 2.8368515968322754}
+01/29/2022 17:22:09 - INFO - codeparrot_training - Step 22129: {'lr': 0.00031266364403630874, 'samples': 4248960, 'steps': 22129, 'loss/train': 1.6768561005592346}
+01/29/2022 17:22:14 - INFO - codeparrot_training - Step 22130: {'lr': 0.0003126478037859472, 'samples': 4249152, 'steps': 22130, 'loss/train': 1.7565520405769348}
+01/29/2022 17:22:19 - INFO - codeparrot_training - Step 22131: {'lr': 0.0003126319632672223, 'samples': 4249344, 'steps': 22131, 'loss/train': 1.9176501631736755}
+01/29/2022 17:22:23 - INFO - codeparrot_training - Step 22132: {'lr': 0.00031261612248020197, 'samples': 4249536, 'steps': 22132, 'loss/train': 1.6004834175109863}
+01/29/2022 17:22:28 - INFO - codeparrot_training - Step 22133: {'lr': 0.00031260028142495404, 'samples': 4249728, 'steps': 22133, 'loss/train': 1.6227040886878967}
+01/29/2022 17:22:32 - INFO - codeparrot_training - Step 22134: {'lr': 0.0003125844401015465, 'samples': 4249920, 'steps': 22134, 'loss/train': 1.4090770483016968}
+01/29/2022 17:22:36 - INFO - codeparrot_training - Step 22135: {'lr': 0.0003125685985100469, 'samples': 4250112, 'steps': 22135, 'loss/train': 2.5610931515693665}
+01/29/2022 17:22:42 - INFO - codeparrot_training - Step 22136: {'lr': 0.0003125527566505234, 'samples': 4250304, 'steps': 22136, 'loss/train': 3.517988920211792}
+01/29/2022 17:22:46 - INFO - codeparrot_training - Step 22137: {'lr': 0.0003125369145230438, 'samples': 4250496, 'steps': 22137, 'loss/train': 2.118595004081726}
+01/29/2022 17:22:50 - INFO - codeparrot_training - Step 22138: {'lr': 0.0003125210721276758, 'samples': 4250688, 'steps': 22138, 'loss/train': 2.0677923560142517}
+01/29/2022 17:22:54 - INFO - codeparrot_training - Step 22139: {'lr': 0.0003125052294644874, 'samples': 4250880, 'steps': 22139, 'loss/train': 1.4147033393383026}
+01/29/2022 17:22:59 - INFO - codeparrot_training - Step 22140: {'lr': 0.0003124893865335466, 'samples': 4251072, 'steps': 22140, 'loss/train': 2.0103389024734497}
+01/29/2022 17:23:06 - INFO - codeparrot_training - Step 22141: {'lr': 0.00031247354333492096, 'samples': 4251264, 'steps': 22141, 'loss/train': 1.762890636920929}
+01/29/2022 17:23:10 - INFO - codeparrot_training - Step 22142: {'lr': 0.00031245769986867845, 'samples': 4251456, 'steps': 22142, 'loss/train': 1.938239336013794}
+01/29/2022 17:23:14 - INFO - codeparrot_training - Step 22143: {'lr': 0.0003124418561348871, 'samples': 4251648, 'steps': 22143, 'loss/train': 1.3896063566207886}
+01/29/2022 17:23:18 - INFO - codeparrot_training - Step 22144: {'lr': 0.0003124260121336146, 'samples': 4251840, 'steps': 22144, 'loss/train': 1.7648577690124512}
+01/29/2022 17:23:23 - INFO - codeparrot_training - Step 22145: {'lr': 0.000312410167864929, 'samples': 4252032, 'steps': 22145, 'loss/train': 0.7795414924621582}
+01/29/2022 17:23:28 - INFO - codeparrot_training - Step 22146: {'lr': 0.00031239432332889796, 'samples': 4252224, 'steps': 22146, 'loss/train': 1.6305744051933289}
+01/29/2022 17:23:32 - INFO - codeparrot_training - Step 22147: {'lr': 0.00031237847852558947, 'samples': 4252416, 'steps': 22147, 'loss/train': 1.5376426577568054}
+01/29/2022 17:23:36 - INFO - codeparrot_training - Step 22148: {'lr': 0.00031236263345507133, 'samples': 4252608, 'steps': 22148, 'loss/train': 2.0374646186828613}
+01/29/2022 17:23:41 - INFO - codeparrot_training - Step 22149: {'lr': 0.0003123467881174116, 'samples': 4252800, 'steps': 22149, 'loss/train': 1.948259711265564}
+01/29/2022 17:23:45 - INFO - codeparrot_training - Step 22150: {'lr': 0.0003123309425126779, 'samples': 4252992, 'steps': 22150, 'loss/train': 1.897462785243988}
+01/29/2022 17:23:50 - INFO - codeparrot_training - Step 22151: {'lr': 0.00031231509664093833, 'samples': 4253184, 'steps': 22151, 'loss/train': 1.369369626045227}
+01/29/2022 17:23:55 - INFO - codeparrot_training - Step 22152: {'lr': 0.00031229925050226063, 'samples': 4253376, 'steps': 22152, 'loss/train': 1.6196731925010681}
+01/29/2022 17:23:59 - INFO - codeparrot_training - Step 22153: {'lr': 0.0003122834040967127, 'samples': 4253568, 'steps': 22153, 'loss/train': 1.067340463399887}
+01/29/2022 17:24:03 - INFO - codeparrot_training - Step 22154: {'lr': 0.00031226755742436255, 'samples': 4253760, 'steps': 22154, 'loss/train': 1.2313039898872375}
+01/29/2022 17:24:07 - INFO - codeparrot_training - Step 22155: {'lr': 0.0003122517104852778, 'samples': 4253952, 'steps': 22155, 'loss/train': 2.15651535987854}
+01/29/2022 17:24:13 - INFO - codeparrot_training - Step 22156: {'lr': 0.0003122358632795266, 'samples': 4254144, 'steps': 22156, 'loss/train': 1.2799761593341827}
+01/29/2022 17:24:17 - INFO - codeparrot_training - Step 22157: {'lr': 0.00031222001580717663, 'samples': 4254336, 'steps': 22157, 'loss/train': 0.9487791359424591}
+01/29/2022 17:24:22 - INFO - codeparrot_training - Step 22158: {'lr': 0.000312204168068296, 'samples': 4254528, 'steps': 22158, 'loss/train': 2.2366417050361633}
+01/29/2022 17:24:26 - INFO - codeparrot_training - Step 22159: {'lr': 0.00031218832006295235, 'samples': 4254720, 'steps': 22159, 'loss/train': 0.8271512389183044}
+01/29/2022 17:24:30 - INFO - codeparrot_training - Step 22160: {'lr': 0.0003121724717912138, 'samples': 4254912, 'steps': 22160, 'loss/train': 1.8140308856964111}
+01/29/2022 17:24:34 - INFO - codeparrot_training - Step 22161: {'lr': 0.000312156623253148, 'samples': 4255104, 'steps': 22161, 'loss/train': 1.1995005011558533}
+01/29/2022 17:24:41 - INFO - codeparrot_training - Step 22162: {'lr': 0.00031214077444882297, 'samples': 4255296, 'steps': 22162, 'loss/train': 2.1305333375930786}
+01/29/2022 17:24:46 - INFO - codeparrot_training - Step 22163: {'lr': 0.0003121249253783067, 'samples': 4255488, 'steps': 22163, 'loss/train': 1.5764726400375366}
+01/29/2022 17:24:50 - INFO - codeparrot_training - Step 22164: {'lr': 0.00031210907604166686, 'samples': 4255680, 'steps': 22164, 'loss/train': 2.166677176952362}
+01/29/2022 17:24:54 - INFO - codeparrot_training - Step 22165: {'lr': 0.0003120932264389715, 'samples': 4255872, 'steps': 22165, 'loss/train': 2.3122923374176025}
+01/29/2022 17:24:58 - INFO - codeparrot_training - Step 22166: {'lr': 0.0003120773765702885, 'samples': 4256064, 'steps': 22166, 'loss/train': 2.731784999370575}
+01/29/2022 17:25:04 - INFO - codeparrot_training - Step 22167: {'lr': 0.00031206152643568577, 'samples': 4256256, 'steps': 22167, 'loss/train': 1.627392053604126}
+01/29/2022 17:25:08 - INFO - codeparrot_training - Step 22168: {'lr': 0.00031204567603523105, 'samples': 4256448, 'steps': 22168, 'loss/train': 1.4719775319099426}
+01/29/2022 17:25:12 - INFO - codeparrot_training - Step 22169: {'lr': 0.00031202982536899246, 'samples': 4256640, 'steps': 22169, 'loss/train': 3.0744305849075317}
+01/29/2022 17:25:17 - INFO - codeparrot_training - Step 22170: {'lr': 0.0003120139744370377, 'samples': 4256832, 'steps': 22170, 'loss/train': 1.5315842628479004}
+01/29/2022 17:25:21 - INFO - codeparrot_training - Step 22171: {'lr': 0.0003119981232394349, 'samples': 4257024, 'steps': 22171, 'loss/train': 2.0551568269729614}
+01/29/2022 17:25:28 - INFO - codeparrot_training - Step 22172: {'lr': 0.0003119822717762517, 'samples': 4257216, 'steps': 22172, 'loss/train': 1.7782668471336365}
+01/29/2022 17:25:32 - INFO - codeparrot_training - Step 22173: {'lr': 0.0003119664200475562, 'samples': 4257408, 'steps': 22173, 'loss/train': 0.7942833602428436}
+01/29/2022 17:25:37 - INFO - codeparrot_training - Step 22174: {'lr': 0.0003119505680534162, 'samples': 4257600, 'steps': 22174, 'loss/train': 2.2435750365257263}
+01/29/2022 17:25:41 - INFO - codeparrot_training - Step 22175: {'lr': 0.00031193471579389967, 'samples': 4257792, 'steps': 22175, 'loss/train': 1.3651418387889862}
+01/29/2022 17:25:45 - INFO - codeparrot_training - Step 22176: {'lr': 0.00031191886326907445, 'samples': 4257984, 'steps': 22176, 'loss/train': 1.7808401584625244}
+01/29/2022 17:25:51 - INFO - codeparrot_training - Step 22177: {'lr': 0.0003119030104790085, 'samples': 4258176, 'steps': 22177, 'loss/train': 2.182638645172119}
+01/29/2022 17:25:55 - INFO - codeparrot_training - Step 22178: {'lr': 0.00031188715742376966, 'samples': 4258368, 'steps': 22178, 'loss/train': 1.304250329732895}
+01/29/2022 17:25:59 - INFO - codeparrot_training - Step 22179: {'lr': 0.0003118713041034259, 'samples': 4258560, 'steps': 22179, 'loss/train': 1.2365865111351013}
+01/29/2022 17:26:03 - INFO - codeparrot_training - Step 22180: {'lr': 0.0003118554505180452, 'samples': 4258752, 'steps': 22180, 'loss/train': 0.9106057584285736}
+01/29/2022 17:26:08 - INFO - codeparrot_training - Step 22181: {'lr': 0.0003118395966676953, 'samples': 4258944, 'steps': 22181, 'loss/train': 1.1767686009407043}
+01/29/2022 17:26:12 - INFO - codeparrot_training - Step 22182: {'lr': 0.00031182374255244426, 'samples': 4259136, 'steps': 22182, 'loss/train': 2.4831928610801697}
+01/29/2022 17:26:17 - INFO - codeparrot_training - Step 22183: {'lr': 0.0003118078881723599, 'samples': 4259328, 'steps': 22183, 'loss/train': 2.1979567408561707}
+01/29/2022 17:26:21 - INFO - codeparrot_training - Step 22184: {'lr': 0.0003117920335275102, 'samples': 4259520, 'steps': 22184, 'loss/train': 1.8036679029464722}
+01/29/2022 17:26:26 - INFO - codeparrot_training - Step 22185: {'lr': 0.000311776178617963, 'samples': 4259712, 'steps': 22185, 'loss/train': 1.111944168806076}
+01/29/2022 17:26:30 - INFO - codeparrot_training - Step 22186: {'lr': 0.0003117603234437864, 'samples': 4259904, 'steps': 22186, 'loss/train': 1.6141878962516785}
+01/29/2022 17:26:34 - INFO - codeparrot_training - Step 22187: {'lr': 0.0003117444680050481, 'samples': 4260096, 'steps': 22187, 'loss/train': 2.6029114723205566}
+01/29/2022 17:26:41 - INFO - codeparrot_training - Step 22188: {'lr': 0.00031172861230181607, 'samples': 4260288, 'steps': 22188, 'loss/train': 1.4934545159339905}
+01/29/2022 17:26:46 - INFO - codeparrot_training - Step 22189: {'lr': 0.0003117127563341583, 'samples': 4260480, 'steps': 22189, 'loss/train': 0.6954122185707092}
+01/29/2022 17:26:50 - INFO - codeparrot_training - Step 22190: {'lr': 0.0003116969001021427, 'samples': 4260672, 'steps': 22190, 'loss/train': 1.2872409224510193}
+01/29/2022 17:26:54 - INFO - codeparrot_training - Step 22191: {'lr': 0.00031168104360583716, 'samples': 4260864, 'steps': 22191, 'loss/train': 1.9373497366905212}
+01/29/2022 17:26:59 - INFO - codeparrot_training - Step 22192: {'lr': 0.00031166518684530966, 'samples': 4261056, 'steps': 22192, 'loss/train': 1.4638495445251465}
+01/29/2022 17:27:03 - INFO - codeparrot_training - Step 22193: {'lr': 0.000311649329820628, 'samples': 4261248, 'steps': 22193, 'loss/train': 1.9079418182373047}
+01/29/2022 17:27:07 - INFO - codeparrot_training - Step 22194: {'lr': 0.0003116334725318602, 'samples': 4261440, 'steps': 22194, 'loss/train': 1.4627099633216858}
+01/29/2022 17:27:13 - INFO - codeparrot_training - Step 22195: {'lr': 0.00031161761497907416, 'samples': 4261632, 'steps': 22195, 'loss/train': 2.284150242805481}
+01/29/2022 17:27:17 - INFO - codeparrot_training - Step 22196: {'lr': 0.00031160175716233793, 'samples': 4261824, 'steps': 22196, 'loss/train': 1.8470045328140259}
+01/29/2022 17:27:21 - INFO - codeparrot_training - Step 22197: {'lr': 0.0003115858990817192, 'samples': 4262016, 'steps': 22197, 'loss/train': 1.015902578830719}
+01/29/2022 17:27:25 - INFO - codeparrot_training - Step 22198: {'lr': 0.0003115700407372861, 'samples': 4262208, 'steps': 22198, 'loss/train': 1.3035476803779602}
+01/29/2022 17:27:29 - INFO - codeparrot_training - Step 22199: {'lr': 0.00031155418212910647, 'samples': 4262400, 'steps': 22199, 'loss/train': 1.7406651377677917}
+01/29/2022 17:27:37 - INFO - codeparrot_training - Step 22200: {'lr': 0.00031153832325724826, 'samples': 4262592, 'steps': 22200, 'loss/train': 1.3886753618717194}
+01/29/2022 17:27:41 - INFO - codeparrot_training - Step 22201: {'lr': 0.0003115224641217795, 'samples': 4262784, 'steps': 22201, 'loss/train': 1.6047552824020386}
+01/29/2022 17:27:45 - INFO - codeparrot_training - Step 22202: {'lr': 0.0003115066047227679, 'samples': 4262976, 'steps': 22202, 'loss/train': 2.1328791975975037}
+01/29/2022 17:27:49 - INFO - codeparrot_training - Step 22203: {'lr': 0.00031149074506028163, 'samples': 4263168, 'steps': 22203, 'loss/train': 2.022697627544403}
+01/29/2022 17:27:54 - INFO - codeparrot_training - Step 22204: {'lr': 0.00031147488513438853, 'samples': 4263360, 'steps': 22204, 'loss/train': 1.7988225817680359}
+01/29/2022 17:27:59 - INFO - codeparrot_training - Step 22205: {'lr': 0.00031145902494515655, 'samples': 4263552, 'steps': 22205, 'loss/train': 2.376934289932251}
+01/29/2022 17:28:03 - INFO - codeparrot_training - Step 22206: {'lr': 0.0003114431644926536, 'samples': 4263744, 'steps': 22206, 'loss/train': 1.2991673648357391}
+01/29/2022 17:28:07 - INFO - codeparrot_training - Step 22207: {'lr': 0.00031142730377694763, 'samples': 4263936, 'steps': 22207, 'loss/train': 1.0995968878269196}
+01/29/2022 17:28:12 - INFO - codeparrot_training - Step 22208: {'lr': 0.0003114114427981066, 'samples': 4264128, 'steps': 22208, 'loss/train': 1.23307666182518}
+01/29/2022 17:28:16 - INFO - codeparrot_training - Step 22209: {'lr': 0.00031139558155619844, 'samples': 4264320, 'steps': 22209, 'loss/train': 2.1707069277763367}
+01/29/2022 17:28:21 - INFO - codeparrot_training - Step 22210: {'lr': 0.0003113797200512912, 'samples': 4264512, 'steps': 22210, 'loss/train': 2.2611259818077087}
+01/29/2022 17:28:25 - INFO - codeparrot_training - Step 22211: {'lr': 0.0003113638582834526, 'samples': 4264704, 'steps': 22211, 'loss/train': 2.2152344584465027}
+01/29/2022 17:28:30 - INFO - codeparrot_training - Step 22212: {'lr': 0.00031134799625275077, 'samples': 4264896, 'steps': 22212, 'loss/train': 1.2821714580059052}
+01/29/2022 17:28:34 - INFO - codeparrot_training - Step 22213: {'lr': 0.0003113321339592536, 'samples': 4265088, 'steps': 22213, 'loss/train': 2.1260620951652527}
+01/29/2022 17:28:38 - INFO - codeparrot_training - Step 22214: {'lr': 0.000311316271403029, 'samples': 4265280, 'steps': 22214, 'loss/train': 1.4978612065315247}
+01/29/2022 17:28:44 - INFO - codeparrot_training - Step 22215: {'lr': 0.00031130040858414506, 'samples': 4265472, 'steps': 22215, 'loss/train': 2.284808099269867}
+01/29/2022 17:28:48 - INFO - codeparrot_training - Step 22216: {'lr': 0.00031128454550266956, 'samples': 4265664, 'steps': 22216, 'loss/train': 0.8572927415370941}
+01/29/2022 17:28:52 - INFO - codeparrot_training - Step 22217: {'lr': 0.0003112686821586706, 'samples': 4265856, 'steps': 22217, 'loss/train': 1.430612862110138}
+01/29/2022 17:28:56 - INFO - codeparrot_training - Step 22218: {'lr': 0.000311252818552216, 'samples': 4266048, 'steps': 22218, 'loss/train': 1.5421258807182312}
+01/29/2022 17:29:01 - INFO - codeparrot_training - Step 22219: {'lr': 0.00031123695468337375, 'samples': 4266240, 'steps': 22219, 'loss/train': 1.5326276421546936}
+01/29/2022 17:29:08 - INFO - codeparrot_training - Step 22220: {'lr': 0.00031122109055221187, 'samples': 4266432, 'steps': 22220, 'loss/train': 1.7968772649765015}
+01/29/2022 17:29:12 - INFO - codeparrot_training - Step 22221: {'lr': 0.00031120522615879834, 'samples': 4266624, 'steps': 22221, 'loss/train': 1.044331133365631}
+01/29/2022 17:29:17 - INFO - codeparrot_training - Step 22222: {'lr': 0.00031118936150320093, 'samples': 4266816, 'steps': 22222, 'loss/train': 1.5359503626823425}
+01/29/2022 17:29:21 - INFO - codeparrot_training - Step 22223: {'lr': 0.00031117349658548783, 'samples': 4267008, 'steps': 22223, 'loss/train': 1.8596057295799255}
+01/29/2022 17:29:25 - INFO - codeparrot_training - Step 22224: {'lr': 0.00031115763140572686, 'samples': 4267200, 'steps': 22224, 'loss/train': 1.3208650946617126}
+01/29/2022 17:29:30 - INFO - codeparrot_training - Step 22225: {'lr': 0.000311141765963986, 'samples': 4267392, 'steps': 22225, 'loss/train': 1.2995184659957886}
+01/29/2022 17:29:35 - INFO - codeparrot_training - Step 22226: {'lr': 0.00031112590026033323, 'samples': 4267584, 'steps': 22226, 'loss/train': 1.6912626028060913}
+01/29/2022 17:29:39 - INFO - codeparrot_training - Step 22227: {'lr': 0.00031111003429483647, 'samples': 4267776, 'steps': 22227, 'loss/train': 1.9819514751434326}
+01/29/2022 17:29:43 - INFO - codeparrot_training - Step 22228: {'lr': 0.00031109416806756387, 'samples': 4267968, 'steps': 22228, 'loss/train': 1.4104354083538055}
+01/29/2022 17:29:47 - INFO - codeparrot_training - Step 22229: {'lr': 0.0003110783015785831, 'samples': 4268160, 'steps': 22229, 'loss/train': 2.4265642762184143}
+01/29/2022 17:29:53 - INFO - codeparrot_training - Step 22230: {'lr': 0.00031106243482796234, 'samples': 4268352, 'steps': 22230, 'loss/train': 1.1850082576274872}
+01/29/2022 17:29:57 - INFO - codeparrot_training - Step 22231: {'lr': 0.0003110465678157695, 'samples': 4268544, 'steps': 22231, 'loss/train': 1.9728258848190308}
+01/29/2022 17:30:01 - INFO - codeparrot_training - Step 22232: {'lr': 0.0003110307005420726, 'samples': 4268736, 'steps': 22232, 'loss/train': 1.3842046558856964}
+01/29/2022 17:30:05 - INFO - codeparrot_training - Step 22233: {'lr': 0.00031101483300693944, 'samples': 4268928, 'steps': 22233, 'loss/train': 1.889630377292633}
+01/29/2022 17:30:10 - INFO - codeparrot_training - Step 22234: {'lr': 0.00031099896521043826, 'samples': 4269120, 'steps': 22234, 'loss/train': 1.580434262752533}
+01/29/2022 17:30:17 - INFO - codeparrot_training - Step 22235: {'lr': 0.0003109830971526367, 'samples': 4269312, 'steps': 22235, 'loss/train': 2.0423081517219543}
+01/29/2022 17:30:21 - INFO - codeparrot_training - Step 22236: {'lr': 0.0003109672288336031, 'samples': 4269504, 'steps': 22236, 'loss/train': 0.8512862026691437}
+01/29/2022 17:30:25 - INFO - codeparrot_training - Step 22237: {'lr': 0.00031095136025340514, 'samples': 4269696, 'steps': 22237, 'loss/train': 1.839228093624115}
+01/29/2022 17:30:29 - INFO - codeparrot_training - Step 22238: {'lr': 0.00031093549141211096, 'samples': 4269888, 'steps': 22238, 'loss/train': 0.4697941839694977}
+01/29/2022 17:30:34 - INFO - codeparrot_training - Step 22239: {'lr': 0.00031091962230978844, 'samples': 4270080, 'steps': 22239, 'loss/train': 1.7374649047851562}
+01/29/2022 17:30:39 - INFO - codeparrot_training - Step 22240: {'lr': 0.0003109037529465056, 'samples': 4270272, 'steps': 22240, 'loss/train': 0.8268062174320221}
+01/29/2022 17:30:43 - INFO - codeparrot_training - Step 22241: {'lr': 0.0003108878833223305, 'samples': 4270464, 'steps': 22241, 'loss/train': 1.6526865363121033}
+01/29/2022 17:30:48 - INFO - codeparrot_training - Step 22242: {'lr': 0.00031087201343733096, 'samples': 4270656, 'steps': 22242, 'loss/train': 1.9559047222137451}
+01/29/2022 17:30:52 - INFO - codeparrot_training - Step 22243: {'lr': 0.00031085614329157515, 'samples': 4270848, 'steps': 22243, 'loss/train': 1.6648249626159668}
+01/29/2022 17:30:56 - INFO - codeparrot_training - Step 22244: {'lr': 0.00031084027288513083, 'samples': 4271040, 'steps': 22244, 'loss/train': 1.9838032722473145}
+01/29/2022 17:31:03 - INFO - codeparrot_training - Step 22245: {'lr': 0.0003108244022180661, 'samples': 4271232, 'steps': 22245, 'loss/train': 1.7340908646583557}
+01/29/2022 17:31:07 - INFO - codeparrot_training - Step 22246: {'lr': 0.000310808531290449, 'samples': 4271424, 'steps': 22246, 'loss/train': 2.201514780521393}
+01/29/2022 17:31:12 - INFO - codeparrot_training - Step 22247: {'lr': 0.00031079266010234746, 'samples': 4271616, 'steps': 22247, 'loss/train': 0.9347780048847198}
+01/29/2022 17:31:16 - INFO - codeparrot_training - Step 22248: {'lr': 0.00031077678865382944, 'samples': 4271808, 'steps': 22248, 'loss/train': 1.3156985342502594}
+01/29/2022 17:31:20 - INFO - codeparrot_training - Step 22249: {'lr': 0.000310760916944963, 'samples': 4272000, 'steps': 22249, 'loss/train': 2.321380913257599}
+01/29/2022 17:31:25 - INFO - codeparrot_training - Step 22250: {'lr': 0.000310745044975816, 'samples': 4272192, 'steps': 22250, 'loss/train': 0.6835193485021591}
+01/29/2022 17:31:30 - INFO - codeparrot_training - Step 22251: {'lr': 0.00031072917274645656, 'samples': 4272384, 'steps': 22251, 'loss/train': 1.3493591845035553}
+01/29/2022 17:31:34 - INFO - codeparrot_training - Step 22252: {'lr': 0.00031071330025695266, 'samples': 4272576, 'steps': 22252, 'loss/train': 1.6082607507705688}
+01/29/2022 17:31:38 - INFO - codeparrot_training - Step 22253: {'lr': 0.0003106974275073722, 'samples': 4272768, 'steps': 22253, 'loss/train': 1.1315776705741882}
+01/29/2022 17:31:42 - INFO - codeparrot_training - Step 22254: {'lr': 0.0003106815544977833, 'samples': 4272960, 'steps': 22254, 'loss/train': 0.9941892921924591}
+01/29/2022 17:31:48 - INFO - codeparrot_training - Step 22255: {'lr': 0.00031066568122825383, 'samples': 4273152, 'steps': 22255, 'loss/train': 0.7491371333599091}
+01/29/2022 17:31:52 - INFO - codeparrot_training - Step 22256: {'lr': 0.0003106498076988519, 'samples': 4273344, 'steps': 22256, 'loss/train': 1.4598349034786224}
+01/29/2022 17:31:56 - INFO - codeparrot_training - Step 22257: {'lr': 0.0003106339339096454, 'samples': 4273536, 'steps': 22257, 'loss/train': 1.4117978811264038}
+01/29/2022 17:32:00 - INFO - codeparrot_training - Step 22258: {'lr': 0.0003106180598607024, 'samples': 4273728, 'steps': 22258, 'loss/train': 1.5366851091384888}
+01/29/2022 17:32:05 - INFO - codeparrot_training - Step 22259: {'lr': 0.00031060218555209094, 'samples': 4273920, 'steps': 22259, 'loss/train': 1.5645892024040222}
+01/29/2022 17:32:10 - INFO - codeparrot_training - Step 22260: {'lr': 0.0003105863109838789, 'samples': 4274112, 'steps': 22260, 'loss/train': 1.5764747858047485}
+01/29/2022 17:32:14 - INFO - codeparrot_training - Step 22261: {'lr': 0.0003105704361561343, 'samples': 4274304, 'steps': 22261, 'loss/train': 0.8120325207710266}
+01/29/2022 17:32:19 - INFO - codeparrot_training - Step 22262: {'lr': 0.00031055456106892526, 'samples': 4274496, 'steps': 22262, 'loss/train': 1.0418902337551117}
+01/29/2022 17:32:23 - INFO - codeparrot_training - Step 22263: {'lr': 0.0003105386857223197, 'samples': 4274688, 'steps': 22263, 'loss/train': 1.3311922252178192}
+01/29/2022 17:32:27 - INFO - codeparrot_training - Step 22264: {'lr': 0.0003105228101163856, 'samples': 4274880, 'steps': 22264, 'loss/train': 0.7109927237033844}
+01/29/2022 17:32:34 - INFO - codeparrot_training - Step 22265: {'lr': 0.000310506934251191, 'samples': 4275072, 'steps': 22265, 'loss/train': 0.811455488204956}
+01/29/2022 17:32:38 - INFO - codeparrot_training - Step 22266: {'lr': 0.0003104910581268039, 'samples': 4275264, 'steps': 22266, 'loss/train': 1.5158439874649048}
+01/29/2022 17:32:43 - INFO - codeparrot_training - Step 22267: {'lr': 0.00031047518174329234, 'samples': 4275456, 'steps': 22267, 'loss/train': 1.293841928243637}
+01/29/2022 17:32:47 - INFO - codeparrot_training - Step 22268: {'lr': 0.00031045930510072427, 'samples': 4275648, 'steps': 22268, 'loss/train': 0.9207302033901215}
+01/29/2022 17:32:51 - INFO - codeparrot_training - Step 22269: {'lr': 0.00031044342819916784, 'samples': 4275840, 'steps': 22269, 'loss/train': 0.7353072613477707}
+01/29/2022 17:32:57 - INFO - codeparrot_training - Step 22270: {'lr': 0.0003104275510386908, 'samples': 4276032, 'steps': 22270, 'loss/train': 1.902368187904358}
+01/29/2022 17:33:01 - INFO - codeparrot_training - Step 22271: {'lr': 0.0003104116736193615, 'samples': 4276224, 'steps': 22271, 'loss/train': 1.2144276201725006}
+01/29/2022 17:33:05 - INFO - codeparrot_training - Step 22272: {'lr': 0.00031039579594124763, 'samples': 4276416, 'steps': 22272, 'loss/train': 1.589021623134613}
+01/29/2022 17:33:09 - INFO - codeparrot_training - Step 22273: {'lr': 0.0003103799180044174, 'samples': 4276608, 'steps': 22273, 'loss/train': 0.9128741025924683}
+01/29/2022 17:33:14 - INFO - codeparrot_training - Step 22274: {'lr': 0.00031036403980893874, 'samples': 4276800, 'steps': 22274, 'loss/train': 1.0794734358787537}
+01/29/2022 17:33:19 - INFO - codeparrot_training - Step 22275: {'lr': 0.0003103481613548797, 'samples': 4276992, 'steps': 22275, 'loss/train': 2.076161205768585}
+01/29/2022 17:33:23 - INFO - codeparrot_training - Step 22276: {'lr': 0.00031033228264230834, 'samples': 4277184, 'steps': 22276, 'loss/train': 1.8049179911613464}
+01/29/2022 17:33:27 - INFO - codeparrot_training - Step 22277: {'lr': 0.0003103164036712926, 'samples': 4277376, 'steps': 22277, 'loss/train': 1.7470189332962036}
+01/29/2022 17:33:32 - INFO - codeparrot_training - Step 22278: {'lr': 0.0003103005244419006, 'samples': 4277568, 'steps': 22278, 'loss/train': 2.076910972595215}
+01/29/2022 17:33:36 - INFO - codeparrot_training - Step 22279: {'lr': 0.00031028464495420026, 'samples': 4277760, 'steps': 22279, 'loss/train': 2.146156132221222}
+01/29/2022 17:33:43 - INFO - codeparrot_training - Step 22280: {'lr': 0.0003102687652082597, 'samples': 4277952, 'steps': 22280, 'loss/train': 0.6180006712675095}
+01/29/2022 17:33:47 - INFO - codeparrot_training - Step 22281: {'lr': 0.00031025288520414686, 'samples': 4278144, 'steps': 22281, 'loss/train': 1.481559008359909}
+01/29/2022 17:33:52 - INFO - codeparrot_training - Step 22282: {'lr': 0.0003102370049419297, 'samples': 4278336, 'steps': 22282, 'loss/train': 1.5847722887992859}
+01/29/2022 17:33:56 - INFO - codeparrot_training - Step 22283: {'lr': 0.0003102211244216764, 'samples': 4278528, 'steps': 22283, 'loss/train': 1.6487449407577515}
+01/29/2022 17:34:00 - INFO - codeparrot_training - Step 22284: {'lr': 0.000310205243643455, 'samples': 4278720, 'steps': 22284, 'loss/train': 1.7624748945236206}
+01/29/2022 17:34:05 - INFO - codeparrot_training - Step 22285: {'lr': 0.00031018936260733337, 'samples': 4278912, 'steps': 22285, 'loss/train': 1.6961397528648376}
+01/29/2022 17:34:10 - INFO - codeparrot_training - Step 22286: {'lr': 0.00031017348131337963, 'samples': 4279104, 'steps': 22286, 'loss/train': 1.9290170073509216}
+01/29/2022 17:34:14 - INFO - codeparrot_training - Step 22287: {'lr': 0.00031015759976166186, 'samples': 4279296, 'steps': 22287, 'loss/train': 1.5726308226585388}
+01/29/2022 17:34:18 - INFO - codeparrot_training - Step 22288: {'lr': 0.00031014171795224794, 'samples': 4279488, 'steps': 22288, 'loss/train': 1.4176044166088104}
+01/29/2022 17:34:22 - INFO - codeparrot_training - Step 22289: {'lr': 0.00031012583588520607, 'samples': 4279680, 'steps': 22289, 'loss/train': 1.9355388879776}
+01/29/2022 17:34:29 - INFO - codeparrot_training - Step 22290: {'lr': 0.00031010995356060416, 'samples': 4279872, 'steps': 22290, 'loss/train': 1.3103908896446228}
+01/29/2022 17:34:34 - INFO - codeparrot_training - Step 22291: {'lr': 0.00031009407097851036, 'samples': 4280064, 'steps': 22291, 'loss/train': 1.4351138770580292}
+01/29/2022 17:34:38 - INFO - codeparrot_training - Step 22292: {'lr': 0.0003100781881389926, 'samples': 4280256, 'steps': 22292, 'loss/train': 0.7290780544281006}
+01/29/2022 17:34:42 - INFO - codeparrot_training - Step 22293: {'lr': 0.00031006230504211895, 'samples': 4280448, 'steps': 22293, 'loss/train': 2.0411185026168823}
+01/29/2022 17:34:46 - INFO - codeparrot_training - Step 22294: {'lr': 0.0003100464216879574, 'samples': 4280640, 'steps': 22294, 'loss/train': 1.146393746137619}
+01/29/2022 17:34:52 - INFO - codeparrot_training - Step 22295: {'lr': 0.0003100305380765762, 'samples': 4280832, 'steps': 22295, 'loss/train': 1.8149932622909546}
+01/29/2022 17:34:56 - INFO - codeparrot_training - Step 22296: {'lr': 0.00031001465420804316, 'samples': 4281024, 'steps': 22296, 'loss/train': 2.023344576358795}
+01/29/2022 17:35:00 - INFO - codeparrot_training - Step 22297: {'lr': 0.0003099987700824264, 'samples': 4281216, 'steps': 22297, 'loss/train': 1.3968597650527954}
+01/29/2022 17:35:04 - INFO - codeparrot_training - Step 22298: {'lr': 0.00030998288569979393, 'samples': 4281408, 'steps': 22298, 'loss/train': 2.2361006140708923}
+01/29/2022 17:35:09 - INFO - codeparrot_training - Step 22299: {'lr': 0.0003099670010602138, 'samples': 4281600, 'steps': 22299, 'loss/train': 1.6231005191802979}
+01/29/2022 17:35:14 - INFO - codeparrot_training - Step 22300: {'lr': 0.00030995111616375417, 'samples': 4281792, 'steps': 22300, 'loss/train': 1.4484961330890656}
+01/29/2022 17:35:18 - INFO - codeparrot_training - Step 22301: {'lr': 0.00030993523101048294, 'samples': 4281984, 'steps': 22301, 'loss/train': 0.8413499593734741}
+01/29/2022 17:35:22 - INFO - codeparrot_training - Step 22302: {'lr': 0.0003099193456004682, 'samples': 4282176, 'steps': 22302, 'loss/train': 0.660102054476738}
+01/29/2022 17:35:27 - INFO - codeparrot_training - Step 22303: {'lr': 0.00030990345993377807, 'samples': 4282368, 'steps': 22303, 'loss/train': 1.9921563863754272}
+01/29/2022 17:35:31 - INFO - codeparrot_training - Step 22304: {'lr': 0.0003098875740104805, 'samples': 4282560, 'steps': 22304, 'loss/train': 2.132050037384033}
+01/29/2022 17:35:38 - INFO - codeparrot_training - Step 22305: {'lr': 0.00030987168783064355, 'samples': 4282752, 'steps': 22305, 'loss/train': 1.6431142687797546}
+01/29/2022 17:35:42 - INFO - codeparrot_training - Step 22306: {'lr': 0.0003098558013943353, 'samples': 4282944, 'steps': 22306, 'loss/train': 2.222215175628662}
+01/29/2022 17:35:46 - INFO - codeparrot_training - Step 22307: {'lr': 0.00030983991470162386, 'samples': 4283136, 'steps': 22307, 'loss/train': 1.9035723209381104}
+01/29/2022 17:35:51 - INFO - codeparrot_training - Step 22308: {'lr': 0.00030982402775257725, 'samples': 4283328, 'steps': 22308, 'loss/train': 2.1157860159873962}
+01/29/2022 17:35:55 - INFO - codeparrot_training - Step 22309: {'lr': 0.0003098081405472634, 'samples': 4283520, 'steps': 22309, 'loss/train': 2.3537063598632812}
+01/29/2022 17:36:00 - INFO - codeparrot_training - Step 22310: {'lr': 0.0003097922530857505, 'samples': 4283712, 'steps': 22310, 'loss/train': 1.9063045978546143}
+01/29/2022 17:36:05 - INFO - codeparrot_training - Step 22311: {'lr': 0.0003097763653681066, 'samples': 4283904, 'steps': 22311, 'loss/train': 1.2291037738323212}
+01/29/2022 17:36:09 - INFO - codeparrot_training - Step 22312: {'lr': 0.00030976047739439974, 'samples': 4284096, 'steps': 22312, 'loss/train': 1.926752507686615}
+01/29/2022 17:36:13 - INFO - codeparrot_training - Step 22313: {'lr': 0.000309744589164698, 'samples': 4284288, 'steps': 22313, 'loss/train': 2.6930195689201355}
+01/29/2022 17:36:17 - INFO - codeparrot_training - Step 22314: {'lr': 0.00030972870067906934, 'samples': 4284480, 'steps': 22314, 'loss/train': 2.834292948246002}
+01/29/2022 17:36:23 - INFO - codeparrot_training - Step 22315: {'lr': 0.00030971281193758197, 'samples': 4284672, 'steps': 22315, 'loss/train': 2.2319204807281494}
+01/29/2022 17:36:27 - INFO - codeparrot_training - Step 22316: {'lr': 0.00030969692294030376, 'samples': 4284864, 'steps': 22316, 'loss/train': 2.0661458373069763}
+01/29/2022 17:36:31 - INFO - codeparrot_training - Step 22317: {'lr': 0.000309681033687303, 'samples': 4285056, 'steps': 22317, 'loss/train': 1.3257966935634613}
+01/29/2022 17:36:35 - INFO - codeparrot_training - Step 22318: {'lr': 0.0003096651441786476, 'samples': 4285248, 'steps': 22318, 'loss/train': 2.265101909637451}
+01/29/2022 17:36:40 - INFO - codeparrot_training - Step 22319: {'lr': 0.0003096492544144056, 'samples': 4285440, 'steps': 22319, 'loss/train': 1.5896835923194885}
+01/29/2022 17:36:45 - INFO - codeparrot_training - Step 22320: {'lr': 0.00030963336439464523, 'samples': 4285632, 'steps': 22320, 'loss/train': 1.6637864112854004}
+01/29/2022 17:36:49 - INFO - codeparrot_training - Step 22321: {'lr': 0.0003096174741194344, 'samples': 4285824, 'steps': 22321, 'loss/train': 1.4753691852092743}
+01/29/2022 17:36:53 - INFO - codeparrot_training - Step 22322: {'lr': 0.00030960158358884127, 'samples': 4286016, 'steps': 22322, 'loss/train': 1.7418606877326965}
+01/29/2022 17:36:58 - INFO - codeparrot_training - Step 22323: {'lr': 0.0003095856928029339, 'samples': 4286208, 'steps': 22323, 'loss/train': 1.9747254252433777}
+01/29/2022 17:37:02 - INFO - codeparrot_training - Step 22324: {'lr': 0.00030956980176178033, 'samples': 4286400, 'steps': 22324, 'loss/train': 1.3094801008701324}
+01/29/2022 17:37:10 - INFO - codeparrot_training - Step 22325: {'lr': 0.00030955391046544865, 'samples': 4286592, 'steps': 22325, 'loss/train': 1.7868545651435852}
+01/29/2022 17:37:14 - INFO - codeparrot_training - Step 22326: {'lr': 0.0003095380189140069, 'samples': 4286784, 'steps': 22326, 'loss/train': 0.8536050617694855}
+01/29/2022 17:37:19 - INFO - codeparrot_training - Step 22327: {'lr': 0.00030952212710752325, 'samples': 4286976, 'steps': 22327, 'loss/train': 2.478178918361664}
+01/29/2022 17:37:23 - INFO - codeparrot_training - Step 22328: {'lr': 0.00030950623504606565, 'samples': 4287168, 'steps': 22328, 'loss/train': 2.2365174293518066}
+01/29/2022 17:37:27 - INFO - codeparrot_training - Step 22329: {'lr': 0.0003094903427297023, 'samples': 4287360, 'steps': 22329, 'loss/train': 1.9033849239349365}
+01/29/2022 17:37:33 - INFO - codeparrot_training - Step 22330: {'lr': 0.00030947445015850114, 'samples': 4287552, 'steps': 22330, 'loss/train': 2.090012848377228}
+01/29/2022 17:37:37 - INFO - codeparrot_training - Step 22331: {'lr': 0.0003094585573325303, 'samples': 4287744, 'steps': 22331, 'loss/train': 1.7306132912635803}
+01/29/2022 17:37:41 - INFO - codeparrot_training - Step 22332: {'lr': 0.00030944266425185794, 'samples': 4287936, 'steps': 22332, 'loss/train': 1.3800647556781769}
+01/29/2022 17:37:45 - INFO - codeparrot_training - Step 22333: {'lr': 0.000309426770916552, 'samples': 4288128, 'steps': 22333, 'loss/train': 1.7704979181289673}
+01/29/2022 17:37:50 - INFO - codeparrot_training - Step 22334: {'lr': 0.0003094108773266808, 'samples': 4288320, 'steps': 22334, 'loss/train': 1.5354090929031372}
+01/29/2022 17:37:54 - INFO - codeparrot_training - Step 22335: {'lr': 0.00030939498348231214, 'samples': 4288512, 'steps': 22335, 'loss/train': 1.1425690054893494}
+01/29/2022 17:37:59 - INFO - codeparrot_training - Step 22336: {'lr': 0.00030937908938351424, 'samples': 4288704, 'steps': 22336, 'loss/train': 1.8589419722557068}
+01/29/2022 17:38:03 - INFO - codeparrot_training - Step 22337: {'lr': 0.0003093631950303552, 'samples': 4288896, 'steps': 22337, 'loss/train': 0.7466406673192978}
+01/29/2022 17:38:08 - INFO - codeparrot_training - Step 22338: {'lr': 0.00030934730042290304, 'samples': 4289088, 'steps': 22338, 'loss/train': 1.763006865978241}
+01/29/2022 17:38:12 - INFO - codeparrot_training - Step 22339: {'lr': 0.00030933140556122597, 'samples': 4289280, 'steps': 22339, 'loss/train': 1.777713418006897}
+01/29/2022 17:38:16 - INFO - codeparrot_training - Step 22340: {'lr': 0.00030931551044539196, 'samples': 4289472, 'steps': 22340, 'loss/train': 2.299822211265564}
+01/29/2022 17:38:23 - INFO - codeparrot_training - Step 22341: {'lr': 0.00030929961507546915, 'samples': 4289664, 'steps': 22341, 'loss/train': 1.2725092470645905}
+01/29/2022 17:38:27 - INFO - codeparrot_training - Step 22342: {'lr': 0.0003092837194515256, 'samples': 4289856, 'steps': 22342, 'loss/train': 1.8722877502441406}
+01/29/2022 17:38:32 - INFO - codeparrot_training - Step 22343: {'lr': 0.00030926782357362944, 'samples': 4290048, 'steps': 22343, 'loss/train': 2.0432660579681396}
+01/29/2022 17:38:36 - INFO - codeparrot_training - Step 22344: {'lr': 0.0003092519274418487, 'samples': 4290240, 'steps': 22344, 'loss/train': 1.704979419708252}
+01/29/2022 17:38:41 - INFO - codeparrot_training - Step 22345: {'lr': 0.0003092360310562516, 'samples': 4290432, 'steps': 22345, 'loss/train': 1.213603287935257}
+01/29/2022 17:38:45 - INFO - codeparrot_training - Step 22346: {'lr': 0.00030922013441690607, 'samples': 4290624, 'steps': 22346, 'loss/train': 1.7939379215240479}
+01/29/2022 17:38:50 - INFO - codeparrot_training - Step 22347: {'lr': 0.0003092042375238803, 'samples': 4290816, 'steps': 22347, 'loss/train': 1.5799673795700073}
+01/29/2022 17:38:54 - INFO - codeparrot_training - Step 22348: {'lr': 0.00030918834037724235, 'samples': 4291008, 'steps': 22348, 'loss/train': 2.3476885557174683}
+01/29/2022 17:38:58 - INFO - codeparrot_training - Step 22349: {'lr': 0.0003091724429770604, 'samples': 4291200, 'steps': 22349, 'loss/train': 1.6843151450157166}
+01/29/2022 17:39:05 - INFO - codeparrot_training - Step 22350: {'lr': 0.0003091565453234025, 'samples': 4291392, 'steps': 22350, 'loss/train': 1.0479763448238373}
+01/29/2022 17:39:10 - INFO - codeparrot_training - Step 22351: {'lr': 0.0003091406474163367, 'samples': 4291584, 'steps': 22351, 'loss/train': 1.6226261258125305}
+01/29/2022 17:39:14 - INFO - codeparrot_training - Step 22352: {'lr': 0.0003091247492559312, 'samples': 4291776, 'steps': 22352, 'loss/train': 0.8822035789489746}
+01/29/2022 17:39:18 - INFO - codeparrot_training - Step 22353: {'lr': 0.00030910885084225396, 'samples': 4291968, 'steps': 22353, 'loss/train': 1.8700103759765625}
+01/29/2022 17:39:22 - INFO - codeparrot_training - Step 22354: {'lr': 0.0003090929521753733, 'samples': 4292160, 'steps': 22354, 'loss/train': 2.0320900082588196}
+01/29/2022 17:39:27 - INFO - codeparrot_training - Step 22355: {'lr': 0.00030907705325535704, 'samples': 4292352, 'steps': 22355, 'loss/train': 2.1835582852363586}
+01/29/2022 17:39:32 - INFO - codeparrot_training - Step 22356: {'lr': 0.0003090611540822736, 'samples': 4292544, 'steps': 22356, 'loss/train': 1.2920957207679749}
+01/29/2022 17:39:36 - INFO - codeparrot_training - Step 22357: {'lr': 0.0003090452546561908, 'samples': 4292736, 'steps': 22357, 'loss/train': 1.5857702493667603}
+01/29/2022 17:39:41 - INFO - codeparrot_training - Step 22358: {'lr': 0.000309029354977177, 'samples': 4292928, 'steps': 22358, 'loss/train': 0.9066223204135895}
+01/29/2022 17:39:45 - INFO - codeparrot_training - Step 22359: {'lr': 0.00030901345504530007, 'samples': 4293120, 'steps': 22359, 'loss/train': 1.1899998486042023}
+01/29/2022 17:39:49 - INFO - codeparrot_training - Step 22360: {'lr': 0.0003089975548606283, 'samples': 4293312, 'steps': 22360, 'loss/train': 0.5644245743751526}
+01/29/2022 17:39:55 - INFO - codeparrot_training - Step 22361: {'lr': 0.0003089816544232298, 'samples': 4293504, 'steps': 22361, 'loss/train': 1.7287184000015259}
+01/29/2022 17:39:59 - INFO - codeparrot_training - Step 22362: {'lr': 0.00030896575373317247, 'samples': 4293696, 'steps': 22362, 'loss/train': 2.126150965690613}
