diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -36518,3 +36518,1009 @@ Use FP16 precision: False
 02/26/2022 00:10:35 - INFO - codeparrot_training - Step 35998: {'lr': 9.783560647896564e-05, 'samples': 18431488, 'steps': 35998, 'loss/train': 0.11420764774084091}
 02/26/2022 00:10:41 - INFO - codeparrot_training - Step 35999: {'lr': 9.7822624287453e-05, 'samples': 18432000, 'steps': 35999, 'loss/train': 1.2246854305267334}
 02/26/2022 00:10:41 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 00:10:59 - WARNING - huggingface_hub.repository - Several commits (36) will be pushed upstream.
+02/26/2022 00:10:59 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 00:11:34 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   aab1012..f71b356  floral-grass-11 -> floral-grass-11
+
+02/26/2022 00:11:39 - INFO - codeparrot_training - Step 36000: {'lr': 9.780964274781984e-05, 'samples': 18432512, 'steps': 36000, 'loss/train': 1.5604214668273926}
+02/26/2022 00:11:44 - INFO - codeparrot_training - Step 36001: {'lr': 9.77966618601219e-05, 'samples': 18433024, 'steps': 36001, 'loss/train': 1.2772327661514282}
+02/26/2022 00:11:48 - INFO - codeparrot_training - Step 36002: {'lr': 9.778368162441476e-05, 'samples': 18433536, 'steps': 36002, 'loss/train': 0.23902873694896698}
+02/26/2022 00:11:55 - INFO - codeparrot_training - Step 36003: {'lr': 9.7770702040754e-05, 'samples': 18434048, 'steps': 36003, 'loss/train': 1.5439203977584839}
+02/26/2022 00:11:58 - INFO - codeparrot_training - Step 36004: {'lr': 9.775772310919509e-05, 'samples': 18434560, 'steps': 36004, 'loss/train': 1.5662668943405151}
+02/26/2022 00:12:04 - INFO - codeparrot_training - Step 36005: {'lr': 9.774474482979388e-05, 'samples': 18435072, 'steps': 36005, 'loss/train': 1.638809084892273}
+02/26/2022 00:12:07 - INFO - codeparrot_training - Step 36006: {'lr': 9.773176720260582e-05, 'samples': 18435584, 'steps': 36006, 'loss/train': 0.26204827427864075}
+02/26/2022 00:12:13 - INFO - codeparrot_training - Step 36007: {'lr': 9.771879022768651e-05, 'samples': 18436096, 'steps': 36007, 'loss/train': 1.5781837701797485}
+02/26/2022 00:12:16 - INFO - codeparrot_training - Step 36008: {'lr': 9.770581390509148e-05, 'samples': 18436608, 'steps': 36008, 'loss/train': 2.4978692531585693}
+02/26/2022 00:12:22 - INFO - codeparrot_training - Step 36009: {'lr': 9.769283823487646e-05, 'samples': 18437120, 'steps': 36009, 'loss/train': 2.2008216381073}
+02/26/2022 00:12:25 - INFO - codeparrot_training - Step 36010: {'lr': 9.767986321709696e-05, 'samples': 18437632, 'steps': 36010, 'loss/train': 1.958353042602539}
+02/26/2022 00:12:31 - INFO - codeparrot_training - Step 36011: {'lr': 9.766688885180854e-05, 'samples': 18438144, 'steps': 36011, 'loss/train': 1.2244277000427246}
+02/26/2022 00:12:34 - INFO - codeparrot_training - Step 36012: {'lr': 9.765391513906673e-05, 'samples': 18438656, 'steps': 36012, 'loss/train': 1.9946836233139038}
+02/26/2022 00:12:41 - INFO - codeparrot_training - Step 36013: {'lr': 9.764094207892726e-05, 'samples': 18439168, 'steps': 36013, 'loss/train': 1.1914106607437134}
+02/26/2022 00:12:44 - INFO - codeparrot_training - Step 36014: {'lr': 9.76279696714456e-05, 'samples': 18439680, 'steps': 36014, 'loss/train': 1.0989683866500854}
+02/26/2022 00:12:50 - INFO - codeparrot_training - Step 36015: {'lr': 9.761499791667736e-05, 'samples': 18440192, 'steps': 36015, 'loss/train': 1.4466044902801514}
+02/26/2022 00:12:53 - INFO - codeparrot_training - Step 36016: {'lr': 9.760202681467798e-05, 'samples': 18440704, 'steps': 36016, 'loss/train': 0.8954662680625916}
+02/26/2022 00:12:59 - INFO - codeparrot_training - Step 36017: {'lr': 9.75890563655032e-05, 'samples': 18441216, 'steps': 36017, 'loss/train': 1.3574097156524658}
+02/26/2022 00:13:02 - INFO - codeparrot_training - Step 36018: {'lr': 9.757608656920844e-05, 'samples': 18441728, 'steps': 36018, 'loss/train': 1.735330581665039}
+02/26/2022 00:13:08 - INFO - codeparrot_training - Step 36019: {'lr': 9.756311742584947e-05, 'samples': 18442240, 'steps': 36019, 'loss/train': 9.147744178771973}
+02/26/2022 00:13:11 - INFO - codeparrot_training - Step 36020: {'lr': 9.755014893548156e-05, 'samples': 18442752, 'steps': 36020, 'loss/train': 0.40673404932022095}
+02/26/2022 00:13:17 - INFO - codeparrot_training - Step 36021: {'lr': 9.753718109816049e-05, 'samples': 18443264, 'steps': 36021, 'loss/train': 2.0481255054473877}
+02/26/2022 00:13:20 - INFO - codeparrot_training - Step 36022: {'lr': 9.752421391394162e-05, 'samples': 18443776, 'steps': 36022, 'loss/train': 1.4590117931365967}
+02/26/2022 00:13:27 - INFO - codeparrot_training - Step 36023: {'lr': 9.751124738288078e-05, 'samples': 18444288, 'steps': 36023, 'loss/train': 1.916585087776184}
+02/26/2022 00:13:30 - INFO - codeparrot_training - Step 36024: {'lr': 9.749828150503314e-05, 'samples': 18444800, 'steps': 36024, 'loss/train': 1.1615715026855469}
+02/26/2022 00:13:36 - INFO - codeparrot_training - Step 36025: {'lr': 9.748531628045456e-05, 'samples': 18445312, 'steps': 36025, 'loss/train': 0.618524968624115}
+02/26/2022 00:13:39 - INFO - codeparrot_training - Step 36026: {'lr': 9.747235170920036e-05, 'samples': 18445824, 'steps': 36026, 'loss/train': 0.7209410667419434}
+02/26/2022 00:13:45 - INFO - codeparrot_training - Step 36027: {'lr': 9.745938779132634e-05, 'samples': 18446336, 'steps': 36027, 'loss/train': 1.6152046918869019}
+02/26/2022 00:13:48 - INFO - codeparrot_training - Step 36028: {'lr': 9.744642452688767e-05, 'samples': 18446848, 'steps': 36028, 'loss/train': 1.5498669147491455}
+02/26/2022 00:13:54 - INFO - codeparrot_training - Step 36029: {'lr': 9.743346191594022e-05, 'samples': 18447360, 'steps': 36029, 'loss/train': 0.20733459293842316}
+02/26/2022 00:13:57 - INFO - codeparrot_training - Step 36030: {'lr': 9.742049995853925e-05, 'samples': 18447872, 'steps': 36030, 'loss/train': 1.7159196138381958}
+02/26/2022 00:14:03 - INFO - codeparrot_training - Step 36031: {'lr': 9.740753865474048e-05, 'samples': 18448384, 'steps': 36031, 'loss/train': 1.3167070150375366}
+02/26/2022 00:14:06 - INFO - codeparrot_training - Step 36032: {'lr': 9.739457800459939e-05, 'samples': 18448896, 'steps': 36032, 'loss/train': 1.5952965021133423}
+02/26/2022 00:14:13 - INFO - codeparrot_training - Step 36033: {'lr': 9.738161800817145e-05, 'samples': 18449408, 'steps': 36033, 'loss/train': 2.0263400077819824}
+02/26/2022 00:14:16 - INFO - codeparrot_training - Step 36034: {'lr': 9.736865866551211e-05, 'samples': 18449920, 'steps': 36034, 'loss/train': 1.3503931760787964}
+02/26/2022 00:14:22 - INFO - codeparrot_training - Step 36035: {'lr': 9.735569997667706e-05, 'samples': 18450432, 'steps': 36035, 'loss/train': 3.332951307296753}
+02/26/2022 00:14:25 - INFO - codeparrot_training - Step 36036: {'lr': 9.734274194172174e-05, 'samples': 18450944, 'steps': 36036, 'loss/train': 1.9804167747497559}
+02/26/2022 00:14:31 - INFO - codeparrot_training - Step 36037: {'lr': 9.732978456070163e-05, 'samples': 18451456, 'steps': 36037, 'loss/train': 1.8700439929962158}
+02/26/2022 00:14:34 - INFO - codeparrot_training - Step 36038: {'lr': 9.731682783367215e-05, 'samples': 18451968, 'steps': 36038, 'loss/train': 2.0220859050750732}
+02/26/2022 00:14:40 - INFO - codeparrot_training - Step 36039: {'lr': 9.7303871760689e-05, 'samples': 18452480, 'steps': 36039, 'loss/train': 1.0624700784683228}
+02/26/2022 00:14:45 - INFO - codeparrot_training - Step 36040: {'lr': 9.729091634180757e-05, 'samples': 18452992, 'steps': 36040, 'loss/train': 1.7062714099884033}
+02/26/2022 00:14:49 - INFO - codeparrot_training - Step 36041: {'lr': 9.727796157708335e-05, 'samples': 18453504, 'steps': 36041, 'loss/train': 1.4486877918243408}
+02/26/2022 00:14:55 - INFO - codeparrot_training - Step 36042: {'lr': 9.726500746657177e-05, 'samples': 18454016, 'steps': 36042, 'loss/train': 0.1205480694770813}
+02/26/2022 00:14:58 - INFO - codeparrot_training - Step 36043: {'lr': 9.72520540103285e-05, 'samples': 18454528, 'steps': 36043, 'loss/train': 1.0018887519836426}
+02/26/2022 00:15:02 - INFO - codeparrot_training - Step 36044: {'lr': 9.72391012084089e-05, 'samples': 18455040, 'steps': 36044, 'loss/train': 0.9732137322425842}
+02/26/2022 00:15:07 - INFO - codeparrot_training - Step 36045: {'lr': 9.722614906086849e-05, 'samples': 18455552, 'steps': 36045, 'loss/train': 1.1737741231918335}
+02/26/2022 00:15:12 - INFO - codeparrot_training - Step 36046: {'lr': 9.721319756776276e-05, 'samples': 18456064, 'steps': 36046, 'loss/train': 1.9837580919265747}
+02/26/2022 00:15:16 - INFO - codeparrot_training - Step 36047: {'lr': 9.72002467291471e-05, 'samples': 18456576, 'steps': 36047, 'loss/train': 2.022613763809204}
+02/26/2022 00:15:23 - INFO - codeparrot_training - Step 36048: {'lr': 9.718729654507713e-05, 'samples': 18457088, 'steps': 36048, 'loss/train': 0.04201718419790268}
+02/26/2022 00:15:26 - INFO - codeparrot_training - Step 36049: {'lr': 9.717434701560827e-05, 'samples': 18457600, 'steps': 36049, 'loss/train': 0.8163706064224243}
+02/26/2022 00:15:32 - INFO - codeparrot_training - Step 36050: {'lr': 9.716139814079594e-05, 'samples': 18458112, 'steps': 36050, 'loss/train': 1.975005030632019}
+02/26/2022 00:15:35 - INFO - codeparrot_training - Step 36051: {'lr': 9.71484499206956e-05, 'samples': 18458624, 'steps': 36051, 'loss/train': 0.0918508991599083}
+02/26/2022 00:15:41 - INFO - codeparrot_training - Step 36052: {'lr': 9.713550235536283e-05, 'samples': 18459136, 'steps': 36052, 'loss/train': 1.9509221315383911}
+02/26/2022 00:15:44 - INFO - codeparrot_training - Step 36053: {'lr': 9.712255544485296e-05, 'samples': 18459648, 'steps': 36053, 'loss/train': 2.816897392272949}
+02/26/2022 00:15:50 - INFO - codeparrot_training - Step 36054: {'lr': 9.710960918922171e-05, 'samples': 18460160, 'steps': 36054, 'loss/train': 2.4247937202453613}
+02/26/2022 00:15:53 - INFO - codeparrot_training - Step 36055: {'lr': 9.709666358852414e-05, 'samples': 18460672, 'steps': 36055, 'loss/train': 1.8876968622207642}
+02/26/2022 00:15:58 - INFO - codeparrot_training - Step 36056: {'lr': 9.708371864281601e-05, 'samples': 18461184, 'steps': 36056, 'loss/train': 2.1317458152770996}
+02/26/2022 00:16:02 - INFO - codeparrot_training - Step 36057: {'lr': 9.707077435215258e-05, 'samples': 18461696, 'steps': 36057, 'loss/train': 1.6023669242858887}
+02/26/2022 00:16:08 - INFO - codeparrot_training - Step 36058: {'lr': 9.705783071658958e-05, 'samples': 18462208, 'steps': 36058, 'loss/train': 0.7936185598373413}
+02/26/2022 00:16:12 - INFO - codeparrot_training - Step 36059: {'lr': 9.704488773618209e-05, 'samples': 18462720, 'steps': 36059, 'loss/train': 1.46337890625}
+02/26/2022 00:16:17 - INFO - codeparrot_training - Step 36060: {'lr': 9.703194541098581e-05, 'samples': 18463232, 'steps': 36060, 'loss/train': 0.6197915077209473}
+02/26/2022 00:16:21 - INFO - codeparrot_training - Step 36061: {'lr': 9.701900374105602e-05, 'samples': 18463744, 'steps': 36061, 'loss/train': 2.0371103286743164}
+02/26/2022 00:16:26 - INFO - codeparrot_training - Step 36062: {'lr': 9.700606272644841e-05, 'samples': 18464256, 'steps': 36062, 'loss/train': 0.45785778760910034}
+02/26/2022 00:16:30 - INFO - codeparrot_training - Step 36063: {'lr': 9.699312236721808e-05, 'samples': 18464768, 'steps': 36063, 'loss/train': 1.5471307039260864}
+02/26/2022 00:16:35 - INFO - codeparrot_training - Step 36064: {'lr': 9.698018266342071e-05, 'samples': 18465280, 'steps': 36064, 'loss/train': 1.319109559059143}
+02/26/2022 00:16:39 - INFO - codeparrot_training - Step 36065: {'lr': 9.696724361511156e-05, 'samples': 18465792, 'steps': 36065, 'loss/train': 2.1715946197509766}
+02/26/2022 00:16:44 - INFO - codeparrot_training - Step 36066: {'lr': 9.695430522234625e-05, 'samples': 18466304, 'steps': 36066, 'loss/train': 2.603968858718872}
+02/26/2022 00:16:48 - INFO - codeparrot_training - Step 36067: {'lr': 9.694136748518007e-05, 'samples': 18466816, 'steps': 36067, 'loss/train': 0.927971363067627}
+02/26/2022 00:16:53 - INFO - codeparrot_training - Step 36068: {'lr': 9.69284304036685e-05, 'samples': 18467328, 'steps': 36068, 'loss/train': 1.7105233669281006}
+02/26/2022 00:16:57 - INFO - codeparrot_training - Step 36069: {'lr': 9.69154939778668e-05, 'samples': 18467840, 'steps': 36069, 'loss/train': 1.2192550897598267}
+02/26/2022 00:17:03 - INFO - codeparrot_training - Step 36070: {'lr': 9.690255820783064e-05, 'samples': 18468352, 'steps': 36070, 'loss/train': 2.7139899730682373}
+02/26/2022 00:17:06 - INFO - codeparrot_training - Step 36071: {'lr': 9.688962309361529e-05, 'samples': 18468864, 'steps': 36071, 'loss/train': 1.2359057664871216}
+02/26/2022 00:17:12 - INFO - codeparrot_training - Step 36072: {'lr': 9.687668863527619e-05, 'samples': 18469376, 'steps': 36072, 'loss/train': 1.5974620580673218}
+02/26/2022 00:17:16 - INFO - codeparrot_training - Step 36073: {'lr': 9.686375483286864e-05, 'samples': 18469888, 'steps': 36073, 'loss/train': 0.8579695820808411}
+02/26/2022 00:17:21 - INFO - codeparrot_training - Step 36074: {'lr': 9.685082168644823e-05, 'samples': 18470400, 'steps': 36074, 'loss/train': 1.3160451650619507}
+02/26/2022 00:17:25 - INFO - codeparrot_training - Step 36075: {'lr': 9.683788919607026e-05, 'samples': 18470912, 'steps': 36075, 'loss/train': 1.2578535079956055}
+02/26/2022 00:17:30 - INFO - codeparrot_training - Step 36076: {'lr': 9.682495736179014e-05, 'samples': 18471424, 'steps': 36076, 'loss/train': 2.295518636703491}
+02/26/2022 00:17:34 - INFO - codeparrot_training - Step 36077: {'lr': 9.681202618366317e-05, 'samples': 18471936, 'steps': 36077, 'loss/train': 1.4106355905532837}
+02/26/2022 00:17:39 - INFO - codeparrot_training - Step 36078: {'lr': 9.679909566174494e-05, 'samples': 18472448, 'steps': 36078, 'loss/train': 1.4337424039840698}
+02/26/2022 00:17:42 - INFO - codeparrot_training - Step 36079: {'lr': 9.678616579609073e-05, 'samples': 18472960, 'steps': 36079, 'loss/train': 1.8247777223587036}
+02/26/2022 00:17:49 - INFO - codeparrot_training - Step 36080: {'lr': 9.677323658675594e-05, 'samples': 18473472, 'steps': 36080, 'loss/train': 1.671035885810852}
+02/26/2022 00:17:52 - INFO - codeparrot_training - Step 36081: {'lr': 9.676030803379585e-05, 'samples': 18473984, 'steps': 36081, 'loss/train': 1.8489046096801758}
+02/26/2022 00:17:58 - INFO - codeparrot_training - Step 36082: {'lr': 9.674738013726603e-05, 'samples': 18474496, 'steps': 36082, 'loss/train': 2.2339859008789062}
+02/26/2022 00:18:01 - INFO - codeparrot_training - Step 36083: {'lr': 9.673445289722177e-05, 'samples': 18475008, 'steps': 36083, 'loss/train': 0.49212533235549927}
+02/26/2022 00:18:07 - INFO - codeparrot_training - Step 36084: {'lr': 9.672152631371848e-05, 'samples': 18475520, 'steps': 36084, 'loss/train': 1.9349168539047241}
+02/26/2022 00:18:10 - INFO - codeparrot_training - Step 36085: {'lr': 9.670860038681136e-05, 'samples': 18476032, 'steps': 36085, 'loss/train': 2.043856620788574}
+02/26/2022 00:18:16 - INFO - codeparrot_training - Step 36086: {'lr': 9.669567511655602e-05, 'samples': 18476544, 'steps': 36086, 'loss/train': 2.254288911819458}
+02/26/2022 00:18:19 - INFO - codeparrot_training - Step 36087: {'lr': 9.668275050300776e-05, 'samples': 18477056, 'steps': 36087, 'loss/train': 2.0351157188415527}
+02/26/2022 00:18:25 - INFO - codeparrot_training - Step 36088: {'lr': 9.666982654622189e-05, 'samples': 18477568, 'steps': 36088, 'loss/train': 1.0550917387008667}
+02/26/2022 00:18:28 - INFO - codeparrot_training - Step 36089: {'lr': 9.66569032462537e-05, 'samples': 18478080, 'steps': 36089, 'loss/train': 1.5271868705749512}
+02/26/2022 00:18:35 - INFO - codeparrot_training - Step 36090: {'lr': 9.664398060315876e-05, 'samples': 18478592, 'steps': 36090, 'loss/train': 1.8364299535751343}
+02/26/2022 00:18:38 - INFO - codeparrot_training - Step 36091: {'lr': 9.663105861699229e-05, 'samples': 18479104, 'steps': 36091, 'loss/train': 1.8867496252059937}
+02/26/2022 00:18:43 - INFO - codeparrot_training - Step 36092: {'lr': 9.661813728780958e-05, 'samples': 18479616, 'steps': 36092, 'loss/train': 1.69874906539917}
+02/26/2022 00:18:47 - INFO - codeparrot_training - Step 36093: {'lr': 9.660521661566615e-05, 'samples': 18480128, 'steps': 36093, 'loss/train': 1.8722460269927979}
+02/26/2022 00:18:52 - INFO - codeparrot_training - Step 36094: {'lr': 9.659229660061728e-05, 'samples': 18480640, 'steps': 36094, 'loss/train': 1.763447880744934}
+02/26/2022 00:18:56 - INFO - codeparrot_training - Step 36095: {'lr': 9.657937724271829e-05, 'samples': 18481152, 'steps': 36095, 'loss/train': 1.0011274814605713}
+02/26/2022 00:19:01 - INFO - codeparrot_training - Step 36096: {'lr': 9.656645854202442e-05, 'samples': 18481664, 'steps': 36096, 'loss/train': 1.1618714332580566}
+02/26/2022 00:19:05 - INFO - codeparrot_training - Step 36097: {'lr': 9.655354049859133e-05, 'samples': 18482176, 'steps': 36097, 'loss/train': 2.0890953540802}
+02/26/2022 00:19:10 - INFO - codeparrot_training - Step 36098: {'lr': 9.654062311247397e-05, 'samples': 18482688, 'steps': 36098, 'loss/train': 1.7578837871551514}
+02/26/2022 00:19:14 - INFO - codeparrot_training - Step 36099: {'lr': 9.652770638372792e-05, 'samples': 18483200, 'steps': 36099, 'loss/train': 1.8363592624664307}
+02/26/2022 00:19:19 - INFO - codeparrot_training - Step 36100: {'lr': 9.651479031240836e-05, 'samples': 18483712, 'steps': 36100, 'loss/train': 2.721839666366577}
+02/26/2022 00:19:23 - INFO - codeparrot_training - Step 36101: {'lr': 9.650187489857079e-05, 'samples': 18484224, 'steps': 36101, 'loss/train': 1.9285857677459717}
+02/26/2022 00:19:28 - INFO - codeparrot_training - Step 36102: {'lr': 9.648896014227046e-05, 'samples': 18484736, 'steps': 36102, 'loss/train': 1.311247706413269}
+02/26/2022 00:19:32 - INFO - codeparrot_training - Step 36103: {'lr': 9.647604604356267e-05, 'samples': 18485248, 'steps': 36103, 'loss/train': 1.5463550090789795}
+02/26/2022 00:19:37 - INFO - codeparrot_training - Step 36104: {'lr': 9.646313260250267e-05, 'samples': 18485760, 'steps': 36104, 'loss/train': 2.2677383422851562}
+02/26/2022 00:19:41 - INFO - codeparrot_training - Step 36105: {'lr': 9.645021981914592e-05, 'samples': 18486272, 'steps': 36105, 'loss/train': 2.1401143074035645}
+02/26/2022 00:19:47 - INFO - codeparrot_training - Step 36106: {'lr': 9.64373076935477e-05, 'samples': 18486784, 'steps': 36106, 'loss/train': 3.245847225189209}
+02/26/2022 00:19:51 - INFO - codeparrot_training - Step 36107: {'lr': 9.642439622576327e-05, 'samples': 18487296, 'steps': 36107, 'loss/train': 1.6988332271575928}
+02/26/2022 00:19:56 - INFO - codeparrot_training - Step 36108: {'lr': 9.641148541584788e-05, 'samples': 18487808, 'steps': 36108, 'loss/train': 1.2002081871032715}
+02/26/2022 00:20:00 - INFO - codeparrot_training - Step 36109: {'lr': 9.639857526385701e-05, 'samples': 18488320, 'steps': 36109, 'loss/train': 1.9421395063400269}
+02/26/2022 00:20:05 - INFO - codeparrot_training - Step 36110: {'lr': 9.638566576984586e-05, 'samples': 18488832, 'steps': 36110, 'loss/train': 1.74909245967865}
+02/26/2022 00:20:09 - INFO - codeparrot_training - Step 36111: {'lr': 9.637275693386974e-05, 'samples': 18489344, 'steps': 36111, 'loss/train': 2.5859227180480957}
+02/26/2022 00:20:14 - INFO - codeparrot_training - Step 36112: {'lr': 9.635984875598389e-05, 'samples': 18489856, 'steps': 36112, 'loss/train': 2.328519821166992}
+02/26/2022 00:20:18 - INFO - codeparrot_training - Step 36113: {'lr': 9.63469412362437e-05, 'samples': 18490368, 'steps': 36113, 'loss/train': 1.3176090717315674}
+02/26/2022 00:20:23 - INFO - codeparrot_training - Step 36114: {'lr': 9.633403437470437e-05, 'samples': 18490880, 'steps': 36114, 'loss/train': 1.733176589012146}
+02/26/2022 00:20:27 - INFO - codeparrot_training - Step 36115: {'lr': 9.632112817142141e-05, 'samples': 18491392, 'steps': 36115, 'loss/train': 1.5681864023208618}
+02/26/2022 00:20:33 - INFO - codeparrot_training - Step 36116: {'lr': 9.630822262644976e-05, 'samples': 18491904, 'steps': 36116, 'loss/train': 1.6882433891296387}
+02/26/2022 00:20:36 - INFO - codeparrot_training - Step 36117: {'lr': 9.629531773984496e-05, 'samples': 18492416, 'steps': 36117, 'loss/train': 1.7538223266601562}
+02/26/2022 00:20:42 - INFO - codeparrot_training - Step 36118: {'lr': 9.628241351166214e-05, 'samples': 18492928, 'steps': 36118, 'loss/train': 1.1836498975753784}
+02/26/2022 00:20:45 - INFO - codeparrot_training - Step 36119: {'lr': 9.626950994195682e-05, 'samples': 18493440, 'steps': 36119, 'loss/train': 1.354833960533142}
+02/26/2022 00:20:51 - INFO - codeparrot_training - Step 36120: {'lr': 9.625660703078392e-05, 'samples': 18493952, 'steps': 36120, 'loss/train': 1.5595625638961792}
