diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -43560,3 +43560,1009 @@ Use FP16 precision: False
 02/26/2022 09:14:58 - INFO - codeparrot_training - Step 42998: {'lr': 2.5796290401286865e-05, 'samples': 22015488, 'steps': 42998, 'loss/train': 1.8694614171981812}
 02/26/2022 09:15:01 - INFO - codeparrot_training - Step 42999: {'lr': 2.578905202883311e-05, 'samples': 22016000, 'steps': 42999, 'loss/train': 1.5767271518707275}
 02/26/2022 09:15:01 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 09:15:19 - WARNING - huggingface_hub.repository - Several commits (43) will be pushed upstream.
+02/26/2022 09:15:19 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 09:15:53 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   2c9f21a..d24e10b  floral-grass-11 -> floral-grass-11
+
+02/26/2022 09:16:00 - INFO - codeparrot_training - Step 43000: {'lr': 2.5781814616827938e-05, 'samples': 22016512, 'steps': 43000, 'loss/train': 1.3371886014938354}
+02/26/2022 09:16:03 - INFO - codeparrot_training - Step 43001: {'lr': 2.5774578165302236e-05, 'samples': 22017024, 'steps': 43001, 'loss/train': 1.2176768779754639}
+02/26/2022 09:16:09 - INFO - codeparrot_training - Step 43002: {'lr': 2.576734267428707e-05, 'samples': 22017536, 'steps': 43002, 'loss/train': 2.634197950363159}
+02/26/2022 09:16:12 - INFO - codeparrot_training - Step 43003: {'lr': 2.5760108143813353e-05, 'samples': 22018048, 'steps': 43003, 'loss/train': 1.4724316596984863}
+02/26/2022 09:16:18 - INFO - codeparrot_training - Step 43004: {'lr': 2.575287457391218e-05, 'samples': 22018560, 'steps': 43004, 'loss/train': 1.85142183303833}
+02/26/2022 09:16:21 - INFO - codeparrot_training - Step 43005: {'lr': 2.5745641964614485e-05, 'samples': 22019072, 'steps': 43005, 'loss/train': 2.0375685691833496}
+02/26/2022 09:16:27 - INFO - codeparrot_training - Step 43006: {'lr': 2.5738410315951256e-05, 'samples': 22019584, 'steps': 43006, 'loss/train': 0.5772395133972168}
+02/26/2022 09:16:32 - INFO - codeparrot_training - Step 43007: {'lr': 2.5731179627953406e-05, 'samples': 22020096, 'steps': 43007, 'loss/train': 1.8390233516693115}
+02/26/2022 09:16:36 - INFO - codeparrot_training - Step 43008: {'lr': 2.572394990065202e-05, 'samples': 22020608, 'steps': 43008, 'loss/train': 3.028017282485962}
+02/26/2022 09:16:41 - INFO - codeparrot_training - Step 43009: {'lr': 2.5716721134078e-05, 'samples': 22021120, 'steps': 43009, 'loss/train': 1.6743240356445312}
+02/26/2022 09:16:45 - INFO - codeparrot_training - Step 43010: {'lr': 2.5709493328262363e-05, 'samples': 22021632, 'steps': 43010, 'loss/train': 0.19287554919719696}
+02/26/2022 09:16:52 - INFO - codeparrot_training - Step 43011: {'lr': 2.5702266483235952e-05, 'samples': 22022144, 'steps': 43011, 'loss/train': 1.4503531455993652}
+02/26/2022 09:16:55 - INFO - codeparrot_training - Step 43012: {'lr': 2.5695040599029883e-05, 'samples': 22022656, 'steps': 43012, 'loss/train': 2.0527825355529785}
+02/26/2022 09:17:01 - INFO - codeparrot_training - Step 43013: {'lr': 2.568781567567502e-05, 'samples': 22023168, 'steps': 43013, 'loss/train': 1.9433341026306152}
+02/26/2022 09:17:04 - INFO - codeparrot_training - Step 43014: {'lr': 2.5680591713202333e-05, 'samples': 22023680, 'steps': 43014, 'loss/train': 0.9118227362632751}
+02/26/2022 09:17:10 - INFO - codeparrot_training - Step 43015: {'lr': 2.567336871164272e-05, 'samples': 22024192, 'steps': 43015, 'loss/train': 1.5988863706588745}
+02/26/2022 09:17:13 - INFO - codeparrot_training - Step 43016: {'lr': 2.5666146671027206e-05, 'samples': 22024704, 'steps': 43016, 'loss/train': 1.2842426300048828}
+02/26/2022 09:17:19 - INFO - codeparrot_training - Step 43017: {'lr': 2.565892559138669e-05, 'samples': 22025216, 'steps': 43017, 'loss/train': 1.4712259769439697}
+02/26/2022 09:17:22 - INFO - codeparrot_training - Step 43018: {'lr': 2.565170547275211e-05, 'samples': 22025728, 'steps': 43018, 'loss/train': 0.6036337018013}
+02/26/2022 09:17:28 - INFO - codeparrot_training - Step 43019: {'lr': 2.5644486315154315e-05, 'samples': 22026240, 'steps': 43019, 'loss/train': 2.047445774078369}
+02/26/2022 09:17:31 - INFO - codeparrot_training - Step 43020: {'lr': 2.563726811862435e-05, 'samples': 22026752, 'steps': 43020, 'loss/train': 2.0443203449249268}
+02/26/2022 09:17:35 - INFO - codeparrot_training - Step 43021: {'lr': 2.5630050883193062e-05, 'samples': 22027264, 'steps': 43021, 'loss/train': 1.2457826137542725}
+02/26/2022 09:17:42 - INFO - codeparrot_training - Step 43022: {'lr': 2.5622834608891475e-05, 'samples': 22027776, 'steps': 43022, 'loss/train': 0.8604428172111511}
+02/26/2022 09:17:48 - INFO - codeparrot_training - Step 43023: {'lr': 2.5615619295750293e-05, 'samples': 22028288, 'steps': 43023, 'loss/train': 1.1581933498382568}
+02/26/2022 09:17:51 - INFO - codeparrot_training - Step 43024: {'lr': 2.560840494380062e-05, 'samples': 22028800, 'steps': 43024, 'loss/train': 0.9329136610031128}
+02/26/2022 09:17:57 - INFO - codeparrot_training - Step 43025: {'lr': 2.560119155307325e-05, 'samples': 22029312, 'steps': 43025, 'loss/train': 2.4660651683807373}
+02/26/2022 09:18:00 - INFO - codeparrot_training - Step 43026: {'lr': 2.5593979123599205e-05, 'samples': 22029824, 'steps': 43026, 'loss/train': 0.03613194078207016}
+02/26/2022 09:18:06 - INFO - codeparrot_training - Step 43027: {'lr': 2.558676765540921e-05, 'samples': 22030336, 'steps': 43027, 'loss/train': 2.0503616333007812}
+02/26/2022 09:18:09 - INFO - codeparrot_training - Step 43028: {'lr': 2.5579557148534272e-05, 'samples': 22030848, 'steps': 43028, 'loss/train': 1.5886479616165161}
+02/26/2022 09:18:15 - INFO - codeparrot_training - Step 43029: {'lr': 2.5572347603005196e-05, 'samples': 22031360, 'steps': 43029, 'loss/train': 1.7415753602981567}
+02/26/2022 09:18:18 - INFO - codeparrot_training - Step 43030: {'lr': 2.5565139018853046e-05, 'samples': 22031872, 'steps': 43030, 'loss/train': 2.2385222911834717}
+02/26/2022 09:18:25 - INFO - codeparrot_training - Step 43031: {'lr': 2.555793139610846e-05, 'samples': 22032384, 'steps': 43031, 'loss/train': 1.4883617162704468}
+02/26/2022 09:18:29 - INFO - codeparrot_training - Step 43032: {'lr': 2.5550724734802466e-05, 'samples': 22032896, 'steps': 43032, 'loss/train': 1.266660213470459}
+02/26/2022 09:18:34 - INFO - codeparrot_training - Step 43033: {'lr': 2.5543519034965878e-05, 'samples': 22033408, 'steps': 43033, 'loss/train': 1.63584566116333}
+02/26/2022 09:18:38 - INFO - codeparrot_training - Step 43034: {'lr': 2.5536314296629613e-05, 'samples': 22033920, 'steps': 43034, 'loss/train': 1.8765552043914795}
+02/26/2022 09:18:43 - INFO - codeparrot_training - Step 43035: {'lr': 2.5529110519824482e-05, 'samples': 22034432, 'steps': 43035, 'loss/train': 1.5633749961853027}
+02/26/2022 09:18:47 - INFO - codeparrot_training - Step 43036: {'lr': 2.5521907704581405e-05, 'samples': 22034944, 'steps': 43036, 'loss/train': 2.941575765609741}
+02/26/2022 09:18:52 - INFO - codeparrot_training - Step 43037: {'lr': 2.551470585093113e-05, 'samples': 22035456, 'steps': 43037, 'loss/train': 2.3214941024780273}
+02/26/2022 09:18:56 - INFO - codeparrot_training - Step 43038: {'lr': 2.550750495890461e-05, 'samples': 22035968, 'steps': 43038, 'loss/train': 1.8605632781982422}
+02/26/2022 09:19:01 - INFO - codeparrot_training - Step 43039: {'lr': 2.550030502853265e-05, 'samples': 22036480, 'steps': 43039, 'loss/train': 2.09035325050354}
+02/26/2022 09:19:05 - INFO - codeparrot_training - Step 43040: {'lr': 2.5493106059846115e-05, 'samples': 22036992, 'steps': 43040, 'loss/train': 2.655060052871704}
+02/26/2022 09:19:08 - INFO - codeparrot_training - Step 43041: {'lr': 2.5485908052875762e-05, 'samples': 22037504, 'steps': 43041, 'loss/train': 1.8697302341461182}
+02/26/2022 09:19:16 - INFO - codeparrot_training - Step 43042: {'lr': 2.547871100765256e-05, 'samples': 22038016, 'steps': 43042, 'loss/train': 2.2321293354034424}
+02/26/2022 09:19:19 - INFO - codeparrot_training - Step 43043: {'lr': 2.547151492420724e-05, 'samples': 22038528, 'steps': 43043, 'loss/train': 1.0352306365966797}
+02/26/2022 09:19:25 - INFO - codeparrot_training - Step 43044: {'lr': 2.546431980257066e-05, 'samples': 22039040, 'steps': 43044, 'loss/train': 2.053722620010376}
+02/26/2022 09:19:28 - INFO - codeparrot_training - Step 43045: {'lr': 2.5457125642773638e-05, 'samples': 22039552, 'steps': 43045, 'loss/train': 1.6874339580535889}
+02/26/2022 09:19:34 - INFO - codeparrot_training - Step 43046: {'lr': 2.5449932444846944e-05, 'samples': 22040064, 'steps': 43046, 'loss/train': 1.937677025794983}
+02/26/2022 09:19:37 - INFO - codeparrot_training - Step 43047: {'lr': 2.544274020882148e-05, 'samples': 22040576, 'steps': 43047, 'loss/train': 2.045397996902466}
+02/26/2022 09:19:43 - INFO - codeparrot_training - Step 43048: {'lr': 2.543554893472802e-05, 'samples': 22041088, 'steps': 43048, 'loss/train': 1.6934360265731812}
+02/26/2022 09:19:48 - INFO - codeparrot_training - Step 43049: {'lr': 2.5428358622597375e-05, 'samples': 22041600, 'steps': 43049, 'loss/train': 2.004577875137329}
+02/26/2022 09:19:52 - INFO - codeparrot_training - Step 43050: {'lr': 2.542116927246024e-05, 'samples': 22042112, 'steps': 43050, 'loss/train': 0.7194834351539612}
+02/26/2022 09:19:57 - INFO - codeparrot_training - Step 43051: {'lr': 2.5413980884347598e-05, 'samples': 22042624, 'steps': 43051, 'loss/train': 1.7380951642990112}
+02/26/2022 09:20:01 - INFO - codeparrot_training - Step 43052: {'lr': 2.5406793458290113e-05, 'samples': 22043136, 'steps': 43052, 'loss/train': 2.375088691711426}
+02/26/2022 09:20:04 - INFO - codeparrot_training - Step 43053: {'lr': 2.539960699431862e-05, 'samples': 22043648, 'steps': 43053, 'loss/train': 1.9548765420913696}
+02/26/2022 09:20:10 - INFO - codeparrot_training - Step 43054: {'lr': 2.5392421492463846e-05, 'samples': 22044160, 'steps': 43054, 'loss/train': 1.586997389793396}
+02/26/2022 09:20:13 - INFO - codeparrot_training - Step 43055: {'lr': 2.5385236952756687e-05, 'samples': 22044672, 'steps': 43055, 'loss/train': 1.9782105684280396}
+02/26/2022 09:20:19 - INFO - codeparrot_training - Step 43056: {'lr': 2.537805337522783e-05, 'samples': 22045184, 'steps': 43056, 'loss/train': 1.9436240196228027}
+02/26/2022 09:20:23 - INFO - codeparrot_training - Step 43057: {'lr': 2.5370870759908073e-05, 'samples': 22045696, 'steps': 43057, 'loss/train': 0.04014909639954567}
+02/26/2022 09:20:30 - INFO - codeparrot_training - Step 43058: {'lr': 2.5363689106828096e-05, 'samples': 22046208, 'steps': 43058, 'loss/train': 1.643044114112854}
+02/26/2022 09:20:33 - INFO - codeparrot_training - Step 43059: {'lr': 2.535650841601883e-05, 'samples': 22046720, 'steps': 43059, 'loss/train': 1.259970784187317}
+02/26/2022 09:20:39 - INFO - codeparrot_training - Step 43060: {'lr': 2.5349328687510888e-05, 'samples': 22047232, 'steps': 43060, 'loss/train': 1.5826419591903687}
+02/26/2022 09:20:42 - INFO - codeparrot_training - Step 43061: {'lr': 2.5342149921335182e-05, 'samples': 22047744, 'steps': 43061, 'loss/train': 1.0303560495376587}
+02/26/2022 09:20:48 - INFO - codeparrot_training - Step 43062: {'lr': 2.533497211752228e-05, 'samples': 22048256, 'steps': 43062, 'loss/train': 2.214817762374878}
+02/26/2022 09:20:53 - INFO - codeparrot_training - Step 43063: {'lr': 2.5327795276103072e-05, 'samples': 22048768, 'steps': 43063, 'loss/train': 1.8114770650863647}
+02/26/2022 09:20:57 - INFO - codeparrot_training - Step 43064: {'lr': 2.5320619397108197e-05, 'samples': 22049280, 'steps': 43064, 'loss/train': 1.6828633546829224}
+02/26/2022 09:21:03 - INFO - codeparrot_training - Step 43065: {'lr': 2.531344448056852e-05, 'samples': 22049792, 'steps': 43065, 'loss/train': 1.882368803024292}
+02/26/2022 09:21:06 - INFO - codeparrot_training - Step 43066: {'lr': 2.530627052651463e-05, 'samples': 22050304, 'steps': 43066, 'loss/train': 1.5169315338134766}
+02/26/2022 09:21:13 - INFO - codeparrot_training - Step 43067: {'lr': 2.529909753497736e-05, 'samples': 22050816, 'steps': 43067, 'loss/train': 0.9377289414405823}
+02/26/2022 09:21:17 - INFO - codeparrot_training - Step 43068: {'lr': 2.5291925505987358e-05, 'samples': 22051328, 'steps': 43068, 'loss/train': 2.026712417602539}
+02/26/2022 09:21:20 - INFO - codeparrot_training - Step 43069: {'lr': 2.5284754439575452e-05, 'samples': 22051840, 'steps': 43069, 'loss/train': 2.1036181449890137}
+02/26/2022 09:21:26 - INFO - codeparrot_training - Step 43070: {'lr': 2.527758433577232e-05, 'samples': 22052352, 'steps': 43070, 'loss/train': 1.6851140260696411}
+02/26/2022 09:21:31 - INFO - codeparrot_training - Step 43071: {'lr': 2.5270415194608627e-05, 'samples': 22052864, 'steps': 43071, 'loss/train': 1.0842164754867554}
+02/26/2022 09:21:35 - INFO - codeparrot_training - Step 43072: {'lr': 2.526324701611507e-05, 'samples': 22053376, 'steps': 43072, 'loss/train': 1.1889334917068481}
+02/26/2022 09:21:40 - INFO - codeparrot_training - Step 43073: {'lr': 2.525607980032246e-05, 'samples': 22053888, 'steps': 43073, 'loss/train': 1.4469133615493774}
+02/26/2022 09:21:44 - INFO - codeparrot_training - Step 43074: {'lr': 2.5248913547261442e-05, 'samples': 22054400, 'steps': 43074, 'loss/train': 1.580714464187622}
+02/26/2022 09:21:49 - INFO - codeparrot_training - Step 43075: {'lr': 2.524174825696271e-05, 'samples': 22054912, 'steps': 43075, 'loss/train': 0.7569447755813599}
+02/26/2022 09:21:53 - INFO - codeparrot_training - Step 43076: {'lr': 2.5234583929456904e-05, 'samples': 22055424, 'steps': 43076, 'loss/train': 1.4094403982162476}
+02/26/2022 09:22:00 - INFO - codeparrot_training - Step 43077: {'lr': 2.522742056477484e-05, 'samples': 22055936, 'steps': 43077, 'loss/train': 1.3814173936843872}
+02/26/2022 09:22:03 - INFO - codeparrot_training - Step 43078: {'lr': 2.5220258162947097e-05, 'samples': 22056448, 'steps': 43078, 'loss/train': 1.5424473285675049}
+02/26/2022 09:22:09 - INFO - codeparrot_training - Step 43079: {'lr': 2.521309672400443e-05, 'samples': 22056960, 'steps': 43079, 'loss/train': 1.7623589038848877}
+02/26/2022 09:22:12 - INFO - codeparrot_training - Step 43080: {'lr': 2.52059362479774e-05, 'samples': 22057472, 'steps': 43080, 'loss/train': 1.5497207641601562}
+02/26/2022 09:22:18 - INFO - codeparrot_training - Step 43081: {'lr': 2.5198776734896818e-05, 'samples': 22057984, 'steps': 43081, 'loss/train': 1.9506393671035767}
+02/26/2022 09:22:21 - INFO - codeparrot_training - Step 43082: {'lr': 2.5191618184793292e-05, 'samples': 22058496, 'steps': 43082, 'loss/train': 1.6275516748428345}
+02/26/2022 09:22:27 - INFO - codeparrot_training - Step 43083: {'lr': 2.5184460597697496e-05, 'samples': 22059008, 'steps': 43083, 'loss/train': 2.3514351844787598}
+02/26/2022 09:22:30 - INFO - codeparrot_training - Step 43084: {'lr': 2.517730397364004e-05, 'samples': 22059520, 'steps': 43084, 'loss/train': 1.8151108026504517}
+02/26/2022 09:22:36 - INFO - codeparrot_training - Step 43085: {'lr': 2.5170148312651658e-05, 'samples': 22060032, 'steps': 43085, 'loss/train': 2.272878885269165}
+02/26/2022 09:22:39 - INFO - codeparrot_training - Step 43086: {'lr': 2.5162993614762985e-05, 'samples': 22060544, 'steps': 43086, 'loss/train': 0.44775155186653137}
+02/26/2022 09:22:47 - INFO - codeparrot_training - Step 43087: {'lr': 2.5155839880004637e-05, 'samples': 22061056, 'steps': 43087, 'loss/train': 0.6676808595657349}
+02/26/2022 09:22:50 - INFO - codeparrot_training - Step 43088: {'lr': 2.514868710840723e-05, 'samples': 22061568, 'steps': 43088, 'loss/train': 2.3862431049346924}
+02/26/2022 09:22:56 - INFO - codeparrot_training - Step 43089: {'lr': 2.5141535300001518e-05, 'samples': 22062080, 'steps': 43089, 'loss/train': 1.9383883476257324}
+02/26/2022 09:22:59 - INFO - codeparrot_training - Step 43090: {'lr': 2.5134384454818028e-05, 'samples': 22062592, 'steps': 43090, 'loss/train': 1.6923736333847046}
+02/26/2022 09:23:05 - INFO - codeparrot_training - Step 43091: {'lr': 2.512723457288746e-05, 'samples': 22063104, 'steps': 43091, 'loss/train': 3.0778558254241943}
+02/26/2022 09:23:08 - INFO - codeparrot_training - Step 43092: {'lr': 2.5120085654240428e-05, 'samples': 22063616, 'steps': 43092, 'loss/train': 2.2247273921966553}
+02/26/2022 09:23:14 - INFO - codeparrot_training - Step 43093: {'lr': 2.5112937698907464e-05, 'samples': 22064128, 'steps': 43093, 'loss/train': 1.5931341648101807}
+02/26/2022 09:23:17 - INFO - codeparrot_training - Step 43094: {'lr': 2.5105790706919317e-05, 'samples': 22064640, 'steps': 43094, 'loss/train': 1.8834272623062134}
+02/26/2022 09:23:23 - INFO - codeparrot_training - Step 43095: {'lr': 2.5098644678306526e-05, 'samples': 22065152, 'steps': 43095, 'loss/train': 1.5963129997253418}
+02/26/2022 09:23:26 - INFO - codeparrot_training - Step 43096: {'lr': 2.509149961309981e-05, 'samples': 22065664, 'steps': 43096, 'loss/train': 2.1256372928619385}
+02/26/2022 09:23:33 - INFO - codeparrot_training - Step 43097: {'lr': 2.5084355511329592e-05, 'samples': 22066176, 'steps': 43097, 'loss/train': 1.1419918537139893}
+02/26/2022 09:23:37 - INFO - codeparrot_training - Step 43098: {'lr': 2.5077212373026626e-05, 'samples': 22066688, 'steps': 43098, 'loss/train': 1.5744365453720093}
+02/26/2022 09:23:42 - INFO - codeparrot_training - Step 43099: {'lr': 2.507007019822141e-05, 'samples': 22067200, 'steps': 43099, 'loss/train': 1.2878320217132568}
+02/26/2022 09:23:46 - INFO - codeparrot_training - Step 43100: {'lr': 2.5062928986944677e-05, 'samples': 22067712, 'steps': 43100, 'loss/train': 3.3926568031311035}
+02/26/2022 09:23:52 - INFO - codeparrot_training - Step 43101: {'lr': 2.505578873922684e-05, 'samples': 22068224, 'steps': 43101, 'loss/train': 1.2551594972610474}
+02/26/2022 09:23:55 - INFO - codeparrot_training - Step 43102: {'lr': 2.5048649455098632e-05, 'samples': 22068736, 'steps': 43102, 'loss/train': 1.7071492671966553}
+02/26/2022 09:24:01 - INFO - codeparrot_training - Step 43103: {'lr': 2.504151113459055e-05, 'samples': 22069248, 'steps': 43103, 'loss/train': 2.2507243156433105}
+02/26/2022 09:24:04 - INFO - codeparrot_training - Step 43104: {'lr': 2.5034373777733267e-05, 'samples': 22069760, 'steps': 43104, 'loss/train': 0.9035642743110657}
+02/26/2022 09:24:10 - INFO - codeparrot_training - Step 43105: {'lr': 2.50272373845572e-05, 'samples': 22070272, 'steps': 43105, 'loss/train': 2.2026631832122803}
+02/26/2022 09:24:13 - INFO - codeparrot_training - Step 43106: {'lr': 2.5020101955093078e-05, 'samples': 22070784, 'steps': 43106, 'loss/train': 2.4340567588806152}
+02/26/2022 09:24:19 - INFO - codeparrot_training - Step 43107: {'lr': 2.5012967489371346e-05, 'samples': 22071296, 'steps': 43107, 'loss/train': 2.2721855640411377}
+02/26/2022 09:24:22 - INFO - codeparrot_training - Step 43108: {'lr': 2.500583398742265e-05, 'samples': 22071808, 'steps': 43108, 'loss/train': 1.805285096168518}
+02/26/2022 09:24:28 - INFO - codeparrot_training - Step 43109: {'lr': 2.4998701449277545e-05, 'samples': 22072320, 'steps': 43109, 'loss/train': 2.4406607151031494}
+02/26/2022 09:24:32 - INFO - codeparrot_training - Step 43110: {'lr': 2.4991569874966537e-05, 'samples': 22072832, 'steps': 43110, 'loss/train': 1.4582270383834839}
+02/26/2022 09:24:37 - INFO - codeparrot_training - Step 43111: {'lr': 2.4984439264520154e-05, 'samples': 22073344, 'steps': 43111, 'loss/train': 1.9722706079483032}
+02/26/2022 09:24:41 - INFO - codeparrot_training - Step 43112: {'lr': 2.497730961796904e-05, 'samples': 22073856, 'steps': 43112, 'loss/train': 1.6750661134719849}
+02/26/2022 09:24:48 - INFO - codeparrot_training - Step 43113: {'lr': 2.4970180935343646e-05, 'samples': 22074368, 'steps': 43113, 'loss/train': 0.39469850063323975}
+02/26/2022 09:24:51 - INFO - codeparrot_training - Step 43114: {'lr': 2.4963053216674582e-05, 'samples': 22074880, 'steps': 43114, 'loss/train': 2.681544780731201}
+02/26/2022 09:24:57 - INFO - codeparrot_training - Step 43115: {'lr': 2.4955926461992268e-05, 'samples': 22075392, 'steps': 43115, 'loss/train': 1.2985799312591553}
+02/26/2022 09:25:00 - INFO - codeparrot_training - Step 43116: {'lr': 2.4948800671327347e-05, 'samples': 22075904, 'steps': 43116, 'loss/train': 2.1947743892669678}
+02/26/2022 09:25:06 - INFO - codeparrot_training - Step 43117: {'lr': 2.494167584471027e-05, 'samples': 22076416, 'steps': 43117, 'loss/train': 1.6213518381118774}
+02/26/2022 09:25:09 - INFO - codeparrot_training - Step 43118: {'lr': 2.4934551982171704e-05, 'samples': 22076928, 'steps': 43118, 'loss/train': 1.6491273641586304}
+02/26/2022 09:25:15 - INFO - codeparrot_training - Step 43119: {'lr': 2.492742908374193e-05, 'samples': 22077440, 'steps': 43119, 'loss/train': 1.3500185012817383}
+02/26/2022 09:25:18 - INFO - codeparrot_training - Step 43120: {'lr': 2.492030714945162e-05, 'samples': 22077952, 'steps': 43120, 'loss/train': 1.6277194023132324}
