diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -45572,3 +45572,1009 @@ Use FP16 precision: False
 02/26/2022 11:51:29 - INFO - codeparrot_training - Step 44998: {'lr': 1.3277988736271397e-05, 'samples': 23039488, 'steps': 44998, 'loss/train': 2.744082450866699}
 02/26/2022 11:51:33 - INFO - codeparrot_training - Step 44999: {'lr': 1.327272767421525e-05, 'samples': 23040000, 'steps': 44999, 'loss/train': 2.3884806632995605}
 02/26/2022 11:51:33 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 11:51:51 - WARNING - huggingface_hub.repository - Several commits (45) will be pushed upstream.
+02/26/2022 11:51:51 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 11:52:41 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   84c8fb0..9a19140  floral-grass-11 -> floral-grass-11
+
+02/26/2022 11:52:49 - INFO - codeparrot_training - Step 45000: {'lr': 1.3267467626223605e-05, 'samples': 23040512, 'steps': 45000, 'loss/train': 2.671565294265747}
+02/26/2022 11:52:52 - INFO - codeparrot_training - Step 45001: {'lr': 1.3262208592318942e-05, 'samples': 23041024, 'steps': 45001, 'loss/train': 2.178778886795044}
+02/26/2022 11:52:58 - INFO - codeparrot_training - Step 45002: {'lr': 1.3256950572523769e-05, 'samples': 23041536, 'steps': 45002, 'loss/train': 1.4157593250274658}
+02/26/2022 11:53:01 - INFO - codeparrot_training - Step 45003: {'lr': 1.3251693566860679e-05, 'samples': 23042048, 'steps': 45003, 'loss/train': 1.25260591506958}
+02/26/2022 11:53:07 - INFO - codeparrot_training - Step 45004: {'lr': 1.3246437575352154e-05, 'samples': 23042560, 'steps': 45004, 'loss/train': 2.4184486865997314}
+02/26/2022 11:53:10 - INFO - codeparrot_training - Step 45005: {'lr': 1.3241182598020679e-05, 'samples': 23043072, 'steps': 45005, 'loss/train': 2.295644998550415}
+02/26/2022 11:53:16 - INFO - codeparrot_training - Step 45006: {'lr': 1.3235928634888871e-05, 'samples': 23043584, 'steps': 45006, 'loss/train': 1.434892177581787}
+02/26/2022 11:53:19 - INFO - codeparrot_training - Step 45007: {'lr': 1.3230675685979077e-05, 'samples': 23044096, 'steps': 45007, 'loss/train': 1.694848656654358}
+02/26/2022 11:53:25 - INFO - codeparrot_training - Step 45008: {'lr': 1.3225423751313942e-05, 'samples': 23044608, 'steps': 45008, 'loss/train': 1.8372126817703247}
+02/26/2022 11:53:28 - INFO - codeparrot_training - Step 45009: {'lr': 1.3220172830915866e-05, 'samples': 23045120, 'steps': 45009, 'loss/train': 1.0034587383270264}
+02/26/2022 11:53:34 - INFO - codeparrot_training - Step 45010: {'lr': 1.3214922924807471e-05, 'samples': 23045632, 'steps': 45010, 'loss/train': 1.565590500831604}
+02/26/2022 11:53:37 - INFO - codeparrot_training - Step 45011: {'lr': 1.3209674033011072e-05, 'samples': 23046144, 'steps': 45011, 'loss/train': 1.5464892387390137}
+02/26/2022 11:53:43 - INFO - codeparrot_training - Step 45012: {'lr': 1.3204426155549259e-05, 'samples': 23046656, 'steps': 45012, 'loss/train': 0.4041423797607422}
+02/26/2022 11:53:46 - INFO - codeparrot_training - Step 45013: {'lr': 1.3199179292444491e-05, 'samples': 23047168, 'steps': 45013, 'loss/train': 1.8522764444351196}
+02/26/2022 11:53:52 - INFO - codeparrot_training - Step 45014: {'lr': 1.3193933443719274e-05, 'samples': 23047680, 'steps': 45014, 'loss/train': 0.8630092740058899}
+02/26/2022 11:53:56 - INFO - codeparrot_training - Step 45015: {'lr': 1.3188688609396011e-05, 'samples': 23048192, 'steps': 45015, 'loss/train': 1.7922227382659912}
+02/26/2022 11:54:01 - INFO - codeparrot_training - Step 45016: {'lr': 1.3183444789497262e-05, 'samples': 23048704, 'steps': 45016, 'loss/train': 1.0154199600219727}
+02/26/2022 11:54:05 - INFO - codeparrot_training - Step 45017: {'lr': 1.3178201984045373e-05, 'samples': 23049216, 'steps': 45017, 'loss/train': 0.5893858671188354}
+02/26/2022 11:54:10 - INFO - codeparrot_training - Step 45018: {'lr': 1.317296019306291e-05, 'samples': 23049728, 'steps': 45018, 'loss/train': 0.25592130422592163}
+02/26/2022 11:54:14 - INFO - codeparrot_training - Step 45019: {'lr': 1.3167719416572272e-05, 'samples': 23050240, 'steps': 45019, 'loss/train': 1.3772392272949219}
+02/26/2022 11:54:19 - INFO - codeparrot_training - Step 45020: {'lr': 1.3162479654595938e-05, 'samples': 23050752, 'steps': 45020, 'loss/train': 2.304441452026367}
+02/26/2022 11:54:23 - INFO - codeparrot_training - Step 45021: {'lr': 1.315724090715631e-05, 'samples': 23051264, 'steps': 45021, 'loss/train': 1.3221042156219482}
+02/26/2022 11:54:28 - INFO - codeparrot_training - Step 45022: {'lr': 1.3152003174275867e-05, 'samples': 23051776, 'steps': 45022, 'loss/train': 1.5573818683624268}
+02/26/2022 11:54:32 - INFO - codeparrot_training - Step 45023: {'lr': 1.3146766455977066e-05, 'samples': 23052288, 'steps': 45023, 'loss/train': 1.6099047660827637}
+02/26/2022 11:54:38 - INFO - codeparrot_training - Step 45024: {'lr': 1.3141530752282276e-05, 'samples': 23052800, 'steps': 45024, 'loss/train': 1.2864468097686768}
+02/26/2022 11:54:42 - INFO - codeparrot_training - Step 45025: {'lr': 1.3136296063213954e-05, 'samples': 23053312, 'steps': 45025, 'loss/train': 1.70708429813385}
+02/26/2022 11:54:47 - INFO - codeparrot_training - Step 45026: {'lr': 1.3131062388794523e-05, 'samples': 23053824, 'steps': 45026, 'loss/train': 1.533958911895752}
+02/26/2022 11:54:51 - INFO - codeparrot_training - Step 45027: {'lr': 1.312582972904644e-05, 'samples': 23054336, 'steps': 45027, 'loss/train': 2.4968132972717285}
+02/26/2022 11:54:56 - INFO - codeparrot_training - Step 45028: {'lr': 1.3120598083992074e-05, 'samples': 23054848, 'steps': 45028, 'loss/train': 2.053069591522217}
+02/26/2022 11:55:00 - INFO - codeparrot_training - Step 45029: {'lr': 1.3115367453653799e-05, 'samples': 23055360, 'steps': 45029, 'loss/train': 1.4023457765579224}
+02/26/2022 11:55:05 - INFO - codeparrot_training - Step 45030: {'lr': 1.311013783805412e-05, 'samples': 23055872, 'steps': 45030, 'loss/train': 1.0853075981140137}
+02/26/2022 11:55:09 - INFO - codeparrot_training - Step 45031: {'lr': 1.3104909237215356e-05, 'samples': 23056384, 'steps': 45031, 'loss/train': 2.683966875076294}
+02/26/2022 11:55:14 - INFO - codeparrot_training - Step 45032: {'lr': 1.3099681651160018e-05, 'samples': 23056896, 'steps': 45032, 'loss/train': 8.609257698059082}
+02/26/2022 11:55:18 - INFO - codeparrot_training - Step 45033: {'lr': 1.3094455079910333e-05, 'samples': 23057408, 'steps': 45033, 'loss/train': 2.4009974002838135}
+02/26/2022 11:55:24 - INFO - codeparrot_training - Step 45034: {'lr': 1.3089229523488816e-05, 'samples': 23057920, 'steps': 45034, 'loss/train': 0.835014820098877}
+02/26/2022 11:55:28 - INFO - codeparrot_training - Step 45035: {'lr': 1.308400498191778e-05, 'samples': 23058432, 'steps': 45035, 'loss/train': 0.5543079376220703}
+02/26/2022 11:55:33 - INFO - codeparrot_training - Step 45036: {'lr': 1.3078781455219708e-05, 'samples': 23058944, 'steps': 45036, 'loss/train': 2.108165979385376}
+02/26/2022 11:55:37 - INFO - codeparrot_training - Step 45037: {'lr': 1.307355894341683e-05, 'samples': 23059456, 'steps': 45037, 'loss/train': 2.0037920475006104}
+02/26/2022 11:55:42 - INFO - codeparrot_training - Step 45038: {'lr': 1.306833744653163e-05, 'samples': 23059968, 'steps': 45038, 'loss/train': 2.6731386184692383}
+02/26/2022 11:55:46 - INFO - codeparrot_training - Step 45039: {'lr': 1.3063116964586425e-05, 'samples': 23060480, 'steps': 45039, 'loss/train': 2.332934856414795}
+02/26/2022 11:55:51 - INFO - codeparrot_training - Step 45040: {'lr': 1.305789749760361e-05, 'samples': 23060992, 'steps': 45040, 'loss/train': 2.214817523956299}
+02/26/2022 11:55:55 - INFO - codeparrot_training - Step 45041: {'lr': 1.3052679045605476e-05, 'samples': 23061504, 'steps': 45041, 'loss/train': 2.0601751804351807}
+02/26/2022 11:56:00 - INFO - codeparrot_training - Step 45042: {'lr': 1.3047461608614475e-05, 'samples': 23062016, 'steps': 45042, 'loss/train': 1.4372636079788208}
+02/26/2022 11:56:04 - INFO - codeparrot_training - Step 45043: {'lr': 1.3042245186652895e-05, 'samples': 23062528, 'steps': 45043, 'loss/train': 2.499464511871338}
+02/26/2022 11:56:10 - INFO - codeparrot_training - Step 45044: {'lr': 1.3037029779743054e-05, 'samples': 23063040, 'steps': 45044, 'loss/train': 2.0009520053863525}
+02/26/2022 11:56:13 - INFO - codeparrot_training - Step 45045: {'lr': 1.3031815387907375e-05, 'samples': 23063552, 'steps': 45045, 'loss/train': 1.4358735084533691}
+02/26/2022 11:56:19 - INFO - codeparrot_training - Step 45046: {'lr': 1.3026602011168176e-05, 'samples': 23064064, 'steps': 45046, 'loss/train': 1.6239103078842163}
+02/26/2022 11:56:22 - INFO - codeparrot_training - Step 45047: {'lr': 1.3021389649547743e-05, 'samples': 23064576, 'steps': 45047, 'loss/train': 1.9477182626724243}
+02/26/2022 11:56:28 - INFO - codeparrot_training - Step 45048: {'lr': 1.3016178303068393e-05, 'samples': 23065088, 'steps': 45048, 'loss/train': 1.7775816917419434}
+02/26/2022 11:56:31 - INFO - codeparrot_training - Step 45049: {'lr': 1.3010967971752524e-05, 'samples': 23065600, 'steps': 45049, 'loss/train': 0.8595528602600098}
+02/26/2022 11:56:37 - INFO - codeparrot_training - Step 45050: {'lr': 1.3005758655622424e-05, 'samples': 23066112, 'steps': 45050, 'loss/train': 1.3412004709243774}
+02/26/2022 11:56:40 - INFO - codeparrot_training - Step 45051: {'lr': 1.300055035470038e-05, 'samples': 23066624, 'steps': 45051, 'loss/train': 0.6212685704231262}
+02/26/2022 11:56:46 - INFO - codeparrot_training - Step 45052: {'lr': 1.2995343069008708e-05, 'samples': 23067136, 'steps': 45052, 'loss/train': 1.173230528831482}
+02/26/2022 11:56:49 - INFO - codeparrot_training - Step 45053: {'lr': 1.2990136798569752e-05, 'samples': 23067648, 'steps': 45053, 'loss/train': 1.4153577089309692}
+02/26/2022 11:56:55 - INFO - codeparrot_training - Step 45054: {'lr': 1.2984931543405771e-05, 'samples': 23068160, 'steps': 45054, 'loss/train': 2.286909580230713}
+02/26/2022 11:56:59 - INFO - codeparrot_training - Step 45055: {'lr': 1.2979727303539108e-05, 'samples': 23068672, 'steps': 45055, 'loss/train': 1.2688261270523071}
+02/26/2022 11:57:04 - INFO - codeparrot_training - Step 45056: {'lr': 1.2974524078991995e-05, 'samples': 23069184, 'steps': 45056, 'loss/train': 2.0816214084625244}
+02/26/2022 11:57:08 - INFO - codeparrot_training - Step 45057: {'lr': 1.2969321869786777e-05, 'samples': 23069696, 'steps': 45057, 'loss/train': 1.7652606964111328}
+02/26/2022 11:57:14 - INFO - codeparrot_training - Step 45058: {'lr': 1.296412067594574e-05, 'samples': 23070208, 'steps': 45058, 'loss/train': 1.4285666942596436}
+02/26/2022 11:57:17 - INFO - codeparrot_training - Step 45059: {'lr': 1.2958920497491144e-05, 'samples': 23070720, 'steps': 45059, 'loss/train': 0.5469043850898743}
+02/26/2022 11:57:23 - INFO - codeparrot_training - Step 45060: {'lr': 1.2953721334445196e-05, 'samples': 23071232, 'steps': 45060, 'loss/train': 1.449249505996704}
+02/26/2022 11:57:26 - INFO - codeparrot_training - Step 45061: {'lr': 1.294852318683029e-05, 'samples': 23071744, 'steps': 45061, 'loss/train': 1.708475947380066}
+02/26/2022 11:57:32 - INFO - codeparrot_training - Step 45062: {'lr': 1.2943326054668663e-05, 'samples': 23072256, 'steps': 45062, 'loss/train': 2.487797737121582}
+02/26/2022 11:57:35 - INFO - codeparrot_training - Step 45063: {'lr': 1.2938129937982519e-05, 'samples': 23072768, 'steps': 45063, 'loss/train': 2.5748367309570312}
+02/26/2022 11:57:40 - INFO - codeparrot_training - Step 45064: {'lr': 1.2932934836794114e-05, 'samples': 23073280, 'steps': 45064, 'loss/train': 2.007824659347534}
+02/26/2022 11:57:44 - INFO - codeparrot_training - Step 45065: {'lr': 1.2927740751125795e-05, 'samples': 23073792, 'steps': 45065, 'loss/train': 1.6610277891159058}
+02/26/2022 11:57:50 - INFO - codeparrot_training - Step 45066: {'lr': 1.2922547680999708e-05, 'samples': 23074304, 'steps': 45066, 'loss/train': 2.152963638305664}
+02/26/2022 11:57:53 - INFO - codeparrot_training - Step 45067: {'lr': 1.2917355626438227e-05, 'samples': 23074816, 'steps': 45067, 'loss/train': 1.9496700763702393}
+02/26/2022 11:57:59 - INFO - codeparrot_training - Step 45068: {'lr': 1.2912164587463442e-05, 'samples': 23075328, 'steps': 45068, 'loss/train': 1.4794533252716064}
+02/26/2022 11:58:02 - INFO - codeparrot_training - Step 45069: {'lr': 1.2906974564097674e-05, 'samples': 23075840, 'steps': 45069, 'loss/train': 1.267409324645996}
+02/26/2022 11:58:08 - INFO - codeparrot_training - Step 45070: {'lr': 1.2901785556363122e-05, 'samples': 23076352, 'steps': 45070, 'loss/train': 0.6815871596336365}
+02/26/2022 11:58:11 - INFO - codeparrot_training - Step 45071: {'lr': 1.2896597564282104e-05, 'samples': 23076864, 'steps': 45071, 'loss/train': 2.6047310829162598}
+02/26/2022 11:58:18 - INFO - codeparrot_training - Step 45072: {'lr': 1.2891410587876712e-05, 'samples': 23077376, 'steps': 45072, 'loss/train': 2.4416463375091553}
+02/26/2022 11:58:21 - INFO - codeparrot_training - Step 45073: {'lr': 1.2886224627169235e-05, 'samples': 23077888, 'steps': 45073, 'loss/train': 1.580350637435913}
+02/26/2022 11:58:27 - INFO - codeparrot_training - Step 45074: {'lr': 1.2881039682181878e-05, 'samples': 23078400, 'steps': 45074, 'loss/train': 2.1410484313964844}
+02/26/2022 11:58:30 - INFO - codeparrot_training - Step 45075: {'lr': 1.28758557529369e-05, 'samples': 23078912, 'steps': 45075, 'loss/train': 2.637892484664917}
+02/26/2022 11:58:36 - INFO - codeparrot_training - Step 45076: {'lr': 1.2870672839456394e-05, 'samples': 23079424, 'steps': 45076, 'loss/train': 1.7528022527694702}
+02/26/2022 11:58:40 - INFO - codeparrot_training - Step 45077: {'lr': 1.2865490941762676e-05, 'samples': 23079936, 'steps': 45077, 'loss/train': 2.098456621170044}
+02/26/2022 11:58:46 - INFO - codeparrot_training - Step 45078: {'lr': 1.2860310059877838e-05, 'samples': 23080448, 'steps': 45078, 'loss/train': 2.971890926361084}
+02/26/2022 11:58:49 - INFO - codeparrot_training - Step 45079: {'lr': 1.2855130193824171e-05, 'samples': 23080960, 'steps': 45079, 'loss/train': 1.1212985515594482}
+02/26/2022 11:58:56 - INFO - codeparrot_training - Step 45080: {'lr': 1.284995134362385e-05, 'samples': 23081472, 'steps': 45080, 'loss/train': 1.6538738012313843}
+02/26/2022 11:58:59 - INFO - codeparrot_training - Step 45081: {'lr': 1.2844773509299024e-05, 'samples': 23081984, 'steps': 45081, 'loss/train': 2.364436149597168}
+02/26/2022 11:59:05 - INFO - codeparrot_training - Step 45082: {'lr': 1.2839596690871841e-05, 'samples': 23082496, 'steps': 45082, 'loss/train': 1.856343150138855}
+02/26/2022 11:59:08 - INFO - codeparrot_training - Step 45083: {'lr': 1.2834420888364535e-05, 'samples': 23083008, 'steps': 45083, 'loss/train': 2.1124069690704346}
+02/26/2022 11:59:14 - INFO - codeparrot_training - Step 45084: {'lr': 1.282924610179928e-05, 'samples': 23083520, 'steps': 45084, 'loss/train': 1.8365552425384521}
+02/26/2022 11:59:17 - INFO - codeparrot_training - Step 45085: {'lr': 1.2824072331198227e-05, 'samples': 23084032, 'steps': 45085, 'loss/train': 1.9485359191894531}
+02/26/2022 11:59:23 - INFO - codeparrot_training - Step 45086: {'lr': 1.2818899576583526e-05, 'samples': 23084544, 'steps': 45086, 'loss/train': 0.6901110410690308}
+02/26/2022 11:59:26 - INFO - codeparrot_training - Step 45087: {'lr': 1.2813727837977324e-05, 'samples': 23085056, 'steps': 45087, 'loss/train': 1.8562246561050415}
+02/26/2022 11:59:32 - INFO - codeparrot_training - Step 45088: {'lr': 1.2808557115401826e-05, 'samples': 23085568, 'steps': 45088, 'loss/train': 1.5247613191604614}
+02/26/2022 11:59:35 - INFO - codeparrot_training - Step 45089: {'lr': 1.2803387408879153e-05, 'samples': 23086080, 'steps': 45089, 'loss/train': 2.16066837310791}
+02/26/2022 11:59:42 - INFO - codeparrot_training - Step 45090: {'lr': 1.2798218718431453e-05, 'samples': 23086592, 'steps': 45090, 'loss/train': 1.5529731512069702}
+02/26/2022 11:59:45 - INFO - codeparrot_training - Step 45091: {'lr': 1.2793051044080822e-05, 'samples': 23087104, 'steps': 45091, 'loss/train': 1.941236972808838}
+02/26/2022 11:59:51 - INFO - codeparrot_training - Step 45092: {'lr': 1.278788438584949e-05, 'samples': 23087616, 'steps': 45092, 'loss/train': 5.634986877441406}
+02/26/2022 11:59:54 - INFO - codeparrot_training - Step 45093: {'lr': 1.2782718743759525e-05, 'samples': 23088128, 'steps': 45093, 'loss/train': 2.2821693420410156}
+02/26/2022 11:59:58 - INFO - codeparrot_training - Step 45094: {'lr': 1.2777554117833073e-05, 'samples': 23088640, 'steps': 45094, 'loss/train': 1.4865084886550903}
+02/26/2022 12:00:04 - INFO - codeparrot_training - Step 45095: {'lr': 1.2772390508092203e-05, 'samples': 23089152, 'steps': 45095, 'loss/train': 1.6523545980453491}
+02/26/2022 12:00:07 - INFO - codeparrot_training - Step 45096: {'lr': 1.2767227914559116e-05, 'samples': 23089664, 'steps': 45096, 'loss/train': 1.8405457735061646}
+02/26/2022 12:00:13 - INFO - codeparrot_training - Step 45097: {'lr': 1.2762066337255907e-05, 'samples': 23090176, 'steps': 45097, 'loss/train': 2.533374547958374}
+02/26/2022 12:00:16 - INFO - codeparrot_training - Step 45098: {'lr': 1.2756905776204698e-05, 'samples': 23090688, 'steps': 45098, 'loss/train': 0.31162989139556885}
+02/26/2022 12:00:22 - INFO - codeparrot_training - Step 45099: {'lr': 1.2751746231427497e-05, 'samples': 23091200, 'steps': 45099, 'loss/train': 1.6668719053268433}
+02/26/2022 12:00:27 - INFO - codeparrot_training - Step 45100: {'lr': 1.2746587702946538e-05, 'samples': 23091712, 'steps': 45100, 'loss/train': 2.486804485321045}
+02/26/2022 12:00:31 - INFO - codeparrot_training - Step 45101: {'lr': 1.2741430190783804e-05, 'samples': 23092224, 'steps': 45101, 'loss/train': 2.341907262802124}
+02/26/2022 12:00:34 - INFO - codeparrot_training - Step 45102: {'lr': 1.2736273694961553e-05, 'samples': 23092736, 'steps': 45102, 'loss/train': 2.0972723960876465}
+02/26/2022 12:00:41 - INFO - codeparrot_training - Step 45103: {'lr': 1.2731118215501658e-05, 'samples': 23093248, 'steps': 45103, 'loss/train': 0.5704176425933838}
+02/26/2022 12:00:44 - INFO - codeparrot_training - Step 45104: {'lr': 1.2725963752426379e-05, 'samples': 23093760, 'steps': 45104, 'loss/train': 0.40021708607673645}
+02/26/2022 12:00:49 - INFO - codeparrot_training - Step 45105: {'lr': 1.2720810305757668e-05, 'samples': 23094272, 'steps': 45105, 'loss/train': 1.5304100513458252}
+02/26/2022 12:00:53 - INFO - codeparrot_training - Step 45106: {'lr': 1.2715657875517733e-05, 'samples': 23094784, 'steps': 45106, 'loss/train': 2.3171169757843018}
+02/26/2022 12:00:59 - INFO - codeparrot_training - Step 45107: {'lr': 1.27105064617285e-05, 'samples': 23095296, 'steps': 45107, 'loss/train': 2.572324752807617}
+02/26/2022 12:01:04 - INFO - codeparrot_training - Step 45108: {'lr': 1.2705356064412172e-05, 'samples': 23095808, 'steps': 45108, 'loss/train': 1.3853480815887451}
+02/26/2022 12:01:07 - INFO - codeparrot_training - Step 45109: {'lr': 1.2700206683590676e-05, 'samples': 23096320, 'steps': 45109, 'loss/train': 1.732122540473938}
+02/26/2022 12:01:13 - INFO - codeparrot_training - Step 45110: {'lr': 1.2695058319286245e-05, 'samples': 23096832, 'steps': 45110, 'loss/train': 2.681973695755005}
+02/26/2022 12:01:16 - INFO - codeparrot_training - Step 45111: {'lr': 1.2689910971520752e-05, 'samples': 23097344, 'steps': 45111, 'loss/train': 1.8421803712844849}
+02/26/2022 12:01:22 - INFO - codeparrot_training - Step 45112: {'lr': 1.2684764640316343e-05, 'samples': 23097856, 'steps': 45112, 'loss/train': 0.5336638689041138}
+02/26/2022 12:01:25 - INFO - codeparrot_training - Step 45113: {'lr': 1.267961932569503e-05, 'samples': 23098368, 'steps': 45113, 'loss/train': 1.389957070350647}
+02/26/2022 12:01:31 - INFO - codeparrot_training - Step 45114: {'lr': 1.2674475027678906e-05, 'samples': 23098880, 'steps': 45114, 'loss/train': 1.7390813827514648}
+02/26/2022 12:01:34 - INFO - codeparrot_training - Step 45115: {'lr': 1.266933174628998e-05, 'samples': 23099392, 'steps': 45115, 'loss/train': 1.7608330249786377}
+02/26/2022 12:01:41 - INFO - codeparrot_training - Step 45116: {'lr': 1.2664189481550236e-05, 'samples': 23099904, 'steps': 45116, 'loss/train': 1.451500415802002}
+02/26/2022 12:01:44 - INFO - codeparrot_training - Step 45117: {'lr': 1.265904823348174e-05, 'samples': 23100416, 'steps': 45117, 'loss/train': 2.1356723308563232}
+02/26/2022 12:01:50 - INFO - codeparrot_training - Step 45118: {'lr': 1.2653908002106557e-05, 'samples': 23100928, 'steps': 45118, 'loss/train': 1.1042168140411377}
+02/26/2022 12:01:53 - INFO - codeparrot_training - Step 45119: {'lr': 1.264876878744664e-05, 'samples': 23101440, 'steps': 45119, 'loss/train': 2.336280107498169}
+02/26/2022 12:01:59 - INFO - codeparrot_training - Step 45120: {'lr': 1.2643630589524058e-05, 'samples': 23101952, 'steps': 45120, 'loss/train': 1.5264019966125488}
