diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -35512,3 +35512,1009 @@ Use FP16 precision: False
 02/25/2022 22:53:07 - INFO - codeparrot_training - Step 34998: {'lr': 0.00011113465271439929, 'samples': 17919488, 'steps': 34998, 'loss/train': 1.1114963293075562}
 02/25/2022 22:53:12 - INFO - codeparrot_training - Step 34999: {'lr': 0.00011112104693229272, 'samples': 17920000, 'steps': 34999, 'loss/train': 2.1389760971069336}
 02/25/2022 22:53:12 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/25/2022 22:53:30 - WARNING - huggingface_hub.repository - Several commits (35) will be pushed upstream.
+02/25/2022 22:53:30 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/25/2022 22:54:05 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   6cccefd..aab1012  floral-grass-11 -> floral-grass-11
+
+02/25/2022 22:54:10 - INFO - codeparrot_training - Step 35000: {'lr': 0.00011110744174509952, 'samples': 17920512, 'steps': 35000, 'loss/train': 1.108235478401184}
+02/25/2022 22:54:16 - INFO - codeparrot_training - Step 35001: {'lr': 0.00011109383715287782, 'samples': 17921024, 'steps': 35001, 'loss/train': 1.5042431354522705}
+02/25/2022 22:54:19 - INFO - codeparrot_training - Step 35002: {'lr': 0.00011108023315568611, 'samples': 17921536, 'steps': 35002, 'loss/train': 1.7382547855377197}
+02/25/2022 22:54:25 - INFO - codeparrot_training - Step 35003: {'lr': 0.00011106662975358256, 'samples': 17922048, 'steps': 35003, 'loss/train': 1.6398383378982544}
+02/25/2022 22:54:29 - INFO - codeparrot_training - Step 35004: {'lr': 0.00011105302694662544, 'samples': 17922560, 'steps': 35004, 'loss/train': 2.1558969020843506}
+02/25/2022 22:54:34 - INFO - codeparrot_training - Step 35005: {'lr': 0.00011103942473487297, 'samples': 17923072, 'steps': 35005, 'loss/train': 2.3898162841796875}
+02/25/2022 22:54:38 - INFO - codeparrot_training - Step 35006: {'lr': 0.00011102582311838358, 'samples': 17923584, 'steps': 35006, 'loss/train': 1.802404522895813}
+02/25/2022 22:54:44 - INFO - codeparrot_training - Step 35007: {'lr': 0.0001110122220972154, 'samples': 17924096, 'steps': 35007, 'loss/train': 0.7885240912437439}
+02/25/2022 22:54:48 - INFO - codeparrot_training - Step 35008: {'lr': 0.00011099862167142674, 'samples': 17924608, 'steps': 35008, 'loss/train': 1.8774694204330444}
+02/25/2022 22:54:53 - INFO - codeparrot_training - Step 35009: {'lr': 0.00011098502184107575, 'samples': 17925120, 'steps': 35009, 'loss/train': 1.8468815088272095}
+02/25/2022 22:54:57 - INFO - codeparrot_training - Step 35010: {'lr': 0.0001109714226062209, 'samples': 17925632, 'steps': 35010, 'loss/train': 1.0692946910858154}
+02/25/2022 22:55:02 - INFO - codeparrot_training - Step 35011: {'lr': 0.0001109578239669203, 'samples': 17926144, 'steps': 35011, 'loss/train': 1.1396328210830688}
+02/25/2022 22:55:06 - INFO - codeparrot_training - Step 35012: {'lr': 0.00011094422592323224, 'samples': 17926656, 'steps': 35012, 'loss/train': 1.6678696870803833}
+02/25/2022 22:55:11 - INFO - codeparrot_training - Step 35013: {'lr': 0.0001109306284752149, 'samples': 17927168, 'steps': 35013, 'loss/train': 1.1875629425048828}
+02/25/2022 22:55:15 - INFO - codeparrot_training - Step 35014: {'lr': 0.00011091703162292666, 'samples': 17927680, 'steps': 35014, 'loss/train': 2.338594436645508}
+02/25/2022 22:55:20 - INFO - codeparrot_training - Step 35015: {'lr': 0.00011090343536642559, 'samples': 17928192, 'steps': 35015, 'loss/train': 1.9597567319869995}
+02/25/2022 22:55:24 - INFO - codeparrot_training - Step 35016: {'lr': 0.00011088983970577021, 'samples': 17928704, 'steps': 35016, 'loss/train': 0.8792717456817627}
+02/25/2022 22:55:29 - INFO - codeparrot_training - Step 35017: {'lr': 0.00011087624464101842, 'samples': 17929216, 'steps': 35017, 'loss/train': 1.7256443500518799}
+02/25/2022 22:55:33 - INFO - codeparrot_training - Step 35018: {'lr': 0.00011086265017222869, 'samples': 17929728, 'steps': 35018, 'loss/train': 1.8258897066116333}
+02/25/2022 22:55:38 - INFO - codeparrot_training - Step 35019: {'lr': 0.00011084905629945912, 'samples': 17930240, 'steps': 35019, 'loss/train': 1.9284745454788208}
+02/25/2022 22:55:42 - INFO - codeparrot_training - Step 35020: {'lr': 0.00011083546302276818, 'samples': 17930752, 'steps': 35020, 'loss/train': 0.4454461932182312}
+02/25/2022 22:55:49 - INFO - codeparrot_training - Step 35021: {'lr': 0.00011082187034221375, 'samples': 17931264, 'steps': 35021, 'loss/train': 0.7002458572387695}
+02/25/2022 22:55:53 - INFO - codeparrot_training - Step 35022: {'lr': 0.0001108082782578543, 'samples': 17931776, 'steps': 35022, 'loss/train': 1.445369839668274}
+02/25/2022 22:55:58 - INFO - codeparrot_training - Step 35023: {'lr': 0.00011079468676974793, 'samples': 17932288, 'steps': 35023, 'loss/train': 2.4348347187042236}
+02/25/2022 22:56:02 - INFO - codeparrot_training - Step 35024: {'lr': 0.0001107810958779531, 'samples': 17932800, 'steps': 35024, 'loss/train': 2.172879934310913}
+02/25/2022 22:56:07 - INFO - codeparrot_training - Step 35025: {'lr': 0.00011076750558252763, 'samples': 17933312, 'steps': 35025, 'loss/train': 1.809644103050232}
+02/25/2022 22:56:11 - INFO - codeparrot_training - Step 35026: {'lr': 0.00011075391588353007, 'samples': 17933824, 'steps': 35026, 'loss/train': 1.2199939489364624}
+02/25/2022 22:56:16 - INFO - codeparrot_training - Step 35027: {'lr': 0.00011074032678101842, 'samples': 17934336, 'steps': 35027, 'loss/train': 2.377208948135376}
+02/25/2022 22:56:20 - INFO - codeparrot_training - Step 35028: {'lr': 0.00011072673827505109, 'samples': 17934848, 'steps': 35028, 'loss/train': 2.5669732093811035}
+02/25/2022 22:56:25 - INFO - codeparrot_training - Step 35029: {'lr': 0.00011071315036568619, 'samples': 17935360, 'steps': 35029, 'loss/train': 1.5660430192947388}
+02/25/2022 22:56:29 - INFO - codeparrot_training - Step 35030: {'lr': 0.00011069956305298193, 'samples': 17935872, 'steps': 35030, 'loss/train': 1.7470132112503052}
+02/25/2022 22:56:34 - INFO - codeparrot_training - Step 35031: {'lr': 0.0001106859763369964, 'samples': 17936384, 'steps': 35031, 'loss/train': 8.209218978881836}
+02/25/2022 22:56:38 - INFO - codeparrot_training - Step 35032: {'lr': 0.00011067239021778803, 'samples': 17936896, 'steps': 35032, 'loss/train': 1.5783425569534302}
+02/25/2022 22:56:45 - INFO - codeparrot_training - Step 35033: {'lr': 0.00011065880469541487, 'samples': 17937408, 'steps': 35033, 'loss/train': 2.4365322589874268}
+02/25/2022 22:56:48 - INFO - codeparrot_training - Step 35034: {'lr': 0.00011064521976993517, 'samples': 17937920, 'steps': 35034, 'loss/train': 0.506431519985199}
+02/25/2022 22:56:54 - INFO - codeparrot_training - Step 35035: {'lr': 0.00011063163544140698, 'samples': 17938432, 'steps': 35035, 'loss/train': 1.369081735610962}
+02/25/2022 22:56:57 - INFO - codeparrot_training - Step 35036: {'lr': 0.0001106180517098887, 'samples': 17938944, 'steps': 35036, 'loss/train': 0.97723788022995}
+02/25/2022 22:57:03 - INFO - codeparrot_training - Step 35037: {'lr': 0.00011060446857543846, 'samples': 17939456, 'steps': 35037, 'loss/train': 2.5222973823547363}
+02/25/2022 22:57:06 - INFO - codeparrot_training - Step 35038: {'lr': 0.00011059088603811438, 'samples': 17939968, 'steps': 35038, 'loss/train': 2.940106153488159}
+02/25/2022 22:57:12 - INFO - codeparrot_training - Step 35039: {'lr': 0.0001105773040979746, 'samples': 17940480, 'steps': 35039, 'loss/train': 0.5782507061958313}
+02/25/2022 22:57:15 - INFO - codeparrot_training - Step 35040: {'lr': 0.00011056372275507747, 'samples': 17940992, 'steps': 35040, 'loss/train': 1.2188520431518555}
+02/25/2022 22:57:21 - INFO - codeparrot_training - Step 35041: {'lr': 0.00011055014200948107, 'samples': 17941504, 'steps': 35041, 'loss/train': 1.243159294128418}
+02/25/2022 22:57:24 - INFO - codeparrot_training - Step 35042: {'lr': 0.00011053656186124358, 'samples': 17942016, 'steps': 35042, 'loss/train': 2.232343912124634}
+02/25/2022 22:57:31 - INFO - codeparrot_training - Step 35043: {'lr': 0.0001105229823104231, 'samples': 17942528, 'steps': 35043, 'loss/train': 1.370757818222046}
+02/25/2022 22:57:35 - INFO - codeparrot_training - Step 35044: {'lr': 0.00011050940335707796, 'samples': 17943040, 'steps': 35044, 'loss/train': 0.0707676112651825}
+02/25/2022 22:57:40 - INFO - codeparrot_training - Step 35045: {'lr': 0.00011049582500126623, 'samples': 17943552, 'steps': 35045, 'loss/train': 1.1718299388885498}
+02/25/2022 22:57:43 - INFO - codeparrot_training - Step 35046: {'lr': 0.0001104822472430461, 'samples': 17944064, 'steps': 35046, 'loss/train': 1.9143428802490234}
+02/25/2022 22:57:49 - INFO - codeparrot_training - Step 35047: {'lr': 0.00011046867008247563, 'samples': 17944576, 'steps': 35047, 'loss/train': 1.6932884454727173}
+02/25/2022 22:57:52 - INFO - codeparrot_training - Step 35048: {'lr': 0.00011045509351961314, 'samples': 17945088, 'steps': 35048, 'loss/train': 2.5862393379211426}
+02/25/2022 22:57:58 - INFO - codeparrot_training - Step 35049: {'lr': 0.00011044151755451676, 'samples': 17945600, 'steps': 35049, 'loss/train': 1.5274930000305176}
+02/25/2022 22:58:02 - INFO - codeparrot_training - Step 35050: {'lr': 0.00011042794218724447, 'samples': 17946112, 'steps': 35050, 'loss/train': 0.43550872802734375}
+02/25/2022 22:58:08 - INFO - codeparrot_training - Step 35051: {'lr': 0.00011041436741785477, 'samples': 17946624, 'steps': 35051, 'loss/train': 1.5613032579421997}
+02/25/2022 22:58:11 - INFO - codeparrot_training - Step 35052: {'lr': 0.0001104007932464054, 'samples': 17947136, 'steps': 35052, 'loss/train': 1.0032163858413696}
+02/25/2022 22:58:15 - INFO - codeparrot_training - Step 35053: {'lr': 0.00011038721967295479, 'samples': 17947648, 'steps': 35053, 'loss/train': 1.9781179428100586}
+02/25/2022 22:58:20 - INFO - codeparrot_training - Step 35054: {'lr': 0.00011037364669756092, 'samples': 17948160, 'steps': 35054, 'loss/train': 1.5532522201538086}
+02/25/2022 22:58:24 - INFO - codeparrot_training - Step 35055: {'lr': 0.0001103600743202822, 'samples': 17948672, 'steps': 35055, 'loss/train': 1.190137505531311}
+02/25/2022 22:58:29 - INFO - codeparrot_training - Step 35056: {'lr': 0.00011034650254117637, 'samples': 17949184, 'steps': 35056, 'loss/train': 1.8900024890899658}
+02/25/2022 22:58:33 - INFO - codeparrot_training - Step 35057: {'lr': 0.0001103329313603019, 'samples': 17949696, 'steps': 35057, 'loss/train': 1.394480586051941}
+02/25/2022 22:58:39 - INFO - codeparrot_training - Step 35058: {'lr': 0.00011031936077771667, 'samples': 17950208, 'steps': 35058, 'loss/train': 1.4922599792480469}
+02/25/2022 22:58:42 - INFO - codeparrot_training - Step 35059: {'lr': 0.00011030579079347916, 'samples': 17950720, 'steps': 35059, 'loss/train': 2.0848727226257324}
+02/25/2022 22:58:49 - INFO - codeparrot_training - Step 35060: {'lr': 0.0001102922214076471, 'samples': 17951232, 'steps': 35060, 'loss/train': 2.190354824066162}
+02/25/2022 22:58:52 - INFO - codeparrot_training - Step 35061: {'lr': 0.0001102786526202789, 'samples': 17951744, 'steps': 35061, 'loss/train': 1.7906067371368408}
+02/25/2022 22:58:58 - INFO - codeparrot_training - Step 35062: {'lr': 0.00011026508443143248, 'samples': 17952256, 'steps': 35062, 'loss/train': 2.0709264278411865}
+02/25/2022 22:59:01 - INFO - codeparrot_training - Step 35063: {'lr': 0.00011025151684116616, 'samples': 17952768, 'steps': 35063, 'loss/train': 2.20731782913208}
+02/25/2022 22:59:07 - INFO - codeparrot_training - Step 35064: {'lr': 0.00011023794984953797, 'samples': 17953280, 'steps': 35064, 'loss/train': 1.8016061782836914}
+02/25/2022 22:59:11 - INFO - codeparrot_training - Step 35065: {'lr': 0.00011022438345660602, 'samples': 17953792, 'steps': 35065, 'loss/train': 1.1277140378952026}
+02/25/2022 22:59:16 - INFO - codeparrot_training - Step 35066: {'lr': 0.00011021081766242836, 'samples': 17954304, 'steps': 35066, 'loss/train': 1.03036367893219}
+02/25/2022 22:59:19 - INFO - codeparrot_training - Step 35067: {'lr': 0.00011019725246706322, 'samples': 17954816, 'steps': 35067, 'loss/train': 1.2721691131591797}
+02/25/2022 22:59:26 - INFO - codeparrot_training - Step 35068: {'lr': 0.00011018368787056868, 'samples': 17955328, 'steps': 35068, 'loss/train': 2.0335490703582764}
+02/25/2022 22:59:29 - INFO - codeparrot_training - Step 35069: {'lr': 0.00011017012387300282, 'samples': 17955840, 'steps': 35069, 'loss/train': 2.016598701477051}
+02/25/2022 22:59:35 - INFO - codeparrot_training - Step 35070: {'lr': 0.00011015656047442366, 'samples': 17956352, 'steps': 35070, 'loss/train': 1.2222919464111328}
+02/25/2022 22:59:39 - INFO - codeparrot_training - Step 35071: {'lr': 0.00011014299767488947, 'samples': 17956864, 'steps': 35071, 'loss/train': 1.6149224042892456}
+02/25/2022 22:59:44 - INFO - codeparrot_training - Step 35072: {'lr': 0.00011012943547445828, 'samples': 17957376, 'steps': 35072, 'loss/train': 2.5101821422576904}
+02/25/2022 22:59:48 - INFO - codeparrot_training - Step 35073: {'lr': 0.00011011587387318819, 'samples': 17957888, 'steps': 35073, 'loss/train': 2.234025239944458}
+02/25/2022 22:59:53 - INFO - codeparrot_training - Step 35074: {'lr': 0.00011010231287113717, 'samples': 17958400, 'steps': 35074, 'loss/train': 1.6715167760849}
+02/25/2022 22:59:57 - INFO - codeparrot_training - Step 35075: {'lr': 0.0001100887524683635, 'samples': 17958912, 'steps': 35075, 'loss/train': 1.2137010097503662}
+02/25/2022 23:00:02 - INFO - codeparrot_training - Step 35076: {'lr': 0.0001100751926649251, 'samples': 17959424, 'steps': 35076, 'loss/train': 2.151000738143921}
+02/25/2022 23:00:06 - INFO - codeparrot_training - Step 35077: {'lr': 0.00011006163346088035, 'samples': 17959936, 'steps': 35077, 'loss/train': 0.8322744369506836}
+02/25/2022 23:00:11 - INFO - codeparrot_training - Step 35078: {'lr': 0.00011004807485628693, 'samples': 17960448, 'steps': 35078, 'loss/train': 1.9726406335830688}
+02/25/2022 23:00:15 - INFO - codeparrot_training - Step 35079: {'lr': 0.00011003451685120319, 'samples': 17960960, 'steps': 35079, 'loss/train': 0.9069707989692688}
+02/25/2022 23:00:22 - INFO - codeparrot_training - Step 35080: {'lr': 0.00011002095944568708, 'samples': 17961472, 'steps': 35080, 'loss/train': 0.8275983929634094}
+02/25/2022 23:00:25 - INFO - codeparrot_training - Step 35081: {'lr': 0.00011000740263979689, 'samples': 17961984, 'steps': 35081, 'loss/train': 1.3482609987258911}
+02/25/2022 23:00:30 - INFO - codeparrot_training - Step 35082: {'lr': 0.00010999384643359033, 'samples': 17962496, 'steps': 35082, 'loss/train': 1.01606285572052}
+02/25/2022 23:00:34 - INFO - codeparrot_training - Step 35083: {'lr': 0.0001099802908271258, 'samples': 17963008, 'steps': 35083, 'loss/train': 1.1263706684112549}
+02/25/2022 23:00:39 - INFO - codeparrot_training - Step 35084: {'lr': 0.00010996673582046124, 'samples': 17963520, 'steps': 35084, 'loss/train': 1.6917088031768799}
+02/25/2022 23:00:43 - INFO - codeparrot_training - Step 35085: {'lr': 0.00010995318141365463, 'samples': 17964032, 'steps': 35085, 'loss/train': 2.3367300033569336}
+02/25/2022 23:00:49 - INFO - codeparrot_training - Step 35086: {'lr': 0.00010993962760676421, 'samples': 17964544, 'steps': 35086, 'loss/train': 1.702036738395691}
+02/25/2022 23:00:52 - INFO - codeparrot_training - Step 35087: {'lr': 0.00010992607439984791, 'samples': 17965056, 'steps': 35087, 'loss/train': 1.744388222694397}
+02/25/2022 23:00:57 - INFO - codeparrot_training - Step 35088: {'lr': 0.00010991252179296387, 'samples': 17965568, 'steps': 35088, 'loss/train': 1.6348741054534912}
+02/25/2022 23:01:01 - INFO - codeparrot_training - Step 35089: {'lr': 0.00010989896978617004, 'samples': 17966080, 'steps': 35089, 'loss/train': 1.9306329488754272}
+02/25/2022 23:01:08 - INFO - codeparrot_training - Step 35090: {'lr': 0.00010988541837952457, 'samples': 17966592, 'steps': 35090, 'loss/train': 1.7279155254364014}
+02/25/2022 23:01:11 - INFO - codeparrot_training - Step 35091: {'lr': 0.00010987186757308551, 'samples': 17967104, 'steps': 35091, 'loss/train': 1.4924261569976807}
+02/25/2022 23:01:17 - INFO - codeparrot_training - Step 35092: {'lr': 0.00010985831736691085, 'samples': 17967616, 'steps': 35092, 'loss/train': 1.0858821868896484}
+02/25/2022 23:01:20 - INFO - codeparrot_training - Step 35093: {'lr': 0.00010984476776105859, 'samples': 17968128, 'steps': 35093, 'loss/train': 2.742690324783325}
+02/25/2022 23:01:26 - INFO - codeparrot_training - Step 35094: {'lr': 0.00010983121875558694, 'samples': 17968640, 'steps': 35094, 'loss/train': 1.5161410570144653}
+02/25/2022 23:01:29 - INFO - codeparrot_training - Step 35095: {'lr': 0.00010981767035055382, 'samples': 17969152, 'steps': 35095, 'loss/train': 1.2386881113052368}
+02/25/2022 23:01:35 - INFO - codeparrot_training - Step 35096: {'lr': 0.00010980412254601729, 'samples': 17969664, 'steps': 35096, 'loss/train': 0.5685620903968811}
+02/25/2022 23:01:38 - INFO - codeparrot_training - Step 35097: {'lr': 0.00010979057534203529, 'samples': 17970176, 'steps': 35097, 'loss/train': 1.8363076448440552}
+02/25/2022 23:01:44 - INFO - codeparrot_training - Step 35098: {'lr': 0.00010977702873866602, 'samples': 17970688, 'steps': 35098, 'loss/train': 2.5595669746398926}
+02/25/2022 23:01:47 - INFO - codeparrot_training - Step 35099: {'lr': 0.00010976348273596745, 'samples': 17971200, 'steps': 35099, 'loss/train': 1.9575541019439697}
+02/25/2022 23:01:53 - INFO - codeparrot_training - Step 35100: {'lr': 0.0001097499373339976, 'samples': 17971712, 'steps': 35100, 'loss/train': 1.759906530380249}
+02/25/2022 23:01:56 - INFO - codeparrot_training - Step 35101: {'lr': 0.00010973639253281437, 'samples': 17972224, 'steps': 35101, 'loss/train': 0.87238609790802}
+02/25/2022 23:02:02 - INFO - codeparrot_training - Step 35102: {'lr': 0.00010972284833247601, 'samples': 17972736, 'steps': 35102, 'loss/train': 1.4301810264587402}
+02/25/2022 23:02:05 - INFO - codeparrot_training - Step 35103: {'lr': 0.00010970930473304041, 'samples': 17973248, 'steps': 35103, 'loss/train': 1.3741896152496338}
+02/25/2022 23:02:11 - INFO - codeparrot_training - Step 35104: {'lr': 0.00010969576173456561, 'samples': 17973760, 'steps': 35104, 'loss/train': 2.065410852432251}
+02/25/2022 23:02:14 - INFO - codeparrot_training - Step 35105: {'lr': 0.00010968221933710951, 'samples': 17974272, 'steps': 35105, 'loss/train': 2.2047057151794434}
+02/25/2022 23:02:21 - INFO - codeparrot_training - Step 35106: {'lr': 0.00010966867754073032, 'samples': 17974784, 'steps': 35106, 'loss/train': 1.1051528453826904}
+02/25/2022 23:02:24 - INFO - codeparrot_training - Step 35107: {'lr': 0.00010965513634548596, 'samples': 17975296, 'steps': 35107, 'loss/train': 1.1823266744613647}
+02/25/2022 23:02:30 - INFO - codeparrot_training - Step 35108: {'lr': 0.00010964159575143445, 'samples': 17975808, 'steps': 35108, 'loss/train': 1.8810521364212036}
+02/25/2022 23:02:33 - INFO - codeparrot_training - Step 35109: {'lr': 0.00010962805575863366, 'samples': 17976320, 'steps': 35109, 'loss/train': 1.592991828918457}
+02/25/2022 23:02:39 - INFO - codeparrot_training - Step 35110: {'lr': 0.00010961451636714182, 'samples': 17976832, 'steps': 35110, 'loss/train': 2.4122071266174316}
+02/25/2022 23:02:42 - INFO - codeparrot_training - Step 35111: {'lr': 0.0001096009775770167, 'samples': 17977344, 'steps': 35111, 'loss/train': 1.9894005060195923}
+02/25/2022 23:02:47 - INFO - codeparrot_training - Step 35112: {'lr': 0.00010958743938831659, 'samples': 17977856, 'steps': 35112, 'loss/train': 1.088016152381897}
+02/25/2022 23:02:51 - INFO - codeparrot_training - Step 35113: {'lr': 0.00010957390180109913, 'samples': 17978368, 'steps': 35113, 'loss/train': 0.5615712404251099}
+02/25/2022 23:02:56 - INFO - codeparrot_training - Step 35114: {'lr': 0.00010956036481542258, 'samples': 17978880, 'steps': 35114, 'loss/train': 2.6407108306884766}
+02/25/2022 23:03:00 - INFO - codeparrot_training - Step 35115: {'lr': 0.00010954682843134472, 'samples': 17979392, 'steps': 35115, 'loss/train': 0.21209581196308136}
+02/25/2022 23:03:06 - INFO - codeparrot_training - Step 35116: {'lr': 0.00010953329264892386, 'samples': 17979904, 'steps': 35116, 'loss/train': 1.5682331323623657}
+02/25/2022 23:03:10 - INFO - codeparrot_training - Step 35117: {'lr': 0.00010951975746821758, 'samples': 17980416, 'steps': 35117, 'loss/train': 1.9463366270065308}
+02/25/2022 23:03:16 - INFO - codeparrot_training - Step 35118: {'lr': 0.00010950622288928413, 'samples': 17980928, 'steps': 35118, 'loss/train': 1.2538259029388428}
+02/25/2022 23:03:19 - INFO - codeparrot_training - Step 35119: {'lr': 0.00010949268891218133, 'samples': 17981440, 'steps': 35119, 'loss/train': 2.1794042587280273}
+02/25/2022 23:03:25 - INFO - codeparrot_training - Step 35120: {'lr': 0.0001094791555369674, 'samples': 17981952, 'steps': 35120, 'loss/train': 1.0278507471084595}
