diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -44566,3 +44566,1009 @@ Use FP16 precision: False
 02/26/2022 10:32:50 - INFO - codeparrot_training - Step 43998: {'lr': 1.9042642136985117e-05, 'samples': 22527488, 'steps': 43998, 'loss/train': 1.2582118511199951}
 02/26/2022 10:32:55 - INFO - codeparrot_training - Step 43999: {'lr': 1.90363790098943e-05, 'samples': 22528000, 'steps': 43999, 'loss/train': 1.9502559900283813}
 02/26/2022 10:32:55 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 10:33:14 - WARNING - huggingface_hub.repository - Several commits (44) will be pushed upstream.
+02/26/2022 10:33:14 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 10:33:49 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   d24e10b..84c8fb0  floral-grass-11 -> floral-grass-11
+
+02/26/2022 10:33:53 - INFO - codeparrot_training - Step 44000: {'lr': 1.9030116872178316e-05, 'samples': 22528512, 'steps': 44000, 'loss/train': 2.3320531845092773}
+02/26/2022 10:33:59 - INFO - codeparrot_training - Step 44001: {'lr': 1.9023855723864002e-05, 'samples': 22529024, 'steps': 44001, 'loss/train': 1.6663116216659546}
+02/26/2022 10:34:02 - INFO - codeparrot_training - Step 44002: {'lr': 1.901759556497809e-05, 'samples': 22529536, 'steps': 44002, 'loss/train': 2.0551815032958984}
+02/26/2022 10:34:10 - INFO - codeparrot_training - Step 44003: {'lr': 1.90113363955475e-05, 'samples': 22530048, 'steps': 44003, 'loss/train': 1.423163890838623}
+02/26/2022 10:34:13 - INFO - codeparrot_training - Step 44004: {'lr': 1.900507821559899e-05, 'samples': 22530560, 'steps': 44004, 'loss/train': 1.4444445371627808}
+02/26/2022 10:34:19 - INFO - codeparrot_training - Step 44005: {'lr': 1.8998821025159375e-05, 'samples': 22531072, 'steps': 44005, 'loss/train': 2.4846723079681396}
+02/26/2022 10:34:22 - INFO - codeparrot_training - Step 44006: {'lr': 1.899256482425543e-05, 'samples': 22531584, 'steps': 44006, 'loss/train': 1.8023244142532349}
+02/26/2022 10:34:28 - INFO - codeparrot_training - Step 44007: {'lr': 1.8986309612914006e-05, 'samples': 22532096, 'steps': 44007, 'loss/train': 0.2879928946495056}
+02/26/2022 10:34:31 - INFO - codeparrot_training - Step 44008: {'lr': 1.8980055391161877e-05, 'samples': 22532608, 'steps': 44008, 'loss/train': 1.7266427278518677}
+02/26/2022 10:34:37 - INFO - codeparrot_training - Step 44009: {'lr': 1.8973802159025862e-05, 'samples': 22533120, 'steps': 44009, 'loss/train': 2.2361907958984375}
+02/26/2022 10:34:40 - INFO - codeparrot_training - Step 44010: {'lr': 1.8967549916532656e-05, 'samples': 22533632, 'steps': 44010, 'loss/train': 0.29530131816864014}
+02/26/2022 10:34:46 - INFO - codeparrot_training - Step 44011: {'lr': 1.8961298663709185e-05, 'samples': 22534144, 'steps': 44011, 'loss/train': 1.5945879220962524}
+02/26/2022 10:34:49 - INFO - codeparrot_training - Step 44012: {'lr': 1.895504840058207e-05, 'samples': 22534656, 'steps': 44012, 'loss/train': 1.5260460376739502}
+02/26/2022 10:34:55 - INFO - codeparrot_training - Step 44013: {'lr': 1.894879912717823e-05, 'samples': 22535168, 'steps': 44013, 'loss/train': 1.5213029384613037}
+02/26/2022 10:34:58 - INFO - codeparrot_training - Step 44014: {'lr': 1.894255084352428e-05, 'samples': 22535680, 'steps': 44014, 'loss/train': 3.3415493965148926}
+02/26/2022 10:35:04 - INFO - codeparrot_training - Step 44015: {'lr': 1.893630354964715e-05, 'samples': 22536192, 'steps': 44015, 'loss/train': 3.2358953952789307}
+02/26/2022 10:35:07 - INFO - codeparrot_training - Step 44016: {'lr': 1.8930057245573506e-05, 'samples': 22536704, 'steps': 44016, 'loss/train': 1.6172772645950317}
+02/26/2022 10:35:13 - INFO - codeparrot_training - Step 44017: {'lr': 1.8923811931330137e-05, 'samples': 22537216, 'steps': 44017, 'loss/train': 0.9790546894073486}
+02/26/2022 10:35:16 - INFO - codeparrot_training - Step 44018: {'lr': 1.891756760694374e-05, 'samples': 22537728, 'steps': 44018, 'loss/train': 2.160904884338379}
+02/26/2022 10:35:24 - INFO - codeparrot_training - Step 44019: {'lr': 1.8911324272441133e-05, 'samples': 22538240, 'steps': 44019, 'loss/train': 1.056497573852539}
+02/26/2022 10:35:27 - INFO - codeparrot_training - Step 44020: {'lr': 1.890508192784904e-05, 'samples': 22538752, 'steps': 44020, 'loss/train': 1.7328752279281616}
+02/26/2022 10:35:33 - INFO - codeparrot_training - Step 44021: {'lr': 1.8898840573194214e-05, 'samples': 22539264, 'steps': 44021, 'loss/train': 1.2247378826141357}
+02/26/2022 10:35:36 - INFO - codeparrot_training - Step 44022: {'lr': 1.889260020850331e-05, 'samples': 22539776, 'steps': 44022, 'loss/train': 3.085550546646118}
+02/26/2022 10:35:42 - INFO - codeparrot_training - Step 44023: {'lr': 1.8886360833803156e-05, 'samples': 22540288, 'steps': 44023, 'loss/train': 1.4534519910812378}
+02/26/2022 10:35:45 - INFO - codeparrot_training - Step 44024: {'lr': 1.8880122449120463e-05, 'samples': 22540800, 'steps': 44024, 'loss/train': 2.2866058349609375}
+02/26/2022 10:35:51 - INFO - codeparrot_training - Step 44025: {'lr': 1.8873885054481926e-05, 'samples': 22541312, 'steps': 44025, 'loss/train': 1.5168508291244507}
+02/26/2022 10:35:54 - INFO - codeparrot_training - Step 44026: {'lr': 1.886764864991425e-05, 'samples': 22541824, 'steps': 44026, 'loss/train': 2.1663818359375}
+02/26/2022 10:36:00 - INFO - codeparrot_training - Step 44027: {'lr': 1.8861413235444213e-05, 'samples': 22542336, 'steps': 44027, 'loss/train': 1.4665101766586304}
+02/26/2022 10:36:03 - INFO - codeparrot_training - Step 44028: {'lr': 1.885517881109841e-05, 'samples': 22542848, 'steps': 44028, 'loss/train': 2.0435917377471924}
+02/26/2022 10:36:11 - INFO - codeparrot_training - Step 44029: {'lr': 1.8848945376903737e-05, 'samples': 22543360, 'steps': 44029, 'loss/train': 0.6146186590194702}
+02/26/2022 10:36:15 - INFO - codeparrot_training - Step 44030: {'lr': 1.8842712932886697e-05, 'samples': 22543872, 'steps': 44030, 'loss/train': 2.2952003479003906}
+02/26/2022 10:36:18 - INFO - codeparrot_training - Step 44031: {'lr': 1.883648147907413e-05, 'samples': 22544384, 'steps': 44031, 'loss/train': 1.5369566679000854}
+02/26/2022 10:36:24 - INFO - codeparrot_training - Step 44032: {'lr': 1.88302510154926e-05, 'samples': 22544896, 'steps': 44032, 'loss/train': 1.8704774379730225}
+02/26/2022 10:36:27 - INFO - codeparrot_training - Step 44033: {'lr': 1.8824021542168973e-05, 'samples': 22545408, 'steps': 44033, 'loss/train': 1.8428127765655518}
+02/26/2022 10:36:33 - INFO - codeparrot_training - Step 44034: {'lr': 1.8817793059129756e-05, 'samples': 22545920, 'steps': 44034, 'loss/train': 2.1620123386383057}
+02/26/2022 10:36:36 - INFO - codeparrot_training - Step 44035: {'lr': 1.8811565566401735e-05, 'samples': 22546432, 'steps': 44035, 'loss/train': 2.19956374168396}
+02/26/2022 10:36:42 - INFO - codeparrot_training - Step 44036: {'lr': 1.8805339064011524e-05, 'samples': 22546944, 'steps': 44036, 'loss/train': 1.258091688156128}
+02/26/2022 10:36:45 - INFO - codeparrot_training - Step 44037: {'lr': 1.8799113551985852e-05, 'samples': 22547456, 'steps': 44037, 'loss/train': 2.4753246307373047}
+02/26/2022 10:36:53 - INFO - codeparrot_training - Step 44038: {'lr': 1.8792889030351367e-05, 'samples': 22547968, 'steps': 44038, 'loss/train': 1.0560418367385864}
+02/26/2022 10:36:56 - INFO - codeparrot_training - Step 44039: {'lr': 1.8786665499134742e-05, 'samples': 22548480, 'steps': 44039, 'loss/train': 4.3891167640686035}
+02/26/2022 10:37:02 - INFO - codeparrot_training - Step 44040: {'lr': 1.8780442958362566e-05, 'samples': 22548992, 'steps': 44040, 'loss/train': 1.6752862930297852}
+02/26/2022 10:37:06 - INFO - codeparrot_training - Step 44041: {'lr': 1.877422140806162e-05, 'samples': 22549504, 'steps': 44041, 'loss/train': 0.7909420728683472}
+02/26/2022 10:37:11 - INFO - codeparrot_training - Step 44042: {'lr': 1.8768000848258472e-05, 'samples': 22550016, 'steps': 44042, 'loss/train': 1.6012473106384277}
+02/26/2022 10:37:15 - INFO - codeparrot_training - Step 44043: {'lr': 1.8761781278979763e-05, 'samples': 22550528, 'steps': 44043, 'loss/train': 1.874677062034607}
+02/26/2022 10:37:20 - INFO - codeparrot_training - Step 44044: {'lr': 1.8755562700252193e-05, 'samples': 22551040, 'steps': 44044, 'loss/train': 1.1454931497573853}
+02/26/2022 10:37:24 - INFO - codeparrot_training - Step 44045: {'lr': 1.87493451121023e-05, 'samples': 22551552, 'steps': 44045, 'loss/train': 1.2387669086456299}
+02/26/2022 10:37:29 - INFO - codeparrot_training - Step 44046: {'lr': 1.874312851455681e-05, 'samples': 22552064, 'steps': 44046, 'loss/train': 2.014160394668579}
+02/26/2022 10:37:33 - INFO - codeparrot_training - Step 44047: {'lr': 1.8736912907642345e-05, 'samples': 22552576, 'steps': 44047, 'loss/train': 1.901646614074707}
+02/26/2022 10:37:39 - INFO - codeparrot_training - Step 44048: {'lr': 1.873069829138552e-05, 'samples': 22553088, 'steps': 44048, 'loss/train': 1.37613046169281}
+02/26/2022 10:37:42 - INFO - codeparrot_training - Step 44049: {'lr': 1.872448466581289e-05, 'samples': 22553600, 'steps': 44049, 'loss/train': 1.799850583076477}
+02/26/2022 10:37:49 - INFO - codeparrot_training - Step 44050: {'lr': 1.8718272030951167e-05, 'samples': 22554112, 'steps': 44050, 'loss/train': 1.1026482582092285}
+02/26/2022 10:37:53 - INFO - codeparrot_training - Step 44051: {'lr': 1.8712060386826934e-05, 'samples': 22554624, 'steps': 44051, 'loss/train': 1.2876883745193481}
+02/26/2022 10:37:58 - INFO - codeparrot_training - Step 44052: {'lr': 1.8705849733466812e-05, 'samples': 22555136, 'steps': 44052, 'loss/train': 1.4196891784667969}
+02/26/2022 10:38:02 - INFO - codeparrot_training - Step 44053: {'lr': 1.8699640070897307e-05, 'samples': 22555648, 'steps': 44053, 'loss/train': 1.5514488220214844}
+02/26/2022 10:38:07 - INFO - codeparrot_training - Step 44054: {'lr': 1.8693431399145172e-05, 'samples': 22556160, 'steps': 44054, 'loss/train': 2.8206305503845215}
+02/26/2022 10:38:11 - INFO - codeparrot_training - Step 44055: {'lr': 1.8687223718236918e-05, 'samples': 22556672, 'steps': 44055, 'loss/train': 1.5404722690582275}
+02/26/2022 10:38:16 - INFO - codeparrot_training - Step 44056: {'lr': 1.8681017028199132e-05, 'samples': 22557184, 'steps': 44056, 'loss/train': 2.372636556625366}
+02/26/2022 10:38:20 - INFO - codeparrot_training - Step 44057: {'lr': 1.8674811329058378e-05, 'samples': 22557696, 'steps': 44057, 'loss/train': 1.7107274532318115}
+02/26/2022 10:38:25 - INFO - codeparrot_training - Step 44058: {'lr': 1.8668606620841326e-05, 'samples': 22558208, 'steps': 44058, 'loss/train': 1.6024490594863892}
+02/26/2022 10:38:29 - INFO - codeparrot_training - Step 44059: {'lr': 1.8662402903574516e-05, 'samples': 22558720, 'steps': 44059, 'loss/train': 0.5669743418693542}
+02/26/2022 10:38:34 - INFO - codeparrot_training - Step 44060: {'lr': 1.8656200177284503e-05, 'samples': 22559232, 'steps': 44060, 'loss/train': 1.5877149105072021}
+02/26/2022 10:38:38 - INFO - codeparrot_training - Step 44061: {'lr': 1.8649998441997856e-05, 'samples': 22559744, 'steps': 44061, 'loss/train': 1.8641678094863892}
+02/26/2022 10:38:43 - INFO - codeparrot_training - Step 44062: {'lr': 1.8643797697741162e-05, 'samples': 22560256, 'steps': 44062, 'loss/train': 1.7686386108398438}
+02/26/2022 10:38:47 - INFO - codeparrot_training - Step 44063: {'lr': 1.8637597944540952e-05, 'samples': 22560768, 'steps': 44063, 'loss/train': 1.1887909173965454}
+02/26/2022 10:38:52 - INFO - codeparrot_training - Step 44064: {'lr': 1.8631399182423876e-05, 'samples': 22561280, 'steps': 44064, 'loss/train': 1.5086917877197266}
+02/26/2022 10:38:56 - INFO - codeparrot_training - Step 44065: {'lr': 1.8625201411416325e-05, 'samples': 22561792, 'steps': 44065, 'loss/train': 1.4407860040664673}
+02/26/2022 10:39:03 - INFO - codeparrot_training - Step 44066: {'lr': 1.8619004631545004e-05, 'samples': 22562304, 'steps': 44066, 'loss/train': 1.3726171255111694}
+02/26/2022 10:39:07 - INFO - codeparrot_training - Step 44067: {'lr': 1.8612808842836337e-05, 'samples': 22562816, 'steps': 44067, 'loss/train': 1.6653813123703003}
+02/26/2022 10:39:12 - INFO - codeparrot_training - Step 44068: {'lr': 1.8606614045317022e-05, 'samples': 22563328, 'steps': 44068, 'loss/train': 2.2861428260803223}
+02/26/2022 10:39:16 - INFO - codeparrot_training - Step 44069: {'lr': 1.86004202390134e-05, 'samples': 22563840, 'steps': 44069, 'loss/train': 2.1283395290374756}
+02/26/2022 10:39:21 - INFO - codeparrot_training - Step 44070: {'lr': 1.859422742395217e-05, 'samples': 22564352, 'steps': 44070, 'loss/train': 1.6871061325073242}
+02/26/2022 10:39:25 - INFO - codeparrot_training - Step 44071: {'lr': 1.858803560015973e-05, 'samples': 22564864, 'steps': 44071, 'loss/train': 2.382408380508423}
+02/26/2022 10:39:30 - INFO - codeparrot_training - Step 44072: {'lr': 1.8581844767662727e-05, 'samples': 22565376, 'steps': 44072, 'loss/train': 1.5408034324645996}
+02/26/2022 10:39:34 - INFO - codeparrot_training - Step 44073: {'lr': 1.857565492648755e-05, 'samples': 22565888, 'steps': 44073, 'loss/train': 1.791016936302185}
+02/26/2022 10:39:39 - INFO - codeparrot_training - Step 44074: {'lr': 1.8569466076660825e-05, 'samples': 22566400, 'steps': 44074, 'loss/train': 1.8555350303649902}
+02/26/2022 10:39:43 - INFO - codeparrot_training - Step 44075: {'lr': 1.8563278218208995e-05, 'samples': 22566912, 'steps': 44075, 'loss/train': 1.4032213687896729}
+02/26/2022 10:39:50 - INFO - codeparrot_training - Step 44076: {'lr': 1.85570913511586e-05, 'samples': 22567424, 'steps': 44076, 'loss/train': 3.2220664024353027}
+02/26/2022 10:39:53 - INFO - codeparrot_training - Step 44077: {'lr': 1.855090547553617e-05, 'samples': 22567936, 'steps': 44077, 'loss/train': 2.0979886054992676}
+02/26/2022 10:39:59 - INFO - codeparrot_training - Step 44078: {'lr': 1.8544720591368134e-05, 'samples': 22568448, 'steps': 44078, 'loss/train': 1.9048371315002441}
+02/26/2022 10:40:02 - INFO - codeparrot_training - Step 44079: {'lr': 1.8538536698680995e-05, 'samples': 22568960, 'steps': 44079, 'loss/train': 2.847787380218506}
+02/26/2022 10:40:08 - INFO - codeparrot_training - Step 44080: {'lr': 1.8532353797501317e-05, 'samples': 22569472, 'steps': 44080, 'loss/train': 1.921404480934143}
+02/26/2022 10:40:12 - INFO - codeparrot_training - Step 44081: {'lr': 1.852617188785552e-05, 'samples': 22569984, 'steps': 44081, 'loss/train': 1.736572504043579}
+02/26/2022 10:40:17 - INFO - codeparrot_training - Step 44082: {'lr': 1.8519990969770113e-05, 'samples': 22570496, 'steps': 44082, 'loss/train': 1.7315726280212402}
+02/26/2022 10:40:20 - INFO - codeparrot_training - Step 44083: {'lr': 1.8513811043271517e-05, 'samples': 22571008, 'steps': 44083, 'loss/train': 1.828129768371582}
+02/26/2022 10:40:26 - INFO - codeparrot_training - Step 44084: {'lr': 1.8507632108386268e-05, 'samples': 22571520, 'steps': 44084, 'loss/train': 1.5297503471374512}
+02/26/2022 10:40:29 - INFO - codeparrot_training - Step 44085: {'lr': 1.8501454165140847e-05, 'samples': 22572032, 'steps': 44085, 'loss/train': 1.3314677476882935}
+02/26/2022 10:40:37 - INFO - codeparrot_training - Step 44086: {'lr': 1.8495277213561674e-05, 'samples': 22572544, 'steps': 44086, 'loss/train': 0.475856214761734}
+02/26/2022 10:40:40 - INFO - codeparrot_training - Step 44087: {'lr': 1.84891012536752e-05, 'samples': 22573056, 'steps': 44087, 'loss/train': 0.9418395757675171}
+02/26/2022 10:40:46 - INFO - codeparrot_training - Step 44088: {'lr': 1.8482926285507905e-05, 'samples': 22573568, 'steps': 44088, 'loss/train': 4.145501136779785}
+02/26/2022 10:40:49 - INFO - codeparrot_training - Step 44089: {'lr': 1.847675230908627e-05, 'samples': 22574080, 'steps': 44089, 'loss/train': 3.4087753295898438}
+02/26/2022 10:40:55 - INFO - codeparrot_training - Step 44090: {'lr': 1.8470579324436714e-05, 'samples': 22574592, 'steps': 44090, 'loss/train': 1.4923803806304932}
+02/26/2022 10:40:58 - INFO - codeparrot_training - Step 44091: {'lr': 1.8464407331585665e-05, 'samples': 22575104, 'steps': 44091, 'loss/train': 1.3637189865112305}
+02/26/2022 10:41:04 - INFO - codeparrot_training - Step 44092: {'lr': 1.845823633055954e-05, 'samples': 22575616, 'steps': 44092, 'loss/train': 0.42889466881752014}
+02/26/2022 10:41:07 - INFO - codeparrot_training - Step 44093: {'lr': 1.8452066321384825e-05, 'samples': 22576128, 'steps': 44093, 'loss/train': 0.9857776165008545}
+02/26/2022 10:41:13 - INFO - codeparrot_training - Step 44094: {'lr': 1.8445897304087966e-05, 'samples': 22576640, 'steps': 44094, 'loss/train': 1.7164846658706665}
+02/26/2022 10:41:17 - INFO - codeparrot_training - Step 44095: {'lr': 1.8439729278695332e-05, 'samples': 22577152, 'steps': 44095, 'loss/train': 1.7889143228530884}
+02/26/2022 10:41:24 - INFO - codeparrot_training - Step 44096: {'lr': 1.843356224523335e-05, 'samples': 22577664, 'steps': 44096, 'loss/train': 1.4949747323989868}
+02/26/2022 10:41:27 - INFO - codeparrot_training - Step 44097: {'lr': 1.8427396203728464e-05, 'samples': 22578176, 'steps': 44097, 'loss/train': 1.7959195375442505}
+02/26/2022 10:41:33 - INFO - codeparrot_training - Step 44098: {'lr': 1.8421231154207075e-05, 'samples': 22578688, 'steps': 44098, 'loss/train': 2.374171257019043}
+02/26/2022 10:41:36 - INFO - codeparrot_training - Step 44099: {'lr': 1.841506709669566e-05, 'samples': 22579200, 'steps': 44099, 'loss/train': 2.2732620239257812}
+02/26/2022 10:41:42 - INFO - codeparrot_training - Step 44100: {'lr': 1.8408904031220476e-05, 'samples': 22579712, 'steps': 44100, 'loss/train': 2.139552354812622}
+02/26/2022 10:41:45 - INFO - codeparrot_training - Step 44101: {'lr': 1.8402741957808056e-05, 'samples': 22580224, 'steps': 44101, 'loss/train': 2.0501139163970947}
+02/26/2022 10:41:51 - INFO - codeparrot_training - Step 44102: {'lr': 1.8396580876484714e-05, 'samples': 22580736, 'steps': 44102, 'loss/train': 2.0415778160095215}
+02/26/2022 10:41:54 - INFO - codeparrot_training - Step 44103: {'lr': 1.8390420787276955e-05, 'samples': 22581248, 'steps': 44103, 'loss/train': 1.4672152996063232}
+02/26/2022 10:42:00 - INFO - codeparrot_training - Step 44104: {'lr': 1.8384261690211014e-05, 'samples': 22581760, 'steps': 44104, 'loss/train': 2.853710174560547}
+02/26/2022 10:42:03 - INFO - codeparrot_training - Step 44105: {'lr': 1.8378103585313388e-05, 'samples': 22582272, 'steps': 44105, 'loss/train': 1.8300520181655884}
+02/26/2022 10:42:09 - INFO - codeparrot_training - Step 44106: {'lr': 1.837194647261037e-05, 'samples': 22582784, 'steps': 44106, 'loss/train': 1.7034151554107666}
+02/26/2022 10:42:12 - INFO - codeparrot_training - Step 44107: {'lr': 1.8365790352128458e-05, 'samples': 22583296, 'steps': 44107, 'loss/train': 8.701147079467773}
+02/26/2022 10:42:18 - INFO - codeparrot_training - Step 44108: {'lr': 1.835963522389389e-05, 'samples': 22583808, 'steps': 44108, 'loss/train': 1.4534105062484741}
+02/26/2022 10:42:21 - INFO - codeparrot_training - Step 44109: {'lr': 1.835348108793311e-05, 'samples': 22584320, 'steps': 44109, 'loss/train': 2.158106565475464}
+02/26/2022 10:42:27 - INFO - codeparrot_training - Step 44110: {'lr': 1.8347327944272435e-05, 'samples': 22584832, 'steps': 44110, 'loss/train': 1.9505138397216797}
+02/26/2022 10:42:30 - INFO - codeparrot_training - Step 44111: {'lr': 1.8341175792938287e-05, 'samples': 22585344, 'steps': 44111, 'loss/train': 0.720163881778717}
+02/26/2022 10:42:38 - INFO - codeparrot_training - Step 44112: {'lr': 1.8335024633956977e-05, 'samples': 22585856, 'steps': 44112, 'loss/train': 1.6401547193527222}
+02/26/2022 10:42:41 - INFO - codeparrot_training - Step 44113: {'lr': 1.8328874467354846e-05, 'samples': 22586368, 'steps': 44113, 'loss/train': 2.9826443195343018}
+02/26/2022 10:42:47 - INFO - codeparrot_training - Step 44114: {'lr': 1.8322725293158233e-05, 'samples': 22586880, 'steps': 44114, 'loss/train': 1.5513803958892822}
+02/26/2022 10:42:50 - INFO - codeparrot_training - Step 44115: {'lr': 1.8316577111393534e-05, 'samples': 22587392, 'steps': 44115, 'loss/train': 1.8280835151672363}
+02/26/2022 10:42:56 - INFO - codeparrot_training - Step 44116: {'lr': 1.8310429922087035e-05, 'samples': 22587904, 'steps': 44116, 'loss/train': 3.183056592941284}
+02/26/2022 10:42:59 - INFO - codeparrot_training - Step 44117: {'lr': 1.8304283725265104e-05, 'samples': 22588416, 'steps': 44117, 'loss/train': 2.2480101585388184}
+02/26/2022 10:43:05 - INFO - codeparrot_training - Step 44118: {'lr': 1.8298138520953995e-05, 'samples': 22588928, 'steps': 44118, 'loss/train': 1.4812006950378418}
+02/26/2022 10:43:08 - INFO - codeparrot_training - Step 44119: {'lr': 1.829199430918016e-05, 'samples': 22589440, 'steps': 44119, 'loss/train': 2.079787015914917}
+02/26/2022 10:43:14 - INFO - codeparrot_training - Step 44120: {'lr': 1.8285851089969803e-05, 'samples': 22589952, 'steps': 44120, 'loss/train': 2.3262670040130615}
