diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -37524,3 +37524,1009 @@ Use FP16 precision: False
 02/26/2022 01:28:07 - INFO - codeparrot_training - Step 36998: {'lr': 8.518815153729173e-05, 'samples': 18943488, 'steps': 36998, 'loss/train': 2.316406011581421}
 02/26/2022 01:28:11 - INFO - codeparrot_training - Step 36999: {'lr': 8.517584852811005e-05, 'samples': 18944000, 'steps': 36999, 'loss/train': 1.4211903810501099}
 02/26/2022 01:28:11 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 01:28:29 - WARNING - huggingface_hub.repository - Several commits (37) will be pushed upstream.
+02/26/2022 01:28:29 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 01:29:05 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   f71b356..333cdce  floral-grass-11 -> floral-grass-11
+
+02/26/2022 01:29:13 - INFO - codeparrot_training - Step 37000: {'lr': 8.516354622498279e-05, 'samples': 18944512, 'steps': 37000, 'loss/train': 1.0448174476623535}
+02/26/2022 01:29:16 - INFO - codeparrot_training - Step 37001: {'lr': 8.515124462796254e-05, 'samples': 18945024, 'steps': 37001, 'loss/train': 2.14421010017395}
+02/26/2022 01:29:22 - INFO - codeparrot_training - Step 37002: {'lr': 8.513894373710201e-05, 'samples': 18945536, 'steps': 37002, 'loss/train': 1.1430444717407227}
+02/26/2022 01:29:25 - INFO - codeparrot_training - Step 37003: {'lr': 8.512664355245384e-05, 'samples': 18946048, 'steps': 37003, 'loss/train': 0.11870545148849487}
+02/26/2022 01:29:31 - INFO - codeparrot_training - Step 37004: {'lr': 8.511434407407082e-05, 'samples': 18946560, 'steps': 37004, 'loss/train': 0.5594982504844666}
+02/26/2022 01:29:35 - INFO - codeparrot_training - Step 37005: {'lr': 8.510204530200561e-05, 'samples': 18947072, 'steps': 37005, 'loss/train': 1.4826196432113647}
+02/26/2022 01:29:40 - INFO - codeparrot_training - Step 37006: {'lr': 8.508974723631088e-05, 'samples': 18947584, 'steps': 37006, 'loss/train': 2.9788272380828857}
+02/26/2022 01:29:44 - INFO - codeparrot_training - Step 37007: {'lr': 8.507744987703923e-05, 'samples': 18948096, 'steps': 37007, 'loss/train': 0.9081933498382568}
+02/26/2022 01:29:49 - INFO - codeparrot_training - Step 37008: {'lr': 8.506515322424349e-05, 'samples': 18948608, 'steps': 37008, 'loss/train': 2.1243278980255127}
+02/26/2022 01:29:53 - INFO - codeparrot_training - Step 37009: {'lr': 8.505285727797627e-05, 'samples': 18949120, 'steps': 37009, 'loss/train': 2.4234466552734375}
+02/26/2022 01:29:58 - INFO - codeparrot_training - Step 37010: {'lr': 8.504056203829022e-05, 'samples': 18949632, 'steps': 37010, 'loss/train': 0.6121158599853516}
+02/26/2022 01:30:02 - INFO - codeparrot_training - Step 37011: {'lr': 8.502826750523793e-05, 'samples': 18950144, 'steps': 37011, 'loss/train': 1.7025648355484009}
+02/26/2022 01:30:07 - INFO - codeparrot_training - Step 37012: {'lr': 8.501597367887226e-05, 'samples': 18950656, 'steps': 37012, 'loss/train': 1.742186188697815}
+02/26/2022 01:30:11 - INFO - codeparrot_training - Step 37013: {'lr': 8.500368055924576e-05, 'samples': 18951168, 'steps': 37013, 'loss/train': 2.372593879699707}
+02/26/2022 01:30:17 - INFO - codeparrot_training - Step 37014: {'lr': 8.49913881464111e-05, 'samples': 18951680, 'steps': 37014, 'loss/train': 1.6525758504867554}
+02/26/2022 01:30:22 - INFO - codeparrot_training - Step 37015: {'lr': 8.497909644042088e-05, 'samples': 18952192, 'steps': 37015, 'loss/train': 0.9016113877296448}
+02/26/2022 01:30:26 - INFO - codeparrot_training - Step 37016: {'lr': 8.496680544132788e-05, 'samples': 18952704, 'steps': 37016, 'loss/train': 1.5941678285598755}
+02/26/2022 01:30:31 - INFO - codeparrot_training - Step 37017: {'lr': 8.495451514918465e-05, 'samples': 18953216, 'steps': 37017, 'loss/train': 0.6739307045936584}
+02/26/2022 01:30:35 - INFO - codeparrot_training - Step 37018: {'lr': 8.494222556404391e-05, 'samples': 18953728, 'steps': 37018, 'loss/train': 1.3178893327713013}
+02/26/2022 01:30:41 - INFO - codeparrot_training - Step 37019: {'lr': 8.492993668595817e-05, 'samples': 18954240, 'steps': 37019, 'loss/train': 1.880487322807312}
+02/26/2022 01:30:44 - INFO - codeparrot_training - Step 37020: {'lr': 8.491764851498024e-05, 'samples': 18954752, 'steps': 37020, 'loss/train': 1.421849012374878}
+02/26/2022 01:30:49 - INFO - codeparrot_training - Step 37021: {'lr': 8.490536105116259e-05, 'samples': 18955264, 'steps': 37021, 'loss/train': 1.0910184383392334}
+02/26/2022 01:30:53 - INFO - codeparrot_training - Step 37022: {'lr': 8.489307429455814e-05, 'samples': 18955776, 'steps': 37022, 'loss/train': 1.6827292442321777}
+02/26/2022 01:30:59 - INFO - codeparrot_training - Step 37023: {'lr': 8.488078824521914e-05, 'samples': 18956288, 'steps': 37023, 'loss/train': 1.7838222980499268}
+02/26/2022 01:31:03 - INFO - codeparrot_training - Step 37024: {'lr': 8.486850290319852e-05, 'samples': 18956800, 'steps': 37024, 'loss/train': 2.0251317024230957}
+02/26/2022 01:31:08 - INFO - codeparrot_training - Step 37025: {'lr': 8.485621826854869e-05, 'samples': 18957312, 'steps': 37025, 'loss/train': 3.424682855606079}
+02/26/2022 01:31:12 - INFO - codeparrot_training - Step 37026: {'lr': 8.484393434132259e-05, 'samples': 18957824, 'steps': 37026, 'loss/train': 1.7839360237121582}
+02/26/2022 01:31:17 - INFO - codeparrot_training - Step 37027: {'lr': 8.483165112157243e-05, 'samples': 18958336, 'steps': 37027, 'loss/train': 2.322451591491699}
+02/26/2022 01:31:21 - INFO - codeparrot_training - Step 37028: {'lr': 8.481936860935113e-05, 'samples': 18958848, 'steps': 37028, 'loss/train': 1.2318921089172363}
+02/26/2022 01:31:27 - INFO - codeparrot_training - Step 37029: {'lr': 8.480708680471111e-05, 'samples': 18959360, 'steps': 37029, 'loss/train': 1.5701014995574951}
+02/26/2022 01:31:30 - INFO - codeparrot_training - Step 37030: {'lr': 8.479480570770525e-05, 'samples': 18959872, 'steps': 37030, 'loss/train': 1.7174086570739746}
+02/26/2022 01:31:34 - INFO - codeparrot_training - Step 37031: {'lr': 8.478252531838581e-05, 'samples': 18960384, 'steps': 37031, 'loss/train': 0.49307626485824585}
+02/26/2022 01:31:39 - INFO - codeparrot_training - Step 37032: {'lr': 8.477024563680566e-05, 'samples': 18960896, 'steps': 37032, 'loss/train': 0.8573851585388184}
+02/26/2022 01:31:43 - INFO - codeparrot_training - Step 37033: {'lr': 8.475796666301721e-05, 'samples': 18961408, 'steps': 37033, 'loss/train': 1.7240451574325562}
+02/26/2022 01:31:48 - INFO - codeparrot_training - Step 37034: {'lr': 8.474568839707328e-05, 'samples': 18961920, 'steps': 37034, 'loss/train': 2.503028392791748}
+02/26/2022 01:31:52 - INFO - codeparrot_training - Step 37035: {'lr': 8.473341083902634e-05, 'samples': 18962432, 'steps': 37035, 'loss/train': 1.8472881317138672}
+02/26/2022 01:31:57 - INFO - codeparrot_training - Step 37036: {'lr': 8.472113398892894e-05, 'samples': 18962944, 'steps': 37036, 'loss/train': 2.088653326034546}
+02/26/2022 01:32:01 - INFO - codeparrot_training - Step 37037: {'lr': 8.470885784683369e-05, 'samples': 18963456, 'steps': 37037, 'loss/train': 0.43235892057418823}
+02/26/2022 01:32:06 - INFO - codeparrot_training - Step 37038: {'lr': 8.469658241279326e-05, 'samples': 18963968, 'steps': 37038, 'loss/train': 1.8186694383621216}
+02/26/2022 01:32:10 - INFO - codeparrot_training - Step 37039: {'lr': 8.46843076868602e-05, 'samples': 18964480, 'steps': 37039, 'loss/train': 1.039555549621582}
+02/26/2022 01:32:15 - INFO - codeparrot_training - Step 37040: {'lr': 8.467203366908707e-05, 'samples': 18964992, 'steps': 37040, 'loss/train': 1.0407304763793945}
+02/26/2022 01:32:19 - INFO - codeparrot_training - Step 37041: {'lr': 8.465976035952635e-05, 'samples': 18965504, 'steps': 37041, 'loss/train': 2.1811790466308594}
+02/26/2022 01:32:25 - INFO - codeparrot_training - Step 37042: {'lr': 8.464748775823081e-05, 'samples': 18966016, 'steps': 37042, 'loss/train': 2.341484546661377}
+02/26/2022 01:32:28 - INFO - codeparrot_training - Step 37043: {'lr': 8.463521586525294e-05, 'samples': 18966528, 'steps': 37043, 'loss/train': 1.6124290227890015}
+02/26/2022 01:32:34 - INFO - codeparrot_training - Step 37044: {'lr': 8.462294468064526e-05, 'samples': 18967040, 'steps': 37044, 'loss/train': 0.9585081934928894}
+02/26/2022 01:32:37 - INFO - codeparrot_training - Step 37045: {'lr': 8.461067420446041e-05, 'samples': 18967552, 'steps': 37045, 'loss/train': 1.5336618423461914}
+02/26/2022 01:32:43 - INFO - codeparrot_training - Step 37046: {'lr': 8.459840443675082e-05, 'samples': 18968064, 'steps': 37046, 'loss/train': 2.356720447540283}
+02/26/2022 01:32:46 - INFO - codeparrot_training - Step 37047: {'lr': 8.458613537756923e-05, 'samples': 18968576, 'steps': 37047, 'loss/train': 2.3372151851654053}
+02/26/2022 01:32:52 - INFO - codeparrot_training - Step 37048: {'lr': 8.457386702696809e-05, 'samples': 18969088, 'steps': 37048, 'loss/train': 0.9617457985877991}
+02/26/2022 01:32:57 - INFO - codeparrot_training - Step 37049: {'lr': 8.4561599385e-05, 'samples': 18969600, 'steps': 37049, 'loss/train': 1.6585125923156738}
+02/26/2022 01:33:01 - INFO - codeparrot_training - Step 37050: {'lr': 8.45493324517174e-05, 'samples': 18970112, 'steps': 37050, 'loss/train': 1.6518136262893677}
+02/26/2022 01:33:08 - INFO - codeparrot_training - Step 37051: {'lr': 8.4537066227173e-05, 'samples': 18970624, 'steps': 37051, 'loss/train': 1.6124365329742432}
+02/26/2022 01:33:11 - INFO - codeparrot_training - Step 37052: {'lr': 8.452480071141927e-05, 'samples': 18971136, 'steps': 37052, 'loss/train': 0.1726987361907959}
+02/26/2022 01:33:17 - INFO - codeparrot_training - Step 37053: {'lr': 8.451253590450875e-05, 'samples': 18971648, 'steps': 37053, 'loss/train': 2.2282145023345947}
+02/26/2022 01:33:20 - INFO - codeparrot_training - Step 37054: {'lr': 8.450027180649389e-05, 'samples': 18972160, 'steps': 37054, 'loss/train': 2.0039875507354736}
+02/26/2022 01:33:26 - INFO - codeparrot_training - Step 37055: {'lr': 8.448800841742741e-05, 'samples': 18972672, 'steps': 37055, 'loss/train': 1.7905770540237427}
+02/26/2022 01:33:29 - INFO - codeparrot_training - Step 37056: {'lr': 8.447574573736172e-05, 'samples': 18973184, 'steps': 37056, 'loss/train': 1.8973604440689087}
+02/26/2022 01:33:35 - INFO - codeparrot_training - Step 37057: {'lr': 8.446348376634937e-05, 'samples': 18973696, 'steps': 37057, 'loss/train': 2.0184977054595947}
+02/26/2022 01:33:38 - INFO - codeparrot_training - Step 37058: {'lr': 8.445122250444284e-05, 'samples': 18974208, 'steps': 37058, 'loss/train': 1.7296326160430908}
+02/26/2022 01:33:44 - INFO - codeparrot_training - Step 37059: {'lr': 8.44389619516948e-05, 'samples': 18974720, 'steps': 37059, 'loss/train': 1.8135051727294922}
+02/26/2022 01:33:47 - INFO - codeparrot_training - Step 37060: {'lr': 8.442670210815756e-05, 'samples': 18975232, 'steps': 37060, 'loss/train': 2.1664235591888428}
+02/26/2022 01:33:53 - INFO - codeparrot_training - Step 37061: {'lr': 8.441444297388393e-05, 'samples': 18975744, 'steps': 37061, 'loss/train': 1.0071851015090942}
+02/26/2022 01:33:56 - INFO - codeparrot_training - Step 37062: {'lr': 8.440218454892606e-05, 'samples': 18976256, 'steps': 37062, 'loss/train': 2.84360408782959}
+02/26/2022 01:34:02 - INFO - codeparrot_training - Step 37063: {'lr': 8.438992683333677e-05, 'samples': 18976768, 'steps': 37063, 'loss/train': 1.911067247390747}
+02/26/2022 01:34:05 - INFO - codeparrot_training - Step 37064: {'lr': 8.437766982716835e-05, 'samples': 18977280, 'steps': 37064, 'loss/train': 1.918104887008667}
+02/26/2022 01:34:12 - INFO - codeparrot_training - Step 37065: {'lr': 8.436541353047356e-05, 'samples': 18977792, 'steps': 37065, 'loss/train': 1.3108104467391968}
+02/26/2022 01:34:16 - INFO - codeparrot_training - Step 37066: {'lr': 8.435315794330456e-05, 'samples': 18978304, 'steps': 37066, 'loss/train': 0.42056167125701904}
+02/26/2022 01:34:21 - INFO - codeparrot_training - Step 37067: {'lr': 8.434090306571415e-05, 'samples': 18978816, 'steps': 37067, 'loss/train': 1.9125841856002808}
+02/26/2022 01:34:25 - INFO - codeparrot_training - Step 37068: {'lr': 8.43286488977546e-05, 'samples': 18979328, 'steps': 37068, 'loss/train': 3.784729242324829}
+02/26/2022 01:34:30 - INFO - codeparrot_training - Step 37069: {'lr': 8.43163954394786e-05, 'samples': 18979840, 'steps': 37069, 'loss/train': 1.6972370147705078}
+02/26/2022 01:34:34 - INFO - codeparrot_training - Step 37070: {'lr': 8.430414269093856e-05, 'samples': 18980352, 'steps': 37070, 'loss/train': 1.8246432542800903}
+02/26/2022 01:34:40 - INFO - codeparrot_training - Step 37071: {'lr': 8.429189065218695e-05, 'samples': 18980864, 'steps': 37071, 'loss/train': 1.7695115804672241}
+02/26/2022 01:34:43 - INFO - codeparrot_training - Step 37072: {'lr': 8.42796393232762e-05, 'samples': 18981376, 'steps': 37072, 'loss/train': 1.4122822284698486}
+02/26/2022 01:34:48 - INFO - codeparrot_training - Step 37073: {'lr': 8.42673887042589e-05, 'samples': 18981888, 'steps': 37073, 'loss/train': 1.8022700548171997}
+02/26/2022 01:34:52 - INFO - codeparrot_training - Step 37074: {'lr': 8.425513879518751e-05, 'samples': 18982400, 'steps': 37074, 'loss/train': 2.4020347595214844}
+02/26/2022 01:34:59 - INFO - codeparrot_training - Step 37075: {'lr': 8.424288959611445e-05, 'samples': 18982912, 'steps': 37075, 'loss/train': 2.6910476684570312}
+02/26/2022 01:35:02 - INFO - codeparrot_training - Step 37076: {'lr': 8.423064110709216e-05, 'samples': 18983424, 'steps': 37076, 'loss/train': 1.259127140045166}
+02/26/2022 01:35:08 - INFO - codeparrot_training - Step 37077: {'lr': 8.421839332817324e-05, 'samples': 18983936, 'steps': 37077, 'loss/train': 2.249293565750122}
+02/26/2022 01:35:11 - INFO - codeparrot_training - Step 37078: {'lr': 8.420614625941009e-05, 'samples': 18984448, 'steps': 37078, 'loss/train': 1.480204463005066}
+02/26/2022 01:35:17 - INFO - codeparrot_training - Step 37079: {'lr': 8.419389990085515e-05, 'samples': 18984960, 'steps': 37079, 'loss/train': 1.628015160560608}
+02/26/2022 01:35:20 - INFO - codeparrot_training - Step 37080: {'lr': 8.41816542525608e-05, 'samples': 18985472, 'steps': 37080, 'loss/train': 1.1605656147003174}
+02/26/2022 01:35:26 - INFO - codeparrot_training - Step 37081: {'lr': 8.41694093145797e-05, 'samples': 18985984, 'steps': 37081, 'loss/train': 0.7357677817344666}
+02/26/2022 01:35:29 - INFO - codeparrot_training - Step 37082: {'lr': 8.415716508696416e-05, 'samples': 18986496, 'steps': 37082, 'loss/train': 2.522886037826538}
+02/26/2022 01:35:35 - INFO - codeparrot_training - Step 37083: {'lr': 8.414492156976669e-05, 'samples': 18987008, 'steps': 37083, 'loss/train': 1.1562961339950562}
+02/26/2022 01:35:38 - INFO - codeparrot_training - Step 37084: {'lr': 8.413267876303964e-05, 'samples': 18987520, 'steps': 37084, 'loss/train': 1.4469730854034424}
+02/26/2022 01:35:45 - INFO - codeparrot_training - Step 37085: {'lr': 8.412043666683559e-05, 'samples': 18988032, 'steps': 37085, 'loss/train': 1.6758081912994385}
+02/26/2022 01:35:48 - INFO - codeparrot_training - Step 37086: {'lr': 8.410819528120692e-05, 'samples': 18988544, 'steps': 37086, 'loss/train': 1.744964599609375}
+02/26/2022 01:35:54 - INFO - codeparrot_training - Step 37087: {'lr': 8.409595460620605e-05, 'samples': 18989056, 'steps': 37087, 'loss/train': 0.8033421635627747}
+02/26/2022 01:35:58 - INFO - codeparrot_training - Step 37088: {'lr': 8.408371464188536e-05, 'samples': 18989568, 'steps': 37088, 'loss/train': 1.8627506494522095}
+02/26/2022 01:36:03 - INFO - codeparrot_training - Step 37089: {'lr': 8.407147538829743e-05, 'samples': 18990080, 'steps': 37089, 'loss/train': 2.1073427200317383}
+02/26/2022 01:36:07 - INFO - codeparrot_training - Step 37090: {'lr': 8.405923684549464e-05, 'samples': 18990592, 'steps': 37090, 'loss/train': 1.9563995599746704}
+02/26/2022 01:36:12 - INFO - codeparrot_training - Step 37091: {'lr': 8.404699901352936e-05, 'samples': 18991104, 'steps': 37091, 'loss/train': 1.3462523221969604}
+02/26/2022 01:36:16 - INFO - codeparrot_training - Step 37092: {'lr': 8.403476189245404e-05, 'samples': 18991616, 'steps': 37092, 'loss/train': 1.350356101989746}
+02/26/2022 01:36:21 - INFO - codeparrot_training - Step 37093: {'lr': 8.4022525482321e-05, 'samples': 18992128, 'steps': 37093, 'loss/train': 1.7961091995239258}
+02/26/2022 01:36:25 - INFO - codeparrot_training - Step 37094: {'lr': 8.401028978318287e-05, 'samples': 18992640, 'steps': 37094, 'loss/train': 2.33398175239563}
+02/26/2022 01:36:31 - INFO - codeparrot_training - Step 37095: {'lr': 8.399805479509188e-05, 'samples': 18993152, 'steps': 37095, 'loss/train': 1.6561747789382935}
+02/26/2022 01:36:35 - INFO - codeparrot_training - Step 37096: {'lr': 8.398582051810064e-05, 'samples': 18993664, 'steps': 37096, 'loss/train': 2.04085111618042}
+02/26/2022 01:36:40 - INFO - codeparrot_training - Step 37097: {'lr': 8.39735869522613e-05, 'samples': 18994176, 'steps': 37097, 'loss/train': 0.5589407086372375}
+02/26/2022 01:36:44 - INFO - codeparrot_training - Step 37098: {'lr': 8.396135409762643e-05, 'samples': 18994688, 'steps': 37098, 'loss/train': 1.7441991567611694}
+02/26/2022 01:36:49 - INFO - codeparrot_training - Step 37099: {'lr': 8.394912195424834e-05, 'samples': 18995200, 'steps': 37099, 'loss/train': 1.0083661079406738}
+02/26/2022 01:36:53 - INFO - codeparrot_training - Step 37100: {'lr': 8.393689052217964e-05, 'samples': 18995712, 'steps': 37100, 'loss/train': 2.9657833576202393}
+02/26/2022 01:36:58 - INFO - codeparrot_training - Step 37101: {'lr': 8.392465980147243e-05, 'samples': 18996224, 'steps': 37101, 'loss/train': 1.6664283275604248}
+02/26/2022 01:37:02 - INFO - codeparrot_training - Step 37102: {'lr': 8.391242979217931e-05, 'samples': 18996736, 'steps': 37102, 'loss/train': 0.24867655336856842}
+02/26/2022 01:37:07 - INFO - codeparrot_training - Step 37103: {'lr': 8.390020049435252e-05, 'samples': 18997248, 'steps': 37103, 'loss/train': 1.1057223081588745}
+02/26/2022 01:37:11 - INFO - codeparrot_training - Step 37104: {'lr': 8.388797190804468e-05, 'samples': 18997760, 'steps': 37104, 'loss/train': 2.328911066055298}
+02/26/2022 01:37:16 - INFO - codeparrot_training - Step 37105: {'lr': 8.387574403330786e-05, 'samples': 18998272, 'steps': 37105, 'loss/train': 1.9053987264633179}
+02/26/2022 01:37:22 - INFO - codeparrot_training - Step 37106: {'lr': 8.38635168701947e-05, 'samples': 18998784, 'steps': 37106, 'loss/train': 1.6752281188964844}
+02/26/2022 01:37:25 - INFO - codeparrot_training - Step 37107: {'lr': 8.385129041875738e-05, 'samples': 18999296, 'steps': 37107, 'loss/train': 1.4451733827590942}
+02/26/2022 01:37:31 - INFO - codeparrot_training - Step 37108: {'lr': 8.383906467904848e-05, 'samples': 18999808, 'steps': 37108, 'loss/train': 0.29254552721977234}
+02/26/2022 01:37:34 - INFO - codeparrot_training - Step 37109: {'lr': 8.382683965112026e-05, 'samples': 19000320, 'steps': 37109, 'loss/train': 1.0294314622879028}
+02/26/2022 01:37:41 - INFO - codeparrot_training - Step 37110: {'lr': 8.381461533502507e-05, 'samples': 19000832, 'steps': 37110, 'loss/train': 2.5762734413146973}
+02/26/2022 01:37:45 - INFO - codeparrot_training - Step 37111: {'lr': 8.380239173081525e-05, 'samples': 19001344, 'steps': 37111, 'loss/train': 1.4563014507293701}
+02/26/2022 01:37:50 - INFO - codeparrot_training - Step 37112: {'lr': 8.379016883854327e-05, 'samples': 19001856, 'steps': 37112, 'loss/train': 1.1990443468093872}
+02/26/2022 01:37:54 - INFO - codeparrot_training - Step 37113: {'lr': 8.377794665826145e-05, 'samples': 19002368, 'steps': 37113, 'loss/train': 2.0943968296051025}
+02/26/2022 01:37:59 - INFO - codeparrot_training - Step 37114: {'lr': 8.37657251900221e-05, 'samples': 19002880, 'steps': 37114, 'loss/train': 0.9344554543495178}
+02/26/2022 01:38:03 - INFO - codeparrot_training - Step 37115: {'lr': 8.37535044338775e-05, 'samples': 19003392, 'steps': 37115, 'loss/train': 1.2807188034057617}
+02/26/2022 01:38:08 - INFO - codeparrot_training - Step 37116: {'lr': 8.374128438988021e-05, 'samples': 19003904, 'steps': 37116, 'loss/train': 1.7306482791900635}
+02/26/2022 01:38:12 - INFO - codeparrot_training - Step 37117: {'lr': 8.372906505808236e-05, 'samples': 19004416, 'steps': 37117, 'loss/train': 1.8291391134262085}
+02/26/2022 01:38:17 - INFO - codeparrot_training - Step 37118: {'lr': 8.371684643853658e-05, 'samples': 19004928, 'steps': 37118, 'loss/train': 0.6260422468185425}
+02/26/2022 01:38:21 - INFO - codeparrot_training - Step 37119: {'lr': 8.370462853129488e-05, 'samples': 19005440, 'steps': 37119, 'loss/train': 2.3483242988586426}
+02/26/2022 01:38:28 - INFO - codeparrot_training - Step 37120: {'lr': 8.369241133640981e-05, 'samples': 19005952, 'steps': 37120, 'loss/train': 1.5019205808639526}