+02/26/2022 12:02:02 - INFO - codeparrot_training - Step 45121: {'lr': 1.2638493408360735e-05, 'samples': 23102464, 'steps': 45121, 'loss/train': 1.165435791015625}
+02/26/2022 12:02:08 - INFO - codeparrot_training - Step 45122: {'lr': 1.2633357243978793e-05, 'samples': 23102976, 'steps': 45122, 'loss/train': 2.081279754638672}
+02/26/2022 12:02:11 - INFO - codeparrot_training - Step 45123: {'lr': 1.262822209640016e-05, 'samples': 23103488, 'steps': 45123, 'loss/train': 0.5477192997932434}
+02/26/2022 12:02:17 - INFO - codeparrot_training - Step 45124: {'lr': 1.2623087965646846e-05, 'samples': 23104000, 'steps': 45124, 'loss/train': 1.7766499519348145}
+02/26/2022 12:02:20 - INFO - codeparrot_training - Step 45125: {'lr': 1.2617954851740832e-05, 'samples': 23104512, 'steps': 45125, 'loss/train': 1.2019747495651245}
+02/26/2022 12:02:26 - INFO - codeparrot_training - Step 45126: {'lr': 1.2612822754704129e-05, 'samples': 23105024, 'steps': 45126, 'loss/train': 1.513781189918518}
+02/26/2022 12:02:29 - INFO - codeparrot_training - Step 45127: {'lr': 1.2607691674558747e-05, 'samples': 23105536, 'steps': 45127, 'loss/train': 0.7065152525901794}
+02/26/2022 12:02:36 - INFO - codeparrot_training - Step 45128: {'lr': 1.2602561611326613e-05, 'samples': 23106048, 'steps': 45128, 'loss/train': 1.7598798274993896}
+02/26/2022 12:02:39 - INFO - codeparrot_training - Step 45129: {'lr': 1.2597432565029682e-05, 'samples': 23106560, 'steps': 45129, 'loss/train': 0.29196837544441223}
+02/26/2022 12:02:45 - INFO - codeparrot_training - Step 45130: {'lr': 1.2592304535690018e-05, 'samples': 23107072, 'steps': 45130, 'loss/train': 1.2039690017700195}
+02/26/2022 12:02:48 - INFO - codeparrot_training - Step 45131: {'lr': 1.2587177523329524e-05, 'samples': 23107584, 'steps': 45131, 'loss/train': 1.0128332376480103}
+02/26/2022 12:02:54 - INFO - codeparrot_training - Step 45132: {'lr': 1.2582051527970178e-05, 'samples': 23108096, 'steps': 45132, 'loss/train': 2.611274242401123}
+02/26/2022 12:02:57 - INFO - codeparrot_training - Step 45133: {'lr': 1.257692654963391e-05, 'samples': 23108608, 'steps': 45133, 'loss/train': 2.5898587703704834}
+02/26/2022 12:03:03 - INFO - codeparrot_training - Step 45134: {'lr': 1.2571802588342728e-05, 'samples': 23109120, 'steps': 45134, 'loss/train': 2.2823262214660645}
+02/26/2022 12:03:06 - INFO - codeparrot_training - Step 45135: {'lr': 1.2566679644118534e-05, 'samples': 23109632, 'steps': 45135, 'loss/train': 1.3068331480026245}
+02/26/2022 12:03:12 - INFO - codeparrot_training - Step 45136: {'lr': 1.2561557716983307e-05, 'samples': 23110144, 'steps': 45136, 'loss/train': 0.837346613407135}
+02/26/2022 12:03:15 - INFO - codeparrot_training - Step 45137: {'lr': 1.2556436806958976e-05, 'samples': 23110656, 'steps': 45137, 'loss/train': 3.0111570358276367}
+02/26/2022 12:03:19 - INFO - codeparrot_training - Step 45138: {'lr': 1.255131691406744e-05, 'samples': 23111168, 'steps': 45138, 'loss/train': 0.5239671468734741}
+02/26/2022 12:03:25 - INFO - codeparrot_training - Step 45139: {'lr': 1.254619803833068e-05, 'samples': 23111680, 'steps': 45139, 'loss/train': 2.8919575214385986}
+02/26/2022 12:03:28 - INFO - codeparrot_training - Step 45140: {'lr': 1.2541080179770569e-05, 'samples': 23112192, 'steps': 45140, 'loss/train': 1.582980751991272}
+02/26/2022 12:03:34 - INFO - codeparrot_training - Step 45141: {'lr': 1.2535963338409173e-05, 'samples': 23112704, 'steps': 45141, 'loss/train': 0.9383519291877747}
+02/26/2022 12:03:37 - INFO - codeparrot_training - Step 45142: {'lr': 1.2530847514268196e-05, 'samples': 23113216, 'steps': 45142, 'loss/train': 0.8160671591758728}
+02/26/2022 12:03:43 - INFO - codeparrot_training - Step 45143: {'lr': 1.252573270736973e-05, 'samples': 23113728, 'steps': 45143, 'loss/train': 1.442929983139038}
+02/26/2022 12:03:49 - INFO - codeparrot_training - Step 45144: {'lr': 1.2520618917735566e-05, 'samples': 23114240, 'steps': 45144, 'loss/train': 1.965590000152588}
+02/26/2022 12:03:52 - INFO - codeparrot_training - Step 45145: {'lr': 1.2515506145387739e-05, 'samples': 23114752, 'steps': 45145, 'loss/train': 0.45528197288513184}
+02/26/2022 12:03:58 - INFO - codeparrot_training - Step 45146: {'lr': 1.2510394390348012e-05, 'samples': 23115264, 'steps': 45146, 'loss/train': 2.4050965309143066}
+02/26/2022 12:04:01 - INFO - codeparrot_training - Step 45147: {'lr': 1.2505283652638366e-05, 'samples': 23115776, 'steps': 45147, 'loss/train': 1.8518065214157104}
+02/26/2022 12:04:07 - INFO - codeparrot_training - Step 45148: {'lr': 1.2500173932280646e-05, 'samples': 23116288, 'steps': 45148, 'loss/train': 2.4183380603790283}
+02/26/2022 12:04:11 - INFO - codeparrot_training - Step 45149: {'lr': 1.249506522929683e-05, 'samples': 23116800, 'steps': 45149, 'loss/train': 2.1279654502868652}
+02/26/2022 12:04:16 - INFO - codeparrot_training - Step 45150: {'lr': 1.2489957543708658e-05, 'samples': 23117312, 'steps': 45150, 'loss/train': 1.2020108699798584}
+02/26/2022 12:04:20 - INFO - codeparrot_training - Step 45151: {'lr': 1.2484850875538135e-05, 'samples': 23117824, 'steps': 45151, 'loss/train': 1.4040043354034424}
+02/26/2022 12:04:25 - INFO - codeparrot_training - Step 45152: {'lr': 1.2479745224807049e-05, 'samples': 23118336, 'steps': 45152, 'loss/train': 2.096092462539673}
+02/26/2022 12:04:29 - INFO - codeparrot_training - Step 45153: {'lr': 1.2474640591537356e-05, 'samples': 23118848, 'steps': 45153, 'loss/train': 2.048578977584839}
+02/26/2022 12:04:34 - INFO - codeparrot_training - Step 45154: {'lr': 1.24695369757509e-05, 'samples': 23119360, 'steps': 45154, 'loss/train': 1.5250835418701172}
+02/26/2022 12:04:38 - INFO - codeparrot_training - Step 45155: {'lr': 1.2464434377469496e-05, 'samples': 23119872, 'steps': 45155, 'loss/train': 0.7809669375419617}
+02/26/2022 12:04:43 - INFO - codeparrot_training - Step 45156: {'lr': 1.245933279671499e-05, 'samples': 23120384, 'steps': 45156, 'loss/train': 2.1010615825653076}
+02/26/2022 12:04:47 - INFO - codeparrot_training - Step 45157: {'lr': 1.2454232233509332e-05, 'samples': 23120896, 'steps': 45157, 'loss/train': 1.7512787580490112}
+02/26/2022 12:04:52 - INFO - codeparrot_training - Step 45158: {'lr': 1.2449132687874315e-05, 'samples': 23121408, 'steps': 45158, 'loss/train': 1.8734853267669678}
+02/26/2022 12:04:56 - INFO - codeparrot_training - Step 45159: {'lr': 1.2444034159831752e-05, 'samples': 23121920, 'steps': 45159, 'loss/train': 1.7184488773345947}
+02/26/2022 12:05:01 - INFO - codeparrot_training - Step 45160: {'lr': 1.2438936649403515e-05, 'samples': 23122432, 'steps': 45160, 'loss/train': 1.4589784145355225}
+02/26/2022 12:05:05 - INFO - codeparrot_training - Step 45161: {'lr': 1.2433840156611448e-05, 'samples': 23122944, 'steps': 45161, 'loss/train': 2.0252017974853516}
+02/26/2022 12:05:11 - INFO - codeparrot_training - Step 45162: {'lr': 1.242874468147734e-05, 'samples': 23123456, 'steps': 45162, 'loss/train': 1.729162573814392}
+02/26/2022 12:05:14 - INFO - codeparrot_training - Step 45163: {'lr': 1.2423650224023143e-05, 'samples': 23123968, 'steps': 45163, 'loss/train': 2.562262535095215}
+02/26/2022 12:05:20 - INFO - codeparrot_training - Step 45164: {'lr': 1.2418556784270508e-05, 'samples': 23124480, 'steps': 45164, 'loss/train': 1.890058994293213}
+02/26/2022 12:05:23 - INFO - codeparrot_training - Step 45165: {'lr': 1.2413464362241362e-05, 'samples': 23124992, 'steps': 45165, 'loss/train': 0.8579510450363159}
+02/26/2022 12:05:29 - INFO - codeparrot_training - Step 45166: {'lr': 1.2408372957957465e-05, 'samples': 23125504, 'steps': 45166, 'loss/train': 0.8872958421707153}
+02/26/2022 12:05:32 - INFO - codeparrot_training - Step 45167: {'lr': 1.2403282571440716e-05, 'samples': 23126016, 'steps': 45167, 'loss/train': 1.5252865552902222}
+02/26/2022 12:05:38 - INFO - codeparrot_training - Step 45168: {'lr': 1.2398193202712821e-05, 'samples': 23126528, 'steps': 45168, 'loss/train': 1.1240007877349854}
+02/26/2022 12:05:41 - INFO - codeparrot_training - Step 45169: {'lr': 1.2393104851795623e-05, 'samples': 23127040, 'steps': 45169, 'loss/train': 1.2660431861877441}
+02/26/2022 12:05:47 - INFO - codeparrot_training - Step 45170: {'lr': 1.2388017518710909e-05, 'samples': 23127552, 'steps': 45170, 'loss/train': 2.0403618812561035}
+02/26/2022 12:05:50 - INFO - codeparrot_training - Step 45171: {'lr': 1.2382931203480497e-05, 'samples': 23128064, 'steps': 45171, 'loss/train': 1.2678524255752563}
+02/26/2022 12:05:57 - INFO - codeparrot_training - Step 45172: {'lr': 1.2377845906126118e-05, 'samples': 23128576, 'steps': 45172, 'loss/train': 1.1919927597045898}
+02/26/2022 12:06:00 - INFO - codeparrot_training - Step 45173: {'lr': 1.2372761626669616e-05, 'samples': 23129088, 'steps': 45173, 'loss/train': 1.4092915058135986}
+02/26/2022 12:06:05 - INFO - codeparrot_training - Step 45174: {'lr': 1.2367678365132778e-05, 'samples': 23129600, 'steps': 45174, 'loss/train': 1.9317097663879395}
+02/26/2022 12:06:09 - INFO - codeparrot_training - Step 45175: {'lr': 1.2362596121537285e-05, 'samples': 23130112, 'steps': 45175, 'loss/train': 2.4590182304382324}
+02/26/2022 12:06:14 - INFO - codeparrot_training - Step 45176: {'lr': 1.2357514895905003e-05, 'samples': 23130624, 'steps': 45176, 'loss/train': 2.421562433242798}
+02/26/2022 12:06:18 - INFO - codeparrot_training - Step 45177: {'lr': 1.2352434688257697e-05, 'samples': 23131136, 'steps': 45177, 'loss/train': 1.9065955877304077}
+02/26/2022 12:06:23 - INFO - codeparrot_training - Step 45178: {'lr': 1.234735549861707e-05, 'samples': 23131648, 'steps': 45178, 'loss/train': 2.655139446258545}
+02/26/2022 12:06:27 - INFO - codeparrot_training - Step 45179: {'lr': 1.234227732700488e-05, 'samples': 23132160, 'steps': 45179, 'loss/train': 0.8640618920326233}
+02/26/2022 12:06:32 - INFO - codeparrot_training - Step 45180: {'lr': 1.2337200173442948e-05, 'samples': 23132672, 'steps': 45180, 'loss/train': 2.1679911613464355}
+02/26/2022 12:06:36 - INFO - codeparrot_training - Step 45181: {'lr': 1.2332124037953003e-05, 'samples': 23133184, 'steps': 45181, 'loss/train': 1.7762260437011719}
+02/26/2022 12:06:42 - INFO - codeparrot_training - Step 45182: {'lr': 1.2327048920556749e-05, 'samples': 23133696, 'steps': 45182, 'loss/train': 3.0028209686279297}
+02/26/2022 12:06:45 - INFO - codeparrot_training - Step 45183: {'lr': 1.2321974821275923e-05, 'samples': 23134208, 'steps': 45183, 'loss/train': 0.267306387424469}
+02/26/2022 12:06:51 - INFO - codeparrot_training - Step 45184: {'lr': 1.2316901740132336e-05, 'samples': 23134720, 'steps': 45184, 'loss/train': 2.4571051597595215}
+02/26/2022 12:06:55 - INFO - codeparrot_training - Step 45185: {'lr': 1.2311829677147612e-05, 'samples': 23135232, 'steps': 45185, 'loss/train': 1.403219223022461}
+02/26/2022 12:07:00 - INFO - codeparrot_training - Step 45186: {'lr': 1.2306758632343567e-05, 'samples': 23135744, 'steps': 45186, 'loss/train': 1.4392234086990356}
+02/26/2022 12:07:04 - INFO - codeparrot_training - Step 45187: {'lr': 1.2301688605741878e-05, 'samples': 23136256, 'steps': 45187, 'loss/train': 1.6896017789840698}
+02/26/2022 12:07:09 - INFO - codeparrot_training - Step 45188: {'lr': 1.2296619597364278e-05, 'samples': 23136768, 'steps': 45188, 'loss/train': 1.8735121488571167}
+02/26/2022 12:07:13 - INFO - codeparrot_training - Step 45189: {'lr': 1.22915516072325e-05, 'samples': 23137280, 'steps': 45189, 'loss/train': 1.7878131866455078}
+02/26/2022 12:07:18 - INFO - codeparrot_training - Step 45190: {'lr': 1.2286484635368244e-05, 'samples': 23137792, 'steps': 45190, 'loss/train': 0.8920719623565674}
+02/26/2022 12:07:22 - INFO - codeparrot_training - Step 45191: {'lr': 1.2281418681793167e-05, 'samples': 23138304, 'steps': 45191, 'loss/train': 1.6619768142700195}
+02/26/2022 12:07:27 - INFO - codeparrot_training - Step 45192: {'lr': 1.2276353746529023e-05, 'samples': 23138816, 'steps': 45192, 'loss/train': 1.7054188251495361}
+02/26/2022 12:07:31 - INFO - codeparrot_training - Step 45193: {'lr': 1.2271289829597493e-05, 'samples': 23139328, 'steps': 45193, 'loss/train': 1.158315658569336}
+02/26/2022 12:07:37 - INFO - codeparrot_training - Step 45194: {'lr': 1.226622693102028e-05, 'samples': 23139840, 'steps': 45194, 'loss/train': 1.4815367460250854}
+02/26/2022 12:07:40 - INFO - codeparrot_training - Step 45195: {'lr': 1.2261165050819034e-05, 'samples': 23140352, 'steps': 45195, 'loss/train': 1.7711548805236816}
+02/26/2022 12:07:46 - INFO - codeparrot_training - Step 45196: {'lr': 1.2256104189015516e-05, 'samples': 23140864, 'steps': 45196, 'loss/train': 1.2785003185272217}
+02/26/2022 12:07:49 - INFO - codeparrot_training - Step 45197: {'lr': 1.2251044345631319e-05, 'samples': 23141376, 'steps': 45197, 'loss/train': 0.9468376040458679}
+02/26/2022 12:07:55 - INFO - codeparrot_training - Step 45198: {'lr': 1.2245985520688175e-05, 'samples': 23141888, 'steps': 45198, 'loss/train': 1.7057024240493774}
+02/26/2022 12:07:58 - INFO - codeparrot_training - Step 45199: {'lr': 1.2240927714207706e-05, 'samples': 23142400, 'steps': 45199, 'loss/train': 1.9232268333435059}
+02/26/2022 12:08:04 - INFO - codeparrot_training - Step 45200: {'lr': 1.2235870926211617e-05, 'samples': 23142912, 'steps': 45200, 'loss/train': 1.2355166673660278}
+02/26/2022 12:08:08 - INFO - codeparrot_training - Step 45201: {'lr': 1.2230815156721532e-05, 'samples': 23143424, 'steps': 45201, 'loss/train': 1.7763829231262207}
+02/26/2022 12:08:14 - INFO - codeparrot_training - Step 45202: {'lr': 1.2225760405759207e-05, 'samples': 23143936, 'steps': 45202, 'loss/train': 2.1136059761047363}
+02/26/2022 12:08:17 - INFO - codeparrot_training - Step 45203: {'lr': 1.2220706673346154e-05, 'samples': 23144448, 'steps': 45203, 'loss/train': 2.2113077640533447}
+02/26/2022 12:08:21 - INFO - codeparrot_training - Step 45204: {'lr': 1.2215653959504136e-05, 'samples': 23144960, 'steps': 45204, 'loss/train': 1.0620841979980469}
+02/26/2022 12:08:26 - INFO - codeparrot_training - Step 45205: {'lr': 1.221060226425469e-05, 'samples': 23145472, 'steps': 45205, 'loss/train': 2.7467172145843506}
+02/26/2022 12:08:30 - INFO - codeparrot_training - Step 45206: {'lr': 1.2205551587619602e-05, 'samples': 23145984, 'steps': 45206, 'loss/train': 1.1453804969787598}
+02/26/2022 12:08:35 - INFO - codeparrot_training - Step 45207: {'lr': 1.2200501929620356e-05, 'samples': 23146496, 'steps': 45207, 'loss/train': 1.5322825908660889}
+02/26/2022 12:08:41 - INFO - codeparrot_training - Step 45208: {'lr': 1.2195453290278686e-05, 'samples': 23147008, 'steps': 45208, 'loss/train': 1.585524559020996}
+02/26/2022 12:08:44 - INFO - codeparrot_training - Step 45209: {'lr': 1.2190405669616157e-05, 'samples': 23147520, 'steps': 45209, 'loss/train': 1.4892983436584473}
+02/26/2022 12:08:50 - INFO - codeparrot_training - Step 45210: {'lr': 1.2185359067654417e-05, 'samples': 23148032, 'steps': 45210, 'loss/train': 1.5818986892700195}
+02/26/2022 12:08:54 - INFO - codeparrot_training - Step 45211: {'lr': 1.2180313484415117e-05, 'samples': 23148544, 'steps': 45211, 'loss/train': 1.7678420543670654}
+02/26/2022 12:08:59 - INFO - codeparrot_training - Step 45212: {'lr': 1.2175268919919823e-05, 'samples': 23149056, 'steps': 45212, 'loss/train': 3.5372023582458496}
+02/26/2022 12:09:03 - INFO - codeparrot_training - Step 45213: {'lr': 1.217022537419013e-05, 'samples': 23149568, 'steps': 45213, 'loss/train': 1.703425645828247}
+02/26/2022 12:09:08 - INFO - codeparrot_training - Step 45214: {'lr': 1.2165182847247713e-05, 'samples': 23150080, 'steps': 45214, 'loss/train': 0.3876635432243347}
+02/26/2022 12:09:12 - INFO - codeparrot_training - Step 45215: {'lr': 1.216014133911411e-05, 'samples': 23150592, 'steps': 45215, 'loss/train': 1.9977920055389404}
+02/26/2022 12:09:17 - INFO - codeparrot_training - Step 45216: {'lr': 1.2155100849810946e-05, 'samples': 23151104, 'steps': 45216, 'loss/train': 1.6331762075424194}
+02/26/2022 12:09:21 - INFO - codeparrot_training - Step 45217: {'lr': 1.2150061379359812e-05, 'samples': 23151616, 'steps': 45217, 'loss/train': 1.7448798418045044}
+02/26/2022 12:09:26 - INFO - codeparrot_training - Step 45218: {'lr': 1.2145022927782273e-05, 'samples': 23152128, 'steps': 45218, 'loss/train': 0.9834668636322021}
+02/26/2022 12:09:30 - INFO - codeparrot_training - Step 45219: {'lr': 1.2139985495099926e-05, 'samples': 23152640, 'steps': 45219, 'loss/train': 2.9776532649993896}
+02/26/2022 12:09:36 - INFO - codeparrot_training - Step 45220: {'lr': 1.213494908133439e-05, 'samples': 23153152, 'steps': 45220, 'loss/train': 1.8694608211517334}
+02/26/2022 12:09:39 - INFO - codeparrot_training - Step 45221: {'lr': 1.2129913686507177e-05, 'samples': 23153664, 'steps': 45221, 'loss/train': 2.2419707775115967}
+02/26/2022 12:09:45 - INFO - codeparrot_training - Step 45222: {'lr': 1.2124879310639852e-05, 'samples': 23154176, 'steps': 45222, 'loss/train': 1.7461345195770264}
+02/26/2022 12:09:48 - INFO - codeparrot_training - Step 45223: {'lr': 1.2119845953754039e-05, 'samples': 23154688, 'steps': 45223, 'loss/train': 2.1057026386260986}
+02/26/2022 12:09:54 - INFO - codeparrot_training - Step 45224: {'lr': 1.2114813615871273e-05, 'samples': 23155200, 'steps': 45224, 'loss/train': 1.709058165550232}
+02/26/2022 12:09:57 - INFO - codeparrot_training - Step 45225: {'lr': 1.2109782297013122e-05, 'samples': 23155712, 'steps': 45225, 'loss/train': 2.0414555072784424}
+02/26/2022 12:10:03 - INFO - codeparrot_training - Step 45226: {'lr': 1.210475199720107e-05, 'samples': 23156224, 'steps': 45226, 'loss/train': 1.4972212314605713}
+02/26/2022 12:10:06 - INFO - codeparrot_training - Step 45227: {'lr': 1.2099722716456763e-05, 'samples': 23156736, 'steps': 45227, 'loss/train': 2.3076183795928955}
+02/26/2022 12:10:12 - INFO - codeparrot_training - Step 45228: {'lr': 1.2094694454801713e-05, 'samples': 23157248, 'steps': 45228, 'loss/train': 2.0042624473571777}
+02/26/2022 12:10:15 - INFO - codeparrot_training - Step 45229: {'lr': 1.2089667212257432e-05, 'samples': 23157760, 'steps': 45229, 'loss/train': 1.3523269891738892}
+02/26/2022 12:10:21 - INFO - codeparrot_training - Step 45230: {'lr': 1.208464098884543e-05, 'samples': 23158272, 'steps': 45230, 'loss/train': 1.8856594562530518}
+02/26/2022 12:10:25 - INFO - codeparrot_training - Step 45231: {'lr': 1.2079615784587328e-05, 'samples': 23158784, 'steps': 45231, 'loss/train': 0.09384184330701828}
+02/26/2022 12:10:31 - INFO - codeparrot_training - Step 45232: {'lr': 1.2074591599504609e-05, 'samples': 23159296, 'steps': 45232, 'loss/train': 1.964783787727356}
+02/26/2022 12:10:34 - INFO - codeparrot_training - Step 45233: {'lr': 1.2069568433618783e-05, 'samples': 23159808, 'steps': 45233, 'loss/train': 2.2132627964019775}
+02/26/2022 12:10:40 - INFO - codeparrot_training - Step 45234: {'lr': 1.2064546286951361e-05, 'samples': 23160320, 'steps': 45234, 'loss/train': 2.107429265975952}
+02/26/2022 12:10:43 - INFO - codeparrot_training - Step 45235: {'lr': 1.2059525159523882e-05, 'samples': 23160832, 'steps': 45235, 'loss/train': 2.46586275100708}
+02/26/2022 12:10:49 - INFO - codeparrot_training - Step 45236: {'lr': 1.20545050513578e-05, 'samples': 23161344, 'steps': 45236, 'loss/train': 2.2400896549224854}
+02/26/2022 12:10:52 - INFO - codeparrot_training - Step 45237: {'lr': 1.2049485962474738e-05, 'samples': 23161856, 'steps': 45237, 'loss/train': 2.2663466930389404}
+02/26/2022 12:10:57 - INFO - codeparrot_training - Step 45238: {'lr': 1.2044467892896066e-05, 'samples': 23162368, 'steps': 45238, 'loss/train': 1.9039990901947021}
+02/26/2022 12:11:01 - INFO - codeparrot_training - Step 45239: {'lr': 1.2039450842643352e-05, 'samples': 23162880, 'steps': 45239, 'loss/train': 0.6447505354881287}
+02/26/2022 12:11:07 - INFO - codeparrot_training - Step 45240: {'lr': 1.2034434811738049e-05, 'samples': 23163392, 'steps': 45240, 'loss/train': 1.5896403789520264}
+02/26/2022 12:11:11 - INFO - codeparrot_training - Step 45241: {'lr': 1.2029419800201725e-05, 'samples': 23163904, 'steps': 45241, 'loss/train': 0.8047797679901123}
+02/26/2022 12:11:16 - INFO - codeparrot_training - Step 45242: {'lr': 1.2024405808055722e-05, 'samples': 23164416, 'steps': 45242, 'loss/train': 2.7478795051574707}
+02/26/2022 12:11:19 - INFO - codeparrot_training - Step 45243: {'lr': 1.2019392835321663e-05, 'samples': 23164928, 'steps': 45243, 'loss/train': 1.7348052263259888}
+02/26/2022 12:11:25 - INFO - codeparrot_training - Step 45244: {'lr': 1.2014380882020892e-05, 'samples': 23165440, 'steps': 45244, 'loss/train': 1.4122836589813232}
+02/26/2022 12:11:28 - INFO - codeparrot_training - Step 45245: {'lr': 1.2009369948175031e-05, 'samples': 23165952, 'steps': 45245, 'loss/train': 2.3270492553710938}