+02/26/2022 10:43:17 - INFO - codeparrot_training - Step 44121: {'lr': 1.8279708863349315e-05, 'samples': 22590464, 'steps': 44121, 'loss/train': 2.307448387145996}
+02/26/2022 10:43:24 - INFO - codeparrot_training - Step 44122: {'lr': 1.8273567629344928e-05, 'samples': 22590976, 'steps': 44122, 'loss/train': 3.5219831466674805}
+02/26/2022 10:43:28 - INFO - codeparrot_training - Step 44123: {'lr': 1.826742738798301e-05, 'samples': 22591488, 'steps': 44123, 'loss/train': 2.1535136699676514}
+02/26/2022 10:43:33 - INFO - codeparrot_training - Step 44124: {'lr': 1.8261288139289874e-05, 'samples': 22592000, 'steps': 44124, 'loss/train': 1.588499903678894}
+02/26/2022 10:43:37 - INFO - codeparrot_training - Step 44125: {'lr': 1.82551498832918e-05, 'samples': 22592512, 'steps': 44125, 'loss/train': 2.4162540435791016}
+02/26/2022 10:43:42 - INFO - codeparrot_training - Step 44126: {'lr': 1.824901262001502e-05, 'samples': 22593024, 'steps': 44126, 'loss/train': 2.277219772338867}
+02/26/2022 10:43:46 - INFO - codeparrot_training - Step 44127: {'lr': 1.8242876349485932e-05, 'samples': 22593536, 'steps': 44127, 'loss/train': 1.1755133867263794}
+02/26/2022 10:43:51 - INFO - codeparrot_training - Step 44128: {'lr': 1.823674107173076e-05, 'samples': 22594048, 'steps': 44128, 'loss/train': 0.45689237117767334}
+02/26/2022 10:43:55 - INFO - codeparrot_training - Step 44129: {'lr': 1.8230606786775793e-05, 'samples': 22594560, 'steps': 44129, 'loss/train': 2.6282219886779785}
+02/26/2022 10:44:01 - INFO - codeparrot_training - Step 44130: {'lr': 1.8224473494647286e-05, 'samples': 22595072, 'steps': 44130, 'loss/train': 1.2036617994308472}
+02/26/2022 10:44:04 - INFO - codeparrot_training - Step 44131: {'lr': 1.8218341195371578e-05, 'samples': 22595584, 'steps': 44131, 'loss/train': 1.94817316532135}
+02/26/2022 10:44:11 - INFO - codeparrot_training - Step 44132: {'lr': 1.8212209888974874e-05, 'samples': 22596096, 'steps': 44132, 'loss/train': 3.273123264312744}
+02/26/2022 10:44:15 - INFO - codeparrot_training - Step 44133: {'lr': 1.8206079575483482e-05, 'samples': 22596608, 'steps': 44133, 'loss/train': 0.9376332759857178}
+02/26/2022 10:44:20 - INFO - codeparrot_training - Step 44134: {'lr': 1.8199950254923603e-05, 'samples': 22597120, 'steps': 44134, 'loss/train': 2.055082321166992}
+02/26/2022 10:44:24 - INFO - codeparrot_training - Step 44135: {'lr': 1.819382192732158e-05, 'samples': 22597632, 'steps': 44135, 'loss/train': 2.9164419174194336}
+02/26/2022 10:44:29 - INFO - codeparrot_training - Step 44136: {'lr': 1.8187694592703586e-05, 'samples': 22598144, 'steps': 44136, 'loss/train': 1.5746794939041138}
+02/26/2022 10:44:33 - INFO - codeparrot_training - Step 44137: {'lr': 1.8181568251095875e-05, 'samples': 22598656, 'steps': 44137, 'loss/train': 2.1007883548736572}
+02/26/2022 10:44:38 - INFO - codeparrot_training - Step 44138: {'lr': 1.817544290252482e-05, 'samples': 22599168, 'steps': 44138, 'loss/train': 1.6616401672363281}
+02/26/2022 10:44:42 - INFO - codeparrot_training - Step 44139: {'lr': 1.816931854701645e-05, 'samples': 22599680, 'steps': 44139, 'loss/train': 3.0018463134765625}
+02/26/2022 10:44:47 - INFO - codeparrot_training - Step 44140: {'lr': 1.8163195184597138e-05, 'samples': 22600192, 'steps': 44140, 'loss/train': 1.820244550704956}
+02/26/2022 10:44:51 - INFO - codeparrot_training - Step 44141: {'lr': 1.8157072815293053e-05, 'samples': 22600704, 'steps': 44141, 'loss/train': 1.670430064201355}
+02/26/2022 10:44:58 - INFO - codeparrot_training - Step 44142: {'lr': 1.8150951439130535e-05, 'samples': 22601216, 'steps': 44142, 'loss/train': 0.9679425954818726}
+02/26/2022 10:45:01 - INFO - codeparrot_training - Step 44143: {'lr': 1.8144831056135626e-05, 'samples': 22601728, 'steps': 44143, 'loss/train': 1.1052898168563843}
+02/26/2022 10:45:07 - INFO - codeparrot_training - Step 44144: {'lr': 1.8138711666334683e-05, 'samples': 22602240, 'steps': 44144, 'loss/train': 1.3196940422058105}
+02/26/2022 10:45:10 - INFO - codeparrot_training - Step 44145: {'lr': 1.8132593269753862e-05, 'samples': 22602752, 'steps': 44145, 'loss/train': 2.3234992027282715}
+02/26/2022 10:45:16 - INFO - codeparrot_training - Step 44146: {'lr': 1.8126475866419385e-05, 'samples': 22603264, 'steps': 44146, 'loss/train': 1.489924430847168}
+02/26/2022 10:45:19 - INFO - codeparrot_training - Step 44147: {'lr': 1.8120359456357484e-05, 'samples': 22603776, 'steps': 44147, 'loss/train': 1.3693950176239014}
+02/26/2022 10:45:25 - INFO - codeparrot_training - Step 44148: {'lr': 1.811424403959433e-05, 'samples': 22604288, 'steps': 44148, 'loss/train': 1.1188582181930542}
+02/26/2022 10:45:28 - INFO - codeparrot_training - Step 44149: {'lr': 1.8108129616156072e-05, 'samples': 22604800, 'steps': 44149, 'loss/train': 1.974057912826538}
+02/26/2022 10:45:34 - INFO - codeparrot_training - Step 44150: {'lr': 1.8102016186068992e-05, 'samples': 22605312, 'steps': 44150, 'loss/train': 1.701819896697998}
+02/26/2022 10:45:37 - INFO - codeparrot_training - Step 44151: {'lr': 1.809590374935924e-05, 'samples': 22605824, 'steps': 44151, 'loss/train': 1.7709394693374634}
+02/26/2022 10:45:43 - INFO - codeparrot_training - Step 44152: {'lr': 1.808979230605301e-05, 'samples': 22606336, 'steps': 44152, 'loss/train': 1.6412721872329712}
+02/26/2022 10:45:48 - INFO - codeparrot_training - Step 44153: {'lr': 1.80836818561764e-05, 'samples': 22606848, 'steps': 44153, 'loss/train': 1.0612002611160278}
+02/26/2022 10:45:52 - INFO - codeparrot_training - Step 44154: {'lr': 1.8077572399755717e-05, 'samples': 22607360, 'steps': 44154, 'loss/train': 2.0660037994384766}
+02/26/2022 10:45:55 - INFO - codeparrot_training - Step 44155: {'lr': 1.8071463936817078e-05, 'samples': 22607872, 'steps': 44155, 'loss/train': 1.5193873643875122}
+02/26/2022 10:46:01 - INFO - codeparrot_training - Step 44156: {'lr': 1.8065356467386635e-05, 'samples': 22608384, 'steps': 44156, 'loss/train': 1.4883440732955933}
+02/26/2022 10:46:04 - INFO - codeparrot_training - Step 44157: {'lr': 1.805924999149053e-05, 'samples': 22608896, 'steps': 44157, 'loss/train': 1.6660815477371216}
+02/26/2022 10:46:12 - INFO - codeparrot_training - Step 44158: {'lr': 1.8053144509154963e-05, 'samples': 22609408, 'steps': 44158, 'loss/train': 1.0225225687026978}
+02/26/2022 10:46:15 - INFO - codeparrot_training - Step 44159: {'lr': 1.804704002040605e-05, 'samples': 22609920, 'steps': 44159, 'loss/train': 2.331615686416626}
+02/26/2022 10:46:21 - INFO - codeparrot_training - Step 44160: {'lr': 1.804093652527003e-05, 'samples': 22610432, 'steps': 44160, 'loss/train': 0.13454057276248932}
+02/26/2022 10:46:26 - INFO - codeparrot_training - Step 44161: {'lr': 1.8034834023772927e-05, 'samples': 22610944, 'steps': 44161, 'loss/train': 1.656209945678711}
+02/26/2022 10:46:30 - INFO - codeparrot_training - Step 44162: {'lr': 1.8028732515940948e-05, 'samples': 22611456, 'steps': 44162, 'loss/train': 1.6359809637069702}
+02/26/2022 10:46:35 - INFO - codeparrot_training - Step 44163: {'lr': 1.802263200180021e-05, 'samples': 22611968, 'steps': 44163, 'loss/train': 1.8141642808914185}
+02/26/2022 10:46:39 - INFO - codeparrot_training - Step 44164: {'lr': 1.801653248137691e-05, 'samples': 22612480, 'steps': 44164, 'loss/train': 0.5752533078193665}
+02/26/2022 10:46:42 - INFO - codeparrot_training - Step 44165: {'lr': 1.8010433954697065e-05, 'samples': 22612992, 'steps': 44165, 'loss/train': 0.09879948198795319}
+02/26/2022 10:46:48 - INFO - codeparrot_training - Step 44166: {'lr': 1.8004336421786864e-05, 'samples': 22613504, 'steps': 44166, 'loss/train': 2.1428003311157227}
+02/26/2022 10:46:55 - INFO - codeparrot_training - Step 44167: {'lr': 1.7998239882672407e-05, 'samples': 22614016, 'steps': 44167, 'loss/train': 1.5720183849334717}
+02/26/2022 10:46:59 - INFO - codeparrot_training - Step 44168: {'lr': 1.799214433737989e-05, 'samples': 22614528, 'steps': 44168, 'loss/train': 1.9422537088394165}
+02/26/2022 10:47:04 - INFO - codeparrot_training - Step 44169: {'lr': 1.7986049785935266e-05, 'samples': 22615040, 'steps': 44169, 'loss/train': 1.6239545345306396}
+02/26/2022 10:47:07 - INFO - codeparrot_training - Step 44170: {'lr': 1.7979956228364792e-05, 'samples': 22615552, 'steps': 44170, 'loss/train': 1.8635287284851074}
+02/26/2022 10:47:13 - INFO - codeparrot_training - Step 44171: {'lr': 1.7973863664694502e-05, 'samples': 22616064, 'steps': 44171, 'loss/train': 1.7481529712677002}
+02/26/2022 10:47:17 - INFO - codeparrot_training - Step 44172: {'lr': 1.7967772094950456e-05, 'samples': 22616576, 'steps': 44172, 'loss/train': 1.626937985420227}
+02/26/2022 10:47:20 - INFO - codeparrot_training - Step 44173: {'lr': 1.7961681519158863e-05, 'samples': 22617088, 'steps': 44173, 'loss/train': 1.8340483903884888}
+02/26/2022 10:47:26 - INFO - codeparrot_training - Step 44174: {'lr': 1.795559193734572e-05, 'samples': 22617600, 'steps': 44174, 'loss/train': 2.196171283721924}
+02/26/2022 10:47:29 - INFO - codeparrot_training - Step 44175: {'lr': 1.7949503349537154e-05, 'samples': 22618112, 'steps': 44175, 'loss/train': 1.8305608034133911}
+02/26/2022 10:47:35 - INFO - codeparrot_training - Step 44176: {'lr': 1.7943415755759167e-05, 'samples': 22618624, 'steps': 44176, 'loss/train': 0.08788692206144333}
+02/26/2022 10:47:42 - INFO - codeparrot_training - Step 44177: {'lr': 1.793732915603796e-05, 'samples': 22619136, 'steps': 44177, 'loss/train': 2.0525529384613037}
+02/26/2022 10:47:45 - INFO - codeparrot_training - Step 44178: {'lr': 1.7931243550399546e-05, 'samples': 22619648, 'steps': 44178, 'loss/train': 1.7399303913116455}
+02/26/2022 10:47:51 - INFO - codeparrot_training - Step 44179: {'lr': 1.7925158938870005e-05, 'samples': 22620160, 'steps': 44179, 'loss/train': 1.8708152770996094}
+02/26/2022 10:47:54 - INFO - codeparrot_training - Step 44180: {'lr': 1.7919075321475327e-05, 'samples': 22620672, 'steps': 44180, 'loss/train': 2.30462646484375}
+02/26/2022 10:48:00 - INFO - codeparrot_training - Step 44181: {'lr': 1.7912992698241677e-05, 'samples': 22621184, 'steps': 44181, 'loss/train': 1.6517940759658813}
+02/26/2022 10:48:03 - INFO - codeparrot_training - Step 44182: {'lr': 1.7906911069195094e-05, 'samples': 22621696, 'steps': 44182, 'loss/train': 1.805814504623413}
+02/26/2022 10:48:09 - INFO - codeparrot_training - Step 44183: {'lr': 1.790083043436158e-05, 'samples': 22622208, 'steps': 44183, 'loss/train': 2.0057175159454346}
+02/26/2022 10:48:12 - INFO - codeparrot_training - Step 44184: {'lr': 1.789475079376718e-05, 'samples': 22622720, 'steps': 44184, 'loss/train': 1.5925467014312744}
+02/26/2022 10:48:18 - INFO - codeparrot_training - Step 44185: {'lr': 1.7888672147438028e-05, 'samples': 22623232, 'steps': 44185, 'loss/train': 2.3879640102386475}
+02/26/2022 10:48:21 - INFO - codeparrot_training - Step 44186: {'lr': 1.7882594495400058e-05, 'samples': 22623744, 'steps': 44186, 'loss/train': 1.004015326499939}
+02/26/2022 10:48:28 - INFO - codeparrot_training - Step 44187: {'lr': 1.787651783767938e-05, 'samples': 22624256, 'steps': 44187, 'loss/train': 2.11611270904541}
+02/26/2022 10:48:32 - INFO - codeparrot_training - Step 44188: {'lr': 1.7870442174301948e-05, 'samples': 22624768, 'steps': 44188, 'loss/train': 0.9738661050796509}
+02/26/2022 10:48:37 - INFO - codeparrot_training - Step 44189: {'lr': 1.7864367505293853e-05, 'samples': 22625280, 'steps': 44189, 'loss/train': 2.6405553817749023}
+02/26/2022 10:48:41 - INFO - codeparrot_training - Step 44190: {'lr': 1.7858293830681104e-05, 'samples': 22625792, 'steps': 44190, 'loss/train': 2.2659103870391846}
+02/26/2022 10:48:46 - INFO - codeparrot_training - Step 44191: {'lr': 1.7852221150489734e-05, 'samples': 22626304, 'steps': 44191, 'loss/train': 2.840360641479492}
+02/26/2022 10:48:50 - INFO - codeparrot_training - Step 44192: {'lr': 1.7846149464745666e-05, 'samples': 22626816, 'steps': 44192, 'loss/train': 1.9779052734375}
+02/26/2022 10:48:56 - INFO - codeparrot_training - Step 44193: {'lr': 1.7840078773475016e-05, 'samples': 22627328, 'steps': 44193, 'loss/train': 2.1348936557769775}
+02/26/2022 10:48:59 - INFO - codeparrot_training - Step 44194: {'lr': 1.783400907670374e-05, 'samples': 22627840, 'steps': 44194, 'loss/train': 2.149169683456421}
+02/26/2022 10:49:04 - INFO - codeparrot_training - Step 44195: {'lr': 1.7827940374457896e-05, 'samples': 22628352, 'steps': 44195, 'loss/train': 1.5813424587249756}
+02/26/2022 10:49:08 - INFO - codeparrot_training - Step 44196: {'lr': 1.7821872666763356e-05, 'samples': 22628864, 'steps': 44196, 'loss/train': 1.56320059299469}
+02/26/2022 10:49:14 - INFO - codeparrot_training - Step 44197: {'lr': 1.7815805953646237e-05, 'samples': 22629376, 'steps': 44197, 'loss/train': 0.8962571024894714}
+02/26/2022 10:49:17 - INFO - codeparrot_training - Step 44198: {'lr': 1.780974023513246e-05, 'samples': 22629888, 'steps': 44198, 'loss/train': 1.8402847051620483}
+02/26/2022 10:49:23 - INFO - codeparrot_training - Step 44199: {'lr': 1.780367551124809e-05, 'samples': 22630400, 'steps': 44199, 'loss/train': 2.4448423385620117}
+02/26/2022 10:49:26 - INFO - codeparrot_training - Step 44200: {'lr': 1.779761178201894e-05, 'samples': 22630912, 'steps': 44200, 'loss/train': 2.2313296794891357}
+02/26/2022 10:49:32 - INFO - codeparrot_training - Step 44201: {'lr': 1.7791549047471157e-05, 'samples': 22631424, 'steps': 44201, 'loss/train': 1.9121071100234985}
+02/26/2022 10:49:35 - INFO - codeparrot_training - Step 44202: {'lr': 1.7785487307630578e-05, 'samples': 22631936, 'steps': 44202, 'loss/train': 1.7731974124908447}
+02/26/2022 10:49:43 - INFO - codeparrot_training - Step 44203: {'lr': 1.7779426562523348e-05, 'samples': 22632448, 'steps': 44203, 'loss/train': 2.4096925258636475}
+02/26/2022 10:49:47 - INFO - codeparrot_training - Step 44204: {'lr': 1.7773366812175202e-05, 'samples': 22632960, 'steps': 44204, 'loss/train': 1.598775029182434}
+02/26/2022 10:49:52 - INFO - codeparrot_training - Step 44205: {'lr': 1.776730805661228e-05, 'samples': 22633472, 'steps': 44205, 'loss/train': 2.3445546627044678}
+02/26/2022 10:49:55 - INFO - codeparrot_training - Step 44206: {'lr': 1.7761250295860425e-05, 'samples': 22633984, 'steps': 44206, 'loss/train': 1.988905429840088}
+02/26/2022 10:50:01 - INFO - codeparrot_training - Step 44207: {'lr': 1.7755193529945645e-05, 'samples': 22634496, 'steps': 44207, 'loss/train': 1.450088620185852}
+02/26/2022 10:50:04 - INFO - codeparrot_training - Step 44208: {'lr': 1.774913775889389e-05, 'samples': 22635008, 'steps': 44208, 'loss/train': 1.1929043531417847}
+02/26/2022 10:50:10 - INFO - codeparrot_training - Step 44209: {'lr': 1.7743082982731085e-05, 'samples': 22635520, 'steps': 44209, 'loss/train': 2.1798148155212402}
+02/26/2022 10:50:13 - INFO - codeparrot_training - Step 44210: {'lr': 1.7737029201483123e-05, 'samples': 22636032, 'steps': 44210, 'loss/train': 1.9419649839401245}
+02/26/2022 10:50:19 - INFO - codeparrot_training - Step 44211: {'lr': 1.7730976415176016e-05, 'samples': 22636544, 'steps': 44211, 'loss/train': 2.2358546257019043}
+02/26/2022 10:50:22 - INFO - codeparrot_training - Step 44212: {'lr': 1.7724924623835625e-05, 'samples': 22637056, 'steps': 44212, 'loss/train': 1.6401432752609253}
+02/26/2022 10:50:30 - INFO - codeparrot_training - Step 44213: {'lr': 1.7718873827487937e-05, 'samples': 22637568, 'steps': 44213, 'loss/train': 1.1242268085479736}
+02/26/2022 10:50:33 - INFO - codeparrot_training - Step 44214: {'lr': 1.771282402615876e-05, 'samples': 22638080, 'steps': 44214, 'loss/train': 0.7484341263771057}
+02/26/2022 10:50:39 - INFO - codeparrot_training - Step 44215: {'lr': 1.7706775219874154e-05, 'samples': 22638592, 'steps': 44215, 'loss/train': 2.0688838958740234}
+02/26/2022 10:50:42 - INFO - codeparrot_training - Step 44216: {'lr': 1.770072740865997e-05, 'samples': 22639104, 'steps': 44216, 'loss/train': 0.730161726474762}
+02/26/2022 10:50:48 - INFO - codeparrot_training - Step 44217: {'lr': 1.769468059254206e-05, 'samples': 22639616, 'steps': 44217, 'loss/train': 2.212062120437622}
+02/26/2022 10:50:51 - INFO - codeparrot_training - Step 44218: {'lr': 1.768863477154642e-05, 'samples': 22640128, 'steps': 44218, 'loss/train': 2.183596611022949}
+02/26/2022 10:50:57 - INFO - codeparrot_training - Step 44219: {'lr': 1.768258994569885e-05, 'samples': 22640640, 'steps': 44219, 'loss/train': 0.9215399622917175}
+02/26/2022 10:51:00 - INFO - codeparrot_training - Step 44220: {'lr': 1.767654611502531e-05, 'samples': 22641152, 'steps': 44220, 'loss/train': 1.0041439533233643}
+02/26/2022 10:51:06 - INFO - codeparrot_training - Step 44221: {'lr': 1.7670503279551692e-05, 'samples': 22641664, 'steps': 44221, 'loss/train': 2.5924010276794434}
+02/26/2022 10:51:09 - INFO - codeparrot_training - Step 44222: {'lr': 1.766446143930389e-05, 'samples': 22642176, 'steps': 44222, 'loss/train': 2.2745039463043213}
+02/26/2022 10:51:15 - INFO - codeparrot_training - Step 44223: {'lr': 1.7658420594307696e-05, 'samples': 22642688, 'steps': 44223, 'loss/train': 2.3978307247161865}
+02/26/2022 10:51:19 - INFO - codeparrot_training - Step 44224: {'lr': 1.7652380744589082e-05, 'samples': 22643200, 'steps': 44224, 'loss/train': 1.124882698059082}
+02/26/2022 10:51:26 - INFO - codeparrot_training - Step 44225: {'lr': 1.7646341890173917e-05, 'samples': 22643712, 'steps': 44225, 'loss/train': 1.7671010494232178}
+02/26/2022 10:51:29 - INFO - codeparrot_training - Step 44226: {'lr': 1.7640304031088015e-05, 'samples': 22644224, 'steps': 44226, 'loss/train': 0.977418839931488}
+02/26/2022 10:51:34 - INFO - codeparrot_training - Step 44227: {'lr': 1.763426716735725e-05, 'samples': 22644736, 'steps': 44227, 'loss/train': 0.3391740322113037}
+02/26/2022 10:51:38 - INFO - codeparrot_training - Step 44228: {'lr': 1.7628231299007536e-05, 'samples': 22645248, 'steps': 44228, 'loss/train': 2.1769115924835205}
+02/26/2022 10:51:43 - INFO - codeparrot_training - Step 44229: {'lr': 1.762219642606466e-05, 'samples': 22645760, 'steps': 44229, 'loss/train': 1.501988172531128}
+02/26/2022 10:51:47 - INFO - codeparrot_training - Step 44230: {'lr': 1.7616162548554554e-05, 'samples': 22646272, 'steps': 44230, 'loss/train': 2.473156213760376}
+02/26/2022 10:51:52 - INFO - codeparrot_training - Step 44231: {'lr': 1.761012966650294e-05, 'samples': 22646784, 'steps': 44231, 'loss/train': 1.0713082551956177}
+02/26/2022 10:51:56 - INFO - codeparrot_training - Step 44232: {'lr': 1.7604097779935796e-05, 'samples': 22647296, 'steps': 44232, 'loss/train': 2.0067172050476074}
+02/26/2022 10:52:01 - INFO - codeparrot_training - Step 44233: {'lr': 1.7598066888878883e-05, 'samples': 22647808, 'steps': 44233, 'loss/train': 1.6761754751205444}
+02/26/2022 10:52:05 - INFO - codeparrot_training - Step 44234: {'lr': 1.7592036993358094e-05, 'samples': 22648320, 'steps': 44234, 'loss/train': 1.9190446138381958}
+02/26/2022 10:52:12 - INFO - codeparrot_training - Step 44235: {'lr': 1.7586008093399165e-05, 'samples': 22648832, 'steps': 44235, 'loss/train': 2.0201337337493896}
+02/26/2022 10:52:15 - INFO - codeparrot_training - Step 44236: {'lr': 1.757998018902804e-05, 'samples': 22649344, 'steps': 44236, 'loss/train': 1.665717363357544}
+02/26/2022 10:52:21 - INFO - codeparrot_training - Step 44237: {'lr': 1.7573953280270395e-05, 'samples': 22649856, 'steps': 44237, 'loss/train': 1.8131917715072632}
+02/26/2022 10:52:25 - INFO - codeparrot_training - Step 44238: {'lr': 1.7567927367152237e-05, 'samples': 22650368, 'steps': 44238, 'loss/train': 2.357252359390259}
+02/26/2022 10:52:30 - INFO - codeparrot_training - Step 44239: {'lr': 1.7561902449699184e-05, 'samples': 22650880, 'steps': 44239, 'loss/train': 1.7787820100784302}
+02/26/2022 10:52:34 - INFO - codeparrot_training - Step 44240: {'lr': 1.7555878527937163e-05, 'samples': 22651392, 'steps': 44240, 'loss/train': 2.1327555179595947}
+02/26/2022 10:52:39 - INFO - codeparrot_training - Step 44241: {'lr': 1.754985560189193e-05, 'samples': 22651904, 'steps': 44241, 'loss/train': 0.6350491642951965}
+02/26/2022 10:52:43 - INFO - codeparrot_training - Step 44242: {'lr': 1.754383367158932e-05, 'samples': 22652416, 'steps': 44242, 'loss/train': 1.1296526193618774}
+02/26/2022 10:52:48 - INFO - codeparrot_training - Step 44243: {'lr': 1.7537812737055124e-05, 'samples': 22652928, 'steps': 44243, 'loss/train': 1.479059100151062}
+02/26/2022 10:52:52 - INFO - codeparrot_training - Step 44244: {'lr': 1.7531792798315128e-05, 'samples': 22653440, 'steps': 44244, 'loss/train': 1.8070179224014282}
+02/26/2022 10:52:57 - INFO - codeparrot_training - Step 44245: {'lr': 1.7525773855395084e-05, 'samples': 22653952, 'steps': 44245, 'loss/train': 2.6320652961730957}
+02/26/2022 10:53:01 - INFO - codeparrot_training - Step 44246: {'lr': 1.7519755908320834e-05, 'samples': 22654464, 'steps': 44246, 'loss/train': 0.6372883915901184}