+02/25/2022 23:03:28 - INFO - codeparrot_training - Step 35121: {'lr': 0.00010946562276370001, 'samples': 17982464, 'steps': 35121, 'loss/train': 1.0653961896896362}
+02/25/2022 23:03:33 - INFO - codeparrot_training - Step 35122: {'lr': 0.00010945209059243733, 'samples': 17982976, 'steps': 35122, 'loss/train': 1.7465919256210327}
+02/25/2022 23:03:37 - INFO - codeparrot_training - Step 35123: {'lr': 0.00010943855902323716, 'samples': 17983488, 'steps': 35123, 'loss/train': 1.8433876037597656}
+02/25/2022 23:03:42 - INFO - codeparrot_training - Step 35124: {'lr': 0.00010942502805615767, 'samples': 17984000, 'steps': 35124, 'loss/train': 1.6061464548110962}
+02/25/2022 23:03:46 - INFO - codeparrot_training - Step 35125: {'lr': 0.00010941149769125675, 'samples': 17984512, 'steps': 35125, 'loss/train': 1.7023378610610962}
+02/25/2022 23:03:53 - INFO - codeparrot_training - Step 35126: {'lr': 0.00010939796792859231, 'samples': 17985024, 'steps': 35126, 'loss/train': 2.8829214572906494}
+02/25/2022 23:03:56 - INFO - codeparrot_training - Step 35127: {'lr': 0.00010938443876822222, 'samples': 17985536, 'steps': 35127, 'loss/train': 1.5216530561447144}
+02/25/2022 23:04:02 - INFO - codeparrot_training - Step 35128: {'lr': 0.00010937091021020464, 'samples': 17986048, 'steps': 35128, 'loss/train': 1.276953101158142}
+02/25/2022 23:04:05 - INFO - codeparrot_training - Step 35129: {'lr': 0.00010935738225459743, 'samples': 17986560, 'steps': 35129, 'loss/train': 2.7647812366485596}
+02/25/2022 23:04:10 - INFO - codeparrot_training - Step 35130: {'lr': 0.00010934385490145857, 'samples': 17987072, 'steps': 35130, 'loss/train': 1.482032060623169}
+02/25/2022 23:04:14 - INFO - codeparrot_training - Step 35131: {'lr': 0.00010933032815084593, 'samples': 17987584, 'steps': 35131, 'loss/train': 1.6871857643127441}
+02/25/2022 23:04:19 - INFO - codeparrot_training - Step 35132: {'lr': 0.00010931680200281741, 'samples': 17988096, 'steps': 35132, 'loss/train': 0.3812721073627472}
+02/25/2022 23:04:23 - INFO - codeparrot_training - Step 35133: {'lr': 0.00010930327645743116, 'samples': 17988608, 'steps': 35133, 'loss/train': 1.7362761497497559}
+02/25/2022 23:04:28 - INFO - codeparrot_training - Step 35134: {'lr': 0.00010928975151474496, 'samples': 17989120, 'steps': 35134, 'loss/train': 1.382032871246338}
+02/25/2022 23:04:32 - INFO - codeparrot_training - Step 35135: {'lr': 0.0001092762271748168, 'samples': 17989632, 'steps': 35135, 'loss/train': 2.2625138759613037}
+02/25/2022 23:04:39 - INFO - codeparrot_training - Step 35136: {'lr': 0.0001092627034377045, 'samples': 17990144, 'steps': 35136, 'loss/train': 2.0682082176208496}
+02/25/2022 23:04:42 - INFO - codeparrot_training - Step 35137: {'lr': 0.0001092491803034662, 'samples': 17990656, 'steps': 35137, 'loss/train': 1.8072658777236938}
+02/25/2022 23:04:47 - INFO - codeparrot_training - Step 35138: {'lr': 0.00010923565777215972, 'samples': 17991168, 'steps': 35138, 'loss/train': 0.4985397756099701}
+02/25/2022 23:04:51 - INFO - codeparrot_training - Step 35139: {'lr': 0.00010922213584384296, 'samples': 17991680, 'steps': 35139, 'loss/train': 1.545931339263916}
+02/25/2022 23:04:56 - INFO - codeparrot_training - Step 35140: {'lr': 0.00010920861451857378, 'samples': 17992192, 'steps': 35140, 'loss/train': 1.739908218383789}
+02/25/2022 23:05:00 - INFO - codeparrot_training - Step 35141: {'lr': 0.0001091950937964103, 'samples': 17992704, 'steps': 35141, 'loss/train': 1.2667440176010132}
+02/25/2022 23:05:05 - INFO - codeparrot_training - Step 35142: {'lr': 0.0001091815736774103, 'samples': 17993216, 'steps': 35142, 'loss/train': 0.4341658353805542}
+02/25/2022 23:05:09 - INFO - codeparrot_training - Step 35143: {'lr': 0.00010916805416163176, 'samples': 17993728, 'steps': 35143, 'loss/train': 1.6816037893295288}
+02/25/2022 23:05:14 - INFO - codeparrot_training - Step 35144: {'lr': 0.00010915453524913243, 'samples': 17994240, 'steps': 35144, 'loss/train': 1.0795174837112427}
+02/25/2022 23:05:18 - INFO - codeparrot_training - Step 35145: {'lr': 0.00010914101693997044, 'samples': 17994752, 'steps': 35145, 'loss/train': 1.9348419904708862}
+02/25/2022 23:05:23 - INFO - codeparrot_training - Step 35146: {'lr': 0.00010912749923420354, 'samples': 17995264, 'steps': 35146, 'loss/train': 1.5501378774642944}
+02/25/2022 23:05:29 - INFO - codeparrot_training - Step 35147: {'lr': 0.00010911398213188986, 'samples': 17995776, 'steps': 35147, 'loss/train': 1.7865114212036133}
+02/25/2022 23:05:32 - INFO - codeparrot_training - Step 35148: {'lr': 0.00010910046563308699, 'samples': 17996288, 'steps': 35148, 'loss/train': 1.5714385509490967}
+02/25/2022 23:05:38 - INFO - codeparrot_training - Step 35149: {'lr': 0.00010908694973785305, 'samples': 17996800, 'steps': 35149, 'loss/train': 1.765008568763733}
+02/25/2022 23:05:42 - INFO - codeparrot_training - Step 35150: {'lr': 0.00010907343444624579, 'samples': 17997312, 'steps': 35150, 'loss/train': 1.4733761548995972}
+02/25/2022 23:05:48 - INFO - codeparrot_training - Step 35151: {'lr': 0.00010905991975832335, 'samples': 17997824, 'steps': 35151, 'loss/train': 2.5241777896881104}
+02/25/2022 23:05:52 - INFO - codeparrot_training - Step 35152: {'lr': 0.00010904640567414331, 'samples': 17998336, 'steps': 35152, 'loss/train': 1.0644471645355225}
+02/25/2022 23:05:58 - INFO - codeparrot_training - Step 35153: {'lr': 0.00010903289219376381, 'samples': 17998848, 'steps': 35153, 'loss/train': 2.402423620223999}
+02/25/2022 23:06:01 - INFO - codeparrot_training - Step 35154: {'lr': 0.00010901937931724251, 'samples': 17999360, 'steps': 35154, 'loss/train': 2.248983144760132}
+02/25/2022 23:06:05 - INFO - codeparrot_training - Step 35155: {'lr': 0.0001090058670446376, 'samples': 17999872, 'steps': 35155, 'loss/train': 2.2496323585510254}
+02/25/2022 23:06:10 - INFO - codeparrot_training - Step 35156: {'lr': 0.00010899235537600663, 'samples': 18000384, 'steps': 35156, 'loss/train': 1.2339597940444946}
+02/25/2022 23:06:14 - INFO - codeparrot_training - Step 35157: {'lr': 0.00010897884431140772, 'samples': 18000896, 'steps': 35157, 'loss/train': 2.013819694519043}
+02/25/2022 23:06:19 - INFO - codeparrot_training - Step 35158: {'lr': 0.00010896533385089857, 'samples': 18001408, 'steps': 35158, 'loss/train': 1.181535243988037}
+02/25/2022 23:06:23 - INFO - codeparrot_training - Step 35159: {'lr': 0.00010895182399453724, 'samples': 18001920, 'steps': 35159, 'loss/train': 1.4936811923980713}
+02/25/2022 23:06:28 - INFO - codeparrot_training - Step 35160: {'lr': 0.0001089383147423815, 'samples': 18002432, 'steps': 35160, 'loss/train': 1.084436058998108}
+02/25/2022 23:06:32 - INFO - codeparrot_training - Step 35161: {'lr': 0.00010892480609448924, 'samples': 18002944, 'steps': 35161, 'loss/train': 1.3056615591049194}
+02/25/2022 23:06:37 - INFO - codeparrot_training - Step 35162: {'lr': 0.0001089112980509182, 'samples': 18003456, 'steps': 35162, 'loss/train': 1.7850764989852905}
+02/25/2022 23:06:41 - INFO - codeparrot_training - Step 35163: {'lr': 0.00010889779061172647, 'samples': 18003968, 'steps': 35163, 'loss/train': 1.6812313795089722}
+02/25/2022 23:06:47 - INFO - codeparrot_training - Step 35164: {'lr': 0.00010888428377697182, 'samples': 18004480, 'steps': 35164, 'loss/train': 1.410692811012268}
+02/25/2022 23:06:51 - INFO - codeparrot_training - Step 35165: {'lr': 0.00010887077754671204, 'samples': 18004992, 'steps': 35165, 'loss/train': 1.505059838294983}
+02/25/2022 23:06:56 - INFO - codeparrot_training - Step 35166: {'lr': 0.00010885727192100498, 'samples': 18005504, 'steps': 35166, 'loss/train': 1.4676064252853394}
+02/25/2022 23:07:00 - INFO - codeparrot_training - Step 35167: {'lr': 0.00010884376689990864, 'samples': 18006016, 'steps': 35167, 'loss/train': 2.4696547985076904}
+02/25/2022 23:07:06 - INFO - codeparrot_training - Step 35168: {'lr': 0.00010883026248348076, 'samples': 18006528, 'steps': 35168, 'loss/train': 1.7583203315734863}
+02/25/2022 23:07:09 - INFO - codeparrot_training - Step 35169: {'lr': 0.0001088167586717792, 'samples': 18007040, 'steps': 35169, 'loss/train': 1.964342474937439}
+02/25/2022 23:07:15 - INFO - codeparrot_training - Step 35170: {'lr': 0.00010880325546486174, 'samples': 18007552, 'steps': 35170, 'loss/train': 2.919741153717041}
+02/25/2022 23:07:18 - INFO - codeparrot_training - Step 35171: {'lr': 0.00010878975286278641, 'samples': 18008064, 'steps': 35171, 'loss/train': 1.266263484954834}
+02/25/2022 23:07:25 - INFO - codeparrot_training - Step 35172: {'lr': 0.00010877625086561091, 'samples': 18008576, 'steps': 35172, 'loss/train': 2.008727550506592}
+02/25/2022 23:07:28 - INFO - codeparrot_training - Step 35173: {'lr': 0.00010876274947339312, 'samples': 18009088, 'steps': 35173, 'loss/train': 1.7658183574676514}
+02/25/2022 23:07:34 - INFO - codeparrot_training - Step 35174: {'lr': 0.00010874924868619079, 'samples': 18009600, 'steps': 35174, 'loss/train': 1.285618782043457}
+02/25/2022 23:07:37 - INFO - codeparrot_training - Step 35175: {'lr': 0.0001087357485040619, 'samples': 18010112, 'steps': 35175, 'loss/train': 2.5516717433929443}
+02/25/2022 23:07:42 - INFO - codeparrot_training - Step 35176: {'lr': 0.0001087222489270642, 'samples': 18010624, 'steps': 35176, 'loss/train': 1.502387285232544}
+02/25/2022 23:07:46 - INFO - codeparrot_training - Step 35177: {'lr': 0.00010870874995525553, 'samples': 18011136, 'steps': 35177, 'loss/train': 1.8763298988342285}
+02/25/2022 23:07:51 - INFO - codeparrot_training - Step 35178: {'lr': 0.0001086952515886937, 'samples': 18011648, 'steps': 35178, 'loss/train': 1.0448024272918701}
+02/25/2022 23:07:55 - INFO - codeparrot_training - Step 35179: {'lr': 0.00010868175382743648, 'samples': 18012160, 'steps': 35179, 'loss/train': 1.884372591972351}
+02/25/2022 23:08:00 - INFO - codeparrot_training - Step 35180: {'lr': 0.00010866825667154182, 'samples': 18012672, 'steps': 35180, 'loss/train': 1.6680703163146973}
+02/25/2022 23:08:04 - INFO - codeparrot_training - Step 35181: {'lr': 0.00010865476012106748, 'samples': 18013184, 'steps': 35181, 'loss/train': 2.2585673332214355}
+02/25/2022 23:08:10 - INFO - codeparrot_training - Step 35182: {'lr': 0.00010864126417607126, 'samples': 18013696, 'steps': 35182, 'loss/train': 2.1255857944488525}
+02/25/2022 23:08:14 - INFO - codeparrot_training - Step 35183: {'lr': 0.00010862776883661087, 'samples': 18014208, 'steps': 35183, 'loss/train': 1.3871996402740479}
+02/25/2022 23:08:19 - INFO - codeparrot_training - Step 35184: {'lr': 0.00010861427410274435, 'samples': 18014720, 'steps': 35184, 'loss/train': 1.528178095817566}
+02/25/2022 23:08:23 - INFO - codeparrot_training - Step 35185: {'lr': 0.00010860077997452927, 'samples': 18015232, 'steps': 35185, 'loss/train': 1.3357255458831787}
+02/25/2022 23:08:28 - INFO - codeparrot_training - Step 35186: {'lr': 0.00010858728645202376, 'samples': 18015744, 'steps': 35186, 'loss/train': 1.7962992191314697}
+02/25/2022 23:08:32 - INFO - codeparrot_training - Step 35187: {'lr': 0.0001085737935352852, 'samples': 18016256, 'steps': 35187, 'loss/train': 3.7248575687408447}
+02/25/2022 23:08:37 - INFO - codeparrot_training - Step 35188: {'lr': 0.00010856030122437172, 'samples': 18016768, 'steps': 35188, 'loss/train': 2.344808578491211}
+02/25/2022 23:08:41 - INFO - codeparrot_training - Step 35189: {'lr': 0.00010854680951934087, 'samples': 18017280, 'steps': 35189, 'loss/train': 1.3696507215499878}
+02/25/2022 23:08:46 - INFO - codeparrot_training - Step 35190: {'lr': 0.0001085333184202508, 'samples': 18017792, 'steps': 35190, 'loss/train': 1.771506905555725}
+02/25/2022 23:08:50 - INFO - codeparrot_training - Step 35191: {'lr': 0.00010851982792715886, 'samples': 18018304, 'steps': 35191, 'loss/train': 2.2414228916168213}
+02/25/2022 23:08:55 - INFO - codeparrot_training - Step 35192: {'lr': 0.00010850633804012314, 'samples': 18018816, 'steps': 35192, 'loss/train': 0.7836082577705383}
+02/25/2022 23:08:59 - INFO - codeparrot_training - Step 35193: {'lr': 0.00010849284875920126, 'samples': 18019328, 'steps': 35193, 'loss/train': 2.035017728805542}
+02/25/2022 23:09:04 - INFO - codeparrot_training - Step 35194: {'lr': 0.00010847936008445116, 'samples': 18019840, 'steps': 35194, 'loss/train': 0.7475261688232422}
+02/25/2022 23:09:08 - INFO - codeparrot_training - Step 35195: {'lr': 0.00010846587201593056, 'samples': 18020352, 'steps': 35195, 'loss/train': 0.2526957392692566}
+02/25/2022 23:09:13 - INFO - codeparrot_training - Step 35196: {'lr': 0.00010845238455369722, 'samples': 18020864, 'steps': 35196, 'loss/train': 1.6900062561035156}
+02/25/2022 23:09:17 - INFO - codeparrot_training - Step 35197: {'lr': 0.00010843889769780882, 'samples': 18021376, 'steps': 35197, 'loss/train': 1.048990249633789}
+02/25/2022 23:09:24 - INFO - codeparrot_training - Step 35198: {'lr': 0.00010842541144832332, 'samples': 18021888, 'steps': 35198, 'loss/train': 0.8782660961151123}
+02/25/2022 23:09:27 - INFO - codeparrot_training - Step 35199: {'lr': 0.00010841192580529841, 'samples': 18022400, 'steps': 35199, 'loss/train': 1.7634166479110718}
+02/25/2022 23:09:33 - INFO - codeparrot_training - Step 35200: {'lr': 0.00010839844076879185, 'samples': 18022912, 'steps': 35200, 'loss/train': 0.7242740392684937}
+02/25/2022 23:09:38 - INFO - codeparrot_training - Step 35201: {'lr': 0.0001083849563388613, 'samples': 18023424, 'steps': 35201, 'loss/train': 1.9755542278289795}
+02/25/2022 23:09:42 - INFO - codeparrot_training - Step 35202: {'lr': 0.00010837147251556473, 'samples': 18023936, 'steps': 35202, 'loss/train': 1.863426923751831}
+02/25/2022 23:09:47 - INFO - codeparrot_training - Step 35203: {'lr': 0.00010835798929895982, 'samples': 18024448, 'steps': 35203, 'loss/train': 1.0128402709960938}
+02/25/2022 23:09:51 - INFO - codeparrot_training - Step 35204: {'lr': 0.00010834450668910428, 'samples': 18024960, 'steps': 35204, 'loss/train': 0.3819665312767029}
+02/25/2022 23:09:56 - INFO - codeparrot_training - Step 35205: {'lr': 0.00010833102468605583, 'samples': 18025472, 'steps': 35205, 'loss/train': 0.9572913646697998}
+02/25/2022 23:10:00 - INFO - codeparrot_training - Step 35206: {'lr': 0.00010831754328987235, 'samples': 18025984, 'steps': 35206, 'loss/train': 1.4223482608795166}
+02/25/2022 23:10:05 - INFO - codeparrot_training - Step 35207: {'lr': 0.00010830406250061143, 'samples': 18026496, 'steps': 35207, 'loss/train': 1.9077860116958618}
+02/25/2022 23:10:09 - INFO - codeparrot_training - Step 35208: {'lr': 0.00010829058231833111, 'samples': 18027008, 'steps': 35208, 'loss/train': 0.6413766145706177}
+02/25/2022 23:10:15 - INFO - codeparrot_training - Step 35209: {'lr': 0.00010827710274308874, 'samples': 18027520, 'steps': 35209, 'loss/train': 3.6506991386413574}
+02/25/2022 23:10:19 - INFO - codeparrot_training - Step 35210: {'lr': 0.00010826362377494236, 'samples': 18028032, 'steps': 35210, 'loss/train': 2.284710645675659}
+02/25/2022 23:10:24 - INFO - codeparrot_training - Step 35211: {'lr': 0.00010825014541394961, 'samples': 18028544, 'steps': 35211, 'loss/train': 1.28914475440979}
+02/25/2022 23:10:28 - INFO - codeparrot_training - Step 35212: {'lr': 0.00010823666766016824, 'samples': 18029056, 'steps': 35212, 'loss/train': 1.1726716756820679}
+02/25/2022 23:10:33 - INFO - codeparrot_training - Step 35213: {'lr': 0.00010822319051365587, 'samples': 18029568, 'steps': 35213, 'loss/train': 1.8452595472335815}
+02/25/2022 23:10:37 - INFO - codeparrot_training - Step 35214: {'lr': 0.00010820971397447044, 'samples': 18030080, 'steps': 35214, 'loss/train': 1.6830847263336182}
+02/25/2022 23:10:42 - INFO - codeparrot_training - Step 35215: {'lr': 0.00010819623804266954, 'samples': 18030592, 'steps': 35215, 'loss/train': 1.6310985088348389}
+02/25/2022 23:10:46 - INFO - codeparrot_training - Step 35216: {'lr': 0.00010818276271831093, 'samples': 18031104, 'steps': 35216, 'loss/train': 1.6741418838500977}
+02/25/2022 23:10:51 - INFO - codeparrot_training - Step 35217: {'lr': 0.00010816928800145226, 'samples': 18031616, 'steps': 35217, 'loss/train': 2.0633130073547363}
+02/25/2022 23:10:55 - INFO - codeparrot_training - Step 35218: {'lr': 0.0001081558138921514, 'samples': 18032128, 'steps': 35218, 'loss/train': 2.2960011959075928}
+02/25/2022 23:11:01 - INFO - codeparrot_training - Step 35219: {'lr': 0.00010814234039046603, 'samples': 18032640, 'steps': 35219, 'loss/train': 0.6507323980331421}
+02/25/2022 23:11:05 - INFO - codeparrot_training - Step 35220: {'lr': 0.0001081288674964537, 'samples': 18033152, 'steps': 35220, 'loss/train': 0.778766393661499}
+02/25/2022 23:11:10 - INFO - codeparrot_training - Step 35221: {'lr': 0.00010811539521017236, 'samples': 18033664, 'steps': 35221, 'loss/train': 1.3828907012939453}
+02/25/2022 23:11:14 - INFO - codeparrot_training - Step 35222: {'lr': 0.00010810192353167958, 'samples': 18034176, 'steps': 35222, 'loss/train': 2.4312429428100586}
+02/25/2022 23:11:19 - INFO - codeparrot_training - Step 35223: {'lr': 0.00010808845246103313, 'samples': 18034688, 'steps': 35223, 'loss/train': 1.6656094789505005}
+02/25/2022 23:11:23 - INFO - codeparrot_training - Step 35224: {'lr': 0.00010807498199829061, 'samples': 18035200, 'steps': 35224, 'loss/train': 1.657833218574524}
+02/25/2022 23:11:28 - INFO - codeparrot_training - Step 35225: {'lr': 0.00010806151214350995, 'samples': 18035712, 'steps': 35225, 'loss/train': 0.8315533399581909}
+02/25/2022 23:11:32 - INFO - codeparrot_training - Step 35226: {'lr': 0.00010804804289674855, 'samples': 18036224, 'steps': 35226, 'loss/train': 0.467655211687088}
+02/25/2022 23:11:37 - INFO - codeparrot_training - Step 35227: {'lr': 0.00010803457425806435, 'samples': 18036736, 'steps': 35227, 'loss/train': 1.8076138496398926}
+02/25/2022 23:11:41 - INFO - codeparrot_training - Step 35228: {'lr': 0.00010802110622751485, 'samples': 18037248, 'steps': 35228, 'loss/train': 1.5313631296157837}
+02/25/2022 23:11:47 - INFO - codeparrot_training - Step 35229: {'lr': 0.00010800763880515802, 'samples': 18037760, 'steps': 35229, 'loss/train': 2.445680856704712}
+02/25/2022 23:11:51 - INFO - codeparrot_training - Step 35230: {'lr': 0.00010799417199105121, 'samples': 18038272, 'steps': 35230, 'loss/train': 1.7447046041488647}
+02/25/2022 23:11:56 - INFO - codeparrot_training - Step 35231: {'lr': 0.00010798070578525235, 'samples': 18038784, 'steps': 35231, 'loss/train': 1.5245343446731567}
+02/25/2022 23:12:00 - INFO - codeparrot_training - Step 35232: {'lr': 0.00010796724018781898, 'samples': 18039296, 'steps': 35232, 'loss/train': 1.9911589622497559}
+02/25/2022 23:12:05 - INFO - codeparrot_training - Step 35233: {'lr': 0.00010795377519880894, 'samples': 18039808, 'steps': 35233, 'loss/train': 0.8974743485450745}
+02/25/2022 23:12:09 - INFO - codeparrot_training - Step 35234: {'lr': 0.00010794031081827982, 'samples': 18040320, 'steps': 35234, 'loss/train': 2.3746330738067627}
+02/25/2022 23:12:14 - INFO - codeparrot_training - Step 35235: {'lr': 0.00010792684704628931, 'samples': 18040832, 'steps': 35235, 'loss/train': 0.8331019282341003}
+02/25/2022 23:12:18 - INFO - codeparrot_training - Step 35236: {'lr': 0.00010791338388289495, 'samples': 18041344, 'steps': 35236, 'loss/train': 1.5934780836105347}
+02/25/2022 23:12:24 - INFO - codeparrot_training - Step 35237: {'lr': 0.00010789992132815466, 'samples': 18041856, 'steps': 35237, 'loss/train': 1.2432156801223755}
+02/25/2022 23:12:27 - INFO - codeparrot_training - Step 35238: {'lr': 0.00010788645938212596, 'samples': 18042368, 'steps': 35238, 'loss/train': 1.760582447052002}
+02/25/2022 23:12:33 - INFO - codeparrot_training - Step 35239: {'lr': 0.00010787299804486656, 'samples': 18042880, 'steps': 35239, 'loss/train': 0.0737852081656456}
+02/25/2022 23:12:36 - INFO - codeparrot_training - Step 35240: {'lr': 0.000107859537316434, 'samples': 18043392, 'steps': 35240, 'loss/train': 1.9848288297653198}
+02/25/2022 23:12:41 - INFO - codeparrot_training - Step 35241: {'lr': 0.00010784607719688616, 'samples': 18043904, 'steps': 35241, 'loss/train': 0.8845615983009338}
+02/25/2022 23:12:45 - INFO - codeparrot_training - Step 35242: {'lr': 0.00010783261768628047, 'samples': 18044416, 'steps': 35242, 'loss/train': 1.969267725944519}
+02/25/2022 23:12:50 - INFO - codeparrot_training - Step 35243: {'lr': 0.00010781915878467491, 'samples': 18044928, 'steps': 35243, 'loss/train': 0.9654189944267273}
+02/25/2022 23:12:54 - INFO - codeparrot_training - Step 35244: {'lr': 0.00010780570049212674, 'samples': 18045440, 'steps': 35244, 'loss/train': 1.4688910245895386}
+02/25/2022 23:13:01 - INFO - codeparrot_training - Step 35245: {'lr': 0.00010779224280869387, 'samples': 18045952, 'steps': 35245, 'loss/train': 1.3934355974197388}