+02/26/2022 12:11:34 - INFO - codeparrot_training - Step 45246: {'lr': 1.2004360033805395e-05, 'samples': 23166464, 'steps': 45246, 'loss/train': 2.6379005908966064}
+02/26/2022 12:11:37 - INFO - codeparrot_training - Step 45247: {'lr': 1.1999351138933524e-05, 'samples': 23166976, 'steps': 45247, 'loss/train': 2.834739923477173}
+02/26/2022 12:11:43 - INFO - codeparrot_training - Step 45248: {'lr': 1.1994343263580843e-05, 'samples': 23167488, 'steps': 45248, 'loss/train': 1.8965092897415161}
+02/26/2022 12:11:46 - INFO - codeparrot_training - Step 45249: {'lr': 1.1989336407768836e-05, 'samples': 23168000, 'steps': 45249, 'loss/train': 1.741094708442688}
+02/26/2022 12:11:52 - INFO - codeparrot_training - Step 45250: {'lr': 1.198433057151893e-05, 'samples': 23168512, 'steps': 45250, 'loss/train': 1.092063069343567}
+02/26/2022 12:11:55 - INFO - codeparrot_training - Step 45251: {'lr': 1.197932575485261e-05, 'samples': 23169024, 'steps': 45251, 'loss/train': 2.190891742706299}
+02/26/2022 12:12:01 - INFO - codeparrot_training - Step 45252: {'lr': 1.1974321957791218e-05, 'samples': 23169536, 'steps': 45252, 'loss/train': 2.755958080291748}
+02/26/2022 12:12:04 - INFO - codeparrot_training - Step 45253: {'lr': 1.1969319180356292e-05, 'samples': 23170048, 'steps': 45253, 'loss/train': 1.9685778617858887}
+02/26/2022 12:12:10 - INFO - codeparrot_training - Step 45254: {'lr': 1.1964317422569204e-05, 'samples': 23170560, 'steps': 45254, 'loss/train': 1.8924490213394165}
+02/26/2022 12:12:14 - INFO - codeparrot_training - Step 45255: {'lr': 1.195931668445141e-05, 'samples': 23171072, 'steps': 45255, 'loss/train': 2.355959415435791}
+02/26/2022 12:12:20 - INFO - codeparrot_training - Step 45256: {'lr': 1.1954316966024282e-05, 'samples': 23171584, 'steps': 45256, 'loss/train': 1.4803780317306519}
+02/26/2022 12:12:24 - INFO - codeparrot_training - Step 45257: {'lr': 1.19493182673093e-05, 'samples': 23172096, 'steps': 45257, 'loss/train': 1.2660396099090576}
+02/26/2022 12:12:30 - INFO - codeparrot_training - Step 45258: {'lr': 1.194432058832784e-05, 'samples': 23172608, 'steps': 45258, 'loss/train': 1.8761061429977417}
+02/26/2022 12:12:33 - INFO - codeparrot_training - Step 45259: {'lr': 1.1939323929101326e-05, 'samples': 23173120, 'steps': 45259, 'loss/train': 1.8708829879760742}
+02/26/2022 12:12:39 - INFO - codeparrot_training - Step 45260: {'lr': 1.1934328289651131e-05, 'samples': 23173632, 'steps': 45260, 'loss/train': 2.60703444480896}
+02/26/2022 12:12:42 - INFO - codeparrot_training - Step 45261: {'lr': 1.192933366999871e-05, 'samples': 23174144, 'steps': 45261, 'loss/train': 0.9393560290336609}
+02/26/2022 12:12:48 - INFO - codeparrot_training - Step 45262: {'lr': 1.1924340070165434e-05, 'samples': 23174656, 'steps': 45262, 'loss/train': 0.1498311161994934}
+02/26/2022 12:12:51 - INFO - codeparrot_training - Step 45263: {'lr': 1.1919347490172677e-05, 'samples': 23175168, 'steps': 45263, 'loss/train': 1.5746078491210938}
+02/26/2022 12:12:57 - INFO - codeparrot_training - Step 45264: {'lr': 1.1914355930041836e-05, 'samples': 23175680, 'steps': 45264, 'loss/train': 1.025717854499817}
+02/26/2022 12:13:00 - INFO - codeparrot_training - Step 45265: {'lr': 1.1909365389794258e-05, 'samples': 23176192, 'steps': 45265, 'loss/train': 2.2685678005218506}
+02/26/2022 12:13:07 - INFO - codeparrot_training - Step 45266: {'lr': 1.1904375869451394e-05, 'samples': 23176704, 'steps': 45266, 'loss/train': 1.3533354997634888}
+02/26/2022 12:13:10 - INFO - codeparrot_training - Step 45267: {'lr': 1.1899387369034592e-05, 'samples': 23177216, 'steps': 45267, 'loss/train': 1.3332960605621338}
+02/26/2022 12:13:16 - INFO - codeparrot_training - Step 45268: {'lr': 1.1894399888565193e-05, 'samples': 23177728, 'steps': 45268, 'loss/train': 2.0228734016418457}
+02/26/2022 12:13:19 - INFO - codeparrot_training - Step 45269: {'lr': 1.1889413428064543e-05, 'samples': 23178240, 'steps': 45269, 'loss/train': 1.8938934803009033}
+02/26/2022 12:13:25 - INFO - codeparrot_training - Step 45270: {'lr': 1.1884427987554097e-05, 'samples': 23178752, 'steps': 45270, 'loss/train': 1.0400285720825195}
+02/26/2022 12:13:30 - INFO - codeparrot_training - Step 45271: {'lr': 1.1879443567055088e-05, 'samples': 23179264, 'steps': 45271, 'loss/train': 1.9002655744552612}
+02/26/2022 12:13:34 - INFO - codeparrot_training - Step 45272: {'lr': 1.1874460166589024e-05, 'samples': 23179776, 'steps': 45272, 'loss/train': 0.550462007522583}
+02/26/2022 12:13:39 - INFO - codeparrot_training - Step 45273: {'lr': 1.1869477786177058e-05, 'samples': 23180288, 'steps': 45273, 'loss/train': 0.43645116686820984}
+02/26/2022 12:13:43 - INFO - codeparrot_training - Step 45274: {'lr': 1.1864496425840699e-05, 'samples': 23180800, 'steps': 45274, 'loss/train': 1.7730686664581299}
+02/26/2022 12:13:49 - INFO - codeparrot_training - Step 45275: {'lr': 1.185951608560118e-05, 'samples': 23181312, 'steps': 45275, 'loss/train': 2.263324737548828}
+02/26/2022 12:13:52 - INFO - codeparrot_training - Step 45276: {'lr': 1.1854536765479929e-05, 'samples': 23181824, 'steps': 45276, 'loss/train': 1.4306524991989136}
+02/26/2022 12:13:58 - INFO - codeparrot_training - Step 45277: {'lr': 1.184955846549815e-05, 'samples': 23182336, 'steps': 45277, 'loss/train': 1.6198973655700684}
+02/26/2022 12:14:02 - INFO - codeparrot_training - Step 45278: {'lr': 1.18445811856773e-05, 'samples': 23182848, 'steps': 45278, 'loss/train': 2.242349624633789}
+02/26/2022 12:14:05 - INFO - codeparrot_training - Step 45279: {'lr': 1.1839604926038584e-05, 'samples': 23183360, 'steps': 45279, 'loss/train': 2.072956085205078}
+02/26/2022 12:14:11 - INFO - codeparrot_training - Step 45280: {'lr': 1.1834629686603455e-05, 'samples': 23183872, 'steps': 45280, 'loss/train': 1.8613369464874268}
+02/26/2022 12:14:15 - INFO - codeparrot_training - Step 45281: {'lr': 1.1829655467393092e-05, 'samples': 23184384, 'steps': 45281, 'loss/train': 1.722774624824524}
+02/26/2022 12:14:20 - INFO - codeparrot_training - Step 45282: {'lr': 1.1824682268428866e-05, 'samples': 23184896, 'steps': 45282, 'loss/train': 0.9641186594963074}
+02/26/2022 12:14:24 - INFO - codeparrot_training - Step 45283: {'lr': 1.1819710089732067e-05, 'samples': 23185408, 'steps': 45283, 'loss/train': 1.4226659536361694}
+02/26/2022 12:14:29 - INFO - codeparrot_training - Step 45284: {'lr': 1.181473893132401e-05, 'samples': 23185920, 'steps': 45284, 'loss/train': 1.9249154329299927}
+02/26/2022 12:14:33 - INFO - codeparrot_training - Step 45285: {'lr': 1.1809768793225983e-05, 'samples': 23186432, 'steps': 45285, 'loss/train': 2.214202404022217}
+02/26/2022 12:14:40 - INFO - codeparrot_training - Step 45286: {'lr': 1.1804799675459276e-05, 'samples': 23186944, 'steps': 45286, 'loss/train': 0.5878369212150574}
+02/26/2022 12:14:43 - INFO - codeparrot_training - Step 45287: {'lr': 1.1799831578045151e-05, 'samples': 23187456, 'steps': 45287, 'loss/train': 0.21804076433181763}
+02/26/2022 12:14:49 - INFO - codeparrot_training - Step 45288: {'lr': 1.179486450100492e-05, 'samples': 23187968, 'steps': 45288, 'loss/train': 2.0877740383148193}
+02/26/2022 12:14:52 - INFO - codeparrot_training - Step 45289: {'lr': 1.1789898444359876e-05, 'samples': 23188480, 'steps': 45289, 'loss/train': 1.3582688570022583}
+02/26/2022 12:14:58 - INFO - codeparrot_training - Step 45290: {'lr': 1.1784933408131248e-05, 'samples': 23188992, 'steps': 45290, 'loss/train': 1.643548846244812}
+02/26/2022 12:15:01 - INFO - codeparrot_training - Step 45291: {'lr': 1.1779969392340273e-05, 'samples': 23189504, 'steps': 45291, 'loss/train': 1.3875877857208252}
+02/26/2022 12:15:07 - INFO - codeparrot_training - Step 45292: {'lr': 1.177500639700832e-05, 'samples': 23190016, 'steps': 45292, 'loss/train': 1.0693222284317017}
+02/26/2022 12:15:10 - INFO - codeparrot_training - Step 45293: {'lr': 1.1770044422156623e-05, 'samples': 23190528, 'steps': 45293, 'loss/train': 1.7935291528701782}
+02/26/2022 12:15:16 - INFO - codeparrot_training - Step 45294: {'lr': 1.176508346780636e-05, 'samples': 23191040, 'steps': 45294, 'loss/train': 1.1440677642822266}
+02/26/2022 12:15:19 - INFO - codeparrot_training - Step 45295: {'lr': 1.1760123533978817e-05, 'samples': 23191552, 'steps': 45295, 'loss/train': 2.1591925621032715}
+02/26/2022 12:15:25 - INFO - codeparrot_training - Step 45296: {'lr': 1.1755164620695314e-05, 'samples': 23192064, 'steps': 45296, 'loss/train': 1.8017840385437012}
+02/26/2022 12:15:28 - INFO - codeparrot_training - Step 45297: {'lr': 1.1750206727976998e-05, 'samples': 23192576, 'steps': 45297, 'loss/train': 2.583359479904175}
+02/26/2022 12:15:34 - INFO - codeparrot_training - Step 45298: {'lr': 1.1745249855845158e-05, 'samples': 23193088, 'steps': 45298, 'loss/train': 1.352054476737976}
+02/26/2022 12:15:37 - INFO - codeparrot_training - Step 45299: {'lr': 1.1740294004320973e-05, 'samples': 23193600, 'steps': 45299, 'loss/train': 0.40386995673179626}
+02/26/2022 12:15:43 - INFO - codeparrot_training - Step 45300: {'lr': 1.1735339173425757e-05, 'samples': 23194112, 'steps': 45300, 'loss/train': 1.9061481952667236}
+02/26/2022 12:15:46 - INFO - codeparrot_training - Step 45301: {'lr': 1.173038536318069e-05, 'samples': 23194624, 'steps': 45301, 'loss/train': 0.03340005502104759}
+02/26/2022 12:15:53 - INFO - codeparrot_training - Step 45302: {'lr': 1.1725432573606975e-05, 'samples': 23195136, 'steps': 45302, 'loss/train': 2.3133888244628906}
+02/26/2022 12:15:56 - INFO - codeparrot_training - Step 45303: {'lr': 1.172048080472582e-05, 'samples': 23195648, 'steps': 45303, 'loss/train': 1.8865264654159546}
+02/26/2022 12:16:01 - INFO - codeparrot_training - Step 45304: {'lr': 1.171553005655851e-05, 'samples': 23196160, 'steps': 45304, 'loss/train': 1.6483254432678223}
+02/26/2022 12:16:05 - INFO - codeparrot_training - Step 45305: {'lr': 1.1710580329126196e-05, 'samples': 23196672, 'steps': 45305, 'loss/train': 1.6172759532928467}
+02/26/2022 12:16:11 - INFO - codeparrot_training - Step 45306: {'lr': 1.1705631622450086e-05, 'samples': 23197184, 'steps': 45306, 'loss/train': 0.6871628165245056}
+02/26/2022 12:16:14 - INFO - codeparrot_training - Step 45307: {'lr': 1.1700683936551355e-05, 'samples': 23197696, 'steps': 45307, 'loss/train': 1.934171438217163}
+02/26/2022 12:16:20 - INFO - codeparrot_training - Step 45308: {'lr': 1.1695737271451263e-05, 'samples': 23198208, 'steps': 45308, 'loss/train': 1.0775582790374756}
+02/26/2022 12:16:23 - INFO - codeparrot_training - Step 45309: {'lr': 1.169079162717096e-05, 'samples': 23198720, 'steps': 45309, 'loss/train': 1.492814064025879}
+02/26/2022 12:16:29 - INFO - codeparrot_training - Step 45310: {'lr': 1.1685847003731598e-05, 'samples': 23199232, 'steps': 45310, 'loss/train': 1.3731632232666016}
+02/26/2022 12:16:32 - INFO - codeparrot_training - Step 45311: {'lr': 1.1680903401154463e-05, 'samples': 23199744, 'steps': 45311, 'loss/train': 1.9341137409210205}
+02/26/2022 12:16:38 - INFO - codeparrot_training - Step 45312: {'lr': 1.1675960819460595e-05, 'samples': 23200256, 'steps': 45312, 'loss/train': 2.4144809246063232}
+02/26/2022 12:16:41 - INFO - codeparrot_training - Step 45313: {'lr': 1.1671019258671285e-05, 'samples': 23200768, 'steps': 45313, 'loss/train': 1.560051441192627}
+02/26/2022 12:16:48 - INFO - codeparrot_training - Step 45314: {'lr': 1.1666078718807593e-05, 'samples': 23201280, 'steps': 45314, 'loss/train': 1.2260018587112427}
+02/26/2022 12:16:51 - INFO - codeparrot_training - Step 45315: {'lr': 1.1661139199890813e-05, 'samples': 23201792, 'steps': 45315, 'loss/train': 0.16942980885505676}
+02/26/2022 12:16:56 - INFO - codeparrot_training - Step 45316: {'lr': 1.1656200701941955e-05, 'samples': 23202304, 'steps': 45316, 'loss/train': 1.7891305685043335}
+02/26/2022 12:17:00 - INFO - codeparrot_training - Step 45317: {'lr': 1.1651263224982305e-05, 'samples': 23202816, 'steps': 45317, 'loss/train': 1.336639404296875}
+02/26/2022 12:17:06 - INFO - codeparrot_training - Step 45318: {'lr': 1.1646326769032934e-05, 'samples': 23203328, 'steps': 45318, 'loss/train': 1.3226101398468018}
+02/26/2022 12:17:09 - INFO - codeparrot_training - Step 45319: {'lr': 1.1641391334115015e-05, 'samples': 23203840, 'steps': 45319, 'loss/train': 1.901580572128296}
+02/26/2022 12:17:14 - INFO - codeparrot_training - Step 45320: {'lr': 1.16364569202497e-05, 'samples': 23204352, 'steps': 45320, 'loss/train': 1.0646063089370728}
+02/26/2022 12:17:18 - INFO - codeparrot_training - Step 45321: {'lr': 1.1631523527458109e-05, 'samples': 23204864, 'steps': 45321, 'loss/train': 1.7868127822875977}
+02/26/2022 12:17:23 - INFO - codeparrot_training - Step 45322: {'lr': 1.1626591155761368e-05, 'samples': 23205376, 'steps': 45322, 'loss/train': 1.2724719047546387}
+02/26/2022 12:17:27 - INFO - codeparrot_training - Step 45323: {'lr': 1.1621659805180623e-05, 'samples': 23205888, 'steps': 45323, 'loss/train': 1.95565664768219}
+02/26/2022 12:17:33 - INFO - codeparrot_training - Step 45324: {'lr': 1.1616729475737025e-05, 'samples': 23206400, 'steps': 45324, 'loss/train': 1.5162220001220703}
+02/26/2022 12:17:36 - INFO - codeparrot_training - Step 45325: {'lr': 1.1611800167451642e-05, 'samples': 23206912, 'steps': 45325, 'loss/train': 0.9925150871276855}
+02/26/2022 12:17:42 - INFO - codeparrot_training - Step 45326: {'lr': 1.1606871880345592e-05, 'samples': 23207424, 'steps': 45326, 'loss/train': 0.8877344727516174}
+02/26/2022 12:17:45 - INFO - codeparrot_training - Step 45327: {'lr': 1.160194461444003e-05, 'samples': 23207936, 'steps': 45327, 'loss/train': 1.5658241510391235}
+02/26/2022 12:17:51 - INFO - codeparrot_training - Step 45328: {'lr': 1.1597018369756018e-05, 'samples': 23208448, 'steps': 45328, 'loss/train': 2.043818235397339}
+02/26/2022 12:17:54 - INFO - codeparrot_training - Step 45329: {'lr': 1.1592093146314709e-05, 'samples': 23208960, 'steps': 45329, 'loss/train': 2.1026744842529297}
+02/26/2022 12:18:00 - INFO - codeparrot_training - Step 45330: {'lr': 1.1587168944137112e-05, 'samples': 23209472, 'steps': 45330, 'loss/train': 2.0544748306274414}
+02/26/2022 12:18:03 - INFO - codeparrot_training - Step 45331: {'lr': 1.1582245763244432e-05, 'samples': 23209984, 'steps': 45331, 'loss/train': 1.0296118259429932}
+02/26/2022 12:18:09 - INFO - codeparrot_training - Step 45332: {'lr': 1.1577323603657652e-05, 'samples': 23210496, 'steps': 45332, 'loss/train': 0.8347918391227722}
+02/26/2022 12:18:12 - INFO - codeparrot_training - Step 45333: {'lr': 1.157240246539798e-05, 'samples': 23211008, 'steps': 45333, 'loss/train': 1.4511092901229858}
+02/26/2022 12:18:19 - INFO - codeparrot_training - Step 45334: {'lr': 1.156748234848637e-05, 'samples': 23211520, 'steps': 45334, 'loss/train': 1.2427624464035034}
+02/26/2022 12:18:22 - INFO - codeparrot_training - Step 45335: {'lr': 1.1562563252943998e-05, 'samples': 23212032, 'steps': 45335, 'loss/train': 2.073885679244995}
+02/26/2022 12:18:28 - INFO - codeparrot_training - Step 45336: {'lr': 1.1557645178791847e-05, 'samples': 23212544, 'steps': 45336, 'loss/train': 1.9329290390014648}
+02/26/2022 12:18:31 - INFO - codeparrot_training - Step 45337: {'lr': 1.1552728126051098e-05, 'samples': 23213056, 'steps': 45337, 'loss/train': 1.5023045539855957}
+02/26/2022 12:18:37 - INFO - codeparrot_training - Step 45338: {'lr': 1.1547812094742677e-05, 'samples': 23213568, 'steps': 45338, 'loss/train': 1.5433142185211182}
+02/26/2022 12:18:40 - INFO - codeparrot_training - Step 45339: {'lr': 1.1542897084887732e-05, 'samples': 23214080, 'steps': 45339, 'loss/train': 1.8299283981323242}
+02/26/2022 12:18:46 - INFO - codeparrot_training - Step 45340: {'lr': 1.1537983096507277e-05, 'samples': 23214592, 'steps': 45340, 'loss/train': 1.4210702180862427}
+02/26/2022 12:18:49 - INFO - codeparrot_training - Step 45341: {'lr': 1.1533070129622458e-05, 'samples': 23215104, 'steps': 45341, 'loss/train': 1.679061770439148}
+02/26/2022 12:18:55 - INFO - codeparrot_training - Step 45342: {'lr': 1.1528158184254151e-05, 'samples': 23215616, 'steps': 45342, 'loss/train': 2.0270400047302246}
+02/26/2022 12:18:58 - INFO - codeparrot_training - Step 45343: {'lr': 1.1523247260423558e-05, 'samples': 23216128, 'steps': 45343, 'loss/train': 2.6155829429626465}
+02/26/2022 12:19:04 - INFO - codeparrot_training - Step 45344: {'lr': 1.1518337358151636e-05, 'samples': 23216640, 'steps': 45344, 'loss/train': 2.6217074394226074}
+02/26/2022 12:19:07 - INFO - codeparrot_training - Step 45345: {'lr': 1.1513428477459398e-05, 'samples': 23217152, 'steps': 45345, 'loss/train': 1.8489625453948975}
+02/26/2022 12:19:13 - INFO - codeparrot_training - Step 45346: {'lr': 1.1508520618367934e-05, 'samples': 23217664, 'steps': 45346, 'loss/train': 2.1252923011779785}
+02/26/2022 12:19:16 - INFO - codeparrot_training - Step 45347: {'lr': 1.1503613780898231e-05, 'samples': 23218176, 'steps': 45347, 'loss/train': 0.4666687250137329}
+02/26/2022 12:19:23 - INFO - codeparrot_training - Step 45348: {'lr': 1.1498707965071325e-05, 'samples': 23218688, 'steps': 45348, 'loss/train': 2.7555248737335205}
+02/26/2022 12:19:26 - INFO - codeparrot_training - Step 45349: {'lr': 1.1493803170908174e-05, 'samples': 23219200, 'steps': 45349, 'loss/train': 2.4711475372314453}
+02/26/2022 12:19:32 - INFO - codeparrot_training - Step 45350: {'lr': 1.1488899398429897e-05, 'samples': 23219712, 'steps': 45350, 'loss/train': 1.7018671035766602}
+02/26/2022 12:19:35 - INFO - codeparrot_training - Step 45351: {'lr': 1.1483996647657424e-05, 'samples': 23220224, 'steps': 45351, 'loss/train': 2.3630166053771973}
+02/26/2022 12:19:41 - INFO - codeparrot_training - Step 45352: {'lr': 1.1479094918611765e-05, 'samples': 23220736, 'steps': 45352, 'loss/train': 1.1727098226547241}
+02/26/2022 12:19:44 - INFO - codeparrot_training - Step 45353: {'lr': 1.1474194211313905e-05, 'samples': 23221248, 'steps': 45353, 'loss/train': 1.8439149856567383}
+02/26/2022 12:19:50 - INFO - codeparrot_training - Step 45354: {'lr': 1.146929452578488e-05, 'samples': 23221760, 'steps': 45354, 'loss/train': 1.3220343589782715}
+02/26/2022 12:19:53 - INFO - codeparrot_training - Step 45355: {'lr': 1.1464395862045645e-05, 'samples': 23222272, 'steps': 45355, 'loss/train': 2.2462637424468994}
+02/26/2022 12:19:59 - INFO - codeparrot_training - Step 45356: {'lr': 1.1459498220117214e-05, 'samples': 23222784, 'steps': 45356, 'loss/train': 2.5816359519958496}
+02/26/2022 12:20:02 - INFO - codeparrot_training - Step 45357: {'lr': 1.1454601600020515e-05, 'samples': 23223296, 'steps': 45357, 'loss/train': 2.331681966781616}
+02/26/2022 12:20:08 - INFO - codeparrot_training - Step 45358: {'lr': 1.1449706001776583e-05, 'samples': 23223808, 'steps': 45358, 'loss/train': 0.09742573648691177}
+02/26/2022 12:20:11 - INFO - codeparrot_training - Step 45359: {'lr': 1.144481142540635e-05, 'samples': 23224320, 'steps': 45359, 'loss/train': 0.7433115839958191}
+02/26/2022 12:20:18 - INFO - codeparrot_training - Step 45360: {'lr': 1.1439917870930795e-05, 'samples': 23224832, 'steps': 45360, 'loss/train': 1.5361292362213135}
+02/26/2022 12:20:21 - INFO - codeparrot_training - Step 45361: {'lr': 1.1435025338370875e-05, 'samples': 23225344, 'steps': 45361, 'loss/train': 2.9006142616271973}
+02/26/2022 12:20:27 - INFO - codeparrot_training - Step 45362: {'lr': 1.1430133827747574e-05, 'samples': 23225856, 'steps': 45362, 'loss/train': 1.2615761756896973}
+02/26/2022 12:20:30 - INFO - codeparrot_training - Step 45363: {'lr': 1.142524333908182e-05, 'samples': 23226368, 'steps': 45363, 'loss/train': 1.749088168144226}
+02/26/2022 12:20:36 - INFO - codeparrot_training - Step 45364: {'lr': 1.1420353872394567e-05, 'samples': 23226880, 'steps': 45364, 'loss/train': 1.0706027746200562}
+02/26/2022 12:20:39 - INFO - codeparrot_training - Step 45365: {'lr': 1.1415465427706717e-05, 'samples': 23227392, 'steps': 45365, 'loss/train': 1.099647045135498}
+02/26/2022 12:20:45 - INFO - codeparrot_training - Step 45366: {'lr': 1.1410578005039308e-05, 'samples': 23227904, 'steps': 45366, 'loss/train': 1.9706231355667114}
+02/26/2022 12:20:48 - INFO - codeparrot_training - Step 45367: {'lr': 1.1405691604413182e-05, 'samples': 23228416, 'steps': 45367, 'loss/train': 1.6307319402694702}
+02/26/2022 12:20:54 - INFO - codeparrot_training - Step 45368: {'lr': 1.1400806225849352e-05, 'samples': 23228928, 'steps': 45368, 'loss/train': 1.7613496780395508}
+02/26/2022 12:20:57 - INFO - codeparrot_training - Step 45369: {'lr': 1.1395921869368664e-05, 'samples': 23229440, 'steps': 45369, 'loss/train': 1.8023744821548462}
+02/26/2022 12:21:04 - INFO - codeparrot_training - Step 45370: {'lr': 1.1391038534992126e-05, 'samples': 23229952, 'steps': 45370, 'loss/train': 3.5374059677124023}
