diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -27561,3 +27561,2506 @@ Use FP16 precision: False
 12/26/2021 06:18:25 - INFO - codeparrot_training - Step 24999: {'lr': 0.00025139503014026, 'samples': 12800000, 'steps': 24999, 'batch_loss/train': 0.850140267983079}
 12/26/2021 06:18:25 - INFO - codeparrot_training - Evaluating and saving model checkpoint
 12/26/2021 06:21:47 - INFO - codeparrot_training - Step 25000: {'loss/eval': 0.7836087942123413, 'perplexity': 2.189358949661255}
+12/26/2021 06:22:05 - WARNING - huggingface_hub.repository - Several commits (8) will be pushed upstream.
+12/26/2021 06:22:18 - INFO - codeparrot_training - Step 25000: {'lr': 0.0002513792672486757, 'samples': 12800512, 'steps': 25000, 'batch_loss/train': 0.7486031828448176}
+12/26/2021 06:22:30 - INFO - codeparrot_training - Step 25001: {'lr': 0.00025136350435160806, 'samples': 12801024, 'steps': 25001, 'batch_loss/train': 0.7181321987882257}
+12/26/2021 06:22:41 - INFO - codeparrot_training - Step 25002: {'lr': 0.0002513477414491195, 'samples': 12801536, 'steps': 25002, 'batch_loss/train': 0.7192767392843962}
+12/26/2021 06:22:51 - INFO - codeparrot_training - Step 25003: {'lr': 0.000251331978541273, 'samples': 12802048, 'steps': 25003, 'batch_loss/train': 0.8425130904652178}
+12/26/2021 06:23:02 - INFO - codeparrot_training - Step 25004: {'lr': 0.0002513162156281309, 'samples': 12802560, 'steps': 25004, 'batch_loss/train': 0.826509122736752}
+12/26/2021 06:23:15 - INFO - codeparrot_training - Step 25005: {'lr': 0.00025130045270975617, 'samples': 12803072, 'steps': 25005, 'batch_loss/train': 0.7450620988383889}
+12/26/2021 06:23:25 - INFO - codeparrot_training - Step 25006: {'lr': 0.0002512846897862112, 'samples': 12803584, 'steps': 25006, 'batch_loss/train': 0.793431447353214}
+12/26/2021 06:23:36 - INFO - codeparrot_training - Step 25007: {'lr': 0.0002512689268575589, 'samples': 12804096, 'steps': 25007, 'batch_loss/train': 0.7859456650912762}
+12/26/2021 06:23:48 - INFO - codeparrot_training - Step 25008: {'lr': 0.00025125316392386174, 'samples': 12804608, 'steps': 25008, 'batch_loss/train': 0.7776422398164868}
+12/26/2021 06:23:59 - INFO - codeparrot_training - Step 25009: {'lr': 0.0002512374009851824, 'samples': 12805120, 'steps': 25009, 'batch_loss/train': 0.7480822606012225}
+12/26/2021 06:24:09 - INFO - codeparrot_training - Step 25010: {'lr': 0.00025122163804158375, 'samples': 12805632, 'steps': 25010, 'batch_loss/train': 0.8441228782758117}
+12/26/2021 06:24:21 - INFO - codeparrot_training - Step 25011: {'lr': 0.00025120587509312834, 'samples': 12806144, 'steps': 25011, 'batch_loss/train': 0.7649983889423311}
+12/26/2021 06:24:32 - INFO - codeparrot_training - Step 25012: {'lr': 0.0002511901121398788, 'samples': 12806656, 'steps': 25012, 'batch_loss/train': 0.730508170556277}
+12/26/2021 06:24:42 - INFO - codeparrot_training - Step 25013: {'lr': 0.0002511743491818978, 'samples': 12807168, 'steps': 25013, 'batch_loss/train': 0.84923015255481}
+12/26/2021 06:24:53 - INFO - codeparrot_training - Step 25014: {'lr': 0.000251158586219248, 'samples': 12807680, 'steps': 25014, 'batch_loss/train': 0.6510013522347435}
+12/26/2021 06:25:06 - INFO - codeparrot_training - Step 25015: {'lr': 0.0002511428232519922, 'samples': 12808192, 'steps': 25015, 'batch_loss/train': 0.774962754920125}
+12/26/2021 06:25:17 - INFO - codeparrot_training - Step 25016: {'lr': 0.0002511270602801929, 'samples': 12808704, 'steps': 25016, 'batch_loss/train': 0.7743064537644386}
+12/26/2021 06:25:27 - INFO - codeparrot_training - Step 25017: {'lr': 0.0002511112973039128, 'samples': 12809216, 'steps': 25017, 'batch_loss/train': 0.7886867765337229}
+12/26/2021 06:25:40 - INFO - codeparrot_training - Step 25018: {'lr': 0.0002510955343232147, 'samples': 12809728, 'steps': 25018, 'batch_loss/train': 0.7100114235654473}
+12/26/2021 06:25:50 - INFO - codeparrot_training - Step 25019: {'lr': 0.0002510797713381611, 'samples': 12810240, 'steps': 25019, 'batch_loss/train': 0.7704456942155957}
+12/26/2021 06:26:01 - INFO - codeparrot_training - Step 25020: {'lr': 0.0002510640083488147, 'samples': 12810752, 'steps': 25020, 'batch_loss/train': 0.6934929229319096}
+12/26/2021 06:26:14 - INFO - codeparrot_training - Step 25021: {'lr': 0.00025104824535523827, 'samples': 12811264, 'steps': 25021, 'batch_loss/train': 0.8820293630706146}
+12/26/2021 06:26:24 - INFO - codeparrot_training - Step 25022: {'lr': 0.00025103248235749443, 'samples': 12811776, 'steps': 25022, 'batch_loss/train': 0.8309427043423057}
+12/26/2021 06:26:35 - INFO - codeparrot_training - Step 25023: {'lr': 0.00025101671935564575, 'samples': 12812288, 'steps': 25023, 'batch_loss/train': 0.6776512796059251}
+12/26/2021 06:26:46 - INFO - codeparrot_training - Step 25024: {'lr': 0.000251000956349755, 'samples': 12812800, 'steps': 25024, 'batch_loss/train': 0.7076684992644005}
+12/26/2021 06:26:58 - INFO - codeparrot_training - Step 25025: {'lr': 0.0002509851933398849, 'samples': 12813312, 'steps': 25025, 'batch_loss/train': 0.8165128990076482}
+12/26/2021 06:27:09 - INFO - codeparrot_training - Step 25026: {'lr': 0.00025096943032609797, 'samples': 12813824, 'steps': 25026, 'batch_loss/train': 0.6828194671543315}
+12/26/2021 06:27:19 - INFO - codeparrot_training - Step 25027: {'lr': 0.00025095366730845697, 'samples': 12814336, 'steps': 25027, 'batch_loss/train': 1.1038661771453917}
+12/26/2021 06:27:31 - INFO - codeparrot_training - Step 25028: {'lr': 0.00025093790428702456, 'samples': 12814848, 'steps': 25028, 'batch_loss/train': 0.89796376042068}
+12/26/2021 06:27:42 - INFO - codeparrot_training - Step 25029: {'lr': 0.0002509221412618635, 'samples': 12815360, 'steps': 25029, 'batch_loss/train': 0.7610187320969999}
+12/26/2021 06:27:52 - INFO - codeparrot_training - Step 25030: {'lr': 0.00025090637823303616, 'samples': 12815872, 'steps': 25030, 'batch_loss/train': 0.9010326601564884}
+12/26/2021 06:28:05 - INFO - codeparrot_training - Step 25031: {'lr': 0.0002508906152006055, 'samples': 12816384, 'steps': 25031, 'batch_loss/train': 0.7293249936774373}
+12/26/2021 06:28:16 - INFO - codeparrot_training - Step 25032: {'lr': 0.00025087485216463405, 'samples': 12816896, 'steps': 25032, 'batch_loss/train': 0.7494586976245046}
+12/26/2021 06:28:27 - INFO - codeparrot_training - Step 25033: {'lr': 0.00025085908912518453, 'samples': 12817408, 'steps': 25033, 'batch_loss/train': 0.6448300629854202}
+12/26/2021 06:28:37 - INFO - codeparrot_training - Step 25034: {'lr': 0.0002508433260823196, 'samples': 12817920, 'steps': 25034, 'batch_loss/train': 0.6479422824922949}
+12/26/2021 06:28:49 - INFO - codeparrot_training - Step 25035: {'lr': 0.0002508275630361019, 'samples': 12818432, 'steps': 25035, 'batch_loss/train': 0.8090747715905309}
+12/26/2021 06:29:00 - INFO - codeparrot_training - Step 25036: {'lr': 0.0002508117999865942, 'samples': 12818944, 'steps': 25036, 'batch_loss/train': 0.8346334446687251}
+12/26/2021 06:29:11 - INFO - codeparrot_training - Step 25037: {'lr': 0.00025079603693385904, 'samples': 12819456, 'steps': 25037, 'batch_loss/train': 0.7675484083592892}
+12/26/2021 06:29:25 - INFO - codeparrot_training - Step 25038: {'lr': 0.00025078027387795914, 'samples': 12819968, 'steps': 25038, 'batch_loss/train': 0.7722112098708749}
+12/26/2021 06:29:35 - INFO - codeparrot_training - Step 25039: {'lr': 0.0002507645108189572, 'samples': 12820480, 'steps': 25039, 'batch_loss/train': 0.8204971980303526}
+12/26/2021 06:29:46 - INFO - codeparrot_training - Step 25040: {'lr': 0.0002507487477569158, 'samples': 12820992, 'steps': 25040, 'batch_loss/train': 0.8561209291219711}
+12/26/2021 06:29:58 - INFO - codeparrot_training - Step 25041: {'lr': 0.0002507329846918977, 'samples': 12821504, 'steps': 25041, 'batch_loss/train': 0.8045361461117864}
+12/26/2021 06:30:09 - INFO - codeparrot_training - Step 25042: {'lr': 0.0002507172216239655, 'samples': 12822016, 'steps': 25042, 'batch_loss/train': 0.8016167664900422}
+12/26/2021 06:30:19 - INFO - codeparrot_training - Step 25043: {'lr': 0.0002507014585531819, 'samples': 12822528, 'steps': 25043, 'batch_loss/train': 1.4926398913376033}
+12/26/2021 06:30:30 - INFO - codeparrot_training - Step 25044: {'lr': 0.0002506856954796096, 'samples': 12823040, 'steps': 25044, 'batch_loss/train': 0.7300555454567075}
+12/26/2021 06:30:42 - INFO - codeparrot_training - Step 25045: {'lr': 0.0002506699324033112, 'samples': 12823552, 'steps': 25045, 'batch_loss/train': 0.7691471693105996}
+12/26/2021 06:30:53 - INFO - codeparrot_training - Step 25046: {'lr': 0.00025065416932434936, 'samples': 12824064, 'steps': 25046, 'batch_loss/train': 0.7339524396229535}
+12/26/2021 06:31:03 - INFO - codeparrot_training - Step 25047: {'lr': 0.0002506384062427869, 'samples': 12824576, 'steps': 25047, 'batch_loss/train': 0.8020094148814678}
+12/26/2021 06:31:17 - INFO - codeparrot_training - Step 25048: {'lr': 0.00025062264315868634, 'samples': 12825088, 'steps': 25048, 'batch_loss/train': 0.8317256150767207}
+12/26/2021 06:31:27 - INFO - codeparrot_training - Step 25049: {'lr': 0.0002506068800721103, 'samples': 12825600, 'steps': 25049, 'batch_loss/train': 0.7649759287014604}
+12/26/2021 06:31:38 - INFO - codeparrot_training - Step 25050: {'lr': 0.0002505911169831216, 'samples': 12826112, 'steps': 25050, 'batch_loss/train': 0.7396461516618729}
+12/26/2021 06:31:50 - INFO - codeparrot_training - Step 25051: {'lr': 0.00025057535389178285, 'samples': 12826624, 'steps': 25051, 'batch_loss/train': 0.68362484453246}
+12/26/2021 06:32:01 - INFO - codeparrot_training - Step 25052: {'lr': 0.0002505595907981568, 'samples': 12827136, 'steps': 25052, 'batch_loss/train': 0.7610949499066919}
+12/26/2021 06:32:11 - INFO - codeparrot_training - Step 25053: {'lr': 0.0002505438277023059, 'samples': 12827648, 'steps': 25053, 'batch_loss/train': 0.7911650519818068}
+12/26/2021 06:32:22 - INFO - codeparrot_training - Step 25054: {'lr': 0.00025052806460429297, 'samples': 12828160, 'steps': 25054, 'batch_loss/train': 0.8082429440692067}
+12/26/2021 06:32:36 - INFO - codeparrot_training - Step 25055: {'lr': 0.0002505123015041806, 'samples': 12828672, 'steps': 25055, 'batch_loss/train': 0.7483472656458616}
+12/26/2021 06:32:47 - INFO - codeparrot_training - Step 25056: {'lr': 0.00025049653840203157, 'samples': 12829184, 'steps': 25056, 'batch_loss/train': 0.7184061212465167}
+12/26/2021 06:32:57 - INFO - codeparrot_training - Step 25057: {'lr': 0.0002504807752979085, 'samples': 12829696, 'steps': 25057, 'batch_loss/train': 0.7280578874051571}
+12/26/2021 06:33:10 - INFO - codeparrot_training - Step 25058: {'lr': 0.00025046501219187404, 'samples': 12830208, 'steps': 25058, 'batch_loss/train': 0.7821520059369504}
+12/26/2021 06:33:20 - INFO - codeparrot_training - Step 25059: {'lr': 0.0002504492490839909, 'samples': 12830720, 'steps': 25059, 'batch_loss/train': 0.7844573194161057}
+12/26/2021 06:33:31 - INFO - codeparrot_training - Step 25060: {'lr': 0.0002504334859743217, 'samples': 12831232, 'steps': 25060, 'batch_loss/train': 0.6604930923786014}
+12/26/2021 06:33:43 - INFO - codeparrot_training - Step 25061: {'lr': 0.0002504177228629291, 'samples': 12831744, 'steps': 25061, 'batch_loss/train': 0.6798931136727333}
+12/26/2021 06:33:54 - INFO - codeparrot_training - Step 25062: {'lr': 0.0002504019597498758, 'samples': 12832256, 'steps': 25062, 'batch_loss/train': 0.7029382777400315}
+12/26/2021 06:34:04 - INFO - codeparrot_training - Step 25063: {'lr': 0.00025038619663522444, 'samples': 12832768, 'steps': 25063, 'batch_loss/train': 0.8242754535749555}
+12/26/2021 06:34:15 - INFO - codeparrot_training - Step 25064: {'lr': 0.00025037043351903776, 'samples': 12833280, 'steps': 25064, 'batch_loss/train': 0.762474010232836}
+12/26/2021 06:34:28 - INFO - codeparrot_training - Step 25065: {'lr': 0.0002503546704013783, 'samples': 12833792, 'steps': 25065, 'batch_loss/train': 0.8155861888080835}
+12/26/2021 06:34:39 - INFO - codeparrot_training - Step 25066: {'lr': 0.0002503389072823089, 'samples': 12834304, 'steps': 25066, 'batch_loss/train': 0.8117594355717301}
+12/26/2021 06:34:49 - INFO - codeparrot_training - Step 25067: {'lr': 0.00025032314416189205, 'samples': 12834816, 'steps': 25067, 'batch_loss/train': 0.6958929365500808}
+12/26/2021 06:35:01 - INFO - codeparrot_training - Step 25068: {'lr': 0.0002503073810401905, 'samples': 12835328, 'steps': 25068, 'batch_loss/train': 0.8559302669018507}
+12/26/2021 06:35:12 - INFO - codeparrot_training - Step 25069: {'lr': 0.000250291617917267, 'samples': 12835840, 'steps': 25069, 'batch_loss/train': 0.7728042753878981}
+12/26/2021 06:35:23 - INFO - codeparrot_training - Step 25070: {'lr': 0.00025027585479318404, 'samples': 12836352, 'steps': 25070, 'batch_loss/train': 0.6847943116445094}
+12/26/2021 06:35:35 - INFO - codeparrot_training - Step 25071: {'lr': 0.0002502600916680045, 'samples': 12836864, 'steps': 25071, 'batch_loss/train': 0.7892649853602052}
+12/26/2021 06:35:45 - INFO - codeparrot_training - Step 25072: {'lr': 0.0002502443285417908, 'samples': 12837376, 'steps': 25072, 'batch_loss/train': 0.8463014224544168}
+12/26/2021 06:35:56 - INFO - codeparrot_training - Step 25073: {'lr': 0.0002502285654146059, 'samples': 12837888, 'steps': 25073, 'batch_loss/train': 0.8712224217306357}
+12/26/2021 06:36:07 - INFO - codeparrot_training - Step 25074: {'lr': 0.00025021280228651216, 'samples': 12838400, 'steps': 25074, 'batch_loss/train': 0.8430936792865396}
+12/26/2021 06:36:19 - INFO - codeparrot_training - Step 25075: {'lr': 0.0002501970391575725, 'samples': 12838912, 'steps': 25075, 'batch_loss/train': 0.8203224837779999}
+12/26/2021 06:36:29 - INFO - codeparrot_training - Step 25076: {'lr': 0.0002501812760278494, 'samples': 12839424, 'steps': 25076, 'batch_loss/train': 0.8165245950222015}
+12/26/2021 06:36:40 - INFO - codeparrot_training - Step 25077: {'lr': 0.0002501655128974057, 'samples': 12839936, 'steps': 25077, 'batch_loss/train': 0.8602843016851693}
+12/26/2021 06:36:53 - INFO - codeparrot_training - Step 25078: {'lr': 0.0002501497497663039, 'samples': 12840448, 'steps': 25078, 'batch_loss/train': 0.8666546577587724}
+12/26/2021 06:37:04 - INFO - codeparrot_training - Step 25079: {'lr': 0.0002501339866346068, 'samples': 12840960, 'steps': 25079, 'batch_loss/train': 0.8109909556806087}
+12/26/2021 06:37:14 - INFO - codeparrot_training - Step 25080: {'lr': 0.000250118223502377, 'samples': 12841472, 'steps': 25080, 'batch_loss/train': 0.8109851409681141}
+12/26/2021 06:37:26 - INFO - codeparrot_training - Step 25081: {'lr': 0.0002501024603696772, 'samples': 12841984, 'steps': 25081, 'batch_loss/train': 0.8219738341867924}
+12/26/2021 06:37:37 - INFO - codeparrot_training - Step 25082: {'lr': 0.0002500866972365701, 'samples': 12842496, 'steps': 25082, 'batch_loss/train': 0.824177891947329}
+12/26/2021 06:37:48 - INFO - codeparrot_training - Step 25083: {'lr': 0.0002500709341031183, 'samples': 12843008, 'steps': 25083, 'batch_loss/train': 0.7640428580343723}
+12/26/2021 06:38:02 - INFO - codeparrot_training - Step 25084: {'lr': 0.0002500551709693845, 'samples': 12843520, 'steps': 25084, 'batch_loss/train': 0.8038536254316568}
+12/26/2021 06:38:12 - INFO - codeparrot_training - Step 25085: {'lr': 0.0002500394078354312, 'samples': 12844032, 'steps': 25085, 'batch_loss/train': 0.7495944211259484}
+12/26/2021 06:38:23 - INFO - codeparrot_training - Step 25086: {'lr': 0.00025002364470132145, 'samples': 12844544, 'steps': 25086, 'batch_loss/train': 0.7743529570288956}
+12/26/2021 06:38:34 - INFO - codeparrot_training - Step 25087: {'lr': 0.0002500078815671176, 'samples': 12845056, 'steps': 25087, 'batch_loss/train': 0.7381405476480722}
+12/26/2021 06:38:46 - INFO - codeparrot_training - Step 25088: {'lr': 0.00024999211843288247, 'samples': 12845568, 'steps': 25088, 'batch_loss/train': 0.8159225331619382}
+12/26/2021 06:38:56 - INFO - codeparrot_training - Step 25089: {'lr': 0.0002499763552986786, 'samples': 12846080, 'steps': 25089, 'batch_loss/train': 0.7501355865970254}
+12/26/2021 06:39:07 - INFO - codeparrot_training - Step 25090: {'lr': 0.0002499605921645688, 'samples': 12846592, 'steps': 25090, 'batch_loss/train': 0.8090883437544107}
+12/26/2021 06:39:19 - INFO - codeparrot_training - Step 25091: {'lr': 0.00024994482903061563, 'samples': 12847104, 'steps': 25091, 'batch_loss/train': 0.8196591201703995}
+12/26/2021 06:39:30 - INFO - codeparrot_training - Step 25092: {'lr': 0.0002499290658968817, 'samples': 12847616, 'steps': 25092, 'batch_loss/train': 0.6432130634784698}
+12/26/2021 06:39:40 - INFO - codeparrot_training - Step 25093: {'lr': 0.00024991330276343, 'samples': 12848128, 'steps': 25093, 'batch_loss/train': 0.7690113387070596}
+12/26/2021 06:39:53 - INFO - codeparrot_training - Step 25094: {'lr': 0.0002498975396303228, 'samples': 12848640, 'steps': 25094, 'batch_loss/train': 0.8569069574587047}
+12/26/2021 06:40:04 - INFO - codeparrot_training - Step 25095: {'lr': 0.00024988177649762296, 'samples': 12849152, 'steps': 25095, 'batch_loss/train': 0.6780301304534078}
+12/26/2021 06:40:15 - INFO - codeparrot_training - Step 25096: {'lr': 0.0002498660133653932, 'samples': 12849664, 'steps': 25096, 'batch_loss/train': 0.8248353535309434}
+12/26/2021 06:40:25 - INFO - codeparrot_training - Step 25097: {'lr': 0.00024985025023369615, 'samples': 12850176, 'steps': 25097, 'batch_loss/train': 0.8412895491346717}
+12/26/2021 06:40:37 - INFO - codeparrot_training - Step 25098: {'lr': 0.0002498344871025944, 'samples': 12850688, 'steps': 25098, 'batch_loss/train': 0.7135917525738478}
+12/26/2021 06:40:48 - INFO - codeparrot_training - Step 25099: {'lr': 0.00024981872397215063, 'samples': 12851200, 'steps': 25099, 'batch_loss/train': 0.7319100480526686}
+12/26/2021 06:40:59 - INFO - codeparrot_training - Step 25100: {'lr': 0.00024980296084242757, 'samples': 12851712, 'steps': 25100, 'batch_loss/train': 0.8136328440159559}
+12/26/2021 06:41:12 - INFO - codeparrot_training - Step 25101: {'lr': 0.00024978719771348785, 'samples': 12852224, 'steps': 25101, 'batch_loss/train': 0.8135204399004579}
+12/26/2021 06:41:22 - INFO - codeparrot_training - Step 25102: {'lr': 0.00024977143458539423, 'samples': 12852736, 'steps': 25102, 'batch_loss/train': 0.8123887374531478}
+12/26/2021 06:41:33 - INFO - codeparrot_training - Step 25103: {'lr': 0.0002497556714582092, 'samples': 12853248, 'steps': 25103, 'batch_loss/train': 0.928447462618351}
+12/26/2021 06:41:45 - INFO - codeparrot_training - Step 25104: {'lr': 0.00024973990833199557, 'samples': 12853760, 'steps': 25104, 'batch_loss/train': 0.8239318253472447}
+12/26/2021 06:41:56 - INFO - codeparrot_training - Step 25105: {'lr': 0.00024972414520681597, 'samples': 12854272, 'steps': 25105, 'batch_loss/train': 0.7832867875695229}
+12/26/2021 06:42:06 - INFO - codeparrot_training - Step 25106: {'lr': 0.00024970838208273304, 'samples': 12854784, 'steps': 25106, 'batch_loss/train': 0.8070274833589792}
+12/26/2021 06:42:17 - INFO - codeparrot_training - Step 25107: {'lr': 0.0002496926189598095, 'samples': 12855296, 'steps': 25107, 'batch_loss/train': 0.8040147256106138}
+12/26/2021 06:42:29 - INFO - codeparrot_training - Step 25108: {'lr': 0.00024967685583810796, 'samples': 12855808, 'steps': 25108, 'batch_loss/train': 0.6460763679351658}
+12/26/2021 06:42:40 - INFO - codeparrot_training - Step 25109: {'lr': 0.0002496610927176912, 'samples': 12856320, 'steps': 25109, 'batch_loss/train': 0.7477070325985551}
+12/26/2021 06:42:50 - INFO - codeparrot_training - Step 25110: {'lr': 0.0002496453295986217, 'samples': 12856832, 'steps': 25110, 'batch_loss/train': 0.8281011367216706}
+12/26/2021 06:43:04 - INFO - codeparrot_training - Step 25111: {'lr': 0.00024962956648096225, 'samples': 12857344, 'steps': 25111, 'batch_loss/train': 0.6982504073530436}
+12/26/2021 06:43:14 - INFO - codeparrot_training - Step 25112: {'lr': 0.0002496138033647756, 'samples': 12857856, 'steps': 25112, 'batch_loss/train': 0.7600904619321227}
+12/26/2021 06:43:25 - INFO - codeparrot_training - Step 25113: {'lr': 0.00024959804025012427, 'samples': 12858368, 'steps': 25113, 'batch_loss/train': 0.7655067087616771}
+12/26/2021 06:43:37 - INFO - codeparrot_training - Step 25114: {'lr': 0.0002495822771370709, 'samples': 12858880, 'steps': 25114, 'batch_loss/train': 0.8308038571849465}
+12/26/2021 06:43:48 - INFO - codeparrot_training - Step 25115: {'lr': 0.00024956651402567836, 'samples': 12859392, 'steps': 25115, 'batch_loss/train': 0.6360157998278737}
+12/26/2021 06:43:58 - INFO - codeparrot_training - Step 25116: {'lr': 0.0002495507509160092, 'samples': 12859904, 'steps': 25116, 'batch_loss/train': 0.7919766688719392}
+12/26/2021 06:44:10 - INFO - codeparrot_training - Step 25117: {'lr': 0.000249534987808126, 'samples': 12860416, 'steps': 25117, 'batch_loss/train': 0.6969919735565782}
+12/26/2021 06:44:21 - INFO - codeparrot_training - Step 25118: {'lr': 0.0002495192247020915, 'samples': 12860928, 'steps': 25118, 'batch_loss/train': 0.7754886150360107}
+12/26/2021 06:44:32 - INFO - codeparrot_training - Step 25119: {'lr': 0.0002495034615979685, 'samples': 12861440, 'steps': 25119, 'batch_loss/train': 0.7734495415352285}
+12/26/2021 06:44:42 - INFO - codeparrot_training - Step 25120: {'lr': 0.0002494876984958194, 'samples': 12861952, 'steps': 25120, 'batch_loss/train': 0.9501345902681351}
+12/26/2021 06:44:54 - INFO - codeparrot_training - Step 25121: {'lr': 0.00024947193539570704, 'samples': 12862464, 'steps': 25121, 'batch_loss/train': 0.6086930404417217}
+12/26/2021 06:45:05 - INFO - codeparrot_training - Step 25122: {'lr': 0.0002494561722976942, 'samples': 12862976, 'steps': 25122, 'batch_loss/train': 0.6249937597895041}
+12/26/2021 06:45:16 - INFO - codeparrot_training - Step 25123: {'lr': 0.0002494404092018433, 'samples': 12863488, 'steps': 25123, 'batch_loss/train': 0.8264021137729287}
+12/26/2021 06:45:29 - INFO - codeparrot_training - Step 25124: {'lr': 0.0002494246461082172, 'samples': 12864000, 'steps': 25124, 'batch_loss/train': 0.7581469791475683}
+12/26/2021 06:45:40 - INFO - codeparrot_training - Step 25125: {'lr': 0.0002494088830168784, 'samples': 12864512, 'steps': 25125, 'batch_loss/train': 0.7333409847924486}
+12/26/2021 06:45:51 - INFO - codeparrot_training - Step 25126: {'lr': 0.0002493931199278897, 'samples': 12865024, 'steps': 25126, 'batch_loss/train': 0.7781597096472979}
+12/26/2021 06:46:01 - INFO - codeparrot_training - Step 25127: {'lr': 0.0002493773568413137, 'samples': 12865536, 'steps': 25127, 'batch_loss/train': 0.8451296836137772}
+12/26/2021 06:46:14 - INFO - codeparrot_training - Step 25128: {'lr': 0.0002493615937572132, 'samples': 12866048, 'steps': 25128, 'batch_loss/train': 0.8305618958547711}
+12/26/2021 06:46:24 - INFO - codeparrot_training - Step 25129: {'lr': 0.0002493458306756506, 'samples': 12866560, 'steps': 25129, 'batch_loss/train': 0.7346009504981339}
+12/26/2021 06:46:35 - INFO - codeparrot_training - Step 25130: {'lr': 0.00024933006759668885, 'samples': 12867072, 'steps': 25130, 'batch_loss/train': 0.8255310375243425}
+12/26/2021 06:46:49 - INFO - codeparrot_training - Step 25131: {'lr': 0.00024931430452039045, 'samples': 12867584, 'steps': 25131, 'batch_loss/train': 0.8082596897147596}
+12/26/2021 06:46:59 - INFO - codeparrot_training - Step 25132: {'lr': 0.0002492985414468181, 'samples': 12868096, 'steps': 25132, 'batch_loss/train': 0.7954719422850758}
+12/26/2021 06:47:10 - INFO - codeparrot_training - Step 25133: {'lr': 0.00024928277837603456, 'samples': 12868608, 'steps': 25133, 'batch_loss/train': 0.8195606218650937}
+12/26/2021 06:47:22 - INFO - codeparrot_training - Step 25134: {'lr': 0.00024926701530810234, 'samples': 12869120, 'steps': 25134, 'batch_loss/train': 0.7434855596511625}
+12/26/2021 06:47:33 - INFO - codeparrot_training - Step 25135: {'lr': 0.0002492512522430843, 'samples': 12869632, 'steps': 25135, 'batch_loss/train': 0.7310621296055615}
+12/26/2021 06:47:43 - INFO - codeparrot_training - Step 25136: {'lr': 0.00024923548918104284, 'samples': 12870144, 'steps': 25136, 'batch_loss/train': 0.8429504497908056}
+12/26/2021 06:47:56 - INFO - codeparrot_training - Step 25137: {'lr': 0.00024921972612204087, 'samples': 12870656, 'steps': 25137, 'batch_loss/train': 0.5821855196263641}
+12/26/2021 06:48:07 - INFO - codeparrot_training - Step 25138: {'lr': 0.00024920396306614103, 'samples': 12871168, 'steps': 25138, 'batch_loss/train': 0.6593735608039424}
+12/26/2021 06:48:17 - INFO - codeparrot_training - Step 25139: {'lr': 0.00024918820001340586, 'samples': 12871680, 'steps': 25139, 'batch_loss/train': 0.852902359329164}
+12/26/2021 06:48:28 - INFO - codeparrot_training - Step 25140: {'lr': 0.00024917243696389804, 'samples': 12872192, 'steps': 25140, 'batch_loss/train': 0.7759818159975111}
+12/26/2021 06:48:42 - INFO - codeparrot_training - Step 25141: {'lr': 0.00024915667391768045, 'samples': 12872704, 'steps': 25141, 'batch_loss/train': 0.7678036973811686}
+12/26/2021 06:48:53 - INFO - codeparrot_training - Step 25142: {'lr': 0.00024914091087481553, 'samples': 12873216, 'steps': 25142, 'batch_loss/train': 0.7463054198306054}
+12/26/2021 06:49:03 - INFO - codeparrot_training - Step 25143: {'lr': 0.00024912514783536596, 'samples': 12873728, 'steps': 25143, 'batch_loss/train': 0.7507673408836126}
+12/26/2021 06:49:15 - INFO - codeparrot_training - Step 25144: {'lr': 0.00024910938479939457, 'samples': 12874240, 'steps': 25144, 'batch_loss/train': 0.8576031904667616}
+12/26/2021 06:49:26 - INFO - codeparrot_training - Step 25145: {'lr': 0.0002490936217669639, 'samples': 12874752, 'steps': 25145, 'batch_loss/train': 0.8438692688941956}
+12/26/2021 06:49:37 - INFO - codeparrot_training - Step 25146: {'lr': 0.00024907785873813664, 'samples': 12875264, 'steps': 25146, 'batch_loss/train': 0.7079036742215976}
+12/26/2021 06:49:49 - INFO - codeparrot_training - Step 25147: {'lr': 0.0002490620957129754, 'samples': 12875776, 'steps': 25147, 'batch_loss/train': 0.8087858906947076}
+12/26/2021 06:49:59 - INFO - codeparrot_training - Step 25148: {'lr': 0.00024904633269154304, 'samples': 12876288, 'steps': 25148, 'batch_loss/train': 0.7815684415400028}
+12/26/2021 06:50:10 - INFO - codeparrot_training - Step 25149: {'lr': 0.00024903056967390204, 'samples': 12876800, 'steps': 25149, 'batch_loss/train': 0.7710384301608428}
+12/26/2021 06:50:20 - INFO - codeparrot_training - Step 25150: {'lr': 0.00024901480666011515, 'samples': 12877312, 'steps': 25150, 'batch_loss/train': 0.70039829518646}
+12/26/2021 06:50:33 - INFO - codeparrot_training - Step 25151: {'lr': 0.00024899904365024495, 'samples': 12877824, 'steps': 25151, 'batch_loss/train': 0.7677603964693844}
+12/26/2021 06:50:43 - INFO - codeparrot_training - Step 25152: {'lr': 0.00024898328064435426, 'samples': 12878336, 'steps': 25152, 'batch_loss/train': 0.710736081469804}
+12/26/2021 06:50:54 - INFO - codeparrot_training - Step 25153: {'lr': 0.00024896751764250563, 'samples': 12878848, 'steps': 25153, 'batch_loss/train': 0.8221833747811615}
+12/26/2021 06:51:08 - INFO - codeparrot_training - Step 25154: {'lr': 0.0002489517546447618, 'samples': 12879360, 'steps': 25154, 'batch_loss/train': 0.7117111319676042}
+12/26/2021 06:51:18 - INFO - codeparrot_training - Step 25155: {'lr': 0.0002489359916511853, 'samples': 12879872, 'steps': 25155, 'batch_loss/train': 0.6154463244602084}
+12/26/2021 06:51:29 - INFO - codeparrot_training - Step 25156: {'lr': 0.00024892022866183894, 'samples': 12880384, 'steps': 25156, 'batch_loss/train': 0.7968365116976202}
+12/26/2021 06:51:40 - INFO - codeparrot_training - Step 25157: {'lr': 0.0002489044656767854, 'samples': 12880896, 'steps': 25157, 'batch_loss/train': 0.7369305486790836}
+12/26/2021 06:51:52 - INFO - codeparrot_training - Step 25158: {'lr': 0.0002488887026960872, 'samples': 12881408, 'steps': 25158, 'batch_loss/train': 0.8034469457343221}
+12/26/2021 06:52:02 - INFO - codeparrot_training - Step 25159: {'lr': 0.0002488729397198071, 'samples': 12881920, 'steps': 25159, 'batch_loss/train': 0.8813261836767197}
+12/26/2021 06:52:13 - INFO - codeparrot_training - Step 25160: {'lr': 0.0002488571767480079, 'samples': 12882432, 'steps': 25160, 'batch_loss/train': 0.7850959744537249}
+12/26/2021 06:52:25 - INFO - codeparrot_training - Step 25161: {'lr': 0.00024884141378075204, 'samples': 12882944, 'steps': 25161, 'batch_loss/train': 0.7900625499896705}
+12/26/2021 06:52:36 - INFO - codeparrot_training - Step 25162: {'lr': 0.0002488256508181022, 'samples': 12883456, 'steps': 25162, 'batch_loss/train': 0.7361171576194465}
+12/26/2021 06:52:46 - INFO - codeparrot_training - Step 25163: {'lr': 0.00024880988786012126, 'samples': 12883968, 'steps': 25163, 'batch_loss/train': 0.7151773679070175}
+12/26/2021 06:53:00 - INFO - codeparrot_training - Step 25164: {'lr': 0.00024879412490687173, 'samples': 12884480, 'steps': 25164, 'batch_loss/train': 0.7105048722587526}
+12/26/2021 06:53:11 - INFO - codeparrot_training - Step 25165: {'lr': 0.00024877836195841627, 'samples': 12884992, 'steps': 25165, 'batch_loss/train': 0.7667989786714315}
+12/26/2021 06:53:21 - INFO - codeparrot_training - Step 25166: {'lr': 0.00024876259901481754, 'samples': 12885504, 'steps': 25166, 'batch_loss/train': 0.7538462462835014}
+12/26/2021 06:53:34 - INFO - codeparrot_training - Step 25167: {'lr': 0.00024874683607613833, 'samples': 12886016, 'steps': 25167, 'batch_loss/train': 0.7253105510026217}
+12/26/2021 06:53:44 - INFO - codeparrot_training - Step 25168: {'lr': 0.0002487310731424412, 'samples': 12886528, 'steps': 25168, 'batch_loss/train': 0.5715466148685664}
+12/26/2021 06:53:55 - INFO - codeparrot_training - Step 25169: {'lr': 0.0002487153102137888, 'samples': 12887040, 'steps': 25169, 'batch_loss/train': 0.7002727170474827}
+12/26/2021 06:54:05 - INFO - codeparrot_training - Step 25170: {'lr': 0.0002486995472902439, 'samples': 12887552, 'steps': 25170, 'batch_loss/train': 0.8182336259633303}
+12/26/2021 06:54:19 - INFO - codeparrot_training - Step 25171: {'lr': 0.00024868378437186914, 'samples': 12888064, 'steps': 25171, 'batch_loss/train': 0.7659212308935821}
+12/26/2021 06:54:30 - INFO - codeparrot_training - Step 25172: {'lr': 0.0002486680214587271, 'samples': 12888576, 'steps': 25172, 'batch_loss/train': 0.7304374193772674}
+12/26/2021 06:54:40 - INFO - codeparrot_training - Step 25173: {'lr': 0.00024865225855088044, 'samples': 12889088, 'steps': 25173, 'batch_loss/train': 0.8240761044435203}
+12/26/2021 06:54:53 - INFO - codeparrot_training - Step 25174: {'lr': 0.000248636495648392, 'samples': 12889600, 'steps': 25174, 'batch_loss/train': 0.8811975559219718}
+12/26/2021 06:55:03 - INFO - codeparrot_training - Step 25175: {'lr': 0.0002486207327513243, 'samples': 12890112, 'steps': 25175, 'batch_loss/train': 0.7502259267494082}
+12/26/2021 06:55:14 - INFO - codeparrot_training - Step 25176: {'lr': 0.0002486049698597401, 'samples': 12890624, 'steps': 25176, 'batch_loss/train': 0.758040162269026}
+12/26/2021 06:55:24 - INFO - codeparrot_training - Step 25177: {'lr': 0.0002485892069737019, 'samples': 12891136, 'steps': 25177, 'batch_loss/train': 0.7791593314614147}
+12/26/2021 06:55:37 - INFO - codeparrot_training - Step 25178: {'lr': 0.0002485734440932726, 'samples': 12891648, 'steps': 25178, 'batch_loss/train': 0.7071819272823632}
+12/26/2021 06:55:47 - INFO - codeparrot_training - Step 25179: {'lr': 0.0002485576812185147, 'samples': 12892160, 'steps': 25179, 'batch_loss/train': 0.7747503183782101}
+12/26/2021 06:55:58 - INFO - codeparrot_training - Step 25180: {'lr': 0.0002485419183494908, 'samples': 12892672, 'steps': 25180, 'batch_loss/train': 0.8664773162454367}
+12/26/2021 06:56:11 - INFO - codeparrot_training - Step 25181: {'lr': 0.00024852615548626384, 'samples': 12893184, 'steps': 25181, 'batch_loss/train': 0.8101812009699643}
+12/26/2021 06:56:21 - INFO - codeparrot_training - Step 25182: {'lr': 0.00024851039262889626, 'samples': 12893696, 'steps': 25182, 'batch_loss/train': 0.6996027354616672}
+12/26/2021 06:56:32 - INFO - codeparrot_training - Step 25183: {'lr': 0.0002484946297774509, 'samples': 12894208, 'steps': 25183, 'batch_loss/train': 0.6872691279277205}
+12/26/2021 06:56:44 - INFO - codeparrot_training - Step 25184: {'lr': 0.0002484788669319902, 'samples': 12894720, 'steps': 25184, 'batch_loss/train': 0.7989580417051911}
+12/26/2021 06:56:55 - INFO - codeparrot_training - Step 25185: {'lr': 0.00024846310409257693, 'samples': 12895232, 'steps': 25185, 'batch_loss/train': 0.6227215928956866}
+12/26/2021 06:57:05 - INFO - codeparrot_training - Step 25186: {'lr': 0.0002484473412592739, 'samples': 12895744, 'steps': 25186, 'batch_loss/train': 0.847647812217474}
+12/26/2021 06:57:16 - INFO - codeparrot_training - Step 25187: {'lr': 0.0002484315784321436, 'samples': 12896256, 'steps': 25187, 'batch_loss/train': 0.8474017744883895}
+12/26/2021 06:57:28 - INFO - codeparrot_training - Step 25188: {'lr': 0.00024841581561124866, 'samples': 12896768, 'steps': 25188, 'batch_loss/train': 0.7684995625168085}
+12/26/2021 06:57:39 - INFO - codeparrot_training - Step 25189: {'lr': 0.00024840005279665195, 'samples': 12897280, 'steps': 25189, 'batch_loss/train': 0.8822960983961821}
+12/26/2021 06:57:49 - INFO - codeparrot_training - Step 25190: {'lr': 0.000248384289988416, 'samples': 12897792, 'steps': 25190, 'batch_loss/train': 0.7692431528121233}
+12/26/2021 06:58:02 - INFO - codeparrot_training - Step 25191: {'lr': 0.0002483685271866035, 'samples': 12898304, 'steps': 25191, 'batch_loss/train': 0.7712049782276154}
+12/26/2021 06:58:13 - INFO - codeparrot_training - Step 25192: {'lr': 0.0002483527643912771, 'samples': 12898816, 'steps': 25192, 'batch_loss/train': 0.7923592193983495}
+12/26/2021 06:58:23 - INFO - codeparrot_training - Step 25193: {'lr': 0.00024833700160249955, 'samples': 12899328, 'steps': 25193, 'batch_loss/train': 0.7270699713844806}
+12/26/2021 06:58:35 - INFO - codeparrot_training - Step 25194: {'lr': 0.0002483212388203334, 'samples': 12899840, 'steps': 25194, 'batch_loss/train': 0.6735094654140994}
+12/26/2021 06:58:46 - INFO - codeparrot_training - Step 25195: {'lr': 0.0002483054760448413, 'samples': 12900352, 'steps': 25195, 'batch_loss/train': 0.8068630769848824}
+12/26/2021 06:58:57 - INFO - codeparrot_training - Step 25196: {'lr': 0.00024828971327608615, 'samples': 12900864, 'steps': 25196, 'batch_loss/train': 0.7272345228120685}
+12/26/2021 06:59:07 - INFO - codeparrot_training - Step 25197: {'lr': 0.0002482739505141304, 'samples': 12901376, 'steps': 25197, 'batch_loss/train': 0.7435132982209325}
+12/26/2021 06:59:19 - INFO - codeparrot_training - Step 25198: {'lr': 0.00024825818775903673, 'samples': 12901888, 'steps': 25198, 'batch_loss/train': 1.0053739519789815}
+12/26/2021 06:59:30 - INFO - codeparrot_training - Step 25199: {'lr': 0.0002482424250108678, 'samples': 12902400, 'steps': 25199, 'batch_loss/train': 0.8014832139015198}
+12/26/2021 06:59:40 - INFO - codeparrot_training - Step 25200: {'lr': 0.0002482266622696864, 'samples': 12902912, 'steps': 25200, 'batch_loss/train': 0.7613943028263748}
+12/26/2021 06:59:54 - INFO - codeparrot_training - Step 25201: {'lr': 0.0002482108995355551, 'samples': 12903424, 'steps': 25201, 'batch_loss/train': 0.6531531689688563}
+12/26/2021 07:00:04 - INFO - codeparrot_training - Step 25202: {'lr': 0.00024819513680853666, 'samples': 12903936, 'steps': 25202, 'batch_loss/train': 0.8714314689859748}
+12/26/2021 07:00:15 - INFO - codeparrot_training - Step 25203: {'lr': 0.0002481793740886935, 'samples': 12904448, 'steps': 25203, 'batch_loss/train': 0.4329336302471347}
+12/26/2021 07:00:28 - INFO - codeparrot_training - Step 25204: {'lr': 0.00024816361137608856, 'samples': 12904960, 'steps': 25204, 'batch_loss/train': 0.715070887003094}
+12/26/2021 07:00:39 - INFO - codeparrot_training - Step 25205: {'lr': 0.00024814784867078447, 'samples': 12905472, 'steps': 25205, 'batch_loss/train': 0.633795278146863}
+12/26/2021 07:00:49 - INFO - codeparrot_training - Step 25206: {'lr': 0.0002481320859728438, 'samples': 12905984, 'steps': 25206, 'batch_loss/train': 0.7792302202433348}
+12/26/2021 07:01:00 - INFO - codeparrot_training - Step 25207: {'lr': 0.00024811632328232923, 'samples': 12906496, 'steps': 25207, 'batch_loss/train': 1.1545892343856394}
+12/26/2021 07:01:12 - INFO - codeparrot_training - Step 25208: {'lr': 0.00024810056059930343, 'samples': 12907008, 'steps': 25208, 'batch_loss/train': 0.8924387460574508}
+12/26/2021 07:01:23 - INFO - codeparrot_training - Step 25209: {'lr': 0.00024808479792382915, 'samples': 12907520, 'steps': 25209, 'batch_loss/train': 0.7427454718854278}
+12/26/2021 07:01:33 - INFO - codeparrot_training - Step 25210: {'lr': 0.0002480690352559689, 'samples': 12908032, 'steps': 25210, 'batch_loss/train': 0.912970220670104}
+12/26/2021 07:01:46 - INFO - codeparrot_training - Step 25211: {'lr': 0.0002480532725957855, 'samples': 12908544, 'steps': 25211, 'batch_loss/train': 0.8438984034582973}
+12/26/2021 07:01:57 - INFO - codeparrot_training - Step 25212: {'lr': 0.00024803750994334157, 'samples': 12909056, 'steps': 25212, 'batch_loss/train': 0.8386733194347471}
+12/26/2021 07:02:07 - INFO - codeparrot_training - Step 25213: {'lr': 0.0002480217472986997, 'samples': 12909568, 'steps': 25213, 'batch_loss/train': 0.818516792729497}
+12/26/2021 07:02:19 - INFO - codeparrot_training - Step 25214: {'lr': 0.00024800598466192256, 'samples': 12910080, 'steps': 25214, 'batch_loss/train': 0.5850690361112356}
+12/26/2021 07:02:30 - INFO - codeparrot_training - Step 25215: {'lr': 0.000247990222033073, 'samples': 12910592, 'steps': 25215, 'batch_loss/train': 0.8122101118788123}
+12/26/2021 07:02:41 - INFO - codeparrot_training - Step 25216: {'lr': 0.00024797445941221355, 'samples': 12911104, 'steps': 25216, 'batch_loss/train': 0.8072720803320408}
+12/26/2021 07:02:51 - INFO - codeparrot_training - Step 25217: {'lr': 0.0002479586967994068, 'samples': 12911616, 'steps': 25217, 'batch_loss/train': 0.7236932758241892}
+12/26/2021 07:03:05 - INFO - codeparrot_training - Step 25218: {'lr': 0.00024794293419471553, 'samples': 12912128, 'steps': 25218, 'batch_loss/train': 0.7611323604360223}
+12/26/2021 07:03:15 - INFO - codeparrot_training - Step 25219: {'lr': 0.00024792717159820244, 'samples': 12912640, 'steps': 25219, 'batch_loss/train': 0.6436840840615332}
+12/26/2021 07:03:26 - INFO - codeparrot_training - Step 25220: {'lr': 0.0002479114090099301, 'samples': 12913152, 'steps': 25220, 'batch_loss/train': 0.7485068994574249}
+12/26/2021 07:03:38 - INFO - codeparrot_training - Step 25221: {'lr': 0.0002478956464299611, 'samples': 12913664, 'steps': 25221, 'batch_loss/train': 0.8766025807708502}
+12/26/2021 07:03:49 - INFO - codeparrot_training - Step 25222: {'lr': 0.00024787988385835836, 'samples': 12914176, 'steps': 25222, 'batch_loss/train': 0.7820430193096399}
+12/26/2021 07:03:59 - INFO - codeparrot_training - Step 25223: {'lr': 0.0002478641212951844, 'samples': 12914688, 'steps': 25223, 'batch_loss/train': 0.86290272930637}
+12/26/2021 07:04:11 - INFO - codeparrot_training - Step 25224: {'lr': 0.0002478483587405018, 'samples': 12915200, 'steps': 25224, 'batch_loss/train': 0.7425351138226688}
+12/26/2021 07:04:22 - INFO - codeparrot_training - Step 25225: {'lr': 0.0002478325961943733, 'samples': 12915712, 'steps': 25225, 'batch_loss/train': 0.7603598572313786}
+12/26/2021 07:04:33 - INFO - codeparrot_training - Step 25226: {'lr': 0.0002478168336568617, 'samples': 12916224, 'steps': 25226, 'batch_loss/train': 0.7730655688792467}
+12/26/2021 07:04:43 - INFO - codeparrot_training - Step 25227: {'lr': 0.0002478010711280295, 'samples': 12916736, 'steps': 25227, 'batch_loss/train': 0.781137814745307}
+12/26/2021 07:04:55 - INFO - codeparrot_training - Step 25228: {'lr': 0.00024778530860793943, 'samples': 12917248, 'steps': 25228, 'batch_loss/train': 0.8078898503445089}
+12/26/2021 07:05:06 - INFO - codeparrot_training - Step 25229: {'lr': 0.000247769546096654, 'samples': 12917760, 'steps': 25229, 'batch_loss/train': 0.7216541916131973}
+12/26/2021 07:05:17 - INFO - codeparrot_training - Step 25230: {'lr': 0.00024775378359423607, 'samples': 12918272, 'steps': 25230, 'batch_loss/train': 0.8242227379232645}
+12/26/2021 07:05:30 - INFO - codeparrot_training - Step 25231: {'lr': 0.0002477380211007484, 'samples': 12918784, 'steps': 25231, 'batch_loss/train': 0.676161892246455}
+12/26/2021 07:05:40 - INFO - codeparrot_training - Step 25232: {'lr': 0.0002477222586162533, 'samples': 12919296, 'steps': 25232, 'batch_loss/train': 0.49211621849099174}
+12/26/2021 07:05:51 - INFO - codeparrot_training - Step 25233: {'lr': 0.00024770649614081375, 'samples': 12919808, 'steps': 25233, 'batch_loss/train': 0.7874887166544795}
+12/26/2021 07:06:03 - INFO - codeparrot_training - Step 25234: {'lr': 0.00024769073367449226, 'samples': 12920320, 'steps': 25234, 'batch_loss/train': 0.8523913593962789}
+12/26/2021 07:06:14 - INFO - codeparrot_training - Step 25235: {'lr': 0.00024767497121735163, 'samples': 12920832, 'steps': 25235, 'batch_loss/train': 0.7446892810985446}
+12/26/2021 07:06:25 - INFO - codeparrot_training - Step 25236: {'lr': 0.0002476592087694543, 'samples': 12921344, 'steps': 25236, 'batch_loss/train': 0.7439605798572302}
+12/26/2021 07:06:35 - INFO - codeparrot_training - Step 25237: {'lr': 0.00024764344633086314, 'samples': 12921856, 'steps': 25237, 'batch_loss/train': 0.8222996983677149}
+12/26/2021 07:06:47 - INFO - codeparrot_training - Step 25238: {'lr': 0.00024762768390164077, 'samples': 12922368, 'steps': 25238, 'batch_loss/train': 0.7443396970629692}
+12/26/2021 07:06:58 - INFO - codeparrot_training - Step 25239: {'lr': 0.0002476119214818498, 'samples': 12922880, 'steps': 25239, 'batch_loss/train': 0.8450577547773719}
+12/26/2021 07:07:09 - INFO - codeparrot_training - Step 25240: {'lr': 0.0002475961590715529, 'samples': 12923392, 'steps': 25240, 'batch_loss/train': 0.7906779842451215}
+12/26/2021 07:07:22 - INFO - codeparrot_training - Step 25241: {'lr': 0.00024758039667081285, 'samples': 12923904, 'steps': 25241, 'batch_loss/train': 0.7624347559176385}
+12/26/2021 07:07:32 - INFO - codeparrot_training - Step 25242: {'lr': 0.0002475646342796923, 'samples': 12924416, 'steps': 25242, 'batch_loss/train': 0.9943456850014627}
+12/26/2021 07:07:43 - INFO - codeparrot_training - Step 25243: {'lr': 0.0002475488718982536, 'samples': 12924928, 'steps': 25243, 'batch_loss/train': 0.7388753894483671}
+12/26/2021 07:07:55 - INFO - codeparrot_training - Step 25244: {'lr': 0.0002475331095265598, 'samples': 12925440, 'steps': 25244, 'batch_loss/train': 0.8480263967067003}
+12/26/2021 07:08:06 - INFO - codeparrot_training - Step 25245: {'lr': 0.00024751734716467347, 'samples': 12925952, 'steps': 25245, 'batch_loss/train': 0.7194511773996055}
+12/26/2021 07:08:16 - INFO - codeparrot_training - Step 25246: {'lr': 0.00024750158481265714, 'samples': 12926464, 'steps': 25246, 'batch_loss/train': 0.7120249404106289}
+12/26/2021 07:08:27 - INFO - codeparrot_training - Step 25247: {'lr': 0.00024748582247057356, 'samples': 12926976, 'steps': 25247, 'batch_loss/train': 0.754669664427638}
+12/26/2021 07:08:40 - INFO - codeparrot_training - Step 25248: {'lr': 0.00024747006013848546, 'samples': 12927488, 'steps': 25248, 'batch_loss/train': 0.7864838857203722}
+12/26/2021 07:08:51 - INFO - codeparrot_training - Step 25249: {'lr': 0.00024745429781645545, 'samples': 12928000, 'steps': 25249, 'batch_loss/train': 0.7990676984190941}
+12/26/2021 07:09:01 - INFO - codeparrot_training - Step 25250: {'lr': 0.0002474385355045462, 'samples': 12928512, 'steps': 25250, 'batch_loss/train': 0.677190980874002}
+12/26/2021 07:09:14 - INFO - codeparrot_training - Step 25251: {'lr': 0.00024742277320282026, 'samples': 12929024, 'steps': 25251, 'batch_loss/train': 0.777458910655696}
+12/26/2021 07:09:24 - INFO - codeparrot_training - Step 25252: {'lr': 0.0002474070109113405, 'samples': 12929536, 'steps': 25252, 'batch_loss/train': 0.7513727061450481}
+12/26/2021 07:09:35 - INFO - codeparrot_training - Step 25253: {'lr': 0.0002473912486301695, 'samples': 12930048, 'steps': 25253, 'batch_loss/train': 0.4703896758146584}
+12/26/2021 07:09:47 - INFO - codeparrot_training - Step 25254: {'lr': 0.00024737548635936975, 'samples': 12930560, 'steps': 25254, 'batch_loss/train': 0.7433117341715842}
+12/26/2021 07:09:58 - INFO - codeparrot_training - Step 25255: {'lr': 0.0002473597240990043, 'samples': 12931072, 'steps': 25255, 'batch_loss/train': 0.7909235283732414}
+12/26/2021 07:10:08 - INFO - codeparrot_training - Step 25256: {'lr': 0.00024734396184913546, 'samples': 12931584, 'steps': 25256, 'batch_loss/train': 0.7707237731665373}
+12/26/2021 07:10:19 - INFO - codeparrot_training - Step 25257: {'lr': 0.00024732819960982606, 'samples': 12932096, 'steps': 25257, 'batch_loss/train': 0.824686860665679}
+12/26/2021 07:10:31 - INFO - codeparrot_training - Step 25258: {'lr': 0.00024731243738113865, 'samples': 12932608, 'steps': 25258, 'batch_loss/train': 0.7180286394432187}
+12/26/2021 07:10:42 - INFO - codeparrot_training - Step 25259: {'lr': 0.00024729667516313605, 'samples': 12933120, 'steps': 25259, 'batch_loss/train': 0.8182902811095119}
+12/26/2021 07:10:52 - INFO - codeparrot_training - Step 25260: {'lr': 0.0002472809129558809, 'samples': 12933632, 'steps': 25260, 'batch_loss/train': 0.6827348773367703}
+12/26/2021 07:11:05 - INFO - codeparrot_training - Step 25261: {'lr': 0.0002472651507594358, 'samples': 12934144, 'steps': 25261, 'batch_loss/train': 0.8835940267890692}
+12/26/2021 07:11:16 - INFO - codeparrot_training - Step 25262: {'lr': 0.0002472493885738633, 'samples': 12934656, 'steps': 25262, 'batch_loss/train': 0.8211430353112519}
+12/26/2021 07:11:26 - INFO - codeparrot_training - Step 25263: {'lr': 0.0002472336263992263, 'samples': 12935168, 'steps': 25263, 'batch_loss/train': 0.8292129868641496}
+12/26/2021 07:11:38 - INFO - codeparrot_training - Step 25264: {'lr': 0.0002472178642355874, 'samples': 12935680, 'steps': 25264, 'batch_loss/train': 0.7430075723677874}
+12/26/2021 07:11:49 - INFO - codeparrot_training - Step 25265: {'lr': 0.0002472021020830091, 'samples': 12936192, 'steps': 25265, 'batch_loss/train': 0.7389349334407598}
+12/26/2021 07:11:59 - INFO - codeparrot_training - Step 25266: {'lr': 0.00024718633994155417, 'samples': 12936704, 'steps': 25266, 'batch_loss/train': 0.7853728900663555}
+12/26/2021 07:12:10 - INFO - codeparrot_training - Step 25267: {'lr': 0.0002471705778112854, 'samples': 12937216, 'steps': 25267, 'batch_loss/train': 0.6541175544261932}
+12/26/2021 07:12:22 - INFO - codeparrot_training - Step 25268: {'lr': 0.0002471548156922654, 'samples': 12937728, 'steps': 25268, 'batch_loss/train': 0.8168372639920563}
+12/26/2021 07:12:33 - INFO - codeparrot_training - Step 25269: {'lr': 0.00024713905358455656, 'samples': 12938240, 'steps': 25269, 'batch_loss/train': 0.8073833100497723}
+12/26/2021 07:12:43 - INFO - codeparrot_training - Step 25270: {'lr': 0.00024712329148822193, 'samples': 12938752, 'steps': 25270, 'batch_loss/train': 0.8600629041902721}
+12/26/2021 07:12:56 - INFO - codeparrot_training - Step 25271: {'lr': 0.00024710752940332396, 'samples': 12939264, 'steps': 25271, 'batch_loss/train': 0.8163968287408352}
+12/26/2021 07:13:07 - INFO - codeparrot_training - Step 25272: {'lr': 0.0002470917673299254, 'samples': 12939776, 'steps': 25272, 'batch_loss/train': 0.7171615394763649}
+12/26/2021 07:13:17 - INFO - codeparrot_training - Step 25273: {'lr': 0.00024707600526808874, 'samples': 12940288, 'steps': 25273, 'batch_loss/train': 0.8787297820672393}
+12/26/2021 07:13:30 - INFO - codeparrot_training - Step 25274: {'lr': 0.00024706024321787695, 'samples': 12940800, 'steps': 25274, 'batch_loss/train': 0.7769947946071625}
+12/26/2021 07:13:40 - INFO - codeparrot_training - Step 25275: {'lr': 0.0002470444811793524, 'samples': 12941312, 'steps': 25275, 'batch_loss/train': 0.7589322440326214}
+12/26/2021 07:13:51 - INFO - codeparrot_training - Step 25276: {'lr': 0.000247028719152578, 'samples': 12941824, 'steps': 25276, 'batch_loss/train': 0.7731988416053355}
+12/26/2021 07:14:04 - INFO - codeparrot_training - Step 25277: {'lr': 0.00024701295713761614, 'samples': 12942336, 'steps': 25277, 'batch_loss/train': 0.7145546902902424}
+12/26/2021 07:14:14 - INFO - codeparrot_training - Step 25278: {'lr': 0.00024699719513452973, 'samples': 12942848, 'steps': 25278, 'batch_loss/train': 0.9208776755258441}
+12/26/2021 07:14:25 - INFO - codeparrot_training - Step 25279: {'lr': 0.00024698143314338136, 'samples': 12943360, 'steps': 25279, 'batch_loss/train': 0.8304054532200098}
+12/26/2021 07:14:36 - INFO - codeparrot_training - Step 25280: {'lr': 0.00024696567116423354, 'samples': 12943872, 'steps': 25280, 'batch_loss/train': 0.7006958639249206}
+12/26/2021 07:14:48 - INFO - codeparrot_training - Step 25281: {'lr': 0.0002469499091971492, 'samples': 12944384, 'steps': 25281, 'batch_loss/train': 0.8345130595844239}
+12/26/2021 07:14:58 - INFO - codeparrot_training - Step 25282: {'lr': 0.00024693414724219086, 'samples': 12944896, 'steps': 25282, 'batch_loss/train': 0.7493009883910418}
+12/26/2021 07:15:09 - INFO - codeparrot_training - Step 25283: {'lr': 0.00024691838529942116, 'samples': 12945408, 'steps': 25283, 'batch_loss/train': 0.7530277851037681}
+12/26/2021 07:15:21 - INFO - codeparrot_training - Step 25284: {'lr': 0.00024690262336890274, 'samples': 12945920, 'steps': 25284, 'batch_loss/train': 0.8222754523158073}
+12/26/2021 07:15:32 - INFO - codeparrot_training - Step 25285: {'lr': 0.0002468868614506984, 'samples': 12946432, 'steps': 25285, 'batch_loss/train': 0.6642490981612355}
+12/26/2021 07:15:42 - INFO - codeparrot_training - Step 25286: {'lr': 0.00024687109954487073, 'samples': 12946944, 'steps': 25286, 'batch_loss/train': 1.4412244409322739}
+12/26/2021 07:15:55 - INFO - codeparrot_training - Step 25287: {'lr': 0.0002468553376514824, 'samples': 12947456, 'steps': 25287, 'batch_loss/train': 1.214528679382056}
+12/26/2021 07:16:06 - INFO - codeparrot_training - Step 25288: {'lr': 0.00024683957577059597, 'samples': 12947968, 'steps': 25288, 'batch_loss/train': 0.8030426502227783}
+12/26/2021 07:16:17 - INFO - codeparrot_training - Step 25289: {'lr': 0.0002468238139022742, 'samples': 12948480, 'steps': 25289, 'batch_loss/train': 0.5838584009325132}
+12/26/2021 07:16:27 - INFO - codeparrot_training - Step 25290: {'lr': 0.0002468080520465799, 'samples': 12948992, 'steps': 25290, 'batch_loss/train': 0.8264772808179259}
+12/26/2021 07:16:40 - INFO - codeparrot_training - Step 25291: {'lr': 0.00024679229020357533, 'samples': 12949504, 'steps': 25291, 'batch_loss/train': 0.7577890955726616}
+12/26/2021 07:16:51 - INFO - codeparrot_training - Step 25292: {'lr': 0.0002467765283733236, 'samples': 12950016, 'steps': 25292, 'batch_loss/train': 0.829571285052225}
+12/26/2021 07:17:01 - INFO - codeparrot_training - Step 25293: {'lr': 0.00024676076655588713, 'samples': 12950528, 'steps': 25293, 'batch_loss/train': 0.935324274469167}
+12/26/2021 07:17:13 - INFO - codeparrot_training - Step 25294: {'lr': 0.0002467450047513286, 'samples': 12951040, 'steps': 25294, 'batch_loss/train': 0.7920938823372126}
+12/26/2021 07:17:24 - INFO - codeparrot_training - Step 25295: {'lr': 0.0002467292429597107, 'samples': 12951552, 'steps': 25295, 'batch_loss/train': 0.7854979118565097}
+12/26/2021 07:17:34 - INFO - codeparrot_training - Step 25296: {'lr': 0.0002467134811810961, 'samples': 12952064, 'steps': 25296, 'batch_loss/train': 0.822670885361731}
+12/26/2021 07:17:46 - INFO - codeparrot_training - Step 25297: {'lr': 0.0002466977194155475, 'samples': 12952576, 'steps': 25297, 'batch_loss/train': 0.770421254914254}
+12/26/2021 07:17:57 - INFO - codeparrot_training - Step 25298: {'lr': 0.0002466819576631275, 'samples': 12953088, 'steps': 25298, 'batch_loss/train': 0.7045935597270727}
+12/26/2021 07:18:08 - INFO - codeparrot_training - Step 25299: {'lr': 0.0002466661959238987, 'samples': 12953600, 'steps': 25299, 'batch_loss/train': 0.7364079630933702}
+12/26/2021 07:18:18 - INFO - codeparrot_training - Step 25300: {'lr': 0.000246650434197924, 'samples': 12954112, 'steps': 25300, 'batch_loss/train': 0.6296161271166056}
+12/26/2021 07:18:31 - INFO - codeparrot_training - Step 25301: {'lr': 0.0002466346724852658, 'samples': 12954624, 'steps': 25301, 'batch_loss/train': 0.8039627065882087}
+12/26/2021 07:18:42 - INFO - codeparrot_training - Step 25302: {'lr': 0.00024661891078598693, 'samples': 12955136, 'steps': 25302, 'batch_loss/train': 0.6877897237427533}
+12/26/2021 07:18:52 - INFO - codeparrot_training - Step 25303: {'lr': 0.0002466031491001499, 'samples': 12955648, 'steps': 25303, 'batch_loss/train': 0.7613191846758127}
+12/26/2021 07:19:05 - INFO - codeparrot_training - Step 25304: {'lr': 0.00024658738742781757, 'samples': 12956160, 'steps': 25304, 'batch_loss/train': 0.8813451244495809}
+12/26/2021 07:19:16 - INFO - codeparrot_training - Step 25305: {'lr': 0.00024657162576905247, 'samples': 12956672, 'steps': 25305, 'batch_loss/train': 0.7283953651203774}
+12/26/2021 07:19:26 - INFO - codeparrot_training - Step 25306: {'lr': 0.00024655586412391725, 'samples': 12957184, 'steps': 25306, 'batch_loss/train': 1.2224285434931517}
+12/26/2021 07:19:38 - INFO - codeparrot_training - Step 25307: {'lr': 0.0002465401024924747, 'samples': 12957696, 'steps': 25307, 'batch_loss/train': 0.7949550431221724}
+12/26/2021 07:19:49 - INFO - codeparrot_training - Step 25308: {'lr': 0.0002465243408747873, 'samples': 12958208, 'steps': 25308, 'batch_loss/train': 0.863728646421805}
+12/26/2021 07:19:59 - INFO - codeparrot_training - Step 25309: {'lr': 0.0002465085792709179, 'samples': 12958720, 'steps': 25309, 'batch_loss/train': 0.9161113444715738}
