diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -40542,3 +40542,1009 @@ Use FP16 precision: False
 02/26/2022 05:21:17 - INFO - codeparrot_training - Step 39998: {'lr': 5.1681588297549986e-05, 'samples': 20479488, 'steps': 39998, 'loss/train': 1.0893386602401733}
 02/26/2022 05:21:23 - INFO - codeparrot_training - Step 39999: {'lr': 5.1671626187582723e-05, 'samples': 20480000, 'steps': 39999, 'loss/train': 2.366724729537964}
 02/26/2022 05:21:23 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 05:21:41 - WARNING - huggingface_hub.repository - Several commits (40) will be pushed upstream.
+02/26/2022 05:21:41 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 05:22:15 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   d2106d5..e25c2d7  floral-grass-11 -> floral-grass-11
+
+02/26/2022 05:22:22 - INFO - codeparrot_training - Step 40000: {'lr': 5.1661664927191235e-05, 'samples': 20480512, 'steps': 40000, 'loss/train': 2.0706076622009277}
+02/26/2022 05:22:25 - INFO - codeparrot_training - Step 40001: {'lr': 5.165170451641815e-05, 'samples': 20481024, 'steps': 40001, 'loss/train': 2.458712577819824}
+02/26/2022 05:22:32 - INFO - codeparrot_training - Step 40002: {'lr': 5.164174495530621e-05, 'samples': 20481536, 'steps': 40002, 'loss/train': 2.227252244949341}
+02/26/2022 05:22:35 - INFO - codeparrot_training - Step 40003: {'lr': 5.163178624389808e-05, 'samples': 20482048, 'steps': 40003, 'loss/train': 2.0496666431427}
+02/26/2022 05:22:39 - INFO - codeparrot_training - Step 40004: {'lr': 5.162182838223639e-05, 'samples': 20482560, 'steps': 40004, 'loss/train': 1.467111349105835}
+02/26/2022 05:22:44 - INFO - codeparrot_training - Step 40005: {'lr': 5.1611871370363774e-05, 'samples': 20483072, 'steps': 40005, 'loss/train': 1.0251861810684204}
+02/26/2022 05:22:48 - INFO - codeparrot_training - Step 40006: {'lr': 5.160191520832283e-05, 'samples': 20483584, 'steps': 40006, 'loss/train': 0.993036687374115}
+02/26/2022 05:22:53 - INFO - codeparrot_training - Step 40007: {'lr': 5.159195989615639e-05, 'samples': 20484096, 'steps': 40007, 'loss/train': 0.9503242373466492}
+02/26/2022 05:22:57 - INFO - codeparrot_training - Step 40008: {'lr': 5.1582005433906954e-05, 'samples': 20484608, 'steps': 40008, 'loss/train': 2.1146693229675293}
+02/26/2022 05:23:02 - INFO - codeparrot_training - Step 40009: {'lr': 5.157205182161723e-05, 'samples': 20485120, 'steps': 40009, 'loss/train': 1.5174983739852905}
+02/26/2022 05:23:08 - INFO - codeparrot_training - Step 40010: {'lr': 5.1562099059329734e-05, 'samples': 20485632, 'steps': 40010, 'loss/train': 2.2172343730926514}
+02/26/2022 05:23:11 - INFO - codeparrot_training - Step 40011: {'lr': 5.155214714708728e-05, 'samples': 20486144, 'steps': 40011, 'loss/train': 1.5754109621047974}
+02/26/2022 05:23:17 - INFO - codeparrot_training - Step 40012: {'lr': 5.154219608493235e-05, 'samples': 20486656, 'steps': 40012, 'loss/train': 1.9428919553756714}
+02/26/2022 05:23:21 - INFO - codeparrot_training - Step 40013: {'lr': 5.1532245872907755e-05, 'samples': 20487168, 'steps': 40013, 'loss/train': 1.9903252124786377}
+02/26/2022 05:23:26 - INFO - codeparrot_training - Step 40014: {'lr': 5.152229651105586e-05, 'samples': 20487680, 'steps': 40014, 'loss/train': 1.218920111656189}
+02/26/2022 05:23:30 - INFO - codeparrot_training - Step 40015: {'lr': 5.151234799941951e-05, 'samples': 20488192, 'steps': 40015, 'loss/train': 0.5999239087104797}
+02/26/2022 05:23:35 - INFO - codeparrot_training - Step 40016: {'lr': 5.1502400338041156e-05, 'samples': 20488704, 'steps': 40016, 'loss/train': 1.3848007917404175}
+02/26/2022 05:23:39 - INFO - codeparrot_training - Step 40017: {'lr': 5.1492453526963605e-05, 'samples': 20489216, 'steps': 40017, 'loss/train': 1.176537275314331}
+02/26/2022 05:23:44 - INFO - codeparrot_training - Step 40018: {'lr': 5.148250756622924e-05, 'samples': 20489728, 'steps': 40018, 'loss/train': 0.8552457690238953}
+02/26/2022 05:23:48 - INFO - codeparrot_training - Step 40019: {'lr': 5.147256245588083e-05, 'samples': 20490240, 'steps': 40019, 'loss/train': 0.6426879167556763}
+02/26/2022 05:23:54 - INFO - codeparrot_training - Step 40020: {'lr': 5.1462618195960876e-05, 'samples': 20490752, 'steps': 40020, 'loss/train': 2.151014566421509}
+02/26/2022 05:23:57 - INFO - codeparrot_training - Step 40021: {'lr': 5.145267478651208e-05, 'samples': 20491264, 'steps': 40021, 'loss/train': 0.49274951219558716}
+02/26/2022 05:24:03 - INFO - codeparrot_training - Step 40022: {'lr': 5.144273222757698e-05, 'samples': 20491776, 'steps': 40022, 'loss/train': 2.256671190261841}
+02/26/2022 05:24:07 - INFO - codeparrot_training - Step 40023: {'lr': 5.143279051919816e-05, 'samples': 20492288, 'steps': 40023, 'loss/train': 1.3380202054977417}
+02/26/2022 05:24:12 - INFO - codeparrot_training - Step 40024: {'lr': 5.142284966141814e-05, 'samples': 20492800, 'steps': 40024, 'loss/train': 0.06912179291248322}
+02/26/2022 05:24:16 - INFO - codeparrot_training - Step 40025: {'lr': 5.1412909654279665e-05, 'samples': 20493312, 'steps': 40025, 'loss/train': 1.2770986557006836}
+02/26/2022 05:24:21 - INFO - codeparrot_training - Step 40026: {'lr': 5.140297049782519e-05, 'samples': 20493824, 'steps': 40026, 'loss/train': 1.455533504486084}
+02/26/2022 05:24:25 - INFO - codeparrot_training - Step 40027: {'lr': 5.139303219209737e-05, 'samples': 20494336, 'steps': 40027, 'loss/train': 2.0087220668792725}
+02/26/2022 05:24:30 - INFO - codeparrot_training - Step 40028: {'lr': 5.1383094737138645e-05, 'samples': 20494848, 'steps': 40028, 'loss/train': 0.9133704304695129}
+02/26/2022 05:24:34 - INFO - codeparrot_training - Step 40029: {'lr': 5.137315813299173e-05, 'samples': 20495360, 'steps': 40029, 'loss/train': 1.5029116868972778}
+02/26/2022 05:24:39 - INFO - codeparrot_training - Step 40030: {'lr': 5.136322237969918e-05, 'samples': 20495872, 'steps': 40030, 'loss/train': 0.5857676863670349}
+02/26/2022 05:24:43 - INFO - codeparrot_training - Step 40031: {'lr': 5.135328747730347e-05, 'samples': 20496384, 'steps': 40031, 'loss/train': 2.138592481613159}
+02/26/2022 05:24:48 - INFO - codeparrot_training - Step 40032: {'lr': 5.134335342584715e-05, 'samples': 20496896, 'steps': 40032, 'loss/train': 1.4250074625015259}
+02/26/2022 05:24:52 - INFO - codeparrot_training - Step 40033: {'lr': 5.133342022537288e-05, 'samples': 20497408, 'steps': 40033, 'loss/train': 1.1370712518692017}
+02/26/2022 05:24:57 - INFO - codeparrot_training - Step 40034: {'lr': 5.132348787592309e-05, 'samples': 20497920, 'steps': 40034, 'loss/train': 0.7452365756034851}
+02/26/2022 05:25:01 - INFO - codeparrot_training - Step 40035: {'lr': 5.131355637754051e-05, 'samples': 20498432, 'steps': 40035, 'loss/train': 1.8102015256881714}
+02/26/2022 05:25:07 - INFO - codeparrot_training - Step 40036: {'lr': 5.130362573026745e-05, 'samples': 20498944, 'steps': 40036, 'loss/train': 1.469778060913086}
+02/26/2022 05:25:10 - INFO - codeparrot_training - Step 40037: {'lr': 5.129369593414662e-05, 'samples': 20499456, 'steps': 40037, 'loss/train': 2.7152087688446045}
+02/26/2022 05:25:16 - INFO - codeparrot_training - Step 40038: {'lr': 5.128376698922052e-05, 'samples': 20499968, 'steps': 40038, 'loss/train': 2.024228572845459}
+02/26/2022 05:25:20 - INFO - codeparrot_training - Step 40039: {'lr': 5.1273838895531636e-05, 'samples': 20500480, 'steps': 40039, 'loss/train': 1.8824213743209839}
+02/26/2022 05:25:25 - INFO - codeparrot_training - Step 40040: {'lr': 5.12639116531225e-05, 'samples': 20500992, 'steps': 40040, 'loss/train': 1.4320727586746216}
+02/26/2022 05:25:29 - INFO - codeparrot_training - Step 40041: {'lr': 5.125398526203567e-05, 'samples': 20501504, 'steps': 40041, 'loss/train': 2.0509164333343506}
+02/26/2022 05:25:34 - INFO - codeparrot_training - Step 40042: {'lr': 5.124405972231369e-05, 'samples': 20502016, 'steps': 40042, 'loss/train': 1.5183888673782349}
+02/26/2022 05:25:38 - INFO - codeparrot_training - Step 40043: {'lr': 5.123413503399904e-05, 'samples': 20502528, 'steps': 40043, 'loss/train': 1.6124614477157593}
+02/26/2022 05:25:43 - INFO - codeparrot_training - Step 40044: {'lr': 5.122421119713419e-05, 'samples': 20503040, 'steps': 40044, 'loss/train': 1.0150903463363647}
+02/26/2022 05:25:47 - INFO - codeparrot_training - Step 40045: {'lr': 5.121428821176174e-05, 'samples': 20503552, 'steps': 40045, 'loss/train': 2.369607448577881}
+02/26/2022 05:25:52 - INFO - codeparrot_training - Step 40046: {'lr': 5.12043660779242e-05, 'samples': 20504064, 'steps': 40046, 'loss/train': 2.4435932636260986}
+02/26/2022 05:25:56 - INFO - codeparrot_training - Step 40047: {'lr': 5.119444479566393e-05, 'samples': 20504576, 'steps': 40047, 'loss/train': 2.146171808242798}
+02/26/2022 05:26:02 - INFO - codeparrot_training - Step 40048: {'lr': 5.118452436502361e-05, 'samples': 20505088, 'steps': 40048, 'loss/train': 1.655824899673462}
+02/26/2022 05:26:06 - INFO - codeparrot_training - Step 40049: {'lr': 5.1174604786045634e-05, 'samples': 20505600, 'steps': 40049, 'loss/train': 2.0627388954162598}
+02/26/2022 05:26:11 - INFO - codeparrot_training - Step 40050: {'lr': 5.116468605877253e-05, 'samples': 20506112, 'steps': 40050, 'loss/train': 1.1675745248794556}
+02/26/2022 05:26:14 - INFO - codeparrot_training - Step 40051: {'lr': 5.1154768183246714e-05, 'samples': 20506624, 'steps': 40051, 'loss/train': 1.636649250984192}
+02/26/2022 05:26:20 - INFO - codeparrot_training - Step 40052: {'lr': 5.1144851159510844e-05, 'samples': 20507136, 'steps': 40052, 'loss/train': 0.8711511492729187}
+02/26/2022 05:26:24 - INFO - codeparrot_training - Step 40053: {'lr': 5.113493498760718e-05, 'samples': 20507648, 'steps': 40053, 'loss/train': 1.1125317811965942}
+02/26/2022 05:26:29 - INFO - codeparrot_training - Step 40054: {'lr': 5.112501966757837e-05, 'samples': 20508160, 'steps': 40054, 'loss/train': 1.3584465980529785}
+02/26/2022 05:26:33 - INFO - codeparrot_training - Step 40055: {'lr': 5.111510519946675e-05, 'samples': 20508672, 'steps': 40055, 'loss/train': 1.6737310886383057}
+02/26/2022 05:26:38 - INFO - codeparrot_training - Step 40056: {'lr': 5.1105191583315e-05, 'samples': 20509184, 'steps': 40056, 'loss/train': 1.2760004997253418}
+02/26/2022 05:26:42 - INFO - codeparrot_training - Step 40057: {'lr': 5.109527881916531e-05, 'samples': 20509696, 'steps': 40057, 'loss/train': 1.088344693183899}
+02/26/2022 05:26:47 - INFO - codeparrot_training - Step 40058: {'lr': 5.1085366907060394e-05, 'samples': 20510208, 'steps': 40058, 'loss/train': 0.9741630554199219}
+02/26/2022 05:26:51 - INFO - codeparrot_training - Step 40059: {'lr': 5.107545584704249e-05, 'samples': 20510720, 'steps': 40059, 'loss/train': 1.393269419670105}
+02/26/2022 05:26:57 - INFO - codeparrot_training - Step 40060: {'lr': 5.106554563915425e-05, 'samples': 20511232, 'steps': 40060, 'loss/train': 1.1468664407730103}
+02/26/2022 05:27:00 - INFO - codeparrot_training - Step 40061: {'lr': 5.1055636283438066e-05, 'samples': 20511744, 'steps': 40061, 'loss/train': 1.1084208488464355}
+02/26/2022 05:27:06 - INFO - codeparrot_training - Step 40062: {'lr': 5.104572777993632e-05, 'samples': 20512256, 'steps': 40062, 'loss/train': 2.5606679916381836}
+02/26/2022 05:27:09 - INFO - codeparrot_training - Step 40063: {'lr': 5.103582012869148e-05, 'samples': 20512768, 'steps': 40063, 'loss/train': 1.2763450145721436}
+02/26/2022 05:27:15 - INFO - codeparrot_training - Step 40064: {'lr': 5.102591332974604e-05, 'samples': 20513280, 'steps': 40064, 'loss/train': 1.368504524230957}
+02/26/2022 05:27:18 - INFO - codeparrot_training - Step 40065: {'lr': 5.101600738314244e-05, 'samples': 20513792, 'steps': 40065, 'loss/train': 1.5549052953720093}
+02/26/2022 05:27:24 - INFO - codeparrot_training - Step 40066: {'lr': 5.100610228892305e-05, 'samples': 20514304, 'steps': 40066, 'loss/train': 1.6941736936569214}
+02/26/2022 05:27:27 - INFO - codeparrot_training - Step 40067: {'lr': 5.099619804713027e-05, 'samples': 20514816, 'steps': 40067, 'loss/train': 2.4828040599823}
+02/26/2022 05:27:33 - INFO - codeparrot_training - Step 40068: {'lr': 5.098629465780666e-05, 'samples': 20515328, 'steps': 40068, 'loss/train': 1.2498379945755005}
+02/26/2022 05:27:36 - INFO - codeparrot_training - Step 40069: {'lr': 5.09763921209945e-05, 'samples': 20515840, 'steps': 40069, 'loss/train': 2.242964506149292}
+02/26/2022 05:27:42 - INFO - codeparrot_training - Step 40070: {'lr': 5.09664904367364e-05, 'samples': 20516352, 'steps': 40070, 'loss/train': 2.03818941116333}
+02/26/2022 05:27:46 - INFO - codeparrot_training - Step 40071: {'lr': 5.0956589605074546e-05, 'samples': 20516864, 'steps': 40071, 'loss/train': 0.3864399492740631}
+02/26/2022 05:27:51 - INFO - codeparrot_training - Step 40072: {'lr': 5.0946689626051505e-05, 'samples': 20517376, 'steps': 40072, 'loss/train': 1.2123653888702393}
+02/26/2022 05:27:57 - INFO - codeparrot_training - Step 40073: {'lr': 5.093679049970959e-05, 'samples': 20517888, 'steps': 40073, 'loss/train': 1.380172848701477}
+02/26/2022 05:28:00 - INFO - codeparrot_training - Step 40074: {'lr': 5.092689222609137e-05, 'samples': 20518400, 'steps': 40074, 'loss/train': 2.345250368118286}
+02/26/2022 05:28:06 - INFO - codeparrot_training - Step 40075: {'lr': 5.091699480523901e-05, 'samples': 20518912, 'steps': 40075, 'loss/train': 1.3724156618118286}
+02/26/2022 05:28:09 - INFO - codeparrot_training - Step 40076: {'lr': 5.0907098237195084e-05, 'samples': 20519424, 'steps': 40076, 'loss/train': 1.8128714561462402}
+02/26/2022 05:28:15 - INFO - codeparrot_training - Step 40077: {'lr': 5.0897202522001865e-05, 'samples': 20519936, 'steps': 40077, 'loss/train': 1.265244483947754}
+02/26/2022 05:28:19 - INFO - codeparrot_training - Step 40078: {'lr': 5.088730765970195e-05, 'samples': 20520448, 'steps': 40078, 'loss/train': 1.8832745552062988}
+02/26/2022 05:28:24 - INFO - codeparrot_training - Step 40079: {'lr': 5.087741365033746e-05, 'samples': 20520960, 'steps': 40079, 'loss/train': 2.4139490127563477}
+02/26/2022 05:28:28 - INFO - codeparrot_training - Step 40080: {'lr': 5.086752049395094e-05, 'samples': 20521472, 'steps': 40080, 'loss/train': 0.6943578720092773}
+02/26/2022 05:28:33 - INFO - codeparrot_training - Step 40081: {'lr': 5.085762819058468e-05, 'samples': 20521984, 'steps': 40081, 'loss/train': 2.4370670318603516}
+02/26/2022 05:28:37 - INFO - codeparrot_training - Step 40082: {'lr': 5.084773674028118e-05, 'samples': 20522496, 'steps': 40082, 'loss/train': 1.584202527999878}
+02/26/2022 05:28:42 - INFO - codeparrot_training - Step 40083: {'lr': 5.083784614308273e-05, 'samples': 20523008, 'steps': 40083, 'loss/train': 1.9010438919067383}
+02/26/2022 05:28:46 - INFO - codeparrot_training - Step 40084: {'lr': 5.082795639903171e-05, 'samples': 20523520, 'steps': 40084, 'loss/train': 1.6217018365859985}
+02/26/2022 05:28:52 - INFO - codeparrot_training - Step 40085: {'lr': 5.0818067508170475e-05, 'samples': 20524032, 'steps': 40085, 'loss/train': 0.5566498041152954}
+02/26/2022 05:28:55 - INFO - codeparrot_training - Step 40086: {'lr': 5.080817947054131e-05, 'samples': 20524544, 'steps': 40086, 'loss/train': 1.7606621980667114}
+02/26/2022 05:29:01 - INFO - codeparrot_training - Step 40087: {'lr': 5.079829228618674e-05, 'samples': 20525056, 'steps': 40087, 'loss/train': 1.311846137046814}
+02/26/2022 05:29:04 - INFO - codeparrot_training - Step 40088: {'lr': 5.078840595514902e-05, 'samples': 20525568, 'steps': 40088, 'loss/train': 1.570685863494873}
+02/26/2022 05:29:10 - INFO - codeparrot_training - Step 40089: {'lr': 5.077852047747053e-05, 'samples': 20526080, 'steps': 40089, 'loss/train': 2.0077085494995117}
+02/26/2022 05:29:13 - INFO - codeparrot_training - Step 40090: {'lr': 5.076863585319352e-05, 'samples': 20526592, 'steps': 40090, 'loss/train': 1.116776943206787}
+02/26/2022 05:29:19 - INFO - codeparrot_training - Step 40091: {'lr': 5.07587520823605e-05, 'samples': 20527104, 'steps': 40091, 'loss/train': 2.1811821460723877}
+02/26/2022 05:29:22 - INFO - codeparrot_training - Step 40092: {'lr': 5.07488691650137e-05, 'samples': 20527616, 'steps': 40092, 'loss/train': 1.6807693243026733}
+02/26/2022 05:29:28 - INFO - codeparrot_training - Step 40093: {'lr': 5.0738987101195454e-05, 'samples': 20528128, 'steps': 40093, 'loss/train': 1.837868332862854}
+02/26/2022 05:29:31 - INFO - codeparrot_training - Step 40094: {'lr': 5.0729105890948064e-05, 'samples': 20528640, 'steps': 40094, 'loss/train': 1.634913682937622}
+02/26/2022 05:29:37 - INFO - codeparrot_training - Step 40095: {'lr': 5.071922553431399e-05, 'samples': 20529152, 'steps': 40095, 'loss/train': 2.005528688430786}
+02/26/2022 05:29:41 - INFO - codeparrot_training - Step 40096: {'lr': 5.070934603133548e-05, 'samples': 20529664, 'steps': 40096, 'loss/train': 2.3615410327911377}
+02/26/2022 05:29:46 - INFO - codeparrot_training - Step 40097: {'lr': 5.069946738205483e-05, 'samples': 20530176, 'steps': 40097, 'loss/train': 0.8500372171401978}
+02/26/2022 05:29:50 - INFO - codeparrot_training - Step 40098: {'lr': 5.068958958651432e-05, 'samples': 20530688, 'steps': 40098, 'loss/train': 1.7104334831237793}
+02/26/2022 05:29:55 - INFO - codeparrot_training - Step 40099: {'lr': 5.067971264475638e-05, 'samples': 20531200, 'steps': 40099, 'loss/train': 0.045369166880846024}
+02/26/2022 05:29:59 - INFO - codeparrot_training - Step 40100: {'lr': 5.066983655682325e-05, 'samples': 20531712, 'steps': 40100, 'loss/train': 1.5760809183120728}
+02/26/2022 05:30:05 - INFO - codeparrot_training - Step 40101: {'lr': 5.065996132275724e-05, 'samples': 20532224, 'steps': 40101, 'loss/train': 2.4488492012023926}
+02/26/2022 05:30:08 - INFO - codeparrot_training - Step 40102: {'lr': 5.065008694260062e-05, 'samples': 20532736, 'steps': 40102, 'loss/train': 1.4471408128738403}
+02/26/2022 05:30:14 - INFO - codeparrot_training - Step 40103: {'lr': 5.06402134163958e-05, 'samples': 20533248, 'steps': 40103, 'loss/train': 0.8726754188537598}
+02/26/2022 05:30:17 - INFO - codeparrot_training - Step 40104: {'lr': 5.063034074418496e-05, 'samples': 20533760, 'steps': 40104, 'loss/train': 1.3631744384765625}
+02/26/2022 05:30:23 - INFO - codeparrot_training - Step 40105: {'lr': 5.062046892601047e-05, 'samples': 20534272, 'steps': 40105, 'loss/train': 1.4810874462127686}
+02/26/2022 05:30:27 - INFO - codeparrot_training - Step 40106: {'lr': 5.06105979619145e-05, 'samples': 20534784, 'steps': 40106, 'loss/train': 1.344446063041687}
+02/26/2022 05:30:32 - INFO - codeparrot_training - Step 40107: {'lr': 5.0600727851939476e-05, 'samples': 20535296, 'steps': 40107, 'loss/train': 1.884863018989563}
+02/26/2022 05:30:36 - INFO - codeparrot_training - Step 40108: {'lr': 5.0590858596127554e-05, 'samples': 20535808, 'steps': 40108, 'loss/train': 1.4678008556365967}
+02/26/2022 05:30:41 - INFO - codeparrot_training - Step 40109: {'lr': 5.0580990194521226e-05, 'samples': 20536320, 'steps': 40109, 'loss/train': 2.0973196029663086}
+02/26/2022 05:30:45 - INFO - codeparrot_training - Step 40110: {'lr': 5.057112264716246e-05, 'samples': 20536832, 'steps': 40110, 'loss/train': 1.0496724843978882}
+02/26/2022 05:30:50 - INFO - codeparrot_training - Step 40111: {'lr': 5.056125595409375e-05, 'samples': 20537344, 'steps': 40111, 'loss/train': 0.8334508538246155}
+02/26/2022 05:30:54 - INFO - codeparrot_training - Step 40112: {'lr': 5.0551390115357225e-05, 'samples': 20537856, 'steps': 40112, 'loss/train': 2.1235551834106445}
+02/26/2022 05:30:59 - INFO - codeparrot_training - Step 40113: {'lr': 5.054152513099533e-05, 'samples': 20538368, 'steps': 40113, 'loss/train': 0.5583958029747009}
+02/26/2022 05:31:03 - INFO - codeparrot_training - Step 40114: {'lr': 5.0531661001050114e-05, 'samples': 20538880, 'steps': 40114, 'loss/train': 1.6286840438842773}
+02/26/2022 05:31:09 - INFO - codeparrot_training - Step 40115: {'lr': 5.052179772556395e-05, 'samples': 20539392, 'steps': 40115, 'loss/train': 1.1722922325134277}
+02/26/2022 05:31:12 - INFO - codeparrot_training - Step 40116: {'lr': 5.051193530457901e-05, 'samples': 20539904, 'steps': 40116, 'loss/train': 1.9946843385696411}
+02/26/2022 05:31:18 - INFO - codeparrot_training - Step 40117: {'lr': 5.050207373813767e-05, 'samples': 20540416, 'steps': 40117, 'loss/train': 1.5008091926574707}
+02/26/2022 05:31:21 - INFO - codeparrot_training - Step 40118: {'lr': 5.04922130262821e-05, 'samples': 20540928, 'steps': 40118, 'loss/train': 1.550695538520813}
+02/26/2022 05:31:27 - INFO - codeparrot_training - Step 40119: {'lr': 5.048235316905453e-05, 'samples': 20541440, 'steps': 40119, 'loss/train': 2.4589316844940186}
+02/26/2022 05:31:30 - INFO - codeparrot_training - Step 40120: {'lr': 5.047249416649713e-05, 'samples': 20541952, 'steps': 40120, 'loss/train': 1.2552516460418701}