+02/26/2022 10:53:06 - INFO - codeparrot_training - Step 44247: {'lr': 1.7513738957118135e-05, 'samples': 22654976, 'steps': 44247, 'loss/train': 1.9252136945724487}
+02/26/2022 10:53:10 - INFO - codeparrot_training - Step 44248: {'lr': 1.7507723001812747e-05, 'samples': 22655488, 'steps': 44248, 'loss/train': 1.052933692932129}
+02/26/2022 10:53:15 - INFO - codeparrot_training - Step 44249: {'lr': 1.75017080424304e-05, 'samples': 22656000, 'steps': 44249, 'loss/train': 1.8560402393341064}
+02/26/2022 10:53:19 - INFO - codeparrot_training - Step 44250: {'lr': 1.7495694078996982e-05, 'samples': 22656512, 'steps': 44250, 'loss/train': 0.8501306176185608}
+02/26/2022 10:53:26 - INFO - codeparrot_training - Step 44251: {'lr': 1.7489681111538147e-05, 'samples': 22657024, 'steps': 44251, 'loss/train': 1.3095051050186157}
+02/26/2022 10:53:30 - INFO - codeparrot_training - Step 44252: {'lr': 1.7483669140079705e-05, 'samples': 22657536, 'steps': 44252, 'loss/train': 1.2871676683425903}
+02/26/2022 10:53:35 - INFO - codeparrot_training - Step 44253: {'lr': 1.7477658164647357e-05, 'samples': 22658048, 'steps': 44253, 'loss/train': 2.25771164894104}
+02/26/2022 10:53:39 - INFO - codeparrot_training - Step 44254: {'lr': 1.747164818526692e-05, 'samples': 22658560, 'steps': 44254, 'loss/train': 1.1048643589019775}
+02/26/2022 10:53:44 - INFO - codeparrot_training - Step 44255: {'lr': 1.746563920196409e-05, 'samples': 22659072, 'steps': 44255, 'loss/train': 1.9347702264785767}
+02/26/2022 10:53:48 - INFO - codeparrot_training - Step 44256: {'lr': 1.7459631214764654e-05, 'samples': 22659584, 'steps': 44256, 'loss/train': 1.8936128616333008}
+02/26/2022 10:53:53 - INFO - codeparrot_training - Step 44257: {'lr': 1.745362422369426e-05, 'samples': 22660096, 'steps': 44257, 'loss/train': 2.983276605606079}
+02/26/2022 10:53:57 - INFO - codeparrot_training - Step 44258: {'lr': 1.7447618228778745e-05, 'samples': 22660608, 'steps': 44258, 'loss/train': 0.921862781047821}
+02/26/2022 10:54:04 - INFO - codeparrot_training - Step 44259: {'lr': 1.7441613230043786e-05, 'samples': 22661120, 'steps': 44259, 'loss/train': 1.5292614698410034}
+02/26/2022 10:54:07 - INFO - codeparrot_training - Step 44260: {'lr': 1.743560922751511e-05, 'samples': 22661632, 'steps': 44260, 'loss/train': 1.538718342781067}
+02/26/2022 10:54:13 - INFO - codeparrot_training - Step 44261: {'lr': 1.7429606221218392e-05, 'samples': 22662144, 'steps': 44261, 'loss/train': 0.7069941759109497}
+02/26/2022 10:54:16 - INFO - codeparrot_training - Step 44262: {'lr': 1.7423604211179444e-05, 'samples': 22662656, 'steps': 44262, 'loss/train': 1.9481362104415894}
+02/26/2022 10:54:22 - INFO - codeparrot_training - Step 44263: {'lr': 1.7417603197423914e-05, 'samples': 22663168, 'steps': 44263, 'loss/train': 0.46221479773521423}
+02/26/2022 10:54:25 - INFO - codeparrot_training - Step 44264: {'lr': 1.741160317997753e-05, 'samples': 22663680, 'steps': 44264, 'loss/train': 2.5421078205108643}
+02/26/2022 10:54:31 - INFO - codeparrot_training - Step 44265: {'lr': 1.740560415886594e-05, 'samples': 22664192, 'steps': 44265, 'loss/train': 0.11504635959863663}
+02/26/2022 10:54:34 - INFO - codeparrot_training - Step 44266: {'lr': 1.73996061341149e-05, 'samples': 22664704, 'steps': 44266, 'loss/train': 1.5536805391311646}
+02/26/2022 10:54:40 - INFO - codeparrot_training - Step 44267: {'lr': 1.739360910575011e-05, 'samples': 22665216, 'steps': 44267, 'loss/train': 1.0991525650024414}
+02/26/2022 10:54:43 - INFO - codeparrot_training - Step 44268: {'lr': 1.7387613073797215e-05, 'samples': 22665728, 'steps': 44268, 'loss/train': 1.2090414762496948}
+02/26/2022 10:54:49 - INFO - codeparrot_training - Step 44269: {'lr': 1.7381618038281978e-05, 'samples': 22666240, 'steps': 44269, 'loss/train': 1.298271656036377}
+02/26/2022 10:54:52 - INFO - codeparrot_training - Step 44270: {'lr': 1.7375623999229955e-05, 'samples': 22666752, 'steps': 44270, 'loss/train': 0.57925945520401}
+02/26/2022 10:55:00 - INFO - codeparrot_training - Step 44271: {'lr': 1.736963095666691e-05, 'samples': 22667264, 'steps': 44271, 'loss/train': 0.42965155839920044}
+02/26/2022 10:55:04 - INFO - codeparrot_training - Step 44272: {'lr': 1.7363638910618486e-05, 'samples': 22667776, 'steps': 44272, 'loss/train': 2.509251356124878}
+02/26/2022 10:55:09 - INFO - codeparrot_training - Step 44273: {'lr': 1.7357647861110438e-05, 'samples': 22668288, 'steps': 44273, 'loss/train': 1.362303614616394}
+02/26/2022 10:55:12 - INFO - codeparrot_training - Step 44274: {'lr': 1.7351657808168248e-05, 'samples': 22668800, 'steps': 44274, 'loss/train': 1.1205365657806396}
+02/26/2022 10:55:18 - INFO - codeparrot_training - Step 44275: {'lr': 1.7345668751817755e-05, 'samples': 22669312, 'steps': 44275, 'loss/train': 2.245718240737915}
+02/26/2022 10:55:21 - INFO - codeparrot_training - Step 44276: {'lr': 1.733968069208447e-05, 'samples': 22669824, 'steps': 44276, 'loss/train': 1.7272087335586548}
+02/26/2022 10:55:27 - INFO - codeparrot_training - Step 44277: {'lr': 1.733369362899423e-05, 'samples': 22670336, 'steps': 44277, 'loss/train': 0.06671199947595596}
+02/26/2022 10:55:31 - INFO - codeparrot_training - Step 44278: {'lr': 1.732770756257246e-05, 'samples': 22670848, 'steps': 44278, 'loss/train': 2.2394421100616455}
+02/26/2022 10:55:36 - INFO - codeparrot_training - Step 44279: {'lr': 1.7321722492844972e-05, 'samples': 22671360, 'steps': 44279, 'loss/train': 1.0178898572921753}
+02/26/2022 10:55:39 - INFO - codeparrot_training - Step 44280: {'lr': 1.7315738419837272e-05, 'samples': 22671872, 'steps': 44280, 'loss/train': 0.20388707518577576}
+02/26/2022 10:55:47 - INFO - codeparrot_training - Step 44281: {'lr': 1.7309755343575122e-05, 'samples': 22672384, 'steps': 44281, 'loss/train': 1.9329198598861694}
+02/26/2022 10:55:51 - INFO - codeparrot_training - Step 44282: {'lr': 1.730377326408408e-05, 'samples': 22672896, 'steps': 44282, 'loss/train': 2.291621685028076}
+02/26/2022 10:55:54 - INFO - codeparrot_training - Step 44283: {'lr': 1.72977921813898e-05, 'samples': 22673408, 'steps': 44283, 'loss/train': 1.476567268371582}
+02/26/2022 10:56:00 - INFO - codeparrot_training - Step 44284: {'lr': 1.7291812095517834e-05, 'samples': 22673920, 'steps': 44284, 'loss/train': 1.6869803667068481}
+02/26/2022 10:56:03 - INFO - codeparrot_training - Step 44285: {'lr': 1.7285833006493917e-05, 'samples': 22674432, 'steps': 44285, 'loss/train': 2.1983492374420166}
+02/26/2022 10:56:09 - INFO - codeparrot_training - Step 44286: {'lr': 1.7279854914343557e-05, 'samples': 22674944, 'steps': 44286, 'loss/train': 2.136845111846924}
+02/26/2022 10:56:12 - INFO - codeparrot_training - Step 44287: {'lr': 1.7273877819092425e-05, 'samples': 22675456, 'steps': 44287, 'loss/train': 1.2943073511123657}
+02/26/2022 10:56:18 - INFO - codeparrot_training - Step 44288: {'lr': 1.726790172076606e-05, 'samples': 22675968, 'steps': 44288, 'loss/train': 1.0539237260818481}
+02/26/2022 10:56:21 - INFO - codeparrot_training - Step 44289: {'lr': 1.7261926619390162e-05, 'samples': 22676480, 'steps': 44289, 'loss/train': 0.6272141337394714}
+02/26/2022 10:56:27 - INFO - codeparrot_training - Step 44290: {'lr': 1.7255952514990236e-05, 'samples': 22676992, 'steps': 44290, 'loss/train': 2.3690638542175293}
+02/26/2022 10:56:30 - INFO - codeparrot_training - Step 44291: {'lr': 1.7249979407591904e-05, 'samples': 22677504, 'steps': 44291, 'loss/train': 1.5071882009506226}
+02/26/2022 10:56:36 - INFO - codeparrot_training - Step 44292: {'lr': 1.7244007297220726e-05, 'samples': 22678016, 'steps': 44292, 'loss/train': 1.57838773727417}
+02/26/2022 10:56:39 - INFO - codeparrot_training - Step 44293: {'lr': 1.723803618390235e-05, 'samples': 22678528, 'steps': 44293, 'loss/train': 1.4604501724243164}
+02/26/2022 10:56:45 - INFO - codeparrot_training - Step 44294: {'lr': 1.7232066067662284e-05, 'samples': 22679040, 'steps': 44294, 'loss/train': 1.5234919786453247}
+02/26/2022 10:56:48 - INFO - codeparrot_training - Step 44295: {'lr': 1.7226096948526227e-05, 'samples': 22679552, 'steps': 44295, 'loss/train': 1.8620376586914062}
+02/26/2022 10:56:54 - INFO - codeparrot_training - Step 44296: {'lr': 1.722012882651955e-05, 'samples': 22680064, 'steps': 44296, 'loss/train': 1.6121175289154053}
+02/26/2022 10:56:57 - INFO - codeparrot_training - Step 44297: {'lr': 1.721416170166798e-05, 'samples': 22680576, 'steps': 44297, 'loss/train': 1.5987181663513184}
+02/26/2022 10:57:05 - INFO - codeparrot_training - Step 44298: {'lr': 1.7208195573996998e-05, 'samples': 22681088, 'steps': 44298, 'loss/train': 0.6864548325538635}
+02/26/2022 10:57:08 - INFO - codeparrot_training - Step 44299: {'lr': 1.7202230443532223e-05, 'samples': 22681600, 'steps': 44299, 'loss/train': 0.8742496371269226}
+02/26/2022 10:57:14 - INFO - codeparrot_training - Step 44300: {'lr': 1.719626631029911e-05, 'samples': 22682112, 'steps': 44300, 'loss/train': 2.184666156768799}
+02/26/2022 10:57:19 - INFO - codeparrot_training - Step 44301: {'lr': 1.7190303174323323e-05, 'samples': 22682624, 'steps': 44301, 'loss/train': 2.066521406173706}
+02/26/2022 10:57:23 - INFO - codeparrot_training - Step 44302: {'lr': 1.7184341035630325e-05, 'samples': 22683136, 'steps': 44302, 'loss/train': 1.77566397190094}
+02/26/2022 10:57:28 - INFO - codeparrot_training - Step 44303: {'lr': 1.717837989424567e-05, 'samples': 22683648, 'steps': 44303, 'loss/train': 1.2108327150344849}
+02/26/2022 10:57:32 - INFO - codeparrot_training - Step 44304: {'lr': 1.717241975019493e-05, 'samples': 22684160, 'steps': 44304, 'loss/train': 1.5419927835464478}
+02/26/2022 10:57:37 - INFO - codeparrot_training - Step 44305: {'lr': 1.7166460603503632e-05, 'samples': 22684672, 'steps': 44305, 'loss/train': 1.5790438652038574}
+02/26/2022 10:57:41 - INFO - codeparrot_training - Step 44306: {'lr': 1.716050245419726e-05, 'samples': 22685184, 'steps': 44306, 'loss/train': 1.9302430152893066}
+02/26/2022 10:57:48 - INFO - codeparrot_training - Step 44307: {'lr': 1.7154545302301346e-05, 'samples': 22685696, 'steps': 44307, 'loss/train': 1.3340981006622314}
+02/26/2022 10:57:52 - INFO - codeparrot_training - Step 44308: {'lr': 1.7148589147841426e-05, 'samples': 22686208, 'steps': 44308, 'loss/train': 1.9165781736373901}
+02/26/2022 10:57:57 - INFO - codeparrot_training - Step 44309: {'lr': 1.7142633990843036e-05, 'samples': 22686720, 'steps': 44309, 'loss/train': 1.89263117313385}
+02/26/2022 10:58:01 - INFO - codeparrot_training - Step 44310: {'lr': 1.7136679831331657e-05, 'samples': 22687232, 'steps': 44310, 'loss/train': 2.063504219055176}
+02/26/2022 10:58:06 - INFO - codeparrot_training - Step 44311: {'lr': 1.713072666933277e-05, 'samples': 22687744, 'steps': 44311, 'loss/train': 1.4691232442855835}
+02/26/2022 10:58:10 - INFO - codeparrot_training - Step 44312: {'lr': 1.7124774504871933e-05, 'samples': 22688256, 'steps': 44312, 'loss/train': 2.1377978324890137}
+02/26/2022 10:58:15 - INFO - codeparrot_training - Step 44313: {'lr': 1.7118823337974625e-05, 'samples': 22688768, 'steps': 44313, 'loss/train': 1.274139404296875}
+02/26/2022 10:58:19 - INFO - codeparrot_training - Step 44314: {'lr': 1.7112873168666333e-05, 'samples': 22689280, 'steps': 44314, 'loss/train': 1.5638521909713745}
+02/26/2022 10:58:24 - INFO - codeparrot_training - Step 44315: {'lr': 1.7106923996972473e-05, 'samples': 22689792, 'steps': 44315, 'loss/train': 1.7805567979812622}
+02/26/2022 10:58:28 - INFO - codeparrot_training - Step 44316: {'lr': 1.7100975822918667e-05, 'samples': 22690304, 'steps': 44316, 'loss/train': 1.4318487644195557}
+02/26/2022 10:58:35 - INFO - codeparrot_training - Step 44317: {'lr': 1.7095028646530313e-05, 'samples': 22690816, 'steps': 44317, 'loss/train': 1.335247278213501}
+02/26/2022 10:58:39 - INFO - codeparrot_training - Step 44318: {'lr': 1.708908246783292e-05, 'samples': 22691328, 'steps': 44318, 'loss/train': 0.7506019473075867}
+02/26/2022 10:58:44 - INFO - codeparrot_training - Step 44319: {'lr': 1.7083137286851875e-05, 'samples': 22691840, 'steps': 44319, 'loss/train': 2.368936777114868}
+02/26/2022 10:58:48 - INFO - codeparrot_training - Step 44320: {'lr': 1.7077193103612777e-05, 'samples': 22692352, 'steps': 44320, 'loss/train': 1.7087756395339966}
+02/26/2022 10:58:53 - INFO - codeparrot_training - Step 44321: {'lr': 1.707124991814099e-05, 'samples': 22692864, 'steps': 44321, 'loss/train': 1.6384295225143433}
+02/26/2022 10:58:57 - INFO - codeparrot_training - Step 44322: {'lr': 1.7065307730462027e-05, 'samples': 22693376, 'steps': 44322, 'loss/train': 2.0050432682037354}
+02/26/2022 10:59:03 - INFO - codeparrot_training - Step 44323: {'lr': 1.7059366540601277e-05, 'samples': 22693888, 'steps': 44323, 'loss/train': 1.3051033020019531}
+02/26/2022 10:59:06 - INFO - codeparrot_training - Step 44324: {'lr': 1.7053426348584283e-05, 'samples': 22694400, 'steps': 44324, 'loss/train': 1.8936129808425903}
+02/26/2022 10:59:11 - INFO - codeparrot_training - Step 44325: {'lr': 1.7047487154436438e-05, 'samples': 22694912, 'steps': 44325, 'loss/train': 2.197650194168091}
+02/26/2022 10:59:15 - INFO - codeparrot_training - Step 44326: {'lr': 1.7041548958183163e-05, 'samples': 22695424, 'steps': 44326, 'loss/train': 1.7790600061416626}
+02/26/2022 10:59:22 - INFO - codeparrot_training - Step 44327: {'lr': 1.7035611759849913e-05, 'samples': 22695936, 'steps': 44327, 'loss/train': 1.5710827112197876}
+02/26/2022 10:59:26 - INFO - codeparrot_training - Step 44328: {'lr': 1.7029675559462137e-05, 'samples': 22696448, 'steps': 44328, 'loss/train': 1.9791598320007324}
+02/26/2022 10:59:31 - INFO - codeparrot_training - Step 44329: {'lr': 1.7023740357045235e-05, 'samples': 22696960, 'steps': 44329, 'loss/train': 1.5320208072662354}
+02/26/2022 10:59:35 - INFO - codeparrot_training - Step 44330: {'lr': 1.7017806152624744e-05, 'samples': 22697472, 'steps': 44330, 'loss/train': 0.6081035137176514}
+02/26/2022 10:59:40 - INFO - codeparrot_training - Step 44331: {'lr': 1.7011872946225888e-05, 'samples': 22697984, 'steps': 44331, 'loss/train': 1.1642194986343384}
+02/26/2022 10:59:44 - INFO - codeparrot_training - Step 44332: {'lr': 1.700594073787423e-05, 'samples': 22698496, 'steps': 44332, 'loss/train': 1.7371748685836792}
+02/26/2022 10:59:49 - INFO - codeparrot_training - Step 44333: {'lr': 1.700000952759509e-05, 'samples': 22699008, 'steps': 44333, 'loss/train': 2.411820411682129}
+02/26/2022 10:59:53 - INFO - codeparrot_training - Step 44334: {'lr': 1.6994079315414022e-05, 'samples': 22699520, 'steps': 44334, 'loss/train': 1.6131850481033325}
+02/26/2022 10:59:58 - INFO - codeparrot_training - Step 44335: {'lr': 1.6988150101356238e-05, 'samples': 22700032, 'steps': 44335, 'loss/train': 2.5622992515563965}
+02/26/2022 11:00:02 - INFO - codeparrot_training - Step 44336: {'lr': 1.6982221885447263e-05, 'samples': 22700544, 'steps': 44336, 'loss/train': 2.8687050342559814}
+02/26/2022 11:00:07 - INFO - codeparrot_training - Step 44337: {'lr': 1.6976294667712443e-05, 'samples': 22701056, 'steps': 44337, 'loss/train': 0.29597508907318115}
+02/26/2022 11:00:11 - INFO - codeparrot_training - Step 44338: {'lr': 1.697036844817723e-05, 'samples': 22701568, 'steps': 44338, 'loss/train': 1.9202061891555786}
+02/26/2022 11:00:16 - INFO - codeparrot_training - Step 44339: {'lr': 1.696444322686691e-05, 'samples': 22702080, 'steps': 44339, 'loss/train': 1.9780937433242798}
+02/26/2022 11:00:20 - INFO - codeparrot_training - Step 44340: {'lr': 1.695851900380696e-05, 'samples': 22702592, 'steps': 44340, 'loss/train': 1.8543251752853394}
+02/26/2022 11:00:25 - INFO - codeparrot_training - Step 44341: {'lr': 1.6952595779022667e-05, 'samples': 22703104, 'steps': 44341, 'loss/train': 1.5015946626663208}
+02/26/2022 11:00:29 - INFO - codeparrot_training - Step 44342: {'lr': 1.694667355253948e-05, 'samples': 22703616, 'steps': 44342, 'loss/train': 1.5951582193374634}
+02/26/2022 11:00:36 - INFO - codeparrot_training - Step 44343: {'lr': 1.6940752324382747e-05, 'samples': 22704128, 'steps': 44343, 'loss/train': 1.6642869710922241}
+02/26/2022 11:00:40 - INFO - codeparrot_training - Step 44344: {'lr': 1.693483209457783e-05, 'samples': 22704640, 'steps': 44344, 'loss/train': 2.298598051071167}
+02/26/2022 11:00:45 - INFO - codeparrot_training - Step 44345: {'lr': 1.6928912863150042e-05, 'samples': 22705152, 'steps': 44345, 'loss/train': 2.067037582397461}
+02/26/2022 11:00:49 - INFO - codeparrot_training - Step 44346: {'lr': 1.692299463012484e-05, 'samples': 22705664, 'steps': 44346, 'loss/train': 1.6908059120178223}
+02/26/2022 11:00:55 - INFO - codeparrot_training - Step 44347: {'lr': 1.69170773955275e-05, 'samples': 22706176, 'steps': 44347, 'loss/train': 1.96216881275177}
+02/26/2022 11:00:58 - INFO - codeparrot_training - Step 44348: {'lr': 1.6911161159383403e-05, 'samples': 22706688, 'steps': 44348, 'loss/train': 0.8237321972846985}
+02/26/2022 11:01:03 - INFO - codeparrot_training - Step 44349: {'lr': 1.6905245921717853e-05, 'samples': 22707200, 'steps': 44349, 'loss/train': 0.5964906811714172}
+02/26/2022 11:01:07 - INFO - codeparrot_training - Step 44350: {'lr': 1.6899331682556195e-05, 'samples': 22707712, 'steps': 44350, 'loss/train': 1.7712651491165161}
+02/26/2022 11:01:12 - INFO - codeparrot_training - Step 44351: {'lr': 1.6893418441923823e-05, 'samples': 22708224, 'steps': 44351, 'loss/train': 1.2285692691802979}
+02/26/2022 11:01:16 - INFO - codeparrot_training - Step 44352: {'lr': 1.6887506199846026e-05, 'samples': 22708736, 'steps': 44352, 'loss/train': 1.5095158815383911}
+02/26/2022 11:01:23 - INFO - codeparrot_training - Step 44353: {'lr': 1.688159495634811e-05, 'samples': 22709248, 'steps': 44353, 'loss/train': 1.5155912637710571}
+02/26/2022 11:01:26 - INFO - codeparrot_training - Step 44354: {'lr': 1.6875684711455398e-05, 'samples': 22709760, 'steps': 44354, 'loss/train': 3.1292927265167236}
+02/26/2022 11:01:32 - INFO - codeparrot_training - Step 44355: {'lr': 1.6869775465193277e-05, 'samples': 22710272, 'steps': 44355, 'loss/train': 0.6916528344154358}
+02/26/2022 11:01:35 - INFO - codeparrot_training - Step 44356: {'lr': 1.686386721758698e-05, 'samples': 22710784, 'steps': 44356, 'loss/train': 1.5902073383331299}
+02/26/2022 11:01:41 - INFO - codeparrot_training - Step 44357: {'lr': 1.685795996866188e-05, 'samples': 22711296, 'steps': 44357, 'loss/train': 2.023514747619629}
+02/26/2022 11:01:44 - INFO - codeparrot_training - Step 44358: {'lr': 1.6852053718443174e-05, 'samples': 22711808, 'steps': 44358, 'loss/train': 2.0408995151519775}
+02/26/2022 11:01:50 - INFO - codeparrot_training - Step 44359: {'lr': 1.6846148466956314e-05, 'samples': 22712320, 'steps': 44359, 'loss/train': 2.2770464420318604}
+02/26/2022 11:01:53 - INFO - codeparrot_training - Step 44360: {'lr': 1.6840244214226503e-05, 'samples': 22712832, 'steps': 44360, 'loss/train': 1.7497460842132568}
+02/26/2022 11:01:59 - INFO - codeparrot_training - Step 44361: {'lr': 1.6834340960279028e-05, 'samples': 22713344, 'steps': 44361, 'loss/train': 1.7307929992675781}
+02/26/2022 11:02:02 - INFO - codeparrot_training - Step 44362: {'lr': 1.682843870513917e-05, 'samples': 22713856, 'steps': 44362, 'loss/train': 1.2015419006347656}
+02/26/2022 11:02:08 - INFO - codeparrot_training - Step 44363: {'lr': 1.6822537448832305e-05, 'samples': 22714368, 'steps': 44363, 'loss/train': 1.3100426197052002}
+02/26/2022 11:02:11 - INFO - codeparrot_training - Step 44364: {'lr': 1.6816637191383573e-05, 'samples': 22714880, 'steps': 44364, 'loss/train': 0.4053735136985779}
+02/26/2022 11:02:19 - INFO - codeparrot_training - Step 44365: {'lr': 1.6810737932818428e-05, 'samples': 22715392, 'steps': 44365, 'loss/train': 2.5232043266296387}
+02/26/2022 11:02:22 - INFO - codeparrot_training - Step 44366: {'lr': 1.6804839673161958e-05, 'samples': 22715904, 'steps': 44366, 'loss/train': 2.31803822517395}
+02/26/2022 11:02:28 - INFO - codeparrot_training - Step 44367: {'lr': 1.679894241243954e-05, 'samples': 22716416, 'steps': 44367, 'loss/train': 0.9804304242134094}
+02/26/2022 11:02:33 - INFO - codeparrot_training - Step 44368: {'lr': 1.679304615067634e-05, 'samples': 22716928, 'steps': 44368, 'loss/train': 1.317438006401062}
+02/26/2022 11:02:37 - INFO - codeparrot_training - Step 44369: {'lr': 1.6787150887897785e-05, 'samples': 22717440, 'steps': 44369, 'loss/train': 1.6659883260726929}
+02/26/2022 11:02:40 - INFO - codeparrot_training - Step 44370: {'lr': 1.6781256624128937e-05, 'samples': 22717952, 'steps': 44370, 'loss/train': 1.5732172727584839}
+02/26/2022 11:02:46 - INFO - codeparrot_training - Step 44371: {'lr': 1.677536335939517e-05, 'samples': 22718464, 'steps': 44371, 'loss/train': 2.2136690616607666}