+12/26/2021 07:20:10 - INFO - codeparrot_training - Step 25310: {'lr': 0.00024649281768092895, 'samples': 12959232, 'steps': 25310, 'batch_loss/train': 0.8621358126401901}
+12/26/2021 07:20:23 - INFO - codeparrot_training - Step 25311: {'lr': 0.0002464770561048833, 'samples': 12959744, 'steps': 25311, 'batch_loss/train': 0.7718858229927719}
+12/26/2021 07:20:33 - INFO - codeparrot_training - Step 25312: {'lr': 0.00024646129454284363, 'samples': 12960256, 'steps': 25312, 'batch_loss/train': 0.8593959324061871}
+12/26/2021 07:20:44 - INFO - codeparrot_training - Step 25313: {'lr': 0.00024644553299487244, 'samples': 12960768, 'steps': 25313, 'batch_loss/train': 0.8555699987336993}
+12/26/2021 07:20:56 - INFO - codeparrot_training - Step 25314: {'lr': 0.0002464297714610324, 'samples': 12961280, 'steps': 25314, 'batch_loss/train': 0.8807461345568299}
+12/26/2021 07:21:07 - INFO - codeparrot_training - Step 25315: {'lr': 0.0002464140099413863, 'samples': 12961792, 'steps': 25315, 'batch_loss/train': 0.8216862613335252}
+12/26/2021 07:21:17 - INFO - codeparrot_training - Step 25316: {'lr': 0.0002463982484359968, 'samples': 12962304, 'steps': 25316, 'batch_loss/train': 0.7079996354877949}
+12/26/2021 07:21:30 - INFO - codeparrot_training - Step 25317: {'lr': 0.0002463824869449264, 'samples': 12962816, 'steps': 25317, 'batch_loss/train': 0.7702031619846821}
+12/26/2021 07:21:41 - INFO - codeparrot_training - Step 25318: {'lr': 0.0002463667254682379, 'samples': 12963328, 'steps': 25318, 'batch_loss/train': 0.7269221104215831}
+12/26/2021 07:21:51 - INFO - codeparrot_training - Step 25319: {'lr': 0.000246350964005994, 'samples': 12963840, 'steps': 25319, 'batch_loss/train': 0.8442498408257961}
+12/26/2021 07:22:02 - INFO - codeparrot_training - Step 25320: {'lr': 0.0002463352025582572, 'samples': 12964352, 'steps': 25320, 'batch_loss/train': 0.7899448396638036}
+12/26/2021 07:22:14 - INFO - codeparrot_training - Step 25321: {'lr': 0.00024631944112509023, 'samples': 12964864, 'steps': 25321, 'batch_loss/train': 0.7599393334239721}
+12/26/2021 07:22:25 - INFO - codeparrot_training - Step 25322: {'lr': 0.00024630367970655587, 'samples': 12965376, 'steps': 25322, 'batch_loss/train': 0.8126710345968604}
+12/26/2021 07:22:35 - INFO - codeparrot_training - Step 25323: {'lr': 0.0002462879183027167, 'samples': 12965888, 'steps': 25323, 'batch_loss/train': 0.7138885674066842}
+12/26/2021 07:22:48 - INFO - codeparrot_training - Step 25324: {'lr': 0.00024627215691363526, 'samples': 12966400, 'steps': 25324, 'batch_loss/train': 0.7698466056026518}
+12/26/2021 07:22:58 - INFO - codeparrot_training - Step 25325: {'lr': 0.0002462563955393743, 'samples': 12966912, 'steps': 25325, 'batch_loss/train': 0.7943705562502146}
+12/26/2021 07:23:09 - INFO - codeparrot_training - Step 25326: {'lr': 0.0002462406341799966, 'samples': 12967424, 'steps': 25326, 'batch_loss/train': 0.7365490397205576}
+12/26/2021 07:23:22 - INFO - codeparrot_training - Step 25327: {'lr': 0.0002462248728355647, 'samples': 12967936, 'steps': 25327, 'batch_loss/train': 0.47858711751177907}
+12/26/2021 07:23:32 - INFO - codeparrot_training - Step 25328: {'lr': 0.0002462091115061412, 'samples': 12968448, 'steps': 25328, 'batch_loss/train': 0.7393220420926809}
+12/26/2021 07:23:43 - INFO - codeparrot_training - Step 25329: {'lr': 0.00024619335019178894, 'samples': 12968960, 'steps': 25329, 'batch_loss/train': 0.7532531747128814}
+12/26/2021 07:23:55 - INFO - codeparrot_training - Step 25330: {'lr': 0.00024617758889257043, 'samples': 12969472, 'steps': 25330, 'batch_loss/train': 0.8638168655452318}
+12/26/2021 07:24:06 - INFO - codeparrot_training - Step 25331: {'lr': 0.0002461618276085485, 'samples': 12969984, 'steps': 25331, 'batch_loss/train': 0.799006418325007}
+12/26/2021 07:24:16 - INFO - codeparrot_training - Step 25332: {'lr': 0.0002461460663397855, 'samples': 12970496, 'steps': 25332, 'batch_loss/train': 0.7566322372294962}
+12/26/2021 07:24:27 - INFO - codeparrot_training - Step 25333: {'lr': 0.00024613030508634443, 'samples': 12971008, 'steps': 25333, 'batch_loss/train': 0.6381411827169359}
+12/26/2021 07:24:39 - INFO - codeparrot_training - Step 25334: {'lr': 0.0002461145438482877, 'samples': 12971520, 'steps': 25334, 'batch_loss/train': 0.7141017545945942}
+12/26/2021 07:24:50 - INFO - codeparrot_training - Step 25335: {'lr': 0.00024609878262567817, 'samples': 12972032, 'steps': 25335, 'batch_loss/train': 0.7904504425823689}
+12/26/2021 07:25:00 - INFO - codeparrot_training - Step 25336: {'lr': 0.0002460830214185783, 'samples': 12972544, 'steps': 25336, 'batch_loss/train': 0.7841252505313605}
+12/26/2021 07:25:12 - INFO - codeparrot_training - Step 25337: {'lr': 0.00024606726022705094, 'samples': 12973056, 'steps': 25337, 'batch_loss/train': 0.798270296305418}
+12/26/2021 07:25:23 - INFO - codeparrot_training - Step 25338: {'lr': 0.0002460514990511587, 'samples': 12973568, 'steps': 25338, 'batch_loss/train': 0.7787195895798504}
+12/26/2021 07:25:34 - INFO - codeparrot_training - Step 25339: {'lr': 0.0002460357378909641, 'samples': 12974080, 'steps': 25339, 'batch_loss/train': 0.8543125931173563}
+12/26/2021 07:25:46 - INFO - codeparrot_training - Step 25340: {'lr': 0.00024601997674652995, 'samples': 12974592, 'steps': 25340, 'batch_loss/train': 0.7798041058704257}
+12/26/2021 07:25:57 - INFO - codeparrot_training - Step 25341: {'lr': 0.00024600421561791894, 'samples': 12975104, 'steps': 25341, 'batch_loss/train': 0.8071639854460955}
+12/26/2021 07:26:08 - INFO - codeparrot_training - Step 25342: {'lr': 0.0002459884545051936, 'samples': 12975616, 'steps': 25342, 'batch_loss/train': 0.6946186758577824}
+12/26/2021 07:26:18 - INFO - codeparrot_training - Step 25343: {'lr': 0.00024597269340841663, 'samples': 12976128, 'steps': 25343, 'batch_loss/train': 0.7040374320931733}
+12/26/2021 07:26:30 - INFO - codeparrot_training - Step 25344: {'lr': 0.0002459569323276507, 'samples': 12976640, 'steps': 25344, 'batch_loss/train': 0.6986176986247301}
+12/26/2021 07:26:41 - INFO - codeparrot_training - Step 25345: {'lr': 0.00024594117126295863, 'samples': 12977152, 'steps': 25345, 'batch_loss/train': 0.8554142005741596}
+12/26/2021 07:26:52 - INFO - codeparrot_training - Step 25346: {'lr': 0.0002459254102144028, 'samples': 12977664, 'steps': 25346, 'batch_loss/train': 0.815017644315958}
+12/26/2021 07:27:04 - INFO - codeparrot_training - Step 25347: {'lr': 0.00024590964918204595, 'samples': 12978176, 'steps': 25347, 'batch_loss/train': 0.7652826309204102}
+12/26/2021 07:27:15 - INFO - codeparrot_training - Step 25348: {'lr': 0.0002458938881659509, 'samples': 12978688, 'steps': 25348, 'batch_loss/train': 0.782454433850944}
+12/26/2021 07:27:26 - INFO - codeparrot_training - Step 25349: {'lr': 0.00024587812716618014, 'samples': 12979200, 'steps': 25349, 'batch_loss/train': 0.7266142340376973}
+12/26/2021 07:27:38 - INFO - codeparrot_training - Step 25350: {'lr': 0.0002458623661827965, 'samples': 12979712, 'steps': 25350, 'batch_loss/train': 0.7426616088487208}
+12/26/2021 07:27:49 - INFO - codeparrot_training - Step 25351: {'lr': 0.0002458466052158624, 'samples': 12980224, 'steps': 25351, 'batch_loss/train': 0.7766115609556437}
+12/26/2021 07:27:59 - INFO - codeparrot_training - Step 25352: {'lr': 0.00024583084426544065, 'samples': 12980736, 'steps': 25352, 'batch_loss/train': 0.8152451182249933}
+12/26/2021 07:28:10 - INFO - codeparrot_training - Step 25353: {'lr': 0.00024581508333159393, 'samples': 12981248, 'steps': 25353, 'batch_loss/train': 0.8470707144588232}
+12/26/2021 07:28:22 - INFO - codeparrot_training - Step 25354: {'lr': 0.0002457993224143848, 'samples': 12981760, 'steps': 25354, 'batch_loss/train': 0.7705894717946649}
+12/26/2021 07:28:32 - INFO - codeparrot_training - Step 25355: {'lr': 0.00024578356151387605, 'samples': 12982272, 'steps': 25355, 'batch_loss/train': 0.7544023925438523}
+12/26/2021 07:28:43 - INFO - codeparrot_training - Step 25356: {'lr': 0.0002457678006301302, 'samples': 12982784, 'steps': 25356, 'batch_loss/train': 0.781107165850699}
+12/26/2021 07:28:56 - INFO - codeparrot_training - Step 25357: {'lr': 0.00024575203976321005, 'samples': 12983296, 'steps': 25357, 'batch_loss/train': 0.9001335194334388}
+12/26/2021 07:29:07 - INFO - codeparrot_training - Step 25358: {'lr': 0.0002457362789131781, 'samples': 12983808, 'steps': 25358, 'batch_loss/train': 0.5494246690068394}
+12/26/2021 07:29:17 - INFO - codeparrot_training - Step 25359: {'lr': 0.00024572051808009716, 'samples': 12984320, 'steps': 25359, 'batch_loss/train': 0.7335309130139649}
+12/26/2021 07:29:30 - INFO - codeparrot_training - Step 25360: {'lr': 0.0002457047572640298, 'samples': 12984832, 'steps': 25360, 'batch_loss/train': 0.860393347509671}
+12/26/2021 07:29:40 - INFO - codeparrot_training - Step 25361: {'lr': 0.0002456889964650388, 'samples': 12985344, 'steps': 25361, 'batch_loss/train': 0.8578019645065069}
+12/26/2021 07:29:51 - INFO - codeparrot_training - Step 25362: {'lr': 0.0002456732356831866, 'samples': 12985856, 'steps': 25362, 'batch_loss/train': 0.7850818978622556}
+12/26/2021 07:30:02 - INFO - codeparrot_training - Step 25363: {'lr': 0.00024565747491853596, 'samples': 12986368, 'steps': 25363, 'batch_loss/train': 0.8587602023035288}
+12/26/2021 07:30:14 - INFO - codeparrot_training - Step 25364: {'lr': 0.00024564171417114963, 'samples': 12986880, 'steps': 25364, 'batch_loss/train': 0.9332420630380511}
+12/26/2021 07:30:25 - INFO - codeparrot_training - Step 25365: {'lr': 0.00024562595344109015, 'samples': 12987392, 'steps': 25365, 'batch_loss/train': 1.3789934376254678}
+12/26/2021 07:30:36 - INFO - codeparrot_training - Step 25366: {'lr': 0.00024561019272842025, 'samples': 12987904, 'steps': 25366, 'batch_loss/train': 0.8737056590616703}
+12/26/2021 07:30:48 - INFO - codeparrot_training - Step 25367: {'lr': 0.00024559443203320264, 'samples': 12988416, 'steps': 25367, 'batch_loss/train': 0.7742352578788996}
+12/26/2021 07:30:58 - INFO - codeparrot_training - Step 25368: {'lr': 0.0002455786713554999, 'samples': 12988928, 'steps': 25368, 'batch_loss/train': 0.8010924705304205}
+12/26/2021 07:31:09 - INFO - codeparrot_training - Step 25369: {'lr': 0.00024556291069537453, 'samples': 12989440, 'steps': 25369, 'batch_loss/train': 0.7124667521566153}
+12/26/2021 07:31:21 - INFO - codeparrot_training - Step 25370: {'lr': 0.0002455471500528894, 'samples': 12989952, 'steps': 25370, 'batch_loss/train': 0.871288635302335}
+12/26/2021 07:31:32 - INFO - codeparrot_training - Step 25371: {'lr': 0.0002455313894281073, 'samples': 12990464, 'steps': 25371, 'batch_loss/train': 0.8234588205814362}
+12/26/2021 07:31:42 - INFO - codeparrot_training - Step 25372: {'lr': 0.00024551562882109055, 'samples': 12990976, 'steps': 25372, 'batch_loss/train': 0.7400173996575177}
+12/26/2021 07:31:53 - INFO - codeparrot_training - Step 25373: {'lr': 0.00024549986823190194, 'samples': 12991488, 'steps': 25373, 'batch_loss/train': 0.693841791478917}
+12/26/2021 07:32:05 - INFO - codeparrot_training - Step 25374: {'lr': 0.00024548410766060425, 'samples': 12992000, 'steps': 25374, 'batch_loss/train': 0.5898843656759709}
+12/26/2021 07:32:16 - INFO - codeparrot_training - Step 25375: {'lr': 0.00024546834710725997, 'samples': 12992512, 'steps': 25375, 'batch_loss/train': 0.7365167494863272}
+12/26/2021 07:32:26 - INFO - codeparrot_training - Step 25376: {'lr': 0.00024545258657193193, 'samples': 12993024, 'steps': 25376, 'batch_loss/train': 0.8548072943231091}
+12/26/2021 07:32:39 - INFO - codeparrot_training - Step 25377: {'lr': 0.0002454368260546826, 'samples': 12993536, 'steps': 25377, 'batch_loss/train': 0.8361872429959476}
+12/26/2021 07:32:50 - INFO - codeparrot_training - Step 25378: {'lr': 0.00024542106555557476, 'samples': 12994048, 'steps': 25378, 'batch_loss/train': 0.8396130031906068}
+12/26/2021 07:33:00 - INFO - codeparrot_training - Step 25379: {'lr': 0.00024540530507467106, 'samples': 12994560, 'steps': 25379, 'batch_loss/train': 0.7665313724428415}
+12/26/2021 07:33:12 - INFO - codeparrot_training - Step 25380: {'lr': 0.000245389544612034, 'samples': 12995072, 'steps': 25380, 'batch_loss/train': 0.7999862506985664}
+12/26/2021 07:33:23 - INFO - codeparrot_training - Step 25381: {'lr': 0.00024537378416772653, 'samples': 12995584, 'steps': 25381, 'batch_loss/train': 0.871319143101573}
+12/26/2021 07:33:34 - INFO - codeparrot_training - Step 25382: {'lr': 0.0002453580237418111, 'samples': 12996096, 'steps': 25382, 'batch_loss/train': 0.8008547341451049}
+12/26/2021 07:33:44 - INFO - codeparrot_training - Step 25383: {'lr': 0.00024534226333435043, 'samples': 12996608, 'steps': 25383, 'batch_loss/train': 0.7540412547532469}
+12/26/2021 07:33:56 - INFO - codeparrot_training - Step 25384: {'lr': 0.0002453265029454071, 'samples': 12997120, 'steps': 25384, 'batch_loss/train': 0.9050165433436632}
+12/26/2021 07:34:07 - INFO - codeparrot_training - Step 25385: {'lr': 0.0002453107425750438, 'samples': 12997632, 'steps': 25385, 'batch_loss/train': 0.7634796863421798}
+12/26/2021 07:34:17 - INFO - codeparrot_training - Step 25386: {'lr': 0.0002452949822233233, 'samples': 12998144, 'steps': 25386, 'batch_loss/train': 0.7556217219680548}
+12/26/2021 07:34:30 - INFO - codeparrot_training - Step 25387: {'lr': 0.00024527922189030826, 'samples': 12998656, 'steps': 25387, 'batch_loss/train': 0.6770554697141051}
+12/26/2021 07:34:41 - INFO - codeparrot_training - Step 25388: {'lr': 0.0002452634615760611, 'samples': 12999168, 'steps': 25388, 'batch_loss/train': 0.8456044206395745}
+12/26/2021 07:34:52 - INFO - codeparrot_training - Step 25389: {'lr': 0.00024524770128064464, 'samples': 12999680, 'steps': 25389, 'batch_loss/train': 0.7407592302188277}
+12/26/2021 07:35:04 - INFO - codeparrot_training - Step 25390: {'lr': 0.0002452319410041216, 'samples': 13000192, 'steps': 25390, 'batch_loss/train': 0.6251990406308323}
+12/26/2021 07:35:14 - INFO - codeparrot_training - Step 25391: {'lr': 0.0002452161807465545, 'samples': 13000704, 'steps': 25391, 'batch_loss/train': 0.7039190374780446}
+12/26/2021 07:35:25 - INFO - codeparrot_training - Step 25392: {'lr': 0.000245200420508006, 'samples': 13001216, 'steps': 25392, 'batch_loss/train': 0.7306415108032525}
+12/26/2021 07:35:36 - INFO - codeparrot_training - Step 25393: {'lr': 0.000245184660288539, 'samples': 13001728, 'steps': 25393, 'batch_loss/train': 0.9142991360276937}
+12/26/2021 07:35:48 - INFO - codeparrot_training - Step 25394: {'lr': 0.0002451689000882158, 'samples': 13002240, 'steps': 25394, 'batch_loss/train': 0.6952954871812835}
+12/26/2021 07:35:59 - INFO - codeparrot_training - Step 25395: {'lr': 0.00024515313990709927, 'samples': 13002752, 'steps': 25395, 'batch_loss/train': 0.6990523962303996}
+12/26/2021 07:36:09 - INFO - codeparrot_training - Step 25396: {'lr': 0.0002451373797452521, 'samples': 13003264, 'steps': 25396, 'batch_loss/train': 1.174984791316092}
+12/26/2021 07:36:22 - INFO - codeparrot_training - Step 25397: {'lr': 0.0002451216196027369, 'samples': 13003776, 'steps': 25397, 'batch_loss/train': 0.8036088240332901}
+12/26/2021 07:36:33 - INFO - codeparrot_training - Step 25398: {'lr': 0.0002451058594796162, 'samples': 13004288, 'steps': 25398, 'batch_loss/train': 0.8061168324202299}
+12/26/2021 07:36:43 - INFO - codeparrot_training - Step 25399: {'lr': 0.00024509009937595267, 'samples': 13004800, 'steps': 25399, 'batch_loss/train': 0.822550181299448}
+12/26/2021 07:36:56 - INFO - codeparrot_training - Step 25400: {'lr': 0.00024507433929180924, 'samples': 13005312, 'steps': 25400, 'batch_loss/train': 0.7763899373821914}
+12/26/2021 07:37:06 - INFO - codeparrot_training - Step 25401: {'lr': 0.0002450585792272483, 'samples': 13005824, 'steps': 25401, 'batch_loss/train': 1.0820992450753693}
+12/26/2021 07:37:17 - INFO - codeparrot_training - Step 25402: {'lr': 0.0002450428191823325, 'samples': 13006336, 'steps': 25402, 'batch_loss/train': 0.8203055632766336}
+12/26/2021 07:37:28 - INFO - codeparrot_training - Step 25403: {'lr': 0.0002450270591571247, 'samples': 13006848, 'steps': 25403, 'batch_loss/train': 0.8230154486373067}
+12/26/2021 07:37:41 - INFO - codeparrot_training - Step 25404: {'lr': 0.0002450112991516875, 'samples': 13007360, 'steps': 25404, 'batch_loss/train': 0.7468617481645197}
+12/26/2021 07:37:51 - INFO - codeparrot_training - Step 25405: {'lr': 0.00024499553916608333, 'samples': 13007872, 'steps': 25405, 'batch_loss/train': 0.8325383504852653}
+12/26/2021 07:38:02 - INFO - codeparrot_training - Step 25406: {'lr': 0.000244979779200375, 'samples': 13008384, 'steps': 25406, 'batch_loss/train': 0.8488864479586482}
+12/26/2021 07:38:14 - INFO - codeparrot_training - Step 25407: {'lr': 0.00024496401925462526, 'samples': 13008896, 'steps': 25407, 'batch_loss/train': 0.7771522486582398}
+12/26/2021 07:38:25 - INFO - codeparrot_training - Step 25408: {'lr': 0.00024494825932889667, 'samples': 13009408, 'steps': 25408, 'batch_loss/train': 0.7495356234721839}
+12/26/2021 07:38:35 - INFO - codeparrot_training - Step 25409: {'lr': 0.0002449324994232519, 'samples': 13009920, 'steps': 25409, 'batch_loss/train': 0.8241537779103965}
+12/26/2021 07:38:46 - INFO - codeparrot_training - Step 25410: {'lr': 0.00024491673953775354, 'samples': 13010432, 'steps': 25410, 'batch_loss/train': 1.1986556188203394}
+12/26/2021 07:38:58 - INFO - codeparrot_training - Step 25411: {'lr': 0.00024490097967246434, 'samples': 13010944, 'steps': 25411, 'batch_loss/train': 0.8233243371360004}
+12/26/2021 07:39:09 - INFO - codeparrot_training - Step 25412: {'lr': 0.00024488521982744696, 'samples': 13011456, 'steps': 25412, 'batch_loss/train': 0.6863092781277373}
+12/26/2021 07:39:19 - INFO - codeparrot_training - Step 25413: {'lr': 0.000244869460002764, 'samples': 13011968, 'steps': 25413, 'batch_loss/train': 0.8579651885665953}
+12/26/2021 07:39:31 - INFO - codeparrot_training - Step 25414: {'lr': 0.00024485370019847804, 'samples': 13012480, 'steps': 25414, 'batch_loss/train': 0.7258054243866354}
+12/26/2021 07:39:42 - INFO - codeparrot_training - Step 25415: {'lr': 0.0002448379404146519, 'samples': 13012992, 'steps': 25415, 'batch_loss/train': 0.8513416517525911}
+12/26/2021 07:39:53 - INFO - codeparrot_training - Step 25416: {'lr': 0.00024482218065134816, 'samples': 13013504, 'steps': 25416, 'batch_loss/train': 0.9008687753230333}
+12/26/2021 07:40:05 - INFO - codeparrot_training - Step 25417: {'lr': 0.00024480642090862943, 'samples': 13014016, 'steps': 25417, 'batch_loss/train': 0.7531140232458711}
+12/26/2021 07:40:16 - INFO - codeparrot_training - Step 25418: {'lr': 0.00024479066118655846, 'samples': 13014528, 'steps': 25418, 'batch_loss/train': 0.8334431545808911}
+12/26/2021 07:40:27 - INFO - codeparrot_training - Step 25419: {'lr': 0.00024477490148519786, 'samples': 13015040, 'steps': 25419, 'batch_loss/train': 0.7741233939304948}
+12/26/2021 07:40:39 - INFO - codeparrot_training - Step 25420: {'lr': 0.00024475914180461026, 'samples': 13015552, 'steps': 25420, 'batch_loss/train': 0.8121870262548327}
+12/26/2021 07:40:49 - INFO - codeparrot_training - Step 25421: {'lr': 0.00024474338214485825, 'samples': 13016064, 'steps': 25421, 'batch_loss/train': 0.8603701549582183}
+12/26/2021 07:41:00 - INFO - codeparrot_training - Step 25422: {'lr': 0.0002447276225060047, 'samples': 13016576, 'steps': 25422, 'batch_loss/train': 0.7138027236796916}
+12/26/2021 07:41:10 - INFO - codeparrot_training - Step 25423: {'lr': 0.00024471186288811214, 'samples': 13017088, 'steps': 25423, 'batch_loss/train': 0.7466053767129779}
+12/26/2021 07:41:23 - INFO - codeparrot_training - Step 25424: {'lr': 0.0002446961032912431, 'samples': 13017600, 'steps': 25424, 'batch_loss/train': 0.9878677083179355}
+12/26/2021 07:41:34 - INFO - codeparrot_training - Step 25425: {'lr': 0.00024468034371546033, 'samples': 13018112, 'steps': 25425, 'batch_loss/train': 1.1608747923746705}
+12/26/2021 07:41:44 - INFO - codeparrot_training - Step 25426: {'lr': 0.00024466458416082666, 'samples': 13018624, 'steps': 25426, 'batch_loss/train': 0.761301476508379}
+12/26/2021 07:41:57 - INFO - codeparrot_training - Step 25427: {'lr': 0.0002446488246274045, 'samples': 13019136, 'steps': 25427, 'batch_loss/train': 0.8101494791917503}
+12/26/2021 07:42:08 - INFO - codeparrot_training - Step 25428: {'lr': 0.0002446330651152565, 'samples': 13019648, 'steps': 25428, 'batch_loss/train': 0.712806137278676}
+12/26/2021 07:42:18 - INFO - codeparrot_training - Step 25429: {'lr': 0.0002446173056244456, 'samples': 13020160, 'steps': 25429, 'batch_loss/train': 0.8665047744289041}
+12/26/2021 07:42:31 - INFO - codeparrot_training - Step 25430: {'lr': 0.0002446015461550341, 'samples': 13020672, 'steps': 25430, 'batch_loss/train': 0.7810465442016721}
+12/26/2021 07:42:41 - INFO - codeparrot_training - Step 25431: {'lr': 0.00024458578670708496, 'samples': 13021184, 'steps': 25431, 'batch_loss/train': 0.8129059700295329}
+12/26/2021 07:42:52 - INFO - codeparrot_training - Step 25432: {'lr': 0.00024457002728066053, 'samples': 13021696, 'steps': 25432, 'batch_loss/train': 0.7552985725924373}
+12/26/2021 07:43:04 - INFO - codeparrot_training - Step 25433: {'lr': 0.00024455426787582375, 'samples': 13022208, 'steps': 25433, 'batch_loss/train': 0.6449605815578252}
+12/26/2021 07:43:15 - INFO - codeparrot_training - Step 25434: {'lr': 0.00024453850849263714, 'samples': 13022720, 'steps': 25434, 'batch_loss/train': 0.8152423533611}
+12/26/2021 07:43:26 - INFO - codeparrot_training - Step 25435: {'lr': 0.00024452274913116335, 'samples': 13023232, 'steps': 25435, 'batch_loss/train': 0.7941770548932254}
+12/26/2021 07:43:36 - INFO - codeparrot_training - Step 25436: {'lr': 0.00024450698979146496, 'samples': 13023744, 'steps': 25436, 'batch_loss/train': 0.8146755611523986}
+12/26/2021 07:43:49 - INFO - codeparrot_training - Step 25437: {'lr': 0.0002444912304736047, 'samples': 13024256, 'steps': 25437, 'batch_loss/train': 0.8922433909028769}
+12/26/2021 07:43:59 - INFO - codeparrot_training - Step 25438: {'lr': 0.00024447547117764537, 'samples': 13024768, 'steps': 25438, 'batch_loss/train': 0.7938921856693923}
+12/26/2021 07:44:10 - INFO - codeparrot_training - Step 25439: {'lr': 0.0002444597119036493, 'samples': 13025280, 'steps': 25439, 'batch_loss/train': 0.900546652264893}
+12/26/2021 07:44:22 - INFO - codeparrot_training - Step 25440: {'lr': 0.0002444439526516794, 'samples': 13025792, 'steps': 25440, 'batch_loss/train': 0.7754444655729458}
+12/26/2021 07:44:33 - INFO - codeparrot_training - Step 25441: {'lr': 0.0002444281934217983, 'samples': 13026304, 'steps': 25441, 'batch_loss/train': 0.7852080389857292}
+12/26/2021 07:44:44 - INFO - codeparrot_training - Step 25442: {'lr': 0.0002444124342140686, 'samples': 13026816, 'steps': 25442, 'batch_loss/train': 0.783188997185789}
+12/26/2021 07:44:54 - INFO - codeparrot_training - Step 25443: {'lr': 0.0002443966750285528, 'samples': 13027328, 'steps': 25443, 'batch_loss/train': 1.3487502527423203}
+12/26/2021 07:45:07 - INFO - codeparrot_training - Step 25444: {'lr': 0.00024438091586531386, 'samples': 13027840, 'steps': 25444, 'batch_loss/train': 0.8172194515354931}
+12/26/2021 07:45:17 - INFO - codeparrot_training - Step 25445: {'lr': 0.0002443651567244143, 'samples': 13028352, 'steps': 25445, 'batch_loss/train': 0.7570237312465906}
+12/26/2021 07:45:28 - INFO - codeparrot_training - Step 25446: {'lr': 0.00024434939760591666, 'samples': 13028864, 'steps': 25446, 'batch_loss/train': 0.8262944389134645}
+12/26/2021 07:45:40 - INFO - codeparrot_training - Step 25447: {'lr': 0.00024433363850988364, 'samples': 13029376, 'steps': 25447, 'batch_loss/train': 0.8412288976833224}
+12/26/2021 07:45:51 - INFO - codeparrot_training - Step 25448: {'lr': 0.000244317879436378, 'samples': 13029888, 'steps': 25448, 'batch_loss/train': 0.708147298777476}
+12/26/2021 07:46:01 - INFO - codeparrot_training - Step 25449: {'lr': 0.0002443021203854624, 'samples': 13030400, 'steps': 25449, 'batch_loss/train': 0.7583611465524882}
+12/26/2021 07:46:14 - INFO - codeparrot_training - Step 25450: {'lr': 0.0002442863613571993, 'samples': 13030912, 'steps': 25450, 'batch_loss/train': 0.7533436673693359}
+12/26/2021 07:46:25 - INFO - codeparrot_training - Step 25451: {'lr': 0.00024427060235165146, 'samples': 13031424, 'steps': 25451, 'batch_loss/train': 0.729934114497155}
+12/26/2021 07:46:35 - INFO - codeparrot_training - Step 25452: {'lr': 0.00024425484336888163, 'samples': 13031936, 'steps': 25452, 'batch_loss/train': 0.8378202025778592}
+12/26/2021 07:46:48 - INFO - codeparrot_training - Step 25453: {'lr': 0.00024423908440895237, 'samples': 13032448, 'steps': 25453, 'batch_loss/train': 0.6327522803330794}
+12/26/2021 07:46:59 - INFO - codeparrot_training - Step 25454: {'lr': 0.0002442233254719262, 'samples': 13032960, 'steps': 25454, 'batch_loss/train': 0.6477282037958503}
+12/26/2021 07:47:09 - INFO - codeparrot_training - Step 25455: {'lr': 0.000244207566557866, 'samples': 13033472, 'steps': 25455, 'batch_loss/train': 0.7823877381160855}
+12/26/2021 07:47:20 - INFO - codeparrot_training - Step 25456: {'lr': 0.0002441918076668344, 'samples': 13033984, 'steps': 25456, 'batch_loss/train': 0.7434593327343464}
+12/26/2021 07:47:32 - INFO - codeparrot_training - Step 25457: {'lr': 0.00024417604879889394, 'samples': 13034496, 'steps': 25457, 'batch_loss/train': 0.7816774221137166}
+12/26/2021 07:47:43 - INFO - codeparrot_training - Step 25458: {'lr': 0.0002441602899541072, 'samples': 13035008, 'steps': 25458, 'batch_loss/train': 1.0845389538444579}
+12/26/2021 07:47:53 - INFO - codeparrot_training - Step 25459: {'lr': 0.0002441445311325371, 'samples': 13035520, 'steps': 25459, 'batch_loss/train': 0.9729837044142187}
+12/26/2021 07:48:05 - INFO - codeparrot_training - Step 25460: {'lr': 0.00024412877233424604, 'samples': 13036032, 'steps': 25460, 'batch_loss/train': 0.7763194935396314}
+12/26/2021 07:48:16 - INFO - codeparrot_training - Step 25461: {'lr': 0.00024411301355929685, 'samples': 13036544, 'steps': 25461, 'batch_loss/train': 0.7907136203721166}
+12/26/2021 07:48:27 - INFO - codeparrot_training - Step 25462: {'lr': 0.00024409725480775197, 'samples': 13037056, 'steps': 25462, 'batch_loss/train': 0.8247030223719776}
+12/26/2021 07:48:40 - INFO - codeparrot_training - Step 25463: {'lr': 0.00024408149607967422, 'samples': 13037568, 'steps': 25463, 'batch_loss/train': 0.7964211283251643}
+12/26/2021 07:48:51 - INFO - codeparrot_training - Step 25464: {'lr': 0.00024406573737512628, 'samples': 13038080, 'steps': 25464, 'batch_loss/train': 0.7149649448692799}
+12/26/2021 07:49:01 - INFO - codeparrot_training - Step 25465: {'lr': 0.00024404997869417058, 'samples': 13038592, 'steps': 25465, 'batch_loss/train': 0.5932726527098566}
+12/26/2021 07:49:12 - INFO - codeparrot_training - Step 25466: {'lr': 0.00024403422003687, 'samples': 13039104, 'steps': 25466, 'batch_loss/train': 0.730967917945236}
+12/26/2021 07:49:24 - INFO - codeparrot_training - Step 25467: {'lr': 0.00024401846140328717, 'samples': 13039616, 'steps': 25467, 'batch_loss/train': 0.774234626442194}
+12/26/2021 07:49:35 - INFO - codeparrot_training - Step 25468: {'lr': 0.00024400270279348468, 'samples': 13040128, 'steps': 25468, 'batch_loss/train': 0.7984460713341832}
+12/26/2021 07:49:45 - INFO - codeparrot_training - Step 25469: {'lr': 0.00024398694420752504, 'samples': 13040640, 'steps': 25469, 'batch_loss/train': 0.7120322035625577}
+12/26/2021 07:49:59 - INFO - codeparrot_training - Step 25470: {'lr': 0.00024397118564547114, 'samples': 13041152, 'steps': 25470, 'batch_loss/train': 0.785693955142051}
+12/26/2021 07:50:09 - INFO - codeparrot_training - Step 25471: {'lr': 0.00024395542710738556, 'samples': 13041664, 'steps': 25471, 'batch_loss/train': 1.5290119735873304}
+12/26/2021 07:50:20 - INFO - codeparrot_training - Step 25472: {'lr': 0.00024393966859333086, 'samples': 13042176, 'steps': 25472, 'batch_loss/train': 0.8014979155850597}
+12/26/2021 07:50:30 - INFO - codeparrot_training - Step 25473: {'lr': 0.0002439239101033697, 'samples': 13042688, 'steps': 25473, 'batch_loss/train': 1.46103885024786}
+12/26/2021 07:50:42 - INFO - codeparrot_training - Step 25474: {'lr': 0.00024390815163756487, 'samples': 13043200, 'steps': 25474, 'batch_loss/train': 1.1255553355440497}
+12/26/2021 07:50:53 - INFO - codeparrot_training - Step 25475: {'lr': 0.00024389239319597892, 'samples': 13043712, 'steps': 25475, 'batch_loss/train': 0.7214068797184154}
+12/26/2021 07:51:04 - INFO - codeparrot_training - Step 25476: {'lr': 0.00024387663477867442, 'samples': 13044224, 'steps': 25476, 'batch_loss/train': 0.7377722505480051}
+12/26/2021 07:51:16 - INFO - codeparrot_training - Step 25477: {'lr': 0.00024386087638571423, 'samples': 13044736, 'steps': 25477, 'batch_loss/train': 0.8908010569866747}
+12/26/2021 07:51:26 - INFO - codeparrot_training - Step 25478: {'lr': 0.00024384511801716087, 'samples': 13045248, 'steps': 25478, 'batch_loss/train': 0.7320580258965492}
+12/26/2021 07:51:37 - INFO - codeparrot_training - Step 25479: {'lr': 0.00024382935967307692, 'samples': 13045760, 'steps': 25479, 'batch_loss/train': 0.6817730193724856}
+12/26/2021 07:51:50 - INFO - codeparrot_training - Step 25480: {'lr': 0.00024381360135352509, 'samples': 13046272, 'steps': 25480, 'batch_loss/train': 0.7772970146033913}
+12/26/2021 07:52:01 - INFO - codeparrot_training - Step 25481: {'lr': 0.00024379784305856813, 'samples': 13046784, 'steps': 25481, 'batch_loss/train': 0.7166177356848493}
+12/26/2021 07:52:11 - INFO - codeparrot_training - Step 25482: {'lr': 0.00024378208478826856, 'samples': 13047296, 'steps': 25482, 'batch_loss/train': 0.6999594018561766}
+12/26/2021 07:52:22 - INFO - codeparrot_training - Step 25483: {'lr': 0.00024376632654268913, 'samples': 13047808, 'steps': 25483, 'batch_loss/train': 0.7762803776422516}
+12/26/2021 07:52:34 - INFO - codeparrot_training - Step 25484: {'lr': 0.00024375056832189233, 'samples': 13048320, 'steps': 25484, 'batch_loss/train': 0.7487888978794217}
+12/26/2021 07:52:45 - INFO - codeparrot_training - Step 25485: {'lr': 0.00024373481012594092, 'samples': 13048832, 'steps': 25485, 'batch_loss/train': 0.7476593474857509}
+12/26/2021 07:52:55 - INFO - codeparrot_training - Step 25486: {'lr': 0.00024371905195489765, 'samples': 13049344, 'steps': 25486, 'batch_loss/train': 0.750846590846777}
+12/26/2021 07:53:08 - INFO - codeparrot_training - Step 25487: {'lr': 0.00024370329380882501, 'samples': 13049856, 'steps': 25487, 'batch_loss/train': 0.7131106555461884}
+12/26/2021 07:53:18 - INFO - codeparrot_training - Step 25488: {'lr': 0.00024368753568778563, 'samples': 13050368, 'steps': 25488, 'batch_loss/train': 0.7658937918022275}
+12/26/2021 07:53:29 - INFO - codeparrot_training - Step 25489: {'lr': 0.00024367177759184225, 'samples': 13050880, 'steps': 25489, 'batch_loss/train': 0.7303954078815877}
+12/26/2021 07:53:41 - INFO - codeparrot_training - Step 25490: {'lr': 0.00024365601952105757, 'samples': 13051392, 'steps': 25490, 'batch_loss/train': 0.7632824928732589}
+12/26/2021 07:53:52 - INFO - codeparrot_training - Step 25491: {'lr': 0.00024364026147549403, 'samples': 13051904, 'steps': 25491, 'batch_loss/train': 0.7004146655090153}
+12/26/2021 07:54:03 - INFO - codeparrot_training - Step 25492: {'lr': 0.0002436245034552145, 'samples': 13052416, 'steps': 25492, 'batch_loss/train': 0.8936387524008751}
+12/26/2021 07:54:13 - INFO - codeparrot_training - Step 25493: {'lr': 0.00024360874546028156, 'samples': 13052928, 'steps': 25493, 'batch_loss/train': 0.8933940138667822}
+12/26/2021 07:54:26 - INFO - codeparrot_training - Step 25494: {'lr': 0.00024359298749075783, 'samples': 13053440, 'steps': 25494, 'batch_loss/train': 0.7924457369372249}
+12/26/2021 07:54:36 - INFO - codeparrot_training - Step 25495: {'lr': 0.00024357722954670586, 'samples': 13053952, 'steps': 25495, 'batch_loss/train': 0.8117532152682543}
+12/26/2021 07:54:47 - INFO - codeparrot_training - Step 25496: {'lr': 0.00024356147162818848, 'samples': 13054464, 'steps': 25496, 'batch_loss/train': 0.8411374874413013}
+12/26/2021 07:54:59 - INFO - codeparrot_training - Step 25497: {'lr': 0.0002435457137352683, 'samples': 13054976, 'steps': 25497, 'batch_loss/train': 0.5607083175564185}
+12/26/2021 07:55:10 - INFO - codeparrot_training - Step 25498: {'lr': 0.00024352995586800785, 'samples': 13055488, 'steps': 25498, 'batch_loss/train': 0.4539109641336836}
+12/26/2021 07:55:20 - INFO - codeparrot_training - Step 25499: {'lr': 0.0002435141980264698, 'samples': 13056000, 'steps': 25499, 'batch_loss/train': 0.7429040626157075}
+12/26/2021 07:55:33 - INFO - codeparrot_training - Step 25500: {'lr': 0.00024349844021071698, 'samples': 13056512, 'steps': 25500, 'batch_loss/train': 0.8486783965490758}
+12/26/2021 07:55:44 - INFO - codeparrot_training - Step 25501: {'lr': 0.00024348268242081183, 'samples': 13057024, 'steps': 25501, 'batch_loss/train': 0.5929707678733394}
+12/26/2021 07:55:54 - INFO - codeparrot_training - Step 25502: {'lr': 0.00024346692465681703, 'samples': 13057536, 'steps': 25502, 'batch_loss/train': 0.7385403085500002}
+12/26/2021 07:56:05 - INFO - codeparrot_training - Step 25503: {'lr': 0.00024345116691879538, 'samples': 13058048, 'steps': 25503, 'batch_loss/train': 0.6885278676636517}
+12/26/2021 07:56:17 - INFO - codeparrot_training - Step 25504: {'lr': 0.00024343540920680943, 'samples': 13058560, 'steps': 25504, 'batch_loss/train': 0.7813744312152267}
+12/26/2021 07:56:28 - INFO - codeparrot_training - Step 25505: {'lr': 0.00024341965152092175, 'samples': 13059072, 'steps': 25505, 'batch_loss/train': 0.7755492059513927}
+12/26/2021 07:56:38 - INFO - codeparrot_training - Step 25506: {'lr': 0.000243403893861195, 'samples': 13059584, 'steps': 25506, 'batch_loss/train': 0.797609587199986}
+12/26/2021 07:56:50 - INFO - codeparrot_training - Step 25507: {'lr': 0.00024338813622769195, 'samples': 13060096, 'steps': 25507, 'batch_loss/train': 0.6656316828448325}
+12/26/2021 07:57:01 - INFO - codeparrot_training - Step 25508: {'lr': 0.00024337237862047517, 'samples': 13060608, 'steps': 25508, 'batch_loss/train': 0.7350138546898961}
+12/26/2021 07:57:12 - INFO - codeparrot_training - Step 25509: {'lr': 0.00024335662103960735, 'samples': 13061120, 'steps': 25509, 'batch_loss/train': 0.6631193880457431}
+12/26/2021 07:57:25 - INFO - codeparrot_training - Step 25510: {'lr': 0.00024334086348515096, 'samples': 13061632, 'steps': 25510, 'batch_loss/train': 0.7939260378479958}
+12/26/2021 07:57:35 - INFO - codeparrot_training - Step 25511: {'lr': 0.00024332510595716887, 'samples': 13062144, 'steps': 25511, 'batch_loss/train': 0.567274977802299}
+12/26/2021 07:57:46 - INFO - codeparrot_training - Step 25512: {'lr': 0.0002433093484557237, 'samples': 13062656, 'steps': 25512, 'batch_loss/train': 0.8364546671509743}
+12/26/2021 07:57:57 - INFO - codeparrot_training - Step 25513: {'lr': 0.0002432935909808779, 'samples': 13063168, 'steps': 25513, 'batch_loss/train': 0.7746496293693781}
+12/26/2021 07:58:09 - INFO - codeparrot_training - Step 25514: {'lr': 0.00024327783353269436, 'samples': 13063680, 'steps': 25514, 'batch_loss/train': 0.6773244286887348}
+12/26/2021 07:58:19 - INFO - codeparrot_training - Step 25515: {'lr': 0.00024326207611123556, 'samples': 13064192, 'steps': 25515, 'batch_loss/train': 0.7349844751879573}
+12/26/2021 07:58:30 - INFO - codeparrot_training - Step 25516: {'lr': 0.00024324631871656428, 'samples': 13064704, 'steps': 25516, 'batch_loss/train': 0.6271973676048219}
+12/26/2021 07:58:43 - INFO - codeparrot_training - Step 25517: {'lr': 0.00024323056134874295, 'samples': 13065216, 'steps': 25517, 'batch_loss/train': 0.7165581309236586}
+12/26/2021 07:58:54 - INFO - codeparrot_training - Step 25518: {'lr': 0.0002432148040078344, 'samples': 13065728, 'steps': 25518, 'batch_loss/train': 0.7643301365897059}
+12/26/2021 07:59:04 - INFO - codeparrot_training - Step 25519: {'lr': 0.0002431990466939013, 'samples': 13066240, 'steps': 25519, 'batch_loss/train': 0.7881850986741483}
+12/26/2021 07:59:16 - INFO - codeparrot_training - Step 25520: {'lr': 0.00024318328940700618, 'samples': 13066752, 'steps': 25520, 'batch_loss/train': 0.6828729519620538}
+12/26/2021 07:59:27 - INFO - codeparrot_training - Step 25521: {'lr': 0.00024316753214721166, 'samples': 13067264, 'steps': 25521, 'batch_loss/train': 0.6760604626033455}
+12/26/2021 07:59:37 - INFO - codeparrot_training - Step 25522: {'lr': 0.00024315177491458057, 'samples': 13067776, 'steps': 25522, 'batch_loss/train': 0.7200795961543918}
+12/26/2021 07:59:48 - INFO - codeparrot_training - Step 25523: {'lr': 0.00024313601770917543, 'samples': 13068288, 'steps': 25523, 'batch_loss/train': 0.8379979776218534}
+12/26/2021 08:00:00 - INFO - codeparrot_training - Step 25524: {'lr': 0.00024312026053105882, 'samples': 13068800, 'steps': 25524, 'batch_loss/train': 0.7415991751477122}
+12/26/2021 08:00:11 - INFO - codeparrot_training - Step 25525: {'lr': 0.0002431045033802934, 'samples': 13069312, 'steps': 25525, 'batch_loss/train': 0.8626988641917706}
+12/26/2021 08:00:21 - INFO - codeparrot_training - Step 25526: {'lr': 0.000243088746256942, 'samples': 13069824, 'steps': 25526, 'batch_loss/train': 0.7116710502887145}
+12/26/2021 08:00:34 - INFO - codeparrot_training - Step 25527: {'lr': 0.0002430729891610671, 'samples': 13070336, 'steps': 25527, 'batch_loss/train': 0.7741246717050672}
+12/26/2021 08:00:45 - INFO - codeparrot_training - Step 25528: {'lr': 0.0002430572320927313, 'samples': 13070848, 'steps': 25528, 'batch_loss/train': 0.7867710562422872}
+12/26/2021 08:00:55 - INFO - codeparrot_training - Step 25529: {'lr': 0.0002430414750519974, 'samples': 13071360, 'steps': 25529, 'batch_loss/train': 0.7652080627158284}
+12/26/2021 08:01:07 - INFO - codeparrot_training - Step 25530: {'lr': 0.00024302571803892802, 'samples': 13071872, 'steps': 25530, 'batch_loss/train': 0.7033906318247318}
+12/26/2021 08:01:18 - INFO - codeparrot_training - Step 25531: {'lr': 0.0002430099610535857, 'samples': 13072384, 'steps': 25531, 'batch_loss/train': 0.7764322590082884}
+12/26/2021 08:01:29 - INFO - codeparrot_training - Step 25532: {'lr': 0.00024299420409603305, 'samples': 13072896, 'steps': 25532, 'batch_loss/train': 0.7878284929320216}
+12/26/2021 08:01:39 - INFO - codeparrot_training - Step 25533: {'lr': 0.00024297844716633294, 'samples': 13073408, 'steps': 25533, 'batch_loss/train': 0.7328105643391609}
+12/26/2021 08:01:51 - INFO - codeparrot_training - Step 25534: {'lr': 0.0002429626902645478, 'samples': 13073920, 'steps': 25534, 'batch_loss/train': 0.7411608172114938}
+12/26/2021 08:02:02 - INFO - codeparrot_training - Step 25535: {'lr': 0.0002429469333907404, 'samples': 13074432, 'steps': 25535, 'batch_loss/train': 0.7631489369086921}
+12/26/2021 08:02:13 - INFO - codeparrot_training - Step 25536: {'lr': 0.00024293117654497322, 'samples': 13074944, 'steps': 25536, 'batch_loss/train': 0.738086523488164}
+12/26/2021 08:02:25 - INFO - codeparrot_training - Step 25537: {'lr': 0.00024291541972730905, 'samples': 13075456, 'steps': 25537, 'batch_loss/train': 0.7535044949036092}
+12/26/2021 08:02:35 - INFO - codeparrot_training - Step 25538: {'lr': 0.00024289966293781058, 'samples': 13075968, 'steps': 25538, 'batch_loss/train': 0.7353521944023669}
+12/26/2021 08:02:46 - INFO - codeparrot_training - Step 25539: {'lr': 0.00024288390617654024, 'samples': 13076480, 'steps': 25539, 'batch_loss/train': 0.7471957316156477}
+12/26/2021 08:02:59 - INFO - codeparrot_training - Step 25540: {'lr': 0.0002428681494435609, 'samples': 13076992, 'steps': 25540, 'batch_loss/train': 1.4619121607393026}
+12/26/2021 08:03:10 - INFO - codeparrot_training - Step 25541: {'lr': 0.0002428523927389351, 'samples': 13077504, 'steps': 25541, 'batch_loss/train': 1.5842913356609643}
+12/26/2021 08:03:20 - INFO - codeparrot_training - Step 25542: {'lr': 0.00024283663606272548, 'samples': 13078016, 'steps': 25542, 'batch_loss/train': 0.7480158703401685}
+12/26/2021 08:03:31 - INFO - codeparrot_training - Step 25543: {'lr': 0.00024282087941499464, 'samples': 13078528, 'steps': 25543, 'batch_loss/train': 0.7749808831140399}
+12/26/2021 08:03:43 - INFO - codeparrot_training - Step 25544: {'lr': 0.0002428051227958053, 'samples': 13079040, 'steps': 25544, 'batch_loss/train': 0.6896400721743703}
+12/26/2021 08:03:54 - INFO - codeparrot_training - Step 25545: {'lr': 0.00024278936620522013, 'samples': 13079552, 'steps': 25545, 'batch_loss/train': 0.8794459467753768}
+12/26/2021 08:04:04 - INFO - codeparrot_training - Step 25546: {'lr': 0.00024277360964330164, 'samples': 13080064, 'steps': 25546, 'batch_loss/train': 0.8534476819913834}
+12/26/2021 08:04:17 - INFO - codeparrot_training - Step 25547: {'lr': 0.00024275785311011252, 'samples': 13080576, 'steps': 25547, 'batch_loss/train': 0.6423857440240681}
+12/26/2021 08:04:27 - INFO - codeparrot_training - Step 25548: {'lr': 0.00024274209660571556, 'samples': 13081088, 'steps': 25548, 'batch_loss/train': 0.8141310745850205}
+12/26/2021 08:04:38 - INFO - codeparrot_training - Step 25549: {'lr': 0.0002427263401301733, 'samples': 13081600, 'steps': 25549, 'batch_loss/train': 0.7233685841783881}
+12/26/2021 08:04:51 - INFO - codeparrot_training - Step 25550: {'lr': 0.0002427105836835482, 'samples': 13082112, 'steps': 25550, 'batch_loss/train': 0.7557749408297241}
+12/26/2021 08:05:01 - INFO - codeparrot_training - Step 25551: {'lr': 0.00024269482726590314, 'samples': 13082624, 'steps': 25551, 'batch_loss/train': 0.7225532999727875}
+12/26/2021 08:05:12 - INFO - codeparrot_training - Step 25552: {'lr': 0.00024267907087730076, 'samples': 13083136, 'steps': 25552, 'batch_loss/train': 0.8197799436748028}
+12/26/2021 08:05:24 - INFO - codeparrot_training - Step 25553: {'lr': 0.00024266331451780357, 'samples': 13083648, 'steps': 25553, 'batch_loss/train': 0.719443621346727}
+12/26/2021 08:05:34 - INFO - codeparrot_training - Step 25554: {'lr': 0.0002426475581874742, 'samples': 13084160, 'steps': 25554, 'batch_loss/train': 0.7277134156320244}
+12/26/2021 08:05:45 - INFO - codeparrot_training - Step 25555: {'lr': 0.00024263180188637549, 'samples': 13084672, 'steps': 25555, 'batch_loss/train': 0.7578696487471461}
+12/26/2021 08:05:56 - INFO - codeparrot_training - Step 25556: {'lr': 0.0002426160456145699, 'samples': 13085184, 'steps': 25556, 'batch_loss/train': 0.8113365648314357}
+12/26/2021 08:06:09 - INFO - codeparrot_training - Step 25557: {'lr': 0.00024260028937212017, 'samples': 13085696, 'steps': 25557, 'batch_loss/train': 0.7272019472438842}
+12/26/2021 08:06:19 - INFO - codeparrot_training - Step 25558: {'lr': 0.0002425845331590888, 'samples': 13086208, 'steps': 25558, 'batch_loss/train': 1.7482003327459097}
+12/26/2021 08:06:30 - INFO - codeparrot_training - Step 25559: {'lr': 0.00024256877697553862, 'samples': 13086720, 'steps': 25559, 'batch_loss/train': 0.6849408550187945}
+12/26/2021 08:06:42 - INFO - codeparrot_training - Step 25560: {'lr': 0.00024255302082153214, 'samples': 13087232, 'steps': 25560, 'batch_loss/train': 0.8043036805465817}
+12/26/2021 08:06:52 - INFO - codeparrot_training - Step 25561: {'lr': 0.00024253726469713208, 'samples': 13087744, 'steps': 25561, 'batch_loss/train': 0.7244565216824412}
+12/26/2021 08:07:03 - INFO - codeparrot_training - Step 25562: {'lr': 0.0002425215086024009, 'samples': 13088256, 'steps': 25562, 'batch_loss/train': 0.7623887090012431}
+12/26/2021 08:07:15 - INFO - codeparrot_training - Step 25563: {'lr': 0.0002425057525374015, 'samples': 13088768, 'steps': 25563, 'batch_loss/train': 0.7684190995059907}
+12/26/2021 08:07:26 - INFO - codeparrot_training - Step 25564: {'lr': 0.00024248999650219638, 'samples': 13089280, 'steps': 25564, 'batch_loss/train': 0.8243449740111828}
+12/26/2021 08:07:36 - INFO - codeparrot_training - Step 25565: {'lr': 0.00024247424049684814, 'samples': 13089792, 'steps': 25565, 'batch_loss/train': 0.7946715466678143}
+12/26/2021 08:07:47 - INFO - codeparrot_training - Step 25566: {'lr': 0.0002424584845214195, 'samples': 13090304, 'steps': 25566, 'batch_loss/train': 0.8640458779409528}
+12/26/2021 08:08:00 - INFO - codeparrot_training - Step 25567: {'lr': 0.00024244272857597315, 'samples': 13090816, 'steps': 25567, 'batch_loss/train': 0.800279809627682}
+12/26/2021 08:08:10 - INFO - codeparrot_training - Step 25568: {'lr': 0.00024242697266057167, 'samples': 13091328, 'steps': 25568, 'batch_loss/train': 0.7536684954538941}
+12/26/2021 08:08:21 - INFO - codeparrot_training - Step 25569: {'lr': 0.00024241121677527755, 'samples': 13091840, 'steps': 25569, 'batch_loss/train': 0.7600521584972739}
+12/26/2021 08:08:33 - INFO - codeparrot_training - Step 25570: {'lr': 0.00024239546092015362, 'samples': 13092352, 'steps': 25570, 'batch_loss/train': 0.9224528511986136}
+12/26/2021 08:08:44 - INFO - codeparrot_training - Step 25571: {'lr': 0.00024237970509526256, 'samples': 13092864, 'steps': 25571, 'batch_loss/train': 0.7183505739085376}
+12/26/2021 08:08:55 - INFO - codeparrot_training - Step 25572: {'lr': 0.00024236394930066683, 'samples': 13093376, 'steps': 25572, 'batch_loss/train': 0.8200003032106906}
+12/26/2021 08:09:07 - INFO - codeparrot_training - Step 25573: {'lr': 0.00024234819353642913, 'samples': 13093888, 'steps': 25573, 'batch_loss/train': 0.710227167699486}
+12/26/2021 08:09:18 - INFO - codeparrot_training - Step 25574: {'lr': 0.0002423324378026122, 'samples': 13094400, 'steps': 25574, 'batch_loss/train': 0.798586618155241}
+12/26/2021 08:09:29 - INFO - codeparrot_training - Step 25575: {'lr': 0.0002423166820992786, 'samples': 13094912, 'steps': 25575, 'batch_loss/train': 0.7765833102166653}
+12/26/2021 08:09:39 - INFO - codeparrot_training - Step 25576: {'lr': 0.00024230092642649087, 'samples': 13095424, 'steps': 25576, 'batch_loss/train': 0.8258945439010859}
+12/26/2021 08:09:51 - INFO - codeparrot_training - Step 25577: {'lr': 0.0002422851707843118, 'samples': 13095936, 'steps': 25577, 'batch_loss/train': 0.8323613165412098}
+12/26/2021 08:10:02 - INFO - codeparrot_training - Step 25578: {'lr': 0.00024226941517280405, 'samples': 13096448, 'steps': 25578, 'batch_loss/train': 0.8673857040703297}
+12/26/2021 08:10:12 - INFO - codeparrot_training - Step 25579: {'lr': 0.0002422536595920301, 'samples': 13096960, 'steps': 25579, 'batch_loss/train': 0.922815608792007}
+12/26/2021 08:10:25 - INFO - codeparrot_training - Step 25580: {'lr': 0.00024223790404205264, 'samples': 13097472, 'steps': 25580, 'batch_loss/train': 0.7278592064976692}
+12/26/2021 08:10:35 - INFO - codeparrot_training - Step 25581: {'lr': 0.00024222214852293448, 'samples': 13097984, 'steps': 25581, 'batch_loss/train': 0.6751248650252819}
+12/26/2021 08:10:46 - INFO - codeparrot_training - Step 25582: {'lr': 0.00024220639303473805, 'samples': 13098496, 'steps': 25582, 'batch_loss/train': 0.7653814805671573}
+12/26/2021 08:10:58 - INFO - codeparrot_training - Step 25583: {'lr': 0.00024219063757752608, 'samples': 13099008, 'steps': 25583, 'batch_loss/train': 0.7525776531547308}
+12/26/2021 08:11:09 - INFO - codeparrot_training - Step 25584: {'lr': 0.0002421748821513611, 'samples': 13099520, 'steps': 25584, 'batch_loss/train': 0.8457002146169543}
+12/26/2021 08:11:19 - INFO - codeparrot_training - Step 25585: {'lr': 0.00024215912675630593, 'samples': 13100032, 'steps': 25585, 'batch_loss/train': 0.9936649659648538}
+12/26/2021 08:11:30 - INFO - codeparrot_training - Step 25586: {'lr': 0.00024214337139242308, 'samples': 13100544, 'steps': 25586, 'batch_loss/train': 0.787125249626115}
+12/26/2021 08:11:44 - INFO - codeparrot_training - Step 25587: {'lr': 0.00024212761605977516, 'samples': 13101056, 'steps': 25587, 'batch_loss/train': 0.753306852420792}
+12/26/2021 08:11:54 - INFO - codeparrot_training - Step 25588: {'lr': 0.00024211186075842494, 'samples': 13101568, 'steps': 25588, 'batch_loss/train': 0.8850392820313573}
+12/26/2021 08:12:05 - INFO - codeparrot_training - Step 25589: {'lr': 0.00024209610548843496, 'samples': 13102080, 'steps': 25589, 'batch_loss/train': 0.8164111841470003}
+12/26/2021 08:12:17 - INFO - codeparrot_training - Step 25590: {'lr': 0.00024208035024986797, 'samples': 13102592, 'steps': 25590, 'batch_loss/train': 0.7593295760452747}
+12/26/2021 08:12:28 - INFO - codeparrot_training - Step 25591: {'lr': 0.00024206459504278634, 'samples': 13103104, 'steps': 25591, 'batch_loss/train': 0.6801232043653727}
+12/26/2021 08:12:38 - INFO - codeparrot_training - Step 25592: {'lr': 0.00024204883986725297, 'samples': 13103616, 'steps': 25592, 'batch_loss/train': 0.8775069303810596}
+12/26/2021 08:12:51 - INFO - codeparrot_training - Step 25593: {'lr': 0.00024203308472333046, 'samples': 13104128, 'steps': 25593, 'batch_loss/train': 0.6741062770597637}
+12/26/2021 08:13:01 - INFO - codeparrot_training - Step 25594: {'lr': 0.00024201732961108143, 'samples': 13104640, 'steps': 25594, 'batch_loss/train': 0.790467849932611}
+12/26/2021 08:13:12 - INFO - codeparrot_training - Step 25595: {'lr': 0.00024200157453056832, 'samples': 13105152, 'steps': 25595, 'batch_loss/train': 0.8367215236648917}
+12/26/2021 08:13:23 - INFO - codeparrot_training - Step 25596: {'lr': 0.00024198581948185402, 'samples': 13105664, 'steps': 25596, 'batch_loss/train': 0.44364497589413077}
+12/26/2021 08:13:35 - INFO - codeparrot_training - Step 25597: {'lr': 0.00024197006446500112, 'samples': 13106176, 'steps': 25597, 'batch_loss/train': 0.7509783655405045}
+12/26/2021 08:13:46 - INFO - codeparrot_training - Step 25598: {'lr': 0.00024195430948007217, 'samples': 13106688, 'steps': 25598, 'batch_loss/train': 0.6636368716135621}
+12/26/2021 08:13:57 - INFO - codeparrot_training - Step 25599: {'lr': 0.00024193855452712975, 'samples': 13107200, 'steps': 25599, 'batch_loss/train': 0.6974060162901878}
+12/26/2021 08:14:09 - INFO - codeparrot_training - Step 25600: {'lr': 0.00024192279960623674, 'samples': 13107712, 'steps': 25600, 'batch_loss/train': 1.0024978760629892}
+12/26/2021 08:14:19 - INFO - codeparrot_training - Step 25601: {'lr': 0.00024190704471745557, 'samples': 13108224, 'steps': 25601, 'batch_loss/train': 0.8435797831043601}
+12/26/2021 08:14:30 - INFO - codeparrot_training - Step 25602: {'lr': 0.00024189128986084888, 'samples': 13108736, 'steps': 25602, 'batch_loss/train': 0.7231028815731406}
+12/26/2021 08:14:43 - INFO - codeparrot_training - Step 25603: {'lr': 0.00024187553503647946, 'samples': 13109248, 'steps': 25603, 'batch_loss/train': 0.618755720090121}
+12/26/2021 08:14:54 - INFO - codeparrot_training - Step 25604: {'lr': 0.00024185978024440986, 'samples': 13109760, 'steps': 25604, 'batch_loss/train': 0.7936783460900187}
+12/26/2021 08:15:04 - INFO - codeparrot_training - Step 25605: {'lr': 0.0002418440254847026, 'samples': 13110272, 'steps': 25605, 'batch_loss/train': 0.6919245980679989}
+12/26/2021 08:15:15 - INFO - codeparrot_training - Step 25606: {'lr': 0.00024182827075742042, 'samples': 13110784, 'steps': 25606, 'batch_loss/train': 0.7902758882846683}
+12/26/2021 08:15:27 - INFO - codeparrot_training - Step 25607: {'lr': 0.00024181251606262603, 'samples': 13111296, 'steps': 25607, 'batch_loss/train': 0.7506070043891668}
+12/26/2021 08:15:37 - INFO - codeparrot_training - Step 25608: {'lr': 0.00024179676140038194, 'samples': 13111808, 'steps': 25608, 'batch_loss/train': 0.7658517234958708}
+12/26/2021 08:15:48 - INFO - codeparrot_training - Step 25609: {'lr': 0.00024178100677075087, 'samples': 13112320, 'steps': 25609, 'batch_loss/train': 0.7822896768338978}
+12/26/2021 08:16:00 - INFO - codeparrot_training - Step 25610: {'lr': 0.00024176525217379531, 'samples': 13112832, 'steps': 25610, 'batch_loss/train': 0.852112396620214}
+12/26/2021 08:16:11 - INFO - codeparrot_training - Step 25611: {'lr': 0.00024174949760957806, 'samples': 13113344, 'steps': 25611, 'batch_loss/train': 0.7652851152233779}
+12/26/2021 08:16:21 - INFO - codeparrot_training - Step 25612: {'lr': 0.00024173374307816174, 'samples': 13113856, 'steps': 25612, 'batch_loss/train': 0.8033472802489996}
+12/26/2021 08:16:34 - INFO - codeparrot_training - Step 25613: {'lr': 0.00024171798857960884, 'samples': 13114368, 'steps': 25613, 'batch_loss/train': 0.8066268293187022}
+12/26/2021 08:16:45 - INFO - codeparrot_training - Step 25614: {'lr': 0.00024170223411398219, 'samples': 13114880, 'steps': 25614, 'batch_loss/train': 0.8279240019619465}
+12/26/2021 08:16:55 - INFO - codeparrot_training - Step 25615: {'lr': 0.00024168647968134425, 'samples': 13115392, 'steps': 25615, 'batch_loss/train': 0.7942458707839251}
+12/26/2021 08:17:07 - INFO - codeparrot_training - Step 25616: {'lr': 0.0002416707252817578, 'samples': 13115904, 'steps': 25616, 'batch_loss/train': 0.8525265846401453}
+12/26/2021 08:17:18 - INFO - codeparrot_training - Step 25617: {'lr': 0.0002416549709152853, 'samples': 13116416, 'steps': 25617, 'batch_loss/train': 0.7663184478878975}
+12/26/2021 08:17:29 - INFO - codeparrot_training - Step 25618: {'lr': 0.00024163921658198955, 'samples': 13116928, 'steps': 25618, 'batch_loss/train': 0.7457417282275856}
+12/26/2021 08:17:39 - INFO - codeparrot_training - Step 25619: {'lr': 0.00024162346228193314, 'samples': 13117440, 'steps': 25619, 'batch_loss/train': 0.8245335537940264}
+12/26/2021 08:17:52 - INFO - codeparrot_training - Step 25620: {'lr': 0.0002416077080151787, 'samples': 13117952, 'steps': 25620, 'batch_loss/train': 1.127496127039194}
+12/26/2021 08:18:02 - INFO - codeparrot_training - Step 25621: {'lr': 0.00024159195378178872, 'samples': 13118464, 'steps': 25621, 'batch_loss/train': 0.7195271225646138}
+12/26/2021 08:18:13 - INFO - codeparrot_training - Step 25622: {'lr': 0.00024157619958182602, 'samples': 13118976, 'steps': 25622, 'batch_loss/train': 0.7661439492367208}
