diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -31488,3 +31488,1009 @@ Use FP16 precision: False
 02/25/2022 17:43:42 - INFO - codeparrot_training - Step 30998: {'lr': 0.00016967112257861721, 'samples': 15871488, 'steps': 30998, 'loss/train': 0.2525671422481537}
 02/25/2022 17:43:47 - INFO - codeparrot_training - Step 30999: {'lr': 0.00016965562795432848, 'samples': 15872000, 'steps': 30999, 'loss/train': 2.7075023651123047}
 02/25/2022 17:43:47 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/25/2022 17:44:05 - WARNING - huggingface_hub.repository - Several commits (31) will be pushed upstream.
+02/25/2022 17:44:05 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/25/2022 17:44:38 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   af58905..59d2b90  floral-grass-11 -> floral-grass-11
+
+02/25/2022 17:44:43 - INFO - codeparrot_training - Step 31000: {'lr': 0.00016964013367420965, 'samples': 15872512, 'steps': 31000, 'loss/train': 0.6951963305473328}
+02/25/2022 17:44:49 - INFO - codeparrot_training - Step 31001: {'lr': 0.00016962463973832688, 'samples': 15873024, 'steps': 31001, 'loss/train': 1.6582273244857788}
+02/25/2022 17:44:52 - INFO - codeparrot_training - Step 31002: {'lr': 0.0001696091461467466, 'samples': 15873536, 'steps': 31002, 'loss/train': 2.0780632495880127}
+02/25/2022 17:44:58 - INFO - codeparrot_training - Step 31003: {'lr': 0.00016959365289953522, 'samples': 15874048, 'steps': 31003, 'loss/train': 2.472001791000366}
+02/25/2022 17:45:01 - INFO - codeparrot_training - Step 31004: {'lr': 0.00016957815999675923, 'samples': 15874560, 'steps': 31004, 'loss/train': 0.43674641847610474}
+02/25/2022 17:45:07 - INFO - codeparrot_training - Step 31005: {'lr': 0.00016956266743848472, 'samples': 15875072, 'steps': 31005, 'loss/train': 1.7334411144256592}
+02/25/2022 17:45:10 - INFO - codeparrot_training - Step 31006: {'lr': 0.00016954717522477824, 'samples': 15875584, 'steps': 31006, 'loss/train': 2.1179020404815674}
+02/25/2022 17:45:16 - INFO - codeparrot_training - Step 31007: {'lr': 0.00016953168335570606, 'samples': 15876096, 'steps': 31007, 'loss/train': 2.2971134185791016}
+02/25/2022 17:45:19 - INFO - codeparrot_training - Step 31008: {'lr': 0.00016951619183133476, 'samples': 15876608, 'steps': 31008, 'loss/train': 1.6437169313430786}
+02/25/2022 17:45:25 - INFO - codeparrot_training - Step 31009: {'lr': 0.0001695007006517304, 'samples': 15877120, 'steps': 31009, 'loss/train': 1.7899909019470215}
+02/25/2022 17:45:28 - INFO - codeparrot_training - Step 31010: {'lr': 0.0001694852098169595, 'samples': 15877632, 'steps': 31010, 'loss/train': 2.6129343509674072}
+02/25/2022 17:45:35 - INFO - codeparrot_training - Step 31011: {'lr': 0.00016946971932708837, 'samples': 15878144, 'steps': 31011, 'loss/train': 1.7559475898742676}
+02/25/2022 17:45:38 - INFO - codeparrot_training - Step 31012: {'lr': 0.00016945422918218338, 'samples': 15878656, 'steps': 31012, 'loss/train': 1.1093412637710571}
+02/25/2022 17:45:43 - INFO - codeparrot_training - Step 31013: {'lr': 0.00016943873938231105, 'samples': 15879168, 'steps': 31013, 'loss/train': 1.1784868240356445}
+02/25/2022 17:45:47 - INFO - codeparrot_training - Step 31014: {'lr': 0.00016942324992753744, 'samples': 15879680, 'steps': 31014, 'loss/train': 1.3519665002822876}
+02/25/2022 17:45:52 - INFO - codeparrot_training - Step 31015: {'lr': 0.00016940776081792908, 'samples': 15880192, 'steps': 31015, 'loss/train': 2.262308359146118}
+02/25/2022 17:45:56 - INFO - codeparrot_training - Step 31016: {'lr': 0.0001693922720535523, 'samples': 15880704, 'steps': 31016, 'loss/train': 2.342503070831299}
+02/25/2022 17:46:02 - INFO - codeparrot_training - Step 31017: {'lr': 0.00016937678363447342, 'samples': 15881216, 'steps': 31017, 'loss/train': 8.604876518249512}
+02/25/2022 17:46:05 - INFO - codeparrot_training - Step 31018: {'lr': 0.00016936129556075876, 'samples': 15881728, 'steps': 31018, 'loss/train': 0.6884424686431885}
+02/25/2022 17:46:11 - INFO - codeparrot_training - Step 31019: {'lr': 0.00016934580783247471, 'samples': 15882240, 'steps': 31019, 'loss/train': 0.8045560121536255}
+02/25/2022 17:46:14 - INFO - codeparrot_training - Step 31020: {'lr': 0.00016933032044968765, 'samples': 15882752, 'steps': 31020, 'loss/train': 1.5906058549880981}
+02/25/2022 17:46:20 - INFO - codeparrot_training - Step 31021: {'lr': 0.00016931483341246385, 'samples': 15883264, 'steps': 31021, 'loss/train': 2.1294853687286377}
+02/25/2022 17:46:24 - INFO - codeparrot_training - Step 31022: {'lr': 0.00016929934672086971, 'samples': 15883776, 'steps': 31022, 'loss/train': 0.7153582572937012}
+02/25/2022 17:46:29 - INFO - codeparrot_training - Step 31023: {'lr': 0.00016928386037497157, 'samples': 15884288, 'steps': 31023, 'loss/train': 2.059047222137451}
+02/25/2022 17:46:33 - INFO - codeparrot_training - Step 31024: {'lr': 0.00016926837437483566, 'samples': 15884800, 'steps': 31024, 'loss/train': 1.5293407440185547}
+02/25/2022 17:46:38 - INFO - codeparrot_training - Step 31025: {'lr': 0.0001692528887205284, 'samples': 15885312, 'steps': 31025, 'loss/train': 1.914209246635437}
+02/25/2022 17:46:42 - INFO - codeparrot_training - Step 31026: {'lr': 0.0001692374034121162, 'samples': 15885824, 'steps': 31026, 'loss/train': 2.1056015491485596}
+02/25/2022 17:46:48 - INFO - codeparrot_training - Step 31027: {'lr': 0.0001692219184496653, 'samples': 15886336, 'steps': 31027, 'loss/train': 1.5960326194763184}
+02/25/2022 17:46:51 - INFO - codeparrot_training - Step 31028: {'lr': 0.00016920643383324201, 'samples': 15886848, 'steps': 31028, 'loss/train': 0.8852987289428711}
+02/25/2022 17:46:56 - INFO - codeparrot_training - Step 31029: {'lr': 0.0001691909495629127, 'samples': 15887360, 'steps': 31029, 'loss/train': 2.0812950134277344}
+02/25/2022 17:47:00 - INFO - codeparrot_training - Step 31030: {'lr': 0.00016917546563874374, 'samples': 15887872, 'steps': 31030, 'loss/train': 2.6640775203704834}
+02/25/2022 17:47:06 - INFO - codeparrot_training - Step 31031: {'lr': 0.00016915998206080147, 'samples': 15888384, 'steps': 31031, 'loss/train': 1.7001160383224487}
+02/25/2022 17:47:10 - INFO - codeparrot_training - Step 31032: {'lr': 0.00016914449882915206, 'samples': 15888896, 'steps': 31032, 'loss/train': 2.2505905628204346}
+02/25/2022 17:47:15 - INFO - codeparrot_training - Step 31033: {'lr': 0.00016912901594386197, 'samples': 15889408, 'steps': 31033, 'loss/train': 1.3172106742858887}
+02/25/2022 17:47:19 - INFO - codeparrot_training - Step 31034: {'lr': 0.00016911353340499748, 'samples': 15889920, 'steps': 31034, 'loss/train': 1.4276812076568604}
+02/25/2022 17:47:24 - INFO - codeparrot_training - Step 31035: {'lr': 0.00016909805121262506, 'samples': 15890432, 'steps': 31035, 'loss/train': 0.8314815163612366}
+02/25/2022 17:47:28 - INFO - codeparrot_training - Step 31036: {'lr': 0.00016908256936681076, 'samples': 15890944, 'steps': 31036, 'loss/train': 2.0956528186798096}
+02/25/2022 17:47:33 - INFO - codeparrot_training - Step 31037: {'lr': 0.00016906708786762104, 'samples': 15891456, 'steps': 31037, 'loss/train': 1.2844158411026}
+02/25/2022 17:47:37 - INFO - codeparrot_training - Step 31038: {'lr': 0.00016905160671512227, 'samples': 15891968, 'steps': 31038, 'loss/train': 1.5763933658599854}
+02/25/2022 17:47:42 - INFO - codeparrot_training - Step 31039: {'lr': 0.00016903612590938075, 'samples': 15892480, 'steps': 31039, 'loss/train': 1.890817403793335}
+02/25/2022 17:47:46 - INFO - codeparrot_training - Step 31040: {'lr': 0.0001690206454504627, 'samples': 15892992, 'steps': 31040, 'loss/train': 2.108973264694214}
+02/25/2022 17:47:51 - INFO - codeparrot_training - Step 31041: {'lr': 0.00016900516533843447, 'samples': 15893504, 'steps': 31041, 'loss/train': 1.5586225986480713}
+02/25/2022 17:47:55 - INFO - codeparrot_training - Step 31042: {'lr': 0.00016898968557336238, 'samples': 15894016, 'steps': 31042, 'loss/train': 2.1700501441955566}
+02/25/2022 17:48:01 - INFO - codeparrot_training - Step 31043: {'lr': 0.00016897420615531289, 'samples': 15894528, 'steps': 31043, 'loss/train': 2.3337037563323975}
+02/25/2022 17:48:04 - INFO - codeparrot_training - Step 31044: {'lr': 0.00016895872708435207, 'samples': 15895040, 'steps': 31044, 'loss/train': 2.0221059322357178}
+02/25/2022 17:48:10 - INFO - codeparrot_training - Step 31045: {'lr': 0.00016894324836054628, 'samples': 15895552, 'steps': 31045, 'loss/train': 1.4884897470474243}
+02/25/2022 17:48:13 - INFO - codeparrot_training - Step 31046: {'lr': 0.00016892776998396193, 'samples': 15896064, 'steps': 31046, 'loss/train': 1.1874433755874634}
+02/25/2022 17:48:19 - INFO - codeparrot_training - Step 31047: {'lr': 0.00016891229195466524, 'samples': 15896576, 'steps': 31047, 'loss/train': 1.7352169752120972}
+02/25/2022 17:48:24 - INFO - codeparrot_training - Step 31048: {'lr': 0.00016889681427272263, 'samples': 15897088, 'steps': 31048, 'loss/train': 1.9792522192001343}
+02/25/2022 17:48:28 - INFO - codeparrot_training - Step 31049: {'lr': 0.00016888133693820023, 'samples': 15897600, 'steps': 31049, 'loss/train': 1.6084520816802979}
+02/25/2022 17:48:33 - INFO - codeparrot_training - Step 31050: {'lr': 0.00016886585995116443, 'samples': 15898112, 'steps': 31050, 'loss/train': 1.772125482559204}
+02/25/2022 17:48:37 - INFO - codeparrot_training - Step 31051: {'lr': 0.00016885038331168153, 'samples': 15898624, 'steps': 31051, 'loss/train': 2.145721912384033}
+02/25/2022 17:48:43 - INFO - codeparrot_training - Step 31052: {'lr': 0.0001688349070198179, 'samples': 15899136, 'steps': 31052, 'loss/train': 2.179814338684082}
+02/25/2022 17:48:46 - INFO - codeparrot_training - Step 31053: {'lr': 0.00016881943107563969, 'samples': 15899648, 'steps': 31053, 'loss/train': 2.506361961364746}
+02/25/2022 17:48:49 - INFO - codeparrot_training - Step 31054: {'lr': 0.00016880395547921322, 'samples': 15900160, 'steps': 31054, 'loss/train': 1.9710701704025269}
+02/25/2022 17:48:55 - INFO - codeparrot_training - Step 31055: {'lr': 0.00016878848023060486, 'samples': 15900672, 'steps': 31055, 'loss/train': 1.7726755142211914}
+02/25/2022 17:48:58 - INFO - codeparrot_training - Step 31056: {'lr': 0.00016877300532988092, 'samples': 15901184, 'steps': 31056, 'loss/train': 1.1743453741073608}
+02/25/2022 17:49:05 - INFO - codeparrot_training - Step 31057: {'lr': 0.0001687575307771076, 'samples': 15901696, 'steps': 31057, 'loss/train': 1.358697533607483}
+02/25/2022 17:49:11 - INFO - codeparrot_training - Step 31058: {'lr': 0.00016874205657235115, 'samples': 15902208, 'steps': 31058, 'loss/train': 2.1262764930725098}
+02/25/2022 17:49:14 - INFO - codeparrot_training - Step 31059: {'lr': 0.00016872658271567794, 'samples': 15902720, 'steps': 31059, 'loss/train': 1.8100907802581787}
+02/25/2022 17:49:20 - INFO - codeparrot_training - Step 31060: {'lr': 0.00016871110920715432, 'samples': 15903232, 'steps': 31060, 'loss/train': 1.5029383897781372}
+02/25/2022 17:49:23 - INFO - codeparrot_training - Step 31061: {'lr': 0.0001686956360468465, 'samples': 15903744, 'steps': 31061, 'loss/train': 1.6959291696548462}
+02/25/2022 17:49:29 - INFO - codeparrot_training - Step 31062: {'lr': 0.00016868016323482067, 'samples': 15904256, 'steps': 31062, 'loss/train': 2.564382791519165}
+02/25/2022 17:49:32 - INFO - codeparrot_training - Step 31063: {'lr': 0.0001686646907711432, 'samples': 15904768, 'steps': 31063, 'loss/train': 1.7366644144058228}
+02/25/2022 17:49:38 - INFO - codeparrot_training - Step 31064: {'lr': 0.00016864921865588045, 'samples': 15905280, 'steps': 31064, 'loss/train': 2.0076510906219482}
+02/25/2022 17:49:41 - INFO - codeparrot_training - Step 31065: {'lr': 0.0001686337468890986, 'samples': 15905792, 'steps': 31065, 'loss/train': 2.0645558834075928}
+02/25/2022 17:49:47 - INFO - codeparrot_training - Step 31066: {'lr': 0.00016861827547086388, 'samples': 15906304, 'steps': 31066, 'loss/train': 2.4808075428009033}
+02/25/2022 17:49:50 - INFO - codeparrot_training - Step 31067: {'lr': 0.0001686028044012427, 'samples': 15906816, 'steps': 31067, 'loss/train': 1.8953602313995361}
+02/25/2022 17:49:56 - INFO - codeparrot_training - Step 31068: {'lr': 0.0001685873336803012, 'samples': 15907328, 'steps': 31068, 'loss/train': 1.5350229740142822}
+02/25/2022 17:50:00 - INFO - codeparrot_training - Step 31069: {'lr': 0.00016857186330810573, 'samples': 15907840, 'steps': 31069, 'loss/train': 1.3350399732589722}
+02/25/2022 17:50:05 - INFO - codeparrot_training - Step 31070: {'lr': 0.00016855639328472257, 'samples': 15908352, 'steps': 31070, 'loss/train': 1.7629777193069458}
+02/25/2022 17:50:09 - INFO - codeparrot_training - Step 31071: {'lr': 0.00016854092361021797, 'samples': 15908864, 'steps': 31071, 'loss/train': 0.61308354139328}
+02/25/2022 17:50:14 - INFO - codeparrot_training - Step 31072: {'lr': 0.0001685254542846581, 'samples': 15909376, 'steps': 31072, 'loss/train': 1.1078263521194458}
+02/25/2022 17:50:18 - INFO - codeparrot_training - Step 31073: {'lr': 0.00016850998530810934, 'samples': 15909888, 'steps': 31073, 'loss/train': 1.7525444030761719}
+02/25/2022 17:50:24 - INFO - codeparrot_training - Step 31074: {'lr': 0.00016849451668063807, 'samples': 15910400, 'steps': 31074, 'loss/train': 2.3823907375335693}
+02/25/2022 17:50:27 - INFO - codeparrot_training - Step 31075: {'lr': 0.00016847904840231026, 'samples': 15910912, 'steps': 31075, 'loss/train': 3.845890522003174}
+02/25/2022 17:50:33 - INFO - codeparrot_training - Step 31076: {'lr': 0.00016846358047319232, 'samples': 15911424, 'steps': 31076, 'loss/train': 1.3276636600494385}
+02/25/2022 17:50:37 - INFO - codeparrot_training - Step 31077: {'lr': 0.0001684481128933505, 'samples': 15911936, 'steps': 31077, 'loss/train': 1.5635157823562622}
+02/25/2022 17:50:43 - INFO - codeparrot_training - Step 31078: {'lr': 0.00016843264566285122, 'samples': 15912448, 'steps': 31078, 'loss/train': 0.15091058611869812}
+02/25/2022 17:50:46 - INFO - codeparrot_training - Step 31079: {'lr': 0.00016841717878176044, 'samples': 15912960, 'steps': 31079, 'loss/train': 2.4018688201904297}
+02/25/2022 17:50:51 - INFO - codeparrot_training - Step 31080: {'lr': 0.0001684017122501446, 'samples': 15913472, 'steps': 31080, 'loss/train': 0.5490176677703857}
+02/25/2022 17:50:55 - INFO - codeparrot_training - Step 31081: {'lr': 0.0001683862460680699, 'samples': 15913984, 'steps': 31081, 'loss/train': 2.7552993297576904}
+02/25/2022 17:51:01 - INFO - codeparrot_training - Step 31082: {'lr': 0.00016837078023560257, 'samples': 15914496, 'steps': 31082, 'loss/train': 1.705405592918396}
+02/25/2022 17:51:04 - INFO - codeparrot_training - Step 31083: {'lr': 0.00016835531475280902, 'samples': 15915008, 'steps': 31083, 'loss/train': 2.1103529930114746}
+02/25/2022 17:51:09 - INFO - codeparrot_training - Step 31084: {'lr': 0.0001683398496197553, 'samples': 15915520, 'steps': 31084, 'loss/train': 2.0248754024505615}
+02/25/2022 17:51:13 - INFO - codeparrot_training - Step 31085: {'lr': 0.00016832438483650768, 'samples': 15916032, 'steps': 31085, 'loss/train': 1.7194762229919434}
+02/25/2022 17:51:19 - INFO - codeparrot_training - Step 31086: {'lr': 0.0001683089204031325, 'samples': 15916544, 'steps': 31086, 'loss/train': 1.9844094514846802}
+02/25/2022 17:51:22 - INFO - codeparrot_training - Step 31087: {'lr': 0.00016829345631969606, 'samples': 15917056, 'steps': 31087, 'loss/train': 1.8127073049545288}
+02/25/2022 17:51:28 - INFO - codeparrot_training - Step 31088: {'lr': 0.00016827799258626442, 'samples': 15917568, 'steps': 31088, 'loss/train': 1.471313238143921}
+02/25/2022 17:51:31 - INFO - codeparrot_training - Step 31089: {'lr': 0.0001682625292029039, 'samples': 15918080, 'steps': 31089, 'loss/train': 1.976351022720337}
+02/25/2022 17:51:37 - INFO - codeparrot_training - Step 31090: {'lr': 0.00016824706616968074, 'samples': 15918592, 'steps': 31090, 'loss/train': 1.6865428686141968}
+02/25/2022 17:51:40 - INFO - codeparrot_training - Step 31091: {'lr': 0.0001682316034866613, 'samples': 15919104, 'steps': 31091, 'loss/train': 2.200965642929077}
+02/25/2022 17:51:46 - INFO - codeparrot_training - Step 31092: {'lr': 0.0001682161411539116, 'samples': 15919616, 'steps': 31092, 'loss/train': 0.759330153465271}
+02/25/2022 17:51:50 - INFO - codeparrot_training - Step 31093: {'lr': 0.000168200679171498, 'samples': 15920128, 'steps': 31093, 'loss/train': 1.1216405630111694}
+02/25/2022 17:51:56 - INFO - codeparrot_training - Step 31094: {'lr': 0.00016818521753948667, 'samples': 15920640, 'steps': 31094, 'loss/train': 1.7577195167541504}
+02/25/2022 17:51:59 - INFO - codeparrot_training - Step 31095: {'lr': 0.00016816975625794396, 'samples': 15921152, 'steps': 31095, 'loss/train': 1.0197408199310303}
+02/25/2022 17:52:05 - INFO - codeparrot_training - Step 31096: {'lr': 0.00016815429532693609, 'samples': 15921664, 'steps': 31096, 'loss/train': 1.8471494913101196}
+02/25/2022 17:52:08 - INFO - codeparrot_training - Step 31097: {'lr': 0.0001681388347465291, 'samples': 15922176, 'steps': 31097, 'loss/train': 1.7075550556182861}
+02/25/2022 17:52:14 - INFO - codeparrot_training - Step 31098: {'lr': 0.00016812337451678938, 'samples': 15922688, 'steps': 31098, 'loss/train': 1.3168261051177979}
+02/25/2022 17:52:17 - INFO - codeparrot_training - Step 31099: {'lr': 0.00016810791463778312, 'samples': 15923200, 'steps': 31099, 'loss/train': 1.2030969858169556}
+02/25/2022 17:52:23 - INFO - codeparrot_training - Step 31100: {'lr': 0.00016809245510957666, 'samples': 15923712, 'steps': 31100, 'loss/train': 1.1703046560287476}
+02/25/2022 17:52:26 - INFO - codeparrot_training - Step 31101: {'lr': 0.000168076995932236, 'samples': 15924224, 'steps': 31101, 'loss/train': 2.145098924636841}
+02/25/2022 17:52:32 - INFO - codeparrot_training - Step 31102: {'lr': 0.0001680615371058275, 'samples': 15924736, 'steps': 31102, 'loss/train': 0.5684425830841064}
+02/25/2022 17:52:35 - INFO - codeparrot_training - Step 31103: {'lr': 0.00016804607863041732, 'samples': 15925248, 'steps': 31103, 'loss/train': 1.3232879638671875}
+02/25/2022 17:52:41 - INFO - codeparrot_training - Step 31104: {'lr': 0.00016803062050607187, 'samples': 15925760, 'steps': 31104, 'loss/train': 1.423197627067566}
+02/25/2022 17:52:45 - INFO - codeparrot_training - Step 31105: {'lr': 0.00016801516273285704, 'samples': 15926272, 'steps': 31105, 'loss/train': 1.9002509117126465}
+02/25/2022 17:52:50 - INFO - codeparrot_training - Step 31106: {'lr': 0.00016799970531083928, 'samples': 15926784, 'steps': 31106, 'loss/train': 1.266247272491455}
+02/25/2022 17:52:54 - INFO - codeparrot_training - Step 31107: {'lr': 0.0001679842482400848, 'samples': 15927296, 'steps': 31107, 'loss/train': 2.3030006885528564}
+02/25/2022 17:52:59 - INFO - codeparrot_training - Step 31108: {'lr': 0.00016796879152065965, 'samples': 15927808, 'steps': 31108, 'loss/train': 2.0188961029052734}
+02/25/2022 17:53:03 - INFO - codeparrot_training - Step 31109: {'lr': 0.0001679533351526302, 'samples': 15928320, 'steps': 31109, 'loss/train': 1.8462263345718384}
+02/25/2022 17:53:08 - INFO - codeparrot_training - Step 31110: {'lr': 0.00016793787913606262, 'samples': 15928832, 'steps': 31110, 'loss/train': 2.0196850299835205}
+02/25/2022 17:53:12 - INFO - codeparrot_training - Step 31111: {'lr': 0.00016792242347102312, 'samples': 15929344, 'steps': 31111, 'loss/train': 1.6382567882537842}
+02/25/2022 17:53:17 - INFO - codeparrot_training - Step 31112: {'lr': 0.00016790696815757787, 'samples': 15929856, 'steps': 31112, 'loss/train': 1.5102472305297852}
+02/25/2022 17:53:21 - INFO - codeparrot_training - Step 31113: {'lr': 0.00016789151319579314, 'samples': 15930368, 'steps': 31113, 'loss/train': 0.6794791221618652}
+02/25/2022 17:53:27 - INFO - codeparrot_training - Step 31114: {'lr': 0.00016787605858573508, 'samples': 15930880, 'steps': 31114, 'loss/train': 1.8656983375549316}
+02/25/2022 17:53:30 - INFO - codeparrot_training - Step 31115: {'lr': 0.00016786060432746986, 'samples': 15931392, 'steps': 31115, 'loss/train': 1.166294813156128}
+02/25/2022 17:53:36 - INFO - codeparrot_training - Step 31116: {'lr': 0.00016784515042106375, 'samples': 15931904, 'steps': 31116, 'loss/train': 1.1743524074554443}
+02/25/2022 17:53:39 - INFO - codeparrot_training - Step 31117: {'lr': 0.0001678296968665829, 'samples': 15932416, 'steps': 31117, 'loss/train': 1.6533632278442383}
+02/25/2022 17:53:45 - INFO - codeparrot_training - Step 31118: {'lr': 0.00016781424366409364, 'samples': 15932928, 'steps': 31118, 'loss/train': 2.6918015480041504}
+02/25/2022 17:53:48 - INFO - codeparrot_training - Step 31119: {'lr': 0.00016779879081366201, 'samples': 15933440, 'steps': 31119, 'loss/train': 0.9125258922576904}
+02/25/2022 17:53:54 - INFO - codeparrot_training - Step 31120: {'lr': 0.00016778333831535418, 'samples': 15933952, 'steps': 31120, 'loss/train': 1.5292145013809204}