+02/26/2022 09:25:24 - INFO - codeparrot_training - Step 43121: {'lr': 2.491318617933122e-05, 'samples': 22078464, 'steps': 43121, 'loss/train': 2.232564687728882}
+02/26/2022 09:25:27 - INFO - codeparrot_training - Step 43122: {'lr': 2.4906066173411347e-05, 'samples': 22078976, 'steps': 43122, 'loss/train': 1.4550281763076782}
+02/26/2022 09:25:35 - INFO - codeparrot_training - Step 43123: {'lr': 2.4898947131722305e-05, 'samples': 22079488, 'steps': 43123, 'loss/train': 1.9753600358963013}
+02/26/2022 09:25:38 - INFO - codeparrot_training - Step 43124: {'lr': 2.489182905429474e-05, 'samples': 22080000, 'steps': 43124, 'loss/train': 0.4699411690235138}
+02/26/2022 09:25:44 - INFO - codeparrot_training - Step 43125: {'lr': 2.4884711941159127e-05, 'samples': 22080512, 'steps': 43125, 'loss/train': 1.6645084619522095}
+02/26/2022 09:25:47 - INFO - codeparrot_training - Step 43126: {'lr': 2.4877595792345885e-05, 'samples': 22081024, 'steps': 43126, 'loss/train': 1.249341607093811}
+02/26/2022 09:25:53 - INFO - codeparrot_training - Step 43127: {'lr': 2.487048060788552e-05, 'samples': 22081536, 'steps': 43127, 'loss/train': 1.398234486579895}
+02/26/2022 09:25:56 - INFO - codeparrot_training - Step 43128: {'lr': 2.486336638780856e-05, 'samples': 22082048, 'steps': 43128, 'loss/train': 0.4382885992527008}
+02/26/2022 09:26:02 - INFO - codeparrot_training - Step 43129: {'lr': 2.4856253132145484e-05, 'samples': 22082560, 'steps': 43129, 'loss/train': 3.2967722415924072}
+02/26/2022 09:26:05 - INFO - codeparrot_training - Step 43130: {'lr': 2.4849140840926655e-05, 'samples': 22083072, 'steps': 43130, 'loss/train': 1.3725587129592896}
+02/26/2022 09:26:11 - INFO - codeparrot_training - Step 43131: {'lr': 2.4842029514182656e-05, 'samples': 22083584, 'steps': 43131, 'loss/train': 1.3670753240585327}
+02/26/2022 09:26:14 - INFO - codeparrot_training - Step 43132: {'lr': 2.4834919151943915e-05, 'samples': 22084096, 'steps': 43132, 'loss/train': 2.967434883117676}
+02/26/2022 09:26:22 - INFO - codeparrot_training - Step 43133: {'lr': 2.4827809754240872e-05, 'samples': 22084608, 'steps': 43133, 'loss/train': 0.5620272159576416}
+02/26/2022 09:26:25 - INFO - codeparrot_training - Step 43134: {'lr': 2.482070132110395e-05, 'samples': 22085120, 'steps': 43134, 'loss/train': 2.092459201812744}
+02/26/2022 09:26:31 - INFO - codeparrot_training - Step 43135: {'lr': 2.481359385256371e-05, 'samples': 22085632, 'steps': 43135, 'loss/train': 2.5158591270446777}
+02/26/2022 09:26:34 - INFO - codeparrot_training - Step 43136: {'lr': 2.4806487348650486e-05, 'samples': 22086144, 'steps': 43136, 'loss/train': 0.18821310997009277}
+02/26/2022 09:26:40 - INFO - codeparrot_training - Step 43137: {'lr': 2.479938180939481e-05, 'samples': 22086656, 'steps': 43137, 'loss/train': 2.0100908279418945}
+02/26/2022 09:26:43 - INFO - codeparrot_training - Step 43138: {'lr': 2.4792277234826993e-05, 'samples': 22087168, 'steps': 43138, 'loss/train': 2.117788791656494}
+02/26/2022 09:26:49 - INFO - codeparrot_training - Step 43139: {'lr': 2.4785173624977647e-05, 'samples': 22087680, 'steps': 43139, 'loss/train': 0.8608589172363281}
+02/26/2022 09:26:52 - INFO - codeparrot_training - Step 43140: {'lr': 2.4778070979877028e-05, 'samples': 22088192, 'steps': 43140, 'loss/train': 0.9559106826782227}
+02/26/2022 09:26:58 - INFO - codeparrot_training - Step 43141: {'lr': 2.4770969299555664e-05, 'samples': 22088704, 'steps': 43141, 'loss/train': 2.077723741531372}
+02/26/2022 09:27:01 - INFO - codeparrot_training - Step 43142: {'lr': 2.476386858404392e-05, 'samples': 22089216, 'steps': 43142, 'loss/train': 1.8173515796661377}
+02/26/2022 09:27:07 - INFO - codeparrot_training - Step 43143: {'lr': 2.4756768833372273e-05, 'samples': 22089728, 'steps': 43143, 'loss/train': 2.4149539470672607}
+02/26/2022 09:27:10 - INFO - codeparrot_training - Step 43144: {'lr': 2.4749670047571114e-05, 'samples': 22090240, 'steps': 43144, 'loss/train': 1.8064653873443604}
+02/26/2022 09:27:17 - INFO - codeparrot_training - Step 43145: {'lr': 2.474257222667084e-05, 'samples': 22090752, 'steps': 43145, 'loss/train': 2.363929271697998}
+02/26/2022 09:27:21 - INFO - codeparrot_training - Step 43146: {'lr': 2.473547537070181e-05, 'samples': 22091264, 'steps': 43146, 'loss/train': 1.602826476097107}
+02/26/2022 09:27:27 - INFO - codeparrot_training - Step 43147: {'lr': 2.472837947969453e-05, 'samples': 22091776, 'steps': 43147, 'loss/train': 2.1847267150878906}
+02/26/2022 09:27:30 - INFO - codeparrot_training - Step 43148: {'lr': 2.4721284553679335e-05, 'samples': 22092288, 'steps': 43148, 'loss/train': 0.6073406934738159}
+02/26/2022 09:27:36 - INFO - codeparrot_training - Step 43149: {'lr': 2.471419059268662e-05, 'samples': 22092800, 'steps': 43149, 'loss/train': 1.8850690126419067}
+02/26/2022 09:27:40 - INFO - codeparrot_training - Step 43150: {'lr': 2.470709759674672e-05, 'samples': 22093312, 'steps': 43150, 'loss/train': 0.9164625406265259}
+02/26/2022 09:27:43 - INFO - codeparrot_training - Step 43151: {'lr': 2.4700005565890136e-05, 'samples': 22093824, 'steps': 43151, 'loss/train': 1.8063238859176636}
+02/26/2022 09:27:48 - INFO - codeparrot_training - Step 43152: {'lr': 2.4692914500147186e-05, 'samples': 22094336, 'steps': 43152, 'loss/train': 0.8614669442176819}
+02/26/2022 09:27:52 - INFO - codeparrot_training - Step 43153: {'lr': 2.4685824399548252e-05, 'samples': 22094848, 'steps': 43153, 'loss/train': 1.6871877908706665}
+02/26/2022 09:27:58 - INFO - codeparrot_training - Step 43154: {'lr': 2.4678735264123647e-05, 'samples': 22095360, 'steps': 43154, 'loss/train': 0.99406898021698}
+02/26/2022 09:28:01 - INFO - codeparrot_training - Step 43155: {'lr': 2.467164709390385e-05, 'samples': 22095872, 'steps': 43155, 'loss/train': 1.5107645988464355}
+02/26/2022 09:28:07 - INFO - codeparrot_training - Step 43156: {'lr': 2.4664559888919108e-05, 'samples': 22096384, 'steps': 43156, 'loss/train': 1.2792789936065674}
+02/26/2022 09:28:10 - INFO - codeparrot_training - Step 43157: {'lr': 2.465747364919993e-05, 'samples': 22096896, 'steps': 43157, 'loss/train': 0.5637739300727844}
+02/26/2022 09:28:15 - INFO - codeparrot_training - Step 43158: {'lr': 2.4650388374776484e-05, 'samples': 22097408, 'steps': 43158, 'loss/train': 1.9193655252456665}
+02/26/2022 09:28:23 - INFO - codeparrot_training - Step 43159: {'lr': 2.46433040656793e-05, 'samples': 22097920, 'steps': 43159, 'loss/train': 0.19971466064453125}
+02/26/2022 09:28:26 - INFO - codeparrot_training - Step 43160: {'lr': 2.4636220721938552e-05, 'samples': 22098432, 'steps': 43160, 'loss/train': 1.2645370960235596}
+02/26/2022 09:28:32 - INFO - codeparrot_training - Step 43161: {'lr': 2.4629138343584795e-05, 'samples': 22098944, 'steps': 43161, 'loss/train': 1.433838129043579}
+02/26/2022 09:28:35 - INFO - codeparrot_training - Step 43162: {'lr': 2.4622056930648146e-05, 'samples': 22099456, 'steps': 43162, 'loss/train': 1.863355040550232}
+02/26/2022 09:28:41 - INFO - codeparrot_training - Step 43163: {'lr': 2.461497648315908e-05, 'samples': 22099968, 'steps': 43163, 'loss/train': 2.170952796936035}
+02/26/2022 09:28:44 - INFO - codeparrot_training - Step 43164: {'lr': 2.460789700114782e-05, 'samples': 22100480, 'steps': 43164, 'loss/train': 3.1801652908325195}
+02/26/2022 09:28:50 - INFO - codeparrot_training - Step 43165: {'lr': 2.4600818484644906e-05, 'samples': 22100992, 'steps': 43165, 'loss/train': 1.2370730638504028}
+02/26/2022 09:28:53 - INFO - codeparrot_training - Step 43166: {'lr': 2.4593740933680387e-05, 'samples': 22101504, 'steps': 43166, 'loss/train': 2.8357667922973633}
+02/26/2022 09:28:59 - INFO - codeparrot_training - Step 43167: {'lr': 2.4586664348284743e-05, 'samples': 22102016, 'steps': 43167, 'loss/train': 1.5399960279464722}
+02/26/2022 09:29:02 - INFO - codeparrot_training - Step 43168: {'lr': 2.4579588728488228e-05, 'samples': 22102528, 'steps': 43168, 'loss/train': 1.034315586090088}
+02/26/2022 09:29:09 - INFO - codeparrot_training - Step 43169: {'lr': 2.4572514074321206e-05, 'samples': 22103040, 'steps': 43169, 'loss/train': 1.7540181875228882}
+02/26/2022 09:29:13 - INFO - codeparrot_training - Step 43170: {'lr': 2.456544038581396e-05, 'samples': 22103552, 'steps': 43170, 'loss/train': 0.8240705132484436}
+02/26/2022 09:29:18 - INFO - codeparrot_training - Step 43171: {'lr': 2.4558367662996796e-05, 'samples': 22104064, 'steps': 43171, 'loss/train': 1.6378718614578247}
+02/26/2022 09:29:22 - INFO - codeparrot_training - Step 43172: {'lr': 2.4551295905899968e-05, 'samples': 22104576, 'steps': 43172, 'loss/train': 2.108032703399658}
+02/26/2022 09:29:28 - INFO - codeparrot_training - Step 43173: {'lr': 2.4544225114553786e-05, 'samples': 22105088, 'steps': 43173, 'loss/train': 1.9060156345367432}
+02/26/2022 09:29:31 - INFO - codeparrot_training - Step 43174: {'lr': 2.453715528898859e-05, 'samples': 22105600, 'steps': 43174, 'loss/train': 0.03589240834116936}
+02/26/2022 09:29:37 - INFO - codeparrot_training - Step 43175: {'lr': 2.45300864292346e-05, 'samples': 22106112, 'steps': 43175, 'loss/train': 1.489375352859497}
+02/26/2022 09:29:40 - INFO - codeparrot_training - Step 43176: {'lr': 2.4523018535322154e-05, 'samples': 22106624, 'steps': 43176, 'loss/train': 2.1516973972320557}
+02/26/2022 09:29:46 - INFO - codeparrot_training - Step 43177: {'lr': 2.4515951607281457e-05, 'samples': 22107136, 'steps': 43177, 'loss/train': 1.1386784315109253}
+02/26/2022 09:29:49 - INFO - codeparrot_training - Step 43178: {'lr': 2.4508885645142865e-05, 'samples': 22107648, 'steps': 43178, 'loss/train': 1.605818748474121}
+02/26/2022 09:29:55 - INFO - codeparrot_training - Step 43179: {'lr': 2.4501820648936608e-05, 'samples': 22108160, 'steps': 43179, 'loss/train': 1.4051034450531006}
+02/26/2022 09:29:58 - INFO - codeparrot_training - Step 43180: {'lr': 2.4494756618692937e-05, 'samples': 22108672, 'steps': 43180, 'loss/train': 1.4265937805175781}
+02/26/2022 09:30:06 - INFO - codeparrot_training - Step 43181: {'lr': 2.4487693554442082e-05, 'samples': 22109184, 'steps': 43181, 'loss/train': 1.743186116218567}
+02/26/2022 09:30:09 - INFO - codeparrot_training - Step 43182: {'lr': 2.4480631456214374e-05, 'samples': 22109696, 'steps': 43182, 'loss/train': 1.8575931787490845}
+02/26/2022 09:30:15 - INFO - codeparrot_training - Step 43183: {'lr': 2.4473570324040047e-05, 'samples': 22110208, 'steps': 43183, 'loss/train': 2.446790933609009}
+02/26/2022 09:30:18 - INFO - codeparrot_training - Step 43184: {'lr': 2.4466510157949318e-05, 'samples': 22110720, 'steps': 43184, 'loss/train': 2.6403350830078125}
+02/26/2022 09:30:24 - INFO - codeparrot_training - Step 43185: {'lr': 2.445945095797242e-05, 'samples': 22111232, 'steps': 43185, 'loss/train': 1.784572958946228}
+02/26/2022 09:30:27 - INFO - codeparrot_training - Step 43186: {'lr': 2.4452392724139656e-05, 'samples': 22111744, 'steps': 43186, 'loss/train': 1.5917832851409912}
+02/26/2022 09:30:33 - INFO - codeparrot_training - Step 43187: {'lr': 2.44453354564812e-05, 'samples': 22112256, 'steps': 43187, 'loss/train': 1.6441023349761963}
+02/26/2022 09:30:36 - INFO - codeparrot_training - Step 43188: {'lr': 2.4438279155027333e-05, 'samples': 22112768, 'steps': 43188, 'loss/train': 1.41212797164917}
+02/26/2022 09:30:42 - INFO - codeparrot_training - Step 43189: {'lr': 2.4431223819808197e-05, 'samples': 22113280, 'steps': 43189, 'loss/train': 0.12104640156030655}
+02/26/2022 09:30:45 - INFO - codeparrot_training - Step 43190: {'lr': 2.44241694508541e-05, 'samples': 22113792, 'steps': 43190, 'loss/train': 1.7831441164016724}
+02/26/2022 09:30:51 - INFO - codeparrot_training - Step 43191: {'lr': 2.4417116048195213e-05, 'samples': 22114304, 'steps': 43191, 'loss/train': 1.0775939226150513}
+02/26/2022 09:30:54 - INFO - codeparrot_training - Step 43192: {'lr': 2.441006361186185e-05, 'samples': 22114816, 'steps': 43192, 'loss/train': 0.931842565536499}
+02/26/2022 09:31:00 - INFO - codeparrot_training - Step 43193: {'lr': 2.4403012141884063e-05, 'samples': 22115328, 'steps': 43193, 'loss/train': 1.607734203338623}
+02/26/2022 09:31:03 - INFO - codeparrot_training - Step 43194: {'lr': 2.4395961638292162e-05, 'samples': 22115840, 'steps': 43194, 'loss/train': 1.892438292503357}
+02/26/2022 09:31:11 - INFO - codeparrot_training - Step 43195: {'lr': 2.4388912101116295e-05, 'samples': 22116352, 'steps': 43195, 'loss/train': 2.179072618484497}
+02/26/2022 09:31:14 - INFO - codeparrot_training - Step 43196: {'lr': 2.4381863530386766e-05, 'samples': 22116864, 'steps': 43196, 'loss/train': 1.7569949626922607}
+02/26/2022 09:31:20 - INFO - codeparrot_training - Step 43197: {'lr': 2.437481592613361e-05, 'samples': 22117376, 'steps': 43197, 'loss/train': 1.214928150177002}
+02/26/2022 09:31:23 - INFO - codeparrot_training - Step 43198: {'lr': 2.436776928838716e-05, 'samples': 22117888, 'steps': 43198, 'loss/train': 2.00553297996521}
+02/26/2022 09:31:29 - INFO - codeparrot_training - Step 43199: {'lr': 2.4360723617177477e-05, 'samples': 22118400, 'steps': 43199, 'loss/train': 0.6758792400360107}
+02/26/2022 09:31:32 - INFO - codeparrot_training - Step 43200: {'lr': 2.43536789125349e-05, 'samples': 22118912, 'steps': 43200, 'loss/train': 2.642744541168213}
+02/26/2022 09:31:38 - INFO - codeparrot_training - Step 43201: {'lr': 2.43466351744894e-05, 'samples': 22119424, 'steps': 43201, 'loss/train': 1.9860972166061401}
+02/26/2022 09:31:41 - INFO - codeparrot_training - Step 43202: {'lr': 2.433959240307135e-05, 'samples': 22119936, 'steps': 43202, 'loss/train': 1.9324525594711304}
+02/26/2022 09:31:47 - INFO - codeparrot_training - Step 43203: {'lr': 2.4332550598310743e-05, 'samples': 22120448, 'steps': 43203, 'loss/train': 1.4562253952026367}
+02/26/2022 09:31:50 - INFO - codeparrot_training - Step 43204: {'lr': 2.4325509760237895e-05, 'samples': 22120960, 'steps': 43204, 'loss/train': 2.2617857456207275}
+02/26/2022 09:31:56 - INFO - codeparrot_training - Step 43205: {'lr': 2.4318469888882893e-05, 'samples': 22121472, 'steps': 43205, 'loss/train': 0.6300603747367859}
+02/26/2022 09:31:59 - INFO - codeparrot_training - Step 43206: {'lr': 2.4311430984275935e-05, 'samples': 22121984, 'steps': 43206, 'loss/train': 1.659628987312317}
+02/26/2022 09:32:06 - INFO - codeparrot_training - Step 43207: {'lr': 2.4304393046447077e-05, 'samples': 22122496, 'steps': 43207, 'loss/train': 0.9781558513641357}
+02/26/2022 09:32:10 - INFO - codeparrot_training - Step 43208: {'lr': 2.4297356075426575e-05, 'samples': 22123008, 'steps': 43208, 'loss/train': 1.295369029045105}
+02/26/2022 09:32:15 - INFO - codeparrot_training - Step 43209: {'lr': 2.4290320071244515e-05, 'samples': 22123520, 'steps': 43209, 'loss/train': 1.990005612373352}
+02/26/2022 09:32:19 - INFO - codeparrot_training - Step 43210: {'lr': 2.4283285033931097e-05, 'samples': 22124032, 'steps': 43210, 'loss/train': 0.5798360705375671}
+02/26/2022 09:32:24 - INFO - codeparrot_training - Step 43211: {'lr': 2.4276250963516322e-05, 'samples': 22124544, 'steps': 43211, 'loss/train': 1.4787448644638062}
+02/26/2022 09:32:28 - INFO - codeparrot_training - Step 43212: {'lr': 2.4269217860030473e-05, 'samples': 22125056, 'steps': 43212, 'loss/train': 1.4362761974334717}
+02/26/2022 09:32:33 - INFO - codeparrot_training - Step 43213: {'lr': 2.426218572350364e-05, 'samples': 22125568, 'steps': 43213, 'loss/train': 1.7281453609466553}
+02/26/2022 09:32:37 - INFO - codeparrot_training - Step 43214: {'lr': 2.4255154553965902e-05, 'samples': 22126080, 'steps': 43214, 'loss/train': 1.4424231052398682}
+02/26/2022 09:32:44 - INFO - codeparrot_training - Step 43215: {'lr': 2.4248124351447354e-05, 'samples': 22126592, 'steps': 43215, 'loss/train': 1.3509424924850464}
+02/26/2022 09:32:48 - INFO - codeparrot_training - Step 43216: {'lr': 2.4241095115978218e-05, 'samples': 22127104, 'steps': 43216, 'loss/train': 1.6565086841583252}
+02/26/2022 09:32:53 - INFO - codeparrot_training - Step 43217: {'lr': 2.4234066847588526e-05, 'samples': 22127616, 'steps': 43217, 'loss/train': 1.8656890392303467}
+02/26/2022 09:32:56 - INFO - codeparrot_training - Step 43218: {'lr': 2.4227039546308394e-05, 'samples': 22128128, 'steps': 43218, 'loss/train': 2.865680456161499}
+02/26/2022 09:33:02 - INFO - codeparrot_training - Step 43219: {'lr': 2.4220013212167907e-05, 'samples': 22128640, 'steps': 43219, 'loss/train': 2.386889934539795}
+02/26/2022 09:33:06 - INFO - codeparrot_training - Step 43220: {'lr': 2.421298784519724e-05, 'samples': 22129152, 'steps': 43220, 'loss/train': 0.08196766674518585}
+02/26/2022 09:33:11 - INFO - codeparrot_training - Step 43221: {'lr': 2.4205963445426414e-05, 'samples': 22129664, 'steps': 43221, 'loss/train': 1.9377459287643433}
+02/26/2022 09:33:15 - INFO - codeparrot_training - Step 43222: {'lr': 2.4198940012885583e-05, 'samples': 22130176, 'steps': 43222, 'loss/train': 1.8808695077896118}
+02/26/2022 09:33:20 - INFO - codeparrot_training - Step 43223: {'lr': 2.4191917547604748e-05, 'samples': 22130688, 'steps': 43223, 'loss/train': 1.5802565813064575}
+02/26/2022 09:33:24 - INFO - codeparrot_training - Step 43224: {'lr': 2.4184896049614018e-05, 'samples': 22131200, 'steps': 43224, 'loss/train': 2.7260637283325195}
+02/26/2022 09:33:29 - INFO - codeparrot_training - Step 43225: {'lr': 2.4177875518943516e-05, 'samples': 22131712, 'steps': 43225, 'loss/train': 3.524568557739258}
+02/26/2022 09:33:33 - INFO - codeparrot_training - Step 43226: {'lr': 2.4170855955623265e-05, 'samples': 22132224, 'steps': 43226, 'loss/train': 1.9323642253875732}
+02/26/2022 09:33:40 - INFO - codeparrot_training - Step 43227: {'lr': 2.4163837359683442e-05, 'samples': 22132736, 'steps': 43227, 'loss/train': 0.7207525372505188}
+02/26/2022 09:33:43 - INFO - codeparrot_training - Step 43228: {'lr': 2.4156819731153938e-05, 'samples': 22133248, 'steps': 43228, 'loss/train': 1.1741726398468018}
+02/26/2022 09:33:49 - INFO - codeparrot_training - Step 43229: {'lr': 2.414980307006495e-05, 'samples': 22133760, 'steps': 43229, 'loss/train': 2.3310253620147705}
+02/26/2022 09:33:52 - INFO - codeparrot_training - Step 43230: {'lr': 2.414278737644643e-05, 'samples': 22134272, 'steps': 43230, 'loss/train': 1.921450138092041}
+02/26/2022 09:33:58 - INFO - codeparrot_training - Step 43231: {'lr': 2.4135772650328597e-05, 'samples': 22134784, 'steps': 43231, 'loss/train': 1.8129687309265137}
+02/26/2022 09:34:01 - INFO - codeparrot_training - Step 43232: {'lr': 2.412875889174129e-05, 'samples': 22135296, 'steps': 43232, 'loss/train': 2.041994571685791}
+02/26/2022 09:34:07 - INFO - codeparrot_training - Step 43233: {'lr': 2.4121746100714713e-05, 'samples': 22135808, 'steps': 43233, 'loss/train': 2.628507614135742}
+02/26/2022 09:34:10 - INFO - codeparrot_training - Step 43234: {'lr': 2.411473427727881e-05, 'samples': 22136320, 'steps': 43234, 'loss/train': 2.3726131916046143}
+02/26/2022 09:34:16 - INFO - codeparrot_training - Step 43235: {'lr': 2.410772342146375e-05, 'samples': 22136832, 'steps': 43235, 'loss/train': 1.8973512649536133}
+02/26/2022 09:34:19 - INFO - codeparrot_training - Step 43236: {'lr': 2.4100713533299374e-05, 'samples': 22137344, 'steps': 43236, 'loss/train': 1.4323657751083374}
+02/26/2022 09:34:25 - INFO - codeparrot_training - Step 43237: {'lr': 2.4093704612815876e-05, 'samples': 22137856, 'steps': 43237, 'loss/train': 0.1868385225534439}
+02/26/2022 09:34:28 - INFO - codeparrot_training - Step 43238: {'lr': 2.408669666004315e-05, 'samples': 22138368, 'steps': 43238, 'loss/train': 1.523546576499939}
+02/26/2022 09:34:34 - INFO - codeparrot_training - Step 43239: {'lr': 2.4079689675011312e-05, 'samples': 22138880, 'steps': 43239, 'loss/train': 0.7568512558937073}
+02/26/2022 09:34:38 - INFO - codeparrot_training - Step 43240: {'lr': 2.4072683657750365e-05, 'samples': 22139392, 'steps': 43240, 'loss/train': 1.989466905593872}
+02/26/2022 09:34:43 - INFO - codeparrot_training - Step 43241: {'lr': 2.4065678608290313e-05, 'samples': 22139904, 'steps': 43241, 'loss/train': 1.2015511989593506}
+02/26/2022 09:34:47 - INFO - codeparrot_training - Step 43242: {'lr': 2.4058674526661078e-05, 'samples': 22140416, 'steps': 43242, 'loss/train': 1.7314058542251587}
+02/26/2022 09:34:54 - INFO - codeparrot_training - Step 43243: {'lr': 2.40516714128928e-05, 'samples': 22140928, 'steps': 43243, 'loss/train': 2.0399346351623535}
+02/26/2022 09:34:57 - INFO - codeparrot_training - Step 43244: {'lr': 2.4044669267015402e-05, 'samples': 22141440, 'steps': 43244, 'loss/train': 1.6607333421707153}
+02/26/2022 09:35:03 - INFO - codeparrot_training - Step 43245: {'lr': 2.4037668089058912e-05, 'samples': 22141952, 'steps': 43245, 'loss/train': 0.6854348182678223}
+02/26/2022 09:35:06 - INFO - codeparrot_training - Step 43246: {'lr': 2.4030667879053225e-05, 'samples': 22142464, 'steps': 43246, 'loss/train': 1.3459415435791016}