+02/25/2022 23:13:04 - INFO - codeparrot_training - Step 35246: {'lr': 0.00010777878573443379, 'samples': 18046464, 'steps': 35246, 'loss/train': 1.5335108041763306}
+02/25/2022 23:13:09 - INFO - codeparrot_training - Step 35247: {'lr': 0.00010776532926940441, 'samples': 18046976, 'steps': 35247, 'loss/train': 1.4982197284698486}
+02/25/2022 23:13:13 - INFO - codeparrot_training - Step 35248: {'lr': 0.00010775187341366301, 'samples': 18047488, 'steps': 35248, 'loss/train': 1.4862370491027832}
+02/25/2022 23:13:18 - INFO - codeparrot_training - Step 35249: {'lr': 0.00010773841816726754, 'samples': 18048000, 'steps': 35249, 'loss/train': 1.0844924449920654}
+02/25/2022 23:13:22 - INFO - codeparrot_training - Step 35250: {'lr': 0.00010772496353027538, 'samples': 18048512, 'steps': 35250, 'loss/train': 1.5317862033843994}
+02/25/2022 23:13:27 - INFO - codeparrot_training - Step 35251: {'lr': 0.00010771150950274453, 'samples': 18049024, 'steps': 35251, 'loss/train': 2.1743288040161133}
+02/25/2022 23:13:31 - INFO - codeparrot_training - Step 35252: {'lr': 0.00010769805608473218, 'samples': 18049536, 'steps': 35252, 'loss/train': 0.7117295861244202}
+02/25/2022 23:13:36 - INFO - codeparrot_training - Step 35253: {'lr': 0.00010768460327629625, 'samples': 18050048, 'steps': 35253, 'loss/train': 1.3391027450561523}
+02/25/2022 23:13:40 - INFO - codeparrot_training - Step 35254: {'lr': 0.0001076711510774942, 'samples': 18050560, 'steps': 35254, 'loss/train': 1.0776550769805908}
+02/25/2022 23:13:47 - INFO - codeparrot_training - Step 35255: {'lr': 0.00010765769948838386, 'samples': 18051072, 'steps': 35255, 'loss/train': 1.361800193786621}
+02/25/2022 23:13:50 - INFO - codeparrot_training - Step 35256: {'lr': 0.00010764424850902271, 'samples': 18051584, 'steps': 35256, 'loss/train': 1.6386168003082275}
+02/25/2022 23:13:56 - INFO - codeparrot_training - Step 35257: {'lr': 0.00010763079813946841, 'samples': 18052096, 'steps': 35257, 'loss/train': 1.8726242780685425}
+02/25/2022 23:13:59 - INFO - codeparrot_training - Step 35258: {'lr': 0.00010761734837977857, 'samples': 18052608, 'steps': 35258, 'loss/train': 1.4635015726089478}
+02/25/2022 23:14:05 - INFO - codeparrot_training - Step 35259: {'lr': 0.0001076038992300107, 'samples': 18053120, 'steps': 35259, 'loss/train': 1.6628472805023193}
+02/25/2022 23:14:08 - INFO - codeparrot_training - Step 35260: {'lr': 0.00010759045069022261, 'samples': 18053632, 'steps': 35260, 'loss/train': 1.4462248086929321}
+02/25/2022 23:14:14 - INFO - codeparrot_training - Step 35261: {'lr': 0.0001075770027604718, 'samples': 18054144, 'steps': 35261, 'loss/train': 2.0075807571411133}
+02/25/2022 23:14:17 - INFO - codeparrot_training - Step 35262: {'lr': 0.00010756355544081589, 'samples': 18054656, 'steps': 35262, 'loss/train': 1.3023672103881836}
+02/25/2022 23:14:23 - INFO - codeparrot_training - Step 35263: {'lr': 0.00010755010873131238, 'samples': 18055168, 'steps': 35263, 'loss/train': 2.8115971088409424}
+02/25/2022 23:14:26 - INFO - codeparrot_training - Step 35264: {'lr': 0.00010753666263201906, 'samples': 18055680, 'steps': 35264, 'loss/train': 1.7383822202682495}
+02/25/2022 23:14:33 - INFO - codeparrot_training - Step 35265: {'lr': 0.00010752321714299345, 'samples': 18056192, 'steps': 35265, 'loss/train': 0.48266932368278503}
+02/25/2022 23:14:36 - INFO - codeparrot_training - Step 35266: {'lr': 0.00010750977226429315, 'samples': 18056704, 'steps': 35266, 'loss/train': 1.9738794565200806}
+02/25/2022 23:14:42 - INFO - codeparrot_training - Step 35267: {'lr': 0.00010749632799597561, 'samples': 18057216, 'steps': 35267, 'loss/train': 1.268410563468933}
+02/25/2022 23:14:45 - INFO - codeparrot_training - Step 35268: {'lr': 0.00010748288433809867, 'samples': 18057728, 'steps': 35268, 'loss/train': 1.5101932287216187}
+02/25/2022 23:14:51 - INFO - codeparrot_training - Step 35269: {'lr': 0.00010746944129071981, 'samples': 18058240, 'steps': 35269, 'loss/train': 1.923219084739685}
+02/25/2022 23:14:56 - INFO - codeparrot_training - Step 35270: {'lr': 0.00010745599885389656, 'samples': 18058752, 'steps': 35270, 'loss/train': 1.9769055843353271}
+02/25/2022 23:15:00 - INFO - codeparrot_training - Step 35271: {'lr': 0.00010744255702768649, 'samples': 18059264, 'steps': 35271, 'loss/train': 1.3716689348220825}
+02/25/2022 23:15:05 - INFO - codeparrot_training - Step 35272: {'lr': 0.00010742911581214734, 'samples': 18059776, 'steps': 35272, 'loss/train': 0.8333598375320435}
+02/25/2022 23:15:09 - INFO - codeparrot_training - Step 35273: {'lr': 0.00010741567520733655, 'samples': 18060288, 'steps': 35273, 'loss/train': 1.40419602394104}
+02/25/2022 23:15:15 - INFO - codeparrot_training - Step 35274: {'lr': 0.00010740223521331177, 'samples': 18060800, 'steps': 35274, 'loss/train': 2.644897937774658}
+02/25/2022 23:15:19 - INFO - codeparrot_training - Step 35275: {'lr': 0.00010738879583013042, 'samples': 18061312, 'steps': 35275, 'loss/train': 0.8513860702514648}
+02/25/2022 23:15:24 - INFO - codeparrot_training - Step 35276: {'lr': 0.00010737535705785028, 'samples': 18061824, 'steps': 35276, 'loss/train': 1.778937816619873}
+02/25/2022 23:15:28 - INFO - codeparrot_training - Step 35277: {'lr': 0.00010736191889652883, 'samples': 18062336, 'steps': 35277, 'loss/train': 1.96829092502594}
+02/25/2022 23:15:33 - INFO - codeparrot_training - Step 35278: {'lr': 0.00010734848134622362, 'samples': 18062848, 'steps': 35278, 'loss/train': 1.4290715456008911}
+02/25/2022 23:15:37 - INFO - codeparrot_training - Step 35279: {'lr': 0.00010733504440699212, 'samples': 18063360, 'steps': 35279, 'loss/train': 1.3900266885757446}
+02/25/2022 23:15:42 - INFO - codeparrot_training - Step 35280: {'lr': 0.0001073216080788921, 'samples': 18063872, 'steps': 35280, 'loss/train': 1.899182915687561}
+02/25/2022 23:15:46 - INFO - codeparrot_training - Step 35281: {'lr': 0.0001073081723619809, 'samples': 18064384, 'steps': 35281, 'loss/train': 0.83708256483078}
+02/25/2022 23:15:51 - INFO - codeparrot_training - Step 35282: {'lr': 0.00010729473725631636, 'samples': 18064896, 'steps': 35282, 'loss/train': 2.265195608139038}
+02/25/2022 23:15:55 - INFO - codeparrot_training - Step 35283: {'lr': 0.00010728130276195569, 'samples': 18065408, 'steps': 35283, 'loss/train': 2.308539867401123}
+02/25/2022 23:16:00 - INFO - codeparrot_training - Step 35284: {'lr': 0.00010726786887895667, 'samples': 18065920, 'steps': 35284, 'loss/train': 2.254373788833618}
+02/25/2022 23:16:04 - INFO - codeparrot_training - Step 35285: {'lr': 0.0001072544356073767, 'samples': 18066432, 'steps': 35285, 'loss/train': 2.101296901702881}
+02/25/2022 23:16:09 - INFO - codeparrot_training - Step 35286: {'lr': 0.00010724100294727357, 'samples': 18066944, 'steps': 35286, 'loss/train': 0.6933637857437134}
+02/25/2022 23:16:13 - INFO - codeparrot_training - Step 35287: {'lr': 0.00010722757089870447, 'samples': 18067456, 'steps': 35287, 'loss/train': 2.2166860103607178}
+02/25/2022 23:16:18 - INFO - codeparrot_training - Step 35288: {'lr': 0.00010721413946172722, 'samples': 18067968, 'steps': 35288, 'loss/train': 1.36432945728302}
+02/25/2022 23:16:22 - INFO - codeparrot_training - Step 35289: {'lr': 0.00010720070863639916, 'samples': 18068480, 'steps': 35289, 'loss/train': 1.2467559576034546}
+02/25/2022 23:16:29 - INFO - codeparrot_training - Step 35290: {'lr': 0.00010718727842277812, 'samples': 18068992, 'steps': 35290, 'loss/train': 1.0521498918533325}
+02/25/2022 23:16:33 - INFO - codeparrot_training - Step 35291: {'lr': 0.00010717384882092124, 'samples': 18069504, 'steps': 35291, 'loss/train': 1.5979316234588623}
+02/25/2022 23:16:38 - INFO - codeparrot_training - Step 35292: {'lr': 0.00010716041983088631, 'samples': 18070016, 'steps': 35292, 'loss/train': 2.316084384918213}
+02/25/2022 23:16:42 - INFO - codeparrot_training - Step 35293: {'lr': 0.00010714699145273071, 'samples': 18070528, 'steps': 35293, 'loss/train': 0.8314520120620728}
+02/25/2022 23:16:47 - INFO - codeparrot_training - Step 35294: {'lr': 0.00010713356368651212, 'samples': 18071040, 'steps': 35294, 'loss/train': 1.5896910429000854}
+02/25/2022 23:16:51 - INFO - codeparrot_training - Step 35295: {'lr': 0.000107120136532288, 'samples': 18071552, 'steps': 35295, 'loss/train': 0.8118664622306824}
+02/25/2022 23:16:56 - INFO - codeparrot_training - Step 35296: {'lr': 0.00010710670999011581, 'samples': 18072064, 'steps': 35296, 'loss/train': 1.6951003074645996}
+02/25/2022 23:17:00 - INFO - codeparrot_training - Step 35297: {'lr': 0.00010709328406005303, 'samples': 18072576, 'steps': 35297, 'loss/train': 2.0402376651763916}
+02/25/2022 23:17:05 - INFO - codeparrot_training - Step 35298: {'lr': 0.00010707985874215734, 'samples': 18073088, 'steps': 35298, 'loss/train': 1.452854871749878}
+02/25/2022 23:17:09 - INFO - codeparrot_training - Step 35299: {'lr': 0.00010706643403648616, 'samples': 18073600, 'steps': 35299, 'loss/train': 2.5785157680511475}
+02/25/2022 23:17:16 - INFO - codeparrot_training - Step 35300: {'lr': 0.00010705300994309697, 'samples': 18074112, 'steps': 35300, 'loss/train': 2.028458595275879}
+02/25/2022 23:17:19 - INFO - codeparrot_training - Step 35301: {'lr': 0.00010703958646204722, 'samples': 18074624, 'steps': 35301, 'loss/train': 1.025451898574829}
+02/25/2022 23:17:25 - INFO - codeparrot_training - Step 35302: {'lr': 0.00010702616359339457, 'samples': 18075136, 'steps': 35302, 'loss/train': 0.8433977961540222}
+02/25/2022 23:17:28 - INFO - codeparrot_training - Step 35303: {'lr': 0.00010701274133719644, 'samples': 18075648, 'steps': 35303, 'loss/train': 2.3498735427856445}
+02/25/2022 23:17:34 - INFO - codeparrot_training - Step 35304: {'lr': 0.00010699931969351032, 'samples': 18076160, 'steps': 35304, 'loss/train': 1.493093490600586}
+02/25/2022 23:17:37 - INFO - codeparrot_training - Step 35305: {'lr': 0.00010698589866239361, 'samples': 18076672, 'steps': 35305, 'loss/train': 1.0323865413665771}
+02/25/2022 23:17:43 - INFO - codeparrot_training - Step 35306: {'lr': 0.000106972478243904, 'samples': 18077184, 'steps': 35306, 'loss/train': 1.8889663219451904}
+02/25/2022 23:17:46 - INFO - codeparrot_training - Step 35307: {'lr': 0.00010695905843809886, 'samples': 18077696, 'steps': 35307, 'loss/train': 1.72743558883667}
+02/25/2022 23:17:52 - INFO - codeparrot_training - Step 35308: {'lr': 0.0001069456392450357, 'samples': 18078208, 'steps': 35308, 'loss/train': 1.4814544916152954}
+02/25/2022 23:17:55 - INFO - codeparrot_training - Step 35309: {'lr': 0.00010693222066477199, 'samples': 18078720, 'steps': 35309, 'loss/train': 1.78554368019104}
+02/25/2022 23:18:02 - INFO - codeparrot_training - Step 35310: {'lr': 0.00010691880269736514, 'samples': 18079232, 'steps': 35310, 'loss/train': 2.375300884246826}
+02/25/2022 23:18:06 - INFO - codeparrot_training - Step 35311: {'lr': 0.00010690538534287278, 'samples': 18079744, 'steps': 35311, 'loss/train': 0.9333917498588562}
+02/25/2022 23:18:11 - INFO - codeparrot_training - Step 35312: {'lr': 0.00010689196860135234, 'samples': 18080256, 'steps': 35312, 'loss/train': 1.6969847679138184}
+02/25/2022 23:18:15 - INFO - codeparrot_training - Step 35313: {'lr': 0.00010687855247286124, 'samples': 18080768, 'steps': 35313, 'loss/train': 2.0370936393737793}
+02/25/2022 23:18:20 - INFO - codeparrot_training - Step 35314: {'lr': 0.00010686513695745689, 'samples': 18081280, 'steps': 35314, 'loss/train': 2.5557234287261963}
+02/25/2022 23:18:24 - INFO - codeparrot_training - Step 35315: {'lr': 0.00010685172205519695, 'samples': 18081792, 'steps': 35315, 'loss/train': 1.0665340423583984}
+02/25/2022 23:18:29 - INFO - codeparrot_training - Step 35316: {'lr': 0.00010683830776613865, 'samples': 18082304, 'steps': 35316, 'loss/train': 1.908714771270752}
+02/25/2022 23:18:33 - INFO - codeparrot_training - Step 35317: {'lr': 0.0001068248940903398, 'samples': 18082816, 'steps': 35317, 'loss/train': 0.5259528160095215}
+02/25/2022 23:18:38 - INFO - codeparrot_training - Step 35318: {'lr': 0.00010681148102785743, 'samples': 18083328, 'steps': 35318, 'loss/train': 1.6868510246276855}
+02/25/2022 23:18:42 - INFO - codeparrot_training - Step 35319: {'lr': 0.00010679806857874933, 'samples': 18083840, 'steps': 35319, 'loss/train': 2.5881948471069336}
+02/25/2022 23:18:49 - INFO - codeparrot_training - Step 35320: {'lr': 0.00010678465674307273, 'samples': 18084352, 'steps': 35320, 'loss/train': 1.679875135421753}
+02/25/2022 23:18:52 - INFO - codeparrot_training - Step 35321: {'lr': 0.00010677124552088535, 'samples': 18084864, 'steps': 35321, 'loss/train': 1.8995732069015503}
+02/25/2022 23:18:58 - INFO - codeparrot_training - Step 35322: {'lr': 0.0001067578349122443, 'samples': 18085376, 'steps': 35322, 'loss/train': 1.9725558757781982}
+02/25/2022 23:19:01 - INFO - codeparrot_training - Step 35323: {'lr': 0.00010674442491720729, 'samples': 18085888, 'steps': 35323, 'loss/train': 0.6953269243240356}
+02/25/2022 23:19:07 - INFO - codeparrot_training - Step 35324: {'lr': 0.00010673101553583159, 'samples': 18086400, 'steps': 35324, 'loss/train': 1.7642712593078613}
+02/25/2022 23:19:10 - INFO - codeparrot_training - Step 35325: {'lr': 0.0001067176067681749, 'samples': 18086912, 'steps': 35325, 'loss/train': 1.4429268836975098}
+02/25/2022 23:19:16 - INFO - codeparrot_training - Step 35326: {'lr': 0.00010670419861429431, 'samples': 18087424, 'steps': 35326, 'loss/train': 0.9882174730300903}
+02/25/2022 23:19:19 - INFO - codeparrot_training - Step 35327: {'lr': 0.00010669079107424753, 'samples': 18087936, 'steps': 35327, 'loss/train': 1.7968116998672485}
+02/25/2022 23:19:25 - INFO - codeparrot_training - Step 35328: {'lr': 0.0001066773841480918, 'samples': 18088448, 'steps': 35328, 'loss/train': 1.209031581878662}
+02/25/2022 23:19:28 - INFO - codeparrot_training - Step 35329: {'lr': 0.00010666397783588475, 'samples': 18088960, 'steps': 35329, 'loss/train': 2.0982604026794434}
+02/25/2022 23:19:34 - INFO - codeparrot_training - Step 35330: {'lr': 0.00010665057213768367, 'samples': 18089472, 'steps': 35330, 'loss/train': 2.2163751125335693}
+02/25/2022 23:19:37 - INFO - codeparrot_training - Step 35331: {'lr': 0.00010663716705354603, 'samples': 18089984, 'steps': 35331, 'loss/train': 0.7657140493392944}
+02/25/2022 23:19:43 - INFO - codeparrot_training - Step 35332: {'lr': 0.00010662376258352919, 'samples': 18090496, 'steps': 35332, 'loss/train': 3.191768169403076}
+02/25/2022 23:19:46 - INFO - codeparrot_training - Step 35333: {'lr': 0.00010661035872769067, 'samples': 18091008, 'steps': 35333, 'loss/train': 2.4975204467773438}
+02/25/2022 23:19:52 - INFO - codeparrot_training - Step 35334: {'lr': 0.00010659695548608785, 'samples': 18091520, 'steps': 35334, 'loss/train': 1.4593830108642578}
+02/25/2022 23:19:55 - INFO - codeparrot_training - Step 35335: {'lr': 0.00010658355285877816, 'samples': 18092032, 'steps': 35335, 'loss/train': 0.7651156187057495}
+02/25/2022 23:20:02 - INFO - codeparrot_training - Step 35336: {'lr': 0.00010657015084581886, 'samples': 18092544, 'steps': 35336, 'loss/train': 1.025732159614563}
+02/25/2022 23:20:06 - INFO - codeparrot_training - Step 35337: {'lr': 0.00010655674944726762, 'samples': 18093056, 'steps': 35337, 'loss/train': 1.3811231851577759}
+02/25/2022 23:20:11 - INFO - codeparrot_training - Step 35338: {'lr': 0.00010654334866318171, 'samples': 18093568, 'steps': 35338, 'loss/train': 2.148015022277832}
+02/25/2022 23:20:15 - INFO - codeparrot_training - Step 35339: {'lr': 0.00010652994849361852, 'samples': 18094080, 'steps': 35339, 'loss/train': 2.7251837253570557}
+02/25/2022 23:20:20 - INFO - codeparrot_training - Step 35340: {'lr': 0.00010651654893863541, 'samples': 18094592, 'steps': 35340, 'loss/train': 1.3495391607284546}
+02/25/2022 23:20:24 - INFO - codeparrot_training - Step 35341: {'lr': 0.00010650314999828992, 'samples': 18095104, 'steps': 35341, 'loss/train': 2.0667080879211426}
+02/25/2022 23:20:29 - INFO - codeparrot_training - Step 35342: {'lr': 0.00010648975167263935, 'samples': 18095616, 'steps': 35342, 'loss/train': 1.8798643350601196}
+02/25/2022 23:20:33 - INFO - codeparrot_training - Step 35343: {'lr': 0.00010647635396174116, 'samples': 18096128, 'steps': 35343, 'loss/train': 0.05614614486694336}
+02/25/2022 23:20:38 - INFO - codeparrot_training - Step 35344: {'lr': 0.00010646295686565257, 'samples': 18096640, 'steps': 35344, 'loss/train': 1.8618308305740356}
+02/25/2022 23:20:42 - INFO - codeparrot_training - Step 35345: {'lr': 0.00010644956038443118, 'samples': 18097152, 'steps': 35345, 'loss/train': 1.8949614763259888}
+02/25/2022 23:20:49 - INFO - codeparrot_training - Step 35346: {'lr': 0.00010643616451813431, 'samples': 18097664, 'steps': 35346, 'loss/train': 1.681885838508606}
+02/25/2022 23:20:52 - INFO - codeparrot_training - Step 35347: {'lr': 0.00010642276926681932, 'samples': 18098176, 'steps': 35347, 'loss/train': 0.7785062789916992}
+02/25/2022 23:20:56 - INFO - codeparrot_training - Step 35348: {'lr': 0.00010640937463054351, 'samples': 18098688, 'steps': 35348, 'loss/train': 1.749366283416748}
+02/25/2022 23:21:01 - INFO - codeparrot_training - Step 35349: {'lr': 0.00010639598060936442, 'samples': 18099200, 'steps': 35349, 'loss/train': 3.2186315059661865}
+02/25/2022 23:21:05 - INFO - codeparrot_training - Step 35350: {'lr': 0.00010638258720333938, 'samples': 18099712, 'steps': 35350, 'loss/train': 1.393624186515808}
+02/25/2022 23:21:10 - INFO - codeparrot_training - Step 35351: {'lr': 0.00010636919441252563, 'samples': 18100224, 'steps': 35351, 'loss/train': 1.5828620195388794}
+02/25/2022 23:21:14 - INFO - codeparrot_training - Step 35352: {'lr': 0.00010635580223698071, 'samples': 18100736, 'steps': 35352, 'loss/train': 2.0323829650878906}
+02/25/2022 23:21:19 - INFO - codeparrot_training - Step 35353: {'lr': 0.00010634241067676196, 'samples': 18101248, 'steps': 35353, 'loss/train': 1.068260908126831}
+02/25/2022 23:21:23 - INFO - codeparrot_training - Step 35354: {'lr': 0.00010632901973192666, 'samples': 18101760, 'steps': 35354, 'loss/train': 2.039034128189087}
+02/25/2022 23:21:28 - INFO - codeparrot_training - Step 35355: {'lr': 0.00010631562940253214, 'samples': 18102272, 'steps': 35355, 'loss/train': 1.6819913387298584}
+02/25/2022 23:21:32 - INFO - codeparrot_training - Step 35356: {'lr': 0.00010630223968863606, 'samples': 18102784, 'steps': 35356, 'loss/train': 1.430169939994812}
+02/25/2022 23:21:39 - INFO - codeparrot_training - Step 35357: {'lr': 0.00010628885059029534, 'samples': 18103296, 'steps': 35357, 'loss/train': 2.618424415588379}
+02/25/2022 23:21:42 - INFO - codeparrot_training - Step 35358: {'lr': 0.00010627546210756763, 'samples': 18103808, 'steps': 35358, 'loss/train': 1.5210448503494263}
+02/25/2022 23:21:48 - INFO - codeparrot_training - Step 35359: {'lr': 0.00010626207424051012, 'samples': 18104320, 'steps': 35359, 'loss/train': 2.4756479263305664}
+02/25/2022 23:21:51 - INFO - codeparrot_training - Step 35360: {'lr': 0.00010624868698918044, 'samples': 18104832, 'steps': 35360, 'loss/train': 0.47095155715942383}
+02/25/2022 23:21:57 - INFO - codeparrot_training - Step 35361: {'lr': 0.00010623530035363557, 'samples': 18105344, 'steps': 35361, 'loss/train': 1.2323811054229736}
+02/25/2022 23:22:00 - INFO - codeparrot_training - Step 35362: {'lr': 0.00010622191433393308, 'samples': 18105856, 'steps': 35362, 'loss/train': 1.3965120315551758}
+02/25/2022 23:22:06 - INFO - codeparrot_training - Step 35363: {'lr': 0.00010620852893013019, 'samples': 18106368, 'steps': 35363, 'loss/train': 2.1996281147003174}
+02/25/2022 23:22:09 - INFO - codeparrot_training - Step 35364: {'lr': 0.00010619514414228437, 'samples': 18106880, 'steps': 35364, 'loss/train': 1.555686593055725}
+02/25/2022 23:22:15 - INFO - codeparrot_training - Step 35365: {'lr': 0.0001061817599704529, 'samples': 18107392, 'steps': 35365, 'loss/train': 2.06492018699646}
+02/25/2022 23:22:18 - INFO - codeparrot_training - Step 35366: {'lr': 0.00010616837641469309, 'samples': 18107904, 'steps': 35366, 'loss/train': 1.7434812784194946}
+02/25/2022 23:22:25 - INFO - codeparrot_training - Step 35367: {'lr': 0.00010615499347506221, 'samples': 18108416, 'steps': 35367, 'loss/train': 1.6285113096237183}
+02/25/2022 23:22:28 - INFO - codeparrot_training - Step 35368: {'lr': 0.00010614161115161777, 'samples': 18108928, 'steps': 35368, 'loss/train': 0.14356619119644165}
+02/25/2022 23:22:34 - INFO - codeparrot_training - Step 35369: {'lr': 0.00010612822944441697, 'samples': 18109440, 'steps': 35369, 'loss/train': 1.6089484691619873}
+02/25/2022 23:22:37 - INFO - codeparrot_training - Step 35370: {'lr': 0.00010611484835351715, 'samples': 18109952, 'steps': 35370, 'loss/train': 0.8704349994659424}
+02/25/2022 23:22:43 - INFO - codeparrot_training - Step 35371: {'lr': 0.0001061014678789755, 'samples': 18110464, 'steps': 35371, 'loss/train': 0.9865365624427795}