+02/25/2022 17:53:57 - INFO - codeparrot_training - Step 31121: {'lr': 0.0001677678861692365, 'samples': 15934464, 'steps': 31121, 'loss/train': 3.9622302055358887}
+02/25/2022 17:54:03 - INFO - codeparrot_training - Step 31122: {'lr': 0.00016775243437537512, 'samples': 15934976, 'steps': 31122, 'loss/train': 2.188889980316162}
+02/25/2022 17:54:06 - INFO - codeparrot_training - Step 31123: {'lr': 0.00016773698293383617, 'samples': 15935488, 'steps': 31123, 'loss/train': 0.6540660262107849}
+02/25/2022 17:54:13 - INFO - codeparrot_training - Step 31124: {'lr': 0.0001677215318446858, 'samples': 15936000, 'steps': 31124, 'loss/train': 1.6189193725585938}
+02/25/2022 17:54:16 - INFO - codeparrot_training - Step 31125: {'lr': 0.00016770608110799028, 'samples': 15936512, 'steps': 31125, 'loss/train': 1.781721830368042}
+02/25/2022 17:54:22 - INFO - codeparrot_training - Step 31126: {'lr': 0.00016769063072381586, 'samples': 15937024, 'steps': 31126, 'loss/train': 2.638826608657837}
+02/25/2022 17:54:25 - INFO - codeparrot_training - Step 31127: {'lr': 0.00016767518069222856, 'samples': 15937536, 'steps': 31127, 'loss/train': 1.7111948728561401}
+02/25/2022 17:54:31 - INFO - codeparrot_training - Step 31128: {'lr': 0.00016765973101329462, 'samples': 15938048, 'steps': 31128, 'loss/train': 1.5463006496429443}
+02/25/2022 17:54:34 - INFO - codeparrot_training - Step 31129: {'lr': 0.0001676442816870803, 'samples': 15938560, 'steps': 31129, 'loss/train': 2.2380406856536865}
+02/25/2022 17:54:40 - INFO - codeparrot_training - Step 31130: {'lr': 0.00016762883271365168, 'samples': 15939072, 'steps': 31130, 'loss/train': 1.604448676109314}
+02/25/2022 17:54:43 - INFO - codeparrot_training - Step 31131: {'lr': 0.00016761338409307508, 'samples': 15939584, 'steps': 31131, 'loss/train': 1.9851630926132202}
+02/25/2022 17:54:49 - INFO - codeparrot_training - Step 31132: {'lr': 0.00016759793582541652, 'samples': 15940096, 'steps': 31132, 'loss/train': 1.4262844324111938}
+02/25/2022 17:54:52 - INFO - codeparrot_training - Step 31133: {'lr': 0.0001675824879107422, 'samples': 15940608, 'steps': 31133, 'loss/train': 1.7848807573318481}
+02/25/2022 17:54:59 - INFO - codeparrot_training - Step 31134: {'lr': 0.00016756704034911834, 'samples': 15941120, 'steps': 31134, 'loss/train': 0.8311417698860168}
+02/25/2022 17:55:02 - INFO - codeparrot_training - Step 31135: {'lr': 0.00016755159314061125, 'samples': 15941632, 'steps': 31135, 'loss/train': 2.46219801902771}
+02/25/2022 17:55:08 - INFO - codeparrot_training - Step 31136: {'lr': 0.00016753614628528678, 'samples': 15942144, 'steps': 31136, 'loss/train': 1.780564308166504}
+02/25/2022 17:55:11 - INFO - codeparrot_training - Step 31137: {'lr': 0.00016752069978321133, 'samples': 15942656, 'steps': 31137, 'loss/train': 1.6061509847640991}
+02/25/2022 17:55:17 - INFO - codeparrot_training - Step 31138: {'lr': 0.000167505253634451, 'samples': 15943168, 'steps': 31138, 'loss/train': 1.6682584285736084}
+02/25/2022 17:55:20 - INFO - codeparrot_training - Step 31139: {'lr': 0.00016748980783907208, 'samples': 15943680, 'steps': 31139, 'loss/train': 0.3699662387371063}
+02/25/2022 17:55:25 - INFO - codeparrot_training - Step 31140: {'lr': 0.00016747436239714048, 'samples': 15944192, 'steps': 31140, 'loss/train': 1.5237154960632324}
+02/25/2022 17:55:29 - INFO - codeparrot_training - Step 31141: {'lr': 0.00016745891730872255, 'samples': 15944704, 'steps': 31141, 'loss/train': 2.326125383377075}
+02/25/2022 17:55:35 - INFO - codeparrot_training - Step 31142: {'lr': 0.00016744347257388438, 'samples': 15945216, 'steps': 31142, 'loss/train': 1.4061886072158813}
+02/25/2022 17:55:38 - INFO - codeparrot_training - Step 31143: {'lr': 0.00016742802819269217, 'samples': 15945728, 'steps': 31143, 'loss/train': 1.7516627311706543}
+02/25/2022 17:55:44 - INFO - codeparrot_training - Step 31144: {'lr': 0.00016741258416521216, 'samples': 15946240, 'steps': 31144, 'loss/train': 1.2640248537063599}
+02/25/2022 17:55:49 - INFO - codeparrot_training - Step 31145: {'lr': 0.00016739714049151034, 'samples': 15946752, 'steps': 31145, 'loss/train': 1.9980137348175049}
+02/25/2022 17:55:53 - INFO - codeparrot_training - Step 31146: {'lr': 0.0001673816971716529, 'samples': 15947264, 'steps': 31146, 'loss/train': 1.8235807418823242}
+02/25/2022 17:55:59 - INFO - codeparrot_training - Step 31147: {'lr': 0.0001673662542057061, 'samples': 15947776, 'steps': 31147, 'loss/train': 1.3574975728988647}
+02/25/2022 17:56:03 - INFO - codeparrot_training - Step 31148: {'lr': 0.00016735081159373604, 'samples': 15948288, 'steps': 31148, 'loss/train': 2.2431135177612305}
+02/25/2022 17:56:08 - INFO - codeparrot_training - Step 31149: {'lr': 0.00016733536933580885, 'samples': 15948800, 'steps': 31149, 'loss/train': 1.345247745513916}
+02/25/2022 17:56:12 - INFO - codeparrot_training - Step 31150: {'lr': 0.0001673199274319906, 'samples': 15949312, 'steps': 31150, 'loss/train': 0.8538775444030762}
+02/25/2022 17:56:17 - INFO - codeparrot_training - Step 31151: {'lr': 0.00016730448588234766, 'samples': 15949824, 'steps': 31151, 'loss/train': 1.5928668975830078}
+02/25/2022 17:56:21 - INFO - codeparrot_training - Step 31152: {'lr': 0.000167289044686946, 'samples': 15950336, 'steps': 31152, 'loss/train': 1.114976406097412}
+02/25/2022 17:56:26 - INFO - codeparrot_training - Step 31153: {'lr': 0.00016727360384585178, 'samples': 15950848, 'steps': 31153, 'loss/train': 2.363546133041382}
+02/25/2022 17:56:30 - INFO - codeparrot_training - Step 31154: {'lr': 0.0001672581633591312, 'samples': 15951360, 'steps': 31154, 'loss/train': 1.1886793375015259}
+02/25/2022 17:56:36 - INFO - codeparrot_training - Step 31155: {'lr': 0.0001672427232268503, 'samples': 15951872, 'steps': 31155, 'loss/train': 2.035230875015259}
+02/25/2022 17:56:39 - INFO - codeparrot_training - Step 31156: {'lr': 0.00016722728344907534, 'samples': 15952384, 'steps': 31156, 'loss/train': 1.4630767107009888}
+02/25/2022 17:56:45 - INFO - codeparrot_training - Step 31157: {'lr': 0.00016721184402587245, 'samples': 15952896, 'steps': 31157, 'loss/train': 1.1839317083358765}
+02/25/2022 17:56:49 - INFO - codeparrot_training - Step 31158: {'lr': 0.00016719640495730772, 'samples': 15953408, 'steps': 31158, 'loss/train': 1.7174227237701416}
+02/25/2022 17:56:54 - INFO - codeparrot_training - Step 31159: {'lr': 0.00016718096624344724, 'samples': 15953920, 'steps': 31159, 'loss/train': 1.1421518325805664}
+02/25/2022 17:56:58 - INFO - codeparrot_training - Step 31160: {'lr': 0.00016716552788435723, 'samples': 15954432, 'steps': 31160, 'loss/train': 1.5369526147842407}
+02/25/2022 17:57:03 - INFO - codeparrot_training - Step 31161: {'lr': 0.0001671500898801039, 'samples': 15954944, 'steps': 31161, 'loss/train': 1.771902084350586}
+02/25/2022 17:57:07 - INFO - codeparrot_training - Step 31162: {'lr': 0.0001671346522307532, 'samples': 15955456, 'steps': 31162, 'loss/train': 2.038527011871338}
+02/25/2022 17:57:12 - INFO - codeparrot_training - Step 31163: {'lr': 0.00016711921493637127, 'samples': 15955968, 'steps': 31163, 'loss/train': 1.1051788330078125}
+02/25/2022 17:57:16 - INFO - codeparrot_training - Step 31164: {'lr': 0.00016710377799702438, 'samples': 15956480, 'steps': 31164, 'loss/train': 1.2831100225448608}
+02/25/2022 17:57:21 - INFO - codeparrot_training - Step 31165: {'lr': 0.00016708834141277863, 'samples': 15956992, 'steps': 31165, 'loss/train': 1.8582972288131714}
+02/25/2022 17:57:25 - INFO - codeparrot_training - Step 31166: {'lr': 0.00016707290518369998, 'samples': 15957504, 'steps': 31166, 'loss/train': 1.6135523319244385}
+02/25/2022 17:57:31 - INFO - codeparrot_training - Step 31167: {'lr': 0.00016705746930985473, 'samples': 15958016, 'steps': 31167, 'loss/train': 1.0429036617279053}
+02/25/2022 17:57:34 - INFO - codeparrot_training - Step 31168: {'lr': 0.0001670420337913089, 'samples': 15958528, 'steps': 31168, 'loss/train': 1.5927619934082031}
+02/25/2022 17:57:40 - INFO - codeparrot_training - Step 31169: {'lr': 0.00016702659862812867, 'samples': 15959040, 'steps': 31169, 'loss/train': 1.475658893585205}
+02/25/2022 17:57:43 - INFO - codeparrot_training - Step 31170: {'lr': 0.00016701116382038023, 'samples': 15959552, 'steps': 31170, 'loss/train': 1.6317473649978638}
+02/25/2022 17:57:49 - INFO - codeparrot_training - Step 31171: {'lr': 0.00016699572936812953, 'samples': 15960064, 'steps': 31171, 'loss/train': 1.924875020980835}
+02/25/2022 17:57:52 - INFO - codeparrot_training - Step 31172: {'lr': 0.00016698029527144277, 'samples': 15960576, 'steps': 31172, 'loss/train': 1.1838500499725342}
+02/25/2022 17:57:58 - INFO - codeparrot_training - Step 31173: {'lr': 0.00016696486153038606, 'samples': 15961088, 'steps': 31173, 'loss/train': 1.4967623949050903}
+02/25/2022 17:58:01 - INFO - codeparrot_training - Step 31174: {'lr': 0.00016694942814502562, 'samples': 15961600, 'steps': 31174, 'loss/train': 1.145328164100647}
+02/25/2022 17:58:07 - INFO - codeparrot_training - Step 31175: {'lr': 0.00016693399511542734, 'samples': 15962112, 'steps': 31175, 'loss/train': 2.7329673767089844}
+02/25/2022 17:58:10 - INFO - codeparrot_training - Step 31176: {'lr': 0.0001669185624416575, 'samples': 15962624, 'steps': 31176, 'loss/train': 1.8396117687225342}
+02/25/2022 17:58:16 - INFO - codeparrot_training - Step 31177: {'lr': 0.00016690313012378212, 'samples': 15963136, 'steps': 31177, 'loss/train': 1.646539568901062}
+02/25/2022 17:58:20 - INFO - codeparrot_training - Step 31178: {'lr': 0.00016688769816186732, 'samples': 15963648, 'steps': 31178, 'loss/train': 1.458148717880249}
+02/25/2022 17:58:25 - INFO - codeparrot_training - Step 31179: {'lr': 0.00016687226655597937, 'samples': 15964160, 'steps': 31179, 'loss/train': 2.8021912574768066}
+02/25/2022 17:58:29 - INFO - codeparrot_training - Step 31180: {'lr': 0.00016685683530618411, 'samples': 15964672, 'steps': 31180, 'loss/train': 0.9809687733650208}
+02/25/2022 17:58:34 - INFO - codeparrot_training - Step 31181: {'lr': 0.00016684140441254776, 'samples': 15965184, 'steps': 31181, 'loss/train': 0.04575040191411972}
+02/25/2022 17:58:38 - INFO - codeparrot_training - Step 31182: {'lr': 0.00016682597387513643, 'samples': 15965696, 'steps': 31182, 'loss/train': 1.824276089668274}
+02/25/2022 17:58:43 - INFO - codeparrot_training - Step 31183: {'lr': 0.00016681054369401636, 'samples': 15966208, 'steps': 31183, 'loss/train': 2.458483934402466}
+02/25/2022 17:58:47 - INFO - codeparrot_training - Step 31184: {'lr': 0.00016679511386925337, 'samples': 15966720, 'steps': 31184, 'loss/train': 2.017644166946411}
+02/25/2022 17:58:52 - INFO - codeparrot_training - Step 31185: {'lr': 0.00016677968440091366, 'samples': 15967232, 'steps': 31185, 'loss/train': 1.9451687335968018}
+02/25/2022 17:58:56 - INFO - codeparrot_training - Step 31186: {'lr': 0.00016676425528906343, 'samples': 15967744, 'steps': 31186, 'loss/train': 1.6944403648376465}
+02/25/2022 17:59:01 - INFO - codeparrot_training - Step 31187: {'lr': 0.00016674882653376877, 'samples': 15968256, 'steps': 31187, 'loss/train': 1.6143730878829956}
+02/25/2022 17:59:05 - INFO - codeparrot_training - Step 31188: {'lr': 0.00016673339813509557, 'samples': 15968768, 'steps': 31188, 'loss/train': 2.245373010635376}
+02/25/2022 17:59:11 - INFO - codeparrot_training - Step 31189: {'lr': 0.00016671797009311007, 'samples': 15969280, 'steps': 31189, 'loss/train': 1.9551475048065186}
+02/25/2022 17:59:17 - INFO - codeparrot_training - Step 31190: {'lr': 0.00016670254240787834, 'samples': 15969792, 'steps': 31190, 'loss/train': 2.5374176502227783}
+02/25/2022 17:59:20 - INFO - codeparrot_training - Step 31191: {'lr': 0.00016668711507946654, 'samples': 15970304, 'steps': 31191, 'loss/train': 1.3559492826461792}
+02/25/2022 17:59:25 - INFO - codeparrot_training - Step 31192: {'lr': 0.0001666716881079407, 'samples': 15970816, 'steps': 31192, 'loss/train': 1.6970767974853516}
+02/25/2022 17:59:29 - INFO - codeparrot_training - Step 31193: {'lr': 0.0001666562614933668, 'samples': 15971328, 'steps': 31193, 'loss/train': 0.76556396484375}
+02/25/2022 17:59:35 - INFO - codeparrot_training - Step 31194: {'lr': 0.00016664083523581106, 'samples': 15971840, 'steps': 31194, 'loss/train': 0.8046711087226868}
+02/25/2022 17:59:38 - INFO - codeparrot_training - Step 31195: {'lr': 0.0001666254093353395, 'samples': 15972352, 'steps': 31195, 'loss/train': 1.5861634016036987}
+02/25/2022 17:59:43 - INFO - codeparrot_training - Step 31196: {'lr': 0.0001666099837920182, 'samples': 15972864, 'steps': 31196, 'loss/train': 1.3574225902557373}
+02/25/2022 17:59:47 - INFO - codeparrot_training - Step 31197: {'lr': 0.00016659455860591323, 'samples': 15973376, 'steps': 31197, 'loss/train': 1.2641143798828125}
+02/25/2022 17:59:52 - INFO - codeparrot_training - Step 31198: {'lr': 0.00016657913377709078, 'samples': 15973888, 'steps': 31198, 'loss/train': 1.5527794361114502}
+02/25/2022 17:59:56 - INFO - codeparrot_training - Step 31199: {'lr': 0.0001665637093056167, 'samples': 15974400, 'steps': 31199, 'loss/train': 1.507619023323059}
+02/25/2022 18:00:02 - INFO - codeparrot_training - Step 31200: {'lr': 0.00016654828519155728, 'samples': 15974912, 'steps': 31200, 'loss/train': 1.67177152633667}
+02/25/2022 18:00:05 - INFO - codeparrot_training - Step 31201: {'lr': 0.00016653286143497852, 'samples': 15975424, 'steps': 31201, 'loss/train': 1.2310543060302734}
+02/25/2022 18:00:11 - INFO - codeparrot_training - Step 31202: {'lr': 0.00016651743803594644, 'samples': 15975936, 'steps': 31202, 'loss/train': 1.9363353252410889}
+02/25/2022 18:00:14 - INFO - codeparrot_training - Step 31203: {'lr': 0.00016650201499452714, 'samples': 15976448, 'steps': 31203, 'loss/train': 1.2398371696472168}
+02/25/2022 18:00:18 - INFO - codeparrot_training - Step 31204: {'lr': 0.00016648659231078666, 'samples': 15976960, 'steps': 31204, 'loss/train': 0.5455965399742126}
+02/25/2022 18:00:24 - INFO - codeparrot_training - Step 31205: {'lr': 0.00016647116998479123, 'samples': 15977472, 'steps': 31205, 'loss/train': 0.49342748522758484}
+02/25/2022 18:00:28 - INFO - codeparrot_training - Step 31206: {'lr': 0.00016645574801660667, 'samples': 15977984, 'steps': 31206, 'loss/train': 2.770235061645508}
+02/25/2022 18:00:33 - INFO - codeparrot_training - Step 31207: {'lr': 0.0001664403264062992, 'samples': 15978496, 'steps': 31207, 'loss/train': 2.9392635822296143}
+02/25/2022 18:00:36 - INFO - codeparrot_training - Step 31208: {'lr': 0.0001664249051539348, 'samples': 15979008, 'steps': 31208, 'loss/train': 1.678147554397583}
+02/25/2022 18:00:42 - INFO - codeparrot_training - Step 31209: {'lr': 0.00016640948425957966, 'samples': 15979520, 'steps': 31209, 'loss/train': 0.8570581674575806}
+02/25/2022 18:00:46 - INFO - codeparrot_training - Step 31210: {'lr': 0.0001663940637232997, 'samples': 15980032, 'steps': 31210, 'loss/train': 1.4423526525497437}
+02/25/2022 18:00:51 - INFO - codeparrot_training - Step 31211: {'lr': 0.00016637864354516097, 'samples': 15980544, 'steps': 31211, 'loss/train': 1.9763888120651245}
+02/25/2022 18:00:55 - INFO - codeparrot_training - Step 31212: {'lr': 0.00016636322372522955, 'samples': 15981056, 'steps': 31212, 'loss/train': 1.3488616943359375}
+02/25/2022 18:01:01 - INFO - codeparrot_training - Step 31213: {'lr': 0.0001663478042635717, 'samples': 15981568, 'steps': 31213, 'loss/train': 1.6343997716903687}
+02/25/2022 18:01:04 - INFO - codeparrot_training - Step 31214: {'lr': 0.00016633238516025316, 'samples': 15982080, 'steps': 31214, 'loss/train': 1.4659315347671509}
+02/25/2022 18:01:10 - INFO - codeparrot_training - Step 31215: {'lr': 0.0001663169664153401, 'samples': 15982592, 'steps': 31215, 'loss/train': 1.0421639680862427}
+02/25/2022 18:01:15 - INFO - codeparrot_training - Step 31216: {'lr': 0.00016630154802889857, 'samples': 15983104, 'steps': 31216, 'loss/train': 3.022994041442871}
+02/25/2022 18:01:19 - INFO - codeparrot_training - Step 31217: {'lr': 0.0001662861300009947, 'samples': 15983616, 'steps': 31217, 'loss/train': 0.4751278758049011}
+02/25/2022 18:01:24 - INFO - codeparrot_training - Step 31218: {'lr': 0.0001662707123316945, 'samples': 15984128, 'steps': 31218, 'loss/train': 1.8457940816879272}
+02/25/2022 18:01:28 - INFO - codeparrot_training - Step 31219: {'lr': 0.0001662552950210639, 'samples': 15984640, 'steps': 31219, 'loss/train': 1.4903556108474731}
+02/25/2022 18:01:33 - INFO - codeparrot_training - Step 31220: {'lr': 0.00016623987806916902, 'samples': 15985152, 'steps': 31220, 'loss/train': 2.1030163764953613}
+02/25/2022 18:01:37 - INFO - codeparrot_training - Step 31221: {'lr': 0.0001662244614760759, 'samples': 15985664, 'steps': 31221, 'loss/train': 2.705338954925537}
+02/25/2022 18:01:42 - INFO - codeparrot_training - Step 31222: {'lr': 0.00016620904524185072, 'samples': 15986176, 'steps': 31222, 'loss/train': 1.33682119846344}
+02/25/2022 18:01:46 - INFO - codeparrot_training - Step 31223: {'lr': 0.00016619362936655927, 'samples': 15986688, 'steps': 31223, 'loss/train': 0.3181931972503662}
+02/25/2022 18:01:52 - INFO - codeparrot_training - Step 31224: {'lr': 0.0001661782138502677, 'samples': 15987200, 'steps': 31224, 'loss/train': 2.599428415298462}
+02/25/2022 18:01:56 - INFO - codeparrot_training - Step 31225: {'lr': 0.00016616279869304201, 'samples': 15987712, 'steps': 31225, 'loss/train': 0.8624588251113892}
+02/25/2022 18:02:01 - INFO - codeparrot_training - Step 31226: {'lr': 0.00016614738389494834, 'samples': 15988224, 'steps': 31226, 'loss/train': 1.3411211967468262}
+02/25/2022 18:02:05 - INFO - codeparrot_training - Step 31227: {'lr': 0.00016613196945605268, 'samples': 15988736, 'steps': 31227, 'loss/train': 2.5998666286468506}
+02/25/2022 18:02:10 - INFO - codeparrot_training - Step 31228: {'lr': 0.00016611655537642097, 'samples': 15989248, 'steps': 31228, 'loss/train': 1.2592430114746094}
+02/25/2022 18:02:14 - INFO - codeparrot_training - Step 31229: {'lr': 0.00016610114165611928, 'samples': 15989760, 'steps': 31229, 'loss/train': 1.348429799079895}
+02/25/2022 18:02:19 - INFO - codeparrot_training - Step 31230: {'lr': 0.00016608572829521367, 'samples': 15990272, 'steps': 31230, 'loss/train': 2.002760410308838}
+02/25/2022 18:02:23 - INFO - codeparrot_training - Step 31231: {'lr': 0.00016607031529377025, 'samples': 15990784, 'steps': 31231, 'loss/train': 1.2538737058639526}
+02/25/2022 18:02:28 - INFO - codeparrot_training - Step 31232: {'lr': 0.00016605490265185485, 'samples': 15991296, 'steps': 31232, 'loss/train': 1.1970288753509521}
+02/25/2022 18:02:32 - INFO - codeparrot_training - Step 31233: {'lr': 0.0001660394903695336, 'samples': 15991808, 'steps': 31233, 'loss/train': 1.6322320699691772}
+02/25/2022 18:02:37 - INFO - codeparrot_training - Step 31234: {'lr': 0.00016602407844687255, 'samples': 15992320, 'steps': 31234, 'loss/train': 2.013763427734375}
+02/25/2022 18:02:41 - INFO - codeparrot_training - Step 31235: {'lr': 0.00016600866688393768, 'samples': 15992832, 'steps': 31235, 'loss/train': 1.3738502264022827}
+02/25/2022 18:02:46 - INFO - codeparrot_training - Step 31236: {'lr': 0.000165993255680795, 'samples': 15993344, 'steps': 31236, 'loss/train': 1.6297922134399414}
+02/25/2022 18:02:50 - INFO - codeparrot_training - Step 31237: {'lr': 0.00016597784483751048, 'samples': 15993856, 'steps': 31237, 'loss/train': 1.540042757987976}
+02/25/2022 18:02:55 - INFO - codeparrot_training - Step 31238: {'lr': 0.00016596243435415027, 'samples': 15994368, 'steps': 31238, 'loss/train': 1.8797917366027832}
+02/25/2022 18:02:59 - INFO - codeparrot_training - Step 31239: {'lr': 0.00016594702423078028, 'samples': 15994880, 'steps': 31239, 'loss/train': 1.9537409543991089}
+02/25/2022 18:03:05 - INFO - codeparrot_training - Step 31240: {'lr': 0.00016593161446746657, 'samples': 15995392, 'steps': 31240, 'loss/train': 2.262458324432373}
+02/25/2022 18:03:09 - INFO - codeparrot_training - Step 31241: {'lr': 0.00016591620506427513, 'samples': 15995904, 'steps': 31241, 'loss/train': 1.32154381275177}
+02/25/2022 18:03:14 - INFO - codeparrot_training - Step 31242: {'lr': 0.00016590079602127192, 'samples': 15996416, 'steps': 31242, 'loss/train': 1.9989919662475586}
+02/25/2022 18:03:18 - INFO - codeparrot_training - Step 31243: {'lr': 0.000165885387338523, 'samples': 15996928, 'steps': 31243, 'loss/train': 1.2249836921691895}
+02/25/2022 18:03:23 - INFO - codeparrot_training - Step 31244: {'lr': 0.0001658699790160944, 'samples': 15997440, 'steps': 31244, 'loss/train': 1.271769642829895}
+02/25/2022 18:03:26 - INFO - codeparrot_training - Step 31245: {'lr': 0.0001658545710540521, 'samples': 15997952, 'steps': 31245, 'loss/train': 1.6355676651000977}