+02/26/2022 01:38:31 - INFO - codeparrot_training - Step 37121: {'lr': 8.368019485393355e-05, 'samples': 19006464, 'steps': 37121, 'loss/train': 1.6771647930145264}
+02/26/2022 01:38:37 - INFO - codeparrot_training - Step 37122: {'lr': 8.366797908391871e-05, 'samples': 19006976, 'steps': 37122, 'loss/train': 2.6337990760803223}
+02/26/2022 01:38:41 - INFO - codeparrot_training - Step 37123: {'lr': 8.365576402641726e-05, 'samples': 19007488, 'steps': 37123, 'loss/train': 1.7889511585235596}
+02/26/2022 01:38:46 - INFO - codeparrot_training - Step 37124: {'lr': 8.364354968148177e-05, 'samples': 19008000, 'steps': 37124, 'loss/train': 1.8414673805236816}
+02/26/2022 01:38:50 - INFO - codeparrot_training - Step 37125: {'lr': 8.363133604916447e-05, 'samples': 19008512, 'steps': 37125, 'loss/train': 1.8435741662979126}
+02/26/2022 01:38:55 - INFO - codeparrot_training - Step 37126: {'lr': 8.361912312951772e-05, 'samples': 19009024, 'steps': 37126, 'loss/train': 1.9183146953582764}
+02/26/2022 01:38:59 - INFO - codeparrot_training - Step 37127: {'lr': 8.360691092259373e-05, 'samples': 19009536, 'steps': 37127, 'loss/train': 1.0897890329360962}
+02/26/2022 01:39:04 - INFO - codeparrot_training - Step 37128: {'lr': 8.3594699428445e-05, 'samples': 19010048, 'steps': 37128, 'loss/train': 2.2542364597320557}
+02/26/2022 01:39:08 - INFO - codeparrot_training - Step 37129: {'lr': 8.358248864712373e-05, 'samples': 19010560, 'steps': 37129, 'loss/train': 1.9320359230041504}
+02/26/2022 01:39:15 - INFO - codeparrot_training - Step 37130: {'lr': 8.357027857868212e-05, 'samples': 19011072, 'steps': 37130, 'loss/train': 2.602360963821411}
+02/26/2022 01:39:18 - INFO - codeparrot_training - Step 37131: {'lr': 8.355806922317271e-05, 'samples': 19011584, 'steps': 37131, 'loss/train': 1.882552146911621}
+02/26/2022 01:39:24 - INFO - codeparrot_training - Step 37132: {'lr': 8.354586058064766e-05, 'samples': 19012096, 'steps': 37132, 'loss/train': 1.3717589378356934}
+02/26/2022 01:39:27 - INFO - codeparrot_training - Step 37133: {'lr': 8.353365265115931e-05, 'samples': 19012608, 'steps': 37133, 'loss/train': 0.8835031986236572}
+02/26/2022 01:39:33 - INFO - codeparrot_training - Step 37134: {'lr': 8.352144543475981e-05, 'samples': 19013120, 'steps': 37134, 'loss/train': 1.2702559232711792}
+02/26/2022 01:39:36 - INFO - codeparrot_training - Step 37135: {'lr': 8.35092389315017e-05, 'samples': 19013632, 'steps': 37135, 'loss/train': 2.1608073711395264}
+02/26/2022 01:39:42 - INFO - codeparrot_training - Step 37136: {'lr': 8.349703314143711e-05, 'samples': 19014144, 'steps': 37136, 'loss/train': 0.49818217754364014}
+02/26/2022 01:39:46 - INFO - codeparrot_training - Step 37137: {'lr': 8.348482806461838e-05, 'samples': 19014656, 'steps': 37137, 'loss/train': 2.246999740600586}
+02/26/2022 01:39:49 - INFO - codeparrot_training - Step 37138: {'lr': 8.347262370109768e-05, 'samples': 19015168, 'steps': 37138, 'loss/train': 1.8392107486724854}
+02/26/2022 01:39:55 - INFO - codeparrot_training - Step 37139: {'lr': 8.346042005092754e-05, 'samples': 19015680, 'steps': 37139, 'loss/train': 2.2710347175598145}
+02/26/2022 01:39:58 - INFO - codeparrot_training - Step 37140: {'lr': 8.344821711415995e-05, 'samples': 19016192, 'steps': 37140, 'loss/train': 4.144713401794434}
+02/26/2022 01:40:04 - INFO - codeparrot_training - Step 37141: {'lr': 8.343601489084738e-05, 'samples': 19016704, 'steps': 37141, 'loss/train': 1.1207575798034668}
+02/26/2022 01:40:07 - INFO - codeparrot_training - Step 37142: {'lr': 8.342381338104191e-05, 'samples': 19017216, 'steps': 37142, 'loss/train': 1.3024550676345825}
+02/26/2022 01:40:14 - INFO - codeparrot_training - Step 37143: {'lr': 8.341161258479605e-05, 'samples': 19017728, 'steps': 37143, 'loss/train': 1.420929193496704}
+02/26/2022 01:40:18 - INFO - codeparrot_training - Step 37144: {'lr': 8.339941250216194e-05, 'samples': 19018240, 'steps': 37144, 'loss/train': 1.2910969257354736}
+02/26/2022 01:40:23 - INFO - codeparrot_training - Step 37145: {'lr': 8.338721313319183e-05, 'samples': 19018752, 'steps': 37145, 'loss/train': 1.5712597370147705}
+02/26/2022 01:40:27 - INFO - codeparrot_training - Step 37146: {'lr': 8.337501447793791e-05, 'samples': 19019264, 'steps': 37146, 'loss/train': 1.859776496887207}
+02/26/2022 01:40:32 - INFO - codeparrot_training - Step 37147: {'lr': 8.336281653645261e-05, 'samples': 19019776, 'steps': 37147, 'loss/train': 0.035972096025943756}
+02/26/2022 01:40:36 - INFO - codeparrot_training - Step 37148: {'lr': 8.33506193087881e-05, 'samples': 19020288, 'steps': 37148, 'loss/train': 1.9353622198104858}
+02/26/2022 01:40:41 - INFO - codeparrot_training - Step 37149: {'lr': 8.333842279499663e-05, 'samples': 19020800, 'steps': 37149, 'loss/train': 0.588671088218689}
+02/26/2022 01:40:45 - INFO - codeparrot_training - Step 37150: {'lr': 8.332622699513032e-05, 'samples': 19021312, 'steps': 37150, 'loss/train': 2.456263780593872}
+02/26/2022 01:40:50 - INFO - codeparrot_training - Step 37151: {'lr': 8.331403190924164e-05, 'samples': 19021824, 'steps': 37151, 'loss/train': 1.5952991247177124}
+02/26/2022 01:40:54 - INFO - codeparrot_training - Step 37152: {'lr': 8.330183753738269e-05, 'samples': 19022336, 'steps': 37152, 'loss/train': 2.2287089824676514}
+02/26/2022 01:40:59 - INFO - codeparrot_training - Step 37153: {'lr': 8.328964387960574e-05, 'samples': 19022848, 'steps': 37153, 'loss/train': 1.2397838830947876}
+02/26/2022 01:41:03 - INFO - codeparrot_training - Step 37154: {'lr': 8.327745093596295e-05, 'samples': 19023360, 'steps': 37154, 'loss/train': 3.152423143386841}
+02/26/2022 01:41:08 - INFO - codeparrot_training - Step 37155: {'lr': 8.32652587065067e-05, 'samples': 19023872, 'steps': 37155, 'loss/train': 1.4935261011123657}
+02/26/2022 01:41:12 - INFO - codeparrot_training - Step 37156: {'lr': 8.325306719128906e-05, 'samples': 19024384, 'steps': 37156, 'loss/train': 1.400776743888855}
+02/26/2022 01:41:18 - INFO - codeparrot_training - Step 37157: {'lr': 8.324087639036248e-05, 'samples': 19024896, 'steps': 37157, 'loss/train': 1.7071540355682373}
+02/26/2022 01:41:24 - INFO - codeparrot_training - Step 37158: {'lr': 8.322868630377889e-05, 'samples': 19025408, 'steps': 37158, 'loss/train': 1.6731566190719604}
+02/26/2022 01:41:27 - INFO - codeparrot_training - Step 37159: {'lr': 8.321649693159072e-05, 'samples': 19025920, 'steps': 37159, 'loss/train': 0.8045114874839783}
+02/26/2022 01:41:33 - INFO - codeparrot_training - Step 37160: {'lr': 8.320430827385004e-05, 'samples': 19026432, 'steps': 37160, 'loss/train': 1.7433161735534668}
+02/26/2022 01:41:36 - INFO - codeparrot_training - Step 37161: {'lr': 8.31921203306093e-05, 'samples': 19026944, 'steps': 37161, 'loss/train': 0.38707539439201355}
+02/26/2022 01:41:42 - INFO - codeparrot_training - Step 37162: {'lr': 8.317993310192041e-05, 'samples': 19027456, 'steps': 37162, 'loss/train': 1.9996532201766968}
+02/26/2022 01:41:45 - INFO - codeparrot_training - Step 37163: {'lr': 8.316774658783577e-05, 'samples': 19027968, 'steps': 37163, 'loss/train': 1.295071005821228}
+02/26/2022 01:41:51 - INFO - codeparrot_training - Step 37164: {'lr': 8.315556078840744e-05, 'samples': 19028480, 'steps': 37164, 'loss/train': 1.472206950187683}
+02/26/2022 01:41:54 - INFO - codeparrot_training - Step 37165: {'lr': 8.314337570368787e-05, 'samples': 19028992, 'steps': 37165, 'loss/train': 1.3874530792236328}
+02/26/2022 01:42:01 - INFO - codeparrot_training - Step 37166: {'lr': 8.313119133372896e-05, 'samples': 19029504, 'steps': 37166, 'loss/train': 1.5878446102142334}
+02/26/2022 01:42:04 - INFO - codeparrot_training - Step 37167: {'lr': 8.31190076785831e-05, 'samples': 19030016, 'steps': 37167, 'loss/train': 2.5487284660339355}
+02/26/2022 01:42:10 - INFO - codeparrot_training - Step 37168: {'lr': 8.310682473830236e-05, 'samples': 19030528, 'steps': 37168, 'loss/train': 1.359023928642273}
+02/26/2022 01:42:13 - INFO - codeparrot_training - Step 37169: {'lr': 8.309464251293905e-05, 'samples': 19031040, 'steps': 37169, 'loss/train': 1.8767603635787964}
+02/26/2022 01:42:19 - INFO - codeparrot_training - Step 37170: {'lr': 8.308246100254529e-05, 'samples': 19031552, 'steps': 37170, 'loss/train': 1.4964475631713867}
+02/26/2022 01:42:22 - INFO - codeparrot_training - Step 37171: {'lr': 8.307028020717328e-05, 'samples': 19032064, 'steps': 37171, 'loss/train': 0.9305400848388672}
+02/26/2022 01:42:28 - INFO - codeparrot_training - Step 37172: {'lr': 8.305810012687518e-05, 'samples': 19032576, 'steps': 37172, 'loss/train': 2.2013516426086426}
+02/26/2022 01:42:31 - INFO - codeparrot_training - Step 37173: {'lr': 8.304592076170309e-05, 'samples': 19033088, 'steps': 37173, 'loss/train': 1.4422435760498047}
+02/26/2022 01:42:37 - INFO - codeparrot_training - Step 37174: {'lr': 8.303374211170933e-05, 'samples': 19033600, 'steps': 37174, 'loss/train': 1.7583857774734497}
+02/26/2022 01:42:40 - INFO - codeparrot_training - Step 37175: {'lr': 8.302156417694601e-05, 'samples': 19034112, 'steps': 37175, 'loss/train': 1.3351573944091797}
+02/26/2022 01:42:46 - INFO - codeparrot_training - Step 37176: {'lr': 8.300938695746529e-05, 'samples': 19034624, 'steps': 37176, 'loss/train': 1.1454607248306274}
+02/26/2022 01:42:49 - INFO - codeparrot_training - Step 37177: {'lr': 8.299721045331923e-05, 'samples': 19035136, 'steps': 37177, 'loss/train': 1.6435911655426025}
+02/26/2022 01:42:56 - INFO - codeparrot_training - Step 37178: {'lr': 8.29850346645602e-05, 'samples': 19035648, 'steps': 37178, 'loss/train': 1.7206287384033203}
+02/26/2022 01:43:00 - INFO - codeparrot_training - Step 37179: {'lr': 8.297285959124023e-05, 'samples': 19036160, 'steps': 37179, 'loss/train': 0.9274430871009827}
+02/26/2022 01:43:05 - INFO - codeparrot_training - Step 37180: {'lr': 8.296068523341152e-05, 'samples': 19036672, 'steps': 37180, 'loss/train': 1.7284140586853027}
+02/26/2022 01:43:09 - INFO - codeparrot_training - Step 37181: {'lr': 8.294851159112606e-05, 'samples': 19037184, 'steps': 37181, 'loss/train': 1.8588786125183105}
+02/26/2022 01:43:15 - INFO - codeparrot_training - Step 37182: {'lr': 8.293633866443625e-05, 'samples': 19037696, 'steps': 37182, 'loss/train': 2.407111644744873}
+02/26/2022 01:43:18 - INFO - codeparrot_training - Step 37183: {'lr': 8.292416645339413e-05, 'samples': 19038208, 'steps': 37183, 'loss/train': 2.17269229888916}
+02/26/2022 01:43:23 - INFO - codeparrot_training - Step 37184: {'lr': 8.29119949580518e-05, 'samples': 19038720, 'steps': 37184, 'loss/train': 0.7622824907302856}
+02/26/2022 01:43:27 - INFO - codeparrot_training - Step 37185: {'lr': 8.289982417846134e-05, 'samples': 19039232, 'steps': 37185, 'loss/train': 0.922652542591095}
+02/26/2022 01:43:33 - INFO - codeparrot_training - Step 37186: {'lr': 8.288765411467508e-05, 'samples': 19039744, 'steps': 37186, 'loss/train': 2.1901113986968994}
+02/26/2022 01:43:36 - INFO - codeparrot_training - Step 37187: {'lr': 8.287548476674502e-05, 'samples': 19040256, 'steps': 37187, 'loss/train': 1.3653470277786255}
+02/26/2022 01:43:43 - INFO - codeparrot_training - Step 37188: {'lr': 8.286331613472334e-05, 'samples': 19040768, 'steps': 37188, 'loss/train': 2.1037349700927734}
+02/26/2022 01:43:47 - INFO - codeparrot_training - Step 37189: {'lr': 8.285114821866204e-05, 'samples': 19041280, 'steps': 37189, 'loss/train': 1.6034964323043823}
+02/26/2022 01:43:52 - INFO - codeparrot_training - Step 37190: {'lr': 8.283898101861345e-05, 'samples': 19041792, 'steps': 37190, 'loss/train': 1.3340699672698975}
+02/26/2022 01:43:56 - INFO - codeparrot_training - Step 37191: {'lr': 8.282681453462948e-05, 'samples': 19042304, 'steps': 37191, 'loss/train': 1.359729528427124}
+02/26/2022 01:44:01 - INFO - codeparrot_training - Step 37192: {'lr': 8.281464876676253e-05, 'samples': 19042816, 'steps': 37192, 'loss/train': 2.09037446975708}
+02/26/2022 01:44:05 - INFO - codeparrot_training - Step 37193: {'lr': 8.280248371506438e-05, 'samples': 19043328, 'steps': 37193, 'loss/train': 1.3655472993850708}
+02/26/2022 01:44:10 - INFO - codeparrot_training - Step 37194: {'lr': 8.279031937958736e-05, 'samples': 19043840, 'steps': 37194, 'loss/train': 2.0710277557373047}
+02/26/2022 01:44:14 - INFO - codeparrot_training - Step 37195: {'lr': 8.277815576038345e-05, 'samples': 19044352, 'steps': 37195, 'loss/train': 1.6258600950241089}
+02/26/2022 01:44:19 - INFO - codeparrot_training - Step 37196: {'lr': 8.276599285750499e-05, 'samples': 19044864, 'steps': 37196, 'loss/train': 1.4014722108840942}
+02/26/2022 01:44:23 - INFO - codeparrot_training - Step 37197: {'lr': 8.275383067100376e-05, 'samples': 19045376, 'steps': 37197, 'loss/train': 1.31800377368927}
+02/26/2022 01:44:28 - INFO - codeparrot_training - Step 37198: {'lr': 8.274166920093207e-05, 'samples': 19045888, 'steps': 37198, 'loss/train': 1.0466171503067017}
+02/26/2022 01:44:32 - INFO - codeparrot_training - Step 37199: {'lr': 8.272950844734186e-05, 'samples': 19046400, 'steps': 37199, 'loss/train': 1.8293547630310059}
+02/26/2022 01:44:37 - INFO - codeparrot_training - Step 37200: {'lr': 8.271734841028553e-05, 'samples': 19046912, 'steps': 37200, 'loss/train': 1.7393453121185303}
+02/26/2022 01:44:41 - INFO - codeparrot_training - Step 37201: {'lr': 8.270518908981476e-05, 'samples': 19047424, 'steps': 37201, 'loss/train': 2.0252933502197266}
+02/26/2022 01:44:46 - INFO - codeparrot_training - Step 37202: {'lr': 8.269303048598195e-05, 'samples': 19047936, 'steps': 37202, 'loss/train': 1.845220923423767}
+02/26/2022 01:44:50 - INFO - codeparrot_training - Step 37203: {'lr': 8.268087259883897e-05, 'samples': 19048448, 'steps': 37203, 'loss/train': 1.7153112888336182}
+02/26/2022 01:44:56 - INFO - codeparrot_training - Step 37204: {'lr': 8.266871542843809e-05, 'samples': 19048960, 'steps': 37204, 'loss/train': 1.5581583976745605}
+02/26/2022 01:45:00 - INFO - codeparrot_training - Step 37205: {'lr': 8.26565589748313e-05, 'samples': 19049472, 'steps': 37205, 'loss/train': 2.1535165309906006}
+02/26/2022 01:45:05 - INFO - codeparrot_training - Step 37206: {'lr': 8.264440323807065e-05, 'samples': 19049984, 'steps': 37206, 'loss/train': 1.6526063680648804}
+02/26/2022 01:45:09 - INFO - codeparrot_training - Step 37207: {'lr': 8.26322482182082e-05, 'samples': 19050496, 'steps': 37207, 'loss/train': 1.8197089433670044}
+02/26/2022 01:45:14 - INFO - codeparrot_training - Step 37208: {'lr': 8.262009391529609e-05, 'samples': 19051008, 'steps': 37208, 'loss/train': 0.7985866665840149}
+02/26/2022 01:45:18 - INFO - codeparrot_training - Step 37209: {'lr': 8.260794032938637e-05, 'samples': 19051520, 'steps': 37209, 'loss/train': 1.6394752264022827}
+02/26/2022 01:45:23 - INFO - codeparrot_training - Step 37210: {'lr': 8.259578746053109e-05, 'samples': 19052032, 'steps': 37210, 'loss/train': 1.7461438179016113}
+02/26/2022 01:45:27 - INFO - codeparrot_training - Step 37211: {'lr': 8.258363530878222e-05, 'samples': 19052544, 'steps': 37211, 'loss/train': 1.7207915782928467}
+02/26/2022 01:45:32 - INFO - codeparrot_training - Step 37212: {'lr': 8.257148387419194e-05, 'samples': 19053056, 'steps': 37212, 'loss/train': 1.7985799312591553}
+02/26/2022 01:45:36 - INFO - codeparrot_training - Step 37213: {'lr': 8.25593331568123e-05, 'samples': 19053568, 'steps': 37213, 'loss/train': 1.9586187601089478}
+02/26/2022 01:45:42 - INFO - codeparrot_training - Step 37214: {'lr': 8.254718315669529e-05, 'samples': 19054080, 'steps': 37214, 'loss/train': 2.7502715587615967}
+02/26/2022 01:45:46 - INFO - codeparrot_training - Step 37215: {'lr': 8.253503387389289e-05, 'samples': 19054592, 'steps': 37215, 'loss/train': 1.0251836776733398}
+02/26/2022 01:45:51 - INFO - codeparrot_training - Step 37216: {'lr': 8.252288530845733e-05, 'samples': 19055104, 'steps': 37216, 'loss/train': 1.1509389877319336}
+02/26/2022 01:45:55 - INFO - codeparrot_training - Step 37217: {'lr': 8.251073746044053e-05, 'samples': 19055616, 'steps': 37217, 'loss/train': 1.3730127811431885}
+02/26/2022 01:46:00 - INFO - codeparrot_training - Step 37218: {'lr': 8.249859032989454e-05, 'samples': 19056128, 'steps': 37218, 'loss/train': 0.052121300250291824}
+02/26/2022 01:46:04 - INFO - codeparrot_training - Step 37219: {'lr': 8.248644391687136e-05, 'samples': 19056640, 'steps': 37219, 'loss/train': 2.6654324531555176}
+02/26/2022 01:46:09 - INFO - codeparrot_training - Step 37220: {'lr': 8.247429822142311e-05, 'samples': 19057152, 'steps': 37220, 'loss/train': 0.9788916707038879}
+02/26/2022 01:46:13 - INFO - codeparrot_training - Step 37221: {'lr': 8.246215324360177e-05, 'samples': 19057664, 'steps': 37221, 'loss/train': 1.7591363191604614}
+02/26/2022 01:46:18 - INFO - codeparrot_training - Step 37222: {'lr': 8.245000898345936e-05, 'samples': 19058176, 'steps': 37222, 'loss/train': 0.528789222240448}
+02/26/2022 01:46:22 - INFO - codeparrot_training - Step 37223: {'lr': 8.243786544104792e-05, 'samples': 19058688, 'steps': 37223, 'loss/train': 0.2821579873561859}
+02/26/2022 01:46:28 - INFO - codeparrot_training - Step 37224: {'lr': 8.242572261641939e-05, 'samples': 19059200, 'steps': 37224, 'loss/train': 3.1132941246032715}
+02/26/2022 01:46:32 - INFO - codeparrot_training - Step 37225: {'lr': 8.241358050962591e-05, 'samples': 19059712, 'steps': 37225, 'loss/train': 0.6978120803833008}
+02/26/2022 01:46:37 - INFO - codeparrot_training - Step 37226: {'lr': 8.240143912071937e-05, 'samples': 19060224, 'steps': 37226, 'loss/train': 1.2478735446929932}
+02/26/2022 01:46:41 - INFO - codeparrot_training - Step 37227: {'lr': 8.238929844975201e-05, 'samples': 19060736, 'steps': 37227, 'loss/train': 2.6061248779296875}
+02/26/2022 01:46:46 - INFO - codeparrot_training - Step 37228: {'lr': 8.237715849677552e-05, 'samples': 19061248, 'steps': 37228, 'loss/train': 0.9349194169044495}
+02/26/2022 01:46:50 - INFO - codeparrot_training - Step 37229: {'lr': 8.236501926184212e-05, 'samples': 19061760, 'steps': 37229, 'loss/train': 0.5367432832717896}
+02/26/2022 01:46:55 - INFO - codeparrot_training - Step 37230: {'lr': 8.235288074500366e-05, 'samples': 19062272, 'steps': 37230, 'loss/train': 1.963213562965393}
+02/26/2022 01:46:59 - INFO - codeparrot_training - Step 37231: {'lr': 8.234074294631241e-05, 'samples': 19062784, 'steps': 37231, 'loss/train': 2.2465946674346924}
+02/26/2022 01:47:04 - INFO - codeparrot_training - Step 37232: {'lr': 8.232860586582e-05, 'samples': 19063296, 'steps': 37232, 'loss/train': 0.9786977171897888}
+02/26/2022 01:47:08 - INFO - codeparrot_training - Step 37233: {'lr': 8.231646950357868e-05, 'samples': 19063808, 'steps': 37233, 'loss/train': 2.409968852996826}
+02/26/2022 01:47:14 - INFO - codeparrot_training - Step 37234: {'lr': 8.230433385964028e-05, 'samples': 19064320, 'steps': 37234, 'loss/train': 1.1553808450698853}
+02/26/2022 01:47:18 - INFO - codeparrot_training - Step 37235: {'lr': 8.229219893405704e-05, 'samples': 19064832, 'steps': 37235, 'loss/train': 1.9721091985702515}
+02/26/2022 01:47:23 - INFO - codeparrot_training - Step 37236: {'lr': 8.22800647268806e-05, 'samples': 19065344, 'steps': 37236, 'loss/train': 1.7504615783691406}
+02/26/2022 01:47:27 - INFO - codeparrot_training - Step 37237: {'lr': 8.22679312381632e-05, 'samples': 19065856, 'steps': 37237, 'loss/train': 1.0252844095230103}
+02/26/2022 01:47:32 - INFO - codeparrot_training - Step 37238: {'lr': 8.22557984679566e-05, 'samples': 19066368, 'steps': 37238, 'loss/train': 1.9307035207748413}
+02/26/2022 01:47:36 - INFO - codeparrot_training - Step 37239: {'lr': 8.224366641631301e-05, 'samples': 19066880, 'steps': 37239, 'loss/train': 2.4583001136779785}
+02/26/2022 01:47:41 - INFO - codeparrot_training - Step 37240: {'lr': 8.223153508328429e-05, 'samples': 19067392, 'steps': 37240, 'loss/train': 1.5907905101776123}
+02/26/2022 01:47:45 - INFO - codeparrot_training - Step 37241: {'lr': 8.221940446892236e-05, 'samples': 19067904, 'steps': 37241, 'loss/train': 0.6465444564819336}
+02/26/2022 01:47:50 - INFO - codeparrot_training - Step 37242: {'lr': 8.220727457327917e-05, 'samples': 19068416, 'steps': 37242, 'loss/train': 1.1750423908233643}
+02/26/2022 01:47:54 - INFO - codeparrot_training - Step 37243: {'lr': 8.219514539640682e-05, 'samples': 19068928, 'steps': 37243, 'loss/train': 2.238308906555176}
+02/26/2022 01:47:59 - INFO - codeparrot_training - Step 37244: {'lr': 8.218301693835719e-05, 'samples': 19069440, 'steps': 37244, 'loss/train': 1.9969607591629028}
+02/26/2022 01:48:03 - INFO - codeparrot_training - Step 37245: {'lr': 8.217088919918223e-05, 'samples': 19069952, 'steps': 37245, 'loss/train': 1.8884025812149048}
