diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -30482,3 +30482,1009 @@ Use FP16 precision: False
 02/25/2022 16:26:13 - INFO - codeparrot_training - Step 29998: {'lr': 0.00018532684912737663, 'samples': 15359488, 'steps': 29998, 'loss/train': 2.062704086303711}
 02/25/2022 16:26:18 - INFO - codeparrot_training - Step 29999: {'lr': 0.00018531104378731973, 'samples': 15360000, 'steps': 29999, 'loss/train': 1.6037700176239014}
 02/25/2022 16:26:18 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/25/2022 16:26:36 - WARNING - huggingface_hub.repository - Several commits (30) will be pushed upstream.
+02/25/2022 16:26:36 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/25/2022 16:27:10 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   95f22fa..af58905  floral-grass-11 -> floral-grass-11
+
+02/25/2022 16:27:15 - INFO - codeparrot_training - Step 30000: {'lr': 0.0001852952387243698, 'samples': 15360512, 'steps': 30000, 'loss/train': 2.1731338500976562}
+02/25/2022 16:27:21 - INFO - codeparrot_training - Step 30001: {'lr': 0.0001852794339385946, 'samples': 15361024, 'steps': 30001, 'loss/train': 1.118362307548523}
+02/25/2022 16:27:24 - INFO - codeparrot_training - Step 30002: {'lr': 0.0001852636294300617, 'samples': 15361536, 'steps': 30002, 'loss/train': 1.3688808679580688}
+02/25/2022 16:27:30 - INFO - codeparrot_training - Step 30003: {'lr': 0.00018524782519883882, 'samples': 15362048, 'steps': 30003, 'loss/train': 2.811722993850708}
+02/25/2022 16:27:34 - INFO - codeparrot_training - Step 30004: {'lr': 0.00018523202124499368, 'samples': 15362560, 'steps': 30004, 'loss/train': 1.7936128377914429}
+02/25/2022 16:27:39 - INFO - codeparrot_training - Step 30005: {'lr': 0.00018521621756859404, 'samples': 15363072, 'steps': 30005, 'loss/train': 0.579941987991333}
+02/25/2022 16:27:43 - INFO - codeparrot_training - Step 30006: {'lr': 0.00018520041416970762, 'samples': 15363584, 'steps': 30006, 'loss/train': 2.729137420654297}
+02/25/2022 16:27:48 - INFO - codeparrot_training - Step 30007: {'lr': 0.00018518461104840195, 'samples': 15364096, 'steps': 30007, 'loss/train': 2.9498283863067627}
+02/25/2022 16:27:52 - INFO - codeparrot_training - Step 30008: {'lr': 0.00018516880820474484, 'samples': 15364608, 'steps': 30008, 'loss/train': 0.9807052612304688}
+02/25/2022 16:27:58 - INFO - codeparrot_training - Step 30009: {'lr': 0.000185153005638804, 'samples': 15365120, 'steps': 30009, 'loss/train': 2.155941963195801}
+02/25/2022 16:28:01 - INFO - codeparrot_training - Step 30010: {'lr': 0.00018513720335064715, 'samples': 15365632, 'steps': 30010, 'loss/train': 1.6149916648864746}
+02/25/2022 16:28:06 - INFO - codeparrot_training - Step 30011: {'lr': 0.00018512140134034182, 'samples': 15366144, 'steps': 30011, 'loss/train': 2.8154728412628174}
+02/25/2022 16:28:10 - INFO - codeparrot_training - Step 30012: {'lr': 0.0001851055996079558, 'samples': 15366656, 'steps': 30012, 'loss/train': 1.0076522827148438}
+02/25/2022 16:28:16 - INFO - codeparrot_training - Step 30013: {'lr': 0.0001850897981535568, 'samples': 15367168, 'steps': 30013, 'loss/train': 1.431535005569458}
+02/25/2022 16:28:20 - INFO - codeparrot_training - Step 30014: {'lr': 0.0001850739969772126, 'samples': 15367680, 'steps': 30014, 'loss/train': 0.21688517928123474}
+02/25/2022 16:28:25 - INFO - codeparrot_training - Step 30015: {'lr': 0.0001850581960789907, 'samples': 15368192, 'steps': 30015, 'loss/train': 2.0599780082702637}
+02/25/2022 16:28:29 - INFO - codeparrot_training - Step 30016: {'lr': 0.00018504239545895885, 'samples': 15368704, 'steps': 30016, 'loss/train': 0.7041217088699341}
+02/25/2022 16:28:34 - INFO - codeparrot_training - Step 30017: {'lr': 0.00018502659511718473, 'samples': 15369216, 'steps': 30017, 'loss/train': 1.148532748222351}
+02/25/2022 16:28:38 - INFO - codeparrot_training - Step 30018: {'lr': 0.00018501079505373614, 'samples': 15369728, 'steps': 30018, 'loss/train': 1.7754034996032715}
+02/25/2022 16:28:43 - INFO - codeparrot_training - Step 30019: {'lr': 0.00018499499526868063, 'samples': 15370240, 'steps': 30019, 'loss/train': 0.9794594645500183}
+02/25/2022 16:28:47 - INFO - codeparrot_training - Step 30020: {'lr': 0.00018497919576208587, 'samples': 15370752, 'steps': 30020, 'loss/train': 2.097980499267578}
+02/25/2022 16:28:53 - INFO - codeparrot_training - Step 30021: {'lr': 0.00018496339653401963, 'samples': 15371264, 'steps': 30021, 'loss/train': 1.9253982305526733}
+02/25/2022 16:28:56 - INFO - codeparrot_training - Step 30022: {'lr': 0.00018494759758454953, 'samples': 15371776, 'steps': 30022, 'loss/train': 2.1501739025115967}
+02/25/2022 16:29:02 - INFO - codeparrot_training - Step 30023: {'lr': 0.0001849317989137433, 'samples': 15372288, 'steps': 30023, 'loss/train': 0.7061108946800232}
+02/25/2022 16:29:06 - INFO - codeparrot_training - Step 30024: {'lr': 0.0001849160005216685, 'samples': 15372800, 'steps': 30024, 'loss/train': 1.9436016082763672}
+02/25/2022 16:29:11 - INFO - codeparrot_training - Step 30025: {'lr': 0.000184900202408393, 'samples': 15373312, 'steps': 30025, 'loss/train': 1.7332801818847656}
+02/25/2022 16:29:15 - INFO - codeparrot_training - Step 30026: {'lr': 0.00018488440457398425, 'samples': 15373824, 'steps': 30026, 'loss/train': 1.6052758693695068}
+02/25/2022 16:29:21 - INFO - codeparrot_training - Step 30027: {'lr': 0.00018486860701851014, 'samples': 15374336, 'steps': 30027, 'loss/train': 1.5934809446334839}
+02/25/2022 16:29:24 - INFO - codeparrot_training - Step 30028: {'lr': 0.00018485280974203823, 'samples': 15374848, 'steps': 30028, 'loss/train': 1.5203590393066406}
+02/25/2022 16:29:29 - INFO - codeparrot_training - Step 30029: {'lr': 0.0001848370127446361, 'samples': 15375360, 'steps': 30029, 'loss/train': 0.6706119775772095}
+02/25/2022 16:29:33 - INFO - codeparrot_training - Step 30030: {'lr': 0.00018482121602637154, 'samples': 15375872, 'steps': 30030, 'loss/train': 1.453978419303894}
+02/25/2022 16:29:38 - INFO - codeparrot_training - Step 30031: {'lr': 0.0001848054195873122, 'samples': 15376384, 'steps': 30031, 'loss/train': 1.6087207794189453}
+02/25/2022 16:29:42 - INFO - codeparrot_training - Step 30032: {'lr': 0.00018478962342752584, 'samples': 15376896, 'steps': 30032, 'loss/train': 1.5389703512191772}
+02/25/2022 16:29:47 - INFO - codeparrot_training - Step 30033: {'lr': 0.0001847738275470799, 'samples': 15377408, 'steps': 30033, 'loss/train': 1.95013427734375}
+02/25/2022 16:29:53 - INFO - codeparrot_training - Step 30034: {'lr': 0.0001847580319460422, 'samples': 15377920, 'steps': 30034, 'loss/train': 2.7690927982330322}
+02/25/2022 16:29:56 - INFO - codeparrot_training - Step 30035: {'lr': 0.00018474223662448037, 'samples': 15378432, 'steps': 30035, 'loss/train': 2.2924814224243164}
+02/25/2022 16:30:02 - INFO - codeparrot_training - Step 30036: {'lr': 0.00018472644158246219, 'samples': 15378944, 'steps': 30036, 'loss/train': 1.685707926750183}
+02/25/2022 16:30:05 - INFO - codeparrot_training - Step 30037: {'lr': 0.0001847106468200551, 'samples': 15379456, 'steps': 30037, 'loss/train': 0.7427926063537598}
+02/25/2022 16:30:11 - INFO - codeparrot_training - Step 30038: {'lr': 0.00018469485233732686, 'samples': 15379968, 'steps': 30038, 'loss/train': 0.8560056686401367}
+02/25/2022 16:30:15 - INFO - codeparrot_training - Step 30039: {'lr': 0.00018467905813434514, 'samples': 15380480, 'steps': 30039, 'loss/train': 2.108710765838623}
+02/25/2022 16:30:20 - INFO - codeparrot_training - Step 30040: {'lr': 0.00018466326421117768, 'samples': 15380992, 'steps': 30040, 'loss/train': 2.2106690406799316}
+02/25/2022 16:30:24 - INFO - codeparrot_training - Step 30041: {'lr': 0.00018464747056789194, 'samples': 15381504, 'steps': 30041, 'loss/train': 1.280234694480896}
+02/25/2022 16:30:29 - INFO - codeparrot_training - Step 30042: {'lr': 0.0001846316772045557, 'samples': 15382016, 'steps': 30042, 'loss/train': 0.9699698090553284}
+02/25/2022 16:30:33 - INFO - codeparrot_training - Step 30043: {'lr': 0.00018461588412123662, 'samples': 15382528, 'steps': 30043, 'loss/train': 1.8953702449798584}
+02/25/2022 16:30:39 - INFO - codeparrot_training - Step 30044: {'lr': 0.00018460009131800233, 'samples': 15383040, 'steps': 30044, 'loss/train': 1.057846188545227}
+02/25/2022 16:30:42 - INFO - codeparrot_training - Step 30045: {'lr': 0.00018458429879492056, 'samples': 15383552, 'steps': 30045, 'loss/train': 1.8462986946105957}
+02/25/2022 16:30:48 - INFO - codeparrot_training - Step 30046: {'lr': 0.00018456850655205876, 'samples': 15384064, 'steps': 30046, 'loss/train': 2.8983097076416016}
+02/25/2022 16:30:51 - INFO - codeparrot_training - Step 30047: {'lr': 0.00018455271458948474, 'samples': 15384576, 'steps': 30047, 'loss/train': 1.0029246807098389}
+02/25/2022 16:30:57 - INFO - codeparrot_training - Step 30048: {'lr': 0.00018453692290726608, 'samples': 15385088, 'steps': 30048, 'loss/train': 2.83892560005188}
+02/25/2022 16:31:01 - INFO - codeparrot_training - Step 30049: {'lr': 0.0001845211315054706, 'samples': 15385600, 'steps': 30049, 'loss/train': 1.5945707559585571}
+02/25/2022 16:31:06 - INFO - codeparrot_training - Step 30050: {'lr': 0.00018450534038416566, 'samples': 15386112, 'steps': 30050, 'loss/train': 1.7259442806243896}
+02/25/2022 16:31:10 - INFO - codeparrot_training - Step 30051: {'lr': 0.00018448954954341907, 'samples': 15386624, 'steps': 30051, 'loss/train': 1.0598602294921875}
+02/25/2022 16:31:15 - INFO - codeparrot_training - Step 30052: {'lr': 0.00018447375898329843, 'samples': 15387136, 'steps': 30052, 'loss/train': 2.3258752822875977}
+02/25/2022 16:31:19 - INFO - codeparrot_training - Step 30053: {'lr': 0.0001844579687038714, 'samples': 15387648, 'steps': 30053, 'loss/train': 1.9298633337020874}
+02/25/2022 16:31:24 - INFO - codeparrot_training - Step 30054: {'lr': 0.00018444217870520574, 'samples': 15388160, 'steps': 30054, 'loss/train': 1.5542007684707642}
+02/25/2022 16:31:28 - INFO - codeparrot_training - Step 30055: {'lr': 0.00018442638898736886, 'samples': 15388672, 'steps': 30055, 'loss/train': 0.4983118176460266}
+02/25/2022 16:31:33 - INFO - codeparrot_training - Step 30056: {'lr': 0.0001844105995504285, 'samples': 15389184, 'steps': 30056, 'loss/train': 1.224015474319458}
+02/25/2022 16:31:37 - INFO - codeparrot_training - Step 30057: {'lr': 0.00018439481039445228, 'samples': 15389696, 'steps': 30057, 'loss/train': 1.2572087049484253}
+02/25/2022 16:31:43 - INFO - codeparrot_training - Step 30058: {'lr': 0.000184379021519508, 'samples': 15390208, 'steps': 30058, 'loss/train': 1.927124261856079}
+02/25/2022 16:31:46 - INFO - codeparrot_training - Step 30059: {'lr': 0.00018436323292566305, 'samples': 15390720, 'steps': 30059, 'loss/train': 4.666149139404297}
+02/25/2022 16:31:52 - INFO - codeparrot_training - Step 30060: {'lr': 0.00018434744461298514, 'samples': 15391232, 'steps': 30060, 'loss/train': 1.5695821046829224}
+02/25/2022 16:31:56 - INFO - codeparrot_training - Step 30061: {'lr': 0.000184331656581542, 'samples': 15391744, 'steps': 30061, 'loss/train': 0.10898319631814957}
+02/25/2022 16:32:01 - INFO - codeparrot_training - Step 30062: {'lr': 0.0001843158688314012, 'samples': 15392256, 'steps': 30062, 'loss/train': 1.9409795999526978}
+02/25/2022 16:32:05 - INFO - codeparrot_training - Step 30063: {'lr': 0.0001843000813626303, 'samples': 15392768, 'steps': 30063, 'loss/train': 1.1655218601226807}
+02/25/2022 16:32:12 - INFO - codeparrot_training - Step 30064: {'lr': 0.000184284294175297, 'samples': 15393280, 'steps': 30064, 'loss/train': 1.15315842628479}
+02/25/2022 16:32:15 - INFO - codeparrot_training - Step 30065: {'lr': 0.00018426850726946896, 'samples': 15393792, 'steps': 30065, 'loss/train': 1.391595721244812}
+02/25/2022 16:32:21 - INFO - codeparrot_training - Step 30066: {'lr': 0.0001842527206452137, 'samples': 15394304, 'steps': 30066, 'loss/train': 0.05137182027101517}
+02/25/2022 16:32:24 - INFO - codeparrot_training - Step 30067: {'lr': 0.00018423693430259897, 'samples': 15394816, 'steps': 30067, 'loss/train': 1.3020931482315063}
+02/25/2022 16:32:30 - INFO - codeparrot_training - Step 30068: {'lr': 0.00018422114824169234, 'samples': 15395328, 'steps': 30068, 'loss/train': 2.1964986324310303}
+02/25/2022 16:32:33 - INFO - codeparrot_training - Step 30069: {'lr': 0.00018420536246256138, 'samples': 15395840, 'steps': 30069, 'loss/train': 1.3870564699172974}
+02/25/2022 16:32:39 - INFO - codeparrot_training - Step 30070: {'lr': 0.00018418957696527373, 'samples': 15396352, 'steps': 30070, 'loss/train': 0.6149616837501526}
+02/25/2022 16:32:42 - INFO - codeparrot_training - Step 30071: {'lr': 0.00018417379174989708, 'samples': 15396864, 'steps': 30071, 'loss/train': 2.8009674549102783}
+02/25/2022 16:32:48 - INFO - codeparrot_training - Step 30072: {'lr': 0.00018415800681649902, 'samples': 15397376, 'steps': 30072, 'loss/train': 1.3102842569351196}
+02/25/2022 16:32:51 - INFO - codeparrot_training - Step 30073: {'lr': 0.00018414222216514708, 'samples': 15397888, 'steps': 30073, 'loss/train': 2.7271907329559326}
+02/25/2022 16:32:58 - INFO - codeparrot_training - Step 30074: {'lr': 0.000184126437795909, 'samples': 15398400, 'steps': 30074, 'loss/train': 1.596235990524292}
+02/25/2022 16:33:01 - INFO - codeparrot_training - Step 30075: {'lr': 0.00018411065370885237, 'samples': 15398912, 'steps': 30075, 'loss/train': 1.2019269466400146}
+02/25/2022 16:33:07 - INFO - codeparrot_training - Step 30076: {'lr': 0.00018409486990404473, 'samples': 15399424, 'steps': 30076, 'loss/train': 1.3568345308303833}
+02/25/2022 16:33:10 - INFO - codeparrot_training - Step 30077: {'lr': 0.00018407908638155374, 'samples': 15399936, 'steps': 30077, 'loss/train': 2.513878583908081}
+02/25/2022 16:33:16 - INFO - codeparrot_training - Step 30078: {'lr': 0.000184063303141447, 'samples': 15400448, 'steps': 30078, 'loss/train': 1.8428399562835693}
+02/25/2022 16:33:19 - INFO - codeparrot_training - Step 30079: {'lr': 0.00018404752018379213, 'samples': 15400960, 'steps': 30079, 'loss/train': 1.758124828338623}
+02/25/2022 16:33:25 - INFO - codeparrot_training - Step 30080: {'lr': 0.00018403173750865685, 'samples': 15401472, 'steps': 30080, 'loss/train': 1.529345989227295}
+02/25/2022 16:33:28 - INFO - codeparrot_training - Step 30081: {'lr': 0.00018401595511610854, 'samples': 15401984, 'steps': 30081, 'loss/train': 1.8872168064117432}
+02/25/2022 16:33:34 - INFO - codeparrot_training - Step 30082: {'lr': 0.00018400017300621493, 'samples': 15402496, 'steps': 30082, 'loss/train': 1.0820785760879517}
+02/25/2022 16:33:37 - INFO - codeparrot_training - Step 30083: {'lr': 0.00018398439117904363, 'samples': 15403008, 'steps': 30083, 'loss/train': 2.262575626373291}
+02/25/2022 16:33:44 - INFO - codeparrot_training - Step 30084: {'lr': 0.00018396860963466234, 'samples': 15403520, 'steps': 30084, 'loss/train': 1.8471931219100952}
+02/25/2022 16:33:48 - INFO - codeparrot_training - Step 30085: {'lr': 0.00018395282837313845, 'samples': 15404032, 'steps': 30085, 'loss/train': 1.480950951576233}
+02/25/2022 16:33:53 - INFO - codeparrot_training - Step 30086: {'lr': 0.0001839370473945397, 'samples': 15404544, 'steps': 30086, 'loss/train': 2.2740652561187744}
+02/25/2022 16:33:57 - INFO - codeparrot_training - Step 30087: {'lr': 0.0001839212666989336, 'samples': 15405056, 'steps': 30087, 'loss/train': 1.5621719360351562}
+02/25/2022 16:34:02 - INFO - codeparrot_training - Step 30088: {'lr': 0.00018390548628638797, 'samples': 15405568, 'steps': 30088, 'loss/train': 0.9227740168571472}
+02/25/2022 16:34:06 - INFO - codeparrot_training - Step 30089: {'lr': 0.00018388970615697013, 'samples': 15406080, 'steps': 30089, 'loss/train': 1.8394383192062378}
+02/25/2022 16:34:12 - INFO - codeparrot_training - Step 30090: {'lr': 0.00018387392631074778, 'samples': 15406592, 'steps': 30090, 'loss/train': 1.1881377696990967}
+02/25/2022 16:34:15 - INFO - codeparrot_training - Step 30091: {'lr': 0.00018385814674778855, 'samples': 15407104, 'steps': 30091, 'loss/train': 1.5937190055847168}
+02/25/2022 16:34:21 - INFO - codeparrot_training - Step 30092: {'lr': 0.00018384236746816002, 'samples': 15407616, 'steps': 30092, 'loss/train': 1.8211866617202759}
+02/25/2022 16:34:24 - INFO - codeparrot_training - Step 30093: {'lr': 0.00018382658847192987, 'samples': 15408128, 'steps': 30093, 'loss/train': 0.9190847873687744}
+02/25/2022 16:34:30 - INFO - codeparrot_training - Step 30094: {'lr': 0.0001838108097591655, 'samples': 15408640, 'steps': 30094, 'loss/train': 2.038853406906128}
+02/25/2022 16:34:33 - INFO - codeparrot_training - Step 30095: {'lr': 0.00018379503132993463, 'samples': 15409152, 'steps': 30095, 'loss/train': 1.6370124816894531}
+02/25/2022 16:34:39 - INFO - codeparrot_training - Step 30096: {'lr': 0.00018377925318430478, 'samples': 15409664, 'steps': 30096, 'loss/train': 1.0914748907089233}
+02/25/2022 16:34:43 - INFO - codeparrot_training - Step 30097: {'lr': 0.00018376347532234372, 'samples': 15410176, 'steps': 30097, 'loss/train': 1.6376678943634033}
+02/25/2022 16:34:48 - INFO - codeparrot_training - Step 30098: {'lr': 0.00018374769774411878, 'samples': 15410688, 'steps': 30098, 'loss/train': 1.219774842262268}
+02/25/2022 16:34:52 - INFO - codeparrot_training - Step 30099: {'lr': 0.00018373192044969766, 'samples': 15411200, 'steps': 30099, 'loss/train': 0.8286318778991699}
+02/25/2022 16:34:57 - INFO - codeparrot_training - Step 30100: {'lr': 0.00018371614343914796, 'samples': 15411712, 'steps': 30100, 'loss/train': 2.9472908973693848}
+02/25/2022 16:35:01 - INFO - codeparrot_training - Step 30101: {'lr': 0.0001837003667125373, 'samples': 15412224, 'steps': 30101, 'loss/train': 1.432822823524475}
+02/25/2022 16:35:06 - INFO - codeparrot_training - Step 30102: {'lr': 0.00018368459026993323, 'samples': 15412736, 'steps': 30102, 'loss/train': 2.500108242034912}
+02/25/2022 16:35:10 - INFO - codeparrot_training - Step 30103: {'lr': 0.00018366881411140327, 'samples': 15413248, 'steps': 30103, 'loss/train': 2.1952555179595947}
+02/25/2022 16:35:15 - INFO - codeparrot_training - Step 30104: {'lr': 0.00018365303823701502, 'samples': 15413760, 'steps': 30104, 'loss/train': 1.3362709283828735}
+02/25/2022 16:35:19 - INFO - codeparrot_training - Step 30105: {'lr': 0.00018363726264683617, 'samples': 15414272, 'steps': 30105, 'loss/train': 2.6070902347564697}
+02/25/2022 16:35:25 - INFO - codeparrot_training - Step 30106: {'lr': 0.0001836214873409342, 'samples': 15414784, 'steps': 30106, 'loss/train': 1.4674382209777832}
+02/25/2022 16:35:28 - INFO - codeparrot_training - Step 30107: {'lr': 0.00018360571231937662, 'samples': 15415296, 'steps': 30107, 'loss/train': 1.2712163925170898}
+02/25/2022 16:35:34 - INFO - codeparrot_training - Step 30108: {'lr': 0.00018358993758223118, 'samples': 15415808, 'steps': 30108, 'loss/train': 8.703932762145996}
+02/25/2022 16:35:37 - INFO - codeparrot_training - Step 30109: {'lr': 0.0001835741631295653, 'samples': 15416320, 'steps': 30109, 'loss/train': 2.3264341354370117}
+02/25/2022 16:35:43 - INFO - codeparrot_training - Step 30110: {'lr': 0.00018355838896144666, 'samples': 15416832, 'steps': 30110, 'loss/train': 1.7961509227752686}
+02/25/2022 16:35:47 - INFO - codeparrot_training - Step 30111: {'lr': 0.00018354261507794273, 'samples': 15417344, 'steps': 30111, 'loss/train': 2.467033624649048}
+02/25/2022 16:35:52 - INFO - codeparrot_training - Step 30112: {'lr': 0.00018352684147912118, 'samples': 15417856, 'steps': 30112, 'loss/train': 0.7049136161804199}
+02/25/2022 16:35:55 - INFO - codeparrot_training - Step 30113: {'lr': 0.00018351106816504948, 'samples': 15418368, 'steps': 30113, 'loss/train': 0.9948156476020813}
+02/25/2022 16:36:01 - INFO - codeparrot_training - Step 30114: {'lr': 0.00018349529513579527, 'samples': 15418880, 'steps': 30114, 'loss/train': 1.4332759380340576}
+02/25/2022 16:36:04 - INFO - codeparrot_training - Step 30115: {'lr': 0.0001834795223914262, 'samples': 15419392, 'steps': 30115, 'loss/train': 1.420559287071228}
+02/25/2022 16:36:10 - INFO - codeparrot_training - Step 30116: {'lr': 0.0001834637499320096, 'samples': 15419904, 'steps': 30116, 'loss/train': 1.6374807357788086}
+02/25/2022 16:36:16 - INFO - codeparrot_training - Step 30117: {'lr': 0.00018344797775761318, 'samples': 15420416, 'steps': 30117, 'loss/train': 2.265371799468994}
+02/25/2022 16:36:19 - INFO - codeparrot_training - Step 30118: {'lr': 0.0001834322058683045, 'samples': 15420928, 'steps': 30118, 'loss/train': 2.8234853744506836}
+02/25/2022 16:36:25 - INFO - codeparrot_training - Step 30119: {'lr': 0.00018341643426415122, 'samples': 15421440, 'steps': 30119, 'loss/train': 1.768696904182434}
+02/25/2022 16:36:28 - INFO - codeparrot_training - Step 30120: {'lr': 0.00018340066294522068, 'samples': 15421952, 'steps': 30120, 'loss/train': 1.665058970451355}