+02/26/2022 09:35:12 - INFO - codeparrot_training - Step 43247: {'lr': 2.4023668637028483e-05, 'samples': 22142976, 'steps': 43247, 'loss/train': 1.8875693082809448}
+02/26/2022 09:35:15 - INFO - codeparrot_training - Step 43248: {'lr': 2.4016670363014582e-05, 'samples': 22143488, 'steps': 43248, 'loss/train': 1.2095513343811035}
+02/26/2022 09:35:21 - INFO - codeparrot_training - Step 43249: {'lr': 2.4009673057041493e-05, 'samples': 22144000, 'steps': 43249, 'loss/train': 1.6082024574279785}
+02/26/2022 09:35:24 - INFO - codeparrot_training - Step 43250: {'lr': 2.4002676719139166e-05, 'samples': 22144512, 'steps': 43250, 'loss/train': 1.9440619945526123}
+02/26/2022 09:35:30 - INFO - codeparrot_training - Step 43251: {'lr': 2.399568134933766e-05, 'samples': 22145024, 'steps': 43251, 'loss/train': 1.5103626251220703}
+02/26/2022 09:35:33 - INFO - codeparrot_training - Step 43252: {'lr': 2.3988686947666815e-05, 'samples': 22145536, 'steps': 43252, 'loss/train': 1.9341816902160645}
+02/26/2022 09:35:41 - INFO - codeparrot_training - Step 43253: {'lr': 2.39816935141568e-05, 'samples': 22146048, 'steps': 43253, 'loss/train': 1.6588099002838135}
+02/26/2022 09:35:44 - INFO - codeparrot_training - Step 43254: {'lr': 2.397470104883734e-05, 'samples': 22146560, 'steps': 43254, 'loss/train': 0.36438480019569397}
+02/26/2022 09:35:50 - INFO - codeparrot_training - Step 43255: {'lr': 2.396770955173852e-05, 'samples': 22147072, 'steps': 43255, 'loss/train': 2.2409348487854004}
+02/26/2022 09:35:53 - INFO - codeparrot_training - Step 43256: {'lr': 2.3960719022890264e-05, 'samples': 22147584, 'steps': 43256, 'loss/train': 1.9302009344100952}
+02/26/2022 09:35:59 - INFO - codeparrot_training - Step 43257: {'lr': 2.3953729462322522e-05, 'samples': 22148096, 'steps': 43257, 'loss/train': 1.5525544881820679}
+02/26/2022 09:36:02 - INFO - codeparrot_training - Step 43258: {'lr': 2.394674087006518e-05, 'samples': 22148608, 'steps': 43258, 'loss/train': 1.984934687614441}
+02/26/2022 09:36:08 - INFO - codeparrot_training - Step 43259: {'lr': 2.3939753246148248e-05, 'samples': 22149120, 'steps': 43259, 'loss/train': 1.010594129562378}
+02/26/2022 09:36:11 - INFO - codeparrot_training - Step 43260: {'lr': 2.3932766590601647e-05, 'samples': 22149632, 'steps': 43260, 'loss/train': 1.0503628253936768}
+02/26/2022 09:36:17 - INFO - codeparrot_training - Step 43261: {'lr': 2.392578090345529e-05, 'samples': 22150144, 'steps': 43261, 'loss/train': 1.0467463731765747}
+02/26/2022 09:36:20 - INFO - codeparrot_training - Step 43262: {'lr': 2.391879618473905e-05, 'samples': 22150656, 'steps': 43262, 'loss/train': 2.2039053440093994}
+02/26/2022 09:36:28 - INFO - codeparrot_training - Step 43263: {'lr': 2.3911812434482928e-05, 'samples': 22151168, 'steps': 43263, 'loss/train': 2.0943071842193604}
+02/26/2022 09:36:31 - INFO - codeparrot_training - Step 43264: {'lr': 2.390482965271684e-05, 'samples': 22151680, 'steps': 43264, 'loss/train': 1.0844749212265015}
+02/26/2022 09:36:37 - INFO - codeparrot_training - Step 43265: {'lr': 2.3897847839470605e-05, 'samples': 22152192, 'steps': 43265, 'loss/train': 1.2232099771499634}
+02/26/2022 09:36:40 - INFO - codeparrot_training - Step 43266: {'lr': 2.3890866994774245e-05, 'samples': 22152704, 'steps': 43266, 'loss/train': 0.8230526447296143}
+02/26/2022 09:36:46 - INFO - codeparrot_training - Step 43267: {'lr': 2.3883887118657604e-05, 'samples': 22153216, 'steps': 43267, 'loss/train': 2.3939335346221924}
+02/26/2022 09:36:49 - INFO - codeparrot_training - Step 43268: {'lr': 2.38769082111506e-05, 'samples': 22153728, 'steps': 43268, 'loss/train': 2.4508931636810303}
+02/26/2022 09:36:55 - INFO - codeparrot_training - Step 43269: {'lr': 2.386993027228307e-05, 'samples': 22154240, 'steps': 43269, 'loss/train': 2.039569139480591}
+02/26/2022 09:36:58 - INFO - codeparrot_training - Step 43270: {'lr': 2.386295330208507e-05, 'samples': 22154752, 'steps': 43270, 'loss/train': 1.6284921169281006}
+02/26/2022 09:37:04 - INFO - codeparrot_training - Step 43271: {'lr': 2.385597730058625e-05, 'samples': 22155264, 'steps': 43271, 'loss/train': 1.267389178276062}
+02/26/2022 09:37:07 - INFO - codeparrot_training - Step 43272: {'lr': 2.3849002267816667e-05, 'samples': 22155776, 'steps': 43272, 'loss/train': 0.8139084577560425}
+02/26/2022 09:37:15 - INFO - codeparrot_training - Step 43273: {'lr': 2.3842028203806098e-05, 'samples': 22156288, 'steps': 43273, 'loss/train': 1.2592933177947998}
+02/26/2022 09:37:18 - INFO - codeparrot_training - Step 43274: {'lr': 2.383505510858458e-05, 'samples': 22156800, 'steps': 43274, 'loss/train': 1.1139179468154907}
+02/26/2022 09:37:24 - INFO - codeparrot_training - Step 43275: {'lr': 2.3828082982181754e-05, 'samples': 22157312, 'steps': 43275, 'loss/train': 1.969099521636963}
+02/26/2022 09:37:27 - INFO - codeparrot_training - Step 43276: {'lr': 2.382111182462765e-05, 'samples': 22157824, 'steps': 43276, 'loss/train': 1.7925219535827637}
+02/26/2022 09:37:33 - INFO - codeparrot_training - Step 43277: {'lr': 2.3814141635952054e-05, 'samples': 22158336, 'steps': 43277, 'loss/train': 2.275580644607544}
+02/26/2022 09:37:36 - INFO - codeparrot_training - Step 43278: {'lr': 2.380717241618488e-05, 'samples': 22158848, 'steps': 43278, 'loss/train': 0.42239734530448914}
+02/26/2022 09:37:42 - INFO - codeparrot_training - Step 43279: {'lr': 2.380020416535597e-05, 'samples': 22159360, 'steps': 43279, 'loss/train': 1.710526943206787}
+02/26/2022 09:37:46 - INFO - codeparrot_training - Step 43280: {'lr': 2.379323688349516e-05, 'samples': 22159872, 'steps': 43280, 'loss/train': 0.8309153318405151}
+02/26/2022 09:37:51 - INFO - codeparrot_training - Step 43281: {'lr': 2.3786270570632233e-05, 'samples': 22160384, 'steps': 43281, 'loss/train': 2.502852439880371}
+02/26/2022 09:37:55 - INFO - codeparrot_training - Step 43282: {'lr': 2.377930522679714e-05, 'samples': 22160896, 'steps': 43282, 'loss/train': 0.5534853339195251}
+02/26/2022 09:38:00 - INFO - codeparrot_training - Step 43283: {'lr': 2.377234085201968e-05, 'samples': 22161408, 'steps': 43283, 'loss/train': 0.08750396966934204}
+02/26/2022 09:38:04 - INFO - codeparrot_training - Step 43284: {'lr': 2.376537744632967e-05, 'samples': 22161920, 'steps': 43284, 'loss/train': 0.8126357793807983}
+02/26/2022 09:38:09 - INFO - codeparrot_training - Step 43285: {'lr': 2.3758415009756895e-05, 'samples': 22162432, 'steps': 43285, 'loss/train': 1.3534290790557861}
+02/26/2022 09:38:13 - INFO - codeparrot_training - Step 43286: {'lr': 2.3751453542331296e-05, 'samples': 22162944, 'steps': 43286, 'loss/train': 1.7971128225326538}
+02/26/2022 09:38:18 - INFO - codeparrot_training - Step 43287: {'lr': 2.3744493044082576e-05, 'samples': 22163456, 'steps': 43287, 'loss/train': 2.2954046726226807}
+02/26/2022 09:38:22 - INFO - codeparrot_training - Step 43288: {'lr': 2.3737533515040682e-05, 'samples': 22163968, 'steps': 43288, 'loss/train': 1.7782648801803589}
+02/26/2022 09:38:29 - INFO - codeparrot_training - Step 43289: {'lr': 2.3730574955235255e-05, 'samples': 22164480, 'steps': 43289, 'loss/train': 1.3095701932907104}
+02/26/2022 09:38:32 - INFO - codeparrot_training - Step 43290: {'lr': 2.3723617364696247e-05, 'samples': 22164992, 'steps': 43290, 'loss/train': 2.352405548095703}
+02/26/2022 09:38:38 - INFO - codeparrot_training - Step 43291: {'lr': 2.3716660743453384e-05, 'samples': 22165504, 'steps': 43291, 'loss/train': 1.0517081022262573}
+02/26/2022 09:38:41 - INFO - codeparrot_training - Step 43292: {'lr': 2.3709705091536555e-05, 'samples': 22166016, 'steps': 43292, 'loss/train': 0.9705249667167664}
+02/26/2022 09:38:47 - INFO - codeparrot_training - Step 43293: {'lr': 2.370275040897543e-05, 'samples': 22166528, 'steps': 43293, 'loss/train': 1.9242444038391113}
+02/26/2022 09:38:50 - INFO - codeparrot_training - Step 43294: {'lr': 2.369579669579988e-05, 'samples': 22167040, 'steps': 43294, 'loss/train': 1.4593243598937988}
+02/26/2022 09:38:56 - INFO - codeparrot_training - Step 43295: {'lr': 2.3688843952039656e-05, 'samples': 22167552, 'steps': 43295, 'loss/train': 2.6275806427001953}
+02/26/2022 09:38:59 - INFO - codeparrot_training - Step 43296: {'lr': 2.3681892177724652e-05, 'samples': 22168064, 'steps': 43296, 'loss/train': 0.8268952369689941}
+02/26/2022 09:39:05 - INFO - codeparrot_training - Step 43297: {'lr': 2.3674941372884478e-05, 'samples': 22168576, 'steps': 43297, 'loss/train': 1.5829060077667236}
+02/26/2022 09:39:08 - INFO - codeparrot_training - Step 43298: {'lr': 2.3667991537549006e-05, 'samples': 22169088, 'steps': 43298, 'loss/train': 1.8122408390045166}
+02/26/2022 09:39:15 - INFO - codeparrot_training - Step 43299: {'lr': 2.3661042671747957e-05, 'samples': 22169600, 'steps': 43299, 'loss/train': 1.3882919549942017}
+02/26/2022 09:39:19 - INFO - codeparrot_training - Step 43300: {'lr': 2.365409477551117e-05, 'samples': 22170112, 'steps': 43300, 'loss/train': 1.8376020193099976}
+02/26/2022 09:39:24 - INFO - codeparrot_training - Step 43301: {'lr': 2.364714784886837e-05, 'samples': 22170624, 'steps': 43301, 'loss/train': 2.0100438594818115}
+02/26/2022 09:39:28 - INFO - codeparrot_training - Step 43302: {'lr': 2.3640201891849315e-05, 'samples': 22171136, 'steps': 43302, 'loss/train': 2.088324546813965}
+02/26/2022 09:39:33 - INFO - codeparrot_training - Step 43303: {'lr': 2.3633256904483786e-05, 'samples': 22171648, 'steps': 43303, 'loss/train': 1.601377248764038}
+02/26/2022 09:39:37 - INFO - codeparrot_training - Step 43304: {'lr': 2.3626312886801423e-05, 'samples': 22172160, 'steps': 43304, 'loss/train': 2.2252700328826904}
+02/26/2022 09:39:43 - INFO - codeparrot_training - Step 43305: {'lr': 2.3619369838832118e-05, 'samples': 22172672, 'steps': 43305, 'loss/train': 3.0573503971099854}
+02/26/2022 09:39:46 - INFO - codeparrot_training - Step 43306: {'lr': 2.3612427760605542e-05, 'samples': 22173184, 'steps': 43306, 'loss/train': 1.2639302015304565}
+02/26/2022 09:39:52 - INFO - codeparrot_training - Step 43307: {'lr': 2.3605486652151426e-05, 'samples': 22173696, 'steps': 43307, 'loss/train': 1.4656106233596802}
+02/26/2022 09:39:55 - INFO - codeparrot_training - Step 43308: {'lr': 2.359854651349949e-05, 'samples': 22174208, 'steps': 43308, 'loss/train': 2.149136543273926}
+02/26/2022 09:40:02 - INFO - codeparrot_training - Step 43309: {'lr': 2.359160734467952e-05, 'samples': 22174720, 'steps': 43309, 'loss/train': 3.0191521644592285}
+02/26/2022 09:40:06 - INFO - codeparrot_training - Step 43310: {'lr': 2.3584669145721217e-05, 'samples': 22175232, 'steps': 43310, 'loss/train': 1.396069049835205}
+02/26/2022 09:40:11 - INFO - codeparrot_training - Step 43311: {'lr': 2.35777319166543e-05, 'samples': 22175744, 'steps': 43311, 'loss/train': 2.0735700130462646}
+02/26/2022 09:40:15 - INFO - codeparrot_training - Step 43312: {'lr': 2.3570795657508442e-05, 'samples': 22176256, 'steps': 43312, 'loss/train': 2.6623117923736572}
+02/26/2022 09:40:21 - INFO - codeparrot_training - Step 43313: {'lr': 2.356386036831343e-05, 'samples': 22176768, 'steps': 43313, 'loss/train': 2.631047487258911}
+02/26/2022 09:40:24 - INFO - codeparrot_training - Step 43314: {'lr': 2.3556926049098952e-05, 'samples': 22177280, 'steps': 43314, 'loss/train': 1.9996322393417358}
+02/26/2022 09:40:28 - INFO - codeparrot_training - Step 43315: {'lr': 2.354999269989469e-05, 'samples': 22177792, 'steps': 43315, 'loss/train': 8.571395874023438}
+02/26/2022 09:40:33 - INFO - codeparrot_training - Step 43316: {'lr': 2.354306032073031e-05, 'samples': 22178304, 'steps': 43316, 'loss/train': 1.7630122900009155}
+02/26/2022 09:40:37 - INFO - codeparrot_training - Step 43317: {'lr': 2.353612891163559e-05, 'samples': 22178816, 'steps': 43317, 'loss/train': 0.7591730356216431}
+02/26/2022 09:40:42 - INFO - codeparrot_training - Step 43318: {'lr': 2.3529198472640206e-05, 'samples': 22179328, 'steps': 43318, 'loss/train': 1.7589277029037476}
+02/26/2022 09:40:46 - INFO - codeparrot_training - Step 43319: {'lr': 2.3522269003773798e-05, 'samples': 22179840, 'steps': 43319, 'loss/train': 1.097955584526062}
+02/26/2022 09:40:53 - INFO - codeparrot_training - Step 43320: {'lr': 2.3515340505066042e-05, 'samples': 22180352, 'steps': 43320, 'loss/train': 2.1457056999206543}
+02/26/2022 09:40:56 - INFO - codeparrot_training - Step 43321: {'lr': 2.3508412976546683e-05, 'samples': 22180864, 'steps': 43321, 'loss/train': 1.7428361177444458}
+02/26/2022 09:41:02 - INFO - codeparrot_training - Step 43322: {'lr': 2.350148641824537e-05, 'samples': 22181376, 'steps': 43322, 'loss/train': 2.8708314895629883}
+02/26/2022 09:41:05 - INFO - codeparrot_training - Step 43323: {'lr': 2.3494560830191775e-05, 'samples': 22181888, 'steps': 43323, 'loss/train': 0.7016080617904663}
+02/26/2022 09:41:11 - INFO - codeparrot_training - Step 43324: {'lr': 2.3487636212415508e-05, 'samples': 22182400, 'steps': 43324, 'loss/train': 1.868202567100525}
+02/26/2022 09:41:14 - INFO - codeparrot_training - Step 43325: {'lr': 2.3480712564946327e-05, 'samples': 22182912, 'steps': 43325, 'loss/train': 1.9066132307052612}
+02/26/2022 09:41:20 - INFO - codeparrot_training - Step 43326: {'lr': 2.347378988781379e-05, 'samples': 22183424, 'steps': 43326, 'loss/train': 3.118809700012207}
+02/26/2022 09:41:23 - INFO - codeparrot_training - Step 43327: {'lr': 2.346686818104771e-05, 'samples': 22183936, 'steps': 43327, 'loss/train': 1.613738775253296}
+02/26/2022 09:41:29 - INFO - codeparrot_training - Step 43328: {'lr': 2.3459947444677553e-05, 'samples': 22184448, 'steps': 43328, 'loss/train': 2.110063314437866}
+02/26/2022 09:41:32 - INFO - codeparrot_training - Step 43329: {'lr': 2.3453027678733085e-05, 'samples': 22184960, 'steps': 43329, 'loss/train': 1.4017025232315063}
+02/26/2022 09:41:40 - INFO - codeparrot_training - Step 43330: {'lr': 2.3446108883243834e-05, 'samples': 22185472, 'steps': 43330, 'loss/train': 2.2838540077209473}
+02/26/2022 09:41:43 - INFO - codeparrot_training - Step 43331: {'lr': 2.3439191058239634e-05, 'samples': 22185984, 'steps': 43331, 'loss/train': 0.2765112817287445}
+02/26/2022 09:41:49 - INFO - codeparrot_training - Step 43332: {'lr': 2.3432274203749908e-05, 'samples': 22186496, 'steps': 43332, 'loss/train': 1.5505725145339966}
+02/26/2022 09:41:52 - INFO - codeparrot_training - Step 43333: {'lr': 2.3425358319804414e-05, 'samples': 22187008, 'steps': 43333, 'loss/train': 1.612720012664795}
+02/26/2022 09:41:58 - INFO - codeparrot_training - Step 43334: {'lr': 2.3418443406432704e-05, 'samples': 22187520, 'steps': 43334, 'loss/train': 2.1533167362213135}
+02/26/2022 09:42:01 - INFO - codeparrot_training - Step 43335: {'lr': 2.3411529463664484e-05, 'samples': 22188032, 'steps': 43335, 'loss/train': 2.0199599266052246}
+02/26/2022 09:42:07 - INFO - codeparrot_training - Step 43336: {'lr': 2.3404616491529306e-05, 'samples': 22188544, 'steps': 43336, 'loss/train': 1.377267837524414}
+02/26/2022 09:42:10 - INFO - codeparrot_training - Step 43337: {'lr': 2.3397704490056792e-05, 'samples': 22189056, 'steps': 43337, 'loss/train': 0.9962981343269348}
+02/26/2022 09:42:16 - INFO - codeparrot_training - Step 43338: {'lr': 2.3390793459276528e-05, 'samples': 22189568, 'steps': 43338, 'loss/train': 8.646394729614258}
+02/26/2022 09:42:19 - INFO - codeparrot_training - Step 43339: {'lr': 2.338388339921818e-05, 'samples': 22190080, 'steps': 43339, 'loss/train': 1.412485957145691}
+02/26/2022 09:42:25 - INFO - codeparrot_training - Step 43340: {'lr': 2.3376974309911343e-05, 'samples': 22190592, 'steps': 43340, 'loss/train': 1.7455847263336182}
+02/26/2022 09:42:28 - INFO - codeparrot_training - Step 43341: {'lr': 2.337006619138557e-05, 'samples': 22191104, 'steps': 43341, 'loss/train': 2.3395497798919678}
+02/26/2022 09:42:34 - INFO - codeparrot_training - Step 43342: {'lr': 2.3363159043670425e-05, 'samples': 22191616, 'steps': 43342, 'loss/train': 0.5850828289985657}
+02/26/2022 09:42:37 - INFO - codeparrot_training - Step 43343: {'lr': 2.3356252866795575e-05, 'samples': 22192128, 'steps': 43343, 'loss/train': 1.5073628425598145}
+02/26/2022 09:42:43 - INFO - codeparrot_training - Step 43344: {'lr': 2.334934766079058e-05, 'samples': 22192640, 'steps': 43344, 'loss/train': 1.417761206626892}
+02/26/2022 09:42:46 - INFO - codeparrot_training - Step 43345: {'lr': 2.3342443425685006e-05, 'samples': 22193152, 'steps': 43345, 'loss/train': 1.6512362957000732}
+02/26/2022 09:42:54 - INFO - codeparrot_training - Step 43346: {'lr': 2.3335540161508407e-05, 'samples': 22193664, 'steps': 43346, 'loss/train': 1.7044094800949097}
+02/26/2022 09:42:57 - INFO - codeparrot_training - Step 43347: {'lr': 2.3328637868290398e-05, 'samples': 22194176, 'steps': 43347, 'loss/train': 0.9756754636764526}
+02/26/2022 09:43:03 - INFO - codeparrot_training - Step 43348: {'lr': 2.332173654606054e-05, 'samples': 22194688, 'steps': 43348, 'loss/train': 2.291933059692383}
+02/26/2022 09:43:07 - INFO - codeparrot_training - Step 43349: {'lr': 2.3314836194848394e-05, 'samples': 22195200, 'steps': 43349, 'loss/train': 1.176311731338501}
+02/26/2022 09:43:12 - INFO - codeparrot_training - Step 43350: {'lr': 2.330793681468352e-05, 'samples': 22195712, 'steps': 43350, 'loss/train': 1.9870012998580933}
+02/26/2022 09:43:15 - INFO - codeparrot_training - Step 43351: {'lr': 2.3301038405595388e-05, 'samples': 22196224, 'steps': 43351, 'loss/train': 1.7480851411819458}
+02/26/2022 09:43:21 - INFO - codeparrot_training - Step 43352: {'lr': 2.3294140967613675e-05, 'samples': 22196736, 'steps': 43352, 'loss/train': 2.084657669067383}
+02/26/2022 09:43:24 - INFO - codeparrot_training - Step 43353: {'lr': 2.3287244500767884e-05, 'samples': 22197248, 'steps': 43353, 'loss/train': 0.8548081517219543}
+02/26/2022 09:43:30 - INFO - codeparrot_training - Step 43354: {'lr': 2.328034900508752e-05, 'samples': 22197760, 'steps': 43354, 'loss/train': 0.9051329493522644}
+02/26/2022 09:43:34 - INFO - codeparrot_training - Step 43355: {'lr': 2.327345448060214e-05, 'samples': 22198272, 'steps': 43355, 'loss/train': 0.10485494881868362}
+02/26/2022 09:43:41 - INFO - codeparrot_training - Step 43356: {'lr': 2.3266560927341306e-05, 'samples': 22198784, 'steps': 43356, 'loss/train': 1.3880500793457031}
+02/26/2022 09:43:44 - INFO - codeparrot_training - Step 43357: {'lr': 2.325966834533455e-05, 'samples': 22199296, 'steps': 43357, 'loss/train': 1.5446600914001465}
+02/26/2022 09:43:50 - INFO - codeparrot_training - Step 43358: {'lr': 2.3252776734611377e-05, 'samples': 22199808, 'steps': 43358, 'loss/train': 2.0897037982940674}
+02/26/2022 09:43:53 - INFO - codeparrot_training - Step 43359: {'lr': 2.324588609520123e-05, 'samples': 22200320, 'steps': 43359, 'loss/train': 0.7792462110519409}
+02/26/2022 09:43:59 - INFO - codeparrot_training - Step 43360: {'lr': 2.3238996427133783e-05, 'samples': 22200832, 'steps': 43360, 'loss/train': 1.8640575408935547}
+02/26/2022 09:44:02 - INFO - codeparrot_training - Step 43361: {'lr': 2.3232107730438408e-05, 'samples': 22201344, 'steps': 43361, 'loss/train': 1.979748010635376}
+02/26/2022 09:44:08 - INFO - codeparrot_training - Step 43362: {'lr': 2.3225220005144763e-05, 'samples': 22201856, 'steps': 43362, 'loss/train': 1.7013325691223145}
+02/26/2022 09:44:12 - INFO - codeparrot_training - Step 43363: {'lr': 2.3218333251282198e-05, 'samples': 22202368, 'steps': 43363, 'loss/train': 0.49709972739219666}
+02/26/2022 09:44:17 - INFO - codeparrot_training - Step 43364: {'lr': 2.321144746888032e-05, 'samples': 22202880, 'steps': 43364, 'loss/train': 1.5686622858047485}
+02/26/2022 09:44:20 - INFO - codeparrot_training - Step 43365: {'lr': 2.3204562657968526e-05, 'samples': 22203392, 'steps': 43365, 'loss/train': 0.9851326942443848}
+02/26/2022 09:44:28 - INFO - codeparrot_training - Step 43366: {'lr': 2.3197678818576485e-05, 'samples': 22203904, 'steps': 43366, 'loss/train': 1.5127288103103638}
+02/26/2022 09:44:31 - INFO - codeparrot_training - Step 43367: {'lr': 2.3190795950733478e-05, 'samples': 22204416, 'steps': 43367, 'loss/train': 2.011246919631958}
+02/26/2022 09:44:37 - INFO - codeparrot_training - Step 43368: {'lr': 2.3183914054469096e-05, 'samples': 22204928, 'steps': 43368, 'loss/train': 0.7503435015678406}
+02/26/2022 09:44:40 - INFO - codeparrot_training - Step 43369: {'lr': 2.3177033129812785e-05, 'samples': 22205440, 'steps': 43369, 'loss/train': 2.5692148208618164}
+02/26/2022 09:44:46 - INFO - codeparrot_training - Step 43370: {'lr': 2.3170153176794135e-05, 'samples': 22205952, 'steps': 43370, 'loss/train': 0.31932225823402405}
+02/26/2022 09:44:49 - INFO - codeparrot_training - Step 43371: {'lr': 2.3163274195442425e-05, 'samples': 22206464, 'steps': 43371, 'loss/train': 1.7933956384658813}