+02/26/2022 12:21:07 - INFO - codeparrot_training - Step 45371: {'lr': 1.1386156222740557e-05, 'samples': 23230464, 'steps': 45371, 'loss/train': 2.701000928878784}
+02/26/2022 12:21:12 - INFO - codeparrot_training - Step 45372: {'lr': 1.1381274932634994e-05, 'samples': 23230976, 'steps': 45372, 'loss/train': 2.431722402572632}
+02/26/2022 12:21:18 - INFO - codeparrot_training - Step 45373: {'lr': 1.1376394664696227e-05, 'samples': 23231488, 'steps': 45373, 'loss/train': 1.1495332717895508}
+02/26/2022 12:21:22 - INFO - codeparrot_training - Step 45374: {'lr': 1.1371515418945239e-05, 'samples': 23232000, 'steps': 45374, 'loss/train': 2.1136183738708496}
+02/26/2022 12:21:25 - INFO - codeparrot_training - Step 45375: {'lr': 1.1366637195402874e-05, 'samples': 23232512, 'steps': 45375, 'loss/train': 1.4459030628204346}
+02/26/2022 12:21:30 - INFO - codeparrot_training - Step 45376: {'lr': 1.1361759994090115e-05, 'samples': 23233024, 'steps': 45376, 'loss/train': 1.4107810258865356}
+02/26/2022 12:21:36 - INFO - codeparrot_training - Step 45377: {'lr': 1.1356883815027751e-05, 'samples': 23233536, 'steps': 45377, 'loss/train': 1.5117871761322021}
+02/26/2022 12:21:39 - INFO - codeparrot_training - Step 45378: {'lr': 1.135200865823674e-05, 'samples': 23234048, 'steps': 45378, 'loss/train': 0.8988096714019775}
+02/26/2022 12:21:46 - INFO - codeparrot_training - Step 45379: {'lr': 1.1347134523737923e-05, 'samples': 23234560, 'steps': 45379, 'loss/train': 1.5875827074050903}
+02/26/2022 12:21:49 - INFO - codeparrot_training - Step 45380: {'lr': 1.134226141155223e-05, 'samples': 23235072, 'steps': 45380, 'loss/train': 2.5511093139648438}
+02/26/2022 12:21:55 - INFO - codeparrot_training - Step 45381: {'lr': 1.1337389321700504e-05, 'samples': 23235584, 'steps': 45381, 'loss/train': 2.0994980335235596}
+02/26/2022 12:21:58 - INFO - codeparrot_training - Step 45382: {'lr': 1.1332518254203616e-05, 'samples': 23236096, 'steps': 45382, 'loss/train': 1.953726053237915}
+02/26/2022 12:22:04 - INFO - codeparrot_training - Step 45383: {'lr': 1.1327648209082413e-05, 'samples': 23236608, 'steps': 45383, 'loss/train': 1.3440583944320679}
+02/26/2022 12:22:08 - INFO - codeparrot_training - Step 45384: {'lr': 1.1322779186357796e-05, 'samples': 23237120, 'steps': 45384, 'loss/train': 2.516601085662842}
+02/26/2022 12:22:11 - INFO - codeparrot_training - Step 45385: {'lr': 1.1317911186050606e-05, 'samples': 23237632, 'steps': 45385, 'loss/train': 1.6852357387542725}
+02/26/2022 12:22:17 - INFO - codeparrot_training - Step 45386: {'lr': 1.1313044208181717e-05, 'samples': 23238144, 'steps': 45386, 'loss/train': 1.2206635475158691}
+02/26/2022 12:22:20 - INFO - codeparrot_training - Step 45387: {'lr': 1.1308178252771888e-05, 'samples': 23238656, 'steps': 45387, 'loss/train': 1.815065622329712}
+02/26/2022 12:22:26 - INFO - codeparrot_training - Step 45388: {'lr': 1.1303313319842078e-05, 'samples': 23239168, 'steps': 45388, 'loss/train': 0.47971311211586}
+02/26/2022 12:22:29 - INFO - codeparrot_training - Step 45389: {'lr': 1.1298449409413076e-05, 'samples': 23239680, 'steps': 45389, 'loss/train': 2.2196054458618164}
+02/26/2022 12:22:35 - INFO - codeparrot_training - Step 45390: {'lr': 1.129358652150575e-05, 'samples': 23240192, 'steps': 45390, 'loss/train': 2.972101926803589}
+02/26/2022 12:22:38 - INFO - codeparrot_training - Step 45391: {'lr': 1.1288724656140837e-05, 'samples': 23240704, 'steps': 45391, 'loss/train': 1.461857795715332}
+02/26/2022 12:22:44 - INFO - codeparrot_training - Step 45392: {'lr': 1.1283863813339262e-05, 'samples': 23241216, 'steps': 45392, 'loss/train': 0.8047551512718201}
+02/26/2022 12:22:47 - INFO - codeparrot_training - Step 45393: {'lr': 1.1279003993121817e-05, 'samples': 23241728, 'steps': 45393, 'loss/train': 0.6176438331604004}
+02/26/2022 12:22:53 - INFO - codeparrot_training - Step 45394: {'lr': 1.1274145195509317e-05, 'samples': 23242240, 'steps': 45394, 'loss/train': 1.6496230363845825}
+02/26/2022 12:22:56 - INFO - codeparrot_training - Step 45395: {'lr': 1.1269287420522578e-05, 'samples': 23242752, 'steps': 45395, 'loss/train': 0.8046233654022217}
+02/26/2022 12:23:02 - INFO - codeparrot_training - Step 45396: {'lr': 1.1264430668182363e-05, 'samples': 23243264, 'steps': 45396, 'loss/train': 1.852026343345642}
+02/26/2022 12:23:06 - INFO - codeparrot_training - Step 45397: {'lr': 1.125957493850957e-05, 'samples': 23243776, 'steps': 45397, 'loss/train': 1.6996564865112305}
+02/26/2022 12:23:11 - INFO - codeparrot_training - Step 45398: {'lr': 1.1254720231524934e-05, 'samples': 23244288, 'steps': 45398, 'loss/train': 1.3712424039840698}
+02/26/2022 12:23:15 - INFO - codeparrot_training - Step 45399: {'lr': 1.1249866547249272e-05, 'samples': 23244800, 'steps': 45399, 'loss/train': 2.1269595623016357}
+02/26/2022 12:23:20 - INFO - codeparrot_training - Step 45400: {'lr': 1.1245013885703342e-05, 'samples': 23245312, 'steps': 45400, 'loss/train': 1.056168556213379}
+02/26/2022 12:23:24 - INFO - codeparrot_training - Step 45401: {'lr': 1.1240162246907992e-05, 'samples': 23245824, 'steps': 45401, 'loss/train': 1.077668309211731}
+02/26/2022 12:23:29 - INFO - codeparrot_training - Step 45402: {'lr': 1.1235311630883982e-05, 'samples': 23246336, 'steps': 45402, 'loss/train': 1.8941786289215088}
+02/26/2022 12:23:33 - INFO - codeparrot_training - Step 45403: {'lr': 1.1230462037652046e-05, 'samples': 23246848, 'steps': 45403, 'loss/train': 1.5298796892166138}
+02/26/2022 12:23:38 - INFO - codeparrot_training - Step 45404: {'lr': 1.1225613467232998e-05, 'samples': 23247360, 'steps': 45404, 'loss/train': 1.7032171487808228}
+02/26/2022 12:23:45 - INFO - codeparrot_training - Step 45405: {'lr': 1.1220765919647602e-05, 'samples': 23247872, 'steps': 45405, 'loss/train': 1.6266518831253052}
+02/26/2022 12:23:48 - INFO - codeparrot_training - Step 45406: {'lr': 1.1215919394916619e-05, 'samples': 23248384, 'steps': 45406, 'loss/train': 0.9353079795837402}
+02/26/2022 12:23:54 - INFO - codeparrot_training - Step 45407: {'lr': 1.1211073893060863e-05, 'samples': 23248896, 'steps': 45407, 'loss/train': 1.487837314605713}
+02/26/2022 12:23:57 - INFO - codeparrot_training - Step 45408: {'lr': 1.1206229414100988e-05, 'samples': 23249408, 'steps': 45408, 'loss/train': 0.3855515718460083}
+02/26/2022 12:24:03 - INFO - codeparrot_training - Step 45409: {'lr': 1.1201385958057836e-05, 'samples': 23249920, 'steps': 45409, 'loss/train': 1.765134334564209}
+02/26/2022 12:24:06 - INFO - codeparrot_training - Step 45410: {'lr': 1.1196543524952057e-05, 'samples': 23250432, 'steps': 45410, 'loss/train': 2.3218090534210205}
+02/26/2022 12:24:12 - INFO - codeparrot_training - Step 45411: {'lr': 1.1191702114804552e-05, 'samples': 23250944, 'steps': 45411, 'loss/train': 0.8160605430603027}
+02/26/2022 12:24:15 - INFO - codeparrot_training - Step 45412: {'lr': 1.1186861727635889e-05, 'samples': 23251456, 'steps': 45412, 'loss/train': 1.7722963094711304}
+02/26/2022 12:24:19 - INFO - codeparrot_training - Step 45413: {'lr': 1.1182022363466909e-05, 'samples': 23251968, 'steps': 45413, 'loss/train': 0.650709867477417}
+02/26/2022 12:24:24 - INFO - codeparrot_training - Step 45414: {'lr': 1.1177184022318293e-05, 'samples': 23252480, 'steps': 45414, 'loss/train': 1.802277684211731}
+02/26/2022 12:24:31 - INFO - codeparrot_training - Step 45415: {'lr': 1.1172346704210857e-05, 'samples': 23252992, 'steps': 45415, 'loss/train': 3.045698881149292}
+02/26/2022 12:24:35 - INFO - codeparrot_training - Step 45416: {'lr': 1.1167510409165166e-05, 'samples': 23253504, 'steps': 45416, 'loss/train': 2.619114875793457}
+02/26/2022 12:24:38 - INFO - codeparrot_training - Step 45417: {'lr': 1.1162675137202067e-05, 'samples': 23254016, 'steps': 45417, 'loss/train': 0.9451178312301636}
+02/26/2022 12:24:44 - INFO - codeparrot_training - Step 45418: {'lr': 1.115784088834218e-05, 'samples': 23254528, 'steps': 45418, 'loss/train': 2.0916895866394043}
+02/26/2022 12:24:47 - INFO - codeparrot_training - Step 45419: {'lr': 1.1153007662606296e-05, 'samples': 23255040, 'steps': 45419, 'loss/train': 1.4889509677886963}
+02/26/2022 12:24:53 - INFO - codeparrot_training - Step 45420: {'lr': 1.1148175460015092e-05, 'samples': 23255552, 'steps': 45420, 'loss/train': 0.9746044278144836}
+02/26/2022 12:24:56 - INFO - codeparrot_training - Step 45421: {'lr': 1.1143344280589273e-05, 'samples': 23256064, 'steps': 45421, 'loss/train': 0.9515894055366516}
+02/26/2022 12:25:02 - INFO - codeparrot_training - Step 45422: {'lr': 1.1138514124349463e-05, 'samples': 23256576, 'steps': 45422, 'loss/train': 1.6756755113601685}
+02/26/2022 12:25:05 - INFO - codeparrot_training - Step 45423: {'lr': 1.113368499131645e-05, 'samples': 23257088, 'steps': 45423, 'loss/train': 3.002045154571533}
+02/26/2022 12:25:10 - INFO - codeparrot_training - Step 45424: {'lr': 1.1128856881510913e-05, 'samples': 23257600, 'steps': 45424, 'loss/train': 2.4058916568756104}
+02/26/2022 12:25:14 - INFO - codeparrot_training - Step 45425: {'lr': 1.1124029794953472e-05, 'samples': 23258112, 'steps': 45425, 'loss/train': 0.14879821240901947}
+02/26/2022 12:25:20 - INFO - codeparrot_training - Step 45426: {'lr': 1.1119203731664807e-05, 'samples': 23258624, 'steps': 45426, 'loss/train': 0.17805123329162598}
+02/26/2022 12:25:26 - INFO - codeparrot_training - Step 45427: {'lr': 1.1114378691665649e-05, 'samples': 23259136, 'steps': 45427, 'loss/train': 1.9532897472381592}
+02/26/2022 12:25:29 - INFO - codeparrot_training - Step 45428: {'lr': 1.1109554674976651e-05, 'samples': 23259648, 'steps': 45428, 'loss/train': 1.526245355606079}
+02/26/2022 12:25:34 - INFO - codeparrot_training - Step 45429: {'lr': 1.1104731681618463e-05, 'samples': 23260160, 'steps': 45429, 'loss/train': 1.6933094263076782}
+02/26/2022 12:25:38 - INFO - codeparrot_training - Step 45430: {'lr': 1.1099909711611705e-05, 'samples': 23260672, 'steps': 45430, 'loss/train': 1.3095364570617676}
+02/26/2022 12:25:43 - INFO - codeparrot_training - Step 45431: {'lr': 1.1095088764977112e-05, 'samples': 23261184, 'steps': 45431, 'loss/train': 1.7271593809127808}
+02/26/2022 12:25:47 - INFO - codeparrot_training - Step 45432: {'lr': 1.1090268841735307e-05, 'samples': 23261696, 'steps': 45432, 'loss/train': 3.4255309104919434}
+02/26/2022 12:25:52 - INFO - codeparrot_training - Step 45433: {'lr': 1.1085449941906912e-05, 'samples': 23262208, 'steps': 45433, 'loss/train': 1.8797121047973633}
+02/26/2022 12:25:56 - INFO - codeparrot_training - Step 45434: {'lr': 1.1080632065512546e-05, 'samples': 23262720, 'steps': 45434, 'loss/train': 1.8912968635559082}
+02/26/2022 12:26:02 - INFO - codeparrot_training - Step 45435: {'lr': 1.1075815212572948e-05, 'samples': 23263232, 'steps': 45435, 'loss/train': 1.7881375551223755}
+02/26/2022 12:26:06 - INFO - codeparrot_training - Step 45436: {'lr': 1.1070999383108682e-05, 'samples': 23263744, 'steps': 45436, 'loss/train': 1.7056219577789307}
+02/26/2022 12:26:11 - INFO - codeparrot_training - Step 45437: {'lr': 1.106618457714037e-05, 'samples': 23264256, 'steps': 45437, 'loss/train': 1.3320502042770386}
+02/26/2022 12:26:15 - INFO - codeparrot_training - Step 45438: {'lr': 1.1061370794688635e-05, 'samples': 23264768, 'steps': 45438, 'loss/train': 2.526350736618042}
+02/26/2022 12:26:20 - INFO - codeparrot_training - Step 45439: {'lr': 1.1056558035774156e-05, 'samples': 23265280, 'steps': 45439, 'loss/train': 1.2630500793457031}
+02/26/2022 12:26:23 - INFO - codeparrot_training - Step 45440: {'lr': 1.105174630041747e-05, 'samples': 23265792, 'steps': 45440, 'loss/train': 2.1425564289093018}
+02/26/2022 12:26:29 - INFO - codeparrot_training - Step 45441: {'lr': 1.104693558863923e-05, 'samples': 23266304, 'steps': 45441, 'loss/train': 2.606959104537964}
+02/26/2022 12:26:33 - INFO - codeparrot_training - Step 45442: {'lr': 1.104212590046011e-05, 'samples': 23266816, 'steps': 45442, 'loss/train': 1.2670127153396606}
+02/26/2022 12:26:38 - INFO - codeparrot_training - Step 45443: {'lr': 1.1037317235900568e-05, 'samples': 23267328, 'steps': 45443, 'loss/train': 2.846410036087036}
+02/26/2022 12:26:41 - INFO - codeparrot_training - Step 45444: {'lr': 1.103250959498131e-05, 'samples': 23267840, 'steps': 45444, 'loss/train': 2.292482852935791}
+02/26/2022 12:26:47 - INFO - codeparrot_training - Step 45445: {'lr': 1.1027702977722875e-05, 'samples': 23268352, 'steps': 45445, 'loss/train': 1.5529601573944092}
+02/26/2022 12:26:50 - INFO - codeparrot_training - Step 45446: {'lr': 1.1022897384145941e-05, 'samples': 23268864, 'steps': 45446, 'loss/train': 1.0820204019546509}
+02/26/2022 12:26:56 - INFO - codeparrot_training - Step 45447: {'lr': 1.1018092814270963e-05, 'samples': 23269376, 'steps': 45447, 'loss/train': 0.5086148977279663}
+02/26/2022 12:27:00 - INFO - codeparrot_training - Step 45448: {'lr': 1.101328926811862e-05, 'samples': 23269888, 'steps': 45448, 'loss/train': 2.6104953289031982}
+02/26/2022 12:27:05 - INFO - codeparrot_training - Step 45449: {'lr': 1.1008486745709423e-05, 'samples': 23270400, 'steps': 45449, 'loss/train': 1.473425030708313}
+02/26/2022 12:27:09 - INFO - codeparrot_training - Step 45450: {'lr': 1.1003685247064076e-05, 'samples': 23270912, 'steps': 45450, 'loss/train': 1.9737762212753296}
+02/26/2022 12:27:15 - INFO - codeparrot_training - Step 45451: {'lr': 1.0998884772202955e-05, 'samples': 23271424, 'steps': 45451, 'loss/train': 0.8413974642753601}
+02/26/2022 12:27:18 - INFO - codeparrot_training - Step 45452: {'lr': 1.0994085321146763e-05, 'samples': 23271936, 'steps': 45452, 'loss/train': 2.181739330291748}
+02/26/2022 12:27:24 - INFO - codeparrot_training - Step 45453: {'lr': 1.0989286893915984e-05, 'samples': 23272448, 'steps': 45453, 'loss/train': 1.670751929283142}
+02/26/2022 12:27:27 - INFO - codeparrot_training - Step 45454: {'lr': 1.0984489490531242e-05, 'samples': 23272960, 'steps': 45454, 'loss/train': 1.9378257989883423}
+02/26/2022 12:27:33 - INFO - codeparrot_training - Step 45455: {'lr': 1.0979693111013045e-05, 'samples': 23273472, 'steps': 45455, 'loss/train': 0.9125990867614746}
+02/26/2022 12:27:36 - INFO - codeparrot_training - Step 45456: {'lr': 1.0974897755381935e-05, 'samples': 23273984, 'steps': 45456, 'loss/train': 2.0668399333953857}
+02/26/2022 12:27:42 - INFO - codeparrot_training - Step 45457: {'lr': 1.097010342365845e-05, 'samples': 23274496, 'steps': 45457, 'loss/train': 1.9885073900222778}
+02/26/2022 12:27:45 - INFO - codeparrot_training - Step 45458: {'lr': 1.0965310115863159e-05, 'samples': 23275008, 'steps': 45458, 'loss/train': 1.798528790473938}
+02/26/2022 12:27:51 - INFO - codeparrot_training - Step 45459: {'lr': 1.096051783201657e-05, 'samples': 23275520, 'steps': 45459, 'loss/train': 2.220317840576172}
+02/26/2022 12:27:54 - INFO - codeparrot_training - Step 45460: {'lr': 1.0955726572139252e-05, 'samples': 23276032, 'steps': 45460, 'loss/train': 1.0446014404296875}
+02/26/2022 12:28:01 - INFO - codeparrot_training - Step 45461: {'lr': 1.0950936336251632e-05, 'samples': 23276544, 'steps': 45461, 'loss/train': 1.6791778802871704}
+02/26/2022 12:28:04 - INFO - codeparrot_training - Step 45462: {'lr': 1.0946147124374361e-05, 'samples': 23277056, 'steps': 45462, 'loss/train': 1.4799556732177734}
+02/26/2022 12:28:09 - INFO - codeparrot_training - Step 45463: {'lr': 1.094135893652784e-05, 'samples': 23277568, 'steps': 45463, 'loss/train': 1.6639615297317505}
+02/26/2022 12:28:13 - INFO - codeparrot_training - Step 45464: {'lr': 1.0936571772732662e-05, 'samples': 23278080, 'steps': 45464, 'loss/train': 2.132904052734375}
+02/26/2022 12:28:18 - INFO - codeparrot_training - Step 45465: {'lr': 1.0931785633009256e-05, 'samples': 23278592, 'steps': 45465, 'loss/train': 2.355464458465576}
+02/26/2022 12:28:22 - INFO - codeparrot_training - Step 45466: {'lr': 1.0927000517378216e-05, 'samples': 23279104, 'steps': 45466, 'loss/train': 1.3249207735061646}
+02/26/2022 12:28:27 - INFO - codeparrot_training - Step 45467: {'lr': 1.0922216425859943e-05, 'samples': 23279616, 'steps': 45467, 'loss/train': 1.9733928442001343}
+02/26/2022 12:28:31 - INFO - codeparrot_training - Step 45468: {'lr': 1.0917433358475031e-05, 'samples': 23280128, 'steps': 45468, 'loss/train': 2.930763006210327}
+02/26/2022 12:28:36 - INFO - codeparrot_training - Step 45469: {'lr': 1.0912651315243883e-05, 'samples': 23280640, 'steps': 45469, 'loss/train': 1.4898325204849243}
+02/26/2022 12:28:40 - INFO - codeparrot_training - Step 45470: {'lr': 1.0907870296187034e-05, 'samples': 23281152, 'steps': 45470, 'loss/train': 1.352459192276001}
+02/26/2022 12:28:46 - INFO - codeparrot_training - Step 45471: {'lr': 1.0903090301324914e-05, 'samples': 23281664, 'steps': 45471, 'loss/train': 2.132077693939209}
+02/26/2022 12:28:49 - INFO - codeparrot_training - Step 45472: {'lr': 1.0898311330678117e-05, 'samples': 23282176, 'steps': 45472, 'loss/train': 1.977565050125122}
+02/26/2022 12:28:55 - INFO - codeparrot_training - Step 45473: {'lr': 1.0893533384266936e-05, 'samples': 23282688, 'steps': 45473, 'loss/train': 1.3192912340164185}
+02/26/2022 12:28:58 - INFO - codeparrot_training - Step 45474: {'lr': 1.0888756462111987e-05, 'samples': 23283200, 'steps': 45474, 'loss/train': 1.3958088159561157}
+02/26/2022 12:29:04 - INFO - codeparrot_training - Step 45475: {'lr': 1.0883980564233676e-05, 'samples': 23283712, 'steps': 45475, 'loss/train': 2.227958917617798}
+02/26/2022 12:29:07 - INFO - codeparrot_training - Step 45476: {'lr': 1.0879205690652428e-05, 'samples': 23284224, 'steps': 45476, 'loss/train': 1.4331892728805542}
+02/26/2022 12:29:13 - INFO - codeparrot_training - Step 45477: {'lr': 1.0874431841388754e-05, 'samples': 23284736, 'steps': 45477, 'loss/train': 1.2371387481689453}
+02/26/2022 12:29:16 - INFO - codeparrot_training - Step 45478: {'lr': 1.0869659016463113e-05, 'samples': 23285248, 'steps': 45478, 'loss/train': 2.2302229404449463}
+02/26/2022 12:29:22 - INFO - codeparrot_training - Step 45479: {'lr': 1.0864887215895902e-05, 'samples': 23285760, 'steps': 45479, 'loss/train': 1.913045048713684}
+02/26/2022 12:29:25 - INFO - codeparrot_training - Step 45480: {'lr': 1.086011643970755e-05, 'samples': 23286272, 'steps': 45480, 'loss/train': 1.6322394609451294}
+02/26/2022 12:29:31 - INFO - codeparrot_training - Step 45481: {'lr': 1.0855346687918543e-05, 'samples': 23286784, 'steps': 45481, 'loss/train': 0.5191765427589417}
+02/26/2022 12:29:35 - INFO - codeparrot_training - Step 45482: {'lr': 1.0850577960549307e-05, 'samples': 23287296, 'steps': 45482, 'loss/train': 0.7215321063995361}
+02/26/2022 12:29:40 - INFO - codeparrot_training - Step 45483: {'lr': 1.0845810257620242e-05, 'samples': 23287808, 'steps': 45483, 'loss/train': 1.601241111755371}
+02/26/2022 12:29:44 - INFO - codeparrot_training - Step 45484: {'lr': 1.0841043579151777e-05, 'samples': 23288320, 'steps': 45484, 'loss/train': 1.6214818954467773}
+02/26/2022 12:29:51 - INFO - codeparrot_training - Step 45485: {'lr': 1.0836277925164368e-05, 'samples': 23288832, 'steps': 45485, 'loss/train': 2.210165500640869}
+02/26/2022 12:29:54 - INFO - codeparrot_training - Step 45486: {'lr': 1.0831513295678385e-05, 'samples': 23289344, 'steps': 45486, 'loss/train': 1.294748067855835}
+02/26/2022 12:30:00 - INFO - codeparrot_training - Step 45487: {'lr': 1.082674969071426e-05, 'samples': 23289856, 'steps': 45487, 'loss/train': 0.06890608370304108}
+02/26/2022 12:30:03 - INFO - codeparrot_training - Step 45488: {'lr': 1.0821987110292364e-05, 'samples': 23290368, 'steps': 45488, 'loss/train': 1.5776761770248413}
+02/26/2022 12:30:06 - INFO - codeparrot_training - Step 45489: {'lr': 1.081722555443318e-05, 'samples': 23290880, 'steps': 45489, 'loss/train': 2.2941017150878906}
+02/26/2022 12:30:12 - INFO - codeparrot_training - Step 45490: {'lr': 1.0812465023157025e-05, 'samples': 23291392, 'steps': 45490, 'loss/train': 1.2923625707626343}
+02/26/2022 12:30:15 - INFO - codeparrot_training - Step 45491: {'lr': 1.0807705516484328e-05, 'samples': 23291904, 'steps': 45491, 'loss/train': 1.4162800312042236}
+02/26/2022 12:30:21 - INFO - codeparrot_training - Step 45492: {'lr': 1.0802947034435435e-05, 'samples': 23292416, 'steps': 45492, 'loss/train': 1.673720359802246}
+02/26/2022 12:30:24 - INFO - codeparrot_training - Step 45493: {'lr': 1.0798189577030798e-05, 'samples': 23292928, 'steps': 45493, 'loss/train': 1.978043556213379}
+02/26/2022 12:30:30 - INFO - codeparrot_training - Step 45494: {'lr': 1.0793433144290766e-05, 'samples': 23293440, 'steps': 45494, 'loss/train': 2.053358554840088}
+02/26/2022 12:30:34 - INFO - codeparrot_training - Step 45495: {'lr': 1.0788677736235708e-05, 'samples': 23293952, 'steps': 45495, 'loss/train': 1.8812906742095947}
+02/26/2022 12:30:40 - INFO - codeparrot_training - Step 45496: {'lr': 1.0783923352885943e-05, 'samples': 23294464, 'steps': 45496, 'loss/train': 2.3655803203582764}