+02/25/2022 16:36:35 - INFO - codeparrot_training - Step 30121: {'lr': 0.00018338489191158057, 'samples': 15422464, 'steps': 30121, 'loss/train': 0.6417982578277588}
+02/25/2022 16:36:38 - INFO - codeparrot_training - Step 30122: {'lr': 0.0001833691211632984, 'samples': 15422976, 'steps': 30122, 'loss/train': 2.398552894592285}
+02/25/2022 16:36:44 - INFO - codeparrot_training - Step 30123: {'lr': 0.0001833533507004419, 'samples': 15423488, 'steps': 30123, 'loss/train': 0.0898883044719696}
+02/25/2022 16:36:47 - INFO - codeparrot_training - Step 30124: {'lr': 0.00018333758052307834, 'samples': 15424000, 'steps': 30124, 'loss/train': 1.6271077394485474}
+02/25/2022 16:36:53 - INFO - codeparrot_training - Step 30125: {'lr': 0.00018332181063127542, 'samples': 15424512, 'steps': 30125, 'loss/train': 2.1353557109832764}
+02/25/2022 16:36:57 - INFO - codeparrot_training - Step 30126: {'lr': 0.00018330604102510074, 'samples': 15425024, 'steps': 30126, 'loss/train': 1.6372706890106201}
+02/25/2022 16:37:02 - INFO - codeparrot_training - Step 30127: {'lr': 0.00018329027170462176, 'samples': 15425536, 'steps': 30127, 'loss/train': 1.4246652126312256}
+02/25/2022 16:37:06 - INFO - codeparrot_training - Step 30128: {'lr': 0.00018327450266990617, 'samples': 15426048, 'steps': 30128, 'loss/train': 2.0891523361206055}
+02/25/2022 16:37:11 - INFO - codeparrot_training - Step 30129: {'lr': 0.00018325873392102131, 'samples': 15426560, 'steps': 30129, 'loss/train': 2.1926770210266113}
+02/25/2022 16:37:15 - INFO - codeparrot_training - Step 30130: {'lr': 0.00018324296545803488, 'samples': 15427072, 'steps': 30130, 'loss/train': 0.9406639933586121}
+02/25/2022 16:37:21 - INFO - codeparrot_training - Step 30131: {'lr': 0.00018322719728101439, 'samples': 15427584, 'steps': 30131, 'loss/train': 1.8344426155090332}
+02/25/2022 16:37:25 - INFO - codeparrot_training - Step 30132: {'lr': 0.00018321142939002743, 'samples': 15428096, 'steps': 30132, 'loss/train': 1.2943284511566162}
+02/25/2022 16:37:30 - INFO - codeparrot_training - Step 30133: {'lr': 0.00018319566178514145, 'samples': 15428608, 'steps': 30133, 'loss/train': 2.014268159866333}
+02/25/2022 16:37:34 - INFO - codeparrot_training - Step 30134: {'lr': 0.00018317989446642403, 'samples': 15429120, 'steps': 30134, 'loss/train': 0.475665807723999}
+02/25/2022 16:37:39 - INFO - codeparrot_training - Step 30135: {'lr': 0.00018316412743394272, 'samples': 15429632, 'steps': 30135, 'loss/train': 2.0615270137786865}
+02/25/2022 16:37:43 - INFO - codeparrot_training - Step 30136: {'lr': 0.0001831483606877652, 'samples': 15430144, 'steps': 30136, 'loss/train': 3.243621826171875}
+02/25/2022 16:37:48 - INFO - codeparrot_training - Step 30137: {'lr': 0.00018313259422795872, 'samples': 15430656, 'steps': 30137, 'loss/train': 1.5465216636657715}
+02/25/2022 16:37:52 - INFO - codeparrot_training - Step 30138: {'lr': 0.000183116828054591, 'samples': 15431168, 'steps': 30138, 'loss/train': 2.214568614959717}
+02/25/2022 16:37:57 - INFO - codeparrot_training - Step 30139: {'lr': 0.00018310106216772954, 'samples': 15431680, 'steps': 30139, 'loss/train': 0.936568558216095}
+02/25/2022 16:38:01 - INFO - codeparrot_training - Step 30140: {'lr': 0.0001830852965674419, 'samples': 15432192, 'steps': 30140, 'loss/train': 1.6838629245758057}
+02/25/2022 16:38:08 - INFO - codeparrot_training - Step 30141: {'lr': 0.00018306953125379568, 'samples': 15432704, 'steps': 30141, 'loss/train': 1.531569480895996}
+02/25/2022 16:38:11 - INFO - codeparrot_training - Step 30142: {'lr': 0.00018305376622685827, 'samples': 15433216, 'steps': 30142, 'loss/train': 1.35321044921875}
+02/25/2022 16:38:17 - INFO - codeparrot_training - Step 30143: {'lr': 0.00018303800148669724, 'samples': 15433728, 'steps': 30143, 'loss/train': 2.2404191493988037}
+02/25/2022 16:38:20 - INFO - codeparrot_training - Step 30144: {'lr': 0.00018302223703338017, 'samples': 15434240, 'steps': 30144, 'loss/train': 1.8038151264190674}
+02/25/2022 16:38:26 - INFO - codeparrot_training - Step 30145: {'lr': 0.00018300647286697465, 'samples': 15434752, 'steps': 30145, 'loss/train': 2.1525137424468994}
+02/25/2022 16:38:29 - INFO - codeparrot_training - Step 30146: {'lr': 0.00018299070898754804, 'samples': 15435264, 'steps': 30146, 'loss/train': 0.23471519351005554}
+02/25/2022 16:38:34 - INFO - codeparrot_training - Step 30147: {'lr': 0.00018297494539516794, 'samples': 15435776, 'steps': 30147, 'loss/train': 1.5956913232803345}
+02/25/2022 16:38:38 - INFO - codeparrot_training - Step 30148: {'lr': 0.00018295918208990195, 'samples': 15436288, 'steps': 30148, 'loss/train': 1.5869932174682617}
+02/25/2022 16:38:43 - INFO - codeparrot_training - Step 30149: {'lr': 0.00018294341907181756, 'samples': 15436800, 'steps': 30149, 'loss/train': 0.8938074111938477}
+02/25/2022 16:38:47 - INFO - codeparrot_training - Step 30150: {'lr': 0.00018292765634098218, 'samples': 15437312, 'steps': 30150, 'loss/train': 1.9406918287277222}
+02/25/2022 16:38:53 - INFO - codeparrot_training - Step 30151: {'lr': 0.00018291189389746344, 'samples': 15437824, 'steps': 30151, 'loss/train': 2.3271782398223877}
+02/25/2022 16:38:57 - INFO - codeparrot_training - Step 30152: {'lr': 0.00018289613174132888, 'samples': 15438336, 'steps': 30152, 'loss/train': 1.7094000577926636}
+02/25/2022 16:39:02 - INFO - codeparrot_training - Step 30153: {'lr': 0.00018288036987264595, 'samples': 15438848, 'steps': 30153, 'loss/train': 1.9177099466323853}
+02/25/2022 16:39:06 - INFO - codeparrot_training - Step 30154: {'lr': 0.00018286460829148224, 'samples': 15439360, 'steps': 30154, 'loss/train': 1.5429911613464355}
+02/25/2022 16:39:11 - INFO - codeparrot_training - Step 30155: {'lr': 0.00018284884699790524, 'samples': 15439872, 'steps': 30155, 'loss/train': 1.754862904548645}
+02/25/2022 16:39:15 - INFO - codeparrot_training - Step 30156: {'lr': 0.0001828330859919824, 'samples': 15440384, 'steps': 30156, 'loss/train': 1.721798300743103}
+02/25/2022 16:39:20 - INFO - codeparrot_training - Step 30157: {'lr': 0.00018281732527378132, 'samples': 15440896, 'steps': 30157, 'loss/train': 1.8194557428359985}
+02/25/2022 16:39:24 - INFO - codeparrot_training - Step 30158: {'lr': 0.0001828015648433695, 'samples': 15441408, 'steps': 30158, 'loss/train': 0.8753766417503357}
+02/25/2022 16:39:29 - INFO - codeparrot_training - Step 30159: {'lr': 0.00018278580470081443, 'samples': 15441920, 'steps': 30159, 'loss/train': 1.3507989645004272}
+02/25/2022 16:39:33 - INFO - codeparrot_training - Step 30160: {'lr': 0.00018277004484618358, 'samples': 15442432, 'steps': 30160, 'loss/train': 1.6233487129211426}
+02/25/2022 16:39:39 - INFO - codeparrot_training - Step 30161: {'lr': 0.00018275428527954457, 'samples': 15442944, 'steps': 30161, 'loss/train': 1.5625134706497192}
+02/25/2022 16:39:45 - INFO - codeparrot_training - Step 30162: {'lr': 0.0001827385260009648, 'samples': 15443456, 'steps': 30162, 'loss/train': 0.8783797025680542}
+02/25/2022 16:39:48 - INFO - codeparrot_training - Step 30163: {'lr': 0.00018272276701051192, 'samples': 15443968, 'steps': 30163, 'loss/train': 1.6139253377914429}
+02/25/2022 16:39:52 - INFO - codeparrot_training - Step 30164: {'lr': 0.00018270700830825325, 'samples': 15444480, 'steps': 30164, 'loss/train': 1.8954344987869263}
+02/25/2022 16:39:57 - INFO - codeparrot_training - Step 30165: {'lr': 0.00018269124989425643, 'samples': 15444992, 'steps': 30165, 'loss/train': 1.599933385848999}
+02/25/2022 16:40:01 - INFO - codeparrot_training - Step 30166: {'lr': 0.00018267549176858889, 'samples': 15445504, 'steps': 30166, 'loss/train': 1.280036211013794}
+02/25/2022 16:40:06 - INFO - codeparrot_training - Step 30167: {'lr': 0.00018265973393131823, 'samples': 15446016, 'steps': 30167, 'loss/train': 0.6672527194023132}
+02/25/2022 16:40:10 - INFO - codeparrot_training - Step 30168: {'lr': 0.00018264397638251183, 'samples': 15446528, 'steps': 30168, 'loss/train': 1.7261629104614258}
+02/25/2022 16:40:15 - INFO - codeparrot_training - Step 30169: {'lr': 0.00018262821912223726, 'samples': 15447040, 'steps': 30169, 'loss/train': 1.4885536432266235}
+02/25/2022 16:40:19 - INFO - codeparrot_training - Step 30170: {'lr': 0.00018261246215056198, 'samples': 15447552, 'steps': 30170, 'loss/train': 1.1082655191421509}
+02/25/2022 16:40:25 - INFO - codeparrot_training - Step 30171: {'lr': 0.00018259670546755364, 'samples': 15448064, 'steps': 30171, 'loss/train': 0.9089599251747131}
+02/25/2022 16:40:28 - INFO - codeparrot_training - Step 30172: {'lr': 0.0001825809490732795, 'samples': 15448576, 'steps': 30172, 'loss/train': 1.4234005212783813}
+02/25/2022 16:40:34 - INFO - codeparrot_training - Step 30173: {'lr': 0.00018256519296780714, 'samples': 15449088, 'steps': 30173, 'loss/train': 1.1667343378067017}
+02/25/2022 16:40:37 - INFO - codeparrot_training - Step 30174: {'lr': 0.0001825494371512041, 'samples': 15449600, 'steps': 30174, 'loss/train': 1.395885705947876}
+02/25/2022 16:40:43 - INFO - codeparrot_training - Step 30175: {'lr': 0.00018253368162353783, 'samples': 15450112, 'steps': 30175, 'loss/train': 0.7092799544334412}
+02/25/2022 16:40:46 - INFO - codeparrot_training - Step 30176: {'lr': 0.00018251792638487597, 'samples': 15450624, 'steps': 30176, 'loss/train': 2.2143876552581787}
+02/25/2022 16:40:53 - INFO - codeparrot_training - Step 30177: {'lr': 0.00018250217143528578, 'samples': 15451136, 'steps': 30177, 'loss/train': 2.141566514968872}
+02/25/2022 16:40:56 - INFO - codeparrot_training - Step 30178: {'lr': 0.00018248641677483485, 'samples': 15451648, 'steps': 30178, 'loss/train': 0.09129561483860016}
+02/25/2022 16:41:02 - INFO - codeparrot_training - Step 30179: {'lr': 0.00018247066240359067, 'samples': 15452160, 'steps': 30179, 'loss/train': 2.096642017364502}
+02/25/2022 16:41:05 - INFO - codeparrot_training - Step 30180: {'lr': 0.00018245490832162085, 'samples': 15452672, 'steps': 30180, 'loss/train': 2.144042730331421}
+02/25/2022 16:41:11 - INFO - codeparrot_training - Step 30181: {'lr': 0.00018243915452899264, 'samples': 15453184, 'steps': 30181, 'loss/train': 1.9913325309753418}
+02/25/2022 16:41:14 - INFO - codeparrot_training - Step 30182: {'lr': 0.00018242340102577365, 'samples': 15453696, 'steps': 30182, 'loss/train': 1.9597300291061401}
+02/25/2022 16:41:20 - INFO - codeparrot_training - Step 30183: {'lr': 0.00018240764781203133, 'samples': 15454208, 'steps': 30183, 'loss/train': 1.1077646017074585}
+02/25/2022 16:41:23 - INFO - codeparrot_training - Step 30184: {'lr': 0.0001823918948878333, 'samples': 15454720, 'steps': 30184, 'loss/train': 2.382155656814575}
+02/25/2022 16:41:29 - INFO - codeparrot_training - Step 30185: {'lr': 0.00018237614225324682, 'samples': 15455232, 'steps': 30185, 'loss/train': 1.6744897365570068}
+02/25/2022 16:41:32 - INFO - codeparrot_training - Step 30186: {'lr': 0.00018236038990833948, 'samples': 15455744, 'steps': 30186, 'loss/train': 1.266266107559204}
+02/25/2022 16:41:38 - INFO - codeparrot_training - Step 30187: {'lr': 0.00018234463785317872, 'samples': 15456256, 'steps': 30187, 'loss/train': 1.7273356914520264}
+02/25/2022 16:41:41 - INFO - codeparrot_training - Step 30188: {'lr': 0.00018232888608783217, 'samples': 15456768, 'steps': 30188, 'loss/train': 1.1520874500274658}
+02/25/2022 16:41:48 - INFO - codeparrot_training - Step 30189: {'lr': 0.00018231313461236713, 'samples': 15457280, 'steps': 30189, 'loss/train': 1.7684825658798218}
+02/25/2022 16:41:53 - INFO - codeparrot_training - Step 30190: {'lr': 0.00018229738342685105, 'samples': 15457792, 'steps': 30190, 'loss/train': 0.6701033711433411}
+02/25/2022 16:41:57 - INFO - codeparrot_training - Step 30191: {'lr': 0.00018228163253135155, 'samples': 15458304, 'steps': 30191, 'loss/train': 1.2233402729034424}
+02/25/2022 16:42:02 - INFO - codeparrot_training - Step 30192: {'lr': 0.00018226588192593604, 'samples': 15458816, 'steps': 30192, 'loss/train': 1.1996300220489502}
+02/25/2022 16:42:06 - INFO - codeparrot_training - Step 30193: {'lr': 0.000182250131610672, 'samples': 15459328, 'steps': 30193, 'loss/train': 1.3650349378585815}
+02/25/2022 16:42:11 - INFO - codeparrot_training - Step 30194: {'lr': 0.00018223438158562684, 'samples': 15459840, 'steps': 30194, 'loss/train': 0.8478029370307922}
+02/25/2022 16:42:15 - INFO - codeparrot_training - Step 30195: {'lr': 0.0001822186318508681, 'samples': 15460352, 'steps': 30195, 'loss/train': 1.6683921813964844}
+02/25/2022 16:42:21 - INFO - codeparrot_training - Step 30196: {'lr': 0.00018220288240646315, 'samples': 15460864, 'steps': 30196, 'loss/train': 0.8926817774772644}
+02/25/2022 16:42:25 - INFO - codeparrot_training - Step 30197: {'lr': 0.00018218713325247963, 'samples': 15461376, 'steps': 30197, 'loss/train': 2.3401453495025635}
+02/25/2022 16:42:30 - INFO - codeparrot_training - Step 30198: {'lr': 0.00018217138438898482, 'samples': 15461888, 'steps': 30198, 'loss/train': 2.5961201190948486}
+02/25/2022 16:42:34 - INFO - codeparrot_training - Step 30199: {'lr': 0.0001821556358160463, 'samples': 15462400, 'steps': 30199, 'loss/train': 0.856624960899353}
+02/25/2022 16:42:39 - INFO - codeparrot_training - Step 30200: {'lr': 0.00018213988753373146, 'samples': 15462912, 'steps': 30200, 'loss/train': 2.5985987186431885}
+02/25/2022 16:42:43 - INFO - codeparrot_training - Step 30201: {'lr': 0.00018212413954210777, 'samples': 15463424, 'steps': 30201, 'loss/train': 2.1162166595458984}
+02/25/2022 16:42:48 - INFO - codeparrot_training - Step 30202: {'lr': 0.00018210839184124277, 'samples': 15463936, 'steps': 30202, 'loss/train': 0.44784021377563477}
+02/25/2022 16:42:52 - INFO - codeparrot_training - Step 30203: {'lr': 0.00018209264443120392, 'samples': 15464448, 'steps': 30203, 'loss/train': 1.6257777214050293}
+02/25/2022 16:42:57 - INFO - codeparrot_training - Step 30204: {'lr': 0.00018207689731205852, 'samples': 15464960, 'steps': 30204, 'loss/train': 1.9447648525238037}
+02/25/2022 16:43:01 - INFO - codeparrot_training - Step 30205: {'lr': 0.0001820611504838741, 'samples': 15465472, 'steps': 30205, 'loss/train': 1.8288772106170654}
+02/25/2022 16:43:06 - INFO - codeparrot_training - Step 30206: {'lr': 0.00018204540394671826, 'samples': 15465984, 'steps': 30206, 'loss/train': 1.1476339101791382}
+02/25/2022 16:43:10 - INFO - codeparrot_training - Step 30207: {'lr': 0.00018202965770065823, 'samples': 15466496, 'steps': 30207, 'loss/train': 2.9437742233276367}
+02/25/2022 16:43:15 - INFO - codeparrot_training - Step 30208: {'lr': 0.0001820139117457616, 'samples': 15467008, 'steps': 30208, 'loss/train': 1.3765581846237183}
+02/25/2022 16:43:19 - INFO - codeparrot_training - Step 30209: {'lr': 0.00018199816608209575, 'samples': 15467520, 'steps': 30209, 'loss/train': 2.2918102741241455}
+02/25/2022 16:43:24 - INFO - codeparrot_training - Step 30210: {'lr': 0.00018198242070972814, 'samples': 15468032, 'steps': 30210, 'loss/train': 0.5545539259910583}
+02/25/2022 16:43:28 - INFO - codeparrot_training - Step 30211: {'lr': 0.00018196667562872637, 'samples': 15468544, 'steps': 30211, 'loss/train': 2.459815740585327}
+02/25/2022 16:43:34 - INFO - codeparrot_training - Step 30212: {'lr': 0.00018195093083915766, 'samples': 15469056, 'steps': 30212, 'loss/train': 1.0434744358062744}
+02/25/2022 16:43:38 - INFO - codeparrot_training - Step 30213: {'lr': 0.00018193518634108954, 'samples': 15469568, 'steps': 30213, 'loss/train': 0.13225218653678894}
+02/25/2022 16:43:43 - INFO - codeparrot_training - Step 30214: {'lr': 0.0001819194421345895, 'samples': 15470080, 'steps': 30214, 'loss/train': 0.9162151217460632}
+02/25/2022 16:43:46 - INFO - codeparrot_training - Step 30215: {'lr': 0.00018190369821972503, 'samples': 15470592, 'steps': 30215, 'loss/train': 2.204357147216797}
+02/25/2022 16:43:52 - INFO - codeparrot_training - Step 30216: {'lr': 0.00018188795459656338, 'samples': 15471104, 'steps': 30216, 'loss/train': 2.0624074935913086}
+02/25/2022 16:43:56 - INFO - codeparrot_training - Step 30217: {'lr': 0.00018187221126517213, 'samples': 15471616, 'steps': 30217, 'loss/train': 1.7415975332260132}
+02/25/2022 16:44:01 - INFO - codeparrot_training - Step 30218: {'lr': 0.0001818564682256187, 'samples': 15472128, 'steps': 30218, 'loss/train': 1.7843259572982788}
+02/25/2022 16:44:05 - INFO - codeparrot_training - Step 30219: {'lr': 0.0001818407254779706, 'samples': 15472640, 'steps': 30219, 'loss/train': 1.7959237098693848}
+02/25/2022 16:44:10 - INFO - codeparrot_training - Step 30220: {'lr': 0.00018182498302229507, 'samples': 15473152, 'steps': 30220, 'loss/train': 1.3945038318634033}
+02/25/2022 16:44:14 - INFO - codeparrot_training - Step 30221: {'lr': 0.00018180924085865968, 'samples': 15473664, 'steps': 30221, 'loss/train': 0.21227788925170898}
+02/25/2022 16:44:20 - INFO - codeparrot_training - Step 30222: {'lr': 0.00018179349898713189, 'samples': 15474176, 'steps': 30222, 'loss/train': 1.6112300157546997}
+02/25/2022 16:44:23 - INFO - codeparrot_training - Step 30223: {'lr': 0.000181777757407779, 'samples': 15474688, 'steps': 30223, 'loss/train': 1.6448408365249634}
+02/25/2022 16:44:29 - INFO - codeparrot_training - Step 30224: {'lr': 0.00018176201612066874, 'samples': 15475200, 'steps': 30224, 'loss/train': 0.7363759279251099}
+02/25/2022 16:44:32 - INFO - codeparrot_training - Step 30225: {'lr': 0.00018174627512586818, 'samples': 15475712, 'steps': 30225, 'loss/train': 1.2504916191101074}
+02/25/2022 16:44:38 - INFO - codeparrot_training - Step 30226: {'lr': 0.00018173053442344493, 'samples': 15476224, 'steps': 30226, 'loss/train': 0.6949149966239929}
+02/25/2022 16:44:41 - INFO - codeparrot_training - Step 30227: {'lr': 0.00018171479401346636, 'samples': 15476736, 'steps': 30227, 'loss/train': 2.5589053630828857}
+02/25/2022 16:44:47 - INFO - codeparrot_training - Step 30228: {'lr': 0.00018169905389600006, 'samples': 15477248, 'steps': 30228, 'loss/train': 2.4627411365509033}
+02/25/2022 16:44:51 - INFO - codeparrot_training - Step 30229: {'lr': 0.00018168331407111322, 'samples': 15477760, 'steps': 30229, 'loss/train': 2.2167105674743652}
+02/25/2022 16:44:56 - INFO - codeparrot_training - Step 30230: {'lr': 0.00018166757453887334, 'samples': 15478272, 'steps': 30230, 'loss/train': 1.3010408878326416}
+02/25/2022 16:44:59 - INFO - codeparrot_training - Step 30231: {'lr': 0.0001816518352993479, 'samples': 15478784, 'steps': 30231, 'loss/train': 1.871283769607544}
+02/25/2022 16:45:06 - INFO - codeparrot_training - Step 30232: {'lr': 0.00018163609635260436, 'samples': 15479296, 'steps': 30232, 'loss/train': 2.2058842182159424}
+02/25/2022 16:45:10 - INFO - codeparrot_training - Step 30233: {'lr': 0.00018162035769871002, 'samples': 15479808, 'steps': 30233, 'loss/train': 3.6509897708892822}
+02/25/2022 16:45:15 - INFO - codeparrot_training - Step 30234: {'lr': 0.00018160461933773234, 'samples': 15480320, 'steps': 30234, 'loss/train': 1.5391018390655518}
+02/25/2022 16:45:19 - INFO - codeparrot_training - Step 30235: {'lr': 0.0001815888812697388, 'samples': 15480832, 'steps': 30235, 'loss/train': 2.2007896900177}
+02/25/2022 16:45:24 - INFO - codeparrot_training - Step 30236: {'lr': 0.00018157314349479672, 'samples': 15481344, 'steps': 30236, 'loss/train': 2.0087921619415283}
+02/25/2022 16:45:28 - INFO - codeparrot_training - Step 30237: {'lr': 0.0001815574060129736, 'samples': 15481856, 'steps': 30237, 'loss/train': 1.4928714036941528}
+02/25/2022 16:45:33 - INFO - codeparrot_training - Step 30238: {'lr': 0.0001815416688243368, 'samples': 15482368, 'steps': 30238, 'loss/train': 1.8697328567504883}
+02/25/2022 16:45:37 - INFO - codeparrot_training - Step 30239: {'lr': 0.00018152593192895377, 'samples': 15482880, 'steps': 30239, 'loss/train': 2.0418519973754883}
+02/25/2022 16:45:42 - INFO - codeparrot_training - Step 30240: {'lr': 0.00018151019532689188, 'samples': 15483392, 'steps': 30240, 'loss/train': 1.92549467086792}
+02/25/2022 16:45:46 - INFO - codeparrot_training - Step 30241: {'lr': 0.00018149445901821862, 'samples': 15483904, 'steps': 30241, 'loss/train': 1.6473177671432495}
+02/25/2022 16:45:51 - INFO - codeparrot_training - Step 30242: {'lr': 0.00018147872300300127, 'samples': 15484416, 'steps': 30242, 'loss/train': 1.3054625988006592}
+02/25/2022 16:45:55 - INFO - codeparrot_training - Step 30243: {'lr': 0.0001814629872813074, 'samples': 15484928, 'steps': 30243, 'loss/train': 1.6677675247192383}
+02/25/2022 16:46:02 - INFO - codeparrot_training - Step 30244: {'lr': 0.00018144725185320423, 'samples': 15485440, 'steps': 30244, 'loss/train': 1.9955741167068481}
+02/25/2022 16:46:05 - INFO - codeparrot_training - Step 30245: {'lr': 0.00018143151671875935, 'samples': 15485952, 'steps': 30245, 'loss/train': 1.2038229703903198}
