diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -29476,3 +29476,1009 @@ Use FP16 precision: False
 02/25/2022 15:08:50 - INFO - codeparrot_training - Step 28998: {'lr': 0.00020125951603697535, 'samples': 14847488, 'steps': 28998, 'loss/train': 1.5860698223114014}
 02/25/2022 15:08:55 - INFO - codeparrot_training - Step 28999: {'lr': 0.0002012434676620429, 'samples': 14848000, 'steps': 28999, 'loss/train': 0.6781620979309082}
 02/25/2022 15:08:55 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/25/2022 15:09:13 - WARNING - huggingface_hub.repository - Several commits (29) will be pushed upstream.
+02/25/2022 15:09:13 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/25/2022 15:09:53 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   739c14e..95f22fa  floral-grass-11 -> floral-grass-11
+
+02/25/2022 15:09:58 - INFO - codeparrot_training - Step 29000: {'lr': 0.00020122741949596797, 'samples': 14848512, 'steps': 29000, 'loss/train': 0.9625966548919678}
+02/25/2022 15:10:03 - INFO - codeparrot_training - Step 29001: {'lr': 0.00020121137153881924, 'samples': 14849024, 'steps': 29001, 'loss/train': 3.0134713649749756}
+02/25/2022 15:10:07 - INFO - codeparrot_training - Step 29002: {'lr': 0.00020119532379066557, 'samples': 14849536, 'steps': 29002, 'loss/train': 1.2822824716567993}
+02/25/2022 15:10:13 - INFO - codeparrot_training - Step 29003: {'lr': 0.0002011792762515757, 'samples': 14850048, 'steps': 29003, 'loss/train': 2.368891477584839}
+02/25/2022 15:10:16 - INFO - codeparrot_training - Step 29004: {'lr': 0.0002011632289216182, 'samples': 14850560, 'steps': 29004, 'loss/train': 2.1708297729492188}
+02/25/2022 15:10:22 - INFO - codeparrot_training - Step 29005: {'lr': 0.00020114718180086195, 'samples': 14851072, 'steps': 29005, 'loss/train': 2.9230146408081055}
+02/25/2022 15:10:25 - INFO - codeparrot_training - Step 29006: {'lr': 0.0002011311348893756, 'samples': 14851584, 'steps': 29006, 'loss/train': 1.6136120557785034}
+02/25/2022 15:10:31 - INFO - codeparrot_training - Step 29007: {'lr': 0.00020111508818722814, 'samples': 14852096, 'steps': 29007, 'loss/train': 4.313667297363281}
+02/25/2022 15:10:35 - INFO - codeparrot_training - Step 29008: {'lr': 0.00020109904169448797, 'samples': 14852608, 'steps': 29008, 'loss/train': 0.40597519278526306}
+02/25/2022 15:10:40 - INFO - codeparrot_training - Step 29009: {'lr': 0.00020108299541122398, 'samples': 14853120, 'steps': 29009, 'loss/train': 1.30682373046875}
+02/25/2022 15:10:44 - INFO - codeparrot_training - Step 29010: {'lr': 0.00020106694933750492, 'samples': 14853632, 'steps': 29010, 'loss/train': 2.0005903244018555}
+02/25/2022 15:10:49 - INFO - codeparrot_training - Step 29011: {'lr': 0.00020105090347339962, 'samples': 14854144, 'steps': 29011, 'loss/train': 1.2578452825546265}
+02/25/2022 15:10:53 - INFO - codeparrot_training - Step 29012: {'lr': 0.00020103485781897658, 'samples': 14854656, 'steps': 29012, 'loss/train': 1.7885220050811768}
+02/25/2022 15:10:58 - INFO - codeparrot_training - Step 29013: {'lr': 0.00020101881237430465, 'samples': 14855168, 'steps': 29013, 'loss/train': 1.321237564086914}
+02/25/2022 15:11:02 - INFO - codeparrot_training - Step 29014: {'lr': 0.00020100276713945259, 'samples': 14855680, 'steps': 29014, 'loss/train': 0.7380515336990356}
+02/25/2022 15:11:08 - INFO - codeparrot_training - Step 29015: {'lr': 0.00020098672211448914, 'samples': 14856192, 'steps': 29015, 'loss/train': 1.0846383571624756}
+02/25/2022 15:11:11 - INFO - codeparrot_training - Step 29016: {'lr': 0.00020097067729948304, 'samples': 14856704, 'steps': 29016, 'loss/train': 1.660983681678772}
+02/25/2022 15:11:17 - INFO - codeparrot_training - Step 29017: {'lr': 0.00020095463269450295, 'samples': 14857216, 'steps': 29017, 'loss/train': 1.8459781408309937}
+02/25/2022 15:11:20 - INFO - codeparrot_training - Step 29018: {'lr': 0.00020093858829961763, 'samples': 14857728, 'steps': 29018, 'loss/train': 2.4474856853485107}
+02/25/2022 15:11:26 - INFO - codeparrot_training - Step 29019: {'lr': 0.00020092254411489585, 'samples': 14858240, 'steps': 29019, 'loss/train': 2.3780765533447266}
+02/25/2022 15:11:29 - INFO - codeparrot_training - Step 29020: {'lr': 0.00020090650014040635, 'samples': 14858752, 'steps': 29020, 'loss/train': 1.4674705266952515}
+02/25/2022 15:11:35 - INFO - codeparrot_training - Step 29021: {'lr': 0.00020089045637621772, 'samples': 14859264, 'steps': 29021, 'loss/train': 1.916792869567871}
+02/25/2022 15:11:38 - INFO - codeparrot_training - Step 29022: {'lr': 0.00020087441282239885, 'samples': 14859776, 'steps': 29022, 'loss/train': 2.1213696002960205}
+02/25/2022 15:11:44 - INFO - codeparrot_training - Step 29023: {'lr': 0.00020085836947901834, 'samples': 14860288, 'steps': 29023, 'loss/train': 1.4662578105926514}
+02/25/2022 15:11:47 - INFO - codeparrot_training - Step 29024: {'lr': 0.00020084232634614503, 'samples': 14860800, 'steps': 29024, 'loss/train': 1.986836552619934}
+02/25/2022 15:11:53 - INFO - codeparrot_training - Step 29025: {'lr': 0.00020082628342384752, 'samples': 14861312, 'steps': 29025, 'loss/train': 2.443887948989868}
+02/25/2022 15:11:56 - INFO - codeparrot_training - Step 29026: {'lr': 0.00020081024071219467, 'samples': 14861824, 'steps': 29026, 'loss/train': 1.7471908330917358}
+02/25/2022 15:12:02 - INFO - codeparrot_training - Step 29027: {'lr': 0.00020079419821125504, 'samples': 14862336, 'steps': 29027, 'loss/train': 2.382305860519409}
+02/25/2022 15:12:05 - INFO - codeparrot_training - Step 29028: {'lr': 0.0002007781559210975, 'samples': 14862848, 'steps': 29028, 'loss/train': 1.7041689157485962}
+02/25/2022 15:12:12 - INFO - codeparrot_training - Step 29029: {'lr': 0.00020076211384179078, 'samples': 14863360, 'steps': 29029, 'loss/train': 1.5009945631027222}
+02/25/2022 15:12:15 - INFO - codeparrot_training - Step 29030: {'lr': 0.0002007460719734034, 'samples': 14863872, 'steps': 29030, 'loss/train': 2.061202049255371}
+02/25/2022 15:12:21 - INFO - codeparrot_training - Step 29031: {'lr': 0.00020073003031600422, 'samples': 14864384, 'steps': 29031, 'loss/train': 1.6635651588439941}
+02/25/2022 15:12:26 - INFO - codeparrot_training - Step 29032: {'lr': 0.00020071398886966198, 'samples': 14864896, 'steps': 29032, 'loss/train': 2.2170233726501465}
+02/25/2022 15:12:30 - INFO - codeparrot_training - Step 29033: {'lr': 0.00020069794763444545, 'samples': 14865408, 'steps': 29033, 'loss/train': 0.6922512650489807}
+02/25/2022 15:12:35 - INFO - codeparrot_training - Step 29034: {'lr': 0.00020068190661042315, 'samples': 14865920, 'steps': 29034, 'loss/train': 1.869996190071106}
+02/25/2022 15:12:39 - INFO - codeparrot_training - Step 29035: {'lr': 0.00020066586579766389, 'samples': 14866432, 'steps': 29035, 'loss/train': 1.3881815671920776}
+02/25/2022 15:12:44 - INFO - codeparrot_training - Step 29036: {'lr': 0.0002006498251962364, 'samples': 14866944, 'steps': 29036, 'loss/train': 2.233100652694702}
+02/25/2022 15:12:48 - INFO - codeparrot_training - Step 29037: {'lr': 0.0002006337848062094, 'samples': 14867456, 'steps': 29037, 'loss/train': 0.6549373269081116}
+02/25/2022 15:12:53 - INFO - codeparrot_training - Step 29038: {'lr': 0.0002006177446276517, 'samples': 14867968, 'steps': 29038, 'loss/train': 2.5598695278167725}
+02/25/2022 15:12:57 - INFO - codeparrot_training - Step 29039: {'lr': 0.00020060170466063178, 'samples': 14868480, 'steps': 29039, 'loss/train': 2.1152689456939697}
+02/25/2022 15:13:03 - INFO - codeparrot_training - Step 29040: {'lr': 0.00020058566490521845, 'samples': 14868992, 'steps': 29040, 'loss/train': 1.7224782705307007}
+02/25/2022 15:13:06 - INFO - codeparrot_training - Step 29041: {'lr': 0.00020056962536148048, 'samples': 14869504, 'steps': 29041, 'loss/train': 1.044476866722107}
+02/25/2022 15:13:12 - INFO - codeparrot_training - Step 29042: {'lr': 0.00020055358602948665, 'samples': 14870016, 'steps': 29042, 'loss/train': 0.026592062786221504}
+02/25/2022 15:13:15 - INFO - codeparrot_training - Step 29043: {'lr': 0.00020053754690930542, 'samples': 14870528, 'steps': 29043, 'loss/train': 1.5946857929229736}
+02/25/2022 15:13:21 - INFO - codeparrot_training - Step 29044: {'lr': 0.00020052150800100565, 'samples': 14871040, 'steps': 29044, 'loss/train': 2.780329704284668}
+02/25/2022 15:13:25 - INFO - codeparrot_training - Step 29045: {'lr': 0.00020050546930465603, 'samples': 14871552, 'steps': 29045, 'loss/train': 0.9196075797080994}
+02/25/2022 15:13:30 - INFO - codeparrot_training - Step 29046: {'lr': 0.00020048943082032533, 'samples': 14872064, 'steps': 29046, 'loss/train': 0.9458131790161133}
+02/25/2022 15:13:34 - INFO - codeparrot_training - Step 29047: {'lr': 0.00020047339254808215, 'samples': 14872576, 'steps': 29047, 'loss/train': 1.834579586982727}
+02/25/2022 15:13:40 - INFO - codeparrot_training - Step 29048: {'lr': 0.0002004573544879952, 'samples': 14873088, 'steps': 29048, 'loss/train': 1.1861038208007812}
+02/25/2022 15:13:44 - INFO - codeparrot_training - Step 29049: {'lr': 0.00020044131664013318, 'samples': 14873600, 'steps': 29049, 'loss/train': 1.1611157655715942}
+02/25/2022 15:13:49 - INFO - codeparrot_training - Step 29050: {'lr': 0.00020042527900456485, 'samples': 14874112, 'steps': 29050, 'loss/train': 1.5956506729125977}
+02/25/2022 15:13:53 - INFO - codeparrot_training - Step 29051: {'lr': 0.00020040924158135897, 'samples': 14874624, 'steps': 29051, 'loss/train': 1.0758423805236816}
+02/25/2022 15:13:58 - INFO - codeparrot_training - Step 29052: {'lr': 0.00020039320437058409, 'samples': 14875136, 'steps': 29052, 'loss/train': 1.390692949295044}
+02/25/2022 15:14:02 - INFO - codeparrot_training - Step 29053: {'lr': 0.00020037716737230892, 'samples': 14875648, 'steps': 29053, 'loss/train': 1.2980892658233643}
+02/25/2022 15:14:08 - INFO - codeparrot_training - Step 29054: {'lr': 0.00020036113058660224, 'samples': 14876160, 'steps': 29054, 'loss/train': 1.3055460453033447}
+02/25/2022 15:14:11 - INFO - codeparrot_training - Step 29055: {'lr': 0.00020034509401353282, 'samples': 14876672, 'steps': 29055, 'loss/train': 0.11481061577796936}
+02/25/2022 15:14:17 - INFO - codeparrot_training - Step 29056: {'lr': 0.00020032905765316914, 'samples': 14877184, 'steps': 29056, 'loss/train': 1.6125483512878418}
+02/25/2022 15:14:20 - INFO - codeparrot_training - Step 29057: {'lr': 0.00020031302150558, 'samples': 14877696, 'steps': 29057, 'loss/train': 2.3205788135528564}
+02/25/2022 15:14:26 - INFO - codeparrot_training - Step 29058: {'lr': 0.0002002969855708341, 'samples': 14878208, 'steps': 29058, 'loss/train': 1.8551654815673828}
+02/25/2022 15:14:29 - INFO - codeparrot_training - Step 29059: {'lr': 0.0002002809498490002, 'samples': 14878720, 'steps': 29059, 'loss/train': 1.5854817628860474}
+02/25/2022 15:14:35 - INFO - codeparrot_training - Step 29060: {'lr': 0.00020026491434014688, 'samples': 14879232, 'steps': 29060, 'loss/train': 2.3203086853027344}
+02/25/2022 15:14:38 - INFO - codeparrot_training - Step 29061: {'lr': 0.00020024887904434286, 'samples': 14879744, 'steps': 29061, 'loss/train': 1.2007865905761719}
+02/25/2022 15:14:44 - INFO - codeparrot_training - Step 29062: {'lr': 0.00020023284396165686, 'samples': 14880256, 'steps': 29062, 'loss/train': 1.4887027740478516}
+02/25/2022 15:14:47 - INFO - codeparrot_training - Step 29063: {'lr': 0.00020021680909215755, 'samples': 14880768, 'steps': 29063, 'loss/train': 2.570343017578125}
+02/25/2022 15:14:53 - INFO - codeparrot_training - Step 29064: {'lr': 0.00020020077443591363, 'samples': 14881280, 'steps': 29064, 'loss/train': 1.6657326221466064}
+02/25/2022 15:14:57 - INFO - codeparrot_training - Step 29065: {'lr': 0.00020018473999299375, 'samples': 14881792, 'steps': 29065, 'loss/train': 2.588829755783081}
+02/25/2022 15:15:03 - INFO - codeparrot_training - Step 29066: {'lr': 0.00020016870576346668, 'samples': 14882304, 'steps': 29066, 'loss/train': 1.1490429639816284}
+02/25/2022 15:15:06 - INFO - codeparrot_training - Step 29067: {'lr': 0.00020015267174740102, 'samples': 14882816, 'steps': 29067, 'loss/train': 1.9579099416732788}
+02/25/2022 15:15:12 - INFO - codeparrot_training - Step 29068: {'lr': 0.0002001366379448655, 'samples': 14883328, 'steps': 29068, 'loss/train': 0.08411605656147003}
+02/25/2022 15:15:15 - INFO - codeparrot_training - Step 29069: {'lr': 0.0002001206043559288, 'samples': 14883840, 'steps': 29069, 'loss/train': 2.3117170333862305}
+02/25/2022 15:15:21 - INFO - codeparrot_training - Step 29070: {'lr': 0.00020010457098065953, 'samples': 14884352, 'steps': 29070, 'loss/train': 1.3876031637191772}
+02/25/2022 15:15:24 - INFO - codeparrot_training - Step 29071: {'lr': 0.00020008853781912646, 'samples': 14884864, 'steps': 29071, 'loss/train': 2.474170207977295}
+02/25/2022 15:15:30 - INFO - codeparrot_training - Step 29072: {'lr': 0.00020007250487139827, 'samples': 14885376, 'steps': 29072, 'loss/train': 0.9779680371284485}
+02/25/2022 15:15:33 - INFO - codeparrot_training - Step 29073: {'lr': 0.00020005647213754363, 'samples': 14885888, 'steps': 29073, 'loss/train': 1.6422357559204102}
+02/25/2022 15:15:39 - INFO - codeparrot_training - Step 29074: {'lr': 0.00020004043961763116, 'samples': 14886400, 'steps': 29074, 'loss/train': 2.2321534156799316}
+02/25/2022 15:15:42 - INFO - codeparrot_training - Step 29075: {'lr': 0.00020002440731172957, 'samples': 14886912, 'steps': 29075, 'loss/train': 1.9346281290054321}
+02/25/2022 15:15:49 - INFO - codeparrot_training - Step 29076: {'lr': 0.00020000837521990755, 'samples': 14887424, 'steps': 29076, 'loss/train': 1.2649942636489868}
+02/25/2022 15:15:52 - INFO - codeparrot_training - Step 29077: {'lr': 0.0001999923433422339, 'samples': 14887936, 'steps': 29077, 'loss/train': 1.595371127128601}
+02/25/2022 15:15:58 - INFO - codeparrot_training - Step 29078: {'lr': 0.00019997631167877706, 'samples': 14888448, 'steps': 29078, 'loss/train': 2.4133219718933105}
+02/25/2022 15:16:01 - INFO - codeparrot_training - Step 29079: {'lr': 0.0001999602802296058, 'samples': 14888960, 'steps': 29079, 'loss/train': 2.6212804317474365}
+02/25/2022 15:16:07 - INFO - codeparrot_training - Step 29080: {'lr': 0.00019994424899478884, 'samples': 14889472, 'steps': 29080, 'loss/train': 1.28313148021698}
+02/25/2022 15:16:10 - INFO - codeparrot_training - Step 29081: {'lr': 0.0001999282179743949, 'samples': 14889984, 'steps': 29081, 'loss/train': 0.08394929021596909}
+02/25/2022 15:16:16 - INFO - codeparrot_training - Step 29082: {'lr': 0.00019991218716849246, 'samples': 14890496, 'steps': 29082, 'loss/train': 2.2583987712860107}
+02/25/2022 15:16:19 - INFO - codeparrot_training - Step 29083: {'lr': 0.00019989615657715033, 'samples': 14891008, 'steps': 29083, 'loss/train': 1.5651559829711914}
+02/25/2022 15:16:25 - INFO - codeparrot_training - Step 29084: {'lr': 0.00019988012620043716, 'samples': 14891520, 'steps': 29084, 'loss/train': 2.0740163326263428}
+02/25/2022 15:16:28 - INFO - codeparrot_training - Step 29085: {'lr': 0.00019986409603842158, 'samples': 14892032, 'steps': 29085, 'loss/train': 2.237011194229126}
+02/25/2022 15:16:35 - INFO - codeparrot_training - Step 29086: {'lr': 0.00019984806609117247, 'samples': 14892544, 'steps': 29086, 'loss/train': 1.4755436182022095}
+02/25/2022 15:16:38 - INFO - codeparrot_training - Step 29087: {'lr': 0.00019983203635875812, 'samples': 14893056, 'steps': 29087, 'loss/train': 1.7659515142440796}
+02/25/2022 15:16:44 - INFO - codeparrot_training - Step 29088: {'lr': 0.00019981600684124744, 'samples': 14893568, 'steps': 29088, 'loss/train': 2.350338935852051}
+02/25/2022 15:16:47 - INFO - codeparrot_training - Step 29089: {'lr': 0.00019979997753870907, 'samples': 14894080, 'steps': 29089, 'loss/train': 2.077547311782837}
+02/25/2022 15:16:53 - INFO - codeparrot_training - Step 29090: {'lr': 0.00019978394845121173, 'samples': 14894592, 'steps': 29090, 'loss/train': 1.985950231552124}
+02/25/2022 15:16:56 - INFO - codeparrot_training - Step 29091: {'lr': 0.00019976791957882392, 'samples': 14895104, 'steps': 29091, 'loss/train': 1.5815422534942627}
+02/25/2022 15:17:02 - INFO - codeparrot_training - Step 29092: {'lr': 0.0001997518909216144, 'samples': 14895616, 'steps': 29092, 'loss/train': 2.0188963413238525}
+02/25/2022 15:17:05 - INFO - codeparrot_training - Step 29093: {'lr': 0.0001997358624796518, 'samples': 14896128, 'steps': 29093, 'loss/train': 2.1548421382904053}
+02/25/2022 15:17:11 - INFO - codeparrot_training - Step 29094: {'lr': 0.00019971983425300495, 'samples': 14896640, 'steps': 29094, 'loss/train': 2.8655264377593994}
+02/25/2022 15:17:14 - INFO - codeparrot_training - Step 29095: {'lr': 0.00019970380624174223, 'samples': 14897152, 'steps': 29095, 'loss/train': 1.9505460262298584}
+02/25/2022 15:17:21 - INFO - codeparrot_training - Step 29096: {'lr': 0.0001996877784459324, 'samples': 14897664, 'steps': 29096, 'loss/train': 1.06796395778656}
+02/25/2022 15:17:27 - INFO - codeparrot_training - Step 29097: {'lr': 0.00019967175086564422, 'samples': 14898176, 'steps': 29097, 'loss/train': 2.072253704071045}
+02/25/2022 15:17:30 - INFO - codeparrot_training - Step 29098: {'lr': 0.00019965572350094625, 'samples': 14898688, 'steps': 29098, 'loss/train': 0.9562056064605713}
+02/25/2022 15:17:36 - INFO - codeparrot_training - Step 29099: {'lr': 0.00019963969635190725, 'samples': 14899200, 'steps': 29099, 'loss/train': 1.3275948762893677}
+02/25/2022 15:17:39 - INFO - codeparrot_training - Step 29100: {'lr': 0.00019962366941859573, 'samples': 14899712, 'steps': 29100, 'loss/train': 1.411186695098877}
+02/25/2022 15:17:44 - INFO - codeparrot_training - Step 29101: {'lr': 0.0001996076427010804, 'samples': 14900224, 'steps': 29101, 'loss/train': 0.07481779903173447}
+02/25/2022 15:17:48 - INFO - codeparrot_training - Step 29102: {'lr': 0.00019959161619943002, 'samples': 14900736, 'steps': 29102, 'loss/train': 1.9677882194519043}
+02/25/2022 15:17:53 - INFO - codeparrot_training - Step 29103: {'lr': 0.00019957558991371308, 'samples': 14901248, 'steps': 29103, 'loss/train': 1.8027527332305908}
+02/25/2022 15:17:57 - INFO - codeparrot_training - Step 29104: {'lr': 0.0001995595638439983, 'samples': 14901760, 'steps': 29104, 'loss/train': 0.16070152819156647}
+02/25/2022 15:18:02 - INFO - codeparrot_training - Step 29105: {'lr': 0.0001995435379903543, 'samples': 14902272, 'steps': 29105, 'loss/train': 1.4582862854003906}
+02/25/2022 15:18:06 - INFO - codeparrot_training - Step 29106: {'lr': 0.00019952751235284983, 'samples': 14902784, 'steps': 29106, 'loss/train': 3.9073355197906494}
+02/25/2022 15:18:11 - INFO - codeparrot_training - Step 29107: {'lr': 0.00019951148693155346, 'samples': 14903296, 'steps': 29107, 'loss/train': 0.9861534833908081}
+02/25/2022 15:18:15 - INFO - codeparrot_training - Step 29108: {'lr': 0.0001994954617265338, 'samples': 14903808, 'steps': 29108, 'loss/train': 1.1505014896392822}
+02/25/2022 15:18:20 - INFO - codeparrot_training - Step 29109: {'lr': 0.00019947943673785963, 'samples': 14904320, 'steps': 29109, 'loss/train': 1.6635161638259888}
+02/25/2022 15:18:24 - INFO - codeparrot_training - Step 29110: {'lr': 0.00019946341196559942, 'samples': 14904832, 'steps': 29110, 'loss/train': 2.057328462600708}
+02/25/2022 15:18:30 - INFO - codeparrot_training - Step 29111: {'lr': 0.00019944738740982192, 'samples': 14905344, 'steps': 29111, 'loss/train': 1.3910659551620483}
+02/25/2022 15:18:34 - INFO - codeparrot_training - Step 29112: {'lr': 0.00019943136307059577, 'samples': 14905856, 'steps': 29112, 'loss/train': 2.2051427364349365}
+02/25/2022 15:18:39 - INFO - codeparrot_training - Step 29113: {'lr': 0.00019941533894798967, 'samples': 14906368, 'steps': 29113, 'loss/train': 1.7288485765457153}
+02/25/2022 15:18:43 - INFO - codeparrot_training - Step 29114: {'lr': 0.00019939931504207208, 'samples': 14906880, 'steps': 29114, 'loss/train': 2.2145586013793945}
+02/25/2022 15:18:48 - INFO - codeparrot_training - Step 29115: {'lr': 0.0001993832913529118, 'samples': 14907392, 'steps': 29115, 'loss/train': 1.4452009201049805}
+02/25/2022 15:18:52 - INFO - codeparrot_training - Step 29116: {'lr': 0.0001993672678805775, 'samples': 14907904, 'steps': 29116, 'loss/train': 2.066502809524536}
+02/25/2022 15:18:57 - INFO - codeparrot_training - Step 29117: {'lr': 0.0001993512446251377, 'samples': 14908416, 'steps': 29117, 'loss/train': 0.6649622321128845}
+02/25/2022 15:19:01 - INFO - codeparrot_training - Step 29118: {'lr': 0.00019933522158666108, 'samples': 14908928, 'steps': 29118, 'loss/train': 3.3403406143188477}
+02/25/2022 15:19:06 - INFO - codeparrot_training - Step 29119: {'lr': 0.00019931919876521622, 'samples': 14909440, 'steps': 29119, 'loss/train': 1.5797144174575806}
+02/25/2022 15:19:10 - INFO - codeparrot_training - Step 29120: {'lr': 0.00019930317616087195, 'samples': 14909952, 'steps': 29120, 'loss/train': 1.4475345611572266}