+02/26/2022 12:30:43 - INFO - codeparrot_training - Step 45497: {'lr': 1.0779169994261956e-05, 'samples': 23294976, 'steps': 45497, 'loss/train': 1.645861029624939}
+02/26/2022 12:30:49 - INFO - codeparrot_training - Step 45498: {'lr': 1.0774417660384034e-05, 'samples': 23295488, 'steps': 45498, 'loss/train': 1.7130159139633179}
+02/26/2022 12:30:52 - INFO - codeparrot_training - Step 45499: {'lr': 1.0769666351272522e-05, 'samples': 23296000, 'steps': 45499, 'loss/train': 2.355069160461426}
+02/26/2022 12:30:58 - INFO - codeparrot_training - Step 45500: {'lr': 1.0764916066947795e-05, 'samples': 23296512, 'steps': 45500, 'loss/train': 0.8156676888465881}
+02/26/2022 12:31:01 - INFO - codeparrot_training - Step 45501: {'lr': 1.0760166807430194e-05, 'samples': 23297024, 'steps': 45501, 'loss/train': 2.6515986919403076}
+02/26/2022 12:31:07 - INFO - codeparrot_training - Step 45502: {'lr': 1.0755418572740067e-05, 'samples': 23297536, 'steps': 45502, 'loss/train': 0.25662875175476074}
+02/26/2022 12:31:11 - INFO - codeparrot_training - Step 45503: {'lr': 1.075067136289784e-05, 'samples': 23298048, 'steps': 45503, 'loss/train': 1.7687629461288452}
+02/26/2022 12:31:16 - INFO - codeparrot_training - Step 45504: {'lr': 1.0745925177923665e-05, 'samples': 23298560, 'steps': 45504, 'loss/train': 4.2185163497924805}
+02/26/2022 12:31:19 - INFO - codeparrot_training - Step 45505: {'lr': 1.0741180017838026e-05, 'samples': 23299072, 'steps': 45505, 'loss/train': 2.6711742877960205}
+02/26/2022 12:31:26 - INFO - codeparrot_training - Step 45506: {'lr': 1.0736435882661156e-05, 'samples': 23299584, 'steps': 45506, 'loss/train': 1.556153416633606}
+02/26/2022 12:31:30 - INFO - codeparrot_training - Step 45507: {'lr': 1.073169277241351e-05, 'samples': 23300096, 'steps': 45507, 'loss/train': 1.7478265762329102}
+02/26/2022 12:31:33 - INFO - codeparrot_training - Step 45508: {'lr': 1.0726950687115211e-05, 'samples': 23300608, 'steps': 45508, 'loss/train': 0.37199845910072327}
+02/26/2022 12:31:39 - INFO - codeparrot_training - Step 45509: {'lr': 1.0722209626786744e-05, 'samples': 23301120, 'steps': 45509, 'loss/train': 0.9415773153305054}
+02/26/2022 12:31:44 - INFO - codeparrot_training - Step 45510: {'lr': 1.0717469591448315e-05, 'samples': 23301632, 'steps': 45510, 'loss/train': 1.8271976709365845}
+02/26/2022 12:31:48 - INFO - codeparrot_training - Step 45511: {'lr': 1.0712730581120322e-05, 'samples': 23302144, 'steps': 45511, 'loss/train': 1.9965388774871826}
+02/26/2022 12:31:53 - INFO - codeparrot_training - Step 45512: {'lr': 1.0707992595822946e-05, 'samples': 23302656, 'steps': 45512, 'loss/train': 1.8485907316207886}
+02/26/2022 12:31:57 - INFO - codeparrot_training - Step 45513: {'lr': 1.0703255635576559e-05, 'samples': 23303168, 'steps': 45513, 'loss/train': 1.9083240032196045}
+02/26/2022 12:32:02 - INFO - codeparrot_training - Step 45514: {'lr': 1.0698519700401422e-05, 'samples': 23303680, 'steps': 45514, 'loss/train': 2.5984535217285156}
+02/26/2022 12:32:06 - INFO - codeparrot_training - Step 45515: {'lr': 1.0693784790317878e-05, 'samples': 23304192, 'steps': 45515, 'loss/train': 1.425366759300232}
+02/26/2022 12:32:11 - INFO - codeparrot_training - Step 45516: {'lr': 1.0689050905346165e-05, 'samples': 23304704, 'steps': 45516, 'loss/train': 1.091313362121582}
+02/26/2022 12:32:15 - INFO - codeparrot_training - Step 45517: {'lr': 1.0684318045506597e-05, 'samples': 23305216, 'steps': 45517, 'loss/train': 1.4275914430618286}
+02/26/2022 12:32:21 - INFO - codeparrot_training - Step 45518: {'lr': 1.0679586210819354e-05, 'samples': 23305728, 'steps': 45518, 'loss/train': 1.656777262687683}
+02/26/2022 12:32:24 - INFO - codeparrot_training - Step 45519: {'lr': 1.0674855401304805e-05, 'samples': 23306240, 'steps': 45519, 'loss/train': 1.0106686353683472}
+02/26/2022 12:32:30 - INFO - codeparrot_training - Step 45520: {'lr': 1.067012561698319e-05, 'samples': 23306752, 'steps': 45520, 'loss/train': 1.3933219909667969}
+02/26/2022 12:32:33 - INFO - codeparrot_training - Step 45521: {'lr': 1.0665396857874764e-05, 'samples': 23307264, 'steps': 45521, 'loss/train': 1.9338077306747437}
+02/26/2022 12:32:39 - INFO - codeparrot_training - Step 45522: {'lr': 1.0660669123999767e-05, 'samples': 23307776, 'steps': 45522, 'loss/train': 1.6024292707443237}
+02/26/2022 12:32:42 - INFO - codeparrot_training - Step 45523: {'lr': 1.0655942415378456e-05, 'samples': 23308288, 'steps': 45523, 'loss/train': 1.117063045501709}
+02/26/2022 12:32:48 - INFO - codeparrot_training - Step 45524: {'lr': 1.0651216732031094e-05, 'samples': 23308800, 'steps': 45524, 'loss/train': 1.8036937713623047}
+02/26/2022 12:32:52 - INFO - codeparrot_training - Step 45525: {'lr': 1.0646492073977915e-05, 'samples': 23309312, 'steps': 45525, 'loss/train': 2.4579007625579834}
+02/26/2022 12:32:55 - INFO - codeparrot_training - Step 45526: {'lr': 1.064176844123918e-05, 'samples': 23309824, 'steps': 45526, 'loss/train': 2.0174057483673096}
+02/26/2022 12:33:01 - INFO - codeparrot_training - Step 45527: {'lr': 1.0637045833835069e-05, 'samples': 23310336, 'steps': 45527, 'loss/train': 0.6836249828338623}
+02/26/2022 12:33:05 - INFO - codeparrot_training - Step 45528: {'lr': 1.0632324251785869e-05, 'samples': 23310848, 'steps': 45528, 'loss/train': 2.2893130779266357}
+02/26/2022 12:33:10 - INFO - codeparrot_training - Step 45529: {'lr': 1.0627603695111787e-05, 'samples': 23311360, 'steps': 45529, 'loss/train': 0.9735177159309387}
+02/26/2022 12:33:14 - INFO - codeparrot_training - Step 45530: {'lr': 1.0622884163833058e-05, 'samples': 23311872, 'steps': 45530, 'loss/train': 1.6222349405288696}
+02/26/2022 12:33:19 - INFO - codeparrot_training - Step 45531: {'lr': 1.061816565796983e-05, 'samples': 23312384, 'steps': 45531, 'loss/train': 1.2825584411621094}
+02/26/2022 12:33:23 - INFO - codeparrot_training - Step 45532: {'lr': 1.0613448177542423e-05, 'samples': 23312896, 'steps': 45532, 'loss/train': 1.4990627765655518}
+02/26/2022 12:33:29 - INFO - codeparrot_training - Step 45533: {'lr': 1.0608731722570958e-05, 'samples': 23313408, 'steps': 45533, 'loss/train': 1.6389977931976318}
+02/26/2022 12:33:32 - INFO - codeparrot_training - Step 45534: {'lr': 1.0604016293075696e-05, 'samples': 23313920, 'steps': 45534, 'loss/train': 1.8013746738433838}
+02/26/2022 12:33:38 - INFO - codeparrot_training - Step 45535: {'lr': 1.0599301889076762e-05, 'samples': 23314432, 'steps': 45535, 'loss/train': 3.0061020851135254}
+02/26/2022 12:33:41 - INFO - codeparrot_training - Step 45536: {'lr': 1.0594588510594445e-05, 'samples': 23314944, 'steps': 45536, 'loss/train': 0.9037851691246033}
+02/26/2022 12:33:47 - INFO - codeparrot_training - Step 45537: {'lr': 1.0589876157648865e-05, 'samples': 23315456, 'steps': 45537, 'loss/train': 1.9245526790618896}
+02/26/2022 12:33:50 - INFO - codeparrot_training - Step 45538: {'lr': 1.0585164830260285e-05, 'samples': 23315968, 'steps': 45538, 'loss/train': 1.0642341375350952}
+02/26/2022 12:33:56 - INFO - codeparrot_training - Step 45539: {'lr': 1.0580454528448774e-05, 'samples': 23316480, 'steps': 45539, 'loss/train': 1.3503497838974}
+02/26/2022 12:33:59 - INFO - codeparrot_training - Step 45540: {'lr': 1.057574525223462e-05, 'samples': 23316992, 'steps': 45540, 'loss/train': 1.8431942462921143}
+02/26/2022 12:34:05 - INFO - codeparrot_training - Step 45541: {'lr': 1.057103700163789e-05, 'samples': 23317504, 'steps': 45541, 'loss/train': 2.5416059494018555}
+02/26/2022 12:34:08 - INFO - codeparrot_training - Step 45542: {'lr': 1.0566329776678874e-05, 'samples': 23318016, 'steps': 45542, 'loss/train': 1.654799461364746}
+02/26/2022 12:34:15 - INFO - codeparrot_training - Step 45543: {'lr': 1.0561623577377639e-05, 'samples': 23318528, 'steps': 45543, 'loss/train': 2.061863899230957}
+02/26/2022 12:34:18 - INFO - codeparrot_training - Step 45544: {'lr': 1.055691840375439e-05, 'samples': 23319040, 'steps': 45544, 'loss/train': 7.775964736938477}
+02/26/2022 12:34:24 - INFO - codeparrot_training - Step 45545: {'lr': 1.0552214255829224e-05, 'samples': 23319552, 'steps': 45545, 'loss/train': 2.3264169692993164}
+02/26/2022 12:34:27 - INFO - codeparrot_training - Step 45546: {'lr': 1.0547511133622428e-05, 'samples': 23320064, 'steps': 45546, 'loss/train': 1.6504597663879395}
+02/26/2022 12:34:33 - INFO - codeparrot_training - Step 45547: {'lr': 1.0542809037153989e-05, 'samples': 23320576, 'steps': 45547, 'loss/train': 1.3006246089935303}
+02/26/2022 12:34:36 - INFO - codeparrot_training - Step 45548: {'lr': 1.0538107966444138e-05, 'samples': 23321088, 'steps': 45548, 'loss/train': 2.3742191791534424}
+02/26/2022 12:34:42 - INFO - codeparrot_training - Step 45549: {'lr': 1.0533407921513e-05, 'samples': 23321600, 'steps': 45549, 'loss/train': 1.9453610181808472}
+02/26/2022 12:34:45 - INFO - codeparrot_training - Step 45550: {'lr': 1.0528708902380696e-05, 'samples': 23322112, 'steps': 45550, 'loss/train': 1.8022223711013794}
+02/26/2022 12:34:51 - INFO - codeparrot_training - Step 45551: {'lr': 1.0524010909067377e-05, 'samples': 23322624, 'steps': 45551, 'loss/train': 0.34935805201530457}
+02/26/2022 12:34:54 - INFO - codeparrot_training - Step 45552: {'lr': 1.0519313941593139e-05, 'samples': 23323136, 'steps': 45552, 'loss/train': 1.9266070127487183}
+02/26/2022 12:35:01 - INFO - codeparrot_training - Step 45553: {'lr': 1.0514617999978104e-05, 'samples': 23323648, 'steps': 45553, 'loss/train': 1.1075153350830078}
+02/26/2022 12:35:05 - INFO - codeparrot_training - Step 45554: {'lr': 1.0509923084242423e-05, 'samples': 23324160, 'steps': 45554, 'loss/train': 2.091653823852539}
+02/26/2022 12:35:10 - INFO - codeparrot_training - Step 45555: {'lr': 1.0505229194406191e-05, 'samples': 23324672, 'steps': 45555, 'loss/train': 1.7227206230163574}
+02/26/2022 12:35:14 - INFO - codeparrot_training - Step 45556: {'lr': 1.0500536330489502e-05, 'samples': 23325184, 'steps': 45556, 'loss/train': 1.3503150939941406}
+02/26/2022 12:35:19 - INFO - codeparrot_training - Step 45557: {'lr': 1.0495844492512451e-05, 'samples': 23325696, 'steps': 45557, 'loss/train': 2.2403905391693115}
+02/26/2022 12:35:23 - INFO - codeparrot_training - Step 45558: {'lr': 1.0491153680495164e-05, 'samples': 23326208, 'steps': 45558, 'loss/train': 1.721691608428955}
+02/26/2022 12:35:28 - INFO - codeparrot_training - Step 45559: {'lr': 1.0486463894457732e-05, 'samples': 23326720, 'steps': 45559, 'loss/train': 2.797123432159424}
+02/26/2022 12:35:32 - INFO - codeparrot_training - Step 45560: {'lr': 1.0481775134420224e-05, 'samples': 23327232, 'steps': 45560, 'loss/train': 0.14029265940189362}
+02/26/2022 12:35:37 - INFO - codeparrot_training - Step 45561: {'lr': 1.0477087400402707e-05, 'samples': 23327744, 'steps': 45561, 'loss/train': 2.30580997467041}
+02/26/2022 12:35:41 - INFO - codeparrot_training - Step 45562: {'lr': 1.0472400692425333e-05, 'samples': 23328256, 'steps': 45562, 'loss/train': 2.5527184009552}
+02/26/2022 12:35:46 - INFO - codeparrot_training - Step 45563: {'lr': 1.0467715010508139e-05, 'samples': 23328768, 'steps': 45563, 'loss/train': 1.0978443622589111}
+02/26/2022 12:35:50 - INFO - codeparrot_training - Step 45564: {'lr': 1.0463030354671165e-05, 'samples': 23329280, 'steps': 45564, 'loss/train': 2.0104455947875977}
+02/26/2022 12:35:56 - INFO - codeparrot_training - Step 45565: {'lr': 1.0458346724934509e-05, 'samples': 23329792, 'steps': 45565, 'loss/train': 1.9947469234466553}
+02/26/2022 12:36:00 - INFO - codeparrot_training - Step 45566: {'lr': 1.0453664121318235e-05, 'samples': 23330304, 'steps': 45566, 'loss/train': 0.20785105228424072}
+02/26/2022 12:36:05 - INFO - codeparrot_training - Step 45567: {'lr': 1.0448982543842412e-05, 'samples': 23330816, 'steps': 45567, 'loss/train': 1.6825751066207886}
+02/26/2022 12:36:09 - INFO - codeparrot_training - Step 45568: {'lr': 1.0444301992527077e-05, 'samples': 23331328, 'steps': 45568, 'loss/train': 2.8616325855255127}
+02/26/2022 12:36:14 - INFO - codeparrot_training - Step 45569: {'lr': 1.0439622467392302e-05, 'samples': 23331840, 'steps': 45569, 'loss/train': 2.049180030822754}
+02/26/2022 12:36:18 - INFO - codeparrot_training - Step 45570: {'lr': 1.0434943968458066e-05, 'samples': 23332352, 'steps': 45570, 'loss/train': 1.535075306892395}
+02/26/2022 12:36:23 - INFO - codeparrot_training - Step 45571: {'lr': 1.0430266495744495e-05, 'samples': 23332864, 'steps': 45571, 'loss/train': 1.9570480585098267}
+02/26/2022 12:36:27 - INFO - codeparrot_training - Step 45572: {'lr': 1.04255900492716e-05, 'samples': 23333376, 'steps': 45572, 'loss/train': 0.7161442637443542}
+02/26/2022 12:36:32 - INFO - codeparrot_training - Step 45573: {'lr': 1.0420914629059391e-05, 'samples': 23333888, 'steps': 45573, 'loss/train': 2.4145686626434326}
+02/26/2022 12:36:36 - INFO - codeparrot_training - Step 45574: {'lr': 1.0416240235127882e-05, 'samples': 23334400, 'steps': 45574, 'loss/train': 1.8772698640823364}
+02/26/2022 12:36:41 - INFO - codeparrot_training - Step 45575: {'lr': 1.0411566867497169e-05, 'samples': 23334912, 'steps': 45575, 'loss/train': 1.1102863550186157}
+02/26/2022 12:36:45 - INFO - codeparrot_training - Step 45576: {'lr': 1.0406894526187177e-05, 'samples': 23335424, 'steps': 45576, 'loss/train': 1.2965295314788818}
+02/26/2022 12:36:50 - INFO - codeparrot_training - Step 45577: {'lr': 1.0402223211218031e-05, 'samples': 23335936, 'steps': 45577, 'loss/train': 0.15410499274730682}
+02/26/2022 12:36:54 - INFO - codeparrot_training - Step 45578: {'lr': 1.0397552922609632e-05, 'samples': 23336448, 'steps': 45578, 'loss/train': 1.8596950769424438}
+02/26/2022 12:36:59 - INFO - codeparrot_training - Step 45579: {'lr': 1.0392883660382046e-05, 'samples': 23336960, 'steps': 45579, 'loss/train': 1.4166401624679565}
+02/26/2022 12:37:03 - INFO - codeparrot_training - Step 45580: {'lr': 1.038821542455523e-05, 'samples': 23337472, 'steps': 45580, 'loss/train': 1.94240140914917}
+02/26/2022 12:37:10 - INFO - codeparrot_training - Step 45581: {'lr': 1.0383548215149307e-05, 'samples': 23337984, 'steps': 45581, 'loss/train': 2.0994479656219482}
+02/26/2022 12:37:13 - INFO - codeparrot_training - Step 45582: {'lr': 1.0378882032184095e-05, 'samples': 23338496, 'steps': 45582, 'loss/train': 3.729790210723877}
+02/26/2022 12:37:19 - INFO - codeparrot_training - Step 45583: {'lr': 1.0374216875679688e-05, 'samples': 23339008, 'steps': 45583, 'loss/train': 1.7085816860198975}
+02/26/2022 12:37:22 - INFO - codeparrot_training - Step 45584: {'lr': 1.0369552745656014e-05, 'samples': 23339520, 'steps': 45584, 'loss/train': 2.3823957443237305}
+02/26/2022 12:37:28 - INFO - codeparrot_training - Step 45585: {'lr': 1.0364889642133141e-05, 'samples': 23340032, 'steps': 45585, 'loss/train': 2.9953062534332275}
+02/26/2022 12:37:31 - INFO - codeparrot_training - Step 45586: {'lr': 1.036022756513097e-05, 'samples': 23340544, 'steps': 45586, 'loss/train': 1.8608744144439697}
+02/26/2022 12:37:37 - INFO - codeparrot_training - Step 45587: {'lr': 1.0355566514669484e-05, 'samples': 23341056, 'steps': 45587, 'loss/train': 1.9176645278930664}
+02/26/2022 12:37:40 - INFO - codeparrot_training - Step 45588: {'lr': 1.035090649076864e-05, 'samples': 23341568, 'steps': 45588, 'loss/train': 1.3425835371017456}
+02/26/2022 12:37:46 - INFO - codeparrot_training - Step 45589: {'lr': 1.0346247493448425e-05, 'samples': 23342080, 'steps': 45589, 'loss/train': 0.6975095868110657}
+02/26/2022 12:37:50 - INFO - codeparrot_training - Step 45590: {'lr': 1.034158952272879e-05, 'samples': 23342592, 'steps': 45590, 'loss/train': 2.5314528942108154}
+02/26/2022 12:37:55 - INFO - codeparrot_training - Step 45591: {'lr': 1.0336932578629692e-05, 'samples': 23343104, 'steps': 45591, 'loss/train': 1.6060001850128174}
+02/26/2022 12:37:59 - INFO - codeparrot_training - Step 45592: {'lr': 1.0332276661171064e-05, 'samples': 23343616, 'steps': 45592, 'loss/train': 3.0115532875061035}
+02/26/2022 12:38:05 - INFO - codeparrot_training - Step 45593: {'lr': 1.0327621770372859e-05, 'samples': 23344128, 'steps': 45593, 'loss/train': 0.7314513325691223}
+02/26/2022 12:38:08 - INFO - codeparrot_training - Step 45594: {'lr': 1.0322967906255032e-05, 'samples': 23344640, 'steps': 45594, 'loss/train': 1.1862751245498657}
+02/26/2022 12:38:12 - INFO - codeparrot_training - Step 45595: {'lr': 1.0318315068837514e-05, 'samples': 23345152, 'steps': 45595, 'loss/train': 3.2045083045959473}
+02/26/2022 12:38:18 - INFO - codeparrot_training - Step 45596: {'lr': 1.0313663258140177e-05, 'samples': 23345664, 'steps': 45596, 'loss/train': 1.443165898323059}
+02/26/2022 12:38:21 - INFO - codeparrot_training - Step 45597: {'lr': 1.0309012474183032e-05, 'samples': 23346176, 'steps': 45597, 'loss/train': 1.5422154664993286}
+02/26/2022 12:38:27 - INFO - codeparrot_training - Step 45598: {'lr': 1.0304362716985955e-05, 'samples': 23346688, 'steps': 45598, 'loss/train': 0.7026045918464661}
+02/26/2022 12:38:30 - INFO - codeparrot_training - Step 45599: {'lr': 1.0299713986568926e-05, 'samples': 23347200, 'steps': 45599, 'loss/train': 2.0096731185913086}
+02/26/2022 12:38:37 - INFO - codeparrot_training - Step 45600: {'lr': 1.0295066282951737e-05, 'samples': 23347712, 'steps': 45600, 'loss/train': 1.1956053972244263}
+02/26/2022 12:38:40 - INFO - codeparrot_training - Step 45601: {'lr': 1.02904196061544e-05, 'samples': 23348224, 'steps': 45601, 'loss/train': 1.2708628177642822}
+02/26/2022 12:38:46 - INFO - codeparrot_training - Step 45602: {'lr': 1.0285773956196814e-05, 'samples': 23348736, 'steps': 45602, 'loss/train': 1.9006245136260986}
+02/26/2022 12:38:49 - INFO - codeparrot_training - Step 45603: {'lr': 1.0281129333098827e-05, 'samples': 23349248, 'steps': 45603, 'loss/train': 0.837859570980072}
+02/26/2022 12:38:55 - INFO - codeparrot_training - Step 45604: {'lr': 1.0276485736880337e-05, 'samples': 23349760, 'steps': 45604, 'loss/train': 1.5647293329238892}
+02/26/2022 12:38:58 - INFO - codeparrot_training - Step 45605: {'lr': 1.02718431675613e-05, 'samples': 23350272, 'steps': 45605, 'loss/train': 1.5073418617248535}
+02/26/2022 12:39:04 - INFO - codeparrot_training - Step 45606: {'lr': 1.0267201625161566e-05, 'samples': 23350784, 'steps': 45606, 'loss/train': 1.3233749866485596}
+02/26/2022 12:39:07 - INFO - codeparrot_training - Step 45607: {'lr': 1.0262561109701001e-05, 'samples': 23351296, 'steps': 45607, 'loss/train': 1.6218550205230713}
+02/26/2022 12:39:13 - INFO - codeparrot_training - Step 45608: {'lr': 1.0257921621199484e-05, 'samples': 23351808, 'steps': 45608, 'loss/train': 1.9633405208587646}
+02/26/2022 12:39:16 - INFO - codeparrot_training - Step 45609: {'lr': 1.0253283159676941e-05, 'samples': 23352320, 'steps': 45609, 'loss/train': 1.0450513362884521}
+02/26/2022 12:39:22 - INFO - codeparrot_training - Step 45610: {'lr': 1.0248645725153189e-05, 'samples': 23352832, 'steps': 45610, 'loss/train': 1.9641263484954834}
+02/26/2022 12:39:25 - INFO - codeparrot_training - Step 45611: {'lr': 1.0244009317648074e-05, 'samples': 23353344, 'steps': 45611, 'loss/train': 2.4277329444885254}
+02/26/2022 12:39:32 - INFO - codeparrot_training - Step 45612: {'lr': 1.0239373937181551e-05, 'samples': 23353856, 'steps': 45612, 'loss/train': 1.9305319786071777}
+02/26/2022 12:39:35 - INFO - codeparrot_training - Step 45613: {'lr': 1.0234739583773383e-05, 'samples': 23354368, 'steps': 45613, 'loss/train': 1.3441033363342285}
+02/26/2022 12:39:41 - INFO - codeparrot_training - Step 45614: {'lr': 1.0230106257443472e-05, 'samples': 23354880, 'steps': 45614, 'loss/train': 1.438027024269104}
+02/26/2022 12:39:46 - INFO - codeparrot_training - Step 45615: {'lr': 1.0225473958211634e-05, 'samples': 23355392, 'steps': 45615, 'loss/train': 1.5967894792556763}
+02/26/2022 12:39:50 - INFO - codeparrot_training - Step 45616: {'lr': 1.0220842686097741e-05, 'samples': 23355904, 'steps': 45616, 'loss/train': 1.7690527439117432}
+02/26/2022 12:39:55 - INFO - codeparrot_training - Step 45617: {'lr': 1.0216212441121615e-05, 'samples': 23356416, 'steps': 45617, 'loss/train': 0.6522355675697327}
+02/26/2022 12:39:59 - INFO - codeparrot_training - Step 45618: {'lr': 1.0211583223303095e-05, 'samples': 23356928, 'steps': 45618, 'loss/train': 2.0137455463409424}
+02/26/2022 12:40:04 - INFO - codeparrot_training - Step 45619: {'lr': 1.0206955032662001e-05, 'samples': 23357440, 'steps': 45619, 'loss/train': 0.7497518062591553}
+02/26/2022 12:40:08 - INFO - codeparrot_training - Step 45620: {'lr': 1.0202327869218208e-05, 'samples': 23357952, 'steps': 45620, 'loss/train': 1.5751317739486694}
+02/26/2022 12:40:13 - INFO - codeparrot_training - Step 45621: {'lr': 1.0197701732991448e-05, 'samples': 23358464, 'steps': 45621, 'loss/train': 1.2662396430969238}