+02/25/2022 16:46:11 - INFO - codeparrot_training - Step 30246: {'lr': 0.00018141578187804008, 'samples': 15486464, 'steps': 30246, 'loss/train': 2.1339006423950195}
+02/25/2022 16:46:14 - INFO - codeparrot_training - Step 30247: {'lr': 0.00018140004733111376, 'samples': 15486976, 'steps': 30247, 'loss/train': 0.936119794845581}
+02/25/2022 16:46:20 - INFO - codeparrot_training - Step 30248: {'lr': 0.00018138431307804784, 'samples': 15487488, 'steps': 30248, 'loss/train': 2.3088300228118896}
+02/25/2022 16:46:23 - INFO - codeparrot_training - Step 30249: {'lr': 0.00018136857911890976, 'samples': 15488000, 'steps': 30249, 'loss/train': 2.0716497898101807}
+02/25/2022 16:46:29 - INFO - codeparrot_training - Step 30250: {'lr': 0.000181352845453767, 'samples': 15488512, 'steps': 30250, 'loss/train': 1.789767861366272}
+02/25/2022 16:46:32 - INFO - codeparrot_training - Step 30251: {'lr': 0.00018133711208268672, 'samples': 15489024, 'steps': 30251, 'loss/train': 2.851513624191284}
+02/25/2022 16:46:38 - INFO - codeparrot_training - Step 30252: {'lr': 0.00018132137900573648, 'samples': 15489536, 'steps': 30252, 'loss/train': 1.2172852754592896}
+02/25/2022 16:46:44 - INFO - codeparrot_training - Step 30253: {'lr': 0.0001813056462229836, 'samples': 15490048, 'steps': 30253, 'loss/train': 3.98465633392334}
+02/25/2022 16:46:47 - INFO - codeparrot_training - Step 30254: {'lr': 0.00018128991373449565, 'samples': 15490560, 'steps': 30254, 'loss/train': 1.589063048362732}
+02/25/2022 16:46:53 - INFO - codeparrot_training - Step 30255: {'lr': 0.00018127418154033982, 'samples': 15491072, 'steps': 30255, 'loss/train': 1.040554165840149}
+02/25/2022 16:46:57 - INFO - codeparrot_training - Step 30256: {'lr': 0.00018125844964058353, 'samples': 15491584, 'steps': 30256, 'loss/train': 0.8568814992904663}
+02/25/2022 16:47:00 - INFO - codeparrot_training - Step 30257: {'lr': 0.00018124271803529425, 'samples': 15492096, 'steps': 30257, 'loss/train': 2.070145606994629}
+02/25/2022 16:47:06 - INFO - codeparrot_training - Step 30258: {'lr': 0.00018122698672453939, 'samples': 15492608, 'steps': 30258, 'loss/train': 2.6813528537750244}
+02/25/2022 16:47:12 - INFO - codeparrot_training - Step 30259: {'lr': 0.0001812112557083862, 'samples': 15493120, 'steps': 30259, 'loss/train': 2.105872869491577}
+02/25/2022 16:47:16 - INFO - codeparrot_training - Step 30260: {'lr': 0.00018119552498690214, 'samples': 15493632, 'steps': 30260, 'loss/train': 1.447824239730835}
+02/25/2022 16:47:19 - INFO - codeparrot_training - Step 30261: {'lr': 0.0001811797945601546, 'samples': 15494144, 'steps': 30261, 'loss/train': 2.2766470909118652}
+02/25/2022 16:47:23 - INFO - codeparrot_training - Step 30262: {'lr': 0.000181164064428211, 'samples': 15494656, 'steps': 30262, 'loss/train': 2.3196182250976562}
+02/25/2022 16:47:28 - INFO - codeparrot_training - Step 30263: {'lr': 0.00018114833459113877, 'samples': 15495168, 'steps': 30263, 'loss/train': 1.186671257019043}
+02/25/2022 16:47:32 - INFO - codeparrot_training - Step 30264: {'lr': 0.00018113260504900512, 'samples': 15495680, 'steps': 30264, 'loss/train': 1.2272653579711914}
+02/25/2022 16:47:37 - INFO - codeparrot_training - Step 30265: {'lr': 0.00018111687580187752, 'samples': 15496192, 'steps': 30265, 'loss/train': 0.9229240417480469}
+02/25/2022 16:47:41 - INFO - codeparrot_training - Step 30266: {'lr': 0.00018110114684982332, 'samples': 15496704, 'steps': 30266, 'loss/train': 2.104604721069336}
+02/25/2022 16:47:46 - INFO - codeparrot_training - Step 30267: {'lr': 0.0001810854181929101, 'samples': 15497216, 'steps': 30267, 'loss/train': 1.3323432207107544}
+02/25/2022 16:47:50 - INFO - codeparrot_training - Step 30268: {'lr': 0.00018106968983120486, 'samples': 15497728, 'steps': 30268, 'loss/train': 1.8889106512069702}
+02/25/2022 16:47:56 - INFO - codeparrot_training - Step 30269: {'lr': 0.00018105396176477527, 'samples': 15498240, 'steps': 30269, 'loss/train': 1.221634030342102}
+02/25/2022 16:47:59 - INFO - codeparrot_training - Step 30270: {'lr': 0.00018103823399368859, 'samples': 15498752, 'steps': 30270, 'loss/train': 1.6162779331207275}
+02/25/2022 16:48:05 - INFO - codeparrot_training - Step 30271: {'lr': 0.0001810225065180122, 'samples': 15499264, 'steps': 30271, 'loss/train': 2.724073648452759}
+02/25/2022 16:48:08 - INFO - codeparrot_training - Step 30272: {'lr': 0.00018100677933781362, 'samples': 15499776, 'steps': 30272, 'loss/train': 2.6476962566375732}
+02/25/2022 16:48:14 - INFO - codeparrot_training - Step 30273: {'lr': 0.00018099105245315995, 'samples': 15500288, 'steps': 30273, 'loss/train': 1.5384633541107178}
+02/25/2022 16:48:17 - INFO - codeparrot_training - Step 30274: {'lr': 0.00018097532586411872, 'samples': 15500800, 'steps': 30274, 'loss/train': 1.2873787879943848}
+02/25/2022 16:48:23 - INFO - codeparrot_training - Step 30275: {'lr': 0.0001809595995707573, 'samples': 15501312, 'steps': 30275, 'loss/train': 1.1256585121154785}
+02/25/2022 16:48:26 - INFO - codeparrot_training - Step 30276: {'lr': 0.0001809438735731431, 'samples': 15501824, 'steps': 30276, 'loss/train': 1.6302610635757446}
+02/25/2022 16:48:32 - INFO - codeparrot_training - Step 30277: {'lr': 0.00018092814787134333, 'samples': 15502336, 'steps': 30277, 'loss/train': 1.5706431865692139}
+02/25/2022 16:48:35 - INFO - codeparrot_training - Step 30278: {'lr': 0.00018091242246542544, 'samples': 15502848, 'steps': 30278, 'loss/train': 2.1778671741485596}
+02/25/2022 16:48:41 - INFO - codeparrot_training - Step 30279: {'lr': 0.00018089669735545682, 'samples': 15503360, 'steps': 30279, 'loss/train': 1.2678357362747192}
+02/25/2022 16:48:45 - INFO - codeparrot_training - Step 30280: {'lr': 0.00018088097254150484, 'samples': 15503872, 'steps': 30280, 'loss/train': 1.8698092699050903}
+02/25/2022 16:48:50 - INFO - codeparrot_training - Step 30281: {'lr': 0.00018086524802363678, 'samples': 15504384, 'steps': 30281, 'loss/train': 1.727962613105774}
+02/25/2022 16:48:54 - INFO - codeparrot_training - Step 30282: {'lr': 0.00018084952380192, 'samples': 15504896, 'steps': 30282, 'loss/train': 1.9903841018676758}
+02/25/2022 16:48:59 - INFO - codeparrot_training - Step 30283: {'lr': 0.00018083379987642202, 'samples': 15505408, 'steps': 30283, 'loss/train': 2.727038860321045}
+02/25/2022 16:49:03 - INFO - codeparrot_training - Step 30284: {'lr': 0.00018081807624720998, 'samples': 15505920, 'steps': 30284, 'loss/train': 1.9048691987991333}
+02/25/2022 16:49:08 - INFO - codeparrot_training - Step 30285: {'lr': 0.00018080235291435137, 'samples': 15506432, 'steps': 30285, 'loss/train': 1.6210334300994873}
+02/25/2022 16:49:12 - INFO - codeparrot_training - Step 30286: {'lr': 0.00018078662987791355, 'samples': 15506944, 'steps': 30286, 'loss/train': 1.8738731145858765}
+02/25/2022 16:49:18 - INFO - codeparrot_training - Step 30287: {'lr': 0.00018077090713796372, 'samples': 15507456, 'steps': 30287, 'loss/train': 1.5903934240341187}
+02/25/2022 16:49:21 - INFO - codeparrot_training - Step 30288: {'lr': 0.00018075518469456943, 'samples': 15507968, 'steps': 30288, 'loss/train': 0.026191094890236855}
+02/25/2022 16:49:27 - INFO - codeparrot_training - Step 30289: {'lr': 0.00018073946254779793, 'samples': 15508480, 'steps': 30289, 'loss/train': 1.4007740020751953}
+02/25/2022 16:49:33 - INFO - codeparrot_training - Step 30290: {'lr': 0.00018072374069771658, 'samples': 15508992, 'steps': 30290, 'loss/train': 1.5824601650238037}
+02/25/2022 16:49:36 - INFO - codeparrot_training - Step 30291: {'lr': 0.00018070801914439271, 'samples': 15509504, 'steps': 30291, 'loss/train': 1.3173692226409912}
+02/25/2022 16:49:42 - INFO - codeparrot_training - Step 30292: {'lr': 0.00018069229788789364, 'samples': 15510016, 'steps': 30292, 'loss/train': 1.6421961784362793}
+02/25/2022 16:49:45 - INFO - codeparrot_training - Step 30293: {'lr': 0.0001806765769282869, 'samples': 15510528, 'steps': 30293, 'loss/train': 0.9650719165802002}
+02/25/2022 16:49:51 - INFO - codeparrot_training - Step 30294: {'lr': 0.00018066085626563958, 'samples': 15511040, 'steps': 30294, 'loss/train': 1.5381430387496948}
+02/25/2022 16:49:54 - INFO - codeparrot_training - Step 30295: {'lr': 0.00018064513590001913, 'samples': 15511552, 'steps': 30295, 'loss/train': 1.9772381782531738}
+02/25/2022 16:50:00 - INFO - codeparrot_training - Step 30296: {'lr': 0.0001806294158314929, 'samples': 15512064, 'steps': 30296, 'loss/train': 1.8971456289291382}
+02/25/2022 16:50:03 - INFO - codeparrot_training - Step 30297: {'lr': 0.0001806136960601282, 'samples': 15512576, 'steps': 30297, 'loss/train': 1.6689056158065796}
+02/25/2022 16:50:09 - INFO - codeparrot_training - Step 30298: {'lr': 0.00018059797658599253, 'samples': 15513088, 'steps': 30298, 'loss/train': 2.2771003246307373}
+02/25/2022 16:50:12 - INFO - codeparrot_training - Step 30299: {'lr': 0.00018058225740915295, 'samples': 15513600, 'steps': 30299, 'loss/train': 0.5486544966697693}
+02/25/2022 16:50:18 - INFO - codeparrot_training - Step 30300: {'lr': 0.00018056653852967697, 'samples': 15514112, 'steps': 30300, 'loss/train': 0.883249819278717}
+02/25/2022 16:50:22 - INFO - codeparrot_training - Step 30301: {'lr': 0.00018055081994763187, 'samples': 15514624, 'steps': 30301, 'loss/train': 2.0749897956848145}
+02/25/2022 16:50:27 - INFO - codeparrot_training - Step 30302: {'lr': 0.00018053510166308513, 'samples': 15515136, 'steps': 30302, 'loss/train': 0.5682500004768372}
+02/25/2022 16:50:31 - INFO - codeparrot_training - Step 30303: {'lr': 0.00018051938367610382, 'samples': 15515648, 'steps': 30303, 'loss/train': 0.42886900901794434}
+02/25/2022 16:50:36 - INFO - codeparrot_training - Step 30304: {'lr': 0.00018050366598675545, 'samples': 15516160, 'steps': 30304, 'loss/train': 1.0510327816009521}
+02/25/2022 16:50:40 - INFO - codeparrot_training - Step 30305: {'lr': 0.00018048794859510726, 'samples': 15516672, 'steps': 30305, 'loss/train': 1.8106976747512817}
+02/25/2022 16:50:45 - INFO - codeparrot_training - Step 30306: {'lr': 0.00018047223150122676, 'samples': 15517184, 'steps': 30306, 'loss/train': 1.6108672618865967}
+02/25/2022 16:50:49 - INFO - codeparrot_training - Step 30307: {'lr': 0.00018045651470518105, 'samples': 15517696, 'steps': 30307, 'loss/train': 1.9195351600646973}
+02/25/2022 16:50:54 - INFO - codeparrot_training - Step 30308: {'lr': 0.00018044079820703752, 'samples': 15518208, 'steps': 30308, 'loss/train': 1.5020885467529297}
+02/25/2022 16:50:58 - INFO - codeparrot_training - Step 30309: {'lr': 0.00018042508200686352, 'samples': 15518720, 'steps': 30309, 'loss/train': 0.6649724841117859}
+02/25/2022 16:51:03 - INFO - codeparrot_training - Step 30310: {'lr': 0.00018040936610472642, 'samples': 15519232, 'steps': 30310, 'loss/train': 1.338865041732788}
+02/25/2022 16:51:07 - INFO - codeparrot_training - Step 30311: {'lr': 0.00018039365050069356, 'samples': 15519744, 'steps': 30311, 'loss/train': 1.6641895771026611}
+02/25/2022 16:51:12 - INFO - codeparrot_training - Step 30312: {'lr': 0.00018037793519483212, 'samples': 15520256, 'steps': 30312, 'loss/train': 1.285415530204773}
+02/25/2022 16:51:16 - INFO - codeparrot_training - Step 30313: {'lr': 0.0001803622201872095, 'samples': 15520768, 'steps': 30313, 'loss/train': 1.5049984455108643}
+02/25/2022 16:51:21 - INFO - codeparrot_training - Step 30314: {'lr': 0.000180346505477893, 'samples': 15521280, 'steps': 30314, 'loss/train': 1.6976279020309448}
+02/25/2022 16:51:25 - INFO - codeparrot_training - Step 30315: {'lr': 0.00018033079106695004, 'samples': 15521792, 'steps': 30315, 'loss/train': 1.6904067993164062}
+02/25/2022 16:51:31 - INFO - codeparrot_training - Step 30316: {'lr': 0.00018031507695444776, 'samples': 15522304, 'steps': 30316, 'loss/train': 1.3612467050552368}
+02/25/2022 16:51:34 - INFO - codeparrot_training - Step 30317: {'lr': 0.00018029936314045358, 'samples': 15522816, 'steps': 30317, 'loss/train': 0.8877971768379211}
+02/25/2022 16:51:40 - INFO - codeparrot_training - Step 30318: {'lr': 0.00018028364962503478, 'samples': 15523328, 'steps': 30318, 'loss/train': 2.2065677642822266}
+02/25/2022 16:51:43 - INFO - codeparrot_training - Step 30319: {'lr': 0.00018026793640825873, 'samples': 15523840, 'steps': 30319, 'loss/train': 1.7737586498260498}
+02/25/2022 16:51:49 - INFO - codeparrot_training - Step 30320: {'lr': 0.0001802522234901927, 'samples': 15524352, 'steps': 30320, 'loss/train': 1.4549020528793335}
+02/25/2022 16:51:52 - INFO - codeparrot_training - Step 30321: {'lr': 0.00018023651087090397, 'samples': 15524864, 'steps': 30321, 'loss/train': 1.487540364265442}
+02/25/2022 16:51:58 - INFO - codeparrot_training - Step 30322: {'lr': 0.00018022079855045987, 'samples': 15525376, 'steps': 30322, 'loss/train': 2.4746651649475098}
+02/25/2022 16:52:01 - INFO - codeparrot_training - Step 30323: {'lr': 0.00018020508652892775, 'samples': 15525888, 'steps': 30323, 'loss/train': 2.464829921722412}
+02/25/2022 16:52:07 - INFO - codeparrot_training - Step 30324: {'lr': 0.00018018937480637488, 'samples': 15526400, 'steps': 30324, 'loss/train': 2.0563764572143555}
+02/25/2022 16:52:13 - INFO - codeparrot_training - Step 30325: {'lr': 0.0001801736633828685, 'samples': 15526912, 'steps': 30325, 'loss/train': 2.1133432388305664}
+02/25/2022 16:52:17 - INFO - codeparrot_training - Step 30326: {'lr': 0.00018015795225847603, 'samples': 15527424, 'steps': 30326, 'loss/train': 1.5929611921310425}
+02/25/2022 16:52:22 - INFO - codeparrot_training - Step 30327: {'lr': 0.00018014224143326468, 'samples': 15527936, 'steps': 30327, 'loss/train': 0.8876922726631165}
+02/25/2022 16:52:25 - INFO - codeparrot_training - Step 30328: {'lr': 0.00018012653090730183, 'samples': 15528448, 'steps': 30328, 'loss/train': 0.8034305572509766}
+02/25/2022 16:52:31 - INFO - codeparrot_training - Step 30329: {'lr': 0.00018011082068065468, 'samples': 15528960, 'steps': 30329, 'loss/train': 1.4819104671478271}
+02/25/2022 16:52:34 - INFO - codeparrot_training - Step 30330: {'lr': 0.00018009511075339063, 'samples': 15529472, 'steps': 30330, 'loss/train': 1.6907936334609985}
+02/25/2022 16:52:40 - INFO - codeparrot_training - Step 30331: {'lr': 0.00018007940112557688, 'samples': 15529984, 'steps': 30331, 'loss/train': 1.6368550062179565}
+02/25/2022 16:52:43 - INFO - codeparrot_training - Step 30332: {'lr': 0.00018006369179728078, 'samples': 15530496, 'steps': 30332, 'loss/train': 2.366891622543335}
+02/25/2022 16:52:49 - INFO - codeparrot_training - Step 30333: {'lr': 0.0001800479827685697, 'samples': 15531008, 'steps': 30333, 'loss/train': 0.8492269515991211}
+02/25/2022 16:52:52 - INFO - codeparrot_training - Step 30334: {'lr': 0.00018003227403951078, 'samples': 15531520, 'steps': 30334, 'loss/train': 2.5445034503936768}
+02/25/2022 16:52:59 - INFO - codeparrot_training - Step 30335: {'lr': 0.00018001656561017135, 'samples': 15532032, 'steps': 30335, 'loss/train': 0.2546338438987732}
+02/25/2022 16:53:02 - INFO - codeparrot_training - Step 30336: {'lr': 0.00018000085748061872, 'samples': 15532544, 'steps': 30336, 'loss/train': 1.3956319093704224}
+02/25/2022 16:53:08 - INFO - codeparrot_training - Step 30337: {'lr': 0.00017998514965092034, 'samples': 15533056, 'steps': 30337, 'loss/train': 1.7547041177749634}
+02/25/2022 16:53:11 - INFO - codeparrot_training - Step 30338: {'lr': 0.0001799694421211432, 'samples': 15533568, 'steps': 30338, 'loss/train': 1.4463893175125122}
+02/25/2022 16:53:17 - INFO - codeparrot_training - Step 30339: {'lr': 0.00017995373489135475, 'samples': 15534080, 'steps': 30339, 'loss/train': 2.6473276615142822}
+02/25/2022 16:53:20 - INFO - codeparrot_training - Step 30340: {'lr': 0.00017993802796162226, 'samples': 15534592, 'steps': 30340, 'loss/train': 2.3492541313171387}
+02/25/2022 16:53:26 - INFO - codeparrot_training - Step 30341: {'lr': 0.0001799223213320131, 'samples': 15535104, 'steps': 30341, 'loss/train': 0.3014780580997467}
+02/25/2022 16:53:29 - INFO - codeparrot_training - Step 30342: {'lr': 0.0001799066150025944, 'samples': 15535616, 'steps': 30342, 'loss/train': 1.133432149887085}
+02/25/2022 16:53:35 - INFO - codeparrot_training - Step 30343: {'lr': 0.00017989090897343346, 'samples': 15536128, 'steps': 30343, 'loss/train': 1.5439274311065674}
+02/25/2022 16:53:38 - INFO - codeparrot_training - Step 30344: {'lr': 0.0001798752032445976, 'samples': 15536640, 'steps': 30344, 'loss/train': 2.585175037384033}
+02/25/2022 16:53:45 - INFO - codeparrot_training - Step 30345: {'lr': 0.00017985949781615412, 'samples': 15537152, 'steps': 30345, 'loss/train': 3.327129364013672}
+02/25/2022 16:53:48 - INFO - codeparrot_training - Step 30346: {'lr': 0.00017984379268817043, 'samples': 15537664, 'steps': 30346, 'loss/train': 2.091188669204712}
+02/25/2022 16:53:54 - INFO - codeparrot_training - Step 30347: {'lr': 0.00017982808786071348, 'samples': 15538176, 'steps': 30347, 'loss/train': 1.7906984090805054}
+02/25/2022 16:53:57 - INFO - codeparrot_training - Step 30348: {'lr': 0.00017981238333385075, 'samples': 15538688, 'steps': 30348, 'loss/train': 1.2289783954620361}
+02/25/2022 16:54:03 - INFO - codeparrot_training - Step 30349: {'lr': 0.00017979667910764945, 'samples': 15539200, 'steps': 30349, 'loss/train': 0.6899938583374023}
+02/25/2022 16:54:06 - INFO - codeparrot_training - Step 30350: {'lr': 0.00017978097518217702, 'samples': 15539712, 'steps': 30350, 'loss/train': 2.473595380783081}
+02/25/2022 16:54:12 - INFO - codeparrot_training - Step 30351: {'lr': 0.0001797652715575005, 'samples': 15540224, 'steps': 30351, 'loss/train': 0.7746754288673401}
+02/25/2022 16:54:15 - INFO - codeparrot_training - Step 30352: {'lr': 0.00017974956823368728, 'samples': 15540736, 'steps': 30352, 'loss/train': 1.9135740995407104}
+02/25/2022 16:54:21 - INFO - codeparrot_training - Step 30353: {'lr': 0.00017973386521080454, 'samples': 15541248, 'steps': 30353, 'loss/train': 2.9119625091552734}
+02/25/2022 16:54:24 - INFO - codeparrot_training - Step 30354: {'lr': 0.00017971816248891972, 'samples': 15541760, 'steps': 30354, 'loss/train': 1.8786157369613647}
+02/25/2022 16:54:30 - INFO - codeparrot_training - Step 30355: {'lr': 0.0001797024600680999, 'samples': 15542272, 'steps': 30355, 'loss/train': 1.6714987754821777}
+02/25/2022 16:54:33 - INFO - codeparrot_training - Step 30356: {'lr': 0.00017968675794841242, 'samples': 15542784, 'steps': 30356, 'loss/train': 0.6088492274284363}
+02/25/2022 16:54:39 - INFO - codeparrot_training - Step 30357: {'lr': 0.00017967105612992453, 'samples': 15543296, 'steps': 30357, 'loss/train': 2.0997838973999023}
+02/25/2022 16:54:42 - INFO - codeparrot_training - Step 30358: {'lr': 0.0001796553546127035, 'samples': 15543808, 'steps': 30358, 'loss/train': 2.0839684009552}
+02/25/2022 16:54:48 - INFO - codeparrot_training - Step 30359: {'lr': 0.0001796396533968167, 'samples': 15544320, 'steps': 30359, 'loss/train': 1.047836422920227}
+02/25/2022 16:54:51 - INFO - codeparrot_training - Step 30360: {'lr': 0.00017962395248233116, 'samples': 15544832, 'steps': 30360, 'loss/train': 0.5357745289802551}
+02/25/2022 16:54:58 - INFO - codeparrot_training - Step 30361: {'lr': 0.0001796082518693143, 'samples': 15545344, 'steps': 30361, 'loss/train': 0.5331573486328125}
+02/25/2022 16:55:02 - INFO - codeparrot_training - Step 30362: {'lr': 0.00017959255155783333, 'samples': 15545856, 'steps': 30362, 'loss/train': 2.125424861907959}
+02/25/2022 16:55:07 - INFO - codeparrot_training - Step 30363: {'lr': 0.00017957685154795558, 'samples': 15546368, 'steps': 30363, 'loss/train': 2.351050853729248}
+02/25/2022 16:55:11 - INFO - codeparrot_training - Step 30364: {'lr': 0.00017956115183974817, 'samples': 15546880, 'steps': 30364, 'loss/train': 0.18436282873153687}
+02/25/2022 16:55:16 - INFO - codeparrot_training - Step 30365: {'lr': 0.0001795454524332784, 'samples': 15547392, 'steps': 30365, 'loss/train': 1.8483694791793823}
+02/25/2022 16:55:20 - INFO - codeparrot_training - Step 30366: {'lr': 0.0001795297533286136, 'samples': 15547904, 'steps': 30366, 'loss/train': 1.4149712324142456}
+02/25/2022 16:55:25 - INFO - codeparrot_training - Step 30367: {'lr': 0.0001795140545258209, 'samples': 15548416, 'steps': 30367, 'loss/train': 2.066742420196533}
+02/25/2022 16:55:29 - INFO - codeparrot_training - Step 30368: {'lr': 0.00017949835602496767, 'samples': 15548928, 'steps': 30368, 'loss/train': 2.4230031967163086}
+02/25/2022 16:55:34 - INFO - codeparrot_training - Step 30369: {'lr': 0.00017948265782612106, 'samples': 15549440, 'steps': 30369, 'loss/train': 1.5072993040084839}
+02/25/2022 16:55:38 - INFO - codeparrot_training - Step 30370: {'lr': 0.00017946695992934836, 'samples': 15549952, 'steps': 30370, 'loss/train': 1.5139480829238892}
+02/25/2022 16:55:45 - INFO - codeparrot_training - Step 30371: {'lr': 0.0001794512623347168, 'samples': 15550464, 'steps': 30371, 'loss/train': 1.3498259782791138}