+02/25/2022 18:03:32 - INFO - codeparrot_training - Step 31246: {'lr': 0.00016583916345246208, 'samples': 15998464, 'steps': 31246, 'loss/train': 1.394234299659729}
+02/25/2022 18:03:35 - INFO - codeparrot_training - Step 31247: {'lr': 0.00016582375621139033, 'samples': 15998976, 'steps': 31247, 'loss/train': 2.311051845550537}
+02/25/2022 18:03:41 - INFO - codeparrot_training - Step 31248: {'lr': 0.000165808349330903, 'samples': 15999488, 'steps': 31248, 'loss/train': 2.1872732639312744}
+02/25/2022 18:03:47 - INFO - codeparrot_training - Step 31249: {'lr': 0.00016579294281106587, 'samples': 16000000, 'steps': 31249, 'loss/train': 0.5892047882080078}
+02/25/2022 18:03:51 - INFO - codeparrot_training - Step 31250: {'lr': 0.000165777536651945, 'samples': 16000512, 'steps': 31250, 'loss/train': 1.1025352478027344}
+02/25/2022 18:03:56 - INFO - codeparrot_training - Step 31251: {'lr': 0.00016576213085360648, 'samples': 16001024, 'steps': 31251, 'loss/train': 1.4451408386230469}
+02/25/2022 18:04:00 - INFO - codeparrot_training - Step 31252: {'lr': 0.00016574672541611618, 'samples': 16001536, 'steps': 31252, 'loss/train': 1.3800761699676514}
+02/25/2022 18:04:05 - INFO - codeparrot_training - Step 31253: {'lr': 0.0001657313203395403, 'samples': 16002048, 'steps': 31253, 'loss/train': 1.1232661008834839}
+02/25/2022 18:04:09 - INFO - codeparrot_training - Step 31254: {'lr': 0.00016571591562394457, 'samples': 16002560, 'steps': 31254, 'loss/train': 1.5900959968566895}
+02/25/2022 18:04:14 - INFO - codeparrot_training - Step 31255: {'lr': 0.00016570051126939505, 'samples': 16003072, 'steps': 31255, 'loss/train': 1.1457480192184448}
+02/25/2022 18:04:18 - INFO - codeparrot_training - Step 31256: {'lr': 0.0001656851072759578, 'samples': 16003584, 'steps': 31256, 'loss/train': 2.219452381134033}
+02/25/2022 18:04:23 - INFO - codeparrot_training - Step 31257: {'lr': 0.00016566970364369893, 'samples': 16004096, 'steps': 31257, 'loss/train': 1.8960309028625488}
+02/25/2022 18:04:27 - INFO - codeparrot_training - Step 31258: {'lr': 0.00016565430037268413, 'samples': 16004608, 'steps': 31258, 'loss/train': 1.7764841318130493}
+02/25/2022 18:04:33 - INFO - codeparrot_training - Step 31259: {'lr': 0.0001656388974629795, 'samples': 16005120, 'steps': 31259, 'loss/train': 1.7179733514785767}
+02/25/2022 18:04:36 - INFO - codeparrot_training - Step 31260: {'lr': 0.0001656234949146511, 'samples': 16005632, 'steps': 31260, 'loss/train': 1.6806409358978271}
+02/25/2022 18:04:42 - INFO - codeparrot_training - Step 31261: {'lr': 0.00016560809272776492, 'samples': 16006144, 'steps': 31261, 'loss/train': 2.0297491550445557}
+02/25/2022 18:04:45 - INFO - codeparrot_training - Step 31262: {'lr': 0.00016559269090238676, 'samples': 16006656, 'steps': 31262, 'loss/train': 0.8652766346931458}
+02/25/2022 18:04:51 - INFO - codeparrot_training - Step 31263: {'lr': 0.00016557728943858275, 'samples': 16007168, 'steps': 31263, 'loss/train': 1.8180917501449585}
+02/25/2022 18:04:54 - INFO - codeparrot_training - Step 31264: {'lr': 0.00016556188833641877, 'samples': 16007680, 'steps': 31264, 'loss/train': 1.8327325582504272}
+02/25/2022 18:05:00 - INFO - codeparrot_training - Step 31265: {'lr': 0.00016554648759596093, 'samples': 16008192, 'steps': 31265, 'loss/train': 1.4832602739334106}
+02/25/2022 18:05:03 - INFO - codeparrot_training - Step 31266: {'lr': 0.00016553108721727516, 'samples': 16008704, 'steps': 31266, 'loss/train': 1.712033987045288}
+02/25/2022 18:05:09 - INFO - codeparrot_training - Step 31267: {'lr': 0.00016551568720042732, 'samples': 16009216, 'steps': 31267, 'loss/train': 0.8692848682403564}
+02/25/2022 18:05:12 - INFO - codeparrot_training - Step 31268: {'lr': 0.00016550028754548342, 'samples': 16009728, 'steps': 31268, 'loss/train': 2.2852439880371094}
+02/25/2022 18:05:18 - INFO - codeparrot_training - Step 31269: {'lr': 0.00016548488825250944, 'samples': 16010240, 'steps': 31269, 'loss/train': 1.9843685626983643}
+02/25/2022 18:05:21 - INFO - codeparrot_training - Step 31270: {'lr': 0.00016546948932157153, 'samples': 16010752, 'steps': 31270, 'loss/train': 1.6326148509979248}
+02/25/2022 18:05:27 - INFO - codeparrot_training - Step 31271: {'lr': 0.00016545409075273532, 'samples': 16011264, 'steps': 31271, 'loss/train': 1.526254415512085}
+02/25/2022 18:05:31 - INFO - codeparrot_training - Step 31272: {'lr': 0.00016543869254606697, 'samples': 16011776, 'steps': 31272, 'loss/train': 1.3565102815628052}
+02/25/2022 18:05:36 - INFO - codeparrot_training - Step 31273: {'lr': 0.0001654232947016324, 'samples': 16012288, 'steps': 31273, 'loss/train': 1.2636314630508423}
+02/25/2022 18:05:40 - INFO - codeparrot_training - Step 31274: {'lr': 0.00016540789721949768, 'samples': 16012800, 'steps': 31274, 'loss/train': 1.8139207363128662}
+02/25/2022 18:05:45 - INFO - codeparrot_training - Step 31275: {'lr': 0.00016539250009972855, 'samples': 16013312, 'steps': 31275, 'loss/train': 1.3705881834030151}
+02/25/2022 18:05:49 - INFO - codeparrot_training - Step 31276: {'lr': 0.00016537710334239115, 'samples': 16013824, 'steps': 31276, 'loss/train': 1.38565194606781}
+02/25/2022 18:05:54 - INFO - codeparrot_training - Step 31277: {'lr': 0.00016536170694755133, 'samples': 16014336, 'steps': 31277, 'loss/train': 0.8240061402320862}
+02/25/2022 18:05:58 - INFO - codeparrot_training - Step 31278: {'lr': 0.0001653463109152751, 'samples': 16014848, 'steps': 31278, 'loss/train': 1.8872151374816895}
+02/25/2022 18:06:03 - INFO - codeparrot_training - Step 31279: {'lr': 0.00016533091524562844, 'samples': 16015360, 'steps': 31279, 'loss/train': 1.7890673875808716}
+02/25/2022 18:06:07 - INFO - codeparrot_training - Step 31280: {'lr': 0.00016531551993867715, 'samples': 16015872, 'steps': 31280, 'loss/train': 0.13869307935237885}
+02/25/2022 18:06:13 - INFO - codeparrot_training - Step 31281: {'lr': 0.00016530012499448737, 'samples': 16016384, 'steps': 31281, 'loss/train': 0.7887718677520752}
+02/25/2022 18:06:16 - INFO - codeparrot_training - Step 31282: {'lr': 0.00016528473041312492, 'samples': 16016896, 'steps': 31282, 'loss/train': 2.1967010498046875}
+02/25/2022 18:06:22 - INFO - codeparrot_training - Step 31283: {'lr': 0.0001652693361946558, 'samples': 16017408, 'steps': 31283, 'loss/train': 1.0679566860198975}
+02/25/2022 18:06:25 - INFO - codeparrot_training - Step 31284: {'lr': 0.00016525394233914592, 'samples': 16017920, 'steps': 31284, 'loss/train': 1.8274058103561401}
+02/25/2022 18:06:31 - INFO - codeparrot_training - Step 31285: {'lr': 0.0001652385488466613, 'samples': 16018432, 'steps': 31285, 'loss/train': 2.4772987365722656}
+02/25/2022 18:06:35 - INFO - codeparrot_training - Step 31286: {'lr': 0.00016522315571726773, 'samples': 16018944, 'steps': 31286, 'loss/train': 0.8367212414741516}
+02/25/2022 18:06:40 - INFO - codeparrot_training - Step 31287: {'lr': 0.00016520776295103127, 'samples': 16019456, 'steps': 31287, 'loss/train': 1.7551041841506958}
+02/25/2022 18:06:44 - INFO - codeparrot_training - Step 31288: {'lr': 0.00016519237054801793, 'samples': 16019968, 'steps': 31288, 'loss/train': 2.6851449012756348}
+02/25/2022 18:06:49 - INFO - codeparrot_training - Step 31289: {'lr': 0.00016517697850829344, 'samples': 16020480, 'steps': 31289, 'loss/train': 1.8984678983688354}
+02/25/2022 18:06:53 - INFO - codeparrot_training - Step 31290: {'lr': 0.00016516158683192383, 'samples': 16020992, 'steps': 31290, 'loss/train': 1.0831626653671265}
+02/25/2022 18:06:58 - INFO - codeparrot_training - Step 31291: {'lr': 0.00016514619551897503, 'samples': 16021504, 'steps': 31291, 'loss/train': 0.8535380363464355}
+02/25/2022 18:07:02 - INFO - codeparrot_training - Step 31292: {'lr': 0.00016513080456951313, 'samples': 16022016, 'steps': 31292, 'loss/train': 0.36174851655960083}
+02/25/2022 18:07:08 - INFO - codeparrot_training - Step 31293: {'lr': 0.00016511541398360377, 'samples': 16022528, 'steps': 31293, 'loss/train': 2.327173948287964}
+02/25/2022 18:07:11 - INFO - codeparrot_training - Step 31294: {'lr': 0.00016510002376131307, 'samples': 16023040, 'steps': 31294, 'loss/train': 2.3153064250946045}
+02/25/2022 18:07:15 - INFO - codeparrot_training - Step 31295: {'lr': 0.00016508463390270685, 'samples': 16023552, 'steps': 31295, 'loss/train': 2.1993751525878906}
+02/25/2022 18:07:22 - INFO - codeparrot_training - Step 31296: {'lr': 0.00016506924440785122, 'samples': 16024064, 'steps': 31296, 'loss/train': 1.4185014963150024}
+02/25/2022 18:07:25 - INFO - codeparrot_training - Step 31297: {'lr': 0.00016505385527681188, 'samples': 16024576, 'steps': 31297, 'loss/train': 1.308010458946228}
+02/25/2022 18:07:31 - INFO - codeparrot_training - Step 31298: {'lr': 0.00016503846650965488, 'samples': 16025088, 'steps': 31298, 'loss/train': 4.247918128967285}
+02/25/2022 18:07:34 - INFO - codeparrot_training - Step 31299: {'lr': 0.00016502307810644607, 'samples': 16025600, 'steps': 31299, 'loss/train': 2.6074845790863037}
+02/25/2022 18:07:40 - INFO - codeparrot_training - Step 31300: {'lr': 0.00016500769006725142, 'samples': 16026112, 'steps': 31300, 'loss/train': 0.8778741955757141}
+02/25/2022 18:07:43 - INFO - codeparrot_training - Step 31301: {'lr': 0.00016499230239213695, 'samples': 16026624, 'steps': 31301, 'loss/train': 1.4500478506088257}
+02/25/2022 18:07:49 - INFO - codeparrot_training - Step 31302: {'lr': 0.00016497691508116837, 'samples': 16027136, 'steps': 31302, 'loss/train': 1.902868390083313}
+02/25/2022 18:07:52 - INFO - codeparrot_training - Step 31303: {'lr': 0.00016496152813441165, 'samples': 16027648, 'steps': 31303, 'loss/train': 1.194953203201294}
+02/25/2022 18:07:58 - INFO - codeparrot_training - Step 31304: {'lr': 0.00016494614155193276, 'samples': 16028160, 'steps': 31304, 'loss/train': 1.6722511053085327}
+02/25/2022 18:08:03 - INFO - codeparrot_training - Step 31305: {'lr': 0.0001649307553337977, 'samples': 16028672, 'steps': 31305, 'loss/train': 1.999638557434082}
+02/25/2022 18:08:07 - INFO - codeparrot_training - Step 31306: {'lr': 0.00016491536948007218, 'samples': 16029184, 'steps': 31306, 'loss/train': 1.7298903465270996}
+02/25/2022 18:08:13 - INFO - codeparrot_training - Step 31307: {'lr': 0.00016489998399082218, 'samples': 16029696, 'steps': 31307, 'loss/train': 1.7059166431427002}
+02/25/2022 18:08:17 - INFO - codeparrot_training - Step 31308: {'lr': 0.00016488459886611362, 'samples': 16030208, 'steps': 31308, 'loss/train': 1.421174168586731}
+02/25/2022 18:08:22 - INFO - codeparrot_training - Step 31309: {'lr': 0.0001648692141060125, 'samples': 16030720, 'steps': 31309, 'loss/train': 0.2620045840740204}
+02/25/2022 18:08:26 - INFO - codeparrot_training - Step 31310: {'lr': 0.0001648538297105846, 'samples': 16031232, 'steps': 31310, 'loss/train': 1.2898287773132324}
+02/25/2022 18:08:31 - INFO - codeparrot_training - Step 31311: {'lr': 0.00016483844567989582, 'samples': 16031744, 'steps': 31311, 'loss/train': 1.1318995952606201}
+02/25/2022 18:08:35 - INFO - codeparrot_training - Step 31312: {'lr': 0.0001648230620140121, 'samples': 16032256, 'steps': 31312, 'loss/train': 2.052733898162842}
+02/25/2022 18:08:40 - INFO - codeparrot_training - Step 31313: {'lr': 0.00016480767871299932, 'samples': 16032768, 'steps': 31313, 'loss/train': 2.0317342281341553}
+02/25/2022 18:08:44 - INFO - codeparrot_training - Step 31314: {'lr': 0.00016479229577692352, 'samples': 16033280, 'steps': 31314, 'loss/train': 1.9304574728012085}
+02/25/2022 18:08:49 - INFO - codeparrot_training - Step 31315: {'lr': 0.00016477691320585032, 'samples': 16033792, 'steps': 31315, 'loss/train': 2.079672336578369}
+02/25/2022 18:08:53 - INFO - codeparrot_training - Step 31316: {'lr': 0.00016476153099984582, 'samples': 16034304, 'steps': 31316, 'loss/train': 1.5480804443359375}
+02/25/2022 18:08:58 - INFO - codeparrot_training - Step 31317: {'lr': 0.0001647461491589758, 'samples': 16034816, 'steps': 31317, 'loss/train': 1.8649154901504517}
+02/25/2022 18:09:02 - INFO - codeparrot_training - Step 31318: {'lr': 0.00016473076768330633, 'samples': 16035328, 'steps': 31318, 'loss/train': 4.1953277587890625}
+02/25/2022 18:09:07 - INFO - codeparrot_training - Step 31319: {'lr': 0.0001647153865729031, 'samples': 16035840, 'steps': 31319, 'loss/train': 0.2611694633960724}
+02/25/2022 18:09:11 - INFO - codeparrot_training - Step 31320: {'lr': 0.00016470000582783205, 'samples': 16036352, 'steps': 31320, 'loss/train': 1.3507462739944458}
+02/25/2022 18:09:17 - INFO - codeparrot_training - Step 31321: {'lr': 0.00016468462544815912, 'samples': 16036864, 'steps': 31321, 'loss/train': 3.5554237365722656}
+02/25/2022 18:09:21 - INFO - codeparrot_training - Step 31322: {'lr': 0.00016466924543395017, 'samples': 16037376, 'steps': 31322, 'loss/train': 2.0216739177703857}
+02/25/2022 18:09:26 - INFO - codeparrot_training - Step 31323: {'lr': 0.00016465386578527107, 'samples': 16037888, 'steps': 31323, 'loss/train': 1.93572199344635}
+02/25/2022 18:09:30 - INFO - codeparrot_training - Step 31324: {'lr': 0.00016463848650218765, 'samples': 16038400, 'steps': 31324, 'loss/train': 2.1210403442382812}
+02/25/2022 18:09:35 - INFO - codeparrot_training - Step 31325: {'lr': 0.00016462310758476596, 'samples': 16038912, 'steps': 31325, 'loss/train': 3.008291482925415}
+02/25/2022 18:09:39 - INFO - codeparrot_training - Step 31326: {'lr': 0.0001646077290330717, 'samples': 16039424, 'steps': 31326, 'loss/train': 0.7228884100914001}
+02/25/2022 18:09:44 - INFO - codeparrot_training - Step 31327: {'lr': 0.00016459235084717084, 'samples': 16039936, 'steps': 31327, 'loss/train': 1.9843220710754395}
+02/25/2022 18:09:47 - INFO - codeparrot_training - Step 31328: {'lr': 0.00016457697302712918, 'samples': 16040448, 'steps': 31328, 'loss/train': 2.1160051822662354}
+02/25/2022 18:09:53 - INFO - codeparrot_training - Step 31329: {'lr': 0.0001645615955730127, 'samples': 16040960, 'steps': 31329, 'loss/train': 2.9800057411193848}
+02/25/2022 18:09:56 - INFO - codeparrot_training - Step 31330: {'lr': 0.00016454621848488714, 'samples': 16041472, 'steps': 31330, 'loss/train': 1.9088553190231323}
+02/25/2022 18:10:02 - INFO - codeparrot_training - Step 31331: {'lr': 0.00016453084176281855, 'samples': 16041984, 'steps': 31331, 'loss/train': 0.1931828409433365}
+02/25/2022 18:10:05 - INFO - codeparrot_training - Step 31332: {'lr': 0.0001645154654068727, 'samples': 16042496, 'steps': 31332, 'loss/train': 1.522559642791748}
+02/25/2022 18:10:12 - INFO - codeparrot_training - Step 31333: {'lr': 0.0001645000894171153, 'samples': 16043008, 'steps': 31333, 'loss/train': 1.8941537141799927}
+02/25/2022 18:10:15 - INFO - codeparrot_training - Step 31334: {'lr': 0.00016448471379361247, 'samples': 16043520, 'steps': 31334, 'loss/train': 1.9524195194244385}
+02/25/2022 18:10:21 - INFO - codeparrot_training - Step 31335: {'lr': 0.00016446933853642991, 'samples': 16044032, 'steps': 31335, 'loss/train': 1.8290770053863525}
+02/25/2022 18:10:24 - INFO - codeparrot_training - Step 31336: {'lr': 0.0001644539636456337, 'samples': 16044544, 'steps': 31336, 'loss/train': 1.6625949144363403}
+02/25/2022 18:10:30 - INFO - codeparrot_training - Step 31337: {'lr': 0.0001644385891212894, 'samples': 16045056, 'steps': 31337, 'loss/train': 1.8477418422698975}
+02/25/2022 18:10:33 - INFO - codeparrot_training - Step 31338: {'lr': 0.000164423214963463, 'samples': 16045568, 'steps': 31338, 'loss/train': 0.9607930183410645}
+02/25/2022 18:10:39 - INFO - codeparrot_training - Step 31339: {'lr': 0.00016440784117222046, 'samples': 16046080, 'steps': 31339, 'loss/train': 1.4734187126159668}
+02/25/2022 18:10:42 - INFO - codeparrot_training - Step 31340: {'lr': 0.0001643924677476276, 'samples': 16046592, 'steps': 31340, 'loss/train': 1.2680000066757202}
+02/25/2022 18:10:49 - INFO - codeparrot_training - Step 31341: {'lr': 0.0001643770946897501, 'samples': 16047104, 'steps': 31341, 'loss/train': 0.39174580574035645}
+02/25/2022 18:10:52 - INFO - codeparrot_training - Step 31342: {'lr': 0.00016436172199865395, 'samples': 16047616, 'steps': 31342, 'loss/train': 0.2295610010623932}
+02/25/2022 18:10:58 - INFO - codeparrot_training - Step 31343: {'lr': 0.00016434634967440502, 'samples': 16048128, 'steps': 31343, 'loss/train': 1.4248918294906616}
+02/25/2022 18:11:01 - INFO - codeparrot_training - Step 31344: {'lr': 0.0001643309777170692, 'samples': 16048640, 'steps': 31344, 'loss/train': 1.4873435497283936}
+02/25/2022 18:11:07 - INFO - codeparrot_training - Step 31345: {'lr': 0.00016431560612671216, 'samples': 16049152, 'steps': 31345, 'loss/train': 1.9297418594360352}
+02/25/2022 18:11:10 - INFO - codeparrot_training - Step 31346: {'lr': 0.0001643002349033999, 'samples': 16049664, 'steps': 31346, 'loss/train': 2.2651169300079346}
+02/25/2022 18:11:16 - INFO - codeparrot_training - Step 31347: {'lr': 0.00016428486404719816, 'samples': 16050176, 'steps': 31347, 'loss/train': 1.835245132446289}
+02/25/2022 18:11:19 - INFO - codeparrot_training - Step 31348: {'lr': 0.00016426949355817288, 'samples': 16050688, 'steps': 31348, 'loss/train': 1.778789758682251}
+02/25/2022 18:11:25 - INFO - codeparrot_training - Step 31349: {'lr': 0.00016425412343638998, 'samples': 16051200, 'steps': 31349, 'loss/train': 2.3909201622009277}
+02/25/2022 18:11:28 - INFO - codeparrot_training - Step 31350: {'lr': 0.00016423875368191506, 'samples': 16051712, 'steps': 31350, 'loss/train': 2.505403518676758}
+02/25/2022 18:11:35 - INFO - codeparrot_training - Step 31351: {'lr': 0.0001642233842948141, 'samples': 16052224, 'steps': 31351, 'loss/train': 1.5558738708496094}
+02/25/2022 18:11:38 - INFO - codeparrot_training - Step 31352: {'lr': 0.00016420801527515294, 'samples': 16052736, 'steps': 31352, 'loss/train': 2.3194472789764404}
+02/25/2022 18:11:44 - INFO - codeparrot_training - Step 31353: {'lr': 0.00016419264662299749, 'samples': 16053248, 'steps': 31353, 'loss/train': 2.3688805103302}
+02/25/2022 18:11:47 - INFO - codeparrot_training - Step 31354: {'lr': 0.0001641772783384134, 'samples': 16053760, 'steps': 31354, 'loss/train': 0.040782712399959564}
+02/25/2022 18:11:53 - INFO - codeparrot_training - Step 31355: {'lr': 0.00016416191042146656, 'samples': 16054272, 'steps': 31355, 'loss/train': 1.2057888507843018}
+02/25/2022 18:11:56 - INFO - codeparrot_training - Step 31356: {'lr': 0.00016414654287222287, 'samples': 16054784, 'steps': 31356, 'loss/train': 1.92635977268219}
+02/25/2022 18:12:02 - INFO - codeparrot_training - Step 31357: {'lr': 0.00016413117569074822, 'samples': 16055296, 'steps': 31357, 'loss/train': 1.574373722076416}
+02/25/2022 18:12:05 - INFO - codeparrot_training - Step 31358: {'lr': 0.00016411580887710825, 'samples': 16055808, 'steps': 31358, 'loss/train': 2.6870310306549072}
+02/25/2022 18:12:11 - INFO - codeparrot_training - Step 31359: {'lr': 0.0001641004424313689, 'samples': 16056320, 'steps': 31359, 'loss/train': 2.15073823928833}
+02/25/2022 18:12:14 - INFO - codeparrot_training - Step 31360: {'lr': 0.00016408507635359598, 'samples': 16056832, 'steps': 31360, 'loss/train': 2.0025622844696045}
+02/25/2022 18:12:20 - INFO - codeparrot_training - Step 31361: {'lr': 0.0001640697106438553, 'samples': 16057344, 'steps': 31361, 'loss/train': 2.7186341285705566}
+02/25/2022 18:12:24 - INFO - codeparrot_training - Step 31362: {'lr': 0.00016405434530221278, 'samples': 16057856, 'steps': 31362, 'loss/train': 1.7994041442871094}
+02/25/2022 18:12:29 - INFO - codeparrot_training - Step 31363: {'lr': 0.00016403898032873405, 'samples': 16058368, 'steps': 31363, 'loss/train': 1.107377052307129}
+02/25/2022 18:12:33 - INFO - codeparrot_training - Step 31364: {'lr': 0.00016402361572348507, 'samples': 16058880, 'steps': 31364, 'loss/train': 0.35256680846214294}
+02/25/2022 18:12:38 - INFO - codeparrot_training - Step 31365: {'lr': 0.00016400825148653165, 'samples': 16059392, 'steps': 31365, 'loss/train': 1.1622322797775269}
+02/25/2022 18:12:42 - INFO - codeparrot_training - Step 31366: {'lr': 0.00016399288761793956, 'samples': 16059904, 'steps': 31366, 'loss/train': 1.2829182147979736}
+02/25/2022 18:12:48 - INFO - codeparrot_training - Step 31367: {'lr': 0.0001639775241177746, 'samples': 16060416, 'steps': 31367, 'loss/train': 1.6835365295410156}
+02/25/2022 18:12:51 - INFO - codeparrot_training - Step 31368: {'lr': 0.0001639621609861026, 'samples': 16060928, 'steps': 31368, 'loss/train': 2.2982916831970215}
+02/25/2022 18:12:57 - INFO - codeparrot_training - Step 31369: {'lr': 0.0001639467982229894, 'samples': 16061440, 'steps': 31369, 'loss/train': 2.454707622528076}
+02/25/2022 18:13:00 - INFO - codeparrot_training - Step 31370: {'lr': 0.00016393143582850084, 'samples': 16061952, 'steps': 31370, 'loss/train': 2.0608131885528564}
+02/25/2022 18:13:06 - INFO - codeparrot_training - Step 31371: {'lr': 0.00016391607380270262, 'samples': 16062464, 'steps': 31371, 'loss/train': 2.734260082244873}