+02/26/2022 11:02:49 - INFO - codeparrot_training - Step 44372: {'lr': 1.676947109372165e-05, 'samples': 22718976, 'steps': 44372, 'loss/train': 1.3972251415252686}
+02/26/2022 11:02:55 - INFO - codeparrot_training - Step 44373: {'lr': 1.6763579827133723e-05, 'samples': 22719488, 'steps': 44373, 'loss/train': 2.0953214168548584}
+02/26/2022 11:02:58 - INFO - codeparrot_training - Step 44374: {'lr': 1.675768955965648e-05, 'samples': 22720000, 'steps': 44374, 'loss/train': 1.6399335861206055}
+02/26/2022 11:03:06 - INFO - codeparrot_training - Step 44375: {'lr': 1.675180029131529e-05, 'samples': 22720512, 'steps': 44375, 'loss/train': 1.7166416645050049}
+02/26/2022 11:03:09 - INFO - codeparrot_training - Step 44376: {'lr': 1.674591202213527e-05, 'samples': 22721024, 'steps': 44376, 'loss/train': 3.4678473472595215}
+02/26/2022 11:03:15 - INFO - codeparrot_training - Step 44377: {'lr': 1.674002475214176e-05, 'samples': 22721536, 'steps': 44377, 'loss/train': 0.4510103166103363}
+02/26/2022 11:03:18 - INFO - codeparrot_training - Step 44378: {'lr': 1.673413848135988e-05, 'samples': 22722048, 'steps': 44378, 'loss/train': 2.1093909740448}
+02/26/2022 11:03:24 - INFO - codeparrot_training - Step 44379: {'lr': 1.6728253209814915e-05, 'samples': 22722560, 'steps': 44379, 'loss/train': 1.158706784248352}
+02/26/2022 11:03:27 - INFO - codeparrot_training - Step 44380: {'lr': 1.6722368937531984e-05, 'samples': 22723072, 'steps': 44380, 'loss/train': 1.2744829654693604}
+02/26/2022 11:03:33 - INFO - codeparrot_training - Step 44381: {'lr': 1.6716485664536372e-05, 'samples': 22723584, 'steps': 44381, 'loss/train': 1.732061505317688}
+02/26/2022 11:03:36 - INFO - codeparrot_training - Step 44382: {'lr': 1.6710603390853284e-05, 'samples': 22724096, 'steps': 44382, 'loss/train': 1.1910959482192993}
+02/26/2022 11:03:41 - INFO - codeparrot_training - Step 44383: {'lr': 1.670472211650789e-05, 'samples': 22724608, 'steps': 44383, 'loss/train': 1.8050498962402344}
+02/26/2022 11:03:47 - INFO - codeparrot_training - Step 44384: {'lr': 1.669884184152534e-05, 'samples': 22725120, 'steps': 44384, 'loss/train': 1.7572084665298462}
+02/26/2022 11:03:50 - INFO - codeparrot_training - Step 44385: {'lr': 1.669296256593092e-05, 'samples': 22725632, 'steps': 44385, 'loss/train': 1.8314539194107056}
+02/26/2022 11:03:56 - INFO - codeparrot_training - Step 44386: {'lr': 1.6687084289749772e-05, 'samples': 22726144, 'steps': 44386, 'loss/train': 1.5001260042190552}
+02/26/2022 11:04:00 - INFO - codeparrot_training - Step 44387: {'lr': 1.6681207013007043e-05, 'samples': 22726656, 'steps': 44387, 'loss/train': 2.005467176437378}
+02/26/2022 11:04:03 - INFO - codeparrot_training - Step 44388: {'lr': 1.667533073572791e-05, 'samples': 22727168, 'steps': 44388, 'loss/train': 0.6919127702713013}
+02/26/2022 11:04:10 - INFO - codeparrot_training - Step 44389: {'lr': 1.6669455457937606e-05, 'samples': 22727680, 'steps': 44389, 'loss/train': 3.4618687629699707}
+02/26/2022 11:04:14 - INFO - codeparrot_training - Step 44390: {'lr': 1.666358117966127e-05, 'samples': 22728192, 'steps': 44390, 'loss/train': 1.8993316888809204}
+02/26/2022 11:04:19 - INFO - codeparrot_training - Step 44391: {'lr': 1.665770790092405e-05, 'samples': 22728704, 'steps': 44391, 'loss/train': 0.6506956219673157}
+02/26/2022 11:04:23 - INFO - codeparrot_training - Step 44392: {'lr': 1.66518356217511e-05, 'samples': 22729216, 'steps': 44392, 'loss/train': 0.6920478940010071}
+02/26/2022 11:04:29 - INFO - codeparrot_training - Step 44393: {'lr': 1.6645964342167613e-05, 'samples': 22729728, 'steps': 44393, 'loss/train': 2.802110195159912}
+02/26/2022 11:04:32 - INFO - codeparrot_training - Step 44394: {'lr': 1.6640094062198714e-05, 'samples': 22730240, 'steps': 44394, 'loss/train': 1.5093233585357666}
+02/26/2022 11:04:38 - INFO - codeparrot_training - Step 44395: {'lr': 1.6634224781869544e-05, 'samples': 22730752, 'steps': 44395, 'loss/train': 2.13940167427063}
+02/26/2022 11:04:41 - INFO - codeparrot_training - Step 44396: {'lr': 1.6628356501205283e-05, 'samples': 22731264, 'steps': 44396, 'loss/train': 2.4444870948791504}
+02/26/2022 11:04:47 - INFO - codeparrot_training - Step 44397: {'lr': 1.6622489220230964e-05, 'samples': 22731776, 'steps': 44397, 'loss/train': 0.8667925000190735}
+02/26/2022 11:04:50 - INFO - codeparrot_training - Step 44398: {'lr': 1.661662293897187e-05, 'samples': 22732288, 'steps': 44398, 'loss/train': 1.5421849489212036}
+02/26/2022 11:04:57 - INFO - codeparrot_training - Step 44399: {'lr': 1.661075765745304e-05, 'samples': 22732800, 'steps': 44399, 'loss/train': 1.5873042345046997}
+02/26/2022 11:05:01 - INFO - codeparrot_training - Step 44400: {'lr': 1.6604893375699592e-05, 'samples': 22733312, 'steps': 44400, 'loss/train': 2.275908946990967}
+02/26/2022 11:05:06 - INFO - codeparrot_training - Step 44401: {'lr': 1.6599030093736644e-05, 'samples': 22733824, 'steps': 44401, 'loss/train': 1.5064443349838257}
+02/26/2022 11:05:10 - INFO - codeparrot_training - Step 44402: {'lr': 1.6593167811589398e-05, 'samples': 22734336, 'steps': 44402, 'loss/train': 0.9278638362884521}
+02/26/2022 11:05:15 - INFO - codeparrot_training - Step 44403: {'lr': 1.6587306529282838e-05, 'samples': 22734848, 'steps': 44403, 'loss/train': 1.7654253244400024}
+02/26/2022 11:05:19 - INFO - codeparrot_training - Step 44404: {'lr': 1.6581446246842243e-05, 'samples': 22735360, 'steps': 44404, 'loss/train': 2.624229907989502}
+02/26/2022 11:05:24 - INFO - codeparrot_training - Step 44405: {'lr': 1.657558696429254e-05, 'samples': 22735872, 'steps': 44405, 'loss/train': 1.9898579120635986}
+02/26/2022 11:05:28 - INFO - codeparrot_training - Step 44406: {'lr': 1.6569728681658904e-05, 'samples': 22736384, 'steps': 44406, 'loss/train': 1.9804118871688843}
+02/26/2022 11:05:34 - INFO - codeparrot_training - Step 44407: {'lr': 1.6563871398966423e-05, 'samples': 22736896, 'steps': 44407, 'loss/train': 0.9044759273529053}
+02/26/2022 11:05:37 - INFO - codeparrot_training - Step 44408: {'lr': 1.655801511624025e-05, 'samples': 22737408, 'steps': 44408, 'loss/train': 1.9096637964248657}
+02/26/2022 11:05:45 - INFO - codeparrot_training - Step 44409: {'lr': 1.655215983350536e-05, 'samples': 22737920, 'steps': 44409, 'loss/train': 2.122568130493164}
+02/26/2022 11:05:48 - INFO - codeparrot_training - Step 44410: {'lr': 1.65463055507869e-05, 'samples': 22738432, 'steps': 44410, 'loss/train': 0.9513373374938965}
+02/26/2022 11:05:54 - INFO - codeparrot_training - Step 44411: {'lr': 1.6540452268109906e-05, 'samples': 22738944, 'steps': 44411, 'loss/train': 1.8396836519241333}
+02/26/2022 11:05:57 - INFO - codeparrot_training - Step 44412: {'lr': 1.6534599985499527e-05, 'samples': 22739456, 'steps': 44412, 'loss/train': 0.6635790467262268}
+02/26/2022 11:06:03 - INFO - codeparrot_training - Step 44413: {'lr': 1.6528748702980796e-05, 'samples': 22739968, 'steps': 44413, 'loss/train': 2.1034364700317383}
+02/26/2022 11:06:06 - INFO - codeparrot_training - Step 44414: {'lr': 1.652289842057875e-05, 'samples': 22740480, 'steps': 44414, 'loss/train': 2.0697436332702637}
+02/26/2022 11:06:12 - INFO - codeparrot_training - Step 44415: {'lr': 1.6517049138318446e-05, 'samples': 22740992, 'steps': 44415, 'loss/train': 1.4194016456604004}
+02/26/2022 11:06:15 - INFO - codeparrot_training - Step 44416: {'lr': 1.6511200856225012e-05, 'samples': 22741504, 'steps': 44416, 'loss/train': 1.4980175495147705}
+02/26/2022 11:06:21 - INFO - codeparrot_training - Step 44417: {'lr': 1.6505353574323423e-05, 'samples': 22742016, 'steps': 44417, 'loss/train': 0.16788151860237122}
+02/26/2022 11:06:24 - INFO - codeparrot_training - Step 44418: {'lr': 1.6499507292638773e-05, 'samples': 22742528, 'steps': 44418, 'loss/train': 2.3986454010009766}
+02/26/2022 11:06:30 - INFO - codeparrot_training - Step 44419: {'lr': 1.6493662011196065e-05, 'samples': 22743040, 'steps': 44419, 'loss/train': 2.2067408561706543}
+02/26/2022 11:06:33 - INFO - codeparrot_training - Step 44420: {'lr': 1.6487817730020365e-05, 'samples': 22743552, 'steps': 44420, 'loss/train': 1.5402960777282715}
+02/26/2022 11:06:39 - INFO - codeparrot_training - Step 44421: {'lr': 1.6481974449136738e-05, 'samples': 22744064, 'steps': 44421, 'loss/train': 1.4308686256408691}
+02/26/2022 11:06:42 - INFO - codeparrot_training - Step 44422: {'lr': 1.6476132168570162e-05, 'samples': 22744576, 'steps': 44422, 'loss/train': 2.275926113128662}
+02/26/2022 11:06:48 - INFO - codeparrot_training - Step 44423: {'lr': 1.6470290888345675e-05, 'samples': 22745088, 'steps': 44423, 'loss/train': 2.815519094467163}
+02/26/2022 11:06:51 - INFO - codeparrot_training - Step 44424: {'lr': 1.6464450608488308e-05, 'samples': 22745600, 'steps': 44424, 'loss/train': 1.0224368572235107}
+02/26/2022 11:06:57 - INFO - codeparrot_training - Step 44425: {'lr': 1.6458611329023073e-05, 'samples': 22746112, 'steps': 44425, 'loss/train': 1.6621652841567993}
+02/26/2022 11:07:00 - INFO - codeparrot_training - Step 44426: {'lr': 1.6452773049975027e-05, 'samples': 22746624, 'steps': 44426, 'loss/train': 0.5357076525688171}
+02/26/2022 11:07:07 - INFO - codeparrot_training - Step 44427: {'lr': 1.6446935771369103e-05, 'samples': 22747136, 'steps': 44427, 'loss/train': 1.5596338510513306}
+02/26/2022 11:07:11 - INFO - codeparrot_training - Step 44428: {'lr': 1.644109949323036e-05, 'samples': 22747648, 'steps': 44428, 'loss/train': 2.2854785919189453}
+02/26/2022 11:07:16 - INFO - codeparrot_training - Step 44429: {'lr': 1.6435264215583807e-05, 'samples': 22748160, 'steps': 44429, 'loss/train': 1.2561132907867432}
+02/26/2022 11:07:20 - INFO - codeparrot_training - Step 44430: {'lr': 1.6429429938454394e-05, 'samples': 22748672, 'steps': 44430, 'loss/train': 2.0563724040985107}
+02/26/2022 11:07:25 - INFO - codeparrot_training - Step 44431: {'lr': 1.6423596661867102e-05, 'samples': 22749184, 'steps': 44431, 'loss/train': 2.307969808578491}
+02/26/2022 11:07:29 - INFO - codeparrot_training - Step 44432: {'lr': 1.6417764385846996e-05, 'samples': 22749696, 'steps': 44432, 'loss/train': 1.527620792388916}
+02/26/2022 11:07:35 - INFO - codeparrot_training - Step 44433: {'lr': 1.6411933110419023e-05, 'samples': 22750208, 'steps': 44433, 'loss/train': 2.0716638565063477}
+02/26/2022 11:07:38 - INFO - codeparrot_training - Step 44434: {'lr': 1.6406102835608144e-05, 'samples': 22750720, 'steps': 44434, 'loss/train': 2.2128257751464844}
+02/26/2022 11:07:45 - INFO - codeparrot_training - Step 44435: {'lr': 1.6400273561439305e-05, 'samples': 22751232, 'steps': 44435, 'loss/train': 1.1949172019958496}
+02/26/2022 11:07:48 - INFO - codeparrot_training - Step 44436: {'lr': 1.639444528793757e-05, 'samples': 22751744, 'steps': 44436, 'loss/train': 1.7842636108398438}
+02/26/2022 11:07:54 - INFO - codeparrot_training - Step 44437: {'lr': 1.6388618015127836e-05, 'samples': 22752256, 'steps': 44437, 'loss/train': 3.123776435852051}
+02/26/2022 11:07:57 - INFO - codeparrot_training - Step 44438: {'lr': 1.638279174303506e-05, 'samples': 22752768, 'steps': 44438, 'loss/train': 1.9284619092941284}
+02/26/2022 11:08:05 - INFO - codeparrot_training - Step 44439: {'lr': 1.637696647168427e-05, 'samples': 22753280, 'steps': 44439, 'loss/train': 2.3993191719055176}
+02/26/2022 11:08:09 - INFO - codeparrot_training - Step 44440: {'lr': 1.6371142201100338e-05, 'samples': 22753792, 'steps': 44440, 'loss/train': 1.1672440767288208}
+02/26/2022 11:08:14 - INFO - codeparrot_training - Step 44441: {'lr': 1.636531893130827e-05, 'samples': 22754304, 'steps': 44441, 'loss/train': 1.2598628997802734}
+02/26/2022 11:08:18 - INFO - codeparrot_training - Step 44442: {'lr': 1.635949666233294e-05, 'samples': 22754816, 'steps': 44442, 'loss/train': 1.899202585220337}
+02/26/2022 11:08:23 - INFO - codeparrot_training - Step 44443: {'lr': 1.6353675394199434e-05, 'samples': 22755328, 'steps': 44443, 'loss/train': 1.6192600727081299}
+02/26/2022 11:08:27 - INFO - codeparrot_training - Step 44444: {'lr': 1.6347855126932515e-05, 'samples': 22755840, 'steps': 44444, 'loss/train': 1.7620224952697754}
+02/26/2022 11:08:32 - INFO - codeparrot_training - Step 44445: {'lr': 1.6342035860557213e-05, 'samples': 22756352, 'steps': 44445, 'loss/train': 1.531874179840088}
+02/26/2022 11:08:36 - INFO - codeparrot_training - Step 44446: {'lr': 1.63362175950984e-05, 'samples': 22756864, 'steps': 44446, 'loss/train': 1.4054603576660156}
+02/26/2022 11:08:41 - INFO - codeparrot_training - Step 44447: {'lr': 1.6330400330581112e-05, 'samples': 22757376, 'steps': 44447, 'loss/train': 0.9037963151931763}
+02/26/2022 11:08:45 - INFO - codeparrot_training - Step 44448: {'lr': 1.6324584067030107e-05, 'samples': 22757888, 'steps': 44448, 'loss/train': 1.8330234289169312}
+02/26/2022 11:08:52 - INFO - codeparrot_training - Step 44449: {'lr': 1.631876880447042e-05, 'samples': 22758400, 'steps': 44449, 'loss/train': 1.782422661781311}
+02/26/2022 11:08:55 - INFO - codeparrot_training - Step 44450: {'lr': 1.6312954542926888e-05, 'samples': 22758912, 'steps': 44450, 'loss/train': 1.4474053382873535}
+02/26/2022 11:09:01 - INFO - codeparrot_training - Step 44451: {'lr': 1.63071412824245e-05, 'samples': 22759424, 'steps': 44451, 'loss/train': 1.9806078672409058}
+02/26/2022 11:09:04 - INFO - codeparrot_training - Step 44452: {'lr': 1.6301329022988114e-05, 'samples': 22759936, 'steps': 44452, 'loss/train': 2.3462915420532227}
+02/26/2022 11:09:10 - INFO - codeparrot_training - Step 44453: {'lr': 1.6295517764642635e-05, 'samples': 22760448, 'steps': 44453, 'loss/train': 0.9762347936630249}
+02/26/2022 11:09:13 - INFO - codeparrot_training - Step 44454: {'lr': 1.6289707507412905e-05, 'samples': 22760960, 'steps': 44454, 'loss/train': 1.8561501502990723}
+02/26/2022 11:09:19 - INFO - codeparrot_training - Step 44455: {'lr': 1.628389825132387e-05, 'samples': 22761472, 'steps': 44455, 'loss/train': 1.9037665128707886}
+02/26/2022 11:09:22 - INFO - codeparrot_training - Step 44456: {'lr': 1.627808999640043e-05, 'samples': 22761984, 'steps': 44456, 'loss/train': 1.3427125215530396}
+02/26/2022 11:09:28 - INFO - codeparrot_training - Step 44457: {'lr': 1.6272282742667427e-05, 'samples': 22762496, 'steps': 44457, 'loss/train': 1.6749818325042725}
+02/26/2022 11:09:31 - INFO - codeparrot_training - Step 44458: {'lr': 1.626647649014973e-05, 'samples': 22763008, 'steps': 44458, 'loss/train': 1.0923738479614258}
+02/26/2022 11:09:39 - INFO - codeparrot_training - Step 44459: {'lr': 1.626067123887223e-05, 'samples': 22763520, 'steps': 44459, 'loss/train': 2.346987724304199}
+02/26/2022 11:09:42 - INFO - codeparrot_training - Step 44460: {'lr': 1.6254866988859778e-05, 'samples': 22764032, 'steps': 44460, 'loss/train': 1.9795336723327637}
+02/26/2022 11:09:48 - INFO - codeparrot_training - Step 44461: {'lr': 1.624906374013732e-05, 'samples': 22764544, 'steps': 44461, 'loss/train': 1.8764890432357788}
+02/26/2022 11:09:51 - INFO - codeparrot_training - Step 44462: {'lr': 1.624326149272959e-05, 'samples': 22765056, 'steps': 44462, 'loss/train': 0.8494221568107605}
+02/26/2022 11:09:57 - INFO - codeparrot_training - Step 44463: {'lr': 1.6237460246661533e-05, 'samples': 22765568, 'steps': 44463, 'loss/train': 2.170180320739746}
+02/26/2022 11:10:00 - INFO - codeparrot_training - Step 44464: {'lr': 1.6231660001957916e-05, 'samples': 22766080, 'steps': 44464, 'loss/train': 1.8503493070602417}
+02/26/2022 11:10:06 - INFO - codeparrot_training - Step 44465: {'lr': 1.6225860758643742e-05, 'samples': 22766592, 'steps': 44465, 'loss/train': 1.6686559915542603}
+02/26/2022 11:10:10 - INFO - codeparrot_training - Step 44466: {'lr': 1.6220062516743627e-05, 'samples': 22767104, 'steps': 44466, 'loss/train': 1.4229120016098022}
+02/26/2022 11:10:16 - INFO - codeparrot_training - Step 44467: {'lr': 1.6214265276282614e-05, 'samples': 22767616, 'steps': 44467, 'loss/train': 2.4440057277679443}
+02/26/2022 11:10:19 - INFO - codeparrot_training - Step 44468: {'lr': 1.6208469037285402e-05, 'samples': 22768128, 'steps': 44468, 'loss/train': 0.8162800073623657}
+02/26/2022 11:10:24 - INFO - codeparrot_training - Step 44469: {'lr': 1.620267379977694e-05, 'samples': 22768640, 'steps': 44469, 'loss/train': 2.914889097213745}
+02/26/2022 11:10:28 - INFO - codeparrot_training - Step 44470: {'lr': 1.6196879563781908e-05, 'samples': 22769152, 'steps': 44470, 'loss/train': 1.7064741849899292}
+02/26/2022 11:10:35 - INFO - codeparrot_training - Step 44471: {'lr': 1.6191086329325227e-05, 'samples': 22769664, 'steps': 44471, 'loss/train': 1.5800516605377197}
+02/26/2022 11:10:39 - INFO - codeparrot_training - Step 44472: {'lr': 1.6185294096431685e-05, 'samples': 22770176, 'steps': 44472, 'loss/train': 0.9188938140869141}
+02/26/2022 11:10:44 - INFO - codeparrot_training - Step 44473: {'lr': 1.6179502865126122e-05, 'samples': 22770688, 'steps': 44473, 'loss/train': 1.3499151468276978}
+02/26/2022 11:10:48 - INFO - codeparrot_training - Step 44474: {'lr': 1.6173712635433323e-05, 'samples': 22771200, 'steps': 44474, 'loss/train': 0.5873643755912781}
+02/26/2022 11:10:53 - INFO - codeparrot_training - Step 44475: {'lr': 1.6167923407378078e-05, 'samples': 22771712, 'steps': 44475, 'loss/train': 1.6316195726394653}
+02/26/2022 11:10:57 - INFO - codeparrot_training - Step 44476: {'lr': 1.6162135180985223e-05, 'samples': 22772224, 'steps': 44476, 'loss/train': 1.1441177129745483}
+02/26/2022 11:11:02 - INFO - codeparrot_training - Step 44477: {'lr': 1.615634795627946e-05, 'samples': 22772736, 'steps': 44477, 'loss/train': 1.8901101350784302}
+02/26/2022 11:11:06 - INFO - codeparrot_training - Step 44478: {'lr': 1.6150561733285717e-05, 'samples': 22773248, 'steps': 44478, 'loss/train': 1.8721638917922974}
+02/26/2022 11:11:11 - INFO - codeparrot_training - Step 44479: {'lr': 1.61447765120287e-05, 'samples': 22773760, 'steps': 44479, 'loss/train': 0.9614415764808655}
+02/26/2022 11:11:15 - INFO - codeparrot_training - Step 44480: {'lr': 1.6138992292533183e-05, 'samples': 22774272, 'steps': 44480, 'loss/train': 1.1102651357650757}
+02/26/2022 11:11:22 - INFO - codeparrot_training - Step 44481: {'lr': 1.613320907482396e-05, 'samples': 22774784, 'steps': 44481, 'loss/train': 2.107565402984619}
+02/26/2022 11:11:25 - INFO - codeparrot_training - Step 44482: {'lr': 1.612742685892582e-05, 'samples': 22775296, 'steps': 44482, 'loss/train': 1.6693578958511353}
+02/26/2022 11:11:31 - INFO - codeparrot_training - Step 44483: {'lr': 1.612164564486354e-05, 'samples': 22775808, 'steps': 44483, 'loss/train': 0.8629449009895325}
+02/26/2022 11:11:34 - INFO - codeparrot_training - Step 44484: {'lr': 1.6115865432661858e-05, 'samples': 22776320, 'steps': 44484, 'loss/train': 1.665207028388977}
+02/26/2022 11:11:40 - INFO - codeparrot_training - Step 44485: {'lr': 1.61100862223455e-05, 'samples': 22776832, 'steps': 44485, 'loss/train': 1.5447379350662231}
+02/26/2022 11:11:43 - INFO - codeparrot_training - Step 44486: {'lr': 1.6104308013939306e-05, 'samples': 22777344, 'steps': 44486, 'loss/train': 1.5844171047210693}
+02/26/2022 11:11:49 - INFO - codeparrot_training - Step 44487: {'lr': 1.609853080746798e-05, 'samples': 22777856, 'steps': 44487, 'loss/train': 1.3804858922958374}
+02/26/2022 11:11:52 - INFO - codeparrot_training - Step 44488: {'lr': 1.6092754602956284e-05, 'samples': 22778368, 'steps': 44488, 'loss/train': 2.5782389640808105}
+02/26/2022 11:11:58 - INFO - codeparrot_training - Step 44489: {'lr': 1.6086979400428913e-05, 'samples': 22778880, 'steps': 44489, 'loss/train': 0.08129153400659561}
+02/26/2022 11:12:02 - INFO - codeparrot_training - Step 44490: {'lr': 1.6081205199910687e-05, 'samples': 22779392, 'steps': 44490, 'loss/train': 1.9636930227279663}
+02/26/2022 11:12:07 - INFO - codeparrot_training - Step 44491: {'lr': 1.6075432001426305e-05, 'samples': 22779904, 'steps': 44491, 'loss/train': 0.9341439604759216}
+02/26/2022 11:12:11 - INFO - codeparrot_training - Step 44492: {'lr': 1.606965980500047e-05, 'samples': 22780416, 'steps': 44492, 'loss/train': 1.3446927070617676}
+02/26/2022 11:12:16 - INFO - codeparrot_training - Step 44493: {'lr': 1.606388861065791e-05, 'samples': 22780928, 'steps': 44493, 'loss/train': 1.7613171339035034}
+02/26/2022 11:12:20 - INFO - codeparrot_training - Step 44494: {'lr': 1.6058118418423418e-05, 'samples': 22781440, 'steps': 44494, 'loss/train': 1.5892575979232788}
+02/26/2022 11:12:27 - INFO - codeparrot_training - Step 44495: {'lr': 1.6052349228321634e-05, 'samples': 22781952, 'steps': 44495, 'loss/train': 1.4850115776062012}
+02/26/2022 11:12:30 - INFO - codeparrot_training - Step 44496: {'lr': 1.6046581040377316e-05, 'samples': 22782464, 'steps': 44496, 'loss/train': 1.5176955461502075}
+02/26/2022 11:12:36 - INFO - codeparrot_training - Step 44497: {'lr': 1.6040813854615115e-05, 'samples': 22782976, 'steps': 44497, 'loss/train': 2.4214513301849365}