+02/25/2022 23:22:47 - INFO - codeparrot_training - Step 35372: {'lr': 0.00010608808802084963, 'samples': 18110976, 'steps': 35372, 'loss/train': 1.3249534368515015}
+02/25/2022 23:22:52 - INFO - codeparrot_training - Step 35373: {'lr': 0.00010607470877919664, 'samples': 18111488, 'steps': 35373, 'loss/train': 1.9026554822921753}
+02/25/2022 23:22:56 - INFO - codeparrot_training - Step 35374: {'lr': 0.00010606133015407393, 'samples': 18112000, 'steps': 35374, 'loss/train': 2.0561776161193848}
+02/25/2022 23:23:01 - INFO - codeparrot_training - Step 35375: {'lr': 0.00010604795214553867, 'samples': 18112512, 'steps': 35375, 'loss/train': 1.7132539749145508}
+02/25/2022 23:23:05 - INFO - codeparrot_training - Step 35376: {'lr': 0.00010603457475364836, 'samples': 18113024, 'steps': 35376, 'loss/train': 1.452529788017273}
+02/25/2022 23:23:11 - INFO - codeparrot_training - Step 35377: {'lr': 0.00010602119797846013, 'samples': 18113536, 'steps': 35377, 'loss/train': 1.4385355710983276}
+02/25/2022 23:23:14 - INFO - codeparrot_training - Step 35378: {'lr': 0.00010600782182003155, 'samples': 18114048, 'steps': 35378, 'loss/train': 2.2396726608276367}
+02/25/2022 23:23:20 - INFO - codeparrot_training - Step 35379: {'lr': 0.00010599444627841956, 'samples': 18114560, 'steps': 35379, 'loss/train': 1.9014734029769897}
+02/25/2022 23:23:23 - INFO - codeparrot_training - Step 35380: {'lr': 0.00010598107135368174, 'samples': 18115072, 'steps': 35380, 'loss/train': 1.8209419250488281}
+02/25/2022 23:23:29 - INFO - codeparrot_training - Step 35381: {'lr': 0.00010596769704587517, 'samples': 18115584, 'steps': 35381, 'loss/train': 1.9149832725524902}
+02/25/2022 23:23:32 - INFO - codeparrot_training - Step 35382: {'lr': 0.00010595432335505745, 'samples': 18116096, 'steps': 35382, 'loss/train': 1.4094489812850952}
+02/25/2022 23:23:39 - INFO - codeparrot_training - Step 35383: {'lr': 0.00010594095028128544, 'samples': 18116608, 'steps': 35383, 'loss/train': 2.4266788959503174}
+02/25/2022 23:23:42 - INFO - codeparrot_training - Step 35384: {'lr': 0.00010592757782461679, 'samples': 18117120, 'steps': 35384, 'loss/train': 1.7293343544006348}
+02/25/2022 23:23:48 - INFO - codeparrot_training - Step 35385: {'lr': 0.00010591420598510856, 'samples': 18117632, 'steps': 35385, 'loss/train': 1.4500473737716675}
+02/25/2022 23:23:51 - INFO - codeparrot_training - Step 35386: {'lr': 0.00010590083476281828, 'samples': 18118144, 'steps': 35386, 'loss/train': 1.6789897680282593}
+02/25/2022 23:23:57 - INFO - codeparrot_training - Step 35387: {'lr': 0.0001058874641578029, 'samples': 18118656, 'steps': 35387, 'loss/train': 1.9555487632751465}
+02/25/2022 23:24:00 - INFO - codeparrot_training - Step 35388: {'lr': 0.00010587409417011995, 'samples': 18119168, 'steps': 35388, 'loss/train': 0.8048198223114014}
+02/25/2022 23:24:06 - INFO - codeparrot_training - Step 35389: {'lr': 0.00010586072479982661, 'samples': 18119680, 'steps': 35389, 'loss/train': 2.1249825954437256}
+02/25/2022 23:24:10 - INFO - codeparrot_training - Step 35390: {'lr': 0.00010584735604698007, 'samples': 18120192, 'steps': 35390, 'loss/train': 0.6738342642784119}
+02/25/2022 23:24:15 - INFO - codeparrot_training - Step 35391: {'lr': 0.00010583398791163779, 'samples': 18120704, 'steps': 35391, 'loss/train': 1.5148173570632935}
+02/25/2022 23:24:19 - INFO - codeparrot_training - Step 35392: {'lr': 0.00010582062039385689, 'samples': 18121216, 'steps': 35392, 'loss/train': 2.119899272918701}
+02/25/2022 23:24:25 - INFO - codeparrot_training - Step 35393: {'lr': 0.00010580725349369469, 'samples': 18121728, 'steps': 35393, 'loss/train': 1.3911043405532837}
+02/25/2022 23:24:29 - INFO - codeparrot_training - Step 35394: {'lr': 0.00010579388721120833, 'samples': 18122240, 'steps': 35394, 'loss/train': 2.0923969745635986}
+02/25/2022 23:24:34 - INFO - codeparrot_training - Step 35395: {'lr': 0.00010578052154645528, 'samples': 18122752, 'steps': 35395, 'loss/train': 2.4774179458618164}
+02/25/2022 23:24:38 - INFO - codeparrot_training - Step 35396: {'lr': 0.00010576715649949268, 'samples': 18123264, 'steps': 35396, 'loss/train': 2.2661447525024414}
+02/25/2022 23:24:43 - INFO - codeparrot_training - Step 35397: {'lr': 0.00010575379207037774, 'samples': 18123776, 'steps': 35397, 'loss/train': 2.8824970722198486}
+02/25/2022 23:24:47 - INFO - codeparrot_training - Step 35398: {'lr': 0.00010574042825916771, 'samples': 18124288, 'steps': 35398, 'loss/train': 2.696951150894165}
+02/25/2022 23:24:52 - INFO - codeparrot_training - Step 35399: {'lr': 0.00010572706506591995, 'samples': 18124800, 'steps': 35399, 'loss/train': 1.6601413488388062}
+02/25/2022 23:24:56 - INFO - codeparrot_training - Step 35400: {'lr': 0.00010571370249069162, 'samples': 18125312, 'steps': 35400, 'loss/train': 2.7954585552215576}
+02/25/2022 23:25:01 - INFO - codeparrot_training - Step 35401: {'lr': 0.00010570034053354, 'samples': 18125824, 'steps': 35401, 'loss/train': 2.2081997394561768}
+02/25/2022 23:25:05 - INFO - codeparrot_training - Step 35402: {'lr': 0.0001056869791945222, 'samples': 18126336, 'steps': 35402, 'loss/train': 1.6273868083953857}
+02/25/2022 23:25:11 - INFO - codeparrot_training - Step 35403: {'lr': 0.00010567361847369561, 'samples': 18126848, 'steps': 35403, 'loss/train': 1.9625701904296875}
+02/25/2022 23:25:15 - INFO - codeparrot_training - Step 35404: {'lr': 0.00010566025837111746, 'samples': 18127360, 'steps': 35404, 'loss/train': 1.2657372951507568}
+02/25/2022 23:25:20 - INFO - codeparrot_training - Step 35405: {'lr': 0.00010564689888684489, 'samples': 18127872, 'steps': 35405, 'loss/train': 0.6094422936439514}
+02/25/2022 23:25:24 - INFO - codeparrot_training - Step 35406: {'lr': 0.00010563354002093511, 'samples': 18128384, 'steps': 35406, 'loss/train': 3.6658592224121094}
+02/25/2022 23:25:29 - INFO - codeparrot_training - Step 35407: {'lr': 0.00010562018177344548, 'samples': 18128896, 'steps': 35407, 'loss/train': 1.5406415462493896}
+02/25/2022 23:25:33 - INFO - codeparrot_training - Step 35408: {'lr': 0.00010560682414443315, 'samples': 18129408, 'steps': 35408, 'loss/train': 1.6650298833847046}
+02/25/2022 23:25:38 - INFO - codeparrot_training - Step 35409: {'lr': 0.00010559346713395531, 'samples': 18129920, 'steps': 35409, 'loss/train': 1.2835291624069214}
+02/25/2022 23:25:42 - INFO - codeparrot_training - Step 35410: {'lr': 0.00010558011074206914, 'samples': 18130432, 'steps': 35410, 'loss/train': 1.1159237623214722}
+02/25/2022 23:25:47 - INFO - codeparrot_training - Step 35411: {'lr': 0.000105566754968832, 'samples': 18130944, 'steps': 35411, 'loss/train': 1.5116472244262695}
+02/25/2022 23:25:51 - INFO - codeparrot_training - Step 35412: {'lr': 0.00010555339981430093, 'samples': 18131456, 'steps': 35412, 'loss/train': 1.252220869064331}
+02/25/2022 23:25:57 - INFO - codeparrot_training - Step 35413: {'lr': 0.00010554004527853341, 'samples': 18131968, 'steps': 35413, 'loss/train': 1.7288554906845093}
+02/25/2022 23:26:01 - INFO - codeparrot_training - Step 35414: {'lr': 0.0001055266913615863, 'samples': 18132480, 'steps': 35414, 'loss/train': 1.1541352272033691}
+02/25/2022 23:26:06 - INFO - codeparrot_training - Step 35415: {'lr': 0.00010551333806351704, 'samples': 18132992, 'steps': 35415, 'loss/train': 2.1531450748443604}
+02/25/2022 23:26:10 - INFO - codeparrot_training - Step 35416: {'lr': 0.00010549998538438265, 'samples': 18133504, 'steps': 35416, 'loss/train': 0.8569403290748596}
+02/25/2022 23:26:16 - INFO - codeparrot_training - Step 35417: {'lr': 0.00010548663332424067, 'samples': 18134016, 'steps': 35417, 'loss/train': 3.468822717666626}
+02/25/2022 23:26:19 - INFO - codeparrot_training - Step 35418: {'lr': 0.00010547328188314784, 'samples': 18134528, 'steps': 35418, 'loss/train': 0.6029434204101562}
+02/25/2022 23:26:25 - INFO - codeparrot_training - Step 35419: {'lr': 0.0001054599310611617, 'samples': 18135040, 'steps': 35419, 'loss/train': 1.1743953227996826}
+02/25/2022 23:26:28 - INFO - codeparrot_training - Step 35420: {'lr': 0.00010544658085833919, 'samples': 18135552, 'steps': 35420, 'loss/train': 0.552441418170929}
+02/25/2022 23:26:34 - INFO - codeparrot_training - Step 35421: {'lr': 0.00010543323127473784, 'samples': 18136064, 'steps': 35421, 'loss/train': 2.759134292602539}
+02/25/2022 23:26:37 - INFO - codeparrot_training - Step 35422: {'lr': 0.00010541988231041444, 'samples': 18136576, 'steps': 35422, 'loss/train': 0.5660808682441711}
+02/25/2022 23:26:44 - INFO - codeparrot_training - Step 35423: {'lr': 0.00010540653396542643, 'samples': 18137088, 'steps': 35423, 'loss/train': 2.1836600303649902}
+02/25/2022 23:26:47 - INFO - codeparrot_training - Step 35424: {'lr': 0.00010539318623983082, 'samples': 18137600, 'steps': 35424, 'loss/train': 1.5300904512405396}
+02/25/2022 23:26:53 - INFO - codeparrot_training - Step 35425: {'lr': 0.00010537983913368499, 'samples': 18138112, 'steps': 35425, 'loss/train': 1.9864635467529297}
+02/25/2022 23:26:56 - INFO - codeparrot_training - Step 35426: {'lr': 0.00010536649264704601, 'samples': 18138624, 'steps': 35426, 'loss/train': 1.4066071510314941}
+02/25/2022 23:27:02 - INFO - codeparrot_training - Step 35427: {'lr': 0.00010535314677997101, 'samples': 18139136, 'steps': 35427, 'loss/train': 2.257549285888672}
+02/25/2022 23:27:06 - INFO - codeparrot_training - Step 35428: {'lr': 0.00010533980153251716, 'samples': 18139648, 'steps': 35428, 'loss/train': 1.54520845413208}
+02/25/2022 23:27:09 - INFO - codeparrot_training - Step 35429: {'lr': 0.00010532645690474172, 'samples': 18140160, 'steps': 35429, 'loss/train': 8.41403579711914}
+02/25/2022 23:27:15 - INFO - codeparrot_training - Step 35430: {'lr': 0.0001053131128967018, 'samples': 18140672, 'steps': 35430, 'loss/train': 0.6244770288467407}
+02/25/2022 23:27:18 - INFO - codeparrot_training - Step 35431: {'lr': 0.0001052997695084546, 'samples': 18141184, 'steps': 35431, 'loss/train': 1.1674411296844482}
+02/25/2022 23:27:24 - INFO - codeparrot_training - Step 35432: {'lr': 0.00010528642674005712, 'samples': 18141696, 'steps': 35432, 'loss/train': 0.9195600152015686}
+02/25/2022 23:27:27 - INFO - codeparrot_training - Step 35433: {'lr': 0.00010527308459156674, 'samples': 18142208, 'steps': 35433, 'loss/train': 2.0221030712127686}
+02/25/2022 23:27:33 - INFO - codeparrot_training - Step 35434: {'lr': 0.00010525974306304048, 'samples': 18142720, 'steps': 35434, 'loss/train': 0.85420823097229}
+02/25/2022 23:27:36 - INFO - codeparrot_training - Step 35435: {'lr': 0.00010524640215453555, 'samples': 18143232, 'steps': 35435, 'loss/train': 1.716320276260376}
+02/25/2022 23:27:42 - INFO - codeparrot_training - Step 35436: {'lr': 0.00010523306186610906, 'samples': 18143744, 'steps': 35436, 'loss/train': 2.248636245727539}
+02/25/2022 23:27:45 - INFO - codeparrot_training - Step 35437: {'lr': 0.00010521972219781805, 'samples': 18144256, 'steps': 35437, 'loss/train': 1.2615989446640015}
+02/25/2022 23:27:50 - INFO - codeparrot_training - Step 35438: {'lr': 0.00010520638314971991, 'samples': 18144768, 'steps': 35438, 'loss/train': 1.4481772184371948}
+02/25/2022 23:27:54 - INFO - codeparrot_training - Step 35439: {'lr': 0.00010519304472187164, 'samples': 18145280, 'steps': 35439, 'loss/train': 1.5319799184799194}
+02/25/2022 23:28:01 - INFO - codeparrot_training - Step 35440: {'lr': 0.00010517970691433035, 'samples': 18145792, 'steps': 35440, 'loss/train': 1.281589388847351}
+02/25/2022 23:28:04 - INFO - codeparrot_training - Step 35441: {'lr': 0.00010516636972715315, 'samples': 18146304, 'steps': 35441, 'loss/train': 1.3462690114974976}
+02/25/2022 23:28:10 - INFO - codeparrot_training - Step 35442: {'lr': 0.00010515303316039732, 'samples': 18146816, 'steps': 35442, 'loss/train': 1.4810951948165894}
+02/25/2022 23:28:13 - INFO - codeparrot_training - Step 35443: {'lr': 0.00010513969721411989, 'samples': 18147328, 'steps': 35443, 'loss/train': 1.9309208393096924}
+02/25/2022 23:28:19 - INFO - codeparrot_training - Step 35444: {'lr': 0.00010512636188837801, 'samples': 18147840, 'steps': 35444, 'loss/train': 1.8213180303573608}
+02/25/2022 23:28:22 - INFO - codeparrot_training - Step 35445: {'lr': 0.00010511302718322871, 'samples': 18148352, 'steps': 35445, 'loss/train': 1.5979044437408447}
+02/25/2022 23:28:28 - INFO - codeparrot_training - Step 35446: {'lr': 0.00010509969309872927, 'samples': 18148864, 'steps': 35446, 'loss/train': 1.7691913843154907}
+02/25/2022 23:28:31 - INFO - codeparrot_training - Step 35447: {'lr': 0.00010508635963493673, 'samples': 18149376, 'steps': 35447, 'loss/train': 1.524527668952942}
+02/25/2022 23:28:37 - INFO - codeparrot_training - Step 35448: {'lr': 0.00010507302679190823, 'samples': 18149888, 'steps': 35448, 'loss/train': 3.2382192611694336}
+02/25/2022 23:28:40 - INFO - codeparrot_training - Step 35449: {'lr': 0.00010505969456970074, 'samples': 18150400, 'steps': 35449, 'loss/train': 1.6202162504196167}
+02/25/2022 23:28:47 - INFO - codeparrot_training - Step 35450: {'lr': 0.00010504636296837161, 'samples': 18150912, 'steps': 35450, 'loss/train': 1.23264479637146}
+02/25/2022 23:28:50 - INFO - codeparrot_training - Step 35451: {'lr': 0.00010503303198797773, 'samples': 18151424, 'steps': 35451, 'loss/train': 2.352633476257324}
+02/25/2022 23:28:56 - INFO - codeparrot_training - Step 35452: {'lr': 0.00010501970162857652, 'samples': 18151936, 'steps': 35452, 'loss/train': 1.6278376579284668}
+02/25/2022 23:28:59 - INFO - codeparrot_training - Step 35453: {'lr': 0.00010500637189022466, 'samples': 18152448, 'steps': 35453, 'loss/train': 1.8849884271621704}
+02/25/2022 23:29:05 - INFO - codeparrot_training - Step 35454: {'lr': 0.00010499304277297955, 'samples': 18152960, 'steps': 35454, 'loss/train': 0.927249550819397}
+02/25/2022 23:29:08 - INFO - codeparrot_training - Step 35455: {'lr': 0.00010497971427689813, 'samples': 18153472, 'steps': 35455, 'loss/train': 0.3041379451751709}
+02/25/2022 23:29:14 - INFO - codeparrot_training - Step 35456: {'lr': 0.00010496638640203774, 'samples': 18153984, 'steps': 35456, 'loss/train': 1.241897463798523}
+02/25/2022 23:29:17 - INFO - codeparrot_training - Step 35457: {'lr': 0.0001049530591484551, 'samples': 18154496, 'steps': 35457, 'loss/train': 1.42458975315094}
+02/25/2022 23:29:23 - INFO - codeparrot_training - Step 35458: {'lr': 0.00010493973251620761, 'samples': 18155008, 'steps': 35458, 'loss/train': 2.3158721923828125}
+02/25/2022 23:29:26 - INFO - codeparrot_training - Step 35459: {'lr': 0.00010492640650535216, 'samples': 18155520, 'steps': 35459, 'loss/train': 1.0699232816696167}
+02/25/2022 23:29:32 - INFO - codeparrot_training - Step 35460: {'lr': 0.00010491308111594597, 'samples': 18156032, 'steps': 35460, 'loss/train': 2.5548694133758545}
+02/25/2022 23:29:35 - INFO - codeparrot_training - Step 35461: {'lr': 0.00010489975634804611, 'samples': 18156544, 'steps': 35461, 'loss/train': 1.7255939245224}
+02/25/2022 23:29:41 - INFO - codeparrot_training - Step 35462: {'lr': 0.0001048864322017096, 'samples': 18157056, 'steps': 35462, 'loss/train': 1.5617070198059082}
+02/25/2022 23:29:44 - INFO - codeparrot_training - Step 35463: {'lr': 0.00010487310867699345, 'samples': 18157568, 'steps': 35463, 'loss/train': 1.327529788017273}
+02/25/2022 23:29:50 - INFO - codeparrot_training - Step 35464: {'lr': 0.00010485978577395492, 'samples': 18158080, 'steps': 35464, 'loss/train': 3.01853346824646}
+02/25/2022 23:29:53 - INFO - codeparrot_training - Step 35465: {'lr': 0.00010484646349265095, 'samples': 18158592, 'steps': 35465, 'loss/train': 0.8646223545074463}
+02/25/2022 23:30:01 - INFO - codeparrot_training - Step 35466: {'lr': 0.00010483314183313866, 'samples': 18159104, 'steps': 35466, 'loss/train': 1.061560034751892}
+02/25/2022 23:30:04 - INFO - codeparrot_training - Step 35467: {'lr': 0.000104819820795475, 'samples': 18159616, 'steps': 35467, 'loss/train': 1.4646320343017578}
+02/25/2022 23:30:10 - INFO - codeparrot_training - Step 35468: {'lr': 0.00010480650037971723, 'samples': 18160128, 'steps': 35468, 'loss/train': 2.607125759124756}
+02/25/2022 23:30:13 - INFO - codeparrot_training - Step 35469: {'lr': 0.0001047931805859223, 'samples': 18160640, 'steps': 35469, 'loss/train': 1.151485800743103}
+02/25/2022 23:30:19 - INFO - codeparrot_training - Step 35470: {'lr': 0.00010477986141414724, 'samples': 18161152, 'steps': 35470, 'loss/train': 0.38980814814567566}
+02/25/2022 23:30:22 - INFO - codeparrot_training - Step 35471: {'lr': 0.00010476654286444912, 'samples': 18161664, 'steps': 35471, 'loss/train': 2.2634479999542236}
+02/25/2022 23:30:28 - INFO - codeparrot_training - Step 35472: {'lr': 0.00010475322493688505, 'samples': 18162176, 'steps': 35472, 'loss/train': 2.209099531173706}
+02/25/2022 23:30:31 - INFO - codeparrot_training - Step 35473: {'lr': 0.00010473990763151206, 'samples': 18162688, 'steps': 35473, 'loss/train': 2.261345386505127}
+02/25/2022 23:30:37 - INFO - codeparrot_training - Step 35474: {'lr': 0.00010472659094838719, 'samples': 18163200, 'steps': 35474, 'loss/train': 2.2410130500793457}
+02/25/2022 23:30:40 - INFO - codeparrot_training - Step 35475: {'lr': 0.00010471327488756738, 'samples': 18163712, 'steps': 35475, 'loss/train': 1.7829253673553467}
+02/25/2022 23:30:47 - INFO - codeparrot_training - Step 35476: {'lr': 0.00010469995944910985, 'samples': 18164224, 'steps': 35476, 'loss/train': 1.9092180728912354}
+02/25/2022 23:30:50 - INFO - codeparrot_training - Step 35477: {'lr': 0.00010468664463307157, 'samples': 18164736, 'steps': 35477, 'loss/train': 1.5434972047805786}
+02/25/2022 23:30:56 - INFO - codeparrot_training - Step 35478: {'lr': 0.00010467333043950956, 'samples': 18165248, 'steps': 35478, 'loss/train': 1.085848093032837}
+02/25/2022 23:30:59 - INFO - codeparrot_training - Step 35479: {'lr': 0.00010466001686848072, 'samples': 18165760, 'steps': 35479, 'loss/train': 2.576321601867676}
+02/25/2022 23:31:05 - INFO - codeparrot_training - Step 35480: {'lr': 0.00010464670392004236, 'samples': 18166272, 'steps': 35480, 'loss/train': 2.178196668624878}
+02/25/2022 23:31:08 - INFO - codeparrot_training - Step 35481: {'lr': 0.00010463339159425137, 'samples': 18166784, 'steps': 35481, 'loss/train': 1.6818640232086182}
+02/25/2022 23:31:13 - INFO - codeparrot_training - Step 35482: {'lr': 0.00010462007989116473, 'samples': 18167296, 'steps': 35482, 'loss/train': 1.6242027282714844}
+02/25/2022 23:31:17 - INFO - codeparrot_training - Step 35483: {'lr': 0.00010460676881083953, 'samples': 18167808, 'steps': 35483, 'loss/train': 2.3885984420776367}
+02/25/2022 23:31:23 - INFO - codeparrot_training - Step 35484: {'lr': 0.00010459345835333267, 'samples': 18168320, 'steps': 35484, 'loss/train': 0.04929136484861374}
+02/25/2022 23:31:26 - INFO - codeparrot_training - Step 35485: {'lr': 0.00010458014851870134, 'samples': 18168832, 'steps': 35485, 'loss/train': 1.3643733263015747}
+02/25/2022 23:31:32 - INFO - codeparrot_training - Step 35486: {'lr': 0.00010456683930700243, 'samples': 18169344, 'steps': 35486, 'loss/train': 1.028764009475708}
+02/25/2022 23:31:36 - INFO - codeparrot_training - Step 35487: {'lr': 0.00010455353071829315, 'samples': 18169856, 'steps': 35487, 'loss/train': 0.9290938377380371}
+02/25/2022 23:31:41 - INFO - codeparrot_training - Step 35488: {'lr': 0.00010454022275263017, 'samples': 18170368, 'steps': 35488, 'loss/train': 2.4605071544647217}
+02/25/2022 23:31:45 - INFO - codeparrot_training - Step 35489: {'lr': 0.00010452691541007078, 'samples': 18170880, 'steps': 35489, 'loss/train': 1.9446767568588257}
+02/25/2022 23:31:50 - INFO - codeparrot_training - Step 35490: {'lr': 0.00010451360869067183, 'samples': 18171392, 'steps': 35490, 'loss/train': 0.18995735049247742}
+02/25/2022 23:31:54 - INFO - codeparrot_training - Step 35491: {'lr': 0.00010450030259449053, 'samples': 18171904, 'steps': 35491, 'loss/train': 2.010329484939575}
+02/25/2022 23:32:00 - INFO - codeparrot_training - Step 35492: {'lr': 0.00010448699712158357, 'samples': 18172416, 'steps': 35492, 'loss/train': 3.019463062286377}
+02/25/2022 23:32:03 - INFO - codeparrot_training - Step 35493: {'lr': 0.0001044736922720082, 'samples': 18172928, 'steps': 35493, 'loss/train': 0.7595480680465698}
+02/25/2022 23:32:08 - INFO - codeparrot_training - Step 35494: {'lr': 0.00010446038804582122, 'samples': 18173440, 'steps': 35494, 'loss/train': 2.158639430999756}
+02/25/2022 23:32:12 - INFO - codeparrot_training - Step 35495: {'lr': 0.0001044470844430799, 'samples': 18173952, 'steps': 35495, 'loss/train': 2.691209077835083}
+02/25/2022 23:32:18 - INFO - codeparrot_training - Step 35496: {'lr': 0.00010443378146384088, 'samples': 18174464, 'steps': 35496, 'loss/train': 1.186385154724121}
