diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -18056,3 +18056,2009 @@ Use FP16 precision: False
 01/27/2022 12:46:44 - INFO - codeparrot_training - Step 17998: {'lr': 0.0003750283395440647, 'samples': 3455808, 'steps': 17998, 'loss/train': 0.6563953310251236}
 01/27/2022 12:46:47 - INFO - codeparrot_training - Step 17999: {'lr': 0.0003750141700397928, 'samples': 3456000, 'steps': 17999, 'loss/train': 0.7130634784698486}
 01/27/2022 12:46:47 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+01/27/2022 12:47:05 - WARNING - huggingface_hub.repository - Several commits (9) will be pushed upstream.
+01/27/2022 12:47:05 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+01/27/2022 12:47:54 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/codeparrot-neo-125M-py
+   6d4e11b..53bbe6d  royal-monkey-12 -> royal-monkey-12
+
+01/27/2022 12:47:58 - INFO - codeparrot_training - Step 18000: {'lr': 0.000375, 'samples': 3456192, 'steps': 18000, 'loss/train': 0.7783340513706207}
+01/27/2022 12:48:01 - INFO - codeparrot_training - Step 18001: {'lr': 0.0003749858294247469, 'samples': 3456384, 'steps': 18001, 'loss/train': 0.3681304603815079}
+01/27/2022 12:48:04 - INFO - codeparrot_training - Step 18002: {'lr': 0.0003749716583140942, 'samples': 3456576, 'steps': 18002, 'loss/train': 1.278022974729538}
+01/27/2022 12:48:08 - INFO - codeparrot_training - Step 18003: {'lr': 0.00037495748666810255, 'samples': 3456768, 'steps': 18003, 'loss/train': 0.6583799421787262}
+01/27/2022 12:48:12 - INFO - codeparrot_training - Step 18004: {'lr': 0.0003749433144868327, 'samples': 3456960, 'steps': 18004, 'loss/train': 0.7187605798244476}
+01/27/2022 12:48:15 - INFO - codeparrot_training - Step 18005: {'lr': 0.00037492914177034544, 'samples': 3457152, 'steps': 18005, 'loss/train': 1.185696691274643}
+01/27/2022 12:48:18 - INFO - codeparrot_training - Step 18006: {'lr': 0.00037491496851870134, 'samples': 3457344, 'steps': 18006, 'loss/train': 0.7499656677246094}
+01/27/2022 12:48:22 - INFO - codeparrot_training - Step 18007: {'lr': 0.0003749007947319612, 'samples': 3457536, 'steps': 18007, 'loss/train': 0.8582116663455963}
+01/27/2022 12:48:25 - INFO - codeparrot_training - Step 18008: {'lr': 0.00037488662041018574, 'samples': 3457728, 'steps': 18008, 'loss/train': 0.7931037247180939}
+01/27/2022 12:48:28 - INFO - codeparrot_training - Step 18009: {'lr': 0.0003748724455534356, 'samples': 3457920, 'steps': 18009, 'loss/train': 0.8969061076641083}
+01/27/2022 12:48:31 - INFO - codeparrot_training - Step 18010: {'lr': 0.0003748582701617716, 'samples': 3458112, 'steps': 18010, 'loss/train': 0.6379171907901764}
+01/27/2022 12:48:34 - INFO - codeparrot_training - Step 18011: {'lr': 0.00037484409423525446, 'samples': 3458304, 'steps': 18011, 'loss/train': 0.7626882791519165}
+01/27/2022 12:48:37 - INFO - codeparrot_training - Step 18012: {'lr': 0.00037482991777394476, 'samples': 3458496, 'steps': 18012, 'loss/train': 0.790972888469696}
+01/27/2022 12:48:42 - INFO - codeparrot_training - Step 18013: {'lr': 0.00037481574077790334, 'samples': 3458688, 'steps': 18013, 'loss/train': 1.1498347520828247}
+01/27/2022 12:48:45 - INFO - codeparrot_training - Step 18014: {'lr': 0.00037480156324719093, 'samples': 3458880, 'steps': 18014, 'loss/train': 0.9776750206947327}
+01/27/2022 12:48:49 - INFO - codeparrot_training - Step 18015: {'lr': 0.00037478738518186835, 'samples': 3459072, 'steps': 18015, 'loss/train': 0.9638514518737793}
+01/27/2022 12:48:52 - INFO - codeparrot_training - Step 18016: {'lr': 0.00037477320658199615, 'samples': 3459264, 'steps': 18016, 'loss/train': 1.1045761406421661}
+01/27/2022 12:48:55 - INFO - codeparrot_training - Step 18017: {'lr': 0.0003747590274476351, 'samples': 3459456, 'steps': 18017, 'loss/train': 0.8492186665534973}
+01/27/2022 12:48:58 - INFO - codeparrot_training - Step 18018: {'lr': 0.0003747448477788461, 'samples': 3459648, 'steps': 18018, 'loss/train': 0.8036980926990509}
+01/27/2022 12:49:01 - INFO - codeparrot_training - Step 18019: {'lr': 0.0003747306675756898, 'samples': 3459840, 'steps': 18019, 'loss/train': 0.5995711237192154}
+01/27/2022 12:49:04 - INFO - codeparrot_training - Step 18020: {'lr': 0.00037471648683822683, 'samples': 3460032, 'steps': 18020, 'loss/train': 0.6237753331661224}
+01/27/2022 12:49:07 - INFO - codeparrot_training - Step 18021: {'lr': 0.00037470230556651814, 'samples': 3460224, 'steps': 18021, 'loss/train': 0.5160147249698639}
+01/27/2022 12:49:14 - INFO - codeparrot_training - Step 18022: {'lr': 0.00037468812376062423, 'samples': 3460416, 'steps': 18022, 'loss/train': 0.6389119327068329}
+01/27/2022 12:49:17 - INFO - codeparrot_training - Step 18023: {'lr': 0.00037467394142060614, 'samples': 3460608, 'steps': 18023, 'loss/train': 1.0112216770648956}
+01/27/2022 12:49:20 - INFO - codeparrot_training - Step 18024: {'lr': 0.0003746597585465243, 'samples': 3460800, 'steps': 18024, 'loss/train': 1.070839673280716}
+01/27/2022 12:49:23 - INFO - codeparrot_training - Step 18025: {'lr': 0.00037464557513843975, 'samples': 3460992, 'steps': 18025, 'loss/train': 0.7032258063554764}
+01/27/2022 12:49:26 - INFO - codeparrot_training - Step 18026: {'lr': 0.0003746313911964132, 'samples': 3461184, 'steps': 18026, 'loss/train': 1.4100111722946167}
+01/27/2022 12:49:29 - INFO - codeparrot_training - Step 18027: {'lr': 0.00037461720672050524, 'samples': 3461376, 'steps': 18027, 'loss/train': 1.2516871690750122}
+01/27/2022 12:49:33 - INFO - codeparrot_training - Step 18028: {'lr': 0.0003746030217107768, 'samples': 3461568, 'steps': 18028, 'loss/train': 1.1812334954738617}
+01/27/2022 12:49:36 - INFO - codeparrot_training - Step 18029: {'lr': 0.0003745888361672885, 'samples': 3461760, 'steps': 18029, 'loss/train': 0.7647760212421417}
+01/27/2022 12:49:39 - INFO - codeparrot_training - Step 18030: {'lr': 0.00037457465009010124, 'samples': 3461952, 'steps': 18030, 'loss/train': 0.8210785388946533}
+01/27/2022 12:49:43 - INFO - codeparrot_training - Step 18031: {'lr': 0.00037456046347927576, 'samples': 3462144, 'steps': 18031, 'loss/train': 0.9385431110858917}
+01/27/2022 12:49:46 - INFO - codeparrot_training - Step 18032: {'lr': 0.0003745462763348727, 'samples': 3462336, 'steps': 18032, 'loss/train': 0.7869641482830048}
+01/27/2022 12:49:50 - INFO - codeparrot_training - Step 18033: {'lr': 0.00037453208865695305, 'samples': 3462528, 'steps': 18033, 'loss/train': 0.7028255313634872}
+01/27/2022 12:49:53 - INFO - codeparrot_training - Step 18034: {'lr': 0.0003745179004455774, 'samples': 3462720, 'steps': 18034, 'loss/train': 0.8299089074134827}
+01/27/2022 12:49:56 - INFO - codeparrot_training - Step 18035: {'lr': 0.00037450371170080673, 'samples': 3462912, 'steps': 18035, 'loss/train': 0.5473497211933136}
+01/27/2022 12:49:59 - INFO - codeparrot_training - Step 18036: {'lr': 0.00037448952242270155, 'samples': 3463104, 'steps': 18036, 'loss/train': 0.5401851385831833}
+01/27/2022 12:50:02 - INFO - codeparrot_training - Step 18037: {'lr': 0.0003744753326113229, 'samples': 3463296, 'steps': 18037, 'loss/train': 1.1735135614871979}
+01/27/2022 12:50:05 - INFO - codeparrot_training - Step 18038: {'lr': 0.00037446114226673136, 'samples': 3463488, 'steps': 18038, 'loss/train': 1.0581133961677551}
+01/27/2022 12:50:08 - INFO - codeparrot_training - Step 18039: {'lr': 0.00037444695138898784, 'samples': 3463680, 'steps': 18039, 'loss/train': 0.9669063091278076}
+01/27/2022 12:50:13 - INFO - codeparrot_training - Step 18040: {'lr': 0.00037443275997815306, 'samples': 3463872, 'steps': 18040, 'loss/train': 1.0502272546291351}
+01/27/2022 12:50:16 - INFO - codeparrot_training - Step 18041: {'lr': 0.00037441856803428785, 'samples': 3464064, 'steps': 18041, 'loss/train': 0.7059443145990372}
+01/27/2022 12:50:20 - INFO - codeparrot_training - Step 18042: {'lr': 0.0003744043755574531, 'samples': 3464256, 'steps': 18042, 'loss/train': 0.4531770497560501}
+01/27/2022 12:50:23 - INFO - codeparrot_training - Step 18043: {'lr': 0.0003743901825477094, 'samples': 3464448, 'steps': 18043, 'loss/train': 0.46784707903862}
+01/27/2022 12:50:26 - INFO - codeparrot_training - Step 18044: {'lr': 0.0003743759890051177, 'samples': 3464640, 'steps': 18044, 'loss/train': 1.2443499863147736}
+01/27/2022 12:50:29 - INFO - codeparrot_training - Step 18045: {'lr': 0.00037436179492973876, 'samples': 3464832, 'steps': 18045, 'loss/train': 1.0601210296154022}
+01/27/2022 12:50:32 - INFO - codeparrot_training - Step 18046: {'lr': 0.0003743476003216334, 'samples': 3465024, 'steps': 18046, 'loss/train': 0.7474950850009918}
+01/27/2022 12:50:35 - INFO - codeparrot_training - Step 18047: {'lr': 0.00037433340518086235, 'samples': 3465216, 'steps': 18047, 'loss/train': 0.8579758107662201}
+01/27/2022 12:50:42 - INFO - codeparrot_training - Step 18048: {'lr': 0.0003743192095074866, 'samples': 3465408, 'steps': 18048, 'loss/train': 1.0479492545127869}
+01/27/2022 12:50:45 - INFO - codeparrot_training - Step 18049: {'lr': 0.0003743050133015666, 'samples': 3465600, 'steps': 18049, 'loss/train': 0.7982334494590759}
+01/27/2022 12:50:48 - INFO - codeparrot_training - Step 18050: {'lr': 0.0003742908165631636, 'samples': 3465792, 'steps': 18050, 'loss/train': 1.0291568040847778}
+01/27/2022 12:50:51 - INFO - codeparrot_training - Step 18051: {'lr': 0.00037427661929233816, 'samples': 3465984, 'steps': 18051, 'loss/train': 0.2324569970369339}
+01/27/2022 12:50:54 - INFO - codeparrot_training - Step 18052: {'lr': 0.00037426242148915113, 'samples': 3466176, 'steps': 18052, 'loss/train': 0.9125169217586517}
+01/27/2022 12:50:57 - INFO - codeparrot_training - Step 18053: {'lr': 0.0003742482231536633, 'samples': 3466368, 'steps': 18053, 'loss/train': 0.34927859902381897}
+01/27/2022 12:51:00 - INFO - codeparrot_training - Step 18054: {'lr': 0.00037423402428593555, 'samples': 3466560, 'steps': 18054, 'loss/train': 0.8629465699195862}
+01/27/2022 12:51:03 - INFO - codeparrot_training - Step 18055: {'lr': 0.00037421982488602875, 'samples': 3466752, 'steps': 18055, 'loss/train': 0.9183895289897919}
+01/27/2022 12:51:07 - INFO - codeparrot_training - Step 18056: {'lr': 0.0003742056249540036, 'samples': 3466944, 'steps': 18056, 'loss/train': 0.9926992356777191}
+01/27/2022 12:51:11 - INFO - codeparrot_training - Step 18057: {'lr': 0.00037419142448992103, 'samples': 3467136, 'steps': 18057, 'loss/train': 0.8707674443721771}
+01/27/2022 12:51:14 - INFO - codeparrot_training - Step 18058: {'lr': 0.0003741772234938418, 'samples': 3467328, 'steps': 18058, 'loss/train': 0.7491530030965805}
+01/27/2022 12:51:17 - INFO - codeparrot_training - Step 18059: {'lr': 0.00037416302196582684, 'samples': 3467520, 'steps': 18059, 'loss/train': 0.9256887137889862}
+01/27/2022 12:51:21 - INFO - codeparrot_training - Step 18060: {'lr': 0.00037414881990593686, 'samples': 3467712, 'steps': 18060, 'loss/train': 0.47925424575805664}
+01/27/2022 12:51:24 - INFO - codeparrot_training - Step 18061: {'lr': 0.0003741346173142328, 'samples': 3467904, 'steps': 18061, 'loss/train': 0.9958025515079498}
+01/27/2022 12:51:27 - INFO - codeparrot_training - Step 18062: {'lr': 0.0003741204141907754, 'samples': 3468096, 'steps': 18062, 'loss/train': 1.2666729390621185}
+01/27/2022 12:51:30 - INFO - codeparrot_training - Step 18063: {'lr': 0.00037410621053562563, 'samples': 3468288, 'steps': 18063, 'loss/train': 0.9110271334648132}
+01/27/2022 12:51:33 - INFO - codeparrot_training - Step 18064: {'lr': 0.0003740920063488442, 'samples': 3468480, 'steps': 18064, 'loss/train': 0.9380240142345428}
+01/27/2022 12:51:36 - INFO - codeparrot_training - Step 18065: {'lr': 0.0003740778016304921, 'samples': 3468672, 'steps': 18065, 'loss/train': 0.7375893294811249}
+01/27/2022 12:51:41 - INFO - codeparrot_training - Step 18066: {'lr': 0.00037406359638063005, 'samples': 3468864, 'steps': 18066, 'loss/train': 0.9161542654037476}
+01/27/2022 12:51:44 - INFO - codeparrot_training - Step 18067: {'lr': 0.00037404939059931884, 'samples': 3469056, 'steps': 18067, 'loss/train': 0.5508367717266083}
+01/27/2022 12:51:47 - INFO - codeparrot_training - Step 18068: {'lr': 0.0003740351842866196, 'samples': 3469248, 'steps': 18068, 'loss/train': 0.9219228029251099}
+01/27/2022 12:51:50 - INFO - codeparrot_training - Step 18069: {'lr': 0.0003740209774425929, 'samples': 3469440, 'steps': 18069, 'loss/train': 0.6948379576206207}
+01/27/2022 12:51:53 - INFO - codeparrot_training - Step 18070: {'lr': 0.00037400677006729977, 'samples': 3469632, 'steps': 18070, 'loss/train': 0.6647333800792694}
+01/27/2022 12:51:56 - INFO - codeparrot_training - Step 18071: {'lr': 0.000373992562160801, 'samples': 3469824, 'steps': 18071, 'loss/train': 0.7119448632001877}
+01/27/2022 12:51:59 - INFO - codeparrot_training - Step 18072: {'lr': 0.0003739783537231575, 'samples': 3470016, 'steps': 18072, 'loss/train': 0.8276842832565308}
+01/27/2022 12:52:03 - INFO - codeparrot_training - Step 18073: {'lr': 0.0003739641447544301, 'samples': 3470208, 'steps': 18073, 'loss/train': 0.7707600295543671}
+01/27/2022 12:52:06 - INFO - codeparrot_training - Step 18074: {'lr': 0.0003739499352546796, 'samples': 3470400, 'steps': 18074, 'loss/train': 0.17535356059670448}
+01/27/2022 12:52:12 - INFO - codeparrot_training - Step 18075: {'lr': 0.00037393572522396697, 'samples': 3470592, 'steps': 18075, 'loss/train': 1.0343612730503082}
+01/27/2022 12:52:15 - INFO - codeparrot_training - Step 18076: {'lr': 0.00037392151466235303, 'samples': 3470784, 'steps': 18076, 'loss/train': 0.9906982183456421}
+01/27/2022 12:52:18 - INFO - codeparrot_training - Step 18077: {'lr': 0.00037390730356989864, 'samples': 3470976, 'steps': 18077, 'loss/train': 0.479432612657547}
+01/27/2022 12:52:22 - INFO - codeparrot_training - Step 18078: {'lr': 0.0003738930919466648, 'samples': 3471168, 'steps': 18078, 'loss/train': 0.6286733597517014}
+01/27/2022 12:52:25 - INFO - codeparrot_training - Step 18079: {'lr': 0.0003738788797927122, 'samples': 3471360, 'steps': 18079, 'loss/train': 0.7387121915817261}
+01/27/2022 12:52:28 - INFO - codeparrot_training - Step 18080: {'lr': 0.0003738646671081019, 'samples': 3471552, 'steps': 18080, 'loss/train': 0.6951380521059036}
+01/27/2022 12:52:31 - INFO - codeparrot_training - Step 18081: {'lr': 0.0003738504538928947, 'samples': 3471744, 'steps': 18081, 'loss/train': 1.4215311706066132}
+01/27/2022 12:52:34 - INFO - codeparrot_training - Step 18082: {'lr': 0.0003738362401471514, 'samples': 3471936, 'steps': 18082, 'loss/train': 0.601099893450737}
+01/27/2022 12:52:39 - INFO - codeparrot_training - Step 18083: {'lr': 0.00037382202587093305, 'samples': 3472128, 'steps': 18083, 'loss/train': 0.6007767766714096}
+01/27/2022 12:52:42 - INFO - codeparrot_training - Step 18084: {'lr': 0.0003738078110643004, 'samples': 3472320, 'steps': 18084, 'loss/train': 0.15557771176099777}
+01/27/2022 12:52:45 - INFO - codeparrot_training - Step 18085: {'lr': 0.0003737935957273144, 'samples': 3472512, 'steps': 18085, 'loss/train': 1.1921128034591675}
+01/27/2022 12:52:48 - INFO - codeparrot_training - Step 18086: {'lr': 0.000373779379860036, 'samples': 3472704, 'steps': 18086, 'loss/train': 0.710585430264473}
+01/27/2022 12:52:51 - INFO - codeparrot_training - Step 18087: {'lr': 0.000373765163462526, 'samples': 3472896, 'steps': 18087, 'loss/train': 0.849108874797821}
+01/27/2022 12:52:54 - INFO - codeparrot_training - Step 18088: {'lr': 0.0003737509465348453, 'samples': 3473088, 'steps': 18088, 'loss/train': 0.37000246345996857}
+01/27/2022 12:52:57 - INFO - codeparrot_training - Step 18089: {'lr': 0.0003737367290770549, 'samples': 3473280, 'steps': 18089, 'loss/train': 0.8966048955917358}
+01/27/2022 12:53:01 - INFO - codeparrot_training - Step 18090: {'lr': 0.0003737225110892156, 'samples': 3473472, 'steps': 18090, 'loss/train': 1.2181785106658936}
+01/27/2022 12:53:04 - INFO - codeparrot_training - Step 18091: {'lr': 0.0003737082925713884, 'samples': 3473664, 'steps': 18091, 'loss/train': 0.509766697883606}
+01/27/2022 12:53:10 - INFO - codeparrot_training - Step 18092: {'lr': 0.00037369407352363417, 'samples': 3473856, 'steps': 18092, 'loss/train': 1.0291243493556976}
+01/27/2022 12:53:13 - INFO - codeparrot_training - Step 18093: {'lr': 0.0003736798539460138, 'samples': 3474048, 'steps': 18093, 'loss/train': 1.848858118057251}
+01/27/2022 12:53:16 - INFO - codeparrot_training - Step 18094: {'lr': 0.00037366563383858814, 'samples': 3474240, 'steps': 18094, 'loss/train': 0.8494852781295776}
+01/27/2022 12:53:19 - INFO - codeparrot_training - Step 18095: {'lr': 0.0003736514132014182, 'samples': 3474432, 'steps': 18095, 'loss/train': 1.011221319437027}
+01/27/2022 12:53:22 - INFO - codeparrot_training - Step 18096: {'lr': 0.0003736371920345649, 'samples': 3474624, 'steps': 18096, 'loss/train': 0.6905400305986404}
+01/27/2022 12:53:26 - INFO - codeparrot_training - Step 18097: {'lr': 0.0003736229703380891, 'samples': 3474816, 'steps': 18097, 'loss/train': 0.8966100811958313}
+01/27/2022 12:53:29 - INFO - codeparrot_training - Step 18098: {'lr': 0.0003736087481120518, 'samples': 3475008, 'steps': 18098, 'loss/train': 1.0186015963554382}
+01/27/2022 12:53:32 - INFO - codeparrot_training - Step 18099: {'lr': 0.0003735945253565138, 'samples': 3475200, 'steps': 18099, 'loss/train': 0.6997658461332321}
+01/27/2022 12:53:35 - INFO - codeparrot_training - Step 18100: {'lr': 0.00037358030207153617, 'samples': 3475392, 'steps': 18100, 'loss/train': 0.7956440448760986}
+01/27/2022 12:53:39 - INFO - codeparrot_training - Step 18101: {'lr': 0.0003735660782571797, 'samples': 3475584, 'steps': 18101, 'loss/train': 0.9437715411186218}
+01/27/2022 12:53:43 - INFO - codeparrot_training - Step 18102: {'lr': 0.00037355185391350546, 'samples': 3475776, 'steps': 18102, 'loss/train': 2.572937250137329}
+01/27/2022 12:53:46 - INFO - codeparrot_training - Step 18103: {'lr': 0.00037353762904057425, 'samples': 3475968, 'steps': 18103, 'loss/train': 0.6230118423700333}
+01/27/2022 12:53:49 - INFO - codeparrot_training - Step 18104: {'lr': 0.00037352340363844706, 'samples': 3476160, 'steps': 18104, 'loss/train': 0.4537196159362793}
+01/27/2022 12:53:52 - INFO - codeparrot_training - Step 18105: {'lr': 0.0003735091777071849, 'samples': 3476352, 'steps': 18105, 'loss/train': 0.8918675780296326}
+01/27/2022 12:53:55 - INFO - codeparrot_training - Step 18106: {'lr': 0.0003734949512468486, 'samples': 3476544, 'steps': 18106, 'loss/train': 0.6889788061380386}
+01/27/2022 12:53:58 - INFO - codeparrot_training - Step 18107: {'lr': 0.0003734807242574991, 'samples': 3476736, 'steps': 18107, 'loss/train': 0.7090197801589966}
+01/27/2022 12:54:01 - INFO - codeparrot_training - Step 18108: {'lr': 0.0003734664967391975, 'samples': 3476928, 'steps': 18108, 'loss/train': 0.6923428326845169}
+01/27/2022 12:54:05 - INFO - codeparrot_training - Step 18109: {'lr': 0.0003734522686920045, 'samples': 3477120, 'steps': 18109, 'loss/train': 0.7861449122428894}
+01/27/2022 12:54:09 - INFO - codeparrot_training - Step 18110: {'lr': 0.00037343804011598125, 'samples': 3477312, 'steps': 18110, 'loss/train': 0.5216350704431534}
+01/27/2022 12:54:12 - INFO - codeparrot_training - Step 18111: {'lr': 0.0003734238110111886, 'samples': 3477504, 'steps': 18111, 'loss/train': 1.1047285795211792}
+01/27/2022 12:54:15 - INFO - codeparrot_training - Step 18112: {'lr': 0.0003734095813776876, 'samples': 3477696, 'steps': 18112, 'loss/train': 0.7929088175296783}
+01/27/2022 12:54:18 - INFO - codeparrot_training - Step 18113: {'lr': 0.00037339535121553907, 'samples': 3477888, 'steps': 18113, 'loss/train': 0.5577153414487839}
+01/27/2022 12:54:22 - INFO - codeparrot_training - Step 18114: {'lr': 0.0003733811205248041, 'samples': 3478080, 'steps': 18114, 'loss/train': 1.6866998076438904}
+01/27/2022 12:54:25 - INFO - codeparrot_training - Step 18115: {'lr': 0.0003733668893055435, 'samples': 3478272, 'steps': 18115, 'loss/train': 0.30324290692806244}
+01/27/2022 12:54:28 - INFO - codeparrot_training - Step 18116: {'lr': 0.00037335265755781844, 'samples': 3478464, 'steps': 18116, 'loss/train': 0.3082219138741493}
+01/27/2022 12:54:31 - INFO - codeparrot_training - Step 18117: {'lr': 0.0003733384252816897, 'samples': 3478656, 'steps': 18117, 'loss/train': 0.972191333770752}
+01/27/2022 12:54:34 - INFO - codeparrot_training - Step 18118: {'lr': 0.0003733241924772183, 'samples': 3478848, 'steps': 18118, 'loss/train': 0.8630913197994232}
+01/27/2022 12:54:39 - INFO - codeparrot_training - Step 18119: {'lr': 0.0003733099591444652, 'samples': 3479040, 'steps': 18119, 'loss/train': 0.6814349591732025}
+01/27/2022 12:54:42 - INFO - codeparrot_training - Step 18120: {'lr': 0.0003732957252834914, 'samples': 3479232, 'steps': 18120, 'loss/train': 0.8279191553592682}
+01/27/2022 12:54:45 - INFO - codeparrot_training - Step 18121: {'lr': 0.0003732814908943579, 'samples': 3479424, 'steps': 18121, 'loss/train': 0.4003446400165558}
+01/27/2022 12:54:48 - INFO - codeparrot_training - Step 18122: {'lr': 0.0003732672559771256, 'samples': 3479616, 'steps': 18122, 'loss/train': 0.5027842372655869}
+01/27/2022 12:54:51 - INFO - codeparrot_training - Step 18123: {'lr': 0.00037325302053185555, 'samples': 3479808, 'steps': 18123, 'loss/train': 0.9951060712337494}
+01/27/2022 12:54:54 - INFO - codeparrot_training - Step 18124: {'lr': 0.0003732387845586087, 'samples': 3480000, 'steps': 18124, 'loss/train': 0.34833555668592453}
+01/27/2022 12:54:57 - INFO - codeparrot_training - Step 18125: {'lr': 0.00037322454805744607, 'samples': 3480192, 'steps': 18125, 'loss/train': 0.10397180914878845}
+01/27/2022 12:55:01 - INFO - codeparrot_training - Step 18126: {'lr': 0.0003732103110284285, 'samples': 3480384, 'steps': 18126, 'loss/train': 0.8677283227443695}
+01/27/2022 12:55:04 - INFO - codeparrot_training - Step 18127: {'lr': 0.00037319607347161715, 'samples': 3480576, 'steps': 18127, 'loss/train': 0.7904294729232788}
+01/27/2022 12:55:10 - INFO - codeparrot_training - Step 18128: {'lr': 0.0003731818353870729, 'samples': 3480768, 'steps': 18128, 'loss/train': 0.6746397167444229}
+01/27/2022 12:55:13 - INFO - codeparrot_training - Step 18129: {'lr': 0.0003731675967748568, 'samples': 3480960, 'steps': 18129, 'loss/train': 0.6195082068443298}
+01/27/2022 12:55:16 - INFO - codeparrot_training - Step 18130: {'lr': 0.0003731533576350298, 'samples': 3481152, 'steps': 18130, 'loss/train': 0.8012702465057373}
+01/27/2022 12:55:19 - INFO - codeparrot_training - Step 18131: {'lr': 0.000373139117967653, 'samples': 3481344, 'steps': 18131, 'loss/train': 0.854013204574585}
+01/27/2022 12:55:23 - INFO - codeparrot_training - Step 18132: {'lr': 0.00037312487777278725, 'samples': 3481536, 'steps': 18132, 'loss/train': 0.3541881814599037}
+01/27/2022 12:55:26 - INFO - codeparrot_training - Step 18133: {'lr': 0.00037311063705049364, 'samples': 3481728, 'steps': 18133, 'loss/train': 0.6126158237457275}
+01/27/2022 12:55:29 - INFO - codeparrot_training - Step 18134: {'lr': 0.00037309639580083314, 'samples': 3481920, 'steps': 18134, 'loss/train': 0.21457500010728836}
+01/27/2022 12:55:32 - INFO - codeparrot_training - Step 18135: {'lr': 0.0003730821540238668, 'samples': 3482112, 'steps': 18135, 'loss/train': 0.8138260245323181}
+01/27/2022 12:55:35 - INFO - codeparrot_training - Step 18136: {'lr': 0.0003730679117196556, 'samples': 3482304, 'steps': 18136, 'loss/train': 0.9166905283927917}
+01/27/2022 12:55:40 - INFO - codeparrot_training - Step 18137: {'lr': 0.0003730536688882605, 'samples': 3482496, 'steps': 18137, 'loss/train': 0.7769080102443695}
+01/27/2022 12:55:43 - INFO - codeparrot_training - Step 18138: {'lr': 0.00037303942552974255, 'samples': 3482688, 'steps': 18138, 'loss/train': 0.7092679738998413}