+02/26/2022 00:20:54 - INFO - codeparrot_training - Step 36121: {'lr': 9.6243704778199e-05, 'samples': 18494464, 'steps': 36121, 'loss/train': 2.852039337158203}
+02/26/2022 00:21:00 - INFO - codeparrot_training - Step 36122: {'lr': 9.623080318425715e-05, 'samples': 18494976, 'steps': 36122, 'loss/train': 2.465924024581909}
+02/26/2022 00:21:03 - INFO - codeparrot_training - Step 36123: {'lr': 9.621790224901387e-05, 'samples': 18495488, 'steps': 36123, 'loss/train': 2.143178939819336}
+02/26/2022 00:21:09 - INFO - codeparrot_training - Step 36124: {'lr': 9.620500197252408e-05, 'samples': 18496000, 'steps': 36124, 'loss/train': 1.795910120010376}
+02/26/2022 00:21:12 - INFO - codeparrot_training - Step 36125: {'lr': 9.619210235484333e-05, 'samples': 18496512, 'steps': 36125, 'loss/train': 0.8504828214645386}
+02/26/2022 00:21:18 - INFO - codeparrot_training - Step 36126: {'lr': 9.617920339602665e-05, 'samples': 18497024, 'steps': 36126, 'loss/train': 1.2204484939575195}
+02/26/2022 00:21:22 - INFO - codeparrot_training - Step 36127: {'lr': 9.616630509612955e-05, 'samples': 18497536, 'steps': 36127, 'loss/train': 1.0036232471466064}
+02/26/2022 00:21:27 - INFO - codeparrot_training - Step 36128: {'lr': 9.615340745520712e-05, 'samples': 18498048, 'steps': 36128, 'loss/train': 1.9192155599594116}
+02/26/2022 00:21:33 - INFO - codeparrot_training - Step 36129: {'lr': 9.614051047331469e-05, 'samples': 18498560, 'steps': 36129, 'loss/train': 1.749711036682129}
+02/26/2022 00:21:36 - INFO - codeparrot_training - Step 36130: {'lr': 9.612761415050741e-05, 'samples': 18499072, 'steps': 36130, 'loss/train': 1.9420716762542725}
+02/26/2022 00:21:42 - INFO - codeparrot_training - Step 36131: {'lr': 9.61147184868405e-05, 'samples': 18499584, 'steps': 36131, 'loss/train': 0.5370197296142578}
+02/26/2022 00:21:45 - INFO - codeparrot_training - Step 36132: {'lr': 9.610182348236934e-05, 'samples': 18500096, 'steps': 36132, 'loss/train': 7.3840813636779785}
+02/26/2022 00:21:51 - INFO - codeparrot_training - Step 36133: {'lr': 9.608892913714912e-05, 'samples': 18500608, 'steps': 36133, 'loss/train': 0.36018943786621094}
+02/26/2022 00:21:54 - INFO - codeparrot_training - Step 36134: {'lr': 9.607603545123505e-05, 'samples': 18501120, 'steps': 36134, 'loss/train': 1.453967809677124}
+02/26/2022 00:22:00 - INFO - codeparrot_training - Step 36135: {'lr': 9.606314242468228e-05, 'samples': 18501632, 'steps': 36135, 'loss/train': 1.7087455987930298}
+02/26/2022 00:22:04 - INFO - codeparrot_training - Step 36136: {'lr': 9.605025005754622e-05, 'samples': 18502144, 'steps': 36136, 'loss/train': 2.5946903228759766}
+02/26/2022 00:22:09 - INFO - codeparrot_training - Step 36137: {'lr': 9.6037358349882e-05, 'samples': 18502656, 'steps': 36137, 'loss/train': 1.105545163154602}
+02/26/2022 00:22:13 - INFO - codeparrot_training - Step 36138: {'lr': 9.602446730174485e-05, 'samples': 18503168, 'steps': 36138, 'loss/train': 3.0374724864959717}
+02/26/2022 00:22:18 - INFO - codeparrot_training - Step 36139: {'lr': 9.601157691318991e-05, 'samples': 18503680, 'steps': 36139, 'loss/train': 0.5579282641410828}
+02/26/2022 00:22:22 - INFO - codeparrot_training - Step 36140: {'lr': 9.599868718427256e-05, 'samples': 18504192, 'steps': 36140, 'loss/train': 2.259603977203369}
+02/26/2022 00:22:27 - INFO - codeparrot_training - Step 36141: {'lr': 9.598579811504792e-05, 'samples': 18504704, 'steps': 36141, 'loss/train': 1.8042490482330322}
+02/26/2022 00:22:31 - INFO - codeparrot_training - Step 36142: {'lr': 9.597290970557124e-05, 'samples': 18505216, 'steps': 36142, 'loss/train': 1.6345702409744263}
+02/26/2022 00:22:36 - INFO - codeparrot_training - Step 36143: {'lr': 9.596002195589759e-05, 'samples': 18505728, 'steps': 36143, 'loss/train': 0.7388705015182495}
+02/26/2022 00:22:40 - INFO - codeparrot_training - Step 36144: {'lr': 9.59471348660824e-05, 'samples': 18506240, 'steps': 36144, 'loss/train': 1.916875958442688}
+02/26/2022 00:22:45 - INFO - codeparrot_training - Step 36145: {'lr': 9.593424843618076e-05, 'samples': 18506752, 'steps': 36145, 'loss/train': 2.252598524093628}
+02/26/2022 00:22:49 - INFO - codeparrot_training - Step 36146: {'lr': 9.592136266624787e-05, 'samples': 18507264, 'steps': 36146, 'loss/train': 2.5656566619873047}
+02/26/2022 00:22:54 - INFO - codeparrot_training - Step 36147: {'lr': 9.590847755633885e-05, 'samples': 18507776, 'steps': 36147, 'loss/train': 2.0132853984832764}
+02/26/2022 00:22:58 - INFO - codeparrot_training - Step 36148: {'lr': 9.589559310650911e-05, 'samples': 18508288, 'steps': 36148, 'loss/train': 1.502086877822876}
+02/26/2022 00:23:03 - INFO - codeparrot_training - Step 36149: {'lr': 9.588270931681367e-05, 'samples': 18508800, 'steps': 36149, 'loss/train': 1.199326992034912}
+02/26/2022 00:23:07 - INFO - codeparrot_training - Step 36150: {'lr': 9.586982618730778e-05, 'samples': 18509312, 'steps': 36150, 'loss/train': 1.222579002380371}
+02/26/2022 00:23:13 - INFO - codeparrot_training - Step 36151: {'lr': 9.585694371804654e-05, 'samples': 18509824, 'steps': 36151, 'loss/train': 1.4391850233078003}
+02/26/2022 00:23:16 - INFO - codeparrot_training - Step 36152: {'lr': 9.584406190908527e-05, 'samples': 18510336, 'steps': 36152, 'loss/train': 0.9116042256355286}
+02/26/2022 00:23:22 - INFO - codeparrot_training - Step 36153: {'lr': 9.5831180760479e-05, 'samples': 18510848, 'steps': 36153, 'loss/train': 2.355449914932251}
+02/26/2022 00:23:25 - INFO - codeparrot_training - Step 36154: {'lr': 9.581830027228319e-05, 'samples': 18511360, 'steps': 36154, 'loss/train': 1.5806246995925903}
+02/26/2022 00:23:31 - INFO - codeparrot_training - Step 36155: {'lr': 9.580542044455265e-05, 'samples': 18511872, 'steps': 36155, 'loss/train': 1.011007308959961}
+02/26/2022 00:23:34 - INFO - codeparrot_training - Step 36156: {'lr': 9.579254127734279e-05, 'samples': 18512384, 'steps': 36156, 'loss/train': 1.5051230192184448}
+02/26/2022 00:23:40 - INFO - codeparrot_training - Step 36157: {'lr': 9.577966277070865e-05, 'samples': 18512896, 'steps': 36157, 'loss/train': 0.9843266606330872}
+02/26/2022 00:23:43 - INFO - codeparrot_training - Step 36158: {'lr': 9.576678492470564e-05, 'samples': 18513408, 'steps': 36158, 'loss/train': 0.6475926041603088}
+02/26/2022 00:23:49 - INFO - codeparrot_training - Step 36159: {'lr': 9.575390773938858e-05, 'samples': 18513920, 'steps': 36159, 'loss/train': 1.8057377338409424}
+02/26/2022 00:23:52 - INFO - codeparrot_training - Step 36160: {'lr': 9.574103121481287e-05, 'samples': 18514432, 'steps': 36160, 'loss/train': 1.9257701635360718}
+02/26/2022 00:23:59 - INFO - codeparrot_training - Step 36161: {'lr': 9.572815535103351e-05, 'samples': 18514944, 'steps': 36161, 'loss/train': 3.0665485858917236}
+02/26/2022 00:24:02 - INFO - codeparrot_training - Step 36162: {'lr': 9.571528014810585e-05, 'samples': 18515456, 'steps': 36162, 'loss/train': 2.3938939571380615}
+02/26/2022 00:24:08 - INFO - codeparrot_training - Step 36163: {'lr': 9.570240560608492e-05, 'samples': 18515968, 'steps': 36163, 'loss/train': 1.0796701908111572}
+02/26/2022 00:24:11 - INFO - codeparrot_training - Step 36164: {'lr': 9.568953172502589e-05, 'samples': 18516480, 'steps': 36164, 'loss/train': 0.8898747563362122}
+02/26/2022 00:24:17 - INFO - codeparrot_training - Step 36165: {'lr': 9.567665850498386e-05, 'samples': 18516992, 'steps': 36165, 'loss/train': 1.6962101459503174}
+02/26/2022 00:24:20 - INFO - codeparrot_training - Step 36166: {'lr': 9.566378594601408e-05, 'samples': 18517504, 'steps': 36166, 'loss/train': 2.162107467651367}
+02/26/2022 00:24:26 - INFO - codeparrot_training - Step 36167: {'lr': 9.56509140481716e-05, 'samples': 18518016, 'steps': 36167, 'loss/train': 1.370734453201294}
+02/26/2022 00:24:29 - INFO - codeparrot_training - Step 36168: {'lr': 9.563804281151164e-05, 'samples': 18518528, 'steps': 36168, 'loss/train': 2.5249743461608887}
+02/26/2022 00:24:35 - INFO - codeparrot_training - Step 36169: {'lr': 9.562517223608918e-05, 'samples': 18519040, 'steps': 36169, 'loss/train': 2.1326749324798584}
+02/26/2022 00:24:38 - INFO - codeparrot_training - Step 36170: {'lr': 9.561230232195959e-05, 'samples': 18519552, 'steps': 36170, 'loss/train': 2.1402714252471924}
+02/26/2022 00:24:44 - INFO - codeparrot_training - Step 36171: {'lr': 9.559943306917781e-05, 'samples': 18520064, 'steps': 36171, 'loss/train': 2.423326015472412}
+02/26/2022 00:24:47 - INFO - codeparrot_training - Step 36172: {'lr': 9.558656447779906e-05, 'samples': 18520576, 'steps': 36172, 'loss/train': 2.217533588409424}
+02/26/2022 00:24:54 - INFO - codeparrot_training - Step 36173: {'lr': 9.557369654787836e-05, 'samples': 18521088, 'steps': 36173, 'loss/train': 2.359816789627075}
+02/26/2022 00:24:57 - INFO - codeparrot_training - Step 36174: {'lr': 9.5560829279471e-05, 'samples': 18521600, 'steps': 36174, 'loss/train': 2.111583948135376}
+02/26/2022 00:25:03 - INFO - codeparrot_training - Step 36175: {'lr': 9.5547962672632e-05, 'samples': 18522112, 'steps': 36175, 'loss/train': 1.4911047220230103}
+02/26/2022 00:25:06 - INFO - codeparrot_training - Step 36176: {'lr': 9.553509672741645e-05, 'samples': 18522624, 'steps': 36176, 'loss/train': 2.1724941730499268}
+02/26/2022 00:25:12 - INFO - codeparrot_training - Step 36177: {'lr': 9.552223144387954e-05, 'samples': 18523136, 'steps': 36177, 'loss/train': 2.2457659244537354}
+02/26/2022 00:25:15 - INFO - codeparrot_training - Step 36178: {'lr': 9.550936682207622e-05, 'samples': 18523648, 'steps': 36178, 'loss/train': 1.2652034759521484}
+02/26/2022 00:25:21 - INFO - codeparrot_training - Step 36179: {'lr': 9.549650286206183e-05, 'samples': 18524160, 'steps': 36179, 'loss/train': 1.9991176128387451}
+02/26/2022 00:25:24 - INFO - codeparrot_training - Step 36180: {'lr': 9.548363956389133e-05, 'samples': 18524672, 'steps': 36180, 'loss/train': 1.9419400691986084}
+02/26/2022 00:25:31 - INFO - codeparrot_training - Step 36181: {'lr': 9.547077692761987e-05, 'samples': 18525184, 'steps': 36181, 'loss/train': 1.9898419380187988}
+02/26/2022 00:25:34 - INFO - codeparrot_training - Step 36182: {'lr': 9.545791495330247e-05, 'samples': 18525696, 'steps': 36182, 'loss/train': 0.9300118088722229}
+02/26/2022 00:25:40 - INFO - codeparrot_training - Step 36183: {'lr': 9.544505364099435e-05, 'samples': 18526208, 'steps': 36183, 'loss/train': 1.7209088802337646}
+02/26/2022 00:25:43 - INFO - codeparrot_training - Step 36184: {'lr': 9.543219299075057e-05, 'samples': 18526720, 'steps': 36184, 'loss/train': 1.4446091651916504}
+02/26/2022 00:25:49 - INFO - codeparrot_training - Step 36185: {'lr': 9.541933300262614e-05, 'samples': 18527232, 'steps': 36185, 'loss/train': 1.3330631256103516}
+02/26/2022 00:25:52 - INFO - codeparrot_training - Step 36186: {'lr': 9.540647367667618e-05, 'samples': 18527744, 'steps': 36186, 'loss/train': 1.9552959203720093}
+02/26/2022 00:25:58 - INFO - codeparrot_training - Step 36187: {'lr': 9.539361501295585e-05, 'samples': 18528256, 'steps': 36187, 'loss/train': 2.862837076187134}
+02/26/2022 00:26:01 - INFO - codeparrot_training - Step 36188: {'lr': 9.53807570115201e-05, 'samples': 18528768, 'steps': 36188, 'loss/train': 1.9843339920043945}
+02/26/2022 00:26:07 - INFO - codeparrot_training - Step 36189: {'lr': 9.536789967242426e-05, 'samples': 18529280, 'steps': 36189, 'loss/train': 2.0549697875976562}
+02/26/2022 00:26:10 - INFO - codeparrot_training - Step 36190: {'lr': 9.535504299572307e-05, 'samples': 18529792, 'steps': 36190, 'loss/train': 1.5688729286193848}
+02/26/2022 00:26:16 - INFO - codeparrot_training - Step 36191: {'lr': 9.534218698147185e-05, 'samples': 18530304, 'steps': 36191, 'loss/train': 0.041098542511463165}
+02/26/2022 00:26:20 - INFO - codeparrot_training - Step 36192: {'lr': 9.532933162972548e-05, 'samples': 18530816, 'steps': 36192, 'loss/train': 1.5200234651565552}
+02/26/2022 00:26:25 - INFO - codeparrot_training - Step 36193: {'lr': 9.531647694053936e-05, 'samples': 18531328, 'steps': 36193, 'loss/train': 1.0544005632400513}
+02/26/2022 00:26:28 - INFO - codeparrot_training - Step 36194: {'lr': 9.530362291396813e-05, 'samples': 18531840, 'steps': 36194, 'loss/train': 2.4101407527923584}
+02/26/2022 00:26:34 - INFO - codeparrot_training - Step 36195: {'lr': 9.529076955006716e-05, 'samples': 18532352, 'steps': 36195, 'loss/train': 1.412487506866455}
+02/26/2022 00:26:40 - INFO - codeparrot_training - Step 36196: {'lr': 9.527791684889131e-05, 'samples': 18532864, 'steps': 36196, 'loss/train': 1.501369595527649}
+02/26/2022 00:26:44 - INFO - codeparrot_training - Step 36197: {'lr': 9.526506481049594e-05, 'samples': 18533376, 'steps': 36197, 'loss/train': 2.132791757583618}
+02/26/2022 00:26:49 - INFO - codeparrot_training - Step 36198: {'lr': 9.525221343493568e-05, 'samples': 18533888, 'steps': 36198, 'loss/train': 0.909018874168396}
+02/26/2022 00:26:53 - INFO - codeparrot_training - Step 36199: {'lr': 9.52393627222659e-05, 'samples': 18534400, 'steps': 36199, 'loss/train': 1.8276982307434082}
+02/26/2022 00:26:58 - INFO - codeparrot_training - Step 36200: {'lr': 9.522651267254148e-05, 'samples': 18534912, 'steps': 36200, 'loss/train': 1.7221652269363403}
+02/26/2022 00:27:02 - INFO - codeparrot_training - Step 36201: {'lr': 9.52136632858176e-05, 'samples': 18535424, 'steps': 36201, 'loss/train': 0.8048366904258728}
+02/26/2022 00:27:07 - INFO - codeparrot_training - Step 36202: {'lr': 9.520081456214924e-05, 'samples': 18535936, 'steps': 36202, 'loss/train': 3.0710699558258057}
+02/26/2022 00:27:11 - INFO - codeparrot_training - Step 36203: {'lr': 9.518796650159146e-05, 'samples': 18536448, 'steps': 36203, 'loss/train': 1.6428643465042114}
+02/26/2022 00:27:16 - INFO - codeparrot_training - Step 36204: {'lr': 9.517511910419916e-05, 'samples': 18536960, 'steps': 36204, 'loss/train': 2.255718946456909}
+02/26/2022 00:27:20 - INFO - codeparrot_training - Step 36205: {'lr': 9.516227237002758e-05, 'samples': 18537472, 'steps': 36205, 'loss/train': 0.04485290125012398}
+02/26/2022 00:27:25 - INFO - codeparrot_training - Step 36206: {'lr': 9.514942629913165e-05, 'samples': 18537984, 'steps': 36206, 'loss/train': 1.8761175870895386}
+02/26/2022 00:27:28 - INFO - codeparrot_training - Step 36207: {'lr': 9.513658089156643e-05, 'samples': 18538496, 'steps': 36207, 'loss/train': 1.4736067056655884}
+02/26/2022 00:27:35 - INFO - codeparrot_training - Step 36208: {'lr': 9.512373614738681e-05, 'samples': 18539008, 'steps': 36208, 'loss/train': 1.3543826341629028}
+02/26/2022 00:27:38 - INFO - codeparrot_training - Step 36209: {'lr': 9.511089206664802e-05, 'samples': 18539520, 'steps': 36209, 'loss/train': 2.117323637008667}
+02/26/2022 00:27:44 - INFO - codeparrot_training - Step 36210: {'lr': 9.5098048649405e-05, 'samples': 18540032, 'steps': 36210, 'loss/train': 1.7350062131881714}
+02/26/2022 00:27:47 - INFO - codeparrot_training - Step 36211: {'lr': 9.508520589571274e-05, 'samples': 18540544, 'steps': 36211, 'loss/train': 1.5164942741394043}
+02/26/2022 00:27:53 - INFO - codeparrot_training - Step 36212: {'lr': 9.50723638056262e-05, 'samples': 18541056, 'steps': 36212, 'loss/train': 1.7740974426269531}
+02/26/2022 00:27:56 - INFO - codeparrot_training - Step 36213: {'lr': 9.505952237920051e-05, 'samples': 18541568, 'steps': 36213, 'loss/train': 1.9600930213928223}
+02/26/2022 00:28:02 - INFO - codeparrot_training - Step 36214: {'lr': 9.504668161649066e-05, 'samples': 18542080, 'steps': 36214, 'loss/train': 1.6778844594955444}
+02/26/2022 00:28:05 - INFO - codeparrot_training - Step 36215: {'lr': 9.503384151755159e-05, 'samples': 18542592, 'steps': 36215, 'loss/train': 1.3575268983840942}
+02/26/2022 00:28:11 - INFO - codeparrot_training - Step 36216: {'lr': 9.502100208243828e-05, 'samples': 18543104, 'steps': 36216, 'loss/train': 2.378061532974243}
+02/26/2022 00:28:15 - INFO - codeparrot_training - Step 36217: {'lr': 9.500816331120585e-05, 'samples': 18543616, 'steps': 36217, 'loss/train': 1.0308067798614502}
+02/26/2022 00:28:21 - INFO - codeparrot_training - Step 36218: {'lr': 9.499532520390925e-05, 'samples': 18544128, 'steps': 36218, 'loss/train': 1.7748454809188843}
+02/26/2022 00:28:24 - INFO - codeparrot_training - Step 36219: {'lr': 9.498248776060345e-05, 'samples': 18544640, 'steps': 36219, 'loss/train': 1.5875684022903442}
+02/26/2022 00:28:30 - INFO - codeparrot_training - Step 36220: {'lr': 9.496965098134335e-05, 'samples': 18545152, 'steps': 36220, 'loss/train': 1.957951545715332}
+02/26/2022 00:28:33 - INFO - codeparrot_training - Step 36221: {'lr': 9.495681486618415e-05, 'samples': 18545664, 'steps': 36221, 'loss/train': 1.8799099922180176}
+02/26/2022 00:28:39 - INFO - codeparrot_training - Step 36222: {'lr': 9.494397941518071e-05, 'samples': 18546176, 'steps': 36222, 'loss/train': 1.6717936992645264}
+02/26/2022 00:28:43 - INFO - codeparrot_training - Step 36223: {'lr': 9.493114462838795e-05, 'samples': 18546688, 'steps': 36223, 'loss/train': 2.255244731903076}
+02/26/2022 00:28:48 - INFO - codeparrot_training - Step 36224: {'lr': 9.491831050586108e-05, 'samples': 18547200, 'steps': 36224, 'loss/train': 2.0785927772521973}
+02/26/2022 00:28:52 - INFO - codeparrot_training - Step 36225: {'lr': 9.490547704765476e-05, 'samples': 18547712, 'steps': 36225, 'loss/train': 1.7624056339263916}
+02/26/2022 00:28:57 - INFO - codeparrot_training - Step 36226: {'lr': 9.489264425382421e-05, 'samples': 18548224, 'steps': 36226, 'loss/train': 0.7971744537353516}
+02/26/2022 00:29:00 - INFO - codeparrot_training - Step 36227: {'lr': 9.487981212442423e-05, 'samples': 18548736, 'steps': 36227, 'loss/train': 2.3726744651794434}
+02/26/2022 00:29:07 - INFO - codeparrot_training - Step 36228: {'lr': 9.486698065951008e-05, 'samples': 18549248, 'steps': 36228, 'loss/train': 1.748160481452942}
+02/26/2022 00:29:10 - INFO - codeparrot_training - Step 36229: {'lr': 9.485414985913631e-05, 'samples': 18549760, 'steps': 36229, 'loss/train': 1.1634185314178467}
+02/26/2022 00:29:16 - INFO - codeparrot_training - Step 36230: {'lr': 9.484131972335822e-05, 'samples': 18550272, 'steps': 36230, 'loss/train': 2.3919410705566406}
+02/26/2022 00:29:19 - INFO - codeparrot_training - Step 36231: {'lr': 9.482849025223053e-05, 'samples': 18550784, 'steps': 36231, 'loss/train': 1.634752631187439}
+02/26/2022 00:29:25 - INFO - codeparrot_training - Step 36232: {'lr': 9.481566144580853e-05, 'samples': 18551296, 'steps': 36232, 'loss/train': 1.9851620197296143}
+02/26/2022 00:29:28 - INFO - codeparrot_training - Step 36233: {'lr': 9.480283330414675e-05, 'samples': 18551808, 'steps': 36233, 'loss/train': 1.6267575025558472}
+02/26/2022 00:29:34 - INFO - codeparrot_training - Step 36234: {'lr': 9.479000582730044e-05, 'samples': 18552320, 'steps': 36234, 'loss/train': 1.9866682291030884}
+02/26/2022 00:29:37 - INFO - codeparrot_training - Step 36235: {'lr': 9.47771790153244e-05, 'samples': 18552832, 'steps': 36235, 'loss/train': 0.9661385416984558}
+02/26/2022 00:29:43 - INFO - codeparrot_training - Step 36236: {'lr': 9.476435286827371e-05, 'samples': 18553344, 'steps': 36236, 'loss/train': 1.5910093784332275}
+02/26/2022 00:29:47 - INFO - codeparrot_training - Step 36237: {'lr': 9.475152738620322e-05, 'samples': 18553856, 'steps': 36237, 'loss/train': 3.230879068374634}
+02/26/2022 00:29:52 - INFO - codeparrot_training - Step 36238: {'lr': 9.473870256916792e-05, 'samples': 18554368, 'steps': 36238, 'loss/train': 2.672098398208618}
+02/26/2022 00:29:56 - INFO - codeparrot_training - Step 36239: {'lr': 9.47258784172226e-05, 'samples': 18554880, 'steps': 36239, 'loss/train': 1.815095067024231}
+02/26/2022 00:30:01 - INFO - codeparrot_training - Step 36240: {'lr': 9.471305493042242e-05, 'samples': 18555392, 'steps': 36240, 'loss/train': 0.35066667199134827}
+02/26/2022 00:30:05 - INFO - codeparrot_training - Step 36241: {'lr': 9.47002321088222e-05, 'samples': 18555904, 'steps': 36241, 'loss/train': 0.9147576093673706}
+02/26/2022 00:30:10 - INFO - codeparrot_training - Step 36242: {'lr': 9.468740995247688e-05, 'samples': 18556416, 'steps': 36242, 'loss/train': 2.4108824729919434}
+02/26/2022 00:30:14 - INFO - codeparrot_training - Step 36243: {'lr': 9.467458846144129e-05, 'samples': 18556928, 'steps': 36243, 'loss/train': 1.6959593296051025}
+02/26/2022 00:30:20 - INFO - codeparrot_training - Step 36244: {'lr': 9.466176763577051e-05, 'samples': 18557440, 'steps': 36244, 'loss/train': 1.2872647047042847}
+02/26/2022 00:30:24 - INFO - codeparrot_training - Step 36245: {'lr': 9.46489474755194e-05, 'samples': 18557952, 'steps': 36245, 'loss/train': 2.546006202697754}
+02/26/2022 00:30:29 - INFO - codeparrot_training - Step 36246: {'lr': 9.463612798074286e-05, 'samples': 18558464, 'steps': 36246, 'loss/train': 2.254347324371338}