+01/29/2022 17:40:03 - INFO - codeparrot_training - Step 22363: {'lr': 0.0003089498527905247, 'samples': 4293888, 'steps': 22363, 'loss/train': 1.7723152041435242}
+01/29/2022 17:40:07 - INFO - codeparrot_training - Step 22364: {'lr': 0.00030893395159535444, 'samples': 4294080, 'steps': 22364, 'loss/train': 0.7271728813648224}
+01/29/2022 17:40:12 - INFO - codeparrot_training - Step 22365: {'lr': 0.00030891805014772987, 'samples': 4294272, 'steps': 22365, 'loss/train': 1.3441012501716614}
+01/29/2022 17:40:16 - INFO - codeparrot_training - Step 22366: {'lr': 0.000308902148447719, 'samples': 4294464, 'steps': 22366, 'loss/train': 2.6977989077568054}
+01/29/2022 17:40:23 - INFO - codeparrot_training - Step 22367: {'lr': 0.00030888624649539015, 'samples': 4294656, 'steps': 22367, 'loss/train': 1.4721782505512238}
+01/29/2022 17:40:27 - INFO - codeparrot_training - Step 22368: {'lr': 0.0003088703442908112, 'samples': 4294848, 'steps': 22368, 'loss/train': 0.4726148396730423}
+01/29/2022 17:40:31 - INFO - codeparrot_training - Step 22369: {'lr': 0.0003088544418340505, 'samples': 4295040, 'steps': 22369, 'loss/train': 1.6314157247543335}
+01/29/2022 17:40:36 - INFO - codeparrot_training - Step 22370: {'lr': 0.000308838539125176, 'samples': 4295232, 'steps': 22370, 'loss/train': 1.34530770778656}
+01/29/2022 17:40:40 - INFO - codeparrot_training - Step 22371: {'lr': 0.00030882263616425587, 'samples': 4295424, 'steps': 22371, 'loss/train': 1.3191091418266296}
+01/29/2022 17:40:45 - INFO - codeparrot_training - Step 22372: {'lr': 0.0003088067329513583, 'samples': 4295616, 'steps': 22372, 'loss/train': 1.6502763032913208}
+01/29/2022 17:40:49 - INFO - codeparrot_training - Step 22373: {'lr': 0.0003087908294865513, 'samples': 4295808, 'steps': 22373, 'loss/train': 1.7070606350898743}
+01/29/2022 17:40:54 - INFO - codeparrot_training - Step 22374: {'lr': 0.00030877492576990306, 'samples': 4296000, 'steps': 22374, 'loss/train': 1.7726672887802124}
+01/29/2022 17:40:58 - INFO - codeparrot_training - Step 22375: {'lr': 0.0003087590218014817, 'samples': 4296192, 'steps': 22375, 'loss/train': 1.4497633874416351}
+01/29/2022 17:41:02 - INFO - codeparrot_training - Step 22376: {'lr': 0.00030874311758135535, 'samples': 4296384, 'steps': 22376, 'loss/train': 2.1837775111198425}
+01/29/2022 17:41:09 - INFO - codeparrot_training - Step 22377: {'lr': 0.00030872721310959216, 'samples': 4296576, 'steps': 22377, 'loss/train': 2.0151114463806152}
+01/29/2022 17:41:14 - INFO - codeparrot_training - Step 22378: {'lr': 0.00030871130838626025, 'samples': 4296768, 'steps': 22378, 'loss/train': 2.0032755732536316}
+01/29/2022 17:41:18 - INFO - codeparrot_training - Step 22379: {'lr': 0.0003086954034114277, 'samples': 4296960, 'steps': 22379, 'loss/train': 1.018265962600708}
+01/29/2022 17:41:22 - INFO - codeparrot_training - Step 22380: {'lr': 0.00030867949818516274, 'samples': 4297152, 'steps': 22380, 'loss/train': 1.2838378250598907}
+01/29/2022 17:41:26 - INFO - codeparrot_training - Step 22381: {'lr': 0.00030866359270753337, 'samples': 4297344, 'steps': 22381, 'loss/train': 1.8699955344200134}
+01/29/2022 17:41:32 - INFO - codeparrot_training - Step 22382: {'lr': 0.00030864768697860784, 'samples': 4297536, 'steps': 22382, 'loss/train': 1.6996850967407227}
+01/29/2022 17:41:36 - INFO - codeparrot_training - Step 22383: {'lr': 0.0003086317809984542, 'samples': 4297728, 'steps': 22383, 'loss/train': 0.8695022463798523}
+01/29/2022 17:41:40 - INFO - codeparrot_training - Step 22384: {'lr': 0.0003086158747671406, 'samples': 4297920, 'steps': 22384, 'loss/train': 1.8485605716705322}
+01/29/2022 17:41:44 - INFO - codeparrot_training - Step 22385: {'lr': 0.0003085999682847353, 'samples': 4298112, 'steps': 22385, 'loss/train': 1.7691872119903564}
+01/29/2022 17:41:49 - INFO - codeparrot_training - Step 22386: {'lr': 0.00030858406155130625, 'samples': 4298304, 'steps': 22386, 'loss/train': 1.9380258321762085}
+01/29/2022 17:41:54 - INFO - codeparrot_training - Step 22387: {'lr': 0.00030856815456692177, 'samples': 4298496, 'steps': 22387, 'loss/train': 0.49799448251724243}
+01/29/2022 17:41:58 - INFO - codeparrot_training - Step 22388: {'lr': 0.00030855224733164987, 'samples': 4298688, 'steps': 22388, 'loss/train': 1.6145399808883667}
+01/29/2022 17:42:02 - INFO - codeparrot_training - Step 22389: {'lr': 0.00030853633984555875, 'samples': 4298880, 'steps': 22389, 'loss/train': 1.6426806449890137}
+01/29/2022 17:42:07 - INFO - codeparrot_training - Step 22390: {'lr': 0.0003085204321087165, 'samples': 4299072, 'steps': 22390, 'loss/train': 1.0875738859176636}
+01/29/2022 17:42:11 - INFO - codeparrot_training - Step 22391: {'lr': 0.00030850452412119135, 'samples': 4299264, 'steps': 22391, 'loss/train': 1.536019206047058}
+01/29/2022 17:42:16 - INFO - codeparrot_training - Step 22392: {'lr': 0.00030848861588305136, 'samples': 4299456, 'steps': 22392, 'loss/train': 2.095602035522461}
+01/29/2022 17:42:20 - INFO - codeparrot_training - Step 22393: {'lr': 0.0003084727073943648, 'samples': 4299648, 'steps': 22393, 'loss/train': 1.5165832042694092}
+01/29/2022 17:42:25 - INFO - codeparrot_training - Step 22394: {'lr': 0.0003084567986551996, 'samples': 4299840, 'steps': 22394, 'loss/train': 1.2301273047924042}
+01/29/2022 17:42:29 - INFO - codeparrot_training - Step 22395: {'lr': 0.0003084408896656241, 'samples': 4300032, 'steps': 22395, 'loss/train': 1.1612260043621063}
+01/29/2022 17:42:33 - INFO - codeparrot_training - Step 22396: {'lr': 0.0003084249804257064, 'samples': 4300224, 'steps': 22396, 'loss/train': 1.5454999208450317}
+01/29/2022 17:42:40 - INFO - codeparrot_training - Step 22397: {'lr': 0.00030840907093551456, 'samples': 4300416, 'steps': 22397, 'loss/train': 1.8851879239082336}
+01/29/2022 17:42:45 - INFO - codeparrot_training - Step 22398: {'lr': 0.0003083931611951169, 'samples': 4300608, 'steps': 22398, 'loss/train': 2.304607629776001}
+01/29/2022 17:42:49 - INFO - codeparrot_training - Step 22399: {'lr': 0.0003083772512045814, 'samples': 4300800, 'steps': 22399, 'loss/train': 0.30224888026714325}
+01/29/2022 17:42:53 - INFO - codeparrot_training - Step 22400: {'lr': 0.0003083613409639764, 'samples': 4300992, 'steps': 22400, 'loss/train': 1.338991105556488}
+01/29/2022 17:42:57 - INFO - codeparrot_training - Step 22401: {'lr': 0.0003083454304733698, 'samples': 4301184, 'steps': 22401, 'loss/train': 2.3591830730438232}
+01/29/2022 17:43:03 - INFO - codeparrot_training - Step 22402: {'lr': 0.00030832951973283, 'samples': 4301376, 'steps': 22402, 'loss/train': 2.3476309776306152}
+01/29/2022 17:43:07 - INFO - codeparrot_training - Step 22403: {'lr': 0.000308313608742425, 'samples': 4301568, 'steps': 22403, 'loss/train': 1.132750242948532}
+01/29/2022 17:43:11 - INFO - codeparrot_training - Step 22404: {'lr': 0.00030829769750222315, 'samples': 4301760, 'steps': 22404, 'loss/train': 1.8457310199737549}
+01/29/2022 17:43:16 - INFO - codeparrot_training - Step 22405: {'lr': 0.00030828178601229235, 'samples': 4301952, 'steps': 22405, 'loss/train': 1.5083738565444946}
+01/29/2022 17:43:21 - INFO - codeparrot_training - Step 22406: {'lr': 0.00030826587427270095, 'samples': 4302144, 'steps': 22406, 'loss/train': 2.1059165596961975}
+01/29/2022 17:43:25 - INFO - codeparrot_training - Step 22407: {'lr': 0.000308249962283517, 'samples': 4302336, 'steps': 22407, 'loss/train': 1.6618793606758118}
+01/29/2022 17:43:29 - INFO - codeparrot_training - Step 22408: {'lr': 0.0003082340500448087, 'samples': 4302528, 'steps': 22408, 'loss/train': 0.7046957910060883}
+01/29/2022 17:43:34 - INFO - codeparrot_training - Step 22409: {'lr': 0.0003082181375566443, 'samples': 4302720, 'steps': 22409, 'loss/train': 2.4371745586395264}
+01/29/2022 17:43:38 - INFO - codeparrot_training - Step 22410: {'lr': 0.0003082022248190918, 'samples': 4302912, 'steps': 22410, 'loss/train': 1.1420212090015411}
+01/29/2022 17:43:45 - INFO - codeparrot_training - Step 22411: {'lr': 0.00030818631183221945, 'samples': 4303104, 'steps': 22411, 'loss/train': 1.1297741532325745}
+01/29/2022 17:43:49 - INFO - codeparrot_training - Step 22412: {'lr': 0.0003081703985960955, 'samples': 4303296, 'steps': 22412, 'loss/train': 1.4712757766246796}
+01/29/2022 17:43:54 - INFO - codeparrot_training - Step 22413: {'lr': 0.00030815448511078796, 'samples': 4303488, 'steps': 22413, 'loss/train': 1.00458624958992}
+01/29/2022 17:43:58 - INFO - codeparrot_training - Step 22414: {'lr': 0.0003081385713763651, 'samples': 4303680, 'steps': 22414, 'loss/train': 1.9919090867042542}
+01/29/2022 17:44:02 - INFO - codeparrot_training - Step 22415: {'lr': 0.00030812265739289503, 'samples': 4303872, 'steps': 22415, 'loss/train': 1.8439516425132751}
+01/29/2022 17:44:07 - INFO - codeparrot_training - Step 22416: {'lr': 0.000308106743160446, 'samples': 4304064, 'steps': 22416, 'loss/train': 2.065950036048889}
+01/29/2022 17:44:12 - INFO - codeparrot_training - Step 22417: {'lr': 0.00030809082867908614, 'samples': 4304256, 'steps': 22417, 'loss/train': 1.742898166179657}
+01/29/2022 17:44:16 - INFO - codeparrot_training - Step 22418: {'lr': 0.0003080749139488836, 'samples': 4304448, 'steps': 22418, 'loss/train': 0.36529695242643356}
+01/29/2022 17:44:20 - INFO - codeparrot_training - Step 22419: {'lr': 0.0003080589989699066, 'samples': 4304640, 'steps': 22419, 'loss/train': 0.9063854813575745}
+01/29/2022 17:44:24 - INFO - codeparrot_training - Step 22420: {'lr': 0.00030804308374222315, 'samples': 4304832, 'steps': 22420, 'loss/train': 1.8355937004089355}
+01/29/2022 17:44:31 - INFO - codeparrot_training - Step 22421: {'lr': 0.00030802716826590164, 'samples': 4305024, 'steps': 22421, 'loss/train': 1.2540223002433777}
+01/29/2022 17:44:36 - INFO - codeparrot_training - Step 22422: {'lr': 0.0003080112525410102, 'samples': 4305216, 'steps': 22422, 'loss/train': 1.1382534205913544}
+01/29/2022 17:44:40 - INFO - codeparrot_training - Step 22423: {'lr': 0.0003079953365676169, 'samples': 4305408, 'steps': 22423, 'loss/train': 2.1711888313293457}
+01/29/2022 17:44:44 - INFO - codeparrot_training - Step 22424: {'lr': 0.00030797942034579013, 'samples': 4305600, 'steps': 22424, 'loss/train': 1.7582064270973206}
+01/29/2022 17:44:48 - INFO - codeparrot_training - Step 22425: {'lr': 0.0003079635038755978, 'samples': 4305792, 'steps': 22425, 'loss/train': 1.579576849937439}
+01/29/2022 17:44:53 - INFO - codeparrot_training - Step 22426: {'lr': 0.0003079475871571083, 'samples': 4305984, 'steps': 22426, 'loss/train': 1.3632395267486572}
+01/29/2022 17:44:58 - INFO - codeparrot_training - Step 22427: {'lr': 0.0003079316701903897, 'samples': 4306176, 'steps': 22427, 'loss/train': 0.3846439719200134}
+01/29/2022 17:45:02 - INFO - codeparrot_training - Step 22428: {'lr': 0.0003079157529755102, 'samples': 4306368, 'steps': 22428, 'loss/train': 1.976223886013031}
+01/29/2022 17:45:06 - INFO - codeparrot_training - Step 22429: {'lr': 0.0003078998355125381, 'samples': 4306560, 'steps': 22429, 'loss/train': 1.2478366792201996}
+01/29/2022 17:45:10 - INFO - codeparrot_training - Step 22430: {'lr': 0.0003078839178015414, 'samples': 4306752, 'steps': 22430, 'loss/train': 0.7927024662494659}
+01/29/2022 17:45:16 - INFO - codeparrot_training - Step 22431: {'lr': 0.0003078679998425884, 'samples': 4306944, 'steps': 22431, 'loss/train': 0.7134803831577301}
+01/29/2022 17:45:20 - INFO - codeparrot_training - Step 22432: {'lr': 0.0003078520816357472, 'samples': 4307136, 'steps': 22432, 'loss/train': 0.6326070874929428}
+01/29/2022 17:45:25 - INFO - codeparrot_training - Step 22433: {'lr': 0.00030783616318108613, 'samples': 4307328, 'steps': 22433, 'loss/train': 0.40483640134334564}
+01/29/2022 17:45:29 - INFO - codeparrot_training - Step 22434: {'lr': 0.0003078202444786733, 'samples': 4307520, 'steps': 22434, 'loss/train': 1.287265419960022}
+01/29/2022 17:45:33 - INFO - codeparrot_training - Step 22435: {'lr': 0.0003078043255285769, 'samples': 4307712, 'steps': 22435, 'loss/train': 2.067869782447815}
+01/29/2022 17:45:40 - INFO - codeparrot_training - Step 22436: {'lr': 0.00030778840633086514, 'samples': 4307904, 'steps': 22436, 'loss/train': 1.4772403836250305}
+01/29/2022 17:45:44 - INFO - codeparrot_training - Step 22437: {'lr': 0.00030777248688560615, 'samples': 4308096, 'steps': 22437, 'loss/train': 1.7748751044273376}
+01/29/2022 17:45:49 - INFO - codeparrot_training - Step 22438: {'lr': 0.0003077565671928682, 'samples': 4308288, 'steps': 22438, 'loss/train': 2.161752998828888}
+01/29/2022 17:45:53 - INFO - codeparrot_training - Step 22439: {'lr': 0.00030774064725271944, 'samples': 4308480, 'steps': 22439, 'loss/train': 1.4943267703056335}
+01/29/2022 17:45:57 - INFO - codeparrot_training - Step 22440: {'lr': 0.00030772472706522806, 'samples': 4308672, 'steps': 22440, 'loss/train': 1.1109226047992706}
+01/29/2022 17:46:02 - INFO - codeparrot_training - Step 22441: {'lr': 0.00030770880663046236, 'samples': 4308864, 'steps': 22441, 'loss/train': 0.37880176305770874}
+01/29/2022 17:46:07 - INFO - codeparrot_training - Step 22442: {'lr': 0.00030769288594849044, 'samples': 4309056, 'steps': 22442, 'loss/train': 1.882781982421875}
+01/29/2022 17:46:11 - INFO - codeparrot_training - Step 22443: {'lr': 0.0003076769650193805, 'samples': 4309248, 'steps': 22443, 'loss/train': 1.352865993976593}
+01/29/2022 17:46:15 - INFO - codeparrot_training - Step 22444: {'lr': 0.0003076610438432007, 'samples': 4309440, 'steps': 22444, 'loss/train': 2.614687979221344}
+01/29/2022 17:46:19 - INFO - codeparrot_training - Step 22445: {'lr': 0.00030764512242001927, 'samples': 4309632, 'steps': 22445, 'loss/train': 1.2921480238437653}
+01/29/2022 17:46:25 - INFO - codeparrot_training - Step 22446: {'lr': 0.00030762920074990457, 'samples': 4309824, 'steps': 22446, 'loss/train': 1.4249204099178314}
+01/29/2022 17:46:29 - INFO - codeparrot_training - Step 22447: {'lr': 0.00030761327883292456, 'samples': 4310016, 'steps': 22447, 'loss/train': 2.6037368774414062}
+01/29/2022 17:46:33 - INFO - codeparrot_training - Step 22448: {'lr': 0.00030759735666914767, 'samples': 4310208, 'steps': 22448, 'loss/train': 1.8982315063476562}
+01/29/2022 17:46:37 - INFO - codeparrot_training - Step 22449: {'lr': 0.00030758143425864187, 'samples': 4310400, 'steps': 22449, 'loss/train': 1.5323554873466492}
+01/29/2022 17:46:42 - INFO - codeparrot_training - Step 22450: {'lr': 0.00030756551160147563, 'samples': 4310592, 'steps': 22450, 'loss/train': 1.5064709186553955}
+01/29/2022 17:46:47 - INFO - codeparrot_training - Step 22451: {'lr': 0.0003075495886977169, 'samples': 4310784, 'steps': 22451, 'loss/train': 1.736323893070221}
+01/29/2022 17:46:51 - INFO - codeparrot_training - Step 22452: {'lr': 0.0003075336655474341, 'samples': 4310976, 'steps': 22452, 'loss/train': 1.7470561265945435}
+01/29/2022 17:46:55 - INFO - codeparrot_training - Step 22453: {'lr': 0.0003075177421506952, 'samples': 4311168, 'steps': 22453, 'loss/train': 2.2157617807388306}
+01/29/2022 17:47:00 - INFO - codeparrot_training - Step 22454: {'lr': 0.0003075018185075687, 'samples': 4311360, 'steps': 22454, 'loss/train': 1.2430177330970764}
+01/29/2022 17:47:04 - INFO - codeparrot_training - Step 22455: {'lr': 0.0003074858946181226, 'samples': 4311552, 'steps': 22455, 'loss/train': 0.9690233767032623}
+01/29/2022 17:47:11 - INFO - codeparrot_training - Step 22456: {'lr': 0.0003074699704824252, 'samples': 4311744, 'steps': 22456, 'loss/train': 1.3582372069358826}
+01/29/2022 17:47:15 - INFO - codeparrot_training - Step 22457: {'lr': 0.0003074540461005447, 'samples': 4311936, 'steps': 22457, 'loss/train': 1.849898099899292}
+01/29/2022 17:47:19 - INFO - codeparrot_training - Step 22458: {'lr': 0.00030743812147254935, 'samples': 4312128, 'steps': 22458, 'loss/train': 1.0980616807937622}
+01/29/2022 17:47:24 - INFO - codeparrot_training - Step 22459: {'lr': 0.0003074221965985073, 'samples': 4312320, 'steps': 22459, 'loss/train': 0.9441007375717163}
+01/29/2022 17:47:28 - INFO - codeparrot_training - Step 22460: {'lr': 0.0003074062714784867, 'samples': 4312512, 'steps': 22460, 'loss/train': 2.294106602668762}
+01/29/2022 17:47:33 - INFO - codeparrot_training - Step 22461: {'lr': 0.000307390346112556, 'samples': 4312704, 'steps': 22461, 'loss/train': 1.6134881973266602}
+01/29/2022 17:47:37 - INFO - codeparrot_training - Step 22462: {'lr': 0.0003073744205007832, 'samples': 4312896, 'steps': 22462, 'loss/train': 1.411789745092392}
+01/29/2022 17:47:42 - INFO - codeparrot_training - Step 22463: {'lr': 0.0003073584946432366, 'samples': 4313088, 'steps': 22463, 'loss/train': 0.2625993937253952}
+01/29/2022 17:47:46 - INFO - codeparrot_training - Step 22464: {'lr': 0.00030734256853998446, 'samples': 4313280, 'steps': 22464, 'loss/train': 1.9448096752166748}
+01/29/2022 17:47:53 - INFO - codeparrot_training - Step 22465: {'lr': 0.00030732664219109497, 'samples': 4313472, 'steps': 22465, 'loss/train': 0.5085471868515015}
+01/29/2022 17:47:57 - INFO - codeparrot_training - Step 22466: {'lr': 0.00030731071559663624, 'samples': 4313664, 'steps': 22466, 'loss/train': 1.8177709579467773}
+01/29/2022 17:48:01 - INFO - codeparrot_training - Step 22467: {'lr': 0.0003072947887566768, 'samples': 4313856, 'steps': 22467, 'loss/train': 1.1331604421138763}
+01/29/2022 17:48:06 - INFO - codeparrot_training - Step 22468: {'lr': 0.0003072788616712845, 'samples': 4314048, 'steps': 22468, 'loss/train': 2.07930064201355}
+01/29/2022 17:48:10 - INFO - codeparrot_training - Step 22469: {'lr': 0.0003072629343405278, 'samples': 4314240, 'steps': 22469, 'loss/train': 1.8916436433792114}
+01/29/2022 17:48:15 - INFO - codeparrot_training - Step 22470: {'lr': 0.00030724700676447485, 'samples': 4314432, 'steps': 22470, 'loss/train': 1.174730658531189}
+01/29/2022 17:48:19 - INFO - codeparrot_training - Step 22471: {'lr': 0.00030723107894319393, 'samples': 4314624, 'steps': 22471, 'loss/train': 1.998643934726715}
+01/29/2022 17:48:24 - INFO - codeparrot_training - Step 22472: {'lr': 0.00030721515087675326, 'samples': 4314816, 'steps': 22472, 'loss/train': 1.112708330154419}
+01/29/2022 17:48:28 - INFO - codeparrot_training - Step 22473: {'lr': 0.00030719922256522105, 'samples': 4315008, 'steps': 22473, 'loss/train': 1.307160347700119}
+01/29/2022 17:48:32 - INFO - codeparrot_training - Step 22474: {'lr': 0.0003071832940086655, 'samples': 4315200, 'steps': 22474, 'loss/train': 0.8839629292488098}
+01/29/2022 17:48:38 - INFO - codeparrot_training - Step 22475: {'lr': 0.0003071673652071549, 'samples': 4315392, 'steps': 22475, 'loss/train': 1.9242262244224548}
+01/29/2022 17:48:42 - INFO - codeparrot_training - Step 22476: {'lr': 0.0003071514361607575, 'samples': 4315584, 'steps': 22476, 'loss/train': 1.864431917667389}
+01/29/2022 17:48:46 - INFO - codeparrot_training - Step 22477: {'lr': 0.0003071355068695414, 'samples': 4315776, 'steps': 22477, 'loss/train': 0.9099798202514648}
+01/29/2022 17:48:50 - INFO - codeparrot_training - Step 22478: {'lr': 0.000307119577333575, 'samples': 4315968, 'steps': 22478, 'loss/train': 1.9412607550621033}
+01/29/2022 17:48:55 - INFO - codeparrot_training - Step 22479: {'lr': 0.0003071036475529264, 'samples': 4316160, 'steps': 22479, 'loss/train': 1.875309705734253}
+01/29/2022 17:49:02 - INFO - codeparrot_training - Step 22480: {'lr': 0.00030708771752766396, 'samples': 4316352, 'steps': 22480, 'loss/train': 1.2843638956546783}
+01/29/2022 17:49:06 - INFO - codeparrot_training - Step 22481: {'lr': 0.0003070717872578558, 'samples': 4316544, 'steps': 22481, 'loss/train': 1.0168274939060211}
+01/29/2022 17:49:10 - INFO - codeparrot_training - Step 22482: {'lr': 0.0003070558567435703, 'samples': 4316736, 'steps': 22482, 'loss/train': 1.393455058336258}
+01/29/2022 17:49:14 - INFO - codeparrot_training - Step 22483: {'lr': 0.00030703992598487564, 'samples': 4316928, 'steps': 22483, 'loss/train': 1.6296265125274658}
+01/29/2022 17:49:19 - INFO - codeparrot_training - Step 22484: {'lr': 0.00030702399498184005, 'samples': 4317120, 'steps': 22484, 'loss/train': 1.5735390186309814}
+01/29/2022 17:49:25 - INFO - codeparrot_training - Step 22485: {'lr': 0.0003070080637345317, 'samples': 4317312, 'steps': 22485, 'loss/train': 1.9801311492919922}
+01/29/2022 17:49:29 - INFO - codeparrot_training - Step 22486: {'lr': 0.00030699213224301896, 'samples': 4317504, 'steps': 22486, 'loss/train': 1.7641382217407227}
+01/29/2022 17:49:33 - INFO - codeparrot_training - Step 22487: {'lr': 0.00030697620050737, 'samples': 4317696, 'steps': 22487, 'loss/train': 2.240931451320648}
+01/29/2022 17:49:37 - INFO - codeparrot_training - Step 22488: {'lr': 0.00030696026852765313, 'samples': 4317888, 'steps': 22488, 'loss/train': 1.8874803185462952}
+01/29/2022 17:49:41 - INFO - codeparrot_training - Step 22489: {'lr': 0.00030694433630393654, 'samples': 4318080, 'steps': 22489, 'loss/train': 0.9857945144176483}
+01/29/2022 17:49:46 - INFO - codeparrot_training - Step 22490: {'lr': 0.00030692840383628845, 'samples': 4318272, 'steps': 22490, 'loss/train': 0.9573308229446411}
+01/29/2022 17:49:51 - INFO - codeparrot_training - Step 22491: {'lr': 0.0003069124711247772, 'samples': 4318464, 'steps': 22491, 'loss/train': 0.8752251863479614}
+01/29/2022 17:49:55 - INFO - codeparrot_training - Step 22492: {'lr': 0.000306896538169471, 'samples': 4318656, 'steps': 22492, 'loss/train': 2.9846282601356506}
+01/29/2022 17:49:59 - INFO - codeparrot_training - Step 22493: {'lr': 0.000306880604970438, 'samples': 4318848, 'steps': 22493, 'loss/train': 1.91996169090271}
+01/29/2022 17:50:04 - INFO - codeparrot_training - Step 22494: {'lr': 0.00030686467152774667, 'samples': 4319040, 'steps': 22494, 'loss/train': 1.4999444782733917}
+01/29/2022 17:50:08 - INFO - codeparrot_training - Step 22495: {'lr': 0.0003068487378414651, 'samples': 4319232, 'steps': 22495, 'loss/train': 1.6021631956100464}
+01/29/2022 17:50:15 - INFO - codeparrot_training - Step 22496: {'lr': 0.0003068328039116616, 'samples': 4319424, 'steps': 22496, 'loss/train': 1.8943807482719421}
+01/29/2022 17:50:19 - INFO - codeparrot_training - Step 22497: {'lr': 0.0003068168697384044, 'samples': 4319616, 'steps': 22497, 'loss/train': 2.3626447319984436}
+01/29/2022 17:50:24 - INFO - codeparrot_training - Step 22498: {'lr': 0.0003068009353217618, 'samples': 4319808, 'steps': 22498, 'loss/train': 1.657164216041565}
+01/29/2022 17:50:28 - INFO - codeparrot_training - Step 22499: {'lr': 0.00030678500066180206, 'samples': 4320000, 'steps': 22499, 'loss/train': 1.6912046670913696}
+01/29/2022 17:50:32 - INFO - codeparrot_training - Step 22500: {'lr': 0.0003067690657585933, 'samples': 4320192, 'steps': 22500, 'loss/train': 1.3313153386116028}
+01/29/2022 17:50:37 - INFO - codeparrot_training - Step 22501: {'lr': 0.000306753130612204, 'samples': 4320384, 'steps': 22501, 'loss/train': 1.508746325969696}
+01/29/2022 17:50:42 - INFO - codeparrot_training - Step 22502: {'lr': 0.0003067371952227022, 'samples': 4320576, 'steps': 22502, 'loss/train': 1.8611767888069153}
+01/29/2022 17:50:46 - INFO - codeparrot_training - Step 22503: {'lr': 0.0003067212595901564, 'samples': 4320768, 'steps': 22503, 'loss/train': 1.7141515016555786}
+01/29/2022 17:50:50 - INFO - codeparrot_training - Step 22504: {'lr': 0.00030670532371463463, 'samples': 4320960, 'steps': 22504, 'loss/train': 0.9104413390159607}
+01/29/2022 17:50:54 - INFO - codeparrot_training - Step 22505: {'lr': 0.0003066893875962053, 'samples': 4321152, 'steps': 22505, 'loss/train': 2.8251532316207886}
+01/29/2022 17:51:00 - INFO - codeparrot_training - Step 22506: {'lr': 0.0003066734512349366, 'samples': 4321344, 'steps': 22506, 'loss/train': 1.5102195739746094}
+01/29/2022 17:51:04 - INFO - codeparrot_training - Step 22507: {'lr': 0.00030665751463089687, 'samples': 4321536, 'steps': 22507, 'loss/train': 1.6490201354026794}
+01/29/2022 17:51:08 - INFO - codeparrot_training - Step 22508: {'lr': 0.0003066415777841543, 'samples': 4321728, 'steps': 22508, 'loss/train': 1.6245834231376648}
+01/29/2022 17:51:12 - INFO - codeparrot_training - Step 22509: {'lr': 0.0003066256406947773, 'samples': 4321920, 'steps': 22509, 'loss/train': 1.3397253155708313}
+01/29/2022 17:51:17 - INFO - codeparrot_training - Step 22510: {'lr': 0.0003066097033628339, 'samples': 4322112, 'steps': 22510, 'loss/train': 2.0956549644470215}
+01/29/2022 17:51:22 - INFO - codeparrot_training - Step 22511: {'lr': 0.0003065937657883926, 'samples': 4322304, 'steps': 22511, 'loss/train': 1.8026633262634277}
+01/29/2022 17:51:26 - INFO - codeparrot_training - Step 22512: {'lr': 0.0003065778279715215, 'samples': 4322496, 'steps': 22512, 'loss/train': 1.776664674282074}
+01/29/2022 17:51:31 - INFO - codeparrot_training - Step 22513: {'lr': 0.000306561889912289, 'samples': 4322688, 'steps': 22513, 'loss/train': 1.276796668767929}
+01/29/2022 17:51:35 - INFO - codeparrot_training - Step 22514: {'lr': 0.0003065459516107633, 'samples': 4322880, 'steps': 22514, 'loss/train': 1.8475749492645264}
+01/29/2022 17:51:39 - INFO - codeparrot_training - Step 22515: {'lr': 0.0003065300130670127, 'samples': 4323072, 'steps': 22515, 'loss/train': 0.7460356950759888}
+01/29/2022 17:51:46 - INFO - codeparrot_training - Step 22516: {'lr': 0.0003065140742811055, 'samples': 4323264, 'steps': 22516, 'loss/train': 1.7624480724334717}
+01/29/2022 17:51:51 - INFO - codeparrot_training - Step 22517: {'lr': 0.0003064981352531099, 'samples': 4323456, 'steps': 22517, 'loss/train': 1.368649274110794}
+01/29/2022 17:51:55 - INFO - codeparrot_training - Step 22518: {'lr': 0.0003064821959830943, 'samples': 4323648, 'steps': 22518, 'loss/train': 0.38379281759262085}
+01/29/2022 17:51:59 - INFO - codeparrot_training - Step 22519: {'lr': 0.00030646625647112686, 'samples': 4323840, 'steps': 22519, 'loss/train': 0.8543847799301147}
+01/29/2022 17:52:03 - INFO - codeparrot_training - Step 22520: {'lr': 0.000306450316717276, 'samples': 4324032, 'steps': 22520, 'loss/train': 1.516962468624115}
+01/29/2022 17:52:09 - INFO - codeparrot_training - Step 22521: {'lr': 0.0003064343767216098, 'samples': 4324224, 'steps': 22521, 'loss/train': 1.367642194032669}
+01/29/2022 17:52:13 - INFO - codeparrot_training - Step 22522: {'lr': 0.00030641843648419664, 'samples': 4324416, 'steps': 22522, 'loss/train': 1.8038250803947449}
+01/29/2022 17:52:17 - INFO - codeparrot_training - Step 22523: {'lr': 0.0003064024960051048, 'samples': 4324608, 'steps': 22523, 'loss/train': 1.4777025282382965}
+01/29/2022 17:52:22 - INFO - codeparrot_training - Step 22524: {'lr': 0.00030638655528440273, 'samples': 4324800, 'steps': 22524, 'loss/train': 1.5362305641174316}
+01/29/2022 17:52:26 - INFO - codeparrot_training - Step 22525: {'lr': 0.0003063706143221584, 'samples': 4324992, 'steps': 22525, 'loss/train': 0.6055576354265213}
+01/29/2022 17:52:33 - INFO - codeparrot_training - Step 22526: {'lr': 0.00030635467311844033, 'samples': 4325184, 'steps': 22526, 'loss/train': 1.963007390499115}
+01/29/2022 17:52:37 - INFO - codeparrot_training - Step 22527: {'lr': 0.00030633873167331674, 'samples': 4325376, 'steps': 22527, 'loss/train': 1.479538768529892}
+01/29/2022 17:52:41 - INFO - codeparrot_training - Step 22528: {'lr': 0.0003063227899868559, 'samples': 4325568, 'steps': 22528, 'loss/train': 1.7092180252075195}
+01/29/2022 17:52:46 - INFO - codeparrot_training - Step 22529: {'lr': 0.00030630684805912613, 'samples': 4325760, 'steps': 22529, 'loss/train': 1.3496848046779633}
+01/29/2022 17:52:50 - INFO - codeparrot_training - Step 22530: {'lr': 0.00030629090589019567, 'samples': 4325952, 'steps': 22530, 'loss/train': 2.225062608718872}
+01/29/2022 17:52:55 - INFO - codeparrot_training - Step 22531: {'lr': 0.00030627496348013285, 'samples': 4326144, 'steps': 22531, 'loss/train': 2.9048418402671814}
+01/29/2022 17:53:00 - INFO - codeparrot_training - Step 22532: {'lr': 0.0003062590208290059, 'samples': 4326336, 'steps': 22532, 'loss/train': 1.9783269166946411}
+01/29/2022 17:53:04 - INFO - codeparrot_training - Step 22533: {'lr': 0.00030624307793688334, 'samples': 4326528, 'steps': 22533, 'loss/train': 1.9878484010696411}
+01/29/2022 17:53:08 - INFO - codeparrot_training - Step 22534: {'lr': 0.00030622713480383314, 'samples': 4326720, 'steps': 22534, 'loss/train': 1.2239999771118164}
+01/29/2022 17:53:12 - INFO - codeparrot_training - Step 22535: {'lr': 0.0003062111914299238, 'samples': 4326912, 'steps': 22535, 'loss/train': 1.7796590924263}
+01/29/2022 17:53:18 - INFO - codeparrot_training - Step 22536: {'lr': 0.0003061952478152236, 'samples': 4327104, 'steps': 22536, 'loss/train': 1.6204179525375366}
+01/29/2022 17:53:22 - INFO - codeparrot_training - Step 22537: {'lr': 0.0003061793039598007, 'samples': 4327296, 'steps': 22537, 'loss/train': 2.0070927143096924}
+01/29/2022 17:53:26 - INFO - codeparrot_training - Step 22538: {'lr': 0.0003061633598637236, 'samples': 4327488, 'steps': 22538, 'loss/train': 0.029809992760419846}
+01/29/2022 17:53:30 - INFO - codeparrot_training - Step 22539: {'lr': 0.00030614741552706045, 'samples': 4327680, 'steps': 22539, 'loss/train': 1.4179799258708954}
+01/29/2022 17:53:35 - INFO - codeparrot_training - Step 22540: {'lr': 0.0003061314709498796, 'samples': 4327872, 'steps': 22540, 'loss/train': 1.3924202620983124}
+01/29/2022 17:53:41 - INFO - codeparrot_training - Step 22541: {'lr': 0.00030611552613224935, 'samples': 4328064, 'steps': 22541, 'loss/train': 1.947918713092804}
+01/29/2022 17:53:45 - INFO - codeparrot_training - Step 22542: {'lr': 0.00030609958107423804, 'samples': 4328256, 'steps': 22542, 'loss/train': 1.474412888288498}
+01/29/2022 17:53:49 - INFO - codeparrot_training - Step 22543: {'lr': 0.0003060836357759139, 'samples': 4328448, 'steps': 22543, 'loss/train': 1.2721078097820282}
+01/29/2022 17:53:53 - INFO - codeparrot_training - Step 22544: {'lr': 0.00030606769023734534, 'samples': 4328640, 'steps': 22544, 'loss/train': 1.742057740688324}
+01/29/2022 17:53:58 - INFO - codeparrot_training - Step 22545: {'lr': 0.0003060517444586005, 'samples': 4328832, 'steps': 22545, 'loss/train': 1.5813894867897034}
+01/29/2022 17:54:05 - INFO - codeparrot_training - Step 22546: {'lr': 0.0003060357984397479, 'samples': 4329024, 'steps': 22546, 'loss/train': 1.7311390042304993}
+01/29/2022 17:54:09 - INFO - codeparrot_training - Step 22547: {'lr': 0.00030601985218085565, 'samples': 4329216, 'steps': 22547, 'loss/train': 1.296610951423645}
+01/29/2022 17:54:14 - INFO - codeparrot_training - Step 22548: {'lr': 0.00030600390568199213, 'samples': 4329408, 'steps': 22548, 'loss/train': 0.9012918770313263}
+01/29/2022 17:54:18 - INFO - codeparrot_training - Step 22549: {'lr': 0.00030598795894322574, 'samples': 4329600, 'steps': 22549, 'loss/train': 1.5588784217834473}
+01/29/2022 17:54:22 - INFO - codeparrot_training - Step 22550: {'lr': 0.00030597201196462466, 'samples': 4329792, 'steps': 22550, 'loss/train': 1.9171850681304932}
+01/29/2022 17:54:27 - INFO - codeparrot_training - Step 22551: {'lr': 0.00030595606474625726, 'samples': 4329984, 'steps': 22551, 'loss/train': 2.574224531650543}
+01/29/2022 17:54:32 - INFO - codeparrot_training - Step 22552: {'lr': 0.00030594011728819184, 'samples': 4330176, 'steps': 22552, 'loss/train': 2.074785053730011}
+01/29/2022 17:54:36 - INFO - codeparrot_training - Step 22553: {'lr': 0.00030592416959049666, 'samples': 4330368, 'steps': 22553, 'loss/train': 1.969417691230774}
+01/29/2022 17:54:40 - INFO - codeparrot_training - Step 22554: {'lr': 0.00030590822165324017, 'samples': 4330560, 'steps': 22554, 'loss/train': 1.4495153725147247}
+01/29/2022 17:54:44 - INFO - codeparrot_training - Step 22555: {'lr': 0.00030589227347649063, 'samples': 4330752, 'steps': 22555, 'loss/train': 2.2211485505104065}
+01/29/2022 17:54:51 - INFO - codeparrot_training - Step 22556: {'lr': 0.00030587632506031624, 'samples': 4330944, 'steps': 22556, 'loss/train': 2.513738751411438}
+01/29/2022 17:54:55 - INFO - codeparrot_training - Step 22557: {'lr': 0.0003058603764047855, 'samples': 4331136, 'steps': 22557, 'loss/train': 2.0188516974449158}
+01/29/2022 17:55:00 - INFO - codeparrot_training - Step 22558: {'lr': 0.00030584442750996666, 'samples': 4331328, 'steps': 22558, 'loss/train': 2.053162693977356}
+01/29/2022 17:55:04 - INFO - codeparrot_training - Step 22559: {'lr': 0.000305828478375928, 'samples': 4331520, 'steps': 22559, 'loss/train': 1.8646305799484253}
+01/29/2022 17:55:08 - INFO - codeparrot_training - Step 22560: {'lr': 0.00030581252900273786, 'samples': 4331712, 'steps': 22560, 'loss/train': 1.633348524570465}
+01/29/2022 17:55:14 - INFO - codeparrot_training - Step 22561: {'lr': 0.0003057965793904646, 'samples': 4331904, 'steps': 22561, 'loss/train': 1.4207072854042053}
+01/29/2022 17:55:19 - INFO - codeparrot_training - Step 22562: {'lr': 0.00030578062953917645, 'samples': 4332096, 'steps': 22562, 'loss/train': 1.6239697337150574}
+01/29/2022 17:55:23 - INFO - codeparrot_training - Step 22563: {'lr': 0.00030576467944894186, 'samples': 4332288, 'steps': 22563, 'loss/train': 1.4147182703018188}
+01/29/2022 17:55:27 - INFO - codeparrot_training - Step 22564: {'lr': 0.000305748729119829, 'samples': 4332480, 'steps': 22564, 'loss/train': 1.7530642747879028}
+01/29/2022 17:55:31 - INFO - codeparrot_training - Step 22565: {'lr': 0.00030573277855190634, 'samples': 4332672, 'steps': 22565, 'loss/train': 0.8965004682540894}
+01/29/2022 17:55:36 - INFO - codeparrot_training - Step 22566: {'lr': 0.0003057168277452422, 'samples': 4332864, 'steps': 22566, 'loss/train': 1.691552460193634}
+01/29/2022 17:55:41 - INFO - codeparrot_training - Step 22567: {'lr': 0.0003057008766999048, 'samples': 4333056, 'steps': 22567, 'loss/train': 2.2041388750076294}
+01/29/2022 17:55:45 - INFO - codeparrot_training - Step 22568: {'lr': 0.0003056849254159625, 'samples': 4333248, 'steps': 22568, 'loss/train': 1.3536492884159088}
+01/29/2022 17:55:50 - INFO - codeparrot_training - Step 22569: {'lr': 0.00030566897389348375, 'samples': 4333440, 'steps': 22569, 'loss/train': 1.8993836045265198}
+01/29/2022 17:55:54 - INFO - codeparrot_training - Step 22570: {'lr': 0.00030565302213253677, 'samples': 4333632, 'steps': 22570, 'loss/train': 1.4931594729423523}
+01/29/2022 17:56:01 - INFO - codeparrot_training - Step 22571: {'lr': 0.0003056370701331899, 'samples': 4333824, 'steps': 22571, 'loss/train': 2.6585533618927}
+01/29/2022 17:56:05 - INFO - codeparrot_training - Step 22572: {'lr': 0.0003056211178955115, 'samples': 4334016, 'steps': 22572, 'loss/train': 0.9932528436183929}
+01/29/2022 17:56:09 - INFO - codeparrot_training - Step 22573: {'lr': 0.00030560516541956983, 'samples': 4334208, 'steps': 22573, 'loss/train': 0.7322928607463837}
+01/29/2022 17:56:14 - INFO - codeparrot_training - Step 22574: {'lr': 0.0003055892127054334, 'samples': 4334400, 'steps': 22574, 'loss/train': 1.8491452932357788}
+01/29/2022 17:56:18 - INFO - codeparrot_training - Step 22575: {'lr': 0.00030557325975317037, 'samples': 4334592, 'steps': 22575, 'loss/train': 1.4021621346473694}
+01/29/2022 17:56:23 - INFO - codeparrot_training - Step 22576: {'lr': 0.00030555730656284914, 'samples': 4334784, 'steps': 22576, 'loss/train': 1.2667794227600098}
+01/29/2022 17:56:28 - INFO - codeparrot_training - Step 22577: {'lr': 0.000305541353134538, 'samples': 4334976, 'steps': 22577, 'loss/train': 1.5846553444862366}
+01/29/2022 17:56:32 - INFO - codeparrot_training - Step 22578: {'lr': 0.0003055253994683054, 'samples': 4335168, 'steps': 22578, 'loss/train': 1.8019521832466125}
+01/29/2022 17:56:36 - INFO - codeparrot_training - Step 22579: {'lr': 0.0003055094455642196, 'samples': 4335360, 'steps': 22579, 'loss/train': 1.7652212977409363}
+01/29/2022 17:56:40 - INFO - codeparrot_training - Step 22580: {'lr': 0.000305493491422349, 'samples': 4335552, 'steps': 22580, 'loss/train': 0.8503713011741638}
+01/29/2022 17:56:46 - INFO - codeparrot_training - Step 22581: {'lr': 0.0003054775370427619, 'samples': 4335744, 'steps': 22581, 'loss/train': 1.9921667575836182}
+01/29/2022 17:56:50 - INFO - codeparrot_training - Step 22582: {'lr': 0.00030546158242552657, 'samples': 4335936, 'steps': 22582, 'loss/train': 1.3752292692661285}
+01/29/2022 17:56:54 - INFO - codeparrot_training - Step 22583: {'lr': 0.00030544562757071154, 'samples': 4336128, 'steps': 22583, 'loss/train': 1.6832209825515747}
+01/29/2022 17:56:58 - INFO - codeparrot_training - Step 22584: {'lr': 0.000305429672478385, 'samples': 4336320, 'steps': 22584, 'loss/train': 2.3017698526382446}
+01/29/2022 17:57:02 - INFO - codeparrot_training - Step 22585: {'lr': 0.0003054137171486153, 'samples': 4336512, 'steps': 22585, 'loss/train': 1.1132575571537018}
+01/29/2022 17:57:10 - INFO - codeparrot_training - Step 22586: {'lr': 0.0003053977615814709, 'samples': 4336704, 'steps': 22586, 'loss/train': 1.9218301177024841}
+01/29/2022 17:57:14 - INFO - codeparrot_training - Step 22587: {'lr': 0.00030538180577702005, 'samples': 4336896, 'steps': 22587, 'loss/train': 2.005050837993622}
+01/29/2022 17:57:18 - INFO - codeparrot_training - Step 22588: {'lr': 0.00030536584973533113, 'samples': 4337088, 'steps': 22588, 'loss/train': 1.9984524250030518}
+01/29/2022 17:57:22 - INFO - codeparrot_training - Step 22589: {'lr': 0.0003053498934564725, 'samples': 4337280, 'steps': 22589, 'loss/train': 1.2944839596748352}
+01/29/2022 17:57:26 - INFO - codeparrot_training - Step 22590: {'lr': 0.00030533393694051256, 'samples': 4337472, 'steps': 22590, 'loss/train': 1.8231110572814941}
+01/29/2022 17:57:32 - INFO - codeparrot_training - Step 22591: {'lr': 0.0003053179801875195, 'samples': 4337664, 'steps': 22591, 'loss/train': 0.9996085166931152}
+01/29/2022 17:57:36 - INFO - codeparrot_training - Step 22592: {'lr': 0.00030530202319756184, 'samples': 4337856, 'steps': 22592, 'loss/train': 1.9952545166015625}
+01/29/2022 17:57:40 - INFO - codeparrot_training - Step 22593: {'lr': 0.0003052860659707078, 'samples': 4338048, 'steps': 22593, 'loss/train': 1.950823724269867}
+01/29/2022 17:57:44 - INFO - codeparrot_training - Step 22594: {'lr': 0.0003052701085070259, 'samples': 4338240, 'steps': 22594, 'loss/train': 1.3987770080566406}
+01/29/2022 17:57:49 - INFO - codeparrot_training - Step 22595: {'lr': 0.00030525415080658437, 'samples': 4338432, 'steps': 22595, 'loss/train': 2.1927865147590637}
+01/29/2022 17:57:54 - INFO - codeparrot_training - Step 22596: {'lr': 0.0003052381928694516, 'samples': 4338624, 'steps': 22596, 'loss/train': 1.451328992843628}