+01/27/2022 12:55:46 - INFO - codeparrot_training - Step 18139: {'lr': 0.0003730251816441628, 'samples': 3482880, 'steps': 18139, 'loss/train': 0.6704412549734116}
+01/27/2022 12:55:49 - INFO - codeparrot_training - Step 18140: {'lr': 0.00037301093723158223, 'samples': 3483072, 'steps': 18140, 'loss/train': 0.9117922782897949}
+01/27/2022 12:55:52 - INFO - codeparrot_training - Step 18141: {'lr': 0.0003729966922920619, 'samples': 3483264, 'steps': 18141, 'loss/train': 0.635067343711853}
+01/27/2022 12:55:55 - INFO - codeparrot_training - Step 18142: {'lr': 0.00037298244682566273, 'samples': 3483456, 'steps': 18142, 'loss/train': 0.9346261918544769}
+01/27/2022 12:55:58 - INFO - codeparrot_training - Step 18143: {'lr': 0.0003729682008324459, 'samples': 3483648, 'steps': 18143, 'loss/train': 0.8442114293575287}
+01/27/2022 12:56:01 - INFO - codeparrot_training - Step 18144: {'lr': 0.00037295395431247223, 'samples': 3483840, 'steps': 18144, 'loss/train': 0.9342556893825531}
+01/27/2022 12:56:05 - INFO - codeparrot_training - Step 18145: {'lr': 0.000372939707265803, 'samples': 3484032, 'steps': 18145, 'loss/train': 1.0331157445907593}
+01/27/2022 12:56:09 - INFO - codeparrot_training - Step 18146: {'lr': 0.000372925459692499, 'samples': 3484224, 'steps': 18146, 'loss/train': 0.24358152598142624}
+01/27/2022 12:56:12 - INFO - codeparrot_training - Step 18147: {'lr': 0.0003729112115926214, 'samples': 3484416, 'steps': 18147, 'loss/train': 0.543109193444252}
+01/27/2022 12:56:15 - INFO - codeparrot_training - Step 18148: {'lr': 0.00037289696296623117, 'samples': 3484608, 'steps': 18148, 'loss/train': 0.7936931848526001}
+01/27/2022 12:56:18 - INFO - codeparrot_training - Step 18149: {'lr': 0.00037288271381338936, 'samples': 3484800, 'steps': 18149, 'loss/train': 0.5351973474025726}
+01/27/2022 12:56:22 - INFO - codeparrot_training - Step 18150: {'lr': 0.0003728684641341571, 'samples': 3484992, 'steps': 18150, 'loss/train': 0.3824924826622009}
+01/27/2022 12:56:25 - INFO - codeparrot_training - Step 18151: {'lr': 0.00037285421392859526, 'samples': 3485184, 'steps': 18151, 'loss/train': 0.8739653527736664}
+01/27/2022 12:56:28 - INFO - codeparrot_training - Step 18152: {'lr': 0.00037283996319676505, 'samples': 3485376, 'steps': 18152, 'loss/train': 0.5760321021080017}
+01/27/2022 12:56:31 - INFO - codeparrot_training - Step 18153: {'lr': 0.0003728257119387274, 'samples': 3485568, 'steps': 18153, 'loss/train': 0.874194324016571}
+01/27/2022 12:56:34 - INFO - codeparrot_training - Step 18154: {'lr': 0.0003728114601545435, 'samples': 3485760, 'steps': 18154, 'loss/train': 0.8250371217727661}
+01/27/2022 12:56:41 - INFO - codeparrot_training - Step 18155: {'lr': 0.0003727972078442742, 'samples': 3485952, 'steps': 18155, 'loss/train': 1.2344775795936584}
+01/27/2022 12:56:44 - INFO - codeparrot_training - Step 18156: {'lr': 0.0003727829550079806, 'samples': 3486144, 'steps': 18156, 'loss/train': 0.8728508055210114}
+01/27/2022 12:56:47 - INFO - codeparrot_training - Step 18157: {'lr': 0.00037276870164572394, 'samples': 3486336, 'steps': 18157, 'loss/train': 0.3032447621226311}
+01/27/2022 12:56:50 - INFO - codeparrot_training - Step 18158: {'lr': 0.00037275444775756506, 'samples': 3486528, 'steps': 18158, 'loss/train': 1.0159862637519836}
+01/27/2022 12:56:53 - INFO - codeparrot_training - Step 18159: {'lr': 0.00037274019334356516, 'samples': 3486720, 'steps': 18159, 'loss/train': 0.6928951442241669}
+01/27/2022 12:56:56 - INFO - codeparrot_training - Step 18160: {'lr': 0.0003727259384037852, 'samples': 3486912, 'steps': 18160, 'loss/train': 0.934124618768692}
+01/27/2022 12:57:00 - INFO - codeparrot_training - Step 18161: {'lr': 0.0003727116829382864, 'samples': 3487104, 'steps': 18161, 'loss/train': 0.9499973952770233}
+01/27/2022 12:57:03 - INFO - codeparrot_training - Step 18162: {'lr': 0.00037269742694712963, 'samples': 3487296, 'steps': 18162, 'loss/train': 0.8307521045207977}
+01/27/2022 12:57:08 - INFO - codeparrot_training - Step 18163: {'lr': 0.000372683170430376, 'samples': 3487488, 'steps': 18163, 'loss/train': 1.0424510836601257}
+01/27/2022 12:57:11 - INFO - codeparrot_training - Step 18164: {'lr': 0.00037266891338808667, 'samples': 3487680, 'steps': 18164, 'loss/train': 0.9442936778068542}
+01/27/2022 12:57:14 - INFO - codeparrot_training - Step 18165: {'lr': 0.0003726546558203227, 'samples': 3487872, 'steps': 18165, 'loss/train': 0.4126611649990082}
+01/27/2022 12:57:17 - INFO - codeparrot_training - Step 18166: {'lr': 0.0003726403977271451, 'samples': 3488064, 'steps': 18166, 'loss/train': 0.6384381204843521}
+01/27/2022 12:57:20 - INFO - codeparrot_training - Step 18167: {'lr': 0.00037262613910861485, 'samples': 3488256, 'steps': 18167, 'loss/train': 0.8885832130908966}
+01/27/2022 12:57:23 - INFO - codeparrot_training - Step 18168: {'lr': 0.0003726118799647933, 'samples': 3488448, 'steps': 18168, 'loss/train': 0.11224985495209694}
+01/27/2022 12:57:26 - INFO - codeparrot_training - Step 18169: {'lr': 0.0003725976202957414, 'samples': 3488640, 'steps': 18169, 'loss/train': 1.015294075012207}
+01/27/2022 12:57:30 - INFO - codeparrot_training - Step 18170: {'lr': 0.0003725833601015202, 'samples': 3488832, 'steps': 18170, 'loss/train': 0.34170931577682495}
+01/27/2022 12:57:33 - INFO - codeparrot_training - Step 18171: {'lr': 0.0003725690993821908, 'samples': 3489024, 'steps': 18171, 'loss/train': 0.8535616099834442}
+01/27/2022 12:57:39 - INFO - codeparrot_training - Step 18172: {'lr': 0.00037255483813781434, 'samples': 3489216, 'steps': 18172, 'loss/train': 0.5719913095235825}
+01/27/2022 12:57:42 - INFO - codeparrot_training - Step 18173: {'lr': 0.00037254057636845186, 'samples': 3489408, 'steps': 18173, 'loss/train': 0.8858694434165955}
+01/27/2022 12:57:45 - INFO - codeparrot_training - Step 18174: {'lr': 0.0003725263140741644, 'samples': 3489600, 'steps': 18174, 'loss/train': 0.8893769681453705}
+01/27/2022 12:57:48 - INFO - codeparrot_training - Step 18175: {'lr': 0.0003725120512550131, 'samples': 3489792, 'steps': 18175, 'loss/train': 0.9640742540359497}
+01/27/2022 12:57:52 - INFO - codeparrot_training - Step 18176: {'lr': 0.00037249778791105916, 'samples': 3489984, 'steps': 18176, 'loss/train': 1.120794653892517}
+01/27/2022 12:57:55 - INFO - codeparrot_training - Step 18177: {'lr': 0.0003724835240423636, 'samples': 3490176, 'steps': 18177, 'loss/train': 0.5758975893259048}
+01/27/2022 12:57:58 - INFO - codeparrot_training - Step 18178: {'lr': 0.0003724692596489874, 'samples': 3490368, 'steps': 18178, 'loss/train': 0.8418464362621307}
+01/27/2022 12:58:01 - INFO - codeparrot_training - Step 18179: {'lr': 0.0003724549947309919, 'samples': 3490560, 'steps': 18179, 'loss/train': 0.25169748812913895}
+01/27/2022 12:58:04 - INFO - codeparrot_training - Step 18180: {'lr': 0.00037244072928843805, 'samples': 3490752, 'steps': 18180, 'loss/train': 0.6230945885181427}
+01/27/2022 12:58:08 - INFO - codeparrot_training - Step 18181: {'lr': 0.000372426463321387, 'samples': 3490944, 'steps': 18181, 'loss/train': 0.6862279325723648}
+01/27/2022 12:58:12 - INFO - codeparrot_training - Step 18182: {'lr': 0.0003724121968298998, 'samples': 3491136, 'steps': 18182, 'loss/train': 0.7366379052400589}
+01/27/2022 12:58:15 - INFO - codeparrot_training - Step 18183: {'lr': 0.0003723979298140377, 'samples': 3491328, 'steps': 18183, 'loss/train': 1.1197682619094849}
+01/27/2022 12:58:18 - INFO - codeparrot_training - Step 18184: {'lr': 0.0003723836622738617, 'samples': 3491520, 'steps': 18184, 'loss/train': 1.1617720127105713}
+01/27/2022 12:58:21 - INFO - codeparrot_training - Step 18185: {'lr': 0.00037236939420943294, 'samples': 3491712, 'steps': 18185, 'loss/train': 0.6545366048812866}
+01/27/2022 12:58:24 - INFO - codeparrot_training - Step 18186: {'lr': 0.0003723551256208125, 'samples': 3491904, 'steps': 18186, 'loss/train': 0.06293873302638531}
+01/27/2022 12:58:27 - INFO - codeparrot_training - Step 18187: {'lr': 0.0003723408565080616, 'samples': 3492096, 'steps': 18187, 'loss/train': 0.9543662667274475}
+01/27/2022 12:58:30 - INFO - codeparrot_training - Step 18188: {'lr': 0.00037232658687124135, 'samples': 3492288, 'steps': 18188, 'loss/train': 0.4376324862241745}
+01/27/2022 12:58:34 - INFO - codeparrot_training - Step 18189: {'lr': 0.0003723123167104128, 'samples': 3492480, 'steps': 18189, 'loss/train': 0.47286437451839447}
+01/27/2022 12:58:38 - INFO - codeparrot_training - Step 18190: {'lr': 0.0003722980460256372, 'samples': 3492672, 'steps': 18190, 'loss/train': 0.9426611065864563}
+01/27/2022 12:58:41 - INFO - codeparrot_training - Step 18191: {'lr': 0.00037228377481697555, 'samples': 3492864, 'steps': 18191, 'loss/train': 0.7290701866149902}
+01/27/2022 12:58:44 - INFO - codeparrot_training - Step 18192: {'lr': 0.0003722695030844891, 'samples': 3493056, 'steps': 18192, 'loss/train': 1.1266021728515625}
+01/27/2022 12:58:47 - INFO - codeparrot_training - Step 18193: {'lr': 0.0003722552308282388, 'samples': 3493248, 'steps': 18193, 'loss/train': 0.6467578858137131}
+01/27/2022 12:58:51 - INFO - codeparrot_training - Step 18194: {'lr': 0.000372240958048286, 'samples': 3493440, 'steps': 18194, 'loss/train': 0.7290550768375397}
+01/27/2022 12:58:54 - INFO - codeparrot_training - Step 18195: {'lr': 0.0003722266847446918, 'samples': 3493632, 'steps': 18195, 'loss/train': 0.6951625943183899}
+01/27/2022 12:58:57 - INFO - codeparrot_training - Step 18196: {'lr': 0.00037221241091751716, 'samples': 3493824, 'steps': 18196, 'loss/train': 1.0365797281265259}
+01/27/2022 12:59:00 - INFO - codeparrot_training - Step 18197: {'lr': 0.00037219813656682346, 'samples': 3494016, 'steps': 18197, 'loss/train': 0.7490474134683609}
+01/27/2022 12:59:03 - INFO - codeparrot_training - Step 18198: {'lr': 0.0003721838616926717, 'samples': 3494208, 'steps': 18198, 'loss/train': 1.2168539464473724}
+01/27/2022 12:59:10 - INFO - codeparrot_training - Step 18199: {'lr': 0.0003721695862951231, 'samples': 3494400, 'steps': 18199, 'loss/train': 1.0521708726882935}
+01/27/2022 12:59:13 - INFO - codeparrot_training - Step 18200: {'lr': 0.0003721553103742388, 'samples': 3494592, 'steps': 18200, 'loss/train': 0.6670533120632172}
+01/27/2022 12:59:16 - INFO - codeparrot_training - Step 18201: {'lr': 0.0003721410339300799, 'samples': 3494784, 'steps': 18201, 'loss/train': 0.35841283947229385}
+01/27/2022 12:59:19 - INFO - codeparrot_training - Step 18202: {'lr': 0.0003721267569627076, 'samples': 3494976, 'steps': 18202, 'loss/train': 0.8085673749446869}
+01/27/2022 12:59:22 - INFO - codeparrot_training - Step 18203: {'lr': 0.00037211247947218306, 'samples': 3495168, 'steps': 18203, 'loss/train': 1.0463350117206573}
+01/27/2022 12:59:25 - INFO - codeparrot_training - Step 18204: {'lr': 0.0003720982014585674, 'samples': 3495360, 'steps': 18204, 'loss/train': 0.4508533626794815}
+01/27/2022 12:59:29 - INFO - codeparrot_training - Step 18205: {'lr': 0.0003720839229219218, 'samples': 3495552, 'steps': 18205, 'loss/train': 0.9725110530853271}
+01/27/2022 12:59:32 - INFO - codeparrot_training - Step 18206: {'lr': 0.00037206964386230754, 'samples': 3495744, 'steps': 18206, 'loss/train': 0.3967955410480499}
+01/27/2022 12:59:36 - INFO - codeparrot_training - Step 18207: {'lr': 0.00037205536427978563, 'samples': 3495936, 'steps': 18207, 'loss/train': 0.04213032592087984}
+01/27/2022 12:59:40 - INFO - codeparrot_training - Step 18208: {'lr': 0.0003720410841744173, 'samples': 3496128, 'steps': 18208, 'loss/train': 0.7325121313333511}
+01/27/2022 12:59:43 - INFO - codeparrot_training - Step 18209: {'lr': 0.0003720268035462637, 'samples': 3496320, 'steps': 18209, 'loss/train': 0.5676687955856323}
+01/27/2022 12:59:46 - INFO - codeparrot_training - Step 18210: {'lr': 0.000372012522395386, 'samples': 3496512, 'steps': 18210, 'loss/train': 0.9903072416782379}
+01/27/2022 12:59:49 - INFO - codeparrot_training - Step 18211: {'lr': 0.00037199824072184546, 'samples': 3496704, 'steps': 18211, 'loss/train': 0.6166180819272995}
+01/27/2022 12:59:52 - INFO - codeparrot_training - Step 18212: {'lr': 0.0003719839585257032, 'samples': 3496896, 'steps': 18212, 'loss/train': 0.903060257434845}
+01/27/2022 12:59:55 - INFO - codeparrot_training - Step 18213: {'lr': 0.00037196967580702036, 'samples': 3497088, 'steps': 18213, 'loss/train': 1.3512671291828156}
+01/27/2022 12:59:58 - INFO - codeparrot_training - Step 18214: {'lr': 0.0003719553925658581, 'samples': 3497280, 'steps': 18214, 'loss/train': 0.31404959410429}
+01/27/2022 13:00:02 - INFO - codeparrot_training - Step 18215: {'lr': 0.00037194110880227777, 'samples': 3497472, 'steps': 18215, 'loss/train': 0.7803693115711212}
+01/27/2022 13:00:06 - INFO - codeparrot_training - Step 18216: {'lr': 0.0003719268245163404, 'samples': 3497664, 'steps': 18216, 'loss/train': 0.5225974023342133}
+01/27/2022 13:00:09 - INFO - codeparrot_training - Step 18217: {'lr': 0.0003719125397081072, 'samples': 3497856, 'steps': 18217, 'loss/train': 1.1301097869873047}
+01/27/2022 13:00:12 - INFO - codeparrot_training - Step 18218: {'lr': 0.00037189825437763946, 'samples': 3498048, 'steps': 18218, 'loss/train': 0.8184877038002014}
+01/27/2022 13:00:15 - INFO - codeparrot_training - Step 18219: {'lr': 0.0003718839685249983, 'samples': 3498240, 'steps': 18219, 'loss/train': 0.49839238822460175}
+01/27/2022 13:00:19 - INFO - codeparrot_training - Step 18220: {'lr': 0.0003718696821502449, 'samples': 3498432, 'steps': 18220, 'loss/train': 0.6624985188245773}
+01/27/2022 13:00:22 - INFO - codeparrot_training - Step 18221: {'lr': 0.0003718553952534405, 'samples': 3498624, 'steps': 18221, 'loss/train': 0.5998035818338394}
+01/27/2022 13:00:25 - INFO - codeparrot_training - Step 18222: {'lr': 0.0003718411078346462, 'samples': 3498816, 'steps': 18222, 'loss/train': 1.0923026204109192}
+01/27/2022 13:00:28 - INFO - codeparrot_training - Step 18223: {'lr': 0.0003718268198939234, 'samples': 3499008, 'steps': 18223, 'loss/train': 0.6527983993291855}
+01/27/2022 13:00:31 - INFO - codeparrot_training - Step 18224: {'lr': 0.0003718125314313331, 'samples': 3499200, 'steps': 18224, 'loss/train': 0.970682680606842}
+01/27/2022 13:00:35 - INFO - codeparrot_training - Step 18225: {'lr': 0.0003717982424469366, 'samples': 3499392, 'steps': 18225, 'loss/train': 0.6771062314510345}
+01/27/2022 13:00:39 - INFO - codeparrot_training - Step 18226: {'lr': 0.00037178395294079516, 'samples': 3499584, 'steps': 18226, 'loss/train': 0.6329424530267715}
+01/27/2022 13:00:42 - INFO - codeparrot_training - Step 18227: {'lr': 0.0003717696629129699, 'samples': 3499776, 'steps': 18227, 'loss/train': 0.42780250310897827}
+01/27/2022 13:00:45 - INFO - codeparrot_training - Step 18228: {'lr': 0.00037175537236352205, 'samples': 3499968, 'steps': 18228, 'loss/train': 1.2889316082000732}
+01/27/2022 13:00:48 - INFO - codeparrot_training - Step 18229: {'lr': 0.0003717410812925129, 'samples': 3500160, 'steps': 18229, 'loss/train': 1.130715250968933}
+01/27/2022 13:00:51 - INFO - codeparrot_training - Step 18230: {'lr': 0.0003717267897000036, 'samples': 3500352, 'steps': 18230, 'loss/train': 0.7970824241638184}
+01/27/2022 13:00:54 - INFO - codeparrot_training - Step 18231: {'lr': 0.0003717124975860554, 'samples': 3500544, 'steps': 18231, 'loss/train': 0.8656455874443054}
+01/27/2022 13:00:57 - INFO - codeparrot_training - Step 18232: {'lr': 0.00037169820495072935, 'samples': 3500736, 'steps': 18232, 'loss/train': 0.24104728549718857}
+01/27/2022 13:01:01 - INFO - codeparrot_training - Step 18233: {'lr': 0.000371683911794087, 'samples': 3500928, 'steps': 18233, 'loss/train': 0.47650909423828125}
+01/27/2022 13:01:07 - INFO - codeparrot_training - Step 18234: {'lr': 0.0003716696181161894, 'samples': 3501120, 'steps': 18234, 'loss/train': 0.8230563104152679}
+01/27/2022 13:01:10 - INFO - codeparrot_training - Step 18235: {'lr': 0.00037165532391709777, 'samples': 3501312, 'steps': 18235, 'loss/train': 0.53752201795578}
+01/27/2022 13:01:13 - INFO - codeparrot_training - Step 18236: {'lr': 0.00037164102919687335, 'samples': 3501504, 'steps': 18236, 'loss/train': 0.5494372844696045}
+01/27/2022 13:01:16 - INFO - codeparrot_training - Step 18237: {'lr': 0.00037162673395557737, 'samples': 3501696, 'steps': 18237, 'loss/train': 0.8688662052154541}
+01/27/2022 13:01:19 - INFO - codeparrot_training - Step 18238: {'lr': 0.0003716124381932711, 'samples': 3501888, 'steps': 18238, 'loss/train': 0.6798280030488968}
+01/27/2022 13:01:22 - INFO - codeparrot_training - Step 18239: {'lr': 0.00037159814191001586, 'samples': 3502080, 'steps': 18239, 'loss/train': 0.7934789657592773}
+01/27/2022 13:01:26 - INFO - codeparrot_training - Step 18240: {'lr': 0.00037158384510587263, 'samples': 3502272, 'steps': 18240, 'loss/train': 0.4958770126104355}
+01/27/2022 13:01:29 - INFO - codeparrot_training - Step 18241: {'lr': 0.0003715695477809029, 'samples': 3502464, 'steps': 18241, 'loss/train': 0.7524427771568298}
+01/27/2022 13:01:33 - INFO - codeparrot_training - Step 18242: {'lr': 0.0003715552499351678, 'samples': 3502656, 'steps': 18242, 'loss/train': 2.2998404502868652}
+01/27/2022 13:01:36 - INFO - codeparrot_training - Step 18243: {'lr': 0.0003715409515687286, 'samples': 3502848, 'steps': 18243, 'loss/train': 0.9852054119110107}
+01/27/2022 13:01:40 - INFO - codeparrot_training - Step 18244: {'lr': 0.00037152665268164664, 'samples': 3503040, 'steps': 18244, 'loss/train': 0.650439977645874}
+01/27/2022 13:01:43 - INFO - codeparrot_training - Step 18245: {'lr': 0.00037151235327398304, 'samples': 3503232, 'steps': 18245, 'loss/train': 0.8506775200366974}
+01/27/2022 13:01:46 - INFO - codeparrot_training - Step 18246: {'lr': 0.000371498053345799, 'samples': 3503424, 'steps': 18246, 'loss/train': 0.22322454303503036}
+01/27/2022 13:01:49 - INFO - codeparrot_training - Step 18247: {'lr': 0.000371483752897156, 'samples': 3503616, 'steps': 18247, 'loss/train': 0.3714127689599991}
+01/27/2022 13:01:52 - INFO - codeparrot_training - Step 18248: {'lr': 0.00037146945192811513, 'samples': 3503808, 'steps': 18248, 'loss/train': 0.9572153985500336}
+01/27/2022 13:01:55 - INFO - codeparrot_training - Step 18249: {'lr': 0.0003714551504387378, 'samples': 3504000, 'steps': 18249, 'loss/train': 0.7633838653564453}
+01/27/2022 13:01:58 - INFO - codeparrot_training - Step 18250: {'lr': 0.000371440848429085, 'samples': 3504192, 'steps': 18250, 'loss/train': 0.5069639682769775}
+01/27/2022 13:02:05 - INFO - codeparrot_training - Step 18251: {'lr': 0.0003714265458992183, 'samples': 3504384, 'steps': 18251, 'loss/train': 0.6772877275943756}
+01/27/2022 13:02:08 - INFO - codeparrot_training - Step 18252: {'lr': 0.00037141224284919876, 'samples': 3504576, 'steps': 18252, 'loss/train': 0.6069973558187485}
+01/27/2022 13:02:11 - INFO - codeparrot_training - Step 18253: {'lr': 0.0003713979392790878, 'samples': 3504768, 'steps': 18253, 'loss/train': 0.6365671008825302}
+01/27/2022 13:02:14 - INFO - codeparrot_training - Step 18254: {'lr': 0.0003713836351889465, 'samples': 3504960, 'steps': 18254, 'loss/train': 1.6882683634757996}
+01/27/2022 13:02:18 - INFO - codeparrot_training - Step 18255: {'lr': 0.00037136933057883636, 'samples': 3505152, 'steps': 18255, 'loss/train': 0.827146589756012}
+01/27/2022 13:02:21 - INFO - codeparrot_training - Step 18256: {'lr': 0.0003713550254488185, 'samples': 3505344, 'steps': 18256, 'loss/train': 0.8606462180614471}
+01/27/2022 13:02:24 - INFO - codeparrot_training - Step 18257: {'lr': 0.0003713407197989543, 'samples': 3505536, 'steps': 18257, 'loss/train': 1.2014252543449402}
+01/27/2022 13:02:27 - INFO - codeparrot_training - Step 18258: {'lr': 0.0003713264136293049, 'samples': 3505728, 'steps': 18258, 'loss/train': 1.072918564081192}
+01/27/2022 13:02:30 - INFO - codeparrot_training - Step 18259: {'lr': 0.0003713121069399317, 'samples': 3505920, 'steps': 18259, 'loss/train': 0.9670339822769165}
+01/27/2022 13:02:35 - INFO - codeparrot_training - Step 18260: {'lr': 0.00037129779973089596, 'samples': 3506112, 'steps': 18260, 'loss/train': 0.8767686188220978}
+01/27/2022 13:02:38 - INFO - codeparrot_training - Step 18261: {'lr': 0.00037128349200225895, 'samples': 3506304, 'steps': 18261, 'loss/train': 0.47849978506565094}
+01/27/2022 13:02:41 - INFO - codeparrot_training - Step 18262: {'lr': 0.000371269183754082, 'samples': 3506496, 'steps': 18262, 'loss/train': 1.108186662197113}
+01/27/2022 13:02:44 - INFO - codeparrot_training - Step 18263: {'lr': 0.00037125487498642636, 'samples': 3506688, 'steps': 18263, 'loss/train': 0.8485361337661743}
+01/27/2022 13:02:47 - INFO - codeparrot_training - Step 18264: {'lr': 0.00037124056569935336, 'samples': 3506880, 'steps': 18264, 'loss/train': 0.4049142301082611}
+01/27/2022 13:02:50 - INFO - codeparrot_training - Step 18265: {'lr': 0.00037122625589292425, 'samples': 3507072, 'steps': 18265, 'loss/train': 1.3758209645748138}
+01/27/2022 13:02:53 - INFO - codeparrot_training - Step 18266: {'lr': 0.0003712119455672004, 'samples': 3507264, 'steps': 18266, 'loss/train': 0.7974366545677185}
+01/27/2022 13:02:57 - INFO - codeparrot_training - Step 18267: {'lr': 0.000371197634722243, 'samples': 3507456, 'steps': 18267, 'loss/train': 1.1401678919792175}
+01/27/2022 13:03:00 - INFO - codeparrot_training - Step 18268: {'lr': 0.0003711833233581134, 'samples': 3507648, 'steps': 18268, 'loss/train': 0.2954166457056999}
+01/27/2022 13:03:04 - INFO - codeparrot_training - Step 18269: {'lr': 0.000371169011474873, 'samples': 3507840, 'steps': 18269, 'loss/train': 0.8411284983158112}
+01/27/2022 13:03:07 - INFO - codeparrot_training - Step 18270: {'lr': 0.00037115469907258303, 'samples': 3508032, 'steps': 18270, 'loss/train': 0.36761871725320816}
+01/27/2022 13:03:11 - INFO - codeparrot_training - Step 18271: {'lr': 0.0003711403861513047, 'samples': 3508224, 'steps': 18271, 'loss/train': 0.4267890751361847}
+01/27/2022 13:03:14 - INFO - codeparrot_training - Step 18272: {'lr': 0.0003711260727110995, 'samples': 3508416, 'steps': 18272, 'loss/train': 1.019156813621521}
+01/27/2022 13:03:17 - INFO - codeparrot_training - Step 18273: {'lr': 0.00037111175875202863, 'samples': 3508608, 'steps': 18273, 'loss/train': 0.8983352780342102}
+01/27/2022 13:03:20 - INFO - codeparrot_training - Step 18274: {'lr': 0.00037109744427415346, 'samples': 3508800, 'steps': 18274, 'loss/train': 0.7388879656791687}
+01/27/2022 13:03:23 - INFO - codeparrot_training - Step 18275: {'lr': 0.0003710831292775353, 'samples': 3508992, 'steps': 18275, 'loss/train': 2.023869574069977}
+01/27/2022 13:03:26 - INFO - codeparrot_training - Step 18276: {'lr': 0.00037106881376223544, 'samples': 3509184, 'steps': 18276, 'loss/train': 1.0476373136043549}
+01/27/2022 13:03:29 - INFO - codeparrot_training - Step 18277: {'lr': 0.00037105449772831527, 'samples': 3509376, 'steps': 18277, 'loss/train': 1.3748307824134827}
+01/27/2022 13:03:35 - INFO - codeparrot_training - Step 18278: {'lr': 0.00037104018117583605, 'samples': 3509568, 'steps': 18278, 'loss/train': 0.7826602756977081}
+01/27/2022 13:03:39 - INFO - codeparrot_training - Step 18279: {'lr': 0.00037102586410485915, 'samples': 3509760, 'steps': 18279, 'loss/train': 0.4434960186481476}
+01/27/2022 13:03:42 - INFO - codeparrot_training - Step 18280: {'lr': 0.00037101154651544583, 'samples': 3509952, 'steps': 18280, 'loss/train': 0.8129192590713501}
+01/27/2022 13:03:45 - INFO - codeparrot_training - Step 18281: {'lr': 0.0003709972284076575, 'samples': 3510144, 'steps': 18281, 'loss/train': 0.8168186545372009}
+01/27/2022 13:03:48 - INFO - codeparrot_training - Step 18282: {'lr': 0.0003709829097815555, 'samples': 3510336, 'steps': 18282, 'loss/train': 1.2640314996242523}