+02/25/2022 16:55:48 - INFO - codeparrot_training - Step 30372: {'lr': 0.00017943556504229364, 'samples': 15550976, 'steps': 30372, 'loss/train': 1.6495883464813232}
+02/25/2022 16:55:54 - INFO - codeparrot_training - Step 30373: {'lr': 0.00017941986805214614, 'samples': 15551488, 'steps': 30373, 'loss/train': 0.6480066180229187}
+02/25/2022 16:55:57 - INFO - codeparrot_training - Step 30374: {'lr': 0.00017940417136434145, 'samples': 15552000, 'steps': 30374, 'loss/train': 1.8776949644088745}
+02/25/2022 16:56:03 - INFO - codeparrot_training - Step 30375: {'lr': 0.0001793884749789469, 'samples': 15552512, 'steps': 30375, 'loss/train': 1.4403725862503052}
+02/25/2022 16:56:06 - INFO - codeparrot_training - Step 30376: {'lr': 0.00017937277889602972, 'samples': 15553024, 'steps': 30376, 'loss/train': 2.1115121841430664}
+02/25/2022 16:56:13 - INFO - codeparrot_training - Step 30377: {'lr': 0.00017935708311565712, 'samples': 15553536, 'steps': 30377, 'loss/train': 2.657585859298706}
+02/25/2022 16:56:16 - INFO - codeparrot_training - Step 30378: {'lr': 0.00017934138763789633, 'samples': 15554048, 'steps': 30378, 'loss/train': 2.2927820682525635}
+02/25/2022 16:56:19 - INFO - codeparrot_training - Step 30379: {'lr': 0.00017932569246281458, 'samples': 15554560, 'steps': 30379, 'loss/train': 1.2131829261779785}
+02/25/2022 16:56:25 - INFO - codeparrot_training - Step 30380: {'lr': 0.0001793099975904791, 'samples': 15555072, 'steps': 30380, 'loss/train': 2.9934585094451904}
+02/25/2022 16:56:28 - INFO - codeparrot_training - Step 30381: {'lr': 0.0001792943030209573, 'samples': 15555584, 'steps': 30381, 'loss/train': 2.340916156768799}
+02/25/2022 16:56:35 - INFO - codeparrot_training - Step 30382: {'lr': 0.0001792786087543161, 'samples': 15556096, 'steps': 30382, 'loss/train': 2.8370912075042725}
+02/25/2022 16:56:38 - INFO - codeparrot_training - Step 30383: {'lr': 0.0001792629147906229, 'samples': 15556608, 'steps': 30383, 'loss/train': 1.3860429525375366}
+02/25/2022 16:56:44 - INFO - codeparrot_training - Step 30384: {'lr': 0.00017924722112994492, 'samples': 15557120, 'steps': 30384, 'loss/train': 1.2881313562393188}
+02/25/2022 16:56:49 - INFO - codeparrot_training - Step 30385: {'lr': 0.0001792315277723495, 'samples': 15557632, 'steps': 30385, 'loss/train': 1.1875712871551514}
+02/25/2022 16:56:53 - INFO - codeparrot_training - Step 30386: {'lr': 0.0001792158347179036, 'samples': 15558144, 'steps': 30386, 'loss/train': 1.2800253629684448}
+02/25/2022 16:56:58 - INFO - codeparrot_training - Step 30387: {'lr': 0.0001792001419666746, 'samples': 15558656, 'steps': 30387, 'loss/train': 3.454211950302124}
+02/25/2022 16:57:02 - INFO - codeparrot_training - Step 30388: {'lr': 0.00017918444951872973, 'samples': 15559168, 'steps': 30388, 'loss/train': 2.270298719406128}
+02/25/2022 16:57:07 - INFO - codeparrot_training - Step 30389: {'lr': 0.00017916875737413628, 'samples': 15559680, 'steps': 30389, 'loss/train': 0.14630478620529175}
+02/25/2022 16:57:10 - INFO - codeparrot_training - Step 30390: {'lr': 0.00017915306553296128, 'samples': 15560192, 'steps': 30390, 'loss/train': 0.29109862446784973}
+02/25/2022 16:57:16 - INFO - codeparrot_training - Step 30391: {'lr': 0.00017913737399527206, 'samples': 15560704, 'steps': 30391, 'loss/train': 1.2275121212005615}
+02/25/2022 16:57:20 - INFO - codeparrot_training - Step 30392: {'lr': 0.00017912168276113582, 'samples': 15561216, 'steps': 30392, 'loss/train': 1.7555817365646362}
+02/25/2022 16:57:25 - INFO - codeparrot_training - Step 30393: {'lr': 0.00017910599183061976, 'samples': 15561728, 'steps': 30393, 'loss/train': 0.1005677655339241}
+02/25/2022 16:57:29 - INFO - codeparrot_training - Step 30394: {'lr': 0.00017909030120379126, 'samples': 15562240, 'steps': 30394, 'loss/train': 1.9121339321136475}
+02/25/2022 16:57:34 - INFO - codeparrot_training - Step 30395: {'lr': 0.00017907461088071728, 'samples': 15562752, 'steps': 30395, 'loss/train': 1.8280134201049805}
+02/25/2022 16:57:38 - INFO - codeparrot_training - Step 30396: {'lr': 0.00017905892086146513, 'samples': 15563264, 'steps': 30396, 'loss/train': 1.857879638671875}
+02/25/2022 16:57:44 - INFO - codeparrot_training - Step 30397: {'lr': 0.00017904323114610206, 'samples': 15563776, 'steps': 30397, 'loss/train': 1.7603507041931152}
+02/25/2022 16:57:47 - INFO - codeparrot_training - Step 30398: {'lr': 0.00017902754173469536, 'samples': 15564288, 'steps': 30398, 'loss/train': 1.5191551446914673}
+02/25/2022 16:57:53 - INFO - codeparrot_training - Step 30399: {'lr': 0.000179011852627312, 'samples': 15564800, 'steps': 30399, 'loss/train': 1.4686052799224854}
+02/25/2022 16:57:56 - INFO - codeparrot_training - Step 30400: {'lr': 0.00017899616382401935, 'samples': 15565312, 'steps': 30400, 'loss/train': 0.4663267433643341}
+02/25/2022 16:58:02 - INFO - codeparrot_training - Step 30401: {'lr': 0.0001789804753248846, 'samples': 15565824, 'steps': 30401, 'loss/train': 2.4124910831451416}
+02/25/2022 16:58:05 - INFO - codeparrot_training - Step 30402: {'lr': 0.00017896478712997499, 'samples': 15566336, 'steps': 30402, 'loss/train': 2.231232166290283}
+02/25/2022 16:58:11 - INFO - codeparrot_training - Step 30403: {'lr': 0.00017894909923935761, 'samples': 15566848, 'steps': 30403, 'loss/train': 1.9327001571655273}
+02/25/2022 16:58:14 - INFO - codeparrot_training - Step 30404: {'lr': 0.00017893341165309973, 'samples': 15567360, 'steps': 30404, 'loss/train': 1.5463478565216064}
+02/25/2022 16:58:20 - INFO - codeparrot_training - Step 30405: {'lr': 0.00017891772437126854, 'samples': 15567872, 'steps': 30405, 'loss/train': 1.7060670852661133}
+02/25/2022 16:58:23 - INFO - codeparrot_training - Step 30406: {'lr': 0.0001789020373939313, 'samples': 15568384, 'steps': 30406, 'loss/train': 1.8175512552261353}
+02/25/2022 16:58:30 - INFO - codeparrot_training - Step 30407: {'lr': 0.00017888635072115517, 'samples': 15568896, 'steps': 30407, 'loss/train': 1.597469687461853}
+02/25/2022 16:58:33 - INFO - codeparrot_training - Step 30408: {'lr': 0.00017887066435300729, 'samples': 15569408, 'steps': 30408, 'loss/train': 2.3395121097564697}
+02/25/2022 16:58:39 - INFO - codeparrot_training - Step 30409: {'lr': 0.00017885497828955488, 'samples': 15569920, 'steps': 30409, 'loss/train': 2.5138654708862305}
+02/25/2022 16:58:42 - INFO - codeparrot_training - Step 30410: {'lr': 0.00017883929253086517, 'samples': 15570432, 'steps': 30410, 'loss/train': 0.031385187059640884}
+02/25/2022 16:58:48 - INFO - codeparrot_training - Step 30411: {'lr': 0.0001788236070770054, 'samples': 15570944, 'steps': 30411, 'loss/train': 1.58345365524292}
+02/25/2022 16:58:51 - INFO - codeparrot_training - Step 30412: {'lr': 0.0001788079219280426, 'samples': 15571456, 'steps': 30412, 'loss/train': 1.1391631364822388}
+02/25/2022 16:58:57 - INFO - codeparrot_training - Step 30413: {'lr': 0.00017879223708404412, 'samples': 15571968, 'steps': 30413, 'loss/train': 2.277214765548706}
+02/25/2022 16:59:00 - INFO - codeparrot_training - Step 30414: {'lr': 0.00017877655254507703, 'samples': 15572480, 'steps': 30414, 'loss/train': 1.3200619220733643}
+02/25/2022 16:59:06 - INFO - codeparrot_training - Step 30415: {'lr': 0.00017876086831120864, 'samples': 15572992, 'steps': 30415, 'loss/train': 2.1675829887390137}
+02/25/2022 16:59:09 - INFO - codeparrot_training - Step 30416: {'lr': 0.00017874518438250596, 'samples': 15573504, 'steps': 30416, 'loss/train': 2.615511894226074}
+02/25/2022 16:59:15 - INFO - codeparrot_training - Step 30417: {'lr': 0.0001787295007590364, 'samples': 15574016, 'steps': 30417, 'loss/train': 2.0820140838623047}
+02/25/2022 16:59:18 - INFO - codeparrot_training - Step 30418: {'lr': 0.00017871381744086694, 'samples': 15574528, 'steps': 30418, 'loss/train': 3.1339359283447266}
+02/25/2022 16:59:24 - INFO - codeparrot_training - Step 30419: {'lr': 0.00017869813442806487, 'samples': 15575040, 'steps': 30419, 'loss/train': 1.7506332397460938}
+02/25/2022 16:59:28 - INFO - codeparrot_training - Step 30420: {'lr': 0.00017868245172069746, 'samples': 15575552, 'steps': 30420, 'loss/train': 2.532691478729248}
+02/25/2022 16:59:33 - INFO - codeparrot_training - Step 30421: {'lr': 0.0001786667693188317, 'samples': 15576064, 'steps': 30421, 'loss/train': 0.6444061398506165}
+02/25/2022 16:59:37 - INFO - codeparrot_training - Step 30422: {'lr': 0.00017865108722253483, 'samples': 15576576, 'steps': 30422, 'loss/train': 2.0239057540893555}
+02/25/2022 16:59:42 - INFO - codeparrot_training - Step 30423: {'lr': 0.000178635405431874, 'samples': 15577088, 'steps': 30423, 'loss/train': 2.295332431793213}
+02/25/2022 16:59:46 - INFO - codeparrot_training - Step 30424: {'lr': 0.00017861972394691655, 'samples': 15577600, 'steps': 30424, 'loss/train': 1.6566064357757568}
+02/25/2022 16:59:52 - INFO - codeparrot_training - Step 30425: {'lr': 0.00017860404276772945, 'samples': 15578112, 'steps': 30425, 'loss/train': 0.6063607335090637}
+02/25/2022 16:59:55 - INFO - codeparrot_training - Step 30426: {'lr': 0.00017858836189438, 'samples': 15578624, 'steps': 30426, 'loss/train': 2.3397676944732666}
+02/25/2022 17:00:01 - INFO - codeparrot_training - Step 30427: {'lr': 0.00017857268132693527, 'samples': 15579136, 'steps': 30427, 'loss/train': 1.2900431156158447}
+02/25/2022 17:00:04 - INFO - codeparrot_training - Step 30428: {'lr': 0.00017855700106546253, 'samples': 15579648, 'steps': 30428, 'loss/train': 1.8489912748336792}
+02/25/2022 17:00:11 - INFO - codeparrot_training - Step 30429: {'lr': 0.00017854132111002903, 'samples': 15580160, 'steps': 30429, 'loss/train': 0.22888192534446716}
+02/25/2022 17:00:16 - INFO - codeparrot_training - Step 30430: {'lr': 0.00017852564146070166, 'samples': 15580672, 'steps': 30430, 'loss/train': 1.965506672859192}
+02/25/2022 17:00:20 - INFO - codeparrot_training - Step 30431: {'lr': 0.00017850996211754778, 'samples': 15581184, 'steps': 30431, 'loss/train': 1.6344127655029297}
+02/25/2022 17:00:25 - INFO - codeparrot_training - Step 30432: {'lr': 0.00017849428308063452, 'samples': 15581696, 'steps': 30432, 'loss/train': 0.405988484621048}
+02/25/2022 17:00:29 - INFO - codeparrot_training - Step 30433: {'lr': 0.00017847860435002915, 'samples': 15582208, 'steps': 30433, 'loss/train': 1.8703230619430542}
+02/25/2022 17:00:34 - INFO - codeparrot_training - Step 30434: {'lr': 0.0001784629259257986, 'samples': 15582720, 'steps': 30434, 'loss/train': 2.751441478729248}
+02/25/2022 17:00:38 - INFO - codeparrot_training - Step 30435: {'lr': 0.0001784472478080102, 'samples': 15583232, 'steps': 30435, 'loss/train': 0.8970136642456055}
+02/25/2022 17:00:43 - INFO - codeparrot_training - Step 30436: {'lr': 0.00017843156999673105, 'samples': 15583744, 'steps': 30436, 'loss/train': 2.1291606426239014}
+02/25/2022 17:00:47 - INFO - codeparrot_training - Step 30437: {'lr': 0.00017841589249202843, 'samples': 15584256, 'steps': 30437, 'loss/train': 1.4831438064575195}
+02/25/2022 17:00:54 - INFO - codeparrot_training - Step 30438: {'lr': 0.0001784002152939693, 'samples': 15584768, 'steps': 30438, 'loss/train': 1.8092501163482666}
+02/25/2022 17:00:57 - INFO - codeparrot_training - Step 30439: {'lr': 0.00017838453840262093, 'samples': 15585280, 'steps': 30439, 'loss/train': 1.9374003410339355}
+02/25/2022 17:01:03 - INFO - codeparrot_training - Step 30440: {'lr': 0.0001783688618180504, 'samples': 15585792, 'steps': 30440, 'loss/train': 1.5201354026794434}
+02/25/2022 17:01:06 - INFO - codeparrot_training - Step 30441: {'lr': 0.00017835318554032498, 'samples': 15586304, 'steps': 30441, 'loss/train': 0.5519653558731079}
+02/25/2022 17:01:11 - INFO - codeparrot_training - Step 30442: {'lr': 0.00017833750956951184, 'samples': 15586816, 'steps': 30442, 'loss/train': 2.6692910194396973}
+02/25/2022 17:01:15 - INFO - codeparrot_training - Step 30443: {'lr': 0.00017832183390567794, 'samples': 15587328, 'steps': 30443, 'loss/train': 1.865640640258789}
+02/25/2022 17:01:20 - INFO - codeparrot_training - Step 30444: {'lr': 0.00017830615854889055, 'samples': 15587840, 'steps': 30444, 'loss/train': 2.4657881259918213}
+02/25/2022 17:01:24 - INFO - codeparrot_training - Step 30445: {'lr': 0.00017829048349921685, 'samples': 15588352, 'steps': 30445, 'loss/train': 1.0936853885650635}
+02/25/2022 17:01:29 - INFO - codeparrot_training - Step 30446: {'lr': 0.000178274808756724, 'samples': 15588864, 'steps': 30446, 'loss/train': 2.338529586791992}
+02/25/2022 17:01:33 - INFO - codeparrot_training - Step 30447: {'lr': 0.000178259134321479, 'samples': 15589376, 'steps': 30447, 'loss/train': 1.8069361448287964}
+02/25/2022 17:01:38 - INFO - codeparrot_training - Step 30448: {'lr': 0.0001782434601935491, 'samples': 15589888, 'steps': 30448, 'loss/train': 0.4443916380405426}
+02/25/2022 17:01:42 - INFO - codeparrot_training - Step 30449: {'lr': 0.0001782277863730014, 'samples': 15590400, 'steps': 30449, 'loss/train': 1.14827299118042}
+02/25/2022 17:01:47 - INFO - codeparrot_training - Step 30450: {'lr': 0.00017821211285990325, 'samples': 15590912, 'steps': 30450, 'loss/train': 2.1597254276275635}
+02/25/2022 17:01:51 - INFO - codeparrot_training - Step 30451: {'lr': 0.00017819643965432145, 'samples': 15591424, 'steps': 30451, 'loss/train': 1.946171522140503}
+02/25/2022 17:01:56 - INFO - codeparrot_training - Step 30452: {'lr': 0.00017818076675632334, 'samples': 15591936, 'steps': 30452, 'loss/train': 0.8167882561683655}
+02/25/2022 17:02:00 - INFO - codeparrot_training - Step 30453: {'lr': 0.00017816509416597603, 'samples': 15592448, 'steps': 30453, 'loss/train': 1.5983017683029175}
+02/25/2022 17:02:07 - INFO - codeparrot_training - Step 30454: {'lr': 0.0001781494218833466, 'samples': 15592960, 'steps': 30454, 'loss/train': 2.4102554321289062}
+02/25/2022 17:02:10 - INFO - codeparrot_training - Step 30455: {'lr': 0.0001781337499085023, 'samples': 15593472, 'steps': 30455, 'loss/train': 0.025332242250442505}
+02/25/2022 17:02:16 - INFO - codeparrot_training - Step 30456: {'lr': 0.00017811807824151014, 'samples': 15593984, 'steps': 30456, 'loss/train': 1.1768425703048706}
+02/25/2022 17:02:19 - INFO - codeparrot_training - Step 30457: {'lr': 0.00017810240688243738, 'samples': 15594496, 'steps': 30457, 'loss/train': 2.0345492362976074}
+02/25/2022 17:02:25 - INFO - codeparrot_training - Step 30458: {'lr': 0.000178086735831351, 'samples': 15595008, 'steps': 30458, 'loss/train': 0.7213736772537231}
+02/25/2022 17:02:28 - INFO - codeparrot_training - Step 30459: {'lr': 0.00017807106508831828, 'samples': 15595520, 'steps': 30459, 'loss/train': 2.2966644763946533}
+02/25/2022 17:02:34 - INFO - codeparrot_training - Step 30460: {'lr': 0.00017805539465340624, 'samples': 15596032, 'steps': 30460, 'loss/train': 2.0306010246276855}
+02/25/2022 17:02:37 - INFO - codeparrot_training - Step 30461: {'lr': 0.000178039724526682, 'samples': 15596544, 'steps': 30461, 'loss/train': 0.999924898147583}
+02/25/2022 17:02:43 - INFO - codeparrot_training - Step 30462: {'lr': 0.00017802405470821275, 'samples': 15597056, 'steps': 30462, 'loss/train': 1.6579713821411133}
+02/25/2022 17:02:46 - INFO - codeparrot_training - Step 30463: {'lr': 0.00017800838519806567, 'samples': 15597568, 'steps': 30463, 'loss/train': 2.093268871307373}
+02/25/2022 17:02:53 - INFO - codeparrot_training - Step 30464: {'lr': 0.0001779927159963078, 'samples': 15598080, 'steps': 30464, 'loss/train': 1.3233659267425537}
+02/25/2022 17:02:56 - INFO - codeparrot_training - Step 30465: {'lr': 0.00017797704710300615, 'samples': 15598592, 'steps': 30465, 'loss/train': 2.512624740600586}
+02/25/2022 17:03:02 - INFO - codeparrot_training - Step 30466: {'lr': 0.00017796137851822802, 'samples': 15599104, 'steps': 30466, 'loss/train': 1.8047192096710205}
+02/25/2022 17:03:05 - INFO - codeparrot_training - Step 30467: {'lr': 0.0001779457102420405, 'samples': 15599616, 'steps': 30467, 'loss/train': 1.194332480430603}
+02/25/2022 17:03:10 - INFO - codeparrot_training - Step 30468: {'lr': 0.00017793004227451071, 'samples': 15600128, 'steps': 30468, 'loss/train': 1.56063973903656}
+02/25/2022 17:03:14 - INFO - codeparrot_training - Step 30469: {'lr': 0.00017791437461570563, 'samples': 15600640, 'steps': 30469, 'loss/train': 3.1466665267944336}
+02/25/2022 17:03:19 - INFO - codeparrot_training - Step 30470: {'lr': 0.00017789870726569256, 'samples': 15601152, 'steps': 30470, 'loss/train': 1.4906448125839233}
+02/25/2022 17:03:23 - INFO - codeparrot_training - Step 30471: {'lr': 0.00017788304022453848, 'samples': 15601664, 'steps': 30471, 'loss/train': 2.7139439582824707}
+02/25/2022 17:03:28 - INFO - codeparrot_training - Step 30472: {'lr': 0.00017786737349231063, 'samples': 15602176, 'steps': 30472, 'loss/train': 1.6032030582427979}
+02/25/2022 17:03:32 - INFO - codeparrot_training - Step 30473: {'lr': 0.000177851707069076, 'samples': 15602688, 'steps': 30473, 'loss/train': 2.177809953689575}
+02/25/2022 17:03:38 - INFO - codeparrot_training - Step 30474: {'lr': 0.00017783604095490172, 'samples': 15603200, 'steps': 30474, 'loss/train': 1.9201726913452148}
+02/25/2022 17:03:41 - INFO - codeparrot_training - Step 30475: {'lr': 0.00017782037514985495, 'samples': 15603712, 'steps': 30475, 'loss/train': 2.48826265335083}
+02/25/2022 17:03:47 - INFO - codeparrot_training - Step 30476: {'lr': 0.0001778047096540027, 'samples': 15604224, 'steps': 30476, 'loss/train': 2.207005739212036}
+02/25/2022 17:03:51 - INFO - codeparrot_training - Step 30477: {'lr': 0.00017778904446741234, 'samples': 15604736, 'steps': 30477, 'loss/train': 1.539368987083435}
+02/25/2022 17:03:56 - INFO - codeparrot_training - Step 30478: {'lr': 0.00017777337959015065, 'samples': 15605248, 'steps': 30478, 'loss/train': 1.6143447160720825}
+02/25/2022 17:03:59 - INFO - codeparrot_training - Step 30479: {'lr': 0.00017775771502228486, 'samples': 15605760, 'steps': 30479, 'loss/train': 1.0108176469802856}
+02/25/2022 17:04:05 - INFO - codeparrot_training - Step 30480: {'lr': 0.00017774205076388205, 'samples': 15606272, 'steps': 30480, 'loss/train': 0.540143609046936}
+02/25/2022 17:04:08 - INFO - codeparrot_training - Step 30481: {'lr': 0.00017772638681500953, 'samples': 15606784, 'steps': 30481, 'loss/train': 1.286149024963379}
+02/25/2022 17:04:14 - INFO - codeparrot_training - Step 30482: {'lr': 0.00017771072317573406, 'samples': 15607296, 'steps': 30482, 'loss/train': 2.2609012126922607}
+02/25/2022 17:04:19 - INFO - codeparrot_training - Step 30483: {'lr': 0.00017769505984612292, 'samples': 15607808, 'steps': 30483, 'loss/train': 1.8654690980911255}
+02/25/2022 17:04:23 - INFO - codeparrot_training - Step 30484: {'lr': 0.00017767939682624317, 'samples': 15608320, 'steps': 30484, 'loss/train': 2.398495674133301}
+02/25/2022 17:04:29 - INFO - codeparrot_training - Step 30485: {'lr': 0.000177663734116162, 'samples': 15608832, 'steps': 30485, 'loss/train': 2.141735315322876}
+02/25/2022 17:04:32 - INFO - codeparrot_training - Step 30486: {'lr': 0.0001776480717159464, 'samples': 15609344, 'steps': 30486, 'loss/train': 2.3866524696350098}
+02/25/2022 17:04:38 - INFO - codeparrot_training - Step 30487: {'lr': 0.0001776324096256634, 'samples': 15609856, 'steps': 30487, 'loss/train': 2.574432849884033}
+02/25/2022 17:04:41 - INFO - codeparrot_training - Step 30488: {'lr': 0.0001776167478453802, 'samples': 15610368, 'steps': 30488, 'loss/train': 1.2526462078094482}
+02/25/2022 17:04:47 - INFO - codeparrot_training - Step 30489: {'lr': 0.0001776010863751639, 'samples': 15610880, 'steps': 30489, 'loss/train': 2.8575878143310547}
+02/25/2022 17:04:51 - INFO - codeparrot_training - Step 30490: {'lr': 0.00017758542521508164, 'samples': 15611392, 'steps': 30490, 'loss/train': 2.34999418258667}
+02/25/2022 17:04:54 - INFO - codeparrot_training - Step 30491: {'lr': 0.00017756976436520032, 'samples': 15611904, 'steps': 30491, 'loss/train': 1.517310380935669}
+02/25/2022 17:05:00 - INFO - codeparrot_training - Step 30492: {'lr': 0.0001775541038255871, 'samples': 15612416, 'steps': 30492, 'loss/train': 1.964463710784912}
+02/25/2022 17:05:05 - INFO - codeparrot_training - Step 30493: {'lr': 0.00017753844359630921, 'samples': 15612928, 'steps': 30493, 'loss/train': 1.4008477926254272}
+02/25/2022 17:05:09 - INFO - codeparrot_training - Step 30494: {'lr': 0.00017752278367743357, 'samples': 15613440, 'steps': 30494, 'loss/train': 1.7811930179595947}
+02/25/2022 17:05:14 - INFO - codeparrot_training - Step 30495: {'lr': 0.0001775071240690273, 'samples': 15613952, 'steps': 30495, 'loss/train': 1.6813509464263916}
+02/25/2022 17:05:18 - INFO - codeparrot_training - Step 30496: {'lr': 0.00017749146477115746, 'samples': 15614464, 'steps': 30496, 'loss/train': 1.497260570526123}
+02/25/2022 17:05:23 - INFO - codeparrot_training - Step 30497: {'lr': 0.0001774758057838912, 'samples': 15614976, 'steps': 30497, 'loss/train': 1.5442038774490356}