+02/26/2022 01:48:08 - INFO - codeparrot_training - Step 37246: {'lr': 8.215876217893379e-05, 'samples': 19070464, 'steps': 37246, 'loss/train': 1.062110424041748}
+02/26/2022 01:48:12 - INFO - codeparrot_training - Step 37247: {'lr': 8.2146635877664e-05, 'samples': 19070976, 'steps': 37247, 'loss/train': 1.6500635147094727}
+02/26/2022 01:48:18 - INFO - codeparrot_training - Step 37248: {'lr': 8.213451029542471e-05, 'samples': 19071488, 'steps': 37248, 'loss/train': 0.808623194694519}
+02/26/2022 01:48:22 - INFO - codeparrot_training - Step 37249: {'lr': 8.212238543226791e-05, 'samples': 19072000, 'steps': 37249, 'loss/train': 2.1174252033233643}
+02/26/2022 01:48:27 - INFO - codeparrot_training - Step 37250: {'lr': 8.211026128824539e-05, 'samples': 19072512, 'steps': 37250, 'loss/train': 1.0835115909576416}
+02/26/2022 01:48:31 - INFO - codeparrot_training - Step 37251: {'lr': 8.209813786340927e-05, 'samples': 19073024, 'steps': 37251, 'loss/train': 1.1820204257965088}
+02/26/2022 01:48:36 - INFO - codeparrot_training - Step 37252: {'lr': 8.208601515781137e-05, 'samples': 19073536, 'steps': 37252, 'loss/train': 2.7489683628082275}
+02/26/2022 01:48:40 - INFO - codeparrot_training - Step 37253: {'lr': 8.207389317150382e-05, 'samples': 19074048, 'steps': 37253, 'loss/train': 2.659041404724121}
+02/26/2022 01:48:45 - INFO - codeparrot_training - Step 37254: {'lr': 8.20617719045382e-05, 'samples': 19074560, 'steps': 37254, 'loss/train': 1.430891752243042}
+02/26/2022 01:48:49 - INFO - codeparrot_training - Step 37255: {'lr': 8.204965135696674e-05, 'samples': 19075072, 'steps': 37255, 'loss/train': 1.758187174797058}
+02/26/2022 01:48:54 - INFO - codeparrot_training - Step 37256: {'lr': 8.203753152884122e-05, 'samples': 19075584, 'steps': 37256, 'loss/train': 1.899909496307373}
+02/26/2022 01:48:58 - INFO - codeparrot_training - Step 37257: {'lr': 8.202541242021358e-05, 'samples': 19076096, 'steps': 37257, 'loss/train': 1.2651687860488892}
+02/26/2022 01:49:04 - INFO - codeparrot_training - Step 37258: {'lr': 8.201329403113566e-05, 'samples': 19076608, 'steps': 37258, 'loss/train': 1.2146570682525635}
+02/26/2022 01:49:07 - INFO - codeparrot_training - Step 37259: {'lr': 8.200117636165952e-05, 'samples': 19077120, 'steps': 37259, 'loss/train': 1.2071138620376587}
+02/26/2022 01:49:13 - INFO - codeparrot_training - Step 37260: {'lr': 8.198905941183701e-05, 'samples': 19077632, 'steps': 37260, 'loss/train': 0.5950890779495239}
+02/26/2022 01:49:17 - INFO - codeparrot_training - Step 37261: {'lr': 8.197694318172e-05, 'samples': 19078144, 'steps': 37261, 'loss/train': 1.4595544338226318}
+02/26/2022 01:49:22 - INFO - codeparrot_training - Step 37262: {'lr': 8.196482767136035e-05, 'samples': 19078656, 'steps': 37262, 'loss/train': 1.4657697677612305}
+02/26/2022 01:49:26 - INFO - codeparrot_training - Step 37263: {'lr': 8.195271288081008e-05, 'samples': 19079168, 'steps': 37263, 'loss/train': 1.6008533239364624}
+02/26/2022 01:49:31 - INFO - codeparrot_training - Step 37264: {'lr': 8.194059881012105e-05, 'samples': 19079680, 'steps': 37264, 'loss/train': 0.9014922380447388}
+02/26/2022 01:49:35 - INFO - codeparrot_training - Step 37265: {'lr': 8.192848545934504e-05, 'samples': 19080192, 'steps': 37265, 'loss/train': 1.6199990510940552}
+02/26/2022 01:49:40 - INFO - codeparrot_training - Step 37266: {'lr': 8.191637282853412e-05, 'samples': 19080704, 'steps': 37266, 'loss/train': 1.428519606590271}
+02/26/2022 01:49:44 - INFO - codeparrot_training - Step 37267: {'lr': 8.190426091774006e-05, 'samples': 19081216, 'steps': 37267, 'loss/train': 1.2228596210479736}
+02/26/2022 01:49:50 - INFO - codeparrot_training - Step 37268: {'lr': 8.189214972701478e-05, 'samples': 19081728, 'steps': 37268, 'loss/train': 1.8246833086013794}
+02/26/2022 01:49:53 - INFO - codeparrot_training - Step 37269: {'lr': 8.188003925641008e-05, 'samples': 19082240, 'steps': 37269, 'loss/train': 0.9061700701713562}
+02/26/2022 01:50:00 - INFO - codeparrot_training - Step 37270: {'lr': 8.186792950597807e-05, 'samples': 19082752, 'steps': 37270, 'loss/train': 1.2816135883331299}
+02/26/2022 01:50:03 - INFO - codeparrot_training - Step 37271: {'lr': 8.185582047577029e-05, 'samples': 19083264, 'steps': 37271, 'loss/train': 1.4242651462554932}
+02/26/2022 01:50:08 - INFO - codeparrot_training - Step 37272: {'lr': 8.184371216583888e-05, 'samples': 19083776, 'steps': 37272, 'loss/train': 1.7851136922836304}
+02/26/2022 01:50:12 - INFO - codeparrot_training - Step 37273: {'lr': 8.18316045762355e-05, 'samples': 19084288, 'steps': 37273, 'loss/train': 2.043785572052002}
+02/26/2022 01:50:18 - INFO - codeparrot_training - Step 37274: {'lr': 8.181949770701233e-05, 'samples': 19084800, 'steps': 37274, 'loss/train': 2.125408411026001}
+02/26/2022 01:50:21 - INFO - codeparrot_training - Step 37275: {'lr': 8.180739155822084e-05, 'samples': 19085312, 'steps': 37275, 'loss/train': 1.7213927507400513}
+02/26/2022 01:50:27 - INFO - codeparrot_training - Step 37276: {'lr': 8.179528612991316e-05, 'samples': 19085824, 'steps': 37276, 'loss/train': 2.7691538333892822}
+02/26/2022 01:50:30 - INFO - codeparrot_training - Step 37277: {'lr': 8.178318142214097e-05, 'samples': 19086336, 'steps': 37277, 'loss/train': 1.141066312789917}
+02/26/2022 01:50:36 - INFO - codeparrot_training - Step 37278: {'lr': 8.177107743495632e-05, 'samples': 19086848, 'steps': 37278, 'loss/train': 2.1283280849456787}
+02/26/2022 01:50:39 - INFO - codeparrot_training - Step 37279: {'lr': 8.175897416841096e-05, 'samples': 19087360, 'steps': 37279, 'loss/train': 0.5556949973106384}
+02/26/2022 01:50:46 - INFO - codeparrot_training - Step 37280: {'lr': 8.174687162255672e-05, 'samples': 19087872, 'steps': 37280, 'loss/train': 0.8359056711196899}
+02/26/2022 01:50:50 - INFO - codeparrot_training - Step 37281: {'lr': 8.173476979744538e-05, 'samples': 19088384, 'steps': 37281, 'loss/train': 0.21715198457241058}
+02/26/2022 01:50:55 - INFO - codeparrot_training - Step 37282: {'lr': 8.172266869312892e-05, 'samples': 19088896, 'steps': 37282, 'loss/train': 1.745500922203064}
+02/26/2022 01:51:01 - INFO - codeparrot_training - Step 37283: {'lr': 8.171056830965911e-05, 'samples': 19089408, 'steps': 37283, 'loss/train': 2.3627512454986572}
+02/26/2022 01:51:04 - INFO - codeparrot_training - Step 37284: {'lr': 8.169846864708783e-05, 'samples': 19089920, 'steps': 37284, 'loss/train': 0.6568889021873474}
+02/26/2022 01:51:10 - INFO - codeparrot_training - Step 37285: {'lr': 8.168636970546675e-05, 'samples': 19090432, 'steps': 37285, 'loss/train': 1.143664002418518}
+02/26/2022 01:51:13 - INFO - codeparrot_training - Step 37286: {'lr': 8.167427148484791e-05, 'samples': 19090944, 'steps': 37286, 'loss/train': 1.9979159832000732}
+02/26/2022 01:51:19 - INFO - codeparrot_training - Step 37287: {'lr': 8.166217398528295e-05, 'samples': 19091456, 'steps': 37287, 'loss/train': 1.9059408903121948}
+02/26/2022 01:51:22 - INFO - codeparrot_training - Step 37288: {'lr': 8.165007720682396e-05, 'samples': 19091968, 'steps': 37288, 'loss/train': 0.2532229721546173}
+02/26/2022 01:51:28 - INFO - codeparrot_training - Step 37289: {'lr': 8.163798114952242e-05, 'samples': 19092480, 'steps': 37289, 'loss/train': 2.5167734622955322}
+02/26/2022 01:51:31 - INFO - codeparrot_training - Step 37290: {'lr': 8.162588581343036e-05, 'samples': 19092992, 'steps': 37290, 'loss/train': 1.857791543006897}
+02/26/2022 01:51:37 - INFO - codeparrot_training - Step 37291: {'lr': 8.16137911985995e-05, 'samples': 19093504, 'steps': 37291, 'loss/train': 0.4462227523326874}
+02/26/2022 01:51:40 - INFO - codeparrot_training - Step 37292: {'lr': 8.160169730508182e-05, 'samples': 19094016, 'steps': 37292, 'loss/train': 1.4387286901474}
+02/26/2022 01:51:46 - INFO - codeparrot_training - Step 37293: {'lr': 8.158960413292884e-05, 'samples': 19094528, 'steps': 37293, 'loss/train': 1.322270154953003}
+02/26/2022 01:51:49 - INFO - codeparrot_training - Step 37294: {'lr': 8.157751168219262e-05, 'samples': 19095040, 'steps': 37294, 'loss/train': 3.5032103061676025}
+02/26/2022 01:51:56 - INFO - codeparrot_training - Step 37295: {'lr': 8.156541995292477e-05, 'samples': 19095552, 'steps': 37295, 'loss/train': 1.9252204895019531}
+02/26/2022 01:52:00 - INFO - codeparrot_training - Step 37296: {'lr': 8.155332894517734e-05, 'samples': 19096064, 'steps': 37296, 'loss/train': 1.2596262693405151}
+02/26/2022 01:52:05 - INFO - codeparrot_training - Step 37297: {'lr': 8.154123865900179e-05, 'samples': 19096576, 'steps': 37297, 'loss/train': 1.458268404006958}
+02/26/2022 01:52:09 - INFO - codeparrot_training - Step 37298: {'lr': 8.152914909445016e-05, 'samples': 19097088, 'steps': 37298, 'loss/train': 2.333829402923584}
+02/26/2022 01:52:14 - INFO - codeparrot_training - Step 37299: {'lr': 8.15170602515741e-05, 'samples': 19097600, 'steps': 37299, 'loss/train': 2.772134304046631}
+02/26/2022 01:52:18 - INFO - codeparrot_training - Step 37300: {'lr': 8.150497213042551e-05, 'samples': 19098112, 'steps': 37300, 'loss/train': 2.3670170307159424}
+02/26/2022 01:52:23 - INFO - codeparrot_training - Step 37301: {'lr': 8.149288473105616e-05, 'samples': 19098624, 'steps': 37301, 'loss/train': 2.0371432304382324}
+02/26/2022 01:52:27 - INFO - codeparrot_training - Step 37302: {'lr': 8.148079805351775e-05, 'samples': 19099136, 'steps': 37302, 'loss/train': 1.5605242252349854}
+02/26/2022 01:52:33 - INFO - codeparrot_training - Step 37303: {'lr': 8.146871209786208e-05, 'samples': 19099648, 'steps': 37303, 'loss/train': 1.6081106662750244}
+02/26/2022 01:52:36 - INFO - codeparrot_training - Step 37304: {'lr': 8.145662686414085e-05, 'samples': 19100160, 'steps': 37304, 'loss/train': 1.1677380800247192}
+02/26/2022 01:52:40 - INFO - codeparrot_training - Step 37305: {'lr': 8.144454235240603e-05, 'samples': 19100672, 'steps': 37305, 'loss/train': 2.038557291030884}
+02/26/2022 01:52:47 - INFO - codeparrot_training - Step 37306: {'lr': 8.143245856270922e-05, 'samples': 19101184, 'steps': 37306, 'loss/train': 2.5647871494293213}
+02/26/2022 01:52:50 - INFO - codeparrot_training - Step 37307: {'lr': 8.142037549510229e-05, 'samples': 19101696, 'steps': 37307, 'loss/train': 1.7239336967468262}
+02/26/2022 01:52:56 - INFO - codeparrot_training - Step 37308: {'lr': 8.14082931496368e-05, 'samples': 19102208, 'steps': 37308, 'loss/train': 2.8829071521759033}
+02/26/2022 01:52:59 - INFO - codeparrot_training - Step 37309: {'lr': 8.139621152636476e-05, 'samples': 19102720, 'steps': 37309, 'loss/train': 2.682356357574463}
+02/26/2022 01:53:05 - INFO - codeparrot_training - Step 37310: {'lr': 8.138413062533781e-05, 'samples': 19103232, 'steps': 37310, 'loss/train': 1.5462348461151123}
+02/26/2022 01:53:08 - INFO - codeparrot_training - Step 37311: {'lr': 8.13720504466077e-05, 'samples': 19103744, 'steps': 37311, 'loss/train': 2.713484048843384}
+02/26/2022 01:53:14 - INFO - codeparrot_training - Step 37312: {'lr': 8.13599709902261e-05, 'samples': 19104256, 'steps': 37312, 'loss/train': 1.5032546520233154}
+02/26/2022 01:53:17 - INFO - codeparrot_training - Step 37313: {'lr': 8.134789225624491e-05, 'samples': 19104768, 'steps': 37313, 'loss/train': 1.992012619972229}
+02/26/2022 01:53:23 - INFO - codeparrot_training - Step 37314: {'lr': 8.133581424471581e-05, 'samples': 19105280, 'steps': 37314, 'loss/train': 1.5557827949523926}
+02/26/2022 01:53:26 - INFO - codeparrot_training - Step 37315: {'lr': 8.132373695569053e-05, 'samples': 19105792, 'steps': 37315, 'loss/train': 1.7976577281951904}
+02/26/2022 01:53:33 - INFO - codeparrot_training - Step 37316: {'lr': 8.131166038922072e-05, 'samples': 19106304, 'steps': 37316, 'loss/train': 1.9578083753585815}
+02/26/2022 01:53:36 - INFO - codeparrot_training - Step 37317: {'lr': 8.129958454535824e-05, 'samples': 19106816, 'steps': 37317, 'loss/train': 1.3418536186218262}
+02/26/2022 01:53:41 - INFO - codeparrot_training - Step 37318: {'lr': 8.12875094241548e-05, 'samples': 19107328, 'steps': 37318, 'loss/train': 1.7534079551696777}
+02/26/2022 01:53:45 - INFO - codeparrot_training - Step 37319: {'lr': 8.12754350256621e-05, 'samples': 19107840, 'steps': 37319, 'loss/train': 2.507652759552002}
+02/26/2022 01:53:51 - INFO - codeparrot_training - Step 37320: {'lr': 8.126336134993176e-05, 'samples': 19108352, 'steps': 37320, 'loss/train': 0.5858209133148193}
+02/26/2022 01:53:54 - INFO - codeparrot_training - Step 37321: {'lr': 8.12512883970157e-05, 'samples': 19108864, 'steps': 37321, 'loss/train': 2.189741373062134}
+02/26/2022 01:54:00 - INFO - codeparrot_training - Step 37322: {'lr': 8.123921616696553e-05, 'samples': 19109376, 'steps': 37322, 'loss/train': 2.0077922344207764}
+02/26/2022 01:54:03 - INFO - codeparrot_training - Step 37323: {'lr': 8.122714465983297e-05, 'samples': 19109888, 'steps': 37323, 'loss/train': 1.1128926277160645}
+02/26/2022 01:54:08 - INFO - codeparrot_training - Step 37324: {'lr': 8.121507387566962e-05, 'samples': 19110400, 'steps': 37324, 'loss/train': 0.3049682676792145}
+02/26/2022 01:54:12 - INFO - codeparrot_training - Step 37325: {'lr': 8.120300381452742e-05, 'samples': 19110912, 'steps': 37325, 'loss/train': 2.042713165283203}
+02/26/2022 01:54:18 - INFO - codeparrot_training - Step 37326: {'lr': 8.119093447645787e-05, 'samples': 19111424, 'steps': 37326, 'loss/train': 1.5282267332077026}
+02/26/2022 01:54:21 - INFO - codeparrot_training - Step 37327: {'lr': 8.117886586151287e-05, 'samples': 19111936, 'steps': 37327, 'loss/train': 0.17124544084072113}
+02/26/2022 01:54:27 - INFO - codeparrot_training - Step 37328: {'lr': 8.116679796974389e-05, 'samples': 19112448, 'steps': 37328, 'loss/train': 2.7843339443206787}
+02/26/2022 01:54:31 - INFO - codeparrot_training - Step 37329: {'lr': 8.115473080120281e-05, 'samples': 19112960, 'steps': 37329, 'loss/train': 0.0338941290974617}
+02/26/2022 01:54:36 - INFO - codeparrot_training - Step 37330: {'lr': 8.114266435594117e-05, 'samples': 19113472, 'steps': 37330, 'loss/train': 1.403425931930542}
+02/26/2022 01:54:39 - INFO - codeparrot_training - Step 37331: {'lr': 8.113059863401087e-05, 'samples': 19113984, 'steps': 37331, 'loss/train': 1.2984315156936646}
+02/26/2022 01:54:45 - INFO - codeparrot_training - Step 37332: {'lr': 8.111853363546331e-05, 'samples': 19114496, 'steps': 37332, 'loss/train': 1.5057514905929565}
+02/26/2022 01:54:48 - INFO - codeparrot_training - Step 37333: {'lr': 8.110646936035044e-05, 'samples': 19115008, 'steps': 37333, 'loss/train': 1.6936734914779663}
+02/26/2022 01:54:54 - INFO - codeparrot_training - Step 37334: {'lr': 8.109440580872371e-05, 'samples': 19115520, 'steps': 37334, 'loss/train': 1.7782814502716064}
+02/26/2022 01:54:57 - INFO - codeparrot_training - Step 37335: {'lr': 8.108234298063502e-05, 'samples': 19116032, 'steps': 37335, 'loss/train': 2.1618940830230713}
+02/26/2022 01:55:03 - INFO - codeparrot_training - Step 37336: {'lr': 8.107028087613592e-05, 'samples': 19116544, 'steps': 37336, 'loss/train': 1.412340760231018}
+02/26/2022 01:55:06 - INFO - codeparrot_training - Step 37337: {'lr': 8.105821949527808e-05, 'samples': 19117056, 'steps': 37337, 'loss/train': 0.7058124542236328}
+02/26/2022 01:55:12 - INFO - codeparrot_training - Step 37338: {'lr': 8.104615883811311e-05, 'samples': 19117568, 'steps': 37338, 'loss/train': 0.9833065867424011}
+02/26/2022 01:55:15 - INFO - codeparrot_training - Step 37339: {'lr': 8.103409890469285e-05, 'samples': 19118080, 'steps': 37339, 'loss/train': 1.828080654144287}
+02/26/2022 01:55:21 - INFO - codeparrot_training - Step 37340: {'lr': 8.102203969506886e-05, 'samples': 19118592, 'steps': 37340, 'loss/train': 1.2084861993789673}
+02/26/2022 01:55:24 - INFO - codeparrot_training - Step 37341: {'lr': 8.100998120929275e-05, 'samples': 19119104, 'steps': 37341, 'loss/train': 2.0466437339782715}
+02/26/2022 01:55:31 - INFO - codeparrot_training - Step 37342: {'lr': 8.099792344741619e-05, 'samples': 19119616, 'steps': 37342, 'loss/train': 2.5775063037872314}
+02/26/2022 01:55:34 - INFO - codeparrot_training - Step 37343: {'lr': 8.09858664094909e-05, 'samples': 19120128, 'steps': 37343, 'loss/train': 1.6297249794006348}
+02/26/2022 01:55:40 - INFO - codeparrot_training - Step 37344: {'lr': 8.09738100955685e-05, 'samples': 19120640, 'steps': 37344, 'loss/train': 1.064758539199829}
+02/26/2022 01:55:43 - INFO - codeparrot_training - Step 37345: {'lr': 8.096175450570065e-05, 'samples': 19121152, 'steps': 37345, 'loss/train': 1.4299209117889404}
+02/26/2022 01:55:49 - INFO - codeparrot_training - Step 37346: {'lr': 8.094969963993887e-05, 'samples': 19121664, 'steps': 37346, 'loss/train': 2.241816282272339}
+02/26/2022 01:55:53 - INFO - codeparrot_training - Step 37347: {'lr': 8.093764549833499e-05, 'samples': 19122176, 'steps': 37347, 'loss/train': 1.838773488998413}
+02/26/2022 01:55:58 - INFO - codeparrot_training - Step 37348: {'lr': 8.092559208094055e-05, 'samples': 19122688, 'steps': 37348, 'loss/train': 2.0230047702789307}
+02/26/2022 01:56:02 - INFO - codeparrot_training - Step 37349: {'lr': 8.091353938780718e-05, 'samples': 19123200, 'steps': 37349, 'loss/train': 1.7273041009902954}
+02/26/2022 01:56:07 - INFO - codeparrot_training - Step 37350: {'lr': 8.090148741898651e-05, 'samples': 19123712, 'steps': 37350, 'loss/train': 1.973228931427002}
+02/26/2022 01:56:11 - INFO - codeparrot_training - Step 37351: {'lr': 8.088943617453009e-05, 'samples': 19124224, 'steps': 37351, 'loss/train': 1.6383157968521118}
+02/26/2022 01:56:18 - INFO - codeparrot_training - Step 37352: {'lr': 8.087738565448974e-05, 'samples': 19124736, 'steps': 37352, 'loss/train': 2.2831029891967773}
+02/26/2022 01:56:21 - INFO - codeparrot_training - Step 37353: {'lr': 8.086533585891692e-05, 'samples': 19125248, 'steps': 37353, 'loss/train': 0.05401711165904999}
+02/26/2022 01:56:27 - INFO - codeparrot_training - Step 37354: {'lr': 8.085328678786333e-05, 'samples': 19125760, 'steps': 37354, 'loss/train': 2.198316812515259}
+02/26/2022 01:56:30 - INFO - codeparrot_training - Step 37355: {'lr': 8.084123844138047e-05, 'samples': 19126272, 'steps': 37355, 'loss/train': 0.8088244795799255}
+02/26/2022 01:56:36 - INFO - codeparrot_training - Step 37356: {'lr': 8.082919081952009e-05, 'samples': 19126784, 'steps': 37356, 'loss/train': 1.7159335613250732}
+02/26/2022 01:56:39 - INFO - codeparrot_training - Step 37357: {'lr': 8.081714392233375e-05, 'samples': 19127296, 'steps': 37357, 'loss/train': 1.7257295846939087}
+02/26/2022 01:56:45 - INFO - codeparrot_training - Step 37358: {'lr': 8.080509774987305e-05, 'samples': 19127808, 'steps': 37358, 'loss/train': 1.7765673398971558}
+02/26/2022 01:56:48 - INFO - codeparrot_training - Step 37359: {'lr': 8.079305230218947e-05, 'samples': 19128320, 'steps': 37359, 'loss/train': 1.519286870956421}
+02/26/2022 01:56:53 - INFO - codeparrot_training - Step 37360: {'lr': 8.078100757933485e-05, 'samples': 19128832, 'steps': 37360, 'loss/train': 1.1347962617874146}
+02/26/2022 01:56:57 - INFO - codeparrot_training - Step 37361: {'lr': 8.076896358136055e-05, 'samples': 19129344, 'steps': 37361, 'loss/train': 1.7465343475341797}
+02/26/2022 01:57:03 - INFO - codeparrot_training - Step 37362: {'lr': 8.075692030831847e-05, 'samples': 19129856, 'steps': 37362, 'loss/train': 1.900414228439331}
+02/26/2022 01:57:07 - INFO - codeparrot_training - Step 37363: {'lr': 8.074487776025979e-05, 'samples': 19130368, 'steps': 37363, 'loss/train': 1.4611653089523315}
+02/26/2022 01:57:12 - INFO - codeparrot_training - Step 37364: {'lr': 8.073283593723644e-05, 'samples': 19130880, 'steps': 37364, 'loss/train': 2.3062353134155273}
+02/26/2022 01:57:15 - INFO - codeparrot_training - Step 37365: {'lr': 8.072079483929975e-05, 'samples': 19131392, 'steps': 37365, 'loss/train': 1.746525526046753}
+02/26/2022 01:57:21 - INFO - codeparrot_training - Step 37366: {'lr': 8.070875446650162e-05, 'samples': 19131904, 'steps': 37366, 'loss/train': 2.2053158283233643}
+02/26/2022 01:57:24 - INFO - codeparrot_training - Step 37367: {'lr': 8.069671481889323e-05, 'samples': 19132416, 'steps': 37367, 'loss/train': 1.556070327758789}
+02/26/2022 01:57:30 - INFO - codeparrot_training - Step 37368: {'lr': 8.068467589652648e-05, 'samples': 19132928, 'steps': 37368, 'loss/train': 2.6845715045928955}
+02/26/2022 01:57:33 - INFO - codeparrot_training - Step 37369: {'lr': 8.067263769945271e-05, 'samples': 19133440, 'steps': 37369, 'loss/train': 1.8082600831985474}
+02/26/2022 01:57:39 - INFO - codeparrot_training - Step 37370: {'lr': 8.066060022772376e-05, 'samples': 19133952, 'steps': 37370, 'loss/train': 1.0379939079284668}
+02/26/2022 01:57:42 - INFO - codeparrot_training - Step 37371: {'lr': 8.064856348139085e-05, 'samples': 19134464, 'steps': 37371, 'loss/train': 1.405431866645813}