+01/27/2022 13:03:51 - INFO - codeparrot_training - Step 18283: {'lr': 0.00037096859063720104, 'samples': 3510528, 'steps': 18283, 'loss/train': 0.7549155950546265}
+01/27/2022 13:03:55 - INFO - codeparrot_training - Step 18284: {'lr': 0.00037095427097465564, 'samples': 3510720, 'steps': 18284, 'loss/train': 0.3178311288356781}
+01/27/2022 13:03:58 - INFO - codeparrot_training - Step 18285: {'lr': 0.0003709399507939805, 'samples': 3510912, 'steps': 18285, 'loss/train': 0.7982871830463409}
+01/27/2022 13:04:01 - INFO - codeparrot_training - Step 18286: {'lr': 0.00037092563009523703, 'samples': 3511104, 'steps': 18286, 'loss/train': 1.0399682521820068}
+01/27/2022 13:04:05 - INFO - codeparrot_training - Step 18287: {'lr': 0.0003709113088784865, 'samples': 3511296, 'steps': 18287, 'loss/train': 0.559084564447403}
+01/27/2022 13:04:08 - INFO - codeparrot_training - Step 18288: {'lr': 0.0003708969871437904, 'samples': 3511488, 'steps': 18288, 'loss/train': 0.5825687348842621}
+01/27/2022 13:04:12 - INFO - codeparrot_training - Step 18289: {'lr': 0.00037088266489120996, 'samples': 3511680, 'steps': 18289, 'loss/train': 0.7663698792457581}
+01/27/2022 13:04:15 - INFO - codeparrot_training - Step 18290: {'lr': 0.0003708683421208066, 'samples': 3511872, 'steps': 18290, 'loss/train': 1.1842447221279144}
+01/27/2022 13:04:18 - INFO - codeparrot_training - Step 18291: {'lr': 0.0003708540188326416, 'samples': 3512064, 'steps': 18291, 'loss/train': 0.2608780637383461}
+01/27/2022 13:04:21 - INFO - codeparrot_training - Step 18292: {'lr': 0.0003708396950267764, 'samples': 3512256, 'steps': 18292, 'loss/train': 0.24833549559116364}
+01/27/2022 13:04:24 - INFO - codeparrot_training - Step 18293: {'lr': 0.00037082537070327225, 'samples': 3512448, 'steps': 18293, 'loss/train': 0.5578436851501465}
+01/27/2022 13:04:27 - INFO - codeparrot_training - Step 18294: {'lr': 0.0003708110458621906, 'samples': 3512640, 'steps': 18294, 'loss/train': 0.6666761934757233}
+01/27/2022 13:04:30 - INFO - codeparrot_training - Step 18295: {'lr': 0.00037079672050359283, 'samples': 3512832, 'steps': 18295, 'loss/train': 0.8401064872741699}
+01/27/2022 13:08:57 - INFO - codeparrot_training - Step 18296: {'lr': 0.00037078239462754023, 'samples': 3513024, 'steps': 18296, 'loss/train': 1.0593761801719666}
+01/27/2022 13:09:00 - INFO - codeparrot_training - Step 18297: {'lr': 0.00037076806823409426, 'samples': 3513216, 'steps': 18297, 'loss/train': 0.8871398270130157}
+01/27/2022 13:09:04 - INFO - codeparrot_training - Step 18298: {'lr': 0.00037075374132331613, 'samples': 3513408, 'steps': 18298, 'loss/train': 0.8174306452274323}
+01/27/2022 13:09:07 - INFO - codeparrot_training - Step 18299: {'lr': 0.0003707394138952674, 'samples': 3513600, 'steps': 18299, 'loss/train': 0.6703898459672928}
+01/27/2022 13:09:10 - INFO - codeparrot_training - Step 18300: {'lr': 0.0003707250859500093, 'samples': 3513792, 'steps': 18300, 'loss/train': 0.6160449385643005}
+01/27/2022 13:09:13 - INFO - codeparrot_training - Step 18301: {'lr': 0.00037071075748760336, 'samples': 3513984, 'steps': 18301, 'loss/train': 0.622194305062294}
+01/27/2022 13:09:16 - INFO - codeparrot_training - Step 18302: {'lr': 0.0003706964285081108, 'samples': 3514176, 'steps': 18302, 'loss/train': 0.5423837006092072}
+01/27/2022 13:09:19 - INFO - codeparrot_training - Step 18303: {'lr': 0.0003706820990115931, 'samples': 3514368, 'steps': 18303, 'loss/train': 0.7538089156150818}
+01/27/2022 13:09:24 - INFO - codeparrot_training - Step 18304: {'lr': 0.00037066776899811153, 'samples': 3514560, 'steps': 18304, 'loss/train': 0.8453518152236938}
+01/27/2022 13:09:27 - INFO - codeparrot_training - Step 18305: {'lr': 0.00037065343846772765, 'samples': 3514752, 'steps': 18305, 'loss/train': 0.6001106053590775}
+01/27/2022 13:09:30 - INFO - codeparrot_training - Step 18306: {'lr': 0.0003706391074205027, 'samples': 3514944, 'steps': 18306, 'loss/train': 0.6567325741052628}
+01/27/2022 13:09:33 - INFO - codeparrot_training - Step 18307: {'lr': 0.00037062477585649814, 'samples': 3515136, 'steps': 18307, 'loss/train': 1.137632042169571}
+01/27/2022 13:09:36 - INFO - codeparrot_training - Step 18308: {'lr': 0.00037061044377577535, 'samples': 3515328, 'steps': 18308, 'loss/train': 0.38940632343292236}
+01/27/2022 13:09:40 - INFO - codeparrot_training - Step 18309: {'lr': 0.00037059611117839565, 'samples': 3515520, 'steps': 18309, 'loss/train': 0.8074138462543488}
+01/27/2022 13:09:43 - INFO - codeparrot_training - Step 18310: {'lr': 0.0003705817780644206, 'samples': 3515712, 'steps': 18310, 'loss/train': 0.8842051327228546}
+01/27/2022 13:09:46 - INFO - codeparrot_training - Step 18311: {'lr': 0.0003705674444339114, 'samples': 3515904, 'steps': 18311, 'loss/train': 0.9325694739818573}
+01/27/2022 13:09:49 - INFO - codeparrot_training - Step 18312: {'lr': 0.0003705531102869297, 'samples': 3516096, 'steps': 18312, 'loss/train': 0.4729570895433426}
+01/27/2022 13:09:53 - INFO - codeparrot_training - Step 18313: {'lr': 0.0003705387756235366, 'samples': 3516288, 'steps': 18313, 'loss/train': 0.929900050163269}
+01/27/2022 13:09:56 - INFO - codeparrot_training - Step 18314: {'lr': 0.00037052444044379375, 'samples': 3516480, 'steps': 18314, 'loss/train': 5.5131672620773315}
+01/27/2022 13:10:00 - INFO - codeparrot_training - Step 18315: {'lr': 0.00037051010474776244, 'samples': 3516672, 'steps': 18315, 'loss/train': 0.8734511733055115}
+01/27/2022 13:10:03 - INFO - codeparrot_training - Step 18316: {'lr': 0.0003704957685355041, 'samples': 3516864, 'steps': 18316, 'loss/train': 0.46326836943626404}
+01/27/2022 13:10:06 - INFO - codeparrot_training - Step 18317: {'lr': 0.00037048143180708014, 'samples': 3517056, 'steps': 18317, 'loss/train': 0.7724949717521667}
+01/27/2022 13:10:09 - INFO - codeparrot_training - Step 18318: {'lr': 0.000370467094562552, 'samples': 3517248, 'steps': 18318, 'loss/train': 1.0280378758907318}
+01/27/2022 13:10:12 - INFO - codeparrot_training - Step 18319: {'lr': 0.0003704527568019811, 'samples': 3517440, 'steps': 18319, 'loss/train': 0.6130871772766113}
+01/27/2022 13:10:16 - INFO - codeparrot_training - Step 18320: {'lr': 0.0003704384185254288, 'samples': 3517632, 'steps': 18320, 'loss/train': 0.7228032499551773}
+01/27/2022 13:10:19 - INFO - codeparrot_training - Step 18321: {'lr': 0.0003704240797329566, 'samples': 3517824, 'steps': 18321, 'loss/train': 1.3469575345516205}
+01/27/2022 13:10:23 - INFO - codeparrot_training - Step 18322: {'lr': 0.00037040974042462584, 'samples': 3518016, 'steps': 18322, 'loss/train': 0.05008471570909023}
+01/27/2022 13:10:26 - INFO - codeparrot_training - Step 18323: {'lr': 0.000370395400600498, 'samples': 3518208, 'steps': 18323, 'loss/train': 0.8208203315734863}
+01/27/2022 13:10:29 - INFO - codeparrot_training - Step 18324: {'lr': 0.00037038106026063457, 'samples': 3518400, 'steps': 18324, 'loss/train': 0.9637604355812073}
+01/27/2022 13:10:33 - INFO - codeparrot_training - Step 18325: {'lr': 0.0003703667194050968, 'samples': 3518592, 'steps': 18325, 'loss/train': 0.5678305774927139}
+01/27/2022 13:10:36 - INFO - codeparrot_training - Step 18326: {'lr': 0.0003703523780339463, 'samples': 3518784, 'steps': 18326, 'loss/train': 0.7692492306232452}
+01/27/2022 13:10:39 - INFO - codeparrot_training - Step 18327: {'lr': 0.0003703380361472444, 'samples': 3518976, 'steps': 18327, 'loss/train': 0.5446786880493164}
+01/27/2022 13:10:42 - INFO - codeparrot_training - Step 18328: {'lr': 0.00037032369374505255, 'samples': 3519168, 'steps': 18328, 'loss/train': 0.5501291155815125}
+01/27/2022 13:10:45 - INFO - codeparrot_training - Step 18329: {'lr': 0.0003703093508274322, 'samples': 3519360, 'steps': 18329, 'loss/train': 0.44551214575767517}
+01/27/2022 13:10:48 - INFO - codeparrot_training - Step 18330: {'lr': 0.0003702950073944448, 'samples': 3519552, 'steps': 18330, 'loss/train': 1.008555293083191}
+01/27/2022 13:10:55 - INFO - codeparrot_training - Step 18331: {'lr': 0.00037028066344615176, 'samples': 3519744, 'steps': 18331, 'loss/train': 0.5455801337957382}
+01/27/2022 13:10:58 - INFO - codeparrot_training - Step 18332: {'lr': 0.0003702663189826146, 'samples': 3519936, 'steps': 18332, 'loss/train': 1.3110808432102203}
+01/27/2022 13:11:01 - INFO - codeparrot_training - Step 18333: {'lr': 0.00037025197400389467, 'samples': 3520128, 'steps': 18333, 'loss/train': 0.8105674982070923}
+01/27/2022 13:11:04 - INFO - codeparrot_training - Step 18334: {'lr': 0.0003702376285100535, 'samples': 3520320, 'steps': 18334, 'loss/train': 0.9312779903411865}
+01/27/2022 13:11:07 - INFO - codeparrot_training - Step 18335: {'lr': 0.00037022328250115244, 'samples': 3520512, 'steps': 18335, 'loss/train': 0.865735799074173}
+01/27/2022 13:11:10 - INFO - codeparrot_training - Step 18336: {'lr': 0.00037020893597725313, 'samples': 3520704, 'steps': 18336, 'loss/train': 0.6463030278682709}
+01/27/2022 13:11:13 - INFO - codeparrot_training - Step 18337: {'lr': 0.0003701945889384168, 'samples': 3520896, 'steps': 18337, 'loss/train': 0.8258965909481049}
+01/27/2022 13:11:17 - INFO - codeparrot_training - Step 18338: {'lr': 0.00037018024138470515, 'samples': 3521088, 'steps': 18338, 'loss/train': 0.9253689050674438}
+01/27/2022 13:11:20 - INFO - codeparrot_training - Step 18339: {'lr': 0.0003701658933161794, 'samples': 3521280, 'steps': 18339, 'loss/train': 0.5940666049718857}
+01/27/2022 13:11:24 - INFO - codeparrot_training - Step 18340: {'lr': 0.00037015154473290113, 'samples': 3521472, 'steps': 18340, 'loss/train': 1.1477303802967072}
+01/27/2022 13:11:27 - INFO - codeparrot_training - Step 18341: {'lr': 0.0003701371956349318, 'samples': 3521664, 'steps': 18341, 'loss/train': 0.6687160581350327}
+01/27/2022 13:11:30 - INFO - codeparrot_training - Step 18342: {'lr': 0.00037012284602233294, 'samples': 3521856, 'steps': 18342, 'loss/train': 0.8382427096366882}
+01/27/2022 13:11:34 - INFO - codeparrot_training - Step 18343: {'lr': 0.0003701084958951659, 'samples': 3522048, 'steps': 18343, 'loss/train': 0.6580657660961151}
+01/27/2022 13:11:37 - INFO - codeparrot_training - Step 18344: {'lr': 0.0003700941452534922, 'samples': 3522240, 'steps': 18344, 'loss/train': 0.5793897360563278}
+01/27/2022 13:11:40 - INFO - codeparrot_training - Step 18345: {'lr': 0.00037007979409737324, 'samples': 3522432, 'steps': 18345, 'loss/train': 1.1894077956676483}
+01/27/2022 13:11:43 - INFO - codeparrot_training - Step 18346: {'lr': 0.0003700654424268707, 'samples': 3522624, 'steps': 18346, 'loss/train': 0.8130575716495514}
+01/27/2022 13:11:46 - INFO - codeparrot_training - Step 18347: {'lr': 0.00037005109024204586, 'samples': 3522816, 'steps': 18347, 'loss/train': 0.9698634445667267}
+01/27/2022 13:11:49 - INFO - codeparrot_training - Step 18348: {'lr': 0.00037003673754296026, 'samples': 3523008, 'steps': 18348, 'loss/train': 0.551543265581131}
+01/27/2022 13:11:54 - INFO - codeparrot_training - Step 18349: {'lr': 0.00037002238432967547, 'samples': 3523200, 'steps': 18349, 'loss/train': 0.5922019332647324}
+01/27/2022 13:11:57 - INFO - codeparrot_training - Step 18350: {'lr': 0.0003700080306022528, 'samples': 3523392, 'steps': 18350, 'loss/train': 1.0134433507919312}
+01/27/2022 13:12:00 - INFO - codeparrot_training - Step 18351: {'lr': 0.00036999367636075386, 'samples': 3523584, 'steps': 18351, 'loss/train': 0.8398018777370453}
+01/27/2022 13:12:03 - INFO - codeparrot_training - Step 18352: {'lr': 0.00036997932160524015, 'samples': 3523776, 'steps': 18352, 'loss/train': 0.8006094396114349}
+01/27/2022 13:12:07 - INFO - codeparrot_training - Step 18353: {'lr': 0.00036996496633577314, 'samples': 3523968, 'steps': 18353, 'loss/train': 0.9544440507888794}
+01/27/2022 13:12:10 - INFO - codeparrot_training - Step 18354: {'lr': 0.00036995061055241426, 'samples': 3524160, 'steps': 18354, 'loss/train': 0.46654120087623596}
+01/27/2022 13:12:13 - INFO - codeparrot_training - Step 18355: {'lr': 0.000369936254255225, 'samples': 3524352, 'steps': 18355, 'loss/train': 0.9321468472480774}
+01/27/2022 13:12:16 - INFO - codeparrot_training - Step 18356: {'lr': 0.000369921897444267, 'samples': 3524544, 'steps': 18356, 'loss/train': 0.9694753289222717}
+01/27/2022 13:12:22 - INFO - codeparrot_training - Step 18357: {'lr': 0.00036990754011960165, 'samples': 3524736, 'steps': 18357, 'loss/train': 0.9052877426147461}
+01/27/2022 13:12:25 - INFO - codeparrot_training - Step 18358: {'lr': 0.0003698931822812905, 'samples': 3524928, 'steps': 18358, 'loss/train': 0.9680974781513214}
+01/27/2022 13:12:28 - INFO - codeparrot_training - Step 18359: {'lr': 0.000369878823929395, 'samples': 3525120, 'steps': 18359, 'loss/train': 1.0036095678806305}
+01/27/2022 13:12:32 - INFO - codeparrot_training - Step 18360: {'lr': 0.00036986446506397666, 'samples': 3525312, 'steps': 18360, 'loss/train': 0.8348698318004608}
+01/27/2022 13:12:35 - INFO - codeparrot_training - Step 18361: {'lr': 0.00036985010568509703, 'samples': 3525504, 'steps': 18361, 'loss/train': 1.1063762605190277}
+01/27/2022 13:12:38 - INFO - codeparrot_training - Step 18362: {'lr': 0.00036983574579281764, 'samples': 3525696, 'steps': 18362, 'loss/train': 0.9636756777763367}
+01/27/2022 13:12:41 - INFO - codeparrot_training - Step 18363: {'lr': 0.0003698213853871999, 'samples': 3525888, 'steps': 18363, 'loss/train': 0.7722897827625275}
+01/27/2022 13:12:44 - INFO - codeparrot_training - Step 18364: {'lr': 0.00036980702446830547, 'samples': 3526080, 'steps': 18364, 'loss/train': 0.31749077886343}
+01/27/2022 13:12:47 - INFO - codeparrot_training - Step 18365: {'lr': 0.0003697926630361957, 'samples': 3526272, 'steps': 18365, 'loss/train': 0.5657761842012405}
+01/27/2022 13:12:52 - INFO - codeparrot_training - Step 18366: {'lr': 0.00036977830109093227, 'samples': 3526464, 'steps': 18366, 'loss/train': 0.2943011149764061}
+01/27/2022 13:12:55 - INFO - codeparrot_training - Step 18367: {'lr': 0.0003697639386325766, 'samples': 3526656, 'steps': 18367, 'loss/train': 1.207328975200653}
+01/27/2022 13:12:58 - INFO - codeparrot_training - Step 18368: {'lr': 0.00036974957566119027, 'samples': 3526848, 'steps': 18368, 'loss/train': 0.7406604588031769}
+01/27/2022 13:13:01 - INFO - codeparrot_training - Step 18369: {'lr': 0.00036973521217683475, 'samples': 3527040, 'steps': 18369, 'loss/train': 2.0340165495872498}
+01/27/2022 13:13:04 - INFO - codeparrot_training - Step 18370: {'lr': 0.00036972084817957164, 'samples': 3527232, 'steps': 18370, 'loss/train': 1.1193589568138123}
+01/27/2022 13:13:08 - INFO - codeparrot_training - Step 18371: {'lr': 0.0003697064836694624, 'samples': 3527424, 'steps': 18371, 'loss/train': 0.8202087879180908}
+01/27/2022 13:13:11 - INFO - codeparrot_training - Step 18372: {'lr': 0.0003696921186465686, 'samples': 3527616, 'steps': 18372, 'loss/train': 0.7900904417037964}
+01/27/2022 13:13:14 - INFO - codeparrot_training - Step 18373: {'lr': 0.00036967775311095186, 'samples': 3527808, 'steps': 18373, 'loss/train': 0.39355936646461487}
+01/27/2022 13:13:17 - INFO - codeparrot_training - Step 18374: {'lr': 0.00036966338706267347, 'samples': 3528000, 'steps': 18374, 'loss/train': 0.9332838356494904}
+01/27/2022 13:13:22 - INFO - codeparrot_training - Step 18375: {'lr': 0.0003696490205017953, 'samples': 3528192, 'steps': 18375, 'loss/train': 0.8089993894100189}
+01/27/2022 13:13:25 - INFO - codeparrot_training - Step 18376: {'lr': 0.00036963465342837855, 'samples': 3528384, 'steps': 18376, 'loss/train': 0.9320121109485626}
+01/27/2022 13:13:28 - INFO - codeparrot_training - Step 18377: {'lr': 0.000369620285842485, 'samples': 3528576, 'steps': 18377, 'loss/train': 1.5324345231056213}
+01/27/2022 13:13:31 - INFO - codeparrot_training - Step 18378: {'lr': 0.00036960591774417613, 'samples': 3528768, 'steps': 18378, 'loss/train': 0.9442209899425507}
+01/27/2022 13:13:34 - INFO - codeparrot_training - Step 18379: {'lr': 0.00036959154913351357, 'samples': 3528960, 'steps': 18379, 'loss/train': 0.5088567137718201}
+01/27/2022 13:13:37 - INFO - codeparrot_training - Step 18380: {'lr': 0.0003695771800105586, 'samples': 3529152, 'steps': 18380, 'loss/train': 0.68269282579422}
+01/27/2022 13:13:40 - INFO - codeparrot_training - Step 18381: {'lr': 0.00036956281037537307, 'samples': 3529344, 'steps': 18381, 'loss/train': 0.5243499130010605}
+01/27/2022 13:13:43 - INFO - codeparrot_training - Step 18382: {'lr': 0.00036954844022801846, 'samples': 3529536, 'steps': 18382, 'loss/train': 0.9105502367019653}
+01/27/2022 13:13:50 - INFO - codeparrot_training - Step 18383: {'lr': 0.00036953406956855624, 'samples': 3529728, 'steps': 18383, 'loss/train': 1.1394043564796448}
+01/27/2022 13:13:53 - INFO - codeparrot_training - Step 18384: {'lr': 0.0003695196983970481, 'samples': 3529920, 'steps': 18384, 'loss/train': 0.4730304479598999}
+01/27/2022 13:13:56 - INFO - codeparrot_training - Step 18385: {'lr': 0.0003695053267135554, 'samples': 3530112, 'steps': 18385, 'loss/train': 0.7863381206989288}
+01/27/2022 13:14:00 - INFO - codeparrot_training - Step 18386: {'lr': 0.00036949095451813997, 'samples': 3530304, 'steps': 18386, 'loss/train': 0.6238634437322617}
+01/27/2022 13:14:03 - INFO - codeparrot_training - Step 18387: {'lr': 0.0003694765818108631, 'samples': 3530496, 'steps': 18387, 'loss/train': 0.4735439121723175}
+01/27/2022 13:14:06 - INFO - codeparrot_training - Step 18388: {'lr': 0.00036946220859178656, 'samples': 3530688, 'steps': 18388, 'loss/train': 0.946388304233551}
+01/27/2022 13:14:09 - INFO - codeparrot_training - Step 18389: {'lr': 0.0003694478348609718, 'samples': 3530880, 'steps': 18389, 'loss/train': 0.7549307942390442}
+01/27/2022 13:14:12 - INFO - codeparrot_training - Step 18390: {'lr': 0.00036943346061848054, 'samples': 3531072, 'steps': 18390, 'loss/train': 0.6156394332647324}
+01/27/2022 13:14:15 - INFO - codeparrot_training - Step 18391: {'lr': 0.00036941908586437416, 'samples': 3531264, 'steps': 18391, 'loss/train': 0.6667010933160782}
+01/27/2022 13:14:20 - INFO - codeparrot_training - Step 18392: {'lr': 0.0003694047105987144, 'samples': 3531456, 'steps': 18392, 'loss/train': 0.555965781211853}
+01/27/2022 13:14:23 - INFO - codeparrot_training - Step 18393: {'lr': 0.00036939033482156277, 'samples': 3531648, 'steps': 18393, 'loss/train': 0.949165016412735}
+01/27/2022 13:14:26 - INFO - codeparrot_training - Step 18394: {'lr': 0.00036937595853298076, 'samples': 3531840, 'steps': 18394, 'loss/train': 1.2641754448413849}
+01/27/2022 13:14:29 - INFO - codeparrot_training - Step 18395: {'lr': 0.0003693615817330302, 'samples': 3532032, 'steps': 18395, 'loss/train': 0.7161440849304199}
+01/27/2022 13:14:32 - INFO - codeparrot_training - Step 18396: {'lr': 0.00036934720442177244, 'samples': 3532224, 'steps': 18396, 'loss/train': 1.1577343940734863}
+01/27/2022 13:14:35 - INFO - codeparrot_training - Step 18397: {'lr': 0.0003693328265992692, 'samples': 3532416, 'steps': 18397, 'loss/train': 0.6496208310127258}
+01/27/2022 13:14:39 - INFO - codeparrot_training - Step 18398: {'lr': 0.000369318448265582, 'samples': 3532608, 'steps': 18398, 'loss/train': 0.8623510301113129}
+01/27/2022 13:14:42 - INFO - codeparrot_training - Step 18399: {'lr': 0.00036930406942077245, 'samples': 3532800, 'steps': 18399, 'loss/train': 0.7060639411211014}
+01/27/2022 13:14:45 - INFO - codeparrot_training - Step 18400: {'lr': 0.0003692896900649021, 'samples': 3532992, 'steps': 18400, 'loss/train': 0.5915030390024185}
+01/27/2022 13:15:01 - INFO - codeparrot_training - Step 18401: {'lr': 0.0003692753101980327, 'samples': 3533184, 'steps': 18401, 'loss/train': 0.8867840766906738}
+01/27/2022 13:15:04 - INFO - codeparrot_training - Step 18402: {'lr': 0.00036926092982022564, 'samples': 3533376, 'steps': 18402, 'loss/train': 0.9333857595920563}
+01/27/2022 13:15:07 - INFO - codeparrot_training - Step 18403: {'lr': 0.0003692465489315427, 'samples': 3533568, 'steps': 18403, 'loss/train': 0.8960094451904297}
+01/27/2022 13:15:11 - INFO - codeparrot_training - Step 18404: {'lr': 0.00036923216753204536, 'samples': 3533760, 'steps': 18404, 'loss/train': 0.15427134186029434}
+01/27/2022 13:15:14 - INFO - codeparrot_training - Step 18405: {'lr': 0.0003692177856217953, 'samples': 3533952, 'steps': 18405, 'loss/train': 0.5520371496677399}
+01/27/2022 13:15:17 - INFO - codeparrot_training - Step 18406: {'lr': 0.00036920340320085413, 'samples': 3534144, 'steps': 18406, 'loss/train': 0.9948255121707916}
+01/27/2022 13:15:20 - INFO - codeparrot_training - Step 18407: {'lr': 0.00036918902026928334, 'samples': 3534336, 'steps': 18407, 'loss/train': 1.038514405488968}
+01/27/2022 13:15:23 - INFO - codeparrot_training - Step 18408: {'lr': 0.00036917463682714473, 'samples': 3534528, 'steps': 18408, 'loss/train': 0.7705506384372711}
+01/27/2022 13:15:26 - INFO - codeparrot_training - Step 18409: {'lr': 0.00036916025287449976, 'samples': 3534720, 'steps': 18409, 'loss/train': 0.6720096319913864}
+01/27/2022 13:15:31 - INFO - codeparrot_training - Step 18410: {'lr': 0.0003691458684114102, 'samples': 3534912, 'steps': 18410, 'loss/train': 0.8152487576007843}
+01/27/2022 13:15:34 - INFO - codeparrot_training - Step 18411: {'lr': 0.00036913148343793744, 'samples': 3535104, 'steps': 18411, 'loss/train': 0.3751904368400574}
+01/27/2022 13:15:37 - INFO - codeparrot_training - Step 18412: {'lr': 0.00036911709795414336, 'samples': 3535296, 'steps': 18412, 'loss/train': 0.49309922754764557}
+01/27/2022 13:15:40 - INFO - codeparrot_training - Step 18413: {'lr': 0.00036910271196008936, 'samples': 3535488, 'steps': 18413, 'loss/train': 0.5364853888750076}
+01/27/2022 13:15:43 - INFO - codeparrot_training - Step 18414: {'lr': 0.0003690883254558372, 'samples': 3535680, 'steps': 18414, 'loss/train': 0.4064796566963196}
+01/27/2022 13:15:46 - INFO - codeparrot_training - Step 18415: {'lr': 0.0003690739384414485, 'samples': 3535872, 'steps': 18415, 'loss/train': 1.0221579372882843}
+01/27/2022 13:15:50 - INFO - codeparrot_training - Step 18416: {'lr': 0.0003690595509169848, 'samples': 3536064, 'steps': 18416, 'loss/train': 1.2742799520492554}
+01/27/2022 13:15:53 - INFO - codeparrot_training - Step 18417: {'lr': 0.00036904516288250786, 'samples': 3536256, 'steps': 18417, 'loss/train': 0.8791683912277222}
+01/27/2022 13:15:56 - INFO - codeparrot_training - Step 18418: {'lr': 0.0003690307743380791, 'samples': 3536448, 'steps': 18418, 'loss/train': 0.7053591459989548}
+01/27/2022 13:16:00 - INFO - codeparrot_training - Step 18419: {'lr': 0.00036901638528376047, 'samples': 3536640, 'steps': 18419, 'loss/train': 0.7030465006828308}
+01/27/2022 13:16:03 - INFO - codeparrot_training - Step 18420: {'lr': 0.00036900199571961336, 'samples': 3536832, 'steps': 18420, 'loss/train': 0.2899126186966896}
+01/27/2022 13:16:06 - INFO - codeparrot_training - Step 18421: {'lr': 0.0003689876056456995, 'samples': 3537024, 'steps': 18421, 'loss/train': 1.067914456129074}
+01/27/2022 13:16:10 - INFO - codeparrot_training - Step 18422: {'lr': 0.0003689732150620805, 'samples': 3537216, 'steps': 18422, 'loss/train': 0.8250411450862885}
+01/27/2022 13:16:13 - INFO - codeparrot_training - Step 18423: {'lr': 0.00036895882396881805, 'samples': 3537408, 'steps': 18423, 'loss/train': 0.9623566567897797}
+01/27/2022 13:16:16 - INFO - codeparrot_training - Step 18424: {'lr': 0.0003689444323659737, 'samples': 3537600, 'steps': 18424, 'loss/train': 0.8414361476898193}
+01/27/2022 13:16:19 - INFO - codeparrot_training - Step 18425: {'lr': 0.00036893004025360926, 'samples': 3537792, 'steps': 18425, 'loss/train': 0.8942971229553223}
+01/27/2022 13:16:22 - INFO - codeparrot_training - Step 18426: {'lr': 0.0003689156476317862, 'samples': 3537984, 'steps': 18426, 'loss/train': 0.10691391304135323}