+02/25/2022 17:05:27 - INFO - codeparrot_training - Step 30498: {'lr': 0.00017746014710729558, 'samples': 15615488, 'steps': 30498, 'loss/train': 1.6781425476074219}
+02/25/2022 17:05:33 - INFO - codeparrot_training - Step 30499: {'lr': 0.00017744448874143758, 'samples': 15616000, 'steps': 30499, 'loss/train': 1.3502012491226196}
+02/25/2022 17:05:36 - INFO - codeparrot_training - Step 30500: {'lr': 0.00017742883068638446, 'samples': 15616512, 'steps': 30500, 'loss/train': 1.7134546041488647}
+02/25/2022 17:05:42 - INFO - codeparrot_training - Step 30501: {'lr': 0.0001774131729422031, 'samples': 15617024, 'steps': 30501, 'loss/train': 1.6110897064208984}
+02/25/2022 17:05:45 - INFO - codeparrot_training - Step 30502: {'lr': 0.00017739751550896067, 'samples': 15617536, 'steps': 30502, 'loss/train': 1.4763764142990112}
+02/25/2022 17:05:51 - INFO - codeparrot_training - Step 30503: {'lr': 0.00017738185838672425, 'samples': 15618048, 'steps': 30503, 'loss/train': 0.5471266508102417}
+02/25/2022 17:05:54 - INFO - codeparrot_training - Step 30504: {'lr': 0.0001773662015755609, 'samples': 15618560, 'steps': 30504, 'loss/train': 0.8086187839508057}
+02/25/2022 17:06:00 - INFO - codeparrot_training - Step 30505: {'lr': 0.00017735054507553766, 'samples': 15619072, 'steps': 30505, 'loss/train': 1.5706974267959595}
+02/25/2022 17:06:03 - INFO - codeparrot_training - Step 30506: {'lr': 0.00017733488888672155, 'samples': 15619584, 'steps': 30506, 'loss/train': 0.7623317837715149}
+02/25/2022 17:06:09 - INFO - codeparrot_training - Step 30507: {'lr': 0.00017731923300917982, 'samples': 15620096, 'steps': 30507, 'loss/train': 0.09161017090082169}
+02/25/2022 17:06:12 - INFO - codeparrot_training - Step 30508: {'lr': 0.00017730357744297938, 'samples': 15620608, 'steps': 30508, 'loss/train': 1.4521379470825195}
+02/25/2022 17:06:18 - INFO - codeparrot_training - Step 30509: {'lr': 0.0001772879221881873, 'samples': 15621120, 'steps': 30509, 'loss/train': 1.8131828308105469}
+02/25/2022 17:06:22 - INFO - codeparrot_training - Step 30510: {'lr': 0.00017727226724487068, 'samples': 15621632, 'steps': 30510, 'loss/train': 1.8775689601898193}
+02/25/2022 17:06:28 - INFO - codeparrot_training - Step 30511: {'lr': 0.00017725661261309666, 'samples': 15622144, 'steps': 30511, 'loss/train': 1.881587266921997}
+02/25/2022 17:06:31 - INFO - codeparrot_training - Step 30512: {'lr': 0.0001772409582929321, 'samples': 15622656, 'steps': 30512, 'loss/train': 1.0350226163864136}
+02/25/2022 17:06:37 - INFO - codeparrot_training - Step 30513: {'lr': 0.00017722530428444417, 'samples': 15623168, 'steps': 30513, 'loss/train': 1.8111931085586548}
+02/25/2022 17:06:40 - INFO - codeparrot_training - Step 30514: {'lr': 0.00017720965058769995, 'samples': 15623680, 'steps': 30514, 'loss/train': 2.302652597427368}
+02/25/2022 17:06:46 - INFO - codeparrot_training - Step 30515: {'lr': 0.00017719399720276646, 'samples': 15624192, 'steps': 30515, 'loss/train': 1.6977014541625977}
+02/25/2022 17:06:49 - INFO - codeparrot_training - Step 30516: {'lr': 0.00017717834412971087, 'samples': 15624704, 'steps': 30516, 'loss/train': 2.1854710578918457}
+02/25/2022 17:06:55 - INFO - codeparrot_training - Step 30517: {'lr': 0.00017716269136860003, 'samples': 15625216, 'steps': 30517, 'loss/train': 1.097787618637085}
+02/25/2022 17:06:58 - INFO - codeparrot_training - Step 30518: {'lr': 0.0001771470389195011, 'samples': 15625728, 'steps': 30518, 'loss/train': 1.9076862335205078}
+02/25/2022 17:07:04 - INFO - codeparrot_training - Step 30519: {'lr': 0.00017713138678248112, 'samples': 15626240, 'steps': 30519, 'loss/train': 2.024108648300171}
+02/25/2022 17:07:08 - INFO - codeparrot_training - Step 30520: {'lr': 0.00017711573495760724, 'samples': 15626752, 'steps': 30520, 'loss/train': 1.320483922958374}
+02/25/2022 17:07:13 - INFO - codeparrot_training - Step 30521: {'lr': 0.0001771000834449463, 'samples': 15627264, 'steps': 30521, 'loss/train': 1.3672194480895996}
+02/25/2022 17:07:17 - INFO - codeparrot_training - Step 30522: {'lr': 0.00017708443224456543, 'samples': 15627776, 'steps': 30522, 'loss/train': 1.7192773818969727}
+02/25/2022 17:07:22 - INFO - codeparrot_training - Step 30523: {'lr': 0.00017706878135653176, 'samples': 15628288, 'steps': 30523, 'loss/train': 1.680015206336975}
+02/25/2022 17:07:26 - INFO - codeparrot_training - Step 30524: {'lr': 0.00017705313078091235, 'samples': 15628800, 'steps': 30524, 'loss/train': 1.914446473121643}
+02/25/2022 17:07:32 - INFO - codeparrot_training - Step 30525: {'lr': 0.00017703748051777401, 'samples': 15629312, 'steps': 30525, 'loss/train': 1.5716078281402588}
+02/25/2022 17:07:35 - INFO - codeparrot_training - Step 30526: {'lr': 0.00017702183056718398, 'samples': 15629824, 'steps': 30526, 'loss/train': 1.9488903284072876}
+02/25/2022 17:07:40 - INFO - codeparrot_training - Step 30527: {'lr': 0.00017700618092920928, 'samples': 15630336, 'steps': 30527, 'loss/train': 2.2982089519500732}
+02/25/2022 17:07:44 - INFO - codeparrot_training - Step 30528: {'lr': 0.00017699053160391686, 'samples': 15630848, 'steps': 30528, 'loss/train': 2.0038654804229736}
+02/25/2022 17:07:50 - INFO - codeparrot_training - Step 30529: {'lr': 0.00017697488259137397, 'samples': 15631360, 'steps': 30529, 'loss/train': 1.3951107263565063}
+02/25/2022 17:07:53 - INFO - codeparrot_training - Step 30530: {'lr': 0.0001769592338916474, 'samples': 15631872, 'steps': 30530, 'loss/train': 1.6343458890914917}
+02/25/2022 17:07:59 - INFO - codeparrot_training - Step 30531: {'lr': 0.00017694358550480424, 'samples': 15632384, 'steps': 30531, 'loss/train': 1.818979024887085}
+02/25/2022 17:08:02 - INFO - codeparrot_training - Step 30532: {'lr': 0.00017692793743091158, 'samples': 15632896, 'steps': 30532, 'loss/train': 1.8359817266464233}
+02/25/2022 17:08:08 - INFO - codeparrot_training - Step 30533: {'lr': 0.00017691228967003654, 'samples': 15633408, 'steps': 30533, 'loss/train': 1.6695060729980469}
+02/25/2022 17:08:11 - INFO - codeparrot_training - Step 30534: {'lr': 0.00017689664222224593, 'samples': 15633920, 'steps': 30534, 'loss/train': 1.3436036109924316}
+02/25/2022 17:08:17 - INFO - codeparrot_training - Step 30535: {'lr': 0.0001768809950876069, 'samples': 15634432, 'steps': 30535, 'loss/train': 1.8788748979568481}
+02/25/2022 17:08:21 - INFO - codeparrot_training - Step 30536: {'lr': 0.00017686534826618646, 'samples': 15634944, 'steps': 30536, 'loss/train': 0.854434072971344}
+02/25/2022 17:08:27 - INFO - codeparrot_training - Step 30537: {'lr': 0.0001768497017580517, 'samples': 15635456, 'steps': 30537, 'loss/train': 0.04006991535425186}
+02/25/2022 17:08:30 - INFO - codeparrot_training - Step 30538: {'lr': 0.00017683405556326962, 'samples': 15635968, 'steps': 30538, 'loss/train': 1.76677405834198}
+02/25/2022 17:08:36 - INFO - codeparrot_training - Step 30539: {'lr': 0.00017681840968190713, 'samples': 15636480, 'steps': 30539, 'loss/train': 0.7660912275314331}
+02/25/2022 17:08:39 - INFO - codeparrot_training - Step 30540: {'lr': 0.00017680276411403135, 'samples': 15636992, 'steps': 30540, 'loss/train': 3.143925428390503}
+02/25/2022 17:08:45 - INFO - codeparrot_training - Step 30541: {'lr': 0.00017678711885970932, 'samples': 15637504, 'steps': 30541, 'loss/train': 2.062979221343994}
+02/25/2022 17:08:48 - INFO - codeparrot_training - Step 30542: {'lr': 0.00017677147391900806, 'samples': 15638016, 'steps': 30542, 'loss/train': 0.058564506471157074}
+02/25/2022 17:08:54 - INFO - codeparrot_training - Step 30543: {'lr': 0.00017675582929199446, 'samples': 15638528, 'steps': 30543, 'loss/train': 0.9685401320457458}
+02/25/2022 17:08:57 - INFO - codeparrot_training - Step 30544: {'lr': 0.00017674018497873566, 'samples': 15639040, 'steps': 30544, 'loss/train': 2.0894265174865723}
+02/25/2022 17:09:04 - INFO - codeparrot_training - Step 30545: {'lr': 0.00017672454097929864, 'samples': 15639552, 'steps': 30545, 'loss/train': 1.9795327186584473}
+02/25/2022 17:09:07 - INFO - codeparrot_training - Step 30546: {'lr': 0.00017670889729375048, 'samples': 15640064, 'steps': 30546, 'loss/train': 1.4028739929199219}
+02/25/2022 17:09:13 - INFO - codeparrot_training - Step 30547: {'lr': 0.00017669325392215808, 'samples': 15640576, 'steps': 30547, 'loss/train': 0.38878780603408813}
+02/25/2022 17:09:16 - INFO - codeparrot_training - Step 30548: {'lr': 0.0001766776108645885, 'samples': 15641088, 'steps': 30548, 'loss/train': 1.4374879598617554}
+02/25/2022 17:09:21 - INFO - codeparrot_training - Step 30549: {'lr': 0.00017666196812110875, 'samples': 15641600, 'steps': 30549, 'loss/train': 2.805933952331543}
+02/25/2022 17:09:25 - INFO - codeparrot_training - Step 30550: {'lr': 0.00017664632569178583, 'samples': 15642112, 'steps': 30550, 'loss/train': 1.6329450607299805}
+02/25/2022 17:09:30 - INFO - codeparrot_training - Step 30551: {'lr': 0.0001766306835766868, 'samples': 15642624, 'steps': 30551, 'loss/train': 2.0024988651275635}
+02/25/2022 17:09:34 - INFO - codeparrot_training - Step 30552: {'lr': 0.0001766150417758786, 'samples': 15643136, 'steps': 30552, 'loss/train': 1.7042843103408813}
+02/25/2022 17:09:39 - INFO - codeparrot_training - Step 30553: {'lr': 0.00017659940028942825, 'samples': 15643648, 'steps': 30553, 'loss/train': 2.143972396850586}
+02/25/2022 17:09:43 - INFO - codeparrot_training - Step 30554: {'lr': 0.00017658375911740277, 'samples': 15644160, 'steps': 30554, 'loss/train': 1.4576095342636108}
+02/25/2022 17:09:48 - INFO - codeparrot_training - Step 30555: {'lr': 0.00017656811825986923, 'samples': 15644672, 'steps': 30555, 'loss/train': 1.351486086845398}
+02/25/2022 17:09:52 - INFO - codeparrot_training - Step 30556: {'lr': 0.00017655247771689446, 'samples': 15645184, 'steps': 30556, 'loss/train': 2.148231029510498}
+02/25/2022 17:09:58 - INFO - codeparrot_training - Step 30557: {'lr': 0.00017653683748854553, 'samples': 15645696, 'steps': 30557, 'loss/train': 1.3356378078460693}
+02/25/2022 17:10:02 - INFO - codeparrot_training - Step 30558: {'lr': 0.00017652119757488949, 'samples': 15646208, 'steps': 30558, 'loss/train': 1.282679796218872}
+02/25/2022 17:10:07 - INFO - codeparrot_training - Step 30559: {'lr': 0.0001765055579759934, 'samples': 15646720, 'steps': 30559, 'loss/train': 1.8678544759750366}
+02/25/2022 17:10:10 - INFO - codeparrot_training - Step 30560: {'lr': 0.00017648991869192405, 'samples': 15647232, 'steps': 30560, 'loss/train': 1.091941237449646}
+02/25/2022 17:10:16 - INFO - codeparrot_training - Step 30561: {'lr': 0.00017647427972274858, 'samples': 15647744, 'steps': 30561, 'loss/train': 2.1383249759674072}
+02/25/2022 17:10:19 - INFO - codeparrot_training - Step 30562: {'lr': 0.00017645864106853392, 'samples': 15648256, 'steps': 30562, 'loss/train': 2.4352896213531494}
+02/25/2022 17:10:25 - INFO - codeparrot_training - Step 30563: {'lr': 0.0001764430027293471, 'samples': 15648768, 'steps': 30563, 'loss/train': 1.6953250169754028}
+02/25/2022 17:10:28 - INFO - codeparrot_training - Step 30564: {'lr': 0.0001764273647052552, 'samples': 15649280, 'steps': 30564, 'loss/train': 2.5382022857666016}
+02/25/2022 17:10:34 - INFO - codeparrot_training - Step 30565: {'lr': 0.000176411726996325, 'samples': 15649792, 'steps': 30565, 'loss/train': 0.9479430913925171}
+02/25/2022 17:10:37 - INFO - codeparrot_training - Step 30566: {'lr': 0.0001763960896026236, 'samples': 15650304, 'steps': 30566, 'loss/train': 1.8191521167755127}
+02/25/2022 17:10:44 - INFO - codeparrot_training - Step 30567: {'lr': 0.00017638045252421798, 'samples': 15650816, 'steps': 30567, 'loss/train': 1.8058654069900513}
+02/25/2022 17:10:47 - INFO - codeparrot_training - Step 30568: {'lr': 0.00017636481576117522, 'samples': 15651328, 'steps': 30568, 'loss/train': 1.1017128229141235}
+02/25/2022 17:10:53 - INFO - codeparrot_training - Step 30569: {'lr': 0.0001763491793135621, 'samples': 15651840, 'steps': 30569, 'loss/train': 2.018596649169922}
+02/25/2022 17:10:56 - INFO - codeparrot_training - Step 30570: {'lr': 0.00017633354318144574, 'samples': 15652352, 'steps': 30570, 'loss/train': 1.7992150783538818}
+02/25/2022 17:11:02 - INFO - codeparrot_training - Step 30571: {'lr': 0.00017631790736489307, 'samples': 15652864, 'steps': 30571, 'loss/train': 0.16332975029945374}
+02/25/2022 17:11:05 - INFO - codeparrot_training - Step 30572: {'lr': 0.00017630227186397118, 'samples': 15653376, 'steps': 30572, 'loss/train': 1.3902952671051025}
+02/25/2022 17:11:11 - INFO - codeparrot_training - Step 30573: {'lr': 0.00017628663667874683, 'samples': 15653888, 'steps': 30573, 'loss/train': 1.0414601564407349}
+02/25/2022 17:11:14 - INFO - codeparrot_training - Step 30574: {'lr': 0.00017627100180928718, 'samples': 15654400, 'steps': 30574, 'loss/train': 2.0970213413238525}
+02/25/2022 17:11:20 - INFO - codeparrot_training - Step 30575: {'lr': 0.0001762553672556591, 'samples': 15654912, 'steps': 30575, 'loss/train': 2.0458486080169678}
+02/25/2022 17:11:23 - INFO - codeparrot_training - Step 30576: {'lr': 0.00017623973301792963, 'samples': 15655424, 'steps': 30576, 'loss/train': 1.5893223285675049}
+02/25/2022 17:11:29 - INFO - codeparrot_training - Step 30577: {'lr': 0.00017622409909616582, 'samples': 15655936, 'steps': 30577, 'loss/train': 1.4068567752838135}
+02/25/2022 17:11:34 - INFO - codeparrot_training - Step 30578: {'lr': 0.00017620846549043446, 'samples': 15656448, 'steps': 30578, 'loss/train': 1.2612806558609009}
+02/25/2022 17:11:38 - INFO - codeparrot_training - Step 30579: {'lr': 0.00017619283220080258, 'samples': 15656960, 'steps': 30579, 'loss/train': 1.2561819553375244}
+02/25/2022 17:11:44 - INFO - codeparrot_training - Step 30580: {'lr': 0.0001761771992273372, 'samples': 15657472, 'steps': 30580, 'loss/train': 1.2824877500534058}
+02/25/2022 17:11:48 - INFO - codeparrot_training - Step 30581: {'lr': 0.00017616156657010532, 'samples': 15657984, 'steps': 30581, 'loss/train': 1.796589970588684}
+02/25/2022 17:11:53 - INFO - codeparrot_training - Step 30582: {'lr': 0.00017614593422917373, 'samples': 15658496, 'steps': 30582, 'loss/train': 1.1098741292953491}
+02/25/2022 17:11:57 - INFO - codeparrot_training - Step 30583: {'lr': 0.00017613030220460957, 'samples': 15659008, 'steps': 30583, 'loss/train': 0.7071213126182556}
+02/25/2022 17:12:02 - INFO - codeparrot_training - Step 30584: {'lr': 0.00017611467049647976, 'samples': 15659520, 'steps': 30584, 'loss/train': 2.6004810333251953}
+02/25/2022 17:12:06 - INFO - codeparrot_training - Step 30585: {'lr': 0.00017609903910485118, 'samples': 15660032, 'steps': 30585, 'loss/train': 1.9710332155227661}
+02/25/2022 17:12:11 - INFO - codeparrot_training - Step 30586: {'lr': 0.0001760834080297909, 'samples': 15660544, 'steps': 30586, 'loss/train': 0.32962745428085327}
+02/25/2022 17:12:15 - INFO - codeparrot_training - Step 30587: {'lr': 0.0001760677772713658, 'samples': 15661056, 'steps': 30587, 'loss/train': 2.142575740814209}
+02/25/2022 17:12:20 - INFO - codeparrot_training - Step 30588: {'lr': 0.00017605214682964287, 'samples': 15661568, 'steps': 30588, 'loss/train': 0.9723318219184875}
+02/25/2022 17:12:24 - INFO - codeparrot_training - Step 30589: {'lr': 0.00017603651670468907, 'samples': 15662080, 'steps': 30589, 'loss/train': 1.0869251489639282}
+02/25/2022 17:12:30 - INFO - codeparrot_training - Step 30590: {'lr': 0.00017602088689657136, 'samples': 15662592, 'steps': 30590, 'loss/train': 3.337860345840454}
+02/25/2022 17:12:34 - INFO - codeparrot_training - Step 30591: {'lr': 0.00017600525740535668, 'samples': 15663104, 'steps': 30591, 'loss/train': 0.8833999633789062}
+02/25/2022 17:12:39 - INFO - codeparrot_training - Step 30592: {'lr': 0.00017598962823111192, 'samples': 15663616, 'steps': 30592, 'loss/train': 1.1802172660827637}
+02/25/2022 17:12:43 - INFO - codeparrot_training - Step 30593: {'lr': 0.00017597399937390414, 'samples': 15664128, 'steps': 30593, 'loss/train': 1.6584898233413696}
+02/25/2022 17:12:48 - INFO - codeparrot_training - Step 30594: {'lr': 0.00017595837083380026, 'samples': 15664640, 'steps': 30594, 'loss/train': 1.86874520778656}
+02/25/2022 17:12:52 - INFO - codeparrot_training - Step 30595: {'lr': 0.0001759427426108672, 'samples': 15665152, 'steps': 30595, 'loss/train': 3.5682506561279297}
+02/25/2022 17:12:57 - INFO - codeparrot_training - Step 30596: {'lr': 0.00017592711470517186, 'samples': 15665664, 'steps': 30596, 'loss/train': 1.6682144403457642}
+02/25/2022 17:13:01 - INFO - codeparrot_training - Step 30597: {'lr': 0.0001759114871167813, 'samples': 15666176, 'steps': 30597, 'loss/train': 0.6292073130607605}
+02/25/2022 17:13:06 - INFO - codeparrot_training - Step 30598: {'lr': 0.00017589585984576235, 'samples': 15666688, 'steps': 30598, 'loss/train': 0.6108160614967346}
+02/25/2022 17:13:10 - INFO - codeparrot_training - Step 30599: {'lr': 0.00017588023289218208, 'samples': 15667200, 'steps': 30599, 'loss/train': 1.347469687461853}
+02/25/2022 17:13:15 - INFO - codeparrot_training - Step 30600: {'lr': 0.0001758646062561073, 'samples': 15667712, 'steps': 30600, 'loss/train': 0.13789142668247223}
+02/25/2022 17:13:19 - INFO - codeparrot_training - Step 30601: {'lr': 0.000175848979937605, 'samples': 15668224, 'steps': 30601, 'loss/train': 1.4443327188491821}
+02/25/2022 17:13:25 - INFO - codeparrot_training - Step 30602: {'lr': 0.0001758333539367421, 'samples': 15668736, 'steps': 30602, 'loss/train': 1.9768928289413452}
+02/25/2022 17:13:28 - INFO - codeparrot_training - Step 30603: {'lr': 0.0001758177282535857, 'samples': 15669248, 'steps': 30603, 'loss/train': 1.5133295059204102}
+02/25/2022 17:13:34 - INFO - codeparrot_training - Step 30604: {'lr': 0.00017580210288820248, 'samples': 15669760, 'steps': 30604, 'loss/train': 3.128160238265991}
+02/25/2022 17:13:37 - INFO - codeparrot_training - Step 30605: {'lr': 0.0001757864778406595, 'samples': 15670272, 'steps': 30605, 'loss/train': 1.8558063507080078}
+02/25/2022 17:13:43 - INFO - codeparrot_training - Step 30606: {'lr': 0.00017577085311102364, 'samples': 15670784, 'steps': 30606, 'loss/train': 0.8960566520690918}
+02/25/2022 17:13:46 - INFO - codeparrot_training - Step 30607: {'lr': 0.00017575522869936203, 'samples': 15671296, 'steps': 30607, 'loss/train': 0.9792315363883972}
+02/25/2022 17:13:53 - INFO - codeparrot_training - Step 30608: {'lr': 0.00017573960460574132, 'samples': 15671808, 'steps': 30608, 'loss/train': 1.8218752145767212}
+02/25/2022 17:13:56 - INFO - codeparrot_training - Step 30609: {'lr': 0.00017572398083022856, 'samples': 15672320, 'steps': 30609, 'loss/train': 0.870544970035553}
+02/25/2022 17:14:02 - INFO - codeparrot_training - Step 30610: {'lr': 0.0001757083573728907, 'samples': 15672832, 'steps': 30610, 'loss/train': 0.1313645988702774}
+02/25/2022 17:14:06 - INFO - codeparrot_training - Step 30611: {'lr': 0.0001756927342337946, 'samples': 15673344, 'steps': 30611, 'loss/train': 1.1649972200393677}
+02/25/2022 17:14:11 - INFO - codeparrot_training - Step 30612: {'lr': 0.0001756771114130074, 'samples': 15673856, 'steps': 30612, 'loss/train': 2.251112937927246}
+02/25/2022 17:14:15 - INFO - codeparrot_training - Step 30613: {'lr': 0.00017566148891059568, 'samples': 15674368, 'steps': 30613, 'loss/train': 0.2405310571193695}
+02/25/2022 17:14:20 - INFO - codeparrot_training - Step 30614: {'lr': 0.0001756458667266266, 'samples': 15674880, 'steps': 30614, 'loss/train': 1.2505325078964233}
+02/25/2022 17:14:24 - INFO - codeparrot_training - Step 30615: {'lr': 0.00017563024486116698, 'samples': 15675392, 'steps': 30615, 'loss/train': 1.6216870546340942}
+02/25/2022 17:14:29 - INFO - codeparrot_training - Step 30616: {'lr': 0.0001756146233142839, 'samples': 15675904, 'steps': 30616, 'loss/train': 1.4520866870880127}
+02/25/2022 17:14:33 - INFO - codeparrot_training - Step 30617: {'lr': 0.00017559900208604404, 'samples': 15676416, 'steps': 30617, 'loss/train': 0.8889603614807129}
+02/25/2022 17:14:39 - INFO - codeparrot_training - Step 30618: {'lr': 0.00017558338117651447, 'samples': 15676928, 'steps': 30618, 'loss/train': 1.3003312349319458}
+02/25/2022 17:14:42 - INFO - codeparrot_training - Step 30619: {'lr': 0.000175567760585762, 'samples': 15677440, 'steps': 30619, 'loss/train': 1.8794946670532227}
+02/25/2022 17:14:48 - INFO - codeparrot_training - Step 30620: {'lr': 0.00017555214031385376, 'samples': 15677952, 'steps': 30620, 'loss/train': 0.7604542970657349}
+02/25/2022 17:14:51 - INFO - codeparrot_training - Step 30621: {'lr': 0.00017553652036085638, 'samples': 15678464, 'steps': 30621, 'loss/train': 1.744019627571106}
+02/25/2022 17:14:57 - INFO - codeparrot_training - Step 30622: {'lr': 0.00017552090072683692, 'samples': 15678976, 'steps': 30622, 'loss/train': 1.0111384391784668}
+02/25/2022 17:15:00 - INFO - codeparrot_training - Step 30623: {'lr': 0.00017550528141186222, 'samples': 15679488, 'steps': 30623, 'loss/train': 0.04356483370065689}