+02/26/2022 00:30:33 - INFO - codeparrot_training - Step 36247: {'lr': 9.462330915149575e-05, 'samples': 18558976, 'steps': 36247, 'loss/train': 1.3925974369049072}
+02/26/2022 00:30:38 - INFO - codeparrot_training - Step 36248: {'lr': 9.461049098783312e-05, 'samples': 18559488, 'steps': 36248, 'loss/train': 1.3173164129257202}
+02/26/2022 00:30:42 - INFO - codeparrot_training - Step 36249: {'lr': 9.459767348980971e-05, 'samples': 18560000, 'steps': 36249, 'loss/train': 1.2947278022766113}
+02/26/2022 00:30:47 - INFO - codeparrot_training - Step 36250: {'lr': 9.458485665748071e-05, 'samples': 18560512, 'steps': 36250, 'loss/train': 1.5315148830413818}
+02/26/2022 00:30:51 - INFO - codeparrot_training - Step 36251: {'lr': 9.457204049090065e-05, 'samples': 18561024, 'steps': 36251, 'loss/train': 1.5976450443267822}
+02/26/2022 00:30:56 - INFO - codeparrot_training - Step 36252: {'lr': 9.45592249901247e-05, 'samples': 18561536, 'steps': 36252, 'loss/train': 1.2676984071731567}
+02/26/2022 00:31:00 - INFO - codeparrot_training - Step 36253: {'lr': 9.45464101552076e-05, 'samples': 18562048, 'steps': 36253, 'loss/train': 2.419299602508545}
+02/26/2022 00:31:06 - INFO - codeparrot_training - Step 36254: {'lr': 9.453359598620448e-05, 'samples': 18562560, 'steps': 36254, 'loss/train': 0.5856847167015076}
+02/26/2022 00:31:09 - INFO - codeparrot_training - Step 36255: {'lr': 9.452078248316989e-05, 'samples': 18563072, 'steps': 36255, 'loss/train': 1.4574521780014038}
+02/26/2022 00:31:15 - INFO - codeparrot_training - Step 36256: {'lr': 9.450796964615902e-05, 'samples': 18563584, 'steps': 36256, 'loss/train': 1.208796501159668}
+02/26/2022 00:31:18 - INFO - codeparrot_training - Step 36257: {'lr': 9.449515747522658e-05, 'samples': 18564096, 'steps': 36257, 'loss/train': 2.0554802417755127}
+02/26/2022 00:31:24 - INFO - codeparrot_training - Step 36258: {'lr': 9.448234597042754e-05, 'samples': 18564608, 'steps': 36258, 'loss/train': 1.1945327520370483}
+02/26/2022 00:31:27 - INFO - codeparrot_training - Step 36259: {'lr': 9.446953513181666e-05, 'samples': 18565120, 'steps': 36259, 'loss/train': 2.031766414642334}
+02/26/2022 00:31:33 - INFO - codeparrot_training - Step 36260: {'lr': 9.445672495944899e-05, 'samples': 18565632, 'steps': 36260, 'loss/train': 1.1894862651824951}
+02/26/2022 00:31:36 - INFO - codeparrot_training - Step 36261: {'lr': 9.44439154533793e-05, 'samples': 18566144, 'steps': 36261, 'loss/train': 1.7685667276382446}
+02/26/2022 00:31:42 - INFO - codeparrot_training - Step 36262: {'lr': 9.443110661366242e-05, 'samples': 18566656, 'steps': 36262, 'loss/train': 2.4416990280151367}
+02/26/2022 00:31:45 - INFO - codeparrot_training - Step 36263: {'lr': 9.441829844035335e-05, 'samples': 18567168, 'steps': 36263, 'loss/train': 2.1631531715393066}
+02/26/2022 00:31:52 - INFO - codeparrot_training - Step 36264: {'lr': 9.440549093350689e-05, 'samples': 18567680, 'steps': 36264, 'loss/train': 1.252649188041687}
+02/26/2022 00:31:57 - INFO - codeparrot_training - Step 36265: {'lr': 9.439268409317791e-05, 'samples': 18568192, 'steps': 36265, 'loss/train': 2.0173165798187256}
+02/26/2022 00:32:01 - INFO - codeparrot_training - Step 36266: {'lr': 9.437987791942115e-05, 'samples': 18568704, 'steps': 36266, 'loss/train': 1.5092350244522095}
+02/26/2022 00:32:06 - INFO - codeparrot_training - Step 36267: {'lr': 9.43670724122917e-05, 'samples': 18569216, 'steps': 36267, 'loss/train': 0.8754414916038513}
+02/26/2022 00:32:10 - INFO - codeparrot_training - Step 36268: {'lr': 9.435426757184426e-05, 'samples': 18569728, 'steps': 36268, 'loss/train': 1.6685688495635986}
+02/26/2022 00:32:16 - INFO - codeparrot_training - Step 36269: {'lr': 9.434146339813373e-05, 'samples': 18570240, 'steps': 36269, 'loss/train': 0.416937917470932}
+02/26/2022 00:32:19 - INFO - codeparrot_training - Step 36270: {'lr': 9.432865989121487e-05, 'samples': 18570752, 'steps': 36270, 'loss/train': 2.1066877841949463}
+02/26/2022 00:32:25 - INFO - codeparrot_training - Step 36271: {'lr': 9.431585705114268e-05, 'samples': 18571264, 'steps': 36271, 'loss/train': 1.7855589389801025}
+02/26/2022 00:32:28 - INFO - codeparrot_training - Step 36272: {'lr': 9.430305487797191e-05, 'samples': 18571776, 'steps': 36272, 'loss/train': 1.211093544960022}
+02/26/2022 00:32:34 - INFO - codeparrot_training - Step 36273: {'lr': 9.429025337175742e-05, 'samples': 18572288, 'steps': 36273, 'loss/train': 1.1095423698425293}
+02/26/2022 00:32:38 - INFO - codeparrot_training - Step 36274: {'lr': 9.427745253255396e-05, 'samples': 18572800, 'steps': 36274, 'loss/train': 1.5531195402145386}
+02/26/2022 00:32:43 - INFO - codeparrot_training - Step 36275: {'lr': 9.42646523604165e-05, 'samples': 18573312, 'steps': 36275, 'loss/train': 2.3040332794189453}
+02/26/2022 00:32:47 - INFO - codeparrot_training - Step 36276: {'lr': 9.425185285539986e-05, 'samples': 18573824, 'steps': 36276, 'loss/train': 2.5518527030944824}
+02/26/2022 00:32:53 - INFO - codeparrot_training - Step 36277: {'lr': 9.423905401755881e-05, 'samples': 18574336, 'steps': 36277, 'loss/train': 1.2939387559890747}
+02/26/2022 00:32:56 - INFO - codeparrot_training - Step 36278: {'lr': 9.422625584694811e-05, 'samples': 18574848, 'steps': 36278, 'loss/train': 2.286909818649292}
+02/26/2022 00:33:00 - INFO - codeparrot_training - Step 36279: {'lr': 9.421345834362274e-05, 'samples': 18575360, 'steps': 36279, 'loss/train': 2.0417141914367676}
+02/26/2022 00:33:05 - INFO - codeparrot_training - Step 36280: {'lr': 9.420066150763748e-05, 'samples': 18575872, 'steps': 36280, 'loss/train': 2.0484418869018555}
+02/26/2022 00:33:09 - INFO - codeparrot_training - Step 36281: {'lr': 9.418786533904708e-05, 'samples': 18576384, 'steps': 36281, 'loss/train': 1.7631926536560059}
+02/26/2022 00:33:14 - INFO - codeparrot_training - Step 36282: {'lr': 9.41750698379063e-05, 'samples': 18576896, 'steps': 36282, 'loss/train': 1.8416720628738403}
+02/26/2022 00:33:18 - INFO - codeparrot_training - Step 36283: {'lr': 9.416227500427015e-05, 'samples': 18577408, 'steps': 36283, 'loss/train': 2.316648006439209}
+02/26/2022 00:33:23 - INFO - codeparrot_training - Step 36284: {'lr': 9.414948083819325e-05, 'samples': 18577920, 'steps': 36284, 'loss/train': 1.7622628211975098}
+02/26/2022 00:33:27 - INFO - codeparrot_training - Step 36285: {'lr': 9.413668733973063e-05, 'samples': 18578432, 'steps': 36285, 'loss/train': 2.356827735900879}
+02/26/2022 00:33:32 - INFO - codeparrot_training - Step 36286: {'lr': 9.412389450893679e-05, 'samples': 18578944, 'steps': 36286, 'loss/train': 1.0693385601043701}
+02/26/2022 00:33:36 - INFO - codeparrot_training - Step 36287: {'lr': 9.411110234586679e-05, 'samples': 18579456, 'steps': 36287, 'loss/train': 2.68483567237854}
+02/26/2022 00:33:41 - INFO - codeparrot_training - Step 36288: {'lr': 9.409831085057521e-05, 'samples': 18579968, 'steps': 36288, 'loss/train': 2.4817874431610107}
+02/26/2022 00:33:45 - INFO - codeparrot_training - Step 36289: {'lr': 9.408552002311716e-05, 'samples': 18580480, 'steps': 36289, 'loss/train': 0.7966066598892212}
+02/26/2022 00:33:51 - INFO - codeparrot_training - Step 36290: {'lr': 9.407272986354703e-05, 'samples': 18580992, 'steps': 36290, 'loss/train': 1.9825495481491089}
+02/26/2022 00:33:54 - INFO - codeparrot_training - Step 36291: {'lr': 9.405994037191996e-05, 'samples': 18581504, 'steps': 36291, 'loss/train': 0.9061518907546997}
+02/26/2022 00:34:00 - INFO - codeparrot_training - Step 36292: {'lr': 9.404715154829044e-05, 'samples': 18582016, 'steps': 36292, 'loss/train': 1.6406949758529663}
+02/26/2022 00:34:03 - INFO - codeparrot_training - Step 36293: {'lr': 9.40343633927136e-05, 'samples': 18582528, 'steps': 36293, 'loss/train': 2.3080708980560303}
+02/26/2022 00:34:09 - INFO - codeparrot_training - Step 36294: {'lr': 9.402157590524385e-05, 'samples': 18583040, 'steps': 36294, 'loss/train': 1.2550100088119507}
+02/26/2022 00:34:12 - INFO - codeparrot_training - Step 36295: {'lr': 9.400878908593621e-05, 'samples': 18583552, 'steps': 36295, 'loss/train': 2.0305874347686768}
+02/26/2022 00:34:18 - INFO - codeparrot_training - Step 36296: {'lr': 9.399600293484533e-05, 'samples': 18584064, 'steps': 36296, 'loss/train': 2.0343048572540283}
+02/26/2022 00:34:21 - INFO - codeparrot_training - Step 36297: {'lr': 9.39832174520261e-05, 'samples': 18584576, 'steps': 36297, 'loss/train': 3.303342819213867}
+02/26/2022 00:34:27 - INFO - codeparrot_training - Step 36298: {'lr': 9.397043263753324e-05, 'samples': 18585088, 'steps': 36298, 'loss/train': 1.1517834663391113}
+02/26/2022 00:34:30 - INFO - codeparrot_training - Step 36299: {'lr': 9.39576484914215e-05, 'samples': 18585600, 'steps': 36299, 'loss/train': 1.7336556911468506}
+02/26/2022 00:34:37 - INFO - codeparrot_training - Step 36300: {'lr': 9.394486501374555e-05, 'samples': 18586112, 'steps': 36300, 'loss/train': 2.379361867904663}
+02/26/2022 00:34:40 - INFO - codeparrot_training - Step 36301: {'lr': 9.393208220456032e-05, 'samples': 18586624, 'steps': 36301, 'loss/train': 1.6731102466583252}
+02/26/2022 00:34:46 - INFO - codeparrot_training - Step 36302: {'lr': 9.391930006392052e-05, 'samples': 18587136, 'steps': 36302, 'loss/train': 2.2060508728027344}
+02/26/2022 00:34:49 - INFO - codeparrot_training - Step 36303: {'lr': 9.390651859188084e-05, 'samples': 18587648, 'steps': 36303, 'loss/train': 2.2408595085144043}
+02/26/2022 00:34:55 - INFO - codeparrot_training - Step 36304: {'lr': 9.389373778849611e-05, 'samples': 18588160, 'steps': 36304, 'loss/train': 1.8166465759277344}
+02/26/2022 00:34:58 - INFO - codeparrot_training - Step 36305: {'lr': 9.388095765382094e-05, 'samples': 18588672, 'steps': 36305, 'loss/train': 1.9427070617675781}
+02/26/2022 00:35:04 - INFO - codeparrot_training - Step 36306: {'lr': 9.386817818791024e-05, 'samples': 18589184, 'steps': 36306, 'loss/train': 1.088804006576538}
+02/26/2022 00:35:07 - INFO - codeparrot_training - Step 36307: {'lr': 9.385539939081872e-05, 'samples': 18589696, 'steps': 36307, 'loss/train': 1.6397230625152588}
+02/26/2022 00:35:13 - INFO - codeparrot_training - Step 36308: {'lr': 9.384262126260107e-05, 'samples': 18590208, 'steps': 36308, 'loss/train': 2.6562230587005615}
+02/26/2022 00:35:16 - INFO - codeparrot_training - Step 36309: {'lr': 9.382984380331194e-05, 'samples': 18590720, 'steps': 36309, 'loss/train': 1.7649052143096924}
+02/26/2022 00:35:23 - INFO - codeparrot_training - Step 36310: {'lr': 9.381706701300627e-05, 'samples': 18591232, 'steps': 36310, 'loss/train': 0.11203889548778534}
+02/26/2022 00:35:26 - INFO - codeparrot_training - Step 36311: {'lr': 9.38042908917387e-05, 'samples': 18591744, 'steps': 36311, 'loss/train': 1.8558820486068726}
+02/26/2022 00:35:32 - INFO - codeparrot_training - Step 36312: {'lr': 9.379151543956397e-05, 'samples': 18592256, 'steps': 36312, 'loss/train': 2.5380797386169434}
+02/26/2022 00:35:35 - INFO - codeparrot_training - Step 36313: {'lr': 9.377874065653665e-05, 'samples': 18592768, 'steps': 36313, 'loss/train': 0.9079977869987488}
+02/26/2022 00:35:41 - INFO - codeparrot_training - Step 36314: {'lr': 9.376596654271172e-05, 'samples': 18593280, 'steps': 36314, 'loss/train': 2.2043566703796387}
+02/26/2022 00:35:45 - INFO - codeparrot_training - Step 36315: {'lr': 9.375319309814375e-05, 'samples': 18593792, 'steps': 36315, 'loss/train': 2.645874500274658}
+02/26/2022 00:35:50 - INFO - codeparrot_training - Step 36316: {'lr': 9.374042032288752e-05, 'samples': 18594304, 'steps': 36316, 'loss/train': 2.0600063800811768}
+02/26/2022 00:35:54 - INFO - codeparrot_training - Step 36317: {'lr': 9.372764821699761e-05, 'samples': 18594816, 'steps': 36317, 'loss/train': 1.090671420097351}
+02/26/2022 00:35:59 - INFO - codeparrot_training - Step 36318: {'lr': 9.371487678052892e-05, 'samples': 18595328, 'steps': 36318, 'loss/train': 0.9171093702316284}
+02/26/2022 00:36:03 - INFO - codeparrot_training - Step 36319: {'lr': 9.370210601353598e-05, 'samples': 18595840, 'steps': 36319, 'loss/train': 1.3418842554092407}
+02/26/2022 00:36:08 - INFO - codeparrot_training - Step 36320: {'lr': 9.368933591607378e-05, 'samples': 18596352, 'steps': 36320, 'loss/train': 3.4095373153686523}
+02/26/2022 00:36:12 - INFO - codeparrot_training - Step 36321: {'lr': 9.367656648819665e-05, 'samples': 18596864, 'steps': 36321, 'loss/train': 2.2872121334075928}
+02/26/2022 00:36:17 - INFO - codeparrot_training - Step 36322: {'lr': 9.366379772995954e-05, 'samples': 18597376, 'steps': 36322, 'loss/train': 2.281195640563965}
+02/26/2022 00:36:21 - INFO - codeparrot_training - Step 36323: {'lr': 9.365102964141701e-05, 'samples': 18597888, 'steps': 36323, 'loss/train': 1.2672016620635986}
+02/26/2022 00:36:26 - INFO - codeparrot_training - Step 36324: {'lr': 9.363826222262397e-05, 'samples': 18598400, 'steps': 36324, 'loss/train': 1.452994465827942}
+02/26/2022 00:36:30 - INFO - codeparrot_training - Step 36325: {'lr': 9.362549547363483e-05, 'samples': 18598912, 'steps': 36325, 'loss/train': 2.0354881286621094}
+02/26/2022 00:36:36 - INFO - codeparrot_training - Step 36326: {'lr': 9.36127293945045e-05, 'samples': 18599424, 'steps': 36326, 'loss/train': 1.5614584684371948}
+02/26/2022 00:36:39 - INFO - codeparrot_training - Step 36327: {'lr': 9.359996398528745e-05, 'samples': 18599936, 'steps': 36327, 'loss/train': 1.912429928779602}
+02/26/2022 00:36:45 - INFO - codeparrot_training - Step 36328: {'lr': 9.35871992460387e-05, 'samples': 18600448, 'steps': 36328, 'loss/train': 0.6544637680053711}
+02/26/2022 00:36:48 - INFO - codeparrot_training - Step 36329: {'lr': 9.357443517681252e-05, 'samples': 18600960, 'steps': 36329, 'loss/train': 2.163517951965332}
+02/26/2022 00:36:54 - INFO - codeparrot_training - Step 36330: {'lr': 9.356167177766389e-05, 'samples': 18601472, 'steps': 36330, 'loss/train': 1.611802577972412}
+02/26/2022 00:36:57 - INFO - codeparrot_training - Step 36331: {'lr': 9.354890904864729e-05, 'samples': 18601984, 'steps': 36331, 'loss/train': 1.443282127380371}
+02/26/2022 00:37:03 - INFO - codeparrot_training - Step 36332: {'lr': 9.353614698981761e-05, 'samples': 18602496, 'steps': 36332, 'loss/train': 1.5898233652114868}
+02/26/2022 00:37:06 - INFO - codeparrot_training - Step 36333: {'lr': 9.352338560122934e-05, 'samples': 18603008, 'steps': 36333, 'loss/train': 0.5311668515205383}
+02/26/2022 00:37:12 - INFO - codeparrot_training - Step 36334: {'lr': 9.351062488293724e-05, 'samples': 18603520, 'steps': 36334, 'loss/train': 1.609014868736267}
+02/26/2022 00:37:16 - INFO - codeparrot_training - Step 36335: {'lr': 9.349786483499582e-05, 'samples': 18604032, 'steps': 36335, 'loss/train': 2.0934019088745117}
+02/26/2022 00:37:22 - INFO - codeparrot_training - Step 36336: {'lr': 9.348510545745995e-05, 'samples': 18604544, 'steps': 36336, 'loss/train': 1.9578651189804077}
+02/26/2022 00:37:25 - INFO - codeparrot_training - Step 36337: {'lr': 9.347234675038419e-05, 'samples': 18605056, 'steps': 36337, 'loss/train': 1.813740849494934}
+02/26/2022 00:37:31 - INFO - codeparrot_training - Step 36338: {'lr': 9.345958871382318e-05, 'samples': 18605568, 'steps': 36338, 'loss/train': 1.6773390769958496}
+02/26/2022 00:37:34 - INFO - codeparrot_training - Step 36339: {'lr': 9.344683134783149e-05, 'samples': 18606080, 'steps': 36339, 'loss/train': 0.47529885172843933}
+02/26/2022 00:37:40 - INFO - codeparrot_training - Step 36340: {'lr': 9.343407465246398e-05, 'samples': 18606592, 'steps': 36340, 'loss/train': 1.8597060441970825}
+02/26/2022 00:37:43 - INFO - codeparrot_training - Step 36341: {'lr': 9.342131862777516e-05, 'samples': 18607104, 'steps': 36341, 'loss/train': 2.439831256866455}
+02/26/2022 00:37:49 - INFO - codeparrot_training - Step 36342: {'lr': 9.340856327381969e-05, 'samples': 18607616, 'steps': 36342, 'loss/train': 1.724426507949829}
+02/26/2022 00:37:52 - INFO - codeparrot_training - Step 36343: {'lr': 9.339580859065214e-05, 'samples': 18608128, 'steps': 36343, 'loss/train': 1.9011871814727783}
+02/26/2022 00:37:58 - INFO - codeparrot_training - Step 36344: {'lr': 9.33830545783273e-05, 'samples': 18608640, 'steps': 36344, 'loss/train': 1.2725675106048584}
+02/26/2022 00:38:01 - INFO - codeparrot_training - Step 36345: {'lr': 9.337030123689972e-05, 'samples': 18609152, 'steps': 36345, 'loss/train': 1.6196274757385254}
+02/26/2022 00:38:07 - INFO - codeparrot_training - Step 36346: {'lr': 9.335754856642404e-05, 'samples': 18609664, 'steps': 36346, 'loss/train': 2.038508892059326}
+02/26/2022 00:38:11 - INFO - codeparrot_training - Step 36347: {'lr': 9.334479656695476e-05, 'samples': 18610176, 'steps': 36347, 'loss/train': 1.5932917594909668}
+02/26/2022 00:38:16 - INFO - codeparrot_training - Step 36348: {'lr': 9.333204523854678e-05, 'samples': 18610688, 'steps': 36348, 'loss/train': 1.1246910095214844}
+02/26/2022 00:38:20 - INFO - codeparrot_training - Step 36349: {'lr': 9.331929458125451e-05, 'samples': 18611200, 'steps': 36349, 'loss/train': 2.1565911769866943}
+02/26/2022 00:38:25 - INFO - codeparrot_training - Step 36350: {'lr': 9.330654459513265e-05, 'samples': 18611712, 'steps': 36350, 'loss/train': 1.7111234664916992}
+02/26/2022 00:38:31 - INFO - codeparrot_training - Step 36351: {'lr': 9.329379528023574e-05, 'samples': 18612224, 'steps': 36351, 'loss/train': 1.4370098114013672}
+02/26/2022 00:38:34 - INFO - codeparrot_training - Step 36352: {'lr': 9.328104663661852e-05, 'samples': 18612736, 'steps': 36352, 'loss/train': 1.0549601316452026}
+02/26/2022 00:38:40 - INFO - codeparrot_training - Step 36353: {'lr': 9.326829866433551e-05, 'samples': 18613248, 'steps': 36353, 'loss/train': 0.15012721717357635}
+02/26/2022 00:38:43 - INFO - codeparrot_training - Step 36354: {'lr': 9.325555136344135e-05, 'samples': 18613760, 'steps': 36354, 'loss/train': 0.8335407972335815}
+02/26/2022 00:38:50 - INFO - codeparrot_training - Step 36355: {'lr': 9.324280473399067e-05, 'samples': 18614272, 'steps': 36355, 'loss/train': 2.3158681392669678}
+02/26/2022 00:38:53 - INFO - codeparrot_training - Step 36356: {'lr': 9.323005877603791e-05, 'samples': 18614784, 'steps': 36356, 'loss/train': 0.5626108050346375}
+02/26/2022 00:38:59 - INFO - codeparrot_training - Step 36357: {'lr': 9.321731348963788e-05, 'samples': 18615296, 'steps': 36357, 'loss/train': 1.7125240564346313}
+02/26/2022 00:39:02 - INFO - codeparrot_training - Step 36358: {'lr': 9.320456887484504e-05, 'samples': 18615808, 'steps': 36358, 'loss/train': 0.712852418422699}
+02/26/2022 00:39:08 - INFO - codeparrot_training - Step 36359: {'lr': 9.319182493171419e-05, 'samples': 18616320, 'steps': 36359, 'loss/train': 1.7100962400436401}
+02/26/2022 00:39:11 - INFO - codeparrot_training - Step 36360: {'lr': 9.317908166029961e-05, 'samples': 18616832, 'steps': 36360, 'loss/train': 1.8680890798568726}
+02/26/2022 00:39:17 - INFO - codeparrot_training - Step 36361: {'lr': 9.316633906065613e-05, 'samples': 18617344, 'steps': 36361, 'loss/train': 0.40120282769203186}
+02/26/2022 00:39:20 - INFO - codeparrot_training - Step 36362: {'lr': 9.315359713283817e-05, 'samples': 18617856, 'steps': 36362, 'loss/train': 2.3107306957244873}
+02/26/2022 00:39:26 - INFO - codeparrot_training - Step 36363: {'lr': 9.314085587690058e-05, 'samples': 18618368, 'steps': 36363, 'loss/train': 1.7699942588806152}
+02/26/2022 00:39:29 - INFO - codeparrot_training - Step 36364: {'lr': 9.312811529289755e-05, 'samples': 18618880, 'steps': 36364, 'loss/train': 2.184447765350342}
+02/26/2022 00:39:35 - INFO - codeparrot_training - Step 36365: {'lr': 9.311537538088396e-05, 'samples': 18619392, 'steps': 36365, 'loss/train': 2.7705628871917725}
+02/26/2022 00:39:38 - INFO - codeparrot_training - Step 36366: {'lr': 9.310263614091421e-05, 'samples': 18619904, 'steps': 36366, 'loss/train': 1.8837342262268066}
+02/26/2022 00:39:44 - INFO - codeparrot_training - Step 36367: {'lr': 9.308989757304303e-05, 'samples': 18620416, 'steps': 36367, 'loss/train': 0.6485057473182678}
+02/26/2022 00:39:47 - INFO - codeparrot_training - Step 36368: {'lr': 9.307715967732491e-05, 'samples': 18620928, 'steps': 36368, 'loss/train': 1.4307818412780762}
+02/26/2022 00:39:53 - INFO - codeparrot_training - Step 36369: {'lr': 9.306442245381439e-05, 'samples': 18621440, 'steps': 36369, 'loss/train': 2.0255372524261475}
+02/26/2022 00:39:56 - INFO - codeparrot_training - Step 36370: {'lr': 9.305168590256599e-05, 'samples': 18621952, 'steps': 36370, 'loss/train': 2.8514492511749268}
+02/26/2022 00:40:03 - INFO - codeparrot_training - Step 36371: {'lr': 9.303895002363439e-05, 'samples': 18622464, 'steps': 36371, 'loss/train': 2.1975536346435547}