+02/26/2022 01:57:49 - INFO - codeparrot_training - Step 37372: {'lr': 8.063652746050582e-05, 'samples': 19134976, 'steps': 37372, 'loss/train': 2.5331814289093018}
+02/26/2022 01:57:52 - INFO - codeparrot_training - Step 37373: {'lr': 8.062449216512003e-05, 'samples': 19135488, 'steps': 37373, 'loss/train': 2.1231932640075684}
+02/26/2022 01:57:59 - INFO - codeparrot_training - Step 37374: {'lr': 8.061245759528521e-05, 'samples': 19136000, 'steps': 37374, 'loss/train': 2.263923168182373}
+02/26/2022 01:58:02 - INFO - codeparrot_training - Step 37375: {'lr': 8.060042375105284e-05, 'samples': 19136512, 'steps': 37375, 'loss/train': 2.292685031890869}
+02/26/2022 01:58:06 - INFO - codeparrot_training - Step 37376: {'lr': 8.058839063247447e-05, 'samples': 19137024, 'steps': 37376, 'loss/train': 2.2224104404449463}
+02/26/2022 01:58:11 - INFO - codeparrot_training - Step 37377: {'lr': 8.057635823960155e-05, 'samples': 19137536, 'steps': 37377, 'loss/train': 1.2549443244934082}
+02/26/2022 01:58:15 - INFO - codeparrot_training - Step 37378: {'lr': 8.05643265724858e-05, 'samples': 19138048, 'steps': 37378, 'loss/train': 1.2091478109359741}
+02/26/2022 01:58:20 - INFO - codeparrot_training - Step 37379: {'lr': 8.055229563117869e-05, 'samples': 19138560, 'steps': 37379, 'loss/train': 1.694042444229126}
+02/26/2022 01:58:24 - INFO - codeparrot_training - Step 37380: {'lr': 8.054026541573172e-05, 'samples': 19139072, 'steps': 37380, 'loss/train': 1.5892401933670044}
+02/26/2022 01:58:29 - INFO - codeparrot_training - Step 37381: {'lr': 8.052823592619637e-05, 'samples': 19139584, 'steps': 37381, 'loss/train': 2.951171398162842}
+02/26/2022 01:58:33 - INFO - codeparrot_training - Step 37382: {'lr': 8.051620716262433e-05, 'samples': 19140096, 'steps': 37382, 'loss/train': 1.8875781297683716}
+02/26/2022 01:58:38 - INFO - codeparrot_training - Step 37383: {'lr': 8.050417912506702e-05, 'samples': 19140608, 'steps': 37383, 'loss/train': 1.9337056875228882}
+02/26/2022 01:58:42 - INFO - codeparrot_training - Step 37384: {'lr': 8.049215181357602e-05, 'samples': 19141120, 'steps': 37384, 'loss/train': 2.507456064224243}
+02/26/2022 01:58:47 - INFO - codeparrot_training - Step 37385: {'lr': 8.048012522820273e-05, 'samples': 19141632, 'steps': 37385, 'loss/train': 0.5179448127746582}
+02/26/2022 01:58:51 - INFO - codeparrot_training - Step 37386: {'lr': 8.046809936899879e-05, 'samples': 19142144, 'steps': 37386, 'loss/train': 1.0809290409088135}
+02/26/2022 01:58:56 - INFO - codeparrot_training - Step 37387: {'lr': 8.045607423601572e-05, 'samples': 19142656, 'steps': 37387, 'loss/train': 1.7661043405532837}
+02/26/2022 01:59:00 - INFO - codeparrot_training - Step 37388: {'lr': 8.044404982930498e-05, 'samples': 19143168, 'steps': 37388, 'loss/train': 0.7763864994049072}
+02/26/2022 01:59:05 - INFO - codeparrot_training - Step 37389: {'lr': 8.043202614891801e-05, 'samples': 19143680, 'steps': 37389, 'loss/train': 0.527213454246521}
+02/26/2022 01:59:09 - INFO - codeparrot_training - Step 37390: {'lr': 8.04200031949065e-05, 'samples': 19144192, 'steps': 37390, 'loss/train': 1.1823621988296509}
+02/26/2022 01:59:15 - INFO - codeparrot_training - Step 37391: {'lr': 8.040798096732182e-05, 'samples': 19144704, 'steps': 37391, 'loss/train': 1.87386155128479}
+02/26/2022 01:59:19 - INFO - codeparrot_training - Step 37392: {'lr': 8.039595946621551e-05, 'samples': 19145216, 'steps': 37392, 'loss/train': 0.8751384019851685}
+02/26/2022 01:59:24 - INFO - codeparrot_training - Step 37393: {'lr': 8.038393869163896e-05, 'samples': 19145728, 'steps': 37393, 'loss/train': 1.0981802940368652}
+02/26/2022 01:59:27 - INFO - codeparrot_training - Step 37394: {'lr': 8.037191864364385e-05, 'samples': 19146240, 'steps': 37394, 'loss/train': 1.4065470695495605}
+02/26/2022 01:59:33 - INFO - codeparrot_training - Step 37395: {'lr': 8.035989932228158e-05, 'samples': 19146752, 'steps': 37395, 'loss/train': 1.101318597793579}
+02/26/2022 01:59:37 - INFO - codeparrot_training - Step 37396: {'lr': 8.034788072760354e-05, 'samples': 19147264, 'steps': 37396, 'loss/train': 0.996615469455719}
+02/26/2022 01:59:42 - INFO - codeparrot_training - Step 37397: {'lr': 8.033586285966148e-05, 'samples': 19147776, 'steps': 37397, 'loss/train': 0.8456780910491943}
+02/26/2022 01:59:45 - INFO - codeparrot_training - Step 37398: {'lr': 8.032384571850656e-05, 'samples': 19148288, 'steps': 37398, 'loss/train': 1.9835975170135498}
+02/26/2022 01:59:52 - INFO - codeparrot_training - Step 37399: {'lr': 8.031182930419048e-05, 'samples': 19148800, 'steps': 37399, 'loss/train': 2.304450035095215}
+02/26/2022 01:59:55 - INFO - codeparrot_training - Step 37400: {'lr': 8.029981361676455e-05, 'samples': 19149312, 'steps': 37400, 'loss/train': 1.8341145515441895}
+02/26/2022 02:00:01 - INFO - codeparrot_training - Step 37401: {'lr': 8.02877986562805e-05, 'samples': 19149824, 'steps': 37401, 'loss/train': 1.2928545475006104}
+02/26/2022 02:00:04 - INFO - codeparrot_training - Step 37402: {'lr': 8.027578442278946e-05, 'samples': 19150336, 'steps': 37402, 'loss/train': 1.7776765823364258}
+02/26/2022 02:00:10 - INFO - codeparrot_training - Step 37403: {'lr': 8.026377091634316e-05, 'samples': 19150848, 'steps': 37403, 'loss/train': 1.263899803161621}
+02/26/2022 02:00:13 - INFO - codeparrot_training - Step 37404: {'lr': 8.025175813699287e-05, 'samples': 19151360, 'steps': 37404, 'loss/train': 1.4260401725769043}
+02/26/2022 02:00:19 - INFO - codeparrot_training - Step 37405: {'lr': 8.023974608479032e-05, 'samples': 19151872, 'steps': 37405, 'loss/train': 1.4067226648330688}
+02/26/2022 02:00:22 - INFO - codeparrot_training - Step 37406: {'lr': 8.022773475978665e-05, 'samples': 19152384, 'steps': 37406, 'loss/train': 2.385246992111206}
+02/26/2022 02:00:28 - INFO - codeparrot_training - Step 37407: {'lr': 8.021572416203351e-05, 'samples': 19152896, 'steps': 37407, 'loss/train': 3.092381715774536}
+02/26/2022 02:00:31 - INFO - codeparrot_training - Step 37408: {'lr': 8.020371429158221e-05, 'samples': 19153408, 'steps': 37408, 'loss/train': 2.0908305644989014}
+02/26/2022 02:00:37 - INFO - codeparrot_training - Step 37409: {'lr': 8.01917051484844e-05, 'samples': 19153920, 'steps': 37409, 'loss/train': 1.589678168296814}
+02/26/2022 02:00:41 - INFO - codeparrot_training - Step 37410: {'lr': 8.017969673279138e-05, 'samples': 19154432, 'steps': 37410, 'loss/train': 1.7165073156356812}
+02/26/2022 02:00:46 - INFO - codeparrot_training - Step 37411: {'lr': 8.01676890445546e-05, 'samples': 19154944, 'steps': 37411, 'loss/train': 2.5316507816314697}
+02/26/2022 02:00:50 - INFO - codeparrot_training - Step 37412: {'lr': 8.015568208382545e-05, 'samples': 19155456, 'steps': 37412, 'loss/train': 1.8428373336791992}
+02/26/2022 02:00:55 - INFO - codeparrot_training - Step 37413: {'lr': 8.014367585065549e-05, 'samples': 19155968, 'steps': 37413, 'loss/train': 2.3472867012023926}
+02/26/2022 02:00:59 - INFO - codeparrot_training - Step 37414: {'lr': 8.01316703450961e-05, 'samples': 19156480, 'steps': 37414, 'loss/train': 1.9508153200149536}
+02/26/2022 02:01:04 - INFO - codeparrot_training - Step 37415: {'lr': 8.011966556719871e-05, 'samples': 19156992, 'steps': 37415, 'loss/train': 1.6075329780578613}
+02/26/2022 02:01:08 - INFO - codeparrot_training - Step 37416: {'lr': 8.010766151701463e-05, 'samples': 19157504, 'steps': 37416, 'loss/train': 2.5764198303222656}
+02/26/2022 02:01:13 - INFO - codeparrot_training - Step 37417: {'lr': 8.009565819459546e-05, 'samples': 19158016, 'steps': 37417, 'loss/train': 2.0447890758514404}
+02/26/2022 02:01:17 - INFO - codeparrot_training - Step 37418: {'lr': 8.008365559999256e-05, 'samples': 19158528, 'steps': 37418, 'loss/train': 1.3697654008865356}
+02/26/2022 02:01:22 - INFO - codeparrot_training - Step 37419: {'lr': 8.00716537332573e-05, 'samples': 19159040, 'steps': 37419, 'loss/train': 1.7474851608276367}
+02/26/2022 02:01:26 - INFO - codeparrot_training - Step 37420: {'lr': 8.005965259444107e-05, 'samples': 19159552, 'steps': 37420, 'loss/train': 2.0200257301330566}
+02/26/2022 02:01:32 - INFO - codeparrot_training - Step 37421: {'lr': 8.004765218359537e-05, 'samples': 19160064, 'steps': 37421, 'loss/train': 1.8225852251052856}
+02/26/2022 02:01:36 - INFO - codeparrot_training - Step 37422: {'lr': 8.003565250077149e-05, 'samples': 19160576, 'steps': 37422, 'loss/train': 1.8792427778244019}
+02/26/2022 02:01:41 - INFO - codeparrot_training - Step 37423: {'lr': 8.00236535460211e-05, 'samples': 19161088, 'steps': 37423, 'loss/train': 1.1113975048065186}
+02/26/2022 02:01:44 - INFO - codeparrot_training - Step 37424: {'lr': 8.001165531939519e-05, 'samples': 19161600, 'steps': 37424, 'loss/train': 1.3480619192123413}
+02/26/2022 02:01:50 - INFO - codeparrot_training - Step 37425: {'lr': 7.999965782094548e-05, 'samples': 19162112, 'steps': 37425, 'loss/train': 2.600320339202881}
+02/26/2022 02:01:53 - INFO - codeparrot_training - Step 37426: {'lr': 7.998766105072316e-05, 'samples': 19162624, 'steps': 37426, 'loss/train': 2.0850894451141357}
+02/26/2022 02:01:59 - INFO - codeparrot_training - Step 37427: {'lr': 7.99756650087799e-05, 'samples': 19163136, 'steps': 37427, 'loss/train': 2.140472650527954}
+02/26/2022 02:02:02 - INFO - codeparrot_training - Step 37428: {'lr': 7.996366969516672e-05, 'samples': 19163648, 'steps': 37428, 'loss/train': 1.5940027236938477}
+02/26/2022 02:02:08 - INFO - codeparrot_training - Step 37429: {'lr': 7.995167510993529e-05, 'samples': 19164160, 'steps': 37429, 'loss/train': 1.5432043075561523}
+02/26/2022 02:02:11 - INFO - codeparrot_training - Step 37430: {'lr': 7.993968125313686e-05, 'samples': 19164672, 'steps': 37430, 'loss/train': 1.2817788124084473}
+02/26/2022 02:02:18 - INFO - codeparrot_training - Step 37431: {'lr': 7.992768812482284e-05, 'samples': 19165184, 'steps': 37431, 'loss/train': 1.318253755569458}
+02/26/2022 02:02:21 - INFO - codeparrot_training - Step 37432: {'lr': 7.991569572504453e-05, 'samples': 19165696, 'steps': 37432, 'loss/train': 1.844303011894226}
+02/26/2022 02:02:27 - INFO - codeparrot_training - Step 37433: {'lr': 7.990370405385348e-05, 'samples': 19166208, 'steps': 37433, 'loss/train': 1.5911951065063477}
+02/26/2022 02:02:30 - INFO - codeparrot_training - Step 37434: {'lr': 7.989171311130092e-05, 'samples': 19166720, 'steps': 37434, 'loss/train': 1.4388668537139893}
+02/26/2022 02:02:36 - INFO - codeparrot_training - Step 37435: {'lr': 7.987972289743814e-05, 'samples': 19167232, 'steps': 37435, 'loss/train': 0.15204450488090515}
+02/26/2022 02:02:39 - INFO - codeparrot_training - Step 37436: {'lr': 7.986773341231673e-05, 'samples': 19167744, 'steps': 37436, 'loss/train': 1.7728265523910522}
+02/26/2022 02:02:45 - INFO - codeparrot_training - Step 37437: {'lr': 7.985574465598791e-05, 'samples': 19168256, 'steps': 37437, 'loss/train': 1.8559874296188354}
+02/26/2022 02:02:48 - INFO - codeparrot_training - Step 37438: {'lr': 7.984375662850304e-05, 'samples': 19168768, 'steps': 37438, 'loss/train': 1.6557707786560059}
+02/26/2022 02:02:54 - INFO - codeparrot_training - Step 37439: {'lr': 7.983176932991343e-05, 'samples': 19169280, 'steps': 37439, 'loss/train': 2.3579235076904297}
+02/26/2022 02:02:57 - INFO - codeparrot_training - Step 37440: {'lr': 7.981978276027054e-05, 'samples': 19169792, 'steps': 37440, 'loss/train': 2.2395734786987305}
+02/26/2022 02:03:03 - INFO - codeparrot_training - Step 37441: {'lr': 7.98077969196257e-05, 'samples': 19170304, 'steps': 37441, 'loss/train': 2.1372907161712646}
+02/26/2022 02:03:06 - INFO - codeparrot_training - Step 37442: {'lr': 7.97958118080302e-05, 'samples': 19170816, 'steps': 37442, 'loss/train': 1.090391993522644}
+02/26/2022 02:03:12 - INFO - codeparrot_training - Step 37443: {'lr': 7.97838274255353e-05, 'samples': 19171328, 'steps': 37443, 'loss/train': 1.7838338613510132}
+02/26/2022 02:03:15 - INFO - codeparrot_training - Step 37444: {'lr': 7.977184377219252e-05, 'samples': 19171840, 'steps': 37444, 'loss/train': 2.1696956157684326}
+02/26/2022 02:03:20 - INFO - codeparrot_training - Step 37445: {'lr': 7.975986084805311e-05, 'samples': 19172352, 'steps': 37445, 'loss/train': 3.2940948009490967}
+02/26/2022 02:03:24 - INFO - codeparrot_training - Step 37446: {'lr': 7.97478786531684e-05, 'samples': 19172864, 'steps': 37446, 'loss/train': 1.753911018371582}
+02/26/2022 02:03:30 - INFO - codeparrot_training - Step 37447: {'lr': 7.973589718758964e-05, 'samples': 19173376, 'steps': 37447, 'loss/train': 2.3564867973327637}
+02/26/2022 02:03:33 - INFO - codeparrot_training - Step 37448: {'lr': 7.972391645136831e-05, 'samples': 19173888, 'steps': 37448, 'loss/train': 1.500327467918396}
+02/26/2022 02:03:39 - INFO - codeparrot_training - Step 37449: {'lr': 7.971193644455566e-05, 'samples': 19174400, 'steps': 37449, 'loss/train': 1.5810664892196655}
+02/26/2022 02:03:42 - INFO - codeparrot_training - Step 37450: {'lr': 7.9699957167203e-05, 'samples': 19174912, 'steps': 37450, 'loss/train': 1.7077873945236206}
+02/26/2022 02:03:48 - INFO - codeparrot_training - Step 37451: {'lr': 7.968797861936153e-05, 'samples': 19175424, 'steps': 37451, 'loss/train': 1.5553562641143799}
+02/26/2022 02:03:51 - INFO - codeparrot_training - Step 37452: {'lr': 7.96760008010828e-05, 'samples': 19175936, 'steps': 37452, 'loss/train': 1.8128317594528198}
+02/26/2022 02:03:57 - INFO - codeparrot_training - Step 37453: {'lr': 7.966402371241799e-05, 'samples': 19176448, 'steps': 37453, 'loss/train': 2.0748281478881836}
+02/26/2022 02:04:00 - INFO - codeparrot_training - Step 37454: {'lr': 7.965204735341842e-05, 'samples': 19176960, 'steps': 37454, 'loss/train': 1.9976259469985962}
+02/26/2022 02:04:06 - INFO - codeparrot_training - Step 37455: {'lr': 7.964007172413529e-05, 'samples': 19177472, 'steps': 37455, 'loss/train': 1.3991798162460327}
+02/26/2022 02:04:09 - INFO - codeparrot_training - Step 37456: {'lr': 7.962809682462008e-05, 'samples': 19177984, 'steps': 37456, 'loss/train': 2.845388889312744}
+02/26/2022 02:04:16 - INFO - codeparrot_training - Step 37457: {'lr': 7.961612265492393e-05, 'samples': 19178496, 'steps': 37457, 'loss/train': 1.387342095375061}
+02/26/2022 02:04:19 - INFO - codeparrot_training - Step 37458: {'lr': 7.960414921509837e-05, 'samples': 19179008, 'steps': 37458, 'loss/train': 2.001033067703247}
+02/26/2022 02:04:25 - INFO - codeparrot_training - Step 37459: {'lr': 7.959217650519438e-05, 'samples': 19179520, 'steps': 37459, 'loss/train': 2.0248870849609375}
+02/26/2022 02:04:28 - INFO - codeparrot_training - Step 37460: {'lr': 7.958020452526346e-05, 'samples': 19180032, 'steps': 37460, 'loss/train': 2.51249098777771}
+02/26/2022 02:04:34 - INFO - codeparrot_training - Step 37461: {'lr': 7.956823327535675e-05, 'samples': 19180544, 'steps': 37461, 'loss/train': 2.244447946548462}
+02/26/2022 02:04:37 - INFO - codeparrot_training - Step 37462: {'lr': 7.955626275552577e-05, 'samples': 19181056, 'steps': 37462, 'loss/train': 2.141691207885742}
+02/26/2022 02:04:43 - INFO - codeparrot_training - Step 37463: {'lr': 7.954429296582149e-05, 'samples': 19181568, 'steps': 37463, 'loss/train': 1.6009907722473145}
+02/26/2022 02:04:46 - INFO - codeparrot_training - Step 37464: {'lr': 7.953232390629539e-05, 'samples': 19182080, 'steps': 37464, 'loss/train': 1.098544716835022}
+02/26/2022 02:04:52 - INFO - codeparrot_training - Step 37465: {'lr': 7.952035557699863e-05, 'samples': 19182592, 'steps': 37465, 'loss/train': 1.5164607763290405}
+02/26/2022 02:04:55 - INFO - codeparrot_training - Step 37466: {'lr': 7.950838797798268e-05, 'samples': 19183104, 'steps': 37466, 'loss/train': 1.4035630226135254}
+02/26/2022 02:05:02 - INFO - codeparrot_training - Step 37467: {'lr': 7.94964211092985e-05, 'samples': 19183616, 'steps': 37467, 'loss/train': 1.9000409841537476}
+02/26/2022 02:05:05 - INFO - codeparrot_training - Step 37468: {'lr': 7.94844549709976e-05, 'samples': 19184128, 'steps': 37468, 'loss/train': 0.5668824315071106}
+02/26/2022 02:05:11 - INFO - codeparrot_training - Step 37469: {'lr': 7.947248956313106e-05, 'samples': 19184640, 'steps': 37469, 'loss/train': 1.2484667301177979}
+02/26/2022 02:05:14 - INFO - codeparrot_training - Step 37470: {'lr': 7.946052488575032e-05, 'samples': 19185152, 'steps': 37470, 'loss/train': 1.3246175050735474}
+02/26/2022 02:05:20 - INFO - codeparrot_training - Step 37471: {'lr': 7.94485609389065e-05, 'samples': 19185664, 'steps': 37471, 'loss/train': 0.7225742936134338}
+02/26/2022 02:05:24 - INFO - codeparrot_training - Step 37472: {'lr': 7.94365977226509e-05, 'samples': 19186176, 'steps': 37472, 'loss/train': 0.3974103331565857}
+02/26/2022 02:05:29 - INFO - codeparrot_training - Step 37473: {'lr': 7.942463523703472e-05, 'samples': 19186688, 'steps': 37473, 'loss/train': 1.729839563369751}
+02/26/2022 02:05:33 - INFO - codeparrot_training - Step 37474: {'lr': 7.941267348210929e-05, 'samples': 19187200, 'steps': 37474, 'loss/train': 1.6778931617736816}
+02/26/2022 02:05:38 - INFO - codeparrot_training - Step 37475: {'lr': 7.940071245792579e-05, 'samples': 19187712, 'steps': 37475, 'loss/train': 1.685838222503662}
+02/26/2022 02:05:44 - INFO - codeparrot_training - Step 37476: {'lr': 7.938875216453548e-05, 'samples': 19188224, 'steps': 37476, 'loss/train': 1.5195034742355347}
+02/26/2022 02:05:47 - INFO - codeparrot_training - Step 37477: {'lr': 7.937679260198949e-05, 'samples': 19188736, 'steps': 37477, 'loss/train': 1.64725923538208}
+02/26/2022 02:05:51 - INFO - codeparrot_training - Step 37478: {'lr': 7.936483377033924e-05, 'samples': 19189248, 'steps': 37478, 'loss/train': 1.3167613744735718}
+02/26/2022 02:05:56 - INFO - codeparrot_training - Step 37479: {'lr': 7.935287566963584e-05, 'samples': 19189760, 'steps': 37479, 'loss/train': 1.630751132965088}
+02/26/2022 02:06:02 - INFO - codeparrot_training - Step 37480: {'lr': 7.934091829993053e-05, 'samples': 19190272, 'steps': 37480, 'loss/train': 1.701886534690857}
+02/26/2022 02:06:06 - INFO - codeparrot_training - Step 37481: {'lr': 7.932896166127456e-05, 'samples': 19190784, 'steps': 37481, 'loss/train': 1.539083480834961}
+02/26/2022 02:06:09 - INFO - codeparrot_training - Step 37482: {'lr': 7.931700575371903e-05, 'samples': 19191296, 'steps': 37482, 'loss/train': 1.2061444520950317}
+02/26/2022 02:06:15 - INFO - codeparrot_training - Step 37483: {'lr': 7.930505057731533e-05, 'samples': 19191808, 'steps': 37483, 'loss/train': 1.1710108518600464}
+02/26/2022 02:06:20 - INFO - codeparrot_training - Step 37484: {'lr': 7.929309613211457e-05, 'samples': 19192320, 'steps': 37484, 'loss/train': 2.493664503097534}
+02/26/2022 02:06:24 - INFO - codeparrot_training - Step 37485: {'lr': 7.928114241816803e-05, 'samples': 19192832, 'steps': 37485, 'loss/train': 1.649251103401184}
+02/26/2022 02:06:29 - INFO - codeparrot_training - Step 37486: {'lr': 7.926918943552674e-05, 'samples': 19193344, 'steps': 37486, 'loss/train': 2.026777744293213}
+02/26/2022 02:06:33 - INFO - codeparrot_training - Step 37487: {'lr': 7.925723718424213e-05, 'samples': 19193856, 'steps': 37487, 'loss/train': 1.492714762687683}
+02/26/2022 02:06:38 - INFO - codeparrot_training - Step 37488: {'lr': 7.924528566436531e-05, 'samples': 19194368, 'steps': 37488, 'loss/train': 1.8701484203338623}
+02/26/2022 02:06:42 - INFO - codeparrot_training - Step 37489: {'lr': 7.923333487594744e-05, 'samples': 19194880, 'steps': 37489, 'loss/train': 1.11849844455719}
+02/26/2022 02:06:47 - INFO - codeparrot_training - Step 37490: {'lr': 7.922138481903968e-05, 'samples': 19195392, 'steps': 37490, 'loss/train': 2.541264057159424}
+02/26/2022 02:06:51 - INFO - codeparrot_training - Step 37491: {'lr': 7.920943549369336e-05, 'samples': 19195904, 'steps': 37491, 'loss/train': 1.1719616651535034}
+02/26/2022 02:06:57 - INFO - codeparrot_training - Step 37492: {'lr': 7.919748689995949e-05, 'samples': 19196416, 'steps': 37492, 'loss/train': 1.9301249980926514}
+02/26/2022 02:07:00 - INFO - codeparrot_training - Step 37493: {'lr': 7.918553903788953e-05, 'samples': 19196928, 'steps': 37493, 'loss/train': 2.1314516067504883}
+02/26/2022 02:07:06 - INFO - codeparrot_training - Step 37494: {'lr': 7.917359190753428e-05, 'samples': 19197440, 'steps': 37494, 'loss/train': 2.2338268756866455}
+02/26/2022 02:07:09 - INFO - codeparrot_training - Step 37495: {'lr': 7.916164550894526e-05, 'samples': 19197952, 'steps': 37495, 'loss/train': 1.6606563329696655}
+02/26/2022 02:07:15 - INFO - codeparrot_training - Step 37496: {'lr': 7.914969984217337e-05, 'samples': 19198464, 'steps': 37496, 'loss/train': 0.9926959276199341}
+02/26/2022 02:07:18 - INFO - codeparrot_training - Step 37497: {'lr': 7.91377549072701e-05, 'samples': 19198976, 'steps': 37497, 'loss/train': 1.6473690271377563}