+02/26/2022 05:31:36 - INFO - codeparrot_training - Step 40121: {'lr': 5.046263601865228e-05, 'samples': 20542464, 'steps': 40121, 'loss/train': 1.1930433511734009}
+02/26/2022 05:31:39 - INFO - codeparrot_training - Step 40122: {'lr': 5.045277872556214e-05, 'samples': 20542976, 'steps': 40122, 'loss/train': 2.5336406230926514}
+02/26/2022 05:31:45 - INFO - codeparrot_training - Step 40123: {'lr': 5.044292228726893e-05, 'samples': 20543488, 'steps': 40123, 'loss/train': 2.752393960952759}
+02/26/2022 05:31:48 - INFO - codeparrot_training - Step 40124: {'lr': 5.043306670381481e-05, 'samples': 20544000, 'steps': 40124, 'loss/train': 1.7778352499008179}
+02/26/2022 05:31:55 - INFO - codeparrot_training - Step 40125: {'lr': 5.0423211975242136e-05, 'samples': 20544512, 'steps': 40125, 'loss/train': 2.417404890060425}
+02/26/2022 05:31:58 - INFO - codeparrot_training - Step 40126: {'lr': 5.041335810159303e-05, 'samples': 20545024, 'steps': 40126, 'loss/train': 1.8781704902648926}
+02/26/2022 05:32:03 - INFO - codeparrot_training - Step 40127: {'lr': 5.040350508290975e-05, 'samples': 20545536, 'steps': 40127, 'loss/train': 1.4455132484436035}
+02/26/2022 05:32:07 - INFO - codeparrot_training - Step 40128: {'lr': 5.0393652919234395e-05, 'samples': 20546048, 'steps': 40128, 'loss/train': 0.7860716581344604}
+02/26/2022 05:32:12 - INFO - codeparrot_training - Step 40129: {'lr': 5.0383801610609294e-05, 'samples': 20546560, 'steps': 40129, 'loss/train': 1.3679593801498413}
+02/26/2022 05:32:16 - INFO - codeparrot_training - Step 40130: {'lr': 5.037395115707663e-05, 'samples': 20547072, 'steps': 40130, 'loss/train': 2.0277838706970215}
+02/26/2022 05:32:22 - INFO - codeparrot_training - Step 40131: {'lr': 5.036410155867857e-05, 'samples': 20547584, 'steps': 40131, 'loss/train': 1.784590244293213}
+02/26/2022 05:32:25 - INFO - codeparrot_training - Step 40132: {'lr': 5.0354252815457275e-05, 'samples': 20548096, 'steps': 40132, 'loss/train': 2.3224823474884033}
+02/26/2022 05:32:31 - INFO - codeparrot_training - Step 40133: {'lr': 5.034440492745493e-05, 'samples': 20548608, 'steps': 40133, 'loss/train': 1.8877272605895996}
+02/26/2022 05:32:34 - INFO - codeparrot_training - Step 40134: {'lr': 5.0334557894713834e-05, 'samples': 20549120, 'steps': 40134, 'loss/train': 1.1653307676315308}
+02/26/2022 05:32:40 - INFO - codeparrot_training - Step 40135: {'lr': 5.0324711717276064e-05, 'samples': 20549632, 'steps': 40135, 'loss/train': 1.8135045766830444}
+02/26/2022 05:32:43 - INFO - codeparrot_training - Step 40136: {'lr': 5.031486639518385e-05, 'samples': 20550144, 'steps': 40136, 'loss/train': 1.6274279356002808}
+02/26/2022 05:32:49 - INFO - codeparrot_training - Step 40137: {'lr': 5.030502192847924e-05, 'samples': 20550656, 'steps': 40137, 'loss/train': 1.4808542728424072}
+02/26/2022 05:32:52 - INFO - codeparrot_training - Step 40138: {'lr': 5.0295178317204595e-05, 'samples': 20551168, 'steps': 40138, 'loss/train': 1.9895950555801392}
+02/26/2022 05:32:58 - INFO - codeparrot_training - Step 40139: {'lr': 5.028533556140199e-05, 'samples': 20551680, 'steps': 40139, 'loss/train': 2.356114387512207}
+02/26/2022 05:33:01 - INFO - codeparrot_training - Step 40140: {'lr': 5.0275493661113584e-05, 'samples': 20552192, 'steps': 40140, 'loss/train': 1.7727941274642944}
+02/26/2022 05:33:08 - INFO - codeparrot_training - Step 40141: {'lr': 5.0265652616381496e-05, 'samples': 20552704, 'steps': 40141, 'loss/train': 1.488487958908081}
+02/26/2022 05:33:12 - INFO - codeparrot_training - Step 40142: {'lr': 5.025581242724797e-05, 'samples': 20553216, 'steps': 40142, 'loss/train': 1.5116512775421143}
+02/26/2022 05:33:17 - INFO - codeparrot_training - Step 40143: {'lr': 5.024597309375506e-05, 'samples': 20553728, 'steps': 40143, 'loss/train': 2.096799850463867}
+02/26/2022 05:33:21 - INFO - codeparrot_training - Step 40144: {'lr': 5.023613461594512e-05, 'samples': 20554240, 'steps': 40144, 'loss/train': 0.38335734605789185}
+02/26/2022 05:33:26 - INFO - codeparrot_training - Step 40145: {'lr': 5.0226296993859996e-05, 'samples': 20554752, 'steps': 40145, 'loss/train': 1.1693456172943115}
+02/26/2022 05:33:30 - INFO - codeparrot_training - Step 40146: {'lr': 5.021646022754206e-05, 'samples': 20555264, 'steps': 40146, 'loss/train': 0.5619250535964966}
+02/26/2022 05:33:35 - INFO - codeparrot_training - Step 40147: {'lr': 5.020662431703329e-05, 'samples': 20555776, 'steps': 40147, 'loss/train': 1.6746019124984741}
+02/26/2022 05:33:39 - INFO - codeparrot_training - Step 40148: {'lr': 5.0196789262376055e-05, 'samples': 20556288, 'steps': 40148, 'loss/train': 1.7113332748413086}
+02/26/2022 05:33:44 - INFO - codeparrot_training - Step 40149: {'lr': 5.018695506361218e-05, 'samples': 20556800, 'steps': 40149, 'loss/train': 2.1674680709838867}
+02/26/2022 05:33:48 - INFO - codeparrot_training - Step 40150: {'lr': 5.017712172078401e-05, 'samples': 20557312, 'steps': 40150, 'loss/train': 2.3900294303894043}
+02/26/2022 05:33:55 - INFO - codeparrot_training - Step 40151: {'lr': 5.0167289233933546e-05, 'samples': 20557824, 'steps': 40151, 'loss/train': 0.1731039136648178}
+02/26/2022 05:33:58 - INFO - codeparrot_training - Step 40152: {'lr': 5.0157457603103055e-05, 'samples': 20558336, 'steps': 40152, 'loss/train': 1.2805061340332031}
+02/26/2022 05:34:04 - INFO - codeparrot_training - Step 40153: {'lr': 5.0147626828334456e-05, 'samples': 20558848, 'steps': 40153, 'loss/train': 2.149085283279419}
+02/26/2022 05:34:07 - INFO - codeparrot_training - Step 40154: {'lr': 5.0137796909670046e-05, 'samples': 20559360, 'steps': 40154, 'loss/train': 1.5128902196884155}
+02/26/2022 05:34:13 - INFO - codeparrot_training - Step 40155: {'lr': 5.0127967847151764e-05, 'samples': 20559872, 'steps': 40155, 'loss/train': 1.9556902647018433}
+02/26/2022 05:34:16 - INFO - codeparrot_training - Step 40156: {'lr': 5.011813964082187e-05, 'samples': 20560384, 'steps': 40156, 'loss/train': 2.357943296432495}
+02/26/2022 05:34:22 - INFO - codeparrot_training - Step 40157: {'lr': 5.010831229072238e-05, 'samples': 20560896, 'steps': 40157, 'loss/train': 0.8816267848014832}
+02/26/2022 05:34:25 - INFO - codeparrot_training - Step 40158: {'lr': 5.0098485796895435e-05, 'samples': 20561408, 'steps': 40158, 'loss/train': 1.043492317199707}
+02/26/2022 05:34:31 - INFO - codeparrot_training - Step 40159: {'lr': 5.0088660159383016e-05, 'samples': 20561920, 'steps': 40159, 'loss/train': 0.5299811959266663}
+02/26/2022 05:34:34 - INFO - codeparrot_training - Step 40160: {'lr': 5.007883537822736e-05, 'samples': 20562432, 'steps': 40160, 'loss/train': 2.145418167114258}
+02/26/2022 05:34:40 - INFO - codeparrot_training - Step 40161: {'lr': 5.006901145347051e-05, 'samples': 20562944, 'steps': 40161, 'loss/train': 0.6495620012283325}
+02/26/2022 05:34:43 - INFO - codeparrot_training - Step 40162: {'lr': 5.0059188385154525e-05, 'samples': 20563456, 'steps': 40162, 'loss/train': 2.6692028045654297}
+02/26/2022 05:34:50 - INFO - codeparrot_training - Step 40163: {'lr': 5.0049366173321395e-05, 'samples': 20563968, 'steps': 40163, 'loss/train': 1.9270213842391968}
+02/26/2022 05:34:54 - INFO - codeparrot_training - Step 40164: {'lr': 5.0039544818013396e-05, 'samples': 20564480, 'steps': 40164, 'loss/train': 1.7030454874038696}
+02/26/2022 05:34:59 - INFO - codeparrot_training - Step 40165: {'lr': 5.0029724319272466e-05, 'samples': 20564992, 'steps': 40165, 'loss/train': 2.9641973972320557}
+02/26/2022 05:35:05 - INFO - codeparrot_training - Step 40166: {'lr': 5.001990467714071e-05, 'samples': 20565504, 'steps': 40166, 'loss/train': 2.1683390140533447}
+02/26/2022 05:35:08 - INFO - codeparrot_training - Step 40167: {'lr': 5.0010085891660115e-05, 'samples': 20566016, 'steps': 40167, 'loss/train': 1.439074993133545}
+02/26/2022 05:35:12 - INFO - codeparrot_training - Step 40168: {'lr': 5.000026796287288e-05, 'samples': 20566528, 'steps': 40168, 'loss/train': 0.9136861562728882}
+02/26/2022 05:35:17 - INFO - codeparrot_training - Step 40169: {'lr': 4.9990450890820964e-05, 'samples': 20567040, 'steps': 40169, 'loss/train': 1.5852373838424683}
+02/26/2022 05:35:21 - INFO - codeparrot_training - Step 40170: {'lr': 4.998063467554648e-05, 'samples': 20567552, 'steps': 40170, 'loss/train': 1.5793037414550781}
+02/26/2022 05:35:26 - INFO - codeparrot_training - Step 40171: {'lr': 4.9970819317091356e-05, 'samples': 20568064, 'steps': 40171, 'loss/train': 1.851469874382019}
+02/26/2022 05:35:30 - INFO - codeparrot_training - Step 40172: {'lr': 4.996100481549781e-05, 'samples': 20568576, 'steps': 40172, 'loss/train': 1.5916630029678345}
+02/26/2022 05:35:36 - INFO - codeparrot_training - Step 40173: {'lr': 4.995119117080779e-05, 'samples': 20569088, 'steps': 40173, 'loss/train': 1.2014036178588867}
+02/26/2022 05:35:39 - INFO - codeparrot_training - Step 40174: {'lr': 4.994137838306334e-05, 'samples': 20569600, 'steps': 40174, 'loss/train': 0.8751251101493835}
+02/26/2022 05:35:45 - INFO - codeparrot_training - Step 40175: {'lr': 4.993156645230643e-05, 'samples': 20570112, 'steps': 40175, 'loss/train': 2.2376980781555176}
+02/26/2022 05:35:48 - INFO - codeparrot_training - Step 40176: {'lr': 4.992175537857924e-05, 'samples': 20570624, 'steps': 40176, 'loss/train': 2.159022092819214}
+02/26/2022 05:35:54 - INFO - codeparrot_training - Step 40177: {'lr': 4.991194516192371e-05, 'samples': 20571136, 'steps': 40177, 'loss/train': 2.358538866043091}
+02/26/2022 05:35:57 - INFO - codeparrot_training - Step 40178: {'lr': 4.990213580238179e-05, 'samples': 20571648, 'steps': 40178, 'loss/train': 0.9790414571762085}
+02/26/2022 05:36:03 - INFO - codeparrot_training - Step 40179: {'lr': 4.9892327299995655e-05, 'samples': 20572160, 'steps': 40179, 'loss/train': 2.0554122924804688}
+02/26/2022 05:36:08 - INFO - codeparrot_training - Step 40180: {'lr': 4.988251965480728e-05, 'samples': 20572672, 'steps': 40180, 'loss/train': 2.4225683212280273}
+02/26/2022 05:36:12 - INFO - codeparrot_training - Step 40181: {'lr': 4.98727128668586e-05, 'samples': 20573184, 'steps': 40181, 'loss/train': 1.674485206604004}
+02/26/2022 05:36:18 - INFO - codeparrot_training - Step 40182: {'lr': 4.986290693619164e-05, 'samples': 20573696, 'steps': 40182, 'loss/train': 2.0279340744018555}
+02/26/2022 05:36:21 - INFO - codeparrot_training - Step 40183: {'lr': 4.985310186284855e-05, 'samples': 20574208, 'steps': 40183, 'loss/train': 1.9559215307235718}
+02/26/2022 05:36:26 - INFO - codeparrot_training - Step 40184: {'lr': 4.9843297646871096e-05, 'samples': 20574720, 'steps': 40184, 'loss/train': 0.44912201166152954}
+02/26/2022 05:36:30 - INFO - codeparrot_training - Step 40185: {'lr': 4.9833494288301474e-05, 'samples': 20575232, 'steps': 40185, 'loss/train': 0.500665009021759}
+02/26/2022 05:36:35 - INFO - codeparrot_training - Step 40186: {'lr': 4.982369178718152e-05, 'samples': 20575744, 'steps': 40186, 'loss/train': 1.400141716003418}
+02/26/2022 05:36:39 - INFO - codeparrot_training - Step 40187: {'lr': 4.981389014355342e-05, 'samples': 20576256, 'steps': 40187, 'loss/train': 2.11801815032959}
+02/26/2022 05:36:45 - INFO - codeparrot_training - Step 40188: {'lr': 4.9804089357458945e-05, 'samples': 20576768, 'steps': 40188, 'loss/train': 2.884381055831909}
+02/26/2022 05:36:48 - INFO - codeparrot_training - Step 40189: {'lr': 4.979428942894024e-05, 'samples': 20577280, 'steps': 40189, 'loss/train': 1.8546353578567505}
+02/26/2022 05:36:54 - INFO - codeparrot_training - Step 40190: {'lr': 4.978449035803917e-05, 'samples': 20577792, 'steps': 40190, 'loss/train': 1.9362952709197998}
+02/26/2022 05:36:57 - INFO - codeparrot_training - Step 40191: {'lr': 4.9774692144797824e-05, 'samples': 20578304, 'steps': 40191, 'loss/train': 1.2781065702438354}
+02/26/2022 05:37:03 - INFO - codeparrot_training - Step 40192: {'lr': 4.976489478925811e-05, 'samples': 20578816, 'steps': 40192, 'loss/train': 1.7723109722137451}
+02/26/2022 05:37:06 - INFO - codeparrot_training - Step 40193: {'lr': 4.975509829146202e-05, 'samples': 20579328, 'steps': 40193, 'loss/train': 2.224621057510376}
+02/26/2022 05:37:12 - INFO - codeparrot_training - Step 40194: {'lr': 4.974530265145144e-05, 'samples': 20579840, 'steps': 40194, 'loss/train': 2.2311601638793945}
+02/26/2022 05:37:15 - INFO - codeparrot_training - Step 40195: {'lr': 4.973550786926845e-05, 'samples': 20580352, 'steps': 40195, 'loss/train': 1.803673505783081}
+02/26/2022 05:37:21 - INFO - codeparrot_training - Step 40196: {'lr': 4.9725713944954956e-05, 'samples': 20580864, 'steps': 40196, 'loss/train': 1.989607810974121}
+02/26/2022 05:37:24 - INFO - codeparrot_training - Step 40197: {'lr': 4.97159208785529e-05, 'samples': 20581376, 'steps': 40197, 'loss/train': 1.81013822555542}
+02/26/2022 05:37:31 - INFO - codeparrot_training - Step 40198: {'lr': 4.970612867010418e-05, 'samples': 20581888, 'steps': 40198, 'loss/train': 1.4249409437179565}
+02/26/2022 05:37:34 - INFO - codeparrot_training - Step 40199: {'lr': 4.969633731965087e-05, 'samples': 20582400, 'steps': 40199, 'loss/train': 1.727388858795166}
+02/26/2022 05:37:40 - INFO - codeparrot_training - Step 40200: {'lr': 4.9686546827234865e-05, 'samples': 20582912, 'steps': 40200, 'loss/train': 1.0296684503555298}
+02/26/2022 05:37:43 - INFO - codeparrot_training - Step 40201: {'lr': 4.967675719289807e-05, 'samples': 20583424, 'steps': 40201, 'loss/train': 2.1278061866760254}
+02/26/2022 05:37:49 - INFO - codeparrot_training - Step 40202: {'lr': 4.966696841668239e-05, 'samples': 20583936, 'steps': 40202, 'loss/train': 2.240410804748535}
+02/26/2022 05:37:52 - INFO - codeparrot_training - Step 40203: {'lr': 4.965718049862985e-05, 'samples': 20584448, 'steps': 40203, 'loss/train': 1.5604584217071533}
+02/26/2022 05:37:58 - INFO - codeparrot_training - Step 40204: {'lr': 4.964739343878227e-05, 'samples': 20584960, 'steps': 40204, 'loss/train': 1.1971962451934814}
+02/26/2022 05:38:01 - INFO - codeparrot_training - Step 40205: {'lr': 4.9637607237181775e-05, 'samples': 20585472, 'steps': 40205, 'loss/train': 0.9253528714179993}
+02/26/2022 05:38:07 - INFO - codeparrot_training - Step 40206: {'lr': 4.962782189387e-05, 'samples': 20585984, 'steps': 40206, 'loss/train': 1.8904399871826172}
+02/26/2022 05:38:10 - INFO - codeparrot_training - Step 40207: {'lr': 4.9618037408889095e-05, 'samples': 20586496, 'steps': 40207, 'loss/train': 1.267303705215454}
+02/26/2022 05:38:17 - INFO - codeparrot_training - Step 40208: {'lr': 4.960825378228082e-05, 'samples': 20587008, 'steps': 40208, 'loss/train': 2.0993943214416504}
+02/26/2022 05:38:20 - INFO - codeparrot_training - Step 40209: {'lr': 4.959847101408729e-05, 'samples': 20587520, 'steps': 40209, 'loss/train': 2.250439405441284}
+02/26/2022 05:38:26 - INFO - codeparrot_training - Step 40210: {'lr': 4.9588689104350134e-05, 'samples': 20588032, 'steps': 40210, 'loss/train': 1.7145178318023682}
+02/26/2022 05:38:29 - INFO - codeparrot_training - Step 40211: {'lr': 4.957890805311147e-05, 'samples': 20588544, 'steps': 40211, 'loss/train': 2.0793075561523438}
+02/26/2022 05:38:35 - INFO - codeparrot_training - Step 40212: {'lr': 4.9569127860413066e-05, 'samples': 20589056, 'steps': 40212, 'loss/train': 1.5559533834457397}
+02/26/2022 05:38:39 - INFO - codeparrot_training - Step 40213: {'lr': 4.955934852629698e-05, 'samples': 20589568, 'steps': 40213, 'loss/train': 2.2094738483428955}
+02/26/2022 05:38:44 - INFO - codeparrot_training - Step 40214: {'lr': 4.9549570050804876e-05, 'samples': 20590080, 'steps': 40214, 'loss/train': 2.9797372817993164}
+02/26/2022 05:38:48 - INFO - codeparrot_training - Step 40215: {'lr': 4.953979243397882e-05, 'samples': 20590592, 'steps': 40215, 'loss/train': 0.03249917924404144}
+02/26/2022 05:38:53 - INFO - codeparrot_training - Step 40216: {'lr': 4.953001567586068e-05, 'samples': 20591104, 'steps': 40216, 'loss/train': 2.1189401149749756}
+02/26/2022 05:38:57 - INFO - codeparrot_training - Step 40217: {'lr': 4.95202397764922e-05, 'samples': 20591616, 'steps': 40217, 'loss/train': 0.28025704622268677}
+02/26/2022 05:39:03 - INFO - codeparrot_training - Step 40218: {'lr': 4.951046473591542e-05, 'samples': 20592128, 'steps': 40218, 'loss/train': 1.8474280834197998}
+02/26/2022 05:39:06 - INFO - codeparrot_training - Step 40219: {'lr': 4.950069055417214e-05, 'samples': 20592640, 'steps': 40219, 'loss/train': 2.675217628479004}
+02/26/2022 05:39:12 - INFO - codeparrot_training - Step 40220: {'lr': 4.949091723130425e-05, 'samples': 20593152, 'steps': 40220, 'loss/train': 2.324949026107788}
+02/26/2022 05:39:15 - INFO - codeparrot_training - Step 40221: {'lr': 4.948114476735355e-05, 'samples': 20593664, 'steps': 40221, 'loss/train': 1.628584861755371}
+02/26/2022 05:39:21 - INFO - codeparrot_training - Step 40222: {'lr': 4.9471373162362e-05, 'samples': 20594176, 'steps': 40222, 'loss/train': 1.1752337217330933}
+02/26/2022 05:39:24 - INFO - codeparrot_training - Step 40223: {'lr': 4.9461602416371405e-05, 'samples': 20594688, 'steps': 40223, 'loss/train': 2.300471782684326}
+02/26/2022 05:39:30 - INFO - codeparrot_training - Step 40224: {'lr': 4.945183252942362e-05, 'samples': 20595200, 'steps': 40224, 'loss/train': 2.1665046215057373}
+02/26/2022 05:39:33 - INFO - codeparrot_training - Step 40225: {'lr': 4.944206350156047e-05, 'samples': 20595712, 'steps': 40225, 'loss/train': 2.160352945327759}
+02/26/2022 05:39:39 - INFO - codeparrot_training - Step 40226: {'lr': 4.943229533282387e-05, 'samples': 20596224, 'steps': 40226, 'loss/train': 1.5185710191726685}
+02/26/2022 05:39:43 - INFO - codeparrot_training - Step 40227: {'lr': 4.942252802325564e-05, 'samples': 20596736, 'steps': 40227, 'loss/train': 2.0852737426757812}
+02/26/2022 05:39:48 - INFO - codeparrot_training - Step 40228: {'lr': 4.941276157289762e-05, 'samples': 20597248, 'steps': 40228, 'loss/train': 1.7140722274780273}
+02/26/2022 05:39:52 - INFO - codeparrot_training - Step 40229: {'lr': 4.9402995981791573e-05, 'samples': 20597760, 'steps': 40229, 'loss/train': 2.2234158515930176}
+02/26/2022 05:39:57 - INFO - codeparrot_training - Step 40230: {'lr': 4.939323124997944e-05, 'samples': 20598272, 'steps': 40230, 'loss/train': 1.0340170860290527}
+02/26/2022 05:40:00 - INFO - codeparrot_training - Step 40231: {'lr': 4.938346737750302e-05, 'samples': 20598784, 'steps': 40231, 'loss/train': 1.3000550270080566}
+02/26/2022 05:40:08 - INFO - codeparrot_training - Step 40232: {'lr': 4.9373704364404106e-05, 'samples': 20599296, 'steps': 40232, 'loss/train': 0.4428895115852356}
+02/26/2022 05:40:12 - INFO - codeparrot_training - Step 40233: {'lr': 4.936394221072449e-05, 'samples': 20599808, 'steps': 40233, 'loss/train': 1.55039644241333}
+02/26/2022 05:40:17 - INFO - codeparrot_training - Step 40234: {'lr': 4.935418091650609e-05, 'samples': 20600320, 'steps': 40234, 'loss/train': 1.3537769317626953}
+02/26/2022 05:40:21 - INFO - codeparrot_training - Step 40235: {'lr': 4.934442048179069e-05, 'samples': 20600832, 'steps': 40235, 'loss/train': 1.1374977827072144}
+02/26/2022 05:40:26 - INFO - codeparrot_training - Step 40236: {'lr': 4.933466090662006e-05, 'samples': 20601344, 'steps': 40236, 'loss/train': 2.037734270095825}
+02/26/2022 05:40:30 - INFO - codeparrot_training - Step 40237: {'lr': 4.932490219103594e-05, 'samples': 20601856, 'steps': 40237, 'loss/train': 1.880330204963684}
+02/26/2022 05:40:35 - INFO - codeparrot_training - Step 40238: {'lr': 4.93151443350803e-05, 'samples': 20602368, 'steps': 40238, 'loss/train': 2.1331236362457275}
+02/26/2022 05:40:39 - INFO - codeparrot_training - Step 40239: {'lr': 4.9305387338794803e-05, 'samples': 20602880, 'steps': 40239, 'loss/train': 1.8540791273117065}
+02/26/2022 05:40:44 - INFO - codeparrot_training - Step 40240: {'lr': 4.929563120222141e-05, 'samples': 20603392, 'steps': 40240, 'loss/train': 2.607050657272339}
+02/26/2022 05:40:48 - INFO - codeparrot_training - Step 40241: {'lr': 4.928587592540168e-05, 'samples': 20603904, 'steps': 40241, 'loss/train': 1.510053277015686}
+02/26/2022 05:40:55 - INFO - codeparrot_training - Step 40242: {'lr': 4.92761215083776e-05, 'samples': 20604416, 'steps': 40242, 'loss/train': 3.6378045082092285}
+02/26/2022 05:40:59 - INFO - codeparrot_training - Step 40243: {'lr': 4.9266367951190796e-05, 'samples': 20604928, 'steps': 40243, 'loss/train': 0.13775387406349182}
+02/26/2022 05:41:04 - INFO - codeparrot_training - Step 40244: {'lr': 4.925661525388328e-05, 'samples': 20605440, 'steps': 40244, 'loss/train': 1.8081549406051636}
+02/26/2022 05:41:08 - INFO - codeparrot_training - Step 40245: {'lr': 4.924686341649653e-05, 'samples': 20605952, 'steps': 40245, 'loss/train': 2.084050178527832}
+02/26/2022 05:41:13 - INFO - codeparrot_training - Step 40246: {'lr': 4.9237112439072555e-05, 'samples': 20606464, 'steps': 40246, 'loss/train': 1.5899858474731445}