+01/29/2022 17:57:58 - INFO - codeparrot_training - Step 22597: {'lr': 0.00030522223469569594, 'samples': 4338816, 'steps': 22597, 'loss/train': 1.8796323537826538}
+01/29/2022 17:58:02 - INFO - codeparrot_training - Step 22598: {'lr': 0.00030520627628538577, 'samples': 4339008, 'steps': 22598, 'loss/train': 2.1469831466674805}
+01/29/2022 17:58:07 - INFO - codeparrot_training - Step 22599: {'lr': 0.0003051903176385894, 'samples': 4339200, 'steps': 22599, 'loss/train': 1.4557562470436096}
+01/29/2022 17:58:11 - INFO - codeparrot_training - Step 22600: {'lr': 0.00030517435875537536, 'samples': 4339392, 'steps': 22600, 'loss/train': 1.2621381282806396}
+01/29/2022 17:58:18 - INFO - codeparrot_training - Step 22601: {'lr': 0.0003051583996358118, 'samples': 4339584, 'steps': 22601, 'loss/train': 1.9712743163108826}
+01/29/2022 17:58:22 - INFO - codeparrot_training - Step 22602: {'lr': 0.00030514244027996705, 'samples': 4339776, 'steps': 22602, 'loss/train': 1.5330138802528381}
+01/29/2022 17:58:26 - INFO - codeparrot_training - Step 22603: {'lr': 0.00030512648068790985, 'samples': 4339968, 'steps': 22603, 'loss/train': 1.8916125297546387}
+01/29/2022 17:58:30 - INFO - codeparrot_training - Step 22604: {'lr': 0.0003051105208597081, 'samples': 4340160, 'steps': 22604, 'loss/train': 1.6457176208496094}
+01/29/2022 17:58:35 - INFO - codeparrot_training - Step 22605: {'lr': 0.00030509456079543044, 'samples': 4340352, 'steps': 22605, 'loss/train': 1.422050803899765}
+01/29/2022 17:58:40 - INFO - codeparrot_training - Step 22606: {'lr': 0.0003050786004951452, 'samples': 4340544, 'steps': 22606, 'loss/train': 0.3679649233818054}
+01/29/2022 17:58:44 - INFO - codeparrot_training - Step 22607: {'lr': 0.00030506263995892075, 'samples': 4340736, 'steps': 22607, 'loss/train': 2.257416844367981}
+01/29/2022 17:58:48 - INFO - codeparrot_training - Step 22608: {'lr': 0.0003050466791868254, 'samples': 4340928, 'steps': 22608, 'loss/train': 1.317636787891388}
+01/29/2022 17:58:53 - INFO - codeparrot_training - Step 22609: {'lr': 0.0003050307181789276, 'samples': 4341120, 'steps': 22609, 'loss/train': 2.2111347913742065}
+01/29/2022 17:58:57 - INFO - codeparrot_training - Step 22610: {'lr': 0.0003050147569352956, 'samples': 4341312, 'steps': 22610, 'loss/train': 2.0741084218025208}
+01/29/2022 17:59:02 - INFO - codeparrot_training - Step 22611: {'lr': 0.0003049987954559979, 'samples': 4341504, 'steps': 22611, 'loss/train': 1.4484303295612335}
+01/29/2022 17:59:06 - INFO - codeparrot_training - Step 22612: {'lr': 0.0003049828337411028, 'samples': 4341696, 'steps': 22612, 'loss/train': 2.2598671317100525}
+01/29/2022 17:59:10 - INFO - codeparrot_training - Step 22613: {'lr': 0.00030496687179067865, 'samples': 4341888, 'steps': 22613, 'loss/train': 1.9846315383911133}
+01/29/2022 17:59:15 - INFO - codeparrot_training - Step 22614: {'lr': 0.000304950909604794, 'samples': 4342080, 'steps': 22614, 'loss/train': 1.234490990638733}
+01/29/2022 17:59:19 - INFO - codeparrot_training - Step 22615: {'lr': 0.000304934947183517, 'samples': 4342272, 'steps': 22615, 'loss/train': 1.2129456102848053}
+01/29/2022 17:59:26 - INFO - codeparrot_training - Step 22616: {'lr': 0.00030491898452691626, 'samples': 4342464, 'steps': 22616, 'loss/train': 1.4212449789047241}
+01/29/2022 17:59:30 - INFO - codeparrot_training - Step 22617: {'lr': 0.0003049030216350599, 'samples': 4342656, 'steps': 22617, 'loss/train': 1.9088489413261414}
+01/29/2022 17:59:34 - INFO - codeparrot_training - Step 22618: {'lr': 0.00030488705850801646, 'samples': 4342848, 'steps': 22618, 'loss/train': 2.5563583374023438}
+01/29/2022 17:59:39 - INFO - codeparrot_training - Step 22619: {'lr': 0.00030487109514585426, 'samples': 4343040, 'steps': 22619, 'loss/train': 1.2687016725540161}
+01/29/2022 17:59:43 - INFO - codeparrot_training - Step 22620: {'lr': 0.0003048551315486418, 'samples': 4343232, 'steps': 22620, 'loss/train': 1.4085635840892792}
+01/29/2022 17:59:48 - INFO - codeparrot_training - Step 22621: {'lr': 0.00030483916771644734, 'samples': 4343424, 'steps': 22621, 'loss/train': 1.7604339122772217}
+01/29/2022 17:59:53 - INFO - codeparrot_training - Step 22622: {'lr': 0.0003048232036493392, 'samples': 4343616, 'steps': 22622, 'loss/train': 1.391908586025238}
+01/29/2022 17:59:57 - INFO - codeparrot_training - Step 22623: {'lr': 0.00030480723934738597, 'samples': 4343808, 'steps': 22623, 'loss/train': 2.3497663736343384}
+01/29/2022 18:00:01 - INFO - codeparrot_training - Step 22624: {'lr': 0.00030479127481065595, 'samples': 4344000, 'steps': 22624, 'loss/train': 0.32307160645723343}
+01/29/2022 18:00:05 - INFO - codeparrot_training - Step 22625: {'lr': 0.0003047753100392174, 'samples': 4344192, 'steps': 22625, 'loss/train': 0.5259219110012054}
+01/29/2022 18:00:10 - INFO - codeparrot_training - Step 22626: {'lr': 0.00030475934503313893, 'samples': 4344384, 'steps': 22626, 'loss/train': 2.3010308146476746}
+01/29/2022 18:00:15 - INFO - codeparrot_training - Step 22627: {'lr': 0.0003047433797924888, 'samples': 4344576, 'steps': 22627, 'loss/train': 1.9353511333465576}
+01/29/2022 18:00:19 - INFO - codeparrot_training - Step 22628: {'lr': 0.00030472741431733535, 'samples': 4344768, 'steps': 22628, 'loss/train': 2.5447248816490173}
+01/29/2022 18:00:23 - INFO - codeparrot_training - Step 22629: {'lr': 0.0003047114486077471, 'samples': 4344960, 'steps': 22629, 'loss/train': 2.856400966644287}
+01/29/2022 18:00:27 - INFO - codeparrot_training - Step 22630: {'lr': 0.0003046954826637923, 'samples': 4345152, 'steps': 22630, 'loss/train': 1.3064830005168915}
+01/29/2022 18:00:35 - INFO - codeparrot_training - Step 22631: {'lr': 0.0003046795164855395, 'samples': 4345344, 'steps': 22631, 'loss/train': 0.995220959186554}
+01/29/2022 18:00:39 - INFO - codeparrot_training - Step 22632: {'lr': 0.000304663550073057, 'samples': 4345536, 'steps': 22632, 'loss/train': 1.7524884939193726}
+01/29/2022 18:00:43 - INFO - codeparrot_training - Step 22633: {'lr': 0.00030464758342641315, 'samples': 4345728, 'steps': 22633, 'loss/train': 1.2462055385112762}
+01/29/2022 18:00:47 - INFO - codeparrot_training - Step 22634: {'lr': 0.00030463161654567647, 'samples': 4345920, 'steps': 22634, 'loss/train': 1.8524560332298279}
+01/29/2022 18:00:52 - INFO - codeparrot_training - Step 22635: {'lr': 0.00030461564943091524, 'samples': 4346112, 'steps': 22635, 'loss/train': 3.009069085121155}
+01/29/2022 18:00:57 - INFO - codeparrot_training - Step 22636: {'lr': 0.0003045996820821979, 'samples': 4346304, 'steps': 22636, 'loss/train': 1.2293091416358948}
+01/29/2022 18:01:01 - INFO - codeparrot_training - Step 22637: {'lr': 0.00030458371449959293, 'samples': 4346496, 'steps': 22637, 'loss/train': 0.9787772297859192}
+01/29/2022 18:01:05 - INFO - codeparrot_training - Step 22638: {'lr': 0.00030456774668316864, 'samples': 4346688, 'steps': 22638, 'loss/train': 2.170802056789398}
+01/29/2022 18:01:09 - INFO - codeparrot_training - Step 22639: {'lr': 0.0003045517786329934, 'samples': 4346880, 'steps': 22639, 'loss/train': 1.1713236272335052}
+01/29/2022 18:01:14 - INFO - codeparrot_training - Step 22640: {'lr': 0.0003045358103491357, 'samples': 4347072, 'steps': 22640, 'loss/train': 1.7208958268165588}
+01/29/2022 18:01:21 - INFO - codeparrot_training - Step 22641: {'lr': 0.00030451984183166384, 'samples': 4347264, 'steps': 22641, 'loss/train': 1.602948009967804}
+01/29/2022 18:01:25 - INFO - codeparrot_training - Step 22642: {'lr': 0.0003045038730806464, 'samples': 4347456, 'steps': 22642, 'loss/train': 2.3609703183174133}
+01/29/2022 18:01:29 - INFO - codeparrot_training - Step 22643: {'lr': 0.00030448790409615155, 'samples': 4347648, 'steps': 22643, 'loss/train': 1.9849843382835388}
+01/29/2022 18:01:33 - INFO - codeparrot_training - Step 22644: {'lr': 0.00030447193487824796, 'samples': 4347840, 'steps': 22644, 'loss/train': 1.5695273280143738}
+01/29/2022 18:01:37 - INFO - codeparrot_training - Step 22645: {'lr': 0.00030445596542700383, 'samples': 4348032, 'steps': 22645, 'loss/train': 1.5953527092933655}
+01/29/2022 18:01:43 - INFO - codeparrot_training - Step 22646: {'lr': 0.00030443999574248764, 'samples': 4348224, 'steps': 22646, 'loss/train': 1.7006077766418457}
+01/29/2022 18:01:47 - INFO - codeparrot_training - Step 22647: {'lr': 0.00030442402582476775, 'samples': 4348416, 'steps': 22647, 'loss/train': 1.3256417512893677}
+01/29/2022 18:01:52 - INFO - codeparrot_training - Step 22648: {'lr': 0.00030440805567391274, 'samples': 4348608, 'steps': 22648, 'loss/train': 1.8449517488479614}
+01/29/2022 18:01:56 - INFO - codeparrot_training - Step 22649: {'lr': 0.00030439208528999074, 'samples': 4348800, 'steps': 22649, 'loss/train': 2.6410857439041138}
+01/29/2022 18:02:00 - INFO - codeparrot_training - Step 22650: {'lr': 0.0003043761146730704, 'samples': 4348992, 'steps': 22650, 'loss/train': 1.1950990855693817}
+01/29/2022 18:02:04 - INFO - codeparrot_training - Step 22651: {'lr': 0.00030436014382322004, 'samples': 4349184, 'steps': 22651, 'loss/train': 2.3110309839248657}
+01/29/2022 18:02:09 - INFO - codeparrot_training - Step 22652: {'lr': 0.00030434417274050805, 'samples': 4349376, 'steps': 22652, 'loss/train': 1.8466633558273315}
+01/29/2022 18:02:14 - INFO - codeparrot_training - Step 22653: {'lr': 0.00030432820142500296, 'samples': 4349568, 'steps': 22653, 'loss/train': 1.9505801796913147}
+01/29/2022 18:02:18 - INFO - codeparrot_training - Step 22654: {'lr': 0.00030431222987677305, 'samples': 4349760, 'steps': 22654, 'loss/train': 1.389021098613739}
+01/29/2022 18:02:22 - INFO - codeparrot_training - Step 22655: {'lr': 0.0003042962580958868, 'samples': 4349952, 'steps': 22655, 'loss/train': 1.6259851455688477}
+01/29/2022 18:02:27 - INFO - codeparrot_training - Step 22656: {'lr': 0.00030428028608241257, 'samples': 4350144, 'steps': 22656, 'loss/train': 1.5154681205749512}
+01/29/2022 18:02:32 - INFO - codeparrot_training - Step 22657: {'lr': 0.00030426431383641893, 'samples': 4350336, 'steps': 22657, 'loss/train': 1.2798936367034912}
+01/29/2022 18:02:36 - INFO - codeparrot_training - Step 22658: {'lr': 0.00030424834135797413, 'samples': 4350528, 'steps': 22658, 'loss/train': 1.3691999316215515}
+01/29/2022 18:02:40 - INFO - codeparrot_training - Step 22659: {'lr': 0.00030423236864714676, 'samples': 4350720, 'steps': 22659, 'loss/train': 1.5329596996307373}
+01/29/2022 18:02:44 - INFO - codeparrot_training - Step 22660: {'lr': 0.00030421639570400505, 'samples': 4350912, 'steps': 22660, 'loss/train': 1.892379641532898}
+01/29/2022 18:02:51 - INFO - codeparrot_training - Step 22661: {'lr': 0.0003042004225286176, 'samples': 4351104, 'steps': 22661, 'loss/train': 2.069654703140259}
+01/29/2022 18:02:56 - INFO - codeparrot_training - Step 22662: {'lr': 0.00030418444912105256, 'samples': 4351296, 'steps': 22662, 'loss/train': 2.0513160824775696}
+01/29/2022 18:03:00 - INFO - codeparrot_training - Step 22663: {'lr': 0.0003041684754813787, 'samples': 4351488, 'steps': 22663, 'loss/train': 1.7221872210502625}
+01/29/2022 18:03:04 - INFO - codeparrot_training - Step 22664: {'lr': 0.0003041525016096643, 'samples': 4351680, 'steps': 22664, 'loss/train': 1.647538661956787}
+01/29/2022 18:03:08 - INFO - codeparrot_training - Step 22665: {'lr': 0.0003041365275059777, 'samples': 4351872, 'steps': 22665, 'loss/train': 2.309602975845337}
+01/29/2022 18:03:12 - INFO - codeparrot_training - Step 22666: {'lr': 0.0003041205531703875, 'samples': 4352064, 'steps': 22666, 'loss/train': 1.45424285531044}
+01/29/2022 18:03:18 - INFO - codeparrot_training - Step 22667: {'lr': 0.0003041045786029619, 'samples': 4352256, 'steps': 22667, 'loss/train': 2.542497217655182}
+01/29/2022 18:03:22 - INFO - codeparrot_training - Step 22668: {'lr': 0.0003040886038037696, 'samples': 4352448, 'steps': 22668, 'loss/train': 1.9844672083854675}
+01/29/2022 18:03:26 - INFO - codeparrot_training - Step 22669: {'lr': 0.00030407262877287883, 'samples': 4352640, 'steps': 22669, 'loss/train': 1.4924677312374115}
+01/29/2022 18:03:31 - INFO - codeparrot_training - Step 22670: {'lr': 0.00030405665351035816, 'samples': 4352832, 'steps': 22670, 'loss/train': 1.7715550661087036}
+01/29/2022 18:03:35 - INFO - codeparrot_training - Step 22671: {'lr': 0.0003040406780162759, 'samples': 4353024, 'steps': 22671, 'loss/train': 2.115138530731201}
+01/29/2022 18:03:40 - INFO - codeparrot_training - Step 22672: {'lr': 0.00030402470229070054, 'samples': 4353216, 'steps': 22672, 'loss/train': 1.7915831208229065}
+01/29/2022 18:03:45 - INFO - codeparrot_training - Step 22673: {'lr': 0.00030400872633370047, 'samples': 4353408, 'steps': 22673, 'loss/train': 1.029433786869049}
+01/29/2022 18:03:49 - INFO - codeparrot_training - Step 22674: {'lr': 0.0003039927501453442, 'samples': 4353600, 'steps': 22674, 'loss/train': 1.8877118825912476}
+01/29/2022 18:03:53 - INFO - codeparrot_training - Step 22675: {'lr': 0.0003039767737257002, 'samples': 4353792, 'steps': 22675, 'loss/train': 0.8239308893680573}
+01/29/2022 18:03:57 - INFO - codeparrot_training - Step 22676: {'lr': 0.0003039607970748368, 'samples': 4353984, 'steps': 22676, 'loss/train': 2.4025379419326782}
+01/29/2022 18:04:04 - INFO - codeparrot_training - Step 22677: {'lr': 0.00030394482019282246, 'samples': 4354176, 'steps': 22677, 'loss/train': 1.7011958956718445}
+01/29/2022 18:04:09 - INFO - codeparrot_training - Step 22678: {'lr': 0.0003039288430797256, 'samples': 4354368, 'steps': 22678, 'loss/train': 1.6839097738265991}
+01/29/2022 18:04:13 - INFO - codeparrot_training - Step 22679: {'lr': 0.0003039128657356147, 'samples': 4354560, 'steps': 22679, 'loss/train': 1.4638222754001617}
+01/29/2022 18:04:17 - INFO - codeparrot_training - Step 22680: {'lr': 0.0003038968881605583, 'samples': 4354752, 'steps': 22680, 'loss/train': 1.6560405492782593}
+01/29/2022 18:04:21 - INFO - codeparrot_training - Step 22681: {'lr': 0.00030388091035462466, 'samples': 4354944, 'steps': 22681, 'loss/train': 0.8733882308006287}
+01/29/2022 18:04:27 - INFO - codeparrot_training - Step 22682: {'lr': 0.00030386493231788234, 'samples': 4355136, 'steps': 22682, 'loss/train': 2.0178637504577637}
+01/29/2022 18:04:31 - INFO - codeparrot_training - Step 22683: {'lr': 0.0003038489540503997, 'samples': 4355328, 'steps': 22683, 'loss/train': 1.6941694021224976}
+01/29/2022 18:04:35 - INFO - codeparrot_training - Step 22684: {'lr': 0.0003038329755522453, 'samples': 4355520, 'steps': 22684, 'loss/train': 0.8521456718444824}
+01/29/2022 18:04:39 - INFO - codeparrot_training - Step 22685: {'lr': 0.0003038169968234875, 'samples': 4355712, 'steps': 22685, 'loss/train': 2.6067540049552917}
+01/29/2022 18:04:43 - INFO - codeparrot_training - Step 22686: {'lr': 0.0003038010178641948, 'samples': 4355904, 'steps': 22686, 'loss/train': 1.5479949116706848}
+01/29/2022 18:04:51 - INFO - codeparrot_training - Step 22687: {'lr': 0.00030378503867443555, 'samples': 4356096, 'steps': 22687, 'loss/train': 2.055567741394043}
+01/29/2022 18:04:55 - INFO - codeparrot_training - Step 22688: {'lr': 0.0003037690592542784, 'samples': 4356288, 'steps': 22688, 'loss/train': 1.2559296190738678}
+01/29/2022 18:04:59 - INFO - codeparrot_training - Step 22689: {'lr': 0.0003037530796037916, 'samples': 4356480, 'steps': 22689, 'loss/train': 1.4546558260917664}
+01/29/2022 18:05:03 - INFO - codeparrot_training - Step 22690: {'lr': 0.0003037370997230436, 'samples': 4356672, 'steps': 22690, 'loss/train': 4.189022541046143}
+01/29/2022 18:05:07 - INFO - codeparrot_training - Step 22691: {'lr': 0.000303721119612103, 'samples': 4356864, 'steps': 22691, 'loss/train': 2.484554171562195}
+01/29/2022 18:05:13 - INFO - codeparrot_training - Step 22692: {'lr': 0.00030370513927103826, 'samples': 4357056, 'steps': 22692, 'loss/train': 0.9178165197372437}
+01/29/2022 18:05:17 - INFO - codeparrot_training - Step 22693: {'lr': 0.0003036891586999176, 'samples': 4357248, 'steps': 22693, 'loss/train': 1.9032052159309387}
+01/29/2022 18:05:21 - INFO - codeparrot_training - Step 22694: {'lr': 0.00030367317789880985, 'samples': 4357440, 'steps': 22694, 'loss/train': 1.4665686786174774}
+01/29/2022 18:05:26 - INFO - codeparrot_training - Step 22695: {'lr': 0.000303657196867783, 'samples': 4357632, 'steps': 22695, 'loss/train': 1.399177998304367}
+01/29/2022 18:05:30 - INFO - codeparrot_training - Step 22696: {'lr': 0.0003036412156069059, 'samples': 4357824, 'steps': 22696, 'loss/train': 0.7604847550392151}
+01/29/2022 18:05:36 - INFO - codeparrot_training - Step 22697: {'lr': 0.00030362523411624686, 'samples': 4358016, 'steps': 22697, 'loss/train': 0.9895264506340027}
+01/29/2022 18:05:40 - INFO - codeparrot_training - Step 22698: {'lr': 0.0003036092523958743, 'samples': 4358208, 'steps': 22698, 'loss/train': 2.0966153740882874}
+01/29/2022 18:05:44 - INFO - codeparrot_training - Step 22699: {'lr': 0.00030359327044585685, 'samples': 4358400, 'steps': 22699, 'loss/train': 1.526306927204132}
+01/29/2022 18:05:48 - INFO - codeparrot_training - Step 22700: {'lr': 0.00030357728826626266, 'samples': 4358592, 'steps': 22700, 'loss/train': 1.9600018858909607}
+01/29/2022 18:05:55 - INFO - codeparrot_training - Step 22701: {'lr': 0.0003035613058571605, 'samples': 4358784, 'steps': 22701, 'loss/train': 1.3661752939224243}
+01/29/2022 18:05:59 - INFO - codeparrot_training - Step 22702: {'lr': 0.00030354532321861865, 'samples': 4358976, 'steps': 22702, 'loss/train': 0.3968551754951477}
+01/29/2022 18:06:03 - INFO - codeparrot_training - Step 22703: {'lr': 0.0003035293403507057, 'samples': 4359168, 'steps': 22703, 'loss/train': 1.915755271911621}
+01/29/2022 18:06:07 - INFO - codeparrot_training - Step 22704: {'lr': 0.00030351335725349, 'samples': 4359360, 'steps': 22704, 'loss/train': 1.3465035259723663}
+01/29/2022 18:06:12 - INFO - codeparrot_training - Step 22705: {'lr': 0.0003034973739270401, 'samples': 4359552, 'steps': 22705, 'loss/train': 2.231407642364502}
+01/29/2022 18:06:17 - INFO - codeparrot_training - Step 22706: {'lr': 0.0003034813903714244, 'samples': 4359744, 'steps': 22706, 'loss/train': 1.1535758972167969}
+01/29/2022 18:06:21 - INFO - codeparrot_training - Step 22707: {'lr': 0.0003034654065867115, 'samples': 4359936, 'steps': 22707, 'loss/train': 0.4374435245990753}
+01/29/2022 18:06:25 - INFO - codeparrot_training - Step 22708: {'lr': 0.0003034494225729697, 'samples': 4360128, 'steps': 22708, 'loss/train': 1.6140959858894348}
+01/29/2022 18:06:29 - INFO - codeparrot_training - Step 22709: {'lr': 0.0003034334383302676, 'samples': 4360320, 'steps': 22709, 'loss/train': 1.8658166527748108}
+01/29/2022 18:06:34 - INFO - codeparrot_training - Step 22710: {'lr': 0.0003034174538586735, 'samples': 4360512, 'steps': 22710, 'loss/train': 1.9266820549964905}
+01/29/2022 18:06:39 - INFO - codeparrot_training - Step 22711: {'lr': 0.00030340146915825605, 'samples': 4360704, 'steps': 22711, 'loss/train': 1.9317008256912231}
+01/29/2022 18:06:43 - INFO - codeparrot_training - Step 22712: {'lr': 0.00030338548422908373, 'samples': 4360896, 'steps': 22712, 'loss/train': 1.4752547442913055}
+01/29/2022 18:06:47 - INFO - codeparrot_training - Step 22713: {'lr': 0.00030336949907122483, 'samples': 4361088, 'steps': 22713, 'loss/train': 1.3171117901802063}
+01/29/2022 18:06:52 - INFO - codeparrot_training - Step 22714: {'lr': 0.0003033535136847481, 'samples': 4361280, 'steps': 22714, 'loss/train': 0.7280515730381012}
+01/29/2022 18:06:56 - INFO - codeparrot_training - Step 22715: {'lr': 0.0003033375280697218, 'samples': 4361472, 'steps': 22715, 'loss/train': 1.7159165740013123}
+01/29/2022 18:07:01 - INFO - codeparrot_training - Step 22716: {'lr': 0.0003033215422262144, 'samples': 4361664, 'steps': 22716, 'loss/train': 1.1653330028057098}
+01/29/2022 18:07:05 - INFO - codeparrot_training - Step 22717: {'lr': 0.0003033055561542945, 'samples': 4361856, 'steps': 22717, 'loss/train': 0.9497595727443695}
+01/29/2022 18:07:09 - INFO - codeparrot_training - Step 22718: {'lr': 0.00030328956985403043, 'samples': 4362048, 'steps': 22718, 'loss/train': 1.3979547321796417}
+01/29/2022 18:07:14 - INFO - codeparrot_training - Step 22719: {'lr': 0.0003032735833254909, 'samples': 4362240, 'steps': 22719, 'loss/train': 1.9712376594543457}
+01/29/2022 18:07:18 - INFO - codeparrot_training - Step 22720: {'lr': 0.0003032575965687442, 'samples': 4362432, 'steps': 22720, 'loss/train': 2.0488414764404297}
+01/29/2022 18:07:25 - INFO - codeparrot_training - Step 22721: {'lr': 0.0003032416095838588, 'samples': 4362624, 'steps': 22721, 'loss/train': 1.5909236669540405}
+01/29/2022 18:07:30 - INFO - codeparrot_training - Step 22722: {'lr': 0.00030322562237090336, 'samples': 4362816, 'steps': 22722, 'loss/train': 1.1047427952289581}
+01/29/2022 18:07:34 - INFO - codeparrot_training - Step 22723: {'lr': 0.00030320963492994616, 'samples': 4363008, 'steps': 22723, 'loss/train': 1.271057903766632}
+01/29/2022 18:07:38 - INFO - codeparrot_training - Step 22724: {'lr': 0.00030319364726105584, 'samples': 4363200, 'steps': 22724, 'loss/train': 1.7241911888122559}
+01/29/2022 18:07:42 - INFO - codeparrot_training - Step 22725: {'lr': 0.00030317765936430086, 'samples': 4363392, 'steps': 22725, 'loss/train': 1.9042607545852661}
+01/29/2022 18:07:47 - INFO - codeparrot_training - Step 22726: {'lr': 0.0003031616712397496, 'samples': 4363584, 'steps': 22726, 'loss/train': 0.8329481184482574}
+01/29/2022 18:07:52 - INFO - codeparrot_training - Step 22727: {'lr': 0.00030314568288747064, 'samples': 4363776, 'steps': 22727, 'loss/train': 2.1147741079330444}
+01/29/2022 18:07:56 - INFO - codeparrot_training - Step 22728: {'lr': 0.00030312969430753244, 'samples': 4363968, 'steps': 22728, 'loss/train': 1.0707303285598755}
+01/29/2022 18:08:00 - INFO - codeparrot_training - Step 22729: {'lr': 0.00030311370550000356, 'samples': 4364160, 'steps': 22729, 'loss/train': 1.3436425924301147}
+01/29/2022 18:08:04 - INFO - codeparrot_training - Step 22730: {'lr': 0.0003030977164649523, 'samples': 4364352, 'steps': 22730, 'loss/train': 2.180356979370117}
+01/29/2022 18:08:10 - INFO - codeparrot_training - Step 22731: {'lr': 0.0003030817272024474, 'samples': 4364544, 'steps': 22731, 'loss/train': 0.8277227282524109}
+01/29/2022 18:08:14 - INFO - codeparrot_training - Step 22732: {'lr': 0.0003030657377125572, 'samples': 4364736, 'steps': 22732, 'loss/train': 1.5730479955673218}
+01/29/2022 18:08:18 - INFO - codeparrot_training - Step 22733: {'lr': 0.0003030497479953503, 'samples': 4364928, 'steps': 22733, 'loss/train': 1.123435914516449}
+01/29/2022 18:08:22 - INFO - codeparrot_training - Step 22734: {'lr': 0.00030303375805089503, 'samples': 4365120, 'steps': 22734, 'loss/train': 1.8944726586341858}
+01/29/2022 18:08:27 - INFO - codeparrot_training - Step 22735: {'lr': 0.00030301776787925995, 'samples': 4365312, 'steps': 22735, 'loss/train': 2.150702476501465}
+01/29/2022 18:08:34 - INFO - codeparrot_training - Step 22736: {'lr': 0.00030300177748051373, 'samples': 4365504, 'steps': 22736, 'loss/train': 1.6872096061706543}
+01/29/2022 18:08:38 - INFO - codeparrot_training - Step 22737: {'lr': 0.0003029857868547246, 'samples': 4365696, 'steps': 22737, 'loss/train': 2.187905788421631}
+01/29/2022 18:08:42 - INFO - codeparrot_training - Step 22738: {'lr': 0.0003029697960019612, 'samples': 4365888, 'steps': 22738, 'loss/train': 2.0301255583763123}
+01/29/2022 18:08:46 - INFO - codeparrot_training - Step 22739: {'lr': 0.0003029538049222921, 'samples': 4366080, 'steps': 22739, 'loss/train': 1.8891257643699646}
+01/29/2022 18:08:51 - INFO - codeparrot_training - Step 22740: {'lr': 0.00030293781361578567, 'samples': 4366272, 'steps': 22740, 'loss/train': 1.9368630051612854}
+01/29/2022 18:08:56 - INFO - codeparrot_training - Step 22741: {'lr': 0.0003029218220825104, 'samples': 4366464, 'steps': 22741, 'loss/train': 2.4061295986175537}
+01/29/2022 18:09:00 - INFO - codeparrot_training - Step 22742: {'lr': 0.00030290583032253486, 'samples': 4366656, 'steps': 22742, 'loss/train': 2.780810058116913}
+01/29/2022 18:09:04 - INFO - codeparrot_training - Step 22743: {'lr': 0.00030288983833592757, 'samples': 4366848, 'steps': 22743, 'loss/train': 2.0040074586868286}
+01/29/2022 18:09:09 - INFO - codeparrot_training - Step 22744: {'lr': 0.00030287384612275704, 'samples': 4367040, 'steps': 22744, 'loss/train': 1.382344901561737}
+01/29/2022 18:09:13 - INFO - codeparrot_training - Step 22745: {'lr': 0.00030285785368309174, 'samples': 4367232, 'steps': 22745, 'loss/train': 1.4574239552021027}
+01/29/2022 18:09:20 - INFO - codeparrot_training - Step 22746: {'lr': 0.0003028418610170001, 'samples': 4367424, 'steps': 22746, 'loss/train': 1.677512526512146}
+01/29/2022 18:09:24 - INFO - codeparrot_training - Step 22747: {'lr': 0.00030282586812455076, 'samples': 4367616, 'steps': 22747, 'loss/train': 1.8202741742134094}
+01/29/2022 18:09:28 - INFO - codeparrot_training - Step 22748: {'lr': 0.00030280987500581213, 'samples': 4367808, 'steps': 22748, 'loss/train': 2.5303276777267456}
+01/29/2022 18:09:33 - INFO - codeparrot_training - Step 22749: {'lr': 0.00030279388166085287, 'samples': 4368000, 'steps': 22749, 'loss/train': 1.4555716216564178}
+01/29/2022 18:09:37 - INFO - codeparrot_training - Step 22750: {'lr': 0.0003027778880897413, 'samples': 4368192, 'steps': 22750, 'loss/train': 1.3192435204982758}
+01/29/2022 18:09:42 - INFO - codeparrot_training - Step 22751: {'lr': 0.000302761894292546, 'samples': 4368384, 'steps': 22751, 'loss/train': 0.8401957154273987}
+01/29/2022 18:09:46 - INFO - codeparrot_training - Step 22752: {'lr': 0.0003027459002693356, 'samples': 4368576, 'steps': 22752, 'loss/train': 1.7822692394256592}
+01/29/2022 18:09:51 - INFO - codeparrot_training - Step 22753: {'lr': 0.00030272990602017843, 'samples': 4368768, 'steps': 22753, 'loss/train': 1.6730920672416687}
+01/29/2022 18:09:55 - INFO - codeparrot_training - Step 22754: {'lr': 0.0003027139115451431, 'samples': 4368960, 'steps': 22754, 'loss/train': 1.4833998084068298}
+01/29/2022 18:09:59 - INFO - codeparrot_training - Step 22755: {'lr': 0.0003026979168442982, 'samples': 4369152, 'steps': 22755, 'loss/train': 1.79094797372818}
+01/29/2022 18:10:04 - INFO - codeparrot_training - Step 22756: {'lr': 0.000302681921917712, 'samples': 4369344, 'steps': 22756, 'loss/train': 1.1190220713615417}
+01/29/2022 18:10:08 - INFO - codeparrot_training - Step 22757: {'lr': 0.00030266592676545326, 'samples': 4369536, 'steps': 22757, 'loss/train': 2.3139299154281616}
+01/29/2022 18:10:13 - INFO - codeparrot_training - Step 22758: {'lr': 0.0003026499313875903, 'samples': 4369728, 'steps': 22758, 'loss/train': 1.6533679962158203}
+01/29/2022 18:10:17 - INFO - codeparrot_training - Step 22759: {'lr': 0.00030263393578419196, 'samples': 4369920, 'steps': 22759, 'loss/train': 1.7058168053627014}
+01/29/2022 18:10:21 - INFO - codeparrot_training - Step 22760: {'lr': 0.0003026179399553264, 'samples': 4370112, 'steps': 22760, 'loss/train': 1.7461187839508057}
+01/29/2022 18:10:28 - INFO - codeparrot_training - Step 22761: {'lr': 0.0003026019439010624, 'samples': 4370304, 'steps': 22761, 'loss/train': 0.9192829728126526}
+01/29/2022 18:10:32 - INFO - codeparrot_training - Step 22762: {'lr': 0.0003025859476214683, 'samples': 4370496, 'steps': 22762, 'loss/train': 1.6669732332229614}
+01/29/2022 18:10:37 - INFO - codeparrot_training - Step 22763: {'lr': 0.00030256995111661275, 'samples': 4370688, 'steps': 22763, 'loss/train': 1.9118657112121582}
+01/29/2022 18:10:41 - INFO - codeparrot_training - Step 22764: {'lr': 0.0003025539543865642, 'samples': 4370880, 'steps': 22764, 'loss/train': 0.9710506796836853}
+01/29/2022 18:10:45 - INFO - codeparrot_training - Step 22765: {'lr': 0.00030253795743139113, 'samples': 4371072, 'steps': 22765, 'loss/train': 1.6971184015274048}
+01/29/2022 18:10:50 - INFO - codeparrot_training - Step 22766: {'lr': 0.0003025219602511622, 'samples': 4371264, 'steps': 22766, 'loss/train': 1.8292352557182312}
+01/29/2022 18:10:55 - INFO - codeparrot_training - Step 22767: {'lr': 0.00030250596284594583, 'samples': 4371456, 'steps': 22767, 'loss/train': 1.5609031319618225}
+01/29/2022 18:10:59 - INFO - codeparrot_training - Step 22768: {'lr': 0.0003024899652158107, 'samples': 4371648, 'steps': 22768, 'loss/train': 1.2597769796848297}
+01/29/2022 18:11:03 - INFO - codeparrot_training - Step 22769: {'lr': 0.00030247396736082506, 'samples': 4371840, 'steps': 22769, 'loss/train': 1.7390357851982117}
+01/29/2022 18:11:07 - INFO - codeparrot_training - Step 22770: {'lr': 0.0003024579692810577, 'samples': 4372032, 'steps': 22770, 'loss/train': 1.6391721367835999}
+01/29/2022 18:11:13 - INFO - codeparrot_training - Step 22771: {'lr': 0.00030244197097657705, 'samples': 4372224, 'steps': 22771, 'loss/train': 2.0379050374031067}
+01/29/2022 18:11:17 - INFO - codeparrot_training - Step 22772: {'lr': 0.0003024259724474516, 'samples': 4372416, 'steps': 22772, 'loss/train': 1.6789941787719727}
+01/29/2022 18:11:21 - INFO - codeparrot_training - Step 22773: {'lr': 0.00030240997369375, 'samples': 4372608, 'steps': 22773, 'loss/train': 1.9578511118888855}
+01/29/2022 18:11:25 - INFO - codeparrot_training - Step 22774: {'lr': 0.0003023939747155406, 'samples': 4372800, 'steps': 22774, 'loss/train': 2.55124694108963}
+01/29/2022 18:11:29 - INFO - codeparrot_training - Step 22775: {'lr': 0.00030237797551289225, 'samples': 4372992, 'steps': 22775, 'loss/train': 2.1401272416114807}
+01/29/2022 18:11:35 - INFO - codeparrot_training - Step 22776: {'lr': 0.0003023619760858731, 'samples': 4373184, 'steps': 22776, 'loss/train': 1.2365278601646423}
+01/29/2022 18:11:39 - INFO - codeparrot_training - Step 22777: {'lr': 0.000302345976434552, 'samples': 4373376, 'steps': 22777, 'loss/train': 1.6188815832138062}
+01/29/2022 18:11:43 - INFO - codeparrot_training - Step 22778: {'lr': 0.0003023299765589973, 'samples': 4373568, 'steps': 22778, 'loss/train': 1.409152328968048}
+01/29/2022 18:11:47 - INFO - codeparrot_training - Step 22779: {'lr': 0.0003023139764592776, 'samples': 4373760, 'steps': 22779, 'loss/train': 1.8776968717575073}
+01/29/2022 18:11:51 - INFO - codeparrot_training - Step 22780: {'lr': 0.0003022979761354614, 'samples': 4373952, 'steps': 22780, 'loss/train': 1.6018590331077576}
+01/29/2022 18:11:59 - INFO - codeparrot_training - Step 22781: {'lr': 0.00030228197558761737, 'samples': 4374144, 'steps': 22781, 'loss/train': 1.7030276656150818}
+01/29/2022 18:12:03 - INFO - codeparrot_training - Step 22782: {'lr': 0.00030226597481581387, 'samples': 4374336, 'steps': 22782, 'loss/train': 1.8905922174453735}
+01/29/2022 18:12:07 - INFO - codeparrot_training - Step 22783: {'lr': 0.0003022499738201195, 'samples': 4374528, 'steps': 22783, 'loss/train': 1.4095516204833984}
+01/29/2022 18:12:11 - INFO - codeparrot_training - Step 22784: {'lr': 0.0003022339726006029, 'samples': 4374720, 'steps': 22784, 'loss/train': 1.3552350103855133}
+01/29/2022 18:12:15 - INFO - codeparrot_training - Step 22785: {'lr': 0.0003022179711573326, 'samples': 4374912, 'steps': 22785, 'loss/train': 1.4603311121463776}
+01/29/2022 18:12:21 - INFO - codeparrot_training - Step 22786: {'lr': 0.000302201969490377, 'samples': 4375104, 'steps': 22786, 'loss/train': 1.8948265314102173}
+01/29/2022 18:12:25 - INFO - codeparrot_training - Step 22787: {'lr': 0.00030218596759980476, 'samples': 4375296, 'steps': 22787, 'loss/train': 1.8659250140190125}
+01/29/2022 18:12:29 - INFO - codeparrot_training - Step 22788: {'lr': 0.00030216996548568443, 'samples': 4375488, 'steps': 22788, 'loss/train': 1.5810002088546753}
+01/29/2022 18:12:34 - INFO - codeparrot_training - Step 22789: {'lr': 0.0003021539631480845, 'samples': 4375680, 'steps': 22789, 'loss/train': 1.231415569782257}
+01/29/2022 18:12:41 - INFO - codeparrot_training - Step 22790: {'lr': 0.0003021379605870736, 'samples': 4375872, 'steps': 22790, 'loss/train': 1.9083879590034485}
+01/29/2022 18:12:45 - INFO - codeparrot_training - Step 22791: {'lr': 0.00030212195780272025, 'samples': 4376064, 'steps': 22791, 'loss/train': 0.31856435537338257}
+01/29/2022 18:12:49 - INFO - codeparrot_training - Step 22792: {'lr': 0.000302105954795093, 'samples': 4376256, 'steps': 22792, 'loss/train': 0.5004651099443436}
+01/29/2022 18:12:53 - INFO - codeparrot_training - Step 22793: {'lr': 0.00030208995156426024, 'samples': 4376448, 'steps': 22793, 'loss/train': 1.8137975335121155}
+01/29/2022 18:12:58 - INFO - codeparrot_training - Step 22794: {'lr': 0.0003020739481102908, 'samples': 4376640, 'steps': 22794, 'loss/train': 1.3704683482646942}
+01/29/2022 18:13:03 - INFO - codeparrot_training - Step 22795: {'lr': 0.00030205794443325296, 'samples': 4376832, 'steps': 22795, 'loss/train': 1.3576910197734833}
+01/29/2022 18:13:07 - INFO - codeparrot_training - Step 22796: {'lr': 0.00030204194053321556, 'samples': 4377024, 'steps': 22796, 'loss/train': 1.5459160208702087}
+01/29/2022 18:13:11 - INFO - codeparrot_training - Step 22797: {'lr': 0.00030202593641024696, 'samples': 4377216, 'steps': 22797, 'loss/train': 1.5351401567459106}
+01/29/2022 18:13:15 - INFO - codeparrot_training - Step 22798: {'lr': 0.0003020099320644157, 'samples': 4377408, 'steps': 22798, 'loss/train': 4.13608717918396}
+01/29/2022 18:13:20 - INFO - codeparrot_training - Step 22799: {'lr': 0.00030199392749579053, 'samples': 4377600, 'steps': 22799, 'loss/train': 1.297102689743042}
+01/29/2022 18:13:25 - INFO - codeparrot_training - Step 22800: {'lr': 0.0003019779227044398, 'samples': 4377792, 'steps': 22800, 'loss/train': 1.0996491014957428}
+01/29/2022 18:13:29 - INFO - codeparrot_training - Step 22801: {'lr': 0.0003019619176904322, 'samples': 4377984, 'steps': 22801, 'loss/train': 1.4743590652942657}
+01/29/2022 18:13:33 - INFO - codeparrot_training - Step 22802: {'lr': 0.00030194591245383625, 'samples': 4378176, 'steps': 22802, 'loss/train': 2.5174614787101746}
+01/29/2022 18:13:37 - INFO - codeparrot_training - Step 22803: {'lr': 0.00030192990699472053, 'samples': 4378368, 'steps': 22803, 'loss/train': 1.0262600183486938}
+01/29/2022 18:13:42 - INFO - codeparrot_training - Step 22804: {'lr': 0.00030191390131315357, 'samples': 4378560, 'steps': 22804, 'loss/train': 1.8829777836799622}
+01/29/2022 18:13:49 - INFO - codeparrot_training - Step 22805: {'lr': 0.0003018978954092039, 'samples': 4378752, 'steps': 22805, 'loss/train': 1.7279403805732727}
+01/29/2022 18:13:53 - INFO - codeparrot_training - Step 22806: {'lr': 0.0003018818892829401, 'samples': 4378944, 'steps': 22806, 'loss/train': 1.3937139809131622}
+01/29/2022 18:13:57 - INFO - codeparrot_training - Step 22807: {'lr': 0.00030186588293443077, 'samples': 4379136, 'steps': 22807, 'loss/train': 1.8231943845748901}
+01/29/2022 18:14:01 - INFO - codeparrot_training - Step 22808: {'lr': 0.0003018498763637445, 'samples': 4379328, 'steps': 22808, 'loss/train': 1.8310550451278687}
+01/29/2022 18:14:06 - INFO - codeparrot_training - Step 22809: {'lr': 0.0003018338695709498, 'samples': 4379520, 'steps': 22809, 'loss/train': 0.640752375125885}
+01/29/2022 18:14:11 - INFO - codeparrot_training - Step 22810: {'lr': 0.0003018178625561153, 'samples': 4379712, 'steps': 22810, 'loss/train': 1.6814045906066895}
+01/29/2022 18:14:15 - INFO - codeparrot_training - Step 22811: {'lr': 0.0003018018553193095, 'samples': 4379904, 'steps': 22811, 'loss/train': 1.8911942839622498}
+01/29/2022 18:14:19 - INFO - codeparrot_training - Step 22812: {'lr': 0.00030178584786060106, 'samples': 4380096, 'steps': 22812, 'loss/train': 1.2076064050197601}
+01/29/2022 18:14:24 - INFO - codeparrot_training - Step 22813: {'lr': 0.00030176984018005836, 'samples': 4380288, 'steps': 22813, 'loss/train': 2.244420289993286}
+01/29/2022 18:14:28 - INFO - codeparrot_training - Step 22814: {'lr': 0.0003017538322777502, 'samples': 4380480, 'steps': 22814, 'loss/train': 1.3622834086418152}
+01/29/2022 18:14:35 - INFO - codeparrot_training - Step 22815: {'lr': 0.00030173782415374503, 'samples': 4380672, 'steps': 22815, 'loss/train': 1.0713537633419037}
+01/29/2022 18:14:39 - INFO - codeparrot_training - Step 22816: {'lr': 0.00030172181580811146, 'samples': 4380864, 'steps': 22816, 'loss/train': 1.9777650833129883}
+01/29/2022 18:14:43 - INFO - codeparrot_training - Step 22817: {'lr': 0.0003017058072409181, 'samples': 4381056, 'steps': 22817, 'loss/train': 1.952803373336792}
+01/29/2022 18:14:48 - INFO - codeparrot_training - Step 22818: {'lr': 0.0003016897984522334, 'samples': 4381248, 'steps': 22818, 'loss/train': 1.1806217730045319}
+01/29/2022 18:14:52 - INFO - codeparrot_training - Step 22819: {'lr': 0.00030167378944212606, 'samples': 4381440, 'steps': 22819, 'loss/train': 1.0166111290454865}
+01/29/2022 18:14:57 - INFO - codeparrot_training - Step 22820: {'lr': 0.0003016577802106645, 'samples': 4381632, 'steps': 22820, 'loss/train': 2.1523951292037964}
+01/29/2022 18:15:01 - INFO - codeparrot_training - Step 22821: {'lr': 0.0003016417707579176, 'samples': 4381824, 'steps': 22821, 'loss/train': 1.309501200914383}
+01/29/2022 18:15:05 - INFO - codeparrot_training - Step 22822: {'lr': 0.00030162576108395364, 'samples': 4382016, 'steps': 22822, 'loss/train': 2.0729949474334717}
+01/29/2022 18:15:10 - INFO - codeparrot_training - Step 22823: {'lr': 0.0003016097511888414, 'samples': 4382208, 'steps': 22823, 'loss/train': 2.202359139919281}
+01/29/2022 18:15:14 - INFO - codeparrot_training - Step 22824: {'lr': 0.0003015937410726493, 'samples': 4382400, 'steps': 22824, 'loss/train': 1.82895165681839}
+01/29/2022 18:15:19 - INFO - codeparrot_training - Step 22825: {'lr': 0.000301577730735446, 'samples': 4382592, 'steps': 22825, 'loss/train': 1.6153437495231628}
+01/29/2022 18:15:23 - INFO - codeparrot_training - Step 22826: {'lr': 0.00030156172017730006, 'samples': 4382784, 'steps': 22826, 'loss/train': 1.3816828429698944}
+01/29/2022 18:15:27 - INFO - codeparrot_training - Step 22827: {'lr': 0.00030154570939828015, 'samples': 4382976, 'steps': 22827, 'loss/train': 0.12029841542243958}
+01/29/2022 18:15:32 - INFO - codeparrot_training - Step 22828: {'lr': 0.0003015296983984547, 'samples': 4383168, 'steps': 22828, 'loss/train': 1.936715841293335}
+01/29/2022 18:15:36 - INFO - codeparrot_training - Step 22829: {'lr': 0.00030151368717789244, 'samples': 4383360, 'steps': 22829, 'loss/train': 1.509410798549652}
+01/29/2022 18:15:41 - INFO - codeparrot_training - Step 22830: {'lr': 0.0003014976757366619, 'samples': 4383552, 'steps': 22830, 'loss/train': 1.2906233668327332}