+02/26/2022 12:40:17 - INFO - codeparrot_training - Step 45622: {'lr': 1.0193076624001624e-05, 'samples': 23358976, 'steps': 45622, 'loss/train': 1.7119075059890747}
+02/26/2022 12:40:22 - INFO - codeparrot_training - Step 45623: {'lr': 1.0188452542268468e-05, 'samples': 23359488, 'steps': 45623, 'loss/train': 1.2963422536849976}
+02/26/2022 12:40:26 - INFO - codeparrot_training - Step 45624: {'lr': 1.018382948781188e-05, 'samples': 23360000, 'steps': 45624, 'loss/train': 1.2522579431533813}
+02/26/2022 12:40:32 - INFO - codeparrot_training - Step 45625: {'lr': 1.0179207460651596e-05, 'samples': 23360512, 'steps': 45625, 'loss/train': 1.1295278072357178}
+02/26/2022 12:40:35 - INFO - codeparrot_training - Step 45626: {'lr': 1.017458646080746e-05, 'samples': 23361024, 'steps': 45626, 'loss/train': 1.0604910850524902}
+02/26/2022 12:40:41 - INFO - codeparrot_training - Step 45627: {'lr': 1.0169966488299181e-05, 'samples': 23361536, 'steps': 45627, 'loss/train': 2.5549004077911377}
+02/26/2022 12:40:44 - INFO - codeparrot_training - Step 45628: {'lr': 1.0165347543146685e-05, 'samples': 23362048, 'steps': 45628, 'loss/train': 1.8073127269744873}
+02/26/2022 12:40:50 - INFO - codeparrot_training - Step 45629: {'lr': 1.0160729625369653e-05, 'samples': 23362560, 'steps': 45629, 'loss/train': 0.17217500507831573}
+02/26/2022 12:40:53 - INFO - codeparrot_training - Step 45630: {'lr': 1.0156112734987899e-05, 'samples': 23363072, 'steps': 45630, 'loss/train': 0.9125199317932129}
+02/26/2022 12:40:59 - INFO - codeparrot_training - Step 45631: {'lr': 1.015149687202116e-05, 'samples': 23363584, 'steps': 45631, 'loss/train': 2.1482908725738525}
+02/26/2022 12:41:02 - INFO - codeparrot_training - Step 45632: {'lr': 1.0146882036489307e-05, 'samples': 23364096, 'steps': 45632, 'loss/train': 1.8630468845367432}
+02/26/2022 12:41:08 - INFO - codeparrot_training - Step 45633: {'lr': 1.0142268228412021e-05, 'samples': 23364608, 'steps': 45633, 'loss/train': 1.496239185333252}
+02/26/2022 12:41:12 - INFO - codeparrot_training - Step 45634: {'lr': 1.0137655447809119e-05, 'samples': 23365120, 'steps': 45634, 'loss/train': 1.337036371231079}
+02/26/2022 12:41:15 - INFO - codeparrot_training - Step 45635: {'lr': 1.0133043694700277e-05, 'samples': 23365632, 'steps': 45635, 'loss/train': 2.2077653408050537}
+02/26/2022 12:41:21 - INFO - codeparrot_training - Step 45636: {'lr': 1.0128432969105373e-05, 'samples': 23366144, 'steps': 45636, 'loss/train': 1.3853659629821777}
+02/26/2022 12:41:24 - INFO - codeparrot_training - Step 45637: {'lr': 1.0123823271044025e-05, 'samples': 23366656, 'steps': 45637, 'loss/train': 1.3224844932556152}
+02/26/2022 12:41:31 - INFO - codeparrot_training - Step 45638: {'lr': 1.0119214600536135e-05, 'samples': 23367168, 'steps': 45638, 'loss/train': 1.3349225521087646}
+02/26/2022 12:41:34 - INFO - codeparrot_training - Step 45639: {'lr': 1.0114606957601302e-05, 'samples': 23367680, 'steps': 45639, 'loss/train': 1.48549222946167}
+02/26/2022 12:41:40 - INFO - codeparrot_training - Step 45640: {'lr': 1.011000034225934e-05, 'samples': 23368192, 'steps': 45640, 'loss/train': 2.0417563915252686}
+02/26/2022 12:41:43 - INFO - codeparrot_training - Step 45641: {'lr': 1.010539475452993e-05, 'samples': 23368704, 'steps': 45641, 'loss/train': 0.6281805634498596}
+02/26/2022 12:41:49 - INFO - codeparrot_training - Step 45642: {'lr': 1.0100790194432912e-05, 'samples': 23369216, 'steps': 45642, 'loss/train': 1.8049238920211792}
+02/26/2022 12:41:52 - INFO - codeparrot_training - Step 45643: {'lr': 1.0096186661987861e-05, 'samples': 23369728, 'steps': 45643, 'loss/train': 0.6389544010162354}
+02/26/2022 12:41:58 - INFO - codeparrot_training - Step 45644: {'lr': 1.009158415721459e-05, 'samples': 23370240, 'steps': 45644, 'loss/train': 1.5484758615493774}
+02/26/2022 12:42:01 - INFO - codeparrot_training - Step 45645: {'lr': 1.0086982680132778e-05, 'samples': 23370752, 'steps': 45645, 'loss/train': 3.094593048095703}
+02/26/2022 12:42:07 - INFO - codeparrot_training - Step 45646: {'lr': 1.008238223076216e-05, 'samples': 23371264, 'steps': 45646, 'loss/train': 2.4040956497192383}
+02/26/2022 12:42:11 - INFO - codeparrot_training - Step 45647: {'lr': 1.0077782809122471e-05, 'samples': 23371776, 'steps': 45647, 'loss/train': 1.758974313735962}
+02/26/2022 12:42:17 - INFO - codeparrot_training - Step 45648: {'lr': 1.0073184415233333e-05, 'samples': 23372288, 'steps': 45648, 'loss/train': 2.1329472064971924}
+02/26/2022 12:42:21 - INFO - codeparrot_training - Step 45649: {'lr': 1.0068587049114507e-05, 'samples': 23372800, 'steps': 45649, 'loss/train': 0.2544361650943756}
+02/26/2022 12:42:26 - INFO - codeparrot_training - Step 45650: {'lr': 1.0063990710785648e-05, 'samples': 23373312, 'steps': 45650, 'loss/train': 1.8752689361572266}
+02/26/2022 12:42:30 - INFO - codeparrot_training - Step 45651: {'lr': 1.0059395400266485e-05, 'samples': 23373824, 'steps': 45651, 'loss/train': 2.842592477798462}
+02/26/2022 12:42:35 - INFO - codeparrot_training - Step 45652: {'lr': 1.0054801117576672e-05, 'samples': 23374336, 'steps': 45652, 'loss/train': 1.9933407306671143}
+02/26/2022 12:42:39 - INFO - codeparrot_training - Step 45653: {'lr': 1.0050207862735916e-05, 'samples': 23374848, 'steps': 45653, 'loss/train': 1.311305284500122}
+02/26/2022 12:42:44 - INFO - codeparrot_training - Step 45654: {'lr': 1.0045615635763838e-05, 'samples': 23375360, 'steps': 45654, 'loss/train': 1.054903507232666}
+02/26/2022 12:42:48 - INFO - codeparrot_training - Step 45655: {'lr': 1.0041024436680174e-05, 'samples': 23375872, 'steps': 45655, 'loss/train': 0.29673346877098083}
+02/26/2022 12:42:53 - INFO - codeparrot_training - Step 45656: {'lr': 1.0036434265504574e-05, 'samples': 23376384, 'steps': 45656, 'loss/train': 2.4695589542388916}
+02/26/2022 12:42:57 - INFO - codeparrot_training - Step 45657: {'lr': 1.0031845122256716e-05, 'samples': 23376896, 'steps': 45657, 'loss/train': 1.6470028162002563}
+02/26/2022 12:43:02 - INFO - codeparrot_training - Step 45658: {'lr': 1.002725700695617e-05, 'samples': 23377408, 'steps': 45658, 'loss/train': 0.3379676043987274}
+02/26/2022 12:43:06 - INFO - codeparrot_training - Step 45659: {'lr': 1.0022669919622695e-05, 'samples': 23377920, 'steps': 45659, 'loss/train': 2.492748498916626}
+02/26/2022 12:43:11 - INFO - codeparrot_training - Step 45660: {'lr': 1.0018083860275889e-05, 'samples': 23378432, 'steps': 45660, 'loss/train': 1.9696780443191528}
+02/26/2022 12:43:15 - INFO - codeparrot_training - Step 45661: {'lr': 1.001349882893543e-05, 'samples': 23378944, 'steps': 45661, 'loss/train': 1.0503180027008057}
+02/26/2022 12:43:21 - INFO - codeparrot_training - Step 45662: {'lr': 1.0008914825620913e-05, 'samples': 23379456, 'steps': 45662, 'loss/train': 0.06641270220279694}
+02/26/2022 12:43:24 - INFO - codeparrot_training - Step 45663: {'lr': 1.0004331850352017e-05, 'samples': 23379968, 'steps': 45663, 'loss/train': 2.2936954498291016}
+02/26/2022 12:43:30 - INFO - codeparrot_training - Step 45664: {'lr': 9.999749903148337e-06, 'samples': 23380480, 'steps': 45664, 'loss/train': 0.952839732170105}
+02/26/2022 12:43:33 - INFO - codeparrot_training - Step 45665: {'lr': 9.995168984029552e-06, 'samples': 23380992, 'steps': 45665, 'loss/train': 1.309648036956787}
+02/26/2022 12:43:39 - INFO - codeparrot_training - Step 45666: {'lr': 9.9905890930152e-06, 'samples': 23381504, 'steps': 45666, 'loss/train': 2.489650011062622}
+02/26/2022 12:43:42 - INFO - codeparrot_training - Step 45667: {'lr': 9.986010230124992e-06, 'samples': 23382016, 'steps': 45667, 'loss/train': 2.2035741806030273}
+02/26/2022 12:43:48 - INFO - codeparrot_training - Step 45668: {'lr': 9.981432395378493e-06, 'samples': 23382528, 'steps': 45668, 'loss/train': 1.8708242177963257}
+02/26/2022 12:43:51 - INFO - codeparrot_training - Step 45669: {'lr': 9.976855588795352e-06, 'samples': 23383040, 'steps': 45669, 'loss/train': 1.7897173166275024}
+02/26/2022 12:43:57 - INFO - codeparrot_training - Step 45670: {'lr': 9.972279810395085e-06, 'samples': 23383552, 'steps': 45670, 'loss/train': 1.3311960697174072}
+02/26/2022 12:44:00 - INFO - codeparrot_training - Step 45671: {'lr': 9.967705060197396e-06, 'samples': 23384064, 'steps': 45671, 'loss/train': 2.46360445022583}
+02/26/2022 12:44:06 - INFO - codeparrot_training - Step 45672: {'lr': 9.963131338221798e-06, 'samples': 23384576, 'steps': 45672, 'loss/train': 1.1261680126190186}
+02/26/2022 12:44:09 - INFO - codeparrot_training - Step 45673: {'lr': 9.958558644487998e-06, 'samples': 23385088, 'steps': 45673, 'loss/train': 1.9089025259017944}
+02/26/2022 12:44:15 - INFO - codeparrot_training - Step 45674: {'lr': 9.953986979015451e-06, 'samples': 23385600, 'steps': 45674, 'loss/train': 1.5332258939743042}
+02/26/2022 12:44:19 - INFO - codeparrot_training - Step 45675: {'lr': 9.949416341823836e-06, 'samples': 23386112, 'steps': 45675, 'loss/train': 1.4545354843139648}
+02/26/2022 12:44:24 - INFO - codeparrot_training - Step 45676: {'lr': 9.944846732932667e-06, 'samples': 23386624, 'steps': 45676, 'loss/train': 2.3801071643829346}
+02/26/2022 12:44:28 - INFO - codeparrot_training - Step 45677: {'lr': 9.940278152361593e-06, 'samples': 23387136, 'steps': 45677, 'loss/train': 2.2263152599334717}
+02/26/2022 12:44:33 - INFO - codeparrot_training - Step 45678: {'lr': 9.935710600130071e-06, 'samples': 23387648, 'steps': 45678, 'loss/train': 1.2591277360916138}
+02/26/2022 12:44:36 - INFO - codeparrot_training - Step 45679: {'lr': 9.93114407625778e-06, 'samples': 23388160, 'steps': 45679, 'loss/train': 1.8275260925292969}
+02/26/2022 12:44:42 - INFO - codeparrot_training - Step 45680: {'lr': 9.926578580764234e-06, 'samples': 23388672, 'steps': 45680, 'loss/train': 0.9706999659538269}
+02/26/2022 12:44:46 - INFO - codeparrot_training - Step 45681: {'lr': 9.922014113669025e-06, 'samples': 23389184, 'steps': 45681, 'loss/train': 2.613708019256592}
+02/26/2022 12:44:51 - INFO - codeparrot_training - Step 45682: {'lr': 9.917450674991612e-06, 'samples': 23389696, 'steps': 45682, 'loss/train': 1.945132851600647}
+02/26/2022 12:44:54 - INFO - codeparrot_training - Step 45683: {'lr': 9.912888264751645e-06, 'samples': 23390208, 'steps': 45683, 'loss/train': 1.57122802734375}
+02/26/2022 12:45:01 - INFO - codeparrot_training - Step 45684: {'lr': 9.908326882968582e-06, 'samples': 23390720, 'steps': 45684, 'loss/train': 1.5135960578918457}
+02/26/2022 12:45:04 - INFO - codeparrot_training - Step 45685: {'lr': 9.903766529662045e-06, 'samples': 23391232, 'steps': 45685, 'loss/train': 2.096989870071411}
+02/26/2022 12:45:10 - INFO - codeparrot_training - Step 45686: {'lr': 9.899207204851546e-06, 'samples': 23391744, 'steps': 45686, 'loss/train': 2.0623619556427}
+02/26/2022 12:45:13 - INFO - codeparrot_training - Step 45687: {'lr': 9.894648908556597e-06, 'samples': 23392256, 'steps': 45687, 'loss/train': 1.8422881364822388}
+02/26/2022 12:45:19 - INFO - codeparrot_training - Step 45688: {'lr': 9.89009164079671e-06, 'samples': 23392768, 'steps': 45688, 'loss/train': 0.7527719736099243}
+02/26/2022 12:45:22 - INFO - codeparrot_training - Step 45689: {'lr': 9.885535401591456e-06, 'samples': 23393280, 'steps': 45689, 'loss/train': 1.548490047454834}
+02/26/2022 12:45:28 - INFO - codeparrot_training - Step 45690: {'lr': 9.880980190960315e-06, 'samples': 23393792, 'steps': 45690, 'loss/train': 1.3348819017410278}
+02/26/2022 12:45:31 - INFO - codeparrot_training - Step 45691: {'lr': 9.876426008922801e-06, 'samples': 23394304, 'steps': 45691, 'loss/train': 2.257467746734619}
+02/26/2022 12:45:37 - INFO - codeparrot_training - Step 45692: {'lr': 9.871872855498399e-06, 'samples': 23394816, 'steps': 45692, 'loss/train': 1.1500526666641235}
+02/26/2022 12:45:40 - INFO - codeparrot_training - Step 45693: {'lr': 9.867320730706674e-06, 'samples': 23395328, 'steps': 45693, 'loss/train': 1.1437115669250488}
+02/26/2022 12:45:47 - INFO - codeparrot_training - Step 45694: {'lr': 9.862769634567114e-06, 'samples': 23395840, 'steps': 45694, 'loss/train': 2.3052122592926025}
+02/26/2022 12:45:50 - INFO - codeparrot_training - Step 45695: {'lr': 9.858219567099175e-06, 'samples': 23396352, 'steps': 45695, 'loss/train': 1.3616516590118408}
+02/26/2022 12:45:56 - INFO - codeparrot_training - Step 45696: {'lr': 9.853670528322367e-06, 'samples': 23396864, 'steps': 45696, 'loss/train': 2.3466379642486572}
+02/26/2022 12:45:59 - INFO - codeparrot_training - Step 45697: {'lr': 9.849122518256205e-06, 'samples': 23397376, 'steps': 45697, 'loss/train': 2.5476202964782715}
+02/26/2022 12:46:05 - INFO - codeparrot_training - Step 45698: {'lr': 9.844575536920113e-06, 'samples': 23397888, 'steps': 45698, 'loss/train': 1.311321496963501}
+02/26/2022 12:46:08 - INFO - codeparrot_training - Step 45699: {'lr': 9.840029584333637e-06, 'samples': 23398400, 'steps': 45699, 'loss/train': 1.832392930984497}
+02/26/2022 12:46:14 - INFO - codeparrot_training - Step 45700: {'lr': 9.835484660516203e-06, 'samples': 23398912, 'steps': 45700, 'loss/train': 1.4693660736083984}
+02/26/2022 12:46:17 - INFO - codeparrot_training - Step 45701: {'lr': 9.83094076548724e-06, 'samples': 23399424, 'steps': 45701, 'loss/train': 0.95565265417099}
+02/26/2022 12:46:23 - INFO - codeparrot_training - Step 45702: {'lr': 9.826397899266315e-06, 'samples': 23399936, 'steps': 45702, 'loss/train': 1.0762507915496826}
+02/26/2022 12:46:26 - INFO - codeparrot_training - Step 45703: {'lr': 9.821856061872858e-06, 'samples': 23400448, 'steps': 45703, 'loss/train': 2.089848041534424}
+02/26/2022 12:46:32 - INFO - codeparrot_training - Step 45704: {'lr': 9.81731525332627e-06, 'samples': 23400960, 'steps': 45704, 'loss/train': 1.6372756958007812}
+02/26/2022 12:46:35 - INFO - codeparrot_training - Step 45705: {'lr': 9.812775473646035e-06, 'samples': 23401472, 'steps': 45705, 'loss/train': 1.3960237503051758}
+02/26/2022 12:46:41 - INFO - codeparrot_training - Step 45706: {'lr': 9.80823672285161e-06, 'samples': 23401984, 'steps': 45706, 'loss/train': 1.920245885848999}
+02/26/2022 12:46:44 - INFO - codeparrot_training - Step 45707: {'lr': 9.803699000962397e-06, 'samples': 23402496, 'steps': 45707, 'loss/train': 1.6112788915634155}
+02/26/2022 12:46:51 - INFO - codeparrot_training - Step 45708: {'lr': 9.799162307997933e-06, 'samples': 23403008, 'steps': 45708, 'loss/train': 2.238391399383545}
+02/26/2022 12:46:54 - INFO - codeparrot_training - Step 45709: {'lr': 9.794626643977539e-06, 'samples': 23403520, 'steps': 45709, 'loss/train': 1.5660067796707153}
+02/26/2022 12:47:00 - INFO - codeparrot_training - Step 45710: {'lr': 9.790092008920698e-06, 'samples': 23404032, 'steps': 45710, 'loss/train': 1.097868800163269}
+02/26/2022 12:47:03 - INFO - codeparrot_training - Step 45711: {'lr': 9.785558402846811e-06, 'samples': 23404544, 'steps': 45711, 'loss/train': 2.1826376914978027}
+02/26/2022 12:47:09 - INFO - codeparrot_training - Step 45712: {'lr': 9.781025825775392e-06, 'samples': 23405056, 'steps': 45712, 'loss/train': 0.9298344850540161}
+02/26/2022 12:47:12 - INFO - codeparrot_training - Step 45713: {'lr': 9.7764942777257e-06, 'samples': 23405568, 'steps': 45713, 'loss/train': 1.4729042053222656}
+02/26/2022 12:47:18 - INFO - codeparrot_training - Step 45714: {'lr': 9.771963758717251e-06, 'samples': 23406080, 'steps': 45714, 'loss/train': 1.8881902694702148}
+02/26/2022 12:47:21 - INFO - codeparrot_training - Step 45715: {'lr': 9.767434268769414e-06, 'samples': 23406592, 'steps': 45715, 'loss/train': 1.3229073286056519}
+02/26/2022 12:47:27 - INFO - codeparrot_training - Step 45716: {'lr': 9.762905807901651e-06, 'samples': 23407104, 'steps': 45716, 'loss/train': 1.5853618383407593}
+02/26/2022 12:47:30 - INFO - codeparrot_training - Step 45717: {'lr': 9.758378376133275e-06, 'samples': 23407616, 'steps': 45717, 'loss/train': 2.241222381591797}
+02/26/2022 12:47:36 - INFO - codeparrot_training - Step 45718: {'lr': 9.753851973483746e-06, 'samples': 23408128, 'steps': 45718, 'loss/train': 0.7458863258361816}
+02/26/2022 12:47:39 - INFO - codeparrot_training - Step 45719: {'lr': 9.74932659997238e-06, 'samples': 23408640, 'steps': 45719, 'loss/train': 1.5619745254516602}
+02/26/2022 12:47:45 - INFO - codeparrot_training - Step 45720: {'lr': 9.744802255618662e-06, 'samples': 23409152, 'steps': 45720, 'loss/train': 1.2937605381011963}
+02/26/2022 12:47:49 - INFO - codeparrot_training - Step 45721: {'lr': 9.74027894044191e-06, 'samples': 23409664, 'steps': 45721, 'loss/train': 1.4370262622833252}
+02/26/2022 12:47:54 - INFO - codeparrot_training - Step 45722: {'lr': 9.735756654461498e-06, 'samples': 23410176, 'steps': 45722, 'loss/train': 2.0643310546875}
+02/26/2022 12:47:57 - INFO - codeparrot_training - Step 45723: {'lr': 9.731235397696797e-06, 'samples': 23410688, 'steps': 45723, 'loss/train': 1.572877287864685}
+02/26/2022 12:48:03 - INFO - codeparrot_training - Step 45724: {'lr': 9.72671517016724e-06, 'samples': 23411200, 'steps': 45724, 'loss/train': 2.330118417739868}
+02/26/2022 12:48:07 - INFO - codeparrot_training - Step 45725: {'lr': 9.722195971892112e-06, 'samples': 23411712, 'steps': 45725, 'loss/train': 0.7508022785186768}
+02/26/2022 12:48:12 - INFO - codeparrot_training - Step 45726: {'lr': 9.71767780289079e-06, 'samples': 23412224, 'steps': 45726, 'loss/train': 1.4067645072937012}
+02/26/2022 12:48:15 - INFO - codeparrot_training - Step 45727: {'lr': 9.713160663182646e-06, 'samples': 23412736, 'steps': 45727, 'loss/train': 1.3673279285430908}
+02/26/2022 12:48:21 - INFO - codeparrot_training - Step 45728: {'lr': 9.708644552787028e-06, 'samples': 23413248, 'steps': 45728, 'loss/train': 1.1845699548721313}
+02/26/2022 12:48:25 - INFO - codeparrot_training - Step 45729: {'lr': 9.704129471723277e-06, 'samples': 23413760, 'steps': 45729, 'loss/train': 1.5683904886245728}
+02/26/2022 12:48:31 - INFO - codeparrot_training - Step 45730: {'lr': 9.699615420010744e-06, 'samples': 23414272, 'steps': 45730, 'loss/train': 1.3461216688156128}
+02/26/2022 12:48:34 - INFO - codeparrot_training - Step 45731: {'lr': 9.695102397668715e-06, 'samples': 23414784, 'steps': 45731, 'loss/train': 2.922683000564575}
+02/26/2022 12:48:40 - INFO - codeparrot_training - Step 45732: {'lr': 9.690590404716592e-06, 'samples': 23415296, 'steps': 45732, 'loss/train': 1.7593162059783936}
+02/26/2022 12:48:43 - INFO - codeparrot_training - Step 45733: {'lr': 9.686079441173668e-06, 'samples': 23415808, 'steps': 45733, 'loss/train': 1.700556755065918}
+02/26/2022 12:48:49 - INFO - codeparrot_training - Step 45734: {'lr': 9.681569507059285e-06, 'samples': 23416320, 'steps': 45734, 'loss/train': 1.8904430866241455}
+02/26/2022 12:48:52 - INFO - codeparrot_training - Step 45735: {'lr': 9.677060602392707e-06, 'samples': 23416832, 'steps': 45735, 'loss/train': 2.1521337032318115}
+02/26/2022 12:48:58 - INFO - codeparrot_training - Step 45736: {'lr': 9.672552727193306e-06, 'samples': 23417344, 'steps': 45736, 'loss/train': 0.817400336265564}
+02/26/2022 12:49:01 - INFO - codeparrot_training - Step 45737: {'lr': 9.668045881480403e-06, 'samples': 23417856, 'steps': 45737, 'loss/train': 1.1943998336791992}
+02/26/2022 12:49:07 - INFO - codeparrot_training - Step 45738: {'lr': 9.663540065273258e-06, 'samples': 23418368, 'steps': 45738, 'loss/train': 0.8992491364479065}
+02/26/2022 12:49:10 - INFO - codeparrot_training - Step 45739: {'lr': 9.659035278591189e-06, 'samples': 23418880, 'steps': 45739, 'loss/train': 2.6421351432800293}
+02/26/2022 12:49:16 - INFO - codeparrot_training - Step 45740: {'lr': 9.654531521453513e-06, 'samples': 23419392, 'steps': 45740, 'loss/train': 1.8610260486602783}
+02/26/2022 12:49:20 - INFO - codeparrot_training - Step 45741: {'lr': 9.650028793879494e-06, 'samples': 23419904, 'steps': 45741, 'loss/train': 0.9102139472961426}
+02/26/2022 12:49:25 - INFO - codeparrot_training - Step 45742: {'lr': 9.645527095888424e-06, 'samples': 23420416, 'steps': 45742, 'loss/train': 2.4732449054718018}
+02/26/2022 12:49:29 - INFO - codeparrot_training - Step 45743: {'lr': 9.64102642749959e-06, 'samples': 23420928, 'steps': 45743, 'loss/train': 1.081050992012024}
+02/26/2022 12:49:34 - INFO - codeparrot_training - Step 45744: {'lr': 9.636526788732308e-06, 'samples': 23421440, 'steps': 45744, 'loss/train': 3.349553108215332}
+02/26/2022 12:49:38 - INFO - codeparrot_training - Step 45745: {'lr': 9.632028179605817e-06, 'samples': 23421952, 'steps': 45745, 'loss/train': 2.085125684738159}
+02/26/2022 12:49:44 - INFO - codeparrot_training - Step 45746: {'lr': 9.627530600139351e-06, 'samples': 23422464, 'steps': 45746, 'loss/train': 2.2292213439941406}
+02/26/2022 12:49:47 - INFO - codeparrot_training - Step 45747: {'lr': 9.62303405035228e-06, 'samples': 23422976, 'steps': 45747, 'loss/train': 1.5764613151550293}