+02/25/2022 15:19:16 - INFO - codeparrot_training - Step 29121: {'lr': 0.0001992871537736967, 'samples': 14910464, 'steps': 29121, 'loss/train': 1.509169101715088}
+02/25/2022 15:19:19 - INFO - codeparrot_training - Step 29122: {'lr': 0.00019927113160375914, 'samples': 14910976, 'steps': 29122, 'loss/train': 1.3574632406234741}
+02/25/2022 15:19:25 - INFO - codeparrot_training - Step 29123: {'lr': 0.00019925510965112798, 'samples': 14911488, 'steps': 29123, 'loss/train': 2.3240227699279785}
+02/25/2022 15:19:28 - INFO - codeparrot_training - Step 29124: {'lr': 0.00019923908791587177, 'samples': 14912000, 'steps': 29124, 'loss/train': 2.033957004547119}
+02/25/2022 15:19:34 - INFO - codeparrot_training - Step 29125: {'lr': 0.00019922306639805933, 'samples': 14912512, 'steps': 29125, 'loss/train': 1.8728193044662476}
+02/25/2022 15:19:37 - INFO - codeparrot_training - Step 29126: {'lr': 0.00019920704509775904, 'samples': 14913024, 'steps': 29126, 'loss/train': 1.3663140535354614}
+02/25/2022 15:19:43 - INFO - codeparrot_training - Step 29127: {'lr': 0.00019919102401503964, 'samples': 14913536, 'steps': 29127, 'loss/train': 2.560606002807617}
+02/25/2022 15:19:46 - INFO - codeparrot_training - Step 29128: {'lr': 0.0001991750031499698, 'samples': 14914048, 'steps': 29128, 'loss/train': 0.6288747787475586}
+02/25/2022 15:19:52 - INFO - codeparrot_training - Step 29129: {'lr': 0.00019915898250261814, 'samples': 14914560, 'steps': 29129, 'loss/train': 1.8293359279632568}
+02/25/2022 15:19:55 - INFO - codeparrot_training - Step 29130: {'lr': 0.0001991429620730532, 'samples': 14915072, 'steps': 29130, 'loss/train': 1.4865890741348267}
+02/25/2022 15:20:02 - INFO - codeparrot_training - Step 29131: {'lr': 0.0001991269418613436, 'samples': 14915584, 'steps': 29131, 'loss/train': 1.2068628072738647}
+02/25/2022 15:20:05 - INFO - codeparrot_training - Step 29132: {'lr': 0.00019911092186755808, 'samples': 14916096, 'steps': 29132, 'loss/train': 3.7082173824310303}
+02/25/2022 15:20:10 - INFO - codeparrot_training - Step 29133: {'lr': 0.0001990949020917653, 'samples': 14916608, 'steps': 29133, 'loss/train': 1.8108493089675903}
+02/25/2022 15:20:14 - INFO - codeparrot_training - Step 29134: {'lr': 0.00019907888253403372, 'samples': 14917120, 'steps': 29134, 'loss/train': 1.4859955310821533}
+02/25/2022 15:20:20 - INFO - codeparrot_training - Step 29135: {'lr': 0.00019906286319443202, 'samples': 14917632, 'steps': 29135, 'loss/train': 0.7468170523643494}
+02/25/2022 15:20:23 - INFO - codeparrot_training - Step 29136: {'lr': 0.00019904684407302882, 'samples': 14918144, 'steps': 29136, 'loss/train': 2.2013497352600098}
+02/25/2022 15:20:29 - INFO - codeparrot_training - Step 29137: {'lr': 0.00019903082516989275, 'samples': 14918656, 'steps': 29137, 'loss/train': 1.5608460903167725}
+02/25/2022 15:20:34 - INFO - codeparrot_training - Step 29138: {'lr': 0.0001990148064850926, 'samples': 14919168, 'steps': 29138, 'loss/train': 1.5630799531936646}
+02/25/2022 15:20:38 - INFO - codeparrot_training - Step 29139: {'lr': 0.0001989987880186967, 'samples': 14919680, 'steps': 29139, 'loss/train': 1.295097827911377}
+02/25/2022 15:20:44 - INFO - codeparrot_training - Step 29140: {'lr': 0.0001989827697707738, 'samples': 14920192, 'steps': 29140, 'loss/train': 2.077880859375}
+02/25/2022 15:20:47 - INFO - codeparrot_training - Step 29141: {'lr': 0.0001989667517413925, 'samples': 14920704, 'steps': 29141, 'loss/train': 0.9966211915016174}
+02/25/2022 15:20:53 - INFO - codeparrot_training - Step 29142: {'lr': 0.00019895073393062153, 'samples': 14921216, 'steps': 29142, 'loss/train': 0.6767629384994507}
+02/25/2022 15:20:56 - INFO - codeparrot_training - Step 29143: {'lr': 0.0001989347163385293, 'samples': 14921728, 'steps': 29143, 'loss/train': 1.9332820177078247}
+02/25/2022 15:21:02 - INFO - codeparrot_training - Step 29144: {'lr': 0.00019891869896518455, 'samples': 14922240, 'steps': 29144, 'loss/train': 2.0521185398101807}
+02/25/2022 15:21:05 - INFO - codeparrot_training - Step 29145: {'lr': 0.00019890268181065585, 'samples': 14922752, 'steps': 29145, 'loss/train': 0.08609345555305481}
+02/25/2022 15:21:11 - INFO - codeparrot_training - Step 29146: {'lr': 0.00019888666487501193, 'samples': 14923264, 'steps': 29146, 'loss/train': 4.454577922821045}
+02/25/2022 15:21:14 - INFO - codeparrot_training - Step 29147: {'lr': 0.00019887064815832126, 'samples': 14923776, 'steps': 29147, 'loss/train': 0.8347738981246948}
+02/25/2022 15:21:20 - INFO - codeparrot_training - Step 29148: {'lr': 0.00019885463166065248, 'samples': 14924288, 'steps': 29148, 'loss/train': 1.6879264116287231}
+02/25/2022 15:21:23 - INFO - codeparrot_training - Step 29149: {'lr': 0.00019883861538207427, 'samples': 14924800, 'steps': 29149, 'loss/train': 2.151691198348999}
+02/25/2022 15:21:29 - INFO - codeparrot_training - Step 29150: {'lr': 0.00019882259932265512, 'samples': 14925312, 'steps': 29150, 'loss/train': 1.461527705192566}
+02/25/2022 15:21:33 - INFO - codeparrot_training - Step 29151: {'lr': 0.00019880658348246375, 'samples': 14925824, 'steps': 29151, 'loss/train': 0.7858912944793701}
+02/25/2022 15:21:36 - INFO - codeparrot_training - Step 29152: {'lr': 0.00019879056786156865, 'samples': 14926336, 'steps': 29152, 'loss/train': 1.5283597707748413}
+02/25/2022 15:21:42 - INFO - codeparrot_training - Step 29153: {'lr': 0.00019877455246003858, 'samples': 14926848, 'steps': 29153, 'loss/train': 1.1754670143127441}
+02/25/2022 15:21:46 - INFO - codeparrot_training - Step 29154: {'lr': 0.00019875853727794202, 'samples': 14927360, 'steps': 29154, 'loss/train': 1.646106481552124}
+02/25/2022 15:21:51 - INFO - codeparrot_training - Step 29155: {'lr': 0.00019874252231534765, 'samples': 14927872, 'steps': 29155, 'loss/train': 1.1370307207107544}
+02/25/2022 15:21:55 - INFO - codeparrot_training - Step 29156: {'lr': 0.00019872650757232397, 'samples': 14928384, 'steps': 29156, 'loss/train': 1.2444932460784912}
+02/25/2022 15:22:00 - INFO - codeparrot_training - Step 29157: {'lr': 0.00019871049304893973, 'samples': 14928896, 'steps': 29157, 'loss/train': 1.2231873273849487}
+02/25/2022 15:22:04 - INFO - codeparrot_training - Step 29158: {'lr': 0.0001986944787452634, 'samples': 14929408, 'steps': 29158, 'loss/train': 1.5079692602157593}
+02/25/2022 15:22:09 - INFO - codeparrot_training - Step 29159: {'lr': 0.0001986784646613636, 'samples': 14929920, 'steps': 29159, 'loss/train': 2.841705799102783}
+02/25/2022 15:22:13 - INFO - codeparrot_training - Step 29160: {'lr': 0.0001986624507973091, 'samples': 14930432, 'steps': 29160, 'loss/train': 1.5490890741348267}
+02/25/2022 15:22:18 - INFO - codeparrot_training - Step 29161: {'lr': 0.00019864643715316823, 'samples': 14930944, 'steps': 29161, 'loss/train': 0.963879406452179}
+02/25/2022 15:22:22 - INFO - codeparrot_training - Step 29162: {'lr': 0.00019863042372900974, 'samples': 14931456, 'steps': 29162, 'loss/train': 1.7841893434524536}
+02/25/2022 15:22:27 - INFO - codeparrot_training - Step 29163: {'lr': 0.00019861441052490216, 'samples': 14931968, 'steps': 29163, 'loss/train': 1.379916787147522}
+02/25/2022 15:22:31 - INFO - codeparrot_training - Step 29164: {'lr': 0.0001985983975409143, 'samples': 14932480, 'steps': 29164, 'loss/train': 1.6264894008636475}
+02/25/2022 15:22:36 - INFO - codeparrot_training - Step 29165: {'lr': 0.00019858238477711443, 'samples': 14932992, 'steps': 29165, 'loss/train': 1.7188951969146729}
+02/25/2022 15:22:40 - INFO - codeparrot_training - Step 29166: {'lr': 0.0001985663722335713, 'samples': 14933504, 'steps': 29166, 'loss/train': 0.8646218180656433}
+02/25/2022 15:22:46 - INFO - codeparrot_training - Step 29167: {'lr': 0.00019855035991035353, 'samples': 14934016, 'steps': 29167, 'loss/train': 2.0107421875}
+02/25/2022 15:22:51 - INFO - codeparrot_training - Step 29168: {'lr': 0.00019853434780752973, 'samples': 14934528, 'steps': 29168, 'loss/train': 1.2153934240341187}
+02/25/2022 15:22:55 - INFO - codeparrot_training - Step 29169: {'lr': 0.0001985183359251684, 'samples': 14935040, 'steps': 29169, 'loss/train': 2.0026893615722656}
+02/25/2022 15:23:00 - INFO - codeparrot_training - Step 29170: {'lr': 0.00019850232426333814, 'samples': 14935552, 'steps': 29170, 'loss/train': 2.3233540058135986}
+02/25/2022 15:23:04 - INFO - codeparrot_training - Step 29171: {'lr': 0.00019848631282210754, 'samples': 14936064, 'steps': 29171, 'loss/train': 0.8999312520027161}
+02/25/2022 15:23:09 - INFO - codeparrot_training - Step 29172: {'lr': 0.00019847030160154527, 'samples': 14936576, 'steps': 29172, 'loss/train': 1.1251603364944458}
+02/25/2022 15:23:13 - INFO - codeparrot_training - Step 29173: {'lr': 0.00019845429060171997, 'samples': 14937088, 'steps': 29173, 'loss/train': 2.0283477306365967}
+02/25/2022 15:23:18 - INFO - codeparrot_training - Step 29174: {'lr': 0.00019843827982269998, 'samples': 14937600, 'steps': 29174, 'loss/train': 0.05940178036689758}
+02/25/2022 15:23:22 - INFO - codeparrot_training - Step 29175: {'lr': 0.00019842226926455403, 'samples': 14938112, 'steps': 29175, 'loss/train': 1.4810538291931152}
+02/25/2022 15:23:28 - INFO - codeparrot_training - Step 29176: {'lr': 0.00019840625892735074, 'samples': 14938624, 'steps': 29176, 'loss/train': 0.6489487886428833}
+02/25/2022 15:23:32 - INFO - codeparrot_training - Step 29177: {'lr': 0.00019839024881115872, 'samples': 14939136, 'steps': 29177, 'loss/train': 1.5926954746246338}
+02/25/2022 15:23:35 - INFO - codeparrot_training - Step 29178: {'lr': 0.0001983742389160464, 'samples': 14939648, 'steps': 29178, 'loss/train': 2.5636343955993652}
+02/25/2022 15:23:41 - INFO - codeparrot_training - Step 29179: {'lr': 0.00019835822924208244, 'samples': 14940160, 'steps': 29179, 'loss/train': 1.109387993812561}
+02/25/2022 15:23:46 - INFO - codeparrot_training - Step 29180: {'lr': 0.00019834221978933542, 'samples': 14940672, 'steps': 29180, 'loss/train': 1.7216293811798096}
+02/25/2022 15:23:50 - INFO - codeparrot_training - Step 29181: {'lr': 0.00019832621055787406, 'samples': 14941184, 'steps': 29181, 'loss/train': 0.7553069591522217}
+02/25/2022 15:23:55 - INFO - codeparrot_training - Step 29182: {'lr': 0.0001983102015477667, 'samples': 14941696, 'steps': 29182, 'loss/train': 1.003391981124878}
+02/25/2022 15:23:59 - INFO - codeparrot_training - Step 29183: {'lr': 0.000198294192759082, 'samples': 14942208, 'steps': 29183, 'loss/train': 1.844997525215149}
+02/25/2022 15:24:04 - INFO - codeparrot_training - Step 29184: {'lr': 0.00019827818419188855, 'samples': 14942720, 'steps': 29184, 'loss/train': 1.0950952768325806}
+02/25/2022 15:24:08 - INFO - codeparrot_training - Step 29185: {'lr': 0.00019826217584625495, 'samples': 14943232, 'steps': 29185, 'loss/train': 1.5986212491989136}
+02/25/2022 15:24:13 - INFO - codeparrot_training - Step 29186: {'lr': 0.00019824616772224987, 'samples': 14943744, 'steps': 29186, 'loss/train': 1.4953640699386597}
+02/25/2022 15:24:17 - INFO - codeparrot_training - Step 29187: {'lr': 0.00019823015981994165, 'samples': 14944256, 'steps': 29187, 'loss/train': 2.254401445388794}
+02/25/2022 15:24:20 - INFO - codeparrot_training - Step 29188: {'lr': 0.00019821415213939903, 'samples': 14944768, 'steps': 29188, 'loss/train': 2.495269775390625}
+02/25/2022 15:24:26 - INFO - codeparrot_training - Step 29189: {'lr': 0.0001981981446806905, 'samples': 14945280, 'steps': 29189, 'loss/train': 1.8421341180801392}
+02/25/2022 15:24:30 - INFO - codeparrot_training - Step 29190: {'lr': 0.0001981821374438848, 'samples': 14945792, 'steps': 29190, 'loss/train': 2.6240923404693604}
+02/25/2022 15:24:35 - INFO - codeparrot_training - Step 29191: {'lr': 0.00019816613042905024, 'samples': 14946304, 'steps': 29191, 'loss/train': 1.244981050491333}
+02/25/2022 15:24:39 - INFO - codeparrot_training - Step 29192: {'lr': 0.0001981501236362555, 'samples': 14946816, 'steps': 29192, 'loss/train': 2.169217109680176}
+02/25/2022 15:24:44 - INFO - codeparrot_training - Step 29193: {'lr': 0.00019813411706556927, 'samples': 14947328, 'steps': 29193, 'loss/train': 1.9341119527816772}
+02/25/2022 15:24:48 - INFO - codeparrot_training - Step 29194: {'lr': 0.00019811811071705993, 'samples': 14947840, 'steps': 29194, 'loss/train': 1.773614525794983}
+02/25/2022 15:24:53 - INFO - codeparrot_training - Step 29195: {'lr': 0.0001981021045907962, 'samples': 14948352, 'steps': 29195, 'loss/train': 1.3399029970169067}
+02/25/2022 15:24:57 - INFO - codeparrot_training - Step 29196: {'lr': 0.00019808609868684652, 'samples': 14948864, 'steps': 29196, 'loss/train': 1.0027785301208496}
+02/25/2022 15:25:02 - INFO - codeparrot_training - Step 29197: {'lr': 0.00019807009300527954, 'samples': 14949376, 'steps': 29197, 'loss/train': 0.34942182898521423}
+02/25/2022 15:25:09 - INFO - codeparrot_training - Step 29198: {'lr': 0.00019805408754616376, 'samples': 14949888, 'steps': 29198, 'loss/train': 1.4731298685073853}
+02/25/2022 15:25:12 - INFO - codeparrot_training - Step 29199: {'lr': 0.00019803808230956782, 'samples': 14950400, 'steps': 29199, 'loss/train': 0.944177508354187}
+02/25/2022 15:25:18 - INFO - codeparrot_training - Step 29200: {'lr': 0.0001980220772955602, 'samples': 14950912, 'steps': 29200, 'loss/train': 0.8870585560798645}
+02/25/2022 15:25:21 - INFO - codeparrot_training - Step 29201: {'lr': 0.00019800607250420948, 'samples': 14951424, 'steps': 29201, 'loss/train': 0.7162878513336182}
+02/25/2022 15:25:27 - INFO - codeparrot_training - Step 29202: {'lr': 0.00019799006793558423, 'samples': 14951936, 'steps': 29202, 'loss/train': 1.5842257738113403}
+02/25/2022 15:25:30 - INFO - codeparrot_training - Step 29203: {'lr': 0.00019797406358975308, 'samples': 14952448, 'steps': 29203, 'loss/train': 1.0783542394638062}
+02/25/2022 15:25:36 - INFO - codeparrot_training - Step 29204: {'lr': 0.00019795805946678453, 'samples': 14952960, 'steps': 29204, 'loss/train': 1.986425518989563}
+02/25/2022 15:25:39 - INFO - codeparrot_training - Step 29205: {'lr': 0.00019794205556674706, 'samples': 14953472, 'steps': 29205, 'loss/train': 1.8771569728851318}
+02/25/2022 15:25:44 - INFO - codeparrot_training - Step 29206: {'lr': 0.0001979260518897093, 'samples': 14953984, 'steps': 29206, 'loss/train': 1.699537992477417}
+02/25/2022 15:25:48 - INFO - codeparrot_training - Step 29207: {'lr': 0.00019791004843573975, 'samples': 14954496, 'steps': 29207, 'loss/train': 1.7367717027664185}
+02/25/2022 15:25:54 - INFO - codeparrot_training - Step 29208: {'lr': 0.00019789404520490717, 'samples': 14955008, 'steps': 29208, 'loss/train': 2.343014717102051}
+02/25/2022 15:25:57 - INFO - codeparrot_training - Step 29209: {'lr': 0.00019787804219727984, 'samples': 14955520, 'steps': 29209, 'loss/train': 1.4697836637496948}
+02/25/2022 15:26:03 - INFO - codeparrot_training - Step 29210: {'lr': 0.0001978620394129264, 'samples': 14956032, 'steps': 29210, 'loss/train': 0.46326011419296265}
+02/25/2022 15:26:06 - INFO - codeparrot_training - Step 29211: {'lr': 0.00019784603685191546, 'samples': 14956544, 'steps': 29211, 'loss/train': 2.3311257362365723}
+02/25/2022 15:26:12 - INFO - codeparrot_training - Step 29212: {'lr': 0.00019783003451431564, 'samples': 14957056, 'steps': 29212, 'loss/train': 1.293544888496399}
+02/25/2022 15:26:15 - INFO - codeparrot_training - Step 29213: {'lr': 0.00019781403240019528, 'samples': 14957568, 'steps': 29213, 'loss/train': 1.6733356714248657}
+02/25/2022 15:26:21 - INFO - codeparrot_training - Step 29214: {'lr': 0.00019779803050962302, 'samples': 14958080, 'steps': 29214, 'loss/train': 1.7695726156234741}
+02/25/2022 15:26:25 - INFO - codeparrot_training - Step 29215: {'lr': 0.00019778202884266743, 'samples': 14958592, 'steps': 29215, 'loss/train': 1.8135424852371216}
+02/25/2022 15:26:30 - INFO - codeparrot_training - Step 29216: {'lr': 0.00019776602739939714, 'samples': 14959104, 'steps': 29216, 'loss/train': 0.8080177307128906}
+02/25/2022 15:26:34 - INFO - codeparrot_training - Step 29217: {'lr': 0.00019775002617988054, 'samples': 14959616, 'steps': 29217, 'loss/train': 1.9845391511917114}
+02/25/2022 15:26:39 - INFO - codeparrot_training - Step 29218: {'lr': 0.00019773402518418617, 'samples': 14960128, 'steps': 29218, 'loss/train': 1.1778076887130737}
+02/25/2022 15:26:43 - INFO - codeparrot_training - Step 29219: {'lr': 0.00019771802441238267, 'samples': 14960640, 'steps': 29219, 'loss/train': 2.2341270446777344}
+02/25/2022 15:26:48 - INFO - codeparrot_training - Step 29220: {'lr': 0.00019770202386453856, 'samples': 14961152, 'steps': 29220, 'loss/train': 2.2118546962738037}
+02/25/2022 15:26:52 - INFO - codeparrot_training - Step 29221: {'lr': 0.0001976860235407225, 'samples': 14961664, 'steps': 29221, 'loss/train': 1.772839903831482}
+02/25/2022 15:26:57 - INFO - codeparrot_training - Step 29222: {'lr': 0.00019767002344100277, 'samples': 14962176, 'steps': 29222, 'loss/train': 2.4914770126342773}
+02/25/2022 15:27:01 - INFO - codeparrot_training - Step 29223: {'lr': 0.00019765402356544803, 'samples': 14962688, 'steps': 29223, 'loss/train': 0.49232980608940125}
+02/25/2022 15:27:07 - INFO - codeparrot_training - Step 29224: {'lr': 0.00019763802391412687, 'samples': 14963200, 'steps': 29224, 'loss/train': 1.1145492792129517}
+02/25/2022 15:27:10 - INFO - codeparrot_training - Step 29225: {'lr': 0.00019762202448710787, 'samples': 14963712, 'steps': 29225, 'loss/train': 0.11281704157590866}
+02/25/2022 15:27:16 - INFO - codeparrot_training - Step 29226: {'lr': 0.00019760602528445936, 'samples': 14964224, 'steps': 29226, 'loss/train': 1.1031262874603271}
+02/25/2022 15:27:19 - INFO - codeparrot_training - Step 29227: {'lr': 0.00019759002630625002, 'samples': 14964736, 'steps': 29227, 'loss/train': 3.045053482055664}
+02/25/2022 15:27:25 - INFO - codeparrot_training - Step 29228: {'lr': 0.00019757402755254838, 'samples': 14965248, 'steps': 29228, 'loss/train': 2.555513620376587}
+02/25/2022 15:27:30 - INFO - codeparrot_training - Step 29229: {'lr': 0.00019755802902342304, 'samples': 14965760, 'steps': 29229, 'loss/train': 1.16560697555542}
+02/25/2022 15:27:34 - INFO - codeparrot_training - Step 29230: {'lr': 0.00019754203071894235, 'samples': 14966272, 'steps': 29230, 'loss/train': 1.9016810655593872}
+02/25/2022 15:27:37 - INFO - codeparrot_training - Step 29231: {'lr': 0.00019752603263917495, 'samples': 14966784, 'steps': 29231, 'loss/train': 0.2916000485420227}
+02/25/2022 15:27:43 - INFO - codeparrot_training - Step 29232: {'lr': 0.0001975100347841894, 'samples': 14967296, 'steps': 29232, 'loss/train': 1.4350587129592896}
+02/25/2022 15:27:46 - INFO - codeparrot_training - Step 29233: {'lr': 0.0001974940371540542, 'samples': 14967808, 'steps': 29233, 'loss/train': 1.8296558856964111}
+02/25/2022 15:27:53 - INFO - codeparrot_training - Step 29234: {'lr': 0.00019747803974883788, 'samples': 14968320, 'steps': 29234, 'loss/train': 1.3263262510299683}
+02/25/2022 15:27:56 - INFO - codeparrot_training - Step 29235: {'lr': 0.0001974620425686089, 'samples': 14968832, 'steps': 29235, 'loss/train': 2.071936845779419}
+02/25/2022 15:28:02 - INFO - codeparrot_training - Step 29236: {'lr': 0.00019744604561343587, 'samples': 14969344, 'steps': 29236, 'loss/train': 1.1329994201660156}
+02/25/2022 15:28:05 - INFO - codeparrot_training - Step 29237: {'lr': 0.00019743004888338734, 'samples': 14969856, 'steps': 29237, 'loss/train': 2.0754735469818115}
+02/25/2022 15:28:11 - INFO - codeparrot_training - Step 29238: {'lr': 0.00019741405237853177, 'samples': 14970368, 'steps': 29238, 'loss/train': 0.10996003448963165}
+02/25/2022 15:28:14 - INFO - codeparrot_training - Step 29239: {'lr': 0.00019739805609893767, 'samples': 14970880, 'steps': 29239, 'loss/train': 2.6604561805725098}
+02/25/2022 15:28:20 - INFO - codeparrot_training - Step 29240: {'lr': 0.00019738206004467362, 'samples': 14971392, 'steps': 29240, 'loss/train': 1.1484644412994385}
+02/25/2022 15:28:23 - INFO - codeparrot_training - Step 29241: {'lr': 0.0001973660642158081, 'samples': 14971904, 'steps': 29241, 'loss/train': 2.341256856918335}
+02/25/2022 15:28:29 - INFO - codeparrot_training - Step 29242: {'lr': 0.00019735006861240962, 'samples': 14972416, 'steps': 29242, 'loss/train': 1.289214015007019}
+02/25/2022 15:28:35 - INFO - codeparrot_training - Step 29243: {'lr': 0.00019733407323454678, 'samples': 14972928, 'steps': 29243, 'loss/train': 1.597387433052063}
+02/25/2022 15:28:38 - INFO - codeparrot_training - Step 29244: {'lr': 0.00019731807808228807, 'samples': 14973440, 'steps': 29244, 'loss/train': 1.108201503753662}
+02/25/2022 15:28:44 - INFO - codeparrot_training - Step 29245: {'lr': 0.00019730208315570192, 'samples': 14973952, 'steps': 29245, 'loss/train': 1.306506872177124}