+02/26/2022 02:07:24 - INFO - codeparrot_training - Step 37498: {'lr': 7.912581070428627e-05, 'samples': 19199488, 'steps': 37498, 'loss/train': 0.8211421370506287}
+02/26/2022 02:07:27 - INFO - codeparrot_training - Step 37499: {'lr': 7.91138672332733e-05, 'samples': 19200000, 'steps': 37499, 'loss/train': 1.2416177988052368}
+02/26/2022 02:07:33 - INFO - codeparrot_training - Step 37500: {'lr': 7.910192449428217e-05, 'samples': 19200512, 'steps': 37500, 'loss/train': 1.3083006143569946}
+02/26/2022 02:07:36 - INFO - codeparrot_training - Step 37501: {'lr': 7.908998248736427e-05, 'samples': 19201024, 'steps': 37501, 'loss/train': 2.146533489227295}
+02/26/2022 02:07:43 - INFO - codeparrot_training - Step 37502: {'lr': 7.907804121257048e-05, 'samples': 19201536, 'steps': 37502, 'loss/train': 1.65902841091156}
+02/26/2022 02:07:46 - INFO - codeparrot_training - Step 37503: {'lr': 7.906610066995216e-05, 'samples': 19202048, 'steps': 37503, 'loss/train': 1.2601209878921509}
+02/26/2022 02:07:52 - INFO - codeparrot_training - Step 37504: {'lr': 7.90541608595603e-05, 'samples': 19202560, 'steps': 37504, 'loss/train': 0.16535258293151855}
+02/26/2022 02:07:55 - INFO - codeparrot_training - Step 37505: {'lr': 7.904222178144621e-05, 'samples': 19203072, 'steps': 37505, 'loss/train': 1.8236409425735474}
+02/26/2022 02:08:01 - INFO - codeparrot_training - Step 37506: {'lr': 7.903028343566096e-05, 'samples': 19203584, 'steps': 37506, 'loss/train': 1.45333731174469}
+02/26/2022 02:08:04 - INFO - codeparrot_training - Step 37507: {'lr': 7.90183458222557e-05, 'samples': 19204096, 'steps': 37507, 'loss/train': 1.0002244710922241}
+02/26/2022 02:08:10 - INFO - codeparrot_training - Step 37508: {'lr': 7.900640894128147e-05, 'samples': 19204608, 'steps': 37508, 'loss/train': 1.2659326791763306}
+02/26/2022 02:08:13 - INFO - codeparrot_training - Step 37509: {'lr': 7.899447279278956e-05, 'samples': 19205120, 'steps': 37509, 'loss/train': 0.9429946541786194}
+02/26/2022 02:08:19 - INFO - codeparrot_training - Step 37510: {'lr': 7.898253737683103e-05, 'samples': 19205632, 'steps': 37510, 'loss/train': 1.514399766921997}
+02/26/2022 02:08:22 - INFO - codeparrot_training - Step 37511: {'lr': 7.897060269345702e-05, 'samples': 19206144, 'steps': 37511, 'loss/train': 0.5376790761947632}
+02/26/2022 02:08:29 - INFO - codeparrot_training - Step 37512: {'lr': 7.895866874271854e-05, 'samples': 19206656, 'steps': 37512, 'loss/train': 2.072449207305908}
+02/26/2022 02:08:32 - INFO - codeparrot_training - Step 37513: {'lr': 7.894673552466689e-05, 'samples': 19207168, 'steps': 37513, 'loss/train': 1.615744709968567}
+02/26/2022 02:08:38 - INFO - codeparrot_training - Step 37514: {'lr': 7.893480303935313e-05, 'samples': 19207680, 'steps': 37514, 'loss/train': 1.763232946395874}
+02/26/2022 02:08:41 - INFO - codeparrot_training - Step 37515: {'lr': 7.892287128682834e-05, 'samples': 19208192, 'steps': 37515, 'loss/train': 1.7276393175125122}
+02/26/2022 02:08:47 - INFO - codeparrot_training - Step 37516: {'lr': 7.891094026714357e-05, 'samples': 19208704, 'steps': 37516, 'loss/train': 0.6663673520088196}
+02/26/2022 02:08:50 - INFO - codeparrot_training - Step 37517: {'lr': 7.889900998035007e-05, 'samples': 19209216, 'steps': 37517, 'loss/train': 2.6416876316070557}
+02/26/2022 02:08:56 - INFO - codeparrot_training - Step 37518: {'lr': 7.888708042649892e-05, 'samples': 19209728, 'steps': 37518, 'loss/train': 1.9058363437652588}
+02/26/2022 02:08:59 - INFO - codeparrot_training - Step 37519: {'lr': 7.887515160564116e-05, 'samples': 19210240, 'steps': 37519, 'loss/train': 1.670300006866455}
+02/26/2022 02:09:05 - INFO - codeparrot_training - Step 37520: {'lr': 7.886322351782782e-05, 'samples': 19210752, 'steps': 37520, 'loss/train': 0.38533204793930054}
+02/26/2022 02:09:08 - INFO - codeparrot_training - Step 37521: {'lr': 7.885129616311018e-05, 'samples': 19211264, 'steps': 37521, 'loss/train': 0.9540395736694336}
+02/26/2022 02:09:14 - INFO - codeparrot_training - Step 37522: {'lr': 7.883936954153922e-05, 'samples': 19211776, 'steps': 37522, 'loss/train': 1.4791840314865112}
+02/26/2022 02:09:17 - INFO - codeparrot_training - Step 37523: {'lr': 7.882744365316608e-05, 'samples': 19212288, 'steps': 37523, 'loss/train': 1.3955113887786865}
+02/26/2022 02:09:23 - INFO - codeparrot_training - Step 37524: {'lr': 7.881551849804175e-05, 'samples': 19212800, 'steps': 37524, 'loss/train': 2.339679718017578}
+02/26/2022 02:09:26 - INFO - codeparrot_training - Step 37525: {'lr': 7.880359407621743e-05, 'samples': 19213312, 'steps': 37525, 'loss/train': 1.4471436738967896}
+02/26/2022 02:09:32 - INFO - codeparrot_training - Step 37526: {'lr': 7.879167038774415e-05, 'samples': 19213824, 'steps': 37526, 'loss/train': 0.8532547950744629}
+02/26/2022 02:09:35 - INFO - codeparrot_training - Step 37527: {'lr': 7.877974743267302e-05, 'samples': 19214336, 'steps': 37527, 'loss/train': 1.6491354703903198}
+02/26/2022 02:09:41 - INFO - codeparrot_training - Step 37528: {'lr': 7.876782521105505e-05, 'samples': 19214848, 'steps': 37528, 'loss/train': 1.108742594718933}
+02/26/2022 02:09:45 - INFO - codeparrot_training - Step 37529: {'lr': 7.875590372294128e-05, 'samples': 19215360, 'steps': 37529, 'loss/train': 3.012669801712036}
+02/26/2022 02:09:50 - INFO - codeparrot_training - Step 37530: {'lr': 7.874398296838292e-05, 'samples': 19215872, 'steps': 37530, 'loss/train': 1.7740246057510376}
+02/26/2022 02:09:54 - INFO - codeparrot_training - Step 37531: {'lr': 7.873206294743088e-05, 'samples': 19216384, 'steps': 37531, 'loss/train': 1.08603036403656}
+02/26/2022 02:10:00 - INFO - codeparrot_training - Step 37532: {'lr': 7.872014366013647e-05, 'samples': 19216896, 'steps': 37532, 'loss/train': 1.7007575035095215}
+02/26/2022 02:10:03 - INFO - codeparrot_training - Step 37533: {'lr': 7.87082251065504e-05, 'samples': 19217408, 'steps': 37533, 'loss/train': 0.8413349390029907}
+02/26/2022 02:10:09 - INFO - codeparrot_training - Step 37534: {'lr': 7.8696307286724e-05, 'samples': 19217920, 'steps': 37534, 'loss/train': 2.635847330093384}
+02/26/2022 02:10:12 - INFO - codeparrot_training - Step 37535: {'lr': 7.868439020070811e-05, 'samples': 19218432, 'steps': 37535, 'loss/train': 2.3311822414398193}
+02/26/2022 02:10:18 - INFO - codeparrot_training - Step 37536: {'lr': 7.867247384855409e-05, 'samples': 19218944, 'steps': 37536, 'loss/train': 1.5154963731765747}
+02/26/2022 02:10:21 - INFO - codeparrot_training - Step 37537: {'lr': 7.866055823031262e-05, 'samples': 19219456, 'steps': 37537, 'loss/train': 1.4162359237670898}
+02/26/2022 02:10:27 - INFO - codeparrot_training - Step 37538: {'lr': 7.864864334603497e-05, 'samples': 19219968, 'steps': 37538, 'loss/train': 0.8373293876647949}
+02/26/2022 02:10:31 - INFO - codeparrot_training - Step 37539: {'lr': 7.863672919577208e-05, 'samples': 19220480, 'steps': 37539, 'loss/train': 2.602374792098999}
+02/26/2022 02:10:36 - INFO - codeparrot_training - Step 37540: {'lr': 7.862481577957517e-05, 'samples': 19220992, 'steps': 37540, 'loss/train': 1.1344972848892212}
+02/26/2022 02:10:40 - INFO - codeparrot_training - Step 37541: {'lr': 7.861290309749494e-05, 'samples': 19221504, 'steps': 37541, 'loss/train': 2.5414211750030518}
+02/26/2022 02:10:45 - INFO - codeparrot_training - Step 37542: {'lr': 7.860099114958275e-05, 'samples': 19222016, 'steps': 37542, 'loss/train': 2.643895387649536}
+02/26/2022 02:10:49 - INFO - codeparrot_training - Step 37543: {'lr': 7.858907993588937e-05, 'samples': 19222528, 'steps': 37543, 'loss/train': 2.0918405055999756}
+02/26/2022 02:10:54 - INFO - codeparrot_training - Step 37544: {'lr': 7.857716945646603e-05, 'samples': 19223040, 'steps': 37544, 'loss/train': 2.726212501525879}
+02/26/2022 02:10:58 - INFO - codeparrot_training - Step 37545: {'lr': 7.856525971136367e-05, 'samples': 19223552, 'steps': 37545, 'loss/train': 0.4558207392692566}
+02/26/2022 02:11:03 - INFO - codeparrot_training - Step 37546: {'lr': 7.855335070063329e-05, 'samples': 19224064, 'steps': 37546, 'loss/train': 0.9627923965454102}
+02/26/2022 02:11:07 - INFO - codeparrot_training - Step 37547: {'lr': 7.854144242432585e-05, 'samples': 19224576, 'steps': 37547, 'loss/train': 2.3472740650177}
+02/26/2022 02:11:13 - INFO - codeparrot_training - Step 37548: {'lr': 7.852953488249248e-05, 'samples': 19225088, 'steps': 37548, 'loss/train': 2.2475218772888184}
+02/26/2022 02:11:16 - INFO - codeparrot_training - Step 37549: {'lr': 7.851762807518414e-05, 'samples': 19225600, 'steps': 37549, 'loss/train': 1.0498695373535156}
+02/26/2022 02:11:22 - INFO - codeparrot_training - Step 37550: {'lr': 7.850572200245185e-05, 'samples': 19226112, 'steps': 37550, 'loss/train': 1.9693659543991089}
+02/26/2022 02:11:25 - INFO - codeparrot_training - Step 37551: {'lr': 7.849381666434652e-05, 'samples': 19226624, 'steps': 37551, 'loss/train': 1.588321328163147}
+02/26/2022 02:11:31 - INFO - codeparrot_training - Step 37552: {'lr': 7.848191206091926e-05, 'samples': 19227136, 'steps': 37552, 'loss/train': 0.9630889892578125}
+02/26/2022 02:11:34 - INFO - codeparrot_training - Step 37553: {'lr': 7.847000819222098e-05, 'samples': 19227648, 'steps': 37553, 'loss/train': 2.2406740188598633}
+02/26/2022 02:11:40 - INFO - codeparrot_training - Step 37554: {'lr': 7.845810505830286e-05, 'samples': 19228160, 'steps': 37554, 'loss/train': 1.8263521194458008}
+02/26/2022 02:11:43 - INFO - codeparrot_training - Step 37555: {'lr': 7.84462026592156e-05, 'samples': 19228672, 'steps': 37555, 'loss/train': 1.951887607574463}
+02/26/2022 02:11:49 - INFO - codeparrot_training - Step 37556: {'lr': 7.84343009950104e-05, 'samples': 19229184, 'steps': 37556, 'loss/train': 1.4607607126235962}
+02/26/2022 02:11:52 - INFO - codeparrot_training - Step 37557: {'lr': 7.842240006573811e-05, 'samples': 19229696, 'steps': 37557, 'loss/train': 2.2483668327331543}
+02/26/2022 02:11:58 - INFO - codeparrot_training - Step 37558: {'lr': 7.841049987144994e-05, 'samples': 19230208, 'steps': 37558, 'loss/train': 1.1881259679794312}
+02/26/2022 02:12:02 - INFO - codeparrot_training - Step 37559: {'lr': 7.839860041219651e-05, 'samples': 19230720, 'steps': 37559, 'loss/train': 1.3830498456954956}
+02/26/2022 02:12:07 - INFO - codeparrot_training - Step 37560: {'lr': 7.838670168802908e-05, 'samples': 19231232, 'steps': 37560, 'loss/train': 0.6501631736755371}
+02/26/2022 02:12:11 - INFO - codeparrot_training - Step 37561: {'lr': 7.837480369899855e-05, 'samples': 19231744, 'steps': 37561, 'loss/train': 1.4882956743240356}
+02/26/2022 02:12:16 - INFO - codeparrot_training - Step 37562: {'lr': 7.836290644515584e-05, 'samples': 19232256, 'steps': 37562, 'loss/train': 2.1507835388183594}
+02/26/2022 02:12:20 - INFO - codeparrot_training - Step 37563: {'lr': 7.835100992655186e-05, 'samples': 19232768, 'steps': 37563, 'loss/train': 0.5640310049057007}
+02/26/2022 02:12:25 - INFO - codeparrot_training - Step 37564: {'lr': 7.833911414323775e-05, 'samples': 19233280, 'steps': 37564, 'loss/train': 1.0250979661941528}
+02/26/2022 02:12:29 - INFO - codeparrot_training - Step 37565: {'lr': 7.832721909526436e-05, 'samples': 19233792, 'steps': 37565, 'loss/train': 2.854789972305298}
+02/26/2022 02:12:34 - INFO - codeparrot_training - Step 37566: {'lr': 7.831532478268256e-05, 'samples': 19234304, 'steps': 37566, 'loss/train': 1.098199725151062}
+02/26/2022 02:12:38 - INFO - codeparrot_training - Step 37567: {'lr': 7.830343120554345e-05, 'samples': 19234816, 'steps': 37567, 'loss/train': 2.814739465713501}
+02/26/2022 02:12:43 - INFO - codeparrot_training - Step 37568: {'lr': 7.829153836389796e-05, 'samples': 19235328, 'steps': 37568, 'loss/train': 1.360663652420044}
+02/26/2022 02:12:47 - INFO - codeparrot_training - Step 37569: {'lr': 7.827964625779699e-05, 'samples': 19235840, 'steps': 37569, 'loss/train': 2.1116974353790283}
+02/26/2022 02:12:52 - INFO - codeparrot_training - Step 37570: {'lr': 7.82677548872914e-05, 'samples': 19236352, 'steps': 37570, 'loss/train': 1.5010920763015747}
+02/26/2022 02:12:56 - INFO - codeparrot_training - Step 37571: {'lr': 7.825586425243228e-05, 'samples': 19236864, 'steps': 37571, 'loss/train': 1.6926275491714478}
+02/26/2022 02:13:02 - INFO - codeparrot_training - Step 37572: {'lr': 7.824397435327051e-05, 'samples': 19237376, 'steps': 37572, 'loss/train': 3.410578966140747}
+02/26/2022 02:13:05 - INFO - codeparrot_training - Step 37573: {'lr': 7.823208518985706e-05, 'samples': 19237888, 'steps': 37573, 'loss/train': 1.316833257675171}
+02/26/2022 02:13:10 - INFO - codeparrot_training - Step 37574: {'lr': 7.822019676224268e-05, 'samples': 19238400, 'steps': 37574, 'loss/train': 1.5929675102233887}
+02/26/2022 02:13:14 - INFO - codeparrot_training - Step 37575: {'lr': 7.82083090704786e-05, 'samples': 19238912, 'steps': 37575, 'loss/train': 1.9950404167175293}
+02/26/2022 02:13:20 - INFO - codeparrot_training - Step 37576: {'lr': 7.819642211461545e-05, 'samples': 19239424, 'steps': 37576, 'loss/train': 0.31540194153785706}
+02/26/2022 02:13:24 - INFO - codeparrot_training - Step 37577: {'lr': 7.818453589470431e-05, 'samples': 19239936, 'steps': 37577, 'loss/train': 1.7159521579742432}
+02/26/2022 02:13:29 - INFO - codeparrot_training - Step 37578: {'lr': 7.817265041079599e-05, 'samples': 19240448, 'steps': 37578, 'loss/train': 1.6884583234786987}
+02/26/2022 02:13:33 - INFO - codeparrot_training - Step 37579: {'lr': 7.816076566294159e-05, 'samples': 19240960, 'steps': 37579, 'loss/train': 1.075323462486267}
+02/26/2022 02:13:38 - INFO - codeparrot_training - Step 37580: {'lr': 7.814888165119186e-05, 'samples': 19241472, 'steps': 37580, 'loss/train': 1.7058758735656738}
+02/26/2022 02:13:44 - INFO - codeparrot_training - Step 37581: {'lr': 7.813699837559777e-05, 'samples': 19241984, 'steps': 37581, 'loss/train': 0.7966915369033813}
+02/26/2022 02:13:47 - INFO - codeparrot_training - Step 37582: {'lr': 7.812511583621012e-05, 'samples': 19242496, 'steps': 37582, 'loss/train': 2.2373738288879395}
+02/26/2022 02:13:53 - INFO - codeparrot_training - Step 37583: {'lr': 7.811323403307999e-05, 'samples': 19243008, 'steps': 37583, 'loss/train': 2.661898374557495}
+02/26/2022 02:13:56 - INFO - codeparrot_training - Step 37584: {'lr': 7.810135296625817e-05, 'samples': 19243520, 'steps': 37584, 'loss/train': 0.9169530868530273}
+02/26/2022 02:14:03 - INFO - codeparrot_training - Step 37585: {'lr': 7.80894726357956e-05, 'samples': 19244032, 'steps': 37585, 'loss/train': 1.4878981113433838}
+02/26/2022 02:14:06 - INFO - codeparrot_training - Step 37586: {'lr': 7.807759304174302e-05, 'samples': 19244544, 'steps': 37586, 'loss/train': 2.100100040435791}
+02/26/2022 02:14:12 - INFO - codeparrot_training - Step 37587: {'lr': 7.806571418415156e-05, 'samples': 19245056, 'steps': 37587, 'loss/train': 0.7248440980911255}
+02/26/2022 02:14:15 - INFO - codeparrot_training - Step 37588: {'lr': 7.805383606307196e-05, 'samples': 19245568, 'steps': 37588, 'loss/train': 1.9430586099624634}
+02/26/2022 02:14:21 - INFO - codeparrot_training - Step 37589: {'lr': 7.804195867855513e-05, 'samples': 19246080, 'steps': 37589, 'loss/train': 2.1219544410705566}
+02/26/2022 02:14:24 - INFO - codeparrot_training - Step 37590: {'lr': 7.803008203065187e-05, 'samples': 19246592, 'steps': 37590, 'loss/train': 2.662882089614868}
+02/26/2022 02:14:30 - INFO - codeparrot_training - Step 37591: {'lr': 7.801820611941322e-05, 'samples': 19247104, 'steps': 37591, 'loss/train': 2.174276828765869}
+02/26/2022 02:14:33 - INFO - codeparrot_training - Step 37592: {'lr': 7.800633094488987e-05, 'samples': 19247616, 'steps': 37592, 'loss/train': 1.1945936679840088}
+02/26/2022 02:14:39 - INFO - codeparrot_training - Step 37593: {'lr': 7.799445650713294e-05, 'samples': 19248128, 'steps': 37593, 'loss/train': 1.6402393579483032}
+02/26/2022 02:14:42 - INFO - codeparrot_training - Step 37594: {'lr': 7.7982582806193e-05, 'samples': 19248640, 'steps': 37594, 'loss/train': 1.4843868017196655}
+02/26/2022 02:14:48 - INFO - codeparrot_training - Step 37595: {'lr': 7.797070984212112e-05, 'samples': 19249152, 'steps': 37595, 'loss/train': 2.216923475265503}
+02/26/2022 02:14:52 - INFO - codeparrot_training - Step 37596: {'lr': 7.795883761496803e-05, 'samples': 19249664, 'steps': 37596, 'loss/train': 0.7403322458267212}
+02/26/2022 02:14:59 - INFO - codeparrot_training - Step 37597: {'lr': 7.794696612478478e-05, 'samples': 19250176, 'steps': 37597, 'loss/train': 1.7882583141326904}
+02/26/2022 02:15:02 - INFO - codeparrot_training - Step 37598: {'lr': 7.793509537162197e-05, 'samples': 19250688, 'steps': 37598, 'loss/train': 2.614980697631836}
+02/26/2022 02:15:08 - INFO - codeparrot_training - Step 37599: {'lr': 7.792322535553062e-05, 'samples': 19251200, 'steps': 37599, 'loss/train': 1.7173736095428467}
+02/26/2022 02:15:11 - INFO - codeparrot_training - Step 37600: {'lr': 7.791135607656147e-05, 'samples': 19251712, 'steps': 37600, 'loss/train': 3.3310883045196533}
+02/26/2022 02:15:17 - INFO - codeparrot_training - Step 37601: {'lr': 7.78994875347655e-05, 'samples': 19252224, 'steps': 37601, 'loss/train': 1.8627843856811523}
+02/26/2022 02:15:20 - INFO - codeparrot_training - Step 37602: {'lr': 7.788761973019348e-05, 'samples': 19252736, 'steps': 37602, 'loss/train': 1.2182908058166504}
+02/26/2022 02:15:26 - INFO - codeparrot_training - Step 37603: {'lr': 7.787575266289626e-05, 'samples': 19253248, 'steps': 37603, 'loss/train': 0.721784770488739}
+02/26/2022 02:15:29 - INFO - codeparrot_training - Step 37604: {'lr': 7.786388633292457e-05, 'samples': 19253760, 'steps': 37604, 'loss/train': 2.258483648300171}
+02/26/2022 02:15:35 - INFO - codeparrot_training - Step 37605: {'lr': 7.78520207403294e-05, 'samples': 19254272, 'steps': 37605, 'loss/train': 2.0320520401000977}
+02/26/2022 02:15:38 - INFO - codeparrot_training - Step 37606: {'lr': 7.784015588516152e-05, 'samples': 19254784, 'steps': 37606, 'loss/train': 0.6851996183395386}
+02/26/2022 02:15:45 - INFO - codeparrot_training - Step 37607: {'lr': 7.782829176747174e-05, 'samples': 19255296, 'steps': 37607, 'loss/train': 1.8158432245254517}
+02/26/2022 02:15:48 - INFO - codeparrot_training - Step 37608: {'lr': 7.78164283873109e-05, 'samples': 19255808, 'steps': 37608, 'loss/train': 1.988140344619751}
+02/26/2022 02:15:54 - INFO - codeparrot_training - Step 37609: {'lr': 7.780456574472971e-05, 'samples': 19256320, 'steps': 37609, 'loss/train': 0.06278786808252335}
+02/26/2022 02:15:57 - INFO - codeparrot_training - Step 37610: {'lr': 7.779270383977916e-05, 'samples': 19256832, 'steps': 37610, 'loss/train': 1.2416313886642456}
+02/26/2022 02:16:03 - INFO - codeparrot_training - Step 37611: {'lr': 7.778084267251001e-05, 'samples': 19257344, 'steps': 37611, 'loss/train': 1.9865598678588867}
+02/26/2022 02:16:06 - INFO - codeparrot_training - Step 37612: {'lr': 7.776898224297302e-05, 'samples': 19257856, 'steps': 37612, 'loss/train': 1.4167735576629639}
+02/26/2022 02:16:12 - INFO - codeparrot_training - Step 37613: {'lr': 7.775712255121895e-05, 'samples': 19258368, 'steps': 37613, 'loss/train': 1.3296536207199097}
+02/26/2022 02:16:15 - INFO - codeparrot_training - Step 37614: {'lr': 7.774526359729876e-05, 'samples': 19258880, 'steps': 37614, 'loss/train': 1.7219594717025757}
+02/26/2022 02:16:21 - INFO - codeparrot_training - Step 37615: {'lr': 7.773340538126314e-05, 'samples': 19259392, 'steps': 37615, 'loss/train': 0.7995109558105469}
+02/26/2022 02:16:24 - INFO - codeparrot_training - Step 37616: {'lr': 7.772154790316294e-05, 'samples': 19259904, 'steps': 37616, 'loss/train': 1.7734417915344238}
+02/26/2022 02:16:30 - INFO - codeparrot_training - Step 37617: {'lr': 7.770969116304883e-05, 'samples': 19260416, 'steps': 37617, 'loss/train': 2.0931289196014404}
+02/26/2022 02:16:33 - INFO - codeparrot_training - Step 37618: {'lr': 7.769783516097179e-05, 'samples': 19260928, 'steps': 37618, 'loss/train': 1.3326913118362427}
+02/26/2022 02:16:40 - INFO - codeparrot_training - Step 37619: {'lr': 7.768597989698248e-05, 'samples': 19261440, 'steps': 37619, 'loss/train': 0.1903793215751648}
+02/26/2022 02:16:43 - INFO - codeparrot_training - Step 37620: {'lr': 7.767412537113175e-05, 'samples': 19261952, 'steps': 37620, 'loss/train': 1.813303828239441}
+02/26/2022 02:16:49 - INFO - codeparrot_training - Step 37621: {'lr': 7.766227158347025e-05, 'samples': 19262464, 'steps': 37621, 'loss/train': 3.632634162902832}
+02/26/2022 02:16:52 - INFO - codeparrot_training - Step 37622: {'lr': 7.765041853404892e-05, 'samples': 19262976, 'steps': 37622, 'loss/train': 3.0631906986236572}