+02/25/2022 17:15:06 - INFO - codeparrot_training - Step 30624: {'lr': 0.00017548966241599933, 'samples': 15680000, 'steps': 30624, 'loss/train': 1.5989981889724731}
+02/25/2022 17:15:09 - INFO - codeparrot_training - Step 30625: {'lr': 0.00017547404373931507, 'samples': 15680512, 'steps': 30625, 'loss/train': 2.618983030319214}
+02/25/2022 17:15:15 - INFO - codeparrot_training - Step 30626: {'lr': 0.0001754584253818763, 'samples': 15681024, 'steps': 30626, 'loss/train': 1.4781371355056763}
+02/25/2022 17:15:18 - INFO - codeparrot_training - Step 30627: {'lr': 0.00017544280734374995, 'samples': 15681536, 'steps': 30627, 'loss/train': 1.9581143856048584}
+02/25/2022 17:15:25 - INFO - codeparrot_training - Step 30628: {'lr': 0.00017542718962500298, 'samples': 15682048, 'steps': 30628, 'loss/train': 1.5257078409194946}
+02/25/2022 17:15:28 - INFO - codeparrot_training - Step 30629: {'lr': 0.00017541157222570228, 'samples': 15682560, 'steps': 30629, 'loss/train': 1.7244051694869995}
+02/25/2022 17:15:34 - INFO - codeparrot_training - Step 30630: {'lr': 0.00017539595514591466, 'samples': 15683072, 'steps': 30630, 'loss/train': 2.757523536682129}
+02/25/2022 17:15:37 - INFO - codeparrot_training - Step 30631: {'lr': 0.00017538033838570711, 'samples': 15683584, 'steps': 30631, 'loss/train': 0.046574871987104416}
+02/25/2022 17:15:43 - INFO - codeparrot_training - Step 30632: {'lr': 0.00017536472194514647, 'samples': 15684096, 'steps': 30632, 'loss/train': 1.671276569366455}
+02/25/2022 17:15:48 - INFO - codeparrot_training - Step 30633: {'lr': 0.0001753491058242997, 'samples': 15684608, 'steps': 30633, 'loss/train': 2.4795539379119873}
+02/25/2022 17:15:52 - INFO - codeparrot_training - Step 30634: {'lr': 0.0001753334900232336, 'samples': 15685120, 'steps': 30634, 'loss/train': 2.11317777633667}
+02/25/2022 17:15:58 - INFO - codeparrot_training - Step 30635: {'lr': 0.00017531787454201517, 'samples': 15685632, 'steps': 30635, 'loss/train': 1.6909476518630981}
+02/25/2022 17:16:01 - INFO - codeparrot_training - Step 30636: {'lr': 0.0001753022593807112, 'samples': 15686144, 'steps': 30636, 'loss/train': 2.088340997695923}
+02/25/2022 17:16:04 - INFO - codeparrot_training - Step 30637: {'lr': 0.00017528664453938863, 'samples': 15686656, 'steps': 30637, 'loss/train': 2.3470187187194824}
+02/25/2022 17:16:11 - INFO - codeparrot_training - Step 30638: {'lr': 0.00017527103001811447, 'samples': 15687168, 'steps': 30638, 'loss/train': 0.9760718941688538}
+02/25/2022 17:16:14 - INFO - codeparrot_training - Step 30639: {'lr': 0.0001752554158169554, 'samples': 15687680, 'steps': 30639, 'loss/train': 1.9636294841766357}
+02/25/2022 17:16:20 - INFO - codeparrot_training - Step 30640: {'lr': 0.00017523980193597836, 'samples': 15688192, 'steps': 30640, 'loss/train': 1.869028925895691}
+02/25/2022 17:16:23 - INFO - codeparrot_training - Step 30641: {'lr': 0.00017522418837525027, 'samples': 15688704, 'steps': 30641, 'loss/train': 2.451420545578003}
+02/25/2022 17:16:29 - INFO - codeparrot_training - Step 30642: {'lr': 0.00017520857513483813, 'samples': 15689216, 'steps': 30642, 'loss/train': 2.1372947692871094}
+02/25/2022 17:16:32 - INFO - codeparrot_training - Step 30643: {'lr': 0.00017519296221480863, 'samples': 15689728, 'steps': 30643, 'loss/train': 2.002358913421631}
+02/25/2022 17:16:38 - INFO - codeparrot_training - Step 30644: {'lr': 0.0001751773496152287, 'samples': 15690240, 'steps': 30644, 'loss/train': 1.2552012205123901}
+02/25/2022 17:16:41 - INFO - codeparrot_training - Step 30645: {'lr': 0.0001751617373361652, 'samples': 15690752, 'steps': 30645, 'loss/train': 2.1874241828918457}
+02/25/2022 17:16:47 - INFO - codeparrot_training - Step 30646: {'lr': 0.0001751461253776851, 'samples': 15691264, 'steps': 30646, 'loss/train': 0.030398182570934296}
+02/25/2022 17:16:50 - INFO - codeparrot_training - Step 30647: {'lr': 0.00017513051373985533, 'samples': 15691776, 'steps': 30647, 'loss/train': 1.8538587093353271}
+02/25/2022 17:16:56 - INFO - codeparrot_training - Step 30648: {'lr': 0.00017511490242274255, 'samples': 15692288, 'steps': 30648, 'loss/train': 3.050234317779541}
+02/25/2022 17:17:01 - INFO - codeparrot_training - Step 30649: {'lr': 0.00017509929142641377, 'samples': 15692800, 'steps': 30649, 'loss/train': 1.6641796827316284}
+02/25/2022 17:17:04 - INFO - codeparrot_training - Step 30650: {'lr': 0.00017508368075093582, 'samples': 15693312, 'steps': 30650, 'loss/train': 2.202838897705078}
+02/25/2022 17:17:10 - INFO - codeparrot_training - Step 30651: {'lr': 0.00017506807039637573, 'samples': 15693824, 'steps': 30651, 'loss/train': 0.5823962092399597}
+02/25/2022 17:17:13 - INFO - codeparrot_training - Step 30652: {'lr': 0.0001750524603628001, 'samples': 15694336, 'steps': 30652, 'loss/train': 1.0130858421325684}
+02/25/2022 17:17:20 - INFO - codeparrot_training - Step 30653: {'lr': 0.0001750368506502759, 'samples': 15694848, 'steps': 30653, 'loss/train': 2.7226009368896484}
+02/25/2022 17:17:23 - INFO - codeparrot_training - Step 30654: {'lr': 0.0001750212412588701, 'samples': 15695360, 'steps': 30654, 'loss/train': 0.23124033212661743}
+02/25/2022 17:17:29 - INFO - codeparrot_training - Step 30655: {'lr': 0.00017500563218864957, 'samples': 15695872, 'steps': 30655, 'loss/train': 0.8997575044631958}
+02/25/2022 17:17:32 - INFO - codeparrot_training - Step 30656: {'lr': 0.00017499002343968097, 'samples': 15696384, 'steps': 30656, 'loss/train': 2.005222797393799}
+02/25/2022 17:17:38 - INFO - codeparrot_training - Step 30657: {'lr': 0.00017497441501203133, 'samples': 15696896, 'steps': 30657, 'loss/train': 1.5784491300582886}
+02/25/2022 17:17:41 - INFO - codeparrot_training - Step 30658: {'lr': 0.00017495880690576742, 'samples': 15697408, 'steps': 30658, 'loss/train': 0.8188974261283875}
+02/25/2022 17:17:47 - INFO - codeparrot_training - Step 30659: {'lr': 0.00017494319912095618, 'samples': 15697920, 'steps': 30659, 'loss/train': 1.558914303779602}
+02/25/2022 17:17:50 - INFO - codeparrot_training - Step 30660: {'lr': 0.00017492759165766458, 'samples': 15698432, 'steps': 30660, 'loss/train': 2.6022419929504395}
+02/25/2022 17:17:56 - INFO - codeparrot_training - Step 30661: {'lr': 0.0001749119845159592, 'samples': 15698944, 'steps': 30661, 'loss/train': 2.226147174835205}
+02/25/2022 17:17:59 - INFO - codeparrot_training - Step 30662: {'lr': 0.00017489637769590706, 'samples': 15699456, 'steps': 30662, 'loss/train': 1.746221899986267}
+02/25/2022 17:18:05 - INFO - codeparrot_training - Step 30663: {'lr': 0.00017488077119757496, 'samples': 15699968, 'steps': 30663, 'loss/train': 1.8763507604599}
+02/25/2022 17:18:09 - INFO - codeparrot_training - Step 30664: {'lr': 0.0001748651650210299, 'samples': 15700480, 'steps': 30664, 'loss/train': 1.6448935270309448}
+02/25/2022 17:18:14 - INFO - codeparrot_training - Step 30665: {'lr': 0.00017484955916633848, 'samples': 15700992, 'steps': 30665, 'loss/train': 0.02377876080572605}
+02/25/2022 17:18:18 - INFO - codeparrot_training - Step 30666: {'lr': 0.00017483395363356775, 'samples': 15701504, 'steps': 30666, 'loss/train': 1.5679619312286377}
+02/25/2022 17:18:23 - INFO - codeparrot_training - Step 30667: {'lr': 0.00017481834842278444, 'samples': 15702016, 'steps': 30667, 'loss/train': 0.10999450832605362}
+02/25/2022 17:18:27 - INFO - codeparrot_training - Step 30668: {'lr': 0.00017480274353405558, 'samples': 15702528, 'steps': 30668, 'loss/train': 0.04276614636182785}
+02/25/2022 17:18:32 - INFO - codeparrot_training - Step 30669: {'lr': 0.0001747871389674478, 'samples': 15703040, 'steps': 30669, 'loss/train': 1.3618367910385132}
+02/25/2022 17:18:36 - INFO - codeparrot_training - Step 30670: {'lr': 0.00017477153472302798, 'samples': 15703552, 'steps': 30670, 'loss/train': 2.252397298812866}
+02/25/2022 17:18:41 - INFO - codeparrot_training - Step 30671: {'lr': 0.0001747559308008631, 'samples': 15704064, 'steps': 30671, 'loss/train': 1.4648576974868774}
+02/25/2022 17:18:45 - INFO - codeparrot_training - Step 30672: {'lr': 0.0001747403272010199, 'samples': 15704576, 'steps': 30672, 'loss/train': 2.3063623905181885}
+02/25/2022 17:18:51 - INFO - codeparrot_training - Step 30673: {'lr': 0.00017472472392356526, 'samples': 15705088, 'steps': 30673, 'loss/train': 0.09646202623844147}
+02/25/2022 17:18:54 - INFO - codeparrot_training - Step 30674: {'lr': 0.00017470912096856594, 'samples': 15705600, 'steps': 30674, 'loss/train': 2.003654718399048}
+02/25/2022 17:19:00 - INFO - codeparrot_training - Step 30675: {'lr': 0.00017469351833608894, 'samples': 15706112, 'steps': 30675, 'loss/train': 1.76346755027771}
+02/25/2022 17:19:04 - INFO - codeparrot_training - Step 30676: {'lr': 0.00017467791602620086, 'samples': 15706624, 'steps': 30676, 'loss/train': 2.195906400680542}
+02/25/2022 17:19:09 - INFO - codeparrot_training - Step 30677: {'lr': 0.00017466231403896878, 'samples': 15707136, 'steps': 30677, 'loss/train': 1.9850581884384155}
+02/25/2022 17:19:13 - INFO - codeparrot_training - Step 30678: {'lr': 0.00017464671237445942, 'samples': 15707648, 'steps': 30678, 'loss/train': 1.6075907945632935}
+02/25/2022 17:19:18 - INFO - codeparrot_training - Step 30679: {'lr': 0.00017463111103273958, 'samples': 15708160, 'steps': 30679, 'loss/train': 2.813119888305664}
+02/25/2022 17:19:21 - INFO - codeparrot_training - Step 30680: {'lr': 0.0001746155100138761, 'samples': 15708672, 'steps': 30680, 'loss/train': 1.9931228160858154}
+02/25/2022 17:19:27 - INFO - codeparrot_training - Step 30681: {'lr': 0.0001745999093179359, 'samples': 15709184, 'steps': 30681, 'loss/train': 2.007575273513794}
+02/25/2022 17:19:30 - INFO - codeparrot_training - Step 30682: {'lr': 0.00017458430894498573, 'samples': 15709696, 'steps': 30682, 'loss/train': 2.521340847015381}
+02/25/2022 17:19:36 - INFO - codeparrot_training - Step 30683: {'lr': 0.00017456870889509242, 'samples': 15710208, 'steps': 30683, 'loss/train': 1.8616303205490112}
+02/25/2022 17:19:39 - INFO - codeparrot_training - Step 30684: {'lr': 0.00017455310916832275, 'samples': 15710720, 'steps': 30684, 'loss/train': 0.9796485304832458}
+02/25/2022 17:19:45 - INFO - codeparrot_training - Step 30685: {'lr': 0.00017453750976474363, 'samples': 15711232, 'steps': 30685, 'loss/train': 0.5382975935935974}
+02/25/2022 17:19:48 - INFO - codeparrot_training - Step 30686: {'lr': 0.00017452191068442196, 'samples': 15711744, 'steps': 30686, 'loss/train': 1.6235179901123047}
+02/25/2022 17:19:54 - INFO - codeparrot_training - Step 30687: {'lr': 0.0001745063119274244, 'samples': 15712256, 'steps': 30687, 'loss/train': 2.1804840564727783}
+02/25/2022 17:19:57 - INFO - codeparrot_training - Step 30688: {'lr': 0.00017449071349381776, 'samples': 15712768, 'steps': 30688, 'loss/train': 1.2717500925064087}
+02/25/2022 17:20:04 - INFO - codeparrot_training - Step 30689: {'lr': 0.00017447511538366895, 'samples': 15713280, 'steps': 30689, 'loss/train': 1.976916790008545}
+02/25/2022 17:20:07 - INFO - codeparrot_training - Step 30690: {'lr': 0.00017445951759704486, 'samples': 15713792, 'steps': 30690, 'loss/train': 1.6629647016525269}
+02/25/2022 17:20:13 - INFO - codeparrot_training - Step 30691: {'lr': 0.00017444392013401213, 'samples': 15714304, 'steps': 30691, 'loss/train': 1.923917531967163}
+02/25/2022 17:20:18 - INFO - codeparrot_training - Step 30692: {'lr': 0.00017442832299463762, 'samples': 15714816, 'steps': 30692, 'loss/train': 1.995125412940979}
+02/25/2022 17:20:21 - INFO - codeparrot_training - Step 30693: {'lr': 0.00017441272617898824, 'samples': 15715328, 'steps': 30693, 'loss/train': 1.0246213674545288}
+02/25/2022 17:20:27 - INFO - codeparrot_training - Step 30694: {'lr': 0.00017439712968713068, 'samples': 15715840, 'steps': 30694, 'loss/train': 1.8903381824493408}
+02/25/2022 17:20:31 - INFO - codeparrot_training - Step 30695: {'lr': 0.00017438153351913193, 'samples': 15716352, 'steps': 30695, 'loss/train': 0.8537416458129883}
+02/25/2022 17:20:36 - INFO - codeparrot_training - Step 30696: {'lr': 0.00017436593767505859, 'samples': 15716864, 'steps': 30696, 'loss/train': 1.5773813724517822}
+02/25/2022 17:20:40 - INFO - codeparrot_training - Step 30697: {'lr': 0.00017435034215497754, 'samples': 15717376, 'steps': 30697, 'loss/train': 1.8662402629852295}
+02/25/2022 17:20:46 - INFO - codeparrot_training - Step 30698: {'lr': 0.0001743347469589556, 'samples': 15717888, 'steps': 30698, 'loss/train': 1.4936827421188354}
+02/25/2022 17:20:49 - INFO - codeparrot_training - Step 30699: {'lr': 0.00017431915208705967, 'samples': 15718400, 'steps': 30699, 'loss/train': 2.2123234272003174}
+02/25/2022 17:20:55 - INFO - codeparrot_training - Step 30700: {'lr': 0.0001743035575393564, 'samples': 15718912, 'steps': 30700, 'loss/train': 1.4433770179748535}
+02/25/2022 17:20:58 - INFO - codeparrot_training - Step 30701: {'lr': 0.00017428796331591263, 'samples': 15719424, 'steps': 30701, 'loss/train': 2.867765188217163}
+02/25/2022 17:21:04 - INFO - codeparrot_training - Step 30702: {'lr': 0.00017427236941679517, 'samples': 15719936, 'steps': 30702, 'loss/train': 1.2954092025756836}
+02/25/2022 17:21:07 - INFO - codeparrot_training - Step 30703: {'lr': 0.00017425677584207095, 'samples': 15720448, 'steps': 30703, 'loss/train': 2.0587944984436035}
+02/25/2022 17:21:13 - INFO - codeparrot_training - Step 30704: {'lr': 0.00017424118259180656, 'samples': 15720960, 'steps': 30704, 'loss/train': 1.3188984394073486}
+02/25/2022 17:21:16 - INFO - codeparrot_training - Step 30705: {'lr': 0.00017422558966606887, 'samples': 15721472, 'steps': 30705, 'loss/train': 1.73975670337677}
+02/25/2022 17:21:22 - INFO - codeparrot_training - Step 30706: {'lr': 0.0001742099970649247, 'samples': 15721984, 'steps': 30706, 'loss/train': 1.804073452949524}
+02/25/2022 17:21:25 - INFO - codeparrot_training - Step 30707: {'lr': 0.00017419440478844082, 'samples': 15722496, 'steps': 30707, 'loss/train': 1.6832637786865234}
+02/25/2022 17:21:31 - INFO - codeparrot_training - Step 30708: {'lr': 0.00017417881283668412, 'samples': 15723008, 'steps': 30708, 'loss/train': 2.0426106452941895}
+02/25/2022 17:21:34 - INFO - codeparrot_training - Step 30709: {'lr': 0.00017416322120972127, 'samples': 15723520, 'steps': 30709, 'loss/train': 0.1720254123210907}
+02/25/2022 17:21:40 - INFO - codeparrot_training - Step 30710: {'lr': 0.00017414762990761908, 'samples': 15724032, 'steps': 30710, 'loss/train': 1.8328558206558228}
+02/25/2022 17:21:44 - INFO - codeparrot_training - Step 30711: {'lr': 0.0001741320389304444, 'samples': 15724544, 'steps': 30711, 'loss/train': 1.8271600008010864}
+02/25/2022 17:21:49 - INFO - codeparrot_training - Step 30712: {'lr': 0.00017411644827826394, 'samples': 15725056, 'steps': 30712, 'loss/train': 1.6123133897781372}
+02/25/2022 17:21:53 - INFO - codeparrot_training - Step 30713: {'lr': 0.0001741008579511445, 'samples': 15725568, 'steps': 30713, 'loss/train': 1.570552110671997}
+02/25/2022 17:21:58 - INFO - codeparrot_training - Step 30714: {'lr': 0.00017408526794915292, 'samples': 15726080, 'steps': 30714, 'loss/train': 1.0803930759429932}
+02/25/2022 17:22:02 - INFO - codeparrot_training - Step 30715: {'lr': 0.00017406967827235598, 'samples': 15726592, 'steps': 30715, 'loss/train': 2.446274995803833}
+02/25/2022 17:22:07 - INFO - codeparrot_training - Step 30716: {'lr': 0.0001740540889208204, 'samples': 15727104, 'steps': 30716, 'loss/train': 1.8607929944992065}
+02/25/2022 17:22:11 - INFO - codeparrot_training - Step 30717: {'lr': 0.00017403849989461296, 'samples': 15727616, 'steps': 30717, 'loss/train': 1.765053153038025}
+02/25/2022 17:22:16 - INFO - codeparrot_training - Step 30718: {'lr': 0.00017402291119380047, 'samples': 15728128, 'steps': 30718, 'loss/train': 2.0131499767303467}
+02/25/2022 17:22:20 - INFO - codeparrot_training - Step 30719: {'lr': 0.0001740073228184497, 'samples': 15728640, 'steps': 30719, 'loss/train': 1.424175500869751}
+02/25/2022 17:22:26 - INFO - codeparrot_training - Step 30720: {'lr': 0.00017399173476862742, 'samples': 15729152, 'steps': 30720, 'loss/train': 3.1434195041656494}
+02/25/2022 17:22:29 - INFO - codeparrot_training - Step 30721: {'lr': 0.00017397614704440045, 'samples': 15729664, 'steps': 30721, 'loss/train': 0.7791910171508789}
+02/25/2022 17:22:35 - INFO - codeparrot_training - Step 30722: {'lr': 0.00017396055964583558, 'samples': 15730176, 'steps': 30722, 'loss/train': 1.2293877601623535}
+02/25/2022 17:22:38 - INFO - codeparrot_training - Step 30723: {'lr': 0.00017394497257299942, 'samples': 15730688, 'steps': 30723, 'loss/train': 2.2874598503112793}
+02/25/2022 17:22:44 - INFO - codeparrot_training - Step 30724: {'lr': 0.00017392938582595885, 'samples': 15731200, 'steps': 30724, 'loss/train': 1.5270161628723145}
+02/25/2022 17:22:47 - INFO - codeparrot_training - Step 30725: {'lr': 0.00017391379940478075, 'samples': 15731712, 'steps': 30725, 'loss/train': 2.097254991531372}
+02/25/2022 17:22:53 - INFO - codeparrot_training - Step 30726: {'lr': 0.00017389821330953165, 'samples': 15732224, 'steps': 30726, 'loss/train': 2.503765821456909}
+02/25/2022 17:22:56 - INFO - codeparrot_training - Step 30727: {'lr': 0.0001738826275402785, 'samples': 15732736, 'steps': 30727, 'loss/train': 3.621640920639038}
+02/25/2022 17:23:02 - INFO - codeparrot_training - Step 30728: {'lr': 0.00017386704209708794, 'samples': 15733248, 'steps': 30728, 'loss/train': 2.242464303970337}
+02/25/2022 17:23:05 - INFO - codeparrot_training - Step 30729: {'lr': 0.0001738514569800269, 'samples': 15733760, 'steps': 30729, 'loss/train': 2.215031147003174}
+02/25/2022 17:23:11 - INFO - codeparrot_training - Step 30730: {'lr': 0.00017383587218916196, 'samples': 15734272, 'steps': 30730, 'loss/train': 1.9182744026184082}
+02/25/2022 17:23:15 - INFO - codeparrot_training - Step 30731: {'lr': 0.00017382028772455996, 'samples': 15734784, 'steps': 30731, 'loss/train': 1.1504204273223877}
+02/25/2022 17:23:20 - INFO - codeparrot_training - Step 30732: {'lr': 0.00017380470358628763, 'samples': 15735296, 'steps': 30732, 'loss/train': 1.8763405084609985}
+02/25/2022 17:23:24 - INFO - codeparrot_training - Step 30733: {'lr': 0.0001737891197744118, 'samples': 15735808, 'steps': 30733, 'loss/train': 1.8437106609344482}
+02/25/2022 17:23:30 - INFO - codeparrot_training - Step 30734: {'lr': 0.00017377353628899927, 'samples': 15736320, 'steps': 30734, 'loss/train': 1.661285400390625}
+02/25/2022 17:23:34 - INFO - codeparrot_training - Step 30735: {'lr': 0.0001737579531301166, 'samples': 15736832, 'steps': 30735, 'loss/train': 1.0417697429656982}
+02/25/2022 17:23:39 - INFO - codeparrot_training - Step 30736: {'lr': 0.00017374237029783062, 'samples': 15737344, 'steps': 30736, 'loss/train': 1.5833008289337158}
+02/25/2022 17:23:43 - INFO - codeparrot_training - Step 30737: {'lr': 0.00017372678779220815, 'samples': 15737856, 'steps': 30737, 'loss/train': 1.7285505533218384}
+02/25/2022 17:23:48 - INFO - codeparrot_training - Step 30738: {'lr': 0.00017371120561331598, 'samples': 15738368, 'steps': 30738, 'loss/train': 1.145124912261963}
+02/25/2022 17:23:52 - INFO - codeparrot_training - Step 30739: {'lr': 0.00017369562376122073, 'samples': 15738880, 'steps': 30739, 'loss/train': 2.554356098175049}
+02/25/2022 17:23:57 - INFO - codeparrot_training - Step 30740: {'lr': 0.00017368004223598912, 'samples': 15739392, 'steps': 30740, 'loss/train': 2.5362050533294678}
+02/25/2022 17:24:01 - INFO - codeparrot_training - Step 30741: {'lr': 0.000173664461037688, 'samples': 15739904, 'steps': 30741, 'loss/train': 2.315133810043335}
+02/25/2022 17:24:06 - INFO - codeparrot_training - Step 30742: {'lr': 0.00017364888016638413, 'samples': 15740416, 'steps': 30742, 'loss/train': 0.9000276327133179}
+02/25/2022 17:24:10 - INFO - codeparrot_training - Step 30743: {'lr': 0.00017363329962214428, 'samples': 15740928, 'steps': 30743, 'loss/train': 1.5901893377304077}
+02/25/2022 17:24:15 - INFO - codeparrot_training - Step 30744: {'lr': 0.00017361771940503502, 'samples': 15741440, 'steps': 30744, 'loss/train': 2.456793785095215}
+02/25/2022 17:24:19 - INFO - codeparrot_training - Step 30745: {'lr': 0.00017360213951512322, 'samples': 15741952, 'steps': 30745, 'loss/train': 1.951997995376587}
+02/25/2022 17:24:26 - INFO - codeparrot_training - Step 30746: {'lr': 0.00017358655995247557, 'samples': 15742464, 'steps': 30746, 'loss/train': 1.5600237846374512}
+02/25/2022 17:24:29 - INFO - codeparrot_training - Step 30747: {'lr': 0.00017357098071715894, 'samples': 15742976, 'steps': 30747, 'loss/train': 1.66391921043396}
+02/25/2022 17:24:35 - INFO - codeparrot_training - Step 30748: {'lr': 0.00017355540180923987, 'samples': 15743488, 'steps': 30748, 'loss/train': 2.0510079860687256}