+02/26/2022 11:12:40 - INFO - codeparrot_training - Step 44498: {'lr': 1.6035047671059816e-05, 'samples': 22783488, 'steps': 44498, 'loss/train': 1.9264425039291382}
+02/26/2022 11:12:45 - INFO - codeparrot_training - Step 44499: {'lr': 1.6029282489736036e-05, 'samples': 22784000, 'steps': 44499, 'loss/train': 0.5881751179695129}
+02/26/2022 11:12:49 - INFO - codeparrot_training - Step 44500: {'lr': 1.6023518310668618e-05, 'samples': 22784512, 'steps': 44500, 'loss/train': 1.1519964933395386}
+02/26/2022 11:12:54 - INFO - codeparrot_training - Step 44501: {'lr': 1.601775513388204e-05, 'samples': 22785024, 'steps': 44501, 'loss/train': 1.3802087306976318}
+02/26/2022 11:12:58 - INFO - codeparrot_training - Step 44502: {'lr': 1.601199295940117e-05, 'samples': 22785536, 'steps': 44502, 'loss/train': 1.8513593673706055}
+02/26/2022 11:13:03 - INFO - codeparrot_training - Step 44503: {'lr': 1.6006231787250552e-05, 'samples': 22786048, 'steps': 44503, 'loss/train': 1.5561649799346924}
+02/26/2022 11:13:06 - INFO - codeparrot_training - Step 44504: {'lr': 1.600047161745505e-05, 'samples': 22786560, 'steps': 44504, 'loss/train': 1.5796668529510498}
+02/26/2022 11:13:12 - INFO - codeparrot_training - Step 44505: {'lr': 1.5994712450039112e-05, 'samples': 22787072, 'steps': 44505, 'loss/train': 2.1288797855377197}
+02/26/2022 11:13:16 - INFO - codeparrot_training - Step 44506: {'lr': 1.5988954285027586e-05, 'samples': 22787584, 'steps': 44506, 'loss/train': 1.2501821517944336}
+02/26/2022 11:13:23 - INFO - codeparrot_training - Step 44507: {'lr': 1.5983197122445004e-05, 'samples': 22788096, 'steps': 44507, 'loss/train': 1.4690552949905396}
+02/26/2022 11:13:26 - INFO - codeparrot_training - Step 44508: {'lr': 1.5977440962316208e-05, 'samples': 22788608, 'steps': 44508, 'loss/train': 0.3202507793903351}
+02/26/2022 11:13:32 - INFO - codeparrot_training - Step 44509: {'lr': 1.5971685804665654e-05, 'samples': 22789120, 'steps': 44509, 'loss/train': 1.7949697971343994}
+02/26/2022 11:13:35 - INFO - codeparrot_training - Step 44510: {'lr': 1.5965931649518122e-05, 'samples': 22789632, 'steps': 44510, 'loss/train': 1.3624011278152466}
+02/26/2022 11:13:41 - INFO - codeparrot_training - Step 44511: {'lr': 1.596017849689818e-05, 'samples': 22790144, 'steps': 44511, 'loss/train': 1.8824430704116821}
+02/26/2022 11:13:45 - INFO - codeparrot_training - Step 44512: {'lr': 1.5954426346830558e-05, 'samples': 22790656, 'steps': 44512, 'loss/train': 1.8574906587600708}
+02/26/2022 11:13:50 - INFO - codeparrot_training - Step 44513: {'lr': 1.5948675199339873e-05, 'samples': 22791168, 'steps': 44513, 'loss/train': 2.2995009422302246}
+02/26/2022 11:13:53 - INFO - codeparrot_training - Step 44514: {'lr': 1.5942925054450718e-05, 'samples': 22791680, 'steps': 44514, 'loss/train': 1.6055896282196045}
+02/26/2022 11:13:59 - INFO - codeparrot_training - Step 44515: {'lr': 1.593717591218774e-05, 'samples': 22792192, 'steps': 44515, 'loss/train': 2.008331298828125}
+02/26/2022 11:14:03 - INFO - codeparrot_training - Step 44516: {'lr': 1.5931427772575585e-05, 'samples': 22792704, 'steps': 44516, 'loss/train': 8.181538581848145}
+02/26/2022 11:14:10 - INFO - codeparrot_training - Step 44517: {'lr': 1.592568063563887e-05, 'samples': 22793216, 'steps': 44517, 'loss/train': 0.44485682249069214}
+02/26/2022 11:14:13 - INFO - codeparrot_training - Step 44518: {'lr': 1.591993450140225e-05, 'samples': 22793728, 'steps': 44518, 'loss/train': 1.7015061378479004}
+02/26/2022 11:14:19 - INFO - codeparrot_training - Step 44519: {'lr': 1.5914189369890226e-05, 'samples': 22794240, 'steps': 44519, 'loss/train': 1.489059567451477}
+02/26/2022 11:14:22 - INFO - codeparrot_training - Step 44520: {'lr': 1.5908445241127527e-05, 'samples': 22794752, 'steps': 44520, 'loss/train': 0.8103165626525879}
+02/26/2022 11:14:28 - INFO - codeparrot_training - Step 44521: {'lr': 1.5902702115138752e-05, 'samples': 22795264, 'steps': 44521, 'loss/train': 2.0369951725006104}
+02/26/2022 11:14:31 - INFO - codeparrot_training - Step 44522: {'lr': 1.5896959991948428e-05, 'samples': 22795776, 'steps': 44522, 'loss/train': 2.053330421447754}
+02/26/2022 11:14:37 - INFO - codeparrot_training - Step 44523: {'lr': 1.589121887158121e-05, 'samples': 22796288, 'steps': 44523, 'loss/train': 2.026573419570923}
+02/26/2022 11:14:40 - INFO - codeparrot_training - Step 44524: {'lr': 1.5885478754061627e-05, 'samples': 22796800, 'steps': 44524, 'loss/train': 0.892111599445343}
+02/26/2022 11:14:46 - INFO - codeparrot_training - Step 44525: {'lr': 1.587973963941433e-05, 'samples': 22797312, 'steps': 44525, 'loss/train': 0.5852540731430054}
+02/26/2022 11:14:49 - INFO - codeparrot_training - Step 44526: {'lr': 1.587400152766391e-05, 'samples': 22797824, 'steps': 44526, 'loss/train': 1.5767624378204346}
+02/26/2022 11:14:56 - INFO - codeparrot_training - Step 44527: {'lr': 1.5868264418834933e-05, 'samples': 22798336, 'steps': 44527, 'loss/train': 1.603103518486023}
+02/26/2022 11:15:00 - INFO - codeparrot_training - Step 44528: {'lr': 1.586252831295193e-05, 'samples': 22798848, 'steps': 44528, 'loss/train': 1.9427510499954224}
+02/26/2022 11:15:05 - INFO - codeparrot_training - Step 44529: {'lr': 1.585679321003952e-05, 'samples': 22799360, 'steps': 44529, 'loss/train': 0.6830424666404724}
+02/26/2022 11:15:09 - INFO - codeparrot_training - Step 44530: {'lr': 1.5851059110122275e-05, 'samples': 22799872, 'steps': 44530, 'loss/train': 1.0778858661651611}
+02/26/2022 11:15:15 - INFO - codeparrot_training - Step 44531: {'lr': 1.584532601322475e-05, 'samples': 22800384, 'steps': 44531, 'loss/train': 1.698647141456604}
+02/26/2022 11:15:18 - INFO - codeparrot_training - Step 44532: {'lr': 1.5839593919371425e-05, 'samples': 22800896, 'steps': 44532, 'loss/train': 0.4144231379032135}
+02/26/2022 11:15:24 - INFO - codeparrot_training - Step 44533: {'lr': 1.583386282858701e-05, 'samples': 22801408, 'steps': 44533, 'loss/train': 1.755776286125183}
+02/26/2022 11:15:27 - INFO - codeparrot_training - Step 44534: {'lr': 1.5828132740895896e-05, 'samples': 22801920, 'steps': 44534, 'loss/train': 1.4762258529663086}
+02/26/2022 11:15:33 - INFO - codeparrot_training - Step 44535: {'lr': 1.5822403656322786e-05, 'samples': 22802432, 'steps': 44535, 'loss/train': 2.0760912895202637}
+02/26/2022 11:15:36 - INFO - codeparrot_training - Step 44536: {'lr': 1.5816675574892082e-05, 'samples': 22802944, 'steps': 44536, 'loss/train': 2.3944971561431885}
+02/26/2022 11:15:42 - INFO - codeparrot_training - Step 44537: {'lr': 1.5810948496628396e-05, 'samples': 22803456, 'steps': 44537, 'loss/train': 2.216747760772705}
+02/26/2022 11:15:46 - INFO - codeparrot_training - Step 44538: {'lr': 1.5805222421556215e-05, 'samples': 22803968, 'steps': 44538, 'loss/train': 2.25126051902771}
+02/26/2022 11:15:51 - INFO - codeparrot_training - Step 44539: {'lr': 1.579949734970018e-05, 'samples': 22804480, 'steps': 44539, 'loss/train': 2.663761854171753}
+02/26/2022 11:15:55 - INFO - codeparrot_training - Step 44540: {'lr': 1.579377328108464e-05, 'samples': 22804992, 'steps': 44540, 'loss/train': 0.8956112265586853}
+02/26/2022 11:16:02 - INFO - codeparrot_training - Step 44541: {'lr': 1.5788050215734235e-05, 'samples': 22805504, 'steps': 44541, 'loss/train': 2.7742488384246826}
+02/26/2022 11:16:05 - INFO - codeparrot_training - Step 44542: {'lr': 1.5782328153673447e-05, 'samples': 22806016, 'steps': 44542, 'loss/train': 1.8063017129898071}
+02/26/2022 11:16:11 - INFO - codeparrot_training - Step 44543: {'lr': 1.5776607094926843e-05, 'samples': 22806528, 'steps': 44543, 'loss/train': 2.1666834354400635}
+02/26/2022 11:16:14 - INFO - codeparrot_training - Step 44544: {'lr': 1.5770887039518816e-05, 'samples': 22807040, 'steps': 44544, 'loss/train': 2.014113187789917}
+02/26/2022 11:16:20 - INFO - codeparrot_training - Step 44545: {'lr': 1.576516798747396e-05, 'samples': 22807552, 'steps': 44545, 'loss/train': 0.9947926998138428}
+02/26/2022 11:16:23 - INFO - codeparrot_training - Step 44546: {'lr': 1.5759449938816727e-05, 'samples': 22808064, 'steps': 44546, 'loss/train': 1.1864845752716064}
+02/26/2022 11:16:29 - INFO - codeparrot_training - Step 44547: {'lr': 1.5753732893571653e-05, 'samples': 22808576, 'steps': 44547, 'loss/train': 1.9343748092651367}
+02/26/2022 11:16:32 - INFO - codeparrot_training - Step 44548: {'lr': 1.5748016851763192e-05, 'samples': 22809088, 'steps': 44548, 'loss/train': 2.335641384124756}
+02/26/2022 11:16:38 - INFO - codeparrot_training - Step 44549: {'lr': 1.5742301813415876e-05, 'samples': 22809600, 'steps': 44549, 'loss/train': 1.488487720489502}
+02/26/2022 11:16:41 - INFO - codeparrot_training - Step 44550: {'lr': 1.5736587778554107e-05, 'samples': 22810112, 'steps': 44550, 'loss/train': 1.9591190814971924}
+02/26/2022 11:16:47 - INFO - codeparrot_training - Step 44551: {'lr': 1.5730874747202418e-05, 'samples': 22810624, 'steps': 44551, 'loss/train': 1.547108769416809}
+02/26/2022 11:16:50 - INFO - codeparrot_training - Step 44552: {'lr': 1.5725162719385315e-05, 'samples': 22811136, 'steps': 44552, 'loss/train': 1.6096739768981934}
+02/26/2022 11:16:54 - INFO - codeparrot_training - Step 44553: {'lr': 1.57194516951272e-05, 'samples': 22811648, 'steps': 44553, 'loss/train': 2.193249464035034}
+02/26/2022 11:17:01 - INFO - codeparrot_training - Step 44554: {'lr': 1.5713741674452526e-05, 'samples': 22812160, 'steps': 44554, 'loss/train': 1.7400882244110107}
+02/26/2022 11:17:05 - INFO - codeparrot_training - Step 44555: {'lr': 1.5708032657385825e-05, 'samples': 22812672, 'steps': 44555, 'loss/train': 2.021022319793701}
+02/26/2022 11:17:10 - INFO - codeparrot_training - Step 44556: {'lr': 1.5702324643951522e-05, 'samples': 22813184, 'steps': 44556, 'loss/train': 1.346304178237915}
+02/26/2022 11:17:14 - INFO - codeparrot_training - Step 44557: {'lr': 1.5696617634174043e-05, 'samples': 22813696, 'steps': 44557, 'loss/train': 1.8561760187149048}
+02/26/2022 11:17:19 - INFO - codeparrot_training - Step 44558: {'lr': 1.5690911628077843e-05, 'samples': 22814208, 'steps': 44558, 'loss/train': 1.738935947418213}
+02/26/2022 11:17:23 - INFO - codeparrot_training - Step 44559: {'lr': 1.56852066256874e-05, 'samples': 22814720, 'steps': 44559, 'loss/train': 1.6938575506210327}
+02/26/2022 11:17:28 - INFO - codeparrot_training - Step 44560: {'lr': 1.5679502627027137e-05, 'samples': 22815232, 'steps': 44560, 'loss/train': 2.8369622230529785}
+02/26/2022 11:17:32 - INFO - codeparrot_training - Step 44561: {'lr': 1.5673799632121484e-05, 'samples': 22815744, 'steps': 44561, 'loss/train': 2.1059303283691406}
+02/26/2022 11:17:37 - INFO - codeparrot_training - Step 44562: {'lr': 1.5668097640994804e-05, 'samples': 22816256, 'steps': 44562, 'loss/train': 1.4703333377838135}
+02/26/2022 11:17:40 - INFO - codeparrot_training - Step 44563: {'lr': 1.566239665367164e-05, 'samples': 22816768, 'steps': 44563, 'loss/train': 1.5801500082015991}
+02/26/2022 11:17:48 - INFO - codeparrot_training - Step 44564: {'lr': 1.565669667017636e-05, 'samples': 22817280, 'steps': 44564, 'loss/train': 2.679043769836426}
+02/26/2022 11:17:51 - INFO - codeparrot_training - Step 44565: {'lr': 1.565099769053338e-05, 'samples': 22817792, 'steps': 44565, 'loss/train': 2.0830371379852295}
+02/26/2022 11:17:57 - INFO - codeparrot_training - Step 44566: {'lr': 1.564529971476708e-05, 'samples': 22818304, 'steps': 44566, 'loss/train': 0.9572405219078064}
+02/26/2022 11:18:01 - INFO - codeparrot_training - Step 44567: {'lr': 1.563960274290191e-05, 'samples': 22818816, 'steps': 44567, 'loss/train': 2.737130641937256}
+02/26/2022 11:18:06 - INFO - codeparrot_training - Step 44568: {'lr': 1.563390677496229e-05, 'samples': 22819328, 'steps': 44568, 'loss/train': 1.5434027910232544}
+02/26/2022 11:18:10 - INFO - codeparrot_training - Step 44569: {'lr': 1.5628211810972543e-05, 'samples': 22819840, 'steps': 44569, 'loss/train': 1.8749423027038574}
+02/26/2022 11:18:15 - INFO - codeparrot_training - Step 44570: {'lr': 1.5622517850957168e-05, 'samples': 22820352, 'steps': 44570, 'loss/train': 1.623382329940796}
+02/26/2022 11:18:19 - INFO - codeparrot_training - Step 44571: {'lr': 1.5616824894940485e-05, 'samples': 22820864, 'steps': 44571, 'loss/train': 1.5651400089263916}
+02/26/2022 11:18:24 - INFO - codeparrot_training - Step 44572: {'lr': 1.5611132942946914e-05, 'samples': 22821376, 'steps': 44572, 'loss/train': 1.0098519325256348}
+02/26/2022 11:18:28 - INFO - codeparrot_training - Step 44573: {'lr': 1.5605441995000774e-05, 'samples': 22821888, 'steps': 44573, 'loss/train': 1.1188361644744873}
+02/26/2022 11:18:35 - INFO - codeparrot_training - Step 44574: {'lr': 1.559975205112657e-05, 'samples': 22822400, 'steps': 44574, 'loss/train': 0.8829925656318665}
+02/26/2022 11:18:38 - INFO - codeparrot_training - Step 44575: {'lr': 1.5594063111348504e-05, 'samples': 22822912, 'steps': 44575, 'loss/train': 1.6833213567733765}
+02/26/2022 11:18:44 - INFO - codeparrot_training - Step 44576: {'lr': 1.5588375175691116e-05, 'samples': 22823424, 'steps': 44576, 'loss/train': 0.3324960768222809}
+02/26/2022 11:18:47 - INFO - codeparrot_training - Step 44577: {'lr': 1.5582688244178633e-05, 'samples': 22823936, 'steps': 44577, 'loss/train': 1.344488501548767}
+02/26/2022 11:18:53 - INFO - codeparrot_training - Step 44578: {'lr': 1.5577002316835563e-05, 'samples': 22824448, 'steps': 44578, 'loss/train': 1.4971399307250977}
+02/26/2022 11:18:57 - INFO - codeparrot_training - Step 44579: {'lr': 1.5571317393686085e-05, 'samples': 22824960, 'steps': 44579, 'loss/train': 1.6542730331420898}
+02/26/2022 11:19:02 - INFO - codeparrot_training - Step 44580: {'lr': 1.5565633474754702e-05, 'samples': 22825472, 'steps': 44580, 'loss/train': 1.607014536857605}
+02/26/2022 11:19:06 - INFO - codeparrot_training - Step 44581: {'lr': 1.555995056006565e-05, 'samples': 22825984, 'steps': 44581, 'loss/train': 2.12030029296875}
+02/26/2022 11:19:11 - INFO - codeparrot_training - Step 44582: {'lr': 1.5554268649643378e-05, 'samples': 22826496, 'steps': 44582, 'loss/train': 0.6701213717460632}
+02/26/2022 11:19:15 - INFO - codeparrot_training - Step 44583: {'lr': 1.5548587743512172e-05, 'samples': 22827008, 'steps': 44583, 'loss/train': 1.2296571731567383}
+02/26/2022 11:19:22 - INFO - codeparrot_training - Step 44584: {'lr': 1.5542907841696347e-05, 'samples': 22827520, 'steps': 44584, 'loss/train': 1.1790920495986938}
+02/26/2022 11:19:25 - INFO - codeparrot_training - Step 44585: {'lr': 1.5537228944220245e-05, 'samples': 22828032, 'steps': 44585, 'loss/train': 0.9743176102638245}
+02/26/2022 11:19:31 - INFO - codeparrot_training - Step 44586: {'lr': 1.5531551051108235e-05, 'samples': 22828544, 'steps': 44586, 'loss/train': 1.2688344717025757}
+02/26/2022 11:19:35 - INFO - codeparrot_training - Step 44587: {'lr': 1.55258741623846e-05, 'samples': 22829056, 'steps': 44587, 'loss/train': 1.8284367322921753}
+02/26/2022 11:19:40 - INFO - codeparrot_training - Step 44588: {'lr': 1.552019827807369e-05, 'samples': 22829568, 'steps': 44588, 'loss/train': 0.9335541129112244}
+02/26/2022 11:19:44 - INFO - codeparrot_training - Step 44589: {'lr': 1.5514523398199724e-05, 'samples': 22830080, 'steps': 44589, 'loss/train': 1.7284339666366577}
+02/26/2022 11:19:50 - INFO - codeparrot_training - Step 44590: {'lr': 1.550884952278714e-05, 'samples': 22830592, 'steps': 44590, 'loss/train': 1.8680366277694702}
+02/26/2022 11:19:54 - INFO - codeparrot_training - Step 44591: {'lr': 1.5503176651860183e-05, 'samples': 22831104, 'steps': 44591, 'loss/train': 1.9334073066711426}
+02/26/2022 11:19:59 - INFO - codeparrot_training - Step 44592: {'lr': 1.5497504785443152e-05, 'samples': 22831616, 'steps': 44592, 'loss/train': 2.0922441482543945}
+02/26/2022 11:20:03 - INFO - codeparrot_training - Step 44593: {'lr': 1.5491833923560294e-05, 'samples': 22832128, 'steps': 44593, 'loss/train': 2.21962571144104}
+02/26/2022 11:20:08 - INFO - codeparrot_training - Step 44594: {'lr': 1.5486164066236014e-05, 'samples': 22832640, 'steps': 44594, 'loss/train': 1.5537835359573364}
+02/26/2022 11:20:12 - INFO - codeparrot_training - Step 44595: {'lr': 1.548049521349451e-05, 'samples': 22833152, 'steps': 44595, 'loss/train': 1.1310774087905884}
+02/26/2022 11:20:17 - INFO - codeparrot_training - Step 44596: {'lr': 1.5474827365360157e-05, 'samples': 22833664, 'steps': 44596, 'loss/train': 1.4799506664276123}
+02/26/2022 11:20:21 - INFO - codeparrot_training - Step 44597: {'lr': 1.5469160521857095e-05, 'samples': 22834176, 'steps': 44597, 'loss/train': 1.1180918216705322}
+02/26/2022 11:20:26 - INFO - codeparrot_training - Step 44598: {'lr': 1.5463494683009725e-05, 'samples': 22834688, 'steps': 44598, 'loss/train': 2.638831615447998}
+02/26/2022 11:20:30 - INFO - codeparrot_training - Step 44599: {'lr': 1.5457829848842226e-05, 'samples': 22835200, 'steps': 44599, 'loss/train': 2.268336296081543}
+02/26/2022 11:20:37 - INFO - codeparrot_training - Step 44600: {'lr': 1.5452166019378987e-05, 'samples': 22835712, 'steps': 44600, 'loss/train': 1.7282999753952026}
+02/26/2022 11:20:41 - INFO - codeparrot_training - Step 44601: {'lr': 1.5446503194644136e-05, 'samples': 22836224, 'steps': 44601, 'loss/train': 1.4947994947433472}
+02/26/2022 11:20:46 - INFO - codeparrot_training - Step 44602: {'lr': 1.544084137466198e-05, 'samples': 22836736, 'steps': 44602, 'loss/train': 1.7952890396118164}
+02/26/2022 11:20:50 - INFO - codeparrot_training - Step 44603: {'lr': 1.543518055945678e-05, 'samples': 22837248, 'steps': 44603, 'loss/train': 2.8607635498046875}
+02/26/2022 11:20:55 - INFO - codeparrot_training - Step 44604: {'lr': 1.542952074905285e-05, 'samples': 22837760, 'steps': 44604, 'loss/train': 2.0992369651794434}
+02/26/2022 11:20:59 - INFO - codeparrot_training - Step 44605: {'lr': 1.5423861943474282e-05, 'samples': 22838272, 'steps': 44605, 'loss/train': 1.0394814014434814}
+02/26/2022 11:21:04 - INFO - codeparrot_training - Step 44606: {'lr': 1.5418204142745447e-05, 'samples': 22838784, 'steps': 44606, 'loss/train': 1.0934089422225952}
+02/26/2022 11:21:08 - INFO - codeparrot_training - Step 44607: {'lr': 1.5412547346890543e-05, 'samples': 22839296, 'steps': 44607, 'loss/train': 2.3242218494415283}
+02/26/2022 11:21:13 - INFO - codeparrot_training - Step 44608: {'lr': 1.540689155593375e-05, 'samples': 22839808, 'steps': 44608, 'loss/train': 1.1360435485839844}
+02/26/2022 11:21:17 - INFO - codeparrot_training - Step 44609: {'lr': 1.540123676989938e-05, 'samples': 22840320, 'steps': 44609, 'loss/train': 2.215571165084839}
+02/26/2022 11:21:24 - INFO - codeparrot_training - Step 44610: {'lr': 1.539558298881161e-05, 'samples': 22840832, 'steps': 44610, 'loss/train': 1.9261587858200073}
+02/26/2022 11:21:27 - INFO - codeparrot_training - Step 44611: {'lr': 1.5389930212694664e-05, 'samples': 22841344, 'steps': 44611, 'loss/train': 0.9890658259391785}
+02/26/2022 11:21:33 - INFO - codeparrot_training - Step 44612: {'lr': 1.5384278441572754e-05, 'samples': 22841856, 'steps': 44612, 'loss/train': 2.372225761413574}
+02/26/2022 11:21:36 - INFO - codeparrot_training - Step 44613: {'lr': 1.5378627675470103e-05, 'samples': 22842368, 'steps': 44613, 'loss/train': 1.7916009426116943}
+02/26/2022 11:21:42 - INFO - codeparrot_training - Step 44614: {'lr': 1.537297791441092e-05, 'samples': 22842880, 'steps': 44614, 'loss/train': 1.7261452674865723}
+02/26/2022 11:21:45 - INFO - codeparrot_training - Step 44615: {'lr': 1.5367329158419404e-05, 'samples': 22843392, 'steps': 44615, 'loss/train': 2.048445701599121}
+02/26/2022 11:21:51 - INFO - codeparrot_training - Step 44616: {'lr': 1.53616814075197e-05, 'samples': 22843904, 'steps': 44616, 'loss/train': 0.3136724829673767}
+02/26/2022 11:21:54 - INFO - codeparrot_training - Step 44617: {'lr': 1.53560346617361e-05, 'samples': 22844416, 'steps': 44617, 'loss/train': 1.753232479095459}
+02/26/2022 11:22:00 - INFO - codeparrot_training - Step 44618: {'lr': 1.5350388921092723e-05, 'samples': 22844928, 'steps': 44618, 'loss/train': 1.9669389724731445}
+02/26/2022 11:22:03 - INFO - codeparrot_training - Step 44619: {'lr': 1.5344744185613768e-05, 'samples': 22845440, 'steps': 44619, 'loss/train': 0.9677431583404541}
+02/26/2022 11:22:11 - INFO - codeparrot_training - Step 44620: {'lr': 1.5339100455323384e-05, 'samples': 22845952, 'steps': 44620, 'loss/train': 1.199379324913025}
+02/26/2022 11:22:14 - INFO - codeparrot_training - Step 44621: {'lr': 1.533345773024583e-05, 'samples': 22846464, 'steps': 44621, 'loss/train': 2.422481060028076}
+02/26/2022 11:22:20 - INFO - codeparrot_training - Step 44622: {'lr': 1.5327816010405227e-05, 'samples': 22846976, 'steps': 44622, 'loss/train': 1.6680974960327148}
+02/26/2022 11:22:23 - INFO - codeparrot_training - Step 44623: {'lr': 1.532217529582572e-05, 'samples': 22847488, 'steps': 44623, 'loss/train': 2.0155797004699707}