+02/26/2022 02:16:58 - INFO - codeparrot_training - Step 37623: {'lr': 7.763856622291848e-05, 'samples': 19263488, 'steps': 37623, 'loss/train': 2.079645872116089}
+02/26/2022 02:17:01 - INFO - codeparrot_training - Step 37624: {'lr': 7.762671465012972e-05, 'samples': 19264000, 'steps': 37624, 'loss/train': 1.6525745391845703}
+02/26/2022 02:17:07 - INFO - codeparrot_training - Step 37625: {'lr': 7.761486381573326e-05, 'samples': 19264512, 'steps': 37625, 'loss/train': 1.8114641904830933}
+02/26/2022 02:17:10 - INFO - codeparrot_training - Step 37626: {'lr': 7.76030137197801e-05, 'samples': 19265024, 'steps': 37626, 'loss/train': 0.35486334562301636}
+02/26/2022 02:17:16 - INFO - codeparrot_training - Step 37627: {'lr': 7.759116436232077e-05, 'samples': 19265536, 'steps': 37627, 'loss/train': 1.0562305450439453}
+02/26/2022 02:17:19 - INFO - codeparrot_training - Step 37628: {'lr': 7.757931574340635e-05, 'samples': 19266048, 'steps': 37628, 'loss/train': 1.7527967691421509}
+02/26/2022 02:17:26 - INFO - codeparrot_training - Step 37629: {'lr': 7.756746786308719e-05, 'samples': 19266560, 'steps': 37629, 'loss/train': 1.2501088380813599}
+02/26/2022 02:17:29 - INFO - codeparrot_training - Step 37630: {'lr': 7.755562072141434e-05, 'samples': 19267072, 'steps': 37630, 'loss/train': 1.4705219268798828}
+02/26/2022 02:17:35 - INFO - codeparrot_training - Step 37631: {'lr': 7.75437743184384e-05, 'samples': 19267584, 'steps': 37631, 'loss/train': 1.6251333951950073}
+02/26/2022 02:17:38 - INFO - codeparrot_training - Step 37632: {'lr': 7.75319286542103e-05, 'samples': 19268096, 'steps': 37632, 'loss/train': 2.0825035572052}
+02/26/2022 02:17:44 - INFO - codeparrot_training - Step 37633: {'lr': 7.75200837287805e-05, 'samples': 19268608, 'steps': 37633, 'loss/train': 2.6459314823150635}
+02/26/2022 02:17:48 - INFO - codeparrot_training - Step 37634: {'lr': 7.750823954219996e-05, 'samples': 19269120, 'steps': 37634, 'loss/train': 1.6984012126922607}
+02/26/2022 02:17:53 - INFO - codeparrot_training - Step 37635: {'lr': 7.749639609451928e-05, 'samples': 19269632, 'steps': 37635, 'loss/train': 1.7171841859817505}
+02/26/2022 02:17:57 - INFO - codeparrot_training - Step 37636: {'lr': 7.748455338578941e-05, 'samples': 19270144, 'steps': 37636, 'loss/train': 1.8325073719024658}
+02/26/2022 02:18:02 - INFO - codeparrot_training - Step 37637: {'lr': 7.747271141606079e-05, 'samples': 19270656, 'steps': 37637, 'loss/train': 1.9868876934051514}
+02/26/2022 02:18:06 - INFO - codeparrot_training - Step 37638: {'lr': 7.746087018538437e-05, 'samples': 19271168, 'steps': 37638, 'loss/train': 0.4312187135219574}
+02/26/2022 02:18:11 - INFO - codeparrot_training - Step 37639: {'lr': 7.744902969381071e-05, 'samples': 19271680, 'steps': 37639, 'loss/train': 1.9650650024414062}
+02/26/2022 02:18:15 - INFO - codeparrot_training - Step 37640: {'lr': 7.743718994139071e-05, 'samples': 19272192, 'steps': 37640, 'loss/train': 1.4446587562561035}
+02/26/2022 02:18:20 - INFO - codeparrot_training - Step 37641: {'lr': 7.742535092817499e-05, 'samples': 19272704, 'steps': 37641, 'loss/train': 1.2791903018951416}
+02/26/2022 02:18:27 - INFO - codeparrot_training - Step 37642: {'lr': 7.741351265421426e-05, 'samples': 19273216, 'steps': 37642, 'loss/train': 1.136134386062622}
+02/26/2022 02:18:30 - INFO - codeparrot_training - Step 37643: {'lr': 7.740167511955915e-05, 'samples': 19273728, 'steps': 37643, 'loss/train': 0.8914410471916199}
+02/26/2022 02:18:37 - INFO - codeparrot_training - Step 37644: {'lr': 7.738983832426056e-05, 'samples': 19274240, 'steps': 37644, 'loss/train': 1.6500110626220703}
+02/26/2022 02:18:40 - INFO - codeparrot_training - Step 37645: {'lr': 7.737800226836905e-05, 'samples': 19274752, 'steps': 37645, 'loss/train': 1.1622395515441895}
+02/26/2022 02:18:43 - INFO - codeparrot_training - Step 37646: {'lr': 7.736616695193541e-05, 'samples': 19275264, 'steps': 37646, 'loss/train': 0.6352105736732483}
+02/26/2022 02:18:49 - INFO - codeparrot_training - Step 37647: {'lr': 7.735433237501018e-05, 'samples': 19275776, 'steps': 37647, 'loss/train': 1.77727472782135}
+02/26/2022 02:18:52 - INFO - codeparrot_training - Step 37648: {'lr': 7.734249853764428e-05, 'samples': 19276288, 'steps': 37648, 'loss/train': 1.2371327877044678}
+02/26/2022 02:18:58 - INFO - codeparrot_training - Step 37649: {'lr': 7.733066543988828e-05, 'samples': 19276800, 'steps': 37649, 'loss/train': 0.880501389503479}
+02/26/2022 02:19:02 - INFO - codeparrot_training - Step 37650: {'lr': 7.731883308179289e-05, 'samples': 19277312, 'steps': 37650, 'loss/train': 1.0649651288986206}
+02/26/2022 02:19:07 - INFO - codeparrot_training - Step 37651: {'lr': 7.730700146340868e-05, 'samples': 19277824, 'steps': 37651, 'loss/train': 1.6635488271713257}
+02/26/2022 02:19:10 - INFO - codeparrot_training - Step 37652: {'lr': 7.729517058478653e-05, 'samples': 19278336, 'steps': 37652, 'loss/train': 1.3024067878723145}
+02/26/2022 02:19:17 - INFO - codeparrot_training - Step 37653: {'lr': 7.728334044597704e-05, 'samples': 19278848, 'steps': 37653, 'loss/train': 2.0576956272125244}
+02/26/2022 02:19:20 - INFO - codeparrot_training - Step 37654: {'lr': 7.727151104703086e-05, 'samples': 19279360, 'steps': 37654, 'loss/train': 1.7574636936187744}
+02/26/2022 02:19:26 - INFO - codeparrot_training - Step 37655: {'lr': 7.725968238799869e-05, 'samples': 19279872, 'steps': 37655, 'loss/train': 2.5031111240386963}
+02/26/2022 02:19:29 - INFO - codeparrot_training - Step 37656: {'lr': 7.724785446893112e-05, 'samples': 19280384, 'steps': 37656, 'loss/train': 1.6008076667785645}
+02/26/2022 02:19:35 - INFO - codeparrot_training - Step 37657: {'lr': 7.723602728987894e-05, 'samples': 19280896, 'steps': 37657, 'loss/train': 1.8773912191390991}
+02/26/2022 02:19:38 - INFO - codeparrot_training - Step 37658: {'lr': 7.722420085089277e-05, 'samples': 19281408, 'steps': 37658, 'loss/train': 2.6788976192474365}
+02/26/2022 02:19:44 - INFO - codeparrot_training - Step 37659: {'lr': 7.721237515202329e-05, 'samples': 19281920, 'steps': 37659, 'loss/train': 0.5612935423851013}
+02/26/2022 02:19:47 - INFO - codeparrot_training - Step 37660: {'lr': 7.7200550193321e-05, 'samples': 19282432, 'steps': 37660, 'loss/train': 2.197359085083008}
+02/26/2022 02:19:53 - INFO - codeparrot_training - Step 37661: {'lr': 7.718872597483679e-05, 'samples': 19282944, 'steps': 37661, 'loss/train': 0.9181150197982788}
+02/26/2022 02:19:56 - INFO - codeparrot_training - Step 37662: {'lr': 7.717690249662113e-05, 'samples': 19283456, 'steps': 37662, 'loss/train': 2.171093463897705}
+02/26/2022 02:20:02 - INFO - codeparrot_training - Step 37663: {'lr': 7.716507975872489e-05, 'samples': 19283968, 'steps': 37663, 'loss/train': 1.2959262132644653}
+02/26/2022 02:20:05 - INFO - codeparrot_training - Step 37664: {'lr': 7.715325776119841e-05, 'samples': 19284480, 'steps': 37664, 'loss/train': 1.7885850667953491}
+02/26/2022 02:20:12 - INFO - codeparrot_training - Step 37665: {'lr': 7.714143650409258e-05, 'samples': 19284992, 'steps': 37665, 'loss/train': 1.8924858570098877}
+02/26/2022 02:20:15 - INFO - codeparrot_training - Step 37666: {'lr': 7.712961598745785e-05, 'samples': 19285504, 'steps': 37666, 'loss/train': 1.885291337966919}
+02/26/2022 02:20:21 - INFO - codeparrot_training - Step 37667: {'lr': 7.711779621134512e-05, 'samples': 19286016, 'steps': 37667, 'loss/train': 2.95695424079895}
+02/26/2022 02:20:24 - INFO - codeparrot_training - Step 37668: {'lr': 7.71059771758047e-05, 'samples': 19286528, 'steps': 37668, 'loss/train': 1.8408385515213013}
+02/26/2022 02:20:30 - INFO - codeparrot_training - Step 37669: {'lr': 7.709415888088744e-05, 'samples': 19287040, 'steps': 37669, 'loss/train': 0.85841304063797}
+02/26/2022 02:20:34 - INFO - codeparrot_training - Step 37670: {'lr': 7.708234132664382e-05, 'samples': 19287552, 'steps': 37670, 'loss/train': 1.4242287874221802}
+02/26/2022 02:20:39 - INFO - codeparrot_training - Step 37671: {'lr': 7.707052451312471e-05, 'samples': 19288064, 'steps': 37671, 'loss/train': 1.4721338748931885}
+02/26/2022 02:20:43 - INFO - codeparrot_training - Step 37672: {'lr': 7.705870844038041e-05, 'samples': 19288576, 'steps': 37672, 'loss/train': 0.7704534530639648}
+02/26/2022 02:20:48 - INFO - codeparrot_training - Step 37673: {'lr': 7.704689310846175e-05, 'samples': 19289088, 'steps': 37673, 'loss/train': 2.886709213256836}
+02/26/2022 02:20:52 - INFO - codeparrot_training - Step 37674: {'lr': 7.70350785174192e-05, 'samples': 19289600, 'steps': 37674, 'loss/train': 0.5283398628234863}
+02/26/2022 02:20:58 - INFO - codeparrot_training - Step 37675: {'lr': 7.702326466730356e-05, 'samples': 19290112, 'steps': 37675, 'loss/train': 1.98975670337677}
+02/26/2022 02:21:02 - INFO - codeparrot_training - Step 37676: {'lr': 7.70114515581653e-05, 'samples': 19290624, 'steps': 37676, 'loss/train': 2.1917145252227783}
+02/26/2022 02:21:07 - INFO - codeparrot_training - Step 37677: {'lr': 7.699963919005506e-05, 'samples': 19291136, 'steps': 37677, 'loss/train': 0.47659945487976074}
+02/26/2022 02:21:11 - INFO - codeparrot_training - Step 37678: {'lr': 7.698782756302336e-05, 'samples': 19291648, 'steps': 37678, 'loss/train': 0.7841382026672363}
+02/26/2022 02:21:16 - INFO - codeparrot_training - Step 37679: {'lr': 7.697601667712092e-05, 'samples': 19292160, 'steps': 37679, 'loss/train': 3.004986524581909}
+02/26/2022 02:21:20 - INFO - codeparrot_training - Step 37680: {'lr': 7.696420653239832e-05, 'samples': 19292672, 'steps': 37680, 'loss/train': 0.6561324596405029}
+02/26/2022 02:21:25 - INFO - codeparrot_training - Step 37681: {'lr': 7.695239712890609e-05, 'samples': 19293184, 'steps': 37681, 'loss/train': 2.267157793045044}
+02/26/2022 02:21:29 - INFO - codeparrot_training - Step 37682: {'lr': 7.694058846669477e-05, 'samples': 19293696, 'steps': 37682, 'loss/train': 2.2618095874786377}
+02/26/2022 02:21:34 - INFO - codeparrot_training - Step 37683: {'lr': 7.692878054581512e-05, 'samples': 19294208, 'steps': 37683, 'loss/train': 2.078975200653076}
+02/26/2022 02:21:38 - INFO - codeparrot_training - Step 37684: {'lr': 7.69169733663176e-05, 'samples': 19294720, 'steps': 37684, 'loss/train': 1.5002537965774536}
+02/26/2022 02:21:44 - INFO - codeparrot_training - Step 37685: {'lr': 7.690516692825278e-05, 'samples': 19295232, 'steps': 37685, 'loss/train': 2.2105298042297363}
+02/26/2022 02:21:47 - INFO - codeparrot_training - Step 37686: {'lr': 7.68933612316712e-05, 'samples': 19295744, 'steps': 37686, 'loss/train': 2.2338786125183105}
+02/26/2022 02:21:53 - INFO - codeparrot_training - Step 37687: {'lr': 7.688155627662357e-05, 'samples': 19296256, 'steps': 37687, 'loss/train': 2.2466061115264893}
+02/26/2022 02:21:56 - INFO - codeparrot_training - Step 37688: {'lr': 7.68697520631604e-05, 'samples': 19296768, 'steps': 37688, 'loss/train': 1.1804167032241821}
+02/26/2022 02:22:02 - INFO - codeparrot_training - Step 37689: {'lr': 7.685794859133222e-05, 'samples': 19297280, 'steps': 37689, 'loss/train': 0.8269443511962891}
+02/26/2022 02:22:05 - INFO - codeparrot_training - Step 37690: {'lr': 7.684614586118954e-05, 'samples': 19297792, 'steps': 37690, 'loss/train': 1.299662709236145}
+02/26/2022 02:22:12 - INFO - codeparrot_training - Step 37691: {'lr': 7.683434387278308e-05, 'samples': 19298304, 'steps': 37691, 'loss/train': 0.918686032295227}
+02/26/2022 02:22:15 - INFO - codeparrot_training - Step 37692: {'lr': 7.682254262616326e-05, 'samples': 19298816, 'steps': 37692, 'loss/train': 2.050319194793701}
+02/26/2022 02:22:21 - INFO - codeparrot_training - Step 37693: {'lr': 7.68107421213807e-05, 'samples': 19299328, 'steps': 37693, 'loss/train': 1.658482551574707}
+02/26/2022 02:22:24 - INFO - codeparrot_training - Step 37694: {'lr': 7.679894235848586e-05, 'samples': 19299840, 'steps': 37694, 'loss/train': 1.7333959341049194}
+02/26/2022 02:22:30 - INFO - codeparrot_training - Step 37695: {'lr': 7.678714333752942e-05, 'samples': 19300352, 'steps': 37695, 'loss/train': 1.998561143875122}
+02/26/2022 02:22:33 - INFO - codeparrot_training - Step 37696: {'lr': 7.677534505856188e-05, 'samples': 19300864, 'steps': 37696, 'loss/train': 0.7671431303024292}
+02/26/2022 02:22:39 - INFO - codeparrot_training - Step 37697: {'lr': 7.676354752163373e-05, 'samples': 19301376, 'steps': 37697, 'loss/train': 2.2496800422668457}
+02/26/2022 02:22:42 - INFO - codeparrot_training - Step 37698: {'lr': 7.675175072679546e-05, 'samples': 19301888, 'steps': 37698, 'loss/train': 2.202409505844116}
+02/26/2022 02:22:48 - INFO - codeparrot_training - Step 37699: {'lr': 7.673995467409775e-05, 'samples': 19302400, 'steps': 37699, 'loss/train': 0.9044036269187927}
+02/26/2022 02:22:51 - INFO - codeparrot_training - Step 37700: {'lr': 7.672815936359106e-05, 'samples': 19302912, 'steps': 37700, 'loss/train': 1.8131860494613647}
+02/26/2022 02:22:58 - INFO - codeparrot_training - Step 37701: {'lr': 7.671636479532584e-05, 'samples': 19303424, 'steps': 37701, 'loss/train': 1.8823350667953491}
+02/26/2022 02:23:01 - INFO - codeparrot_training - Step 37702: {'lr': 7.670457096935279e-05, 'samples': 19303936, 'steps': 37702, 'loss/train': 1.5045844316482544}
+02/26/2022 02:23:07 - INFO - codeparrot_training - Step 37703: {'lr': 7.66927778857223e-05, 'samples': 19304448, 'steps': 37703, 'loss/train': 1.8278326988220215}
+02/26/2022 02:23:10 - INFO - codeparrot_training - Step 37704: {'lr': 7.668098554448493e-05, 'samples': 19304960, 'steps': 37704, 'loss/train': 2.500993251800537}
+02/26/2022 02:23:16 - INFO - codeparrot_training - Step 37705: {'lr': 7.666919394569111e-05, 'samples': 19305472, 'steps': 37705, 'loss/train': 0.6172981858253479}
+02/26/2022 02:23:19 - INFO - codeparrot_training - Step 37706: {'lr': 7.665740308939156e-05, 'samples': 19305984, 'steps': 37706, 'loss/train': 1.8391003608703613}
+02/26/2022 02:23:25 - INFO - codeparrot_training - Step 37707: {'lr': 7.664561297563653e-05, 'samples': 19306496, 'steps': 37707, 'loss/train': 0.7069413661956787}
+02/26/2022 02:23:28 - INFO - codeparrot_training - Step 37708: {'lr': 7.663382360447672e-05, 'samples': 19307008, 'steps': 37708, 'loss/train': 1.7706489562988281}
+02/26/2022 02:23:34 - INFO - codeparrot_training - Step 37709: {'lr': 7.662203497596246e-05, 'samples': 19307520, 'steps': 37709, 'loss/train': 1.852295160293579}
+02/26/2022 02:23:37 - INFO - codeparrot_training - Step 37710: {'lr': 7.661024709014445e-05, 'samples': 19308032, 'steps': 37710, 'loss/train': 1.259118914604187}
+02/26/2022 02:23:44 - INFO - codeparrot_training - Step 37711: {'lr': 7.659845994707304e-05, 'samples': 19308544, 'steps': 37711, 'loss/train': 1.4487167596817017}
+02/26/2022 02:23:47 - INFO - codeparrot_training - Step 37712: {'lr': 7.65866735467988e-05, 'samples': 19309056, 'steps': 37712, 'loss/train': 0.6487449407577515}
+02/26/2022 02:23:53 - INFO - codeparrot_training - Step 37713: {'lr': 7.657488788937211e-05, 'samples': 19309568, 'steps': 37713, 'loss/train': 2.396380662918091}
+02/26/2022 02:23:56 - INFO - codeparrot_training - Step 37714: {'lr': 7.65631029748436e-05, 'samples': 19310080, 'steps': 37714, 'loss/train': 1.8077212572097778}
+02/26/2022 02:24:02 - INFO - codeparrot_training - Step 37715: {'lr': 7.655131880326369e-05, 'samples': 19310592, 'steps': 37715, 'loss/train': 1.7396163940429688}
+02/26/2022 02:24:05 - INFO - codeparrot_training - Step 37716: {'lr': 7.653953537468287e-05, 'samples': 19311104, 'steps': 37716, 'loss/train': 1.4670031070709229}
+02/26/2022 02:24:11 - INFO - codeparrot_training - Step 37717: {'lr': 7.652775268915149e-05, 'samples': 19311616, 'steps': 37717, 'loss/train': 1.8817789554595947}
+02/26/2022 02:24:14 - INFO - codeparrot_training - Step 37718: {'lr': 7.651597074672023e-05, 'samples': 19312128, 'steps': 37718, 'loss/train': 0.9007397890090942}
+02/26/2022 02:24:20 - INFO - codeparrot_training - Step 37719: {'lr': 7.650418954743946e-05, 'samples': 19312640, 'steps': 37719, 'loss/train': 1.0696829557418823}
+02/26/2022 02:24:23 - INFO - codeparrot_training - Step 37720: {'lr': 7.649240909135965e-05, 'samples': 19313152, 'steps': 37720, 'loss/train': 1.502488613128662}
+02/26/2022 02:24:29 - INFO - codeparrot_training - Step 37721: {'lr': 7.648062937853117e-05, 'samples': 19313664, 'steps': 37721, 'loss/train': 1.6990318298339844}
+02/26/2022 02:24:32 - INFO - codeparrot_training - Step 37722: {'lr': 7.646885040900466e-05, 'samples': 19314176, 'steps': 37722, 'loss/train': 2.4838478565216064}
+02/26/2022 02:24:38 - INFO - codeparrot_training - Step 37723: {'lr': 7.645707218283044e-05, 'samples': 19314688, 'steps': 37723, 'loss/train': 1.113002896308899}
+02/26/2022 02:24:41 - INFO - codeparrot_training - Step 37724: {'lr': 7.644529470005917e-05, 'samples': 19315200, 'steps': 37724, 'loss/train': 1.4262864589691162}
+02/26/2022 02:24:49 - INFO - codeparrot_training - Step 37725: {'lr': 7.643351796074097e-05, 'samples': 19315712, 'steps': 37725, 'loss/train': 1.3646433353424072}
+02/26/2022 02:24:52 - INFO - codeparrot_training - Step 37726: {'lr': 7.642174196492654e-05, 'samples': 19316224, 'steps': 37726, 'loss/train': 0.8107653856277466}
+02/26/2022 02:24:57 - INFO - codeparrot_training - Step 37727: {'lr': 7.64099667126662e-05, 'samples': 19316736, 'steps': 37727, 'loss/train': 0.8023613691329956}
+02/26/2022 02:25:01 - INFO - codeparrot_training - Step 37728: {'lr': 7.63981922040106e-05, 'samples': 19317248, 'steps': 37728, 'loss/train': 1.1433221101760864}
+02/26/2022 02:25:06 - INFO - codeparrot_training - Step 37729: {'lr': 7.638641843900984e-05, 'samples': 19317760, 'steps': 37729, 'loss/train': 1.3264213800430298}
+02/26/2022 02:25:12 - INFO - codeparrot_training - Step 37730: {'lr': 7.63746454177146e-05, 'samples': 19318272, 'steps': 37730, 'loss/train': 1.3353325128555298}
+02/26/2022 02:25:16 - INFO - codeparrot_training - Step 37731: {'lr': 7.63628731401752e-05, 'samples': 19318784, 'steps': 37731, 'loss/train': 1.8617967367172241}
+02/26/2022 02:25:19 - INFO - codeparrot_training - Step 37732: {'lr': 7.635110160644227e-05, 'samples': 19319296, 'steps': 37732, 'loss/train': 1.646790623664856}
+02/26/2022 02:25:25 - INFO - codeparrot_training - Step 37733: {'lr': 7.633933081656592e-05, 'samples': 19319808, 'steps': 37733, 'loss/train': 1.2764478921890259}
+02/26/2022 02:25:28 - INFO - codeparrot_training - Step 37734: {'lr': 7.63275607705968e-05, 'samples': 19320320, 'steps': 37734, 'loss/train': 1.2741814851760864}
+02/26/2022 02:25:34 - INFO - codeparrot_training - Step 37735: {'lr': 7.631579146858525e-05, 'samples': 19320832, 'steps': 37735, 'loss/train': 1.9060357809066772}
+02/26/2022 02:25:37 - INFO - codeparrot_training - Step 37736: {'lr': 7.630402291058164e-05, 'samples': 19321344, 'steps': 37736, 'loss/train': 1.738125205039978}
+02/26/2022 02:25:44 - INFO - codeparrot_training - Step 37737: {'lr': 7.62922550966365e-05, 'samples': 19321856, 'steps': 37737, 'loss/train': 2.2915990352630615}
+02/26/2022 02:25:48 - INFO - codeparrot_training - Step 37738: {'lr': 7.628048802680021e-05, 'samples': 19322368, 'steps': 37738, 'loss/train': 1.769662857055664}
+02/26/2022 02:25:53 - INFO - codeparrot_training - Step 37739: {'lr': 7.626872170112315e-05, 'samples': 19322880, 'steps': 37739, 'loss/train': 1.2634891271591187}
+02/26/2022 02:25:57 - INFO - codeparrot_training - Step 37740: {'lr': 7.625695611965564e-05, 'samples': 19323392, 'steps': 37740, 'loss/train': 0.36607810854911804}
+02/26/2022 02:26:02 - INFO - codeparrot_training - Step 37741: {'lr': 7.624519128244823e-05, 'samples': 19323904, 'steps': 37741, 'loss/train': 1.370645523071289}
+02/26/2022 02:26:06 - INFO - codeparrot_training - Step 37742: {'lr': 7.623342718955123e-05, 'samples': 19324416, 'steps': 37742, 'loss/train': 1.4751585721969604}
+02/26/2022 02:26:11 - INFO - codeparrot_training - Step 37743: {'lr': 7.622166384101506e-05, 'samples': 19324928, 'steps': 37743, 'loss/train': 1.82878577709198}
+02/26/2022 02:26:15 - INFO - codeparrot_training - Step 37744: {'lr': 7.620990123689004e-05, 'samples': 19325440, 'steps': 37744, 'loss/train': 1.614038348197937}
+02/26/2022 02:26:20 - INFO - codeparrot_training - Step 37745: {'lr': 7.619813937722667e-05, 'samples': 19325952, 'steps': 37745, 'loss/train': 1.630193829536438}
+02/26/2022 02:26:23 - INFO - codeparrot_training - Step 37746: {'lr': 7.61863782620753e-05, 'samples': 19326464, 'steps': 37746, 'loss/train': 1.2569080591201782}
+02/26/2022 02:26:30 - INFO - codeparrot_training - Step 37747: {'lr': 7.617461789148627e-05, 'samples': 19326976, 'steps': 37747, 'loss/train': 0.7265498638153076}
+02/26/2022 02:26:33 - INFO - codeparrot_training - Step 37748: {'lr': 7.616285826550995e-05, 'samples': 19327488, 'steps': 37748, 'loss/train': 2.1613547801971436}