+02/25/2022 17:24:40 - INFO - codeparrot_training - Step 30749: {'lr': 0.0001735398232287852, 'samples': 15744000, 'steps': 30749, 'loss/train': 2.2184791564941406}
+02/25/2022 17:24:44 - INFO - codeparrot_training - Step 30750: {'lr': 0.0001735242449758616, 'samples': 15744512, 'steps': 30750, 'loss/train': 2.3929343223571777}
+02/25/2022 17:24:49 - INFO - codeparrot_training - Step 30751: {'lr': 0.00017350866705053596, 'samples': 15745024, 'steps': 30751, 'loss/train': 2.2873752117156982}
+02/25/2022 17:24:53 - INFO - codeparrot_training - Step 30752: {'lr': 0.00017349308945287484, 'samples': 15745536, 'steps': 30752, 'loss/train': 1.360449194908142}
+02/25/2022 17:24:58 - INFO - codeparrot_training - Step 30753: {'lr': 0.000173477512182945, 'samples': 15746048, 'steps': 30753, 'loss/train': 2.048659324645996}
+02/25/2022 17:25:02 - INFO - codeparrot_training - Step 30754: {'lr': 0.00017346193524081323, 'samples': 15746560, 'steps': 30754, 'loss/train': 1.7202531099319458}
+02/25/2022 17:25:08 - INFO - codeparrot_training - Step 30755: {'lr': 0.00017344635862654624, 'samples': 15747072, 'steps': 30755, 'loss/train': 0.8473336696624756}
+02/25/2022 17:25:12 - INFO - codeparrot_training - Step 30756: {'lr': 0.00017343078234021076, 'samples': 15747584, 'steps': 30756, 'loss/train': 1.283828854560852}
+02/25/2022 17:25:17 - INFO - codeparrot_training - Step 30757: {'lr': 0.00017341520638187346, 'samples': 15748096, 'steps': 30757, 'loss/train': 1.289860725402832}
+02/25/2022 17:25:21 - INFO - codeparrot_training - Step 30758: {'lr': 0.00017339963075160112, 'samples': 15748608, 'steps': 30758, 'loss/train': 2.3887338638305664}
+02/25/2022 17:25:26 - INFO - codeparrot_training - Step 30759: {'lr': 0.0001733840554494604, 'samples': 15749120, 'steps': 30759, 'loss/train': 0.19671623408794403}
+02/25/2022 17:25:30 - INFO - codeparrot_training - Step 30760: {'lr': 0.00017336848047551812, 'samples': 15749632, 'steps': 30760, 'loss/train': 1.5675932168960571}
+02/25/2022 17:25:35 - INFO - codeparrot_training - Step 30761: {'lr': 0.00017335290582984087, 'samples': 15750144, 'steps': 30761, 'loss/train': 1.2926170825958252}
+02/25/2022 17:25:39 - INFO - codeparrot_training - Step 30762: {'lr': 0.00017333733151249549, 'samples': 15750656, 'steps': 30762, 'loss/train': 1.3450815677642822}
+02/25/2022 17:25:44 - INFO - codeparrot_training - Step 30763: {'lr': 0.00017332175752354857, 'samples': 15751168, 'steps': 30763, 'loss/train': 0.8815481066703796}
+02/25/2022 17:25:48 - INFO - codeparrot_training - Step 30764: {'lr': 0.00017330618386306697, 'samples': 15751680, 'steps': 30764, 'loss/train': 2.398009777069092}
+02/25/2022 17:25:55 - INFO - codeparrot_training - Step 30765: {'lr': 0.00017329061053111724, 'samples': 15752192, 'steps': 30765, 'loss/train': 2.5809693336486816}
+02/25/2022 17:25:58 - INFO - codeparrot_training - Step 30766: {'lr': 0.00017327503752776624, 'samples': 15752704, 'steps': 30766, 'loss/train': 1.4935481548309326}
+02/25/2022 17:26:04 - INFO - codeparrot_training - Step 30767: {'lr': 0.00017325946485308058, 'samples': 15753216, 'steps': 30767, 'loss/train': 2.256352663040161}
+02/25/2022 17:26:07 - INFO - codeparrot_training - Step 30768: {'lr': 0.00017324389250712702, 'samples': 15753728, 'steps': 30768, 'loss/train': 1.2551031112670898}
+02/25/2022 17:26:13 - INFO - codeparrot_training - Step 30769: {'lr': 0.00017322832048997233, 'samples': 15754240, 'steps': 30769, 'loss/train': 1.4574698209762573}
+02/25/2022 17:26:16 - INFO - codeparrot_training - Step 30770: {'lr': 0.00017321274880168302, 'samples': 15754752, 'steps': 30770, 'loss/train': 0.5906045436859131}
+02/25/2022 17:26:22 - INFO - codeparrot_training - Step 30771: {'lr': 0.00017319717744232593, 'samples': 15755264, 'steps': 30771, 'loss/train': 1.4669440984725952}
+02/25/2022 17:26:25 - INFO - codeparrot_training - Step 30772: {'lr': 0.00017318160641196777, 'samples': 15755776, 'steps': 30772, 'loss/train': 1.480870008468628}
+02/25/2022 17:26:31 - INFO - codeparrot_training - Step 30773: {'lr': 0.0001731660357106753, 'samples': 15756288, 'steps': 30773, 'loss/train': 0.9543995261192322}
+02/25/2022 17:26:34 - INFO - codeparrot_training - Step 30774: {'lr': 0.000173150465338515, 'samples': 15756800, 'steps': 30774, 'loss/train': 1.2802789211273193}
+02/25/2022 17:26:39 - INFO - codeparrot_training - Step 30775: {'lr': 0.00017313489529555373, 'samples': 15757312, 'steps': 30775, 'loss/train': 2.2093520164489746}
+02/25/2022 17:26:43 - INFO - codeparrot_training - Step 30776: {'lr': 0.00017311932558185817, 'samples': 15757824, 'steps': 30776, 'loss/train': 1.8110259771347046}
+02/25/2022 17:26:48 - INFO - codeparrot_training - Step 30777: {'lr': 0.00017310375619749508, 'samples': 15758336, 'steps': 30777, 'loss/train': 1.2443310022354126}
+02/25/2022 17:26:52 - INFO - codeparrot_training - Step 30778: {'lr': 0.000173088187142531, 'samples': 15758848, 'steps': 30778, 'loss/train': 1.9007090330123901}
+02/25/2022 17:26:58 - INFO - codeparrot_training - Step 30779: {'lr': 0.0001730726184170327, 'samples': 15759360, 'steps': 30779, 'loss/train': 1.0733305215835571}
+02/25/2022 17:27:01 - INFO - codeparrot_training - Step 30780: {'lr': 0.00017305705002106686, 'samples': 15759872, 'steps': 30780, 'loss/train': 1.1906702518463135}
+02/25/2022 17:27:07 - INFO - codeparrot_training - Step 30781: {'lr': 0.00017304148195470021, 'samples': 15760384, 'steps': 30781, 'loss/train': 1.8607711791992188}
+02/25/2022 17:27:10 - INFO - codeparrot_training - Step 30782: {'lr': 0.00017302591421799953, 'samples': 15760896, 'steps': 30782, 'loss/train': 1.4264580011367798}
+02/25/2022 17:27:16 - INFO - codeparrot_training - Step 30783: {'lr': 0.00017301034681103128, 'samples': 15761408, 'steps': 30783, 'loss/train': 0.6306841969490051}
+02/25/2022 17:27:19 - INFO - codeparrot_training - Step 30784: {'lr': 0.00017299477973386224, 'samples': 15761920, 'steps': 30784, 'loss/train': 1.4932963848114014}
+02/25/2022 17:27:25 - INFO - codeparrot_training - Step 30785: {'lr': 0.0001729792129865591, 'samples': 15762432, 'steps': 30785, 'loss/train': 2.1149187088012695}
+02/25/2022 17:27:29 - INFO - codeparrot_training - Step 30786: {'lr': 0.00017296364656918867, 'samples': 15762944, 'steps': 30786, 'loss/train': 2.4017672538757324}
+02/25/2022 17:27:34 - INFO - codeparrot_training - Step 30787: {'lr': 0.00017294808048181742, 'samples': 15763456, 'steps': 30787, 'loss/train': 1.8204617500305176}
+02/25/2022 17:27:38 - INFO - codeparrot_training - Step 30788: {'lr': 0.00017293251472451216, 'samples': 15763968, 'steps': 30788, 'loss/train': 1.8793182373046875}
+02/25/2022 17:27:43 - INFO - codeparrot_training - Step 30789: {'lr': 0.00017291694929733948, 'samples': 15764480, 'steps': 30789, 'loss/train': 0.851350724697113}
+02/25/2022 17:27:47 - INFO - codeparrot_training - Step 30790: {'lr': 0.00017290138420036623, 'samples': 15764992, 'steps': 30790, 'loss/train': 2.3421406745910645}
+02/25/2022 17:27:53 - INFO - codeparrot_training - Step 30791: {'lr': 0.00017288581943365888, 'samples': 15765504, 'steps': 30791, 'loss/train': 0.8024457097053528}
+02/25/2022 17:27:56 - INFO - codeparrot_training - Step 30792: {'lr': 0.00017287025499728418, 'samples': 15766016, 'steps': 30792, 'loss/train': 0.7061970233917236}
+02/25/2022 17:28:02 - INFO - codeparrot_training - Step 30793: {'lr': 0.00017285469089130883, 'samples': 15766528, 'steps': 30793, 'loss/train': 0.47067826986312866}
+02/25/2022 17:28:05 - INFO - codeparrot_training - Step 30794: {'lr': 0.00017283912711579946, 'samples': 15767040, 'steps': 30794, 'loss/train': 2.7015745639801025}
+02/25/2022 17:28:11 - INFO - codeparrot_training - Step 30795: {'lr': 0.00017282356367082292, 'samples': 15767552, 'steps': 30795, 'loss/train': 1.3296995162963867}
+02/25/2022 17:28:14 - INFO - codeparrot_training - Step 30796: {'lr': 0.00017280800055644559, 'samples': 15768064, 'steps': 30796, 'loss/train': 1.2685362100601196}
+02/25/2022 17:28:20 - INFO - codeparrot_training - Step 30797: {'lr': 0.00017279243777273425, 'samples': 15768576, 'steps': 30797, 'loss/train': 1.469246745109558}
+02/25/2022 17:28:23 - INFO - codeparrot_training - Step 30798: {'lr': 0.00017277687531975567, 'samples': 15769088, 'steps': 30798, 'loss/train': 1.4838075637817383}
+02/25/2022 17:28:29 - INFO - codeparrot_training - Step 30799: {'lr': 0.00017276131319757644, 'samples': 15769600, 'steps': 30799, 'loss/train': 1.4600505828857422}
+02/25/2022 17:28:32 - INFO - codeparrot_training - Step 30800: {'lr': 0.00017274575140626317, 'samples': 15770112, 'steps': 30800, 'loss/train': 0.3699166178703308}
+02/25/2022 17:28:39 - INFO - codeparrot_training - Step 30801: {'lr': 0.00017273018994588257, 'samples': 15770624, 'steps': 30801, 'loss/train': 0.6264426112174988}
+02/25/2022 17:28:42 - INFO - codeparrot_training - Step 30802: {'lr': 0.0001727146288165013, 'samples': 15771136, 'steps': 30802, 'loss/train': 1.7573096752166748}
+02/25/2022 17:28:48 - INFO - codeparrot_training - Step 30803: {'lr': 0.00017269906801818603, 'samples': 15771648, 'steps': 30803, 'loss/train': 2.4332785606384277}
+02/25/2022 17:28:51 - INFO - codeparrot_training - Step 30804: {'lr': 0.00017268350755100345, 'samples': 15772160, 'steps': 30804, 'loss/train': 1.8508580923080444}
+02/25/2022 17:28:57 - INFO - codeparrot_training - Step 30805: {'lr': 0.00017266794741502013, 'samples': 15772672, 'steps': 30805, 'loss/train': 1.782822608947754}
+02/25/2022 17:29:00 - INFO - codeparrot_training - Step 30806: {'lr': 0.00017265238761030277, 'samples': 15773184, 'steps': 30806, 'loss/train': 1.3476858139038086}
+02/25/2022 17:29:06 - INFO - codeparrot_training - Step 30807: {'lr': 0.000172636828136918, 'samples': 15773696, 'steps': 30807, 'loss/train': 2.2343435287475586}
+02/25/2022 17:29:09 - INFO - codeparrot_training - Step 30808: {'lr': 0.00017262126899493256, 'samples': 15774208, 'steps': 30808, 'loss/train': 0.7837716341018677}
+02/25/2022 17:29:15 - INFO - codeparrot_training - Step 30809: {'lr': 0.000172605710184413, 'samples': 15774720, 'steps': 30809, 'loss/train': 2.136336326599121}
+02/25/2022 17:29:18 - INFO - codeparrot_training - Step 30810: {'lr': 0.00017259015170542597, 'samples': 15775232, 'steps': 30810, 'loss/train': 1.6527882814407349}
+02/25/2022 17:29:25 - INFO - codeparrot_training - Step 30811: {'lr': 0.00017257459355803816, 'samples': 15775744, 'steps': 30811, 'loss/train': 2.1016275882720947}
+02/25/2022 17:29:28 - INFO - codeparrot_training - Step 30812: {'lr': 0.00017255903574231625, 'samples': 15776256, 'steps': 30812, 'loss/train': 2.0298404693603516}
+02/25/2022 17:29:34 - INFO - codeparrot_training - Step 30813: {'lr': 0.00017254347825832684, 'samples': 15776768, 'steps': 30813, 'loss/train': 1.8232800960540771}
+02/25/2022 17:29:37 - INFO - codeparrot_training - Step 30814: {'lr': 0.00017252792110613652, 'samples': 15777280, 'steps': 30814, 'loss/train': 8.336360931396484}
+02/25/2022 17:29:43 - INFO - codeparrot_training - Step 30815: {'lr': 0.00017251236428581196, 'samples': 15777792, 'steps': 30815, 'loss/train': 1.1440787315368652}
+02/25/2022 17:29:46 - INFO - codeparrot_training - Step 30816: {'lr': 0.00017249680779741987, 'samples': 15778304, 'steps': 30816, 'loss/train': 1.5922085046768188}
+02/25/2022 17:29:52 - INFO - codeparrot_training - Step 30817: {'lr': 0.0001724812516410269, 'samples': 15778816, 'steps': 30817, 'loss/train': 1.9258041381835938}
+02/25/2022 17:29:55 - INFO - codeparrot_training - Step 30818: {'lr': 0.00017246569581669957, 'samples': 15779328, 'steps': 30818, 'loss/train': 1.764691710472107}
+02/25/2022 17:30:01 - INFO - codeparrot_training - Step 30819: {'lr': 0.00017245014032450456, 'samples': 15779840, 'steps': 30819, 'loss/train': 0.8716151118278503}
+02/25/2022 17:30:05 - INFO - codeparrot_training - Step 30820: {'lr': 0.00017243458516450855, 'samples': 15780352, 'steps': 30820, 'loss/train': 2.1852333545684814}
+02/25/2022 17:30:10 - INFO - codeparrot_training - Step 30821: {'lr': 0.00017241903033677822, 'samples': 15780864, 'steps': 30821, 'loss/train': 1.449008584022522}
+02/25/2022 17:30:14 - INFO - codeparrot_training - Step 30822: {'lr': 0.00017240347584138006, 'samples': 15781376, 'steps': 30822, 'loss/train': 1.9347764253616333}
+02/25/2022 17:30:19 - INFO - codeparrot_training - Step 30823: {'lr': 0.00017238792167838073, 'samples': 15781888, 'steps': 30823, 'loss/train': 2.350262403488159}
+02/25/2022 17:30:25 - INFO - codeparrot_training - Step 30824: {'lr': 0.00017237236784784692, 'samples': 15782400, 'steps': 30824, 'loss/train': 1.3809736967086792}
+02/25/2022 17:30:28 - INFO - codeparrot_training - Step 30825: {'lr': 0.00017235681434984532, 'samples': 15782912, 'steps': 30825, 'loss/train': 1.7339568138122559}
+02/25/2022 17:30:34 - INFO - codeparrot_training - Step 30826: {'lr': 0.00017234126118444238, 'samples': 15783424, 'steps': 30826, 'loss/train': 1.257758617401123}
+02/25/2022 17:30:38 - INFO - codeparrot_training - Step 30827: {'lr': 0.00017232570835170482, 'samples': 15783936, 'steps': 30827, 'loss/train': 1.5525656938552856}
+02/25/2022 17:30:43 - INFO - codeparrot_training - Step 30828: {'lr': 0.00017231015585169928, 'samples': 15784448, 'steps': 30828, 'loss/train': 1.4200551509857178}
+02/25/2022 17:30:47 - INFO - codeparrot_training - Step 30829: {'lr': 0.00017229460368449234, 'samples': 15784960, 'steps': 30829, 'loss/train': 1.3803961277008057}
+02/25/2022 17:30:52 - INFO - codeparrot_training - Step 30830: {'lr': 0.00017227905185015075, 'samples': 15785472, 'steps': 30830, 'loss/train': 1.514911413192749}
+02/25/2022 17:30:56 - INFO - codeparrot_training - Step 30831: {'lr': 0.00017226350034874096, 'samples': 15785984, 'steps': 30831, 'loss/train': 3.714461088180542}
+02/25/2022 17:31:01 - INFO - codeparrot_training - Step 30832: {'lr': 0.0001722479491803296, 'samples': 15786496, 'steps': 30832, 'loss/train': 2.6190905570983887}
+02/25/2022 17:31:05 - INFO - codeparrot_training - Step 30833: {'lr': 0.00017223239834498333, 'samples': 15787008, 'steps': 30833, 'loss/train': 1.3432352542877197}
+02/25/2022 17:31:10 - INFO - codeparrot_training - Step 30834: {'lr': 0.0001722168478427689, 'samples': 15787520, 'steps': 30834, 'loss/train': 1.4192978143692017}
+02/25/2022 17:31:14 - INFO - codeparrot_training - Step 30835: {'lr': 0.0001722012976737527, 'samples': 15788032, 'steps': 30835, 'loss/train': 1.192130446434021}
+02/25/2022 17:31:20 - INFO - codeparrot_training - Step 30836: {'lr': 0.0001721857478380014, 'samples': 15788544, 'steps': 30836, 'loss/train': 2.4973342418670654}
+02/25/2022 17:31:23 - INFO - codeparrot_training - Step 30837: {'lr': 0.00017217019833558168, 'samples': 15789056, 'steps': 30837, 'loss/train': 2.3958022594451904}
+02/25/2022 17:31:29 - INFO - codeparrot_training - Step 30838: {'lr': 0.0001721546491665602, 'samples': 15789568, 'steps': 30838, 'loss/train': 0.4230477809906006}
+02/25/2022 17:31:32 - INFO - codeparrot_training - Step 30839: {'lr': 0.0001721391003310034, 'samples': 15790080, 'steps': 30839, 'loss/train': 1.9959704875946045}
+02/25/2022 17:31:38 - INFO - codeparrot_training - Step 30840: {'lr': 0.00017212355182897799, 'samples': 15790592, 'steps': 30840, 'loss/train': 2.0177571773529053}
+02/25/2022 17:31:41 - INFO - codeparrot_training - Step 30841: {'lr': 0.0001721080036605505, 'samples': 15791104, 'steps': 30841, 'loss/train': 1.3669719696044922}
+02/25/2022 17:31:47 - INFO - codeparrot_training - Step 30842: {'lr': 0.0001720924558257877, 'samples': 15791616, 'steps': 30842, 'loss/train': 1.751502275466919}
+02/25/2022 17:31:50 - INFO - codeparrot_training - Step 30843: {'lr': 0.00017207690832475602, 'samples': 15792128, 'steps': 30843, 'loss/train': 2.3211817741394043}
+02/25/2022 17:31:56 - INFO - codeparrot_training - Step 30844: {'lr': 0.0001720613611575221, 'samples': 15792640, 'steps': 30844, 'loss/train': 2.121504783630371}
+02/25/2022 17:31:59 - INFO - codeparrot_training - Step 30845: {'lr': 0.00017204581432415255, 'samples': 15793152, 'steps': 30845, 'loss/train': 1.253933072090149}
+02/25/2022 17:32:06 - INFO - codeparrot_training - Step 30846: {'lr': 0.00017203026782471403, 'samples': 15793664, 'steps': 30846, 'loss/train': 1.7700453996658325}
+02/25/2022 17:32:09 - INFO - codeparrot_training - Step 30847: {'lr': 0.00017201472165927312, 'samples': 15794176, 'steps': 30847, 'loss/train': 0.7101238369941711}
+02/25/2022 17:32:15 - INFO - codeparrot_training - Step 30848: {'lr': 0.00017199917582789631, 'samples': 15794688, 'steps': 30848, 'loss/train': 1.5261906385421753}
+02/25/2022 17:32:18 - INFO - codeparrot_training - Step 30849: {'lr': 0.0001719836303306503, 'samples': 15795200, 'steps': 30849, 'loss/train': 1.2616701126098633}
+02/25/2022 17:32:24 - INFO - codeparrot_training - Step 30850: {'lr': 0.00017196808516760159, 'samples': 15795712, 'steps': 30850, 'loss/train': 1.6685019731521606}
+02/25/2022 17:32:27 - INFO - codeparrot_training - Step 30851: {'lr': 0.00017195254033881686, 'samples': 15796224, 'steps': 30851, 'loss/train': 3.0583178997039795}
+02/25/2022 17:32:33 - INFO - codeparrot_training - Step 30852: {'lr': 0.00017193699584436268, 'samples': 15796736, 'steps': 30852, 'loss/train': 1.7429862022399902}
+02/25/2022 17:32:36 - INFO - codeparrot_training - Step 30853: {'lr': 0.0001719214516843056, 'samples': 15797248, 'steps': 30853, 'loss/train': 1.7161043882369995}
+02/25/2022 17:32:42 - INFO - codeparrot_training - Step 30854: {'lr': 0.00017190590785871225, 'samples': 15797760, 'steps': 30854, 'loss/train': 0.8770737648010254}
+02/25/2022 17:32:45 - INFO - codeparrot_training - Step 30855: {'lr': 0.00017189036436764913, 'samples': 15798272, 'steps': 30855, 'loss/train': 0.0209540706127882}
+02/25/2022 17:32:51 - INFO - codeparrot_training - Step 30856: {'lr': 0.00017187482121118304, 'samples': 15798784, 'steps': 30856, 'loss/train': 2.1271042823791504}
+02/25/2022 17:32:55 - INFO - codeparrot_training - Step 30857: {'lr': 0.00017185927838938027, 'samples': 15799296, 'steps': 30857, 'loss/train': 0.4109536409378052}
+02/25/2022 17:33:00 - INFO - codeparrot_training - Step 30858: {'lr': 0.00017184373590230752, 'samples': 15799808, 'steps': 30858, 'loss/train': 1.4277530908584595}
+02/25/2022 17:33:04 - INFO - codeparrot_training - Step 30859: {'lr': 0.0001718281937500314, 'samples': 15800320, 'steps': 30859, 'loss/train': 1.2299360036849976}
+02/25/2022 17:33:09 - INFO - codeparrot_training - Step 30860: {'lr': 0.00017181265193261865, 'samples': 15800832, 'steps': 30860, 'loss/train': 1.243561863899231}
+02/25/2022 17:33:13 - INFO - codeparrot_training - Step 30861: {'lr': 0.00017179711045013546, 'samples': 15801344, 'steps': 30861, 'loss/train': 2.1546435356140137}
+02/25/2022 17:33:19 - INFO - codeparrot_training - Step 30862: {'lr': 0.00017178156930264867, 'samples': 15801856, 'steps': 30862, 'loss/train': 1.813400387763977}
+02/25/2022 17:33:23 - INFO - codeparrot_training - Step 30863: {'lr': 0.00017176602849022484, 'samples': 15802368, 'steps': 30863, 'loss/train': 1.7000378370285034}
+02/25/2022 17:33:28 - INFO - codeparrot_training - Step 30864: {'lr': 0.00017175048801293041, 'samples': 15802880, 'steps': 30864, 'loss/train': 2.5158183574676514}
+02/25/2022 17:33:32 - INFO - codeparrot_training - Step 30865: {'lr': 0.00017173494787083222, 'samples': 15803392, 'steps': 30865, 'loss/train': 0.9467108845710754}
+02/25/2022 17:33:37 - INFO - codeparrot_training - Step 30866: {'lr': 0.00017171940806399652, 'samples': 15803904, 'steps': 30866, 'loss/train': 1.1335984468460083}
+02/25/2022 17:33:41 - INFO - codeparrot_training - Step 30867: {'lr': 0.00017170386859249005, 'samples': 15804416, 'steps': 30867, 'loss/train': 1.988924503326416}
+02/25/2022 17:33:47 - INFO - codeparrot_training - Step 30868: {'lr': 0.0001716883294563793, 'samples': 15804928, 'steps': 30868, 'loss/train': 2.3729653358459473}
+02/25/2022 17:33:50 - INFO - codeparrot_training - Step 30869: {'lr': 0.00017167279065573104, 'samples': 15805440, 'steps': 30869, 'loss/train': 2.2912564277648926}
+02/25/2022 17:33:56 - INFO - codeparrot_training - Step 30870: {'lr': 0.00017165725219061152, 'samples': 15805952, 'steps': 30870, 'loss/train': 1.8677479028701782}
+02/25/2022 17:33:59 - INFO - codeparrot_training - Step 30871: {'lr': 0.00017164171406108749, 'samples': 15806464, 'steps': 30871, 'loss/train': 2.0796620845794678}
+02/25/2022 17:34:06 - INFO - codeparrot_training - Step 30872: {'lr': 0.00017162617626722545, 'samples': 15806976, 'steps': 30872, 'loss/train': 2.152864694595337}
+02/25/2022 17:34:09 - INFO - codeparrot_training - Step 30873: {'lr': 0.00017161063880909212, 'samples': 15807488, 'steps': 30873, 'loss/train': 2.406723976135254}
+02/25/2022 17:34:15 - INFO - codeparrot_training - Step 30874: {'lr': 0.0001715951016867538, 'samples': 15808000, 'steps': 30874, 'loss/train': 2.6085996627807617}