+02/25/2022 23:32:21 - INFO - codeparrot_training - Step 35497: {'lr': 0.00010442047910816144, 'samples': 18174976, 'steps': 35497, 'loss/train': 1.7737675905227661}
+02/25/2022 23:32:27 - INFO - codeparrot_training - Step 35498: {'lr': 0.00010440717737609831, 'samples': 18175488, 'steps': 35498, 'loss/train': 1.436472773551941}
+02/25/2022 23:32:30 - INFO - codeparrot_training - Step 35499: {'lr': 0.00010439387626770869, 'samples': 18176000, 'steps': 35499, 'loss/train': 2.16960072517395}
+02/25/2022 23:32:36 - INFO - codeparrot_training - Step 35500: {'lr': 0.0001043805757830495, 'samples': 18176512, 'steps': 35500, 'loss/train': 1.3573946952819824}
+02/25/2022 23:32:39 - INFO - codeparrot_training - Step 35501: {'lr': 0.00010436727592217766, 'samples': 18177024, 'steps': 35501, 'loss/train': 1.2822918891906738}
+02/25/2022 23:32:45 - INFO - codeparrot_training - Step 35502: {'lr': 0.00010435397668515007, 'samples': 18177536, 'steps': 35502, 'loss/train': 1.9201580286026}
+02/25/2022 23:32:48 - INFO - codeparrot_training - Step 35503: {'lr': 0.00010434067807202391, 'samples': 18178048, 'steps': 35503, 'loss/train': 2.1618285179138184}
+02/25/2022 23:32:54 - INFO - codeparrot_training - Step 35504: {'lr': 0.00010432738008285602, 'samples': 18178560, 'steps': 35504, 'loss/train': 1.8175995349884033}
+02/25/2022 23:32:59 - INFO - codeparrot_training - Step 35505: {'lr': 0.00010431408271770338, 'samples': 18179072, 'steps': 35505, 'loss/train': 2.1174609661102295}
+02/25/2022 23:33:03 - INFO - codeparrot_training - Step 35506: {'lr': 0.00010430078597662284, 'samples': 18179584, 'steps': 35506, 'loss/train': 2.102395534515381}
+02/25/2022 23:33:08 - INFO - codeparrot_training - Step 35507: {'lr': 0.0001042874898596716, 'samples': 18180096, 'steps': 35507, 'loss/train': 1.876882553100586}
+02/25/2022 23:33:12 - INFO - codeparrot_training - Step 35508: {'lr': 0.00010427419436690644, 'samples': 18180608, 'steps': 35508, 'loss/train': 1.7557848691940308}
+02/25/2022 23:33:15 - INFO - codeparrot_training - Step 35509: {'lr': 0.00010426089949838438, 'samples': 18181120, 'steps': 35509, 'loss/train': 1.5969310998916626}
+02/25/2022 23:33:21 - INFO - codeparrot_training - Step 35510: {'lr': 0.00010424760525416227, 'samples': 18181632, 'steps': 35510, 'loss/train': 2.2206850051879883}
+02/25/2022 23:33:25 - INFO - codeparrot_training - Step 35511: {'lr': 0.0001042343116342972, 'samples': 18182144, 'steps': 35511, 'loss/train': 2.0760226249694824}
+02/25/2022 23:33:31 - INFO - codeparrot_training - Step 35512: {'lr': 0.00010422101863884598, 'samples': 18182656, 'steps': 35512, 'loss/train': 1.0528029203414917}
+02/25/2022 23:33:35 - INFO - codeparrot_training - Step 35513: {'lr': 0.00010420772626786582, 'samples': 18183168, 'steps': 35513, 'loss/train': 2.6579325199127197}
+02/25/2022 23:33:40 - INFO - codeparrot_training - Step 35514: {'lr': 0.00010419443452141325, 'samples': 18183680, 'steps': 35514, 'loss/train': 1.6120131015777588}
+02/25/2022 23:33:44 - INFO - codeparrot_training - Step 35515: {'lr': 0.00010418114339954554, 'samples': 18184192, 'steps': 35515, 'loss/train': 1.038468599319458}
+02/25/2022 23:33:49 - INFO - codeparrot_training - Step 35516: {'lr': 0.00010416785290231951, 'samples': 18184704, 'steps': 35516, 'loss/train': 1.4201387166976929}
+02/25/2022 23:33:52 - INFO - codeparrot_training - Step 35517: {'lr': 0.00010415456302979209, 'samples': 18185216, 'steps': 35517, 'loss/train': 2.7407658100128174}
+02/25/2022 23:33:58 - INFO - codeparrot_training - Step 35518: {'lr': 0.00010414127378202009, 'samples': 18185728, 'steps': 35518, 'loss/train': 3.2500479221343994}
+02/25/2022 23:34:01 - INFO - codeparrot_training - Step 35519: {'lr': 0.0001041279851590607, 'samples': 18186240, 'steps': 35519, 'loss/train': 0.6206653714179993}
+02/25/2022 23:34:07 - INFO - codeparrot_training - Step 35520: {'lr': 0.00010411469716097066, 'samples': 18186752, 'steps': 35520, 'loss/train': 1.6798497438430786}
+02/25/2022 23:34:10 - INFO - codeparrot_training - Step 35521: {'lr': 0.00010410140978780685, 'samples': 18187264, 'steps': 35521, 'loss/train': 1.5994287729263306}
+02/25/2022 23:34:17 - INFO - codeparrot_training - Step 35522: {'lr': 0.00010408812303962639, 'samples': 18187776, 'steps': 35522, 'loss/train': 1.5160788297653198}
+02/25/2022 23:34:20 - INFO - codeparrot_training - Step 35523: {'lr': 0.00010407483691648606, 'samples': 18188288, 'steps': 35523, 'loss/train': 0.4246375858783722}
+02/25/2022 23:34:26 - INFO - codeparrot_training - Step 35524: {'lr': 0.00010406155141844279, 'samples': 18188800, 'steps': 35524, 'loss/train': 2.1877810955047607}
+02/25/2022 23:34:29 - INFO - codeparrot_training - Step 35525: {'lr': 0.00010404826654555341, 'samples': 18189312, 'steps': 35525, 'loss/train': 0.7963857054710388}
+02/25/2022 23:34:35 - INFO - codeparrot_training - Step 35526: {'lr': 0.00010403498229787497, 'samples': 18189824, 'steps': 35526, 'loss/train': 1.8368951082229614}
+02/25/2022 23:34:38 - INFO - codeparrot_training - Step 35527: {'lr': 0.00010402169867546435, 'samples': 18190336, 'steps': 35527, 'loss/train': 1.940693736076355}
+02/25/2022 23:34:44 - INFO - codeparrot_training - Step 35528: {'lr': 0.00010400841567837843, 'samples': 18190848, 'steps': 35528, 'loss/train': 1.2978272438049316}
+02/25/2022 23:34:47 - INFO - codeparrot_training - Step 35529: {'lr': 0.00010399513330667398, 'samples': 18191360, 'steps': 35529, 'loss/train': 0.6375097632408142}
+02/25/2022 23:34:52 - INFO - codeparrot_training - Step 35530: {'lr': 0.00010398185156040812, 'samples': 18191872, 'steps': 35530, 'loss/train': 2.410601854324341}
+02/25/2022 23:34:56 - INFO - codeparrot_training - Step 35531: {'lr': 0.00010396857043963764, 'samples': 18192384, 'steps': 35531, 'loss/train': 2.1480660438537598}
+02/25/2022 23:35:03 - INFO - codeparrot_training - Step 35532: {'lr': 0.00010395528994441944, 'samples': 18192896, 'steps': 35532, 'loss/train': 1.7839181423187256}
+02/25/2022 23:35:07 - INFO - codeparrot_training - Step 35533: {'lr': 0.0001039420100748103, 'samples': 18193408, 'steps': 35533, 'loss/train': 0.6810885071754456}
+02/25/2022 23:35:12 - INFO - codeparrot_training - Step 35534: {'lr': 0.00010392873083086731, 'samples': 18193920, 'steps': 35534, 'loss/train': 3.1297543048858643}
+02/25/2022 23:35:16 - INFO - codeparrot_training - Step 35535: {'lr': 0.00010391545221264725, 'samples': 18194432, 'steps': 35535, 'loss/train': 0.7706844806671143}
+02/25/2022 23:35:21 - INFO - codeparrot_training - Step 35536: {'lr': 0.00010390217422020701, 'samples': 18194944, 'steps': 35536, 'loss/train': 1.6830925941467285}
+02/25/2022 23:35:25 - INFO - codeparrot_training - Step 35537: {'lr': 0.00010388889685360339, 'samples': 18195456, 'steps': 35537, 'loss/train': 0.2152283489704132}
+02/25/2022 23:35:30 - INFO - codeparrot_training - Step 35538: {'lr': 0.00010387562011289343, 'samples': 18195968, 'steps': 35538, 'loss/train': 1.922654628753662}
+02/25/2022 23:35:34 - INFO - codeparrot_training - Step 35539: {'lr': 0.00010386234399813388, 'samples': 18196480, 'steps': 35539, 'loss/train': 1.8588982820510864}
+02/25/2022 23:35:39 - INFO - codeparrot_training - Step 35540: {'lr': 0.00010384906850938167, 'samples': 18196992, 'steps': 35540, 'loss/train': 2.6272544860839844}
+02/25/2022 23:35:43 - INFO - codeparrot_training - Step 35541: {'lr': 0.00010383579364669354, 'samples': 18197504, 'steps': 35541, 'loss/train': 2.1056222915649414}
+02/25/2022 23:35:49 - INFO - codeparrot_training - Step 35542: {'lr': 0.00010382251941012655, 'samples': 18198016, 'steps': 35542, 'loss/train': 1.777288794517517}
+02/25/2022 23:35:52 - INFO - codeparrot_training - Step 35543: {'lr': 0.00010380924579973747, 'samples': 18198528, 'steps': 35543, 'loss/train': 1.0001877546310425}
+02/25/2022 23:35:58 - INFO - codeparrot_training - Step 35544: {'lr': 0.00010379597281558314, 'samples': 18199040, 'steps': 35544, 'loss/train': 1.594598650932312}
+02/25/2022 23:36:01 - INFO - codeparrot_training - Step 35545: {'lr': 0.0001037827004577204, 'samples': 18199552, 'steps': 35545, 'loss/train': 2.4374799728393555}
+02/25/2022 23:36:07 - INFO - codeparrot_training - Step 35546: {'lr': 0.00010376942872620618, 'samples': 18200064, 'steps': 35546, 'loss/train': 1.3352670669555664}
+02/25/2022 23:36:10 - INFO - codeparrot_training - Step 35547: {'lr': 0.00010375615762109725, 'samples': 18200576, 'steps': 35547, 'loss/train': 2.118468999862671}
+02/25/2022 23:36:16 - INFO - codeparrot_training - Step 35548: {'lr': 0.00010374288714245064, 'samples': 18201088, 'steps': 35548, 'loss/train': 1.0579332113265991}
+02/25/2022 23:36:19 - INFO - codeparrot_training - Step 35549: {'lr': 0.0001037296172903229, 'samples': 18201600, 'steps': 35549, 'loss/train': 2.0325934886932373}
+02/25/2022 23:36:25 - INFO - codeparrot_training - Step 35550: {'lr': 0.00010371634806477112, 'samples': 18202112, 'steps': 35550, 'loss/train': 1.336356282234192}
+02/25/2022 23:36:28 - INFO - codeparrot_training - Step 35551: {'lr': 0.00010370307946585197, 'samples': 18202624, 'steps': 35551, 'loss/train': 1.1712565422058105}
+02/25/2022 23:36:34 - INFO - codeparrot_training - Step 35552: {'lr': 0.00010368981149362256, 'samples': 18203136, 'steps': 35552, 'loss/train': 2.8321943283081055}
+02/25/2022 23:36:37 - INFO - codeparrot_training - Step 35553: {'lr': 0.00010367654414813935, 'samples': 18203648, 'steps': 35553, 'loss/train': 1.9090839624404907}
+02/25/2022 23:36:43 - INFO - codeparrot_training - Step 35554: {'lr': 0.00010366327742945944, 'samples': 18204160, 'steps': 35554, 'loss/train': 0.8347397446632385}
+02/25/2022 23:36:46 - INFO - codeparrot_training - Step 35555: {'lr': 0.00010365001133763948, 'samples': 18204672, 'steps': 35555, 'loss/train': 1.8070425987243652}
+02/25/2022 23:36:52 - INFO - codeparrot_training - Step 35556: {'lr': 0.00010363674587273662, 'samples': 18205184, 'steps': 35556, 'loss/train': 1.78842294216156}
+02/25/2022 23:36:55 - INFO - codeparrot_training - Step 35557: {'lr': 0.00010362348103480726, 'samples': 18205696, 'steps': 35557, 'loss/train': 0.8642957210540771}
+02/25/2022 23:37:01 - INFO - codeparrot_training - Step 35558: {'lr': 0.00010361021682390854, 'samples': 18206208, 'steps': 35558, 'loss/train': 1.9542944431304932}
+02/25/2022 23:37:05 - INFO - codeparrot_training - Step 35559: {'lr': 0.00010359695324009705, 'samples': 18206720, 'steps': 35559, 'loss/train': 0.6702874898910522}
+02/25/2022 23:37:10 - INFO - codeparrot_training - Step 35560: {'lr': 0.00010358369028342985, 'samples': 18207232, 'steps': 35560, 'loss/train': 1.0588432550430298}
+02/25/2022 23:37:14 - INFO - codeparrot_training - Step 35561: {'lr': 0.0001035704279539636, 'samples': 18207744, 'steps': 35561, 'loss/train': 1.6165931224822998}
+02/25/2022 23:37:19 - INFO - codeparrot_training - Step 35562: {'lr': 0.00010355716625175513, 'samples': 18208256, 'steps': 35562, 'loss/train': 1.4471553564071655}
+02/25/2022 23:37:23 - INFO - codeparrot_training - Step 35563: {'lr': 0.00010354390517686118, 'samples': 18208768, 'steps': 35563, 'loss/train': 2.2126753330230713}
+02/25/2022 23:37:29 - INFO - codeparrot_training - Step 35564: {'lr': 0.00010353064472933873, 'samples': 18209280, 'steps': 35564, 'loss/train': 2.165555238723755}
+02/25/2022 23:37:32 - INFO - codeparrot_training - Step 35565: {'lr': 0.00010351738490924451, 'samples': 18209792, 'steps': 35565, 'loss/train': 2.593285322189331}
+02/25/2022 23:37:37 - INFO - codeparrot_training - Step 35566: {'lr': 0.00010350412571663525, 'samples': 18210304, 'steps': 35566, 'loss/train': 1.0473862886428833}
+02/25/2022 23:37:41 - INFO - codeparrot_training - Step 35567: {'lr': 0.00010349086715156784, 'samples': 18210816, 'steps': 35567, 'loss/train': 1.6462459564208984}
+02/25/2022 23:37:47 - INFO - codeparrot_training - Step 35568: {'lr': 0.00010347760921409894, 'samples': 18211328, 'steps': 35568, 'loss/train': 2.1878182888031006}
+02/25/2022 23:37:50 - INFO - codeparrot_training - Step 35569: {'lr': 0.0001034643519042855, 'samples': 18211840, 'steps': 35569, 'loss/train': 1.4180275201797485}
+02/25/2022 23:37:56 - INFO - codeparrot_training - Step 35570: {'lr': 0.00010345109522218426, 'samples': 18212352, 'steps': 35570, 'loss/train': 1.5729774236679077}
+02/25/2022 23:37:59 - INFO - codeparrot_training - Step 35571: {'lr': 0.000103437839167852, 'samples': 18212864, 'steps': 35571, 'loss/train': 1.1597007513046265}
+02/25/2022 23:38:05 - INFO - codeparrot_training - Step 35572: {'lr': 0.00010342458374134539, 'samples': 18213376, 'steps': 35572, 'loss/train': 2.199892520904541}
+02/25/2022 23:38:08 - INFO - codeparrot_training - Step 35573: {'lr': 0.00010341132894272143, 'samples': 18213888, 'steps': 35573, 'loss/train': 1.2314226627349854}
+02/25/2022 23:38:14 - INFO - codeparrot_training - Step 35574: {'lr': 0.00010339807477203678, 'samples': 18214400, 'steps': 35574, 'loss/train': 2.3173766136169434}
+02/25/2022 23:38:17 - INFO - codeparrot_training - Step 35575: {'lr': 0.00010338482122934822, 'samples': 18214912, 'steps': 35575, 'loss/train': 1.3393115997314453}
+02/25/2022 23:38:23 - INFO - codeparrot_training - Step 35576: {'lr': 0.00010337156831471245, 'samples': 18215424, 'steps': 35576, 'loss/train': 2.509542465209961}
+02/25/2022 23:38:26 - INFO - codeparrot_training - Step 35577: {'lr': 0.0001033583160281864, 'samples': 18215936, 'steps': 35577, 'loss/train': 1.648756742477417}
+02/25/2022 23:38:33 - INFO - codeparrot_training - Step 35578: {'lr': 0.00010334506436982679, 'samples': 18216448, 'steps': 35578, 'loss/train': 1.7115774154663086}
+02/25/2022 23:38:37 - INFO - codeparrot_training - Step 35579: {'lr': 0.00010333181333969033, 'samples': 18216960, 'steps': 35579, 'loss/train': 1.1724892854690552}
+02/25/2022 23:38:42 - INFO - codeparrot_training - Step 35580: {'lr': 0.00010331856293783373, 'samples': 18217472, 'steps': 35580, 'loss/train': 1.849133014678955}
+02/25/2022 23:38:46 - INFO - codeparrot_training - Step 35581: {'lr': 0.00010330531316431391, 'samples': 18217984, 'steps': 35581, 'loss/train': 2.23974609375}
+02/25/2022 23:38:51 - INFO - codeparrot_training - Step 35582: {'lr': 0.00010329206401918748, 'samples': 18218496, 'steps': 35582, 'loss/train': 1.8942471742630005}
+02/25/2022 23:38:54 - INFO - codeparrot_training - Step 35583: {'lr': 0.00010327881550251142, 'samples': 18219008, 'steps': 35583, 'loss/train': 1.1225790977478027}
+02/25/2022 23:39:00 - INFO - codeparrot_training - Step 35584: {'lr': 0.00010326556761434213, 'samples': 18219520, 'steps': 35584, 'loss/train': 2.266474723815918}
+02/25/2022 23:39:03 - INFO - codeparrot_training - Step 35585: {'lr': 0.0001032523203547367, 'samples': 18220032, 'steps': 35585, 'loss/train': 1.83694326877594}
+02/25/2022 23:39:09 - INFO - codeparrot_training - Step 35586: {'lr': 0.00010323907372375161, 'samples': 18220544, 'steps': 35586, 'loss/train': 1.8051525354385376}
+02/25/2022 23:39:13 - INFO - codeparrot_training - Step 35587: {'lr': 0.00010322582772144388, 'samples': 18221056, 'steps': 35587, 'loss/train': 1.1442192792892456}
+02/25/2022 23:39:18 - INFO - codeparrot_training - Step 35588: {'lr': 0.00010321258234786996, 'samples': 18221568, 'steps': 35588, 'loss/train': 1.355179786682129}
+02/25/2022 23:39:22 - INFO - codeparrot_training - Step 35589: {'lr': 0.00010319933760308677, 'samples': 18222080, 'steps': 35589, 'loss/train': 0.8855079412460327}
+02/25/2022 23:39:28 - INFO - codeparrot_training - Step 35590: {'lr': 0.00010318609348715096, 'samples': 18222592, 'steps': 35590, 'loss/train': 1.1857198476791382}
+02/25/2022 23:39:31 - INFO - codeparrot_training - Step 35591: {'lr': 0.00010317285000011948, 'samples': 18223104, 'steps': 35591, 'loss/train': 2.095787286758423}
+02/25/2022 23:39:37 - INFO - codeparrot_training - Step 35592: {'lr': 0.00010315960714204869, 'samples': 18223616, 'steps': 35592, 'loss/train': 1.4601294994354248}
+02/25/2022 23:39:40 - INFO - codeparrot_training - Step 35593: {'lr': 0.00010314636491299562, 'samples': 18224128, 'steps': 35593, 'loss/train': 1.440686583518982}
+02/25/2022 23:39:46 - INFO - codeparrot_training - Step 35594: {'lr': 0.00010313312331301677, 'samples': 18224640, 'steps': 35594, 'loss/train': 2.279529571533203}
+02/25/2022 23:39:51 - INFO - codeparrot_training - Step 35595: {'lr': 0.00010311988234216912, 'samples': 18225152, 'steps': 35595, 'loss/train': 1.548601508140564}
+02/25/2022 23:39:55 - INFO - codeparrot_training - Step 35596: {'lr': 0.00010310664200050923, 'samples': 18225664, 'steps': 35596, 'loss/train': 1.823628544807434}
+02/25/2022 23:40:00 - INFO - codeparrot_training - Step 35597: {'lr': 0.00010309340228809383, 'samples': 18226176, 'steps': 35597, 'loss/train': 1.5251606702804565}
+02/25/2022 23:40:04 - INFO - codeparrot_training - Step 35598: {'lr': 0.0001030801632049796, 'samples': 18226688, 'steps': 35598, 'loss/train': 1.4135679006576538}
+02/25/2022 23:40:09 - INFO - codeparrot_training - Step 35599: {'lr': 0.00010306692475122334, 'samples': 18227200, 'steps': 35599, 'loss/train': 2.0083534717559814}
+02/25/2022 23:40:13 - INFO - codeparrot_training - Step 35600: {'lr': 0.00010305368692688174, 'samples': 18227712, 'steps': 35600, 'loss/train': 0.426821768283844}
+02/25/2022 23:40:18 - INFO - codeparrot_training - Step 35601: {'lr': 0.00010304044973201148, 'samples': 18228224, 'steps': 35601, 'loss/train': 1.6690462827682495}
+02/25/2022 23:40:22 - INFO - codeparrot_training - Step 35602: {'lr': 0.00010302721316666919, 'samples': 18228736, 'steps': 35602, 'loss/train': 2.3260042667388916}
+02/25/2022 23:40:28 - INFO - codeparrot_training - Step 35603: {'lr': 0.00010301397723091171, 'samples': 18229248, 'steps': 35603, 'loss/train': 1.3810107707977295}
+02/25/2022 23:40:31 - INFO - codeparrot_training - Step 35604: {'lr': 0.00010300074192479573, 'samples': 18229760, 'steps': 35604, 'loss/train': 1.0180336236953735}
+02/25/2022 23:40:37 - INFO - codeparrot_training - Step 35605: {'lr': 0.00010298750724837786, 'samples': 18230272, 'steps': 35605, 'loss/train': 2.1794779300689697}
+02/25/2022 23:40:40 - INFO - codeparrot_training - Step 35606: {'lr': 0.0001029742732017147, 'samples': 18230784, 'steps': 35606, 'loss/train': 4.077032089233398}
+02/25/2022 23:40:46 - INFO - codeparrot_training - Step 35607: {'lr': 0.00010296103978486321, 'samples': 18231296, 'steps': 35607, 'loss/train': 1.3326951265335083}
+02/25/2022 23:40:49 - INFO - codeparrot_training - Step 35608: {'lr': 0.00010294780699787993, 'samples': 18231808, 'steps': 35608, 'loss/train': 2.435358762741089}
+02/25/2022 23:40:55 - INFO - codeparrot_training - Step 35609: {'lr': 0.00010293457484082155, 'samples': 18232320, 'steps': 35609, 'loss/train': 2.8295412063598633}
+02/25/2022 23:40:58 - INFO - codeparrot_training - Step 35610: {'lr': 0.00010292134331374465, 'samples': 18232832, 'steps': 35610, 'loss/train': 1.5999478101730347}
+02/25/2022 23:41:04 - INFO - codeparrot_training - Step 35611: {'lr': 0.00010290811241670611, 'samples': 18233344, 'steps': 35611, 'loss/train': 1.3185818195343018}
+02/25/2022 23:41:07 - INFO - codeparrot_training - Step 35612: {'lr': 0.0001028948821497625, 'samples': 18233856, 'steps': 35612, 'loss/train': 2.8693010807037354}
+02/25/2022 23:41:13 - INFO - codeparrot_training - Step 35613: {'lr': 0.00010288165251297052, 'samples': 18234368, 'steps': 35613, 'loss/train': 2.090181350708008}
+02/25/2022 23:41:16 - INFO - codeparrot_training - Step 35614: {'lr': 0.00010286842350638681, 'samples': 18234880, 'steps': 35614, 'loss/train': 0.7473689317703247}
+02/25/2022 23:41:22 - INFO - codeparrot_training - Step 35615: {'lr': 0.00010285519513006796, 'samples': 18235392, 'steps': 35615, 'loss/train': 1.6938201189041138}
+02/25/2022 23:41:26 - INFO - codeparrot_training - Step 35616: {'lr': 0.00010284196738407084, 'samples': 18235904, 'steps': 35616, 'loss/train': 1.5739587545394897}
+02/25/2022 23:41:31 - INFO - codeparrot_training - Step 35617: {'lr': 0.00010282874026845191, 'samples': 18236416, 'steps': 35617, 'loss/train': 1.7888811826705933}
+02/25/2022 23:41:35 - INFO - codeparrot_training - Step 35618: {'lr': 0.00010281551378326811, 'samples': 18236928, 'steps': 35618, 'loss/train': 0.931207001209259}
+02/25/2022 23:41:40 - INFO - codeparrot_training - Step 35619: {'lr': 0.00010280228792857571, 'samples': 18237440, 'steps': 35619, 'loss/train': 2.0260813236236572}
+02/25/2022 23:41:44 - INFO - codeparrot_training - Step 35620: {'lr': 0.00010278906270443168, 'samples': 18237952, 'steps': 35620, 'loss/train': 1.224518060684204}
+02/25/2022 23:41:49 - INFO - codeparrot_training - Step 35621: {'lr': 0.00010277583811089247, 'samples': 18238464, 'steps': 35621, 'loss/train': 2.755197286605835}
+02/25/2022 23:41:53 - INFO - codeparrot_training - Step 35622: {'lr': 0.000102762614148015, 'samples': 18238976, 'steps': 35622, 'loss/train': 1.1081726551055908}
