diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -34506,3 +34506,1009 @@ Use FP16 precision: False
 02/25/2022 21:35:37 - INFO - codeparrot_training - Step 33998: {'lr': 0.00012502834168590582, 'samples': 17407488, 'steps': 33998, 'loss/train': 2.855187177658081}
 02/25/2022 21:35:43 - INFO - codeparrot_training - Step 33999: {'lr': 0.00012501417057525317, 'samples': 17408000, 'steps': 33999, 'loss/train': 1.9884394407272339}
 02/25/2022 21:35:43 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/25/2022 21:36:00 - WARNING - huggingface_hub.repository - Several commits (34) will be pushed upstream.
+02/25/2022 21:36:00 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/25/2022 21:36:35 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   6b82659..6cccefd  floral-grass-11 -> floral-grass-11
+
+02/25/2022 21:36:39 - INFO - codeparrot_training - Step 34000: {'lr': 0.00012500000000000006, 'samples': 17408512, 'steps': 34000, 'loss/train': 1.4243096113204956}
+02/25/2022 21:36:47 - INFO - codeparrot_training - Step 34001: {'lr': 0.00012498582996020715, 'samples': 17409024, 'steps': 34001, 'loss/train': 0.8388751745223999}
+02/25/2022 21:36:50 - INFO - codeparrot_training - Step 34002: {'lr': 0.0001249716604559354, 'samples': 17409536, 'steps': 34002, 'loss/train': 1.6036185026168823}
+02/25/2022 21:36:56 - INFO - codeparrot_training - Step 34003: {'lr': 0.00012495749148724533, 'samples': 17410048, 'steps': 34003, 'loss/train': 1.8250926733016968}
+02/25/2022 21:36:59 - INFO - codeparrot_training - Step 34004: {'lr': 0.00012494332305419765, 'samples': 17410560, 'steps': 34004, 'loss/train': 2.837171792984009}
+02/25/2022 21:37:05 - INFO - codeparrot_training - Step 34005: {'lr': 0.00012492915515685307, 'samples': 17411072, 'steps': 34005, 'loss/train': 1.3692988157272339}
+02/25/2022 21:37:08 - INFO - codeparrot_training - Step 34006: {'lr': 0.00012491498779527218, 'samples': 17411584, 'steps': 34006, 'loss/train': 1.5175013542175293}
+02/25/2022 21:37:14 - INFO - codeparrot_training - Step 34007: {'lr': 0.0001249008209695159, 'samples': 17412096, 'steps': 34007, 'loss/train': 0.8875550627708435}
+02/25/2022 21:37:17 - INFO - codeparrot_training - Step 34008: {'lr': 0.00012488665467964472, 'samples': 17412608, 'steps': 34008, 'loss/train': 0.9341545701026917}
+02/25/2022 21:37:23 - INFO - codeparrot_training - Step 34009: {'lr': 0.0001248724889257194, 'samples': 17413120, 'steps': 34009, 'loss/train': 2.840592384338379}
+02/25/2022 21:37:26 - INFO - codeparrot_training - Step 34010: {'lr': 0.00012485832370780052, 'samples': 17413632, 'steps': 34010, 'loss/train': 1.1056419610977173}
+02/25/2022 21:37:33 - INFO - codeparrot_training - Step 34011: {'lr': 0.00012484415902594892, 'samples': 17414144, 'steps': 34011, 'loss/train': 1.3155574798583984}
+02/25/2022 21:37:36 - INFO - codeparrot_training - Step 34012: {'lr': 0.00012482999488022515, 'samples': 17414656, 'steps': 34012, 'loss/train': 1.75338613986969}
+02/25/2022 21:37:42 - INFO - codeparrot_training - Step 34013: {'lr': 0.00012481583127069008, 'samples': 17415168, 'steps': 34013, 'loss/train': 0.644737184047699}
+02/25/2022 21:37:45 - INFO - codeparrot_training - Step 34014: {'lr': 0.00012480166819740408, 'samples': 17415680, 'steps': 34014, 'loss/train': 1.4448992013931274}
+02/25/2022 21:37:51 - INFO - codeparrot_training - Step 34015: {'lr': 0.00012478750566042804, 'samples': 17416192, 'steps': 34015, 'loss/train': 1.5743814706802368}
+02/25/2022 21:37:54 - INFO - codeparrot_training - Step 34016: {'lr': 0.00012477334365982248, 'samples': 17416704, 'steps': 34016, 'loss/train': 0.9618198871612549}
+02/25/2022 21:38:00 - INFO - codeparrot_training - Step 34017: {'lr': 0.00012475918219564836, 'samples': 17417216, 'steps': 34017, 'loss/train': 2.1817731857299805}
+02/25/2022 21:38:03 - INFO - codeparrot_training - Step 34018: {'lr': 0.00012474502126796593, 'samples': 17417728, 'steps': 34018, 'loss/train': 1.1738075017929077}
+02/25/2022 21:38:09 - INFO - codeparrot_training - Step 34019: {'lr': 0.00012473086087683616, 'samples': 17418240, 'steps': 34019, 'loss/train': 2.341294765472412}
+02/25/2022 21:38:12 - INFO - codeparrot_training - Step 34020: {'lr': 0.00012471670102231952, 'samples': 17418752, 'steps': 34020, 'loss/train': 2.4172229766845703}
+02/25/2022 21:38:18 - INFO - codeparrot_training - Step 34021: {'lr': 0.00012470254170447685, 'samples': 17419264, 'steps': 34021, 'loss/train': 1.9748722314834595}
+02/25/2022 21:38:21 - INFO - codeparrot_training - Step 34022: {'lr': 0.00012468838292336868, 'samples': 17419776, 'steps': 34022, 'loss/train': 2.4044883251190186}
+02/25/2022 21:38:28 - INFO - codeparrot_training - Step 34023: {'lr': 0.00012467422467905568, 'samples': 17420288, 'steps': 34023, 'loss/train': 2.3403255939483643}
+02/25/2022 21:38:31 - INFO - codeparrot_training - Step 34024: {'lr': 0.00012466006697159842, 'samples': 17420800, 'steps': 34024, 'loss/train': 1.6199828386306763}
+02/25/2022 21:38:37 - INFO - codeparrot_training - Step 34025: {'lr': 0.00012464590980105776, 'samples': 17421312, 'steps': 34025, 'loss/train': 2.4142913818359375}
+02/25/2022 21:38:40 - INFO - codeparrot_training - Step 34026: {'lr': 0.00012463175316749419, 'samples': 17421824, 'steps': 34026, 'loss/train': 2.1441690921783447}
+02/25/2022 21:38:46 - INFO - codeparrot_training - Step 34027: {'lr': 0.00012461759707096837, 'samples': 17422336, 'steps': 34027, 'loss/train': 2.3603148460388184}
+02/25/2022 21:38:49 - INFO - codeparrot_training - Step 34028: {'lr': 0.00012460344151154088, 'samples': 17422848, 'steps': 34028, 'loss/train': 1.9076552391052246}
+02/25/2022 21:38:55 - INFO - codeparrot_training - Step 34029: {'lr': 0.0001245892864892725, 'samples': 17423360, 'steps': 34029, 'loss/train': 0.810051441192627}
+02/25/2022 21:38:58 - INFO - codeparrot_training - Step 34030: {'lr': 0.0001245751320042238, 'samples': 17423872, 'steps': 34030, 'loss/train': 2.6465001106262207}
+02/25/2022 21:39:04 - INFO - codeparrot_training - Step 34031: {'lr': 0.00012456097805645542, 'samples': 17424384, 'steps': 34031, 'loss/train': 1.9230165481567383}
+02/25/2022 21:39:07 - INFO - codeparrot_training - Step 34032: {'lr': 0.00012454682464602788, 'samples': 17424896, 'steps': 34032, 'loss/train': 1.5525074005126953}
+02/25/2022 21:39:13 - INFO - codeparrot_training - Step 34033: {'lr': 0.000124532671773002, 'samples': 17425408, 'steps': 34033, 'loss/train': 1.8166000843048096}
+02/25/2022 21:39:16 - INFO - codeparrot_training - Step 34034: {'lr': 0.00012451851943743821, 'samples': 17425920, 'steps': 34034, 'loss/train': 1.4815582036972046}
+02/25/2022 21:39:22 - INFO - codeparrot_training - Step 34035: {'lr': 0.00012450436763939747, 'samples': 17426432, 'steps': 34035, 'loss/train': 1.8599796295166016}
+02/25/2022 21:39:25 - INFO - codeparrot_training - Step 34036: {'lr': 0.00012449021637893993, 'samples': 17426944, 'steps': 34036, 'loss/train': 1.8132766485214233}
+02/25/2022 21:39:32 - INFO - codeparrot_training - Step 34037: {'lr': 0.00012447606565612655, 'samples': 17427456, 'steps': 34037, 'loss/train': 2.1141202449798584}
+02/25/2022 21:39:35 - INFO - codeparrot_training - Step 34038: {'lr': 0.00012446191547101787, 'samples': 17427968, 'steps': 34038, 'loss/train': 0.9231016635894775}
+02/25/2022 21:39:41 - INFO - codeparrot_training - Step 34039: {'lr': 0.00012444776582367446, 'samples': 17428480, 'steps': 34039, 'loss/train': 0.9292402863502502}
+02/25/2022 21:39:44 - INFO - codeparrot_training - Step 34040: {'lr': 0.00012443361671415687, 'samples': 17428992, 'steps': 34040, 'loss/train': 1.6624534130096436}
+02/25/2022 21:39:50 - INFO - codeparrot_training - Step 34041: {'lr': 0.0001244194681425259, 'samples': 17429504, 'steps': 34041, 'loss/train': 2.331366539001465}
+02/25/2022 21:39:53 - INFO - codeparrot_training - Step 34042: {'lr': 0.000124405320108842, 'samples': 17430016, 'steps': 34042, 'loss/train': 2.037729024887085}
+02/25/2022 21:39:59 - INFO - codeparrot_training - Step 34043: {'lr': 0.00012439117261316588, 'samples': 17430528, 'steps': 34043, 'loss/train': 1.4858005046844482}
+02/25/2022 21:40:02 - INFO - codeparrot_training - Step 34044: {'lr': 0.00012437702565555798, 'samples': 17431040, 'steps': 34044, 'loss/train': 1.946694254875183}
+02/25/2022 21:40:08 - INFO - codeparrot_training - Step 34045: {'lr': 0.0001243628792360791, 'samples': 17431552, 'steps': 34045, 'loss/train': 1.1625895500183105}
+02/25/2022 21:40:11 - INFO - codeparrot_training - Step 34046: {'lr': 0.00012434873335478977, 'samples': 17432064, 'steps': 34046, 'loss/train': 0.3992147743701935}
+02/25/2022 21:40:18 - INFO - codeparrot_training - Step 34047: {'lr': 0.00012433458801175046, 'samples': 17432576, 'steps': 34047, 'loss/train': 1.2809768915176392}
+02/25/2022 21:40:21 - INFO - codeparrot_training - Step 34048: {'lr': 0.00012432044320702197, 'samples': 17433088, 'steps': 34048, 'loss/train': 1.73301362991333}
+02/25/2022 21:40:27 - INFO - codeparrot_training - Step 34049: {'lr': 0.0001243062989406648, 'samples': 17433600, 'steps': 34049, 'loss/train': 2.468609094619751}
+02/25/2022 21:40:30 - INFO - codeparrot_training - Step 34050: {'lr': 0.00012429215521273955, 'samples': 17434112, 'steps': 34050, 'loss/train': 1.0004311800003052}
+02/25/2022 21:40:36 - INFO - codeparrot_training - Step 34051: {'lr': 0.00012427801202330666, 'samples': 17434624, 'steps': 34051, 'loss/train': 1.8955143690109253}
+02/25/2022 21:40:39 - INFO - codeparrot_training - Step 34052: {'lr': 0.00012426386937242705, 'samples': 17435136, 'steps': 34052, 'loss/train': 2.399646282196045}
+02/25/2022 21:40:45 - INFO - codeparrot_training - Step 34053: {'lr': 0.0001242497272601609, 'samples': 17435648, 'steps': 34053, 'loss/train': 1.944251537322998}
+02/25/2022 21:40:48 - INFO - codeparrot_training - Step 34054: {'lr': 0.0001242355856865691, 'samples': 17436160, 'steps': 34054, 'loss/train': 0.9422216415405273}
+02/25/2022 21:40:54 - INFO - codeparrot_training - Step 34055: {'lr': 0.000124221444651712, 'samples': 17436672, 'steps': 34055, 'loss/train': 1.0910838842391968}
+02/25/2022 21:40:57 - INFO - codeparrot_training - Step 34056: {'lr': 0.0001242073041556505, 'samples': 17437184, 'steps': 34056, 'loss/train': 1.9763319492340088}
+02/25/2022 21:41:04 - INFO - codeparrot_training - Step 34057: {'lr': 0.00012419316419844475, 'samples': 17437696, 'steps': 34057, 'loss/train': 1.1914631128311157}
+02/25/2022 21:41:07 - INFO - codeparrot_training - Step 34058: {'lr': 0.00012417902478015566, 'samples': 17438208, 'steps': 34058, 'loss/train': 1.6025583744049072}
+02/25/2022 21:41:13 - INFO - codeparrot_training - Step 34059: {'lr': 0.00012416488590084358, 'samples': 17438720, 'steps': 34059, 'loss/train': 0.6146384477615356}
+02/25/2022 21:41:16 - INFO - codeparrot_training - Step 34060: {'lr': 0.00012415074756056923, 'samples': 17439232, 'steps': 34060, 'loss/train': 1.5017462968826294}
+02/25/2022 21:41:22 - INFO - codeparrot_training - Step 34061: {'lr': 0.00012413660975939316, 'samples': 17439744, 'steps': 34061, 'loss/train': 1.7307779788970947}
+02/25/2022 21:41:25 - INFO - codeparrot_training - Step 34062: {'lr': 0.00012412247249737586, 'samples': 17440256, 'steps': 34062, 'loss/train': 1.2051664590835571}
+02/25/2022 21:41:31 - INFO - codeparrot_training - Step 34063: {'lr': 0.00012410833577457785, 'samples': 17440768, 'steps': 34063, 'loss/train': 1.3565372228622437}
+02/25/2022 21:41:35 - INFO - codeparrot_training - Step 34064: {'lr': 0.0001240941995910598, 'samples': 17441280, 'steps': 34064, 'loss/train': 2.0141043663024902}
+02/25/2022 21:41:41 - INFO - codeparrot_training - Step 34065: {'lr': 0.00012408006394688226, 'samples': 17441792, 'steps': 34065, 'loss/train': 1.417724847793579}
+02/25/2022 21:41:44 - INFO - codeparrot_training - Step 34066: {'lr': 0.0001240659288421057, 'samples': 17442304, 'steps': 34066, 'loss/train': 2.277214765548706}
+02/25/2022 21:41:51 - INFO - codeparrot_training - Step 34067: {'lr': 0.00012405179427679068, 'samples': 17442816, 'steps': 34067, 'loss/train': 1.711771011352539}
+02/25/2022 21:41:54 - INFO - codeparrot_training - Step 34068: {'lr': 0.00012403766025099783, 'samples': 17443328, 'steps': 34068, 'loss/train': 0.7656440138816833}
+02/25/2022 21:42:00 - INFO - codeparrot_training - Step 34069: {'lr': 0.00012402352676478757, 'samples': 17443840, 'steps': 34069, 'loss/train': 2.886143684387207}
+02/25/2022 21:42:03 - INFO - codeparrot_training - Step 34070: {'lr': 0.00012400939381822072, 'samples': 17444352, 'steps': 34070, 'loss/train': 1.0049716234207153}
+02/25/2022 21:42:09 - INFO - codeparrot_training - Step 34071: {'lr': 0.00012399526141135739, 'samples': 17444864, 'steps': 34071, 'loss/train': 2.914724588394165}
+02/25/2022 21:42:13 - INFO - codeparrot_training - Step 34072: {'lr': 0.00012398112954425846, 'samples': 17445376, 'steps': 34072, 'loss/train': 2.105363607406616}
+02/25/2022 21:42:18 - INFO - codeparrot_training - Step 34073: {'lr': 0.0001239669982169843, 'samples': 17445888, 'steps': 34073, 'loss/train': 2.1937716007232666}
+02/25/2022 21:42:22 - INFO - codeparrot_training - Step 34074: {'lr': 0.00012395286742959566, 'samples': 17446400, 'steps': 34074, 'loss/train': 0.2757807672023773}
+02/25/2022 21:42:27 - INFO - codeparrot_training - Step 34075: {'lr': 0.0001239387371821527, 'samples': 17446912, 'steps': 34075, 'loss/train': 0.8021522760391235}
+02/25/2022 21:42:31 - INFO - codeparrot_training - Step 34076: {'lr': 0.00012392460747471628, 'samples': 17447424, 'steps': 34076, 'loss/train': 0.6329545378684998}
+02/25/2022 21:42:36 - INFO - codeparrot_training - Step 34077: {'lr': 0.00012391047830734672, 'samples': 17447936, 'steps': 34077, 'loss/train': 2.679614543914795}
+02/25/2022 21:42:40 - INFO - codeparrot_training - Step 34078: {'lr': 0.0001238963496801048, 'samples': 17448448, 'steps': 34078, 'loss/train': 1.8143888711929321}
+02/25/2022 21:42:45 - INFO - codeparrot_training - Step 34079: {'lr': 0.0001238822215930507, 'samples': 17448960, 'steps': 34079, 'loss/train': 1.9522573947906494}
+02/25/2022 21:42:49 - INFO - codeparrot_training - Step 34080: {'lr': 0.00012386809404624523, 'samples': 17449472, 'steps': 34080, 'loss/train': 0.7795721888542175}
+02/25/2022 21:42:54 - INFO - codeparrot_training - Step 34081: {'lr': 0.00012385396703974865, 'samples': 17449984, 'steps': 34081, 'loss/train': 2.5577166080474854}
+02/25/2022 21:42:58 - INFO - codeparrot_training - Step 34082: {'lr': 0.00012383984057362174, 'samples': 17450496, 'steps': 34082, 'loss/train': 2.729487419128418}
+02/25/2022 21:43:03 - INFO - codeparrot_training - Step 34083: {'lr': 0.00012382571464792492, 'samples': 17451008, 'steps': 34083, 'loss/train': 1.553989052772522}
+02/25/2022 21:43:07 - INFO - codeparrot_training - Step 34084: {'lr': 0.00012381158926271863, 'samples': 17451520, 'steps': 34084, 'loss/train': 1.9220963716506958}
+02/25/2022 21:43:13 - INFO - codeparrot_training - Step 34085: {'lr': 0.00012379746441806345, 'samples': 17452032, 'steps': 34085, 'loss/train': 1.4028511047363281}
+02/25/2022 21:43:17 - INFO - codeparrot_training - Step 34086: {'lr': 0.00012378334011401977, 'samples': 17452544, 'steps': 34086, 'loss/train': 1.713929295539856}
+02/25/2022 21:43:22 - INFO - codeparrot_training - Step 34087: {'lr': 0.00012376921635064825, 'samples': 17453056, 'steps': 34087, 'loss/train': 2.0141427516937256}
+02/25/2022 21:43:26 - INFO - codeparrot_training - Step 34088: {'lr': 0.00012375509312800934, 'samples': 17453568, 'steps': 34088, 'loss/train': 1.5257792472839355}
+02/25/2022 21:43:31 - INFO - codeparrot_training - Step 34089: {'lr': 0.0001237409704461635, 'samples': 17454080, 'steps': 34089, 'loss/train': 2.831057548522949}
+02/25/2022 21:43:35 - INFO - codeparrot_training - Step 34090: {'lr': 0.00012372684830517117, 'samples': 17454592, 'steps': 34090, 'loss/train': 1.6461092233657837}
+02/25/2022 21:43:40 - INFO - codeparrot_training - Step 34091: {'lr': 0.00012371272670509304, 'samples': 17455104, 'steps': 34091, 'loss/train': 1.003557801246643}
+02/25/2022 21:43:44 - INFO - codeparrot_training - Step 34092: {'lr': 0.00012369860564598946, 'samples': 17455616, 'steps': 34092, 'loss/train': 1.6743559837341309}
+02/25/2022 21:43:50 - INFO - codeparrot_training - Step 34093: {'lr': 0.00012368448512792095, 'samples': 17456128, 'steps': 34093, 'loss/train': 2.147613048553467}
+02/25/2022 21:43:54 - INFO - codeparrot_training - Step 34094: {'lr': 0.00012367036515094793, 'samples': 17456640, 'steps': 34094, 'loss/train': 2.846097946166992}
+02/25/2022 21:43:59 - INFO - codeparrot_training - Step 34095: {'lr': 0.00012365624571513103, 'samples': 17457152, 'steps': 34095, 'loss/train': 1.2241394519805908}
+02/25/2022 21:44:03 - INFO - codeparrot_training - Step 34096: {'lr': 0.00012364212682053067, 'samples': 17457664, 'steps': 34096, 'loss/train': 2.141115188598633}
+02/25/2022 21:44:08 - INFO - codeparrot_training - Step 34097: {'lr': 0.0001236280084672073, 'samples': 17458176, 'steps': 34097, 'loss/train': 2.5461888313293457}
+02/25/2022 21:44:12 - INFO - codeparrot_training - Step 34098: {'lr': 0.0001236138906552213, 'samples': 17458688, 'steps': 34098, 'loss/train': 0.9235637187957764}
+02/25/2022 21:44:17 - INFO - codeparrot_training - Step 34099: {'lr': 0.0001235997733846334, 'samples': 17459200, 'steps': 34099, 'loss/train': 1.9525666236877441}
+02/25/2022 21:44:21 - INFO - codeparrot_training - Step 34100: {'lr': 0.0001235856566555039, 'samples': 17459712, 'steps': 34100, 'loss/train': 1.4226510524749756}
+02/25/2022 21:44:27 - INFO - codeparrot_training - Step 34101: {'lr': 0.0001235715404678933, 'samples': 17460224, 'steps': 34101, 'loss/train': 1.412358283996582}
+02/25/2022 21:44:30 - INFO - codeparrot_training - Step 34102: {'lr': 0.00012355742482186198, 'samples': 17460736, 'steps': 34102, 'loss/train': 1.0481382608413696}
+02/25/2022 21:44:37 - INFO - codeparrot_training - Step 34103: {'lr': 0.00012354330971747063, 'samples': 17461248, 'steps': 34103, 'loss/train': 2.2695329189300537}
+02/25/2022 21:44:41 - INFO - codeparrot_training - Step 34104: {'lr': 0.00012352919515477956, 'samples': 17461760, 'steps': 34104, 'loss/train': 1.5227618217468262}
+02/25/2022 21:44:46 - INFO - codeparrot_training - Step 34105: {'lr': 0.0001235150811338493, 'samples': 17462272, 'steps': 34105, 'loss/train': 1.0407601594924927}
+02/25/2022 21:44:50 - INFO - codeparrot_training - Step 34106: {'lr': 0.0001235009676547401, 'samples': 17462784, 'steps': 34106, 'loss/train': 1.8043757677078247}
+02/25/2022 21:44:55 - INFO - codeparrot_training - Step 34107: {'lr': 0.00012348685471751277, 'samples': 17463296, 'steps': 34107, 'loss/train': 0.4909404218196869}
+02/25/2022 21:44:59 - INFO - codeparrot_training - Step 34108: {'lr': 0.0001234727423222274, 'samples': 17463808, 'steps': 34108, 'loss/train': 1.8534376621246338}
+02/25/2022 21:45:04 - INFO - codeparrot_training - Step 34109: {'lr': 0.00012345863046894488, 'samples': 17464320, 'steps': 34109, 'loss/train': 1.036161184310913}
+02/25/2022 21:45:08 - INFO - codeparrot_training - Step 34110: {'lr': 0.00012344451915772517, 'samples': 17464832, 'steps': 34110, 'loss/train': 1.4591485261917114}
+02/25/2022 21:45:13 - INFO - codeparrot_training - Step 34111: {'lr': 0.00012343040838862907, 'samples': 17465344, 'steps': 34111, 'loss/train': 1.9940106868743896}
+02/25/2022 21:45:17 - INFO - codeparrot_training - Step 34112: {'lr': 0.0001234162981617168, 'samples': 17465856, 'steps': 34112, 'loss/train': 1.9499503374099731}
+02/25/2022 21:45:22 - INFO - codeparrot_training - Step 34113: {'lr': 0.00012340218847704915, 'samples': 17466368, 'steps': 34113, 'loss/train': 1.8292899131774902}
+02/25/2022 21:45:25 - INFO - codeparrot_training - Step 34114: {'lr': 0.0001233880793346861, 'samples': 17466880, 'steps': 34114, 'loss/train': 1.6427180767059326}
+02/25/2022 21:45:33 - INFO - codeparrot_training - Step 34115: {'lr': 0.0001233739707346884, 'samples': 17467392, 'steps': 34115, 'loss/train': 0.5149872303009033}
+02/25/2022 21:45:36 - INFO - codeparrot_training - Step 34116: {'lr': 0.00012335986267711632, 'samples': 17467904, 'steps': 34116, 'loss/train': 1.7666956186294556}
+02/25/2022 21:45:42 - INFO - codeparrot_training - Step 34117: {'lr': 0.00012334575516203048, 'samples': 17468416, 'steps': 34117, 'loss/train': 1.8671932220458984}
+02/25/2022 21:45:45 - INFO - codeparrot_training - Step 34118: {'lr': 0.0001233316481894912, 'samples': 17468928, 'steps': 34118, 'loss/train': 1.2974510192871094}
+02/25/2022 21:45:51 - INFO - codeparrot_training - Step 34119: {'lr': 0.0001233175417595589, 'samples': 17469440, 'steps': 34119, 'loss/train': 0.8795772790908813}
+02/25/2022 21:45:54 - INFO - codeparrot_training - Step 34120: {'lr': 0.00012330343587229394, 'samples': 17469952, 'steps': 34120, 'loss/train': 0.1629948765039444}