+02/26/2022 05:41:17 - INFO - codeparrot_training - Step 40247: {'lr': 4.9227362321652954e-05, 'samples': 20606976, 'steps': 40247, 'loss/train': 0.9158003926277161}
+02/26/2022 05:41:22 - INFO - codeparrot_training - Step 40248: {'lr': 4.9217613064279714e-05, 'samples': 20607488, 'steps': 40248, 'loss/train': 1.5574148893356323}
+02/26/2022 05:41:26 - INFO - codeparrot_training - Step 40249: {'lr': 4.920786466699434e-05, 'samples': 20608000, 'steps': 40249, 'loss/train': 2.0158865451812744}
+02/26/2022 05:41:31 - INFO - codeparrot_training - Step 40250: {'lr': 4.919811712983879e-05, 'samples': 20608512, 'steps': 40250, 'loss/train': 1.5035561323165894}
+02/26/2022 05:41:35 - INFO - codeparrot_training - Step 40251: {'lr': 4.918837045285468e-05, 'samples': 20609024, 'steps': 40251, 'loss/train': 2.6910500526428223}
+02/26/2022 05:41:42 - INFO - codeparrot_training - Step 40252: {'lr': 4.917862463608389e-05, 'samples': 20609536, 'steps': 40252, 'loss/train': 2.20007061958313}
+02/26/2022 05:41:45 - INFO - codeparrot_training - Step 40253: {'lr': 4.9168879679568095e-05, 'samples': 20610048, 'steps': 40253, 'loss/train': 1.6671972274780273}
+02/26/2022 05:41:51 - INFO - codeparrot_training - Step 40254: {'lr': 4.915913558334906e-05, 'samples': 20610560, 'steps': 40254, 'loss/train': 2.116628646850586}
+02/26/2022 05:41:55 - INFO - codeparrot_training - Step 40255: {'lr': 4.914939234746843e-05, 'samples': 20611072, 'steps': 40255, 'loss/train': 2.0490267276763916}
+02/26/2022 05:42:00 - INFO - codeparrot_training - Step 40256: {'lr': 4.91396499719681e-05, 'samples': 20611584, 'steps': 40256, 'loss/train': 1.2523854970932007}
+02/26/2022 05:42:04 - INFO - codeparrot_training - Step 40257: {'lr': 4.912990845688972e-05, 'samples': 20612096, 'steps': 40257, 'loss/train': 1.367032766342163}
+02/26/2022 05:42:09 - INFO - codeparrot_training - Step 40258: {'lr': 4.912016780227504e-05, 'samples': 20612608, 'steps': 40258, 'loss/train': 1.9894146919250488}
+02/26/2022 05:42:13 - INFO - codeparrot_training - Step 40259: {'lr': 4.91104280081657e-05, 'samples': 20613120, 'steps': 40259, 'loss/train': 2.3943426609039307}
+02/26/2022 05:42:18 - INFO - codeparrot_training - Step 40260: {'lr': 4.910068907460358e-05, 'samples': 20613632, 'steps': 40260, 'loss/train': 2.610714912414551}
+02/26/2022 05:42:22 - INFO - codeparrot_training - Step 40261: {'lr': 4.90909510016303e-05, 'samples': 20614144, 'steps': 40261, 'loss/train': 2.104128360748291}
+02/26/2022 05:42:27 - INFO - codeparrot_training - Step 40262: {'lr': 4.908121378928759e-05, 'samples': 20614656, 'steps': 40262, 'loss/train': 0.2344641238451004}
+02/26/2022 05:42:31 - INFO - codeparrot_training - Step 40263: {'lr': 4.907147743761717e-05, 'samples': 20615168, 'steps': 40263, 'loss/train': 1.5972188711166382}
+02/26/2022 05:42:38 - INFO - codeparrot_training - Step 40264: {'lr': 4.906174194666069e-05, 'samples': 20615680, 'steps': 40264, 'loss/train': 1.8211365938186646}
+02/26/2022 05:42:42 - INFO - codeparrot_training - Step 40265: {'lr': 4.905200731645992e-05, 'samples': 20616192, 'steps': 40265, 'loss/train': 1.0623358488082886}
+02/26/2022 05:42:47 - INFO - codeparrot_training - Step 40266: {'lr': 4.904227354705659e-05, 'samples': 20616704, 'steps': 40266, 'loss/train': 2.013404130935669}
+02/26/2022 05:42:51 - INFO - codeparrot_training - Step 40267: {'lr': 4.903254063849233e-05, 'samples': 20617216, 'steps': 40267, 'loss/train': 1.8484675884246826}
+02/26/2022 05:42:56 - INFO - codeparrot_training - Step 40268: {'lr': 4.902280859080876e-05, 'samples': 20617728, 'steps': 40268, 'loss/train': 2.3819262981414795}
+02/26/2022 05:43:00 - INFO - codeparrot_training - Step 40269: {'lr': 4.901307740404776e-05, 'samples': 20618240, 'steps': 40269, 'loss/train': 1.7495012283325195}
+02/26/2022 05:43:05 - INFO - codeparrot_training - Step 40270: {'lr': 4.900334707825091e-05, 'samples': 20618752, 'steps': 40270, 'loss/train': 1.4493780136108398}
+02/26/2022 05:43:09 - INFO - codeparrot_training - Step 40271: {'lr': 4.899361761345988e-05, 'samples': 20619264, 'steps': 40271, 'loss/train': 0.7620633244514465}
+02/26/2022 05:43:14 - INFO - codeparrot_training - Step 40272: {'lr': 4.898388900971634e-05, 'samples': 20619776, 'steps': 40272, 'loss/train': 1.3191192150115967}
+02/26/2022 05:43:18 - INFO - codeparrot_training - Step 40273: {'lr': 4.897416126706203e-05, 'samples': 20620288, 'steps': 40273, 'loss/train': 1.7160799503326416}
+02/26/2022 05:43:23 - INFO - codeparrot_training - Step 40274: {'lr': 4.8964434385538574e-05, 'samples': 20620800, 'steps': 40274, 'loss/train': 1.4717628955841064}
+02/26/2022 05:43:27 - INFO - codeparrot_training - Step 40275: {'lr': 4.8954708365187646e-05, 'samples': 20621312, 'steps': 40275, 'loss/train': 2.196061849594116}
+02/26/2022 05:43:32 - INFO - codeparrot_training - Step 40276: {'lr': 4.894498320605084e-05, 'samples': 20621824, 'steps': 40276, 'loss/train': 1.260366439819336}
+02/26/2022 05:43:36 - INFO - codeparrot_training - Step 40277: {'lr': 4.893525890816997e-05, 'samples': 20622336, 'steps': 40277, 'loss/train': 1.4206507205963135}
+02/26/2022 05:43:43 - INFO - codeparrot_training - Step 40278: {'lr': 4.8925535471586515e-05, 'samples': 20622848, 'steps': 40278, 'loss/train': 0.799635112285614}
+02/26/2022 05:43:46 - INFO - codeparrot_training - Step 40279: {'lr': 4.8915812896342363e-05, 'samples': 20623360, 'steps': 40279, 'loss/train': 4.539821624755859}
+02/26/2022 05:43:52 - INFO - codeparrot_training - Step 40280: {'lr': 4.890609118247888e-05, 'samples': 20623872, 'steps': 40280, 'loss/train': 2.939774990081787}
+02/26/2022 05:43:55 - INFO - codeparrot_training - Step 40281: {'lr': 4.889637033003794e-05, 'samples': 20624384, 'steps': 40281, 'loss/train': 0.7319871783256531}
+02/26/2022 05:44:01 - INFO - codeparrot_training - Step 40282: {'lr': 4.888665033906098e-05, 'samples': 20624896, 'steps': 40282, 'loss/train': 1.9546502828598022}
+02/26/2022 05:44:04 - INFO - codeparrot_training - Step 40283: {'lr': 4.887693120958991e-05, 'samples': 20625408, 'steps': 40283, 'loss/train': 1.8348294496536255}
+02/26/2022 05:44:10 - INFO - codeparrot_training - Step 40284: {'lr': 4.8867212941666075e-05, 'samples': 20625920, 'steps': 40284, 'loss/train': 1.8929299116134644}
+02/26/2022 05:44:13 - INFO - codeparrot_training - Step 40285: {'lr': 4.885749553533128e-05, 'samples': 20626432, 'steps': 40285, 'loss/train': 1.8761297464370728}
+02/26/2022 05:44:19 - INFO - codeparrot_training - Step 40286: {'lr': 4.8847778990627065e-05, 'samples': 20626944, 'steps': 40286, 'loss/train': 1.6109845638275146}
+02/26/2022 05:44:22 - INFO - codeparrot_training - Step 40287: {'lr': 4.883806330759513e-05, 'samples': 20627456, 'steps': 40287, 'loss/train': 1.7778126001358032}
+02/26/2022 05:44:29 - INFO - codeparrot_training - Step 40288: {'lr': 4.882834848627707e-05, 'samples': 20627968, 'steps': 40288, 'loss/train': 1.9402728080749512}
+02/26/2022 05:44:33 - INFO - codeparrot_training - Step 40289: {'lr': 4.8818634526714475e-05, 'samples': 20628480, 'steps': 40289, 'loss/train': 1.888139009475708}
+02/26/2022 05:44:39 - INFO - codeparrot_training - Step 40290: {'lr': 4.880892142894891e-05, 'samples': 20628992, 'steps': 40290, 'loss/train': 1.9593098163604736}
+02/26/2022 05:44:42 - INFO - codeparrot_training - Step 40291: {'lr': 4.879920919302211e-05, 'samples': 20629504, 'steps': 40291, 'loss/train': 2.0112409591674805}
+02/26/2022 05:44:47 - INFO - codeparrot_training - Step 40292: {'lr': 4.87894978189756e-05, 'samples': 20630016, 'steps': 40292, 'loss/train': 1.7236138582229614}
+02/26/2022 05:44:51 - INFO - codeparrot_training - Step 40293: {'lr': 4.877978730685098e-05, 'samples': 20630528, 'steps': 40293, 'loss/train': 2.652998208999634}
+02/26/2022 05:44:57 - INFO - codeparrot_training - Step 40294: {'lr': 4.877007765668981e-05, 'samples': 20631040, 'steps': 40294, 'loss/train': 2.97902774810791}
+02/26/2022 05:45:00 - INFO - codeparrot_training - Step 40295: {'lr': 4.87603688685338e-05, 'samples': 20631552, 'steps': 40295, 'loss/train': 0.0480986088514328}
+02/26/2022 05:45:06 - INFO - codeparrot_training - Step 40296: {'lr': 4.875066094242445e-05, 'samples': 20632064, 'steps': 40296, 'loss/train': 1.9318784475326538}
+02/26/2022 05:45:13 - INFO - codeparrot_training - Step 40297: {'lr': 4.874095387840338e-05, 'samples': 20632576, 'steps': 40297, 'loss/train': 1.6873944997787476}
+02/26/2022 05:45:16 - INFO - codeparrot_training - Step 40298: {'lr': 4.8731247676512076e-05, 'samples': 20633088, 'steps': 40298, 'loss/train': 2.169782876968384}
+02/26/2022 05:45:22 - INFO - codeparrot_training - Step 40299: {'lr': 4.8721542336792266e-05, 'samples': 20633600, 'steps': 40299, 'loss/train': 1.590254783630371}
+02/26/2022 05:45:25 - INFO - codeparrot_training - Step 40300: {'lr': 4.871183785928546e-05, 'samples': 20634112, 'steps': 40300, 'loss/train': 0.49390989542007446}
+02/26/2022 05:45:31 - INFO - codeparrot_training - Step 40301: {'lr': 4.8702134244033216e-05, 'samples': 20634624, 'steps': 40301, 'loss/train': 1.6930512189865112}
+02/26/2022 05:45:34 - INFO - codeparrot_training - Step 40302: {'lr': 4.8692431491077075e-05, 'samples': 20635136, 'steps': 40302, 'loss/train': 1.3124794960021973}
+02/26/2022 05:45:40 - INFO - codeparrot_training - Step 40303: {'lr': 4.868272960045866e-05, 'samples': 20635648, 'steps': 40303, 'loss/train': 2.376016139984131}
+02/26/2022 05:45:43 - INFO - codeparrot_training - Step 40304: {'lr': 4.867302857221953e-05, 'samples': 20636160, 'steps': 40304, 'loss/train': 2.35194730758667}
+02/26/2022 05:45:49 - INFO - codeparrot_training - Step 40305: {'lr': 4.8663328406401226e-05, 'samples': 20636672, 'steps': 40305, 'loss/train': 1.8355700969696045}
+02/26/2022 05:45:52 - INFO - codeparrot_training - Step 40306: {'lr': 4.8653629103045214e-05, 'samples': 20637184, 'steps': 40306, 'loss/train': 2.4623522758483887}
+02/26/2022 05:46:00 - INFO - codeparrot_training - Step 40307: {'lr': 4.8643930662193166e-05, 'samples': 20637696, 'steps': 40307, 'loss/train': 1.516063928604126}
+02/26/2022 05:46:03 - INFO - codeparrot_training - Step 40308: {'lr': 4.863423308388659e-05, 'samples': 20638208, 'steps': 40308, 'loss/train': 1.8128044605255127}
+02/26/2022 05:46:09 - INFO - codeparrot_training - Step 40309: {'lr': 4.862453636816705e-05, 'samples': 20638720, 'steps': 40309, 'loss/train': 1.3424099683761597}
+02/26/2022 05:46:12 - INFO - codeparrot_training - Step 40310: {'lr': 4.861484051507603e-05, 'samples': 20639232, 'steps': 40310, 'loss/train': 1.806500792503357}
+02/26/2022 05:46:18 - INFO - codeparrot_training - Step 40311: {'lr': 4.8605145524655e-05, 'samples': 20639744, 'steps': 40311, 'loss/train': 1.4328997135162354}
+02/26/2022 05:46:21 - INFO - codeparrot_training - Step 40312: {'lr': 4.8595451396945685e-05, 'samples': 20640256, 'steps': 40312, 'loss/train': 0.896848738193512}
+02/26/2022 05:46:27 - INFO - codeparrot_training - Step 40313: {'lr': 4.85857581319894e-05, 'samples': 20640768, 'steps': 40313, 'loss/train': 3.0436997413635254}
+02/26/2022 05:46:30 - INFO - codeparrot_training - Step 40314: {'lr': 4.857606572982792e-05, 'samples': 20641280, 'steps': 40314, 'loss/train': 1.6188586950302124}
+02/26/2022 05:46:36 - INFO - codeparrot_training - Step 40315: {'lr': 4.856637419050247e-05, 'samples': 20641792, 'steps': 40315, 'loss/train': 1.3449711799621582}
+02/26/2022 05:46:39 - INFO - codeparrot_training - Step 40316: {'lr': 4.855668351405479e-05, 'samples': 20642304, 'steps': 40316, 'loss/train': 1.1660408973693848}
+02/26/2022 05:46:45 - INFO - codeparrot_training - Step 40317: {'lr': 4.854699370052626e-05, 'samples': 20642816, 'steps': 40317, 'loss/train': 1.2835907936096191}
+02/26/2022 05:46:48 - INFO - codeparrot_training - Step 40318: {'lr': 4.8537304749958544e-05, 'samples': 20643328, 'steps': 40318, 'loss/train': 2.2396318912506104}
+02/26/2022 05:46:54 - INFO - codeparrot_training - Step 40319: {'lr': 4.8527616662392914e-05, 'samples': 20643840, 'steps': 40319, 'loss/train': 2.143279790878296}
+02/26/2022 05:46:57 - INFO - codeparrot_training - Step 40320: {'lr': 4.851792943787109e-05, 'samples': 20644352, 'steps': 40320, 'loss/train': 1.4868887662887573}
+02/26/2022 05:47:03 - INFO - codeparrot_training - Step 40321: {'lr': 4.850824307643439e-05, 'samples': 20644864, 'steps': 40321, 'loss/train': 1.2822750806808472}
+02/26/2022 05:47:06 - INFO - codeparrot_training - Step 40322: {'lr': 4.849855757812455e-05, 'samples': 20645376, 'steps': 40322, 'loss/train': 2.142866611480713}
+02/26/2022 05:47:13 - INFO - codeparrot_training - Step 40323: {'lr': 4.848887294298274e-05, 'samples': 20645888, 'steps': 40323, 'loss/train': 1.1699045896530151}
+02/26/2022 05:47:17 - INFO - codeparrot_training - Step 40324: {'lr': 4.8479189171050706e-05, 'samples': 20646400, 'steps': 40324, 'loss/train': 0.7557324767112732}
+02/26/2022 05:47:22 - INFO - codeparrot_training - Step 40325: {'lr': 4.846950626236976e-05, 'samples': 20646912, 'steps': 40325, 'loss/train': 1.879967212677002}
+02/26/2022 05:47:26 - INFO - codeparrot_training - Step 40326: {'lr': 4.84598242169815e-05, 'samples': 20647424, 'steps': 40326, 'loss/train': 2.1193058490753174}
+02/26/2022 05:47:31 - INFO - codeparrot_training - Step 40327: {'lr': 4.845014303492739e-05, 'samples': 20647936, 'steps': 40327, 'loss/train': 0.20236289501190186}
+02/26/2022 05:47:35 - INFO - codeparrot_training - Step 40328: {'lr': 4.844046271624886e-05, 'samples': 20648448, 'steps': 40328, 'loss/train': 1.8047990798950195}
+02/26/2022 05:47:41 - INFO - codeparrot_training - Step 40329: {'lr': 4.843078326098732e-05, 'samples': 20648960, 'steps': 40329, 'loss/train': 1.8903441429138184}
+02/26/2022 05:47:44 - INFO - codeparrot_training - Step 40330: {'lr': 4.842110466918434e-05, 'samples': 20649472, 'steps': 40330, 'loss/train': 3.083993673324585}
+02/26/2022 05:47:50 - INFO - codeparrot_training - Step 40331: {'lr': 4.841142694088138e-05, 'samples': 20649984, 'steps': 40331, 'loss/train': 1.4143931865692139}
+02/26/2022 05:47:53 - INFO - codeparrot_training - Step 40332: {'lr': 4.8401750076119827e-05, 'samples': 20650496, 'steps': 40332, 'loss/train': 1.3540617227554321}
+02/26/2022 05:47:58 - INFO - codeparrot_training - Step 40333: {'lr': 4.839207407494109e-05, 'samples': 20651008, 'steps': 40333, 'loss/train': 2.186516523361206}
+02/26/2022 05:48:02 - INFO - codeparrot_training - Step 40334: {'lr': 4.838239893738677e-05, 'samples': 20651520, 'steps': 40334, 'loss/train': 1.4937856197357178}
+02/26/2022 05:48:09 - INFO - codeparrot_training - Step 40335: {'lr': 4.837272466349818e-05, 'samples': 20652032, 'steps': 40335, 'loss/train': 1.642652153968811}
+02/26/2022 05:48:12 - INFO - codeparrot_training - Step 40336: {'lr': 4.836305125331694e-05, 'samples': 20652544, 'steps': 40336, 'loss/train': 1.775629997253418}
+02/26/2022 05:48:18 - INFO - codeparrot_training - Step 40337: {'lr': 4.835337870688422e-05, 'samples': 20653056, 'steps': 40337, 'loss/train': 1.382092833518982}
+02/26/2022 05:48:24 - INFO - codeparrot_training - Step 40338: {'lr': 4.834370702424165e-05, 'samples': 20653568, 'steps': 40338, 'loss/train': 2.4745702743530273}
+02/26/2022 05:48:27 - INFO - codeparrot_training - Step 40339: {'lr': 4.833403620543056e-05, 'samples': 20654080, 'steps': 40339, 'loss/train': 1.7895512580871582}
+02/26/2022 05:48:33 - INFO - codeparrot_training - Step 40340: {'lr': 4.8324366250492553e-05, 'samples': 20654592, 'steps': 40340, 'loss/train': 1.7502992153167725}
+02/26/2022 05:48:36 - INFO - codeparrot_training - Step 40341: {'lr': 4.8314697159468764e-05, 'samples': 20655104, 'steps': 40341, 'loss/train': 1.076488971710205}
+02/26/2022 05:48:41 - INFO - codeparrot_training - Step 40342: {'lr': 4.8305028932400854e-05, 'samples': 20655616, 'steps': 40342, 'loss/train': 1.9662400484085083}
+02/26/2022 05:48:45 - INFO - codeparrot_training - Step 40343: {'lr': 4.829536156933018e-05, 'samples': 20656128, 'steps': 40343, 'loss/train': 1.3128210306167603}
+02/26/2022 05:48:52 - INFO - codeparrot_training - Step 40344: {'lr': 4.8285695070298094e-05, 'samples': 20656640, 'steps': 40344, 'loss/train': 1.4712445735931396}
+02/26/2022 05:48:56 - INFO - codeparrot_training - Step 40345: {'lr': 4.827602943534598e-05, 'samples': 20657152, 'steps': 40345, 'loss/train': 1.8644556999206543}
+02/26/2022 05:49:01 - INFO - codeparrot_training - Step 40346: {'lr': 4.826636466451537e-05, 'samples': 20657664, 'steps': 40346, 'loss/train': 1.7470310926437378}
+02/26/2022 05:49:05 - INFO - codeparrot_training - Step 40347: {'lr': 4.82567007578476e-05, 'samples': 20658176, 'steps': 40347, 'loss/train': 2.2780463695526123}
+02/26/2022 05:49:11 - INFO - codeparrot_training - Step 40348: {'lr': 4.8247037715384e-05, 'samples': 20658688, 'steps': 40348, 'loss/train': 2.12211537361145}
+02/26/2022 05:49:14 - INFO - codeparrot_training - Step 40349: {'lr': 4.823737553716609e-05, 'samples': 20659200, 'steps': 40349, 'loss/train': 5.402233123779297}
+02/26/2022 05:49:18 - INFO - codeparrot_training - Step 40350: {'lr': 4.8227714223235185e-05, 'samples': 20659712, 'steps': 40350, 'loss/train': 8.754063606262207}
+02/26/2022 05:49:23 - INFO - codeparrot_training - Step 40351: {'lr': 4.82180537736327e-05, 'samples': 20660224, 'steps': 40351, 'loss/train': 1.3048845529556274}
+02/26/2022 05:49:27 - INFO - codeparrot_training - Step 40352: {'lr': 4.820839418839992e-05, 'samples': 20660736, 'steps': 40352, 'loss/train': 1.3804601430892944}
+02/26/2022 05:49:32 - INFO - codeparrot_training - Step 40353: {'lr': 4.819873546757836e-05, 'samples': 20661248, 'steps': 40353, 'loss/train': 1.3691766262054443}
+02/26/2022 05:49:36 - INFO - codeparrot_training - Step 40354: {'lr': 4.818907761120936e-05, 'samples': 20661760, 'steps': 40354, 'loss/train': 1.8946551084518433}
+02/26/2022 05:49:43 - INFO - codeparrot_training - Step 40355: {'lr': 4.817942061933425e-05, 'samples': 20662272, 'steps': 40355, 'loss/train': 2.6421525478363037}
+02/26/2022 05:49:46 - INFO - codeparrot_training - Step 40356: {'lr': 4.816976449199437e-05, 'samples': 20662784, 'steps': 40356, 'loss/train': 2.05658221244812}
+02/26/2022 05:49:52 - INFO - codeparrot_training - Step 40357: {'lr': 4.816010922923125e-05, 'samples': 20663296, 'steps': 40357, 'loss/train': 1.6093169450759888}
+02/26/2022 05:49:55 - INFO - codeparrot_training - Step 40358: {'lr': 4.8150454831086014e-05, 'samples': 20663808, 'steps': 40358, 'loss/train': 1.247710108757019}
+02/26/2022 05:50:01 - INFO - codeparrot_training - Step 40359: {'lr': 4.8140801297600175e-05, 'samples': 20664320, 'steps': 40359, 'loss/train': 1.6580345630645752}
+02/26/2022 05:50:04 - INFO - codeparrot_training - Step 40360: {'lr': 4.8131148628815016e-05, 'samples': 20664832, 'steps': 40360, 'loss/train': 2.1830527782440186}
+02/26/2022 05:50:10 - INFO - codeparrot_training - Step 40361: {'lr': 4.812149682477196e-05, 'samples': 20665344, 'steps': 40361, 'loss/train': 2.115399122238159}
+02/26/2022 05:50:13 - INFO - codeparrot_training - Step 40362: {'lr': 4.811184588551232e-05, 'samples': 20665856, 'steps': 40362, 'loss/train': 0.8488881587982178}
+02/26/2022 05:50:19 - INFO - codeparrot_training - Step 40363: {'lr': 4.810219581107739e-05, 'samples': 20666368, 'steps': 40363, 'loss/train': 2.124912738800049}
+02/26/2022 05:50:26 - INFO - codeparrot_training - Step 40364: {'lr': 4.809254660150852e-05, 'samples': 20666880, 'steps': 40364, 'loss/train': 2.0002939701080322}
+02/26/2022 05:50:30 - INFO - codeparrot_training - Step 40365: {'lr': 4.808289825684711e-05, 'samples': 20667392, 'steps': 40365, 'loss/train': 1.3772315979003906}
+02/26/2022 05:50:35 - INFO - codeparrot_training - Step 40366: {'lr': 4.8073250777134456e-05, 'samples': 20667904, 'steps': 40366, 'loss/train': 2.5757248401641846}
+02/26/2022 05:50:39 - INFO - codeparrot_training - Step 40367: {'lr': 4.8063604162411875e-05, 'samples': 20668416, 'steps': 40367, 'loss/train': 1.1028145551681519}
+02/26/2022 05:50:44 - INFO - codeparrot_training - Step 40368: {'lr': 4.8053958412720615e-05, 'samples': 20668928, 'steps': 40368, 'loss/train': 0.9846283197402954}
+02/26/2022 05:50:48 - INFO - codeparrot_training - Step 40369: {'lr': 4.804431352810215e-05, 'samples': 20669440, 'steps': 40369, 'loss/train': 0.6975376009941101}
+02/26/2022 05:50:54 - INFO - codeparrot_training - Step 40370: {'lr': 4.8034669508597726e-05, 'samples': 20669952, 'steps': 40370, 'loss/train': 1.8684865236282349}
+02/26/2022 05:50:57 - INFO - codeparrot_training - Step 40371: {'lr': 4.802502635424863e-05, 'samples': 20670464, 'steps': 40371, 'loss/train': 2.1115877628326416}