+01/29/2022 18:15:46 - INFO - codeparrot_training - Step 22831: {'lr': 0.0003014816640748316, 'samples': 4383744, 'steps': 22831, 'loss/train': 0.45537570118904114}
+01/29/2022 18:15:50 - INFO - codeparrot_training - Step 22832: {'lr': 0.00030146565219247033, 'samples': 4383936, 'steps': 22832, 'loss/train': 2.0401943922042847}
+01/29/2022 18:15:54 - INFO - codeparrot_training - Step 22833: {'lr': 0.0003014496400896465, 'samples': 4384128, 'steps': 22833, 'loss/train': 2.1697239875793457}
+01/29/2022 18:15:58 - INFO - codeparrot_training - Step 22834: {'lr': 0.0003014336277664287, 'samples': 4384320, 'steps': 22834, 'loss/train': 1.8260303735733032}
+01/29/2022 18:16:05 - INFO - codeparrot_training - Step 22835: {'lr': 0.0003014176152228856, 'samples': 4384512, 'steps': 22835, 'loss/train': 0.5460586398839951}
+01/29/2022 18:16:09 - INFO - codeparrot_training - Step 22836: {'lr': 0.00030140160245908584, 'samples': 4384704, 'steps': 22836, 'loss/train': 1.803138256072998}
+01/29/2022 18:16:13 - INFO - codeparrot_training - Step 22837: {'lr': 0.0003013855894750978, 'samples': 4384896, 'steps': 22837, 'loss/train': 1.4124256074428558}
+01/29/2022 18:16:17 - INFO - codeparrot_training - Step 22838: {'lr': 0.0003013695762709903, 'samples': 4385088, 'steps': 22838, 'loss/train': 1.141856074333191}
+01/29/2022 18:16:22 - INFO - codeparrot_training - Step 22839: {'lr': 0.0003013535628468318, 'samples': 4385280, 'steps': 22839, 'loss/train': 1.673801064491272}
+01/29/2022 18:16:27 - INFO - codeparrot_training - Step 22840: {'lr': 0.000301337549202691, 'samples': 4385472, 'steps': 22840, 'loss/train': 1.7270865440368652}
+01/29/2022 18:16:32 - INFO - codeparrot_training - Step 22841: {'lr': 0.0003013215353386364, 'samples': 4385664, 'steps': 22841, 'loss/train': 1.0781209766864777}
+01/29/2022 18:16:36 - INFO - codeparrot_training - Step 22842: {'lr': 0.00030130552125473667, 'samples': 4385856, 'steps': 22842, 'loss/train': 1.826519250869751}
+01/29/2022 18:16:40 - INFO - codeparrot_training - Step 22843: {'lr': 0.00030128950695106034, 'samples': 4386048, 'steps': 22843, 'loss/train': 1.1155222356319427}
+01/29/2022 18:16:44 - INFO - codeparrot_training - Step 22844: {'lr': 0.00030127349242767607, 'samples': 4386240, 'steps': 22844, 'loss/train': 1.4012559950351715}
+01/29/2022 18:16:49 - INFO - codeparrot_training - Step 22845: {'lr': 0.0003012574776846524, 'samples': 4386432, 'steps': 22845, 'loss/train': 1.1858426928520203}
+01/29/2022 18:16:54 - INFO - codeparrot_training - Step 22846: {'lr': 0.00030124146272205804, 'samples': 4386624, 'steps': 22846, 'loss/train': 1.4142619371414185}
+01/29/2022 18:16:58 - INFO - codeparrot_training - Step 22847: {'lr': 0.00030122544753996143, 'samples': 4386816, 'steps': 22847, 'loss/train': 1.6412867903709412}
+01/29/2022 18:17:02 - INFO - codeparrot_training - Step 22848: {'lr': 0.00030120943213843136, 'samples': 4387008, 'steps': 22848, 'loss/train': 1.6413654685020447}
+01/29/2022 18:17:06 - INFO - codeparrot_training - Step 22849: {'lr': 0.0003011934165175363, 'samples': 4387200, 'steps': 22849, 'loss/train': 0.9747190475463867}
+01/29/2022 18:17:13 - INFO - codeparrot_training - Step 22850: {'lr': 0.00030117740067734495, 'samples': 4387392, 'steps': 22850, 'loss/train': 0.5448332726955414}
+01/29/2022 18:17:17 - INFO - codeparrot_training - Step 22851: {'lr': 0.0003011613846179258, 'samples': 4387584, 'steps': 22851, 'loss/train': 1.674925446510315}
+01/29/2022 18:17:22 - INFO - codeparrot_training - Step 22852: {'lr': 0.0003011453683393476, 'samples': 4387776, 'steps': 22852, 'loss/train': 1.2593805491924286}
+01/29/2022 18:17:26 - INFO - codeparrot_training - Step 22853: {'lr': 0.0003011293518416788, 'samples': 4387968, 'steps': 22853, 'loss/train': 1.5952216386795044}
+01/29/2022 18:17:30 - INFO - codeparrot_training - Step 22854: {'lr': 0.00030111333512498813, 'samples': 4388160, 'steps': 22854, 'loss/train': 1.69060617685318}
+01/29/2022 18:17:35 - INFO - codeparrot_training - Step 22855: {'lr': 0.00030109731818934413, 'samples': 4388352, 'steps': 22855, 'loss/train': 2.325329840183258}
+01/29/2022 18:17:39 - INFO - codeparrot_training - Step 22856: {'lr': 0.00030108130103481554, 'samples': 4388544, 'steps': 22856, 'loss/train': 0.9819927513599396}
+01/29/2022 18:17:44 - INFO - codeparrot_training - Step 22857: {'lr': 0.0003010652836614707, 'samples': 4388736, 'steps': 22857, 'loss/train': 1.9060855507850647}
+01/29/2022 18:17:48 - INFO - codeparrot_training - Step 22858: {'lr': 0.00030104926606937856, 'samples': 4388928, 'steps': 22858, 'loss/train': 2.0269762873649597}
+01/29/2022 18:17:52 - INFO - codeparrot_training - Step 22859: {'lr': 0.0003010332482586075, 'samples': 4389120, 'steps': 22859, 'loss/train': 0.5463093817234039}
+01/29/2022 18:17:57 - INFO - codeparrot_training - Step 22860: {'lr': 0.0003010172302292263, 'samples': 4389312, 'steps': 22860, 'loss/train': 1.5653629302978516}
+01/29/2022 18:18:01 - INFO - codeparrot_training - Step 22861: {'lr': 0.00030100121198130335, 'samples': 4389504, 'steps': 22861, 'loss/train': 2.037302255630493}
+01/29/2022 18:18:06 - INFO - codeparrot_training - Step 22862: {'lr': 0.0003009851935149075, 'samples': 4389696, 'steps': 22862, 'loss/train': 0.9454594552516937}
+01/29/2022 18:18:10 - INFO - codeparrot_training - Step 22863: {'lr': 0.0003009691748301072, 'samples': 4389888, 'steps': 22863, 'loss/train': 1.4355264902114868}
+01/29/2022 18:18:14 - INFO - codeparrot_training - Step 22864: {'lr': 0.0003009531559269713, 'samples': 4390080, 'steps': 22864, 'loss/train': 1.7166073322296143}
+01/29/2022 18:18:21 - INFO - codeparrot_training - Step 22865: {'lr': 0.00030093713680556805, 'samples': 4390272, 'steps': 22865, 'loss/train': 0.43563611805438995}
+01/29/2022 18:18:26 - INFO - codeparrot_training - Step 22866: {'lr': 0.0003009211174659664, 'samples': 4390464, 'steps': 22866, 'loss/train': 1.112350881099701}
+01/29/2022 18:18:30 - INFO - codeparrot_training - Step 22867: {'lr': 0.00030090509790823476, 'samples': 4390656, 'steps': 22867, 'loss/train': 1.817123293876648}
+01/29/2022 18:18:34 - INFO - codeparrot_training - Step 22868: {'lr': 0.0003008890781324419, 'samples': 4390848, 'steps': 22868, 'loss/train': 1.3850375711917877}
+01/29/2022 18:18:38 - INFO - codeparrot_training - Step 22869: {'lr': 0.0003008730581386564, 'samples': 4391040, 'steps': 22869, 'loss/train': 1.5528718829154968}
+01/29/2022 18:18:44 - INFO - codeparrot_training - Step 22870: {'lr': 0.00030085703792694687, 'samples': 4391232, 'steps': 22870, 'loss/train': 2.2865429520606995}
+01/29/2022 18:18:48 - INFO - codeparrot_training - Step 22871: {'lr': 0.00030084101749738195, 'samples': 4391424, 'steps': 22871, 'loss/train': 1.5292319655418396}
+01/29/2022 18:18:52 - INFO - codeparrot_training - Step 22872: {'lr': 0.00030082499685003025, 'samples': 4391616, 'steps': 22872, 'loss/train': 1.5616831183433533}
+01/29/2022 18:18:56 - INFO - codeparrot_training - Step 22873: {'lr': 0.0003008089759849604, 'samples': 4391808, 'steps': 22873, 'loss/train': 0.3576483204960823}
+01/29/2022 18:19:00 - INFO - codeparrot_training - Step 22874: {'lr': 0.000300792954902241, 'samples': 4392000, 'steps': 22874, 'loss/train': 1.2270965874195099}
+01/29/2022 18:19:08 - INFO - codeparrot_training - Step 22875: {'lr': 0.00030077693360194076, 'samples': 4392192, 'steps': 22875, 'loss/train': 1.4320040345191956}
+01/29/2022 18:19:12 - INFO - codeparrot_training - Step 22876: {'lr': 0.0003007609120841282, 'samples': 4392384, 'steps': 22876, 'loss/train': 1.3605059087276459}
+01/29/2022 18:19:16 - INFO - codeparrot_training - Step 22877: {'lr': 0.0003007448903488721, 'samples': 4392576, 'steps': 22877, 'loss/train': 1.3770816028118134}
+01/29/2022 18:19:21 - INFO - codeparrot_training - Step 22878: {'lr': 0.00030072886839624093, 'samples': 4392768, 'steps': 22878, 'loss/train': 1.3724685609340668}
+01/29/2022 18:19:25 - INFO - codeparrot_training - Step 22879: {'lr': 0.0003007128462263034, 'samples': 4392960, 'steps': 22879, 'loss/train': 1.7986995577812195}
+01/29/2022 18:19:30 - INFO - codeparrot_training - Step 22880: {'lr': 0.0003006968238391281, 'samples': 4393152, 'steps': 22880, 'loss/train': 1.3728920817375183}
+01/29/2022 18:19:34 - INFO - codeparrot_training - Step 22881: {'lr': 0.00030068080123478376, 'samples': 4393344, 'steps': 22881, 'loss/train': 0.7580476999282837}
+01/29/2022 18:19:38 - INFO - codeparrot_training - Step 22882: {'lr': 0.000300664778413339, 'samples': 4393536, 'steps': 22882, 'loss/train': 1.8079531788825989}
+01/29/2022 18:19:43 - INFO - codeparrot_training - Step 22883: {'lr': 0.00030064875537486236, 'samples': 4393728, 'steps': 22883, 'loss/train': 2.0214276909828186}
+01/29/2022 18:19:47 - INFO - codeparrot_training - Step 22884: {'lr': 0.00030063273211942254, 'samples': 4393920, 'steps': 22884, 'loss/train': 1.3957232236862183}
+01/29/2022 18:19:52 - INFO - codeparrot_training - Step 22885: {'lr': 0.0003006167086470882, 'samples': 4394112, 'steps': 22885, 'loss/train': 2.091041922569275}
+01/29/2022 18:19:56 - INFO - codeparrot_training - Step 22886: {'lr': 0.00030060068495792793, 'samples': 4394304, 'steps': 22886, 'loss/train': 1.520008385181427}
+01/29/2022 18:20:01 - INFO - codeparrot_training - Step 22887: {'lr': 0.0003005846610520104, 'samples': 4394496, 'steps': 22887, 'loss/train': 1.8482022285461426}
+01/29/2022 18:20:05 - INFO - codeparrot_training - Step 22888: {'lr': 0.00030056863692940426, 'samples': 4394688, 'steps': 22888, 'loss/train': 1.130009651184082}
+01/29/2022 18:20:09 - INFO - codeparrot_training - Step 22889: {'lr': 0.00030055261259017807, 'samples': 4394880, 'steps': 22889, 'loss/train': 1.6096422672271729}
+01/29/2022 18:20:14 - INFO - codeparrot_training - Step 22890: {'lr': 0.00030053658803440064, 'samples': 4395072, 'steps': 22890, 'loss/train': 1.81514310836792}
+01/29/2022 18:20:18 - INFO - codeparrot_training - Step 22891: {'lr': 0.00030052056326214046, 'samples': 4395264, 'steps': 22891, 'loss/train': 1.5317895412445068}
+01/29/2022 18:20:23 - INFO - codeparrot_training - Step 22892: {'lr': 0.00030050453827346627, 'samples': 4395456, 'steps': 22892, 'loss/train': 1.9805434942245483}
+01/29/2022 18:20:27 - INFO - codeparrot_training - Step 22893: {'lr': 0.0003004885130684467, 'samples': 4395648, 'steps': 22893, 'loss/train': 1.3283092081546783}
+01/29/2022 18:20:31 - INFO - codeparrot_training - Step 22894: {'lr': 0.00030047248764715023, 'samples': 4395840, 'steps': 22894, 'loss/train': 1.4930495917797089}
+01/29/2022 18:20:37 - INFO - codeparrot_training - Step 22895: {'lr': 0.0003004564620096457, 'samples': 4396032, 'steps': 22895, 'loss/train': 2.0945271849632263}
+01/29/2022 18:20:42 - INFO - codeparrot_training - Step 22896: {'lr': 0.00030044043615600174, 'samples': 4396224, 'steps': 22896, 'loss/train': 1.868647813796997}
+01/29/2022 18:20:46 - INFO - codeparrot_training - Step 22897: {'lr': 0.000300424410086287, 'samples': 4396416, 'steps': 22897, 'loss/train': 0.6828644424676895}
+01/29/2022 18:20:50 - INFO - codeparrot_training - Step 22898: {'lr': 0.00030040838380057005, 'samples': 4396608, 'steps': 22898, 'loss/train': 1.6245787739753723}
+01/29/2022 18:20:54 - INFO - codeparrot_training - Step 22899: {'lr': 0.00030039235729891964, 'samples': 4396800, 'steps': 22899, 'loss/train': 0.88378044962883}
+01/29/2022 18:21:00 - INFO - codeparrot_training - Step 22900: {'lr': 0.0003003763305814043, 'samples': 4396992, 'steps': 22900, 'loss/train': 1.8239076733589172}
+01/29/2022 18:21:04 - INFO - codeparrot_training - Step 22901: {'lr': 0.00030036030364809284, 'samples': 4397184, 'steps': 22901, 'loss/train': 1.773227334022522}
+01/29/2022 18:21:08 - INFO - codeparrot_training - Step 22902: {'lr': 0.00030034427649905377, 'samples': 4397376, 'steps': 22902, 'loss/train': 1.754398226737976}
+01/29/2022 18:21:12 - INFO - codeparrot_training - Step 22903: {'lr': 0.0003003282491343559, 'samples': 4397568, 'steps': 22903, 'loss/train': 2.2365134954452515}
+01/29/2022 18:21:20 - INFO - codeparrot_training - Step 22904: {'lr': 0.00030031222155406763, 'samples': 4397760, 'steps': 22904, 'loss/train': 0.7728082537651062}
+01/29/2022 18:21:24 - INFO - codeparrot_training - Step 22905: {'lr': 0.00030029619375825784, 'samples': 4397952, 'steps': 22905, 'loss/train': 1.6769290566444397}
+01/29/2022 18:21:28 - INFO - codeparrot_training - Step 22906: {'lr': 0.00030028016574699517, 'samples': 4398144, 'steps': 22906, 'loss/train': 1.7948432564735413}
+01/29/2022 18:21:33 - INFO - codeparrot_training - Step 22907: {'lr': 0.0003002641375203482, 'samples': 4398336, 'steps': 22907, 'loss/train': 1.1459676325321198}
+01/29/2022 18:21:37 - INFO - codeparrot_training - Step 22908: {'lr': 0.0003002481090783856, 'samples': 4398528, 'steps': 22908, 'loss/train': 1.7443251013755798}
+01/29/2022 18:21:41 - INFO - codeparrot_training - Step 22909: {'lr': 0.0003002320804211761, 'samples': 4398720, 'steps': 22909, 'loss/train': 0.48318345844745636}
+01/29/2022 18:21:46 - INFO - codeparrot_training - Step 22910: {'lr': 0.00030021605154878836, 'samples': 4398912, 'steps': 22910, 'loss/train': 0.28127557039260864}
+01/29/2022 18:21:51 - INFO - codeparrot_training - Step 22911: {'lr': 0.0003002000224612909, 'samples': 4399104, 'steps': 22911, 'loss/train': 1.9011738896369934}
+01/29/2022 18:21:55 - INFO - codeparrot_training - Step 22912: {'lr': 0.0003001839931587526, 'samples': 4399296, 'steps': 22912, 'loss/train': 1.5495071411132812}
+01/29/2022 18:21:59 - INFO - codeparrot_training - Step 22913: {'lr': 0.0003001679636412419, 'samples': 4399488, 'steps': 22913, 'loss/train': 1.8258352875709534}
+01/29/2022 18:22:04 - INFO - codeparrot_training - Step 22914: {'lr': 0.0003001519339088277, 'samples': 4399680, 'steps': 22914, 'loss/train': 1.5052485466003418}
+01/29/2022 18:22:09 - INFO - codeparrot_training - Step 22915: {'lr': 0.00030013590396157843, 'samples': 4399872, 'steps': 22915, 'loss/train': 1.5976653099060059}
+01/29/2022 18:22:13 - INFO - codeparrot_training - Step 22916: {'lr': 0.0003001198737995628, 'samples': 4400064, 'steps': 22916, 'loss/train': 2.8728012442588806}
+01/29/2022 18:22:17 - INFO - codeparrot_training - Step 22917: {'lr': 0.0003001038434228497, 'samples': 4400256, 'steps': 22917, 'loss/train': 2.209022283554077}
+01/29/2022 18:22:21 - INFO - codeparrot_training - Step 22918: {'lr': 0.00030008781283150755, 'samples': 4400448, 'steps': 22918, 'loss/train': 1.693701982498169}
+01/29/2022 18:22:29 - INFO - codeparrot_training - Step 22919: {'lr': 0.0003000717820256052, 'samples': 4400640, 'steps': 22919, 'loss/train': 1.3219149112701416}
+01/29/2022 18:22:33 - INFO - codeparrot_training - Step 22920: {'lr': 0.00030005575100521117, 'samples': 4400832, 'steps': 22920, 'loss/train': 1.7044764161109924}
+01/29/2022 18:22:38 - INFO - codeparrot_training - Step 22921: {'lr': 0.0003000397197703942, 'samples': 4401024, 'steps': 22921, 'loss/train': 2.485175371170044}
+01/29/2022 18:22:42 - INFO - codeparrot_training - Step 22922: {'lr': 0.00030002368832122295, 'samples': 4401216, 'steps': 22922, 'loss/train': 1.2345252335071564}
+01/29/2022 18:22:46 - INFO - codeparrot_training - Step 22923: {'lr': 0.00030000765665776617, 'samples': 4401408, 'steps': 22923, 'loss/train': 2.157676041126251}
+01/29/2022 18:22:51 - INFO - codeparrot_training - Step 22924: {'lr': 0.0002999916247800924, 'samples': 4401600, 'steps': 22924, 'loss/train': 1.7933852076530457}
+01/29/2022 18:22:55 - INFO - codeparrot_training - Step 22925: {'lr': 0.00029997559268827044, 'samples': 4401792, 'steps': 22925, 'loss/train': 1.7508324980735779}
+01/29/2022 18:23:00 - INFO - codeparrot_training - Step 22926: {'lr': 0.0002999595603823689, 'samples': 4401984, 'steps': 22926, 'loss/train': 1.3589553236961365}
+01/29/2022 18:23:04 - INFO - codeparrot_training - Step 22927: {'lr': 0.00029994352786245643, 'samples': 4402176, 'steps': 22927, 'loss/train': 0.9397151470184326}
+01/29/2022 18:23:08 - INFO - codeparrot_training - Step 22928: {'lr': 0.0002999274951286017, 'samples': 4402368, 'steps': 22928, 'loss/train': 1.820344090461731}
+01/29/2022 18:23:13 - INFO - codeparrot_training - Step 22929: {'lr': 0.0002999114621808735, 'samples': 4402560, 'steps': 22929, 'loss/train': 1.874934732913971}
+01/29/2022 18:23:17 - INFO - codeparrot_training - Step 22930: {'lr': 0.0002998954290193405, 'samples': 4402752, 'steps': 22930, 'loss/train': 0.8018859028816223}
+01/29/2022 18:23:22 - INFO - codeparrot_training - Step 22931: {'lr': 0.00029987939564407124, 'samples': 4402944, 'steps': 22931, 'loss/train': 1.581015408039093}
+01/29/2022 18:23:26 - INFO - codeparrot_training - Step 22932: {'lr': 0.00029986336205513456, 'samples': 4403136, 'steps': 22932, 'loss/train': 0.8631349503993988}
+01/29/2022 18:23:30 - INFO - codeparrot_training - Step 22933: {'lr': 0.00029984732825259904, 'samples': 4403328, 'steps': 22933, 'loss/train': 1.6551636457443237}
+01/29/2022 18:23:37 - INFO - codeparrot_training - Step 22934: {'lr': 0.00029983129423653333, 'samples': 4403520, 'steps': 22934, 'loss/train': 1.7456670999526978}
+01/29/2022 18:23:41 - INFO - codeparrot_training - Step 22935: {'lr': 0.00029981526000700626, 'samples': 4403712, 'steps': 22935, 'loss/train': 0.8519167900085449}
+01/29/2022 18:23:46 - INFO - codeparrot_training - Step 22936: {'lr': 0.0002997992255640864, 'samples': 4403904, 'steps': 22936, 'loss/train': 1.9465434551239014}
+01/29/2022 18:23:50 - INFO - codeparrot_training - Step 22937: {'lr': 0.0002997831909078425, 'samples': 4404096, 'steps': 22937, 'loss/train': 1.8705162405967712}
+01/29/2022 18:23:54 - INFO - codeparrot_training - Step 22938: {'lr': 0.00029976715603834315, 'samples': 4404288, 'steps': 22938, 'loss/train': 1.7924251556396484}
+01/29/2022 18:24:00 - INFO - codeparrot_training - Step 22939: {'lr': 0.00029975112095565723, 'samples': 4404480, 'steps': 22939, 'loss/train': 0.8445884585380554}
+01/29/2022 18:24:04 - INFO - codeparrot_training - Step 22940: {'lr': 0.00029973508565985316, 'samples': 4404672, 'steps': 22940, 'loss/train': 2.1401864290237427}
+01/29/2022 18:24:08 - INFO - codeparrot_training - Step 22941: {'lr': 0.0002997190501509999, 'samples': 4404864, 'steps': 22941, 'loss/train': 2.229790985584259}
+01/29/2022 18:24:12 - INFO - codeparrot_training - Step 22942: {'lr': 0.00029970301442916594, 'samples': 4405056, 'steps': 22942, 'loss/train': 1.0542238354682922}
+01/29/2022 18:24:17 - INFO - codeparrot_training - Step 22943: {'lr': 0.00029968697849442006, 'samples': 4405248, 'steps': 22943, 'loss/train': 1.2307583391666412}
+01/29/2022 18:24:22 - INFO - codeparrot_training - Step 22944: {'lr': 0.0002996709423468309, 'samples': 4405440, 'steps': 22944, 'loss/train': 1.6441705226898193}
+01/29/2022 18:24:26 - INFO - codeparrot_training - Step 22945: {'lr': 0.00029965490598646727, 'samples': 4405632, 'steps': 22945, 'loss/train': 2.2551414370536804}
+01/29/2022 18:24:30 - INFO - codeparrot_training - Step 22946: {'lr': 0.00029963886941339774, 'samples': 4405824, 'steps': 22946, 'loss/train': 1.3485913574695587}
+01/29/2022 18:24:35 - INFO - codeparrot_training - Step 22947: {'lr': 0.0002996228326276911, 'samples': 4406016, 'steps': 22947, 'loss/train': 1.8700127005577087}
+01/29/2022 18:24:39 - INFO - codeparrot_training - Step 22948: {'lr': 0.0002996067956294159, 'samples': 4406208, 'steps': 22948, 'loss/train': 0.9711973965167999}
+01/29/2022 18:24:44 - INFO - codeparrot_training - Step 22949: {'lr': 0.0002995907584186411, 'samples': 4406400, 'steps': 22949, 'loss/train': 3.2543052434921265}
+01/29/2022 18:24:48 - INFO - codeparrot_training - Step 22950: {'lr': 0.00029957472099543516, 'samples': 4406592, 'steps': 22950, 'loss/train': 1.555031955242157}
+01/29/2022 18:24:52 - INFO - codeparrot_training - Step 22951: {'lr': 0.00029955868335986686, 'samples': 4406784, 'steps': 22951, 'loss/train': 1.9286640286445618}
+01/29/2022 18:24:57 - INFO - codeparrot_training - Step 22952: {'lr': 0.0002995426455120049, 'samples': 4406976, 'steps': 22952, 'loss/train': 1.962409257888794}
+01/29/2022 18:25:01 - INFO - codeparrot_training - Step 22953: {'lr': 0.0002995266074519179, 'samples': 4407168, 'steps': 22953, 'loss/train': 1.0626579523086548}
+01/29/2022 18:25:08 - INFO - codeparrot_training - Step 22954: {'lr': 0.00029951056917967476, 'samples': 4407360, 'steps': 22954, 'loss/train': 1.4268065392971039}
+01/29/2022 18:25:12 - INFO - codeparrot_training - Step 22955: {'lr': 0.000299494530695344, 'samples': 4407552, 'steps': 22955, 'loss/train': 0.8782104849815369}
+01/29/2022 18:25:16 - INFO - codeparrot_training - Step 22956: {'lr': 0.0002994784919989944, 'samples': 4407744, 'steps': 22956, 'loss/train': 0.9012706875801086}
+01/29/2022 18:25:20 - INFO - codeparrot_training - Step 22957: {'lr': 0.00029946245309069464, 'samples': 4407936, 'steps': 22957, 'loss/train': 2.026245653629303}
+01/29/2022 18:25:25 - INFO - codeparrot_training - Step 22958: {'lr': 0.0002994464139705135, 'samples': 4408128, 'steps': 22958, 'loss/train': 2.137509047985077}
+01/29/2022 18:25:30 - INFO - codeparrot_training - Step 22959: {'lr': 0.00029943037463851953, 'samples': 4408320, 'steps': 22959, 'loss/train': 2.0421212911605835}
+01/29/2022 18:25:34 - INFO - codeparrot_training - Step 22960: {'lr': 0.00029941433509478153, 'samples': 4408512, 'steps': 22960, 'loss/train': 0.18693409860134125}
+01/29/2022 18:25:38 - INFO - codeparrot_training - Step 22961: {'lr': 0.00029939829533936823, 'samples': 4408704, 'steps': 22961, 'loss/train': 2.1646254658699036}
+01/29/2022 18:25:42 - INFO - codeparrot_training - Step 22962: {'lr': 0.00029938225537234843, 'samples': 4408896, 'steps': 22962, 'loss/train': 1.7223967909812927}
+01/29/2022 18:25:47 - INFO - codeparrot_training - Step 22963: {'lr': 0.0002993662151937906, 'samples': 4409088, 'steps': 22963, 'loss/train': 1.6265230178833008}
+01/29/2022 18:25:54 - INFO - codeparrot_training - Step 22964: {'lr': 0.00029935017480376357, 'samples': 4409280, 'steps': 22964, 'loss/train': 2.0113776326179504}
+01/29/2022 18:25:58 - INFO - codeparrot_training - Step 22965: {'lr': 0.00029933413420233615, 'samples': 4409472, 'steps': 22965, 'loss/train': 1.1648696064949036}
+01/29/2022 18:26:02 - INFO - codeparrot_training - Step 22966: {'lr': 0.0002993180933895769, 'samples': 4409664, 'steps': 22966, 'loss/train': 1.8579869270324707}
+01/29/2022 18:26:06 - INFO - codeparrot_training - Step 22967: {'lr': 0.00029930205236555464, 'samples': 4409856, 'steps': 22967, 'loss/train': 2.1379626989364624}
+01/29/2022 18:26:11 - INFO - codeparrot_training - Step 22968: {'lr': 0.000299286011130338, 'samples': 4410048, 'steps': 22968, 'loss/train': 1.8259559869766235}
+01/29/2022 18:26:16 - INFO - codeparrot_training - Step 22969: {'lr': 0.00029926996968399576, 'samples': 4410240, 'steps': 22969, 'loss/train': 1.6146785616874695}
+01/29/2022 18:26:20 - INFO - codeparrot_training - Step 22970: {'lr': 0.0002992539280265966, 'samples': 4410432, 'steps': 22970, 'loss/train': 1.68508780002594}
+01/29/2022 18:26:24 - INFO - codeparrot_training - Step 22971: {'lr': 0.00029923788615820936, 'samples': 4410624, 'steps': 22971, 'loss/train': 1.6271243691444397}
+01/29/2022 18:26:29 - INFO - codeparrot_training - Step 22972: {'lr': 0.0002992218440789025, 'samples': 4410816, 'steps': 22972, 'loss/train': 1.5803046226501465}
+01/29/2022 18:26:33 - INFO - codeparrot_training - Step 22973: {'lr': 0.00029920580178874497, 'samples': 4411008, 'steps': 22973, 'loss/train': 2.113477349281311}
+01/29/2022 18:26:39 - INFO - codeparrot_training - Step 22974: {'lr': 0.00029918975928780537, 'samples': 4411200, 'steps': 22974, 'loss/train': 0.9925240874290466}
+01/29/2022 18:26:44 - INFO - codeparrot_training - Step 22975: {'lr': 0.0002991737165761525, 'samples': 4411392, 'steps': 22975, 'loss/train': 2.055527687072754}
+01/29/2022 18:26:48 - INFO - codeparrot_training - Step 22976: {'lr': 0.000299157673653855, 'samples': 4411584, 'steps': 22976, 'loss/train': 0.9820162653923035}
+01/29/2022 18:26:52 - INFO - codeparrot_training - Step 22977: {'lr': 0.0002991416305209817, 'samples': 4411776, 'steps': 22977, 'loss/train': 1.787769377231598}
+01/29/2022 18:26:56 - INFO - codeparrot_training - Step 22978: {'lr': 0.0002991255871776012, 'samples': 4411968, 'steps': 22978, 'loss/train': 1.492217481136322}
+01/29/2022 18:27:02 - INFO - codeparrot_training - Step 22979: {'lr': 0.0002991095436237823, 'samples': 4412160, 'steps': 22979, 'loss/train': 0.47994978725910187}
+01/29/2022 18:27:06 - INFO - codeparrot_training - Step 22980: {'lr': 0.00029909349985959377, 'samples': 4412352, 'steps': 22980, 'loss/train': 3.39680278301239}
+01/29/2022 18:27:10 - INFO - codeparrot_training - Step 22981: {'lr': 0.00029907745588510416, 'samples': 4412544, 'steps': 22981, 'loss/train': 0.5173554718494415}
+01/29/2022 18:27:15 - INFO - codeparrot_training - Step 22982: {'lr': 0.00029906141170038243, 'samples': 4412736, 'steps': 22982, 'loss/train': 1.4966984689235687}
+01/29/2022 18:27:19 - INFO - codeparrot_training - Step 22983: {'lr': 0.00029904536730549706, 'samples': 4412928, 'steps': 22983, 'loss/train': 0.878593236207962}
+01/29/2022 18:27:25 - INFO - codeparrot_training - Step 22984: {'lr': 0.00029902932270051705, 'samples': 4413120, 'steps': 22984, 'loss/train': 2.097237467765808}
+01/29/2022 18:27:29 - INFO - codeparrot_training - Step 22985: {'lr': 0.00029901327788551087, 'samples': 4413312, 'steps': 22985, 'loss/train': 1.1548473536968231}
+01/29/2022 18:27:33 - INFO - codeparrot_training - Step 22986: {'lr': 0.0002989972328605475, 'samples': 4413504, 'steps': 22986, 'loss/train': 1.7361928224563599}
+01/29/2022 18:27:37 - INFO - codeparrot_training - Step 22987: {'lr': 0.0002989811876256954, 'samples': 4413696, 'steps': 22987, 'loss/train': 0.9563450217247009}
+01/29/2022 18:27:42 - INFO - codeparrot_training - Step 22988: {'lr': 0.0002989651421810235, 'samples': 4413888, 'steps': 22988, 'loss/train': 2.6399272084236145}
+01/29/2022 18:27:48 - INFO - codeparrot_training - Step 22989: {'lr': 0.00029894909652660053, 'samples': 4414080, 'steps': 22989, 'loss/train': 1.6022130846977234}
+01/29/2022 18:27:53 - INFO - codeparrot_training - Step 22990: {'lr': 0.0002989330506624951, 'samples': 4414272, 'steps': 22990, 'loss/train': 1.8688042759895325}
+01/29/2022 18:27:57 - INFO - codeparrot_training - Step 22991: {'lr': 0.0002989170045887761, 'samples': 4414464, 'steps': 22991, 'loss/train': 2.324509084224701}
+01/29/2022 18:28:01 - INFO - codeparrot_training - Step 22992: {'lr': 0.00029890095830551204, 'samples': 4414656, 'steps': 22992, 'loss/train': 1.6287261843681335}
+01/29/2022 18:28:05 - INFO - codeparrot_training - Step 22993: {'lr': 0.00029888491181277195, 'samples': 4414848, 'steps': 22993, 'loss/train': 1.263989120721817}
+01/29/2022 18:28:10 - INFO - codeparrot_training - Step 22994: {'lr': 0.00029886886511062434, 'samples': 4415040, 'steps': 22994, 'loss/train': 1.062888264656067}
+01/29/2022 18:28:15 - INFO - codeparrot_training - Step 22995: {'lr': 0.0002988528181991381, 'samples': 4415232, 'steps': 22995, 'loss/train': 1.923064112663269}
+01/29/2022 18:28:19 - INFO - codeparrot_training - Step 22996: {'lr': 0.00029883677107838183, 'samples': 4415424, 'steps': 22996, 'loss/train': 2.177878439426422}
+01/29/2022 18:28:23 - INFO - codeparrot_training - Step 22997: {'lr': 0.0002988207237484244, 'samples': 4415616, 'steps': 22997, 'loss/train': 1.8306997418403625}
+01/29/2022 18:28:27 - INFO - codeparrot_training - Step 22998: {'lr': 0.0002988046762093344, 'samples': 4415808, 'steps': 22998, 'loss/train': 1.8007375001907349}
+01/29/2022 18:28:33 - INFO - codeparrot_training - Step 22999: {'lr': 0.00029878862846118075, 'samples': 4416000, 'steps': 22999, 'loss/train': 1.5074341893196106}
+01/29/2022 18:28:37 - INFO - codeparrot_training - Step 23000: {'lr': 0.0002987725805040321, 'samples': 4416192, 'steps': 23000, 'loss/train': 0.9491569697856903}
+01/29/2022 18:28:41 - INFO - codeparrot_training - Step 23001: {'lr': 0.00029875653233795715, 'samples': 4416384, 'steps': 23001, 'loss/train': 5.5430556535720825}
+01/29/2022 18:28:46 - INFO - codeparrot_training - Step 23002: {'lr': 0.0002987404839630248, 'samples': 4416576, 'steps': 23002, 'loss/train': 2.128088593482971}
+01/29/2022 18:28:50 - INFO - codeparrot_training - Step 23003: {'lr': 0.00029872443537930357, 'samples': 4416768, 'steps': 23003, 'loss/train': 1.970357358455658}
+01/29/2022 18:28:54 - INFO - codeparrot_training - Step 23004: {'lr': 0.0002987083865868624, 'samples': 4416960, 'steps': 23004, 'loss/train': 1.778695821762085}
+01/29/2022 18:29:00 - INFO - codeparrot_training - Step 23005: {'lr': 0.0002986923375857699, 'samples': 4417152, 'steps': 23005, 'loss/train': 1.370459407567978}
+01/29/2022 18:29:04 - INFO - codeparrot_training - Step 23006: {'lr': 0.00029867628837609503, 'samples': 4417344, 'steps': 23006, 'loss/train': 1.4542391002178192}
+01/29/2022 18:29:08 - INFO - codeparrot_training - Step 23007: {'lr': 0.0002986602389579062, 'samples': 4417536, 'steps': 23007, 'loss/train': 1.631394624710083}
+01/29/2022 18:29:12 - INFO - codeparrot_training - Step 23008: {'lr': 0.0002986441893312726, 'samples': 4417728, 'steps': 23008, 'loss/train': 2.1123728156089783}
+01/29/2022 18:29:19 - INFO - codeparrot_training - Step 23009: {'lr': 0.0002986281394962626, 'samples': 4417920, 'steps': 23009, 'loss/train': 0.7910929620265961}
+01/29/2022 18:29:24 - INFO - codeparrot_training - Step 23010: {'lr': 0.00029861208945294507, 'samples': 4418112, 'steps': 23010, 'loss/train': 1.1339275538921356}
+01/29/2022 18:29:28 - INFO - codeparrot_training - Step 23011: {'lr': 0.00029859603920138876, 'samples': 4418304, 'steps': 23011, 'loss/train': 1.200260192155838}
+01/29/2022 18:29:32 - INFO - codeparrot_training - Step 23012: {'lr': 0.00029857998874166253, 'samples': 4418496, 'steps': 23012, 'loss/train': 1.6534844040870667}
+01/29/2022 18:29:36 - INFO - codeparrot_training - Step 23013: {'lr': 0.00029856393807383504, 'samples': 4418688, 'steps': 23013, 'loss/train': 0.534333810210228}
+01/29/2022 18:29:42 - INFO - codeparrot_training - Step 23014: {'lr': 0.000298547887197975, 'samples': 4418880, 'steps': 23014, 'loss/train': 1.8801698684692383}
+01/29/2022 18:29:46 - INFO - codeparrot_training - Step 23015: {'lr': 0.0002985318361141513, 'samples': 4419072, 'steps': 23015, 'loss/train': 0.6082141846418381}
+01/29/2022 18:29:50 - INFO - codeparrot_training - Step 23016: {'lr': 0.0002985157848224326, 'samples': 4419264, 'steps': 23016, 'loss/train': 3.625856637954712}
+01/29/2022 18:29:54 - INFO - codeparrot_training - Step 23017: {'lr': 0.00029849973332288763, 'samples': 4419456, 'steps': 23017, 'loss/train': 1.5938211679458618}
+01/29/2022 18:29:59 - INFO - codeparrot_training - Step 23018: {'lr': 0.00029848368161558526, 'samples': 4419648, 'steps': 23018, 'loss/train': 1.8360373377799988}
+01/29/2022 18:30:04 - INFO - codeparrot_training - Step 23019: {'lr': 0.0002984676297005942, 'samples': 4419840, 'steps': 23019, 'loss/train': 1.194874495267868}
+01/29/2022 18:30:08 - INFO - codeparrot_training - Step 23020: {'lr': 0.0002984515775779832, 'samples': 4420032, 'steps': 23020, 'loss/train': 2.4280391335487366}
+01/29/2022 18:30:12 - INFO - codeparrot_training - Step 23021: {'lr': 0.00029843552524782104, 'samples': 4420224, 'steps': 23021, 'loss/train': 1.140060156583786}
+01/29/2022 18:30:17 - INFO - codeparrot_training - Step 23022: {'lr': 0.00029841947271017647, 'samples': 4420416, 'steps': 23022, 'loss/train': 1.230110228061676}
+01/29/2022 18:30:21 - INFO - codeparrot_training - Step 23023: {'lr': 0.0002984034199651182, 'samples': 4420608, 'steps': 23023, 'loss/train': 1.8766189813613892}
+01/29/2022 18:30:29 - INFO - codeparrot_training - Step 23024: {'lr': 0.00029838736701271514, 'samples': 4420800, 'steps': 23024, 'loss/train': 2.3643380999565125}
+01/29/2022 18:30:33 - INFO - codeparrot_training - Step 23025: {'lr': 0.00029837131385303587, 'samples': 4420992, 'steps': 23025, 'loss/train': 1.4539784789085388}
+01/29/2022 18:30:37 - INFO - codeparrot_training - Step 23026: {'lr': 0.0002983552604861493, 'samples': 4421184, 'steps': 23026, 'loss/train': 1.1441902220249176}
+01/29/2022 18:30:41 - INFO - codeparrot_training - Step 23027: {'lr': 0.0002983392069121241, 'samples': 4421376, 'steps': 23027, 'loss/train': 1.4763380885124207}
+01/29/2022 18:30:45 - INFO - codeparrot_training - Step 23028: {'lr': 0.00029832315313102915, 'samples': 4421568, 'steps': 23028, 'loss/train': 0.8269141316413879}
+01/29/2022 18:30:50 - INFO - codeparrot_training - Step 23029: {'lr': 0.00029830709914293306, 'samples': 4421760, 'steps': 23029, 'loss/train': 2.123270273208618}
+01/29/2022 18:30:55 - INFO - codeparrot_training - Step 23030: {'lr': 0.00029829104494790483, 'samples': 4421952, 'steps': 23030, 'loss/train': 1.2172421514987946}
+01/29/2022 18:30:59 - INFO - codeparrot_training - Step 23031: {'lr': 0.00029827499054601306, 'samples': 4422144, 'steps': 23031, 'loss/train': 1.655597984790802}
+01/29/2022 18:31:03 - INFO - codeparrot_training - Step 23032: {'lr': 0.0002982589359373265, 'samples': 4422336, 'steps': 23032, 'loss/train': 1.7984713912010193}
+01/29/2022 18:31:07 - INFO - codeparrot_training - Step 23033: {'lr': 0.000298242881121914, 'samples': 4422528, 'steps': 23033, 'loss/train': 1.8164291381835938}
+01/29/2022 18:31:12 - INFO - codeparrot_training - Step 23034: {'lr': 0.00029822682609984436, 'samples': 4422720, 'steps': 23034, 'loss/train': 1.6172073483467102}
+01/29/2022 18:31:19 - INFO - codeparrot_training - Step 23035: {'lr': 0.00029821077087118625, 'samples': 4422912, 'steps': 23035, 'loss/train': 2.2695103883743286}
+01/29/2022 18:31:23 - INFO - codeparrot_training - Step 23036: {'lr': 0.00029819471543600856, 'samples': 4423104, 'steps': 23036, 'loss/train': 2.0083000659942627}
+01/29/2022 18:31:28 - INFO - codeparrot_training - Step 23037: {'lr': 0.00029817865979437996, 'samples': 4423296, 'steps': 23037, 'loss/train': 1.5501455068588257}
+01/29/2022 18:31:32 - INFO - codeparrot_training - Step 23038: {'lr': 0.0002981626039463693, 'samples': 4423488, 'steps': 23038, 'loss/train': 1.0201160609722137}
+01/29/2022 18:31:36 - INFO - codeparrot_training - Step 23039: {'lr': 0.0002981465478920454, 'samples': 4423680, 'steps': 23039, 'loss/train': 2.1868144869804382}
+01/29/2022 18:31:41 - INFO - codeparrot_training - Step 23040: {'lr': 0.00029813049163147687, 'samples': 4423872, 'steps': 23040, 'loss/train': 0.36695674806833267}
+01/29/2022 18:31:46 - INFO - codeparrot_training - Step 23041: {'lr': 0.0002981144351647327, 'samples': 4424064, 'steps': 23041, 'loss/train': 1.1891387701034546}
+01/29/2022 18:31:50 - INFO - codeparrot_training - Step 23042: {'lr': 0.0002980983784918815, 'samples': 4424256, 'steps': 23042, 'loss/train': 2.0426174998283386}
+01/29/2022 18:31:54 - INFO - codeparrot_training - Step 23043: {'lr': 0.0002980823216129921, 'samples': 4424448, 'steps': 23043, 'loss/train': 1.8515739440917969}
+01/29/2022 18:31:58 - INFO - codeparrot_training - Step 23044: {'lr': 0.00029806626452813333, 'samples': 4424640, 'steps': 23044, 'loss/train': 1.966461718082428}
+01/29/2022 18:32:04 - INFO - codeparrot_training - Step 23045: {'lr': 0.000298050207237374, 'samples': 4424832, 'steps': 23045, 'loss/train': 1.403796672821045}
+01/29/2022 18:32:08 - INFO - codeparrot_training - Step 23046: {'lr': 0.0002980341497407828, 'samples': 4425024, 'steps': 23046, 'loss/train': 1.0729482471942902}
+01/29/2022 18:32:12 - INFO - codeparrot_training - Step 23047: {'lr': 0.0002980180920384286, 'samples': 4425216, 'steps': 23047, 'loss/train': 2.042766809463501}
+01/29/2022 18:32:17 - INFO - codeparrot_training - Step 23048: {'lr': 0.00029800203413038, 'samples': 4425408, 'steps': 23048, 'loss/train': 1.2288187444210052}
+01/29/2022 18:32:21 - INFO - codeparrot_training - Step 23049: {'lr': 0.000297985976016706, 'samples': 4425600, 'steps': 23049, 'loss/train': 1.3425982296466827}
+01/29/2022 18:32:26 - INFO - codeparrot_training - Step 23050: {'lr': 0.0002979699176974754, 'samples': 4425792, 'steps': 23050, 'loss/train': 1.9713072180747986}
+01/29/2022 18:32:30 - INFO - codeparrot_training - Step 23051: {'lr': 0.0002979538591727568, 'samples': 4425984, 'steps': 23051, 'loss/train': 1.8561809062957764}
+01/29/2022 18:32:35 - INFO - codeparrot_training - Step 23052: {'lr': 0.00029793780044261916, 'samples': 4426176, 'steps': 23052, 'loss/train': 1.2474626004695892}
+01/29/2022 18:32:39 - INFO - codeparrot_training - Step 23053: {'lr': 0.0002979217415071311, 'samples': 4426368, 'steps': 23053, 'loss/train': 1.851986825466156}
+01/29/2022 18:32:43 - INFO - codeparrot_training - Step 23054: {'lr': 0.00029790568236636166, 'samples': 4426560, 'steps': 23054, 'loss/train': 1.8090705871582031}
+01/29/2022 18:32:50 - INFO - codeparrot_training - Step 23055: {'lr': 0.0002978896230203794, 'samples': 4426752, 'steps': 23055, 'loss/train': 1.6334649324417114}
+01/29/2022 18:32:54 - INFO - codeparrot_training - Step 23056: {'lr': 0.0002978735634692533, 'samples': 4426944, 'steps': 23056, 'loss/train': 1.376938372850418}
+01/29/2022 18:32:58 - INFO - codeparrot_training - Step 23057: {'lr': 0.0002978575037130519, 'samples': 4427136, 'steps': 23057, 'loss/train': 2.3455894589424133}
+01/29/2022 18:33:03 - INFO - codeparrot_training - Step 23058: {'lr': 0.0002978414437518443, 'samples': 4427328, 'steps': 23058, 'loss/train': 1.396323412656784}
+01/29/2022 18:33:07 - INFO - codeparrot_training - Step 23059: {'lr': 0.00029782538358569905, 'samples': 4427520, 'steps': 23059, 'loss/train': 1.0653463304042816}
+01/29/2022 18:33:12 - INFO - codeparrot_training - Step 23060: {'lr': 0.0002978093232146851, 'samples': 4427712, 'steps': 23060, 'loss/train': 0.8534327745437622}
+01/29/2022 18:33:16 - INFO - codeparrot_training - Step 23061: {'lr': 0.00029779326263887113, 'samples': 4427904, 'steps': 23061, 'loss/train': 2.0489278435707092}
+01/29/2022 18:33:21 - INFO - codeparrot_training - Step 23062: {'lr': 0.00029777720185832605, 'samples': 4428096, 'steps': 23062, 'loss/train': 2.8205305337905884}
+01/29/2022 18:33:25 - INFO - codeparrot_training - Step 23063: {'lr': 0.0002977611408731186, 'samples': 4428288, 'steps': 23063, 'loss/train': 2.5672622323036194}
+01/29/2022 18:33:29 - INFO - codeparrot_training - Step 23064: {'lr': 0.0002977450796833176, 'samples': 4428480, 'steps': 23064, 'loss/train': 0.8126356601715088}