+01/27/2022 13:16:27 - INFO - codeparrot_training - Step 18427: {'lr': 0.0003689012545005664, 'samples': 3538176, 'steps': 18427, 'loss/train': 0.19674481451511383}
+01/27/2022 13:16:30 - INFO - codeparrot_training - Step 18428: {'lr': 0.0003688868608600113, 'samples': 3538368, 'steps': 18428, 'loss/train': 1.3016538619995117}
+01/27/2022 13:16:33 - INFO - codeparrot_training - Step 18429: {'lr': 0.0003688724667101826, 'samples': 3538560, 'steps': 18429, 'loss/train': 1.0530186295509338}
+01/27/2022 13:16:36 - INFO - codeparrot_training - Step 18430: {'lr': 0.0003688580720511421, 'samples': 3538752, 'steps': 18430, 'loss/train': 1.062100499868393}
+01/27/2022 13:16:39 - INFO - codeparrot_training - Step 18431: {'lr': 0.0003688436768829512, 'samples': 3538944, 'steps': 18431, 'loss/train': 0.25318510830402374}
+01/27/2022 13:16:42 - INFO - codeparrot_training - Step 18432: {'lr': 0.0003688292812056719, 'samples': 3539136, 'steps': 18432, 'loss/train': 0.714407354593277}
+01/27/2022 13:16:46 - INFO - codeparrot_training - Step 18433: {'lr': 0.00036881488501936554, 'samples': 3539328, 'steps': 18433, 'loss/train': 0.6477169543504715}
+01/27/2022 13:16:49 - INFO - codeparrot_training - Step 18434: {'lr': 0.00036880048832409407, 'samples': 3539520, 'steps': 18434, 'loss/train': 1.0486215949058533}
+01/27/2022 13:16:52 - INFO - codeparrot_training - Step 18435: {'lr': 0.000368786091119919, 'samples': 3539712, 'steps': 18435, 'loss/train': 0.04033800680190325}
+01/27/2022 13:16:57 - INFO - codeparrot_training - Step 18436: {'lr': 0.00036877169340690204, 'samples': 3539904, 'steps': 18436, 'loss/train': 0.5420503467321396}
+01/27/2022 13:17:01 - INFO - codeparrot_training - Step 18437: {'lr': 0.0003687572951851048, 'samples': 3540096, 'steps': 18437, 'loss/train': 1.2846967577934265}
+01/27/2022 13:17:04 - INFO - codeparrot_training - Step 18438: {'lr': 0.0003687428964545891, 'samples': 3540288, 'steps': 18438, 'loss/train': 0.5260314792394638}
+01/27/2022 13:17:07 - INFO - codeparrot_training - Step 18439: {'lr': 0.00036872849721541643, 'samples': 3540480, 'steps': 18439, 'loss/train': 3.7471060752868652}
+01/27/2022 13:17:10 - INFO - codeparrot_training - Step 18440: {'lr': 0.0003687140974676486, 'samples': 3540672, 'steps': 18440, 'loss/train': 0.8952222168445587}
+01/27/2022 13:17:13 - INFO - codeparrot_training - Step 18441: {'lr': 0.00036869969721134736, 'samples': 3540864, 'steps': 18441, 'loss/train': 0.8834848701953888}
+01/27/2022 13:17:16 - INFO - codeparrot_training - Step 18442: {'lr': 0.0003686852964465742, 'samples': 3541056, 'steps': 18442, 'loss/train': 0.8104639649391174}
+01/27/2022 13:17:19 - INFO - codeparrot_training - Step 18443: {'lr': 0.000368670895173391, 'samples': 3541248, 'steps': 18443, 'loss/train': 0.7299737334251404}
+01/27/2022 13:17:23 - INFO - codeparrot_training - Step 18444: {'lr': 0.00036865649339185935, 'samples': 3541440, 'steps': 18444, 'loss/train': 0.8304727971553802}
+01/27/2022 13:17:27 - INFO - codeparrot_training - Step 18445: {'lr': 0.000368642091102041, 'samples': 3541632, 'steps': 18445, 'loss/train': 0.8680945336818695}
+01/27/2022 13:17:30 - INFO - codeparrot_training - Step 18446: {'lr': 0.0003686276883039975, 'samples': 3541824, 'steps': 18446, 'loss/train': 0.8274680078029633}
+01/27/2022 13:17:33 - INFO - codeparrot_training - Step 18447: {'lr': 0.0003686132849977908, 'samples': 3542016, 'steps': 18447, 'loss/train': 1.2690901458263397}
+01/27/2022 13:17:36 - INFO - codeparrot_training - Step 18448: {'lr': 0.0003685988811834823, 'samples': 3542208, 'steps': 18448, 'loss/train': 1.0444246530532837}
+01/27/2022 13:17:40 - INFO - codeparrot_training - Step 18449: {'lr': 0.00036858447686113395, 'samples': 3542400, 'steps': 18449, 'loss/train': 0.48551349341869354}
+01/27/2022 13:17:43 - INFO - codeparrot_training - Step 18450: {'lr': 0.0003685700720308073, 'samples': 3542592, 'steps': 18450, 'loss/train': 1.148313581943512}
+01/27/2022 13:17:46 - INFO - codeparrot_training - Step 18451: {'lr': 0.0003685556666925641, 'samples': 3542784, 'steps': 18451, 'loss/train': 0.43067827820777893}
+01/27/2022 13:17:49 - INFO - codeparrot_training - Step 18452: {'lr': 0.0003685412608464661, 'samples': 3542976, 'steps': 18452, 'loss/train': 0.9486543238162994}
+01/27/2022 13:17:52 - INFO - codeparrot_training - Step 18453: {'lr': 0.00036852685449257505, 'samples': 3543168, 'steps': 18453, 'loss/train': 1.1631980538368225}
+01/27/2022 13:17:57 - INFO - codeparrot_training - Step 18454: {'lr': 0.00036851244763095247, 'samples': 3543360, 'steps': 18454, 'loss/train': 1.2007796466350555}
+01/27/2022 13:18:00 - INFO - codeparrot_training - Step 18455: {'lr': 0.0003684980402616603, 'samples': 3543552, 'steps': 18455, 'loss/train': 0.4465082734823227}
+01/27/2022 13:18:03 - INFO - codeparrot_training - Step 18456: {'lr': 0.0003684836323847601, 'samples': 3543744, 'steps': 18456, 'loss/train': 0.4656393527984619}
+01/27/2022 13:18:06 - INFO - codeparrot_training - Step 18457: {'lr': 0.0003684692240003137, 'samples': 3543936, 'steps': 18457, 'loss/train': 1.0898792445659637}
+01/27/2022 13:18:09 - INFO - codeparrot_training - Step 18458: {'lr': 0.00036845481510838264, 'samples': 3544128, 'steps': 18458, 'loss/train': 0.5729221701622009}
+01/27/2022 13:18:12 - INFO - codeparrot_training - Step 18459: {'lr': 0.00036844040570902886, 'samples': 3544320, 'steps': 18459, 'loss/train': 0.5986530035734177}
+01/27/2022 13:18:15 - INFO - codeparrot_training - Step 18460: {'lr': 0.00036842599580231395, 'samples': 3544512, 'steps': 18460, 'loss/train': 0.6716265231370926}
+01/27/2022 13:18:19 - INFO - codeparrot_training - Step 18461: {'lr': 0.0003684115853882997, 'samples': 3544704, 'steps': 18461, 'loss/train': 0.7140794992446899}
+01/27/2022 13:18:25 - INFO - codeparrot_training - Step 18462: {'lr': 0.00036839717446704787, 'samples': 3544896, 'steps': 18462, 'loss/train': 0.9006646871566772}
+01/27/2022 13:18:28 - INFO - codeparrot_training - Step 18463: {'lr': 0.00036838276303862, 'samples': 3545088, 'steps': 18463, 'loss/train': 0.7033320218324661}
+01/27/2022 13:18:31 - INFO - codeparrot_training - Step 18464: {'lr': 0.00036836835110307803, 'samples': 3545280, 'steps': 18464, 'loss/train': 0.2690603658556938}
+01/27/2022 13:18:34 - INFO - codeparrot_training - Step 18465: {'lr': 0.0003683539386604837, 'samples': 3545472, 'steps': 18465, 'loss/train': 0.9282865226268768}
+01/27/2022 13:18:37 - INFO - codeparrot_training - Step 18466: {'lr': 0.00036833952571089856, 'samples': 3545664, 'steps': 18466, 'loss/train': 1.0644635260105133}
+01/27/2022 13:18:41 - INFO - codeparrot_training - Step 18467: {'lr': 0.0003683251122543846, 'samples': 3545856, 'steps': 18467, 'loss/train': 0.31714700907468796}
+01/27/2022 13:18:44 - INFO - codeparrot_training - Step 18468: {'lr': 0.0003683106982910033, 'samples': 3546048, 'steps': 18468, 'loss/train': 0.41816264390945435}
+01/27/2022 13:18:47 - INFO - codeparrot_training - Step 18469: {'lr': 0.0003682962838208166, 'samples': 3546240, 'steps': 18469, 'loss/train': 0.8327123522758484}
+01/27/2022 13:18:50 - INFO - codeparrot_training - Step 18470: {'lr': 0.0003682818688438862, 'samples': 3546432, 'steps': 18470, 'loss/train': 0.9555867612361908}
+01/27/2022 13:18:55 - INFO - codeparrot_training - Step 18471: {'lr': 0.00036826745336027383, 'samples': 3546624, 'steps': 18471, 'loss/train': 0.7848193645477295}
+01/27/2022 13:18:58 - INFO - codeparrot_training - Step 18472: {'lr': 0.0003682530373700412, 'samples': 3546816, 'steps': 18472, 'loss/train': 1.506401538848877}
+01/27/2022 13:19:01 - INFO - codeparrot_training - Step 18473: {'lr': 0.00036823862087325017, 'samples': 3547008, 'steps': 18473, 'loss/train': 0.494672566652298}
+01/27/2022 13:19:04 - INFO - codeparrot_training - Step 18474: {'lr': 0.00036822420386996237, 'samples': 3547200, 'steps': 18474, 'loss/train': 0.9348134994506836}
+01/27/2022 13:19:07 - INFO - codeparrot_training - Step 18475: {'lr': 0.0003682097863602397, 'samples': 3547392, 'steps': 18475, 'loss/train': 1.4449903070926666}
+01/27/2022 13:19:10 - INFO - codeparrot_training - Step 18476: {'lr': 0.00036819536834414374, 'samples': 3547584, 'steps': 18476, 'loss/train': 1.107170730829239}
+01/27/2022 13:19:13 - INFO - codeparrot_training - Step 18477: {'lr': 0.0003681809498217364, 'samples': 3547776, 'steps': 18477, 'loss/train': 0.47888150811195374}
+01/27/2022 13:19:17 - INFO - codeparrot_training - Step 18478: {'lr': 0.0003681665307930794, 'samples': 3547968, 'steps': 18478, 'loss/train': 0.9961900413036346}
+01/27/2022 13:19:20 - INFO - codeparrot_training - Step 18479: {'lr': 0.0003681521112582345, 'samples': 3548160, 'steps': 18479, 'loss/train': 0.8545778095722198}
+01/27/2022 13:19:26 - INFO - codeparrot_training - Step 18480: {'lr': 0.00036813769121726354, 'samples': 3548352, 'steps': 18480, 'loss/train': 1.0871217548847198}
+01/27/2022 13:19:29 - INFO - codeparrot_training - Step 18481: {'lr': 0.00036812327067022813, 'samples': 3548544, 'steps': 18481, 'loss/train': 1.4231591820716858}
+01/27/2022 13:19:33 - INFO - codeparrot_training - Step 18482: {'lr': 0.00036810884961719015, 'samples': 3548736, 'steps': 18482, 'loss/train': 0.8970577418804169}
+01/27/2022 13:19:36 - INFO - codeparrot_training - Step 18483: {'lr': 0.0003680944280582114, 'samples': 3548928, 'steps': 18483, 'loss/train': 0.7538943886756897}
+01/27/2022 13:19:39 - INFO - codeparrot_training - Step 18484: {'lr': 0.0003680800059933536, 'samples': 3549120, 'steps': 18484, 'loss/train': 0.3099707365036011}
+01/27/2022 13:19:42 - INFO - codeparrot_training - Step 18485: {'lr': 0.00036806558342267854, 'samples': 3549312, 'steps': 18485, 'loss/train': 0.5717348009347916}
+01/27/2022 13:19:45 - INFO - codeparrot_training - Step 18486: {'lr': 0.0003680511603462481, 'samples': 3549504, 'steps': 18486, 'loss/train': 1.7002184987068176}
+01/27/2022 13:19:48 - INFO - codeparrot_training - Step 18487: {'lr': 0.00036803673676412386, 'samples': 3549696, 'steps': 18487, 'loss/train': 0.905717521905899}
+01/27/2022 13:19:51 - INFO - codeparrot_training - Step 18488: {'lr': 0.00036802231267636773, 'samples': 3549888, 'steps': 18488, 'loss/train': 0.7133515477180481}
+01/27/2022 13:19:56 - INFO - codeparrot_training - Step 18489: {'lr': 0.0003680078880830415, 'samples': 3550080, 'steps': 18489, 'loss/train': 0.4598179757595062}
+01/27/2022 13:19:59 - INFO - codeparrot_training - Step 18490: {'lr': 0.000367993462984207, 'samples': 3550272, 'steps': 18490, 'loss/train': 0.8263865411281586}
+01/27/2022 13:20:02 - INFO - codeparrot_training - Step 18491: {'lr': 0.0003679790373799259, 'samples': 3550464, 'steps': 18491, 'loss/train': 0.7833853662014008}
+01/27/2022 13:20:05 - INFO - codeparrot_training - Step 18492: {'lr': 0.0003679646112702601, 'samples': 3550656, 'steps': 18492, 'loss/train': 0.8143737316131592}
+01/27/2022 13:20:09 - INFO - codeparrot_training - Step 18493: {'lr': 0.0003679501846552714, 'samples': 3550848, 'steps': 18493, 'loss/train': 0.46310846507549286}
+01/27/2022 13:20:12 - INFO - codeparrot_training - Step 18494: {'lr': 0.00036793575753502153, 'samples': 3551040, 'steps': 18494, 'loss/train': 0.9339428544044495}
+01/27/2022 13:20:15 - INFO - codeparrot_training - Step 18495: {'lr': 0.0003679213299095723, 'samples': 3551232, 'steps': 18495, 'loss/train': 0.9411144554615021}
+01/27/2022 13:20:18 - INFO - codeparrot_training - Step 18496: {'lr': 0.00036790690177898556, 'samples': 3551424, 'steps': 18496, 'loss/train': 1.0962677299976349}
+01/27/2022 13:20:21 - INFO - codeparrot_training - Step 18497: {'lr': 0.00036789247314332306, 'samples': 3551616, 'steps': 18497, 'loss/train': 0.6498252600431442}
+01/27/2022 13:20:26 - INFO - codeparrot_training - Step 18498: {'lr': 0.00036787804400264666, 'samples': 3551808, 'steps': 18498, 'loss/train': 1.3845148086547852}
+01/27/2022 13:20:29 - INFO - codeparrot_training - Step 18499: {'lr': 0.00036786361435701823, 'samples': 3552000, 'steps': 18499, 'loss/train': 0.033090163953602314}
+01/27/2022 13:20:32 - INFO - codeparrot_training - Step 18500: {'lr': 0.0003678491842064995, 'samples': 3552192, 'steps': 18500, 'loss/train': 0.6647753119468689}
+01/27/2022 13:20:35 - INFO - codeparrot_training - Step 18501: {'lr': 0.00036783475355115213, 'samples': 3552384, 'steps': 18501, 'loss/train': 1.0583834052085876}
+01/27/2022 13:20:38 - INFO - codeparrot_training - Step 18502: {'lr': 0.0003678203223910382, 'samples': 3552576, 'steps': 18502, 'loss/train': 0.8878505229949951}
+01/27/2022 13:20:41 - INFO - codeparrot_training - Step 18503: {'lr': 0.0003678058907262194, 'samples': 3552768, 'steps': 18503, 'loss/train': 0.7923762202262878}
+01/27/2022 13:20:45 - INFO - codeparrot_training - Step 18504: {'lr': 0.00036779145855675763, 'samples': 3552960, 'steps': 18504, 'loss/train': 1.051880657672882}
+01/27/2022 13:20:48 - INFO - codeparrot_training - Step 18505: {'lr': 0.00036777702588271455, 'samples': 3553152, 'steps': 18505, 'loss/train': 0.7767634391784668}
+01/27/2022 13:20:54 - INFO - codeparrot_training - Step 18506: {'lr': 0.0003677625927041522, 'samples': 3553344, 'steps': 18506, 'loss/train': 0.3506992533802986}
+01/27/2022 13:20:57 - INFO - codeparrot_training - Step 18507: {'lr': 0.0003677481590211322, 'samples': 3553536, 'steps': 18507, 'loss/train': 0.7095646262168884}
+01/27/2022 13:21:00 - INFO - codeparrot_training - Step 18508: {'lr': 0.0003677337248337165, 'samples': 3553728, 'steps': 18508, 'loss/train': 0.9872158169746399}
+01/27/2022 13:21:04 - INFO - codeparrot_training - Step 18509: {'lr': 0.0003677192901419669, 'samples': 3553920, 'steps': 18509, 'loss/train': 0.8386205434799194}
+01/27/2022 13:21:07 - INFO - codeparrot_training - Step 18510: {'lr': 0.0003677048549459453, 'samples': 3554112, 'steps': 18510, 'loss/train': 0.8049817979335785}
+01/27/2022 13:21:10 - INFO - codeparrot_training - Step 18511: {'lr': 0.00036769041924571345, 'samples': 3554304, 'steps': 18511, 'loss/train': 0.3683631867170334}
+01/27/2022 13:21:13 - INFO - codeparrot_training - Step 18512: {'lr': 0.0003676759830413332, 'samples': 3554496, 'steps': 18512, 'loss/train': 1.0704875886440277}
+01/27/2022 13:21:16 - INFO - codeparrot_training - Step 18513: {'lr': 0.00036766154633286635, 'samples': 3554688, 'steps': 18513, 'loss/train': 1.798645555973053}
+01/27/2022 13:21:19 - INFO - codeparrot_training - Step 18514: {'lr': 0.00036764710912037487, 'samples': 3554880, 'steps': 18514, 'loss/train': 0.9899064302444458}
+01/27/2022 13:21:24 - INFO - codeparrot_training - Step 18515: {'lr': 0.00036763267140392053, 'samples': 3555072, 'steps': 18515, 'loss/train': 0.763509213924408}
+01/27/2022 13:21:27 - INFO - codeparrot_training - Step 18516: {'lr': 0.0003676182331835651, 'samples': 3555264, 'steps': 18516, 'loss/train': 0.8135026395320892}
+01/27/2022 13:21:30 - INFO - codeparrot_training - Step 18517: {'lr': 0.00036760379445937067, 'samples': 3555456, 'steps': 18517, 'loss/train': 0.7770262062549591}
+01/27/2022 13:21:33 - INFO - codeparrot_training - Step 18518: {'lr': 0.0003675893552313988, 'samples': 3555648, 'steps': 18518, 'loss/train': 0.912979781627655}
+01/27/2022 13:21:36 - INFO - codeparrot_training - Step 18519: {'lr': 0.0003675749154997115, 'samples': 3555840, 'steps': 18519, 'loss/train': 0.651005744934082}
+01/27/2022 13:21:39 - INFO - codeparrot_training - Step 18520: {'lr': 0.00036756047526437057, 'samples': 3556032, 'steps': 18520, 'loss/train': 1.1127969324588776}
+01/27/2022 13:21:43 - INFO - codeparrot_training - Step 18521: {'lr': 0.00036754603452543796, 'samples': 3556224, 'steps': 18521, 'loss/train': 0.14142419770359993}
+01/27/2022 13:21:46 - INFO - codeparrot_training - Step 18522: {'lr': 0.00036753159328297536, 'samples': 3556416, 'steps': 18522, 'loss/train': 0.6588319391012192}
+01/27/2022 13:21:49 - INFO - codeparrot_training - Step 18523: {'lr': 0.00036751715153704483, 'samples': 3556608, 'steps': 18523, 'loss/train': 0.7903733253479004}
+01/27/2022 13:21:53 - INFO - codeparrot_training - Step 18524: {'lr': 0.0003675027092877081, 'samples': 3556800, 'steps': 18524, 'loss/train': 0.34571119397878647}
+01/27/2022 13:21:56 - INFO - codeparrot_training - Step 18525: {'lr': 0.0003674882665350271, 'samples': 3556992, 'steps': 18525, 'loss/train': 0.4288294315338135}
+01/27/2022 13:22:00 - INFO - codeparrot_training - Step 18526: {'lr': 0.0003674738232790636, 'samples': 3557184, 'steps': 18526, 'loss/train': 0.8233058452606201}
+01/27/2022 13:22:03 - INFO - codeparrot_training - Step 18527: {'lr': 0.0003674593795198796, 'samples': 3557376, 'steps': 18527, 'loss/train': 1.0996398031711578}
+01/27/2022 13:22:06 - INFO - codeparrot_training - Step 18528: {'lr': 0.00036744493525753697, 'samples': 3557568, 'steps': 18528, 'loss/train': 0.36309508234262466}
+01/27/2022 13:22:09 - INFO - codeparrot_training - Step 18529: {'lr': 0.00036743049049209743, 'samples': 3557760, 'steps': 18529, 'loss/train': 0.15456416085362434}
+01/27/2022 13:22:12 - INFO - codeparrot_training - Step 18530: {'lr': 0.00036741604522362304, 'samples': 3557952, 'steps': 18530, 'loss/train': 0.7188944667577744}
+01/27/2022 13:22:15 - INFO - codeparrot_training - Step 18531: {'lr': 0.00036740159945217556, 'samples': 3558144, 'steps': 18531, 'loss/train': 0.7921497523784637}
+01/27/2022 13:22:19 - INFO - codeparrot_training - Step 18532: {'lr': 0.0003673871531778169, 'samples': 3558336, 'steps': 18532, 'loss/train': 0.558163583278656}
+01/27/2022 13:22:25 - INFO - codeparrot_training - Step 18533: {'lr': 0.00036737270640060894, 'samples': 3558528, 'steps': 18533, 'loss/train': 0.7594650685787201}
+01/27/2022 13:22:28 - INFO - codeparrot_training - Step 18534: {'lr': 0.0003673582591206136, 'samples': 3558720, 'steps': 18534, 'loss/train': 0.5239275991916656}
+01/27/2022 13:22:31 - INFO - codeparrot_training - Step 18535: {'lr': 0.00036734381133789277, 'samples': 3558912, 'steps': 18535, 'loss/train': 1.0863468647003174}
+01/27/2022 13:22:34 - INFO - codeparrot_training - Step 18536: {'lr': 0.00036732936305250826, 'samples': 3559104, 'steps': 18536, 'loss/train': 0.8445269465446472}
+01/27/2022 13:22:37 - INFO - codeparrot_training - Step 18537: {'lr': 0.00036731491426452204, 'samples': 3559296, 'steps': 18537, 'loss/train': 1.1128745377063751}
+01/27/2022 13:22:41 - INFO - codeparrot_training - Step 18538: {'lr': 0.00036730046497399587, 'samples': 3559488, 'steps': 18538, 'loss/train': 0.07015471905469894}
+01/27/2022 13:22:44 - INFO - codeparrot_training - Step 18539: {'lr': 0.0003672860151809919, 'samples': 3559680, 'steps': 18539, 'loss/train': 0.7753913104534149}
+01/27/2022 13:22:47 - INFO - codeparrot_training - Step 18540: {'lr': 0.0003672715648855718, 'samples': 3559872, 'steps': 18540, 'loss/train': 1.1139108538627625}
+01/27/2022 13:22:50 - INFO - codeparrot_training - Step 18541: {'lr': 0.00036725711408779765, 'samples': 3560064, 'steps': 18541, 'loss/train': 0.8612324595451355}
+01/27/2022 13:22:55 - INFO - codeparrot_training - Step 18542: {'lr': 0.0003672426627877312, 'samples': 3560256, 'steps': 18542, 'loss/train': 0.7775594294071198}
+01/27/2022 13:22:58 - INFO - codeparrot_training - Step 18543: {'lr': 0.0003672282109854344, 'samples': 3560448, 'steps': 18543, 'loss/train': 0.8499189019203186}
+01/27/2022 13:23:01 - INFO - codeparrot_training - Step 18544: {'lr': 0.00036721375868096925, 'samples': 3560640, 'steps': 18544, 'loss/train': 0.4589877426624298}
+01/27/2022 13:23:04 - INFO - codeparrot_training - Step 18545: {'lr': 0.00036719930587439744, 'samples': 3560832, 'steps': 18545, 'loss/train': 0.2323261946439743}
+01/27/2022 13:23:07 - INFO - codeparrot_training - Step 18546: {'lr': 0.00036718485256578116, 'samples': 3561024, 'steps': 18546, 'loss/train': 0.5779815316200256}
+01/27/2022 13:23:10 - INFO - codeparrot_training - Step 18547: {'lr': 0.00036717039875518203, 'samples': 3561216, 'steps': 18547, 'loss/train': 0.8667841851711273}
+01/27/2022 13:23:13 - INFO - codeparrot_training - Step 18548: {'lr': 0.00036715594444266224, 'samples': 3561408, 'steps': 18548, 'loss/train': 0.5163145065307617}
+01/27/2022 13:23:17 - INFO - codeparrot_training - Step 18549: {'lr': 0.00036714148962828353, 'samples': 3561600, 'steps': 18549, 'loss/train': 1.079548716545105}
+01/27/2022 13:23:20 - INFO - codeparrot_training - Step 18550: {'lr': 0.0003671270343121079, 'samples': 3561792, 'steps': 18550, 'loss/train': 1.720865249633789}
+01/27/2022 13:23:24 - INFO - codeparrot_training - Step 18551: {'lr': 0.0003671125784941972, 'samples': 3561984, 'steps': 18551, 'loss/train': 1.0143998265266418}
+01/27/2022 13:23:27 - INFO - codeparrot_training - Step 18552: {'lr': 0.00036709812217461347, 'samples': 3562176, 'steps': 18552, 'loss/train': 0.9244347810745239}
+01/27/2022 13:23:30 - INFO - codeparrot_training - Step 18553: {'lr': 0.0003670836653534185, 'samples': 3562368, 'steps': 18553, 'loss/train': 0.6659983098506927}
+01/27/2022 13:23:34 - INFO - codeparrot_training - Step 18554: {'lr': 0.0003670692080306743, 'samples': 3562560, 'steps': 18554, 'loss/train': 1.708454668521881}
+01/27/2022 13:23:37 - INFO - codeparrot_training - Step 18555: {'lr': 0.0003670547502064429, 'samples': 3562752, 'steps': 18555, 'loss/train': 0.37841659784317017}
+01/27/2022 13:23:40 - INFO - codeparrot_training - Step 18556: {'lr': 0.000367040291880786, 'samples': 3562944, 'steps': 18556, 'loss/train': 0.89719158411026}
+01/27/2022 13:23:43 - INFO - codeparrot_training - Step 18557: {'lr': 0.0003670258330537656, 'samples': 3563136, 'steps': 18557, 'loss/train': 1.2081689536571503}
+01/27/2022 13:23:46 - INFO - codeparrot_training - Step 18558: {'lr': 0.0003670113737254438, 'samples': 3563328, 'steps': 18558, 'loss/train': 0.94808229804039}
+01/27/2022 13:23:49 - INFO - codeparrot_training - Step 18559: {'lr': 0.0003669969138958824, 'samples': 3563520, 'steps': 18559, 'loss/train': 0.33627015352249146}
+01/27/2022 13:23:55 - INFO - codeparrot_training - Step 18560: {'lr': 0.00036698245356514336, 'samples': 3563712, 'steps': 18560, 'loss/train': 0.9891756176948547}
+01/27/2022 13:23:59 - INFO - codeparrot_training - Step 18561: {'lr': 0.00036696799273328864, 'samples': 3563904, 'steps': 18561, 'loss/train': 0.883620411157608}
+01/27/2022 13:24:02 - INFO - codeparrot_training - Step 18562: {'lr': 0.0003669535314003802, 'samples': 3564096, 'steps': 18562, 'loss/train': 1.363692283630371}
+01/27/2022 13:24:05 - INFO - codeparrot_training - Step 18563: {'lr': 0.00036693906956647996, 'samples': 3564288, 'steps': 18563, 'loss/train': 0.8274165987968445}
+01/27/2022 13:24:08 - INFO - codeparrot_training - Step 18564: {'lr': 0.0003669246072316498, 'samples': 3564480, 'steps': 18564, 'loss/train': 0.9090515971183777}
+01/27/2022 13:24:11 - INFO - codeparrot_training - Step 18565: {'lr': 0.00036691014439595187, 'samples': 3564672, 'steps': 18565, 'loss/train': 0.8282375335693359}
+01/27/2022 13:24:14 - INFO - codeparrot_training - Step 18566: {'lr': 0.00036689568105944794, 'samples': 3564864, 'steps': 18566, 'loss/train': 0.7493789345026016}
+01/27/2022 13:24:17 - INFO - codeparrot_training - Step 18567: {'lr': 0.0003668812172222001, 'samples': 3565056, 'steps': 18567, 'loss/train': 0.3905888646841049}
+01/27/2022 13:24:22 - INFO - codeparrot_training - Step 18568: {'lr': 0.0003668667528842702, 'samples': 3565248, 'steps': 18568, 'loss/train': 0.983105331659317}
+01/27/2022 13:24:25 - INFO - codeparrot_training - Step 18569: {'lr': 0.0003668522880457202, 'samples': 3565440, 'steps': 18569, 'loss/train': 0.7160105556249619}
+01/27/2022 13:24:28 - INFO - codeparrot_training - Step 18570: {'lr': 0.0003668378227066121, 'samples': 3565632, 'steps': 18570, 'loss/train': 1.2158815562725067}