+12/26/2021 08:18:25 - INFO - codeparrot_training - Step 25623: {'lr': 0.00024156044541535324, 'samples': 13119488, 'steps': 25623, 'batch_loss/train': 0.8303496846929193}
+12/26/2021 08:18:36 - INFO - codeparrot_training - Step 25624: {'lr': 0.0002415446912824328, 'samples': 13120000, 'steps': 25624, 'batch_loss/train': 0.6610793732106686}
+12/26/2021 08:18:47 - INFO - codeparrot_training - Step 25625: {'lr': 0.00024152893718312754, 'samples': 13120512, 'steps': 25625, 'batch_loss/train': 0.803855866426602}
+12/26/2021 08:18:59 - INFO - codeparrot_training - Step 25626: {'lr': 0.00024151318311750012, 'samples': 13121024, 'steps': 25626, 'batch_loss/train': 0.895193855278194}
+12/26/2021 08:19:10 - INFO - codeparrot_training - Step 25627: {'lr': 0.00024149742908561298, 'samples': 13121536, 'steps': 25627, 'batch_loss/train': 0.7403868939727545}
+12/26/2021 08:19:20 - INFO - codeparrot_training - Step 25628: {'lr': 0.00024148167508752883, 'samples': 13122048, 'steps': 25628, 'batch_loss/train': 0.7678479219321162}
+12/26/2021 08:19:31 - INFO - codeparrot_training - Step 25629: {'lr': 0.0002414659211233104, 'samples': 13122560, 'steps': 25629, 'batch_loss/train': 0.7657164447009563}
+12/26/2021 08:19:43 - INFO - codeparrot_training - Step 25630: {'lr': 0.00024145016719302027, 'samples': 13123072, 'steps': 25630, 'batch_loss/train': 0.7660337975248694}
+12/26/2021 08:19:54 - INFO - codeparrot_training - Step 25631: {'lr': 0.00024143441329672096, 'samples': 13123584, 'steps': 25631, 'batch_loss/train': 0.8475355925038457}
+12/26/2021 08:20:04 - INFO - codeparrot_training - Step 25632: {'lr': 0.00024141865943447515, 'samples': 13124096, 'steps': 25632, 'batch_loss/train': 0.7750682914629579}
+12/26/2021 08:20:16 - INFO - codeparrot_training - Step 25633: {'lr': 0.00024140290560634565, 'samples': 13124608, 'steps': 25633, 'batch_loss/train': 0.7848186660557985}
+12/26/2021 08:20:27 - INFO - codeparrot_training - Step 25634: {'lr': 0.00024138715181239487, 'samples': 13125120, 'steps': 25634, 'batch_loss/train': 0.686711935326457}
+12/26/2021 08:20:38 - INFO - codeparrot_training - Step 25635: {'lr': 0.00024137139805268556, 'samples': 13125632, 'steps': 25635, 'batch_loss/train': 0.7536587476497516}
+12/26/2021 08:20:50 - INFO - codeparrot_training - Step 25636: {'lr': 0.00024135564432728023, 'samples': 13126144, 'steps': 25636, 'batch_loss/train': 0.8220540774054825}
+12/26/2021 08:21:01 - INFO - codeparrot_training - Step 25637: {'lr': 0.00024133989063624165, 'samples': 13126656, 'steps': 25637, 'batch_loss/train': 0.8285356843844056}
+12/26/2021 08:21:12 - INFO - codeparrot_training - Step 25638: {'lr': 0.00024132413697963245, 'samples': 13127168, 'steps': 25638, 'batch_loss/train': 0.8406379446387291}
+12/26/2021 08:21:22 - INFO - codeparrot_training - Step 25639: {'lr': 0.00024130838335751509, 'samples': 13127680, 'steps': 25639, 'batch_loss/train': 0.7448771416675299}
+12/26/2021 08:21:34 - INFO - codeparrot_training - Step 25640: {'lr': 0.0002412926297699524, 'samples': 13128192, 'steps': 25640, 'batch_loss/train': 0.7736082436749712}
+12/26/2021 08:21:45 - INFO - codeparrot_training - Step 25641: {'lr': 0.00024127687621700692, 'samples': 13128704, 'steps': 25641, 'batch_loss/train': 0.7507635420188308}
+12/26/2021 08:21:56 - INFO - codeparrot_training - Step 25642: {'lr': 0.00024126112269874132, 'samples': 13129216, 'steps': 25642, 'batch_loss/train': 0.8442133297212422}
+12/26/2021 08:22:08 - INFO - codeparrot_training - Step 25643: {'lr': 0.0002412453692152181, 'samples': 13129728, 'steps': 25643, 'batch_loss/train': 0.7651243424043059}
+12/26/2021 08:22:19 - INFO - codeparrot_training - Step 25644: {'lr': 0.0002412296157665, 'samples': 13130240, 'steps': 25644, 'batch_loss/train': 0.7152375257574022}
+12/26/2021 08:22:30 - INFO - codeparrot_training - Step 25645: {'lr': 0.00024121386235264973, 'samples': 13130752, 'steps': 25645, 'batch_loss/train': 0.800598953384906}
+12/26/2021 08:22:42 - INFO - codeparrot_training - Step 25646: {'lr': 0.00024119810897372973, 'samples': 13131264, 'steps': 25646, 'batch_loss/train': 0.7278207242488861}
+12/26/2021 08:22:53 - INFO - codeparrot_training - Step 25647: {'lr': 0.0002411823556298027, 'samples': 13131776, 'steps': 25647, 'batch_loss/train': 0.7406101655215025}
+12/26/2021 08:23:03 - INFO - codeparrot_training - Step 25648: {'lr': 0.0002411666023209314, 'samples': 13132288, 'steps': 25648, 'batch_loss/train': 0.7755062775686383}
+12/26/2021 08:23:15 - INFO - codeparrot_training - Step 25649: {'lr': 0.00024115084904717836, 'samples': 13132800, 'steps': 25649, 'batch_loss/train': 0.6750445789657533}
+12/26/2021 08:23:26 - INFO - codeparrot_training - Step 25650: {'lr': 0.0002411350958086061, 'samples': 13133312, 'steps': 25650, 'batch_loss/train': 0.8506113239564002}
+12/26/2021 08:23:37 - INFO - codeparrot_training - Step 25651: {'lr': 0.00024111934260527738, 'samples': 13133824, 'steps': 25651, 'batch_loss/train': 0.7383399084210396}
+12/26/2021 08:23:47 - INFO - codeparrot_training - Step 25652: {'lr': 0.0002411035894372549, 'samples': 13134336, 'steps': 25652, 'batch_loss/train': 0.8143528802320361}
+12/26/2021 08:24:00 - INFO - codeparrot_training - Step 25653: {'lr': 0.00024108783630460108, 'samples': 13134848, 'steps': 25653, 'batch_loss/train': 0.8548581749200821}
+12/26/2021 08:24:11 - INFO - codeparrot_training - Step 25654: {'lr': 0.00024107208320737862, 'samples': 13135360, 'steps': 25654, 'batch_loss/train': 0.7694508843123913}
+12/26/2021 08:24:22 - INFO - codeparrot_training - Step 25655: {'lr': 0.0002410563301456503, 'samples': 13135872, 'steps': 25655, 'batch_loss/train': 0.6568312533199787}
+12/26/2021 08:24:34 - INFO - codeparrot_training - Step 25656: {'lr': 0.00024104057711947857, 'samples': 13136384, 'steps': 25656, 'batch_loss/train': 0.7584679732099175}
+12/26/2021 08:24:44 - INFO - codeparrot_training - Step 25657: {'lr': 0.0002410248241289262, 'samples': 13136896, 'steps': 25657, 'batch_loss/train': 0.7533673010766506}
+12/26/2021 08:24:55 - INFO - codeparrot_training - Step 25658: {'lr': 0.00024100907117405562, 'samples': 13137408, 'steps': 25658, 'batch_loss/train': 1.001582687254995}
+12/26/2021 08:25:08 - INFO - codeparrot_training - Step 25659: {'lr': 0.0002409933182549297, 'samples': 13137920, 'steps': 25659, 'batch_loss/train': 0.7891017072834074}
+12/26/2021 08:25:19 - INFO - codeparrot_training - Step 25660: {'lr': 0.0002409775653716109, 'samples': 13138432, 'steps': 25660, 'batch_loss/train': 0.8817252553999424}
+12/26/2021 08:25:29 - INFO - codeparrot_training - Step 25661: {'lr': 0.0002409618125241618, 'samples': 13138944, 'steps': 25661, 'batch_loss/train': 0.5441136875888333}
+12/26/2021 08:25:40 - INFO - codeparrot_training - Step 25662: {'lr': 0.00024094605971264525, 'samples': 13139456, 'steps': 25662, 'batch_loss/train': 0.517150899511762}
+12/26/2021 08:25:52 - INFO - codeparrot_training - Step 25663: {'lr': 0.00024093030693712368, 'samples': 13139968, 'steps': 25663, 'batch_loss/train': 0.7682713996618986}
+12/26/2021 08:26:02 - INFO - codeparrot_training - Step 25664: {'lr': 0.00024091455419765988, 'samples': 13140480, 'steps': 25664, 'batch_loss/train': 0.8016022838419303}
+12/26/2021 08:26:13 - INFO - codeparrot_training - Step 25665: {'lr': 0.00024089880149431622, 'samples': 13140992, 'steps': 25665, 'batch_loss/train': 0.7519746003672481}
+12/26/2021 08:26:26 - INFO - codeparrot_training - Step 25666: {'lr': 0.00024088304882715562, 'samples': 13141504, 'steps': 25666, 'batch_loss/train': 0.7858880301937461}
+12/26/2021 08:26:36 - INFO - codeparrot_training - Step 25667: {'lr': 0.00024086729619624055, 'samples': 13142016, 'steps': 25667, 'batch_loss/train': 0.6504157148301601}
+12/26/2021 08:26:47 - INFO - codeparrot_training - Step 25668: {'lr': 0.00024085154360163367, 'samples': 13142528, 'steps': 25668, 'batch_loss/train': 0.8548199199140072}
+12/26/2021 08:26:59 - INFO - codeparrot_training - Step 25669: {'lr': 0.00024083579104339751, 'samples': 13143040, 'steps': 25669, 'batch_loss/train': 0.6942332901526242}
+12/26/2021 08:27:10 - INFO - codeparrot_training - Step 25670: {'lr': 0.00024082003852159484, 'samples': 13143552, 'steps': 25670, 'batch_loss/train': 0.7351198044489138}
+12/26/2021 08:27:20 - INFO - codeparrot_training - Step 25671: {'lr': 0.00024080428603628825, 'samples': 13144064, 'steps': 25671, 'batch_loss/train': 0.7527138320729136}
+12/26/2021 08:27:33 - INFO - codeparrot_training - Step 25672: {'lr': 0.00024078853358754036, 'samples': 13144576, 'steps': 25672, 'batch_loss/train': 0.8404470253735781}
+12/26/2021 08:27:44 - INFO - codeparrot_training - Step 25673: {'lr': 0.00024077278117541368, 'samples': 13145088, 'steps': 25673, 'batch_loss/train': 0.692695727571845}
+12/26/2021 08:27:54 - INFO - codeparrot_training - Step 25674: {'lr': 0.00024075702879997106, 'samples': 13145600, 'steps': 25674, 'batch_loss/train': 0.6913813119754195}
+12/26/2021 08:28:05 - INFO - codeparrot_training - Step 25675: {'lr': 0.000240741276461275, 'samples': 13146112, 'steps': 25675, 'batch_loss/train': 0.7882913071662188}
+12/26/2021 08:28:17 - INFO - codeparrot_training - Step 25676: {'lr': 0.00024072552415938802, 'samples': 13146624, 'steps': 25676, 'batch_loss/train': 0.7930404988583177}
+12/26/2021 08:28:28 - INFO - codeparrot_training - Step 25677: {'lr': 0.0002407097718943729, 'samples': 13147136, 'steps': 25677, 'batch_loss/train': 1.2901873313821852}
+12/26/2021 08:28:38 - INFO - codeparrot_training - Step 25678: {'lr': 0.00024069401966629229, 'samples': 13147648, 'steps': 25678, 'batch_loss/train': 0.702502197586}
+12/26/2021 08:28:51 - INFO - codeparrot_training - Step 25679: {'lr': 0.00024067826747520868, 'samples': 13148160, 'steps': 25679, 'batch_loss/train': 0.7899191472679377}
+12/26/2021 08:29:01 - INFO - codeparrot_training - Step 25680: {'lr': 0.00024066251532118473, 'samples': 13148672, 'steps': 25680, 'batch_loss/train': 0.7923524482175708}
+12/26/2021 08:29:12 - INFO - codeparrot_training - Step 25681: {'lr': 0.00024064676320428318, 'samples': 13149184, 'steps': 25681, 'batch_loss/train': 0.6240557470591739}
+12/26/2021 08:29:25 - INFO - codeparrot_training - Step 25682: {'lr': 0.0002406310111245665, 'samples': 13149696, 'steps': 25682, 'batch_loss/train': 0.8028945070691407}
+12/26/2021 08:29:35 - INFO - codeparrot_training - Step 25683: {'lr': 0.00024061525908209748, 'samples': 13150208, 'steps': 25683, 'batch_loss/train': 1.1717665675096214}
+12/26/2021 08:29:46 - INFO - codeparrot_training - Step 25684: {'lr': 0.00024059950707693847, 'samples': 13150720, 'steps': 25684, 'batch_loss/train': 1.4787830892018974}
+12/26/2021 08:29:57 - INFO - codeparrot_training - Step 25685: {'lr': 0.00024058375510915242, 'samples': 13151232, 'steps': 25685, 'batch_loss/train': 0.9157687267288566}
+12/26/2021 08:30:09 - INFO - codeparrot_training - Step 25686: {'lr': 0.00024056800317880176, 'samples': 13151744, 'steps': 25686, 'batch_loss/train': 0.7809232342988253}
+12/26/2021 08:30:19 - INFO - codeparrot_training - Step 25687: {'lr': 0.0002405522512859491, 'samples': 13152256, 'steps': 25687, 'batch_loss/train': 0.8665598444640636}
+12/26/2021 08:30:30 - INFO - codeparrot_training - Step 25688: {'lr': 0.00024053649943065722, 'samples': 13152768, 'steps': 25688, 'batch_loss/train': 0.7689689036924392}
+12/26/2021 08:30:43 - INFO - codeparrot_training - Step 25689: {'lr': 0.0002405207476129886, 'samples': 13153280, 'steps': 25689, 'batch_loss/train': 0.789722110144794}
+12/26/2021 08:30:53 - INFO - codeparrot_training - Step 25690: {'lr': 0.00024050499583300595, 'samples': 13153792, 'steps': 25690, 'batch_loss/train': 0.8239286374300718}
+12/26/2021 08:31:04 - INFO - codeparrot_training - Step 25691: {'lr': 0.00024048924409077175, 'samples': 13154304, 'steps': 25691, 'batch_loss/train': 0.5402957508340478}
+12/26/2021 08:31:16 - INFO - codeparrot_training - Step 25692: {'lr': 0.00024047349238634876, 'samples': 13154816, 'steps': 25692, 'batch_loss/train': 0.8345676432363689}
+12/26/2021 08:31:27 - INFO - codeparrot_training - Step 25693: {'lr': 0.00024045774071979964, 'samples': 13155328, 'steps': 25693, 'batch_loss/train': 0.7936202669516206}
+12/26/2021 08:31:37 - INFO - codeparrot_training - Step 25694: {'lr': 0.00024044198909118695, 'samples': 13155840, 'steps': 25694, 'batch_loss/train': 0.7949494943022728}
+12/26/2021 08:31:48 - INFO - codeparrot_training - Step 25695: {'lr': 0.00024042623750057315, 'samples': 13156352, 'steps': 25695, 'batch_loss/train': 0.8667920362204313}
+12/26/2021 08:32:00 - INFO - codeparrot_training - Step 25696: {'lr': 0.00024041048594802112, 'samples': 13156864, 'steps': 25696, 'batch_loss/train': 0.78974160249345}
+12/26/2021 08:32:11 - INFO - codeparrot_training - Step 25697: {'lr': 0.00024039473443359343, 'samples': 13157376, 'steps': 25697, 'batch_loss/train': 0.8008343577384949}
+12/26/2021 08:32:21 - INFO - codeparrot_training - Step 25698: {'lr': 0.0002403789829573525, 'samples': 13157888, 'steps': 25698, 'batch_loss/train': 0.7565186815336347}
+12/26/2021 08:32:34 - INFO - codeparrot_training - Step 25699: {'lr': 0.0002403632315193612, 'samples': 13158400, 'steps': 25699, 'batch_loss/train': 0.7636766098439693}
+12/26/2021 08:32:45 - INFO - codeparrot_training - Step 25700: {'lr': 0.00024034748011968207, 'samples': 13158912, 'steps': 25700, 'batch_loss/train': 0.8322267290204763}
+12/26/2021 08:32:55 - INFO - codeparrot_training - Step 25701: {'lr': 0.00024033172875837766, 'samples': 13159424, 'steps': 25701, 'batch_loss/train': 0.8310914281755686}
+12/26/2021 08:33:07 - INFO - codeparrot_training - Step 25702: {'lr': 0.00024031597743551063, 'samples': 13159936, 'steps': 25702, 'batch_loss/train': 0.8530555604957044}
+12/26/2021 08:33:18 - INFO - codeparrot_training - Step 25703: {'lr': 0.0002403002261511437, 'samples': 13160448, 'steps': 25703, 'batch_loss/train': 0.8359483107924461}
+12/26/2021 08:33:28 - INFO - codeparrot_training - Step 25704: {'lr': 0.0002402844749053394, 'samples': 13160960, 'steps': 25704, 'batch_loss/train': 0.7300987932831049}
+12/26/2021 08:33:40 - INFO - codeparrot_training - Step 25705: {'lr': 0.00024026872369816032, 'samples': 13161472, 'steps': 25705, 'batch_loss/train': 0.7146923262625933}
+12/26/2021 08:33:51 - INFO - codeparrot_training - Step 25706: {'lr': 0.00024025297252966904, 'samples': 13161984, 'steps': 25706, 'batch_loss/train': 0.7606488345190883}
+12/26/2021 08:34:02 - INFO - codeparrot_training - Step 25707: {'lr': 0.0002402372213999284, 'samples': 13162496, 'steps': 25707, 'batch_loss/train': 0.851669903844595}
+12/26/2021 08:34:12 - INFO - codeparrot_training - Step 25708: {'lr': 0.00024022147030900083, 'samples': 13163008, 'steps': 25708, 'batch_loss/train': 0.743648887379095}
+12/26/2021 08:34:24 - INFO - codeparrot_training - Step 25709: {'lr': 0.00024020571925694904, 'samples': 13163520, 'steps': 25709, 'batch_loss/train': 0.861418925691396}
+12/26/2021 08:34:35 - INFO - codeparrot_training - Step 25710: {'lr': 0.0002401899682438355, 'samples': 13164032, 'steps': 25710, 'batch_loss/train': 0.7006671763956547}
+12/26/2021 08:34:46 - INFO - codeparrot_training - Step 25711: {'lr': 0.00024017421726972305, 'samples': 13164544, 'steps': 25711, 'batch_loss/train': 0.7296668440103531}
+12/26/2021 08:34:58 - INFO - codeparrot_training - Step 25712: {'lr': 0.00024015846633467416, 'samples': 13165056, 'steps': 25712, 'batch_loss/train': 0.7852527648210526}
+12/26/2021 08:35:09 - INFO - codeparrot_training - Step 25713: {'lr': 0.00024014271543875145, 'samples': 13165568, 'steps': 25713, 'batch_loss/train': 0.8590681655332446}
+12/26/2021 08:35:20 - INFO - codeparrot_training - Step 25714: {'lr': 0.00024012696458201765, 'samples': 13166080, 'steps': 25714, 'batch_loss/train': 0.7225767835043371}
+12/26/2021 08:35:32 - INFO - codeparrot_training - Step 25715: {'lr': 0.00024011121376453528, 'samples': 13166592, 'steps': 25715, 'batch_loss/train': 0.7293378482572734}
+12/26/2021 08:35:42 - INFO - codeparrot_training - Step 25716: {'lr': 0.00024009546298636706, 'samples': 13167104, 'steps': 25716, 'batch_loss/train': 0.7592993788421154}
+12/26/2021 08:35:53 - INFO - codeparrot_training - Step 25717: {'lr': 0.0002400797122475754, 'samples': 13167616, 'steps': 25717, 'batch_loss/train': 0.7924859141930938}
+12/26/2021 08:36:03 - INFO - codeparrot_training - Step 25718: {'lr': 0.00024006396154822312, 'samples': 13168128, 'steps': 25718, 'batch_loss/train': 0.6861547706648707}
+12/26/2021 08:36:15 - INFO - codeparrot_training - Step 25719: {'lr': 0.00024004821088837288, 'samples': 13168640, 'steps': 25719, 'batch_loss/train': 0.7934864368289709}
+12/26/2021 08:36:26 - INFO - codeparrot_training - Step 25720: {'lr': 0.00024003246026808715, 'samples': 13169152, 'steps': 25720, 'batch_loss/train': 0.7968322103843093}
+12/26/2021 08:36:37 - INFO - codeparrot_training - Step 25721: {'lr': 0.00024001670968742847, 'samples': 13169664, 'steps': 25721, 'batch_loss/train': 0.7081986400298774}
+12/26/2021 08:36:50 - INFO - codeparrot_training - Step 25722: {'lr': 0.00024000095914645966, 'samples': 13170176, 'steps': 25722, 'batch_loss/train': 0.7227232349105179}
+12/26/2021 08:37:00 - INFO - codeparrot_training - Step 25723: {'lr': 0.00023998520864524332, 'samples': 13170688, 'steps': 25723, 'batch_loss/train': 0.6848343564197421}
+12/26/2021 08:37:11 - INFO - codeparrot_training - Step 25724: {'lr': 0.0002399694581838419, 'samples': 13171200, 'steps': 25724, 'batch_loss/train': 0.7336029461584985}
+12/26/2021 08:37:23 - INFO - codeparrot_training - Step 25725: {'lr': 0.00023995370776231818, 'samples': 13171712, 'steps': 25725, 'batch_loss/train': 0.6575454506091774}
+12/26/2021 08:37:34 - INFO - codeparrot_training - Step 25726: {'lr': 0.00023993795738073477, 'samples': 13172224, 'steps': 25726, 'batch_loss/train': 0.7492589924950153}
+12/26/2021 08:37:45 - INFO - codeparrot_training - Step 25727: {'lr': 0.00023992220703915418, 'samples': 13172736, 'steps': 25727, 'batch_loss/train': 0.7203901568427682}
+12/26/2021 08:37:57 - INFO - codeparrot_training - Step 25728: {'lr': 0.00023990645673763905, 'samples': 13173248, 'steps': 25728, 'batch_loss/train': 0.7796481673722155}
+12/26/2021 08:38:08 - INFO - codeparrot_training - Step 25729: {'lr': 0.00023989070647625213, 'samples': 13173760, 'steps': 25729, 'batch_loss/train': 0.7440155036747456}
+12/26/2021 08:38:19 - INFO - codeparrot_training - Step 25730: {'lr': 0.000239874956255056, 'samples': 13174272, 'steps': 25730, 'batch_loss/train': 0.7862240625545382}
+12/26/2021 08:38:29 - INFO - codeparrot_training - Step 25731: {'lr': 0.0002398592060741131, 'samples': 13174784, 'steps': 25731, 'batch_loss/train': 0.7974992776289582}
+12/26/2021 08:38:41 - INFO - codeparrot_training - Step 25732: {'lr': 0.00023984345593348614, 'samples': 13175296, 'steps': 25732, 'batch_loss/train': 0.6980411889962852}
+12/26/2021 08:38:52 - INFO - codeparrot_training - Step 25733: {'lr': 0.00023982770583323785, 'samples': 13175808, 'steps': 25733, 'batch_loss/train': 0.7091995552182198}
+12/26/2021 08:39:03 - INFO - codeparrot_training - Step 25734: {'lr': 0.00023981195577343073, 'samples': 13176320, 'steps': 25734, 'batch_loss/train': 0.872275173664093}
+12/26/2021 08:39:15 - INFO - codeparrot_training - Step 25735: {'lr': 0.00023979620575412735, 'samples': 13176832, 'steps': 25735, 'batch_loss/train': 0.7051110533066094}
+12/26/2021 08:39:25 - INFO - codeparrot_training - Step 25736: {'lr': 0.00023978045577539053, 'samples': 13177344, 'steps': 25736, 'batch_loss/train': 0.7495652941288427}
+12/26/2021 08:39:36 - INFO - codeparrot_training - Step 25737: {'lr': 0.00023976470583728272, 'samples': 13177856, 'steps': 25737, 'batch_loss/train': 1.0153612131252885}
+12/26/2021 08:39:49 - INFO - codeparrot_training - Step 25738: {'lr': 0.00023974895593986662, 'samples': 13178368, 'steps': 25738, 'batch_loss/train': 0.7433601086959243}
+12/26/2021 08:39:59 - INFO - codeparrot_training - Step 25739: {'lr': 0.00023973320608320468, 'samples': 13178880, 'steps': 25739, 'batch_loss/train': 0.6806671074591577}
+12/26/2021 08:40:10 - INFO - codeparrot_training - Step 25740: {'lr': 0.00023971745626735975, 'samples': 13179392, 'steps': 25740, 'batch_loss/train': 1.0124928411096334}
+12/26/2021 08:40:21 - INFO - codeparrot_training - Step 25741: {'lr': 0.00023970170649239428, 'samples': 13179904, 'steps': 25741, 'batch_loss/train': 0.7584285056218505}
+12/26/2021 08:40:33 - INFO - codeparrot_training - Step 25742: {'lr': 0.000239685956758371, 'samples': 13180416, 'steps': 25742, 'batch_loss/train': 1.0049521503970027}
+12/26/2021 08:40:44 - INFO - codeparrot_training - Step 25743: {'lr': 0.00023967020706535235, 'samples': 13180928, 'steps': 25743, 'batch_loss/train': 0.7149849545676261}
+12/26/2021 08:40:54 - INFO - codeparrot_training - Step 25744: {'lr': 0.00023965445741340108, 'samples': 13181440, 'steps': 25744, 'batch_loss/train': 0.835332652553916}
+12/26/2021 08:41:07 - INFO - codeparrot_training - Step 25745: {'lr': 0.00023963870780257986, 'samples': 13181952, 'steps': 25745, 'batch_loss/train': 0.7892172420397401}
+12/26/2021 08:41:17 - INFO - codeparrot_training - Step 25746: {'lr': 0.00023962295823295115, 'samples': 13182464, 'steps': 25746, 'batch_loss/train': 0.6809890040312894}
+12/26/2021 08:41:28 - INFO - codeparrot_training - Step 25747: {'lr': 0.00023960720870457765, 'samples': 13182976, 'steps': 25747, 'batch_loss/train': 0.6467705350369215}
+12/26/2021 08:41:40 - INFO - codeparrot_training - Step 25748: {'lr': 0.00023959145921752202, 'samples': 13183488, 'steps': 25748, 'batch_loss/train': 0.7079080189578235}
+12/26/2021 08:41:51 - INFO - codeparrot_training - Step 25749: {'lr': 0.0002395757097718468, 'samples': 13184000, 'steps': 25749, 'batch_loss/train': 0.7931945724412799}
+12/26/2021 08:42:02 - INFO - codeparrot_training - Step 25750: {'lr': 0.00023955996036761454, 'samples': 13184512, 'steps': 25750, 'batch_loss/train': 0.7731549628078938}
+12/26/2021 08:42:12 - INFO - codeparrot_training - Step 25751: {'lr': 0.000239544211004888, 'samples': 13185024, 'steps': 25751, 'batch_loss/train': 0.7282419791445136}
+12/26/2021 08:42:25 - INFO - codeparrot_training - Step 25752: {'lr': 0.00023952846168372977, 'samples': 13185536, 'steps': 25752, 'batch_loss/train': 0.851398472674191}
+12/26/2021 08:42:35 - INFO - codeparrot_training - Step 25753: {'lr': 0.00023951271240420235, 'samples': 13186048, 'steps': 25753, 'batch_loss/train': 0.7699234905885532}
+12/26/2021 08:42:46 - INFO - codeparrot_training - Step 25754: {'lr': 0.0002394969631663684, 'samples': 13186560, 'steps': 25754, 'batch_loss/train': 0.6961660687811673}
+12/26/2021 08:42:58 - INFO - codeparrot_training - Step 25755: {'lr': 0.00023948121397029067, 'samples': 13187072, 'steps': 25755, 'batch_loss/train': 0.487789791659452}
+12/26/2021 08:43:09 - INFO - codeparrot_training - Step 25756: {'lr': 0.00023946546481603165, 'samples': 13187584, 'steps': 25756, 'batch_loss/train': 0.9084380222484469}
+12/26/2021 08:43:19 - INFO - codeparrot_training - Step 25757: {'lr': 0.0002394497157036539, 'samples': 13188096, 'steps': 25757, 'batch_loss/train': 0.7611568029969931}
+12/26/2021 08:43:33 - INFO - codeparrot_training - Step 25758: {'lr': 0.00023943396663322005, 'samples': 13188608, 'steps': 25758, 'batch_loss/train': 0.7315767034888268}
+12/26/2021 08:43:43 - INFO - codeparrot_training - Step 25759: {'lr': 0.00023941821760479289, 'samples': 13189120, 'steps': 25759, 'batch_loss/train': 0.7697617029771209}
+12/26/2021 08:43:54 - INFO - codeparrot_training - Step 25760: {'lr': 0.00023940246861843482, 'samples': 13189632, 'steps': 25760, 'batch_loss/train': 0.630233125470113}
+12/26/2021 08:44:05 - INFO - codeparrot_training - Step 25761: {'lr': 0.00023938671967420848, 'samples': 13190144, 'steps': 25761, 'batch_loss/train': 0.5011241778265685}
+12/26/2021 08:44:17 - INFO - codeparrot_training - Step 25762: {'lr': 0.00023937097077217665, 'samples': 13190656, 'steps': 25762, 'batch_loss/train': 0.6652533113956451}
+12/26/2021 08:44:28 - INFO - codeparrot_training - Step 25763: {'lr': 0.00023935522191240177, 'samples': 13191168, 'steps': 25763, 'batch_loss/train': 0.6427395481150597}
+12/26/2021 08:44:38 - INFO - codeparrot_training - Step 25764: {'lr': 0.00023933947309494655, 'samples': 13191680, 'steps': 25764, 'batch_loss/train': 0.807606347836554}
+12/26/2021 08:44:50 - INFO - codeparrot_training - Step 25765: {'lr': 0.00023932372431987347, 'samples': 13192192, 'steps': 25765, 'batch_loss/train': 0.7295577796176076}
+12/26/2021 08:45:01 - INFO - codeparrot_training - Step 25766: {'lr': 0.00023930797558724534, 'samples': 13192704, 'steps': 25766, 'batch_loss/train': 0.7307689823210239}
+12/26/2021 08:45:12 - INFO - codeparrot_training - Step 25767: {'lr': 0.0002392922268971246, 'samples': 13193216, 'steps': 25767, 'batch_loss/train': 0.5970146679319441}
+12/26/2021 08:45:24 - INFO - codeparrot_training - Step 25768: {'lr': 0.000239276478249574, 'samples': 13193728, 'steps': 25768, 'batch_loss/train': 0.7473883754573762}
+12/26/2021 08:45:35 - INFO - codeparrot_training - Step 25769: {'lr': 0.00023926072964465593, 'samples': 13194240, 'steps': 25769, 'batch_loss/train': 0.9938931385986507}
+12/26/2021 08:45:46 - INFO - codeparrot_training - Step 25770: {'lr': 0.0002392449810824332, 'samples': 13194752, 'steps': 25770, 'batch_loss/train': 0.9848896004259586}
+12/26/2021 08:45:56 - INFO - codeparrot_training - Step 25771: {'lr': 0.00023922923256296843, 'samples': 13195264, 'steps': 25771, 'batch_loss/train': 0.8097131885588169}
+12/26/2021 08:46:08 - INFO - codeparrot_training - Step 25772: {'lr': 0.00023921348408632406, 'samples': 13195776, 'steps': 25772, 'batch_loss/train': 0.772377263288945}
+12/26/2021 08:46:19 - INFO - codeparrot_training - Step 25773: {'lr': 0.00023919773565256287, 'samples': 13196288, 'steps': 25773, 'batch_loss/train': 0.7120174127630889}
+12/26/2021 08:46:30 - INFO - codeparrot_training - Step 25774: {'lr': 0.00023918198726174744, 'samples': 13196800, 'steps': 25774, 'batch_loss/train': 0.723727410659194}
+12/26/2021 08:46:42 - INFO - codeparrot_training - Step 25775: {'lr': 0.00023916623891394036, 'samples': 13197312, 'steps': 25775, 'batch_loss/train': 0.7646713042631745}
+12/26/2021 08:46:52 - INFO - codeparrot_training - Step 25776: {'lr': 0.00023915049060920407, 'samples': 13197824, 'steps': 25776, 'batch_loss/train': 0.7677471470087767}
+12/26/2021 08:47:03 - INFO - codeparrot_training - Step 25777: {'lr': 0.00023913474234760144, 'samples': 13198336, 'steps': 25777, 'batch_loss/train': 0.7229087149025872}
+12/26/2021 08:47:13 - INFO - codeparrot_training - Step 25778: {'lr': 0.000239118994129195, 'samples': 13198848, 'steps': 25778, 'batch_loss/train': 0.7394619351252913}
+12/26/2021 08:47:26 - INFO - codeparrot_training - Step 25779: {'lr': 0.00023910324595404728, 'samples': 13199360, 'steps': 25779, 'batch_loss/train': 0.7986871553584933}
+12/26/2021 08:47:36 - INFO - codeparrot_training - Step 25780: {'lr': 0.0002390874978222209, 'samples': 13199872, 'steps': 25780, 'batch_loss/train': 0.8150971699506044}
+12/26/2021 08:47:47 - INFO - codeparrot_training - Step 25781: {'lr': 0.0002390717497337786, 'samples': 13200384, 'steps': 25781, 'batch_loss/train': 0.8640148262493312}
+12/26/2021 08:47:59 - INFO - codeparrot_training - Step 25782: {'lr': 0.00023905600168878287, 'samples': 13200896, 'steps': 25782, 'batch_loss/train': 0.7916410621255636}
+12/26/2021 08:48:10 - INFO - codeparrot_training - Step 25783: {'lr': 0.00023904025368729634, 'samples': 13201408, 'steps': 25783, 'batch_loss/train': 0.8282918771728873}
+12/26/2021 08:48:21 - INFO - codeparrot_training - Step 25784: {'lr': 0.00023902450572938156, 'samples': 13201920, 'steps': 25784, 'batch_loss/train': 0.7876042947173119}
+12/26/2021 08:48:33 - INFO - codeparrot_training - Step 25785: {'lr': 0.0002390087578151013, 'samples': 13202432, 'steps': 25785, 'batch_loss/train': 0.8113891645334661}
+12/26/2021 08:48:43 - INFO - codeparrot_training - Step 25786: {'lr': 0.00023899300994451804, 'samples': 13202944, 'steps': 25786, 'batch_loss/train': 0.77538335416466}
+12/26/2021 08:48:54 - INFO - codeparrot_training - Step 25787: {'lr': 0.00023897726211769433, 'samples': 13203456, 'steps': 25787, 'batch_loss/train': 0.7206165418028831}
+12/26/2021 08:49:07 - INFO - codeparrot_training - Step 25788: {'lr': 0.00023896151433469296, 'samples': 13203968, 'steps': 25788, 'batch_loss/train': 0.7424169920850545}
+12/26/2021 08:49:17 - INFO - codeparrot_training - Step 25789: {'lr': 0.00023894576659557644, 'samples': 13204480, 'steps': 25789, 'batch_loss/train': 0.7823118790984154}
+12/26/2021 08:49:28 - INFO - codeparrot_training - Step 25790: {'lr': 0.00023893001890040738, 'samples': 13204992, 'steps': 25790, 'batch_loss/train': 0.8082421176368371}
+12/26/2021 08:49:39 - INFO - codeparrot_training - Step 25791: {'lr': 0.00023891427124924828, 'samples': 13205504, 'steps': 25791, 'batch_loss/train': 0.7648593168705702}
+12/26/2021 08:49:51 - INFO - codeparrot_training - Step 25792: {'lr': 0.00023889852364216194, 'samples': 13206016, 'steps': 25792, 'batch_loss/train': 0.8382271435111761}
+12/26/2021 08:50:01 - INFO - codeparrot_training - Step 25793: {'lr': 0.0002388827760792109, 'samples': 13206528, 'steps': 25793, 'batch_loss/train': 0.7468791650608182}
+12/26/2021 08:50:12 - INFO - codeparrot_training - Step 25794: {'lr': 0.00023886702856045777, 'samples': 13207040, 'steps': 25794, 'batch_loss/train': 0.72382235317491}
+12/26/2021 08:50:24 - INFO - codeparrot_training - Step 25795: {'lr': 0.00023885128108596502, 'samples': 13207552, 'steps': 25795, 'batch_loss/train': 0.7529358807951212}
+12/26/2021 08:50:35 - INFO - codeparrot_training - Step 25796: {'lr': 0.00023883553365579538, 'samples': 13208064, 'steps': 25796, 'batch_loss/train': 0.7787098162807524}
+12/26/2021 08:50:45 - INFO - codeparrot_training - Step 25797: {'lr': 0.00023881978627001156, 'samples': 13208576, 'steps': 25797, 'batch_loss/train': 0.8660402758978307}
+12/26/2021 08:50:58 - INFO - codeparrot_training - Step 25798: {'lr': 0.0002388040389286759, 'samples': 13209088, 'steps': 25798, 'batch_loss/train': 0.7861004546284676}
+12/26/2021 08:51:09 - INFO - codeparrot_training - Step 25799: {'lr': 0.00023878829163185121, 'samples': 13209600, 'steps': 25799, 'batch_loss/train': 0.7302084341645241}
+12/26/2021 08:51:19 - INFO - codeparrot_training - Step 25800: {'lr': 0.0002387725443796001, 'samples': 13210112, 'steps': 25800, 'batch_loss/train': 0.7636531237512827}
+12/26/2021 08:51:30 - INFO - codeparrot_training - Step 25801: {'lr': 0.00023875679717198512, 'samples': 13210624, 'steps': 25801, 'batch_loss/train': 0.7649942100979388}
+12/26/2021 08:51:42 - INFO - codeparrot_training - Step 25802: {'lr': 0.00023874105000906877, 'samples': 13211136, 'steps': 25802, 'batch_loss/train': 0.782144655007869}
+12/26/2021 08:51:53 - INFO - codeparrot_training - Step 25803: {'lr': 0.00023872530289091375, 'samples': 13211648, 'steps': 25803, 'batch_loss/train': 0.8992117829620838}
+12/26/2021 08:52:03 - INFO - codeparrot_training - Step 25804: {'lr': 0.0002387095558175828, 'samples': 13212160, 'steps': 25804, 'batch_loss/train': 0.7227761764079332}
+12/26/2021 08:52:16 - INFO - codeparrot_training - Step 25805: {'lr': 0.00023869380878913828, 'samples': 13212672, 'steps': 25805, 'batch_loss/train': 0.7792542732786387}
+12/26/2021 08:52:26 - INFO - codeparrot_training - Step 25806: {'lr': 0.0002386780618056429, 'samples': 13213184, 'steps': 25806, 'batch_loss/train': 0.7791832450311631}
+12/26/2021 08:52:37 - INFO - codeparrot_training - Step 25807: {'lr': 0.00023866231486715934, 'samples': 13213696, 'steps': 25807, 'batch_loss/train': 0.7875376876909286}
+12/26/2021 08:52:49 - INFO - codeparrot_training - Step 25808: {'lr': 0.0002386465679737501, 'samples': 13214208, 'steps': 25808, 'batch_loss/train': 0.731395436450839}
+12/26/2021 08:53:00 - INFO - codeparrot_training - Step 25809: {'lr': 0.00023863082112547776, 'samples': 13214720, 'steps': 25809, 'batch_loss/train': 0.8849211251363158}
+12/26/2021 08:53:10 - INFO - codeparrot_training - Step 25810: {'lr': 0.0002386150743224051, 'samples': 13215232, 'steps': 25810, 'batch_loss/train': 0.7273187150713056}
+12/26/2021 08:53:22 - INFO - codeparrot_training - Step 25811: {'lr': 0.0002385993275645946, 'samples': 13215744, 'steps': 25811, 'batch_loss/train': 0.719786380417645}
+12/26/2021 08:53:33 - INFO - codeparrot_training - Step 25812: {'lr': 0.00023858358085210882, 'samples': 13216256, 'steps': 25812, 'batch_loss/train': 0.7613806575536728}
+12/26/2021 08:53:44 - INFO - codeparrot_training - Step 25813: {'lr': 0.00023856783418501034, 'samples': 13216768, 'steps': 25813, 'batch_loss/train': 0.8490590648725629}
+12/26/2021 08:53:54 - INFO - codeparrot_training - Step 25814: {'lr': 0.00023855208756336197, 'samples': 13217280, 'steps': 25814, 'batch_loss/train': 0.7466192059218884}
+12/26/2021 08:54:06 - INFO - codeparrot_training - Step 25815: {'lr': 0.0002385363409872261, 'samples': 13217792, 'steps': 25815, 'batch_loss/train': 0.8462620107457042}
+12/26/2021 08:54:17 - INFO - codeparrot_training - Step 25816: {'lr': 0.00023852059445666547, 'samples': 13218304, 'steps': 25816, 'batch_loss/train': 0.8822855362668633}
+12/26/2021 08:54:28 - INFO - codeparrot_training - Step 25817: {'lr': 0.00023850484797174253, 'samples': 13218816, 'steps': 25817, 'batch_loss/train': 0.6205046270042658}
+12/26/2021 08:54:40 - INFO - codeparrot_training - Step 25818: {'lr': 0.00023848910153252, 'samples': 13219328, 'steps': 25818, 'batch_loss/train': 0.6986976908519864}
+12/26/2021 08:54:51 - INFO - codeparrot_training - Step 25819: {'lr': 0.00023847335513906053, 'samples': 13219840, 'steps': 25819, 'batch_loss/train': 0.7838326701894403}
+12/26/2021 08:55:02 - INFO - codeparrot_training - Step 25820: {'lr': 0.00023845760879142664, 'samples': 13220352, 'steps': 25820, 'batch_loss/train': 0.7708400280680507}
+12/26/2021 08:55:14 - INFO - codeparrot_training - Step 25821: {'lr': 0.00023844186248968082, 'samples': 13220864, 'steps': 25821, 'batch_loss/train': 0.5903663435019553}
+12/26/2021 08:55:24 - INFO - codeparrot_training - Step 25822: {'lr': 0.00023842611623388585, 'samples': 13221376, 'steps': 25822, 'batch_loss/train': 0.7859106743708253}
+12/26/2021 08:55:35 - INFO - codeparrot_training - Step 25823: {'lr': 0.00023841037002410434, 'samples': 13221888, 'steps': 25823, 'batch_loss/train': 0.7558905808255076}
+12/26/2021 08:55:46 - INFO - codeparrot_training - Step 25824: {'lr': 0.0002383946238603987, 'samples': 13222400, 'steps': 25824, 'batch_loss/train': 0.7717930786311626}
+12/26/2021 08:55:58 - INFO - codeparrot_training - Step 25825: {'lr': 0.00023837887774283173, 'samples': 13222912, 'steps': 25825, 'batch_loss/train': 0.8487730016931891}
+12/26/2021 08:56:08 - INFO - codeparrot_training - Step 25826: {'lr': 0.000238363131671466, 'samples': 13223424, 'steps': 25826, 'batch_loss/train': 0.7383453054353595}
+12/26/2021 08:56:19 - INFO - codeparrot_training - Step 25827: {'lr': 0.000238347385646364, 'samples': 13223936, 'steps': 25827, 'batch_loss/train': 0.7573372656479478}
+12/26/2021 08:56:32 - INFO - codeparrot_training - Step 25828: {'lr': 0.00023833163966758834, 'samples': 13224448, 'steps': 25828, 'batch_loss/train': 0.7068978953175247}
+12/26/2021 08:56:42 - INFO - codeparrot_training - Step 25829: {'lr': 0.0002383158937352018, 'samples': 13224960, 'steps': 25829, 'batch_loss/train': 0.7752053984440863}
+12/26/2021 08:56:53 - INFO - codeparrot_training - Step 25830: {'lr': 0.0002383001478492668, 'samples': 13225472, 'steps': 25830, 'batch_loss/train': 0.678603402338922}
+12/26/2021 08:57:05 - INFO - codeparrot_training - Step 25831: {'lr': 0.00023828440200984602, 'samples': 13225984, 'steps': 25831, 'batch_loss/train': 0.6619744659401476}
+12/26/2021 08:57:16 - INFO - codeparrot_training - Step 25832: {'lr': 0.00023826865621700194, 'samples': 13226496, 'steps': 25832, 'batch_loss/train': 0.7074231074657291}
+12/26/2021 08:57:26 - INFO - codeparrot_training - Step 25833: {'lr': 0.00023825291047079736, 'samples': 13227008, 'steps': 25833, 'batch_loss/train': 0.7026512789307162}
+12/26/2021 08:57:37 - INFO - codeparrot_training - Step 25834: {'lr': 0.00023823716477129473, 'samples': 13227520, 'steps': 25834, 'batch_loss/train': 0.7529293678235263}
+12/26/2021 08:57:50 - INFO - codeparrot_training - Step 25835: {'lr': 0.00023822141911855664, 'samples': 13228032, 'steps': 25835, 'batch_loss/train': 0.7667767368257046}
+12/26/2021 08:58:00 - INFO - codeparrot_training - Step 25836: {'lr': 0.00023820567351264582, 'samples': 13228544, 'steps': 25836, 'batch_loss/train': 0.7498479164205492}
+12/26/2021 08:58:11 - INFO - codeparrot_training - Step 25837: {'lr': 0.00023818992795362475, 'samples': 13229056, 'steps': 25837, 'batch_loss/train': 0.751997341401875}
+12/26/2021 08:58:23 - INFO - codeparrot_training - Step 25838: {'lr': 0.00023817418244155613, 'samples': 13229568, 'steps': 25838, 'batch_loss/train': 0.8761371374130249}
+12/26/2021 08:58:34 - INFO - codeparrot_training - Step 25839: {'lr': 0.00023815843697650238, 'samples': 13230080, 'steps': 25839, 'batch_loss/train': 0.8168506021611392}
+12/26/2021 08:58:44 - INFO - codeparrot_training - Step 25840: {'lr': 0.00023814269155852634, 'samples': 13230592, 'steps': 25840, 'batch_loss/train': 0.7627277803840116}
+12/26/2021 08:58:57 - INFO - codeparrot_training - Step 25841: {'lr': 0.00023812694618769036, 'samples': 13231104, 'steps': 25841, 'batch_loss/train': 0.6352652052883059}
+12/26/2021 08:59:07 - INFO - codeparrot_training - Step 25842: {'lr': 0.0002381112008640573, 'samples': 13231616, 'steps': 25842, 'batch_loss/train': 0.769064961001277}
+12/26/2021 08:59:18 - INFO - codeparrot_training - Step 25843: {'lr': 0.00023809545558768947, 'samples': 13232128, 'steps': 25843, 'batch_loss/train': 0.7181727685965598}
+12/26/2021 08:59:28 - INFO - codeparrot_training - Step 25844: {'lr': 0.00023807971035864968, 'samples': 13232640, 'steps': 25844, 'batch_loss/train': 0.8240876458585262}
+12/26/2021 08:59:41 - INFO - codeparrot_training - Step 25845: {'lr': 0.0002380639651770005, 'samples': 13233152, 'steps': 25845, 'batch_loss/train': 0.8549575665965676}
+12/26/2021 08:59:52 - INFO - codeparrot_training - Step 25846: {'lr': 0.00023804822004280435, 'samples': 13233664, 'steps': 25846, 'batch_loss/train': 0.7134215317200869}
+12/26/2021 09:00:02 - INFO - codeparrot_training - Step 25847: {'lr': 0.00023803247495612413, 'samples': 13234176, 'steps': 25847, 'batch_loss/train': 0.8119685752317309}
+12/26/2021 09:00:14 - INFO - codeparrot_training - Step 25848: {'lr': 0.0002380167299170222, 'samples': 13234688, 'steps': 25848, 'batch_loss/train': 0.7307017343118787}
+12/26/2021 09:00:25 - INFO - codeparrot_training - Step 25849: {'lr': 0.00023800098492556126, 'samples': 13235200, 'steps': 25849, 'batch_loss/train': 0.7084963535889983}
+12/26/2021 09:00:36 - INFO - codeparrot_training - Step 25850: {'lr': 0.00023798523998180376, 'samples': 13235712, 'steps': 25850, 'batch_loss/train': 0.7355574136599898}
+12/26/2021 09:00:48 - INFO - codeparrot_training - Step 25851: {'lr': 0.0002379694950858125, 'samples': 13236224, 'steps': 25851, 'batch_loss/train': 0.813627430703491}
+12/26/2021 09:00:58 - INFO - codeparrot_training - Step 25852: {'lr': 0.00023795375023765, 'samples': 13236736, 'steps': 25852, 'batch_loss/train': 0.7095377296209335}
+12/26/2021 09:01:09 - INFO - codeparrot_training - Step 25853: {'lr': 0.0002379380054373788, 'samples': 13237248, 'steps': 25853, 'batch_loss/train': 0.7851838804781437}
+12/26/2021 09:01:21 - INFO - codeparrot_training - Step 25854: {'lr': 0.0002379222606850615, 'samples': 13237760, 'steps': 25854, 'batch_loss/train': 0.7029807581566274}
+12/26/2021 09:01:32 - INFO - codeparrot_training - Step 25855: {'lr': 0.00023790651598076085, 'samples': 13238272, 'steps': 25855, 'batch_loss/train': 0.7082315306179225}
+12/26/2021 09:01:42 - INFO - codeparrot_training - Step 25856: {'lr': 0.0002378907713245393, 'samples': 13238784, 'steps': 25856, 'batch_loss/train': 0.7536236932501197}
+12/26/2021 09:01:53 - INFO - codeparrot_training - Step 25857: {'lr': 0.00023787502671645942, 'samples': 13239296, 'steps': 25857, 'batch_loss/train': 0.9224378038197756}
+12/26/2021 09:02:06 - INFO - codeparrot_training - Step 25858: {'lr': 0.0002378592821565838, 'samples': 13239808, 'steps': 25858, 'batch_loss/train': 0.8491421900689602}
+12/26/2021 09:02:16 - INFO - codeparrot_training - Step 25859: {'lr': 0.0002378435376449752, 'samples': 13240320, 'steps': 25859, 'batch_loss/train': 0.7726580509915948}
+12/26/2021 09:02:27 - INFO - codeparrot_training - Step 25860: {'lr': 0.00023782779318169605, 'samples': 13240832, 'steps': 25860, 'batch_loss/train': 0.8032284388318658}
+12/26/2021 09:02:39 - INFO - codeparrot_training - Step 25861: {'lr': 0.00023781204876680896, 'samples': 13241344, 'steps': 25861, 'batch_loss/train': 0.7835508196149021}
+12/26/2021 09:02:50 - INFO - codeparrot_training - Step 25862: {'lr': 0.00023779630440037665, 'samples': 13241856, 'steps': 25862, 'batch_loss/train': 0.6777599202468991}
+12/26/2021 09:03:00 - INFO - codeparrot_training - Step 25863: {'lr': 0.00023778056008246157, 'samples': 13242368, 'steps': 25863, 'batch_loss/train': 0.6783703099936247}
+12/26/2021 09:03:13 - INFO - codeparrot_training - Step 25864: {'lr': 0.00023776481581312643, 'samples': 13242880, 'steps': 25864, 'batch_loss/train': 0.7382772710407153}
+12/26/2021 09:03:24 - INFO - codeparrot_training - Step 25865: {'lr': 0.00023774907159243363, 'samples': 13243392, 'steps': 25865, 'batch_loss/train': 0.7440869733691216}
+12/26/2021 09:03:35 - INFO - codeparrot_training - Step 25866: {'lr': 0.00023773332742044607, 'samples': 13243904, 'steps': 25866, 'batch_loss/train': 0.6469409526325762}
+12/26/2021 09:03:45 - INFO - codeparrot_training - Step 25867: {'lr': 0.00023771758329722606, 'samples': 13244416, 'steps': 25867, 'batch_loss/train': 0.8313260339200497}
+12/26/2021 09:03:57 - INFO - codeparrot_training - Step 25868: {'lr': 0.00023770183922283637, 'samples': 13244928, 'steps': 25868, 'batch_loss/train': 0.6554475215962157}
+12/26/2021 09:04:08 - INFO - codeparrot_training - Step 25869: {'lr': 0.0002376860951973394, 'samples': 13245440, 'steps': 25869, 'batch_loss/train': 0.7330520222894847}
+12/26/2021 09:04:19 - INFO - codeparrot_training - Step 25870: {'lr': 0.00023767035122079793, 'samples': 13245952, 'steps': 25870, 'batch_loss/train': 0.7343311649747193}
+12/26/2021 09:04:31 - INFO - codeparrot_training - Step 25871: {'lr': 0.00023765460729327457, 'samples': 13246464, 'steps': 25871, 'batch_loss/train': 0.8181927315890789}
+12/26/2021 09:04:41 - INFO - codeparrot_training - Step 25872: {'lr': 0.0002376388634148317, 'samples': 13246976, 'steps': 25872, 'batch_loss/train': 0.7867958163842559}
+12/26/2021 09:04:52 - INFO - codeparrot_training - Step 25873: {'lr': 0.00023762311958553205, 'samples': 13247488, 'steps': 25873, 'batch_loss/train': 0.7787940478883684}
+12/26/2021 09:05:05 - INFO - codeparrot_training - Step 25874: {'lr': 0.00023760737580543834, 'samples': 13248000, 'steps': 25874, 'batch_loss/train': 0.7329162256792188}
+12/26/2021 09:05:16 - INFO - codeparrot_training - Step 25875: {'lr': 0.000237591632074613, 'samples': 13248512, 'steps': 25875, 'batch_loss/train': 0.712734192609787}
+12/26/2021 09:05:26 - INFO - codeparrot_training - Step 25876: {'lr': 0.0002375758883931185, 'samples': 13249024, 'steps': 25876, 'batch_loss/train': 0.9029819583520293}
+12/26/2021 09:05:37 - INFO - codeparrot_training - Step 25877: {'lr': 0.00023756014476101765, 'samples': 13249536, 'steps': 25877, 'batch_loss/train': 0.7182923723012209}
+12/26/2021 09:05:49 - INFO - codeparrot_training - Step 25878: {'lr': 0.00023754440117837303, 'samples': 13250048, 'steps': 25878, 'batch_loss/train': 0.7883221320807934}
+12/26/2021 09:06:00 - INFO - codeparrot_training - Step 25879: {'lr': 0.0002375286576452471, 'samples': 13250560, 'steps': 25879, 'batch_loss/train': 0.7622475530952215}
+12/26/2021 09:06:10 - INFO - codeparrot_training - Step 25880: {'lr': 0.00023751291416170247, 'samples': 13251072, 'steps': 25880, 'batch_loss/train': 0.850754152983427}
+12/26/2021 09:06:23 - INFO - codeparrot_training - Step 25881: {'lr': 0.00023749717072780188, 'samples': 13251584, 'steps': 25881, 'batch_loss/train': 0.6617951535154134}
+12/26/2021 09:06:34 - INFO - codeparrot_training - Step 25882: {'lr': 0.0002374814273436078, 'samples': 13252096, 'steps': 25882, 'batch_loss/train': 0.7948501383652911}
+12/26/2021 09:06:44 - INFO - codeparrot_training - Step 25883: {'lr': 0.00023746568400918275, 'samples': 13252608, 'steps': 25883, 'batch_loss/train': 0.7859710892662406}
+12/26/2021 09:06:55 - INFO - codeparrot_training - Step 25884: {'lr': 0.00023744994072458952, 'samples': 13253120, 'steps': 25884, 'batch_loss/train': 0.7411525533534586}
+12/26/2021 09:07:07 - INFO - codeparrot_training - Step 25885: {'lr': 0.0002374341974898906, 'samples': 13253632, 'steps': 25885, 'batch_loss/train': 0.6558716711588204}
+12/26/2021 09:07:18 - INFO - codeparrot_training - Step 25886: {'lr': 0.0002374184543051485, 'samples': 13254144, 'steps': 25886, 'batch_loss/train': 0.7384914932772517}
+12/26/2021 09:07:28 - INFO - codeparrot_training - Step 25887: {'lr': 0.0002374027111704258, 'samples': 13254656, 'steps': 25887, 'batch_loss/train': 0.8627694901078939}
+12/26/2021 09:07:40 - INFO - codeparrot_training - Step 25888: {'lr': 0.0002373869680857853, 'samples': 13255168, 'steps': 25888, 'batch_loss/train': 0.7836383730173111}
+12/26/2021 09:07:51 - INFO - codeparrot_training - Step 25889: {'lr': 0.0002373712250512894, 'samples': 13255680, 'steps': 25889, 'batch_loss/train': 0.8167732283473015}
+12/26/2021 09:08:02 - INFO - codeparrot_training - Step 25890: {'lr': 0.00023735548206700079, 'samples': 13256192, 'steps': 25890, 'batch_loss/train': 0.8129525622352958}
+12/26/2021 09:08:14 - INFO - codeparrot_training - Step 25891: {'lr': 0.0002373397391329819, 'samples': 13256704, 'steps': 25891, 'batch_loss/train': 0.7314910553395748}
+12/26/2021 09:08:24 - INFO - codeparrot_training - Step 25892: {'lr': 0.00023732399624929555, 'samples': 13257216, 'steps': 25892, 'batch_loss/train': 0.7945600261446089}
+12/26/2021 09:08:35 - INFO - codeparrot_training - Step 25893: {'lr': 0.00023730825341600412, 'samples': 13257728, 'steps': 25893, 'batch_loss/train': 0.8338272827677429}
+12/26/2021 09:08:48 - INFO - codeparrot_training - Step 25894: {'lr': 0.00023729251063317033, 'samples': 13258240, 'steps': 25894, 'batch_loss/train': 0.9272001590579748}
+12/26/2021 09:08:58 - INFO - codeparrot_training - Step 25895: {'lr': 0.00023727676790085664, 'samples': 13258752, 'steps': 25895, 'batch_loss/train': 0.7245870530605316}
+12/26/2021 09:09:09 - INFO - codeparrot_training - Step 25896: {'lr': 0.00023726102521912576, 'samples': 13259264, 'steps': 25896, 'batch_loss/train': 0.7669850520323962}
+12/26/2021 09:09:19 - INFO - codeparrot_training - Step 25897: {'lr': 0.0002372452825880403, 'samples': 13259776, 'steps': 25897, 'batch_loss/train': 0.7699030581861734}
+12/26/2021 09:09:34 - INFO - codeparrot_training - Step 25898: {'lr': 0.00023722954000766268, 'samples': 13260288, 'steps': 25898, 'batch_loss/train': 0.7679054755717516}
+12/26/2021 09:09:45 - INFO - codeparrot_training - Step 25899: {'lr': 0.0002372137974780556, 'samples': 13260800, 'steps': 25899, 'batch_loss/train': 0.7533565675839782}
+12/26/2021 09:09:55 - INFO - codeparrot_training - Step 25900: {'lr': 0.0002371980549992817, 'samples': 13261312, 'steps': 25900, 'batch_loss/train': 0.697374670766294}
+12/26/2021 09:10:07 - INFO - codeparrot_training - Step 25901: {'lr': 0.00023718231257140355, 'samples': 13261824, 'steps': 25901, 'batch_loss/train': 0.7607488394714892}
+12/26/2021 09:10:18 - INFO - codeparrot_training - Step 25902: {'lr': 0.0002371665701944835, 'samples': 13262336, 'steps': 25902, 'batch_loss/train': 0.670893486472778}
+12/26/2021 09:10:29 - INFO - codeparrot_training - Step 25903: {'lr': 0.00023715082786858444, 'samples': 13262848, 'steps': 25903, 'batch_loss/train': 0.7957008937373757}
+12/26/2021 09:10:41 - INFO - codeparrot_training - Step 25904: {'lr': 0.00023713508559376886, 'samples': 13263360, 'steps': 25904, 'batch_loss/train': 0.7625911957584321}
+12/26/2021 09:10:51 - INFO - codeparrot_training - Step 25905: {'lr': 0.00023711934337009927, 'samples': 13263872, 'steps': 25905, 'batch_loss/train': 0.7578310491517186}
+12/26/2021 09:11:02 - INFO - codeparrot_training - Step 25906: {'lr': 0.00023710360119763826, 'samples': 13264384, 'steps': 25906, 'batch_loss/train': 0.6637306208722293}
+12/26/2021 09:11:16 - INFO - codeparrot_training - Step 25907: {'lr': 0.00023708785907644857, 'samples': 13264896, 'steps': 25907, 'batch_loss/train': 0.7968278415501118}
+12/26/2021 09:11:27 - INFO - codeparrot_training - Step 25908: {'lr': 0.00023707211700659263, 'samples': 13265408, 'steps': 25908, 'batch_loss/train': 0.7519544120877981}
+12/26/2021 09:11:37 - INFO - codeparrot_training - Step 25909: {'lr': 0.000237056374988133, 'samples': 13265920, 'steps': 25909, 'batch_loss/train': 0.674628120730631}
+12/26/2021 09:11:48 - INFO - codeparrot_training - Step 25910: {'lr': 0.00023704063302113244, 'samples': 13266432, 'steps': 25910, 'batch_loss/train': 0.8337775748805143}
+12/26/2021 09:12:00 - INFO - codeparrot_training - Step 25911: {'lr': 0.00023702489110565345, 'samples': 13266944, 'steps': 25911, 'batch_loss/train': 0.7792388579109684}
+12/26/2021 09:12:10 - INFO - codeparrot_training - Step 25912: {'lr': 0.0002370091492417585, 'samples': 13267456, 'steps': 25912, 'batch_loss/train': 0.675048332195729}
+12/26/2021 09:12:21 - INFO - codeparrot_training - Step 25913: {'lr': 0.00023699340742951025, 'samples': 13267968, 'steps': 25913, 'batch_loss/train': 0.8270281180739403}
+12/26/2021 09:12:35 - INFO - codeparrot_training - Step 25914: {'lr': 0.00023697766566897142, 'samples': 13268480, 'steps': 25914, 'batch_loss/train': 0.7361671247635968}
+12/26/2021 09:12:46 - INFO - codeparrot_training - Step 25915: {'lr': 0.00023696192396020438, 'samples': 13268992, 'steps': 25915, 'batch_loss/train': 0.7710491635371}
+12/26/2021 09:12:56 - INFO - codeparrot_training - Step 25916: {'lr': 0.0002369461823032719, 'samples': 13269504, 'steps': 25916, 'batch_loss/train': 0.7704849233850837}
+12/26/2021 09:13:08 - INFO - codeparrot_training - Step 25917: {'lr': 0.00023693044069823636, 'samples': 13270016, 'steps': 25917, 'batch_loss/train': 0.7608193596825004}
+12/26/2021 09:13:19 - INFO - codeparrot_training - Step 25918: {'lr': 0.00023691469914516048, 'samples': 13270528, 'steps': 25918, 'batch_loss/train': 0.7649617120623589}
+12/26/2021 09:13:29 - INFO - codeparrot_training - Step 25919: {'lr': 0.00023689895764410687, 'samples': 13271040, 'steps': 25919, 'batch_loss/train': 0.7301172437146306}
+12/26/2021 09:13:40 - INFO - codeparrot_training - Step 25920: {'lr': 0.00023688321619513798, 'samples': 13271552, 'steps': 25920, 'batch_loss/train': 0.8097895570099354}
+12/26/2021 09:13:52 - INFO - codeparrot_training - Step 25921: {'lr': 0.00023686747479831656, 'samples': 13272064, 'steps': 25921, 'batch_loss/train': 0.7223944151774049}
+12/26/2021 09:14:03 - INFO - codeparrot_training - Step 25922: {'lr': 0.00023685173345370502, 'samples': 13272576, 'steps': 25922, 'batch_loss/train': 0.826703604310751}
+12/26/2021 09:14:13 - INFO - codeparrot_training - Step 25923: {'lr': 0.0002368359921613661, 'samples': 13273088, 'steps': 25923, 'batch_loss/train': 0.7204307373613119}
+12/26/2021 09:14:28 - INFO - codeparrot_training - Step 25924: {'lr': 0.00023682025092136222, 'samples': 13273600, 'steps': 25924, 'batch_loss/train': 0.6914122509770095}
+12/26/2021 09:14:38 - INFO - codeparrot_training - Step 25925: {'lr': 0.0002368045097337561, 'samples': 13274112, 'steps': 25925, 'batch_loss/train': 0.7630156809464097}
+12/26/2021 09:14:49 - INFO - codeparrot_training - Step 25926: {'lr': 0.00023678876859861028, 'samples': 13274624, 'steps': 25926, 'batch_loss/train': 0.8214589990675449}
+12/26/2021 09:15:01 - INFO - codeparrot_training - Step 25927: {'lr': 0.0002367730275159873, 'samples': 13275136, 'steps': 25927, 'batch_loss/train': 0.7388475835323334}
+12/26/2021 09:15:11 - INFO - codeparrot_training - Step 25928: {'lr': 0.00023675728648594977, 'samples': 13275648, 'steps': 25928, 'batch_loss/train': 0.7671865252777934}
+12/26/2021 09:15:22 - INFO - codeparrot_training - Step 25929: {'lr': 0.00023674154550856025, 'samples': 13276160, 'steps': 25929, 'batch_loss/train': 0.7363605930004269}
+12/26/2021 09:15:33 - INFO - codeparrot_training - Step 25930: {'lr': 0.0002367258045838814, 'samples': 13276672, 'steps': 25930, 'batch_loss/train': 0.7293666321784258}
+12/26/2021 09:15:45 - INFO - codeparrot_training - Step 25931: {'lr': 0.0002367100637119757, 'samples': 13277184, 'steps': 25931, 'batch_loss/train': 0.7950822487473488}
+12/26/2021 09:15:55 - INFO - codeparrot_training - Step 25932: {'lr': 0.0002366943228929057, 'samples': 13277696, 'steps': 25932, 'batch_loss/train': 0.6805221736431122}
+12/26/2021 09:16:06 - INFO - codeparrot_training - Step 25933: {'lr': 0.00023667858212673415, 'samples': 13278208, 'steps': 25933, 'batch_loss/train': 0.6444833055138588}
+12/26/2021 09:16:18 - INFO - codeparrot_training - Step 25934: {'lr': 0.00023666284141352352, 'samples': 13278720, 'steps': 25934, 'batch_loss/train': 0.7770900866016746}
+12/26/2021 09:16:29 - INFO - codeparrot_training - Step 25935: {'lr': 0.0002366471007533363, 'samples': 13279232, 'steps': 25935, 'batch_loss/train': 0.5607092686695978}