+02/25/2022 17:34:18 - INFO - codeparrot_training - Step 30875: {'lr': 0.00017157956490027715, 'samples': 15808512, 'steps': 30875, 'loss/train': 1.7170699834823608}
+02/25/2022 17:34:24 - INFO - codeparrot_training - Step 30876: {'lr': 0.00017156402844972874, 'samples': 15809024, 'steps': 30876, 'loss/train': 0.03191710636019707}
+02/25/2022 17:34:27 - INFO - codeparrot_training - Step 30877: {'lr': 0.00017154849233517517, 'samples': 15809536, 'steps': 30877, 'loss/train': 2.4711148738861084}
+02/25/2022 17:34:33 - INFO - codeparrot_training - Step 30878: {'lr': 0.000171532956556683, 'samples': 15810048, 'steps': 30878, 'loss/train': 1.783501386642456}
+02/25/2022 17:34:36 - INFO - codeparrot_training - Step 30879: {'lr': 0.00017151742111431867, 'samples': 15810560, 'steps': 30879, 'loss/train': 2.419020414352417}
+02/25/2022 17:34:42 - INFO - codeparrot_training - Step 30880: {'lr': 0.00017150188600814875, 'samples': 15811072, 'steps': 30880, 'loss/train': 1.9733247756958008}
+02/25/2022 17:34:45 - INFO - codeparrot_training - Step 30881: {'lr': 0.00017148635123823985, 'samples': 15811584, 'steps': 30881, 'loss/train': 0.3049890697002411}
+02/25/2022 17:34:52 - INFO - codeparrot_training - Step 30882: {'lr': 0.0001714708168046586, 'samples': 15812096, 'steps': 30882, 'loss/train': 1.6930447816848755}
+02/25/2022 17:34:55 - INFO - codeparrot_training - Step 30883: {'lr': 0.0001714552827074713, 'samples': 15812608, 'steps': 30883, 'loss/train': 2.405334711074829}
+02/25/2022 17:35:01 - INFO - codeparrot_training - Step 30884: {'lr': 0.00017143974894674464, 'samples': 15813120, 'steps': 30884, 'loss/train': 2.5295002460479736}
+02/25/2022 17:35:04 - INFO - codeparrot_training - Step 30885: {'lr': 0.00017142421552254517, 'samples': 15813632, 'steps': 30885, 'loss/train': 1.4440367221832275}
+02/25/2022 17:35:10 - INFO - codeparrot_training - Step 30886: {'lr': 0.0001714086824349395, 'samples': 15814144, 'steps': 30886, 'loss/train': 1.6534053087234497}
+02/25/2022 17:35:13 - INFO - codeparrot_training - Step 30887: {'lr': 0.00017139314968399395, 'samples': 15814656, 'steps': 30887, 'loss/train': 1.4282454252243042}
+02/25/2022 17:35:19 - INFO - codeparrot_training - Step 30888: {'lr': 0.0001713776172697752, 'samples': 15815168, 'steps': 30888, 'loss/train': 0.889448344707489}
+02/25/2022 17:35:22 - INFO - codeparrot_training - Step 30889: {'lr': 0.00017136208519234986, 'samples': 15815680, 'steps': 30889, 'loss/train': 1.6427979469299316}
+02/25/2022 17:35:28 - INFO - codeparrot_training - Step 30890: {'lr': 0.0001713465534517843, 'samples': 15816192, 'steps': 30890, 'loss/train': 2.5924878120422363}
+02/25/2022 17:35:31 - INFO - codeparrot_training - Step 30891: {'lr': 0.00017133102204814515, 'samples': 15816704, 'steps': 30891, 'loss/train': 1.9412628412246704}
+02/25/2022 17:35:37 - INFO - codeparrot_training - Step 30892: {'lr': 0.0001713154909814989, 'samples': 15817216, 'steps': 30892, 'loss/train': 1.671139121055603}
+02/25/2022 17:35:41 - INFO - codeparrot_training - Step 30893: {'lr': 0.00017129996025191217, 'samples': 15817728, 'steps': 30893, 'loss/train': 1.4168286323547363}
+02/25/2022 17:35:46 - INFO - codeparrot_training - Step 30894: {'lr': 0.00017128442985945137, 'samples': 15818240, 'steps': 30894, 'loss/train': 2.154322862625122}
+02/25/2022 17:35:50 - INFO - codeparrot_training - Step 30895: {'lr': 0.0001712688998041831, 'samples': 15818752, 'steps': 30895, 'loss/train': 1.8076807260513306}
+02/25/2022 17:35:55 - INFO - codeparrot_training - Step 30896: {'lr': 0.00017125337008617387, 'samples': 15819264, 'steps': 30896, 'loss/train': 1.6764289140701294}
+02/25/2022 17:35:59 - INFO - codeparrot_training - Step 30897: {'lr': 0.00017123784070549014, 'samples': 15819776, 'steps': 30897, 'loss/train': 2.3929100036621094}
+02/25/2022 17:36:04 - INFO - codeparrot_training - Step 30898: {'lr': 0.00017122231166219847, 'samples': 15820288, 'steps': 30898, 'loss/train': 2.544685125350952}
+02/25/2022 17:36:08 - INFO - codeparrot_training - Step 30899: {'lr': 0.0001712067829563655, 'samples': 15820800, 'steps': 30899, 'loss/train': 1.954620361328125}
+02/25/2022 17:36:13 - INFO - codeparrot_training - Step 30900: {'lr': 0.00017119125458805765, 'samples': 15821312, 'steps': 30900, 'loss/train': 2.143437385559082}
+02/25/2022 17:36:17 - INFO - codeparrot_training - Step 30901: {'lr': 0.00017117572655734136, 'samples': 15821824, 'steps': 30901, 'loss/train': 1.8220163583755493}
+02/25/2022 17:36:22 - INFO - codeparrot_training - Step 30902: {'lr': 0.00017116019886428326, 'samples': 15822336, 'steps': 30902, 'loss/train': 1.794333815574646}
+02/25/2022 17:36:28 - INFO - codeparrot_training - Step 30903: {'lr': 0.0001711446715089498, 'samples': 15822848, 'steps': 30903, 'loss/train': 1.8099684715270996}
+02/25/2022 17:36:31 - INFO - codeparrot_training - Step 30904: {'lr': 0.00017112914449140764, 'samples': 15823360, 'steps': 30904, 'loss/train': 1.5666985511779785}
+02/25/2022 17:36:37 - INFO - codeparrot_training - Step 30905: {'lr': 0.0001711136178117231, 'samples': 15823872, 'steps': 30905, 'loss/train': 1.0048363208770752}
+02/25/2022 17:36:40 - INFO - codeparrot_training - Step 30906: {'lr': 0.00017109809146996274, 'samples': 15824384, 'steps': 30906, 'loss/train': 2.2085349559783936}
+02/25/2022 17:36:46 - INFO - codeparrot_training - Step 30907: {'lr': 0.00017108256546619313, 'samples': 15824896, 'steps': 30907, 'loss/train': 1.774291753768921}
+02/25/2022 17:36:50 - INFO - codeparrot_training - Step 30908: {'lr': 0.00017106703980048084, 'samples': 15825408, 'steps': 30908, 'loss/train': 1.3725571632385254}
+02/25/2022 17:36:55 - INFO - codeparrot_training - Step 30909: {'lr': 0.00017105151447289218, 'samples': 15825920, 'steps': 30909, 'loss/train': 1.9665273427963257}
+02/25/2022 17:36:59 - INFO - codeparrot_training - Step 30910: {'lr': 0.0001710359894834938, 'samples': 15826432, 'steps': 30910, 'loss/train': 2.044612169265747}
+02/25/2022 17:37:04 - INFO - codeparrot_training - Step 30911: {'lr': 0.0001710204648323521, 'samples': 15826944, 'steps': 30911, 'loss/train': 2.160705089569092}
+02/25/2022 17:37:08 - INFO - codeparrot_training - Step 30912: {'lr': 0.00017100494051953372, 'samples': 15827456, 'steps': 30912, 'loss/train': 2.6525821685791016}
+02/25/2022 17:37:13 - INFO - codeparrot_training - Step 30913: {'lr': 0.00017098941654510513, 'samples': 15827968, 'steps': 30913, 'loss/train': 1.2803303003311157}
+02/25/2022 17:37:17 - INFO - codeparrot_training - Step 30914: {'lr': 0.00017097389290913275, 'samples': 15828480, 'steps': 30914, 'loss/train': 1.8472870588302612}
+02/25/2022 17:37:22 - INFO - codeparrot_training - Step 30915: {'lr': 0.00017095836961168308, 'samples': 15828992, 'steps': 30915, 'loss/train': 8.416533470153809}
+02/25/2022 17:37:26 - INFO - codeparrot_training - Step 30916: {'lr': 0.00017094284665282266, 'samples': 15829504, 'steps': 30916, 'loss/train': 3.010324239730835}
+02/25/2022 17:37:32 - INFO - codeparrot_training - Step 30917: {'lr': 0.0001709273240326181, 'samples': 15830016, 'steps': 30917, 'loss/train': 1.221379280090332}
+02/25/2022 17:37:35 - INFO - codeparrot_training - Step 30918: {'lr': 0.00017091180175113566, 'samples': 15830528, 'steps': 30918, 'loss/train': 2.2346441745758057}
+02/25/2022 17:37:41 - INFO - codeparrot_training - Step 30919: {'lr': 0.00017089627980844193, 'samples': 15831040, 'steps': 30919, 'loss/train': 0.8587213754653931}
+02/25/2022 17:37:44 - INFO - codeparrot_training - Step 30920: {'lr': 0.00017088075820460348, 'samples': 15831552, 'steps': 30920, 'loss/train': 1.1487531661987305}
+02/25/2022 17:37:50 - INFO - codeparrot_training - Step 30921: {'lr': 0.00017086523693968678, 'samples': 15832064, 'steps': 30921, 'loss/train': 1.5996801853179932}
+02/25/2022 17:37:53 - INFO - codeparrot_training - Step 30922: {'lr': 0.0001708497160137582, 'samples': 15832576, 'steps': 30922, 'loss/train': 1.2667784690856934}
+02/25/2022 17:37:59 - INFO - codeparrot_training - Step 30923: {'lr': 0.00017083419542688428, 'samples': 15833088, 'steps': 30923, 'loss/train': 0.20584948360919952}
+02/25/2022 17:38:02 - INFO - codeparrot_training - Step 30924: {'lr': 0.00017081867517913156, 'samples': 15833600, 'steps': 30924, 'loss/train': 1.4849941730499268}
+02/25/2022 17:38:08 - INFO - codeparrot_training - Step 30925: {'lr': 0.00017080315527056645, 'samples': 15834112, 'steps': 30925, 'loss/train': 1.971928596496582}
+02/25/2022 17:38:11 - INFO - codeparrot_training - Step 30926: {'lr': 0.0001707876357012556, 'samples': 15834624, 'steps': 30926, 'loss/train': 0.6870985627174377}
+02/25/2022 17:38:18 - INFO - codeparrot_training - Step 30927: {'lr': 0.00017077211647126527, 'samples': 15835136, 'steps': 30927, 'loss/train': 2.1315577030181885}
+02/25/2022 17:38:21 - INFO - codeparrot_training - Step 30928: {'lr': 0.00017075659758066205, 'samples': 15835648, 'steps': 30928, 'loss/train': 0.5204908847808838}
+02/25/2022 17:38:27 - INFO - codeparrot_training - Step 30929: {'lr': 0.00017074107902951245, 'samples': 15836160, 'steps': 30929, 'loss/train': 1.8238492012023926}
+02/25/2022 17:38:30 - INFO - codeparrot_training - Step 30930: {'lr': 0.00017072556081788287, 'samples': 15836672, 'steps': 30930, 'loss/train': 2.8112919330596924}
+02/25/2022 17:38:36 - INFO - codeparrot_training - Step 30931: {'lr': 0.0001707100429458398, 'samples': 15837184, 'steps': 30931, 'loss/train': 1.8597627878189087}
+02/25/2022 17:38:39 - INFO - codeparrot_training - Step 30932: {'lr': 0.00017069452541344972, 'samples': 15837696, 'steps': 30932, 'loss/train': 3.509359121322632}
+02/25/2022 17:38:45 - INFO - codeparrot_training - Step 30933: {'lr': 0.00017067900822077914, 'samples': 15838208, 'steps': 30933, 'loss/train': 2.451078176498413}
+02/25/2022 17:38:48 - INFO - codeparrot_training - Step 30934: {'lr': 0.00017066349136789452, 'samples': 15838720, 'steps': 30934, 'loss/train': 0.6647193431854248}
+02/25/2022 17:38:54 - INFO - codeparrot_training - Step 30935: {'lr': 0.00017064797485486226, 'samples': 15839232, 'steps': 30935, 'loss/train': 0.14275012910366058}
+02/25/2022 17:38:57 - INFO - codeparrot_training - Step 30936: {'lr': 0.00017063245868174893, 'samples': 15839744, 'steps': 30936, 'loss/train': 1.5685423612594604}
+02/25/2022 17:39:03 - INFO - codeparrot_training - Step 30937: {'lr': 0.00017061694284862088, 'samples': 15840256, 'steps': 30937, 'loss/train': 2.4612739086151123}
+02/25/2022 17:39:06 - INFO - codeparrot_training - Step 30938: {'lr': 0.00017060142735554463, 'samples': 15840768, 'steps': 30938, 'loss/train': 2.703680992126465}
+02/25/2022 17:39:13 - INFO - codeparrot_training - Step 30939: {'lr': 0.00017058591220258672, 'samples': 15841280, 'steps': 30939, 'loss/train': 1.4875937700271606}
+02/25/2022 17:39:16 - INFO - codeparrot_training - Step 30940: {'lr': 0.00017057039738981356, 'samples': 15841792, 'steps': 30940, 'loss/train': 1.9587736129760742}
+02/25/2022 17:39:22 - INFO - codeparrot_training - Step 30941: {'lr': 0.00017055488291729153, 'samples': 15842304, 'steps': 30941, 'loss/train': 2.813506841659546}
+02/25/2022 17:39:25 - INFO - codeparrot_training - Step 30942: {'lr': 0.00017053936878508717, 'samples': 15842816, 'steps': 30942, 'loss/train': 0.8924906253814697}
+02/25/2022 17:39:31 - INFO - codeparrot_training - Step 30943: {'lr': 0.000170523854993267, 'samples': 15843328, 'steps': 30943, 'loss/train': 2.669710636138916}
+02/25/2022 17:39:34 - INFO - codeparrot_training - Step 30944: {'lr': 0.00017050834154189732, 'samples': 15843840, 'steps': 30944, 'loss/train': 2.0535473823547363}
+02/25/2022 17:39:40 - INFO - codeparrot_training - Step 30945: {'lr': 0.00017049282843104464, 'samples': 15844352, 'steps': 30945, 'loss/train': 5.883098602294922}
+02/25/2022 17:39:43 - INFO - codeparrot_training - Step 30946: {'lr': 0.00017047731566077545, 'samples': 15844864, 'steps': 30946, 'loss/train': 1.8980615139007568}
+02/25/2022 17:39:49 - INFO - codeparrot_training - Step 30947: {'lr': 0.0001704618032311563, 'samples': 15845376, 'steps': 30947, 'loss/train': 1.850115418434143}
+02/25/2022 17:39:53 - INFO - codeparrot_training - Step 30948: {'lr': 0.00017044629114225341, 'samples': 15845888, 'steps': 30948, 'loss/train': 0.18369010090827942}
+02/25/2022 17:39:58 - INFO - codeparrot_training - Step 30949: {'lr': 0.00017043077939413339, 'samples': 15846400, 'steps': 30949, 'loss/train': 1.6518300771713257}
+02/25/2022 17:40:02 - INFO - codeparrot_training - Step 30950: {'lr': 0.0001704152679868626, 'samples': 15846912, 'steps': 30950, 'loss/train': 1.1802380084991455}
+02/25/2022 17:40:07 - INFO - codeparrot_training - Step 30951: {'lr': 0.00017039975692050753, 'samples': 15847424, 'steps': 30951, 'loss/train': 2.2024519443511963}
+02/25/2022 17:40:11 - INFO - codeparrot_training - Step 30952: {'lr': 0.00017038424619513471, 'samples': 15847936, 'steps': 30952, 'loss/train': 1.9004453420639038}
+02/25/2022 17:40:16 - INFO - codeparrot_training - Step 30953: {'lr': 0.00017036873581081045, 'samples': 15848448, 'steps': 30953, 'loss/train': 2.141775131225586}
+02/25/2022 17:40:20 - INFO - codeparrot_training - Step 30954: {'lr': 0.00017035322576760122, 'samples': 15848960, 'steps': 30954, 'loss/train': 1.3196666240692139}
+02/25/2022 17:40:26 - INFO - codeparrot_training - Step 30955: {'lr': 0.00017033771606557346, 'samples': 15849472, 'steps': 30955, 'loss/train': 0.9540612101554871}
+02/25/2022 17:40:29 - INFO - codeparrot_training - Step 30956: {'lr': 0.00017032220670479376, 'samples': 15849984, 'steps': 30956, 'loss/train': 1.969914436340332}
+02/25/2022 17:40:35 - INFO - codeparrot_training - Step 30957: {'lr': 0.0001703066976853283, 'samples': 15850496, 'steps': 30957, 'loss/train': 0.7234318852424622}
+02/25/2022 17:40:38 - INFO - codeparrot_training - Step 30958: {'lr': 0.00017029118900724363, 'samples': 15851008, 'steps': 30958, 'loss/train': 1.6052916049957275}
+02/25/2022 17:40:44 - INFO - codeparrot_training - Step 30959: {'lr': 0.00017027568067060623, 'samples': 15851520, 'steps': 30959, 'loss/train': 2.1360130310058594}
+02/25/2022 17:40:47 - INFO - codeparrot_training - Step 30960: {'lr': 0.0001702601726754825, 'samples': 15852032, 'steps': 30960, 'loss/train': 3.0374183654785156}
+02/25/2022 17:40:53 - INFO - codeparrot_training - Step 30961: {'lr': 0.00017024466502193892, 'samples': 15852544, 'steps': 30961, 'loss/train': 2.400423049926758}
+02/25/2022 17:40:56 - INFO - codeparrot_training - Step 30962: {'lr': 0.00017022915771004182, 'samples': 15853056, 'steps': 30962, 'loss/train': 1.2397576570510864}
+02/25/2022 17:41:02 - INFO - codeparrot_training - Step 30963: {'lr': 0.00017021365073985767, 'samples': 15853568, 'steps': 30963, 'loss/train': 2.007594347000122}
+02/25/2022 17:41:05 - INFO - codeparrot_training - Step 30964: {'lr': 0.0001701981441114529, 'samples': 15854080, 'steps': 30964, 'loss/train': 1.2358102798461914}
+02/25/2022 17:41:12 - INFO - codeparrot_training - Step 30965: {'lr': 0.000170182637824894, 'samples': 15854592, 'steps': 30965, 'loss/train': 1.9479891061782837}
+02/25/2022 17:41:15 - INFO - codeparrot_training - Step 30966: {'lr': 0.0001701671318802473, 'samples': 15855104, 'steps': 30966, 'loss/train': 2.5740203857421875}
+02/25/2022 17:41:21 - INFO - codeparrot_training - Step 30967: {'lr': 0.00017015162627757925, 'samples': 15855616, 'steps': 30967, 'loss/train': 1.6464446783065796}
+02/25/2022 17:41:24 - INFO - codeparrot_training - Step 30968: {'lr': 0.00017013612101695623, 'samples': 15856128, 'steps': 30968, 'loss/train': 1.8661909103393555}
+02/25/2022 17:41:30 - INFO - codeparrot_training - Step 30969: {'lr': 0.00017012061609844486, 'samples': 15856640, 'steps': 30969, 'loss/train': 1.9088335037231445}
+02/25/2022 17:41:33 - INFO - codeparrot_training - Step 30970: {'lr': 0.00017010511152211127, 'samples': 15857152, 'steps': 30970, 'loss/train': 1.4334156513214111}
+02/25/2022 17:41:39 - INFO - codeparrot_training - Step 30971: {'lr': 0.00017008960728802204, 'samples': 15857664, 'steps': 30971, 'loss/train': 2.141254425048828}
+02/25/2022 17:41:42 - INFO - codeparrot_training - Step 30972: {'lr': 0.0001700741033962435, 'samples': 15858176, 'steps': 30972, 'loss/train': 1.1563220024108887}
+02/25/2022 17:41:48 - INFO - codeparrot_training - Step 30973: {'lr': 0.00017005859984684223, 'samples': 15858688, 'steps': 30973, 'loss/train': 0.5578577518463135}
+02/25/2022 17:41:51 - INFO - codeparrot_training - Step 30974: {'lr': 0.00017004309663988456, 'samples': 15859200, 'steps': 30974, 'loss/train': 1.5617010593414307}
+02/25/2022 17:41:57 - INFO - codeparrot_training - Step 30975: {'lr': 0.00017002759377543675, 'samples': 15859712, 'steps': 30975, 'loss/train': 8.43612289428711}
+02/25/2022 17:42:01 - INFO - codeparrot_training - Step 30976: {'lr': 0.00017001209125356543, 'samples': 15860224, 'steps': 30976, 'loss/train': 1.6823818683624268}
+02/25/2022 17:42:06 - INFO - codeparrot_training - Step 30977: {'lr': 0.00016999658907433684, 'samples': 15860736, 'steps': 30977, 'loss/train': 1.3721647262573242}
+02/25/2022 17:42:10 - INFO - codeparrot_training - Step 30978: {'lr': 0.0001699810872378175, 'samples': 15861248, 'steps': 30978, 'loss/train': 0.9308111667633057}
+02/25/2022 17:42:16 - INFO - codeparrot_training - Step 30979: {'lr': 0.00016996558574407377, 'samples': 15861760, 'steps': 30979, 'loss/train': 1.0938304662704468}
+02/25/2022 17:42:19 - INFO - codeparrot_training - Step 30980: {'lr': 0.00016995008459317208, 'samples': 15862272, 'steps': 30980, 'loss/train': 1.901100516319275}
+02/25/2022 17:42:25 - INFO - codeparrot_training - Step 30981: {'lr': 0.00016993458378517876, 'samples': 15862784, 'steps': 30981, 'loss/train': 1.3357551097869873}
+02/25/2022 17:42:28 - INFO - codeparrot_training - Step 30982: {'lr': 0.00016991908332016033, 'samples': 15863296, 'steps': 30982, 'loss/train': 1.954347014427185}
+02/25/2022 17:42:34 - INFO - codeparrot_training - Step 30983: {'lr': 0.00016990358319818308, 'samples': 15863808, 'steps': 30983, 'loss/train': 1.605167031288147}
+02/25/2022 17:42:37 - INFO - codeparrot_training - Step 30984: {'lr': 0.0001698880834193134, 'samples': 15864320, 'steps': 30984, 'loss/train': 0.49428653717041016}
+02/25/2022 17:42:43 - INFO - codeparrot_training - Step 30985: {'lr': 0.00016987258398361776, 'samples': 15864832, 'steps': 30985, 'loss/train': 2.966676712036133}
+02/25/2022 17:42:49 - INFO - codeparrot_training - Step 30986: {'lr': 0.0001698570848911625, 'samples': 15865344, 'steps': 30986, 'loss/train': 1.2860169410705566}
+02/25/2022 17:42:52 - INFO - codeparrot_training - Step 30987: {'lr': 0.00016984158614201416, 'samples': 15865856, 'steps': 30987, 'loss/train': 0.14702078700065613}
+02/25/2022 17:42:58 - INFO - codeparrot_training - Step 30988: {'lr': 0.0001698260877362389, 'samples': 15866368, 'steps': 30988, 'loss/train': 1.084276556968689}
+02/25/2022 17:43:01 - INFO - codeparrot_training - Step 30989: {'lr': 0.00016981058967390321, 'samples': 15866880, 'steps': 30989, 'loss/train': 2.4190120697021484}
+02/25/2022 17:43:07 - INFO - codeparrot_training - Step 30990: {'lr': 0.00016979509195507353, 'samples': 15867392, 'steps': 30990, 'loss/train': 0.3845204710960388}
+02/25/2022 17:43:10 - INFO - codeparrot_training - Step 30991: {'lr': 0.0001697795945798163, 'samples': 15867904, 'steps': 30991, 'loss/train': 1.586069107055664}
+02/25/2022 17:43:16 - INFO - codeparrot_training - Step 30992: {'lr': 0.00016976409754819767, 'samples': 15868416, 'steps': 30992, 'loss/train': 2.422858238220215}
+02/25/2022 17:43:20 - INFO - codeparrot_training - Step 30993: {'lr': 0.00016974860086028422, 'samples': 15868928, 'steps': 30993, 'loss/train': 2.136409282684326}
+02/25/2022 17:43:23 - INFO - codeparrot_training - Step 30994: {'lr': 0.00016973310451614225, 'samples': 15869440, 'steps': 30994, 'loss/train': 1.3699604272842407}
+02/25/2022 17:43:29 - INFO - codeparrot_training - Step 30995: {'lr': 0.00016971760851583828, 'samples': 15869952, 'steps': 30995, 'loss/train': 2.20263409614563}
+02/25/2022 17:43:33 - INFO - codeparrot_training - Step 30996: {'lr': 0.0001697021128594385, 'samples': 15870464, 'steps': 30996, 'loss/train': 1.7646503448486328}
+02/25/2022 17:43:38 - INFO - codeparrot_training - Step 30997: {'lr': 0.00016968661754700933, 'samples': 15870976, 'steps': 30997, 'loss/train': 1.811256766319275}
+02/25/2022 17:43:42 - INFO - codeparrot_training - Step 30998: {'lr': 0.00016967112257861721, 'samples': 15871488, 'steps': 30998, 'loss/train': 0.2525671422481537}
+02/25/2022 17:43:47 - INFO - codeparrot_training - Step 30999: {'lr': 0.00016965562795432848, 'samples': 15872000, 'steps': 30999, 'loss/train': 2.7075023651123047}
+02/25/2022 17:43:47 - INFO - codeparrot_training - Evaluating and saving model checkpoint