+02/25/2022 21:46:00 - INFO - codeparrot_training - Step 34121: {'lr': 0.00012328933052775694, 'samples': 17470464, 'steps': 34121, 'loss/train': 1.5280299186706543}
+02/25/2022 21:46:03 - INFO - codeparrot_training - Step 34122: {'lr': 0.00012327522572600823, 'samples': 17470976, 'steps': 34122, 'loss/train': 1.272597074508667}
+02/25/2022 21:46:09 - INFO - codeparrot_training - Step 34123: {'lr': 0.00012326112146710815, 'samples': 17471488, 'steps': 34123, 'loss/train': 0.7454657554626465}
+02/25/2022 21:46:14 - INFO - codeparrot_training - Step 34124: {'lr': 0.00012324701775111714, 'samples': 17472000, 'steps': 34124, 'loss/train': 2.791496753692627}
+02/25/2022 21:46:18 - INFO - codeparrot_training - Step 34125: {'lr': 0.00012323291457809574, 'samples': 17472512, 'steps': 34125, 'loss/train': 1.6492421627044678}
+02/25/2022 21:46:21 - INFO - codeparrot_training - Step 34126: {'lr': 0.00012321881194810426, 'samples': 17473024, 'steps': 34126, 'loss/train': 2.4348554611206055}
+02/25/2022 21:46:27 - INFO - codeparrot_training - Step 34127: {'lr': 0.00012320470986120315, 'samples': 17473536, 'steps': 34127, 'loss/train': 2.2779111862182617}
+02/25/2022 21:46:30 - INFO - codeparrot_training - Step 34128: {'lr': 0.00012319060831745272, 'samples': 17474048, 'steps': 34128, 'loss/train': 2.267850399017334}
+02/25/2022 21:46:37 - INFO - codeparrot_training - Step 34129: {'lr': 0.0001231765073169135, 'samples': 17474560, 'steps': 34129, 'loss/train': 0.024210814386606216}
+02/25/2022 21:46:41 - INFO - codeparrot_training - Step 34130: {'lr': 0.00012316240685964592, 'samples': 17475072, 'steps': 34130, 'loss/train': 1.3733785152435303}
+02/25/2022 21:46:46 - INFO - codeparrot_training - Step 34131: {'lr': 0.00012314830694571028, 'samples': 17475584, 'steps': 34131, 'loss/train': 1.220655918121338}
+02/25/2022 21:46:50 - INFO - codeparrot_training - Step 34132: {'lr': 0.000123134207575167, 'samples': 17476096, 'steps': 34132, 'loss/train': 2.375781536102295}
+02/25/2022 21:46:55 - INFO - codeparrot_training - Step 34133: {'lr': 0.0001231201087480764, 'samples': 17476608, 'steps': 34133, 'loss/train': 0.9140170812606812}
+02/25/2022 21:46:59 - INFO - codeparrot_training - Step 34134: {'lr': 0.00012310601046449906, 'samples': 17477120, 'steps': 34134, 'loss/train': 0.8746749758720398}
+02/25/2022 21:47:04 - INFO - codeparrot_training - Step 34135: {'lr': 0.0001230919127244953, 'samples': 17477632, 'steps': 34135, 'loss/train': 1.2982467412948608}
+02/25/2022 21:47:10 - INFO - codeparrot_training - Step 34136: {'lr': 0.0001230778155281255, 'samples': 17478144, 'steps': 34136, 'loss/train': 1.7328559160232544}
+02/25/2022 21:47:13 - INFO - codeparrot_training - Step 34137: {'lr': 0.0001230637188754499, 'samples': 17478656, 'steps': 34137, 'loss/train': 2.060474395751953}
+02/25/2022 21:47:16 - INFO - codeparrot_training - Step 34138: {'lr': 0.00012304962276652915, 'samples': 17479168, 'steps': 34138, 'loss/train': 2.488842487335205}
+02/25/2022 21:47:22 - INFO - codeparrot_training - Step 34139: {'lr': 0.00012303552720142346, 'samples': 17479680, 'steps': 34139, 'loss/train': 2.090043067932129}
+02/25/2022 21:47:29 - INFO - codeparrot_training - Step 34140: {'lr': 0.00012302143218019328, 'samples': 17480192, 'steps': 34140, 'loss/train': 1.0664960145950317}
+02/25/2022 21:47:33 - INFO - codeparrot_training - Step 34141: {'lr': 0.00012300733770289889, 'samples': 17480704, 'steps': 34141, 'loss/train': 2.038123607635498}
+02/25/2022 21:47:38 - INFO - codeparrot_training - Step 34142: {'lr': 0.0001229932437696008, 'samples': 17481216, 'steps': 34142, 'loss/train': 1.902867317199707}
+02/25/2022 21:47:42 - INFO - codeparrot_training - Step 34143: {'lr': 0.00012297915038035926, 'samples': 17481728, 'steps': 34143, 'loss/train': 2.5757765769958496}
+02/25/2022 21:47:47 - INFO - codeparrot_training - Step 34144: {'lr': 0.00012296505753523488, 'samples': 17482240, 'steps': 34144, 'loss/train': 1.0799211263656616}
+02/25/2022 21:47:51 - INFO - codeparrot_training - Step 34145: {'lr': 0.00012295096523428766, 'samples': 17482752, 'steps': 34145, 'loss/train': 2.9951064586639404}
+02/25/2022 21:47:56 - INFO - codeparrot_training - Step 34146: {'lr': 0.00012293687347757828, 'samples': 17483264, 'steps': 34146, 'loss/train': 1.6820772886276245}
+02/25/2022 21:47:59 - INFO - codeparrot_training - Step 34147: {'lr': 0.00012292278226516687, 'samples': 17483776, 'steps': 34147, 'loss/train': 1.5084384679794312}
+02/25/2022 21:48:05 - INFO - codeparrot_training - Step 34148: {'lr': 0.00012290869159711413, 'samples': 17484288, 'steps': 34148, 'loss/train': 1.3514103889465332}
+02/25/2022 21:48:08 - INFO - codeparrot_training - Step 34149: {'lr': 0.00012289460147347998, 'samples': 17484800, 'steps': 34149, 'loss/train': 1.5390719175338745}
+02/25/2022 21:48:15 - INFO - codeparrot_training - Step 34150: {'lr': 0.0001228805118943251, 'samples': 17485312, 'steps': 34150, 'loss/train': 1.820581078529358}
+02/25/2022 21:48:19 - INFO - codeparrot_training - Step 34151: {'lr': 0.00012286642285970964, 'samples': 17485824, 'steps': 34151, 'loss/train': 2.831040382385254}
+02/25/2022 21:48:24 - INFO - codeparrot_training - Step 34152: {'lr': 0.00012285233436969427, 'samples': 17486336, 'steps': 34152, 'loss/train': 1.1981221437454224}
+02/25/2022 21:48:28 - INFO - codeparrot_training - Step 34153: {'lr': 0.0001228382464243389, 'samples': 17486848, 'steps': 34153, 'loss/train': 2.040548801422119}
+02/25/2022 21:48:33 - INFO - codeparrot_training - Step 34154: {'lr': 0.00012282415902370423, 'samples': 17487360, 'steps': 34154, 'loss/train': 8.590916633605957}
+02/25/2022 21:48:37 - INFO - codeparrot_training - Step 34155: {'lr': 0.00012281007216785033, 'samples': 17487872, 'steps': 34155, 'loss/train': 2.0622220039367676}
+02/25/2022 21:48:42 - INFO - codeparrot_training - Step 34156: {'lr': 0.00012279598585683785, 'samples': 17488384, 'steps': 34156, 'loss/train': 1.634106159210205}
+02/25/2022 21:48:46 - INFO - codeparrot_training - Step 34157: {'lr': 0.00012278190009072694, 'samples': 17488896, 'steps': 34157, 'loss/train': 1.7318800687789917}
+02/25/2022 21:48:51 - INFO - codeparrot_training - Step 34158: {'lr': 0.000122767814869578, 'samples': 17489408, 'steps': 34158, 'loss/train': 1.5939161777496338}
+02/25/2022 21:48:54 - INFO - codeparrot_training - Step 34159: {'lr': 0.00012275373019345124, 'samples': 17489920, 'steps': 34159, 'loss/train': 1.2504925727844238}
+02/25/2022 21:49:01 - INFO - codeparrot_training - Step 34160: {'lr': 0.00012273964606240718, 'samples': 17490432, 'steps': 34160, 'loss/train': 2.122093677520752}
+02/25/2022 21:49:05 - INFO - codeparrot_training - Step 34161: {'lr': 0.00012272556247650606, 'samples': 17490944, 'steps': 34161, 'loss/train': 2.037853717803955}
+02/25/2022 21:49:10 - INFO - codeparrot_training - Step 34162: {'lr': 0.00012271147943580823, 'samples': 17491456, 'steps': 34162, 'loss/train': 1.8190076351165771}
+02/25/2022 21:49:14 - INFO - codeparrot_training - Step 34163: {'lr': 0.00012269739694037392, 'samples': 17491968, 'steps': 34163, 'loss/train': 2.6141421794891357}
+02/25/2022 21:49:20 - INFO - codeparrot_training - Step 34164: {'lr': 0.0001226833149902636, 'samples': 17492480, 'steps': 34164, 'loss/train': 2.3264358043670654}
+02/25/2022 21:49:23 - INFO - codeparrot_training - Step 34165: {'lr': 0.00012266923358553755, 'samples': 17492992, 'steps': 34165, 'loss/train': 1.684666395187378}
+02/25/2022 21:49:27 - INFO - codeparrot_training - Step 34166: {'lr': 0.00012265515272625606, 'samples': 17493504, 'steps': 34166, 'loss/train': 3.095144271850586}
+02/25/2022 21:49:32 - INFO - codeparrot_training - Step 34167: {'lr': 0.00012264107241247938, 'samples': 17494016, 'steps': 34167, 'loss/train': 1.890511155128479}
+02/25/2022 21:49:36 - INFO - codeparrot_training - Step 34168: {'lr': 0.000122626992644268, 'samples': 17494528, 'steps': 34168, 'loss/train': 2.21645450592041}
+02/25/2022 21:49:41 - INFO - codeparrot_training - Step 34169: {'lr': 0.00012261291342168212, 'samples': 17495040, 'steps': 34169, 'loss/train': 2.033522129058838}
+02/25/2022 21:49:44 - INFO - codeparrot_training - Step 34170: {'lr': 0.00012259883474478208, 'samples': 17495552, 'steps': 34170, 'loss/train': 2.2934815883636475}
+02/25/2022 21:49:50 - INFO - codeparrot_training - Step 34171: {'lr': 0.00012258475661362808, 'samples': 17496064, 'steps': 34171, 'loss/train': 1.2459129095077515}
+02/25/2022 21:49:54 - INFO - codeparrot_training - Step 34172: {'lr': 0.0001225706790282806, 'samples': 17496576, 'steps': 34172, 'loss/train': 0.9784200191497803}
+02/25/2022 21:49:59 - INFO - codeparrot_training - Step 34173: {'lr': 0.0001225566019887999, 'samples': 17497088, 'steps': 34173, 'loss/train': 1.7030739784240723}
+02/25/2022 21:50:03 - INFO - codeparrot_training - Step 34174: {'lr': 0.00012254252549524626, 'samples': 17497600, 'steps': 34174, 'loss/train': 1.4188021421432495}
+02/25/2022 21:50:08 - INFO - codeparrot_training - Step 34175: {'lr': 0.00012252844954767982, 'samples': 17498112, 'steps': 34175, 'loss/train': 1.7169009447097778}
+02/25/2022 21:50:11 - INFO - codeparrot_training - Step 34176: {'lr': 0.00012251437414616118, 'samples': 17498624, 'steps': 34176, 'loss/train': 1.7522066831588745}
+02/25/2022 21:50:19 - INFO - codeparrot_training - Step 34177: {'lr': 0.00012250029929075044, 'samples': 17499136, 'steps': 34177, 'loss/train': 0.4757154583930969}
+02/25/2022 21:50:22 - INFO - codeparrot_training - Step 34178: {'lr': 0.00012248622498150786, 'samples': 17499648, 'steps': 34178, 'loss/train': 1.1543480157852173}
+02/25/2022 21:50:27 - INFO - codeparrot_training - Step 34179: {'lr': 0.0001224721512184939, 'samples': 17500160, 'steps': 34179, 'loss/train': 0.6923044323921204}
+02/25/2022 21:50:31 - INFO - codeparrot_training - Step 34180: {'lr': 0.0001224580780017687, 'samples': 17500672, 'steps': 34180, 'loss/train': 2.0192697048187256}
+02/25/2022 21:50:37 - INFO - codeparrot_training - Step 34181: {'lr': 0.00012244400533139264, 'samples': 17501184, 'steps': 34181, 'loss/train': 1.540331482887268}
+02/25/2022 21:50:40 - INFO - codeparrot_training - Step 34182: {'lr': 0.00012242993320742585, 'samples': 17501696, 'steps': 34182, 'loss/train': 1.4672720432281494}
+02/25/2022 21:50:45 - INFO - codeparrot_training - Step 34183: {'lr': 0.00012241586162992893, 'samples': 17502208, 'steps': 34183, 'loss/train': 1.8776568174362183}
+02/25/2022 21:50:49 - INFO - codeparrot_training - Step 34184: {'lr': 0.00012240179059896171, 'samples': 17502720, 'steps': 34184, 'loss/train': 1.552356243133545}
+02/25/2022 21:50:54 - INFO - codeparrot_training - Step 34185: {'lr': 0.00012238772011458486, 'samples': 17503232, 'steps': 34185, 'loss/train': 0.6263514161109924}
+02/25/2022 21:50:58 - INFO - codeparrot_training - Step 34186: {'lr': 0.00012237365017685835, 'samples': 17503744, 'steps': 34186, 'loss/train': 1.6560841798782349}
+02/25/2022 21:51:05 - INFO - codeparrot_training - Step 34187: {'lr': 0.0001223595807858428, 'samples': 17504256, 'steps': 34187, 'loss/train': 0.9910959601402283}
+02/25/2022 21:51:08 - INFO - codeparrot_training - Step 34188: {'lr': 0.00012234551194159808, 'samples': 17504768, 'steps': 34188, 'loss/train': 1.9278590679168701}
+02/25/2022 21:51:14 - INFO - codeparrot_training - Step 34189: {'lr': 0.00012233144364418475, 'samples': 17505280, 'steps': 34189, 'loss/train': 2.078850030899048}
+02/25/2022 21:51:17 - INFO - codeparrot_training - Step 34190: {'lr': 0.0001223173758936629, 'samples': 17505792, 'steps': 34190, 'loss/train': 1.0764309167861938}
+02/25/2022 21:51:23 - INFO - codeparrot_training - Step 34191: {'lr': 0.00012230330869009292, 'samples': 17506304, 'steps': 34191, 'loss/train': 1.2835830450057983}
+02/25/2022 21:51:26 - INFO - codeparrot_training - Step 34192: {'lr': 0.00012228924203353507, 'samples': 17506816, 'steps': 34192, 'loss/train': 0.9146235585212708}
+02/25/2022 21:51:32 - INFO - codeparrot_training - Step 34193: {'lr': 0.00012227517592404953, 'samples': 17507328, 'steps': 34193, 'loss/train': 1.3004958629608154}
+02/25/2022 21:51:35 - INFO - codeparrot_training - Step 34194: {'lr': 0.00012226111036169648, 'samples': 17507840, 'steps': 34194, 'loss/train': 1.843943476676941}
+02/25/2022 21:51:42 - INFO - codeparrot_training - Step 34195: {'lr': 0.00012224704534653636, 'samples': 17508352, 'steps': 34195, 'loss/train': 0.8542377948760986}
+02/25/2022 21:51:45 - INFO - codeparrot_training - Step 34196: {'lr': 0.00012223298087862936, 'samples': 17508864, 'steps': 34196, 'loss/train': 3.116443157196045}
+02/25/2022 21:51:51 - INFO - codeparrot_training - Step 34197: {'lr': 0.00012221891695803565, 'samples': 17509376, 'steps': 34197, 'loss/train': 0.0853521078824997}
+02/25/2022 21:51:54 - INFO - codeparrot_training - Step 34198: {'lr': 0.00012220485358481548, 'samples': 17509888, 'steps': 34198, 'loss/train': 2.267965316772461}
+02/25/2022 21:52:00 - INFO - codeparrot_training - Step 34199: {'lr': 0.00012219079075902921, 'samples': 17510400, 'steps': 34199, 'loss/train': 1.4474376440048218}
+02/25/2022 21:52:03 - INFO - codeparrot_training - Step 34200: {'lr': 0.00012217672848073701, 'samples': 17510912, 'steps': 34200, 'loss/train': 0.03651644289493561}
+02/25/2022 21:52:09 - INFO - codeparrot_training - Step 34201: {'lr': 0.0001221626667499991, 'samples': 17511424, 'steps': 34201, 'loss/train': 0.019949259236454964}
+02/25/2022 21:52:12 - INFO - codeparrot_training - Step 34202: {'lr': 0.00012214860556687568, 'samples': 17511936, 'steps': 34202, 'loss/train': 2.273759365081787}
+02/25/2022 21:52:18 - INFO - codeparrot_training - Step 34203: {'lr': 0.0001221345449314271, 'samples': 17512448, 'steps': 34203, 'loss/train': 2.4573237895965576}
+02/25/2022 21:52:24 - INFO - codeparrot_training - Step 34204: {'lr': 0.00012212048484371344, 'samples': 17512960, 'steps': 34204, 'loss/train': 1.961094856262207}
+02/25/2022 21:52:27 - INFO - codeparrot_training - Step 34205: {'lr': 0.0001221064253037952, 'samples': 17513472, 'steps': 34205, 'loss/train': 1.192826747894287}
+02/25/2022 21:52:33 - INFO - codeparrot_training - Step 34206: {'lr': 0.00012209236631173227, 'samples': 17513984, 'steps': 34206, 'loss/train': 0.6642242074012756}
+02/25/2022 21:52:36 - INFO - codeparrot_training - Step 34207: {'lr': 0.0001220783078675851, 'samples': 17514496, 'steps': 34207, 'loss/train': 2.4273128509521484}
+02/25/2022 21:52:42 - INFO - codeparrot_training - Step 34208: {'lr': 0.00012206424997141371, 'samples': 17515008, 'steps': 34208, 'loss/train': 2.315173625946045}
+02/25/2022 21:52:45 - INFO - codeparrot_training - Step 34209: {'lr': 0.00012205019262327868, 'samples': 17515520, 'steps': 34209, 'loss/train': 0.1208794116973877}
+02/25/2022 21:52:51 - INFO - codeparrot_training - Step 34210: {'lr': 0.00012203613582323978, 'samples': 17516032, 'steps': 34210, 'loss/train': 0.37559983134269714}
+02/25/2022 21:52:54 - INFO - codeparrot_training - Step 34211: {'lr': 0.00012202207957135755, 'samples': 17516544, 'steps': 34211, 'loss/train': 1.8713535070419312}
+02/25/2022 21:53:01 - INFO - codeparrot_training - Step 34212: {'lr': 0.000122008023867692, 'samples': 17517056, 'steps': 34212, 'loss/train': 2.0705177783966064}
+02/25/2022 21:53:04 - INFO - codeparrot_training - Step 34213: {'lr': 0.00012199396871230362, 'samples': 17517568, 'steps': 34213, 'loss/train': 1.9738880395889282}
+02/25/2022 21:53:10 - INFO - codeparrot_training - Step 34214: {'lr': 0.00012197991410525225, 'samples': 17518080, 'steps': 34214, 'loss/train': 2.5504486560821533}
+02/25/2022 21:53:13 - INFO - codeparrot_training - Step 34215: {'lr': 0.00012196586004659835, 'samples': 17518592, 'steps': 34215, 'loss/train': 1.540117621421814}
+02/25/2022 21:53:19 - INFO - codeparrot_training - Step 34216: {'lr': 0.00012195180653640206, 'samples': 17519104, 'steps': 34216, 'loss/train': 1.542211651802063}
+02/25/2022 21:53:22 - INFO - codeparrot_training - Step 34217: {'lr': 0.00012193775357472348, 'samples': 17519616, 'steps': 34217, 'loss/train': 0.9172749519348145}
+02/25/2022 21:53:28 - INFO - codeparrot_training - Step 34218: {'lr': 0.00012192370116162299, 'samples': 17520128, 'steps': 34218, 'loss/train': 1.5847469568252563}
+02/25/2022 21:53:31 - INFO - codeparrot_training - Step 34219: {'lr': 0.0001219096492971607, 'samples': 17520640, 'steps': 34219, 'loss/train': 2.484056234359741}
+02/25/2022 21:53:37 - INFO - codeparrot_training - Step 34220: {'lr': 0.00012189559798139682, 'samples': 17521152, 'steps': 34220, 'loss/train': 2.0310730934143066}
+02/25/2022 21:53:40 - INFO - codeparrot_training - Step 34221: {'lr': 0.0001218815472143914, 'samples': 17521664, 'steps': 34221, 'loss/train': 1.2230699062347412}
+02/25/2022 21:53:47 - INFO - codeparrot_training - Step 34222: {'lr': 0.00012186749699620484, 'samples': 17522176, 'steps': 34222, 'loss/train': 1.8630653619766235}
+02/25/2022 21:53:51 - INFO - codeparrot_training - Step 34223: {'lr': 0.00012185344732689723, 'samples': 17522688, 'steps': 34223, 'loss/train': 0.0635826364159584}
+02/25/2022 21:53:56 - INFO - codeparrot_training - Step 34224: {'lr': 0.00012183939820652878, 'samples': 17523200, 'steps': 34224, 'loss/train': 0.801608681678772}
+02/25/2022 21:54:00 - INFO - codeparrot_training - Step 34225: {'lr': 0.00012182534963515956, 'samples': 17523712, 'steps': 34225, 'loss/train': 0.1870885193347931}
+02/25/2022 21:54:05 - INFO - codeparrot_training - Step 34226: {'lr': 0.00012181130161284992, 'samples': 17524224, 'steps': 34226, 'loss/train': 0.91904217004776}
+02/25/2022 21:54:09 - INFO - codeparrot_training - Step 34227: {'lr': 0.00012179725413965997, 'samples': 17524736, 'steps': 34227, 'loss/train': 1.6117677688598633}
+02/25/2022 21:54:14 - INFO - codeparrot_training - Step 34228: {'lr': 0.0001217832072156499, 'samples': 17525248, 'steps': 34228, 'loss/train': 0.7262179851531982}
+02/25/2022 21:54:18 - INFO - codeparrot_training - Step 34229: {'lr': 0.00012176916084087971, 'samples': 17525760, 'steps': 34229, 'loss/train': 1.738732933998108}
+02/25/2022 21:54:23 - INFO - codeparrot_training - Step 34230: {'lr': 0.00012175511501540984, 'samples': 17526272, 'steps': 34230, 'loss/train': 1.9498666524887085}
+02/25/2022 21:54:27 - INFO - codeparrot_training - Step 34231: {'lr': 0.00012174106973930035, 'samples': 17526784, 'steps': 34231, 'loss/train': 0.5765262842178345}
+02/25/2022 21:54:33 - INFO - codeparrot_training - Step 34232: {'lr': 0.00012172702501261138, 'samples': 17527296, 'steps': 34232, 'loss/train': 1.3165740966796875}
+02/25/2022 21:54:37 - INFO - codeparrot_training - Step 34233: {'lr': 0.000121712980835403, 'samples': 17527808, 'steps': 34233, 'loss/train': 1.34809410572052}
+02/25/2022 21:54:42 - INFO - codeparrot_training - Step 34234: {'lr': 0.00012169893720773556, 'samples': 17528320, 'steps': 34234, 'loss/train': 2.336918354034424}
+02/25/2022 21:54:46 - INFO - codeparrot_training - Step 34235: {'lr': 0.00012168489412966916, 'samples': 17528832, 'steps': 34235, 'loss/train': 1.5499122142791748}
+02/25/2022 21:54:51 - INFO - codeparrot_training - Step 34236: {'lr': 0.0001216708516012639, 'samples': 17529344, 'steps': 34236, 'loss/train': 1.5450172424316406}
+02/25/2022 21:54:55 - INFO - codeparrot_training - Step 34237: {'lr': 0.00012165680962257988, 'samples': 17529856, 'steps': 34237, 'loss/train': 1.0320149660110474}
+02/25/2022 21:55:00 - INFO - codeparrot_training - Step 34238: {'lr': 0.00012164276819367742, 'samples': 17530368, 'steps': 34238, 'loss/train': 1.023850917816162}
+02/25/2022 21:55:04 - INFO - codeparrot_training - Step 34239: {'lr': 0.0001216287273146165, 'samples': 17530880, 'steps': 34239, 'loss/train': 1.1832797527313232}
+02/25/2022 21:55:09 - INFO - codeparrot_training - Step 34240: {'lr': 0.00012161468698545755, 'samples': 17531392, 'steps': 34240, 'loss/train': 1.522051215171814}
+02/25/2022 21:55:13 - INFO - codeparrot_training - Step 34241: {'lr': 0.00012160064720626029, 'samples': 17531904, 'steps': 34241, 'loss/train': 1.848900318145752}
+02/25/2022 21:55:19 - INFO - codeparrot_training - Step 34242: {'lr': 0.0001215866079770852, 'samples': 17532416, 'steps': 34242, 'loss/train': 4.349810600280762}
+02/25/2022 21:55:23 - INFO - codeparrot_training - Step 34243: {'lr': 0.00012157256929799223, 'samples': 17532928, 'steps': 34243, 'loss/train': 1.546586275100708}
+02/25/2022 21:55:28 - INFO - codeparrot_training - Step 34244: {'lr': 0.00012155853116904178, 'samples': 17533440, 'steps': 34244, 'loss/train': 2.323444128036499}
+02/25/2022 21:55:32 - INFO - codeparrot_training - Step 34245: {'lr': 0.00012154449359029357, 'samples': 17533952, 'steps': 34245, 'loss/train': 1.5288819074630737}
+02/25/2022 21:55:37 - INFO - codeparrot_training - Step 34246: {'lr': 0.00012153045656180808, 'samples': 17534464, 'steps': 34246, 'loss/train': 1.9346307516098022}