+02/26/2022 00:40:06 - INFO - codeparrot_training - Step 36372: {'lr': 9.30262148170741e-05, 'samples': 18622976, 'steps': 36372, 'loss/train': 1.1395689249038696}
+02/26/2022 00:40:12 - INFO - codeparrot_training - Step 36373: {'lr': 9.301348028293965e-05, 'samples': 18623488, 'steps': 36373, 'loss/train': 0.3569713234901428}
+02/26/2022 00:40:15 - INFO - codeparrot_training - Step 36374: {'lr': 9.300074642128554e-05, 'samples': 18624000, 'steps': 36374, 'loss/train': 1.5336700677871704}
+02/26/2022 00:40:21 - INFO - codeparrot_training - Step 36375: {'lr': 9.298801323216646e-05, 'samples': 18624512, 'steps': 36375, 'loss/train': 2.599266290664673}
+02/26/2022 00:40:24 - INFO - codeparrot_training - Step 36376: {'lr': 9.297528071563685e-05, 'samples': 18625024, 'steps': 36376, 'loss/train': 1.296905755996704}
+02/26/2022 00:40:30 - INFO - codeparrot_training - Step 36377: {'lr': 9.296254887175132e-05, 'samples': 18625536, 'steps': 36377, 'loss/train': 1.8133147954940796}
+02/26/2022 00:40:33 - INFO - codeparrot_training - Step 36378: {'lr': 9.294981770056424e-05, 'samples': 18626048, 'steps': 36378, 'loss/train': 2.8033108711242676}
+02/26/2022 00:40:39 - INFO - codeparrot_training - Step 36379: {'lr': 9.293708720213037e-05, 'samples': 18626560, 'steps': 36379, 'loss/train': 0.46975770592689514}
+02/26/2022 00:40:42 - INFO - codeparrot_training - Step 36380: {'lr': 9.292435737650406e-05, 'samples': 18627072, 'steps': 36380, 'loss/train': 1.9406933784484863}
+02/26/2022 00:40:48 - INFO - codeparrot_training - Step 36381: {'lr': 9.291162822374011e-05, 'samples': 18627584, 'steps': 36381, 'loss/train': 2.463886260986328}
+02/26/2022 00:40:52 - INFO - codeparrot_training - Step 36382: {'lr': 9.289889974389268e-05, 'samples': 18628096, 'steps': 36382, 'loss/train': 1.7609314918518066}
+02/26/2022 00:40:57 - INFO - codeparrot_training - Step 36383: {'lr': 9.288617193701654e-05, 'samples': 18628608, 'steps': 36383, 'loss/train': 2.3828985691070557}
+02/26/2022 00:41:01 - INFO - codeparrot_training - Step 36384: {'lr': 9.287344480316617e-05, 'samples': 18629120, 'steps': 36384, 'loss/train': 1.5279086828231812}
+02/26/2022 00:41:06 - INFO - codeparrot_training - Step 36385: {'lr': 9.28607183423961e-05, 'samples': 18629632, 'steps': 36385, 'loss/train': 2.533825159072876}
+02/26/2022 00:41:10 - INFO - codeparrot_training - Step 36386: {'lr': 9.28479925547607e-05, 'samples': 18630144, 'steps': 36386, 'loss/train': 0.9208022356033325}
+02/26/2022 00:41:16 - INFO - codeparrot_training - Step 36387: {'lr': 9.283526744031467e-05, 'samples': 18630656, 'steps': 36387, 'loss/train': 1.663558840751648}
+02/26/2022 00:41:19 - INFO - codeparrot_training - Step 36388: {'lr': 9.282254299911247e-05, 'samples': 18631168, 'steps': 36388, 'loss/train': 3.1090292930603027}
+02/26/2022 00:41:25 - INFO - codeparrot_training - Step 36389: {'lr': 9.28098192312086e-05, 'samples': 18631680, 'steps': 36389, 'loss/train': 1.2936574220657349}
+02/26/2022 00:41:28 - INFO - codeparrot_training - Step 36390: {'lr': 9.279709613665743e-05, 'samples': 18632192, 'steps': 36390, 'loss/train': 1.8255300521850586}
+02/26/2022 00:41:34 - INFO - codeparrot_training - Step 36391: {'lr': 9.278437371551368e-05, 'samples': 18632704, 'steps': 36391, 'loss/train': 1.5036951303482056}
+02/26/2022 00:41:38 - INFO - codeparrot_training - Step 36392: {'lr': 9.277165196783177e-05, 'samples': 18633216, 'steps': 36392, 'loss/train': 2.361042022705078}
+02/26/2022 00:41:43 - INFO - codeparrot_training - Step 36393: {'lr': 9.275893089366607e-05, 'samples': 18633728, 'steps': 36393, 'loss/train': 1.6003185510635376}
+02/26/2022 00:41:47 - INFO - codeparrot_training - Step 36394: {'lr': 9.274621049307128e-05, 'samples': 18634240, 'steps': 36394, 'loss/train': 1.0404809713363647}
+02/26/2022 00:41:52 - INFO - codeparrot_training - Step 36395: {'lr': 9.273349076610177e-05, 'samples': 18634752, 'steps': 36395, 'loss/train': 2.1912903785705566}
+02/26/2022 00:41:55 - INFO - codeparrot_training - Step 36396: {'lr': 9.272077171281207e-05, 'samples': 18635264, 'steps': 36396, 'loss/train': 0.31599026918411255}
+02/26/2022 00:42:01 - INFO - codeparrot_training - Step 36397: {'lr': 9.270805333325655e-05, 'samples': 18635776, 'steps': 36397, 'loss/train': 0.7981106042861938}
+02/26/2022 00:42:04 - INFO - codeparrot_training - Step 36398: {'lr': 9.269533562748989e-05, 'samples': 18636288, 'steps': 36398, 'loss/train': 1.8272963762283325}
+02/26/2022 00:42:10 - INFO - codeparrot_training - Step 36399: {'lr': 9.268261859556643e-05, 'samples': 18636800, 'steps': 36399, 'loss/train': 1.8302654027938843}
+02/26/2022 00:42:13 - INFO - codeparrot_training - Step 36400: {'lr': 9.266990223754068e-05, 'samples': 18637312, 'steps': 36400, 'loss/train': 1.2698239088058472}
+02/26/2022 00:42:19 - INFO - codeparrot_training - Step 36401: {'lr': 9.265718655346703e-05, 'samples': 18637824, 'steps': 36401, 'loss/train': 0.3177264630794525}
+02/26/2022 00:42:22 - INFO - codeparrot_training - Step 36402: {'lr': 9.264447154340022e-05, 'samples': 18638336, 'steps': 36402, 'loss/train': 1.8804055452346802}
+02/26/2022 00:42:29 - INFO - codeparrot_training - Step 36403: {'lr': 9.263175720739434e-05, 'samples': 18638848, 'steps': 36403, 'loss/train': 2.492642641067505}
+02/26/2022 00:42:32 - INFO - codeparrot_training - Step 36404: {'lr': 9.261904354550413e-05, 'samples': 18639360, 'steps': 36404, 'loss/train': 1.6978983879089355}
+02/26/2022 00:42:38 - INFO - codeparrot_training - Step 36405: {'lr': 9.260633055778389e-05, 'samples': 18639872, 'steps': 36405, 'loss/train': 1.8529647588729858}
+02/26/2022 00:42:41 - INFO - codeparrot_training - Step 36406: {'lr': 9.259361824428822e-05, 'samples': 18640384, 'steps': 36406, 'loss/train': 1.9268027544021606}
+02/26/2022 00:42:46 - INFO - codeparrot_training - Step 36407: {'lr': 9.258090660507152e-05, 'samples': 18640896, 'steps': 36407, 'loss/train': 1.407084584236145}
+02/26/2022 00:42:50 - INFO - codeparrot_training - Step 36408: {'lr': 9.25681956401882e-05, 'samples': 18641408, 'steps': 36408, 'loss/train': 1.5694947242736816}
+02/26/2022 00:42:55 - INFO - codeparrot_training - Step 36409: {'lr': 9.255548534969268e-05, 'samples': 18641920, 'steps': 36409, 'loss/train': 1.7173643112182617}
+02/26/2022 00:42:59 - INFO - codeparrot_training - Step 36410: {'lr': 9.254277573363957e-05, 'samples': 18642432, 'steps': 36410, 'loss/train': 1.9942305088043213}
+02/26/2022 00:43:04 - INFO - codeparrot_training - Step 36411: {'lr': 9.253006679208317e-05, 'samples': 18642944, 'steps': 36411, 'loss/train': 1.5869961977005005}
+02/26/2022 00:43:08 - INFO - codeparrot_training - Step 36412: {'lr': 9.251735852507798e-05, 'samples': 18643456, 'steps': 36412, 'loss/train': 1.273870587348938}
+02/26/2022 00:43:13 - INFO - codeparrot_training - Step 36413: {'lr': 9.250465093267834e-05, 'samples': 18643968, 'steps': 36413, 'loss/train': 1.7111687660217285}
+02/26/2022 00:43:19 - INFO - codeparrot_training - Step 36414: {'lr': 9.249194401493884e-05, 'samples': 18644480, 'steps': 36414, 'loss/train': 1.0950794219970703}
+02/26/2022 00:43:22 - INFO - codeparrot_training - Step 36415: {'lr': 9.247923777191386e-05, 'samples': 18644992, 'steps': 36415, 'loss/train': 1.7551685571670532}
+02/26/2022 00:43:29 - INFO - codeparrot_training - Step 36416: {'lr': 9.246653220365778e-05, 'samples': 18645504, 'steps': 36416, 'loss/train': 1.7830021381378174}
+02/26/2022 00:43:32 - INFO - codeparrot_training - Step 36417: {'lr': 9.245382731022497e-05, 'samples': 18646016, 'steps': 36417, 'loss/train': 1.981549859046936}
+02/26/2022 00:43:38 - INFO - codeparrot_training - Step 36418: {'lr': 9.244112309167005e-05, 'samples': 18646528, 'steps': 36418, 'loss/train': 1.9376245737075806}
+02/26/2022 00:43:41 - INFO - codeparrot_training - Step 36419: {'lr': 9.24284195480472e-05, 'samples': 18647040, 'steps': 36419, 'loss/train': 5.026699542999268}
+02/26/2022 00:43:47 - INFO - codeparrot_training - Step 36420: {'lr': 9.241571667941117e-05, 'samples': 18647552, 'steps': 36420, 'loss/train': 1.808791995048523}
+02/26/2022 00:43:50 - INFO - codeparrot_training - Step 36421: {'lr': 9.2403014485816e-05, 'samples': 18648064, 'steps': 36421, 'loss/train': 0.2704356610774994}
+02/26/2022 00:43:56 - INFO - codeparrot_training - Step 36422: {'lr': 9.239031296731634e-05, 'samples': 18648576, 'steps': 36422, 'loss/train': 1.584717035293579}
+02/26/2022 00:43:59 - INFO - codeparrot_training - Step 36423: {'lr': 9.237761212396648e-05, 'samples': 18649088, 'steps': 36423, 'loss/train': 1.7895535230636597}
+02/26/2022 00:44:05 - INFO - codeparrot_training - Step 36424: {'lr': 9.2364911955821e-05, 'samples': 18649600, 'steps': 36424, 'loss/train': 1.4520173072814941}
+02/26/2022 00:44:08 - INFO - codeparrot_training - Step 36425: {'lr': 9.235221246293405e-05, 'samples': 18650112, 'steps': 36425, 'loss/train': 1.1449350118637085}
+02/26/2022 00:44:14 - INFO - codeparrot_training - Step 36426: {'lr': 9.233951364536025e-05, 'samples': 18650624, 'steps': 36426, 'loss/train': 1.5596548318862915}
+02/26/2022 00:44:18 - INFO - codeparrot_training - Step 36427: {'lr': 9.23268155031538e-05, 'samples': 18651136, 'steps': 36427, 'loss/train': 2.1618642807006836}
+02/26/2022 00:44:23 - INFO - codeparrot_training - Step 36428: {'lr': 9.23141180363693e-05, 'samples': 18651648, 'steps': 36428, 'loss/train': 0.4900175631046295}
+02/26/2022 00:44:27 - INFO - codeparrot_training - Step 36429: {'lr': 9.230142124506105e-05, 'samples': 18652160, 'steps': 36429, 'loss/train': 1.1107112169265747}
+02/26/2022 00:44:32 - INFO - codeparrot_training - Step 36430: {'lr': 9.228872512928344e-05, 'samples': 18652672, 'steps': 36430, 'loss/train': 1.955566644668579}
+02/26/2022 00:44:36 - INFO - codeparrot_training - Step 36431: {'lr': 9.227602968909077e-05, 'samples': 18653184, 'steps': 36431, 'loss/train': 0.5552517175674438}
+02/26/2022 00:44:41 - INFO - codeparrot_training - Step 36432: {'lr': 9.226333492453759e-05, 'samples': 18653696, 'steps': 36432, 'loss/train': 2.1072685718536377}
+02/26/2022 00:44:45 - INFO - codeparrot_training - Step 36433: {'lr': 9.225064083567819e-05, 'samples': 18654208, 'steps': 36433, 'loss/train': 1.8738425970077515}
+02/26/2022 00:44:50 - INFO - codeparrot_training - Step 36434: {'lr': 9.223794742256694e-05, 'samples': 18654720, 'steps': 36434, 'loss/train': 3.0810461044311523}
+02/26/2022 00:44:54 - INFO - codeparrot_training - Step 36435: {'lr': 9.222525468525825e-05, 'samples': 18655232, 'steps': 36435, 'loss/train': 1.5632967948913574}
+02/26/2022 00:44:59 - INFO - codeparrot_training - Step 36436: {'lr': 9.221256262380637e-05, 'samples': 18655744, 'steps': 36436, 'loss/train': 8.585833549499512}
+02/26/2022 00:45:03 - INFO - codeparrot_training - Step 36437: {'lr': 9.219987123826587e-05, 'samples': 18656256, 'steps': 36437, 'loss/train': 2.015568733215332}
+02/26/2022 00:45:09 - INFO - codeparrot_training - Step 36438: {'lr': 9.218718052869099e-05, 'samples': 18656768, 'steps': 36438, 'loss/train': 1.0608787536621094}
+02/26/2022 00:45:12 - INFO - codeparrot_training - Step 36439: {'lr': 9.217449049513615e-05, 'samples': 18657280, 'steps': 36439, 'loss/train': 2.932211399078369}
+02/26/2022 00:45:18 - INFO - codeparrot_training - Step 36440: {'lr': 9.216180113765556e-05, 'samples': 18657792, 'steps': 36440, 'loss/train': 1.40361487865448}
+02/26/2022 00:45:21 - INFO - codeparrot_training - Step 36441: {'lr': 9.21491124563038e-05, 'samples': 18658304, 'steps': 36441, 'loss/train': 2.0718863010406494}
+02/26/2022 00:45:27 - INFO - codeparrot_training - Step 36442: {'lr': 9.213642445113513e-05, 'samples': 18658816, 'steps': 36442, 'loss/train': 2.466803789138794}
+02/26/2022 00:45:30 - INFO - codeparrot_training - Step 36443: {'lr': 9.212373712220388e-05, 'samples': 18659328, 'steps': 36443, 'loss/train': 1.2844488620758057}
+02/26/2022 00:45:36 - INFO - codeparrot_training - Step 36444: {'lr': 9.21110504695643e-05, 'samples': 18659840, 'steps': 36444, 'loss/train': 1.6582192182540894}
+02/26/2022 00:45:39 - INFO - codeparrot_training - Step 36445: {'lr': 9.209836449327095e-05, 'samples': 18660352, 'steps': 36445, 'loss/train': 2.2200734615325928}
+02/26/2022 00:45:45 - INFO - codeparrot_training - Step 36446: {'lr': 9.208567919337806e-05, 'samples': 18660864, 'steps': 36446, 'loss/train': 1.795745611190796}
+02/26/2022 00:45:48 - INFO - codeparrot_training - Step 36447: {'lr': 9.207299456993998e-05, 'samples': 18661376, 'steps': 36447, 'loss/train': 1.5260186195373535}
+02/26/2022 00:45:55 - INFO - codeparrot_training - Step 36448: {'lr': 9.206031062301095e-05, 'samples': 18661888, 'steps': 36448, 'loss/train': 1.56020987033844}
+02/26/2022 00:45:58 - INFO - codeparrot_training - Step 36449: {'lr': 9.204762735264552e-05, 'samples': 18662400, 'steps': 36449, 'loss/train': 1.3249993324279785}
+02/26/2022 00:46:04 - INFO - codeparrot_training - Step 36450: {'lr': 9.203494475889787e-05, 'samples': 18662912, 'steps': 36450, 'loss/train': 0.8002824187278748}
+02/26/2022 00:46:07 - INFO - codeparrot_training - Step 36451: {'lr': 9.202226284182236e-05, 'samples': 18663424, 'steps': 36451, 'loss/train': 1.8464992046356201}
+02/26/2022 00:46:13 - INFO - codeparrot_training - Step 36452: {'lr': 9.200958160147322e-05, 'samples': 18663936, 'steps': 36452, 'loss/train': 1.8326163291931152}
+02/26/2022 00:46:16 - INFO - codeparrot_training - Step 36453: {'lr': 9.199690103790495e-05, 'samples': 18664448, 'steps': 36453, 'loss/train': 2.0581140518188477}
+02/26/2022 00:46:22 - INFO - codeparrot_training - Step 36454: {'lr': 9.19842211511717e-05, 'samples': 18664960, 'steps': 36454, 'loss/train': 1.4904792308807373}
+02/26/2022 00:46:25 - INFO - codeparrot_training - Step 36455: {'lr': 9.197154194132807e-05, 'samples': 18665472, 'steps': 36455, 'loss/train': 2.4088711738586426}
+02/26/2022 00:46:31 - INFO - codeparrot_training - Step 36456: {'lr': 9.195886340842797e-05, 'samples': 18665984, 'steps': 36456, 'loss/train': 1.2211743593215942}
+02/26/2022 00:46:34 - INFO - codeparrot_training - Step 36457: {'lr': 9.194618555252601e-05, 'samples': 18666496, 'steps': 36457, 'loss/train': 1.9535833597183228}
+02/26/2022 00:46:40 - INFO - codeparrot_training - Step 36458: {'lr': 9.193350837367631e-05, 'samples': 18667008, 'steps': 36458, 'loss/train': 2.4182329177856445}
+02/26/2022 00:46:43 - INFO - codeparrot_training - Step 36459: {'lr': 9.192083187193345e-05, 'samples': 18667520, 'steps': 36459, 'loss/train': 2.333261013031006}
+02/26/2022 00:46:49 - INFO - codeparrot_training - Step 36460: {'lr': 9.190815604735139e-05, 'samples': 18668032, 'steps': 36460, 'loss/train': 1.7774379253387451}
+02/26/2022 00:46:52 - INFO - codeparrot_training - Step 36461: {'lr': 9.189548089998464e-05, 'samples': 18668544, 'steps': 36461, 'loss/train': 0.9777742028236389}
+02/26/2022 00:46:58 - INFO - codeparrot_training - Step 36462: {'lr': 9.188280642988738e-05, 'samples': 18669056, 'steps': 36462, 'loss/train': 2.45379900932312}
+02/26/2022 00:47:01 - INFO - codeparrot_training - Step 36463: {'lr': 9.187013263711417e-05, 'samples': 18669568, 'steps': 36463, 'loss/train': 2.6046814918518066}
+02/26/2022 00:47:08 - INFO - codeparrot_training - Step 36464: {'lr': 9.185745952171889e-05, 'samples': 18670080, 'steps': 36464, 'loss/train': 1.002368688583374}
+02/26/2022 00:47:11 - INFO - codeparrot_training - Step 36465: {'lr': 9.184478708375616e-05, 'samples': 18670592, 'steps': 36465, 'loss/train': 2.8651559352874756}
+02/26/2022 00:47:17 - INFO - codeparrot_training - Step 36466: {'lr': 9.183211532328e-05, 'samples': 18671104, 'steps': 36466, 'loss/train': 1.9365592002868652}
+02/26/2022 00:47:20 - INFO - codeparrot_training - Step 36467: {'lr': 9.181944424034497e-05, 'samples': 18671616, 'steps': 36467, 'loss/train': 1.973009467124939}
+02/26/2022 00:47:26 - INFO - codeparrot_training - Step 36468: {'lr': 9.180677383500518e-05, 'samples': 18672128, 'steps': 36468, 'loss/train': 1.2666012048721313}
+02/26/2022 00:47:31 - INFO - codeparrot_training - Step 36469: {'lr': 9.179410410731498e-05, 'samples': 18672640, 'steps': 36469, 'loss/train': 0.8861597180366516}
+02/26/2022 00:47:35 - INFO - codeparrot_training - Step 36470: {'lr': 9.178143505732847e-05, 'samples': 18673152, 'steps': 36470, 'loss/train': 1.7103712558746338}
+02/26/2022 00:47:40 - INFO - codeparrot_training - Step 36471: {'lr': 9.176876668510018e-05, 'samples': 18673664, 'steps': 36471, 'loss/train': 1.3528631925582886}
+02/26/2022 00:47:44 - INFO - codeparrot_training - Step 36472: {'lr': 9.175609899068421e-05, 'samples': 18674176, 'steps': 36472, 'loss/train': 2.02816104888916}
+02/26/2022 00:47:50 - INFO - codeparrot_training - Step 36473: {'lr': 9.17434319741349e-05, 'samples': 18674688, 'steps': 36473, 'loss/train': 0.8615384697914124}
+02/26/2022 00:47:53 - INFO - codeparrot_training - Step 36474: {'lr': 9.173076563550636e-05, 'samples': 18675200, 'steps': 36474, 'loss/train': 2.8255438804626465}
+02/26/2022 00:47:59 - INFO - codeparrot_training - Step 36475: {'lr': 9.171809997485305e-05, 'samples': 18675712, 'steps': 36475, 'loss/train': 2.03831148147583}
+02/26/2022 00:48:02 - INFO - codeparrot_training - Step 36476: {'lr': 9.170543499222917e-05, 'samples': 18676224, 'steps': 36476, 'loss/train': 1.6320687532424927}
+02/26/2022 00:48:08 - INFO - codeparrot_training - Step 36477: {'lr': 9.169277068768891e-05, 'samples': 18676736, 'steps': 36477, 'loss/train': 1.1638494729995728}
+02/26/2022 00:48:11 - INFO - codeparrot_training - Step 36478: {'lr': 9.168010706128649e-05, 'samples': 18677248, 'steps': 36478, 'loss/train': 0.7139557003974915}
+02/26/2022 00:48:17 - INFO - codeparrot_training - Step 36479: {'lr': 9.166744411307629e-05, 'samples': 18677760, 'steps': 36479, 'loss/train': 2.1932663917541504}
+02/26/2022 00:48:20 - INFO - codeparrot_training - Step 36480: {'lr': 9.165478184311248e-05, 'samples': 18678272, 'steps': 36480, 'loss/train': 1.1974139213562012}
+02/26/2022 00:48:26 - INFO - codeparrot_training - Step 36481: {'lr': 9.164212025144933e-05, 'samples': 18678784, 'steps': 36481, 'loss/train': 1.3745049238204956}
+02/26/2022 00:48:29 - INFO - codeparrot_training - Step 36482: {'lr': 9.162945933814101e-05, 'samples': 18679296, 'steps': 36482, 'loss/train': 1.2544997930526733}
+02/26/2022 00:48:36 - INFO - codeparrot_training - Step 36483: {'lr': 9.161679910324175e-05, 'samples': 18679808, 'steps': 36483, 'loss/train': 2.548205852508545}
+02/26/2022 00:48:39 - INFO - codeparrot_training - Step 36484: {'lr': 9.160413954680591e-05, 'samples': 18680320, 'steps': 36484, 'loss/train': 1.778460144996643}
+02/26/2022 00:48:45 - INFO - codeparrot_training - Step 36485: {'lr': 9.159148066888761e-05, 'samples': 18680832, 'steps': 36485, 'loss/train': 2.4942657947540283}
+02/26/2022 00:48:48 - INFO - codeparrot_training - Step 36486: {'lr': 9.157882246954113e-05, 'samples': 18681344, 'steps': 36486, 'loss/train': 0.688693642616272}
+02/26/2022 00:48:54 - INFO - codeparrot_training - Step 36487: {'lr': 9.156616494882059e-05, 'samples': 18681856, 'steps': 36487, 'loss/train': 0.7565783262252808}
+02/26/2022 00:48:57 - INFO - codeparrot_training - Step 36488: {'lr': 9.155350810678037e-05, 'samples': 18682368, 'steps': 36488, 'loss/train': 1.8944106101989746}
+02/26/2022 00:49:03 - INFO - codeparrot_training - Step 36489: {'lr': 9.154085194347453e-05, 'samples': 18682880, 'steps': 36489, 'loss/train': 0.5719866752624512}
+02/26/2022 00:49:06 - INFO - codeparrot_training - Step 36490: {'lr': 9.152819645895752e-05, 'samples': 18683392, 'steps': 36490, 'loss/train': 1.1436283588409424}
+02/26/2022 00:49:12 - INFO - codeparrot_training - Step 36491: {'lr': 9.151554165328324e-05, 'samples': 18683904, 'steps': 36491, 'loss/train': 2.089611053466797}
+02/26/2022 00:49:15 - INFO - codeparrot_training - Step 36492: {'lr': 9.150288752650612e-05, 'samples': 18684416, 'steps': 36492, 'loss/train': 0.33290091156959534}
+02/26/2022 00:49:21 - INFO - codeparrot_training - Step 36493: {'lr': 9.149023407868023e-05, 'samples': 18684928, 'steps': 36493, 'loss/train': 1.9095097780227661}
+02/26/2022 00:49:25 - INFO - codeparrot_training - Step 36494: {'lr': 9.147758130986004e-05, 'samples': 18685440, 'steps': 36494, 'loss/train': 1.8613122701644897}
+02/26/2022 00:49:30 - INFO - codeparrot_training - Step 36495: {'lr': 9.146492922009933e-05, 'samples': 18685952, 'steps': 36495, 'loss/train': 0.5094005465507507}
+02/26/2022 00:49:34 - INFO - codeparrot_training - Step 36496: {'lr': 9.145227780945264e-05, 'samples': 18686464, 'steps': 36496, 'loss/train': 0.2982470393180847}
+02/26/2022 00:49:40 - INFO - codeparrot_training - Step 36497: {'lr': 9.143962707797396e-05, 'samples': 18686976, 'steps': 36497, 'loss/train': 1.2405245304107666}