+02/26/2022 05:51:00 - INFO - codeparrot_training - Step 40372: {'lr': 4.801538406509612e-05, 'samples': 20670976, 'steps': 40372, 'loss/train': 1.3481011390686035}
+02/26/2022 05:51:06 - INFO - codeparrot_training - Step 40373: {'lr': 4.8005742641181656e-05, 'samples': 20671488, 'steps': 40373, 'loss/train': 1.3121333122253418}
+02/26/2022 05:51:09 - INFO - codeparrot_training - Step 40374: {'lr': 4.799610208254635e-05, 'samples': 20672000, 'steps': 40374, 'loss/train': 1.856300711631775}
+02/26/2022 05:51:15 - INFO - codeparrot_training - Step 40375: {'lr': 4.7986462389231766e-05, 'samples': 20672512, 'steps': 40375, 'loss/train': 1.6852102279663086}
+02/26/2022 05:51:19 - INFO - codeparrot_training - Step 40376: {'lr': 4.797682356127886e-05, 'samples': 20673024, 'steps': 40376, 'loss/train': 3.256267547607422}
+02/26/2022 05:51:24 - INFO - codeparrot_training - Step 40377: {'lr': 4.79671855987292e-05, 'samples': 20673536, 'steps': 40377, 'loss/train': 1.380620002746582}
+02/26/2022 05:51:27 - INFO - codeparrot_training - Step 40378: {'lr': 4.795754850162387e-05, 'samples': 20674048, 'steps': 40378, 'loss/train': 1.87704336643219}
+02/26/2022 05:51:33 - INFO - codeparrot_training - Step 40379: {'lr': 4.794791227000439e-05, 'samples': 20674560, 'steps': 40379, 'loss/train': 0.9133546948432922}
+02/26/2022 05:51:36 - INFO - codeparrot_training - Step 40380: {'lr': 4.7938276903911754e-05, 'samples': 20675072, 'steps': 40380, 'loss/train': 1.7998042106628418}
+02/26/2022 05:51:44 - INFO - codeparrot_training - Step 40381: {'lr': 4.7928642403387456e-05, 'samples': 20675584, 'steps': 40381, 'loss/train': 1.4639952182769775}
+02/26/2022 05:51:47 - INFO - codeparrot_training - Step 40382: {'lr': 4.791900876847261e-05, 'samples': 20676096, 'steps': 40382, 'loss/train': 1.407915472984314}
+02/26/2022 05:51:53 - INFO - codeparrot_training - Step 40383: {'lr': 4.790937599920872e-05, 'samples': 20676608, 'steps': 40383, 'loss/train': 1.1989779472351074}
+02/26/2022 05:51:56 - INFO - codeparrot_training - Step 40384: {'lr': 4.789974409563674e-05, 'samples': 20677120, 'steps': 40384, 'loss/train': 1.5867364406585693}
+02/26/2022 05:52:02 - INFO - codeparrot_training - Step 40385: {'lr': 4.789011305779814e-05, 'samples': 20677632, 'steps': 40385, 'loss/train': 1.241894245147705}
+02/26/2022 05:52:05 - INFO - codeparrot_training - Step 40386: {'lr': 4.788048288573407e-05, 'samples': 20678144, 'steps': 40386, 'loss/train': 0.5858150720596313}
+02/26/2022 05:52:11 - INFO - codeparrot_training - Step 40387: {'lr': 4.7870853579485896e-05, 'samples': 20678656, 'steps': 40387, 'loss/train': 1.0563205480575562}
+02/26/2022 05:52:14 - INFO - codeparrot_training - Step 40388: {'lr': 4.7861225139094774e-05, 'samples': 20679168, 'steps': 40388, 'loss/train': 1.2035837173461914}
+02/26/2022 05:52:20 - INFO - codeparrot_training - Step 40389: {'lr': 4.7851597564602e-05, 'samples': 20679680, 'steps': 40389, 'loss/train': 1.666856288909912}
+02/26/2022 05:52:23 - INFO - codeparrot_training - Step 40390: {'lr': 4.78419708560488e-05, 'samples': 20680192, 'steps': 40390, 'loss/train': 1.10679030418396}
+02/26/2022 05:52:30 - INFO - codeparrot_training - Step 40391: {'lr': 4.783234501347633e-05, 'samples': 20680704, 'steps': 40391, 'loss/train': 0.051086753606796265}
+02/26/2022 05:52:34 - INFO - codeparrot_training - Step 40392: {'lr': 4.7822720036925975e-05, 'samples': 20681216, 'steps': 40392, 'loss/train': 1.8546780347824097}
+02/26/2022 05:52:39 - INFO - codeparrot_training - Step 40393: {'lr': 4.781309592643887e-05, 'samples': 20681728, 'steps': 40393, 'loss/train': 2.0365428924560547}
+02/26/2022 05:52:45 - INFO - codeparrot_training - Step 40394: {'lr': 4.780347268205626e-05, 'samples': 20682240, 'steps': 40394, 'loss/train': 1.6743618249893188}
+02/26/2022 05:52:48 - INFO - codeparrot_training - Step 40395: {'lr': 4.7793850303819334e-05, 'samples': 20682752, 'steps': 40395, 'loss/train': 0.9641406536102295}
+02/26/2022 05:52:52 - INFO - codeparrot_training - Step 40396: {'lr': 4.7784228791769386e-05, 'samples': 20683264, 'steps': 40396, 'loss/train': 1.098073124885559}
+02/26/2022 05:52:57 - INFO - codeparrot_training - Step 40397: {'lr': 4.777460814594758e-05, 'samples': 20683776, 'steps': 40397, 'loss/train': 0.37153273820877075}
+02/26/2022 05:53:03 - INFO - codeparrot_training - Step 40398: {'lr': 4.776498836639515e-05, 'samples': 20684288, 'steps': 40398, 'loss/train': 1.2333284616470337}
+02/26/2022 05:53:06 - INFO - codeparrot_training - Step 40399: {'lr': 4.775536945315323e-05, 'samples': 20684800, 'steps': 40399, 'loss/train': 0.9079501032829285}
+02/26/2022 05:53:13 - INFO - codeparrot_training - Step 40400: {'lr': 4.7745751406263163e-05, 'samples': 20685312, 'steps': 40400, 'loss/train': 1.0404052734375}
+02/26/2022 05:53:16 - INFO - codeparrot_training - Step 40401: {'lr': 4.773613422576606e-05, 'samples': 20685824, 'steps': 40401, 'loss/train': 1.9159491062164307}
+02/26/2022 05:53:22 - INFO - codeparrot_training - Step 40402: {'lr': 4.7726517911703124e-05, 'samples': 20686336, 'steps': 40402, 'loss/train': 1.39814031124115}
+02/26/2022 05:53:25 - INFO - codeparrot_training - Step 40403: {'lr': 4.7716902464115504e-05, 'samples': 20686848, 'steps': 40403, 'loss/train': 2.0115244388580322}
+02/26/2022 05:53:31 - INFO - codeparrot_training - Step 40404: {'lr': 4.770728788304451e-05, 'samples': 20687360, 'steps': 40404, 'loss/train': 0.49342501163482666}
+02/26/2022 05:53:34 - INFO - codeparrot_training - Step 40405: {'lr': 4.769767416853127e-05, 'samples': 20687872, 'steps': 40405, 'loss/train': 2.1932361125946045}
+02/26/2022 05:53:40 - INFO - codeparrot_training - Step 40406: {'lr': 4.768806132061693e-05, 'samples': 20688384, 'steps': 40406, 'loss/train': 2.0098977088928223}
+02/26/2022 05:53:43 - INFO - codeparrot_training - Step 40407: {'lr': 4.767844933934265e-05, 'samples': 20688896, 'steps': 40407, 'loss/train': 1.7678189277648926}
+02/26/2022 05:53:47 - INFO - codeparrot_training - Step 40408: {'lr': 4.7668838224749714e-05, 'samples': 20689408, 'steps': 40408, 'loss/train': 1.4884628057479858}
+02/26/2022 05:53:53 - INFO - codeparrot_training - Step 40409: {'lr': 4.7659227976879134e-05, 'samples': 20689920, 'steps': 40409, 'loss/train': 1.2438020706176758}
+02/26/2022 05:53:56 - INFO - codeparrot_training - Step 40410: {'lr': 4.764961859577233e-05, 'samples': 20690432, 'steps': 40410, 'loss/train': 2.1484482288360596}
+02/26/2022 05:54:02 - INFO - codeparrot_training - Step 40411: {'lr': 4.764001008147015e-05, 'samples': 20690944, 'steps': 40411, 'loss/train': 0.8591733574867249}
+02/26/2022 05:54:05 - INFO - codeparrot_training - Step 40412: {'lr': 4.7630402434014006e-05, 'samples': 20691456, 'steps': 40412, 'loss/train': 1.1374257802963257}
+02/26/2022 05:54:11 - INFO - codeparrot_training - Step 40413: {'lr': 4.762079565344488e-05, 'samples': 20691968, 'steps': 40413, 'loss/train': 0.7427809238433838}
+02/26/2022 05:54:15 - INFO - codeparrot_training - Step 40414: {'lr': 4.761118973980413e-05, 'samples': 20692480, 'steps': 40414, 'loss/train': 0.22924913465976715}
+02/26/2022 05:54:20 - INFO - codeparrot_training - Step 40415: {'lr': 4.7601584693132636e-05, 'samples': 20692992, 'steps': 40415, 'loss/train': 1.5678691864013672}
+02/26/2022 05:54:24 - INFO - codeparrot_training - Step 40416: {'lr': 4.759198051347177e-05, 'samples': 20693504, 'steps': 40416, 'loss/train': 1.2099350690841675}
+02/26/2022 05:54:29 - INFO - codeparrot_training - Step 40417: {'lr': 4.7582377200862504e-05, 'samples': 20694016, 'steps': 40417, 'loss/train': 1.4026752710342407}
+02/26/2022 05:54:33 - INFO - codeparrot_training - Step 40418: {'lr': 4.757277475534621e-05, 'samples': 20694528, 'steps': 40418, 'loss/train': 1.8972070217132568}
+02/26/2022 05:54:38 - INFO - codeparrot_training - Step 40419: {'lr': 4.7563173176963734e-05, 'samples': 20695040, 'steps': 40419, 'loss/train': 2.4594573974609375}
+02/26/2022 05:54:42 - INFO - codeparrot_training - Step 40420: {'lr': 4.7553572465756396e-05, 'samples': 20695552, 'steps': 40420, 'loss/train': 1.88540518283844}
+02/26/2022 05:54:47 - INFO - codeparrot_training - Step 40421: {'lr': 4.7543972621765213e-05, 'samples': 20696064, 'steps': 40421, 'loss/train': 0.7198459506034851}
+02/26/2022 05:54:51 - INFO - codeparrot_training - Step 40422: {'lr': 4.7534373645031435e-05, 'samples': 20696576, 'steps': 40422, 'loss/train': 1.031506896018982}
+02/26/2022 05:54:56 - INFO - codeparrot_training - Step 40423: {'lr': 4.752477553559612e-05, 'samples': 20697088, 'steps': 40423, 'loss/train': 2.101296901702881}
+02/26/2022 05:55:00 - INFO - codeparrot_training - Step 40424: {'lr': 4.7515178293500354e-05, 'samples': 20697600, 'steps': 40424, 'loss/train': 1.5176054239273071}
+02/26/2022 05:55:05 - INFO - codeparrot_training - Step 40425: {'lr': 4.7505581918785206e-05, 'samples': 20698112, 'steps': 40425, 'loss/train': 1.1849312782287598}
+02/26/2022 05:55:09 - INFO - codeparrot_training - Step 40426: {'lr': 4.7495986411491915e-05, 'samples': 20698624, 'steps': 40426, 'loss/train': 2.3570902347564697}
+02/26/2022 05:55:15 - INFO - codeparrot_training - Step 40427: {'lr': 4.7486391771661505e-05, 'samples': 20699136, 'steps': 40427, 'loss/train': 1.1125259399414062}
+02/26/2022 05:55:18 - INFO - codeparrot_training - Step 40428: {'lr': 4.74767979993351e-05, 'samples': 20699648, 'steps': 40428, 'loss/train': 2.4073944091796875}
+02/26/2022 05:55:24 - INFO - codeparrot_training - Step 40429: {'lr': 4.746720509455371e-05, 'samples': 20700160, 'steps': 40429, 'loss/train': 2.1372411251068115}
+02/26/2022 05:55:27 - INFO - codeparrot_training - Step 40430: {'lr': 4.745761305735857e-05, 'samples': 20700672, 'steps': 40430, 'loss/train': 1.3745031356811523}
+02/26/2022 05:55:33 - INFO - codeparrot_training - Step 40431: {'lr': 4.744802188779071e-05, 'samples': 20701184, 'steps': 40431, 'loss/train': 1.794533371925354}
+02/26/2022 05:55:37 - INFO - codeparrot_training - Step 40432: {'lr': 4.74384315858912e-05, 'samples': 20701696, 'steps': 40432, 'loss/train': 1.7580296993255615}
+02/26/2022 05:55:42 - INFO - codeparrot_training - Step 40433: {'lr': 4.742884215170107e-05, 'samples': 20702208, 'steps': 40433, 'loss/train': 1.7651078701019287}
+02/26/2022 05:55:46 - INFO - codeparrot_training - Step 40434: {'lr': 4.74192535852615e-05, 'samples': 20702720, 'steps': 40434, 'loss/train': 0.9584648013114929}
+02/26/2022 05:55:51 - INFO - codeparrot_training - Step 40435: {'lr': 4.740966588661355e-05, 'samples': 20703232, 'steps': 40435, 'loss/train': 1.8396974802017212}
+02/26/2022 05:55:55 - INFO - codeparrot_training - Step 40436: {'lr': 4.740007905579824e-05, 'samples': 20703744, 'steps': 40436, 'loss/train': 2.1046106815338135}
+02/26/2022 05:56:00 - INFO - codeparrot_training - Step 40437: {'lr': 4.739049309285667e-05, 'samples': 20704256, 'steps': 40437, 'loss/train': 2.063457727432251}
+02/26/2022 05:56:04 - INFO - codeparrot_training - Step 40438: {'lr': 4.738090799782982e-05, 'samples': 20704768, 'steps': 40438, 'loss/train': 2.6092586517333984}
+02/26/2022 05:56:10 - INFO - codeparrot_training - Step 40439: {'lr': 4.737132377075889e-05, 'samples': 20705280, 'steps': 40439, 'loss/train': 1.60270094871521}
+02/26/2022 05:56:13 - INFO - codeparrot_training - Step 40440: {'lr': 4.736174041168487e-05, 'samples': 20705792, 'steps': 40440, 'loss/train': 1.2395126819610596}
+02/26/2022 05:56:19 - INFO - codeparrot_training - Step 40441: {'lr': 4.735215792064882e-05, 'samples': 20706304, 'steps': 40441, 'loss/train': 1.4732745885849}
+02/26/2022 05:56:22 - INFO - codeparrot_training - Step 40442: {'lr': 4.734257629769168e-05, 'samples': 20706816, 'steps': 40442, 'loss/train': 1.563485860824585}
+02/26/2022 05:56:28 - INFO - codeparrot_training - Step 40443: {'lr': 4.733299554285467e-05, 'samples': 20707328, 'steps': 40443, 'loss/train': 2.1985204219818115}
+02/26/2022 05:56:31 - INFO - codeparrot_training - Step 40444: {'lr': 4.73234156561787e-05, 'samples': 20707840, 'steps': 40444, 'loss/train': 0.5319942831993103}
+02/26/2022 05:56:37 - INFO - codeparrot_training - Step 40445: {'lr': 4.7313836637704996e-05, 'samples': 20708352, 'steps': 40445, 'loss/train': 2.4166722297668457}
+02/26/2022 05:56:40 - INFO - codeparrot_training - Step 40446: {'lr': 4.73042584874743e-05, 'samples': 20708864, 'steps': 40446, 'loss/train': 2.171186923980713}
+02/26/2022 05:56:46 - INFO - codeparrot_training - Step 40447: {'lr': 4.729468120552788e-05, 'samples': 20709376, 'steps': 40447, 'loss/train': 1.713104248046875}
+02/26/2022 05:56:49 - INFO - codeparrot_training - Step 40448: {'lr': 4.7285104791906617e-05, 'samples': 20709888, 'steps': 40448, 'loss/train': 1.9176017045974731}
+02/26/2022 05:56:56 - INFO - codeparrot_training - Step 40449: {'lr': 4.727552924665171e-05, 'samples': 20710400, 'steps': 40449, 'loss/train': 1.979875922203064}
+02/26/2022 05:57:01 - INFO - codeparrot_training - Step 40450: {'lr': 4.7265954569803955e-05, 'samples': 20710912, 'steps': 40450, 'loss/train': 3.5115139484405518}
+02/26/2022 05:57:05 - INFO - codeparrot_training - Step 40451: {'lr': 4.725638076140451e-05, 'samples': 20711424, 'steps': 40451, 'loss/train': 1.6537821292877197}
+02/26/2022 05:57:10 - INFO - codeparrot_training - Step 40452: {'lr': 4.7246807821494317e-05, 'samples': 20711936, 'steps': 40452, 'loss/train': 1.6634182929992676}
+02/26/2022 05:57:14 - INFO - codeparrot_training - Step 40453: {'lr': 4.723723575011454e-05, 'samples': 20712448, 'steps': 40453, 'loss/train': 1.9879093170166016}
+02/26/2022 05:57:19 - INFO - codeparrot_training - Step 40454: {'lr': 4.7227664547305924e-05, 'samples': 20712960, 'steps': 40454, 'loss/train': 1.7678773403167725}
+02/26/2022 05:57:23 - INFO - codeparrot_training - Step 40455: {'lr': 4.721809421310966e-05, 'samples': 20713472, 'steps': 40455, 'loss/train': 1.6923011541366577}
+02/26/2022 05:57:28 - INFO - codeparrot_training - Step 40456: {'lr': 4.720852474756665e-05, 'samples': 20713984, 'steps': 40456, 'loss/train': 1.6868033409118652}
+02/26/2022 05:57:32 - INFO - codeparrot_training - Step 40457: {'lr': 4.719895615071798e-05, 'samples': 20714496, 'steps': 40457, 'loss/train': 1.614898920059204}
+02/26/2022 05:57:35 - INFO - codeparrot_training - Step 40458: {'lr': 4.718938842260459e-05, 'samples': 20715008, 'steps': 40458, 'loss/train': 1.5221036672592163}
+02/26/2022 05:57:41 - INFO - codeparrot_training - Step 40459: {'lr': 4.7179821563267446e-05, 'samples': 20715520, 'steps': 40459, 'loss/train': 1.9841527938842773}
+02/26/2022 05:57:45 - INFO - codeparrot_training - Step 40460: {'lr': 4.7170255572747485e-05, 'samples': 20716032, 'steps': 40460, 'loss/train': 1.0592180490493774}
+02/26/2022 05:57:50 - INFO - codeparrot_training - Step 40461: {'lr': 4.7160690451085814e-05, 'samples': 20716544, 'steps': 40461, 'loss/train': 1.9526796340942383}
+02/26/2022 05:57:54 - INFO - codeparrot_training - Step 40462: {'lr': 4.715112619832335e-05, 'samples': 20717056, 'steps': 40462, 'loss/train': 0.7734876871109009}
+02/26/2022 05:57:59 - INFO - codeparrot_training - Step 40463: {'lr': 4.714156281450102e-05, 'samples': 20717568, 'steps': 40463, 'loss/train': 1.2418992519378662}
+02/26/2022 05:58:06 - INFO - codeparrot_training - Step 40464: {'lr': 4.7132000299659774e-05, 'samples': 20718080, 'steps': 40464, 'loss/train': 1.4783622026443481}
+02/26/2022 05:58:09 - INFO - codeparrot_training - Step 40465: {'lr': 4.712243865384067e-05, 'samples': 20718592, 'steps': 40465, 'loss/train': 1.2277157306671143}
+02/26/2022 05:58:13 - INFO - codeparrot_training - Step 40466: {'lr': 4.7112877877084624e-05, 'samples': 20719104, 'steps': 40466, 'loss/train': 0.5284483432769775}
+02/26/2022 05:58:18 - INFO - codeparrot_training - Step 40467: {'lr': 4.7103317969432596e-05, 'samples': 20719616, 'steps': 40467, 'loss/train': 1.923523187637329}
+02/26/2022 05:58:22 - INFO - codeparrot_training - Step 40468: {'lr': 4.709375893092546e-05, 'samples': 20720128, 'steps': 40468, 'loss/train': 1.4221651554107666}
+02/26/2022 05:58:27 - INFO - codeparrot_training - Step 40469: {'lr': 4.708420076160427e-05, 'samples': 20720640, 'steps': 40469, 'loss/train': 1.3970476388931274}
+02/26/2022 05:58:31 - INFO - codeparrot_training - Step 40470: {'lr': 4.707464346150997e-05, 'samples': 20721152, 'steps': 40470, 'loss/train': 1.513027548789978}
+02/26/2022 05:58:36 - INFO - codeparrot_training - Step 40471: {'lr': 4.706508703068343e-05, 'samples': 20721664, 'steps': 40471, 'loss/train': 1.8742296695709229}
+02/26/2022 05:58:40 - INFO - codeparrot_training - Step 40472: {'lr': 4.705553146916558e-05, 'samples': 20722176, 'steps': 40472, 'loss/train': 1.8326905965805054}
+02/26/2022 05:58:46 - INFO - codeparrot_training - Step 40473: {'lr': 4.704597677699743e-05, 'samples': 20722688, 'steps': 40473, 'loss/train': 1.5975911617279053}
+02/26/2022 05:58:50 - INFO - codeparrot_training - Step 40474: {'lr': 4.7036422954219894e-05, 'samples': 20723200, 'steps': 40474, 'loss/train': 0.06833074986934662}
+02/26/2022 05:58:55 - INFO - codeparrot_training - Step 40475: {'lr': 4.702687000087385e-05, 'samples': 20723712, 'steps': 40475, 'loss/train': 1.5726796388626099}
+02/26/2022 05:58:59 - INFO - codeparrot_training - Step 40476: {'lr': 4.70173179170002e-05, 'samples': 20724224, 'steps': 40476, 'loss/train': 1.153774380683899}
+02/26/2022 05:59:04 - INFO - codeparrot_training - Step 40477: {'lr': 4.700776670263995e-05, 'samples': 20724736, 'steps': 40477, 'loss/train': 2.6383652687072754}
+02/26/2022 05:59:08 - INFO - codeparrot_training - Step 40478: {'lr': 4.699821635783399e-05, 'samples': 20725248, 'steps': 40478, 'loss/train': 1.3899964094161987}
+02/26/2022 05:59:13 - INFO - codeparrot_training - Step 40479: {'lr': 4.698866688262321e-05, 'samples': 20725760, 'steps': 40479, 'loss/train': 0.6454765200614929}
+02/26/2022 05:59:17 - INFO - codeparrot_training - Step 40480: {'lr': 4.6979118277048426e-05, 'samples': 20726272, 'steps': 40480, 'loss/train': 1.4977879524230957}
+02/26/2022 05:59:22 - INFO - codeparrot_training - Step 40481: {'lr': 4.6969570541150725e-05, 'samples': 20726784, 'steps': 40481, 'loss/train': 2.570521831512451}
+02/26/2022 05:59:26 - INFO - codeparrot_training - Step 40482: {'lr': 4.69600236749709e-05, 'samples': 20727296, 'steps': 40482, 'loss/train': 0.8650080561637878}
+02/26/2022 05:59:32 - INFO - codeparrot_training - Step 40483: {'lr': 4.695047767854982e-05, 'samples': 20727808, 'steps': 40483, 'loss/train': 1.0418413877487183}
+02/26/2022 05:59:35 - INFO - codeparrot_training - Step 40484: {'lr': 4.694093255192847e-05, 'samples': 20728320, 'steps': 40484, 'loss/train': 1.1293816566467285}
+02/26/2022 05:59:41 - INFO - codeparrot_training - Step 40485: {'lr': 4.693138829514768e-05, 'samples': 20728832, 'steps': 40485, 'loss/train': 1.1596565246582031}
+02/26/2022 05:59:44 - INFO - codeparrot_training - Step 40486: {'lr': 4.6921844908248326e-05, 'samples': 20729344, 'steps': 40486, 'loss/train': 2.1983063220977783}
+02/26/2022 05:59:50 - INFO - codeparrot_training - Step 40487: {'lr': 4.691230239127126e-05, 'samples': 20729856, 'steps': 40487, 'loss/train': 2.0347418785095215}
+02/26/2022 05:59:53 - INFO - codeparrot_training - Step 40488: {'lr': 4.690276074425753e-05, 'samples': 20730368, 'steps': 40488, 'loss/train': 0.9864697456359863}
+02/26/2022 05:59:59 - INFO - codeparrot_training - Step 40489: {'lr': 4.689321996724777e-05, 'samples': 20730880, 'steps': 40489, 'loss/train': 1.7229087352752686}
+02/26/2022 06:00:02 - INFO - codeparrot_training - Step 40490: {'lr': 4.688368006028299e-05, 'samples': 20731392, 'steps': 40490, 'loss/train': 2.7362678050994873}
+02/26/2022 06:00:08 - INFO - codeparrot_training - Step 40491: {'lr': 4.687414102340398e-05, 'samples': 20731904, 'steps': 40491, 'loss/train': 0.0761817991733551}
+02/26/2022 06:00:11 - INFO - codeparrot_training - Step 40492: {'lr': 4.686460285665173e-05, 'samples': 20732416, 'steps': 40492, 'loss/train': 1.157053828239441}
+02/26/2022 06:00:17 - INFO - codeparrot_training - Step 40493: {'lr': 4.6855065560066996e-05, 'samples': 20732928, 'steps': 40493, 'loss/train': 1.4385110139846802}
+02/26/2022 06:00:20 - INFO - codeparrot_training - Step 40494: {'lr': 4.684552913369067e-05, 'samples': 20733440, 'steps': 40494, 'loss/train': 1.5717777013778687}
+02/26/2022 06:00:26 - INFO - codeparrot_training - Step 40495: {'lr': 4.683599357756352e-05, 'samples': 20733952, 'steps': 40495, 'loss/train': 1.8374850749969482}
+02/26/2022 06:00:30 - INFO - codeparrot_training - Step 40496: {'lr': 4.6826458891726513e-05, 'samples': 20734464, 'steps': 40496, 'loss/train': 0.260576069355011}
+02/26/2022 06:00:35 - INFO - codeparrot_training - Step 40497: {'lr': 4.6816925076220454e-05, 'samples': 20734976, 'steps': 40497, 'loss/train': 0.3610433340072632}