+02/25/2022 15:28:48 - INFO - codeparrot_training - Step 29246: {'lr': 0.0001972860884548569, 'samples': 14974464, 'steps': 29246, 'loss/train': 1.9439525604248047}
+02/25/2022 15:28:51 - INFO - codeparrot_training - Step 29247: {'lr': 0.00019727009397982166, 'samples': 14974976, 'steps': 29247, 'loss/train': 1.989899754524231}
+02/25/2022 15:28:57 - INFO - codeparrot_training - Step 29248: {'lr': 0.0001972540997306645, 'samples': 14975488, 'steps': 29248, 'loss/train': 1.3352165222167969}
+02/25/2022 15:29:00 - INFO - codeparrot_training - Step 29249: {'lr': 0.000197238105707454, 'samples': 14976000, 'steps': 29249, 'loss/train': 1.0574227571487427}
+02/25/2022 15:29:06 - INFO - codeparrot_training - Step 29250: {'lr': 0.00019722211191025872, 'samples': 14976512, 'steps': 29250, 'loss/train': 1.839902639389038}
+02/25/2022 15:29:09 - INFO - codeparrot_training - Step 29251: {'lr': 0.00019720611833914725, 'samples': 14977024, 'steps': 29251, 'loss/train': 0.0934521034359932}
+02/25/2022 15:29:15 - INFO - codeparrot_training - Step 29252: {'lr': 0.0001971901249941879, 'samples': 14977536, 'steps': 29252, 'loss/train': 1.4760456085205078}
+02/25/2022 15:29:18 - INFO - codeparrot_training - Step 29253: {'lr': 0.00019717413187544927, 'samples': 14978048, 'steps': 29253, 'loss/train': 2.273589611053467}
+02/25/2022 15:29:24 - INFO - codeparrot_training - Step 29254: {'lr': 0.00019715813898299993, 'samples': 14978560, 'steps': 29254, 'loss/train': 1.7405003309249878}
+02/25/2022 15:29:27 - INFO - codeparrot_training - Step 29255: {'lr': 0.0001971421463169083, 'samples': 14979072, 'steps': 29255, 'loss/train': 2.853459596633911}
+02/25/2022 15:29:32 - INFO - codeparrot_training - Step 29256: {'lr': 0.00019712615387724306, 'samples': 14979584, 'steps': 29256, 'loss/train': 1.4149203300476074}
+02/25/2022 15:29:36 - INFO - codeparrot_training - Step 29257: {'lr': 0.00019711016166407247, 'samples': 14980096, 'steps': 29257, 'loss/train': 1.1924718618392944}
+02/25/2022 15:29:41 - INFO - codeparrot_training - Step 29258: {'lr': 0.00019709416967746515, 'samples': 14980608, 'steps': 29258, 'loss/train': 0.6543747186660767}
+02/25/2022 15:29:48 - INFO - codeparrot_training - Step 29259: {'lr': 0.0001970781779174896, 'samples': 14981120, 'steps': 29259, 'loss/train': 1.014675498008728}
+02/25/2022 15:29:52 - INFO - codeparrot_training - Step 29260: {'lr': 0.00019706218638421448, 'samples': 14981632, 'steps': 29260, 'loss/train': 2.5467488765716553}
+02/25/2022 15:29:57 - INFO - codeparrot_training - Step 29261: {'lr': 0.00019704619507770799, 'samples': 14982144, 'steps': 29261, 'loss/train': 2.1105871200561523}
+02/25/2022 15:30:01 - INFO - codeparrot_training - Step 29262: {'lr': 0.00019703020399803881, 'samples': 14982656, 'steps': 29262, 'loss/train': 3.2613344192504883}
+02/25/2022 15:30:06 - INFO - codeparrot_training - Step 29263: {'lr': 0.0001970142131452754, 'samples': 14983168, 'steps': 29263, 'loss/train': 1.2367548942565918}
+02/25/2022 15:30:10 - INFO - codeparrot_training - Step 29264: {'lr': 0.0001969982225194864, 'samples': 14983680, 'steps': 29264, 'loss/train': 1.9033775329589844}
+02/25/2022 15:30:15 - INFO - codeparrot_training - Step 29265: {'lr': 0.00019698223212074006, 'samples': 14984192, 'steps': 29265, 'loss/train': 1.407662034034729}
+02/25/2022 15:30:19 - INFO - codeparrot_training - Step 29266: {'lr': 0.00019696624194910504, 'samples': 14984704, 'steps': 29266, 'loss/train': 1.8565514087677002}
+02/25/2022 15:30:24 - INFO - codeparrot_training - Step 29267: {'lr': 0.00019695025200464978, 'samples': 14985216, 'steps': 29267, 'loss/train': 1.8945730924606323}
+02/25/2022 15:30:28 - INFO - codeparrot_training - Step 29268: {'lr': 0.00019693426228744278, 'samples': 14985728, 'steps': 29268, 'loss/train': 1.8950185775756836}
+02/25/2022 15:30:34 - INFO - codeparrot_training - Step 29269: {'lr': 0.00019691827279755267, 'samples': 14986240, 'steps': 29269, 'loss/train': 0.8250755667686462}
+02/25/2022 15:30:38 - INFO - codeparrot_training - Step 29270: {'lr': 0.0001969022835350477, 'samples': 14986752, 'steps': 29270, 'loss/train': 3.0484871864318848}
+02/25/2022 15:30:43 - INFO - codeparrot_training - Step 29271: {'lr': 0.0001968862944999965, 'samples': 14987264, 'steps': 29271, 'loss/train': 0.8977957963943481}
+02/25/2022 15:30:47 - INFO - codeparrot_training - Step 29272: {'lr': 0.00019687030569246755, 'samples': 14987776, 'steps': 29272, 'loss/train': 2.0800747871398926}
+02/25/2022 15:30:52 - INFO - codeparrot_training - Step 29273: {'lr': 0.00019685431711252943, 'samples': 14988288, 'steps': 29273, 'loss/train': 2.110466957092285}
+02/25/2022 15:30:56 - INFO - codeparrot_training - Step 29274: {'lr': 0.00019683832876025048, 'samples': 14988800, 'steps': 29274, 'loss/train': 1.6376875638961792}
+02/25/2022 15:31:01 - INFO - codeparrot_training - Step 29275: {'lr': 0.0001968223406356992, 'samples': 14989312, 'steps': 29275, 'loss/train': 1.8917607069015503}
+02/25/2022 15:31:05 - INFO - codeparrot_training - Step 29276: {'lr': 0.00019680635273894415, 'samples': 14989824, 'steps': 29276, 'loss/train': 1.366753101348877}
+02/25/2022 15:31:10 - INFO - codeparrot_training - Step 29277: {'lr': 0.0001967903650700539, 'samples': 14990336, 'steps': 29277, 'loss/train': 1.572314739227295}
+02/25/2022 15:31:13 - INFO - codeparrot_training - Step 29278: {'lr': 0.00019677437762909673, 'samples': 14990848, 'steps': 29278, 'loss/train': 1.0510770082473755}
+02/25/2022 15:31:20 - INFO - codeparrot_training - Step 29279: {'lr': 0.00019675839041614118, 'samples': 14991360, 'steps': 29279, 'loss/train': 2.573420763015747}
+02/25/2022 15:31:24 - INFO - codeparrot_training - Step 29280: {'lr': 0.00019674240343125587, 'samples': 14991872, 'steps': 29280, 'loss/train': 1.5276926755905151}
+02/25/2022 15:31:29 - INFO - codeparrot_training - Step 29281: {'lr': 0.00019672641667450915, 'samples': 14992384, 'steps': 29281, 'loss/train': 0.049334824085235596}
+02/25/2022 15:31:33 - INFO - codeparrot_training - Step 29282: {'lr': 0.0001967104301459696, 'samples': 14992896, 'steps': 29282, 'loss/train': 1.5207090377807617}
+02/25/2022 15:31:38 - INFO - codeparrot_training - Step 29283: {'lr': 0.00019669444384570557, 'samples': 14993408, 'steps': 29283, 'loss/train': 1.9674229621887207}
+02/25/2022 15:31:42 - INFO - codeparrot_training - Step 29284: {'lr': 0.0001966784577737857, 'samples': 14993920, 'steps': 29284, 'loss/train': 1.933316946029663}
+02/25/2022 15:31:47 - INFO - codeparrot_training - Step 29285: {'lr': 0.00019666247193027829, 'samples': 14994432, 'steps': 29285, 'loss/train': 3.544548749923706}
+02/25/2022 15:31:51 - INFO - codeparrot_training - Step 29286: {'lr': 0.00019664648631525199, 'samples': 14994944, 'steps': 29286, 'loss/train': 2.274874687194824}
+02/25/2022 15:31:56 - INFO - codeparrot_training - Step 29287: {'lr': 0.00019663050092877515, 'samples': 14995456, 'steps': 29287, 'loss/train': 1.3459421396255493}
+02/25/2022 15:32:00 - INFO - codeparrot_training - Step 29288: {'lr': 0.00019661451577091633, 'samples': 14995968, 'steps': 29288, 'loss/train': 1.2078479528427124}
+02/25/2022 15:32:06 - INFO - codeparrot_training - Step 29289: {'lr': 0.0001965985308417439, 'samples': 14996480, 'steps': 29289, 'loss/train': 1.757480502128601}
+02/25/2022 15:32:09 - INFO - codeparrot_training - Step 29290: {'lr': 0.0001965825461413265, 'samples': 14996992, 'steps': 29290, 'loss/train': 0.9850122332572937}
+02/25/2022 15:32:15 - INFO - codeparrot_training - Step 29291: {'lr': 0.00019656656166973252, 'samples': 14997504, 'steps': 29291, 'loss/train': 6.274078369140625}
+02/25/2022 15:32:18 - INFO - codeparrot_training - Step 29292: {'lr': 0.00019655057742703032, 'samples': 14998016, 'steps': 29292, 'loss/train': 2.1278178691864014}
+02/25/2022 15:32:24 - INFO - codeparrot_training - Step 29293: {'lr': 0.00019653459341328853, 'samples': 14998528, 'steps': 29293, 'loss/train': 1.459757685661316}
+02/25/2022 15:32:27 - INFO - codeparrot_training - Step 29294: {'lr': 0.00019651860962857553, 'samples': 14999040, 'steps': 29294, 'loss/train': 2.2623355388641357}
+02/25/2022 15:32:33 - INFO - codeparrot_training - Step 29295: {'lr': 0.00019650262607295995, 'samples': 14999552, 'steps': 29295, 'loss/train': 2.201338529586792}
+02/25/2022 15:32:37 - INFO - codeparrot_training - Step 29296: {'lr': 0.00019648664274651002, 'samples': 15000064, 'steps': 29296, 'loss/train': 2.0335028171539307}
+02/25/2022 15:32:42 - INFO - codeparrot_training - Step 29297: {'lr': 0.00019647065964929433, 'samples': 15000576, 'steps': 29297, 'loss/train': 2.2171027660369873}
+02/25/2022 15:32:46 - INFO - codeparrot_training - Step 29298: {'lr': 0.00019645467678138133, 'samples': 15001088, 'steps': 29298, 'loss/train': 1.4360616207122803}
+02/25/2022 15:32:51 - INFO - codeparrot_training - Step 29299: {'lr': 0.00019643869414283957, 'samples': 15001600, 'steps': 29299, 'loss/train': 1.6715625524520874}
+02/25/2022 15:32:55 - INFO - codeparrot_training - Step 29300: {'lr': 0.00019642271173373735, 'samples': 15002112, 'steps': 29300, 'loss/train': 1.9193049669265747}
+02/25/2022 15:33:00 - INFO - codeparrot_training - Step 29301: {'lr': 0.00019640672955414324, 'samples': 15002624, 'steps': 29301, 'loss/train': 1.0728017091751099}
+02/25/2022 15:33:04 - INFO - codeparrot_training - Step 29302: {'lr': 0.00019639074760412567, 'samples': 15003136, 'steps': 29302, 'loss/train': 0.8722072243690491}
+02/25/2022 15:33:09 - INFO - codeparrot_training - Step 29303: {'lr': 0.00019637476588375312, 'samples': 15003648, 'steps': 29303, 'loss/train': 2.069180488586426}
+02/25/2022 15:33:13 - INFO - codeparrot_training - Step 29304: {'lr': 0.00019635878439309412, 'samples': 15004160, 'steps': 29304, 'loss/train': 0.8410064578056335}
+02/25/2022 15:33:19 - INFO - codeparrot_training - Step 29305: {'lr': 0.00019634280313221697, 'samples': 15004672, 'steps': 29305, 'loss/train': 1.8132877349853516}
+02/25/2022 15:33:22 - INFO - codeparrot_training - Step 29306: {'lr': 0.00019632682210119021, 'samples': 15005184, 'steps': 29306, 'loss/train': 0.051910340785980225}
+02/25/2022 15:33:28 - INFO - codeparrot_training - Step 29307: {'lr': 0.0001963108413000823, 'samples': 15005696, 'steps': 29307, 'loss/train': 1.4317172765731812}
+02/25/2022 15:33:31 - INFO - codeparrot_training - Step 29308: {'lr': 0.00019629486072896183, 'samples': 15006208, 'steps': 29308, 'loss/train': 2.275067090988159}
+02/25/2022 15:33:37 - INFO - codeparrot_training - Step 29309: {'lr': 0.00019627888038789698, 'samples': 15006720, 'steps': 29309, 'loss/train': 1.3946512937545776}
+02/25/2022 15:33:40 - INFO - codeparrot_training - Step 29310: {'lr': 0.00019626290027695636, 'samples': 15007232, 'steps': 29310, 'loss/train': 0.10922368615865707}
+02/25/2022 15:33:46 - INFO - codeparrot_training - Step 29311: {'lr': 0.00019624692039620843, 'samples': 15007744, 'steps': 29311, 'loss/train': 0.8043493628501892}
+02/25/2022 15:33:50 - INFO - codeparrot_training - Step 29312: {'lr': 0.00019623094074572173, 'samples': 15008256, 'steps': 29312, 'loss/train': 1.902226448059082}
+02/25/2022 15:33:55 - INFO - codeparrot_training - Step 29313: {'lr': 0.00019621496132556446, 'samples': 15008768, 'steps': 29313, 'loss/train': 1.8014180660247803}
+02/25/2022 15:33:58 - INFO - codeparrot_training - Step 29314: {'lr': 0.00019619898213580527, 'samples': 15009280, 'steps': 29314, 'loss/train': 2.7981128692626953}
+02/25/2022 15:34:04 - INFO - codeparrot_training - Step 29315: {'lr': 0.0001961830031765125, 'samples': 15009792, 'steps': 29315, 'loss/train': 1.6252634525299072}
+02/25/2022 15:34:07 - INFO - codeparrot_training - Step 29316: {'lr': 0.0001961670244477547, 'samples': 15010304, 'steps': 29316, 'loss/train': 1.3770695924758911}
+02/25/2022 15:34:14 - INFO - codeparrot_training - Step 29317: {'lr': 0.00019615104594960037, 'samples': 15010816, 'steps': 29317, 'loss/train': 1.3650835752487183}
+02/25/2022 15:34:17 - INFO - codeparrot_training - Step 29318: {'lr': 0.00019613506768211776, 'samples': 15011328, 'steps': 29318, 'loss/train': 0.790747344493866}
+02/25/2022 15:34:23 - INFO - codeparrot_training - Step 29319: {'lr': 0.0001961190896453754, 'samples': 15011840, 'steps': 29319, 'loss/train': 0.8436750769615173}
+02/25/2022 15:34:26 - INFO - codeparrot_training - Step 29320: {'lr': 0.0001961031118394418, 'samples': 15012352, 'steps': 29320, 'loss/train': 1.0435761213302612}
+02/25/2022 15:34:32 - INFO - codeparrot_training - Step 29321: {'lr': 0.00019608713426438534, 'samples': 15012864, 'steps': 29321, 'loss/train': 1.2243672609329224}
+02/25/2022 15:34:35 - INFO - codeparrot_training - Step 29322: {'lr': 0.00019607115692027446, 'samples': 15013376, 'steps': 29322, 'loss/train': 1.5739738941192627}
+02/25/2022 15:34:41 - INFO - codeparrot_training - Step 29323: {'lr': 0.00019605517980717763, 'samples': 15013888, 'steps': 29323, 'loss/train': 1.5121190547943115}
+02/25/2022 15:34:44 - INFO - codeparrot_training - Step 29324: {'lr': 0.0001960392029251633, 'samples': 15014400, 'steps': 29324, 'loss/train': 1.2646921873092651}
+02/25/2022 15:34:50 - INFO - codeparrot_training - Step 29325: {'lr': 0.00019602322627429992, 'samples': 15014912, 'steps': 29325, 'loss/train': 1.9140151739120483}
+02/25/2022 15:34:56 - INFO - codeparrot_training - Step 29326: {'lr': 0.0001960072498546558, 'samples': 15015424, 'steps': 29326, 'loss/train': 2.251607894897461}
+02/25/2022 15:34:59 - INFO - codeparrot_training - Step 29327: {'lr': 0.00019599127366629957, 'samples': 15015936, 'steps': 29327, 'loss/train': 1.3281522989273071}
+02/25/2022 15:35:05 - INFO - codeparrot_training - Step 29328: {'lr': 0.0001959752977092995, 'samples': 15016448, 'steps': 29328, 'loss/train': 1.7271817922592163}
+02/25/2022 15:35:08 - INFO - codeparrot_training - Step 29329: {'lr': 0.00019595932198372413, 'samples': 15016960, 'steps': 29329, 'loss/train': 1.743116021156311}
+02/25/2022 15:35:14 - INFO - codeparrot_training - Step 29330: {'lr': 0.00019594334648964187, 'samples': 15017472, 'steps': 29330, 'loss/train': 1.4380152225494385}
+02/25/2022 15:35:18 - INFO - codeparrot_training - Step 29331: {'lr': 0.00019592737122712118, 'samples': 15017984, 'steps': 29331, 'loss/train': 2.723742723464966}
+02/25/2022 15:35:21 - INFO - codeparrot_training - Step 29332: {'lr': 0.00019591139619623042, 'samples': 15018496, 'steps': 29332, 'loss/train': 1.4156464338302612}
+02/25/2022 15:35:27 - INFO - codeparrot_training - Step 29333: {'lr': 0.00019589542139703803, 'samples': 15019008, 'steps': 29333, 'loss/train': 0.8630192875862122}
+02/25/2022 15:35:30 - INFO - codeparrot_training - Step 29334: {'lr': 0.00019587944682961263, 'samples': 15019520, 'steps': 29334, 'loss/train': 1.9504932165145874}
+02/25/2022 15:35:36 - INFO - codeparrot_training - Step 29335: {'lr': 0.00019586347249402235, 'samples': 15020032, 'steps': 29335, 'loss/train': 1.0513288974761963}
+02/25/2022 15:35:39 - INFO - codeparrot_training - Step 29336: {'lr': 0.00019584749839033575, 'samples': 15020544, 'steps': 29336, 'loss/train': 2.233471155166626}
+02/25/2022 15:35:45 - INFO - codeparrot_training - Step 29337: {'lr': 0.0001958315245186213, 'samples': 15021056, 'steps': 29337, 'loss/train': 1.8266321420669556}
+02/25/2022 15:35:49 - INFO - codeparrot_training - Step 29338: {'lr': 0.00019581555087894747, 'samples': 15021568, 'steps': 29338, 'loss/train': 1.6159883737564087}
+02/25/2022 15:35:54 - INFO - codeparrot_training - Step 29339: {'lr': 0.00019579957747138256, 'samples': 15022080, 'steps': 29339, 'loss/train': 1.0751569271087646}
+02/25/2022 15:35:58 - INFO - codeparrot_training - Step 29340: {'lr': 0.00019578360429599502, 'samples': 15022592, 'steps': 29340, 'loss/train': 1.4261159896850586}
+02/25/2022 15:36:03 - INFO - codeparrot_training - Step 29341: {'lr': 0.0001957676313528533, 'samples': 15023104, 'steps': 29341, 'loss/train': 1.552079677581787}
+02/25/2022 15:36:09 - INFO - codeparrot_training - Step 29342: {'lr': 0.00019575165864202582, 'samples': 15023616, 'steps': 29342, 'loss/train': 1.9177117347717285}
+02/25/2022 15:36:12 - INFO - codeparrot_training - Step 29343: {'lr': 0.00019573568616358114, 'samples': 15024128, 'steps': 29343, 'loss/train': 1.2271636724472046}
+02/25/2022 15:36:18 - INFO - codeparrot_training - Step 29344: {'lr': 0.00019571971391758744, 'samples': 15024640, 'steps': 29344, 'loss/train': 2.9983057975769043}
+02/25/2022 15:36:21 - INFO - codeparrot_training - Step 29345: {'lr': 0.0001957037419041132, 'samples': 15025152, 'steps': 29345, 'loss/train': 0.7506029009819031}
+02/25/2022 15:36:27 - INFO - codeparrot_training - Step 29346: {'lr': 0.00019568777012322696, 'samples': 15025664, 'steps': 29346, 'loss/train': 1.364060878753662}
+02/25/2022 15:36:30 - INFO - codeparrot_training - Step 29347: {'lr': 0.00019567179857499714, 'samples': 15026176, 'steps': 29347, 'loss/train': 2.2714734077453613}
+02/25/2022 15:36:36 - INFO - codeparrot_training - Step 29348: {'lr': 0.00019565582725949198, 'samples': 15026688, 'steps': 29348, 'loss/train': 1.7118453979492188}
+02/25/2022 15:36:39 - INFO - codeparrot_training - Step 29349: {'lr': 0.00019563985617678, 'samples': 15027200, 'steps': 29349, 'loss/train': 1.5523933172225952}
+02/25/2022 15:36:45 - INFO - codeparrot_training - Step 29350: {'lr': 0.00019562388532692963, 'samples': 15027712, 'steps': 29350, 'loss/train': 1.0967680215835571}
+02/25/2022 15:36:48 - INFO - codeparrot_training - Step 29351: {'lr': 0.00019560791471000925, 'samples': 15028224, 'steps': 29351, 'loss/train': 1.3886933326721191}
+02/25/2022 15:36:54 - INFO - codeparrot_training - Step 29352: {'lr': 0.0001955919443260874, 'samples': 15028736, 'steps': 29352, 'loss/train': 0.7895808219909668}
+02/25/2022 15:36:58 - INFO - codeparrot_training - Step 29353: {'lr': 0.0001955759741752323, 'samples': 15029248, 'steps': 29353, 'loss/train': 1.2784093618392944}
+02/25/2022 15:37:03 - INFO - codeparrot_training - Step 29354: {'lr': 0.00019556000425751242, 'samples': 15029760, 'steps': 29354, 'loss/train': 1.3381541967391968}
+02/25/2022 15:37:07 - INFO - codeparrot_training - Step 29355: {'lr': 0.00019554403457299618, 'samples': 15030272, 'steps': 29355, 'loss/train': 2.130079507827759}
+02/25/2022 15:37:12 - INFO - codeparrot_training - Step 29356: {'lr': 0.00019552806512175214, 'samples': 15030784, 'steps': 29356, 'loss/train': 1.572493553161621}
+02/25/2022 15:37:16 - INFO - codeparrot_training - Step 29357: {'lr': 0.00019551209590384848, 'samples': 15031296, 'steps': 29357, 'loss/train': 1.359002947807312}
+02/25/2022 15:37:21 - INFO - codeparrot_training - Step 29358: {'lr': 0.00019549612691935363, 'samples': 15031808, 'steps': 29358, 'loss/train': 2.0528948307037354}
+02/25/2022 15:37:25 - INFO - codeparrot_training - Step 29359: {'lr': 0.00019548015816833614, 'samples': 15032320, 'steps': 29359, 'loss/train': 1.699252963066101}
+02/25/2022 15:37:30 - INFO - codeparrot_training - Step 29360: {'lr': 0.00019546418965086444, 'samples': 15032832, 'steps': 29360, 'loss/train': 1.3062719106674194}
+02/25/2022 15:37:34 - INFO - codeparrot_training - Step 29361: {'lr': 0.00019544822136700668, 'samples': 15033344, 'steps': 29361, 'loss/train': 1.5943058729171753}
+02/25/2022 15:37:40 - INFO - codeparrot_training - Step 29362: {'lr': 0.00019543225331683145, 'samples': 15033856, 'steps': 29362, 'loss/train': 3.369662046432495}
+02/25/2022 15:37:43 - INFO - codeparrot_training - Step 29363: {'lr': 0.0001954162855004071, 'samples': 15034368, 'steps': 29363, 'loss/train': 1.9614683389663696}
+02/25/2022 15:37:49 - INFO - codeparrot_training - Step 29364: {'lr': 0.00019540031791780214, 'samples': 15034880, 'steps': 29364, 'loss/train': 2.183573007583618}
+02/25/2022 15:37:52 - INFO - codeparrot_training - Step 29365: {'lr': 0.00019538435056908485, 'samples': 15035392, 'steps': 29365, 'loss/train': 1.6338070631027222}
+02/25/2022 15:37:58 - INFO - codeparrot_training - Step 29366: {'lr': 0.00019536838345432362, 'samples': 15035904, 'steps': 29366, 'loss/train': 0.5155447721481323}
+02/25/2022 15:38:01 - INFO - codeparrot_training - Step 29367: {'lr': 0.00019535241657358694, 'samples': 15036416, 'steps': 29367, 'loss/train': 1.9426361322402954}
+02/25/2022 15:38:07 - INFO - codeparrot_training - Step 29368: {'lr': 0.0001953364499269431, 'samples': 15036928, 'steps': 29368, 'loss/train': 0.9669619798660278}
+02/25/2022 15:38:10 - INFO - codeparrot_training - Step 29369: {'lr': 0.0001953204835144606, 'samples': 15037440, 'steps': 29369, 'loss/train': 2.10241436958313}
+02/25/2022 15:38:16 - INFO - codeparrot_training - Step 29370: {'lr': 0.0001953045173362077, 'samples': 15037952, 'steps': 29370, 'loss/train': 1.0091904401779175}
+02/25/2022 15:38:19 - INFO - codeparrot_training - Step 29371: {'lr': 0.000195288551392253, 'samples': 15038464, 'steps': 29371, 'loss/train': 2.2979555130004883}