+12/26/2021 09:16:39 - INFO - codeparrot_training - Step 25936: {'lr': 0.00023663136014623527, 'samples': 13279744, 'steps': 25936, 'batch_loss/train': 0.7606370359426364}
+12/26/2021 09:16:54 - INFO - codeparrot_training - Step 25937: {'lr': 0.0002366156195922829, 'samples': 13280256, 'steps': 25937, 'batch_loss/train': 0.7363438149914145}
+12/26/2021 09:17:04 - INFO - codeparrot_training - Step 25938: {'lr': 0.00023659987909154174, 'samples': 13280768, 'steps': 25938, 'batch_loss/train': 0.8617282379418612}
+12/26/2021 09:17:15 - INFO - codeparrot_training - Step 25939: {'lr': 0.00023658413864407435, 'samples': 13281280, 'steps': 25939, 'batch_loss/train': 1.059578905813396}
+12/26/2021 09:17:25 - INFO - codeparrot_training - Step 25940: {'lr': 0.00023656839824994346, 'samples': 13281792, 'steps': 25940, 'batch_loss/train': 0.7155702370218933}
+12/26/2021 09:17:38 - INFO - codeparrot_training - Step 25941: {'lr': 0.0002365526579092115, 'samples': 13282304, 'steps': 25941, 'batch_loss/train': 0.7176461787894368}
+12/26/2021 09:17:48 - INFO - codeparrot_training - Step 25942: {'lr': 0.00023653691762194113, 'samples': 13282816, 'steps': 25942, 'batch_loss/train': 0.831014025490731}
+12/26/2021 09:17:59 - INFO - codeparrot_training - Step 25943: {'lr': 0.0002365211773881948, 'samples': 13283328, 'steps': 25943, 'batch_loss/train': 0.7375680790282786}
+12/26/2021 09:18:13 - INFO - codeparrot_training - Step 25944: {'lr': 0.0002365054372080352, 'samples': 13283840, 'steps': 25944, 'batch_loss/train': 0.8618763815611601}
+12/26/2021 09:18:23 - INFO - codeparrot_training - Step 25945: {'lr': 0.00023648969708152498, 'samples': 13284352, 'steps': 25945, 'batch_loss/train': 0.746865161228925}
+12/26/2021 09:18:34 - INFO - codeparrot_training - Step 25946: {'lr': 0.00023647395700872648, 'samples': 13284864, 'steps': 25946, 'batch_loss/train': 0.7379702075850219}
+12/26/2021 09:18:46 - INFO - codeparrot_training - Step 25947: {'lr': 0.00023645821698970256, 'samples': 13285376, 'steps': 25947, 'batch_loss/train': 0.7563121132552624}
+12/26/2021 09:18:57 - INFO - codeparrot_training - Step 25948: {'lr': 0.0002364424770245156, 'samples': 13285888, 'steps': 25948, 'batch_loss/train': 0.7647718004882336}
+12/26/2021 09:19:07 - INFO - codeparrot_training - Step 25949: {'lr': 0.00023642673711322826, 'samples': 13286400, 'steps': 25949, 'batch_loss/train': 0.7338793538510799}
+12/26/2021 09:19:18 - INFO - codeparrot_training - Step 25950: {'lr': 0.00023641099725590301, 'samples': 13286912, 'steps': 25950, 'batch_loss/train': 0.7441261947387829}
+12/26/2021 09:19:30 - INFO - codeparrot_training - Step 25951: {'lr': 0.00023639525745260253, 'samples': 13287424, 'steps': 25951, 'batch_loss/train': 0.8491028444841504}
+12/26/2021 09:19:41 - INFO - codeparrot_training - Step 25952: {'lr': 0.00023637951770338946, 'samples': 13287936, 'steps': 25952, 'batch_loss/train': 1.0665181400254369}
+12/26/2021 09:19:52 - INFO - codeparrot_training - Step 25953: {'lr': 0.0002363637780083262, 'samples': 13288448, 'steps': 25953, 'batch_loss/train': 1.3581379484385252}
+12/26/2021 09:20:05 - INFO - codeparrot_training - Step 25954: {'lr': 0.00023634803836747537, 'samples': 13288960, 'steps': 25954, 'batch_loss/train': 0.7752924566157162}
+12/26/2021 09:20:16 - INFO - codeparrot_training - Step 25955: {'lr': 0.0002363322987808997, 'samples': 13289472, 'steps': 25955, 'batch_loss/train': 0.7328214468434453}
+12/26/2021 09:20:27 - INFO - codeparrot_training - Step 25956: {'lr': 0.00023631655924866163, 'samples': 13289984, 'steps': 25956, 'batch_loss/train': 0.7878120760433376}
+12/26/2021 09:20:39 - INFO - codeparrot_training - Step 25957: {'lr': 0.00023630081977082365, 'samples': 13290496, 'steps': 25957, 'batch_loss/train': 0.8095325324684381}
+12/26/2021 09:20:49 - INFO - codeparrot_training - Step 25958: {'lr': 0.0002362850803474485, 'samples': 13291008, 'steps': 25958, 'batch_loss/train': 0.7506893844110891}
+12/26/2021 09:21:00 - INFO - codeparrot_training - Step 25959: {'lr': 0.00023626934097859873, 'samples': 13291520, 'steps': 25959, 'batch_loss/train': 0.8044554013758898}
+12/26/2021 09:21:11 - INFO - codeparrot_training - Step 25960: {'lr': 0.00023625360166433686, 'samples': 13292032, 'steps': 25960, 'batch_loss/train': 0.7885770266875625}
+12/26/2021 09:21:23 - INFO - codeparrot_training - Step 25961: {'lr': 0.00023623786240472539, 'samples': 13292544, 'steps': 25961, 'batch_loss/train': 0.8216549977660179}
+12/26/2021 09:21:34 - INFO - codeparrot_training - Step 25962: {'lr': 0.0002362221231998271, 'samples': 13293056, 'steps': 25962, 'batch_loss/train': 0.7208397784270346}
+12/26/2021 09:21:44 - INFO - codeparrot_training - Step 25963: {'lr': 0.00023620638404970437, 'samples': 13293568, 'steps': 25963, 'batch_loss/train': 0.7875577285885811}
+12/26/2021 09:21:58 - INFO - codeparrot_training - Step 25964: {'lr': 0.00023619064495441995, 'samples': 13294080, 'steps': 25964, 'batch_loss/train': 0.7668369561433792}
+12/26/2021 09:22:09 - INFO - codeparrot_training - Step 25965: {'lr': 0.0002361749059140362, 'samples': 13294592, 'steps': 25965, 'batch_loss/train': 0.7741670468822122}
+12/26/2021 09:22:19 - INFO - codeparrot_training - Step 25966: {'lr': 0.00023615916692861587, 'samples': 13295104, 'steps': 25966, 'batch_loss/train': 0.7739833360537887}
+12/26/2021 09:22:31 - INFO - codeparrot_training - Step 25967: {'lr': 0.0002361434279982214, 'samples': 13295616, 'steps': 25967, 'batch_loss/train': 0.7451002122834325}
+12/26/2021 09:22:42 - INFO - codeparrot_training - Step 25968: {'lr': 0.0002361276891229155, 'samples': 13296128, 'steps': 25968, 'batch_loss/train': 0.8506747931241989}
+12/26/2021 09:22:53 - INFO - codeparrot_training - Step 25969: {'lr': 0.0002361119503027606, 'samples': 13296640, 'steps': 25969, 'batch_loss/train': 0.9083570409566164}
+12/26/2021 09:23:07 - INFO - codeparrot_training - Step 25970: {'lr': 0.0002360962115378194, 'samples': 13297152, 'steps': 25970, 'batch_loss/train': 0.8969914205372334}
+12/26/2021 09:23:17 - INFO - codeparrot_training - Step 25971: {'lr': 0.00023608047282815443, 'samples': 13297664, 'steps': 25971, 'batch_loss/train': 0.8583004020620137}
+12/26/2021 09:23:28 - INFO - codeparrot_training - Step 25972: {'lr': 0.00023606473417382812, 'samples': 13298176, 'steps': 25972, 'batch_loss/train': 0.7221879176795483}
+12/26/2021 09:23:38 - INFO - codeparrot_training - Step 25973: {'lr': 0.00023604899557490332, 'samples': 13298688, 'steps': 25973, 'batch_loss/train': 0.6990219606086612}
+12/26/2021 09:23:50 - INFO - codeparrot_training - Step 25974: {'lr': 0.00023603325703144235, 'samples': 13299200, 'steps': 25974, 'batch_loss/train': 0.8094442496076226}
+12/26/2021 09:24:01 - INFO - codeparrot_training - Step 25975: {'lr': 0.00023601751854350795, 'samples': 13299712, 'steps': 25975, 'batch_loss/train': 0.7896806411445141}
+12/26/2021 09:24:12 - INFO - codeparrot_training - Step 25976: {'lr': 0.00023600178011116252, 'samples': 13300224, 'steps': 25976, 'batch_loss/train': 0.714960609562695}
+12/26/2021 09:24:24 - INFO - codeparrot_training - Step 25977: {'lr': 0.00023598604173446879, 'samples': 13300736, 'steps': 25977, 'batch_loss/train': 0.7862966675311327}
+12/26/2021 09:24:34 - INFO - codeparrot_training - Step 25978: {'lr': 0.00023597030341348927, 'samples': 13301248, 'steps': 25978, 'batch_loss/train': 0.8549827896058559}
+12/26/2021 09:24:45 - INFO - codeparrot_training - Step 25979: {'lr': 0.0002359545651482865, 'samples': 13301760, 'steps': 25979, 'batch_loss/train': 0.7247585998848081}
+12/26/2021 09:24:58 - INFO - codeparrot_training - Step 25980: {'lr': 0.000235938826938923, 'samples': 13302272, 'steps': 25980, 'batch_loss/train': 0.7524679191410542}
+12/26/2021 09:25:08 - INFO - codeparrot_training - Step 25981: {'lr': 0.0002359230887854616, 'samples': 13302784, 'steps': 25981, 'batch_loss/train': 0.5509963142685592}
+12/26/2021 09:25:19 - INFO - codeparrot_training - Step 25982: {'lr': 0.00023590735068796463, 'samples': 13303296, 'steps': 25982, 'batch_loss/train': 0.6975228009978309}
+12/26/2021 09:25:33 - INFO - codeparrot_training - Step 25983: {'lr': 0.00023589161264649458, 'samples': 13303808, 'steps': 25983, 'batch_loss/train': 0.6339583437656984}
+12/26/2021 09:25:44 - INFO - codeparrot_training - Step 25984: {'lr': 0.00023587587466111423, 'samples': 13304320, 'steps': 25984, 'batch_loss/train': 0.8780158488079906}
+12/26/2021 09:25:54 - INFO - codeparrot_training - Step 25985: {'lr': 0.00023586013673188613, 'samples': 13304832, 'steps': 25985, 'batch_loss/train': 0.7409118125215173}
+12/26/2021 09:26:05 - INFO - codeparrot_training - Step 25986: {'lr': 0.00023584439885887272, 'samples': 13305344, 'steps': 25986, 'batch_loss/train': 0.6298027783632278}
+12/26/2021 09:26:17 - INFO - codeparrot_training - Step 25987: {'lr': 0.0002358286610421366, 'samples': 13305856, 'steps': 25987, 'batch_loss/train': 0.7174604749307036}
+12/26/2021 09:26:27 - INFO - codeparrot_training - Step 25988: {'lr': 0.00023581292328174046, 'samples': 13306368, 'steps': 25988, 'batch_loss/train': 0.6211636929656379}
+12/26/2021 09:26:38 - INFO - codeparrot_training - Step 25989: {'lr': 0.00023579718557774674, 'samples': 13306880, 'steps': 25989, 'batch_loss/train': 0.7626046957448125}
+12/26/2021 09:26:50 - INFO - codeparrot_training - Step 25990: {'lr': 0.0002357814479302181, 'samples': 13307392, 'steps': 25990, 'batch_loss/train': 0.7834404250606894}
+12/26/2021 09:27:01 - INFO - codeparrot_training - Step 25991: {'lr': 0.00023576571033921692, 'samples': 13307904, 'steps': 25991, 'batch_loss/train': 1.0344557026401162}
+12/26/2021 09:27:12 - INFO - codeparrot_training - Step 25992: {'lr': 0.00023574997280480602, 'samples': 13308416, 'steps': 25992, 'batch_loss/train': 0.7838697091210634}
+12/26/2021 09:27:25 - INFO - codeparrot_training - Step 25993: {'lr': 0.00023573423532704784, 'samples': 13308928, 'steps': 25993, 'batch_loss/train': 0.8149074874818325}
+12/26/2021 09:27:36 - INFO - codeparrot_training - Step 25994: {'lr': 0.00023571849790600486, 'samples': 13309440, 'steps': 25994, 'batch_loss/train': 0.8304463140666485}
+12/26/2021 09:27:47 - INFO - codeparrot_training - Step 25995: {'lr': 0.00023570276054173984, 'samples': 13309952, 'steps': 25995, 'batch_loss/train': 0.7654845351353288}
+12/26/2021 09:27:57 - INFO - codeparrot_training - Step 25996: {'lr': 0.0002356870232343152, 'samples': 13310464, 'steps': 25996, 'batch_loss/train': 0.679148836992681}
+12/26/2021 09:28:09 - INFO - codeparrot_training - Step 25997: {'lr': 0.00023567128598379363, 'samples': 13310976, 'steps': 25997, 'batch_loss/train': 0.6932398732751608}
+12/26/2021 09:28:20 - INFO - codeparrot_training - Step 25998: {'lr': 0.0002356555487902375, 'samples': 13311488, 'steps': 25998, 'batch_loss/train': 0.7532311007380486}
+12/26/2021 09:28:31 - INFO - codeparrot_training - Step 25999: {'lr': 0.00023563981165370952, 'samples': 13312000, 'steps': 25999, 'batch_loss/train': 0.7428278885781765}
+12/26/2021 09:28:45 - INFO - codeparrot_training - Step 26000: {'lr': 0.0002356240745742723, 'samples': 13312512, 'steps': 26000, 'batch_loss/train': 0.7819683216512203}
+12/26/2021 09:28:55 - INFO - codeparrot_training - Step 26001: {'lr': 0.0002356083375519883, 'samples': 13313024, 'steps': 26001, 'batch_loss/train': 1.5837554666213691}
+12/26/2021 09:29:06 - INFO - codeparrot_training - Step 26002: {'lr': 0.00023559260058692006, 'samples': 13313536, 'steps': 26002, 'batch_loss/train': 1.6329371524043381}
+12/26/2021 09:29:18 - INFO - codeparrot_training - Step 26003: {'lr': 0.00023557686367913023, 'samples': 13314048, 'steps': 26003, 'batch_loss/train': 0.7240577936172485}
+12/26/2021 09:29:29 - INFO - codeparrot_training - Step 26004: {'lr': 0.0002355611268286814, 'samples': 13314560, 'steps': 26004, 'batch_loss/train': 0.8276777900755405}
+12/26/2021 09:29:39 - INFO - codeparrot_training - Step 26005: {'lr': 0.000235545390035636, 'samples': 13315072, 'steps': 26005, 'batch_loss/train': 0.8475555153563619}
+12/26/2021 09:29:50 - INFO - codeparrot_training - Step 26006: {'lr': 0.00023552965330005666, 'samples': 13315584, 'steps': 26006, 'batch_loss/train': 0.8010385651141405}
+12/26/2021 09:30:02 - INFO - codeparrot_training - Step 26007: {'lr': 0.00023551391662200602, 'samples': 13316096, 'steps': 26007, 'batch_loss/train': 0.831780334468931}
+12/26/2021 09:30:12 - INFO - codeparrot_training - Step 26008: {'lr': 0.00023549818000154655, 'samples': 13316608, 'steps': 26008, 'batch_loss/train': 0.8252571113407612}
+12/26/2021 09:30:23 - INFO - codeparrot_training - Step 26009: {'lr': 0.0002354824434387408, 'samples': 13317120, 'steps': 26009, 'batch_loss/train': 0.8603667933493853}
+12/26/2021 09:30:37 - INFO - codeparrot_training - Step 26010: {'lr': 0.00023546670693365144, 'samples': 13317632, 'steps': 26010, 'batch_loss/train': 0.8298094915226102}
+12/26/2021 09:30:47 - INFO - codeparrot_training - Step 26011: {'lr': 0.00023545097048634098, 'samples': 13318144, 'steps': 26011, 'batch_loss/train': 0.7826192174106836}
+12/26/2021 09:30:58 - INFO - codeparrot_training - Step 26012: {'lr': 0.00023543523409687192, 'samples': 13318656, 'steps': 26012, 'batch_loss/train': 0.7158940387889743}
+12/26/2021 09:31:10 - INFO - codeparrot_training - Step 26013: {'lr': 0.00023541949776530684, 'samples': 13319168, 'steps': 26013, 'batch_loss/train': 0.7622501326259226}
+12/26/2021 09:31:21 - INFO - codeparrot_training - Step 26014: {'lr': 0.0002354037614917084, 'samples': 13319680, 'steps': 26014, 'batch_loss/train': 0.8237082115374506}
+12/26/2021 09:31:31 - INFO - codeparrot_training - Step 26015: {'lr': 0.00023538802527613907, 'samples': 13320192, 'steps': 26015, 'batch_loss/train': 0.678959489800036}
+12/26/2021 09:31:42 - INFO - codeparrot_training - Step 26016: {'lr': 0.00023537228911866146, 'samples': 13320704, 'steps': 26016, 'batch_loss/train': 0.8331500729545951}
+12/26/2021 09:31:56 - INFO - codeparrot_training - Step 26017: {'lr': 0.000235356553019338, 'samples': 13321216, 'steps': 26017, 'batch_loss/train': 0.7271498269401491}
+12/26/2021 09:32:07 - INFO - codeparrot_training - Step 26018: {'lr': 0.0002353408169782315, 'samples': 13321728, 'steps': 26018, 'batch_loss/train': 0.7428227672353387}
+12/26/2021 09:32:17 - INFO - codeparrot_training - Step 26019: {'lr': 0.00023532508099540427, 'samples': 13322240, 'steps': 26019, 'batch_loss/train': 0.7161113130860031}
+12/26/2021 09:32:29 - INFO - codeparrot_training - Step 26020: {'lr': 0.00023530934507091898, 'samples': 13322752, 'steps': 26020, 'batch_loss/train': 0.6813697521574795}
+12/26/2021 09:32:40 - INFO - codeparrot_training - Step 26021: {'lr': 0.00023529360920483824, 'samples': 13323264, 'steps': 26021, 'batch_loss/train': 0.7968770060688257}
+12/26/2021 09:32:51 - INFO - codeparrot_training - Step 26022: {'lr': 0.00023527787339722454, 'samples': 13323776, 'steps': 26022, 'batch_loss/train': 0.6951539788860828}
+12/26/2021 09:33:03 - INFO - codeparrot_training - Step 26023: {'lr': 0.0002352621376481405, 'samples': 13324288, 'steps': 26023, 'batch_loss/train': 0.8986020470038056}
+12/26/2021 09:33:13 - INFO - codeparrot_training - Step 26024: {'lr': 0.00023524640195764853, 'samples': 13324800, 'steps': 26024, 'batch_loss/train': 0.789143449626863}
+12/26/2021 09:33:24 - INFO - codeparrot_training - Step 26025: {'lr': 0.00023523066632581132, 'samples': 13325312, 'steps': 26025, 'batch_loss/train': 0.6362986410968006}
+12/26/2021 09:33:36 - INFO - codeparrot_training - Step 26026: {'lr': 0.0002352149307526915, 'samples': 13325824, 'steps': 26026, 'batch_loss/train': 0.8891735468059778}
+12/26/2021 09:33:47 - INFO - codeparrot_training - Step 26027: {'lr': 0.0002351991952383515, 'samples': 13326336, 'steps': 26027, 'batch_loss/train': 0.7402698043733835}
+12/26/2021 09:33:57 - INFO - codeparrot_training - Step 26028: {'lr': 0.00023518345978285383, 'samples': 13326848, 'steps': 26028, 'batch_loss/train': 0.9123258572071791}
+12/26/2021 09:34:08 - INFO - codeparrot_training - Step 26029: {'lr': 0.0002351677243862612, 'samples': 13327360, 'steps': 26029, 'batch_loss/train': 0.6906130453571677}
+12/26/2021 09:34:22 - INFO - codeparrot_training - Step 26030: {'lr': 0.0002351519890486361, 'samples': 13327872, 'steps': 26030, 'batch_loss/train': 0.7377040348947048}
+12/26/2021 09:34:32 - INFO - codeparrot_training - Step 26031: {'lr': 0.000235136253770041, 'samples': 13328384, 'steps': 26031, 'batch_loss/train': 0.7830670922994614}
+12/26/2021 09:34:43 - INFO - codeparrot_training - Step 26032: {'lr': 0.00023512051855053862, 'samples': 13328896, 'steps': 26032, 'batch_loss/train': 0.7655929527245462}
+12/26/2021 09:34:55 - INFO - codeparrot_training - Step 26033: {'lr': 0.00023510478339019145, 'samples': 13329408, 'steps': 26033, 'batch_loss/train': 0.7757431399077177}
+12/26/2021 09:35:06 - INFO - codeparrot_training - Step 26034: {'lr': 0.00023508904828906202, 'samples': 13329920, 'steps': 26034, 'batch_loss/train': 0.6988633163273335}
+12/26/2021 09:35:16 - INFO - codeparrot_training - Step 26035: {'lr': 0.0002350733132472128, 'samples': 13330432, 'steps': 26035, 'batch_loss/train': 0.8000454464927316}
+12/26/2021 09:35:29 - INFO - codeparrot_training - Step 26036: {'lr': 0.00023505757826470662, 'samples': 13330944, 'steps': 26036, 'batch_loss/train': 0.749291087500751}
+12/26/2021 09:35:39 - INFO - codeparrot_training - Step 26037: {'lr': 0.00023504184334160582, 'samples': 13331456, 'steps': 26037, 'batch_loss/train': 0.8156044669449329}
+12/26/2021 09:35:50 - INFO - codeparrot_training - Step 26038: {'lr': 0.000235026108477973, 'samples': 13331968, 'steps': 26038, 'batch_loss/train': 0.6475465341936797}
+12/26/2021 09:36:00 - INFO - codeparrot_training - Step 26039: {'lr': 0.00023501037367387062, 'samples': 13332480, 'steps': 26039, 'batch_loss/train': 0.7890656762756407}
+12/26/2021 09:36:15 - INFO - codeparrot_training - Step 26040: {'lr': 0.00023499463892936147, 'samples': 13332992, 'steps': 26040, 'batch_loss/train': 0.7555847116746008}
+12/26/2021 09:36:25 - INFO - codeparrot_training - Step 26041: {'lr': 0.0002349789042445079, 'samples': 13333504, 'steps': 26041, 'batch_loss/train': 0.7939869952388108}
+12/26/2021 09:36:36 - INFO - codeparrot_training - Step 26042: {'lr': 0.0002349631696193726, 'samples': 13334016, 'steps': 26042, 'batch_loss/train': 0.7804129258729517}
+12/26/2021 09:36:48 - INFO - codeparrot_training - Step 26043: {'lr': 0.00023494743505401796, 'samples': 13334528, 'steps': 26043, 'batch_loss/train': 0.779556754976511}
+12/26/2021 09:36:59 - INFO - codeparrot_training - Step 26044: {'lr': 0.00023493170054850665, 'samples': 13335040, 'steps': 26044, 'batch_loss/train': 0.6590242731035687}
+12/26/2021 09:37:09 - INFO - codeparrot_training - Step 26045: {'lr': 0.00023491596610290127, 'samples': 13335552, 'steps': 26045, 'batch_loss/train': 0.8654554346576333}
+12/26/2021 09:37:21 - INFO - codeparrot_training - Step 26046: {'lr': 0.0002349002317172642, 'samples': 13336064, 'steps': 26046, 'batch_loss/train': 0.7215089579112828}
+12/26/2021 09:37:32 - INFO - codeparrot_training - Step 26047: {'lr': 0.00023488449739165825, 'samples': 13336576, 'steps': 26047, 'batch_loss/train': 0.5504820933565497}
+12/26/2021 09:37:43 - INFO - codeparrot_training - Step 26048: {'lr': 0.00023486876312614577, 'samples': 13337088, 'steps': 26048, 'batch_loss/train': 0.8309954688884318}
+12/26/2021 09:37:53 - INFO - codeparrot_training - Step 26049: {'lr': 0.0002348530289207894, 'samples': 13337600, 'steps': 26049, 'batch_loss/train': 0.7403324274346232}
+12/26/2021 09:38:07 - INFO - codeparrot_training - Step 26050: {'lr': 0.00023483729477565162, 'samples': 13338112, 'steps': 26050, 'batch_loss/train': 0.6441928073763847}
+12/26/2021 09:38:17 - INFO - codeparrot_training - Step 26051: {'lr': 0.00023482156069079503, 'samples': 13338624, 'steps': 26051, 'batch_loss/train': 0.7687936997972429}
+12/26/2021 09:38:28 - INFO - codeparrot_training - Step 26052: {'lr': 0.00023480582666628226, 'samples': 13339136, 'steps': 26052, 'batch_loss/train': 0.7480375675368123}
+12/26/2021 09:38:40 - INFO - codeparrot_training - Step 26053: {'lr': 0.0002347900927021757, 'samples': 13339648, 'steps': 26053, 'batch_loss/train': 0.7344648099970073}
+12/26/2021 09:38:51 - INFO - codeparrot_training - Step 26054: {'lr': 0.00023477435879853798, 'samples': 13340160, 'steps': 26054, 'batch_loss/train': 0.7587902992963791}
+12/26/2021 09:39:01 - INFO - codeparrot_training - Step 26055: {'lr': 0.00023475862495543173, 'samples': 13340672, 'steps': 26055, 'batch_loss/train': 0.6749400754924864}
+12/26/2021 09:39:13 - INFO - codeparrot_training - Step 26056: {'lr': 0.00023474289117291947, 'samples': 13341184, 'steps': 26056, 'batch_loss/train': 0.7667188689811155}
+12/26/2021 09:39:24 - INFO - codeparrot_training - Step 26057: {'lr': 0.00023472715745106358, 'samples': 13341696, 'steps': 26057, 'batch_loss/train': 0.8194458084180951}
+12/26/2021 09:39:35 - INFO - codeparrot_training - Step 26058: {'lr': 0.00023471142378992679, 'samples': 13342208, 'steps': 26058, 'batch_loss/train': 0.853099474683404}
+12/26/2021 09:39:45 - INFO - codeparrot_training - Step 26059: {'lr': 0.0002346956901895717, 'samples': 13342720, 'steps': 26059, 'batch_loss/train': 0.8446651850827038}
+12/26/2021 09:39:59 - INFO - codeparrot_training - Step 26060: {'lr': 0.00023467995665006066, 'samples': 13343232, 'steps': 26060, 'batch_loss/train': 0.7404141239821911}
+12/26/2021 09:40:10 - INFO - codeparrot_training - Step 26061: {'lr': 0.00023466422317145632, 'samples': 13343744, 'steps': 26061, 'batch_loss/train': 0.7121214317157865}
+12/26/2021 09:40:20 - INFO - codeparrot_training - Step 26062: {'lr': 0.00023464848975382133, 'samples': 13344256, 'steps': 26062, 'batch_loss/train': 0.7340856790542603}
+12/26/2021 09:40:32 - INFO - codeparrot_training - Step 26063: {'lr': 0.00023463275639721816, 'samples': 13344768, 'steps': 26063, 'batch_loss/train': 0.7486339574679732}
+12/26/2021 09:40:43 - INFO - codeparrot_training - Step 26064: {'lr': 0.00023461702310170927, 'samples': 13345280, 'steps': 26064, 'batch_loss/train': 0.729817243758589}
+12/26/2021 09:40:54 - INFO - codeparrot_training - Step 26065: {'lr': 0.00023460128986735725, 'samples': 13345792, 'steps': 26065, 'batch_loss/train': 0.6613902288954705}
+12/26/2021 09:41:06 - INFO - codeparrot_training - Step 26066: {'lr': 0.0002345855566942248, 'samples': 13346304, 'steps': 26066, 'batch_loss/train': 0.7246125438250601}
+12/26/2021 09:41:16 - INFO - codeparrot_training - Step 26067: {'lr': 0.00023456982358237427, 'samples': 13346816, 'steps': 26067, 'batch_loss/train': 0.6446796543896198}
+12/26/2021 09:41:27 - INFO - codeparrot_training - Step 26068: {'lr': 0.00023455409053186826, 'samples': 13347328, 'steps': 26068, 'batch_loss/train': 0.7702003885060549}
+12/26/2021 09:41:37 - INFO - codeparrot_training - Step 26069: {'lr': 0.00023453835754276946, 'samples': 13347840, 'steps': 26069, 'batch_loss/train': 0.7305349982343614}
+12/26/2021 09:41:51 - INFO - codeparrot_training - Step 26070: {'lr': 0.00023452262461514027, 'samples': 13348352, 'steps': 26070, 'batch_loss/train': 0.7061587243806571}
+12/26/2021 09:42:02 - INFO - codeparrot_training - Step 26071: {'lr': 0.00023450689174904334, 'samples': 13348864, 'steps': 26071, 'batch_loss/train': 0.8305439404211938}
+12/26/2021 09:42:13 - INFO - codeparrot_training - Step 26072: {'lr': 0.00023449115894454104, 'samples': 13349376, 'steps': 26072, 'batch_loss/train': 0.597798774484545}
+12/26/2021 09:42:25 - INFO - codeparrot_training - Step 26073: {'lr': 0.00023447542620169614, 'samples': 13349888, 'steps': 26073, 'batch_loss/train': 0.7754335361532867}
+12/26/2021 09:42:35 - INFO - codeparrot_training - Step 26074: {'lr': 0.00023445969352057104, 'samples': 13350400, 'steps': 26074, 'batch_loss/train': 0.7986839694203809}
+12/26/2021 09:42:46 - INFO - codeparrot_training - Step 26075: {'lr': 0.00023444396090122836, 'samples': 13350912, 'steps': 26075, 'batch_loss/train': 0.7641231298912317}
+12/26/2021 09:43:00 - INFO - codeparrot_training - Step 26076: {'lr': 0.00023442822834373055, 'samples': 13351424, 'steps': 26076, 'batch_loss/train': 0.8496071497211233}
+12/26/2021 09:43:11 - INFO - codeparrot_training - Step 26077: {'lr': 0.00023441249584814022, 'samples': 13351936, 'steps': 26077, 'batch_loss/train': 0.800064499489963}
+12/26/2021 09:43:21 - INFO - codeparrot_training - Step 26078: {'lr': 0.00023439676341452002, 'samples': 13352448, 'steps': 26078, 'batch_loss/train': 0.7726354412734509}
+12/26/2021 09:43:33 - INFO - codeparrot_training - Step 26079: {'lr': 0.00023438103104293234, 'samples': 13352960, 'steps': 26079, 'batch_loss/train': 0.8180374763906002}
+12/26/2021 09:43:44 - INFO - codeparrot_training - Step 26080: {'lr': 0.0002343652987334397, 'samples': 13353472, 'steps': 26080, 'batch_loss/train': 0.7575042750686407}
+12/26/2021 09:43:54 - INFO - codeparrot_training - Step 26081: {'lr': 0.00023434956648610487, 'samples': 13353984, 'steps': 26081, 'batch_loss/train': 0.8903489392250776}
+12/26/2021 09:44:05 - INFO - codeparrot_training - Step 26082: {'lr': 0.00023433383430099025, 'samples': 13354496, 'steps': 26082, 'batch_loss/train': 0.7336709103547037}
+12/26/2021 09:44:17 - INFO - codeparrot_training - Step 26083: {'lr': 0.00023431810217815828, 'samples': 13355008, 'steps': 26083, 'batch_loss/train': 0.8542579677887261}
+12/26/2021 09:44:28 - INFO - codeparrot_training - Step 26084: {'lr': 0.00023430237011767165, 'samples': 13355520, 'steps': 26084, 'batch_loss/train': 0.6742868754081428}
+12/26/2021 09:44:38 - INFO - codeparrot_training - Step 26085: {'lr': 0.00023428663811959295, 'samples': 13356032, 'steps': 26085, 'batch_loss/train': 0.6988393776118755}
+12/26/2021 09:44:52 - INFO - codeparrot_training - Step 26086: {'lr': 0.00023427090618398456, 'samples': 13356544, 'steps': 26086, 'batch_loss/train': 0.7064858681987971}
+12/26/2021 09:45:03 - INFO - codeparrot_training - Step 26087: {'lr': 0.00023425517431090908, 'samples': 13357056, 'steps': 26087, 'batch_loss/train': 0.811085226945579}
+12/26/2021 09:45:13 - INFO - codeparrot_training - Step 26088: {'lr': 0.00023423944250042917, 'samples': 13357568, 'steps': 26088, 'batch_loss/train': 0.7687523546628654}
+12/26/2021 09:45:25 - INFO - codeparrot_training - Step 26089: {'lr': 0.00023422371075260722, 'samples': 13358080, 'steps': 26089, 'batch_loss/train': 0.758080936037004}
+12/26/2021 09:45:36 - INFO - codeparrot_training - Step 26090: {'lr': 0.00023420797906750594, 'samples': 13358592, 'steps': 26090, 'batch_loss/train': 0.7081567756831646}
+12/26/2021 09:45:47 - INFO - codeparrot_training - Step 26091: {'lr': 0.00023419224744518764, 'samples': 13359104, 'steps': 26091, 'batch_loss/train': 0.6830850336700678}
+12/26/2021 09:45:57 - INFO - codeparrot_training - Step 26092: {'lr': 0.0002341765158857151, 'samples': 13359616, 'steps': 26092, 'batch_loss/train': 0.6773157633724622}
+12/26/2021 09:46:09 - INFO - codeparrot_training - Step 26093: {'lr': 0.0002341607843891507, 'samples': 13360128, 'steps': 26093, 'batch_loss/train': 0.8158318211790174}
+12/26/2021 09:46:20 - INFO - codeparrot_training - Step 26094: {'lr': 0.000234145052955557, 'samples': 13360640, 'steps': 26094, 'batch_loss/train': 0.6335561532177962}
+12/26/2021 09:46:31 - INFO - codeparrot_training - Step 26095: {'lr': 0.0002341293215849967, 'samples': 13361152, 'steps': 26095, 'batch_loss/train': 0.6869053670670837}
+12/26/2021 09:46:43 - INFO - codeparrot_training - Step 26096: {'lr': 0.00023411359027753216, 'samples': 13361664, 'steps': 26096, 'batch_loss/train': 0.740988768171519}
+12/26/2021 09:46:54 - INFO - codeparrot_training - Step 26097: {'lr': 0.00023409785903322604, 'samples': 13362176, 'steps': 26097, 'batch_loss/train': 0.7090284409932792}
+12/26/2021 09:47:04 - INFO - codeparrot_training - Step 26098: {'lr': 0.0002340821278521407, 'samples': 13362688, 'steps': 26098, 'batch_loss/train': 0.6597109218128026}
+12/26/2021 09:47:18 - INFO - codeparrot_training - Step 26099: {'lr': 0.00023406639673433888, 'samples': 13363200, 'steps': 26099, 'batch_loss/train': 1.1865664194338024}
+12/26/2021 09:47:29 - INFO - codeparrot_training - Step 26100: {'lr': 0.0002340506656798831, 'samples': 13363712, 'steps': 26100, 'batch_loss/train': 0.7233932334929705}
+12/26/2021 09:47:39 - INFO - codeparrot_training - Step 26101: {'lr': 0.00023403493468883587, 'samples': 13364224, 'steps': 26101, 'batch_loss/train': 0.6929620201699436}
+12/26/2021 09:47:51 - INFO - codeparrot_training - Step 26102: {'lr': 0.00023401920376125957, 'samples': 13364736, 'steps': 26102, 'batch_loss/train': 0.7442413624376059}
+12/26/2021 09:48:02 - INFO - codeparrot_training - Step 26103: {'lr': 0.00023400347289721695, 'samples': 13365248, 'steps': 26103, 'batch_loss/train': 0.6916662137955427}
+12/26/2021 09:48:13 - INFO - codeparrot_training - Step 26104: {'lr': 0.00023398774209677054, 'samples': 13365760, 'steps': 26104, 'batch_loss/train': 0.8064131606370211}
+12/26/2021 09:48:23 - INFO - codeparrot_training - Step 26105: {'lr': 0.00023397201135998269, 'samples': 13366272, 'steps': 26105, 'batch_loss/train': 0.7870007515884936}
+12/26/2021 09:48:35 - INFO - codeparrot_training - Step 26106: {'lr': 0.00023395628068691615, 'samples': 13366784, 'steps': 26106, 'batch_loss/train': 0.7503221407532692}
+12/26/2021 09:48:46 - INFO - codeparrot_training - Step 26107: {'lr': 0.00023394055007763345, 'samples': 13367296, 'steps': 26107, 'batch_loss/train': 0.7024766054237261}
+12/26/2021 09:48:57 - INFO - codeparrot_training - Step 26108: {'lr': 0.00023392481953219703, 'samples': 13367808, 'steps': 26108, 'batch_loss/train': 0.9176308214082383}
+12/26/2021 09:49:11 - INFO - codeparrot_training - Step 26109: {'lr': 0.00023390908905066937, 'samples': 13368320, 'steps': 26109, 'batch_loss/train': 0.7125283456407487}
+12/26/2021 09:49:21 - INFO - codeparrot_training - Step 26110: {'lr': 0.00023389335863311313, 'samples': 13368832, 'steps': 26110, 'batch_loss/train': 0.8522607963532209}
+12/26/2021 09:49:32 - INFO - codeparrot_training - Step 26111: {'lr': 0.00023387762827959088, 'samples': 13369344, 'steps': 26111, 'batch_loss/train': 0.8357624094933271}
+12/26/2021 09:49:44 - INFO - codeparrot_training - Step 26112: {'lr': 0.00023386189799016506, 'samples': 13369856, 'steps': 26112, 'batch_loss/train': 0.6610377198085189}
+12/26/2021 09:49:55 - INFO - codeparrot_training - Step 26113: {'lr': 0.00023384616776489815, 'samples': 13370368, 'steps': 26113, 'batch_loss/train': 0.8656371487304568}
+12/26/2021 09:50:05 - INFO - codeparrot_training - Step 26114: {'lr': 0.0002338304376038529, 'samples': 13370880, 'steps': 26114, 'batch_loss/train': 0.8674363326281309}
+12/26/2021 09:50:16 - INFO - codeparrot_training - Step 26115: {'lr': 0.00023381470750709167, 'samples': 13371392, 'steps': 26115, 'batch_loss/train': 0.8453086540102959}
+12/26/2021 09:50:30 - INFO - codeparrot_training - Step 26116: {'lr': 0.0002337989774746771, 'samples': 13371904, 'steps': 26116, 'batch_loss/train': 0.7349693961441517}
+12/26/2021 09:50:40 - INFO - codeparrot_training - Step 26117: {'lr': 0.00023378324750667156, 'samples': 13372416, 'steps': 26117, 'batch_loss/train': 0.8253106623888016}
+12/26/2021 09:50:51 - INFO - codeparrot_training - Step 26118: {'lr': 0.00023376751760313785, 'samples': 13372928, 'steps': 26118, 'batch_loss/train': 0.8211213289760053}
+12/26/2021 09:51:03 - INFO - codeparrot_training - Step 26119: {'lr': 0.00023375178776413827, 'samples': 13373440, 'steps': 26119, 'batch_loss/train': 0.7695653578266501}
+12/26/2021 09:51:14 - INFO - codeparrot_training - Step 26120: {'lr': 0.0002337360579897354, 'samples': 13373952, 'steps': 26120, 'batch_loss/train': 0.7769173076376319}
+12/26/2021 09:51:24 - INFO - codeparrot_training - Step 26121: {'lr': 0.00023372032827999195, 'samples': 13374464, 'steps': 26121, 'batch_loss/train': 0.8235906669870019}
+12/26/2021 09:51:36 - INFO - codeparrot_training - Step 26122: {'lr': 0.00023370459863497026, 'samples': 13374976, 'steps': 26122, 'batch_loss/train': 0.7366911545395851}
+12/26/2021 09:51:47 - INFO - codeparrot_training - Step 26123: {'lr': 0.000233688869054733, 'samples': 13375488, 'steps': 26123, 'batch_loss/train': 0.721054854337126}
+12/26/2021 09:51:57 - INFO - codeparrot_training - Step 26124: {'lr': 0.00023367313953934253, 'samples': 13376000, 'steps': 26124, 'batch_loss/train': 0.7429492813535035}
+12/26/2021 09:52:08 - INFO - codeparrot_training - Step 26125: {'lr': 0.0002336574100888615, 'samples': 13376512, 'steps': 26125, 'batch_loss/train': 0.8412161711603403}
+12/26/2021 09:52:22 - INFO - codeparrot_training - Step 26126: {'lr': 0.00023364168070335255, 'samples': 13377024, 'steps': 26126, 'batch_loss/train': 0.7101463684812188}
+12/26/2021 09:52:33 - INFO - codeparrot_training - Step 26127: {'lr': 0.00023362595138287808, 'samples': 13377536, 'steps': 26127, 'batch_loss/train': 0.7577155167236924}
+12/26/2021 09:52:43 - INFO - codeparrot_training - Step 26128: {'lr': 0.0002336102221275005, 'samples': 13378048, 'steps': 26128, 'batch_loss/train': 0.8114373916760087}
+12/26/2021 09:52:56 - INFO - codeparrot_training - Step 26129: {'lr': 0.0002335944929372826, 'samples': 13378560, 'steps': 26129, 'batch_loss/train': 0.75711471401155}
+12/26/2021 09:53:07 - INFO - codeparrot_training - Step 26130: {'lr': 0.00023357876381228685, 'samples': 13379072, 'steps': 26130, 'batch_loss/train': 1.3573623867705464}
+12/26/2021 09:53:18 - INFO - codeparrot_training - Step 26131: {'lr': 0.0002335630347525756, 'samples': 13379584, 'steps': 26131, 'batch_loss/train': 0.7126883128657937}
+12/26/2021 09:53:30 - INFO - codeparrot_training - Step 26132: {'lr': 0.00023354730575821157, 'samples': 13380096, 'steps': 26132, 'batch_loss/train': 0.7333954321220517}
+12/26/2021 09:53:40 - INFO - codeparrot_training - Step 26133: {'lr': 0.0002335315768292573, 'samples': 13380608, 'steps': 26133, 'batch_loss/train': 0.8198731241282076}
+12/26/2021 09:53:51 - INFO - codeparrot_training - Step 26134: {'lr': 0.0002335158479657752, 'samples': 13381120, 'steps': 26134, 'batch_loss/train': 0.7400715332478285}
+12/26/2021 09:54:01 - INFO - codeparrot_training - Step 26135: {'lr': 0.00023350011916782785, 'samples': 13381632, 'steps': 26135, 'batch_loss/train': 0.7217797404155135}
+12/26/2021 09:54:16 - INFO - codeparrot_training - Step 26136: {'lr': 0.00023348439043547787, 'samples': 13382144, 'steps': 26136, 'batch_loss/train': 0.7197559922933578}
+12/26/2021 09:54:27 - INFO - codeparrot_training - Step 26137: {'lr': 0.00023346866176878774, 'samples': 13382656, 'steps': 26137, 'batch_loss/train': 0.754384396597743}
+12/26/2021 09:54:37 - INFO - codeparrot_training - Step 26138: {'lr': 0.00023345293316781989, 'samples': 13383168, 'steps': 26138, 'batch_loss/train': 0.6137346024625003}
+12/26/2021 09:54:49 - INFO - codeparrot_training - Step 26139: {'lr': 0.00023343720463263687, 'samples': 13383680, 'steps': 26139, 'batch_loss/train': 0.7604665872640908}
+12/26/2021 09:55:00 - INFO - codeparrot_training - Step 26140: {'lr': 0.0002334214761633014, 'samples': 13384192, 'steps': 26140, 'batch_loss/train': 0.7813973668962717}
+12/26/2021 09:55:11 - INFO - codeparrot_training - Step 26141: {'lr': 0.00023340574775987583, 'samples': 13384704, 'steps': 26141, 'batch_loss/train': 0.8254727032035589}
+12/26/2021 09:55:23 - INFO - codeparrot_training - Step 26142: {'lr': 0.00023339001942242267, 'samples': 13385216, 'steps': 26142, 'batch_loss/train': 0.761193691752851}
+12/26/2021 09:55:33 - INFO - codeparrot_training - Step 26143: {'lr': 0.00023337429115100465, 'samples': 13385728, 'steps': 26143, 'batch_loss/train': 0.8571447683498263}
+12/26/2021 09:55:44 - INFO - codeparrot_training - Step 26144: {'lr': 0.00023335856294568407, 'samples': 13386240, 'steps': 26144, 'batch_loss/train': 0.7415509894490242}
+12/26/2021 09:55:55 - INFO - codeparrot_training - Step 26145: {'lr': 0.00023334283480652365, 'samples': 13386752, 'steps': 26145, 'batch_loss/train': 0.8814348978921771}
+12/26/2021 09:56:09 - INFO - codeparrot_training - Step 26146: {'lr': 0.00023332710673358573, 'samples': 13387264, 'steps': 26146, 'batch_loss/train': 1.092443136498332}
+12/26/2021 09:56:19 - INFO - codeparrot_training - Step 26147: {'lr': 0.00023331137872693302, 'samples': 13387776, 'steps': 26147, 'batch_loss/train': 0.7659530355595052}
+12/26/2021 09:56:30 - INFO - codeparrot_training - Step 26148: {'lr': 0.00023329565078662796, 'samples': 13388288, 'steps': 26148, 'batch_loss/train': 0.7718757493421435}
+12/26/2021 09:56:42 - INFO - codeparrot_training - Step 26149: {'lr': 0.00023327992291273308, 'samples': 13388800, 'steps': 26149, 'batch_loss/train': 0.752516588428989}
+12/26/2021 09:56:52 - INFO - codeparrot_training - Step 26150: {'lr': 0.00023326419510531082, 'samples': 13389312, 'steps': 26150, 'batch_loss/train': 0.8019016687758267}
+12/26/2021 09:57:03 - INFO - codeparrot_training - Step 26151: {'lr': 0.0002332484673644239, 'samples': 13389824, 'steps': 26151, 'batch_loss/train': 0.8583212997764349}
+12/26/2021 09:57:15 - INFO - codeparrot_training - Step 26152: {'lr': 0.00023323273969013476, 'samples': 13390336, 'steps': 26152, 'batch_loss/train': 0.8662036973983049}
+12/26/2021 09:57:26 - INFO - codeparrot_training - Step 26153: {'lr': 0.00023321701208250592, 'samples': 13390848, 'steps': 26153, 'batch_loss/train': 0.8236299380660057}
+12/26/2021 09:57:37 - INFO - codeparrot_training - Step 26154: {'lr': 0.0002332012845415998, 'samples': 13391360, 'steps': 26154, 'batch_loss/train': 0.6762142558582127}
+12/26/2021 09:57:47 - INFO - codeparrot_training - Step 26155: {'lr': 0.0002331855570674791, 'samples': 13391872, 'steps': 26155, 'batch_loss/train': 0.6285018532071263}
+12/26/2021 09:58:01 - INFO - codeparrot_training - Step 26156: {'lr': 0.00023316982966020632, 'samples': 13392384, 'steps': 26156, 'batch_loss/train': 0.7696425663307309}
+12/26/2021 09:58:12 - INFO - codeparrot_training - Step 26157: {'lr': 0.0002331541023198438, 'samples': 13392896, 'steps': 26157, 'batch_loss/train': 0.8106358023360372}
+12/26/2021 09:58:23 - INFO - codeparrot_training - Step 26158: {'lr': 0.0002331383750464543, 'samples': 13393408, 'steps': 26158, 'batch_loss/train': 0.7853950676508248}
+12/26/2021 09:58:35 - INFO - codeparrot_training - Step 26159: {'lr': 0.0002331226478401003, 'samples': 13393920, 'steps': 26159, 'batch_loss/train': 0.5293515503872186}
+12/26/2021 09:58:45 - INFO - codeparrot_training - Step 26160: {'lr': 0.00023310692070084422, 'samples': 13394432, 'steps': 26160, 'batch_loss/train': 0.7831840617582202}
+12/26/2021 09:58:56 - INFO - codeparrot_training - Step 26161: {'lr': 0.00023309119362874858, 'samples': 13394944, 'steps': 26161, 'batch_loss/train': 0.7708150092512369}
+12/26/2021 09:59:08 - INFO - codeparrot_training - Step 26162: {'lr': 0.0002330754666238761, 'samples': 13395456, 'steps': 26162, 'batch_loss/train': 0.7340543884783983}
+12/26/2021 09:59:19 - INFO - codeparrot_training - Step 26163: {'lr': 0.0002330597396862892, 'samples': 13395968, 'steps': 26163, 'batch_loss/train': 0.7614497162867337}
+12/26/2021 09:59:29 - INFO - codeparrot_training - Step 26164: {'lr': 0.0002330440128160503, 'samples': 13396480, 'steps': 26164, 'batch_loss/train': 0.7221005982137285}
+12/26/2021 09:59:43 - INFO - codeparrot_training - Step 26165: {'lr': 0.0002330282860132219, 'samples': 13396992, 'steps': 26165, 'batch_loss/train': 0.7697407845407724}
+12/26/2021 09:59:54 - INFO - codeparrot_training - Step 26166: {'lr': 0.0002330125592778668, 'samples': 13397504, 'steps': 26166, 'batch_loss/train': 0.7741874142084271}
+12/26/2021 10:00:05 - INFO - codeparrot_training - Step 26167: {'lr': 0.00023299683261004725, 'samples': 13398016, 'steps': 26167, 'batch_loss/train': 0.8488091416656971}
+12/26/2021 10:00:15 - INFO - codeparrot_training - Step 26168: {'lr': 0.00023298110600982586, 'samples': 13398528, 'steps': 26168, 'batch_loss/train': 0.771504445001483}
+12/26/2021 10:00:27 - INFO - codeparrot_training - Step 26169: {'lr': 0.00023296537947726525, 'samples': 13399040, 'steps': 26169, 'batch_loss/train': 0.6329452376812696}
+12/26/2021 10:00:38 - INFO - codeparrot_training - Step 26170: {'lr': 0.00023294965301242783, 'samples': 13399552, 'steps': 26170, 'batch_loss/train': 0.7216535108163953}
+12/26/2021 10:00:49 - INFO - codeparrot_training - Step 26171: {'lr': 0.00023293392661537615, 'samples': 13400064, 'steps': 26171, 'batch_loss/train': 0.7001497126184404}
+12/26/2021 10:01:02 - INFO - codeparrot_training - Step 26172: {'lr': 0.00023291820028617267, 'samples': 13400576, 'steps': 26172, 'batch_loss/train': 0.6967901097377762}
+12/26/2021 10:01:13 - INFO - codeparrot_training - Step 26173: {'lr': 0.00023290247402488009, 'samples': 13401088, 'steps': 26173, 'batch_loss/train': 0.7278851922601461}
+12/26/2021 10:01:24 - INFO - codeparrot_training - Step 26174: {'lr': 0.00023288674783156075, 'samples': 13401600, 'steps': 26174, 'batch_loss/train': 0.7752253534272313}
+12/26/2021 10:01:36 - INFO - codeparrot_training - Step 26175: {'lr': 0.00023287102170627727, 'samples': 13402112, 'steps': 26175, 'batch_loss/train': 0.7098590465029702}
+12/26/2021 10:01:46 - INFO - codeparrot_training - Step 26176: {'lr': 0.0002328552956490921, 'samples': 13402624, 'steps': 26176, 'batch_loss/train': 0.7505971062928438}
+12/26/2021 10:01:57 - INFO - codeparrot_training - Step 26177: {'lr': 0.00023283956966006777, 'samples': 13403136, 'steps': 26177, 'batch_loss/train': 0.730323847848922}
+12/26/2021 10:02:07 - INFO - codeparrot_training - Step 26178: {'lr': 0.00023282384373926693, 'samples': 13403648, 'steps': 26178, 'batch_loss/train': 0.8090007407590747}
+12/26/2021 10:02:19 - INFO - codeparrot_training - Step 26179: {'lr': 0.00023280811788675189, 'samples': 13404160, 'steps': 26179, 'batch_loss/train': 0.7712122884113342}
+12/26/2021 10:02:30 - INFO - codeparrot_training - Step 26180: {'lr': 0.00023279239210258532, 'samples': 13404672, 'steps': 26180, 'batch_loss/train': 0.7078364416956902}
+12/26/2021 10:02:41 - INFO - codeparrot_training - Step 26181: {'lr': 0.0002327766663868298, 'samples': 13405184, 'steps': 26181, 'batch_loss/train': 0.8172575352946296}
+12/26/2021 10:02:53 - INFO - codeparrot_training - Step 26182: {'lr': 0.00023276094073954772, 'samples': 13405696, 'steps': 26182, 'batch_loss/train': 0.7967284377664328}
+12/26/2021 10:03:04 - INFO - codeparrot_training - Step 26183: {'lr': 0.0002327452151608015, 'samples': 13406208, 'steps': 26183, 'batch_loss/train': 0.7490208093076944}
+12/26/2021 10:03:14 - INFO - codeparrot_training - Step 26184: {'lr': 0.00023272948965065389, 'samples': 13406720, 'steps': 26184, 'batch_loss/train': 0.7314787735231221}
+12/26/2021 10:03:28 - INFO - codeparrot_training - Step 26185: {'lr': 0.0002327137642091673, 'samples': 13407232, 'steps': 26185, 'batch_loss/train': 0.7268679328262806}
+12/26/2021 10:03:39 - INFO - codeparrot_training - Step 26186: {'lr': 0.00023269803883640424, 'samples': 13407744, 'steps': 26186, 'batch_loss/train': 0.7260758108459413}
+12/26/2021 10:03:49 - INFO - codeparrot_training - Step 26187: {'lr': 0.0002326823135324272, 'samples': 13408256, 'steps': 26187, 'batch_loss/train': 0.669847896322608}
+12/26/2021 10:04:00 - INFO - codeparrot_training - Step 26188: {'lr': 0.00023266658829729887, 'samples': 13408768, 'steps': 26188, 'batch_loss/train': 0.7562977713532746}
+12/26/2021 10:04:12 - INFO - codeparrot_training - Step 26189: {'lr': 0.00023265086313108154, 'samples': 13409280, 'steps': 26189, 'batch_loss/train': 0.695277632214129}
+12/26/2021 10:04:23 - INFO - codeparrot_training - Step 26190: {'lr': 0.0002326351380338379, 'samples': 13409792, 'steps': 26190, 'batch_loss/train': 0.8684347374364734}
+12/26/2021 10:04:33 - INFO - codeparrot_training - Step 26191: {'lr': 0.0002326194130056303, 'samples': 13410304, 'steps': 26191, 'batch_loss/train': 0.8933281591162086}
+12/26/2021 10:04:45 - INFO - codeparrot_training - Step 26192: {'lr': 0.00023260368804652143, 'samples': 13410816, 'steps': 26192, 'batch_loss/train': 0.7761816796846688}
+12/26/2021 10:04:56 - INFO - codeparrot_training - Step 26193: {'lr': 0.0002325879631565737, 'samples': 13411328, 'steps': 26193, 'batch_loss/train': 0.7965296604670584}
+12/26/2021 10:05:07 - INFO - codeparrot_training - Step 26194: {'lr': 0.0002325722383358496, 'samples': 13411840, 'steps': 26194, 'batch_loss/train': 0.8024264238774776}
+12/26/2021 10:05:20 - INFO - codeparrot_training - Step 26195: {'lr': 0.0002325565135844118, 'samples': 13412352, 'steps': 26195, 'batch_loss/train': 0.798708651214838}
+12/26/2021 10:05:31 - INFO - codeparrot_training - Step 26196: {'lr': 0.00023254078890232264, 'samples': 13412864, 'steps': 26196, 'batch_loss/train': 0.651177337160334}
+12/26/2021 10:05:42 - INFO - codeparrot_training - Step 26197: {'lr': 0.0002325250642896448, 'samples': 13413376, 'steps': 26197, 'batch_loss/train': 0.7126981876790524}
+12/26/2021 10:05:52 - INFO - codeparrot_training - Step 26198: {'lr': 0.00023250933974644055, 'samples': 13413888, 'steps': 26198, 'batch_loss/train': 0.7137439642101526}
+12/26/2021 10:06:05 - INFO - codeparrot_training - Step 26199: {'lr': 0.0002324936152727727, 'samples': 13414400, 'steps': 26199, 'batch_loss/train': 0.7095487420447171}
+12/26/2021 10:06:16 - INFO - codeparrot_training - Step 26200: {'lr': 0.00023247789086870357, 'samples': 13414912, 'steps': 26200, 'batch_loss/train': 0.6828748772386461}
+12/26/2021 10:06:26 - INFO - codeparrot_training - Step 26201: {'lr': 0.0002324621665342958, 'samples': 13415424, 'steps': 26201, 'batch_loss/train': 0.5886604227125645}
+12/26/2021 10:06:40 - INFO - codeparrot_training - Step 26202: {'lr': 0.00023244644226961172, 'samples': 13415936, 'steps': 26202, 'batch_loss/train': 0.6795970303937793}
+12/26/2021 10:06:51 - INFO - codeparrot_training - Step 26203: {'lr': 0.000232430718074714, 'samples': 13416448, 'steps': 26203, 'batch_loss/train': 0.6193226324976422}
+12/26/2021 10:07:02 - INFO - codeparrot_training - Step 26204: {'lr': 0.00023241499394966518, 'samples': 13416960, 'steps': 26204, 'batch_loss/train': 0.8341347416862845}
+12/26/2021 10:07:14 - INFO - codeparrot_training - Step 26205: {'lr': 0.0002323992698945276, 'samples': 13417472, 'steps': 26205, 'batch_loss/train': 0.7065936243161559}
+12/26/2021 10:07:24 - INFO - codeparrot_training - Step 26206: {'lr': 0.0002323835459093639, 'samples': 13417984, 'steps': 26206, 'batch_loss/train': 0.7896255105733871}
+12/26/2021 10:07:35 - INFO - codeparrot_training - Step 26207: {'lr': 0.00023236782199423665, 'samples': 13418496, 'steps': 26207, 'batch_loss/train': 0.7958039687946439}
+12/26/2021 10:07:47 - INFO - codeparrot_training - Step 26208: {'lr': 0.00023235209814920826, 'samples': 13419008, 'steps': 26208, 'batch_loss/train': 0.6834685415960848}
+12/26/2021 10:07:58 - INFO - codeparrot_training - Step 26209: {'lr': 0.00023233637437434116, 'samples': 13419520, 'steps': 26209, 'batch_loss/train': 0.7255755057558417}
+12/26/2021 10:08:09 - INFO - codeparrot_training - Step 26210: {'lr': 0.00023232065066969805, 'samples': 13420032, 'steps': 26210, 'batch_loss/train': 0.8582848599180579}
+12/26/2021 10:08:19 - INFO - codeparrot_training - Step 26211: {'lr': 0.00023230492703534136, 'samples': 13420544, 'steps': 26211, 'batch_loss/train': 0.7397546181455255}
+12/26/2021 10:08:33 - INFO - codeparrot_training - Step 26212: {'lr': 0.00023228920347133358, 'samples': 13421056, 'steps': 26212, 'batch_loss/train': 0.7721071289852262}
+12/26/2021 10:08:44 - INFO - codeparrot_training - Step 26213: {'lr': 0.00023227347997773718, 'samples': 13421568, 'steps': 26213, 'batch_loss/train': 0.8341947020962834}
+12/26/2021 10:08:54 - INFO - codeparrot_training - Step 26214: {'lr': 0.00023225775655461484, 'samples': 13422080, 'steps': 26214, 'batch_loss/train': 0.7484061205759645}
+12/26/2021 10:09:07 - INFO - codeparrot_training - Step 26215: {'lr': 0.00023224203320202893, 'samples': 13422592, 'steps': 26215, 'batch_loss/train': 0.7408265848644078}
+12/26/2021 10:09:17 - INFO - codeparrot_training - Step 26216: {'lr': 0.0002322263099200419, 'samples': 13423104, 'steps': 26216, 'batch_loss/train': 0.8158200024627149}
+12/26/2021 10:09:28 - INFO - codeparrot_training - Step 26217: {'lr': 0.00023221058670871644, 'samples': 13423616, 'steps': 26217, 'batch_loss/train': 0.7274777055718005}
+12/26/2021 10:09:40 - INFO - codeparrot_training - Step 26218: {'lr': 0.00023219486356811504, 'samples': 13424128, 'steps': 26218, 'batch_loss/train': 0.6769930575974286}
+12/26/2021 10:09:51 - INFO - codeparrot_training - Step 26219: {'lr': 0.00023217914049830003, 'samples': 13424640, 'steps': 26219, 'batch_loss/train': 0.7066082265228033}
+12/26/2021 10:10:01 - INFO - codeparrot_training - Step 26220: {'lr': 0.00023216341749933398, 'samples': 13425152, 'steps': 26220, 'batch_loss/train': 0.7364610154181719}
+12/26/2021 10:10:12 - INFO - codeparrot_training - Step 26221: {'lr': 0.00023214769457127957, 'samples': 13425664, 'steps': 26221, 'batch_loss/train': 0.7984052752144635}
+12/26/2021 10:10:24 - INFO - codeparrot_training - Step 26222: {'lr': 0.00023213197171419915, 'samples': 13426176, 'steps': 26222, 'batch_loss/train': 0.8635175498202443}
+12/26/2021 10:10:35 - INFO - codeparrot_training - Step 26223: {'lr': 0.00023211624892815528, 'samples': 13426688, 'steps': 26223, 'batch_loss/train': 0.8087545540183783}
+12/26/2021 10:10:45 - INFO - codeparrot_training - Step 26224: {'lr': 0.00023210052621321035, 'samples': 13427200, 'steps': 26224, 'batch_loss/train': 1.0036026174202561}
+12/26/2021 10:10:59 - INFO - codeparrot_training - Step 26225: {'lr': 0.00023208480356942703, 'samples': 13427712, 'steps': 26225, 'batch_loss/train': 0.7077645503450185}
+12/26/2021 10:11:10 - INFO - codeparrot_training - Step 26226: {'lr': 0.00023206908099686783, 'samples': 13428224, 'steps': 26226, 'batch_loss/train': 0.8674269486218691}
+12/26/2021 10:11:21 - INFO - codeparrot_training - Step 26227: {'lr': 0.00023205335849559518, 'samples': 13428736, 'steps': 26227, 'batch_loss/train': 0.6503856899216771}
+12/26/2021 10:11:33 - INFO - codeparrot_training - Step 26228: {'lr': 0.0002320376360656715, 'samples': 13429248, 'steps': 26228, 'batch_loss/train': 0.7031209585256875}
+12/26/2021 10:11:43 - INFO - codeparrot_training - Step 26229: {'lr': 0.00023202191370715945, 'samples': 13429760, 'steps': 26229, 'batch_loss/train': 0.860745926387608}
+12/26/2021 10:11:54 - INFO - codeparrot_training - Step 26230: {'lr': 0.00023200619142012152, 'samples': 13430272, 'steps': 26230, 'batch_loss/train': 0.7567672715522349}
+12/26/2021 10:12:04 - INFO - codeparrot_training - Step 26231: {'lr': 0.0002319904692046201, 'samples': 13430784, 'steps': 26231, 'batch_loss/train': 0.901835210621357}
+12/26/2021 10:12:18 - INFO - codeparrot_training - Step 26232: {'lr': 0.0002319747470607178, 'samples': 13431296, 'steps': 26232, 'batch_loss/train': 0.7115955180488527}
+12/26/2021 10:12:29 - INFO - codeparrot_training - Step 26233: {'lr': 0.00023195902498847717, 'samples': 13431808, 'steps': 26233, 'batch_loss/train': 0.7389371367171407}
+12/26/2021 10:12:39 - INFO - codeparrot_training - Step 26234: {'lr': 0.0002319433029879606, 'samples': 13432320, 'steps': 26234, 'batch_loss/train': 0.7826099436497316}
+12/26/2021 10:12:51 - INFO - codeparrot_training - Step 26235: {'lr': 0.0002319275810592306, 'samples': 13432832, 'steps': 26235, 'batch_loss/train': 0.771770715713501}
+12/26/2021 10:13:02 - INFO - codeparrot_training - Step 26236: {'lr': 0.00023191185920234977, 'samples': 13433344, 'steps': 26236, 'batch_loss/train': 0.7349578198045492}
+12/26/2021 10:13:13 - INFO - codeparrot_training - Step 26237: {'lr': 0.00023189613741738058, 'samples': 13433856, 'steps': 26237, 'batch_loss/train': 0.7713393960148096}
+12/26/2021 10:13:25 - INFO - codeparrot_training - Step 26238: {'lr': 0.00023188041570438547, 'samples': 13434368, 'steps': 26238, 'batch_loss/train': 0.7941971952095628}
+12/26/2021 10:13:35 - INFO - codeparrot_training - Step 26239: {'lr': 0.0002318646940634269, 'samples': 13434880, 'steps': 26239, 'batch_loss/train': 0.7853837659931742}
+12/26/2021 10:13:46 - INFO - codeparrot_training - Step 26240: {'lr': 0.00023184897249456762, 'samples': 13435392, 'steps': 26240, 'batch_loss/train': 0.7345951870083809}
+12/26/2021 10:13:59 - INFO - codeparrot_training - Step 26241: {'lr': 0.0002318332509978699, 'samples': 13435904, 'steps': 26241, 'batch_loss/train': 0.7863642331212759}
+12/26/2021 10:14:10 - INFO - codeparrot_training - Step 26242: {'lr': 0.00023181752957339625, 'samples': 13436416, 'steps': 26242, 'batch_loss/train': 0.7956974180415273}
+12/26/2021 10:14:21 - INFO - codeparrot_training - Step 26243: {'lr': 0.00023180180822120934, 'samples': 13436928, 'steps': 26243, 'batch_loss/train': 0.8023375663906336}
+12/26/2021 10:14:31 - INFO - codeparrot_training - Step 26244: {'lr': 0.00023178608694137158, 'samples': 13437440, 'steps': 26244, 'batch_loss/train': 0.7816259292885661}
+12/26/2021 10:14:43 - INFO - codeparrot_training - Step 26245: {'lr': 0.0002317703657339454, 'samples': 13437952, 'steps': 26245, 'batch_loss/train': 0.8157386677339673}
+12/26/2021 10:14:54 - INFO - codeparrot_training - Step 26246: {'lr': 0.00023175464459899327, 'samples': 13438464, 'steps': 26246, 'batch_loss/train': 0.7845200505107641}
+12/26/2021 10:15:05 - INFO - codeparrot_training - Step 26247: {'lr': 0.00023173892353657794, 'samples': 13438976, 'steps': 26247, 'batch_loss/train': 0.6888174738269299}
+12/26/2021 10:15:17 - INFO - codeparrot_training - Step 26248: {'lr': 0.0002317232025467617, 'samples': 13439488, 'steps': 26248, 'batch_loss/train': 0.6556850392371416}