+02/25/2022 21:55:41 - INFO - codeparrot_training - Step 34247: {'lr': 0.00012151642008364524, 'samples': 17534976, 'steps': 34247, 'loss/train': 2.1254489421844482}
+02/25/2022 21:55:46 - INFO - codeparrot_training - Step 34248: {'lr': 0.00012150238415586542, 'samples': 17535488, 'steps': 34248, 'loss/train': 1.7623099088668823}
+02/25/2022 21:55:50 - INFO - codeparrot_training - Step 34249: {'lr': 0.00012148834877852841, 'samples': 17536000, 'steps': 34249, 'loss/train': 1.994359016418457}
+02/25/2022 21:55:55 - INFO - codeparrot_training - Step 34250: {'lr': 0.0001214743139516946, 'samples': 17536512, 'steps': 34250, 'loss/train': 1.7585986852645874}
+02/25/2022 21:55:59 - INFO - codeparrot_training - Step 34251: {'lr': 0.00012146027967542391, 'samples': 17537024, 'steps': 34251, 'loss/train': 1.7589515447616577}
+02/25/2022 21:56:04 - INFO - codeparrot_training - Step 34252: {'lr': 0.00012144624594977669, 'samples': 17537536, 'steps': 34252, 'loss/train': 2.556988000869751}
+02/25/2022 21:56:08 - INFO - codeparrot_training - Step 34253: {'lr': 0.00012143221277481289, 'samples': 17538048, 'steps': 34253, 'loss/train': 1.621940016746521}
+02/25/2022 21:56:14 - INFO - codeparrot_training - Step 34254: {'lr': 0.0001214181801505927, 'samples': 17538560, 'steps': 34254, 'loss/train': 2.0679194927215576}
+02/25/2022 21:56:17 - INFO - codeparrot_training - Step 34255: {'lr': 0.0001214041480771761, 'samples': 17539072, 'steps': 34255, 'loss/train': 2.2575337886810303}
+02/25/2022 21:56:22 - INFO - codeparrot_training - Step 34256: {'lr': 0.00012139011655462338, 'samples': 17539584, 'steps': 34256, 'loss/train': 1.7101577520370483}
+02/25/2022 21:56:26 - INFO - codeparrot_training - Step 34257: {'lr': 0.00012137608558299456, 'samples': 17540096, 'steps': 34257, 'loss/train': 0.5349771976470947}
+02/25/2022 21:56:31 - INFO - codeparrot_training - Step 34258: {'lr': 0.00012136205516234974, 'samples': 17540608, 'steps': 34258, 'loss/train': 1.7355486154556274}
+02/25/2022 21:56:35 - INFO - codeparrot_training - Step 34259: {'lr': 0.00012134802529274897, 'samples': 17541120, 'steps': 34259, 'loss/train': 0.9616206288337708}
+02/25/2022 21:56:41 - INFO - codeparrot_training - Step 34260: {'lr': 0.00012133399597425246, 'samples': 17541632, 'steps': 34260, 'loss/train': 2.388927459716797}
+02/25/2022 21:56:45 - INFO - codeparrot_training - Step 34261: {'lr': 0.00012131996720692026, 'samples': 17542144, 'steps': 34261, 'loss/train': 1.3599094152450562}
+02/25/2022 21:56:50 - INFO - codeparrot_training - Step 34262: {'lr': 0.00012130593899081249, 'samples': 17542656, 'steps': 34262, 'loss/train': 2.1522176265716553}
+02/25/2022 21:56:54 - INFO - codeparrot_training - Step 34263: {'lr': 0.00012129191132598919, 'samples': 17543168, 'steps': 34263, 'loss/train': 2.4992573261260986}
+02/25/2022 21:56:59 - INFO - codeparrot_training - Step 34264: {'lr': 0.00012127788421251038, 'samples': 17543680, 'steps': 34264, 'loss/train': 2.077789068222046}
+02/25/2022 21:57:03 - INFO - codeparrot_training - Step 34265: {'lr': 0.00012126385765043635, 'samples': 17544192, 'steps': 34265, 'loss/train': 1.3943564891815186}
+02/25/2022 21:57:08 - INFO - codeparrot_training - Step 34266: {'lr': 0.00012124983163982706, 'samples': 17544704, 'steps': 34266, 'loss/train': 8.703642845153809}
+02/25/2022 21:57:12 - INFO - codeparrot_training - Step 34267: {'lr': 0.00012123580618074262, 'samples': 17545216, 'steps': 34267, 'loss/train': 2.1163244247436523}
+02/25/2022 21:57:17 - INFO - codeparrot_training - Step 34268: {'lr': 0.00012122178127324298, 'samples': 17545728, 'steps': 34268, 'loss/train': 1.6408638954162598}
+02/25/2022 21:57:21 - INFO - codeparrot_training - Step 34269: {'lr': 0.00012120775691738845, 'samples': 17546240, 'steps': 34269, 'loss/train': 2.154496192932129}
+02/25/2022 21:57:27 - INFO - codeparrot_training - Step 34270: {'lr': 0.000121193733113239, 'samples': 17546752, 'steps': 34270, 'loss/train': 2.46219539642334}
+02/25/2022 21:57:31 - INFO - codeparrot_training - Step 34271: {'lr': 0.0001211797098608547, 'samples': 17547264, 'steps': 34271, 'loss/train': 1.1195275783538818}
+02/25/2022 21:57:36 - INFO - codeparrot_training - Step 34272: {'lr': 0.0001211656871602955, 'samples': 17547776, 'steps': 34272, 'loss/train': 0.8301324844360352}
+02/25/2022 21:57:40 - INFO - codeparrot_training - Step 34273: {'lr': 0.00012115166501162168, 'samples': 17548288, 'steps': 34273, 'loss/train': 0.9998868703842163}
+02/25/2022 21:57:45 - INFO - codeparrot_training - Step 34274: {'lr': 0.00012113764341489323, 'samples': 17548800, 'steps': 34274, 'loss/train': 1.7557333707809448}
+02/25/2022 21:57:49 - INFO - codeparrot_training - Step 34275: {'lr': 0.00012112362237017018, 'samples': 17549312, 'steps': 34275, 'loss/train': 2.2235615253448486}
+02/25/2022 21:57:54 - INFO - codeparrot_training - Step 34276: {'lr': 0.00012110960187751255, 'samples': 17549824, 'steps': 34276, 'loss/train': 1.8932520151138306}
+02/25/2022 21:57:58 - INFO - codeparrot_training - Step 34277: {'lr': 0.0001210955819369805, 'samples': 17550336, 'steps': 34277, 'loss/train': 3.0477845668792725}
+02/25/2022 21:58:03 - INFO - codeparrot_training - Step 34278: {'lr': 0.000121081562548634, 'samples': 17550848, 'steps': 34278, 'loss/train': 1.579720377922058}
+02/25/2022 21:58:07 - INFO - codeparrot_training - Step 34279: {'lr': 0.00012106754371253331, 'samples': 17551360, 'steps': 34279, 'loss/train': 2.263288974761963}
+02/25/2022 21:58:13 - INFO - codeparrot_training - Step 34280: {'lr': 0.00012105352542873815, 'samples': 17551872, 'steps': 34280, 'loss/train': 1.1526834964752197}
+02/25/2022 21:58:17 - INFO - codeparrot_training - Step 34281: {'lr': 0.00012103950769730881, 'samples': 17552384, 'steps': 34281, 'loss/train': 0.21607467532157898}
+02/25/2022 21:58:22 - INFO - codeparrot_training - Step 34282: {'lr': 0.00012102549051830521, 'samples': 17552896, 'steps': 34282, 'loss/train': 1.692558765411377}
+02/25/2022 21:58:28 - INFO - codeparrot_training - Step 34283: {'lr': 0.00012101147389178762, 'samples': 17553408, 'steps': 34283, 'loss/train': 1.4710392951965332}
+02/25/2022 21:58:31 - INFO - codeparrot_training - Step 34284: {'lr': 0.00012099745781781574, 'samples': 17553920, 'steps': 34284, 'loss/train': 1.6302279233932495}
+02/25/2022 21:58:37 - INFO - codeparrot_training - Step 34285: {'lr': 0.00012098344229644989, 'samples': 17554432, 'steps': 34285, 'loss/train': 2.4527428150177}
+02/25/2022 21:58:40 - INFO - codeparrot_training - Step 34286: {'lr': 0.0001209694273277499, 'samples': 17554944, 'steps': 34286, 'loss/train': 0.20489583909511566}
+02/25/2022 21:58:46 - INFO - codeparrot_training - Step 34287: {'lr': 0.000120955412911776, 'samples': 17555456, 'steps': 34287, 'loss/train': 2.0604114532470703}
+02/25/2022 21:58:49 - INFO - codeparrot_training - Step 34288: {'lr': 0.00012094139904858814, 'samples': 17555968, 'steps': 34288, 'loss/train': 2.0850396156311035}
+02/25/2022 21:58:55 - INFO - codeparrot_training - Step 34289: {'lr': 0.00012092738573824636, 'samples': 17556480, 'steps': 34289, 'loss/train': 2.975572109222412}
+02/25/2022 21:58:59 - INFO - codeparrot_training - Step 34290: {'lr': 0.00012091337298081054, 'samples': 17556992, 'steps': 34290, 'loss/train': 1.118453025817871}
+02/25/2022 21:59:04 - INFO - codeparrot_training - Step 34291: {'lr': 0.00012089936077634098, 'samples': 17557504, 'steps': 34291, 'loss/train': 2.4356935024261475}
+02/25/2022 21:59:08 - INFO - codeparrot_training - Step 34292: {'lr': 0.00012088534912489754, 'samples': 17558016, 'steps': 34292, 'loss/train': 2.5965936183929443}
+02/25/2022 21:59:13 - INFO - codeparrot_training - Step 34293: {'lr': 0.00012087133802654027, 'samples': 17558528, 'steps': 34293, 'loss/train': 1.1163548231124878}
+02/25/2022 21:59:17 - INFO - codeparrot_training - Step 34294: {'lr': 0.00012085732748132908, 'samples': 17559040, 'steps': 34294, 'loss/train': 1.2215179204940796}
+02/25/2022 21:59:22 - INFO - codeparrot_training - Step 34295: {'lr': 0.0001208433174893242, 'samples': 17559552, 'steps': 34295, 'loss/train': 1.8212321996688843}
+02/25/2022 21:59:26 - INFO - codeparrot_training - Step 34296: {'lr': 0.00012082930805058553, 'samples': 17560064, 'steps': 34296, 'loss/train': 1.0686501264572144}
+02/25/2022 21:59:31 - INFO - codeparrot_training - Step 34297: {'lr': 0.00012081529916517309, 'samples': 17560576, 'steps': 34297, 'loss/train': 1.660291075706482}
+02/25/2022 21:59:35 - INFO - codeparrot_training - Step 34298: {'lr': 0.00012080129083314682, 'samples': 17561088, 'steps': 34298, 'loss/train': 2.1139681339263916}
+02/25/2022 21:59:40 - INFO - codeparrot_training - Step 34299: {'lr': 0.00012078728305456685, 'samples': 17561600, 'steps': 34299, 'loss/train': 0.8264246582984924}
+02/25/2022 21:59:44 - INFO - codeparrot_training - Step 34300: {'lr': 0.00012077327582949313, 'samples': 17562112, 'steps': 34300, 'loss/train': 1.4739696979522705}
+02/25/2022 21:59:49 - INFO - codeparrot_training - Step 34301: {'lr': 0.00012075926915798568, 'samples': 17562624, 'steps': 34301, 'loss/train': 2.240708827972412}
+02/25/2022 21:59:53 - INFO - codeparrot_training - Step 34302: {'lr': 0.00012074526304010438, 'samples': 17563136, 'steps': 34302, 'loss/train': 2.087946891784668}
+02/25/2022 22:00:00 - INFO - codeparrot_training - Step 34303: {'lr': 0.00012073125747590943, 'samples': 17563648, 'steps': 34303, 'loss/train': 1.0485347509384155}
+02/25/2022 22:00:03 - INFO - codeparrot_training - Step 34304: {'lr': 0.00012071725246546073, 'samples': 17564160, 'steps': 34304, 'loss/train': 0.7819700837135315}
+02/25/2022 22:00:09 - INFO - codeparrot_training - Step 34305: {'lr': 0.00012070324800881826, 'samples': 17564672, 'steps': 34305, 'loss/train': 0.3822716772556305}
+02/25/2022 22:00:12 - INFO - codeparrot_training - Step 34306: {'lr': 0.00012068924410604194, 'samples': 17565184, 'steps': 34306, 'loss/train': 1.613389492034912}
+02/25/2022 22:00:18 - INFO - codeparrot_training - Step 34307: {'lr': 0.00012067524075719191, 'samples': 17565696, 'steps': 34307, 'loss/train': 1.0227255821228027}
+02/25/2022 22:00:21 - INFO - codeparrot_training - Step 34308: {'lr': 0.00012066123796232808, 'samples': 17566208, 'steps': 34308, 'loss/train': 1.2361652851104736}
+02/25/2022 22:00:27 - INFO - codeparrot_training - Step 34309: {'lr': 0.00012064723572151045, 'samples': 17566720, 'steps': 34309, 'loss/train': 0.935239851474762}
+02/25/2022 22:00:30 - INFO - codeparrot_training - Step 34310: {'lr': 0.00012063323403479901, 'samples': 17567232, 'steps': 34310, 'loss/train': 1.5385384559631348}
+02/25/2022 22:00:36 - INFO - codeparrot_training - Step 34311: {'lr': 0.0001206192329022536, 'samples': 17567744, 'steps': 34311, 'loss/train': 2.8794918060302734}
+02/25/2022 22:00:40 - INFO - codeparrot_training - Step 34312: {'lr': 0.00012060523232393438, 'samples': 17568256, 'steps': 34312, 'loss/train': 0.7928222417831421}
+02/25/2022 22:00:45 - INFO - codeparrot_training - Step 34313: {'lr': 0.00012059123229990121, 'samples': 17568768, 'steps': 34313, 'loss/train': 1.6463814973831177}
+02/25/2022 22:00:49 - INFO - codeparrot_training - Step 34314: {'lr': 0.00012057723283021426, 'samples': 17569280, 'steps': 34314, 'loss/train': 1.168013095855713}
+02/25/2022 22:00:56 - INFO - codeparrot_training - Step 34315: {'lr': 0.0001205632339149332, 'samples': 17569792, 'steps': 34315, 'loss/train': 1.4635422229766846}
+02/25/2022 22:00:59 - INFO - codeparrot_training - Step 34316: {'lr': 0.0001205492355541182, 'samples': 17570304, 'steps': 34316, 'loss/train': 2.195033311843872}
+02/25/2022 22:01:05 - INFO - codeparrot_training - Step 34317: {'lr': 0.0001205352377478291, 'samples': 17570816, 'steps': 34317, 'loss/train': 8.661101341247559}
+02/25/2022 22:01:08 - INFO - codeparrot_training - Step 34318: {'lr': 0.00012052124049612611, 'samples': 17571328, 'steps': 34318, 'loss/train': 1.613338828086853}
+02/25/2022 22:01:14 - INFO - codeparrot_training - Step 34319: {'lr': 0.00012050724379906883, 'samples': 17571840, 'steps': 34319, 'loss/train': 2.1264991760253906}
+02/25/2022 22:01:17 - INFO - codeparrot_training - Step 34320: {'lr': 0.00012049324765671748, 'samples': 17572352, 'steps': 34320, 'loss/train': 1.6607111692428589}
+02/25/2022 22:01:23 - INFO - codeparrot_training - Step 34321: {'lr': 0.00012047925206913185, 'samples': 17572864, 'steps': 34321, 'loss/train': 1.7710572481155396}
+02/25/2022 22:01:26 - INFO - codeparrot_training - Step 34322: {'lr': 0.00012046525703637217, 'samples': 17573376, 'steps': 34322, 'loss/train': 2.024055242538452}
+02/25/2022 22:01:32 - INFO - codeparrot_training - Step 34323: {'lr': 0.000120451262558498, 'samples': 17573888, 'steps': 34323, 'loss/train': 2.007153272628784}
+02/25/2022 22:01:35 - INFO - codeparrot_training - Step 34324: {'lr': 0.00012043726863556959, 'samples': 17574400, 'steps': 34324, 'loss/train': 3.044768810272217}
+02/25/2022 22:01:42 - INFO - codeparrot_training - Step 34325: {'lr': 0.00012042327526764668, 'samples': 17574912, 'steps': 34325, 'loss/train': 1.6636451482772827}
+02/25/2022 22:01:45 - INFO - codeparrot_training - Step 34326: {'lr': 0.00012040928245478941, 'samples': 17575424, 'steps': 34326, 'loss/train': 1.5949853658676147}
+02/25/2022 22:01:51 - INFO - codeparrot_training - Step 34327: {'lr': 0.00012039529019705762, 'samples': 17575936, 'steps': 34327, 'loss/train': 2.127155303955078}
+02/25/2022 22:01:54 - INFO - codeparrot_training - Step 34328: {'lr': 0.00012038129849451124, 'samples': 17576448, 'steps': 34328, 'loss/train': 0.9926344752311707}
+02/25/2022 22:02:00 - INFO - codeparrot_training - Step 34329: {'lr': 0.00012036730734721014, 'samples': 17576960, 'steps': 34329, 'loss/train': 2.0554497241973877}
+02/25/2022 22:02:03 - INFO - codeparrot_training - Step 34330: {'lr': 0.00012035331675521442, 'samples': 17577472, 'steps': 34330, 'loss/train': 1.907273769378662}
+02/25/2022 22:02:09 - INFO - codeparrot_training - Step 34331: {'lr': 0.00012033932671858394, 'samples': 17577984, 'steps': 34331, 'loss/train': 2.640911340713501}
+02/25/2022 22:02:12 - INFO - codeparrot_training - Step 34332: {'lr': 0.00012032533723737859, 'samples': 17578496, 'steps': 34332, 'loss/train': 0.9879363775253296}
+02/25/2022 22:02:18 - INFO - codeparrot_training - Step 34333: {'lr': 0.00012031134831165824, 'samples': 17579008, 'steps': 34333, 'loss/train': 1.335127592086792}
+02/25/2022 22:02:21 - INFO - codeparrot_training - Step 34334: {'lr': 0.00012029735994148297, 'samples': 17579520, 'steps': 34334, 'loss/train': 0.9870880842208862}
+02/25/2022 22:02:28 - INFO - codeparrot_training - Step 34335: {'lr': 0.00012028337212691254, 'samples': 17580032, 'steps': 34335, 'loss/train': 1.6604045629501343}
+02/25/2022 22:02:31 - INFO - codeparrot_training - Step 34336: {'lr': 0.00012026938486800714, 'samples': 17580544, 'steps': 34336, 'loss/train': 2.1227593421936035}
+02/25/2022 22:02:37 - INFO - codeparrot_training - Step 34337: {'lr': 0.00012025539816482634, 'samples': 17581056, 'steps': 34337, 'loss/train': 1.1126620769500732}
+02/25/2022 22:02:40 - INFO - codeparrot_training - Step 34338: {'lr': 0.00012024141201743027, 'samples': 17581568, 'steps': 34338, 'loss/train': 0.6179150938987732}
+02/25/2022 22:02:46 - INFO - codeparrot_training - Step 34339: {'lr': 0.00012022742642587872, 'samples': 17582080, 'steps': 34339, 'loss/train': 1.9983370304107666}
+02/25/2022 22:02:49 - INFO - codeparrot_training - Step 34340: {'lr': 0.00012021344139023186, 'samples': 17582592, 'steps': 34340, 'loss/train': 1.8165253400802612}
+02/25/2022 22:02:55 - INFO - codeparrot_training - Step 34341: {'lr': 0.00012019945691054918, 'samples': 17583104, 'steps': 34341, 'loss/train': 1.3494136333465576}
+02/25/2022 22:02:58 - INFO - codeparrot_training - Step 34342: {'lr': 0.00012018547298689092, 'samples': 17583616, 'steps': 34342, 'loss/train': 2.2382593154907227}
+02/25/2022 22:03:03 - INFO - codeparrot_training - Step 34343: {'lr': 0.0001201714896193169, 'samples': 17584128, 'steps': 34343, 'loss/train': 1.6111931800842285}
+02/25/2022 22:03:07 - INFO - codeparrot_training - Step 34344: {'lr': 0.00012015750680788698, 'samples': 17584640, 'steps': 34344, 'loss/train': 1.4953608512878418}
+02/25/2022 22:03:13 - INFO - codeparrot_training - Step 34345: {'lr': 0.00012014352455266094, 'samples': 17585152, 'steps': 34345, 'loss/train': 1.877808690071106}
+02/25/2022 22:03:17 - INFO - codeparrot_training - Step 34346: {'lr': 0.00012012954285369895, 'samples': 17585664, 'steps': 34346, 'loss/train': 2.4097847938537598}
+02/25/2022 22:03:22 - INFO - codeparrot_training - Step 34347: {'lr': 0.00012011556171106075, 'samples': 17586176, 'steps': 34347, 'loss/train': 0.13597442209720612}
+02/25/2022 22:03:26 - INFO - codeparrot_training - Step 34348: {'lr': 0.00012010158112480612, 'samples': 17586688, 'steps': 34348, 'loss/train': 2.021965503692627}
+02/25/2022 22:03:31 - INFO - codeparrot_training - Step 34349: {'lr': 0.0001200876010949952, 'samples': 17587200, 'steps': 34349, 'loss/train': 1.5330291986465454}
+02/25/2022 22:03:35 - INFO - codeparrot_training - Step 34350: {'lr': 0.00012007362162168772, 'samples': 17587712, 'steps': 34350, 'loss/train': 2.091512680053711}
+02/25/2022 22:03:40 - INFO - codeparrot_training - Step 34351: {'lr': 0.0001200596427049436, 'samples': 17588224, 'steps': 34351, 'loss/train': 1.4846479892730713}
+02/25/2022 22:03:44 - INFO - codeparrot_training - Step 34352: {'lr': 0.00012004566434482261, 'samples': 17588736, 'steps': 34352, 'loss/train': 1.3995633125305176}
+02/25/2022 22:03:49 - INFO - codeparrot_training - Step 34353: {'lr': 0.00012003168654138482, 'samples': 17589248, 'steps': 34353, 'loss/train': 0.061498068273067474}
+02/25/2022 22:03:53 - INFO - codeparrot_training - Step 34354: {'lr': 0.00012001770929469003, 'samples': 17589760, 'steps': 34354, 'loss/train': 1.894551157951355}
+02/25/2022 22:03:58 - INFO - codeparrot_training - Step 34355: {'lr': 0.00012000373260479807, 'samples': 17590272, 'steps': 34355, 'loss/train': 1.7367933988571167}
+02/25/2022 22:04:02 - INFO - codeparrot_training - Step 34356: {'lr': 0.00011998975647176876, 'samples': 17590784, 'steps': 34356, 'loss/train': 1.460475206375122}
+02/25/2022 22:04:07 - INFO - codeparrot_training - Step 34357: {'lr': 0.00011997578089566228, 'samples': 17591296, 'steps': 34357, 'loss/train': 1.8767181634902954}
+02/25/2022 22:04:11 - INFO - codeparrot_training - Step 34358: {'lr': 0.00011996180587653801, 'samples': 17591808, 'steps': 34358, 'loss/train': 3.2255258560180664}
+02/25/2022 22:04:16 - INFO - codeparrot_training - Step 34359: {'lr': 0.00011994783141445622, 'samples': 17592320, 'steps': 34359, 'loss/train': 1.0728782415390015}
+02/25/2022 22:04:20 - INFO - codeparrot_training - Step 34360: {'lr': 0.00011993385750947649, 'samples': 17592832, 'steps': 34360, 'loss/train': 1.1422830820083618}
+02/25/2022 22:04:26 - INFO - codeparrot_training - Step 34361: {'lr': 0.00011991988416165894, 'samples': 17593344, 'steps': 34361, 'loss/train': 1.5575486421585083}
+02/25/2022 22:04:30 - INFO - codeparrot_training - Step 34362: {'lr': 0.00011990591137106329, 'samples': 17593856, 'steps': 34362, 'loss/train': 1.9952211380004883}
+02/25/2022 22:04:35 - INFO - codeparrot_training - Step 34363: {'lr': 0.0001198919391377494, 'samples': 17594368, 'steps': 34363, 'loss/train': 1.8301459550857544}
+02/25/2022 22:04:39 - INFO - codeparrot_training - Step 34364: {'lr': 0.00011987796746177704, 'samples': 17594880, 'steps': 34364, 'loss/train': 1.5398815870285034}
+02/25/2022 22:04:44 - INFO - codeparrot_training - Step 34365: {'lr': 0.00011986399634320625, 'samples': 17595392, 'steps': 34365, 'loss/train': 2.0846877098083496}
+02/25/2022 22:04:48 - INFO - codeparrot_training - Step 34366: {'lr': 0.00011985002578209678, 'samples': 17595904, 'steps': 34366, 'loss/train': 1.6592000722885132}
+02/25/2022 22:04:53 - INFO - codeparrot_training - Step 34367: {'lr': 0.00011983605577850845, 'samples': 17596416, 'steps': 34367, 'loss/train': 0.9313715696334839}
+02/25/2022 22:04:57 - INFO - codeparrot_training - Step 34368: {'lr': 0.00011982208633250105, 'samples': 17596928, 'steps': 34368, 'loss/train': 2.610107660293579}
+02/25/2022 22:05:02 - INFO - codeparrot_training - Step 34369: {'lr': 0.00011980811744413458, 'samples': 17597440, 'steps': 34369, 'loss/train': 0.9194834232330322}
+02/25/2022 22:05:06 - INFO - codeparrot_training - Step 34370: {'lr': 0.00011979414911346878, 'samples': 17597952, 'steps': 34370, 'loss/train': 1.716711163520813}
+02/25/2022 22:05:12 - INFO - codeparrot_training - Step 34371: {'lr': 0.00011978018134056353, 'samples': 17598464, 'steps': 34371, 'loss/train': 0.9407782554626465}