+02/25/2022 15:38:25 - INFO - codeparrot_training - Step 29372: {'lr': 0.0001952725856826647, 'samples': 15038976, 'steps': 29372, 'loss/train': 2.239908218383789}
+02/25/2022 15:38:29 - INFO - codeparrot_training - Step 29373: {'lr': 0.00019525662020751129, 'samples': 15039488, 'steps': 29373, 'loss/train': 1.7329139709472656}
+02/25/2022 15:38:34 - INFO - codeparrot_training - Step 29374: {'lr': 0.00019524065496686114, 'samples': 15040000, 'steps': 29374, 'loss/train': 1.4030593633651733}
+02/25/2022 15:38:38 - INFO - codeparrot_training - Step 29375: {'lr': 0.00019522468996078257, 'samples': 15040512, 'steps': 29375, 'loss/train': 2.3346424102783203}
+02/25/2022 15:38:43 - INFO - codeparrot_training - Step 29376: {'lr': 0.00019520872518934406, 'samples': 15041024, 'steps': 29376, 'loss/train': 2.5046608448028564}
+02/25/2022 15:38:47 - INFO - codeparrot_training - Step 29377: {'lr': 0.00019519276065261399, 'samples': 15041536, 'steps': 29377, 'loss/train': 1.0978639125823975}
+02/25/2022 15:38:52 - INFO - codeparrot_training - Step 29378: {'lr': 0.00019517679635066082, 'samples': 15042048, 'steps': 29378, 'loss/train': 1.665866494178772}
+02/25/2022 15:38:56 - INFO - codeparrot_training - Step 29379: {'lr': 0.00019516083228355275, 'samples': 15042560, 'steps': 29379, 'loss/train': 1.7864251136779785}
+02/25/2022 15:39:01 - INFO - codeparrot_training - Step 29380: {'lr': 0.00019514486845135823, 'samples': 15043072, 'steps': 29380, 'loss/train': 2.371065616607666}
+02/25/2022 15:39:04 - INFO - codeparrot_training - Step 29381: {'lr': 0.0001951289048541457, 'samples': 15043584, 'steps': 29381, 'loss/train': 2.5646724700927734}
+02/25/2022 15:39:11 - INFO - codeparrot_training - Step 29382: {'lr': 0.0001951129414919836, 'samples': 15044096, 'steps': 29382, 'loss/train': 1.8295261859893799}
+02/25/2022 15:39:14 - INFO - codeparrot_training - Step 29383: {'lr': 0.00019509697836494016, 'samples': 15044608, 'steps': 29383, 'loss/train': 2.626011848449707}
+02/25/2022 15:39:20 - INFO - codeparrot_training - Step 29384: {'lr': 0.00019508101547308383, 'samples': 15045120, 'steps': 29384, 'loss/train': 1.493629813194275}
+02/25/2022 15:39:25 - INFO - codeparrot_training - Step 29385: {'lr': 0.00019506505281648295, 'samples': 15045632, 'steps': 29385, 'loss/train': 1.805092453956604}
+02/25/2022 15:39:29 - INFO - codeparrot_training - Step 29386: {'lr': 0.00019504909039520609, 'samples': 15046144, 'steps': 29386, 'loss/train': 1.8686336278915405}
+02/25/2022 15:39:34 - INFO - codeparrot_training - Step 29387: {'lr': 0.00019503312820932133, 'samples': 15046656, 'steps': 29387, 'loss/train': 3.2920823097229004}
+02/25/2022 15:39:38 - INFO - codeparrot_training - Step 29388: {'lr': 0.00019501716625889724, 'samples': 15047168, 'steps': 29388, 'loss/train': 1.1561026573181152}
+02/25/2022 15:39:43 - INFO - codeparrot_training - Step 29389: {'lr': 0.0001950012045440021, 'samples': 15047680, 'steps': 29389, 'loss/train': 2.1098732948303223}
+02/25/2022 15:39:47 - INFO - codeparrot_training - Step 29390: {'lr': 0.00019498524306470438, 'samples': 15048192, 'steps': 29390, 'loss/train': 1.7553433179855347}
+02/25/2022 15:39:53 - INFO - codeparrot_training - Step 29391: {'lr': 0.00019496928182107252, 'samples': 15048704, 'steps': 29391, 'loss/train': 0.07703979313373566}
+02/25/2022 15:39:57 - INFO - codeparrot_training - Step 29392: {'lr': 0.00019495332081317463, 'samples': 15049216, 'steps': 29392, 'loss/train': 2.7321858406066895}
+02/25/2022 15:40:02 - INFO - codeparrot_training - Step 29393: {'lr': 0.00019493736004107926, 'samples': 15049728, 'steps': 29393, 'loss/train': 1.851007342338562}
+02/25/2022 15:40:06 - INFO - codeparrot_training - Step 29394: {'lr': 0.00019492139950485476, 'samples': 15050240, 'steps': 29394, 'loss/train': 1.576150894165039}
+02/25/2022 15:40:11 - INFO - codeparrot_training - Step 29395: {'lr': 0.0001949054392045696, 'samples': 15050752, 'steps': 29395, 'loss/train': 1.3481794595718384}
+02/25/2022 15:40:15 - INFO - codeparrot_training - Step 29396: {'lr': 0.00019488947914029193, 'samples': 15051264, 'steps': 29396, 'loss/train': 2.079875946044922}
+02/25/2022 15:40:20 - INFO - codeparrot_training - Step 29397: {'lr': 0.00019487351931209024, 'samples': 15051776, 'steps': 29397, 'loss/train': 2.4200375080108643}
+02/25/2022 15:40:24 - INFO - codeparrot_training - Step 29398: {'lr': 0.00019485755972003288, 'samples': 15052288, 'steps': 29398, 'loss/train': 1.5261116027832031}
+02/25/2022 15:40:29 - INFO - codeparrot_training - Step 29399: {'lr': 0.00019484160036418836, 'samples': 15052800, 'steps': 29399, 'loss/train': 2.0629138946533203}
+02/25/2022 15:40:33 - INFO - codeparrot_training - Step 29400: {'lr': 0.00019482564124462476, 'samples': 15053312, 'steps': 29400, 'loss/train': 2.50455641746521}
+02/25/2022 15:40:38 - INFO - codeparrot_training - Step 29401: {'lr': 0.00019480968236141062, 'samples': 15053824, 'steps': 29401, 'loss/train': 1.1354399919509888}
+02/25/2022 15:40:41 - INFO - codeparrot_training - Step 29402: {'lr': 0.00019479372371461427, 'samples': 15054336, 'steps': 29402, 'loss/train': 2.0294108390808105}
+02/25/2022 15:40:47 - INFO - codeparrot_training - Step 29403: {'lr': 0.00019477776530430407, 'samples': 15054848, 'steps': 29403, 'loss/train': 0.8522345423698425}
+02/25/2022 15:40:50 - INFO - codeparrot_training - Step 29404: {'lr': 0.00019476180713054851, 'samples': 15055360, 'steps': 29404, 'loss/train': 3.1087419986724854}
+02/25/2022 15:40:56 - INFO - codeparrot_training - Step 29405: {'lr': 0.0001947458491934157, 'samples': 15055872, 'steps': 29405, 'loss/train': 2.4730608463287354}
+02/25/2022 15:40:59 - INFO - codeparrot_training - Step 29406: {'lr': 0.00019472989149297415, 'samples': 15056384, 'steps': 29406, 'loss/train': 1.6713955402374268}
+02/25/2022 15:41:05 - INFO - codeparrot_training - Step 29407: {'lr': 0.00019471393402929223, 'samples': 15056896, 'steps': 29407, 'loss/train': 1.9496151208877563}
+02/25/2022 15:41:09 - INFO - codeparrot_training - Step 29408: {'lr': 0.00019469797680243827, 'samples': 15057408, 'steps': 29408, 'loss/train': 1.1949479579925537}
+02/25/2022 15:41:14 - INFO - codeparrot_training - Step 29409: {'lr': 0.00019468201981248057, 'samples': 15057920, 'steps': 29409, 'loss/train': 1.6739237308502197}
+02/25/2022 15:41:18 - INFO - codeparrot_training - Step 29410: {'lr': 0.0001946660630594875, 'samples': 15058432, 'steps': 29410, 'loss/train': 1.9693725109100342}
+02/25/2022 15:41:24 - INFO - codeparrot_training - Step 29411: {'lr': 0.00019465010654352751, 'samples': 15058944, 'steps': 29411, 'loss/train': 2.389209508895874}
+02/25/2022 15:41:27 - INFO - codeparrot_training - Step 29412: {'lr': 0.00019463415026466885, 'samples': 15059456, 'steps': 29412, 'loss/train': 0.26547902822494507}
+02/25/2022 15:41:33 - INFO - codeparrot_training - Step 29413: {'lr': 0.00019461819422297998, 'samples': 15059968, 'steps': 29413, 'loss/train': 2.3412394523620605}
+02/25/2022 15:41:36 - INFO - codeparrot_training - Step 29414: {'lr': 0.00019460223841852908, 'samples': 15060480, 'steps': 29414, 'loss/train': 0.8751606345176697}
+02/25/2022 15:41:42 - INFO - codeparrot_training - Step 29415: {'lr': 0.00019458628285138469, 'samples': 15060992, 'steps': 29415, 'loss/train': 1.8667467832565308}
+02/25/2022 15:41:45 - INFO - codeparrot_training - Step 29416: {'lr': 0.00019457032752161503, 'samples': 15061504, 'steps': 29416, 'loss/train': 1.6235064268112183}
+02/25/2022 15:41:52 - INFO - codeparrot_training - Step 29417: {'lr': 0.0001945543724292885, 'samples': 15062016, 'steps': 29417, 'loss/train': 1.1945921182632446}
+02/25/2022 15:41:55 - INFO - codeparrot_training - Step 29418: {'lr': 0.00019453841757447344, 'samples': 15062528, 'steps': 29418, 'loss/train': 1.0461317300796509}
+02/25/2022 15:42:00 - INFO - codeparrot_training - Step 29419: {'lr': 0.00019452246295723814, 'samples': 15063040, 'steps': 29419, 'loss/train': 1.671424150466919}
+02/25/2022 15:42:04 - INFO - codeparrot_training - Step 29420: {'lr': 0.00019450650857765102, 'samples': 15063552, 'steps': 29420, 'loss/train': 1.7188565731048584}
+02/25/2022 15:42:09 - INFO - codeparrot_training - Step 29421: {'lr': 0.0001944905544357804, 'samples': 15064064, 'steps': 29421, 'loss/train': 0.5750018358230591}
+02/25/2022 15:42:13 - INFO - codeparrot_training - Step 29422: {'lr': 0.00019447460053169464, 'samples': 15064576, 'steps': 29422, 'loss/train': 1.7493497133255005}
+02/25/2022 15:42:19 - INFO - codeparrot_training - Step 29423: {'lr': 0.00019445864686546202, 'samples': 15065088, 'steps': 29423, 'loss/train': 2.12671160697937}
+02/25/2022 15:42:24 - INFO - codeparrot_training - Step 29424: {'lr': 0.0001944426934371509, 'samples': 15065600, 'steps': 29424, 'loss/train': 1.4279041290283203}
+02/25/2022 15:42:28 - INFO - codeparrot_training - Step 29425: {'lr': 0.00019442674024682967, 'samples': 15066112, 'steps': 29425, 'loss/train': 1.9587042331695557}
+02/25/2022 15:42:33 - INFO - codeparrot_training - Step 29426: {'lr': 0.0001944107872945667, 'samples': 15066624, 'steps': 29426, 'loss/train': 1.8342225551605225}
+02/25/2022 15:42:36 - INFO - codeparrot_training - Step 29427: {'lr': 0.0001943948345804302, 'samples': 15067136, 'steps': 29427, 'loss/train': 1.7993742227554321}
+02/25/2022 15:42:43 - INFO - codeparrot_training - Step 29428: {'lr': 0.00019437888210448854, 'samples': 15067648, 'steps': 29428, 'loss/train': 2.1386380195617676}
+02/25/2022 15:42:46 - INFO - codeparrot_training - Step 29429: {'lr': 0.00019436292986681008, 'samples': 15068160, 'steps': 29429, 'loss/train': 1.4390922784805298}
+02/25/2022 15:42:52 - INFO - codeparrot_training - Step 29430: {'lr': 0.0001943469778674633, 'samples': 15068672, 'steps': 29430, 'loss/train': 1.1585594415664673}
+02/25/2022 15:42:55 - INFO - codeparrot_training - Step 29431: {'lr': 0.0001943310261065163, 'samples': 15069184, 'steps': 29431, 'loss/train': 2.0845022201538086}
+02/25/2022 15:43:01 - INFO - codeparrot_training - Step 29432: {'lr': 0.00019431507458403749, 'samples': 15069696, 'steps': 29432, 'loss/train': 3.0166690349578857}
+02/25/2022 15:43:04 - INFO - codeparrot_training - Step 29433: {'lr': 0.0001942991233000952, 'samples': 15070208, 'steps': 29433, 'loss/train': 0.1814434826374054}
+02/25/2022 15:43:10 - INFO - codeparrot_training - Step 29434: {'lr': 0.0001942831722547579, 'samples': 15070720, 'steps': 29434, 'loss/train': 1.883793592453003}
+02/25/2022 15:43:13 - INFO - codeparrot_training - Step 29435: {'lr': 0.0001942672214480937, 'samples': 15071232, 'steps': 29435, 'loss/train': 1.7238143682479858}
+02/25/2022 15:43:19 - INFO - codeparrot_training - Step 29436: {'lr': 0.00019425127088017102, 'samples': 15071744, 'steps': 29436, 'loss/train': 1.0666111707687378}
+02/25/2022 15:43:22 - INFO - codeparrot_training - Step 29437: {'lr': 0.0001942353205510582, 'samples': 15072256, 'steps': 29437, 'loss/train': 1.8013737201690674}
+02/25/2022 15:43:29 - INFO - codeparrot_training - Step 29438: {'lr': 0.00019421937046082356, 'samples': 15072768, 'steps': 29438, 'loss/train': 2.0546984672546387}
+02/25/2022 15:43:32 - INFO - codeparrot_training - Step 29439: {'lr': 0.0001942034206095355, 'samples': 15073280, 'steps': 29439, 'loss/train': 1.9984092712402344}
+02/25/2022 15:43:36 - INFO - codeparrot_training - Step 29440: {'lr': 0.0001941874709972622, 'samples': 15073792, 'steps': 29440, 'loss/train': 2.270327091217041}
+02/25/2022 15:43:41 - INFO - codeparrot_training - Step 29441: {'lr': 0.00019417152162407204, 'samples': 15074304, 'steps': 29441, 'loss/train': 0.8403199315071106}
+02/25/2022 15:43:45 - INFO - codeparrot_training - Step 29442: {'lr': 0.00019415557249003333, 'samples': 15074816, 'steps': 29442, 'loss/train': 1.4506248235702515}
+02/25/2022 15:43:50 - INFO - codeparrot_training - Step 29443: {'lr': 0.00019413962359521452, 'samples': 15075328, 'steps': 29443, 'loss/train': 1.1142593622207642}
+02/25/2022 15:43:54 - INFO - codeparrot_training - Step 29444: {'lr': 0.00019412367493968374, 'samples': 15075840, 'steps': 29444, 'loss/train': 1.184648871421814}
+02/25/2022 15:43:59 - INFO - codeparrot_training - Step 29445: {'lr': 0.0001941077265235094, 'samples': 15076352, 'steps': 29445, 'loss/train': 0.9934272170066833}
+02/25/2022 15:44:05 - INFO - codeparrot_training - Step 29446: {'lr': 0.00019409177834675982, 'samples': 15076864, 'steps': 29446, 'loss/train': 1.4621813297271729}
+02/25/2022 15:44:08 - INFO - codeparrot_training - Step 29447: {'lr': 0.00019407583040950338, 'samples': 15077376, 'steps': 29447, 'loss/train': 1.6000694036483765}
+02/25/2022 15:44:12 - INFO - codeparrot_training - Step 29448: {'lr': 0.00019405988271180825, 'samples': 15077888, 'steps': 29448, 'loss/train': 9.644929885864258}
+02/25/2022 15:44:17 - INFO - codeparrot_training - Step 29449: {'lr': 0.0001940439352537428, 'samples': 15078400, 'steps': 29449, 'loss/train': 1.9958058595657349}
+02/25/2022 15:44:21 - INFO - codeparrot_training - Step 29450: {'lr': 0.00019402798803537538, 'samples': 15078912, 'steps': 29450, 'loss/train': 1.8296359777450562}
+02/25/2022 15:44:26 - INFO - codeparrot_training - Step 29451: {'lr': 0.00019401204105677433, 'samples': 15079424, 'steps': 29451, 'loss/train': 1.334689974784851}
+02/25/2022 15:44:30 - INFO - codeparrot_training - Step 29452: {'lr': 0.0001939960943180079, 'samples': 15079936, 'steps': 29452, 'loss/train': 1.03721022605896}
+02/25/2022 15:44:36 - INFO - codeparrot_training - Step 29453: {'lr': 0.0001939801478191444, 'samples': 15080448, 'steps': 29453, 'loss/train': 1.6068592071533203}
+02/25/2022 15:44:41 - INFO - codeparrot_training - Step 29454: {'lr': 0.00019396420156025213, 'samples': 15080960, 'steps': 29454, 'loss/train': 1.4263707399368286}
+02/25/2022 15:44:45 - INFO - codeparrot_training - Step 29455: {'lr': 0.0001939482555413995, 'samples': 15081472, 'steps': 29455, 'loss/train': 1.5901168584823608}
+02/25/2022 15:44:50 - INFO - codeparrot_training - Step 29456: {'lr': 0.00019393230976265475, 'samples': 15081984, 'steps': 29456, 'loss/train': 1.0355278253555298}
+02/25/2022 15:44:54 - INFO - codeparrot_training - Step 29457: {'lr': 0.0001939163642240861, 'samples': 15082496, 'steps': 29457, 'loss/train': 2.120537757873535}
+02/25/2022 15:44:59 - INFO - codeparrot_training - Step 29458: {'lr': 0.00019390041892576202, 'samples': 15083008, 'steps': 29458, 'loss/train': 1.9373555183410645}
+02/25/2022 15:45:03 - INFO - codeparrot_training - Step 29459: {'lr': 0.00019388447386775066, 'samples': 15083520, 'steps': 29459, 'loss/train': 1.7710665464401245}
+02/25/2022 15:45:08 - INFO - codeparrot_training - Step 29460: {'lr': 0.0001938685290501204, 'samples': 15084032, 'steps': 29460, 'loss/train': 1.3460595607757568}
+02/25/2022 15:45:12 - INFO - codeparrot_training - Step 29461: {'lr': 0.00019385258447293958, 'samples': 15084544, 'steps': 29461, 'loss/train': 0.07201723009347916}
+02/25/2022 15:45:18 - INFO - codeparrot_training - Step 29462: {'lr': 0.00019383664013627647, 'samples': 15085056, 'steps': 29462, 'loss/train': 1.4551562070846558}
+02/25/2022 15:45:21 - INFO - codeparrot_training - Step 29463: {'lr': 0.0001938206960401993, 'samples': 15085568, 'steps': 29463, 'loss/train': 0.5092214345932007}
+02/25/2022 15:45:27 - INFO - codeparrot_training - Step 29464: {'lr': 0.00019380475218477644, 'samples': 15086080, 'steps': 29464, 'loss/train': 2.0804381370544434}
+02/25/2022 15:45:30 - INFO - codeparrot_training - Step 29465: {'lr': 0.0001937888085700763, 'samples': 15086592, 'steps': 29465, 'loss/train': 2.38860821723938}
+02/25/2022 15:45:36 - INFO - codeparrot_training - Step 29466: {'lr': 0.00019377286519616692, 'samples': 15087104, 'steps': 29466, 'loss/train': 1.8411568403244019}
+02/25/2022 15:45:39 - INFO - codeparrot_training - Step 29467: {'lr': 0.00019375692206311673, 'samples': 15087616, 'steps': 29467, 'loss/train': 0.9012255072593689}
+02/25/2022 15:45:45 - INFO - codeparrot_training - Step 29468: {'lr': 0.00019374097917099404, 'samples': 15088128, 'steps': 29468, 'loss/train': 1.1848349571228027}
+02/25/2022 15:45:48 - INFO - codeparrot_training - Step 29469: {'lr': 0.00019372503651986721, 'samples': 15088640, 'steps': 29469, 'loss/train': 2.2058956623077393}
+02/25/2022 15:45:54 - INFO - codeparrot_training - Step 29470: {'lr': 0.0001937090941098044, 'samples': 15089152, 'steps': 29470, 'loss/train': 1.4756314754486084}
+02/25/2022 15:45:57 - INFO - codeparrot_training - Step 29471: {'lr': 0.0001936931519408739, 'samples': 15089664, 'steps': 29471, 'loss/train': 0.39692986011505127}
+02/25/2022 15:46:04 - INFO - codeparrot_training - Step 29472: {'lr': 0.00019367721001314412, 'samples': 15090176, 'steps': 29472, 'loss/train': 1.7813066244125366}
+02/25/2022 15:46:07 - INFO - codeparrot_training - Step 29473: {'lr': 0.00019366126832668325, 'samples': 15090688, 'steps': 29473, 'loss/train': 2.5521273612976074}
+02/25/2022 15:46:13 - INFO - codeparrot_training - Step 29474: {'lr': 0.00019364532688155974, 'samples': 15091200, 'steps': 29474, 'loss/train': 1.4444141387939453}
+02/25/2022 15:46:16 - INFO - codeparrot_training - Step 29475: {'lr': 0.00019362938567784162, 'samples': 15091712, 'steps': 29475, 'loss/train': 1.8973220586776733}
+02/25/2022 15:46:22 - INFO - codeparrot_training - Step 29476: {'lr': 0.00019361344471559733, 'samples': 15092224, 'steps': 29476, 'loss/train': 2.45803165435791}
+02/25/2022 15:46:25 - INFO - codeparrot_training - Step 29477: {'lr': 0.00019359750399489513, 'samples': 15092736, 'steps': 29477, 'loss/train': 0.8903579711914062}
+02/25/2022 15:46:31 - INFO - codeparrot_training - Step 29478: {'lr': 0.00019358156351580343, 'samples': 15093248, 'steps': 29478, 'loss/train': 1.7931666374206543}
+02/25/2022 15:46:34 - INFO - codeparrot_training - Step 29479: {'lr': 0.00019356562327839028, 'samples': 15093760, 'steps': 29479, 'loss/train': 2.4955785274505615}
+02/25/2022 15:46:40 - INFO - codeparrot_training - Step 29480: {'lr': 0.0001935496832827241, 'samples': 15094272, 'steps': 29480, 'loss/train': 2.1500279903411865}
+02/25/2022 15:46:43 - INFO - codeparrot_training - Step 29481: {'lr': 0.00019353374352887312, 'samples': 15094784, 'steps': 29481, 'loss/train': 1.5815048217773438}
+02/25/2022 15:46:49 - INFO - codeparrot_training - Step 29482: {'lr': 0.00019351780401690577, 'samples': 15095296, 'steps': 29482, 'loss/train': 2.6938416957855225}
+02/25/2022 15:46:52 - INFO - codeparrot_training - Step 29483: {'lr': 0.00019350186474689015, 'samples': 15095808, 'steps': 29483, 'loss/train': 0.2253606915473938}
+02/25/2022 15:46:58 - INFO - codeparrot_training - Step 29484: {'lr': 0.00019348592571889455, 'samples': 15096320, 'steps': 29484, 'loss/train': 1.763107180595398}
+02/25/2022 15:47:02 - INFO - codeparrot_training - Step 29485: {'lr': 0.00019346998693298728, 'samples': 15096832, 'steps': 29485, 'loss/train': 2.435241937637329}
+02/25/2022 15:47:07 - INFO - codeparrot_training - Step 29486: {'lr': 0.0001934540483892367, 'samples': 15097344, 'steps': 29486, 'loss/train': 2.925766706466675}
+02/25/2022 15:47:11 - INFO - codeparrot_training - Step 29487: {'lr': 0.0001934381100877111, 'samples': 15097856, 'steps': 29487, 'loss/train': 1.9237571954727173}
+02/25/2022 15:47:17 - INFO - codeparrot_training - Step 29488: {'lr': 0.00019342217202847857, 'samples': 15098368, 'steps': 29488, 'loss/train': 1.8808329105377197}
+02/25/2022 15:47:20 - INFO - codeparrot_training - Step 29489: {'lr': 0.00019340623421160746, 'samples': 15098880, 'steps': 29489, 'loss/train': 0.7876949310302734}
+02/25/2022 15:47:26 - INFO - codeparrot_training - Step 29490: {'lr': 0.00019339029663716613, 'samples': 15099392, 'steps': 29490, 'loss/train': 1.7941373586654663}
+02/25/2022 15:47:29 - INFO - codeparrot_training - Step 29491: {'lr': 0.00019337435930522284, 'samples': 15099904, 'steps': 29491, 'loss/train': 1.0998692512512207}
+02/25/2022 15:47:35 - INFO - codeparrot_training - Step 29492: {'lr': 0.00019335842221584573, 'samples': 15100416, 'steps': 29492, 'loss/train': 1.1510341167449951}
+02/25/2022 15:47:38 - INFO - codeparrot_training - Step 29493: {'lr': 0.00019334248536910314, 'samples': 15100928, 'steps': 29493, 'loss/train': 1.662171483039856}
+02/25/2022 15:47:44 - INFO - codeparrot_training - Step 29494: {'lr': 0.0001933265487650634, 'samples': 15101440, 'steps': 29494, 'loss/train': 0.19030936062335968}
+02/25/2022 15:47:47 - INFO - codeparrot_training - Step 29495: {'lr': 0.0001933106124037948, 'samples': 15101952, 'steps': 29495, 'loss/train': 2.00911808013916}
+02/25/2022 15:47:53 - INFO - codeparrot_training - Step 29496: {'lr': 0.00019329467628536543, 'samples': 15102464, 'steps': 29496, 'loss/train': 1.2284132242202759}
+02/25/2022 15:47:56 - INFO - codeparrot_training - Step 29497: {'lr': 0.00019327874040984367, 'samples': 15102976, 'steps': 29497, 'loss/train': 1.6208912134170532}