+02/26/2022 11:22:29 - INFO - codeparrot_training - Step 44624: {'lr': 1.5316535586531482e-05, 'samples': 22848000, 'steps': 44624, 'loss/train': 0.9815978407859802}
+02/26/2022 11:22:32 - INFO - codeparrot_training - Step 44625: {'lr': 1.531089688254672e-05, 'samples': 22848512, 'steps': 44625, 'loss/train': 2.4685962200164795}
+02/26/2022 11:22:38 - INFO - codeparrot_training - Step 44626: {'lr': 1.5305259183895527e-05, 'samples': 22849024, 'steps': 44626, 'loss/train': 1.4522428512573242}
+02/26/2022 11:22:41 - INFO - codeparrot_training - Step 44627: {'lr': 1.5299622490602098e-05, 'samples': 22849536, 'steps': 44627, 'loss/train': 1.910078763961792}
+02/26/2022 11:22:47 - INFO - codeparrot_training - Step 44628: {'lr': 1.5293986802690533e-05, 'samples': 22850048, 'steps': 44628, 'loss/train': 1.6691898107528687}
+02/26/2022 11:22:50 - INFO - codeparrot_training - Step 44629: {'lr': 1.5288352120185033e-05, 'samples': 22850560, 'steps': 44629, 'loss/train': 1.1679444313049316}
+02/26/2022 11:22:58 - INFO - codeparrot_training - Step 44630: {'lr': 1.528271844310966e-05, 'samples': 22851072, 'steps': 44630, 'loss/train': 1.4711912870407104}
+02/26/2022 11:23:01 - INFO - codeparrot_training - Step 44631: {'lr': 1.5277085771488676e-05, 'samples': 22851584, 'steps': 44631, 'loss/train': 1.7267378568649292}
+02/26/2022 11:23:07 - INFO - codeparrot_training - Step 44632: {'lr': 1.5271454105346028e-05, 'samples': 22852096, 'steps': 44632, 'loss/train': 1.5025639533996582}
+02/26/2022 11:23:10 - INFO - codeparrot_training - Step 44633: {'lr': 1.5265823444706007e-05, 'samples': 22852608, 'steps': 44633, 'loss/train': 8.783032417297363}
+02/26/2022 11:23:16 - INFO - codeparrot_training - Step 44634: {'lr': 1.526019378959259e-05, 'samples': 22853120, 'steps': 44634, 'loss/train': 1.2788257598876953}
+02/26/2022 11:23:19 - INFO - codeparrot_training - Step 44635: {'lr': 1.525456514003004e-05, 'samples': 22853632, 'steps': 44635, 'loss/train': 1.1093562841415405}
+02/26/2022 11:23:25 - INFO - codeparrot_training - Step 44636: {'lr': 1.5248937496042337e-05, 'samples': 22854144, 'steps': 44636, 'loss/train': 1.7389986515045166}
+02/26/2022 11:23:28 - INFO - codeparrot_training - Step 44637: {'lr': 1.5243310857653681e-05, 'samples': 22854656, 'steps': 44637, 'loss/train': 0.5567581057548523}
+02/26/2022 11:23:34 - INFO - codeparrot_training - Step 44638: {'lr': 1.5237685224888081e-05, 'samples': 22855168, 'steps': 44638, 'loss/train': 2.364442825317383}
+02/26/2022 11:23:37 - INFO - codeparrot_training - Step 44639: {'lr': 1.5232060597769797e-05, 'samples': 22855680, 'steps': 44639, 'loss/train': 1.1462093591690063}
+02/26/2022 11:23:43 - INFO - codeparrot_training - Step 44640: {'lr': 1.5226436976322727e-05, 'samples': 22856192, 'steps': 44640, 'loss/train': 1.674525260925293}
+02/26/2022 11:23:46 - INFO - codeparrot_training - Step 44641: {'lr': 1.5220814360571072e-05, 'samples': 22856704, 'steps': 44641, 'loss/train': 3.4201090335845947}
+02/26/2022 11:23:52 - INFO - codeparrot_training - Step 44642: {'lr': 1.5215192750538897e-05, 'samples': 22857216, 'steps': 44642, 'loss/train': 1.2100610733032227}
+02/26/2022 11:23:55 - INFO - codeparrot_training - Step 44643: {'lr': 1.5209572146250295e-05, 'samples': 22857728, 'steps': 44643, 'loss/train': 1.5242233276367188}
+02/26/2022 11:24:00 - INFO - codeparrot_training - Step 44644: {'lr': 1.5203952547729328e-05, 'samples': 22858240, 'steps': 44644, 'loss/train': 1.9821656942367554}
+02/26/2022 11:24:04 - INFO - codeparrot_training - Step 44645: {'lr': 1.5198333955000088e-05, 'samples': 22858752, 'steps': 44645, 'loss/train': 1.4836583137512207}
+02/26/2022 11:24:11 - INFO - codeparrot_training - Step 44646: {'lr': 1.5192716368086585e-05, 'samples': 22859264, 'steps': 44646, 'loss/train': 1.4414249658584595}
+02/26/2022 11:24:15 - INFO - codeparrot_training - Step 44647: {'lr': 1.5187099787012965e-05, 'samples': 22859776, 'steps': 44647, 'loss/train': 1.9191162586212158}
+02/26/2022 11:24:20 - INFO - codeparrot_training - Step 44648: {'lr': 1.5181484211803238e-05, 'samples': 22860288, 'steps': 44648, 'loss/train': 1.0874743461608887}
+02/26/2022 11:24:23 - INFO - codeparrot_training - Step 44649: {'lr': 1.5175869642481494e-05, 'samples': 22860800, 'steps': 44649, 'loss/train': 1.8841404914855957}
+02/26/2022 11:24:29 - INFO - codeparrot_training - Step 44650: {'lr': 1.5170256079071687e-05, 'samples': 22861312, 'steps': 44650, 'loss/train': 1.0572062730789185}
+02/26/2022 11:24:32 - INFO - codeparrot_training - Step 44651: {'lr': 1.5164643521597993e-05, 'samples': 22861824, 'steps': 44651, 'loss/train': 1.6270796060562134}
+02/26/2022 11:24:38 - INFO - codeparrot_training - Step 44652: {'lr': 1.5159031970084391e-05, 'samples': 22862336, 'steps': 44652, 'loss/train': 0.7014657258987427}
+02/26/2022 11:24:41 - INFO - codeparrot_training - Step 44653: {'lr': 1.5153421424554947e-05, 'samples': 22862848, 'steps': 44653, 'loss/train': 0.9855467677116394}
+02/26/2022 11:24:47 - INFO - codeparrot_training - Step 44654: {'lr': 1.5147811885033641e-05, 'samples': 22863360, 'steps': 44654, 'loss/train': 1.1339077949523926}
+02/26/2022 11:24:51 - INFO - codeparrot_training - Step 44655: {'lr': 1.514220335154451e-05, 'samples': 22863872, 'steps': 44655, 'loss/train': 1.8605103492736816}
+02/26/2022 11:24:58 - INFO - codeparrot_training - Step 44656: {'lr': 1.5136595824111642e-05, 'samples': 22864384, 'steps': 44656, 'loss/train': 1.6077550649642944}
+02/26/2022 11:25:01 - INFO - codeparrot_training - Step 44657: {'lr': 1.5130989302759025e-05, 'samples': 22864896, 'steps': 44657, 'loss/train': 2.056802749633789}
+02/26/2022 11:25:07 - INFO - codeparrot_training - Step 44658: {'lr': 1.512538378751066e-05, 'samples': 22865408, 'steps': 44658, 'loss/train': 2.618234634399414}
+02/26/2022 11:25:10 - INFO - codeparrot_training - Step 44659: {'lr': 1.5119779278390561e-05, 'samples': 22865920, 'steps': 44659, 'loss/train': 2.2130839824676514}
+02/26/2022 11:25:16 - INFO - codeparrot_training - Step 44660: {'lr': 1.5114175775422761e-05, 'samples': 22866432, 'steps': 44660, 'loss/train': 1.6912882328033447}
+02/26/2022 11:25:19 - INFO - codeparrot_training - Step 44661: {'lr': 1.5108573278631243e-05, 'samples': 22866944, 'steps': 44661, 'loss/train': 1.4367542266845703}
+02/26/2022 11:25:25 - INFO - codeparrot_training - Step 44662: {'lr': 1.5102971788040014e-05, 'samples': 22867456, 'steps': 44662, 'loss/train': 1.1090903282165527}
+02/26/2022 11:25:28 - INFO - codeparrot_training - Step 44663: {'lr': 1.509737130367303e-05, 'samples': 22867968, 'steps': 44663, 'loss/train': 1.3047922849655151}
+02/26/2022 11:25:34 - INFO - codeparrot_training - Step 44664: {'lr': 1.509177182555435e-05, 'samples': 22868480, 'steps': 44664, 'loss/train': 1.6728254556655884}
+02/26/2022 11:25:37 - INFO - codeparrot_training - Step 44665: {'lr': 1.5086173353707932e-05, 'samples': 22868992, 'steps': 44665, 'loss/train': 2.834684371948242}
+02/26/2022 11:25:44 - INFO - codeparrot_training - Step 44666: {'lr': 1.5080575888157755e-05, 'samples': 22869504, 'steps': 44666, 'loss/train': 0.025383299216628075}
+02/26/2022 11:25:48 - INFO - codeparrot_training - Step 44667: {'lr': 1.5074979428927743e-05, 'samples': 22870016, 'steps': 44667, 'loss/train': 2.3409533500671387}
+02/26/2022 11:25:53 - INFO - codeparrot_training - Step 44668: {'lr': 1.506938397604199e-05, 'samples': 22870528, 'steps': 44668, 'loss/train': 2.192551374435425}
+02/26/2022 11:25:57 - INFO - codeparrot_training - Step 44669: {'lr': 1.5063789529524336e-05, 'samples': 22871040, 'steps': 44669, 'loss/train': 1.894571304321289}
+02/26/2022 11:26:02 - INFO - codeparrot_training - Step 44670: {'lr': 1.5058196089398901e-05, 'samples': 22871552, 'steps': 44670, 'loss/train': 1.6740198135375977}
+02/26/2022 11:26:06 - INFO - codeparrot_training - Step 44671: {'lr': 1.5052603655689445e-05, 'samples': 22872064, 'steps': 44671, 'loss/train': 1.7244845628738403}
+02/26/2022 11:26:11 - INFO - codeparrot_training - Step 44672: {'lr': 1.5047012228420088e-05, 'samples': 22872576, 'steps': 44672, 'loss/train': 1.27073073387146}
+02/26/2022 11:26:15 - INFO - codeparrot_training - Step 44673: {'lr': 1.5041421807614697e-05, 'samples': 22873088, 'steps': 44673, 'loss/train': 1.600583791732788}
+02/26/2022 11:26:20 - INFO - codeparrot_training - Step 44674: {'lr': 1.5035832393297311e-05, 'samples': 22873600, 'steps': 44674, 'loss/train': 0.4095900356769562}
+02/26/2022 11:26:24 - INFO - codeparrot_training - Step 44675: {'lr': 1.5030243985491743e-05, 'samples': 22874112, 'steps': 44675, 'loss/train': 1.4403547048568726}
+02/26/2022 11:26:31 - INFO - codeparrot_training - Step 44676: {'lr': 1.5024656584222029e-05, 'samples': 22874624, 'steps': 44676, 'loss/train': 1.2183821201324463}
+02/26/2022 11:26:34 - INFO - codeparrot_training - Step 44677: {'lr': 1.5019070189512014e-05, 'samples': 22875136, 'steps': 44677, 'loss/train': 2.425318717956543}
+02/26/2022 11:26:40 - INFO - codeparrot_training - Step 44678: {'lr': 1.5013484801385758e-05, 'samples': 22875648, 'steps': 44678, 'loss/train': 2.4563510417938232}
+02/26/2022 11:26:44 - INFO - codeparrot_training - Step 44679: {'lr': 1.5007900419867104e-05, 'samples': 22876160, 'steps': 44679, 'loss/train': 1.4586529731750488}
+02/26/2022 11:26:49 - INFO - codeparrot_training - Step 44680: {'lr': 1.5002317044979979e-05, 'samples': 22876672, 'steps': 44680, 'loss/train': 2.079988479614258}
+02/26/2022 11:26:53 - INFO - codeparrot_training - Step 44681: {'lr': 1.499673467674828e-05, 'samples': 22877184, 'steps': 44681, 'loss/train': 0.6317312717437744}
+02/26/2022 11:26:58 - INFO - codeparrot_training - Step 44682: {'lr': 1.4991153315195987e-05, 'samples': 22877696, 'steps': 44682, 'loss/train': 0.9572232961654663}
+02/26/2022 11:27:02 - INFO - codeparrot_training - Step 44683: {'lr': 1.4985572960346971e-05, 'samples': 22878208, 'steps': 44683, 'loss/train': 2.1569111347198486}
+02/26/2022 11:27:07 - INFO - codeparrot_training - Step 44684: {'lr': 1.497999361222513e-05, 'samples': 22878720, 'steps': 44684, 'loss/train': 1.4307478666305542}
+02/26/2022 11:27:11 - INFO - codeparrot_training - Step 44685: {'lr': 1.497441527085433e-05, 'samples': 22879232, 'steps': 44685, 'loss/train': 1.980157732963562}
+02/26/2022 11:27:16 - INFO - codeparrot_training - Step 44686: {'lr': 1.4968837936258528e-05, 'samples': 22879744, 'steps': 44686, 'loss/train': 1.2960748672485352}
+02/26/2022 11:27:20 - INFO - codeparrot_training - Step 44687: {'lr': 1.496326160846162e-05, 'samples': 22880256, 'steps': 44687, 'loss/train': 0.7699154019355774}
+02/26/2022 11:27:26 - INFO - codeparrot_training - Step 44688: {'lr': 1.4957686287487449e-05, 'samples': 22880768, 'steps': 44688, 'loss/train': 2.26187801361084}
+02/26/2022 11:27:29 - INFO - codeparrot_training - Step 44689: {'lr': 1.4952111973359884e-05, 'samples': 22881280, 'steps': 44689, 'loss/train': 0.2561168670654297}
+02/26/2022 11:27:33 - INFO - codeparrot_training - Step 44690: {'lr': 1.4946538666102877e-05, 'samples': 22881792, 'steps': 44690, 'loss/train': 1.0869572162628174}
+02/26/2022 11:27:38 - INFO - codeparrot_training - Step 44691: {'lr': 1.4940966365740272e-05, 'samples': 22882304, 'steps': 44691, 'loss/train': 1.7562129497528076}
+02/26/2022 11:27:42 - INFO - codeparrot_training - Step 44692: {'lr': 1.493539507229591e-05, 'samples': 22882816, 'steps': 44692, 'loss/train': 1.228003978729248}
+02/26/2022 11:27:49 - INFO - codeparrot_training - Step 44693: {'lr': 1.4929824785793634e-05, 'samples': 22883328, 'steps': 44693, 'loss/train': 2.1903302669525146}
+02/26/2022 11:27:53 - INFO - codeparrot_training - Step 44694: {'lr': 1.4924255506257395e-05, 'samples': 22883840, 'steps': 44694, 'loss/train': 0.9961408972740173}
+02/26/2022 11:27:58 - INFO - codeparrot_training - Step 44695: {'lr': 1.491868723371098e-05, 'samples': 22884352, 'steps': 44695, 'loss/train': 2.0601065158843994}
+02/26/2022 11:28:02 - INFO - codeparrot_training - Step 44696: {'lr': 1.4913119968178291e-05, 'samples': 22884864, 'steps': 44696, 'loss/train': 1.5648421049118042}
+02/26/2022 11:28:07 - INFO - codeparrot_training - Step 44697: {'lr': 1.4907553709683108e-05, 'samples': 22885376, 'steps': 44697, 'loss/train': 1.8498815298080444}
+02/26/2022 11:28:11 - INFO - codeparrot_training - Step 44698: {'lr': 1.4901988458249332e-05, 'samples': 22885888, 'steps': 44698, 'loss/train': 1.4192838668823242}
+02/26/2022 11:28:16 - INFO - codeparrot_training - Step 44699: {'lr': 1.4896424213900777e-05, 'samples': 22886400, 'steps': 44699, 'loss/train': 1.2419383525848389}
+02/26/2022 11:28:20 - INFO - codeparrot_training - Step 44700: {'lr': 1.4890860976661313e-05, 'samples': 22886912, 'steps': 44700, 'loss/train': 1.0514472723007202}
+02/26/2022 11:28:25 - INFO - codeparrot_training - Step 44701: {'lr': 1.4885298746554727e-05, 'samples': 22887424, 'steps': 44701, 'loss/train': 1.8465126752853394}
+02/26/2022 11:28:29 - INFO - codeparrot_training - Step 44702: {'lr': 1.487973752360483e-05, 'samples': 22887936, 'steps': 44702, 'loss/train': 1.3820991516113281}
+02/26/2022 11:28:36 - INFO - codeparrot_training - Step 44703: {'lr': 1.4874177307835523e-05, 'samples': 22888448, 'steps': 44703, 'loss/train': 1.9503690004348755}
+02/26/2022 11:28:40 - INFO - codeparrot_training - Step 44704: {'lr': 1.4868618099270536e-05, 'samples': 22888960, 'steps': 44704, 'loss/train': 1.6652113199234009}
+02/26/2022 11:28:45 - INFO - codeparrot_training - Step 44705: {'lr': 1.4863059897933823e-05, 'samples': 22889472, 'steps': 44705, 'loss/train': 0.6835692524909973}
+02/26/2022 11:28:49 - INFO - codeparrot_training - Step 44706: {'lr': 1.4857502703849002e-05, 'samples': 22889984, 'steps': 44706, 'loss/train': 1.8001538515090942}
+02/26/2022 11:28:54 - INFO - codeparrot_training - Step 44707: {'lr': 1.4851946517039999e-05, 'samples': 22890496, 'steps': 44707, 'loss/train': 1.6719951629638672}
+02/26/2022 11:28:58 - INFO - codeparrot_training - Step 44708: {'lr': 1.4846391337530574e-05, 'samples': 22891008, 'steps': 44708, 'loss/train': 2.054715156555176}
+02/26/2022 11:29:04 - INFO - codeparrot_training - Step 44709: {'lr': 1.4840837165344595e-05, 'samples': 22891520, 'steps': 44709, 'loss/train': 0.8846263289451599}
+02/26/2022 11:29:07 - INFO - codeparrot_training - Step 44710: {'lr': 1.483528400050574e-05, 'samples': 22892032, 'steps': 44710, 'loss/train': 1.3819191455841064}
+02/26/2022 11:29:13 - INFO - codeparrot_training - Step 44711: {'lr': 1.4829731843037902e-05, 'samples': 22892544, 'steps': 44711, 'loss/train': 1.6048927307128906}
+02/26/2022 11:29:16 - INFO - codeparrot_training - Step 44712: {'lr': 1.4824180692964762e-05, 'samples': 22893056, 'steps': 44712, 'loss/train': 1.6939247846603394}
+02/26/2022 11:29:23 - INFO - codeparrot_training - Step 44713: {'lr': 1.4818630550310214e-05, 'samples': 22893568, 'steps': 44713, 'loss/train': 0.9921132326126099}
+02/26/2022 11:29:27 - INFO - codeparrot_training - Step 44714: {'lr': 1.4813081415097934e-05, 'samples': 22894080, 'steps': 44714, 'loss/train': 1.8767247200012207}
+02/26/2022 11:29:33 - INFO - codeparrot_training - Step 44715: {'lr': 1.4807533287351738e-05, 'samples': 22894592, 'steps': 44715, 'loss/train': 1.3981701135635376}
+02/26/2022 11:29:37 - INFO - codeparrot_training - Step 44716: {'lr': 1.4801986167095382e-05, 'samples': 22895104, 'steps': 44716, 'loss/train': 1.323700189590454}
+02/26/2022 11:29:40 - INFO - codeparrot_training - Step 44717: {'lr': 1.4796440054352656e-05, 'samples': 22895616, 'steps': 44717, 'loss/train': 3.6197593212127686}
+02/26/2022 11:29:46 - INFO - codeparrot_training - Step 44718: {'lr': 1.4790894949147288e-05, 'samples': 22896128, 'steps': 44718, 'loss/train': 1.5982636213302612}
+02/26/2022 11:29:49 - INFO - codeparrot_training - Step 44719: {'lr': 1.4785350851503037e-05, 'samples': 22896640, 'steps': 44719, 'loss/train': 0.8854488730430603}
+02/26/2022 11:29:55 - INFO - codeparrot_training - Step 44720: {'lr': 1.4779807761443637e-05, 'samples': 22897152, 'steps': 44720, 'loss/train': 1.3149296045303345}
+02/26/2022 11:29:58 - INFO - codeparrot_training - Step 44721: {'lr': 1.477426567899287e-05, 'samples': 22897664, 'steps': 44721, 'loss/train': 1.3728604316711426}
+02/26/2022 11:30:04 - INFO - codeparrot_training - Step 44722: {'lr': 1.4768724604174444e-05, 'samples': 22898176, 'steps': 44722, 'loss/train': 2.3593132495880127}
+02/26/2022 11:30:07 - INFO - codeparrot_training - Step 44723: {'lr': 1.4763184537012142e-05, 'samples': 22898688, 'steps': 44723, 'loss/train': 1.705065369606018}
+02/26/2022 11:30:14 - INFO - codeparrot_training - Step 44724: {'lr': 1.4757645477529585e-05, 'samples': 22899200, 'steps': 44724, 'loss/train': 1.4917598962783813}
+02/26/2022 11:30:18 - INFO - codeparrot_training - Step 44725: {'lr': 1.4752107425750643e-05, 'samples': 22899712, 'steps': 44725, 'loss/train': 2.5256097316741943}
+02/26/2022 11:30:23 - INFO - codeparrot_training - Step 44726: {'lr': 1.4746570381698936e-05, 'samples': 22900224, 'steps': 44726, 'loss/train': 5.361443996429443}
+02/26/2022 11:30:27 - INFO - codeparrot_training - Step 44727: {'lr': 1.474103434539828e-05, 'samples': 22900736, 'steps': 44727, 'loss/train': 1.583449125289917}
+02/26/2022 11:30:32 - INFO - codeparrot_training - Step 44728: {'lr': 1.4735499316872265e-05, 'samples': 22901248, 'steps': 44728, 'loss/train': 1.9465041160583496}
+02/26/2022 11:30:36 - INFO - codeparrot_training - Step 44729: {'lr': 1.4729965296144704e-05, 'samples': 22901760, 'steps': 44729, 'loss/train': 1.7254831790924072}
+02/26/2022 11:30:41 - INFO - codeparrot_training - Step 44730: {'lr': 1.472443228323922e-05, 'samples': 22902272, 'steps': 44730, 'loss/train': 0.29090437293052673}
+02/26/2022 11:30:45 - INFO - codeparrot_training - Step 44731: {'lr': 1.4718900278179652e-05, 'samples': 22902784, 'steps': 44731, 'loss/train': 1.8774993419647217}
+02/26/2022 11:30:50 - INFO - codeparrot_training - Step 44732: {'lr': 1.4713369280989513e-05, 'samples': 22903296, 'steps': 44732, 'loss/train': 1.6809306144714355}
+02/26/2022 11:30:54 - INFO - codeparrot_training - Step 44733: {'lr': 1.4707839291692616e-05, 'samples': 22903808, 'steps': 44733, 'loss/train': 2.260448694229126}
+02/26/2022 11:31:01 - INFO - codeparrot_training - Step 44734: {'lr': 1.4702310310312633e-05, 'samples': 22904320, 'steps': 44734, 'loss/train': 2.3479745388031006}
+02/26/2022 11:31:04 - INFO - codeparrot_training - Step 44735: {'lr': 1.4696782336873243e-05, 'samples': 22904832, 'steps': 44735, 'loss/train': 2.1346280574798584}
+02/26/2022 11:31:10 - INFO - codeparrot_training - Step 44736: {'lr': 1.4691255371398094e-05, 'samples': 22905344, 'steps': 44736, 'loss/train': 1.2090847492218018}
+02/26/2022 11:31:13 - INFO - codeparrot_training - Step 44737: {'lr': 1.4685729413910886e-05, 'samples': 22905856, 'steps': 44737, 'loss/train': 2.1135356426239014}
+02/26/2022 11:31:19 - INFO - codeparrot_training - Step 44738: {'lr': 1.4680204464435326e-05, 'samples': 22906368, 'steps': 44738, 'loss/train': 0.7936621904373169}
+02/26/2022 11:31:22 - INFO - codeparrot_training - Step 44739: {'lr': 1.4674680522995003e-05, 'samples': 22906880, 'steps': 44739, 'loss/train': 2.1766417026519775}
+02/26/2022 11:31:28 - INFO - codeparrot_training - Step 44740: {'lr': 1.4669157589613652e-05, 'samples': 22907392, 'steps': 44740, 'loss/train': 2.0970025062561035}
+02/26/2022 11:31:31 - INFO - codeparrot_training - Step 44741: {'lr': 1.4663635664314917e-05, 'samples': 22907904, 'steps': 44741, 'loss/train': 1.4811991453170776}
+02/26/2022 11:31:37 - INFO - codeparrot_training - Step 44742: {'lr': 1.4658114747122448e-05, 'samples': 22908416, 'steps': 44742, 'loss/train': 1.4360178709030151}
+02/26/2022 11:31:40 - INFO - codeparrot_training - Step 44743: {'lr': 1.4652594838059835e-05, 'samples': 22908928, 'steps': 44743, 'loss/train': 0.7907061576843262}
+02/26/2022 11:31:46 - INFO - codeparrot_training - Step 44744: {'lr': 1.4647075937150811e-05, 'samples': 22909440, 'steps': 44744, 'loss/train': 0.9782797694206238}
+02/26/2022 11:31:49 - INFO - codeparrot_training - Step 44745: {'lr': 1.4641558044418967e-05, 'samples': 22909952, 'steps': 44745, 'loss/train': 0.5114972591400146}
+02/26/2022 11:31:55 - INFO - codeparrot_training - Step 44746: {'lr': 1.4636041159887981e-05, 'samples': 22910464, 'steps': 44746, 'loss/train': 1.0731189250946045}
+02/26/2022 11:31:59 - INFO - codeparrot_training - Step 44747: {'lr': 1.4630525283581386e-05, 'samples': 22910976, 'steps': 44747, 'loss/train': 1.8338072299957275}
+02/26/2022 11:32:04 - INFO - codeparrot_training - Step 44748: {'lr': 1.4625010415522999e-05, 'samples': 22911488, 'steps': 44748, 'loss/train': 1.7852144241333008}