+02/25/2022 18:13:09 - INFO - codeparrot_training - Step 31372: {'lr': 0.00016390071214566067, 'samples': 16062976, 'steps': 31372, 'loss/train': 2.8478658199310303}
+02/25/2022 18:13:15 - INFO - codeparrot_training - Step 31373: {'lr': 0.00016388535085744063, 'samples': 16063488, 'steps': 31373, 'loss/train': 2.019195556640625}
+02/25/2022 18:13:19 - INFO - codeparrot_training - Step 31374: {'lr': 0.00016386998993810843, 'samples': 16064000, 'steps': 31374, 'loss/train': 2.0332553386688232}
+02/25/2022 18:13:24 - INFO - codeparrot_training - Step 31375: {'lr': 0.0001638546293877299, 'samples': 16064512, 'steps': 31375, 'loss/train': 1.3325759172439575}
+02/25/2022 18:13:28 - INFO - codeparrot_training - Step 31376: {'lr': 0.00016383926920637078, 'samples': 16065024, 'steps': 31376, 'loss/train': 1.8542050123214722}
+02/25/2022 18:13:34 - INFO - codeparrot_training - Step 31377: {'lr': 0.00016382390939409684, 'samples': 16065536, 'steps': 31377, 'loss/train': 0.6890619993209839}
+02/25/2022 18:13:38 - INFO - codeparrot_training - Step 31378: {'lr': 0.00016380854995097388, 'samples': 16066048, 'steps': 31378, 'loss/train': 0.48866522312164307}
+02/25/2022 18:13:43 - INFO - codeparrot_training - Step 31379: {'lr': 0.00016379319087706784, 'samples': 16066560, 'steps': 31379, 'loss/train': 1.8394485712051392}
+02/25/2022 18:13:47 - INFO - codeparrot_training - Step 31380: {'lr': 0.00016377783217244428, 'samples': 16067072, 'steps': 31380, 'loss/train': 1.0201244354248047}
+02/25/2022 18:13:52 - INFO - codeparrot_training - Step 31381: {'lr': 0.0001637624738371691, 'samples': 16067584, 'steps': 31381, 'loss/train': 1.376876711845398}
+02/25/2022 18:13:56 - INFO - codeparrot_training - Step 31382: {'lr': 0.00016374711587130813, 'samples': 16068096, 'steps': 31382, 'loss/train': 1.0548646450042725}
+02/25/2022 18:14:01 - INFO - codeparrot_training - Step 31383: {'lr': 0.0001637317582749272, 'samples': 16068608, 'steps': 31383, 'loss/train': 1.7436797618865967}
+02/25/2022 18:14:05 - INFO - codeparrot_training - Step 31384: {'lr': 0.00016371640104809193, 'samples': 16069120, 'steps': 31384, 'loss/train': 1.2995902299880981}
+02/25/2022 18:14:10 - INFO - codeparrot_training - Step 31385: {'lr': 0.00016370104419086823, 'samples': 16069632, 'steps': 31385, 'loss/train': 1.227754831314087}
+02/25/2022 18:14:14 - INFO - codeparrot_training - Step 31386: {'lr': 0.0001636856877033218, 'samples': 16070144, 'steps': 31386, 'loss/train': 1.9077762365341187}
+02/25/2022 18:14:19 - INFO - codeparrot_training - Step 31387: {'lr': 0.00016367033158551852, 'samples': 16070656, 'steps': 31387, 'loss/train': 0.8147440552711487}
+02/25/2022 18:14:23 - INFO - codeparrot_training - Step 31388: {'lr': 0.00016365497583752423, 'samples': 16071168, 'steps': 31388, 'loss/train': 1.2274079322814941}
+02/25/2022 18:14:29 - INFO - codeparrot_training - Step 31389: {'lr': 0.00016363962045940449, 'samples': 16071680, 'steps': 31389, 'loss/train': 2.4627585411071777}
+02/25/2022 18:14:33 - INFO - codeparrot_training - Step 31390: {'lr': 0.00016362426545122517, 'samples': 16072192, 'steps': 31390, 'loss/train': 1.0315213203430176}
+02/25/2022 18:14:38 - INFO - codeparrot_training - Step 31391: {'lr': 0.0001636089108130521, 'samples': 16072704, 'steps': 31391, 'loss/train': 1.514729380607605}
+02/25/2022 18:14:42 - INFO - codeparrot_training - Step 31392: {'lr': 0.00016359355654495112, 'samples': 16073216, 'steps': 31392, 'loss/train': 2.878905773162842}
+02/25/2022 18:14:47 - INFO - codeparrot_training - Step 31393: {'lr': 0.0001635782026469878, 'samples': 16073728, 'steps': 31393, 'loss/train': 2.0227460861206055}
+02/25/2022 18:14:51 - INFO - codeparrot_training - Step 31394: {'lr': 0.00016356284911922802, 'samples': 16074240, 'steps': 31394, 'loss/train': 1.5283938646316528}
+02/25/2022 18:14:56 - INFO - codeparrot_training - Step 31395: {'lr': 0.00016354749596173758, 'samples': 16074752, 'steps': 31395, 'loss/train': 2.339381694793701}
+02/25/2022 18:15:00 - INFO - codeparrot_training - Step 31396: {'lr': 0.00016353214317458214, 'samples': 16075264, 'steps': 31396, 'loss/train': 1.9438034296035767}
+02/25/2022 18:15:05 - INFO - codeparrot_training - Step 31397: {'lr': 0.0001635167907578277, 'samples': 16075776, 'steps': 31397, 'loss/train': 0.17732861638069153}
+02/25/2022 18:15:09 - INFO - codeparrot_training - Step 31398: {'lr': 0.00016350143871153976, 'samples': 16076288, 'steps': 31398, 'loss/train': 1.9937018156051636}
+02/25/2022 18:15:14 - INFO - codeparrot_training - Step 31399: {'lr': 0.00016348608703578416, 'samples': 16076800, 'steps': 31399, 'loss/train': 1.9904096126556396}
+02/25/2022 18:15:18 - INFO - codeparrot_training - Step 31400: {'lr': 0.0001634707357306267, 'samples': 16077312, 'steps': 31400, 'loss/train': 1.1059869527816772}
+02/25/2022 18:15:23 - INFO - codeparrot_training - Step 31401: {'lr': 0.00016345538479613326, 'samples': 16077824, 'steps': 31401, 'loss/train': 2.338977336883545}
+02/25/2022 18:15:27 - INFO - codeparrot_training - Step 31402: {'lr': 0.00016344003423236937, 'samples': 16078336, 'steps': 31402, 'loss/train': 1.0222047567367554}
+02/25/2022 18:15:33 - INFO - codeparrot_training - Step 31403: {'lr': 0.00016342468403940087, 'samples': 16078848, 'steps': 31403, 'loss/train': 1.3300868272781372}
+02/25/2022 18:15:36 - INFO - codeparrot_training - Step 31404: {'lr': 0.00016340933421729354, 'samples': 16079360, 'steps': 31404, 'loss/train': 2.7973134517669678}
+02/25/2022 18:15:42 - INFO - codeparrot_training - Step 31405: {'lr': 0.00016339398476611323, 'samples': 16079872, 'steps': 31405, 'loss/train': 0.03932815417647362}
+02/25/2022 18:15:45 - INFO - codeparrot_training - Step 31406: {'lr': 0.00016337863568592547, 'samples': 16080384, 'steps': 31406, 'loss/train': 1.5377815961837769}
+02/25/2022 18:15:51 - INFO - codeparrot_training - Step 31407: {'lr': 0.00016336328697679614, 'samples': 16080896, 'steps': 31407, 'loss/train': 1.4611084461212158}
+02/25/2022 18:15:55 - INFO - codeparrot_training - Step 31408: {'lr': 0.00016334793863879098, 'samples': 16081408, 'steps': 31408, 'loss/train': 2.497556209564209}
+02/25/2022 18:16:00 - INFO - codeparrot_training - Step 31409: {'lr': 0.00016333259067197582, 'samples': 16081920, 'steps': 31409, 'loss/train': 1.7908082008361816}
+02/25/2022 18:16:04 - INFO - codeparrot_training - Step 31410: {'lr': 0.00016331724307641626, 'samples': 16082432, 'steps': 31410, 'loss/train': 1.9612091779708862}
+02/25/2022 18:16:09 - INFO - codeparrot_training - Step 31411: {'lr': 0.00016330189585217809, 'samples': 16082944, 'steps': 31411, 'loss/train': 1.506575107574463}
+02/25/2022 18:16:13 - INFO - codeparrot_training - Step 31412: {'lr': 0.0001632865489993271, 'samples': 16083456, 'steps': 31412, 'loss/train': 0.2556702494621277}
+02/25/2022 18:16:18 - INFO - codeparrot_training - Step 31413: {'lr': 0.00016327120251792898, 'samples': 16083968, 'steps': 31413, 'loss/train': 1.2654744386672974}
+02/25/2022 18:16:22 - INFO - codeparrot_training - Step 31414: {'lr': 0.00016325585640804952, 'samples': 16084480, 'steps': 31414, 'loss/train': 2.5947704315185547}
+02/25/2022 18:16:28 - INFO - codeparrot_training - Step 31415: {'lr': 0.00016324051066975438, 'samples': 16084992, 'steps': 31415, 'loss/train': 3.0848538875579834}
+02/25/2022 18:16:31 - INFO - codeparrot_training - Step 31416: {'lr': 0.00016322516530310939, 'samples': 16085504, 'steps': 31416, 'loss/train': 2.0074641704559326}
+02/25/2022 18:16:37 - INFO - codeparrot_training - Step 31417: {'lr': 0.00016320982030818015, 'samples': 16086016, 'steps': 31417, 'loss/train': 2.3168277740478516}
+02/25/2022 18:16:40 - INFO - codeparrot_training - Step 31418: {'lr': 0.0001631944756850326, 'samples': 16086528, 'steps': 31418, 'loss/train': 1.6213756799697876}
+02/25/2022 18:16:46 - INFO - codeparrot_training - Step 31419: {'lr': 0.0001631791314337323, 'samples': 16087040, 'steps': 31419, 'loss/train': 2.0430426597595215}
+02/25/2022 18:16:49 - INFO - codeparrot_training - Step 31420: {'lr': 0.00016316378755434503, 'samples': 16087552, 'steps': 31420, 'loss/train': 0.6122271418571472}
+02/25/2022 18:16:55 - INFO - codeparrot_training - Step 31421: {'lr': 0.00016314844404693646, 'samples': 16088064, 'steps': 31421, 'loss/train': 2.4425461292266846}
+02/25/2022 18:16:58 - INFO - codeparrot_training - Step 31422: {'lr': 0.0001631331009115724, 'samples': 16088576, 'steps': 31422, 'loss/train': 0.983812689781189}
+02/25/2022 18:17:04 - INFO - codeparrot_training - Step 31423: {'lr': 0.0001631177581483187, 'samples': 16089088, 'steps': 31423, 'loss/train': 2.4030473232269287}
+02/25/2022 18:17:07 - INFO - codeparrot_training - Step 31424: {'lr': 0.00016310241575724077, 'samples': 16089600, 'steps': 31424, 'loss/train': 1.9232796430587769}
+02/25/2022 18:17:13 - INFO - codeparrot_training - Step 31425: {'lr': 0.00016308707373840453, 'samples': 16090112, 'steps': 31425, 'loss/train': 1.9096519947052002}
+02/25/2022 18:17:17 - INFO - codeparrot_training - Step 31426: {'lr': 0.00016307173209187569, 'samples': 16090624, 'steps': 31426, 'loss/train': 1.6862930059432983}
+02/25/2022 18:17:23 - INFO - codeparrot_training - Step 31427: {'lr': 0.00016305639081772, 'samples': 16091136, 'steps': 31427, 'loss/train': 0.8484938144683838}
+02/25/2022 18:17:26 - INFO - codeparrot_training - Step 31428: {'lr': 0.00016304104991600305, 'samples': 16091648, 'steps': 31428, 'loss/train': 1.4951131343841553}
+02/25/2022 18:17:32 - INFO - codeparrot_training - Step 31429: {'lr': 0.00016302570938679062, 'samples': 16092160, 'steps': 31429, 'loss/train': 2.608872652053833}
+02/25/2022 18:17:35 - INFO - codeparrot_training - Step 31430: {'lr': 0.00016301036923014844, 'samples': 16092672, 'steps': 31430, 'loss/train': 1.015969157218933}
+02/25/2022 18:17:40 - INFO - codeparrot_training - Step 31431: {'lr': 0.0001629950294461423, 'samples': 16093184, 'steps': 31431, 'loss/train': 0.4371148645877838}
+02/25/2022 18:17:46 - INFO - codeparrot_training - Step 31432: {'lr': 0.00016297969003483775, 'samples': 16093696, 'steps': 31432, 'loss/train': 2.5788638591766357}
+02/25/2022 18:17:49 - INFO - codeparrot_training - Step 31433: {'lr': 0.00016296435099630058, 'samples': 16094208, 'steps': 31433, 'loss/train': 2.3890678882598877}
+02/25/2022 18:17:56 - INFO - codeparrot_training - Step 31434: {'lr': 0.00016294901233059647, 'samples': 16094720, 'steps': 31434, 'loss/train': 1.4159319400787354}
+02/25/2022 18:17:59 - INFO - codeparrot_training - Step 31435: {'lr': 0.00016293367403779116, 'samples': 16095232, 'steps': 31435, 'loss/train': 2.2135400772094727}
+02/25/2022 18:18:05 - INFO - codeparrot_training - Step 31436: {'lr': 0.00016291833611795046, 'samples': 16095744, 'steps': 31436, 'loss/train': 2.3922173976898193}
+02/25/2022 18:18:08 - INFO - codeparrot_training - Step 31437: {'lr': 0.00016290299857113982, 'samples': 16096256, 'steps': 31437, 'loss/train': 1.6987357139587402}
+02/25/2022 18:18:13 - INFO - codeparrot_training - Step 31438: {'lr': 0.00016288766139742512, 'samples': 16096768, 'steps': 31438, 'loss/train': 1.840745449066162}
+02/25/2022 18:18:17 - INFO - codeparrot_training - Step 31439: {'lr': 0.00016287232459687197, 'samples': 16097280, 'steps': 31439, 'loss/train': 1.9439425468444824}
+02/25/2022 18:18:22 - INFO - codeparrot_training - Step 31440: {'lr': 0.00016285698816954626, 'samples': 16097792, 'steps': 31440, 'loss/train': 2.0155298709869385}
+02/25/2022 18:18:26 - INFO - codeparrot_training - Step 31441: {'lr': 0.0001628416521155134, 'samples': 16098304, 'steps': 31441, 'loss/train': 1.8807417154312134}
+02/25/2022 18:18:32 - INFO - codeparrot_training - Step 31442: {'lr': 0.00016282631643483924, 'samples': 16098816, 'steps': 31442, 'loss/train': 1.6496045589447021}
+02/25/2022 18:18:35 - INFO - codeparrot_training - Step 31443: {'lr': 0.00016281098112758949, 'samples': 16099328, 'steps': 31443, 'loss/train': 0.7996290326118469}
+02/25/2022 18:18:41 - INFO - codeparrot_training - Step 31444: {'lr': 0.0001627956461938298, 'samples': 16099840, 'steps': 31444, 'loss/train': 0.5370369553565979}
+02/25/2022 18:18:44 - INFO - codeparrot_training - Step 31445: {'lr': 0.00016278031163362595, 'samples': 16100352, 'steps': 31445, 'loss/train': 1.209165096282959}
+02/25/2022 18:18:50 - INFO - codeparrot_training - Step 31446: {'lr': 0.00016276497744704347, 'samples': 16100864, 'steps': 31446, 'loss/train': 2.0173287391662598}
+02/25/2022 18:18:53 - INFO - codeparrot_training - Step 31447: {'lr': 0.00016274964363414813, 'samples': 16101376, 'steps': 31447, 'loss/train': 0.4982140064239502}
+02/25/2022 18:18:59 - INFO - codeparrot_training - Step 31448: {'lr': 0.00016273431019500558, 'samples': 16101888, 'steps': 31448, 'loss/train': 1.2553842067718506}
+02/25/2022 18:19:02 - INFO - codeparrot_training - Step 31449: {'lr': 0.00016271897712968165, 'samples': 16102400, 'steps': 31449, 'loss/train': 0.6253923177719116}
+02/25/2022 18:19:09 - INFO - codeparrot_training - Step 31450: {'lr': 0.00016270364443824183, 'samples': 16102912, 'steps': 31450, 'loss/train': 2.9799771308898926}
+02/25/2022 18:19:12 - INFO - codeparrot_training - Step 31451: {'lr': 0.00016268831212075182, 'samples': 16103424, 'steps': 31451, 'loss/train': 1.1534920930862427}
+02/25/2022 18:19:18 - INFO - codeparrot_training - Step 31452: {'lr': 0.00016267298017727744, 'samples': 16103936, 'steps': 31452, 'loss/train': 1.5033506155014038}
+02/25/2022 18:19:21 - INFO - codeparrot_training - Step 31453: {'lr': 0.00016265764860788428, 'samples': 16104448, 'steps': 31453, 'loss/train': 1.624739170074463}
+02/25/2022 18:19:27 - INFO - codeparrot_training - Step 31454: {'lr': 0.000162642317412638, 'samples': 16104960, 'steps': 31454, 'loss/train': 2.84039306640625}
+02/25/2022 18:19:30 - INFO - codeparrot_training - Step 31455: {'lr': 0.00016262698659160428, 'samples': 16105472, 'steps': 31455, 'loss/train': 1.781067132949829}
+02/25/2022 18:19:36 - INFO - codeparrot_training - Step 31456: {'lr': 0.00016261165614484885, 'samples': 16105984, 'steps': 31456, 'loss/train': 2.2916932106018066}
+02/25/2022 18:19:39 - INFO - codeparrot_training - Step 31457: {'lr': 0.00016259632607243727, 'samples': 16106496, 'steps': 31457, 'loss/train': 1.4510235786437988}
+02/25/2022 18:19:45 - INFO - codeparrot_training - Step 31458: {'lr': 0.00016258099637443536, 'samples': 16107008, 'steps': 31458, 'loss/train': 0.9987894296646118}
+02/25/2022 18:19:48 - INFO - codeparrot_training - Step 31459: {'lr': 0.00016256566705090864, 'samples': 16107520, 'steps': 31459, 'loss/train': 2.0802817344665527}
+02/25/2022 18:19:54 - INFO - codeparrot_training - Step 31460: {'lr': 0.00016255033810192284, 'samples': 16108032, 'steps': 31460, 'loss/train': 1.784799575805664}
+02/25/2022 18:19:58 - INFO - codeparrot_training - Step 31461: {'lr': 0.00016253500952754363, 'samples': 16108544, 'steps': 31461, 'loss/train': 2.1759583950042725}
+02/25/2022 18:20:03 - INFO - codeparrot_training - Step 31462: {'lr': 0.00016251968132783667, 'samples': 16109056, 'steps': 31462, 'loss/train': 1.330784559249878}
+02/25/2022 18:20:07 - INFO - codeparrot_training - Step 31463: {'lr': 0.00016250435350286766, 'samples': 16109568, 'steps': 31463, 'loss/train': 2.2426869869232178}
+02/25/2022 18:20:12 - INFO - codeparrot_training - Step 31464: {'lr': 0.00016248902605270216, 'samples': 16110080, 'steps': 31464, 'loss/train': 2.0271759033203125}
+02/25/2022 18:20:16 - INFO - codeparrot_training - Step 31465: {'lr': 0.00016247369897740588, 'samples': 16110592, 'steps': 31465, 'loss/train': 1.1936595439910889}
+02/25/2022 18:20:22 - INFO - codeparrot_training - Step 31466: {'lr': 0.00016245837227704457, 'samples': 16111104, 'steps': 31466, 'loss/train': 2.5051722526550293}
+02/25/2022 18:20:25 - INFO - codeparrot_training - Step 31467: {'lr': 0.0001624430459516837, 'samples': 16111616, 'steps': 31467, 'loss/train': 1.6761746406555176}
+02/25/2022 18:20:31 - INFO - codeparrot_training - Step 31468: {'lr': 0.000162427720001389, 'samples': 16112128, 'steps': 31468, 'loss/train': 0.8441166281700134}
+02/25/2022 18:20:34 - INFO - codeparrot_training - Step 31469: {'lr': 0.00016241239442622618, 'samples': 16112640, 'steps': 31469, 'loss/train': 1.2163292169570923}
+02/25/2022 18:20:41 - INFO - codeparrot_training - Step 31470: {'lr': 0.00016239706922626082, 'samples': 16113152, 'steps': 31470, 'loss/train': 1.8924944400787354}
+02/25/2022 18:20:44 - INFO - codeparrot_training - Step 31471: {'lr': 0.00016238174440155872, 'samples': 16113664, 'steps': 31471, 'loss/train': 0.9663808345794678}
+02/25/2022 18:20:49 - INFO - codeparrot_training - Step 31472: {'lr': 0.0001623664199521853, 'samples': 16114176, 'steps': 31472, 'loss/train': 2.631345748901367}
+02/25/2022 18:20:53 - INFO - codeparrot_training - Step 31473: {'lr': 0.00016235109587820628, 'samples': 16114688, 'steps': 31473, 'loss/train': 1.7658894062042236}
+02/25/2022 18:20:58 - INFO - codeparrot_training - Step 31474: {'lr': 0.00016233577217968738, 'samples': 16115200, 'steps': 31474, 'loss/train': 2.008976936340332}
+02/25/2022 18:21:02 - INFO - codeparrot_training - Step 31475: {'lr': 0.00016232044885669428, 'samples': 16115712, 'steps': 31475, 'loss/train': 2.281757354736328}
+02/25/2022 18:21:08 - INFO - codeparrot_training - Step 31476: {'lr': 0.00016230512590929242, 'samples': 16116224, 'steps': 31476, 'loss/train': 1.5633317232131958}
+02/25/2022 18:21:11 - INFO - codeparrot_training - Step 31477: {'lr': 0.00016228980333754756, 'samples': 16116736, 'steps': 31477, 'loss/train': 0.8711519837379456}
+02/25/2022 18:21:17 - INFO - codeparrot_training - Step 31478: {'lr': 0.00016227448114152532, 'samples': 16117248, 'steps': 31478, 'loss/train': 0.7697080373764038}
+02/25/2022 18:21:20 - INFO - codeparrot_training - Step 31479: {'lr': 0.00016225915932129148, 'samples': 16117760, 'steps': 31479, 'loss/train': 1.5787307024002075}
+02/25/2022 18:21:26 - INFO - codeparrot_training - Step 31480: {'lr': 0.00016224383787691142, 'samples': 16118272, 'steps': 31480, 'loss/train': 1.3579158782958984}
+02/25/2022 18:21:30 - INFO - codeparrot_training - Step 31481: {'lr': 0.00016222851680845087, 'samples': 16118784, 'steps': 31481, 'loss/train': 2.413877248764038}
+02/25/2022 18:21:35 - INFO - codeparrot_training - Step 31482: {'lr': 0.0001622131961159755, 'samples': 16119296, 'steps': 31482, 'loss/train': 0.7387328147888184}
+02/25/2022 18:21:39 - INFO - codeparrot_training - Step 31483: {'lr': 0.00016219787579955096, 'samples': 16119808, 'steps': 31483, 'loss/train': 1.0056289434432983}
+02/25/2022 18:21:44 - INFO - codeparrot_training - Step 31484: {'lr': 0.0001621825558592429, 'samples': 16120320, 'steps': 31484, 'loss/train': 1.7875761985778809}
+02/25/2022 18:21:48 - INFO - codeparrot_training - Step 31485: {'lr': 0.0001621672362951168, 'samples': 16120832, 'steps': 31485, 'loss/train': 1.2262378931045532}
+02/25/2022 18:21:53 - INFO - codeparrot_training - Step 31486: {'lr': 0.00016215191710723838, 'samples': 16121344, 'steps': 31486, 'loss/train': 1.5009336471557617}
+02/25/2022 18:21:57 - INFO - codeparrot_training - Step 31487: {'lr': 0.00016213659829567324, 'samples': 16121856, 'steps': 31487, 'loss/train': 1.916374683380127}
+02/25/2022 18:22:02 - INFO - codeparrot_training - Step 31488: {'lr': 0.00016212127986048707, 'samples': 16122368, 'steps': 31488, 'loss/train': 2.270653009414673}
+02/25/2022 18:22:06 - INFO - codeparrot_training - Step 31489: {'lr': 0.0001621059618017454, 'samples': 16122880, 'steps': 31489, 'loss/train': 1.6212043762207031}
+02/25/2022 18:22:11 - INFO - codeparrot_training - Step 31490: {'lr': 0.00016209064411951384, 'samples': 16123392, 'steps': 31490, 'loss/train': 1.4815562963485718}
+02/25/2022 18:22:15 - INFO - codeparrot_training - Step 31491: {'lr': 0.00016207532681385806, 'samples': 16123904, 'steps': 31491, 'loss/train': 2.0144875049591064}
+02/25/2022 18:22:20 - INFO - codeparrot_training - Step 31492: {'lr': 0.0001620600098848437, 'samples': 16124416, 'steps': 31492, 'loss/train': 1.268331527709961}
+02/25/2022 18:22:24 - INFO - codeparrot_training - Step 31493: {'lr': 0.00016204469333253634, 'samples': 16124928, 'steps': 31493, 'loss/train': 1.5123624801635742}
+02/25/2022 18:22:29 - INFO - codeparrot_training - Step 31494: {'lr': 0.00016202937715700157, 'samples': 16125440, 'steps': 31494, 'loss/train': 1.5088878870010376}
+02/25/2022 18:22:33 - INFO - codeparrot_training - Step 31495: {'lr': 0.00016201406135830498, 'samples': 16125952, 'steps': 31495, 'loss/train': 2.3948800563812256}
+02/25/2022 18:22:39 - INFO - codeparrot_training - Step 31496: {'lr': 0.00016199874593651227, 'samples': 16126464, 'steps': 31496, 'loss/train': 1.7384083271026611}
+02/25/2022 18:22:42 - INFO - codeparrot_training - Step 31497: {'lr': 0.00016198343089168899, 'samples': 16126976, 'steps': 31497, 'loss/train': 2.2056689262390137}