+02/26/2022 00:49:43 - INFO - codeparrot_training - Step 36498: {'lr': 9.142697702571775e-05, 'samples': 18687488, 'steps': 36498, 'loss/train': 1.6004984378814697}
+02/26/2022 00:49:49 - INFO - codeparrot_training - Step 36499: {'lr': 9.141432765273783e-05, 'samples': 18688000, 'steps': 36499, 'loss/train': 2.5078694820404053}
+02/26/2022 00:49:52 - INFO - codeparrot_training - Step 36500: {'lr': 9.140167895908866e-05, 'samples': 18688512, 'steps': 36500, 'loss/train': 1.4973552227020264}
+02/26/2022 00:49:58 - INFO - codeparrot_training - Step 36501: {'lr': 9.138903094482426e-05, 'samples': 18689024, 'steps': 36501, 'loss/train': 2.0895373821258545}
+02/26/2022 00:50:01 - INFO - codeparrot_training - Step 36502: {'lr': 9.137638360999898e-05, 'samples': 18689536, 'steps': 36502, 'loss/train': 1.540652871131897}
+02/26/2022 00:50:07 - INFO - codeparrot_training - Step 36503: {'lr': 9.136373695466687e-05, 'samples': 18690048, 'steps': 36503, 'loss/train': 1.1501940488815308}
+02/26/2022 00:50:10 - INFO - codeparrot_training - Step 36504: {'lr': 9.135109097888217e-05, 'samples': 18690560, 'steps': 36504, 'loss/train': 2.350374698638916}
+02/26/2022 00:50:16 - INFO - codeparrot_training - Step 36505: {'lr': 9.133844568269892e-05, 'samples': 18691072, 'steps': 36505, 'loss/train': 2.0685176849365234}
+02/26/2022 00:50:19 - INFO - codeparrot_training - Step 36506: {'lr': 9.13258010661715e-05, 'samples': 18691584, 'steps': 36506, 'loss/train': 1.2391624450683594}
+02/26/2022 00:50:26 - INFO - codeparrot_training - Step 36507: {'lr': 9.131315712935392e-05, 'samples': 18692096, 'steps': 36507, 'loss/train': 1.4086748361587524}
+02/26/2022 00:50:29 - INFO - codeparrot_training - Step 36508: {'lr': 9.130051387230043e-05, 'samples': 18692608, 'steps': 36508, 'loss/train': 1.7807008028030396}
+02/26/2022 00:50:34 - INFO - codeparrot_training - Step 36509: {'lr': 9.128787129506502e-05, 'samples': 18693120, 'steps': 36509, 'loss/train': 1.27669095993042}
+02/26/2022 00:50:38 - INFO - codeparrot_training - Step 36510: {'lr': 9.127522939770208e-05, 'samples': 18693632, 'steps': 36510, 'loss/train': 1.4267017841339111}
+02/26/2022 00:50:43 - INFO - codeparrot_training - Step 36511: {'lr': 9.126258818026565e-05, 'samples': 18694144, 'steps': 36511, 'loss/train': 1.5238736867904663}
+02/26/2022 00:50:47 - INFO - codeparrot_training - Step 36512: {'lr': 9.124994764280989e-05, 'samples': 18694656, 'steps': 36512, 'loss/train': 1.1365036964416504}
+02/26/2022 00:50:52 - INFO - codeparrot_training - Step 36513: {'lr': 9.123730778538886e-05, 'samples': 18695168, 'steps': 36513, 'loss/train': 1.2731398344039917}
+02/26/2022 00:50:56 - INFO - codeparrot_training - Step 36514: {'lr': 9.122466860805687e-05, 'samples': 18695680, 'steps': 36514, 'loss/train': 1.5848681926727295}
+02/26/2022 00:51:01 - INFO - codeparrot_training - Step 36515: {'lr': 9.121203011086799e-05, 'samples': 18696192, 'steps': 36515, 'loss/train': 2.597829818725586}
+02/26/2022 00:51:05 - INFO - codeparrot_training - Step 36516: {'lr': 9.119939229387636e-05, 'samples': 18696704, 'steps': 36516, 'loss/train': 0.7429189085960388}
+02/26/2022 00:51:10 - INFO - codeparrot_training - Step 36517: {'lr': 9.1186755157136e-05, 'samples': 18697216, 'steps': 36517, 'loss/train': 1.1402100324630737}
+02/26/2022 00:51:14 - INFO - codeparrot_training - Step 36518: {'lr': 9.117411870070128e-05, 'samples': 18697728, 'steps': 36518, 'loss/train': 0.9129608869552612}
+02/26/2022 00:51:20 - INFO - codeparrot_training - Step 36519: {'lr': 9.116148292462614e-05, 'samples': 18698240, 'steps': 36519, 'loss/train': 0.587259829044342}
+02/26/2022 00:51:23 - INFO - codeparrot_training - Step 36520: {'lr': 9.114884782896482e-05, 'samples': 18698752, 'steps': 36520, 'loss/train': 1.1782442331314087}
+02/26/2022 00:51:29 - INFO - codeparrot_training - Step 36521: {'lr': 9.113621341377129e-05, 'samples': 18699264, 'steps': 36521, 'loss/train': 2.414233684539795}
+02/26/2022 00:51:32 - INFO - codeparrot_training - Step 36522: {'lr': 9.112357967909988e-05, 'samples': 18699776, 'steps': 36522, 'loss/train': 0.5109829306602478}
+02/26/2022 00:51:38 - INFO - codeparrot_training - Step 36523: {'lr': 9.111094662500458e-05, 'samples': 18700288, 'steps': 36523, 'loss/train': 0.5842905640602112}
+02/26/2022 00:51:41 - INFO - codeparrot_training - Step 36524: {'lr': 9.109831425153956e-05, 'samples': 18700800, 'steps': 36524, 'loss/train': 2.1260931491851807}
+02/26/2022 00:51:46 - INFO - codeparrot_training - Step 36525: {'lr': 9.10856825587588e-05, 'samples': 18701312, 'steps': 36525, 'loss/train': 1.952371597290039}
+02/26/2022 00:51:50 - INFO - codeparrot_training - Step 36526: {'lr': 9.107305154671658e-05, 'samples': 18701824, 'steps': 36526, 'loss/train': 1.3889007568359375}
+02/26/2022 00:51:55 - INFO - codeparrot_training - Step 36527: {'lr': 9.106042121546698e-05, 'samples': 18702336, 'steps': 36527, 'loss/train': 1.3719687461853027}
+02/26/2022 00:51:59 - INFO - codeparrot_training - Step 36528: {'lr': 9.104779156506395e-05, 'samples': 18702848, 'steps': 36528, 'loss/train': 2.0373573303222656}
+02/26/2022 00:52:07 - INFO - codeparrot_training - Step 36529: {'lr': 9.10351625955619e-05, 'samples': 18703360, 'steps': 36529, 'loss/train': 1.4478174448013306}
+02/26/2022 00:52:10 - INFO - codeparrot_training - Step 36530: {'lr': 9.102253430701458e-05, 'samples': 18703872, 'steps': 36530, 'loss/train': 1.9549286365509033}
+02/26/2022 00:52:16 - INFO - codeparrot_training - Step 36531: {'lr': 9.100990669947629e-05, 'samples': 18704384, 'steps': 36531, 'loss/train': 1.2932859659194946}
+02/26/2022 00:52:19 - INFO - codeparrot_training - Step 36532: {'lr': 9.099727977300101e-05, 'samples': 18704896, 'steps': 36532, 'loss/train': 0.42060935497283936}
+02/26/2022 00:52:25 - INFO - codeparrot_training - Step 36533: {'lr': 9.098465352764308e-05, 'samples': 18705408, 'steps': 36533, 'loss/train': 2.394883871078491}
+02/26/2022 00:52:28 - INFO - codeparrot_training - Step 36534: {'lr': 9.097202796345619e-05, 'samples': 18705920, 'steps': 36534, 'loss/train': 1.678720474243164}
+02/26/2022 00:52:34 - INFO - codeparrot_training - Step 36535: {'lr': 9.095940308049474e-05, 'samples': 18706432, 'steps': 36535, 'loss/train': 2.3407063484191895}
+02/26/2022 00:52:37 - INFO - codeparrot_training - Step 36536: {'lr': 9.094677887881264e-05, 'samples': 18706944, 'steps': 36536, 'loss/train': 2.2407546043395996}
+02/26/2022 00:52:43 - INFO - codeparrot_training - Step 36537: {'lr': 9.093415535846411e-05, 'samples': 18707456, 'steps': 36537, 'loss/train': 1.5770275592803955}
+02/26/2022 00:52:46 - INFO - codeparrot_training - Step 36538: {'lr': 9.092153251950314e-05, 'samples': 18707968, 'steps': 36538, 'loss/train': 1.9819976091384888}
+02/26/2022 00:52:52 - INFO - codeparrot_training - Step 36539: {'lr': 9.090891036198381e-05, 'samples': 18708480, 'steps': 36539, 'loss/train': 2.4673635959625244}
+02/26/2022 00:52:55 - INFO - codeparrot_training - Step 36540: {'lr': 9.089628888596011e-05, 'samples': 18708992, 'steps': 36540, 'loss/train': 1.967557668685913}
+02/26/2022 00:53:02 - INFO - codeparrot_training - Step 36541: {'lr': 9.088366809148627e-05, 'samples': 18709504, 'steps': 36541, 'loss/train': 1.6730071306228638}
+02/26/2022 00:53:06 - INFO - codeparrot_training - Step 36542: {'lr': 9.087104797861628e-05, 'samples': 18710016, 'steps': 36542, 'loss/train': 1.5909526348114014}
+02/26/2022 00:53:11 - INFO - codeparrot_training - Step 36543: {'lr': 9.085842854740418e-05, 'samples': 18710528, 'steps': 36543, 'loss/train': 1.136153221130371}
+02/26/2022 00:53:15 - INFO - codeparrot_training - Step 36544: {'lr': 9.084580979790396e-05, 'samples': 18711040, 'steps': 36544, 'loss/train': 1.3035250902175903}
+02/26/2022 00:53:20 - INFO - codeparrot_training - Step 36545: {'lr': 9.083319173016986e-05, 'samples': 18711552, 'steps': 36545, 'loss/train': 1.2188856601715088}
+02/26/2022 00:53:24 - INFO - codeparrot_training - Step 36546: {'lr': 9.082057434425578e-05, 'samples': 18712064, 'steps': 36546, 'loss/train': 0.9534711837768555}
+02/26/2022 00:53:29 - INFO - codeparrot_training - Step 36547: {'lr': 9.080795764021585e-05, 'samples': 18712576, 'steps': 36547, 'loss/train': 2.044180393218994}
+02/26/2022 00:53:33 - INFO - codeparrot_training - Step 36548: {'lr': 9.079534161810396e-05, 'samples': 18713088, 'steps': 36548, 'loss/train': 1.6012495756149292}
+02/26/2022 00:53:39 - INFO - codeparrot_training - Step 36549: {'lr': 9.07827262779744e-05, 'samples': 18713600, 'steps': 36549, 'loss/train': 1.5406208038330078}
+02/26/2022 00:53:42 - INFO - codeparrot_training - Step 36550: {'lr': 9.077011161988097e-05, 'samples': 18714112, 'steps': 36550, 'loss/train': 1.481308937072754}
+02/26/2022 00:53:49 - INFO - codeparrot_training - Step 36551: {'lr': 9.0757497643878e-05, 'samples': 18714624, 'steps': 36551, 'loss/train': 2.161729574203491}
+02/26/2022 00:53:55 - INFO - codeparrot_training - Step 36552: {'lr': 9.074488435001918e-05, 'samples': 18715136, 'steps': 36552, 'loss/train': 2.0254125595092773}
+02/26/2022 00:53:58 - INFO - codeparrot_training - Step 36553: {'lr': 9.07322717383588e-05, 'samples': 18715648, 'steps': 36553, 'loss/train': 2.609917640686035}
+02/26/2022 00:54:04 - INFO - codeparrot_training - Step 36554: {'lr': 9.071965980895069e-05, 'samples': 18716160, 'steps': 36554, 'loss/train': 2.134484052658081}
+02/26/2022 00:54:07 - INFO - codeparrot_training - Step 36555: {'lr': 9.070704856184913e-05, 'samples': 18716672, 'steps': 36555, 'loss/train': 2.4393458366394043}
+02/26/2022 00:54:13 - INFO - codeparrot_training - Step 36556: {'lr': 9.069443799710786e-05, 'samples': 18717184, 'steps': 36556, 'loss/train': 2.3811662197113037}
+02/26/2022 00:54:16 - INFO - codeparrot_training - Step 36557: {'lr': 9.068182811478113e-05, 'samples': 18717696, 'steps': 36557, 'loss/train': 1.608520269393921}
+02/26/2022 00:54:22 - INFO - codeparrot_training - Step 36558: {'lr': 9.066921891492275e-05, 'samples': 18718208, 'steps': 36558, 'loss/train': 0.580219030380249}
+02/26/2022 00:54:25 - INFO - codeparrot_training - Step 36559: {'lr': 9.065661039758699e-05, 'samples': 18718720, 'steps': 36559, 'loss/train': 1.437660574913025}
+02/26/2022 00:54:32 - INFO - codeparrot_training - Step 36560: {'lr': 9.064400256282756e-05, 'samples': 18719232, 'steps': 36560, 'loss/train': 1.2620149850845337}
+02/26/2022 00:54:36 - INFO - codeparrot_training - Step 36561: {'lr': 9.063139541069872e-05, 'samples': 18719744, 'steps': 36561, 'loss/train': 0.777113676071167}
+02/26/2022 00:54:41 - INFO - codeparrot_training - Step 36562: {'lr': 9.061878894125439e-05, 'samples': 18720256, 'steps': 36562, 'loss/train': 2.1638152599334717}
+02/26/2022 00:54:45 - INFO - codeparrot_training - Step 36563: {'lr': 9.060618315454843e-05, 'samples': 18720768, 'steps': 36563, 'loss/train': 1.0005766153335571}
+02/26/2022 00:54:50 - INFO - codeparrot_training - Step 36564: {'lr': 9.059357805063509e-05, 'samples': 18721280, 'steps': 36564, 'loss/train': 1.9134806394577026}
+02/26/2022 00:54:54 - INFO - codeparrot_training - Step 36565: {'lr': 9.058097362956825e-05, 'samples': 18721792, 'steps': 36565, 'loss/train': 1.4803436994552612}
+02/26/2022 00:54:59 - INFO - codeparrot_training - Step 36566: {'lr': 9.056836989140185e-05, 'samples': 18722304, 'steps': 36566, 'loss/train': 2.0330381393432617}
+02/26/2022 00:55:03 - INFO - codeparrot_training - Step 36567: {'lr': 9.055576683618987e-05, 'samples': 18722816, 'steps': 36567, 'loss/train': 1.4669479131698608}
+02/26/2022 00:55:08 - INFO - codeparrot_training - Step 36568: {'lr': 9.054316446398647e-05, 'samples': 18723328, 'steps': 36568, 'loss/train': 1.5934340953826904}
+02/26/2022 00:55:12 - INFO - codeparrot_training - Step 36569: {'lr': 9.053056277484547e-05, 'samples': 18723840, 'steps': 36569, 'loss/train': 1.7027442455291748}
+02/26/2022 00:55:20 - INFO - codeparrot_training - Step 36570: {'lr': 9.051796176882093e-05, 'samples': 18724352, 'steps': 36570, 'loss/train': 0.7654800415039062}
+02/26/2022 00:55:23 - INFO - codeparrot_training - Step 36571: {'lr': 9.050536144596672e-05, 'samples': 18724864, 'steps': 36571, 'loss/train': 1.9175941944122314}
+02/26/2022 00:55:29 - INFO - codeparrot_training - Step 36572: {'lr': 9.049276180633698e-05, 'samples': 18725376, 'steps': 36572, 'loss/train': 2.3417036533355713}
+02/26/2022 00:55:32 - INFO - codeparrot_training - Step 36573: {'lr': 9.048016284998559e-05, 'samples': 18725888, 'steps': 36573, 'loss/train': 1.097623348236084}
+02/26/2022 00:55:38 - INFO - codeparrot_training - Step 36574: {'lr': 9.046756457696653e-05, 'samples': 18726400, 'steps': 36574, 'loss/train': 1.2813806533813477}
+02/26/2022 00:55:41 - INFO - codeparrot_training - Step 36575: {'lr': 9.045496698733368e-05, 'samples': 18726912, 'steps': 36575, 'loss/train': 2.026169538497925}
+02/26/2022 00:55:47 - INFO - codeparrot_training - Step 36576: {'lr': 9.044237008114117e-05, 'samples': 18727424, 'steps': 36576, 'loss/train': 1.244621992111206}
+02/26/2022 00:55:50 - INFO - codeparrot_training - Step 36577: {'lr': 9.042977385844289e-05, 'samples': 18727936, 'steps': 36577, 'loss/train': 0.2353709638118744}
+02/26/2022 00:55:56 - INFO - codeparrot_training - Step 36578: {'lr': 9.04171783192928e-05, 'samples': 18728448, 'steps': 36578, 'loss/train': 2.1739213466644287}
+02/26/2022 00:55:59 - INFO - codeparrot_training - Step 36579: {'lr': 9.040458346374475e-05, 'samples': 18728960, 'steps': 36579, 'loss/train': 2.2177698612213135}
+02/26/2022 00:56:05 - INFO - codeparrot_training - Step 36580: {'lr': 9.039198929185286e-05, 'samples': 18729472, 'steps': 36580, 'loss/train': 1.8375288248062134}
+02/26/2022 00:56:08 - INFO - codeparrot_training - Step 36581: {'lr': 9.037939580367102e-05, 'samples': 18729984, 'steps': 36581, 'loss/train': 2.5287203788757324}
+02/26/2022 00:56:14 - INFO - codeparrot_training - Step 36582: {'lr': 9.036680299925315e-05, 'samples': 18730496, 'steps': 36582, 'loss/train': 1.8921520709991455}
+02/26/2022 00:56:17 - INFO - codeparrot_training - Step 36583: {'lr': 9.035421087865314e-05, 'samples': 18731008, 'steps': 36583, 'loss/train': 1.2827168703079224}
+02/26/2022 00:56:23 - INFO - codeparrot_training - Step 36584: {'lr': 9.034161944192506e-05, 'samples': 18731520, 'steps': 36584, 'loss/train': 2.0836844444274902}
+02/26/2022 00:56:26 - INFO - codeparrot_training - Step 36585: {'lr': 9.032902868912274e-05, 'samples': 18732032, 'steps': 36585, 'loss/train': 2.06073260307312}
+02/26/2022 00:56:33 - INFO - codeparrot_training - Step 36586: {'lr': 9.031643862030028e-05, 'samples': 18732544, 'steps': 36586, 'loss/train': 2.0865678787231445}
+02/26/2022 00:56:37 - INFO - codeparrot_training - Step 36587: {'lr': 9.030384923551136e-05, 'samples': 18733056, 'steps': 36587, 'loss/train': 0.4919186234474182}
+02/26/2022 00:56:42 - INFO - codeparrot_training - Step 36588: {'lr': 9.02912605348101e-05, 'samples': 18733568, 'steps': 36588, 'loss/train': 1.92879056930542}
+02/26/2022 00:56:46 - INFO - codeparrot_training - Step 36589: {'lr': 9.027867251825031e-05, 'samples': 18734080, 'steps': 36589, 'loss/train': 1.7919151782989502}
+02/26/2022 00:56:51 - INFO - codeparrot_training - Step 36590: {'lr': 9.026608518588612e-05, 'samples': 18734592, 'steps': 36590, 'loss/train': 1.3220763206481934}
+02/26/2022 00:56:55 - INFO - codeparrot_training - Step 36591: {'lr': 9.025349853777113e-05, 'samples': 18735104, 'steps': 36591, 'loss/train': 0.5083885788917542}
+02/26/2022 00:57:00 - INFO - codeparrot_training - Step 36592: {'lr': 9.02409125739595e-05, 'samples': 18735616, 'steps': 36592, 'loss/train': 1.3132753372192383}
+02/26/2022 00:57:04 - INFO - codeparrot_training - Step 36593: {'lr': 9.0228327294505e-05, 'samples': 18736128, 'steps': 36593, 'loss/train': 0.5286920666694641}
+02/26/2022 00:57:09 - INFO - codeparrot_training - Step 36594: {'lr': 9.021574269946179e-05, 'samples': 18736640, 'steps': 36594, 'loss/train': 0.41230759024620056}
+02/26/2022 00:57:13 - INFO - codeparrot_training - Step 36595: {'lr': 9.020315878888344e-05, 'samples': 18737152, 'steps': 36595, 'loss/train': 0.9003009796142578}
+02/26/2022 00:57:20 - INFO - codeparrot_training - Step 36596: {'lr': 9.019057556282406e-05, 'samples': 18737664, 'steps': 36596, 'loss/train': 1.450286626815796}
+02/26/2022 00:57:24 - INFO - codeparrot_training - Step 36597: {'lr': 9.017799302133747e-05, 'samples': 18738176, 'steps': 36597, 'loss/train': 2.3623502254486084}
+02/26/2022 00:57:29 - INFO - codeparrot_training - Step 36598: {'lr': 9.016541116447766e-05, 'samples': 18738688, 'steps': 36598, 'loss/train': 0.5349457263946533}
+02/26/2022 00:57:33 - INFO - codeparrot_training - Step 36599: {'lr': 9.015282999229848e-05, 'samples': 18739200, 'steps': 36599, 'loss/train': 1.5680148601531982}
+02/26/2022 00:57:38 - INFO - codeparrot_training - Step 36600: {'lr': 9.014024950485383e-05, 'samples': 18739712, 'steps': 36600, 'loss/train': 1.4445619583129883}
+02/26/2022 00:57:41 - INFO - codeparrot_training - Step 36601: {'lr': 9.012766970219752e-05, 'samples': 18740224, 'steps': 36601, 'loss/train': 2.0936193466186523}
+02/26/2022 00:57:47 - INFO - codeparrot_training - Step 36602: {'lr': 9.011509058438358e-05, 'samples': 18740736, 'steps': 36602, 'loss/train': 1.6281620264053345}
+02/26/2022 00:57:51 - INFO - codeparrot_training - Step 36603: {'lr': 9.010251215146581e-05, 'samples': 18741248, 'steps': 36603, 'loss/train': 1.5327023267745972}
+02/26/2022 00:57:56 - INFO - codeparrot_training - Step 36604: {'lr': 9.008993440349811e-05, 'samples': 18741760, 'steps': 36604, 'loss/train': 1.826502799987793}
+02/26/2022 00:58:00 - INFO - codeparrot_training - Step 36605: {'lr': 9.007735734053426e-05, 'samples': 18742272, 'steps': 36605, 'loss/train': 0.7369265556335449}
+02/26/2022 00:58:07 - INFO - codeparrot_training - Step 36606: {'lr': 9.006478096262836e-05, 'samples': 18742784, 'steps': 36606, 'loss/train': 1.9530997276306152}
+02/26/2022 00:58:11 - INFO - codeparrot_training - Step 36607: {'lr': 9.00522052698341e-05, 'samples': 18743296, 'steps': 36607, 'loss/train': 1.1178377866744995}
+02/26/2022 00:58:16 - INFO - codeparrot_training - Step 36608: {'lr': 9.003963026220543e-05, 'samples': 18743808, 'steps': 36608, 'loss/train': 1.75657057762146}
+02/26/2022 00:58:20 - INFO - codeparrot_training - Step 36609: {'lr': 9.002705593979618e-05, 'samples': 18744320, 'steps': 36609, 'loss/train': 2.128976345062256}
+02/26/2022 00:58:25 - INFO - codeparrot_training - Step 36610: {'lr': 9.001448230266015e-05, 'samples': 18744832, 'steps': 36610, 'loss/train': 1.520991325378418}
+02/26/2022 00:58:28 - INFO - codeparrot_training - Step 36611: {'lr': 9.000190935085137e-05, 'samples': 18745344, 'steps': 36611, 'loss/train': 1.533866047859192}
+02/26/2022 00:58:34 - INFO - codeparrot_training - Step 36612: {'lr': 8.998933708442359e-05, 'samples': 18745856, 'steps': 36612, 'loss/train': 2.0014543533325195}
+02/26/2022 00:58:37 - INFO - codeparrot_training - Step 36613: {'lr': 8.997676550343067e-05, 'samples': 18746368, 'steps': 36613, 'loss/train': 1.734210729598999}
+02/26/2022 00:58:43 - INFO - codeparrot_training - Step 36614: {'lr': 8.996419460792641e-05, 'samples': 18746880, 'steps': 36614, 'loss/train': 2.078361749649048}
+02/26/2022 00:58:46 - INFO - codeparrot_training - Step 36615: {'lr': 8.995162439796478e-05, 'samples': 18747392, 'steps': 36615, 'loss/train': 0.3727642595767975}
+02/26/2022 00:58:54 - INFO - codeparrot_training - Step 36616: {'lr': 8.99390548735996e-05, 'samples': 18747904, 'steps': 36616, 'loss/train': 1.3749072551727295}
+02/26/2022 00:58:57 - INFO - codeparrot_training - Step 36617: {'lr': 8.992648603488468e-05, 'samples': 18748416, 'steps': 36617, 'loss/train': 0.5264908075332642}
+02/26/2022 00:59:03 - INFO - codeparrot_training - Step 36618: {'lr': 8.991391788187376e-05, 'samples': 18748928, 'steps': 36618, 'loss/train': 2.08695125579834}
+02/26/2022 00:59:06 - INFO - codeparrot_training - Step 36619: {'lr': 8.990135041462088e-05, 'samples': 18749440, 'steps': 36619, 'loss/train': 1.4376933574676514}
+02/26/2022 00:59:12 - INFO - codeparrot_training - Step 36620: {'lr': 8.98887836331798e-05, 'samples': 18749952, 'steps': 36620, 'loss/train': 2.2350635528564453}
+02/26/2022 00:59:15 - INFO - codeparrot_training - Step 36621: {'lr': 8.987621753760431e-05, 'samples': 18750464, 'steps': 36621, 'loss/train': 1.6088550090789795}
+02/26/2022 00:59:21 - INFO - codeparrot_training - Step 36622: {'lr': 8.986365212794815e-05, 'samples': 18750976, 'steps': 36622, 'loss/train': 1.5629996061325073}
+02/26/2022 00:59:24 - INFO - codeparrot_training - Step 36623: {'lr': 8.985108740426538e-05, 'samples': 18751488, 'steps': 36623, 'loss/train': 1.1186065673828125}