+02/26/2022 06:00:39 - INFO - codeparrot_training - Step 40498: {'lr': 4.6807392131086175e-05, 'samples': 20735488, 'steps': 40498, 'loss/train': 6.080682277679443}
+02/26/2022 06:00:44 - INFO - codeparrot_training - Step 40499: {'lr': 4.679786005636444e-05, 'samples': 20736000, 'steps': 40499, 'loss/train': 0.0848667174577713}
+02/26/2022 06:00:48 - INFO - codeparrot_training - Step 40500: {'lr': 4.6788328852096216e-05, 'samples': 20736512, 'steps': 40500, 'loss/train': 1.1825543642044067}
+02/26/2022 06:00:53 - INFO - codeparrot_training - Step 40501: {'lr': 4.677879851832226e-05, 'samples': 20737024, 'steps': 40501, 'loss/train': 2.076835870742798}
+02/26/2022 06:00:57 - INFO - codeparrot_training - Step 40502: {'lr': 4.676926905508339e-05, 'samples': 20737536, 'steps': 40502, 'loss/train': 1.152147889137268}
+02/26/2022 06:01:02 - INFO - codeparrot_training - Step 40503: {'lr': 4.675974046242037e-05, 'samples': 20738048, 'steps': 40503, 'loss/train': 1.346023440361023}
+02/26/2022 06:01:06 - INFO - codeparrot_training - Step 40504: {'lr': 4.675021274037416e-05, 'samples': 20738560, 'steps': 40504, 'loss/train': 1.8701070547103882}
+02/26/2022 06:01:12 - INFO - codeparrot_training - Step 40505: {'lr': 4.674068588898545e-05, 'samples': 20739072, 'steps': 40505, 'loss/train': 0.7971094250679016}
+02/26/2022 06:01:16 - INFO - codeparrot_training - Step 40506: {'lr': 4.6731159908295214e-05, 'samples': 20739584, 'steps': 40506, 'loss/train': 1.0658234357833862}
+02/26/2022 06:01:21 - INFO - codeparrot_training - Step 40507: {'lr': 4.672163479834401e-05, 'samples': 20740096, 'steps': 40507, 'loss/train': 1.5536221265792847}
+02/26/2022 06:01:25 - INFO - codeparrot_training - Step 40508: {'lr': 4.671211055917285e-05, 'samples': 20740608, 'steps': 40508, 'loss/train': 2.74406361579895}
+02/26/2022 06:01:30 - INFO - codeparrot_training - Step 40509: {'lr': 4.67025871908224e-05, 'samples': 20741120, 'steps': 40509, 'loss/train': 1.5952783823013306}
+02/26/2022 06:01:34 - INFO - codeparrot_training - Step 40510: {'lr': 4.669306469333362e-05, 'samples': 20741632, 'steps': 40510, 'loss/train': 2.4164438247680664}
+02/26/2022 06:01:39 - INFO - codeparrot_training - Step 40511: {'lr': 4.6683543066747076e-05, 'samples': 20742144, 'steps': 40511, 'loss/train': 1.5712264776229858}
+02/26/2022 06:01:43 - INFO - codeparrot_training - Step 40512: {'lr': 4.667402231110374e-05, 'samples': 20742656, 'steps': 40512, 'loss/train': 1.611841082572937}
+02/26/2022 06:01:48 - INFO - codeparrot_training - Step 40513: {'lr': 4.6664502426444266e-05, 'samples': 20743168, 'steps': 40513, 'loss/train': 2.0886714458465576}
+02/26/2022 06:01:52 - INFO - codeparrot_training - Step 40514: {'lr': 4.665498341280963e-05, 'samples': 20743680, 'steps': 40514, 'loss/train': 2.2555325031280518}
+02/26/2022 06:01:57 - INFO - codeparrot_training - Step 40515: {'lr': 4.664546527024033e-05, 'samples': 20744192, 'steps': 40515, 'loss/train': 0.9298934936523438}
+02/26/2022 06:02:01 - INFO - codeparrot_training - Step 40516: {'lr': 4.663594799877735e-05, 'samples': 20744704, 'steps': 40516, 'loss/train': 1.6815675497055054}
+02/26/2022 06:02:07 - INFO - codeparrot_training - Step 40517: {'lr': 4.662643159846133e-05, 'samples': 20745216, 'steps': 40517, 'loss/train': 2.065317153930664}
+02/26/2022 06:02:11 - INFO - codeparrot_training - Step 40518: {'lr': 4.6616916069333185e-05, 'samples': 20745728, 'steps': 40518, 'loss/train': 3.2296125888824463}
+02/26/2022 06:02:16 - INFO - codeparrot_training - Step 40519: {'lr': 4.6607401411433575e-05, 'samples': 20746240, 'steps': 40519, 'loss/train': 2.0922257900238037}
+02/26/2022 06:02:20 - INFO - codeparrot_training - Step 40520: {'lr': 4.659788762480327e-05, 'samples': 20746752, 'steps': 40520, 'loss/train': 1.606555461883545}
+02/26/2022 06:02:25 - INFO - codeparrot_training - Step 40521: {'lr': 4.6588374709483014e-05, 'samples': 20747264, 'steps': 40521, 'loss/train': 2.468932867050171}
+02/26/2022 06:02:29 - INFO - codeparrot_training - Step 40522: {'lr': 4.6578862665513534e-05, 'samples': 20747776, 'steps': 40522, 'loss/train': 2.378203868865967}
+02/26/2022 06:02:34 - INFO - codeparrot_training - Step 40523: {'lr': 4.6569351492935676e-05, 'samples': 20748288, 'steps': 40523, 'loss/train': 1.4238327741622925}
+02/26/2022 06:02:38 - INFO - codeparrot_training - Step 40524: {'lr': 4.6559841191790106e-05, 'samples': 20748800, 'steps': 40524, 'loss/train': 2.251905918121338}
+02/26/2022 06:02:43 - INFO - codeparrot_training - Step 40525: {'lr': 4.655033176211757e-05, 'samples': 20749312, 'steps': 40525, 'loss/train': 1.4019033908843994}
+02/26/2022 06:02:47 - INFO - codeparrot_training - Step 40526: {'lr': 4.654082320395875e-05, 'samples': 20749824, 'steps': 40526, 'loss/train': 1.7638624906539917}
+02/26/2022 06:02:52 - INFO - codeparrot_training - Step 40527: {'lr': 4.653131551735451e-05, 'samples': 20750336, 'steps': 40527, 'loss/train': 1.9404500722885132}
+02/26/2022 06:02:56 - INFO - codeparrot_training - Step 40528: {'lr': 4.652180870234551e-05, 'samples': 20750848, 'steps': 40528, 'loss/train': 1.8244398832321167}
+02/26/2022 06:03:02 - INFO - codeparrot_training - Step 40529: {'lr': 4.6512302758972445e-05, 'samples': 20751360, 'steps': 40529, 'loss/train': 1.048850178718567}
+02/26/2022 06:03:08 - INFO - codeparrot_training - Step 40530: {'lr': 4.6502797687276025e-05, 'samples': 20751872, 'steps': 40530, 'loss/train': 1.8237043619155884}
+02/26/2022 06:03:11 - INFO - codeparrot_training - Step 40531: {'lr': 4.649329348729706e-05, 'samples': 20752384, 'steps': 40531, 'loss/train': 0.8328178524971008}
+02/26/2022 06:03:15 - INFO - codeparrot_training - Step 40532: {'lr': 4.648379015907619e-05, 'samples': 20752896, 'steps': 40532, 'loss/train': 0.9888048768043518}
+02/26/2022 06:03:20 - INFO - codeparrot_training - Step 40533: {'lr': 4.647428770265416e-05, 'samples': 20753408, 'steps': 40533, 'loss/train': 0.6856012940406799}
+02/26/2022 06:03:26 - INFO - codeparrot_training - Step 40534: {'lr': 4.646478611807156e-05, 'samples': 20753920, 'steps': 40534, 'loss/train': 2.7985596656799316}
+02/26/2022 06:03:29 - INFO - codeparrot_training - Step 40535: {'lr': 4.645528540536928e-05, 'samples': 20754432, 'steps': 40535, 'loss/train': 1.9000080823898315}
+02/26/2022 06:03:35 - INFO - codeparrot_training - Step 40536: {'lr': 4.644578556458792e-05, 'samples': 20754944, 'steps': 40536, 'loss/train': 2.3702783584594727}
+02/26/2022 06:03:38 - INFO - codeparrot_training - Step 40537: {'lr': 4.643628659576818e-05, 'samples': 20755456, 'steps': 40537, 'loss/train': 1.701398491859436}
+02/26/2022 06:03:44 - INFO - codeparrot_training - Step 40538: {'lr': 4.642678849895068e-05, 'samples': 20755968, 'steps': 40538, 'loss/train': 2.4501144886016846}
+02/26/2022 06:03:47 - INFO - codeparrot_training - Step 40539: {'lr': 4.641729127417624e-05, 'samples': 20756480, 'steps': 40539, 'loss/train': 0.7281582951545715}
+02/26/2022 06:03:53 - INFO - codeparrot_training - Step 40540: {'lr': 4.640779492148547e-05, 'samples': 20756992, 'steps': 40540, 'loss/train': 0.8816997408866882}
+02/26/2022 06:03:57 - INFO - codeparrot_training - Step 40541: {'lr': 4.639829944091905e-05, 'samples': 20757504, 'steps': 40541, 'loss/train': 2.528623580932617}
+02/26/2022 06:04:02 - INFO - codeparrot_training - Step 40542: {'lr': 4.6388804832517645e-05, 'samples': 20758016, 'steps': 40542, 'loss/train': 1.9930883646011353}
+02/26/2022 06:04:06 - INFO - codeparrot_training - Step 40543: {'lr': 4.6379311096321986e-05, 'samples': 20758528, 'steps': 40543, 'loss/train': 1.8037618398666382}
+02/26/2022 06:04:11 - INFO - codeparrot_training - Step 40544: {'lr': 4.636981823237263e-05, 'samples': 20759040, 'steps': 40544, 'loss/train': 1.8401269912719727}
+02/26/2022 06:04:15 - INFO - codeparrot_training - Step 40545: {'lr': 4.6360326240710435e-05, 'samples': 20759552, 'steps': 40545, 'loss/train': 1.8866084814071655}
+02/26/2022 06:04:20 - INFO - codeparrot_training - Step 40546: {'lr': 4.6350835121375836e-05, 'samples': 20760064, 'steps': 40546, 'loss/train': 1.5532060861587524}
+02/26/2022 06:04:24 - INFO - codeparrot_training - Step 40547: {'lr': 4.634134487440964e-05, 'samples': 20760576, 'steps': 40547, 'loss/train': 1.2604817152023315}
+02/26/2022 06:04:29 - INFO - codeparrot_training - Step 40548: {'lr': 4.633185549985239e-05, 'samples': 20761088, 'steps': 40548, 'loss/train': 1.6716194152832031}
+02/26/2022 06:04:33 - INFO - codeparrot_training - Step 40549: {'lr': 4.632236699774492e-05, 'samples': 20761600, 'steps': 40549, 'loss/train': 2.2486982345581055}
+02/26/2022 06:04:38 - INFO - codeparrot_training - Step 40550: {'lr': 4.6312879368127645e-05, 'samples': 20762112, 'steps': 40550, 'loss/train': 0.9296627044677734}
+02/26/2022 06:04:42 - INFO - codeparrot_training - Step 40551: {'lr': 4.630339261104138e-05, 'samples': 20762624, 'steps': 40551, 'loss/train': 1.1658196449279785}
+02/26/2022 06:04:48 - INFO - codeparrot_training - Step 40552: {'lr': 4.629390672652661e-05, 'samples': 20763136, 'steps': 40552, 'loss/train': 0.9665114283561707}
+02/26/2022 06:04:51 - INFO - codeparrot_training - Step 40553: {'lr': 4.628442171462416e-05, 'samples': 20763648, 'steps': 40553, 'loss/train': 0.6147657036781311}
+02/26/2022 06:04:57 - INFO - codeparrot_training - Step 40554: {'lr': 4.627493757537454e-05, 'samples': 20764160, 'steps': 40554, 'loss/train': 1.8051081895828247}
+02/26/2022 06:05:00 - INFO - codeparrot_training - Step 40555: {'lr': 4.62654543088184e-05, 'samples': 20764672, 'steps': 40555, 'loss/train': 1.947871446609497}
+02/26/2022 06:05:06 - INFO - codeparrot_training - Step 40556: {'lr': 4.625597191499631e-05, 'samples': 20765184, 'steps': 40556, 'loss/train': 1.5963685512542725}
+02/26/2022 06:05:09 - INFO - codeparrot_training - Step 40557: {'lr': 4.6246490393948984e-05, 'samples': 20765696, 'steps': 40557, 'loss/train': 1.5938471555709839}
+02/26/2022 06:05:15 - INFO - codeparrot_training - Step 40558: {'lr': 4.6237009745717015e-05, 'samples': 20766208, 'steps': 40558, 'loss/train': 5.4820146560668945}
+02/26/2022 06:05:18 - INFO - codeparrot_training - Step 40559: {'lr': 4.6227529970340965e-05, 'samples': 20766720, 'steps': 40559, 'loss/train': 2.2246623039245605}
+02/26/2022 06:05:24 - INFO - codeparrot_training - Step 40560: {'lr': 4.621805106786142e-05, 'samples': 20767232, 'steps': 40560, 'loss/train': 1.1664386987686157}
+02/26/2022 06:05:27 - INFO - codeparrot_training - Step 40561: {'lr': 4.62085730383191e-05, 'samples': 20767744, 'steps': 40561, 'loss/train': 0.8692867159843445}
+02/26/2022 06:05:34 - INFO - codeparrot_training - Step 40562: {'lr': 4.619909588175455e-05, 'samples': 20768256, 'steps': 40562, 'loss/train': 1.664820909500122}
+02/26/2022 06:05:37 - INFO - codeparrot_training - Step 40563: {'lr': 4.618961959820836e-05, 'samples': 20768768, 'steps': 40563, 'loss/train': 1.2737410068511963}
+02/26/2022 06:05:42 - INFO - codeparrot_training - Step 40564: {'lr': 4.618014418772104e-05, 'samples': 20769280, 'steps': 40564, 'loss/train': 1.7915550470352173}
+02/26/2022 06:05:46 - INFO - codeparrot_training - Step 40565: {'lr': 4.617066965033334e-05, 'samples': 20769792, 'steps': 40565, 'loss/train': 1.1754978895187378}
+02/26/2022 06:05:52 - INFO - codeparrot_training - Step 40566: {'lr': 4.6161195986085787e-05, 'samples': 20770304, 'steps': 40566, 'loss/train': 1.2529526948928833}
+02/26/2022 06:05:55 - INFO - codeparrot_training - Step 40567: {'lr': 4.6151723195018925e-05, 'samples': 20770816, 'steps': 40567, 'loss/train': 2.1452362537384033}
+02/26/2022 06:06:01 - INFO - codeparrot_training - Step 40568: {'lr': 4.614225127717334e-05, 'samples': 20771328, 'steps': 40568, 'loss/train': 1.314723253250122}
+02/26/2022 06:06:04 - INFO - codeparrot_training - Step 40569: {'lr': 4.6132780232589574e-05, 'samples': 20771840, 'steps': 40569, 'loss/train': 1.7965604066848755}
+02/26/2022 06:06:09 - INFO - codeparrot_training - Step 40570: {'lr': 4.612331006130829e-05, 'samples': 20772352, 'steps': 40570, 'loss/train': 0.4550262987613678}
+02/26/2022 06:06:13 - INFO - codeparrot_training - Step 40571: {'lr': 4.611384076337003e-05, 'samples': 20772864, 'steps': 40571, 'loss/train': 2.019219398498535}
+02/26/2022 06:06:18 - INFO - codeparrot_training - Step 40572: {'lr': 4.6104372338815315e-05, 'samples': 20773376, 'steps': 40572, 'loss/train': 2.0042831897735596}
+02/26/2022 06:06:22 - INFO - codeparrot_training - Step 40573: {'lr': 4.609490478768466e-05, 'samples': 20773888, 'steps': 40573, 'loss/train': 1.4293339252471924}
+02/26/2022 06:06:27 - INFO - codeparrot_training - Step 40574: {'lr': 4.608543811001875e-05, 'samples': 20774400, 'steps': 40574, 'loss/train': 1.8459911346435547}
+02/26/2022 06:06:31 - INFO - codeparrot_training - Step 40575: {'lr': 4.607597230585808e-05, 'samples': 20774912, 'steps': 40575, 'loss/train': 1.714013695716858}
+02/26/2022 06:06:37 - INFO - codeparrot_training - Step 40576: {'lr': 4.606650737524321e-05, 'samples': 20775424, 'steps': 40576, 'loss/train': 1.852400302886963}
+02/26/2022 06:06:41 - INFO - codeparrot_training - Step 40577: {'lr': 4.6057043318214585e-05, 'samples': 20775936, 'steps': 40577, 'loss/train': 1.2687244415283203}
+02/26/2022 06:06:46 - INFO - codeparrot_training - Step 40578: {'lr': 4.604758013481289e-05, 'samples': 20776448, 'steps': 40578, 'loss/train': 0.03738047555088997}
+02/26/2022 06:06:50 - INFO - codeparrot_training - Step 40579: {'lr': 4.6038117825078526e-05, 'samples': 20776960, 'steps': 40579, 'loss/train': 2.158421754837036}
+02/26/2022 06:06:55 - INFO - codeparrot_training - Step 40580: {'lr': 4.6028656389052236e-05, 'samples': 20777472, 'steps': 40580, 'loss/train': 2.041349172592163}
+02/26/2022 06:06:59 - INFO - codeparrot_training - Step 40581: {'lr': 4.60191958267743e-05, 'samples': 20777984, 'steps': 40581, 'loss/train': 1.8198038339614868}
+02/26/2022 06:07:04 - INFO - codeparrot_training - Step 40582: {'lr': 4.6009736138285406e-05, 'samples': 20778496, 'steps': 40582, 'loss/train': 3.0360329151153564}
+02/26/2022 06:07:08 - INFO - codeparrot_training - Step 40583: {'lr': 4.6000277323625964e-05, 'samples': 20779008, 'steps': 40583, 'loss/train': 1.5629966259002686}
+02/26/2022 06:07:13 - INFO - codeparrot_training - Step 40584: {'lr': 4.59908193828367e-05, 'samples': 20779520, 'steps': 40584, 'loss/train': 1.730132818222046}
+02/26/2022 06:07:17 - INFO - codeparrot_training - Step 40585: {'lr': 4.598136231595784e-05, 'samples': 20780032, 'steps': 40585, 'loss/train': 1.7915922403335571}
+02/26/2022 06:07:22 - INFO - codeparrot_training - Step 40586: {'lr': 4.5971906123030095e-05, 'samples': 20780544, 'steps': 40586, 'loss/train': 2.3261685371398926}
+02/26/2022 06:07:26 - INFO - codeparrot_training - Step 40587: {'lr': 4.596245080409386e-05, 'samples': 20781056, 'steps': 40587, 'loss/train': 0.7651938199996948}
+02/26/2022 06:07:32 - INFO - codeparrot_training - Step 40588: {'lr': 4.595299635918984e-05, 'samples': 20781568, 'steps': 40588, 'loss/train': 1.8871077299118042}
+02/26/2022 06:07:35 - INFO - codeparrot_training - Step 40589: {'lr': 4.594354278835827e-05, 'samples': 20782080, 'steps': 40589, 'loss/train': 1.2763116359710693}
+02/26/2022 06:07:41 - INFO - codeparrot_training - Step 40590: {'lr': 4.593409009163982e-05, 'samples': 20782592, 'steps': 40590, 'loss/train': 1.8814716339111328}
+02/26/2022 06:07:45 - INFO - codeparrot_training - Step 40591: {'lr': 4.5924638269074864e-05, 'samples': 20783104, 'steps': 40591, 'loss/train': 1.482607364654541}
+02/26/2022 06:07:50 - INFO - codeparrot_training - Step 40592: {'lr': 4.5915187320704016e-05, 'samples': 20783616, 'steps': 40592, 'loss/train': 1.1465833187103271}
+02/26/2022 06:07:54 - INFO - codeparrot_training - Step 40593: {'lr': 4.590573724656771e-05, 'samples': 20784128, 'steps': 40593, 'loss/train': 1.9058486223220825}
+02/26/2022 06:07:59 - INFO - codeparrot_training - Step 40594: {'lr': 4.5896288046706396e-05, 'samples': 20784640, 'steps': 40594, 'loss/train': 1.572718620300293}
+02/26/2022 06:08:03 - INFO - codeparrot_training - Step 40595: {'lr': 4.5886839721160533e-05, 'samples': 20785152, 'steps': 40595, 'loss/train': 1.552043080329895}
+02/26/2022 06:08:08 - INFO - codeparrot_training - Step 40596: {'lr': 4.5877392269970705e-05, 'samples': 20785664, 'steps': 40596, 'loss/train': 0.4863733649253845}
+02/26/2022 06:08:12 - INFO - codeparrot_training - Step 40597: {'lr': 4.5867945693177285e-05, 'samples': 20786176, 'steps': 40597, 'loss/train': 2.2209222316741943}
+02/26/2022 06:08:18 - INFO - codeparrot_training - Step 40598: {'lr': 4.585849999082079e-05, 'samples': 20786688, 'steps': 40598, 'loss/train': 2.3790736198425293}
+02/26/2022 06:08:21 - INFO - codeparrot_training - Step 40599: {'lr': 4.5849055162941605e-05, 'samples': 20787200, 'steps': 40599, 'loss/train': 0.9918031692504883}
+02/26/2022 06:08:27 - INFO - codeparrot_training - Step 40600: {'lr': 4.5839611209580275e-05, 'samples': 20787712, 'steps': 40600, 'loss/train': 2.91489315032959}
+02/26/2022 06:08:30 - INFO - codeparrot_training - Step 40601: {'lr': 4.5830168130777225e-05, 'samples': 20788224, 'steps': 40601, 'loss/train': 0.4606277048587799}
+02/26/2022 06:08:36 - INFO - codeparrot_training - Step 40602: {'lr': 4.5820725926572934e-05, 'samples': 20788736, 'steps': 40602, 'loss/train': 1.4059597253799438}
+02/26/2022 06:08:40 - INFO - codeparrot_training - Step 40603: {'lr': 4.581128459700773e-05, 'samples': 20789248, 'steps': 40603, 'loss/train': 2.222914695739746}
+02/26/2022 06:08:43 - INFO - codeparrot_training - Step 40604: {'lr': 4.5801844142122214e-05, 'samples': 20789760, 'steps': 40604, 'loss/train': 3.2750346660614014}
+02/26/2022 06:08:49 - INFO - codeparrot_training - Step 40605: {'lr': 4.579240456195677e-05, 'samples': 20790272, 'steps': 40605, 'loss/train': 0.3980376124382019}
+02/26/2022 06:08:54 - INFO - codeparrot_training - Step 40606: {'lr': 4.578296585655181e-05, 'samples': 20790784, 'steps': 40606, 'loss/train': 1.935340166091919}
+02/26/2022 06:08:58 - INFO - codeparrot_training - Step 40607: {'lr': 4.57735280259477e-05, 'samples': 20791296, 'steps': 40607, 'loss/train': 2.185854911804199}
+02/26/2022 06:09:03 - INFO - codeparrot_training - Step 40608: {'lr': 4.5764091070185035e-05, 'samples': 20791808, 'steps': 40608, 'loss/train': 1.781665325164795}
+02/26/2022 06:09:07 - INFO - codeparrot_training - Step 40609: {'lr': 4.5754654989304143e-05, 'samples': 20792320, 'steps': 40609, 'loss/train': 1.6466495990753174}
+02/26/2022 06:09:12 - INFO - codeparrot_training - Step 40610: {'lr': 4.574521978334545e-05, 'samples': 20792832, 'steps': 40610, 'loss/train': 2.1167585849761963}
+02/26/2022 06:09:15 - INFO - codeparrot_training - Step 40611: {'lr': 4.573578545234933e-05, 'samples': 20793344, 'steps': 40611, 'loss/train': 0.7844367027282715}
+02/26/2022 06:09:21 - INFO - codeparrot_training - Step 40612: {'lr': 4.572635199635627e-05, 'samples': 20793856, 'steps': 40612, 'loss/train': 1.0828701257705688}
+02/26/2022 06:09:24 - INFO - codeparrot_training - Step 40613: {'lr': 4.5716919415406677e-05, 'samples': 20794368, 'steps': 40613, 'loss/train': 1.0152088403701782}
+02/26/2022 06:09:31 - INFO - codeparrot_training - Step 40614: {'lr': 4.570748770954089e-05, 'samples': 20794880, 'steps': 40614, 'loss/train': 2.0802741050720215}
+02/26/2022 06:09:35 - INFO - codeparrot_training - Step 40615: {'lr': 4.569805687879946e-05, 'samples': 20795392, 'steps': 40615, 'loss/train': 1.5632034540176392}
+02/26/2022 06:09:40 - INFO - codeparrot_training - Step 40616: {'lr': 4.5688626923222564e-05, 'samples': 20795904, 'steps': 40616, 'loss/train': 1.374606728553772}
+02/26/2022 06:09:44 - INFO - codeparrot_training - Step 40617: {'lr': 4.5679197842850766e-05, 'samples': 20796416, 'steps': 40617, 'loss/train': 1.8106578588485718}
+02/26/2022 06:09:49 - INFO - codeparrot_training - Step 40618: {'lr': 4.5669769637724346e-05, 'samples': 20796928, 'steps': 40618, 'loss/train': 1.1967626810073853}
+02/26/2022 06:09:53 - INFO - codeparrot_training - Step 40619: {'lr': 4.566034230788388e-05, 'samples': 20797440, 'steps': 40619, 'loss/train': 2.0095179080963135}
+02/26/2022 06:09:58 - INFO - codeparrot_training - Step 40620: {'lr': 4.56509158533695e-05, 'samples': 20797952, 'steps': 40620, 'loss/train': 0.8096858859062195}
+02/26/2022 06:10:02 - INFO - codeparrot_training - Step 40621: {'lr': 4.5641490274221766e-05, 'samples': 20798464, 'steps': 40621, 'loss/train': 2.2266359329223633}
+02/26/2022 06:10:07 - INFO - codeparrot_training - Step 40622: {'lr': 4.5632065570480916e-05, 'samples': 20798976, 'steps': 40622, 'loss/train': 1.4248467683792114}
+02/26/2022 06:10:11 - INFO - codeparrot_training - Step 40623: {'lr': 4.562264174218755e-05, 'samples': 20799488, 'steps': 40623, 'loss/train': 2.23813796043396}