+12/26/2021 10:15:28 - INFO - codeparrot_training - Step 26249: {'lr': 0.00023170748162960715, 'samples': 13440000, 'steps': 26249, 'batch_loss/train': 0.8230776488780975}
+12/26/2021 10:15:38 - INFO - codeparrot_training - Step 26250: {'lr': 0.00023169176078517665, 'samples': 13440512, 'steps': 26250, 'batch_loss/train': 0.7073634809348732}
+12/26/2021 10:15:52 - INFO - codeparrot_training - Step 26251: {'lr': 0.00023167604001353282, 'samples': 13441024, 'steps': 26251, 'batch_loss/train': 0.4805913553573191}
+12/26/2021 10:16:03 - INFO - codeparrot_training - Step 26252: {'lr': 0.0002316603193147382, 'samples': 13441536, 'steps': 26252, 'batch_loss/train': 0.7876241244375706}
+12/26/2021 10:16:14 - INFO - codeparrot_training - Step 26253: {'lr': 0.00023164459868885514, 'samples': 13442048, 'steps': 26253, 'batch_loss/train': 0.7730848230421543}
+12/26/2021 10:16:24 - INFO - codeparrot_training - Step 26254: {'lr': 0.00023162887813594634, 'samples': 13442560, 'steps': 26254, 'batch_loss/train': 0.6762714367359877}
+12/26/2021 10:16:36 - INFO - codeparrot_training - Step 26255: {'lr': 0.0002316131576560741, 'samples': 13443072, 'steps': 26255, 'batch_loss/train': 0.70591601729393}
+12/26/2021 10:16:47 - INFO - codeparrot_training - Step 26256: {'lr': 0.00023159743724930108, 'samples': 13443584, 'steps': 26256, 'batch_loss/train': 0.7158820340409875}
+12/26/2021 10:16:58 - INFO - codeparrot_training - Step 26257: {'lr': 0.00023158171691568954, 'samples': 13444096, 'steps': 26257, 'batch_loss/train': 0.8887211736291647}
+12/26/2021 10:17:10 - INFO - codeparrot_training - Step 26258: {'lr': 0.00023156599665530227, 'samples': 13444608, 'steps': 26258, 'batch_loss/train': 0.8374080625362694}
+12/26/2021 10:17:20 - INFO - codeparrot_training - Step 26259: {'lr': 0.00023155027646820165, 'samples': 13445120, 'steps': 26259, 'batch_loss/train': 0.8579171122983098}
+12/26/2021 10:17:31 - INFO - codeparrot_training - Step 26260: {'lr': 0.0002315345563544501, 'samples': 13445632, 'steps': 26260, 'batch_loss/train': 0.7153625497594476}
+12/26/2021 10:17:43 - INFO - codeparrot_training - Step 26261: {'lr': 0.00023151883631411014, 'samples': 13446144, 'steps': 26261, 'batch_loss/train': 0.7794278869405389}
+12/26/2021 10:17:54 - INFO - codeparrot_training - Step 26262: {'lr': 0.0002315031163472444, 'samples': 13446656, 'steps': 26262, 'batch_loss/train': 0.49711071548517793}
+12/26/2021 10:18:04 - INFO - codeparrot_training - Step 26263: {'lr': 0.0002314873964539153, 'samples': 13447168, 'steps': 26263, 'batch_loss/train': 0.8196134075988084}
+12/26/2021 10:18:15 - INFO - codeparrot_training - Step 26264: {'lr': 0.00023147167663418528, 'samples': 13447680, 'steps': 26264, 'batch_loss/train': 0.814304718747735}
+12/26/2021 10:18:29 - INFO - codeparrot_training - Step 26265: {'lr': 0.00023145595688811682, 'samples': 13448192, 'steps': 26265, 'batch_loss/train': 0.6988451024517417}
+12/26/2021 10:18:40 - INFO - codeparrot_training - Step 26266: {'lr': 0.0002314402372157726, 'samples': 13448704, 'steps': 26266, 'batch_loss/train': 0.718789691105485}
+12/26/2021 10:18:50 - INFO - codeparrot_training - Step 26267: {'lr': 0.0002314245176172149, 'samples': 13449216, 'steps': 26267, 'batch_loss/train': 0.8405181029811502}
+12/26/2021 10:19:02 - INFO - codeparrot_training - Step 26268: {'lr': 0.00023140879809250626, 'samples': 13449728, 'steps': 26268, 'batch_loss/train': 0.7701915372163057}
+12/26/2021 10:19:13 - INFO - codeparrot_training - Step 26269: {'lr': 0.00023139307864170934, 'samples': 13450240, 'steps': 26269, 'batch_loss/train': 0.634485955350101}
+12/26/2021 10:19:24 - INFO - codeparrot_training - Step 26270: {'lr': 0.00023137735926488643, 'samples': 13450752, 'steps': 26270, 'batch_loss/train': 0.761842031031847}
+12/26/2021 10:19:37 - INFO - codeparrot_training - Step 26271: {'lr': 0.00023136163996210018, 'samples': 13451264, 'steps': 26271, 'batch_loss/train': 0.8154570814222097}
+12/26/2021 10:19:48 - INFO - codeparrot_training - Step 26272: {'lr': 0.0002313459207334129, 'samples': 13451776, 'steps': 26272, 'batch_loss/train': 0.7310328297317028}
+12/26/2021 10:19:59 - INFO - codeparrot_training - Step 26273: {'lr': 0.0002313302015788873, 'samples': 13452288, 'steps': 26273, 'batch_loss/train': 0.930685460800305}
+12/26/2021 10:20:09 - INFO - codeparrot_training - Step 26274: {'lr': 0.00023131448249858572, 'samples': 13452800, 'steps': 26274, 'batch_loss/train': 0.7044536017347127}
+12/26/2021 10:20:22 - INFO - codeparrot_training - Step 26275: {'lr': 0.00023129876349257077, 'samples': 13453312, 'steps': 26275, 'batch_loss/train': 0.7879202831536531}
+12/26/2021 10:20:32 - INFO - codeparrot_training - Step 26276: {'lr': 0.00023128304456090476, 'samples': 13453824, 'steps': 26276, 'batch_loss/train': 0.7817599903792143}
+12/26/2021 10:20:43 - INFO - codeparrot_training - Step 26277: {'lr': 0.00023126732570365034, 'samples': 13454336, 'steps': 26277, 'batch_loss/train': 0.7262054271996021}
+12/26/2021 10:20:55 - INFO - codeparrot_training - Step 26278: {'lr': 0.00023125160692087003, 'samples': 13454848, 'steps': 26278, 'batch_loss/train': 0.6550607895478606}
+12/26/2021 10:21:06 - INFO - codeparrot_training - Step 26279: {'lr': 0.0002312358882126261, 'samples': 13455360, 'steps': 26279, 'batch_loss/train': 0.787059148773551}
+12/26/2021 10:21:17 - INFO - codeparrot_training - Step 26280: {'lr': 0.0002312201695789813, 'samples': 13455872, 'steps': 26280, 'batch_loss/train': 0.7623834004625678}
+12/26/2021 10:21:30 - INFO - codeparrot_training - Step 26281: {'lr': 0.00023120445101999803, 'samples': 13456384, 'steps': 26281, 'batch_loss/train': 0.8411388462409377}
+12/26/2021 10:21:41 - INFO - codeparrot_training - Step 26282: {'lr': 0.0002311887325357388, 'samples': 13456896, 'steps': 26282, 'batch_loss/train': 0.8290560180321336}
+12/26/2021 10:21:52 - INFO - codeparrot_training - Step 26283: {'lr': 0.00023117301412626596, 'samples': 13457408, 'steps': 26283, 'batch_loss/train': 1.7429401744157076}
+12/26/2021 10:22:02 - INFO - codeparrot_training - Step 26284: {'lr': 0.00023115729579164214, 'samples': 13457920, 'steps': 26284, 'batch_loss/train': 0.832346623763442}
+12/26/2021 10:22:15 - INFO - codeparrot_training - Step 26285: {'lr': 0.00023114157753192986, 'samples': 13458432, 'steps': 26285, 'batch_loss/train': 0.8304202407598495}
+12/26/2021 10:22:25 - INFO - codeparrot_training - Step 26286: {'lr': 0.0002311258593471915, 'samples': 13458944, 'steps': 26286, 'batch_loss/train': 0.7541996911168098}
+12/26/2021 10:22:36 - INFO - codeparrot_training - Step 26287: {'lr': 0.00023111014123748953, 'samples': 13459456, 'steps': 26287, 'batch_loss/train': 0.7054243567399681}
+12/26/2021 10:22:50 - INFO - codeparrot_training - Step 26288: {'lr': 0.0002310944232028866, 'samples': 13459968, 'steps': 26288, 'batch_loss/train': 0.7156015376094729}
+12/26/2021 10:23:01 - INFO - codeparrot_training - Step 26289: {'lr': 0.00023107870524344512, 'samples': 13460480, 'steps': 26289, 'batch_loss/train': 0.7899724338203669}
+12/26/2021 10:23:11 - INFO - codeparrot_training - Step 26290: {'lr': 0.00023106298735922748, 'samples': 13460992, 'steps': 26290, 'batch_loss/train': 0.8630540855228901}
+12/26/2021 10:23:23 - INFO - codeparrot_training - Step 26291: {'lr': 0.00023104726955029627, 'samples': 13461504, 'steps': 26291, 'batch_loss/train': 0.7461704602465034}
+12/26/2021 10:23:34 - INFO - codeparrot_training - Step 26292: {'lr': 0.00023103155181671404, 'samples': 13462016, 'steps': 26292, 'batch_loss/train': 0.6873550089076161}
+12/26/2021 10:23:45 - INFO - codeparrot_training - Step 26293: {'lr': 0.00023101583415854313, 'samples': 13462528, 'steps': 26293, 'batch_loss/train': 0.779627209994942}
+12/26/2021 10:23:55 - INFO - codeparrot_training - Step 26294: {'lr': 0.000231000116575846, 'samples': 13463040, 'steps': 26294, 'batch_loss/train': 0.7277526403777301}
+12/26/2021 10:24:07 - INFO - codeparrot_training - Step 26295: {'lr': 0.0002309843990686854, 'samples': 13463552, 'steps': 26295, 'batch_loss/train': 0.6764907971955836}
+12/26/2021 10:24:18 - INFO - codeparrot_training - Step 26296: {'lr': 0.00023096868163712358, 'samples': 13464064, 'steps': 26296, 'batch_loss/train': 0.7934698138851672}
+12/26/2021 10:24:28 - INFO - codeparrot_training - Step 26297: {'lr': 0.00023095296428122318, 'samples': 13464576, 'steps': 26297, 'batch_loss/train': 0.7428763592615724}
+12/26/2021 10:24:41 - INFO - codeparrot_training - Step 26298: {'lr': 0.00023093724700104642, 'samples': 13465088, 'steps': 26298, 'batch_loss/train': 0.7903498820960522}
+12/26/2021 10:24:51 - INFO - codeparrot_training - Step 26299: {'lr': 0.00023092152979665615, 'samples': 13465600, 'steps': 26299, 'batch_loss/train': 0.8566628266125917}
+12/26/2021 10:25:02 - INFO - codeparrot_training - Step 26300: {'lr': 0.00023090581266811458, 'samples': 13466112, 'steps': 26300, 'batch_loss/train': 0.7581474827602506}
+12/26/2021 10:25:15 - INFO - codeparrot_training - Step 26301: {'lr': 0.00023089009561548434, 'samples': 13466624, 'steps': 26301, 'batch_loss/train': 0.8490446079522371}
+12/26/2021 10:25:26 - INFO - codeparrot_training - Step 26302: {'lr': 0.00023087437863882776, 'samples': 13467136, 'steps': 26302, 'batch_loss/train': 0.8227810487151146}
+12/26/2021 10:25:37 - INFO - codeparrot_training - Step 26303: {'lr': 0.0002308586617382075, 'samples': 13467648, 'steps': 26303, 'batch_loss/train': 0.6958869432564825}
+12/26/2021 10:25:49 - INFO - codeparrot_training - Step 26304: {'lr': 0.00023084294491368602, 'samples': 13468160, 'steps': 26304, 'batch_loss/train': 0.7977168820798397}
+12/26/2021 10:26:00 - INFO - codeparrot_training - Step 26305: {'lr': 0.00023082722816532565, 'samples': 13468672, 'steps': 26305, 'batch_loss/train': 0.7480731443502009}
+12/26/2021 10:26:10 - INFO - codeparrot_training - Step 26306: {'lr': 0.00023081151149318901, 'samples': 13469184, 'steps': 26306, 'batch_loss/train': 0.820012026000768}
+12/26/2021 10:26:21 - INFO - codeparrot_training - Step 26307: {'lr': 0.00023079579489733865, 'samples': 13469696, 'steps': 26307, 'batch_loss/train': 0.7486750222742558}
+12/26/2021 10:26:33 - INFO - codeparrot_training - Step 26308: {'lr': 0.00023078007837783695, 'samples': 13470208, 'steps': 26308, 'batch_loss/train': 0.8535340204834938}
+12/26/2021 10:26:43 - INFO - codeparrot_training - Step 26309: {'lr': 0.00023076436193474628, 'samples': 13470720, 'steps': 26309, 'batch_loss/train': 0.6990560758858919}
+12/26/2021 10:26:54 - INFO - codeparrot_training - Step 26310: {'lr': 0.0002307486455681293, 'samples': 13471232, 'steps': 26310, 'batch_loss/train': 0.7054136539809406}
+12/26/2021 10:27:08 - INFO - codeparrot_training - Step 26311: {'lr': 0.00023073292927804849, 'samples': 13471744, 'steps': 26311, 'batch_loss/train': 0.7281540732365102}
+12/26/2021 10:27:19 - INFO - codeparrot_training - Step 26312: {'lr': 0.00023071721306456623, 'samples': 13472256, 'steps': 26312, 'batch_loss/train': 0.6711474629119039}
+12/26/2021 10:27:29 - INFO - codeparrot_training - Step 26313: {'lr': 0.00023070149692774498, 'samples': 13472768, 'steps': 26313, 'batch_loss/train': 0.6797333809081465}
+12/26/2021 10:27:41 - INFO - codeparrot_training - Step 26314: {'lr': 0.0002306857808676474, 'samples': 13473280, 'steps': 26314, 'batch_loss/train': 0.7671933649107814}
+12/26/2021 10:27:52 - INFO - codeparrot_training - Step 26315: {'lr': 0.00023067006488433585, 'samples': 13473792, 'steps': 26315, 'batch_loss/train': 0.737089947797358}
+12/26/2021 10:28:03 - INFO - codeparrot_training - Step 26316: {'lr': 0.0002306543489778727, 'samples': 13474304, 'steps': 26316, 'batch_loss/train': 0.7085490059107542}
+12/26/2021 10:28:14 - INFO - codeparrot_training - Step 26317: {'lr': 0.00023063863314832074, 'samples': 13474816, 'steps': 26317, 'batch_loss/train': 0.7392154978588223}
+12/26/2021 10:28:28 - INFO - codeparrot_training - Step 26318: {'lr': 0.0002306229173957422, 'samples': 13475328, 'steps': 26318, 'batch_loss/train': 0.7095002988353372}
+12/26/2021 10:28:38 - INFO - codeparrot_training - Step 26319: {'lr': 0.00023060720172019962, 'samples': 13475840, 'steps': 26319, 'batch_loss/train': 0.5098478496074677}
+12/26/2021 10:28:49 - INFO - codeparrot_training - Step 26320: {'lr': 0.00023059148612175541, 'samples': 13476352, 'steps': 26320, 'batch_loss/train': 0.8426100062206388}
+12/26/2021 10:29:01 - INFO - codeparrot_training - Step 26321: {'lr': 0.0002305757706004722, 'samples': 13476864, 'steps': 26321, 'batch_loss/train': 0.7349335672333837}
+12/26/2021 10:29:12 - INFO - codeparrot_training - Step 26322: {'lr': 0.00023056005515641238, 'samples': 13477376, 'steps': 26322, 'batch_loss/train': 0.7051230588112958}
+12/26/2021 10:29:22 - INFO - codeparrot_training - Step 26323: {'lr': 0.00023054433978963846, 'samples': 13477888, 'steps': 26323, 'batch_loss/train': 0.68692380504217}
+12/26/2021 10:29:35 - INFO - codeparrot_training - Step 26324: {'lr': 0.0002305286245002128, 'samples': 13478400, 'steps': 26324, 'batch_loss/train': 0.7896018885076046}
+12/26/2021 10:29:46 - INFO - codeparrot_training - Step 26325: {'lr': 0.00023051290928819803, 'samples': 13478912, 'steps': 26325, 'batch_loss/train': 0.618038518470712}
+12/26/2021 10:29:56 - INFO - codeparrot_training - Step 26326: {'lr': 0.00023049719415365667, 'samples': 13479424, 'steps': 26326, 'batch_loss/train': 0.8173243775963783}
+12/26/2021 10:30:07 - INFO - codeparrot_training - Step 26327: {'lr': 0.00023048147909665096, 'samples': 13479936, 'steps': 26327, 'batch_loss/train': 1.0979345124214888}
+12/26/2021 10:30:20 - INFO - codeparrot_training - Step 26328: {'lr': 0.00023046576411724364, 'samples': 13480448, 'steps': 26328, 'batch_loss/train': 0.7573218373581767}
+12/26/2021 10:30:30 - INFO - codeparrot_training - Step 26329: {'lr': 0.00023045004921549698, 'samples': 13480960, 'steps': 26329, 'batch_loss/train': 0.7515492091188207}
+12/26/2021 10:30:41 - INFO - codeparrot_training - Step 26330: {'lr': 0.0002304343343914736, 'samples': 13481472, 'steps': 26330, 'batch_loss/train': 0.7778468923643231}
+12/26/2021 10:30:55 - INFO - codeparrot_training - Step 26331: {'lr': 0.00023041861964523584, 'samples': 13481984, 'steps': 26331, 'batch_loss/train': 0.7588240881450474}
+12/26/2021 10:31:06 - INFO - codeparrot_training - Step 26332: {'lr': 0.0002304029049768463, 'samples': 13482496, 'steps': 26332, 'batch_loss/train': 0.7967244577594101}
+12/26/2021 10:31:16 - INFO - codeparrot_training - Step 26333: {'lr': 0.00023038719038636745, 'samples': 13483008, 'steps': 26333, 'batch_loss/train': 0.6590705807320774}
+12/26/2021 10:31:28 - INFO - codeparrot_training - Step 26334: {'lr': 0.00023037147587386177, 'samples': 13483520, 'steps': 26334, 'batch_loss/train': 0.6673543965443969}
+12/26/2021 10:31:39 - INFO - codeparrot_training - Step 26335: {'lr': 0.00023035576143939153, 'samples': 13484032, 'steps': 26335, 'batch_loss/train': 0.6809943860862404}
+12/26/2021 10:31:50 - INFO - codeparrot_training - Step 26336: {'lr': 0.00023034004708301942, 'samples': 13484544, 'steps': 26336, 'batch_loss/train': 0.7674567709909752}
+12/26/2021 10:32:00 - INFO - codeparrot_training - Step 26337: {'lr': 0.00023032433280480795, 'samples': 13485056, 'steps': 26337, 'batch_loss/train': 0.8203878425993025}
+12/26/2021 10:32:12 - INFO - codeparrot_training - Step 26338: {'lr': 0.00023030861860481942, 'samples': 13485568, 'steps': 26338, 'batch_loss/train': 0.7808148190379143}
+12/26/2021 10:32:23 - INFO - codeparrot_training - Step 26339: {'lr': 0.00023029290448311636, 'samples': 13486080, 'steps': 26339, 'batch_loss/train': 0.699969279114157}
+12/26/2021 10:32:34 - INFO - codeparrot_training - Step 26340: {'lr': 0.00023027719043976132, 'samples': 13486592, 'steps': 26340, 'batch_loss/train': 0.7212994303554296}
+12/26/2021 10:32:48 - INFO - codeparrot_training - Step 26341: {'lr': 0.00023026147647481674, 'samples': 13487104, 'steps': 26341, 'batch_loss/train': 0.7416748786345124}
+12/26/2021 10:32:58 - INFO - codeparrot_training - Step 26342: {'lr': 0.00023024576258834497, 'samples': 13487616, 'steps': 26342, 'batch_loss/train': 0.5976918248925358}
+12/26/2021 10:33:09 - INFO - codeparrot_training - Step 26343: {'lr': 0.00023023004878040875, 'samples': 13488128, 'steps': 26343, 'batch_loss/train': 0.7260345639660954}
+12/26/2021 10:33:21 - INFO - codeparrot_training - Step 26344: {'lr': 0.00023021433505107037, 'samples': 13488640, 'steps': 26344, 'batch_loss/train': 0.8050139779224992}
+12/26/2021 10:33:32 - INFO - codeparrot_training - Step 26345: {'lr': 0.00023019862140039225, 'samples': 13489152, 'steps': 26345, 'batch_loss/train': 0.8074117798241787}
+12/26/2021 10:33:42 - INFO - codeparrot_training - Step 26346: {'lr': 0.0002301829078284369, 'samples': 13489664, 'steps': 26346, 'batch_loss/train': 0.5613970353151672}
+12/26/2021 10:33:53 - INFO - codeparrot_training - Step 26347: {'lr': 0.00023016719433526693, 'samples': 13490176, 'steps': 26347, 'batch_loss/train': 0.6612517446046695}
+12/26/2021 10:34:07 - INFO - codeparrot_training - Step 26348: {'lr': 0.00023015148092094464, 'samples': 13490688, 'steps': 26348, 'batch_loss/train': 0.8071663295850158}
+12/26/2021 10:34:18 - INFO - codeparrot_training - Step 26349: {'lr': 0.00023013576758553262, 'samples': 13491200, 'steps': 26349, 'batch_loss/train': 0.73896240349859}
+12/26/2021 10:34:28 - INFO - codeparrot_training - Step 26350: {'lr': 0.0002301200543290932, 'samples': 13491712, 'steps': 26350, 'batch_loss/train': 0.7885100422427058}
+12/26/2021 10:34:40 - INFO - codeparrot_training - Step 26351: {'lr': 0.000230104341151689, 'samples': 13492224, 'steps': 26351, 'batch_loss/train': 0.7444465421140194}
+12/26/2021 10:34:51 - INFO - codeparrot_training - Step 26352: {'lr': 0.00023008862805338245, 'samples': 13492736, 'steps': 26352, 'batch_loss/train': 0.769721882417798}
+12/26/2021 10:35:02 - INFO - codeparrot_training - Step 26353: {'lr': 0.00023007291503423588, 'samples': 13493248, 'steps': 26353, 'batch_loss/train': 0.7396321590058506}
+12/26/2021 10:35:15 - INFO - codeparrot_training - Step 26354: {'lr': 0.00023005720209431203, 'samples': 13493760, 'steps': 26354, 'batch_loss/train': 0.8253393084742129}
+12/26/2021 10:35:26 - INFO - codeparrot_training - Step 26355: {'lr': 0.00023004148923367313, 'samples': 13494272, 'steps': 26355, 'batch_loss/train': 0.8429589280858636}
+12/26/2021 10:35:37 - INFO - codeparrot_training - Step 26356: {'lr': 0.0002300257764523818, 'samples': 13494784, 'steps': 26356, 'batch_loss/train': 0.7415354885160923}
+12/26/2021 10:35:47 - INFO - codeparrot_training - Step 26357: {'lr': 0.00023001006375050032, 'samples': 13495296, 'steps': 26357, 'batch_loss/train': 0.7677646148949862}
+12/26/2021 10:35:59 - INFO - codeparrot_training - Step 26358: {'lr': 0.00022999435112809133, 'samples': 13495808, 'steps': 26358, 'batch_loss/train': 0.7555956775322556}
+12/26/2021 10:36:10 - INFO - codeparrot_training - Step 26359: {'lr': 0.0002299786385852173, 'samples': 13496320, 'steps': 26359, 'batch_loss/train': 0.6908195983851328}
+12/26/2021 10:36:21 - INFO - codeparrot_training - Step 26360: {'lr': 0.00022996292612194062, 'samples': 13496832, 'steps': 26360, 'batch_loss/train': 0.6932414500042796}
+12/26/2021 10:36:33 - INFO - codeparrot_training - Step 26361: {'lr': 0.00022994721373832368, 'samples': 13497344, 'steps': 26361, 'batch_loss/train': 0.8173355339094996}
+12/26/2021 10:36:44 - INFO - codeparrot_training - Step 26362: {'lr': 0.0002299315014344292, 'samples': 13497856, 'steps': 26362, 'batch_loss/train': 0.730623094830662}
+12/26/2021 10:36:54 - INFO - codeparrot_training - Step 26363: {'lr': 0.00022991578921031946, 'samples': 13498368, 'steps': 26363, 'batch_loss/train': 0.6293481294997036}
+12/26/2021 10:37:06 - INFO - codeparrot_training - Step 26364: {'lr': 0.00022990007706605686, 'samples': 13498880, 'steps': 26364, 'batch_loss/train': 0.7629873035475612}
+12/26/2021 10:37:17 - INFO - codeparrot_training - Step 26365: {'lr': 0.00022988436500170404, 'samples': 13499392, 'steps': 26365, 'batch_loss/train': 0.818441272713244}
+12/26/2021 10:37:28 - INFO - codeparrot_training - Step 26366: {'lr': 0.0002298686530173234, 'samples': 13499904, 'steps': 26366, 'batch_loss/train': 0.7308141118846834}
+12/26/2021 10:37:40 - INFO - codeparrot_training - Step 26367: {'lr': 0.00022985294111297738, 'samples': 13500416, 'steps': 26367, 'batch_loss/train': 0.7774022333323956}
+12/26/2021 10:37:51 - INFO - codeparrot_training - Step 26368: {'lr': 0.00022983722928872837, 'samples': 13500928, 'steps': 26368, 'batch_loss/train': 0.6795993172563612}
+12/26/2021 10:38:01 - INFO - codeparrot_training - Step 26369: {'lr': 0.00022982151754463907, 'samples': 13501440, 'steps': 26369, 'batch_loss/train': 0.7356705875135958}
+12/26/2021 10:38:12 - INFO - codeparrot_training - Step 26370: {'lr': 0.00022980580588077176, 'samples': 13501952, 'steps': 26370, 'batch_loss/train': 0.7589896535500884}
+12/26/2021 10:38:26 - INFO - codeparrot_training - Step 26371: {'lr': 0.0002297900942971889, 'samples': 13502464, 'steps': 26371, 'batch_loss/train': 0.7688696039840579}
+12/26/2021 10:38:37 - INFO - codeparrot_training - Step 26372: {'lr': 0.00022977438279395297, 'samples': 13502976, 'steps': 26372, 'batch_loss/train': 0.8107136506587267}
+12/26/2021 10:38:47 - INFO - codeparrot_training - Step 26373: {'lr': 0.00022975867137112652, 'samples': 13503488, 'steps': 26373, 'batch_loss/train': 0.8367642010562122}
+12/26/2021 10:38:59 - INFO - codeparrot_training - Step 26374: {'lr': 0.00022974296002877193, 'samples': 13504000, 'steps': 26374, 'batch_loss/train': 0.7689884295687079}
+12/26/2021 10:39:10 - INFO - codeparrot_training - Step 26375: {'lr': 0.00022972724876695172, 'samples': 13504512, 'steps': 26375, 'batch_loss/train': 0.6898940037935972}
+12/26/2021 10:39:21 - INFO - codeparrot_training - Step 26376: {'lr': 0.0002297115375857282, 'samples': 13505024, 'steps': 26376, 'batch_loss/train': 0.735118858050555}
+12/26/2021 10:39:34 - INFO - codeparrot_training - Step 26377: {'lr': 0.00022969582648516403, 'samples': 13505536, 'steps': 26377, 'batch_loss/train': 0.8226137352176011}
+12/26/2021 10:39:45 - INFO - codeparrot_training - Step 26378: {'lr': 0.00022968011546532159, 'samples': 13506048, 'steps': 26378, 'batch_loss/train': 0.7421882287599146}
+12/26/2021 10:39:55 - INFO - codeparrot_training - Step 26379: {'lr': 0.00022966440452626324, 'samples': 13506560, 'steps': 26379, 'batch_loss/train': 0.662020830437541}
+12/26/2021 10:40:06 - INFO - codeparrot_training - Step 26380: {'lr': 0.00022964869366805168, 'samples': 13507072, 'steps': 26380, 'batch_loss/train': 0.7234824607148767}
+12/26/2021 10:40:18 - INFO - codeparrot_training - Step 26381: {'lr': 0.00022963298289074916, 'samples': 13507584, 'steps': 26381, 'batch_loss/train': 0.8299737265333533}
+12/26/2021 10:40:28 - INFO - codeparrot_training - Step 26382: {'lr': 0.00022961727219441828, 'samples': 13508096, 'steps': 26382, 'batch_loss/train': 0.8051335453055799}
+12/26/2021 10:40:39 - INFO - codeparrot_training - Step 26383: {'lr': 0.0002296015615791213, 'samples': 13508608, 'steps': 26383, 'batch_loss/train': 0.7475555017590523}
+12/26/2021 10:40:51 - INFO - codeparrot_training - Step 26384: {'lr': 0.00022958585104492085, 'samples': 13509120, 'steps': 26384, 'batch_loss/train': 0.6993989022448659}
+12/26/2021 10:41:02 - INFO - codeparrot_training - Step 26385: {'lr': 0.00022957014059187943, 'samples': 13509632, 'steps': 26385, 'batch_loss/train': 0.7126201307401061}
+12/26/2021 10:41:12 - INFO - codeparrot_training - Step 26386: {'lr': 0.00022955443022005937, 'samples': 13510144, 'steps': 26386, 'batch_loss/train': 1.0187919670715928}
+12/26/2021 10:41:26 - INFO - codeparrot_training - Step 26387: {'lr': 0.0002295387199295231, 'samples': 13510656, 'steps': 26387, 'batch_loss/train': 0.8527535377070308}
+12/26/2021 10:41:37 - INFO - codeparrot_training - Step 26388: {'lr': 0.00022952300972033327, 'samples': 13511168, 'steps': 26388, 'batch_loss/train': 0.76249875780195}
+12/26/2021 10:41:47 - INFO - codeparrot_training - Step 26389: {'lr': 0.00022950729959255223, 'samples': 13511680, 'steps': 26389, 'batch_loss/train': 0.7615324249491096}
+12/26/2021 10:41:58 - INFO - codeparrot_training - Step 26390: {'lr': 0.0002294915895462423, 'samples': 13512192, 'steps': 26390, 'batch_loss/train': 0.7242501545697451}
+12/26/2021 10:42:10 - INFO - codeparrot_training - Step 26391: {'lr': 0.00022947587958146613, 'samples': 13512704, 'steps': 26391, 'batch_loss/train': 0.7946090510813519}
+12/26/2021 10:42:21 - INFO - codeparrot_training - Step 26392: {'lr': 0.00022946016969828617, 'samples': 13513216, 'steps': 26392, 'batch_loss/train': 0.8139668125659227}
+12/26/2021 10:42:32 - INFO - codeparrot_training - Step 26393: {'lr': 0.00022944445989676478, 'samples': 13513728, 'steps': 26393, 'batch_loss/train': 0.9001809414476156}
+12/26/2021 10:42:45 - INFO - codeparrot_training - Step 26394: {'lr': 0.00022942875017696436, 'samples': 13514240, 'steps': 26394, 'batch_loss/train': 0.8559263704810292}
+12/26/2021 10:42:56 - INFO - codeparrot_training - Step 26395: {'lr': 0.00022941304053894759, 'samples': 13514752, 'steps': 26395, 'batch_loss/train': 0.7711971933022141}
+12/26/2021 10:43:07 - INFO - codeparrot_training - Step 26396: {'lr': 0.00022939733098277678, 'samples': 13515264, 'steps': 26396, 'batch_loss/train': 0.6844325183192268}
+12/26/2021 10:43:19 - INFO - codeparrot_training - Step 26397: {'lr': 0.0002293816215085144, 'samples': 13515776, 'steps': 26397, 'batch_loss/train': 0.7345785340294242}
+12/26/2021 10:43:29 - INFO - codeparrot_training - Step 26398: {'lr': 0.0002293659121162228, 'samples': 13516288, 'steps': 26398, 'batch_loss/train': 0.7602941747754812}
+12/26/2021 10:43:40 - INFO - codeparrot_training - Step 26399: {'lr': 0.0002293502028059647, 'samples': 13516800, 'steps': 26399, 'batch_loss/train': 0.8566155969165266}
+12/26/2021 10:43:52 - INFO - codeparrot_training - Step 26400: {'lr': 0.00022933449357780234, 'samples': 13517312, 'steps': 26400, 'batch_loss/train': 0.7915117861703038}
+12/26/2021 10:44:03 - INFO - codeparrot_training - Step 26401: {'lr': 0.00022931878443179814, 'samples': 13517824, 'steps': 26401, 'batch_loss/train': 0.7434261776506901}
+12/26/2021 10:44:13 - INFO - codeparrot_training - Step 26402: {'lr': 0.00022930307536801476, 'samples': 13518336, 'steps': 26402, 'batch_loss/train': 0.8669106159359217}
+12/26/2021 10:44:24 - INFO - codeparrot_training - Step 26403: {'lr': 0.00022928736638651447, 'samples': 13518848, 'steps': 26403, 'batch_loss/train': 0.8835240163025446}
+12/26/2021 10:44:36 - INFO - codeparrot_training - Step 26404: {'lr': 0.0002292716574873599, 'samples': 13519360, 'steps': 26404, 'batch_loss/train': 0.7323910668492317}
+12/26/2021 10:44:47 - INFO - codeparrot_training - Step 26405: {'lr': 0.00022925594867061324, 'samples': 13519872, 'steps': 26405, 'batch_loss/train': 0.6597953522577882}
+12/26/2021 10:44:57 - INFO - codeparrot_training - Step 26406: {'lr': 0.00022924023993633718, 'samples': 13520384, 'steps': 26406, 'batch_loss/train': 0.7838473506271839}
+12/26/2021 10:45:12 - INFO - codeparrot_training - Step 26407: {'lr': 0.0002292245312845941, 'samples': 13520896, 'steps': 26407, 'batch_loss/train': 0.7996250081341714}
+12/26/2021 10:45:22 - INFO - codeparrot_training - Step 26408: {'lr': 0.00022920882271544647, 'samples': 13521408, 'steps': 26408, 'batch_loss/train': 0.8739825012162328}
+12/26/2021 10:45:33 - INFO - codeparrot_training - Step 26409: {'lr': 0.00022919311422895657, 'samples': 13521920, 'steps': 26409, 'batch_loss/train': 0.6460348220425658}
+12/26/2021 10:45:45 - INFO - codeparrot_training - Step 26410: {'lr': 0.0002291774058251871, 'samples': 13522432, 'steps': 26410, 'batch_loss/train': 0.7829908579587936}
+12/26/2021 10:45:55 - INFO - codeparrot_training - Step 26411: {'lr': 0.00022916169750420042, 'samples': 13522944, 'steps': 26411, 'batch_loss/train': 0.7151548396795988}
+12/26/2021 10:46:06 - INFO - codeparrot_training - Step 26412: {'lr': 0.0002291459892660589, 'samples': 13523456, 'steps': 26412, 'batch_loss/train': 0.7838390865363181}
+12/26/2021 10:46:17 - INFO - codeparrot_training - Step 26413: {'lr': 0.00022913028111082502, 'samples': 13523968, 'steps': 26413, 'batch_loss/train': 0.7295641186647117}
+12/26/2021 10:46:29 - INFO - codeparrot_training - Step 26414: {'lr': 0.00022911457303856133, 'samples': 13524480, 'steps': 26414, 'batch_loss/train': 0.7370542755816132}
+12/26/2021 10:46:40 - INFO - codeparrot_training - Step 26415: {'lr': 0.00022909886504933024, 'samples': 13524992, 'steps': 26415, 'batch_loss/train': 0.6185264443629421}
+12/26/2021 10:46:50 - INFO - codeparrot_training - Step 26416: {'lr': 0.0002290831571431941, 'samples': 13525504, 'steps': 26416, 'batch_loss/train': 0.7544240991119295}
+12/26/2021 10:47:04 - INFO - codeparrot_training - Step 26417: {'lr': 0.00022906744932021545, 'samples': 13526016, 'steps': 26417, 'batch_loss/train': 0.723677912261337}
+12/26/2021 10:47:15 - INFO - codeparrot_training - Step 26418: {'lr': 0.00022905174158045676, 'samples': 13526528, 'steps': 26418, 'batch_loss/train': 0.783828909508884}
+12/26/2021 10:47:26 - INFO - codeparrot_training - Step 26419: {'lr': 0.00022903603392398039, 'samples': 13527040, 'steps': 26419, 'batch_loss/train': 0.7433942770585418}
+12/26/2021 10:47:38 - INFO - codeparrot_training - Step 26420: {'lr': 0.0002290203263508488, 'samples': 13527552, 'steps': 26420, 'batch_loss/train': 0.6123960199183784}
+12/26/2021 10:47:48 - INFO - codeparrot_training - Step 26421: {'lr': 0.00022900461886112456, 'samples': 13528064, 'steps': 26421, 'batch_loss/train': 0.6970004476606846}
+12/26/2021 10:47:59 - INFO - codeparrot_training - Step 26422: {'lr': 0.00022898891145486998, 'samples': 13528576, 'steps': 26422, 'batch_loss/train': 0.7805371802533045}
+12/26/2021 10:48:10 - INFO - codeparrot_training - Step 26423: {'lr': 0.0002289732041321476, 'samples': 13529088, 'steps': 26423, 'batch_loss/train': 0.7916224440559745}
+12/26/2021 10:48:24 - INFO - codeparrot_training - Step 26424: {'lr': 0.0002289574968930197, 'samples': 13529600, 'steps': 26424, 'batch_loss/train': 1.1662898873910308}
+12/26/2021 10:48:34 - INFO - codeparrot_training - Step 26425: {'lr': 0.00022894178973754899, 'samples': 13530112, 'steps': 26425, 'batch_loss/train': 0.713077993132174}
+12/26/2021 10:48:45 - INFO - codeparrot_training - Step 26426: {'lr': 0.00022892608266579767, 'samples': 13530624, 'steps': 26426, 'batch_loss/train': 0.7753490759059787}
+12/26/2021 10:48:57 - INFO - codeparrot_training - Step 26427: {'lr': 0.00022891037567782826, 'samples': 13531136, 'steps': 26427, 'batch_loss/train': 0.7921460054349154}
+12/26/2021 10:49:08 - INFO - codeparrot_training - Step 26428: {'lr': 0.00022889466877370335, 'samples': 13531648, 'steps': 26428, 'batch_loss/train': 0.6522731338627636}
+12/26/2021 10:49:18 - INFO - codeparrot_training - Step 26429: {'lr': 0.0002288789619534852, 'samples': 13532160, 'steps': 26429, 'batch_loss/train': 0.673866315279156}
+12/26/2021 10:49:29 - INFO - codeparrot_training - Step 26430: {'lr': 0.0002288632552172364, 'samples': 13532672, 'steps': 26430, 'batch_loss/train': 0.7447085967287421}
+12/26/2021 10:49:41 - INFO - codeparrot_training - Step 26431: {'lr': 0.0002288475485650192, 'samples': 13533184, 'steps': 26431, 'batch_loss/train': 0.7540964977815747}
+12/26/2021 10:49:52 - INFO - codeparrot_training - Step 26432: {'lr': 0.0002288318419968962, 'samples': 13533696, 'steps': 26432, 'batch_loss/train': 0.6712830968899652}
+12/26/2021 10:50:02 - INFO - codeparrot_training - Step 26433: {'lr': 0.00022881613551292988, 'samples': 13534208, 'steps': 26433, 'batch_loss/train': 0.6730381542583928}
+12/26/2021 10:50:16 - INFO - codeparrot_training - Step 26434: {'lr': 0.00022880042911318257, 'samples': 13534720, 'steps': 26434, 'batch_loss/train': 0.7790232702391222}
+12/26/2021 10:50:27 - INFO - codeparrot_training - Step 26435: {'lr': 0.00022878472279771667, 'samples': 13535232, 'steps': 26435, 'batch_loss/train': 0.770393610233441}
+12/26/2021 10:50:38 - INFO - codeparrot_training - Step 26436: {'lr': 0.00022876901656659475, 'samples': 13535744, 'steps': 26436, 'batch_loss/train': 0.7680381932295859}
+12/26/2021 10:50:50 - INFO - codeparrot_training - Step 26437: {'lr': 0.00022875331041987925, 'samples': 13536256, 'steps': 26437, 'batch_loss/train': 0.8224147474393249}
+12/26/2021 10:51:00 - INFO - codeparrot_training - Step 26438: {'lr': 0.00022873760435763248, 'samples': 13536768, 'steps': 26438, 'batch_loss/train': 0.7256800676696002}
+12/26/2021 10:51:11 - INFO - codeparrot_training - Step 26439: {'lr': 0.000228721898379917, 'samples': 13537280, 'steps': 26439, 'batch_loss/train': 0.6698136394843459}
+12/26/2021 10:51:23 - INFO - codeparrot_training - Step 26440: {'lr': 0.00022870619248679526, 'samples': 13537792, 'steps': 26440, 'batch_loss/train': 0.7587280552834272}
+12/26/2021 10:51:34 - INFO - codeparrot_training - Step 26441: {'lr': 0.00022869048667832962, 'samples': 13538304, 'steps': 26441, 'batch_loss/train': 0.6718999044969678}
+12/26/2021 10:51:44 - INFO - codeparrot_training - Step 26442: {'lr': 0.0002286747809545825, 'samples': 13538816, 'steps': 26442, 'batch_loss/train': 0.704154082108289}
+12/26/2021 10:51:55 - INFO - codeparrot_training - Step 26443: {'lr': 0.00022865907531561652, 'samples': 13539328, 'steps': 26443, 'batch_loss/train': 0.7484538580756634}
+12/26/2021 10:52:07 - INFO - codeparrot_training - Step 26444: {'lr': 0.000228643369761494, 'samples': 13539840, 'steps': 26444, 'batch_loss/train': 0.7924815067090094}
+12/26/2021 10:52:18 - INFO - codeparrot_training - Step 26445: {'lr': 0.0002286276642922773, 'samples': 13540352, 'steps': 26445, 'batch_loss/train': 0.7965269037522376}
+12/26/2021 10:52:29 - INFO - codeparrot_training - Step 26446: {'lr': 0.00022861195890802892, 'samples': 13540864, 'steps': 26446, 'batch_loss/train': 0.6562613951973617}
+12/26/2021 10:52:42 - INFO - codeparrot_training - Step 26447: {'lr': 0.00022859625360881138, 'samples': 13541376, 'steps': 26447, 'batch_loss/train': 0.8279432775452733}
+12/26/2021 10:52:53 - INFO - codeparrot_training - Step 26448: {'lr': 0.00022858054839468704, 'samples': 13541888, 'steps': 26448, 'batch_loss/train': 0.7995092160999775}
+12/26/2021 10:53:04 - INFO - codeparrot_training - Step 26449: {'lr': 0.00022856484326571835, 'samples': 13542400, 'steps': 26449, 'batch_loss/train': 0.6666943458840251}
+12/26/2021 10:53:16 - INFO - codeparrot_training - Step 26450: {'lr': 0.00022854913822196768, 'samples': 13542912, 'steps': 26450, 'batch_loss/train': 0.7700927336700261}
+12/26/2021 10:53:26 - INFO - codeparrot_training - Step 26451: {'lr': 0.0002285334332634976, 'samples': 13543424, 'steps': 26451, 'batch_loss/train': 0.7685429723933339}
+12/26/2021 10:53:37 - INFO - codeparrot_training - Step 26452: {'lr': 0.00022851772839037053, 'samples': 13543936, 'steps': 26452, 'batch_loss/train': 0.8479480971582234}
+12/26/2021 10:53:48 - INFO - codeparrot_training - Step 26453: {'lr': 0.00022850202360264873, 'samples': 13544448, 'steps': 26453, 'batch_loss/train': 0.8691898789256811}
+12/26/2021 10:54:00 - INFO - codeparrot_training - Step 26454: {'lr': 0.00022848631890039494, 'samples': 13544960, 'steps': 26454, 'batch_loss/train': 0.7406675467500463}
+12/26/2021 10:54:10 - INFO - codeparrot_training - Step 26455: {'lr': 0.0002284706142836713, 'samples': 13545472, 'steps': 26455, 'batch_loss/train': 0.7297307663247921}
+12/26/2021 10:54:21 - INFO - codeparrot_training - Step 26456: {'lr': 0.00022845490975254042, 'samples': 13545984, 'steps': 26456, 'batch_loss/train': 0.8356396723538637}
+12/26/2021 10:54:35 - INFO - codeparrot_training - Step 26457: {'lr': 0.00022843920530706463, 'samples': 13546496, 'steps': 26457, 'batch_loss/train': 0.6902227960526943}
+12/26/2021 10:54:45 - INFO - codeparrot_training - Step 26458: {'lr': 0.00022842350094730643, 'samples': 13547008, 'steps': 26458, 'batch_loss/train': 0.7863526865839958}
+12/26/2021 10:54:56 - INFO - codeparrot_training - Step 26459: {'lr': 0.00022840779667332832, 'samples': 13547520, 'steps': 26459, 'batch_loss/train': 0.8528694165870547}
+12/26/2021 10:55:08 - INFO - codeparrot_training - Step 26460: {'lr': 0.00022839209248519266, 'samples': 13548032, 'steps': 26460, 'batch_loss/train': 0.8173585655167699}
+12/26/2021 10:55:19 - INFO - codeparrot_training - Step 26461: {'lr': 0.00022837638838296175, 'samples': 13548544, 'steps': 26461, 'batch_loss/train': 0.7685276011470705}
+12/26/2021 10:55:29 - INFO - codeparrot_training - Step 26462: {'lr': 0.0002283606843666982, 'samples': 13549056, 'steps': 26462, 'batch_loss/train': 0.6973487767390907}
+12/26/2021 10:55:43 - INFO - codeparrot_training - Step 26463: {'lr': 0.00022834498043646447, 'samples': 13549568, 'steps': 26463, 'batch_loss/train': 0.762430218514055}
+12/26/2021 10:55:54 - INFO - codeparrot_training - Step 26464: {'lr': 0.00022832927659232282, 'samples': 13550080, 'steps': 26464, 'batch_loss/train': 0.7232519425451756}
+12/26/2021 10:56:05 - INFO - codeparrot_training - Step 26465: {'lr': 0.00022831357283433585, 'samples': 13550592, 'steps': 26465, 'batch_loss/train': 0.8598402077332139}
+12/26/2021 10:56:15 - INFO - codeparrot_training - Step 26466: {'lr': 0.00022829786916256595, 'samples': 13551104, 'steps': 26466, 'batch_loss/train': 0.7833577319979668}
+12/26/2021 10:56:27 - INFO - codeparrot_training - Step 26467: {'lr': 0.00022828216557707552, 'samples': 13551616, 'steps': 26467, 'batch_loss/train': 0.7480811825953424}
+12/26/2021 10:56:38 - INFO - codeparrot_training - Step 26468: {'lr': 0.0002282664620779269, 'samples': 13552128, 'steps': 26468, 'batch_loss/train': 0.768919677939266}
+12/26/2021 10:56:49 - INFO - codeparrot_training - Step 26469: {'lr': 0.00022825075866518275, 'samples': 13552640, 'steps': 26469, 'batch_loss/train': 0.7537871890235692}
+12/26/2021 10:57:01 - INFO - codeparrot_training - Step 26470: {'lr': 0.00022823505533890533, 'samples': 13553152, 'steps': 26470, 'batch_loss/train': 0.7461477373726666}
+12/26/2021 10:57:11 - INFO - codeparrot_training - Step 26471: {'lr': 0.0002282193520991571, 'samples': 13553664, 'steps': 26471, 'batch_loss/train': 0.6697752242907882}
+12/26/2021 10:57:22 - INFO - codeparrot_training - Step 26472: {'lr': 0.00022820364894600046, 'samples': 13554176, 'steps': 26472, 'batch_loss/train': 0.7280890559777617}
+12/26/2021 10:57:36 - INFO - codeparrot_training - Step 26473: {'lr': 0.00022818794587949797, 'samples': 13554688, 'steps': 26473, 'batch_loss/train': 0.7584407897666097}
+12/26/2021 10:57:46 - INFO - codeparrot_training - Step 26474: {'lr': 0.00022817224289971188, 'samples': 13555200, 'steps': 26474, 'batch_loss/train': 0.743313170503825}
+12/26/2021 10:57:57 - INFO - codeparrot_training - Step 26475: {'lr': 0.00022815654000670472, 'samples': 13555712, 'steps': 26475, 'batch_loss/train': 0.7699325624853373}
+12/26/2021 10:58:08 - INFO - codeparrot_training - Step 26476: {'lr': 0.000228140837200539, 'samples': 13556224, 'steps': 26476, 'batch_loss/train': 0.7444582832977176}
+12/26/2021 10:58:20 - INFO - codeparrot_training - Step 26477: {'lr': 0.00022812513448127697, 'samples': 13556736, 'steps': 26477, 'batch_loss/train': 0.7263819458894432}
+12/26/2021 10:58:30 - INFO - codeparrot_training - Step 26478: {'lr': 0.00022810943184898124, 'samples': 13557248, 'steps': 26478, 'batch_loss/train': 0.6607896109344438}
+12/26/2021 10:58:41 - INFO - codeparrot_training - Step 26479: {'lr': 0.000228093729303714, 'samples': 13557760, 'steps': 26479, 'batch_loss/train': 0.7659159004688263}
+12/26/2021 10:58:53 - INFO - codeparrot_training - Step 26480: {'lr': 0.00022807802684553797, 'samples': 13558272, 'steps': 26480, 'batch_loss/train': 0.6615124945528805}
+12/26/2021 10:59:04 - INFO - codeparrot_training - Step 26481: {'lr': 0.00022806232447451533, 'samples': 13558784, 'steps': 26481, 'batch_loss/train': 0.8849879186600447}
+12/26/2021 10:59:14 - INFO - codeparrot_training - Step 26482: {'lr': 0.00022804662219070868, 'samples': 13559296, 'steps': 26482, 'batch_loss/train': 0.8220686316490173}
+12/26/2021 10:59:26 - INFO - codeparrot_training - Step 26483: {'lr': 0.00022803091999418028, 'samples': 13559808, 'steps': 26483, 'batch_loss/train': 0.7542761135846376}
+12/26/2021 10:59:37 - INFO - codeparrot_training - Step 26484: {'lr': 0.00022801521788499266, 'samples': 13560320, 'steps': 26484, 'batch_loss/train': 0.7142631788738072}
+12/26/2021 10:59:48 - INFO - codeparrot_training - Step 26485: {'lr': 0.00022799951586320832, 'samples': 13560832, 'steps': 26485, 'batch_loss/train': 0.7874725563451648}
+12/26/2021 10:59:58 - INFO - codeparrot_training - Step 26486: {'lr': 0.00022798381392888953, 'samples': 13561344, 'steps': 26486, 'batch_loss/train': 0.76888239197433}
+12/26/2021 11:00:12 - INFO - codeparrot_training - Step 26487: {'lr': 0.00022796811208209873, 'samples': 13561856, 'steps': 26487, 'batch_loss/train': 0.7201489664148539}
+12/26/2021 11:00:23 - INFO - codeparrot_training - Step 26488: {'lr': 0.0002279524103228985, 'samples': 13562368, 'steps': 26488, 'batch_loss/train': 0.8560398947447538}
+12/26/2021 11:00:33 - INFO - codeparrot_training - Step 26489: {'lr': 0.00022793670865135118, 'samples': 13562880, 'steps': 26489, 'batch_loss/train': 0.7499525765888393}
+12/26/2021 11:00:46 - INFO - codeparrot_training - Step 26490: {'lr': 0.00022792100706751906, 'samples': 13563392, 'steps': 26490, 'batch_loss/train': 0.7551544792950153}
+12/26/2021 11:00:56 - INFO - codeparrot_training - Step 26491: {'lr': 0.0002279053055714647, 'samples': 13563904, 'steps': 26491, 'batch_loss/train': 0.7085969555191696}
+12/26/2021 11:01:07 - INFO - codeparrot_training - Step 26492: {'lr': 0.0002278896041632506, 'samples': 13564416, 'steps': 26492, 'batch_loss/train': 0.7493948573246598}
+12/26/2021 11:01:22 - INFO - codeparrot_training - Step 26493: {'lr': 0.00022787390284293905, 'samples': 13564928, 'steps': 26493, 'batch_loss/train': 0.925752729177475}
+12/26/2021 11:01:32 - INFO - codeparrot_training - Step 26494: {'lr': 0.0002278582016105924, 'samples': 13565440, 'steps': 26494, 'batch_loss/train': 0.7138329073786736}
+12/26/2021 11:01:43 - INFO - codeparrot_training - Step 26495: {'lr': 0.00022784250046627331, 'samples': 13565952, 'steps': 26495, 'batch_loss/train': 1.5136159602552652}
+12/26/2021 11:01:54 - INFO - codeparrot_training - Step 26496: {'lr': 0.00022782679941004403, 'samples': 13566464, 'steps': 26496, 'batch_loss/train': 0.6118942002067342}
+12/26/2021 11:02:06 - INFO - codeparrot_training - Step 26497: {'lr': 0.00022781109844196708, 'samples': 13566976, 'steps': 26497, 'batch_loss/train': 0.7639608486206271}
+12/26/2021 11:02:16 - INFO - codeparrot_training - Step 26498: {'lr': 0.0002277953975621047, 'samples': 13567488, 'steps': 26498, 'batch_loss/train': 0.6714569074683823}
+12/26/2021 11:02:27 - INFO - codeparrot_training - Step 26499: {'lr': 0.0002277796967705195, 'samples': 13568000, 'steps': 26499, 'batch_loss/train': 0.6376433650148101}
+12/26/2021 11:02:39 - INFO - codeparrot_training - Step 26500: {'lr': 0.00022776399606727385, 'samples': 13568512, 'steps': 26500, 'batch_loss/train': 0.7740937070921063}
+12/26/2021 11:02:50 - INFO - codeparrot_training - Step 26501: {'lr': 0.00022774829545243007, 'samples': 13569024, 'steps': 26501, 'batch_loss/train': 0.6921401573345065}
+12/26/2021 11:03:00 - INFO - codeparrot_training - Step 26502: {'lr': 0.00022773259492605077, 'samples': 13569536, 'steps': 26502, 'batch_loss/train': 0.7668122593313456}
+12/26/2021 11:03:11 - INFO - codeparrot_training - Step 26503: {'lr': 0.00022771689448819823, 'samples': 13570048, 'steps': 26503, 'batch_loss/train': 0.7857512142509222}
+12/26/2021 11:03:25 - INFO - codeparrot_training - Step 26504: {'lr': 0.00022770119413893498, 'samples': 13570560, 'steps': 26504, 'batch_loss/train': 0.7736318088136613}
+12/26/2021 11:03:36 - INFO - codeparrot_training - Step 26505: {'lr': 0.0002276854938783232, 'samples': 13571072, 'steps': 26505, 'batch_loss/train': 1.2800234400201589}
+12/26/2021 11:03:46 - INFO - codeparrot_training - Step 26506: {'lr': 0.0002276697937064256, 'samples': 13571584, 'steps': 26506, 'batch_loss/train': 0.7278192150406539}
+12/26/2021 11:03:58 - INFO - codeparrot_training - Step 26507: {'lr': 0.0002276540936233044, 'samples': 13572096, 'steps': 26507, 'batch_loss/train': 0.864836947992444}
+12/26/2021 11:04:09 - INFO - codeparrot_training - Step 26508: {'lr': 0.0002276383936290222, 'samples': 13572608, 'steps': 26508, 'batch_loss/train': 0.9033354455605149}
+12/26/2021 11:04:20 - INFO - codeparrot_training - Step 26509: {'lr': 0.00022762269372364114, 'samples': 13573120, 'steps': 26509, 'batch_loss/train': 0.6352962167002261}
+12/26/2021 11:04:33 - INFO - codeparrot_training - Step 26510: {'lr': 0.00022760699390722387, 'samples': 13573632, 'steps': 26510, 'batch_loss/train': 0.7612182899611071}
+12/26/2021 11:04:44 - INFO - codeparrot_training - Step 26511: {'lr': 0.00022759129417983278, 'samples': 13574144, 'steps': 26511, 'batch_loss/train': 0.745813270797953}
+12/26/2021 11:04:55 - INFO - codeparrot_training - Step 26512: {'lr': 0.00022757559454153015, 'samples': 13574656, 'steps': 26512, 'batch_loss/train': 0.7687159804627299}
+12/26/2021 11:05:05 - INFO - codeparrot_training - Step 26513: {'lr': 0.00022755989499237853, 'samples': 13575168, 'steps': 26513, 'batch_loss/train': 0.785527334548533}
+12/26/2021 11:05:17 - INFO - codeparrot_training - Step 26514: {'lr': 0.00022754419553244035, 'samples': 13575680, 'steps': 26514, 'batch_loss/train': 0.8516025329008698}
+12/26/2021 11:05:28 - INFO - codeparrot_training - Step 26515: {'lr': 0.00022752849616177798, 'samples': 13576192, 'steps': 26515, 'batch_loss/train': 0.7767160208895802}
+12/26/2021 11:05:39 - INFO - codeparrot_training - Step 26516: {'lr': 0.0002275127968804537, 'samples': 13576704, 'steps': 26516, 'batch_loss/train': 0.8032967804465443}
+12/26/2021 11:05:51 - INFO - codeparrot_training - Step 26517: {'lr': 0.00022749709768853012, 'samples': 13577216, 'steps': 26517, 'batch_loss/train': 0.70370087120682}
+12/26/2021 11:06:01 - INFO - codeparrot_training - Step 26518: {'lr': 0.00022748139858606964, 'samples': 13577728, 'steps': 26518, 'batch_loss/train': 0.7096171411685646}
+12/26/2021 11:06:12 - INFO - codeparrot_training - Step 26519: {'lr': 0.00022746569957313455, 'samples': 13578240, 'steps': 26519, 'batch_loss/train': 0.7398445997387171}
+12/26/2021 11:06:24 - INFO - codeparrot_training - Step 26520: {'lr': 0.00022745000064978727, 'samples': 13578752, 'steps': 26520, 'batch_loss/train': 0.7186222551390529}
+12/26/2021 11:06:35 - INFO - codeparrot_training - Step 26521: {'lr': 0.0002274343018160904, 'samples': 13579264, 'steps': 26521, 'batch_loss/train': 0.7609301079064608}
+12/26/2021 11:06:45 - INFO - codeparrot_training - Step 26522: {'lr': 0.00022741860307210615, 'samples': 13579776, 'steps': 26522, 'batch_loss/train': 0.7549688220024109}
+12/26/2021 11:06:59 - INFO - codeparrot_training - Step 26523: {'lr': 0.00022740290441789708, 'samples': 13580288, 'steps': 26523, 'batch_loss/train': 0.7560655237175524}
+12/26/2021 11:07:10 - INFO - codeparrot_training - Step 26524: {'lr': 0.0002273872058535254, 'samples': 13580800, 'steps': 26524, 'batch_loss/train': 0.7806376731023192}
+12/26/2021 11:07:21 - INFO - codeparrot_training - Step 26525: {'lr': 0.00022737150737905379, 'samples': 13581312, 'steps': 26525, 'batch_loss/train': 0.8453270504251122}
+12/26/2021 11:07:31 - INFO - codeparrot_training - Step 26526: {'lr': 0.00022735580899454444, 'samples': 13581824, 'steps': 26526, 'batch_loss/train': 0.7303385250270367}
+12/26/2021 11:07:44 - INFO - codeparrot_training - Step 26527: {'lr': 0.0002273401107000598, 'samples': 13582336, 'steps': 26527, 'batch_loss/train': 0.7048224434256554}
+12/26/2021 11:07:54 - INFO - codeparrot_training - Step 26528: {'lr': 0.00022732441249566243, 'samples': 13582848, 'steps': 26528, 'batch_loss/train': 0.7646640362218022}
+12/26/2021 11:08:05 - INFO - codeparrot_training - Step 26529: {'lr': 0.0002273087143814146, 'samples': 13583360, 'steps': 26529, 'batch_loss/train': 0.7545663490891457}
+12/26/2021 11:08:17 - INFO - codeparrot_training - Step 26530: {'lr': 0.0002272930163573788, 'samples': 13583872, 'steps': 26530, 'batch_loss/train': 0.6981466608121991}
+12/26/2021 11:08:28 - INFO - codeparrot_training - Step 26531: {'lr': 0.00022727731842361727, 'samples': 13584384, 'steps': 26531, 'batch_loss/train': 0.7109900552313775}
+12/26/2021 11:08:38 - INFO - codeparrot_training - Step 26532: {'lr': 0.0002272616205801926, 'samples': 13584896, 'steps': 26532, 'batch_loss/train': 0.7645823499187827}
+12/26/2021 11:08:49 - INFO - codeparrot_training - Step 26533: {'lr': 0.0002272459228271672, 'samples': 13585408, 'steps': 26533, 'batch_loss/train': 0.8640383742749691}
+12/26/2021 11:09:03 - INFO - codeparrot_training - Step 26534: {'lr': 0.00022723022516460344, 'samples': 13585920, 'steps': 26534, 'batch_loss/train': 0.623320407117717}
+12/26/2021 11:09:13 - INFO - codeparrot_training - Step 26535: {'lr': 0.00022721452759256355, 'samples': 13586432, 'steps': 26535, 'batch_loss/train': 0.7744660452008247}
+12/26/2021 11:09:24 - INFO - codeparrot_training - Step 26536: {'lr': 0.00022719883011111018, 'samples': 13586944, 'steps': 26536, 'batch_loss/train': 0.730665713082999}
+12/26/2021 11:09:36 - INFO - codeparrot_training - Step 26537: {'lr': 0.0002271831327203057, 'samples': 13587456, 'steps': 26537, 'batch_loss/train': 0.776857624296099}
+12/26/2021 11:09:47 - INFO - codeparrot_training - Step 26538: {'lr': 0.00022716743542021238, 'samples': 13587968, 'steps': 26538, 'batch_loss/train': 0.864201239310205}
+12/26/2021 11:09:57 - INFO - codeparrot_training - Step 26539: {'lr': 0.00022715173821089273, 'samples': 13588480, 'steps': 26539, 'batch_loss/train': 0.6712399171665311}
+12/26/2021 11:10:11 - INFO - codeparrot_training - Step 26540: {'lr': 0.0002271360410924092, 'samples': 13588992, 'steps': 26540, 'batch_loss/train': 0.8223260752856731}
+12/26/2021 11:10:22 - INFO - codeparrot_training - Step 26541: {'lr': 0.00022712034406482414, 'samples': 13589504, 'steps': 26541, 'batch_loss/train': 0.6726165945874527}
+12/26/2021 11:10:33 - INFO - codeparrot_training - Step 26542: {'lr': 0.00022710464712819985, 'samples': 13590016, 'steps': 26542, 'batch_loss/train': 0.7840409353375435}
+12/26/2021 11:10:45 - INFO - codeparrot_training - Step 26543: {'lr': 0.00022708895028259894, 'samples': 13590528, 'steps': 26543, 'batch_loss/train': 0.8387900311499834}
+12/26/2021 11:10:55 - INFO - codeparrot_training - Step 26544: {'lr': 0.00022707325352808373, 'samples': 13591040, 'steps': 26544, 'batch_loss/train': 0.7366657396778464}
+12/26/2021 11:11:06 - INFO - codeparrot_training - Step 26545: {'lr': 0.00022705755686471655, 'samples': 13591552, 'steps': 26545, 'batch_loss/train': 0.7931143064051867}
+12/26/2021 11:11:17 - INFO - codeparrot_training - Step 26546: {'lr': 0.0002270418602925598, 'samples': 13592064, 'steps': 26546, 'batch_loss/train': 0.7599034733138978}
+12/26/2021 11:11:29 - INFO - codeparrot_training - Step 26547: {'lr': 0.00022702616381167607, 'samples': 13592576, 'steps': 26547, 'batch_loss/train': 0.6938266381621361}
+12/26/2021 11:11:40 - INFO - codeparrot_training - Step 26548: {'lr': 0.00022701046742212757, 'samples': 13593088, 'steps': 26548, 'batch_loss/train': 0.6176656854804605}
+12/26/2021 11:11:50 - INFO - codeparrot_training - Step 26549: {'lr': 0.00022699477112397674, 'samples': 13593600, 'steps': 26549, 'batch_loss/train': 0.7595840459689498}
+12/26/2021 11:12:02 - INFO - codeparrot_training - Step 26550: {'lr': 0.0002269790749172861, 'samples': 13594112, 'steps': 26550, 'batch_loss/train': 0.6899700262583792}
+12/26/2021 11:12:13 - INFO - codeparrot_training - Step 26551: {'lr': 0.00022696337880211798, 'samples': 13594624, 'steps': 26551, 'batch_loss/train': 0.6756789352511987}
+12/26/2021 11:12:24 - INFO - codeparrot_training - Step 26552: {'lr': 0.0002269476827785347, 'samples': 13595136, 'steps': 26552, 'batch_loss/train': 0.7096376288682222}
+12/26/2021 11:12:38 - INFO - codeparrot_training - Step 26553: {'lr': 0.00022693198684659865, 'samples': 13595648, 'steps': 26553, 'batch_loss/train': 0.695090712280944}
+12/26/2021 11:12:48 - INFO - codeparrot_training - Step 26554: {'lr': 0.00022691629100637245, 'samples': 13596160, 'steps': 26554, 'batch_loss/train': 0.9113818516489118}
+12/26/2021 11:12:59 - INFO - codeparrot_training - Step 26555: {'lr': 0.0002269005952579183, 'samples': 13596672, 'steps': 26555, 'batch_loss/train': 0.8460070549044758}
+12/26/2021 11:13:09 - INFO - codeparrot_training - Step 26556: {'lr': 0.00022688489960129873, 'samples': 13597184, 'steps': 26556, 'batch_loss/train': 0.7651677643880248}
+12/26/2021 11:13:21 - INFO - codeparrot_training - Step 26557: {'lr': 0.00022686920403657595, 'samples': 13597696, 'steps': 26557, 'batch_loss/train': 0.6803085539722815}
+12/26/2021 11:13:32 - INFO - codeparrot_training - Step 26558: {'lr': 0.00022685350856381254, 'samples': 13598208, 'steps': 26558, 'batch_loss/train': 0.7868963205255568}
+12/26/2021 11:13:43 - INFO - codeparrot_training - Step 26559: {'lr': 0.0002268378131830709, 'samples': 13598720, 'steps': 26559, 'batch_loss/train': 0.7616252226289362}
+12/26/2021 11:13:55 - INFO - codeparrot_training - Step 26560: {'lr': 0.00022682211789441337, 'samples': 13599232, 'steps': 26560, 'batch_loss/train': 0.6687498043756932}
+12/26/2021 11:14:05 - INFO - codeparrot_training - Step 26561: {'lr': 0.00022680642269790223, 'samples': 13599744, 'steps': 26561, 'batch_loss/train': 0.8808200443163514}