+02/26/2022 00:59:30 - INFO - codeparrot_training - Step 36624: {'lr': 8.983852336660957e-05, 'samples': 18752000, 'steps': 36624, 'loss/train': 1.2251043319702148}
+02/26/2022 00:59:33 - INFO - codeparrot_training - Step 36625: {'lr': 8.982596001503488e-05, 'samples': 18752512, 'steps': 36625, 'loss/train': 1.1691792011260986}
+02/26/2022 00:59:39 - INFO - codeparrot_training - Step 36626: {'lr': 8.981339734959473e-05, 'samples': 18753024, 'steps': 36626, 'loss/train': 0.4061290919780731}
+02/26/2022 00:59:43 - INFO - codeparrot_training - Step 36627: {'lr': 8.98008353703432e-05, 'samples': 18753536, 'steps': 36627, 'loss/train': 0.4246062934398651}
+02/26/2022 00:59:48 - INFO - codeparrot_training - Step 36628: {'lr': 8.978827407733395e-05, 'samples': 18754048, 'steps': 36628, 'loss/train': 2.015063524246216}
+02/26/2022 00:59:52 - INFO - codeparrot_training - Step 36629: {'lr': 8.977571347062102e-05, 'samples': 18754560, 'steps': 36629, 'loss/train': 1.1896742582321167}
+02/26/2022 00:59:55 - INFO - codeparrot_training - Step 36630: {'lr': 8.97631535502579e-05, 'samples': 18755072, 'steps': 36630, 'loss/train': 1.0434242486953735}
+02/26/2022 01:00:01 - INFO - codeparrot_training - Step 36631: {'lr': 8.975059431629862e-05, 'samples': 18755584, 'steps': 36631, 'loss/train': 0.6546241044998169}
+02/26/2022 01:00:08 - INFO - codeparrot_training - Step 36632: {'lr': 8.973803576879683e-05, 'samples': 18756096, 'steps': 36632, 'loss/train': 2.605837106704712}
+02/26/2022 01:00:11 - INFO - codeparrot_training - Step 36633: {'lr': 8.972547790780658e-05, 'samples': 18756608, 'steps': 36633, 'loss/train': 1.5168198347091675}
+02/26/2022 01:00:17 - INFO - codeparrot_training - Step 36634: {'lr': 8.971292073338133e-05, 'samples': 18757120, 'steps': 36634, 'loss/train': 0.7329345941543579}
+02/26/2022 01:00:20 - INFO - codeparrot_training - Step 36635: {'lr': 8.970036424557509e-05, 'samples': 18757632, 'steps': 36635, 'loss/train': 1.1443909406661987}
+02/26/2022 01:00:26 - INFO - codeparrot_training - Step 36636: {'lr': 8.968780844444152e-05, 'samples': 18758144, 'steps': 36636, 'loss/train': 0.39262306690216064}
+02/26/2022 01:00:29 - INFO - codeparrot_training - Step 36637: {'lr': 8.967525333003455e-05, 'samples': 18758656, 'steps': 36637, 'loss/train': 1.4619505405426025}
+02/26/2022 01:00:35 - INFO - codeparrot_training - Step 36638: {'lr': 8.966269890240788e-05, 'samples': 18759168, 'steps': 36638, 'loss/train': 2.2366981506347656}
+02/26/2022 01:00:38 - INFO - codeparrot_training - Step 36639: {'lr': 8.965014516161532e-05, 'samples': 18759680, 'steps': 36639, 'loss/train': 1.7728203535079956}
+02/26/2022 01:00:44 - INFO - codeparrot_training - Step 36640: {'lr': 8.963759210771053e-05, 'samples': 18760192, 'steps': 36640, 'loss/train': 2.0320136547088623}
+02/26/2022 01:00:47 - INFO - codeparrot_training - Step 36641: {'lr': 8.962503974074743e-05, 'samples': 18760704, 'steps': 36641, 'loss/train': 2.209712266921997}
+02/26/2022 01:00:55 - INFO - codeparrot_training - Step 36642: {'lr': 8.961248806077976e-05, 'samples': 18761216, 'steps': 36642, 'loss/train': 1.7144718170166016}
+02/26/2022 01:00:58 - INFO - codeparrot_training - Step 36643: {'lr': 8.959993706786126e-05, 'samples': 18761728, 'steps': 36643, 'loss/train': 1.9846676588058472}
+02/26/2022 01:01:04 - INFO - codeparrot_training - Step 36644: {'lr': 8.958738676204562e-05, 'samples': 18762240, 'steps': 36644, 'loss/train': 2.2778046131134033}
+02/26/2022 01:01:07 - INFO - codeparrot_training - Step 36645: {'lr': 8.957483714338674e-05, 'samples': 18762752, 'steps': 36645, 'loss/train': 1.777929425239563}
+02/26/2022 01:01:13 - INFO - codeparrot_training - Step 36646: {'lr': 8.956228821193835e-05, 'samples': 18763264, 'steps': 36646, 'loss/train': 1.7810192108154297}
+02/26/2022 01:01:16 - INFO - codeparrot_training - Step 36647: {'lr': 8.954973996775415e-05, 'samples': 18763776, 'steps': 36647, 'loss/train': 1.401321291923523}
+02/26/2022 01:01:22 - INFO - codeparrot_training - Step 36648: {'lr': 8.953719241088781e-05, 'samples': 18764288, 'steps': 36648, 'loss/train': 1.5986754894256592}
+02/26/2022 01:01:25 - INFO - codeparrot_training - Step 36649: {'lr': 8.952464554139331e-05, 'samples': 18764800, 'steps': 36649, 'loss/train': 1.8167191743850708}
+02/26/2022 01:01:31 - INFO - codeparrot_training - Step 36650: {'lr': 8.951209935932425e-05, 'samples': 18765312, 'steps': 36650, 'loss/train': 0.5065308809280396}
+02/26/2022 01:01:34 - INFO - codeparrot_training - Step 36651: {'lr': 8.949955386473441e-05, 'samples': 18765824, 'steps': 36651, 'loss/train': 1.5306487083435059}
+02/26/2022 01:01:41 - INFO - codeparrot_training - Step 36652: {'lr': 8.94870090576774e-05, 'samples': 18766336, 'steps': 36652, 'loss/train': 2.083111524581909}
+02/26/2022 01:01:45 - INFO - codeparrot_training - Step 36653: {'lr': 8.947446493820719e-05, 'samples': 18766848, 'steps': 36653, 'loss/train': 2.302685499191284}
+02/26/2022 01:01:50 - INFO - codeparrot_training - Step 36654: {'lr': 8.946192150637739e-05, 'samples': 18767360, 'steps': 36654, 'loss/train': 2.792943000793457}
+02/26/2022 01:01:54 - INFO - codeparrot_training - Step 36655: {'lr': 8.944937876224173e-05, 'samples': 18767872, 'steps': 36655, 'loss/train': 1.550115942955017}
+02/26/2022 01:01:59 - INFO - codeparrot_training - Step 36656: {'lr': 8.94368367058539e-05, 'samples': 18768384, 'steps': 36656, 'loss/train': 0.9503949880599976}
+02/26/2022 01:02:03 - INFO - codeparrot_training - Step 36657: {'lr': 8.942429533726773e-05, 'samples': 18768896, 'steps': 36657, 'loss/train': 2.2719216346740723}
+02/26/2022 01:02:08 - INFO - codeparrot_training - Step 36658: {'lr': 8.941175465653692e-05, 'samples': 18769408, 'steps': 36658, 'loss/train': 1.3227165937423706}
+02/26/2022 01:02:12 - INFO - codeparrot_training - Step 36659: {'lr': 8.939921466371503e-05, 'samples': 18769920, 'steps': 36659, 'loss/train': 2.4320244789123535}
+02/26/2022 01:02:17 - INFO - codeparrot_training - Step 36660: {'lr': 8.938667535885611e-05, 'samples': 18770432, 'steps': 36660, 'loss/train': 1.371164321899414}
+02/26/2022 01:02:21 - INFO - codeparrot_training - Step 36661: {'lr': 8.937413674201351e-05, 'samples': 18770944, 'steps': 36661, 'loss/train': 2.8989126682281494}
+02/26/2022 01:02:26 - INFO - codeparrot_training - Step 36662: {'lr': 8.936159881324118e-05, 'samples': 18771456, 'steps': 36662, 'loss/train': 1.17244291305542}
+02/26/2022 01:02:30 - INFO - codeparrot_training - Step 36663: {'lr': 8.934906157259265e-05, 'samples': 18771968, 'steps': 36663, 'loss/train': 2.791363477706909}
+02/26/2022 01:02:37 - INFO - codeparrot_training - Step 36664: {'lr': 8.93365250201219e-05, 'samples': 18772480, 'steps': 36664, 'loss/train': 1.1157643795013428}
+02/26/2022 01:02:40 - INFO - codeparrot_training - Step 36665: {'lr': 8.932398915588232e-05, 'samples': 18772992, 'steps': 36665, 'loss/train': 1.331910252571106}
+02/26/2022 01:02:46 - INFO - codeparrot_training - Step 36666: {'lr': 8.931145397992781e-05, 'samples': 18773504, 'steps': 36666, 'loss/train': 2.3801474571228027}
+02/26/2022 01:02:50 - INFO - codeparrot_training - Step 36667: {'lr': 8.929891949231195e-05, 'samples': 18774016, 'steps': 36667, 'loss/train': 0.6971074938774109}
+02/26/2022 01:02:55 - INFO - codeparrot_training - Step 36668: {'lr': 8.928638569308862e-05, 'samples': 18774528, 'steps': 36668, 'loss/train': 1.557456374168396}
+02/26/2022 01:02:59 - INFO - codeparrot_training - Step 36669: {'lr': 8.927385258231124e-05, 'samples': 18775040, 'steps': 36669, 'loss/train': 1.8067389726638794}
+02/26/2022 01:03:04 - INFO - codeparrot_training - Step 36670: {'lr': 8.926132016003371e-05, 'samples': 18775552, 'steps': 36670, 'loss/train': 2.0618882179260254}
+02/26/2022 01:03:08 - INFO - codeparrot_training - Step 36671: {'lr': 8.924878842630957e-05, 'samples': 18776064, 'steps': 36671, 'loss/train': 2.746779441833496}
+02/26/2022 01:03:13 - INFO - codeparrot_training - Step 36672: {'lr': 8.923625738119265e-05, 'samples': 18776576, 'steps': 36672, 'loss/train': 2.2812204360961914}
+02/26/2022 01:03:17 - INFO - codeparrot_training - Step 36673: {'lr': 8.922372702473655e-05, 'samples': 18777088, 'steps': 36673, 'loss/train': 2.902531385421753}
+02/26/2022 01:03:22 - INFO - codeparrot_training - Step 36674: {'lr': 8.921119735699496e-05, 'samples': 18777600, 'steps': 36674, 'loss/train': 2.0630908012390137}
+02/26/2022 01:03:26 - INFO - codeparrot_training - Step 36675: {'lr': 8.919866837802146e-05, 'samples': 18778112, 'steps': 36675, 'loss/train': 1.6360176801681519}
+02/26/2022 01:03:31 - INFO - codeparrot_training - Step 36676: {'lr': 8.91861400878699e-05, 'samples': 18778624, 'steps': 36676, 'loss/train': 2.1341984272003174}
+02/26/2022 01:03:35 - INFO - codeparrot_training - Step 36677: {'lr': 8.917361248659384e-05, 'samples': 18779136, 'steps': 36677, 'loss/train': 1.956908106803894}
+02/26/2022 01:03:42 - INFO - codeparrot_training - Step 36678: {'lr': 8.916108557424696e-05, 'samples': 18779648, 'steps': 36678, 'loss/train': 2.20556378364563}
+02/26/2022 01:03:45 - INFO - codeparrot_training - Step 36679: {'lr': 8.914855935088281e-05, 'samples': 18780160, 'steps': 36679, 'loss/train': 0.5835480093955994}
+02/26/2022 01:03:51 - INFO - codeparrot_training - Step 36680: {'lr': 8.913603381655528e-05, 'samples': 18780672, 'steps': 36680, 'loss/train': 1.3446823358535767}
+02/26/2022 01:03:54 - INFO - codeparrot_training - Step 36681: {'lr': 8.912350897131788e-05, 'samples': 18781184, 'steps': 36681, 'loss/train': 1.9582816362380981}
+02/26/2022 01:04:00 - INFO - codeparrot_training - Step 36682: {'lr': 8.911098481522429e-05, 'samples': 18781696, 'steps': 36682, 'loss/train': 2.449564218521118}
+02/26/2022 01:04:03 - INFO - codeparrot_training - Step 36683: {'lr': 8.909846134832805e-05, 'samples': 18782208, 'steps': 36683, 'loss/train': 1.4189239740371704}
+02/26/2022 01:04:09 - INFO - codeparrot_training - Step 36684: {'lr': 8.908593857068301e-05, 'samples': 18782720, 'steps': 36684, 'loss/train': 1.3596614599227905}
+02/26/2022 01:04:12 - INFO - codeparrot_training - Step 36685: {'lr': 8.907341648234262e-05, 'samples': 18783232, 'steps': 36685, 'loss/train': 2.7265517711639404}
+02/26/2022 01:04:18 - INFO - codeparrot_training - Step 36686: {'lr': 8.906089508336079e-05, 'samples': 18783744, 'steps': 36686, 'loss/train': 2.148716926574707}
+02/26/2022 01:04:21 - INFO - codeparrot_training - Step 36687: {'lr': 8.904837437379082e-05, 'samples': 18784256, 'steps': 36687, 'loss/train': 0.6689602136611938}
+02/26/2022 01:04:29 - INFO - codeparrot_training - Step 36688: {'lr': 8.903585435368658e-05, 'samples': 18784768, 'steps': 36688, 'loss/train': 1.9057968854904175}
+02/26/2022 01:04:32 - INFO - codeparrot_training - Step 36689: {'lr': 8.902333502310154e-05, 'samples': 18785280, 'steps': 36689, 'loss/train': 0.8908526301383972}
+02/26/2022 01:04:38 - INFO - codeparrot_training - Step 36690: {'lr': 8.90108163820896e-05, 'samples': 18785792, 'steps': 36690, 'loss/train': 1.8692876100540161}
+02/26/2022 01:04:41 - INFO - codeparrot_training - Step 36691: {'lr': 8.899829843070404e-05, 'samples': 18786304, 'steps': 36691, 'loss/train': 1.663933515548706}
+02/26/2022 01:04:47 - INFO - codeparrot_training - Step 36692: {'lr': 8.89857811689987e-05, 'samples': 18786816, 'steps': 36692, 'loss/train': 0.9016863107681274}
+02/26/2022 01:04:50 - INFO - codeparrot_training - Step 36693: {'lr': 8.897326459702715e-05, 'samples': 18787328, 'steps': 36693, 'loss/train': 2.07425856590271}
+02/26/2022 01:04:56 - INFO - codeparrot_training - Step 36694: {'lr': 8.896074871484295e-05, 'samples': 18787840, 'steps': 36694, 'loss/train': 0.817865788936615}
+02/26/2022 01:04:59 - INFO - codeparrot_training - Step 36695: {'lr': 8.894823352249982e-05, 'samples': 18788352, 'steps': 36695, 'loss/train': 1.2361661195755005}
+02/26/2022 01:05:05 - INFO - codeparrot_training - Step 36696: {'lr': 8.893571902005132e-05, 'samples': 18788864, 'steps': 36696, 'loss/train': 1.9286699295043945}
+02/26/2022 01:05:08 - INFO - codeparrot_training - Step 36697: {'lr': 8.892320520755107e-05, 'samples': 18789376, 'steps': 36697, 'loss/train': 1.6497316360473633}
+02/26/2022 01:05:14 - INFO - codeparrot_training - Step 36698: {'lr': 8.891069208505256e-05, 'samples': 18789888, 'steps': 36698, 'loss/train': 1.2243497371673584}
+02/26/2022 01:05:17 - INFO - codeparrot_training - Step 36699: {'lr': 8.88981796526096e-05, 'samples': 18790400, 'steps': 36699, 'loss/train': 2.0028512477874756}
+02/26/2022 01:05:23 - INFO - codeparrot_training - Step 36700: {'lr': 8.88856679102757e-05, 'samples': 18790912, 'steps': 36700, 'loss/train': 1.0738894939422607}
+02/26/2022 01:05:27 - INFO - codeparrot_training - Step 36701: {'lr': 8.887315685810437e-05, 'samples': 18791424, 'steps': 36701, 'loss/train': 0.9110260605812073}
+02/26/2022 01:05:32 - INFO - codeparrot_training - Step 36702: {'lr': 8.886064649614924e-05, 'samples': 18791936, 'steps': 36702, 'loss/train': 1.0963810682296753}
+02/26/2022 01:05:36 - INFO - codeparrot_training - Step 36703: {'lr': 8.884813682446402e-05, 'samples': 18792448, 'steps': 36703, 'loss/train': 1.3647912740707397}
+02/26/2022 01:05:41 - INFO - codeparrot_training - Step 36704: {'lr': 8.88356278431022e-05, 'samples': 18792960, 'steps': 36704, 'loss/train': 1.3876621723175049}
+02/26/2022 01:05:45 - INFO - codeparrot_training - Step 36705: {'lr': 8.882311955211736e-05, 'samples': 18793472, 'steps': 36705, 'loss/train': 0.4826810657978058}
+02/26/2022 01:05:50 - INFO - codeparrot_training - Step 36706: {'lr': 8.881061195156303e-05, 'samples': 18793984, 'steps': 36706, 'loss/train': 3.0438857078552246}
+02/26/2022 01:05:56 - INFO - codeparrot_training - Step 36707: {'lr': 8.879810504149293e-05, 'samples': 18794496, 'steps': 36707, 'loss/train': 1.1649420261383057}
+02/26/2022 01:05:59 - INFO - codeparrot_training - Step 36708: {'lr': 8.878559882196058e-05, 'samples': 18795008, 'steps': 36708, 'loss/train': 1.2484266757965088}
+02/26/2022 01:06:06 - INFO - codeparrot_training - Step 36709: {'lr': 8.87730932930195e-05, 'samples': 18795520, 'steps': 36709, 'loss/train': 2.1707355976104736}
+02/26/2022 01:06:09 - INFO - codeparrot_training - Step 36710: {'lr': 8.876058845472323e-05, 'samples': 18796032, 'steps': 36710, 'loss/train': 2.2836008071899414}
+02/26/2022 01:06:15 - INFO - codeparrot_training - Step 36711: {'lr': 8.874808430712547e-05, 'samples': 18796544, 'steps': 36711, 'loss/train': 0.06388260424137115}
+02/26/2022 01:06:18 - INFO - codeparrot_training - Step 36712: {'lr': 8.873558085027975e-05, 'samples': 18797056, 'steps': 36712, 'loss/train': 1.3427910804748535}
+02/26/2022 01:06:24 - INFO - codeparrot_training - Step 36713: {'lr': 8.872307808423954e-05, 'samples': 18797568, 'steps': 36713, 'loss/train': 1.4716030359268188}
+02/26/2022 01:06:27 - INFO - codeparrot_training - Step 36714: {'lr': 8.871057600905841e-05, 'samples': 18798080, 'steps': 36714, 'loss/train': 1.0129289627075195}
+02/26/2022 01:06:33 - INFO - codeparrot_training - Step 36715: {'lr': 8.869807462479001e-05, 'samples': 18798592, 'steps': 36715, 'loss/train': 1.4471043348312378}
+02/26/2022 01:06:36 - INFO - codeparrot_training - Step 36716: {'lr': 8.868557393148787e-05, 'samples': 18799104, 'steps': 36716, 'loss/train': 2.0507829189300537}
+02/26/2022 01:06:42 - INFO - codeparrot_training - Step 36717: {'lr': 8.867307392920551e-05, 'samples': 18799616, 'steps': 36717, 'loss/train': 0.945280134677887}
+02/26/2022 01:06:45 - INFO - codeparrot_training - Step 36718: {'lr': 8.866057461799637e-05, 'samples': 18800128, 'steps': 36718, 'loss/train': 1.4961342811584473}
+02/26/2022 01:06:51 - INFO - codeparrot_training - Step 36719: {'lr': 8.864807599791419e-05, 'samples': 18800640, 'steps': 36719, 'loss/train': 2.179086446762085}
+02/26/2022 01:06:54 - INFO - codeparrot_training - Step 36720: {'lr': 8.863557806901232e-05, 'samples': 18801152, 'steps': 36720, 'loss/train': 0.6030438542366028}
+02/26/2022 01:07:00 - INFO - codeparrot_training - Step 36721: {'lr': 8.862308083134457e-05, 'samples': 18801664, 'steps': 36721, 'loss/train': 3.1088778972625732}
+02/26/2022 01:07:03 - INFO - codeparrot_training - Step 36722: {'lr': 8.861058428496413e-05, 'samples': 18802176, 'steps': 36722, 'loss/train': 1.686672329902649}
+02/26/2022 01:07:10 - INFO - codeparrot_training - Step 36723: {'lr': 8.859808842992478e-05, 'samples': 18802688, 'steps': 36723, 'loss/train': 1.7433375120162964}
+02/26/2022 01:07:13 - INFO - codeparrot_training - Step 36724: {'lr': 8.858559326627991e-05, 'samples': 18803200, 'steps': 36724, 'loss/train': 1.9560959339141846}
+02/26/2022 01:07:19 - INFO - codeparrot_training - Step 36725: {'lr': 8.857309879408323e-05, 'samples': 18803712, 'steps': 36725, 'loss/train': 0.5793426632881165}
+02/26/2022 01:07:23 - INFO - codeparrot_training - Step 36726: {'lr': 8.8560605013388e-05, 'samples': 18804224, 'steps': 36726, 'loss/train': 1.6196162700653076}
+02/26/2022 01:07:28 - INFO - codeparrot_training - Step 36727: {'lr': 8.854811192424797e-05, 'samples': 18804736, 'steps': 36727, 'loss/train': 2.195920467376709}
+02/26/2022 01:07:32 - INFO - codeparrot_training - Step 36728: {'lr': 8.853561952671646e-05, 'samples': 18805248, 'steps': 36728, 'loss/train': 2.598487615585327}
+02/26/2022 01:07:37 - INFO - codeparrot_training - Step 36729: {'lr': 8.852312782084723e-05, 'samples': 18805760, 'steps': 36729, 'loss/train': 1.9568482637405396}
+02/26/2022 01:07:41 - INFO - codeparrot_training - Step 36730: {'lr': 8.85106368066935e-05, 'samples': 18806272, 'steps': 36730, 'loss/train': 2.3036105632781982}
+02/26/2022 01:07:46 - INFO - codeparrot_training - Step 36731: {'lr': 8.849814648430901e-05, 'samples': 18806784, 'steps': 36731, 'loss/train': 1.8940600156784058}
+02/26/2022 01:07:50 - INFO - codeparrot_training - Step 36732: {'lr': 8.84856568537471e-05, 'samples': 18807296, 'steps': 36732, 'loss/train': 0.8471187949180603}
+02/26/2022 01:07:55 - INFO - codeparrot_training - Step 36733: {'lr': 8.84731679150614e-05, 'samples': 18807808, 'steps': 36733, 'loss/train': 1.2096806764602661}
+02/26/2022 01:07:59 - INFO - codeparrot_training - Step 36734: {'lr': 8.84606796683054e-05, 'samples': 18808320, 'steps': 36734, 'loss/train': 1.26979660987854}
+02/26/2022 01:08:05 - INFO - codeparrot_training - Step 36735: {'lr': 8.844819211353252e-05, 'samples': 18808832, 'steps': 36735, 'loss/train': 0.9659348726272583}
+02/26/2022 01:08:09 - INFO - codeparrot_training - Step 36736: {'lr': 8.843570525079623e-05, 'samples': 18809344, 'steps': 36736, 'loss/train': 1.104225754737854}
+02/26/2022 01:08:14 - INFO - codeparrot_training - Step 36737: {'lr': 8.842321908015017e-05, 'samples': 18809856, 'steps': 36737, 'loss/train': 2.7877025604248047}
+02/26/2022 01:08:18 - INFO - codeparrot_training - Step 36738: {'lr': 8.841073360164769e-05, 'samples': 18810368, 'steps': 36738, 'loss/train': 1.5737988948822021}
+02/26/2022 01:08:23 - INFO - codeparrot_training - Step 36739: {'lr': 8.839824881534234e-05, 'samples': 18810880, 'steps': 36739, 'loss/train': 0.966422975063324}
+02/26/2022 01:08:27 - INFO - codeparrot_training - Step 36740: {'lr': 8.838576472128757e-05, 'samples': 18811392, 'steps': 36740, 'loss/train': 1.9636882543563843}
+02/26/2022 01:08:32 - INFO - codeparrot_training - Step 36741: {'lr': 8.837328131953679e-05, 'samples': 18811904, 'steps': 36741, 'loss/train': 1.583749771118164}
+02/26/2022 01:08:36 - INFO - codeparrot_training - Step 36742: {'lr': 8.836079861014365e-05, 'samples': 18812416, 'steps': 36742, 'loss/train': 2.1364831924438477}
+02/26/2022 01:08:41 - INFO - codeparrot_training - Step 36743: {'lr': 8.83483165931615e-05, 'samples': 18812928, 'steps': 36743, 'loss/train': 1.0170072317123413}
+02/26/2022 01:08:45 - INFO - codeparrot_training - Step 36744: {'lr': 8.833583526864383e-05, 'samples': 18813440, 'steps': 36744, 'loss/train': 1.1249419450759888}
+02/26/2022 01:08:52 - INFO - codeparrot_training - Step 36745: {'lr': 8.832335463664404e-05, 'samples': 18813952, 'steps': 36745, 'loss/train': 1.7335306406021118}
+02/26/2022 01:08:55 - INFO - codeparrot_training - Step 36746: {'lr': 8.831087469721571e-05, 'samples': 18814464, 'steps': 36746, 'loss/train': 1.799824595451355}
+02/26/2022 01:09:01 - INFO - codeparrot_training - Step 36747: {'lr': 8.82983954504123e-05, 'samples': 18814976, 'steps': 36747, 'loss/train': 1.6073668003082275}
+02/26/2022 01:09:04 - INFO - codeparrot_training - Step 36748: {'lr': 8.828591689628718e-05, 'samples': 18815488, 'steps': 36748, 'loss/train': 1.3555233478546143}