+02/26/2022 06:10:17 - INFO - codeparrot_training - Step 40624: {'lr': 4.5613218789381765e-05, 'samples': 20800000, 'steps': 40624, 'loss/train': 2.587691307067871}
+02/26/2022 06:10:21 - INFO - codeparrot_training - Step 40625: {'lr': 4.56037967121041e-05, 'samples': 20800512, 'steps': 40625, 'loss/train': 1.147760272026062}
+02/26/2022 06:10:26 - INFO - codeparrot_training - Step 40626: {'lr': 4.559437551039483e-05, 'samples': 20801024, 'steps': 40626, 'loss/train': 1.448170781135559}
+02/26/2022 06:10:30 - INFO - codeparrot_training - Step 40627: {'lr': 4.5584955184294395e-05, 'samples': 20801536, 'steps': 40627, 'loss/train': 2.316469669342041}
+02/26/2022 06:10:36 - INFO - codeparrot_training - Step 40628: {'lr': 4.55755357338431e-05, 'samples': 20802048, 'steps': 40628, 'loss/train': 0.6476761698722839}
+02/26/2022 06:10:39 - INFO - codeparrot_training - Step 40629: {'lr': 4.5566117159081296e-05, 'samples': 20802560, 'steps': 40629, 'loss/train': 2.0663208961486816}
+02/26/2022 06:10:44 - INFO - codeparrot_training - Step 40630: {'lr': 4.5556699460049283e-05, 'samples': 20803072, 'steps': 40630, 'loss/train': 1.1277186870574951}
+02/26/2022 06:10:48 - INFO - codeparrot_training - Step 40631: {'lr': 4.5547282636787505e-05, 'samples': 20803584, 'steps': 40631, 'loss/train': 2.1589252948760986}
+02/26/2022 06:10:53 - INFO - codeparrot_training - Step 40632: {'lr': 4.553786668933624e-05, 'samples': 20804096, 'steps': 40632, 'loss/train': 2.241276264190674}
+02/26/2022 06:10:57 - INFO - codeparrot_training - Step 40633: {'lr': 4.552845161773583e-05, 'samples': 20804608, 'steps': 40633, 'loss/train': 1.279417872428894}
+02/26/2022 06:11:03 - INFO - codeparrot_training - Step 40634: {'lr': 4.551903742202657e-05, 'samples': 20805120, 'steps': 40634, 'loss/train': 2.058643341064453}
+02/26/2022 06:11:06 - INFO - codeparrot_training - Step 40635: {'lr': 4.5509624102248845e-05, 'samples': 20805632, 'steps': 40635, 'loss/train': 0.7822037935256958}
+02/26/2022 06:11:12 - INFO - codeparrot_training - Step 40636: {'lr': 4.550021165844298e-05, 'samples': 20806144, 'steps': 40636, 'loss/train': 1.352022409439087}
+02/26/2022 06:11:15 - INFO - codeparrot_training - Step 40637: {'lr': 4.5490800090649264e-05, 'samples': 20806656, 'steps': 40637, 'loss/train': 1.8850128650665283}
+02/26/2022 06:11:21 - INFO - codeparrot_training - Step 40638: {'lr': 4.5481389398907976e-05, 'samples': 20807168, 'steps': 40638, 'loss/train': 2.1383614540100098}
+02/26/2022 06:11:24 - INFO - codeparrot_training - Step 40639: {'lr': 4.547197958325952e-05, 'samples': 20807680, 'steps': 40639, 'loss/train': 1.354177713394165}
+02/26/2022 06:11:30 - INFO - codeparrot_training - Step 40640: {'lr': 4.54625706437441e-05, 'samples': 20808192, 'steps': 40640, 'loss/train': 2.9157562255859375}
+02/26/2022 06:11:33 - INFO - codeparrot_training - Step 40641: {'lr': 4.545316258040222e-05, 'samples': 20808704, 'steps': 40641, 'loss/train': 1.508209228515625}
+02/26/2022 06:11:39 - INFO - codeparrot_training - Step 40642: {'lr': 4.544375539327389e-05, 'samples': 20809216, 'steps': 40642, 'loss/train': 1.7163926362991333}
+02/26/2022 06:11:42 - INFO - codeparrot_training - Step 40643: {'lr': 4.543434908239963e-05, 'samples': 20809728, 'steps': 40643, 'loss/train': 1.3316614627838135}
+02/26/2022 06:11:49 - INFO - codeparrot_training - Step 40644: {'lr': 4.54249436478196e-05, 'samples': 20810240, 'steps': 40644, 'loss/train': 1.2661103010177612}
+02/26/2022 06:11:52 - INFO - codeparrot_training - Step 40645: {'lr': 4.5415539089574296e-05, 'samples': 20810752, 'steps': 40645, 'loss/train': 2.5089473724365234}
+02/26/2022 06:11:58 - INFO - codeparrot_training - Step 40646: {'lr': 4.540613540770372e-05, 'samples': 20811264, 'steps': 40646, 'loss/train': 1.5070173740386963}
+02/26/2022 06:12:01 - INFO - codeparrot_training - Step 40647: {'lr': 4.539673260224833e-05, 'samples': 20811776, 'steps': 40647, 'loss/train': 1.2447699308395386}
+02/26/2022 06:12:07 - INFO - codeparrot_training - Step 40648: {'lr': 4.5387330673248404e-05, 'samples': 20812288, 'steps': 40648, 'loss/train': 1.0220937728881836}
+02/26/2022 06:12:10 - INFO - codeparrot_training - Step 40649: {'lr': 4.537792962074416e-05, 'samples': 20812800, 'steps': 40649, 'loss/train': 0.5093560218811035}
+02/26/2022 06:12:16 - INFO - codeparrot_training - Step 40650: {'lr': 4.5368529444775846e-05, 'samples': 20813312, 'steps': 40650, 'loss/train': 1.0306366682052612}
+02/26/2022 06:12:19 - INFO - codeparrot_training - Step 40651: {'lr': 4.535913014538384e-05, 'samples': 20813824, 'steps': 40651, 'loss/train': 1.6842639446258545}
+02/26/2022 06:12:25 - INFO - codeparrot_training - Step 40652: {'lr': 4.534973172260831e-05, 'samples': 20814336, 'steps': 40652, 'loss/train': 1.7435857057571411}
+02/26/2022 06:12:29 - INFO - codeparrot_training - Step 40653: {'lr': 4.53403341764895e-05, 'samples': 20814848, 'steps': 40653, 'loss/train': 2.4779903888702393}
+02/26/2022 06:12:34 - INFO - codeparrot_training - Step 40654: {'lr': 4.5330937507067795e-05, 'samples': 20815360, 'steps': 40654, 'loss/train': 1.433372974395752}
+02/26/2022 06:12:38 - INFO - codeparrot_training - Step 40655: {'lr': 4.532154171438333e-05, 'samples': 20815872, 'steps': 40655, 'loss/train': 1.8924967050552368}
+02/26/2022 06:12:43 - INFO - codeparrot_training - Step 40656: {'lr': 4.531214679847639e-05, 'samples': 20816384, 'steps': 40656, 'loss/train': 0.5280120968818665}
+02/26/2022 06:12:47 - INFO - codeparrot_training - Step 40657: {'lr': 4.5302752759387166e-05, 'samples': 20816896, 'steps': 40657, 'loss/train': 2.0612668991088867}
+02/26/2022 06:12:52 - INFO - codeparrot_training - Step 40658: {'lr': 4.529335959715602e-05, 'samples': 20817408, 'steps': 40658, 'loss/train': 1.9786375761032104}
+02/26/2022 06:12:56 - INFO - codeparrot_training - Step 40659: {'lr': 4.528396731182313e-05, 'samples': 20817920, 'steps': 40659, 'loss/train': 3.2058160305023193}
+02/26/2022 06:13:03 - INFO - codeparrot_training - Step 40660: {'lr': 4.5274575903428706e-05, 'samples': 20818432, 'steps': 40660, 'loss/train': 2.008648157119751}
+02/26/2022 06:13:06 - INFO - codeparrot_training - Step 40661: {'lr': 4.526518537201291e-05, 'samples': 20818944, 'steps': 40661, 'loss/train': 0.45992931723594666}
+02/26/2022 06:13:12 - INFO - codeparrot_training - Step 40662: {'lr': 4.525579571761612e-05, 'samples': 20819456, 'steps': 40662, 'loss/train': 1.9788191318511963}
+02/26/2022 06:13:15 - INFO - codeparrot_training - Step 40663: {'lr': 4.524640694027851e-05, 'samples': 20819968, 'steps': 40663, 'loss/train': 1.5528110265731812}
+02/26/2022 06:13:21 - INFO - codeparrot_training - Step 40664: {'lr': 4.523701904004027e-05, 'samples': 20820480, 'steps': 40664, 'loss/train': 1.4869941473007202}
+02/26/2022 06:13:24 - INFO - codeparrot_training - Step 40665: {'lr': 4.522763201694152e-05, 'samples': 20820992, 'steps': 40665, 'loss/train': 1.5976141691207886}
+02/26/2022 06:13:29 - INFO - codeparrot_training - Step 40666: {'lr': 4.521824587102269e-05, 'samples': 20821504, 'steps': 40666, 'loss/train': 1.2229297161102295}
+02/26/2022 06:13:33 - INFO - codeparrot_training - Step 40667: {'lr': 4.520886060232382e-05, 'samples': 20822016, 'steps': 40667, 'loss/train': 1.2224538326263428}
+02/26/2022 06:13:38 - INFO - codeparrot_training - Step 40668: {'lr': 4.519947621088519e-05, 'samples': 20822528, 'steps': 40668, 'loss/train': 1.6681156158447266}
+02/26/2022 06:13:42 - INFO - codeparrot_training - Step 40669: {'lr': 4.51900926967469e-05, 'samples': 20823040, 'steps': 40669, 'loss/train': 1.7371987104415894}
+02/26/2022 06:13:48 - INFO - codeparrot_training - Step 40670: {'lr': 4.518071005994931e-05, 'samples': 20823552, 'steps': 40670, 'loss/train': 2.11722469329834}
+02/26/2022 06:13:52 - INFO - codeparrot_training - Step 40671: {'lr': 4.517132830053247e-05, 'samples': 20824064, 'steps': 40671, 'loss/train': 2.6994738578796387}
+02/26/2022 06:13:57 - INFO - codeparrot_training - Step 40672: {'lr': 4.516194741853666e-05, 'samples': 20824576, 'steps': 40672, 'loss/train': 2.239804267883301}
+02/26/2022 06:14:01 - INFO - codeparrot_training - Step 40673: {'lr': 4.515256741400195e-05, 'samples': 20825088, 'steps': 40673, 'loss/train': 1.952246904373169}
+02/26/2022 06:14:06 - INFO - codeparrot_training - Step 40674: {'lr': 4.514318828696864e-05, 'samples': 20825600, 'steps': 40674, 'loss/train': 1.093126893043518}
+02/26/2022 06:14:10 - INFO - codeparrot_training - Step 40675: {'lr': 4.513381003747682e-05, 'samples': 20826112, 'steps': 40675, 'loss/train': 2.2619130611419678}
+02/26/2022 06:14:15 - INFO - codeparrot_training - Step 40676: {'lr': 4.5124432665566816e-05, 'samples': 20826624, 'steps': 40676, 'loss/train': 1.5883575677871704}
+02/26/2022 06:14:19 - INFO - codeparrot_training - Step 40677: {'lr': 4.511505617127859e-05, 'samples': 20827136, 'steps': 40677, 'loss/train': 2.1372218132019043}
+02/26/2022 06:14:24 - INFO - codeparrot_training - Step 40678: {'lr': 4.5105680554652444e-05, 'samples': 20827648, 'steps': 40678, 'loss/train': 0.2200467884540558}
+02/26/2022 06:14:28 - INFO - codeparrot_training - Step 40679: {'lr': 4.509630581572844e-05, 'samples': 20828160, 'steps': 40679, 'loss/train': 1.5311400890350342}
+02/26/2022 06:14:34 - INFO - codeparrot_training - Step 40680: {'lr': 4.5086931954546936e-05, 'samples': 20828672, 'steps': 40680, 'loss/train': 1.8666329383850098}
+02/26/2022 06:14:38 - INFO - codeparrot_training - Step 40681: {'lr': 4.507755897114779e-05, 'samples': 20829184, 'steps': 40681, 'loss/train': 0.5599453449249268}
+02/26/2022 06:14:43 - INFO - codeparrot_training - Step 40682: {'lr': 4.50681868655714e-05, 'samples': 20829696, 'steps': 40682, 'loss/train': 1.1868550777435303}
+02/26/2022 06:14:47 - INFO - codeparrot_training - Step 40683: {'lr': 4.505881563785777e-05, 'samples': 20830208, 'steps': 40683, 'loss/train': 2.5260062217712402}
+02/26/2022 06:14:52 - INFO - codeparrot_training - Step 40684: {'lr': 4.504944528804722e-05, 'samples': 20830720, 'steps': 40684, 'loss/train': 1.2936135530471802}
+02/26/2022 06:14:56 - INFO - codeparrot_training - Step 40685: {'lr': 4.504007581617961e-05, 'samples': 20831232, 'steps': 40685, 'loss/train': 1.6963977813720703}
+02/26/2022 06:15:01 - INFO - codeparrot_training - Step 40686: {'lr': 4.503070722229533e-05, 'samples': 20831744, 'steps': 40686, 'loss/train': 0.7575234770774841}
+02/26/2022 06:15:05 - INFO - codeparrot_training - Step 40687: {'lr': 4.5021339506434333e-05, 'samples': 20832256, 'steps': 40687, 'loss/train': 2.60663104057312}
+02/26/2022 06:15:10 - INFO - codeparrot_training - Step 40688: {'lr': 4.501197266863691e-05, 'samples': 20832768, 'steps': 40688, 'loss/train': 0.827700674533844}
+02/26/2022 06:15:14 - INFO - codeparrot_training - Step 40689: {'lr': 4.50026067089431e-05, 'samples': 20833280, 'steps': 40689, 'loss/train': 1.858444333076477}
+02/26/2022 06:15:20 - INFO - codeparrot_training - Step 40690: {'lr': 4.4993241627393015e-05, 'samples': 20833792, 'steps': 40690, 'loss/train': 0.8139358758926392}
+02/26/2022 06:15:24 - INFO - codeparrot_training - Step 40691: {'lr': 4.498387742402674e-05, 'samples': 20834304, 'steps': 40691, 'loss/train': 1.3843085765838623}
+02/26/2022 06:15:29 - INFO - codeparrot_training - Step 40692: {'lr': 4.4974514098884476e-05, 'samples': 20834816, 'steps': 40692, 'loss/train': 1.636030912399292}
+02/26/2022 06:15:33 - INFO - codeparrot_training - Step 40693: {'lr': 4.496515165200632e-05, 'samples': 20835328, 'steps': 40693, 'loss/train': 0.40758177638053894}
+02/26/2022 06:15:38 - INFO - codeparrot_training - Step 40694: {'lr': 4.495579008343234e-05, 'samples': 20835840, 'steps': 40694, 'loss/train': 1.7806816101074219}
+02/26/2022 06:15:42 - INFO - codeparrot_training - Step 40695: {'lr': 4.494642939320262e-05, 'samples': 20836352, 'steps': 40695, 'loss/train': 2.207427740097046}
+02/26/2022 06:15:47 - INFO - codeparrot_training - Step 40696: {'lr': 4.493706958135724e-05, 'samples': 20836864, 'steps': 40696, 'loss/train': 0.799826443195343}
+02/26/2022 06:15:51 - INFO - codeparrot_training - Step 40697: {'lr': 4.492771064793641e-05, 'samples': 20837376, 'steps': 40697, 'loss/train': 2.2678768634796143}
+02/26/2022 06:15:56 - INFO - codeparrot_training - Step 40698: {'lr': 4.4918352592980135e-05, 'samples': 20837888, 'steps': 40698, 'loss/train': 1.369477391242981}
+02/26/2022 06:16:00 - INFO - codeparrot_training - Step 40699: {'lr': 4.490899541652854e-05, 'samples': 20838400, 'steps': 40699, 'loss/train': 0.9359695315361023}
+02/26/2022 06:16:05 - INFO - codeparrot_training - Step 40700: {'lr': 4.4899639118621604e-05, 'samples': 20838912, 'steps': 40700, 'loss/train': 2.0658695697784424}
+02/26/2022 06:16:09 - INFO - codeparrot_training - Step 40701: {'lr': 4.489028369929954e-05, 'samples': 20839424, 'steps': 40701, 'loss/train': 1.4029302597045898}
+02/26/2022 06:16:14 - INFO - codeparrot_training - Step 40702: {'lr': 4.48809291586024e-05, 'samples': 20839936, 'steps': 40702, 'loss/train': 0.7578195929527283}
+02/26/2022 06:16:18 - INFO - codeparrot_training - Step 40703: {'lr': 4.4871575496570196e-05, 'samples': 20840448, 'steps': 40703, 'loss/train': 0.972649097442627}
+02/26/2022 06:16:23 - INFO - codeparrot_training - Step 40704: {'lr': 4.4862222713242974e-05, 'samples': 20840960, 'steps': 40704, 'loss/train': 1.363482117652893}
+02/26/2022 06:16:27 - INFO - codeparrot_training - Step 40705: {'lr': 4.4852870808660935e-05, 'samples': 20841472, 'steps': 40705, 'loss/train': 1.8993037939071655}
+02/26/2022 06:16:33 - INFO - codeparrot_training - Step 40706: {'lr': 4.4843519782864025e-05, 'samples': 20841984, 'steps': 40706, 'loss/train': 1.552755355834961}
+02/26/2022 06:16:36 - INFO - codeparrot_training - Step 40707: {'lr': 4.483416963589235e-05, 'samples': 20842496, 'steps': 40707, 'loss/train': 2.6266660690307617}
+02/26/2022 06:16:42 - INFO - codeparrot_training - Step 40708: {'lr': 4.482482036778587e-05, 'samples': 20843008, 'steps': 40708, 'loss/train': 1.4920932054519653}
+02/26/2022 06:16:45 - INFO - codeparrot_training - Step 40709: {'lr': 4.481547197858479e-05, 'samples': 20843520, 'steps': 40709, 'loss/train': 3.0304181575775146}
+02/26/2022 06:16:51 - INFO - codeparrot_training - Step 40710: {'lr': 4.480612446832899e-05, 'samples': 20844032, 'steps': 40710, 'loss/train': 1.311439037322998}
+02/26/2022 06:16:54 - INFO - codeparrot_training - Step 40711: {'lr': 4.4796777837058736e-05, 'samples': 20844544, 'steps': 40711, 'loss/train': 1.6831220388412476}
+02/26/2022 06:17:00 - INFO - codeparrot_training - Step 40712: {'lr': 4.4787432084813814e-05, 'samples': 20845056, 'steps': 40712, 'loss/train': 1.8098589181900024}
+02/26/2022 06:17:03 - INFO - codeparrot_training - Step 40713: {'lr': 4.47780872116344e-05, 'samples': 20845568, 'steps': 40713, 'loss/train': 0.9848004579544067}
+02/26/2022 06:17:09 - INFO - codeparrot_training - Step 40714: {'lr': 4.476874321756047e-05, 'samples': 20846080, 'steps': 40714, 'loss/train': 1.3946572542190552}
+02/26/2022 06:17:12 - INFO - codeparrot_training - Step 40715: {'lr': 4.475940010263219e-05, 'samples': 20846592, 'steps': 40715, 'loss/train': 2.43644380569458}
+02/26/2022 06:17:19 - INFO - codeparrot_training - Step 40716: {'lr': 4.475005786688932e-05, 'samples': 20847104, 'steps': 40716, 'loss/train': 1.3501920700073242}
+02/26/2022 06:17:22 - INFO - codeparrot_training - Step 40717: {'lr': 4.474071651037209e-05, 'samples': 20847616, 'steps': 40717, 'loss/train': 1.3525301218032837}
+02/26/2022 06:17:28 - INFO - codeparrot_training - Step 40718: {'lr': 4.4731376033120416e-05, 'samples': 20848128, 'steps': 40718, 'loss/train': 1.175323724746704}
+02/26/2022 06:17:31 - INFO - codeparrot_training - Step 40719: {'lr': 4.4722036435174474e-05, 'samples': 20848640, 'steps': 40719, 'loss/train': 1.742234706878662}
+02/26/2022 06:17:37 - INFO - codeparrot_training - Step 40720: {'lr': 4.471269771657399e-05, 'samples': 20849152, 'steps': 40720, 'loss/train': 2.0646002292633057}
+02/26/2022 06:17:40 - INFO - codeparrot_training - Step 40721: {'lr': 4.470335987735921e-05, 'samples': 20849664, 'steps': 40721, 'loss/train': 2.9056432247161865}
+02/26/2022 06:17:46 - INFO - codeparrot_training - Step 40722: {'lr': 4.469402291756999e-05, 'samples': 20850176, 'steps': 40722, 'loss/train': 1.9018305540084839}
+02/26/2022 06:17:49 - INFO - codeparrot_training - Step 40723: {'lr': 4.468468683724644e-05, 'samples': 20850688, 'steps': 40723, 'loss/train': 1.7955676317214966}
+02/26/2022 06:17:55 - INFO - codeparrot_training - Step 40724: {'lr': 4.4675351636428466e-05, 'samples': 20851200, 'steps': 40724, 'loss/train': 2.1126933097839355}
+02/26/2022 06:17:58 - INFO - codeparrot_training - Step 40725: {'lr': 4.466601731515613e-05, 'samples': 20851712, 'steps': 40725, 'loss/train': 0.5542493462562561}
+02/26/2022 06:18:04 - INFO - codeparrot_training - Step 40726: {'lr': 4.46566838734693e-05, 'samples': 20852224, 'steps': 40726, 'loss/train': 2.639909267425537}
+02/26/2022 06:18:08 - INFO - codeparrot_training - Step 40727: {'lr': 4.4647351311408066e-05, 'samples': 20852736, 'steps': 40727, 'loss/train': 0.20054863393306732}
+02/26/2022 06:18:13 - INFO - codeparrot_training - Step 40728: {'lr': 4.46380196290124e-05, 'samples': 20853248, 'steps': 40728, 'loss/train': 1.1855874061584473}
+02/26/2022 06:18:17 - INFO - codeparrot_training - Step 40729: {'lr': 4.462868882632226e-05, 'samples': 20853760, 'steps': 40729, 'loss/train': 2.0065133571624756}
+02/26/2022 06:18:22 - INFO - codeparrot_training - Step 40730: {'lr': 4.461935890337751e-05, 'samples': 20854272, 'steps': 40730, 'loss/train': 1.4884397983551025}
+02/26/2022 06:18:26 - INFO - codeparrot_training - Step 40731: {'lr': 4.461002986021831e-05, 'samples': 20854784, 'steps': 40731, 'loss/train': 2.6291041374206543}
+02/26/2022 06:18:31 - INFO - codeparrot_training - Step 40732: {'lr': 4.460070169688449e-05, 'samples': 20855296, 'steps': 40732, 'loss/train': 2.4353983402252197}
+02/26/2022 06:18:35 - INFO - codeparrot_training - Step 40733: {'lr': 4.459137441341607e-05, 'samples': 20855808, 'steps': 40733, 'loss/train': 1.6389790773391724}
+02/26/2022 06:18:41 - INFO - codeparrot_training - Step 40734: {'lr': 4.458204800985288e-05, 'samples': 20856320, 'steps': 40734, 'loss/train': 1.2082011699676514}
+02/26/2022 06:18:44 - INFO - codeparrot_training - Step 40735: {'lr': 4.457272248623506e-05, 'samples': 20856832, 'steps': 40735, 'loss/train': 1.411363124847412}
+02/26/2022 06:18:50 - INFO - codeparrot_training - Step 40736: {'lr': 4.456339784260246e-05, 'samples': 20857344, 'steps': 40736, 'loss/train': 1.9699811935424805}
+02/26/2022 06:18:54 - INFO - codeparrot_training - Step 40737: {'lr': 4.4554074078995034e-05, 'samples': 20857856, 'steps': 40737, 'loss/train': 0.40651679039001465}
+02/26/2022 06:18:59 - INFO - codeparrot_training - Step 40738: {'lr': 4.454475119545265e-05, 'samples': 20858368, 'steps': 40738, 'loss/train': 1.990043044090271}
+02/26/2022 06:19:03 - INFO - codeparrot_training - Step 40739: {'lr': 4.453542919201536e-05, 'samples': 20858880, 'steps': 40739, 'loss/train': 1.5787826776504517}
+02/26/2022 06:19:08 - INFO - codeparrot_training - Step 40740: {'lr': 4.4526108068723065e-05, 'samples': 20859392, 'steps': 40740, 'loss/train': 0.30250421166419983}
+02/26/2022 06:19:12 - INFO - codeparrot_training - Step 40741: {'lr': 4.451678782561569e-05, 'samples': 20859904, 'steps': 40741, 'loss/train': 1.5993998050689697}
+02/26/2022 06:19:17 - INFO - codeparrot_training - Step 40742: {'lr': 4.450746846273307e-05, 'samples': 20860416, 'steps': 40742, 'loss/train': 2.2453672885894775}
+02/26/2022 06:19:21 - INFO - codeparrot_training - Step 40743: {'lr': 4.4498149980115255e-05, 'samples': 20860928, 'steps': 40743, 'loss/train': 1.0461004972457886}
+02/26/2022 06:19:27 - INFO - codeparrot_training - Step 40744: {'lr': 4.448883237780213e-05, 'samples': 20861440, 'steps': 40744, 'loss/train': 0.5661978721618652}
+02/26/2022 06:19:30 - INFO - codeparrot_training - Step 40745: {'lr': 4.4479515655833583e-05, 'samples': 20861952, 'steps': 40745, 'loss/train': 1.9539790153503418}
+02/26/2022 06:19:35 - INFO - codeparrot_training - Step 40746: {'lr': 4.4470199814249544e-05, 'samples': 20862464, 'steps': 40746, 'loss/train': 2.5866692066192627}
+02/26/2022 06:19:39 - INFO - codeparrot_training - Step 40747: {'lr': 4.4460884853089825e-05, 'samples': 20862976, 'steps': 40747, 'loss/train': 2.005302667617798}
+02/26/2022 06:19:45 - INFO - codeparrot_training - Step 40748: {'lr': 4.4451570772394475e-05, 'samples': 20863488, 'steps': 40748, 'loss/train': 1.261811375617981}