+02/26/2022 09:44:55 - INFO - codeparrot_training - Step 43372: {'lr': 2.3156396185787244e-05, 'samples': 22206976, 'steps': 43372, 'loss/train': 1.6930683851242065}
+02/26/2022 09:44:58 - INFO - codeparrot_training - Step 43373: {'lr': 2.314951914785801e-05, 'samples': 22207488, 'steps': 43373, 'loss/train': 2.106527090072632}
+02/26/2022 09:45:04 - INFO - codeparrot_training - Step 43374: {'lr': 2.3142643081684234e-05, 'samples': 22208000, 'steps': 43374, 'loss/train': 0.9784080386161804}
+02/26/2022 09:45:09 - INFO - codeparrot_training - Step 43375: {'lr': 2.313576798729536e-05, 'samples': 22208512, 'steps': 43375, 'loss/train': 1.622563123703003}
+02/26/2022 09:45:13 - INFO - codeparrot_training - Step 43376: {'lr': 2.312889386472078e-05, 'samples': 22209024, 'steps': 43376, 'loss/train': 1.9017900228500366}
+02/26/2022 09:45:19 - INFO - codeparrot_training - Step 43377: {'lr': 2.312202071398997e-05, 'samples': 22209536, 'steps': 43377, 'loss/train': 0.9812229871749878}
+02/26/2022 09:45:22 - INFO - codeparrot_training - Step 43378: {'lr': 2.3115148535132414e-05, 'samples': 22210048, 'steps': 43378, 'loss/train': 0.97850501537323}
+02/26/2022 09:45:28 - INFO - codeparrot_training - Step 43379: {'lr': 2.3108277328177522e-05, 'samples': 22210560, 'steps': 43379, 'loss/train': 1.0578759908676147}
+02/26/2022 09:45:31 - INFO - codeparrot_training - Step 43380: {'lr': 2.3101407093154726e-05, 'samples': 22211072, 'steps': 43380, 'loss/train': 1.8145272731781006}
+02/26/2022 09:45:35 - INFO - codeparrot_training - Step 43381: {'lr': 2.309453783009338e-05, 'samples': 22211584, 'steps': 43381, 'loss/train': 1.930153489112854}
+02/26/2022 09:45:42 - INFO - codeparrot_training - Step 43382: {'lr': 2.3087669539023054e-05, 'samples': 22212096, 'steps': 43382, 'loss/train': 1.6122373342514038}
+02/26/2022 09:45:45 - INFO - codeparrot_training - Step 43383: {'lr': 2.3080802219973108e-05, 'samples': 22212608, 'steps': 43383, 'loss/train': 1.7784168720245361}
+02/26/2022 09:45:51 - INFO - codeparrot_training - Step 43384: {'lr': 2.3073935872972933e-05, 'samples': 22213120, 'steps': 43384, 'loss/train': 1.5295833349227905}
+02/26/2022 09:45:54 - INFO - codeparrot_training - Step 43385: {'lr': 2.306707049805193e-05, 'samples': 22213632, 'steps': 43385, 'loss/train': 1.7486096620559692}
+02/26/2022 09:46:01 - INFO - codeparrot_training - Step 43386: {'lr': 2.3060206095239568e-05, 'samples': 22214144, 'steps': 43386, 'loss/train': 2.384612798690796}
+02/26/2022 09:46:05 - INFO - codeparrot_training - Step 43387: {'lr': 2.3053342664565215e-05, 'samples': 22214656, 'steps': 43387, 'loss/train': 2.0957627296447754}
+02/26/2022 09:46:10 - INFO - codeparrot_training - Step 43388: {'lr': 2.304648020605829e-05, 'samples': 22215168, 'steps': 43388, 'loss/train': 2.064143657684326}
+02/26/2022 09:46:13 - INFO - codeparrot_training - Step 43389: {'lr': 2.3039618719748135e-05, 'samples': 22215680, 'steps': 43389, 'loss/train': 1.608757734298706}
+02/26/2022 09:46:19 - INFO - codeparrot_training - Step 43390: {'lr': 2.3032758205664224e-05, 'samples': 22216192, 'steps': 43390, 'loss/train': 2.877185106277466}
+02/26/2022 09:46:23 - INFO - codeparrot_training - Step 43391: {'lr': 2.302589866383592e-05, 'samples': 22216704, 'steps': 43391, 'loss/train': 1.7679684162139893}
+02/26/2022 09:46:28 - INFO - codeparrot_training - Step 43392: {'lr': 2.3019040094292565e-05, 'samples': 22217216, 'steps': 43392, 'loss/train': 0.6853064298629761}
+02/26/2022 09:46:31 - INFO - codeparrot_training - Step 43393: {'lr': 2.301218249706355e-05, 'samples': 22217728, 'steps': 43393, 'loss/train': 0.793110191822052}
+02/26/2022 09:46:37 - INFO - codeparrot_training - Step 43394: {'lr': 2.3005325872178296e-05, 'samples': 22218240, 'steps': 43394, 'loss/train': 0.7027353048324585}
+02/26/2022 09:46:41 - INFO - codeparrot_training - Step 43395: {'lr': 2.299847021966617e-05, 'samples': 22218752, 'steps': 43395, 'loss/train': 0.8983511924743652}
+02/26/2022 09:46:47 - INFO - codeparrot_training - Step 43396: {'lr': 2.299161553955645e-05, 'samples': 22219264, 'steps': 43396, 'loss/train': 0.4964800775051117}
+02/26/2022 09:46:51 - INFO - codeparrot_training - Step 43397: {'lr': 2.298476183187867e-05, 'samples': 22219776, 'steps': 43397, 'loss/train': 1.0525320768356323}
+02/26/2022 09:46:56 - INFO - codeparrot_training - Step 43398: {'lr': 2.2977909096661974e-05, 'samples': 22220288, 'steps': 43398, 'loss/train': 1.3047536611557007}
+02/26/2022 09:47:00 - INFO - codeparrot_training - Step 43399: {'lr': 2.2971057333935895e-05, 'samples': 22220800, 'steps': 43399, 'loss/train': 1.5811887979507446}
+02/26/2022 09:47:05 - INFO - codeparrot_training - Step 43400: {'lr': 2.296420654372966e-05, 'samples': 22221312, 'steps': 43400, 'loss/train': 1.232550859451294}
+02/26/2022 09:47:08 - INFO - codeparrot_training - Step 43401: {'lr': 2.295735672607277e-05, 'samples': 22221824, 'steps': 43401, 'loss/train': 2.2581450939178467}
+02/26/2022 09:47:14 - INFO - codeparrot_training - Step 43402: {'lr': 2.295050788099437e-05, 'samples': 22222336, 'steps': 43402, 'loss/train': 1.3651314973831177}
+02/26/2022 09:47:17 - INFO - codeparrot_training - Step 43403: {'lr': 2.2943660008523963e-05, 'samples': 22222848, 'steps': 43403, 'loss/train': 1.9363576173782349}
+02/26/2022 09:47:23 - INFO - codeparrot_training - Step 43404: {'lr': 2.2936813108690752e-05, 'samples': 22223360, 'steps': 43404, 'loss/train': 1.8281817436218262}
+02/26/2022 09:47:26 - INFO - codeparrot_training - Step 43405: {'lr': 2.2929967181524235e-05, 'samples': 22223872, 'steps': 43405, 'loss/train': 1.103778600692749}
+02/26/2022 09:47:32 - INFO - codeparrot_training - Step 43406: {'lr': 2.2923122227053532e-05, 'samples': 22224384, 'steps': 43406, 'loss/train': 2.184523820877075}
+02/26/2022 09:47:36 - INFO - codeparrot_training - Step 43407: {'lr': 2.291627824530812e-05, 'samples': 22224896, 'steps': 43407, 'loss/train': 2.1645805835723877}
+02/26/2022 09:47:42 - INFO - codeparrot_training - Step 43408: {'lr': 2.2909435236317222e-05, 'samples': 22225408, 'steps': 43408, 'loss/train': 1.6693423986434937}
+02/26/2022 09:47:45 - INFO - codeparrot_training - Step 43409: {'lr': 2.2902593200110234e-05, 'samples': 22225920, 'steps': 43409, 'loss/train': 1.8765085935592651}
+02/26/2022 09:47:51 - INFO - codeparrot_training - Step 43410: {'lr': 2.2895752136716434e-05, 'samples': 22226432, 'steps': 43410, 'loss/train': 0.6838909983634949}
+02/26/2022 09:47:54 - INFO - codeparrot_training - Step 43411: {'lr': 2.288891204616511e-05, 'samples': 22226944, 'steps': 43411, 'loss/train': 1.6518410444259644}
+02/26/2022 09:48:00 - INFO - codeparrot_training - Step 43412: {'lr': 2.2882072928485515e-05, 'samples': 22227456, 'steps': 43412, 'loss/train': 1.6013000011444092}
+02/26/2022 09:48:03 - INFO - codeparrot_training - Step 43413: {'lr': 2.2875234783707067e-05, 'samples': 22227968, 'steps': 43413, 'loss/train': 3.5163512229919434}
+02/26/2022 09:48:09 - INFO - codeparrot_training - Step 43414: {'lr': 2.2868397611858994e-05, 'samples': 22228480, 'steps': 43414, 'loss/train': 1.1242539882659912}
+02/26/2022 09:48:12 - INFO - codeparrot_training - Step 43415: {'lr': 2.2861561412970576e-05, 'samples': 22228992, 'steps': 43415, 'loss/train': 2.2537989616394043}
+02/26/2022 09:48:18 - INFO - codeparrot_training - Step 43416: {'lr': 2.2854726187071046e-05, 'samples': 22229504, 'steps': 43416, 'loss/train': 1.7038037776947021}
+02/26/2022 09:48:22 - INFO - codeparrot_training - Step 43417: {'lr': 2.284789193418982e-05, 'samples': 22230016, 'steps': 43417, 'loss/train': 1.4832667112350464}
+02/26/2022 09:48:27 - INFO - codeparrot_training - Step 43418: {'lr': 2.2841058654356068e-05, 'samples': 22230528, 'steps': 43418, 'loss/train': 3.860633611679077}
+02/26/2022 09:48:31 - INFO - codeparrot_training - Step 43419: {'lr': 2.2834226347599103e-05, 'samples': 22231040, 'steps': 43419, 'loss/train': 0.5670018792152405}
+02/26/2022 09:48:36 - INFO - codeparrot_training - Step 43420: {'lr': 2.2827395013948123e-05, 'samples': 22231552, 'steps': 43420, 'loss/train': 2.107569932937622}
+02/26/2022 09:48:40 - INFO - codeparrot_training - Step 43421: {'lr': 2.282056465343249e-05, 'samples': 22232064, 'steps': 43421, 'loss/train': 1.589634656906128}
+02/26/2022 09:48:45 - INFO - codeparrot_training - Step 43422: {'lr': 2.281373526608138e-05, 'samples': 22232576, 'steps': 43422, 'loss/train': 1.9231767654418945}
+02/26/2022 09:48:49 - INFO - codeparrot_training - Step 43423: {'lr': 2.2806906851924186e-05, 'samples': 22233088, 'steps': 43423, 'loss/train': 0.6679226160049438}
+02/26/2022 09:48:54 - INFO - codeparrot_training - Step 43424: {'lr': 2.2800079410989966e-05, 'samples': 22233600, 'steps': 43424, 'loss/train': 2.052579641342163}
+02/26/2022 09:48:57 - INFO - codeparrot_training - Step 43425: {'lr': 2.2793252943308114e-05, 'samples': 22234112, 'steps': 43425, 'loss/train': 0.2066899687051773}
+02/26/2022 09:49:03 - INFO - codeparrot_training - Step 43426: {'lr': 2.278642744890777e-05, 'samples': 22234624, 'steps': 43426, 'loss/train': 1.3800404071807861}
+02/26/2022 09:49:07 - INFO - codeparrot_training - Step 43427: {'lr': 2.2779602927818305e-05, 'samples': 22235136, 'steps': 43427, 'loss/train': 2.0924065113067627}
+02/26/2022 09:49:12 - INFO - codeparrot_training - Step 43428: {'lr': 2.2772779380068802e-05, 'samples': 22235648, 'steps': 43428, 'loss/train': 1.7433634996414185}
+02/26/2022 09:49:15 - INFO - codeparrot_training - Step 43429: {'lr': 2.2765956805688576e-05, 'samples': 22236160, 'steps': 43429, 'loss/train': 2.530639410018921}
+02/26/2022 09:49:21 - INFO - codeparrot_training - Step 43430: {'lr': 2.275913520470685e-05, 'samples': 22236672, 'steps': 43430, 'loss/train': 1.3204100131988525}
+02/26/2022 09:49:24 - INFO - codeparrot_training - Step 43431: {'lr': 2.275231457715285e-05, 'samples': 22237184, 'steps': 43431, 'loss/train': 1.4688754081726074}
+02/26/2022 09:49:31 - INFO - codeparrot_training - Step 43432: {'lr': 2.2745494923055694e-05, 'samples': 22237696, 'steps': 43432, 'loss/train': 1.9970577955245972}
+02/26/2022 09:49:35 - INFO - codeparrot_training - Step 43433: {'lr': 2.2738676242444744e-05, 'samples': 22238208, 'steps': 43433, 'loss/train': 0.5037768483161926}
+02/26/2022 09:49:40 - INFO - codeparrot_training - Step 43434: {'lr': 2.273185853534912e-05, 'samples': 22238720, 'steps': 43434, 'loss/train': 1.9678643941879272}
+02/26/2022 09:49:44 - INFO - codeparrot_training - Step 43435: {'lr': 2.272504180179802e-05, 'samples': 22239232, 'steps': 43435, 'loss/train': 1.108302354812622}
+02/26/2022 09:49:49 - INFO - codeparrot_training - Step 43436: {'lr': 2.2718226041820724e-05, 'samples': 22239744, 'steps': 43436, 'loss/train': 2.790809392929077}
+02/26/2022 09:49:52 - INFO - codeparrot_training - Step 43437: {'lr': 2.2711411255446374e-05, 'samples': 22240256, 'steps': 43437, 'loss/train': 1.8792283535003662}
+02/26/2022 09:49:58 - INFO - codeparrot_training - Step 43438: {'lr': 2.2704597442704173e-05, 'samples': 22240768, 'steps': 43438, 'loss/train': 1.9621168375015259}
+02/26/2022 09:50:01 - INFO - codeparrot_training - Step 43439: {'lr': 2.2697784603623235e-05, 'samples': 22241280, 'steps': 43439, 'loss/train': 1.5062484741210938}
+02/26/2022 09:50:07 - INFO - codeparrot_training - Step 43440: {'lr': 2.2690972738232866e-05, 'samples': 22241792, 'steps': 43440, 'loss/train': 0.8874250650405884}
+02/26/2022 09:50:14 - INFO - codeparrot_training - Step 43441: {'lr': 2.268416184656219e-05, 'samples': 22242304, 'steps': 43441, 'loss/train': 2.6573238372802734}
+02/26/2022 09:50:17 - INFO - codeparrot_training - Step 43442: {'lr': 2.26773519286404e-05, 'samples': 22242816, 'steps': 43442, 'loss/train': 2.785123825073242}
+02/26/2022 09:50:23 - INFO - codeparrot_training - Step 43443: {'lr': 2.2670542984496583e-05, 'samples': 22243328, 'steps': 43443, 'loss/train': 1.7695703506469727}
+02/26/2022 09:50:26 - INFO - codeparrot_training - Step 43444: {'lr': 2.2663735014160026e-05, 'samples': 22243840, 'steps': 43444, 'loss/train': 1.756841778755188}
+02/26/2022 09:50:32 - INFO - codeparrot_training - Step 43445: {'lr': 2.265692801765984e-05, 'samples': 22244352, 'steps': 43445, 'loss/train': 1.8958699703216553}
+02/26/2022 09:50:35 - INFO - codeparrot_training - Step 43446: {'lr': 2.2650121995025198e-05, 'samples': 22244864, 'steps': 43446, 'loss/train': 2.4778921604156494}
+02/26/2022 09:50:41 - INFO - codeparrot_training - Step 43447: {'lr': 2.2643316946285186e-05, 'samples': 22245376, 'steps': 43447, 'loss/train': 1.6054095029830933}
+02/26/2022 09:50:44 - INFO - codeparrot_training - Step 43448: {'lr': 2.263651287146906e-05, 'samples': 22245888, 'steps': 43448, 'loss/train': 1.0989842414855957}
+02/26/2022 09:50:50 - INFO - codeparrot_training - Step 43449: {'lr': 2.2629709770605912e-05, 'samples': 22246400, 'steps': 43449, 'loss/train': 1.7166266441345215}
+02/26/2022 09:50:53 - INFO - codeparrot_training - Step 43450: {'lr': 2.2622907643724903e-05, 'samples': 22246912, 'steps': 43450, 'loss/train': 1.7227504253387451}
+02/26/2022 09:50:59 - INFO - codeparrot_training - Step 43451: {'lr': 2.2616106490855076e-05, 'samples': 22247424, 'steps': 43451, 'loss/train': 0.9361081123352051}
+02/26/2022 09:51:02 - INFO - codeparrot_training - Step 43452: {'lr': 2.2609306312025735e-05, 'samples': 22247936, 'steps': 43452, 'loss/train': 8.648787498474121}
+02/26/2022 09:51:09 - INFO - codeparrot_training - Step 43453: {'lr': 2.2602507107265886e-05, 'samples': 22248448, 'steps': 43453, 'loss/train': 2.183933973312378}
+02/26/2022 09:51:12 - INFO - codeparrot_training - Step 43454: {'lr': 2.2595708876604727e-05, 'samples': 22248960, 'steps': 43454, 'loss/train': 1.903673529624939}
+02/26/2022 09:51:18 - INFO - codeparrot_training - Step 43455: {'lr': 2.2588911620071262e-05, 'samples': 22249472, 'steps': 43455, 'loss/train': 1.6323210000991821}
+02/26/2022 09:51:21 - INFO - codeparrot_training - Step 43456: {'lr': 2.2582115337694746e-05, 'samples': 22249984, 'steps': 43456, 'loss/train': 2.4872117042541504}
+02/26/2022 09:51:27 - INFO - codeparrot_training - Step 43457: {'lr': 2.2575320029504186e-05, 'samples': 22250496, 'steps': 43457, 'loss/train': 2.29856276512146}
+02/26/2022 09:51:30 - INFO - codeparrot_training - Step 43458: {'lr': 2.2568525695528834e-05, 'samples': 22251008, 'steps': 43458, 'loss/train': 2.756319284439087}
+02/26/2022 09:51:36 - INFO - codeparrot_training - Step 43459: {'lr': 2.2561732335797615e-05, 'samples': 22251520, 'steps': 43459, 'loss/train': 1.7654610872268677}
+02/26/2022 09:51:39 - INFO - codeparrot_training - Step 43460: {'lr': 2.255493995033975e-05, 'samples': 22252032, 'steps': 43460, 'loss/train': 1.4436110258102417}
+02/26/2022 09:51:45 - INFO - codeparrot_training - Step 43461: {'lr': 2.254814853918427e-05, 'samples': 22252544, 'steps': 43461, 'loss/train': 1.920397400856018}
+02/26/2022 09:51:48 - INFO - codeparrot_training - Step 43462: {'lr': 2.2541358102360383e-05, 'samples': 22253056, 'steps': 43462, 'loss/train': 2.2659895420074463}
+02/26/2022 09:51:54 - INFO - codeparrot_training - Step 43463: {'lr': 2.2534568639897e-05, 'samples': 22253568, 'steps': 43463, 'loss/train': 0.08842244744300842}
+02/26/2022 09:51:58 - INFO - codeparrot_training - Step 43464: {'lr': 2.252778015182333e-05, 'samples': 22254080, 'steps': 43464, 'loss/train': 3.660236358642578}
+02/26/2022 09:52:03 - INFO - codeparrot_training - Step 43465: {'lr': 2.252099263816837e-05, 'samples': 22254592, 'steps': 43465, 'loss/train': 1.9817352294921875}
+02/26/2022 09:52:07 - INFO - codeparrot_training - Step 43466: {'lr': 2.2514206098961348e-05, 'samples': 22255104, 'steps': 43466, 'loss/train': 1.95352041721344}
+02/26/2022 09:52:12 - INFO - codeparrot_training - Step 43467: {'lr': 2.250742053423116e-05, 'samples': 22255616, 'steps': 43467, 'loss/train': 2.5268237590789795}
+02/26/2022 09:52:16 - INFO - codeparrot_training - Step 43468: {'lr': 2.2500635944006976e-05, 'samples': 22256128, 'steps': 43468, 'loss/train': 2.38124418258667}
+02/26/2022 09:52:21 - INFO - codeparrot_training - Step 43469: {'lr': 2.249385232831777e-05, 'samples': 22256640, 'steps': 43469, 'loss/train': 1.3344988822937012}
+02/26/2022 09:52:25 - INFO - codeparrot_training - Step 43470: {'lr': 2.248706968719272e-05, 'samples': 22257152, 'steps': 43470, 'loss/train': 2.3130009174346924}
+02/26/2022 09:52:30 - INFO - codeparrot_training - Step 43471: {'lr': 2.2480288020660793e-05, 'samples': 22257664, 'steps': 43471, 'loss/train': 1.9131823778152466}
+02/26/2022 09:52:34 - INFO - codeparrot_training - Step 43472: {'lr': 2.2473507328751085e-05, 'samples': 22258176, 'steps': 43472, 'loss/train': 1.6652742624282837}
+02/26/2022 09:52:39 - INFO - codeparrot_training - Step 43473: {'lr': 2.24667276114926e-05, 'samples': 22258688, 'steps': 43473, 'loss/train': 1.5640021562576294}
+02/26/2022 09:52:43 - INFO - codeparrot_training - Step 43474: {'lr': 2.245994886891442e-05, 'samples': 22259200, 'steps': 43474, 'loss/train': 1.9226869344711304}
+02/26/2022 09:52:49 - INFO - codeparrot_training - Step 43475: {'lr': 2.2453171101045584e-05, 'samples': 22259712, 'steps': 43475, 'loss/train': 2.1402430534362793}
+02/26/2022 09:52:52 - INFO - codeparrot_training - Step 43476: {'lr': 2.244639430791509e-05, 'samples': 22260224, 'steps': 43476, 'loss/train': 1.4616705179214478}
+02/26/2022 09:52:58 - INFO - codeparrot_training - Step 43477: {'lr': 2.243961848955195e-05, 'samples': 22260736, 'steps': 43477, 'loss/train': 0.03049241192638874}
+02/26/2022 09:53:02 - INFO - codeparrot_training - Step 43478: {'lr': 2.2432843645985273e-05, 'samples': 22261248, 'steps': 43478, 'loss/train': 0.6343022584915161}
+02/26/2022 09:53:07 - INFO - codeparrot_training - Step 43479: {'lr': 2.242606977724404e-05, 'samples': 22261760, 'steps': 43479, 'loss/train': 3.0201783180236816}
+02/26/2022 09:53:11 - INFO - codeparrot_training - Step 43480: {'lr': 2.2419296883357254e-05, 'samples': 22262272, 'steps': 43480, 'loss/train': 0.8773584365844727}
+02/26/2022 09:53:16 - INFO - codeparrot_training - Step 43481: {'lr': 2.2412524964353948e-05, 'samples': 22262784, 'steps': 43481, 'loss/train': 1.2619014978408813}
+02/26/2022 09:53:20 - INFO - codeparrot_training - Step 43482: {'lr': 2.2405754020263043e-05, 'samples': 22263296, 'steps': 43482, 'loss/train': 1.7896931171417236}
+02/26/2022 09:53:25 - INFO - codeparrot_training - Step 43483: {'lr': 2.2398984051113708e-05, 'samples': 22263808, 'steps': 43483, 'loss/train': 1.3970919847488403}
+02/26/2022 09:53:29 - INFO - codeparrot_training - Step 43484: {'lr': 2.239221505693481e-05, 'samples': 22264320, 'steps': 43484, 'loss/train': 1.3139902353286743}
+02/26/2022 09:53:34 - INFO - codeparrot_training - Step 43485: {'lr': 2.2385447037755412e-05, 'samples': 22264832, 'steps': 43485, 'loss/train': 0.5896218419075012}
+02/26/2022 09:53:38 - INFO - codeparrot_training - Step 43486: {'lr': 2.2378679993604435e-05, 'samples': 22265344, 'steps': 43486, 'loss/train': 2.613023281097412}
+02/26/2022 09:53:44 - INFO - codeparrot_training - Step 43487: {'lr': 2.237191392451096e-05, 'samples': 22265856, 'steps': 43487, 'loss/train': 2.0351693630218506}
+02/26/2022 09:53:48 - INFO - codeparrot_training - Step 43488: {'lr': 2.2365148830503918e-05, 'samples': 22266368, 'steps': 43488, 'loss/train': 2.0091772079467773}
+02/26/2022 09:53:51 - INFO - codeparrot_training - Step 43489: {'lr': 2.2358384711612307e-05, 'samples': 22266880, 'steps': 43489, 'loss/train': 1.3579537868499756}
+02/26/2022 09:53:57 - INFO - codeparrot_training - Step 43490: {'lr': 2.2351621567865022e-05, 'samples': 22267392, 'steps': 43490, 'loss/train': 1.9796391725540161}
+02/26/2022 09:54:01 - INFO - codeparrot_training - Step 43491: {'lr': 2.234485939929118e-05, 'samples': 22267904, 'steps': 43491, 'loss/train': 0.026413578540086746}
+02/26/2022 09:54:06 - INFO - codeparrot_training - Step 43492: {'lr': 2.2338098205919615e-05, 'samples': 22268416, 'steps': 43492, 'loss/train': 0.3130205273628235}
+02/26/2022 09:54:10 - INFO - codeparrot_training - Step 43493: {'lr': 2.2331337987779447e-05, 'samples': 22268928, 'steps': 43493, 'loss/train': 1.8519543409347534}
+02/26/2022 09:54:15 - INFO - codeparrot_training - Step 43494: {'lr': 2.2324578744899432e-05, 'samples': 22269440, 'steps': 43494, 'loss/train': 1.9269509315490723}
+02/26/2022 09:54:19 - INFO - codeparrot_training - Step 43495: {'lr': 2.2317820477308653e-05, 'samples': 22269952, 'steps': 43495, 'loss/train': 2.3492367267608643}
+02/26/2022 09:54:24 - INFO - codeparrot_training - Step 43496: {'lr': 2.2311063185036007e-05, 'samples': 22270464, 'steps': 43496, 'loss/train': 1.2519398927688599}
+02/26/2022 09:54:28 - INFO - codeparrot_training - Step 43497: {'lr': 2.230430686811058e-05, 'samples': 22270976, 'steps': 43497, 'loss/train': 2.450395107269287}