+02/25/2022 18:22:48 - INFO - codeparrot_training - Step 31498: {'lr': 0.00016196811622390067, 'samples': 16127488, 'steps': 31498, 'loss/train': 2.7153568267822266}
+02/25/2022 18:22:51 - INFO - codeparrot_training - Step 31499: {'lr': 0.00016195280193321304, 'samples': 16128000, 'steps': 31499, 'loss/train': 1.393386721611023}
+02/25/2022 18:22:57 - INFO - codeparrot_training - Step 31500: {'lr': 0.00016193748801969163, 'samples': 16128512, 'steps': 31500, 'loss/train': 1.6365031003952026}
+02/25/2022 18:23:00 - INFO - codeparrot_training - Step 31501: {'lr': 0.0001619221744834021, 'samples': 16129024, 'steps': 31501, 'loss/train': 2.2532615661621094}
+02/25/2022 18:23:06 - INFO - codeparrot_training - Step 31502: {'lr': 0.00016190686132440995, 'samples': 16129536, 'steps': 31502, 'loss/train': 1.7475076913833618}
+02/25/2022 18:23:09 - INFO - codeparrot_training - Step 31503: {'lr': 0.00016189154854278089, 'samples': 16130048, 'steps': 31503, 'loss/train': 2.1678028106689453}
+02/25/2022 18:23:15 - INFO - codeparrot_training - Step 31504: {'lr': 0.00016187623613858038, 'samples': 16130560, 'steps': 31504, 'loss/train': 1.0537303686141968}
+02/25/2022 18:23:18 - INFO - codeparrot_training - Step 31505: {'lr': 0.00016186092411187413, 'samples': 16131072, 'steps': 31505, 'loss/train': 1.9704115390777588}
+02/25/2022 18:23:24 - INFO - codeparrot_training - Step 31506: {'lr': 0.00016184561246272778, 'samples': 16131584, 'steps': 31506, 'loss/train': 1.3763046264648438}
+02/25/2022 18:23:28 - INFO - codeparrot_training - Step 31507: {'lr': 0.00016183030119120673, 'samples': 16132096, 'steps': 31507, 'loss/train': 3.0875656604766846}
+02/25/2022 18:23:33 - INFO - codeparrot_training - Step 31508: {'lr': 0.0001618149902973767, 'samples': 16132608, 'steps': 31508, 'loss/train': 1.5790398120880127}
+02/25/2022 18:23:37 - INFO - codeparrot_training - Step 31509: {'lr': 0.00016179967978130317, 'samples': 16133120, 'steps': 31509, 'loss/train': 1.7692136764526367}
+02/25/2022 18:23:42 - INFO - codeparrot_training - Step 31510: {'lr': 0.00016178436964305197, 'samples': 16133632, 'steps': 31510, 'loss/train': 1.2031954526901245}
+02/25/2022 18:23:46 - INFO - codeparrot_training - Step 31511: {'lr': 0.0001617690598826884, 'samples': 16134144, 'steps': 31511, 'loss/train': 1.9572832584381104}
+02/25/2022 18:23:51 - INFO - codeparrot_training - Step 31512: {'lr': 0.00016175375050027818, 'samples': 16134656, 'steps': 31512, 'loss/train': 1.5301895141601562}
+02/25/2022 18:23:55 - INFO - codeparrot_training - Step 31513: {'lr': 0.00016173844149588684, 'samples': 16135168, 'steps': 31513, 'loss/train': 1.6924278736114502}
+02/25/2022 18:24:00 - INFO - codeparrot_training - Step 31514: {'lr': 0.00016172313286958012, 'samples': 16135680, 'steps': 31514, 'loss/train': 1.571447730064392}
+02/25/2022 18:24:04 - INFO - codeparrot_training - Step 31515: {'lr': 0.00016170782462142337, 'samples': 16136192, 'steps': 31515, 'loss/train': 1.613728404045105}
+02/25/2022 18:24:10 - INFO - codeparrot_training - Step 31516: {'lr': 0.00016169251675148222, 'samples': 16136704, 'steps': 31516, 'loss/train': 1.9871997833251953}
+02/25/2022 18:24:14 - INFO - codeparrot_training - Step 31517: {'lr': 0.00016167720925982232, 'samples': 16137216, 'steps': 31517, 'loss/train': 1.9205714464187622}
+02/25/2022 18:24:19 - INFO - codeparrot_training - Step 31518: {'lr': 0.00016166190214650923, 'samples': 16137728, 'steps': 31518, 'loss/train': 1.247193455696106}
+02/25/2022 18:24:23 - INFO - codeparrot_training - Step 31519: {'lr': 0.0001616465954116086, 'samples': 16138240, 'steps': 31519, 'loss/train': 1.2244220972061157}
+02/25/2022 18:24:28 - INFO - codeparrot_training - Step 31520: {'lr': 0.00016163128905518576, 'samples': 16138752, 'steps': 31520, 'loss/train': 1.8758896589279175}
+02/25/2022 18:24:31 - INFO - codeparrot_training - Step 31521: {'lr': 0.0001616159830773064, 'samples': 16139264, 'steps': 31521, 'loss/train': 2.3563461303710938}
+02/25/2022 18:24:37 - INFO - codeparrot_training - Step 31522: {'lr': 0.00016160067747803618, 'samples': 16139776, 'steps': 31522, 'loss/train': 2.1000547409057617}
+02/25/2022 18:24:40 - INFO - codeparrot_training - Step 31523: {'lr': 0.00016158537225744064, 'samples': 16140288, 'steps': 31523, 'loss/train': 2.04306697845459}
+02/25/2022 18:24:46 - INFO - codeparrot_training - Step 31524: {'lr': 0.0001615700674155852, 'samples': 16140800, 'steps': 31524, 'loss/train': 2.198699951171875}
+02/25/2022 18:24:49 - INFO - codeparrot_training - Step 31525: {'lr': 0.00016155476295253552, 'samples': 16141312, 'steps': 31525, 'loss/train': 1.1795037984848022}
+02/25/2022 18:24:56 - INFO - codeparrot_training - Step 31526: {'lr': 0.00016153945886835714, 'samples': 16141824, 'steps': 31526, 'loss/train': 1.513329029083252}
+02/25/2022 18:24:59 - INFO - codeparrot_training - Step 31527: {'lr': 0.0001615241551631157, 'samples': 16142336, 'steps': 31527, 'loss/train': 1.8119617700576782}
+02/25/2022 18:25:05 - INFO - codeparrot_training - Step 31528: {'lr': 0.00016150885183687663, 'samples': 16142848, 'steps': 31528, 'loss/train': 1.5955504179000854}
+02/25/2022 18:25:10 - INFO - codeparrot_training - Step 31529: {'lr': 0.00016149354888970552, 'samples': 16143360, 'steps': 31529, 'loss/train': 1.851017951965332}
+02/25/2022 18:25:14 - INFO - codeparrot_training - Step 31530: {'lr': 0.00016147824632166793, 'samples': 16143872, 'steps': 31530, 'loss/train': 2.9062321186065674}
+02/25/2022 18:25:19 - INFO - codeparrot_training - Step 31531: {'lr': 0.00016146294413282945, 'samples': 16144384, 'steps': 31531, 'loss/train': 2.50898814201355}
+02/25/2022 18:25:23 - INFO - codeparrot_training - Step 31532: {'lr': 0.0001614476423232557, 'samples': 16144896, 'steps': 31532, 'loss/train': 1.60396146774292}
+02/25/2022 18:25:29 - INFO - codeparrot_training - Step 31533: {'lr': 0.00016143234089301202, 'samples': 16145408, 'steps': 31533, 'loss/train': 1.9290982484817505}
+02/25/2022 18:25:32 - INFO - codeparrot_training - Step 31534: {'lr': 0.00016141703984216406, 'samples': 16145920, 'steps': 31534, 'loss/train': 2.207388162612915}
+02/25/2022 18:25:36 - INFO - codeparrot_training - Step 31535: {'lr': 0.00016140173917077738, 'samples': 16146432, 'steps': 31535, 'loss/train': 2.1177854537963867}
+02/25/2022 18:25:41 - INFO - codeparrot_training - Step 31536: {'lr': 0.00016138643887891763, 'samples': 16146944, 'steps': 31536, 'loss/train': 2.578043222427368}
+02/25/2022 18:25:45 - INFO - codeparrot_training - Step 31537: {'lr': 0.00016137113896665011, 'samples': 16147456, 'steps': 31537, 'loss/train': 0.777461051940918}
+02/25/2022 18:25:50 - INFO - codeparrot_training - Step 31538: {'lr': 0.00016135583943404052, 'samples': 16147968, 'steps': 31538, 'loss/train': 0.9611170291900635}
+02/25/2022 18:25:54 - INFO - codeparrot_training - Step 31539: {'lr': 0.00016134054028115442, 'samples': 16148480, 'steps': 31539, 'loss/train': 1.5688283443450928}
+02/25/2022 18:25:59 - INFO - codeparrot_training - Step 31540: {'lr': 0.00016132524150805726, 'samples': 16148992, 'steps': 31540, 'loss/train': 1.6839861869812012}
+02/25/2022 18:26:03 - INFO - codeparrot_training - Step 31541: {'lr': 0.00016130994311481462, 'samples': 16149504, 'steps': 31541, 'loss/train': 1.8979129791259766}
+02/25/2022 18:26:09 - INFO - codeparrot_training - Step 31542: {'lr': 0.00016129464510149195, 'samples': 16150016, 'steps': 31542, 'loss/train': 1.0789316892623901}
+02/25/2022 18:26:13 - INFO - codeparrot_training - Step 31543: {'lr': 0.00016127934746815493, 'samples': 16150528, 'steps': 31543, 'loss/train': 1.0093352794647217}
+02/25/2022 18:26:18 - INFO - codeparrot_training - Step 31544: {'lr': 0.00016126405021486896, 'samples': 16151040, 'steps': 31544, 'loss/train': 2.8517873287200928}
+02/25/2022 18:26:22 - INFO - codeparrot_training - Step 31545: {'lr': 0.0001612487533416997, 'samples': 16151552, 'steps': 31545, 'loss/train': 1.5938973426818848}
+02/25/2022 18:26:27 - INFO - codeparrot_training - Step 31546: {'lr': 0.00016123345684871255, 'samples': 16152064, 'steps': 31546, 'loss/train': 1.7922611236572266}
+02/25/2022 18:26:31 - INFO - codeparrot_training - Step 31547: {'lr': 0.00016121816073597306, 'samples': 16152576, 'steps': 31547, 'loss/train': 1.693039059638977}
+02/25/2022 18:26:36 - INFO - codeparrot_training - Step 31548: {'lr': 0.00016120286500354678, 'samples': 16153088, 'steps': 31548, 'loss/train': 2.9492082595825195}
+02/25/2022 18:26:40 - INFO - codeparrot_training - Step 31549: {'lr': 0.00016118756965149928, 'samples': 16153600, 'steps': 31549, 'loss/train': 2.2102174758911133}
+02/25/2022 18:26:45 - INFO - codeparrot_training - Step 31550: {'lr': 0.00016117227467989602, 'samples': 16154112, 'steps': 31550, 'loss/train': 1.0934679508209229}
+02/25/2022 18:26:49 - INFO - codeparrot_training - Step 31551: {'lr': 0.00016115698008880251, 'samples': 16154624, 'steps': 31551, 'loss/train': 1.2507045269012451}
+02/25/2022 18:26:54 - INFO - codeparrot_training - Step 31552: {'lr': 0.00016114168587828427, 'samples': 16155136, 'steps': 31552, 'loss/train': 1.506596326828003}
+02/25/2022 18:26:58 - INFO - codeparrot_training - Step 31553: {'lr': 0.0001611263920484068, 'samples': 16155648, 'steps': 31553, 'loss/train': 2.0190672874450684}
+02/25/2022 18:27:04 - INFO - codeparrot_training - Step 31554: {'lr': 0.0001611110985992358, 'samples': 16156160, 'steps': 31554, 'loss/train': 2.8666253089904785}
+02/25/2022 18:27:07 - INFO - codeparrot_training - Step 31555: {'lr': 0.00016109580553083647, 'samples': 16156672, 'steps': 31555, 'loss/train': 1.9381645917892456}
+02/25/2022 18:27:13 - INFO - codeparrot_training - Step 31556: {'lr': 0.00016108051284327452, 'samples': 16157184, 'steps': 31556, 'loss/train': 1.572195053100586}
+02/25/2022 18:27:16 - INFO - codeparrot_training - Step 31557: {'lr': 0.0001610652205366154, 'samples': 16157696, 'steps': 31557, 'loss/train': 1.6516542434692383}
+02/25/2022 18:27:22 - INFO - codeparrot_training - Step 31558: {'lr': 0.00016104992861092475, 'samples': 16158208, 'steps': 31558, 'loss/train': 1.7021163702011108}
+02/25/2022 18:27:25 - INFO - codeparrot_training - Step 31559: {'lr': 0.00016103463706626783, 'samples': 16158720, 'steps': 31559, 'loss/train': 1.7618345022201538}
+02/25/2022 18:27:31 - INFO - codeparrot_training - Step 31560: {'lr': 0.0001610193459027103, 'samples': 16159232, 'steps': 31560, 'loss/train': 2.249755382537842}
+02/25/2022 18:27:37 - INFO - codeparrot_training - Step 31561: {'lr': 0.00016100405512031762, 'samples': 16159744, 'steps': 31561, 'loss/train': 2.2130277156829834}
+02/25/2022 18:27:40 - INFO - codeparrot_training - Step 31562: {'lr': 0.00016098876471915547, 'samples': 16160256, 'steps': 31562, 'loss/train': 1.775108814239502}
+02/25/2022 18:27:46 - INFO - codeparrot_training - Step 31563: {'lr': 0.00016097347469928904, 'samples': 16160768, 'steps': 31563, 'loss/train': 2.009779930114746}
+02/25/2022 18:27:50 - INFO - codeparrot_training - Step 31564: {'lr': 0.00016095818506078397, 'samples': 16161280, 'steps': 31564, 'loss/train': 1.1456416845321655}
+02/25/2022 18:27:55 - INFO - codeparrot_training - Step 31565: {'lr': 0.00016094289580370576, 'samples': 16161792, 'steps': 31565, 'loss/train': 3.4641997814178467}
+02/25/2022 18:27:59 - INFO - codeparrot_training - Step 31566: {'lr': 0.00016092760692811993, 'samples': 16162304, 'steps': 31566, 'loss/train': 2.318700075149536}
+02/25/2022 18:28:04 - INFO - codeparrot_training - Step 31567: {'lr': 0.000160912318434092, 'samples': 16162816, 'steps': 31567, 'loss/train': 2.204702854156494}
+02/25/2022 18:28:08 - INFO - codeparrot_training - Step 31568: {'lr': 0.00016089703032168734, 'samples': 16163328, 'steps': 31568, 'loss/train': 1.3072080612182617}
+02/25/2022 18:28:13 - INFO - codeparrot_training - Step 31569: {'lr': 0.00016088174259097152, 'samples': 16163840, 'steps': 31569, 'loss/train': 0.9428216218948364}
+02/25/2022 18:28:17 - INFO - codeparrot_training - Step 31570: {'lr': 0.00016086645524200998, 'samples': 16164352, 'steps': 31570, 'loss/train': 0.3764359652996063}
+02/25/2022 18:28:22 - INFO - codeparrot_training - Step 31571: {'lr': 0.0001608511682748684, 'samples': 16164864, 'steps': 31571, 'loss/train': 1.476344108581543}
+02/25/2022 18:28:26 - INFO - codeparrot_training - Step 31572: {'lr': 0.00016083588168961196, 'samples': 16165376, 'steps': 31572, 'loss/train': 1.2195860147476196}
+02/25/2022 18:28:32 - INFO - codeparrot_training - Step 31573: {'lr': 0.0001608205954863063, 'samples': 16165888, 'steps': 31573, 'loss/train': 3.0609970092773438}
+02/25/2022 18:28:35 - INFO - codeparrot_training - Step 31574: {'lr': 0.0001608053096650169, 'samples': 16166400, 'steps': 31574, 'loss/train': 0.3347488045692444}
+02/25/2022 18:28:41 - INFO - codeparrot_training - Step 31575: {'lr': 0.00016079002422580936, 'samples': 16166912, 'steps': 31575, 'loss/train': 1.8849388360977173}
+02/25/2022 18:28:45 - INFO - codeparrot_training - Step 31576: {'lr': 0.0001607747391687489, 'samples': 16167424, 'steps': 31576, 'loss/train': 0.7627377510070801}
+02/25/2022 18:28:50 - INFO - codeparrot_training - Step 31577: {'lr': 0.0001607594544939011, 'samples': 16167936, 'steps': 31577, 'loss/train': 1.2533965110778809}
+02/25/2022 18:28:54 - INFO - codeparrot_training - Step 31578: {'lr': 0.00016074417020133148, 'samples': 16168448, 'steps': 31578, 'loss/train': 1.9179221391677856}
+02/25/2022 18:28:59 - INFO - codeparrot_training - Step 31579: {'lr': 0.00016072888629110554, 'samples': 16168960, 'steps': 31579, 'loss/train': 1.9761630296707153}
+02/25/2022 18:29:03 - INFO - codeparrot_training - Step 31580: {'lr': 0.00016071360276328874, 'samples': 16169472, 'steps': 31580, 'loss/train': 1.5991648435592651}
+02/25/2022 18:29:08 - INFO - codeparrot_training - Step 31581: {'lr': 0.0001606983196179464, 'samples': 16169984, 'steps': 31581, 'loss/train': 1.6161880493164062}
+02/25/2022 18:29:11 - INFO - codeparrot_training - Step 31582: {'lr': 0.00016068303685514413, 'samples': 16170496, 'steps': 31582, 'loss/train': 1.508562445640564}
+02/25/2022 18:29:17 - INFO - codeparrot_training - Step 31583: {'lr': 0.00016066775447494744, 'samples': 16171008, 'steps': 31583, 'loss/train': 1.707594394683838}
+02/25/2022 18:29:20 - INFO - codeparrot_training - Step 31584: {'lr': 0.0001606524724774217, 'samples': 16171520, 'steps': 31584, 'loss/train': 2.6192173957824707}
+02/25/2022 18:29:26 - INFO - codeparrot_training - Step 31585: {'lr': 0.00016063719086263235, 'samples': 16172032, 'steps': 31585, 'loss/train': 1.3823825120925903}
+02/25/2022 18:29:29 - INFO - codeparrot_training - Step 31586: {'lr': 0.00016062190963064496, 'samples': 16172544, 'steps': 31586, 'loss/train': 1.595389485359192}
+02/25/2022 18:29:35 - INFO - codeparrot_training - Step 31587: {'lr': 0.00016060662878152488, 'samples': 16173056, 'steps': 31587, 'loss/train': 1.7498188018798828}
+02/25/2022 18:29:38 - INFO - codeparrot_training - Step 31588: {'lr': 0.00016059134831533767, 'samples': 16173568, 'steps': 31588, 'loss/train': 1.6914992332458496}
+02/25/2022 18:29:45 - INFO - codeparrot_training - Step 31589: {'lr': 0.00016057606823214867, 'samples': 16174080, 'steps': 31589, 'loss/train': 1.4717464447021484}
+02/25/2022 18:29:49 - INFO - codeparrot_training - Step 31590: {'lr': 0.00016056078853202344, 'samples': 16174592, 'steps': 31590, 'loss/train': 1.5320957899093628}
+02/25/2022 18:29:54 - INFO - codeparrot_training - Step 31591: {'lr': 0.00016054550921502735, 'samples': 16175104, 'steps': 31591, 'loss/train': 1.8790576457977295}
+02/25/2022 18:29:57 - INFO - codeparrot_training - Step 31592: {'lr': 0.00016053023028122587, 'samples': 16175616, 'steps': 31592, 'loss/train': 1.9117008447647095}
+02/25/2022 18:30:03 - INFO - codeparrot_training - Step 31593: {'lr': 0.00016051495173068458, 'samples': 16176128, 'steps': 31593, 'loss/train': 2.3490333557128906}
+02/25/2022 18:30:07 - INFO - codeparrot_training - Step 31594: {'lr': 0.0001604996735634688, 'samples': 16176640, 'steps': 31594, 'loss/train': 1.184212565422058}
+02/25/2022 18:30:12 - INFO - codeparrot_training - Step 31595: {'lr': 0.0001604843957796439, 'samples': 16177152, 'steps': 31595, 'loss/train': 2.103588104248047}
+02/25/2022 18:30:16 - INFO - codeparrot_training - Step 31596: {'lr': 0.0001604691183792755, 'samples': 16177664, 'steps': 31596, 'loss/train': 1.6357944011688232}
+02/25/2022 18:30:21 - INFO - codeparrot_training - Step 31597: {'lr': 0.00016045384136242902, 'samples': 16178176, 'steps': 31597, 'loss/train': 1.512934923171997}
+02/25/2022 18:30:25 - INFO - codeparrot_training - Step 31598: {'lr': 0.00016043856472916976, 'samples': 16178688, 'steps': 31598, 'loss/train': 1.5081853866577148}
+02/25/2022 18:30:32 - INFO - codeparrot_training - Step 31599: {'lr': 0.00016042328847956324, 'samples': 16179200, 'steps': 31599, 'loss/train': 2.2163279056549072}
+02/25/2022 18:30:35 - INFO - codeparrot_training - Step 31600: {'lr': 0.00016040801261367493, 'samples': 16179712, 'steps': 31600, 'loss/train': 0.700271725654602}
+02/25/2022 18:30:41 - INFO - codeparrot_training - Step 31601: {'lr': 0.00016039273713157023, 'samples': 16180224, 'steps': 31601, 'loss/train': 1.7624937295913696}
+02/25/2022 18:30:44 - INFO - codeparrot_training - Step 31602: {'lr': 0.00016037746203331472, 'samples': 16180736, 'steps': 31602, 'loss/train': 0.7927564978599548}
+02/25/2022 18:30:50 - INFO - codeparrot_training - Step 31603: {'lr': 0.00016036218731897356, 'samples': 16181248, 'steps': 31603, 'loss/train': 2.5134196281433105}
+02/25/2022 18:30:53 - INFO - codeparrot_training - Step 31604: {'lr': 0.00016034691298861238, 'samples': 16181760, 'steps': 31604, 'loss/train': 1.9081127643585205}
+02/25/2022 18:30:58 - INFO - codeparrot_training - Step 31605: {'lr': 0.00016033163904229648, 'samples': 16182272, 'steps': 31605, 'loss/train': 2.1580874919891357}
+02/25/2022 18:31:02 - INFO - codeparrot_training - Step 31606: {'lr': 0.00016031636548009153, 'samples': 16182784, 'steps': 31606, 'loss/train': 1.7244893312454224}
+02/25/2022 18:31:08 - INFO - codeparrot_training - Step 31607: {'lr': 0.00016030109230206265, 'samples': 16183296, 'steps': 31607, 'loss/train': 1.2768585681915283}
+02/25/2022 18:31:11 - INFO - codeparrot_training - Step 31608: {'lr': 0.00016028581950827542, 'samples': 16183808, 'steps': 31608, 'loss/train': 0.056688010692596436}
+02/25/2022 18:31:17 - INFO - codeparrot_training - Step 31609: {'lr': 0.00016027054709879529, 'samples': 16184320, 'steps': 31609, 'loss/train': 1.9658455848693848}
+02/25/2022 18:31:21 - INFO - codeparrot_training - Step 31610: {'lr': 0.0001602552750736877, 'samples': 16184832, 'steps': 31610, 'loss/train': 1.5458626747131348}
+02/25/2022 18:31:26 - INFO - codeparrot_training - Step 31611: {'lr': 0.00016024000343301792, 'samples': 16185344, 'steps': 31611, 'loss/train': 1.6033605337142944}
+02/25/2022 18:31:30 - INFO - codeparrot_training - Step 31612: {'lr': 0.0001602247321768515, 'samples': 16185856, 'steps': 31612, 'loss/train': 1.0204944610595703}
+02/25/2022 18:31:35 - INFO - codeparrot_training - Step 31613: {'lr': 0.00016020946130525376, 'samples': 16186368, 'steps': 31613, 'loss/train': 2.261725425720215}
+02/25/2022 18:31:39 - INFO - codeparrot_training - Step 31614: {'lr': 0.00016019419081829024, 'samples': 16186880, 'steps': 31614, 'loss/train': 1.6130387783050537}
+02/25/2022 18:31:44 - INFO - codeparrot_training - Step 31615: {'lr': 0.00016017892071602637, 'samples': 16187392, 'steps': 31615, 'loss/train': 1.3018862009048462}
+02/25/2022 18:31:48 - INFO - codeparrot_training - Step 31616: {'lr': 0.00016016365099852736, 'samples': 16187904, 'steps': 31616, 'loss/train': 1.9669933319091797}
+02/25/2022 18:31:53 - INFO - codeparrot_training - Step 31617: {'lr': 0.00016014838166585873, 'samples': 16188416, 'steps': 31617, 'loss/train': 1.4091639518737793}
+02/25/2022 18:31:57 - INFO - codeparrot_training - Step 31618: {'lr': 0.00016013311271808594, 'samples': 16188928, 'steps': 31618, 'loss/train': 1.8106836080551147}
+02/25/2022 18:32:03 - INFO - codeparrot_training - Step 31619: {'lr': 0.00016011784415527445, 'samples': 16189440, 'steps': 31619, 'loss/train': 1.875355839729309}
+02/25/2022 18:32:06 - INFO - codeparrot_training - Step 31620: {'lr': 0.00016010257597748944, 'samples': 16189952, 'steps': 31620, 'loss/train': 1.0046404600143433}
+02/25/2022 18:32:12 - INFO - codeparrot_training - Step 31621: {'lr': 0.0001600873081847965, 'samples': 16190464, 'steps': 31621, 'loss/train': 1.0814626216888428}
+02/25/2022 18:32:15 - INFO - codeparrot_training - Step 31622: {'lr': 0.00016007204077726094, 'samples': 16190976, 'steps': 31622, 'loss/train': 2.0401828289031982}