+01/29/2022 18:33:35 - INFO - codeparrot_training - Step 23065: {'lr': 0.0002977290182889918, 'samples': 4428672, 'steps': 23065, 'loss/train': 1.7380022406578064}
+01/29/2022 18:33:39 - INFO - codeparrot_training - Step 23066: {'lr': 0.00029771295669021, 'samples': 4428864, 'steps': 23066, 'loss/train': 1.7631163001060486}
+01/29/2022 18:33:44 - INFO - codeparrot_training - Step 23067: {'lr': 0.00029769689488704117, 'samples': 4429056, 'steps': 23067, 'loss/train': 1.384735643863678}
+01/29/2022 18:33:48 - INFO - codeparrot_training - Step 23068: {'lr': 0.00029768083287955394, 'samples': 4429248, 'steps': 23068, 'loss/train': 1.9085624814033508}
+01/29/2022 18:33:52 - INFO - codeparrot_training - Step 23069: {'lr': 0.00029766477066781716, 'samples': 4429440, 'steps': 23069, 'loss/train': 1.8446475863456726}
+01/29/2022 18:33:56 - INFO - codeparrot_training - Step 23070: {'lr': 0.0002976487082518996, 'samples': 4429632, 'steps': 23070, 'loss/train': 1.3955824077129364}
+01/29/2022 18:34:03 - INFO - codeparrot_training - Step 23071: {'lr': 0.00029763264563187016, 'samples': 4429824, 'steps': 23071, 'loss/train': 0.9198881685733795}
+01/29/2022 18:34:08 - INFO - codeparrot_training - Step 23072: {'lr': 0.0002976165828077975, 'samples': 4430016, 'steps': 23072, 'loss/train': 1.3852568864822388}
+01/29/2022 18:34:12 - INFO - codeparrot_training - Step 23073: {'lr': 0.00029760051977975057, 'samples': 4430208, 'steps': 23073, 'loss/train': 1.3958100378513336}
+01/29/2022 18:34:16 - INFO - codeparrot_training - Step 23074: {'lr': 0.00029758445654779814, 'samples': 4430400, 'steps': 23074, 'loss/train': 2.0058860778808594}
+01/29/2022 18:34:20 - INFO - codeparrot_training - Step 23075: {'lr': 0.000297568393112009, 'samples': 4430592, 'steps': 23075, 'loss/train': 1.4790295958518982}
+01/29/2022 18:34:25 - INFO - codeparrot_training - Step 23076: {'lr': 0.00029755232947245195, 'samples': 4430784, 'steps': 23076, 'loss/train': 1.613834023475647}
+01/29/2022 18:34:30 - INFO - codeparrot_training - Step 23077: {'lr': 0.0002975362656291958, 'samples': 4430976, 'steps': 23077, 'loss/train': 0.9335027933120728}
+01/29/2022 18:34:34 - INFO - codeparrot_training - Step 23078: {'lr': 0.0002975202015823095, 'samples': 4431168, 'steps': 23078, 'loss/train': 1.6172756552696228}
+01/29/2022 18:34:38 - INFO - codeparrot_training - Step 23079: {'lr': 0.0002975041373318617, 'samples': 4431360, 'steps': 23079, 'loss/train': 2.1997944116592407}
+01/29/2022 18:34:42 - INFO - codeparrot_training - Step 23080: {'lr': 0.0002974880728779212, 'samples': 4431552, 'steps': 23080, 'loss/train': 1.6990166902542114}
+01/29/2022 18:34:49 - INFO - codeparrot_training - Step 23081: {'lr': 0.00029747200822055684, 'samples': 4431744, 'steps': 23081, 'loss/train': 1.6540609002113342}
+01/29/2022 18:34:54 - INFO - codeparrot_training - Step 23082: {'lr': 0.0002974559433598376, 'samples': 4431936, 'steps': 23082, 'loss/train': 0.5006837099790573}
+01/29/2022 18:34:58 - INFO - codeparrot_training - Step 23083: {'lr': 0.0002974398782958321, 'samples': 4432128, 'steps': 23083, 'loss/train': 1.4754895269870758}
+01/29/2022 18:35:02 - INFO - codeparrot_training - Step 23084: {'lr': 0.00029742381302860923, 'samples': 4432320, 'steps': 23084, 'loss/train': 1.0189997255802155}
+01/29/2022 18:35:06 - INFO - codeparrot_training - Step 23085: {'lr': 0.00029740774755823777, 'samples': 4432512, 'steps': 23085, 'loss/train': 1.7615439891815186}
+01/29/2022 18:35:10 - INFO - codeparrot_training - Step 23086: {'lr': 0.0002973916818847866, 'samples': 4432704, 'steps': 23086, 'loss/train': 1.0611320436000824}
+01/29/2022 18:35:16 - INFO - codeparrot_training - Step 23087: {'lr': 0.00029737561600832454, 'samples': 4432896, 'steps': 23087, 'loss/train': 1.8515342473983765}
+01/29/2022 18:35:20 - INFO - codeparrot_training - Step 23088: {'lr': 0.00029735954992892035, 'samples': 4433088, 'steps': 23088, 'loss/train': 1.1151388585567474}
+01/29/2022 18:35:24 - INFO - codeparrot_training - Step 23089: {'lr': 0.00029734348364664285, 'samples': 4433280, 'steps': 23089, 'loss/train': 0.4310491383075714}
+01/29/2022 18:35:28 - INFO - codeparrot_training - Step 23090: {'lr': 0.000297327417161561, 'samples': 4433472, 'steps': 23090, 'loss/train': 2.0430171489715576}
+01/29/2022 18:35:33 - INFO - codeparrot_training - Step 23091: {'lr': 0.0002973113504737435, 'samples': 4433664, 'steps': 23091, 'loss/train': 1.6617136001586914}
+01/29/2022 18:35:38 - INFO - codeparrot_training - Step 23092: {'lr': 0.00029729528358325914, 'samples': 4433856, 'steps': 23092, 'loss/train': 3.135055661201477}
+01/29/2022 18:35:42 - INFO - codeparrot_training - Step 23093: {'lr': 0.00029727921649017687, 'samples': 4434048, 'steps': 23093, 'loss/train': 1.7480605244636536}
+01/29/2022 18:35:46 - INFO - codeparrot_training - Step 23094: {'lr': 0.0002972631491945653, 'samples': 4434240, 'steps': 23094, 'loss/train': 0.6088943481445312}
+01/29/2022 18:35:51 - INFO - codeparrot_training - Step 23095: {'lr': 0.00029724708169649364, 'samples': 4434432, 'steps': 23095, 'loss/train': 1.4190166890621185}
+01/29/2022 18:35:55 - INFO - codeparrot_training - Step 23096: {'lr': 0.0002972310139960303, 'samples': 4434624, 'steps': 23096, 'loss/train': 1.4016354382038116}
+01/29/2022 18:36:02 - INFO - codeparrot_training - Step 23097: {'lr': 0.00029721494609324435, 'samples': 4434816, 'steps': 23097, 'loss/train': 2.0085148215293884}
+01/29/2022 18:36:06 - INFO - codeparrot_training - Step 23098: {'lr': 0.00029719887798820453, 'samples': 4435008, 'steps': 23098, 'loss/train': 0.6903316229581833}
+01/29/2022 18:36:10 - INFO - codeparrot_training - Step 23099: {'lr': 0.00029718280968097976, 'samples': 4435200, 'steps': 23099, 'loss/train': 1.5519681572914124}
+01/29/2022 18:36:15 - INFO - codeparrot_training - Step 23100: {'lr': 0.00029716674117163884, 'samples': 4435392, 'steps': 23100, 'loss/train': 1.7665636539459229}
+01/29/2022 18:36:19 - INFO - codeparrot_training - Step 23101: {'lr': 0.0002971506724602505, 'samples': 4435584, 'steps': 23101, 'loss/train': 2.001471519470215}
+01/29/2022 18:36:24 - INFO - codeparrot_training - Step 23102: {'lr': 0.0002971346035468837, 'samples': 4435776, 'steps': 23102, 'loss/train': 2.2460432052612305}
+01/29/2022 18:36:28 - INFO - codeparrot_training - Step 23103: {'lr': 0.0002971185344316072, 'samples': 4435968, 'steps': 23103, 'loss/train': 1.6324945092201233}
+01/29/2022 18:36:32 - INFO - codeparrot_training - Step 23104: {'lr': 0.0002971024651144899, 'samples': 4436160, 'steps': 23104, 'loss/train': 1.6673092246055603}
+01/29/2022 18:36:37 - INFO - codeparrot_training - Step 23105: {'lr': 0.0002970863955956005, 'samples': 4436352, 'steps': 23105, 'loss/train': 1.6129724979400635}
+01/29/2022 18:36:41 - INFO - codeparrot_training - Step 23106: {'lr': 0.00029707032587500805, 'samples': 4436544, 'steps': 23106, 'loss/train': 1.5977883338928223}
+01/29/2022 18:36:47 - INFO - codeparrot_training - Step 23107: {'lr': 0.00029705425595278126, 'samples': 4436736, 'steps': 23107, 'loss/train': 1.5122979283332825}
+01/29/2022 18:36:51 - INFO - codeparrot_training - Step 23108: {'lr': 0.0002970381858289889, 'samples': 4436928, 'steps': 23108, 'loss/train': 1.386611670255661}
+01/29/2022 18:36:55 - INFO - codeparrot_training - Step 23109: {'lr': 0.0002970221155037, 'samples': 4437120, 'steps': 23109, 'loss/train': 2.368249475955963}
+01/29/2022 18:36:59 - INFO - codeparrot_training - Step 23110: {'lr': 0.0002970060449769832, 'samples': 4437312, 'steps': 23110, 'loss/train': 1.9223610162734985}
+01/29/2022 18:37:03 - INFO - codeparrot_training - Step 23111: {'lr': 0.00029698997424890746, 'samples': 4437504, 'steps': 23111, 'loss/train': 1.6111754179000854}
+01/29/2022 18:37:08 - INFO - codeparrot_training - Step 23112: {'lr': 0.0002969739033195415, 'samples': 4437696, 'steps': 23112, 'loss/train': 1.9728158712387085}
+01/29/2022 18:37:14 - INFO - codeparrot_training - Step 23113: {'lr': 0.0002969578321889544, 'samples': 4437888, 'steps': 23113, 'loss/train': 1.6741247177124023}
+01/29/2022 18:37:18 - INFO - codeparrot_training - Step 23114: {'lr': 0.00029694176085721474, 'samples': 4438080, 'steps': 23114, 'loss/train': 1.405007243156433}
+01/29/2022 18:37:22 - INFO - codeparrot_training - Step 23115: {'lr': 0.0002969256893243916, 'samples': 4438272, 'steps': 23115, 'loss/train': 1.5891265869140625}
+01/29/2022 18:37:27 - INFO - codeparrot_training - Step 23116: {'lr': 0.0002969096175905536, 'samples': 4438464, 'steps': 23116, 'loss/train': 2.036029100418091}
+01/29/2022 18:37:31 - INFO - codeparrot_training - Step 23117: {'lr': 0.00029689354565576976, 'samples': 4438656, 'steps': 23117, 'loss/train': 2.907109558582306}
+01/29/2022 18:37:36 - INFO - codeparrot_training - Step 23118: {'lr': 0.0002968774735201088, 'samples': 4438848, 'steps': 23118, 'loss/train': 1.739692747592926}
+01/29/2022 18:37:40 - INFO - codeparrot_training - Step 23119: {'lr': 0.0002968614011836397, 'samples': 4439040, 'steps': 23119, 'loss/train': 0.09935282170772552}
+01/29/2022 18:37:45 - INFO - codeparrot_training - Step 23120: {'lr': 0.0002968453286464312, 'samples': 4439232, 'steps': 23120, 'loss/train': 1.437962919473648}
+01/29/2022 18:37:49 - INFO - codeparrot_training - Step 23121: {'lr': 0.00029682925590855213, 'samples': 4439424, 'steps': 23121, 'loss/train': 1.4750226438045502}
+01/29/2022 18:37:53 - INFO - codeparrot_training - Step 23122: {'lr': 0.0002968131829700715, 'samples': 4439616, 'steps': 23122, 'loss/train': 0.8548922538757324}
+01/29/2022 18:37:58 - INFO - codeparrot_training - Step 23123: {'lr': 0.000296797109831058, 'samples': 4439808, 'steps': 23123, 'loss/train': 1.9634762406349182}
+01/29/2022 18:38:03 - INFO - codeparrot_training - Step 23124: {'lr': 0.00029678103649158057, 'samples': 4440000, 'steps': 23124, 'loss/train': 2.0205230712890625}
+01/29/2022 18:38:07 - INFO - codeparrot_training - Step 23125: {'lr': 0.00029676496295170804, 'samples': 4440192, 'steps': 23125, 'loss/train': 1.1049490571022034}
+01/29/2022 18:38:11 - INFO - codeparrot_training - Step 23126: {'lr': 0.0002967488892115092, 'samples': 4440384, 'steps': 23126, 'loss/train': 0.7669596076011658}
+01/29/2022 18:38:15 - INFO - codeparrot_training - Step 23127: {'lr': 0.000296732815271053, 'samples': 4440576, 'steps': 23127, 'loss/train': 1.0560089349746704}
+01/29/2022 18:38:23 - INFO - codeparrot_training - Step 23128: {'lr': 0.00029671674113040833, 'samples': 4440768, 'steps': 23128, 'loss/train': 1.059216320514679}
+01/29/2022 18:38:27 - INFO - codeparrot_training - Step 23129: {'lr': 0.00029670066678964385, 'samples': 4440960, 'steps': 23129, 'loss/train': 2.0115979313850403}
+01/29/2022 18:38:31 - INFO - codeparrot_training - Step 23130: {'lr': 0.0002966845922488286, 'samples': 4441152, 'steps': 23130, 'loss/train': 1.3042857348918915}
+01/29/2022 18:38:35 - INFO - codeparrot_training - Step 23131: {'lr': 0.00029666851750803137, 'samples': 4441344, 'steps': 23131, 'loss/train': 1.5222651958465576}
+01/29/2022 18:38:39 - INFO - codeparrot_training - Step 23132: {'lr': 0.00029665244256732107, 'samples': 4441536, 'steps': 23132, 'loss/train': 1.9952389597892761}
+01/29/2022 18:38:45 - INFO - codeparrot_training - Step 23133: {'lr': 0.0002966363674267665, 'samples': 4441728, 'steps': 23133, 'loss/train': 1.5430419445037842}
+01/29/2022 18:38:49 - INFO - codeparrot_training - Step 23134: {'lr': 0.00029662029208643646, 'samples': 4441920, 'steps': 23134, 'loss/train': 6.626663446426392}
+01/29/2022 18:38:53 - INFO - codeparrot_training - Step 23135: {'lr': 0.0002966042165464, 'samples': 4442112, 'steps': 23135, 'loss/train': 1.9223134517669678}
+01/29/2022 18:38:57 - INFO - codeparrot_training - Step 23136: {'lr': 0.0002965881408067258, 'samples': 4442304, 'steps': 23136, 'loss/train': 1.1105023920536041}
+01/29/2022 18:39:03 - INFO - codeparrot_training - Step 23137: {'lr': 0.0002965720648674829, 'samples': 4442496, 'steps': 23137, 'loss/train': 1.785583734512329}
+01/29/2022 18:39:07 - INFO - codeparrot_training - Step 23138: {'lr': 0.00029655598872873994, 'samples': 4442688, 'steps': 23138, 'loss/train': 1.8721386194229126}
+01/29/2022 18:39:11 - INFO - codeparrot_training - Step 23139: {'lr': 0.000296539912390566, 'samples': 4442880, 'steps': 23139, 'loss/train': 1.4665687680244446}
+01/29/2022 18:39:15 - INFO - codeparrot_training - Step 23140: {'lr': 0.0002965238358530298, 'samples': 4443072, 'steps': 23140, 'loss/train': 1.5840913653373718}
+01/29/2022 18:39:20 - INFO - codeparrot_training - Step 23141: {'lr': 0.0002965077591162003, 'samples': 4443264, 'steps': 23141, 'loss/train': 3.4325594902038574}
+01/29/2022 18:39:24 - INFO - codeparrot_training - Step 23142: {'lr': 0.00029649168218014627, 'samples': 4443456, 'steps': 23142, 'loss/train': 1.5362436175346375}
+01/29/2022 18:39:31 - INFO - codeparrot_training - Step 23143: {'lr': 0.0002964756050449367, 'samples': 4443648, 'steps': 23143, 'loss/train': 1.7773287892341614}
+01/29/2022 18:39:35 - INFO - codeparrot_training - Step 23144: {'lr': 0.0002964595277106403, 'samples': 4443840, 'steps': 23144, 'loss/train': 1.7885060906410217}
+01/29/2022 18:39:40 - INFO - codeparrot_training - Step 23145: {'lr': 0.0002964434501773262, 'samples': 4444032, 'steps': 23145, 'loss/train': 1.2465850710868835}
+01/29/2022 18:39:44 - INFO - codeparrot_training - Step 23146: {'lr': 0.00029642737244506295, 'samples': 4444224, 'steps': 23146, 'loss/train': 1.2899384200572968}
+01/29/2022 18:39:48 - INFO - codeparrot_training - Step 23147: {'lr': 0.0002964112945139196, 'samples': 4444416, 'steps': 23147, 'loss/train': 1.8332045674324036}
+01/29/2022 18:39:53 - INFO - codeparrot_training - Step 23148: {'lr': 0.0002963952163839651, 'samples': 4444608, 'steps': 23148, 'loss/train': 1.7137954831123352}
+01/29/2022 18:39:58 - INFO - codeparrot_training - Step 23149: {'lr': 0.00029637913805526816, 'samples': 4444800, 'steps': 23149, 'loss/train': 2.582074463367462}
+01/29/2022 18:40:02 - INFO - codeparrot_training - Step 23150: {'lr': 0.0002963630595278977, 'samples': 4444992, 'steps': 23150, 'loss/train': 1.6220165491104126}
+01/29/2022 18:40:06 - INFO - codeparrot_training - Step 23151: {'lr': 0.0002963469808019227, 'samples': 4445184, 'steps': 23151, 'loss/train': 1.0220195353031158}
+01/29/2022 18:40:10 - INFO - codeparrot_training - Step 23152: {'lr': 0.00029633090187741187, 'samples': 4445376, 'steps': 23152, 'loss/train': 1.4293614327907562}
+01/29/2022 18:40:16 - INFO - codeparrot_training - Step 23153: {'lr': 0.00029631482275443424, 'samples': 4445568, 'steps': 23153, 'loss/train': 1.9287464618682861}
+01/29/2022 18:40:20 - INFO - codeparrot_training - Step 23154: {'lr': 0.0002962987434330585, 'samples': 4445760, 'steps': 23154, 'loss/train': 1.8952546119689941}
+01/29/2022 18:40:24 - INFO - codeparrot_training - Step 23155: {'lr': 0.0002962826639133538, 'samples': 4445952, 'steps': 23155, 'loss/train': 1.8633751273155212}
+01/29/2022 18:40:28 - INFO - codeparrot_training - Step 23156: {'lr': 0.00029626658419538873, 'samples': 4446144, 'steps': 23156, 'loss/train': 2.079039216041565}
+01/29/2022 18:40:32 - INFO - codeparrot_training - Step 23157: {'lr': 0.0002962505042792324, 'samples': 4446336, 'steps': 23157, 'loss/train': 3.2094790935516357}
+01/29/2022 18:40:39 - INFO - codeparrot_training - Step 23158: {'lr': 0.0002962344241649535, 'samples': 4446528, 'steps': 23158, 'loss/train': 2.5170240998268127}
+01/29/2022 18:40:44 - INFO - codeparrot_training - Step 23159: {'lr': 0.000296218343852621, 'samples': 4446720, 'steps': 23159, 'loss/train': 2.5689695477485657}
+01/29/2022 18:40:48 - INFO - codeparrot_training - Step 23160: {'lr': 0.0002962022633423039, 'samples': 4446912, 'steps': 23160, 'loss/train': 1.6599812507629395}
+01/29/2022 18:40:52 - INFO - codeparrot_training - Step 23161: {'lr': 0.00029618618263407094, 'samples': 4447104, 'steps': 23161, 'loss/train': 1.6597378849983215}
+01/29/2022 18:40:56 - INFO - codeparrot_training - Step 23162: {'lr': 0.00029617010172799095, 'samples': 4447296, 'steps': 23162, 'loss/train': 0.7564560770988464}
+01/29/2022 18:41:01 - INFO - codeparrot_training - Step 23163: {'lr': 0.00029615402062413307, 'samples': 4447488, 'steps': 23163, 'loss/train': 1.0986165404319763}
+01/29/2022 18:41:06 - INFO - codeparrot_training - Step 23164: {'lr': 0.00029613793932256583, 'samples': 4447680, 'steps': 23164, 'loss/train': 0.9762145578861237}
+01/29/2022 18:41:10 - INFO - codeparrot_training - Step 23165: {'lr': 0.00029612185782335845, 'samples': 4447872, 'steps': 23165, 'loss/train': 1.694276511669159}
+01/29/2022 18:41:14 - INFO - codeparrot_training - Step 23166: {'lr': 0.00029610577612657963, 'samples': 4448064, 'steps': 23166, 'loss/train': 2.2981384992599487}
+01/29/2022 18:41:18 - INFO - codeparrot_training - Step 23167: {'lr': 0.0002960896942322983, 'samples': 4448256, 'steps': 23167, 'loss/train': 2.0583161115646362}
+01/29/2022 18:41:25 - INFO - codeparrot_training - Step 23168: {'lr': 0.0002960736121405834, 'samples': 4448448, 'steps': 23168, 'loss/train': 1.8880535960197449}
+01/29/2022 18:41:30 - INFO - codeparrot_training - Step 23169: {'lr': 0.00029605752985150367, 'samples': 4448640, 'steps': 23169, 'loss/train': 0.3782931715250015}
+01/29/2022 18:41:34 - INFO - codeparrot_training - Step 23170: {'lr': 0.00029604144736512816, 'samples': 4448832, 'steps': 23170, 'loss/train': 2.4504014253616333}
+01/29/2022 18:41:38 - INFO - codeparrot_training - Step 23171: {'lr': 0.00029602536468152575, 'samples': 4449024, 'steps': 23171, 'loss/train': 1.1112326681613922}
+01/29/2022 18:41:42 - INFO - codeparrot_training - Step 23172: {'lr': 0.00029600928180076525, 'samples': 4449216, 'steps': 23172, 'loss/train': 1.1164332926273346}
+01/29/2022 18:41:48 - INFO - codeparrot_training - Step 23173: {'lr': 0.0002959931987229156, 'samples': 4449408, 'steps': 23173, 'loss/train': 1.7164258360862732}
+01/29/2022 18:41:52 - INFO - codeparrot_training - Step 23174: {'lr': 0.0002959771154480457, 'samples': 4449600, 'steps': 23174, 'loss/train': 2.0142742395401}
+01/29/2022 18:41:56 - INFO - codeparrot_training - Step 23175: {'lr': 0.0002959610319762244, 'samples': 4449792, 'steps': 23175, 'loss/train': 1.7904317378997803}
+01/29/2022 18:42:00 - INFO - codeparrot_training - Step 23176: {'lr': 0.0002959449483075207, 'samples': 4449984, 'steps': 23176, 'loss/train': 1.7117793560028076}
+01/29/2022 18:42:05 - INFO - codeparrot_training - Step 23177: {'lr': 0.0002959288644420034, 'samples': 4450176, 'steps': 23177, 'loss/train': 1.7281857132911682}
+01/29/2022 18:42:10 - INFO - codeparrot_training - Step 23178: {'lr': 0.00029591278037974147, 'samples': 4450368, 'steps': 23178, 'loss/train': 1.9885963797569275}
+01/29/2022 18:42:14 - INFO - codeparrot_training - Step 23179: {'lr': 0.0002958966961208037, 'samples': 4450560, 'steps': 23179, 'loss/train': 1.896978735923767}
+01/29/2022 18:42:18 - INFO - codeparrot_training - Step 23180: {'lr': 0.0002958806116652591, 'samples': 4450752, 'steps': 23180, 'loss/train': 2.2111555337905884}
+01/29/2022 18:42:22 - INFO - codeparrot_training - Step 23181: {'lr': 0.0002958645270131765, 'samples': 4450944, 'steps': 23181, 'loss/train': 1.2742020785808563}
+01/29/2022 18:42:27 - INFO - codeparrot_training - Step 23182: {'lr': 0.0002958484421646248, 'samples': 4451136, 'steps': 23182, 'loss/train': 0.9614979028701782}
+01/29/2022 18:42:32 - INFO - codeparrot_training - Step 23183: {'lr': 0.000295832357119673, 'samples': 4451328, 'steps': 23183, 'loss/train': 1.1897641718387604}
+01/29/2022 18:42:36 - INFO - codeparrot_training - Step 23184: {'lr': 0.00029581627187838993, 'samples': 4451520, 'steps': 23184, 'loss/train': 1.7033300399780273}
+01/29/2022 18:42:40 - INFO - codeparrot_training - Step 23185: {'lr': 0.0002958001864408445, 'samples': 4451712, 'steps': 23185, 'loss/train': 2.6749479174613953}
+01/29/2022 18:42:44 - INFO - codeparrot_training - Step 23186: {'lr': 0.0002957841008071056, 'samples': 4451904, 'steps': 23186, 'loss/train': 1.611595094203949}
+01/29/2022 18:42:49 - INFO - codeparrot_training - Step 23187: {'lr': 0.0002957680149772422, 'samples': 4452096, 'steps': 23187, 'loss/train': 0.3499368578195572}
+01/29/2022 18:42:56 - INFO - codeparrot_training - Step 23188: {'lr': 0.000295751928951323, 'samples': 4452288, 'steps': 23188, 'loss/train': 1.6294294595718384}
+01/29/2022 18:43:00 - INFO - codeparrot_training - Step 23189: {'lr': 0.0002957358427294172, 'samples': 4452480, 'steps': 23189, 'loss/train': 1.1983872950077057}
+01/29/2022 18:43:04 - INFO - codeparrot_training - Step 23190: {'lr': 0.00029571975631159355, 'samples': 4452672, 'steps': 23190, 'loss/train': 0.7314898073673248}
+01/29/2022 18:43:08 - INFO - codeparrot_training - Step 23191: {'lr': 0.000295703669697921, 'samples': 4452864, 'steps': 23191, 'loss/train': 0.9694561958312988}
+01/29/2022 18:43:12 - INFO - codeparrot_training - Step 23192: {'lr': 0.0002956875828884684, 'samples': 4453056, 'steps': 23192, 'loss/train': 0.7855068147182465}
+01/29/2022 18:43:18 - INFO - codeparrot_training - Step 23193: {'lr': 0.0002956714958833047, 'samples': 4453248, 'steps': 23193, 'loss/train': 1.4939375817775726}
+01/29/2022 18:43:22 - INFO - codeparrot_training - Step 23194: {'lr': 0.00029565540868249884, 'samples': 4453440, 'steps': 23194, 'loss/train': 1.352573812007904}
+01/29/2022 18:43:26 - INFO - codeparrot_training - Step 23195: {'lr': 0.0002956393212861197, 'samples': 4453632, 'steps': 23195, 'loss/train': 2.3461192846298218}
+01/29/2022 18:43:30 - INFO - codeparrot_training - Step 23196: {'lr': 0.0002956232336942362, 'samples': 4453824, 'steps': 23196, 'loss/train': 1.3016488552093506}
+01/29/2022 18:43:36 - INFO - codeparrot_training - Step 23197: {'lr': 0.0002956071459069173, 'samples': 4454016, 'steps': 23197, 'loss/train': 1.7917349338531494}
+01/29/2022 18:43:40 - INFO - codeparrot_training - Step 23198: {'lr': 0.0002955910579242318, 'samples': 4454208, 'steps': 23198, 'loss/train': 2.090689480304718}
+01/29/2022 18:43:44 - INFO - codeparrot_training - Step 23199: {'lr': 0.0002955749697462487, 'samples': 4454400, 'steps': 23199, 'loss/train': 1.257991075515747}
+01/29/2022 18:43:48 - INFO - codeparrot_training - Step 23200: {'lr': 0.0002955588813730369, 'samples': 4454592, 'steps': 23200, 'loss/train': 2.010055661201477}
+01/29/2022 18:43:52 - INFO - codeparrot_training - Step 23201: {'lr': 0.0002955427928046653, 'samples': 4454784, 'steps': 23201, 'loss/train': 1.7041840553283691}
+01/29/2022 18:43:57 - INFO - codeparrot_training - Step 23202: {'lr': 0.0002955267040412029, 'samples': 4454976, 'steps': 23202, 'loss/train': 1.703305184841156}
+01/29/2022 18:44:02 - INFO - codeparrot_training - Step 23203: {'lr': 0.0002955106150827185, 'samples': 4455168, 'steps': 23203, 'loss/train': 2.1972561478614807}
+01/29/2022 18:44:07 - INFO - codeparrot_training - Step 23204: {'lr': 0.0002954945259292811, 'samples': 4455360, 'steps': 23204, 'loss/train': 0.8391433954238892}
+01/29/2022 18:44:11 - INFO - codeparrot_training - Step 23205: {'lr': 0.0002954784365809596, 'samples': 4455552, 'steps': 23205, 'loss/train': 2.0140337347984314}
+01/29/2022 18:44:15 - INFO - codeparrot_training - Step 23206: {'lr': 0.0002954623470378229, 'samples': 4455744, 'steps': 23206, 'loss/train': 0.3081681579351425}
+01/29/2022 18:44:19 - INFO - codeparrot_training - Step 23207: {'lr': 0.00029544625729993997, 'samples': 4455936, 'steps': 23207, 'loss/train': 1.7021459341049194}
+01/29/2022 18:44:27 - INFO - codeparrot_training - Step 23208: {'lr': 0.0002954301673673797, 'samples': 4456128, 'steps': 23208, 'loss/train': 1.9546167254447937}
+01/29/2022 18:44:31 - INFO - codeparrot_training - Step 23209: {'lr': 0.00029541407724021095, 'samples': 4456320, 'steps': 23209, 'loss/train': 1.6733815670013428}
+01/29/2022 18:44:36 - INFO - codeparrot_training - Step 23210: {'lr': 0.00029539798691850274, 'samples': 4456512, 'steps': 23210, 'loss/train': 0.9961519539356232}
+01/29/2022 18:44:40 - INFO - codeparrot_training - Step 23211: {'lr': 0.00029538189640232406, 'samples': 4456704, 'steps': 23211, 'loss/train': 1.0113953053951263}
+01/29/2022 18:44:44 - INFO - codeparrot_training - Step 23212: {'lr': 0.00029536580569174364, 'samples': 4456896, 'steps': 23212, 'loss/train': 1.078325629234314}
+01/29/2022 18:44:49 - INFO - codeparrot_training - Step 23213: {'lr': 0.0002953497147868306, 'samples': 4457088, 'steps': 23213, 'loss/train': 1.4259792566299438}
+01/29/2022 18:44:53 - INFO - codeparrot_training - Step 23214: {'lr': 0.0002953336236876537, 'samples': 4457280, 'steps': 23214, 'loss/train': 2.5550538897514343}
+01/29/2022 18:44:58 - INFO - codeparrot_training - Step 23215: {'lr': 0.000295317532394282, 'samples': 4457472, 'steps': 23215, 'loss/train': 0.9792680740356445}
+01/29/2022 18:45:02 - INFO - codeparrot_training - Step 23216: {'lr': 0.0002953014409067844, 'samples': 4457664, 'steps': 23216, 'loss/train': 1.2682044804096222}
+01/29/2022 18:45:06 - INFO - codeparrot_training - Step 23217: {'lr': 0.00029528534922522974, 'samples': 4457856, 'steps': 23217, 'loss/train': 1.8119970560073853}
+01/29/2022 18:45:14 - INFO - codeparrot_training - Step 23218: {'lr': 0.00029526925734968705, 'samples': 4458048, 'steps': 23218, 'loss/train': 1.2626051902770996}
+01/29/2022 18:45:18 - INFO - codeparrot_training - Step 23219: {'lr': 0.0002952531652802252, 'samples': 4458240, 'steps': 23219, 'loss/train': 1.4585267901420593}
+01/29/2022 18:45:22 - INFO - codeparrot_training - Step 23220: {'lr': 0.00029523707301691327, 'samples': 4458432, 'steps': 23220, 'loss/train': 1.9848253726959229}
+01/29/2022 18:45:26 - INFO - codeparrot_training - Step 23221: {'lr': 0.00029522098055982, 'samples': 4458624, 'steps': 23221, 'loss/train': 2.4584553837776184}
+01/29/2022 18:45:30 - INFO - codeparrot_training - Step 23222: {'lr': 0.00029520488790901446, 'samples': 4458816, 'steps': 23222, 'loss/train': 1.193576842546463}
+01/29/2022 18:45:36 - INFO - codeparrot_training - Step 23223: {'lr': 0.00029518879506456556, 'samples': 4459008, 'steps': 23223, 'loss/train': 1.2331117987632751}
+01/29/2022 18:45:40 - INFO - codeparrot_training - Step 23224: {'lr': 0.0002951727020265421, 'samples': 4459200, 'steps': 23224, 'loss/train': 1.643989384174347}
+01/29/2022 18:45:44 - INFO - codeparrot_training - Step 23225: {'lr': 0.0002951566087950132, 'samples': 4459392, 'steps': 23225, 'loss/train': 1.7927902936935425}
+01/29/2022 18:45:49 - INFO - codeparrot_training - Step 23226: {'lr': 0.0002951405153700477, 'samples': 4459584, 'steps': 23226, 'loss/train': 1.6314463019371033}
+01/29/2022 18:45:53 - INFO - codeparrot_training - Step 23227: {'lr': 0.0002951244217517146, 'samples': 4459776, 'steps': 23227, 'loss/train': 1.9097882509231567}
+01/29/2022 18:45:58 - INFO - codeparrot_training - Step 23228: {'lr': 0.0002951083279400828, 'samples': 4459968, 'steps': 23228, 'loss/train': 1.0329337120056152}
+01/29/2022 18:46:02 - INFO - codeparrot_training - Step 23229: {'lr': 0.00029509223393522124, 'samples': 4460160, 'steps': 23229, 'loss/train': 0.8805965781211853}
+01/29/2022 18:46:07 - INFO - codeparrot_training - Step 23230: {'lr': 0.00029507613973719883, 'samples': 4460352, 'steps': 23230, 'loss/train': 1.7975038290023804}
+01/29/2022 18:46:11 - INFO - codeparrot_training - Step 23231: {'lr': 0.0002950600453460846, 'samples': 4460544, 'steps': 23231, 'loss/train': 0.7211261093616486}
+01/29/2022 18:46:15 - INFO - codeparrot_training - Step 23232: {'lr': 0.0002950439507619474, 'samples': 4460736, 'steps': 23232, 'loss/train': 0.39314796030521393}
+01/29/2022 18:46:22 - INFO - codeparrot_training - Step 23233: {'lr': 0.00029502785598485624, 'samples': 4460928, 'steps': 23233, 'loss/train': 0.741805762052536}
+01/29/2022 18:46:26 - INFO - codeparrot_training - Step 23234: {'lr': 0.00029501176101488004, 'samples': 4461120, 'steps': 23234, 'loss/train': 1.6169832944869995}
+01/29/2022 18:46:31 - INFO - codeparrot_training - Step 23235: {'lr': 0.00029499566585208776, 'samples': 4461312, 'steps': 23235, 'loss/train': 1.8583529591560364}
+01/29/2022 18:46:35 - INFO - codeparrot_training - Step 23236: {'lr': 0.00029497957049654834, 'samples': 4461504, 'steps': 23236, 'loss/train': 1.118738830089569}
+01/29/2022 18:46:39 - INFO - codeparrot_training - Step 23237: {'lr': 0.0002949634749483307, 'samples': 4461696, 'steps': 23237, 'loss/train': 0.2517586424946785}
+01/29/2022 18:46:44 - INFO - codeparrot_training - Step 23238: {'lr': 0.0002949473792075039, 'samples': 4461888, 'steps': 23238, 'loss/train': 1.4251549243927002}
+01/29/2022 18:46:48 - INFO - codeparrot_training - Step 23239: {'lr': 0.00029493128327413664, 'samples': 4462080, 'steps': 23239, 'loss/train': 1.3065955638885498}
+01/29/2022 18:46:53 - INFO - codeparrot_training - Step 23240: {'lr': 0.0002949151871482982, 'samples': 4462272, 'steps': 23240, 'loss/train': 1.6747446656227112}
+01/29/2022 18:46:57 - INFO - codeparrot_training - Step 23241: {'lr': 0.00029489909083005723, 'samples': 4462464, 'steps': 23241, 'loss/train': 1.3836702704429626}
+01/29/2022 18:47:01 - INFO - codeparrot_training - Step 23242: {'lr': 0.0002948829943194829, 'samples': 4462656, 'steps': 23242, 'loss/train': 1.32831010222435}
+01/29/2022 18:47:09 - INFO - codeparrot_training - Step 23243: {'lr': 0.00029486689761664396, 'samples': 4462848, 'steps': 23243, 'loss/train': 2.1221668124198914}
+01/29/2022 18:47:13 - INFO - codeparrot_training - Step 23244: {'lr': 0.0002948508007216096, 'samples': 4463040, 'steps': 23244, 'loss/train': 1.1438261568546295}
+01/29/2022 18:47:17 - INFO - codeparrot_training - Step 23245: {'lr': 0.00029483470363444856, 'samples': 4463232, 'steps': 23245, 'loss/train': 1.6109785437583923}
+01/29/2022 18:47:21 - INFO - codeparrot_training - Step 23246: {'lr': 0.00029481860635522994, 'samples': 4463424, 'steps': 23246, 'loss/train': 1.984331488609314}
+01/29/2022 18:47:26 - INFO - codeparrot_training - Step 23247: {'lr': 0.0002948025088840226, 'samples': 4463616, 'steps': 23247, 'loss/train': 1.5337965488433838}
+01/29/2022 18:47:31 - INFO - codeparrot_training - Step 23248: {'lr': 0.0002947864112208956, 'samples': 4463808, 'steps': 23248, 'loss/train': 1.7255542874336243}
+01/29/2022 18:47:35 - INFO - codeparrot_training - Step 23249: {'lr': 0.0002947703133659178, 'samples': 4464000, 'steps': 23249, 'loss/train': 2.4917845129966736}
+01/29/2022 18:47:39 - INFO - codeparrot_training - Step 23250: {'lr': 0.00029475421531915827, 'samples': 4464192, 'steps': 23250, 'loss/train': 1.0447227358818054}
+01/29/2022 18:47:43 - INFO - codeparrot_training - Step 23251: {'lr': 0.00029473811708068576, 'samples': 4464384, 'steps': 23251, 'loss/train': 0.9773289263248444}
+01/29/2022 18:47:48 - INFO - codeparrot_training - Step 23252: {'lr': 0.0002947220186505694, 'samples': 4464576, 'steps': 23252, 'loss/train': 1.086462289094925}
+01/29/2022 18:47:53 - INFO - codeparrot_training - Step 23253: {'lr': 0.00029470592002887815, 'samples': 4464768, 'steps': 23253, 'loss/train': 1.6926743388175964}
+01/29/2022 18:47:57 - INFO - codeparrot_training - Step 23254: {'lr': 0.00029468982121568096, 'samples': 4464960, 'steps': 23254, 'loss/train': 1.1080255508422852}
+01/29/2022 18:48:01 - INFO - codeparrot_training - Step 23255: {'lr': 0.0002946737222110467, 'samples': 4465152, 'steps': 23255, 'loss/train': 1.4396816790103912}
+01/29/2022 18:48:06 - INFO - codeparrot_training - Step 23256: {'lr': 0.0002946576230150444, 'samples': 4465344, 'steps': 23256, 'loss/train': 1.1534215807914734}
+01/29/2022 18:48:10 - INFO - codeparrot_training - Step 23257: {'lr': 0.00029464152362774305, 'samples': 4465536, 'steps': 23257, 'loss/train': 1.4394406378269196}
+01/29/2022 18:48:15 - INFO - codeparrot_training - Step 23258: {'lr': 0.00029462542404921156, 'samples': 4465728, 'steps': 23258, 'loss/train': 1.5305541157722473}
+01/29/2022 18:48:19 - INFO - codeparrot_training - Step 23259: {'lr': 0.00029460932427951897, 'samples': 4465920, 'steps': 23259, 'loss/train': 2.0810272693634033}
+01/29/2022 18:48:23 - INFO - codeparrot_training - Step 23260: {'lr': 0.00029459322431873416, 'samples': 4466112, 'steps': 23260, 'loss/train': 1.840428650379181}
+01/29/2022 18:48:28 - INFO - codeparrot_training - Step 23261: {'lr': 0.00029457712416692617, 'samples': 4466304, 'steps': 23261, 'loss/train': 1.1708111464977264}
+01/29/2022 18:48:32 - INFO - codeparrot_training - Step 23262: {'lr': 0.0002945610238241639, 'samples': 4466496, 'steps': 23262, 'loss/train': 1.6881346106529236}
+01/29/2022 18:48:39 - INFO - codeparrot_training - Step 23263: {'lr': 0.0002945449232905164, 'samples': 4466688, 'steps': 23263, 'loss/train': 2.1594282388687134}
+01/29/2022 18:48:43 - INFO - codeparrot_training - Step 23264: {'lr': 0.0002945288225660525, 'samples': 4466880, 'steps': 23264, 'loss/train': 1.6376920938491821}
+01/29/2022 18:48:48 - INFO - codeparrot_training - Step 23265: {'lr': 0.00029451272165084144, 'samples': 4467072, 'steps': 23265, 'loss/train': 1.9564931988716125}
+01/29/2022 18:48:52 - INFO - codeparrot_training - Step 23266: {'lr': 0.00029449662054495184, 'samples': 4467264, 'steps': 23266, 'loss/train': 1.4839029908180237}
+01/29/2022 18:48:56 - INFO - codeparrot_training - Step 23267: {'lr': 0.0002944805192484529, 'samples': 4467456, 'steps': 23267, 'loss/train': 1.2274959683418274}
+01/29/2022 18:49:02 - INFO - codeparrot_training - Step 23268: {'lr': 0.0002944644177614136, 'samples': 4467648, 'steps': 23268, 'loss/train': 1.8114550709724426}
+01/29/2022 18:49:06 - INFO - codeparrot_training - Step 23269: {'lr': 0.00029444831608390276, 'samples': 4467840, 'steps': 23269, 'loss/train': 1.6136836409568787}
+01/29/2022 18:49:10 - INFO - codeparrot_training - Step 23270: {'lr': 0.0002944322142159895, 'samples': 4468032, 'steps': 23270, 'loss/train': 1.7151551842689514}
+01/29/2022 18:49:14 - INFO - codeparrot_training - Step 23271: {'lr': 0.0002944161121577427, 'samples': 4468224, 'steps': 23271, 'loss/train': 1.9233795404434204}
+01/29/2022 18:49:18 - INFO - codeparrot_training - Step 23272: {'lr': 0.0002944000099092314, 'samples': 4468416, 'steps': 23272, 'loss/train': 1.7427728176116943}
+01/29/2022 18:49:24 - INFO - codeparrot_training - Step 23273: {'lr': 0.0002943839074705246, 'samples': 4468608, 'steps': 23273, 'loss/train': 1.8949243426322937}
+01/29/2022 18:49:28 - INFO - codeparrot_training - Step 23274: {'lr': 0.00029436780484169125, 'samples': 4468800, 'steps': 23274, 'loss/train': 2.0702097415924072}
+01/29/2022 18:49:32 - INFO - codeparrot_training - Step 23275: {'lr': 0.0002943517020228003, 'samples': 4468992, 'steps': 23275, 'loss/train': 0.8338253796100616}
+01/29/2022 18:49:36 - INFO - codeparrot_training - Step 23276: {'lr': 0.00029433559901392067, 'samples': 4469184, 'steps': 23276, 'loss/train': 1.8925509452819824}
+01/29/2022 18:49:40 - INFO - codeparrot_training - Step 23277: {'lr': 0.0002943194958151214, 'samples': 4469376, 'steps': 23277, 'loss/train': 0.9840590357780457}
+01/29/2022 18:49:48 - INFO - codeparrot_training - Step 23278: {'lr': 0.00029430339242647157, 'samples': 4469568, 'steps': 23278, 'loss/train': 1.6704506278038025}
+01/29/2022 18:49:53 - INFO - codeparrot_training - Step 23279: {'lr': 0.00029428728884804, 'samples': 4469760, 'steps': 23279, 'loss/train': 0.9872241318225861}
+01/29/2022 18:49:57 - INFO - codeparrot_training - Step 23280: {'lr': 0.0002942711850798959, 'samples': 4469952, 'steps': 23280, 'loss/train': 1.8851932883262634}
+01/29/2022 18:50:01 - INFO - codeparrot_training - Step 23281: {'lr': 0.00029425508112210794, 'samples': 4470144, 'steps': 23281, 'loss/train': 1.5206989645957947}
+01/29/2022 18:50:05 - INFO - codeparrot_training - Step 23282: {'lr': 0.00029423897697474535, 'samples': 4470336, 'steps': 23282, 'loss/train': 0.7305037379264832}
+01/29/2022 18:50:10 - INFO - codeparrot_training - Step 23283: {'lr': 0.00029422287263787696, 'samples': 4470528, 'steps': 23283, 'loss/train': 1.5482491850852966}
+01/29/2022 18:50:15 - INFO - codeparrot_training - Step 23284: {'lr': 0.0002942067681115719, 'samples': 4470720, 'steps': 23284, 'loss/train': 2.070922315120697}
+01/29/2022 18:50:19 - INFO - codeparrot_training - Step 23285: {'lr': 0.0002941906633958989, 'samples': 4470912, 'steps': 23285, 'loss/train': 0.38009458780288696}
+01/29/2022 18:50:23 - INFO - codeparrot_training - Step 23286: {'lr': 0.0002941745584909274, 'samples': 4471104, 'steps': 23286, 'loss/train': 1.2880283296108246}
+01/29/2022 18:50:27 - INFO - codeparrot_training - Step 23287: {'lr': 0.00029415845339672596, 'samples': 4471296, 'steps': 23287, 'loss/train': 0.8239124715328217}
+01/29/2022 18:50:33 - INFO - codeparrot_training - Step 23288: {'lr': 0.00029414234811336377, 'samples': 4471488, 'steps': 23288, 'loss/train': 1.5126171112060547}
+01/29/2022 18:50:37 - INFO - codeparrot_training - Step 23289: {'lr': 0.00029412624264090983, 'samples': 4471680, 'steps': 23289, 'loss/train': 0.28543784469366074}
+01/29/2022 18:50:41 - INFO - codeparrot_training - Step 23290: {'lr': 0.00029411013697943294, 'samples': 4471872, 'steps': 23290, 'loss/train': 2.0454028844833374}
+01/29/2022 18:50:45 - INFO - codeparrot_training - Step 23291: {'lr': 0.00029409403112900227, 'samples': 4472064, 'steps': 23291, 'loss/train': 0.943126380443573}
+01/29/2022 18:50:49 - INFO - codeparrot_training - Step 23292: {'lr': 0.00029407792508968683, 'samples': 4472256, 'steps': 23292, 'loss/train': 1.8001209497451782}
+01/29/2022 18:50:57 - INFO - codeparrot_training - Step 23293: {'lr': 0.0002940618188615555, 'samples': 4472448, 'steps': 23293, 'loss/train': 2.1383376717567444}
+01/29/2022 18:51:01 - INFO - codeparrot_training - Step 23294: {'lr': 0.0002940457124446774, 'samples': 4472640, 'steps': 23294, 'loss/train': 1.557030200958252}
+01/29/2022 18:51:05 - INFO - codeparrot_training - Step 23295: {'lr': 0.0002940296058391214, 'samples': 4472832, 'steps': 23295, 'loss/train': 1.4101864993572235}
+01/29/2022 18:51:09 - INFO - codeparrot_training - Step 23296: {'lr': 0.00029401349904495654, 'samples': 4473024, 'steps': 23296, 'loss/train': 1.62547767162323}
+01/29/2022 18:51:15 - INFO - codeparrot_training - Step 23297: {'lr': 0.00029399739206225186, 'samples': 4473216, 'steps': 23297, 'loss/train': 1.8235584497451782}
+01/29/2022 18:51:19 - INFO - codeparrot_training - Step 23298: {'lr': 0.00029398128489107635, 'samples': 4473408, 'steps': 23298, 'loss/train': 1.682007372379303}