+02/26/2022 06:19:48 - INFO - codeparrot_training - Step 40749: {'lr': 4.444225757220327e-05, 'samples': 20864000, 'steps': 40749, 'loss/train': 1.837463617324829}
+02/26/2022 06:19:54 - INFO - codeparrot_training - Step 40750: {'lr': 4.4432945252556286e-05, 'samples': 20864512, 'steps': 40750, 'loss/train': 2.495427131652832}
+02/26/2022 06:19:57 - INFO - codeparrot_training - Step 40751: {'lr': 4.4423633813493156e-05, 'samples': 20865024, 'steps': 40751, 'loss/train': 2.406679630279541}
+02/26/2022 06:20:03 - INFO - codeparrot_training - Step 40752: {'lr': 4.441432325505398e-05, 'samples': 20865536, 'steps': 40752, 'loss/train': 1.4487475156784058}
+02/26/2022 06:20:07 - INFO - codeparrot_training - Step 40753: {'lr': 4.4405013577278495e-05, 'samples': 20866048, 'steps': 40753, 'loss/train': 1.676520824432373}
+02/26/2022 06:20:13 - INFO - codeparrot_training - Step 40754: {'lr': 4.439570478020677e-05, 'samples': 20866560, 'steps': 40754, 'loss/train': 1.4793343544006348}
+02/26/2022 06:20:16 - INFO - codeparrot_training - Step 40755: {'lr': 4.438639686387841e-05, 'samples': 20867072, 'steps': 40755, 'loss/train': 2.4943671226501465}
+02/26/2022 06:20:24 - INFO - codeparrot_training - Step 40756: {'lr': 4.4377089828333525e-05, 'samples': 20867584, 'steps': 40756, 'loss/train': 2.4255130290985107}
+02/26/2022 06:20:27 - INFO - codeparrot_training - Step 40757: {'lr': 4.436778367361183e-05, 'samples': 20868096, 'steps': 40757, 'loss/train': 1.2586151361465454}
+02/26/2022 06:20:33 - INFO - codeparrot_training - Step 40758: {'lr': 4.4358478399753376e-05, 'samples': 20868608, 'steps': 40758, 'loss/train': 2.4424874782562256}
+02/26/2022 06:20:36 - INFO - codeparrot_training - Step 40759: {'lr': 4.434917400679778e-05, 'samples': 20869120, 'steps': 40759, 'loss/train': 0.986875057220459}
+02/26/2022 06:20:42 - INFO - codeparrot_training - Step 40760: {'lr': 4.433987049478508e-05, 'samples': 20869632, 'steps': 40760, 'loss/train': 0.9369402527809143}
+02/26/2022 06:20:45 - INFO - codeparrot_training - Step 40761: {'lr': 4.4330567863755e-05, 'samples': 20870144, 'steps': 40761, 'loss/train': 2.6631221771240234}
+02/26/2022 06:20:51 - INFO - codeparrot_training - Step 40762: {'lr': 4.432126611374754e-05, 'samples': 20870656, 'steps': 40762, 'loss/train': 1.162684679031372}
+02/26/2022 06:20:54 - INFO - codeparrot_training - Step 40763: {'lr': 4.4311965244802465e-05, 'samples': 20871168, 'steps': 40763, 'loss/train': 1.3134104013442993}
+02/26/2022 06:21:00 - INFO - codeparrot_training - Step 40764: {'lr': 4.430266525695961e-05, 'samples': 20871680, 'steps': 40764, 'loss/train': 2.3260109424591064}
+02/26/2022 06:21:03 - INFO - codeparrot_training - Step 40765: {'lr': 4.429336615025875e-05, 'samples': 20872192, 'steps': 40765, 'loss/train': 1.045145034790039}
+02/26/2022 06:21:11 - INFO - codeparrot_training - Step 40766: {'lr': 4.428406792473988e-05, 'samples': 20872704, 'steps': 40766, 'loss/train': 1.8641666173934937}
+02/26/2022 06:21:14 - INFO - codeparrot_training - Step 40767: {'lr': 4.427477058044271e-05, 'samples': 20873216, 'steps': 40767, 'loss/train': 2.126070261001587}
+02/26/2022 06:21:20 - INFO - codeparrot_training - Step 40768: {'lr': 4.426547411740711e-05, 'samples': 20873728, 'steps': 40768, 'loss/train': 0.6071975231170654}
+02/26/2022 06:21:24 - INFO - codeparrot_training - Step 40769: {'lr': 4.425617853567282e-05, 'samples': 20874240, 'steps': 40769, 'loss/train': 1.2382758855819702}
+02/26/2022 06:21:27 - INFO - codeparrot_training - Step 40770: {'lr': 4.424688383527981e-05, 'samples': 20874752, 'steps': 40770, 'loss/train': 2.252837657928467}
+02/26/2022 06:21:33 - INFO - codeparrot_training - Step 40771: {'lr': 4.423759001626773e-05, 'samples': 20875264, 'steps': 40771, 'loss/train': 0.39704710245132446}
+02/26/2022 06:21:36 - INFO - codeparrot_training - Step 40772: {'lr': 4.4228297078676625e-05, 'samples': 20875776, 'steps': 40772, 'loss/train': 1.6951396465301514}
+02/26/2022 06:21:42 - INFO - codeparrot_training - Step 40773: {'lr': 4.421900502254603e-05, 'samples': 20876288, 'steps': 40773, 'loss/train': 1.009183406829834}
+02/26/2022 06:21:45 - INFO - codeparrot_training - Step 40774: {'lr': 4.4209713847915915e-05, 'samples': 20876800, 'steps': 40774, 'loss/train': 2.698479175567627}
+02/26/2022 06:21:51 - INFO - codeparrot_training - Step 40775: {'lr': 4.420042355482601e-05, 'samples': 20877312, 'steps': 40775, 'loss/train': 1.1100869178771973}
+02/26/2022 06:21:54 - INFO - codeparrot_training - Step 40776: {'lr': 4.4191134143316244e-05, 'samples': 20877824, 'steps': 40776, 'loss/train': 2.363595724105835}
+02/26/2022 06:22:02 - INFO - codeparrot_training - Step 40777: {'lr': 4.418184561342617e-05, 'samples': 20878336, 'steps': 40777, 'loss/train': 1.3484740257263184}
+02/26/2022 06:22:05 - INFO - codeparrot_training - Step 40778: {'lr': 4.417255796519581e-05, 'samples': 20878848, 'steps': 40778, 'loss/train': 1.744852066040039}
+02/26/2022 06:22:11 - INFO - codeparrot_training - Step 40779: {'lr': 4.416327119866484e-05, 'samples': 20879360, 'steps': 40779, 'loss/train': 1.7797449827194214}
+02/26/2022 06:22:14 - INFO - codeparrot_training - Step 40780: {'lr': 4.415398531387305e-05, 'samples': 20879872, 'steps': 40780, 'loss/train': 1.4905694723129272}
+02/26/2022 06:22:20 - INFO - codeparrot_training - Step 40781: {'lr': 4.4144700310860154e-05, 'samples': 20880384, 'steps': 40781, 'loss/train': 1.4019602537155151}
+02/26/2022 06:22:25 - INFO - codeparrot_training - Step 40782: {'lr': 4.4135416189666074e-05, 'samples': 20880896, 'steps': 40782, 'loss/train': 1.5768240690231323}
+02/26/2022 06:22:29 - INFO - codeparrot_training - Step 40783: {'lr': 4.41261329503305e-05, 'samples': 20881408, 'steps': 40783, 'loss/train': 1.9067399501800537}
+02/26/2022 06:22:35 - INFO - codeparrot_training - Step 40784: {'lr': 4.411685059289314e-05, 'samples': 20881920, 'steps': 40784, 'loss/train': 1.9370423555374146}
+02/26/2022 06:22:38 - INFO - codeparrot_training - Step 40785: {'lr': 4.410756911739386e-05, 'samples': 20882432, 'steps': 40785, 'loss/train': 2.0855767726898193}
+02/26/2022 06:22:45 - INFO - codeparrot_training - Step 40786: {'lr': 4.409828852387238e-05, 'samples': 20882944, 'steps': 40786, 'loss/train': 1.1551172733306885}
+02/26/2022 06:22:49 - INFO - codeparrot_training - Step 40787: {'lr': 4.408900881236846e-05, 'samples': 20883456, 'steps': 40787, 'loss/train': 1.6368690729141235}
+02/26/2022 06:22:54 - INFO - codeparrot_training - Step 40788: {'lr': 4.407972998292178e-05, 'samples': 20883968, 'steps': 40788, 'loss/train': 1.2518243789672852}
+02/26/2022 06:22:58 - INFO - codeparrot_training - Step 40789: {'lr': 4.4070452035572226e-05, 'samples': 20884480, 'steps': 40789, 'loss/train': 0.770409345626831}
+02/26/2022 06:23:04 - INFO - codeparrot_training - Step 40790: {'lr': 4.406117497035944e-05, 'samples': 20884992, 'steps': 40790, 'loss/train': 1.7084684371948242}
+02/26/2022 06:23:07 - INFO - codeparrot_training - Step 40791: {'lr': 4.4051898787323215e-05, 'samples': 20885504, 'steps': 40791, 'loss/train': 0.6682916283607483}
+02/26/2022 06:23:13 - INFO - codeparrot_training - Step 40792: {'lr': 4.404262348650317e-05, 'samples': 20886016, 'steps': 40792, 'loss/train': 1.2871159315109253}
+02/26/2022 06:23:16 - INFO - codeparrot_training - Step 40793: {'lr': 4.4033349067939264e-05, 'samples': 20886528, 'steps': 40793, 'loss/train': 1.8911652565002441}
+02/26/2022 06:23:22 - INFO - codeparrot_training - Step 40794: {'lr': 4.4024075531670955e-05, 'samples': 20887040, 'steps': 40794, 'loss/train': 1.4692856073379517}
+02/26/2022 06:23:25 - INFO - codeparrot_training - Step 40795: {'lr': 4.401480287773818e-05, 'samples': 20887552, 'steps': 40795, 'loss/train': 2.4328737258911133}
+02/26/2022 06:23:32 - INFO - codeparrot_training - Step 40796: {'lr': 4.4005531106180495e-05, 'samples': 20888064, 'steps': 40796, 'loss/train': 0.5882765650749207}
+02/26/2022 06:23:36 - INFO - codeparrot_training - Step 40797: {'lr': 4.399626021703779e-05, 'samples': 20888576, 'steps': 40797, 'loss/train': 0.7727590799331665}
+02/26/2022 06:23:41 - INFO - codeparrot_training - Step 40798: {'lr': 4.3986990210349686e-05, 'samples': 20889088, 'steps': 40798, 'loss/train': 1.250331997871399}
+02/26/2022 06:23:45 - INFO - codeparrot_training - Step 40799: {'lr': 4.397772108615592e-05, 'samples': 20889600, 'steps': 40799, 'loss/train': 0.5298184156417847}
+02/26/2022 06:23:50 - INFO - codeparrot_training - Step 40800: {'lr': 4.3968452844496076e-05, 'samples': 20890112, 'steps': 40800, 'loss/train': 0.34215831756591797}
+02/26/2022 06:23:54 - INFO - codeparrot_training - Step 40801: {'lr': 4.395918548541003e-05, 'samples': 20890624, 'steps': 40801, 'loss/train': 0.5815662741661072}
+02/26/2022 06:23:59 - INFO - codeparrot_training - Step 40802: {'lr': 4.394991900893744e-05, 'samples': 20891136, 'steps': 40802, 'loss/train': 1.3196675777435303}
+02/26/2022 06:24:03 - INFO - codeparrot_training - Step 40803: {'lr': 4.3940653415117934e-05, 'samples': 20891648, 'steps': 40803, 'loss/train': 1.2838267087936401}
+02/26/2022 06:24:08 - INFO - codeparrot_training - Step 40804: {'lr': 4.3931388703991185e-05, 'samples': 20892160, 'steps': 40804, 'loss/train': 0.6708071827888489}
+02/26/2022 06:24:12 - INFO - codeparrot_training - Step 40805: {'lr': 4.392212487559699e-05, 'samples': 20892672, 'steps': 40805, 'loss/train': 1.4132819175720215}
+02/26/2022 06:24:18 - INFO - codeparrot_training - Step 40806: {'lr': 4.391286192997499e-05, 'samples': 20893184, 'steps': 40806, 'loss/train': 1.7606216669082642}
+02/26/2022 06:24:21 - INFO - codeparrot_training - Step 40807: {'lr': 4.390359986716483e-05, 'samples': 20893696, 'steps': 40807, 'loss/train': 1.7435531616210938}
+02/26/2022 06:24:27 - INFO - codeparrot_training - Step 40808: {'lr': 4.389433868720616e-05, 'samples': 20894208, 'steps': 40808, 'loss/train': 2.457872152328491}
+02/26/2022 06:24:30 - INFO - codeparrot_training - Step 40809: {'lr': 4.388507839013875e-05, 'samples': 20894720, 'steps': 40809, 'loss/train': 2.2805721759796143}
+02/26/2022 06:24:36 - INFO - codeparrot_training - Step 40810: {'lr': 4.387581897600215e-05, 'samples': 20895232, 'steps': 40810, 'loss/train': 1.54039466381073}
+02/26/2022 06:24:39 - INFO - codeparrot_training - Step 40811: {'lr': 4.386656044483622e-05, 'samples': 20895744, 'steps': 40811, 'loss/train': 1.3677804470062256}
+02/26/2022 06:24:47 - INFO - codeparrot_training - Step 40812: {'lr': 4.385730279668038e-05, 'samples': 20896256, 'steps': 40812, 'loss/train': 1.6924481391906738}
+02/26/2022 06:24:50 - INFO - codeparrot_training - Step 40813: {'lr': 4.384804603157444e-05, 'samples': 20896768, 'steps': 40813, 'loss/train': 1.5936757326126099}
+02/26/2022 06:24:56 - INFO - codeparrot_training - Step 40814: {'lr': 4.383879014955794e-05, 'samples': 20897280, 'steps': 40814, 'loss/train': 1.200735330581665}
+02/26/2022 06:24:59 - INFO - codeparrot_training - Step 40815: {'lr': 4.3829535150670756e-05, 'samples': 20897792, 'steps': 40815, 'loss/train': 1.162596583366394}
+02/26/2022 06:25:05 - INFO - codeparrot_training - Step 40816: {'lr': 4.3820281034952224e-05, 'samples': 20898304, 'steps': 40816, 'loss/train': 1.3583128452301025}
+02/26/2022 06:25:08 - INFO - codeparrot_training - Step 40817: {'lr': 4.38110278024422e-05, 'samples': 20898816, 'steps': 40817, 'loss/train': 0.8648126721382141}
+02/26/2022 06:25:14 - INFO - codeparrot_training - Step 40818: {'lr': 4.38017754531802e-05, 'samples': 20899328, 'steps': 40818, 'loss/train': 1.525160789489746}
+02/26/2022 06:25:18 - INFO - codeparrot_training - Step 40819: {'lr': 4.379252398720601e-05, 'samples': 20899840, 'steps': 40819, 'loss/train': 1.1953896284103394}
+02/26/2022 06:25:23 - INFO - codeparrot_training - Step 40820: {'lr': 4.378327340455915e-05, 'samples': 20900352, 'steps': 40820, 'loss/train': 1.8413482904434204}
+02/26/2022 06:25:27 - INFO - codeparrot_training - Step 40821: {'lr': 4.377402370527925e-05, 'samples': 20900864, 'steps': 40821, 'loss/train': 0.7372773289680481}
+02/26/2022 06:25:34 - INFO - codeparrot_training - Step 40822: {'lr': 4.376477488940592e-05, 'samples': 20901376, 'steps': 40822, 'loss/train': 1.7024935483932495}
+02/26/2022 06:25:37 - INFO - codeparrot_training - Step 40823: {'lr': 4.375552695697885e-05, 'samples': 20901888, 'steps': 40823, 'loss/train': 1.5873041152954102}
+02/26/2022 06:25:43 - INFO - codeparrot_training - Step 40824: {'lr': 4.374627990803759e-05, 'samples': 20902400, 'steps': 40824, 'loss/train': 1.8936855792999268}
+02/26/2022 06:25:46 - INFO - codeparrot_training - Step 40825: {'lr': 4.3737033742621807e-05, 'samples': 20902912, 'steps': 40825, 'loss/train': 1.7708910703659058}
+02/26/2022 06:25:52 - INFO - codeparrot_training - Step 40826: {'lr': 4.372778846077105e-05, 'samples': 20903424, 'steps': 40826, 'loss/train': 1.6903440952301025}
+02/26/2022 06:25:55 - INFO - codeparrot_training - Step 40827: {'lr': 4.37185440625249e-05, 'samples': 20903936, 'steps': 40827, 'loss/train': 1.6371549367904663}
+02/26/2022 06:26:01 - INFO - codeparrot_training - Step 40828: {'lr': 4.3709300547923074e-05, 'samples': 20904448, 'steps': 40828, 'loss/train': 0.6397308111190796}
+02/26/2022 06:26:04 - INFO - codeparrot_training - Step 40829: {'lr': 4.37000579170051e-05, 'samples': 20904960, 'steps': 40829, 'loss/train': 0.825473427772522}
+02/26/2022 06:26:10 - INFO - codeparrot_training - Step 40830: {'lr': 4.369081616981058e-05, 'samples': 20905472, 'steps': 40830, 'loss/train': 1.4700621366500854}
+02/26/2022 06:26:13 - INFO - codeparrot_training - Step 40831: {'lr': 4.3681575306379004e-05, 'samples': 20905984, 'steps': 40831, 'loss/train': 2.26362943649292}
+02/26/2022 06:26:19 - INFO - codeparrot_training - Step 40832: {'lr': 4.367233532675011e-05, 'samples': 20906496, 'steps': 40832, 'loss/train': 1.8314299583435059}
+02/26/2022 06:26:22 - INFO - codeparrot_training - Step 40833: {'lr': 4.366309623096343e-05, 'samples': 20907008, 'steps': 40833, 'loss/train': 1.544396996498108}
+02/26/2022 06:26:30 - INFO - codeparrot_training - Step 40834: {'lr': 4.365385801905852e-05, 'samples': 20907520, 'steps': 40834, 'loss/train': 1.4938440322875977}
+02/26/2022 06:26:33 - INFO - codeparrot_training - Step 40835: {'lr': 4.364462069107489e-05, 'samples': 20908032, 'steps': 40835, 'loss/train': 2.0050759315490723}
+02/26/2022 06:26:39 - INFO - codeparrot_training - Step 40836: {'lr': 4.3635384247052265e-05, 'samples': 20908544, 'steps': 40836, 'loss/train': 2.5611331462860107}
+02/26/2022 06:26:42 - INFO - codeparrot_training - Step 40837: {'lr': 4.36261486870301e-05, 'samples': 20909056, 'steps': 40837, 'loss/train': 0.5659295916557312}
+02/26/2022 06:26:48 - INFO - codeparrot_training - Step 40838: {'lr': 4.3616914011048004e-05, 'samples': 20909568, 'steps': 40838, 'loss/train': 1.9685627222061157}
+02/26/2022 06:26:51 - INFO - codeparrot_training - Step 40839: {'lr': 4.360768021914543e-05, 'samples': 20910080, 'steps': 40839, 'loss/train': 1.1939127445220947}
+02/26/2022 06:26:57 - INFO - codeparrot_training - Step 40840: {'lr': 4.359844731136209e-05, 'samples': 20910592, 'steps': 40840, 'loss/train': 1.7248051166534424}
+02/26/2022 06:27:00 - INFO - codeparrot_training - Step 40841: {'lr': 4.3589215287737486e-05, 'samples': 20911104, 'steps': 40841, 'loss/train': 2.264495372772217}
+02/26/2022 06:27:06 - INFO - codeparrot_training - Step 40842: {'lr': 4.35799841483111e-05, 'samples': 20911616, 'steps': 40842, 'loss/train': 1.7720999717712402}
+02/26/2022 06:27:09 - INFO - codeparrot_training - Step 40843: {'lr': 4.357075389312248e-05, 'samples': 20912128, 'steps': 40843, 'loss/train': 1.3957901000976562}
+02/26/2022 06:27:15 - INFO - codeparrot_training - Step 40844: {'lr': 4.356152452221127e-05, 'samples': 20912640, 'steps': 40844, 'loss/train': 1.2353039979934692}
+02/26/2022 06:27:18 - INFO - codeparrot_training - Step 40845: {'lr': 4.3552296035616836e-05, 'samples': 20913152, 'steps': 40845, 'loss/train': 2.1513874530792236}
+02/26/2022 06:27:24 - INFO - codeparrot_training - Step 40846: {'lr': 4.3543068433378965e-05, 'samples': 20913664, 'steps': 40846, 'loss/train': 0.04748530685901642}
+02/26/2022 06:27:27 - INFO - codeparrot_training - Step 40847: {'lr': 4.3533841715536895e-05, 'samples': 20914176, 'steps': 40847, 'loss/train': 1.8627108335494995}
+02/26/2022 06:27:35 - INFO - codeparrot_training - Step 40848: {'lr': 4.352461588213036e-05, 'samples': 20914688, 'steps': 40848, 'loss/train': 1.798604130744934}
+02/26/2022 06:27:38 - INFO - codeparrot_training - Step 40849: {'lr': 4.3515390933198756e-05, 'samples': 20915200, 'steps': 40849, 'loss/train': 4.01839017868042}
+02/26/2022 06:27:44 - INFO - codeparrot_training - Step 40850: {'lr': 4.350616686878175e-05, 'samples': 20915712, 'steps': 40850, 'loss/train': 0.030324609950184822}
+02/26/2022 06:27:47 - INFO - codeparrot_training - Step 40851: {'lr': 4.349694368891865e-05, 'samples': 20916224, 'steps': 40851, 'loss/train': 1.670116901397705}
+02/26/2022 06:27:53 - INFO - codeparrot_training - Step 40852: {'lr': 4.348772139364912e-05, 'samples': 20916736, 'steps': 40852, 'loss/train': 2.041996479034424}
+02/26/2022 06:27:56 - INFO - codeparrot_training - Step 40853: {'lr': 4.3478499983012565e-05, 'samples': 20917248, 'steps': 40853, 'loss/train': 2.494936466217041}
+02/26/2022 06:28:02 - INFO - codeparrot_training - Step 40854: {'lr': 4.346927945704865e-05, 'samples': 20917760, 'steps': 40854, 'loss/train': 2.2606112957000732}
+02/26/2022 06:28:05 - INFO - codeparrot_training - Step 40855: {'lr': 4.346005981579665e-05, 'samples': 20918272, 'steps': 40855, 'loss/train': 1.6669801473617554}
+02/26/2022 06:28:11 - INFO - codeparrot_training - Step 40856: {'lr': 4.345084105929622e-05, 'samples': 20918784, 'steps': 40856, 'loss/train': 2.2037057876586914}
+02/26/2022 06:28:14 - INFO - codeparrot_training - Step 40857: {'lr': 4.3441623187586766e-05, 'samples': 20919296, 'steps': 40857, 'loss/train': 1.3628287315368652}
+02/26/2022 06:28:22 - INFO - codeparrot_training - Step 40858: {'lr': 4.343240620070785e-05, 'samples': 20919808, 'steps': 40858, 'loss/train': 1.6588469743728638}
+02/26/2022 06:28:25 - INFO - codeparrot_training - Step 40859: {'lr': 4.342319009869891e-05, 'samples': 20920320, 'steps': 40859, 'loss/train': 1.1480106115341187}
+02/26/2022 06:28:30 - INFO - codeparrot_training - Step 40860: {'lr': 4.341397488159943e-05, 'samples': 20920832, 'steps': 40860, 'loss/train': 2.2994632720947266}
+02/26/2022 06:28:34 - INFO - codeparrot_training - Step 40861: {'lr': 4.3404760549448804e-05, 'samples': 20921344, 'steps': 40861, 'loss/train': 1.8536527156829834}
+02/26/2022 06:28:39 - INFO - codeparrot_training - Step 40862: {'lr': 4.339554710228666e-05, 'samples': 20921856, 'steps': 40862, 'loss/train': 2.179905652999878}
+02/26/2022 06:28:43 - INFO - codeparrot_training - Step 40863: {'lr': 4.33863345401524e-05, 'samples': 20922368, 'steps': 40863, 'loss/train': 1.7461615800857544}
+02/26/2022 06:28:48 - INFO - codeparrot_training - Step 40864: {'lr': 4.337712286308546e-05, 'samples': 20922880, 'steps': 40864, 'loss/train': 0.3026457726955414}
+02/26/2022 06:28:52 - INFO - codeparrot_training - Step 40865: {'lr': 4.3367912071125256e-05, 'samples': 20923392, 'steps': 40865, 'loss/train': 1.9761264324188232}
+02/26/2022 06:28:57 - INFO - codeparrot_training - Step 40866: {'lr': 4.335870216431137e-05, 'samples': 20923904, 'steps': 40866, 'loss/train': 2.746539831161499}
+02/26/2022 06:29:01 - INFO - codeparrot_training - Step 40867: {'lr': 4.334949314268319e-05, 'samples': 20924416, 'steps': 40867, 'loss/train': 1.6884409189224243}
+02/26/2022 06:29:08 - INFO - codeparrot_training - Step 40868: {'lr': 4.334028500628015e-05, 'samples': 20924928, 'steps': 40868, 'loss/train': 1.7547966241836548}
+02/26/2022 06:29:11 - INFO - codeparrot_training - Step 40869: {'lr': 4.333107775514167e-05, 'samples': 20925440, 'steps': 40869, 'loss/train': 1.5080362558364868}
+02/26/2022 06:29:17 - INFO - codeparrot_training - Step 40870: {'lr': 4.332187138930726e-05, 'samples': 20925952, 'steps': 40870, 'loss/train': 2.6367745399475098}
+02/26/2022 06:29:20 - INFO - codeparrot_training - Step 40871: {'lr': 4.331266590881636e-05, 'samples': 20926464, 'steps': 40871, 'loss/train': 1.408315658569336}
+02/26/2022 06:29:26 - INFO - codeparrot_training - Step 40872: {'lr': 4.330346131370835e-05, 'samples': 20926976, 'steps': 40872, 'loss/train': 1.6436982154846191}
+02/26/2022 06:29:29 - INFO - codeparrot_training - Step 40873: {'lr': 4.329425760402267e-05, 'samples': 20927488, 'steps': 40873, 'loss/train': 1.768805980682373}
+02/26/2022 06:29:35 - INFO - codeparrot_training - Step 40874: {'lr': 4.328505477979869e-05, 'samples': 20928000, 'steps': 40874, 'loss/train': 0.49441030621528625}