+02/25/2022 23:41:58 - INFO - codeparrot_training - Step 35623: {'lr': 0.00010274939081585552, 'samples': 18239488, 'steps': 35623, 'loss/train': 1.8059111833572388}
+02/25/2022 23:42:02 - INFO - codeparrot_training - Step 35624: {'lr': 0.00010273616811447103, 'samples': 18240000, 'steps': 35624, 'loss/train': 0.6258696913719177}
+02/25/2022 23:42:08 - INFO - codeparrot_training - Step 35625: {'lr': 0.00010272294604391791, 'samples': 18240512, 'steps': 35625, 'loss/train': 1.4776450395584106}
+02/25/2022 23:42:11 - INFO - codeparrot_training - Step 35626: {'lr': 0.00010270972460425313, 'samples': 18241024, 'steps': 35626, 'loss/train': 1.4091289043426514}
+02/25/2022 23:42:17 - INFO - codeparrot_training - Step 35627: {'lr': 0.00010269650379553294, 'samples': 18241536, 'steps': 35627, 'loss/train': 2.6466026306152344}
+02/25/2022 23:42:20 - INFO - codeparrot_training - Step 35628: {'lr': 0.00010268328361781423, 'samples': 18242048, 'steps': 35628, 'loss/train': 1.0453681945800781}
+02/25/2022 23:42:26 - INFO - codeparrot_training - Step 35629: {'lr': 0.00010267006407115346, 'samples': 18242560, 'steps': 35629, 'loss/train': 2.5418148040771484}
+02/25/2022 23:42:30 - INFO - codeparrot_training - Step 35630: {'lr': 0.00010265684515560745, 'samples': 18243072, 'steps': 35630, 'loss/train': 8.621676445007324}
+02/25/2022 23:42:33 - INFO - codeparrot_training - Step 35631: {'lr': 0.00010264362687123274, 'samples': 18243584, 'steps': 35631, 'loss/train': 3.1544809341430664}
+02/25/2022 23:42:39 - INFO - codeparrot_training - Step 35632: {'lr': 0.00010263040921808592, 'samples': 18244096, 'steps': 35632, 'loss/train': 1.3971437215805054}
+02/25/2022 23:42:44 - INFO - codeparrot_training - Step 35633: {'lr': 0.00010261719219622356, 'samples': 18244608, 'steps': 35633, 'loss/train': 1.9479739665985107}
+02/25/2022 23:42:48 - INFO - codeparrot_training - Step 35634: {'lr': 0.00010260397580570244, 'samples': 18245120, 'steps': 35634, 'loss/train': 1.7317043542861938}
+02/25/2022 23:42:54 - INFO - codeparrot_training - Step 35635: {'lr': 0.00010259076004657905, 'samples': 18245632, 'steps': 35635, 'loss/train': 2.0960140228271484}
+02/25/2022 23:42:57 - INFO - codeparrot_training - Step 35636: {'lr': 0.00010257754491891009, 'samples': 18246144, 'steps': 35636, 'loss/train': 1.5212658643722534}
+02/25/2022 23:43:03 - INFO - codeparrot_training - Step 35637: {'lr': 0.00010256433042275199, 'samples': 18246656, 'steps': 35637, 'loss/train': 2.1963188648223877}
+02/25/2022 23:43:06 - INFO - codeparrot_training - Step 35638: {'lr': 0.00010255111655816157, 'samples': 18247168, 'steps': 35638, 'loss/train': 2.380225419998169}
+02/25/2022 23:43:12 - INFO - codeparrot_training - Step 35639: {'lr': 0.00010253790332519536, 'samples': 18247680, 'steps': 35639, 'loss/train': 1.555729627609253}
+02/25/2022 23:43:16 - INFO - codeparrot_training - Step 35640: {'lr': 0.00010252469072390994, 'samples': 18248192, 'steps': 35640, 'loss/train': 2.052687406539917}
+02/25/2022 23:43:21 - INFO - codeparrot_training - Step 35641: {'lr': 0.00010251147875436185, 'samples': 18248704, 'steps': 35641, 'loss/train': 1.7617690563201904}
+02/25/2022 23:43:25 - INFO - codeparrot_training - Step 35642: {'lr': 0.00010249826741660781, 'samples': 18249216, 'steps': 35642, 'loss/train': 2.667482852935791}
+02/25/2022 23:43:30 - INFO - codeparrot_training - Step 35643: {'lr': 0.00010248505671070427, 'samples': 18249728, 'steps': 35643, 'loss/train': 1.1361576318740845}
+02/25/2022 23:43:34 - INFO - codeparrot_training - Step 35644: {'lr': 0.00010247184663670812, 'samples': 18250240, 'steps': 35644, 'loss/train': 2.1625943183898926}
+02/25/2022 23:43:39 - INFO - codeparrot_training - Step 35645: {'lr': 0.00010245863719467552, 'samples': 18250752, 'steps': 35645, 'loss/train': 1.7517374753952026}
+02/25/2022 23:43:43 - INFO - codeparrot_training - Step 35646: {'lr': 0.00010244542838466337, 'samples': 18251264, 'steps': 35646, 'loss/train': 1.5311287641525269}
+02/25/2022 23:43:48 - INFO - codeparrot_training - Step 35647: {'lr': 0.00010243222020672812, 'samples': 18251776, 'steps': 35647, 'loss/train': 2.803532600402832}
+02/25/2022 23:43:52 - INFO - codeparrot_training - Step 35648: {'lr': 0.00010241901266092644, 'samples': 18252288, 'steps': 35648, 'loss/train': 2.301795482635498}
+02/25/2022 23:43:57 - INFO - codeparrot_training - Step 35649: {'lr': 0.0001024058057473147, 'samples': 18252800, 'steps': 35649, 'loss/train': 1.5233286619186401}
+02/25/2022 23:44:00 - INFO - codeparrot_training - Step 35650: {'lr': 0.00010239259946594976, 'samples': 18253312, 'steps': 35650, 'loss/train': 2.147953748703003}
+02/25/2022 23:44:07 - INFO - codeparrot_training - Step 35651: {'lr': 0.00010237939381688801, 'samples': 18253824, 'steps': 35651, 'loss/train': 0.2026338279247284}
+02/25/2022 23:44:10 - INFO - codeparrot_training - Step 35652: {'lr': 0.00010236618880018609, 'samples': 18254336, 'steps': 35652, 'loss/train': 1.5804471969604492}
+02/25/2022 23:44:16 - INFO - codeparrot_training - Step 35653: {'lr': 0.00010235298441590044, 'samples': 18254848, 'steps': 35653, 'loss/train': 2.2368381023406982}
+02/25/2022 23:44:19 - INFO - codeparrot_training - Step 35654: {'lr': 0.0001023397806640878, 'samples': 18255360, 'steps': 35654, 'loss/train': 1.8397973775863647}
+02/25/2022 23:44:25 - INFO - codeparrot_training - Step 35655: {'lr': 0.00010232657754480467, 'samples': 18255872, 'steps': 35655, 'loss/train': 2.2479805946350098}
+02/25/2022 23:44:28 - INFO - codeparrot_training - Step 35656: {'lr': 0.0001023133750581075, 'samples': 18256384, 'steps': 35656, 'loss/train': 1.7291384935379028}
+02/25/2022 23:44:34 - INFO - codeparrot_training - Step 35657: {'lr': 0.00010230017320405302, 'samples': 18256896, 'steps': 35657, 'loss/train': 0.8899371027946472}
+02/25/2022 23:44:37 - INFO - codeparrot_training - Step 35658: {'lr': 0.00010228697198269771, 'samples': 18257408, 'steps': 35658, 'loss/train': 0.680962860584259}
+02/25/2022 23:44:43 - INFO - codeparrot_training - Step 35659: {'lr': 0.00010227377139409808, 'samples': 18257920, 'steps': 35659, 'loss/train': 2.230807304382324}
+02/25/2022 23:44:47 - INFO - codeparrot_training - Step 35660: {'lr': 0.00010226057143831064, 'samples': 18258432, 'steps': 35660, 'loss/train': 2.2269339561462402}
+02/25/2022 23:44:53 - INFO - codeparrot_training - Step 35661: {'lr': 0.00010224737211539217, 'samples': 18258944, 'steps': 35661, 'loss/train': 1.005497932434082}
+02/25/2022 23:44:56 - INFO - codeparrot_training - Step 35662: {'lr': 0.00010223417342539884, 'samples': 18259456, 'steps': 35662, 'loss/train': 1.2220706939697266}
+02/25/2022 23:45:02 - INFO - codeparrot_training - Step 35663: {'lr': 0.0001022209753683875, 'samples': 18259968, 'steps': 35663, 'loss/train': 1.8131171464920044}
+02/25/2022 23:45:06 - INFO - codeparrot_training - Step 35664: {'lr': 0.00010220777794441449, 'samples': 18260480, 'steps': 35664, 'loss/train': 0.9779194593429565}
+02/25/2022 23:45:11 - INFO - codeparrot_training - Step 35665: {'lr': 0.00010219458115353661, 'samples': 18260992, 'steps': 35665, 'loss/train': 2.191406011581421}
+02/25/2022 23:45:15 - INFO - codeparrot_training - Step 35666: {'lr': 0.00010218138499581003, 'samples': 18261504, 'steps': 35666, 'loss/train': 1.5816563367843628}
+02/25/2022 23:45:20 - INFO - codeparrot_training - Step 35667: {'lr': 0.00010216818947129155, 'samples': 18262016, 'steps': 35667, 'loss/train': 1.7287938594818115}
+02/25/2022 23:45:24 - INFO - codeparrot_training - Step 35668: {'lr': 0.00010215499458003752, 'samples': 18262528, 'steps': 35668, 'loss/train': 2.4390480518341064}
+02/25/2022 23:45:29 - INFO - codeparrot_training - Step 35669: {'lr': 0.00010214180032210463, 'samples': 18263040, 'steps': 35669, 'loss/train': 1.2931214570999146}
+02/25/2022 23:45:33 - INFO - codeparrot_training - Step 35670: {'lr': 0.0001021286066975493, 'samples': 18263552, 'steps': 35670, 'loss/train': 1.8427566289901733}
+02/25/2022 23:45:39 - INFO - codeparrot_training - Step 35671: {'lr': 0.0001021154137064281, 'samples': 18264064, 'steps': 35671, 'loss/train': 1.9134794473648071}
+02/25/2022 23:45:42 - INFO - codeparrot_training - Step 35672: {'lr': 0.0001021022213487974, 'samples': 18264576, 'steps': 35672, 'loss/train': 1.5787392854690552}
+02/25/2022 23:45:48 - INFO - codeparrot_training - Step 35673: {'lr': 0.00010208902962471391, 'samples': 18265088, 'steps': 35673, 'loss/train': 1.4939777851104736}
+02/25/2022 23:45:51 - INFO - codeparrot_training - Step 35674: {'lr': 0.00010207583853423405, 'samples': 18265600, 'steps': 35674, 'loss/train': 1.661348819732666}
+02/25/2022 23:45:57 - INFO - codeparrot_training - Step 35675: {'lr': 0.00010206264807741434, 'samples': 18266112, 'steps': 35675, 'loss/train': 1.9807336330413818}
+02/25/2022 23:46:00 - INFO - codeparrot_training - Step 35676: {'lr': 0.00010204945825431114, 'samples': 18266624, 'steps': 35676, 'loss/train': 3.0842208862304688}
+02/25/2022 23:46:06 - INFO - codeparrot_training - Step 35677: {'lr': 0.0001020362690649812, 'samples': 18267136, 'steps': 35677, 'loss/train': 1.785537838935852}
+02/25/2022 23:46:10 - INFO - codeparrot_training - Step 35678: {'lr': 0.0001020230805094808, 'samples': 18267648, 'steps': 35678, 'loss/train': 1.5328142642974854}
+02/25/2022 23:46:15 - INFO - codeparrot_training - Step 35679: {'lr': 0.00010200989258786674, 'samples': 18268160, 'steps': 35679, 'loss/train': 1.5087634325027466}
+02/25/2022 23:46:18 - INFO - codeparrot_training - Step 35680: {'lr': 0.00010199670530019511, 'samples': 18268672, 'steps': 35680, 'loss/train': 1.4962003231048584}
+02/25/2022 23:46:25 - INFO - codeparrot_training - Step 35681: {'lr': 0.00010198351864652269, 'samples': 18269184, 'steps': 35681, 'loss/train': 2.130740165710449}
+02/25/2022 23:46:28 - INFO - codeparrot_training - Step 35682: {'lr': 0.00010197033262690577, 'samples': 18269696, 'steps': 35682, 'loss/train': 2.107067108154297}
+02/25/2022 23:46:34 - INFO - codeparrot_training - Step 35683: {'lr': 0.00010195714724140115, 'samples': 18270208, 'steps': 35683, 'loss/train': 1.1449522972106934}
+02/25/2022 23:46:37 - INFO - codeparrot_training - Step 35684: {'lr': 0.00010194396249006491, 'samples': 18270720, 'steps': 35684, 'loss/train': 2.2388737201690674}
+02/25/2022 23:46:43 - INFO - codeparrot_training - Step 35685: {'lr': 0.00010193077837295381, 'samples': 18271232, 'steps': 35685, 'loss/train': 2.305042266845703}
+02/25/2022 23:46:46 - INFO - codeparrot_training - Step 35686: {'lr': 0.00010191759489012415, 'samples': 18271744, 'steps': 35686, 'loss/train': 0.840194582939148}
+02/25/2022 23:46:52 - INFO - codeparrot_training - Step 35687: {'lr': 0.00010190441204163267, 'samples': 18272256, 'steps': 35687, 'loss/train': 1.3436988592147827}
+02/25/2022 23:46:55 - INFO - codeparrot_training - Step 35688: {'lr': 0.00010189122982753549, 'samples': 18272768, 'steps': 35688, 'loss/train': 0.8906335234642029}
+02/25/2022 23:47:01 - INFO - codeparrot_training - Step 35689: {'lr': 0.00010187804824788932, 'samples': 18273280, 'steps': 35689, 'loss/train': 1.6520715951919556}
+02/25/2022 23:47:04 - INFO - codeparrot_training - Step 35690: {'lr': 0.00010186486730275049, 'samples': 18273792, 'steps': 35690, 'loss/train': 1.9635087251663208}
+02/25/2022 23:47:10 - INFO - codeparrot_training - Step 35691: {'lr': 0.00010185168699217561, 'samples': 18274304, 'steps': 35691, 'loss/train': 2.116649866104126}
+02/25/2022 23:47:13 - INFO - codeparrot_training - Step 35692: {'lr': 0.00010183850731622108, 'samples': 18274816, 'steps': 35692, 'loss/train': 0.07796463370323181}
+02/25/2022 23:47:19 - INFO - codeparrot_training - Step 35693: {'lr': 0.00010182532827494332, 'samples': 18275328, 'steps': 35693, 'loss/train': 0.04188638925552368}
+02/25/2022 23:47:22 - INFO - codeparrot_training - Step 35694: {'lr': 0.00010181214986839879, 'samples': 18275840, 'steps': 35694, 'loss/train': 1.5360817909240723}
+02/25/2022 23:47:28 - INFO - codeparrot_training - Step 35695: {'lr': 0.00010179897209664388, 'samples': 18276352, 'steps': 35695, 'loss/train': 2.1865015029907227}
+02/25/2022 23:47:31 - INFO - codeparrot_training - Step 35696: {'lr': 0.0001017857949597352, 'samples': 18276864, 'steps': 35696, 'loss/train': 1.3727922439575195}
+02/25/2022 23:47:38 - INFO - codeparrot_training - Step 35697: {'lr': 0.00010177261845772912, 'samples': 18277376, 'steps': 35697, 'loss/train': 0.8442731499671936}
+02/25/2022 23:47:41 - INFO - codeparrot_training - Step 35698: {'lr': 0.00010175944259068207, 'samples': 18277888, 'steps': 35698, 'loss/train': 2.0390079021453857}
+02/25/2022 23:47:47 - INFO - codeparrot_training - Step 35699: {'lr': 0.0001017462673586504, 'samples': 18278400, 'steps': 35699, 'loss/train': 2.4970924854278564}
+02/25/2022 23:47:50 - INFO - codeparrot_training - Step 35700: {'lr': 0.00010173309276169074, 'samples': 18278912, 'steps': 35700, 'loss/train': 1.4240550994873047}
+02/25/2022 23:47:56 - INFO - codeparrot_training - Step 35701: {'lr': 0.00010171991879985942, 'samples': 18279424, 'steps': 35701, 'loss/train': 0.8915626406669617}
+02/25/2022 23:47:59 - INFO - codeparrot_training - Step 35702: {'lr': 0.00010170674547321291, 'samples': 18279936, 'steps': 35702, 'loss/train': 1.6054563522338867}
+02/25/2022 23:48:05 - INFO - codeparrot_training - Step 35703: {'lr': 0.00010169357278180749, 'samples': 18280448, 'steps': 35703, 'loss/train': 2.487030029296875}
+02/25/2022 23:48:08 - INFO - codeparrot_training - Step 35704: {'lr': 0.00010168040072569984, 'samples': 18280960, 'steps': 35704, 'loss/train': 1.916651725769043}
+02/25/2022 23:48:14 - INFO - codeparrot_training - Step 35705: {'lr': 0.00010166722930494623, 'samples': 18281472, 'steps': 35705, 'loss/train': 1.5042604207992554}
+02/25/2022 23:48:17 - INFO - codeparrot_training - Step 35706: {'lr': 0.0001016540585196031, 'samples': 18281984, 'steps': 35706, 'loss/train': 0.22606606781482697}
+02/25/2022 23:48:23 - INFO - codeparrot_training - Step 35707: {'lr': 0.00010164088836972682, 'samples': 18282496, 'steps': 35707, 'loss/train': 1.6904124021530151}
+02/25/2022 23:48:27 - INFO - codeparrot_training - Step 35708: {'lr': 0.00010162771885537392, 'samples': 18283008, 'steps': 35708, 'loss/train': 1.1760166883468628}
+02/25/2022 23:48:32 - INFO - codeparrot_training - Step 35709: {'lr': 0.00010161454997660081, 'samples': 18283520, 'steps': 35709, 'loss/train': 1.9237961769104004}
+02/25/2022 23:48:36 - INFO - codeparrot_training - Step 35710: {'lr': 0.00010160138173346379, 'samples': 18284032, 'steps': 35710, 'loss/train': 1.5310009717941284}
+02/25/2022 23:48:41 - INFO - codeparrot_training - Step 35711: {'lr': 0.00010158821412601926, 'samples': 18284544, 'steps': 35711, 'loss/train': 1.596812129020691}
+02/25/2022 23:48:45 - INFO - codeparrot_training - Step 35712: {'lr': 0.00010157504715432381, 'samples': 18285056, 'steps': 35712, 'loss/train': 2.7541723251342773}
+02/25/2022 23:48:50 - INFO - codeparrot_training - Step 35713: {'lr': 0.0001015618808184337, 'samples': 18285568, 'steps': 35713, 'loss/train': 2.4386332035064697}
+02/25/2022 23:48:54 - INFO - codeparrot_training - Step 35714: {'lr': 0.00010154871511840535, 'samples': 18286080, 'steps': 35714, 'loss/train': 1.5850285291671753}
+02/25/2022 23:48:59 - INFO - codeparrot_training - Step 35715: {'lr': 0.0001015355500542951, 'samples': 18286592, 'steps': 35715, 'loss/train': 1.33933687210083}
+02/25/2022 23:49:03 - INFO - codeparrot_training - Step 35716: {'lr': 0.00010152238562615951, 'samples': 18287104, 'steps': 35716, 'loss/train': 1.891749382019043}
+02/25/2022 23:49:09 - INFO - codeparrot_training - Step 35717: {'lr': 0.00010150922183405475, 'samples': 18287616, 'steps': 35717, 'loss/train': 1.4227796792984009}
+02/25/2022 23:49:12 - INFO - codeparrot_training - Step 35718: {'lr': 0.00010149605867803755, 'samples': 18288128, 'steps': 35718, 'loss/train': 1.137722373008728}
+02/25/2022 23:49:18 - INFO - codeparrot_training - Step 35719: {'lr': 0.00010148289615816386, 'samples': 18288640, 'steps': 35719, 'loss/train': 0.9836340546607971}
+02/25/2022 23:49:21 - INFO - codeparrot_training - Step 35720: {'lr': 0.00010146973427449039, 'samples': 18289152, 'steps': 35720, 'loss/train': 2.1329128742218018}
+02/25/2022 23:49:27 - INFO - codeparrot_training - Step 35721: {'lr': 0.00010145657302707334, 'samples': 18289664, 'steps': 35721, 'loss/train': 2.545548439025879}
+02/25/2022 23:49:30 - INFO - codeparrot_training - Step 35722: {'lr': 0.00010144341241596933, 'samples': 18290176, 'steps': 35722, 'loss/train': 1.2810003757476807}
+02/25/2022 23:49:36 - INFO - codeparrot_training - Step 35723: {'lr': 0.00010143025244123439, 'samples': 18290688, 'steps': 35723, 'loss/train': 0.42229577898979187}
+02/25/2022 23:49:39 - INFO - codeparrot_training - Step 35724: {'lr': 0.00010141709310292516, 'samples': 18291200, 'steps': 35724, 'loss/train': 1.2862200736999512}
+02/25/2022 23:49:44 - INFO - codeparrot_training - Step 35725: {'lr': 0.00010140393440109783, 'samples': 18291712, 'steps': 35725, 'loss/train': 0.7427252531051636}
+02/25/2022 23:49:48 - INFO - codeparrot_training - Step 35726: {'lr': 0.00010139077633580896, 'samples': 18292224, 'steps': 35726, 'loss/train': 1.8521952629089355}
+02/25/2022 23:49:53 - INFO - codeparrot_training - Step 35727: {'lr': 0.00010137761890711481, 'samples': 18292736, 'steps': 35727, 'loss/train': 1.9859143495559692}
+02/25/2022 23:49:57 - INFO - codeparrot_training - Step 35728: {'lr': 0.00010136446211507175, 'samples': 18293248, 'steps': 35728, 'loss/train': 2.207202672958374}
+02/25/2022 23:50:02 - INFO - codeparrot_training - Step 35729: {'lr': 0.00010135130595973602, 'samples': 18293760, 'steps': 35729, 'loss/train': 2.231048345565796}
+02/25/2022 23:50:06 - INFO - codeparrot_training - Step 35730: {'lr': 0.0001013381504411642, 'samples': 18294272, 'steps': 35730, 'loss/train': 1.4255168437957764}
+02/25/2022 23:50:11 - INFO - codeparrot_training - Step 35731: {'lr': 0.00010132499555941257, 'samples': 18294784, 'steps': 35731, 'loss/train': 3.725872278213501}
+02/25/2022 23:50:15 - INFO - codeparrot_training - Step 35732: {'lr': 0.00010131184131453741, 'samples': 18295296, 'steps': 35732, 'loss/train': 2.2639212608337402}
+02/25/2022 23:50:21 - INFO - codeparrot_training - Step 35733: {'lr': 0.00010129868770659503, 'samples': 18295808, 'steps': 35733, 'loss/train': 2.052513837814331}
+02/25/2022 23:50:24 - INFO - codeparrot_training - Step 35734: {'lr': 0.00010128553473564197, 'samples': 18296320, 'steps': 35734, 'loss/train': 0.46537500619888306}
+02/25/2022 23:50:30 - INFO - codeparrot_training - Step 35735: {'lr': 0.00010127238240173443, 'samples': 18296832, 'steps': 35735, 'loss/train': 2.5970702171325684}
+02/25/2022 23:50:35 - INFO - codeparrot_training - Step 35736: {'lr': 0.0001012592307049288, 'samples': 18297344, 'steps': 35736, 'loss/train': 2.2687795162200928}
+02/25/2022 23:50:39 - INFO - codeparrot_training - Step 35737: {'lr': 0.00010124607964528127, 'samples': 18297856, 'steps': 35737, 'loss/train': 1.0605558156967163}
+02/25/2022 23:50:44 - INFO - codeparrot_training - Step 35738: {'lr': 0.0001012329292228484, 'samples': 18298368, 'steps': 35738, 'loss/train': 2.531243324279785}
+02/25/2022 23:50:48 - INFO - codeparrot_training - Step 35739: {'lr': 0.00010121977943768643, 'samples': 18298880, 'steps': 35739, 'loss/train': 1.646443486213684}
+02/25/2022 23:50:53 - INFO - codeparrot_training - Step 35740: {'lr': 0.00010120663028985169, 'samples': 18299392, 'steps': 35740, 'loss/train': 0.739367663860321}
+02/25/2022 23:50:57 - INFO - codeparrot_training - Step 35741: {'lr': 0.00010119348177940049, 'samples': 18299904, 'steps': 35741, 'loss/train': 1.4047653675079346}
+02/25/2022 23:51:04 - INFO - codeparrot_training - Step 35742: {'lr': 0.00010118033390638906, 'samples': 18300416, 'steps': 35742, 'loss/train': 2.459845781326294}
+02/25/2022 23:51:07 - INFO - codeparrot_training - Step 35743: {'lr': 0.00010116718667087394, 'samples': 18300928, 'steps': 35743, 'loss/train': 2.2868740558624268}
+02/25/2022 23:51:11 - INFO - codeparrot_training - Step 35744: {'lr': 0.00010115404007291131, 'samples': 18301440, 'steps': 35744, 'loss/train': 1.2523095607757568}
+02/25/2022 23:51:16 - INFO - codeparrot_training - Step 35745: {'lr': 0.00010114089411255754, 'samples': 18301952, 'steps': 35745, 'loss/train': 1.5059741735458374}
+02/25/2022 23:51:20 - INFO - codeparrot_training - Step 35746: {'lr': 0.00010112774878986877, 'samples': 18302464, 'steps': 35746, 'loss/train': 1.524960994720459}
+02/25/2022 23:51:25 - INFO - codeparrot_training - Step 35747: {'lr': 0.00010111460410490158, 'samples': 18302976, 'steps': 35747, 'loss/train': 0.3220663368701935}