+01/27/2022 13:24:32 - INFO - codeparrot_training - Step 18571: {'lr': 0.00036682335686700796, 'samples': 3565824, 'steps': 18571, 'loss/train': 1.1649152040481567}
+01/27/2022 13:24:35 - INFO - codeparrot_training - Step 18572: {'lr': 0.00036680889052696954, 'samples': 3566016, 'steps': 18572, 'loss/train': 0.8653494715690613}
+01/27/2022 13:24:38 - INFO - codeparrot_training - Step 18573: {'lr': 0.00036679442368655897, 'samples': 3566208, 'steps': 18573, 'loss/train': 0.34480221569538116}
+01/27/2022 13:24:41 - INFO - codeparrot_training - Step 18574: {'lr': 0.00036677995634583815, 'samples': 3566400, 'steps': 18574, 'loss/train': 0.24878805130720139}
+01/27/2022 13:24:44 - INFO - codeparrot_training - Step 18575: {'lr': 0.0003667654885048691, 'samples': 3566592, 'steps': 18575, 'loss/train': 0.2256404086947441}
+01/27/2022 13:24:47 - INFO - codeparrot_training - Step 18576: {'lr': 0.00036675102016371386, 'samples': 3566784, 'steps': 18576, 'loss/train': 0.36383774131536484}
+01/27/2022 13:24:50 - INFO - codeparrot_training - Step 18577: {'lr': 0.0003667365513224342, 'samples': 3566976, 'steps': 18577, 'loss/train': 0.7024054974317551}
+01/27/2022 13:24:55 - INFO - codeparrot_training - Step 18578: {'lr': 0.0003667220819810923, 'samples': 3567168, 'steps': 18578, 'loss/train': 0.7457302361726761}
+01/27/2022 13:24:58 - INFO - codeparrot_training - Step 18579: {'lr': 0.00036670761213975, 'samples': 3567360, 'steps': 18579, 'loss/train': 0.8167273700237274}
+01/27/2022 13:25:01 - INFO - codeparrot_training - Step 18580: {'lr': 0.0003666931417984694, 'samples': 3567552, 'steps': 18580, 'loss/train': 1.1267026662826538}
+01/27/2022 13:25:05 - INFO - codeparrot_training - Step 18581: {'lr': 0.00036667867095731244, 'samples': 3567744, 'steps': 18581, 'loss/train': 0.4053625613451004}
+01/27/2022 13:25:08 - INFO - codeparrot_training - Step 18582: {'lr': 0.0003666641996163411, 'samples': 3567936, 'steps': 18582, 'loss/train': 0.8739952147006989}
+01/27/2022 13:25:11 - INFO - codeparrot_training - Step 18583: {'lr': 0.0003666497277756173, 'samples': 3568128, 'steps': 18583, 'loss/train': 0.9031223952770233}
+01/27/2022 13:25:14 - INFO - codeparrot_training - Step 18584: {'lr': 0.0003666352554352032, 'samples': 3568320, 'steps': 18584, 'loss/train': 0.46945054829120636}
+01/27/2022 13:25:17 - INFO - codeparrot_training - Step 18585: {'lr': 0.0003666207825951606, 'samples': 3568512, 'steps': 18585, 'loss/train': 0.7193960398435593}
+01/27/2022 13:25:24 - INFO - codeparrot_training - Step 18586: {'lr': 0.00036660630925555173, 'samples': 3568704, 'steps': 18586, 'loss/train': 1.0392120480537415}
+01/27/2022 13:25:27 - INFO - codeparrot_training - Step 18587: {'lr': 0.0003665918354164384, 'samples': 3568896, 'steps': 18587, 'loss/train': 0.7959800362586975}
+01/27/2022 13:25:30 - INFO - codeparrot_training - Step 18588: {'lr': 0.00036657736107788264, 'samples': 3569088, 'steps': 18588, 'loss/train': 0.19492493569850922}
+01/27/2022 13:25:33 - INFO - codeparrot_training - Step 18589: {'lr': 0.00036656288623994647, 'samples': 3569280, 'steps': 18589, 'loss/train': 0.8235041499137878}
+01/27/2022 13:25:36 - INFO - codeparrot_training - Step 18590: {'lr': 0.000366548410902692, 'samples': 3569472, 'steps': 18590, 'loss/train': 0.9968390464782715}
+01/27/2022 13:25:39 - INFO - codeparrot_training - Step 18591: {'lr': 0.00036653393506618106, 'samples': 3569664, 'steps': 18591, 'loss/train': 0.9153388738632202}
+01/27/2022 13:25:43 - INFO - codeparrot_training - Step 18592: {'lr': 0.00036651945873047574, 'samples': 3569856, 'steps': 18592, 'loss/train': 1.03916797041893}
+01/27/2022 13:25:46 - INFO - codeparrot_training - Step 18593: {'lr': 0.0003665049818956381, 'samples': 3570048, 'steps': 18593, 'loss/train': 0.7216233462095261}
+01/27/2022 13:25:49 - INFO - codeparrot_training - Step 18594: {'lr': 0.0003664905045617301, 'samples': 3570240, 'steps': 18594, 'loss/train': 1.0269894003868103}
+01/27/2022 13:25:52 - INFO - codeparrot_training - Step 18595: {'lr': 0.0003664760267288138, 'samples': 3570432, 'steps': 18595, 'loss/train': 0.11333409324288368}
+01/27/2022 13:25:57 - INFO - codeparrot_training - Step 18596: {'lr': 0.0003664615483969511, 'samples': 3570624, 'steps': 18596, 'loss/train': 0.3270769268274307}
+01/27/2022 13:26:00 - INFO - codeparrot_training - Step 18597: {'lr': 0.0003664470695662042, 'samples': 3570816, 'steps': 18597, 'loss/train': 0.3343578726053238}
+01/27/2022 13:26:03 - INFO - codeparrot_training - Step 18598: {'lr': 0.000366432590236635, 'samples': 3571008, 'steps': 18598, 'loss/train': 0.7080526649951935}
+01/27/2022 13:26:06 - INFO - codeparrot_training - Step 18599: {'lr': 0.0003664181104083055, 'samples': 3571200, 'steps': 18599, 'loss/train': 0.6257230192422867}
+01/27/2022 13:26:09 - INFO - codeparrot_training - Step 18600: {'lr': 0.00036640363008127785, 'samples': 3571392, 'steps': 18600, 'loss/train': 1.24184051156044}
+01/27/2022 13:26:13 - INFO - codeparrot_training - Step 18601: {'lr': 0.000366389149255614, 'samples': 3571584, 'steps': 18601, 'loss/train': 0.9200806617736816}
+01/27/2022 13:26:16 - INFO - codeparrot_training - Step 18602: {'lr': 0.00036637466793137605, 'samples': 3571776, 'steps': 18602, 'loss/train': 1.1185944378376007}
+01/27/2022 13:26:19 - INFO - codeparrot_training - Step 18603: {'lr': 0.0003663601861086259, 'samples': 3571968, 'steps': 18603, 'loss/train': 0.9236532747745514}
+01/27/2022 13:26:22 - INFO - codeparrot_training - Step 18604: {'lr': 0.00036634570378742565, 'samples': 3572160, 'steps': 18604, 'loss/train': 1.0166428685188293}
+01/27/2022 13:26:26 - INFO - codeparrot_training - Step 18605: {'lr': 0.00036633122096783736, 'samples': 3572352, 'steps': 18605, 'loss/train': 0.7063952386379242}
+01/27/2022 13:26:30 - INFO - codeparrot_training - Step 18606: {'lr': 0.00036631673764992307, 'samples': 3572544, 'steps': 18606, 'loss/train': 0.438681498169899}
+01/27/2022 13:26:33 - INFO - codeparrot_training - Step 18607: {'lr': 0.00036630225383374476, 'samples': 3572736, 'steps': 18607, 'loss/train': 0.7608331739902496}
+01/27/2022 13:26:36 - INFO - codeparrot_training - Step 18608: {'lr': 0.0003662877695193646, 'samples': 3572928, 'steps': 18608, 'loss/train': 0.675964817404747}
+01/27/2022 13:26:39 - INFO - codeparrot_training - Step 18609: {'lr': 0.0003662732847068445, 'samples': 3573120, 'steps': 18609, 'loss/train': 0.8739599883556366}
+01/27/2022 13:26:42 - INFO - codeparrot_training - Step 18610: {'lr': 0.00036625879939624663, 'samples': 3573312, 'steps': 18610, 'loss/train': 0.993057131767273}
+01/27/2022 13:26:45 - INFO - codeparrot_training - Step 18611: {'lr': 0.000366244313587633, 'samples': 3573504, 'steps': 18611, 'loss/train': 0.4305894076824188}
+01/27/2022 13:26:48 - INFO - codeparrot_training - Step 18612: {'lr': 0.0003662298272810655, 'samples': 3573696, 'steps': 18612, 'loss/train': 0.924425482749939}
+01/27/2022 13:26:52 - INFO - codeparrot_training - Step 18613: {'lr': 0.00036621534047660647, 'samples': 3573888, 'steps': 18613, 'loss/train': 1.1073550879955292}
+01/27/2022 13:26:56 - INFO - codeparrot_training - Step 18614: {'lr': 0.00036620085317431777, 'samples': 3574080, 'steps': 18614, 'loss/train': 0.37392374128103256}
+01/27/2022 13:26:59 - INFO - codeparrot_training - Step 18615: {'lr': 0.0003661863653742615, 'samples': 3574272, 'steps': 18615, 'loss/train': 0.4281790852546692}
+01/27/2022 13:27:02 - INFO - codeparrot_training - Step 18616: {'lr': 0.0003661718770764998, 'samples': 3574464, 'steps': 18616, 'loss/train': 0.8818663358688354}
+01/27/2022 13:27:05 - INFO - codeparrot_training - Step 18617: {'lr': 0.00036615738828109465, 'samples': 3574656, 'steps': 18617, 'loss/train': 0.669090673327446}
+01/27/2022 13:27:09 - INFO - codeparrot_training - Step 18618: {'lr': 0.00036614289898810804, 'samples': 3574848, 'steps': 18618, 'loss/train': 1.1144830584526062}
+01/27/2022 13:27:12 - INFO - codeparrot_training - Step 18619: {'lr': 0.00036612840919760225, 'samples': 3575040, 'steps': 18619, 'loss/train': 0.6481812000274658}
+01/27/2022 13:27:15 - INFO - codeparrot_training - Step 18620: {'lr': 0.00036611391890963913, 'samples': 3575232, 'steps': 18620, 'loss/train': 0.9705536663532257}
+01/27/2022 13:27:18 - INFO - codeparrot_training - Step 18621: {'lr': 0.00036609942812428087, 'samples': 3575424, 'steps': 18621, 'loss/train': 0.8682281076908112}
+01/27/2022 13:27:21 - INFO - codeparrot_training - Step 18622: {'lr': 0.00036608493684158963, 'samples': 3575616, 'steps': 18622, 'loss/train': 0.9095510244369507}
+01/27/2022 13:27:28 - INFO - codeparrot_training - Step 18623: {'lr': 0.0003660704450616272, 'samples': 3575808, 'steps': 18623, 'loss/train': 0.9048404395580292}
+01/27/2022 13:27:31 - INFO - codeparrot_training - Step 18624: {'lr': 0.00036605595278445605, 'samples': 3576000, 'steps': 18624, 'loss/train': 0.7642213404178619}
+01/27/2022 13:27:34 - INFO - codeparrot_training - Step 18625: {'lr': 0.0003660414600101379, 'samples': 3576192, 'steps': 18625, 'loss/train': 1.0566789507865906}
+01/27/2022 13:27:37 - INFO - codeparrot_training - Step 18626: {'lr': 0.00036602696673873505, 'samples': 3576384, 'steps': 18626, 'loss/train': 0.9470058381557465}
+01/27/2022 13:27:40 - INFO - codeparrot_training - Step 18627: {'lr': 0.00036601247297030943, 'samples': 3576576, 'steps': 18627, 'loss/train': 1.667824923992157}
+01/27/2022 13:27:43 - INFO - codeparrot_training - Step 18628: {'lr': 0.00036599797870492327, 'samples': 3576768, 'steps': 18628, 'loss/train': 1.2337820827960968}
+01/27/2022 13:27:47 - INFO - codeparrot_training - Step 18629: {'lr': 0.0003659834839426387, 'samples': 3576960, 'steps': 18629, 'loss/train': 0.7781722247600555}
+01/27/2022 13:27:50 - INFO - codeparrot_training - Step 18630: {'lr': 0.0003659689886835176, 'samples': 3577152, 'steps': 18630, 'loss/train': 1.1178059577941895}
+01/27/2022 13:27:53 - INFO - codeparrot_training - Step 18631: {'lr': 0.00036595449292762215, 'samples': 3577344, 'steps': 18631, 'loss/train': 1.1897429823875427}
+01/27/2022 13:27:57 - INFO - codeparrot_training - Step 18632: {'lr': 0.00036593999667501457, 'samples': 3577536, 'steps': 18632, 'loss/train': 0.5518336147069931}
+01/27/2022 13:28:00 - INFO - codeparrot_training - Step 18633: {'lr': 0.0003659254999257568, 'samples': 3577728, 'steps': 18633, 'loss/train': 0.7995511293411255}
+01/27/2022 13:28:04 - INFO - codeparrot_training - Step 18634: {'lr': 0.000365911002679911, 'samples': 3577920, 'steps': 18634, 'loss/train': 0.7052022367715836}
+01/27/2022 13:28:07 - INFO - codeparrot_training - Step 18635: {'lr': 0.00036589650493753937, 'samples': 3578112, 'steps': 18635, 'loss/train': 1.0413710474967957}
+01/27/2022 13:28:10 - INFO - codeparrot_training - Step 18636: {'lr': 0.00036588200669870376, 'samples': 3578304, 'steps': 18636, 'loss/train': 0.924974262714386}
+01/27/2022 13:28:13 - INFO - codeparrot_training - Step 18637: {'lr': 0.0003658675079634665, 'samples': 3578496, 'steps': 18637, 'loss/train': 0.9689501523971558}
+01/27/2022 13:28:16 - INFO - codeparrot_training - Step 18638: {'lr': 0.0003658530087318896, 'samples': 3578688, 'steps': 18638, 'loss/train': 0.5844961702823639}
+01/27/2022 13:28:19 - INFO - codeparrot_training - Step 18639: {'lr': 0.00036583850900403527, 'samples': 3578880, 'steps': 18639, 'loss/train': 0.6997938752174377}
+01/27/2022 13:28:22 - INFO - codeparrot_training - Step 18640: {'lr': 0.00036582400877996547, 'samples': 3579072, 'steps': 18640, 'loss/train': 0.745918795466423}
+01/27/2022 13:28:29 - INFO - codeparrot_training - Step 18641: {'lr': 0.0003658095080597424, 'samples': 3579264, 'steps': 18641, 'loss/train': 0.5814971476793289}
+01/27/2022 13:28:32 - INFO - codeparrot_training - Step 18642: {'lr': 0.0003657950068434282, 'samples': 3579456, 'steps': 18642, 'loss/train': 0.7293587476015091}
+01/27/2022 13:28:35 - INFO - codeparrot_training - Step 18643: {'lr': 0.000365780505131085, 'samples': 3579648, 'steps': 18643, 'loss/train': 0.8553249835968018}
+01/27/2022 13:28:38 - INFO - codeparrot_training - Step 18644: {'lr': 0.00036576600292277477, 'samples': 3579840, 'steps': 18644, 'loss/train': 0.5599940121173859}
+01/27/2022 13:28:41 - INFO - codeparrot_training - Step 18645: {'lr': 0.00036575150021855987, 'samples': 3580032, 'steps': 18645, 'loss/train': 0.7239399701356888}
+01/27/2022 13:28:44 - INFO - codeparrot_training - Step 18646: {'lr': 0.00036573699701850223, 'samples': 3580224, 'steps': 18646, 'loss/train': 1.087976485490799}
+01/27/2022 13:28:47 - INFO - codeparrot_training - Step 18647: {'lr': 0.000365722493322664, 'samples': 3580416, 'steps': 18647, 'loss/train': 0.9544008672237396}
+01/27/2022 13:28:51 - INFO - codeparrot_training - Step 18648: {'lr': 0.0003657079891311075, 'samples': 3580608, 'steps': 18648, 'loss/train': 0.16105610132217407}
+01/27/2022 13:28:54 - INFO - codeparrot_training - Step 18649: {'lr': 0.00036569348444389456, 'samples': 3580800, 'steps': 18649, 'loss/train': 0.7022166699171066}
+01/27/2022 13:28:58 - INFO - codeparrot_training - Step 18650: {'lr': 0.00036567897926108756, 'samples': 3580992, 'steps': 18650, 'loss/train': 0.6962032914161682}
+01/27/2022 13:29:01 - INFO - codeparrot_training - Step 18651: {'lr': 0.00036566447358274846, 'samples': 3581184, 'steps': 18651, 'loss/train': 1.0919153094291687}
+01/27/2022 13:29:05 - INFO - codeparrot_training - Step 18652: {'lr': 0.0003656499674089396, 'samples': 3581376, 'steps': 18652, 'loss/train': 1.0218662023544312}
+01/27/2022 13:29:08 - INFO - codeparrot_training - Step 18653: {'lr': 0.0003656354607397229, 'samples': 3581568, 'steps': 18653, 'loss/train': 0.6575139462947845}
+01/27/2022 13:29:11 - INFO - codeparrot_training - Step 18654: {'lr': 0.00036562095357516066, 'samples': 3581760, 'steps': 18654, 'loss/train': 0.6128004938364029}
+01/27/2022 13:29:14 - INFO - codeparrot_training - Step 18655: {'lr': 0.00036560644591531496, 'samples': 3581952, 'steps': 18655, 'loss/train': 0.8874427378177643}
+01/27/2022 13:29:17 - INFO - codeparrot_training - Step 18656: {'lr': 0.00036559193776024794, 'samples': 3582144, 'steps': 18656, 'loss/train': 1.0205704271793365}
+01/27/2022 13:29:20 - INFO - codeparrot_training - Step 18657: {'lr': 0.0003655774291100218, 'samples': 3582336, 'steps': 18657, 'loss/train': 0.9981542229652405}
+01/27/2022 13:29:25 - INFO - codeparrot_training - Step 18658: {'lr': 0.0003655629199646986, 'samples': 3582528, 'steps': 18658, 'loss/train': 0.2609299197793007}
+01/27/2022 13:29:28 - INFO - codeparrot_training - Step 18659: {'lr': 0.00036554841032434063, 'samples': 3582720, 'steps': 18659, 'loss/train': 1.1488370597362518}
+01/27/2022 13:29:31 - INFO - codeparrot_training - Step 18660: {'lr': 0.00036553390018900984, 'samples': 3582912, 'steps': 18660, 'loss/train': 1.1021738648414612}
+01/27/2022 13:29:34 - INFO - codeparrot_training - Step 18661: {'lr': 0.0003655193895587686, 'samples': 3583104, 'steps': 18661, 'loss/train': 0.4428962767124176}
+01/27/2022 13:29:37 - INFO - codeparrot_training - Step 18662: {'lr': 0.000365504878433679, 'samples': 3583296, 'steps': 18662, 'loss/train': 0.7026685774326324}
+01/27/2022 13:29:40 - INFO - codeparrot_training - Step 18663: {'lr': 0.00036549036681380307, 'samples': 3583488, 'steps': 18663, 'loss/train': 0.5013768374919891}
+01/27/2022 13:29:44 - INFO - codeparrot_training - Step 18664: {'lr': 0.00036547585469920316, 'samples': 3583680, 'steps': 18664, 'loss/train': 0.7646104395389557}
+01/27/2022 13:29:47 - INFO - codeparrot_training - Step 18665: {'lr': 0.00036546134208994137, 'samples': 3583872, 'steps': 18665, 'loss/train': 1.031805843114853}
+01/27/2022 13:29:50 - INFO - codeparrot_training - Step 18666: {'lr': 0.00036544682898607977, 'samples': 3584064, 'steps': 18666, 'loss/train': 0.7117861211299896}
+01/27/2022 13:29:56 - INFO - codeparrot_training - Step 18667: {'lr': 0.00036543231538768066, 'samples': 3584256, 'steps': 18667, 'loss/train': 0.49831853806972504}
+01/27/2022 13:29:59 - INFO - codeparrot_training - Step 18668: {'lr': 0.00036541780129480616, 'samples': 3584448, 'steps': 18668, 'loss/train': 0.73811374604702}
+01/27/2022 13:30:03 - INFO - codeparrot_training - Step 18669: {'lr': 0.0003654032867075185, 'samples': 3584640, 'steps': 18669, 'loss/train': 0.36278707534074783}
+01/27/2022 13:30:06 - INFO - codeparrot_training - Step 18670: {'lr': 0.00036538877162587975, 'samples': 3584832, 'steps': 18670, 'loss/train': 1.0078959167003632}
+01/27/2022 13:30:09 - INFO - codeparrot_training - Step 18671: {'lr': 0.00036537425604995214, 'samples': 3585024, 'steps': 18671, 'loss/train': 0.8687787652015686}
+01/27/2022 13:30:12 - INFO - codeparrot_training - Step 18672: {'lr': 0.00036535973997979787, 'samples': 3585216, 'steps': 18672, 'loss/train': 0.6364780962467194}
+01/27/2022 13:30:15 - INFO - codeparrot_training - Step 18673: {'lr': 0.0003653452234154791, 'samples': 3585408, 'steps': 18673, 'loss/train': 0.8360619843006134}
+01/27/2022 13:30:18 - INFO - codeparrot_training - Step 18674: {'lr': 0.000365330706357058, 'samples': 3585600, 'steps': 18674, 'loss/train': 0.37529660761356354}
+01/27/2022 13:30:21 - INFO - codeparrot_training - Step 18675: {'lr': 0.0003653161888045968, 'samples': 3585792, 'steps': 18675, 'loss/train': 1.113235205411911}
+01/27/2022 13:30:26 - INFO - codeparrot_training - Step 18676: {'lr': 0.0003653016707581577, 'samples': 3585984, 'steps': 18676, 'loss/train': 0.7460271120071411}
+01/27/2022 13:30:29 - INFO - codeparrot_training - Step 18677: {'lr': 0.00036528715221780276, 'samples': 3586176, 'steps': 18677, 'loss/train': 0.5702787190675735}
+01/27/2022 13:30:32 - INFO - codeparrot_training - Step 18678: {'lr': 0.0003652726331835944, 'samples': 3586368, 'steps': 18678, 'loss/train': 0.6129147559404373}
+01/27/2022 13:30:35 - INFO - codeparrot_training - Step 18679: {'lr': 0.00036525811365559457, 'samples': 3586560, 'steps': 18679, 'loss/train': 0.4998675584793091}
+01/27/2022 13:30:39 - INFO - codeparrot_training - Step 18680: {'lr': 0.0003652435936338656, 'samples': 3586752, 'steps': 18680, 'loss/train': 0.89584019780159}
+01/27/2022 13:30:42 - INFO - codeparrot_training - Step 18681: {'lr': 0.0003652290731184697, 'samples': 3586944, 'steps': 18681, 'loss/train': 0.7347650974988937}
+01/27/2022 13:30:45 - INFO - codeparrot_training - Step 18682: {'lr': 0.000365214552109469, 'samples': 3587136, 'steps': 18682, 'loss/train': 0.6465654373168945}
+01/27/2022 13:30:48 - INFO - codeparrot_training - Step 18683: {'lr': 0.0003652000306069258, 'samples': 3587328, 'steps': 18683, 'loss/train': 0.6111679226160049}
+01/27/2022 13:30:51 - INFO - codeparrot_training - Step 18684: {'lr': 0.00036518550861090217, 'samples': 3587520, 'steps': 18684, 'loss/train': 0.91005939245224}
+01/27/2022 13:30:58 - INFO - codeparrot_training - Step 18685: {'lr': 0.0003651709861214605, 'samples': 3587712, 'steps': 18685, 'loss/train': 0.5589654743671417}
+01/27/2022 13:31:01 - INFO - codeparrot_training - Step 18686: {'lr': 0.0003651564631386628, 'samples': 3587904, 'steps': 18686, 'loss/train': 0.9091226756572723}
+01/27/2022 13:31:04 - INFO - codeparrot_training - Step 18687: {'lr': 0.0003651419396625714, 'samples': 3588096, 'steps': 18687, 'loss/train': 0.9352650046348572}
+01/27/2022 13:31:07 - INFO - codeparrot_training - Step 18688: {'lr': 0.0003651274156932485, 'samples': 3588288, 'steps': 18688, 'loss/train': 0.5225381702184677}
+01/27/2022 13:31:10 - INFO - codeparrot_training - Step 18689: {'lr': 0.00036511289123075636, 'samples': 3588480, 'steps': 18689, 'loss/train': 1.6766395568847656}
+01/27/2022 13:31:13 - INFO - codeparrot_training - Step 18690: {'lr': 0.0003650983662751571, 'samples': 3588672, 'steps': 18690, 'loss/train': 0.8424912393093109}
+01/27/2022 13:31:16 - INFO - codeparrot_training - Step 18691: {'lr': 0.000365083840826513, 'samples': 3588864, 'steps': 18691, 'loss/train': 0.7372515946626663}
+01/27/2022 13:31:19 - INFO - codeparrot_training - Step 18692: {'lr': 0.00036506931488488627, 'samples': 3589056, 'steps': 18692, 'loss/train': 0.7959060072898865}
+01/27/2022 13:31:23 - INFO - codeparrot_training - Step 18693: {'lr': 0.0003650547884503391, 'samples': 3589248, 'steps': 18693, 'loss/train': 0.9670954942703247}
+01/27/2022 13:31:27 - INFO - codeparrot_training - Step 18694: {'lr': 0.0003650402615229338, 'samples': 3589440, 'steps': 18694, 'loss/train': 0.7482847720384598}
+01/27/2022 13:31:30 - INFO - codeparrot_training - Step 18695: {'lr': 0.00036502573410273243, 'samples': 3589632, 'steps': 18695, 'loss/train': 0.3130272701382637}
+01/27/2022 13:31:34 - INFO - codeparrot_training - Step 18696: {'lr': 0.0003650112061897975, 'samples': 3589824, 'steps': 18696, 'loss/train': 0.6871450692415237}
+01/27/2022 13:31:37 - INFO - codeparrot_training - Step 18697: {'lr': 0.000364996677784191, 'samples': 3590016, 'steps': 18697, 'loss/train': 0.45088519155979156}
+01/27/2022 13:31:40 - INFO - codeparrot_training - Step 18698: {'lr': 0.00036498214888597524, 'samples': 3590208, 'steps': 18698, 'loss/train': 0.5250630229711533}
+01/27/2022 13:31:43 - INFO - codeparrot_training - Step 18699: {'lr': 0.0003649676194952125, 'samples': 3590400, 'steps': 18699, 'loss/train': 0.9163219034671783}
+01/27/2022 13:31:46 - INFO - codeparrot_training - Step 18700: {'lr': 0.000364953089611965, 'samples': 3590592, 'steps': 18700, 'loss/train': 0.8193682730197906}
+01/27/2022 13:31:49 - INFO - codeparrot_training - Step 18701: {'lr': 0.00036493855923629495, 'samples': 3590784, 'steps': 18701, 'loss/train': 0.6986059248447418}
+01/27/2022 13:31:52 - INFO - codeparrot_training - Step 18702: {'lr': 0.00036492402836826453, 'samples': 3590976, 'steps': 18702, 'loss/train': 1.3268450796604156}
+01/27/2022 13:31:58 - INFO - codeparrot_training - Step 18703: {'lr': 0.0003649094970079362, 'samples': 3591168, 'steps': 18703, 'loss/train': 0.9038107395172119}
+01/27/2022 13:32:01 - INFO - codeparrot_training - Step 18704: {'lr': 0.000364894965155372, 'samples': 3591360, 'steps': 18704, 'loss/train': 1.0555518865585327}
+01/27/2022 13:32:04 - INFO - codeparrot_training - Step 18705: {'lr': 0.0003648804328106342, 'samples': 3591552, 'steps': 18705, 'loss/train': 1.0112284719944}
+01/27/2022 13:32:07 - INFO - codeparrot_training - Step 18706: {'lr': 0.0003648658999737852, 'samples': 3591744, 'steps': 18706, 'loss/train': 0.7170833051204681}
+01/27/2022 13:32:10 - INFO - codeparrot_training - Step 18707: {'lr': 0.0003648513666448871, 'samples': 3591936, 'steps': 18707, 'loss/train': 0.7326899617910385}
+01/27/2022 13:32:14 - INFO - codeparrot_training - Step 18708: {'lr': 0.0003648368328240022, 'samples': 3592128, 'steps': 18708, 'loss/train': 0.82109534740448}
+01/27/2022 13:32:17 - INFO - codeparrot_training - Step 18709: {'lr': 0.00036482229851119287, 'samples': 3592320, 'steps': 18709, 'loss/train': 1.2437048256397247}
+01/27/2022 13:32:20 - INFO - codeparrot_training - Step 18710: {'lr': 0.0003648077637065212, 'samples': 3592512, 'steps': 18710, 'loss/train': 2.1557591557502747}
+01/27/2022 13:32:23 - INFO - codeparrot_training - Step 18711: {'lr': 0.00036479322841004953, 'samples': 3592704, 'steps': 18711, 'loss/train': 1.6870177388191223}
+01/27/2022 13:32:26 - INFO - codeparrot_training - Step 18712: {'lr': 0.0003647786926218401, 'samples': 3592896, 'steps': 18712, 'loss/train': 0.620644211769104}
+01/27/2022 13:32:31 - INFO - codeparrot_training - Step 18713: {'lr': 0.00036476415634195523, 'samples': 3593088, 'steps': 18713, 'loss/train': 0.5742541998624802}
+01/27/2022 13:32:34 - INFO - codeparrot_training - Step 18714: {'lr': 0.00036474961957045715, 'samples': 3593280, 'steps': 18714, 'loss/train': 1.0669689774513245}
+01/27/2022 13:32:37 - INFO - codeparrot_training - Step 18715: {'lr': 0.00036473508230740816, 'samples': 3593472, 'steps': 18715, 'loss/train': 1.1857856512069702}