+12/26/2021 11:14:16 - INFO - codeparrot_training - Step 26562: {'lr': 0.00022679072759360002, 'samples': 13600256, 'steps': 26562, 'batch_loss/train': 0.7626407607458532}
+12/26/2021 11:14:30 - INFO - codeparrot_training - Step 26563: {'lr': 0.00022677503258156922, 'samples': 13600768, 'steps': 26563, 'batch_loss/train': 0.7620911984704435}
+12/26/2021 11:14:40 - INFO - codeparrot_training - Step 26564: {'lr': 0.00022675933766187198, 'samples': 13601280, 'steps': 26564, 'batch_loss/train': 0.7069011814892292}
+12/26/2021 11:14:51 - INFO - codeparrot_training - Step 26565: {'lr': 0.0002267436428345709, 'samples': 13601792, 'steps': 26565, 'batch_loss/train': 0.6958777036052197}
+12/26/2021 11:15:02 - INFO - codeparrot_training - Step 26566: {'lr': 0.0002267279480997284, 'samples': 13602304, 'steps': 26566, 'batch_loss/train': 0.7242936184629798}
+12/26/2021 11:15:14 - INFO - codeparrot_training - Step 26567: {'lr': 0.0002267122534574067, 'samples': 13602816, 'steps': 26567, 'batch_loss/train': 0.8843832355923951}
+12/26/2021 11:15:24 - INFO - codeparrot_training - Step 26568: {'lr': 0.00022669655890766825, 'samples': 13603328, 'steps': 26568, 'batch_loss/train': 0.6611906029284}
+12/26/2021 11:15:35 - INFO - codeparrot_training - Step 26569: {'lr': 0.00022668086445057557, 'samples': 13603840, 'steps': 26569, 'batch_loss/train': 0.7161659366684034}
+12/26/2021 11:15:49 - INFO - codeparrot_training - Step 26570: {'lr': 0.00022666517008619098, 'samples': 13604352, 'steps': 26570, 'batch_loss/train': 0.7489704107865691}
+12/26/2021 11:16:00 - INFO - codeparrot_training - Step 26571: {'lr': 0.0002266494758145768, 'samples': 13604864, 'steps': 26571, 'batch_loss/train': 1.5984522392973304}
+12/26/2021 11:16:10 - INFO - codeparrot_training - Step 26572: {'lr': 0.00022663378163579546, 'samples': 13605376, 'steps': 26572, 'batch_loss/train': 0.8257952490821481}
+12/26/2021 11:16:22 - INFO - codeparrot_training - Step 26573: {'lr': 0.00022661808754990945, 'samples': 13605888, 'steps': 26573, 'batch_loss/train': 0.9459136473014951}
+12/26/2021 11:16:33 - INFO - codeparrot_training - Step 26574: {'lr': 0.0002266023935569811, 'samples': 13606400, 'steps': 26574, 'batch_loss/train': 0.85934903845191}
+12/26/2021 11:16:43 - INFO - codeparrot_training - Step 26575: {'lr': 0.00022658669965707268, 'samples': 13606912, 'steps': 26575, 'batch_loss/train': 0.7798464661464095}
+12/26/2021 11:16:55 - INFO - codeparrot_training - Step 26576: {'lr': 0.00022657100585024684, 'samples': 13607424, 'steps': 26576, 'batch_loss/train': 0.7788379229605198}
+12/26/2021 11:17:06 - INFO - codeparrot_training - Step 26577: {'lr': 0.00022655531213656578, 'samples': 13607936, 'steps': 26577, 'batch_loss/train': 0.841097554191947}
+12/26/2021 11:17:17 - INFO - codeparrot_training - Step 26578: {'lr': 0.00022653961851609202, 'samples': 13608448, 'steps': 26578, 'batch_loss/train': 0.7865645289421082}
+12/26/2021 11:17:27 - INFO - codeparrot_training - Step 26579: {'lr': 0.00022652392498888772, 'samples': 13608960, 'steps': 26579, 'batch_loss/train': 0.7991254054941237}
+12/26/2021 11:17:41 - INFO - codeparrot_training - Step 26580: {'lr': 0.00022650823155501558, 'samples': 13609472, 'steps': 26580, 'batch_loss/train': 0.7274698519613594}
+12/26/2021 11:17:52 - INFO - codeparrot_training - Step 26581: {'lr': 0.00022649253821453778, 'samples': 13609984, 'steps': 26581, 'batch_loss/train': 0.8712316863238811}
+12/26/2021 11:18:03 - INFO - codeparrot_training - Step 26582: {'lr': 0.0002264768449675168, 'samples': 13610496, 'steps': 26582, 'batch_loss/train': 0.7444622442126274}
+12/26/2021 11:18:15 - INFO - codeparrot_training - Step 26583: {'lr': 0.0002264611518140149, 'samples': 13611008, 'steps': 26583, 'batch_loss/train': 0.8177956002764404}
+12/26/2021 11:18:25 - INFO - codeparrot_training - Step 26584: {'lr': 0.00022644545875409467, 'samples': 13611520, 'steps': 26584, 'batch_loss/train': 0.807553025893867}
+12/26/2021 11:18:36 - INFO - codeparrot_training - Step 26585: {'lr': 0.00022642976578781844, 'samples': 13612032, 'steps': 26585, 'batch_loss/train': 0.6557619000668637}
+12/26/2021 11:18:48 - INFO - codeparrot_training - Step 26586: {'lr': 0.00022641407291524844, 'samples': 13612544, 'steps': 26586, 'batch_loss/train': 0.6432518749497831}
+12/26/2021 11:18:59 - INFO - codeparrot_training - Step 26587: {'lr': 0.0002263983801364472, 'samples': 13613056, 'steps': 26587, 'batch_loss/train': 0.7612902484834194}
+12/26/2021 11:19:09 - INFO - codeparrot_training - Step 26588: {'lr': 0.0002263826874514772, 'samples': 13613568, 'steps': 26588, 'batch_loss/train': 0.6587547059170902}
+12/26/2021 11:19:20 - INFO - codeparrot_training - Step 26589: {'lr': 0.0002263669948604007, 'samples': 13614080, 'steps': 26589, 'batch_loss/train': 0.7823539674282074}
+12/26/2021 11:19:34 - INFO - codeparrot_training - Step 26590: {'lr': 0.00022635130236327996, 'samples': 13614592, 'steps': 26590, 'batch_loss/train': 0.7286267848103307}
+12/26/2021 11:19:45 - INFO - codeparrot_training - Step 26591: {'lr': 0.00022633560996017764, 'samples': 13615104, 'steps': 26591, 'batch_loss/train': 0.7290628448245116}
+12/26/2021 11:19:56 - INFO - codeparrot_training - Step 26592: {'lr': 0.00022631991765115602, 'samples': 13615616, 'steps': 26592, 'batch_loss/train': 0.6645289361476898}
+12/26/2021 11:20:08 - INFO - codeparrot_training - Step 26593: {'lr': 0.00022630422543627737, 'samples': 13616128, 'steps': 26593, 'batch_loss/train': 0.7166994784492999}
+12/26/2021 11:20:18 - INFO - codeparrot_training - Step 26594: {'lr': 0.00022628853331560416, 'samples': 13616640, 'steps': 26594, 'batch_loss/train': 0.7923867651261389}
+12/26/2021 11:20:29 - INFO - codeparrot_training - Step 26595: {'lr': 0.0002262728412891989, 'samples': 13617152, 'steps': 26595, 'batch_loss/train': 0.8837126903235912}
+12/26/2021 11:20:41 - INFO - codeparrot_training - Step 26596: {'lr': 0.00022625714935712386, 'samples': 13617664, 'steps': 26596, 'batch_loss/train': 0.6655160079244524}
+12/26/2021 11:20:52 - INFO - codeparrot_training - Step 26597: {'lr': 0.00022624145751944135, 'samples': 13618176, 'steps': 26597, 'batch_loss/train': 0.7083052834495902}
+12/26/2021 11:21:02 - INFO - codeparrot_training - Step 26598: {'lr': 0.0002262257657762138, 'samples': 13618688, 'steps': 26598, 'batch_loss/train': 0.8130045523867011}
+12/26/2021 11:21:13 - INFO - codeparrot_training - Step 26599: {'lr': 0.00022621007412750374, 'samples': 13619200, 'steps': 26599, 'batch_loss/train': 0.718199753202498}
+12/26/2021 11:21:27 - INFO - codeparrot_training - Step 26600: {'lr': 0.00022619438257337338, 'samples': 13619712, 'steps': 26600, 'batch_loss/train': 0.8561946467962116}
+12/26/2021 11:21:38 - INFO - codeparrot_training - Step 26601: {'lr': 0.00022617869111388512, 'samples': 13620224, 'steps': 26601, 'batch_loss/train': 0.6388814570382237}
+12/26/2021 11:21:48 - INFO - codeparrot_training - Step 26602: {'lr': 0.00022616299974910148, 'samples': 13620736, 'steps': 26602, 'batch_loss/train': 0.8357412731274962}
+12/26/2021 11:22:00 - INFO - codeparrot_training - Step 26603: {'lr': 0.00022614730847908473, 'samples': 13621248, 'steps': 26603, 'batch_loss/train': 0.746918854303658}
+12/26/2021 11:22:11 - INFO - codeparrot_training - Step 26604: {'lr': 0.0002261316173038973, 'samples': 13621760, 'steps': 26604, 'batch_loss/train': 0.7787495879456401}
+12/26/2021 11:22:21 - INFO - codeparrot_training - Step 26605: {'lr': 0.00022611592622360144, 'samples': 13622272, 'steps': 26605, 'batch_loss/train': 0.7462692093104124}
+12/26/2021 11:22:33 - INFO - codeparrot_training - Step 26606: {'lr': 0.00022610023523825977, 'samples': 13622784, 'steps': 26606, 'batch_loss/train': 0.84342792769894}
+12/26/2021 11:22:44 - INFO - codeparrot_training - Step 26607: {'lr': 0.00022608454434793447, 'samples': 13623296, 'steps': 26607, 'batch_loss/train': 0.7468075631186366}
+12/26/2021 11:22:55 - INFO - codeparrot_training - Step 26608: {'lr': 0.00022606885355268806, 'samples': 13623808, 'steps': 26608, 'batch_loss/train': 0.6925795243587345}
+12/26/2021 11:23:05 - INFO - codeparrot_training - Step 26609: {'lr': 0.0002260531628525827, 'samples': 13624320, 'steps': 26609, 'batch_loss/train': 0.7717750249430537}
+12/26/2021 11:23:19 - INFO - codeparrot_training - Step 26610: {'lr': 0.00022603747224768104, 'samples': 13624832, 'steps': 26610, 'batch_loss/train': 0.7392938584089279}
+12/26/2021 11:23:30 - INFO - codeparrot_training - Step 26611: {'lr': 0.00022602178173804536, 'samples': 13625344, 'steps': 26611, 'batch_loss/train': 0.8145723892375827}
+12/26/2021 11:23:41 - INFO - codeparrot_training - Step 26612: {'lr': 0.0002260060913237379, 'samples': 13625856, 'steps': 26612, 'batch_loss/train': 0.7459759095218033}
+12/26/2021 11:23:53 - INFO - codeparrot_training - Step 26613: {'lr': 0.00022599040100482128, 'samples': 13626368, 'steps': 26613, 'batch_loss/train': 0.7151271470356733}
+12/26/2021 11:24:04 - INFO - codeparrot_training - Step 26614: {'lr': 0.00022597471078135772, 'samples': 13626880, 'steps': 26614, 'batch_loss/train': 0.4973503132350743}
+12/26/2021 11:24:14 - INFO - codeparrot_training - Step 26615: {'lr': 0.0002259590206534097, 'samples': 13627392, 'steps': 26615, 'batch_loss/train': 0.6818685787147842}
+12/26/2021 11:24:26 - INFO - codeparrot_training - Step 26616: {'lr': 0.0002259433306210394, 'samples': 13627904, 'steps': 26616, 'batch_loss/train': 0.6848952546715736}
+12/26/2021 11:24:37 - INFO - codeparrot_training - Step 26617: {'lr': 0.00022592764068430943, 'samples': 13628416, 'steps': 26617, 'batch_loss/train': 0.8665241161361337}
+12/26/2021 11:24:47 - INFO - codeparrot_training - Step 26618: {'lr': 0.0002259119508432821, 'samples': 13628928, 'steps': 26618, 'batch_loss/train': 0.7802820652723312}
+12/26/2021 11:24:58 - INFO - codeparrot_training - Step 26619: {'lr': 0.0002258962610980197, 'samples': 13629440, 'steps': 26619, 'batch_loss/train': 0.8334549386054277}
+12/26/2021 11:25:12 - INFO - codeparrot_training - Step 26620: {'lr': 0.00022588057144858463, 'samples': 13629952, 'steps': 26620, 'batch_loss/train': 0.6809018300846219}
+12/26/2021 11:25:23 - INFO - codeparrot_training - Step 26621: {'lr': 0.00022586488189503938, 'samples': 13630464, 'steps': 26621, 'batch_loss/train': 0.7751767290756106}
+12/26/2021 11:25:33 - INFO - codeparrot_training - Step 26622: {'lr': 0.0002258491924374462, 'samples': 13630976, 'steps': 26622, 'batch_loss/train': 0.8207745882682502}
+12/26/2021 11:25:45 - INFO - codeparrot_training - Step 26623: {'lr': 0.00022583350307586747, 'samples': 13631488, 'steps': 26623, 'batch_loss/train': 0.7057975311763585}
+12/26/2021 11:25:56 - INFO - codeparrot_training - Step 26624: {'lr': 0.0002258178138103657, 'samples': 13632000, 'steps': 26624, 'batch_loss/train': 0.8451768991071731}
+12/26/2021 11:26:07 - INFO - codeparrot_training - Step 26625: {'lr': 0.00022580212464100318, 'samples': 13632512, 'steps': 26625, 'batch_loss/train': 0.8224200764670968}
+12/26/2021 11:26:19 - INFO - codeparrot_training - Step 26626: {'lr': 0.00022578643556784222, 'samples': 13633024, 'steps': 26626, 'batch_loss/train': 0.638866692665033}
+12/26/2021 11:26:29 - INFO - codeparrot_training - Step 26627: {'lr': 0.0002257707465909452, 'samples': 13633536, 'steps': 26627, 'batch_loss/train': 0.8850031257607043}
+12/26/2021 11:26:40 - INFO - codeparrot_training - Step 26628: {'lr': 0.00022575505771037466, 'samples': 13634048, 'steps': 26628, 'batch_loss/train': 0.8070859452709556}
+12/26/2021 11:26:51 - INFO - codeparrot_training - Step 26629: {'lr': 0.0002257393689261928, 'samples': 13634560, 'steps': 26629, 'batch_loss/train': 0.8175071962177753}
+12/26/2021 11:27:03 - INFO - codeparrot_training - Step 26630: {'lr': 0.00022572368023846207, 'samples': 13635072, 'steps': 26630, 'batch_loss/train': 1.5824670451693237}
+12/26/2021 11:27:13 - INFO - codeparrot_training - Step 26631: {'lr': 0.00022570799164724472, 'samples': 13635584, 'steps': 26631, 'batch_loss/train': 0.595354356628377}
+12/26/2021 11:27:24 - INFO - codeparrot_training - Step 26632: {'lr': 0.0002256923031526033, 'samples': 13636096, 'steps': 26632, 'batch_loss/train': 0.8010539840906858}
+12/26/2021 11:27:38 - INFO - codeparrot_training - Step 26633: {'lr': 0.00022567661475460016, 'samples': 13636608, 'steps': 26633, 'batch_loss/train': 0.7642057281918824}
+12/26/2021 11:27:49 - INFO - codeparrot_training - Step 26634: {'lr': 0.0002256609264532976, 'samples': 13637120, 'steps': 26634, 'batch_loss/train': 0.7622585487551987}
+12/26/2021 11:27:59 - INFO - codeparrot_training - Step 26635: {'lr': 0.0002256452382487579, 'samples': 13637632, 'steps': 26635, 'batch_loss/train': 0.7689282284118235}
+12/26/2021 11:28:11 - INFO - codeparrot_training - Step 26636: {'lr': 0.00022562955014104362, 'samples': 13638144, 'steps': 26636, 'batch_loss/train': 0.8065306320786476}
+12/26/2021 11:28:22 - INFO - codeparrot_training - Step 26637: {'lr': 0.00022561386213021707, 'samples': 13638656, 'steps': 26637, 'batch_loss/train': 0.829422214999795}
+12/26/2021 11:28:32 - INFO - codeparrot_training - Step 26638: {'lr': 0.0002255981742163405, 'samples': 13639168, 'steps': 26638, 'batch_loss/train': 0.8035866566933692}
+12/26/2021 11:28:43 - INFO - codeparrot_training - Step 26639: {'lr': 0.0002255824863994764, 'samples': 13639680, 'steps': 26639, 'batch_loss/train': 0.7132150302059017}
+12/26/2021 11:28:58 - INFO - codeparrot_training - Step 26640: {'lr': 0.0002255667986796872, 'samples': 13640192, 'steps': 26640, 'batch_loss/train': 0.7334228330291808}
+12/26/2021 11:29:08 - INFO - codeparrot_training - Step 26641: {'lr': 0.00022555111105703517, 'samples': 13640704, 'steps': 26641, 'batch_loss/train': 0.6450730999931693}
+12/26/2021 11:29:19 - INFO - codeparrot_training - Step 26642: {'lr': 0.0002255354235315826, 'samples': 13641216, 'steps': 26642, 'batch_loss/train': 0.7570244250819087}
+12/26/2021 11:29:31 - INFO - codeparrot_training - Step 26643: {'lr': 0.00022551973610339198, 'samples': 13641728, 'steps': 26643, 'batch_loss/train': 0.6819639666937292}
+12/26/2021 11:29:42 - INFO - codeparrot_training - Step 26644: {'lr': 0.00022550404877252572, 'samples': 13642240, 'steps': 26644, 'batch_loss/train': 0.7604140918701887}
+12/26/2021 11:29:52 - INFO - codeparrot_training - Step 26645: {'lr': 0.00022548836153904605, 'samples': 13642752, 'steps': 26645, 'batch_loss/train': 0.7297080112621188}
+12/26/2021 11:30:04 - INFO - codeparrot_training - Step 26646: {'lr': 0.00022547267440301534, 'samples': 13643264, 'steps': 26646, 'batch_loss/train': 0.7699466068297625}
+12/26/2021 11:30:15 - INFO - codeparrot_training - Step 26647: {'lr': 0.00022545698736449612, 'samples': 13643776, 'steps': 26647, 'batch_loss/train': 0.7487853011116385}
+12/26/2021 11:30:26 - INFO - codeparrot_training - Step 26648: {'lr': 0.00022544130042355062, 'samples': 13644288, 'steps': 26648, 'batch_loss/train': 0.8196489382535219}
+12/26/2021 11:30:36 - INFO - codeparrot_training - Step 26649: {'lr': 0.00022542561358024116, 'samples': 13644800, 'steps': 26649, 'batch_loss/train': 0.7440110882744193}
+12/26/2021 11:30:50 - INFO - codeparrot_training - Step 26650: {'lr': 0.0002254099268346303, 'samples': 13645312, 'steps': 26650, 'batch_loss/train': 0.6838004253804684}
+12/26/2021 11:31:01 - INFO - codeparrot_training - Step 26651: {'lr': 0.00022539424018678029, 'samples': 13645824, 'steps': 26651, 'batch_loss/train': 0.9201954072341323}
+12/26/2021 11:31:12 - INFO - codeparrot_training - Step 26652: {'lr': 0.00022537855363675342, 'samples': 13646336, 'steps': 26652, 'batch_loss/train': 0.8465639613568783}
+12/26/2021 11:31:24 - INFO - codeparrot_training - Step 26653: {'lr': 0.00022536286718461207, 'samples': 13646848, 'steps': 26653, 'batch_loss/train': 0.7024266084190458}
+12/26/2021 11:31:34 - INFO - codeparrot_training - Step 26654: {'lr': 0.00022534718083041876, 'samples': 13647360, 'steps': 26654, 'batch_loss/train': 0.7138100657612085}
+12/26/2021 11:31:45 - INFO - codeparrot_training - Step 26655: {'lr': 0.00022533149457423573, 'samples': 13647872, 'steps': 26655, 'batch_loss/train': 0.7176647423766553}
+12/26/2021 11:31:59 - INFO - codeparrot_training - Step 26656: {'lr': 0.00022531580841612543, 'samples': 13648384, 'steps': 26656, 'batch_loss/train': 0.7364725302322768}
+12/26/2021 11:32:09 - INFO - codeparrot_training - Step 26657: {'lr': 0.00022530012235615, 'samples': 13648896, 'steps': 26657, 'batch_loss/train': 0.7039200107101351}
+12/26/2021 11:32:20 - INFO - codeparrot_training - Step 26658: {'lr': 0.00022528443639437202, 'samples': 13649408, 'steps': 26658, 'batch_loss/train': 0.7581022684462368}
+12/26/2021 11:32:32 - INFO - codeparrot_training - Step 26659: {'lr': 0.00022526875053085387, 'samples': 13649920, 'steps': 26659, 'batch_loss/train': 0.7669555004686117}
+12/26/2021 11:32:43 - INFO - codeparrot_training - Step 26660: {'lr': 0.00022525306476565767, 'samples': 13650432, 'steps': 26660, 'batch_loss/train': 0.9768393528647721}
+12/26/2021 11:32:53 - INFO - codeparrot_training - Step 26661: {'lr': 0.00022523737909884608, 'samples': 13650944, 'steps': 26661, 'batch_loss/train': 0.7221408439800143}
+12/26/2021 11:33:04 - INFO - codeparrot_training - Step 26662: {'lr': 0.00022522169353048128, 'samples': 13651456, 'steps': 26662, 'batch_loss/train': 0.6863712761551142}
+12/26/2021 11:33:16 - INFO - codeparrot_training - Step 26663: {'lr': 0.00022520600806062572, 'samples': 13651968, 'steps': 26663, 'batch_loss/train': 0.7765477066859603}
+12/26/2021 11:33:27 - INFO - codeparrot_training - Step 26664: {'lr': 0.0002251903226893416, 'samples': 13652480, 'steps': 26664, 'batch_loss/train': 0.7658288162201643}
+12/26/2021 11:33:37 - INFO - codeparrot_training - Step 26665: {'lr': 0.00022517463741669142, 'samples': 13652992, 'steps': 26665, 'batch_loss/train': 0.681660661008209}
+12/26/2021 11:33:50 - INFO - codeparrot_training - Step 26666: {'lr': 0.00022515895224273758, 'samples': 13653504, 'steps': 26666, 'batch_loss/train': 0.8009716200176626}
+12/26/2021 11:34:00 - INFO - codeparrot_training - Step 26667: {'lr': 0.0002251432671675423, 'samples': 13654016, 'steps': 26667, 'batch_loss/train': 0.7883411543443799}
+12/26/2021 11:34:11 - INFO - codeparrot_training - Step 26668: {'lr': 0.00022512758219116796, 'samples': 13654528, 'steps': 26668, 'batch_loss/train': 0.7048510778695345}
+12/26/2021 11:34:25 - INFO - codeparrot_training - Step 26669: {'lr': 0.00022511189731367706, 'samples': 13655040, 'steps': 26669, 'batch_loss/train': 0.804938163375482}
+12/26/2021 11:34:36 - INFO - codeparrot_training - Step 26670: {'lr': 0.00022509621253513187, 'samples': 13655552, 'steps': 26670, 'batch_loss/train': 0.6188855724176392}
+12/26/2021 11:34:47 - INFO - codeparrot_training - Step 26671: {'lr': 0.0002250805278555947, 'samples': 13656064, 'steps': 26671, 'batch_loss/train': 0.8471671580336988}
+12/26/2021 11:34:57 - INFO - codeparrot_training - Step 26672: {'lr': 0.00022506484327512787, 'samples': 13656576, 'steps': 26672, 'batch_loss/train': 0.8000881494954228}
+12/26/2021 11:35:10 - INFO - codeparrot_training - Step 26673: {'lr': 0.00022504915879379393, 'samples': 13657088, 'steps': 26673, 'batch_loss/train': 0.7564933103858493}
+12/26/2021 11:35:20 - INFO - codeparrot_training - Step 26674: {'lr': 0.00022503347441165503, 'samples': 13657600, 'steps': 26674, 'batch_loss/train': 0.6293267777946312}
+12/26/2021 11:35:31 - INFO - codeparrot_training - Step 26675: {'lr': 0.00022501779012877357, 'samples': 13658112, 'steps': 26675, 'batch_loss/train': 0.7573028784245253}
+12/26/2021 11:35:43 - INFO - codeparrot_training - Step 26676: {'lr': 0.00022500210594521205, 'samples': 13658624, 'steps': 26676, 'batch_loss/train': 0.779009803198278}
+12/26/2021 11:35:53 - INFO - codeparrot_training - Step 26677: {'lr': 0.00022498642186103267, 'samples': 13659136, 'steps': 26677, 'batch_loss/train': 0.8309233764884993}
+12/26/2021 11:36:04 - INFO - codeparrot_training - Step 26678: {'lr': 0.0002249707378762979, 'samples': 13659648, 'steps': 26678, 'batch_loss/train': 0.7156615585554391}
+12/26/2021 11:36:18 - INFO - codeparrot_training - Step 26679: {'lr': 0.00022495505399106985, 'samples': 13660160, 'steps': 26679, 'batch_loss/train': 0.8159506050869823}
+12/26/2021 11:36:29 - INFO - codeparrot_training - Step 26680: {'lr': 0.00022493937020541123, 'samples': 13660672, 'steps': 26680, 'batch_loss/train': 1.4519646824337542}
+12/26/2021 11:36:39 - INFO - codeparrot_training - Step 26681: {'lr': 0.00022492368651938415, 'samples': 13661184, 'steps': 26681, 'batch_loss/train': 0.5447830036282539}
+12/26/2021 11:36:50 - INFO - codeparrot_training - Step 26682: {'lr': 0.00022490800293305105, 'samples': 13661696, 'steps': 26682, 'batch_loss/train': 0.7201319967862219}
+12/26/2021 11:37:02 - INFO - codeparrot_training - Step 26683: {'lr': 0.00022489231944647416, 'samples': 13662208, 'steps': 26683, 'batch_loss/train': 0.7191772379446775}
+12/26/2021 11:37:13 - INFO - codeparrot_training - Step 26684: {'lr': 0.00022487663605971598, 'samples': 13662720, 'steps': 26684, 'batch_loss/train': 0.6983474022708833}
+12/26/2021 11:37:23 - INFO - codeparrot_training - Step 26685: {'lr': 0.00022486095277283885, 'samples': 13663232, 'steps': 26685, 'batch_loss/train': 0.7209075298160315}
+12/26/2021 11:37:36 - INFO - codeparrot_training - Step 26686: {'lr': 0.00022484526958590497, 'samples': 13663744, 'steps': 26686, 'batch_loss/train': 0.7649381848168559}
+12/26/2021 11:37:46 - INFO - codeparrot_training - Step 26687: {'lr': 0.00022482958649897695, 'samples': 13664256, 'steps': 26687, 'batch_loss/train': 0.7653354583308101}
+12/26/2021 11:37:57 - INFO - codeparrot_training - Step 26688: {'lr': 0.00022481390351211688, 'samples': 13664768, 'steps': 26688, 'batch_loss/train': 0.7406483432278037}
+12/26/2021 11:38:11 - INFO - codeparrot_training - Step 26689: {'lr': 0.00022479822062538733, 'samples': 13665280, 'steps': 26689, 'batch_loss/train': 0.5849062938941643}
+12/26/2021 11:38:22 - INFO - codeparrot_training - Step 26690: {'lr': 0.0002247825378388504, 'samples': 13665792, 'steps': 26690, 'batch_loss/train': 0.8757971944287419}
+12/26/2021 11:38:32 - INFO - codeparrot_training - Step 26691: {'lr': 0.00022476685515256863, 'samples': 13666304, 'steps': 26691, 'batch_loss/train': 0.6977055622264743}
+12/26/2021 11:38:44 - INFO - codeparrot_training - Step 26692: {'lr': 0.00022475117256660437, 'samples': 13666816, 'steps': 26692, 'batch_loss/train': 0.700093568302691}
+12/26/2021 11:38:55 - INFO - codeparrot_training - Step 26693: {'lr': 0.00022473549008101987, 'samples': 13667328, 'steps': 26693, 'batch_loss/train': 0.7777432007715106}
+12/26/2021 11:39:05 - INFO - codeparrot_training - Step 26694: {'lr': 0.00022471980769587745, 'samples': 13667840, 'steps': 26694, 'batch_loss/train': 0.7593953646719456}
+12/26/2021 11:39:16 - INFO - codeparrot_training - Step 26695: {'lr': 0.00022470412541123966, 'samples': 13668352, 'steps': 26695, 'batch_loss/train': 0.7959694918245077}
+12/26/2021 11:39:30 - INFO - codeparrot_training - Step 26696: {'lr': 0.0002246884432271687, 'samples': 13668864, 'steps': 26696, 'batch_loss/train': 0.9149260297417641}
+12/26/2021 11:39:41 - INFO - codeparrot_training - Step 26697: {'lr': 0.00022467276114372683, 'samples': 13669376, 'steps': 26697, 'batch_loss/train': 0.8135582394897938}
+12/26/2021 11:39:51 - INFO - codeparrot_training - Step 26698: {'lr': 0.0002246570791609766, 'samples': 13669888, 'steps': 26698, 'batch_loss/train': 0.7809630110859871}
+12/26/2021 11:40:03 - INFO - codeparrot_training - Step 26699: {'lr': 0.00022464139727898023, 'samples': 13670400, 'steps': 26699, 'batch_loss/train': 0.7451390996575356}
+12/26/2021 11:40:14 - INFO - codeparrot_training - Step 26700: {'lr': 0.00022462571549780006, 'samples': 13670912, 'steps': 26700, 'batch_loss/train': 0.7020670101046562}
+12/26/2021 11:40:25 - INFO - codeparrot_training - Step 26701: {'lr': 0.00022461003381749842, 'samples': 13671424, 'steps': 26701, 'batch_loss/train': 0.7398744001984596}
+12/26/2021 11:40:37 - INFO - codeparrot_training - Step 26702: {'lr': 0.00022459435223813783, 'samples': 13671936, 'steps': 26702, 'batch_loss/train': 0.6029589595273137}
+12/26/2021 11:40:47 - INFO - codeparrot_training - Step 26703: {'lr': 0.0002245786707597804, 'samples': 13672448, 'steps': 26703, 'batch_loss/train': 0.8532277485355735}
+12/26/2021 11:40:58 - INFO - codeparrot_training - Step 26704: {'lr': 0.00022456298938248865, 'samples': 13672960, 'steps': 26704, 'batch_loss/train': 0.8129167756997049}
+12/26/2021 11:41:09 - INFO - codeparrot_training - Step 26705: {'lr': 0.00022454730810632476, 'samples': 13673472, 'steps': 26705, 'batch_loss/train': 0.7387951135169715}
+12/26/2021 11:41:21 - INFO - codeparrot_training - Step 26706: {'lr': 0.00022453162693135124, 'samples': 13673984, 'steps': 26706, 'batch_loss/train': 0.7568303346633911}
+12/26/2021 11:41:31 - INFO - codeparrot_training - Step 26707: {'lr': 0.00022451594585763034, 'samples': 13674496, 'steps': 26707, 'batch_loss/train': 0.8917929581366479}
+12/26/2021 11:41:42 - INFO - codeparrot_training - Step 26708: {'lr': 0.00022450026488522444, 'samples': 13675008, 'steps': 26708, 'batch_loss/train': 0.769305182620883}
+12/26/2021 11:41:56 - INFO - codeparrot_training - Step 26709: {'lr': 0.00022448458401419576, 'samples': 13675520, 'steps': 26709, 'batch_loss/train': 0.9500228678807616}
+12/26/2021 11:42:06 - INFO - codeparrot_training - Step 26710: {'lr': 0.0002244689032446068, 'samples': 13676032, 'steps': 26710, 'batch_loss/train': 0.6765034454874694}
+12/26/2021 11:42:17 - INFO - codeparrot_training - Step 26711: {'lr': 0.00022445322257651988, 'samples': 13676544, 'steps': 26711, 'batch_loss/train': 0.8417146757710725}
+12/26/2021 11:42:29 - INFO - codeparrot_training - Step 26712: {'lr': 0.00022443754200999718, 'samples': 13677056, 'steps': 26712, 'batch_loss/train': 0.7223571226932108}
+12/26/2021 11:42:40 - INFO - codeparrot_training - Step 26713: {'lr': 0.00022442186154510125, 'samples': 13677568, 'steps': 26713, 'batch_loss/train': 0.8219403270632029}
+12/26/2021 11:42:50 - INFO - codeparrot_training - Step 26714: {'lr': 0.00022440618118189438, 'samples': 13678080, 'steps': 26714, 'batch_loss/train': 0.7166420973371714}
+12/26/2021 11:43:02 - INFO - codeparrot_training - Step 26715: {'lr': 0.00022439050092043887, 'samples': 13678592, 'steps': 26715, 'batch_loss/train': 0.7734335358254611}
+12/26/2021 11:43:13 - INFO - codeparrot_training - Step 26716: {'lr': 0.00022437482076079692, 'samples': 13679104, 'steps': 26716, 'batch_loss/train': 0.8076036614365876}
+12/26/2021 11:43:24 - INFO - codeparrot_training - Step 26717: {'lr': 0.0002243591407030311, 'samples': 13679616, 'steps': 26717, 'batch_loss/train': 0.7045322360936552}
+12/26/2021 11:43:34 - INFO - codeparrot_training - Step 26718: {'lr': 0.00022434346074720369, 'samples': 13680128, 'steps': 26718, 'batch_loss/train': 0.675122354645282}
+12/26/2021 11:43:48 - INFO - codeparrot_training - Step 26719: {'lr': 0.00022432778089337698, 'samples': 13680640, 'steps': 26719, 'batch_loss/train': 0.7730833645910025}
+12/26/2021 11:43:59 - INFO - codeparrot_training - Step 26720: {'lr': 0.0002243121011416132, 'samples': 13681152, 'steps': 26720, 'batch_loss/train': 0.7504207687452435}
+12/26/2021 11:44:09 - INFO - codeparrot_training - Step 26721: {'lr': 0.00022429642149197497, 'samples': 13681664, 'steps': 26721, 'batch_loss/train': 0.7857560955453664}
+12/26/2021 11:44:21 - INFO - codeparrot_training - Step 26722: {'lr': 0.00022428074194452436, 'samples': 13682176, 'steps': 26722, 'batch_loss/train': 0.7278433381579816}
+12/26/2021 11:44:32 - INFO - codeparrot_training - Step 26723: {'lr': 0.0002242650624993238, 'samples': 13682688, 'steps': 26723, 'batch_loss/train': 0.7392128771170974}
+12/26/2021 11:44:43 - INFO - codeparrot_training - Step 26724: {'lr': 0.00022424938315643574, 'samples': 13683200, 'steps': 26724, 'batch_loss/train': 0.7947757253423333}
+12/26/2021 11:44:56 - INFO - codeparrot_training - Step 26725: {'lr': 0.00022423370391592238, 'samples': 13683712, 'steps': 26725, 'batch_loss/train': 0.698815044015646}
+12/26/2021 11:45:07 - INFO - codeparrot_training - Step 26726: {'lr': 0.00022421802477784602, 'samples': 13684224, 'steps': 26726, 'batch_loss/train': 0.7945384818594903}
+12/26/2021 11:45:17 - INFO - codeparrot_training - Step 26727: {'lr': 0.00022420234574226902, 'samples': 13684736, 'steps': 26727, 'batch_loss/train': 0.672198268584907}
+12/26/2021 11:45:28 - INFO - codeparrot_training - Step 26728: {'lr': 0.0002241866668092539, 'samples': 13685248, 'steps': 26728, 'batch_loss/train': 0.7528452281258069}
+12/26/2021 11:45:40 - INFO - codeparrot_training - Step 26729: {'lr': 0.00022417098797886277, 'samples': 13685760, 'steps': 26729, 'batch_loss/train': 0.8097270308062434}
+12/26/2021 11:45:51 - INFO - codeparrot_training - Step 26730: {'lr': 0.00022415530925115807, 'samples': 13686272, 'steps': 26730, 'batch_loss/train': 0.6633328453171998}
+12/26/2021 11:46:01 - INFO - codeparrot_training - Step 26731: {'lr': 0.000224139630626202, 'samples': 13686784, 'steps': 26731, 'batch_loss/train': 0.6709648980759084}
+12/26/2021 11:46:13 - INFO - codeparrot_training - Step 26732: {'lr': 0.00022412395210405712, 'samples': 13687296, 'steps': 26732, 'batch_loss/train': 0.574543856899254}
+12/26/2021 11:46:24 - INFO - codeparrot_training - Step 26733: {'lr': 0.00022410827368478557, 'samples': 13687808, 'steps': 26733, 'batch_loss/train': 0.711560832336545}
+12/26/2021 11:46:35 - INFO - codeparrot_training - Step 26734: {'lr': 0.00022409259536844967, 'samples': 13688320, 'steps': 26734, 'batch_loss/train': 0.6947666211053729}
+12/26/2021 11:46:48 - INFO - codeparrot_training - Step 26735: {'lr': 0.00022407691715511198, 'samples': 13688832, 'steps': 26735, 'batch_loss/train': 0.8241451736539602}
+12/26/2021 11:46:59 - INFO - codeparrot_training - Step 26736: {'lr': 0.0002240612390448346, 'samples': 13689344, 'steps': 26736, 'batch_loss/train': 0.8094647875986993}
+12/26/2021 11:47:10 - INFO - codeparrot_training - Step 26737: {'lr': 0.00022404556103768002, 'samples': 13689856, 'steps': 26737, 'batch_loss/train': 0.728834742680192}
+12/26/2021 11:47:20 - INFO - codeparrot_training - Step 26738: {'lr': 0.00022402988313371033, 'samples': 13690368, 'steps': 26738, 'batch_loss/train': 0.7604713602922857}
+12/26/2021 11:47:32 - INFO - codeparrot_training - Step 26739: {'lr': 0.0002240142053329881, 'samples': 13690880, 'steps': 26739, 'batch_loss/train': 0.672586562926881}
+12/26/2021 11:47:43 - INFO - codeparrot_training - Step 26740: {'lr': 0.00022399852763557566, 'samples': 13691392, 'steps': 26740, 'batch_loss/train': 0.8304130518808961}
+12/26/2021 11:47:54 - INFO - codeparrot_training - Step 26741: {'lr': 0.00022398285004153525, 'samples': 13691904, 'steps': 26741, 'batch_loss/train': 0.7769549109507352}
+12/26/2021 11:48:06 - INFO - codeparrot_training - Step 26742: {'lr': 0.00022396717255092905, 'samples': 13692416, 'steps': 26742, 'batch_loss/train': 0.6626744035165757}
+12/26/2021 11:48:17 - INFO - codeparrot_training - Step 26743: {'lr': 0.00022395149516381963, 'samples': 13692928, 'steps': 26743, 'batch_loss/train': 0.46685352141503245}
+12/26/2021 11:48:27 - INFO - codeparrot_training - Step 26744: {'lr': 0.0002239358178802693, 'samples': 13693440, 'steps': 26744, 'batch_loss/train': 0.7280251104384661}
+12/26/2021 11:48:40 - INFO - codeparrot_training - Step 26745: {'lr': 0.00022392014070034022, 'samples': 13693952, 'steps': 26745, 'batch_loss/train': 0.7345063565298915}
+12/26/2021 11:48:50 - INFO - codeparrot_training - Step 26746: {'lr': 0.00022390446362409477, 'samples': 13694464, 'steps': 26746, 'batch_loss/train': 0.9260724182240665}
+12/26/2021 11:49:01 - INFO - codeparrot_training - Step 26747: {'lr': 0.00022388878665159545, 'samples': 13694976, 'steps': 26747, 'batch_loss/train': 0.8287439672276378}
+12/26/2021 11:49:11 - INFO - codeparrot_training - Step 26748: {'lr': 0.00022387310978290437, 'samples': 13695488, 'steps': 26748, 'batch_loss/train': 0.6539232429349795}
+12/26/2021 11:49:26 - INFO - codeparrot_training - Step 26749: {'lr': 0.00022385743301808392, 'samples': 13696000, 'steps': 26749, 'batch_loss/train': 0.7139667572919279}
+12/26/2021 11:49:36 - INFO - codeparrot_training - Step 26750: {'lr': 0.00022384175635719653, 'samples': 13696512, 'steps': 26750, 'batch_loss/train': 1.0195169858634472}
+12/26/2021 11:49:47 - INFO - codeparrot_training - Step 26751: {'lr': 0.00022382607980030446, 'samples': 13697024, 'steps': 26751, 'batch_loss/train': 0.7679197704419494}
+12/26/2021 11:49:59 - INFO - codeparrot_training - Step 26752: {'lr': 0.00022381040334746994, 'samples': 13697536, 'steps': 26752, 'batch_loss/train': 0.8465987355448306}
+12/26/2021 11:50:09 - INFO - codeparrot_training - Step 26753: {'lr': 0.00022379472699875534, 'samples': 13698048, 'steps': 26753, 'batch_loss/train': 0.7641827864572406}
+12/26/2021 11:50:20 - INFO - codeparrot_training - Step 26754: {'lr': 0.00022377905075422308, 'samples': 13698560, 'steps': 26754, 'batch_loss/train': 0.813064175657928}
+12/26/2021 11:50:34 - INFO - codeparrot_training - Step 26755: {'lr': 0.0002237633746139354, 'samples': 13699072, 'steps': 26755, 'batch_loss/train': 0.7422439004294574}
+12/26/2021 11:50:45 - INFO - codeparrot_training - Step 26756: {'lr': 0.00022374769857795465, 'samples': 13699584, 'steps': 26756, 'batch_loss/train': 0.7358318055048585}
+12/26/2021 11:50:55 - INFO - codeparrot_training - Step 26757: {'lr': 0.00022373202264634306, 'samples': 13700096, 'steps': 26757, 'batch_loss/train': 0.6917812356259674}
+12/26/2021 11:51:06 - INFO - codeparrot_training - Step 26758: {'lr': 0.00022371634681916307, 'samples': 13700608, 'steps': 26758, 'batch_loss/train': 0.7568380394950509}
+12/26/2021 11:51:18 - INFO - codeparrot_training - Step 26759: {'lr': 0.00022370067109647702, 'samples': 13701120, 'steps': 26759, 'batch_loss/train': 0.7143628867343068}
+12/26/2021 11:51:28 - INFO - codeparrot_training - Step 26760: {'lr': 0.00022368499547834706, 'samples': 13701632, 'steps': 26760, 'batch_loss/train': 0.8075491851195693}
+12/26/2021 11:51:39 - INFO - codeparrot_training - Step 26761: {'lr': 0.00022366931996483573, 'samples': 13702144, 'steps': 26761, 'batch_loss/train': 0.7242178930900991}
+12/26/2021 11:51:51 - INFO - codeparrot_training - Step 26762: {'lr': 0.00022365364455600518, 'samples': 13702656, 'steps': 26762, 'batch_loss/train': 0.8126704003661871}
+12/26/2021 11:52:02 - INFO - codeparrot_training - Step 26763: {'lr': 0.00022363796925191788, 'samples': 13703168, 'steps': 26763, 'batch_loss/train': 0.8090120700653642}
+12/26/2021 11:52:12 - INFO - codeparrot_training - Step 26764: {'lr': 0.0002236222940526359, 'samples': 13703680, 'steps': 26764, 'batch_loss/train': 0.7950878329575062}
+12/26/2021 11:52:26 - INFO - codeparrot_training - Step 26765: {'lr': 0.0002236066189582218, 'samples': 13704192, 'steps': 26765, 'batch_loss/train': 0.7440197328105569}
+12/26/2021 11:52:36 - INFO - codeparrot_training - Step 26766: {'lr': 0.00022359094396873789, 'samples': 13704704, 'steps': 26766, 'batch_loss/train': 0.8267139233648777}
+12/26/2021 11:52:47 - INFO - codeparrot_training - Step 26767: {'lr': 0.00022357526908424634, 'samples': 13705216, 'steps': 26767, 'batch_loss/train': 0.7268555879127234}
+12/26/2021 11:53:00 - INFO - codeparrot_training - Step 26768: {'lr': 0.00022355959430480954, 'samples': 13705728, 'steps': 26768, 'batch_loss/train': 0.7351593929342926}
+12/26/2021 11:53:11 - INFO - codeparrot_training - Step 26769: {'lr': 0.0002235439196304898, 'samples': 13706240, 'steps': 26769, 'batch_loss/train': 0.6706378739327192}
+12/26/2021 11:53:21 - INFO - codeparrot_training - Step 26770: {'lr': 0.00022352824506134951, 'samples': 13706752, 'steps': 26770, 'batch_loss/train': 0.9205794259905815}
+12/26/2021 11:53:32 - INFO - codeparrot_training - Step 26771: {'lr': 0.00022351257059745084, 'samples': 13707264, 'steps': 26771, 'batch_loss/train': 0.7534484402276576}
+12/26/2021 11:53:44 - INFO - codeparrot_training - Step 26772: {'lr': 0.00022349689623885622, 'samples': 13707776, 'steps': 26772, 'batch_loss/train': 0.8009936893358827}
+12/26/2021 11:53:54 - INFO - codeparrot_training - Step 26773: {'lr': 0.00022348122198562797, 'samples': 13708288, 'steps': 26773, 'batch_loss/train': 0.8281614175066352}
+12/26/2021 11:54:05 - INFO - codeparrot_training - Step 26774: {'lr': 0.00022346554783782835, 'samples': 13708800, 'steps': 26774, 'batch_loss/train': 0.7124905314994976}
+12/26/2021 11:54:17 - INFO - codeparrot_training - Step 26775: {'lr': 0.00022344987379551963, 'samples': 13709312, 'steps': 26775, 'batch_loss/train': 0.802395431092009}
+12/26/2021 11:54:28 - INFO - codeparrot_training - Step 26776: {'lr': 0.00022343419985876428, 'samples': 13709824, 'steps': 26776, 'batch_loss/train': 0.7627839820925146}
+12/26/2021 11:54:38 - INFO - codeparrot_training - Step 26777: {'lr': 0.00022341852602762453, 'samples': 13710336, 'steps': 26777, 'batch_loss/train': 0.8239222292322665}
+12/26/2021 11:54:51 - INFO - codeparrot_training - Step 26778: {'lr': 0.00022340285230216262, 'samples': 13710848, 'steps': 26778, 'batch_loss/train': 0.8001159750856459}
+12/26/2021 11:55:02 - INFO - codeparrot_training - Step 26779: {'lr': 0.00022338717868244088, 'samples': 13711360, 'steps': 26779, 'batch_loss/train': 0.8617224283516407}
+12/26/2021 11:55:12 - INFO - codeparrot_training - Step 26780: {'lr': 0.00022337150516852177, 'samples': 13711872, 'steps': 26780, 'batch_loss/train': 0.735003415029496}
+12/26/2021 11:55:23 - INFO - codeparrot_training - Step 26781: {'lr': 0.00022335583176046745, 'samples': 13712384, 'steps': 26781, 'batch_loss/train': 0.7173898508772254}
+12/26/2021 11:55:35 - INFO - codeparrot_training - Step 26782: {'lr': 0.00022334015845834035, 'samples': 13712896, 'steps': 26782, 'batch_loss/train': 0.7272152462974191}
+12/26/2021 11:55:46 - INFO - codeparrot_training - Step 26783: {'lr': 0.00022332448526220258, 'samples': 13713408, 'steps': 26783, 'batch_loss/train': 0.8619802165776491}
+12/26/2021 11:55:57 - INFO - codeparrot_training - Step 26784: {'lr': 0.00022330881217211667, 'samples': 13713920, 'steps': 26784, 'batch_loss/train': 0.7869294080883265}
+12/26/2021 11:56:09 - INFO - codeparrot_training - Step 26785: {'lr': 0.00022329313918814487, 'samples': 13714432, 'steps': 26785, 'batch_loss/train': 0.7647852562367916}
+12/26/2021 11:56:19 - INFO - codeparrot_training - Step 26786: {'lr': 0.00022327746631034938, 'samples': 13714944, 'steps': 26786, 'batch_loss/train': 0.7661078488454223}
+12/26/2021 11:56:30 - INFO - codeparrot_training - Step 26787: {'lr': 0.00022326179353879266, 'samples': 13715456, 'steps': 26787, 'batch_loss/train': 0.7528681075200438}
+12/26/2021 11:56:43 - INFO - codeparrot_training - Step 26788: {'lr': 0.00022324612087353695, 'samples': 13715968, 'steps': 26788, 'batch_loss/train': 0.7136106750695035}
+12/26/2021 11:56:53 - INFO - codeparrot_training - Step 26789: {'lr': 0.0002232304483146446, 'samples': 13716480, 'steps': 26789, 'batch_loss/train': 0.7386757186613977}
+12/26/2021 11:57:04 - INFO - codeparrot_training - Step 26790: {'lr': 0.00022321477586217776, 'samples': 13716992, 'steps': 26790, 'batch_loss/train': 0.8724896181374788}
+12/26/2021 11:57:14 - INFO - codeparrot_training - Step 26791: {'lr': 0.00022319910351619894, 'samples': 13717504, 'steps': 26791, 'batch_loss/train': 0.6794714301358908}
+12/26/2021 11:57:27 - INFO - codeparrot_training - Step 26792: {'lr': 0.00022318343127677038, 'samples': 13718016, 'steps': 26792, 'batch_loss/train': 2.850297162309289}
+12/26/2021 11:57:38 - INFO - codeparrot_training - Step 26793: {'lr': 0.00022316775914395432, 'samples': 13718528, 'steps': 26793, 'batch_loss/train': 0.7929796017706394}
+12/26/2021 11:57:48 - INFO - codeparrot_training - Step 26794: {'lr': 0.0002231520871178131, 'samples': 13719040, 'steps': 26794, 'batch_loss/train': 0.8509122906252742}
+12/26/2021 11:58:01 - INFO - codeparrot_training - Step 26795: {'lr': 0.0002231364151984091, 'samples': 13719552, 'steps': 26795, 'batch_loss/train': 0.7834293236956}
+12/26/2021 11:58:11 - INFO - codeparrot_training - Step 26796: {'lr': 0.00022312074338580462, 'samples': 13720064, 'steps': 26796, 'batch_loss/train': 0.8372762817889452}
+12/26/2021 11:58:22 - INFO - codeparrot_training - Step 26797: {'lr': 0.0002231050716800618, 'samples': 13720576, 'steps': 26797, 'batch_loss/train': 0.7053414818365127}
+12/26/2021 11:58:34 - INFO - codeparrot_training - Step 26798: {'lr': 0.00022308940008124308, 'samples': 13721088, 'steps': 26798, 'batch_loss/train': 0.6875747516751289}
+12/26/2021 11:58:45 - INFO - codeparrot_training - Step 26799: {'lr': 0.00022307372858941082, 'samples': 13721600, 'steps': 26799, 'batch_loss/train': 0.8327781856060028}
+12/26/2021 11:58:56 - INFO - codeparrot_training - Step 26800: {'lr': 0.00022305805720462718, 'samples': 13722112, 'steps': 26800, 'batch_loss/train': 0.8501297235488892}
+12/26/2021 11:59:06 - INFO - codeparrot_training - Step 26801: {'lr': 0.0002230423859269545, 'samples': 13722624, 'steps': 26801, 'batch_loss/train': 0.7178016903344542}
+12/26/2021 11:59:18 - INFO - codeparrot_training - Step 26802: {'lr': 0.00022302671475645522, 'samples': 13723136, 'steps': 26802, 'batch_loss/train': 0.7639355631545186}
+12/26/2021 11:59:29 - INFO - codeparrot_training - Step 26803: {'lr': 0.00022301104369319146, 'samples': 13723648, 'steps': 26803, 'batch_loss/train': 0.6907674754038453}
+12/26/2021 11:59:39 - INFO - codeparrot_training - Step 26804: {'lr': 0.0002229953727372257, 'samples': 13724160, 'steps': 26804, 'batch_loss/train': 0.8076934600248933}
+12/26/2021 11:59:52 - INFO - codeparrot_training - Step 26805: {'lr': 0.00022297970188862, 'samples': 13724672, 'steps': 26805, 'batch_loss/train': 1.6821157545782626}
+12/26/2021 12:00:03 - INFO - codeparrot_training - Step 26806: {'lr': 0.0002229640311474369, 'samples': 13725184, 'steps': 26806, 'batch_loss/train': 0.7718113958835602}
+12/26/2021 12:00:14 - INFO - codeparrot_training - Step 26807: {'lr': 0.00022294836051373857, 'samples': 13725696, 'steps': 26807, 'batch_loss/train': 0.773678632453084}
+12/26/2021 12:00:26 - INFO - codeparrot_training - Step 26808: {'lr': 0.0002229326899875873, 'samples': 13726208, 'steps': 26808, 'batch_loss/train': 0.7622329127043486}
+12/26/2021 12:00:36 - INFO - codeparrot_training - Step 26809: {'lr': 0.00022291701956904554, 'samples': 13726720, 'steps': 26809, 'batch_loss/train': 0.8234318783506751}
+12/26/2021 12:00:47 - INFO - codeparrot_training - Step 26810: {'lr': 0.0002229013492581754, 'samples': 13727232, 'steps': 26810, 'batch_loss/train': 0.7606566408649087}
+12/26/2021 12:00:58 - INFO - codeparrot_training - Step 26811: {'lr': 0.00022288567905503934, 'samples': 13727744, 'steps': 26811, 'batch_loss/train': 0.8458014251664281}
+12/26/2021 12:01:10 - INFO - codeparrot_training - Step 26812: {'lr': 0.00022287000895969948, 'samples': 13728256, 'steps': 26812, 'batch_loss/train': 0.8317914055660367}
+12/26/2021 12:01:20 - INFO - codeparrot_training - Step 26813: {'lr': 0.0002228543389722183, 'samples': 13728768, 'steps': 26813, 'batch_loss/train': 0.7181926819030195}
+12/26/2021 12:01:31 - INFO - codeparrot_training - Step 26814: {'lr': 0.00022283866909265798, 'samples': 13729280, 'steps': 26814, 'batch_loss/train': 0.7828273391351104}
+12/26/2021 12:01:43 - INFO - codeparrot_training - Step 26815: {'lr': 0.0002228229993210809, 'samples': 13729792, 'steps': 26815, 'batch_loss/train': 0.8542177546769381}
+12/26/2021 12:01:54 - INFO - codeparrot_training - Step 26816: {'lr': 0.00022280732965754925, 'samples': 13730304, 'steps': 26816, 'batch_loss/train': 0.7761054458096623}
+12/26/2021 12:02:05 - INFO - codeparrot_training - Step 26817: {'lr': 0.0002227916601021254, 'samples': 13730816, 'steps': 26817, 'batch_loss/train': 0.45147480204468593}
+12/26/2021 12:02:17 - INFO - codeparrot_training - Step 26818: {'lr': 0.0002227759906548717, 'samples': 13731328, 'steps': 26818, 'batch_loss/train': 0.780628441600129}
+12/26/2021 12:02:28 - INFO - codeparrot_training - Step 26819: {'lr': 0.00022276032131585027, 'samples': 13731840, 'steps': 26819, 'batch_loss/train': 0.584349409211427}
+12/26/2021 12:02:39 - INFO - codeparrot_training - Step 26820: {'lr': 0.00022274465208512356, 'samples': 13732352, 'steps': 26820, 'batch_loss/train': 0.8435436283471063}
+12/26/2021 12:02:49 - INFO - codeparrot_training - Step 26821: {'lr': 0.0002227289829627539, 'samples': 13732864, 'steps': 26821, 'batch_loss/train': 0.7436623591929674}
+12/26/2021 12:03:01 - INFO - codeparrot_training - Step 26822: {'lr': 0.0002227133139488035, 'samples': 13733376, 'steps': 26822, 'batch_loss/train': 0.7117613069713116}
+12/26/2021 12:03:12 - INFO - codeparrot_training - Step 26823: {'lr': 0.00022269764504333455, 'samples': 13733888, 'steps': 26823, 'batch_loss/train': 0.747182690538466}
+12/26/2021 12:03:23 - INFO - codeparrot_training - Step 26824: {'lr': 0.00022268197624640952, 'samples': 13734400, 'steps': 26824, 'batch_loss/train': 0.7623042110353708}
+12/26/2021 12:03:35 - INFO - codeparrot_training - Step 26825: {'lr': 0.00022266630755809067, 'samples': 13734912, 'steps': 26825, 'batch_loss/train': 0.7098951985244639}
+12/26/2021 12:03:46 - INFO - codeparrot_training - Step 26826: {'lr': 0.00022265063897844025, 'samples': 13735424, 'steps': 26826, 'batch_loss/train': 0.8082050601951778}
+12/26/2021 12:03:57 - INFO - codeparrot_training - Step 26827: {'lr': 0.00022263497050752047, 'samples': 13735936, 'steps': 26827, 'batch_loss/train': 0.6800264082849026}
+12/26/2021 12:04:09 - INFO - codeparrot_training - Step 26828: {'lr': 0.0002226193021453938, 'samples': 13736448, 'steps': 26828, 'batch_loss/train': 0.7263939605327323}
+12/26/2021 12:04:19 - INFO - codeparrot_training - Step 26829: {'lr': 0.00022260363389212242, 'samples': 13736960, 'steps': 26829, 'batch_loss/train': 0.8407434308901429}
+12/26/2021 12:04:30 - INFO - codeparrot_training - Step 26830: {'lr': 0.0002225879657477687, 'samples': 13737472, 'steps': 26830, 'batch_loss/train': 0.8158363876864314}
+12/26/2021 12:04:40 - INFO - codeparrot_training - Step 26831: {'lr': 0.00022257229771239479, 'samples': 13737984, 'steps': 26831, 'batch_loss/train': 0.6993914027698338}
+12/26/2021 12:04:53 - INFO - codeparrot_training - Step 26832: {'lr': 0.00022255662978606317, 'samples': 13738496, 'steps': 26832, 'batch_loss/train': 0.8400634983554482}
+12/26/2021 12:05:03 - INFO - codeparrot_training - Step 26833: {'lr': 0.00022254096196883594, 'samples': 13739008, 'steps': 26833, 'batch_loss/train': 0.766527112107724}
+12/26/2021 12:05:14 - INFO - codeparrot_training - Step 26834: {'lr': 0.00022252529426077544, 'samples': 13739520, 'steps': 26834, 'batch_loss/train': 0.773870796430856}
+12/26/2021 12:05:27 - INFO - codeparrot_training - Step 26835: {'lr': 0.00022250962666194412, 'samples': 13740032, 'steps': 26835, 'batch_loss/train': 1.0740570342168212}
+12/26/2021 12:05:37 - INFO - codeparrot_training - Step 26836: {'lr': 0.00022249395917240409, 'samples': 13740544, 'steps': 26836, 'batch_loss/train': 0.7754856683313847}
+12/26/2021 12:05:48 - INFO - codeparrot_training - Step 26837: {'lr': 0.00022247829179221774, 'samples': 13741056, 'steps': 26837, 'batch_loss/train': 0.7529646942857653}
+12/26/2021 12:06:00 - INFO - codeparrot_training - Step 26838: {'lr': 0.00022246262452144718, 'samples': 13741568, 'steps': 26838, 'batch_loss/train': 0.7397330235689878}
+12/26/2021 12:06:11 - INFO - codeparrot_training - Step 26839: {'lr': 0.00022244695736015491, 'samples': 13742080, 'steps': 26839, 'batch_loss/train': 0.767320801387541}
+12/26/2021 12:06:21 - INFO - codeparrot_training - Step 26840: {'lr': 0.00022243129030840317, 'samples': 13742592, 'steps': 26840, 'batch_loss/train': 0.8042468512430787}
+12/26/2021 12:06:32 - INFO - codeparrot_training - Step 26841: {'lr': 0.0002224156233662542, 'samples': 13743104, 'steps': 26841, 'batch_loss/train': 0.784406702965498}
+12/26/2021 12:06:45 - INFO - codeparrot_training - Step 26842: {'lr': 0.0002223999565337702, 'samples': 13743616, 'steps': 26842, 'batch_loss/train': 0.8355715675279498}
+12/26/2021 12:06:55 - INFO - codeparrot_training - Step 26843: {'lr': 0.00022238428981101362, 'samples': 13744128, 'steps': 26843, 'batch_loss/train': 0.8051990633830428}
+12/26/2021 12:07:06 - INFO - codeparrot_training - Step 26844: {'lr': 0.00022236862319804674, 'samples': 13744640, 'steps': 26844, 'batch_loss/train': 0.7813010402023792}
+12/26/2021 12:07:19 - INFO - codeparrot_training - Step 26845: {'lr': 0.00022235295669493165, 'samples': 13745152, 'steps': 26845, 'batch_loss/train': 0.7996572474949062}
+12/26/2021 12:07:30 - INFO - codeparrot_training - Step 26846: {'lr': 0.00022233729030173084, 'samples': 13745664, 'steps': 26846, 'batch_loss/train': 0.6430337727069855}
+12/26/2021 12:07:41 - INFO - codeparrot_training - Step 26847: {'lr': 0.00022232162401850653, 'samples': 13746176, 'steps': 26847, 'batch_loss/train': 1.0970399267971516}
+12/26/2021 12:07:51 - INFO - codeparrot_training - Step 26848: {'lr': 0.00022230595784532097, 'samples': 13746688, 'steps': 26848, 'batch_loss/train': 1.1000310322269797}
+12/26/2021 12:08:03 - INFO - codeparrot_training - Step 26849: {'lr': 0.00022229029178223638, 'samples': 13747200, 'steps': 26849, 'batch_loss/train': 0.8026437042281032}
+12/26/2021 12:08:14 - INFO - codeparrot_training - Step 26850: {'lr': 0.0002222746258293153, 'samples': 13747712, 'steps': 26850, 'batch_loss/train': 0.7355893284548074}
+12/26/2021 12:08:24 - INFO - codeparrot_training - Step 26851: {'lr': 0.0002222589599866198, 'samples': 13748224, 'steps': 26851, 'batch_loss/train': 0.7877711039036512}
+12/26/2021 12:08:37 - INFO - codeparrot_training - Step 26852: {'lr': 0.00022224329425421215, 'samples': 13748736, 'steps': 26852, 'batch_loss/train': 0.6617656713351607}
+12/26/2021 12:08:47 - INFO - codeparrot_training - Step 26853: {'lr': 0.0002222276286321546, 'samples': 13749248, 'steps': 26853, 'batch_loss/train': 0.7899280283600092}
+12/26/2021 12:08:58 - INFO - codeparrot_training - Step 26854: {'lr': 0.00022221196312050966, 'samples': 13749760, 'steps': 26854, 'batch_loss/train': 0.7865721830166876}
+12/26/2021 12:09:11 - INFO - codeparrot_training - Step 26855: {'lr': 0.0002221962977193394, 'samples': 13750272, 'steps': 26855, 'batch_loss/train': 0.5480280756019056}
+12/26/2021 12:09:21 - INFO - codeparrot_training - Step 26856: {'lr': 0.00022218063242870607, 'samples': 13750784, 'steps': 26856, 'batch_loss/train': 0.7746148332953453}
+12/26/2021 12:09:32 - INFO - codeparrot_training - Step 26857: {'lr': 0.0002221649672486722, 'samples': 13751296, 'steps': 26857, 'batch_loss/train': 0.757344757206738}
+12/26/2021 12:09:44 - INFO - codeparrot_training - Step 26858: {'lr': 0.00022214930217929988, 'samples': 13751808, 'steps': 26858, 'batch_loss/train': 0.7879700260236859}
+12/26/2021 12:09:55 - INFO - codeparrot_training - Step 26859: {'lr': 0.00022213363722065136, 'samples': 13752320, 'steps': 26859, 'batch_loss/train': 0.7987445006147027}
+12/26/2021 12:10:05 - INFO - codeparrot_training - Step 26860: {'lr': 0.00022211797237278895, 'samples': 13752832, 'steps': 26860, 'batch_loss/train': 0.8768093446269631}
+12/26/2021 12:10:16 - INFO - codeparrot_training - Step 26861: {'lr': 0.00022210230763577504, 'samples': 13753344, 'steps': 26861, 'batch_loss/train': 0.7112472879234701}
+12/26/2021 12:10:28 - INFO - codeparrot_training - Step 26862: {'lr': 0.00022208664300967177, 'samples': 13753856, 'steps': 26862, 'batch_loss/train': 0.6738438033498824}
+12/26/2021 12:10:38 - INFO - codeparrot_training - Step 26863: {'lr': 0.00022207097849454152, 'samples': 13754368, 'steps': 26863, 'batch_loss/train': 0.692893841303885}
+12/26/2021 12:10:49 - INFO - codeparrot_training - Step 26864: {'lr': 0.00022205531409044642, 'samples': 13754880, 'steps': 26864, 'batch_loss/train': 0.8456500321626663}
+12/26/2021 12:11:02 - INFO - codeparrot_training - Step 26865: {'lr': 0.0002220396497974489, 'samples': 13755392, 'steps': 26865, 'batch_loss/train': 0.6859007223974913}
+12/26/2021 12:11:12 - INFO - codeparrot_training - Step 26866: {'lr': 0.00022202398561561122, 'samples': 13755904, 'steps': 26866, 'batch_loss/train': 0.7807025155052543}
+12/26/2021 12:11:23 - INFO - codeparrot_training - Step 26867: {'lr': 0.0002220083215449956, 'samples': 13756416, 'steps': 26867, 'batch_loss/train': 0.7742145815864205}
+12/26/2021 12:11:35 - INFO - codeparrot_training - Step 26868: {'lr': 0.0002219926575856642, 'samples': 13756928, 'steps': 26868, 'batch_loss/train': 0.7385332202538848}
+12/26/2021 12:11:46 - INFO - codeparrot_training - Step 26869: {'lr': 0.0002219769937376795, 'samples': 13757440, 'steps': 26869, 'batch_loss/train': 0.7771449470892549}
+12/26/2021 12:11:56 - INFO - codeparrot_training - Step 26870: {'lr': 0.00022196133000110374, 'samples': 13757952, 'steps': 26870, 'batch_loss/train': 0.7919516563415527}
+12/26/2021 12:12:07 - INFO - codeparrot_training - Step 26871: {'lr': 0.00022194566637599904, 'samples': 13758464, 'steps': 26871, 'batch_loss/train': 0.7733302777633071}
+12/26/2021 12:12:20 - INFO - codeparrot_training - Step 26872: {'lr': 0.0002219300028624278, 'samples': 13758976, 'steps': 26872, 'batch_loss/train': 1.1943327018525451}
+12/26/2021 12:12:30 - INFO - codeparrot_training - Step 26873: {'lr': 0.00022191433946045237, 'samples': 13759488, 'steps': 26873, 'batch_loss/train': 0.8430076334625483}
+12/26/2021 12:12:41 - INFO - codeparrot_training - Step 26874: {'lr': 0.00022189867617013485, 'samples': 13760000, 'steps': 26874, 'batch_loss/train': 0.7361236289143562}