+02/26/2022 11:32:08 - INFO - codeparrot_training - Step 44749: {'lr': 1.4619496555736216e-05, 'samples': 22912000, 'steps': 44749, 'loss/train': 2.174665689468384}
+02/26/2022 11:32:13 - INFO - codeparrot_training - Step 44750: {'lr': 1.4613983704244827e-05, 'samples': 22912512, 'steps': 44750, 'loss/train': 1.481002926826477}
+02/26/2022 11:32:17 - INFO - codeparrot_training - Step 44751: {'lr': 1.4608471861072337e-05, 'samples': 22913024, 'steps': 44751, 'loss/train': 2.3604652881622314}
+02/26/2022 11:32:24 - INFO - codeparrot_training - Step 44752: {'lr': 1.4602961026242479e-05, 'samples': 22913536, 'steps': 44752, 'loss/train': 2.2931647300720215}
+02/26/2022 11:32:27 - INFO - codeparrot_training - Step 44753: {'lr': 1.4597451199778761e-05, 'samples': 22914048, 'steps': 44753, 'loss/train': 0.5501608848571777}
+02/26/2022 11:32:33 - INFO - codeparrot_training - Step 44754: {'lr': 1.4591942381704803e-05, 'samples': 22914560, 'steps': 44754, 'loss/train': 1.8694555759429932}
+02/26/2022 11:32:36 - INFO - codeparrot_training - Step 44755: {'lr': 1.45864345720442e-05, 'samples': 22915072, 'steps': 44755, 'loss/train': 2.1331048011779785}
+02/26/2022 11:32:42 - INFO - codeparrot_training - Step 44756: {'lr': 1.4580927770820568e-05, 'samples': 22915584, 'steps': 44756, 'loss/train': 1.9914377927780151}
+02/26/2022 11:32:45 - INFO - codeparrot_training - Step 44757: {'lr': 1.457542197805753e-05, 'samples': 22916096, 'steps': 44757, 'loss/train': 2.684089422225952}
+02/26/2022 11:32:51 - INFO - codeparrot_training - Step 44758: {'lr': 1.4569917193778593e-05, 'samples': 22916608, 'steps': 44758, 'loss/train': 1.728378415107727}
+02/26/2022 11:32:54 - INFO - codeparrot_training - Step 44759: {'lr': 1.4564413418007349e-05, 'samples': 22917120, 'steps': 44759, 'loss/train': 2.331244707107544}
+02/26/2022 11:33:02 - INFO - codeparrot_training - Step 44760: {'lr': 1.455891065076742e-05, 'samples': 22917632, 'steps': 44760, 'loss/train': 2.3482542037963867}
+02/26/2022 11:33:05 - INFO - codeparrot_training - Step 44761: {'lr': 1.455340889208237e-05, 'samples': 22918144, 'steps': 44761, 'loss/train': 0.6966246366500854}
+02/26/2022 11:33:11 - INFO - codeparrot_training - Step 44762: {'lr': 1.4547908141975763e-05, 'samples': 22918656, 'steps': 44762, 'loss/train': 1.5532572269439697}
+02/26/2022 11:33:14 - INFO - codeparrot_training - Step 44763: {'lr': 1.4542408400471108e-05, 'samples': 22919168, 'steps': 44763, 'loss/train': 1.6490448713302612}
+02/26/2022 11:33:20 - INFO - codeparrot_training - Step 44764: {'lr': 1.4536909667592052e-05, 'samples': 22919680, 'steps': 44764, 'loss/train': 1.952669382095337}
+02/26/2022 11:33:23 - INFO - codeparrot_training - Step 44765: {'lr': 1.453141194336205e-05, 'samples': 22920192, 'steps': 44765, 'loss/train': 2.195852756500244}
+02/26/2022 11:33:29 - INFO - codeparrot_training - Step 44766: {'lr': 1.4525915227804804e-05, 'samples': 22920704, 'steps': 44766, 'loss/train': 1.8344160318374634}
+02/26/2022 11:33:32 - INFO - codeparrot_training - Step 44767: {'lr': 1.4520419520943683e-05, 'samples': 22921216, 'steps': 44767, 'loss/train': 1.6205499172210693}
+02/26/2022 11:33:38 - INFO - codeparrot_training - Step 44768: {'lr': 1.4514924822802367e-05, 'samples': 22921728, 'steps': 44768, 'loss/train': 0.5510485768318176}
+02/26/2022 11:33:41 - INFO - codeparrot_training - Step 44769: {'lr': 1.4509431133404278e-05, 'samples': 22922240, 'steps': 44769, 'loss/train': 1.9467521905899048}
+02/26/2022 11:33:49 - INFO - codeparrot_training - Step 44770: {'lr': 1.450393845277309e-05, 'samples': 22922752, 'steps': 44770, 'loss/train': 1.3639692068099976}
+02/26/2022 11:33:52 - INFO - codeparrot_training - Step 44771: {'lr': 1.4498446780932179e-05, 'samples': 22923264, 'steps': 44771, 'loss/train': 1.977573275566101}
+02/26/2022 11:33:58 - INFO - codeparrot_training - Step 44772: {'lr': 1.449295611790516e-05, 'samples': 22923776, 'steps': 44772, 'loss/train': 1.6996464729309082}
+02/26/2022 11:34:01 - INFO - codeparrot_training - Step 44773: {'lr': 1.4487466463715516e-05, 'samples': 22924288, 'steps': 44773, 'loss/train': 1.724143624305725}
+02/26/2022 11:34:07 - INFO - codeparrot_training - Step 44774: {'lr': 1.4481977818386866e-05, 'samples': 22924800, 'steps': 44774, 'loss/train': 1.9280657768249512}
+02/26/2022 11:34:10 - INFO - codeparrot_training - Step 44775: {'lr': 1.4476490181942554e-05, 'samples': 22925312, 'steps': 44775, 'loss/train': 2.70141339302063}
+02/26/2022 11:34:16 - INFO - codeparrot_training - Step 44776: {'lr': 1.4471003554406197e-05, 'samples': 22925824, 'steps': 44776, 'loss/train': 1.3111261129379272}
+02/26/2022 11:34:19 - INFO - codeparrot_training - Step 44777: {'lr': 1.4465517935801254e-05, 'samples': 22926336, 'steps': 44777, 'loss/train': 2.184539794921875}
+02/26/2022 11:34:25 - INFO - codeparrot_training - Step 44778: {'lr': 1.4460033326151257e-05, 'samples': 22926848, 'steps': 44778, 'loss/train': 2.2106947898864746}
+02/26/2022 11:34:28 - INFO - codeparrot_training - Step 44779: {'lr': 1.4454549725479716e-05, 'samples': 22927360, 'steps': 44779, 'loss/train': 0.666465163230896}
+02/26/2022 11:34:36 - INFO - codeparrot_training - Step 44780: {'lr': 1.4449067133810057e-05, 'samples': 22927872, 'steps': 44780, 'loss/train': 1.6523337364196777}
+02/26/2022 11:34:39 - INFO - codeparrot_training - Step 44781: {'lr': 1.4443585551165817e-05, 'samples': 22928384, 'steps': 44781, 'loss/train': 1.522513508796692}
+02/26/2022 11:34:45 - INFO - codeparrot_training - Step 44782: {'lr': 1.443810497757042e-05, 'samples': 22928896, 'steps': 44782, 'loss/train': 0.9574680328369141}
+02/26/2022 11:34:49 - INFO - codeparrot_training - Step 44783: {'lr': 1.4432625413047407e-05, 'samples': 22929408, 'steps': 44783, 'loss/train': 3.3301377296447754}
+02/26/2022 11:34:52 - INFO - codeparrot_training - Step 44784: {'lr': 1.4427146857620228e-05, 'samples': 22929920, 'steps': 44784, 'loss/train': 1.077704668045044}
+02/26/2022 11:34:58 - INFO - codeparrot_training - Step 44785: {'lr': 1.4421669311312363e-05, 'samples': 22930432, 'steps': 44785, 'loss/train': 1.9786309003829956}
+02/26/2022 11:35:02 - INFO - codeparrot_training - Step 44786: {'lr': 1.4416192774147213e-05, 'samples': 22930944, 'steps': 44786, 'loss/train': 2.3678464889526367}
+02/26/2022 11:35:05 - INFO - codeparrot_training - Step 44787: {'lr': 1.4410717246148341e-05, 'samples': 22931456, 'steps': 44787, 'loss/train': 1.082016110420227}
+02/26/2022 11:35:11 - INFO - codeparrot_training - Step 44788: {'lr': 1.4405242727339145e-05, 'samples': 22931968, 'steps': 44788, 'loss/train': 2.3670239448547363}
+02/26/2022 11:35:14 - INFO - codeparrot_training - Step 44789: {'lr': 1.4399769217743052e-05, 'samples': 22932480, 'steps': 44789, 'loss/train': 2.005310297012329}
+02/26/2022 11:35:20 - INFO - codeparrot_training - Step 44790: {'lr': 1.4394296717383516e-05, 'samples': 22932992, 'steps': 44790, 'loss/train': 1.7135089635849}
+02/26/2022 11:35:23 - INFO - codeparrot_training - Step 44791: {'lr': 1.438882522628404e-05, 'samples': 22933504, 'steps': 44791, 'loss/train': 1.2183119058609009}
+02/26/2022 11:35:29 - INFO - codeparrot_training - Step 44792: {'lr': 1.4383354744468031e-05, 'samples': 22934016, 'steps': 44792, 'loss/train': 1.9528683423995972}
+02/26/2022 11:35:32 - INFO - codeparrot_training - Step 44793: {'lr': 1.4377885271958907e-05, 'samples': 22934528, 'steps': 44793, 'loss/train': 3.095869779586792}
+02/26/2022 11:35:38 - INFO - codeparrot_training - Step 44794: {'lr': 1.4372416808780042e-05, 'samples': 22935040, 'steps': 44794, 'loss/train': 1.505162000656128}
+02/26/2022 11:35:41 - INFO - codeparrot_training - Step 44795: {'lr': 1.4366949354955e-05, 'samples': 22935552, 'steps': 44795, 'loss/train': 2.209728479385376}
+02/26/2022 11:35:47 - INFO - codeparrot_training - Step 44796: {'lr': 1.4361482910507095e-05, 'samples': 22936064, 'steps': 44796, 'loss/train': 1.2412135601043701}
+02/26/2022 11:35:51 - INFO - codeparrot_training - Step 44797: {'lr': 1.4356017475459782e-05, 'samples': 22936576, 'steps': 44797, 'loss/train': 0.9276085495948792}
+02/26/2022 11:35:58 - INFO - codeparrot_training - Step 44798: {'lr': 1.435055304983643e-05, 'samples': 22937088, 'steps': 44798, 'loss/train': 0.4544755220413208}
+02/26/2022 11:36:01 - INFO - codeparrot_training - Step 44799: {'lr': 1.4345089633660519e-05, 'samples': 22937600, 'steps': 44799, 'loss/train': 1.5171109437942505}
+02/26/2022 11:36:07 - INFO - codeparrot_training - Step 44800: {'lr': 1.4339627226955393e-05, 'samples': 22938112, 'steps': 44800, 'loss/train': 1.1359609365463257}
+02/26/2022 11:36:10 - INFO - codeparrot_training - Step 44801: {'lr': 1.4334165829744534e-05, 'samples': 22938624, 'steps': 44801, 'loss/train': 0.08711253851652145}
+02/26/2022 11:36:16 - INFO - codeparrot_training - Step 44802: {'lr': 1.4328705442051199e-05, 'samples': 22939136, 'steps': 44802, 'loss/train': 2.0219101905822754}
+02/26/2022 11:36:19 - INFO - codeparrot_training - Step 44803: {'lr': 1.4323246063898898e-05, 'samples': 22939648, 'steps': 44803, 'loss/train': 1.6056190729141235}
+02/26/2022 11:36:25 - INFO - codeparrot_training - Step 44804: {'lr': 1.4317787695310918e-05, 'samples': 22940160, 'steps': 44804, 'loss/train': 1.6616590023040771}
+02/26/2022 11:36:28 - INFO - codeparrot_training - Step 44805: {'lr': 1.4312330336310797e-05, 'samples': 22940672, 'steps': 44805, 'loss/train': 2.397465229034424}
+02/26/2022 11:36:34 - INFO - codeparrot_training - Step 44806: {'lr': 1.4306873986921737e-05, 'samples': 22941184, 'steps': 44806, 'loss/train': 2.635054111480713}
+02/26/2022 11:36:37 - INFO - codeparrot_training - Step 44807: {'lr': 1.430141864716722e-05, 'samples': 22941696, 'steps': 44807, 'loss/train': 1.749933123588562}
+02/26/2022 11:36:45 - INFO - codeparrot_training - Step 44808: {'lr': 1.429596431707053e-05, 'samples': 22942208, 'steps': 44808, 'loss/train': 2.2393569946289062}
+02/26/2022 11:36:48 - INFO - codeparrot_training - Step 44809: {'lr': 1.4290510996655182e-05, 'samples': 22942720, 'steps': 44809, 'loss/train': 1.040702223777771}
+02/26/2022 11:36:54 - INFO - codeparrot_training - Step 44810: {'lr': 1.4285058685944347e-05, 'samples': 22943232, 'steps': 44810, 'loss/train': 1.6356934309005737}
+02/26/2022 11:36:57 - INFO - codeparrot_training - Step 44811: {'lr': 1.4279607384961535e-05, 'samples': 22943744, 'steps': 44811, 'loss/train': 2.3489394187927246}
+02/26/2022 11:37:03 - INFO - codeparrot_training - Step 44812: {'lr': 1.4274157093729978e-05, 'samples': 22944256, 'steps': 44812, 'loss/train': 1.4525930881500244}
+02/26/2022 11:37:06 - INFO - codeparrot_training - Step 44813: {'lr': 1.4268707812273102e-05, 'samples': 22944768, 'steps': 44813, 'loss/train': 1.2331892251968384}
+02/26/2022 11:37:12 - INFO - codeparrot_training - Step 44814: {'lr': 1.426325954061425e-05, 'samples': 22945280, 'steps': 44814, 'loss/train': 1.5273914337158203}
+02/26/2022 11:37:15 - INFO - codeparrot_training - Step 44815: {'lr': 1.4257812278776761e-05, 'samples': 22945792, 'steps': 44815, 'loss/train': 1.8169147968292236}
+02/26/2022 11:37:21 - INFO - codeparrot_training - Step 44816: {'lr': 1.425236602678387e-05, 'samples': 22946304, 'steps': 44816, 'loss/train': 1.6976604461669922}
+02/26/2022 11:37:24 - INFO - codeparrot_training - Step 44817: {'lr': 1.4246920784659029e-05, 'samples': 22946816, 'steps': 44817, 'loss/train': 2.6473615169525146}
+02/26/2022 11:37:32 - INFO - codeparrot_training - Step 44818: {'lr': 1.4241476552425552e-05, 'samples': 22947328, 'steps': 44818, 'loss/train': 2.080000877380371}
+02/26/2022 11:37:35 - INFO - codeparrot_training - Step 44819: {'lr': 1.4236033330106701e-05, 'samples': 22947840, 'steps': 44819, 'loss/train': 2.176170825958252}
+02/26/2022 11:37:41 - INFO - codeparrot_training - Step 44820: {'lr': 1.4230591117725788e-05, 'samples': 22948352, 'steps': 44820, 'loss/train': 1.4614702463150024}
+02/26/2022 11:37:44 - INFO - codeparrot_training - Step 44821: {'lr': 1.4225149915306184e-05, 'samples': 22948864, 'steps': 44821, 'loss/train': 1.2352185249328613}
+02/26/2022 11:37:50 - INFO - codeparrot_training - Step 44822: {'lr': 1.4219709722871177e-05, 'samples': 22949376, 'steps': 44822, 'loss/train': 1.61871337890625}
+02/26/2022 11:37:53 - INFO - codeparrot_training - Step 44823: {'lr': 1.421427054044408e-05, 'samples': 22949888, 'steps': 44823, 'loss/train': 0.9498617053031921}
+02/26/2022 11:37:59 - INFO - codeparrot_training - Step 44824: {'lr': 1.4208832368048124e-05, 'samples': 22950400, 'steps': 44824, 'loss/train': 1.7852410078048706}
+02/26/2022 11:38:02 - INFO - codeparrot_training - Step 44825: {'lr': 1.4203395205706681e-05, 'samples': 22950912, 'steps': 44825, 'loss/train': 2.3695123195648193}
+02/26/2022 11:38:07 - INFO - codeparrot_training - Step 44826: {'lr': 1.4197959053443039e-05, 'samples': 22951424, 'steps': 44826, 'loss/train': 1.4653220176696777}
+02/26/2022 11:38:11 - INFO - codeparrot_training - Step 44827: {'lr': 1.419252391128048e-05, 'samples': 22951936, 'steps': 44827, 'loss/train': 1.5267162322998047}
+02/26/2022 11:38:18 - INFO - codeparrot_training - Step 44828: {'lr': 1.4187089779242212e-05, 'samples': 22952448, 'steps': 44828, 'loss/train': 1.3887460231781006}
+02/26/2022 11:38:22 - INFO - codeparrot_training - Step 44829: {'lr': 1.4181656657351604e-05, 'samples': 22952960, 'steps': 44829, 'loss/train': 1.1708043813705444}
+02/26/2022 11:38:27 - INFO - codeparrot_training - Step 44830: {'lr': 1.4176224545631917e-05, 'samples': 22953472, 'steps': 44830, 'loss/train': 0.2806812822818756}
+02/26/2022 11:38:31 - INFO - codeparrot_training - Step 44831: {'lr': 1.4170793444106405e-05, 'samples': 22953984, 'steps': 44831, 'loss/train': 3.100749969482422}
+02/26/2022 11:38:36 - INFO - codeparrot_training - Step 44832: {'lr': 1.4165363352798305e-05, 'samples': 22954496, 'steps': 44832, 'loss/train': 1.7939221858978271}
+02/26/2022 11:38:40 - INFO - codeparrot_training - Step 44833: {'lr': 1.4159934271730873e-05, 'samples': 22955008, 'steps': 44833, 'loss/train': 1.3883085250854492}
+02/26/2022 11:38:45 - INFO - codeparrot_training - Step 44834: {'lr': 1.4154506200927453e-05, 'samples': 22955520, 'steps': 44834, 'loss/train': 1.2299606800079346}
+02/26/2022 11:38:49 - INFO - codeparrot_training - Step 44835: {'lr': 1.4149079140411163e-05, 'samples': 22956032, 'steps': 44835, 'loss/train': 2.4087371826171875}
+02/26/2022 11:38:54 - INFO - codeparrot_training - Step 44836: {'lr': 1.4143653090205432e-05, 'samples': 22956544, 'steps': 44836, 'loss/train': 3.22493577003479}
+02/26/2022 11:38:58 - INFO - codeparrot_training - Step 44837: {'lr': 1.4138228050333323e-05, 'samples': 22957056, 'steps': 44837, 'loss/train': 1.6801273822784424}
+02/26/2022 11:39:03 - INFO - codeparrot_training - Step 44838: {'lr': 1.413280402081818e-05, 'samples': 22957568, 'steps': 44838, 'loss/train': 0.3348790109157562}
+02/26/2022 11:39:07 - INFO - codeparrot_training - Step 44839: {'lr': 1.4127381001683176e-05, 'samples': 22958080, 'steps': 44839, 'loss/train': 1.442067265510559}
+02/26/2022 11:39:12 - INFO - codeparrot_training - Step 44840: {'lr': 1.4121958992951628e-05, 'samples': 22958592, 'steps': 44840, 'loss/train': 2.6171531677246094}
+02/26/2022 11:39:16 - INFO - codeparrot_training - Step 44841: {'lr': 1.4116537994646655e-05, 'samples': 22959104, 'steps': 44841, 'loss/train': 1.5465601682662964}
+02/26/2022 11:39:21 - INFO - codeparrot_training - Step 44842: {'lr': 1.4111118006791573e-05, 'samples': 22959616, 'steps': 44842, 'loss/train': 1.819445013999939}
+02/26/2022 11:39:25 - INFO - codeparrot_training - Step 44843: {'lr': 1.4105699029409502e-05, 'samples': 22960128, 'steps': 44843, 'loss/train': 1.1812183856964111}
+02/26/2022 11:39:32 - INFO - codeparrot_training - Step 44844: {'lr': 1.4100281062523812e-05, 'samples': 22960640, 'steps': 44844, 'loss/train': 0.9964702129364014}
+02/26/2022 11:39:35 - INFO - codeparrot_training - Step 44845: {'lr': 1.4094864106157512e-05, 'samples': 22961152, 'steps': 44845, 'loss/train': 2.3635566234588623}
+02/26/2022 11:39:41 - INFO - codeparrot_training - Step 44846: {'lr': 1.4089448160333946e-05, 'samples': 22961664, 'steps': 44846, 'loss/train': 0.991076648235321}
+02/26/2022 11:39:45 - INFO - codeparrot_training - Step 44847: {'lr': 1.4084033225076259e-05, 'samples': 22962176, 'steps': 44847, 'loss/train': 1.69138765335083}
+02/26/2022 11:39:50 - INFO - codeparrot_training - Step 44848: {'lr': 1.4078619300407685e-05, 'samples': 22962688, 'steps': 44848, 'loss/train': 1.7338188886642456}
+02/26/2022 11:39:53 - INFO - codeparrot_training - Step 44849: {'lr': 1.4073206386351372e-05, 'samples': 22963200, 'steps': 44849, 'loss/train': 1.3852802515029907}
+02/26/2022 11:39:59 - INFO - codeparrot_training - Step 44850: {'lr': 1.406779448293055e-05, 'samples': 22963712, 'steps': 44850, 'loss/train': 2.29586124420166}
+02/26/2022 11:40:02 - INFO - codeparrot_training - Step 44851: {'lr': 1.4062383590168342e-05, 'samples': 22964224, 'steps': 44851, 'loss/train': 2.5437958240509033}
+02/26/2022 11:40:08 - INFO - codeparrot_training - Step 44852: {'lr': 1.4056973708088006e-05, 'samples': 22964736, 'steps': 44852, 'loss/train': 2.7314047813415527}
+02/26/2022 11:40:12 - INFO - codeparrot_training - Step 44853: {'lr': 1.405156483671266e-05, 'samples': 22965248, 'steps': 44853, 'loss/train': 2.251418352127075}
+02/26/2022 11:40:19 - INFO - codeparrot_training - Step 44854: {'lr': 1.4046156976065483e-05, 'samples': 22965760, 'steps': 44854, 'loss/train': 0.707883358001709}
+02/26/2022 11:40:23 - INFO - codeparrot_training - Step 44855: {'lr': 1.4040750126169621e-05, 'samples': 22966272, 'steps': 44855, 'loss/train': 1.8748682737350464}
+02/26/2022 11:40:29 - INFO - codeparrot_training - Step 44856: {'lr': 1.4035344287048307e-05, 'samples': 22966784, 'steps': 44856, 'loss/train': 1.1896917819976807}
+02/26/2022 11:40:32 - INFO - codeparrot_training - Step 44857: {'lr': 1.4029939458724632e-05, 'samples': 22967296, 'steps': 44857, 'loss/train': 1.3778964281082153}
+02/26/2022 11:40:38 - INFO - codeparrot_training - Step 44858: {'lr': 1.4024535641221775e-05, 'samples': 22967808, 'steps': 44858, 'loss/train': 1.0070184469223022}
+02/26/2022 11:40:41 - INFO - codeparrot_training - Step 44859: {'lr': 1.4019132834562825e-05, 'samples': 22968320, 'steps': 44859, 'loss/train': 2.2810237407684326}
+02/26/2022 11:40:47 - INFO - codeparrot_training - Step 44860: {'lr': 1.4013731038771043e-05, 'samples': 22968832, 'steps': 44860, 'loss/train': 2.036309242248535}
+02/26/2022 11:40:50 - INFO - codeparrot_training - Step 44861: {'lr': 1.4008330253869438e-05, 'samples': 22969344, 'steps': 44861, 'loss/train': 0.9027878642082214}
+02/26/2022 11:40:56 - INFO - codeparrot_training - Step 44862: {'lr': 1.4002930479881325e-05, 'samples': 22969856, 'steps': 44862, 'loss/train': 1.34867525100708}
+02/26/2022 11:40:59 - INFO - codeparrot_training - Step 44863: {'lr': 1.3997531716829603e-05, 'samples': 22970368, 'steps': 44863, 'loss/train': 2.2102622985839844}
+02/26/2022 11:41:07 - INFO - codeparrot_training - Step 44864: {'lr': 1.3992133964737585e-05, 'samples': 22970880, 'steps': 44864, 'loss/train': 1.5386334657669067}
+02/26/2022 11:41:10 - INFO - codeparrot_training - Step 44865: {'lr': 1.398673722362831e-05, 'samples': 22971392, 'steps': 44865, 'loss/train': 1.4795500040054321}
+02/26/2022 11:41:16 - INFO - codeparrot_training - Step 44866: {'lr': 1.3981341493524897e-05, 'samples': 22971904, 'steps': 44866, 'loss/train': 2.3604021072387695}
+02/26/2022 11:41:19 - INFO - codeparrot_training - Step 44867: {'lr': 1.3975946774450437e-05, 'samples': 22972416, 'steps': 44867, 'loss/train': 1.7739827632904053}
+02/26/2022 11:41:25 - INFO - codeparrot_training - Step 44868: {'lr': 1.3970553066428138e-05, 'samples': 22972928, 'steps': 44868, 'loss/train': 0.7035548686981201}
+02/26/2022 11:41:28 - INFO - codeparrot_training - Step 44869: {'lr': 1.3965160369481034e-05, 'samples': 22973440, 'steps': 44869, 'loss/train': 1.1788643598556519}
+02/26/2022 11:41:34 - INFO - codeparrot_training - Step 44870: {'lr': 1.3959768683632218e-05, 'samples': 22973952, 'steps': 44870, 'loss/train': 0.8870359659194946}
+02/26/2022 11:41:37 - INFO - codeparrot_training - Step 44871: {'lr': 1.3954378008904755e-05, 'samples': 22974464, 'steps': 44871, 'loss/train': 0.2659461498260498}
+02/26/2022 11:41:43 - INFO - codeparrot_training - Step 44872: {'lr': 1.394898834532185e-05, 'samples': 22974976, 'steps': 44872, 'loss/train': 0.4549286663532257}
+02/26/2022 11:41:46 - INFO - codeparrot_training - Step 44873: {'lr': 1.394359969290651e-05, 'samples': 22975488, 'steps': 44873, 'loss/train': 1.9302061796188354}
+02/26/2022 11:41:52 - INFO - codeparrot_training - Step 44874: {'lr': 1.3938212051681803e-05, 'samples': 22976000, 'steps': 44874, 'loss/train': 0.11666565388441086}