+01/27/2022 13:32:40 - INFO - codeparrot_training - Step 18716: {'lr': 0.00036472054455287053, 'samples': 3593664, 'steps': 18716, 'loss/train': 1.4861932396888733}
+01/27/2022 13:32:43 - INFO - codeparrot_training - Step 18717: {'lr': 0.0003647060063069064, 'samples': 3593856, 'steps': 18717, 'loss/train': 0.9604891240596771}
+01/27/2022 13:32:46 - INFO - codeparrot_training - Step 18718: {'lr': 0.0003646914675695783, 'samples': 3594048, 'steps': 18718, 'loss/train': 1.24799644947052}
+01/27/2022 13:32:49 - INFO - codeparrot_training - Step 18719: {'lr': 0.0003646769283409483, 'samples': 3594240, 'steps': 18719, 'loss/train': 0.6126604825258255}
+01/27/2022 13:32:53 - INFO - codeparrot_training - Step 18720: {'lr': 0.0003646623886210788, 'samples': 3594432, 'steps': 18720, 'loss/train': 0.8666823506355286}
+01/27/2022 13:32:56 - INFO - codeparrot_training - Step 18721: {'lr': 0.00036464784841003196, 'samples': 3594624, 'steps': 18721, 'loss/train': 0.6993962377309799}
+01/27/2022 13:33:02 - INFO - codeparrot_training - Step 18722: {'lr': 0.0003646333077078702, 'samples': 3594816, 'steps': 18722, 'loss/train': 0.9736563563346863}
+01/27/2022 13:33:05 - INFO - codeparrot_training - Step 18723: {'lr': 0.0003646187665146557, 'samples': 3595008, 'steps': 18723, 'loss/train': 1.071164309978485}
+01/27/2022 13:33:08 - INFO - codeparrot_training - Step 18724: {'lr': 0.00036460422483045084, 'samples': 3595200, 'steps': 18724, 'loss/train': 0.7410981506109238}
+01/27/2022 13:33:11 - INFO - codeparrot_training - Step 18725: {'lr': 0.0003645896826553178, 'samples': 3595392, 'steps': 18725, 'loss/train': 0.8594644367694855}
+01/27/2022 13:33:14 - INFO - codeparrot_training - Step 18726: {'lr': 0.0003645751399893191, 'samples': 3595584, 'steps': 18726, 'loss/train': 0.9975924789905548}
+01/27/2022 13:33:17 - INFO - codeparrot_training - Step 18727: {'lr': 0.0003645605968325167, 'samples': 3595776, 'steps': 18727, 'loss/train': 0.5188171416521072}
+01/27/2022 13:33:21 - INFO - codeparrot_training - Step 18728: {'lr': 0.00036454605318497323, 'samples': 3595968, 'steps': 18728, 'loss/train': 0.9689260125160217}
+01/27/2022 13:33:24 - INFO - codeparrot_training - Step 18729: {'lr': 0.00036453150904675074, 'samples': 3596160, 'steps': 18729, 'loss/train': 0.9440776705741882}
+01/27/2022 13:33:27 - INFO - codeparrot_training - Step 18730: {'lr': 0.00036451696441791164, 'samples': 3596352, 'steps': 18730, 'loss/train': 0.797278493642807}
+01/27/2022 13:33:31 - INFO - codeparrot_training - Step 18731: {'lr': 0.0003645024192985183, 'samples': 3596544, 'steps': 18731, 'loss/train': 0.5250157713890076}
+01/27/2022 13:33:34 - INFO - codeparrot_training - Step 18732: {'lr': 0.0003644878736886329, 'samples': 3596736, 'steps': 18732, 'loss/train': 1.03001868724823}
+01/27/2022 13:33:37 - INFO - codeparrot_training - Step 18733: {'lr': 0.0003644733275883179, 'samples': 3596928, 'steps': 18733, 'loss/train': 1.6762311458587646}
+01/27/2022 13:33:41 - INFO - codeparrot_training - Step 18734: {'lr': 0.00036445878099763534, 'samples': 3597120, 'steps': 18734, 'loss/train': 0.5927459746599197}
+01/27/2022 13:33:44 - INFO - codeparrot_training - Step 18735: {'lr': 0.00036444423391664783, 'samples': 3597312, 'steps': 18735, 'loss/train': 0.7062072604894638}
+01/27/2022 13:33:47 - INFO - codeparrot_training - Step 18736: {'lr': 0.0003644296863454175, 'samples': 3597504, 'steps': 18736, 'loss/train': 1.0655791461467743}
+01/27/2022 13:33:50 - INFO - codeparrot_training - Step 18737: {'lr': 0.0003644151382840068, 'samples': 3597696, 'steps': 18737, 'loss/train': 0.4314012676477432}
+01/27/2022 13:33:53 - INFO - codeparrot_training - Step 18738: {'lr': 0.00036440058973247793, 'samples': 3597888, 'steps': 18738, 'loss/train': 0.9579728543758392}
+01/27/2022 13:33:56 - INFO - codeparrot_training - Step 18739: {'lr': 0.0003643860406908933, 'samples': 3598080, 'steps': 18739, 'loss/train': 0.8362397253513336}
+01/27/2022 13:34:01 - INFO - codeparrot_training - Step 18740: {'lr': 0.0003643714911593151, 'samples': 3598272, 'steps': 18740, 'loss/train': 0.6335264146327972}
+01/27/2022 13:34:04 - INFO - codeparrot_training - Step 18741: {'lr': 0.00036435694113780585, 'samples': 3598464, 'steps': 18741, 'loss/train': 0.9242131412029266}
+01/27/2022 13:34:07 - INFO - codeparrot_training - Step 18742: {'lr': 0.0003643423906264277, 'samples': 3598656, 'steps': 18742, 'loss/train': 1.0014483332633972}
+01/27/2022 13:34:10 - INFO - codeparrot_training - Step 18743: {'lr': 0.0003643278396252431, 'samples': 3598848, 'steps': 18743, 'loss/train': 0.5943898111581802}
+01/27/2022 13:34:14 - INFO - codeparrot_training - Step 18744: {'lr': 0.0003643132881343144, 'samples': 3599040, 'steps': 18744, 'loss/train': 1.176072746515274}
+01/27/2022 13:34:17 - INFO - codeparrot_training - Step 18745: {'lr': 0.00036429873615370374, 'samples': 3599232, 'steps': 18745, 'loss/train': 0.8328054249286652}
+01/27/2022 13:34:20 - INFO - codeparrot_training - Step 18746: {'lr': 0.00036428418368347363, 'samples': 3599424, 'steps': 18746, 'loss/train': 1.1894085109233856}
+01/27/2022 13:34:23 - INFO - codeparrot_training - Step 18747: {'lr': 0.0003642696307236864, 'samples': 3599616, 'steps': 18747, 'loss/train': 1.0547876358032227}
+01/27/2022 13:34:29 - INFO - codeparrot_training - Step 18748: {'lr': 0.0003642550772744044, 'samples': 3599808, 'steps': 18748, 'loss/train': 0.9486473500728607}
+01/27/2022 13:34:32 - INFO - codeparrot_training - Step 18749: {'lr': 0.0003642405233356898, 'samples': 3600000, 'steps': 18749, 'loss/train': 0.77495938539505}
+01/27/2022 13:34:35 - INFO - codeparrot_training - Step 18750: {'lr': 0.00036422596890760517, 'samples': 3600192, 'steps': 18750, 'loss/train': 0.7965442836284637}
+01/27/2022 13:34:38 - INFO - codeparrot_training - Step 18751: {'lr': 0.00036421141399021274, 'samples': 3600384, 'steps': 18751, 'loss/train': 0.8701559901237488}
+01/27/2022 13:34:42 - INFO - codeparrot_training - Step 18752: {'lr': 0.00036419685858357485, 'samples': 3600576, 'steps': 18752, 'loss/train': 0.5557377487421036}
+01/27/2022 13:34:45 - INFO - codeparrot_training - Step 18753: {'lr': 0.00036418230268775393, 'samples': 3600768, 'steps': 18753, 'loss/train': 0.7191696614027023}
+01/27/2022 13:34:48 - INFO - codeparrot_training - Step 18754: {'lr': 0.0003641677463028123, 'samples': 3600960, 'steps': 18754, 'loss/train': 0.8291792571544647}
+01/27/2022 13:34:51 - INFO - codeparrot_training - Step 18755: {'lr': 0.0003641531894288122, 'samples': 3601152, 'steps': 18755, 'loss/train': 0.7096597999334335}
+01/27/2022 13:34:54 - INFO - codeparrot_training - Step 18756: {'lr': 0.0003641386320658161, 'samples': 3601344, 'steps': 18756, 'loss/train': 0.775615006685257}
+01/27/2022 13:34:59 - INFO - codeparrot_training - Step 18757: {'lr': 0.00036412407421388646, 'samples': 3601536, 'steps': 18757, 'loss/train': 0.9413596987724304}
+01/27/2022 13:35:02 - INFO - codeparrot_training - Step 18758: {'lr': 0.00036410951587308545, 'samples': 3601728, 'steps': 18758, 'loss/train': 0.18826409429311752}
+01/27/2022 13:35:05 - INFO - codeparrot_training - Step 18759: {'lr': 0.00036409495704347553, 'samples': 3601920, 'steps': 18759, 'loss/train': 0.732478603720665}
+01/27/2022 13:35:08 - INFO - codeparrot_training - Step 18760: {'lr': 0.000364080397725119, 'samples': 3602112, 'steps': 18760, 'loss/train': 1.0027509033679962}
+01/27/2022 13:35:11 - INFO - codeparrot_training - Step 18761: {'lr': 0.00036406583791807824, 'samples': 3602304, 'steps': 18761, 'loss/train': 0.8675553202629089}
+01/27/2022 13:35:14 - INFO - codeparrot_training - Step 18762: {'lr': 0.0003640512776224157, 'samples': 3602496, 'steps': 18762, 'loss/train': 0.8951370120048523}
+01/27/2022 13:35:17 - INFO - codeparrot_training - Step 18763: {'lr': 0.0003640367168381937, 'samples': 3602688, 'steps': 18763, 'loss/train': 1.2560701668262482}
+01/27/2022 13:35:21 - INFO - codeparrot_training - Step 18764: {'lr': 0.0003640221555654747, 'samples': 3602880, 'steps': 18764, 'loss/train': 0.775719165802002}
+01/27/2022 13:35:24 - INFO - codeparrot_training - Step 18765: {'lr': 0.00036400759380432083, 'samples': 3603072, 'steps': 18765, 'loss/train': 1.0565375089645386}
+01/27/2022 13:35:29 - INFO - codeparrot_training - Step 18766: {'lr': 0.00036399303155479476, 'samples': 3603264, 'steps': 18766, 'loss/train': 0.8546505868434906}
+01/27/2022 13:35:32 - INFO - codeparrot_training - Step 18767: {'lr': 0.00036397846881695866, 'samples': 3603456, 'steps': 18767, 'loss/train': 0.7693002820014954}
+01/27/2022 13:35:35 - INFO - codeparrot_training - Step 18768: {'lr': 0.00036396390559087505, 'samples': 3603648, 'steps': 18768, 'loss/train': 0.7569660544395447}
+01/27/2022 13:35:38 - INFO - codeparrot_training - Step 18769: {'lr': 0.0003639493418766062, 'samples': 3603840, 'steps': 18769, 'loss/train': 0.8837428987026215}
+01/27/2022 13:35:42 - INFO - codeparrot_training - Step 18770: {'lr': 0.0003639347776742146, 'samples': 3604032, 'steps': 18770, 'loss/train': 1.2759531140327454}
+01/27/2022 13:35:45 - INFO - codeparrot_training - Step 18771: {'lr': 0.00036392021298376257, 'samples': 3604224, 'steps': 18771, 'loss/train': 0.7366109937429428}
+01/27/2022 13:35:48 - INFO - codeparrot_training - Step 18772: {'lr': 0.0003639056478053125, 'samples': 3604416, 'steps': 18772, 'loss/train': 1.2128584384918213}
+01/27/2022 13:35:51 - INFO - codeparrot_training - Step 18773: {'lr': 0.0003638910821389268, 'samples': 3604608, 'steps': 18773, 'loss/train': 1.1933401823043823}
+01/27/2022 13:35:54 - INFO - codeparrot_training - Step 18774: {'lr': 0.0003638765159846679, 'samples': 3604800, 'steps': 18774, 'loss/train': 1.0711041390895844}
+01/27/2022 13:35:58 - INFO - codeparrot_training - Step 18775: {'lr': 0.0003638619493425982, 'samples': 3604992, 'steps': 18775, 'loss/train': 0.6152843087911606}
+01/27/2022 13:36:02 - INFO - codeparrot_training - Step 18776: {'lr': 0.00036384738221278, 'samples': 3605184, 'steps': 18776, 'loss/train': 0.710540235042572}
+01/27/2022 13:36:05 - INFO - codeparrot_training - Step 18777: {'lr': 0.0003638328145952758, 'samples': 3605376, 'steps': 18777, 'loss/train': 0.49129919707775116}
+01/27/2022 13:36:08 - INFO - codeparrot_training - Step 18778: {'lr': 0.0003638182464901479, 'samples': 3605568, 'steps': 18778, 'loss/train': 0.6175988763570786}
+01/27/2022 13:36:11 - INFO - codeparrot_training - Step 18779: {'lr': 0.00036380367789745887, 'samples': 3605760, 'steps': 18779, 'loss/train': 0.9006537795066833}
+01/27/2022 13:36:14 - INFO - codeparrot_training - Step 18780: {'lr': 0.00036378910881727095, 'samples': 3605952, 'steps': 18780, 'loss/train': 0.742204338312149}
+01/27/2022 13:36:17 - INFO - codeparrot_training - Step 18781: {'lr': 0.00036377453924964665, 'samples': 3606144, 'steps': 18781, 'loss/train': 0.37506620585918427}
+01/27/2022 13:36:20 - INFO - codeparrot_training - Step 18782: {'lr': 0.0003637599691946484, 'samples': 3606336, 'steps': 18782, 'loss/train': 0.5798609554767609}
+01/27/2022 13:36:25 - INFO - codeparrot_training - Step 18783: {'lr': 0.00036374539865233847, 'samples': 3606528, 'steps': 18783, 'loss/train': 0.6621811240911484}
+01/27/2022 13:36:28 - INFO - codeparrot_training - Step 18784: {'lr': 0.00036373082762277943, 'samples': 3606720, 'steps': 18784, 'loss/train': 1.415568083524704}
+01/27/2022 13:36:31 - INFO - codeparrot_training - Step 18785: {'lr': 0.00036371625610603366, 'samples': 3606912, 'steps': 18785, 'loss/train': 0.9581551551818848}
+01/27/2022 13:36:34 - INFO - codeparrot_training - Step 18786: {'lr': 0.0003637016841021635, 'samples': 3607104, 'steps': 18786, 'loss/train': 0.4889935255050659}
+01/27/2022 13:36:38 - INFO - codeparrot_training - Step 18787: {'lr': 0.00036368711161123145, 'samples': 3607296, 'steps': 18787, 'loss/train': 0.9422788023948669}
+01/27/2022 13:36:41 - INFO - codeparrot_training - Step 18788: {'lr': 0.0003636725386332999, 'samples': 3607488, 'steps': 18788, 'loss/train': 0.8423256576061249}
+01/27/2022 13:36:44 - INFO - codeparrot_training - Step 18789: {'lr': 0.0003636579651684313, 'samples': 3607680, 'steps': 18789, 'loss/train': 0.8653550148010254}
+01/27/2022 13:36:47 - INFO - codeparrot_training - Step 18790: {'lr': 0.0003636433912166881, 'samples': 3607872, 'steps': 18790, 'loss/train': 1.208050400018692}
+01/27/2022 13:36:50 - INFO - codeparrot_training - Step 18791: {'lr': 0.00036362881677813266, 'samples': 3608064, 'steps': 18791, 'loss/train': 0.6179940551519394}
+01/27/2022 13:36:55 - INFO - codeparrot_training - Step 18792: {'lr': 0.00036361424185282743, 'samples': 3608256, 'steps': 18792, 'loss/train': 0.8245638012886047}
+01/27/2022 13:36:58 - INFO - codeparrot_training - Step 18793: {'lr': 0.00036359966644083485, 'samples': 3608448, 'steps': 18793, 'loss/train': 1.0332440435886383}
+01/27/2022 13:37:02 - INFO - codeparrot_training - Step 18794: {'lr': 0.0003635850905422175, 'samples': 3608640, 'steps': 18794, 'loss/train': 0.8307240307331085}
+01/27/2022 13:37:05 - INFO - codeparrot_training - Step 18795: {'lr': 0.00036357051415703755, 'samples': 3608832, 'steps': 18795, 'loss/train': 0.9462301433086395}
+01/27/2022 13:37:08 - INFO - codeparrot_training - Step 18796: {'lr': 0.00036355593728535767, 'samples': 3609024, 'steps': 18796, 'loss/train': 0.4840237498283386}
+01/27/2022 13:37:11 - INFO - codeparrot_training - Step 18797: {'lr': 0.00036354135992724013, 'samples': 3609216, 'steps': 18797, 'loss/train': 0.5692341774702072}
+01/27/2022 13:37:14 - INFO - codeparrot_training - Step 18798: {'lr': 0.0003635267820827476, 'samples': 3609408, 'steps': 18798, 'loss/train': 0.33349259197711945}
+01/27/2022 13:37:17 - INFO - codeparrot_training - Step 18799: {'lr': 0.0003635122037519422, 'samples': 3609600, 'steps': 18799, 'loss/train': 1.8738741874694824}
+01/27/2022 13:37:20 - INFO - codeparrot_training - Step 18800: {'lr': 0.00036349762493488667, 'samples': 3609792, 'steps': 18800, 'loss/train': 0.7156499773263931}
+01/27/2022 13:37:25 - INFO - codeparrot_training - Step 18801: {'lr': 0.0003634830456316434, 'samples': 3609984, 'steps': 18801, 'loss/train': 0.4695017337799072}
+01/27/2022 13:37:28 - INFO - codeparrot_training - Step 18802: {'lr': 0.00036346846584227473, 'samples': 3610176, 'steps': 18802, 'loss/train': 1.1469914317131042}
+01/27/2022 13:37:31 - INFO - codeparrot_training - Step 18803: {'lr': 0.0003634538855668432, 'samples': 3610368, 'steps': 18803, 'loss/train': 0.538662001490593}
+01/27/2022 13:37:35 - INFO - codeparrot_training - Step 18804: {'lr': 0.00036343930480541123, 'samples': 3610560, 'steps': 18804, 'loss/train': 0.9001286029815674}
+01/27/2022 13:37:38 - INFO - codeparrot_training - Step 18805: {'lr': 0.0003634247235580413, 'samples': 3610752, 'steps': 18805, 'loss/train': 0.62286776304245}
+01/27/2022 13:37:41 - INFO - codeparrot_training - Step 18806: {'lr': 0.0003634101418247959, 'samples': 3610944, 'steps': 18806, 'loss/train': 0.7698130309581757}
+01/27/2022 13:37:44 - INFO - codeparrot_training - Step 18807: {'lr': 0.0003633955596057374, 'samples': 3611136, 'steps': 18807, 'loss/train': 0.6932909488677979}
+01/27/2022 13:37:47 - INFO - codeparrot_training - Step 18808: {'lr': 0.00036338097690092843, 'samples': 3611328, 'steps': 18808, 'loss/train': 1.0684490203857422}
+01/27/2022 13:37:50 - INFO - codeparrot_training - Step 18809: {'lr': 0.0003633663937104313, 'samples': 3611520, 'steps': 18809, 'loss/train': 0.8909383714199066}
+01/27/2022 13:37:55 - INFO - codeparrot_training - Step 18810: {'lr': 0.0003633518100343085, 'samples': 3611712, 'steps': 18810, 'loss/train': 0.6677201986312866}
+01/27/2022 13:37:59 - INFO - codeparrot_training - Step 18811: {'lr': 0.0003633372258726226, 'samples': 3611904, 'steps': 18811, 'loss/train': 1.0450266301631927}
+01/27/2022 13:38:02 - INFO - codeparrot_training - Step 18812: {'lr': 0.00036332264122543594, 'samples': 3612096, 'steps': 18812, 'loss/train': 0.80339714884758}
+01/27/2022 13:38:05 - INFO - codeparrot_training - Step 18813: {'lr': 0.0003633080560928111, 'samples': 3612288, 'steps': 18813, 'loss/train': 1.0937435030937195}
+01/27/2022 13:38:08 - INFO - codeparrot_training - Step 18814: {'lr': 0.0003632934704748106, 'samples': 3612480, 'steps': 18814, 'loss/train': 0.829008936882019}
+01/27/2022 13:38:11 - INFO - codeparrot_training - Step 18815: {'lr': 0.00036327888437149674, 'samples': 3612672, 'steps': 18815, 'loss/train': 0.7295287102460861}
+01/27/2022 13:38:14 - INFO - codeparrot_training - Step 18816: {'lr': 0.00036326429778293223, 'samples': 3612864, 'steps': 18816, 'loss/train': 0.8600471019744873}
+01/27/2022 13:38:17 - INFO - codeparrot_training - Step 18817: {'lr': 0.00036324971070917934, 'samples': 3613056, 'steps': 18817, 'loss/train': 0.9477980732917786}
+01/27/2022 13:38:21 - INFO - codeparrot_training - Step 18818: {'lr': 0.00036323512315030067, 'samples': 3613248, 'steps': 18818, 'loss/train': 0.7393689751625061}
+01/27/2022 13:38:25 - INFO - codeparrot_training - Step 18819: {'lr': 0.0003632205351063587, 'samples': 3613440, 'steps': 18819, 'loss/train': 0.8390335142612457}
+01/27/2022 13:38:28 - INFO - codeparrot_training - Step 18820: {'lr': 0.0003632059465774159, 'samples': 3613632, 'steps': 18820, 'loss/train': 0.8245196342468262}
+01/27/2022 13:38:31 - INFO - codeparrot_training - Step 18821: {'lr': 0.0003631913575635348, 'samples': 3613824, 'steps': 18821, 'loss/train': 0.6637702435255051}
+01/27/2022 13:38:34 - INFO - codeparrot_training - Step 18822: {'lr': 0.00036317676806477784, 'samples': 3614016, 'steps': 18822, 'loss/train': 0.9189677238464355}
+01/27/2022 13:38:38 - INFO - codeparrot_training - Step 18823: {'lr': 0.0003631621780812075, 'samples': 3614208, 'steps': 18823, 'loss/train': 0.6329809427261353}
+01/27/2022 13:38:41 - INFO - codeparrot_training - Step 18824: {'lr': 0.00036314758761288643, 'samples': 3614400, 'steps': 18824, 'loss/train': 0.9567583501338959}
+01/27/2022 13:38:44 - INFO - codeparrot_training - Step 18825: {'lr': 0.0003631329966598769, 'samples': 3614592, 'steps': 18825, 'loss/train': 1.1656697988510132}
+01/27/2022 13:38:47 - INFO - codeparrot_training - Step 18826: {'lr': 0.0003631184052222416, 'samples': 3614784, 'steps': 18826, 'loss/train': 0.7824828028678894}
+01/27/2022 13:38:52 - INFO - codeparrot_training - Step 18827: {'lr': 0.00036310381330004296, 'samples': 3614976, 'steps': 18827, 'loss/train': 0.7137200385332108}
+01/27/2022 13:38:55 - INFO - codeparrot_training - Step 18828: {'lr': 0.0003630892208933435, 'samples': 3615168, 'steps': 18828, 'loss/train': 0.1095789335668087}
+01/27/2022 13:38:58 - INFO - codeparrot_training - Step 18829: {'lr': 0.00036307462800220575, 'samples': 3615360, 'steps': 18829, 'loss/train': 0.680888369679451}
+01/27/2022 13:39:01 - INFO - codeparrot_training - Step 18830: {'lr': 0.0003630600346266922, 'samples': 3615552, 'steps': 18830, 'loss/train': 0.5850623399019241}
+01/27/2022 13:39:05 - INFO - codeparrot_training - Step 18831: {'lr': 0.00036304544076686527, 'samples': 3615744, 'steps': 18831, 'loss/train': 1.2204163670539856}
+01/27/2022 13:39:08 - INFO - codeparrot_training - Step 18832: {'lr': 0.0003630308464227877, 'samples': 3615936, 'steps': 18832, 'loss/train': 0.881042093038559}
+01/27/2022 13:39:11 - INFO - codeparrot_training - Step 18833: {'lr': 0.0003630162515945218, 'samples': 3616128, 'steps': 18833, 'loss/train': 0.3298681899905205}
+01/27/2022 13:39:14 - INFO - codeparrot_training - Step 18834: {'lr': 0.00036300165628213015, 'samples': 3616320, 'steps': 18834, 'loss/train': 0.7369388490915298}
+01/27/2022 13:39:17 - INFO - codeparrot_training - Step 18835: {'lr': 0.0003629870604856754, 'samples': 3616512, 'steps': 18835, 'loss/train': 0.978228360414505}
+01/27/2022 13:39:20 - INFO - codeparrot_training - Step 18836: {'lr': 0.0003629724642052198, 'samples': 3616704, 'steps': 18836, 'loss/train': 0.09339878149330616}
+01/27/2022 13:39:25 - INFO - codeparrot_training - Step 18837: {'lr': 0.00036295786744082616, 'samples': 3616896, 'steps': 18837, 'loss/train': 0.40193402767181396}
+01/27/2022 13:39:28 - INFO - codeparrot_training - Step 18838: {'lr': 0.0003629432701925568, 'samples': 3617088, 'steps': 18838, 'loss/train': 0.832101970911026}
+01/27/2022 13:39:31 - INFO - codeparrot_training - Step 18839: {'lr': 0.0003629286724604744, 'samples': 3617280, 'steps': 18839, 'loss/train': 0.9168409109115601}
+01/27/2022 13:39:34 - INFO - codeparrot_training - Step 18840: {'lr': 0.0003629140742446414, 'samples': 3617472, 'steps': 18840, 'loss/train': 0.7681308388710022}
+01/27/2022 13:39:37 - INFO - codeparrot_training - Step 18841: {'lr': 0.00036289947554512034, 'samples': 3617664, 'steps': 18841, 'loss/train': 1.1409349143505096}
+01/27/2022 13:39:40 - INFO - codeparrot_training - Step 18842: {'lr': 0.0003628848763619738, 'samples': 3617856, 'steps': 18842, 'loss/train': 0.7533544600009918}
+01/27/2022 13:39:44 - INFO - codeparrot_training - Step 18843: {'lr': 0.0003628702766952643, 'samples': 3618048, 'steps': 18843, 'loss/train': 0.8874721527099609}
+01/27/2022 13:39:47 - INFO - codeparrot_training - Step 18844: {'lr': 0.00036285567654505433, 'samples': 3618240, 'steps': 18844, 'loss/train': 1.0935246348381042}
+01/27/2022 13:39:50 - INFO - codeparrot_training - Step 18845: {'lr': 0.00036284107591140653, 'samples': 3618432, 'steps': 18845, 'loss/train': 0.4806362986564636}
+01/27/2022 13:39:55 - INFO - codeparrot_training - Step 18846: {'lr': 0.0003628264747943834, 'samples': 3618624, 'steps': 18846, 'loss/train': 0.5903149992227554}
+01/27/2022 13:39:58 - INFO - codeparrot_training - Step 18847: {'lr': 0.0003628118731940475, 'samples': 3618816, 'steps': 18847, 'loss/train': 0.5096534639596939}
+01/27/2022 13:40:01 - INFO - codeparrot_training - Step 18848: {'lr': 0.00036279727111046127, 'samples': 3619008, 'steps': 18848, 'loss/train': 0.44203828275203705}
+01/27/2022 13:40:04 - INFO - codeparrot_training - Step 18849: {'lr': 0.0003627826685436874, 'samples': 3619200, 'steps': 18849, 'loss/train': 0.7903436422348022}
+01/27/2022 13:40:08 - INFO - codeparrot_training - Step 18850: {'lr': 0.00036276806549378836, 'samples': 3619392, 'steps': 18850, 'loss/train': 1.6324648261070251}
+01/27/2022 13:40:11 - INFO - codeparrot_training - Step 18851: {'lr': 0.0003627534619608268, 'samples': 3619584, 'steps': 18851, 'loss/train': 1.0682023465633392}
+01/27/2022 13:40:14 - INFO - codeparrot_training - Step 18852: {'lr': 0.00036273885794486514, 'samples': 3619776, 'steps': 18852, 'loss/train': 1.0202310383319855}
+01/27/2022 13:40:17 - INFO - codeparrot_training - Step 18853: {'lr': 0.00036272425344596607, 'samples': 3619968, 'steps': 18853, 'loss/train': 0.5887914597988129}
+01/27/2022 13:40:20 - INFO - codeparrot_training - Step 18854: {'lr': 0.000362709648464192, 'samples': 3620160, 'steps': 18854, 'loss/train': 0.939776211977005}
+01/27/2022 13:40:25 - INFO - codeparrot_training - Step 18855: {'lr': 0.00036269504299960573, 'samples': 3620352, 'steps': 18855, 'loss/train': 0.7116459310054779}
+01/27/2022 13:40:28 - INFO - codeparrot_training - Step 18856: {'lr': 0.00036268043705226953, 'samples': 3620544, 'steps': 18856, 'loss/train': 0.6280943602323532}
+01/27/2022 13:40:31 - INFO - codeparrot_training - Step 18857: {'lr': 0.00036266583062224625, 'samples': 3620736, 'steps': 18857, 'loss/train': 0.6962577849626541}
+01/27/2022 13:40:34 - INFO - codeparrot_training - Step 18858: {'lr': 0.0003626512237095982, 'samples': 3620928, 'steps': 18858, 'loss/train': 0.7823440432548523}
+01/27/2022 13:40:37 - INFO - codeparrot_training - Step 18859: {'lr': 0.00036263661631438814, 'samples': 3621120, 'steps': 18859, 'loss/train': 0.47869330644607544}