+02/25/2022 15:48:03 - INFO - codeparrot_training - Step 29498: {'lr': 0.00019326280477729782, 'samples': 15103488, 'steps': 29498, 'loss/train': 2.196002960205078}
+02/25/2022 15:48:06 - INFO - codeparrot_training - Step 29499: {'lr': 0.00019324686938779607, 'samples': 15104000, 'steps': 29499, 'loss/train': 1.610750675201416}
+02/25/2022 15:48:11 - INFO - codeparrot_training - Step 29500: {'lr': 0.00019323093424140672, 'samples': 15104512, 'steps': 29500, 'loss/train': 2.109882354736328}
+02/25/2022 15:48:15 - INFO - codeparrot_training - Step 29501: {'lr': 0.000193214999338198, 'samples': 15105024, 'steps': 29501, 'loss/train': 2.144434928894043}
+02/25/2022 15:48:20 - INFO - codeparrot_training - Step 29502: {'lr': 0.00019319906467823823, 'samples': 15105536, 'steps': 29502, 'loss/train': 2.1958749294281006}
+02/25/2022 15:48:24 - INFO - codeparrot_training - Step 29503: {'lr': 0.0001931831302615956, 'samples': 15106048, 'steps': 29503, 'loss/train': 1.651770830154419}
+02/25/2022 15:48:29 - INFO - codeparrot_training - Step 29504: {'lr': 0.00019316719608833844, 'samples': 15106560, 'steps': 29504, 'loss/train': 1.9949365854263306}
+02/25/2022 15:48:33 - INFO - codeparrot_training - Step 29505: {'lr': 0.00019315126215853495, 'samples': 15107072, 'steps': 29505, 'loss/train': 1.9250798225402832}
+02/25/2022 15:48:39 - INFO - codeparrot_training - Step 29506: {'lr': 0.00019313532847225334, 'samples': 15107584, 'steps': 29506, 'loss/train': 1.9694836139678955}
+02/25/2022 15:48:42 - INFO - codeparrot_training - Step 29507: {'lr': 0.00019311939502956196, 'samples': 15108096, 'steps': 29507, 'loss/train': 2.1735928058624268}
+02/25/2022 15:48:48 - INFO - codeparrot_training - Step 29508: {'lr': 0.00019310346183052906, 'samples': 15108608, 'steps': 29508, 'loss/train': 0.9547359943389893}
+02/25/2022 15:48:52 - INFO - codeparrot_training - Step 29509: {'lr': 0.00019308752887522288, 'samples': 15109120, 'steps': 29509, 'loss/train': 0.9513463377952576}
+02/25/2022 15:48:58 - INFO - codeparrot_training - Step 29510: {'lr': 0.00019307159616371159, 'samples': 15109632, 'steps': 29510, 'loss/train': 2.1413819789886475}
+02/25/2022 15:49:01 - INFO - codeparrot_training - Step 29511: {'lr': 0.0001930556636960635, 'samples': 15110144, 'steps': 29511, 'loss/train': 2.062797784805298}
+02/25/2022 15:49:07 - INFO - codeparrot_training - Step 29512: {'lr': 0.00019303973147234688, 'samples': 15110656, 'steps': 29512, 'loss/train': 1.8565301895141602}
+02/25/2022 15:49:10 - INFO - codeparrot_training - Step 29513: {'lr': 0.00019302379949263004, 'samples': 15111168, 'steps': 29513, 'loss/train': 2.2574539184570312}
+02/25/2022 15:49:16 - INFO - codeparrot_training - Step 29514: {'lr': 0.0001930078677569811, 'samples': 15111680, 'steps': 29514, 'loss/train': 2.2472429275512695}
+02/25/2022 15:49:19 - INFO - codeparrot_training - Step 29515: {'lr': 0.0001929919362654683, 'samples': 15112192, 'steps': 29515, 'loss/train': 2.3249053955078125}
+02/25/2022 15:49:25 - INFO - codeparrot_training - Step 29516: {'lr': 0.00019297600501816, 'samples': 15112704, 'steps': 29516, 'loss/train': 1.8411891460418701}
+02/25/2022 15:49:28 - INFO - codeparrot_training - Step 29517: {'lr': 0.00019296007401512446, 'samples': 15113216, 'steps': 29517, 'loss/train': 2.0202596187591553}
+02/25/2022 15:49:34 - INFO - codeparrot_training - Step 29518: {'lr': 0.00019294414325642974, 'samples': 15113728, 'steps': 29518, 'loss/train': 2.335059881210327}
+02/25/2022 15:49:37 - INFO - codeparrot_training - Step 29519: {'lr': 0.00019292821274214417, 'samples': 15114240, 'steps': 29519, 'loss/train': 2.269178867340088}
+02/25/2022 15:49:43 - INFO - codeparrot_training - Step 29520: {'lr': 0.00019291228247233605, 'samples': 15114752, 'steps': 29520, 'loss/train': 1.6253571510314941}
+02/25/2022 15:49:47 - INFO - codeparrot_training - Step 29521: {'lr': 0.0001928963524470736, 'samples': 15115264, 'steps': 29521, 'loss/train': 1.1992791891098022}
+02/25/2022 15:49:54 - INFO - codeparrot_training - Step 29522: {'lr': 0.00019288042266642512, 'samples': 15115776, 'steps': 29522, 'loss/train': 0.9252859354019165}
+02/25/2022 15:49:57 - INFO - codeparrot_training - Step 29523: {'lr': 0.00019286449313045867, 'samples': 15116288, 'steps': 29523, 'loss/train': 1.6279892921447754}
+02/25/2022 15:50:01 - INFO - codeparrot_training - Step 29524: {'lr': 0.0001928485638392426, 'samples': 15116800, 'steps': 29524, 'loss/train': 1.7121692895889282}
+02/25/2022 15:50:06 - INFO - codeparrot_training - Step 29525: {'lr': 0.00019283263479284512, 'samples': 15117312, 'steps': 29525, 'loss/train': 1.0754340887069702}
+02/25/2022 15:50:10 - INFO - codeparrot_training - Step 29526: {'lr': 0.0001928167059913346, 'samples': 15117824, 'steps': 29526, 'loss/train': 1.8617494106292725}
+02/25/2022 15:50:15 - INFO - codeparrot_training - Step 29527: {'lr': 0.00019280077743477904, 'samples': 15118336, 'steps': 29527, 'loss/train': 2.3506860733032227}
+02/25/2022 15:50:19 - INFO - codeparrot_training - Step 29528: {'lr': 0.00019278484912324678, 'samples': 15118848, 'steps': 29528, 'loss/train': 2.4082860946655273}
+02/25/2022 15:50:24 - INFO - codeparrot_training - Step 29529: {'lr': 0.00019276892105680606, 'samples': 15119360, 'steps': 29529, 'loss/train': 1.2569336891174316}
+02/25/2022 15:50:30 - INFO - codeparrot_training - Step 29530: {'lr': 0.00019275299323552524, 'samples': 15119872, 'steps': 29530, 'loss/train': 2.0977346897125244}
+02/25/2022 15:50:33 - INFO - codeparrot_training - Step 29531: {'lr': 0.00019273706565947228, 'samples': 15120384, 'steps': 29531, 'loss/train': 1.2070053815841675}
+02/25/2022 15:50:40 - INFO - codeparrot_training - Step 29532: {'lr': 0.00019272113832871556, 'samples': 15120896, 'steps': 29532, 'loss/train': 2.329711675643921}
+02/25/2022 15:50:43 - INFO - codeparrot_training - Step 29533: {'lr': 0.0001927052112433233, 'samples': 15121408, 'steps': 29533, 'loss/train': 1.1488196849822998}
+02/25/2022 15:50:49 - INFO - codeparrot_training - Step 29534: {'lr': 0.0001926892844033637, 'samples': 15121920, 'steps': 29534, 'loss/train': 1.0951377153396606}
+02/25/2022 15:50:52 - INFO - codeparrot_training - Step 29535: {'lr': 0.00019267335780890515, 'samples': 15122432, 'steps': 29535, 'loss/train': 5.3136820793151855}
+02/25/2022 15:50:57 - INFO - codeparrot_training - Step 29536: {'lr': 0.0001926574314600156, 'samples': 15122944, 'steps': 29536, 'loss/train': 2.738659620285034}
+02/25/2022 15:51:01 - INFO - codeparrot_training - Step 29537: {'lr': 0.00019264150535676342, 'samples': 15123456, 'steps': 29537, 'loss/train': 0.5020974278450012}
+02/25/2022 15:51:06 - INFO - codeparrot_training - Step 29538: {'lr': 0.00019262557949921685, 'samples': 15123968, 'steps': 29538, 'loss/train': 0.9084689617156982}
+02/25/2022 15:51:10 - INFO - codeparrot_training - Step 29539: {'lr': 0.0001926096538874441, 'samples': 15124480, 'steps': 29539, 'loss/train': 1.8927985429763794}
+02/25/2022 15:51:15 - INFO - codeparrot_training - Step 29540: {'lr': 0.0001925937285215133, 'samples': 15124992, 'steps': 29540, 'loss/train': 1.073237657546997}
+02/25/2022 15:51:19 - INFO - codeparrot_training - Step 29541: {'lr': 0.00019257780340149275, 'samples': 15125504, 'steps': 29541, 'loss/train': 2.1037042140960693}
+02/25/2022 15:51:25 - INFO - codeparrot_training - Step 29542: {'lr': 0.0001925618785274507, 'samples': 15126016, 'steps': 29542, 'loss/train': 1.9683488607406616}
+02/25/2022 15:51:29 - INFO - codeparrot_training - Step 29543: {'lr': 0.00019254595389945535, 'samples': 15126528, 'steps': 29543, 'loss/train': 0.8652065396308899}
+02/25/2022 15:51:34 - INFO - codeparrot_training - Step 29544: {'lr': 0.0001925300295175748, 'samples': 15127040, 'steps': 29544, 'loss/train': 1.3263882398605347}
+02/25/2022 15:51:37 - INFO - codeparrot_training - Step 29545: {'lr': 0.00019251410538187745, 'samples': 15127552, 'steps': 29545, 'loss/train': 2.102966070175171}
+02/25/2022 15:51:43 - INFO - codeparrot_training - Step 29546: {'lr': 0.00019249818149243134, 'samples': 15128064, 'steps': 29546, 'loss/train': 1.1589629650115967}
+02/25/2022 15:51:46 - INFO - codeparrot_training - Step 29547: {'lr': 0.00019248225784930477, 'samples': 15128576, 'steps': 29547, 'loss/train': 1.2822563648223877}
+02/25/2022 15:51:52 - INFO - codeparrot_training - Step 29548: {'lr': 0.00019246633445256598, 'samples': 15129088, 'steps': 29548, 'loss/train': 1.6994547843933105}
+02/25/2022 15:51:55 - INFO - codeparrot_training - Step 29549: {'lr': 0.00019245041130228313, 'samples': 15129600, 'steps': 29549, 'loss/train': 1.9016213417053223}
+02/25/2022 15:52:01 - INFO - codeparrot_training - Step 29550: {'lr': 0.00019243448839852443, 'samples': 15130112, 'steps': 29550, 'loss/train': 2.334958076477051}
+02/25/2022 15:52:04 - INFO - codeparrot_training - Step 29551: {'lr': 0.00019241856574135808, 'samples': 15130624, 'steps': 29551, 'loss/train': 1.5798158645629883}
+02/25/2022 15:52:10 - INFO - codeparrot_training - Step 29552: {'lr': 0.00019240264333085245, 'samples': 15131136, 'steps': 29552, 'loss/train': 4.411936283111572}
+02/25/2022 15:52:14 - INFO - codeparrot_training - Step 29553: {'lr': 0.00019238672116707545, 'samples': 15131648, 'steps': 29553, 'loss/train': 0.17052246630191803}
+02/25/2022 15:52:20 - INFO - codeparrot_training - Step 29554: {'lr': 0.00019237079925009547, 'samples': 15132160, 'steps': 29554, 'loss/train': 2.920450210571289}
+02/25/2022 15:52:23 - INFO - codeparrot_training - Step 29555: {'lr': 0.00019235487757998069, 'samples': 15132672, 'steps': 29555, 'loss/train': 2.9069278240203857}
+02/25/2022 15:52:29 - INFO - codeparrot_training - Step 29556: {'lr': 0.00019233895615679942, 'samples': 15133184, 'steps': 29556, 'loss/train': 3.27300763130188}
+02/25/2022 15:52:32 - INFO - codeparrot_training - Step 29557: {'lr': 0.00019232303498061963, 'samples': 15133696, 'steps': 29557, 'loss/train': 0.46360430121421814}
+02/25/2022 15:52:38 - INFO - codeparrot_training - Step 29558: {'lr': 0.00019230711405150962, 'samples': 15134208, 'steps': 29558, 'loss/train': 2.1411924362182617}
+02/25/2022 15:52:41 - INFO - codeparrot_training - Step 29559: {'lr': 0.00019229119336953765, 'samples': 15134720, 'steps': 29559, 'loss/train': 0.42131873965263367}
+02/25/2022 15:52:47 - INFO - codeparrot_training - Step 29560: {'lr': 0.00019227527293477187, 'samples': 15135232, 'steps': 29560, 'loss/train': 2.43323016166687}
+02/25/2022 15:52:50 - INFO - codeparrot_training - Step 29561: {'lr': 0.00019225935274728062, 'samples': 15135744, 'steps': 29561, 'loss/train': 0.8900261521339417}
+02/25/2022 15:52:56 - INFO - codeparrot_training - Step 29562: {'lr': 0.00019224343280713186, 'samples': 15136256, 'steps': 29562, 'loss/train': 0.5050694346427917}
+02/25/2022 15:52:59 - INFO - codeparrot_training - Step 29563: {'lr': 0.00019222751311439386, 'samples': 15136768, 'steps': 29563, 'loss/train': 1.0143170356750488}
+02/25/2022 15:53:05 - INFO - codeparrot_training - Step 29564: {'lr': 0.00019221159366913487, 'samples': 15137280, 'steps': 29564, 'loss/train': 0.9791566133499146}
+02/25/2022 15:53:08 - INFO - codeparrot_training - Step 29565: {'lr': 0.00019219567447142317, 'samples': 15137792, 'steps': 29565, 'loss/train': 1.795219898223877}
+02/25/2022 15:53:14 - INFO - codeparrot_training - Step 29566: {'lr': 0.00019217975552132674, 'samples': 15138304, 'steps': 29566, 'loss/train': 1.645015835762024}
+02/25/2022 15:53:17 - INFO - codeparrot_training - Step 29567: {'lr': 0.00019216383681891386, 'samples': 15138816, 'steps': 29567, 'loss/train': 2.1263914108276367}
+02/25/2022 15:53:24 - INFO - codeparrot_training - Step 29568: {'lr': 0.00019214791836425277, 'samples': 15139328, 'steps': 29568, 'loss/train': 1.382897138595581}
+02/25/2022 15:53:27 - INFO - codeparrot_training - Step 29569: {'lr': 0.00019213200015741162, 'samples': 15139840, 'steps': 29569, 'loss/train': 1.7313216924667358}
+02/25/2022 15:53:33 - INFO - codeparrot_training - Step 29570: {'lr': 0.0001921160821984587, 'samples': 15140352, 'steps': 29570, 'loss/train': 2.05067777633667}
+02/25/2022 15:53:36 - INFO - codeparrot_training - Step 29571: {'lr': 0.000192100164487462, 'samples': 15140864, 'steps': 29571, 'loss/train': 1.8695744276046753}
+02/25/2022 15:53:42 - INFO - codeparrot_training - Step 29572: {'lr': 0.00019208424702448978, 'samples': 15141376, 'steps': 29572, 'loss/train': 1.1515432596206665}
+02/25/2022 15:53:45 - INFO - codeparrot_training - Step 29573: {'lr': 0.0001920683298096103, 'samples': 15141888, 'steps': 29573, 'loss/train': 1.4647974967956543}
+02/25/2022 15:53:51 - INFO - codeparrot_training - Step 29574: {'lr': 0.0001920524128428918, 'samples': 15142400, 'steps': 29574, 'loss/train': 1.6085587739944458}
+02/25/2022 15:53:54 - INFO - codeparrot_training - Step 29575: {'lr': 0.00019203649612440225, 'samples': 15142912, 'steps': 29575, 'loss/train': 2.2466847896575928}
+02/25/2022 15:54:00 - INFO - codeparrot_training - Step 29576: {'lr': 0.00019202057965420993, 'samples': 15143424, 'steps': 29576, 'loss/train': 2.081496000289917}
+02/25/2022 15:54:03 - INFO - codeparrot_training - Step 29577: {'lr': 0.00019200466343238305, 'samples': 15143936, 'steps': 29577, 'loss/train': 2.046339750289917}
+02/25/2022 15:54:09 - INFO - codeparrot_training - Step 29578: {'lr': 0.00019198874745898986, 'samples': 15144448, 'steps': 29578, 'loss/train': 1.7536154985427856}
+02/25/2022 15:54:12 - INFO - codeparrot_training - Step 29579: {'lr': 0.00019197283173409837, 'samples': 15144960, 'steps': 29579, 'loss/train': 0.6974378228187561}
+02/25/2022 15:54:19 - INFO - codeparrot_training - Step 29580: {'lr': 0.00019195691625777686, 'samples': 15145472, 'steps': 29580, 'loss/train': 2.0256919860839844}
+02/25/2022 15:54:22 - INFO - codeparrot_training - Step 29581: {'lr': 0.0001919410010300935, 'samples': 15145984, 'steps': 29581, 'loss/train': 1.1672112941741943}
+02/25/2022 15:54:28 - INFO - codeparrot_training - Step 29582: {'lr': 0.00019192508605111647, 'samples': 15146496, 'steps': 29582, 'loss/train': 1.1593413352966309}
+02/25/2022 15:54:31 - INFO - codeparrot_training - Step 29583: {'lr': 0.00019190917132091393, 'samples': 15147008, 'steps': 29583, 'loss/train': 1.7980064153671265}
+02/25/2022 15:54:37 - INFO - codeparrot_training - Step 29584: {'lr': 0.000191893256839554, 'samples': 15147520, 'steps': 29584, 'loss/train': 0.8032127022743225}
+02/25/2022 15:54:40 - INFO - codeparrot_training - Step 29585: {'lr': 0.00019187734260710498, 'samples': 15148032, 'steps': 29585, 'loss/train': 1.225101351737976}
+02/25/2022 15:54:45 - INFO - codeparrot_training - Step 29586: {'lr': 0.00019186142862363491, 'samples': 15148544, 'steps': 29586, 'loss/train': 1.4679492712020874}
+02/25/2022 15:54:49 - INFO - codeparrot_training - Step 29587: {'lr': 0.00019184551488921208, 'samples': 15149056, 'steps': 29587, 'loss/train': 0.7630745768547058}
+02/25/2022 15:54:54 - INFO - codeparrot_training - Step 29588: {'lr': 0.00019182960140390454, 'samples': 15149568, 'steps': 29588, 'loss/train': 1.9443163871765137}
+02/25/2022 15:54:58 - INFO - codeparrot_training - Step 29589: {'lr': 0.00019181368816778056, 'samples': 15150080, 'steps': 29589, 'loss/train': 2.063915729522705}
+02/25/2022 15:55:04 - INFO - codeparrot_training - Step 29590: {'lr': 0.0001917977751809082, 'samples': 15150592, 'steps': 29590, 'loss/train': 2.680968999862671}
+02/25/2022 15:55:08 - INFO - codeparrot_training - Step 29591: {'lr': 0.00019178186244335577, 'samples': 15151104, 'steps': 29591, 'loss/train': 1.933132290840149}
+02/25/2022 15:55:13 - INFO - codeparrot_training - Step 29592: {'lr': 0.00019176594995519135, 'samples': 15151616, 'steps': 29592, 'loss/train': 2.1474061012268066}
+02/25/2022 15:55:17 - INFO - codeparrot_training - Step 29593: {'lr': 0.00019175003771648307, 'samples': 15152128, 'steps': 29593, 'loss/train': 1.4122283458709717}
+02/25/2022 15:55:22 - INFO - codeparrot_training - Step 29594: {'lr': 0.0001917341257272991, 'samples': 15152640, 'steps': 29594, 'loss/train': 2.515367031097412}
+02/25/2022 15:55:26 - INFO - codeparrot_training - Step 29595: {'lr': 0.00019171821398770764, 'samples': 15153152, 'steps': 29595, 'loss/train': 2.163191556930542}
+02/25/2022 15:55:31 - INFO - codeparrot_training - Step 29596: {'lr': 0.00019170230249777696, 'samples': 15153664, 'steps': 29596, 'loss/train': 0.5583831071853638}
+02/25/2022 15:55:35 - INFO - codeparrot_training - Step 29597: {'lr': 0.00019168639125757497, 'samples': 15154176, 'steps': 29597, 'loss/train': 1.909466028213501}
+02/25/2022 15:55:40 - INFO - codeparrot_training - Step 29598: {'lr': 0.00019167048026716999, 'samples': 15154688, 'steps': 29598, 'loss/train': 1.9099794626235962}
+02/25/2022 15:55:44 - INFO - codeparrot_training - Step 29599: {'lr': 0.00019165456952663015, 'samples': 15155200, 'steps': 29599, 'loss/train': 0.5206064581871033}
+02/25/2022 15:55:50 - INFO - codeparrot_training - Step 29600: {'lr': 0.00019163865903602372, 'samples': 15155712, 'steps': 29600, 'loss/train': 1.6779024600982666}
+02/25/2022 15:55:54 - INFO - codeparrot_training - Step 29601: {'lr': 0.00019162274879541864, 'samples': 15156224, 'steps': 29601, 'loss/train': 2.0416195392608643}
+02/25/2022 15:55:59 - INFO - codeparrot_training - Step 29602: {'lr': 0.00019160683880488314, 'samples': 15156736, 'steps': 29602, 'loss/train': 1.99224853515625}
+02/25/2022 15:56:03 - INFO - codeparrot_training - Step 29603: {'lr': 0.00019159092906448542, 'samples': 15157248, 'steps': 29603, 'loss/train': 1.9454398155212402}
+02/25/2022 15:56:08 - INFO - codeparrot_training - Step 29604: {'lr': 0.00019157501957429372, 'samples': 15157760, 'steps': 29604, 'loss/train': 0.9746183156967163}
+02/25/2022 15:56:12 - INFO - codeparrot_training - Step 29605: {'lr': 0.00019155911033437597, 'samples': 15158272, 'steps': 29605, 'loss/train': 2.7291858196258545}
+02/25/2022 15:56:17 - INFO - codeparrot_training - Step 29606: {'lr': 0.00019154320134480044, 'samples': 15158784, 'steps': 29606, 'loss/train': 1.805185079574585}
+02/25/2022 15:56:21 - INFO - codeparrot_training - Step 29607: {'lr': 0.0001915272926056353, 'samples': 15159296, 'steps': 29607, 'loss/train': 1.6780824661254883}
+02/25/2022 15:56:26 - INFO - codeparrot_training - Step 29608: {'lr': 0.00019151138411694862, 'samples': 15159808, 'steps': 29608, 'loss/train': 1.4848872423171997}
+02/25/2022 15:56:30 - INFO - codeparrot_training - Step 29609: {'lr': 0.0001914954758788087, 'samples': 15160320, 'steps': 29609, 'loss/train': 2.320767879486084}
+02/25/2022 15:56:36 - INFO - codeparrot_training - Step 29610: {'lr': 0.00019147956789128352, 'samples': 15160832, 'steps': 29610, 'loss/train': 2.444650173187256}
+02/25/2022 15:56:39 - INFO - codeparrot_training - Step 29611: {'lr': 0.00019146366015444126, 'samples': 15161344, 'steps': 29611, 'loss/train': 1.49068284034729}
+02/25/2022 15:56:45 - INFO - codeparrot_training - Step 29612: {'lr': 0.00019144775266835012, 'samples': 15161856, 'steps': 29612, 'loss/train': 1.059166431427002}
+02/25/2022 15:56:48 - INFO - codeparrot_training - Step 29613: {'lr': 0.00019143184543307833, 'samples': 15162368, 'steps': 29613, 'loss/train': 1.8150075674057007}
+02/25/2022 15:56:54 - INFO - codeparrot_training - Step 29614: {'lr': 0.00019141593844869376, 'samples': 15162880, 'steps': 29614, 'loss/train': 1.6683028936386108}
+02/25/2022 15:56:57 - INFO - codeparrot_training - Step 29615: {'lr': 0.00019140003171526474, 'samples': 15163392, 'steps': 29615, 'loss/train': 1.4695719480514526}
+02/25/2022 15:57:03 - INFO - codeparrot_training - Step 29616: {'lr': 0.00019138412523285936, 'samples': 15163904, 'steps': 29616, 'loss/train': 1.930280327796936}
+02/25/2022 15:57:06 - INFO - codeparrot_training - Step 29617: {'lr': 0.0001913682190015458, 'samples': 15164416, 'steps': 29617, 'loss/train': 2.031759738922119}
+02/25/2022 15:57:12 - INFO - codeparrot_training - Step 29618: {'lr': 0.00019135231302139227, 'samples': 15164928, 'steps': 29618, 'loss/train': 0.242783322930336}
+02/25/2022 15:57:15 - INFO - codeparrot_training - Step 29619: {'lr': 0.00019133640729246672, 'samples': 15165440, 'steps': 29619, 'loss/train': 1.2959227561950684}
+02/25/2022 15:57:21 - INFO - codeparrot_training - Step 29620: {'lr': 0.00019132050181483736, 'samples': 15165952, 'steps': 29620, 'loss/train': 1.4619669914245605}
+02/25/2022 15:57:26 - INFO - codeparrot_training - Step 29621: {'lr': 0.0001913045965885723, 'samples': 15166464, 'steps': 29621, 'loss/train': 2.0575144290924072}
+02/25/2022 15:57:30 - INFO - codeparrot_training - Step 29622: {'lr': 0.00019128869161373985, 'samples': 15166976, 'steps': 29622, 'loss/train': 1.9949980974197388}