+02/26/2022 09:54:35 - INFO - codeparrot_training - Step 43498: {'lr': 2.229755152656107e-05, 'samples': 22271488, 'steps': 43498, 'loss/train': 0.159735769033432}
+02/26/2022 09:54:38 - INFO - codeparrot_training - Step 43499: {'lr': 2.2290797160416628e-05, 'samples': 22272000, 'steps': 43499, 'loss/train': 0.40374743938446045}
+02/26/2022 09:54:44 - INFO - codeparrot_training - Step 43500: {'lr': 2.2284043769706025e-05, 'samples': 22272512, 'steps': 43500, 'loss/train': 1.3648148775100708}
+02/26/2022 09:54:47 - INFO - codeparrot_training - Step 43501: {'lr': 2.2277291354458385e-05, 'samples': 22273024, 'steps': 43501, 'loss/train': 1.0472849607467651}
+02/26/2022 09:54:53 - INFO - codeparrot_training - Step 43502: {'lr': 2.2270539914702433e-05, 'samples': 22273536, 'steps': 43502, 'loss/train': 1.6023541688919067}
+02/26/2022 09:54:56 - INFO - codeparrot_training - Step 43503: {'lr': 2.2263789450467197e-05, 'samples': 22274048, 'steps': 43503, 'loss/train': 2.093118667602539}
+02/26/2022 09:55:02 - INFO - codeparrot_training - Step 43504: {'lr': 2.225703996178155e-05, 'samples': 22274560, 'steps': 43504, 'loss/train': 0.5411143898963928}
+02/26/2022 09:55:05 - INFO - codeparrot_training - Step 43505: {'lr': 2.2250291448674438e-05, 'samples': 22275072, 'steps': 43505, 'loss/train': 2.3279290199279785}
+02/26/2022 09:55:11 - INFO - codeparrot_training - Step 43506: {'lr': 2.2243543911174808e-05, 'samples': 22275584, 'steps': 43506, 'loss/train': 1.6749930381774902}
+02/26/2022 09:55:14 - INFO - codeparrot_training - Step 43507: {'lr': 2.2236797349311473e-05, 'samples': 22276096, 'steps': 43507, 'loss/train': 1.1816524267196655}
+02/26/2022 09:55:20 - INFO - codeparrot_training - Step 43508: {'lr': 2.2230051763113353e-05, 'samples': 22276608, 'steps': 43508, 'loss/train': 1.7725175619125366}
+02/26/2022 09:55:23 - INFO - codeparrot_training - Step 43509: {'lr': 2.2223307152609394e-05, 'samples': 22277120, 'steps': 43509, 'loss/train': 1.676793098449707}
+02/26/2022 09:55:29 - INFO - codeparrot_training - Step 43510: {'lr': 2.2216563517828468e-05, 'samples': 22277632, 'steps': 43510, 'loss/train': 1.7410683631896973}
+02/26/2022 09:55:32 - INFO - codeparrot_training - Step 43511: {'lr': 2.2209820858799468e-05, 'samples': 22278144, 'steps': 43511, 'loss/train': 1.521414041519165}
+02/26/2022 09:55:38 - INFO - codeparrot_training - Step 43512: {'lr': 2.22030791755512e-05, 'samples': 22278656, 'steps': 43512, 'loss/train': 1.7858301401138306}
+02/26/2022 09:55:41 - INFO - codeparrot_training - Step 43513: {'lr': 2.219633846811267e-05, 'samples': 22279168, 'steps': 43513, 'loss/train': 1.272739052772522}
+02/26/2022 09:55:48 - INFO - codeparrot_training - Step 43514: {'lr': 2.218959873651269e-05, 'samples': 22279680, 'steps': 43514, 'loss/train': 1.957121729850769}
+02/26/2022 09:55:51 - INFO - codeparrot_training - Step 43515: {'lr': 2.2182859980780122e-05, 'samples': 22280192, 'steps': 43515, 'loss/train': 1.0967785120010376}
+02/26/2022 09:55:57 - INFO - codeparrot_training - Step 43516: {'lr': 2.2176122200943784e-05, 'samples': 22280704, 'steps': 43516, 'loss/train': 2.4115428924560547}
+02/26/2022 09:56:00 - INFO - codeparrot_training - Step 43517: {'lr': 2.2169385397032644e-05, 'samples': 22281216, 'steps': 43517, 'loss/train': 1.3720903396606445}
+02/26/2022 09:56:06 - INFO - codeparrot_training - Step 43518: {'lr': 2.2162649569075543e-05, 'samples': 22281728, 'steps': 43518, 'loss/train': 1.6678321361541748}
+02/26/2022 09:56:09 - INFO - codeparrot_training - Step 43519: {'lr': 2.2155914717101265e-05, 'samples': 22282240, 'steps': 43519, 'loss/train': 1.8068166971206665}
+02/26/2022 09:56:15 - INFO - codeparrot_training - Step 43520: {'lr': 2.214918084113868e-05, 'samples': 22282752, 'steps': 43520, 'loss/train': 2.0317234992980957}
+02/26/2022 09:56:18 - INFO - codeparrot_training - Step 43521: {'lr': 2.2142447941216697e-05, 'samples': 22283264, 'steps': 43521, 'loss/train': 2.336134433746338}
+02/26/2022 09:56:24 - INFO - codeparrot_training - Step 43522: {'lr': 2.213571601736411e-05, 'samples': 22283776, 'steps': 43522, 'loss/train': 0.7021390199661255}
+02/26/2022 09:56:27 - INFO - codeparrot_training - Step 43523: {'lr': 2.2128985069609754e-05, 'samples': 22284288, 'steps': 43523, 'loss/train': 1.630194067955017}
+02/26/2022 09:56:33 - INFO - codeparrot_training - Step 43524: {'lr': 2.2122255097982437e-05, 'samples': 22284800, 'steps': 43524, 'loss/train': 0.6716061234474182}
+02/26/2022 09:56:37 - INFO - codeparrot_training - Step 43525: {'lr': 2.211552610251105e-05, 'samples': 22285312, 'steps': 43525, 'loss/train': 1.8806359767913818}
+02/26/2022 09:56:42 - INFO - codeparrot_training - Step 43526: {'lr': 2.2108798083224384e-05, 'samples': 22285824, 'steps': 43526, 'loss/train': 1.6183557510375977}
+02/26/2022 09:56:46 - INFO - codeparrot_training - Step 43527: {'lr': 2.21020710401513e-05, 'samples': 22286336, 'steps': 43527, 'loss/train': 1.3140428066253662}
+02/26/2022 09:56:51 - INFO - codeparrot_training - Step 43528: {'lr': 2.2095344973320554e-05, 'samples': 22286848, 'steps': 43528, 'loss/train': 2.096261978149414}
+02/26/2022 09:56:55 - INFO - codeparrot_training - Step 43529: {'lr': 2.2088619882760925e-05, 'samples': 22287360, 'steps': 43529, 'loss/train': 2.4248530864715576}
+02/26/2022 09:57:00 - INFO - codeparrot_training - Step 43530: {'lr': 2.2081895768501337e-05, 'samples': 22287872, 'steps': 43530, 'loss/train': 0.021893871948122978}
+02/26/2022 09:57:04 - INFO - codeparrot_training - Step 43531: {'lr': 2.2075172630570488e-05, 'samples': 22288384, 'steps': 43531, 'loss/train': 1.6067407131195068}
+02/26/2022 09:57:09 - INFO - codeparrot_training - Step 43532: {'lr': 2.2068450468997302e-05, 'samples': 22288896, 'steps': 43532, 'loss/train': 0.09487932175397873}
+02/26/2022 09:57:15 - INFO - codeparrot_training - Step 43533: {'lr': 2.2061729283810422e-05, 'samples': 22289408, 'steps': 43533, 'loss/train': 1.2157052755355835}
+02/26/2022 09:57:19 - INFO - codeparrot_training - Step 43534: {'lr': 2.2055009075038767e-05, 'samples': 22289920, 'steps': 43534, 'loss/train': 2.417781352996826}
+02/26/2022 09:57:24 - INFO - codeparrot_training - Step 43535: {'lr': 2.2048289842711013e-05, 'samples': 22290432, 'steps': 43535, 'loss/train': 1.621270775794983}
+02/26/2022 09:57:28 - INFO - codeparrot_training - Step 43536: {'lr': 2.20415715868561e-05, 'samples': 22290944, 'steps': 43536, 'loss/train': 0.9649053812026978}
+02/26/2022 09:57:33 - INFO - codeparrot_training - Step 43537: {'lr': 2.203485430750263e-05, 'samples': 22291456, 'steps': 43537, 'loss/train': 1.3556554317474365}
+02/26/2022 09:57:37 - INFO - codeparrot_training - Step 43538: {'lr': 2.202813800467948e-05, 'samples': 22291968, 'steps': 43538, 'loss/train': 0.5001811385154724}
+02/26/2022 09:57:42 - INFO - codeparrot_training - Step 43539: {'lr': 2.2021422678415365e-05, 'samples': 22292480, 'steps': 43539, 'loss/train': 1.3180254697799683}
+02/26/2022 09:57:46 - INFO - codeparrot_training - Step 43540: {'lr': 2.201470832873917e-05, 'samples': 22292992, 'steps': 43540, 'loss/train': 2.034135103225708}
+02/26/2022 09:57:51 - INFO - codeparrot_training - Step 43541: {'lr': 2.2007994955679482e-05, 'samples': 22293504, 'steps': 43541, 'loss/train': 1.5188831090927124}
+02/26/2022 09:57:54 - INFO - codeparrot_training - Step 43542: {'lr': 2.2001282559265202e-05, 'samples': 22294016, 'steps': 43542, 'loss/train': 2.059890031814575}
+02/26/2022 09:58:01 - INFO - codeparrot_training - Step 43543: {'lr': 2.1994571139524966e-05, 'samples': 22294528, 'steps': 43543, 'loss/train': 2.3666932582855225}
+02/26/2022 09:58:04 - INFO - codeparrot_training - Step 43544: {'lr': 2.1987860696487644e-05, 'samples': 22295040, 'steps': 43544, 'loss/train': 2.2725865840911865}
+02/26/2022 09:58:10 - INFO - codeparrot_training - Step 43545: {'lr': 2.1981151230181906e-05, 'samples': 22295552, 'steps': 43545, 'loss/train': 1.6586047410964966}
+02/26/2022 09:58:13 - INFO - codeparrot_training - Step 43546: {'lr': 2.1974442740636535e-05, 'samples': 22296064, 'steps': 43546, 'loss/train': 1.552412509918213}
+02/26/2022 09:58:19 - INFO - codeparrot_training - Step 43547: {'lr': 2.1967735227880203e-05, 'samples': 22296576, 'steps': 43547, 'loss/train': 2.3389294147491455}
+02/26/2022 09:58:22 - INFO - codeparrot_training - Step 43548: {'lr': 2.196102869194172e-05, 'samples': 22297088, 'steps': 43548, 'loss/train': 2.385468006134033}
+02/26/2022 09:58:28 - INFO - codeparrot_training - Step 43549: {'lr': 2.1954323132849786e-05, 'samples': 22297600, 'steps': 43549, 'loss/train': 2.4554502964019775}
+02/26/2022 09:58:31 - INFO - codeparrot_training - Step 43550: {'lr': 2.1947618550633096e-05, 'samples': 22298112, 'steps': 43550, 'loss/train': 2.7939515113830566}
+02/26/2022 09:58:37 - INFO - codeparrot_training - Step 43551: {'lr': 2.1940914945320384e-05, 'samples': 22298624, 'steps': 43551, 'loss/train': 1.6824169158935547}
+02/26/2022 09:58:40 - INFO - codeparrot_training - Step 43552: {'lr': 2.19342123169404e-05, 'samples': 22299136, 'steps': 43552, 'loss/train': 0.776712954044342}
+02/26/2022 09:58:46 - INFO - codeparrot_training - Step 43553: {'lr': 2.192751066552179e-05, 'samples': 22299648, 'steps': 43553, 'loss/train': 1.4674180746078491}
+02/26/2022 09:58:49 - INFO - codeparrot_training - Step 43554: {'lr': 2.1920809991093387e-05, 'samples': 22300160, 'steps': 43554, 'loss/train': 2.2019076347351074}
+02/26/2022 09:58:55 - INFO - codeparrot_training - Step 43555: {'lr': 2.1914110293683733e-05, 'samples': 22300672, 'steps': 43555, 'loss/train': 0.9122690558433533}
+02/26/2022 09:58:58 - INFO - codeparrot_training - Step 43556: {'lr': 2.190741157332163e-05, 'samples': 22301184, 'steps': 43556, 'loss/train': 2.2633981704711914}
+02/26/2022 09:59:02 - INFO - codeparrot_training - Step 43557: {'lr': 2.1900713830035724e-05, 'samples': 22301696, 'steps': 43557, 'loss/train': 1.1077462434768677}
+02/26/2022 09:59:08 - INFO - codeparrot_training - Step 43558: {'lr': 2.18940170638548e-05, 'samples': 22302208, 'steps': 43558, 'loss/train': 1.3015111684799194}
+02/26/2022 09:59:11 - INFO - codeparrot_training - Step 43559: {'lr': 2.1887321274807416e-05, 'samples': 22302720, 'steps': 43559, 'loss/train': 0.40784981846809387}
+02/26/2022 09:59:17 - INFO - codeparrot_training - Step 43560: {'lr': 2.1880626462922326e-05, 'samples': 22303232, 'steps': 43560, 'loss/train': 1.8171428442001343}
+02/26/2022 09:59:21 - INFO - codeparrot_training - Step 43561: {'lr': 2.1873932628228204e-05, 'samples': 22303744, 'steps': 43561, 'loss/train': 2.3191559314727783}
+02/26/2022 09:59:26 - INFO - codeparrot_training - Step 43562: {'lr': 2.1867239770753723e-05, 'samples': 22304256, 'steps': 43562, 'loss/train': 1.5726062059402466}
+02/26/2022 09:59:30 - INFO - codeparrot_training - Step 43563: {'lr': 2.186054789052752e-05, 'samples': 22304768, 'steps': 43563, 'loss/train': 0.5699591636657715}
+02/26/2022 09:59:36 - INFO - codeparrot_training - Step 43564: {'lr': 2.1853856987578302e-05, 'samples': 22305280, 'steps': 43564, 'loss/train': 1.682510256767273}
+02/26/2022 09:59:39 - INFO - codeparrot_training - Step 43565: {'lr': 2.1847167061934737e-05, 'samples': 22305792, 'steps': 43565, 'loss/train': 1.5512629747390747}
+02/26/2022 09:59:45 - INFO - codeparrot_training - Step 43566: {'lr': 2.1840478113625413e-05, 'samples': 22306304, 'steps': 43566, 'loss/train': 1.0707979202270508}
+02/26/2022 09:59:48 - INFO - codeparrot_training - Step 43567: {'lr': 2.1833790142679055e-05, 'samples': 22306816, 'steps': 43567, 'loss/train': 1.5064936876296997}
+02/26/2022 09:59:54 - INFO - codeparrot_training - Step 43568: {'lr': 2.1827103149124312e-05, 'samples': 22307328, 'steps': 43568, 'loss/train': 0.9083361625671387}
+02/26/2022 09:59:57 - INFO - codeparrot_training - Step 43569: {'lr': 2.1820417132989795e-05, 'samples': 22307840, 'steps': 43569, 'loss/train': 1.3410167694091797}
+02/26/2022 10:00:03 - INFO - codeparrot_training - Step 43570: {'lr': 2.1813732094304124e-05, 'samples': 22308352, 'steps': 43570, 'loss/train': 2.022702217102051}
+02/26/2022 10:00:07 - INFO - codeparrot_training - Step 43571: {'lr': 2.1807048033095994e-05, 'samples': 22308864, 'steps': 43571, 'loss/train': 2.2553858757019043}
+02/26/2022 10:00:12 - INFO - codeparrot_training - Step 43572: {'lr': 2.1800364949394027e-05, 'samples': 22309376, 'steps': 43572, 'loss/train': 1.5829468965530396}
+02/26/2022 10:00:16 - INFO - codeparrot_training - Step 43573: {'lr': 2.179368284322683e-05, 'samples': 22309888, 'steps': 43573, 'loss/train': 1.0391923189163208}
+02/26/2022 10:00:22 - INFO - codeparrot_training - Step 43574: {'lr': 2.1787001714622974e-05, 'samples': 22310400, 'steps': 43574, 'loss/train': 2.8245155811309814}
+02/26/2022 10:00:25 - INFO - codeparrot_training - Step 43575: {'lr': 2.1780321563611233e-05, 'samples': 22310912, 'steps': 43575, 'loss/train': 2.0118017196655273}
+02/26/2022 10:00:31 - INFO - codeparrot_training - Step 43576: {'lr': 2.1773642390220062e-05, 'samples': 22311424, 'steps': 43576, 'loss/train': 1.3278862237930298}
+02/26/2022 10:00:34 - INFO - codeparrot_training - Step 43577: {'lr': 2.1766964194478157e-05, 'samples': 22311936, 'steps': 43577, 'loss/train': 1.2484737634658813}
+02/26/2022 10:00:40 - INFO - codeparrot_training - Step 43578: {'lr': 2.1760286976414053e-05, 'samples': 22312448, 'steps': 43578, 'loss/train': 2.48964786529541}
+02/26/2022 10:00:43 - INFO - codeparrot_training - Step 43579: {'lr': 2.175361073605647e-05, 'samples': 22312960, 'steps': 43579, 'loss/train': 0.3831650912761688}
+02/26/2022 10:00:49 - INFO - codeparrot_training - Step 43580: {'lr': 2.1746935473433927e-05, 'samples': 22313472, 'steps': 43580, 'loss/train': 1.5037912130355835}
+02/26/2022 10:00:53 - INFO - codeparrot_training - Step 43581: {'lr': 2.1740261188575056e-05, 'samples': 22313984, 'steps': 43581, 'loss/train': 1.6030086278915405}
+02/26/2022 10:00:59 - INFO - codeparrot_training - Step 43582: {'lr': 2.1733587881508366e-05, 'samples': 22314496, 'steps': 43582, 'loss/train': 1.0923399925231934}
+02/26/2022 10:01:02 - INFO - codeparrot_training - Step 43583: {'lr': 2.1726915552262556e-05, 'samples': 22315008, 'steps': 43583, 'loss/train': 2.1169118881225586}
+02/26/2022 10:01:08 - INFO - codeparrot_training - Step 43584: {'lr': 2.1720244200866128e-05, 'samples': 22315520, 'steps': 43584, 'loss/train': 2.002955436706543}
+02/26/2022 10:01:11 - INFO - codeparrot_training - Step 43585: {'lr': 2.1713573827347706e-05, 'samples': 22316032, 'steps': 43585, 'loss/train': 1.750496745109558}
+02/26/2022 10:01:17 - INFO - codeparrot_training - Step 43586: {'lr': 2.1706904431735812e-05, 'samples': 22316544, 'steps': 43586, 'loss/train': 1.5691057443618774}
+02/26/2022 10:01:20 - INFO - codeparrot_training - Step 43587: {'lr': 2.170023601405907e-05, 'samples': 22317056, 'steps': 43587, 'loss/train': 1.8770387172698975}
+02/26/2022 10:01:26 - INFO - codeparrot_training - Step 43588: {'lr': 2.169356857434604e-05, 'samples': 22317568, 'steps': 43588, 'loss/train': 2.1737616062164307}
+02/26/2022 10:01:29 - INFO - codeparrot_training - Step 43589: {'lr': 2.168690211262525e-05, 'samples': 22318080, 'steps': 43589, 'loss/train': 1.8542524576187134}
+02/26/2022 10:01:35 - INFO - codeparrot_training - Step 43590: {'lr': 2.1680236628925238e-05, 'samples': 22318592, 'steps': 43590, 'loss/train': 1.8732661008834839}
+02/26/2022 10:01:39 - INFO - codeparrot_training - Step 43591: {'lr': 2.1673572123274643e-05, 'samples': 22319104, 'steps': 43591, 'loss/train': 2.0041987895965576}
+02/26/2022 10:01:45 - INFO - codeparrot_training - Step 43592: {'lr': 2.1666908595701917e-05, 'samples': 22319616, 'steps': 43592, 'loss/train': 1.4893574714660645}
+02/26/2022 10:01:48 - INFO - codeparrot_training - Step 43593: {'lr': 2.1660246046235733e-05, 'samples': 22320128, 'steps': 43593, 'loss/train': 0.6406717896461487}
+02/26/2022 10:01:52 - INFO - codeparrot_training - Step 43594: {'lr': 2.165358447490448e-05, 'samples': 22320640, 'steps': 43594, 'loss/train': 0.5837734341621399}
+02/26/2022 10:01:57 - INFO - codeparrot_training - Step 43595: {'lr': 2.164692388173678e-05, 'samples': 22321152, 'steps': 43595, 'loss/train': 2.337806224822998}
+02/26/2022 10:02:03 - INFO - codeparrot_training - Step 43596: {'lr': 2.1640264266761106e-05, 'samples': 22321664, 'steps': 43596, 'loss/train': 3.754213571548462}
+02/26/2022 10:02:06 - INFO - codeparrot_training - Step 43597: {'lr': 2.163360563000613e-05, 'samples': 22322176, 'steps': 43597, 'loss/train': 1.6627697944641113}
+02/26/2022 10:02:10 - INFO - codeparrot_training - Step 43598: {'lr': 2.1626947971500193e-05, 'samples': 22322688, 'steps': 43598, 'loss/train': 0.7337480783462524}
+02/26/2022 10:02:15 - INFO - codeparrot_training - Step 43599: {'lr': 2.162029129127191e-05, 'samples': 22323200, 'steps': 43599, 'loss/train': 2.1815218925476074}
+02/26/2022 10:02:19 - INFO - codeparrot_training - Step 43600: {'lr': 2.1613635589349755e-05, 'samples': 22323712, 'steps': 43600, 'loss/train': 0.5456998348236084}
+02/26/2022 10:02:24 - INFO - codeparrot_training - Step 43601: {'lr': 2.1606980865762293e-05, 'samples': 22324224, 'steps': 43601, 'loss/train': 1.6283646821975708}
+02/26/2022 10:02:28 - INFO - codeparrot_training - Step 43602: {'lr': 2.1600327120538028e-05, 'samples': 22324736, 'steps': 43602, 'loss/train': 1.1674281358718872}
+02/26/2022 10:02:33 - INFO - codeparrot_training - Step 43603: {'lr': 2.1593674353705433e-05, 'samples': 22325248, 'steps': 43603, 'loss/train': 1.5113232135772705}
+02/26/2022 10:02:37 - INFO - codeparrot_training - Step 43604: {'lr': 2.1587022565292935e-05, 'samples': 22325760, 'steps': 43604, 'loss/train': 1.2222647666931152}
+02/26/2022 10:02:42 - INFO - codeparrot_training - Step 43605: {'lr': 2.1580371755329174e-05, 'samples': 22326272, 'steps': 43605, 'loss/train': 2.115407705307007}
+02/26/2022 10:02:48 - INFO - codeparrot_training - Step 43606: {'lr': 2.1573721923842577e-05, 'samples': 22326784, 'steps': 43606, 'loss/train': 1.6032240390777588}
+02/26/2022 10:02:51 - INFO - codeparrot_training - Step 43607: {'lr': 2.1567073070861585e-05, 'samples': 22327296, 'steps': 43607, 'loss/train': 0.5503872632980347}
+02/26/2022 10:02:58 - INFO - codeparrot_training - Step 43608: {'lr': 2.156042519641474e-05, 'samples': 22327808, 'steps': 43608, 'loss/train': 1.4683725833892822}
+02/26/2022 10:03:01 - INFO - codeparrot_training - Step 43609: {'lr': 2.1553778300530458e-05, 'samples': 22328320, 'steps': 43609, 'loss/train': 1.0390137434005737}
+02/26/2022 10:03:07 - INFO - codeparrot_training - Step 43610: {'lr': 2.1547132383237272e-05, 'samples': 22328832, 'steps': 43610, 'loss/train': 1.2063630819320679}
+02/26/2022 10:03:10 - INFO - codeparrot_training - Step 43611: {'lr': 2.1540487444563663e-05, 'samples': 22329344, 'steps': 43611, 'loss/train': 2.2851593494415283}
+02/26/2022 10:03:14 - INFO - codeparrot_training - Step 43612: {'lr': 2.1533843484538047e-05, 'samples': 22329856, 'steps': 43612, 'loss/train': 2.853872060775757}
+02/26/2022 10:03:19 - INFO - codeparrot_training - Step 43613: {'lr': 2.152720050318885e-05, 'samples': 22330368, 'steps': 43613, 'loss/train': 2.212364435195923}
+02/26/2022 10:03:23 - INFO - codeparrot_training - Step 43614: {'lr': 2.15205585005446e-05, 'samples': 22330880, 'steps': 43614, 'loss/train': 0.7002944946289062}
+02/26/2022 10:03:28 - INFO - codeparrot_training - Step 43615: {'lr': 2.1513917476633755e-05, 'samples': 22331392, 'steps': 43615, 'loss/train': 0.7644915580749512}
+02/26/2022 10:03:32 - INFO - codeparrot_training - Step 43616: {'lr': 2.150727743148473e-05, 'samples': 22331904, 'steps': 43616, 'loss/train': 1.8617172241210938}
+02/26/2022 10:03:38 - INFO - codeparrot_training - Step 43617: {'lr': 2.1500638365125918e-05, 'samples': 22332416, 'steps': 43617, 'loss/train': 1.6511237621307373}
+02/26/2022 10:03:41 - INFO - codeparrot_training - Step 43618: {'lr': 2.1494000277585885e-05, 'samples': 22332928, 'steps': 43618, 'loss/train': 0.6012080311775208}
+02/26/2022 10:03:48 - INFO - codeparrot_training - Step 43619: {'lr': 2.1487363168892965e-05, 'samples': 22333440, 'steps': 43619, 'loss/train': 2.4701123237609863}
+02/26/2022 10:03:51 - INFO - codeparrot_training - Step 43620: {'lr': 2.148072703907564e-05, 'samples': 22333952, 'steps': 43620, 'loss/train': 1.5888080596923828}
+02/26/2022 10:03:56 - INFO - codeparrot_training - Step 43621: {'lr': 2.1474091888162265e-05, 'samples': 22334464, 'steps': 43621, 'loss/train': 1.536924123764038}
+02/26/2022 10:04:00 - INFO - codeparrot_training - Step 43622: {'lr': 2.146745771618136e-05, 'samples': 22334976, 'steps': 43622, 'loss/train': 1.5509012937545776}
+02/26/2022 10:04:05 - INFO - codeparrot_training - Step 43623: {'lr': 2.146082452316131e-05, 'samples': 22335488, 'steps': 43623, 'loss/train': 1.5179985761642456}