+02/26/2022 06:29:38 - INFO - codeparrot_training - Step 40875: {'lr': 4.327585284107597e-05, 'samples': 20928512, 'steps': 40875, 'loss/train': 0.9649516344070435}
+02/26/2022 06:29:44 - INFO - codeparrot_training - Step 40876: {'lr': 4.326665178789385e-05, 'samples': 20929024, 'steps': 40876, 'loss/train': 1.2714118957519531}
+02/26/2022 06:29:47 - INFO - codeparrot_training - Step 40877: {'lr': 4.325745162029174e-05, 'samples': 20929536, 'steps': 40877, 'loss/train': 2.574646472930908}
+02/26/2022 06:29:55 - INFO - codeparrot_training - Step 40878: {'lr': 4.324825233830901e-05, 'samples': 20930048, 'steps': 40878, 'loss/train': 1.4813597202301025}
+02/26/2022 06:29:58 - INFO - codeparrot_training - Step 40879: {'lr': 4.323905394198519e-05, 'samples': 20930560, 'steps': 40879, 'loss/train': 1.8668663501739502}
+02/26/2022 06:30:04 - INFO - codeparrot_training - Step 40880: {'lr': 4.3229856431359515e-05, 'samples': 20931072, 'steps': 40880, 'loss/train': 1.1853092908859253}
+02/26/2022 06:30:07 - INFO - codeparrot_training - Step 40881: {'lr': 4.3220659806471635e-05, 'samples': 20931584, 'steps': 40881, 'loss/train': 2.443833112716675}
+02/26/2022 06:30:13 - INFO - codeparrot_training - Step 40882: {'lr': 4.3211464067360654e-05, 'samples': 20932096, 'steps': 40882, 'loss/train': 1.5212701559066772}
+02/26/2022 06:30:16 - INFO - codeparrot_training - Step 40883: {'lr': 4.3202269214066144e-05, 'samples': 20932608, 'steps': 40883, 'loss/train': 2.31856107711792}
+02/26/2022 06:30:22 - INFO - codeparrot_training - Step 40884: {'lr': 4.319307524662738e-05, 'samples': 20933120, 'steps': 40884, 'loss/train': 2.0456039905548096}
+02/26/2022 06:30:25 - INFO - codeparrot_training - Step 40885: {'lr': 4.318388216508395e-05, 'samples': 20933632, 'steps': 40885, 'loss/train': 1.1022002696990967}
+02/26/2022 06:30:31 - INFO - codeparrot_training - Step 40886: {'lr': 4.317468996947499e-05, 'samples': 20934144, 'steps': 40886, 'loss/train': 0.9232212901115417}
+02/26/2022 06:30:34 - INFO - codeparrot_training - Step 40887: {'lr': 4.3165498659840014e-05, 'samples': 20934656, 'steps': 40887, 'loss/train': 8.901434898376465}
+02/26/2022 06:30:40 - INFO - codeparrot_training - Step 40888: {'lr': 4.3156308236218326e-05, 'samples': 20935168, 'steps': 40888, 'loss/train': 1.868941068649292}
+02/26/2022 06:30:43 - INFO - codeparrot_training - Step 40889: {'lr': 4.314711869864946e-05, 'samples': 20935680, 'steps': 40889, 'loss/train': 1.9776207208633423}
+02/26/2022 06:30:49 - INFO - codeparrot_training - Step 40890: {'lr': 4.3137930047172506e-05, 'samples': 20936192, 'steps': 40890, 'loss/train': 1.3485157489776611}
+02/26/2022 06:30:52 - INFO - codeparrot_training - Step 40891: {'lr': 4.312874228182706e-05, 'samples': 20936704, 'steps': 40891, 'loss/train': 3.263789415359497}
+02/26/2022 06:30:58 - INFO - codeparrot_training - Step 40892: {'lr': 4.3119555402652334e-05, 'samples': 20937216, 'steps': 40892, 'loss/train': 0.8976107835769653}
+02/26/2022 06:31:01 - INFO - codeparrot_training - Step 40893: {'lr': 4.311036940968777e-05, 'samples': 20937728, 'steps': 40893, 'loss/train': 2.054450511932373}
+02/26/2022 06:31:07 - INFO - codeparrot_training - Step 40894: {'lr': 4.310118430297272e-05, 'samples': 20938240, 'steps': 40894, 'loss/train': 2.3945393562316895}
+02/26/2022 06:31:10 - INFO - codeparrot_training - Step 40895: {'lr': 4.309200008254649e-05, 'samples': 20938752, 'steps': 40895, 'loss/train': 1.1406890153884888}
+02/26/2022 06:31:17 - INFO - codeparrot_training - Step 40896: {'lr': 4.308281674844836e-05, 'samples': 20939264, 'steps': 40896, 'loss/train': 1.7404903173446655}
+02/26/2022 06:31:21 - INFO - codeparrot_training - Step 40897: {'lr': 4.307363430071781e-05, 'samples': 20939776, 'steps': 40897, 'loss/train': 2.1969399452209473}
+02/26/2022 06:31:26 - INFO - codeparrot_training - Step 40898: {'lr': 4.3064452739394104e-05, 'samples': 20940288, 'steps': 40898, 'loss/train': 1.557886004447937}
+02/26/2022 06:31:30 - INFO - codeparrot_training - Step 40899: {'lr': 4.3055272064516566e-05, 'samples': 20940800, 'steps': 40899, 'loss/train': 2.0844814777374268}
+02/26/2022 06:31:35 - INFO - codeparrot_training - Step 40900: {'lr': 4.3046092276124466e-05, 'samples': 20941312, 'steps': 40900, 'loss/train': 1.954880714416504}
+02/26/2022 06:31:41 - INFO - codeparrot_training - Step 40901: {'lr': 4.3036913374257276e-05, 'samples': 20941824, 'steps': 40901, 'loss/train': 1.1112169027328491}
+02/26/2022 06:31:44 - INFO - codeparrot_training - Step 40902: {'lr': 4.302773535895421e-05, 'samples': 20942336, 'steps': 40902, 'loss/train': 1.0091986656188965}
+02/26/2022 06:31:50 - INFO - codeparrot_training - Step 40903: {'lr': 4.3018558230254604e-05, 'samples': 20942848, 'steps': 40903, 'loss/train': 1.8580636978149414}
+02/26/2022 06:31:53 - INFO - codeparrot_training - Step 40904: {'lr': 4.3009381988197707e-05, 'samples': 20943360, 'steps': 40904, 'loss/train': 1.3948968648910522}
+02/26/2022 06:32:00 - INFO - codeparrot_training - Step 40905: {'lr': 4.3000206632822975e-05, 'samples': 20943872, 'steps': 40905, 'loss/train': 2.0072360038757324}
+02/26/2022 06:32:04 - INFO - codeparrot_training - Step 40906: {'lr': 4.2991032164169605e-05, 'samples': 20944384, 'steps': 40906, 'loss/train': 2.0293049812316895}
+02/26/2022 06:32:09 - INFO - codeparrot_training - Step 40907: {'lr': 4.298185858227693e-05, 'samples': 20944896, 'steps': 40907, 'loss/train': 1.7484631538391113}
+02/26/2022 06:32:13 - INFO - codeparrot_training - Step 40908: {'lr': 4.2972685887184174e-05, 'samples': 20945408, 'steps': 40908, 'loss/train': 3.557563304901123}
+02/26/2022 06:32:18 - INFO - codeparrot_training - Step 40909: {'lr': 4.296351407893076e-05, 'samples': 20945920, 'steps': 40909, 'loss/train': 2.3417558670043945}
+02/26/2022 06:32:22 - INFO - codeparrot_training - Step 40910: {'lr': 4.2954343157555894e-05, 'samples': 20946432, 'steps': 40910, 'loss/train': 0.9013440012931824}
+02/26/2022 06:32:27 - INFO - codeparrot_training - Step 40911: {'lr': 4.2945173123098895e-05, 'samples': 20946944, 'steps': 40911, 'loss/train': 2.2168171405792236}
+02/26/2022 06:32:31 - INFO - codeparrot_training - Step 40912: {'lr': 4.293600397559896e-05, 'samples': 20947456, 'steps': 40912, 'loss/train': 0.04781728610396385}
+02/26/2022 06:32:38 - INFO - codeparrot_training - Step 40913: {'lr': 4.292683571509548e-05, 'samples': 20947968, 'steps': 40913, 'loss/train': 2.33477783203125}
+02/26/2022 06:32:42 - INFO - codeparrot_training - Step 40914: {'lr': 4.291766834162772e-05, 'samples': 20948480, 'steps': 40914, 'loss/train': 2.3277344703674316}
+02/26/2022 06:32:47 - INFO - codeparrot_training - Step 40915: {'lr': 4.290850185523487e-05, 'samples': 20948992, 'steps': 40915, 'loss/train': 1.4409483671188354}
+02/26/2022 06:32:51 - INFO - codeparrot_training - Step 40916: {'lr': 4.289933625595621e-05, 'samples': 20949504, 'steps': 40916, 'loss/train': 2.034222364425659}
+02/26/2022 06:32:56 - INFO - codeparrot_training - Step 40917: {'lr': 4.289017154383107e-05, 'samples': 20950016, 'steps': 40917, 'loss/train': 1.5203218460083008}
+02/26/2022 06:33:00 - INFO - codeparrot_training - Step 40918: {'lr': 4.288100771889869e-05, 'samples': 20950528, 'steps': 40918, 'loss/train': 2.0772271156311035}
+02/26/2022 06:33:05 - INFO - codeparrot_training - Step 40919: {'lr': 4.287184478119824e-05, 'samples': 20951040, 'steps': 40919, 'loss/train': 1.5221832990646362}
+02/26/2022 06:33:09 - INFO - codeparrot_training - Step 40920: {'lr': 4.2862682730769154e-05, 'samples': 20951552, 'steps': 40920, 'loss/train': 1.3872801065444946}
+02/26/2022 06:33:14 - INFO - codeparrot_training - Step 40921: {'lr': 4.2853521567650435e-05, 'samples': 20952064, 'steps': 40921, 'loss/train': 1.5734838247299194}
+02/26/2022 06:33:18 - INFO - codeparrot_training - Step 40922: {'lr': 4.2844361291881515e-05, 'samples': 20952576, 'steps': 40922, 'loss/train': 1.678175926208496}
+02/26/2022 06:33:23 - INFO - codeparrot_training - Step 40923: {'lr': 4.283520190350151e-05, 'samples': 20953088, 'steps': 40923, 'loss/train': 1.7431775331497192}
+02/26/2022 06:33:27 - INFO - codeparrot_training - Step 40924: {'lr': 4.282604340254984e-05, 'samples': 20953600, 'steps': 40924, 'loss/train': 1.6529545783996582}
+02/26/2022 06:33:33 - INFO - codeparrot_training - Step 40925: {'lr': 4.281688578906551e-05, 'samples': 20954112, 'steps': 40925, 'loss/train': 1.1516813039779663}
+02/26/2022 06:33:37 - INFO - codeparrot_training - Step 40926: {'lr': 4.280772906308789e-05, 'samples': 20954624, 'steps': 40926, 'loss/train': 0.6605219841003418}
+02/26/2022 06:33:42 - INFO - codeparrot_training - Step 40927: {'lr': 4.279857322465611e-05, 'samples': 20955136, 'steps': 40927, 'loss/train': 0.5205278396606445}
+02/26/2022 06:33:46 - INFO - codeparrot_training - Step 40928: {'lr': 4.278941827380953e-05, 'samples': 20955648, 'steps': 40928, 'loss/train': 1.8531779050827026}
+02/26/2022 06:33:51 - INFO - codeparrot_training - Step 40929: {'lr': 4.2780264210587254e-05, 'samples': 20956160, 'steps': 40929, 'loss/train': 1.3304390907287598}
+02/26/2022 06:33:55 - INFO - codeparrot_training - Step 40930: {'lr': 4.2771111035028564e-05, 'samples': 20956672, 'steps': 40930, 'loss/train': 1.2356091737747192}
+02/26/2022 06:34:00 - INFO - codeparrot_training - Step 40931: {'lr': 4.276195874717253e-05, 'samples': 20957184, 'steps': 40931, 'loss/train': 0.5766004920005798}
+02/26/2022 06:34:04 - INFO - codeparrot_training - Step 40932: {'lr': 4.275280734705855e-05, 'samples': 20957696, 'steps': 40932, 'loss/train': 1.6547880172729492}
+02/26/2022 06:34:09 - INFO - codeparrot_training - Step 40933: {'lr': 4.274365683472572e-05, 'samples': 20958208, 'steps': 40933, 'loss/train': 2.637953042984009}
+02/26/2022 06:34:13 - INFO - codeparrot_training - Step 40934: {'lr': 4.273450721021327e-05, 'samples': 20958720, 'steps': 40934, 'loss/train': 1.4736648797988892}
+02/26/2022 06:34:18 - INFO - codeparrot_training - Step 40935: {'lr': 4.272535847356032e-05, 'samples': 20959232, 'steps': 40935, 'loss/train': 1.8897048234939575}
+02/26/2022 06:34:22 - INFO - codeparrot_training - Step 40936: {'lr': 4.271621062480616e-05, 'samples': 20959744, 'steps': 40936, 'loss/train': 1.274722695350647}
+02/26/2022 06:34:27 - INFO - codeparrot_training - Step 40937: {'lr': 4.2707063663989936e-05, 'samples': 20960256, 'steps': 40937, 'loss/train': 2.474787712097168}
+02/26/2022 06:34:31 - INFO - codeparrot_training - Step 40938: {'lr': 4.269791759115082e-05, 'samples': 20960768, 'steps': 40938, 'loss/train': 0.8456509709358215}
+02/26/2022 06:34:38 - INFO - codeparrot_training - Step 40939: {'lr': 4.268877240632796e-05, 'samples': 20961280, 'steps': 40939, 'loss/train': 2.729900360107422}
+02/26/2022 06:34:41 - INFO - codeparrot_training - Step 40940: {'lr': 4.267962810956061e-05, 'samples': 20961792, 'steps': 40940, 'loss/train': 1.9282358884811401}
+02/26/2022 06:34:47 - INFO - codeparrot_training - Step 40941: {'lr': 4.2670484700887855e-05, 'samples': 20962304, 'steps': 40941, 'loss/train': 1.52631676197052}
+02/26/2022 06:34:50 - INFO - codeparrot_training - Step 40942: {'lr': 4.2661342180349035e-05, 'samples': 20962816, 'steps': 40942, 'loss/train': 1.077835202217102}
+02/26/2022 06:34:56 - INFO - codeparrot_training - Step 40943: {'lr': 4.265220054798305e-05, 'samples': 20963328, 'steps': 40943, 'loss/train': 1.5397757291793823}
+02/26/2022 06:34:59 - INFO - codeparrot_training - Step 40944: {'lr': 4.2643059803829266e-05, 'samples': 20963840, 'steps': 40944, 'loss/train': 1.5199321508407593}
+02/26/2022 06:35:05 - INFO - codeparrot_training - Step 40945: {'lr': 4.2633919947926705e-05, 'samples': 20964352, 'steps': 40945, 'loss/train': 1.4621610641479492}
+02/26/2022 06:35:08 - INFO - codeparrot_training - Step 40946: {'lr': 4.262478098031469e-05, 'samples': 20964864, 'steps': 40946, 'loss/train': 2.241328239440918}
+02/26/2022 06:35:14 - INFO - codeparrot_training - Step 40947: {'lr': 4.261564290103217e-05, 'samples': 20965376, 'steps': 40947, 'loss/train': 2.9173591136932373}
+02/26/2022 06:35:17 - INFO - codeparrot_training - Step 40948: {'lr': 4.260650571011845e-05, 'samples': 20965888, 'steps': 40948, 'loss/train': 1.6773098707199097}
+02/26/2022 06:35:23 - INFO - codeparrot_training - Step 40949: {'lr': 4.2597369407612524e-05, 'samples': 20966400, 'steps': 40949, 'loss/train': 2.913886308670044}
+02/26/2022 06:35:26 - INFO - codeparrot_training - Step 40950: {'lr': 4.258823399355374e-05, 'samples': 20966912, 'steps': 40950, 'loss/train': 2.2451696395874023}
+02/26/2022 06:35:33 - INFO - codeparrot_training - Step 40951: {'lr': 4.257909946798097e-05, 'samples': 20967424, 'steps': 40951, 'loss/train': 1.7844198942184448}
+02/26/2022 06:35:37 - INFO - codeparrot_training - Step 40952: {'lr': 4.256996583093356e-05, 'samples': 20967936, 'steps': 40952, 'loss/train': 0.5329646468162537}
+02/26/2022 06:35:42 - INFO - codeparrot_training - Step 40953: {'lr': 4.256083308245051e-05, 'samples': 20968448, 'steps': 40953, 'loss/train': 2.287973403930664}
+02/26/2022 06:35:45 - INFO - codeparrot_training - Step 40954: {'lr': 4.255170122257096e-05, 'samples': 20968960, 'steps': 40954, 'loss/train': 1.462609887123108}
+02/26/2022 06:35:51 - INFO - codeparrot_training - Step 40955: {'lr': 4.2542570251334096e-05, 'samples': 20969472, 'steps': 40955, 'loss/train': 1.0726749897003174}
+02/26/2022 06:35:55 - INFO - codeparrot_training - Step 40956: {'lr': 4.253344016877897e-05, 'samples': 20969984, 'steps': 40956, 'loss/train': 1.830863118171692}
+02/26/2022 06:36:00 - INFO - codeparrot_training - Step 40957: {'lr': 4.252431097494472e-05, 'samples': 20970496, 'steps': 40957, 'loss/train': 1.4729727506637573}
+02/26/2022 06:36:04 - INFO - codeparrot_training - Step 40958: {'lr': 4.2515182669870395e-05, 'samples': 20971008, 'steps': 40958, 'loss/train': 2.107224464416504}
+02/26/2022 06:36:09 - INFO - codeparrot_training - Step 40959: {'lr': 4.2506055253595186e-05, 'samples': 20971520, 'steps': 40959, 'loss/train': 1.4418028593063354}
+02/26/2022 06:36:13 - INFO - codeparrot_training - Step 40960: {'lr': 4.2496928726158154e-05, 'samples': 20972032, 'steps': 40960, 'loss/train': 1.3643921613693237}
+02/26/2022 06:36:19 - INFO - codeparrot_training - Step 40961: {'lr': 4.248780308759839e-05, 'samples': 20972544, 'steps': 40961, 'loss/train': 1.6517362594604492}
+02/26/2022 06:36:22 - INFO - codeparrot_training - Step 40962: {'lr': 4.2478678337954935e-05, 'samples': 20973056, 'steps': 40962, 'loss/train': 2.0048162937164307}
+02/26/2022 06:36:28 - INFO - codeparrot_training - Step 40963: {'lr': 4.246955447726697e-05, 'samples': 20973568, 'steps': 40963, 'loss/train': 1.8953315019607544}
+02/26/2022 06:36:31 - INFO - codeparrot_training - Step 40964: {'lr': 4.246043150557355e-05, 'samples': 20974080, 'steps': 40964, 'loss/train': 1.8628745079040527}
+02/26/2022 06:36:37 - INFO - codeparrot_training - Step 40965: {'lr': 4.245130942291373e-05, 'samples': 20974592, 'steps': 40965, 'loss/train': 1.5999078750610352}
+02/26/2022 06:36:41 - INFO - codeparrot_training - Step 40966: {'lr': 4.2442188229326526e-05, 'samples': 20975104, 'steps': 40966, 'loss/train': 1.1961758136749268}
+02/26/2022 06:36:46 - INFO - codeparrot_training - Step 40967: {'lr': 4.243306792485116e-05, 'samples': 20975616, 'steps': 40967, 'loss/train': 1.8956518173217773}
+02/26/2022 06:36:50 - INFO - codeparrot_training - Step 40968: {'lr': 4.242394850952661e-05, 'samples': 20976128, 'steps': 40968, 'loss/train': 1.1217402219772339}
+02/26/2022 06:36:55 - INFO - codeparrot_training - Step 40969: {'lr': 4.241482998339197e-05, 'samples': 20976640, 'steps': 40969, 'loss/train': 1.3871036767959595}
+02/26/2022 06:36:59 - INFO - codeparrot_training - Step 40970: {'lr': 4.240571234648621e-05, 'samples': 20977152, 'steps': 40970, 'loss/train': 1.8450934886932373}
+02/26/2022 06:37:04 - INFO - codeparrot_training - Step 40971: {'lr': 4.2396595598848505e-05, 'samples': 20977664, 'steps': 40971, 'loss/train': 2.474783182144165}
+02/26/2022 06:37:08 - INFO - codeparrot_training - Step 40972: {'lr': 4.238747974051787e-05, 'samples': 20978176, 'steps': 40972, 'loss/train': 1.4702836275100708}
+02/26/2022 06:37:13 - INFO - codeparrot_training - Step 40973: {'lr': 4.2378364771533366e-05, 'samples': 20978688, 'steps': 40973, 'loss/train': 1.8752280473709106}
+02/26/2022 06:37:17 - INFO - codeparrot_training - Step 40974: {'lr': 4.2369250691933955e-05, 'samples': 20979200, 'steps': 40974, 'loss/train': 1.3975797891616821}
+02/26/2022 06:37:23 - INFO - codeparrot_training - Step 40975: {'lr': 4.236013750175879e-05, 'samples': 20979712, 'steps': 40975, 'loss/train': 1.2419483661651611}
+02/26/2022 06:37:26 - INFO - codeparrot_training - Step 40976: {'lr': 4.2351025201046804e-05, 'samples': 20980224, 'steps': 40976, 'loss/train': 1.7040468454360962}
+02/26/2022 06:37:32 - INFO - codeparrot_training - Step 40977: {'lr': 4.2341913789837235e-05, 'samples': 20980736, 'steps': 40977, 'loss/train': 1.8875755071640015}
+02/26/2022 06:37:35 - INFO - codeparrot_training - Step 40978: {'lr': 4.2332803268168826e-05, 'samples': 20981248, 'steps': 40978, 'loss/train': 1.6545933485031128}
+02/26/2022 06:37:41 - INFO - codeparrot_training - Step 40979: {'lr': 4.23236936360808e-05, 'samples': 20981760, 'steps': 40979, 'loss/train': 0.6189851760864258}
+02/26/2022 06:37:44 - INFO - codeparrot_training - Step 40980: {'lr': 4.231458489361206e-05, 'samples': 20982272, 'steps': 40980, 'loss/train': 0.4759628176689148}
+02/26/2022 06:37:50 - INFO - codeparrot_training - Step 40981: {'lr': 4.230547704080184e-05, 'samples': 20982784, 'steps': 40981, 'loss/train': 1.9501804113388062}
+02/26/2022 06:37:53 - INFO - codeparrot_training - Step 40982: {'lr': 4.229637007768888e-05, 'samples': 20983296, 'steps': 40982, 'loss/train': 2.1409239768981934}
+02/26/2022 06:37:59 - INFO - codeparrot_training - Step 40983: {'lr': 4.228726400431237e-05, 'samples': 20983808, 'steps': 40983, 'loss/train': 1.6045479774475098}
+02/26/2022 06:38:02 - INFO - codeparrot_training - Step 40984: {'lr': 4.2278158820711205e-05, 'samples': 20984320, 'steps': 40984, 'loss/train': 0.621855616569519}
+02/26/2022 06:38:08 - INFO - codeparrot_training - Step 40985: {'lr': 4.226905452692456e-05, 'samples': 20984832, 'steps': 40985, 'loss/train': 1.9347269535064697}
+02/26/2022 06:38:11 - INFO - codeparrot_training - Step 40986: {'lr': 4.225995112299122e-05, 'samples': 20985344, 'steps': 40986, 'loss/train': 1.9254193305969238}
+02/26/2022 06:38:18 - INFO - codeparrot_training - Step 40987: {'lr': 4.225084860895031e-05, 'samples': 20985856, 'steps': 40987, 'loss/train': 1.4470301866531372}
+02/26/2022 06:38:21 - INFO - codeparrot_training - Step 40988: {'lr': 4.224174698484079e-05, 'samples': 20986368, 'steps': 40988, 'loss/train': 2.0797104835510254}
+02/26/2022 06:38:27 - INFO - codeparrot_training - Step 40989: {'lr': 4.2232646250701666e-05, 'samples': 20986880, 'steps': 40989, 'loss/train': 2.3309998512268066}
+02/26/2022 06:38:30 - INFO - codeparrot_training - Step 40990: {'lr': 4.2223546406571944e-05, 'samples': 20987392, 'steps': 40990, 'loss/train': 0.6763462424278259}
+02/26/2022 06:38:36 - INFO - codeparrot_training - Step 40991: {'lr': 4.221444745249056e-05, 'samples': 20987904, 'steps': 40991, 'loss/train': 2.250633955001831}
+02/26/2022 06:38:39 - INFO - codeparrot_training - Step 40992: {'lr': 4.220534938849643e-05, 'samples': 20988416, 'steps': 40992, 'loss/train': 1.4728803634643555}
+02/26/2022 06:38:44 - INFO - codeparrot_training - Step 40993: {'lr': 4.219625221462869e-05, 'samples': 20988928, 'steps': 40993, 'loss/train': 2.3220465183258057}
+02/26/2022 06:38:48 - INFO - codeparrot_training - Step 40994: {'lr': 4.21871559309262e-05, 'samples': 20989440, 'steps': 40994, 'loss/train': 1.1009777784347534}
+02/26/2022 06:38:53 - INFO - codeparrot_training - Step 40995: {'lr': 4.217806053742795e-05, 'samples': 20989952, 'steps': 40995, 'loss/train': 0.7125465869903564}
+02/26/2022 06:38:57 - INFO - codeparrot_training - Step 40996: {'lr': 4.216896603417286e-05, 'samples': 20990464, 'steps': 40996, 'loss/train': 1.5867514610290527}
+02/26/2022 06:39:04 - INFO - codeparrot_training - Step 40997: {'lr': 4.215987242119995e-05, 'samples': 20990976, 'steps': 40997, 'loss/train': 0.9622074961662292}
+02/26/2022 06:39:07 - INFO - codeparrot_training - Step 40998: {'lr': 4.2150779698548165e-05, 'samples': 20991488, 'steps': 40998, 'loss/train': 1.6610727310180664}
+02/26/2022 06:39:13 - INFO - codeparrot_training - Step 40999: {'lr': 4.214168786625644e-05, 'samples': 20992000, 'steps': 40999, 'loss/train': 1.473726749420166}
+02/26/2022 06:39:13 - INFO - codeparrot_training - Evaluating and saving model checkpoint