+02/25/2022 23:51:29 - INFO - codeparrot_training - Step 35748: {'lr': 0.00010110146005771215, 'samples': 18303488, 'steps': 35748, 'loss/train': 1.0911425352096558}
+02/25/2022 23:51:34 - INFO - codeparrot_training - Step 35749: {'lr': 0.00010108831664835676, 'samples': 18304000, 'steps': 35749, 'loss/train': 0.6062576770782471}
+02/25/2022 23:51:38 - INFO - codeparrot_training - Step 35750: {'lr': 0.00010107517387689166, 'samples': 18304512, 'steps': 35750, 'loss/train': 0.12423809617757797}
+02/25/2022 23:51:43 - INFO - codeparrot_training - Step 35751: {'lr': 0.00010106203174337334, 'samples': 18305024, 'steps': 35751, 'loss/train': 1.6128947734832764}
+02/25/2022 23:51:47 - INFO - codeparrot_training - Step 35752: {'lr': 0.00010104889024785785, 'samples': 18305536, 'steps': 35752, 'loss/train': 3.344428777694702}
+02/25/2022 23:51:53 - INFO - codeparrot_training - Step 35753: {'lr': 0.00010103574939040184, 'samples': 18306048, 'steps': 35753, 'loss/train': 1.7074198722839355}
+02/25/2022 23:51:56 - INFO - codeparrot_training - Step 35754: {'lr': 0.00010102260917106115, 'samples': 18306560, 'steps': 35754, 'loss/train': 2.4637317657470703}
+02/25/2022 23:52:02 - INFO - codeparrot_training - Step 35755: {'lr': 0.00010100946958989244, 'samples': 18307072, 'steps': 35755, 'loss/train': 1.056294322013855}
+02/25/2022 23:52:05 - INFO - codeparrot_training - Step 35756: {'lr': 0.0001009963306469517, 'samples': 18307584, 'steps': 35756, 'loss/train': 0.06776978075504303}
+02/25/2022 23:52:11 - INFO - codeparrot_training - Step 35757: {'lr': 0.0001009831923422956, 'samples': 18308096, 'steps': 35757, 'loss/train': 1.934425950050354}
+02/25/2022 23:52:14 - INFO - codeparrot_training - Step 35758: {'lr': 0.00010097005467598, 'samples': 18308608, 'steps': 35758, 'loss/train': 1.629753828048706}
+02/25/2022 23:52:20 - INFO - codeparrot_training - Step 35759: {'lr': 0.00010095691764806142, 'samples': 18309120, 'steps': 35759, 'loss/train': 1.4768569469451904}
+02/25/2022 23:52:23 - INFO - codeparrot_training - Step 35760: {'lr': 0.00010094378125859602, 'samples': 18309632, 'steps': 35760, 'loss/train': 1.0320724248886108}
+02/25/2022 23:52:29 - INFO - codeparrot_training - Step 35761: {'lr': 0.00010093064550764031, 'samples': 18310144, 'steps': 35761, 'loss/train': 1.607743501663208}
+02/25/2022 23:52:32 - INFO - codeparrot_training - Step 35762: {'lr': 0.00010091751039525019, 'samples': 18310656, 'steps': 35762, 'loss/train': 1.2424111366271973}
+02/25/2022 23:52:38 - INFO - codeparrot_training - Step 35763: {'lr': 0.00010090437592148218, 'samples': 18311168, 'steps': 35763, 'loss/train': 2.506679058074951}
+02/25/2022 23:52:41 - INFO - codeparrot_training - Step 35764: {'lr': 0.00010089124208639242, 'samples': 18311680, 'steps': 35764, 'loss/train': 1.5359445810317993}
+02/25/2022 23:52:48 - INFO - codeparrot_training - Step 35765: {'lr': 0.00010087810889003732, 'samples': 18312192, 'steps': 35765, 'loss/train': 1.8538669347763062}
+02/25/2022 23:52:51 - INFO - codeparrot_training - Step 35766: {'lr': 0.00010086497633247304, 'samples': 18312704, 'steps': 35766, 'loss/train': 1.8985295295715332}
+02/25/2022 23:52:57 - INFO - codeparrot_training - Step 35767: {'lr': 0.00010085184441375586, 'samples': 18313216, 'steps': 35767, 'loss/train': 1.0780740976333618}
+02/25/2022 23:53:00 - INFO - codeparrot_training - Step 35768: {'lr': 0.00010083871313394191, 'samples': 18313728, 'steps': 35768, 'loss/train': 1.8520212173461914}
+02/25/2022 23:53:06 - INFO - codeparrot_training - Step 35769: {'lr': 0.00010082558249308766, 'samples': 18314240, 'steps': 35769, 'loss/train': 1.1098905801773071}
+02/25/2022 23:53:09 - INFO - codeparrot_training - Step 35770: {'lr': 0.00010081245249124923, 'samples': 18314752, 'steps': 35770, 'loss/train': 1.4739210605621338}
+02/25/2022 23:53:15 - INFO - codeparrot_training - Step 35771: {'lr': 0.0001007993231284829, 'samples': 18315264, 'steps': 35771, 'loss/train': 1.1698055267333984}
+02/25/2022 23:53:18 - INFO - codeparrot_training - Step 35772: {'lr': 0.0001007861944048448, 'samples': 18315776, 'steps': 35772, 'loss/train': 2.637181520462036}
+02/25/2022 23:53:24 - INFO - codeparrot_training - Step 35773: {'lr': 0.00010077306632039132, 'samples': 18316288, 'steps': 35773, 'loss/train': 2.039930820465088}
+02/25/2022 23:53:27 - INFO - codeparrot_training - Step 35774: {'lr': 0.00010075993887517868, 'samples': 18316800, 'steps': 35774, 'loss/train': 1.1873440742492676}
+02/25/2022 23:53:33 - INFO - codeparrot_training - Step 35775: {'lr': 0.00010074681206926306, 'samples': 18317312, 'steps': 35775, 'loss/train': 2.5722475051879883}
+02/25/2022 23:53:36 - INFO - codeparrot_training - Step 35776: {'lr': 0.0001007336859027006, 'samples': 18317824, 'steps': 35776, 'loss/train': 2.2421562671661377}
+02/25/2022 23:53:42 - INFO - codeparrot_training - Step 35777: {'lr': 0.00010072056037554772, 'samples': 18318336, 'steps': 35777, 'loss/train': 1.755372166633606}
+02/25/2022 23:53:45 - INFO - codeparrot_training - Step 35778: {'lr': 0.00010070743548786057, 'samples': 18318848, 'steps': 35778, 'loss/train': 0.9578810334205627}
+02/25/2022 23:53:52 - INFO - codeparrot_training - Step 35779: {'lr': 0.00010069431123969536, 'samples': 18319360, 'steps': 35779, 'loss/train': 2.17842698097229}
+02/25/2022 23:53:55 - INFO - codeparrot_training - Step 35780: {'lr': 0.00010068118763110824, 'samples': 18319872, 'steps': 35780, 'loss/train': 3.354618549346924}
+02/25/2022 23:54:01 - INFO - codeparrot_training - Step 35781: {'lr': 0.00010066806466215556, 'samples': 18320384, 'steps': 35781, 'loss/train': 1.9873430728912354}
+02/25/2022 23:54:04 - INFO - codeparrot_training - Step 35782: {'lr': 0.0001006549423328935, 'samples': 18320896, 'steps': 35782, 'loss/train': 0.951806902885437}
+02/25/2022 23:54:10 - INFO - codeparrot_training - Step 35783: {'lr': 0.0001006418206433782, 'samples': 18321408, 'steps': 35783, 'loss/train': 2.1334714889526367}
+02/25/2022 23:54:13 - INFO - codeparrot_training - Step 35784: {'lr': 0.00010062869959366586, 'samples': 18321920, 'steps': 35784, 'loss/train': 0.36228030920028687}
+02/25/2022 23:54:19 - INFO - codeparrot_training - Step 35785: {'lr': 0.0001006155791838128, 'samples': 18322432, 'steps': 35785, 'loss/train': 1.8500031232833862}
+02/25/2022 23:54:22 - INFO - codeparrot_training - Step 35786: {'lr': 0.0001006024594138752, 'samples': 18322944, 'steps': 35786, 'loss/train': 2.4462132453918457}
+02/25/2022 23:54:28 - INFO - codeparrot_training - Step 35787: {'lr': 0.00010058934028390912, 'samples': 18323456, 'steps': 35787, 'loss/train': 2.0870611667633057}
+02/25/2022 23:54:31 - INFO - codeparrot_training - Step 35788: {'lr': 0.00010057622179397103, 'samples': 18323968, 'steps': 35788, 'loss/train': 1.8753844499588013}
+02/25/2022 23:54:37 - INFO - codeparrot_training - Step 35789: {'lr': 0.0001005631039441168, 'samples': 18324480, 'steps': 35789, 'loss/train': 1.2309633493423462}
+02/25/2022 23:54:43 - INFO - codeparrot_training - Step 35790: {'lr': 0.00010054998673440286, 'samples': 18324992, 'steps': 35790, 'loss/train': 1.9031633138656616}
+02/25/2022 23:54:47 - INFO - codeparrot_training - Step 35791: {'lr': 0.00010053687016488522, 'samples': 18325504, 'steps': 35791, 'loss/train': 1.7304120063781738}
+02/25/2022 23:54:52 - INFO - codeparrot_training - Step 35792: {'lr': 0.00010052375423562038, 'samples': 18326016, 'steps': 35792, 'loss/train': 1.9291925430297852}
+02/25/2022 23:54:56 - INFO - codeparrot_training - Step 35793: {'lr': 0.00010051063894666409, 'samples': 18326528, 'steps': 35793, 'loss/train': 3.257054090499878}
+02/25/2022 23:55:01 - INFO - codeparrot_training - Step 35794: {'lr': 0.00010049752429807288, 'samples': 18327040, 'steps': 35794, 'loss/train': 1.846071481704712}
+02/25/2022 23:55:05 - INFO - codeparrot_training - Step 35795: {'lr': 0.00010048441028990272, 'samples': 18327552, 'steps': 35795, 'loss/train': 1.2694448232650757}
+02/25/2022 23:55:10 - INFO - codeparrot_training - Step 35796: {'lr': 0.00010047129692221003, 'samples': 18328064, 'steps': 35796, 'loss/train': 2.1736929416656494}
+02/25/2022 23:55:14 - INFO - codeparrot_training - Step 35797: {'lr': 0.00010045818419505062, 'samples': 18328576, 'steps': 35797, 'loss/train': 1.857295274734497}
+02/25/2022 23:55:19 - INFO - codeparrot_training - Step 35798: {'lr': 0.00010044507210848097, 'samples': 18329088, 'steps': 35798, 'loss/train': 2.3312344551086426}
+02/25/2022 23:55:23 - INFO - codeparrot_training - Step 35799: {'lr': 0.00010043196066255709, 'samples': 18329600, 'steps': 35799, 'loss/train': 1.7168599367141724}
+02/25/2022 23:55:29 - INFO - codeparrot_training - Step 35800: {'lr': 0.00010041884985733524, 'samples': 18330112, 'steps': 35800, 'loss/train': 1.7620187997817993}
+02/25/2022 23:55:33 - INFO - codeparrot_training - Step 35801: {'lr': 0.00010040573969287155, 'samples': 18330624, 'steps': 35801, 'loss/train': 1.7488982677459717}
+02/25/2022 23:55:38 - INFO - codeparrot_training - Step 35802: {'lr': 0.00010039263016922218, 'samples': 18331136, 'steps': 35802, 'loss/train': 2.3864548206329346}
+02/25/2022 23:55:41 - INFO - codeparrot_training - Step 35803: {'lr': 0.0001003795212864432, 'samples': 18331648, 'steps': 35803, 'loss/train': 1.5186033248901367}
+02/25/2022 23:55:47 - INFO - codeparrot_training - Step 35804: {'lr': 0.0001003664130445909, 'samples': 18332160, 'steps': 35804, 'loss/train': 0.7373254299163818}
+02/25/2022 23:55:50 - INFO - codeparrot_training - Step 35805: {'lr': 0.0001003533054437214, 'samples': 18332672, 'steps': 35805, 'loss/train': 2.089855670928955}
+02/25/2022 23:55:56 - INFO - codeparrot_training - Step 35806: {'lr': 0.00010034019848389081, 'samples': 18333184, 'steps': 35806, 'loss/train': 1.377484917640686}
+02/25/2022 23:55:59 - INFO - codeparrot_training - Step 35807: {'lr': 0.00010032709216515518, 'samples': 18333696, 'steps': 35807, 'loss/train': 1.7374728918075562}
+02/25/2022 23:56:05 - INFO - codeparrot_training - Step 35808: {'lr': 0.00010031398648757087, 'samples': 18334208, 'steps': 35808, 'loss/train': 2.625351667404175}
+02/25/2022 23:56:08 - INFO - codeparrot_training - Step 35809: {'lr': 0.00010030088145119392, 'samples': 18334720, 'steps': 35809, 'loss/train': 2.052825689315796}
+02/25/2022 23:56:15 - INFO - codeparrot_training - Step 35810: {'lr': 0.00010028777705608045, 'samples': 18335232, 'steps': 35810, 'loss/train': 1.3945361375808716}
+02/25/2022 23:56:19 - INFO - codeparrot_training - Step 35811: {'lr': 0.0001002746733022865, 'samples': 18335744, 'steps': 35811, 'loss/train': 0.5266461372375488}
+02/25/2022 23:56:24 - INFO - codeparrot_training - Step 35812: {'lr': 0.00010026157018986839, 'samples': 18336256, 'steps': 35812, 'loss/train': 1.0185898542404175}
+02/25/2022 23:56:27 - INFO - codeparrot_training - Step 35813: {'lr': 0.0001002484677188821, 'samples': 18336768, 'steps': 35813, 'loss/train': 1.2480287551879883}
+02/25/2022 23:56:33 - INFO - codeparrot_training - Step 35814: {'lr': 0.00010023536588938401, 'samples': 18337280, 'steps': 35814, 'loss/train': 1.4092193841934204}
+02/25/2022 23:56:36 - INFO - codeparrot_training - Step 35815: {'lr': 0.00010022226470142986, 'samples': 18337792, 'steps': 35815, 'loss/train': 2.3569858074188232}
+02/25/2022 23:56:42 - INFO - codeparrot_training - Step 35816: {'lr': 0.00010020916415507605, 'samples': 18338304, 'steps': 35816, 'loss/train': 2.0819053649902344}
+02/25/2022 23:56:45 - INFO - codeparrot_training - Step 35817: {'lr': 0.00010019606425037855, 'samples': 18338816, 'steps': 35817, 'loss/train': 1.334726333618164}
+02/25/2022 23:56:51 - INFO - codeparrot_training - Step 35818: {'lr': 0.00010018296498739371, 'samples': 18339328, 'steps': 35818, 'loss/train': 1.0104502439498901}
+02/25/2022 23:56:54 - INFO - codeparrot_training - Step 35819: {'lr': 0.00010016986636617731, 'samples': 18339840, 'steps': 35819, 'loss/train': 0.9667519330978394}
+02/25/2022 23:57:00 - INFO - codeparrot_training - Step 35820: {'lr': 0.0001001567683867857, 'samples': 18340352, 'steps': 35820, 'loss/train': 2.3153321743011475}
+02/25/2022 23:57:04 - INFO - codeparrot_training - Step 35821: {'lr': 0.00010014367104927491, 'samples': 18340864, 'steps': 35821, 'loss/train': 1.448493480682373}
+02/25/2022 23:57:09 - INFO - codeparrot_training - Step 35822: {'lr': 0.00010013057435370104, 'samples': 18341376, 'steps': 35822, 'loss/train': 1.8445961475372314}
+02/25/2022 23:57:13 - INFO - codeparrot_training - Step 35823: {'lr': 0.00010011747830012014, 'samples': 18341888, 'steps': 35823, 'loss/train': 0.7104392647743225}
+02/25/2022 23:57:18 - INFO - codeparrot_training - Step 35824: {'lr': 0.00010010438288858845, 'samples': 18342400, 'steps': 35824, 'loss/train': 1.7712794542312622}
+02/25/2022 23:57:21 - INFO - codeparrot_training - Step 35825: {'lr': 0.00010009128811916197, 'samples': 18342912, 'steps': 35825, 'loss/train': 1.2480518817901611}
+02/25/2022 23:57:28 - INFO - codeparrot_training - Step 35826: {'lr': 0.00010007819399189674, 'samples': 18343424, 'steps': 35826, 'loss/train': 1.5449738502502441}
+02/25/2022 23:57:32 - INFO - codeparrot_training - Step 35827: {'lr': 0.000100065100506849, 'samples': 18343936, 'steps': 35827, 'loss/train': 1.6615954637527466}
+02/25/2022 23:57:37 - INFO - codeparrot_training - Step 35828: {'lr': 0.00010005200766407476, 'samples': 18344448, 'steps': 35828, 'loss/train': 2.4104394912719727}
+02/25/2022 23:57:41 - INFO - codeparrot_training - Step 35829: {'lr': 0.00010003891546363012, 'samples': 18344960, 'steps': 35829, 'loss/train': 1.866317868232727}
+02/25/2022 23:57:46 - INFO - codeparrot_training - Step 35830: {'lr': 0.00010002582390557105, 'samples': 18345472, 'steps': 35830, 'loss/train': 1.8755161762237549}
+02/25/2022 23:57:50 - INFO - codeparrot_training - Step 35831: {'lr': 0.00010001273298995381, 'samples': 18345984, 'steps': 35831, 'loss/train': 0.6883083581924438}
+02/25/2022 23:57:55 - INFO - codeparrot_training - Step 35832: {'lr': 9.99996427168344e-05, 'samples': 18346496, 'steps': 35832, 'loss/train': 1.780527949333191}
+02/25/2022 23:57:59 - INFO - codeparrot_training - Step 35833: {'lr': 9.998655308626888e-05, 'samples': 18347008, 'steps': 35833, 'loss/train': 2.493551254272461}
+02/25/2022 23:58:04 - INFO - codeparrot_training - Step 35834: {'lr': 9.997346409831326e-05, 'samples': 18347520, 'steps': 35834, 'loss/train': 1.2785578966140747}
+02/25/2022 23:58:08 - INFO - codeparrot_training - Step 35835: {'lr': 9.996037575302375e-05, 'samples': 18348032, 'steps': 35835, 'loss/train': 3.2638285160064697}
+02/25/2022 23:58:14 - INFO - codeparrot_training - Step 35836: {'lr': 9.994728805045635e-05, 'samples': 18348544, 'steps': 35836, 'loss/train': 1.2130745649337769}
+02/25/2022 23:58:18 - INFO - codeparrot_training - Step 35837: {'lr': 9.993420099066717e-05, 'samples': 18349056, 'steps': 35837, 'loss/train': 2.0023648738861084}
+02/25/2022 23:58:23 - INFO - codeparrot_training - Step 35838: {'lr': 9.992111457371208e-05, 'samples': 18349568, 'steps': 35838, 'loss/train': 2.543809413909912}
+02/25/2022 23:58:29 - INFO - codeparrot_training - Step 35839: {'lr': 9.990802879964739e-05, 'samples': 18350080, 'steps': 35839, 'loss/train': 2.232409954071045}
+02/25/2022 23:58:32 - INFO - codeparrot_training - Step 35840: {'lr': 9.989494366852902e-05, 'samples': 18350592, 'steps': 35840, 'loss/train': 3.030867338180542}
+02/25/2022 23:58:36 - INFO - codeparrot_training - Step 35841: {'lr': 9.988185918041307e-05, 'samples': 18351104, 'steps': 35841, 'loss/train': 0.17709724605083466}
+02/25/2022 23:58:41 - INFO - codeparrot_training - Step 35842: {'lr': 9.986877533535546e-05, 'samples': 18351616, 'steps': 35842, 'loss/train': 0.7245816588401794}
+02/25/2022 23:58:48 - INFO - codeparrot_training - Step 35843: {'lr': 9.985569213341247e-05, 'samples': 18352128, 'steps': 35843, 'loss/train': 1.4777520895004272}
+02/25/2022 23:58:51 - INFO - codeparrot_training - Step 35844: {'lr': 9.984260957463998e-05, 'samples': 18352640, 'steps': 35844, 'loss/train': 0.8061791062355042}
+02/25/2022 23:58:54 - INFO - codeparrot_training - Step 35845: {'lr': 9.98295276590941e-05, 'samples': 18353152, 'steps': 35845, 'loss/train': 1.0470699071884155}
+02/25/2022 23:59:00 - INFO - codeparrot_training - Step 35846: {'lr': 9.981644638683075e-05, 'samples': 18353664, 'steps': 35846, 'loss/train': 0.7400323152542114}
+02/25/2022 23:59:06 - INFO - codeparrot_training - Step 35847: {'lr': 9.98033657579061e-05, 'samples': 18354176, 'steps': 35847, 'loss/train': 1.6745328903198242}
+02/25/2022 23:59:09 - INFO - codeparrot_training - Step 35848: {'lr': 9.979028577237609e-05, 'samples': 18354688, 'steps': 35848, 'loss/train': 0.5688425302505493}
+02/25/2022 23:59:15 - INFO - codeparrot_training - Step 35849: {'lr': 9.977720643029697e-05, 'samples': 18355200, 'steps': 35849, 'loss/train': 1.9610005617141724}
+02/25/2022 23:59:18 - INFO - codeparrot_training - Step 35850: {'lr': 9.976412773172441e-05, 'samples': 18355712, 'steps': 35850, 'loss/train': 1.489272952079773}
+02/25/2022 23:59:24 - INFO - codeparrot_training - Step 35851: {'lr': 9.975104967671473e-05, 'samples': 18356224, 'steps': 35851, 'loss/train': 0.8180992603302002}
+02/25/2022 23:59:27 - INFO - codeparrot_training - Step 35852: {'lr': 9.973797226532372e-05, 'samples': 18356736, 'steps': 35852, 'loss/train': 0.04300791025161743}
+02/25/2022 23:59:31 - INFO - codeparrot_training - Step 35853: {'lr': 9.972489549760772e-05, 'samples': 18357248, 'steps': 35853, 'loss/train': 0.9210607409477234}
+02/25/2022 23:59:36 - INFO - codeparrot_training - Step 35854: {'lr': 9.971181937362239e-05, 'samples': 18357760, 'steps': 35854, 'loss/train': 2.4391281604766846}
+02/25/2022 23:59:42 - INFO - codeparrot_training - Step 35855: {'lr': 9.969874389342396e-05, 'samples': 18358272, 'steps': 35855, 'loss/train': 1.3115605115890503}
+02/25/2022 23:59:46 - INFO - codeparrot_training - Step 35856: {'lr': 9.968566905706832e-05, 'samples': 18358784, 'steps': 35856, 'loss/train': 2.1106553077697754}
+02/25/2022 23:59:51 - INFO - codeparrot_training - Step 35857: {'lr': 9.967259486461172e-05, 'samples': 18359296, 'steps': 35857, 'loss/train': 3.029512882232666}
+02/25/2022 23:59:55 - INFO - codeparrot_training - Step 35858: {'lr': 9.96595213161098e-05, 'samples': 18359808, 'steps': 35858, 'loss/train': 1.8139848709106445}
+02/26/2022 00:00:00 - INFO - codeparrot_training - Step 35859: {'lr': 9.964644841161885e-05, 'samples': 18360320, 'steps': 35859, 'loss/train': 2.35768461227417}
+02/26/2022 00:00:04 - INFO - codeparrot_training - Step 35860: {'lr': 9.963337615119466e-05, 'samples': 18360832, 'steps': 35860, 'loss/train': 1.69540274143219}
+02/26/2022 00:00:09 - INFO - codeparrot_training - Step 35861: {'lr': 9.962030453489346e-05, 'samples': 18361344, 'steps': 35861, 'loss/train': 1.393372654914856}
+02/26/2022 00:00:13 - INFO - codeparrot_training - Step 35862: {'lr': 9.960723356277113e-05, 'samples': 18361856, 'steps': 35862, 'loss/train': 2.2338507175445557}
+02/26/2022 00:00:18 - INFO - codeparrot_training - Step 35863: {'lr': 9.959416323488363e-05, 'samples': 18362368, 'steps': 35863, 'loss/train': 0.9011073708534241}
+02/26/2022 00:00:22 - INFO - codeparrot_training - Step 35864: {'lr': 9.958109355128688e-05, 'samples': 18362880, 'steps': 35864, 'loss/train': 4.287140846252441}
+02/26/2022 00:00:25 - INFO - codeparrot_training - Step 35865: {'lr': 9.956802451203709e-05, 'samples': 18363392, 'steps': 35865, 'loss/train': 1.5284981727600098}
+02/26/2022 00:00:31 - INFO - codeparrot_training - Step 35866: {'lr': 9.955495611719009e-05, 'samples': 18363904, 'steps': 35866, 'loss/train': 3.0469791889190674}
+02/26/2022 00:00:34 - INFO - codeparrot_training - Step 35867: {'lr': 9.954188836680189e-05, 'samples': 18364416, 'steps': 35867, 'loss/train': 2.0611648559570312}
+02/26/2022 00:00:40 - INFO - codeparrot_training - Step 35868: {'lr': 9.952882126092835e-05, 'samples': 18364928, 'steps': 35868, 'loss/train': 2.296166181564331}
+02/26/2022 00:00:44 - INFO - codeparrot_training - Step 35869: {'lr': 9.951575479962569e-05, 'samples': 18365440, 'steps': 35869, 'loss/train': 2.1618359088897705}
+02/26/2022 00:00:49 - INFO - codeparrot_training - Step 35870: {'lr': 9.950268898294972e-05, 'samples': 18365952, 'steps': 35870, 'loss/train': 1.6320720911026}
+02/26/2022 00:00:53 - INFO - codeparrot_training - Step 35871: {'lr': 9.948962381095647e-05, 'samples': 18366464, 'steps': 35871, 'loss/train': 1.590567946434021}
+02/26/2022 00:01:00 - INFO - codeparrot_training - Step 35872: {'lr': 9.947655928370186e-05, 'samples': 18366976, 'steps': 35872, 'loss/train': 0.12426839768886566}
+02/26/2022 00:01:03 - INFO - codeparrot_training - Step 35873: {'lr': 9.94634954012418e-05, 'samples': 18367488, 'steps': 35873, 'loss/train': 0.4463624954223633}