+02/25/2022 18:32:21 - INFO - codeparrot_training - Step 31623: {'lr': 0.00016005677375494835, 'samples': 16191488, 'steps': 31623, 'loss/train': 1.9461112022399902}
+02/25/2022 18:32:24 - INFO - codeparrot_training - Step 31624: {'lr': 0.00016004150711792383, 'samples': 16192000, 'steps': 31624, 'loss/train': 0.0671166405081749}
+02/25/2022 18:32:30 - INFO - codeparrot_training - Step 31625: {'lr': 0.00016002624086625296, 'samples': 16192512, 'steps': 31625, 'loss/train': 2.3351097106933594}
+02/25/2022 18:32:33 - INFO - codeparrot_training - Step 31626: {'lr': 0.00016001097500000113, 'samples': 16193024, 'steps': 31626, 'loss/train': 1.3996005058288574}
+02/25/2022 18:32:39 - INFO - codeparrot_training - Step 31627: {'lr': 0.0001599957095192337, 'samples': 16193536, 'steps': 31627, 'loss/train': 2.0448317527770996}
+02/25/2022 18:32:42 - INFO - codeparrot_training - Step 31628: {'lr': 0.0001599804444240161, 'samples': 16194048, 'steps': 31628, 'loss/train': 2.384078025817871}
+02/25/2022 18:32:48 - INFO - codeparrot_training - Step 31629: {'lr': 0.00015996517971441363, 'samples': 16194560, 'steps': 31629, 'loss/train': 1.5013035535812378}
+02/25/2022 18:32:51 - INFO - codeparrot_training - Step 31630: {'lr': 0.0001599499153904918, 'samples': 16195072, 'steps': 31630, 'loss/train': 1.3970874547958374}
+02/25/2022 18:32:57 - INFO - codeparrot_training - Step 31631: {'lr': 0.0001599346514523159, 'samples': 16195584, 'steps': 31631, 'loss/train': 0.038425154983997345}
+02/25/2022 18:33:00 - INFO - codeparrot_training - Step 31632: {'lr': 0.00015991938789995137, 'samples': 16196096, 'steps': 31632, 'loss/train': 2.1713266372680664}
+02/25/2022 18:33:06 - INFO - codeparrot_training - Step 31633: {'lr': 0.00015990412473346354, 'samples': 16196608, 'steps': 31633, 'loss/train': 1.9602012634277344}
+02/25/2022 18:33:09 - INFO - codeparrot_training - Step 31634: {'lr': 0.0001598888619529179, 'samples': 16197120, 'steps': 31634, 'loss/train': 0.5772212147712708}
+02/25/2022 18:33:15 - INFO - codeparrot_training - Step 31635: {'lr': 0.00015987359955837972, 'samples': 16197632, 'steps': 31635, 'loss/train': 1.5683538913726807}
+02/25/2022 18:33:19 - INFO - codeparrot_training - Step 31636: {'lr': 0.00015985833754991442, 'samples': 16198144, 'steps': 31636, 'loss/train': 0.8712681531906128}
+02/25/2022 18:33:24 - INFO - codeparrot_training - Step 31637: {'lr': 0.0001598430759275874, 'samples': 16198656, 'steps': 31637, 'loss/train': 1.6394400596618652}
+02/25/2022 18:33:30 - INFO - codeparrot_training - Step 31638: {'lr': 0.000159827814691464, 'samples': 16199168, 'steps': 31638, 'loss/train': 2.8779196739196777}
+02/25/2022 18:33:33 - INFO - codeparrot_training - Step 31639: {'lr': 0.0001598125538416096, 'samples': 16199680, 'steps': 31639, 'loss/train': 0.8959330320358276}
+02/25/2022 18:33:39 - INFO - codeparrot_training - Step 31640: {'lr': 0.00015979729337808955, 'samples': 16200192, 'steps': 31640, 'loss/train': 1.1992782354354858}
+02/25/2022 18:33:43 - INFO - codeparrot_training - Step 31641: {'lr': 0.00015978203330096935, 'samples': 16200704, 'steps': 31641, 'loss/train': 2.318474769592285}
+02/25/2022 18:33:46 - INFO - codeparrot_training - Step 31642: {'lr': 0.00015976677361031422, 'samples': 16201216, 'steps': 31642, 'loss/train': 1.9080734252929688}
+02/25/2022 18:33:51 - INFO - codeparrot_training - Step 31643: {'lr': 0.00015975151430618954, 'samples': 16201728, 'steps': 31643, 'loss/train': 1.6840846538543701}
+02/25/2022 18:33:58 - INFO - codeparrot_training - Step 31644: {'lr': 0.00015973625538866072, 'samples': 16202240, 'steps': 31644, 'loss/train': 2.330674886703491}
+02/25/2022 18:34:01 - INFO - codeparrot_training - Step 31645: {'lr': 0.00015972099685779322, 'samples': 16202752, 'steps': 31645, 'loss/train': 1.6319199800491333}
+02/25/2022 18:34:07 - INFO - codeparrot_training - Step 31646: {'lr': 0.0001597057387136522, 'samples': 16203264, 'steps': 31646, 'loss/train': 1.3397094011306763}
+02/25/2022 18:34:10 - INFO - codeparrot_training - Step 31647: {'lr': 0.00015969048095630314, 'samples': 16203776, 'steps': 31647, 'loss/train': 1.62205171585083}
+02/25/2022 18:34:16 - INFO - codeparrot_training - Step 31648: {'lr': 0.00015967522358581137, 'samples': 16204288, 'steps': 31648, 'loss/train': 0.8132610321044922}
+02/25/2022 18:34:19 - INFO - codeparrot_training - Step 31649: {'lr': 0.00015965996660224235, 'samples': 16204800, 'steps': 31649, 'loss/train': 1.024165153503418}
+02/25/2022 18:34:25 - INFO - codeparrot_training - Step 31650: {'lr': 0.00015964471000566127, 'samples': 16205312, 'steps': 31650, 'loss/train': 2.103924036026001}
+02/25/2022 18:34:28 - INFO - codeparrot_training - Step 31651: {'lr': 0.0001596294537961336, 'samples': 16205824, 'steps': 31651, 'loss/train': 2.021979808807373}
+02/25/2022 18:34:34 - INFO - codeparrot_training - Step 31652: {'lr': 0.00015961419797372455, 'samples': 16206336, 'steps': 31652, 'loss/train': 0.940769374370575}
+02/25/2022 18:34:37 - INFO - codeparrot_training - Step 31653: {'lr': 0.00015959894253849967, 'samples': 16206848, 'steps': 31653, 'loss/train': 2.1182608604431152}
+02/25/2022 18:34:43 - INFO - codeparrot_training - Step 31654: {'lr': 0.00015958368749052426, 'samples': 16207360, 'steps': 31654, 'loss/train': 1.481539011001587}
+02/25/2022 18:34:47 - INFO - codeparrot_training - Step 31655: {'lr': 0.00015956843282986354, 'samples': 16207872, 'steps': 31655, 'loss/train': 2.612964391708374}
+02/25/2022 18:34:52 - INFO - codeparrot_training - Step 31656: {'lr': 0.00015955317855658298, 'samples': 16208384, 'steps': 31656, 'loss/train': 1.4114587306976318}
+02/25/2022 18:34:56 - INFO - codeparrot_training - Step 31657: {'lr': 0.00015953792467074778, 'samples': 16208896, 'steps': 31657, 'loss/train': 1.0481922626495361}
+02/25/2022 18:35:02 - INFO - codeparrot_training - Step 31658: {'lr': 0.0001595226711724236, 'samples': 16209408, 'steps': 31658, 'loss/train': 1.016542911529541}
+02/25/2022 18:35:06 - INFO - codeparrot_training - Step 31659: {'lr': 0.0001595074180616754, 'samples': 16209920, 'steps': 31659, 'loss/train': 2.3109352588653564}
+02/25/2022 18:35:09 - INFO - codeparrot_training - Step 31660: {'lr': 0.0001594921653385687, 'samples': 16210432, 'steps': 31660, 'loss/train': 2.574134111404419}
+02/25/2022 18:35:14 - INFO - codeparrot_training - Step 31661: {'lr': 0.0001594769130031688, 'samples': 16210944, 'steps': 31661, 'loss/train': 2.1193525791168213}
+02/25/2022 18:35:18 - INFO - codeparrot_training - Step 31662: {'lr': 0.0001594616610555411, 'samples': 16211456, 'steps': 31662, 'loss/train': 2.043381452560425}
+02/25/2022 18:35:24 - INFO - codeparrot_training - Step 31663: {'lr': 0.000159446409495751, 'samples': 16211968, 'steps': 31663, 'loss/train': 1.0986430644989014}
+02/25/2022 18:35:27 - INFO - codeparrot_training - Step 31664: {'lr': 0.0001594311583238636, 'samples': 16212480, 'steps': 31664, 'loss/train': 1.4999703168869019}
+02/25/2022 18:35:33 - INFO - codeparrot_training - Step 31665: {'lr': 0.00015941590753994434, 'samples': 16212992, 'steps': 31665, 'loss/train': 8.629182815551758}
+02/25/2022 18:35:36 - INFO - codeparrot_training - Step 31666: {'lr': 0.0001594006571440586, 'samples': 16213504, 'steps': 31666, 'loss/train': 1.6565648317337036}
+02/25/2022 18:35:42 - INFO - codeparrot_training - Step 31667: {'lr': 0.00015938540713627178, 'samples': 16214016, 'steps': 31667, 'loss/train': 1.8821903467178345}
+02/25/2022 18:35:46 - INFO - codeparrot_training - Step 31668: {'lr': 0.00015937015751664902, 'samples': 16214528, 'steps': 31668, 'loss/train': 1.2966912984848022}
+02/25/2022 18:35:51 - INFO - codeparrot_training - Step 31669: {'lr': 0.00015935490828525566, 'samples': 16215040, 'steps': 31669, 'loss/train': 2.6232807636260986}
+02/25/2022 18:35:55 - INFO - codeparrot_training - Step 31670: {'lr': 0.0001593396594421572, 'samples': 16215552, 'steps': 31670, 'loss/train': 3.301279067993164}
+02/25/2022 18:36:00 - INFO - codeparrot_training - Step 31671: {'lr': 0.00015932441098741884, 'samples': 16216064, 'steps': 31671, 'loss/train': 2.928297281265259}
+02/25/2022 18:36:04 - INFO - codeparrot_training - Step 31672: {'lr': 0.00015930916292110583, 'samples': 16216576, 'steps': 31672, 'loss/train': 1.9819281101226807}
+02/25/2022 18:36:09 - INFO - codeparrot_training - Step 31673: {'lr': 0.0001592939152432836, 'samples': 16217088, 'steps': 31673, 'loss/train': 1.8660078048706055}
+02/25/2022 18:36:13 - INFO - codeparrot_training - Step 31674: {'lr': 0.00015927866795401746, 'samples': 16217600, 'steps': 31674, 'loss/train': 9.75402545928955}
+02/25/2022 18:36:18 - INFO - codeparrot_training - Step 31675: {'lr': 0.00015926342105337268, 'samples': 16218112, 'steps': 31675, 'loss/train': 1.33323073387146}
+02/25/2022 18:36:22 - INFO - codeparrot_training - Step 31676: {'lr': 0.00015924817454141462, 'samples': 16218624, 'steps': 31676, 'loss/train': 2.3078315258026123}
+02/25/2022 18:36:27 - INFO - codeparrot_training - Step 31677: {'lr': 0.0001592329284182086, 'samples': 16219136, 'steps': 31677, 'loss/train': 1.7296061515808105}
+02/25/2022 18:36:31 - INFO - codeparrot_training - Step 31678: {'lr': 0.0001592176826838198, 'samples': 16219648, 'steps': 31678, 'loss/train': 1.7072107791900635}
+02/25/2022 18:36:36 - INFO - codeparrot_training - Step 31679: {'lr': 0.00015920243733831362, 'samples': 16220160, 'steps': 31679, 'loss/train': 0.7049659490585327}
+02/25/2022 18:36:40 - INFO - codeparrot_training - Step 31680: {'lr': 0.00015918719238175544, 'samples': 16220672, 'steps': 31680, 'loss/train': 2.18479061126709}
+02/25/2022 18:36:46 - INFO - codeparrot_training - Step 31681: {'lr': 0.0001591719478142105, 'samples': 16221184, 'steps': 31681, 'loss/train': 1.0697076320648193}
+02/25/2022 18:36:49 - INFO - codeparrot_training - Step 31682: {'lr': 0.00015915670363574402, 'samples': 16221696, 'steps': 31682, 'loss/train': 1.19609534740448}
+02/25/2022 18:36:55 - INFO - codeparrot_training - Step 31683: {'lr': 0.00015914145984642137, 'samples': 16222208, 'steps': 31683, 'loss/train': 1.8251248598098755}
+02/25/2022 18:36:58 - INFO - codeparrot_training - Step 31684: {'lr': 0.00015912621644630797, 'samples': 16222720, 'steps': 31684, 'loss/train': 1.3311402797698975}
+02/25/2022 18:37:04 - INFO - codeparrot_training - Step 31685: {'lr': 0.0001591109734354689, 'samples': 16223232, 'steps': 31685, 'loss/train': 0.0562140978872776}
+02/25/2022 18:37:07 - INFO - codeparrot_training - Step 31686: {'lr': 0.00015909573081396956, 'samples': 16223744, 'steps': 31686, 'loss/train': 0.07935921847820282}
+02/25/2022 18:37:13 - INFO - codeparrot_training - Step 31687: {'lr': 0.00015908048858187524, 'samples': 16224256, 'steps': 31687, 'loss/train': 1.1900615692138672}
+02/25/2022 18:37:17 - INFO - codeparrot_training - Step 31688: {'lr': 0.00015906524673925125, 'samples': 16224768, 'steps': 31688, 'loss/train': 1.9659942388534546}
+02/25/2022 18:37:22 - INFO - codeparrot_training - Step 31689: {'lr': 0.00015905000528616296, 'samples': 16225280, 'steps': 31689, 'loss/train': 1.279463768005371}
+02/25/2022 18:37:26 - INFO - codeparrot_training - Step 31690: {'lr': 0.00015903476422267547, 'samples': 16225792, 'steps': 31690, 'loss/train': 2.544485569000244}
+02/25/2022 18:37:31 - INFO - codeparrot_training - Step 31691: {'lr': 0.00015901952354885413, 'samples': 16226304, 'steps': 31691, 'loss/train': 0.4169861674308777}
+02/25/2022 18:37:35 - INFO - codeparrot_training - Step 31692: {'lr': 0.0001590042832647643, 'samples': 16226816, 'steps': 31692, 'loss/train': 1.5177192687988281}
+02/25/2022 18:37:41 - INFO - codeparrot_training - Step 31693: {'lr': 0.0001589890433704713, 'samples': 16227328, 'steps': 31693, 'loss/train': 2.1188693046569824}
+02/25/2022 18:37:45 - INFO - codeparrot_training - Step 31694: {'lr': 0.00015897380386604027, 'samples': 16227840, 'steps': 31694, 'loss/train': 2.1663472652435303}
+02/25/2022 18:37:51 - INFO - codeparrot_training - Step 31695: {'lr': 0.00015895856475153658, 'samples': 16228352, 'steps': 31695, 'loss/train': 1.648962140083313}
+02/25/2022 18:37:54 - INFO - codeparrot_training - Step 31696: {'lr': 0.00015894332602702543, 'samples': 16228864, 'steps': 31696, 'loss/train': 1.9686440229415894}
+02/25/2022 18:38:00 - INFO - codeparrot_training - Step 31697: {'lr': 0.0001589280876925723, 'samples': 16229376, 'steps': 31697, 'loss/train': 1.9946438074111938}
+02/25/2022 18:38:03 - INFO - codeparrot_training - Step 31698: {'lr': 0.00015891284974824222, 'samples': 16229888, 'steps': 31698, 'loss/train': 1.570151686668396}
+02/25/2022 18:38:09 - INFO - codeparrot_training - Step 31699: {'lr': 0.00015889761219410053, 'samples': 16230400, 'steps': 31699, 'loss/train': 2.239841938018799}
+02/25/2022 18:38:12 - INFO - codeparrot_training - Step 31700: {'lr': 0.0001588823750302126, 'samples': 16230912, 'steps': 31700, 'loss/train': 1.6195874214172363}
+02/25/2022 18:38:17 - INFO - codeparrot_training - Step 31701: {'lr': 0.00015886713825664358, 'samples': 16231424, 'steps': 31701, 'loss/train': 2.39121413230896}
+02/25/2022 18:38:21 - INFO - codeparrot_training - Step 31702: {'lr': 0.00015885190187345894, 'samples': 16231936, 'steps': 31702, 'loss/train': 1.454297423362732}
+02/25/2022 18:38:27 - INFO - codeparrot_training - Step 31703: {'lr': 0.0001588366658807237, 'samples': 16232448, 'steps': 31703, 'loss/train': 1.4006729125976562}
+02/25/2022 18:38:31 - INFO - codeparrot_training - Step 31704: {'lr': 0.00015882143027850325, 'samples': 16232960, 'steps': 31704, 'loss/train': 1.6190322637557983}
+02/25/2022 18:38:36 - INFO - codeparrot_training - Step 31705: {'lr': 0.00015880619506686283, 'samples': 16233472, 'steps': 31705, 'loss/train': 0.7933560609817505}
+02/25/2022 18:38:40 - INFO - codeparrot_training - Step 31706: {'lr': 0.0001587909602458678, 'samples': 16233984, 'steps': 31706, 'loss/train': 8.767669677734375}
+02/25/2022 18:38:45 - INFO - codeparrot_training - Step 31707: {'lr': 0.00015877572581558325, 'samples': 16234496, 'steps': 31707, 'loss/train': 1.5217331647872925}
+02/25/2022 18:38:49 - INFO - codeparrot_training - Step 31708: {'lr': 0.0001587604917760745, 'samples': 16235008, 'steps': 31708, 'loss/train': 2.2403440475463867}
+02/25/2022 18:38:54 - INFO - codeparrot_training - Step 31709: {'lr': 0.00015874525812740686, 'samples': 16235520, 'steps': 31709, 'loss/train': 0.8287572860717773}
+02/25/2022 18:38:58 - INFO - codeparrot_training - Step 31710: {'lr': 0.00015873002486964562, 'samples': 16236032, 'steps': 31710, 'loss/train': 2.5754573345184326}
+02/25/2022 18:39:03 - INFO - codeparrot_training - Step 31711: {'lr': 0.00015871479200285594, 'samples': 16236544, 'steps': 31711, 'loss/train': 0.15411376953125}
+02/25/2022 18:39:07 - INFO - codeparrot_training - Step 31712: {'lr': 0.00015869955952710308, 'samples': 16237056, 'steps': 31712, 'loss/train': 1.8623647689819336}
+02/25/2022 18:39:13 - INFO - codeparrot_training - Step 31713: {'lr': 0.00015868432744245225, 'samples': 16237568, 'steps': 31713, 'loss/train': 0.5052101016044617}
+02/25/2022 18:39:16 - INFO - codeparrot_training - Step 31714: {'lr': 0.00015866909574896887, 'samples': 16238080, 'steps': 31714, 'loss/train': 1.67640221118927}
+02/25/2022 18:39:22 - INFO - codeparrot_training - Step 31715: {'lr': 0.0001586538644467181, 'samples': 16238592, 'steps': 31715, 'loss/train': 1.3188472986221313}
+02/25/2022 18:39:25 - INFO - codeparrot_training - Step 31716: {'lr': 0.00015863863353576512, 'samples': 16239104, 'steps': 31716, 'loss/train': 1.0435198545455933}
+02/25/2022 18:39:31 - INFO - codeparrot_training - Step 31717: {'lr': 0.00015862340301617522, 'samples': 16239616, 'steps': 31717, 'loss/train': 1.3111354112625122}
+02/25/2022 18:39:34 - INFO - codeparrot_training - Step 31718: {'lr': 0.00015860817288801364, 'samples': 16240128, 'steps': 31718, 'loss/train': 1.2448370456695557}
+02/25/2022 18:39:40 - INFO - codeparrot_training - Step 31719: {'lr': 0.00015859294315134564, 'samples': 16240640, 'steps': 31719, 'loss/train': 2.3127501010894775}
+02/25/2022 18:39:43 - INFO - codeparrot_training - Step 31720: {'lr': 0.00015857771380623642, 'samples': 16241152, 'steps': 31720, 'loss/train': 1.96444571018219}
+02/25/2022 18:39:49 - INFO - codeparrot_training - Step 31721: {'lr': 0.00015856248485275132, 'samples': 16241664, 'steps': 31721, 'loss/train': 1.3329789638519287}
+02/25/2022 18:39:52 - INFO - codeparrot_training - Step 31722: {'lr': 0.00015854725629095543, 'samples': 16242176, 'steps': 31722, 'loss/train': 1.278171420097351}
+02/25/2022 18:39:59 - INFO - codeparrot_training - Step 31723: {'lr': 0.00015853202812091406, 'samples': 16242688, 'steps': 31723, 'loss/train': 0.5689896941184998}
+02/25/2022 18:40:02 - INFO - codeparrot_training - Step 31724: {'lr': 0.0001585168003426925, 'samples': 16243200, 'steps': 31724, 'loss/train': 2.6579689979553223}
+02/25/2022 18:40:08 - INFO - codeparrot_training - Step 31725: {'lr': 0.00015850157295635585, 'samples': 16243712, 'steps': 31725, 'loss/train': 1.4840666055679321}
+02/25/2022 18:40:11 - INFO - codeparrot_training - Step 31726: {'lr': 0.0001584863459619694, 'samples': 16244224, 'steps': 31726, 'loss/train': 1.70193612575531}
+02/25/2022 18:40:17 - INFO - codeparrot_training - Step 31727: {'lr': 0.0001584711193595984, 'samples': 16244736, 'steps': 31727, 'loss/train': 2.2464041709899902}
+02/25/2022 18:40:20 - INFO - codeparrot_training - Step 31728: {'lr': 0.00015845589314930815, 'samples': 16245248, 'steps': 31728, 'loss/train': 3.001596450805664}
+02/25/2022 18:40:25 - INFO - codeparrot_training - Step 31729: {'lr': 0.0001584406673311637, 'samples': 16245760, 'steps': 31729, 'loss/train': 2.058844804763794}
+02/25/2022 18:40:29 - INFO - codeparrot_training - Step 31730: {'lr': 0.0001584254419052303, 'samples': 16246272, 'steps': 31730, 'loss/train': 1.9087986946105957}
+02/25/2022 18:40:34 - INFO - codeparrot_training - Step 31731: {'lr': 0.0001584102168715733, 'samples': 16246784, 'steps': 31731, 'loss/train': 2.9825191497802734}
+02/25/2022 18:40:38 - INFO - codeparrot_training - Step 31732: {'lr': 0.0001583949922302579, 'samples': 16247296, 'steps': 31732, 'loss/train': 1.3170325756072998}
+02/25/2022 18:40:44 - INFO - codeparrot_training - Step 31733: {'lr': 0.00015837976798134916, 'samples': 16247808, 'steps': 31733, 'loss/train': 2.037655830383301}
+02/25/2022 18:40:47 - INFO - codeparrot_training - Step 31734: {'lr': 0.00015836454412491248, 'samples': 16248320, 'steps': 31734, 'loss/train': 1.617730975151062}
+02/25/2022 18:40:53 - INFO - codeparrot_training - Step 31735: {'lr': 0.00015834932066101292, 'samples': 16248832, 'steps': 31735, 'loss/train': 1.6753456592559814}
+02/25/2022 18:40:56 - INFO - codeparrot_training - Step 31736: {'lr': 0.0001583340975897158, 'samples': 16249344, 'steps': 31736, 'loss/train': 2.129549503326416}
+02/25/2022 18:41:02 - INFO - codeparrot_training - Step 31737: {'lr': 0.00015831887491108638, 'samples': 16249856, 'steps': 31737, 'loss/train': 1.9992746114730835}
+02/25/2022 18:41:05 - INFO - codeparrot_training - Step 31738: {'lr': 0.0001583036526251897, 'samples': 16250368, 'steps': 31738, 'loss/train': 0.10700304806232452}
+02/25/2022 18:41:11 - INFO - codeparrot_training - Step 31739: {'lr': 0.00015828843073209103, 'samples': 16250880, 'steps': 31739, 'loss/train': 2.000544786453247}
+02/25/2022 18:41:14 - INFO - codeparrot_training - Step 31740: {'lr': 0.00015827320923185562, 'samples': 16251392, 'steps': 31740, 'loss/train': 1.155184030532837}
+02/25/2022 18:41:20 - INFO - codeparrot_training - Step 31741: {'lr': 0.0001582579881245488, 'samples': 16251904, 'steps': 31741, 'loss/train': 1.5773965120315552}
+02/25/2022 18:41:23 - INFO - codeparrot_training - Step 31742: {'lr': 0.00015824276741023544, 'samples': 16252416, 'steps': 31742, 'loss/train': 1.9481098651885986}
+02/25/2022 18:41:29 - INFO - codeparrot_training - Step 31743: {'lr': 0.00015822754708898096, 'samples': 16252928, 'steps': 31743, 'loss/train': 1.9054780006408691}
+02/25/2022 18:41:33 - INFO - codeparrot_training - Step 31744: {'lr': 0.00015821232716085054, 'samples': 16253440, 'steps': 31744, 'loss/train': 1.865513801574707}
+02/25/2022 18:41:38 - INFO - codeparrot_training - Step 31745: {'lr': 0.00015819710762590946, 'samples': 16253952, 'steps': 31745, 'loss/train': 1.294236183166504}
+02/25/2022 18:41:41 - INFO - codeparrot_training - Step 31746: {'lr': 0.00015818188848422274, 'samples': 16254464, 'steps': 31746, 'loss/train': 1.5553377866744995}
+02/25/2022 18:41:47 - INFO - codeparrot_training - Step 31747: {'lr': 0.00015816666973585564, 'samples': 16254976, 'steps': 31747, 'loss/train': 2.326873779296875}
+02/25/2022 18:41:51 - INFO - codeparrot_training - Step 31748: {'lr': 0.00015815145138087336, 'samples': 16255488, 'steps': 31748, 'loss/train': 1.71562922000885}