+02/25/2022 15:57:35 - INFO - codeparrot_training - Step 29623: {'lr': 0.0001912727868904079, 'samples': 15167488, 'steps': 29623, 'loss/train': 1.4256365299224854}
+02/25/2022 15:57:39 - INFO - codeparrot_training - Step 29624: {'lr': 0.00019125688241864464, 'samples': 15168000, 'steps': 29624, 'loss/train': 1.6261653900146484}
+02/25/2022 15:57:45 - INFO - codeparrot_training - Step 29625: {'lr': 0.00019124097819851835, 'samples': 15168512, 'steps': 29625, 'loss/train': 2.0156874656677246}
+02/25/2022 15:57:48 - INFO - codeparrot_training - Step 29626: {'lr': 0.00019122507423009703, 'samples': 15169024, 'steps': 29626, 'loss/train': 1.8535398244857788}
+02/25/2022 15:57:54 - INFO - codeparrot_training - Step 29627: {'lr': 0.00019120917051344878, 'samples': 15169536, 'steps': 29627, 'loss/train': 1.51189124584198}
+02/25/2022 15:57:57 - INFO - codeparrot_training - Step 29628: {'lr': 0.00019119326704864176, 'samples': 15170048, 'steps': 29628, 'loss/train': 1.3984068632125854}
+02/25/2022 15:58:03 - INFO - codeparrot_training - Step 29629: {'lr': 0.0001911773638357442, 'samples': 15170560, 'steps': 29629, 'loss/train': 1.2846662998199463}
+02/25/2022 15:58:06 - INFO - codeparrot_training - Step 29630: {'lr': 0.00019116146087482403, 'samples': 15171072, 'steps': 29630, 'loss/train': 1.957324743270874}
+02/25/2022 15:58:12 - INFO - codeparrot_training - Step 29631: {'lr': 0.00019114555816594956, 'samples': 15171584, 'steps': 29631, 'loss/train': 2.1340928077697754}
+02/25/2022 15:58:15 - INFO - codeparrot_training - Step 29632: {'lr': 0.00019112965570918883, 'samples': 15172096, 'steps': 29632, 'loss/train': 2.267246723175049}
+02/25/2022 15:58:22 - INFO - codeparrot_training - Step 29633: {'lr': 0.00019111375350460991, 'samples': 15172608, 'steps': 29633, 'loss/train': 2.070429801940918}
+02/25/2022 15:58:25 - INFO - codeparrot_training - Step 29634: {'lr': 0.00019109785155228096, 'samples': 15173120, 'steps': 29634, 'loss/train': 0.7860600352287292}
+02/25/2022 15:58:31 - INFO - codeparrot_training - Step 29635: {'lr': 0.00019108194985227017, 'samples': 15173632, 'steps': 29635, 'loss/train': 1.2916388511657715}
+02/25/2022 15:58:34 - INFO - codeparrot_training - Step 29636: {'lr': 0.00019106604840464562, 'samples': 15174144, 'steps': 29636, 'loss/train': 1.569584608078003}
+02/25/2022 15:58:40 - INFO - codeparrot_training - Step 29637: {'lr': 0.00019105014720947533, 'samples': 15174656, 'steps': 29637, 'loss/train': 1.7774972915649414}
+02/25/2022 15:58:43 - INFO - codeparrot_training - Step 29638: {'lr': 0.00019103424626682746, 'samples': 15175168, 'steps': 29638, 'loss/train': 1.153586745262146}
+02/25/2022 15:58:49 - INFO - codeparrot_training - Step 29639: {'lr': 0.00019101834557677034, 'samples': 15175680, 'steps': 29639, 'loss/train': 2.7185306549072266}
+02/25/2022 15:58:52 - INFO - codeparrot_training - Step 29640: {'lr': 0.00019100244513937175, 'samples': 15176192, 'steps': 29640, 'loss/train': 2.2741262912750244}
+02/25/2022 15:58:58 - INFO - codeparrot_training - Step 29641: {'lr': 0.00019098654495469991, 'samples': 15176704, 'steps': 29641, 'loss/train': 1.2731679677963257}
+02/25/2022 15:59:01 - INFO - codeparrot_training - Step 29642: {'lr': 0.00019097064502282302, 'samples': 15177216, 'steps': 29642, 'loss/train': 1.6074756383895874}
+02/25/2022 15:59:07 - INFO - codeparrot_training - Step 29643: {'lr': 0.00019095474534380912, 'samples': 15177728, 'steps': 29643, 'loss/train': 1.6521459817886353}
+02/25/2022 15:59:10 - INFO - codeparrot_training - Step 29644: {'lr': 0.00019093884591772647, 'samples': 15178240, 'steps': 29644, 'loss/train': 1.7591832876205444}
+02/25/2022 15:59:17 - INFO - codeparrot_training - Step 29645: {'lr': 0.00019092294674464294, 'samples': 15178752, 'steps': 29645, 'loss/train': 1.0980162620544434}
+02/25/2022 15:59:20 - INFO - codeparrot_training - Step 29646: {'lr': 0.00019090704782462675, 'samples': 15179264, 'steps': 29646, 'loss/train': 2.2975497245788574}
+02/25/2022 15:59:24 - INFO - codeparrot_training - Step 29647: {'lr': 0.000190891149157746, 'samples': 15179776, 'steps': 29647, 'loss/train': 1.4089785814285278}
+02/25/2022 15:59:29 - INFO - codeparrot_training - Step 29648: {'lr': 0.0001908752507440689, 'samples': 15180288, 'steps': 29648, 'loss/train': 1.8052784204483032}
+02/25/2022 15:59:32 - INFO - codeparrot_training - Step 29649: {'lr': 0.00019085935258366334, 'samples': 15180800, 'steps': 29649, 'loss/train': 1.8595654964447021}
+02/25/2022 15:59:38 - INFO - codeparrot_training - Step 29650: {'lr': 0.00019084345467659754, 'samples': 15181312, 'steps': 29650, 'loss/train': 1.6302999258041382}
+02/25/2022 15:59:44 - INFO - codeparrot_training - Step 29651: {'lr': 0.00019082755702293957, 'samples': 15181824, 'steps': 29651, 'loss/train': 1.4600595235824585}
+02/25/2022 15:59:47 - INFO - codeparrot_training - Step 29652: {'lr': 0.00019081165962275771, 'samples': 15182336, 'steps': 29652, 'loss/train': 1.6980838775634766}
+02/25/2022 15:59:52 - INFO - codeparrot_training - Step 29653: {'lr': 0.0001907957624761198, 'samples': 15182848, 'steps': 29653, 'loss/train': 1.9229475259780884}
+02/25/2022 15:59:56 - INFO - codeparrot_training - Step 29654: {'lr': 0.00019077986558309402, 'samples': 15183360, 'steps': 29654, 'loss/train': 1.6512038707733154}
+02/25/2022 16:00:02 - INFO - codeparrot_training - Step 29655: {'lr': 0.00019076396894374847, 'samples': 15183872, 'steps': 29655, 'loss/train': 1.4573386907577515}
+02/25/2022 16:00:05 - INFO - codeparrot_training - Step 29656: {'lr': 0.0001907480725581513, 'samples': 15184384, 'steps': 29656, 'loss/train': 1.6344777345657349}
+02/25/2022 16:00:11 - INFO - codeparrot_training - Step 29657: {'lr': 0.00019073217642637068, 'samples': 15184896, 'steps': 29657, 'loss/train': 1.3940215110778809}
+02/25/2022 16:00:14 - INFO - codeparrot_training - Step 29658: {'lr': 0.00019071628054847446, 'samples': 15185408, 'steps': 29658, 'loss/train': 1.614280343055725}
+02/25/2022 16:00:20 - INFO - codeparrot_training - Step 29659: {'lr': 0.0001907003849245309, 'samples': 15185920, 'steps': 29659, 'loss/train': 1.55470871925354}
+02/25/2022 16:00:23 - INFO - codeparrot_training - Step 29660: {'lr': 0.00019068448955460805, 'samples': 15186432, 'steps': 29660, 'loss/train': 1.646599531173706}
+02/25/2022 16:00:29 - INFO - codeparrot_training - Step 29661: {'lr': 0.0001906685944387741, 'samples': 15186944, 'steps': 29661, 'loss/train': 2.3054816722869873}
+02/25/2022 16:00:32 - INFO - codeparrot_training - Step 29662: {'lr': 0.00019065269957709697, 'samples': 15187456, 'steps': 29662, 'loss/train': 0.7035332918167114}
+02/25/2022 16:00:38 - INFO - codeparrot_training - Step 29663: {'lr': 0.00019063680496964484, 'samples': 15187968, 'steps': 29663, 'loss/train': 2.4976718425750732}
+02/25/2022 16:00:41 - INFO - codeparrot_training - Step 29664: {'lr': 0.00019062091061648577, 'samples': 15188480, 'steps': 29664, 'loss/train': 1.5096818208694458}
+02/25/2022 16:00:47 - INFO - codeparrot_training - Step 29665: {'lr': 0.00019060501651768796, 'samples': 15188992, 'steps': 29665, 'loss/train': 2.1369731426239014}
+02/25/2022 16:00:50 - INFO - codeparrot_training - Step 29666: {'lr': 0.0001905891226733193, 'samples': 15189504, 'steps': 29666, 'loss/train': 0.33467504382133484}
+02/25/2022 16:00:56 - INFO - codeparrot_training - Step 29667: {'lr': 0.000190573229083448, 'samples': 15190016, 'steps': 29667, 'loss/train': 1.9126348495483398}
+02/25/2022 16:01:00 - INFO - codeparrot_training - Step 29668: {'lr': 0.00019055733574814207, 'samples': 15190528, 'steps': 29668, 'loss/train': 0.6411345601081848}
+02/25/2022 16:01:05 - INFO - codeparrot_training - Step 29669: {'lr': 0.00019054144266746973, 'samples': 15191040, 'steps': 29669, 'loss/train': 2.2083663940429688}
+02/25/2022 16:01:09 - INFO - codeparrot_training - Step 29670: {'lr': 0.00019052554984149895, 'samples': 15191552, 'steps': 29670, 'loss/train': 2.313786268234253}
+02/25/2022 16:01:12 - INFO - codeparrot_training - Step 29671: {'lr': 0.0001905096572702978, 'samples': 15192064, 'steps': 29671, 'loss/train': 1.0212492942810059}
+02/25/2022 16:01:18 - INFO - codeparrot_training - Step 29672: {'lr': 0.0001904937649539344, 'samples': 15192576, 'steps': 29672, 'loss/train': 2.3255090713500977}
+02/25/2022 16:01:22 - INFO - codeparrot_training - Step 29673: {'lr': 0.0001904778728924768, 'samples': 15193088, 'steps': 29673, 'loss/train': 1.4593840837478638}
+02/25/2022 16:01:27 - INFO - codeparrot_training - Step 29674: {'lr': 0.0001904619810859931, 'samples': 15193600, 'steps': 29674, 'loss/train': 2.344268321990967}
+02/25/2022 16:01:31 - INFO - codeparrot_training - Step 29675: {'lr': 0.00019044608953455136, 'samples': 15194112, 'steps': 29675, 'loss/train': 1.2264584302902222}
+02/25/2022 16:01:36 - INFO - codeparrot_training - Step 29676: {'lr': 0.0001904301982382197, 'samples': 15194624, 'steps': 29676, 'loss/train': 1.537766695022583}
+02/25/2022 16:01:40 - INFO - codeparrot_training - Step 29677: {'lr': 0.0001904143071970661, 'samples': 15195136, 'steps': 29677, 'loss/train': 2.050400972366333}
+02/25/2022 16:01:45 - INFO - codeparrot_training - Step 29678: {'lr': 0.0001903984164111587, 'samples': 15195648, 'steps': 29678, 'loss/train': 0.5048926472663879}
+02/25/2022 16:01:49 - INFO - codeparrot_training - Step 29679: {'lr': 0.00019038252588056558, 'samples': 15196160, 'steps': 29679, 'loss/train': 2.1396775245666504}
+02/25/2022 16:01:54 - INFO - codeparrot_training - Step 29680: {'lr': 0.00019036663560535483, 'samples': 15196672, 'steps': 29680, 'loss/train': 1.6611448526382446}
+02/25/2022 16:02:01 - INFO - codeparrot_training - Step 29681: {'lr': 0.0001903507455855944, 'samples': 15197184, 'steps': 29681, 'loss/train': 1.4303745031356812}
+02/25/2022 16:02:04 - INFO - codeparrot_training - Step 29682: {'lr': 0.00019033485582135244, 'samples': 15197696, 'steps': 29682, 'loss/train': 1.0385608673095703}
+02/25/2022 16:02:09 - INFO - codeparrot_training - Step 29683: {'lr': 0.00019031896631269713, 'samples': 15198208, 'steps': 29683, 'loss/train': 1.137509822845459}
+02/25/2022 16:02:13 - INFO - codeparrot_training - Step 29684: {'lr': 0.00019030307705969628, 'samples': 15198720, 'steps': 29684, 'loss/train': 0.40029680728912354}
+02/25/2022 16:02:18 - INFO - codeparrot_training - Step 29685: {'lr': 0.00019028718806241813, 'samples': 15199232, 'steps': 29685, 'loss/train': 1.3261994123458862}
+02/25/2022 16:02:22 - INFO - codeparrot_training - Step 29686: {'lr': 0.00019027129932093067, 'samples': 15199744, 'steps': 29686, 'loss/train': 1.7638357877731323}
+02/25/2022 16:02:28 - INFO - codeparrot_training - Step 29687: {'lr': 0.00019025541083530213, 'samples': 15200256, 'steps': 29687, 'loss/train': 2.1726949214935303}
+02/25/2022 16:02:31 - INFO - codeparrot_training - Step 29688: {'lr': 0.00019023952260560032, 'samples': 15200768, 'steps': 29688, 'loss/train': 1.969434142112732}
+02/25/2022 16:02:37 - INFO - codeparrot_training - Step 29689: {'lr': 0.0001902236346318934, 'samples': 15201280, 'steps': 29689, 'loss/train': 1.3506629467010498}
+02/25/2022 16:02:40 - INFO - codeparrot_training - Step 29690: {'lr': 0.00019020774691424946, 'samples': 15201792, 'steps': 29690, 'loss/train': 1.8096134662628174}
+02/25/2022 16:02:46 - INFO - codeparrot_training - Step 29691: {'lr': 0.00019019185945273655, 'samples': 15202304, 'steps': 29691, 'loss/train': 1.6162241697311401}
+02/25/2022 16:02:50 - INFO - codeparrot_training - Step 29692: {'lr': 0.00019017597224742287, 'samples': 15202816, 'steps': 29692, 'loss/train': 1.4075875282287598}
+02/25/2022 16:02:55 - INFO - codeparrot_training - Step 29693: {'lr': 0.00019016008529837617, 'samples': 15203328, 'steps': 29693, 'loss/train': 1.740389347076416}
+02/25/2022 16:02:59 - INFO - codeparrot_training - Step 29694: {'lr': 0.00019014419860566468, 'samples': 15203840, 'steps': 29694, 'loss/train': 1.511661410331726}
+02/25/2022 16:03:04 - INFO - codeparrot_training - Step 29695: {'lr': 0.00019012831216935644, 'samples': 15204352, 'steps': 29695, 'loss/train': 0.8117575645446777}
+02/25/2022 16:03:08 - INFO - codeparrot_training - Step 29696: {'lr': 0.0001901124259895196, 'samples': 15204864, 'steps': 29696, 'loss/train': 2.414487600326538}
+02/25/2022 16:03:13 - INFO - codeparrot_training - Step 29697: {'lr': 0.000190096540066222, 'samples': 15205376, 'steps': 29697, 'loss/train': 1.35346257686615}
+02/25/2022 16:03:17 - INFO - codeparrot_training - Step 29698: {'lr': 0.00019008065439953182, 'samples': 15205888, 'steps': 29698, 'loss/train': 8.438970565795898}
+02/25/2022 16:03:22 - INFO - codeparrot_training - Step 29699: {'lr': 0.00019006476898951707, 'samples': 15206400, 'steps': 29699, 'loss/train': 2.0487515926361084}
+02/25/2022 16:03:26 - INFO - codeparrot_training - Step 29700: {'lr': 0.00019004888383624595, 'samples': 15206912, 'steps': 29700, 'loss/train': 1.7922396659851074}
+02/25/2022 16:03:32 - INFO - codeparrot_training - Step 29701: {'lr': 0.00019003299893978622, 'samples': 15207424, 'steps': 29701, 'loss/train': 1.2144306898117065}
+02/25/2022 16:03:35 - INFO - codeparrot_training - Step 29702: {'lr': 0.00019001711430020613, 'samples': 15207936, 'steps': 29702, 'loss/train': 1.787750244140625}
+02/25/2022 16:03:41 - INFO - codeparrot_training - Step 29703: {'lr': 0.00019000122991757367, 'samples': 15208448, 'steps': 29703, 'loss/train': 1.7397289276123047}
+02/25/2022 16:03:45 - INFO - codeparrot_training - Step 29704: {'lr': 0.00018998534579195685, 'samples': 15208960, 'steps': 29704, 'loss/train': 2.310115098953247}
+02/25/2022 16:03:48 - INFO - codeparrot_training - Step 29705: {'lr': 0.0001899694619234239, 'samples': 15209472, 'steps': 29705, 'loss/train': 2.9601659774780273}
+02/25/2022 16:03:54 - INFO - codeparrot_training - Step 29706: {'lr': 0.00018995357831204257, 'samples': 15209984, 'steps': 29706, 'loss/train': 1.3097292184829712}
+02/25/2022 16:03:57 - INFO - codeparrot_training - Step 29707: {'lr': 0.00018993769495788109, 'samples': 15210496, 'steps': 29707, 'loss/train': 1.6220203638076782}
+02/25/2022 16:04:02 - INFO - codeparrot_training - Step 29708: {'lr': 0.00018992181186100744, 'samples': 15211008, 'steps': 29708, 'loss/train': 1.0518784523010254}
+02/25/2022 16:04:06 - INFO - codeparrot_training - Step 29709: {'lr': 0.00018990592902148973, 'samples': 15211520, 'steps': 29709, 'loss/train': 1.4611061811447144}
+02/25/2022 16:04:11 - INFO - codeparrot_training - Step 29710: {'lr': 0.00018989004643939588, 'samples': 15212032, 'steps': 29710, 'loss/train': 1.481844425201416}
+02/25/2022 16:04:15 - INFO - codeparrot_training - Step 29711: {'lr': 0.00018987416411479397, 'samples': 15212544, 'steps': 29711, 'loss/train': 2.341019630432129}
+02/25/2022 16:04:20 - INFO - codeparrot_training - Step 29712: {'lr': 0.00018985828204775204, 'samples': 15213056, 'steps': 29712, 'loss/train': 2.197713613510132}
+02/25/2022 16:04:24 - INFO - codeparrot_training - Step 29713: {'lr': 0.00018984240023833826, 'samples': 15213568, 'steps': 29713, 'loss/train': 1.190040946006775}
+02/25/2022 16:04:30 - INFO - codeparrot_training - Step 29714: {'lr': 0.00018982651868662044, 'samples': 15214080, 'steps': 29714, 'loss/train': 1.5683180093765259}
+02/25/2022 16:04:33 - INFO - codeparrot_training - Step 29715: {'lr': 0.00018981063739266667, 'samples': 15214592, 'steps': 29715, 'loss/train': 1.4655780792236328}
+02/25/2022 16:04:39 - INFO - codeparrot_training - Step 29716: {'lr': 0.0001897947563565451, 'samples': 15215104, 'steps': 29716, 'loss/train': 2.7867512702941895}
+02/25/2022 16:04:42 - INFO - codeparrot_training - Step 29717: {'lr': 0.00018977887557832356, 'samples': 15215616, 'steps': 29717, 'loss/train': 0.9705775380134583}
+02/25/2022 16:04:48 - INFO - codeparrot_training - Step 29718: {'lr': 0.00018976299505807031, 'samples': 15216128, 'steps': 29718, 'loss/train': 2.2573275566101074}
+02/25/2022 16:04:54 - INFO - codeparrot_training - Step 29719: {'lr': 0.00018974711479585323, 'samples': 15216640, 'steps': 29719, 'loss/train': 1.5685479640960693}
+02/25/2022 16:04:57 - INFO - codeparrot_training - Step 29720: {'lr': 0.00018973123479174036, 'samples': 15217152, 'steps': 29720, 'loss/train': 1.150031328201294}
+02/25/2022 16:05:03 - INFO - codeparrot_training - Step 29721: {'lr': 0.00018971535504579972, 'samples': 15217664, 'steps': 29721, 'loss/train': 2.5471959114074707}
+02/25/2022 16:05:06 - INFO - codeparrot_training - Step 29722: {'lr': 0.0001896994755580994, 'samples': 15218176, 'steps': 29722, 'loss/train': 2.855038642883301}
+02/25/2022 16:05:12 - INFO - codeparrot_training - Step 29723: {'lr': 0.0001896835963287074, 'samples': 15218688, 'steps': 29723, 'loss/train': 1.5021971464157104}
+02/25/2022 16:05:15 - INFO - codeparrot_training - Step 29724: {'lr': 0.00018966771735769167, 'samples': 15219200, 'steps': 29724, 'loss/train': 2.3879992961883545}
+02/25/2022 16:05:21 - INFO - codeparrot_training - Step 29725: {'lr': 0.00018965183864512027, 'samples': 15219712, 'steps': 29725, 'loss/train': 2.1341824531555176}
+02/25/2022 16:05:24 - INFO - codeparrot_training - Step 29726: {'lr': 0.00018963596019106125, 'samples': 15220224, 'steps': 29726, 'loss/train': 1.0678937435150146}
+02/25/2022 16:05:31 - INFO - codeparrot_training - Step 29727: {'lr': 0.0001896200819955827, 'samples': 15220736, 'steps': 29727, 'loss/train': 2.567450761795044}
+02/25/2022 16:05:34 - INFO - codeparrot_training - Step 29728: {'lr': 0.00018960420405875244, 'samples': 15221248, 'steps': 29728, 'loss/train': 1.4132837057113647}
+02/25/2022 16:05:40 - INFO - codeparrot_training - Step 29729: {'lr': 0.00018958832638063858, 'samples': 15221760, 'steps': 29729, 'loss/train': 0.6862665414810181}
+02/25/2022 16:05:43 - INFO - codeparrot_training - Step 29730: {'lr': 0.00018957244896130915, 'samples': 15222272, 'steps': 29730, 'loss/train': 1.5507272481918335}
+02/25/2022 16:05:49 - INFO - codeparrot_training - Step 29731: {'lr': 0.00018955657180083228, 'samples': 15222784, 'steps': 29731, 'loss/train': 1.1663380861282349}
+02/25/2022 16:05:52 - INFO - codeparrot_training - Step 29732: {'lr': 0.00018954069489927574, 'samples': 15223296, 'steps': 29732, 'loss/train': 0.8850343823432922}
+02/25/2022 16:05:58 - INFO - codeparrot_training - Step 29733: {'lr': 0.0001895248182567077, 'samples': 15223808, 'steps': 29733, 'loss/train': 1.8836568593978882}
+02/25/2022 16:06:01 - INFO - codeparrot_training - Step 29734: {'lr': 0.0001895089418731961, 'samples': 15224320, 'steps': 29734, 'loss/train': 0.9981403946876526}
+02/25/2022 16:06:07 - INFO - codeparrot_training - Step 29735: {'lr': 0.0001894930657488091, 'samples': 15224832, 'steps': 29735, 'loss/train': 2.9812047481536865}
+02/25/2022 16:06:10 - INFO - codeparrot_training - Step 29736: {'lr': 0.0001894771898836145, 'samples': 15225344, 'steps': 29736, 'loss/train': 0.3940974175930023}
+02/25/2022 16:06:17 - INFO - codeparrot_training - Step 29737: {'lr': 0.00018946131427768039, 'samples': 15225856, 'steps': 29737, 'loss/train': 1.7756891250610352}
+02/25/2022 16:06:21 - INFO - codeparrot_training - Step 29738: {'lr': 0.00018944543893107478, 'samples': 15226368, 'steps': 29738, 'loss/train': 1.3673560619354248}
+02/25/2022 16:06:26 - INFO - codeparrot_training - Step 29739: {'lr': 0.00018942956384386566, 'samples': 15226880, 'steps': 29739, 'loss/train': 2.3634588718414307}
+02/25/2022 16:06:30 - INFO - codeparrot_training - Step 29740: {'lr': 0.00018941368901612121, 'samples': 15227392, 'steps': 29740, 'loss/train': 1.5838547945022583}
+02/25/2022 16:06:35 - INFO - codeparrot_training - Step 29741: {'lr': 0.00018939781444790916, 'samples': 15227904, 'steps': 29741, 'loss/train': 2.242109775543213}
+02/25/2022 16:06:39 - INFO - codeparrot_training - Step 29742: {'lr': 0.0001893819401392976, 'samples': 15228416, 'steps': 29742, 'loss/train': 0.8927356004714966}
+02/25/2022 16:06:44 - INFO - codeparrot_training - Step 29743: {'lr': 0.0001893660660903546, 'samples': 15228928, 'steps': 29743, 'loss/train': 1.2056217193603516}
+02/25/2022 16:06:48 - INFO - codeparrot_training - Step 29744: {'lr': 0.0001893501923011482, 'samples': 15229440, 'steps': 29744, 'loss/train': 1.2550374269485474}
+02/25/2022 16:06:53 - INFO - codeparrot_training - Step 29745: {'lr': 0.00018933431877174624, 'samples': 15229952, 'steps': 29745, 'loss/train': 1.7424426078796387}
+02/25/2022 16:06:57 - INFO - codeparrot_training - Step 29746: {'lr': 0.00018931844550221675, 'samples': 15230464, 'steps': 29746, 'loss/train': 0.5975298285484314}
+02/25/2022 16:07:03 - INFO - codeparrot_training - Step 29747: {'lr': 0.00018930257249262776, 'samples': 15230976, 'steps': 29747, 'loss/train': 0.07923591136932373}
+02/25/2022 16:07:06 - INFO - codeparrot_training - Step 29748: {'lr': 0.00018928669974304746, 'samples': 15231488, 'steps': 29748, 'loss/train': 1.3820666074752808}