+02/26/2022 00:01:09 - INFO - codeparrot_training - Step 35874: {'lr': 9.945043216363239e-05, 'samples': 18368000, 'steps': 35874, 'loss/train': 1.449934959411621}
+02/26/2022 00:01:12 - INFO - codeparrot_training - Step 35875: {'lr': 9.943736957092955e-05, 'samples': 18368512, 'steps': 35875, 'loss/train': 1.1564255952835083}
+02/26/2022 00:01:18 - INFO - codeparrot_training - Step 35876: {'lr': 9.942430762318919e-05, 'samples': 18369024, 'steps': 35876, 'loss/train': 1.4119863510131836}
+02/26/2022 00:01:21 - INFO - codeparrot_training - Step 35877: {'lr': 9.941124632046722e-05, 'samples': 18369536, 'steps': 35877, 'loss/train': 1.3367893695831299}
+02/26/2022 00:01:27 - INFO - codeparrot_training - Step 35878: {'lr': 9.939818566281971e-05, 'samples': 18370048, 'steps': 35878, 'loss/train': 1.0903595685958862}
+02/26/2022 00:01:30 - INFO - codeparrot_training - Step 35879: {'lr': 9.938512565030258e-05, 'samples': 18370560, 'steps': 35879, 'loss/train': 2.6931049823760986}
+02/26/2022 00:01:36 - INFO - codeparrot_training - Step 35880: {'lr': 9.937206628297172e-05, 'samples': 18371072, 'steps': 35880, 'loss/train': 1.529502272605896}
+02/26/2022 00:01:39 - INFO - codeparrot_training - Step 35881: {'lr': 9.935900756088304e-05, 'samples': 18371584, 'steps': 35881, 'loss/train': 0.259307861328125}
+02/26/2022 00:01:46 - INFO - codeparrot_training - Step 35882: {'lr': 9.934594948409259e-05, 'samples': 18372096, 'steps': 35882, 'loss/train': 1.7002257108688354}
+02/26/2022 00:01:50 - INFO - codeparrot_training - Step 35883: {'lr': 9.933289205265627e-05, 'samples': 18372608, 'steps': 35883, 'loss/train': 0.8881520628929138}
+02/26/2022 00:01:55 - INFO - codeparrot_training - Step 35884: {'lr': 9.931983526662999e-05, 'samples': 18373120, 'steps': 35884, 'loss/train': 0.8898929357528687}
+02/26/2022 00:01:58 - INFO - codeparrot_training - Step 35885: {'lr': 9.930677912606961e-05, 'samples': 18373632, 'steps': 35885, 'loss/train': 1.0810331106185913}
+02/26/2022 00:02:04 - INFO - codeparrot_training - Step 35886: {'lr': 9.929372363103124e-05, 'samples': 18374144, 'steps': 35886, 'loss/train': 2.11832594871521}
+02/26/2022 00:02:07 - INFO - codeparrot_training - Step 35887: {'lr': 9.928066878157061e-05, 'samples': 18374656, 'steps': 35887, 'loss/train': 0.5292294025421143}
+02/26/2022 00:02:13 - INFO - codeparrot_training - Step 35888: {'lr': 9.926761457774389e-05, 'samples': 18375168, 'steps': 35888, 'loss/train': 0.9705906510353088}
+02/26/2022 00:02:16 - INFO - codeparrot_training - Step 35889: {'lr': 9.92545610196067e-05, 'samples': 18375680, 'steps': 35889, 'loss/train': 2.221918821334839}
+02/26/2022 00:02:23 - INFO - codeparrot_training - Step 35890: {'lr': 9.924150810721516e-05, 'samples': 18376192, 'steps': 35890, 'loss/train': 2.068521499633789}
+02/26/2022 00:02:26 - INFO - codeparrot_training - Step 35891: {'lr': 9.922845584062506e-05, 'samples': 18376704, 'steps': 35891, 'loss/train': 0.6108127236366272}
+02/26/2022 00:02:32 - INFO - codeparrot_training - Step 35892: {'lr': 9.921540421989256e-05, 'samples': 18377216, 'steps': 35892, 'loss/train': 3.333068609237671}
+02/26/2022 00:02:36 - INFO - codeparrot_training - Step 35893: {'lr': 9.920235324507321e-05, 'samples': 18377728, 'steps': 35893, 'loss/train': 1.1117558479309082}
+02/26/2022 00:02:41 - INFO - codeparrot_training - Step 35894: {'lr': 9.918930291622317e-05, 'samples': 18378240, 'steps': 35894, 'loss/train': 1.6782842874526978}
+02/26/2022 00:02:45 - INFO - codeparrot_training - Step 35895: {'lr': 9.917625323339819e-05, 'samples': 18378752, 'steps': 35895, 'loss/train': 2.5834484100341797}
+02/26/2022 00:02:50 - INFO - codeparrot_training - Step 35896: {'lr': 9.916320419665434e-05, 'samples': 18379264, 'steps': 35896, 'loss/train': 2.003950834274292}
+02/26/2022 00:02:53 - INFO - codeparrot_training - Step 35897: {'lr': 9.915015580604741e-05, 'samples': 18379776, 'steps': 35897, 'loss/train': 1.8916932344436646}
+02/26/2022 00:02:59 - INFO - codeparrot_training - Step 35898: {'lr': 9.913710806163334e-05, 'samples': 18380288, 'steps': 35898, 'loss/train': 2.0987513065338135}
+02/26/2022 00:03:03 - INFO - codeparrot_training - Step 35899: {'lr': 9.912406096346787e-05, 'samples': 18380800, 'steps': 35899, 'loss/train': 1.5130975246429443}
+02/26/2022 00:03:08 - INFO - codeparrot_training - Step 35900: {'lr': 9.911101451160715e-05, 'samples': 18381312, 'steps': 35900, 'loss/train': 0.556807816028595}
+02/26/2022 00:03:11 - INFO - codeparrot_training - Step 35901: {'lr': 9.909796870610691e-05, 'samples': 18381824, 'steps': 35901, 'loss/train': 0.40556639432907104}
+02/26/2022 00:03:18 - INFO - codeparrot_training - Step 35902: {'lr': 9.908492354702303e-05, 'samples': 18382336, 'steps': 35902, 'loss/train': 1.4766087532043457}
+02/26/2022 00:03:21 - INFO - codeparrot_training - Step 35903: {'lr': 9.907187903441137e-05, 'samples': 18382848, 'steps': 35903, 'loss/train': 0.6808434128761292}
+02/26/2022 00:03:27 - INFO - codeparrot_training - Step 35904: {'lr': 9.905883516832793e-05, 'samples': 18383360, 'steps': 35904, 'loss/train': 2.0826196670532227}
+02/26/2022 00:03:30 - INFO - codeparrot_training - Step 35905: {'lr': 9.904579194882848e-05, 'samples': 18383872, 'steps': 35905, 'loss/train': 2.017265558242798}
+02/26/2022 00:03:36 - INFO - codeparrot_training - Step 35906: {'lr': 9.903274937596896e-05, 'samples': 18384384, 'steps': 35906, 'loss/train': 0.5058174729347229}
+02/26/2022 00:03:39 - INFO - codeparrot_training - Step 35907: {'lr': 9.90197074498051e-05, 'samples': 18384896, 'steps': 35907, 'loss/train': 1.1410186290740967}
+02/26/2022 00:03:45 - INFO - codeparrot_training - Step 35908: {'lr': 9.900666617039294e-05, 'samples': 18385408, 'steps': 35908, 'loss/train': 1.5514171123504639}
+02/26/2022 00:03:48 - INFO - codeparrot_training - Step 35909: {'lr': 9.899362553778832e-05, 'samples': 18385920, 'steps': 35909, 'loss/train': 0.8423264026641846}
+02/26/2022 00:03:53 - INFO - codeparrot_training - Step 35910: {'lr': 9.898058555204703e-05, 'samples': 18386432, 'steps': 35910, 'loss/train': 2.1960935592651367}
+02/26/2022 00:03:57 - INFO - codeparrot_training - Step 35911: {'lr': 9.896754621322487e-05, 'samples': 18386944, 'steps': 35911, 'loss/train': 2.146303176879883}
+02/26/2022 00:04:03 - INFO - codeparrot_training - Step 35912: {'lr': 9.895450752137788e-05, 'samples': 18387456, 'steps': 35912, 'loss/train': 1.890219807624817}
+02/26/2022 00:04:06 - INFO - codeparrot_training - Step 35913: {'lr': 9.894146947656182e-05, 'samples': 18387968, 'steps': 35913, 'loss/train': 2.2780332565307617}
+02/26/2022 00:04:12 - INFO - codeparrot_training - Step 35914: {'lr': 9.892843207883254e-05, 'samples': 18388480, 'steps': 35914, 'loss/train': 1.6994646787643433}
+02/26/2022 00:04:15 - INFO - codeparrot_training - Step 35915: {'lr': 9.891539532824582e-05, 'samples': 18388992, 'steps': 35915, 'loss/train': 1.5538941621780396}
+02/26/2022 00:04:20 - INFO - codeparrot_training - Step 35916: {'lr': 9.89023592248576e-05, 'samples': 18389504, 'steps': 35916, 'loss/train': 1.7740012407302856}
+02/26/2022 00:04:27 - INFO - codeparrot_training - Step 35917: {'lr': 9.888932376872373e-05, 'samples': 18390016, 'steps': 35917, 'loss/train': 1.4152382612228394}
+02/26/2022 00:04:31 - INFO - codeparrot_training - Step 35918: {'lr': 9.887628895990003e-05, 'samples': 18390528, 'steps': 35918, 'loss/train': 2.090456485748291}
+02/26/2022 00:04:34 - INFO - codeparrot_training - Step 35919: {'lr': 9.886325479844229e-05, 'samples': 18391040, 'steps': 35919, 'loss/train': 0.2509082555770874}
+02/26/2022 00:04:40 - INFO - codeparrot_training - Step 35920: {'lr': 9.885022128440629e-05, 'samples': 18391552, 'steps': 35920, 'loss/train': 1.359215497970581}
+02/26/2022 00:04:43 - INFO - codeparrot_training - Step 35921: {'lr': 9.883718841784806e-05, 'samples': 18392064, 'steps': 35921, 'loss/train': 1.5009562969207764}
+02/26/2022 00:04:49 - INFO - codeparrot_training - Step 35922: {'lr': 9.88241561988232e-05, 'samples': 18392576, 'steps': 35922, 'loss/train': 0.68912273645401}
+02/26/2022 00:04:52 - INFO - codeparrot_training - Step 35923: {'lr': 9.881112462738784e-05, 'samples': 18393088, 'steps': 35923, 'loss/train': 1.9958263635635376}
+02/26/2022 00:04:58 - INFO - codeparrot_training - Step 35924: {'lr': 9.879809370359744e-05, 'samples': 18393600, 'steps': 35924, 'loss/train': 2.1054980754852295}
+02/26/2022 00:05:01 - INFO - codeparrot_training - Step 35925: {'lr': 9.878506342750806e-05, 'samples': 18394112, 'steps': 35925, 'loss/train': 0.9638531804084778}
+02/26/2022 00:05:07 - INFO - codeparrot_training - Step 35926: {'lr': 9.877203379917541e-05, 'samples': 18394624, 'steps': 35926, 'loss/train': 2.6006088256835938}
+02/26/2022 00:05:10 - INFO - codeparrot_training - Step 35927: {'lr': 9.875900481865549e-05, 'samples': 18395136, 'steps': 35927, 'loss/train': 1.6636947393417358}
+02/26/2022 00:05:16 - INFO - codeparrot_training - Step 35928: {'lr': 9.874597648600378e-05, 'samples': 18395648, 'steps': 35928, 'loss/train': 0.9224361777305603}
+02/26/2022 00:05:20 - INFO - codeparrot_training - Step 35929: {'lr': 9.87329488012764e-05, 'samples': 18396160, 'steps': 35929, 'loss/train': 1.7337384223937988}
+02/26/2022 00:05:25 - INFO - codeparrot_training - Step 35930: {'lr': 9.871992176452894e-05, 'samples': 18396672, 'steps': 35930, 'loss/train': 1.2812590599060059}
+02/26/2022 00:05:29 - INFO - codeparrot_training - Step 35931: {'lr': 9.870689537581745e-05, 'samples': 18397184, 'steps': 35931, 'loss/train': 1.6524097919464111}
+02/26/2022 00:05:34 - INFO - codeparrot_training - Step 35932: {'lr': 9.869386963519742e-05, 'samples': 18397696, 'steps': 35932, 'loss/train': 1.4401382207870483}
+02/26/2022 00:05:38 - INFO - codeparrot_training - Step 35933: {'lr': 9.868084454272488e-05, 'samples': 18398208, 'steps': 35933, 'loss/train': 1.5436817407608032}
+02/26/2022 00:05:43 - INFO - codeparrot_training - Step 35934: {'lr': 9.866782009845549e-05, 'samples': 18398720, 'steps': 35934, 'loss/train': 2.2125649452209473}
+02/26/2022 00:05:47 - INFO - codeparrot_training - Step 35935: {'lr': 9.86547963024452e-05, 'samples': 18399232, 'steps': 35935, 'loss/train': 1.5236769914627075}
+02/26/2022 00:05:53 - INFO - codeparrot_training - Step 35936: {'lr': 9.864177315474967e-05, 'samples': 18399744, 'steps': 35936, 'loss/train': 1.952535629272461}
+02/26/2022 00:05:56 - INFO - codeparrot_training - Step 35937: {'lr': 9.862875065542476e-05, 'samples': 18400256, 'steps': 35937, 'loss/train': 2.332662582397461}
+02/26/2022 00:06:02 - INFO - codeparrot_training - Step 35938: {'lr': 9.86157288045261e-05, 'samples': 18400768, 'steps': 35938, 'loss/train': 1.490973711013794}
+02/26/2022 00:06:06 - INFO - codeparrot_training - Step 35939: {'lr': 9.860270760210968e-05, 'samples': 18401280, 'steps': 35939, 'loss/train': 1.6946594715118408}
+02/26/2022 00:06:11 - INFO - codeparrot_training - Step 35940: {'lr': 9.858968704823117e-05, 'samples': 18401792, 'steps': 35940, 'loss/train': 1.5306942462921143}
+02/26/2022 00:06:15 - INFO - codeparrot_training - Step 35941: {'lr': 9.857666714294641e-05, 'samples': 18402304, 'steps': 35941, 'loss/train': 2.0050270557403564}
+02/26/2022 00:06:20 - INFO - codeparrot_training - Step 35942: {'lr': 9.8563647886311e-05, 'samples': 18402816, 'steps': 35942, 'loss/train': 1.5935672521591187}
+02/26/2022 00:06:23 - INFO - codeparrot_training - Step 35943: {'lr': 9.855062927838093e-05, 'samples': 18403328, 'steps': 35943, 'loss/train': 1.4004801511764526}
+02/26/2022 00:06:29 - INFO - codeparrot_training - Step 35944: {'lr': 9.85376113192118e-05, 'samples': 18403840, 'steps': 35944, 'loss/train': 0.8524960875511169}
+02/26/2022 00:06:33 - INFO - codeparrot_training - Step 35945: {'lr': 9.85245940088596e-05, 'samples': 18404352, 'steps': 35945, 'loss/train': 1.702237844467163}
+02/26/2022 00:06:38 - INFO - codeparrot_training - Step 35946: {'lr': 9.851157734737978e-05, 'samples': 18404864, 'steps': 35946, 'loss/train': 2.018848180770874}
+02/26/2022 00:06:42 - INFO - codeparrot_training - Step 35947: {'lr': 9.849856133482834e-05, 'samples': 18405376, 'steps': 35947, 'loss/train': 1.7983595132827759}
+02/26/2022 00:06:47 - INFO - codeparrot_training - Step 35948: {'lr': 9.848554597126088e-05, 'samples': 18405888, 'steps': 35948, 'loss/train': 2.495162010192871}
+02/26/2022 00:06:51 - INFO - codeparrot_training - Step 35949: {'lr': 9.847253125673342e-05, 'samples': 18406400, 'steps': 35949, 'loss/train': 1.4930616617202759}
+02/26/2022 00:06:56 - INFO - codeparrot_training - Step 35950: {'lr': 9.84595171913013e-05, 'samples': 18406912, 'steps': 35950, 'loss/train': 1.4348050355911255}
+02/26/2022 00:07:00 - INFO - codeparrot_training - Step 35951: {'lr': 9.844650377502063e-05, 'samples': 18407424, 'steps': 35951, 'loss/train': 1.898951768875122}
+02/26/2022 00:07:05 - INFO - codeparrot_training - Step 35952: {'lr': 9.843349100794696e-05, 'samples': 18407936, 'steps': 35952, 'loss/train': 1.524706244468689}
+02/26/2022 00:07:09 - INFO - codeparrot_training - Step 35953: {'lr': 9.842047889013613e-05, 'samples': 18408448, 'steps': 35953, 'loss/train': 1.0840580463409424}
+02/26/2022 00:07:15 - INFO - codeparrot_training - Step 35954: {'lr': 9.840746742164372e-05, 'samples': 18408960, 'steps': 35954, 'loss/train': 2.1585886478424072}
+02/26/2022 00:07:18 - INFO - codeparrot_training - Step 35955: {'lr': 9.83944566025257e-05, 'samples': 18409472, 'steps': 35955, 'loss/train': 2.505445718765259}
+02/26/2022 00:07:24 - INFO - codeparrot_training - Step 35956: {'lr': 9.838144643283767e-05, 'samples': 18409984, 'steps': 35956, 'loss/train': 1.520660400390625}
+02/26/2022 00:07:27 - INFO - codeparrot_training - Step 35957: {'lr': 9.836843691263528e-05, 'samples': 18410496, 'steps': 35957, 'loss/train': 2.3834221363067627}
+02/26/2022 00:07:33 - INFO - codeparrot_training - Step 35958: {'lr': 9.835542804197448e-05, 'samples': 18411008, 'steps': 35958, 'loss/train': 2.029059648513794}
+02/26/2022 00:07:36 - INFO - codeparrot_training - Step 35959: {'lr': 9.834241982091082e-05, 'samples': 18411520, 'steps': 35959, 'loss/train': 1.187527060508728}
+02/26/2022 00:07:42 - INFO - codeparrot_training - Step 35960: {'lr': 9.832941224950012e-05, 'samples': 18412032, 'steps': 35960, 'loss/train': 1.8057677745819092}
+02/26/2022 00:07:45 - INFO - codeparrot_training - Step 35961: {'lr': 9.831640532779795e-05, 'samples': 18412544, 'steps': 35961, 'loss/train': 2.1662979125976562}
+02/26/2022 00:07:51 - INFO - codeparrot_training - Step 35962: {'lr': 9.830339905586025e-05, 'samples': 18413056, 'steps': 35962, 'loss/train': 2.6305973529815674}
+02/26/2022 00:07:54 - INFO - codeparrot_training - Step 35963: {'lr': 9.82903934337426e-05, 'samples': 18413568, 'steps': 35963, 'loss/train': 1.7157042026519775}
+02/26/2022 00:08:01 - INFO - codeparrot_training - Step 35964: {'lr': 9.827738846150073e-05, 'samples': 18414080, 'steps': 35964, 'loss/train': 1.8077280521392822}
+02/26/2022 00:08:04 - INFO - codeparrot_training - Step 35965: {'lr': 9.826438413919026e-05, 'samples': 18414592, 'steps': 35965, 'loss/train': 1.522325873374939}
+02/26/2022 00:08:10 - INFO - codeparrot_training - Step 35966: {'lr': 9.825138046686716e-05, 'samples': 18415104, 'steps': 35966, 'loss/train': 2.0576868057250977}
+02/26/2022 00:08:13 - INFO - codeparrot_training - Step 35967: {'lr': 9.82383774445868e-05, 'samples': 18415616, 'steps': 35967, 'loss/train': 1.4208050966262817}
+02/26/2022 00:08:19 - INFO - codeparrot_training - Step 35968: {'lr': 9.822537507240512e-05, 'samples': 18416128, 'steps': 35968, 'loss/train': 3.1818759441375732}
+02/26/2022 00:08:22 - INFO - codeparrot_training - Step 35969: {'lr': 9.821237335037763e-05, 'samples': 18416640, 'steps': 35969, 'loss/train': 1.4404741525650024}
+02/26/2022 00:08:28 - INFO - codeparrot_training - Step 35970: {'lr': 9.819937227856027e-05, 'samples': 18417152, 'steps': 35970, 'loss/train': 2.3471426963806152}
+02/26/2022 00:08:31 - INFO - codeparrot_training - Step 35971: {'lr': 9.818637185700855e-05, 'samples': 18417664, 'steps': 35971, 'loss/train': 0.8370423316955566}
+02/26/2022 00:08:37 - INFO - codeparrot_training - Step 35972: {'lr': 9.817337208577823e-05, 'samples': 18418176, 'steps': 35972, 'loss/train': 1.9797887802124023}
+02/26/2022 00:08:40 - INFO - codeparrot_training - Step 35973: {'lr': 9.816037296492489e-05, 'samples': 18418688, 'steps': 35973, 'loss/train': 1.2125232219696045}
+02/26/2022 00:08:46 - INFO - codeparrot_training - Step 35974: {'lr': 9.814737449450437e-05, 'samples': 18419200, 'steps': 35974, 'loss/train': 1.9455817937850952}
+02/26/2022 00:08:50 - INFO - codeparrot_training - Step 35975: {'lr': 9.813437667457229e-05, 'samples': 18419712, 'steps': 35975, 'loss/train': 2.243952751159668}
+02/26/2022 00:08:55 - INFO - codeparrot_training - Step 35976: {'lr': 9.812137950518432e-05, 'samples': 18420224, 'steps': 35976, 'loss/train': 0.57295161485672}
+02/26/2022 00:08:59 - INFO - codeparrot_training - Step 35977: {'lr': 9.810838298639605e-05, 'samples': 18420736, 'steps': 35977, 'loss/train': 1.1149544715881348}
+02/26/2022 00:09:04 - INFO - codeparrot_training - Step 35978: {'lr': 9.80953871182633e-05, 'samples': 18421248, 'steps': 35978, 'loss/train': 1.590366005897522}
+02/26/2022 00:09:08 - INFO - codeparrot_training - Step 35979: {'lr': 9.80823919008417e-05, 'samples': 18421760, 'steps': 35979, 'loss/train': 2.021833658218384}
+02/26/2022 00:09:13 - INFO - codeparrot_training - Step 35980: {'lr': 9.806939733418687e-05, 'samples': 18422272, 'steps': 35980, 'loss/train': 1.6028356552124023}
+02/26/2022 00:09:17 - INFO - codeparrot_training - Step 35981: {'lr': 9.805640341835442e-05, 'samples': 18422784, 'steps': 35981, 'loss/train': 1.319348931312561}
+02/26/2022 00:09:22 - INFO - codeparrot_training - Step 35982: {'lr': 9.804341015340021e-05, 'samples': 18423296, 'steps': 35982, 'loss/train': 1.474488377571106}
+02/26/2022 00:09:26 - INFO - codeparrot_training - Step 35983: {'lr': 9.803041753937967e-05, 'samples': 18423808, 'steps': 35983, 'loss/train': 2.3666207790374756}
+02/26/2022 00:09:32 - INFO - codeparrot_training - Step 35984: {'lr': 9.801742557634872e-05, 'samples': 18424320, 'steps': 35984, 'loss/train': 1.7574007511138916}
+02/26/2022 00:09:35 - INFO - codeparrot_training - Step 35985: {'lr': 9.80044342643627e-05, 'samples': 18424832, 'steps': 35985, 'loss/train': 1.3362573385238647}
+02/26/2022 00:09:41 - INFO - codeparrot_training - Step 35986: {'lr': 9.799144360347751e-05, 'samples': 18425344, 'steps': 35986, 'loss/train': 1.7247356176376343}
+02/26/2022 00:09:44 - INFO - codeparrot_training - Step 35987: {'lr': 9.79784535937486e-05, 'samples': 18425856, 'steps': 35987, 'loss/train': 2.2092444896698}
+02/26/2022 00:09:50 - INFO - codeparrot_training - Step 35988: {'lr': 9.796546423523192e-05, 'samples': 18426368, 'steps': 35988, 'loss/train': 0.9692898988723755}
+02/26/2022 00:09:53 - INFO - codeparrot_training - Step 35989: {'lr': 9.795247552798273e-05, 'samples': 18426880, 'steps': 35989, 'loss/train': 0.5010618567466736}
+02/26/2022 00:09:59 - INFO - codeparrot_training - Step 35990: {'lr': 9.793948747205694e-05, 'samples': 18427392, 'steps': 35990, 'loss/train': 1.6082537174224854}
+02/26/2022 00:10:04 - INFO - codeparrot_training - Step 35991: {'lr': 9.792650006751002e-05, 'samples': 18427904, 'steps': 35991, 'loss/train': 1.2630656957626343}
+02/26/2022 00:10:08 - INFO - codeparrot_training - Step 35992: {'lr': 9.791351331439779e-05, 'samples': 18428416, 'steps': 35992, 'loss/train': 1.6579424142837524}
+02/26/2022 00:10:13 - INFO - codeparrot_training - Step 35993: {'lr': 9.790052721277576e-05, 'samples': 18428928, 'steps': 35993, 'loss/train': 2.244624614715576}
+02/26/2022 00:10:17 - INFO - codeparrot_training - Step 35994: {'lr': 9.788754176269957e-05, 'samples': 18429440, 'steps': 35994, 'loss/train': 1.616926670074463}
+02/26/2022 00:10:22 - INFO - codeparrot_training - Step 35995: {'lr': 9.787455696422476e-05, 'samples': 18429952, 'steps': 35995, 'loss/train': 1.8336987495422363}
+02/26/2022 00:10:26 - INFO - codeparrot_training - Step 35996: {'lr': 9.786157281740712e-05, 'samples': 18430464, 'steps': 35996, 'loss/train': 2.339150905609131}
+02/26/2022 00:10:31 - INFO - codeparrot_training - Step 35997: {'lr': 9.784858932230222e-05, 'samples': 18430976, 'steps': 35997, 'loss/train': 1.881528615951538}
+02/26/2022 00:10:35 - INFO - codeparrot_training - Step 35998: {'lr': 9.783560647896564e-05, 'samples': 18431488, 'steps': 35998, 'loss/train': 0.11420764774084091}
+02/26/2022 00:10:41 - INFO - codeparrot_training - Step 35999: {'lr': 9.7822624287453e-05, 'samples': 18432000, 'steps': 35999, 'loss/train': 1.2246854305267334}
+02/26/2022 00:10:41 - INFO - codeparrot_training - Evaluating and saving model checkpoint