+02/26/2022 02:26:39 - INFO - codeparrot_training - Step 37749: {'lr': 7.615109938419679e-05, 'samples': 19328000, 'steps': 37749, 'loss/train': 1.4838377237319946}
+02/26/2022 02:26:42 - INFO - codeparrot_training - Step 37750: {'lr': 7.613934124759711e-05, 'samples': 19328512, 'steps': 37750, 'loss/train': 1.4032288789749146}
+02/26/2022 02:26:48 - INFO - codeparrot_training - Step 37751: {'lr': 7.612758385576133e-05, 'samples': 19329024, 'steps': 37751, 'loss/train': 1.520467758178711}
+02/26/2022 02:26:51 - INFO - codeparrot_training - Step 37752: {'lr': 7.611582720873963e-05, 'samples': 19329536, 'steps': 37752, 'loss/train': 0.9275469779968262}
+02/26/2022 02:26:57 - INFO - codeparrot_training - Step 37753: {'lr': 7.610407130658264e-05, 'samples': 19330048, 'steps': 37753, 'loss/train': 1.8000197410583496}
+02/26/2022 02:27:00 - INFO - codeparrot_training - Step 37754: {'lr': 7.609231614934057e-05, 'samples': 19330560, 'steps': 37754, 'loss/train': 2.0067830085754395}
+02/26/2022 02:27:06 - INFO - codeparrot_training - Step 37755: {'lr': 7.60805617370638e-05, 'samples': 19331072, 'steps': 37755, 'loss/train': 1.082345962524414}
+02/26/2022 02:27:09 - INFO - codeparrot_training - Step 37756: {'lr': 7.60688080698026e-05, 'samples': 19331584, 'steps': 37756, 'loss/train': 1.7399219274520874}
+02/26/2022 02:27:16 - INFO - codeparrot_training - Step 37757: {'lr': 7.605705514760746e-05, 'samples': 19332096, 'steps': 37757, 'loss/train': 1.2099108695983887}
+02/26/2022 02:27:20 - INFO - codeparrot_training - Step 37758: {'lr': 7.604530297052866e-05, 'samples': 19332608, 'steps': 37758, 'loss/train': 1.954397201538086}
+02/26/2022 02:27:25 - INFO - codeparrot_training - Step 37759: {'lr': 7.603355153861657e-05, 'samples': 19333120, 'steps': 37759, 'loss/train': 2.264927864074707}
+02/26/2022 02:27:29 - INFO - codeparrot_training - Step 37760: {'lr': 7.602180085192142e-05, 'samples': 19333632, 'steps': 37760, 'loss/train': 2.3893911838531494}
+02/26/2022 02:27:35 - INFO - codeparrot_training - Step 37761: {'lr': 7.60100509104937e-05, 'samples': 19334144, 'steps': 37761, 'loss/train': 1.7324268817901611}
+02/26/2022 02:27:38 - INFO - codeparrot_training - Step 37762: {'lr': 7.59983017143836e-05, 'samples': 19334656, 'steps': 37762, 'loss/train': 0.6631697416305542}
+02/26/2022 02:27:44 - INFO - codeparrot_training - Step 37763: {'lr': 7.598655326364167e-05, 'samples': 19335168, 'steps': 37763, 'loss/train': 1.6210612058639526}
+02/26/2022 02:27:47 - INFO - codeparrot_training - Step 37764: {'lr': 7.597480555831796e-05, 'samples': 19335680, 'steps': 37764, 'loss/train': 0.8646954298019409}
+02/26/2022 02:27:53 - INFO - codeparrot_training - Step 37765: {'lr': 7.5963058598463e-05, 'samples': 19336192, 'steps': 37765, 'loss/train': 2.4253830909729004}
+02/26/2022 02:27:56 - INFO - codeparrot_training - Step 37766: {'lr': 7.595131238412693e-05, 'samples': 19336704, 'steps': 37766, 'loss/train': 1.2177038192749023}
+02/26/2022 02:28:02 - INFO - codeparrot_training - Step 37767: {'lr': 7.593956691536032e-05, 'samples': 19337216, 'steps': 37767, 'loss/train': 2.2936832904815674}
+02/26/2022 02:28:05 - INFO - codeparrot_training - Step 37768: {'lr': 7.592782219221323e-05, 'samples': 19337728, 'steps': 37768, 'loss/train': 0.20163801312446594}
+02/26/2022 02:28:11 - INFO - codeparrot_training - Step 37769: {'lr': 7.591607821473612e-05, 'samples': 19338240, 'steps': 37769, 'loss/train': 2.099280595779419}
+02/26/2022 02:28:14 - INFO - codeparrot_training - Step 37770: {'lr': 7.590433498297916e-05, 'samples': 19338752, 'steps': 37770, 'loss/train': 1.8411799669265747}
+02/26/2022 02:28:20 - INFO - codeparrot_training - Step 37771: {'lr': 7.589259249699285e-05, 'samples': 19339264, 'steps': 37771, 'loss/train': 2.408233404159546}
+02/26/2022 02:28:23 - INFO - codeparrot_training - Step 37772: {'lr': 7.588085075682738e-05, 'samples': 19339776, 'steps': 37772, 'loss/train': 1.186763048171997}
+02/26/2022 02:28:30 - INFO - codeparrot_training - Step 37773: {'lr': 7.586910976253306e-05, 'samples': 19340288, 'steps': 37773, 'loss/train': 1.0824908018112183}
+02/26/2022 02:28:35 - INFO - codeparrot_training - Step 37774: {'lr': 7.585736951416012e-05, 'samples': 19340800, 'steps': 37774, 'loss/train': 1.772605538368225}
+02/26/2022 02:28:39 - INFO - codeparrot_training - Step 37775: {'lr': 7.584563001175895e-05, 'samples': 19341312, 'steps': 37775, 'loss/train': 1.5589966773986816}
+02/26/2022 02:28:44 - INFO - codeparrot_training - Step 37776: {'lr': 7.583389125537982e-05, 'samples': 19341824, 'steps': 37776, 'loss/train': 1.0156264305114746}
+02/26/2022 02:28:48 - INFO - codeparrot_training - Step 37777: {'lr': 7.5822153245073e-05, 'samples': 19342336, 'steps': 37777, 'loss/train': 2.102470874786377}
+02/26/2022 02:28:53 - INFO - codeparrot_training - Step 37778: {'lr': 7.58104159808887e-05, 'samples': 19342848, 'steps': 37778, 'loss/train': 1.3758769035339355}
+02/26/2022 02:28:57 - INFO - codeparrot_training - Step 37779: {'lr': 7.579867946287733e-05, 'samples': 19343360, 'steps': 37779, 'loss/train': 1.672695279121399}
+02/26/2022 02:29:02 - INFO - codeparrot_training - Step 37780: {'lr': 7.57869436910891e-05, 'samples': 19343872, 'steps': 37780, 'loss/train': 1.244953989982605}
+02/26/2022 02:29:06 - INFO - codeparrot_training - Step 37781: {'lr': 7.577520866557428e-05, 'samples': 19344384, 'steps': 37781, 'loss/train': 2.3688089847564697}
+02/26/2022 02:29:12 - INFO - codeparrot_training - Step 37782: {'lr': 7.57634743863831e-05, 'samples': 19344896, 'steps': 37782, 'loss/train': 1.2336266040802002}
+02/26/2022 02:29:16 - INFO - codeparrot_training - Step 37783: {'lr': 7.57517408535659e-05, 'samples': 19345408, 'steps': 37783, 'loss/train': 1.7861788272857666}
+02/26/2022 02:29:21 - INFO - codeparrot_training - Step 37784: {'lr': 7.574000806717294e-05, 'samples': 19345920, 'steps': 37784, 'loss/train': 2.7549006938934326}
+02/26/2022 02:29:25 - INFO - codeparrot_training - Step 37785: {'lr': 7.572827602725443e-05, 'samples': 19346432, 'steps': 37785, 'loss/train': 2.124006748199463}
+02/26/2022 02:29:30 - INFO - codeparrot_training - Step 37786: {'lr': 7.571654473386067e-05, 'samples': 19346944, 'steps': 37786, 'loss/train': 2.021242141723633}
+02/26/2022 02:29:34 - INFO - codeparrot_training - Step 37787: {'lr': 7.570481418704181e-05, 'samples': 19347456, 'steps': 37787, 'loss/train': 2.0368247032165527}
+02/26/2022 02:29:40 - INFO - codeparrot_training - Step 37788: {'lr': 7.569308438684824e-05, 'samples': 19347968, 'steps': 37788, 'loss/train': 2.628750801086426}
+02/26/2022 02:29:43 - INFO - codeparrot_training - Step 37789: {'lr': 7.568135533333018e-05, 'samples': 19348480, 'steps': 37789, 'loss/train': 1.37528395652771}
+02/26/2022 02:29:49 - INFO - codeparrot_training - Step 37790: {'lr': 7.566962702653781e-05, 'samples': 19348992, 'steps': 37790, 'loss/train': 1.5383455753326416}
+02/26/2022 02:29:52 - INFO - codeparrot_training - Step 37791: {'lr': 7.565789946652135e-05, 'samples': 19349504, 'steps': 37791, 'loss/train': 1.5342360734939575}
+02/26/2022 02:29:58 - INFO - codeparrot_training - Step 37792: {'lr': 7.564617265333115e-05, 'samples': 19350016, 'steps': 37792, 'loss/train': 2.489480972290039}
+02/26/2022 02:30:02 - INFO - codeparrot_training - Step 37793: {'lr': 7.563444658701741e-05, 'samples': 19350528, 'steps': 37793, 'loss/train': 1.9754176139831543}
+02/26/2022 02:30:07 - INFO - codeparrot_training - Step 37794: {'lr': 7.562272126763029e-05, 'samples': 19351040, 'steps': 37794, 'loss/train': 2.429378032684326}
+02/26/2022 02:30:11 - INFO - codeparrot_training - Step 37795: {'lr': 7.561099669522001e-05, 'samples': 19351552, 'steps': 37795, 'loss/train': 2.4170801639556885}
+02/26/2022 02:30:16 - INFO - codeparrot_training - Step 37796: {'lr': 7.559927286983692e-05, 'samples': 19352064, 'steps': 37796, 'loss/train': 2.2970399856567383}
+02/26/2022 02:30:20 - INFO - codeparrot_training - Step 37797: {'lr': 7.558754979153107e-05, 'samples': 19352576, 'steps': 37797, 'loss/train': 1.7543789148330688}
+02/26/2022 02:30:25 - INFO - codeparrot_training - Step 37798: {'lr': 7.557582746035297e-05, 'samples': 19353088, 'steps': 37798, 'loss/train': 1.4067046642303467}
+02/26/2022 02:30:29 - INFO - codeparrot_training - Step 37799: {'lr': 7.556410587635245e-05, 'samples': 19353600, 'steps': 37799, 'loss/train': 1.6838634014129639}
+02/26/2022 02:30:34 - INFO - codeparrot_training - Step 37800: {'lr': 7.555238503958001e-05, 'samples': 19354112, 'steps': 37800, 'loss/train': 0.7755517363548279}
+02/26/2022 02:30:38 - INFO - codeparrot_training - Step 37801: {'lr': 7.554066495008566e-05, 'samples': 19354624, 'steps': 37801, 'loss/train': 1.8027853965759277}
+02/26/2022 02:30:43 - INFO - codeparrot_training - Step 37802: {'lr': 7.552894560791987e-05, 'samples': 19355136, 'steps': 37802, 'loss/train': 1.1575514078140259}
+02/26/2022 02:30:47 - INFO - codeparrot_training - Step 37803: {'lr': 7.551722701313251e-05, 'samples': 19355648, 'steps': 37803, 'loss/train': 2.10512375831604}
+02/26/2022 02:30:53 - INFO - codeparrot_training - Step 37804: {'lr': 7.550550916577404e-05, 'samples': 19356160, 'steps': 37804, 'loss/train': 1.02147376537323}
+02/26/2022 02:30:56 - INFO - codeparrot_training - Step 37805: {'lr': 7.549379206589446e-05, 'samples': 19356672, 'steps': 37805, 'loss/train': 0.7600228190422058}
+02/26/2022 02:31:02 - INFO - codeparrot_training - Step 37806: {'lr': 7.548207571354426e-05, 'samples': 19357184, 'steps': 37806, 'loss/train': 1.8661831617355347}
+02/26/2022 02:31:05 - INFO - codeparrot_training - Step 37807: {'lr': 7.547036010877323e-05, 'samples': 19357696, 'steps': 37807, 'loss/train': 1.5922367572784424}
+02/26/2022 02:31:11 - INFO - codeparrot_training - Step 37808: {'lr': 7.545864525163188e-05, 'samples': 19358208, 'steps': 37808, 'loss/train': 2.172541379928589}
+02/26/2022 02:31:14 - INFO - codeparrot_training - Step 37809: {'lr': 7.544693114217016e-05, 'samples': 19358720, 'steps': 37809, 'loss/train': 1.2649165391921997}
+02/26/2022 02:31:20 - INFO - codeparrot_training - Step 37810: {'lr': 7.543521778043843e-05, 'samples': 19359232, 'steps': 37810, 'loss/train': 1.7339946031570435}
+02/26/2022 02:31:23 - INFO - codeparrot_training - Step 37811: {'lr': 7.542350516648683e-05, 'samples': 19359744, 'steps': 37811, 'loss/train': 1.2232105731964111}
+02/26/2022 02:31:29 - INFO - codeparrot_training - Step 37812: {'lr': 7.541179330036549e-05, 'samples': 19360256, 'steps': 37812, 'loss/train': 0.4428284168243408}
+02/26/2022 02:31:32 - INFO - codeparrot_training - Step 37813: {'lr': 7.540008218212448e-05, 'samples': 19360768, 'steps': 37813, 'loss/train': 2.109645128250122}
+02/26/2022 02:31:38 - INFO - codeparrot_training - Step 37814: {'lr': 7.538837181181418e-05, 'samples': 19361280, 'steps': 37814, 'loss/train': 1.935546636581421}
+02/26/2022 02:31:41 - INFO - codeparrot_training - Step 37815: {'lr': 7.537666218948466e-05, 'samples': 19361792, 'steps': 37815, 'loss/train': 2.405400514602661}
+02/26/2022 02:31:47 - INFO - codeparrot_training - Step 37816: {'lr': 7.536495331518606e-05, 'samples': 19362304, 'steps': 37816, 'loss/train': 1.3233942985534668}
+02/26/2022 02:31:50 - INFO - codeparrot_training - Step 37817: {'lr': 7.535324518896846e-05, 'samples': 19362816, 'steps': 37817, 'loss/train': 2.205601453781128}
+02/26/2022 02:31:57 - INFO - codeparrot_training - Step 37818: {'lr': 7.534153781088218e-05, 'samples': 19363328, 'steps': 37818, 'loss/train': 2.071200132369995}
+02/26/2022 02:32:00 - INFO - codeparrot_training - Step 37819: {'lr': 7.53298311809773e-05, 'samples': 19363840, 'steps': 37819, 'loss/train': 2.0881972312927246}
+02/26/2022 02:32:06 - INFO - codeparrot_training - Step 37820: {'lr': 7.531812529930399e-05, 'samples': 19364352, 'steps': 37820, 'loss/train': 2.171382188796997}
+02/26/2022 02:32:09 - INFO - codeparrot_training - Step 37821: {'lr': 7.530642016591225e-05, 'samples': 19364864, 'steps': 37821, 'loss/train': 0.10827948153018951}
+02/26/2022 02:32:15 - INFO - codeparrot_training - Step 37822: {'lr': 7.52947157808524e-05, 'samples': 19365376, 'steps': 37822, 'loss/train': 1.5215182304382324}
+02/26/2022 02:32:18 - INFO - codeparrot_training - Step 37823: {'lr': 7.528301214417452e-05, 'samples': 19365888, 'steps': 37823, 'loss/train': 1.8233108520507812}
+02/26/2022 02:32:24 - INFO - codeparrot_training - Step 37824: {'lr': 7.527130925592876e-05, 'samples': 19366400, 'steps': 37824, 'loss/train': 1.275500774383545}
+02/26/2022 02:32:27 - INFO - codeparrot_training - Step 37825: {'lr': 7.525960711616514e-05, 'samples': 19366912, 'steps': 37825, 'loss/train': 1.5067020654678345}
+02/26/2022 02:32:33 - INFO - codeparrot_training - Step 37826: {'lr': 7.524790572493396e-05, 'samples': 19367424, 'steps': 37826, 'loss/train': 0.7007200121879578}
+02/26/2022 02:32:36 - INFO - codeparrot_training - Step 37827: {'lr': 7.523620508228524e-05, 'samples': 19367936, 'steps': 37827, 'loss/train': 8.622401237487793}
+02/26/2022 02:32:43 - INFO - codeparrot_training - Step 37828: {'lr': 7.522450518826912e-05, 'samples': 19368448, 'steps': 37828, 'loss/train': 1.0185248851776123}
+02/26/2022 02:32:46 - INFO - codeparrot_training - Step 37829: {'lr': 7.521280604293568e-05, 'samples': 19368960, 'steps': 37829, 'loss/train': 1.9488813877105713}
+02/26/2022 02:32:52 - INFO - codeparrot_training - Step 37830: {'lr': 7.520110764633511e-05, 'samples': 19369472, 'steps': 37830, 'loss/train': 0.44856682419776917}
+02/26/2022 02:32:55 - INFO - codeparrot_training - Step 37831: {'lr': 7.518940999851753e-05, 'samples': 19369984, 'steps': 37831, 'loss/train': 1.596879243850708}
+02/26/2022 02:33:01 - INFO - codeparrot_training - Step 37832: {'lr': 7.517771309953292e-05, 'samples': 19370496, 'steps': 37832, 'loss/train': 2.1883862018585205}
+02/26/2022 02:33:04 - INFO - codeparrot_training - Step 37833: {'lr': 7.516601694943162e-05, 'samples': 19371008, 'steps': 37833, 'loss/train': 1.9536049365997314}
+02/26/2022 02:33:10 - INFO - codeparrot_training - Step 37834: {'lr': 7.515432154826346e-05, 'samples': 19371520, 'steps': 37834, 'loss/train': 2.092209577560425}
+02/26/2022 02:33:13 - INFO - codeparrot_training - Step 37835: {'lr': 7.51426268960787e-05, 'samples': 19372032, 'steps': 37835, 'loss/train': 1.6627986431121826}
+02/26/2022 02:33:19 - INFO - codeparrot_training - Step 37836: {'lr': 7.513093299292736e-05, 'samples': 19372544, 'steps': 37836, 'loss/train': 2.3882265090942383}
+02/26/2022 02:33:22 - INFO - codeparrot_training - Step 37837: {'lr': 7.511923983885971e-05, 'samples': 19373056, 'steps': 37837, 'loss/train': 0.8633599877357483}
+02/26/2022 02:33:28 - INFO - codeparrot_training - Step 37838: {'lr': 7.510754743392555e-05, 'samples': 19373568, 'steps': 37838, 'loss/train': 1.60392427444458}
+02/26/2022 02:33:31 - INFO - codeparrot_training - Step 37839: {'lr': 7.509585577817524e-05, 'samples': 19374080, 'steps': 37839, 'loss/train': 1.9865578413009644}
+02/26/2022 02:33:38 - INFO - codeparrot_training - Step 37840: {'lr': 7.508416487165862e-05, 'samples': 19374592, 'steps': 37840, 'loss/train': 2.0998520851135254}
+02/26/2022 02:33:41 - INFO - codeparrot_training - Step 37841: {'lr': 7.507247471442607e-05, 'samples': 19375104, 'steps': 37841, 'loss/train': 2.170912981033325}
+02/26/2022 02:33:47 - INFO - codeparrot_training - Step 37842: {'lr': 7.506078530652736e-05, 'samples': 19375616, 'steps': 37842, 'loss/train': 1.8825217485427856}
+02/26/2022 02:33:50 - INFO - codeparrot_training - Step 37843: {'lr': 7.504909664801273e-05, 'samples': 19376128, 'steps': 37843, 'loss/train': 1.8935712575912476}
+02/26/2022 02:33:56 - INFO - codeparrot_training - Step 37844: {'lr': 7.503740873893217e-05, 'samples': 19376640, 'steps': 37844, 'loss/train': 1.1253705024719238}
+02/26/2022 02:33:59 - INFO - codeparrot_training - Step 37845: {'lr': 7.502572157933585e-05, 'samples': 19377152, 'steps': 37845, 'loss/train': 0.06850094348192215}
+02/26/2022 02:34:05 - INFO - codeparrot_training - Step 37846: {'lr': 7.50140351692738e-05, 'samples': 19377664, 'steps': 37846, 'loss/train': 0.3980892300605774}
+02/26/2022 02:34:08 - INFO - codeparrot_training - Step 37847: {'lr': 7.500234950879603e-05, 'samples': 19378176, 'steps': 37847, 'loss/train': 0.7753595113754272}
+02/26/2022 02:34:14 - INFO - codeparrot_training - Step 37848: {'lr': 7.499066459795258e-05, 'samples': 19378688, 'steps': 37848, 'loss/train': 0.8248132467269897}
+02/26/2022 02:34:17 - INFO - codeparrot_training - Step 37849: {'lr': 7.49789804367936e-05, 'samples': 19379200, 'steps': 37849, 'loss/train': 1.2214226722717285}
+02/26/2022 02:34:24 - INFO - codeparrot_training - Step 37850: {'lr': 7.496729702536912e-05, 'samples': 19379712, 'steps': 37850, 'loss/train': 1.7030028104782104}
+02/26/2022 02:34:27 - INFO - codeparrot_training - Step 37851: {'lr': 7.495561436372916e-05, 'samples': 19380224, 'steps': 37851, 'loss/train': 2.2792766094207764}
+02/26/2022 02:34:32 - INFO - codeparrot_training - Step 37852: {'lr': 7.494393245192366e-05, 'samples': 19380736, 'steps': 37852, 'loss/train': 2.000772714614868}
+02/26/2022 02:34:36 - INFO - codeparrot_training - Step 37853: {'lr': 7.493225129000286e-05, 'samples': 19381248, 'steps': 37853, 'loss/train': 1.6239792108535767}
+02/26/2022 02:34:41 - INFO - codeparrot_training - Step 37854: {'lr': 7.492057087801671e-05, 'samples': 19381760, 'steps': 37854, 'loss/train': 1.1903672218322754}
+02/26/2022 02:34:45 - INFO - codeparrot_training - Step 37855: {'lr': 7.490889121601523e-05, 'samples': 19382272, 'steps': 37855, 'loss/train': 1.0262376070022583}
+02/26/2022 02:34:50 - INFO - codeparrot_training - Step 37856: {'lr': 7.489721230404842e-05, 'samples': 19382784, 'steps': 37856, 'loss/train': 0.7008168697357178}
+02/26/2022 02:34:54 - INFO - codeparrot_training - Step 37857: {'lr': 7.48855341421664e-05, 'samples': 19383296, 'steps': 37857, 'loss/train': 1.4924386739730835}
+02/26/2022 02:34:59 - INFO - codeparrot_training - Step 37858: {'lr': 7.487385673041904e-05, 'samples': 19383808, 'steps': 37858, 'loss/train': 1.9464226961135864}
+02/26/2022 02:35:03 - INFO - codeparrot_training - Step 37859: {'lr': 7.486218006885668e-05, 'samples': 19384320, 'steps': 37859, 'loss/train': 1.0982117652893066}
+02/26/2022 02:35:08 - INFO - codeparrot_training - Step 37860: {'lr': 7.485050415752892e-05, 'samples': 19384832, 'steps': 37860, 'loss/train': 2.342677593231201}
+02/26/2022 02:35:12 - INFO - codeparrot_training - Step 37861: {'lr': 7.483882899648611e-05, 'samples': 19385344, 'steps': 37861, 'loss/train': 1.7296215295791626}
+02/26/2022 02:35:17 - INFO - codeparrot_training - Step 37862: {'lr': 7.482715458577802e-05, 'samples': 19385856, 'steps': 37862, 'loss/train': 1.3387377262115479}
+02/26/2022 02:35:21 - INFO - codeparrot_training - Step 37863: {'lr': 7.481548092545496e-05, 'samples': 19386368, 'steps': 37863, 'loss/train': 2.5989415645599365}
+02/26/2022 02:35:27 - INFO - codeparrot_training - Step 37864: {'lr': 7.480380801556657e-05, 'samples': 19386880, 'steps': 37864, 'loss/train': 2.5908684730529785}
+02/26/2022 02:35:30 - INFO - codeparrot_training - Step 37865: {'lr': 7.479213585616315e-05, 'samples': 19387392, 'steps': 37865, 'loss/train': 2.1272919178009033}
+02/26/2022 02:35:36 - INFO - codeparrot_training - Step 37866: {'lr': 7.478046444729456e-05, 'samples': 19387904, 'steps': 37866, 'loss/train': 1.7978525161743164}
+02/26/2022 02:35:40 - INFO - codeparrot_training - Step 37867: {'lr': 7.476879378901075e-05, 'samples': 19388416, 'steps': 37867, 'loss/train': 3.0519626140594482}
+02/26/2022 02:35:45 - INFO - codeparrot_training - Step 37868: {'lr': 7.475712388136185e-05, 'samples': 19388928, 'steps': 37868, 'loss/train': 1.9512221813201904}
+02/26/2022 02:35:49 - INFO - codeparrot_training - Step 37869: {'lr': 7.474545472439781e-05, 'samples': 19389440, 'steps': 37869, 'loss/train': 0.9405859112739563}
+02/26/2022 02:35:54 - INFO - codeparrot_training - Step 37870: {'lr': 7.473378631816857e-05, 'samples': 19389952, 'steps': 37870, 'loss/train': 3.8095788955688477}
+02/26/2022 02:35:58 - INFO - codeparrot_training - Step 37871: {'lr': 7.472211866272405e-05, 'samples': 19390464, 'steps': 37871, 'loss/train': 2.5191195011138916}
+02/26/2022 02:36:03 - INFO - codeparrot_training - Step 37872: {'lr': 7.471045175811442e-05, 'samples': 19390976, 'steps': 37872, 'loss/train': 1.3548946380615234}
+02/26/2022 02:36:07 - INFO - codeparrot_training - Step 37873: {'lr': 7.469878560438956e-05, 'samples': 19391488, 'steps': 37873, 'loss/train': 1.8850998878479004}
+02/26/2022 02:36:13 - INFO - codeparrot_training - Step 37874: {'lr': 7.46871202015994e-05, 'samples': 19392000, 'steps': 37874, 'loss/train': 1.2202781438827515}