+02/26/2022 11:41:55 - INFO - codeparrot_training - Step 44875: {'lr': 1.3932825421670875e-05, 'samples': 22976512, 'steps': 44875, 'loss/train': 1.7317155599594116}
+02/26/2022 11:42:03 - INFO - codeparrot_training - Step 44876: {'lr': 1.3927439802896762e-05, 'samples': 22977024, 'steps': 44876, 'loss/train': 1.5664552450180054}
+02/26/2022 11:42:06 - INFO - codeparrot_training - Step 44877: {'lr': 1.3922055195382532e-05, 'samples': 22977536, 'steps': 44877, 'loss/train': 1.6684882640838623}
+02/26/2022 11:42:12 - INFO - codeparrot_training - Step 44878: {'lr': 1.3916671599151192e-05, 'samples': 22978048, 'steps': 44878, 'loss/train': 2.0447936058044434}
+02/26/2022 11:42:15 - INFO - codeparrot_training - Step 44879: {'lr': 1.3911289014225976e-05, 'samples': 22978560, 'steps': 44879, 'loss/train': 1.8103251457214355}
+02/26/2022 11:42:21 - INFO - codeparrot_training - Step 44880: {'lr': 1.390590744062975e-05, 'samples': 22979072, 'steps': 44880, 'loss/train': 2.3412225246429443}
+02/26/2022 11:42:24 - INFO - codeparrot_training - Step 44881: {'lr': 1.3900526878385666e-05, 'samples': 22979584, 'steps': 44881, 'loss/train': 2.4388139247894287}
+02/26/2022 11:42:30 - INFO - codeparrot_training - Step 44882: {'lr': 1.3895147327516733e-05, 'samples': 22980096, 'steps': 44882, 'loss/train': 1.9126757383346558}
+02/26/2022 11:42:33 - INFO - codeparrot_training - Step 44883: {'lr': 1.3889768788046097e-05, 'samples': 22980608, 'steps': 44883, 'loss/train': 2.176185131072998}
+02/26/2022 11:42:39 - INFO - codeparrot_training - Step 44884: {'lr': 1.3884391259996632e-05, 'samples': 22981120, 'steps': 44884, 'loss/train': 1.6940706968307495}
+02/26/2022 11:42:42 - INFO - codeparrot_training - Step 44885: {'lr': 1.387901474339151e-05, 'samples': 22981632, 'steps': 44885, 'loss/train': 1.2070517539978027}
+02/26/2022 11:42:48 - INFO - codeparrot_training - Step 44886: {'lr': 1.387363923825366e-05, 'samples': 22982144, 'steps': 44886, 'loss/train': 0.8737601041793823}
+02/26/2022 11:42:51 - INFO - codeparrot_training - Step 44887: {'lr': 1.3868264744606229e-05, 'samples': 22982656, 'steps': 44887, 'loss/train': 1.9998855590820312}
+02/26/2022 11:42:57 - INFO - codeparrot_training - Step 44888: {'lr': 1.3862891262472144e-05, 'samples': 22983168, 'steps': 44888, 'loss/train': 2.0377228260040283}
+02/26/2022 11:43:00 - INFO - codeparrot_training - Step 44889: {'lr': 1.3857518791874441e-05, 'samples': 22983680, 'steps': 44889, 'loss/train': 0.19210028648376465}
+02/26/2022 11:43:07 - INFO - codeparrot_training - Step 44890: {'lr': 1.385214733283613e-05, 'samples': 22984192, 'steps': 44890, 'loss/train': 0.9174827933311462}
+02/26/2022 11:43:11 - INFO - codeparrot_training - Step 44891: {'lr': 1.3846776885380247e-05, 'samples': 22984704, 'steps': 44891, 'loss/train': 1.795182466506958}
+02/26/2022 11:43:16 - INFO - codeparrot_training - Step 44892: {'lr': 1.3841407449529803e-05, 'samples': 22985216, 'steps': 44892, 'loss/train': 0.7177965044975281}
+02/26/2022 11:43:20 - INFO - codeparrot_training - Step 44893: {'lr': 1.3836039025307778e-05, 'samples': 22985728, 'steps': 44893, 'loss/train': 1.2636419534683228}
+02/26/2022 11:43:25 - INFO - codeparrot_training - Step 44894: {'lr': 1.3830671612737156e-05, 'samples': 22986240, 'steps': 44894, 'loss/train': 2.26444411277771}
+02/26/2022 11:43:29 - INFO - codeparrot_training - Step 44895: {'lr': 1.3825305211840944e-05, 'samples': 22986752, 'steps': 44895, 'loss/train': 1.291749358177185}
+02/26/2022 11:43:34 - INFO - codeparrot_training - Step 44896: {'lr': 1.3819939822642125e-05, 'samples': 22987264, 'steps': 44896, 'loss/train': 1.6828207969665527}
+02/26/2022 11:43:38 - INFO - codeparrot_training - Step 44897: {'lr': 1.3814575445163735e-05, 'samples': 22987776, 'steps': 44897, 'loss/train': 2.33162522315979}
+02/26/2022 11:43:43 - INFO - codeparrot_training - Step 44898: {'lr': 1.3809212079428645e-05, 'samples': 22988288, 'steps': 44898, 'loss/train': 2.1885838508605957}
+02/26/2022 11:43:47 - INFO - codeparrot_training - Step 44899: {'lr': 1.3803849725459922e-05, 'samples': 22988800, 'steps': 44899, 'loss/train': 2.8827908039093018}
+02/26/2022 11:43:54 - INFO - codeparrot_training - Step 44900: {'lr': 1.3798488383280488e-05, 'samples': 22989312, 'steps': 44900, 'loss/train': 1.083183765411377}
+02/26/2022 11:43:57 - INFO - codeparrot_training - Step 44901: {'lr': 1.3793128052913384e-05, 'samples': 22989824, 'steps': 44901, 'loss/train': 1.840578556060791}
+02/26/2022 11:44:03 - INFO - codeparrot_training - Step 44902: {'lr': 1.3787768734381451e-05, 'samples': 22990336, 'steps': 44902, 'loss/train': 1.6816685199737549}
+02/26/2022 11:44:09 - INFO - codeparrot_training - Step 44903: {'lr': 1.3782410427707726e-05, 'samples': 22990848, 'steps': 44903, 'loss/train': 1.649601697921753}
+02/26/2022 11:44:12 - INFO - codeparrot_training - Step 44904: {'lr': 1.3777053132915135e-05, 'samples': 22991360, 'steps': 44904, 'loss/train': 0.4503486454486847}
+02/26/2022 11:44:15 - INFO - codeparrot_training - Step 44905: {'lr': 1.3771696850026688e-05, 'samples': 22991872, 'steps': 44905, 'loss/train': 1.1975904703140259}
+02/26/2022 11:44:21 - INFO - codeparrot_training - Step 44906: {'lr': 1.3766341579065227e-05, 'samples': 22992384, 'steps': 44906, 'loss/train': 1.4130488634109497}
+02/26/2022 11:44:26 - INFO - codeparrot_training - Step 44907: {'lr': 1.3760987320053791e-05, 'samples': 22992896, 'steps': 44907, 'loss/train': 1.5262269973754883}
+02/26/2022 11:44:30 - INFO - codeparrot_training - Step 44908: {'lr': 1.3755634073015221e-05, 'samples': 22993408, 'steps': 44908, 'loss/train': 1.9153634309768677}
+02/26/2022 11:44:36 - INFO - codeparrot_training - Step 44909: {'lr': 1.3750281837972528e-05, 'samples': 22993920, 'steps': 44909, 'loss/train': 2.522473096847534}
+02/26/2022 11:44:39 - INFO - codeparrot_training - Step 44910: {'lr': 1.3744930614948608e-05, 'samples': 22994432, 'steps': 44910, 'loss/train': 0.5706866979598999}
+02/26/2022 11:44:46 - INFO - codeparrot_training - Step 44911: {'lr': 1.3739580403966418e-05, 'samples': 22994944, 'steps': 44911, 'loss/train': 1.0238890647888184}
+02/26/2022 11:44:50 - INFO - codeparrot_training - Step 44912: {'lr': 1.3734231205048826e-05, 'samples': 22995456, 'steps': 44912, 'loss/train': 1.5381544828414917}
+02/26/2022 11:44:55 - INFO - codeparrot_training - Step 44913: {'lr': 1.3728883018218701e-05, 'samples': 22995968, 'steps': 44913, 'loss/train': 0.21254390478134155}
+02/26/2022 11:44:59 - INFO - codeparrot_training - Step 44914: {'lr': 1.3723535843499086e-05, 'samples': 22996480, 'steps': 44914, 'loss/train': 2.314822196960449}
+02/26/2022 11:45:05 - INFO - codeparrot_training - Step 44915: {'lr': 1.3718189680912819e-05, 'samples': 22996992, 'steps': 44915, 'loss/train': 1.8591642379760742}
+02/26/2022 11:45:08 - INFO - codeparrot_training - Step 44916: {'lr': 1.3712844530482771e-05, 'samples': 22997504, 'steps': 44916, 'loss/train': 1.4261101484298706}
+02/26/2022 11:45:11 - INFO - codeparrot_training - Step 44917: {'lr': 1.370750039223187e-05, 'samples': 22998016, 'steps': 44917, 'loss/train': 2.20125412940979}
+02/26/2022 11:45:17 - INFO - codeparrot_training - Step 44918: {'lr': 1.3702157266183012e-05, 'samples': 22998528, 'steps': 44918, 'loss/train': 0.04204076528549194}
+02/26/2022 11:45:23 - INFO - codeparrot_training - Step 44919: {'lr': 1.369681515235907e-05, 'samples': 22999040, 'steps': 44919, 'loss/train': 1.4420565366744995}
+02/26/2022 11:45:26 - INFO - codeparrot_training - Step 44920: {'lr': 1.369147405078297e-05, 'samples': 22999552, 'steps': 44920, 'loss/train': 1.529695749282837}
+02/26/2022 11:45:33 - INFO - codeparrot_training - Step 44921: {'lr': 1.3686133961477498e-05, 'samples': 23000064, 'steps': 44921, 'loss/train': 2.391444683074951}
+02/26/2022 11:45:37 - INFO - codeparrot_training - Step 44922: {'lr': 1.3680794884465635e-05, 'samples': 23000576, 'steps': 44922, 'loss/train': 2.1998631954193115}
+02/26/2022 11:45:40 - INFO - codeparrot_training - Step 44923: {'lr': 1.3675456819770227e-05, 'samples': 23001088, 'steps': 44923, 'loss/train': 1.3627636432647705}
+02/26/2022 11:45:46 - INFO - codeparrot_training - Step 44924: {'lr': 1.3670119767414085e-05, 'samples': 23001600, 'steps': 44924, 'loss/train': 0.7539050579071045}
+02/26/2022 11:45:52 - INFO - codeparrot_training - Step 44925: {'lr': 1.3664783727420083e-05, 'samples': 23002112, 'steps': 44925, 'loss/train': 1.9430809020996094}
+02/26/2022 11:45:55 - INFO - codeparrot_training - Step 44926: {'lr': 1.3659448699811145e-05, 'samples': 23002624, 'steps': 44926, 'loss/train': 1.7160112857818604}
+02/26/2022 11:46:01 - INFO - codeparrot_training - Step 44927: {'lr': 1.3654114684610085e-05, 'samples': 23003136, 'steps': 44927, 'loss/train': 1.6460176706314087}
+02/26/2022 11:46:04 - INFO - codeparrot_training - Step 44928: {'lr': 1.364878168183975e-05, 'samples': 23003648, 'steps': 44928, 'loss/train': 1.4424197673797607}
+02/26/2022 11:46:10 - INFO - codeparrot_training - Step 44929: {'lr': 1.3643449691522953e-05, 'samples': 23004160, 'steps': 44929, 'loss/train': 1.2292176485061646}
+02/26/2022 11:46:13 - INFO - codeparrot_training - Step 44930: {'lr': 1.363811871368259e-05, 'samples': 23004672, 'steps': 44930, 'loss/train': 1.455814003944397}
+02/26/2022 11:46:19 - INFO - codeparrot_training - Step 44931: {'lr': 1.3632788748341452e-05, 'samples': 23005184, 'steps': 44931, 'loss/train': 2.065498113632202}
+02/26/2022 11:46:22 - INFO - codeparrot_training - Step 44932: {'lr': 1.3627459795522435e-05, 'samples': 23005696, 'steps': 44932, 'loss/train': 1.2602853775024414}
+02/26/2022 11:46:28 - INFO - codeparrot_training - Step 44933: {'lr': 1.3622131855248243e-05, 'samples': 23006208, 'steps': 44933, 'loss/train': 1.2619924545288086}
+02/26/2022 11:46:31 - INFO - codeparrot_training - Step 44934: {'lr': 1.361680492754186e-05, 'samples': 23006720, 'steps': 44934, 'loss/train': 1.9449198246002197}
+02/26/2022 11:46:37 - INFO - codeparrot_training - Step 44935: {'lr': 1.3611479012425959e-05, 'samples': 23007232, 'steps': 44935, 'loss/train': 1.1407411098480225}
+02/26/2022 11:46:40 - INFO - codeparrot_training - Step 44936: {'lr': 1.3606154109923497e-05, 'samples': 23007744, 'steps': 44936, 'loss/train': 1.7550444602966309}
+02/26/2022 11:46:47 - INFO - codeparrot_training - Step 44937: {'lr': 1.3600830220057148e-05, 'samples': 23008256, 'steps': 44937, 'loss/train': 1.7893890142440796}
+02/26/2022 11:46:51 - INFO - codeparrot_training - Step 44938: {'lr': 1.359550734284981e-05, 'samples': 23008768, 'steps': 44938, 'loss/train': 2.1626994609832764}
+02/26/2022 11:46:56 - INFO - codeparrot_training - Step 44939: {'lr': 1.359018547832419e-05, 'samples': 23009280, 'steps': 44939, 'loss/train': 2.213036060333252}
+02/26/2022 11:47:00 - INFO - codeparrot_training - Step 44940: {'lr': 1.3584864626503241e-05, 'samples': 23009792, 'steps': 44940, 'loss/train': 2.429441213607788}
+02/26/2022 11:47:05 - INFO - codeparrot_training - Step 44941: {'lr': 1.3579544787409608e-05, 'samples': 23010304, 'steps': 44941, 'loss/train': 0.3529791235923767}
+02/26/2022 11:47:09 - INFO - codeparrot_training - Step 44942: {'lr': 1.357422596106614e-05, 'samples': 23010816, 'steps': 44942, 'loss/train': 1.4843324422836304}
+02/26/2022 11:47:14 - INFO - codeparrot_training - Step 44943: {'lr': 1.356890814749559e-05, 'samples': 23011328, 'steps': 44943, 'loss/train': 2.6002357006073}
+02/26/2022 11:47:18 - INFO - codeparrot_training - Step 44944: {'lr': 1.3563591346720804e-05, 'samples': 23011840, 'steps': 44944, 'loss/train': 2.2776660919189453}
+02/26/2022 11:47:23 - INFO - codeparrot_training - Step 44945: {'lr': 1.3558275558764488e-05, 'samples': 23012352, 'steps': 44945, 'loss/train': 2.0558323860168457}
+02/26/2022 11:47:26 - INFO - codeparrot_training - Step 44946: {'lr': 1.3552960783649481e-05, 'samples': 23012864, 'steps': 44946, 'loss/train': 1.8615316152572632}
+02/26/2022 11:47:34 - INFO - codeparrot_training - Step 44947: {'lr': 1.3547647021398435e-05, 'samples': 23013376, 'steps': 44947, 'loss/train': 1.7510186433792114}
+02/26/2022 11:47:37 - INFO - codeparrot_training - Step 44948: {'lr': 1.3542334272034245e-05, 'samples': 23013888, 'steps': 44948, 'loss/train': 2.126037359237671}
+02/26/2022 11:47:43 - INFO - codeparrot_training - Step 44949: {'lr': 1.3537022535579618e-05, 'samples': 23014400, 'steps': 44949, 'loss/train': 1.4729446172714233}
+02/26/2022 11:47:46 - INFO - codeparrot_training - Step 44950: {'lr': 1.3531711812057284e-05, 'samples': 23014912, 'steps': 44950, 'loss/train': 1.7517380714416504}
+02/26/2022 11:47:52 - INFO - codeparrot_training - Step 44951: {'lr': 1.3526402101489977e-05, 'samples': 23015424, 'steps': 44951, 'loss/train': 1.76023268699646}
+02/26/2022 11:47:55 - INFO - codeparrot_training - Step 44952: {'lr': 1.3521093403900508e-05, 'samples': 23015936, 'steps': 44952, 'loss/train': 1.7382863759994507}
+02/26/2022 11:48:01 - INFO - codeparrot_training - Step 44953: {'lr': 1.3515785719311586e-05, 'samples': 23016448, 'steps': 44953, 'loss/train': 1.0642966032028198}
+02/26/2022 11:48:04 - INFO - codeparrot_training - Step 44954: {'lr': 1.3510479047745966e-05, 'samples': 23016960, 'steps': 44954, 'loss/train': 1.8943291902542114}
+02/26/2022 11:48:10 - INFO - codeparrot_training - Step 44955: {'lr': 1.35051733892263e-05, 'samples': 23017472, 'steps': 44955, 'loss/train': 2.3570845127105713}
+02/26/2022 11:48:13 - INFO - codeparrot_training - Step 44956: {'lr': 1.3499868743775401e-05, 'samples': 23017984, 'steps': 44956, 'loss/train': 1.9444695711135864}
+02/26/2022 11:48:20 - INFO - codeparrot_training - Step 44957: {'lr': 1.3494565111416002e-05, 'samples': 23018496, 'steps': 44957, 'loss/train': 1.4855417013168335}
+02/26/2022 11:48:24 - INFO - codeparrot_training - Step 44958: {'lr': 1.3489262492170778e-05, 'samples': 23019008, 'steps': 44958, 'loss/train': 0.9789879322052002}
+02/26/2022 11:48:29 - INFO - codeparrot_training - Step 44959: {'lr': 1.3483960886062435e-05, 'samples': 23019520, 'steps': 44959, 'loss/train': 1.6357616186141968}
+02/26/2022 11:48:33 - INFO - codeparrot_training - Step 44960: {'lr': 1.3478660293113675e-05, 'samples': 23020032, 'steps': 44960, 'loss/train': 1.8695590496063232}
+02/26/2022 11:48:38 - INFO - codeparrot_training - Step 44961: {'lr': 1.347336071334726e-05, 'samples': 23020544, 'steps': 44961, 'loss/train': 2.5499942302703857}
+02/26/2022 11:48:42 - INFO - codeparrot_training - Step 44962: {'lr': 1.3468062146785865e-05, 'samples': 23021056, 'steps': 44962, 'loss/train': 1.955837607383728}
+02/26/2022 11:48:47 - INFO - codeparrot_training - Step 44963: {'lr': 1.3462764593452192e-05, 'samples': 23021568, 'steps': 44963, 'loss/train': 1.4898210763931274}
+02/26/2022 11:48:51 - INFO - codeparrot_training - Step 44964: {'lr': 1.3457468053368893e-05, 'samples': 23022080, 'steps': 44964, 'loss/train': 2.160562753677368}
+02/26/2022 11:48:57 - INFO - codeparrot_training - Step 44965: {'lr': 1.3452172526558726e-05, 'samples': 23022592, 'steps': 44965, 'loss/train': 1.4715397357940674}
+02/26/2022 11:49:00 - INFO - codeparrot_training - Step 44966: {'lr': 1.344687801304431e-05, 'samples': 23023104, 'steps': 44966, 'loss/train': 1.6743314266204834}
+02/26/2022 11:49:07 - INFO - codeparrot_training - Step 44967: {'lr': 1.3441584512848381e-05, 'samples': 23023616, 'steps': 44967, 'loss/train': 1.6812307834625244}
+02/26/2022 11:49:10 - INFO - codeparrot_training - Step 44968: {'lr': 1.3436292025993558e-05, 'samples': 23024128, 'steps': 44968, 'loss/train': 1.242566466331482}
+02/26/2022 11:49:16 - INFO - codeparrot_training - Step 44969: {'lr': 1.343100055250257e-05, 'samples': 23024640, 'steps': 44969, 'loss/train': 0.8106839656829834}
+02/26/2022 11:49:19 - INFO - codeparrot_training - Step 44970: {'lr': 1.3425710092398013e-05, 'samples': 23025152, 'steps': 44970, 'loss/train': 1.2639392614364624}
+02/26/2022 11:49:25 - INFO - codeparrot_training - Step 44971: {'lr': 1.3420420645702675e-05, 'samples': 23025664, 'steps': 44971, 'loss/train': 1.9180103540420532}
+02/26/2022 11:49:28 - INFO - codeparrot_training - Step 44972: {'lr': 1.3415132212439062e-05, 'samples': 23026176, 'steps': 44972, 'loss/train': 2.346921443939209}
+02/26/2022 11:49:34 - INFO - codeparrot_training - Step 44973: {'lr': 1.3409844792629939e-05, 'samples': 23026688, 'steps': 44973, 'loss/train': 1.6661607027053833}
+02/26/2022 11:49:37 - INFO - codeparrot_training - Step 44974: {'lr': 1.3404558386297867e-05, 'samples': 23027200, 'steps': 44974, 'loss/train': 1.7475236654281616}
+02/26/2022 11:49:43 - INFO - codeparrot_training - Step 44975: {'lr': 1.3399272993465605e-05, 'samples': 23027712, 'steps': 44975, 'loss/train': 2.166574716567993}
+02/26/2022 11:49:47 - INFO - codeparrot_training - Step 44976: {'lr': 1.3393988614155666e-05, 'samples': 23028224, 'steps': 44976, 'loss/train': 0.2191791534423828}
+02/26/2022 11:49:53 - INFO - codeparrot_training - Step 44977: {'lr': 1.3388705248390782e-05, 'samples': 23028736, 'steps': 44977, 'loss/train': 0.2064363956451416}
+02/26/2022 11:49:56 - INFO - codeparrot_training - Step 44978: {'lr': 1.3383422896193515e-05, 'samples': 23029248, 'steps': 44978, 'loss/train': 1.958709955215454}
+02/26/2022 11:50:02 - INFO - codeparrot_training - Step 44979: {'lr': 1.3378141557586599e-05, 'samples': 23029760, 'steps': 44979, 'loss/train': 0.9487249851226807}
+02/26/2022 11:50:05 - INFO - codeparrot_training - Step 44980: {'lr': 1.3372861232592542e-05, 'samples': 23030272, 'steps': 44980, 'loss/train': 0.9992273449897766}
+02/26/2022 11:50:11 - INFO - codeparrot_training - Step 44981: {'lr': 1.3367581921234023e-05, 'samples': 23030784, 'steps': 44981, 'loss/train': 0.7073620557785034}
+02/26/2022 11:50:14 - INFO - codeparrot_training - Step 44982: {'lr': 1.3362303623533606e-05, 'samples': 23031296, 'steps': 44982, 'loss/train': 1.50754976272583}
+02/26/2022 11:50:20 - INFO - codeparrot_training - Step 44983: {'lr': 1.3357026339513967e-05, 'samples': 23031808, 'steps': 44983, 'loss/train': 0.4032500982284546}
+02/26/2022 11:50:23 - INFO - codeparrot_training - Step 44984: {'lr': 1.3351750069197699e-05, 'samples': 23032320, 'steps': 44984, 'loss/train': 2.0182955265045166}
+02/26/2022 11:50:29 - INFO - codeparrot_training - Step 44985: {'lr': 1.3346474812607396e-05, 'samples': 23032832, 'steps': 44985, 'loss/train': 2.0755176544189453}
+02/26/2022 11:50:34 - INFO - codeparrot_training - Step 44986: {'lr': 1.3341200569765621e-05, 'samples': 23033344, 'steps': 44986, 'loss/train': 2.204955577850342}
+02/26/2022 11:50:38 - INFO - codeparrot_training - Step 44987: {'lr': 1.3335927340694997e-05, 'samples': 23033856, 'steps': 44987, 'loss/train': 1.1490033864974976}
+02/26/2022 11:50:44 - INFO - codeparrot_training - Step 44988: {'lr': 1.3330655125418145e-05, 'samples': 23034368, 'steps': 44988, 'loss/train': 0.2702863812446594}
+02/26/2022 11:50:47 - INFO - codeparrot_training - Step 44989: {'lr': 1.3325383923957601e-05, 'samples': 23034880, 'steps': 44989, 'loss/train': 2.4160664081573486}
+02/26/2022 11:50:53 - INFO - codeparrot_training - Step 44990: {'lr': 1.3320113736335931e-05, 'samples': 23035392, 'steps': 44990, 'loss/train': 1.412811517715454}
+02/26/2022 11:50:56 - INFO - codeparrot_training - Step 44991: {'lr': 1.3314844562575757e-05, 'samples': 23035904, 'steps': 44991, 'loss/train': 1.1590325832366943}
+02/26/2022 11:51:02 - INFO - codeparrot_training - Step 44992: {'lr': 1.3309576402699641e-05, 'samples': 23036416, 'steps': 44992, 'loss/train': 1.7436206340789795}
+02/26/2022 11:51:05 - INFO - codeparrot_training - Step 44993: {'lr': 1.3304309256730152e-05, 'samples': 23036928, 'steps': 44993, 'loss/train': 2.0711193084716797}
+02/26/2022 11:51:11 - INFO - codeparrot_training - Step 44994: {'lr': 1.3299043124689798e-05, 'samples': 23037440, 'steps': 44994, 'loss/train': 1.8494958877563477}
+02/26/2022 11:51:14 - INFO - codeparrot_training - Step 44995: {'lr': 1.3293778006601225e-05, 'samples': 23037952, 'steps': 44995, 'loss/train': 1.8960152864456177}
+02/26/2022 11:51:20 - INFO - codeparrot_training - Step 44996: {'lr': 1.3288513902486921e-05, 'samples': 23038464, 'steps': 44996, 'loss/train': 1.916840672492981}
+02/26/2022 11:51:23 - INFO - codeparrot_training - Step 44997: {'lr': 1.3283250812369474e-05, 'samples': 23038976, 'steps': 44997, 'loss/train': 1.4744246006011963}
+02/26/2022 11:51:29 - INFO - codeparrot_training - Step 44998: {'lr': 1.3277988736271397e-05, 'samples': 23039488, 'steps': 44998, 'loss/train': 2.744082450866699}
+02/26/2022 11:51:33 - INFO - codeparrot_training - Step 44999: {'lr': 1.327272767421525e-05, 'samples': 23040000, 'steps': 44999, 'loss/train': 2.3884806632995605}
+02/26/2022 11:51:33 - INFO - codeparrot_training - Evaluating and saving model checkpoint