+02/25/2022 22:05:16 - INFO - codeparrot_training - Step 34372: {'lr': 0.00011976621412547853, 'samples': 17598976, 'steps': 34372, 'loss/train': 1.4251182079315186}
+02/25/2022 22:05:21 - INFO - codeparrot_training - Step 34373: {'lr': 0.00011975224746827379, 'samples': 17599488, 'steps': 34373, 'loss/train': 2.514246702194214}
+02/25/2022 22:05:25 - INFO - codeparrot_training - Step 34374: {'lr': 0.000119738281369009, 'samples': 17600000, 'steps': 34374, 'loss/train': 1.043410062789917}
+02/25/2022 22:05:30 - INFO - codeparrot_training - Step 34375: {'lr': 0.0001197243158277442, 'samples': 17600512, 'steps': 34375, 'loss/train': 0.3327135443687439}
+02/25/2022 22:05:34 - INFO - codeparrot_training - Step 34376: {'lr': 0.0001197103508445389, 'samples': 17601024, 'steps': 34376, 'loss/train': 1.5241411924362183}
+02/25/2022 22:05:39 - INFO - codeparrot_training - Step 34377: {'lr': 0.00011969638641945319, 'samples': 17601536, 'steps': 34377, 'loss/train': 2.4000070095062256}
+02/25/2022 22:05:43 - INFO - codeparrot_training - Step 34378: {'lr': 0.00011968242255254669, 'samples': 17602048, 'steps': 34378, 'loss/train': 1.7739942073822021}
+02/25/2022 22:05:48 - INFO - codeparrot_training - Step 34379: {'lr': 0.00011966845924387948, 'samples': 17602560, 'steps': 34379, 'loss/train': 2.6915862560272217}
+02/25/2022 22:05:55 - INFO - codeparrot_training - Step 34380: {'lr': 0.00011965449649351103, 'samples': 17603072, 'steps': 34380, 'loss/train': 1.2555384635925293}
+02/25/2022 22:05:58 - INFO - codeparrot_training - Step 34381: {'lr': 0.0001196405343015014, 'samples': 17603584, 'steps': 34381, 'loss/train': 1.3932793140411377}
+02/25/2022 22:06:04 - INFO - codeparrot_training - Step 34382: {'lr': 0.00011962657266791027, 'samples': 17604096, 'steps': 34382, 'loss/train': 0.9549170136451721}
+02/25/2022 22:06:07 - INFO - codeparrot_training - Step 34383: {'lr': 0.00011961261159279768, 'samples': 17604608, 'steps': 34383, 'loss/train': 1.380858302116394}
+02/25/2022 22:06:11 - INFO - codeparrot_training - Step 34384: {'lr': 0.00011959865107622306, 'samples': 17605120, 'steps': 34384, 'loss/train': 1.988647699356079}
+02/25/2022 22:06:16 - INFO - codeparrot_training - Step 34385: {'lr': 0.00011958469111824649, 'samples': 17605632, 'steps': 34385, 'loss/train': 1.6716965436935425}
+02/25/2022 22:06:20 - INFO - codeparrot_training - Step 34386: {'lr': 0.00011957073171892763, 'samples': 17606144, 'steps': 34386, 'loss/train': 1.7977254390716553}
+02/25/2022 22:06:25 - INFO - codeparrot_training - Step 34387: {'lr': 0.00011955677287832642, 'samples': 17606656, 'steps': 34387, 'loss/train': 1.5108684301376343}
+02/25/2022 22:06:29 - INFO - codeparrot_training - Step 34388: {'lr': 0.00011954281459650257, 'samples': 17607168, 'steps': 34388, 'loss/train': 2.588124990463257}
+02/25/2022 22:06:36 - INFO - codeparrot_training - Step 34389: {'lr': 0.00011952885687351589, 'samples': 17607680, 'steps': 34389, 'loss/train': 2.1757864952087402}
+02/25/2022 22:06:39 - INFO - codeparrot_training - Step 34390: {'lr': 0.00011951489970942614, 'samples': 17608192, 'steps': 34390, 'loss/train': 1.5195791721343994}
+02/25/2022 22:06:45 - INFO - codeparrot_training - Step 34391: {'lr': 0.00011950094310429304, 'samples': 17608704, 'steps': 34391, 'loss/train': 1.5775372982025146}
+02/25/2022 22:06:48 - INFO - codeparrot_training - Step 34392: {'lr': 0.00011948698705817657, 'samples': 17609216, 'steps': 34392, 'loss/train': 1.64582097530365}
+02/25/2022 22:06:54 - INFO - codeparrot_training - Step 34393: {'lr': 0.0001194730315711364, 'samples': 17609728, 'steps': 34393, 'loss/train': 0.4548017382621765}
+02/25/2022 22:06:57 - INFO - codeparrot_training - Step 34394: {'lr': 0.0001194590766432323, 'samples': 17610240, 'steps': 34394, 'loss/train': 3.014071226119995}
+02/25/2022 22:07:03 - INFO - codeparrot_training - Step 34395: {'lr': 0.00011944512227452398, 'samples': 17610752, 'steps': 34395, 'loss/train': 1.0806884765625}
+02/25/2022 22:07:08 - INFO - codeparrot_training - Step 34396: {'lr': 0.0001194311684650714, 'samples': 17611264, 'steps': 34396, 'loss/train': 1.7794002294540405}
+02/25/2022 22:07:12 - INFO - codeparrot_training - Step 34397: {'lr': 0.00011941721521493424, 'samples': 17611776, 'steps': 34397, 'loss/train': 1.7507359981536865}
+02/25/2022 22:07:15 - INFO - codeparrot_training - Step 34398: {'lr': 0.00011940326252417225, 'samples': 17612288, 'steps': 34398, 'loss/train': 1.9027122259140015}
+02/25/2022 22:07:22 - INFO - codeparrot_training - Step 34399: {'lr': 0.00011938931039284511, 'samples': 17612800, 'steps': 34399, 'loss/train': 1.7898634672164917}
+02/25/2022 22:07:25 - INFO - codeparrot_training - Step 34400: {'lr': 0.00011937535882101281, 'samples': 17613312, 'steps': 34400, 'loss/train': 1.5655403137207031}
+02/25/2022 22:07:31 - INFO - codeparrot_training - Step 34401: {'lr': 0.00011936140780873497, 'samples': 17613824, 'steps': 34401, 'loss/train': 1.9456537961959839}
+02/25/2022 22:07:34 - INFO - codeparrot_training - Step 34402: {'lr': 0.00011934745735607138, 'samples': 17614336, 'steps': 34402, 'loss/train': 1.0505439043045044}
+02/25/2022 22:07:40 - INFO - codeparrot_training - Step 34403: {'lr': 0.00011933350746308172, 'samples': 17614848, 'steps': 34403, 'loss/train': 1.397525668144226}
+02/25/2022 22:07:43 - INFO - codeparrot_training - Step 34404: {'lr': 0.00011931955812982593, 'samples': 17615360, 'steps': 34404, 'loss/train': 1.1197293996810913}
+02/25/2022 22:07:49 - INFO - codeparrot_training - Step 34405: {'lr': 0.00011930560935636361, 'samples': 17615872, 'steps': 34405, 'loss/train': 1.566070795059204}
+02/25/2022 22:07:52 - INFO - codeparrot_training - Step 34406: {'lr': 0.0001192916611427546, 'samples': 17616384, 'steps': 34406, 'loss/train': 2.6954689025878906}
+02/25/2022 22:07:58 - INFO - codeparrot_training - Step 34407: {'lr': 0.0001192777134890585, 'samples': 17616896, 'steps': 34407, 'loss/train': 0.8532391786575317}
+02/25/2022 22:08:01 - INFO - codeparrot_training - Step 34408: {'lr': 0.00011926376639533526, 'samples': 17617408, 'steps': 34408, 'loss/train': 2.053250551223755}
+02/25/2022 22:08:07 - INFO - codeparrot_training - Step 34409: {'lr': 0.00011924981986164443, 'samples': 17617920, 'steps': 34409, 'loss/train': 2.6227874755859375}
+02/25/2022 22:08:11 - INFO - codeparrot_training - Step 34410: {'lr': 0.00011923587388804605, 'samples': 17618432, 'steps': 34410, 'loss/train': 3.803899049758911}
+02/25/2022 22:08:16 - INFO - codeparrot_training - Step 34411: {'lr': 0.00011922192847459948, 'samples': 17618944, 'steps': 34411, 'loss/train': 0.18625019490718842}
+02/25/2022 22:08:20 - INFO - codeparrot_training - Step 34412: {'lr': 0.00011920798362136472, 'samples': 17619456, 'steps': 34412, 'loss/train': 0.25925758481025696}
+02/25/2022 22:08:25 - INFO - codeparrot_training - Step 34413: {'lr': 0.00011919403932840136, 'samples': 17619968, 'steps': 34413, 'loss/train': 1.047576904296875}
+02/25/2022 22:08:29 - INFO - codeparrot_training - Step 34414: {'lr': 0.00011918009559576937, 'samples': 17620480, 'steps': 34414, 'loss/train': 1.256148099899292}
+02/25/2022 22:08:34 - INFO - codeparrot_training - Step 34415: {'lr': 0.00011916615242352814, 'samples': 17620992, 'steps': 34415, 'loss/train': 1.1419812440872192}
+02/25/2022 22:08:38 - INFO - codeparrot_training - Step 34416: {'lr': 0.00011915220981173763, 'samples': 17621504, 'steps': 34416, 'loss/train': 1.4046618938446045}
+02/25/2022 22:08:43 - INFO - codeparrot_training - Step 34417: {'lr': 0.00011913826776045741, 'samples': 17622016, 'steps': 34417, 'loss/train': 1.9058088064193726}
+02/25/2022 22:08:47 - INFO - codeparrot_training - Step 34418: {'lr': 0.00011912432626974751, 'samples': 17622528, 'steps': 34418, 'loss/train': 1.6757359504699707}
+02/25/2022 22:08:53 - INFO - codeparrot_training - Step 34419: {'lr': 0.00011911038533966726, 'samples': 17623040, 'steps': 34419, 'loss/train': 1.6711442470550537}
+02/25/2022 22:08:57 - INFO - codeparrot_training - Step 34420: {'lr': 0.00011909644497027663, 'samples': 17623552, 'steps': 34420, 'loss/train': 1.3533649444580078}
+02/25/2022 22:09:02 - INFO - codeparrot_training - Step 34421: {'lr': 0.00011908250516163516, 'samples': 17624064, 'steps': 34421, 'loss/train': 1.3884401321411133}
+02/25/2022 22:09:06 - INFO - codeparrot_training - Step 34422: {'lr': 0.00011906856591380278, 'samples': 17624576, 'steps': 34422, 'loss/train': 0.41252240538597107}
+02/25/2022 22:09:11 - INFO - codeparrot_training - Step 34423: {'lr': 0.0001190546272268391, 'samples': 17625088, 'steps': 34423, 'loss/train': 2.0497453212738037}
+02/25/2022 22:09:15 - INFO - codeparrot_training - Step 34424: {'lr': 0.00011904068910080379, 'samples': 17625600, 'steps': 34424, 'loss/train': 0.7143895030021667}
+02/25/2022 22:09:20 - INFO - codeparrot_training - Step 34425: {'lr': 0.00011902675153575651, 'samples': 17626112, 'steps': 34425, 'loss/train': 2.6272928714752197}
+02/25/2022 22:09:24 - INFO - codeparrot_training - Step 34426: {'lr': 0.00011901281453175711, 'samples': 17626624, 'steps': 34426, 'loss/train': 1.0482078790664673}
+02/25/2022 22:09:29 - INFO - codeparrot_training - Step 34427: {'lr': 0.00011899887808886522, 'samples': 17627136, 'steps': 34427, 'loss/train': 1.4988723993301392}
+02/25/2022 22:09:32 - INFO - codeparrot_training - Step 34428: {'lr': 0.00011898494220714057, 'samples': 17627648, 'steps': 34428, 'loss/train': 0.9011014699935913}
+02/25/2022 22:09:38 - INFO - codeparrot_training - Step 34429: {'lr': 0.0001189710068866427, 'samples': 17628160, 'steps': 34429, 'loss/train': 1.652367353439331}
+02/25/2022 22:09:42 - INFO - codeparrot_training - Step 34430: {'lr': 0.00011895707212743154, 'samples': 17628672, 'steps': 34430, 'loss/train': 2.3249340057373047}
+02/25/2022 22:09:48 - INFO - codeparrot_training - Step 34431: {'lr': 0.00011894313792956666, 'samples': 17629184, 'steps': 34431, 'loss/train': 1.4634212255477905}
+02/25/2022 22:09:51 - INFO - codeparrot_training - Step 34432: {'lr': 0.00011892920429310778, 'samples': 17629696, 'steps': 34432, 'loss/train': 1.5412893295288086}
+02/25/2022 22:09:57 - INFO - codeparrot_training - Step 34433: {'lr': 0.00011891527121811446, 'samples': 17630208, 'steps': 34433, 'loss/train': 2.243695020675659}
+02/25/2022 22:10:01 - INFO - codeparrot_training - Step 34434: {'lr': 0.00011890133870464658, 'samples': 17630720, 'steps': 34434, 'loss/train': 1.4952956438064575}
+02/25/2022 22:10:06 - INFO - codeparrot_training - Step 34435: {'lr': 0.00011888740675276377, 'samples': 17631232, 'steps': 34435, 'loss/train': 2.9891302585601807}
+02/25/2022 22:10:10 - INFO - codeparrot_training - Step 34436: {'lr': 0.00011887347536252565, 'samples': 17631744, 'steps': 34436, 'loss/train': 1.4947948455810547}
+02/25/2022 22:10:15 - INFO - codeparrot_training - Step 34437: {'lr': 0.00011885954453399192, 'samples': 17632256, 'steps': 34437, 'loss/train': 1.708880066871643}
+02/25/2022 22:10:19 - INFO - codeparrot_training - Step 34438: {'lr': 0.00011884561426722218, 'samples': 17632768, 'steps': 34438, 'loss/train': 1.7759019136428833}
+02/25/2022 22:10:24 - INFO - codeparrot_training - Step 34439: {'lr': 0.00011883168456227628, 'samples': 17633280, 'steps': 34439, 'loss/train': 1.7213183641433716}
+02/25/2022 22:10:28 - INFO - codeparrot_training - Step 34440: {'lr': 0.00011881775541921378, 'samples': 17633792, 'steps': 34440, 'loss/train': 2.7807064056396484}
+02/25/2022 22:10:33 - INFO - codeparrot_training - Step 34441: {'lr': 0.00011880382683809435, 'samples': 17634304, 'steps': 34441, 'loss/train': 2.012974262237549}
+02/25/2022 22:10:37 - INFO - codeparrot_training - Step 34442: {'lr': 0.00011878989881897758, 'samples': 17634816, 'steps': 34442, 'loss/train': 1.3133208751678467}
+02/25/2022 22:10:42 - INFO - codeparrot_training - Step 34443: {'lr': 0.00011877597136192334, 'samples': 17635328, 'steps': 34443, 'loss/train': 1.5778939723968506}
+02/25/2022 22:10:46 - INFO - codeparrot_training - Step 34444: {'lr': 0.00011876204446699104, 'samples': 17635840, 'steps': 34444, 'loss/train': 2.15189266204834}
+02/25/2022 22:10:52 - INFO - codeparrot_training - Step 34445: {'lr': 0.0001187481181342407, 'samples': 17636352, 'steps': 34445, 'loss/train': 1.7066816091537476}
+02/25/2022 22:10:55 - INFO - codeparrot_training - Step 34446: {'lr': 0.0001187341923637315, 'samples': 17636864, 'steps': 34446, 'loss/train': 1.9746837615966797}
+02/25/2022 22:11:01 - INFO - codeparrot_training - Step 34447: {'lr': 0.00011872026715552348, 'samples': 17637376, 'steps': 34447, 'loss/train': 1.8586677312850952}
+02/25/2022 22:11:05 - INFO - codeparrot_training - Step 34448: {'lr': 0.00011870634250967604, 'samples': 17637888, 'steps': 34448, 'loss/train': 1.7036064863204956}
+02/25/2022 22:11:10 - INFO - codeparrot_training - Step 34449: {'lr': 0.00011869241842624912, 'samples': 17638400, 'steps': 34449, 'loss/train': 0.5855467915534973}
+02/25/2022 22:11:14 - INFO - codeparrot_training - Step 34450: {'lr': 0.00011867849490530199, 'samples': 17638912, 'steps': 34450, 'loss/train': 0.8826522827148438}
+02/25/2022 22:11:19 - INFO - codeparrot_training - Step 34451: {'lr': 0.00011866457194689462, 'samples': 17639424, 'steps': 34451, 'loss/train': 1.8082689046859741}
+02/25/2022 22:11:23 - INFO - codeparrot_training - Step 34452: {'lr': 0.0001186506495510864, 'samples': 17639936, 'steps': 34452, 'loss/train': 1.526912808418274}
+02/25/2022 22:11:28 - INFO - codeparrot_training - Step 34453: {'lr': 0.00011863672771793727, 'samples': 17640448, 'steps': 34453, 'loss/train': 1.8211246728897095}
+02/25/2022 22:11:32 - INFO - codeparrot_training - Step 34454: {'lr': 0.0001186228064475065, 'samples': 17640960, 'steps': 34454, 'loss/train': 1.8936933279037476}
+02/25/2022 22:11:38 - INFO - codeparrot_training - Step 34455: {'lr': 0.00011860888573985401, 'samples': 17641472, 'steps': 34455, 'loss/train': 1.4745776653289795}
+02/25/2022 22:11:41 - INFO - codeparrot_training - Step 34456: {'lr': 0.00011859496559503924, 'samples': 17641984, 'steps': 34456, 'loss/train': 1.9079605340957642}
+02/25/2022 22:11:47 - INFO - codeparrot_training - Step 34457: {'lr': 0.000118581046013122, 'samples': 17642496, 'steps': 34457, 'loss/train': 0.6239862442016602}
+02/25/2022 22:11:50 - INFO - codeparrot_training - Step 34458: {'lr': 0.00011856712699416183, 'samples': 17643008, 'steps': 34458, 'loss/train': 2.514531373977661}
+02/25/2022 22:11:56 - INFO - codeparrot_training - Step 34459: {'lr': 0.00011855320853821835, 'samples': 17643520, 'steps': 34459, 'loss/train': 1.8926079273223877}
+02/25/2022 22:11:59 - INFO - codeparrot_training - Step 34460: {'lr': 0.0001185392906453511, 'samples': 17644032, 'steps': 34460, 'loss/train': 1.8797709941864014}
+02/25/2022 22:12:05 - INFO - codeparrot_training - Step 34461: {'lr': 0.00011852537331561988, 'samples': 17644544, 'steps': 34461, 'loss/train': 1.1673511266708374}
+02/25/2022 22:12:08 - INFO - codeparrot_training - Step 34462: {'lr': 0.00011851145654908421, 'samples': 17645056, 'steps': 34462, 'loss/train': 1.8389017581939697}
+02/25/2022 22:12:14 - INFO - codeparrot_training - Step 34463: {'lr': 0.0001184975403458037, 'samples': 17645568, 'steps': 34463, 'loss/train': 2.3040175437927246}
+02/25/2022 22:12:17 - INFO - codeparrot_training - Step 34464: {'lr': 0.0001184836247058379, 'samples': 17646080, 'steps': 34464, 'loss/train': 1.7416069507598877}
+02/25/2022 22:12:24 - INFO - codeparrot_training - Step 34465: {'lr': 0.00011846970962924655, 'samples': 17646592, 'steps': 34465, 'loss/train': 1.2095332145690918}
+02/25/2022 22:12:27 - INFO - codeparrot_training - Step 34466: {'lr': 0.00011845579511608922, 'samples': 17647104, 'steps': 34466, 'loss/train': 1.3330409526824951}
+02/25/2022 22:12:33 - INFO - codeparrot_training - Step 34467: {'lr': 0.00011844188116642548, 'samples': 17647616, 'steps': 34467, 'loss/train': 0.02262728475034237}
+02/25/2022 22:12:36 - INFO - codeparrot_training - Step 34468: {'lr': 0.00011842796778031487, 'samples': 17648128, 'steps': 34468, 'loss/train': 1.7267838716506958}
+02/25/2022 22:12:42 - INFO - codeparrot_training - Step 34469: {'lr': 0.00011841405495781715, 'samples': 17648640, 'steps': 34469, 'loss/train': 1.9224244356155396}
+02/25/2022 22:12:45 - INFO - codeparrot_training - Step 34470: {'lr': 0.00011840014269899186, 'samples': 17649152, 'steps': 34470, 'loss/train': 0.3081327974796295}
+02/25/2022 22:12:51 - INFO - codeparrot_training - Step 34471: {'lr': 0.00011838623100389853, 'samples': 17649664, 'steps': 34471, 'loss/train': 1.6698603630065918}
+02/25/2022 22:12:54 - INFO - codeparrot_training - Step 34472: {'lr': 0.00011837231987259672, 'samples': 17650176, 'steps': 34472, 'loss/train': 2.2035703659057617}
+02/25/2022 22:13:00 - INFO - codeparrot_training - Step 34473: {'lr': 0.00011835840930514618, 'samples': 17650688, 'steps': 34473, 'loss/train': 2.0116829872131348}
+02/25/2022 22:13:03 - INFO - codeparrot_training - Step 34474: {'lr': 0.00011834449930160643, 'samples': 17651200, 'steps': 34474, 'loss/train': 1.978700876235962}
+02/25/2022 22:13:09 - INFO - codeparrot_training - Step 34475: {'lr': 0.00011833058986203704, 'samples': 17651712, 'steps': 34475, 'loss/train': 2.1101579666137695}
+02/25/2022 22:13:13 - INFO - codeparrot_training - Step 34476: {'lr': 0.00011831668098649751, 'samples': 17652224, 'steps': 34476, 'loss/train': 0.505397617816925}
+02/25/2022 22:13:18 - INFO - codeparrot_training - Step 34477: {'lr': 0.00011830277267504758, 'samples': 17652736, 'steps': 34477, 'loss/train': 2.005589723587036}
+02/25/2022 22:13:22 - INFO - codeparrot_training - Step 34478: {'lr': 0.00011828886492774674, 'samples': 17653248, 'steps': 34478, 'loss/train': 2.6046063899993896}
+02/25/2022 22:13:27 - INFO - codeparrot_training - Step 34479: {'lr': 0.0001182749577446546, 'samples': 17653760, 'steps': 34479, 'loss/train': 1.8999263048171997}
+02/25/2022 22:13:31 - INFO - codeparrot_training - Step 34480: {'lr': 0.0001182610511258306, 'samples': 17654272, 'steps': 34480, 'loss/train': 1.9366304874420166}
+02/25/2022 22:13:36 - INFO - codeparrot_training - Step 34481: {'lr': 0.00011824714507133455, 'samples': 17654784, 'steps': 34481, 'loss/train': 1.6994245052337646}
+02/25/2022 22:13:40 - INFO - codeparrot_training - Step 34482: {'lr': 0.00011823323958122586, 'samples': 17655296, 'steps': 34482, 'loss/train': 0.6163861155509949}
+02/25/2022 22:13:45 - INFO - codeparrot_training - Step 34483: {'lr': 0.00011821933465556406, 'samples': 17655808, 'steps': 34483, 'loss/train': 1.8265265226364136}
+02/25/2022 22:13:51 - INFO - codeparrot_training - Step 34484: {'lr': 0.00011820543029440887, 'samples': 17656320, 'steps': 34484, 'loss/train': 2.1469321250915527}
+02/25/2022 22:13:54 - INFO - codeparrot_training - Step 34485: {'lr': 0.00011819152649781978, 'samples': 17656832, 'steps': 34485, 'loss/train': 1.2261356115341187}
+02/25/2022 22:13:58 - INFO - codeparrot_training - Step 34486: {'lr': 0.00011817762326585631, 'samples': 17657344, 'steps': 34486, 'loss/train': 1.5056616067886353}
+02/25/2022 22:14:03 - INFO - codeparrot_training - Step 34487: {'lr': 0.00011816372059857797, 'samples': 17657856, 'steps': 34487, 'loss/train': 1.2098602056503296}
+02/25/2022 22:14:07 - INFO - codeparrot_training - Step 34488: {'lr': 0.00011814981849604459, 'samples': 17658368, 'steps': 34488, 'loss/train': 1.6687403917312622}
+02/25/2022 22:14:12 - INFO - codeparrot_training - Step 34489: {'lr': 0.0001181359169583153, 'samples': 17658880, 'steps': 34489, 'loss/train': 2.331040859222412}
+02/25/2022 22:14:16 - INFO - codeparrot_training - Step 34490: {'lr': 0.00011812201598544998, 'samples': 17659392, 'steps': 34490, 'loss/train': 2.013587474822998}
+02/25/2022 22:14:22 - INFO - codeparrot_training - Step 34491: {'lr': 0.00011810811557750797, 'samples': 17659904, 'steps': 34491, 'loss/train': 8.733814239501953}
+02/25/2022 22:14:25 - INFO - codeparrot_training - Step 34492: {'lr': 0.00011809421573454901, 'samples': 17660416, 'steps': 34492, 'loss/train': 1.5935531854629517}
+02/25/2022 22:14:31 - INFO - codeparrot_training - Step 34493: {'lr': 0.00011808031645663253, 'samples': 17660928, 'steps': 34493, 'loss/train': 1.9496949911117554}
+02/25/2022 22:14:34 - INFO - codeparrot_training - Step 34494: {'lr': 0.0001180664177438181, 'samples': 17661440, 'steps': 34494, 'loss/train': 1.5631678104400635}
+02/25/2022 22:14:40 - INFO - codeparrot_training - Step 34495: {'lr': 0.00011805251959616517, 'samples': 17661952, 'steps': 34495, 'loss/train': 2.1765522956848145}
+02/25/2022 22:14:43 - INFO - codeparrot_training - Step 34496: {'lr': 0.00011803862201373342, 'samples': 17662464, 'steps': 34496, 'loss/train': 1.3541117906570435}
+02/25/2022 22:14:49 - INFO - codeparrot_training - Step 34497: {'lr': 0.0001180247249965823, 'samples': 17662976, 'steps': 34497, 'loss/train': 1.4002761840820312}