+02/25/2022 18:41:57 - INFO - codeparrot_training - Step 31749: {'lr': 0.0001581362334193411, 'samples': 16256000, 'steps': 31749, 'loss/train': 2.299311637878418}
+02/25/2022 18:42:00 - INFO - codeparrot_training - Step 31750: {'lr': 0.00015812101585132416, 'samples': 16256512, 'steps': 31750, 'loss/train': 1.9052296876907349}
+02/25/2022 18:42:06 - INFO - codeparrot_training - Step 31751: {'lr': 0.00015810579867688748, 'samples': 16257024, 'steps': 31751, 'loss/train': 0.9062367081642151}
+02/25/2022 18:42:11 - INFO - codeparrot_training - Step 31752: {'lr': 0.00015809058189609644, 'samples': 16257536, 'steps': 31752, 'loss/train': 0.9063776731491089}
+02/25/2022 18:42:15 - INFO - codeparrot_training - Step 31753: {'lr': 0.00015807536550901607, 'samples': 16258048, 'steps': 31753, 'loss/train': 0.7376357913017273}
+02/25/2022 18:42:20 - INFO - codeparrot_training - Step 31754: {'lr': 0.0001580601495157118, 'samples': 16258560, 'steps': 31754, 'loss/train': 1.2490119934082031}
+02/25/2022 18:42:24 - INFO - codeparrot_training - Step 31755: {'lr': 0.00015804493391624857, 'samples': 16259072, 'steps': 31755, 'loss/train': 2.912867307662964}
+02/25/2022 18:42:27 - INFO - codeparrot_training - Step 31756: {'lr': 0.00015802971871069164, 'samples': 16259584, 'steps': 31756, 'loss/train': 2.0779659748077393}
+02/25/2022 18:42:33 - INFO - codeparrot_training - Step 31757: {'lr': 0.00015801450389910622, 'samples': 16260096, 'steps': 31757, 'loss/train': 1.400538444519043}
+02/25/2022 18:42:36 - INFO - codeparrot_training - Step 31758: {'lr': 0.00015799928948155746, 'samples': 16260608, 'steps': 31758, 'loss/train': 1.424329400062561}
+02/25/2022 18:42:42 - INFO - codeparrot_training - Step 31759: {'lr': 0.0001579840754581105, 'samples': 16261120, 'steps': 31759, 'loss/train': 2.344059944152832}
+02/25/2022 18:42:46 - INFO - codeparrot_training - Step 31760: {'lr': 0.00015796886182883053, 'samples': 16261632, 'steps': 31760, 'loss/train': 0.9455052614212036}
+02/25/2022 18:42:52 - INFO - codeparrot_training - Step 31761: {'lr': 0.00015795364859378275, 'samples': 16262144, 'steps': 31761, 'loss/train': 1.7467272281646729}
+02/25/2022 18:42:55 - INFO - codeparrot_training - Step 31762: {'lr': 0.00015793843575303228, 'samples': 16262656, 'steps': 31762, 'loss/train': 0.9557493329048157}
+02/25/2022 18:43:01 - INFO - codeparrot_training - Step 31763: {'lr': 0.0001579232233066444, 'samples': 16263168, 'steps': 31763, 'loss/train': 1.8905915021896362}
+02/25/2022 18:43:04 - INFO - codeparrot_training - Step 31764: {'lr': 0.00015790801125468418, 'samples': 16263680, 'steps': 31764, 'loss/train': 2.09942889213562}
+02/25/2022 18:43:09 - INFO - codeparrot_training - Step 31765: {'lr': 0.00015789279959721675, 'samples': 16264192, 'steps': 31765, 'loss/train': 2.7502262592315674}
+02/25/2022 18:43:13 - INFO - codeparrot_training - Step 31766: {'lr': 0.0001578775883343073, 'samples': 16264704, 'steps': 31766, 'loss/train': 1.4074475765228271}
+02/25/2022 18:43:18 - INFO - codeparrot_training - Step 31767: {'lr': 0.00015786237746602105, 'samples': 16265216, 'steps': 31767, 'loss/train': 1.8503464460372925}
+02/25/2022 18:43:22 - INFO - codeparrot_training - Step 31768: {'lr': 0.00015784716699242317, 'samples': 16265728, 'steps': 31768, 'loss/train': 2.2772557735443115}
+02/25/2022 18:43:28 - INFO - codeparrot_training - Step 31769: {'lr': 0.00015783195691357866, 'samples': 16266240, 'steps': 31769, 'loss/train': 1.5083093643188477}
+02/25/2022 18:43:31 - INFO - codeparrot_training - Step 31770: {'lr': 0.0001578167472295528, 'samples': 16266752, 'steps': 31770, 'loss/train': 1.286312222480774}
+02/25/2022 18:43:37 - INFO - codeparrot_training - Step 31771: {'lr': 0.00015780153794041075, 'samples': 16267264, 'steps': 31771, 'loss/train': 1.2470710277557373}
+02/25/2022 18:43:40 - INFO - codeparrot_training - Step 31772: {'lr': 0.0001577863290462177, 'samples': 16267776, 'steps': 31772, 'loss/train': 0.790630042552948}
+02/25/2022 18:43:46 - INFO - codeparrot_training - Step 31773: {'lr': 0.00015777112054703862, 'samples': 16268288, 'steps': 31773, 'loss/train': 1.1662843227386475}
+02/25/2022 18:43:52 - INFO - codeparrot_training - Step 31774: {'lr': 0.0001577559124429388, 'samples': 16268800, 'steps': 31774, 'loss/train': 0.5784724354743958}
+02/25/2022 18:43:55 - INFO - codeparrot_training - Step 31775: {'lr': 0.0001577407047339834, 'samples': 16269312, 'steps': 31775, 'loss/train': 1.2439956665039062}
+02/25/2022 18:44:00 - INFO - codeparrot_training - Step 31776: {'lr': 0.00015772549742023757, 'samples': 16269824, 'steps': 31776, 'loss/train': 1.5234769582748413}
+02/25/2022 18:44:04 - INFO - codeparrot_training - Step 31777: {'lr': 0.00015771029050176634, 'samples': 16270336, 'steps': 31777, 'loss/train': 2.2709524631500244}
+02/25/2022 18:44:09 - INFO - codeparrot_training - Step 31778: {'lr': 0.00015769508397863485, 'samples': 16270848, 'steps': 31778, 'loss/train': 1.2925472259521484}
+02/25/2022 18:44:13 - INFO - codeparrot_training - Step 31779: {'lr': 0.00015767987785090837, 'samples': 16271360, 'steps': 31779, 'loss/train': 1.24466073513031}
+02/25/2022 18:44:18 - INFO - codeparrot_training - Step 31780: {'lr': 0.0001576646721186521, 'samples': 16271872, 'steps': 31780, 'loss/train': 0.8119126558303833}
+02/25/2022 18:44:22 - INFO - codeparrot_training - Step 31781: {'lr': 0.00015764946678193092, 'samples': 16272384, 'steps': 31781, 'loss/train': 1.8556114435195923}
+02/25/2022 18:44:27 - INFO - codeparrot_training - Step 31782: {'lr': 0.00015763426184081008, 'samples': 16272896, 'steps': 31782, 'loss/train': 0.774508535861969}
+02/25/2022 18:44:31 - INFO - codeparrot_training - Step 31783: {'lr': 0.00015761905729535478, 'samples': 16273408, 'steps': 31783, 'loss/train': 1.894516944885254}
+02/25/2022 18:44:38 - INFO - codeparrot_training - Step 31784: {'lr': 0.00015760385314563007, 'samples': 16273920, 'steps': 31784, 'loss/train': 2.374850034713745}
+02/25/2022 18:44:41 - INFO - codeparrot_training - Step 31785: {'lr': 0.00015758864939170126, 'samples': 16274432, 'steps': 31785, 'loss/train': 1.2760549783706665}
+02/25/2022 18:44:47 - INFO - codeparrot_training - Step 31786: {'lr': 0.0001575734460336332, 'samples': 16274944, 'steps': 31786, 'loss/train': 2.0600438117980957}
+02/25/2022 18:44:51 - INFO - codeparrot_training - Step 31787: {'lr': 0.00015755824307149113, 'samples': 16275456, 'steps': 31787, 'loss/train': 2.0675594806671143}
+02/25/2022 18:44:56 - INFO - codeparrot_training - Step 31788: {'lr': 0.00015754304050534022, 'samples': 16275968, 'steps': 31788, 'loss/train': 1.8726649284362793}
+02/25/2022 18:45:00 - INFO - codeparrot_training - Step 31789: {'lr': 0.00015752783833524565, 'samples': 16276480, 'steps': 31789, 'loss/train': 2.101017951965332}
+02/25/2022 18:45:05 - INFO - codeparrot_training - Step 31790: {'lr': 0.00015751263656127234, 'samples': 16276992, 'steps': 31790, 'loss/train': 1.1056339740753174}
+02/25/2022 18:45:09 - INFO - codeparrot_training - Step 31791: {'lr': 0.0001574974351834856, 'samples': 16277504, 'steps': 31791, 'loss/train': 1.7317559719085693}
+02/25/2022 18:45:14 - INFO - codeparrot_training - Step 31792: {'lr': 0.00015748223420195042, 'samples': 16278016, 'steps': 31792, 'loss/train': 1.4771580696105957}
+02/25/2022 18:45:17 - INFO - codeparrot_training - Step 31793: {'lr': 0.00015746703361673207, 'samples': 16278528, 'steps': 31793, 'loss/train': 1.4978071451187134}
+02/25/2022 18:45:24 - INFO - codeparrot_training - Step 31794: {'lr': 0.00015745183342789546, 'samples': 16279040, 'steps': 31794, 'loss/train': 1.4167848825454712}
+02/25/2022 18:45:28 - INFO - codeparrot_training - Step 31795: {'lr': 0.0001574366336355058, 'samples': 16279552, 'steps': 31795, 'loss/train': 0.8790462613105774}
+02/25/2022 18:45:33 - INFO - codeparrot_training - Step 31796: {'lr': 0.00015742143423962823, 'samples': 16280064, 'steps': 31796, 'loss/train': 1.5654898881912231}
+02/25/2022 18:45:37 - INFO - codeparrot_training - Step 31797: {'lr': 0.00015740623524032794, 'samples': 16280576, 'steps': 31797, 'loss/train': 1.3439390659332275}
+02/25/2022 18:45:42 - INFO - codeparrot_training - Step 31798: {'lr': 0.00015739103663766984, 'samples': 16281088, 'steps': 31798, 'loss/train': 0.5737715363502502}
+02/25/2022 18:45:45 - INFO - codeparrot_training - Step 31799: {'lr': 0.00015737583843171914, 'samples': 16281600, 'steps': 31799, 'loss/train': 1.4663745164871216}
+02/25/2022 18:45:51 - INFO - codeparrot_training - Step 31800: {'lr': 0.00015736064062254093, 'samples': 16282112, 'steps': 31800, 'loss/train': 1.7405738830566406}
+02/25/2022 18:45:55 - INFO - codeparrot_training - Step 31801: {'lr': 0.00015734544321020037, 'samples': 16282624, 'steps': 31801, 'loss/train': 1.870730996131897}
+02/25/2022 18:46:00 - INFO - codeparrot_training - Step 31802: {'lr': 0.0001573302461947625, 'samples': 16283136, 'steps': 31802, 'loss/train': 2.4010608196258545}
+02/25/2022 18:46:03 - INFO - codeparrot_training - Step 31803: {'lr': 0.0001573150495762924, 'samples': 16283648, 'steps': 31803, 'loss/train': 2.220834493637085}
+02/25/2022 18:46:10 - INFO - codeparrot_training - Step 31804: {'lr': 0.0001572998533548552, 'samples': 16284160, 'steps': 31804, 'loss/train': 2.6818807125091553}
+02/25/2022 18:46:14 - INFO - codeparrot_training - Step 31805: {'lr': 0.00015728465753051598, 'samples': 16284672, 'steps': 31805, 'loss/train': 1.5541250705718994}
+02/25/2022 18:46:19 - INFO - codeparrot_training - Step 31806: {'lr': 0.0001572694621033399, 'samples': 16285184, 'steps': 31806, 'loss/train': 1.9580897092819214}
+02/25/2022 18:46:23 - INFO - codeparrot_training - Step 31807: {'lr': 0.00015725426707339195, 'samples': 16285696, 'steps': 31807, 'loss/train': 2.7889318466186523}
+02/25/2022 18:46:28 - INFO - codeparrot_training - Step 31808: {'lr': 0.0001572390724407373, 'samples': 16286208, 'steps': 31808, 'loss/train': 1.3352032899856567}
+02/25/2022 18:46:32 - INFO - codeparrot_training - Step 31809: {'lr': 0.00015722387820544097, 'samples': 16286720, 'steps': 31809, 'loss/train': 1.9415841102600098}
+02/25/2022 18:46:37 - INFO - codeparrot_training - Step 31810: {'lr': 0.00015720868436756808, 'samples': 16287232, 'steps': 31810, 'loss/train': 1.6297929286956787}
+02/25/2022 18:46:41 - INFO - codeparrot_training - Step 31811: {'lr': 0.00015719349092718383, 'samples': 16287744, 'steps': 31811, 'loss/train': 1.4024025201797485}
+02/25/2022 18:46:46 - INFO - codeparrot_training - Step 31812: {'lr': 0.00015717829788435307, 'samples': 16288256, 'steps': 31812, 'loss/train': 2.3132739067077637}
+02/25/2022 18:46:50 - INFO - codeparrot_training - Step 31813: {'lr': 0.00015716310523914104, 'samples': 16288768, 'steps': 31813, 'loss/train': 1.8912065029144287}
+02/25/2022 18:46:55 - INFO - codeparrot_training - Step 31814: {'lr': 0.00015714791299161276, 'samples': 16289280, 'steps': 31814, 'loss/train': 1.8868147134780884}
+02/25/2022 18:47:02 - INFO - codeparrot_training - Step 31815: {'lr': 0.00015713272114183346, 'samples': 16289792, 'steps': 31815, 'loss/train': 0.5052993893623352}
+02/25/2022 18:47:05 - INFO - codeparrot_training - Step 31816: {'lr': 0.00015711752968986797, 'samples': 16290304, 'steps': 31816, 'loss/train': 1.3647977113723755}
+02/25/2022 18:47:11 - INFO - codeparrot_training - Step 31817: {'lr': 0.00015710233863578153, 'samples': 16290816, 'steps': 31817, 'loss/train': 2.4201929569244385}
+02/25/2022 18:47:14 - INFO - codeparrot_training - Step 31818: {'lr': 0.00015708714797963913, 'samples': 16291328, 'steps': 31818, 'loss/train': 0.685491144657135}
+02/25/2022 18:47:20 - INFO - codeparrot_training - Step 31819: {'lr': 0.0001570719577215059, 'samples': 16291840, 'steps': 31819, 'loss/train': 1.572088360786438}
+02/25/2022 18:47:23 - INFO - codeparrot_training - Step 31820: {'lr': 0.00015705676786144702, 'samples': 16292352, 'steps': 31820, 'loss/train': 2.3007888793945312}
+02/25/2022 18:47:29 - INFO - codeparrot_training - Step 31821: {'lr': 0.00015704157839952732, 'samples': 16292864, 'steps': 31821, 'loss/train': 0.7093051671981812}
+02/25/2022 18:47:32 - INFO - codeparrot_training - Step 31822: {'lr': 0.000157026389335812, 'samples': 16293376, 'steps': 31822, 'loss/train': 2.0013861656188965}
+02/25/2022 18:47:38 - INFO - codeparrot_training - Step 31823: {'lr': 0.00015701120067036607, 'samples': 16293888, 'steps': 31823, 'loss/train': 0.8477587699890137}
+02/25/2022 18:47:41 - INFO - codeparrot_training - Step 31824: {'lr': 0.00015699601240325473, 'samples': 16294400, 'steps': 31824, 'loss/train': 1.998630166053772}
+02/25/2022 18:47:47 - INFO - codeparrot_training - Step 31825: {'lr': 0.00015698082453454283, 'samples': 16294912, 'steps': 31825, 'loss/train': 0.9047789573669434}
+02/25/2022 18:47:50 - INFO - codeparrot_training - Step 31826: {'lr': 0.00015696563706429556, 'samples': 16295424, 'steps': 31826, 'loss/train': 2.0305511951446533}
+02/25/2022 18:47:56 - INFO - codeparrot_training - Step 31827: {'lr': 0.00015695044999257797, 'samples': 16295936, 'steps': 31827, 'loss/train': 1.7714321613311768}
+02/25/2022 18:47:59 - INFO - codeparrot_training - Step 31828: {'lr': 0.0001569352633194552, 'samples': 16296448, 'steps': 31828, 'loss/train': 2.3439695835113525}
+02/25/2022 18:48:03 - INFO - codeparrot_training - Step 31829: {'lr': 0.00015692007704499209, 'samples': 16296960, 'steps': 31829, 'loss/train': 1.3583475351333618}
+02/25/2022 18:48:10 - INFO - codeparrot_training - Step 31830: {'lr': 0.00015690489116925382, 'samples': 16297472, 'steps': 31830, 'loss/train': 2.480048656463623}
+02/25/2022 18:48:13 - INFO - codeparrot_training - Step 31831: {'lr': 0.00015688970569230544, 'samples': 16297984, 'steps': 31831, 'loss/train': 0.7590587735176086}
+02/25/2022 18:48:17 - INFO - codeparrot_training - Step 31832: {'lr': 0.000156874520614212, 'samples': 16298496, 'steps': 31832, 'loss/train': 1.787411093711853}
+02/25/2022 18:48:22 - INFO - codeparrot_training - Step 31833: {'lr': 0.00015685933593503865, 'samples': 16299008, 'steps': 31833, 'loss/train': 1.5617295503616333}
+02/25/2022 18:48:28 - INFO - codeparrot_training - Step 31834: {'lr': 0.00015684415165485022, 'samples': 16299520, 'steps': 31834, 'loss/train': 1.5835037231445312}
+02/25/2022 18:48:31 - INFO - codeparrot_training - Step 31835: {'lr': 0.00015682896777371186, 'samples': 16300032, 'steps': 31835, 'loss/train': 2.327470541000366}
+02/25/2022 18:48:37 - INFO - codeparrot_training - Step 31836: {'lr': 0.00015681378429168865, 'samples': 16300544, 'steps': 31836, 'loss/train': 1.2230695486068726}
+02/25/2022 18:48:40 - INFO - codeparrot_training - Step 31837: {'lr': 0.00015679860120884566, 'samples': 16301056, 'steps': 31837, 'loss/train': 1.0115549564361572}
+02/25/2022 18:48:44 - INFO - codeparrot_training - Step 31838: {'lr': 0.00015678341852524776, 'samples': 16301568, 'steps': 31838, 'loss/train': 1.486220359802246}
+02/25/2022 18:48:49 - INFO - codeparrot_training - Step 31839: {'lr': 0.00015676823624096012, 'samples': 16302080, 'steps': 31839, 'loss/train': 2.473174571990967}
+02/25/2022 18:48:56 - INFO - codeparrot_training - Step 31840: {'lr': 0.00015675305435604775, 'samples': 16302592, 'steps': 31840, 'loss/train': 1.6202027797698975}
+02/25/2022 18:48:59 - INFO - codeparrot_training - Step 31841: {'lr': 0.0001567378728705758, 'samples': 16303104, 'steps': 31841, 'loss/train': 1.4797914028167725}
+02/25/2022 18:49:05 - INFO - codeparrot_training - Step 31842: {'lr': 0.00015672269178460904, 'samples': 16303616, 'steps': 31842, 'loss/train': 2.4769647121429443}
+02/25/2022 18:49:08 - INFO - codeparrot_training - Step 31843: {'lr': 0.00015670751109821268, 'samples': 16304128, 'steps': 31843, 'loss/train': 1.3690781593322754}
+02/25/2022 18:49:14 - INFO - codeparrot_training - Step 31844: {'lr': 0.0001566923308114518, 'samples': 16304640, 'steps': 31844, 'loss/train': 1.8152060508728027}
+02/25/2022 18:49:17 - INFO - codeparrot_training - Step 31845: {'lr': 0.0001566771509243912, 'samples': 16305152, 'steps': 31845, 'loss/train': 2.0704500675201416}
+02/25/2022 18:49:23 - INFO - codeparrot_training - Step 31846: {'lr': 0.0001566619714370962, 'samples': 16305664, 'steps': 31846, 'loss/train': 1.2604831457138062}
+02/25/2022 18:49:26 - INFO - codeparrot_training - Step 31847: {'lr': 0.00015664679234963158, 'samples': 16306176, 'steps': 31847, 'loss/train': 2.9137635231018066}
+02/25/2022 18:49:32 - INFO - codeparrot_training - Step 31848: {'lr': 0.00015663161366206253, 'samples': 16306688, 'steps': 31848, 'loss/train': 1.4659334421157837}
+02/25/2022 18:49:35 - INFO - codeparrot_training - Step 31849: {'lr': 0.00015661643537445393, 'samples': 16307200, 'steps': 31849, 'loss/train': 1.1687768697738647}
+02/25/2022 18:49:41 - INFO - codeparrot_training - Step 31850: {'lr': 0.00015660125748687094, 'samples': 16307712, 'steps': 31850, 'loss/train': 2.955305576324463}
+02/25/2022 18:49:44 - INFO - codeparrot_training - Step 31851: {'lr': 0.00015658607999937847, 'samples': 16308224, 'steps': 31851, 'loss/train': 2.2020745277404785}
+02/25/2022 18:49:51 - INFO - codeparrot_training - Step 31852: {'lr': 0.00015657090291204157, 'samples': 16308736, 'steps': 31852, 'loss/train': 1.3224120140075684}
+02/25/2022 18:49:54 - INFO - codeparrot_training - Step 31853: {'lr': 0.00015655572622492523, 'samples': 16309248, 'steps': 31853, 'loss/train': 0.5819587707519531}
+02/25/2022 18:50:00 - INFO - codeparrot_training - Step 31854: {'lr': 0.00015654054993809454, 'samples': 16309760, 'steps': 31854, 'loss/train': 1.0389347076416016}
+02/25/2022 18:50:03 - INFO - codeparrot_training - Step 31855: {'lr': 0.00015652537405161444, 'samples': 16310272, 'steps': 31855, 'loss/train': 1.2572869062423706}
+02/25/2022 18:50:09 - INFO - codeparrot_training - Step 31856: {'lr': 0.00015651019856554994, 'samples': 16310784, 'steps': 31856, 'loss/train': 1.8519079685211182}
+02/25/2022 18:50:12 - INFO - codeparrot_training - Step 31857: {'lr': 0.00015649502347996603, 'samples': 16311296, 'steps': 31857, 'loss/train': 1.3159236907958984}
+02/25/2022 18:50:18 - INFO - codeparrot_training - Step 31858: {'lr': 0.00015647984879492777, 'samples': 16311808, 'steps': 31858, 'loss/train': 1.7824357748031616}
+02/25/2022 18:50:21 - INFO - codeparrot_training - Step 31859: {'lr': 0.00015646467451050023, 'samples': 16312320, 'steps': 31859, 'loss/train': 8.676250457763672}
+02/25/2022 18:50:27 - INFO - codeparrot_training - Step 31860: {'lr': 0.00015644950062674822, 'samples': 16312832, 'steps': 31860, 'loss/train': 2.1044960021972656}
+02/25/2022 18:50:30 - INFO - codeparrot_training - Step 31861: {'lr': 0.00015643432714373685, 'samples': 16313344, 'steps': 31861, 'loss/train': 1.0503605604171753}
+02/25/2022 18:50:36 - INFO - codeparrot_training - Step 31862: {'lr': 0.00015641915406153112, 'samples': 16313856, 'steps': 31862, 'loss/train': 0.8284381628036499}
+02/25/2022 18:50:40 - INFO - codeparrot_training - Step 31863: {'lr': 0.00015640398138019613, 'samples': 16314368, 'steps': 31863, 'loss/train': 1.6933575868606567}
+02/25/2022 18:50:45 - INFO - codeparrot_training - Step 31864: {'lr': 0.00015638880909979664, 'samples': 16314880, 'steps': 31864, 'loss/train': 1.7500853538513184}
+02/25/2022 18:50:49 - INFO - codeparrot_training - Step 31865: {'lr': 0.00015637363722039777, 'samples': 16315392, 'steps': 31865, 'loss/train': 1.5358532667160034}
+02/25/2022 18:50:54 - INFO - codeparrot_training - Step 31866: {'lr': 0.00015635846574206449, 'samples': 16315904, 'steps': 31866, 'loss/train': 2.0121545791625977}
+02/25/2022 18:50:58 - INFO - codeparrot_training - Step 31867: {'lr': 0.00015634329466486182, 'samples': 16316416, 'steps': 31867, 'loss/train': 2.1975982189178467}
+02/25/2022 18:51:03 - INFO - codeparrot_training - Step 31868: {'lr': 0.00015632812398885487, 'samples': 16316928, 'steps': 31868, 'loss/train': 1.0573875904083252}
+02/25/2022 18:51:07 - INFO - codeparrot_training - Step 31869: {'lr': 0.00015631295371410837, 'samples': 16317440, 'steps': 31869, 'loss/train': 1.326983094215393}
+02/25/2022 18:51:12 - INFO - codeparrot_training - Step 31870: {'lr': 0.0001562977838406874, 'samples': 16317952, 'steps': 31870, 'loss/train': 2.1941564083099365}
+02/25/2022 18:51:16 - INFO - codeparrot_training - Step 31871: {'lr': 0.000156282614368657, 'samples': 16318464, 'steps': 31871, 'loss/train': 1.5488618612289429}
+02/25/2022 18:51:21 - INFO - codeparrot_training - Step 31872: {'lr': 0.00015626744529808223, 'samples': 16318976, 'steps': 31872, 'loss/train': 1.1792395114898682}
+02/25/2022 18:51:25 - INFO - codeparrot_training - Step 31873: {'lr': 0.00015625227662902782, 'samples': 16319488, 'steps': 31873, 'loss/train': 1.9254331588745117}
+02/25/2022 18:51:30 - INFO - codeparrot_training - Step 31874: {'lr': 0.00015623710836155892, 'samples': 16320000, 'steps': 31874, 'loss/train': 1.8640425205230713}