+12/26/2021 12:12:53 - INFO - codeparrot_training - Step 26875: {'lr': 0.0002218830129915375, 'samples': 13760512, 'steps': 26875, 'batch_loss/train': 0.777714729309082}
+12/26/2021 12:13:04 - INFO - codeparrot_training - Step 26876: {'lr': 0.00022186734992472285, 'samples': 13761024, 'steps': 26876, 'batch_loss/train': 0.9971666326746345}
+12/26/2021 12:13:14 - INFO - codeparrot_training - Step 26877: {'lr': 0.00022185168696975293, 'samples': 13761536, 'steps': 26877, 'batch_loss/train': 0.977806318551302}
+12/26/2021 12:13:26 - INFO - codeparrot_training - Step 26878: {'lr': 0.00022183602412669003, 'samples': 13762048, 'steps': 26878, 'batch_loss/train': 1.143821744248271}
+12/26/2021 12:13:37 - INFO - codeparrot_training - Step 26879: {'lr': 0.00022182036139559643, 'samples': 13762560, 'steps': 26879, 'batch_loss/train': 0.814595403149724}
+12/26/2021 12:13:48 - INFO - codeparrot_training - Step 26880: {'lr': 0.00022180469877653455, 'samples': 13763072, 'steps': 26880, 'batch_loss/train': 0.7866146340966225}
+12/26/2021 12:13:58 - INFO - codeparrot_training - Step 26881: {'lr': 0.0002217890362695665, 'samples': 13763584, 'steps': 26881, 'batch_loss/train': 0.6592147927731276}
+12/26/2021 12:14:16 - INFO - codeparrot_training - Step 26882: {'lr': 0.00022177337387475449, 'samples': 13764096, 'steps': 26882, 'batch_loss/train': 0.7920673051849008}
+12/26/2021 12:14:26 - INFO - codeparrot_training - Step 26883: {'lr': 0.000221757711592161, 'samples': 13764608, 'steps': 26883, 'batch_loss/train': 0.819673668127507}
+12/26/2021 12:14:37 - INFO - codeparrot_training - Step 26884: {'lr': 0.00022174204942184813, 'samples': 13765120, 'steps': 26884, 'batch_loss/train': 0.8063693894073367}
+12/26/2021 12:14:49 - INFO - codeparrot_training - Step 26885: {'lr': 0.0002217263873638783, 'samples': 13765632, 'steps': 26885, 'batch_loss/train': 0.9522474677069113}
+12/26/2021 12:15:00 - INFO - codeparrot_training - Step 26886: {'lr': 0.00022171072541831352, 'samples': 13766144, 'steps': 26886, 'batch_loss/train': 0.7217081125709228}
+12/26/2021 12:15:10 - INFO - codeparrot_training - Step 26887: {'lr': 0.00022169506358521635, 'samples': 13766656, 'steps': 26887, 'batch_loss/train': 0.7735015712678432}
+12/26/2021 12:15:23 - INFO - codeparrot_training - Step 26888: {'lr': 0.00022167940186464886, 'samples': 13767168, 'steps': 26888, 'batch_loss/train': 0.6956087579019368}
+12/26/2021 12:15:34 - INFO - codeparrot_training - Step 26889: {'lr': 0.00022166374025667343, 'samples': 13767680, 'steps': 26889, 'batch_loss/train': 0.7983135674148798}
+12/26/2021 12:15:44 - INFO - codeparrot_training - Step 26890: {'lr': 0.00022164807876135214, 'samples': 13768192, 'steps': 26890, 'batch_loss/train': 0.6301028807647526}
+12/26/2021 12:15:55 - INFO - codeparrot_training - Step 26891: {'lr': 0.00022163241737874748, 'samples': 13768704, 'steps': 26891, 'batch_loss/train': 0.8184030428528786}
+12/26/2021 12:16:07 - INFO - codeparrot_training - Step 26892: {'lr': 0.00022161675610892162, 'samples': 13769216, 'steps': 26892, 'batch_loss/train': 0.8600868084467947}
+12/26/2021 12:16:18 - INFO - codeparrot_training - Step 26893: {'lr': 0.00022160109495193676, 'samples': 13769728, 'steps': 26893, 'batch_loss/train': 1.0054925084114075}
+12/26/2021 12:16:28 - INFO - codeparrot_training - Step 26894: {'lr': 0.00022158543390785523, 'samples': 13770240, 'steps': 26894, 'batch_loss/train': 0.6680535636842251}
+12/26/2021 12:16:41 - INFO - codeparrot_training - Step 26895: {'lr': 0.00022156977297673938, 'samples': 13770752, 'steps': 26895, 'batch_loss/train': 0.6908709881827235}
+12/26/2021 12:16:51 - INFO - codeparrot_training - Step 26896: {'lr': 0.00022155411215865137, 'samples': 13771264, 'steps': 26896, 'batch_loss/train': 0.6544442421291023}
+12/26/2021 12:17:02 - INFO - codeparrot_training - Step 26897: {'lr': 0.00022153845145365337, 'samples': 13771776, 'steps': 26897, 'batch_loss/train': 0.7569029438309371}
+12/26/2021 12:17:14 - INFO - codeparrot_training - Step 26898: {'lr': 0.00022152279086180778, 'samples': 13772288, 'steps': 26898, 'batch_loss/train': 0.8283262569457293}
+12/26/2021 12:17:25 - INFO - codeparrot_training - Step 26899: {'lr': 0.00022150713038317688, 'samples': 13772800, 'steps': 26899, 'batch_loss/train': 0.6689525116235018}
+12/26/2021 12:17:35 - INFO - codeparrot_training - Step 26900: {'lr': 0.0002214914700178228, 'samples': 13773312, 'steps': 26900, 'batch_loss/train': 0.675526135135442}
+12/26/2021 12:17:46 - INFO - codeparrot_training - Step 26901: {'lr': 0.00022147580976580782, 'samples': 13773824, 'steps': 26901, 'batch_loss/train': 0.7552426075562835}
+12/26/2021 12:17:58 - INFO - codeparrot_training - Step 26902: {'lr': 0.00022146014962719438, 'samples': 13774336, 'steps': 26902, 'batch_loss/train': 0.8152258079499006}
+12/26/2021 12:18:09 - INFO - codeparrot_training - Step 26903: {'lr': 0.00022144448960204462, 'samples': 13774848, 'steps': 26903, 'batch_loss/train': 0.6719792718067765}
+12/26/2021 12:18:19 - INFO - codeparrot_training - Step 26904: {'lr': 0.0002214288296904207, 'samples': 13775360, 'steps': 26904, 'batch_loss/train': 0.7214266150258482}
+12/26/2021 12:18:32 - INFO - codeparrot_training - Step 26905: {'lr': 0.0002214131698923849, 'samples': 13775872, 'steps': 26905, 'batch_loss/train': 0.820925259962678}
+12/26/2021 12:18:43 - INFO - codeparrot_training - Step 26906: {'lr': 0.00022139751020799968, 'samples': 13776384, 'steps': 26906, 'batch_loss/train': 0.7527413037605584}
+12/26/2021 12:18:53 - INFO - codeparrot_training - Step 26907: {'lr': 0.00022138185063732708, 'samples': 13776896, 'steps': 26907, 'batch_loss/train': 0.6894842847250402}
+12/26/2021 12:19:05 - INFO - codeparrot_training - Step 26908: {'lr': 0.00022136619118042942, 'samples': 13777408, 'steps': 26908, 'batch_loss/train': 0.7620302494615316}
+12/26/2021 12:19:16 - INFO - codeparrot_training - Step 26909: {'lr': 0.00022135053183736905, 'samples': 13777920, 'steps': 26909, 'batch_loss/train': 1.0828076107427478}
+12/26/2021 12:19:27 - INFO - codeparrot_training - Step 26910: {'lr': 0.00022133487260820808, 'samples': 13778432, 'steps': 26910, 'batch_loss/train': 0.8808481618762016}
+12/26/2021 12:19:38 - INFO - codeparrot_training - Step 26911: {'lr': 0.00022131921349300893, 'samples': 13778944, 'steps': 26911, 'batch_loss/train': 0.7959563378244638}
+12/26/2021 12:19:50 - INFO - codeparrot_training - Step 26912: {'lr': 0.0002213035544918336, 'samples': 13779456, 'steps': 26912, 'batch_loss/train': 0.7652749232947826}
+12/26/2021 12:20:01 - INFO - codeparrot_training - Step 26913: {'lr': 0.00022128789560474466, 'samples': 13779968, 'steps': 26913, 'batch_loss/train': 0.8190188645385206}
+12/26/2021 12:20:11 - INFO - codeparrot_training - Step 26914: {'lr': 0.00022127223683180415, 'samples': 13780480, 'steps': 26914, 'batch_loss/train': 0.7522309748455882}
+12/26/2021 12:20:24 - INFO - codeparrot_training - Step 26915: {'lr': 0.0002212565781730744, 'samples': 13780992, 'steps': 26915, 'batch_loss/train': 0.7729719779454172}
+12/26/2021 12:20:34 - INFO - codeparrot_training - Step 26916: {'lr': 0.00022124091962861755, 'samples': 13781504, 'steps': 26916, 'batch_loss/train': 0.8246688917279243}
+12/26/2021 12:20:45 - INFO - codeparrot_training - Step 26917: {'lr': 0.00022122526119849602, 'samples': 13782016, 'steps': 26917, 'batch_loss/train': 0.7731840568594635}
+12/26/2021 12:20:57 - INFO - codeparrot_training - Step 26918: {'lr': 0.00022120960288277201, 'samples': 13782528, 'steps': 26918, 'batch_loss/train': 0.7323749829083681}
+12/26/2021 12:21:08 - INFO - codeparrot_training - Step 26919: {'lr': 0.00022119394468150764, 'samples': 13783040, 'steps': 26919, 'batch_loss/train': 0.59223563503474}
+12/26/2021 12:21:19 - INFO - codeparrot_training - Step 26920: {'lr': 0.00022117828659476532, 'samples': 13783552, 'steps': 26920, 'batch_loss/train': 0.7884430647827685}
+12/26/2021 12:21:31 - INFO - codeparrot_training - Step 26921: {'lr': 0.0002211626286226073, 'samples': 13784064, 'steps': 26921, 'batch_loss/train': 1.1021805945783854}
+12/26/2021 12:21:42 - INFO - codeparrot_training - Step 26922: {'lr': 0.00022114697076509583, 'samples': 13784576, 'steps': 26922, 'batch_loss/train': 0.7746934378519654}
+12/26/2021 12:21:53 - INFO - codeparrot_training - Step 26923: {'lr': 0.00022113131302229294, 'samples': 13785088, 'steps': 26923, 'batch_loss/train': 0.8097754147602245}
+12/26/2021 12:22:03 - INFO - codeparrot_training - Step 26924: {'lr': 0.00022111565539426113, 'samples': 13785600, 'steps': 26924, 'batch_loss/train': 0.6941768364049494}
+12/26/2021 12:22:15 - INFO - codeparrot_training - Step 26925: {'lr': 0.00022109999788106263, 'samples': 13786112, 'steps': 26925, 'batch_loss/train': 0.7548412219621241}
+12/26/2021 12:22:26 - INFO - codeparrot_training - Step 26926: {'lr': 0.00022108434048275955, 'samples': 13786624, 'steps': 26926, 'batch_loss/train': 0.748389377258718}
+12/26/2021 12:22:36 - INFO - codeparrot_training - Step 26927: {'lr': 0.00022106868319941416, 'samples': 13787136, 'steps': 26927, 'batch_loss/train': 0.7457506963983178}
+12/26/2021 12:22:50 - INFO - codeparrot_training - Step 26928: {'lr': 0.0002210530260310889, 'samples': 13787648, 'steps': 26928, 'batch_loss/train': 0.7677046461030841}
+12/26/2021 12:23:00 - INFO - codeparrot_training - Step 26929: {'lr': 0.00022103736897784583, 'samples': 13788160, 'steps': 26929, 'batch_loss/train': 0.7903059343807399}
+12/26/2021 12:23:11 - INFO - codeparrot_training - Step 26930: {'lr': 0.00022102171203974713, 'samples': 13788672, 'steps': 26930, 'batch_loss/train': 0.5825669071637094}
+12/26/2021 12:23:21 - INFO - codeparrot_training - Step 26931: {'lr': 0.00022100605521685533, 'samples': 13789184, 'steps': 26931, 'batch_loss/train': 1.4838593173772097}
+12/26/2021 12:23:34 - INFO - codeparrot_training - Step 26932: {'lr': 0.00022099039850923246, 'samples': 13789696, 'steps': 26932, 'batch_loss/train': 0.8256744435057044}
+12/26/2021 12:23:44 - INFO - codeparrot_training - Step 26933: {'lr': 0.0002209747419169408, 'samples': 13790208, 'steps': 26933, 'batch_loss/train': 0.9038686603307724}
+12/26/2021 12:23:55 - INFO - codeparrot_training - Step 26934: {'lr': 0.00022095908544004253, 'samples': 13790720, 'steps': 26934, 'batch_loss/train': 0.8370175184682012}
+12/26/2021 12:24:07 - INFO - codeparrot_training - Step 26935: {'lr': 0.0002209434290786001, 'samples': 13791232, 'steps': 26935, 'batch_loss/train': 0.8107154318131506}
+12/26/2021 12:24:17 - INFO - codeparrot_training - Step 26936: {'lr': 0.00022092777283267553, 'samples': 13791744, 'steps': 26936, 'batch_loss/train': 0.678151513915509}
+12/26/2021 12:24:28 - INFO - codeparrot_training - Step 26937: {'lr': 0.00022091211670233123, 'samples': 13792256, 'steps': 26937, 'batch_loss/train': 0.8012394467368722}
+12/26/2021 12:24:40 - INFO - codeparrot_training - Step 26938: {'lr': 0.00022089646068762931, 'samples': 13792768, 'steps': 26938, 'batch_loss/train': 0.7890409119427204}
+12/26/2021 12:24:51 - INFO - codeparrot_training - Step 26939: {'lr': 0.00022088080478863207, 'samples': 13793280, 'steps': 26939, 'batch_loss/train': 0.7706225030124187}
+12/26/2021 12:25:02 - INFO - codeparrot_training - Step 26940: {'lr': 0.00022086514900540184, 'samples': 13793792, 'steps': 26940, 'batch_loss/train': 0.7992990287020802}
+12/26/2021 12:25:14 - INFO - codeparrot_training - Step 26941: {'lr': 0.00022084949333800075, 'samples': 13794304, 'steps': 26941, 'batch_loss/train': 0.9773413995280862}
+12/26/2021 12:25:25 - INFO - codeparrot_training - Step 26942: {'lr': 0.00022083383778649096, 'samples': 13794816, 'steps': 26942, 'batch_loss/train': 0.7390576009638608}
+12/26/2021 12:25:36 - INFO - codeparrot_training - Step 26943: {'lr': 0.0002208181823509349, 'samples': 13795328, 'steps': 26943, 'batch_loss/train': 0.8162013106048107}
+12/26/2021 12:25:48 - INFO - codeparrot_training - Step 26944: {'lr': 0.00022080252703139474, 'samples': 13795840, 'steps': 26944, 'batch_loss/train': 0.7922199815511703}
+12/26/2021 12:25:58 - INFO - codeparrot_training - Step 26945: {'lr': 0.00022078687182793265, 'samples': 13796352, 'steps': 26945, 'batch_loss/train': 0.8028006684035063}
+12/26/2021 12:26:09 - INFO - codeparrot_training - Step 26946: {'lr': 0.00022077121674061095, 'samples': 13796864, 'steps': 26946, 'batch_loss/train': 0.7663517042528838}
+12/26/2021 12:26:19 - INFO - codeparrot_training - Step 26947: {'lr': 0.00022075556176949192, 'samples': 13797376, 'steps': 26947, 'batch_loss/train': 0.8363300557248294}
+12/26/2021 12:26:32 - INFO - codeparrot_training - Step 26948: {'lr': 0.00022073990691463775, 'samples': 13797888, 'steps': 26948, 'batch_loss/train': 0.7961086360737681}
+12/26/2021 12:26:42 - INFO - codeparrot_training - Step 26949: {'lr': 0.00022072425217611053, 'samples': 13798400, 'steps': 26949, 'batch_loss/train': 0.9001552425324917}
+12/26/2021 12:26:53 - INFO - codeparrot_training - Step 26950: {'lr': 0.00022070859755397268, 'samples': 13798912, 'steps': 26950, 'batch_loss/train': 0.81758813560009}
+12/26/2021 12:27:06 - INFO - codeparrot_training - Step 26951: {'lr': 0.00022069294304828647, 'samples': 13799424, 'steps': 26951, 'batch_loss/train': 0.8528676414862275}
+12/26/2021 12:27:16 - INFO - codeparrot_training - Step 26952: {'lr': 0.00022067728865911396, 'samples': 13799936, 'steps': 26952, 'batch_loss/train': 0.8304093349725008}
+12/26/2021 12:27:27 - INFO - codeparrot_training - Step 26953: {'lr': 0.00022066163438651742, 'samples': 13800448, 'steps': 26953, 'batch_loss/train': 0.7794530354440212}
+12/26/2021 12:27:39 - INFO - codeparrot_training - Step 26954: {'lr': 0.0002206459802305593, 'samples': 13800960, 'steps': 26954, 'batch_loss/train': 0.7005040431395173}
+12/26/2021 12:27:50 - INFO - codeparrot_training - Step 26955: {'lr': 0.0002206303261913016, 'samples': 13801472, 'steps': 26955, 'batch_loss/train': 0.743810866959393}
+12/26/2021 12:28:00 - INFO - codeparrot_training - Step 26956: {'lr': 0.00022061467226880657, 'samples': 13801984, 'steps': 26956, 'batch_loss/train': 0.6777687013964169}
+12/26/2021 12:28:11 - INFO - codeparrot_training - Step 26957: {'lr': 0.00022059901846313663, 'samples': 13802496, 'steps': 26957, 'batch_loss/train': 0.7706391140818596}
+12/26/2021 12:28:24 - INFO - codeparrot_training - Step 26958: {'lr': 0.0002205833647743539, 'samples': 13803008, 'steps': 26958, 'batch_loss/train': 0.7633831053972244}
+12/26/2021 12:28:34 - INFO - codeparrot_training - Step 26959: {'lr': 0.00022056771120252053, 'samples': 13803520, 'steps': 26959, 'batch_loss/train': 0.7450809633592144}
+12/26/2021 12:28:45 - INFO - codeparrot_training - Step 26960: {'lr': 0.00022055205774769877, 'samples': 13804032, 'steps': 26960, 'batch_loss/train': 0.7906965763540938}
+12/26/2021 12:28:57 - INFO - codeparrot_training - Step 26961: {'lr': 0.00022053640440995102, 'samples': 13804544, 'steps': 26961, 'batch_loss/train': 0.7080852688523009}
+12/26/2021 12:29:07 - INFO - codeparrot_training - Step 26962: {'lr': 0.00022052075118933937, 'samples': 13805056, 'steps': 26962, 'batch_loss/train': 0.8069335306063294}
+12/26/2021 12:29:18 - INFO - codeparrot_training - Step 26963: {'lr': 0.00022050509808592612, 'samples': 13805568, 'steps': 26963, 'batch_loss/train': 0.7631840375252068}
+12/26/2021 12:29:30 - INFO - codeparrot_training - Step 26964: {'lr': 0.00022048944509977335, 'samples': 13806080, 'steps': 26964, 'batch_loss/train': 0.7285250294953585}
+12/26/2021 12:29:41 - INFO - codeparrot_training - Step 26965: {'lr': 0.00022047379223094347, 'samples': 13806592, 'steps': 26965, 'batch_loss/train': 0.6962743575568311}
+12/26/2021 12:29:52 - INFO - codeparrot_training - Step 26966: {'lr': 0.00022045813947949865, 'samples': 13807104, 'steps': 26966, 'batch_loss/train': 0.876103950664401}
+12/26/2021 12:30:02 - INFO - codeparrot_training - Step 26967: {'lr': 0.00022044248684550106, 'samples': 13807616, 'steps': 26967, 'batch_loss/train': 0.8048037430271506}
+12/26/2021 12:30:15 - INFO - codeparrot_training - Step 26968: {'lr': 0.0002204268343290131, 'samples': 13808128, 'steps': 26968, 'batch_loss/train': 0.739431219175458}
+12/26/2021 12:30:26 - INFO - codeparrot_training - Step 26969: {'lr': 0.0002204111819300968, 'samples': 13808640, 'steps': 26969, 'batch_loss/train': 1.096088821592275}
+12/26/2021 12:30:37 - INFO - codeparrot_training - Step 26970: {'lr': 0.00022039552964881452, 'samples': 13809152, 'steps': 26970, 'batch_loss/train': 0.8523690355941653}
+12/26/2021 12:30:49 - INFO - codeparrot_training - Step 26971: {'lr': 0.00022037987748522832, 'samples': 13809664, 'steps': 26971, 'batch_loss/train': 0.9263286064378917}
+12/26/2021 12:30:59 - INFO - codeparrot_training - Step 26972: {'lr': 0.0002203642254394006, 'samples': 13810176, 'steps': 26972, 'batch_loss/train': 0.7456491100601852}
+12/26/2021 12:31:10 - INFO - codeparrot_training - Step 26973: {'lr': 0.00022034857351139358, 'samples': 13810688, 'steps': 26973, 'batch_loss/train': 0.63673903234303}
+12/26/2021 12:31:22 - INFO - codeparrot_training - Step 26974: {'lr': 0.0002203329217012694, 'samples': 13811200, 'steps': 26974, 'batch_loss/train': 0.7526087742298841}
+12/26/2021 12:31:33 - INFO - codeparrot_training - Step 26975: {'lr': 0.00022031727000909027, 'samples': 13811712, 'steps': 26975, 'batch_loss/train': 0.84923948161304}
+12/26/2021 12:31:43 - INFO - codeparrot_training - Step 26976: {'lr': 0.00022030161843491857, 'samples': 13812224, 'steps': 26976, 'batch_loss/train': 0.8585964189842343}
+12/26/2021 12:31:54 - INFO - codeparrot_training - Step 26977: {'lr': 0.00022028596697881643, 'samples': 13812736, 'steps': 26977, 'batch_loss/train': 0.7541988678276539}
+12/26/2021 12:32:06 - INFO - codeparrot_training - Step 26978: {'lr': 0.00022027031564084598, 'samples': 13813248, 'steps': 26978, 'batch_loss/train': 0.7926886482164264}
+12/26/2021 12:32:17 - INFO - codeparrot_training - Step 26979: {'lr': 0.0002202546644210695, 'samples': 13813760, 'steps': 26979, 'batch_loss/train': 0.7955847587436438}
+12/26/2021 12:32:28 - INFO - codeparrot_training - Step 26980: {'lr': 0.00022023901331954935, 'samples': 13814272, 'steps': 26980, 'batch_loss/train': 0.7680556392297149}
+12/26/2021 12:32:41 - INFO - codeparrot_training - Step 26981: {'lr': 0.00022022336233634758, 'samples': 13814784, 'steps': 26981, 'batch_loss/train': 0.9030226496979594}
+12/26/2021 12:32:51 - INFO - codeparrot_training - Step 26982: {'lr': 0.00022020771147152643, 'samples': 13815296, 'steps': 26982, 'batch_loss/train': 0.7730351767968386}
+12/26/2021 12:33:02 - INFO - codeparrot_training - Step 26983: {'lr': 0.00022019206072514827, 'samples': 13815808, 'steps': 26983, 'batch_loss/train': 0.7203350877389312}
+12/26/2021 12:33:14 - INFO - codeparrot_training - Step 26984: {'lr': 0.0002201764100972752, 'samples': 13816320, 'steps': 26984, 'batch_loss/train': 0.8322451766580343}
+12/26/2021 12:33:25 - INFO - codeparrot_training - Step 26985: {'lr': 0.00022016075958796946, 'samples': 13816832, 'steps': 26985, 'batch_loss/train': 0.8818021677434444}
+12/26/2021 12:33:35 - INFO - codeparrot_training - Step 26986: {'lr': 0.00022014510919729323, 'samples': 13817344, 'steps': 26986, 'batch_loss/train': 0.7441869508475065}
+12/26/2021 12:33:46 - INFO - codeparrot_training - Step 26987: {'lr': 0.00022012945892530884, 'samples': 13817856, 'steps': 26987, 'batch_loss/train': 0.8289701147004962}
+12/26/2021 12:33:59 - INFO - codeparrot_training - Step 26988: {'lr': 0.0002201138087720784, 'samples': 13818368, 'steps': 26988, 'batch_loss/train': 0.6286798673681915}
+12/26/2021 12:34:10 - INFO - codeparrot_training - Step 26989: {'lr': 0.00022009815873766421, 'samples': 13818880, 'steps': 26989, 'batch_loss/train': 0.7761516403406858}
+12/26/2021 12:34:21 - INFO - codeparrot_training - Step 26990: {'lr': 0.00022008250882212836, 'samples': 13819392, 'steps': 26990, 'batch_loss/train': 0.7179976042825729}
+12/26/2021 12:34:33 - INFO - codeparrot_training - Step 26991: {'lr': 0.0002200668590255332, 'samples': 13819904, 'steps': 26991, 'batch_loss/train': 0.6701143195386976}
+12/26/2021 12:34:43 - INFO - codeparrot_training - Step 26992: {'lr': 0.00022005120934794097, 'samples': 13820416, 'steps': 26992, 'batch_loss/train': 0.7933606458827853}
+12/26/2021 12:34:54 - INFO - codeparrot_training - Step 26993: {'lr': 0.00022003555978941373, 'samples': 13820928, 'steps': 26993, 'batch_loss/train': 0.7773209363222122}
+12/26/2021 12:35:07 - INFO - codeparrot_training - Step 26994: {'lr': 0.00022001991035001386, 'samples': 13821440, 'steps': 26994, 'batch_loss/train': 0.7822571038268507}
+12/26/2021 12:35:17 - INFO - codeparrot_training - Step 26995: {'lr': 0.00022000426102980347, 'samples': 13821952, 'steps': 26995, 'batch_loss/train': 1.0510398243786767}
+12/26/2021 12:35:28 - INFO - codeparrot_training - Step 26996: {'lr': 0.0002199886118288449, 'samples': 13822464, 'steps': 26996, 'batch_loss/train': 0.8635897166095674}
+12/26/2021 12:35:39 - INFO - codeparrot_training - Step 26997: {'lr': 0.00021997296274720013, 'samples': 13822976, 'steps': 26997, 'batch_loss/train': 0.725394444540143}
+12/26/2021 12:35:52 - INFO - codeparrot_training - Step 26998: {'lr': 0.00021995731378493159, 'samples': 13823488, 'steps': 26998, 'batch_loss/train': 1.0469727548770607}
+12/26/2021 12:36:03 - INFO - codeparrot_training - Step 26999: {'lr': 0.00021994166494210146, 'samples': 13824000, 'steps': 26999, 'batch_loss/train': 1.1291398098692298}
+12/26/2021 12:36:13 - INFO - codeparrot_training - Step 27000: {'lr': 0.00021992601621877188, 'samples': 13824512, 'steps': 27000, 'batch_loss/train': 0.6775836439337581}
+12/26/2021 12:36:26 - INFO - codeparrot_training - Step 27001: {'lr': 0.0002199103676150051, 'samples': 13825024, 'steps': 27001, 'batch_loss/train': 0.7505150595679879}
+12/26/2021 12:36:36 - INFO - codeparrot_training - Step 27002: {'lr': 0.00021989471913086336, 'samples': 13825536, 'steps': 27002, 'batch_loss/train': 0.6063736071810126}
+12/26/2021 12:36:47 - INFO - codeparrot_training - Step 27003: {'lr': 0.0002198790707664089, 'samples': 13826048, 'steps': 27003, 'batch_loss/train': 0.7102157928748056}
+12/26/2021 12:36:59 - INFO - codeparrot_training - Step 27004: {'lr': 0.00021986342252170373, 'samples': 13826560, 'steps': 27004, 'batch_loss/train': 2.4254546919837594}
+12/26/2021 12:37:10 - INFO - codeparrot_training - Step 27005: {'lr': 0.0002198477743968103, 'samples': 13827072, 'steps': 27005, 'batch_loss/train': 0.793127171928063}
+12/26/2021 12:37:20 - INFO - codeparrot_training - Step 27006: {'lr': 0.00021983212639179082, 'samples': 13827584, 'steps': 27006, 'batch_loss/train': 0.7204168578609824}
+12/26/2021 12:37:31 - INFO - codeparrot_training - Step 27007: {'lr': 0.00021981647850670733, 'samples': 13828096, 'steps': 27007, 'batch_loss/train': 0.838599038310349}
+12/26/2021 12:37:44 - INFO - codeparrot_training - Step 27008: {'lr': 0.00021980083074162206, 'samples': 13828608, 'steps': 27008, 'batch_loss/train': 0.8126303539611399}
+12/26/2021 12:37:54 - INFO - codeparrot_training - Step 27009: {'lr': 0.00021978518309659738, 'samples': 13829120, 'steps': 27009, 'batch_loss/train': 0.7605645968578756}
+12/26/2021 12:38:05 - INFO - codeparrot_training - Step 27010: {'lr': 0.00021976953557169537, 'samples': 13829632, 'steps': 27010, 'batch_loss/train': 0.8270937083289027}
+12/26/2021 12:38:17 - INFO - codeparrot_training - Step 27011: {'lr': 0.00021975388816697833, 'samples': 13830144, 'steps': 27011, 'batch_loss/train': 0.8533422630280256}
+12/26/2021 12:38:27 - INFO - codeparrot_training - Step 27012: {'lr': 0.0002197382408825083, 'samples': 13830656, 'steps': 27012, 'batch_loss/train': 0.8064243197441101}
+12/26/2021 12:38:38 - INFO - codeparrot_training - Step 27013: {'lr': 0.0002197225937183477, 'samples': 13831168, 'steps': 27013, 'batch_loss/train': 0.8330345312133431}
+12/26/2021 12:38:51 - INFO - codeparrot_training - Step 27014: {'lr': 0.00021970694667455858, 'samples': 13831680, 'steps': 27014, 'batch_loss/train': 0.8708139136433601}
+12/26/2021 12:39:02 - INFO - codeparrot_training - Step 27015: {'lr': 0.00021969129975120326, 'samples': 13832192, 'steps': 27015, 'batch_loss/train': 0.7529397895559669}
+12/26/2021 12:39:12 - INFO - codeparrot_training - Step 27016: {'lr': 0.0002196756529483438, 'samples': 13832704, 'steps': 27016, 'batch_loss/train': 0.7541760453023016}
+12/26/2021 12:39:23 - INFO - codeparrot_training - Step 27017: {'lr': 0.00021966000626604253, 'samples': 13833216, 'steps': 27017, 'batch_loss/train': 0.7897997591644526}
+12/26/2021 12:39:35 - INFO - codeparrot_training - Step 27018: {'lr': 0.00021964435970436166, 'samples': 13833728, 'steps': 27018, 'batch_loss/train': 0.8469058303162456}
+12/26/2021 12:39:45 - INFO - codeparrot_training - Step 27019: {'lr': 0.00021962871326336325, 'samples': 13834240, 'steps': 27019, 'batch_loss/train': 0.7272483045235276}
+12/26/2021 12:39:56 - INFO - codeparrot_training - Step 27020: {'lr': 0.00021961306694310968, 'samples': 13834752, 'steps': 27020, 'batch_loss/train': 0.7006875933147967}
+12/26/2021 12:40:08 - INFO - codeparrot_training - Step 27021: {'lr': 0.00021959742074366311, 'samples': 13835264, 'steps': 27021, 'batch_loss/train': 0.7315167980268598}
+12/26/2021 12:40:19 - INFO - codeparrot_training - Step 27022: {'lr': 0.00021958177466508574, 'samples': 13835776, 'steps': 27022, 'batch_loss/train': 0.7836270860861987}
+12/26/2021 12:40:29 - INFO - codeparrot_training - Step 27023: {'lr': 0.00021956612870743965, 'samples': 13836288, 'steps': 27023, 'batch_loss/train': 1.781600458547473}
+12/26/2021 12:40:42 - INFO - codeparrot_training - Step 27024: {'lr': 0.00021955048287078716, 'samples': 13836800, 'steps': 27024, 'batch_loss/train': 0.7434368012472987}
+12/26/2021 12:40:52 - INFO - codeparrot_training - Step 27025: {'lr': 0.0002195348371551905, 'samples': 13837312, 'steps': 27025, 'batch_loss/train': 0.8770156526006758}
+12/26/2021 12:41:03 - INFO - codeparrot_training - Step 27026: {'lr': 0.00021951919156071182, 'samples': 13837824, 'steps': 27026, 'batch_loss/train': 0.7966187102720141}
+12/26/2021 12:41:13 - INFO - codeparrot_training - Step 27027: {'lr': 0.00021950354608741323, 'samples': 13838336, 'steps': 27027, 'batch_loss/train': 0.787079163827002}
+12/26/2021 12:41:26 - INFO - codeparrot_training - Step 27028: {'lr': 0.00021948790073535716, 'samples': 13838848, 'steps': 27028, 'batch_loss/train': 0.7692256653681397}
+12/26/2021 12:41:37 - INFO - codeparrot_training - Step 27029: {'lr': 0.0002194722555046056, 'samples': 13839360, 'steps': 27029, 'batch_loss/train': 0.8831538306549191}
+12/26/2021 12:41:47 - INFO - codeparrot_training - Step 27030: {'lr': 0.0002194566103952208, 'samples': 13839872, 'steps': 27030, 'batch_loss/train': 0.8083289889618754}
+12/26/2021 12:42:00 - INFO - codeparrot_training - Step 27031: {'lr': 0.00021944096540726506, 'samples': 13840384, 'steps': 27031, 'batch_loss/train': 0.562446370255202}
+12/26/2021 12:42:10 - INFO - codeparrot_training - Step 27032: {'lr': 0.00021942532054080053, 'samples': 13840896, 'steps': 27032, 'batch_loss/train': 0.6989931841380894}
+12/26/2021 12:42:21 - INFO - codeparrot_training - Step 27033: {'lr': 0.0002194096757958893, 'samples': 13841408, 'steps': 27033, 'batch_loss/train': 0.8117665471509099}
+12/26/2021 12:42:33 - INFO - codeparrot_training - Step 27034: {'lr': 0.0002193940311725936, 'samples': 13841920, 'steps': 27034, 'batch_loss/train': 0.7164223592262715}
+12/26/2021 12:42:44 - INFO - codeparrot_training - Step 27035: {'lr': 0.0002193783866709758, 'samples': 13842432, 'steps': 27035, 'batch_loss/train': 1.6427463321015239}
+12/26/2021 12:42:55 - INFO - codeparrot_training - Step 27036: {'lr': 0.00021936274229109794, 'samples': 13842944, 'steps': 27036, 'batch_loss/train': 0.8429020373150706}
+12/26/2021 12:43:05 - INFO - codeparrot_training - Step 27037: {'lr': 0.00021934709803302224, 'samples': 13843456, 'steps': 27037, 'batch_loss/train': 0.9469894208014011}
+12/26/2021 12:43:18 - INFO - codeparrot_training - Step 27038: {'lr': 0.00021933145389681083, 'samples': 13843968, 'steps': 27038, 'batch_loss/train': 0.8318827801849693}
+12/26/2021 12:43:29 - INFO - codeparrot_training - Step 27039: {'lr': 0.0002193158098825261, 'samples': 13844480, 'steps': 27039, 'batch_loss/train': 0.7812955733388662}
+12/26/2021 12:43:39 - INFO - codeparrot_training - Step 27040: {'lr': 0.00021930016599023007, 'samples': 13844992, 'steps': 27040, 'batch_loss/train': 0.7668367670848966}
+12/26/2021 12:43:51 - INFO - codeparrot_training - Step 27041: {'lr': 0.0002192845222199849, 'samples': 13845504, 'steps': 27041, 'batch_loss/train': 0.7606734749861062}
+12/26/2021 12:44:02 - INFO - codeparrot_training - Step 27042: {'lr': 0.00021926887857185302, 'samples': 13846016, 'steps': 27042, 'batch_loss/train': 0.7399984924122691}
+12/26/2021 12:44:13 - INFO - codeparrot_training - Step 27043: {'lr': 0.00021925323504589638, 'samples': 13846528, 'steps': 27043, 'batch_loss/train': 0.6473904829472303}
+12/26/2021 12:44:25 - INFO - codeparrot_training - Step 27044: {'lr': 0.00021923759164217736, 'samples': 13847040, 'steps': 27044, 'batch_loss/train': 0.8271099627017975}
+12/26/2021 12:44:36 - INFO - codeparrot_training - Step 27045: {'lr': 0.00021922194836075792, 'samples': 13847552, 'steps': 27045, 'batch_loss/train': 0.7703213612549007}
+12/26/2021 12:44:47 - INFO - codeparrot_training - Step 27046: {'lr': 0.00021920630520170045, 'samples': 13848064, 'steps': 27046, 'batch_loss/train': 0.779523809440434}
+12/26/2021 12:44:59 - INFO - codeparrot_training - Step 27047: {'lr': 0.00021919066216506716, 'samples': 13848576, 'steps': 27047, 'batch_loss/train': 0.7274111602455378}
+12/26/2021 12:45:09 - INFO - codeparrot_training - Step 27048: {'lr': 0.00021917501925092014, 'samples': 13849088, 'steps': 27048, 'batch_loss/train': 0.7483838414773345}
+12/26/2021 12:45:20 - INFO - codeparrot_training - Step 27049: {'lr': 0.00021915937645932153, 'samples': 13849600, 'steps': 27049, 'batch_loss/train': 0.7487923111766577}
+12/26/2021 12:45:30 - INFO - codeparrot_training - Step 27050: {'lr': 0.0002191437337903336, 'samples': 13850112, 'steps': 27050, 'batch_loss/train': 0.8818279793485999}
+12/26/2021 12:45:43 - INFO - codeparrot_training - Step 27051: {'lr': 0.00021912809124401863, 'samples': 13850624, 'steps': 27051, 'batch_loss/train': 0.8660522634163499}
+12/26/2021 12:45:54 - INFO - codeparrot_training - Step 27052: {'lr': 0.00021911244882043864, 'samples': 13851136, 'steps': 27052, 'batch_loss/train': 0.8110558970365673}
+12/26/2021 12:46:04 - INFO - codeparrot_training - Step 27053: {'lr': 0.00021909680651965586, 'samples': 13851648, 'steps': 27053, 'batch_loss/train': 1.012129764072597}
+12/26/2021 12:46:18 - INFO - codeparrot_training - Step 27054: {'lr': 0.0002190811643417326, 'samples': 13852160, 'steps': 27054, 'batch_loss/train': 0.7084256396628916}
+12/26/2021 12:46:28 - INFO - codeparrot_training - Step 27055: {'lr': 0.00021906552228673088, 'samples': 13852672, 'steps': 27055, 'batch_loss/train': 1.434898822568357}
+12/26/2021 12:46:39 - INFO - codeparrot_training - Step 27056: {'lr': 0.00021904988035471293, 'samples': 13853184, 'steps': 27056, 'batch_loss/train': 1.0496738450601697}
+12/26/2021 12:46:50 - INFO - codeparrot_training - Step 27057: {'lr': 0.00021903423854574108, 'samples': 13853696, 'steps': 27057, 'batch_loss/train': 1.036179605871439}
+12/26/2021 12:47:02 - INFO - codeparrot_training - Step 27058: {'lr': 0.0002190185968598774, 'samples': 13854208, 'steps': 27058, 'batch_loss/train': 0.8004870277363807}
+12/26/2021 12:47:12 - INFO - codeparrot_training - Step 27059: {'lr': 0.00021900295529718404, 'samples': 13854720, 'steps': 27059, 'batch_loss/train': 0.588340733665973}
+12/26/2021 12:47:23 - INFO - codeparrot_training - Step 27060: {'lr': 0.00021898731385772315, 'samples': 13855232, 'steps': 27060, 'batch_loss/train': 0.7730168383568525}
+12/26/2021 12:47:35 - INFO - codeparrot_training - Step 27061: {'lr': 0.00021897167254155713, 'samples': 13855744, 'steps': 27061, 'batch_loss/train': 0.809382401406765}
+12/26/2021 12:47:46 - INFO - codeparrot_training - Step 27062: {'lr': 0.00021895603134874792, 'samples': 13856256, 'steps': 27062, 'batch_loss/train': 0.7451087338849902}
+12/26/2021 12:47:56 - INFO - codeparrot_training - Step 27063: {'lr': 0.0002189403902793579, 'samples': 13856768, 'steps': 27063, 'batch_loss/train': 0.7126258290372789}
+12/26/2021 12:48:07 - INFO - codeparrot_training - Step 27064: {'lr': 0.00021892474933344903, 'samples': 13857280, 'steps': 27064, 'batch_loss/train': 0.7846895977854729}
+12/26/2021 12:48:19 - INFO - codeparrot_training - Step 27065: {'lr': 0.0002189091085110837, 'samples': 13857792, 'steps': 27065, 'batch_loss/train': 0.7160997511819005}
+12/26/2021 12:48:30 - INFO - codeparrot_training - Step 27066: {'lr': 0.00021889346781232404, 'samples': 13858304, 'steps': 27066, 'batch_loss/train': 0.7794391997158527}
+12/26/2021 12:48:40 - INFO - codeparrot_training - Step 27067: {'lr': 0.00021887782723723214, 'samples': 13858816, 'steps': 27067, 'batch_loss/train': 0.8421553252264857}
+12/26/2021 12:48:53 - INFO - codeparrot_training - Step 27068: {'lr': 0.00021886218678587033, 'samples': 13859328, 'steps': 27068, 'batch_loss/train': 0.7937546847388148}
+12/26/2021 12:49:04 - INFO - codeparrot_training - Step 27069: {'lr': 0.00021884654645830065, 'samples': 13859840, 'steps': 27069, 'batch_loss/train': 0.745559782255441}
+12/26/2021 12:49:14 - INFO - codeparrot_training - Step 27070: {'lr': 0.00021883090625458542, 'samples': 13860352, 'steps': 27070, 'batch_loss/train': 0.781346004921943}
+12/26/2021 12:49:27 - INFO - codeparrot_training - Step 27071: {'lr': 0.00021881526617478658, 'samples': 13860864, 'steps': 27071, 'batch_loss/train': 0.8135878723114729}
+12/26/2021 12:49:37 - INFO - codeparrot_training - Step 27072: {'lr': 0.00021879962621896654, 'samples': 13861376, 'steps': 27072, 'batch_loss/train': 0.7833338738419116}
+12/26/2021 12:49:48 - INFO - codeparrot_training - Step 27073: {'lr': 0.0002187839863871875, 'samples': 13861888, 'steps': 27073, 'batch_loss/train': 0.7291790745221078}
+12/26/2021 12:50:01 - INFO - codeparrot_training - Step 27074: {'lr': 0.00021876834667951145, 'samples': 13862400, 'steps': 27074, 'batch_loss/train': 0.8123081708326936}
+12/26/2021 12:50:11 - INFO - codeparrot_training - Step 27075: {'lr': 0.0002187527070960006, 'samples': 13862912, 'steps': 27075, 'batch_loss/train': 0.7491811811923981}
+12/26/2021 12:50:22 - INFO - codeparrot_training - Step 27076: {'lr': 0.0002187370676367173, 'samples': 13863424, 'steps': 27076, 'batch_loss/train': 0.781294490210712}
+12/26/2021 12:50:33 - INFO - codeparrot_training - Step 27077: {'lr': 0.00021872142830172366, 'samples': 13863936, 'steps': 27077, 'batch_loss/train': 0.793626700527966}
+12/26/2021 12:50:45 - INFO - codeparrot_training - Step 27078: {'lr': 0.00021870578909108165, 'samples': 13864448, 'steps': 27078, 'batch_loss/train': 0.6436593462713063}
+12/26/2021 12:50:55 - INFO - codeparrot_training - Step 27079: {'lr': 0.00021869015000485368, 'samples': 13864960, 'steps': 27079, 'batch_loss/train': 0.7208319630008191}
+12/26/2021 12:51:06 - INFO - codeparrot_training - Step 27080: {'lr': 0.0002186745110431019, 'samples': 13865472, 'steps': 27080, 'batch_loss/train': 0.7263997981790453}
+12/26/2021 12:51:18 - INFO - codeparrot_training - Step 27081: {'lr': 0.0002186588722058884, 'samples': 13865984, 'steps': 27081, 'batch_loss/train': 0.9427215843461454}
+12/26/2021 12:51:29 - INFO - codeparrot_training - Step 27082: {'lr': 0.0002186432334932753, 'samples': 13866496, 'steps': 27082, 'batch_loss/train': 0.8414271725341678}
+12/26/2021 12:51:39 - INFO - codeparrot_training - Step 27083: {'lr': 0.00021862759490532503, 'samples': 13867008, 'steps': 27083, 'batch_loss/train': 0.8108413126319647}
+12/26/2021 12:51:52 - INFO - codeparrot_training - Step 27084: {'lr': 0.00021861195644209952, 'samples': 13867520, 'steps': 27084, 'batch_loss/train': 0.7967210188508034}
+12/26/2021 12:52:03 - INFO - codeparrot_training - Step 27085: {'lr': 0.00021859631810366105, 'samples': 13868032, 'steps': 27085, 'batch_loss/train': 0.7856118390336633}
+12/26/2021 12:52:14 - INFO - codeparrot_training - Step 27086: {'lr': 0.00021858067989007164, 'samples': 13868544, 'steps': 27086, 'batch_loss/train': 0.7972414479590952}
+12/26/2021 12:52:24 - INFO - codeparrot_training - Step 27087: {'lr': 0.00021856504180139372, 'samples': 13869056, 'steps': 27087, 'batch_loss/train': 0.8154855170287192}
+12/26/2021 12:52:36 - INFO - codeparrot_training - Step 27088: {'lr': 0.00021854940383768927, 'samples': 13869568, 'steps': 27088, 'batch_loss/train': 0.8462633369490504}
+12/26/2021 12:52:47 - INFO - codeparrot_training - Step 27089: {'lr': 0.00021853376599902056, 'samples': 13870080, 'steps': 27089, 'batch_loss/train': 0.7942904438823462}
+12/26/2021 12:52:58 - INFO - codeparrot_training - Step 27090: {'lr': 0.0002185181282854496, 'samples': 13870592, 'steps': 27090, 'batch_loss/train': 0.7880741157568991}
+12/26/2021 12:53:10 - INFO - codeparrot_training - Step 27091: {'lr': 0.00021850249069703875, 'samples': 13871104, 'steps': 27091, 'batch_loss/train': 0.7183612436638214}
+12/26/2021 12:53:21 - INFO - codeparrot_training - Step 27092: {'lr': 0.00021848685323385016, 'samples': 13871616, 'steps': 27092, 'batch_loss/train': 0.8345005924347788}
+12/26/2021 12:53:32 - INFO - codeparrot_training - Step 27093: {'lr': 0.00021847121589594582, 'samples': 13872128, 'steps': 27093, 'batch_loss/train': 0.8004102967679501}
+12/26/2021 12:53:44 - INFO - codeparrot_training - Step 27094: {'lr': 0.00021845557868338812, 'samples': 13872640, 'steps': 27094, 'batch_loss/train': 0.7121539828367531}
+12/26/2021 12:53:54 - INFO - codeparrot_training - Step 27095: {'lr': 0.0002184399415962391, 'samples': 13873152, 'steps': 27095, 'batch_loss/train': 0.7235813681036234}
+12/26/2021 12:54:05 - INFO - codeparrot_training - Step 27096: {'lr': 0.00021842430463456098, 'samples': 13873664, 'steps': 27096, 'batch_loss/train': 0.7366345808841288}
+12/26/2021 12:54:16 - INFO - codeparrot_training - Step 27097: {'lr': 0.0002184086677984158, 'samples': 13874176, 'steps': 27097, 'batch_loss/train': 0.8298656931146979}
+12/26/2021 12:54:28 - INFO - codeparrot_training - Step 27098: {'lr': 0.0002183930310878659, 'samples': 13874688, 'steps': 27098, 'batch_loss/train': 0.7614655075594783}
+12/26/2021 12:54:38 - INFO - codeparrot_training - Step 27099: {'lr': 0.0002183773945029734, 'samples': 13875200, 'steps': 27099, 'batch_loss/train': 1.5876053732354194}
+12/26/2021 12:54:49 - INFO - codeparrot_training - Step 27100: {'lr': 0.00021836175804380042, 'samples': 13875712, 'steps': 27100, 'batch_loss/train': 0.7100616097450256}
+12/26/2021 12:55:01 - INFO - codeparrot_training - Step 27101: {'lr': 0.0002183461217104091, 'samples': 13876224, 'steps': 27101, 'batch_loss/train': 0.6580079422565177}
+12/26/2021 12:55:12 - INFO - codeparrot_training - Step 27102: {'lr': 0.00021833048550286171, 'samples': 13876736, 'steps': 27102, 'batch_loss/train': 0.7946018704678863}
+12/26/2021 12:55:22 - INFO - codeparrot_training - Step 27103: {'lr': 0.00021831484942122037, 'samples': 13877248, 'steps': 27103, 'batch_loss/train': 0.7266000350937247}
+12/26/2021 12:55:35 - INFO - codeparrot_training - Step 27104: {'lr': 0.00021829921346554716, 'samples': 13877760, 'steps': 27104, 'batch_loss/train': 0.8187220357358456}
+12/26/2021 12:55:46 - INFO - codeparrot_training - Step 27105: {'lr': 0.00021828357763590427, 'samples': 13878272, 'steps': 27105, 'batch_loss/train': 0.8175321104936302}
+12/26/2021 12:55:56 - INFO - codeparrot_training - Step 27106: {'lr': 0.00021826794193235404, 'samples': 13878784, 'steps': 27106, 'batch_loss/train': 0.8699501026421785}
+12/26/2021 12:56:07 - INFO - codeparrot_training - Step 27107: {'lr': 0.0002182523063549584, 'samples': 13879296, 'steps': 27107, 'batch_loss/train': 0.6648680404759943}
+12/26/2021 12:56:19 - INFO - codeparrot_training - Step 27108: {'lr': 0.00021823667090377954, 'samples': 13879808, 'steps': 27108, 'batch_loss/train': 0.8258742885664105}
+12/26/2021 12:56:30 - INFO - codeparrot_training - Step 27109: {'lr': 0.00021822103557887982, 'samples': 13880320, 'steps': 27109, 'batch_loss/train': 0.7412862260825932}
+12/26/2021 12:56:40 - INFO - codeparrot_training - Step 27110: {'lr': 0.0002182054003803212, 'samples': 13880832, 'steps': 27110, 'batch_loss/train': 0.9426129348576069}
+12/26/2021 12:56:52 - INFO - codeparrot_training - Step 27111: {'lr': 0.00021818976530816591, 'samples': 13881344, 'steps': 27111, 'batch_loss/train': 0.9827627884224057}
+12/26/2021 12:57:03 - INFO - codeparrot_training - Step 27112: {'lr': 0.00021817413036247606, 'samples': 13881856, 'steps': 27112, 'batch_loss/train': 0.811134030809626}
+12/26/2021 12:57:14 - INFO - codeparrot_training - Step 27113: {'lr': 0.00021815849554331394, 'samples': 13882368, 'steps': 27113, 'batch_loss/train': 0.7723307642154396}
+12/26/2021 12:57:26 - INFO - codeparrot_training - Step 27114: {'lr': 0.00021814286085074155, 'samples': 13882880, 'steps': 27114, 'batch_loss/train': 0.7148374242242426}
+12/26/2021 12:57:37 - INFO - codeparrot_training - Step 27115: {'lr': 0.00021812722628482107, 'samples': 13883392, 'steps': 27115, 'batch_loss/train': 0.6243322346126661}
+12/26/2021 12:57:48 - INFO - codeparrot_training - Step 27116: {'lr': 0.0002181115918456148, 'samples': 13883904, 'steps': 27116, 'batch_loss/train': 0.8197688583750278}
+12/26/2021 12:58:00 - INFO - codeparrot_training - Step 27117: {'lr': 0.00021809595753318476, 'samples': 13884416, 'steps': 27117, 'batch_loss/train': 0.7262420190963894}
+12/26/2021 12:58:10 - INFO - codeparrot_training - Step 27118: {'lr': 0.0002180803233475932, 'samples': 13884928, 'steps': 27118, 'batch_loss/train': 0.821019237395376}
+12/26/2021 12:58:21 - INFO - codeparrot_training - Step 27119: {'lr': 0.0002180646892889021, 'samples': 13885440, 'steps': 27119, 'batch_loss/train': 0.7356160450726748}
+12/26/2021 12:58:31 - INFO - codeparrot_training - Step 27120: {'lr': 0.0002180490553571738, 'samples': 13885952, 'steps': 27120, 'batch_loss/train': 0.7506029678042978}
+12/26/2021 12:58:44 - INFO - codeparrot_training - Step 27121: {'lr': 0.00021803342155247047, 'samples': 13886464, 'steps': 27121, 'batch_loss/train': 0.7654706547036767}
+12/26/2021 12:58:55 - INFO - codeparrot_training - Step 27122: {'lr': 0.00021801778787485415, 'samples': 13886976, 'steps': 27122, 'batch_loss/train': 0.7212541177868843}
+12/26/2021 12:59:06 - INFO - codeparrot_training - Step 27123: {'lr': 0.00021800215432438695, 'samples': 13887488, 'steps': 27123, 'batch_loss/train': 0.7998914867639542}
+12/26/2021 12:59:18 - INFO - codeparrot_training - Step 27124: {'lr': 0.00021798652090113113, 'samples': 13888000, 'steps': 27124, 'batch_loss/train': 0.840723967179656}
+12/26/2021 12:59:28 - INFO - codeparrot_training - Step 27125: {'lr': 0.0002179708876051489, 'samples': 13888512, 'steps': 27125, 'batch_loss/train': 0.7694286247715354}
+12/26/2021 12:59:39 - INFO - codeparrot_training - Step 27126: {'lr': 0.00021795525443650223, 'samples': 13889024, 'steps': 27126, 'batch_loss/train': 0.6716906378278509}
+12/26/2021 12:59:51 - INFO - codeparrot_training - Step 27127: {'lr': 0.00021793962139525337, 'samples': 13889536, 'steps': 27127, 'batch_loss/train': 0.8311259895563126}
+12/26/2021 13:00:02 - INFO - codeparrot_training - Step 27128: {'lr': 0.00021792398848146454, 'samples': 13890048, 'steps': 27128, 'batch_loss/train': 0.7572008902207017}
+12/26/2021 13:00:13 - INFO - codeparrot_training - Step 27129: {'lr': 0.00021790835569519783, 'samples': 13890560, 'steps': 27129, 'batch_loss/train': 0.5487365419976413}
+12/26/2021 13:00:23 - INFO - codeparrot_training - Step 27130: {'lr': 0.00021789272303651526, 'samples': 13891072, 'steps': 27130, 'batch_loss/train': 0.7727763800648972}
+12/26/2021 13:00:36 - INFO - codeparrot_training - Step 27131: {'lr': 0.00021787709050547917, 'samples': 13891584, 'steps': 27131, 'batch_loss/train': 0.722957078833133}
+12/26/2021 13:00:47 - INFO - codeparrot_training - Step 27132: {'lr': 0.0002178614581021517, 'samples': 13892096, 'steps': 27132, 'batch_loss/train': 0.7817056505009532}
+12/26/2021 13:00:57 - INFO - codeparrot_training - Step 27133: {'lr': 0.00021784582582659484, 'samples': 13892608, 'steps': 27133, 'batch_loss/train': 0.7707389332354069}
+12/26/2021 13:01:10 - INFO - codeparrot_training - Step 27134: {'lr': 0.0002178301936788708, 'samples': 13893120, 'steps': 27134, 'batch_loss/train': 0.7625471208011732}
+12/26/2021 13:01:20 - INFO - codeparrot_training - Step 27135: {'lr': 0.0002178145616590419, 'samples': 13893632, 'steps': 27135, 'batch_loss/train': 0.8191851545125246}
+12/26/2021 13:01:31 - INFO - codeparrot_training - Step 27136: {'lr': 0.00021779892976717008, 'samples': 13894144, 'steps': 27136, 'batch_loss/train': 0.7463194709271193}
+12/26/2021 13:01:43 - INFO - codeparrot_training - Step 27137: {'lr': 0.00021778329800331762, 'samples': 13894656, 'steps': 27137, 'batch_loss/train': 0.7906886297278106}
+12/26/2021 13:01:54 - INFO - codeparrot_training - Step 27138: {'lr': 0.00021776766636754646, 'samples': 13895168, 'steps': 27138, 'batch_loss/train': 0.8154989648610353}
+12/26/2021 13:02:04 - INFO - codeparrot_training - Step 27139: {'lr': 0.000217752034859919, 'samples': 13895680, 'steps': 27139, 'batch_loss/train': 0.8336400042753667}
+12/26/2021 13:02:15 - INFO - codeparrot_training - Step 27140: {'lr': 0.00021773640348049728, 'samples': 13896192, 'steps': 27140, 'batch_loss/train': 0.7406328469514847}
+12/26/2021 13:02:27 - INFO - codeparrot_training - Step 27141: {'lr': 0.00021772077222934335, 'samples': 13896704, 'steps': 27141, 'batch_loss/train': 0.5720992251299322}
+12/26/2021 13:02:37 - INFO - codeparrot_training - Step 27142: {'lr': 0.0002177051411065195, 'samples': 13897216, 'steps': 27142, 'batch_loss/train': 0.7950016250833869}
+12/26/2021 13:02:48 - INFO - codeparrot_training - Step 27143: {'lr': 0.00021768951011208782, 'samples': 13897728, 'steps': 27143, 'batch_loss/train': 0.7916628890670836}
+12/26/2021 13:03:01 - INFO - codeparrot_training - Step 27144: {'lr': 0.00021767387924611048, 'samples': 13898240, 'steps': 27144, 'batch_loss/train': 0.67946937517263}
+12/26/2021 13:03:12 - INFO - codeparrot_training - Step 27145: {'lr': 0.0002176582485086495, 'samples': 13898752, 'steps': 27145, 'batch_loss/train': 0.7375022559426725}
+12/26/2021 13:03:22 - INFO - codeparrot_training - Step 27146: {'lr': 0.00021764261789976715, 'samples': 13899264, 'steps': 27146, 'batch_loss/train': 0.8013742389157414}
+12/26/2021 13:03:34 - INFO - codeparrot_training - Step 27147: {'lr': 0.00021762698741952557, 'samples': 13899776, 'steps': 27147, 'batch_loss/train': 0.7948544169776142}
+12/26/2021 13:03:45 - INFO - codeparrot_training - Step 27148: {'lr': 0.0002176113570679869, 'samples': 13900288, 'steps': 27148, 'batch_loss/train': 0.7543308027088642}
+12/26/2021 13:03:55 - INFO - codeparrot_training - Step 27149: {'lr': 0.00021759572684521312, 'samples': 13900800, 'steps': 27149, 'batch_loss/train': 0.7966751968488097}
+12/26/2021 13:04:08 - INFO - codeparrot_training - Step 27150: {'lr': 0.00021758009675126654, 'samples': 13901312, 'steps': 27150, 'batch_loss/train': 0.682664658408612}
+12/26/2021 13:04:18 - INFO - codeparrot_training - Step 27151: {'lr': 0.00021756446678620933, 'samples': 13901824, 'steps': 27151, 'batch_loss/train': 0.7275812588632107}
+12/26/2021 13:04:29 - INFO - codeparrot_training - Step 27152: {'lr': 0.00021754883695010343, 'samples': 13902336, 'steps': 27152, 'batch_loss/train': 0.7778588319197297}
+12/26/2021 13:04:40 - INFO - codeparrot_training - Step 27153: {'lr': 0.00021753320724301117, 'samples': 13902848, 'steps': 27153, 'batch_loss/train': 0.8127177143469453}
+12/26/2021 13:04:52 - INFO - codeparrot_training - Step 27154: {'lr': 0.00021751757766499467, 'samples': 13903360, 'steps': 27154, 'batch_loss/train': 0.6921047195792198}
+12/26/2021 13:05:03 - INFO - codeparrot_training - Step 27155: {'lr': 0.00021750194821611594, 'samples': 13903872, 'steps': 27155, 'batch_loss/train': 0.8137591991107911}
+12/26/2021 13:05:14 - INFO - codeparrot_training - Step 27156: {'lr': 0.00021748631889643717, 'samples': 13904384, 'steps': 27156, 'batch_loss/train': 0.5954364044591784}
+12/26/2021 13:05:26 - INFO - codeparrot_training - Step 27157: {'lr': 0.0002174706897060206, 'samples': 13904896, 'steps': 27157, 'batch_loss/train': 0.7736724596470594}
+12/26/2021 13:05:36 - INFO - codeparrot_training - Step 27158: {'lr': 0.00021745506064492829, 'samples': 13905408, 'steps': 27158, 'batch_loss/train': 0.8878686367534101}
+12/26/2021 13:05:47 - INFO - codeparrot_training - Step 27159: {'lr': 0.00021743943171322234, 'samples': 13905920, 'steps': 27159, 'batch_loss/train': 0.729830470867455}
+12/26/2021 13:05:59 - INFO - codeparrot_training - Step 27160: {'lr': 0.0002174238029109648, 'samples': 13906432, 'steps': 27160, 'batch_loss/train': 0.7685317504219711}
+12/26/2021 13:06:10 - INFO - codeparrot_training - Step 27161: {'lr': 0.0002174081742382181, 'samples': 13906944, 'steps': 27161, 'batch_loss/train': 0.7988546201959252}
+12/26/2021 13:06:21 - INFO - codeparrot_training - Step 27162: {'lr': 0.0002173925456950441, 'samples': 13907456, 'steps': 27162, 'batch_loss/train': 0.7338656370993704}
+12/26/2021 13:06:31 - INFO - codeparrot_training - Step 27163: {'lr': 0.0002173769172815051, 'samples': 13907968, 'steps': 27163, 'batch_loss/train': 0.8197123003192246}
+12/26/2021 13:06:44 - INFO - codeparrot_training - Step 27164: {'lr': 0.00021736128899766302, 'samples': 13908480, 'steps': 27164, 'batch_loss/train': 1.5042116409167647}
+12/26/2021 13:06:54 - INFO - codeparrot_training - Step 27165: {'lr': 0.00021734566084358022, 'samples': 13908992, 'steps': 27165, 'batch_loss/train': 0.5574335539713502}
+12/26/2021 13:07:05 - INFO - codeparrot_training - Step 27166: {'lr': 0.0002173300328193188, 'samples': 13909504, 'steps': 27166, 'batch_loss/train': 0.6928272880613804}
+12/26/2021 13:07:17 - INFO - codeparrot_training - Step 27167: {'lr': 0.0002173144049249407, 'samples': 13910016, 'steps': 27167, 'batch_loss/train': 0.7365514598786831}
+12/26/2021 13:07:27 - INFO - codeparrot_training - Step 27168: {'lr': 0.00021729877716050828, 'samples': 13910528, 'steps': 27168, 'batch_loss/train': 0.7901913430541754}
+12/26/2021 13:07:38 - INFO - codeparrot_training - Step 27169: {'lr': 0.00021728314952608354, 'samples': 13911040, 'steps': 27169, 'batch_loss/train': 0.7468491735635325}
+12/26/2021 13:07:51 - INFO - codeparrot_training - Step 27170: {'lr': 0.00021726752202172873, 'samples': 13911552, 'steps': 27170, 'batch_loss/train': 0.8004070268943906}
+12/26/2021 13:08:01 - INFO - codeparrot_training - Step 27171: {'lr': 0.00021725189464750572, 'samples': 13912064, 'steps': 27171, 'batch_loss/train': 0.770708717405796}
+12/26/2021 13:08:12 - INFO - codeparrot_training - Step 27172: {'lr': 0.0002172362674034769, 'samples': 13912576, 'steps': 27172, 'batch_loss/train': 0.5760654994519427}
+12/26/2021 13:08:23 - INFO - codeparrot_training - Step 27173: {'lr': 0.0002172206402897044, 'samples': 13913088, 'steps': 27173, 'batch_loss/train': 0.8182404469698668}
+12/26/2021 13:08:35 - INFO - codeparrot_training - Step 27174: {'lr': 0.0002172050133062502, 'samples': 13913600, 'steps': 27174, 'batch_loss/train': 0.7858980847522616}
+12/26/2021 13:08:46 - INFO - codeparrot_training - Step 27175: {'lr': 0.0002171893864531764, 'samples': 13914112, 'steps': 27175, 'batch_loss/train': 0.6581841219449416}
+12/26/2021 13:08:56 - INFO - codeparrot_training - Step 27176: {'lr': 0.00021717375973054527, 'samples': 13914624, 'steps': 27176, 'batch_loss/train': 0.8197712432593107}
+12/26/2021 13:09:08 - INFO - codeparrot_training - Step 27177: {'lr': 0.00021715813313841892, 'samples': 13915136, 'steps': 27177, 'batch_loss/train': 0.719864524435252}
+12/26/2021 13:09:19 - INFO - codeparrot_training - Step 27178: {'lr': 0.0002171425066768593, 'samples': 13915648, 'steps': 27178, 'batch_loss/train': 0.5885572559782304}
+12/26/2021 13:09:30 - INFO - codeparrot_training - Step 27179: {'lr': 0.00021712688034592875, 'samples': 13916160, 'steps': 27179, 'batch_loss/train': 0.8334966888651252}
+12/26/2021 13:09:42 - INFO - codeparrot_training - Step 27180: {'lr': 0.00021711125414568939, 'samples': 13916672, 'steps': 27180, 'batch_loss/train': 0.7438372685573995}
+12/26/2021 13:09:52 - INFO - codeparrot_training - Step 27181: {'lr': 0.00021709562807620318, 'samples': 13917184, 'steps': 27181, 'batch_loss/train': 0.7454559940379113}
+12/26/2021 13:10:03 - INFO - codeparrot_training - Step 27182: {'lr': 0.00021708000213753226, 'samples': 13917696, 'steps': 27182, 'batch_loss/train': 0.825958036351949}
+12/26/2021 13:10:16 - INFO - codeparrot_training - Step 27183: {'lr': 0.00021706437632973896, 'samples': 13918208, 'steps': 27183, 'batch_loss/train': 0.7852405402809381}
+12/26/2021 13:10:26 - INFO - codeparrot_training - Step 27184: {'lr': 0.00021704875065288527, 'samples': 13918720, 'steps': 27184, 'batch_loss/train': 0.7660761141451076}
+12/26/2021 13:10:37 - INFO - codeparrot_training - Step 27185: {'lr': 0.0002170331251070332, 'samples': 13919232, 'steps': 27185, 'batch_loss/train': 0.9318880988284945}
+12/26/2021 13:10:47 - INFO - codeparrot_training - Step 27186: {'lr': 0.00021701749969224495, 'samples': 13919744, 'steps': 27186, 'batch_loss/train': 0.7673905971460044}
+12/26/2021 13:10:59 - INFO - codeparrot_training - Step 27187: {'lr': 0.0002170018744085828, 'samples': 13920256, 'steps': 27187, 'batch_loss/train': 0.8371842717751861}