+02/25/2022 22:14:52 - INFO - codeparrot_training - Step 34498: {'lr': 0.00011801082854477136, 'samples': 17663488, 'steps': 34498, 'loss/train': 1.5403581857681274}
+02/25/2022 22:14:58 - INFO - codeparrot_training - Step 34499: {'lr': 0.00011799693265836003, 'samples': 17664000, 'steps': 34499, 'loss/train': 1.3468595743179321}
+02/25/2022 22:15:01 - INFO - codeparrot_training - Step 34500: {'lr': 0.00011798303733740801, 'samples': 17664512, 'steps': 34500, 'loss/train': 3.1558167934417725}
+02/25/2022 22:15:08 - INFO - codeparrot_training - Step 34501: {'lr': 0.00011796914258197475, 'samples': 17665024, 'steps': 34501, 'loss/train': 2.0643725395202637}
+02/25/2022 22:15:11 - INFO - codeparrot_training - Step 34502: {'lr': 0.00011795524839211974, 'samples': 17665536, 'steps': 34502, 'loss/train': 1.1217124462127686}
+02/25/2022 22:15:17 - INFO - codeparrot_training - Step 34503: {'lr': 0.00011794135476790243, 'samples': 17666048, 'steps': 34503, 'loss/train': 2.5557968616485596}
+02/25/2022 22:15:20 - INFO - codeparrot_training - Step 34504: {'lr': 0.00011792746170938254, 'samples': 17666560, 'steps': 34504, 'loss/train': 1.7109984159469604}
+02/25/2022 22:15:26 - INFO - codeparrot_training - Step 34505: {'lr': 0.00011791356921661936, 'samples': 17667072, 'steps': 34505, 'loss/train': 2.3901760578155518}
+02/25/2022 22:15:29 - INFO - codeparrot_training - Step 34506: {'lr': 0.00011789967728967269, 'samples': 17667584, 'steps': 34506, 'loss/train': 0.5616239309310913}
+02/25/2022 22:15:35 - INFO - codeparrot_training - Step 34507: {'lr': 0.00011788578592860166, 'samples': 17668096, 'steps': 34507, 'loss/train': 2.199445962905884}
+02/25/2022 22:15:38 - INFO - codeparrot_training - Step 34508: {'lr': 0.00011787189513346607, 'samples': 17668608, 'steps': 34508, 'loss/train': 0.9049603939056396}
+02/25/2022 22:15:44 - INFO - codeparrot_training - Step 34509: {'lr': 0.00011785800490432525, 'samples': 17669120, 'steps': 34509, 'loss/train': 1.4657635688781738}
+02/25/2022 22:15:47 - INFO - codeparrot_training - Step 34510: {'lr': 0.00011784411524123895, 'samples': 17669632, 'steps': 34510, 'loss/train': 2.1990153789520264}
+02/25/2022 22:15:53 - INFO - codeparrot_training - Step 34511: {'lr': 0.0001178302261442663, 'samples': 17670144, 'steps': 34511, 'loss/train': 1.5574162006378174}
+02/25/2022 22:15:57 - INFO - codeparrot_training - Step 34512: {'lr': 0.00011781633761346707, 'samples': 17670656, 'steps': 34512, 'loss/train': 2.137705087661743}
+02/25/2022 22:16:02 - INFO - codeparrot_training - Step 34513: {'lr': 0.00011780244964890063, 'samples': 17671168, 'steps': 34513, 'loss/train': 1.1875412464141846}
+02/25/2022 22:16:06 - INFO - codeparrot_training - Step 34514: {'lr': 0.00011778856225062668, 'samples': 17671680, 'steps': 34514, 'loss/train': 1.5434225797653198}
+02/25/2022 22:16:11 - INFO - codeparrot_training - Step 34515: {'lr': 0.00011777467541870438, 'samples': 17672192, 'steps': 34515, 'loss/train': 2.895737886428833}
+02/25/2022 22:16:15 - INFO - codeparrot_training - Step 34516: {'lr': 0.00011776078915319343, 'samples': 17672704, 'steps': 34516, 'loss/train': 0.9499350190162659}
+02/25/2022 22:16:20 - INFO - codeparrot_training - Step 34517: {'lr': 0.00011774690345415323, 'samples': 17673216, 'steps': 34517, 'loss/train': 1.9704746007919312}
+02/25/2022 22:16:24 - INFO - codeparrot_training - Step 34518: {'lr': 0.00011773301832164338, 'samples': 17673728, 'steps': 34518, 'loss/train': 1.9685696363449097}
+02/25/2022 22:16:29 - INFO - codeparrot_training - Step 34519: {'lr': 0.00011771913375572327, 'samples': 17674240, 'steps': 34519, 'loss/train': 1.28207528591156}
+02/25/2022 22:16:33 - INFO - codeparrot_training - Step 34520: {'lr': 0.00011770524975645239, 'samples': 17674752, 'steps': 34520, 'loss/train': 1.3010730743408203}
+02/25/2022 22:16:38 - INFO - codeparrot_training - Step 34521: {'lr': 0.00011769136632389019, 'samples': 17675264, 'steps': 34521, 'loss/train': 1.799863338470459}
+02/25/2022 22:16:44 - INFO - codeparrot_training - Step 34522: {'lr': 0.00011767748345809609, 'samples': 17675776, 'steps': 34522, 'loss/train': 2.1449368000030518}
+02/25/2022 22:16:48 - INFO - codeparrot_training - Step 34523: {'lr': 0.00011766360115912975, 'samples': 17676288, 'steps': 34523, 'loss/train': 2.3421308994293213}
+02/25/2022 22:16:53 - INFO - codeparrot_training - Step 34524: {'lr': 0.0001176497194270505, 'samples': 17676800, 'steps': 34524, 'loss/train': 1.8416329622268677}
+02/25/2022 22:16:57 - INFO - codeparrot_training - Step 34525: {'lr': 0.00011763583826191781, 'samples': 17677312, 'steps': 34525, 'loss/train': 2.042531728744507}
+02/25/2022 22:17:00 - INFO - codeparrot_training - Step 34526: {'lr': 0.00011762195766379107, 'samples': 17677824, 'steps': 34526, 'loss/train': 8.58413314819336}
+02/25/2022 22:17:06 - INFO - codeparrot_training - Step 34527: {'lr': 0.00011760807763272993, 'samples': 17678336, 'steps': 34527, 'loss/train': 1.0910606384277344}
+02/25/2022 22:17:09 - INFO - codeparrot_training - Step 34528: {'lr': 0.00011759419816879374, 'samples': 17678848, 'steps': 34528, 'loss/train': 1.202926754951477}
+02/25/2022 22:17:15 - INFO - codeparrot_training - Step 34529: {'lr': 0.00011758031927204199, 'samples': 17679360, 'steps': 34529, 'loss/train': 2.0078704357147217}
+02/25/2022 22:17:19 - INFO - codeparrot_training - Step 34530: {'lr': 0.000117566440942534, 'samples': 17679872, 'steps': 34530, 'loss/train': 1.1683276891708374}
+02/25/2022 22:17:24 - INFO - codeparrot_training - Step 34531: {'lr': 0.00011755256318032939, 'samples': 17680384, 'steps': 34531, 'loss/train': 2.735950469970703}
+02/25/2022 22:17:28 - INFO - codeparrot_training - Step 34532: {'lr': 0.00011753868598548756, 'samples': 17680896, 'steps': 34532, 'loss/train': 1.4186333417892456}
+02/25/2022 22:17:33 - INFO - codeparrot_training - Step 34533: {'lr': 0.00011752480935806795, 'samples': 17681408, 'steps': 34533, 'loss/train': 2.1476638317108154}
+02/25/2022 22:17:37 - INFO - codeparrot_training - Step 34534: {'lr': 0.00011751093329812986, 'samples': 17681920, 'steps': 34534, 'loss/train': 1.9056293964385986}
+02/25/2022 22:17:42 - INFO - codeparrot_training - Step 34535: {'lr': 0.000117497057805733, 'samples': 17682432, 'steps': 34535, 'loss/train': 0.7239101529121399}
+02/25/2022 22:17:45 - INFO - codeparrot_training - Step 34536: {'lr': 0.00011748318288093664, 'samples': 17682944, 'steps': 34536, 'loss/train': 2.2706823348999023}
+02/25/2022 22:17:52 - INFO - codeparrot_training - Step 34537: {'lr': 0.00011746930852380028, 'samples': 17683456, 'steps': 34537, 'loss/train': 1.4279333353042603}
+02/25/2022 22:17:56 - INFO - codeparrot_training - Step 34538: {'lr': 0.0001174554347343832, 'samples': 17683968, 'steps': 34538, 'loss/train': 1.6547961235046387}
+02/25/2022 22:18:01 - INFO - codeparrot_training - Step 34539: {'lr': 0.00011744156151274505, 'samples': 17684480, 'steps': 34539, 'loss/train': 1.0733511447906494}
+02/25/2022 22:18:07 - INFO - codeparrot_training - Step 34540: {'lr': 0.00011742768885894515, 'samples': 17684992, 'steps': 34540, 'loss/train': 2.257333517074585}
+02/25/2022 22:18:10 - INFO - codeparrot_training - Step 34541: {'lr': 0.00011741381677304294, 'samples': 17685504, 'steps': 34541, 'loss/train': 1.711201548576355}
+02/25/2022 22:18:16 - INFO - codeparrot_training - Step 34542: {'lr': 0.00011739994525509772, 'samples': 17686016, 'steps': 34542, 'loss/train': 1.2990092039108276}
+02/25/2022 22:18:19 - INFO - codeparrot_training - Step 34543: {'lr': 0.00011738607430516915, 'samples': 17686528, 'steps': 34543, 'loss/train': 0.5107671022415161}
+02/25/2022 22:18:25 - INFO - codeparrot_training - Step 34544: {'lr': 0.00011737220392331644, 'samples': 17687040, 'steps': 34544, 'loss/train': 1.5232945680618286}
+02/25/2022 22:18:28 - INFO - codeparrot_training - Step 34545: {'lr': 0.00011735833410959929, 'samples': 17687552, 'steps': 34545, 'loss/train': 2.2349820137023926}
+02/25/2022 22:18:35 - INFO - codeparrot_training - Step 34546: {'lr': 0.00011734446486407671, 'samples': 17688064, 'steps': 34546, 'loss/train': 1.894779086112976}
+02/25/2022 22:18:38 - INFO - codeparrot_training - Step 34547: {'lr': 0.00011733059618680841, 'samples': 17688576, 'steps': 34547, 'loss/train': 1.9319714307785034}
+02/25/2022 22:18:44 - INFO - codeparrot_training - Step 34548: {'lr': 0.00011731672807785362, 'samples': 17689088, 'steps': 34548, 'loss/train': 1.4587323665618896}
+02/25/2022 22:18:47 - INFO - codeparrot_training - Step 34549: {'lr': 0.00011730286053727201, 'samples': 17689600, 'steps': 34549, 'loss/train': 1.6989874839782715}
+02/25/2022 22:18:53 - INFO - codeparrot_training - Step 34550: {'lr': 0.00011728899356512265, 'samples': 17690112, 'steps': 34550, 'loss/train': 1.7564151287078857}
+02/25/2022 22:18:56 - INFO - codeparrot_training - Step 34551: {'lr': 0.00011727512716146519, 'samples': 17690624, 'steps': 34551, 'loss/train': 0.7209196090698242}
+02/25/2022 22:19:02 - INFO - codeparrot_training - Step 34552: {'lr': 0.00011726126132635884, 'samples': 17691136, 'steps': 34552, 'loss/train': 1.901749610900879}
+02/25/2022 22:19:05 - INFO - codeparrot_training - Step 34553: {'lr': 0.00011724739605986318, 'samples': 17691648, 'steps': 34553, 'loss/train': 1.8601866960525513}
+02/25/2022 22:19:11 - INFO - codeparrot_training - Step 34554: {'lr': 0.00011723353136203752, 'samples': 17692160, 'steps': 34554, 'loss/train': 1.503511667251587}
+02/25/2022 22:19:14 - INFO - codeparrot_training - Step 34555: {'lr': 0.00011721966723294128, 'samples': 17692672, 'steps': 34555, 'loss/train': 2.1501471996307373}
+02/25/2022 22:19:19 - INFO - codeparrot_training - Step 34556: {'lr': 0.0001172058036726337, 'samples': 17693184, 'steps': 34556, 'loss/train': 2.0608112812042236}
+02/25/2022 22:19:23 - INFO - codeparrot_training - Step 34557: {'lr': 0.00011719194068117439, 'samples': 17693696, 'steps': 34557, 'loss/train': 1.8894293308258057}
+02/25/2022 22:19:30 - INFO - codeparrot_training - Step 34558: {'lr': 0.00011717807825862264, 'samples': 17694208, 'steps': 34558, 'loss/train': 2.048642635345459}
+02/25/2022 22:19:33 - INFO - codeparrot_training - Step 34559: {'lr': 0.00011716421640503783, 'samples': 17694720, 'steps': 34559, 'loss/train': 1.0017175674438477}
+02/25/2022 22:19:39 - INFO - codeparrot_training - Step 34560: {'lr': 0.00011715035512047925, 'samples': 17695232, 'steps': 34560, 'loss/train': 1.1716833114624023}
+02/25/2022 22:19:42 - INFO - codeparrot_training - Step 34561: {'lr': 0.00011713649440500646, 'samples': 17695744, 'steps': 34561, 'loss/train': 1.7081788778305054}
+02/25/2022 22:19:48 - INFO - codeparrot_training - Step 34562: {'lr': 0.00011712263425867872, 'samples': 17696256, 'steps': 34562, 'loss/train': 1.4829665422439575}
+02/25/2022 22:19:51 - INFO - codeparrot_training - Step 34563: {'lr': 0.00011710877468155543, 'samples': 17696768, 'steps': 34563, 'loss/train': 1.4972842931747437}
+02/25/2022 22:19:57 - INFO - codeparrot_training - Step 34564: {'lr': 0.00011709491567369584, 'samples': 17697280, 'steps': 34564, 'loss/train': 0.6604659557342529}
+02/25/2022 22:20:00 - INFO - codeparrot_training - Step 34565: {'lr': 0.00011708105723515955, 'samples': 17697792, 'steps': 34565, 'loss/train': 1.2514607906341553}
+02/25/2022 22:20:06 - INFO - codeparrot_training - Step 34566: {'lr': 0.0001170671993660058, 'samples': 17698304, 'steps': 34566, 'loss/train': 2.00887393951416}
+02/25/2022 22:20:09 - INFO - codeparrot_training - Step 34567: {'lr': 0.00011705334206629396, 'samples': 17698816, 'steps': 34567, 'loss/train': 2.506592035293579}
+02/25/2022 22:20:15 - INFO - codeparrot_training - Step 34568: {'lr': 0.00011703948533608339, 'samples': 17699328, 'steps': 34568, 'loss/train': 1.84834623336792}
+02/25/2022 22:20:19 - INFO - codeparrot_training - Step 34569: {'lr': 0.00011702562917543332, 'samples': 17699840, 'steps': 34569, 'loss/train': 1.971256971359253}
+02/25/2022 22:20:24 - INFO - codeparrot_training - Step 34570: {'lr': 0.00011701177358440334, 'samples': 17700352, 'steps': 34570, 'loss/train': 1.0615979433059692}
+02/25/2022 22:20:28 - INFO - codeparrot_training - Step 34571: {'lr': 0.00011699791856305272, 'samples': 17700864, 'steps': 34571, 'loss/train': 2.1620712280273438}
+02/25/2022 22:20:33 - INFO - codeparrot_training - Step 34572: {'lr': 0.00011698406411144074, 'samples': 17701376, 'steps': 34572, 'loss/train': 2.0336813926696777}
+02/25/2022 22:20:37 - INFO - codeparrot_training - Step 34573: {'lr': 0.00011697021022962673, 'samples': 17701888, 'steps': 34573, 'loss/train': 1.7189326286315918}
+02/25/2022 22:20:42 - INFO - codeparrot_training - Step 34574: {'lr': 0.00011695635691767018, 'samples': 17702400, 'steps': 34574, 'loss/train': 1.8887455463409424}
+02/25/2022 22:20:46 - INFO - codeparrot_training - Step 34575: {'lr': 0.00011694250417563036, 'samples': 17702912, 'steps': 34575, 'loss/train': 2.6055195331573486}
+02/25/2022 22:20:51 - INFO - codeparrot_training - Step 34576: {'lr': 0.00011692865200356659, 'samples': 17703424, 'steps': 34576, 'loss/train': 2.192671298980713}
+02/25/2022 22:20:55 - INFO - codeparrot_training - Step 34577: {'lr': 0.00011691480040153812, 'samples': 17703936, 'steps': 34577, 'loss/train': 1.7293568849563599}
+02/25/2022 22:21:01 - INFO - codeparrot_training - Step 34578: {'lr': 0.00011690094936960449, 'samples': 17704448, 'steps': 34578, 'loss/train': 1.6918243169784546}
+02/25/2022 22:21:04 - INFO - codeparrot_training - Step 34579: {'lr': 0.00011688709890782484, 'samples': 17704960, 'steps': 34579, 'loss/train': 1.848863959312439}
+02/25/2022 22:21:09 - INFO - codeparrot_training - Step 34580: {'lr': 0.00011687324901625879, 'samples': 17705472, 'steps': 34580, 'loss/train': 1.8401762247085571}
+02/25/2022 22:21:13 - INFO - codeparrot_training - Step 34581: {'lr': 0.00011685939969496526, 'samples': 17705984, 'steps': 34581, 'loss/train': 2.2630977630615234}
+02/25/2022 22:21:20 - INFO - codeparrot_training - Step 34582: {'lr': 0.00011684555094400388, 'samples': 17706496, 'steps': 34582, 'loss/train': 2.4304563999176025}
+02/25/2022 22:21:23 - INFO - codeparrot_training - Step 34583: {'lr': 0.0001168317027634338, 'samples': 17707008, 'steps': 34583, 'loss/train': 1.4651545286178589}
+02/25/2022 22:21:29 - INFO - codeparrot_training - Step 34584: {'lr': 0.00011681785515331458, 'samples': 17707520, 'steps': 34584, 'loss/train': 1.0978509187698364}
+02/25/2022 22:21:32 - INFO - codeparrot_training - Step 34585: {'lr': 0.0001168040081137052, 'samples': 17708032, 'steps': 34585, 'loss/train': 1.6135120391845703}
+02/25/2022 22:21:38 - INFO - codeparrot_training - Step 34586: {'lr': 0.00011679016164466528, 'samples': 17708544, 'steps': 34586, 'loss/train': 0.91129469871521}
+02/25/2022 22:21:41 - INFO - codeparrot_training - Step 34587: {'lr': 0.00011677631574625389, 'samples': 17709056, 'steps': 34587, 'loss/train': 1.7425577640533447}
+02/25/2022 22:21:47 - INFO - codeparrot_training - Step 34588: {'lr': 0.00011676247041853063, 'samples': 17709568, 'steps': 34588, 'loss/train': 1.8360234498977661}
+02/25/2022 22:21:51 - INFO - codeparrot_training - Step 34589: {'lr': 0.00011674862566155448, 'samples': 17710080, 'steps': 34589, 'loss/train': 1.300140142440796}
+02/25/2022 22:21:56 - INFO - codeparrot_training - Step 34590: {'lr': 0.00011673478147538497, 'samples': 17710592, 'steps': 34590, 'loss/train': 1.4613616466522217}
+02/25/2022 22:22:00 - INFO - codeparrot_training - Step 34591: {'lr': 0.00011672093786008128, 'samples': 17711104, 'steps': 34591, 'loss/train': 0.02263191156089306}
+02/25/2022 22:22:05 - INFO - codeparrot_training - Step 34592: {'lr': 0.00011670709481570285, 'samples': 17711616, 'steps': 34592, 'loss/train': 2.3177847862243652}
+02/25/2022 22:22:09 - INFO - codeparrot_training - Step 34593: {'lr': 0.0001166932523423089, 'samples': 17712128, 'steps': 34593, 'loss/train': 2.3077192306518555}
+02/25/2022 22:22:15 - INFO - codeparrot_training - Step 34594: {'lr': 0.00011667941043995873, 'samples': 17712640, 'steps': 34594, 'loss/train': 0.9222477078437805}
+02/25/2022 22:22:19 - INFO - codeparrot_training - Step 34595: {'lr': 0.00011666556910871151, 'samples': 17713152, 'steps': 34595, 'loss/train': 2.0848236083984375}
+02/25/2022 22:22:24 - INFO - codeparrot_training - Step 34596: {'lr': 0.00011665172834862678, 'samples': 17713664, 'steps': 34596, 'loss/train': 1.283647060394287}
+02/25/2022 22:22:28 - INFO - codeparrot_training - Step 34597: {'lr': 0.0001166378881597637, 'samples': 17714176, 'steps': 34597, 'loss/train': 1.5610575675964355}
+02/25/2022 22:22:33 - INFO - codeparrot_training - Step 34598: {'lr': 0.00011662404854218156, 'samples': 17714688, 'steps': 34598, 'loss/train': 1.3010587692260742}
+02/25/2022 22:22:37 - INFO - codeparrot_training - Step 34599: {'lr': 0.00011661020949593954, 'samples': 17715200, 'steps': 34599, 'loss/train': 1.7915834188461304}
+02/25/2022 22:22:42 - INFO - codeparrot_training - Step 34600: {'lr': 0.00011659637102109713, 'samples': 17715712, 'steps': 34600, 'loss/train': 2.0628342628479004}
+02/25/2022 22:22:46 - INFO - codeparrot_training - Step 34601: {'lr': 0.0001165825331177135, 'samples': 17716224, 'steps': 34601, 'loss/train': 0.058083467185497284}
+02/25/2022 22:22:52 - INFO - codeparrot_training - Step 34602: {'lr': 0.00011656869578584792, 'samples': 17716736, 'steps': 34602, 'loss/train': 2.5492336750030518}
+02/25/2022 22:22:55 - INFO - codeparrot_training - Step 34603: {'lr': 0.0001165548590255596, 'samples': 17717248, 'steps': 34603, 'loss/train': 1.669998288154602}
+02/25/2022 22:23:01 - INFO - codeparrot_training - Step 34604: {'lr': 0.00011654102283690798, 'samples': 17717760, 'steps': 34604, 'loss/train': 1.1803606748580933}
+02/25/2022 22:23:04 - INFO - codeparrot_training - Step 34605: {'lr': 0.00011652718721995226, 'samples': 17718272, 'steps': 34605, 'loss/train': 2.2157907485961914}
+02/25/2022 22:23:10 - INFO - codeparrot_training - Step 34606: {'lr': 0.00011651335217475167, 'samples': 17718784, 'steps': 34606, 'loss/train': 3.046508550643921}
+02/25/2022 22:23:13 - INFO - codeparrot_training - Step 34607: {'lr': 0.00011649951770136538, 'samples': 17719296, 'steps': 34607, 'loss/train': 2.4108340740203857}
+02/25/2022 22:23:19 - INFO - codeparrot_training - Step 34608: {'lr': 0.00011648568379985289, 'samples': 17719808, 'steps': 34608, 'loss/train': 1.2670422792434692}
+02/25/2022 22:23:22 - INFO - codeparrot_training - Step 34609: {'lr': 0.00011647185047027334, 'samples': 17720320, 'steps': 34609, 'loss/train': 1.6605414152145386}
+02/25/2022 22:23:28 - INFO - codeparrot_training - Step 34610: {'lr': 0.00011645801771268596, 'samples': 17720832, 'steps': 34610, 'loss/train': 0.9107034206390381}
+02/25/2022 22:23:31 - INFO - codeparrot_training - Step 34611: {'lr': 0.00011644418552714994, 'samples': 17721344, 'steps': 34611, 'loss/train': 1.0117378234863281}
+02/25/2022 22:23:37 - INFO - codeparrot_training - Step 34612: {'lr': 0.0001164303539137247, 'samples': 17721856, 'steps': 34612, 'loss/train': 2.227492332458496}
+02/25/2022 22:23:41 - INFO - codeparrot_training - Step 34613: {'lr': 0.00011641652287246942, 'samples': 17722368, 'steps': 34613, 'loss/train': 1.8931323289871216}
+02/25/2022 22:23:46 - INFO - codeparrot_training - Step 34614: {'lr': 0.00011640269240344326, 'samples': 17722880, 'steps': 34614, 'loss/train': 1.9377555847167969}
+02/25/2022 22:23:50 - INFO - codeparrot_training - Step 34615: {'lr': 0.00011638886250670572, 'samples': 17723392, 'steps': 34615, 'loss/train': 1.1202654838562012}
+02/25/2022 22:23:55 - INFO - codeparrot_training - Step 34616: {'lr': 0.00011637503318231568, 'samples': 17723904, 'steps': 34616, 'loss/train': 3.031440019607544}
+02/25/2022 22:23:59 - INFO - codeparrot_training - Step 34617: {'lr': 0.00011636120443033265, 'samples': 17724416, 'steps': 34617, 'loss/train': 0.8041306734085083}
+02/25/2022 22:24:04 - INFO - codeparrot_training - Step 34618: {'lr': 0.0001163473762508157, 'samples': 17724928, 'steps': 34618, 'loss/train': 2.0794122219085693}
+02/25/2022 22:24:08 - INFO - codeparrot_training - Step 34619: {'lr': 0.00011633354864382437, 'samples': 17725440, 'steps': 34619, 'loss/train': 1.8044875860214233}
+02/25/2022 22:24:13 - INFO - codeparrot_training - Step 34620: {'lr': 0.00011631972160941745, 'samples': 17725952, 'steps': 34620, 'loss/train': 0.1137242391705513}
+02/25/2022 22:24:17 - INFO - codeparrot_training - Step 34621: {'lr': 0.00011630589514765449, 'samples': 17726464, 'steps': 34621, 'loss/train': 1.539244294166565}
+02/25/2022 22:24:22 - INFO - codeparrot_training - Step 34622: {'lr': 0.00011629206925859454, 'samples': 17726976, 'steps': 34622, 'loss/train': 1.1114946603775024}
+02/25/2022 22:24:26 - INFO - codeparrot_training - Step 34623: {'lr': 0.00011627824394229708, 'samples': 17727488, 'steps': 34623, 'loss/train': 2.2846133708953857}