+01/27/2022 13:40:41 - INFO - codeparrot_training - Step 18860: {'lr': 0.0003626220084366786, 'samples': 3621312, 'steps': 18860, 'loss/train': 1.1611580550670624}
+01/27/2022 13:40:44 - INFO - codeparrot_training - Step 18861: {'lr': 0.00036260740007653216, 'samples': 3621504, 'steps': 18861, 'loss/train': 0.47757433354854584}
+01/27/2022 13:40:47 - INFO - codeparrot_training - Step 18862: {'lr': 0.00036259279123401127, 'samples': 3621696, 'steps': 18862, 'loss/train': 0.8478726446628571}
+01/27/2022 13:40:50 - INFO - codeparrot_training - Step 18863: {'lr': 0.0003625781819091787, 'samples': 3621888, 'steps': 18863, 'loss/train': 0.915984570980072}
+01/27/2022 13:40:55 - INFO - codeparrot_training - Step 18864: {'lr': 0.0003625635721020969, 'samples': 3622080, 'steps': 18864, 'loss/train': 0.7631275355815887}
+01/27/2022 13:40:59 - INFO - codeparrot_training - Step 18865: {'lr': 0.00036254896181282846, 'samples': 3622272, 'steps': 18865, 'loss/train': 0.8254813849925995}
+01/27/2022 13:41:02 - INFO - codeparrot_training - Step 18866: {'lr': 0.0003625343510414362, 'samples': 3622464, 'steps': 18866, 'loss/train': 0.2981877252459526}
+01/27/2022 13:41:05 - INFO - codeparrot_training - Step 18867: {'lr': 0.0003625197397879823, 'samples': 3622656, 'steps': 18867, 'loss/train': 0.7992415130138397}
+01/27/2022 13:41:08 - INFO - codeparrot_training - Step 18868: {'lr': 0.0003625051280525297, 'samples': 3622848, 'steps': 18868, 'loss/train': 0.7890004813671112}
+01/27/2022 13:41:11 - INFO - codeparrot_training - Step 18869: {'lr': 0.00036249051583514075, 'samples': 3623040, 'steps': 18869, 'loss/train': 0.30140114575624466}
+01/27/2022 13:41:14 - INFO - codeparrot_training - Step 18870: {'lr': 0.00036247590313587827, 'samples': 3623232, 'steps': 18870, 'loss/train': 0.8993377089500427}
+01/27/2022 13:41:17 - INFO - codeparrot_training - Step 18871: {'lr': 0.0003624612899548046, 'samples': 3623424, 'steps': 18871, 'loss/train': 0.7864416539669037}
+01/27/2022 13:41:21 - INFO - codeparrot_training - Step 18872: {'lr': 0.0003624466762919826, 'samples': 3623616, 'steps': 18872, 'loss/train': 0.17003965750336647}
+01/27/2022 13:41:26 - INFO - codeparrot_training - Step 18873: {'lr': 0.00036243206214747466, 'samples': 3623808, 'steps': 18873, 'loss/train': 0.8461900949478149}
+01/27/2022 13:41:29 - INFO - codeparrot_training - Step 18874: {'lr': 0.00036241744752134347, 'samples': 3624000, 'steps': 18874, 'loss/train': 0.7302502244710922}
+01/27/2022 13:41:32 - INFO - codeparrot_training - Step 18875: {'lr': 0.0003624028324136517, 'samples': 3624192, 'steps': 18875, 'loss/train': 0.6616382449865341}
+01/27/2022 13:41:35 - INFO - codeparrot_training - Step 18876: {'lr': 0.00036238821682446176, 'samples': 3624384, 'steps': 18876, 'loss/train': 0.5507943481206894}
+01/27/2022 13:41:39 - INFO - codeparrot_training - Step 18877: {'lr': 0.0003623736007538365, 'samples': 3624576, 'steps': 18877, 'loss/train': 1.0090661644935608}
+01/27/2022 13:41:42 - INFO - codeparrot_training - Step 18878: {'lr': 0.00036235898420183837, 'samples': 3624768, 'steps': 18878, 'loss/train': 1.3205365240573883}
+01/27/2022 13:41:45 - INFO - codeparrot_training - Step 18879: {'lr': 0.0003623443671685301, 'samples': 3624960, 'steps': 18879, 'loss/train': 0.8895893096923828}
+01/27/2022 13:41:48 - INFO - codeparrot_training - Step 18880: {'lr': 0.00036232974965397414, 'samples': 3625152, 'steps': 18880, 'loss/train': 0.820611834526062}
+01/27/2022 13:41:51 - INFO - codeparrot_training - Step 18881: {'lr': 0.00036231513165823324, 'samples': 3625344, 'steps': 18881, 'loss/train': 1.1017330884933472}
+01/27/2022 13:41:54 - INFO - codeparrot_training - Step 18882: {'lr': 0.00036230051318137, 'samples': 3625536, 'steps': 18882, 'loss/train': 1.0374796092510223}
+01/27/2022 13:41:59 - INFO - codeparrot_training - Step 18883: {'lr': 0.00036228589422344703, 'samples': 3625728, 'steps': 18883, 'loss/train': 1.045265793800354}
+01/27/2022 13:42:02 - INFO - codeparrot_training - Step 18884: {'lr': 0.0003622712747845269, 'samples': 3625920, 'steps': 18884, 'loss/train': 1.0883330404758453}
+01/27/2022 13:42:05 - INFO - codeparrot_training - Step 18885: {'lr': 0.0003622566548646723, 'samples': 3626112, 'steps': 18885, 'loss/train': 0.5596411675214767}
+01/27/2022 13:42:08 - INFO - codeparrot_training - Step 18886: {'lr': 0.00036224203446394584, 'samples': 3626304, 'steps': 18886, 'loss/train': 0.825402706861496}
+01/27/2022 13:42:11 - INFO - codeparrot_training - Step 18887: {'lr': 0.00036222741358241014, 'samples': 3626496, 'steps': 18887, 'loss/train': 1.164488822221756}
+01/27/2022 13:42:15 - INFO - codeparrot_training - Step 18888: {'lr': 0.0003622127922201278, 'samples': 3626688, 'steps': 18888, 'loss/train': 0.663980171084404}
+01/27/2022 13:42:18 - INFO - codeparrot_training - Step 18889: {'lr': 0.0003621981703771616, 'samples': 3626880, 'steps': 18889, 'loss/train': 1.1918239295482635}
+01/27/2022 13:42:21 - INFO - codeparrot_training - Step 18890: {'lr': 0.0003621835480535739, 'samples': 3627072, 'steps': 18890, 'loss/train': 0.7366197109222412}
+01/27/2022 13:42:24 - INFO - codeparrot_training - Step 18891: {'lr': 0.00036216892524942764, 'samples': 3627264, 'steps': 18891, 'loss/train': 0.6760221272706985}
+01/27/2022 13:42:30 - INFO - codeparrot_training - Step 18892: {'lr': 0.00036215430196478526, 'samples': 3627456, 'steps': 18892, 'loss/train': 0.7292076498270035}
+01/27/2022 13:42:33 - INFO - codeparrot_training - Step 18893: {'lr': 0.0003621396781997095, 'samples': 3627648, 'steps': 18893, 'loss/train': 0.3497614413499832}
+01/27/2022 13:42:36 - INFO - codeparrot_training - Step 18894: {'lr': 0.00036212505395426297, 'samples': 3627840, 'steps': 18894, 'loss/train': 0.41707755625247955}
+01/27/2022 13:42:39 - INFO - codeparrot_training - Step 18895: {'lr': 0.0003621104292285082, 'samples': 3628032, 'steps': 18895, 'loss/train': 0.9399104118347168}
+01/27/2022 13:42:42 - INFO - codeparrot_training - Step 18896: {'lr': 0.0003620958040225081, 'samples': 3628224, 'steps': 18896, 'loss/train': 1.2782314717769623}
+01/27/2022 13:42:45 - INFO - codeparrot_training - Step 18897: {'lr': 0.0003620811783363251, 'samples': 3628416, 'steps': 18897, 'loss/train': 1.1344084739685059}
+01/27/2022 13:42:49 - INFO - codeparrot_training - Step 18898: {'lr': 0.0003620665521700219, 'samples': 3628608, 'steps': 18898, 'loss/train': 0.7369822114706039}
+01/27/2022 13:42:52 - INFO - codeparrot_training - Step 18899: {'lr': 0.00036205192552366124, 'samples': 3628800, 'steps': 18899, 'loss/train': 0.7895850241184235}
+01/27/2022 13:42:55 - INFO - codeparrot_training - Step 18900: {'lr': 0.0003620372983973057, 'samples': 3628992, 'steps': 18900, 'loss/train': 0.3569754436612129}
+01/27/2022 13:42:59 - INFO - codeparrot_training - Step 18901: {'lr': 0.00036202267079101793, 'samples': 3629184, 'steps': 18901, 'loss/train': 0.5586162507534027}
+01/27/2022 13:43:02 - INFO - codeparrot_training - Step 18902: {'lr': 0.0003620080427048605, 'samples': 3629376, 'steps': 18902, 'loss/train': 2.2975423336029053}
+01/27/2022 13:43:06 - INFO - codeparrot_training - Step 18903: {'lr': 0.00036199341413889637, 'samples': 3629568, 'steps': 18903, 'loss/train': 0.9832028746604919}
+01/27/2022 13:43:09 - INFO - codeparrot_training - Step 18904: {'lr': 0.00036197878509318794, 'samples': 3629760, 'steps': 18904, 'loss/train': 0.6668833047151566}
+01/27/2022 13:43:12 - INFO - codeparrot_training - Step 18905: {'lr': 0.00036196415556779795, 'samples': 3629952, 'steps': 18905, 'loss/train': 0.9059524834156036}
+01/27/2022 13:43:15 - INFO - codeparrot_training - Step 18906: {'lr': 0.00036194952556278906, 'samples': 3630144, 'steps': 18906, 'loss/train': 0.8595348000526428}
+01/27/2022 13:43:18 - INFO - codeparrot_training - Step 18907: {'lr': 0.00036193489507822395, 'samples': 3630336, 'steps': 18907, 'loss/train': 0.8740520775318146}
+01/27/2022 13:43:21 - INFO - codeparrot_training - Step 18908: {'lr': 0.0003619202641141652, 'samples': 3630528, 'steps': 18908, 'loss/train': 0.5755103230476379}
+01/27/2022 13:43:24 - INFO - codeparrot_training - Step 18909: {'lr': 0.00036190563267067576, 'samples': 3630720, 'steps': 18909, 'loss/train': 0.7722661793231964}
+01/27/2022 13:43:29 - INFO - codeparrot_training - Step 18910: {'lr': 0.000361891000747818, 'samples': 3630912, 'steps': 18910, 'loss/train': 0.9319781363010406}
+01/27/2022 13:43:32 - INFO - codeparrot_training - Step 18911: {'lr': 0.0003618763683456548, 'samples': 3631104, 'steps': 18911, 'loss/train': 0.6650138944387436}
+01/27/2022 13:43:35 - INFO - codeparrot_training - Step 18912: {'lr': 0.00036186173546424874, 'samples': 3631296, 'steps': 18912, 'loss/train': 0.7348363995552063}
+01/27/2022 13:43:38 - INFO - codeparrot_training - Step 18913: {'lr': 0.00036184710210366243, 'samples': 3631488, 'steps': 18913, 'loss/train': 0.9715542197227478}
+01/27/2022 13:43:41 - INFO - codeparrot_training - Step 18914: {'lr': 0.0003618324682639588, 'samples': 3631680, 'steps': 18914, 'loss/train': 0.46541498601436615}
+01/27/2022 13:43:44 - INFO - codeparrot_training - Step 18915: {'lr': 0.0003618178339452003, 'samples': 3631872, 'steps': 18915, 'loss/train': 0.9175418615341187}
+01/27/2022 13:43:48 - INFO - codeparrot_training - Step 18916: {'lr': 0.00036180319914744973, 'samples': 3632064, 'steps': 18916, 'loss/train': 1.0600241124629974}
+01/27/2022 13:43:51 - INFO - codeparrot_training - Step 18917: {'lr': 0.00036178856387076967, 'samples': 3632256, 'steps': 18917, 'loss/train': 0.8647804856300354}
+01/27/2022 13:43:56 - INFO - codeparrot_training - Step 18918: {'lr': 0.00036177392811522304, 'samples': 3632448, 'steps': 18918, 'loss/train': 0.8574380278587341}
+01/27/2022 13:43:59 - INFO - codeparrot_training - Step 18919: {'lr': 0.0003617592918808723, 'samples': 3632640, 'steps': 18919, 'loss/train': 0.8900513648986816}
+01/27/2022 13:44:02 - INFO - codeparrot_training - Step 18920: {'lr': 0.0003617446551677803, 'samples': 3632832, 'steps': 18920, 'loss/train': 1.0787131190299988}
+01/27/2022 13:44:05 - INFO - codeparrot_training - Step 18921: {'lr': 0.0003617300179760096, 'samples': 3633024, 'steps': 18921, 'loss/train': 0.9045417308807373}
+01/27/2022 13:44:08 - INFO - codeparrot_training - Step 18922: {'lr': 0.0003617153803056231, 'samples': 3633216, 'steps': 18922, 'loss/train': 0.9795723259449005}
+01/27/2022 13:44:12 - INFO - codeparrot_training - Step 18923: {'lr': 0.0003617007421566833, 'samples': 3633408, 'steps': 18923, 'loss/train': 0.9649868309497833}
+01/27/2022 13:44:15 - INFO - codeparrot_training - Step 18924: {'lr': 0.00036168610352925294, 'samples': 3633600, 'steps': 18924, 'loss/train': 0.7029768973588943}
+01/27/2022 13:44:18 - INFO - codeparrot_training - Step 18925: {'lr': 0.00036167146442339486, 'samples': 3633792, 'steps': 18925, 'loss/train': 0.8442369103431702}
+01/27/2022 13:44:21 - INFO - codeparrot_training - Step 18926: {'lr': 0.00036165682483917165, 'samples': 3633984, 'steps': 18926, 'loss/train': 0.8422721028327942}
+01/27/2022 13:44:25 - INFO - codeparrot_training - Step 18927: {'lr': 0.0003616421847766461, 'samples': 3634176, 'steps': 18927, 'loss/train': 1.0568844079971313}
+01/27/2022 13:44:28 - INFO - codeparrot_training - Step 18928: {'lr': 0.00036162754423588086, 'samples': 3634368, 'steps': 18928, 'loss/train': 0.8323201239109039}
+01/27/2022 13:44:32 - INFO - codeparrot_training - Step 18929: {'lr': 0.0003616129032169387, 'samples': 3634560, 'steps': 18929, 'loss/train': 0.9011093974113464}
+01/27/2022 13:44:35 - INFO - codeparrot_training - Step 18930: {'lr': 0.00036159826171988223, 'samples': 3634752, 'steps': 18930, 'loss/train': 0.6811513602733612}
+01/27/2022 13:44:38 - INFO - codeparrot_training - Step 18931: {'lr': 0.00036158361974477434, 'samples': 3634944, 'steps': 18931, 'loss/train': 0.8369676768779755}
+01/27/2022 13:44:41 - INFO - codeparrot_training - Step 18932: {'lr': 0.0003615689772916776, 'samples': 3635136, 'steps': 18932, 'loss/train': 1.076085090637207}
+01/27/2022 13:44:44 - INFO - codeparrot_training - Step 18933: {'lr': 0.00036155433436065477, 'samples': 3635328, 'steps': 18933, 'loss/train': 1.0611059367656708}
+01/27/2022 13:44:47 - INFO - codeparrot_training - Step 18934: {'lr': 0.0003615396909517686, 'samples': 3635520, 'steps': 18934, 'loss/train': 0.7097843885421753}
+01/27/2022 13:44:50 - INFO - codeparrot_training - Step 18935: {'lr': 0.00036152504706508183, 'samples': 3635712, 'steps': 18935, 'loss/train': 1.059912621974945}
+01/27/2022 13:44:55 - INFO - codeparrot_training - Step 18936: {'lr': 0.00036151040270065716, 'samples': 3635904, 'steps': 18936, 'loss/train': 0.47979260981082916}
+01/27/2022 13:44:58 - INFO - codeparrot_training - Step 18937: {'lr': 0.00036149575785855736, 'samples': 3636096, 'steps': 18937, 'loss/train': 0.9593153893947601}
+01/27/2022 13:45:01 - INFO - codeparrot_training - Step 18938: {'lr': 0.00036148111253884514, 'samples': 3636288, 'steps': 18938, 'loss/train': 1.1108843386173248}
+01/27/2022 13:45:04 - INFO - codeparrot_training - Step 18939: {'lr': 0.00036146646674158317, 'samples': 3636480, 'steps': 18939, 'loss/train': 0.7289961129426956}
+01/27/2022 13:45:07 - INFO - codeparrot_training - Step 18940: {'lr': 0.00036145182046683433, 'samples': 3636672, 'steps': 18940, 'loss/train': 0.7776516973972321}
+01/27/2022 13:45:11 - INFO - codeparrot_training - Step 18941: {'lr': 0.0003614371737146612, 'samples': 3636864, 'steps': 18941, 'loss/train': 1.096402645111084}
+01/27/2022 13:45:14 - INFO - codeparrot_training - Step 18942: {'lr': 0.0003614225264851266, 'samples': 3637056, 'steps': 18942, 'loss/train': 0.8235321342945099}
+01/27/2022 13:45:17 - INFO - codeparrot_training - Step 18943: {'lr': 0.00036140787877829336, 'samples': 3637248, 'steps': 18943, 'loss/train': 0.9459818601608276}
+01/27/2022 13:45:20 - INFO - codeparrot_training - Step 18944: {'lr': 0.0003613932305942241, 'samples': 3637440, 'steps': 18944, 'loss/train': 0.3675884082913399}
+01/27/2022 13:45:25 - INFO - codeparrot_training - Step 18945: {'lr': 0.0003613785819329816, 'samples': 3637632, 'steps': 18945, 'loss/train': 0.5106977820396423}
+01/27/2022 13:45:28 - INFO - codeparrot_training - Step 18946: {'lr': 0.00036136393279462866, 'samples': 3637824, 'steps': 18946, 'loss/train': 0.9714540839195251}
+01/27/2022 13:45:31 - INFO - codeparrot_training - Step 18947: {'lr': 0.00036134928317922796, 'samples': 3638016, 'steps': 18947, 'loss/train': 0.8406849503517151}
+01/27/2022 13:45:34 - INFO - codeparrot_training - Step 18948: {'lr': 0.0003613346330868423, 'samples': 3638208, 'steps': 18948, 'loss/train': 0.9906325936317444}
+01/27/2022 13:45:37 - INFO - codeparrot_training - Step 18949: {'lr': 0.0003613199825175344, 'samples': 3638400, 'steps': 18949, 'loss/train': 0.8849116265773773}
+01/27/2022 13:45:40 - INFO - codeparrot_training - Step 18950: {'lr': 0.0003613053314713671, 'samples': 3638592, 'steps': 18950, 'loss/train': 0.5645587742328644}
+01/27/2022 13:45:43 - INFO - codeparrot_training - Step 18951: {'lr': 0.00036129067994840303, 'samples': 3638784, 'steps': 18951, 'loss/train': 0.6950496733188629}
+01/27/2022 13:45:47 - INFO - codeparrot_training - Step 18952: {'lr': 0.0003612760279487051, 'samples': 3638976, 'steps': 18952, 'loss/train': 0.6127288788557053}
+01/27/2022 13:45:50 - INFO - codeparrot_training - Step 18953: {'lr': 0.00036126137547233593, 'samples': 3639168, 'steps': 18953, 'loss/train': 0.7855439186096191}
+01/27/2022 13:45:55 - INFO - codeparrot_training - Step 18954: {'lr': 0.00036124672251935843, 'samples': 3639360, 'steps': 18954, 'loss/train': 0.9309447705745697}
+01/27/2022 13:45:58 - INFO - codeparrot_training - Step 18955: {'lr': 0.00036123206908983527, 'samples': 3639552, 'steps': 18955, 'loss/train': 0.7673503160476685}
+01/27/2022 13:46:01 - INFO - codeparrot_training - Step 18956: {'lr': 0.00036121741518382915, 'samples': 3639744, 'steps': 18956, 'loss/train': 1.1198194921016693}
+01/27/2022 13:46:04 - INFO - codeparrot_training - Step 18957: {'lr': 0.0003612027608014031, 'samples': 3639936, 'steps': 18957, 'loss/train': 1.1756084561347961}
+01/27/2022 13:46:07 - INFO - codeparrot_training - Step 18958: {'lr': 0.0003611881059426196, 'samples': 3640128, 'steps': 18958, 'loss/train': 0.609635666012764}
+01/27/2022 13:46:11 - INFO - codeparrot_training - Step 18959: {'lr': 0.00036117345060754163, 'samples': 3640320, 'steps': 18959, 'loss/train': 0.9670417606830597}
+01/27/2022 13:46:14 - INFO - codeparrot_training - Step 18960: {'lr': 0.00036115879479623185, 'samples': 3640512, 'steps': 18960, 'loss/train': 0.428711861371994}
+01/27/2022 13:46:17 - INFO - codeparrot_training - Step 18961: {'lr': 0.00036114413850875317, 'samples': 3640704, 'steps': 18961, 'loss/train': 0.8494731187820435}
+01/27/2022 13:46:21 - INFO - codeparrot_training - Step 18962: {'lr': 0.00036112948174516824, 'samples': 3640896, 'steps': 18962, 'loss/train': 0.9791832268238068}
+01/27/2022 13:46:25 - INFO - codeparrot_training - Step 18963: {'lr': 0.0003611148245055399, 'samples': 3641088, 'steps': 18963, 'loss/train': 0.8136836886405945}
+01/27/2022 13:46:28 - INFO - codeparrot_training - Step 18964: {'lr': 0.000361100166789931, 'samples': 3641280, 'steps': 18964, 'loss/train': 0.5980373471975327}
+01/27/2022 13:46:31 - INFO - codeparrot_training - Step 18965: {'lr': 0.0003610855085984041, 'samples': 3641472, 'steps': 18965, 'loss/train': 0.989222377538681}
+01/27/2022 13:46:34 - INFO - codeparrot_training - Step 18966: {'lr': 0.0003610708499310223, 'samples': 3641664, 'steps': 18966, 'loss/train': 0.7451415359973907}
+01/27/2022 13:46:37 - INFO - codeparrot_training - Step 18967: {'lr': 0.0003610561907878482, 'samples': 3641856, 'steps': 18967, 'loss/train': 0.5936509519815445}
+01/27/2022 13:46:40 - INFO - codeparrot_training - Step 18968: {'lr': 0.00036104153116894465, 'samples': 3642048, 'steps': 18968, 'loss/train': 0.5204677730798721}
+01/27/2022 13:46:44 - INFO - codeparrot_training - Step 18969: {'lr': 0.00036102687107437444, 'samples': 3642240, 'steps': 18969, 'loss/train': 0.5699532777070999}
+01/27/2022 13:46:47 - INFO - codeparrot_training - Step 18970: {'lr': 0.0003610122105042004, 'samples': 3642432, 'steps': 18970, 'loss/train': 0.9158046841621399}
+01/27/2022 13:46:52 - INFO - codeparrot_training - Step 18971: {'lr': 0.0003609975494584852, 'samples': 3642624, 'steps': 18971, 'loss/train': 0.8137737214565277}
+01/27/2022 13:46:55 - INFO - codeparrot_training - Step 18972: {'lr': 0.0003609828879372918, 'samples': 3642816, 'steps': 18972, 'loss/train': 0.7449458241462708}
+01/27/2022 13:46:59 - INFO - codeparrot_training - Step 18973: {'lr': 0.000360968225940683, 'samples': 3643008, 'steps': 18973, 'loss/train': 0.19324050843715668}
+01/27/2022 13:47:02 - INFO - codeparrot_training - Step 18974: {'lr': 0.00036095356346872156, 'samples': 3643200, 'steps': 18974, 'loss/train': 0.5550824850797653}
+01/27/2022 13:47:05 - INFO - codeparrot_training - Step 18975: {'lr': 0.0003609389005214702, 'samples': 3643392, 'steps': 18975, 'loss/train': 0.43701164424419403}
+01/27/2022 13:47:08 - INFO - codeparrot_training - Step 18976: {'lr': 0.0003609242370989919, 'samples': 3643584, 'steps': 18976, 'loss/train': 0.7050240486860275}
+01/27/2022 13:47:11 - INFO - codeparrot_training - Step 18977: {'lr': 0.0003609095732013493, 'samples': 3643776, 'steps': 18977, 'loss/train': 0.9056119322776794}
+01/27/2022 13:47:14 - INFO - codeparrot_training - Step 18978: {'lr': 0.0003608949088286054, 'samples': 3643968, 'steps': 18978, 'loss/train': 2.31142258644104}
+01/27/2022 13:47:17 - INFO - codeparrot_training - Step 18979: {'lr': 0.00036088024398082285, 'samples': 3644160, 'steps': 18979, 'loss/train': 0.725172758102417}
+01/27/2022 13:47:20 - INFO - codeparrot_training - Step 18980: {'lr': 0.00036086557865806464, 'samples': 3644352, 'steps': 18980, 'loss/train': 0.04154461808502674}
+01/27/2022 13:47:25 - INFO - codeparrot_training - Step 18981: {'lr': 0.0003608509128603934, 'samples': 3644544, 'steps': 18981, 'loss/train': 1.191240280866623}
+01/27/2022 13:47:28 - INFO - codeparrot_training - Step 18982: {'lr': 0.0003608362465878721, 'samples': 3644736, 'steps': 18982, 'loss/train': 0.9917411506175995}
+01/27/2022 13:47:31 - INFO - codeparrot_training - Step 18983: {'lr': 0.0003608215798405635, 'samples': 3644928, 'steps': 18983, 'loss/train': 0.9173887073993683}
+01/27/2022 13:47:34 - INFO - codeparrot_training - Step 18984: {'lr': 0.0003608069126185304, 'samples': 3645120, 'steps': 18984, 'loss/train': 0.6697044521570206}
+01/27/2022 13:47:37 - INFO - codeparrot_training - Step 18985: {'lr': 0.0003607922449218357, 'samples': 3645312, 'steps': 18985, 'loss/train': 0.7588180303573608}
+01/27/2022 13:47:41 - INFO - codeparrot_training - Step 18986: {'lr': 0.00036077757675054224, 'samples': 3645504, 'steps': 18986, 'loss/train': 0.7266744375228882}
+01/27/2022 13:47:44 - INFO - codeparrot_training - Step 18987: {'lr': 0.0003607629081047128, 'samples': 3645696, 'steps': 18987, 'loss/train': 0.8560867309570312}
+01/27/2022 13:47:47 - INFO - codeparrot_training - Step 18988: {'lr': 0.0003607482389844102, 'samples': 3645888, 'steps': 18988, 'loss/train': 0.6191274225711823}
+01/27/2022 13:47:50 - INFO - codeparrot_training - Step 18989: {'lr': 0.0003607335693896973, 'samples': 3646080, 'steps': 18989, 'loss/train': 0.6858907341957092}
+01/27/2022 13:47:54 - INFO - codeparrot_training - Step 18990: {'lr': 0.00036071889932063706, 'samples': 3646272, 'steps': 18990, 'loss/train': 0.49461235105991364}
+01/27/2022 13:47:58 - INFO - codeparrot_training - Step 18991: {'lr': 0.0003607042287772921, 'samples': 3646464, 'steps': 18991, 'loss/train': 0.6854437440633774}
+01/27/2022 13:48:01 - INFO - codeparrot_training - Step 18992: {'lr': 0.0003606895577597254, 'samples': 3646656, 'steps': 18992, 'loss/train': 0.9008650481700897}
+01/27/2022 13:48:04 - INFO - codeparrot_training - Step 18993: {'lr': 0.0003606748862679998, 'samples': 3646848, 'steps': 18993, 'loss/train': 1.001380294561386}
+01/27/2022 13:48:07 - INFO - codeparrot_training - Step 18994: {'lr': 0.00036066021430217817, 'samples': 3647040, 'steps': 18994, 'loss/train': 0.5234911143779755}
+01/27/2022 13:48:10 - INFO - codeparrot_training - Step 18995: {'lr': 0.00036064554186232323, 'samples': 3647232, 'steps': 18995, 'loss/train': 1.3328780829906464}
+01/27/2022 13:48:13 - INFO - codeparrot_training - Step 18996: {'lr': 0.0003606308689484979, 'samples': 3647424, 'steps': 18996, 'loss/train': 0.8555132746696472}
+01/27/2022 13:48:16 - INFO - codeparrot_training - Step 18997: {'lr': 0.0003606161955607652, 'samples': 3647616, 'steps': 18997, 'loss/train': 0.9142524003982544}
+01/27/2022 13:48:22 - INFO - codeparrot_training - Step 18998: {'lr': 0.0003606015216991877, 'samples': 3647808, 'steps': 18998, 'loss/train': 0.4983205050230026}
+01/27/2022 13:48:25 - INFO - codeparrot_training - Step 18999: {'lr': 0.0003605868473638285, 'samples': 3648000, 'steps': 18999, 'loss/train': 0.4881068766117096}
+01/27/2022 13:48:28 - INFO - codeparrot_training - Step 19000: {'lr': 0.00036057217255475036, 'samples': 3648192, 'steps': 19000, 'loss/train': 0.0435930322855711}
+01/27/2022 13:48:31 - INFO - codeparrot_training - Step 19001: {'lr': 0.0003605574972720161, 'samples': 3648384, 'steps': 19001, 'loss/train': 0.39314237236976624}
+01/27/2022 13:48:34 - INFO - codeparrot_training - Step 19002: {'lr': 0.00036054282151568864, 'samples': 3648576, 'steps': 19002, 'loss/train': 1.0290627479553223}
+01/27/2022 13:48:38 - INFO - codeparrot_training - Step 19003: {'lr': 0.0003605281452858308, 'samples': 3648768, 'steps': 19003, 'loss/train': 0.8745445311069489}