+02/25/2022 16:07:12 - INFO - codeparrot_training - Step 29749: {'lr': 0.0001892708272535435, 'samples': 15232000, 'steps': 29749, 'loss/train': 1.4223815202713013}
+02/25/2022 16:07:15 - INFO - codeparrot_training - Step 29750: {'lr': 0.00018925495502418406, 'samples': 15232512, 'steps': 29750, 'loss/train': 1.9621717929840088}
+02/25/2022 16:07:21 - INFO - codeparrot_training - Step 29751: {'lr': 0.00018923908305503706, 'samples': 15233024, 'steps': 29751, 'loss/train': 2.3909809589385986}
+02/25/2022 16:07:24 - INFO - codeparrot_training - Step 29752: {'lr': 0.00018922321134617055, 'samples': 15233536, 'steps': 29752, 'loss/train': 1.416527271270752}
+02/25/2022 16:07:30 - INFO - codeparrot_training - Step 29753: {'lr': 0.00018920733989765263, 'samples': 15234048, 'steps': 29753, 'loss/train': 1.3958063125610352}
+02/25/2022 16:07:33 - INFO - codeparrot_training - Step 29754: {'lr': 0.00018919146870955103, 'samples': 15234560, 'steps': 29754, 'loss/train': 1.333345651626587}
+02/25/2022 16:07:39 - INFO - codeparrot_training - Step 29755: {'lr': 0.00018917559778193387, 'samples': 15235072, 'steps': 29755, 'loss/train': 1.0761003494262695}
+02/25/2022 16:07:42 - INFO - codeparrot_training - Step 29756: {'lr': 0.00018915972711486923, 'samples': 15235584, 'steps': 29756, 'loss/train': 1.9296795129776}
+02/25/2022 16:07:48 - INFO - codeparrot_training - Step 29757: {'lr': 0.000189143856708425, 'samples': 15236096, 'steps': 29757, 'loss/train': 1.7058906555175781}
+02/25/2022 16:07:51 - INFO - codeparrot_training - Step 29758: {'lr': 0.00018912798656266905, 'samples': 15236608, 'steps': 29758, 'loss/train': 1.8765857219696045}
+02/25/2022 16:07:57 - INFO - codeparrot_training - Step 29759: {'lr': 0.00018911211667766952, 'samples': 15237120, 'steps': 29759, 'loss/train': 1.8540867567062378}
+02/25/2022 16:08:02 - INFO - codeparrot_training - Step 29760: {'lr': 0.00018909624705349437, 'samples': 15237632, 'steps': 29760, 'loss/train': 1.778419017791748}
+02/25/2022 16:08:06 - INFO - codeparrot_training - Step 29761: {'lr': 0.0001890803776902116, 'samples': 15238144, 'steps': 29761, 'loss/train': 1.5091160535812378}
+02/25/2022 16:08:11 - INFO - codeparrot_training - Step 29762: {'lr': 0.00018906450858788908, 'samples': 15238656, 'steps': 29762, 'loss/train': 1.6516075134277344}
+02/25/2022 16:08:15 - INFO - codeparrot_training - Step 29763: {'lr': 0.0001890486397465949, 'samples': 15239168, 'steps': 29763, 'loss/train': 1.1941330432891846}
+02/25/2022 16:08:21 - INFO - codeparrot_training - Step 29764: {'lr': 0.00018903277116639694, 'samples': 15239680, 'steps': 29764, 'loss/train': 1.5156298875808716}
+02/25/2022 16:08:24 - INFO - codeparrot_training - Step 29765: {'lr': 0.00018901690284736322, 'samples': 15240192, 'steps': 29765, 'loss/train': 1.579389214515686}
+02/25/2022 16:08:30 - INFO - codeparrot_training - Step 29766: {'lr': 0.0001890010347895618, 'samples': 15240704, 'steps': 29766, 'loss/train': 2.8881235122680664}
+02/25/2022 16:08:33 - INFO - codeparrot_training - Step 29767: {'lr': 0.00018898516699306057, 'samples': 15241216, 'steps': 29767, 'loss/train': 1.1252330541610718}
+02/25/2022 16:08:39 - INFO - codeparrot_training - Step 29768: {'lr': 0.00018896929945792746, 'samples': 15241728, 'steps': 29768, 'loss/train': 0.9265627861022949}
+02/25/2022 16:08:43 - INFO - codeparrot_training - Step 29769: {'lr': 0.00018895343218423047, 'samples': 15242240, 'steps': 29769, 'loss/train': 0.06501796841621399}
+02/25/2022 16:08:48 - INFO - codeparrot_training - Step 29770: {'lr': 0.00018893756517203772, 'samples': 15242752, 'steps': 29770, 'loss/train': 1.299767017364502}
+02/25/2022 16:08:51 - INFO - codeparrot_training - Step 29771: {'lr': 0.00018892169842141694, 'samples': 15243264, 'steps': 29771, 'loss/train': 3.187598943710327}
+02/25/2022 16:08:57 - INFO - codeparrot_training - Step 29772: {'lr': 0.00018890583193243622, 'samples': 15243776, 'steps': 29772, 'loss/train': 0.026948869228363037}
+02/25/2022 16:09:00 - INFO - codeparrot_training - Step 29773: {'lr': 0.0001888899657051635, 'samples': 15244288, 'steps': 29773, 'loss/train': 2.140406370162964}
+02/25/2022 16:09:07 - INFO - codeparrot_training - Step 29774: {'lr': 0.00018887409973966686, 'samples': 15244800, 'steps': 29774, 'loss/train': 1.4617607593536377}
+02/25/2022 16:09:10 - INFO - codeparrot_training - Step 29775: {'lr': 0.0001888582340360141, 'samples': 15245312, 'steps': 29775, 'loss/train': 1.877639889717102}
+02/25/2022 16:09:16 - INFO - codeparrot_training - Step 29776: {'lr': 0.0001888423685942732, 'samples': 15245824, 'steps': 29776, 'loss/train': 1.2678022384643555}
+02/25/2022 16:09:19 - INFO - codeparrot_training - Step 29777: {'lr': 0.0001888265034145122, 'samples': 15246336, 'steps': 29777, 'loss/train': 1.7724556922912598}
+02/25/2022 16:09:25 - INFO - codeparrot_training - Step 29778: {'lr': 0.00018881063849679903, 'samples': 15246848, 'steps': 29778, 'loss/train': 2.483582019805908}
+02/25/2022 16:09:28 - INFO - codeparrot_training - Step 29779: {'lr': 0.00018879477384120178, 'samples': 15247360, 'steps': 29779, 'loss/train': 1.9547593593597412}
+02/25/2022 16:09:34 - INFO - codeparrot_training - Step 29780: {'lr': 0.00018877890944778814, 'samples': 15247872, 'steps': 29780, 'loss/train': 2.127516269683838}
+02/25/2022 16:09:37 - INFO - codeparrot_training - Step 29781: {'lr': 0.00018876304531662626, 'samples': 15248384, 'steps': 29781, 'loss/train': 1.4735227823257446}
+02/25/2022 16:09:43 - INFO - codeparrot_training - Step 29782: {'lr': 0.00018874718144778403, 'samples': 15248896, 'steps': 29782, 'loss/train': 8.687186241149902}
+02/25/2022 16:09:46 - INFO - codeparrot_training - Step 29783: {'lr': 0.00018873131784132952, 'samples': 15249408, 'steps': 29783, 'loss/train': 2.3048532009124756}
+02/25/2022 16:09:52 - INFO - codeparrot_training - Step 29784: {'lr': 0.00018871545449733048, 'samples': 15249920, 'steps': 29784, 'loss/train': 1.9706388711929321}
+02/25/2022 16:09:56 - INFO - codeparrot_training - Step 29785: {'lr': 0.00018869959141585497, 'samples': 15250432, 'steps': 29785, 'loss/train': 2.3992395401000977}
+02/25/2022 16:10:01 - INFO - codeparrot_training - Step 29786: {'lr': 0.00018868372859697096, 'samples': 15250944, 'steps': 29786, 'loss/train': 2.2633328437805176}
+02/25/2022 16:10:05 - INFO - codeparrot_training - Step 29787: {'lr': 0.00018866786604074636, 'samples': 15251456, 'steps': 29787, 'loss/train': 2.0165457725524902}
+02/25/2022 16:10:10 - INFO - codeparrot_training - Step 29788: {'lr': 0.0001886520037472493, 'samples': 15251968, 'steps': 29788, 'loss/train': 1.6836521625518799}
+02/25/2022 16:10:14 - INFO - codeparrot_training - Step 29789: {'lr': 0.00018863614171654743, 'samples': 15252480, 'steps': 29789, 'loss/train': 3.434588670730591}
+02/25/2022 16:10:19 - INFO - codeparrot_training - Step 29790: {'lr': 0.00018862027994870887, 'samples': 15252992, 'steps': 29790, 'loss/train': 1.123921275138855}
+02/25/2022 16:10:23 - INFO - codeparrot_training - Step 29791: {'lr': 0.0001886044184438015, 'samples': 15253504, 'steps': 29791, 'loss/train': 1.8841543197631836}
+02/25/2022 16:10:28 - INFO - codeparrot_training - Step 29792: {'lr': 0.00018858855720189346, 'samples': 15254016, 'steps': 29792, 'loss/train': 2.458669900894165}
+02/25/2022 16:10:32 - INFO - codeparrot_training - Step 29793: {'lr': 0.00018857269622305243, 'samples': 15254528, 'steps': 29793, 'loss/train': 0.5346103310585022}
+02/25/2022 16:10:38 - INFO - codeparrot_training - Step 29794: {'lr': 0.00018855683550734643, 'samples': 15255040, 'steps': 29794, 'loss/train': 1.3085296154022217}
+02/25/2022 16:10:41 - INFO - codeparrot_training - Step 29795: {'lr': 0.00018854097505484346, 'samples': 15255552, 'steps': 29795, 'loss/train': 1.4604032039642334}
+02/25/2022 16:10:47 - INFO - codeparrot_training - Step 29796: {'lr': 0.00018852511486561156, 'samples': 15256064, 'steps': 29796, 'loss/train': 1.089059591293335}
+02/25/2022 16:10:50 - INFO - codeparrot_training - Step 29797: {'lr': 0.0001885092549397184, 'samples': 15256576, 'steps': 29797, 'loss/train': 1.461957573890686}
+02/25/2022 16:10:56 - INFO - codeparrot_training - Step 29798: {'lr': 0.0001884933952772321, 'samples': 15257088, 'steps': 29798, 'loss/train': 1.9711018800735474}
+02/25/2022 16:10:59 - INFO - codeparrot_training - Step 29799: {'lr': 0.00018847753587822053, 'samples': 15257600, 'steps': 29799, 'loss/train': 1.9362515211105347}
+02/25/2022 16:11:05 - INFO - codeparrot_training - Step 29800: {'lr': 0.00018846167674275175, 'samples': 15258112, 'steps': 29800, 'loss/train': 1.3598463535308838}
+02/25/2022 16:11:08 - INFO - codeparrot_training - Step 29801: {'lr': 0.0001884458178708936, 'samples': 15258624, 'steps': 29801, 'loss/train': 1.8957009315490723}
+02/25/2022 16:11:14 - INFO - codeparrot_training - Step 29802: {'lr': 0.00018842995926271393, 'samples': 15259136, 'steps': 29802, 'loss/train': 2.2282369136810303}
+02/25/2022 16:11:17 - INFO - codeparrot_training - Step 29803: {'lr': 0.00018841410091828087, 'samples': 15259648, 'steps': 29803, 'loss/train': 1.5105842351913452}
+02/25/2022 16:11:23 - INFO - codeparrot_training - Step 29804: {'lr': 0.00018839824283766216, 'samples': 15260160, 'steps': 29804, 'loss/train': 1.987215280532837}
+02/25/2022 16:11:26 - INFO - codeparrot_training - Step 29805: {'lr': 0.00018838238502092585, 'samples': 15260672, 'steps': 29805, 'loss/train': 1.5523611307144165}
+02/25/2022 16:11:32 - INFO - codeparrot_training - Step 29806: {'lr': 0.00018836652746813984, 'samples': 15261184, 'steps': 29806, 'loss/train': 1.0876392126083374}
+02/25/2022 16:11:35 - INFO - codeparrot_training - Step 29807: {'lr': 0.00018835067017937206, 'samples': 15261696, 'steps': 29807, 'loss/train': 2.51267409324646}
+02/25/2022 16:11:41 - INFO - codeparrot_training - Step 29808: {'lr': 0.0001883348131546904, 'samples': 15262208, 'steps': 29808, 'loss/train': 1.921204686164856}
+02/25/2022 16:11:44 - INFO - codeparrot_training - Step 29809: {'lr': 0.00018831895639416288, 'samples': 15262720, 'steps': 29809, 'loss/train': 2.3218698501586914}
+02/25/2022 16:11:50 - INFO - codeparrot_training - Step 29810: {'lr': 0.00018830309989785736, 'samples': 15263232, 'steps': 29810, 'loss/train': 1.2432307004928589}
+02/25/2022 16:11:54 - INFO - codeparrot_training - Step 29811: {'lr': 0.00018828724366584172, 'samples': 15263744, 'steps': 29811, 'loss/train': 1.2498656511306763}
+02/25/2022 16:12:00 - INFO - codeparrot_training - Step 29812: {'lr': 0.00018827138769818391, 'samples': 15264256, 'steps': 29812, 'loss/train': 1.3791866302490234}
+02/25/2022 16:12:03 - INFO - codeparrot_training - Step 29813: {'lr': 0.0001882555319949519, 'samples': 15264768, 'steps': 29813, 'loss/train': 1.654335379600525}
+02/25/2022 16:12:09 - INFO - codeparrot_training - Step 29814: {'lr': 0.00018823967655621368, 'samples': 15265280, 'steps': 29814, 'loss/train': 0.4004649519920349}
+02/25/2022 16:12:12 - INFO - codeparrot_training - Step 29815: {'lr': 0.000188223821382037, 'samples': 15265792, 'steps': 29815, 'loss/train': 1.8729444742202759}
+02/25/2022 16:12:18 - INFO - codeparrot_training - Step 29816: {'lr': 0.00018820796647248982, 'samples': 15266304, 'steps': 29816, 'loss/train': 1.3981703519821167}
+02/25/2022 16:12:21 - INFO - codeparrot_training - Step 29817: {'lr': 0.00018819211182764006, 'samples': 15266816, 'steps': 29817, 'loss/train': 3.234819173812866}
+02/25/2022 16:12:27 - INFO - codeparrot_training - Step 29818: {'lr': 0.0001881762574475558, 'samples': 15267328, 'steps': 29818, 'loss/train': 1.0996278524398804}
+02/25/2022 16:12:30 - INFO - codeparrot_training - Step 29819: {'lr': 0.00018816040333230472, 'samples': 15267840, 'steps': 29819, 'loss/train': 1.0675584077835083}
+02/25/2022 16:12:37 - INFO - codeparrot_training - Step 29820: {'lr': 0.00018814454948195487, 'samples': 15268352, 'steps': 29820, 'loss/train': 2.1080851554870605}
+02/25/2022 16:12:40 - INFO - codeparrot_training - Step 29821: {'lr': 0.0001881286958965741, 'samples': 15268864, 'steps': 29821, 'loss/train': 1.9768056869506836}
+02/25/2022 16:12:45 - INFO - codeparrot_training - Step 29822: {'lr': 0.00018811284257623043, 'samples': 15269376, 'steps': 29822, 'loss/train': 1.400217056274414}
+02/25/2022 16:12:49 - INFO - codeparrot_training - Step 29823: {'lr': 0.0001880969895209916, 'samples': 15269888, 'steps': 29823, 'loss/train': 0.8915282487869263}
+02/25/2022 16:12:55 - INFO - codeparrot_training - Step 29824: {'lr': 0.00018808113673092564, 'samples': 15270400, 'steps': 29824, 'loss/train': 1.4705545902252197}
+02/25/2022 16:12:58 - INFO - codeparrot_training - Step 29825: {'lr': 0.0001880652842061004, 'samples': 15270912, 'steps': 29825, 'loss/train': 1.9595308303833008}
+02/25/2022 16:13:04 - INFO - codeparrot_training - Step 29826: {'lr': 0.0001880494319465838, 'samples': 15271424, 'steps': 29826, 'loss/train': 1.4071155786514282}
+02/25/2022 16:13:07 - INFO - codeparrot_training - Step 29827: {'lr': 0.00018803357995244392, 'samples': 15271936, 'steps': 29827, 'loss/train': 1.5162628889083862}
+02/25/2022 16:13:13 - INFO - codeparrot_training - Step 29828: {'lr': 0.00018801772822374835, 'samples': 15272448, 'steps': 29828, 'loss/train': 2.0283267498016357}
+02/25/2022 16:13:16 - INFO - codeparrot_training - Step 29829: {'lr': 0.0001880018767605652, 'samples': 15272960, 'steps': 29829, 'loss/train': 1.5230984687805176}
+02/25/2022 16:13:22 - INFO - codeparrot_training - Step 29830: {'lr': 0.00018798602556296227, 'samples': 15273472, 'steps': 29830, 'loss/train': 2.1076390743255615}
+02/25/2022 16:13:26 - INFO - codeparrot_training - Step 29831: {'lr': 0.00018797017463100766, 'samples': 15273984, 'steps': 29831, 'loss/train': 2.198701858520508}
+02/25/2022 16:13:31 - INFO - codeparrot_training - Step 29832: {'lr': 0.000187954323964769, 'samples': 15274496, 'steps': 29832, 'loss/train': 0.7351523637771606}
+02/25/2022 16:13:35 - INFO - codeparrot_training - Step 29833: {'lr': 0.00018793847356431432, 'samples': 15275008, 'steps': 29833, 'loss/train': 2.197909355163574}
+02/25/2022 16:13:40 - INFO - codeparrot_training - Step 29834: {'lr': 0.0001879226234297115, 'samples': 15275520, 'steps': 29834, 'loss/train': 1.4007171392440796}
+02/25/2022 16:13:46 - INFO - codeparrot_training - Step 29835: {'lr': 0.00018790677356102847, 'samples': 15276032, 'steps': 29835, 'loss/train': 1.749891996383667}
+02/25/2022 16:13:49 - INFO - codeparrot_training - Step 29836: {'lr': 0.0001878909239583332, 'samples': 15276544, 'steps': 29836, 'loss/train': 1.1606929302215576}
+02/25/2022 16:13:55 - INFO - codeparrot_training - Step 29837: {'lr': 0.0001878750746216934, 'samples': 15277056, 'steps': 29837, 'loss/train': 2.2587380409240723}
+02/25/2022 16:13:58 - INFO - codeparrot_training - Step 29838: {'lr': 0.000187859225551177, 'samples': 15277568, 'steps': 29838, 'loss/train': 2.2455410957336426}
+02/25/2022 16:14:04 - INFO - codeparrot_training - Step 29839: {'lr': 0.00018784337674685201, 'samples': 15278080, 'steps': 29839, 'loss/train': 2.0633840560913086}
+02/25/2022 16:14:08 - INFO - codeparrot_training - Step 29840: {'lr': 0.00018782752820878634, 'samples': 15278592, 'steps': 29840, 'loss/train': 1.5649887323379517}
+02/25/2022 16:14:13 - INFO - codeparrot_training - Step 29841: {'lr': 0.0001878116799370477, 'samples': 15279104, 'steps': 29841, 'loss/train': 1.1138519048690796}
+02/25/2022 16:14:17 - INFO - codeparrot_training - Step 29842: {'lr': 0.00018779583193170407, 'samples': 15279616, 'steps': 29842, 'loss/train': 1.6223907470703125}
+02/25/2022 16:14:22 - INFO - codeparrot_training - Step 29843: {'lr': 0.00018777998419282338, 'samples': 15280128, 'steps': 29843, 'loss/train': 2.567366600036621}
+02/25/2022 16:14:26 - INFO - codeparrot_training - Step 29844: {'lr': 0.0001877641367204735, 'samples': 15280640, 'steps': 29844, 'loss/train': 1.1580398082733154}
+02/25/2022 16:14:31 - INFO - codeparrot_training - Step 29845: {'lr': 0.00018774828951472223, 'samples': 15281152, 'steps': 29845, 'loss/train': 2.1439363956451416}
+02/25/2022 16:14:35 - INFO - codeparrot_training - Step 29846: {'lr': 0.00018773244257563754, 'samples': 15281664, 'steps': 29846, 'loss/train': 1.6922271251678467}
+02/25/2022 16:14:40 - INFO - codeparrot_training - Step 29847: {'lr': 0.00018771659590328732, 'samples': 15282176, 'steps': 29847, 'loss/train': 1.844558835029602}
+02/25/2022 16:14:43 - INFO - codeparrot_training - Step 29848: {'lr': 0.0001877007494977394, 'samples': 15282688, 'steps': 29848, 'loss/train': 1.942375659942627}
+02/25/2022 16:14:49 - INFO - codeparrot_training - Step 29849: {'lr': 0.0001876849033590617, 'samples': 15283200, 'steps': 29849, 'loss/train': 1.2581619024276733}
+02/25/2022 16:14:52 - INFO - codeparrot_training - Step 29850: {'lr': 0.00018766905748732212, 'samples': 15283712, 'steps': 29850, 'loss/train': 2.7367358207702637}
+02/25/2022 16:14:58 - INFO - codeparrot_training - Step 29851: {'lr': 0.00018765321188258846, 'samples': 15284224, 'steps': 29851, 'loss/train': 1.8691809177398682}
+02/25/2022 16:15:02 - INFO - codeparrot_training - Step 29852: {'lr': 0.00018763736654492863, 'samples': 15284736, 'steps': 29852, 'loss/train': 1.7424983978271484}
+02/25/2022 16:15:07 - INFO - codeparrot_training - Step 29853: {'lr': 0.0001876215214744106, 'samples': 15285248, 'steps': 29853, 'loss/train': 2.612626314163208}
+02/25/2022 16:15:11 - INFO - codeparrot_training - Step 29854: {'lr': 0.00018760567667110213, 'samples': 15285760, 'steps': 29854, 'loss/train': 1.677430510520935}
+02/25/2022 16:15:17 - INFO - codeparrot_training - Step 29855: {'lr': 0.00018758983213507108, 'samples': 15286272, 'steps': 29855, 'loss/train': 1.66335928440094}
+02/25/2022 16:15:20 - INFO - codeparrot_training - Step 29856: {'lr': 0.00018757398786638536, 'samples': 15286784, 'steps': 29856, 'loss/train': 1.9988110065460205}
+02/25/2022 16:15:26 - INFO - codeparrot_training - Step 29857: {'lr': 0.00018755814386511296, 'samples': 15287296, 'steps': 29857, 'loss/train': 2.092353343963623}
+02/25/2022 16:15:29 - INFO - codeparrot_training - Step 29858: {'lr': 0.00018754230013132156, 'samples': 15287808, 'steps': 29858, 'loss/train': 0.9833886623382568}
+02/25/2022 16:15:35 - INFO - codeparrot_training - Step 29859: {'lr': 0.0001875264566650791, 'samples': 15288320, 'steps': 29859, 'loss/train': 1.6744505167007446}
+02/25/2022 16:15:38 - INFO - codeparrot_training - Step 29860: {'lr': 0.00018751061346645347, 'samples': 15288832, 'steps': 29860, 'loss/train': 1.8313595056533813}
+02/25/2022 16:15:44 - INFO - codeparrot_training - Step 29861: {'lr': 0.00018749477053551253, 'samples': 15289344, 'steps': 29861, 'loss/train': 0.6466661095619202}
+02/25/2022 16:15:47 - INFO - codeparrot_training - Step 29862: {'lr': 0.00018747892787232423, 'samples': 15289856, 'steps': 29862, 'loss/train': 1.9630857706069946}
+02/25/2022 16:15:53 - INFO - codeparrot_training - Step 29863: {'lr': 0.0001874630854769563, 'samples': 15290368, 'steps': 29863, 'loss/train': 2.6020519733428955}
+02/25/2022 16:15:56 - INFO - codeparrot_training - Step 29864: {'lr': 0.0001874472433494766, 'samples': 15290880, 'steps': 29864, 'loss/train': 1.771794080734253}
+02/25/2022 16:16:02 - INFO - codeparrot_training - Step 29865: {'lr': 0.00018743140148995307, 'samples': 15291392, 'steps': 29865, 'loss/train': 1.7185612916946411}
+02/25/2022 16:16:06 - INFO - codeparrot_training - Step 29866: {'lr': 0.00018741555989845366, 'samples': 15291904, 'steps': 29866, 'loss/train': 1.4438657760620117}
+02/25/2022 16:16:11 - INFO - codeparrot_training - Step 29867: {'lr': 0.00018739971857504597, 'samples': 15292416, 'steps': 29867, 'loss/train': 1.6309866905212402}
+02/25/2022 16:16:15 - INFO - codeparrot_training - Step 29868: {'lr': 0.00018738387751979804, 'samples': 15292928, 'steps': 29868, 'loss/train': 0.8186879754066467}
+02/25/2022 16:16:20 - INFO - codeparrot_training - Step 29869: {'lr': 0.0001873680367327777, 'samples': 15293440, 'steps': 29869, 'loss/train': 2.065586566925049}
+02/25/2022 16:16:26 - INFO - codeparrot_training - Step 29870: {'lr': 0.0001873521962140529, 'samples': 15293952, 'steps': 29870, 'loss/train': 1.1738897562026978}
+02/25/2022 16:16:29 - INFO - codeparrot_training - Step 29871: {'lr': 0.0001873363559636913, 'samples': 15294464, 'steps': 29871, 'loss/train': 1.2587348222732544}
+02/25/2022 16:16:35 - INFO - codeparrot_training - Step 29872: {'lr': 0.00018732051598176086, 'samples': 15294976, 'steps': 29872, 'loss/train': 2.7985191345214844}
+02/25/2022 16:16:38 - INFO - codeparrot_training - Step 29873: {'lr': 0.00018730467626832943, 'samples': 15295488, 'steps': 29873, 'loss/train': 1.3569531440734863}
+02/25/2022 16:16:44 - INFO - codeparrot_training - Step 29874: {'lr': 0.00018728883682346482, 'samples': 15296000, 'steps': 29874, 'loss/train': 0.3122064471244812}