+02/26/2022 10:04:09 - INFO - codeparrot_training - Step 43624: {'lr': 2.145419230913051e-05, 'samples': 22336000, 'steps': 43624, 'loss/train': 1.7410173416137695}
+02/26/2022 10:04:15 - INFO - codeparrot_training - Step 43625: {'lr': 2.144756107411733e-05, 'samples': 22336512, 'steps': 43625, 'loss/train': 1.8667746782302856}
+02/26/2022 10:04:18 - INFO - codeparrot_training - Step 43626: {'lr': 2.1440930818150295e-05, 'samples': 22337024, 'steps': 43626, 'loss/train': 1.4250669479370117}
+02/26/2022 10:04:24 - INFO - codeparrot_training - Step 43627: {'lr': 2.1434301541257667e-05, 'samples': 22337536, 'steps': 43627, 'loss/train': 1.4384822845458984}
+02/26/2022 10:04:27 - INFO - codeparrot_training - Step 43628: {'lr': 2.1427673243468004e-05, 'samples': 22338048, 'steps': 43628, 'loss/train': 1.5117721557617188}
+02/26/2022 10:04:33 - INFO - codeparrot_training - Step 43629: {'lr': 2.1421045924809508e-05, 'samples': 22338560, 'steps': 43629, 'loss/train': 1.7281244993209839}
+02/26/2022 10:04:37 - INFO - codeparrot_training - Step 43630: {'lr': 2.1414419585310735e-05, 'samples': 22339072, 'steps': 43630, 'loss/train': 1.583446741104126}
+02/26/2022 10:04:42 - INFO - codeparrot_training - Step 43631: {'lr': 2.1407794224999942e-05, 'samples': 22339584, 'steps': 43631, 'loss/train': 1.3835880756378174}
+02/26/2022 10:04:46 - INFO - codeparrot_training - Step 43632: {'lr': 2.1401169843905694e-05, 'samples': 22340096, 'steps': 43632, 'loss/train': 1.801163911819458}
+02/26/2022 10:04:51 - INFO - codeparrot_training - Step 43633: {'lr': 2.139454644205613e-05, 'samples': 22340608, 'steps': 43633, 'loss/train': 2.3483221530914307}
+02/26/2022 10:04:55 - INFO - codeparrot_training - Step 43634: {'lr': 2.138792401947978e-05, 'samples': 22341120, 'steps': 43634, 'loss/train': 2.1796042919158936}
+02/26/2022 10:05:00 - INFO - codeparrot_training - Step 43635: {'lr': 2.1381302576204937e-05, 'samples': 22341632, 'steps': 43635, 'loss/train': 1.7966972589492798}
+02/26/2022 10:05:04 - INFO - codeparrot_training - Step 43636: {'lr': 2.1374682112260097e-05, 'samples': 22342144, 'steps': 43636, 'loss/train': 1.9575128555297852}
+02/26/2022 10:05:09 - INFO - codeparrot_training - Step 43637: {'lr': 2.1368062627673405e-05, 'samples': 22342656, 'steps': 43637, 'loss/train': 2.6049258708953857}
+02/26/2022 10:05:13 - INFO - codeparrot_training - Step 43638: {'lr': 2.13614441224734e-05, 'samples': 22343168, 'steps': 43638, 'loss/train': 1.7215691804885864}
+02/26/2022 10:05:19 - INFO - codeparrot_training - Step 43639: {'lr': 2.135482659668833e-05, 'samples': 22343680, 'steps': 43639, 'loss/train': 1.0972094535827637}
+02/26/2022 10:05:22 - INFO - codeparrot_training - Step 43640: {'lr': 2.1348210050346596e-05, 'samples': 22344192, 'steps': 43640, 'loss/train': 1.209674596786499}
+02/26/2022 10:05:28 - INFO - codeparrot_training - Step 43641: {'lr': 2.1341594483476556e-05, 'samples': 22344704, 'steps': 43641, 'loss/train': 1.6548192501068115}
+02/26/2022 10:05:31 - INFO - codeparrot_training - Step 43642: {'lr': 2.13349798961065e-05, 'samples': 22345216, 'steps': 43642, 'loss/train': 1.3497061729431152}
+02/26/2022 10:05:37 - INFO - codeparrot_training - Step 43643: {'lr': 2.1328366288264734e-05, 'samples': 22345728, 'steps': 43643, 'loss/train': 1.309401273727417}
+02/26/2022 10:05:40 - INFO - codeparrot_training - Step 43644: {'lr': 2.132175365997968e-05, 'samples': 22346240, 'steps': 43644, 'loss/train': 1.948535442352295}
+02/26/2022 10:05:46 - INFO - codeparrot_training - Step 43645: {'lr': 2.1315142011279626e-05, 'samples': 22346752, 'steps': 43645, 'loss/train': 1.441284418106079}
+02/26/2022 10:05:49 - INFO - codeparrot_training - Step 43646: {'lr': 2.1308531342192904e-05, 'samples': 22347264, 'steps': 43646, 'loss/train': 1.3826335668563843}
+02/26/2022 10:05:55 - INFO - codeparrot_training - Step 43647: {'lr': 2.1301921652747747e-05, 'samples': 22347776, 'steps': 43647, 'loss/train': 1.5711207389831543}
+02/26/2022 10:05:58 - INFO - codeparrot_training - Step 43648: {'lr': 2.12953129429726e-05, 'samples': 22348288, 'steps': 43648, 'loss/train': 2.4665095806121826}
+02/26/2022 10:06:04 - INFO - codeparrot_training - Step 43649: {'lr': 2.1288705212895692e-05, 'samples': 22348800, 'steps': 43649, 'loss/train': 2.434887170791626}
+02/26/2022 10:06:07 - INFO - codeparrot_training - Step 43650: {'lr': 2.1282098462545362e-05, 'samples': 22349312, 'steps': 43650, 'loss/train': 1.5098457336425781}
+02/26/2022 10:06:13 - INFO - codeparrot_training - Step 43651: {'lr': 2.127549269194984e-05, 'samples': 22349824, 'steps': 43651, 'loss/train': 1.7913058996200562}
+02/26/2022 10:06:16 - INFO - codeparrot_training - Step 43652: {'lr': 2.126888790113754e-05, 'samples': 22350336, 'steps': 43652, 'loss/train': 0.6994002461433411}
+02/26/2022 10:06:22 - INFO - codeparrot_training - Step 43653: {'lr': 2.126228409013667e-05, 'samples': 22350848, 'steps': 43653, 'loss/train': 1.750333309173584}
+02/26/2022 10:06:26 - INFO - codeparrot_training - Step 43654: {'lr': 2.1255681258975562e-05, 'samples': 22351360, 'steps': 43654, 'loss/train': 1.9532947540283203}
+02/26/2022 10:06:31 - INFO - codeparrot_training - Step 43655: {'lr': 2.1249079407682476e-05, 'samples': 22351872, 'steps': 43655, 'loss/train': 1.5574061870574951}
+02/26/2022 10:06:35 - INFO - codeparrot_training - Step 43656: {'lr': 2.1242478536285638e-05, 'samples': 22352384, 'steps': 43656, 'loss/train': 2.4159815311431885}
+02/26/2022 10:06:40 - INFO - codeparrot_training - Step 43657: {'lr': 2.1235878644813437e-05, 'samples': 22352896, 'steps': 43657, 'loss/train': 1.7896896600723267}
+02/26/2022 10:06:44 - INFO - codeparrot_training - Step 43658: {'lr': 2.122927973329411e-05, 'samples': 22353408, 'steps': 43658, 'loss/train': 2.0833568572998047}
+02/26/2022 10:06:50 - INFO - codeparrot_training - Step 43659: {'lr': 2.122268180175588e-05, 'samples': 22353920, 'steps': 43659, 'loss/train': 2.0339841842651367}
+02/26/2022 10:06:53 - INFO - codeparrot_training - Step 43660: {'lr': 2.1216084850226998e-05, 'samples': 22354432, 'steps': 43660, 'loss/train': 1.7435036897659302}
+02/26/2022 10:06:58 - INFO - codeparrot_training - Step 43661: {'lr': 2.1209488878735833e-05, 'samples': 22354944, 'steps': 43661, 'loss/train': 2.865453004837036}
+02/26/2022 10:07:02 - INFO - codeparrot_training - Step 43662: {'lr': 2.1202893887310503e-05, 'samples': 22355456, 'steps': 43662, 'loss/train': 1.6757209300994873}
+02/26/2022 10:07:08 - INFO - codeparrot_training - Step 43663: {'lr': 2.1196299875979403e-05, 'samples': 22355968, 'steps': 43663, 'loss/train': 1.7064749002456665}
+02/26/2022 10:07:11 - INFO - codeparrot_training - Step 43664: {'lr': 2.1189706844770618e-05, 'samples': 22356480, 'steps': 43664, 'loss/train': 1.3960672616958618}
+02/26/2022 10:07:18 - INFO - codeparrot_training - Step 43665: {'lr': 2.1183114793712517e-05, 'samples': 22356992, 'steps': 43665, 'loss/train': 1.3924715518951416}
+02/26/2022 10:07:21 - INFO - codeparrot_training - Step 43666: {'lr': 2.1176523722833268e-05, 'samples': 22357504, 'steps': 43666, 'loss/train': 1.9153714179992676}
+02/26/2022 10:07:27 - INFO - codeparrot_training - Step 43667: {'lr': 2.1169933632161186e-05, 'samples': 22358016, 'steps': 43667, 'loss/train': 2.349346876144409}
+02/26/2022 10:07:30 - INFO - codeparrot_training - Step 43668: {'lr': 2.1163344521724385e-05, 'samples': 22358528, 'steps': 43668, 'loss/train': 2.1138269901275635}
+02/26/2022 10:07:36 - INFO - codeparrot_training - Step 43669: {'lr': 2.1156756391551178e-05, 'samples': 22359040, 'steps': 43669, 'loss/train': 1.7477502822875977}
+02/26/2022 10:07:39 - INFO - codeparrot_training - Step 43670: {'lr': 2.1150169241669736e-05, 'samples': 22359552, 'steps': 43670, 'loss/train': 2.5730812549591064}
+02/26/2022 10:07:45 - INFO - codeparrot_training - Step 43671: {'lr': 2.1143583072108398e-05, 'samples': 22360064, 'steps': 43671, 'loss/train': 0.9646165370941162}
+02/26/2022 10:07:48 - INFO - codeparrot_training - Step 43672: {'lr': 2.1136997882895166e-05, 'samples': 22360576, 'steps': 43672, 'loss/train': 1.162533164024353}
+02/26/2022 10:07:54 - INFO - codeparrot_training - Step 43673: {'lr': 2.1130413674058407e-05, 'samples': 22361088, 'steps': 43673, 'loss/train': 1.509287714958191}
+02/26/2022 10:07:57 - INFO - codeparrot_training - Step 43674: {'lr': 2.112383044562624e-05, 'samples': 22361600, 'steps': 43674, 'loss/train': 1.2716783285140991}
+02/26/2022 10:08:04 - INFO - codeparrot_training - Step 43675: {'lr': 2.1117248197626976e-05, 'samples': 22362112, 'steps': 43675, 'loss/train': 2.0076851844787598}
+02/26/2022 10:08:07 - INFO - codeparrot_training - Step 43676: {'lr': 2.111066693008873e-05, 'samples': 22362624, 'steps': 43676, 'loss/train': 1.412522315979004}
+02/26/2022 10:08:13 - INFO - codeparrot_training - Step 43677: {'lr': 2.11040866430397e-05, 'samples': 22363136, 'steps': 43677, 'loss/train': 0.7914162874221802}
+02/26/2022 10:08:16 - INFO - codeparrot_training - Step 43678: {'lr': 2.109750733650803e-05, 'samples': 22363648, 'steps': 43678, 'loss/train': 1.700441837310791}
+02/26/2022 10:08:22 - INFO - codeparrot_training - Step 43679: {'lr': 2.109092901052201e-05, 'samples': 22364160, 'steps': 43679, 'loss/train': 1.3637765645980835}
+02/26/2022 10:08:25 - INFO - codeparrot_training - Step 43680: {'lr': 2.1084351665109745e-05, 'samples': 22364672, 'steps': 43680, 'loss/train': 1.7694653272628784}
+02/26/2022 10:08:31 - INFO - codeparrot_training - Step 43681: {'lr': 2.1077775300299446e-05, 'samples': 22365184, 'steps': 43681, 'loss/train': 1.0543186664581299}
+02/26/2022 10:08:34 - INFO - codeparrot_training - Step 43682: {'lr': 2.107119991611922e-05, 'samples': 22365696, 'steps': 43682, 'loss/train': 3.1474647521972656}
+02/26/2022 10:08:40 - INFO - codeparrot_training - Step 43683: {'lr': 2.106462551259733e-05, 'samples': 22366208, 'steps': 43683, 'loss/train': 2.513465166091919}
+02/26/2022 10:08:43 - INFO - codeparrot_training - Step 43684: {'lr': 2.105805208976186e-05, 'samples': 22366720, 'steps': 43684, 'loss/train': 1.6875081062316895}
+02/26/2022 10:08:49 - INFO - codeparrot_training - Step 43685: {'lr': 2.1051479647641035e-05, 'samples': 22367232, 'steps': 43685, 'loss/train': 1.676844596862793}
+02/26/2022 10:08:53 - INFO - codeparrot_training - Step 43686: {'lr': 2.1044908186262895e-05, 'samples': 22367744, 'steps': 43686, 'loss/train': 3.0588529109954834}
+02/26/2022 10:08:58 - INFO - codeparrot_training - Step 43687: {'lr': 2.1038337705655748e-05, 'samples': 22368256, 'steps': 43687, 'loss/train': 1.7073297500610352}
+02/26/2022 10:09:02 - INFO - codeparrot_training - Step 43688: {'lr': 2.1031768205847624e-05, 'samples': 22368768, 'steps': 43688, 'loss/train': 1.9989266395568848}
+02/26/2022 10:09:07 - INFO - codeparrot_training - Step 43689: {'lr': 2.1025199686866725e-05, 'samples': 22369280, 'steps': 43689, 'loss/train': 2.059293270111084}
+02/26/2022 10:09:11 - INFO - codeparrot_training - Step 43690: {'lr': 2.1018632148741113e-05, 'samples': 22369792, 'steps': 43690, 'loss/train': 1.9883449077606201}
+02/26/2022 10:09:16 - INFO - codeparrot_training - Step 43691: {'lr': 2.1012065591499014e-05, 'samples': 22370304, 'steps': 43691, 'loss/train': 0.9303855299949646}
+02/26/2022 10:09:20 - INFO - codeparrot_training - Step 43692: {'lr': 2.1005500015168487e-05, 'samples': 22370816, 'steps': 43692, 'loss/train': 1.8114391565322876}
+02/26/2022 10:09:25 - INFO - codeparrot_training - Step 43693: {'lr': 2.099893541977771e-05, 'samples': 22371328, 'steps': 43693, 'loss/train': 1.7076948881149292}
+02/26/2022 10:09:29 - INFO - codeparrot_training - Step 43694: {'lr': 2.099237180535474e-05, 'samples': 22371840, 'steps': 43694, 'loss/train': 0.3772452473640442}
+02/26/2022 10:09:34 - INFO - codeparrot_training - Step 43695: {'lr': 2.0985809171927772e-05, 'samples': 22372352, 'steps': 43695, 'loss/train': 1.763005018234253}
+02/26/2022 10:09:40 - INFO - codeparrot_training - Step 43696: {'lr': 2.0979247519524876e-05, 'samples': 22372864, 'steps': 43696, 'loss/train': 1.1091537475585938}
+02/26/2022 10:09:43 - INFO - codeparrot_training - Step 43697: {'lr': 2.0972686848174133e-05, 'samples': 22373376, 'steps': 43697, 'loss/train': 2.7304844856262207}
+02/26/2022 10:09:47 - INFO - codeparrot_training - Step 43698: {'lr': 2.0966127157903665e-05, 'samples': 22373888, 'steps': 43698, 'loss/train': 1.645606517791748}
+02/26/2022 10:09:52 - INFO - codeparrot_training - Step 43699: {'lr': 2.095956844874161e-05, 'samples': 22374400, 'steps': 43699, 'loss/train': 1.3573448657989502}
+02/26/2022 10:09:56 - INFO - codeparrot_training - Step 43700: {'lr': 2.0953010720716037e-05, 'samples': 22374912, 'steps': 43700, 'loss/train': 1.6097822189331055}
+02/26/2022 10:10:02 - INFO - codeparrot_training - Step 43701: {'lr': 2.0946453973855e-05, 'samples': 22375424, 'steps': 43701, 'loss/train': 1.5295767784118652}
+02/26/2022 10:10:06 - INFO - codeparrot_training - Step 43702: {'lr': 2.0939898208186675e-05, 'samples': 22375936, 'steps': 43702, 'loss/train': 1.5707643032073975}
+02/26/2022 10:10:12 - INFO - codeparrot_training - Step 43703: {'lr': 2.0933343423739066e-05, 'samples': 22376448, 'steps': 43703, 'loss/train': 1.3826990127563477}
+02/26/2022 10:10:15 - INFO - codeparrot_training - Step 43704: {'lr': 2.092678962054029e-05, 'samples': 22376960, 'steps': 43704, 'loss/train': 2.2209463119506836}
+02/26/2022 10:10:21 - INFO - codeparrot_training - Step 43705: {'lr': 2.092023679861835e-05, 'samples': 22377472, 'steps': 43705, 'loss/train': 1.704289436340332}
+02/26/2022 10:10:24 - INFO - codeparrot_training - Step 43706: {'lr': 2.0913684958001472e-05, 'samples': 22377984, 'steps': 43706, 'loss/train': 1.503014087677002}
+02/26/2022 10:10:30 - INFO - codeparrot_training - Step 43707: {'lr': 2.0907134098717555e-05, 'samples': 22378496, 'steps': 43707, 'loss/train': 1.75319242477417}
+02/26/2022 10:10:34 - INFO - codeparrot_training - Step 43708: {'lr': 2.0900584220794744e-05, 'samples': 22379008, 'steps': 43708, 'loss/train': 2.362241744995117}
+02/26/2022 10:10:39 - INFO - codeparrot_training - Step 43709: {'lr': 2.0894035324261067e-05, 'samples': 22379520, 'steps': 43709, 'loss/train': 2.19525146484375}
+02/26/2022 10:10:43 - INFO - codeparrot_training - Step 43710: {'lr': 2.088748740914462e-05, 'samples': 22380032, 'steps': 43710, 'loss/train': 1.2324306964874268}
+02/26/2022 10:10:49 - INFO - codeparrot_training - Step 43711: {'lr': 2.0880940475473425e-05, 'samples': 22380544, 'steps': 43711, 'loss/train': 1.3014642000198364}
+02/26/2022 10:10:52 - INFO - codeparrot_training - Step 43712: {'lr': 2.0874394523275526e-05, 'samples': 22381056, 'steps': 43712, 'loss/train': 1.6936137676239014}
+02/26/2022 10:10:58 - INFO - codeparrot_training - Step 43713: {'lr': 2.086784955257895e-05, 'samples': 22381568, 'steps': 43713, 'loss/train': 1.50624418258667}
+02/26/2022 10:11:01 - INFO - codeparrot_training - Step 43714: {'lr': 2.0861305563411758e-05, 'samples': 22382080, 'steps': 43714, 'loss/train': 1.9275946617126465}
+02/26/2022 10:11:07 - INFO - codeparrot_training - Step 43715: {'lr': 2.0854762555801988e-05, 'samples': 22382592, 'steps': 43715, 'loss/train': 1.298649549484253}
+02/26/2022 10:11:10 - INFO - codeparrot_training - Step 43716: {'lr': 2.084822052977764e-05, 'samples': 22383104, 'steps': 43716, 'loss/train': 1.914510726928711}
+02/26/2022 10:11:16 - INFO - codeparrot_training - Step 43717: {'lr': 2.084167948536672e-05, 'samples': 22383616, 'steps': 43717, 'loss/train': 2.3082385063171387}
+02/26/2022 10:11:19 - INFO - codeparrot_training - Step 43718: {'lr': 2.083513942259732e-05, 'samples': 22384128, 'steps': 43718, 'loss/train': 2.0877203941345215}
+02/26/2022 10:11:25 - INFO - codeparrot_training - Step 43719: {'lr': 2.0828600341497415e-05, 'samples': 22384640, 'steps': 43719, 'loss/train': 2.9534261226654053}
+02/26/2022 10:11:28 - INFO - codeparrot_training - Step 43720: {'lr': 2.0822062242095014e-05, 'samples': 22385152, 'steps': 43720, 'loss/train': 0.9044831395149231}
+02/26/2022 10:11:34 - INFO - codeparrot_training - Step 43721: {'lr': 2.0815525124418087e-05, 'samples': 22385664, 'steps': 43721, 'loss/train': 2.0298373699188232}
+02/26/2022 10:11:38 - INFO - codeparrot_training - Step 43722: {'lr': 2.080898898849473e-05, 'samples': 22386176, 'steps': 43722, 'loss/train': 1.7623164653778076}
+02/26/2022 10:11:43 - INFO - codeparrot_training - Step 43723: {'lr': 2.0802453834352834e-05, 'samples': 22386688, 'steps': 43723, 'loss/train': 1.785817265510559}
+02/26/2022 10:11:47 - INFO - codeparrot_training - Step 43724: {'lr': 2.0795919662020518e-05, 'samples': 22387200, 'steps': 43724, 'loss/train': 1.8084228038787842}
+02/26/2022 10:11:52 - INFO - codeparrot_training - Step 43725: {'lr': 2.0789386471525647e-05, 'samples': 22387712, 'steps': 43725, 'loss/train': 0.9158523082733154}
+02/26/2022 10:11:56 - INFO - codeparrot_training - Step 43726: {'lr': 2.0782854262896277e-05, 'samples': 22388224, 'steps': 43726, 'loss/train': 1.596768856048584}
+02/26/2022 10:12:01 - INFO - codeparrot_training - Step 43727: {'lr': 2.0776323036160337e-05, 'samples': 22388736, 'steps': 43727, 'loss/train': 1.9071859121322632}
+02/26/2022 10:12:05 - INFO - codeparrot_training - Step 43728: {'lr': 2.0769792791345942e-05, 'samples': 22389248, 'steps': 43728, 'loss/train': 1.6885617971420288}
+02/26/2022 10:12:10 - INFO - codeparrot_training - Step 43729: {'lr': 2.0763263528480875e-05, 'samples': 22389760, 'steps': 43729, 'loss/train': 1.3484243154525757}
+02/26/2022 10:12:14 - INFO - codeparrot_training - Step 43730: {'lr': 2.0756735247593222e-05, 'samples': 22390272, 'steps': 43730, 'loss/train': 1.6374986171722412}
+02/26/2022 10:12:20 - INFO - codeparrot_training - Step 43731: {'lr': 2.075020794871091e-05, 'samples': 22390784, 'steps': 43731, 'loss/train': 1.3780603408813477}
+02/26/2022 10:12:24 - INFO - codeparrot_training - Step 43732: {'lr': 2.0743681631861965e-05, 'samples': 22391296, 'steps': 43732, 'loss/train': 1.3645790815353394}
+02/26/2022 10:12:29 - INFO - codeparrot_training - Step 43733: {'lr': 2.0737156297074233e-05, 'samples': 22391808, 'steps': 43733, 'loss/train': 2.2477939128875732}
+02/26/2022 10:12:33 - INFO - codeparrot_training - Step 43734: {'lr': 2.0730631944375745e-05, 'samples': 22392320, 'steps': 43734, 'loss/train': 0.0358416773378849}
+02/26/2022 10:12:38 - INFO - codeparrot_training - Step 43735: {'lr': 2.072410857379445e-05, 'samples': 22392832, 'steps': 43735, 'loss/train': 0.8678162693977356}
+02/26/2022 10:12:42 - INFO - codeparrot_training - Step 43736: {'lr': 2.071758618535821e-05, 'samples': 22393344, 'steps': 43736, 'loss/train': 1.9455173015594482}
+02/26/2022 10:12:47 - INFO - codeparrot_training - Step 43737: {'lr': 2.0711064779095068e-05, 'samples': 22393856, 'steps': 43737, 'loss/train': 2.4433224201202393}
+02/26/2022 10:12:51 - INFO - codeparrot_training - Step 43738: {'lr': 2.070454435503294e-05, 'samples': 22394368, 'steps': 43738, 'loss/train': 1.611151933670044}
+02/26/2022 10:12:56 - INFO - codeparrot_training - Step 43739: {'lr': 2.069802491319969e-05, 'samples': 22394880, 'steps': 43739, 'loss/train': 1.200068712234497}
+02/26/2022 10:13:00 - INFO - codeparrot_training - Step 43740: {'lr': 2.0691506453623277e-05, 'samples': 22395392, 'steps': 43740, 'loss/train': 1.3425055742263794}
+02/26/2022 10:13:05 - INFO - codeparrot_training - Step 43741: {'lr': 2.0684988976331696e-05, 'samples': 22395904, 'steps': 43741, 'loss/train': 1.4413142204284668}
+02/26/2022 10:13:09 - INFO - codeparrot_training - Step 43742: {'lr': 2.0678472481352793e-05, 'samples': 22396416, 'steps': 43742, 'loss/train': 1.354150652885437}
+02/26/2022 10:13:14 - INFO - codeparrot_training - Step 43743: {'lr': 2.067195696871449e-05, 'samples': 22396928, 'steps': 43743, 'loss/train': 2.612377405166626}
+02/26/2022 10:13:18 - INFO - codeparrot_training - Step 43744: {'lr': 2.066544243844465e-05, 'samples': 22397440, 'steps': 43744, 'loss/train': 0.7962895035743713}
+02/26/2022 10:13:23 - INFO - codeparrot_training - Step 43745: {'lr': 2.0658928890571277e-05, 'samples': 22397952, 'steps': 43745, 'loss/train': 1.8486801385879517}
+02/26/2022 10:13:26 - INFO - codeparrot_training - Step 43746: {'lr': 2.0652416325122243e-05, 'samples': 22398464, 'steps': 43746, 'loss/train': 1.0842667818069458}
+02/26/2022 10:13:33 - INFO - codeparrot_training - Step 43747: {'lr': 2.064590474212544e-05, 'samples': 22398976, 'steps': 43747, 'loss/train': 0.8362195491790771}
+02/26/2022 10:13:36 - INFO - codeparrot_training - Step 43748: {'lr': 2.0639394141608704e-05, 'samples': 22399488, 'steps': 43748, 'loss/train': 1.8332408666610718}