+02/26/2022 01:09:10 - INFO - codeparrot_training - Step 36749: {'lr': 8.827343903489377e-05, 'samples': 18816000, 'steps': 36749, 'loss/train': 1.284003734588623}
+02/26/2022 01:09:13 - INFO - codeparrot_training - Step 36750: {'lr': 8.826096186628569e-05, 'samples': 18816512, 'steps': 36750, 'loss/train': 1.3173454999923706}
+02/26/2022 01:09:19 - INFO - codeparrot_training - Step 36751: {'lr': 8.824848539051627e-05, 'samples': 18817024, 'steps': 36751, 'loss/train': 1.413901448249817}
+02/26/2022 01:09:22 - INFO - codeparrot_training - Step 36752: {'lr': 8.8236009607639e-05, 'samples': 18817536, 'steps': 36752, 'loss/train': 1.5600212812423706}
+02/26/2022 01:09:27 - INFO - codeparrot_training - Step 36753: {'lr': 8.822353451770718e-05, 'samples': 18818048, 'steps': 36753, 'loss/train': 0.9365037083625793}
+02/26/2022 01:09:31 - INFO - codeparrot_training - Step 36754: {'lr': 8.821106012077449e-05, 'samples': 18818560, 'steps': 36754, 'loss/train': 1.6110575199127197}
+02/26/2022 01:09:37 - INFO - codeparrot_training - Step 36755: {'lr': 8.819858641689413e-05, 'samples': 18819072, 'steps': 36755, 'loss/train': 0.7587934136390686}
+02/26/2022 01:09:41 - INFO - codeparrot_training - Step 36756: {'lr': 8.818611340611985e-05, 'samples': 18819584, 'steps': 36756, 'loss/train': 2.1823596954345703}
+02/26/2022 01:09:46 - INFO - codeparrot_training - Step 36757: {'lr': 8.817364108850468e-05, 'samples': 18820096, 'steps': 36757, 'loss/train': 0.9404318928718567}
+02/26/2022 01:09:50 - INFO - codeparrot_training - Step 36758: {'lr': 8.816116946410232e-05, 'samples': 18820608, 'steps': 36758, 'loss/train': 3.9288594722747803}
+02/26/2022 01:09:55 - INFO - codeparrot_training - Step 36759: {'lr': 8.814869853296609e-05, 'samples': 18821120, 'steps': 36759, 'loss/train': 1.5254809856414795}
+02/26/2022 01:09:59 - INFO - codeparrot_training - Step 36760: {'lr': 8.813622829514956e-05, 'samples': 18821632, 'steps': 36760, 'loss/train': 1.4608752727508545}
+02/26/2022 01:10:04 - INFO - codeparrot_training - Step 36761: {'lr': 8.812375875070588e-05, 'samples': 18822144, 'steps': 36761, 'loss/train': 1.009140968322754}
+02/26/2022 01:10:08 - INFO - codeparrot_training - Step 36762: {'lr': 8.81112898996887e-05, 'samples': 18822656, 'steps': 36762, 'loss/train': 1.6667149066925049}
+02/26/2022 01:10:13 - INFO - codeparrot_training - Step 36763: {'lr': 8.809882174215125e-05, 'samples': 18823168, 'steps': 36763, 'loss/train': 1.8655192852020264}
+02/26/2022 01:10:17 - INFO - codeparrot_training - Step 36764: {'lr': 8.808635427814723e-05, 'samples': 18823680, 'steps': 36764, 'loss/train': 1.089234471321106}
+02/26/2022 01:10:22 - INFO - codeparrot_training - Step 36765: {'lr': 8.807388750772968e-05, 'samples': 18824192, 'steps': 36765, 'loss/train': 1.9756888151168823}
+02/26/2022 01:10:26 - INFO - codeparrot_training - Step 36766: {'lr': 8.806142143095228e-05, 'samples': 18824704, 'steps': 36766, 'loss/train': 1.5609052181243896}
+02/26/2022 01:10:31 - INFO - codeparrot_training - Step 36767: {'lr': 8.80489560478682e-05, 'samples': 18825216, 'steps': 36767, 'loss/train': 1.8127546310424805}
+02/26/2022 01:10:35 - INFO - codeparrot_training - Step 36768: {'lr': 8.803649135853109e-05, 'samples': 18825728, 'steps': 36768, 'loss/train': 1.8154958486557007}
+02/26/2022 01:10:40 - INFO - codeparrot_training - Step 36769: {'lr': 8.802402736299419e-05, 'samples': 18826240, 'steps': 36769, 'loss/train': 1.4463987350463867}
+02/26/2022 01:10:47 - INFO - codeparrot_training - Step 36770: {'lr': 8.801156406131095e-05, 'samples': 18826752, 'steps': 36770, 'loss/train': 0.7040348649024963}
+02/26/2022 01:10:50 - INFO - codeparrot_training - Step 36771: {'lr': 8.799910145353466e-05, 'samples': 18827264, 'steps': 36771, 'loss/train': 0.79520183801651}
+02/26/2022 01:10:54 - INFO - codeparrot_training - Step 36772: {'lr': 8.798663953971884e-05, 'samples': 18827776, 'steps': 36772, 'loss/train': 1.9044753313064575}
+02/26/2022 01:10:59 - INFO - codeparrot_training - Step 36773: {'lr': 8.79741783199168e-05, 'samples': 18828288, 'steps': 36773, 'loss/train': 1.3131247758865356}
+02/26/2022 01:11:03 - INFO - codeparrot_training - Step 36774: {'lr': 8.796171779418194e-05, 'samples': 18828800, 'steps': 36774, 'loss/train': 1.9512815475463867}
+02/26/2022 01:11:09 - INFO - codeparrot_training - Step 36775: {'lr': 8.794925796256756e-05, 'samples': 18829312, 'steps': 36775, 'loss/train': 1.364993929862976}
+02/26/2022 01:11:12 - INFO - codeparrot_training - Step 36776: {'lr': 8.793679882512717e-05, 'samples': 18829824, 'steps': 36776, 'loss/train': 0.08782120048999786}
+02/26/2022 01:11:18 - INFO - codeparrot_training - Step 36777: {'lr': 8.792434038191408e-05, 'samples': 18830336, 'steps': 36777, 'loss/train': 1.3393462896347046}
+02/26/2022 01:11:21 - INFO - codeparrot_training - Step 36778: {'lr': 8.791188263298164e-05, 'samples': 18830848, 'steps': 36778, 'loss/train': 1.8954521417617798}
+02/26/2022 01:11:26 - INFO - codeparrot_training - Step 36779: {'lr': 8.789942557838316e-05, 'samples': 18831360, 'steps': 36779, 'loss/train': 1.1856162548065186}
+02/26/2022 01:11:33 - INFO - codeparrot_training - Step 36780: {'lr': 8.788696921817213e-05, 'samples': 18831872, 'steps': 36780, 'loss/train': 2.0256221294403076}
+02/26/2022 01:11:37 - INFO - codeparrot_training - Step 36781: {'lr': 8.787451355240186e-05, 'samples': 18832384, 'steps': 36781, 'loss/train': 1.7585585117340088}
+02/26/2022 01:11:40 - INFO - codeparrot_training - Step 36782: {'lr': 8.78620585811257e-05, 'samples': 18832896, 'steps': 36782, 'loss/train': 2.2875008583068848}
+02/26/2022 01:11:46 - INFO - codeparrot_training - Step 36783: {'lr': 8.784960430439689e-05, 'samples': 18833408, 'steps': 36783, 'loss/train': 2.3009443283081055}
+02/26/2022 01:11:49 - INFO - codeparrot_training - Step 36784: {'lr': 8.7837150722269e-05, 'samples': 18833920, 'steps': 36784, 'loss/train': 3.1083948612213135}
+02/26/2022 01:11:55 - INFO - codeparrot_training - Step 36785: {'lr': 8.782469783479524e-05, 'samples': 18834432, 'steps': 36785, 'loss/train': 1.133135199546814}
+02/26/2022 01:11:58 - INFO - codeparrot_training - Step 36786: {'lr': 8.7812245642029e-05, 'samples': 18834944, 'steps': 36786, 'loss/train': 2.467921733856201}
+02/26/2022 01:12:04 - INFO - codeparrot_training - Step 36787: {'lr': 8.779979414402359e-05, 'samples': 18835456, 'steps': 36787, 'loss/train': 2.1750290393829346}
+02/26/2022 01:12:07 - INFO - codeparrot_training - Step 36788: {'lr': 8.778734334083226e-05, 'samples': 18835968, 'steps': 36788, 'loss/train': 2.211529016494751}
+02/26/2022 01:12:13 - INFO - codeparrot_training - Step 36789: {'lr': 8.777489323250854e-05, 'samples': 18836480, 'steps': 36789, 'loss/train': 1.9000332355499268}
+02/26/2022 01:12:16 - INFO - codeparrot_training - Step 36790: {'lr': 8.776244381910567e-05, 'samples': 18836992, 'steps': 36790, 'loss/train': 1.1563711166381836}
+02/26/2022 01:12:22 - INFO - codeparrot_training - Step 36791: {'lr': 8.774999510067697e-05, 'samples': 18837504, 'steps': 36791, 'loss/train': 1.0664424896240234}
+02/26/2022 01:12:26 - INFO - codeparrot_training - Step 36792: {'lr': 8.773754707727568e-05, 'samples': 18838016, 'steps': 36792, 'loss/train': 1.1581929922103882}
+02/26/2022 01:12:31 - INFO - codeparrot_training - Step 36793: {'lr': 8.772509974895529e-05, 'samples': 18838528, 'steps': 36793, 'loss/train': 2.2678403854370117}
+02/26/2022 01:12:35 - INFO - codeparrot_training - Step 36794: {'lr': 8.771265311576898e-05, 'samples': 18839040, 'steps': 36794, 'loss/train': 2.21399188041687}
+02/26/2022 01:12:40 - INFO - codeparrot_training - Step 36795: {'lr': 8.77002071777703e-05, 'samples': 18839552, 'steps': 36795, 'loss/train': 1.5858330726623535}
+02/26/2022 01:12:44 - INFO - codeparrot_training - Step 36796: {'lr': 8.768776193501221e-05, 'samples': 18840064, 'steps': 36796, 'loss/train': 1.632371425628662}
+02/26/2022 01:12:49 - INFO - codeparrot_training - Step 36797: {'lr': 8.767531738754831e-05, 'samples': 18840576, 'steps': 36797, 'loss/train': 1.3312007188796997}
+02/26/2022 01:12:53 - INFO - codeparrot_training - Step 36798: {'lr': 8.766287353543174e-05, 'samples': 18841088, 'steps': 36798, 'loss/train': 0.6778756976127625}
+02/26/2022 01:12:58 - INFO - codeparrot_training - Step 36799: {'lr': 8.765043037871601e-05, 'samples': 18841600, 'steps': 36799, 'loss/train': 2.2973077297210693}
+02/26/2022 01:13:02 - INFO - codeparrot_training - Step 36800: {'lr': 8.763798791745412e-05, 'samples': 18842112, 'steps': 36800, 'loss/train': 1.3411606550216675}
+02/26/2022 01:13:08 - INFO - codeparrot_training - Step 36801: {'lr': 8.762554615169959e-05, 'samples': 18842624, 'steps': 36801, 'loss/train': 1.8979246616363525}
+02/26/2022 01:13:11 - INFO - codeparrot_training - Step 36802: {'lr': 8.761310508150563e-05, 'samples': 18843136, 'steps': 36802, 'loss/train': 2.303459405899048}
+02/26/2022 01:13:17 - INFO - codeparrot_training - Step 36803: {'lr': 8.760066470692563e-05, 'samples': 18843648, 'steps': 36803, 'loss/train': 1.838030219078064}
+02/26/2022 01:13:20 - INFO - codeparrot_training - Step 36804: {'lr': 8.758822502801281e-05, 'samples': 18844160, 'steps': 36804, 'loss/train': 2.050367593765259}
+02/26/2022 01:13:26 - INFO - codeparrot_training - Step 36805: {'lr': 8.757578604482045e-05, 'samples': 18844672, 'steps': 36805, 'loss/train': 1.5041393041610718}
+02/26/2022 01:13:30 - INFO - codeparrot_training - Step 36806: {'lr': 8.756334775740176e-05, 'samples': 18845184, 'steps': 36806, 'loss/train': 2.2302026748657227}
+02/26/2022 01:13:35 - INFO - codeparrot_training - Step 36807: {'lr': 8.75509101658102e-05, 'samples': 18845696, 'steps': 36807, 'loss/train': 1.511582612991333}
+02/26/2022 01:13:39 - INFO - codeparrot_training - Step 36808: {'lr': 8.753847327009897e-05, 'samples': 18846208, 'steps': 36808, 'loss/train': 0.8627186417579651}
+02/26/2022 01:13:44 - INFO - codeparrot_training - Step 36809: {'lr': 8.752603707032133e-05, 'samples': 18846720, 'steps': 36809, 'loss/train': 1.1052284240722656}
+02/26/2022 01:13:48 - INFO - codeparrot_training - Step 36810: {'lr': 8.751360156653049e-05, 'samples': 18847232, 'steps': 36810, 'loss/train': 1.7874600887298584}
+02/26/2022 01:13:53 - INFO - codeparrot_training - Step 36811: {'lr': 8.750116675877984e-05, 'samples': 18847744, 'steps': 36811, 'loss/train': 2.403961658477783}
+02/26/2022 01:13:57 - INFO - codeparrot_training - Step 36812: {'lr': 8.748873264712259e-05, 'samples': 18848256, 'steps': 36812, 'loss/train': 0.9178921580314636}
+02/26/2022 01:14:00 - INFO - codeparrot_training - Step 36813: {'lr': 8.747629923161205e-05, 'samples': 18848768, 'steps': 36813, 'loss/train': 1.6454524993896484}
+02/26/2022 01:14:06 - INFO - codeparrot_training - Step 36814: {'lr': 8.746386651230131e-05, 'samples': 18849280, 'steps': 36814, 'loss/train': 0.9535394906997681}
+02/26/2022 01:14:11 - INFO - codeparrot_training - Step 36815: {'lr': 8.745143448924383e-05, 'samples': 18849792, 'steps': 36815, 'loss/train': 1.245305061340332}
+02/26/2022 01:14:15 - INFO - codeparrot_training - Step 36816: {'lr': 8.743900316249273e-05, 'samples': 18850304, 'steps': 36816, 'loss/train': 1.4459348917007446}
+02/26/2022 01:14:21 - INFO - codeparrot_training - Step 36817: {'lr': 8.742657253210151e-05, 'samples': 18850816, 'steps': 36817, 'loss/train': 2.023543357849121}
+02/26/2022 01:14:24 - INFO - codeparrot_training - Step 36818: {'lr': 8.741414259812302e-05, 'samples': 18851328, 'steps': 36818, 'loss/train': 1.181449055671692}
+02/26/2022 01:14:30 - INFO - codeparrot_training - Step 36819: {'lr': 8.740171336061082e-05, 'samples': 18851840, 'steps': 36819, 'loss/train': 1.7929778099060059}
+02/26/2022 01:14:33 - INFO - codeparrot_training - Step 36820: {'lr': 8.738928481961806e-05, 'samples': 18852352, 'steps': 36820, 'loss/train': 1.7656954526901245}
+02/26/2022 01:14:37 - INFO - codeparrot_training - Step 36821: {'lr': 8.737685697519796e-05, 'samples': 18852864, 'steps': 36821, 'loss/train': 2.5808839797973633}
+02/26/2022 01:14:42 - INFO - codeparrot_training - Step 36822: {'lr': 8.736442982740366e-05, 'samples': 18853376, 'steps': 36822, 'loss/train': 1.9175353050231934}
+02/26/2022 01:14:48 - INFO - codeparrot_training - Step 36823: {'lr': 8.73520033762886e-05, 'samples': 18853888, 'steps': 36823, 'loss/train': 1.0916659832000732}
+02/26/2022 01:14:51 - INFO - codeparrot_training - Step 36824: {'lr': 8.733957762190593e-05, 'samples': 18854400, 'steps': 36824, 'loss/train': 2.4224956035614014}
+02/26/2022 01:14:57 - INFO - codeparrot_training - Step 36825: {'lr': 8.732715256430884e-05, 'samples': 18854912, 'steps': 36825, 'loss/train': 0.5221654772758484}
+02/26/2022 01:15:00 - INFO - codeparrot_training - Step 36826: {'lr': 8.731472820355049e-05, 'samples': 18855424, 'steps': 36826, 'loss/train': 0.7488956451416016}
+02/26/2022 01:15:07 - INFO - codeparrot_training - Step 36827: {'lr': 8.730230453968427e-05, 'samples': 18855936, 'steps': 36827, 'loss/train': 1.9354008436203003}
+02/26/2022 01:15:10 - INFO - codeparrot_training - Step 36828: {'lr': 8.728988157276333e-05, 'samples': 18856448, 'steps': 36828, 'loss/train': 1.7492464780807495}
+02/26/2022 01:15:16 - INFO - codeparrot_training - Step 36829: {'lr': 8.727745930284075e-05, 'samples': 18856960, 'steps': 36829, 'loss/train': 2.198659896850586}
+02/26/2022 01:15:19 - INFO - codeparrot_training - Step 36830: {'lr': 8.726503772997e-05, 'samples': 18857472, 'steps': 36830, 'loss/train': 0.5789958834648132}
+02/26/2022 01:15:24 - INFO - codeparrot_training - Step 36831: {'lr': 8.725261685420411e-05, 'samples': 18857984, 'steps': 36831, 'loss/train': 2.080350875854492}
+02/26/2022 01:15:28 - INFO - codeparrot_training - Step 36832: {'lr': 8.724019667559635e-05, 'samples': 18858496, 'steps': 36832, 'loss/train': 0.3867916166782379}
+02/26/2022 01:15:34 - INFO - codeparrot_training - Step 36833: {'lr': 8.722777719419984e-05, 'samples': 18859008, 'steps': 36833, 'loss/train': 2.3656091690063477}
+02/26/2022 01:15:37 - INFO - codeparrot_training - Step 36834: {'lr': 8.721535841006798e-05, 'samples': 18859520, 'steps': 36834, 'loss/train': 1.1518471240997314}
+02/26/2022 01:15:43 - INFO - codeparrot_training - Step 36835: {'lr': 8.720294032325368e-05, 'samples': 18860032, 'steps': 36835, 'loss/train': 2.0127668380737305}
+02/26/2022 01:15:46 - INFO - codeparrot_training - Step 36836: {'lr': 8.719052293381036e-05, 'samples': 18860544, 'steps': 36836, 'loss/train': 2.148756265640259}
+02/26/2022 01:15:52 - INFO - codeparrot_training - Step 36837: {'lr': 8.717810624179109e-05, 'samples': 18861056, 'steps': 36837, 'loss/train': 1.2076683044433594}
+02/26/2022 01:15:56 - INFO - codeparrot_training - Step 36838: {'lr': 8.716569024724925e-05, 'samples': 18861568, 'steps': 36838, 'loss/train': 3.046046257019043}
+02/26/2022 01:16:01 - INFO - codeparrot_training - Step 36839: {'lr': 8.715327495023773e-05, 'samples': 18862080, 'steps': 36839, 'loss/train': 1.2962663173675537}
+02/26/2022 01:16:05 - INFO - codeparrot_training - Step 36840: {'lr': 8.714086035080996e-05, 'samples': 18862592, 'steps': 36840, 'loss/train': 1.513439416885376}
+02/26/2022 01:16:11 - INFO - codeparrot_training - Step 36841: {'lr': 8.712844644901893e-05, 'samples': 18863104, 'steps': 36841, 'loss/train': 1.9734536409378052}
+02/26/2022 01:16:14 - INFO - codeparrot_training - Step 36842: {'lr': 8.711603324491802e-05, 'samples': 18863616, 'steps': 36842, 'loss/train': 1.1246708631515503}
+02/26/2022 01:16:20 - INFO - codeparrot_training - Step 36843: {'lr': 8.71036207385603e-05, 'samples': 18864128, 'steps': 36843, 'loss/train': 1.082332968711853}
+02/26/2022 01:16:23 - INFO - codeparrot_training - Step 36844: {'lr': 8.709120892999895e-05, 'samples': 18864640, 'steps': 36844, 'loss/train': 2.752357244491577}
+02/26/2022 01:16:29 - INFO - codeparrot_training - Step 36845: {'lr': 8.707879781928702e-05, 'samples': 18865152, 'steps': 36845, 'loss/train': 1.7152692079544067}
+02/26/2022 01:16:32 - INFO - codeparrot_training - Step 36846: {'lr': 8.70663874064779e-05, 'samples': 18865664, 'steps': 36846, 'loss/train': 1.0590592622756958}
+02/26/2022 01:16:39 - INFO - codeparrot_training - Step 36847: {'lr': 8.705397769162462e-05, 'samples': 18866176, 'steps': 36847, 'loss/train': 2.0598013401031494}
+02/26/2022 01:16:42 - INFO - codeparrot_training - Step 36848: {'lr': 8.704156867478036e-05, 'samples': 18866688, 'steps': 36848, 'loss/train': 1.2578845024108887}
+02/26/2022 01:16:48 - INFO - codeparrot_training - Step 36849: {'lr': 8.70291603559982e-05, 'samples': 18867200, 'steps': 36849, 'loss/train': 1.9242167472839355}
+02/26/2022 01:16:51 - INFO - codeparrot_training - Step 36850: {'lr': 8.701675273533145e-05, 'samples': 18867712, 'steps': 36850, 'loss/train': 2.022484302520752}
+02/26/2022 01:16:57 - INFO - codeparrot_training - Step 36851: {'lr': 8.700434581283309e-05, 'samples': 18868224, 'steps': 36851, 'loss/train': 1.3832817077636719}
+02/26/2022 01:17:00 - INFO - codeparrot_training - Step 36852: {'lr': 8.699193958855653e-05, 'samples': 18868736, 'steps': 36852, 'loss/train': 1.7992764711380005}
+02/26/2022 01:17:06 - INFO - codeparrot_training - Step 36853: {'lr': 8.697953406255458e-05, 'samples': 18869248, 'steps': 36853, 'loss/train': 0.9926881194114685}
+02/26/2022 01:17:09 - INFO - codeparrot_training - Step 36854: {'lr': 8.696712923488059e-05, 'samples': 18869760, 'steps': 36854, 'loss/train': 2.096977472305298}
+02/26/2022 01:17:15 - INFO - codeparrot_training - Step 36855: {'lr': 8.69547251055876e-05, 'samples': 18870272, 'steps': 36855, 'loss/train': 2.0804827213287354}
+02/26/2022 01:17:18 - INFO - codeparrot_training - Step 36856: {'lr': 8.694232167472896e-05, 'samples': 18870784, 'steps': 36856, 'loss/train': 1.3874249458312988}
+02/26/2022 01:17:24 - INFO - codeparrot_training - Step 36857: {'lr': 8.692991894235744e-05, 'samples': 18871296, 'steps': 36857, 'loss/train': 2.0656490325927734}
+02/26/2022 01:17:27 - INFO - codeparrot_training - Step 36858: {'lr': 8.691751690852647e-05, 'samples': 18871808, 'steps': 36858, 'loss/train': 1.6641432046890259}
+02/26/2022 01:17:33 - INFO - codeparrot_training - Step 36859: {'lr': 8.690511557328901e-05, 'samples': 18872320, 'steps': 36859, 'loss/train': 2.2094168663024902}
+02/26/2022 01:17:37 - INFO - codeparrot_training - Step 36860: {'lr': 8.689271493669836e-05, 'samples': 18872832, 'steps': 36860, 'loss/train': 3.599497079849243}
+02/26/2022 01:17:40 - INFO - codeparrot_training - Step 36861: {'lr': 8.688031499880741e-05, 'samples': 18873344, 'steps': 36861, 'loss/train': 1.0254249572753906}
+02/26/2022 01:17:46 - INFO - codeparrot_training - Step 36862: {'lr': 8.686791575966946e-05, 'samples': 18873856, 'steps': 36862, 'loss/train': 4.905345916748047}
+02/26/2022 01:17:49 - INFO - codeparrot_training - Step 36863: {'lr': 8.685551721933749e-05, 'samples': 18874368, 'steps': 36863, 'loss/train': 1.6642621755599976}
+02/26/2022 01:17:55 - INFO - codeparrot_training - Step 36864: {'lr': 8.684311937786474e-05, 'samples': 18874880, 'steps': 36864, 'loss/train': 2.1865453720092773}
+02/26/2022 01:17:59 - INFO - codeparrot_training - Step 36865: {'lr': 8.683072223530427e-05, 'samples': 18875392, 'steps': 36865, 'loss/train': 0.7889824509620667}
+02/26/2022 01:18:04 - INFO - codeparrot_training - Step 36866: {'lr': 8.681832579170917e-05, 'samples': 18875904, 'steps': 36866, 'loss/train': 1.3934391736984253}
+02/26/2022 01:18:08 - INFO - codeparrot_training - Step 36867: {'lr': 8.680593004713255e-05, 'samples': 18876416, 'steps': 36867, 'loss/train': 1.8681923151016235}
+02/26/2022 01:18:13 - INFO - codeparrot_training - Step 36868: {'lr': 8.679353500162743e-05, 'samples': 18876928, 'steps': 36868, 'loss/train': 0.5904476642608643}
+02/26/2022 01:18:17 - INFO - codeparrot_training - Step 36869: {'lr': 8.678114065524706e-05, 'samples': 18877440, 'steps': 36869, 'loss/train': 1.245560646057129}
+02/26/2022 01:18:22 - INFO - codeparrot_training - Step 36870: {'lr': 8.676874700804443e-05, 'samples': 18877952, 'steps': 36870, 'loss/train': 2.2014451026916504}
+02/26/2022 01:18:26 - INFO - codeparrot_training - Step 36871: {'lr': 8.675635406007268e-05, 'samples': 18878464, 'steps': 36871, 'loss/train': 1.8092398643493652}
+02/26/2022 01:18:31 - INFO - codeparrot_training - Step 36872: {'lr': 8.67439618113848e-05, 'samples': 18878976, 'steps': 36872, 'loss/train': 2.4203686714172363}
+02/26/2022 01:18:35 - INFO - codeparrot_training - Step 36873: {'lr': 8.673157026203401e-05, 'samples': 18879488, 'steps': 36873, 'loss/train': 1.5392042398452759}
+02/26/2022 01:18:41 - INFO - codeparrot_training - Step 36874: {'lr': 8.671917941207335e-05, 'samples': 18880000, 'steps': 36874, 'loss/train': 1.6053953170776367}