+02/25/2022 22:24:31 - INFO - codeparrot_training - Step 34624: {'lr': 0.000116264419198821, 'samples': 17728000, 'steps': 34624, 'loss/train': 1.5699506998062134}
+02/25/2022 22:24:35 - INFO - codeparrot_training - Step 34625: {'lr': 0.00011625059502822574, 'samples': 17728512, 'steps': 34625, 'loss/train': 0.9542575478553772}
+02/25/2022 22:24:40 - INFO - codeparrot_training - Step 34626: {'lr': 0.0001162367714305704, 'samples': 17729024, 'steps': 34626, 'loss/train': 1.475652813911438}
+02/25/2022 22:24:43 - INFO - codeparrot_training - Step 34627: {'lr': 0.00011622294840591432, 'samples': 17729536, 'steps': 34627, 'loss/train': 1.5890648365020752}
+02/25/2022 22:24:50 - INFO - codeparrot_training - Step 34628: {'lr': 0.00011620912595431668, 'samples': 17730048, 'steps': 34628, 'loss/train': 1.347261905670166}
+02/25/2022 22:24:53 - INFO - codeparrot_training - Step 34629: {'lr': 0.00011619530407583665, 'samples': 17730560, 'steps': 34629, 'loss/train': 0.06105469539761543}
+02/25/2022 22:24:59 - INFO - codeparrot_training - Step 34630: {'lr': 0.00011618148277053334, 'samples': 17731072, 'steps': 34630, 'loss/train': 1.517052412033081}
+02/25/2022 22:25:02 - INFO - codeparrot_training - Step 34631: {'lr': 0.00011616766203846618, 'samples': 17731584, 'steps': 34631, 'loss/train': 1.7262964248657227}
+02/25/2022 22:25:07 - INFO - codeparrot_training - Step 34632: {'lr': 0.00011615384187969424, 'samples': 17732096, 'steps': 34632, 'loss/train': 1.9341062307357788}
+02/25/2022 22:25:11 - INFO - codeparrot_training - Step 34633: {'lr': 0.00011614002229427676, 'samples': 17732608, 'steps': 34633, 'loss/train': 1.9326890707015991}
+02/25/2022 22:25:16 - INFO - codeparrot_training - Step 34634: {'lr': 0.00011612620328227283, 'samples': 17733120, 'steps': 34634, 'loss/train': 0.2123493254184723}
+02/25/2022 22:25:22 - INFO - codeparrot_training - Step 34635: {'lr': 0.00011611238484374181, 'samples': 17733632, 'steps': 34635, 'loss/train': 2.0048444271087646}
+02/25/2022 22:25:26 - INFO - codeparrot_training - Step 34636: {'lr': 0.00011609856697874283, 'samples': 17734144, 'steps': 34636, 'loss/train': 1.5604960918426514}
+02/25/2022 22:25:29 - INFO - codeparrot_training - Step 34637: {'lr': 0.00011608474968733507, 'samples': 17734656, 'steps': 34637, 'loss/train': 1.8149012327194214}
+02/25/2022 22:25:35 - INFO - codeparrot_training - Step 34638: {'lr': 0.00011607093296957763, 'samples': 17735168, 'steps': 34638, 'loss/train': 2.2279791831970215}
+02/25/2022 22:25:38 - INFO - codeparrot_training - Step 34639: {'lr': 0.00011605711682552986, 'samples': 17735680, 'steps': 34639, 'loss/train': 2.379338502883911}
+02/25/2022 22:25:45 - INFO - codeparrot_training - Step 34640: {'lr': 0.00011604330125525078, 'samples': 17736192, 'steps': 34640, 'loss/train': 1.7767943143844604}
+02/25/2022 22:25:48 - INFO - codeparrot_training - Step 34641: {'lr': 0.00011602948625879986, 'samples': 17736704, 'steps': 34641, 'loss/train': 1.3086358308792114}
+02/25/2022 22:25:53 - INFO - codeparrot_training - Step 34642: {'lr': 0.0001160156718362359, 'samples': 17737216, 'steps': 34642, 'loss/train': 1.9066476821899414}
+02/25/2022 22:25:57 - INFO - codeparrot_training - Step 34643: {'lr': 0.00011600185798761833, 'samples': 17737728, 'steps': 34643, 'loss/train': 1.8030650615692139}
+02/25/2022 22:26:03 - INFO - codeparrot_training - Step 34644: {'lr': 0.00011598804471300617, 'samples': 17738240, 'steps': 34644, 'loss/train': 2.8161821365356445}
+02/25/2022 22:26:06 - INFO - codeparrot_training - Step 34645: {'lr': 0.00011597423201245888, 'samples': 17738752, 'steps': 34645, 'loss/train': 2.0014748573303223}
+02/25/2022 22:26:11 - INFO - codeparrot_training - Step 34646: {'lr': 0.0001159604198860352, 'samples': 17739264, 'steps': 34646, 'loss/train': 1.6776434183120728}
+02/25/2022 22:26:15 - INFO - codeparrot_training - Step 34647: {'lr': 0.00011594660833379464, 'samples': 17739776, 'steps': 34647, 'loss/train': 1.5947719812393188}
+02/25/2022 22:26:20 - INFO - codeparrot_training - Step 34648: {'lr': 0.0001159327973557962, 'samples': 17740288, 'steps': 34648, 'loss/train': 1.512926697731018}
+02/25/2022 22:26:24 - INFO - codeparrot_training - Step 34649: {'lr': 0.00011591898695209913, 'samples': 17740800, 'steps': 34649, 'loss/train': 2.541478395462036}
+02/25/2022 22:26:30 - INFO - codeparrot_training - Step 34650: {'lr': 0.00011590517712276244, 'samples': 17741312, 'steps': 34650, 'loss/train': 2.7670252323150635}
+02/25/2022 22:26:33 - INFO - codeparrot_training - Step 34651: {'lr': 0.00011589136786784548, 'samples': 17741824, 'steps': 34651, 'loss/train': 1.616425633430481}
+02/25/2022 22:26:39 - INFO - codeparrot_training - Step 34652: {'lr': 0.0001158775591874073, 'samples': 17742336, 'steps': 34652, 'loss/train': 1.431565523147583}
+02/25/2022 22:26:42 - INFO - codeparrot_training - Step 34653: {'lr': 0.00011586375108150702, 'samples': 17742848, 'steps': 34653, 'loss/train': 0.9353484511375427}
+02/25/2022 22:26:48 - INFO - codeparrot_training - Step 34654: {'lr': 0.00011584994355020389, 'samples': 17743360, 'steps': 34654, 'loss/train': 1.4318761825561523}
+02/25/2022 22:26:54 - INFO - codeparrot_training - Step 34655: {'lr': 0.00011583613659355702, 'samples': 17743872, 'steps': 34655, 'loss/train': 2.1980092525482178}
+02/25/2022 22:26:57 - INFO - codeparrot_training - Step 34656: {'lr': 0.00011582233021162553, 'samples': 17744384, 'steps': 34656, 'loss/train': 2.157557487487793}
+02/25/2022 22:27:03 - INFO - codeparrot_training - Step 34657: {'lr': 0.00011580852440446848, 'samples': 17744896, 'steps': 34657, 'loss/train': 1.8558359146118164}
+02/25/2022 22:27:06 - INFO - codeparrot_training - Step 34658: {'lr': 0.00011579471917214519, 'samples': 17745408, 'steps': 34658, 'loss/train': 0.5897127389907837}
+02/25/2022 22:27:12 - INFO - codeparrot_training - Step 34659: {'lr': 0.00011578091451471471, 'samples': 17745920, 'steps': 34659, 'loss/train': 1.6939104795455933}
+02/25/2022 22:27:15 - INFO - codeparrot_training - Step 34660: {'lr': 0.00011576711043223617, 'samples': 17746432, 'steps': 34660, 'loss/train': 1.783022403717041}
+02/25/2022 22:27:21 - INFO - codeparrot_training - Step 34661: {'lr': 0.00011575330692476863, 'samples': 17746944, 'steps': 34661, 'loss/train': 0.041671521961688995}
+02/25/2022 22:27:24 - INFO - codeparrot_training - Step 34662: {'lr': 0.00011573950399237138, 'samples': 17747456, 'steps': 34662, 'loss/train': 1.9889122247695923}
+02/25/2022 22:27:30 - INFO - codeparrot_training - Step 34663: {'lr': 0.00011572570163510346, 'samples': 17747968, 'steps': 34663, 'loss/train': 2.0284688472747803}
+02/25/2022 22:27:33 - INFO - codeparrot_training - Step 34664: {'lr': 0.000115711899853024, 'samples': 17748480, 'steps': 34664, 'loss/train': 1.3339725732803345}
+02/25/2022 22:27:39 - INFO - codeparrot_training - Step 34665: {'lr': 0.00011569809864619202, 'samples': 17748992, 'steps': 34665, 'loss/train': 1.8196130990982056}
+02/25/2022 22:27:43 - INFO - codeparrot_training - Step 34666: {'lr': 0.00011568429801466684, 'samples': 17749504, 'steps': 34666, 'loss/train': 1.3567323684692383}
+02/25/2022 22:27:48 - INFO - codeparrot_training - Step 34667: {'lr': 0.00011567049795850749, 'samples': 17750016, 'steps': 34667, 'loss/train': 2.4590413570404053}
+02/25/2022 22:27:52 - INFO - codeparrot_training - Step 34668: {'lr': 0.00011565669847777305, 'samples': 17750528, 'steps': 34668, 'loss/train': 2.0494492053985596}
+02/25/2022 22:27:57 - INFO - codeparrot_training - Step 34669: {'lr': 0.00011564289957252255, 'samples': 17751040, 'steps': 34669, 'loss/train': 0.3943789005279541}
+02/25/2022 22:28:01 - INFO - codeparrot_training - Step 34670: {'lr': 0.0001156291012428153, 'samples': 17751552, 'steps': 34670, 'loss/train': 1.8019856214523315}
+02/25/2022 22:28:06 - INFO - codeparrot_training - Step 34671: {'lr': 0.00011561530348871036, 'samples': 17752064, 'steps': 34671, 'loss/train': 1.2965737581253052}
+02/25/2022 22:28:10 - INFO - codeparrot_training - Step 34672: {'lr': 0.00011560150631026675, 'samples': 17752576, 'steps': 34672, 'loss/train': 1.956367015838623}
+02/25/2022 22:28:15 - INFO - codeparrot_training - Step 34673: {'lr': 0.00011558770970754351, 'samples': 17753088, 'steps': 34673, 'loss/train': 1.4930142164230347}
+02/25/2022 22:28:19 - INFO - codeparrot_training - Step 34674: {'lr': 0.00011557391368059997, 'samples': 17753600, 'steps': 34674, 'loss/train': 1.1915210485458374}
+02/25/2022 22:28:25 - INFO - codeparrot_training - Step 34675: {'lr': 0.00011556011822949497, 'samples': 17754112, 'steps': 34675, 'loss/train': 0.9868058562278748}
+02/25/2022 22:28:28 - INFO - codeparrot_training - Step 34676: {'lr': 0.00011554632335428795, 'samples': 17754624, 'steps': 34676, 'loss/train': 2.591519832611084}
+02/25/2022 22:28:34 - INFO - codeparrot_training - Step 34677: {'lr': 0.00011553252905503755, 'samples': 17755136, 'steps': 34677, 'loss/train': 1.2586069107055664}
+02/25/2022 22:28:37 - INFO - codeparrot_training - Step 34678: {'lr': 0.00011551873533180323, 'samples': 17755648, 'steps': 34678, 'loss/train': 1.4948464632034302}
+02/25/2022 22:28:43 - INFO - codeparrot_training - Step 34679: {'lr': 0.00011550494218464386, 'samples': 17756160, 'steps': 34679, 'loss/train': 2.0157244205474854}
+02/25/2022 22:28:46 - INFO - codeparrot_training - Step 34680: {'lr': 0.00011549114961361879, 'samples': 17756672, 'steps': 34680, 'loss/train': 1.7757574319839478}
+02/25/2022 22:28:52 - INFO - codeparrot_training - Step 34681: {'lr': 0.00011547735761878672, 'samples': 17757184, 'steps': 34681, 'loss/train': 2.003019094467163}
+02/25/2022 22:28:56 - INFO - codeparrot_training - Step 34682: {'lr': 0.00011546356620020704, 'samples': 17757696, 'steps': 34682, 'loss/train': 1.2299540042877197}
+02/25/2022 22:29:01 - INFO - codeparrot_training - Step 34683: {'lr': 0.00011544977535793863, 'samples': 17758208, 'steps': 34683, 'loss/train': 1.5901049375534058}
+02/25/2022 22:29:05 - INFO - codeparrot_training - Step 34684: {'lr': 0.00011543598509204084, 'samples': 17758720, 'steps': 34684, 'loss/train': 2.0335536003112793}
+02/25/2022 22:29:11 - INFO - codeparrot_training - Step 34685: {'lr': 0.00011542219540257237, 'samples': 17759232, 'steps': 34685, 'loss/train': 1.7262800931930542}
+02/25/2022 22:29:14 - INFO - codeparrot_training - Step 34686: {'lr': 0.00011540840628959259, 'samples': 17759744, 'steps': 34686, 'loss/train': 1.1418932676315308}
+02/25/2022 22:29:20 - INFO - codeparrot_training - Step 34687: {'lr': 0.00011539461775316035, 'samples': 17760256, 'steps': 34687, 'loss/train': 1.444786787033081}
+02/25/2022 22:29:23 - INFO - codeparrot_training - Step 34688: {'lr': 0.00011538082979333495, 'samples': 17760768, 'steps': 34688, 'loss/train': 1.8199483156204224}
+02/25/2022 22:29:29 - INFO - codeparrot_training - Step 34689: {'lr': 0.00011536704241017529, 'samples': 17761280, 'steps': 34689, 'loss/train': 1.733020544052124}
+02/25/2022 22:29:32 - INFO - codeparrot_training - Step 34690: {'lr': 0.00011535325560374049, 'samples': 17761792, 'steps': 34690, 'loss/train': 1.4590998888015747}
+02/25/2022 22:29:38 - INFO - codeparrot_training - Step 34691: {'lr': 0.00011533946937408948, 'samples': 17762304, 'steps': 34691, 'loss/train': 1.5719199180603027}
+02/25/2022 22:29:42 - INFO - codeparrot_training - Step 34692: {'lr': 0.00011532568372128155, 'samples': 17762816, 'steps': 34692, 'loss/train': 1.649390459060669}
+02/25/2022 22:29:47 - INFO - codeparrot_training - Step 34693: {'lr': 0.00011531189864537558, 'samples': 17763328, 'steps': 34693, 'loss/train': 2.069047451019287}
+02/25/2022 22:29:50 - INFO - codeparrot_training - Step 34694: {'lr': 0.0001152981141464307, 'samples': 17763840, 'steps': 34694, 'loss/train': 1.9279448986053467}
+02/25/2022 22:29:57 - INFO - codeparrot_training - Step 34695: {'lr': 0.0001152843302245059, 'samples': 17764352, 'steps': 34695, 'loss/train': 1.2418310642242432}
+02/25/2022 22:30:00 - INFO - codeparrot_training - Step 34696: {'lr': 0.00011527054687966021, 'samples': 17764864, 'steps': 34696, 'loss/train': 0.9884203672409058}
+02/25/2022 22:30:05 - INFO - codeparrot_training - Step 34697: {'lr': 0.00011525676411195277, 'samples': 17765376, 'steps': 34697, 'loss/train': 1.4402081966400146}
+02/25/2022 22:30:09 - INFO - codeparrot_training - Step 34698: {'lr': 0.00011524298192144256, 'samples': 17765888, 'steps': 34698, 'loss/train': 1.8921897411346436}
+02/25/2022 22:30:14 - INFO - codeparrot_training - Step 34699: {'lr': 0.00011522920030818865, 'samples': 17766400, 'steps': 34699, 'loss/train': 2.6496877670288086}
+02/25/2022 22:30:18 - INFO - codeparrot_training - Step 34700: {'lr': 0.00011521541927224994, 'samples': 17766912, 'steps': 34700, 'loss/train': 1.0045610666275024}
+02/25/2022 22:30:23 - INFO - codeparrot_training - Step 34701: {'lr': 0.00011520163881368568, 'samples': 17767424, 'steps': 34701, 'loss/train': 2.375739812850952}
+02/25/2022 22:30:27 - INFO - codeparrot_training - Step 34702: {'lr': 0.0001151878589325548, 'samples': 17767936, 'steps': 34702, 'loss/train': 2.2931594848632812}
+02/25/2022 22:30:32 - INFO - codeparrot_training - Step 34703: {'lr': 0.0001151740796289163, 'samples': 17768448, 'steps': 34703, 'loss/train': 2.1976895332336426}
+02/25/2022 22:30:36 - INFO - codeparrot_training - Step 34704: {'lr': 0.00011516030090282914, 'samples': 17768960, 'steps': 34704, 'loss/train': 3.053875684738159}
+02/25/2022 22:30:41 - INFO - codeparrot_training - Step 34705: {'lr': 0.00011514652275435255, 'samples': 17769472, 'steps': 34705, 'loss/train': 1.0335137844085693}
+02/25/2022 22:30:45 - INFO - codeparrot_training - Step 34706: {'lr': 0.00011513274518354544, 'samples': 17769984, 'steps': 34706, 'loss/train': 2.4896655082702637}
+02/25/2022 22:30:50 - INFO - codeparrot_training - Step 34707: {'lr': 0.0001151189681904668, 'samples': 17770496, 'steps': 34707, 'loss/train': 1.9550503492355347}
+02/25/2022 22:30:54 - INFO - codeparrot_training - Step 34708: {'lr': 0.00011510519177517559, 'samples': 17771008, 'steps': 34708, 'loss/train': 0.1877146065235138}
+02/25/2022 22:30:59 - INFO - codeparrot_training - Step 34709: {'lr': 0.00011509141593773098, 'samples': 17771520, 'steps': 34709, 'loss/train': 1.8265786170959473}
+02/25/2022 22:31:03 - INFO - codeparrot_training - Step 34710: {'lr': 0.00011507764067819185, 'samples': 17772032, 'steps': 34710, 'loss/train': 1.7760233879089355}
+02/25/2022 22:31:10 - INFO - codeparrot_training - Step 34711: {'lr': 0.00011506386599661741, 'samples': 17772544, 'steps': 34711, 'loss/train': 1.6839982271194458}
+02/25/2022 22:31:13 - INFO - codeparrot_training - Step 34712: {'lr': 0.00011505009189306636, 'samples': 17773056, 'steps': 34712, 'loss/train': 1.2640389204025269}
+02/25/2022 22:31:19 - INFO - codeparrot_training - Step 34713: {'lr': 0.00011503631836759795, 'samples': 17773568, 'steps': 34713, 'loss/train': 1.7982176542282104}
+02/25/2022 22:31:22 - INFO - codeparrot_training - Step 34714: {'lr': 0.00011502254542027102, 'samples': 17774080, 'steps': 34714, 'loss/train': 1.723174810409546}
+02/25/2022 22:31:28 - INFO - codeparrot_training - Step 34715: {'lr': 0.00011500877305114479, 'samples': 17774592, 'steps': 34715, 'loss/train': 1.2586579322814941}
+02/25/2022 22:31:32 - INFO - codeparrot_training - Step 34716: {'lr': 0.00011499500126027798, 'samples': 17775104, 'steps': 34716, 'loss/train': 0.7825419902801514}
+02/25/2022 22:31:37 - INFO - codeparrot_training - Step 34717: {'lr': 0.00011498123004772976, 'samples': 17775616, 'steps': 34717, 'loss/train': 1.3392101526260376}
+02/25/2022 22:31:41 - INFO - codeparrot_training - Step 34718: {'lr': 0.00011496745941355902, 'samples': 17776128, 'steps': 34718, 'loss/train': 2.122418165206909}
+02/25/2022 22:31:46 - INFO - codeparrot_training - Step 34719: {'lr': 0.00011495368935782497, 'samples': 17776640, 'steps': 34719, 'loss/train': 2.154677391052246}
+02/25/2022 22:31:50 - INFO - codeparrot_training - Step 34720: {'lr': 0.00011493991988058625, 'samples': 17777152, 'steps': 34720, 'loss/train': 1.484230399131775}
+02/25/2022 22:31:56 - INFO - codeparrot_training - Step 34721: {'lr': 0.00011492615098190213, 'samples': 17777664, 'steps': 34721, 'loss/train': 2.2542247772216797}
+02/25/2022 22:32:00 - INFO - codeparrot_training - Step 34722: {'lr': 0.00011491238266183138, 'samples': 17778176, 'steps': 34722, 'loss/train': 2.3974907398223877}
+02/25/2022 22:32:05 - INFO - codeparrot_training - Step 34723: {'lr': 0.00011489861492043319, 'samples': 17778688, 'steps': 34723, 'loss/train': 1.3617750406265259}
+02/25/2022 22:32:09 - INFO - codeparrot_training - Step 34724: {'lr': 0.00011488484775776645, 'samples': 17779200, 'steps': 34724, 'loss/train': 2.085836887359619}
+02/25/2022 22:32:14 - INFO - codeparrot_training - Step 34725: {'lr': 0.00011487108117389008, 'samples': 17779712, 'steps': 34725, 'loss/train': 1.6692780256271362}
+02/25/2022 22:32:18 - INFO - codeparrot_training - Step 34726: {'lr': 0.00011485731516886303, 'samples': 17780224, 'steps': 34726, 'loss/train': 0.5327391028404236}
+02/25/2022 22:32:23 - INFO - codeparrot_training - Step 34727: {'lr': 0.00011484354974274441, 'samples': 17780736, 'steps': 34727, 'loss/train': 1.2682796716690063}
+02/25/2022 22:32:27 - INFO - codeparrot_training - Step 34728: {'lr': 0.00011482978489559312, 'samples': 17781248, 'steps': 34728, 'loss/train': 2.6547932624816895}
+02/25/2022 22:32:32 - INFO - codeparrot_training - Step 34729: {'lr': 0.0001148160206274681, 'samples': 17781760, 'steps': 34729, 'loss/train': 1.0286450386047363}
+02/25/2022 22:32:36 - INFO - codeparrot_training - Step 34730: {'lr': 0.00011480225693842822, 'samples': 17782272, 'steps': 34730, 'loss/train': 1.509169101715088}
+02/25/2022 22:32:42 - INFO - codeparrot_training - Step 34731: {'lr': 0.00011478849382853263, 'samples': 17782784, 'steps': 34731, 'loss/train': 1.5795807838439941}
+02/25/2022 22:32:45 - INFO - codeparrot_training - Step 34732: {'lr': 0.00011477473129784022, 'samples': 17783296, 'steps': 34732, 'loss/train': 1.7360645532608032}
+02/25/2022 22:32:51 - INFO - codeparrot_training - Step 34733: {'lr': 0.00011476096934640992, 'samples': 17783808, 'steps': 34733, 'loss/train': 1.1455860137939453}
+02/25/2022 22:32:54 - INFO - codeparrot_training - Step 34734: {'lr': 0.00011474720797430057, 'samples': 17784320, 'steps': 34734, 'loss/train': 1.6169661283493042}
+02/25/2022 22:33:00 - INFO - codeparrot_training - Step 34735: {'lr': 0.00011473344718157133, 'samples': 17784832, 'steps': 34735, 'loss/train': 1.66375732421875}
+02/25/2022 22:33:03 - INFO - codeparrot_training - Step 34736: {'lr': 0.00011471968696828106, 'samples': 17785344, 'steps': 34736, 'loss/train': 3.9603748321533203}
+02/25/2022 22:33:09 - INFO - codeparrot_training - Step 34737: {'lr': 0.0001147059273344887, 'samples': 17785856, 'steps': 34737, 'loss/train': 1.7098735570907593}
+02/25/2022 22:33:12 - INFO - codeparrot_training - Step 34738: {'lr': 0.00011469216828025309, 'samples': 17786368, 'steps': 34738, 'loss/train': 0.5596854090690613}
+02/25/2022 22:33:18 - INFO - codeparrot_training - Step 34739: {'lr': 0.00011467840980563335, 'samples': 17786880, 'steps': 34739, 'loss/train': 2.477786064147949}
+02/25/2022 22:33:21 - INFO - codeparrot_training - Step 34740: {'lr': 0.00011466465191068834, 'samples': 17787392, 'steps': 34740, 'loss/train': 0.8669219017028809}
+02/25/2022 22:33:28 - INFO - codeparrot_training - Step 34741: {'lr': 0.00011465089459547698, 'samples': 17787904, 'steps': 34741, 'loss/train': 1.8605616092681885}
+02/25/2022 22:33:31 - INFO - codeparrot_training - Step 34742: {'lr': 0.00011463713786005813, 'samples': 17788416, 'steps': 34742, 'loss/train': 1.132358193397522}
+02/25/2022 22:33:37 - INFO - codeparrot_training - Step 34743: {'lr': 0.00011462338170449088, 'samples': 17788928, 'steps': 34743, 'loss/train': 1.6693097352981567}
+02/25/2022 22:33:40 - INFO - codeparrot_training - Step 34744: {'lr': 0.00011460962612883408, 'samples': 17789440, 'steps': 34744, 'loss/train': 2.387767791748047}
+02/25/2022 22:33:46 - INFO - codeparrot_training - Step 34745: {'lr': 0.00011459587113314665, 'samples': 17789952, 'steps': 34745, 'loss/train': 1.6871999502182007}
+02/25/2022 22:33:49 - INFO - codeparrot_training - Step 34746: {'lr': 0.00011458211671748753, 'samples': 17790464, 'steps': 34746, 'loss/train': 2.442793607711792}
+02/25/2022 22:33:55 - INFO - codeparrot_training - Step 34747: {'lr': 0.0001145683628819155, 'samples': 17790976, 'steps': 34747, 'loss/train': 1.7778996229171753}
+02/25/2022 22:33:58 - INFO - codeparrot_training - Step 34748: {'lr': 0.0001145546096264897, 'samples': 17791488, 'steps': 34748, 'loss/train': 0.8230723142623901}