+01/29/2022 18:51:23 - INFO - codeparrot_training - Step 23299: {'lr': 0.0002939651775314989, 'samples': 4473600, 'steps': 23299, 'loss/train': 1.9396677017211914}
+01/29/2022 18:51:27 - INFO - codeparrot_training - Step 23300: {'lr': 0.0002939490699835887, 'samples': 4473792, 'steps': 23300, 'loss/train': 1.7669304013252258}
+01/29/2022 18:51:32 - INFO - codeparrot_training - Step 23301: {'lr': 0.0002939329622474145, 'samples': 4473984, 'steps': 23301, 'loss/train': 2.681039571762085}
+01/29/2022 18:51:36 - INFO - codeparrot_training - Step 23302: {'lr': 0.00029391685432304554, 'samples': 4474176, 'steps': 23302, 'loss/train': 1.3685748875141144}
+01/29/2022 18:51:43 - INFO - codeparrot_training - Step 23303: {'lr': 0.0002939007462105507, 'samples': 4474368, 'steps': 23303, 'loss/train': 1.1285144984722137}
+01/29/2022 18:51:48 - INFO - codeparrot_training - Step 23304: {'lr': 0.000293884637909999, 'samples': 4474560, 'steps': 23304, 'loss/train': 0.9999127686023712}
+01/29/2022 18:51:52 - INFO - codeparrot_training - Step 23305: {'lr': 0.00029386852942145944, 'samples': 4474752, 'steps': 23305, 'loss/train': 0.5849349349737167}
+01/29/2022 18:51:56 - INFO - codeparrot_training - Step 23306: {'lr': 0.000293852420745001, 'samples': 4474944, 'steps': 23306, 'loss/train': 1.226283073425293}
+01/29/2022 18:52:00 - INFO - codeparrot_training - Step 23307: {'lr': 0.0002938363118806928, 'samples': 4475136, 'steps': 23307, 'loss/train': 1.9939870834350586}
+01/29/2022 18:52:06 - INFO - codeparrot_training - Step 23308: {'lr': 0.00029382020282860366, 'samples': 4475328, 'steps': 23308, 'loss/train': 1.4510125815868378}
+01/29/2022 18:52:10 - INFO - codeparrot_training - Step 23309: {'lr': 0.00029380409358880277, 'samples': 4475520, 'steps': 23309, 'loss/train': 1.4119104444980621}
+01/29/2022 18:52:14 - INFO - codeparrot_training - Step 23310: {'lr': 0.000293787984161359, 'samples': 4475712, 'steps': 23310, 'loss/train': 1.4977566003799438}
+01/29/2022 18:52:18 - INFO - codeparrot_training - Step 23311: {'lr': 0.0002937718745463414, 'samples': 4475904, 'steps': 23311, 'loss/train': 2.011587381362915}
+01/29/2022 18:52:24 - INFO - codeparrot_training - Step 23312: {'lr': 0.00029375576474381903, 'samples': 4476096, 'steps': 23312, 'loss/train': 1.610706388950348}
+01/29/2022 18:52:28 - INFO - codeparrot_training - Step 23313: {'lr': 0.0002937396547538609, 'samples': 4476288, 'steps': 23313, 'loss/train': 1.1237922012805939}
+01/29/2022 18:52:32 - INFO - codeparrot_training - Step 23314: {'lr': 0.00029372354457653585, 'samples': 4476480, 'steps': 23314, 'loss/train': 7.362727403640747}
+01/29/2022 18:52:36 - INFO - codeparrot_training - Step 23315: {'lr': 0.00029370743421191305, 'samples': 4476672, 'steps': 23315, 'loss/train': 2.5959951281547546}
+01/29/2022 18:52:40 - INFO - codeparrot_training - Step 23316: {'lr': 0.0002936913236600616, 'samples': 4476864, 'steps': 23316, 'loss/train': 1.9263877272605896}
+01/29/2022 18:52:45 - INFO - codeparrot_training - Step 23317: {'lr': 0.0002936752129210503, 'samples': 4477056, 'steps': 23317, 'loss/train': 1.5166170001029968}
+01/29/2022 18:52:52 - INFO - codeparrot_training - Step 23318: {'lr': 0.00029365910199494823, 'samples': 4477248, 'steps': 23318, 'loss/train': 3.065377950668335}
+01/29/2022 18:52:57 - INFO - codeparrot_training - Step 23319: {'lr': 0.00029364299088182445, 'samples': 4477440, 'steps': 23319, 'loss/train': 3.064932703971863}
+01/29/2022 18:53:01 - INFO - codeparrot_training - Step 23320: {'lr': 0.000293626879581748, 'samples': 4477632, 'steps': 23320, 'loss/train': 1.877602458000183}
+01/29/2022 18:53:05 - INFO - codeparrot_training - Step 23321: {'lr': 0.00029361076809478784, 'samples': 4477824, 'steps': 23321, 'loss/train': 1.4004883468151093}
+01/29/2022 18:53:09 - INFO - codeparrot_training - Step 23322: {'lr': 0.00029359465642101297, 'samples': 4478016, 'steps': 23322, 'loss/train': 1.3742044866085052}
+01/29/2022 18:53:15 - INFO - codeparrot_training - Step 23323: {'lr': 0.0002935785445604924, 'samples': 4478208, 'steps': 23323, 'loss/train': 1.5735953450202942}
+01/29/2022 18:53:19 - INFO - codeparrot_training - Step 23324: {'lr': 0.0002935624325132953, 'samples': 4478400, 'steps': 23324, 'loss/train': 1.1457508206367493}
+01/29/2022 18:53:23 - INFO - codeparrot_training - Step 23325: {'lr': 0.0002935463202794905, 'samples': 4478592, 'steps': 23325, 'loss/train': 1.3250703513622284}
+01/29/2022 18:53:27 - INFO - codeparrot_training - Step 23326: {'lr': 0.00029353020785914707, 'samples': 4478784, 'steps': 23326, 'loss/train': 1.8481310606002808}
+01/29/2022 18:53:31 - INFO - codeparrot_training - Step 23327: {'lr': 0.00029351409525233416, 'samples': 4478976, 'steps': 23327, 'loss/train': 1.072963535785675}
+01/29/2022 18:53:37 - INFO - codeparrot_training - Step 23328: {'lr': 0.0002934979824591205, 'samples': 4479168, 'steps': 23328, 'loss/train': 1.936540961265564}
+01/29/2022 18:53:41 - INFO - codeparrot_training - Step 23329: {'lr': 0.0002934818694795755, 'samples': 4479360, 'steps': 23329, 'loss/train': 1.6431094408035278}
+01/29/2022 18:53:45 - INFO - codeparrot_training - Step 23330: {'lr': 0.0002934657563137678, 'samples': 4479552, 'steps': 23330, 'loss/train': 1.199665904045105}
+01/29/2022 18:53:49 - INFO - codeparrot_training - Step 23331: {'lr': 0.00029344964296176666, 'samples': 4479744, 'steps': 23331, 'loss/train': 1.3808862268924713}
+01/29/2022 18:53:53 - INFO - codeparrot_training - Step 23332: {'lr': 0.00029343352942364106, 'samples': 4479936, 'steps': 23332, 'loss/train': 2.2962589859962463}
+01/29/2022 18:53:59 - INFO - codeparrot_training - Step 23333: {'lr': 0.00029341741569946007, 'samples': 4480128, 'steps': 23333, 'loss/train': 2.3964579105377197}
+01/29/2022 18:54:03 - INFO - codeparrot_training - Step 23334: {'lr': 0.00029340130178929255, 'samples': 4480320, 'steps': 23334, 'loss/train': 1.3287532031536102}
+01/29/2022 18:54:07 - INFO - codeparrot_training - Step 23335: {'lr': 0.0002933851876932077, 'samples': 4480512, 'steps': 23335, 'loss/train': 0.911943107843399}
+01/29/2022 18:54:12 - INFO - codeparrot_training - Step 23336: {'lr': 0.00029336907341127443, 'samples': 4480704, 'steps': 23336, 'loss/train': 2.4040390849113464}
+01/29/2022 18:54:16 - INFO - codeparrot_training - Step 23337: {'lr': 0.0002933529589435619, 'samples': 4480896, 'steps': 23337, 'loss/train': 1.0912826657295227}
+01/29/2022 18:54:20 - INFO - codeparrot_training - Step 23338: {'lr': 0.000293336844290139, 'samples': 4481088, 'steps': 23338, 'loss/train': 0.9360928237438202}
+01/29/2022 18:54:27 - INFO - codeparrot_training - Step 23339: {'lr': 0.0002933207294510748, 'samples': 4481280, 'steps': 23339, 'loss/train': 4.495097279548645}
+01/29/2022 18:54:32 - INFO - codeparrot_training - Step 23340: {'lr': 0.0002933046144264384, 'samples': 4481472, 'steps': 23340, 'loss/train': 2.0485569834709167}
+01/29/2022 18:54:36 - INFO - codeparrot_training - Step 23341: {'lr': 0.0002932884992162987, 'samples': 4481664, 'steps': 23341, 'loss/train': 1.872409164905548}
+01/29/2022 18:54:40 - INFO - codeparrot_training - Step 23342: {'lr': 0.00029327238382072495, 'samples': 4481856, 'steps': 23342, 'loss/train': 1.8960413932800293}
+01/29/2022 18:54:44 - INFO - codeparrot_training - Step 23343: {'lr': 0.000293256268239786, 'samples': 4482048, 'steps': 23343, 'loss/train': 1.773486614227295}
+01/29/2022 18:54:48 - INFO - codeparrot_training - Step 23344: {'lr': 0.00029324015247355093, 'samples': 4482240, 'steps': 23344, 'loss/train': 1.8087866306304932}
+01/29/2022 18:54:54 - INFO - codeparrot_training - Step 23345: {'lr': 0.0002932240365220887, 'samples': 4482432, 'steps': 23345, 'loss/train': 1.599815011024475}
+01/29/2022 18:54:58 - INFO - codeparrot_training - Step 23346: {'lr': 0.00029320792038546855, 'samples': 4482624, 'steps': 23346, 'loss/train': 1.7377837300300598}
+01/29/2022 18:55:02 - INFO - codeparrot_training - Step 23347: {'lr': 0.0002931918040637593, 'samples': 4482816, 'steps': 23347, 'loss/train': 1.5408373475074768}
+01/29/2022 18:55:07 - INFO - codeparrot_training - Step 23348: {'lr': 0.0002931756875570301, 'samples': 4483008, 'steps': 23348, 'loss/train': 0.7731020450592041}
+01/29/2022 18:55:15 - INFO - codeparrot_training - Step 23349: {'lr': 0.0002931595708653499, 'samples': 4483200, 'steps': 23349, 'loss/train': 2.102031111717224}
+01/29/2022 18:55:19 - INFO - codeparrot_training - Step 23350: {'lr': 0.00029314345398878796, 'samples': 4483392, 'steps': 23350, 'loss/train': 2.515945851802826}
+01/29/2022 18:55:23 - INFO - codeparrot_training - Step 23351: {'lr': 0.00029312733692741307, 'samples': 4483584, 'steps': 23351, 'loss/train': 1.2912079095840454}
+01/29/2022 18:55:27 - INFO - codeparrot_training - Step 23352: {'lr': 0.00029311121968129435, 'samples': 4483776, 'steps': 23352, 'loss/train': 2.1166725754737854}
+01/29/2022 18:55:31 - INFO - codeparrot_training - Step 23353: {'lr': 0.00029309510225050087, 'samples': 4483968, 'steps': 23353, 'loss/train': 1.8947714567184448}
+01/29/2022 18:55:36 - INFO - codeparrot_training - Step 23354: {'lr': 0.00029307898463510164, 'samples': 4484160, 'steps': 23354, 'loss/train': 1.6648356914520264}
+01/29/2022 18:55:40 - INFO - codeparrot_training - Step 23355: {'lr': 0.00029306286683516573, 'samples': 4484352, 'steps': 23355, 'loss/train': 1.5453450679779053}
+01/29/2022 18:55:45 - INFO - codeparrot_training - Step 23356: {'lr': 0.00029304674885076215, 'samples': 4484544, 'steps': 23356, 'loss/train': 1.0176503956317902}
+01/29/2022 18:55:49 - INFO - codeparrot_training - Step 23357: {'lr': 0.00029303063068196006, 'samples': 4484736, 'steps': 23357, 'loss/train': 1.9858978986740112}
+01/29/2022 18:55:53 - INFO - codeparrot_training - Step 23358: {'lr': 0.0002930145123288283, 'samples': 4484928, 'steps': 23358, 'loss/train': 1.7870965003967285}
+01/29/2022 18:55:57 - INFO - codeparrot_training - Step 23359: {'lr': 0.00029299839379143613, 'samples': 4485120, 'steps': 23359, 'loss/train': 2.196174681186676}
+01/29/2022 18:56:02 - INFO - codeparrot_training - Step 23360: {'lr': 0.0002929822750698524, 'samples': 4485312, 'steps': 23360, 'loss/train': 1.3759199380874634}
+01/29/2022 18:56:07 - INFO - codeparrot_training - Step 23361: {'lr': 0.0002929661561641463, 'samples': 4485504, 'steps': 23361, 'loss/train': 2.1584295630455017}
+01/29/2022 18:56:11 - INFO - codeparrot_training - Step 23362: {'lr': 0.0002929500370743868, 'samples': 4485696, 'steps': 23362, 'loss/train': 1.3625726401805878}
+01/29/2022 18:56:15 - INFO - codeparrot_training - Step 23363: {'lr': 0.000292933917800643, 'samples': 4485888, 'steps': 23363, 'loss/train': 1.2797605991363525}
+01/29/2022 18:56:19 - INFO - codeparrot_training - Step 23364: {'lr': 0.0002929177983429839, 'samples': 4486080, 'steps': 23364, 'loss/train': 2.8687732815742493}
+01/29/2022 18:56:26 - INFO - codeparrot_training - Step 23365: {'lr': 0.0002929016787014786, 'samples': 4486272, 'steps': 23365, 'loss/train': 2.208058476448059}
+01/29/2022 18:56:30 - INFO - codeparrot_training - Step 23366: {'lr': 0.0002928855588761962, 'samples': 4486464, 'steps': 23366, 'loss/train': 2.0985467433929443}
+01/29/2022 18:56:35 - INFO - codeparrot_training - Step 23367: {'lr': 0.0002928694388672056, 'samples': 4486656, 'steps': 23367, 'loss/train': 1.923707127571106}
+01/29/2022 18:56:39 - INFO - codeparrot_training - Step 23368: {'lr': 0.00029285331867457597, 'samples': 4486848, 'steps': 23368, 'loss/train': 2.032481074333191}
+01/29/2022 18:56:43 - INFO - codeparrot_training - Step 23369: {'lr': 0.00029283719829837636, 'samples': 4487040, 'steps': 23369, 'loss/train': 1.8166193962097168}
+01/29/2022 18:56:49 - INFO - codeparrot_training - Step 23370: {'lr': 0.0002928210777386757, 'samples': 4487232, 'steps': 23370, 'loss/train': 0.4718962758779526}
+01/29/2022 18:56:53 - INFO - codeparrot_training - Step 23371: {'lr': 0.0002928049569955432, 'samples': 4487424, 'steps': 23371, 'loss/train': 1.201136827468872}
+01/29/2022 18:56:57 - INFO - codeparrot_training - Step 23372: {'lr': 0.0002927888360690478, 'samples': 4487616, 'steps': 23372, 'loss/train': 1.5234006643295288}
+01/29/2022 18:57:01 - INFO - codeparrot_training - Step 23373: {'lr': 0.00029277271495925874, 'samples': 4487808, 'steps': 23373, 'loss/train': 1.2596802413463593}
+01/29/2022 18:57:06 - INFO - codeparrot_training - Step 23374: {'lr': 0.00029275659366624493, 'samples': 4488000, 'steps': 23374, 'loss/train': 0.7350783348083496}
+01/29/2022 18:57:11 - INFO - codeparrot_training - Step 23375: {'lr': 0.00029274047219007534, 'samples': 4488192, 'steps': 23375, 'loss/train': 0.9206145107746124}
+01/29/2022 18:57:15 - INFO - codeparrot_training - Step 23376: {'lr': 0.00029272435053081917, 'samples': 4488384, 'steps': 23376, 'loss/train': 1.7407103776931763}
+01/29/2022 18:57:20 - INFO - codeparrot_training - Step 23377: {'lr': 0.0002927082286885455, 'samples': 4488576, 'steps': 23377, 'loss/train': 1.094814956188202}
+01/29/2022 18:57:24 - INFO - codeparrot_training - Step 23378: {'lr': 0.0002926921066633233, 'samples': 4488768, 'steps': 23378, 'loss/train': 1.7803650498390198}
+01/29/2022 18:57:28 - INFO - codeparrot_training - Step 23379: {'lr': 0.0002926759844552217, 'samples': 4488960, 'steps': 23379, 'loss/train': 2.378867447376251}
+01/29/2022 18:57:32 - INFO - codeparrot_training - Step 23380: {'lr': 0.0002926598620643097, 'samples': 4489152, 'steps': 23380, 'loss/train': 0.060898203402757645}
+01/29/2022 18:57:39 - INFO - codeparrot_training - Step 23381: {'lr': 0.0002926437394906564, 'samples': 4489344, 'steps': 23381, 'loss/train': 1.8095946907997131}
+01/29/2022 18:57:43 - INFO - codeparrot_training - Step 23382: {'lr': 0.00029262761673433087, 'samples': 4489536, 'steps': 23382, 'loss/train': 0.8861842453479767}
+01/29/2022 18:57:48 - INFO - codeparrot_training - Step 23383: {'lr': 0.0002926114937954022, 'samples': 4489728, 'steps': 23383, 'loss/train': 1.2276331186294556}
+01/29/2022 18:57:52 - INFO - codeparrot_training - Step 23384: {'lr': 0.0002925953706739394, 'samples': 4489920, 'steps': 23384, 'loss/train': 1.8839187026023865}
+01/29/2022 18:57:56 - INFO - codeparrot_training - Step 23385: {'lr': 0.0002925792473700116, 'samples': 4490112, 'steps': 23385, 'loss/train': 1.9476351141929626}
+01/29/2022 18:58:01 - INFO - codeparrot_training - Step 23386: {'lr': 0.00029256312388368773, 'samples': 4490304, 'steps': 23386, 'loss/train': 1.7674071192741394}
+01/29/2022 18:58:05 - INFO - codeparrot_training - Step 23387: {'lr': 0.000292547000215037, 'samples': 4490496, 'steps': 23387, 'loss/train': 1.41392183303833}
+01/29/2022 18:58:10 - INFO - codeparrot_training - Step 23388: {'lr': 0.0002925308763641284, 'samples': 4490688, 'steps': 23388, 'loss/train': 2.2168931365013123}
+01/29/2022 18:58:14 - INFO - codeparrot_training - Step 23389: {'lr': 0.00029251475233103104, 'samples': 4490880, 'steps': 23389, 'loss/train': 1.4965595304965973}
+01/29/2022 18:58:18 - INFO - codeparrot_training - Step 23390: {'lr': 0.000292498628115814, 'samples': 4491072, 'steps': 23390, 'loss/train': 1.8176636695861816}
+01/29/2022 18:58:23 - INFO - codeparrot_training - Step 23391: {'lr': 0.00029248250371854636, 'samples': 4491264, 'steps': 23391, 'loss/train': 2.2741209268569946}
+01/29/2022 18:58:28 - INFO - codeparrot_training - Step 23392: {'lr': 0.0002924663791392971, 'samples': 4491456, 'steps': 23392, 'loss/train': 1.4217088222503662}
+01/29/2022 18:58:32 - INFO - codeparrot_training - Step 23393: {'lr': 0.0002924502543781354, 'samples': 4491648, 'steps': 23393, 'loss/train': 2.0711830258369446}
+01/29/2022 18:58:36 - INFO - codeparrot_training - Step 23394: {'lr': 0.0002924341294351303, 'samples': 4491840, 'steps': 23394, 'loss/train': 1.3741532564163208}
+01/29/2022 18:58:40 - INFO - codeparrot_training - Step 23395: {'lr': 0.0002924180043103508, 'samples': 4492032, 'steps': 23395, 'loss/train': 1.5941179990768433}
+01/29/2022 18:58:47 - INFO - codeparrot_training - Step 23396: {'lr': 0.0002924018790038662, 'samples': 4492224, 'steps': 23396, 'loss/train': 5.665289282798767}
+01/29/2022 18:58:51 - INFO - codeparrot_training - Step 23397: {'lr': 0.0002923857535157452, 'samples': 4492416, 'steps': 23397, 'loss/train': 1.8933770656585693}
+01/29/2022 18:58:56 - INFO - codeparrot_training - Step 23398: {'lr': 0.00029236962784605727, 'samples': 4492608, 'steps': 23398, 'loss/train': 1.6870286464691162}
+01/29/2022 18:59:00 - INFO - codeparrot_training - Step 23399: {'lr': 0.00029235350199487115, 'samples': 4492800, 'steps': 23399, 'loss/train': 1.9260896444320679}
+01/29/2022 18:59:04 - INFO - codeparrot_training - Step 23400: {'lr': 0.0002923373759622561, 'samples': 4492992, 'steps': 23400, 'loss/train': 1.4178017377853394}
+01/29/2022 18:59:09 - INFO - codeparrot_training - Step 23401: {'lr': 0.00029232124974828126, 'samples': 4493184, 'steps': 23401, 'loss/train': 2.1675020456314087}
+01/29/2022 18:59:14 - INFO - codeparrot_training - Step 23402: {'lr': 0.0002923051233530156, 'samples': 4493376, 'steps': 23402, 'loss/train': 1.398179680109024}
+01/29/2022 18:59:18 - INFO - codeparrot_training - Step 23403: {'lr': 0.00029228899677652826, 'samples': 4493568, 'steps': 23403, 'loss/train': 1.1035617291927338}
+01/29/2022 18:59:22 - INFO - codeparrot_training - Step 23404: {'lr': 0.0002922728700188882, 'samples': 4493760, 'steps': 23404, 'loss/train': 0.8353758752346039}
+01/29/2022 18:59:26 - INFO - codeparrot_training - Step 23405: {'lr': 0.00029225674308016464, 'samples': 4493952, 'steps': 23405, 'loss/train': 0.5655635297298431}
+01/29/2022 18:59:31 - INFO - codeparrot_training - Step 23406: {'lr': 0.00029224061596042663, 'samples': 4494144, 'steps': 23406, 'loss/train': 1.6903476119041443}
+01/29/2022 18:59:35 - INFO - codeparrot_training - Step 23407: {'lr': 0.0002922244886597432, 'samples': 4494336, 'steps': 23407, 'loss/train': 1.7021645307540894}
+01/29/2022 18:59:40 - INFO - codeparrot_training - Step 23408: {'lr': 0.00029220836117818346, 'samples': 4494528, 'steps': 23408, 'loss/train': 0.03564136661589146}
+01/29/2022 18:59:44 - INFO - codeparrot_training - Step 23409: {'lr': 0.00029219223351581653, 'samples': 4494720, 'steps': 23409, 'loss/train': 2.229494333267212}
+01/29/2022 18:59:48 - INFO - codeparrot_training - Step 23410: {'lr': 0.00029217610567271147, 'samples': 4494912, 'steps': 23410, 'loss/train': 2.0820257663726807}
+01/29/2022 18:59:55 - INFO - codeparrot_training - Step 23411: {'lr': 0.00029215997764893734, 'samples': 4495104, 'steps': 23411, 'loss/train': 2.163722813129425}
+01/29/2022 18:59:59 - INFO - codeparrot_training - Step 23412: {'lr': 0.00029214384944456327, 'samples': 4495296, 'steps': 23412, 'loss/train': 1.4853903651237488}
+01/29/2022 19:00:04 - INFO - codeparrot_training - Step 23413: {'lr': 0.00029212772105965837, 'samples': 4495488, 'steps': 23413, 'loss/train': 1.4210830628871918}
+01/29/2022 19:00:08 - INFO - codeparrot_training - Step 23414: {'lr': 0.0002921115924942916, 'samples': 4495680, 'steps': 23414, 'loss/train': 1.2766062319278717}
+01/29/2022 19:00:12 - INFO - codeparrot_training - Step 23415: {'lr': 0.0002920954637485322, 'samples': 4495872, 'steps': 23415, 'loss/train': 1.5876399278640747}
+01/29/2022 19:00:17 - INFO - codeparrot_training - Step 23416: {'lr': 0.0002920793348224493, 'samples': 4496064, 'steps': 23416, 'loss/train': 1.630038321018219}
+01/29/2022 19:00:22 - INFO - codeparrot_training - Step 23417: {'lr': 0.0002920632057161117, 'samples': 4496256, 'steps': 23417, 'loss/train': 2.072730839252472}
+01/29/2022 19:00:26 - INFO - codeparrot_training - Step 23418: {'lr': 0.00029204707642958876, 'samples': 4496448, 'steps': 23418, 'loss/train': 1.281270056962967}
+01/29/2022 19:00:30 - INFO - codeparrot_training - Step 23419: {'lr': 0.0002920309469629495, 'samples': 4496640, 'steps': 23419, 'loss/train': 1.8569276332855225}
+01/29/2022 19:00:34 - INFO - codeparrot_training - Step 23420: {'lr': 0.000292014817316263, 'samples': 4496832, 'steps': 23420, 'loss/train': 0.9491633176803589}
+01/29/2022 19:00:41 - INFO - codeparrot_training - Step 23421: {'lr': 0.00029199868748959836, 'samples': 4497024, 'steps': 23421, 'loss/train': 1.0134774148464203}
+01/29/2022 19:00:46 - INFO - codeparrot_training - Step 23422: {'lr': 0.00029198255748302473, 'samples': 4497216, 'steps': 23422, 'loss/train': 2.256779193878174}
+01/29/2022 19:00:50 - INFO - codeparrot_training - Step 23423: {'lr': 0.00029196642729661113, 'samples': 4497408, 'steps': 23423, 'loss/train': 2.050343155860901}
+01/29/2022 19:00:54 - INFO - codeparrot_training - Step 23424: {'lr': 0.0002919502969304266, 'samples': 4497600, 'steps': 23424, 'loss/train': 1.6850246787071228}
+01/29/2022 19:00:58 - INFO - codeparrot_training - Step 23425: {'lr': 0.0002919341663845404, 'samples': 4497792, 'steps': 23425, 'loss/train': 1.2777585983276367}
+01/29/2022 19:01:03 - INFO - codeparrot_training - Step 23426: {'lr': 0.00029191803565902153, 'samples': 4497984, 'steps': 23426, 'loss/train': 1.154289722442627}
+01/29/2022 19:01:08 - INFO - codeparrot_training - Step 23427: {'lr': 0.00029190190475393915, 'samples': 4498176, 'steps': 23427, 'loss/train': 1.3202232420444489}
+01/29/2022 19:01:12 - INFO - codeparrot_training - Step 23428: {'lr': 0.0002918857736693622, 'samples': 4498368, 'steps': 23428, 'loss/train': 1.4825566112995148}
+01/29/2022 19:01:16 - INFO - codeparrot_training - Step 23429: {'lr': 0.00029186964240536, 'samples': 4498560, 'steps': 23429, 'loss/train': 2.413850426673889}
+01/29/2022 19:01:20 - INFO - codeparrot_training - Step 23430: {'lr': 0.0002918535109620015, 'samples': 4498752, 'steps': 23430, 'loss/train': 0.3662059083580971}
+01/29/2022 19:01:25 - INFO - codeparrot_training - Step 23431: {'lr': 0.0002918373793393559, 'samples': 4498944, 'steps': 23431, 'loss/train': 1.5947660207748413}
+01/29/2022 19:01:30 - INFO - codeparrot_training - Step 23432: {'lr': 0.0002918212475374922, 'samples': 4499136, 'steps': 23432, 'loss/train': 1.6040446758270264}
+01/29/2022 19:01:34 - INFO - codeparrot_training - Step 23433: {'lr': 0.0002918051155564796, 'samples': 4499328, 'steps': 23433, 'loss/train': 1.9761255383491516}
+01/29/2022 19:01:38 - INFO - codeparrot_training - Step 23434: {'lr': 0.0002917889833963871, 'samples': 4499520, 'steps': 23434, 'loss/train': 2.336623191833496}
+01/29/2022 19:01:42 - INFO - codeparrot_training - Step 23435: {'lr': 0.00029177285105728393, 'samples': 4499712, 'steps': 23435, 'loss/train': 1.6058651804924011}
+01/29/2022 19:01:49 - INFO - codeparrot_training - Step 23436: {'lr': 0.00029175671853923907, 'samples': 4499904, 'steps': 23436, 'loss/train': 1.0747953057289124}
+01/29/2022 19:01:53 - INFO - codeparrot_training - Step 23437: {'lr': 0.00029174058584232167, 'samples': 4500096, 'steps': 23437, 'loss/train': 1.886991262435913}
+01/29/2022 19:01:58 - INFO - codeparrot_training - Step 23438: {'lr': 0.00029172445296660096, 'samples': 4500288, 'steps': 23438, 'loss/train': 1.4761460423469543}
+01/29/2022 19:02:02 - INFO - codeparrot_training - Step 23439: {'lr': 0.0002917083199121459, 'samples': 4500480, 'steps': 23439, 'loss/train': 1.9325183629989624}
+01/29/2022 19:02:06 - INFO - codeparrot_training - Step 23440: {'lr': 0.00029169218667902556, 'samples': 4500672, 'steps': 23440, 'loss/train': 1.462662398815155}
+01/29/2022 19:02:11 - INFO - codeparrot_training - Step 23441: {'lr': 0.0002916760532673092, 'samples': 4500864, 'steps': 23441, 'loss/train': 1.06521937251091}
+01/29/2022 19:02:16 - INFO - codeparrot_training - Step 23442: {'lr': 0.0002916599196770659, 'samples': 4501056, 'steps': 23442, 'loss/train': 1.634423553943634}
+01/29/2022 19:02:20 - INFO - codeparrot_training - Step 23443: {'lr': 0.00029164378590836466, 'samples': 4501248, 'steps': 23443, 'loss/train': 2.208073318004608}
+01/29/2022 19:02:24 - INFO - codeparrot_training - Step 23444: {'lr': 0.0002916276519612747, 'samples': 4501440, 'steps': 23444, 'loss/train': 1.4235206544399261}
+01/29/2022 19:02:28 - INFO - codeparrot_training - Step 23445: {'lr': 0.00029161151783586506, 'samples': 4501632, 'steps': 23445, 'loss/train': 1.533090054988861}
+01/29/2022 19:02:33 - INFO - codeparrot_training - Step 23446: {'lr': 0.0002915953835322049, 'samples': 4501824, 'steps': 23446, 'loss/train': 1.6342588663101196}
+01/29/2022 19:02:37 - INFO - codeparrot_training - Step 23447: {'lr': 0.0002915792490503634, 'samples': 4502016, 'steps': 23447, 'loss/train': 0.733200877904892}
+01/29/2022 19:02:42 - INFO - codeparrot_training - Step 23448: {'lr': 0.0002915631143904095, 'samples': 4502208, 'steps': 23448, 'loss/train': 1.6744832396507263}
+01/29/2022 19:02:46 - INFO - codeparrot_training - Step 23449: {'lr': 0.0002915469795524125, 'samples': 4502400, 'steps': 23449, 'loss/train': 1.7502082586288452}
+01/29/2022 19:02:50 - INFO - codeparrot_training - Step 23450: {'lr': 0.00029153084453644135, 'samples': 4502592, 'steps': 23450, 'loss/train': 1.2573341131210327}
+01/29/2022 19:02:57 - INFO - codeparrot_training - Step 23451: {'lr': 0.0002915147093425653, 'samples': 4502784, 'steps': 23451, 'loss/train': 0.8749573230743408}
+01/29/2022 19:03:01 - INFO - codeparrot_training - Step 23452: {'lr': 0.0002914985739708534, 'samples': 4502976, 'steps': 23452, 'loss/train': 1.524575114250183}
+01/29/2022 19:03:06 - INFO - codeparrot_training - Step 23453: {'lr': 0.00029148243842137486, 'samples': 4503168, 'steps': 23453, 'loss/train': 1.3035745024681091}
+01/29/2022 19:03:10 - INFO - codeparrot_training - Step 23454: {'lr': 0.0002914663026941986, 'samples': 4503360, 'steps': 23454, 'loss/train': 1.414121925830841}
+01/29/2022 19:03:14 - INFO - codeparrot_training - Step 23455: {'lr': 0.00029145016678939394, 'samples': 4503552, 'steps': 23455, 'loss/train': 1.3035902380943298}
+01/29/2022 19:03:19 - INFO - codeparrot_training - Step 23456: {'lr': 0.00029143403070702994, 'samples': 4503744, 'steps': 23456, 'loss/train': 1.3982045352458954}
+01/29/2022 19:03:24 - INFO - codeparrot_training - Step 23457: {'lr': 0.0002914178944471757, 'samples': 4503936, 'steps': 23457, 'loss/train': 1.4357421398162842}
+01/29/2022 19:03:28 - INFO - codeparrot_training - Step 23458: {'lr': 0.00029140175800990034, 'samples': 4504128, 'steps': 23458, 'loss/train': 1.547856867313385}
+01/29/2022 19:03:32 - INFO - codeparrot_training - Step 23459: {'lr': 0.000291385621395273, 'samples': 4504320, 'steps': 23459, 'loss/train': 2.009180009365082}
+01/29/2022 19:03:36 - INFO - codeparrot_training - Step 23460: {'lr': 0.0002913694846033628, 'samples': 4504512, 'steps': 23460, 'loss/train': 1.4154377281665802}
+01/29/2022 19:03:42 - INFO - codeparrot_training - Step 23461: {'lr': 0.00029135334763423883, 'samples': 4504704, 'steps': 23461, 'loss/train': 1.318810522556305}
+01/29/2022 19:03:46 - INFO - codeparrot_training - Step 23462: {'lr': 0.0002913372104879703, 'samples': 4504896, 'steps': 23462, 'loss/train': 2.385393261909485}
+01/29/2022 19:03:50 - INFO - codeparrot_training - Step 23463: {'lr': 0.00029132107316462625, 'samples': 4505088, 'steps': 23463, 'loss/train': 1.3901743590831757}
+01/29/2022 19:03:54 - INFO - codeparrot_training - Step 23464: {'lr': 0.00029130493566427586, 'samples': 4505280, 'steps': 23464, 'loss/train': 1.726375937461853}
+01/29/2022 19:03:59 - INFO - codeparrot_training - Step 23465: {'lr': 0.0002912887979869882, 'samples': 4505472, 'steps': 23465, 'loss/train': 1.1280149817466736}
+01/29/2022 19:04:06 - INFO - codeparrot_training - Step 23466: {'lr': 0.00029127266013283254, 'samples': 4505664, 'steps': 23466, 'loss/train': 1.5110414028167725}
+01/29/2022 19:04:10 - INFO - codeparrot_training - Step 23467: {'lr': 0.0002912565221018778, 'samples': 4505856, 'steps': 23467, 'loss/train': 1.598167598247528}
+01/29/2022 19:04:14 - INFO - codeparrot_training - Step 23468: {'lr': 0.00029124038389419325, 'samples': 4506048, 'steps': 23468, 'loss/train': 1.4244639873504639}
+01/29/2022 19:04:19 - INFO - codeparrot_training - Step 23469: {'lr': 0.00029122424550984805, 'samples': 4506240, 'steps': 23469, 'loss/train': 2.148797035217285}
+01/29/2022 19:04:23 - INFO - codeparrot_training - Step 23470: {'lr': 0.00029120810694891126, 'samples': 4506432, 'steps': 23470, 'loss/train': 1.1480562686920166}
+01/29/2022 19:04:28 - INFO - codeparrot_training - Step 23471: {'lr': 0.00029119196821145193, 'samples': 4506624, 'steps': 23471, 'loss/train': 1.9497599601745605}
+01/29/2022 19:04:32 - INFO - codeparrot_training - Step 23472: {'lr': 0.00029117582929753935, 'samples': 4506816, 'steps': 23472, 'loss/train': 1.742314338684082}
+01/29/2022 19:04:36 - INFO - codeparrot_training - Step 23473: {'lr': 0.00029115969020724263, 'samples': 4507008, 'steps': 23473, 'loss/train': 0.9851269125938416}
+01/29/2022 19:04:40 - INFO - codeparrot_training - Step 23474: {'lr': 0.0002911435509406308, 'samples': 4507200, 'steps': 23474, 'loss/train': 1.542772114276886}
+01/29/2022 19:04:45 - INFO - codeparrot_training - Step 23475: {'lr': 0.0002911274114977731, 'samples': 4507392, 'steps': 23475, 'loss/train': 2.123226821422577}
+01/29/2022 19:04:50 - INFO - codeparrot_training - Step 23476: {'lr': 0.0002911112718787386, 'samples': 4507584, 'steps': 23476, 'loss/train': 1.1383949518203735}
+01/29/2022 19:04:54 - INFO - codeparrot_training - Step 23477: {'lr': 0.0002910951320835965, 'samples': 4507776, 'steps': 23477, 'loss/train': 1.2849175035953522}
+01/29/2022 19:04:58 - INFO - codeparrot_training - Step 23478: {'lr': 0.0002910789921124159, 'samples': 4507968, 'steps': 23478, 'loss/train': 1.966187596321106}
+01/29/2022 19:05:03 - INFO - codeparrot_training - Step 23479: {'lr': 0.00029106285196526594, 'samples': 4508160, 'steps': 23479, 'loss/train': 0.7822984457015991}
+01/29/2022 19:05:07 - INFO - codeparrot_training - Step 23480: {'lr': 0.00029104671164221574, 'samples': 4508352, 'steps': 23480, 'loss/train': 1.988376259803772}
+01/29/2022 19:05:14 - INFO - codeparrot_training - Step 23481: {'lr': 0.0002910305711433345, 'samples': 4508544, 'steps': 23481, 'loss/train': 1.0579802691936493}
+01/29/2022 19:05:18 - INFO - codeparrot_training - Step 23482: {'lr': 0.00029101443046869127, 'samples': 4508736, 'steps': 23482, 'loss/train': 1.953819751739502}
+01/29/2022 19:05:22 - INFO - codeparrot_training - Step 23483: {'lr': 0.0002909982896183553, 'samples': 4508928, 'steps': 23483, 'loss/train': 0.9124419987201691}
+01/29/2022 19:05:27 - INFO - codeparrot_training - Step 23484: {'lr': 0.0002909821485923956, 'samples': 4509120, 'steps': 23484, 'loss/train': 2.545562446117401}
+01/29/2022 19:05:31 - INFO - codeparrot_training - Step 23485: {'lr': 0.0002909660073908814, 'samples': 4509312, 'steps': 23485, 'loss/train': 2.3818620443344116}
+01/29/2022 19:05:36 - INFO - codeparrot_training - Step 23486: {'lr': 0.00029094986601388195, 'samples': 4509504, 'steps': 23486, 'loss/train': 2.2068700790405273}
+01/29/2022 19:05:40 - INFO - codeparrot_training - Step 23487: {'lr': 0.00029093372446146613, 'samples': 4509696, 'steps': 23487, 'loss/train': 1.413188874721527}
+01/29/2022 19:05:44 - INFO - codeparrot_training - Step 23488: {'lr': 0.0002909175827337033, 'samples': 4509888, 'steps': 23488, 'loss/train': 1.9557716846466064}
+01/29/2022 19:05:49 - INFO - codeparrot_training - Step 23489: {'lr': 0.00029090144083066254, 'samples': 4510080, 'steps': 23489, 'loss/train': 1.7378736734390259}
+01/29/2022 19:05:53 - INFO - codeparrot_training - Step 23490: {'lr': 0.00029088529875241306, 'samples': 4510272, 'steps': 23490, 'loss/train': 1.7797958850860596}
+01/29/2022 19:05:58 - INFO - codeparrot_training - Step 23491: {'lr': 0.00029086915649902383, 'samples': 4510464, 'steps': 23491, 'loss/train': 2.420977234840393}
+01/29/2022 19:06:02 - INFO - codeparrot_training - Step 23492: {'lr': 0.0002908530140705642, 'samples': 4510656, 'steps': 23492, 'loss/train': 1.1180883944034576}
+01/29/2022 19:06:06 - INFO - codeparrot_training - Step 23493: {'lr': 0.0002908368714671031, 'samples': 4510848, 'steps': 23493, 'loss/train': 2.1636418104171753}
+01/29/2022 19:06:11 - INFO - codeparrot_training - Step 23494: {'lr': 0.00029082072868870997, 'samples': 4511040, 'steps': 23494, 'loss/train': 1.6476813554763794}
+01/29/2022 19:06:15 - INFO - codeparrot_training - Step 23495: {'lr': 0.00029080458573545366, 'samples': 4511232, 'steps': 23495, 'loss/train': 0.6330626159906387}
+01/29/2022 19:06:22 - INFO - codeparrot_training - Step 23496: {'lr': 0.0002907884426074036, 'samples': 4511424, 'steps': 23496, 'loss/train': 0.10907256603240967}
+01/29/2022 19:06:26 - INFO - codeparrot_training - Step 23497: {'lr': 0.0002907722993046287, 'samples': 4511616, 'steps': 23497, 'loss/train': 1.4759130477905273}
+01/29/2022 19:06:31 - INFO - codeparrot_training - Step 23498: {'lr': 0.0002907561558271983, 'samples': 4511808, 'steps': 23498, 'loss/train': 1.6287830471992493}
+01/29/2022 19:06:35 - INFO - codeparrot_training - Step 23499: {'lr': 0.0002907400121751814, 'samples': 4512000, 'steps': 23499, 'loss/train': 1.3182591497898102}
+01/29/2022 19:06:39 - INFO - codeparrot_training - Step 23500: {'lr': 0.0002907238683486472, 'samples': 4512192, 'steps': 23500, 'loss/train': 1.9748808145523071}
+01/29/2022 19:06:44 - INFO - codeparrot_training - Step 23501: {'lr': 0.0002907077243476649, 'samples': 4512384, 'steps': 23501, 'loss/train': 1.4046264588832855}
+01/29/2022 19:06:48 - INFO - codeparrot_training - Step 23502: {'lr': 0.0002906915801723037, 'samples': 4512576, 'steps': 23502, 'loss/train': 1.0986999571323395}
+01/29/2022 19:06:53 - INFO - codeparrot_training - Step 23503: {'lr': 0.00029067543582263265, 'samples': 4512768, 'steps': 23503, 'loss/train': 2.2202517986297607}
+01/29/2022 19:06:57 - INFO - codeparrot_training - Step 23504: {'lr': 0.00029065929129872095, 'samples': 4512960, 'steps': 23504, 'loss/train': 1.6832519173622131}
+01/29/2022 19:07:01 - INFO - codeparrot_training - Step 23505: {'lr': 0.0002906431466006378, 'samples': 4513152, 'steps': 23505, 'loss/train': 1.800916850566864}
+01/29/2022 19:07:06 - INFO - codeparrot_training - Step 23506: {'lr': 0.0002906270017284522, 'samples': 4513344, 'steps': 23506, 'loss/train': 2.2701990008354187}
+01/29/2022 19:07:10 - INFO - codeparrot_training - Step 23507: {'lr': 0.0002906108566822335, 'samples': 4513536, 'steps': 23507, 'loss/train': 1.4367680847644806}
+01/29/2022 19:07:15 - INFO - codeparrot_training - Step 23508: {'lr': 0.0002905947114620508, 'samples': 4513728, 'steps': 23508, 'loss/train': 1.2189191579818726}
+01/29/2022 19:07:19 - INFO - codeparrot_training - Step 23509: {'lr': 0.0002905785660679732, 'samples': 4513920, 'steps': 23509, 'loss/train': 0.5008187144994736}
+01/29/2022 19:07:23 - INFO - codeparrot_training - Step 23510: {'lr': 0.00029056242050006995, 'samples': 4514112, 'steps': 23510, 'loss/train': 0.8042642176151276}
+01/29/2022 19:07:28 - INFO - codeparrot_training - Step 23511: {'lr': 0.0002905462747584101, 'samples': 4514304, 'steps': 23511, 'loss/train': 2.079659879207611}
+01/29/2022 19:07:32 - INFO - codeparrot_training - Step 23512: {'lr': 0.00029053012884306295, 'samples': 4514496, 'steps': 23512, 'loss/train': 1.538853943347931}
+01/29/2022 19:07:37 - INFO - codeparrot_training - Step 23513: {'lr': 0.00029051398275409756, 'samples': 4514688, 'steps': 23513, 'loss/train': 1.7433398365974426}
+01/29/2022 19:07:41 - INFO - codeparrot_training - Step 23514: {'lr': 0.0002904978364915832, 'samples': 4514880, 'steps': 23514, 'loss/train': 2.0995720624923706}
+01/29/2022 19:07:45 - INFO - codeparrot_training - Step 23515: {'lr': 0.0002904816900555889, 'samples': 4515072, 'steps': 23515, 'loss/train': 2.2261358499526978}
+01/29/2022 19:07:55 - INFO - codeparrot_training - Step 23516: {'lr': 0.000290465543446184, 'samples': 4515264, 'steps': 23516, 'loss/train': 1.4080660343170166}
+01/29/2022 19:08:00 - INFO - codeparrot_training - Step 23517: {'lr': 0.0002904493966634374, 'samples': 4515456, 'steps': 23517, 'loss/train': 2.0122350454330444}
+01/29/2022 19:08:04 - INFO - codeparrot_training - Step 23518: {'lr': 0.0002904332497074186, 'samples': 4515648, 'steps': 23518, 'loss/train': 1.3460581004619598}
+01/29/2022 19:08:08 - INFO - codeparrot_training - Step 23519: {'lr': 0.00029041710257819643, 'samples': 4515840, 'steps': 23519, 'loss/train': 2.007681369781494}
+01/29/2022 19:08:12 - INFO - codeparrot_training - Step 23520: {'lr': 0.0002904009552758403, 'samples': 4516032, 'steps': 23520, 'loss/train': 1.9106823205947876}
+01/29/2022 19:08:18 - INFO - codeparrot_training - Step 23521: {'lr': 0.0002903848078004193, 'samples': 4516224, 'steps': 23521, 'loss/train': 1.7514288425445557}
+01/29/2022 19:08:22 - INFO - codeparrot_training - Step 23522: {'lr': 0.00029036866015200256, 'samples': 4516416, 'steps': 23522, 'loss/train': 1.829567313194275}
+01/29/2022 19:08:26 - INFO - codeparrot_training - Step 23523: {'lr': 0.0002903525123306594, 'samples': 4516608, 'steps': 23523, 'loss/train': 1.5408071279525757}
+01/29/2022 19:08:30 - INFO - codeparrot_training - Step 23524: {'lr': 0.0002903363643364588, 'samples': 4516800, 'steps': 23524, 'loss/train': 1.0977406203746796}
+01/29/2022 19:08:34 - INFO - codeparrot_training - Step 23525: {'lr': 0.0002903202161694701, 'samples': 4516992, 'steps': 23525, 'loss/train': 1.4390417039394379}
+01/29/2022 19:08:42 - INFO - codeparrot_training - Step 23526: {'lr': 0.0002903040678297624, 'samples': 4517184, 'steps': 23526, 'loss/train': 1.7109475135803223}
+01/29/2022 19:08:46 - INFO - codeparrot_training - Step 23527: {'lr': 0.00029028791931740483, 'samples': 4517376, 'steps': 23527, 'loss/train': 1.2296702563762665}
+01/29/2022 19:08:50 - INFO - codeparrot_training - Step 23528: {'lr': 0.0002902717706324666, 'samples': 4517568, 'steps': 23528, 'loss/train': 2.0463956594467163}
+01/29/2022 19:08:54 - INFO - codeparrot_training - Step 23529: {'lr': 0.00029025562177501695, 'samples': 4517760, 'steps': 23529, 'loss/train': 1.7227948307991028}
+01/29/2022 19:08:58 - INFO - codeparrot_training - Step 23530: {'lr': 0.0002902394727451249, 'samples': 4517952, 'steps': 23530, 'loss/train': 1.3101400136947632}
+01/29/2022 19:09:04 - INFO - codeparrot_training - Step 23531: {'lr': 0.00029022332354285986, 'samples': 4518144, 'steps': 23531, 'loss/train': 1.4097506403923035}
+01/29/2022 19:09:08 - INFO - codeparrot_training - Step 23532: {'lr': 0.00029020717416829075, 'samples': 4518336, 'steps': 23532, 'loss/train': 1.9474639892578125}