+02/26/2022 12:49:53 - INFO - codeparrot_training - Step 45748: {'lr': 9.618538530263731e-06, 'samples': 23423488, 'steps': 45748, 'loss/train': 1.3521808385849}
+02/26/2022 12:49:56 - INFO - codeparrot_training - Step 45749: {'lr': 9.614044039893077e-06, 'samples': 23424000, 'steps': 45749, 'loss/train': 2.0272583961486816}
+02/26/2022 12:50:02 - INFO - codeparrot_training - Step 45750: {'lr': 9.609550579259497e-06, 'samples': 23424512, 'steps': 45750, 'loss/train': 2.5384294986724854}
+02/26/2022 12:50:05 - INFO - codeparrot_training - Step 45751: {'lr': 9.605058148382335e-06, 'samples': 23425024, 'steps': 45751, 'loss/train': 1.9197872877120972}
+02/26/2022 12:50:10 - INFO - codeparrot_training - Step 45752: {'lr': 9.600566747280714e-06, 'samples': 23425536, 'steps': 45752, 'loss/train': 0.6177244186401367}
+02/26/2022 12:50:14 - INFO - codeparrot_training - Step 45753: {'lr': 9.596076375973955e-06, 'samples': 23426048, 'steps': 45753, 'loss/train': 2.314452886581421}
+02/26/2022 12:50:19 - INFO - codeparrot_training - Step 45754: {'lr': 9.591587034481237e-06, 'samples': 23426560, 'steps': 45754, 'loss/train': 1.8757658004760742}
+02/26/2022 12:50:23 - INFO - codeparrot_training - Step 45755: {'lr': 9.587098722821847e-06, 'samples': 23427072, 'steps': 45755, 'loss/train': 1.6897307634353638}
+02/26/2022 12:50:29 - INFO - codeparrot_training - Step 45756: {'lr': 9.582611441014965e-06, 'samples': 23427584, 'steps': 45756, 'loss/train': 1.489898920059204}
+02/26/2022 12:50:33 - INFO - codeparrot_training - Step 45757: {'lr': 9.578125189079855e-06, 'samples': 23428096, 'steps': 45757, 'loss/train': 2.992708683013916}
+02/26/2022 12:50:38 - INFO - codeparrot_training - Step 45758: {'lr': 9.573639967035697e-06, 'samples': 23428608, 'steps': 45758, 'loss/train': 1.3026658296585083}
+02/26/2022 12:50:41 - INFO - codeparrot_training - Step 45759: {'lr': 9.56915577490175e-06, 'samples': 23429120, 'steps': 45759, 'loss/train': 2.2049973011016846}
+02/26/2022 12:50:47 - INFO - codeparrot_training - Step 45760: {'lr': 9.564672612697168e-06, 'samples': 23429632, 'steps': 45760, 'loss/train': 1.5544084310531616}
+02/26/2022 12:50:51 - INFO - codeparrot_training - Step 45761: {'lr': 9.56019048044121e-06, 'samples': 23430144, 'steps': 45761, 'loss/train': 1.6291383504867554}
+02/26/2022 12:50:56 - INFO - codeparrot_training - Step 45762: {'lr': 9.555709378153004e-06, 'samples': 23430656, 'steps': 45762, 'loss/train': 1.9854035377502441}
+02/26/2022 12:51:00 - INFO - codeparrot_training - Step 45763: {'lr': 9.55122930585181e-06, 'samples': 23431168, 'steps': 45763, 'loss/train': 1.219187617301941}
+02/26/2022 12:51:05 - INFO - codeparrot_training - Step 45764: {'lr': 9.546750263556808e-06, 'samples': 23431680, 'steps': 45764, 'loss/train': 1.121265172958374}
+02/26/2022 12:51:09 - INFO - codeparrot_training - Step 45765: {'lr': 9.542272251287176e-06, 'samples': 23432192, 'steps': 45765, 'loss/train': 1.8777953386306763}
+02/26/2022 12:51:15 - INFO - codeparrot_training - Step 45766: {'lr': 9.537795269062066e-06, 'samples': 23432704, 'steps': 45766, 'loss/train': 0.7203315496444702}
+02/26/2022 12:51:18 - INFO - codeparrot_training - Step 45767: {'lr': 9.53331931690074e-06, 'samples': 23433216, 'steps': 45767, 'loss/train': 1.976548433303833}
+02/26/2022 12:51:24 - INFO - codeparrot_training - Step 45768: {'lr': 9.52884439482224e-06, 'samples': 23433728, 'steps': 45768, 'loss/train': 2.1859312057495117}
+02/26/2022 12:51:27 - INFO - codeparrot_training - Step 45769: {'lr': 9.524370502845908e-06, 'samples': 23434240, 'steps': 45769, 'loss/train': 1.7748059034347534}
+02/26/2022 12:51:33 - INFO - codeparrot_training - Step 45770: {'lr': 9.519897640990733e-06, 'samples': 23434752, 'steps': 45770, 'loss/train': 1.8329581022262573}
+02/26/2022 12:51:36 - INFO - codeparrot_training - Step 45771: {'lr': 9.515425809276002e-06, 'samples': 23435264, 'steps': 45771, 'loss/train': 0.9880490899085999}
+02/26/2022 12:51:42 - INFO - codeparrot_training - Step 45772: {'lr': 9.510955007720784e-06, 'samples': 23435776, 'steps': 45772, 'loss/train': 1.8825429677963257}
+02/26/2022 12:51:45 - INFO - codeparrot_training - Step 45773: {'lr': 9.506485236344342e-06, 'samples': 23436288, 'steps': 45773, 'loss/train': 1.7602052688598633}
+02/26/2022 12:51:51 - INFO - codeparrot_training - Step 45774: {'lr': 9.502016495165688e-06, 'samples': 23436800, 'steps': 45774, 'loss/train': 2.645533800125122}
+02/26/2022 12:51:54 - INFO - codeparrot_training - Step 45775: {'lr': 9.497548784204058e-06, 'samples': 23437312, 'steps': 45775, 'loss/train': 2.595423698425293}
+02/26/2022 12:52:01 - INFO - codeparrot_training - Step 45776: {'lr': 9.493082103478518e-06, 'samples': 23437824, 'steps': 45776, 'loss/train': 0.023891957476735115}
+02/26/2022 12:52:04 - INFO - codeparrot_training - Step 45777: {'lr': 9.488616453008304e-06, 'samples': 23438336, 'steps': 45777, 'loss/train': 1.465423583984375}
+02/26/2022 12:52:10 - INFO - codeparrot_training - Step 45778: {'lr': 9.484151832812454e-06, 'samples': 23438848, 'steps': 45778, 'loss/train': 1.2099231481552124}
+02/26/2022 12:52:13 - INFO - codeparrot_training - Step 45779: {'lr': 9.479688242910122e-06, 'samples': 23439360, 'steps': 45779, 'loss/train': 2.161647081375122}
+02/26/2022 12:52:19 - INFO - codeparrot_training - Step 45780: {'lr': 9.47522568332046e-06, 'samples': 23439872, 'steps': 45780, 'loss/train': 2.580091953277588}
+02/26/2022 12:52:22 - INFO - codeparrot_training - Step 45781: {'lr': 9.470764154062505e-06, 'samples': 23440384, 'steps': 45781, 'loss/train': 2.033205270767212}
+02/26/2022 12:52:28 - INFO - codeparrot_training - Step 45782: {'lr': 9.466303655155439e-06, 'samples': 23440896, 'steps': 45782, 'loss/train': 1.76809561252594}
+02/26/2022 12:52:31 - INFO - codeparrot_training - Step 45783: {'lr': 9.461844186618385e-06, 'samples': 23441408, 'steps': 45783, 'loss/train': 1.7237430810928345}
+02/26/2022 12:52:37 - INFO - codeparrot_training - Step 45784: {'lr': 9.457385748470382e-06, 'samples': 23441920, 'steps': 45784, 'loss/train': 1.134592056274414}
+02/26/2022 12:52:41 - INFO - codeparrot_training - Step 45785: {'lr': 9.452928340730554e-06, 'samples': 23442432, 'steps': 45785, 'loss/train': 1.6695693731307983}
+02/26/2022 12:52:47 - INFO - codeparrot_training - Step 45786: {'lr': 9.448471963417998e-06, 'samples': 23442944, 'steps': 45786, 'loss/train': 1.2193089723587036}
+02/26/2022 12:52:50 - INFO - codeparrot_training - Step 45787: {'lr': 9.444016616551837e-06, 'samples': 23443456, 'steps': 45787, 'loss/train': 0.04601413011550903}
+02/26/2022 12:52:56 - INFO - codeparrot_training - Step 45788: {'lr': 9.439562300151112e-06, 'samples': 23443968, 'steps': 45788, 'loss/train': 1.4977885484695435}
+02/26/2022 12:52:59 - INFO - codeparrot_training - Step 45789: {'lr': 9.43510901423486e-06, 'samples': 23444480, 'steps': 45789, 'loss/train': 2.6571083068847656}
+02/26/2022 12:53:05 - INFO - codeparrot_training - Step 45790: {'lr': 9.430656758822264e-06, 'samples': 23444992, 'steps': 45790, 'loss/train': 2.8447751998901367}
+02/26/2022 12:53:08 - INFO - codeparrot_training - Step 45791: {'lr': 9.426205533932363e-06, 'samples': 23445504, 'steps': 45791, 'loss/train': 0.9111694693565369}
+02/26/2022 12:53:14 - INFO - codeparrot_training - Step 45792: {'lr': 9.42175533958417e-06, 'samples': 23446016, 'steps': 45792, 'loss/train': 0.949516773223877}
+02/26/2022 12:53:17 - INFO - codeparrot_training - Step 45793: {'lr': 9.41730617579678e-06, 'samples': 23446528, 'steps': 45793, 'loss/train': 0.7893884778022766}
+02/26/2022 12:53:22 - INFO - codeparrot_training - Step 45794: {'lr': 9.41285804258929e-06, 'samples': 23447040, 'steps': 45794, 'loss/train': 1.7310644388198853}
+02/26/2022 12:53:26 - INFO - codeparrot_training - Step 45795: {'lr': 9.40841093998071e-06, 'samples': 23447552, 'steps': 45795, 'loss/train': 1.2345225811004639}
+02/26/2022 12:53:32 - INFO - codeparrot_training - Step 45796: {'lr': 9.403964867990111e-06, 'samples': 23448064, 'steps': 45796, 'loss/train': 0.19714412093162537}
+02/26/2022 12:53:35 - INFO - codeparrot_training - Step 45797: {'lr': 9.399519826636478e-06, 'samples': 23448576, 'steps': 45797, 'loss/train': 2.241469621658325}
+02/26/2022 12:53:40 - INFO - codeparrot_training - Step 45798: {'lr': 9.395075815938958e-06, 'samples': 23449088, 'steps': 45798, 'loss/train': 0.5774608850479126}
+02/26/2022 12:53:44 - INFO - codeparrot_training - Step 45799: {'lr': 9.390632835916513e-06, 'samples': 23449600, 'steps': 45799, 'loss/train': 1.9616197347640991}
+02/26/2022 12:53:50 - INFO - codeparrot_training - Step 45800: {'lr': 9.386190886588208e-06, 'samples': 23450112, 'steps': 45800, 'loss/train': 3.375455379486084}
+02/26/2022 12:53:53 - INFO - codeparrot_training - Step 45801: {'lr': 9.381749967973002e-06, 'samples': 23450624, 'steps': 45801, 'loss/train': 2.8787474632263184}
+02/26/2022 12:54:00 - INFO - codeparrot_training - Step 45802: {'lr': 9.377310080090047e-06, 'samples': 23451136, 'steps': 45802, 'loss/train': 2.1386520862579346}
+02/26/2022 12:54:03 - INFO - codeparrot_training - Step 45803: {'lr': 9.372871222958213e-06, 'samples': 23451648, 'steps': 45803, 'loss/train': 1.2434087991714478}
+02/26/2022 12:54:09 - INFO - codeparrot_training - Step 45804: {'lr': 9.368433396596681e-06, 'samples': 23452160, 'steps': 45804, 'loss/train': 1.6879916191101074}
+02/26/2022 12:54:12 - INFO - codeparrot_training - Step 45805: {'lr': 9.363996601024299e-06, 'samples': 23452672, 'steps': 45805, 'loss/train': 0.04310918599367142}
+02/26/2022 12:54:18 - INFO - codeparrot_training - Step 45806: {'lr': 9.359560836260189e-06, 'samples': 23453184, 'steps': 45806, 'loss/train': 1.6436880826950073}
+02/26/2022 12:54:21 - INFO - codeparrot_training - Step 45807: {'lr': 9.35512610232328e-06, 'samples': 23453696, 'steps': 45807, 'loss/train': 2.1185781955718994}
+02/26/2022 12:54:27 - INFO - codeparrot_training - Step 45808: {'lr': 9.35069239923264e-06, 'samples': 23454208, 'steps': 45808, 'loss/train': 1.0220075845718384}
+02/26/2022 12:54:30 - INFO - codeparrot_training - Step 45809: {'lr': 9.346259727007173e-06, 'samples': 23454720, 'steps': 45809, 'loss/train': 1.7280954122543335}
+02/26/2022 12:54:36 - INFO - codeparrot_training - Step 45810: {'lr': 9.341828085665915e-06, 'samples': 23455232, 'steps': 45810, 'loss/train': 1.114738941192627}
+02/26/2022 12:54:39 - INFO - codeparrot_training - Step 45811: {'lr': 9.337397475227855e-06, 'samples': 23455744, 'steps': 45811, 'loss/train': 1.9941314458847046}
+02/26/2022 12:54:45 - INFO - codeparrot_training - Step 45812: {'lr': 9.332967895712002e-06, 'samples': 23456256, 'steps': 45812, 'loss/train': 2.0828349590301514}
+02/26/2022 12:54:49 - INFO - codeparrot_training - Step 45813: {'lr': 9.328539347137233e-06, 'samples': 23456768, 'steps': 45813, 'loss/train': 0.07467823475599289}
+02/26/2022 12:54:54 - INFO - codeparrot_training - Step 45814: {'lr': 9.324111829522641e-06, 'samples': 23457280, 'steps': 45814, 'loss/train': 2.331038236618042}
+02/26/2022 12:54:58 - INFO - codeparrot_training - Step 45815: {'lr': 9.319685342887074e-06, 'samples': 23457792, 'steps': 45815, 'loss/train': 2.171290636062622}
+02/26/2022 12:55:03 - INFO - codeparrot_training - Step 45816: {'lr': 9.315259887249572e-06, 'samples': 23458304, 'steps': 45816, 'loss/train': 1.9826956987380981}
+02/26/2022 12:55:07 - INFO - codeparrot_training - Step 45817: {'lr': 9.310835462629092e-06, 'samples': 23458816, 'steps': 45817, 'loss/train': 2.1105475425720215}
+02/26/2022 12:55:12 - INFO - codeparrot_training - Step 45818: {'lr': 9.306412069044562e-06, 'samples': 23459328, 'steps': 45818, 'loss/train': 2.259345054626465}
+02/26/2022 12:55:16 - INFO - codeparrot_training - Step 45819: {'lr': 9.301989706514886e-06, 'samples': 23459840, 'steps': 45819, 'loss/train': 2.4962852001190186}
+02/26/2022 12:55:21 - INFO - codeparrot_training - Step 45820: {'lr': 9.2975683750591e-06, 'samples': 23460352, 'steps': 45820, 'loss/train': 2.236016273498535}
+02/26/2022 12:55:25 - INFO - codeparrot_training - Step 45821: {'lr': 9.293148074696084e-06, 'samples': 23460864, 'steps': 45821, 'loss/train': 2.167633295059204}
+02/26/2022 12:55:31 - INFO - codeparrot_training - Step 45822: {'lr': 9.288728805444818e-06, 'samples': 23461376, 'steps': 45822, 'loss/train': 2.3480963706970215}
+02/26/2022 12:55:34 - INFO - codeparrot_training - Step 45823: {'lr': 9.28431056732415e-06, 'samples': 23461888, 'steps': 45823, 'loss/train': 1.847638487815857}
+02/26/2022 12:55:40 - INFO - codeparrot_training - Step 45824: {'lr': 9.279893360353093e-06, 'samples': 23462400, 'steps': 45824, 'loss/train': 3.16902494430542}
+02/26/2022 12:55:43 - INFO - codeparrot_training - Step 45825: {'lr': 9.275477184550546e-06, 'samples': 23462912, 'steps': 45825, 'loss/train': 1.3458139896392822}
+02/26/2022 12:55:49 - INFO - codeparrot_training - Step 45826: {'lr': 9.271062039935412e-06, 'samples': 23463424, 'steps': 45826, 'loss/train': 0.8282353281974792}
+02/26/2022 12:55:52 - INFO - codeparrot_training - Step 45827: {'lr': 9.266647926526595e-06, 'samples': 23463936, 'steps': 45827, 'loss/train': 1.0621026754379272}
+02/26/2022 12:55:58 - INFO - codeparrot_training - Step 45828: {'lr': 9.26223484434302e-06, 'samples': 23464448, 'steps': 45828, 'loss/train': 2.1471261978149414}
+02/26/2022 12:56:02 - INFO - codeparrot_training - Step 45829: {'lr': 9.257822793403591e-06, 'samples': 23464960, 'steps': 45829, 'loss/train': 2.161721706390381}
+02/26/2022 12:56:05 - INFO - codeparrot_training - Step 45830: {'lr': 9.253411773727211e-06, 'samples': 23465472, 'steps': 45830, 'loss/train': 2.138962745666504}
+02/26/2022 12:56:11 - INFO - codeparrot_training - Step 45831: {'lr': 9.24900178533275e-06, 'samples': 23465984, 'steps': 45831, 'loss/train': 0.8545023798942566}
+02/26/2022 12:56:14 - INFO - codeparrot_training - Step 45832: {'lr': 9.244592828239112e-06, 'samples': 23466496, 'steps': 45832, 'loss/train': 1.726701021194458}
+02/26/2022 12:56:21 - INFO - codeparrot_training - Step 45833: {'lr': 9.240184902465226e-06, 'samples': 23467008, 'steps': 45833, 'loss/train': 1.7529367208480835}
+02/26/2022 12:56:24 - INFO - codeparrot_training - Step 45834: {'lr': 9.23577800802991e-06, 'samples': 23467520, 'steps': 45834, 'loss/train': 0.9927670955657959}
+02/26/2022 12:56:30 - INFO - codeparrot_training - Step 45835: {'lr': 9.231372144952094e-06, 'samples': 23468032, 'steps': 45835, 'loss/train': 1.4131137132644653}
+02/26/2022 12:56:33 - INFO - codeparrot_training - Step 45836: {'lr': 9.226967313250595e-06, 'samples': 23468544, 'steps': 45836, 'loss/train': 0.463163822889328}
+02/26/2022 12:56:39 - INFO - codeparrot_training - Step 45837: {'lr': 9.222563512944343e-06, 'samples': 23469056, 'steps': 45837, 'loss/train': 1.06686532497406}
+02/26/2022 12:56:42 - INFO - codeparrot_training - Step 45838: {'lr': 9.218160744052157e-06, 'samples': 23469568, 'steps': 45838, 'loss/train': 1.561614751815796}
+02/26/2022 12:56:48 - INFO - codeparrot_training - Step 45839: {'lr': 9.213759006592936e-06, 'samples': 23470080, 'steps': 45839, 'loss/train': 2.109992265701294}
+02/26/2022 12:56:51 - INFO - codeparrot_training - Step 45840: {'lr': 9.209358300585474e-06, 'samples': 23470592, 'steps': 45840, 'loss/train': 1.4622077941894531}
+02/26/2022 12:56:57 - INFO - codeparrot_training - Step 45841: {'lr': 9.204958626048671e-06, 'samples': 23471104, 'steps': 45841, 'loss/train': 1.6549198627471924}
+02/26/2022 12:57:00 - INFO - codeparrot_training - Step 45842: {'lr': 9.200559983001372e-06, 'samples': 23471616, 'steps': 45842, 'loss/train': 1.8383203744888306}
+02/26/2022 12:57:07 - INFO - codeparrot_training - Step 45843: {'lr': 9.196162371462451e-06, 'samples': 23472128, 'steps': 45843, 'loss/train': 1.9243422746658325}
+02/26/2022 12:57:10 - INFO - codeparrot_training - Step 45844: {'lr': 9.191765791450645e-06, 'samples': 23472640, 'steps': 45844, 'loss/train': 1.528111457824707}
+02/26/2022 12:57:16 - INFO - codeparrot_training - Step 45845: {'lr': 9.187370242984883e-06, 'samples': 23473152, 'steps': 45845, 'loss/train': 0.7781782746315002}
+02/26/2022 12:57:19 - INFO - codeparrot_training - Step 45846: {'lr': 9.182975726083925e-06, 'samples': 23473664, 'steps': 45846, 'loss/train': 1.097321629524231}
+02/26/2022 12:57:25 - INFO - codeparrot_training - Step 45847: {'lr': 9.178582240766703e-06, 'samples': 23474176, 'steps': 45847, 'loss/train': 1.3071389198303223}
+02/26/2022 12:57:28 - INFO - codeparrot_training - Step 45848: {'lr': 9.174189787051896e-06, 'samples': 23474688, 'steps': 45848, 'loss/train': 1.2330803871154785}
+02/26/2022 12:57:34 - INFO - codeparrot_training - Step 45849: {'lr': 9.169798364958404e-06, 'samples': 23475200, 'steps': 45849, 'loss/train': 1.8518048524856567}
+02/26/2022 12:57:37 - INFO - codeparrot_training - Step 45850: {'lr': 9.165407974505019e-06, 'samples': 23475712, 'steps': 45850, 'loss/train': 1.329521894454956}
+02/26/2022 12:57:43 - INFO - codeparrot_training - Step 45851: {'lr': 9.161018615710587e-06, 'samples': 23476224, 'steps': 45851, 'loss/train': 1.731979489326477}
+02/26/2022 12:57:46 - INFO - codeparrot_training - Step 45852: {'lr': 9.156630288593842e-06, 'samples': 23476736, 'steps': 45852, 'loss/train': 2.024141550064087}
+02/26/2022 12:57:52 - INFO - codeparrot_training - Step 45853: {'lr': 9.152242993173632e-06, 'samples': 23477248, 'steps': 45853, 'loss/train': 2.0704832077026367}
+02/26/2022 12:57:56 - INFO - codeparrot_training - Step 45854: {'lr': 9.147856729468717e-06, 'samples': 23477760, 'steps': 45854, 'loss/train': 1.230162501335144}
+02/26/2022 12:58:01 - INFO - codeparrot_training - Step 45855: {'lr': 9.143471497497946e-06, 'samples': 23478272, 'steps': 45855, 'loss/train': 1.0638129711151123}
+02/26/2022 12:58:05 - INFO - codeparrot_training - Step 45856: {'lr': 9.139087297280024e-06, 'samples': 23478784, 'steps': 45856, 'loss/train': 1.7369027137756348}
+02/26/2022 12:58:10 - INFO - codeparrot_training - Step 45857: {'lr': 9.134704128833798e-06, 'samples': 23479296, 'steps': 45857, 'loss/train': 2.1756985187530518}
+02/26/2022 12:58:14 - INFO - codeparrot_training - Step 45858: {'lr': 9.130321992177976e-06, 'samples': 23479808, 'steps': 45858, 'loss/train': 1.2524319887161255}
+02/26/2022 12:58:20 - INFO - codeparrot_training - Step 45859: {'lr': 9.125940887331402e-06, 'samples': 23480320, 'steps': 45859, 'loss/train': 1.6620538234710693}
+02/26/2022 12:58:23 - INFO - codeparrot_training - Step 45860: {'lr': 9.121560814312813e-06, 'samples': 23480832, 'steps': 45860, 'loss/train': 1.9836480617523193}
+02/26/2022 12:58:29 - INFO - codeparrot_training - Step 45861: {'lr': 9.11718177314097e-06, 'samples': 23481344, 'steps': 45861, 'loss/train': 1.1807059049606323}
+02/26/2022 12:58:33 - INFO - codeparrot_training - Step 45862: {'lr': 9.11280376383461e-06, 'samples': 23481856, 'steps': 45862, 'loss/train': 1.1456806659698486}
+02/26/2022 12:58:38 - INFO - codeparrot_training - Step 45863: {'lr': 9.10842678641252e-06, 'samples': 23482368, 'steps': 45863, 'loss/train': 1.6775380373001099}
+02/26/2022 12:58:41 - INFO - codeparrot_training - Step 45864: {'lr': 9.10405084089344e-06, 'samples': 23482880, 'steps': 45864, 'loss/train': 1.735140323638916}
+02/26/2022 12:58:47 - INFO - codeparrot_training - Step 45865: {'lr': 9.099675927296131e-06, 'samples': 23483392, 'steps': 45865, 'loss/train': 1.1546891927719116}
+02/26/2022 12:58:51 - INFO - codeparrot_training - Step 45866: {'lr': 9.09530204563927e-06, 'samples': 23483904, 'steps': 45866, 'loss/train': 0.9374484419822693}
+02/26/2022 12:58:56 - INFO - codeparrot_training - Step 45867: {'lr': 9.090929195941678e-06, 'samples': 23484416, 'steps': 45867, 'loss/train': 1.7938917875289917}
+02/26/2022 12:59:00 - INFO - codeparrot_training - Step 45868: {'lr': 9.08655737822206e-06, 'samples': 23484928, 'steps': 45868, 'loss/train': 1.1096880435943604}
+02/26/2022 12:59:05 - INFO - codeparrot_training - Step 45869: {'lr': 9.082186592499097e-06, 'samples': 23485440, 'steps': 45869, 'loss/train': 2.4260993003845215}
+02/26/2022 12:59:08 - INFO - codeparrot_training - Step 45870: {'lr': 9.077816838791552e-06, 'samples': 23485952, 'steps': 45870, 'loss/train': 0.9869159460067749}
+02/26/2022 12:59:15 - INFO - codeparrot_training - Step 45871: {'lr': 9.073448117118132e-06, 'samples': 23486464, 'steps': 45871, 'loss/train': 2.30340576171875}
+02/26/2022 12:59:18 - INFO - codeparrot_training - Step 45872: {'lr': 9.069080427497572e-06, 'samples': 23486976, 'steps': 45872, 'loss/train': 3.0496270656585693}
+02/26/2022 12:59:24 - INFO - codeparrot_training - Step 45873: {'lr': 9.064713769948579e-06, 'samples': 23487488, 'steps': 45873, 'loss/train': 1.0741225481033325}