+02/25/2022 22:34:04 - INFO - codeparrot_training - Step 34749: {'lr': 0.00011454085695126887, 'samples': 17792000, 'steps': 34749, 'loss/train': 2.565108299255371}
+02/25/2022 22:34:07 - INFO - codeparrot_training - Step 34750: {'lr': 0.00011452710485631215, 'samples': 17792512, 'steps': 34750, 'loss/train': 1.8774287700653076}
+02/25/2022 22:34:13 - INFO - codeparrot_training - Step 34751: {'lr': 0.00011451335334167815, 'samples': 17793024, 'steps': 34751, 'loss/train': 0.710401713848114}
+02/25/2022 22:34:16 - INFO - codeparrot_training - Step 34752: {'lr': 0.00011449960240742596, 'samples': 17793536, 'steps': 34752, 'loss/train': 1.6455191373825073}
+02/25/2022 22:34:22 - INFO - codeparrot_training - Step 34753: {'lr': 0.00011448585205361439, 'samples': 17794048, 'steps': 34753, 'loss/train': 0.8933877348899841}
+02/25/2022 22:34:25 - INFO - codeparrot_training - Step 34754: {'lr': 0.00011447210228030258, 'samples': 17794560, 'steps': 34754, 'loss/train': 2.0767128467559814}
+02/25/2022 22:34:31 - INFO - codeparrot_training - Step 34755: {'lr': 0.00011445835308754906, 'samples': 17795072, 'steps': 34755, 'loss/train': 1.4412498474121094}
+02/25/2022 22:34:34 - INFO - codeparrot_training - Step 34756: {'lr': 0.00011444460447541302, 'samples': 17795584, 'steps': 34756, 'loss/train': 2.297098159790039}
+02/25/2022 22:34:40 - INFO - codeparrot_training - Step 34757: {'lr': 0.00011443085644395315, 'samples': 17796096, 'steps': 34757, 'loss/train': 1.4687154293060303}
+02/25/2022 22:34:44 - INFO - codeparrot_training - Step 34758: {'lr': 0.00011441710899322863, 'samples': 17796608, 'steps': 34758, 'loss/train': 0.5483139753341675}
+02/25/2022 22:34:50 - INFO - codeparrot_training - Step 34759: {'lr': 0.00011440336212329796, 'samples': 17797120, 'steps': 34759, 'loss/train': 1.6640715599060059}
+02/25/2022 22:34:53 - INFO - codeparrot_training - Step 34760: {'lr': 0.00011438961583422036, 'samples': 17797632, 'steps': 34760, 'loss/train': 1.5567317008972168}
+02/25/2022 22:34:59 - INFO - codeparrot_training - Step 34761: {'lr': 0.00011437587012605446, 'samples': 17798144, 'steps': 34761, 'loss/train': 2.008934497833252}
+02/25/2022 22:35:02 - INFO - codeparrot_training - Step 34762: {'lr': 0.00011436212499885937, 'samples': 17798656, 'steps': 34762, 'loss/train': 1.55619215965271}
+02/25/2022 22:35:08 - INFO - codeparrot_training - Step 34763: {'lr': 0.00011434838045269388, 'samples': 17799168, 'steps': 34763, 'loss/train': 1.3606946468353271}
+02/25/2022 22:35:11 - INFO - codeparrot_training - Step 34764: {'lr': 0.00011433463648761682, 'samples': 17799680, 'steps': 34764, 'loss/train': 1.671073317527771}
+02/25/2022 22:35:17 - INFO - codeparrot_training - Step 34765: {'lr': 0.00011432089310368704, 'samples': 17800192, 'steps': 34765, 'loss/train': 2.27286434173584}
+02/25/2022 22:35:20 - INFO - codeparrot_training - Step 34766: {'lr': 0.00011430715030096356, 'samples': 17800704, 'steps': 34766, 'loss/train': 1.3739622831344604}
+02/25/2022 22:35:26 - INFO - codeparrot_training - Step 34767: {'lr': 0.00011429340807950514, 'samples': 17801216, 'steps': 34767, 'loss/train': 0.9082155227661133}
+02/25/2022 22:35:29 - INFO - codeparrot_training - Step 34768: {'lr': 0.00011427966643937068, 'samples': 17801728, 'steps': 34768, 'loss/train': 2.0918383598327637}
+02/25/2022 22:35:35 - INFO - codeparrot_training - Step 34769: {'lr': 0.00011426592538061895, 'samples': 17802240, 'steps': 34769, 'loss/train': 1.7435027360916138}
+02/25/2022 22:35:39 - INFO - codeparrot_training - Step 34770: {'lr': 0.00011425218490330896, 'samples': 17802752, 'steps': 34770, 'loss/train': 0.36293432116508484}
+02/25/2022 22:35:44 - INFO - codeparrot_training - Step 34771: {'lr': 0.00011423844500749944, 'samples': 17803264, 'steps': 34771, 'loss/train': 2.0686099529266357}
+02/25/2022 22:35:48 - INFO - codeparrot_training - Step 34772: {'lr': 0.00011422470569324949, 'samples': 17803776, 'steps': 34772, 'loss/train': 0.8506045937538147}
+02/25/2022 22:35:53 - INFO - codeparrot_training - Step 34773: {'lr': 0.00011421096696061758, 'samples': 17804288, 'steps': 34773, 'loss/train': 0.5906069874763489}
+02/25/2022 22:35:57 - INFO - codeparrot_training - Step 34774: {'lr': 0.00011419722880966291, 'samples': 17804800, 'steps': 34774, 'loss/train': 1.8113291263580322}
+02/25/2022 22:36:02 - INFO - codeparrot_training - Step 34775: {'lr': 0.00011418349124044405, 'samples': 17805312, 'steps': 34775, 'loss/train': 1.4793859720230103}
+02/25/2022 22:36:06 - INFO - codeparrot_training - Step 34776: {'lr': 0.0001141697542530202, 'samples': 17805824, 'steps': 34776, 'loss/train': 1.440613031387329}
+02/25/2022 22:36:12 - INFO - codeparrot_training - Step 34777: {'lr': 0.0001141560178474498, 'samples': 17806336, 'steps': 34777, 'loss/train': 1.905440092086792}
+02/25/2022 22:36:15 - INFO - codeparrot_training - Step 34778: {'lr': 0.00011414228202379195, 'samples': 17806848, 'steps': 34778, 'loss/train': 1.4701566696166992}
+02/25/2022 22:36:21 - INFO - codeparrot_training - Step 34779: {'lr': 0.00011412854678210546, 'samples': 17807360, 'steps': 34779, 'loss/train': 2.088491678237915}
+02/25/2022 22:36:25 - INFO - codeparrot_training - Step 34780: {'lr': 0.00011411481212244912, 'samples': 17807872, 'steps': 34780, 'loss/train': 1.9860289096832275}
+02/25/2022 22:36:30 - INFO - codeparrot_training - Step 34781: {'lr': 0.00011410107804488168, 'samples': 17808384, 'steps': 34781, 'loss/train': 0.901138961315155}
+02/25/2022 22:36:34 - INFO - codeparrot_training - Step 34782: {'lr': 0.00011408734454946218, 'samples': 17808896, 'steps': 34782, 'loss/train': 2.3694934844970703}
+02/25/2022 22:36:39 - INFO - codeparrot_training - Step 34783: {'lr': 0.00011407361163624933, 'samples': 17809408, 'steps': 34783, 'loss/train': 1.860734462738037}
+02/25/2022 22:36:43 - INFO - codeparrot_training - Step 34784: {'lr': 0.00011405987930530184, 'samples': 17809920, 'steps': 34784, 'loss/train': 1.1248018741607666}
+02/25/2022 22:36:48 - INFO - codeparrot_training - Step 34785: {'lr': 0.0001140461475566788, 'samples': 17810432, 'steps': 34785, 'loss/train': 1.4287890195846558}
+02/25/2022 22:36:52 - INFO - codeparrot_training - Step 34786: {'lr': 0.00011403241639043887, 'samples': 17810944, 'steps': 34786, 'loss/train': 3.293961763381958}
+02/25/2022 22:36:57 - INFO - codeparrot_training - Step 34787: {'lr': 0.0001140186858066409, 'samples': 17811456, 'steps': 34787, 'loss/train': 1.580237865447998}
+02/25/2022 22:37:01 - INFO - codeparrot_training - Step 34788: {'lr': 0.00011400495580534362, 'samples': 17811968, 'steps': 34788, 'loss/train': 1.9166027307510376}
+02/25/2022 22:37:07 - INFO - codeparrot_training - Step 34789: {'lr': 0.000113991226386606, 'samples': 17812480, 'steps': 34789, 'loss/train': 3.150430917739868}
+02/25/2022 22:37:10 - INFO - codeparrot_training - Step 34790: {'lr': 0.00011397749755048678, 'samples': 17812992, 'steps': 34790, 'loss/train': 2.3032455444335938}
+02/25/2022 22:37:16 - INFO - codeparrot_training - Step 34791: {'lr': 0.00011396376929704477, 'samples': 17813504, 'steps': 34791, 'loss/train': 1.786041021347046}
+02/25/2022 22:37:19 - INFO - codeparrot_training - Step 34792: {'lr': 0.00011395004162633868, 'samples': 17814016, 'steps': 34792, 'loss/train': 1.5370835065841675}
+02/25/2022 22:37:25 - INFO - codeparrot_training - Step 34793: {'lr': 0.00011393631453842763, 'samples': 17814528, 'steps': 34793, 'loss/train': 1.2027186155319214}
+02/25/2022 22:37:28 - INFO - codeparrot_training - Step 34794: {'lr': 0.00011392258803336996, 'samples': 17815040, 'steps': 34794, 'loss/train': 1.5637550354003906}
+02/25/2022 22:37:34 - INFO - codeparrot_training - Step 34795: {'lr': 0.00011390886211122484, 'samples': 17815552, 'steps': 34795, 'loss/train': 0.5180971026420593}
+02/25/2022 22:37:38 - INFO - codeparrot_training - Step 34796: {'lr': 0.00011389513677205084, 'samples': 17816064, 'steps': 34796, 'loss/train': 0.08174548298120499}
+02/25/2022 22:37:43 - INFO - codeparrot_training - Step 34797: {'lr': 0.00011388141201590696, 'samples': 17816576, 'steps': 34797, 'loss/train': 0.4705628454685211}
+02/25/2022 22:37:47 - INFO - codeparrot_training - Step 34798: {'lr': 0.0001138676878428519, 'samples': 17817088, 'steps': 34798, 'loss/train': 0.3046160936355591}
+02/25/2022 22:37:52 - INFO - codeparrot_training - Step 34799: {'lr': 0.00011385396425294439, 'samples': 17817600, 'steps': 34799, 'loss/train': 1.660393238067627}
+02/25/2022 22:37:56 - INFO - codeparrot_training - Step 34800: {'lr': 0.00011384024124624323, 'samples': 17818112, 'steps': 34800, 'loss/train': 0.9168550968170166}
+02/25/2022 22:38:01 - INFO - codeparrot_training - Step 34801: {'lr': 0.00011382651882280728, 'samples': 17818624, 'steps': 34801, 'loss/train': 1.9195979833602905}
+02/25/2022 22:38:08 - INFO - codeparrot_training - Step 34802: {'lr': 0.00011381279698269533, 'samples': 17819136, 'steps': 34802, 'loss/train': 2.484697103500366}
+02/25/2022 22:38:11 - INFO - codeparrot_training - Step 34803: {'lr': 0.00011379907572596609, 'samples': 17819648, 'steps': 34803, 'loss/train': 2.1058056354522705}
+02/25/2022 22:38:17 - INFO - codeparrot_training - Step 34804: {'lr': 0.00011378535505267825, 'samples': 17820160, 'steps': 34804, 'loss/train': 1.8874750137329102}
+02/25/2022 22:38:21 - INFO - codeparrot_training - Step 34805: {'lr': 0.00011377163496289084, 'samples': 17820672, 'steps': 34805, 'loss/train': 0.70106041431427}
+02/25/2022 22:38:24 - INFO - codeparrot_training - Step 34806: {'lr': 0.00011375791545666245, 'samples': 17821184, 'steps': 34806, 'loss/train': 1.9727753400802612}
+02/25/2022 22:38:30 - INFO - codeparrot_training - Step 34807: {'lr': 0.0001137441965340519, 'samples': 17821696, 'steps': 34807, 'loss/train': 1.4847525358200073}
+02/25/2022 22:38:33 - INFO - codeparrot_training - Step 34808: {'lr': 0.00011373047819511783, 'samples': 17822208, 'steps': 34808, 'loss/train': 2.4324989318847656}
+02/25/2022 22:38:39 - INFO - codeparrot_training - Step 34809: {'lr': 0.00011371676043991924, 'samples': 17822720, 'steps': 34809, 'loss/train': 2.353549003601074}
+02/25/2022 22:38:42 - INFO - codeparrot_training - Step 34810: {'lr': 0.00011370304326851469, 'samples': 17823232, 'steps': 34810, 'loss/train': 1.6818206310272217}
+02/25/2022 22:38:48 - INFO - codeparrot_training - Step 34811: {'lr': 0.00011368932668096318, 'samples': 17823744, 'steps': 34811, 'loss/train': 2.0532338619232178}
+02/25/2022 22:38:51 - INFO - codeparrot_training - Step 34812: {'lr': 0.00011367561067732316, 'samples': 17824256, 'steps': 34812, 'loss/train': 2.611616373062134}
+02/25/2022 22:38:57 - INFO - codeparrot_training - Step 34813: {'lr': 0.00011366189525765361, 'samples': 17824768, 'steps': 34813, 'loss/train': 2.118795871734619}
+02/25/2022 22:39:00 - INFO - codeparrot_training - Step 34814: {'lr': 0.00011364818042201311, 'samples': 17825280, 'steps': 34814, 'loss/train': 2.0578081607818604}
+02/25/2022 22:39:07 - INFO - codeparrot_training - Step 34815: {'lr': 0.00011363446617046069, 'samples': 17825792, 'steps': 34815, 'loss/train': 0.41846537590026855}
+02/25/2022 22:39:10 - INFO - codeparrot_training - Step 34816: {'lr': 0.00011362075250305473, 'samples': 17826304, 'steps': 34816, 'loss/train': 1.268330454826355}
+02/25/2022 22:39:16 - INFO - codeparrot_training - Step 34817: {'lr': 0.00011360703941985429, 'samples': 17826816, 'steps': 34817, 'loss/train': 2.0518436431884766}
+02/25/2022 22:39:19 - INFO - codeparrot_training - Step 34818: {'lr': 0.00011359332692091784, 'samples': 17827328, 'steps': 34818, 'loss/train': 0.8262566328048706}
+02/25/2022 22:39:25 - INFO - codeparrot_training - Step 34819: {'lr': 0.00011357961500630439, 'samples': 17827840, 'steps': 34819, 'loss/train': 0.8251410126686096}
+02/25/2022 22:39:28 - INFO - codeparrot_training - Step 34820: {'lr': 0.00011356590367607253, 'samples': 17828352, 'steps': 34820, 'loss/train': 1.5320590734481812}
+02/25/2022 22:39:34 - INFO - codeparrot_training - Step 34821: {'lr': 0.00011355219293028104, 'samples': 17828864, 'steps': 34821, 'loss/train': 2.3063979148864746}
+02/25/2022 22:39:37 - INFO - codeparrot_training - Step 34822: {'lr': 0.0001135384827689885, 'samples': 17829376, 'steps': 34822, 'loss/train': 1.833272933959961}
+02/25/2022 22:39:43 - INFO - codeparrot_training - Step 34823: {'lr': 0.00011352477319225391, 'samples': 17829888, 'steps': 34823, 'loss/train': 0.49192261695861816}
+02/25/2022 22:39:46 - INFO - codeparrot_training - Step 34824: {'lr': 0.00011351106420013583, 'samples': 17830400, 'steps': 34824, 'loss/train': 1.8050167560577393}
+02/25/2022 22:39:53 - INFO - codeparrot_training - Step 34825: {'lr': 0.000113497355792693, 'samples': 17830912, 'steps': 34825, 'loss/train': 1.7704142332077026}
+02/25/2022 22:39:56 - INFO - codeparrot_training - Step 34826: {'lr': 0.00011348364796998417, 'samples': 17831424, 'steps': 34826, 'loss/train': 2.008364677429199}
+02/25/2022 22:40:02 - INFO - codeparrot_training - Step 34827: {'lr': 0.00011346994073206792, 'samples': 17831936, 'steps': 34827, 'loss/train': 1.5049372911453247}
+02/25/2022 22:40:06 - INFO - codeparrot_training - Step 34828: {'lr': 0.00011345623407900322, 'samples': 17832448, 'steps': 34828, 'loss/train': 2.518047332763672}
+02/25/2022 22:40:11 - INFO - codeparrot_training - Step 34829: {'lr': 0.00011344252801084864, 'samples': 17832960, 'steps': 34829, 'loss/train': 2.15262508392334}
+02/25/2022 22:40:14 - INFO - codeparrot_training - Step 34830: {'lr': 0.0001134288225276629, 'samples': 17833472, 'steps': 34830, 'loss/train': 1.8653783798217773}
+02/25/2022 22:40:20 - INFO - codeparrot_training - Step 34831: {'lr': 0.00011341511762950466, 'samples': 17833984, 'steps': 34831, 'loss/train': 1.3218104839324951}
+02/25/2022 22:40:26 - INFO - codeparrot_training - Step 34832: {'lr': 0.00011340141331643275, 'samples': 17834496, 'steps': 34832, 'loss/train': 2.3935933113098145}
+02/25/2022 22:40:29 - INFO - codeparrot_training - Step 34833: {'lr': 0.00011338770958850581, 'samples': 17835008, 'steps': 34833, 'loss/train': 2.6254355907440186}
+02/25/2022 22:40:35 - INFO - codeparrot_training - Step 34834: {'lr': 0.00011337400644578255, 'samples': 17835520, 'steps': 34834, 'loss/train': 2.383310556411743}
+02/25/2022 22:40:39 - INFO - codeparrot_training - Step 34835: {'lr': 0.00011336030388832156, 'samples': 17836032, 'steps': 34835, 'loss/train': 1.4206146001815796}
+02/25/2022 22:40:42 - INFO - codeparrot_training - Step 34836: {'lr': 0.00011334660191618174, 'samples': 17836544, 'steps': 34836, 'loss/train': 0.7846734523773193}
+02/25/2022 22:40:48 - INFO - codeparrot_training - Step 34837: {'lr': 0.0001133329005294217, 'samples': 17837056, 'steps': 34837, 'loss/train': 0.32985156774520874}
+02/25/2022 22:40:51 - INFO - codeparrot_training - Step 34838: {'lr': 0.0001133191997281001, 'samples': 17837568, 'steps': 34838, 'loss/train': 2.610053300857544}
+02/25/2022 22:40:56 - INFO - codeparrot_training - Step 34839: {'lr': 0.00011330549951227556, 'samples': 17838080, 'steps': 34839, 'loss/train': 1.4487978219985962}
+02/25/2022 22:41:00 - INFO - codeparrot_training - Step 34840: {'lr': 0.00011329179988200694, 'samples': 17838592, 'steps': 34840, 'loss/train': 1.882456660270691}
+02/25/2022 22:41:06 - INFO - codeparrot_training - Step 34841: {'lr': 0.00011327810083735285, 'samples': 17839104, 'steps': 34841, 'loss/train': 1.6914174556732178}
+02/25/2022 22:41:10 - INFO - codeparrot_training - Step 34842: {'lr': 0.00011326440237837196, 'samples': 17839616, 'steps': 34842, 'loss/train': 1.529268503189087}
+02/25/2022 22:41:15 - INFO - codeparrot_training - Step 34843: {'lr': 0.00011325070450512285, 'samples': 17840128, 'steps': 34843, 'loss/train': 1.900531530380249}
+02/25/2022 22:41:19 - INFO - codeparrot_training - Step 34844: {'lr': 0.00011323700721766439, 'samples': 17840640, 'steps': 34844, 'loss/train': 2.039630651473999}
+02/25/2022 22:41:24 - INFO - codeparrot_training - Step 34845: {'lr': 0.00011322331051605509, 'samples': 17841152, 'steps': 34845, 'loss/train': 1.405947208404541}
+02/25/2022 22:41:28 - INFO - codeparrot_training - Step 34846: {'lr': 0.00011320961440035386, 'samples': 17841664, 'steps': 34846, 'loss/train': 2.322772264480591}
+02/25/2022 22:41:33 - INFO - codeparrot_training - Step 34847: {'lr': 0.00011319591887061903, 'samples': 17842176, 'steps': 34847, 'loss/train': 1.8700200319290161}
+02/25/2022 22:41:37 - INFO - codeparrot_training - Step 34848: {'lr': 0.00011318222392690956, 'samples': 17842688, 'steps': 34848, 'loss/train': 2.925299644470215}
+02/25/2022 22:41:42 - INFO - codeparrot_training - Step 34849: {'lr': 0.00011316852956928386, 'samples': 17843200, 'steps': 34849, 'loss/train': 1.8161739110946655}
+02/25/2022 22:41:46 - INFO - codeparrot_training - Step 34850: {'lr': 0.00011315483579780094, 'samples': 17843712, 'steps': 34850, 'loss/train': 1.7367805242538452}
+02/25/2022 22:41:52 - INFO - codeparrot_training - Step 34851: {'lr': 0.00011314114261251906, 'samples': 17844224, 'steps': 34851, 'loss/train': 2.5559041500091553}
+02/25/2022 22:41:56 - INFO - codeparrot_training - Step 34852: {'lr': 0.00011312745001349717, 'samples': 17844736, 'steps': 34852, 'loss/train': 2.043379545211792}
+02/25/2022 22:42:01 - INFO - codeparrot_training - Step 34853: {'lr': 0.00011311375800079371, 'samples': 17845248, 'steps': 34853, 'loss/train': 0.7614732384681702}
+02/25/2022 22:42:05 - INFO - codeparrot_training - Step 34854: {'lr': 0.00011310006657446764, 'samples': 17845760, 'steps': 34854, 'loss/train': 0.9589431285858154}
+02/25/2022 22:42:10 - INFO - codeparrot_training - Step 34855: {'lr': 0.00011308637573457725, 'samples': 17846272, 'steps': 34855, 'loss/train': 1.9918023347854614}
+02/25/2022 22:42:14 - INFO - codeparrot_training - Step 34856: {'lr': 0.00011307268548118141, 'samples': 17846784, 'steps': 34856, 'loss/train': 0.6873029470443726}
+02/25/2022 22:42:19 - INFO - codeparrot_training - Step 34857: {'lr': 0.00011305899581433862, 'samples': 17847296, 'steps': 34857, 'loss/train': 1.7289001941680908}
+02/25/2022 22:42:23 - INFO - codeparrot_training - Step 34858: {'lr': 0.00011304530673410773, 'samples': 17847808, 'steps': 34858, 'loss/train': 1.7044283151626587}
+02/25/2022 22:42:28 - INFO - codeparrot_training - Step 34859: {'lr': 0.00011303161824054723, 'samples': 17848320, 'steps': 34859, 'loss/train': 2.1292171478271484}
+02/25/2022 22:42:32 - INFO - codeparrot_training - Step 34860: {'lr': 0.00011301793033371579, 'samples': 17848832, 'steps': 34860, 'loss/train': 1.9011460542678833}
+02/25/2022 22:42:38 - INFO - codeparrot_training - Step 34861: {'lr': 0.00011300424301367196, 'samples': 17849344, 'steps': 34861, 'loss/train': 2.2141149044036865}
+02/25/2022 22:42:41 - INFO - codeparrot_training - Step 34862: {'lr': 0.00011299055628047452, 'samples': 17849856, 'steps': 34862, 'loss/train': 1.8247283697128296}
+02/25/2022 22:42:47 - INFO - codeparrot_training - Step 34863: {'lr': 0.00011297687013418206, 'samples': 17850368, 'steps': 34863, 'loss/train': 1.9151647090911865}
+02/25/2022 22:42:50 - INFO - codeparrot_training - Step 34864: {'lr': 0.00011296318457485316, 'samples': 17850880, 'steps': 34864, 'loss/train': 2.1352217197418213}
+02/25/2022 22:42:56 - INFO - codeparrot_training - Step 34865: {'lr': 0.00011294949960254639, 'samples': 17851392, 'steps': 34865, 'loss/train': 1.6991958618164062}
+02/25/2022 22:42:59 - INFO - codeparrot_training - Step 34866: {'lr': 0.00011293581521732052, 'samples': 17851904, 'steps': 34866, 'loss/train': 1.3736205101013184}
+02/25/2022 22:43:05 - INFO - codeparrot_training - Step 34867: {'lr': 0.00011292213141923407, 'samples': 17852416, 'steps': 34867, 'loss/train': 2.411273717880249}
+02/25/2022 22:43:08 - INFO - codeparrot_training - Step 34868: {'lr': 0.00011290844820834572, 'samples': 17852928, 'steps': 34868, 'loss/train': 1.9816040992736816}
+02/25/2022 22:43:14 - INFO - codeparrot_training - Step 34869: {'lr': 0.00011289476558471396, 'samples': 17853440, 'steps': 34869, 'loss/train': 2.042404890060425}
+02/25/2022 22:43:17 - INFO - codeparrot_training - Step 34870: {'lr': 0.00011288108354839755, 'samples': 17853952, 'steps': 34870, 'loss/train': 2.1103172302246094}
+02/25/2022 22:43:23 - INFO - codeparrot_training - Step 34871: {'lr': 0.00011286740209945507, 'samples': 17854464, 'steps': 34871, 'loss/train': 1.5293422937393188}
+02/25/2022 22:43:27 - INFO - codeparrot_training - Step 34872: {'lr': 0.00011285372123794507, 'samples': 17854976, 'steps': 34872, 'loss/train': 0.029536200687289238}
+02/25/2022 22:43:32 - INFO - codeparrot_training - Step 34873: {'lr': 0.00011284004096392619, 'samples': 17855488, 'steps': 34873, 'loss/train': 1.5277615785598755}
+02/25/2022 22:43:36 - INFO - codeparrot_training - Step 34874: {'lr': 0.0001128263612774569, 'samples': 17856000, 'steps': 34874, 'loss/train': 2.2245821952819824}