+02/26/2022 10:13:42 - INFO - codeparrot_training - Step 43749: {'lr': 2.0632884523599986e-05, 'samples': 22400000, 'steps': 43749, 'loss/train': 2.9326047897338867}
+02/26/2022 10:13:47 - INFO - codeparrot_training - Step 43750: {'lr': 2.0626375888127186e-05, 'samples': 22400512, 'steps': 43750, 'loss/train': 0.7321078181266785}
+02/26/2022 10:13:51 - INFO - codeparrot_training - Step 43751: {'lr': 2.0619868235218164e-05, 'samples': 22401024, 'steps': 43751, 'loss/train': 2.223775863647461}
+02/26/2022 10:13:56 - INFO - codeparrot_training - Step 43752: {'lr': 2.0613361564900735e-05, 'samples': 22401536, 'steps': 43752, 'loss/train': 1.9427775144577026}
+02/26/2022 10:14:00 - INFO - codeparrot_training - Step 43753: {'lr': 2.0606855877202873e-05, 'samples': 22402048, 'steps': 43753, 'loss/train': 2.0571823120117188}
+02/26/2022 10:14:06 - INFO - codeparrot_training - Step 43754: {'lr': 2.0600351172152392e-05, 'samples': 22402560, 'steps': 43754, 'loss/train': 1.2308913469314575}
+02/26/2022 10:14:09 - INFO - codeparrot_training - Step 43755: {'lr': 2.059384744977716e-05, 'samples': 22403072, 'steps': 43755, 'loss/train': 1.4520541429519653}
+02/26/2022 10:14:16 - INFO - codeparrot_training - Step 43756: {'lr': 2.0587344710105012e-05, 'samples': 22403584, 'steps': 43756, 'loss/train': 1.9432131052017212}
+02/26/2022 10:14:20 - INFO - codeparrot_training - Step 43757: {'lr': 2.058084295316387e-05, 'samples': 22404096, 'steps': 43757, 'loss/train': 1.9516907930374146}
+02/26/2022 10:14:25 - INFO - codeparrot_training - Step 43758: {'lr': 2.0574342178981547e-05, 'samples': 22404608, 'steps': 43758, 'loss/train': 2.027433156967163}
+02/26/2022 10:14:29 - INFO - codeparrot_training - Step 43759: {'lr': 2.056784238758591e-05, 'samples': 22405120, 'steps': 43759, 'loss/train': 1.9764513969421387}
+02/26/2022 10:14:34 - INFO - codeparrot_training - Step 43760: {'lr': 2.0561343579004716e-05, 'samples': 22405632, 'steps': 43760, 'loss/train': 2.2532684803009033}
+02/26/2022 10:14:38 - INFO - codeparrot_training - Step 43761: {'lr': 2.055484575326591e-05, 'samples': 22406144, 'steps': 43761, 'loss/train': 0.8994282484054565}
+02/26/2022 10:14:43 - INFO - codeparrot_training - Step 43762: {'lr': 2.054834891039728e-05, 'samples': 22406656, 'steps': 43762, 'loss/train': 1.210808277130127}
+02/26/2022 10:14:47 - INFO - codeparrot_training - Step 43763: {'lr': 2.0541853050426716e-05, 'samples': 22407168, 'steps': 43763, 'loss/train': 1.6927560567855835}
+02/26/2022 10:14:52 - INFO - codeparrot_training - Step 43764: {'lr': 2.0535358173381923e-05, 'samples': 22407680, 'steps': 43764, 'loss/train': 2.2049617767333984}
+02/26/2022 10:14:56 - INFO - codeparrot_training - Step 43765: {'lr': 2.052886427929082e-05, 'samples': 22408192, 'steps': 43765, 'loss/train': 2.969558000564575}
+02/26/2022 10:15:02 - INFO - codeparrot_training - Step 43766: {'lr': 2.0522371368181166e-05, 'samples': 22408704, 'steps': 43766, 'loss/train': 1.4443888664245605}
+02/26/2022 10:15:05 - INFO - codeparrot_training - Step 43767: {'lr': 2.0515879440080905e-05, 'samples': 22409216, 'steps': 43767, 'loss/train': 2.103226661682129}
+02/26/2022 10:15:11 - INFO - codeparrot_training - Step 43768: {'lr': 2.050938849501763e-05, 'samples': 22409728, 'steps': 43768, 'loss/train': 1.6351176500320435}
+02/26/2022 10:15:15 - INFO - codeparrot_training - Step 43769: {'lr': 2.0502898533019315e-05, 'samples': 22410240, 'steps': 43769, 'loss/train': 1.7441072463989258}
+02/26/2022 10:15:20 - INFO - codeparrot_training - Step 43770: {'lr': 2.0496409554113694e-05, 'samples': 22410752, 'steps': 43770, 'loss/train': 1.6881424188613892}
+02/26/2022 10:15:23 - INFO - codeparrot_training - Step 43771: {'lr': 2.04899215583286e-05, 'samples': 22411264, 'steps': 43771, 'loss/train': 2.434337615966797}
+02/26/2022 10:15:29 - INFO - codeparrot_training - Step 43772: {'lr': 2.0483434545691792e-05, 'samples': 22411776, 'steps': 43772, 'loss/train': 3.2211732864379883}
+02/26/2022 10:15:33 - INFO - codeparrot_training - Step 43773: {'lr': 2.0476948516231082e-05, 'samples': 22412288, 'steps': 43773, 'loss/train': 1.474689245223999}
+02/26/2022 10:15:38 - INFO - codeparrot_training - Step 43774: {'lr': 2.0470463469974193e-05, 'samples': 22412800, 'steps': 43774, 'loss/train': 2.4542195796966553}
+02/26/2022 10:15:42 - INFO - codeparrot_training - Step 43775: {'lr': 2.0463979406949023e-05, 'samples': 22413312, 'steps': 43775, 'loss/train': 1.8150604963302612}
+02/26/2022 10:15:47 - INFO - codeparrot_training - Step 43776: {'lr': 2.0457496327183245e-05, 'samples': 22413824, 'steps': 43776, 'loss/train': 1.1934990882873535}
+02/26/2022 10:15:51 - INFO - codeparrot_training - Step 43777: {'lr': 2.0451014230704696e-05, 'samples': 22414336, 'steps': 43777, 'loss/train': 2.025015354156494}
+02/26/2022 10:15:56 - INFO - codeparrot_training - Step 43778: {'lr': 2.044453311754105e-05, 'samples': 22414848, 'steps': 43778, 'loss/train': 1.4752459526062012}
+02/26/2022 10:16:00 - INFO - codeparrot_training - Step 43779: {'lr': 2.043805298772017e-05, 'samples': 22415360, 'steps': 43779, 'loss/train': 2.145583391189575}
+02/26/2022 10:16:05 - INFO - codeparrot_training - Step 43780: {'lr': 2.0431573841269812e-05, 'samples': 22415872, 'steps': 43780, 'loss/train': 1.887131690979004}
+02/26/2022 10:16:08 - INFO - codeparrot_training - Step 43781: {'lr': 2.0425095678217653e-05, 'samples': 22416384, 'steps': 43781, 'loss/train': 1.146106243133545}
+02/26/2022 10:16:15 - INFO - codeparrot_training - Step 43782: {'lr': 2.0418618498591474e-05, 'samples': 22416896, 'steps': 43782, 'loss/train': 2.37729811668396}
+02/26/2022 10:16:19 - INFO - codeparrot_training - Step 43783: {'lr': 2.0412142302419055e-05, 'samples': 22417408, 'steps': 43783, 'loss/train': 1.6423615217208862}
+02/26/2022 10:16:24 - INFO - codeparrot_training - Step 43784: {'lr': 2.040566708972813e-05, 'samples': 22417920, 'steps': 43784, 'loss/train': 2.652238607406616}
+02/26/2022 10:16:28 - INFO - codeparrot_training - Step 43785: {'lr': 2.0399192860546397e-05, 'samples': 22418432, 'steps': 43785, 'loss/train': 2.1094307899475098}
+02/26/2022 10:16:33 - INFO - codeparrot_training - Step 43786: {'lr': 2.0392719614901634e-05, 'samples': 22418944, 'steps': 43786, 'loss/train': 1.3287436962127686}
+02/26/2022 10:16:37 - INFO - codeparrot_training - Step 43787: {'lr': 2.0386247352821518e-05, 'samples': 22419456, 'steps': 43787, 'loss/train': 0.2299940288066864}
+02/26/2022 10:16:42 - INFO - codeparrot_training - Step 43788: {'lr': 2.0379776074333835e-05, 'samples': 22419968, 'steps': 43788, 'loss/train': 2.07014536857605}
+02/26/2022 10:16:46 - INFO - codeparrot_training - Step 43789: {'lr': 2.0373305779466305e-05, 'samples': 22420480, 'steps': 43789, 'loss/train': 1.465870976448059}
+02/26/2022 10:16:51 - INFO - codeparrot_training - Step 43790: {'lr': 2.036683646824658e-05, 'samples': 22420992, 'steps': 43790, 'loss/train': 1.9039493799209595}
+02/26/2022 10:16:55 - INFO - codeparrot_training - Step 43791: {'lr': 2.0360368140702386e-05, 'samples': 22421504, 'steps': 43791, 'loss/train': 1.6106352806091309}
+02/26/2022 10:17:02 - INFO - codeparrot_training - Step 43792: {'lr': 2.0353900796861504e-05, 'samples': 22422016, 'steps': 43792, 'loss/train': 2.552624464035034}
+02/26/2022 10:17:05 - INFO - codeparrot_training - Step 43793: {'lr': 2.0347434436751605e-05, 'samples': 22422528, 'steps': 43793, 'loss/train': 2.0767154693603516}
+02/26/2022 10:17:11 - INFO - codeparrot_training - Step 43794: {'lr': 2.034096906040034e-05, 'samples': 22423040, 'steps': 43794, 'loss/train': 1.2512662410736084}
+02/26/2022 10:17:14 - INFO - codeparrot_training - Step 43795: {'lr': 2.0334504667835403e-05, 'samples': 22423552, 'steps': 43795, 'loss/train': 1.1248472929000854}
+02/26/2022 10:17:19 - INFO - codeparrot_training - Step 43796: {'lr': 2.0328041259084578e-05, 'samples': 22424064, 'steps': 43796, 'loss/train': 2.4028737545013428}
+02/26/2022 10:17:23 - INFO - codeparrot_training - Step 43797: {'lr': 2.0321578834175457e-05, 'samples': 22424576, 'steps': 43797, 'loss/train': 1.6904635429382324}
+02/26/2022 10:17:28 - INFO - codeparrot_training - Step 43798: {'lr': 2.0315117393135822e-05, 'samples': 22425088, 'steps': 43798, 'loss/train': 1.0677117109298706}
+02/26/2022 10:17:32 - INFO - codeparrot_training - Step 43799: {'lr': 2.0308656935993237e-05, 'samples': 22425600, 'steps': 43799, 'loss/train': 1.1503819227218628}
+02/26/2022 10:17:37 - INFO - codeparrot_training - Step 43800: {'lr': 2.030219746277545e-05, 'samples': 22426112, 'steps': 43800, 'loss/train': 1.6439311504364014}
+02/26/2022 10:17:41 - INFO - codeparrot_training - Step 43801: {'lr': 2.0295738973510057e-05, 'samples': 22426624, 'steps': 43801, 'loss/train': 1.3753182888031006}
+02/26/2022 10:17:46 - INFO - codeparrot_training - Step 43802: {'lr': 2.0289281468224896e-05, 'samples': 22427136, 'steps': 43802, 'loss/train': 2.3771438598632812}
+02/26/2022 10:17:50 - INFO - codeparrot_training - Step 43803: {'lr': 2.0282824946947416e-05, 'samples': 22427648, 'steps': 43803, 'loss/train': 1.1201207637786865}
+02/26/2022 10:17:56 - INFO - codeparrot_training - Step 43804: {'lr': 2.0276369409705398e-05, 'samples': 22428160, 'steps': 43804, 'loss/train': 2.048480749130249}
+02/26/2022 10:18:00 - INFO - codeparrot_training - Step 43805: {'lr': 2.0269914856526434e-05, 'samples': 22428672, 'steps': 43805, 'loss/train': 1.1324502229690552}
+02/26/2022 10:18:05 - INFO - codeparrot_training - Step 43806: {'lr': 2.0263461287438306e-05, 'samples': 22429184, 'steps': 43806, 'loss/train': 2.0590100288391113}
+02/26/2022 10:18:09 - INFO - codeparrot_training - Step 43807: {'lr': 2.0257008702468467e-05, 'samples': 22429696, 'steps': 43807, 'loss/train': 1.0743722915649414}
+02/26/2022 10:18:14 - INFO - codeparrot_training - Step 43808: {'lr': 2.0250557101644697e-05, 'samples': 22430208, 'steps': 43808, 'loss/train': 1.3453279733657837}
+02/26/2022 10:18:18 - INFO - codeparrot_training - Step 43809: {'lr': 2.024410648499453e-05, 'samples': 22430720, 'steps': 43809, 'loss/train': 4.067195415496826}
+02/26/2022 10:18:23 - INFO - codeparrot_training - Step 43810: {'lr': 2.0237656852545692e-05, 'samples': 22431232, 'steps': 43810, 'loss/train': 2.2540249824523926}
+02/26/2022 10:18:27 - INFO - codeparrot_training - Step 43811: {'lr': 2.0231208204325806e-05, 'samples': 22431744, 'steps': 43811, 'loss/train': 0.5296192765235901}
+02/26/2022 10:18:32 - INFO - codeparrot_training - Step 43812: {'lr': 2.022476054036243e-05, 'samples': 22432256, 'steps': 43812, 'loss/train': 1.5596256256103516}
+02/26/2022 10:18:36 - INFO - codeparrot_training - Step 43813: {'lr': 2.0218313860683207e-05, 'samples': 22432768, 'steps': 43813, 'loss/train': 1.3284732103347778}
+02/26/2022 10:18:42 - INFO - codeparrot_training - Step 43814: {'lr': 2.0211868165315784e-05, 'samples': 22433280, 'steps': 43814, 'loss/train': 2.014993190765381}
+02/26/2022 10:18:45 - INFO - codeparrot_training - Step 43815: {'lr': 2.020542345428775e-05, 'samples': 22433792, 'steps': 43815, 'loss/train': 0.9022366404533386}
+02/26/2022 10:18:51 - INFO - codeparrot_training - Step 43816: {'lr': 2.0198979727626743e-05, 'samples': 22434304, 'steps': 43816, 'loss/train': 1.9093165397644043}
+02/26/2022 10:18:54 - INFO - codeparrot_training - Step 43817: {'lr': 2.019253698536025e-05, 'samples': 22434816, 'steps': 43817, 'loss/train': 0.9826906323432922}
+02/26/2022 10:19:00 - INFO - codeparrot_training - Step 43818: {'lr': 2.018609522751602e-05, 'samples': 22435328, 'steps': 43818, 'loss/train': 1.0435973405838013}
+02/26/2022 10:19:03 - INFO - codeparrot_training - Step 43819: {'lr': 2.0179654454121594e-05, 'samples': 22435840, 'steps': 43819, 'loss/train': 1.8581339120864868}
+02/26/2022 10:19:09 - INFO - codeparrot_training - Step 43820: {'lr': 2.0173214665204552e-05, 'samples': 22436352, 'steps': 43820, 'loss/train': 2.1677000522613525}
+02/26/2022 10:19:12 - INFO - codeparrot_training - Step 43821: {'lr': 2.0166775860792435e-05, 'samples': 22436864, 'steps': 43821, 'loss/train': 1.340701937675476}
+02/26/2022 10:19:18 - INFO - codeparrot_training - Step 43822: {'lr': 2.0160338040912912e-05, 'samples': 22437376, 'steps': 43822, 'loss/train': 0.30107632279396057}
+02/26/2022 10:19:21 - INFO - codeparrot_training - Step 43823: {'lr': 2.0153901205593517e-05, 'samples': 22437888, 'steps': 43823, 'loss/train': 1.8809938430786133}
+02/26/2022 10:19:27 - INFO - codeparrot_training - Step 43824: {'lr': 2.014746535486181e-05, 'samples': 22438400, 'steps': 43824, 'loss/train': 1.2629835605621338}
+02/26/2022 10:19:30 - INFO - codeparrot_training - Step 43825: {'lr': 2.0141030488745354e-05, 'samples': 22438912, 'steps': 43825, 'loss/train': 2.004812240600586}
+02/26/2022 10:19:36 - INFO - codeparrot_training - Step 43826: {'lr': 2.0134596607271765e-05, 'samples': 22439424, 'steps': 43826, 'loss/train': 2.4527204036712646}
+02/26/2022 10:19:39 - INFO - codeparrot_training - Step 43827: {'lr': 2.0128163710468577e-05, 'samples': 22439936, 'steps': 43827, 'loss/train': 2.2179901599884033}
+02/26/2022 10:19:45 - INFO - codeparrot_training - Step 43828: {'lr': 2.0121731798363324e-05, 'samples': 22440448, 'steps': 43828, 'loss/train': 2.125952959060669}
+02/26/2022 10:19:49 - INFO - codeparrot_training - Step 43829: {'lr': 2.0115300870983567e-05, 'samples': 22440960, 'steps': 43829, 'loss/train': 1.1136139631271362}
+02/26/2022 10:19:54 - INFO - codeparrot_training - Step 43830: {'lr': 2.0108870928356893e-05, 'samples': 22441472, 'steps': 43830, 'loss/train': 1.0993467569351196}
+02/26/2022 10:19:58 - INFO - codeparrot_training - Step 43831: {'lr': 2.010244197051081e-05, 'samples': 22441984, 'steps': 43831, 'loss/train': 2.5315487384796143}
+02/26/2022 10:20:03 - INFO - codeparrot_training - Step 43832: {'lr': 2.0096013997472823e-05, 'samples': 22442496, 'steps': 43832, 'loss/train': 2.315072536468506}
+02/26/2022 10:20:07 - INFO - codeparrot_training - Step 43833: {'lr': 2.0089587009270606e-05, 'samples': 22443008, 'steps': 43833, 'loss/train': 1.696114182472229}
+02/26/2022 10:20:12 - INFO - codeparrot_training - Step 43834: {'lr': 2.0083161005931522e-05, 'samples': 22443520, 'steps': 43834, 'loss/train': 1.0973819494247437}
+02/26/2022 10:20:16 - INFO - codeparrot_training - Step 43835: {'lr': 2.007673598748319e-05, 'samples': 22444032, 'steps': 43835, 'loss/train': 1.7770588397979736}
+02/26/2022 10:20:21 - INFO - codeparrot_training - Step 43836: {'lr': 2.0070311953953062e-05, 'samples': 22444544, 'steps': 43836, 'loss/train': 2.0600974559783936}
+02/26/2022 10:20:25 - INFO - codeparrot_training - Step 43837: {'lr': 2.006388890536881e-05, 'samples': 22445056, 'steps': 43837, 'loss/train': 1.2696826457977295}
+02/26/2022 10:20:30 - INFO - codeparrot_training - Step 43838: {'lr': 2.0057466841757744e-05, 'samples': 22445568, 'steps': 43838, 'loss/train': 1.8642808198928833}
+02/26/2022 10:20:34 - INFO - codeparrot_training - Step 43839: {'lr': 2.0051045763147535e-05, 'samples': 22446080, 'steps': 43839, 'loss/train': 1.5182468891143799}
+02/26/2022 10:20:40 - INFO - codeparrot_training - Step 43840: {'lr': 2.0044625669565582e-05, 'samples': 22446592, 'steps': 43840, 'loss/train': 1.523544192314148}
+02/26/2022 10:20:44 - INFO - codeparrot_training - Step 43841: {'lr': 2.0038206561039524e-05, 'samples': 22447104, 'steps': 43841, 'loss/train': 2.371044874191284}
+02/26/2022 10:20:49 - INFO - codeparrot_training - Step 43842: {'lr': 2.0031788437596676e-05, 'samples': 22447616, 'steps': 43842, 'loss/train': 0.7976179718971252}
+02/26/2022 10:20:53 - INFO - codeparrot_training - Step 43843: {'lr': 2.0025371299264655e-05, 'samples': 22448128, 'steps': 43843, 'loss/train': 1.791491985321045}
+02/26/2022 10:20:58 - INFO - codeparrot_training - Step 43844: {'lr': 2.0018955146070882e-05, 'samples': 22448640, 'steps': 43844, 'loss/train': 1.8317155838012695}
+02/26/2022 10:21:02 - INFO - codeparrot_training - Step 43845: {'lr': 2.0012539978042922e-05, 'samples': 22449152, 'steps': 43845, 'loss/train': 0.15922223031520844}
+02/26/2022 10:21:07 - INFO - codeparrot_training - Step 43846: {'lr': 2.000612579520822e-05, 'samples': 22449664, 'steps': 43846, 'loss/train': 1.7359062433242798}
+02/26/2022 10:21:11 - INFO - codeparrot_training - Step 43847: {'lr': 1.9999712597594227e-05, 'samples': 22450176, 'steps': 43847, 'loss/train': 1.3494253158569336}
+02/26/2022 10:21:16 - INFO - codeparrot_training - Step 43848: {'lr': 1.9993300385228396e-05, 'samples': 22450688, 'steps': 43848, 'loss/train': 1.2517229318618774}
+02/26/2022 10:21:20 - INFO - codeparrot_training - Step 43849: {'lr': 1.998688915813829e-05, 'samples': 22451200, 'steps': 43849, 'loss/train': 2.7270681858062744}
+02/26/2022 10:21:26 - INFO - codeparrot_training - Step 43850: {'lr': 1.9980478916351297e-05, 'samples': 22451712, 'steps': 43850, 'loss/train': 1.3634605407714844}
+02/26/2022 10:21:29 - INFO - codeparrot_training - Step 43851: {'lr': 1.99740696598949e-05, 'samples': 22452224, 'steps': 43851, 'loss/train': 2.0398809909820557}
+02/26/2022 10:21:35 - INFO - codeparrot_training - Step 43852: {'lr': 1.9967661388796493e-05, 'samples': 22452736, 'steps': 43852, 'loss/train': 1.5950738191604614}
+02/26/2022 10:21:40 - INFO - codeparrot_training - Step 43853: {'lr': 1.9961254103083638e-05, 'samples': 22453248, 'steps': 43853, 'loss/train': 2.6485507488250732}
+02/26/2022 10:21:44 - INFO - codeparrot_training - Step 43854: {'lr': 1.9954847802783698e-05, 'samples': 22453760, 'steps': 43854, 'loss/train': 2.2636215686798096}
+02/26/2022 10:21:47 - INFO - codeparrot_training - Step 43855: {'lr': 1.9948442487924156e-05, 'samples': 22454272, 'steps': 43855, 'loss/train': 1.1294500827789307}
+02/26/2022 10:21:53 - INFO - codeparrot_training - Step 43856: {'lr': 1.9942038158532405e-05, 'samples': 22454784, 'steps': 43856, 'loss/train': 1.655468225479126}
+02/26/2022 10:21:58 - INFO - codeparrot_training - Step 43857: {'lr': 1.993563481463595e-05, 'samples': 22455296, 'steps': 43857, 'loss/train': 0.7238036394119263}
+02/26/2022 10:22:02 - INFO - codeparrot_training - Step 43858: {'lr': 1.992923245626213e-05, 'samples': 22455808, 'steps': 43858, 'loss/train': 1.7627135515213013}
+02/26/2022 10:22:08 - INFO - codeparrot_training - Step 43859: {'lr': 1.9922831083438535e-05, 'samples': 22456320, 'steps': 43859, 'loss/train': 0.20615874230861664}
+02/26/2022 10:22:11 - INFO - codeparrot_training - Step 43860: {'lr': 1.9916430696192368e-05, 'samples': 22456832, 'steps': 43860, 'loss/train': 1.5028220415115356}
+02/26/2022 10:22:17 - INFO - codeparrot_training - Step 43861: {'lr': 1.9910031294551183e-05, 'samples': 22457344, 'steps': 43861, 'loss/train': 2.1154861450195312}
+02/26/2022 10:22:20 - INFO - codeparrot_training - Step 43862: {'lr': 1.9903632878542327e-05, 'samples': 22457856, 'steps': 43862, 'loss/train': 1.9556525945663452}
+02/26/2022 10:22:26 - INFO - codeparrot_training - Step 43863: {'lr': 1.9897235448193358e-05, 'samples': 22458368, 'steps': 43863, 'loss/train': 1.813613772392273}
+02/26/2022 10:22:29 - INFO - codeparrot_training - Step 43864: {'lr': 1.9890839003531446e-05, 'samples': 22458880, 'steps': 43864, 'loss/train': 2.005075454711914}
+02/26/2022 10:22:35 - INFO - codeparrot_training - Step 43865: {'lr': 1.9884443544584156e-05, 'samples': 22459392, 'steps': 43865, 'loss/train': 1.1705480813980103}
+02/26/2022 10:22:38 - INFO - codeparrot_training - Step 43866: {'lr': 1.9878049071378856e-05, 'samples': 22459904, 'steps': 43866, 'loss/train': 1.6140512228012085}
+02/26/2022 10:22:44 - INFO - codeparrot_training - Step 43867: {'lr': 1.9871655583942883e-05, 'samples': 22460416, 'steps': 43867, 'loss/train': 1.451699137687683}
+02/26/2022 10:22:47 - INFO - codeparrot_training - Step 43868: {'lr': 1.9865263082303687e-05, 'samples': 22460928, 'steps': 43868, 'loss/train': 1.4963024854660034}
+02/26/2022 10:22:53 - INFO - codeparrot_training - Step 43869: {'lr': 1.9858871566488633e-05, 'samples': 22461440, 'steps': 43869, 'loss/train': 0.3987034857273102}
+02/26/2022 10:22:56 - INFO - codeparrot_training - Step 43870: {'lr': 1.9852481036525123e-05, 'samples': 22461952, 'steps': 43870, 'loss/train': 1.4994840621948242}
+02/26/2022 10:23:02 - INFO - codeparrot_training - Step 43871: {'lr': 1.9846091492440434e-05, 'samples': 22462464, 'steps': 43871, 'loss/train': 1.8571969270706177}
+02/26/2022 10:23:05 - INFO - codeparrot_training - Step 43872: {'lr': 1.9839702934262045e-05, 'samples': 22462976, 'steps': 43872, 'loss/train': 0.23138749599456787}
+02/26/2022 10:23:11 - INFO - codeparrot_training - Step 43873: {'lr': 1.9833315362017297e-05, 'samples': 22463488, 'steps': 43873, 'loss/train': 1.7051478624343872}
+02/26/2022 10:23:15 - INFO - codeparrot_training - Step 43874: {'lr': 1.9826928775733527e-05, 'samples': 22464000, 'steps': 43874, 'loss/train': 1.7428762912750244}