+02/25/2022 18:51:34 - INFO - codeparrot_training - Step 31875: {'lr': 0.00015622194049574048, 'samples': 16320512, 'steps': 31875, 'loss/train': 0.4795461595058441}
+02/25/2022 18:51:40 - INFO - codeparrot_training - Step 31876: {'lr': 0.00015620677303163755, 'samples': 16321024, 'steps': 31876, 'loss/train': 1.900381326675415}
+02/25/2022 18:51:44 - INFO - codeparrot_training - Step 31877: {'lr': 0.00015619160596931492, 'samples': 16321536, 'steps': 31877, 'loss/train': 0.43327316641807556}
+02/25/2022 18:51:49 - INFO - codeparrot_training - Step 31878: {'lr': 0.00015617643930883768, 'samples': 16322048, 'steps': 31878, 'loss/train': 1.862421989440918}
+02/25/2022 18:51:53 - INFO - codeparrot_training - Step 31879: {'lr': 0.00015616127305027077, 'samples': 16322560, 'steps': 31879, 'loss/train': 1.6216498613357544}
+02/25/2022 18:51:58 - INFO - codeparrot_training - Step 31880: {'lr': 0.0001561461071936792, 'samples': 16323072, 'steps': 31880, 'loss/train': 2.3397865295410156}
+02/25/2022 18:52:02 - INFO - codeparrot_training - Step 31881: {'lr': 0.000156130941739128, 'samples': 16323584, 'steps': 31881, 'loss/train': 2.486037254333496}
+02/25/2022 18:52:07 - INFO - codeparrot_training - Step 31882: {'lr': 0.0001561157766866819, 'samples': 16324096, 'steps': 31882, 'loss/train': 0.6346574425697327}
+02/25/2022 18:52:11 - INFO - codeparrot_training - Step 31883: {'lr': 0.000156100612036406, 'samples': 16324608, 'steps': 31883, 'loss/train': 0.8969175219535828}
+02/25/2022 18:52:16 - INFO - codeparrot_training - Step 31884: {'lr': 0.00015608544778836534, 'samples': 16325120, 'steps': 31884, 'loss/train': 1.0674989223480225}
+02/25/2022 18:52:20 - INFO - codeparrot_training - Step 31885: {'lr': 0.0001560702839426248, 'samples': 16325632, 'steps': 31885, 'loss/train': 1.240315556526184}
+02/25/2022 18:52:26 - INFO - codeparrot_training - Step 31886: {'lr': 0.00015605512049924927, 'samples': 16326144, 'steps': 31886, 'loss/train': 0.6611651182174683}
+02/25/2022 18:52:30 - INFO - codeparrot_training - Step 31887: {'lr': 0.0001560399574583038, 'samples': 16326656, 'steps': 31887, 'loss/train': 1.142682671546936}
+02/25/2022 18:52:35 - INFO - codeparrot_training - Step 31888: {'lr': 0.00015602479481985332, 'samples': 16327168, 'steps': 31888, 'loss/train': 1.334628701210022}
+02/25/2022 18:52:39 - INFO - codeparrot_training - Step 31889: {'lr': 0.00015600963258396284, 'samples': 16327680, 'steps': 31889, 'loss/train': 1.9676766395568848}
+02/25/2022 18:52:44 - INFO - codeparrot_training - Step 31890: {'lr': 0.00015599447075069714, 'samples': 16328192, 'steps': 31890, 'loss/train': 1.572779893875122}
+02/25/2022 18:52:50 - INFO - codeparrot_training - Step 31891: {'lr': 0.0001559793093201213, 'samples': 16328704, 'steps': 31891, 'loss/train': 2.2284061908721924}
+02/25/2022 18:52:53 - INFO - codeparrot_training - Step 31892: {'lr': 0.0001559641482923003, 'samples': 16329216, 'steps': 31892, 'loss/train': 1.5746623277664185}
+02/25/2022 18:52:59 - INFO - codeparrot_training - Step 31893: {'lr': 0.00015594898766729898, 'samples': 16329728, 'steps': 31893, 'loss/train': 1.9416314363479614}
+02/25/2022 18:53:02 - INFO - codeparrot_training - Step 31894: {'lr': 0.00015593382744518236, 'samples': 16330240, 'steps': 31894, 'loss/train': 1.0915865898132324}
+02/25/2022 18:53:08 - INFO - codeparrot_training - Step 31895: {'lr': 0.00015591866762601538, 'samples': 16330752, 'steps': 31895, 'loss/train': 2.1597647666931152}
+02/25/2022 18:53:11 - INFO - codeparrot_training - Step 31896: {'lr': 0.0001559035082098629, 'samples': 16331264, 'steps': 31896, 'loss/train': 1.8096113204956055}
+02/25/2022 18:53:18 - INFO - codeparrot_training - Step 31897: {'lr': 0.00015588834919678991, 'samples': 16331776, 'steps': 31897, 'loss/train': 0.5843462347984314}
+02/25/2022 18:53:21 - INFO - codeparrot_training - Step 31898: {'lr': 0.0001558731905868614, 'samples': 16332288, 'steps': 31898, 'loss/train': 1.7148973941802979}
+02/25/2022 18:53:26 - INFO - codeparrot_training - Step 31899: {'lr': 0.00015585803238014223, 'samples': 16332800, 'steps': 31899, 'loss/train': 1.8135031461715698}
+02/25/2022 18:53:30 - INFO - codeparrot_training - Step 31900: {'lr': 0.00015584287457669733, 'samples': 16333312, 'steps': 31900, 'loss/train': 1.3195728063583374}
+02/25/2022 18:53:35 - INFO - codeparrot_training - Step 31901: {'lr': 0.00015582771717659167, 'samples': 16333824, 'steps': 31901, 'loss/train': 1.6785763502120972}
+02/25/2022 18:53:39 - INFO - codeparrot_training - Step 31902: {'lr': 0.00015581256017989023, 'samples': 16334336, 'steps': 31902, 'loss/train': 0.8784569501876831}
+02/25/2022 18:53:44 - INFO - codeparrot_training - Step 31903: {'lr': 0.00015579740358665777, 'samples': 16334848, 'steps': 31903, 'loss/train': 1.6951404809951782}
+02/25/2022 18:53:48 - INFO - codeparrot_training - Step 31904: {'lr': 0.00015578224739695937, 'samples': 16335360, 'steps': 31904, 'loss/train': 1.5694926977157593}
+02/25/2022 18:53:54 - INFO - codeparrot_training - Step 31905: {'lr': 0.0001557670916108599, 'samples': 16335872, 'steps': 31905, 'loss/train': 2.8908092975616455}
+02/25/2022 18:53:57 - INFO - codeparrot_training - Step 31906: {'lr': 0.00015575193622842425, 'samples': 16336384, 'steps': 31906, 'loss/train': 0.05563532933592796}
+02/25/2022 18:54:03 - INFO - codeparrot_training - Step 31907: {'lr': 0.00015573678124971747, 'samples': 16336896, 'steps': 31907, 'loss/train': 1.988373041152954}
+02/25/2022 18:54:07 - INFO - codeparrot_training - Step 31908: {'lr': 0.00015572162667480433, 'samples': 16337408, 'steps': 31908, 'loss/train': 2.0349233150482178}
+02/25/2022 18:54:12 - INFO - codeparrot_training - Step 31909: {'lr': 0.00015570647250374978, 'samples': 16337920, 'steps': 31909, 'loss/train': 0.9201633334159851}
+02/25/2022 18:54:16 - INFO - codeparrot_training - Step 31910: {'lr': 0.00015569131873661873, 'samples': 16338432, 'steps': 31910, 'loss/train': 2.3175086975097656}
+02/25/2022 18:54:21 - INFO - codeparrot_training - Step 31911: {'lr': 0.0001556761653734763, 'samples': 16338944, 'steps': 31911, 'loss/train': 1.4183411598205566}
+02/25/2022 18:54:25 - INFO - codeparrot_training - Step 31912: {'lr': 0.00015566101241438706, 'samples': 16339456, 'steps': 31912, 'loss/train': 1.8293401002883911}
+02/25/2022 18:54:30 - INFO - codeparrot_training - Step 31913: {'lr': 0.0001556458598594161, 'samples': 16339968, 'steps': 31913, 'loss/train': 1.4242357015609741}
+02/25/2022 18:54:34 - INFO - codeparrot_training - Step 31914: {'lr': 0.00015563070770862835, 'samples': 16340480, 'steps': 31914, 'loss/train': 2.116787910461426}
+02/25/2022 18:54:39 - INFO - codeparrot_training - Step 31915: {'lr': 0.00015561555596208876, 'samples': 16340992, 'steps': 31915, 'loss/train': 2.064756155014038}
+02/25/2022 18:54:42 - INFO - codeparrot_training - Step 31916: {'lr': 0.00015560040461986204, 'samples': 16341504, 'steps': 31916, 'loss/train': 1.232796549797058}
+02/25/2022 18:54:48 - INFO - codeparrot_training - Step 31917: {'lr': 0.00015558525368201327, 'samples': 16342016, 'steps': 31917, 'loss/train': 2.007220983505249}
+02/25/2022 18:54:51 - INFO - codeparrot_training - Step 31918: {'lr': 0.00015557010314860723, 'samples': 16342528, 'steps': 31918, 'loss/train': 1.057302713394165}
+02/25/2022 18:54:57 - INFO - codeparrot_training - Step 31919: {'lr': 0.0001555549530197089, 'samples': 16343040, 'steps': 31919, 'loss/train': 1.8840572834014893}
+02/25/2022 18:55:00 - INFO - codeparrot_training - Step 31920: {'lr': 0.00015553980329538326, 'samples': 16343552, 'steps': 31920, 'loss/train': 2.9769749641418457}
+02/25/2022 18:55:06 - INFO - codeparrot_training - Step 31921: {'lr': 0.00015552465397569503, 'samples': 16344064, 'steps': 31921, 'loss/train': 1.6371641159057617}
+02/25/2022 18:55:09 - INFO - codeparrot_training - Step 31922: {'lr': 0.00015550950506070915, 'samples': 16344576, 'steps': 31922, 'loss/train': 1.532838225364685}
+02/25/2022 18:55:16 - INFO - codeparrot_training - Step 31923: {'lr': 0.00015549435655049057, 'samples': 16345088, 'steps': 31923, 'loss/train': 0.9875187277793884}
+02/25/2022 18:55:19 - INFO - codeparrot_training - Step 31924: {'lr': 0.00015547920844510428, 'samples': 16345600, 'steps': 31924, 'loss/train': 1.5907374620437622}
+02/25/2022 18:55:25 - INFO - codeparrot_training - Step 31925: {'lr': 0.0001554640607446149, 'samples': 16346112, 'steps': 31925, 'loss/train': 2.4162392616271973}
+02/25/2022 18:55:28 - INFO - codeparrot_training - Step 31926: {'lr': 0.00015544891344908748, 'samples': 16346624, 'steps': 31926, 'loss/train': 2.298130989074707}
+02/25/2022 18:55:34 - INFO - codeparrot_training - Step 31927: {'lr': 0.00015543376655858688, 'samples': 16347136, 'steps': 31927, 'loss/train': 1.5891966819763184}
+02/25/2022 18:55:37 - INFO - codeparrot_training - Step 31928: {'lr': 0.00015541862007317807, 'samples': 16347648, 'steps': 31928, 'loss/train': 2.1017534732818604}
+02/25/2022 18:55:42 - INFO - codeparrot_training - Step 31929: {'lr': 0.00015540347399292586, 'samples': 16348160, 'steps': 31929, 'loss/train': 1.8119382858276367}
+02/25/2022 18:55:46 - INFO - codeparrot_training - Step 31930: {'lr': 0.00015538832831789507, 'samples': 16348672, 'steps': 31930, 'loss/train': 1.8796244859695435}
+02/25/2022 18:55:51 - INFO - codeparrot_training - Step 31931: {'lr': 0.00015537318304815063, 'samples': 16349184, 'steps': 31931, 'loss/train': 0.5294276475906372}
+02/25/2022 18:55:55 - INFO - codeparrot_training - Step 31932: {'lr': 0.00015535803818375748, 'samples': 16349696, 'steps': 31932, 'loss/train': 2.3740463256835938}
+02/25/2022 18:56:01 - INFO - codeparrot_training - Step 31933: {'lr': 0.00015534289372478045, 'samples': 16350208, 'steps': 31933, 'loss/train': 0.30523836612701416}
+02/25/2022 18:56:05 - INFO - codeparrot_training - Step 31934: {'lr': 0.00015532774967128439, 'samples': 16350720, 'steps': 31934, 'loss/train': 1.152420163154602}
+02/25/2022 18:56:10 - INFO - codeparrot_training - Step 31935: {'lr': 0.0001553126060233342, 'samples': 16351232, 'steps': 31935, 'loss/train': 2.621337413787842}
+02/25/2022 18:56:14 - INFO - codeparrot_training - Step 31936: {'lr': 0.00015529746278099467, 'samples': 16351744, 'steps': 31936, 'loss/train': 1.2015964984893799}
+02/25/2022 18:56:19 - INFO - codeparrot_training - Step 31937: {'lr': 0.00015528231994433082, 'samples': 16352256, 'steps': 31937, 'loss/train': 0.0820598155260086}
+02/25/2022 18:56:23 - INFO - codeparrot_training - Step 31938: {'lr': 0.00015526717751340735, 'samples': 16352768, 'steps': 31938, 'loss/train': 2.08146595954895}
+02/25/2022 18:56:28 - INFO - codeparrot_training - Step 31939: {'lr': 0.00015525203548828932, 'samples': 16353280, 'steps': 31939, 'loss/train': 2.4808058738708496}
+02/25/2022 18:56:32 - INFO - codeparrot_training - Step 31940: {'lr': 0.0001552368938690414, 'samples': 16353792, 'steps': 31940, 'loss/train': 1.5145026445388794}
+02/25/2022 18:56:37 - INFO - codeparrot_training - Step 31941: {'lr': 0.00015522175265572852, 'samples': 16354304, 'steps': 31941, 'loss/train': 0.05511852353811264}
+02/25/2022 18:56:41 - INFO - codeparrot_training - Step 31942: {'lr': 0.00015520661184841567, 'samples': 16354816, 'steps': 31942, 'loss/train': 1.681165337562561}
+02/25/2022 18:56:48 - INFO - codeparrot_training - Step 31943: {'lr': 0.0001551914714471675, 'samples': 16355328, 'steps': 31943, 'loss/train': 2.3063130378723145}
+02/25/2022 18:56:51 - INFO - codeparrot_training - Step 31944: {'lr': 0.00015517633145204895, 'samples': 16355840, 'steps': 31944, 'loss/train': 1.2548444271087646}
+02/25/2022 18:56:57 - INFO - codeparrot_training - Step 31945: {'lr': 0.00015516119186312488, 'samples': 16356352, 'steps': 31945, 'loss/train': 1.2488266229629517}
+02/25/2022 18:57:00 - INFO - codeparrot_training - Step 31946: {'lr': 0.00015514605268046028, 'samples': 16356864, 'steps': 31946, 'loss/train': 1.8956669569015503}
+02/25/2022 18:57:06 - INFO - codeparrot_training - Step 31947: {'lr': 0.00015513091390411977, 'samples': 16357376, 'steps': 31947, 'loss/train': 1.4030108451843262}
+02/25/2022 18:57:09 - INFO - codeparrot_training - Step 31948: {'lr': 0.00015511577553416826, 'samples': 16357888, 'steps': 31948, 'loss/train': 2.4183309078216553}
+02/25/2022 18:57:15 - INFO - codeparrot_training - Step 31949: {'lr': 0.00015510063757067066, 'samples': 16358400, 'steps': 31949, 'loss/train': 2.086169719696045}
+02/25/2022 18:57:20 - INFO - codeparrot_training - Step 31950: {'lr': 0.0001550855000136919, 'samples': 16358912, 'steps': 31950, 'loss/train': 1.7187268733978271}
+02/25/2022 18:57:23 - INFO - codeparrot_training - Step 31951: {'lr': 0.0001550703628632966, 'samples': 16359424, 'steps': 31951, 'loss/train': 1.9003520011901855}
+02/25/2022 18:57:30 - INFO - codeparrot_training - Step 31952: {'lr': 0.00015505522611954976, 'samples': 16359936, 'steps': 31952, 'loss/train': 1.767127275466919}
+02/25/2022 18:57:34 - INFO - codeparrot_training - Step 31953: {'lr': 0.00015504008978251614, 'samples': 16360448, 'steps': 31953, 'loss/train': 0.5519925355911255}
+02/25/2022 18:57:39 - INFO - codeparrot_training - Step 31954: {'lr': 0.00015502495385226066, 'samples': 16360960, 'steps': 31954, 'loss/train': 2.0983614921569824}
+02/25/2022 18:57:43 - INFO - codeparrot_training - Step 31955: {'lr': 0.00015500981832884819, 'samples': 16361472, 'steps': 31955, 'loss/train': 1.8321764469146729}
+02/25/2022 18:57:48 - INFO - codeparrot_training - Step 31956: {'lr': 0.00015499468321234343, 'samples': 16361984, 'steps': 31956, 'loss/train': 1.2970038652420044}
+02/25/2022 18:57:52 - INFO - codeparrot_training - Step 31957: {'lr': 0.00015497954850281125, 'samples': 16362496, 'steps': 31957, 'loss/train': 2.4311492443084717}
+02/25/2022 18:57:57 - INFO - codeparrot_training - Step 31958: {'lr': 0.00015496441420031652, 'samples': 16363008, 'steps': 31958, 'loss/train': 1.999210238456726}
+02/25/2022 18:58:00 - INFO - codeparrot_training - Step 31959: {'lr': 0.00015494928030492417, 'samples': 16363520, 'steps': 31959, 'loss/train': 0.7683048248291016}
+02/25/2022 18:58:06 - INFO - codeparrot_training - Step 31960: {'lr': 0.00015493414681669882, 'samples': 16364032, 'steps': 31960, 'loss/train': 1.3272879123687744}
+02/25/2022 18:58:10 - INFO - codeparrot_training - Step 31961: {'lr': 0.0001549190137357054, 'samples': 16364544, 'steps': 31961, 'loss/train': 1.0985316038131714}
+02/25/2022 18:58:15 - INFO - codeparrot_training - Step 31962: {'lr': 0.0001549038810620087, 'samples': 16365056, 'steps': 31962, 'loss/train': 1.879543662071228}
+02/25/2022 18:58:19 - INFO - codeparrot_training - Step 31963: {'lr': 0.0001548887487956737, 'samples': 16365568, 'steps': 31963, 'loss/train': 1.8775640726089478}
+02/25/2022 18:58:24 - INFO - codeparrot_training - Step 31964: {'lr': 0.000154873616936765, 'samples': 16366080, 'steps': 31964, 'loss/train': 2.780747175216675}
+02/25/2022 18:58:27 - INFO - codeparrot_training - Step 31965: {'lr': 0.00015485848548534752, 'samples': 16366592, 'steps': 31965, 'loss/train': 1.7711981534957886}
+02/25/2022 18:58:33 - INFO - codeparrot_training - Step 31966: {'lr': 0.00015484335444148606, 'samples': 16367104, 'steps': 31966, 'loss/train': 2.0191216468811035}
+02/25/2022 18:58:36 - INFO - codeparrot_training - Step 31967: {'lr': 0.00015482822380524547, 'samples': 16367616, 'steps': 31967, 'loss/train': 1.8072110414505005}
+02/25/2022 18:58:43 - INFO - codeparrot_training - Step 31968: {'lr': 0.00015481309357669065, 'samples': 16368128, 'steps': 31968, 'loss/train': 1.5220261812210083}
+02/25/2022 18:58:46 - INFO - codeparrot_training - Step 31969: {'lr': 0.00015479796375588624, 'samples': 16368640, 'steps': 31969, 'loss/train': 2.519467353820801}
+02/25/2022 18:58:52 - INFO - codeparrot_training - Step 31970: {'lr': 0.00015478283434289708, 'samples': 16369152, 'steps': 31970, 'loss/train': 1.261850118637085}
+02/25/2022 18:58:55 - INFO - codeparrot_training - Step 31971: {'lr': 0.00015476770533778806, 'samples': 16369664, 'steps': 31971, 'loss/train': 1.8854120969772339}
+02/25/2022 18:59:00 - INFO - codeparrot_training - Step 31972: {'lr': 0.000154752576740624, 'samples': 16370176, 'steps': 31972, 'loss/train': 1.5968600511550903}
+02/25/2022 18:59:04 - INFO - codeparrot_training - Step 31973: {'lr': 0.0001547374485514696, 'samples': 16370688, 'steps': 31973, 'loss/train': 0.3152819871902466}
+02/25/2022 18:59:10 - INFO - codeparrot_training - Step 31974: {'lr': 0.00015472232077038973, 'samples': 16371200, 'steps': 31974, 'loss/train': 1.1850675344467163}
+02/25/2022 18:59:13 - INFO - codeparrot_training - Step 31975: {'lr': 0.00015470719339744917, 'samples': 16371712, 'steps': 31975, 'loss/train': 1.9640557765960693}
+02/25/2022 18:59:18 - INFO - codeparrot_training - Step 31976: {'lr': 0.00015469206643271274, 'samples': 16372224, 'steps': 31976, 'loss/train': 2.8242547512054443}
+02/25/2022 18:59:22 - INFO - codeparrot_training - Step 31977: {'lr': 0.00015467693987624527, 'samples': 16372736, 'steps': 31977, 'loss/train': 1.5759389400482178}
+02/25/2022 18:59:28 - INFO - codeparrot_training - Step 31978: {'lr': 0.00015466181372811148, 'samples': 16373248, 'steps': 31978, 'loss/train': 1.1964836120605469}
+02/25/2022 18:59:32 - INFO - codeparrot_training - Step 31979: {'lr': 0.0001546466879883762, 'samples': 16373760, 'steps': 31979, 'loss/train': 1.551389217376709}
+02/25/2022 18:59:37 - INFO - codeparrot_training - Step 31980: {'lr': 0.00015463156265710428, 'samples': 16374272, 'steps': 31980, 'loss/train': 1.5725407600402832}
+02/25/2022 18:59:41 - INFO - codeparrot_training - Step 31981: {'lr': 0.00015461643773436045, 'samples': 16374784, 'steps': 31981, 'loss/train': 1.3243732452392578}
+02/25/2022 18:59:46 - INFO - codeparrot_training - Step 31982: {'lr': 0.00015460131322020954, 'samples': 16375296, 'steps': 31982, 'loss/train': 0.8847343325614929}
+02/25/2022 18:59:50 - INFO - codeparrot_training - Step 31983: {'lr': 0.00015458618911471622, 'samples': 16375808, 'steps': 31983, 'loss/train': 1.9870924949645996}
+02/25/2022 18:59:55 - INFO - codeparrot_training - Step 31984: {'lr': 0.00015457106541794541, 'samples': 16376320, 'steps': 31984, 'loss/train': 1.9988124370574951}
+02/25/2022 18:59:59 - INFO - codeparrot_training - Step 31985: {'lr': 0.00015455594212996187, 'samples': 16376832, 'steps': 31985, 'loss/train': 3.596651077270508}
+02/25/2022 19:00:04 - INFO - codeparrot_training - Step 31986: {'lr': 0.0001545408192508304, 'samples': 16377344, 'steps': 31986, 'loss/train': 1.3388316631317139}
+02/25/2022 19:00:08 - INFO - codeparrot_training - Step 31987: {'lr': 0.00015452569678061568, 'samples': 16377856, 'steps': 31987, 'loss/train': 2.265784978866577}
+02/25/2022 19:00:14 - INFO - codeparrot_training - Step 31988: {'lr': 0.00015451057471938258, 'samples': 16378368, 'steps': 31988, 'loss/train': 1.4022538661956787}
+02/25/2022 19:00:18 - INFO - codeparrot_training - Step 31989: {'lr': 0.0001544954530671958, 'samples': 16378880, 'steps': 31989, 'loss/train': 0.9224123954772949}
+02/25/2022 19:00:23 - INFO - codeparrot_training - Step 31990: {'lr': 0.0001544803318241203, 'samples': 16379392, 'steps': 31990, 'loss/train': 1.9672799110412598}
+02/25/2022 19:00:27 - INFO - codeparrot_training - Step 31991: {'lr': 0.00015446521099022066, 'samples': 16379904, 'steps': 31991, 'loss/train': 2.0309693813323975}
+02/25/2022 19:00:32 - INFO - codeparrot_training - Step 31992: {'lr': 0.00015445009056556168, 'samples': 16380416, 'steps': 31992, 'loss/train': 2.064556360244751}
+02/25/2022 19:00:36 - INFO - codeparrot_training - Step 31993: {'lr': 0.00015443497055020821, 'samples': 16380928, 'steps': 31993, 'loss/train': 1.3361923694610596}
+02/25/2022 19:00:41 - INFO - codeparrot_training - Step 31994: {'lr': 0.00015441985094422506, 'samples': 16381440, 'steps': 31994, 'loss/train': 2.0053908824920654}
+02/25/2022 19:00:45 - INFO - codeparrot_training - Step 31995: {'lr': 0.00015440473174767682, 'samples': 16381952, 'steps': 31995, 'loss/train': 0.2217193841934204}
+02/25/2022 19:00:50 - INFO - codeparrot_training - Step 31996: {'lr': 0.00015438961296062833, 'samples': 16382464, 'steps': 31996, 'loss/train': 1.7954959869384766}
+02/25/2022 19:00:54 - INFO - codeparrot_training - Step 31997: {'lr': 0.0001543744945831444, 'samples': 16382976, 'steps': 31997, 'loss/train': 1.6838881969451904}
+02/25/2022 19:00:59 - INFO - codeparrot_training - Step 31998: {'lr': 0.0001543593766152899, 'samples': 16383488, 'steps': 31998, 'loss/train': 1.7603875398635864}
+02/25/2022 19:01:03 - INFO - codeparrot_training - Step 31999: {'lr': 0.00015434425905712934, 'samples': 16384000, 'steps': 31999, 'loss/train': 1.2924124002456665}
+02/25/2022 19:01:03 - INFO - codeparrot_training - Evaluating and saving model checkpoint