+02/25/2022 22:43:41 - INFO - codeparrot_training - Step 34875: {'lr': 0.00011281268217859606, 'samples': 17856512, 'steps': 34875, 'loss/train': 2.654569387435913}
+02/25/2022 22:43:45 - INFO - codeparrot_training - Step 34876: {'lr': 0.00011279900366740209, 'samples': 17857024, 'steps': 34876, 'loss/train': 1.8947581052780151}
+02/25/2022 22:43:50 - INFO - codeparrot_training - Step 34877: {'lr': 0.00011278532574393366, 'samples': 17857536, 'steps': 34877, 'loss/train': 2.1822595596313477}
+02/25/2022 22:43:54 - INFO - codeparrot_training - Step 34878: {'lr': 0.0001127716484082492, 'samples': 17858048, 'steps': 34878, 'loss/train': 2.4381775856018066}
+02/25/2022 22:43:59 - INFO - codeparrot_training - Step 34879: {'lr': 0.00011275797166040752, 'samples': 17858560, 'steps': 34879, 'loss/train': 2.6193206310272217}
+02/25/2022 22:44:03 - INFO - codeparrot_training - Step 34880: {'lr': 0.00011274429550046702, 'samples': 17859072, 'steps': 34880, 'loss/train': 2.463113784790039}
+02/25/2022 22:44:08 - INFO - codeparrot_training - Step 34881: {'lr': 0.00011273061992848657, 'samples': 17859584, 'steps': 34881, 'loss/train': 1.1435399055480957}
+02/25/2022 22:44:12 - INFO - codeparrot_training - Step 34882: {'lr': 0.00011271694494452436, 'samples': 17860096, 'steps': 34882, 'loss/train': 1.7033058404922485}
+02/25/2022 22:44:17 - INFO - codeparrot_training - Step 34883: {'lr': 0.00011270327054863927, 'samples': 17860608, 'steps': 34883, 'loss/train': 1.3589262962341309}
+02/25/2022 22:44:21 - INFO - codeparrot_training - Step 34884: {'lr': 0.00011268959674088966, 'samples': 17861120, 'steps': 34884, 'loss/train': 1.501327395439148}
+02/25/2022 22:44:26 - INFO - codeparrot_training - Step 34885: {'lr': 0.00011267592352133444, 'samples': 17861632, 'steps': 34885, 'loss/train': 2.5504422187805176}
+02/25/2022 22:44:30 - INFO - codeparrot_training - Step 34886: {'lr': 0.00011266225089003173, 'samples': 17862144, 'steps': 34886, 'loss/train': 0.5927507281303406}
+02/25/2022 22:44:36 - INFO - codeparrot_training - Step 34887: {'lr': 0.00011264857884704045, 'samples': 17862656, 'steps': 34887, 'loss/train': 2.2058801651000977}
+02/25/2022 22:44:39 - INFO - codeparrot_training - Step 34888: {'lr': 0.00011263490739241894, 'samples': 17863168, 'steps': 34888, 'loss/train': 1.3723721504211426}
+02/25/2022 22:44:45 - INFO - codeparrot_training - Step 34889: {'lr': 0.00011262123652622605, 'samples': 17863680, 'steps': 34889, 'loss/train': 1.5350292921066284}
+02/25/2022 22:44:48 - INFO - codeparrot_training - Step 34890: {'lr': 0.00011260756624851997, 'samples': 17864192, 'steps': 34890, 'loss/train': 2.0353267192840576}
+02/25/2022 22:44:54 - INFO - codeparrot_training - Step 34891: {'lr': 0.00011259389655935953, 'samples': 17864704, 'steps': 34891, 'loss/train': 2.067101240158081}
+02/25/2022 22:44:57 - INFO - codeparrot_training - Step 34892: {'lr': 0.00011258022745880315, 'samples': 17865216, 'steps': 34892, 'loss/train': 2.164034128189087}
+02/25/2022 22:45:03 - INFO - codeparrot_training - Step 34893: {'lr': 0.00011256655894690951, 'samples': 17865728, 'steps': 34893, 'loss/train': 0.8312376737594604}
+02/25/2022 22:45:06 - INFO - codeparrot_training - Step 34894: {'lr': 0.00011255289102373708, 'samples': 17866240, 'steps': 34894, 'loss/train': 0.8810471296310425}
+02/25/2022 22:45:12 - INFO - codeparrot_training - Step 34895: {'lr': 0.00011253922368934446, 'samples': 17866752, 'steps': 34895, 'loss/train': 2.441986560821533}
+02/25/2022 22:45:15 - INFO - codeparrot_training - Step 34896: {'lr': 0.00011252555694379005, 'samples': 17867264, 'steps': 34896, 'loss/train': 2.1777114868164062}
+02/25/2022 22:45:22 - INFO - codeparrot_training - Step 34897: {'lr': 0.00011251189078713255, 'samples': 17867776, 'steps': 34897, 'loss/train': 0.8940674066543579}
+02/25/2022 22:45:25 - INFO - codeparrot_training - Step 34898: {'lr': 0.00011249822521943051, 'samples': 17868288, 'steps': 34898, 'loss/train': 2.1172637939453125}
+02/25/2022 22:45:31 - INFO - codeparrot_training - Step 34899: {'lr': 0.00011248456024074239, 'samples': 17868800, 'steps': 34899, 'loss/train': 1.5558733940124512}
+02/25/2022 22:45:34 - INFO - codeparrot_training - Step 34900: {'lr': 0.00011247089585112666, 'samples': 17869312, 'steps': 34900, 'loss/train': 1.7457056045532227}
+02/25/2022 22:45:40 - INFO - codeparrot_training - Step 34901: {'lr': 0.000112457232050642, 'samples': 17869824, 'steps': 34901, 'loss/train': 2.31425142288208}
+02/25/2022 22:45:43 - INFO - codeparrot_training - Step 34902: {'lr': 0.00011244356883934692, 'samples': 17870336, 'steps': 34902, 'loss/train': 2.0756053924560547}
+02/25/2022 22:45:49 - INFO - codeparrot_training - Step 34903: {'lr': 0.0001124299062172999, 'samples': 17870848, 'steps': 34903, 'loss/train': 2.2840089797973633}
+02/25/2022 22:45:52 - INFO - codeparrot_training - Step 34904: {'lr': 0.0001124162441845594, 'samples': 17871360, 'steps': 34904, 'loss/train': 1.2529807090759277}
+02/25/2022 22:45:58 - INFO - codeparrot_training - Step 34905: {'lr': 0.0001124025827411841, 'samples': 17871872, 'steps': 34905, 'loss/train': 1.1864484548568726}
+02/25/2022 22:46:01 - INFO - codeparrot_training - Step 34906: {'lr': 0.00011238892188723245, 'samples': 17872384, 'steps': 34906, 'loss/train': 1.8505655527114868}
+02/25/2022 22:46:07 - INFO - codeparrot_training - Step 34907: {'lr': 0.00011237526162276296, 'samples': 17872896, 'steps': 34907, 'loss/train': 2.631504774093628}
+02/25/2022 22:46:11 - INFO - codeparrot_training - Step 34908: {'lr': 0.00011236160194783404, 'samples': 17873408, 'steps': 34908, 'loss/train': 0.26429206132888794}
+02/25/2022 22:46:16 - INFO - codeparrot_training - Step 34909: {'lr': 0.00011234794286250443, 'samples': 17873920, 'steps': 34909, 'loss/train': 1.490645408630371}
+02/25/2022 22:46:20 - INFO - codeparrot_training - Step 34910: {'lr': 0.00011233428436683252, 'samples': 17874432, 'steps': 34910, 'loss/train': 1.4881706237792969}
+02/25/2022 22:46:25 - INFO - codeparrot_training - Step 34911: {'lr': 0.00011232062646087682, 'samples': 17874944, 'steps': 34911, 'loss/train': 1.62655508518219}
+02/25/2022 22:46:31 - INFO - codeparrot_training - Step 34912: {'lr': 0.00011230696914469573, 'samples': 17875456, 'steps': 34912, 'loss/train': 2.2832798957824707}
+02/25/2022 22:46:34 - INFO - codeparrot_training - Step 34913: {'lr': 0.00011229331241834798, 'samples': 17875968, 'steps': 34913, 'loss/train': 1.8659603595733643}
+02/25/2022 22:46:38 - INFO - codeparrot_training - Step 34914: {'lr': 0.00011227965628189194, 'samples': 17876480, 'steps': 34914, 'loss/train': 1.4666457176208496}
+02/25/2022 22:46:43 - INFO - codeparrot_training - Step 34915: {'lr': 0.00011226600073538609, 'samples': 17876992, 'steps': 34915, 'loss/train': 0.9396272301673889}
+02/25/2022 22:46:47 - INFO - codeparrot_training - Step 34916: {'lr': 0.0001122523457788889, 'samples': 17877504, 'steps': 34916, 'loss/train': 2.5893049240112305}
+02/25/2022 22:46:52 - INFO - codeparrot_training - Step 34917: {'lr': 0.00011223869141245902, 'samples': 17878016, 'steps': 34917, 'loss/train': 0.9040366411209106}
+02/25/2022 22:46:59 - INFO - codeparrot_training - Step 34918: {'lr': 0.00011222503763615483, 'samples': 17878528, 'steps': 34918, 'loss/train': 2.832584857940674}
+02/25/2022 22:47:02 - INFO - codeparrot_training - Step 34919: {'lr': 0.00011221138445003473, 'samples': 17879040, 'steps': 34919, 'loss/train': 3.3160455226898193}
+02/25/2022 22:47:06 - INFO - codeparrot_training - Step 34920: {'lr': 0.00011219773185415749, 'samples': 17879552, 'steps': 34920, 'loss/train': 2.2828028202056885}
+02/25/2022 22:47:11 - INFO - codeparrot_training - Step 34921: {'lr': 0.00011218407984858126, 'samples': 17880064, 'steps': 34921, 'loss/train': 1.2578774690628052}
+02/25/2022 22:47:15 - INFO - codeparrot_training - Step 34922: {'lr': 0.00011217042843336472, 'samples': 17880576, 'steps': 34922, 'loss/train': 1.8347923755645752}
+02/25/2022 22:47:20 - INFO - codeparrot_training - Step 34923: {'lr': 0.00011215677760856622, 'samples': 17881088, 'steps': 34923, 'loss/train': 1.6547961235046387}
+02/25/2022 22:47:24 - INFO - codeparrot_training - Step 34924: {'lr': 0.00011214312737424451, 'samples': 17881600, 'steps': 34924, 'loss/train': 1.731743335723877}
+02/25/2022 22:47:29 - INFO - codeparrot_training - Step 34925: {'lr': 0.00011212947773045767, 'samples': 17882112, 'steps': 34925, 'loss/train': 0.5325666069984436}
+02/25/2022 22:47:33 - INFO - codeparrot_training - Step 34926: {'lr': 0.00011211582867726447, 'samples': 17882624, 'steps': 34926, 'loss/train': 1.871329426765442}
+02/25/2022 22:47:38 - INFO - codeparrot_training - Step 34927: {'lr': 0.00011210218021472318, 'samples': 17883136, 'steps': 34927, 'loss/train': 2.5138485431671143}
+02/25/2022 22:47:42 - INFO - codeparrot_training - Step 34928: {'lr': 0.00011208853234289245, 'samples': 17883648, 'steps': 34928, 'loss/train': 1.7874820232391357}
+02/25/2022 22:47:49 - INFO - codeparrot_training - Step 34929: {'lr': 0.00011207488506183067, 'samples': 17884160, 'steps': 34929, 'loss/train': 1.4505573511123657}
+02/25/2022 22:47:52 - INFO - codeparrot_training - Step 34930: {'lr': 0.00011206123837159626, 'samples': 17884672, 'steps': 34930, 'loss/train': 1.51365327835083}
+02/25/2022 22:47:58 - INFO - codeparrot_training - Step 34931: {'lr': 0.00011204759227224762, 'samples': 17885184, 'steps': 34931, 'loss/train': 2.6081128120422363}
+02/25/2022 22:48:01 - INFO - codeparrot_training - Step 34932: {'lr': 0.0001120339467638434, 'samples': 17885696, 'steps': 34932, 'loss/train': 2.1144351959228516}
+02/25/2022 22:48:07 - INFO - codeparrot_training - Step 34933: {'lr': 0.00011202030184644194, 'samples': 17886208, 'steps': 34933, 'loss/train': 2.384094476699829}
+02/25/2022 22:48:10 - INFO - codeparrot_training - Step 34934: {'lr': 0.00011200665752010169, 'samples': 17886720, 'steps': 34934, 'loss/train': 2.0969252586364746}
+02/25/2022 22:48:15 - INFO - codeparrot_training - Step 34935: {'lr': 0.000111993013784881, 'samples': 17887232, 'steps': 34935, 'loss/train': 0.46304118633270264}
+02/25/2022 22:48:19 - INFO - codeparrot_training - Step 34936: {'lr': 0.00011197937064083852, 'samples': 17887744, 'steps': 34936, 'loss/train': 0.5004669427871704}
+02/25/2022 22:48:24 - INFO - codeparrot_training - Step 34937: {'lr': 0.00011196572808803258, 'samples': 17888256, 'steps': 34937, 'loss/train': 2.3549468517303467}
+02/25/2022 22:48:28 - INFO - codeparrot_training - Step 34938: {'lr': 0.00011195208612652163, 'samples': 17888768, 'steps': 34938, 'loss/train': 1.580697774887085}
+02/25/2022 22:48:34 - INFO - codeparrot_training - Step 34939: {'lr': 0.00011193844475636403, 'samples': 17889280, 'steps': 34939, 'loss/train': 0.3215804398059845}
+02/25/2022 22:48:38 - INFO - codeparrot_training - Step 34940: {'lr': 0.00011192480397761836, 'samples': 17889792, 'steps': 34940, 'loss/train': 1.3663384914398193}
+02/25/2022 22:48:43 - INFO - codeparrot_training - Step 34941: {'lr': 0.00011191116379034291, 'samples': 17890304, 'steps': 34941, 'loss/train': 1.5452347993850708}
+02/25/2022 22:48:47 - INFO - codeparrot_training - Step 34942: {'lr': 0.00011189752419459637, 'samples': 17890816, 'steps': 34942, 'loss/train': 1.679283857345581}
+02/25/2022 22:48:52 - INFO - codeparrot_training - Step 34943: {'lr': 0.00011188388519043679, 'samples': 17891328, 'steps': 34943, 'loss/train': 1.6597061157226562}
+02/25/2022 22:48:56 - INFO - codeparrot_training - Step 34944: {'lr': 0.0001118702467779229, 'samples': 17891840, 'steps': 34944, 'loss/train': 1.6378945112228394}
+02/25/2022 22:49:01 - INFO - codeparrot_training - Step 34945: {'lr': 0.00011185660895711292, 'samples': 17892352, 'steps': 34945, 'loss/train': 1.4472392797470093}
+02/25/2022 22:49:05 - INFO - codeparrot_training - Step 34946: {'lr': 0.00011184297172806554, 'samples': 17892864, 'steps': 34946, 'loss/train': 1.899036169052124}
+02/25/2022 22:49:10 - INFO - codeparrot_training - Step 34947: {'lr': 0.00011182933509083881, 'samples': 17893376, 'steps': 34947, 'loss/train': 2.451996088027954}
+02/25/2022 22:49:14 - INFO - codeparrot_training - Step 34948: {'lr': 0.00011181569904549144, 'samples': 17893888, 'steps': 34948, 'loss/train': 1.5890038013458252}
+02/25/2022 22:49:19 - INFO - codeparrot_training - Step 34949: {'lr': 0.00011180206359208162, 'samples': 17894400, 'steps': 34949, 'loss/train': 3.1941628456115723}
+02/25/2022 22:49:23 - INFO - codeparrot_training - Step 34950: {'lr': 0.00011178842873066808, 'samples': 17894912, 'steps': 34950, 'loss/train': 2.425978899002075}
+02/25/2022 22:49:30 - INFO - codeparrot_training - Step 34951: {'lr': 0.00011177479446130884, 'samples': 17895424, 'steps': 34951, 'loss/train': 1.8784880638122559}
+02/25/2022 22:49:33 - INFO - codeparrot_training - Step 34952: {'lr': 0.00011176116078406257, 'samples': 17895936, 'steps': 34952, 'loss/train': 1.4919337034225464}
+02/25/2022 22:49:38 - INFO - codeparrot_training - Step 34953: {'lr': 0.00011174752769898757, 'samples': 17896448, 'steps': 34953, 'loss/train': 2.8363733291625977}
+02/25/2022 22:49:42 - INFO - codeparrot_training - Step 34954: {'lr': 0.00011173389520614222, 'samples': 17896960, 'steps': 34954, 'loss/train': 1.3076075315475464}
+02/25/2022 22:49:48 - INFO - codeparrot_training - Step 34955: {'lr': 0.00011172026330558501, 'samples': 17897472, 'steps': 34955, 'loss/train': 1.2258182764053345}
+02/25/2022 22:49:51 - INFO - codeparrot_training - Step 34956: {'lr': 0.00011170663199737429, 'samples': 17897984, 'steps': 34956, 'loss/train': 1.2298986911773682}
+02/25/2022 22:49:56 - INFO - codeparrot_training - Step 34957: {'lr': 0.00011169300128156848, 'samples': 17898496, 'steps': 34957, 'loss/train': 1.129446029663086}
+02/25/2022 22:50:00 - INFO - codeparrot_training - Step 34958: {'lr': 0.00011167937115822583, 'samples': 17899008, 'steps': 34958, 'loss/train': 1.7038719654083252}
+02/25/2022 22:50:07 - INFO - codeparrot_training - Step 34959: {'lr': 0.0001116657416274049, 'samples': 17899520, 'steps': 34959, 'loss/train': 2.1519057750701904}
+02/25/2022 22:50:10 - INFO - codeparrot_training - Step 34960: {'lr': 0.000111652112689164, 'samples': 17900032, 'steps': 34960, 'loss/train': 0.7725964784622192}
+02/25/2022 22:50:16 - INFO - codeparrot_training - Step 34961: {'lr': 0.00011163848434356156, 'samples': 17900544, 'steps': 34961, 'loss/train': 2.222297430038452}
+02/25/2022 22:50:19 - INFO - codeparrot_training - Step 34962: {'lr': 0.0001116248565906558, 'samples': 17901056, 'steps': 34962, 'loss/train': 1.4858317375183105}
+02/25/2022 22:50:25 - INFO - codeparrot_training - Step 34963: {'lr': 0.0001116112294305053, 'samples': 17901568, 'steps': 34963, 'loss/train': 1.138819932937622}
+02/25/2022 22:50:28 - INFO - codeparrot_training - Step 34964: {'lr': 0.00011159760286316836, 'samples': 17902080, 'steps': 34964, 'loss/train': 1.6068347692489624}
+02/25/2022 22:50:34 - INFO - codeparrot_training - Step 34965: {'lr': 0.00011158397688870333, 'samples': 17902592, 'steps': 34965, 'loss/train': 1.5345022678375244}
+02/25/2022 22:50:38 - INFO - codeparrot_training - Step 34966: {'lr': 0.00011157035150716854, 'samples': 17903104, 'steps': 34966, 'loss/train': 1.7706342935562134}
+02/25/2022 22:50:43 - INFO - codeparrot_training - Step 34967: {'lr': 0.00011155672671862247, 'samples': 17903616, 'steps': 34967, 'loss/train': 2.6073594093322754}
+02/25/2022 22:50:47 - INFO - codeparrot_training - Step 34968: {'lr': 0.00011154310252312341, 'samples': 17904128, 'steps': 34968, 'loss/train': 2.2426705360412598}
+02/25/2022 22:50:52 - INFO - codeparrot_training - Step 34969: {'lr': 0.00011152947892072976, 'samples': 17904640, 'steps': 34969, 'loss/train': 1.2264734506607056}
+02/25/2022 22:50:56 - INFO - codeparrot_training - Step 34970: {'lr': 0.00011151585591149973, 'samples': 17905152, 'steps': 34970, 'loss/train': 8.609617233276367}
+02/25/2022 22:51:03 - INFO - codeparrot_training - Step 34971: {'lr': 0.00011150223349549193, 'samples': 17905664, 'steps': 34971, 'loss/train': 2.39105486869812}
+02/25/2022 22:51:06 - INFO - codeparrot_training - Step 34972: {'lr': 0.00011148861167276455, 'samples': 17906176, 'steps': 34972, 'loss/train': 1.4714531898498535}
+02/25/2022 22:51:12 - INFO - codeparrot_training - Step 34973: {'lr': 0.000111474990443376, 'samples': 17906688, 'steps': 34973, 'loss/train': 0.8845553994178772}
+02/25/2022 22:51:15 - INFO - codeparrot_training - Step 34974: {'lr': 0.0001114613698073845, 'samples': 17907200, 'steps': 34974, 'loss/train': 1.391049861907959}
+02/25/2022 22:51:21 - INFO - codeparrot_training - Step 34975: {'lr': 0.00011144774976484859, 'samples': 17907712, 'steps': 34975, 'loss/train': 1.3737519979476929}
+02/25/2022 22:51:24 - INFO - codeparrot_training - Step 34976: {'lr': 0.00011143413031582644, 'samples': 17908224, 'steps': 34976, 'loss/train': 1.7991303205490112}
+02/25/2022 22:51:30 - INFO - codeparrot_training - Step 34977: {'lr': 0.00011142051146037666, 'samples': 17908736, 'steps': 34977, 'loss/train': 1.9011895656585693}
+02/25/2022 22:51:33 - INFO - codeparrot_training - Step 34978: {'lr': 0.0001114068931985572, 'samples': 17909248, 'steps': 34978, 'loss/train': 2.7613868713378906}
+02/25/2022 22:51:39 - INFO - codeparrot_training - Step 34979: {'lr': 0.00011139327553042672, 'samples': 17909760, 'steps': 34979, 'loss/train': 2.097351312637329}
+02/25/2022 22:51:42 - INFO - codeparrot_training - Step 34980: {'lr': 0.00011137965845604333, 'samples': 17910272, 'steps': 34980, 'loss/train': 1.0668549537658691}
+02/25/2022 22:51:48 - INFO - codeparrot_training - Step 34981: {'lr': 0.00011136604197546566, 'samples': 17910784, 'steps': 34981, 'loss/train': 1.2242546081542969}
+02/25/2022 22:51:51 - INFO - codeparrot_training - Step 34982: {'lr': 0.00011135242608875166, 'samples': 17911296, 'steps': 34982, 'loss/train': 0.8575018644332886}
+02/25/2022 22:51:57 - INFO - codeparrot_training - Step 34983: {'lr': 0.00011133881079595992, 'samples': 17911808, 'steps': 34983, 'loss/train': 2.0313644409179688}
+02/25/2022 22:52:00 - INFO - codeparrot_training - Step 34984: {'lr': 0.0001113251960971486, 'samples': 17912320, 'steps': 34984, 'loss/train': 2.2020838260650635}
+02/25/2022 22:52:06 - INFO - codeparrot_training - Step 34985: {'lr': 0.00011131158199237628, 'samples': 17912832, 'steps': 34985, 'loss/train': 0.9864962100982666}
+02/25/2022 22:52:10 - INFO - codeparrot_training - Step 34986: {'lr': 0.00011129796848170095, 'samples': 17913344, 'steps': 34986, 'loss/train': 3.3161728382110596}
+02/25/2022 22:52:17 - INFO - codeparrot_training - Step 34987: {'lr': 0.00011128435556518115, 'samples': 17913856, 'steps': 34987, 'loss/train': 1.9074406623840332}
+02/25/2022 22:52:20 - INFO - codeparrot_training - Step 34988: {'lr': 0.00011127074324287504, 'samples': 17914368, 'steps': 34988, 'loss/train': 0.2186937928199768}
+02/25/2022 22:52:26 - INFO - codeparrot_training - Step 34989: {'lr': 0.0001112571315148411, 'samples': 17914880, 'steps': 34989, 'loss/train': 2.095627784729004}
+02/25/2022 22:52:29 - INFO - codeparrot_training - Step 34990: {'lr': 0.00011124352038113755, 'samples': 17915392, 'steps': 34990, 'loss/train': 1.5660802125930786}
+02/25/2022 22:52:35 - INFO - codeparrot_training - Step 34991: {'lr': 0.00011122990984182268, 'samples': 17915904, 'steps': 34991, 'loss/train': 1.0819975137710571}
+02/25/2022 22:52:38 - INFO - codeparrot_training - Step 34992: {'lr': 0.00011121629989695475, 'samples': 17916416, 'steps': 34992, 'loss/train': 1.5947767496109009}
+02/25/2022 22:52:44 - INFO - codeparrot_training - Step 34993: {'lr': 0.00011120269054659221, 'samples': 17916928, 'steps': 34993, 'loss/train': 2.1122138500213623}
+02/25/2022 22:52:47 - INFO - codeparrot_training - Step 34994: {'lr': 0.00011118908179079326, 'samples': 17917440, 'steps': 34994, 'loss/train': 1.865601658821106}
+02/25/2022 22:52:54 - INFO - codeparrot_training - Step 34995: {'lr': 0.00011117547362961622, 'samples': 17917952, 'steps': 34995, 'loss/train': 1.8611263036727905}
+02/25/2022 22:52:58 - INFO - codeparrot_training - Step 34996: {'lr': 0.0001111618660631193, 'samples': 17918464, 'steps': 34996, 'loss/train': 1.7628575563430786}
+02/25/2022 22:53:03 - INFO - codeparrot_training - Step 34997: {'lr': 0.00011114825909136089, 'samples': 17918976, 'steps': 34997, 'loss/train': 1.7161481380462646}
+02/25/2022 22:53:07 - INFO - codeparrot_training - Step 34998: {'lr': 0.00011113465271439929, 'samples': 17919488, 'steps': 34998, 'loss/train': 1.1114963293075562}
+02/25/2022 22:53:12 - INFO - codeparrot_training - Step 34999: {'lr': 0.00011112104693229272, 'samples': 17920000, 'steps': 34999, 'loss/train': 2.1389760971069336}
+02/25/2022 22:53:12 - INFO - codeparrot_training - Evaluating and saving model checkpoint