+02/26/2022 02:36:16 - INFO - codeparrot_training - Step 37875: {'lr': 7.46754555497939e-05, 'samples': 19392512, 'steps': 37875, 'loss/train': 1.0568088293075562}
+02/26/2022 02:36:22 - INFO - codeparrot_training - Step 37876: {'lr': 7.466379164902315e-05, 'samples': 19393024, 'steps': 37876, 'loss/train': 1.5135680437088013}
+02/26/2022 02:36:25 - INFO - codeparrot_training - Step 37877: {'lr': 7.465212849933706e-05, 'samples': 19393536, 'steps': 37877, 'loss/train': 2.227048397064209}
+02/26/2022 02:36:31 - INFO - codeparrot_training - Step 37878: {'lr': 7.464046610078553e-05, 'samples': 19394048, 'steps': 37878, 'loss/train': 1.6113837957382202}
+02/26/2022 02:36:35 - INFO - codeparrot_training - Step 37879: {'lr': 7.462880445341852e-05, 'samples': 19394560, 'steps': 37879, 'loss/train': 0.5117477774620056}
+02/26/2022 02:36:40 - INFO - codeparrot_training - Step 37880: {'lr': 7.461714355728607e-05, 'samples': 19395072, 'steps': 37880, 'loss/train': 1.7949235439300537}
+02/26/2022 02:36:44 - INFO - codeparrot_training - Step 37881: {'lr': 7.460548341243811e-05, 'samples': 19395584, 'steps': 37881, 'loss/train': 0.3060033917427063}
+02/26/2022 02:36:49 - INFO - codeparrot_training - Step 37882: {'lr': 7.459382401892456e-05, 'samples': 19396096, 'steps': 37882, 'loss/train': 0.9207753539085388}
+02/26/2022 02:36:52 - INFO - codeparrot_training - Step 37883: {'lr': 7.458216537679529e-05, 'samples': 19396608, 'steps': 37883, 'loss/train': 2.17743182182312}
+02/26/2022 02:36:58 - INFO - codeparrot_training - Step 37884: {'lr': 7.45705074861004e-05, 'samples': 19397120, 'steps': 37884, 'loss/train': 2.0331780910491943}
+02/26/2022 02:37:01 - INFO - codeparrot_training - Step 37885: {'lr': 7.455885034688974e-05, 'samples': 19397632, 'steps': 37885, 'loss/train': 1.174649715423584}
+02/26/2022 02:37:08 - INFO - codeparrot_training - Step 37886: {'lr': 7.454719395921328e-05, 'samples': 19398144, 'steps': 37886, 'loss/train': 0.8099073171615601}
+02/26/2022 02:37:11 - INFO - codeparrot_training - Step 37887: {'lr': 7.453553832312085e-05, 'samples': 19398656, 'steps': 37887, 'loss/train': 1.5410692691802979}
+02/26/2022 02:37:17 - INFO - codeparrot_training - Step 37888: {'lr': 7.452388343866251e-05, 'samples': 19399168, 'steps': 37888, 'loss/train': 1.5804073810577393}
+02/26/2022 02:37:20 - INFO - codeparrot_training - Step 37889: {'lr': 7.451222930588817e-05, 'samples': 19399680, 'steps': 37889, 'loss/train': 0.32616671919822693}
+02/26/2022 02:37:26 - INFO - codeparrot_training - Step 37890: {'lr': 7.450057592484768e-05, 'samples': 19400192, 'steps': 37890, 'loss/train': 2.31925630569458}
+02/26/2022 02:37:29 - INFO - codeparrot_training - Step 37891: {'lr': 7.448892329559092e-05, 'samples': 19400704, 'steps': 37891, 'loss/train': 0.8049443364143372}
+02/26/2022 02:37:35 - INFO - codeparrot_training - Step 37892: {'lr': 7.447727141816798e-05, 'samples': 19401216, 'steps': 37892, 'loss/train': 1.657503604888916}
+02/26/2022 02:37:38 - INFO - codeparrot_training - Step 37893: {'lr': 7.44656202926286e-05, 'samples': 19401728, 'steps': 37893, 'loss/train': 1.398797869682312}
+02/26/2022 02:37:44 - INFO - codeparrot_training - Step 37894: {'lr': 7.445396991902289e-05, 'samples': 19402240, 'steps': 37894, 'loss/train': 0.8342295289039612}
+02/26/2022 02:37:50 - INFO - codeparrot_training - Step 37895: {'lr': 7.44423202974005e-05, 'samples': 19402752, 'steps': 37895, 'loss/train': 1.3070570230484009}
+02/26/2022 02:37:53 - INFO - codeparrot_training - Step 37896: {'lr': 7.443067142781155e-05, 'samples': 19403264, 'steps': 37896, 'loss/train': 0.8890364766120911}
+02/26/2022 02:37:59 - INFO - codeparrot_training - Step 37897: {'lr': 7.441902331030576e-05, 'samples': 19403776, 'steps': 37897, 'loss/train': 3.535931348800659}
+02/26/2022 02:38:02 - INFO - codeparrot_training - Step 37898: {'lr': 7.440737594493327e-05, 'samples': 19404288, 'steps': 37898, 'loss/train': 1.6163150072097778}
+02/26/2022 02:38:06 - INFO - codeparrot_training - Step 37899: {'lr': 7.439572933174366e-05, 'samples': 19404800, 'steps': 37899, 'loss/train': 2.1310107707977295}
+02/26/2022 02:38:11 - INFO - codeparrot_training - Step 37900: {'lr': 7.43840834707871e-05, 'samples': 19405312, 'steps': 37900, 'loss/train': 0.8816977739334106}
+02/26/2022 02:38:17 - INFO - codeparrot_training - Step 37901: {'lr': 7.437243836211324e-05, 'samples': 19405824, 'steps': 37901, 'loss/train': 2.5828487873077393}
+02/26/2022 02:38:21 - INFO - codeparrot_training - Step 37902: {'lr': 7.436079400577228e-05, 'samples': 19406336, 'steps': 37902, 'loss/train': 1.9444694519042969}
+02/26/2022 02:38:24 - INFO - codeparrot_training - Step 37903: {'lr': 7.434915040181373e-05, 'samples': 19406848, 'steps': 37903, 'loss/train': 1.9368579387664795}
+02/26/2022 02:38:30 - INFO - codeparrot_training - Step 37904: {'lr': 7.433750755028773e-05, 'samples': 19407360, 'steps': 37904, 'loss/train': 1.4657213687896729}
+02/26/2022 02:38:33 - INFO - codeparrot_training - Step 37905: {'lr': 7.432586545124399e-05, 'samples': 19407872, 'steps': 37905, 'loss/train': 2.437809705734253}
+02/26/2022 02:38:38 - INFO - codeparrot_training - Step 37906: {'lr': 7.431422410473254e-05, 'samples': 19408384, 'steps': 37906, 'loss/train': 2.106069803237915}
+02/26/2022 02:38:42 - INFO - codeparrot_training - Step 37907: {'lr': 7.430258351080316e-05, 'samples': 19408896, 'steps': 37907, 'loss/train': 0.8649512529373169}
+02/26/2022 02:38:47 - INFO - codeparrot_training - Step 37908: {'lr': 7.429094366950573e-05, 'samples': 19409408, 'steps': 37908, 'loss/train': 0.7029323577880859}
+02/26/2022 02:38:51 - INFO - codeparrot_training - Step 37909: {'lr': 7.427930458089002e-05, 'samples': 19409920, 'steps': 37909, 'loss/train': 2.547236680984497}
+02/26/2022 02:38:56 - INFO - codeparrot_training - Step 37910: {'lr': 7.426766624500602e-05, 'samples': 19410432, 'steps': 37910, 'loss/train': 1.298637866973877}
+02/26/2022 02:39:00 - INFO - codeparrot_training - Step 37911: {'lr': 7.42560286619036e-05, 'samples': 19410944, 'steps': 37911, 'loss/train': 3.5130388736724854}
+02/26/2022 02:39:07 - INFO - codeparrot_training - Step 37912: {'lr': 7.42443918316325e-05, 'samples': 19411456, 'steps': 37912, 'loss/train': 1.2513548135757446}
+02/26/2022 02:39:10 - INFO - codeparrot_training - Step 37913: {'lr': 7.423275575424262e-05, 'samples': 19411968, 'steps': 37913, 'loss/train': 1.935681700706482}
+02/26/2022 02:39:16 - INFO - codeparrot_training - Step 37914: {'lr': 7.422112042978373e-05, 'samples': 19412480, 'steps': 37914, 'loss/train': 0.6941214799880981}
+02/26/2022 02:39:19 - INFO - codeparrot_training - Step 37915: {'lr': 7.420948585830584e-05, 'samples': 19412992, 'steps': 37915, 'loss/train': 2.090805768966675}
+02/26/2022 02:39:25 - INFO - codeparrot_training - Step 37916: {'lr': 7.419785203985868e-05, 'samples': 19413504, 'steps': 37916, 'loss/train': 3.969656229019165}
+02/26/2022 02:39:28 - INFO - codeparrot_training - Step 37917: {'lr': 7.41862189744921e-05, 'samples': 19414016, 'steps': 37917, 'loss/train': 0.5874213576316833}
+02/26/2022 02:39:34 - INFO - codeparrot_training - Step 37918: {'lr': 7.417458666225586e-05, 'samples': 19414528, 'steps': 37918, 'loss/train': 1.4047911167144775}
+02/26/2022 02:39:37 - INFO - codeparrot_training - Step 37919: {'lr': 7.416295510319992e-05, 'samples': 19415040, 'steps': 37919, 'loss/train': 1.738824725151062}
+02/26/2022 02:39:43 - INFO - codeparrot_training - Step 37920: {'lr': 7.415132429737406e-05, 'samples': 19415552, 'steps': 37920, 'loss/train': 2.721315622329712}
+02/26/2022 02:39:46 - INFO - codeparrot_training - Step 37921: {'lr': 7.413969424482808e-05, 'samples': 19416064, 'steps': 37921, 'loss/train': 1.6940052509307861}
+02/26/2022 02:39:53 - INFO - codeparrot_training - Step 37922: {'lr': 7.412806494561175e-05, 'samples': 19416576, 'steps': 37922, 'loss/train': 0.5788756012916565}
+02/26/2022 02:39:56 - INFO - codeparrot_training - Step 37923: {'lr': 7.4116436399775e-05, 'samples': 19417088, 'steps': 37923, 'loss/train': 1.7050844430923462}
+02/26/2022 02:40:02 - INFO - codeparrot_training - Step 37924: {'lr': 7.41048086073676e-05, 'samples': 19417600, 'steps': 37924, 'loss/train': 1.9376347064971924}
+02/26/2022 02:40:05 - INFO - codeparrot_training - Step 37925: {'lr': 7.409318156843933e-05, 'samples': 19418112, 'steps': 37925, 'loss/train': 1.453546166419983}
+02/26/2022 02:40:11 - INFO - codeparrot_training - Step 37926: {'lr': 7.408155528303995e-05, 'samples': 19418624, 'steps': 37926, 'loss/train': 1.5567615032196045}
+02/26/2022 02:40:14 - INFO - codeparrot_training - Step 37927: {'lr': 7.40699297512194e-05, 'samples': 19419136, 'steps': 37927, 'loss/train': 1.801559567451477}
+02/26/2022 02:40:20 - INFO - codeparrot_training - Step 37928: {'lr': 7.405830497302732e-05, 'samples': 19419648, 'steps': 37928, 'loss/train': 1.6155083179473877}
+02/26/2022 02:40:24 - INFO - codeparrot_training - Step 37929: {'lr': 7.404668094851378e-05, 'samples': 19420160, 'steps': 37929, 'loss/train': 1.728258490562439}
+02/26/2022 02:40:29 - INFO - codeparrot_training - Step 37930: {'lr': 7.40350576777282e-05, 'samples': 19420672, 'steps': 37930, 'loss/train': 1.5946191549301147}
+02/26/2022 02:40:33 - INFO - codeparrot_training - Step 37931: {'lr': 7.402343516072063e-05, 'samples': 19421184, 'steps': 37931, 'loss/train': 1.9840030670166016}
+02/26/2022 02:40:39 - INFO - codeparrot_training - Step 37932: {'lr': 7.401181339754074e-05, 'samples': 19421696, 'steps': 37932, 'loss/train': 0.9495724439620972}
+02/26/2022 02:40:42 - INFO - codeparrot_training - Step 37933: {'lr': 7.400019238823849e-05, 'samples': 19422208, 'steps': 37933, 'loss/train': 1.8813179731369019}
+02/26/2022 02:40:48 - INFO - codeparrot_training - Step 37934: {'lr': 7.39885721328634e-05, 'samples': 19422720, 'steps': 37934, 'loss/train': 0.4933006465435028}
+02/26/2022 02:40:51 - INFO - codeparrot_training - Step 37935: {'lr': 7.397695263146542e-05, 'samples': 19423232, 'steps': 37935, 'loss/train': 2.935905933380127}
+02/26/2022 02:40:57 - INFO - codeparrot_training - Step 37936: {'lr': 7.396533388409421e-05, 'samples': 19423744, 'steps': 37936, 'loss/train': 1.4275503158569336}
+02/26/2022 02:41:00 - INFO - codeparrot_training - Step 37937: {'lr': 7.395371589079977e-05, 'samples': 19424256, 'steps': 37937, 'loss/train': 2.7695472240448}
+02/26/2022 02:41:06 - INFO - codeparrot_training - Step 37938: {'lr': 7.394209865163157e-05, 'samples': 19424768, 'steps': 37938, 'loss/train': 2.240042209625244}
+02/26/2022 02:41:09 - INFO - codeparrot_training - Step 37939: {'lr': 7.393048216663959e-05, 'samples': 19425280, 'steps': 37939, 'loss/train': 0.4156077802181244}
+02/26/2022 02:41:14 - INFO - codeparrot_training - Step 37940: {'lr': 7.391886643587343e-05, 'samples': 19425792, 'steps': 37940, 'loss/train': 1.6348804235458374}
+02/26/2022 02:41:21 - INFO - codeparrot_training - Step 37941: {'lr': 7.390725145938302e-05, 'samples': 19426304, 'steps': 37941, 'loss/train': 2.4212088584899902}
+02/26/2022 02:41:24 - INFO - codeparrot_training - Step 37942: {'lr': 7.389563723721801e-05, 'samples': 19426816, 'steps': 37942, 'loss/train': 1.5889395475387573}
+02/26/2022 02:41:29 - INFO - codeparrot_training - Step 37943: {'lr': 7.388402376942818e-05, 'samples': 19427328, 'steps': 37943, 'loss/train': 1.5325666666030884}
+02/26/2022 02:41:33 - INFO - codeparrot_training - Step 37944: {'lr': 7.387241105606321e-05, 'samples': 19427840, 'steps': 37944, 'loss/train': 1.6129833459854126}
+02/26/2022 02:41:38 - INFO - codeparrot_training - Step 37945: {'lr': 7.386079909717297e-05, 'samples': 19428352, 'steps': 37945, 'loss/train': 2.318693161010742}
+02/26/2022 02:41:42 - INFO - codeparrot_training - Step 37946: {'lr': 7.384918789280715e-05, 'samples': 19428864, 'steps': 37946, 'loss/train': 1.3313490152359009}
+02/26/2022 02:41:48 - INFO - codeparrot_training - Step 37947: {'lr': 7.383757744301545e-05, 'samples': 19429376, 'steps': 37947, 'loss/train': 1.7612372636795044}
+02/26/2022 02:41:51 - INFO - codeparrot_training - Step 37948: {'lr': 7.382596774784756e-05, 'samples': 19429888, 'steps': 37948, 'loss/train': 2.1369247436523438}
+02/26/2022 02:41:57 - INFO - codeparrot_training - Step 37949: {'lr': 7.381435880735334e-05, 'samples': 19430400, 'steps': 37949, 'loss/train': 1.8385356664657593}
+02/26/2022 02:42:00 - INFO - codeparrot_training - Step 37950: {'lr': 7.38027506215825e-05, 'samples': 19430912, 'steps': 37950, 'loss/train': 2.474994421005249}
+02/26/2022 02:42:06 - INFO - codeparrot_training - Step 37951: {'lr': 7.379114319058471e-05, 'samples': 19431424, 'steps': 37951, 'loss/train': 0.9505074620246887}
+02/26/2022 02:42:09 - INFO - codeparrot_training - Step 37952: {'lr': 7.377953651440964e-05, 'samples': 19431936, 'steps': 37952, 'loss/train': 1.0113667249679565}
+02/26/2022 02:42:15 - INFO - codeparrot_training - Step 37953: {'lr': 7.376793059310713e-05, 'samples': 19432448, 'steps': 37953, 'loss/train': 1.8876501321792603}
+02/26/2022 02:42:18 - INFO - codeparrot_training - Step 37954: {'lr': 7.375632542672688e-05, 'samples': 19432960, 'steps': 37954, 'loss/train': 2.267726421356201}
+02/26/2022 02:42:24 - INFO - codeparrot_training - Step 37955: {'lr': 7.374472101531854e-05, 'samples': 19433472, 'steps': 37955, 'loss/train': 1.8480150699615479}
+02/26/2022 02:42:28 - INFO - codeparrot_training - Step 37956: {'lr': 7.373311735893178e-05, 'samples': 19433984, 'steps': 37956, 'loss/train': 2.2423174381256104}
+02/26/2022 02:42:34 - INFO - codeparrot_training - Step 37957: {'lr': 7.372151445761646e-05, 'samples': 19434496, 'steps': 37957, 'loss/train': 0.9533421993255615}
+02/26/2022 02:42:37 - INFO - codeparrot_training - Step 37958: {'lr': 7.370991231142222e-05, 'samples': 19435008, 'steps': 37958, 'loss/train': 1.7286713123321533}
+02/26/2022 02:42:41 - INFO - codeparrot_training - Step 37959: {'lr': 7.369831092039869e-05, 'samples': 19435520, 'steps': 37959, 'loss/train': 0.22948433458805084}
+02/26/2022 02:42:46 - INFO - codeparrot_training - Step 37960: {'lr': 7.368671028459564e-05, 'samples': 19436032, 'steps': 37960, 'loss/train': 2.0543406009674072}
+02/26/2022 02:42:52 - INFO - codeparrot_training - Step 37961: {'lr': 7.367511040406266e-05, 'samples': 19436544, 'steps': 37961, 'loss/train': 1.7658185958862305}
+02/26/2022 02:42:55 - INFO - codeparrot_training - Step 37962: {'lr': 7.366351127884957e-05, 'samples': 19437056, 'steps': 37962, 'loss/train': 0.878489077091217}
+02/26/2022 02:42:59 - INFO - codeparrot_training - Step 37963: {'lr': 7.365191290900603e-05, 'samples': 19437568, 'steps': 37963, 'loss/train': 2.6228792667388916}
+02/26/2022 02:43:04 - INFO - codeparrot_training - Step 37964: {'lr': 7.364031529458171e-05, 'samples': 19438080, 'steps': 37964, 'loss/train': 1.7497022151947021}
+02/26/2022 02:43:08 - INFO - codeparrot_training - Step 37965: {'lr': 7.362871843562616e-05, 'samples': 19438592, 'steps': 37965, 'loss/train': 2.4399523735046387}
+02/26/2022 02:43:13 - INFO - codeparrot_training - Step 37966: {'lr': 7.361712233218928e-05, 'samples': 19439104, 'steps': 37966, 'loss/train': 1.2031710147857666}
+02/26/2022 02:43:17 - INFO - codeparrot_training - Step 37967: {'lr': 7.360552698432055e-05, 'samples': 19439616, 'steps': 37967, 'loss/train': 0.023997321724891663}
+02/26/2022 02:43:23 - INFO - codeparrot_training - Step 37968: {'lr': 7.359393239206991e-05, 'samples': 19440128, 'steps': 37968, 'loss/train': 2.0471606254577637}
+02/26/2022 02:43:27 - INFO - codeparrot_training - Step 37969: {'lr': 7.358233855548668e-05, 'samples': 19440640, 'steps': 37969, 'loss/train': 2.9501149654388428}
+02/26/2022 02:43:32 - INFO - codeparrot_training - Step 37970: {'lr': 7.357074547462076e-05, 'samples': 19441152, 'steps': 37970, 'loss/train': 1.5002484321594238}
+02/26/2022 02:43:36 - INFO - codeparrot_training - Step 37971: {'lr': 7.355915314952169e-05, 'samples': 19441664, 'steps': 37971, 'loss/train': 1.643397569656372}
+02/26/2022 02:43:42 - INFO - codeparrot_training - Step 37972: {'lr': 7.354756158023931e-05, 'samples': 19442176, 'steps': 37972, 'loss/train': 0.3950980007648468}
+02/26/2022 02:43:45 - INFO - codeparrot_training - Step 37973: {'lr': 7.3535970766823e-05, 'samples': 19442688, 'steps': 37973, 'loss/train': 1.9400101900100708}
+02/26/2022 02:43:51 - INFO - codeparrot_training - Step 37974: {'lr': 7.352438070932262e-05, 'samples': 19443200, 'steps': 37974, 'loss/train': 2.783689260482788}
+02/26/2022 02:43:54 - INFO - codeparrot_training - Step 37975: {'lr': 7.351279140778772e-05, 'samples': 19443712, 'steps': 37975, 'loss/train': 1.078197956085205}
+02/26/2022 02:43:59 - INFO - codeparrot_training - Step 37976: {'lr': 7.350120286226803e-05, 'samples': 19444224, 'steps': 37976, 'loss/train': 2.2455835342407227}
+02/26/2022 02:44:03 - INFO - codeparrot_training - Step 37977: {'lr': 7.348961507281316e-05, 'samples': 19444736, 'steps': 37977, 'loss/train': 1.970812201499939}
+02/26/2022 02:44:09 - INFO - codeparrot_training - Step 37978: {'lr': 7.347802803947271e-05, 'samples': 19445248, 'steps': 37978, 'loss/train': 1.2487633228302002}
+02/26/2022 02:44:13 - INFO - codeparrot_training - Step 37979: {'lr': 7.346644176229625e-05, 'samples': 19445760, 'steps': 37979, 'loss/train': 1.3906623125076294}
+02/26/2022 02:44:18 - INFO - codeparrot_training - Step 37980: {'lr': 7.34548562413336e-05, 'samples': 19446272, 'steps': 37980, 'loss/train': 0.7225399017333984}
+02/26/2022 02:44:24 - INFO - codeparrot_training - Step 37981: {'lr': 7.344327147663427e-05, 'samples': 19446784, 'steps': 37981, 'loss/train': 2.2374472618103027}
+02/26/2022 02:44:27 - INFO - codeparrot_training - Step 37982: {'lr': 7.34316874682479e-05, 'samples': 19447296, 'steps': 37982, 'loss/train': 2.211420774459839}
+02/26/2022 02:44:33 - INFO - codeparrot_training - Step 37983: {'lr': 7.342010421622405e-05, 'samples': 19447808, 'steps': 37983, 'loss/train': 1.4731392860412598}
+02/26/2022 02:44:36 - INFO - codeparrot_training - Step 37984: {'lr': 7.34085217206125e-05, 'samples': 19448320, 'steps': 37984, 'loss/train': 1.5622638463974}
+02/26/2022 02:44:41 - INFO - codeparrot_training - Step 37985: {'lr': 7.339693998146274e-05, 'samples': 19448832, 'steps': 37985, 'loss/train': 1.8136826753616333}
+02/26/2022 02:44:45 - INFO - codeparrot_training - Step 37986: {'lr': 7.338535899882439e-05, 'samples': 19449344, 'steps': 37986, 'loss/train': 2.1253254413604736}
+02/26/2022 02:44:51 - INFO - codeparrot_training - Step 37987: {'lr': 7.337377877274703e-05, 'samples': 19449856, 'steps': 37987, 'loss/train': 0.3222852349281311}
+02/26/2022 02:44:54 - INFO - codeparrot_training - Step 37988: {'lr': 7.336219930328042e-05, 'samples': 19450368, 'steps': 37988, 'loss/train': 0.4808877110481262}
+02/26/2022 02:45:00 - INFO - codeparrot_training - Step 37989: {'lr': 7.335062059047396e-05, 'samples': 19450880, 'steps': 37989, 'loss/train': 0.1336582601070404}
+02/26/2022 02:45:04 - INFO - codeparrot_training - Step 37990: {'lr': 7.333904263437748e-05, 'samples': 19451392, 'steps': 37990, 'loss/train': 2.6639249324798584}
+02/26/2022 02:45:09 - INFO - codeparrot_training - Step 37991: {'lr': 7.332746543504032e-05, 'samples': 19451904, 'steps': 37991, 'loss/train': 1.6894506216049194}
+02/26/2022 02:45:13 - INFO - codeparrot_training - Step 37992: {'lr': 7.331588899251227e-05, 'samples': 19452416, 'steps': 37992, 'loss/train': 2.1091763973236084}
+02/26/2022 02:45:18 - INFO - codeparrot_training - Step 37993: {'lr': 7.330431330684285e-05, 'samples': 19452928, 'steps': 37993, 'loss/train': 2.9126014709472656}
+02/26/2022 02:45:22 - INFO - codeparrot_training - Step 37994: {'lr': 7.329273837808162e-05, 'samples': 19453440, 'steps': 37994, 'loss/train': 2.5056447982788086}
+02/26/2022 02:45:27 - INFO - codeparrot_training - Step 37995: {'lr': 7.328116420627811e-05, 'samples': 19453952, 'steps': 37995, 'loss/train': 0.940467119216919}
+02/26/2022 02:45:30 - INFO - codeparrot_training - Step 37996: {'lr': 7.32695907914821e-05, 'samples': 19454464, 'steps': 37996, 'loss/train': 0.4037728011608124}
+02/26/2022 02:45:36 - INFO - codeparrot_training - Step 37997: {'lr': 7.3258018133743e-05, 'samples': 19454976, 'steps': 37997, 'loss/train': 0.9487450122833252}
+02/26/2022 02:45:39 - INFO - codeparrot_training - Step 37998: {'lr': 7.324644623311046e-05, 'samples': 19455488, 'steps': 37998, 'loss/train': 0.07264846563339233}
+02/26/2022 02:45:45 - INFO - codeparrot_training - Step 37999: {'lr': 7.323487508963392e-05, 'samples': 19456000, 'steps': 37999, 'loss/train': 1.960349678993225}
+02/26/2022 02:45:45 - INFO - codeparrot_training - Evaluating and saving model checkpoint