+02/26/2022 12:59:27 - INFO - codeparrot_training - Step 45874: {'lr': 9.060348144489833e-06, 'samples': 23488000, 'steps': 45874, 'loss/train': 1.533676028251648}
+02/26/2022 12:59:33 - INFO - codeparrot_training - Step 45875: {'lr': 9.05598355114004e-06, 'samples': 23488512, 'steps': 45875, 'loss/train': 2.1415016651153564}
+02/26/2022 12:59:36 - INFO - codeparrot_training - Step 45876: {'lr': 9.051619989917909e-06, 'samples': 23489024, 'steps': 45876, 'loss/train': 0.37592703104019165}
+02/26/2022 12:59:42 - INFO - codeparrot_training - Step 45877: {'lr': 9.047257460842118e-06, 'samples': 23489536, 'steps': 45877, 'loss/train': 0.7004016637802124}
+02/26/2022 12:59:45 - INFO - codeparrot_training - Step 45878: {'lr': 9.04289596393143e-06, 'samples': 23490048, 'steps': 45878, 'loss/train': 1.8229241371154785}
+02/26/2022 12:59:51 - INFO - codeparrot_training - Step 45879: {'lr': 9.038535499204387e-06, 'samples': 23490560, 'steps': 45879, 'loss/train': 1.2572519779205322}
+02/26/2022 12:59:54 - INFO - codeparrot_training - Step 45880: {'lr': 9.034176066679779e-06, 'samples': 23491072, 'steps': 45880, 'loss/train': 2.00264835357666}
+02/26/2022 13:00:00 - INFO - codeparrot_training - Step 45881: {'lr': 9.0298176663762e-06, 'samples': 23491584, 'steps': 45881, 'loss/train': 1.9450125694274902}
+02/26/2022 13:00:04 - INFO - codeparrot_training - Step 45882: {'lr': 9.025460298312443e-06, 'samples': 23492096, 'steps': 45882, 'loss/train': 1.8951421976089478}
+02/26/2022 13:00:09 - INFO - codeparrot_training - Step 45883: {'lr': 9.021103962507022e-06, 'samples': 23492608, 'steps': 45883, 'loss/train': 0.12006185203790665}
+02/26/2022 13:00:13 - INFO - codeparrot_training - Step 45884: {'lr': 9.016748658978723e-06, 'samples': 23493120, 'steps': 45884, 'loss/train': 1.4472486972808838}
+02/26/2022 13:00:18 - INFO - codeparrot_training - Step 45885: {'lr': 9.01239438774612e-06, 'samples': 23493632, 'steps': 45885, 'loss/train': 0.9986704587936401}
+02/26/2022 13:00:22 - INFO - codeparrot_training - Step 45886: {'lr': 9.008041148827944e-06, 'samples': 23494144, 'steps': 45886, 'loss/train': 1.5340802669525146}
+02/26/2022 13:00:27 - INFO - codeparrot_training - Step 45887: {'lr': 9.003688942242737e-06, 'samples': 23494656, 'steps': 45887, 'loss/train': 1.8583381175994873}
+02/26/2022 13:00:31 - INFO - codeparrot_training - Step 45888: {'lr': 8.999337768009235e-06, 'samples': 23495168, 'steps': 45888, 'loss/train': 1.8377958536148071}
+02/26/2022 13:00:36 - INFO - codeparrot_training - Step 45889: {'lr': 8.994987626146007e-06, 'samples': 23495680, 'steps': 45889, 'loss/train': 1.835382103919983}
+02/26/2022 13:00:40 - INFO - codeparrot_training - Step 45890: {'lr': 8.990638516671757e-06, 'samples': 23496192, 'steps': 45890, 'loss/train': 2.2052927017211914}
+02/26/2022 13:00:46 - INFO - codeparrot_training - Step 45891: {'lr': 8.986290439605083e-06, 'samples': 23496704, 'steps': 45891, 'loss/train': 1.654951572418213}
+02/26/2022 13:00:50 - INFO - codeparrot_training - Step 45892: {'lr': 8.981943394964637e-06, 'samples': 23497216, 'steps': 45892, 'loss/train': 2.0534675121307373}
+02/26/2022 13:00:55 - INFO - codeparrot_training - Step 45893: {'lr': 8.97759738276896e-06, 'samples': 23497728, 'steps': 45893, 'loss/train': 1.6069839000701904}
+02/26/2022 13:00:59 - INFO - codeparrot_training - Step 45894: {'lr': 8.97325240303673e-06, 'samples': 23498240, 'steps': 45894, 'loss/train': 0.6512133479118347}
+02/26/2022 13:01:04 - INFO - codeparrot_training - Step 45895: {'lr': 8.968908455786601e-06, 'samples': 23498752, 'steps': 45895, 'loss/train': 2.1880362033843994}
+02/26/2022 13:01:08 - INFO - codeparrot_training - Step 45896: {'lr': 8.964565541037084e-06, 'samples': 23499264, 'steps': 45896, 'loss/train': 0.9493980407714844}
+02/26/2022 13:01:13 - INFO - codeparrot_training - Step 45897: {'lr': 8.960223658806831e-06, 'samples': 23499776, 'steps': 45897, 'loss/train': 2.1147823333740234}
+02/26/2022 13:01:17 - INFO - codeparrot_training - Step 45898: {'lr': 8.955882809114468e-06, 'samples': 23500288, 'steps': 45898, 'loss/train': 1.4750529527664185}
+02/26/2022 13:01:22 - INFO - codeparrot_training - Step 45899: {'lr': 8.951542991978562e-06, 'samples': 23500800, 'steps': 45899, 'loss/train': 1.3232852220535278}
+02/26/2022 13:01:28 - INFO - codeparrot_training - Step 45900: {'lr': 8.947204207417681e-06, 'samples': 23501312, 'steps': 45900, 'loss/train': 1.0434068441390991}
+02/26/2022 13:01:31 - INFO - codeparrot_training - Step 45901: {'lr': 8.942866455450448e-06, 'samples': 23501824, 'steps': 45901, 'loss/train': 1.7676472663879395}
+02/26/2022 13:01:37 - INFO - codeparrot_training - Step 45902: {'lr': 8.938529736095435e-06, 'samples': 23502336, 'steps': 45902, 'loss/train': 2.1735122203826904}
+02/26/2022 13:01:40 - INFO - codeparrot_training - Step 45903: {'lr': 8.93419404937118e-06, 'samples': 23502848, 'steps': 45903, 'loss/train': 1.2070268392562866}
+02/26/2022 13:01:46 - INFO - codeparrot_training - Step 45904: {'lr': 8.929859395296363e-06, 'samples': 23503360, 'steps': 45904, 'loss/train': 0.3817344605922699}
+02/26/2022 13:01:50 - INFO - codeparrot_training - Step 45905: {'lr': 8.925525773889414e-06, 'samples': 23503872, 'steps': 45905, 'loss/train': 0.6035551428794861}
+02/26/2022 13:01:55 - INFO - codeparrot_training - Step 45906: {'lr': 8.921193185168986e-06, 'samples': 23504384, 'steps': 45906, 'loss/train': 1.1104191541671753}
+02/26/2022 13:01:59 - INFO - codeparrot_training - Step 45907: {'lr': 8.916861629153616e-06, 'samples': 23504896, 'steps': 45907, 'loss/train': 2.7395594120025635}
+02/26/2022 13:02:04 - INFO - codeparrot_training - Step 45908: {'lr': 8.912531105861876e-06, 'samples': 23505408, 'steps': 45908, 'loss/train': 2.068023204803467}
+02/26/2022 13:02:08 - INFO - codeparrot_training - Step 45909: {'lr': 8.90820161531225e-06, 'samples': 23505920, 'steps': 45909, 'loss/train': 1.781388521194458}
+02/26/2022 13:02:13 - INFO - codeparrot_training - Step 45910: {'lr': 8.903873157523362e-06, 'samples': 23506432, 'steps': 45910, 'loss/train': 0.5364334583282471}
+02/26/2022 13:02:17 - INFO - codeparrot_training - Step 45911: {'lr': 8.899545732513754e-06, 'samples': 23506944, 'steps': 45911, 'loss/train': 2.1767072677612305}
+02/26/2022 13:02:22 - INFO - codeparrot_training - Step 45912: {'lr': 8.89521934030188e-06, 'samples': 23507456, 'steps': 45912, 'loss/train': 1.7564771175384521}
+02/26/2022 13:02:26 - INFO - codeparrot_training - Step 45913: {'lr': 8.89089398090634e-06, 'samples': 23507968, 'steps': 45913, 'loss/train': 1.9760472774505615}
+02/26/2022 13:02:31 - INFO - codeparrot_training - Step 45914: {'lr': 8.886569654345672e-06, 'samples': 23508480, 'steps': 45914, 'loss/train': 2.5871505737304688}
+02/26/2022 13:02:35 - INFO - codeparrot_training - Step 45915: {'lr': 8.88224636063839e-06, 'samples': 23508992, 'steps': 45915, 'loss/train': 2.2539894580841064}
+02/26/2022 13:02:41 - INFO - codeparrot_training - Step 45916: {'lr': 8.877924099802954e-06, 'samples': 23509504, 'steps': 45916, 'loss/train': 1.4650945663452148}
+02/26/2022 13:02:44 - INFO - codeparrot_training - Step 45917: {'lr': 8.873602871857955e-06, 'samples': 23510016, 'steps': 45917, 'loss/train': 2.014577865600586}
+02/26/2022 13:02:50 - INFO - codeparrot_training - Step 45918: {'lr': 8.869282676821882e-06, 'samples': 23510528, 'steps': 45918, 'loss/train': 1.9250085353851318}
+02/26/2022 13:02:53 - INFO - codeparrot_training - Step 45919: {'lr': 8.86496351471322e-06, 'samples': 23511040, 'steps': 45919, 'loss/train': 1.820523738861084}
+02/26/2022 13:02:59 - INFO - codeparrot_training - Step 45920: {'lr': 8.860645385550481e-06, 'samples': 23511552, 'steps': 45920, 'loss/train': 1.5361367464065552}
+02/26/2022 13:03:02 - INFO - codeparrot_training - Step 45921: {'lr': 8.856328289352177e-06, 'samples': 23512064, 'steps': 45921, 'loss/train': 0.7257646918296814}
+02/26/2022 13:03:08 - INFO - codeparrot_training - Step 45922: {'lr': 8.852012226136796e-06, 'samples': 23512576, 'steps': 45922, 'loss/train': 0.7444840669631958}
+02/26/2022 13:03:11 - INFO - codeparrot_training - Step 45923: {'lr': 8.847697195922822e-06, 'samples': 23513088, 'steps': 45923, 'loss/train': 2.3029396533966064}
+02/26/2022 13:03:17 - INFO - codeparrot_training - Step 45924: {'lr': 8.843383198728711e-06, 'samples': 23513600, 'steps': 45924, 'loss/train': 1.55702805519104}
+02/26/2022 13:03:20 - INFO - codeparrot_training - Step 45925: {'lr': 8.839070234572977e-06, 'samples': 23514112, 'steps': 45925, 'loss/train': 1.0189182758331299}
+02/26/2022 13:03:26 - INFO - codeparrot_training - Step 45926: {'lr': 8.834758303474133e-06, 'samples': 23514624, 'steps': 45926, 'loss/train': 2.2082135677337646}
+02/26/2022 13:03:30 - INFO - codeparrot_training - Step 45927: {'lr': 8.830447405450582e-06, 'samples': 23515136, 'steps': 45927, 'loss/train': 1.8855373859405518}
+02/26/2022 13:03:35 - INFO - codeparrot_training - Step 45928: {'lr': 8.826137540520778e-06, 'samples': 23515648, 'steps': 45928, 'loss/train': 2.0455801486968994}
+02/26/2022 13:03:39 - INFO - codeparrot_training - Step 45929: {'lr': 8.821828708703238e-06, 'samples': 23516160, 'steps': 45929, 'loss/train': 1.8347651958465576}
+02/26/2022 13:03:42 - INFO - codeparrot_training - Step 45930: {'lr': 8.817520910016447e-06, 'samples': 23516672, 'steps': 45930, 'loss/train': 1.90438711643219}
+02/26/2022 13:03:48 - INFO - codeparrot_training - Step 45931: {'lr': 8.813214144478776e-06, 'samples': 23517184, 'steps': 45931, 'loss/train': 2.3967397212982178}
+02/26/2022 13:03:53 - INFO - codeparrot_training - Step 45932: {'lr': 8.808908412108685e-06, 'samples': 23517696, 'steps': 45932, 'loss/train': 2.6964433193206787}
+02/26/2022 13:03:57 - INFO - codeparrot_training - Step 45933: {'lr': 8.804603712924686e-06, 'samples': 23518208, 'steps': 45933, 'loss/train': 1.3708142042160034}
+02/26/2022 13:04:02 - INFO - codeparrot_training - Step 45934: {'lr': 8.800300046945154e-06, 'samples': 23518720, 'steps': 45934, 'loss/train': 2.43684720993042}
+02/26/2022 13:04:06 - INFO - codeparrot_training - Step 45935: {'lr': 8.795997414188573e-06, 'samples': 23519232, 'steps': 45935, 'loss/train': 0.43741047382354736}
+02/26/2022 13:04:12 - INFO - codeparrot_training - Step 45936: {'lr': 8.791695814673289e-06, 'samples': 23519744, 'steps': 45936, 'loss/train': 2.609468460083008}
+02/26/2022 13:04:16 - INFO - codeparrot_training - Step 45937: {'lr': 8.787395248417846e-06, 'samples': 23520256, 'steps': 45937, 'loss/train': 2.0758635997772217}
+02/26/2022 13:04:21 - INFO - codeparrot_training - Step 45938: {'lr': 8.783095715440558e-06, 'samples': 23520768, 'steps': 45938, 'loss/train': 1.8781394958496094}
+02/26/2022 13:04:24 - INFO - codeparrot_training - Step 45939: {'lr': 8.778797215759971e-06, 'samples': 23521280, 'steps': 45939, 'loss/train': 2.147948980331421}
+02/26/2022 13:04:30 - INFO - codeparrot_training - Step 45940: {'lr': 8.774499749394344e-06, 'samples': 23521792, 'steps': 45940, 'loss/train': 2.4299027919769287}
+02/26/2022 13:04:33 - INFO - codeparrot_training - Step 45941: {'lr': 8.77020331636219e-06, 'samples': 23522304, 'steps': 45941, 'loss/train': 1.6396194696426392}
+02/26/2022 13:04:39 - INFO - codeparrot_training - Step 45942: {'lr': 8.765907916681887e-06, 'samples': 23522816, 'steps': 45942, 'loss/train': 2.0596086978912354}
+02/26/2022 13:04:42 - INFO - codeparrot_training - Step 45943: {'lr': 8.761613550371861e-06, 'samples': 23523328, 'steps': 45943, 'loss/train': 0.8581370711326599}
+02/26/2022 13:04:48 - INFO - codeparrot_training - Step 45944: {'lr': 8.757320217450432e-06, 'samples': 23523840, 'steps': 45944, 'loss/train': 1.881097674369812}
+02/26/2022 13:04:51 - INFO - codeparrot_training - Step 45945: {'lr': 8.753027917936084e-06, 'samples': 23524352, 'steps': 45945, 'loss/train': 1.4257121086120605}
+02/26/2022 13:04:57 - INFO - codeparrot_training - Step 45946: {'lr': 8.748736651847112e-06, 'samples': 23524864, 'steps': 45946, 'loss/train': 2.3621208667755127}
+02/26/2022 13:05:01 - INFO - codeparrot_training - Step 45947: {'lr': 8.744446419201996e-06, 'samples': 23525376, 'steps': 45947, 'loss/train': 2.346371650695801}
+02/26/2022 13:05:06 - INFO - codeparrot_training - Step 45948: {'lr': 8.74015722001903e-06, 'samples': 23525888, 'steps': 45948, 'loss/train': 1.1491217613220215}
+02/26/2022 13:05:09 - INFO - codeparrot_training - Step 45949: {'lr': 8.735869054316642e-06, 'samples': 23526400, 'steps': 45949, 'loss/train': 2.288332939147949}
+02/26/2022 13:05:15 - INFO - codeparrot_training - Step 45950: {'lr': 8.731581922113152e-06, 'samples': 23526912, 'steps': 45950, 'loss/train': 2.613100290298462}
+02/26/2022 13:05:18 - INFO - codeparrot_training - Step 45951: {'lr': 8.72729582342699e-06, 'samples': 23527424, 'steps': 45951, 'loss/train': 0.8736313581466675}
+02/26/2022 13:05:25 - INFO - codeparrot_training - Step 45952: {'lr': 8.723010758276446e-06, 'samples': 23527936, 'steps': 45952, 'loss/train': 2.044598340988159}
+02/26/2022 13:05:28 - INFO - codeparrot_training - Step 45953: {'lr': 8.718726726679948e-06, 'samples': 23528448, 'steps': 45953, 'loss/train': 0.5576093792915344}
+02/26/2022 13:05:34 - INFO - codeparrot_training - Step 45954: {'lr': 8.714443728655763e-06, 'samples': 23528960, 'steps': 45954, 'loss/train': 2.2565579414367676}
+02/26/2022 13:05:37 - INFO - codeparrot_training - Step 45955: {'lr': 8.710161764222319e-06, 'samples': 23529472, 'steps': 45955, 'loss/train': 1.1954596042633057}
+02/26/2022 13:05:43 - INFO - codeparrot_training - Step 45956: {'lr': 8.705880833397934e-06, 'samples': 23529984, 'steps': 45956, 'loss/train': 1.9215234518051147}
+02/26/2022 13:05:46 - INFO - codeparrot_training - Step 45957: {'lr': 8.701600936200927e-06, 'samples': 23530496, 'steps': 45957, 'loss/train': 1.510150671005249}
+02/26/2022 13:05:52 - INFO - codeparrot_training - Step 45958: {'lr': 8.697322072649617e-06, 'samples': 23531008, 'steps': 45958, 'loss/train': 1.6921449899673462}
+02/26/2022 13:05:55 - INFO - codeparrot_training - Step 45959: {'lr': 8.69304424276235e-06, 'samples': 23531520, 'steps': 45959, 'loss/train': 2.34102725982666}
+02/26/2022 13:06:01 - INFO - codeparrot_training - Step 45960: {'lr': 8.6887674465575e-06, 'samples': 23532032, 'steps': 45960, 'loss/train': 2.7821972370147705}
+02/26/2022 13:06:04 - INFO - codeparrot_training - Step 45961: {'lr': 8.68449168405333e-06, 'samples': 23532544, 'steps': 45961, 'loss/train': 2.6380436420440674}
+02/26/2022 13:06:10 - INFO - codeparrot_training - Step 45962: {'lr': 8.680216955268161e-06, 'samples': 23533056, 'steps': 45962, 'loss/train': 0.8853145241737366}
+02/26/2022 13:06:14 - INFO - codeparrot_training - Step 45963: {'lr': 8.675943260220309e-06, 'samples': 23533568, 'steps': 45963, 'loss/train': 1.9960031509399414}
+02/26/2022 13:06:19 - INFO - codeparrot_training - Step 45964: {'lr': 8.671670598928095e-06, 'samples': 23534080, 'steps': 45964, 'loss/train': 1.354811429977417}
+02/26/2022 13:06:23 - INFO - codeparrot_training - Step 45965: {'lr': 8.667398971409834e-06, 'samples': 23534592, 'steps': 45965, 'loss/train': 2.2799642086029053}
+02/26/2022 13:06:28 - INFO - codeparrot_training - Step 45966: {'lr': 8.663128377683794e-06, 'samples': 23535104, 'steps': 45966, 'loss/train': 1.219589114189148}
+02/26/2022 13:06:32 - INFO - codeparrot_training - Step 45967: {'lr': 8.658858817768233e-06, 'samples': 23535616, 'steps': 45967, 'loss/train': 0.9523189663887024}
+02/26/2022 13:06:37 - INFO - codeparrot_training - Step 45968: {'lr': 8.65459029168153e-06, 'samples': 23536128, 'steps': 45968, 'loss/train': 1.415152907371521}
+02/26/2022 13:06:41 - INFO - codeparrot_training - Step 45969: {'lr': 8.650322799441945e-06, 'samples': 23536640, 'steps': 45969, 'loss/train': 1.439796805381775}
+02/26/2022 13:06:46 - INFO - codeparrot_training - Step 45970: {'lr': 8.646056341067714e-06, 'samples': 23537152, 'steps': 45970, 'loss/train': 1.2613742351531982}
+02/26/2022 13:06:50 - INFO - codeparrot_training - Step 45971: {'lr': 8.64179091657713e-06, 'samples': 23537664, 'steps': 45971, 'loss/train': 2.420267105102539}
+02/26/2022 13:06:56 - INFO - codeparrot_training - Step 45972: {'lr': 8.637526525988481e-06, 'samples': 23538176, 'steps': 45972, 'loss/train': 1.1902494430541992}
+02/26/2022 13:06:59 - INFO - codeparrot_training - Step 45973: {'lr': 8.633263169320005e-06, 'samples': 23538688, 'steps': 45973, 'loss/train': 1.8855726718902588}
+02/26/2022 13:07:05 - INFO - codeparrot_training - Step 45974: {'lr': 8.629000846590045e-06, 'samples': 23539200, 'steps': 45974, 'loss/train': 1.6354668140411377}
+02/26/2022 13:07:11 - INFO - codeparrot_training - Step 45975: {'lr': 8.624739557816758e-06, 'samples': 23539712, 'steps': 45975, 'loss/train': 2.1895813941955566}
+02/26/2022 13:07:14 - INFO - codeparrot_training - Step 45976: {'lr': 8.620479303018458e-06, 'samples': 23540224, 'steps': 45976, 'loss/train': 1.7519341707229614}
+02/26/2022 13:07:18 - INFO - codeparrot_training - Step 45977: {'lr': 8.616220082213355e-06, 'samples': 23540736, 'steps': 45977, 'loss/train': 2.413083553314209}
+02/26/2022 13:07:23 - INFO - codeparrot_training - Step 45978: {'lr': 8.611961895419767e-06, 'samples': 23541248, 'steps': 45978, 'loss/train': 1.788680076599121}
+02/26/2022 13:07:29 - INFO - codeparrot_training - Step 45979: {'lr': 8.60770474265582e-06, 'samples': 23541760, 'steps': 45979, 'loss/train': 2.295016050338745}
+02/26/2022 13:07:32 - INFO - codeparrot_training - Step 45980: {'lr': 8.603448623939857e-06, 'samples': 23542272, 'steps': 45980, 'loss/train': 1.5782240629196167}
+02/26/2022 13:07:39 - INFO - codeparrot_training - Step 45981: {'lr': 8.599193539290034e-06, 'samples': 23542784, 'steps': 45981, 'loss/train': 2.5019900798797607}
+02/26/2022 13:07:42 - INFO - codeparrot_training - Step 45982: {'lr': 8.594939488724695e-06, 'samples': 23543296, 'steps': 45982, 'loss/train': 0.2256821244955063}
+02/26/2022 13:07:46 - INFO - codeparrot_training - Step 45983: {'lr': 8.590686472261882e-06, 'samples': 23543808, 'steps': 45983, 'loss/train': 2.6255862712860107}
+02/26/2022 13:07:51 - INFO - codeparrot_training - Step 45984: {'lr': 8.58643448991997e-06, 'samples': 23544320, 'steps': 45984, 'loss/train': 0.9441692233085632}
+02/26/2022 13:07:55 - INFO - codeparrot_training - Step 45985: {'lr': 8.582183541717081e-06, 'samples': 23544832, 'steps': 45985, 'loss/train': 0.22554410994052887}
+02/26/2022 13:08:00 - INFO - codeparrot_training - Step 45986: {'lr': 8.57793362767148e-06, 'samples': 23545344, 'steps': 45986, 'loss/train': 0.9938243627548218}
+02/26/2022 13:08:04 - INFO - codeparrot_training - Step 45987: {'lr': 8.573684747801347e-06, 'samples': 23545856, 'steps': 45987, 'loss/train': 0.31699928641319275}
+02/26/2022 13:08:09 - INFO - codeparrot_training - Step 45988: {'lr': 8.569436902124888e-06, 'samples': 23546368, 'steps': 45988, 'loss/train': 0.4721957743167877}
+02/26/2022 13:08:13 - INFO - codeparrot_training - Step 45989: {'lr': 8.565190090660286e-06, 'samples': 23546880, 'steps': 45989, 'loss/train': 2.2545809745788574}
+02/26/2022 13:08:18 - INFO - codeparrot_training - Step 45990: {'lr': 8.560944313425772e-06, 'samples': 23547392, 'steps': 45990, 'loss/train': 1.7034109830856323}
+02/26/2022 13:08:24 - INFO - codeparrot_training - Step 45991: {'lr': 8.556699570439503e-06, 'samples': 23547904, 'steps': 45991, 'loss/train': 1.6170326471328735}
+02/26/2022 13:08:27 - INFO - codeparrot_training - Step 45992: {'lr': 8.552455861719655e-06, 'samples': 23548416, 'steps': 45992, 'loss/train': 1.6514886617660522}
+02/26/2022 13:08:33 - INFO - codeparrot_training - Step 45993: {'lr': 8.548213187284409e-06, 'samples': 23548928, 'steps': 45993, 'loss/train': 1.1597504615783691}
+02/26/2022 13:08:36 - INFO - codeparrot_training - Step 45994: {'lr': 8.543971547151946e-06, 'samples': 23549440, 'steps': 45994, 'loss/train': 1.3295881748199463}
+02/26/2022 13:08:42 - INFO - codeparrot_training - Step 45995: {'lr': 8.539730941340473e-06, 'samples': 23549952, 'steps': 45995, 'loss/train': 1.5939793586730957}
+02/26/2022 13:08:45 - INFO - codeparrot_training - Step 45996: {'lr': 8.535491369868087e-06, 'samples': 23550464, 'steps': 45996, 'loss/train': 1.8474210500717163}
+02/26/2022 13:08:51 - INFO - codeparrot_training - Step 45997: {'lr': 8.531252832752967e-06, 'samples': 23550976, 'steps': 45997, 'loss/train': 1.1621639728546143}
+02/26/2022 13:08:55 - INFO - codeparrot_training - Step 45998: {'lr': 8.52701533001332e-06, 'samples': 23551488, 'steps': 45998, 'loss/train': 1.7301279306411743}
+02/26/2022 13:09:00 - INFO - codeparrot_training - Step 45999: {'lr': 8.522778861667273e-06, 'samples': 23552000, 'steps': 45999, 'loss/train': 1.087491512298584}
+02/26/2022 13:09:00 - INFO - codeparrot_training - Evaluating and saving model checkpoint