+02/25/2022 16:16:47 - INFO - codeparrot_training - Step 29875: {'lr': 0.00018727299764723505, 'samples': 15296512, 'steps': 29875, 'loss/train': 1.2292375564575195}
+02/25/2022 16:16:53 - INFO - codeparrot_training - Step 29876: {'lr': 0.0001872571587397077, 'samples': 15297024, 'steps': 29876, 'loss/train': 1.993888020515442}
+02/25/2022 16:16:57 - INFO - codeparrot_training - Step 29877: {'lr': 0.00018724132010095084, 'samples': 15297536, 'steps': 29877, 'loss/train': 1.970518708229065}
+02/25/2022 16:17:03 - INFO - codeparrot_training - Step 29878: {'lr': 0.00018722548173103216, 'samples': 15298048, 'steps': 29878, 'loss/train': 2.3291068077087402}
+02/25/2022 16:17:06 - INFO - codeparrot_training - Step 29879: {'lr': 0.00018720964363001975, 'samples': 15298560, 'steps': 29879, 'loss/train': 0.7262600064277649}
+02/25/2022 16:17:12 - INFO - codeparrot_training - Step 29880: {'lr': 0.00018719380579798112, 'samples': 15299072, 'steps': 29880, 'loss/train': 2.0507781505584717}
+02/25/2022 16:17:15 - INFO - codeparrot_training - Step 29881: {'lr': 0.00018717796823498434, 'samples': 15299584, 'steps': 29881, 'loss/train': 1.4369564056396484}
+02/25/2022 16:17:21 - INFO - codeparrot_training - Step 29882: {'lr': 0.00018716213094109715, 'samples': 15300096, 'steps': 29882, 'loss/train': 2.242964267730713}
+02/25/2022 16:17:24 - INFO - codeparrot_training - Step 29883: {'lr': 0.00018714629391638758, 'samples': 15300608, 'steps': 29883, 'loss/train': 1.3677055835723877}
+02/25/2022 16:17:30 - INFO - codeparrot_training - Step 29884: {'lr': 0.0001871304571609232, 'samples': 15301120, 'steps': 29884, 'loss/train': 1.5618127584457397}
+02/25/2022 16:17:33 - INFO - codeparrot_training - Step 29885: {'lr': 0.00018711462067477197, 'samples': 15301632, 'steps': 29885, 'loss/train': 2.276064157485962}
+02/25/2022 16:17:40 - INFO - codeparrot_training - Step 29886: {'lr': 0.00018709878445800177, 'samples': 15302144, 'steps': 29886, 'loss/train': 0.7911723852157593}
+02/25/2022 16:17:43 - INFO - codeparrot_training - Step 29887: {'lr': 0.0001870829485106804, 'samples': 15302656, 'steps': 29887, 'loss/train': 0.036495327949523926}
+02/25/2022 16:17:49 - INFO - codeparrot_training - Step 29888: {'lr': 0.00018706711283287576, 'samples': 15303168, 'steps': 29888, 'loss/train': 1.5549920797348022}
+02/25/2022 16:17:52 - INFO - codeparrot_training - Step 29889: {'lr': 0.00018705127742465555, 'samples': 15303680, 'steps': 29889, 'loss/train': 1.387128233909607}
+02/25/2022 16:17:57 - INFO - codeparrot_training - Step 29890: {'lr': 0.00018703544228608775, 'samples': 15304192, 'steps': 29890, 'loss/train': 1.522879958152771}
+02/25/2022 16:18:01 - INFO - codeparrot_training - Step 29891: {'lr': 0.00018701960741724003, 'samples': 15304704, 'steps': 29891, 'loss/train': 2.4289002418518066}
+02/25/2022 16:18:07 - INFO - codeparrot_training - Step 29892: {'lr': 0.00018700377281818037, 'samples': 15305216, 'steps': 29892, 'loss/train': 0.4750358462333679}
+02/25/2022 16:18:10 - INFO - codeparrot_training - Step 29893: {'lr': 0.00018698793848897656, 'samples': 15305728, 'steps': 29893, 'loss/train': 1.9496701955795288}
+02/25/2022 16:18:15 - INFO - codeparrot_training - Step 29894: {'lr': 0.00018697210442969637, 'samples': 15306240, 'steps': 29894, 'loss/train': 2.053495407104492}
+02/25/2022 16:18:19 - INFO - codeparrot_training - Step 29895: {'lr': 0.00018695627064040766, 'samples': 15306752, 'steps': 29895, 'loss/train': 1.6171382665634155}
+02/25/2022 16:18:25 - INFO - codeparrot_training - Step 29896: {'lr': 0.00018694043712117828, 'samples': 15307264, 'steps': 29896, 'loss/train': 1.5707803964614868}
+02/25/2022 16:18:28 - INFO - codeparrot_training - Step 29897: {'lr': 0.00018692460387207616, 'samples': 15307776, 'steps': 29897, 'loss/train': 1.1517490148544312}
+02/25/2022 16:18:34 - INFO - codeparrot_training - Step 29898: {'lr': 0.0001869087708931689, 'samples': 15308288, 'steps': 29898, 'loss/train': 3.257429838180542}
+02/25/2022 16:18:37 - INFO - codeparrot_training - Step 29899: {'lr': 0.00018689293818452437, 'samples': 15308800, 'steps': 29899, 'loss/train': 1.906327486038208}
+02/25/2022 16:18:43 - INFO - codeparrot_training - Step 29900: {'lr': 0.00018687710574621051, 'samples': 15309312, 'steps': 29900, 'loss/train': 2.666914701461792}
+02/25/2022 16:18:47 - INFO - codeparrot_training - Step 29901: {'lr': 0.0001868612735782952, 'samples': 15309824, 'steps': 29901, 'loss/train': 1.3863444328308105}
+02/25/2022 16:18:53 - INFO - codeparrot_training - Step 29902: {'lr': 0.000186845441680846, 'samples': 15310336, 'steps': 29902, 'loss/train': 1.2310504913330078}
+02/25/2022 16:18:57 - INFO - codeparrot_training - Step 29903: {'lr': 0.0001868296100539309, 'samples': 15310848, 'steps': 29903, 'loss/train': 1.6549617052078247}
+02/25/2022 16:19:02 - INFO - codeparrot_training - Step 29904: {'lr': 0.00018681377869761767, 'samples': 15311360, 'steps': 29904, 'loss/train': 1.6674079895019531}
+02/25/2022 16:19:06 - INFO - codeparrot_training - Step 29905: {'lr': 0.00018679794761197426, 'samples': 15311872, 'steps': 29905, 'loss/train': 0.11368287354707718}
+02/25/2022 16:19:11 - INFO - codeparrot_training - Step 29906: {'lr': 0.00018678211679706828, 'samples': 15312384, 'steps': 29906, 'loss/train': 1.024168610572815}
+02/25/2022 16:19:15 - INFO - codeparrot_training - Step 29907: {'lr': 0.00018676628625296764, 'samples': 15312896, 'steps': 29907, 'loss/train': 2.765545606613159}
+02/25/2022 16:19:20 - INFO - codeparrot_training - Step 29908: {'lr': 0.00018675045597974014, 'samples': 15313408, 'steps': 29908, 'loss/train': 1.0893880128860474}
+02/25/2022 16:19:24 - INFO - codeparrot_training - Step 29909: {'lr': 0.0001867346259774536, 'samples': 15313920, 'steps': 29909, 'loss/train': 1.813191294670105}
+02/25/2022 16:19:29 - INFO - codeparrot_training - Step 29910: {'lr': 0.00018671879624617592, 'samples': 15314432, 'steps': 29910, 'loss/train': 1.8024532794952393}
+02/25/2022 16:19:33 - INFO - codeparrot_training - Step 29911: {'lr': 0.00018670296678597475, 'samples': 15314944, 'steps': 29911, 'loss/train': 1.6160545349121094}
+02/25/2022 16:19:39 - INFO - codeparrot_training - Step 29912: {'lr': 0.00018668713759691796, 'samples': 15315456, 'steps': 29912, 'loss/train': 1.8470666408538818}
+02/25/2022 16:19:43 - INFO - codeparrot_training - Step 29913: {'lr': 0.0001866713086790734, 'samples': 15315968, 'steps': 29913, 'loss/train': 1.7411481142044067}
+02/25/2022 16:19:48 - INFO - codeparrot_training - Step 29914: {'lr': 0.00018665548003250892, 'samples': 15316480, 'steps': 29914, 'loss/train': 1.9195104837417603}
+02/25/2022 16:19:52 - INFO - codeparrot_training - Step 29915: {'lr': 0.00018663965165729212, 'samples': 15316992, 'steps': 29915, 'loss/train': 2.3886818885803223}
+02/25/2022 16:19:57 - INFO - codeparrot_training - Step 29916: {'lr': 0.00018662382355349097, 'samples': 15317504, 'steps': 29916, 'loss/train': 1.9789098501205444}
+02/25/2022 16:20:01 - INFO - codeparrot_training - Step 29917: {'lr': 0.00018660799572117322, 'samples': 15318016, 'steps': 29917, 'loss/train': 2.1190907955169678}
+02/25/2022 16:20:06 - INFO - codeparrot_training - Step 29918: {'lr': 0.00018659216816040681, 'samples': 15318528, 'steps': 29918, 'loss/train': 1.5890998840332031}
+02/25/2022 16:20:10 - INFO - codeparrot_training - Step 29919: {'lr': 0.0001865763408712593, 'samples': 15319040, 'steps': 29919, 'loss/train': 0.4515683650970459}
+02/25/2022 16:20:15 - INFO - codeparrot_training - Step 29920: {'lr': 0.0001865605138537986, 'samples': 15319552, 'steps': 29920, 'loss/train': 1.984484076499939}
+02/25/2022 16:20:18 - INFO - codeparrot_training - Step 29921: {'lr': 0.00018654468710809253, 'samples': 15320064, 'steps': 29921, 'loss/train': 1.7808912992477417}
+02/25/2022 16:20:25 - INFO - codeparrot_training - Step 29922: {'lr': 0.00018652886063420888, 'samples': 15320576, 'steps': 29922, 'loss/train': 1.6435508728027344}
+02/25/2022 16:20:31 - INFO - codeparrot_training - Step 29923: {'lr': 0.0001865130344322155, 'samples': 15321088, 'steps': 29923, 'loss/train': 1.0891015529632568}
+02/25/2022 16:20:34 - INFO - codeparrot_training - Step 29924: {'lr': 0.00018649720850218005, 'samples': 15321600, 'steps': 29924, 'loss/train': 1.3649837970733643}
+02/25/2022 16:20:40 - INFO - codeparrot_training - Step 29925: {'lr': 0.00018648138284417038, 'samples': 15322112, 'steps': 29925, 'loss/train': 2.976989984512329}
+02/25/2022 16:20:43 - INFO - codeparrot_training - Step 29926: {'lr': 0.00018646555745825428, 'samples': 15322624, 'steps': 29926, 'loss/train': 1.773169994354248}
+02/25/2022 16:20:47 - INFO - codeparrot_training - Step 29927: {'lr': 0.0001864497323444997, 'samples': 15323136, 'steps': 29927, 'loss/train': 0.02213803492486477}
+02/25/2022 16:20:52 - INFO - codeparrot_training - Step 29928: {'lr': 0.00018643390750297417, 'samples': 15323648, 'steps': 29928, 'loss/train': 1.4953334331512451}
+02/25/2022 16:20:56 - INFO - codeparrot_training - Step 29929: {'lr': 0.00018641808293374558, 'samples': 15324160, 'steps': 29929, 'loss/train': 1.6758784055709839}
+02/25/2022 16:21:02 - INFO - codeparrot_training - Step 29930: {'lr': 0.00018640225863688178, 'samples': 15324672, 'steps': 29930, 'loss/train': 1.9071877002716064}
+02/25/2022 16:21:05 - INFO - codeparrot_training - Step 29931: {'lr': 0.00018638643461245052, 'samples': 15325184, 'steps': 29931, 'loss/train': 0.02334209717810154}
+02/25/2022 16:21:12 - INFO - codeparrot_training - Step 29932: {'lr': 0.00018637061086051955, 'samples': 15325696, 'steps': 29932, 'loss/train': 1.3272088766098022}
+02/25/2022 16:21:15 - INFO - codeparrot_training - Step 29933: {'lr': 0.00018635478738115662, 'samples': 15326208, 'steps': 29933, 'loss/train': 1.2069669961929321}
+02/25/2022 16:21:21 - INFO - codeparrot_training - Step 29934: {'lr': 0.0001863389641744297, 'samples': 15326720, 'steps': 29934, 'loss/train': 2.0012550354003906}
+02/25/2022 16:21:24 - INFO - codeparrot_training - Step 29935: {'lr': 0.00018632314124040635, 'samples': 15327232, 'steps': 29935, 'loss/train': 1.572792887687683}
+02/25/2022 16:21:30 - INFO - codeparrot_training - Step 29936: {'lr': 0.00018630731857915452, 'samples': 15327744, 'steps': 29936, 'loss/train': 2.4771556854248047}
+02/25/2022 16:21:33 - INFO - codeparrot_training - Step 29937: {'lr': 0.00018629149619074186, 'samples': 15328256, 'steps': 29937, 'loss/train': 1.747625708580017}
+02/25/2022 16:21:39 - INFO - codeparrot_training - Step 29938: {'lr': 0.00018627567407523615, 'samples': 15328768, 'steps': 29938, 'loss/train': 1.184655785560608}
+02/25/2022 16:21:44 - INFO - codeparrot_training - Step 29939: {'lr': 0.00018625985223270526, 'samples': 15329280, 'steps': 29939, 'loss/train': 2.02947998046875}
+02/25/2022 16:21:48 - INFO - codeparrot_training - Step 29940: {'lr': 0.00018624403066321696, 'samples': 15329792, 'steps': 29940, 'loss/train': 0.8118106722831726}
+02/25/2022 16:21:54 - INFO - codeparrot_training - Step 29941: {'lr': 0.00018622820936683898, 'samples': 15330304, 'steps': 29941, 'loss/train': 0.9568519592285156}
+02/25/2022 16:21:58 - INFO - codeparrot_training - Step 29942: {'lr': 0.00018621238834363906, 'samples': 15330816, 'steps': 29942, 'loss/train': 1.9766855239868164}
+02/25/2022 16:22:01 - INFO - codeparrot_training - Step 29943: {'lr': 0.00018619656759368504, 'samples': 15331328, 'steps': 29943, 'loss/train': 1.9341517686843872}
+02/25/2022 16:22:07 - INFO - codeparrot_training - Step 29944: {'lr': 0.00018618074711704462, 'samples': 15331840, 'steps': 29944, 'loss/train': 1.0070363283157349}
+02/25/2022 16:22:10 - INFO - codeparrot_training - Step 29945: {'lr': 0.00018616492691378572, 'samples': 15332352, 'steps': 29945, 'loss/train': 1.2417387962341309}
+02/25/2022 16:22:16 - INFO - codeparrot_training - Step 29946: {'lr': 0.00018614910698397591, 'samples': 15332864, 'steps': 29946, 'loss/train': 2.1367080211639404}
+02/25/2022 16:22:19 - INFO - codeparrot_training - Step 29947: {'lr': 0.00018613328732768305, 'samples': 15333376, 'steps': 29947, 'loss/train': 1.1206823587417603}
+02/25/2022 16:22:25 - INFO - codeparrot_training - Step 29948: {'lr': 0.00018611746794497492, 'samples': 15333888, 'steps': 29948, 'loss/train': 1.4667154550552368}
+02/25/2022 16:22:28 - INFO - codeparrot_training - Step 29949: {'lr': 0.00018610164883591936, 'samples': 15334400, 'steps': 29949, 'loss/train': 1.8145860433578491}
+02/25/2022 16:22:34 - INFO - codeparrot_training - Step 29950: {'lr': 0.00018608583000058395, 'samples': 15334912, 'steps': 29950, 'loss/train': 0.9149900674819946}
+02/25/2022 16:22:37 - INFO - codeparrot_training - Step 29951: {'lr': 0.00018607001143903658, 'samples': 15335424, 'steps': 29951, 'loss/train': 8.858648300170898}
+02/25/2022 16:22:43 - INFO - codeparrot_training - Step 29952: {'lr': 0.0001860541931513449, 'samples': 15335936, 'steps': 29952, 'loss/train': 1.905663251876831}
+02/25/2022 16:22:46 - INFO - codeparrot_training - Step 29953: {'lr': 0.00018603837513757692, 'samples': 15336448, 'steps': 29953, 'loss/train': 1.3951255083084106}
+02/25/2022 16:22:52 - INFO - codeparrot_training - Step 29954: {'lr': 0.0001860225573978001, 'samples': 15336960, 'steps': 29954, 'loss/train': 1.651943564414978}
+02/25/2022 16:22:56 - INFO - codeparrot_training - Step 29955: {'lr': 0.00018600673993208233, 'samples': 15337472, 'steps': 29955, 'loss/train': 1.7826783657073975}
+02/25/2022 16:23:01 - INFO - codeparrot_training - Step 29956: {'lr': 0.0001859909227404914, 'samples': 15337984, 'steps': 29956, 'loss/train': 1.4395363330841064}
+02/25/2022 16:23:05 - INFO - codeparrot_training - Step 29957: {'lr': 0.000185975105823095, 'samples': 15338496, 'steps': 29957, 'loss/train': 0.8780284523963928}
+02/25/2022 16:23:11 - INFO - codeparrot_training - Step 29958: {'lr': 0.000185959289179961, 'samples': 15339008, 'steps': 29958, 'loss/train': 1.5655696392059326}
+02/25/2022 16:23:14 - INFO - codeparrot_training - Step 29959: {'lr': 0.00018594347281115703, 'samples': 15339520, 'steps': 29959, 'loss/train': 1.1873611211776733}
+02/25/2022 16:23:20 - INFO - codeparrot_training - Step 29960: {'lr': 0.00018592765671675081, 'samples': 15340032, 'steps': 29960, 'loss/train': 1.6124969720840454}
+02/25/2022 16:23:23 - INFO - codeparrot_training - Step 29961: {'lr': 0.00018591184089681025, 'samples': 15340544, 'steps': 29961, 'loss/train': 1.7341086864471436}
+02/25/2022 16:23:29 - INFO - codeparrot_training - Step 29962: {'lr': 0.00018589602535140303, 'samples': 15341056, 'steps': 29962, 'loss/train': 1.8591160774230957}
+02/25/2022 16:23:34 - INFO - codeparrot_training - Step 29963: {'lr': 0.00018588021008059683, 'samples': 15341568, 'steps': 29963, 'loss/train': 2.252802848815918}
+02/25/2022 16:23:38 - INFO - codeparrot_training - Step 29964: {'lr': 0.00018586439508445942, 'samples': 15342080, 'steps': 29964, 'loss/train': 1.975447416305542}
+02/25/2022 16:23:43 - INFO - codeparrot_training - Step 29965: {'lr': 0.00018584858036305862, 'samples': 15342592, 'steps': 29965, 'loss/train': 0.9785515666007996}
+02/25/2022 16:23:47 - INFO - codeparrot_training - Step 29966: {'lr': 0.0001858327659164622, 'samples': 15343104, 'steps': 29966, 'loss/train': 1.7467445135116577}
+02/25/2022 16:23:53 - INFO - codeparrot_training - Step 29967: {'lr': 0.00018581695174473773, 'samples': 15343616, 'steps': 29967, 'loss/train': 2.2341299057006836}
+02/25/2022 16:23:56 - INFO - codeparrot_training - Step 29968: {'lr': 0.00018580113784795306, 'samples': 15344128, 'steps': 29968, 'loss/train': 2.1433560848236084}
+02/25/2022 16:24:02 - INFO - codeparrot_training - Step 29969: {'lr': 0.00018578532422617593, 'samples': 15344640, 'steps': 29969, 'loss/train': 1.321009635925293}
+02/25/2022 16:24:05 - INFO - codeparrot_training - Step 29970: {'lr': 0.00018576951087947413, 'samples': 15345152, 'steps': 29970, 'loss/train': 1.6169174909591675}
+02/25/2022 16:24:11 - INFO - codeparrot_training - Step 29971: {'lr': 0.0001857536978079154, 'samples': 15345664, 'steps': 29971, 'loss/train': 1.31284499168396}
+02/25/2022 16:24:15 - INFO - codeparrot_training - Step 29972: {'lr': 0.0001857378850115673, 'samples': 15346176, 'steps': 29972, 'loss/train': 1.9284205436706543}
+02/25/2022 16:24:20 - INFO - codeparrot_training - Step 29973: {'lr': 0.00018572207249049774, 'samples': 15346688, 'steps': 29973, 'loss/train': 1.9425864219665527}
+02/25/2022 16:24:24 - INFO - codeparrot_training - Step 29974: {'lr': 0.00018570626024477443, 'samples': 15347200, 'steps': 29974, 'loss/train': 2.069139003753662}
+02/25/2022 16:24:29 - INFO - codeparrot_training - Step 29975: {'lr': 0.00018569044827446512, 'samples': 15347712, 'steps': 29975, 'loss/train': 1.0136204957962036}
+02/25/2022 16:24:33 - INFO - codeparrot_training - Step 29976: {'lr': 0.00018567463657963746, 'samples': 15348224, 'steps': 29976, 'loss/train': 0.7865616083145142}
+02/25/2022 16:24:39 - INFO - codeparrot_training - Step 29977: {'lr': 0.0001856588251603592, 'samples': 15348736, 'steps': 29977, 'loss/train': 1.5731464624404907}
+02/25/2022 16:24:42 - INFO - codeparrot_training - Step 29978: {'lr': 0.0001856430140166982, 'samples': 15349248, 'steps': 29978, 'loss/train': 1.0307871103286743}
+02/25/2022 16:24:48 - INFO - codeparrot_training - Step 29979: {'lr': 0.00018562720314872205, 'samples': 15349760, 'steps': 29979, 'loss/train': 1.8848859071731567}
+02/25/2022 16:24:51 - INFO - codeparrot_training - Step 29980: {'lr': 0.00018561139255649848, 'samples': 15350272, 'steps': 29980, 'loss/train': 1.8795145750045776}
+02/25/2022 16:24:57 - INFO - codeparrot_training - Step 29981: {'lr': 0.00018559558224009532, 'samples': 15350784, 'steps': 29981, 'loss/train': 1.0919599533081055}
+02/25/2022 16:25:00 - INFO - codeparrot_training - Step 29982: {'lr': 0.00018557977219958016, 'samples': 15351296, 'steps': 29982, 'loss/train': 1.9145148992538452}
+02/25/2022 16:25:06 - INFO - codeparrot_training - Step 29983: {'lr': 0.00018556396243502084, 'samples': 15351808, 'steps': 29983, 'loss/train': 1.362343668937683}
+02/25/2022 16:25:09 - INFO - codeparrot_training - Step 29984: {'lr': 0.00018554815294648505, 'samples': 15352320, 'steps': 29984, 'loss/train': 4.103315830230713}
+02/25/2022 16:25:15 - INFO - codeparrot_training - Step 29985: {'lr': 0.00018553234373404055, 'samples': 15352832, 'steps': 29985, 'loss/train': 0.41832172870635986}
+02/25/2022 16:25:18 - INFO - codeparrot_training - Step 29986: {'lr': 0.00018551653479775493, 'samples': 15353344, 'steps': 29986, 'loss/train': 0.9002289175987244}
+02/25/2022 16:25:24 - INFO - codeparrot_training - Step 29987: {'lr': 0.00018550072613769604, 'samples': 15353856, 'steps': 29987, 'loss/train': 2.3278775215148926}
+02/25/2022 16:25:27 - INFO - codeparrot_training - Step 29988: {'lr': 0.0001854849177539316, 'samples': 15354368, 'steps': 29988, 'loss/train': 1.7065527439117432}
+02/25/2022 16:25:33 - INFO - codeparrot_training - Step 29989: {'lr': 0.00018546910964652925, 'samples': 15354880, 'steps': 29989, 'loss/train': 1.6275399923324585}
+02/25/2022 16:25:37 - INFO - codeparrot_training - Step 29990: {'lr': 0.0001854533018155567, 'samples': 15355392, 'steps': 29990, 'loss/train': 2.2125675678253174}
+02/25/2022 16:25:42 - INFO - codeparrot_training - Step 29991: {'lr': 0.00018543749426108174, 'samples': 15355904, 'steps': 29991, 'loss/train': 1.320961356163025}
+02/25/2022 16:25:46 - INFO - codeparrot_training - Step 29992: {'lr': 0.00018542168698317202, 'samples': 15356416, 'steps': 29992, 'loss/train': 1.1490603685379028}
+02/25/2022 16:25:51 - INFO - codeparrot_training - Step 29993: {'lr': 0.0001854058799818954, 'samples': 15356928, 'steps': 29993, 'loss/train': 1.6798889636993408}
+02/25/2022 16:25:55 - INFO - codeparrot_training - Step 29994: {'lr': 0.00018539007325731939, 'samples': 15357440, 'steps': 29994, 'loss/train': 0.7727236747741699}
+02/25/2022 16:26:00 - INFO - codeparrot_training - Step 29995: {'lr': 0.00018537426680951177, 'samples': 15357952, 'steps': 29995, 'loss/train': 2.1184613704681396}
+02/25/2022 16:26:04 - INFO - codeparrot_training - Step 29996: {'lr': 0.00018535846063854027, 'samples': 15358464, 'steps': 29996, 'loss/train': 2.0593769550323486}
+02/25/2022 16:26:09 - INFO - codeparrot_training - Step 29997: {'lr': 0.00018534265474447274, 'samples': 15358976, 'steps': 29997, 'loss/train': 1.5953224897384644}
+02/25/2022 16:26:13 - INFO - codeparrot_training - Step 29998: {'lr': 0.00018532684912737663, 'samples': 15359488, 'steps': 29998, 'loss/train': 2.062704086303711}
+02/25/2022 16:26:18 - INFO - codeparrot_training - Step 29999: {'lr': 0.00018531104378731973, 'samples': 15360000, 'steps': 29999, 'loss/train': 1.6037700176239014}
+02/25/2022 16:26:18 - INFO - codeparrot_training - Evaluating and saving model checkpoint