+02/26/2022 01:18:44 - INFO - codeparrot_training - Step 36875: {'lr': 8.670678926155587e-05, 'samples': 18880512, 'steps': 36875, 'loss/train': 2.9327821731567383}
+02/26/2022 01:18:50 - INFO - codeparrot_training - Step 36876: {'lr': 8.669439981053457e-05, 'samples': 18881024, 'steps': 36876, 'loss/train': 1.974166750907898}
+02/26/2022 01:18:54 - INFO - codeparrot_training - Step 36877: {'lr': 8.668201105906268e-05, 'samples': 18881536, 'steps': 36877, 'loss/train': 0.8197054862976074}
+02/26/2022 01:19:00 - INFO - codeparrot_training - Step 36878: {'lr': 8.666962300719319e-05, 'samples': 18882048, 'steps': 36878, 'loss/train': 1.4822155237197876}
+02/26/2022 01:19:03 - INFO - codeparrot_training - Step 36879: {'lr': 8.665723565497918e-05, 'samples': 18882560, 'steps': 36879, 'loss/train': 1.7086328268051147}
+02/26/2022 01:19:09 - INFO - codeparrot_training - Step 36880: {'lr': 8.664484900247364e-05, 'samples': 18883072, 'steps': 36880, 'loss/train': 1.2913063764572144}
+02/26/2022 01:19:12 - INFO - codeparrot_training - Step 36881: {'lr': 8.663246304972976e-05, 'samples': 18883584, 'steps': 36881, 'loss/train': 1.814562201499939}
+02/26/2022 01:19:18 - INFO - codeparrot_training - Step 36882: {'lr': 8.662007779680053e-05, 'samples': 18884096, 'steps': 36882, 'loss/train': 2.63313889503479}
+02/26/2022 01:19:21 - INFO - codeparrot_training - Step 36883: {'lr': 8.660769324373902e-05, 'samples': 18884608, 'steps': 36883, 'loss/train': 2.911949634552002}
+02/26/2022 01:19:27 - INFO - codeparrot_training - Step 36884: {'lr': 8.659530939059818e-05, 'samples': 18885120, 'steps': 36884, 'loss/train': 1.2410475015640259}
+02/26/2022 01:19:31 - INFO - codeparrot_training - Step 36885: {'lr': 8.658292623743124e-05, 'samples': 18885632, 'steps': 36885, 'loss/train': 2.073704242706299}
+02/26/2022 01:19:36 - INFO - codeparrot_training - Step 36886: {'lr': 8.657054378429113e-05, 'samples': 18886144, 'steps': 36886, 'loss/train': 1.582241177558899}
+02/26/2022 01:19:40 - INFO - codeparrot_training - Step 36887: {'lr': 8.655816203123096e-05, 'samples': 18886656, 'steps': 36887, 'loss/train': 1.8512474298477173}
+02/26/2022 01:19:45 - INFO - codeparrot_training - Step 36888: {'lr': 8.65457809783036e-05, 'samples': 18887168, 'steps': 36888, 'loss/train': 0.6571885943412781}
+02/26/2022 01:19:49 - INFO - codeparrot_training - Step 36889: {'lr': 8.653340062556231e-05, 'samples': 18887680, 'steps': 36889, 'loss/train': 1.2929080724716187}
+02/26/2022 01:19:55 - INFO - codeparrot_training - Step 36890: {'lr': 8.652102097305994e-05, 'samples': 18888192, 'steps': 36890, 'loss/train': 0.9289174675941467}
+02/26/2022 01:19:58 - INFO - codeparrot_training - Step 36891: {'lr': 8.650864202084977e-05, 'samples': 18888704, 'steps': 36891, 'loss/train': 1.818631887435913}
+02/26/2022 01:20:01 - INFO - codeparrot_training - Step 36892: {'lr': 8.649626376898451e-05, 'samples': 18889216, 'steps': 36892, 'loss/train': 0.7039647102355957}
+02/26/2022 01:20:07 - INFO - codeparrot_training - Step 36893: {'lr': 8.64838862175174e-05, 'samples': 18889728, 'steps': 36893, 'loss/train': 1.8207097053527832}
+02/26/2022 01:20:11 - INFO - codeparrot_training - Step 36894: {'lr': 8.647150936650134e-05, 'samples': 18890240, 'steps': 36894, 'loss/train': 2.6992299556732178}
+02/26/2022 01:20:16 - INFO - codeparrot_training - Step 36895: {'lr': 8.645913321598956e-05, 'samples': 18890752, 'steps': 36895, 'loss/train': 1.1495805978775024}
+02/26/2022 01:20:20 - INFO - codeparrot_training - Step 36896: {'lr': 8.644675776603475e-05, 'samples': 18891264, 'steps': 36896, 'loss/train': 0.7655107378959656}
+02/26/2022 01:20:26 - INFO - codeparrot_training - Step 36897: {'lr': 8.643438301669019e-05, 'samples': 18891776, 'steps': 36897, 'loss/train': 2.3657946586608887}
+02/26/2022 01:20:29 - INFO - codeparrot_training - Step 36898: {'lr': 8.64220089680087e-05, 'samples': 18892288, 'steps': 36898, 'loss/train': 1.6852672100067139}
+02/26/2022 01:20:35 - INFO - codeparrot_training - Step 36899: {'lr': 8.640963562004356e-05, 'samples': 18892800, 'steps': 36899, 'loss/train': 2.520089864730835}
+02/26/2022 01:20:38 - INFO - codeparrot_training - Step 36900: {'lr': 8.639726297284742e-05, 'samples': 18893312, 'steps': 36900, 'loss/train': 1.196007251739502}
+02/26/2022 01:20:44 - INFO - codeparrot_training - Step 36901: {'lr': 8.638489102647354e-05, 'samples': 18893824, 'steps': 36901, 'loss/train': 2.1488497257232666}
+02/26/2022 01:20:47 - INFO - codeparrot_training - Step 36902: {'lr': 8.637251978097476e-05, 'samples': 18894336, 'steps': 36902, 'loss/train': 2.2404894828796387}
+02/26/2022 01:20:53 - INFO - codeparrot_training - Step 36903: {'lr': 8.636014923640423e-05, 'samples': 18894848, 'steps': 36903, 'loss/train': 2.3863933086395264}
+02/26/2022 01:20:56 - INFO - codeparrot_training - Step 36904: {'lr': 8.634777939281485e-05, 'samples': 18895360, 'steps': 36904, 'loss/train': 1.0175422430038452}
+02/26/2022 01:21:02 - INFO - codeparrot_training - Step 36905: {'lr': 8.633541025025962e-05, 'samples': 18895872, 'steps': 36905, 'loss/train': 1.083966612815857}
+02/26/2022 01:21:05 - INFO - codeparrot_training - Step 36906: {'lr': 8.632304180879145e-05, 'samples': 18896384, 'steps': 36906, 'loss/train': 1.6595673561096191}
+02/26/2022 01:21:11 - INFO - codeparrot_training - Step 36907: {'lr': 8.631067406846346e-05, 'samples': 18896896, 'steps': 36907, 'loss/train': 1.9447115659713745}
+02/26/2022 01:21:14 - INFO - codeparrot_training - Step 36908: {'lr': 8.629830702932856e-05, 'samples': 18897408, 'steps': 36908, 'loss/train': 1.941929578781128}
+02/26/2022 01:21:20 - INFO - codeparrot_training - Step 36909: {'lr': 8.628594069143975e-05, 'samples': 18897920, 'steps': 36909, 'loss/train': 1.0319911241531372}
+02/26/2022 01:21:26 - INFO - codeparrot_training - Step 36910: {'lr': 8.62735750548499e-05, 'samples': 18898432, 'steps': 36910, 'loss/train': 1.553390383720398}
+02/26/2022 01:21:29 - INFO - codeparrot_training - Step 36911: {'lr': 8.626121011961216e-05, 'samples': 18898944, 'steps': 36911, 'loss/train': 0.03581497445702553}
+02/26/2022 01:21:35 - INFO - codeparrot_training - Step 36912: {'lr': 8.624884588577938e-05, 'samples': 18899456, 'steps': 36912, 'loss/train': 1.3789877891540527}
+02/26/2022 01:21:38 - INFO - codeparrot_training - Step 36913: {'lr': 8.623648235340457e-05, 'samples': 18899968, 'steps': 36913, 'loss/train': 1.325390338897705}
+02/26/2022 01:21:44 - INFO - codeparrot_training - Step 36914: {'lr': 8.622411952254055e-05, 'samples': 18900480, 'steps': 36914, 'loss/train': 1.1682175397872925}
+02/26/2022 01:21:47 - INFO - codeparrot_training - Step 36915: {'lr': 8.621175739324052e-05, 'samples': 18900992, 'steps': 36915, 'loss/train': 1.6988774538040161}
+02/26/2022 01:21:53 - INFO - codeparrot_training - Step 36916: {'lr': 8.619939596555729e-05, 'samples': 18901504, 'steps': 36916, 'loss/train': 1.4303603172302246}
+02/26/2022 01:21:56 - INFO - codeparrot_training - Step 36917: {'lr': 8.618703523954385e-05, 'samples': 18902016, 'steps': 36917, 'loss/train': 0.629348635673523}
+02/26/2022 01:22:02 - INFO - codeparrot_training - Step 36918: {'lr': 8.617467521525313e-05, 'samples': 18902528, 'steps': 36918, 'loss/train': 2.0985968112945557}
+02/26/2022 01:22:05 - INFO - codeparrot_training - Step 36919: {'lr': 8.616231589273799e-05, 'samples': 18903040, 'steps': 36919, 'loss/train': 1.2436678409576416}
+02/26/2022 01:22:12 - INFO - codeparrot_training - Step 36920: {'lr': 8.614995727205155e-05, 'samples': 18903552, 'steps': 36920, 'loss/train': 1.4784159660339355}
+02/26/2022 01:22:15 - INFO - codeparrot_training - Step 36921: {'lr': 8.613759935324667e-05, 'samples': 18904064, 'steps': 36921, 'loss/train': 1.9755879640579224}
+02/26/2022 01:22:20 - INFO - codeparrot_training - Step 36922: {'lr': 8.612524213637626e-05, 'samples': 18904576, 'steps': 36922, 'loss/train': 1.7446744441986084}
+02/26/2022 01:22:24 - INFO - codeparrot_training - Step 36923: {'lr': 8.611288562149322e-05, 'samples': 18905088, 'steps': 36923, 'loss/train': 1.886895775794983}
+02/26/2022 01:22:30 - INFO - codeparrot_training - Step 36924: {'lr': 8.610052980865062e-05, 'samples': 18905600, 'steps': 36924, 'loss/train': 2.2142117023468018}
+02/26/2022 01:22:33 - INFO - codeparrot_training - Step 36925: {'lr': 8.608817469790123e-05, 'samples': 18906112, 'steps': 36925, 'loss/train': 2.1308860778808594}
+02/26/2022 01:22:39 - INFO - codeparrot_training - Step 36926: {'lr': 8.607582028929819e-05, 'samples': 18906624, 'steps': 36926, 'loss/train': 1.9334596395492554}
+02/26/2022 01:22:42 - INFO - codeparrot_training - Step 36927: {'lr': 8.606346658289413e-05, 'samples': 18907136, 'steps': 36927, 'loss/train': 1.5525330305099487}
+02/26/2022 01:22:48 - INFO - codeparrot_training - Step 36928: {'lr': 8.60511135787422e-05, 'samples': 18907648, 'steps': 36928, 'loss/train': 0.7826526761054993}
+02/26/2022 01:22:51 - INFO - codeparrot_training - Step 36929: {'lr': 8.603876127689517e-05, 'samples': 18908160, 'steps': 36929, 'loss/train': 1.7909104824066162}
+02/26/2022 01:22:57 - INFO - codeparrot_training - Step 36930: {'lr': 8.602640967740618e-05, 'samples': 18908672, 'steps': 36930, 'loss/train': 3.16336989402771}
+02/26/2022 01:23:00 - INFO - codeparrot_training - Step 36931: {'lr': 8.601405878032783e-05, 'samples': 18909184, 'steps': 36931, 'loss/train': 1.7063947916030884}
+02/26/2022 01:23:06 - INFO - codeparrot_training - Step 36932: {'lr': 8.600170858571326e-05, 'samples': 18909696, 'steps': 36932, 'loss/train': 1.1706607341766357}
+02/26/2022 01:23:10 - INFO - codeparrot_training - Step 36933: {'lr': 8.598935909361521e-05, 'samples': 18910208, 'steps': 36933, 'loss/train': 1.4866220951080322}
+02/26/2022 01:23:15 - INFO - codeparrot_training - Step 36934: {'lr': 8.597701030408683e-05, 'samples': 18910720, 'steps': 36934, 'loss/train': 1.1321985721588135}
+02/26/2022 01:23:19 - INFO - codeparrot_training - Step 36935: {'lr': 8.596466221718069e-05, 'samples': 18911232, 'steps': 36935, 'loss/train': 1.9462562799453735}
+02/26/2022 01:23:24 - INFO - codeparrot_training - Step 36936: {'lr': 8.595231483294993e-05, 'samples': 18911744, 'steps': 36936, 'loss/train': 1.0940117835998535}
+02/26/2022 01:23:28 - INFO - codeparrot_training - Step 36937: {'lr': 8.593996815144728e-05, 'samples': 18912256, 'steps': 36937, 'loss/train': 2.583690643310547}
+02/26/2022 01:23:33 - INFO - codeparrot_training - Step 36938: {'lr': 8.59276221727258e-05, 'samples': 18912768, 'steps': 36938, 'loss/train': 1.597976565361023}
+02/26/2022 01:23:37 - INFO - codeparrot_training - Step 36939: {'lr': 8.591527689683828e-05, 'samples': 18913280, 'steps': 36939, 'loss/train': 1.8345625400543213}
+02/26/2022 01:23:42 - INFO - codeparrot_training - Step 36940: {'lr': 8.590293232383762e-05, 'samples': 18913792, 'steps': 36940, 'loss/train': 1.2770590782165527}
+02/26/2022 01:23:45 - INFO - codeparrot_training - Step 36941: {'lr': 8.58905884537766e-05, 'samples': 18914304, 'steps': 36941, 'loss/train': 0.027937322854995728}
+02/26/2022 01:23:51 - INFO - codeparrot_training - Step 36942: {'lr': 8.587824528670826e-05, 'samples': 18914816, 'steps': 36942, 'loss/train': 2.7899630069732666}
+02/26/2022 01:23:54 - INFO - codeparrot_training - Step 36943: {'lr': 8.586590282268542e-05, 'samples': 18915328, 'steps': 36943, 'loss/train': 1.5152673721313477}
+02/26/2022 01:24:00 - INFO - codeparrot_training - Step 36944: {'lr': 8.585356106176093e-05, 'samples': 18915840, 'steps': 36944, 'loss/train': 1.5155937671661377}
+02/26/2022 01:24:03 - INFO - codeparrot_training - Step 36945: {'lr': 8.584122000398758e-05, 'samples': 18916352, 'steps': 36945, 'loss/train': 1.8368003368377686}
+02/26/2022 01:24:10 - INFO - codeparrot_training - Step 36946: {'lr': 8.58288796494184e-05, 'samples': 18916864, 'steps': 36946, 'loss/train': 1.3719526529312134}
+02/26/2022 01:24:13 - INFO - codeparrot_training - Step 36947: {'lr': 8.581653999810617e-05, 'samples': 18917376, 'steps': 36947, 'loss/train': 1.2114650011062622}
+02/26/2022 01:24:19 - INFO - codeparrot_training - Step 36948: {'lr': 8.580420105010375e-05, 'samples': 18917888, 'steps': 36948, 'loss/train': 2.3056693077087402}
+02/26/2022 01:24:22 - INFO - codeparrot_training - Step 36949: {'lr': 8.579186280546392e-05, 'samples': 18918400, 'steps': 36949, 'loss/train': 1.5895808935165405}
+02/26/2022 01:24:28 - INFO - codeparrot_training - Step 36950: {'lr': 8.577952526423969e-05, 'samples': 18918912, 'steps': 36950, 'loss/train': 1.3941559791564941}
+02/26/2022 01:24:31 - INFO - codeparrot_training - Step 36951: {'lr': 8.576718842648379e-05, 'samples': 18919424, 'steps': 36951, 'loss/train': 1.3350311517715454}
+02/26/2022 01:24:37 - INFO - codeparrot_training - Step 36952: {'lr': 8.575485229224914e-05, 'samples': 18919936, 'steps': 36952, 'loss/train': 0.5477108359336853}
+02/26/2022 01:24:40 - INFO - codeparrot_training - Step 36953: {'lr': 8.574251686158846e-05, 'samples': 18920448, 'steps': 36953, 'loss/train': 1.8919907808303833}
+02/26/2022 01:24:46 - INFO - codeparrot_training - Step 36954: {'lr': 8.573018213455477e-05, 'samples': 18920960, 'steps': 36954, 'loss/train': 2.1822316646575928}
+02/26/2022 01:24:49 - INFO - codeparrot_training - Step 36955: {'lr': 8.571784811120081e-05, 'samples': 18921472, 'steps': 36955, 'loss/train': 2.1003291606903076}
+02/26/2022 01:24:55 - INFO - codeparrot_training - Step 36956: {'lr': 8.570551479157942e-05, 'samples': 18921984, 'steps': 36956, 'loss/train': 2.2563424110412598}
+02/26/2022 01:24:58 - INFO - codeparrot_training - Step 36957: {'lr': 8.569318217574334e-05, 'samples': 18922496, 'steps': 36957, 'loss/train': 1.9402010440826416}
+02/26/2022 01:25:04 - INFO - codeparrot_training - Step 36958: {'lr': 8.568085026374561e-05, 'samples': 18923008, 'steps': 36958, 'loss/train': 1.490246295928955}
+02/26/2022 01:25:08 - INFO - codeparrot_training - Step 36959: {'lr': 8.56685190556389e-05, 'samples': 18923520, 'steps': 36959, 'loss/train': 1.853089451789856}
+02/26/2022 01:25:13 - INFO - codeparrot_training - Step 36960: {'lr': 8.565618855147602e-05, 'samples': 18924032, 'steps': 36960, 'loss/train': 1.6487221717834473}
+02/26/2022 01:25:17 - INFO - codeparrot_training - Step 36961: {'lr': 8.564385875130992e-05, 'samples': 18924544, 'steps': 36961, 'loss/train': 0.9966661930084229}
+02/26/2022 01:25:22 - INFO - codeparrot_training - Step 36962: {'lr': 8.563152965519336e-05, 'samples': 18925056, 'steps': 36962, 'loss/train': 1.129966378211975}
+02/26/2022 01:25:26 - INFO - codeparrot_training - Step 36963: {'lr': 8.561920126317909e-05, 'samples': 18925568, 'steps': 36963, 'loss/train': 0.6994410157203674}
+02/26/2022 01:25:31 - INFO - codeparrot_training - Step 36964: {'lr': 8.560687357531993e-05, 'samples': 18926080, 'steps': 36964, 'loss/train': 1.9713150262832642}
+02/26/2022 01:25:35 - INFO - codeparrot_training - Step 36965: {'lr': 8.559454659166888e-05, 'samples': 18926592, 'steps': 36965, 'loss/train': 0.9345439672470093}
+02/26/2022 01:25:40 - INFO - codeparrot_training - Step 36966: {'lr': 8.558222031227838e-05, 'samples': 18927104, 'steps': 36966, 'loss/train': 2.3028151988983154}
+02/26/2022 01:25:47 - INFO - codeparrot_training - Step 36967: {'lr': 8.556989473720156e-05, 'samples': 18927616, 'steps': 36967, 'loss/train': 2.0832371711730957}
+02/26/2022 01:25:50 - INFO - codeparrot_training - Step 36968: {'lr': 8.555756986649099e-05, 'samples': 18928128, 'steps': 36968, 'loss/train': 1.4604074954986572}
+02/26/2022 01:25:53 - INFO - codeparrot_training - Step 36969: {'lr': 8.554524570019978e-05, 'samples': 18928640, 'steps': 36969, 'loss/train': 1.4824813604354858}
+02/26/2022 01:25:59 - INFO - codeparrot_training - Step 36970: {'lr': 8.553292223838033e-05, 'samples': 18929152, 'steps': 36970, 'loss/train': 1.9281446933746338}
+02/26/2022 01:26:05 - INFO - codeparrot_training - Step 36971: {'lr': 8.552059948108567e-05, 'samples': 18929664, 'steps': 36971, 'loss/train': 1.9268885850906372}
+02/26/2022 01:26:08 - INFO - codeparrot_training - Step 36972: {'lr': 8.550827742836851e-05, 'samples': 18930176, 'steps': 36972, 'loss/train': 1.6829442977905273}
+02/26/2022 01:26:14 - INFO - codeparrot_training - Step 36973: {'lr': 8.549595608028168e-05, 'samples': 18930688, 'steps': 36973, 'loss/train': 1.2781786918640137}
+02/26/2022 01:26:17 - INFO - codeparrot_training - Step 36974: {'lr': 8.548363543687798e-05, 'samples': 18931200, 'steps': 36974, 'loss/train': 2.119123697280884}
+02/26/2022 01:26:23 - INFO - codeparrot_training - Step 36975: {'lr': 8.547131549821014e-05, 'samples': 18931712, 'steps': 36975, 'loss/train': 0.3199837803840637}
+02/26/2022 01:26:26 - INFO - codeparrot_training - Step 36976: {'lr': 8.545899626433085e-05, 'samples': 18932224, 'steps': 36976, 'loss/train': 2.2008097171783447}
+02/26/2022 01:26:32 - INFO - codeparrot_training - Step 36977: {'lr': 8.544667773529305e-05, 'samples': 18932736, 'steps': 36977, 'loss/train': 1.7141129970550537}
+02/26/2022 01:26:36 - INFO - codeparrot_training - Step 36978: {'lr': 8.543435991114942e-05, 'samples': 18933248, 'steps': 36978, 'loss/train': 1.878537654876709}
+02/26/2022 01:26:41 - INFO - codeparrot_training - Step 36979: {'lr': 8.542204279195276e-05, 'samples': 18933760, 'steps': 36979, 'loss/train': 1.8371224403381348}
+02/26/2022 01:26:45 - INFO - codeparrot_training - Step 36980: {'lr': 8.540972637775571e-05, 'samples': 18934272, 'steps': 36980, 'loss/train': 1.2525502443313599}
+02/26/2022 01:26:50 - INFO - codeparrot_training - Step 36981: {'lr': 8.539741066861123e-05, 'samples': 18934784, 'steps': 36981, 'loss/train': 1.5267858505249023}
+02/26/2022 01:26:54 - INFO - codeparrot_training - Step 36982: {'lr': 8.538509566457198e-05, 'samples': 18935296, 'steps': 36982, 'loss/train': 1.4636080265045166}
+02/26/2022 01:26:59 - INFO - codeparrot_training - Step 36983: {'lr': 8.537278136569068e-05, 'samples': 18935808, 'steps': 36983, 'loss/train': 1.436458706855774}
+02/26/2022 01:27:03 - INFO - codeparrot_training - Step 36984: {'lr': 8.536046777202003e-05, 'samples': 18936320, 'steps': 36984, 'loss/train': 1.3397808074951172}
+02/26/2022 01:27:08 - INFO - codeparrot_training - Step 36985: {'lr': 8.534815488361297e-05, 'samples': 18936832, 'steps': 36985, 'loss/train': 1.1611132621765137}
+02/26/2022 01:27:12 - INFO - codeparrot_training - Step 36986: {'lr': 8.533584270052203e-05, 'samples': 18937344, 'steps': 36986, 'loss/train': 1.9433910846710205}
+02/26/2022 01:27:17 - INFO - codeparrot_training - Step 36987: {'lr': 8.53235312228002e-05, 'samples': 18937856, 'steps': 36987, 'loss/train': 1.7604655027389526}
+02/26/2022 01:27:21 - INFO - codeparrot_training - Step 36988: {'lr': 8.531122045049993e-05, 'samples': 18938368, 'steps': 36988, 'loss/train': 2.3688790798187256}
+02/26/2022 01:27:27 - INFO - codeparrot_training - Step 36989: {'lr': 8.529891038367418e-05, 'samples': 18938880, 'steps': 36989, 'loss/train': 0.8820657730102539}
+02/26/2022 01:27:30 - INFO - codeparrot_training - Step 36990: {'lr': 8.528660102237551e-05, 'samples': 18939392, 'steps': 36990, 'loss/train': 1.9192264080047607}
+02/26/2022 01:27:34 - INFO - codeparrot_training - Step 36991: {'lr': 8.52742923666569e-05, 'samples': 18939904, 'steps': 36991, 'loss/train': 1.7725179195404053}
+02/26/2022 01:27:39 - INFO - codeparrot_training - Step 36992: {'lr': 8.526198441657077e-05, 'samples': 18940416, 'steps': 36992, 'loss/train': 2.414793014526367}
+02/26/2022 01:27:43 - INFO - codeparrot_training - Step 36993: {'lr': 8.524967717217006e-05, 'samples': 18940928, 'steps': 36993, 'loss/train': 2.566577434539795}
+02/26/2022 01:27:49 - INFO - codeparrot_training - Step 36994: {'lr': 8.523737063350733e-05, 'samples': 18941440, 'steps': 36994, 'loss/train': 1.655949354171753}
+02/26/2022 01:27:53 - INFO - codeparrot_training - Step 36995: {'lr': 8.522506480063555e-05, 'samples': 18941952, 'steps': 36995, 'loss/train': 1.919616937637329}
+02/26/2022 01:27:58 - INFO - codeparrot_training - Step 36996: {'lr': 8.521275967360711e-05, 'samples': 18942464, 'steps': 36996, 'loss/train': 2.190977096557617}
+02/26/2022 01:28:02 - INFO - codeparrot_training - Step 36997: {'lr': 8.520045525247497e-05, 'samples': 18942976, 'steps': 36997, 'loss/train': 0.4945140480995178}
+02/26/2022 01:28:07 - INFO - codeparrot_training - Step 36998: {'lr': 8.518815153729173e-05, 'samples': 18943488, 'steps': 36998, 'loss/train': 2.316406011581421}
+02/26/2022 01:28:11 - INFO - codeparrot_training - Step 36999: {'lr': 8.517584852811005e-05, 'samples': 18944000, 'steps': 36999, 'loss/train': 1.4211903810501099}
+02/26/2022 01:28:11 - INFO - codeparrot_training - Evaluating and saving model checkpoint