+01/29/2022 19:09:12 - INFO - codeparrot_training - Step 23533: {'lr': 0.0002901910246214869, 'samples': 4518528, 'steps': 23533, 'loss/train': 0.5948719829320908}
+01/29/2022 19:09:17 - INFO - codeparrot_training - Step 23534: {'lr': 0.0002901748749025175, 'samples': 4518720, 'steps': 23534, 'loss/train': 1.0492455661296844}
+01/29/2022 19:09:21 - INFO - codeparrot_training - Step 23535: {'lr': 0.00029015872501145164, 'samples': 4518912, 'steps': 23535, 'loss/train': 2.3044983744621277}
+01/29/2022 19:09:26 - INFO - codeparrot_training - Step 23536: {'lr': 0.00029014257494835863, 'samples': 4519104, 'steps': 23536, 'loss/train': 1.0451931059360504}
+01/29/2022 19:09:30 - INFO - codeparrot_training - Step 23537: {'lr': 0.0002901264247133075, 'samples': 4519296, 'steps': 23537, 'loss/train': 1.7027868032455444}
+01/29/2022 19:09:34 - INFO - codeparrot_training - Step 23538: {'lr': 0.00029011027430636755, 'samples': 4519488, 'steps': 23538, 'loss/train': 2.0271987318992615}
+01/29/2022 19:09:39 - INFO - codeparrot_training - Step 23539: {'lr': 0.00029009412372760793, 'samples': 4519680, 'steps': 23539, 'loss/train': 0.6688603609800339}
+01/29/2022 19:09:43 - INFO - codeparrot_training - Step 23540: {'lr': 0.00029007797297709784, 'samples': 4519872, 'steps': 23540, 'loss/train': 2.1581037640571594}
+01/29/2022 19:09:50 - INFO - codeparrot_training - Step 23541: {'lr': 0.00029006182205490634, 'samples': 4520064, 'steps': 23541, 'loss/train': 1.0117990672588348}
+01/29/2022 19:09:54 - INFO - codeparrot_training - Step 23542: {'lr': 0.0002900456709611028, 'samples': 4520256, 'steps': 23542, 'loss/train': 1.702073335647583}
+01/29/2022 19:09:58 - INFO - codeparrot_training - Step 23543: {'lr': 0.0002900295196957563, 'samples': 4520448, 'steps': 23543, 'loss/train': 2.1143898367881775}
+01/29/2022 19:10:02 - INFO - codeparrot_training - Step 23544: {'lr': 0.00029001336825893603, 'samples': 4520640, 'steps': 23544, 'loss/train': 1.314928561449051}
+01/29/2022 19:10:07 - INFO - codeparrot_training - Step 23545: {'lr': 0.0002899972166507112, 'samples': 4520832, 'steps': 23545, 'loss/train': 1.609722912311554}
+01/29/2022 19:10:12 - INFO - codeparrot_training - Step 23546: {'lr': 0.000289981064871151, 'samples': 4521024, 'steps': 23546, 'loss/train': 1.8978588581085205}
+01/29/2022 19:10:16 - INFO - codeparrot_training - Step 23547: {'lr': 0.00028996491292032465, 'samples': 4521216, 'steps': 23547, 'loss/train': 0.7476308941841125}
+01/29/2022 19:10:20 - INFO - codeparrot_training - Step 23548: {'lr': 0.00028994876079830125, 'samples': 4521408, 'steps': 23548, 'loss/train': 1.5288351774215698}
+01/29/2022 19:10:24 - INFO - codeparrot_training - Step 23549: {'lr': 0.00028993260850515015, 'samples': 4521600, 'steps': 23549, 'loss/train': 0.5534289926290512}
+01/29/2022 19:10:29 - INFO - codeparrot_training - Step 23550: {'lr': 0.0002899164560409403, 'samples': 4521792, 'steps': 23550, 'loss/train': 1.5859671235084534}
+01/29/2022 19:10:36 - INFO - codeparrot_training - Step 23551: {'lr': 0.00028990030340574107, 'samples': 4521984, 'steps': 23551, 'loss/train': 2.0358683466911316}
+01/29/2022 19:10:40 - INFO - codeparrot_training - Step 23552: {'lr': 0.0002898841505996216, 'samples': 4522176, 'steps': 23552, 'loss/train': 2.123853027820587}
+01/29/2022 19:10:44 - INFO - codeparrot_training - Step 23553: {'lr': 0.0002898679976226511, 'samples': 4522368, 'steps': 23553, 'loss/train': 2.005721390247345}
+01/29/2022 19:10:49 - INFO - codeparrot_training - Step 23554: {'lr': 0.00028985184447489874, 'samples': 4522560, 'steps': 23554, 'loss/train': 2.650689125061035}
+01/29/2022 19:10:53 - INFO - codeparrot_training - Step 23555: {'lr': 0.00028983569115643385, 'samples': 4522752, 'steps': 23555, 'loss/train': 1.0001804530620575}
+01/29/2022 19:10:58 - INFO - codeparrot_training - Step 23556: {'lr': 0.0002898195376673254, 'samples': 4522944, 'steps': 23556, 'loss/train': 1.5973536372184753}
+01/29/2022 19:11:02 - INFO - codeparrot_training - Step 23557: {'lr': 0.0002898033840076427, 'samples': 4523136, 'steps': 23557, 'loss/train': 1.8827658891677856}
+01/29/2022 19:11:06 - INFO - codeparrot_training - Step 23558: {'lr': 0.00028978723017745496, 'samples': 4523328, 'steps': 23558, 'loss/train': 1.9671898484230042}
+01/29/2022 19:11:10 - INFO - codeparrot_training - Step 23559: {'lr': 0.0002897710761768313, 'samples': 4523520, 'steps': 23559, 'loss/train': 0.7853535711765289}
+01/29/2022 19:11:15 - INFO - codeparrot_training - Step 23560: {'lr': 0.00028975492200584106, 'samples': 4523712, 'steps': 23560, 'loss/train': 2.0321706533432007}
+01/29/2022 19:11:20 - INFO - codeparrot_training - Step 23561: {'lr': 0.00028973876766455334, 'samples': 4523904, 'steps': 23561, 'loss/train': 2.468182861804962}
+01/29/2022 19:11:24 - INFO - codeparrot_training - Step 23562: {'lr': 0.00028972261315303736, 'samples': 4524096, 'steps': 23562, 'loss/train': 1.599661946296692}
+01/29/2022 19:11:28 - INFO - codeparrot_training - Step 23563: {'lr': 0.0002897064584713623, 'samples': 4524288, 'steps': 23563, 'loss/train': 1.8730634450912476}
+01/29/2022 19:11:32 - INFO - codeparrot_training - Step 23564: {'lr': 0.0002896903036195974, 'samples': 4524480, 'steps': 23564, 'loss/train': 0.20208900421857834}
+01/29/2022 19:11:37 - INFO - codeparrot_training - Step 23565: {'lr': 0.0002896741485978118, 'samples': 4524672, 'steps': 23565, 'loss/train': 1.4883933663368225}
+01/29/2022 19:11:44 - INFO - codeparrot_training - Step 23566: {'lr': 0.0002896579934060748, 'samples': 4524864, 'steps': 23566, 'loss/train': 1.0558072328567505}
+01/29/2022 19:11:48 - INFO - codeparrot_training - Step 23567: {'lr': 0.00028964183804445554, 'samples': 4525056, 'steps': 23567, 'loss/train': 1.8807506561279297}
+01/29/2022 19:11:52 - INFO - codeparrot_training - Step 23568: {'lr': 0.00028962568251302326, 'samples': 4525248, 'steps': 23568, 'loss/train': 2.109527885913849}
+01/29/2022 19:11:56 - INFO - codeparrot_training - Step 23569: {'lr': 0.00028960952681184705, 'samples': 4525440, 'steps': 23569, 'loss/train': 2.0506719946861267}
+01/29/2022 19:12:01 - INFO - codeparrot_training - Step 23570: {'lr': 0.00028959337094099617, 'samples': 4525632, 'steps': 23570, 'loss/train': 2.146498739719391}
+01/29/2022 19:12:06 - INFO - codeparrot_training - Step 23571: {'lr': 0.00028957721490054, 'samples': 4525824, 'steps': 23571, 'loss/train': 2.682684302330017}
+01/29/2022 19:12:10 - INFO - codeparrot_training - Step 23572: {'lr': 0.00028956105869054757, 'samples': 4526016, 'steps': 23572, 'loss/train': 1.5999755859375}
+01/29/2022 19:12:14 - INFO - codeparrot_training - Step 23573: {'lr': 0.0002895449023110881, 'samples': 4526208, 'steps': 23573, 'loss/train': 1.8513761758804321}
+01/29/2022 19:12:19 - INFO - codeparrot_training - Step 23574: {'lr': 0.00028952874576223083, 'samples': 4526400, 'steps': 23574, 'loss/train': 1.1775789856910706}
+01/29/2022 19:12:23 - INFO - codeparrot_training - Step 23575: {'lr': 0.000289512589044045, 'samples': 4526592, 'steps': 23575, 'loss/train': 1.6110325455665588}
+01/29/2022 19:12:28 - INFO - codeparrot_training - Step 23576: {'lr': 0.0002894964321565997, 'samples': 4526784, 'steps': 23576, 'loss/train': 2.0046944618225098}
+01/29/2022 19:12:32 - INFO - codeparrot_training - Step 23577: {'lr': 0.00028948027509996435, 'samples': 4526976, 'steps': 23577, 'loss/train': 1.1311782896518707}
+01/29/2022 19:12:36 - INFO - codeparrot_training - Step 23578: {'lr': 0.0002894641178742079, 'samples': 4527168, 'steps': 23578, 'loss/train': 2.17206734418869}
+01/29/2022 19:12:41 - INFO - codeparrot_training - Step 23579: {'lr': 0.0002894479604793998, 'samples': 4527360, 'steps': 23579, 'loss/train': 1.757177710533142}
+01/29/2022 19:12:45 - INFO - codeparrot_training - Step 23580: {'lr': 0.0002894318029156091, 'samples': 4527552, 'steps': 23580, 'loss/train': 2.186424493789673}
+01/29/2022 19:12:50 - INFO - codeparrot_training - Step 23581: {'lr': 0.0002894156451829051, 'samples': 4527744, 'steps': 23581, 'loss/train': 1.0172752439975739}
+01/29/2022 19:12:54 - INFO - codeparrot_training - Step 23582: {'lr': 0.000289399487281357, 'samples': 4527936, 'steps': 23582, 'loss/train': 1.7343057990074158}
+01/29/2022 19:12:59 - INFO - codeparrot_training - Step 23583: {'lr': 0.000289383329211034, 'samples': 4528128, 'steps': 23583, 'loss/train': 1.3295190632343292}
+01/29/2022 19:13:03 - INFO - codeparrot_training - Step 23584: {'lr': 0.0002893671709720054, 'samples': 4528320, 'steps': 23584, 'loss/train': 2.192379355430603}
+01/29/2022 19:13:10 - INFO - codeparrot_training - Step 23585: {'lr': 0.0002893510125643403, 'samples': 4528512, 'steps': 23585, 'loss/train': 1.437978744506836}
+01/29/2022 19:13:14 - INFO - codeparrot_training - Step 23586: {'lr': 0.0002893348539881079, 'samples': 4528704, 'steps': 23586, 'loss/train': 1.7104053497314453}
+01/29/2022 19:13:18 - INFO - codeparrot_training - Step 23587: {'lr': 0.0002893186952433775, 'samples': 4528896, 'steps': 23587, 'loss/train': 1.288418859243393}
+01/29/2022 19:13:23 - INFO - codeparrot_training - Step 23588: {'lr': 0.00028930253633021826, 'samples': 4529088, 'steps': 23588, 'loss/train': 1.702209234237671}
+01/29/2022 19:13:27 - INFO - codeparrot_training - Step 23589: {'lr': 0.0002892863772486995, 'samples': 4529280, 'steps': 23589, 'loss/train': 1.1587603390216827}
+01/29/2022 19:13:32 - INFO - codeparrot_training - Step 23590: {'lr': 0.0002892702179988903, 'samples': 4529472, 'steps': 23590, 'loss/train': 1.911069631576538}
+01/29/2022 19:13:36 - INFO - codeparrot_training - Step 23591: {'lr': 0.00028925405858086007, 'samples': 4529664, 'steps': 23591, 'loss/train': 1.5337275266647339}
+01/29/2022 19:13:40 - INFO - codeparrot_training - Step 23592: {'lr': 0.0002892378989946779, 'samples': 4529856, 'steps': 23592, 'loss/train': 2.29597145318985}
+01/29/2022 19:13:45 - INFO - codeparrot_training - Step 23593: {'lr': 0.00028922173924041296, 'samples': 4530048, 'steps': 23593, 'loss/train': 1.511126697063446}
+01/29/2022 19:13:49 - INFO - codeparrot_training - Step 23594: {'lr': 0.00028920557931813454, 'samples': 4530240, 'steps': 23594, 'loss/train': 2.927673876285553}
+01/29/2022 19:13:56 - INFO - codeparrot_training - Step 23595: {'lr': 0.00028918941922791185, 'samples': 4530432, 'steps': 23595, 'loss/train': 1.7779861092567444}
+01/29/2022 19:14:00 - INFO - codeparrot_training - Step 23596: {'lr': 0.00028917325896981417, 'samples': 4530624, 'steps': 23596, 'loss/train': 1.590164601802826}
+01/29/2022 19:14:05 - INFO - codeparrot_training - Step 23597: {'lr': 0.0002891570985439108, 'samples': 4530816, 'steps': 23597, 'loss/train': 2.3961090445518494}
+01/29/2022 19:14:09 - INFO - codeparrot_training - Step 23598: {'lr': 0.00028914093795027063, 'samples': 4531008, 'steps': 23598, 'loss/train': 1.22804456949234}
+01/29/2022 19:14:13 - INFO - codeparrot_training - Step 23599: {'lr': 0.0002891247771889633, 'samples': 4531200, 'steps': 23599, 'loss/train': 0.6672954261302948}
+01/29/2022 19:14:17 - INFO - codeparrot_training - Step 23600: {'lr': 0.00028910861626005774, 'samples': 4531392, 'steps': 23600, 'loss/train': 2.1550764441490173}
+01/29/2022 19:14:23 - INFO - codeparrot_training - Step 23601: {'lr': 0.0002890924551636234, 'samples': 4531584, 'steps': 23601, 'loss/train': 1.7952173352241516}
+01/29/2022 19:14:27 - INFO - codeparrot_training - Step 23602: {'lr': 0.00028907629389972924, 'samples': 4531776, 'steps': 23602, 'loss/train': 2.0719361901283264}
+01/29/2022 19:14:31 - INFO - codeparrot_training - Step 23603: {'lr': 0.00028906013246844474, 'samples': 4531968, 'steps': 23603, 'loss/train': 1.9146162271499634}
+01/29/2022 19:14:35 - INFO - codeparrot_training - Step 23604: {'lr': 0.000289043970869839, 'samples': 4532160, 'steps': 23604, 'loss/train': 1.4733530580997467}
+01/29/2022 19:14:40 - INFO - codeparrot_training - Step 23605: {'lr': 0.0002890278091039813, 'samples': 4532352, 'steps': 23605, 'loss/train': 1.8041006326675415}
+01/29/2022 19:14:45 - INFO - codeparrot_training - Step 23606: {'lr': 0.00028901164717094085, 'samples': 4532544, 'steps': 23606, 'loss/train': 1.6420056223869324}
+01/29/2022 19:14:49 - INFO - codeparrot_training - Step 23607: {'lr': 0.0002889954850707869, 'samples': 4532736, 'steps': 23607, 'loss/train': 1.7639518976211548}
+01/29/2022 19:14:53 - INFO - codeparrot_training - Step 23608: {'lr': 0.0002889793228035887, 'samples': 4532928, 'steps': 23608, 'loss/train': 1.0020563006401062}
+01/29/2022 19:14:57 - INFO - codeparrot_training - Step 23609: {'lr': 0.0002889631603694154, 'samples': 4533120, 'steps': 23609, 'loss/train': 1.9632509350776672}
+01/29/2022 19:15:02 - INFO - codeparrot_training - Step 23610: {'lr': 0.00028894699776833637, 'samples': 4533312, 'steps': 23610, 'loss/train': 2.0334352254867554}
+01/29/2022 19:15:09 - INFO - codeparrot_training - Step 23611: {'lr': 0.0002889308350004207, 'samples': 4533504, 'steps': 23611, 'loss/train': 2.1508902311325073}
+01/29/2022 19:15:13 - INFO - codeparrot_training - Step 23612: {'lr': 0.00028891467206573773, 'samples': 4533696, 'steps': 23612, 'loss/train': 2.1062373518943787}
+01/29/2022 19:15:17 - INFO - codeparrot_training - Step 23613: {'lr': 0.00028889850896435666, 'samples': 4533888, 'steps': 23613, 'loss/train': 1.3582224547863007}
+01/29/2022 19:15:21 - INFO - codeparrot_training - Step 23614: {'lr': 0.00028888234569634673, 'samples': 4534080, 'steps': 23614, 'loss/train': 1.3187231719493866}
+01/29/2022 19:15:26 - INFO - codeparrot_training - Step 23615: {'lr': 0.00028886618226177716, 'samples': 4534272, 'steps': 23615, 'loss/train': 1.0428127348423004}
+01/29/2022 19:15:31 - INFO - codeparrot_training - Step 23616: {'lr': 0.00028885001866071723, 'samples': 4534464, 'steps': 23616, 'loss/train': 1.4518929719924927}
+01/29/2022 19:15:35 - INFO - codeparrot_training - Step 23617: {'lr': 0.0002888338548932361, 'samples': 4534656, 'steps': 23617, 'loss/train': 1.3859311044216156}
+01/29/2022 19:15:39 - INFO - codeparrot_training - Step 23618: {'lr': 0.00028881769095940316, 'samples': 4534848, 'steps': 23618, 'loss/train': 1.0455459952354431}
+01/29/2022 19:15:43 - INFO - codeparrot_training - Step 23619: {'lr': 0.0002888015268592875, 'samples': 4535040, 'steps': 23619, 'loss/train': 1.89580500125885}
+01/29/2022 19:15:48 - INFO - codeparrot_training - Step 23620: {'lr': 0.0002887853625929584, 'samples': 4535232, 'steps': 23620, 'loss/train': 2.0361133217811584}
+01/29/2022 19:15:53 - INFO - codeparrot_training - Step 23621: {'lr': 0.00028876919816048517, 'samples': 4535424, 'steps': 23621, 'loss/train': 1.0005371868610382}
+01/29/2022 19:15:57 - INFO - codeparrot_training - Step 23622: {'lr': 0.00028875303356193697, 'samples': 4535616, 'steps': 23622, 'loss/train': 1.7976120114326477}
+01/29/2022 19:16:01 - INFO - codeparrot_training - Step 23623: {'lr': 0.0002887368687973831, 'samples': 4535808, 'steps': 23623, 'loss/train': 1.565843939781189}
+01/29/2022 19:16:06 - INFO - codeparrot_training - Step 23624: {'lr': 0.00028872070386689274, 'samples': 4536000, 'steps': 23624, 'loss/train': 1.8201919198036194}
+01/29/2022 19:16:10 - INFO - codeparrot_training - Step 23625: {'lr': 0.00028870453877053527, 'samples': 4536192, 'steps': 23625, 'loss/train': 2.30889755487442}
+01/29/2022 19:16:17 - INFO - codeparrot_training - Step 23626: {'lr': 0.00028868837350837977, 'samples': 4536384, 'steps': 23626, 'loss/train': 0.7231777310371399}
+01/29/2022 19:16:21 - INFO - codeparrot_training - Step 23627: {'lr': 0.0002886722080804956, 'samples': 4536576, 'steps': 23627, 'loss/train': 0.1940544918179512}
+01/29/2022 19:16:25 - INFO - codeparrot_training - Step 23628: {'lr': 0.000288656042486952, 'samples': 4536768, 'steps': 23628, 'loss/train': 2.192134737968445}
+01/29/2022 19:16:29 - INFO - codeparrot_training - Step 23629: {'lr': 0.0002886398767278181, 'samples': 4536960, 'steps': 23629, 'loss/train': 1.3044951260089874}
+01/29/2022 19:16:34 - INFO - codeparrot_training - Step 23630: {'lr': 0.0002886237108031633, 'samples': 4537152, 'steps': 23630, 'loss/train': 1.27476105093956}
+01/29/2022 19:16:39 - INFO - codeparrot_training - Step 23631: {'lr': 0.0002886075447130568, 'samples': 4537344, 'steps': 23631, 'loss/train': 1.9324843883514404}
+01/29/2022 19:16:43 - INFO - codeparrot_training - Step 23632: {'lr': 0.00028859137845756784, 'samples': 4537536, 'steps': 23632, 'loss/train': 1.5678538084030151}
+01/29/2022 19:16:47 - INFO - codeparrot_training - Step 23633: {'lr': 0.00028857521203676565, 'samples': 4537728, 'steps': 23633, 'loss/train': 1.8970937132835388}
+01/29/2022 19:16:51 - INFO - codeparrot_training - Step 23634: {'lr': 0.0002885590454507195, 'samples': 4537920, 'steps': 23634, 'loss/train': 2.4781769514083862}
+01/29/2022 19:16:56 - INFO - codeparrot_training - Step 23635: {'lr': 0.00028854287869949856, 'samples': 4538112, 'steps': 23635, 'loss/train': 1.6633277535438538}
+01/29/2022 19:17:01 - INFO - codeparrot_training - Step 23636: {'lr': 0.00028852671178317233, 'samples': 4538304, 'steps': 23636, 'loss/train': 1.2453313171863556}
+01/29/2022 19:17:05 - INFO - codeparrot_training - Step 23637: {'lr': 0.00028851054470180977, 'samples': 4538496, 'steps': 23637, 'loss/train': 1.7195470333099365}
+01/29/2022 19:17:09 - INFO - codeparrot_training - Step 23638: {'lr': 0.0002884943774554803, 'samples': 4538688, 'steps': 23638, 'loss/train': 0.6917873471975327}
+01/29/2022 19:17:14 - INFO - codeparrot_training - Step 23639: {'lr': 0.0002884782100442531, 'samples': 4538880, 'steps': 23639, 'loss/train': 1.4804065525531769}
+01/29/2022 19:17:18 - INFO - codeparrot_training - Step 23640: {'lr': 0.0002884620424681976, 'samples': 4539072, 'steps': 23640, 'loss/train': 1.6097263097763062}
+01/29/2022 19:17:25 - INFO - codeparrot_training - Step 23641: {'lr': 0.0002884458747273828, 'samples': 4539264, 'steps': 23641, 'loss/train': 1.4032634496688843}
+01/29/2022 19:17:29 - INFO - codeparrot_training - Step 23642: {'lr': 0.0002884297068218781, 'samples': 4539456, 'steps': 23642, 'loss/train': 1.9821245670318604}
+01/29/2022 19:17:33 - INFO - codeparrot_training - Step 23643: {'lr': 0.00028841353875175274, 'samples': 4539648, 'steps': 23643, 'loss/train': 0.9634590446949005}
+01/29/2022 19:17:38 - INFO - codeparrot_training - Step 23644: {'lr': 0.000288397370517076, 'samples': 4539840, 'steps': 23644, 'loss/train': 1.4657481908798218}
+01/29/2022 19:17:42 - INFO - codeparrot_training - Step 23645: {'lr': 0.0002883812021179171, 'samples': 4540032, 'steps': 23645, 'loss/train': 1.4241676926612854}
+01/29/2022 19:17:47 - INFO - codeparrot_training - Step 23646: {'lr': 0.0002883650335543453, 'samples': 4540224, 'steps': 23646, 'loss/train': 1.4703748226165771}
+01/29/2022 19:17:51 - INFO - codeparrot_training - Step 23647: {'lr': 0.00028834886482643, 'samples': 4540416, 'steps': 23647, 'loss/train': 1.4033105671405792}
+01/29/2022 19:17:55 - INFO - codeparrot_training - Step 23648: {'lr': 0.00028833269593424017, 'samples': 4540608, 'steps': 23648, 'loss/train': 1.2777427732944489}
+01/29/2022 19:18:00 - INFO - codeparrot_training - Step 23649: {'lr': 0.0002883165268778454, 'samples': 4540800, 'steps': 23649, 'loss/train': 1.685349941253662}
+01/29/2022 19:18:04 - INFO - codeparrot_training - Step 23650: {'lr': 0.00028830035765731464, 'samples': 4540992, 'steps': 23650, 'loss/train': 2.9324432015419006}
+01/29/2022 19:18:09 - INFO - codeparrot_training - Step 23651: {'lr': 0.0002882841882727174, 'samples': 4541184, 'steps': 23651, 'loss/train': 1.2459416091442108}
+01/29/2022 19:18:13 - INFO - codeparrot_training - Step 23652: {'lr': 0.00028826801872412284, 'samples': 4541376, 'steps': 23652, 'loss/train': 2.5213379859924316}
+01/29/2022 19:18:17 - INFO - codeparrot_training - Step 23653: {'lr': 0.00028825184901160023, 'samples': 4541568, 'steps': 23653, 'loss/train': 2.100342035293579}
+01/29/2022 19:18:22 - INFO - codeparrot_training - Step 23654: {'lr': 0.00028823567913521884, 'samples': 4541760, 'steps': 23654, 'loss/train': 2.1708794832229614}
+01/29/2022 19:18:26 - INFO - codeparrot_training - Step 23655: {'lr': 0.0002882195090950479, 'samples': 4541952, 'steps': 23655, 'loss/train': 1.84884774684906}
+01/29/2022 19:18:33 - INFO - codeparrot_training - Step 23656: {'lr': 0.00028820333889115684, 'samples': 4542144, 'steps': 23656, 'loss/train': 0.8114182949066162}
+01/29/2022 19:18:38 - INFO - codeparrot_training - Step 23657: {'lr': 0.0002881871685236147, 'samples': 4542336, 'steps': 23657, 'loss/train': 0.4368104785680771}
+01/29/2022 19:18:42 - INFO - codeparrot_training - Step 23658: {'lr': 0.0002881709979924909, 'samples': 4542528, 'steps': 23658, 'loss/train': 1.8903606534004211}
+01/29/2022 19:18:46 - INFO - codeparrot_training - Step 23659: {'lr': 0.00028815482729785467, 'samples': 4542720, 'steps': 23659, 'loss/train': 0.898266077041626}
+01/29/2022 19:18:50 - INFO - codeparrot_training - Step 23660: {'lr': 0.00028813865643977527, 'samples': 4542912, 'steps': 23660, 'loss/train': 1.8084413409233093}
+01/29/2022 19:18:55 - INFO - codeparrot_training - Step 23661: {'lr': 0.000288122485418322, 'samples': 4543104, 'steps': 23661, 'loss/train': 1.5018451809883118}
+01/29/2022 19:19:00 - INFO - codeparrot_training - Step 23662: {'lr': 0.0002881063142335641, 'samples': 4543296, 'steps': 23662, 'loss/train': 2.0607866048812866}
+01/29/2022 19:19:04 - INFO - codeparrot_training - Step 23663: {'lr': 0.0002880901428855708, 'samples': 4543488, 'steps': 23663, 'loss/train': 0.8942987322807312}
+01/29/2022 19:19:08 - INFO - codeparrot_training - Step 23664: {'lr': 0.0002880739713744114, 'samples': 4543680, 'steps': 23664, 'loss/train': 1.0721558332443237}
+01/29/2022 19:19:12 - INFO - codeparrot_training - Step 23665: {'lr': 0.00028805779970015525, 'samples': 4543872, 'steps': 23665, 'loss/train': 0.6317817270755768}
+01/29/2022 19:19:17 - INFO - codeparrot_training - Step 23666: {'lr': 0.00028804162786287156, 'samples': 4544064, 'steps': 23666, 'loss/train': 1.6330751180648804}
+01/29/2022 19:19:22 - INFO - codeparrot_training - Step 23667: {'lr': 0.0002880254558626297, 'samples': 4544256, 'steps': 23667, 'loss/train': 1.716603398323059}
+01/29/2022 19:19:26 - INFO - codeparrot_training - Step 23668: {'lr': 0.0002880092836994987, 'samples': 4544448, 'steps': 23668, 'loss/train': 0.9604514837265015}
+01/29/2022 19:19:30 - INFO - codeparrot_training - Step 23669: {'lr': 0.0002879931113735482, 'samples': 4544640, 'steps': 23669, 'loss/train': 1.6209686994552612}
+01/29/2022 19:19:34 - INFO - codeparrot_training - Step 23670: {'lr': 0.0002879769388848471, 'samples': 4544832, 'steps': 23670, 'loss/train': 2.439521849155426}
+01/29/2022 19:19:41 - INFO - codeparrot_training - Step 23671: {'lr': 0.0002879607662334649, 'samples': 4545024, 'steps': 23671, 'loss/train': 1.1243546605110168}
+01/29/2022 19:19:45 - INFO - codeparrot_training - Step 23672: {'lr': 0.0002879445934194709, 'samples': 4545216, 'steps': 23672, 'loss/train': 1.020241767168045}
+01/29/2022 19:19:50 - INFO - codeparrot_training - Step 23673: {'lr': 0.0002879284204429342, 'samples': 4545408, 'steps': 23673, 'loss/train': 0.5058597028255463}
+01/29/2022 19:19:54 - INFO - codeparrot_training - Step 23674: {'lr': 0.0002879122473039243, 'samples': 4545600, 'steps': 23674, 'loss/train': 2.0158562064170837}
+01/29/2022 19:19:58 - INFO - codeparrot_training - Step 23675: {'lr': 0.0002878960740025103, 'samples': 4545792, 'steps': 23675, 'loss/train': 1.4889556467533112}
+01/29/2022 19:20:04 - INFO - codeparrot_training - Step 23676: {'lr': 0.00028787990053876156, 'samples': 4545984, 'steps': 23676, 'loss/train': 1.0995689928531647}
+01/29/2022 19:20:08 - INFO - codeparrot_training - Step 23677: {'lr': 0.00028786372691274735, 'samples': 4546176, 'steps': 23677, 'loss/train': 0.9373913705348969}
+01/29/2022 19:20:12 - INFO - codeparrot_training - Step 23678: {'lr': 0.00028784755312453704, 'samples': 4546368, 'steps': 23678, 'loss/train': 1.635275959968567}
+01/29/2022 19:20:16 - INFO - codeparrot_training - Step 23679: {'lr': 0.0002878313791741997, 'samples': 4546560, 'steps': 23679, 'loss/train': 1.503688931465149}
+01/29/2022 19:20:20 - INFO - codeparrot_training - Step 23680: {'lr': 0.00028781520506180485, 'samples': 4546752, 'steps': 23680, 'loss/train': 1.2362979054450989}
+01/29/2022 19:20:26 - INFO - codeparrot_training - Step 23681: {'lr': 0.0002877990307874216, 'samples': 4546944, 'steps': 23681, 'loss/train': 2.723007917404175}
+01/29/2022 19:20:30 - INFO - codeparrot_training - Step 23682: {'lr': 0.00028778285635111935, 'samples': 4547136, 'steps': 23682, 'loss/train': 1.6047493815422058}
+01/29/2022 19:20:34 - INFO - codeparrot_training - Step 23683: {'lr': 0.0002877666817529673, 'samples': 4547328, 'steps': 23683, 'loss/train': 1.4539070427417755}
+01/29/2022 19:20:38 - INFO - codeparrot_training - Step 23684: {'lr': 0.0002877505069930348, 'samples': 4547520, 'steps': 23684, 'loss/train': 1.8380354046821594}
+01/29/2022 19:20:43 - INFO - codeparrot_training - Step 23685: {'lr': 0.0002877343320713911, 'samples': 4547712, 'steps': 23685, 'loss/train': 2.0078691244125366}
+01/29/2022 19:20:50 - INFO - codeparrot_training - Step 23686: {'lr': 0.0002877181569881055, 'samples': 4547904, 'steps': 23686, 'loss/train': 1.735472023487091}
+01/29/2022 19:20:54 - INFO - codeparrot_training - Step 23687: {'lr': 0.00028770198174324737, 'samples': 4548096, 'steps': 23687, 'loss/train': 1.4380398094654083}
+01/29/2022 19:20:58 - INFO - codeparrot_training - Step 23688: {'lr': 0.00028768580633688586, 'samples': 4548288, 'steps': 23688, 'loss/train': 1.53056001663208}
+01/29/2022 19:21:02 - INFO - codeparrot_training - Step 23689: {'lr': 0.00028766963076909033, 'samples': 4548480, 'steps': 23689, 'loss/train': 1.2517696022987366}
+01/29/2022 19:21:07 - INFO - codeparrot_training - Step 23690: {'lr': 0.00028765345503993, 'samples': 4548672, 'steps': 23690, 'loss/train': 1.582670509815216}
+01/29/2022 19:21:12 - INFO - codeparrot_training - Step 23691: {'lr': 0.0002876372791494743, 'samples': 4548864, 'steps': 23691, 'loss/train': 1.9783621430397034}
+01/29/2022 19:21:16 - INFO - codeparrot_training - Step 23692: {'lr': 0.00028762110309779246, 'samples': 4549056, 'steps': 23692, 'loss/train': 0.7423478364944458}
+01/29/2022 19:21:20 - INFO - codeparrot_training - Step 23693: {'lr': 0.0002876049268849537, 'samples': 4549248, 'steps': 23693, 'loss/train': 1.3109329640865326}
+01/29/2022 19:21:24 - INFO - codeparrot_training - Step 23694: {'lr': 0.0002875887505110274, 'samples': 4549440, 'steps': 23694, 'loss/train': 1.7189496159553528}
+01/29/2022 19:21:29 - INFO - codeparrot_training - Step 23695: {'lr': 0.00028757257397608285, 'samples': 4549632, 'steps': 23695, 'loss/train': 0.1487370729446411}
+01/29/2022 19:21:34 - INFO - codeparrot_training - Step 23696: {'lr': 0.0002875563972801893, 'samples': 4549824, 'steps': 23696, 'loss/train': 0.7647911310195923}
+01/29/2022 19:21:38 - INFO - codeparrot_training - Step 23697: {'lr': 0.0002875402204234161, 'samples': 4550016, 'steps': 23697, 'loss/train': 0.8909223675727844}
+01/29/2022 19:21:42 - INFO - codeparrot_training - Step 23698: {'lr': 0.0002875240434058324, 'samples': 4550208, 'steps': 23698, 'loss/train': 1.7898362874984741}
+01/29/2022 19:21:46 - INFO - codeparrot_training - Step 23699: {'lr': 0.0002875078662275078, 'samples': 4550400, 'steps': 23699, 'loss/train': 2.1902164220809937}
+01/29/2022 19:21:51 - INFO - codeparrot_training - Step 23700: {'lr': 0.00028749168888851125, 'samples': 4550592, 'steps': 23700, 'loss/train': 1.5495579242706299}
+01/29/2022 19:21:58 - INFO - codeparrot_training - Step 23701: {'lr': 0.0002874755113889123, 'samples': 4550784, 'steps': 23701, 'loss/train': 1.3093827366828918}
+01/29/2022 19:22:02 - INFO - codeparrot_training - Step 23702: {'lr': 0.0002874593337287801, 'samples': 4550976, 'steps': 23702, 'loss/train': 1.3923381865024567}
+01/29/2022 19:22:06 - INFO - codeparrot_training - Step 23703: {'lr': 0.00028744315590818406, 'samples': 4551168, 'steps': 23703, 'loss/train': 0.7172324806451797}
+01/29/2022 19:22:10 - INFO - codeparrot_training - Step 23704: {'lr': 0.0002874269779271934, 'samples': 4551360, 'steps': 23704, 'loss/train': 2.0920695662498474}
+01/29/2022 19:22:15 - INFO - codeparrot_training - Step 23705: {'lr': 0.0002874107997858775, 'samples': 4551552, 'steps': 23705, 'loss/train': 1.9279126524925232}
+01/29/2022 19:22:20 - INFO - codeparrot_training - Step 23706: {'lr': 0.00028739462148430554, 'samples': 4551744, 'steps': 23706, 'loss/train': 0.4689645320177078}
+01/29/2022 19:22:24 - INFO - codeparrot_training - Step 23707: {'lr': 0.0002873784430225469, 'samples': 4551936, 'steps': 23707, 'loss/train': 1.6635469794273376}
+01/29/2022 19:22:28 - INFO - codeparrot_training - Step 23708: {'lr': 0.000287362264400671, 'samples': 4552128, 'steps': 23708, 'loss/train': 1.9938942790031433}
+01/29/2022 19:22:32 - INFO - codeparrot_training - Step 23709: {'lr': 0.00028734608561874686, 'samples': 4552320, 'steps': 23709, 'loss/train': 1.080033391714096}
+01/29/2022 19:22:37 - INFO - codeparrot_training - Step 23710: {'lr': 0.0002873299066768441, 'samples': 4552512, 'steps': 23710, 'loss/train': 2.1379358768463135}
+01/29/2022 19:22:42 - INFO - codeparrot_training - Step 23711: {'lr': 0.0002873137275750317, 'samples': 4552704, 'steps': 23711, 'loss/train': 1.4008140563964844}
+01/29/2022 19:22:46 - INFO - codeparrot_training - Step 23712: {'lr': 0.0002872975483133793, 'samples': 4552896, 'steps': 23712, 'loss/train': 2.440418243408203}
+01/29/2022 19:22:50 - INFO - codeparrot_training - Step 23713: {'lr': 0.00028728136889195595, 'samples': 4553088, 'steps': 23713, 'loss/train': 1.624002456665039}
+01/29/2022 19:22:54 - INFO - codeparrot_training - Step 23714: {'lr': 0.0002872651893108311, 'samples': 4553280, 'steps': 23714, 'loss/train': 2.178945779800415}
+01/29/2022 19:22:59 - INFO - codeparrot_training - Step 23715: {'lr': 0.000287249009570074, 'samples': 4553472, 'steps': 23715, 'loss/train': 1.3063002526760101}
+01/29/2022 19:23:06 - INFO - codeparrot_training - Step 23716: {'lr': 0.000287232829669754, 'samples': 4553664, 'steps': 23716, 'loss/train': 1.0008048713207245}
+01/29/2022 19:23:10 - INFO - codeparrot_training - Step 23717: {'lr': 0.0002872166496099403, 'samples': 4553856, 'steps': 23717, 'loss/train': 1.2476704716682434}
+01/29/2022 19:23:14 - INFO - codeparrot_training - Step 23718: {'lr': 0.0002872004693907024, 'samples': 4554048, 'steps': 23718, 'loss/train': 1.477052092552185}
+01/29/2022 19:23:18 - INFO - codeparrot_training - Step 23719: {'lr': 0.0002871842890121094, 'samples': 4554240, 'steps': 23719, 'loss/train': 4.027720928192139}
+01/29/2022 19:23:23 - INFO - codeparrot_training - Step 23720: {'lr': 0.0002871681084742308, 'samples': 4554432, 'steps': 23720, 'loss/train': 1.648006796836853}
+01/29/2022 19:23:28 - INFO - codeparrot_training - Step 23721: {'lr': 0.0002871519277771358, 'samples': 4554624, 'steps': 23721, 'loss/train': 1.3107689023017883}
+01/29/2022 19:23:32 - INFO - codeparrot_training - Step 23722: {'lr': 0.0002871357469208937, 'samples': 4554816, 'steps': 23722, 'loss/train': 2.1247002482414246}
+01/29/2022 19:23:36 - INFO - codeparrot_training - Step 23723: {'lr': 0.0002871195659055739, 'samples': 4555008, 'steps': 23723, 'loss/train': 1.9375355243682861}
+01/29/2022 19:23:40 - INFO - codeparrot_training - Step 23724: {'lr': 0.0002871033847312456, 'samples': 4555200, 'steps': 23724, 'loss/train': 1.8329717516899109}
+01/29/2022 19:23:45 - INFO - codeparrot_training - Step 23725: {'lr': 0.0002870872033979782, 'samples': 4555392, 'steps': 23725, 'loss/train': 0.8448777794837952}
+01/29/2022 19:23:52 - INFO - codeparrot_training - Step 23726: {'lr': 0.000287071021905841, 'samples': 4555584, 'steps': 23726, 'loss/train': 1.02071812748909}
+01/29/2022 19:23:56 - INFO - codeparrot_training - Step 23727: {'lr': 0.00028705484025490333, 'samples': 4555776, 'steps': 23727, 'loss/train': 1.5726622939109802}
+01/29/2022 19:24:00 - INFO - codeparrot_training - Step 23728: {'lr': 0.0002870386584452345, 'samples': 4555968, 'steps': 23728, 'loss/train': 1.6203492879867554}
+01/29/2022 19:24:05 - INFO - codeparrot_training - Step 23729: {'lr': 0.00028702247647690383, 'samples': 4556160, 'steps': 23729, 'loss/train': 1.7342959642410278}
+01/29/2022 19:24:09 - INFO - codeparrot_training - Step 23730: {'lr': 0.0002870062943499806, 'samples': 4556352, 'steps': 23730, 'loss/train': 1.0411593317985535}
+01/29/2022 19:24:14 - INFO - codeparrot_training - Step 23731: {'lr': 0.0002869901120645341, 'samples': 4556544, 'steps': 23731, 'loss/train': 0.5245552808046341}
+01/29/2022 19:24:18 - INFO - codeparrot_training - Step 23732: {'lr': 0.0002869739296206338, 'samples': 4556736, 'steps': 23732, 'loss/train': 2.3226277828216553}
+01/29/2022 19:24:22 - INFO - codeparrot_training - Step 23733: {'lr': 0.0002869577470183489, 'samples': 4556928, 'steps': 23733, 'loss/train': 1.2922874987125397}
+01/29/2022 19:24:27 - INFO - codeparrot_training - Step 23734: {'lr': 0.00028694156425774874, 'samples': 4557120, 'steps': 23734, 'loss/train': 2.087547719478607}
+01/29/2022 19:24:31 - INFO - codeparrot_training - Step 23735: {'lr': 0.00028692538133890267, 'samples': 4557312, 'steps': 23735, 'loss/train': 0.8362129926681519}
+01/29/2022 19:24:36 - INFO - codeparrot_training - Step 23736: {'lr': 0.00028690919826188, 'samples': 4557504, 'steps': 23736, 'loss/train': 2.1496256589889526}
+01/29/2022 19:24:40 - INFO - codeparrot_training - Step 23737: {'lr': 0.00028689301502674995, 'samples': 4557696, 'steps': 23737, 'loss/train': 1.5346455574035645}
+01/29/2022 19:24:44 - INFO - codeparrot_training - Step 23738: {'lr': 0.00028687683163358197, 'samples': 4557888, 'steps': 23738, 'loss/train': 0.8892926573753357}
+01/29/2022 19:24:49 - INFO - codeparrot_training - Step 23739: {'lr': 0.00028686064808244546, 'samples': 4558080, 'steps': 23739, 'loss/train': 2.0266106128692627}
+01/29/2022 19:24:53 - INFO - codeparrot_training - Step 23740: {'lr': 0.0002868444643734095, 'samples': 4558272, 'steps': 23740, 'loss/train': 1.892072081565857}
+01/29/2022 19:24:58 - INFO - codeparrot_training - Step 23741: {'lr': 0.00028682828050654365, 'samples': 4558464, 'steps': 23741, 'loss/train': 2.1624661087989807}
+01/29/2022 19:25:02 - INFO - codeparrot_training - Step 23742: {'lr': 0.0002868120964819171, 'samples': 4558656, 'steps': 23742, 'loss/train': 1.6489394903182983}
+01/29/2022 19:25:07 - INFO - codeparrot_training - Step 23743: {'lr': 0.0002867959122995992, 'samples': 4558848, 'steps': 23743, 'loss/train': 0.7138748914003372}
+01/29/2022 19:25:11 - INFO - codeparrot_training - Step 23744: {'lr': 0.0002867797279596593, 'samples': 4559040, 'steps': 23744, 'loss/train': 0.12176494672894478}
+01/29/2022 19:25:15 - INFO - codeparrot_training - Step 23745: {'lr': 0.0002867635434621668, 'samples': 4559232, 'steps': 23745, 'loss/train': 1.2917332649230957}
+01/29/2022 19:25:22 - INFO - codeparrot_training - Step 23746: {'lr': 0.0002867473588071909, 'samples': 4559424, 'steps': 23746, 'loss/train': 1.7220810055732727}
+01/29/2022 19:25:26 - INFO - codeparrot_training - Step 23747: {'lr': 0.00028673117399480096, 'samples': 4559616, 'steps': 23747, 'loss/train': 1.638944149017334}
+01/29/2022 19:25:30 - INFO - codeparrot_training - Step 23748: {'lr': 0.00028671498902506636, 'samples': 4559808, 'steps': 23748, 'loss/train': 1.7566072940826416}
+01/29/2022 19:25:35 - INFO - codeparrot_training - Step 23749: {'lr': 0.00028669880389805647, 'samples': 4560000, 'steps': 23749, 'loss/train': 1.0826734006404877}
+01/29/2022 19:25:39 - INFO - codeparrot_training - Step 23750: {'lr': 0.00028668261861384045, 'samples': 4560192, 'steps': 23750, 'loss/train': 1.6565784215927124}
+01/29/2022 19:25:44 - INFO - codeparrot_training - Step 23751: {'lr': 0.00028666643317248777, 'samples': 4560384, 'steps': 23751, 'loss/train': 2.1252753138542175}
+01/29/2022 19:25:48 - INFO - codeparrot_training - Step 23752: {'lr': 0.00028665024757406775, 'samples': 4560576, 'steps': 23752, 'loss/train': 0.653294026851654}
+01/29/2022 19:25:52 - INFO - codeparrot_training - Step 23753: {'lr': 0.0002866340618186497, 'samples': 4560768, 'steps': 23753, 'loss/train': 2.214840888977051}
+01/29/2022 19:25:57 - INFO - codeparrot_training - Step 23754: {'lr': 0.00028661787590630297, 'samples': 4560960, 'steps': 23754, 'loss/train': 1.1759944260120392}
+01/29/2022 19:26:01 - INFO - codeparrot_training - Step 23755: {'lr': 0.00028660168983709683, 'samples': 4561152, 'steps': 23755, 'loss/train': 1.716211974620819}
+01/29/2022 19:26:06 - INFO - codeparrot_training - Step 23756: {'lr': 0.00028658550361110075, 'samples': 4561344, 'steps': 23756, 'loss/train': 1.3853089213371277}
+01/29/2022 19:26:10 - INFO - codeparrot_training - Step 23757: {'lr': 0.000286569317228384, 'samples': 4561536, 'steps': 23757, 'loss/train': 1.2177854776382446}
+01/29/2022 19:26:14 - INFO - codeparrot_training - Step 23758: {'lr': 0.00028655313068901586, 'samples': 4561728, 'steps': 23758, 'loss/train': 1.4455585777759552}
+01/29/2022 19:26:19 - INFO - codeparrot_training - Step 23759: {'lr': 0.0002865369439930657, 'samples': 4561920, 'steps': 23759, 'loss/train': 1.3968193531036377}
+01/29/2022 19:26:23 - INFO - codeparrot_training - Step 23760: {'lr': 0.00028652075714060294, 'samples': 4562112, 'steps': 23760, 'loss/train': 1.2114752233028412}
+01/29/2022 19:26:30 - INFO - codeparrot_training - Step 23761: {'lr': 0.0002865045701316968, 'samples': 4562304, 'steps': 23761, 'loss/train': 1.4791530668735504}
+01/29/2022 19:26:34 - INFO - codeparrot_training - Step 23762: {'lr': 0.00028648838296641666, 'samples': 4562496, 'steps': 23762, 'loss/train': 1.3298044502735138}
+01/29/2022 19:26:38 - INFO - codeparrot_training - Step 23763: {'lr': 0.00028647219564483195, 'samples': 4562688, 'steps': 23763, 'loss/train': 1.910481870174408}
+01/29/2022 19:26:42 - INFO - codeparrot_training - Step 23764: {'lr': 0.00028645600816701186, 'samples': 4562880, 'steps': 23764, 'loss/train': 1.7682952880859375}
+01/29/2022 19:26:47 - INFO - codeparrot_training - Step 23765: {'lr': 0.00028643982053302584, 'samples': 4563072, 'steps': 23765, 'loss/train': 1.5082772970199585}
+01/29/2022 19:26:52 - INFO - codeparrot_training - Step 23766: {'lr': 0.00028642363274294317, 'samples': 4563264, 'steps': 23766, 'loss/train': 2.0256861448287964}