+02/26/2022 10:23:20 - INFO - codeparrot_training - Step 43875: {'lr': 1.9820543175438076e-05, 'samples': 22464512, 'steps': 43875, 'loss/train': 1.8086543083190918}
+02/26/2022 10:23:24 - INFO - codeparrot_training - Step 43876: {'lr': 1.9814158561158368e-05, 'samples': 22465024, 'steps': 43876, 'loss/train': 2.5017173290252686}
+02/26/2022 10:23:29 - INFO - codeparrot_training - Step 43877: {'lr': 1.980777493292171e-05, 'samples': 22465536, 'steps': 43877, 'loss/train': 0.5445043444633484}
+02/26/2022 10:23:33 - INFO - codeparrot_training - Step 43878: {'lr': 1.9801392290755448e-05, 'samples': 22466048, 'steps': 43878, 'loss/train': 1.9074139595031738}
+02/26/2022 10:23:38 - INFO - codeparrot_training - Step 43879: {'lr': 1.979501063468689e-05, 'samples': 22466560, 'steps': 43879, 'loss/train': 0.796567440032959}
+02/26/2022 10:23:42 - INFO - codeparrot_training - Step 43880: {'lr': 1.9788629964743454e-05, 'samples': 22467072, 'steps': 43880, 'loss/train': 2.2069852352142334}
+02/26/2022 10:23:47 - INFO - codeparrot_training - Step 43881: {'lr': 1.97822502809524e-05, 'samples': 22467584, 'steps': 43881, 'loss/train': 1.8154195547103882}
+02/26/2022 10:23:51 - INFO - codeparrot_training - Step 43882: {'lr': 1.9775871583341127e-05, 'samples': 22468096, 'steps': 43882, 'loss/train': 1.9115105867385864}
+02/26/2022 10:23:56 - INFO - codeparrot_training - Step 43883: {'lr': 1.976949387193683e-05, 'samples': 22468608, 'steps': 43883, 'loss/train': 2.654982328414917}
+02/26/2022 10:24:00 - INFO - codeparrot_training - Step 43884: {'lr': 1.9763117146766985e-05, 'samples': 22469120, 'steps': 43884, 'loss/train': 2.313267230987549}
+02/26/2022 10:24:06 - INFO - codeparrot_training - Step 43885: {'lr': 1.9756741407858825e-05, 'samples': 22469632, 'steps': 43885, 'loss/train': 1.5166712999343872}
+02/26/2022 10:24:09 - INFO - codeparrot_training - Step 43886: {'lr': 1.9750366655239687e-05, 'samples': 22470144, 'steps': 43886, 'loss/train': 0.6908068060874939}
+02/26/2022 10:24:15 - INFO - codeparrot_training - Step 43887: {'lr': 1.9743992888936803e-05, 'samples': 22470656, 'steps': 43887, 'loss/train': 0.4037047028541565}
+02/26/2022 10:24:18 - INFO - codeparrot_training - Step 43888: {'lr': 1.973762010897759e-05, 'samples': 22471168, 'steps': 43888, 'loss/train': 2.2648537158966064}
+02/26/2022 10:24:24 - INFO - codeparrot_training - Step 43889: {'lr': 1.9731248315389306e-05, 'samples': 22471680, 'steps': 43889, 'loss/train': 1.0088422298431396}
+02/26/2022 10:24:27 - INFO - codeparrot_training - Step 43890: {'lr': 1.9724877508199235e-05, 'samples': 22472192, 'steps': 43890, 'loss/train': 1.3535374402999878}
+02/26/2022 10:24:33 - INFO - codeparrot_training - Step 43891: {'lr': 1.9718507687434606e-05, 'samples': 22472704, 'steps': 43891, 'loss/train': 2.247473955154419}
+02/26/2022 10:24:36 - INFO - codeparrot_training - Step 43892: {'lr': 1.971213885312284e-05, 'samples': 22473216, 'steps': 43892, 'loss/train': 2.134769916534424}
+02/26/2022 10:24:42 - INFO - codeparrot_training - Step 43893: {'lr': 1.9705771005291084e-05, 'samples': 22473728, 'steps': 43893, 'loss/train': 2.107015609741211}
+02/26/2022 10:24:46 - INFO - codeparrot_training - Step 43894: {'lr': 1.969940414396676e-05, 'samples': 22474240, 'steps': 43894, 'loss/train': 1.4998337030410767}
+02/26/2022 10:24:52 - INFO - codeparrot_training - Step 43895: {'lr': 1.9693038269176984e-05, 'samples': 22474752, 'steps': 43895, 'loss/train': 1.8322309255599976}
+02/26/2022 10:24:55 - INFO - codeparrot_training - Step 43896: {'lr': 1.968667338094915e-05, 'samples': 22475264, 'steps': 43896, 'loss/train': 2.7887208461761475}
+02/26/2022 10:25:01 - INFO - codeparrot_training - Step 43897: {'lr': 1.9680309479310405e-05, 'samples': 22475776, 'steps': 43897, 'loss/train': 2.078357696533203}
+02/26/2022 10:25:04 - INFO - codeparrot_training - Step 43898: {'lr': 1.9673946564288197e-05, 'samples': 22476288, 'steps': 43898, 'loss/train': 2.2084178924560547}
+02/26/2022 10:25:10 - INFO - codeparrot_training - Step 43899: {'lr': 1.9667584635909562e-05, 'samples': 22476800, 'steps': 43899, 'loss/train': 1.5692343711853027}
+02/26/2022 10:25:13 - INFO - codeparrot_training - Step 43900: {'lr': 1.9661223694201895e-05, 'samples': 22477312, 'steps': 43900, 'loss/train': 1.3374872207641602}
+02/26/2022 10:25:19 - INFO - codeparrot_training - Step 43901: {'lr': 1.9654863739192367e-05, 'samples': 22477824, 'steps': 43901, 'loss/train': 0.5901874303817749}
+02/26/2022 10:25:22 - INFO - codeparrot_training - Step 43902: {'lr': 1.964850477090835e-05, 'samples': 22478336, 'steps': 43902, 'loss/train': 1.7448763847351074}
+02/26/2022 10:25:28 - INFO - codeparrot_training - Step 43903: {'lr': 1.9642146789376897e-05, 'samples': 22478848, 'steps': 43903, 'loss/train': 1.2017637491226196}
+02/26/2022 10:25:31 - INFO - codeparrot_training - Step 43904: {'lr': 1.963578979462541e-05, 'samples': 22479360, 'steps': 43904, 'loss/train': 0.21360686421394348}
+02/26/2022 10:25:37 - INFO - codeparrot_training - Step 43905: {'lr': 1.962943378668097e-05, 'samples': 22479872, 'steps': 43905, 'loss/train': 2.2890353202819824}
+02/26/2022 10:25:41 - INFO - codeparrot_training - Step 43906: {'lr': 1.9623078765570958e-05, 'samples': 22480384, 'steps': 43906, 'loss/train': 1.4535874128341675}
+02/26/2022 10:25:46 - INFO - codeparrot_training - Step 43907: {'lr': 1.9616724731322504e-05, 'samples': 22480896, 'steps': 43907, 'loss/train': 2.1506361961364746}
+02/26/2022 10:25:50 - INFO - codeparrot_training - Step 43908: {'lr': 1.9610371683962874e-05, 'samples': 22481408, 'steps': 43908, 'loss/train': 2.1257059574127197}
+02/26/2022 10:25:55 - INFO - codeparrot_training - Step 43909: {'lr': 1.960401962351918e-05, 'samples': 22481920, 'steps': 43909, 'loss/train': 1.7139891386032104}
+02/26/2022 10:25:59 - INFO - codeparrot_training - Step 43910: {'lr': 1.9597668550018767e-05, 'samples': 22482432, 'steps': 43910, 'loss/train': 1.9152476787567139}
+02/26/2022 10:26:04 - INFO - codeparrot_training - Step 43911: {'lr': 1.959131846348877e-05, 'samples': 22482944, 'steps': 43911, 'loss/train': 2.175318956375122}
+02/26/2022 10:26:08 - INFO - codeparrot_training - Step 43912: {'lr': 1.958496936395643e-05, 'samples': 22483456, 'steps': 43912, 'loss/train': 1.6276425123214722}
+02/26/2022 10:26:13 - INFO - codeparrot_training - Step 43913: {'lr': 1.957862125144888e-05, 'samples': 22483968, 'steps': 43913, 'loss/train': 1.1410492658615112}
+02/26/2022 10:26:17 - INFO - codeparrot_training - Step 43914: {'lr': 1.9572274125993326e-05, 'samples': 22484480, 'steps': 43914, 'loss/train': 2.458616256713867}
+02/26/2022 10:26:25 - INFO - codeparrot_training - Step 43915: {'lr': 1.9565927987617027e-05, 'samples': 22484992, 'steps': 43915, 'loss/train': 0.8795093894004822}
+02/26/2022 10:26:28 - INFO - codeparrot_training - Step 43916: {'lr': 1.9559582836347094e-05, 'samples': 22485504, 'steps': 43916, 'loss/train': 2.233673572540283}
+02/26/2022 10:26:32 - INFO - codeparrot_training - Step 43917: {'lr': 1.955323867221076e-05, 'samples': 22486016, 'steps': 43917, 'loss/train': 1.1261332035064697}
+02/26/2022 10:26:37 - INFO - codeparrot_training - Step 43918: {'lr': 1.954689549523514e-05, 'samples': 22486528, 'steps': 43918, 'loss/train': 1.9638397693634033}
+02/26/2022 10:26:41 - INFO - codeparrot_training - Step 43919: {'lr': 1.9540553305447462e-05, 'samples': 22487040, 'steps': 43919, 'loss/train': 2.019672155380249}
+02/26/2022 10:26:46 - INFO - codeparrot_training - Step 43920: {'lr': 1.95342121028749e-05, 'samples': 22487552, 'steps': 43920, 'loss/train': 1.5649964809417725}
+02/26/2022 10:26:50 - INFO - codeparrot_training - Step 43921: {'lr': 1.952787188754457e-05, 'samples': 22488064, 'steps': 43921, 'loss/train': 2.075366497039795}
+02/26/2022 10:26:56 - INFO - codeparrot_training - Step 43922: {'lr': 1.9521532659483614e-05, 'samples': 22488576, 'steps': 43922, 'loss/train': 0.8432846665382385}
+02/26/2022 10:26:59 - INFO - codeparrot_training - Step 43923: {'lr': 1.9515194418719268e-05, 'samples': 22489088, 'steps': 43923, 'loss/train': 1.9711413383483887}
+02/26/2022 10:27:05 - INFO - codeparrot_training - Step 43924: {'lr': 1.950885716527867e-05, 'samples': 22489600, 'steps': 43924, 'loss/train': 1.4739127159118652}
+02/26/2022 10:27:08 - INFO - codeparrot_training - Step 43925: {'lr': 1.9502520899188913e-05, 'samples': 22490112, 'steps': 43925, 'loss/train': 2.1115334033966064}
+02/26/2022 10:27:15 - INFO - codeparrot_training - Step 43926: {'lr': 1.9496185620477113e-05, 'samples': 22490624, 'steps': 43926, 'loss/train': 2.2978007793426514}
+02/26/2022 10:27:19 - INFO - codeparrot_training - Step 43927: {'lr': 1.9489851329170526e-05, 'samples': 22491136, 'steps': 43927, 'loss/train': 1.0188546180725098}
+02/26/2022 10:27:24 - INFO - codeparrot_training - Step 43928: {'lr': 1.948351802529616e-05, 'samples': 22491648, 'steps': 43928, 'loss/train': 2.3698296546936035}
+02/26/2022 10:27:28 - INFO - codeparrot_training - Step 43929: {'lr': 1.9477185708881297e-05, 'samples': 22492160, 'steps': 43929, 'loss/train': 1.9153876304626465}
+02/26/2022 10:27:33 - INFO - codeparrot_training - Step 43930: {'lr': 1.9470854379952913e-05, 'samples': 22492672, 'steps': 43930, 'loss/train': 1.6623371839523315}
+02/26/2022 10:27:37 - INFO - codeparrot_training - Step 43931: {'lr': 1.9464524038538185e-05, 'samples': 22493184, 'steps': 43931, 'loss/train': 1.2903441190719604}
+02/26/2022 10:27:42 - INFO - codeparrot_training - Step 43932: {'lr': 1.9458194684664226e-05, 'samples': 22493696, 'steps': 43932, 'loss/train': 1.928707242012024}
+02/26/2022 10:27:46 - INFO - codeparrot_training - Step 43933: {'lr': 1.9451866318358214e-05, 'samples': 22494208, 'steps': 43933, 'loss/train': 0.9612748026847839}
+02/26/2022 10:27:51 - INFO - codeparrot_training - Step 43934: {'lr': 1.9445538939647122e-05, 'samples': 22494720, 'steps': 43934, 'loss/train': 1.7725878953933716}
+02/26/2022 10:27:59 - INFO - codeparrot_training - Step 43935: {'lr': 1.9439212548558182e-05, 'samples': 22495232, 'steps': 43935, 'loss/train': 2.034044027328491}
+02/26/2022 10:28:02 - INFO - codeparrot_training - Step 43936: {'lr': 1.94328871451184e-05, 'samples': 22495744, 'steps': 43936, 'loss/train': 1.49234139919281}
+02/26/2022 10:28:08 - INFO - codeparrot_training - Step 43937: {'lr': 1.9426562729355003e-05, 'samples': 22496256, 'steps': 43937, 'loss/train': 1.8445290327072144}
+02/26/2022 10:28:11 - INFO - codeparrot_training - Step 43938: {'lr': 1.9420239301294912e-05, 'samples': 22496768, 'steps': 43938, 'loss/train': 2.1561474800109863}
+02/26/2022 10:28:17 - INFO - codeparrot_training - Step 43939: {'lr': 1.941391686096533e-05, 'samples': 22497280, 'steps': 43939, 'loss/train': 0.9168580174446106}
+02/26/2022 10:28:20 - INFO - codeparrot_training - Step 43940: {'lr': 1.940759540839329e-05, 'samples': 22497792, 'steps': 43940, 'loss/train': 2.79150652885437}
+02/26/2022 10:28:26 - INFO - codeparrot_training - Step 43941: {'lr': 1.9401274943605913e-05, 'samples': 22498304, 'steps': 43941, 'loss/train': 1.4240717887878418}
+02/26/2022 10:28:29 - INFO - codeparrot_training - Step 43942: {'lr': 1.9394955466630254e-05, 'samples': 22498816, 'steps': 43942, 'loss/train': 0.761337399482727}
+02/26/2022 10:28:35 - INFO - codeparrot_training - Step 43943: {'lr': 1.938863697749338e-05, 'samples': 22499328, 'steps': 43943, 'loss/train': 2.0061745643615723}
+02/26/2022 10:28:38 - INFO - codeparrot_training - Step 43944: {'lr': 1.938231947622232e-05, 'samples': 22499840, 'steps': 43944, 'loss/train': 1.974061369895935}
+02/26/2022 10:28:44 - INFO - codeparrot_training - Step 43945: {'lr': 1.9376002962844195e-05, 'samples': 22500352, 'steps': 43945, 'loss/train': 0.44279757142066956}
+02/26/2022 10:28:47 - INFO - codeparrot_training - Step 43946: {'lr': 1.936968743738607e-05, 'samples': 22500864, 'steps': 43946, 'loss/train': 1.695180058479309}
+02/26/2022 10:28:54 - INFO - codeparrot_training - Step 43947: {'lr': 1.9363372899874944e-05, 'samples': 22501376, 'steps': 43947, 'loss/train': 2.4326558113098145}
+02/26/2022 10:28:58 - INFO - codeparrot_training - Step 43948: {'lr': 1.935705935033788e-05, 'samples': 22501888, 'steps': 43948, 'loss/train': 1.4776768684387207}
+02/26/2022 10:29:03 - INFO - codeparrot_training - Step 43949: {'lr': 1.935074678880197e-05, 'samples': 22502400, 'steps': 43949, 'loss/train': 1.9536120891571045}
+02/26/2022 10:29:07 - INFO - codeparrot_training - Step 43950: {'lr': 1.9344435215294193e-05, 'samples': 22502912, 'steps': 43950, 'loss/train': 1.8860721588134766}
+02/26/2022 10:29:12 - INFO - codeparrot_training - Step 43951: {'lr': 1.9338124629841664e-05, 'samples': 22503424, 'steps': 43951, 'loss/train': 0.06276270002126694}
+02/26/2022 10:29:16 - INFO - codeparrot_training - Step 43952: {'lr': 1.9331815032471277e-05, 'samples': 22503936, 'steps': 43952, 'loss/train': 1.4646512269973755}
+02/26/2022 10:29:21 - INFO - codeparrot_training - Step 43953: {'lr': 1.932550642321021e-05, 'samples': 22504448, 'steps': 43953, 'loss/train': 1.6647932529449463}
+02/26/2022 10:29:25 - INFO - codeparrot_training - Step 43954: {'lr': 1.9319198802085436e-05, 'samples': 22504960, 'steps': 43954, 'loss/train': 1.5435038805007935}
+02/26/2022 10:29:32 - INFO - codeparrot_training - Step 43955: {'lr': 1.9312892169123962e-05, 'samples': 22505472, 'steps': 43955, 'loss/train': 1.6119916439056396}
+02/26/2022 10:29:36 - INFO - codeparrot_training - Step 43956: {'lr': 1.930658652435274e-05, 'samples': 22505984, 'steps': 43956, 'loss/train': 2.1861534118652344}
+02/26/2022 10:29:41 - INFO - codeparrot_training - Step 43957: {'lr': 1.930028186779892e-05, 'samples': 22506496, 'steps': 43957, 'loss/train': 1.6345672607421875}
+02/26/2022 10:29:44 - INFO - codeparrot_training - Step 43958: {'lr': 1.9293978199489414e-05, 'samples': 22507008, 'steps': 43958, 'loss/train': 1.0778197050094604}
+02/26/2022 10:29:50 - INFO - codeparrot_training - Step 43959: {'lr': 1.9287675519451266e-05, 'samples': 22507520, 'steps': 43959, 'loss/train': 1.6340985298156738}
+02/26/2022 10:29:53 - INFO - codeparrot_training - Step 43960: {'lr': 1.9281373827711446e-05, 'samples': 22508032, 'steps': 43960, 'loss/train': 0.7536647915840149}
+02/26/2022 10:29:59 - INFO - codeparrot_training - Step 43961: {'lr': 1.927507312429694e-05, 'samples': 22508544, 'steps': 43961, 'loss/train': 0.5697888135910034}
+02/26/2022 10:30:03 - INFO - codeparrot_training - Step 43962: {'lr': 1.9268773409234773e-05, 'samples': 22509056, 'steps': 43962, 'loss/train': 2.946394205093384}
+02/26/2022 10:30:08 - INFO - codeparrot_training - Step 43963: {'lr': 1.9262474682551905e-05, 'samples': 22509568, 'steps': 43963, 'loss/train': 1.8599181175231934}
+02/26/2022 10:30:12 - INFO - codeparrot_training - Step 43964: {'lr': 1.9256176944275367e-05, 'samples': 22510080, 'steps': 43964, 'loss/train': 1.7560877799987793}
+02/26/2022 10:30:17 - INFO - codeparrot_training - Step 43965: {'lr': 1.9249880194432024e-05, 'samples': 22510592, 'steps': 43965, 'loss/train': 2.406370162963867}
+02/26/2022 10:30:21 - INFO - codeparrot_training - Step 43966: {'lr': 1.9243584433048965e-05, 'samples': 22511104, 'steps': 43966, 'loss/train': 2.0007286071777344}
+02/26/2022 10:30:26 - INFO - codeparrot_training - Step 43967: {'lr': 1.923728966015309e-05, 'samples': 22511616, 'steps': 43967, 'loss/train': 1.6675264835357666}
+02/26/2022 10:30:30 - INFO - codeparrot_training - Step 43968: {'lr': 1.9230995875771483e-05, 'samples': 22512128, 'steps': 43968, 'loss/train': 1.068717360496521}
+02/26/2022 10:30:35 - INFO - codeparrot_training - Step 43969: {'lr': 1.9224703079930905e-05, 'samples': 22512640, 'steps': 43969, 'loss/train': 1.3827135562896729}
+02/26/2022 10:30:39 - INFO - codeparrot_training - Step 43970: {'lr': 1.921841127265847e-05, 'samples': 22513152, 'steps': 43970, 'loss/train': 0.5793749690055847}
+02/26/2022 10:30:46 - INFO - codeparrot_training - Step 43971: {'lr': 1.9212120453981047e-05, 'samples': 22513664, 'steps': 43971, 'loss/train': 1.5587520599365234}
+02/26/2022 10:30:50 - INFO - codeparrot_training - Step 43972: {'lr': 1.9205830623925696e-05, 'samples': 22514176, 'steps': 43972, 'loss/train': 1.8233580589294434}
+02/26/2022 10:30:55 - INFO - codeparrot_training - Step 43973: {'lr': 1.9199541782519174e-05, 'samples': 22514688, 'steps': 43973, 'loss/train': 0.9353955388069153}
+02/26/2022 10:30:59 - INFO - codeparrot_training - Step 43974: {'lr': 1.91932539297886e-05, 'samples': 22515200, 'steps': 43974, 'loss/train': 1.533398985862732}
+02/26/2022 10:31:04 - INFO - codeparrot_training - Step 43975: {'lr': 1.9186967065760786e-05, 'samples': 22515712, 'steps': 43975, 'loss/train': 2.2748358249664307}
+02/26/2022 10:31:07 - INFO - codeparrot_training - Step 43976: {'lr': 1.9180681190462763e-05, 'samples': 22516224, 'steps': 43976, 'loss/train': 1.627120018005371}
+02/26/2022 10:31:13 - INFO - codeparrot_training - Step 43977: {'lr': 1.9174396303921398e-05, 'samples': 22516736, 'steps': 43977, 'loss/train': 1.7510101795196533}
+02/26/2022 10:31:17 - INFO - codeparrot_training - Step 43978: {'lr': 1.9168112406163618e-05, 'samples': 22517248, 'steps': 43978, 'loss/train': 0.9625411629676819}
+02/26/2022 10:31:22 - INFO - codeparrot_training - Step 43979: {'lr': 1.9161829497216315e-05, 'samples': 22517760, 'steps': 43979, 'loss/train': 2.05888032913208}
+02/26/2022 10:31:25 - INFO - codeparrot_training - Step 43980: {'lr': 1.915554757710647e-05, 'samples': 22518272, 'steps': 43980, 'loss/train': 1.0286204814910889}
+02/26/2022 10:31:33 - INFO - codeparrot_training - Step 43981: {'lr': 1.9149266645860976e-05, 'samples': 22518784, 'steps': 43981, 'loss/train': 1.733374834060669}
+02/26/2022 10:31:36 - INFO - codeparrot_training - Step 43982: {'lr': 1.91429867035067e-05, 'samples': 22519296, 'steps': 43982, 'loss/train': 2.0034515857696533}
+02/26/2022 10:31:42 - INFO - codeparrot_training - Step 43983: {'lr': 1.9136707750070536e-05, 'samples': 22519808, 'steps': 43983, 'loss/train': 2.4939634799957275}
+02/26/2022 10:31:45 - INFO - codeparrot_training - Step 43984: {'lr': 1.9130429785579438e-05, 'samples': 22520320, 'steps': 43984, 'loss/train': 0.7421519160270691}
+02/26/2022 10:31:51 - INFO - codeparrot_training - Step 43985: {'lr': 1.9124152810060275e-05, 'samples': 22520832, 'steps': 43985, 'loss/train': 3.264716386795044}
+02/26/2022 10:31:54 - INFO - codeparrot_training - Step 43986: {'lr': 1.911787682353991e-05, 'samples': 22521344, 'steps': 43986, 'loss/train': 1.2016572952270508}
+02/26/2022 10:32:00 - INFO - codeparrot_training - Step 43987: {'lr': 1.9111601826045237e-05, 'samples': 22521856, 'steps': 43987, 'loss/train': 1.7218455076217651}
+02/26/2022 10:32:03 - INFO - codeparrot_training - Step 43988: {'lr': 1.9105327817603186e-05, 'samples': 22522368, 'steps': 43988, 'loss/train': 1.9870600700378418}
+02/26/2022 10:32:09 - INFO - codeparrot_training - Step 43989: {'lr': 1.9099054798240535e-05, 'samples': 22522880, 'steps': 43989, 'loss/train': 1.014692783355713}
+02/26/2022 10:32:12 - INFO - codeparrot_training - Step 43990: {'lr': 1.909278276798429e-05, 'samples': 22523392, 'steps': 43990, 'loss/train': 1.6703882217407227}
+02/26/2022 10:32:18 - INFO - codeparrot_training - Step 43991: {'lr': 1.9086511726861182e-05, 'samples': 22523904, 'steps': 43991, 'loss/train': 2.2936129570007324}
+02/26/2022 10:32:21 - INFO - codeparrot_training - Step 43992: {'lr': 1.9080241674898162e-05, 'samples': 22524416, 'steps': 43992, 'loss/train': 1.3780274391174316}
+02/26/2022 10:32:28 - INFO - codeparrot_training - Step 43993: {'lr': 1.9073972612122072e-05, 'samples': 22524928, 'steps': 43993, 'loss/train': 1.5659040212631226}
+02/26/2022 10:32:32 - INFO - codeparrot_training - Step 43994: {'lr': 1.9067704538559745e-05, 'samples': 22525440, 'steps': 43994, 'loss/train': 1.0398471355438232}
+02/26/2022 10:32:37 - INFO - codeparrot_training - Step 43995: {'lr': 1.9061437454238025e-05, 'samples': 22525952, 'steps': 43995, 'loss/train': 1.5446808338165283}
+02/26/2022 10:32:41 - INFO - codeparrot_training - Step 43996: {'lr': 1.9055171359183777e-05, 'samples': 22526464, 'steps': 43996, 'loss/train': 2.550922393798828}
+02/26/2022 10:32:46 - INFO - codeparrot_training - Step 43997: {'lr': 1.904890625342387e-05, 'samples': 22526976, 'steps': 43997, 'loss/train': 2.6649065017700195}
+02/26/2022 10:32:50 - INFO - codeparrot_training - Step 43998: {'lr': 1.9042642136985117e-05, 'samples': 22527488, 'steps': 43998, 'loss/train': 1.2582118511199951}
+02/26/2022 10:32:55 - INFO - codeparrot_training - Step 43999: {'lr': 1.90363790098943e-05, 'samples': 22528000, 'steps': 43999, 'loss/train': 1.9502559900283813}
+02/26/2022 10:32:55 - INFO - codeparrot_training - Evaluating and saving model checkpoint