+12/26/2021 13:11:10 - INFO - codeparrot_training - Step 27188: {'lr': 0.00021698624925610866, 'samples': 13920768, 'steps': 27188, 'batch_loss/train': 0.8226602403447032}
+12/26/2021 13:11:21 - INFO - codeparrot_training - Step 27189: {'lr': 0.00021697062423488468, 'samples': 13921280, 'steps': 27189, 'batch_loss/train': 0.8650006866082549}
+12/26/2021 13:11:33 - INFO - codeparrot_training - Step 27190: {'lr': 0.00021695499934497314, 'samples': 13921792, 'steps': 27190, 'batch_loss/train': 0.7525924732908607}
+12/26/2021 13:11:44 - INFO - codeparrot_training - Step 27191: {'lr': 0.00021693937458643598, 'samples': 13922304, 'steps': 27191, 'batch_loss/train': 0.7396597884362563}
+12/26/2021 13:11:55 - INFO - codeparrot_training - Step 27192: {'lr': 0.00021692374995933546, 'samples': 13922816, 'steps': 27192, 'batch_loss/train': 0.7689241981133819}
+12/26/2021 13:12:07 - INFO - codeparrot_training - Step 27193: {'lr': 0.00021690812546373348, 'samples': 13923328, 'steps': 27193, 'batch_loss/train': 0.6986834856215864}
+12/26/2021 13:12:17 - INFO - codeparrot_training - Step 27194: {'lr': 0.00021689250109969242, 'samples': 13923840, 'steps': 27194, 'batch_loss/train': 0.7162663314957172}
+12/26/2021 13:12:28 - INFO - codeparrot_training - Step 27195: {'lr': 0.00021687687686727423, 'samples': 13924352, 'steps': 27195, 'batch_loss/train': 0.7787895798683167}
+12/26/2021 13:12:40 - INFO - codeparrot_training - Step 27196: {'lr': 0.00021686125276654112, 'samples': 13924864, 'steps': 27196, 'batch_loss/train': 0.7925372673198581}
+12/26/2021 13:12:51 - INFO - codeparrot_training - Step 27197: {'lr': 0.00021684562879755503, 'samples': 13925376, 'steps': 27197, 'batch_loss/train': 0.7754449415951967}
+12/26/2021 13:13:01 - INFO - codeparrot_training - Step 27198: {'lr': 0.00021683000496037827, 'samples': 13925888, 'steps': 27198, 'batch_loss/train': 0.6875720005482435}
+12/26/2021 13:13:12 - INFO - codeparrot_training - Step 27199: {'lr': 0.00021681438125507294, 'samples': 13926400, 'steps': 27199, 'batch_loss/train': 0.7437633546069264}
+12/26/2021 13:13:25 - INFO - codeparrot_training - Step 27200: {'lr': 0.00021679875768170105, 'samples': 13926912, 'steps': 27200, 'batch_loss/train': 0.7322270143777132}
+12/26/2021 13:13:35 - INFO - codeparrot_training - Step 27201: {'lr': 0.0002167831342403247, 'samples': 13927424, 'steps': 27201, 'batch_loss/train': 0.7838300270959735}
+12/26/2021 13:13:46 - INFO - codeparrot_training - Step 27202: {'lr': 0.00021676751093100617, 'samples': 13927936, 'steps': 27202, 'batch_loss/train': 0.716859613719862}
+12/26/2021 13:13:58 - INFO - codeparrot_training - Step 27203: {'lr': 0.0002167518877538075, 'samples': 13928448, 'steps': 27203, 'batch_loss/train': 0.7750191250815988}
+12/26/2021 13:14:08 - INFO - codeparrot_training - Step 27204: {'lr': 0.00021673626470879062, 'samples': 13928960, 'steps': 27204, 'batch_loss/train': 0.6689856229932047}
+12/26/2021 13:14:19 - INFO - codeparrot_training - Step 27205: {'lr': 0.00021672064179601786, 'samples': 13929472, 'steps': 27205, 'batch_loss/train': 0.8102872051531449}
+12/26/2021 13:14:31 - INFO - codeparrot_training - Step 27206: {'lr': 0.00021670501901555135, 'samples': 13929984, 'steps': 27206, 'batch_loss/train': 0.7617429653182626}
+12/26/2021 13:14:42 - INFO - codeparrot_training - Step 27207: {'lr': 0.000216689396367453, 'samples': 13930496, 'steps': 27207, 'batch_loss/train': 0.6580883885035291}
+12/26/2021 13:14:52 - INFO - codeparrot_training - Step 27208: {'lr': 0.00021667377385178502, 'samples': 13931008, 'steps': 27208, 'batch_loss/train': 0.5714455780107528}
+12/26/2021 13:15:03 - INFO - codeparrot_training - Step 27209: {'lr': 0.00021665815146860963, 'samples': 13931520, 'steps': 27209, 'batch_loss/train': 0.7166253803297877}
+12/26/2021 13:15:17 - INFO - codeparrot_training - Step 27210: {'lr': 0.00021664252921798884, 'samples': 13932032, 'steps': 27210, 'batch_loss/train': 0.7633572886697948}
+12/26/2021 13:15:27 - INFO - codeparrot_training - Step 27211: {'lr': 0.0002166269070999847, 'samples': 13932544, 'steps': 27211, 'batch_loss/train': 0.7582415184006095}
+12/26/2021 13:15:38 - INFO - codeparrot_training - Step 27212: {'lr': 0.00021661128511465933, 'samples': 13933056, 'steps': 27212, 'batch_loss/train': 1.2208243769127876}
+12/26/2021 13:15:50 - INFO - codeparrot_training - Step 27213: {'lr': 0.00021659566326207496, 'samples': 13933568, 'steps': 27213, 'batch_loss/train': 0.7081458461470902}
+12/26/2021 13:16:01 - INFO - codeparrot_training - Step 27214: {'lr': 0.0002165800415422936, 'samples': 13934080, 'steps': 27214, 'batch_loss/train': 0.8444514228031039}
+12/26/2021 13:16:11 - INFO - codeparrot_training - Step 27215: {'lr': 0.00021656441995537734, 'samples': 13934592, 'steps': 27215, 'batch_loss/train': 0.7960063051432371}
+12/26/2021 13:16:24 - INFO - codeparrot_training - Step 27216: {'lr': 0.00021654879850138838, 'samples': 13935104, 'steps': 27216, 'batch_loss/train': 0.6514921197667718}
+12/26/2021 13:16:35 - INFO - codeparrot_training - Step 27217: {'lr': 0.00021653317718038875, 'samples': 13935616, 'steps': 27217, 'batch_loss/train': 0.7056590393185616}
+12/26/2021 13:16:45 - INFO - codeparrot_training - Step 27218: {'lr': 0.00021651755599244062, 'samples': 13936128, 'steps': 27218, 'batch_loss/train': 0.7420390956103802}
+12/26/2021 13:16:56 - INFO - codeparrot_training - Step 27219: {'lr': 0.00021650193493760592, 'samples': 13936640, 'steps': 27219, 'batch_loss/train': 0.8523889114148915}
+12/26/2021 13:17:08 - INFO - codeparrot_training - Step 27220: {'lr': 0.00021648631401594698, 'samples': 13937152, 'steps': 27220, 'batch_loss/train': 0.6987639035796747}
+12/26/2021 13:17:19 - INFO - codeparrot_training - Step 27221: {'lr': 0.00021647069322752578, 'samples': 13937664, 'steps': 27221, 'batch_loss/train': 0.8319571511819959}
+12/26/2021 13:17:30 - INFO - codeparrot_training - Step 27222: {'lr': 0.0002164550725724045, 'samples': 13938176, 'steps': 27222, 'batch_loss/train': 0.7248290125280619}
+12/26/2021 13:17:42 - INFO - codeparrot_training - Step 27223: {'lr': 0.00021643945205064504, 'samples': 13938688, 'steps': 27223, 'batch_loss/train': 0.7175038680434227}
+12/26/2021 13:17:53 - INFO - codeparrot_training - Step 27224: {'lr': 0.00021642383166230972, 'samples': 13939200, 'steps': 27224, 'batch_loss/train': 0.7992959674447775}
+12/26/2021 13:18:03 - INFO - codeparrot_training - Step 27225: {'lr': 0.00021640821140746062, 'samples': 13939712, 'steps': 27225, 'batch_loss/train': 0.7977613476105034}
+12/26/2021 13:18:16 - INFO - codeparrot_training - Step 27226: {'lr': 0.00021639259128615967, 'samples': 13940224, 'steps': 27226, 'batch_loss/train': 0.5686143399798311}
+12/26/2021 13:18:27 - INFO - codeparrot_training - Step 27227: {'lr': 0.00021637697129846915, 'samples': 13940736, 'steps': 27227, 'batch_loss/train': 0.7828495283611119}
+12/26/2021 13:18:37 - INFO - codeparrot_training - Step 27228: {'lr': 0.00021636135144445116, 'samples': 13941248, 'steps': 27228, 'batch_loss/train': 0.769777289358899}
+12/26/2021 13:18:48 - INFO - codeparrot_training - Step 27229: {'lr': 0.00021634573172416772, 'samples': 13941760, 'steps': 27229, 'batch_loss/train': 0.6886118231341243}
+12/26/2021 13:19:00 - INFO - codeparrot_training - Step 27230: {'lr': 0.00021633011213768083, 'samples': 13942272, 'steps': 27230, 'batch_loss/train': 0.7176682483404875}
+12/26/2021 13:19:11 - INFO - codeparrot_training - Step 27231: {'lr': 0.00021631449268505278, 'samples': 13942784, 'steps': 27231, 'batch_loss/train': 0.6014870899380185}
+12/26/2021 13:19:21 - INFO - codeparrot_training - Step 27232: {'lr': 0.00021629887336634563, 'samples': 13943296, 'steps': 27232, 'batch_loss/train': 0.7714622672647238}
+12/26/2021 13:19:35 - INFO - codeparrot_training - Step 27233: {'lr': 0.00021628325418162136, 'samples': 13943808, 'steps': 27233, 'batch_loss/train': 0.8038583733141422}
+12/26/2021 13:19:45 - INFO - codeparrot_training - Step 27234: {'lr': 0.0002162676351309421, 'samples': 13944320, 'steps': 27234, 'batch_loss/train': 0.8422315088100731}
+12/26/2021 13:19:56 - INFO - codeparrot_training - Step 27235: {'lr': 0.0002162520162143701, 'samples': 13944832, 'steps': 27235, 'batch_loss/train': 0.6787443356588483}
+12/26/2021 13:20:08 - INFO - codeparrot_training - Step 27236: {'lr': 0.0002162363974319673, 'samples': 13945344, 'steps': 27236, 'batch_loss/train': 0.5963813392445445}
+12/26/2021 13:20:18 - INFO - codeparrot_training - Step 27237: {'lr': 0.00021622077878379587, 'samples': 13945856, 'steps': 27237, 'batch_loss/train': 0.6830056535545737}
+12/26/2021 13:20:29 - INFO - codeparrot_training - Step 27238: {'lr': 0.00021620516026991777, 'samples': 13946368, 'steps': 27238, 'batch_loss/train': 0.6639145319350064}
+12/26/2021 13:20:41 - INFO - codeparrot_training - Step 27239: {'lr': 0.0002161895418903953, 'samples': 13946880, 'steps': 27239, 'batch_loss/train': 0.6495601636124775}
+12/26/2021 13:20:52 - INFO - codeparrot_training - Step 27240: {'lr': 0.00021617392364529038, 'samples': 13947392, 'steps': 27240, 'batch_loss/train': 0.6775653185322881}
+12/26/2021 13:21:02 - INFO - codeparrot_training - Step 27241: {'lr': 0.00021615830553466513, 'samples': 13947904, 'steps': 27241, 'batch_loss/train': 0.7794518987648189}
+12/26/2021 13:21:13 - INFO - codeparrot_training - Step 27242: {'lr': 0.00021614268755858177, 'samples': 13948416, 'steps': 27242, 'batch_loss/train': 0.7755796652054414}
+12/26/2021 13:21:25 - INFO - codeparrot_training - Step 27243: {'lr': 0.00021612706971710225, 'samples': 13948928, 'steps': 27243, 'batch_loss/train': 0.7977935504168272}
+12/26/2021 13:21:36 - INFO - codeparrot_training - Step 27244: {'lr': 0.00021611145201028877, 'samples': 13949440, 'steps': 27244, 'batch_loss/train': 0.75275884475559}
+12/26/2021 13:21:46 - INFO - codeparrot_training - Step 27245: {'lr': 0.00021609583443820324, 'samples': 13949952, 'steps': 27245, 'batch_loss/train': 0.6399912384222262}
+12/26/2021 13:21:59 - INFO - codeparrot_training - Step 27246: {'lr': 0.00021608021700090794, 'samples': 13950464, 'steps': 27246, 'batch_loss/train': 0.7015347694978118}
+12/26/2021 13:22:10 - INFO - codeparrot_training - Step 27247: {'lr': 0.00021606459969846496, 'samples': 13950976, 'steps': 27247, 'batch_loss/train': 0.7046430506743491}
+12/26/2021 13:22:20 - INFO - codeparrot_training - Step 27248: {'lr': 0.0002160489825309363, 'samples': 13951488, 'steps': 27248, 'batch_loss/train': 0.6316765104420483}
+12/26/2021 13:22:32 - INFO - codeparrot_training - Step 27249: {'lr': 0.00021603336549838392, 'samples': 13952000, 'steps': 27249, 'batch_loss/train': 0.7750612022355199}
+12/26/2021 13:22:43 - INFO - codeparrot_training - Step 27250: {'lr': 0.00021601774860087014, 'samples': 13952512, 'steps': 27250, 'batch_loss/train': 0.7403640220873058}
+12/26/2021 13:22:54 - INFO - codeparrot_training - Step 27251: {'lr': 0.00021600213183845703, 'samples': 13953024, 'steps': 27251, 'batch_loss/train': 0.8010115143842995}
+12/26/2021 13:23:04 - INFO - codeparrot_training - Step 27252: {'lr': 0.0002159865152112065, 'samples': 13953536, 'steps': 27252, 'batch_loss/train': 0.7791456738486886}
+12/26/2021 13:23:16 - INFO - codeparrot_training - Step 27253: {'lr': 0.00021597089871918076, 'samples': 13954048, 'steps': 27253, 'batch_loss/train': 0.7514928611926734}
+12/26/2021 13:23:27 - INFO - codeparrot_training - Step 27254: {'lr': 0.00021595528236244196, 'samples': 13954560, 'steps': 27254, 'batch_loss/train': 0.7237394331023097}
+12/26/2021 13:23:38 - INFO - codeparrot_training - Step 27255: {'lr': 0.0002159396661410521, 'samples': 13955072, 'steps': 27255, 'batch_loss/train': 0.7462077271193266}
+12/26/2021 13:23:51 - INFO - codeparrot_training - Step 27256: {'lr': 0.00021592405005507314, 'samples': 13955584, 'steps': 27256, 'batch_loss/train': 0.7743469309061766}
+12/26/2021 13:24:02 - INFO - codeparrot_training - Step 27257: {'lr': 0.0002159084341045674, 'samples': 13956096, 'steps': 27257, 'batch_loss/train': 0.7982581984251738}
+12/26/2021 13:24:12 - INFO - codeparrot_training - Step 27258: {'lr': 0.00021589281828959683, 'samples': 13956608, 'steps': 27258, 'batch_loss/train': 0.8406065278686583}
+12/26/2021 13:24:24 - INFO - codeparrot_training - Step 27259: {'lr': 0.00021587720261022353, 'samples': 13957120, 'steps': 27259, 'batch_loss/train': 0.462055156705901}
+12/26/2021 13:24:35 - INFO - codeparrot_training - Step 27260: {'lr': 0.00021586158706650955, 'samples': 13957632, 'steps': 27260, 'batch_loss/train': 0.6888150159502402}
+12/26/2021 13:24:46 - INFO - codeparrot_training - Step 27261: {'lr': 0.00021584597165851706, 'samples': 13958144, 'steps': 27261, 'batch_loss/train': 0.6735057001933455}
+12/26/2021 13:24:58 - INFO - codeparrot_training - Step 27262: {'lr': 0.0002158303563863081, 'samples': 13958656, 'steps': 27262, 'batch_loss/train': 0.8011096132104285}
+12/26/2021 13:25:08 - INFO - codeparrot_training - Step 27263: {'lr': 0.00021581474124994466, 'samples': 13959168, 'steps': 27263, 'batch_loss/train': 0.6348018195712939}
+12/26/2021 13:25:19 - INFO - codeparrot_training - Step 27264: {'lr': 0.000215799126249489, 'samples': 13959680, 'steps': 27264, 'batch_loss/train': 0.7797064799815416}
+12/26/2021 13:25:30 - INFO - codeparrot_training - Step 27265: {'lr': 0.0002157835113850031, 'samples': 13960192, 'steps': 27265, 'batch_loss/train': 0.7268147207796574}
+12/26/2021 13:25:42 - INFO - codeparrot_training - Step 27266: {'lr': 0.00021576789665654896, 'samples': 13960704, 'steps': 27266, 'batch_loss/train': 0.7752494341693819}
+12/26/2021 13:25:53 - INFO - codeparrot_training - Step 27267: {'lr': 0.00021575228206418873, 'samples': 13961216, 'steps': 27267, 'batch_loss/train': 0.6358629220048897}
+12/26/2021 13:26:04 - INFO - codeparrot_training - Step 27268: {'lr': 0.0002157366676079846, 'samples': 13961728, 'steps': 27268, 'batch_loss/train': 0.7664312198758125}
+12/26/2021 13:26:16 - INFO - codeparrot_training - Step 27269: {'lr': 0.00021572105328799846, 'samples': 13962240, 'steps': 27269, 'batch_loss/train': 0.6792155737057328}
+12/26/2021 13:26:26 - INFO - codeparrot_training - Step 27270: {'lr': 0.00021570543910429254, 'samples': 13962752, 'steps': 27270, 'batch_loss/train': 0.7756683919578791}
+12/26/2021 13:26:37 - INFO - codeparrot_training - Step 27271: {'lr': 0.00021568982505692876, 'samples': 13963264, 'steps': 27271, 'batch_loss/train': 0.8135379033628851}
+12/26/2021 13:26:50 - INFO - codeparrot_training - Step 27272: {'lr': 0.0002156742111459693, 'samples': 13963776, 'steps': 27272, 'batch_loss/train': 0.568958580144681}
+12/26/2021 13:27:01 - INFO - codeparrot_training - Step 27273: {'lr': 0.00021565859737147626, 'samples': 13964288, 'steps': 27273, 'batch_loss/train': 0.7800338980741799}
+12/26/2021 13:27:11 - INFO - codeparrot_training - Step 27274: {'lr': 0.00021564298373351172, 'samples': 13964800, 'steps': 27274, 'batch_loss/train': 1.727379010990262}
+12/26/2021 13:27:22 - INFO - codeparrot_training - Step 27275: {'lr': 0.00021562737023213754, 'samples': 13965312, 'steps': 27275, 'batch_loss/train': 0.7799513605423272}
+12/26/2021 13:27:34 - INFO - codeparrot_training - Step 27276: {'lr': 0.000215611756867416, 'samples': 13965824, 'steps': 27276, 'batch_loss/train': 0.7438491573557258}
+12/26/2021 13:27:45 - INFO - codeparrot_training - Step 27277: {'lr': 0.00021559614363940926, 'samples': 13966336, 'steps': 27277, 'batch_loss/train': 0.7821432148339227}
+12/26/2021 13:27:55 - INFO - codeparrot_training - Step 27278: {'lr': 0.0002155805305481791, 'samples': 13966848, 'steps': 27278, 'batch_loss/train': 0.7806840687990189}
+12/26/2021 13:28:07 - INFO - codeparrot_training - Step 27279: {'lr': 0.00021556491759378783, 'samples': 13967360, 'steps': 27279, 'batch_loss/train': 0.7307316893711686}
+12/26/2021 13:28:18 - INFO - codeparrot_training - Step 27280: {'lr': 0.0002155493047762975, 'samples': 13967872, 'steps': 27280, 'batch_loss/train': 0.9692784659564495}
+12/26/2021 13:28:29 - INFO - codeparrot_training - Step 27281: {'lr': 0.00021553369209577006, 'samples': 13968384, 'steps': 27281, 'batch_loss/train': 0.7906979527324438}
+12/26/2021 13:28:41 - INFO - codeparrot_training - Step 27282: {'lr': 0.0002155180795522676, 'samples': 13968896, 'steps': 27282, 'batch_loss/train': 0.7414973890408874}
+12/26/2021 13:28:52 - INFO - codeparrot_training - Step 27283: {'lr': 0.00021550246714585234, 'samples': 13969408, 'steps': 27283, 'batch_loss/train': 0.7714155362918973}
+12/26/2021 13:29:03 - INFO - codeparrot_training - Step 27284: {'lr': 0.00021548685487658624, 'samples': 13969920, 'steps': 27284, 'batch_loss/train': 0.7386543401516974}
+12/26/2021 13:29:13 - INFO - codeparrot_training - Step 27285: {'lr': 0.00021547124274453133, 'samples': 13970432, 'steps': 27285, 'batch_loss/train': 0.9896892933174968}
+12/26/2021 13:29:25 - INFO - codeparrot_training - Step 27286: {'lr': 0.00021545563074974966, 'samples': 13970944, 'steps': 27286, 'batch_loss/train': 0.8365190960466862}
+12/26/2021 13:29:36 - INFO - codeparrot_training - Step 27287: {'lr': 0.00021544001889230348, 'samples': 13971456, 'steps': 27287, 'batch_loss/train': 0.8277174970135093}
+12/26/2021 13:29:46 - INFO - codeparrot_training - Step 27288: {'lr': 0.00021542440717225466, 'samples': 13971968, 'steps': 27288, 'batch_loss/train': 0.721927460283041}
+12/26/2021 13:29:59 - INFO - codeparrot_training - Step 27289: {'lr': 0.0002154087955896653, 'samples': 13972480, 'steps': 27289, 'batch_loss/train': 0.6463753036223352}
+12/26/2021 13:30:09 - INFO - codeparrot_training - Step 27290: {'lr': 0.0002153931841445976, 'samples': 13972992, 'steps': 27290, 'batch_loss/train': 0.7168538812547922}
+12/26/2021 13:30:20 - INFO - codeparrot_training - Step 27291: {'lr': 0.0002153775728371135, 'samples': 13973504, 'steps': 27291, 'batch_loss/train': 0.6692867192905396}
+12/26/2021 13:30:32 - INFO - codeparrot_training - Step 27292: {'lr': 0.00021536196166727515, 'samples': 13974016, 'steps': 27292, 'batch_loss/train': 0.83882902469486}
+12/26/2021 13:30:43 - INFO - codeparrot_training - Step 27293: {'lr': 0.00021534635063514442, 'samples': 13974528, 'steps': 27293, 'batch_loss/train': 0.7865318530239165}
+12/26/2021 13:30:53 - INFO - codeparrot_training - Step 27294: {'lr': 0.00021533073974078367, 'samples': 13975040, 'steps': 27294, 'batch_loss/train': 0.7282458022236824}
+12/26/2021 13:31:04 - INFO - codeparrot_training - Step 27295: {'lr': 0.00021531512898425472, 'samples': 13975552, 'steps': 27295, 'batch_loss/train': 0.9085077131167054}
+12/26/2021 13:31:17 - INFO - codeparrot_training - Step 27296: {'lr': 0.0002152995183656198, 'samples': 13976064, 'steps': 27296, 'batch_loss/train': 0.7575462497770786}
+12/26/2021 13:31:27 - INFO - codeparrot_training - Step 27297: {'lr': 0.0002152839078849408, 'samples': 13976576, 'steps': 27297, 'batch_loss/train': 0.780556540004909}
+12/26/2021 13:31:38 - INFO - codeparrot_training - Step 27298: {'lr': 0.0002152682975422799, 'samples': 13977088, 'steps': 27298, 'batch_loss/train': 1.1239968948066235}
+12/26/2021 13:31:50 - INFO - codeparrot_training - Step 27299: {'lr': 0.00021525268733769919, 'samples': 13977600, 'steps': 27299, 'batch_loss/train': 0.7615221943706274}
+12/26/2021 13:32:00 - INFO - codeparrot_training - Step 27300: {'lr': 0.00021523707727126056, 'samples': 13978112, 'steps': 27300, 'batch_loss/train': 0.7327823960222304}
+12/26/2021 13:32:11 - INFO - codeparrot_training - Step 27301: {'lr': 0.0002152214673430263, 'samples': 13978624, 'steps': 27301, 'batch_loss/train': 0.6219471669755876}
+12/26/2021 13:32:24 - INFO - codeparrot_training - Step 27302: {'lr': 0.0002152058575530583, 'samples': 13979136, 'steps': 27302, 'batch_loss/train': 0.7311607603915036}
+12/26/2021 13:32:35 - INFO - codeparrot_training - Step 27303: {'lr': 0.00021519024790141872, 'samples': 13979648, 'steps': 27303, 'batch_loss/train': 0.7338210600428283}
+12/26/2021 13:32:45 - INFO - codeparrot_training - Step 27304: {'lr': 0.00021517463838816947, 'samples': 13980160, 'steps': 27304, 'batch_loss/train': 0.7143482840619981}
+12/26/2021 13:32:57 - INFO - codeparrot_training - Step 27305: {'lr': 0.00021515902901337276, 'samples': 13980672, 'steps': 27305, 'batch_loss/train': 0.795198455452919}
+12/26/2021 13:33:08 - INFO - codeparrot_training - Step 27306: {'lr': 0.00021514341977709066, 'samples': 13981184, 'steps': 27306, 'batch_loss/train': 0.7574572199955583}
+12/26/2021 13:33:19 - INFO - codeparrot_training - Step 27307: {'lr': 0.00021512781067938507, 'samples': 13981696, 'steps': 27307, 'batch_loss/train': 0.7312184358015656}
+12/26/2021 13:33:29 - INFO - codeparrot_training - Step 27308: {'lr': 0.00021511220172031815, 'samples': 13982208, 'steps': 27308, 'batch_loss/train': 0.6593869583448395}
+12/26/2021 13:33:41 - INFO - codeparrot_training - Step 27309: {'lr': 0.000215096592899952, 'samples': 13982720, 'steps': 27309, 'batch_loss/train': 0.8406616430729628}
+12/26/2021 13:33:52 - INFO - codeparrot_training - Step 27310: {'lr': 0.00021508098421834862, 'samples': 13983232, 'steps': 27310, 'batch_loss/train': 0.6879583867266774}
+12/26/2021 13:34:02 - INFO - codeparrot_training - Step 27311: {'lr': 0.00021506537567557, 'samples': 13983744, 'steps': 27311, 'batch_loss/train': 0.9272620435804129}
+12/26/2021 13:34:15 - INFO - codeparrot_training - Step 27312: {'lr': 0.00021504976727167825, 'samples': 13984256, 'steps': 27312, 'batch_loss/train': 0.8133239904418588}
+12/26/2021 13:34:26 - INFO - codeparrot_training - Step 27313: {'lr': 0.0002150341590067355, 'samples': 13984768, 'steps': 27313, 'batch_loss/train': 0.7298646392300725}
+12/26/2021 13:34:37 - INFO - codeparrot_training - Step 27314: {'lr': 0.00021501855088080368, 'samples': 13985280, 'steps': 27314, 'batch_loss/train': 0.6734913196414709}
+12/26/2021 13:34:49 - INFO - codeparrot_training - Step 27315: {'lr': 0.00021500294289394483, 'samples': 13985792, 'steps': 27315, 'batch_loss/train': 0.7497005555778742}
+12/26/2021 13:34:59 - INFO - codeparrot_training - Step 27316: {'lr': 0.0002149873350462212, 'samples': 13986304, 'steps': 27316, 'batch_loss/train': 0.7670929473824799}
+12/26/2021 13:35:10 - INFO - codeparrot_training - Step 27317: {'lr': 0.00021497172733769466, 'samples': 13986816, 'steps': 27317, 'batch_loss/train': 0.7528836000710726}
+12/26/2021 13:35:20 - INFO - codeparrot_training - Step 27318: {'lr': 0.00021495611976842733, 'samples': 13987328, 'steps': 27318, 'batch_loss/train': 0.7487867968156934}
+12/26/2021 13:35:33 - INFO - codeparrot_training - Step 27319: {'lr': 0.00021494051233848113, 'samples': 13987840, 'steps': 27319, 'batch_loss/train': 0.6081765084527433}
+12/26/2021 13:35:44 - INFO - codeparrot_training - Step 27320: {'lr': 0.00021492490504791833, 'samples': 13988352, 'steps': 27320, 'batch_loss/train': 0.982107064453885}
+12/26/2021 13:35:55 - INFO - codeparrot_training - Step 27321: {'lr': 0.0002149092978968008, 'samples': 13988864, 'steps': 27321, 'batch_loss/train': 0.7989742499776185}
+12/26/2021 13:36:07 - INFO - codeparrot_training - Step 27322: {'lr': 0.00021489369088519074, 'samples': 13989376, 'steps': 27322, 'batch_loss/train': 0.7249123575165868}
+12/26/2021 13:36:17 - INFO - codeparrot_training - Step 27323: {'lr': 0.00021487808401314995, 'samples': 13989888, 'steps': 27323, 'batch_loss/train': 0.903818441554904}
+12/26/2021 13:36:28 - INFO - codeparrot_training - Step 27324: {'lr': 0.00021486247728074072, 'samples': 13990400, 'steps': 27324, 'batch_loss/train': 0.7822946012020111}
+12/26/2021 13:36:40 - INFO - codeparrot_training - Step 27325: {'lr': 0.00021484687068802505, 'samples': 13990912, 'steps': 27325, 'batch_loss/train': 0.7887480761855841}
+12/26/2021 13:36:51 - INFO - codeparrot_training - Step 27326: {'lr': 0.0002148312642350649, 'samples': 13991424, 'steps': 27326, 'batch_loss/train': 0.7507677581161261}
+12/26/2021 13:37:01 - INFO - codeparrot_training - Step 27327: {'lr': 0.00021481565792192236, 'samples': 13991936, 'steps': 27327, 'batch_loss/train': 0.7593656322569586}
+12/26/2021 13:37:12 - INFO - codeparrot_training - Step 27328: {'lr': 0.00021480005174865952, 'samples': 13992448, 'steps': 27328, 'batch_loss/train': 0.7345631305361167}
+12/26/2021 13:37:24 - INFO - codeparrot_training - Step 27329: {'lr': 0.00021478444571533843, 'samples': 13992960, 'steps': 27329, 'batch_loss/train': 0.7742540074978024}
+12/26/2021 13:37:35 - INFO - codeparrot_training - Step 27330: {'lr': 0.00021476883982202095, 'samples': 13993472, 'steps': 27330, 'batch_loss/train': 0.7496516294777393}
+12/26/2021 13:37:45 - INFO - codeparrot_training - Step 27331: {'lr': 0.0002147532340687693, 'samples': 13993984, 'steps': 27331, 'batch_loss/train': 0.7428221050649881}
+12/26/2021 13:37:58 - INFO - codeparrot_training - Step 27332: {'lr': 0.00021473762845564557, 'samples': 13994496, 'steps': 27332, 'batch_loss/train': 0.7683177078142762}
+12/26/2021 13:38:09 - INFO - codeparrot_training - Step 27333: {'lr': 0.00021472202298271166, 'samples': 13995008, 'steps': 27333, 'batch_loss/train': 0.6582209975458682}
+12/26/2021 13:38:19 - INFO - codeparrot_training - Step 27334: {'lr': 0.00021470641765002958, 'samples': 13995520, 'steps': 27334, 'batch_loss/train': 0.7059986557578668}
+12/26/2021 13:38:32 - INFO - codeparrot_training - Step 27335: {'lr': 0.00021469081245766158, 'samples': 13996032, 'steps': 27335, 'batch_loss/train': 0.6611622218042612}
+12/26/2021 13:38:42 - INFO - codeparrot_training - Step 27336: {'lr': 0.00021467520740566954, 'samples': 13996544, 'steps': 27336, 'batch_loss/train': 0.6859114109538496}
+12/26/2021 13:38:53 - INFO - codeparrot_training - Step 27337: {'lr': 0.0002146596024941155, 'samples': 13997056, 'steps': 27337, 'batch_loss/train': 0.837390573695302}
+12/26/2021 13:39:05 - INFO - codeparrot_training - Step 27338: {'lr': 0.0002146439977230616, 'samples': 13997568, 'steps': 27338, 'batch_loss/train': 0.8234539162367582}
+12/26/2021 13:39:15 - INFO - codeparrot_training - Step 27339: {'lr': 0.00021462839309256983, 'samples': 13998080, 'steps': 27339, 'batch_loss/train': 0.7875824719667435}
+12/26/2021 13:39:26 - INFO - codeparrot_training - Step 27340: {'lr': 0.00021461278860270217, 'samples': 13998592, 'steps': 27340, 'batch_loss/train': 0.8308965414762497}
+12/26/2021 13:39:37 - INFO - codeparrot_training - Step 27341: {'lr': 0.0002145971842535206, 'samples': 13999104, 'steps': 27341, 'batch_loss/train': 0.6995325153693557}
+12/26/2021 13:39:49 - INFO - codeparrot_training - Step 27342: {'lr': 0.00021458158004508744, 'samples': 13999616, 'steps': 27342, 'batch_loss/train': 0.7458971869200468}
+12/26/2021 13:40:00 - INFO - codeparrot_training - Step 27343: {'lr': 0.00021456597597746443, 'samples': 14000128, 'steps': 27343, 'batch_loss/train': 0.7875542612746358}
+12/26/2021 13:40:10 - INFO - codeparrot_training - Step 27344: {'lr': 0.00021455037205071382, 'samples': 14000640, 'steps': 27344, 'batch_loss/train': 0.7613002881407738}
+12/26/2021 13:40:23 - INFO - codeparrot_training - Step 27345: {'lr': 0.0002145347682648974, 'samples': 14001152, 'steps': 27345, 'batch_loss/train': 0.6832750528119504}
+12/26/2021 13:40:33 - INFO - codeparrot_training - Step 27346: {'lr': 0.00021451916462007748, 'samples': 14001664, 'steps': 27346, 'batch_loss/train': 0.7787856794893742}
+12/26/2021 13:40:44 - INFO - codeparrot_training - Step 27347: {'lr': 0.0002145035611163159, 'samples': 14002176, 'steps': 27347, 'batch_loss/train': 0.6276735719293356}
+12/26/2021 13:40:57 - INFO - codeparrot_training - Step 27348: {'lr': 0.00021448795775367484, 'samples': 14002688, 'steps': 27348, 'batch_loss/train': 0.7256981087848544}
+12/26/2021 13:41:07 - INFO - codeparrot_training - Step 27349: {'lr': 0.00021447235453221611, 'samples': 14003200, 'steps': 27349, 'batch_loss/train': 0.7397038876079023}
+12/26/2021 13:41:18 - INFO - codeparrot_training - Step 27350: {'lr': 0.00021445675145200194, 'samples': 14003712, 'steps': 27350, 'batch_loss/train': 0.834719086997211}
+12/26/2021 13:41:29 - INFO - codeparrot_training - Step 27351: {'lr': 0.00021444114851309438, 'samples': 14004224, 'steps': 27351, 'batch_loss/train': 0.8950158753432333}
+12/26/2021 13:41:41 - INFO - codeparrot_training - Step 27352: {'lr': 0.00021442554571555526, 'samples': 14004736, 'steps': 27352, 'batch_loss/train': 0.7443508412688971}
+12/26/2021 13:41:51 - INFO - codeparrot_training - Step 27353: {'lr': 0.00021440994305944676, 'samples': 14005248, 'steps': 27353, 'batch_loss/train': 0.7121713790111244}
+12/26/2021 13:42:02 - INFO - codeparrot_training - Step 27354: {'lr': 0.00021439434054483102, 'samples': 14005760, 'steps': 27354, 'batch_loss/train': 0.7105630626901984}
+12/26/2021 13:42:14 - INFO - codeparrot_training - Step 27355: {'lr': 0.0002143787381717699, 'samples': 14006272, 'steps': 27355, 'batch_loss/train': 0.698831470683217}
+12/26/2021 13:42:25 - INFO - codeparrot_training - Step 27356: {'lr': 0.00021436313594032537, 'samples': 14006784, 'steps': 27356, 'batch_loss/train': 0.7169869374483824}
+12/26/2021 13:42:35 - INFO - codeparrot_training - Step 27357: {'lr': 0.00021434753385055958, 'samples': 14007296, 'steps': 27357, 'batch_loss/train': 0.7608541250228882}
+12/26/2021 13:42:48 - INFO - codeparrot_training - Step 27358: {'lr': 0.00021433193190253458, 'samples': 14007808, 'steps': 27358, 'batch_loss/train': 0.6376302824355662}
+12/26/2021 13:42:59 - INFO - codeparrot_training - Step 27359: {'lr': 0.00021431633009631236, 'samples': 14008320, 'steps': 27359, 'batch_loss/train': 0.8062478937208652}
+12/26/2021 13:43:09 - INFO - codeparrot_training - Step 27360: {'lr': 0.00021430072843195482, 'samples': 14008832, 'steps': 27360, 'batch_loss/train': 0.6433372311294079}
+12/26/2021 13:43:20 - INFO - codeparrot_training - Step 27361: {'lr': 0.00021428512690952424, 'samples': 14009344, 'steps': 27361, 'batch_loss/train': 0.6805545720271766}
+12/26/2021 13:43:32 - INFO - codeparrot_training - Step 27362: {'lr': 0.00021426952552908247, 'samples': 14009856, 'steps': 27362, 'batch_loss/train': 0.7056457819417119}
+12/26/2021 13:43:43 - INFO - codeparrot_training - Step 27363: {'lr': 0.00021425392429069148, 'samples': 14010368, 'steps': 27363, 'batch_loss/train': 0.7411837382242084}
+12/26/2021 13:43:53 - INFO - codeparrot_training - Step 27364: {'lr': 0.00021423832319441354, 'samples': 14010880, 'steps': 27364, 'batch_loss/train': 0.7282661478966475}
+12/26/2021 13:44:05 - INFO - codeparrot_training - Step 27365: {'lr': 0.0002142227222403105, 'samples': 14011392, 'steps': 27365, 'batch_loss/train': 0.7000159891322255}
+12/26/2021 13:44:16 - INFO - codeparrot_training - Step 27366: {'lr': 0.00021420712142844434, 'samples': 14011904, 'steps': 27366, 'batch_loss/train': 0.7185401599854231}
+12/26/2021 13:44:27 - INFO - codeparrot_training - Step 27367: {'lr': 0.00021419152075887712, 'samples': 14012416, 'steps': 27367, 'batch_loss/train': 0.8292774418368936}
+12/26/2021 13:44:39 - INFO - codeparrot_training - Step 27368: {'lr': 0.00021417592023167102, 'samples': 14012928, 'steps': 27368, 'batch_loss/train': 0.775236853864044}
+12/26/2021 13:44:49 - INFO - codeparrot_training - Step 27369: {'lr': 0.00021416031984688782, 'samples': 14013440, 'steps': 27369, 'batch_loss/train': 0.7054244875907898}
+12/26/2021 13:45:00 - INFO - codeparrot_training - Step 27370: {'lr': 0.00021414471960458974, 'samples': 14013952, 'steps': 27370, 'batch_loss/train': 0.793754521291703}
+12/26/2021 13:45:13 - INFO - codeparrot_training - Step 27371: {'lr': 0.00021412911950483866, 'samples': 14014464, 'steps': 27371, 'batch_loss/train': 0.7846129550598562}
+12/26/2021 13:45:23 - INFO - codeparrot_training - Step 27372: {'lr': 0.00021411351954769665, 'samples': 14014976, 'steps': 27372, 'batch_loss/train': 0.6844465694739483}
+12/26/2021 13:45:34 - INFO - codeparrot_training - Step 27373: {'lr': 0.00021409791973322578, 'samples': 14015488, 'steps': 27373, 'batch_loss/train': 0.7256845971569419}
+12/26/2021 13:45:44 - INFO - codeparrot_training - Step 27374: {'lr': 0.000214082320061488, 'samples': 14016000, 'steps': 27374, 'batch_loss/train': 0.757585521787405}
+12/26/2021 13:45:57 - INFO - codeparrot_training - Step 27375: {'lr': 0.00021406672053254536, 'samples': 14016512, 'steps': 27375, 'batch_loss/train': 0.7790449121966958}
+12/26/2021 13:46:07 - INFO - codeparrot_training - Step 27376: {'lr': 0.00021405112114645985, 'samples': 14017024, 'steps': 27376, 'batch_loss/train': 0.6959700677543879}
+12/26/2021 13:46:18 - INFO - codeparrot_training - Step 27377: {'lr': 0.0002140355219032936, 'samples': 14017536, 'steps': 27377, 'batch_loss/train': 0.4756632625358179}
+12/26/2021 13:46:31 - INFO - codeparrot_training - Step 27378: {'lr': 0.0002140199228031084, 'samples': 14018048, 'steps': 27378, 'batch_loss/train': 0.7108594734454527}
+12/26/2021 13:46:42 - INFO - codeparrot_training - Step 27379: {'lr': 0.00021400432384596646, 'samples': 14018560, 'steps': 27379, 'batch_loss/train': 0.854108223807998}
+12/26/2021 13:46:52 - INFO - codeparrot_training - Step 27380: {'lr': 0.00021398872503192978, 'samples': 14019072, 'steps': 27380, 'batch_loss/train': 0.7838931204751134}
+12/26/2021 13:47:04 - INFO - codeparrot_training - Step 27381: {'lr': 0.0002139731263610603, 'samples': 14019584, 'steps': 27381, 'batch_loss/train': 0.8206744994968176}
+12/26/2021 13:47:15 - INFO - codeparrot_training - Step 27382: {'lr': 0.00021395752783341998, 'samples': 14020096, 'steps': 27382, 'batch_loss/train': 0.8776143128052354}
+12/26/2021 13:47:25 - INFO - codeparrot_training - Step 27383: {'lr': 0.00021394192944907103, 'samples': 14020608, 'steps': 27383, 'batch_loss/train': 0.7816627025604248}
+12/26/2021 13:47:36 - INFO - codeparrot_training - Step 27384: {'lr': 0.0002139263312080754, 'samples': 14021120, 'steps': 27384, 'batch_loss/train': 0.7261860747821629}
+12/26/2021 13:47:48 - INFO - codeparrot_training - Step 27385: {'lr': 0.00021391073311049496, 'samples': 14021632, 'steps': 27385, 'batch_loss/train': 0.7626169072464108}
+12/26/2021 13:47:59 - INFO - codeparrot_training - Step 27386: {'lr': 0.00021389513515639177, 'samples': 14022144, 'steps': 27386, 'batch_loss/train': 0.6116462723584846}
+12/26/2021 13:48:10 - INFO - codeparrot_training - Step 27387: {'lr': 0.000213879537345828, 'samples': 14022656, 'steps': 27387, 'batch_loss/train': 0.8628039751201868}
+12/26/2021 13:48:22 - INFO - codeparrot_training - Step 27388: {'lr': 0.00021386393967886552, 'samples': 14023168, 'steps': 27388, 'batch_loss/train': 0.7532918192446232}
+12/26/2021 13:48:33 - INFO - codeparrot_training - Step 27389: {'lr': 0.0002138483421555663, 'samples': 14023680, 'steps': 27389, 'batch_loss/train': 0.7642232980579138}
+12/26/2021 13:48:44 - INFO - codeparrot_training - Step 27390: {'lr': 0.00021383274477599251, 'samples': 14024192, 'steps': 27390, 'batch_loss/train': 0.7356998829636723}
+12/26/2021 13:48:56 - INFO - codeparrot_training - Step 27391: {'lr': 0.0002138171475402061, 'samples': 14024704, 'steps': 27391, 'batch_loss/train': 0.7197974873706698}
+12/26/2021 13:49:06 - INFO - codeparrot_training - Step 27392: {'lr': 0.00021380155044826898, 'samples': 14025216, 'steps': 27392, 'batch_loss/train': 0.7219215063378215}
+12/26/2021 13:49:17 - INFO - codeparrot_training - Step 27393: {'lr': 0.0002137859535002432, 'samples': 14025728, 'steps': 27393, 'batch_loss/train': 0.715513383038342}
+12/26/2021 13:49:28 - INFO - codeparrot_training - Step 27394: {'lr': 0.00021377035669619085, 'samples': 14026240, 'steps': 27394, 'batch_loss/train': 0.6809238884598017}
+12/26/2021 13:49:40 - INFO - codeparrot_training - Step 27395: {'lr': 0.00021375476003617386, 'samples': 14026752, 'steps': 27395, 'batch_loss/train': 0.725053469534032}
+12/26/2021 13:49:50 - INFO - codeparrot_training - Step 27396: {'lr': 0.00021373916352025434, 'samples': 14027264, 'steps': 27396, 'batch_loss/train': 0.8133892444893718}
+12/26/2021 13:50:01 - INFO - codeparrot_training - Step 27397: {'lr': 0.00021372356714849405, 'samples': 14027776, 'steps': 27397, 'batch_loss/train': 0.767081051133573}
+12/26/2021 13:50:14 - INFO - codeparrot_training - Step 27398: {'lr': 0.00021370797092095523, 'samples': 14028288, 'steps': 27398, 'batch_loss/train': 0.815512964501977}
+12/26/2021 13:50:24 - INFO - codeparrot_training - Step 27399: {'lr': 0.0002136923748376999, 'samples': 14028800, 'steps': 27399, 'batch_loss/train': 0.7347484575584531}
+12/26/2021 13:50:35 - INFO - codeparrot_training - Step 27400: {'lr': 0.00021367677889878983, 'samples': 14029312, 'steps': 27400, 'batch_loss/train': 0.7425196873955429}
+12/26/2021 13:50:47 - INFO - codeparrot_training - Step 27401: {'lr': 0.00021366118310428732, 'samples': 14029824, 'steps': 27401, 'batch_loss/train': 0.6715479507111013}
+12/26/2021 13:50:58 - INFO - codeparrot_training - Step 27402: {'lr': 0.00021364558745425412, 'samples': 14030336, 'steps': 27402, 'batch_loss/train': 0.7701253080740571}
+12/26/2021 13:51:08 - INFO - codeparrot_training - Step 27403: {'lr': 0.00021362999194875242, 'samples': 14030848, 'steps': 27403, 'batch_loss/train': 0.698065287899226}
+12/26/2021 13:51:21 - INFO - codeparrot_training - Step 27404: {'lr': 0.00021361439658784404, 'samples': 14031360, 'steps': 27404, 'batch_loss/train': 0.7713523618876934}
+12/26/2021 13:51:31 - INFO - codeparrot_training - Step 27405: {'lr': 0.00021359880137159117, 'samples': 14031872, 'steps': 27405, 'batch_loss/train': 0.7367908740416169}
+12/26/2021 13:51:42 - INFO - codeparrot_training - Step 27406: {'lr': 0.0002135832063000557, 'samples': 14032384, 'steps': 27406, 'batch_loss/train': 0.723023840226233}
+12/26/2021 13:51:53 - INFO - codeparrot_training - Step 27407: {'lr': 0.00021356761137329963, 'samples': 14032896, 'steps': 27407, 'batch_loss/train': 0.5365917473100126}
+12/26/2021 13:52:05 - INFO - codeparrot_training - Step 27408: {'lr': 0.00021355201659138494, 'samples': 14033408, 'steps': 27408, 'batch_loss/train': 0.7949535623774864}
+12/26/2021 13:52:16 - INFO - codeparrot_training - Step 27409: {'lr': 0.00021353642195437377, 'samples': 14033920, 'steps': 27409, 'batch_loss/train': 0.7375349141657352}
+12/26/2021 13:52:27 - INFO - codeparrot_training - Step 27410: {'lr': 0.000213520827462328, 'samples': 14034432, 'steps': 27410, 'batch_loss/train': 0.7822269746102393}
+12/26/2021 13:52:39 - INFO - codeparrot_training - Step 27411: {'lr': 0.00021350523311530956, 'samples': 14034944, 'steps': 27411, 'batch_loss/train': 0.8198210634291172}
+12/26/2021 13:52:49 - INFO - codeparrot_training - Step 27412: {'lr': 0.00021348963891338056, 'samples': 14035456, 'steps': 27412, 'batch_loss/train': 0.6518556969240308}
+12/26/2021 13:53:00 - INFO - codeparrot_training - Step 27413: {'lr': 0.00021347404485660303, 'samples': 14035968, 'steps': 27413, 'batch_loss/train': 0.625104307604488}
+12/26/2021 13:53:12 - INFO - codeparrot_training - Step 27414: {'lr': 0.00021345845094503888, 'samples': 14036480, 'steps': 27414, 'batch_loss/train': 0.6648952420218848}
+12/26/2021 13:53:23 - INFO - codeparrot_training - Step 27415: {'lr': 0.00021344285717875007, 'samples': 14036992, 'steps': 27415, 'batch_loss/train': 0.5767999533563852}
+12/26/2021 13:53:33 - INFO - codeparrot_training - Step 27416: {'lr': 0.00021342726355779874, 'samples': 14037504, 'steps': 27416, 'batch_loss/train': 0.8002425227314234}
+12/26/2021 13:53:44 - INFO - codeparrot_training - Step 27417: {'lr': 0.00021341167008224676, 'samples': 14038016, 'steps': 27417, 'batch_loss/train': 0.7636952525936067}
+12/26/2021 13:53:57 - INFO - codeparrot_training - Step 27418: {'lr': 0.00021339607675215623, 'samples': 14038528, 'steps': 27418, 'batch_loss/train': 0.7894334369339049}
+12/26/2021 13:54:07 - INFO - codeparrot_training - Step 27419: {'lr': 0.00021338048356758896, 'samples': 14039040, 'steps': 27419, 'batch_loss/train': 0.751232647919096}
+12/26/2021 13:54:18 - INFO - codeparrot_training - Step 27420: {'lr': 0.00021336489052860717, 'samples': 14039552, 'steps': 27420, 'batch_loss/train': 0.6808071164414287}
+12/26/2021 13:54:30 - INFO - codeparrot_training - Step 27421: {'lr': 0.00021334929763527268, 'samples': 14040064, 'steps': 27421, 'batch_loss/train': 0.8142749294638634}
+12/26/2021 13:54:41 - INFO - codeparrot_training - Step 27422: {'lr': 0.0002133337048876476, 'samples': 14040576, 'steps': 27422, 'batch_loss/train': 0.8602611385285854}
+12/26/2021 13:54:52 - INFO - codeparrot_training - Step 27423: {'lr': 0.00021331811228579375, 'samples': 14041088, 'steps': 27423, 'batch_loss/train': 0.7152037741616368}
+12/26/2021 13:55:04 - INFO - codeparrot_training - Step 27424: {'lr': 0.00021330251982977334, 'samples': 14041600, 'steps': 27424, 'batch_loss/train': 0.6925508691929281}
+12/26/2021 13:55:14 - INFO - codeparrot_training - Step 27425: {'lr': 0.00021328692751964822, 'samples': 14042112, 'steps': 27425, 'batch_loss/train': 0.7436532387509942}
+12/26/2021 13:55:25 - INFO - codeparrot_training - Step 27426: {'lr': 0.00021327133535548038, 'samples': 14042624, 'steps': 27426, 'batch_loss/train': 0.46217688638716936}
+12/26/2021 13:55:38 - INFO - codeparrot_training - Step 27427: {'lr': 0.00021325574333733188, 'samples': 14043136, 'steps': 27427, 'batch_loss/train': 0.6827804450877011}
+12/26/2021 13:55:48 - INFO - codeparrot_training - Step 27428: {'lr': 0.0002132401514652647, 'samples': 14043648, 'steps': 27428, 'batch_loss/train': 0.8060486786998808}
+12/26/2021 13:55:59 - INFO - codeparrot_training - Step 27429: {'lr': 0.00021322455973934082, 'samples': 14044160, 'steps': 27429, 'batch_loss/train': 0.7341833962127566}
+12/26/2021 13:56:10 - INFO - codeparrot_training - Step 27430: {'lr': 0.0002132089681596221, 'samples': 14044672, 'steps': 27430, 'batch_loss/train': 0.7362618185579777}
+12/26/2021 13:56:22 - INFO - codeparrot_training - Step 27431: {'lr': 0.00021319337672617066, 'samples': 14045184, 'steps': 27431, 'batch_loss/train': 0.758668229333125}
+12/26/2021 13:56:33 - INFO - codeparrot_training - Step 27432: {'lr': 0.00021317778543904854, 'samples': 14045696, 'steps': 27432, 'batch_loss/train': 0.6630243993131444}
+12/26/2021 13:56:43 - INFO - codeparrot_training - Step 27433: {'lr': 0.00021316219429831756, 'samples': 14046208, 'steps': 27433, 'batch_loss/train': 0.7078460613265634}
+12/26/2021 13:56:56 - INFO - codeparrot_training - Step 27434: {'lr': 0.00021314660330403975, 'samples': 14046720, 'steps': 27434, 'batch_loss/train': 0.7466688342392445}
+12/26/2021 13:57:07 - INFO - codeparrot_training - Step 27435: {'lr': 0.00021313101245627725, 'samples': 14047232, 'steps': 27435, 'batch_loss/train': 0.7231939774937928}
+12/26/2021 13:57:17 - INFO - codeparrot_training - Step 27436: {'lr': 0.00021311542175509192, 'samples': 14047744, 'steps': 27436, 'batch_loss/train': 0.8363915020599961}
+12/26/2021 13:57:28 - INFO - codeparrot_training - Step 27437: {'lr': 0.0002130998312005456, 'samples': 14048256, 'steps': 27437, 'batch_loss/train': 0.7254895297810435}
+12/26/2021 13:57:40 - INFO - codeparrot_training - Step 27438: {'lr': 0.0002130842407927005, 'samples': 14048768, 'steps': 27438, 'batch_loss/train': 0.6996960295364261}
+12/26/2021 13:57:51 - INFO - codeparrot_training - Step 27439: {'lr': 0.00021306865053161855, 'samples': 14049280, 'steps': 27439, 'batch_loss/train': 0.8057536543346941}
+12/26/2021 13:58:01 - INFO - codeparrot_training - Step 27440: {'lr': 0.0002130530604173617, 'samples': 14049792, 'steps': 27440, 'batch_loss/train': 0.6870970940217376}
+12/26/2021 13:58:14 - INFO - codeparrot_training - Step 27441: {'lr': 0.00021303747044999183, 'samples': 14050304, 'steps': 27441, 'batch_loss/train': 0.7393595790490508}
+12/26/2021 13:58:24 - INFO - codeparrot_training - Step 27442: {'lr': 0.00021302188062957114, 'samples': 14050816, 'steps': 27442, 'batch_loss/train': 0.7588392328470945}
+12/26/2021 13:58:35 - INFO - codeparrot_training - Step 27443: {'lr': 0.00021300629095616144, 'samples': 14051328, 'steps': 27443, 'batch_loss/train': 0.7193299857899547}
+12/26/2021 13:58:47 - INFO - codeparrot_training - Step 27444: {'lr': 0.00021299070142982482, 'samples': 14051840, 'steps': 27444, 'batch_loss/train': 0.8250915680546314}
+12/26/2021 13:58:58 - INFO - codeparrot_training - Step 27445: {'lr': 0.00021297511205062308, 'samples': 14052352, 'steps': 27445, 'batch_loss/train': 1.2071020039729774}
+12/26/2021 13:59:09 - INFO - codeparrot_training - Step 27446: {'lr': 0.0002129595228186184, 'samples': 14052864, 'steps': 27446, 'batch_loss/train': 0.6906975638121367}
+12/26/2021 13:59:19 - INFO - codeparrot_training - Step 27447: {'lr': 0.00021294393373387266, 'samples': 14053376, 'steps': 27447, 'batch_loss/train': 0.8235173812136054}
+12/26/2021 13:59:32 - INFO - codeparrot_training - Step 27448: {'lr': 0.0002129283447964478, 'samples': 14053888, 'steps': 27448, 'batch_loss/train': 0.759198906365782}
+12/26/2021 13:59:43 - INFO - codeparrot_training - Step 27449: {'lr': 0.0002129127560064059, 'samples': 14054400, 'steps': 27449, 'batch_loss/train': 0.7453299295157194}
+12/26/2021 13:59:53 - INFO - codeparrot_training - Step 27450: {'lr': 0.00021289716736380885, 'samples': 14054912, 'steps': 27450, 'batch_loss/train': 0.803541692905128}
+12/26/2021 14:00:05 - INFO - codeparrot_training - Step 27451: {'lr': 0.00021288157886871872, 'samples': 14055424, 'steps': 27451, 'batch_loss/train': 0.7352191489189863}
+12/26/2021 14:00:16 - INFO - codeparrot_training - Step 27452: {'lr': 0.00021286599052119728, 'samples': 14055936, 'steps': 27452, 'batch_loss/train': 0.6392095913179219}
+12/26/2021 14:00:26 - INFO - codeparrot_training - Step 27453: {'lr': 0.00021285040232130672, 'samples': 14056448, 'steps': 27453, 'batch_loss/train': 0.7259219083935022}
+12/26/2021 14:00:39 - INFO - codeparrot_training - Step 27454: {'lr': 0.00021283481426910897, 'samples': 14056960, 'steps': 27454, 'batch_loss/train': 0.7768657645210624}
+12/26/2021 14:00:49 - INFO - codeparrot_training - Step 27455: {'lr': 0.000212819226364666, 'samples': 14057472, 'steps': 27455, 'batch_loss/train': 0.7818382414989173}
+12/26/2021 14:01:00 - INFO - codeparrot_training - Step 27456: {'lr': 0.0002128036386080396, 'samples': 14057984, 'steps': 27456, 'batch_loss/train': 0.7406553188338876}
+12/26/2021 14:01:13 - INFO - codeparrot_training - Step 27457: {'lr': 0.00021278805099929194, 'samples': 14058496, 'steps': 27457, 'batch_loss/train': 0.8145861243829131}
+12/26/2021 14:01:23 - INFO - codeparrot_training - Step 27458: {'lr': 0.000212772463538485, 'samples': 14059008, 'steps': 27458, 'batch_loss/train': 0.7333097690716386}
+12/26/2021 14:01:34 - INFO - codeparrot_training - Step 27459: {'lr': 0.00021275687622568064, 'samples': 14059520, 'steps': 27459, 'batch_loss/train': 0.7677144100889564}
+12/26/2021 14:01:45 - INFO - codeparrot_training - Step 27460: {'lr': 0.00021274128906094083, 'samples': 14060032, 'steps': 27460, 'batch_loss/train': 0.7111782357096672}
+12/26/2021 14:01:57 - INFO - codeparrot_training - Step 27461: {'lr': 0.0002127257020443277, 'samples': 14060544, 'steps': 27461, 'batch_loss/train': 0.7449412643909454}
+12/26/2021 14:02:07 - INFO - codeparrot_training - Step 27462: {'lr': 0.00021271011517590306, 'samples': 14061056, 'steps': 27462, 'batch_loss/train': 0.7833515191450715}
+12/26/2021 14:02:18 - INFO - codeparrot_training - Step 27463: {'lr': 0.00021269452845572884, 'samples': 14061568, 'steps': 27463, 'batch_loss/train': 0.8655624212697148}
+12/26/2021 14:02:31 - INFO - codeparrot_training - Step 27464: {'lr': 0.0002126789418838672, 'samples': 14062080, 'steps': 27464, 'batch_loss/train': 0.7563475044444203}
+12/26/2021 14:02:42 - INFO - codeparrot_training - Step 27465: {'lr': 0.00021266335546038002, 'samples': 14062592, 'steps': 27465, 'batch_loss/train': 0.7239959571743384}
+12/26/2021 14:02:52 - INFO - codeparrot_training - Step 27466: {'lr': 0.00021264776918532917, 'samples': 14063104, 'steps': 27466, 'batch_loss/train': 0.7107904488220811}
+12/26/2021 14:03:05 - INFO - codeparrot_training - Step 27467: {'lr': 0.0002126321830587766, 'samples': 14063616, 'steps': 27467, 'batch_loss/train': 0.8896985407918692}
+12/26/2021 14:03:15 - INFO - codeparrot_training - Step 27468: {'lr': 0.00021261659708078453, 'samples': 14064128, 'steps': 27468, 'batch_loss/train': 1.0209396263526287}
+12/26/2021 14:03:26 - INFO - codeparrot_training - Step 27469: {'lr': 0.00021260101125141466, 'samples': 14064640, 'steps': 27469, 'batch_loss/train': 0.7232308965176344}
+12/26/2021 14:03:37 - INFO - codeparrot_training - Step 27470: {'lr': 0.0002125854255707291, 'samples': 14065152, 'steps': 27470, 'batch_loss/train': 0.7214729031547904}
+12/26/2021 14:03:49 - INFO - codeparrot_training - Step 27471: {'lr': 0.00021256984003878966, 'samples': 14065664, 'steps': 27471, 'batch_loss/train': 0.7331533394753933}
+12/26/2021 14:04:00 - INFO - codeparrot_training - Step 27472: {'lr': 0.00021255425465565848, 'samples': 14066176, 'steps': 27472, 'batch_loss/train': 0.7935335780493915}
+12/26/2021 14:04:10 - INFO - codeparrot_training - Step 27473: {'lr': 0.00021253866942139746, 'samples': 14066688, 'steps': 27473, 'batch_loss/train': 0.7163341604173183}
+12/26/2021 14:04:22 - INFO - codeparrot_training - Step 27474: {'lr': 0.00021252308433606845, 'samples': 14067200, 'steps': 27474, 'batch_loss/train': 0.8058529701083899}
+12/26/2021 14:04:33 - INFO - codeparrot_training - Step 27475: {'lr': 0.00021250749939973363, 'samples': 14067712, 'steps': 27475, 'batch_loss/train': 0.8155926428735256}
+12/26/2021 14:04:44 - INFO - codeparrot_training - Step 27476: {'lr': 0.00021249191461245476, 'samples': 14068224, 'steps': 27476, 'batch_loss/train': 0.7240930711850524}
+12/26/2021 14:04:57 - INFO - codeparrot_training - Step 27477: {'lr': 0.00021247632997429395, 'samples': 14068736, 'steps': 27477, 'batch_loss/train': 0.7854027720168233}
+12/26/2021 14:05:07 - INFO - codeparrot_training - Step 27478: {'lr': 0.00021246074548531298, 'samples': 14069248, 'steps': 27478, 'batch_loss/train': 0.6895593856461346}
+12/26/2021 14:05:18 - INFO - codeparrot_training - Step 27479: {'lr': 0.00021244516114557395, 'samples': 14069760, 'steps': 27479, 'batch_loss/train': 0.7933257790282369}
+12/26/2021 14:05:29 - INFO - codeparrot_training - Step 27480: {'lr': 0.0002124295769551388, 'samples': 14070272, 'steps': 27480, 'batch_loss/train': 0.7846604458754882}
+12/26/2021 14:05:41 - INFO - codeparrot_training - Step 27481: {'lr': 0.00021241399291406954, 'samples': 14070784, 'steps': 27481, 'batch_loss/train': 0.6970454277470708}
+12/26/2021 14:05:51 - INFO - codeparrot_training - Step 27482: {'lr': 0.00021239840902242792, 'samples': 14071296, 'steps': 27482, 'batch_loss/train': 0.7796653024852276}
+12/26/2021 14:06:02 - INFO - codeparrot_training - Step 27483: {'lr': 0.00021238282528027603, 'samples': 14071808, 'steps': 27483, 'batch_loss/train': 0.764684995636344}
+12/26/2021 14:06:14 - INFO - codeparrot_training - Step 27484: {'lr': 0.00021236724168767593, 'samples': 14072320, 'steps': 27484, 'batch_loss/train': 0.7123764264397323}
+12/26/2021 14:06:25 - INFO - codeparrot_training - Step 27485: {'lr': 0.00021235165824468933, 'samples': 14072832, 'steps': 27485, 'batch_loss/train': 0.6907447972334921}
+12/26/2021 14:06:35 - INFO - codeparrot_training - Step 27486: {'lr': 0.00021233607495137842, 'samples': 14073344, 'steps': 27486, 'batch_loss/train': 0.527449149871245}
+12/26/2021 14:06:48 - INFO - codeparrot_training - Step 27487: {'lr': 0.00021232049180780504, 'samples': 14073856, 'steps': 27487, 'batch_loss/train': 0.7105193589814007}
+12/26/2021 14:06:59 - INFO - codeparrot_training - Step 27488: {'lr': 0.00021230490881403116, 'samples': 14074368, 'steps': 27488, 'batch_loss/train': 0.7007274520583451}
+12/26/2021 14:07:09 - INFO - codeparrot_training - Step 27489: {'lr': 0.00021228932597011865, 'samples': 14074880, 'steps': 27489, 'batch_loss/train': 0.7285040142014623}
+12/26/2021 14:07:21 - INFO - codeparrot_training - Step 27490: {'lr': 0.00021227374327612966, 'samples': 14075392, 'steps': 27490, 'batch_loss/train': 0.7185186264105141}
+12/26/2021 14:07:32 - INFO - codeparrot_training - Step 27491: {'lr': 0.000212258160732126, 'samples': 14075904, 'steps': 27491, 'batch_loss/train': 0.8200795892626047}
+12/26/2021 14:07:43 - INFO - codeparrot_training - Step 27492: {'lr': 0.0002122425783381696, 'samples': 14076416, 'steps': 27492, 'batch_loss/train': 0.7352728857658803}
+12/26/2021 14:07:53 - INFO - codeparrot_training - Step 27493: {'lr': 0.00021222699609432243, 'samples': 14076928, 'steps': 27493, 'batch_loss/train': 0.7604974801652133}
+12/26/2021 14:08:06 - INFO - codeparrot_training - Step 27494: {'lr': 0.00021221141400064653, 'samples': 14077440, 'steps': 27494, 'batch_loss/train': 0.7580869700759649}
+12/26/2021 14:08:17 - INFO - codeparrot_training - Step 27495: {'lr': 0.00021219583205720375, 'samples': 14077952, 'steps': 27495, 'batch_loss/train': 0.7945199115201831}
+12/26/2021 14:08:27 - INFO - codeparrot_training - Step 27496: {'lr': 0.0002121802502640561, 'samples': 14078464, 'steps': 27496, 'batch_loss/train': 0.5193465107586235}
+12/26/2021 14:08:40 - INFO - codeparrot_training - Step 27497: {'lr': 0.0002121646686212654, 'samples': 14078976, 'steps': 27497, 'batch_loss/train': 0.7732472894713283}
+12/26/2021 14:08:50 - INFO - codeparrot_training - Step 27498: {'lr': 0.00021214908712889374, 'samples': 14079488, 'steps': 27498, 'batch_loss/train': 0.5588060114532709}
+12/26/2021 14:09:01 - INFO - codeparrot_training - Step 27499: {'lr': 0.00021213350578700308, 'samples': 14080000, 'steps': 27499, 'batch_loss/train': 0.7931059999391437}
+12/26/2021 14:09:01 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+12/26/2021 14:12:22 - INFO - codeparrot_training - Step 27500: {'loss/eval': 0.7743321061134338, 'perplexity': 2.169142961502075}