+01/29/2022 19:26:56 - INFO - codeparrot_training - Step 23767: {'lr': 0.0002864074447968333, 'samples': 4563456, 'steps': 23767, 'loss/train': 1.7174050211906433}
+01/29/2022 19:27:00 - INFO - codeparrot_training - Step 23768: {'lr': 0.0002863912566947654, 'samples': 4563648, 'steps': 23768, 'loss/train': 0.7418636083602905}
+01/29/2022 19:27:04 - INFO - codeparrot_training - Step 23769: {'lr': 0.000286375068436809, 'samples': 4563840, 'steps': 23769, 'loss/train': 1.9094980359077454}
+01/29/2022 19:27:09 - INFO - codeparrot_training - Step 23770: {'lr': 0.00028635888002303324, 'samples': 4564032, 'steps': 23770, 'loss/train': 1.7938088178634644}
+01/29/2022 19:27:16 - INFO - codeparrot_training - Step 23771: {'lr': 0.00028634269145350765, 'samples': 4564224, 'steps': 23771, 'loss/train': 1.3138563930988312}
+01/29/2022 19:27:20 - INFO - codeparrot_training - Step 23772: {'lr': 0.00028632650272830153, 'samples': 4564416, 'steps': 23772, 'loss/train': 1.6065797209739685}
+01/29/2022 19:27:24 - INFO - codeparrot_training - Step 23773: {'lr': 0.00028631031384748426, 'samples': 4564608, 'steps': 23773, 'loss/train': 1.501475214958191}
+01/29/2022 19:27:28 - INFO - codeparrot_training - Step 23774: {'lr': 0.000286294124811125, 'samples': 4564800, 'steps': 23774, 'loss/train': 1.3008224666118622}
+01/29/2022 19:27:32 - INFO - codeparrot_training - Step 23775: {'lr': 0.0002862779356192933, 'samples': 4564992, 'steps': 23775, 'loss/train': 1.6909782886505127}
+01/29/2022 19:27:38 - INFO - codeparrot_training - Step 23776: {'lr': 0.0002862617462720584, 'samples': 4565184, 'steps': 23776, 'loss/train': 0.7725518345832825}
+01/29/2022 19:27:42 - INFO - codeparrot_training - Step 23777: {'lr': 0.00028624555676948975, 'samples': 4565376, 'steps': 23777, 'loss/train': 1.2059840261936188}
+01/29/2022 19:27:46 - INFO - codeparrot_training - Step 23778: {'lr': 0.00028622936711165665, 'samples': 4565568, 'steps': 23778, 'loss/train': 4.271234035491943}
+01/29/2022 19:27:51 - INFO - codeparrot_training - Step 23779: {'lr': 0.00028621317729862837, 'samples': 4565760, 'steps': 23779, 'loss/train': 1.623085856437683}
+01/29/2022 19:27:56 - INFO - codeparrot_training - Step 23780: {'lr': 0.00028619698733047444, 'samples': 4565952, 'steps': 23780, 'loss/train': 1.645267367362976}
+01/29/2022 19:28:00 - INFO - codeparrot_training - Step 23781: {'lr': 0.0002861807972072641, 'samples': 4566144, 'steps': 23781, 'loss/train': 1.791879415512085}
+01/29/2022 19:28:04 - INFO - codeparrot_training - Step 23782: {'lr': 0.0002861646069290667, 'samples': 4566336, 'steps': 23782, 'loss/train': 1.2174807786941528}
+01/29/2022 19:28:08 - INFO - codeparrot_training - Step 23783: {'lr': 0.0002861484164959515, 'samples': 4566528, 'steps': 23783, 'loss/train': 1.5905728340148926}
+01/29/2022 19:28:13 - INFO - codeparrot_training - Step 23784: {'lr': 0.0002861322259079881, 'samples': 4566720, 'steps': 23784, 'loss/train': 1.6962445378303528}
+01/29/2022 19:28:20 - INFO - codeparrot_training - Step 23785: {'lr': 0.00028611603516524566, 'samples': 4566912, 'steps': 23785, 'loss/train': 2.1118746399879456}
+01/29/2022 19:28:24 - INFO - codeparrot_training - Step 23786: {'lr': 0.00028609984426779364, 'samples': 4567104, 'steps': 23786, 'loss/train': 1.920285165309906}
+01/29/2022 19:28:28 - INFO - codeparrot_training - Step 23787: {'lr': 0.0002860836532157012, 'samples': 4567296, 'steps': 23787, 'loss/train': 1.371019721031189}
+01/29/2022 19:28:32 - INFO - codeparrot_training - Step 23788: {'lr': 0.000286067462009038, 'samples': 4567488, 'steps': 23788, 'loss/train': 1.7535801529884338}
+01/29/2022 19:28:36 - INFO - codeparrot_training - Step 23789: {'lr': 0.00028605127064787315, 'samples': 4567680, 'steps': 23789, 'loss/train': 1.6762797832489014}
+01/29/2022 19:28:42 - INFO - codeparrot_training - Step 23790: {'lr': 0.00028603507913227617, 'samples': 4567872, 'steps': 23790, 'loss/train': 1.8376827836036682}
+01/29/2022 19:28:46 - INFO - codeparrot_training - Step 23791: {'lr': 0.00028601888746231633, 'samples': 4568064, 'steps': 23791, 'loss/train': 7.076035737991333}
+01/29/2022 19:28:50 - INFO - codeparrot_training - Step 23792: {'lr': 0.00028600269563806304, 'samples': 4568256, 'steps': 23792, 'loss/train': 2.0563488006591797}
+01/29/2022 19:28:54 - INFO - codeparrot_training - Step 23793: {'lr': 0.0002859865036595856, 'samples': 4568448, 'steps': 23793, 'loss/train': 1.8380098342895508}
+01/29/2022 19:28:59 - INFO - codeparrot_training - Step 23794: {'lr': 0.0002859703115269534, 'samples': 4568640, 'steps': 23794, 'loss/train': 1.7151578664779663}
+01/29/2022 19:29:04 - INFO - codeparrot_training - Step 23795: {'lr': 0.0002859541192402359, 'samples': 4568832, 'steps': 23795, 'loss/train': 1.7719531059265137}
+01/29/2022 19:29:08 - INFO - codeparrot_training - Step 23796: {'lr': 0.00028593792679950227, 'samples': 4569024, 'steps': 23796, 'loss/train': 1.539851188659668}
+01/29/2022 19:29:12 - INFO - codeparrot_training - Step 23797: {'lr': 0.00028592173420482206, 'samples': 4569216, 'steps': 23797, 'loss/train': 1.4888718724250793}
+01/29/2022 19:29:17 - INFO - codeparrot_training - Step 23798: {'lr': 0.0002859055414562644, 'samples': 4569408, 'steps': 23798, 'loss/train': 1.5513405203819275}
+01/29/2022 19:29:21 - INFO - codeparrot_training - Step 23799: {'lr': 0.00028588934855389885, 'samples': 4569600, 'steps': 23799, 'loss/train': 1.4778040945529938}
+01/29/2022 19:29:27 - INFO - codeparrot_training - Step 23800: {'lr': 0.0002858731554977948, 'samples': 4569792, 'steps': 23800, 'loss/train': 1.8277117609977722}
+01/29/2022 19:29:31 - INFO - codeparrot_training - Step 23801: {'lr': 0.00028585696228802153, 'samples': 4569984, 'steps': 23801, 'loss/train': 2.074763596057892}
+01/29/2022 19:29:35 - INFO - codeparrot_training - Step 23802: {'lr': 0.0002858407689246484, 'samples': 4570176, 'steps': 23802, 'loss/train': 1.2895670235157013}
+01/29/2022 19:29:39 - INFO - codeparrot_training - Step 23803: {'lr': 0.0002858245754077448, 'samples': 4570368, 'steps': 23803, 'loss/train': 0.1970665007829666}
+01/29/2022 19:29:44 - INFO - codeparrot_training - Step 23804: {'lr': 0.0002858083817373801, 'samples': 4570560, 'steps': 23804, 'loss/train': 0.7052941471338272}
+01/29/2022 19:29:51 - INFO - codeparrot_training - Step 23805: {'lr': 0.00028579218791362367, 'samples': 4570752, 'steps': 23805, 'loss/train': 1.0911725163459778}
+01/29/2022 19:29:55 - INFO - codeparrot_training - Step 23806: {'lr': 0.0002857759939365449, 'samples': 4570944, 'steps': 23806, 'loss/train': 1.2738105654716492}
+01/29/2022 19:29:59 - INFO - codeparrot_training - Step 23807: {'lr': 0.000285759799806213, 'samples': 4571136, 'steps': 23807, 'loss/train': 2.2049102783203125}
+01/29/2022 19:30:03 - INFO - codeparrot_training - Step 23808: {'lr': 0.00028574360552269755, 'samples': 4571328, 'steps': 23808, 'loss/train': 1.3672296702861786}
+01/29/2022 19:30:07 - INFO - codeparrot_training - Step 23809: {'lr': 0.0002857274110860679, 'samples': 4571520, 'steps': 23809, 'loss/train': 1.8350075483322144}
+01/29/2022 19:30:13 - INFO - codeparrot_training - Step 23810: {'lr': 0.00028571121649639337, 'samples': 4571712, 'steps': 23810, 'loss/train': 1.603756070137024}
+01/29/2022 19:30:17 - INFO - codeparrot_training - Step 23811: {'lr': 0.0002856950217537432, 'samples': 4571904, 'steps': 23811, 'loss/train': 0.9147336781024933}
+01/29/2022 19:30:22 - INFO - codeparrot_training - Step 23812: {'lr': 0.000285678826858187, 'samples': 4572096, 'steps': 23812, 'loss/train': 1.9245082139968872}
+01/29/2022 19:30:26 - INFO - codeparrot_training - Step 23813: {'lr': 0.000285662631809794, 'samples': 4572288, 'steps': 23813, 'loss/train': 2.3410520553588867}
+01/29/2022 19:30:30 - INFO - codeparrot_training - Step 23814: {'lr': 0.0002856464366086336, 'samples': 4572480, 'steps': 23814, 'loss/train': 1.513264775276184}
+01/29/2022 19:30:37 - INFO - codeparrot_training - Step 23815: {'lr': 0.0002856302412547752, 'samples': 4572672, 'steps': 23815, 'loss/train': 1.2241946160793304}
+01/29/2022 19:30:41 - INFO - codeparrot_training - Step 23816: {'lr': 0.0002856140457482882, 'samples': 4572864, 'steps': 23816, 'loss/train': 1.3713993430137634}
+01/29/2022 19:30:45 - INFO - codeparrot_training - Step 23817: {'lr': 0.0002855978500892419, 'samples': 4573056, 'steps': 23817, 'loss/train': 0.9049513041973114}
+01/29/2022 19:30:49 - INFO - codeparrot_training - Step 23818: {'lr': 0.00028558165427770567, 'samples': 4573248, 'steps': 23818, 'loss/train': 2.162637948989868}
+01/29/2022 19:30:54 - INFO - codeparrot_training - Step 23819: {'lr': 0.00028556545831374903, 'samples': 4573440, 'steps': 23819, 'loss/train': 0.9633722305297852}
+01/29/2022 19:30:59 - INFO - codeparrot_training - Step 23820: {'lr': 0.0002855492621974411, 'samples': 4573632, 'steps': 23820, 'loss/train': 2.0418970584869385}
+01/29/2022 19:31:03 - INFO - codeparrot_training - Step 23821: {'lr': 0.0002855330659288516, 'samples': 4573824, 'steps': 23821, 'loss/train': 0.2941695749759674}
+01/29/2022 19:31:07 - INFO - codeparrot_training - Step 23822: {'lr': 0.00028551686950804964, 'samples': 4574016, 'steps': 23822, 'loss/train': 1.7458866834640503}
+01/29/2022 19:31:11 - INFO - codeparrot_training - Step 23823: {'lr': 0.0002855006729351046, 'samples': 4574208, 'steps': 23823, 'loss/train': 1.0646314322948456}
+01/29/2022 19:31:16 - INFO - codeparrot_training - Step 23824: {'lr': 0.0002854844762100861, 'samples': 4574400, 'steps': 23824, 'loss/train': 1.6914478540420532}
+01/29/2022 19:31:21 - INFO - codeparrot_training - Step 23825: {'lr': 0.0002854682793330633, 'samples': 4574592, 'steps': 23825, 'loss/train': 1.7167373299598694}
+01/29/2022 19:31:25 - INFO - codeparrot_training - Step 23826: {'lr': 0.0002854520823041057, 'samples': 4574784, 'steps': 23826, 'loss/train': 1.381077915430069}
+01/29/2022 19:31:29 - INFO - codeparrot_training - Step 23827: {'lr': 0.0002854358851232826, 'samples': 4574976, 'steps': 23827, 'loss/train': 1.4398386776447296}
+01/29/2022 19:31:34 - INFO - codeparrot_training - Step 23828: {'lr': 0.0002854196877906635, 'samples': 4575168, 'steps': 23828, 'loss/train': 1.9371782541275024}
+01/29/2022 19:31:38 - INFO - codeparrot_training - Step 23829: {'lr': 0.00028540349030631756, 'samples': 4575360, 'steps': 23829, 'loss/train': 1.711251139640808}
+01/29/2022 19:31:45 - INFO - codeparrot_training - Step 23830: {'lr': 0.0002853872926703144, 'samples': 4575552, 'steps': 23830, 'loss/train': 0.7924161851406097}
+01/29/2022 19:31:49 - INFO - codeparrot_training - Step 23831: {'lr': 0.0002853710948827233, 'samples': 4575744, 'steps': 23831, 'loss/train': 1.3826949298381805}
+01/29/2022 19:31:54 - INFO - codeparrot_training - Step 23832: {'lr': 0.00028535489694361365, 'samples': 4575936, 'steps': 23832, 'loss/train': 1.1010820269584656}
+01/29/2022 19:31:58 - INFO - codeparrot_training - Step 23833: {'lr': 0.00028533869885305485, 'samples': 4576128, 'steps': 23833, 'loss/train': 1.9540046453475952}
+01/29/2022 19:32:02 - INFO - codeparrot_training - Step 23834: {'lr': 0.0002853225006111163, 'samples': 4576320, 'steps': 23834, 'loss/train': 0.8004572689533234}
+01/29/2022 19:32:06 - INFO - codeparrot_training - Step 23835: {'lr': 0.00028530630221786736, 'samples': 4576512, 'steps': 23835, 'loss/train': 6.822926759719849}
+01/29/2022 19:32:12 - INFO - codeparrot_training - Step 23836: {'lr': 0.00028529010367337745, 'samples': 4576704, 'steps': 23836, 'loss/train': 2.3045502305030823}
+01/29/2022 19:32:16 - INFO - codeparrot_training - Step 23837: {'lr': 0.0002852739049777159, 'samples': 4576896, 'steps': 23837, 'loss/train': 1.4196816086769104}
+01/29/2022 19:32:20 - INFO - codeparrot_training - Step 23838: {'lr': 0.00028525770613095213, 'samples': 4577088, 'steps': 23838, 'loss/train': 1.6034406423568726}
+01/29/2022 19:32:24 - INFO - codeparrot_training - Step 23839: {'lr': 0.00028524150713315566, 'samples': 4577280, 'steps': 23839, 'loss/train': 1.4373348355293274}
+01/29/2022 19:32:28 - INFO - codeparrot_training - Step 23840: {'lr': 0.00028522530798439564, 'samples': 4577472, 'steps': 23840, 'loss/train': 1.5730960965156555}
+01/29/2022 19:32:36 - INFO - codeparrot_training - Step 23841: {'lr': 0.0002852091086847417, 'samples': 4577664, 'steps': 23841, 'loss/train': 2.0838144421577454}
+01/29/2022 19:32:40 - INFO - codeparrot_training - Step 23842: {'lr': 0.000285192909234263, 'samples': 4577856, 'steps': 23842, 'loss/train': 1.7983176112174988}
+01/29/2022 19:32:45 - INFO - codeparrot_training - Step 23843: {'lr': 0.0002851767096330291, 'samples': 4578048, 'steps': 23843, 'loss/train': 1.8224149346351624}
+01/29/2022 19:32:49 - INFO - codeparrot_training - Step 23844: {'lr': 0.00028516050988110935, 'samples': 4578240, 'steps': 23844, 'loss/train': 1.6363706588745117}
+01/29/2022 19:32:53 - INFO - codeparrot_training - Step 23845: {'lr': 0.00028514430997857317, 'samples': 4578432, 'steps': 23845, 'loss/train': 1.4135805666446686}
+01/29/2022 19:32:57 - INFO - codeparrot_training - Step 23846: {'lr': 0.0002851281099254899, 'samples': 4578624, 'steps': 23846, 'loss/train': 2.20774644613266}
+01/29/2022 19:33:02 - INFO - codeparrot_training - Step 23847: {'lr': 0.000285111909721929, 'samples': 4578816, 'steps': 23847, 'loss/train': 1.1829372346401215}
+01/29/2022 19:33:07 - INFO - codeparrot_training - Step 23848: {'lr': 0.0002850957093679597, 'samples': 4579008, 'steps': 23848, 'loss/train': 1.3758874833583832}
+01/29/2022 19:33:11 - INFO - codeparrot_training - Step 23849: {'lr': 0.0002850795088636516, 'samples': 4579200, 'steps': 23849, 'loss/train': 1.052115261554718}
+01/29/2022 19:33:15 - INFO - codeparrot_training - Step 23850: {'lr': 0.0002850633082090741, 'samples': 4579392, 'steps': 23850, 'loss/train': 1.8243750929832458}
+01/29/2022 19:33:19 - INFO - codeparrot_training - Step 23851: {'lr': 0.00028504710740429647, 'samples': 4579584, 'steps': 23851, 'loss/train': 1.0301754176616669}
+01/29/2022 19:33:25 - INFO - codeparrot_training - Step 23852: {'lr': 0.0002850309064493882, 'samples': 4579776, 'steps': 23852, 'loss/train': 0.9863284528255463}
+01/29/2022 19:33:29 - INFO - codeparrot_training - Step 23853: {'lr': 0.00028501470534441855, 'samples': 4579968, 'steps': 23853, 'loss/train': 0.96084925532341}
+01/29/2022 19:33:33 - INFO - codeparrot_training - Step 23854: {'lr': 0.00028499850408945704, 'samples': 4580160, 'steps': 23854, 'loss/train': 1.420459359884262}
+01/29/2022 19:33:37 - INFO - codeparrot_training - Step 23855: {'lr': 0.0002849823026845731, 'samples': 4580352, 'steps': 23855, 'loss/train': 2.1820918321609497}
+01/29/2022 19:33:42 - INFO - codeparrot_training - Step 23856: {'lr': 0.00028496610112983605, 'samples': 4580544, 'steps': 23856, 'loss/train': 1.1218160390853882}
+01/29/2022 19:33:47 - INFO - codeparrot_training - Step 23857: {'lr': 0.00028494989942531537, 'samples': 4580736, 'steps': 23857, 'loss/train': 1.9560842514038086}
+01/29/2022 19:33:51 - INFO - codeparrot_training - Step 23858: {'lr': 0.0002849336975710804, 'samples': 4580928, 'steps': 23858, 'loss/train': 1.6119248270988464}
+01/29/2022 19:33:55 - INFO - codeparrot_training - Step 23859: {'lr': 0.0002849174955672005, 'samples': 4581120, 'steps': 23859, 'loss/train': 1.2113977074623108}
+01/29/2022 19:34:00 - INFO - codeparrot_training - Step 23860: {'lr': 0.0002849012934137452, 'samples': 4581312, 'steps': 23860, 'loss/train': 1.7344120144844055}
+01/29/2022 19:34:04 - INFO - codeparrot_training - Step 23861: {'lr': 0.0002848850911107838, 'samples': 4581504, 'steps': 23861, 'loss/train': 1.6292213201522827}
+01/29/2022 19:34:11 - INFO - codeparrot_training - Step 23862: {'lr': 0.0002848688886583858, 'samples': 4581696, 'steps': 23862, 'loss/train': 1.101608008146286}
+01/29/2022 19:34:16 - INFO - codeparrot_training - Step 23863: {'lr': 0.0002848526860566205, 'samples': 4581888, 'steps': 23863, 'loss/train': 0.5139185786247253}
+01/29/2022 19:34:20 - INFO - codeparrot_training - Step 23864: {'lr': 0.00028483648330555737, 'samples': 4582080, 'steps': 23864, 'loss/train': 1.9260830283164978}
+01/29/2022 19:34:24 - INFO - codeparrot_training - Step 23865: {'lr': 0.0002848202804052659, 'samples': 4582272, 'steps': 23865, 'loss/train': 1.8986132740974426}
+01/29/2022 19:34:28 - INFO - codeparrot_training - Step 23866: {'lr': 0.00028480407735581527, 'samples': 4582464, 'steps': 23866, 'loss/train': 1.4476490914821625}
+01/29/2022 19:34:34 - INFO - codeparrot_training - Step 23867: {'lr': 0.00028478787415727515, 'samples': 4582656, 'steps': 23867, 'loss/train': 1.9308829307556152}
+01/29/2022 19:34:38 - INFO - codeparrot_training - Step 23868: {'lr': 0.00028477167080971465, 'samples': 4582848, 'steps': 23868, 'loss/train': 2.2217541933059692}
+01/29/2022 19:34:42 - INFO - codeparrot_training - Step 23869: {'lr': 0.0002847554673132035, 'samples': 4583040, 'steps': 23869, 'loss/train': 2.3728623390197754}
+01/29/2022 19:34:47 - INFO - codeparrot_training - Step 23870: {'lr': 0.00028473926366781095, 'samples': 4583232, 'steps': 23870, 'loss/train': 1.3863560557365417}
+01/29/2022 19:34:51 - INFO - codeparrot_training - Step 23871: {'lr': 0.00028472305987360636, 'samples': 4583424, 'steps': 23871, 'loss/train': 2.0632824897766113}
+01/29/2022 19:34:55 - INFO - codeparrot_training - Step 23872: {'lr': 0.00028470685593065927, 'samples': 4583616, 'steps': 23872, 'loss/train': 1.2877518832683563}
+01/29/2022 19:35:00 - INFO - codeparrot_training - Step 23873: {'lr': 0.00028469065183903895, 'samples': 4583808, 'steps': 23873, 'loss/train': 1.3399710059165955}
+01/29/2022 19:35:05 - INFO - codeparrot_training - Step 23874: {'lr': 0.000284674447598815, 'samples': 4584000, 'steps': 23874, 'loss/train': 0.10148592665791512}
+01/29/2022 19:35:09 - INFO - codeparrot_training - Step 23875: {'lr': 0.00028465824321005667, 'samples': 4584192, 'steps': 23875, 'loss/train': 1.0485747456550598}
+01/29/2022 19:35:13 - INFO - codeparrot_training - Step 23876: {'lr': 0.0002846420386728334, 'samples': 4584384, 'steps': 23876, 'loss/train': 1.7209099531173706}
+01/29/2022 19:35:17 - INFO - codeparrot_training - Step 23877: {'lr': 0.0002846258339872147, 'samples': 4584576, 'steps': 23877, 'loss/train': 1.7283049821853638}
+01/29/2022 19:35:24 - INFO - codeparrot_training - Step 23878: {'lr': 0.0002846096291532699, 'samples': 4584768, 'steps': 23878, 'loss/train': 1.7433993816375732}
+01/29/2022 19:35:29 - INFO - codeparrot_training - Step 23879: {'lr': 0.0002845934241710684, 'samples': 4584960, 'steps': 23879, 'loss/train': 2.1004786491394043}
+01/29/2022 19:35:33 - INFO - codeparrot_training - Step 23880: {'lr': 0.0002845772190406798, 'samples': 4585152, 'steps': 23880, 'loss/train': 6.667693376541138}
+01/29/2022 19:35:37 - INFO - codeparrot_training - Step 23881: {'lr': 0.0002845610137621732, 'samples': 4585344, 'steps': 23881, 'loss/train': 0.3637164607644081}
+01/29/2022 19:35:41 - INFO - codeparrot_training - Step 23882: {'lr': 0.0002845448083356183, 'samples': 4585536, 'steps': 23882, 'loss/train': 1.4367566406726837}
+01/29/2022 19:35:47 - INFO - codeparrot_training - Step 23883: {'lr': 0.00028452860276108436, 'samples': 4585728, 'steps': 23883, 'loss/train': 0.30817843973636627}
+01/29/2022 19:35:51 - INFO - codeparrot_training - Step 23884: {'lr': 0.0002845123970386408, 'samples': 4585920, 'steps': 23884, 'loss/train': 0.657759815454483}
+01/29/2022 19:35:55 - INFO - codeparrot_training - Step 23885: {'lr': 0.00028449619116835715, 'samples': 4586112, 'steps': 23885, 'loss/train': 2.3292782306671143}
+01/29/2022 19:35:59 - INFO - codeparrot_training - Step 23886: {'lr': 0.0002844799851503028, 'samples': 4586304, 'steps': 23886, 'loss/train': 1.1195893585681915}
+01/29/2022 19:36:04 - INFO - codeparrot_training - Step 23887: {'lr': 0.0002844637789845471, 'samples': 4586496, 'steps': 23887, 'loss/train': 2.125841438770294}
+01/29/2022 19:36:09 - INFO - codeparrot_training - Step 23888: {'lr': 0.0002844475726711595, 'samples': 4586688, 'steps': 23888, 'loss/train': 2.2391082644462585}
+01/29/2022 19:36:13 - INFO - codeparrot_training - Step 23889: {'lr': 0.0002844313662102095, 'samples': 4586880, 'steps': 23889, 'loss/train': 1.4712892770767212}
+01/29/2022 19:36:17 - INFO - codeparrot_training - Step 23890: {'lr': 0.0002844151596017665, 'samples': 4587072, 'steps': 23890, 'loss/train': 2.506456196308136}
+01/29/2022 19:36:21 - INFO - codeparrot_training - Step 23891: {'lr': 0.0002843989528458997, 'samples': 4587264, 'steps': 23891, 'loss/train': 1.4225609600543976}
+01/29/2022 19:36:26 - INFO - codeparrot_training - Step 23892: {'lr': 0.0002843827459426789, 'samples': 4587456, 'steps': 23892, 'loss/train': 0.7289436310529709}
+01/29/2022 19:36:33 - INFO - codeparrot_training - Step 23893: {'lr': 0.00028436653889217316, 'samples': 4587648, 'steps': 23893, 'loss/train': 1.9818010926246643}
+01/29/2022 19:36:37 - INFO - codeparrot_training - Step 23894: {'lr': 0.00028435033169445223, 'samples': 4587840, 'steps': 23894, 'loss/train': 1.9826531410217285}
+01/29/2022 19:36:41 - INFO - codeparrot_training - Step 23895: {'lr': 0.0002843341243495853, 'samples': 4588032, 'steps': 23895, 'loss/train': 1.2097082734107971}
+01/29/2022 19:36:46 - INFO - codeparrot_training - Step 23896: {'lr': 0.0002843179168576419, 'samples': 4588224, 'steps': 23896, 'loss/train': 1.5351024270057678}
+01/29/2022 19:36:50 - INFO - codeparrot_training - Step 23897: {'lr': 0.00028430170921869147, 'samples': 4588416, 'steps': 23897, 'loss/train': 1.1531972587108612}
+01/29/2022 19:36:55 - INFO - codeparrot_training - Step 23898: {'lr': 0.0002842855014328034, 'samples': 4588608, 'steps': 23898, 'loss/train': 1.8952388763427734}
+01/29/2022 19:37:00 - INFO - codeparrot_training - Step 23899: {'lr': 0.0002842692935000471, 'samples': 4588800, 'steps': 23899, 'loss/train': 1.4552857875823975}
+01/29/2022 19:37:04 - INFO - codeparrot_training - Step 23900: {'lr': 0.00028425308542049207, 'samples': 4588992, 'steps': 23900, 'loss/train': 0.5217408835887909}
+01/29/2022 19:37:08 - INFO - codeparrot_training - Step 23901: {'lr': 0.0002842368771942077, 'samples': 4589184, 'steps': 23901, 'loss/train': 2.226067364215851}
+01/29/2022 19:37:12 - INFO - codeparrot_training - Step 23902: {'lr': 0.00028422066882126336, 'samples': 4589376, 'steps': 23902, 'loss/train': 3.2934683561325073}
+01/29/2022 19:37:19 - INFO - codeparrot_training - Step 23903: {'lr': 0.0002842044603017285, 'samples': 4589568, 'steps': 23903, 'loss/train': 1.824639916419983}
+01/29/2022 19:37:24 - INFO - codeparrot_training - Step 23904: {'lr': 0.00028418825163567275, 'samples': 4589760, 'steps': 23904, 'loss/train': 1.7718345522880554}
+01/29/2022 19:37:28 - INFO - codeparrot_training - Step 23905: {'lr': 0.0002841720428231653, 'samples': 4589952, 'steps': 23905, 'loss/train': 1.5791105031967163}
+01/29/2022 19:37:32 - INFO - codeparrot_training - Step 23906: {'lr': 0.00028415583386427566, 'samples': 4590144, 'steps': 23906, 'loss/train': 1.253352016210556}
+01/29/2022 19:37:36 - INFO - codeparrot_training - Step 23907: {'lr': 0.0002841396247590733, 'samples': 4590336, 'steps': 23907, 'loss/train': 1.9852674007415771}
+01/29/2022 19:37:41 - INFO - codeparrot_training - Step 23908: {'lr': 0.00028412341550762755, 'samples': 4590528, 'steps': 23908, 'loss/train': 2.7583696246147156}
+01/29/2022 19:37:46 - INFO - codeparrot_training - Step 23909: {'lr': 0.00028410720611000804, 'samples': 4590720, 'steps': 23909, 'loss/train': 2.0297282338142395}
+01/29/2022 19:37:50 - INFO - codeparrot_training - Step 23910: {'lr': 0.000284090996566284, 'samples': 4590912, 'steps': 23910, 'loss/train': 1.7018296122550964}
+01/29/2022 19:37:54 - INFO - codeparrot_training - Step 23911: {'lr': 0.00028407478687652503, 'samples': 4591104, 'steps': 23911, 'loss/train': 1.6421799659729004}
+01/29/2022 19:37:58 - INFO - codeparrot_training - Step 23912: {'lr': 0.0002840585770408004, 'samples': 4591296, 'steps': 23912, 'loss/train': 4.3295817375183105}
+01/29/2022 19:38:03 - INFO - codeparrot_training - Step 23913: {'lr': 0.00028404236705917974, 'samples': 4591488, 'steps': 23913, 'loss/train': 1.6491249203681946}
+01/29/2022 19:38:08 - INFO - codeparrot_training - Step 23914: {'lr': 0.00028402615693173236, 'samples': 4591680, 'steps': 23914, 'loss/train': 1.2246925234794617}
+01/29/2022 19:38:12 - INFO - codeparrot_training - Step 23915: {'lr': 0.00028400994665852777, 'samples': 4591872, 'steps': 23915, 'loss/train': 1.3568209111690521}
+01/29/2022 19:38:16 - INFO - codeparrot_training - Step 23916: {'lr': 0.00028399373623963525, 'samples': 4592064, 'steps': 23916, 'loss/train': 1.4262498915195465}
+01/29/2022 19:38:20 - INFO - codeparrot_training - Step 23917: {'lr': 0.0002839775256751244, 'samples': 4592256, 'steps': 23917, 'loss/train': 2.5017465949058533}
+01/29/2022 19:38:28 - INFO - codeparrot_training - Step 23918: {'lr': 0.00028396131496506466, 'samples': 4592448, 'steps': 23918, 'loss/train': 7.339595317840576}
+01/29/2022 19:38:32 - INFO - codeparrot_training - Step 23919: {'lr': 0.00028394510410952544, 'samples': 4592640, 'steps': 23919, 'loss/train': 1.2172236442565918}
+01/29/2022 19:38:36 - INFO - codeparrot_training - Step 23920: {'lr': 0.0002839288931085761, 'samples': 4592832, 'steps': 23920, 'loss/train': 1.8071399331092834}
+01/29/2022 19:38:40 - INFO - codeparrot_training - Step 23921: {'lr': 0.0002839126819622862, 'samples': 4593024, 'steps': 23921, 'loss/train': 2.278142273426056}
+01/29/2022 19:38:44 - INFO - codeparrot_training - Step 23922: {'lr': 0.00028389647067072517, 'samples': 4593216, 'steps': 23922, 'loss/train': 2.8620150089263916}
+01/29/2022 19:38:50 - INFO - codeparrot_training - Step 23923: {'lr': 0.00028388025923396234, 'samples': 4593408, 'steps': 23923, 'loss/train': 1.8065795302391052}
+01/29/2022 19:38:54 - INFO - codeparrot_training - Step 23924: {'lr': 0.0002838640476520673, 'samples': 4593600, 'steps': 23924, 'loss/train': 1.7671501636505127}
+01/29/2022 19:38:58 - INFO - codeparrot_training - Step 23925: {'lr': 0.00028384783592510945, 'samples': 4593792, 'steps': 23925, 'loss/train': 7.2844648361206055}
+01/29/2022 19:39:03 - INFO - codeparrot_training - Step 23926: {'lr': 0.00028383162405315823, 'samples': 4593984, 'steps': 23926, 'loss/train': 1.9071187376976013}
+01/29/2022 19:39:07 - INFO - codeparrot_training - Step 23927: {'lr': 0.00028381541203628295, 'samples': 4594176, 'steps': 23927, 'loss/train': 1.6116403341293335}
+01/29/2022 19:39:12 - INFO - codeparrot_training - Step 23928: {'lr': 0.0002837991998745533, 'samples': 4594368, 'steps': 23928, 'loss/train': 1.4861798286437988}
+01/29/2022 19:39:16 - INFO - codeparrot_training - Step 23929: {'lr': 0.0002837829875680386, 'samples': 4594560, 'steps': 23929, 'loss/train': 1.7700303196907043}
+01/29/2022 19:39:21 - INFO - codeparrot_training - Step 23930: {'lr': 0.00028376677511680827, 'samples': 4594752, 'steps': 23930, 'loss/train': 1.864320158958435}
+01/29/2022 19:39:25 - INFO - codeparrot_training - Step 23931: {'lr': 0.0002837505625209318, 'samples': 4594944, 'steps': 23931, 'loss/train': 2.405535399913788}
+01/29/2022 19:39:29 - INFO - codeparrot_training - Step 23932: {'lr': 0.0002837343497804787, 'samples': 4595136, 'steps': 23932, 'loss/train': 1.9334238767623901}
+01/29/2022 19:39:34 - INFO - codeparrot_training - Step 23933: {'lr': 0.0002837181368955183, 'samples': 4595328, 'steps': 23933, 'loss/train': 1.53024423122406}
+01/29/2022 19:39:38 - INFO - codeparrot_training - Step 23934: {'lr': 0.0002837019238661201, 'samples': 4595520, 'steps': 23934, 'loss/train': 1.4726594388484955}
+01/29/2022 19:39:43 - INFO - codeparrot_training - Step 23935: {'lr': 0.00028368571069235354, 'samples': 4595712, 'steps': 23935, 'loss/train': 1.70952308177948}
+01/29/2022 19:39:47 - INFO - codeparrot_training - Step 23936: {'lr': 0.00028366949737428814, 'samples': 4595904, 'steps': 23936, 'loss/train': 2.256426215171814}
+01/29/2022 19:39:51 - INFO - codeparrot_training - Step 23937: {'lr': 0.00028365328391199334, 'samples': 4596096, 'steps': 23937, 'loss/train': 1.5048179626464844}
+01/29/2022 19:39:58 - INFO - codeparrot_training - Step 23938: {'lr': 0.0002836370703055385, 'samples': 4596288, 'steps': 23938, 'loss/train': 0.08158080093562603}
+01/29/2022 19:40:02 - INFO - codeparrot_training - Step 23939: {'lr': 0.0002836208565549932, 'samples': 4596480, 'steps': 23939, 'loss/train': 1.6059486865997314}
+01/29/2022 19:40:07 - INFO - codeparrot_training - Step 23940: {'lr': 0.00028360464266042674, 'samples': 4596672, 'steps': 23940, 'loss/train': 1.704684555530548}
+01/29/2022 19:40:11 - INFO - codeparrot_training - Step 23941: {'lr': 0.00028358842862190873, 'samples': 4596864, 'steps': 23941, 'loss/train': 1.7739797830581665}
+01/29/2022 19:40:15 - INFO - codeparrot_training - Step 23942: {'lr': 0.00028357221443950847, 'samples': 4597056, 'steps': 23942, 'loss/train': 1.3783408105373383}
+01/29/2022 19:40:19 - INFO - codeparrot_training - Step 23943: {'lr': 0.00028355600011329557, 'samples': 4597248, 'steps': 23943, 'loss/train': 1.5278881788253784}
+01/29/2022 19:40:24 - INFO - codeparrot_training - Step 23944: {'lr': 0.00028353978564333936, 'samples': 4597440, 'steps': 23944, 'loss/train': 1.6490326523780823}
+01/29/2022 19:40:29 - INFO - codeparrot_training - Step 23945: {'lr': 0.0002835235710297094, 'samples': 4597632, 'steps': 23945, 'loss/train': 3.061728358268738}
+01/29/2022 19:40:33 - INFO - codeparrot_training - Step 23946: {'lr': 0.0002835073562724751, 'samples': 4597824, 'steps': 23946, 'loss/train': 1.5444656610488892}
+01/29/2022 19:40:37 - INFO - codeparrot_training - Step 23947: {'lr': 0.00028349114137170593, 'samples': 4598016, 'steps': 23947, 'loss/train': 0.9231298863887787}
+01/29/2022 19:40:41 - INFO - codeparrot_training - Step 23948: {'lr': 0.0002834749263274714, 'samples': 4598208, 'steps': 23948, 'loss/train': 1.6378228068351746}
+01/29/2022 19:40:48 - INFO - codeparrot_training - Step 23949: {'lr': 0.00028345871113984086, 'samples': 4598400, 'steps': 23949, 'loss/train': 1.4845108687877655}
+01/29/2022 19:40:52 - INFO - codeparrot_training - Step 23950: {'lr': 0.0002834424958088838, 'samples': 4598592, 'steps': 23950, 'loss/train': 1.4202479124069214}
+01/29/2022 19:40:57 - INFO - codeparrot_training - Step 23951: {'lr': 0.00028342628033466974, 'samples': 4598784, 'steps': 23951, 'loss/train': 1.2285007238388062}
+01/29/2022 19:41:01 - INFO - codeparrot_training - Step 23952: {'lr': 0.00028341006471726816, 'samples': 4598976, 'steps': 23952, 'loss/train': 2.1908154487609863}
+01/29/2022 19:41:05 - INFO - codeparrot_training - Step 23953: {'lr': 0.0002833938489567484, 'samples': 4599168, 'steps': 23953, 'loss/train': 1.482117623090744}
+01/29/2022 19:41:11 - INFO - codeparrot_training - Step 23954: {'lr': 0.00028337763305318, 'samples': 4599360, 'steps': 23954, 'loss/train': 0.8394040167331696}
+01/29/2022 19:41:15 - INFO - codeparrot_training - Step 23955: {'lr': 0.00028336141700663244, 'samples': 4599552, 'steps': 23955, 'loss/train': 1.9457117915153503}
+01/29/2022 19:41:19 - INFO - codeparrot_training - Step 23956: {'lr': 0.00028334520081717507, 'samples': 4599744, 'steps': 23956, 'loss/train': 1.0480142533779144}
+01/29/2022 19:41:24 - INFO - codeparrot_training - Step 23957: {'lr': 0.0002833289844848776, 'samples': 4599936, 'steps': 23957, 'loss/train': 0.804270476102829}
+01/29/2022 19:41:28 - INFO - codeparrot_training - Step 23958: {'lr': 0.0002833127680098092, 'samples': 4600128, 'steps': 23958, 'loss/train': 1.8483906984329224}
+01/29/2022 19:41:32 - INFO - codeparrot_training - Step 23959: {'lr': 0.0002832965513920396, 'samples': 4600320, 'steps': 23959, 'loss/train': 1.3476849496364594}
+01/29/2022 19:41:37 - INFO - codeparrot_training - Step 23960: {'lr': 0.0002832803346316381, 'samples': 4600512, 'steps': 23960, 'loss/train': 1.3335632979869843}
+01/29/2022 19:41:41 - INFO - codeparrot_training - Step 23961: {'lr': 0.0002832641177286742, 'samples': 4600704, 'steps': 23961, 'loss/train': 1.489419847726822}
+01/29/2022 19:41:46 - INFO - codeparrot_training - Step 23962: {'lr': 0.0002832479006832174, 'samples': 4600896, 'steps': 23962, 'loss/train': 2.156388223171234}
+01/29/2022 19:41:50 - INFO - codeparrot_training - Step 23963: {'lr': 0.0002832316834953372, 'samples': 4601088, 'steps': 23963, 'loss/train': 1.30601567029953}
+01/29/2022 19:41:54 - INFO - codeparrot_training - Step 23964: {'lr': 0.0002832154661651029, 'samples': 4601280, 'steps': 23964, 'loss/train': 1.8446361422538757}
+01/29/2022 19:42:01 - INFO - codeparrot_training - Step 23965: {'lr': 0.00028319924869258425, 'samples': 4601472, 'steps': 23965, 'loss/train': 1.301607370376587}
+01/29/2022 19:42:05 - INFO - codeparrot_training - Step 23966: {'lr': 0.0002831830310778504, 'samples': 4601664, 'steps': 23966, 'loss/train': 1.3221142888069153}
+01/29/2022 19:42:10 - INFO - codeparrot_training - Step 23967: {'lr': 0.0002831668133209711, 'samples': 4601856, 'steps': 23967, 'loss/train': 1.6231216192245483}
+01/29/2022 19:42:14 - INFO - codeparrot_training - Step 23968: {'lr': 0.0002831505954220156, 'samples': 4602048, 'steps': 23968, 'loss/train': 1.7397499680519104}
+01/29/2022 19:42:18 - INFO - codeparrot_training - Step 23969: {'lr': 0.00028313437738105353, 'samples': 4602240, 'steps': 23969, 'loss/train': 1.442326694726944}
+01/29/2022 19:42:23 - INFO - codeparrot_training - Step 23970: {'lr': 0.0002831181591981543, 'samples': 4602432, 'steps': 23970, 'loss/train': 2.5649591088294983}
+01/29/2022 19:42:27 - INFO - codeparrot_training - Step 23971: {'lr': 0.0002831019408733874, 'samples': 4602624, 'steps': 23971, 'loss/train': 1.7808603644371033}
+01/29/2022 19:42:32 - INFO - codeparrot_training - Step 23972: {'lr': 0.00028308572240682233, 'samples': 4602816, 'steps': 23972, 'loss/train': 2.0594301223754883}
+01/29/2022 19:42:36 - INFO - codeparrot_training - Step 23973: {'lr': 0.00028306950379852844, 'samples': 4603008, 'steps': 23973, 'loss/train': 1.7291666865348816}
+01/29/2022 19:42:40 - INFO - codeparrot_training - Step 23974: {'lr': 0.0002830532850485754, 'samples': 4603200, 'steps': 23974, 'loss/train': 1.3687503933906555}
+01/29/2022 19:42:46 - INFO - codeparrot_training - Step 23975: {'lr': 0.0002830370661570325, 'samples': 4603392, 'steps': 23975, 'loss/train': 1.598259687423706}
+01/29/2022 19:42:50 - INFO - codeparrot_training - Step 23976: {'lr': 0.00028302084712396937, 'samples': 4603584, 'steps': 23976, 'loss/train': 1.7897295355796814}
+01/29/2022 19:42:54 - INFO - codeparrot_training - Step 23977: {'lr': 0.00028300462794945535, 'samples': 4603776, 'steps': 23977, 'loss/train': 1.8059313297271729}
+01/29/2022 19:42:58 - INFO - codeparrot_training - Step 23978: {'lr': 0.00028298840863356006, 'samples': 4603968, 'steps': 23978, 'loss/train': 1.3636030554771423}
+01/29/2022 19:43:03 - INFO - codeparrot_training - Step 23979: {'lr': 0.0002829721891763529, 'samples': 4604160, 'steps': 23979, 'loss/train': 1.709848701953888}
+01/29/2022 19:43:10 - INFO - codeparrot_training - Step 23980: {'lr': 0.00028295596957790325, 'samples': 4604352, 'steps': 23980, 'loss/train': 0.4057074040174484}
+01/29/2022 19:43:14 - INFO - codeparrot_training - Step 23981: {'lr': 0.0002829397498382808, 'samples': 4604544, 'steps': 23981, 'loss/train': 1.948394000530243}
+01/29/2022 19:43:18 - INFO - codeparrot_training - Step 23982: {'lr': 0.00028292352995755487, 'samples': 4604736, 'steps': 23982, 'loss/train': 1.4339393377304077}
+01/29/2022 19:43:22 - INFO - codeparrot_training - Step 23983: {'lr': 0.000282907309935795, 'samples': 4604928, 'steps': 23983, 'loss/train': 2.09488445520401}
+01/29/2022 19:43:27 - INFO - codeparrot_training - Step 23984: {'lr': 0.00028289108977307066, 'samples': 4605120, 'steps': 23984, 'loss/train': 1.6891711950302124}
+01/29/2022 19:43:32 - INFO - codeparrot_training - Step 23985: {'lr': 0.00028287486946945137, 'samples': 4605312, 'steps': 23985, 'loss/train': 1.6502711176872253}
+01/29/2022 19:43:36 - INFO - codeparrot_training - Step 23986: {'lr': 0.0002828586490250065, 'samples': 4605504, 'steps': 23986, 'loss/train': 1.5294495820999146}
+01/29/2022 19:43:40 - INFO - codeparrot_training - Step 23987: {'lr': 0.00028284242843980566, 'samples': 4605696, 'steps': 23987, 'loss/train': 1.2680242359638214}
+01/29/2022 19:43:44 - INFO - codeparrot_training - Step 23988: {'lr': 0.00028282620771391824, 'samples': 4605888, 'steps': 23988, 'loss/train': 2.1197129487991333}
+01/29/2022 19:43:49 - INFO - codeparrot_training - Step 23989: {'lr': 0.00028280998684741387, 'samples': 4606080, 'steps': 23989, 'loss/train': 1.4383395910263062}
+01/29/2022 19:43:54 - INFO - codeparrot_training - Step 23990: {'lr': 0.00028279376584036187, 'samples': 4606272, 'steps': 23990, 'loss/train': 1.322454571723938}
+01/29/2022 19:43:58 - INFO - codeparrot_training - Step 23991: {'lr': 0.0002827775446928318, 'samples': 4606464, 'steps': 23991, 'loss/train': 0.16137124970555305}
+01/29/2022 19:44:02 - INFO - codeparrot_training - Step 23992: {'lr': 0.00028276132340489306, 'samples': 4606656, 'steps': 23992, 'loss/train': 1.6955174803733826}
+01/29/2022 19:44:06 - INFO - codeparrot_training - Step 23993: {'lr': 0.0002827451019766153, 'samples': 4606848, 'steps': 23993, 'loss/train': 3.67300021648407}
+01/29/2022 19:44:11 - INFO - codeparrot_training - Step 23994: {'lr': 0.00028272888040806795, 'samples': 4607040, 'steps': 23994, 'loss/train': 1.2293110191822052}
+01/29/2022 19:44:18 - INFO - codeparrot_training - Step 23995: {'lr': 0.0002827126586993204, 'samples': 4607232, 'steps': 23995, 'loss/train': 2.323569595813751}
+01/29/2022 19:44:22 - INFO - codeparrot_training - Step 23996: {'lr': 0.0002826964368504422, 'samples': 4607424, 'steps': 23996, 'loss/train': 1.1858670115470886}
+01/29/2022 19:44:26 - INFO - codeparrot_training - Step 23997: {'lr': 0.0002826802148615029, 'samples': 4607616, 'steps': 23997, 'loss/train': 2.8211867809295654}
+01/29/2022 19:44:31 - INFO - codeparrot_training - Step 23998: {'lr': 0.00028266399273257193, 'samples': 4607808, 'steps': 23998, 'loss/train': 2.0148983001708984}
+01/29/2022 19:44:35 - INFO - codeparrot_training - Step 23999: {'lr': 0.0002826477704637188, 'samples': 4608000, 'steps': 23999, 'loss/train': 2.0956292152404785}
+01/29/2022 19:44:35 - INFO - codeparrot_training - Evaluating and saving model checkpoint