+01/27/2022 13:48:41 - INFO - codeparrot_training - Step 19004: {'lr': 0.00036051346858250556, 'samples': 3648960, 'steps': 19004, 'loss/train': 0.8420693278312683}
+01/27/2022 13:48:44 - INFO - codeparrot_training - Step 19005: {'lr': 0.00036049879140577566, 'samples': 3649152, 'steps': 19005, 'loss/train': 0.863182783126831}
+01/27/2022 13:48:47 - INFO - codeparrot_training - Step 19006: {'lr': 0.000360484113755704, 'samples': 3649344, 'steps': 19006, 'loss/train': 0.7923385798931122}
+01/27/2022 13:48:51 - INFO - codeparrot_training - Step 19007: {'lr': 0.0003604694356323536, 'samples': 3649536, 'steps': 19007, 'loss/train': 0.7654018700122833}
+01/27/2022 13:48:55 - INFO - codeparrot_training - Step 19008: {'lr': 0.00036045475703578705, 'samples': 3649728, 'steps': 19008, 'loss/train': 0.7069433033466339}
+01/27/2022 13:48:58 - INFO - codeparrot_training - Step 19009: {'lr': 0.0003604400779660674, 'samples': 3649920, 'steps': 19009, 'loss/train': 0.45912525057792664}
+01/27/2022 13:49:01 - INFO - codeparrot_training - Step 19010: {'lr': 0.00036042539842325765, 'samples': 3650112, 'steps': 19010, 'loss/train': 0.9914131164550781}
+01/27/2022 13:49:04 - INFO - codeparrot_training - Step 19011: {'lr': 0.0003604107184074205, 'samples': 3650304, 'steps': 19011, 'loss/train': 0.5886460840702057}
+01/27/2022 13:49:07 - INFO - codeparrot_training - Step 19012: {'lr': 0.0003603960379186189, 'samples': 3650496, 'steps': 19012, 'loss/train': 1.14814093708992}
+01/27/2022 13:49:10 - INFO - codeparrot_training - Step 19013: {'lr': 0.0003603813569569157, 'samples': 3650688, 'steps': 19013, 'loss/train': 1.1260447204113007}
+01/27/2022 13:49:13 - INFO - codeparrot_training - Step 19014: {'lr': 0.0003603666755223739, 'samples': 3650880, 'steps': 19014, 'loss/train': 0.7492021322250366}
+01/27/2022 13:49:17 - INFO - codeparrot_training - Step 19015: {'lr': 0.0003603519936150562, 'samples': 3651072, 'steps': 19015, 'loss/train': 0.8979611098766327}
+01/27/2022 13:49:21 - INFO - codeparrot_training - Step 19016: {'lr': 0.00036033731123502567, 'samples': 3651264, 'steps': 19016, 'loss/train': 1.0122809708118439}
+01/27/2022 13:49:24 - INFO - codeparrot_training - Step 19017: {'lr': 0.00036032262838234507, 'samples': 3651456, 'steps': 19017, 'loss/train': 0.8825485110282898}
+01/27/2022 13:49:27 - INFO - codeparrot_training - Step 19018: {'lr': 0.0003603079450570775, 'samples': 3651648, 'steps': 19018, 'loss/train': 1.0657289922237396}
+01/27/2022 13:49:30 - INFO - codeparrot_training - Step 19019: {'lr': 0.00036029326125928556, 'samples': 3651840, 'steps': 19019, 'loss/train': 1.4763169884681702}
+01/27/2022 13:49:34 - INFO - codeparrot_training - Step 19020: {'lr': 0.00036027857698903235, 'samples': 3652032, 'steps': 19020, 'loss/train': 1.1911181509494781}
+01/27/2022 13:49:37 - INFO - codeparrot_training - Step 19021: {'lr': 0.00036026389224638077, 'samples': 3652224, 'steps': 19021, 'loss/train': 1.02083221077919}
+01/27/2022 13:49:40 - INFO - codeparrot_training - Step 19022: {'lr': 0.00036024920703139375, 'samples': 3652416, 'steps': 19022, 'loss/train': 0.09290242753922939}
+01/27/2022 13:49:43 - INFO - codeparrot_training - Step 19023: {'lr': 0.00036023452134413406, 'samples': 3652608, 'steps': 19023, 'loss/train': 1.0322981178760529}
+01/27/2022 13:49:46 - INFO - codeparrot_training - Step 19024: {'lr': 0.00036021983518466465, 'samples': 3652800, 'steps': 19024, 'loss/train': 0.6855015009641647}
+01/27/2022 13:49:51 - INFO - codeparrot_training - Step 19025: {'lr': 0.00036020514855304855, 'samples': 3652992, 'steps': 19025, 'loss/train': 0.4881555140018463}
+01/27/2022 13:49:55 - INFO - codeparrot_training - Step 19026: {'lr': 0.00036019046144934855, 'samples': 3653184, 'steps': 19026, 'loss/train': 1.2355031669139862}
+01/27/2022 13:49:58 - INFO - codeparrot_training - Step 19027: {'lr': 0.0003601757738736275, 'samples': 3653376, 'steps': 19027, 'loss/train': 0.7823030054569244}
+01/27/2022 13:50:01 - INFO - codeparrot_training - Step 19028: {'lr': 0.0003601610858259485, 'samples': 3653568, 'steps': 19028, 'loss/train': 0.9367031157016754}
+01/27/2022 13:50:04 - INFO - codeparrot_training - Step 19029: {'lr': 0.0003601463973063745, 'samples': 3653760, 'steps': 19029, 'loss/train': 0.7031680047512054}
+01/27/2022 13:50:07 - INFO - codeparrot_training - Step 19030: {'lr': 0.0003601317083149682, 'samples': 3653952, 'steps': 19030, 'loss/train': 0.9601144194602966}
+01/27/2022 13:50:10 - INFO - codeparrot_training - Step 19031: {'lr': 0.0003601170188517926, 'samples': 3654144, 'steps': 19031, 'loss/train': 0.7337204664945602}
+01/27/2022 13:50:13 - INFO - codeparrot_training - Step 19032: {'lr': 0.0003601023289169107, 'samples': 3654336, 'steps': 19032, 'loss/train': 0.8907444477081299}
+01/27/2022 13:50:18 - INFO - codeparrot_training - Step 19033: {'lr': 0.00036008763851038534, 'samples': 3654528, 'steps': 19033, 'loss/train': 1.5230421423912048}
+01/27/2022 13:50:21 - INFO - codeparrot_training - Step 19034: {'lr': 0.0003600729476322796, 'samples': 3654720, 'steps': 19034, 'loss/train': 1.2411752343177795}
+01/27/2022 13:50:24 - INFO - codeparrot_training - Step 19035: {'lr': 0.0003600582562826562, 'samples': 3654912, 'steps': 19035, 'loss/train': 0.2017224356532097}
+01/27/2022 13:50:27 - INFO - codeparrot_training - Step 19036: {'lr': 0.0003600435644615783, 'samples': 3655104, 'steps': 19036, 'loss/train': 0.534413605928421}
+01/27/2022 13:50:30 - INFO - codeparrot_training - Step 19037: {'lr': 0.0003600288721691085, 'samples': 3655296, 'steps': 19037, 'loss/train': 0.5584934055805206}
+01/27/2022 13:50:34 - INFO - codeparrot_training - Step 19038: {'lr': 0.0003600141794053102, 'samples': 3655488, 'steps': 19038, 'loss/train': 0.5955703854560852}
+01/27/2022 13:50:37 - INFO - codeparrot_training - Step 19039: {'lr': 0.00035999948617024594, 'samples': 3655680, 'steps': 19039, 'loss/train': 0.8366945385932922}
+01/27/2022 13:50:40 - INFO - codeparrot_training - Step 19040: {'lr': 0.0003599847924639788, 'samples': 3655872, 'steps': 19040, 'loss/train': 0.7094031572341919}
+01/27/2022 13:50:43 - INFO - codeparrot_training - Step 19041: {'lr': 0.0003599700982865718, 'samples': 3656064, 'steps': 19041, 'loss/train': 0.9101461172103882}
+01/27/2022 13:50:47 - INFO - codeparrot_training - Step 19042: {'lr': 0.00035995540363808776, 'samples': 3656256, 'steps': 19042, 'loss/train': 0.7267244160175323}
+01/27/2022 13:50:51 - INFO - codeparrot_training - Step 19043: {'lr': 0.0003599407085185897, 'samples': 3656448, 'steps': 19043, 'loss/train': 0.6993515342473984}
+01/27/2022 13:50:54 - INFO - codeparrot_training - Step 19044: {'lr': 0.00035992601292814065, 'samples': 3656640, 'steps': 19044, 'loss/train': 1.615009367465973}
+01/27/2022 13:50:57 - INFO - codeparrot_training - Step 19045: {'lr': 0.00035991131686680333, 'samples': 3656832, 'steps': 19045, 'loss/train': 0.5625324994325638}
+01/27/2022 13:51:00 - INFO - codeparrot_training - Step 19046: {'lr': 0.00035989662033464087, 'samples': 3657024, 'steps': 19046, 'loss/train': 0.8538473546504974}
+01/27/2022 13:51:03 - INFO - codeparrot_training - Step 19047: {'lr': 0.0003598819233317162, 'samples': 3657216, 'steps': 19047, 'loss/train': 0.7711475193500519}
+01/27/2022 13:51:06 - INFO - codeparrot_training - Step 19048: {'lr': 0.0003598672258580923, 'samples': 3657408, 'steps': 19048, 'loss/train': 0.745492234826088}
+01/27/2022 13:51:09 - INFO - codeparrot_training - Step 19049: {'lr': 0.00035985252791383203, 'samples': 3657600, 'steps': 19049, 'loss/train': 0.8675601482391357}
+01/27/2022 13:51:13 - INFO - codeparrot_training - Step 19050: {'lr': 0.0003598378294989984, 'samples': 3657792, 'steps': 19050, 'loss/train': 0.928381472826004}
+01/27/2022 13:51:17 - INFO - codeparrot_training - Step 19051: {'lr': 0.00035982313061365444, 'samples': 3657984, 'steps': 19051, 'loss/train': 0.837149441242218}
+01/27/2022 13:51:20 - INFO - codeparrot_training - Step 19052: {'lr': 0.00035980843125786306, 'samples': 3658176, 'steps': 19052, 'loss/train': 0.7052355855703354}
+01/27/2022 13:51:23 - INFO - codeparrot_training - Step 19053: {'lr': 0.00035979373143168726, 'samples': 3658368, 'steps': 19053, 'loss/train': 0.8489227294921875}
+01/27/2022 13:51:26 - INFO - codeparrot_training - Step 19054: {'lr': 0.0003597790311351898, 'samples': 3658560, 'steps': 19054, 'loss/train': 0.7496937811374664}
+01/27/2022 13:51:30 - INFO - codeparrot_training - Step 19055: {'lr': 0.00035976433036843405, 'samples': 3658752, 'steps': 19055, 'loss/train': 0.8982219099998474}
+01/27/2022 13:51:33 - INFO - codeparrot_training - Step 19056: {'lr': 0.00035974962913148267, 'samples': 3658944, 'steps': 19056, 'loss/train': 0.6266970634460449}
+01/27/2022 13:51:36 - INFO - codeparrot_training - Step 19057: {'lr': 0.0003597349274243987, 'samples': 3659136, 'steps': 19057, 'loss/train': 0.7434812933206558}
+01/27/2022 13:51:39 - INFO - codeparrot_training - Step 19058: {'lr': 0.0003597202252472452, 'samples': 3659328, 'steps': 19058, 'loss/train': 1.1896097660064697}
+01/27/2022 13:51:42 - INFO - codeparrot_training - Step 19059: {'lr': 0.00035970552260008505, 'samples': 3659520, 'steps': 19059, 'loss/train': 0.5243093222379684}
+01/27/2022 13:51:47 - INFO - codeparrot_training - Step 19060: {'lr': 0.00035969081948298133, 'samples': 3659712, 'steps': 19060, 'loss/train': 0.4153429716825485}
+01/27/2022 13:51:50 - INFO - codeparrot_training - Step 19061: {'lr': 0.0003596761158959969, 'samples': 3659904, 'steps': 19061, 'loss/train': 0.989717423915863}
+01/27/2022 13:51:53 - INFO - codeparrot_training - Step 19062: {'lr': 0.00035966141183919485, 'samples': 3660096, 'steps': 19062, 'loss/train': 0.8926084935665131}
+01/27/2022 13:51:57 - INFO - codeparrot_training - Step 19063: {'lr': 0.0003596467073126382, 'samples': 3660288, 'steps': 19063, 'loss/train': 0.9105120599269867}
+01/27/2022 13:52:00 - INFO - codeparrot_training - Step 19064: {'lr': 0.0003596320023163898, 'samples': 3660480, 'steps': 19064, 'loss/train': 0.9508336186408997}
+01/27/2022 13:52:03 - INFO - codeparrot_training - Step 19065: {'lr': 0.0003596172968505127, 'samples': 3660672, 'steps': 19065, 'loss/train': 0.7974936068058014}
+01/27/2022 13:52:06 - INFO - codeparrot_training - Step 19066: {'lr': 0.0003596025909150699, 'samples': 3660864, 'steps': 19066, 'loss/train': 0.5297851860523224}
+01/27/2022 13:52:09 - INFO - codeparrot_training - Step 19067: {'lr': 0.00035958788451012446, 'samples': 3661056, 'steps': 19067, 'loss/train': 0.4985020011663437}
+01/27/2022 13:52:12 - INFO - codeparrot_training - Step 19068: {'lr': 0.0003595731776357393, 'samples': 3661248, 'steps': 19068, 'loss/train': 1.0202693939208984}
+01/27/2022 13:52:17 - INFO - codeparrot_training - Step 19069: {'lr': 0.00035955847029197746, 'samples': 3661440, 'steps': 19069, 'loss/train': 0.8018979728221893}
+01/27/2022 13:52:20 - INFO - codeparrot_training - Step 19070: {'lr': 0.0003595437624789019, 'samples': 3661632, 'steps': 19070, 'loss/train': 0.645546555519104}
+01/27/2022 13:52:23 - INFO - codeparrot_training - Step 19071: {'lr': 0.00035952905419657565, 'samples': 3661824, 'steps': 19071, 'loss/train': 0.8564887940883636}
+01/27/2022 13:52:26 - INFO - codeparrot_training - Step 19072: {'lr': 0.0003595143454450617, 'samples': 3662016, 'steps': 19072, 'loss/train': 1.148466020822525}
+01/27/2022 13:52:30 - INFO - codeparrot_training - Step 19073: {'lr': 0.0003594996362244231, 'samples': 3662208, 'steps': 19073, 'loss/train': 1.0687452256679535}
+01/27/2022 13:52:33 - INFO - codeparrot_training - Step 19074: {'lr': 0.0003594849265347228, 'samples': 3662400, 'steps': 19074, 'loss/train': 0.4887618273496628}
+01/27/2022 13:52:36 - INFO - codeparrot_training - Step 19075: {'lr': 0.00035947021637602384, 'samples': 3662592, 'steps': 19075, 'loss/train': 0.9588407278060913}
+01/27/2022 13:52:39 - INFO - codeparrot_training - Step 19076: {'lr': 0.0003594555057483892, 'samples': 3662784, 'steps': 19076, 'loss/train': 0.8967203199863434}
+01/27/2022 13:52:44 - INFO - codeparrot_training - Step 19077: {'lr': 0.00035944079465188194, 'samples': 3662976, 'steps': 19077, 'loss/train': 0.2185160368680954}
+01/27/2022 13:52:47 - INFO - codeparrot_training - Step 19078: {'lr': 0.0003594260830865651, 'samples': 3663168, 'steps': 19078, 'loss/train': 0.46099019050598145}
+01/27/2022 13:52:50 - INFO - codeparrot_training - Step 19079: {'lr': 0.00035941137105250173, 'samples': 3663360, 'steps': 19079, 'loss/train': 1.1373668611049652}
+01/27/2022 13:52:53 - INFO - codeparrot_training - Step 19080: {'lr': 0.00035939665854975466, 'samples': 3663552, 'steps': 19080, 'loss/train': 0.66378253698349}
+01/27/2022 13:52:57 - INFO - codeparrot_training - Step 19081: {'lr': 0.0003593819455783871, 'samples': 3663744, 'steps': 19081, 'loss/train': 1.3406189382076263}
+01/27/2022 13:53:00 - INFO - codeparrot_training - Step 19082: {'lr': 0.000359367232138462, 'samples': 3663936, 'steps': 19082, 'loss/train': 1.1992820799350739}
+01/27/2022 13:53:03 - INFO - codeparrot_training - Step 19083: {'lr': 0.00035935251823004244, 'samples': 3664128, 'steps': 19083, 'loss/train': 0.7739851176738739}
+01/27/2022 13:53:06 - INFO - codeparrot_training - Step 19084: {'lr': 0.00035933780385319136, 'samples': 3664320, 'steps': 19084, 'loss/train': 0.7469961494207382}
+01/27/2022 13:53:09 - INFO - codeparrot_training - Step 19085: {'lr': 0.0003593230890079719, 'samples': 3664512, 'steps': 19085, 'loss/train': 1.1714991331100464}
+01/27/2022 13:53:13 - INFO - codeparrot_training - Step 19086: {'lr': 0.0003593083736944471, 'samples': 3664704, 'steps': 19086, 'loss/train': 0.23480625450611115}
+01/27/2022 13:53:17 - INFO - codeparrot_training - Step 19087: {'lr': 0.00035929365791267974, 'samples': 3664896, 'steps': 19087, 'loss/train': 1.1517069339752197}
+01/27/2022 13:53:20 - INFO - codeparrot_training - Step 19088: {'lr': 0.00035927894166273323, 'samples': 3665088, 'steps': 19088, 'loss/train': 0.8544479012489319}
+01/27/2022 13:53:23 - INFO - codeparrot_training - Step 19089: {'lr': 0.00035926422494467035, 'samples': 3665280, 'steps': 19089, 'loss/train': 0.6294917464256287}
+01/27/2022 13:53:26 - INFO - codeparrot_training - Step 19090: {'lr': 0.0003592495077585543, 'samples': 3665472, 'steps': 19090, 'loss/train': 0.713222399353981}
+01/27/2022 13:53:29 - INFO - codeparrot_training - Step 19091: {'lr': 0.000359234790104448, 'samples': 3665664, 'steps': 19091, 'loss/train': 1.0695366561412811}
+01/27/2022 13:53:32 - INFO - codeparrot_training - Step 19092: {'lr': 0.0003592200719824146, 'samples': 3665856, 'steps': 19092, 'loss/train': 1.0995365381240845}
+01/27/2022 13:53:36 - INFO - codeparrot_training - Step 19093: {'lr': 0.000359205353392517, 'samples': 3666048, 'steps': 19093, 'loss/train': 0.9446450471878052}
+01/27/2022 13:53:39 - INFO - codeparrot_training - Step 19094: {'lr': 0.00035919063433481835, 'samples': 3666240, 'steps': 19094, 'loss/train': 1.3749322593212128}
+01/27/2022 13:53:43 - INFO - codeparrot_training - Step 19095: {'lr': 0.0003591759148093818, 'samples': 3666432, 'steps': 19095, 'loss/train': 0.9402890503406525}
+01/27/2022 13:53:46 - INFO - codeparrot_training - Step 19096: {'lr': 0.0003591611948162702, 'samples': 3666624, 'steps': 19096, 'loss/train': 1.349437415599823}
+01/27/2022 13:53:49 - INFO - codeparrot_training - Step 19097: {'lr': 0.0003591464743555467, 'samples': 3666816, 'steps': 19097, 'loss/train': 0.48575757443904877}
+01/27/2022 13:53:52 - INFO - codeparrot_training - Step 19098: {'lr': 0.0003591317534272744, 'samples': 3667008, 'steps': 19098, 'loss/train': 0.35403746366500854}
+01/27/2022 13:53:56 - INFO - codeparrot_training - Step 19099: {'lr': 0.0003591170320315164, 'samples': 3667200, 'steps': 19099, 'loss/train': 0.8728524148464203}
+01/27/2022 13:53:59 - INFO - codeparrot_training - Step 19100: {'lr': 0.0003591023101683355, 'samples': 3667392, 'steps': 19100, 'loss/train': 0.30339378118515015}
+01/27/2022 13:54:02 - INFO - codeparrot_training - Step 19101: {'lr': 0.00035908758783779506, 'samples': 3667584, 'steps': 19101, 'loss/train': 1.1293379366397858}
+01/27/2022 13:54:05 - INFO - codeparrot_training - Step 19102: {'lr': 0.000359072865039958, 'samples': 3667776, 'steps': 19102, 'loss/train': 0.705081894993782}
+01/27/2022 13:54:08 - INFO - codeparrot_training - Step 19103: {'lr': 0.0003590581417748875, 'samples': 3667968, 'steps': 19103, 'loss/train': 1.2135385572910309}
+01/27/2022 13:54:13 - INFO - codeparrot_training - Step 19104: {'lr': 0.0003590434180426465, 'samples': 3668160, 'steps': 19104, 'loss/train': 0.8362473249435425}
+01/27/2022 13:54:16 - INFO - codeparrot_training - Step 19105: {'lr': 0.00035902869384329803, 'samples': 3668352, 'steps': 19105, 'loss/train': 1.2956298887729645}
+01/27/2022 13:54:19 - INFO - codeparrot_training - Step 19106: {'lr': 0.0003590139691769054, 'samples': 3668544, 'steps': 19106, 'loss/train': 0.624491885304451}
+01/27/2022 13:54:23 - INFO - codeparrot_training - Step 19107: {'lr': 0.00035899924404353146, 'samples': 3668736, 'steps': 19107, 'loss/train': 0.7572492063045502}
+01/27/2022 13:54:26 - INFO - codeparrot_training - Step 19108: {'lr': 0.00035898451844323937, 'samples': 3668928, 'steps': 19108, 'loss/train': 1.0209040939807892}
+01/27/2022 13:54:29 - INFO - codeparrot_training - Step 19109: {'lr': 0.0003589697923760923, 'samples': 3669120, 'steps': 19109, 'loss/train': 0.586201474070549}
+01/27/2022 13:54:32 - INFO - codeparrot_training - Step 19110: {'lr': 0.00035895506584215307, 'samples': 3669312, 'steps': 19110, 'loss/train': 0.6085498631000519}
+01/27/2022 13:54:35 - INFO - codeparrot_training - Step 19111: {'lr': 0.0003589403388414851, 'samples': 3669504, 'steps': 19111, 'loss/train': 0.6207830160856247}
+01/27/2022 13:54:38 - INFO - codeparrot_training - Step 19112: {'lr': 0.0003589256113741513, 'samples': 3669696, 'steps': 19112, 'loss/train': 0.7016380727291107}
+01/27/2022 13:54:43 - INFO - codeparrot_training - Step 19113: {'lr': 0.00035891088344021464, 'samples': 3669888, 'steps': 19113, 'loss/train': 0.8857230842113495}
+01/27/2022 13:54:46 - INFO - codeparrot_training - Step 19114: {'lr': 0.00035889615503973847, 'samples': 3670080, 'steps': 19114, 'loss/train': 1.049599975347519}
+01/27/2022 13:54:49 - INFO - codeparrot_training - Step 19115: {'lr': 0.00035888142617278567, 'samples': 3670272, 'steps': 19115, 'loss/train': 0.7049127817153931}
+01/27/2022 13:54:52 - INFO - codeparrot_training - Step 19116: {'lr': 0.00035886669683941945, 'samples': 3670464, 'steps': 19116, 'loss/train': 0.6181523948907852}
+01/27/2022 13:54:55 - INFO - codeparrot_training - Step 19117: {'lr': 0.00035885196703970286, 'samples': 3670656, 'steps': 19117, 'loss/train': 0.781467854976654}
+01/27/2022 13:54:59 - INFO - codeparrot_training - Step 19118: {'lr': 0.00035883723677369893, 'samples': 3670848, 'steps': 19118, 'loss/train': 0.8513934016227722}
+01/27/2022 13:55:02 - INFO - codeparrot_training - Step 19119: {'lr': 0.000358822506041471, 'samples': 3671040, 'steps': 19119, 'loss/train': 1.1049672961235046}
+01/27/2022 13:55:05 - INFO - codeparrot_training - Step 19120: {'lr': 0.0003588077748430819, 'samples': 3671232, 'steps': 19120, 'loss/train': 1.1470887064933777}
+01/27/2022 13:55:10 - INFO - codeparrot_training - Step 19121: {'lr': 0.00035879304317859486, 'samples': 3671424, 'steps': 19121, 'loss/train': 0.9517691731452942}
+01/27/2022 13:55:13 - INFO - codeparrot_training - Step 19122: {'lr': 0.00035877831104807296, 'samples': 3671616, 'steps': 19122, 'loss/train': 1.0932510495185852}
+01/27/2022 13:55:16 - INFO - codeparrot_training - Step 19123: {'lr': 0.00035876357845157934, 'samples': 3671808, 'steps': 19123, 'loss/train': 0.37284714728593826}
+01/27/2022 13:55:19 - INFO - codeparrot_training - Step 19124: {'lr': 0.00035874884538917705, 'samples': 3672000, 'steps': 19124, 'loss/train': 0.09017263166606426}
+01/27/2022 13:55:23 - INFO - codeparrot_training - Step 19125: {'lr': 0.0003587341118609293, 'samples': 3672192, 'steps': 19125, 'loss/train': 0.6465041488409042}
+01/27/2022 13:55:26 - INFO - codeparrot_training - Step 19126: {'lr': 0.00035871937786689914, 'samples': 3672384, 'steps': 19126, 'loss/train': 0.5750724971294403}
+01/27/2022 13:55:29 - INFO - codeparrot_training - Step 19127: {'lr': 0.00035870464340714966, 'samples': 3672576, 'steps': 19127, 'loss/train': 1.1021609902381897}
+01/27/2022 13:55:32 - INFO - codeparrot_training - Step 19128: {'lr': 0.00035868990848174397, 'samples': 3672768, 'steps': 19128, 'loss/train': 0.9781957268714905}
+01/27/2022 13:55:35 - INFO - codeparrot_training - Step 19129: {'lr': 0.00035867517309074527, 'samples': 3672960, 'steps': 19129, 'loss/train': 0.8018022179603577}
+01/27/2022 13:55:40 - INFO - codeparrot_training - Step 19130: {'lr': 0.0003586604372342166, 'samples': 3673152, 'steps': 19130, 'loss/train': 0.8628755807876587}
+01/27/2022 13:55:43 - INFO - codeparrot_training - Step 19131: {'lr': 0.0003586457009122211, 'samples': 3673344, 'steps': 19131, 'loss/train': 0.8672328293323517}
+01/27/2022 13:55:46 - INFO - codeparrot_training - Step 19132: {'lr': 0.000358630964124822, 'samples': 3673536, 'steps': 19132, 'loss/train': 0.8878940641880035}
+01/27/2022 13:55:49 - INFO - codeparrot_training - Step 19133: {'lr': 0.0003586162268720823, 'samples': 3673728, 'steps': 19133, 'loss/train': 1.323710560798645}
+01/27/2022 13:55:52 - INFO - codeparrot_training - Step 19134: {'lr': 0.00035860148915406513, 'samples': 3673920, 'steps': 19134, 'loss/train': 0.5600125640630722}
+01/27/2022 13:55:55 - INFO - codeparrot_training - Step 19135: {'lr': 0.00035858675097083373, 'samples': 3674112, 'steps': 19135, 'loss/train': 0.899318128824234}
+01/27/2022 13:55:59 - INFO - codeparrot_training - Step 19136: {'lr': 0.0003585720123224512, 'samples': 3674304, 'steps': 19136, 'loss/train': 0.7964322566986084}
+01/27/2022 13:56:02 - INFO - codeparrot_training - Step 19137: {'lr': 0.0003585572732089806, 'samples': 3674496, 'steps': 19137, 'loss/train': 1.0546216070652008}
+01/27/2022 13:56:05 - INFO - codeparrot_training - Step 19138: {'lr': 0.00035854253363048507, 'samples': 3674688, 'steps': 19138, 'loss/train': 1.372391402721405}
+01/27/2022 13:56:10 - INFO - codeparrot_training - Step 19139: {'lr': 0.00035852779358702783, 'samples': 3674880, 'steps': 19139, 'loss/train': 0.2930922880768776}
+01/27/2022 13:56:13 - INFO - codeparrot_training - Step 19140: {'lr': 0.00035851305307867197, 'samples': 3675072, 'steps': 19140, 'loss/train': 0.6616984158754349}
+01/27/2022 13:56:16 - INFO - codeparrot_training - Step 19141: {'lr': 0.0003584983121054807, 'samples': 3675264, 'steps': 19141, 'loss/train': 2.0498435497283936}
+01/27/2022 13:56:19 - INFO - codeparrot_training - Step 19142: {'lr': 0.00035848357066751703, 'samples': 3675456, 'steps': 19142, 'loss/train': 0.5872955471277237}
+01/27/2022 13:56:22 - INFO - codeparrot_training - Step 19143: {'lr': 0.0003584688287648442, 'samples': 3675648, 'steps': 19143, 'loss/train': 0.7771088182926178}
+01/27/2022 13:56:25 - INFO - codeparrot_training - Step 19144: {'lr': 0.00035845408639752544, 'samples': 3675840, 'steps': 19144, 'loss/train': 0.8948843479156494}
+01/27/2022 13:56:29 - INFO - codeparrot_training - Step 19145: {'lr': 0.00035843934356562376, 'samples': 3676032, 'steps': 19145, 'loss/train': 0.4033498764038086}
+01/27/2022 13:56:32 - INFO - codeparrot_training - Step 19146: {'lr': 0.00035842460026920233, 'samples': 3676224, 'steps': 19146, 'loss/train': 0.5796536207199097}
+01/27/2022 13:56:35 - INFO - codeparrot_training - Step 19147: {'lr': 0.00035840985650832435, 'samples': 3676416, 'steps': 19147, 'loss/train': 1.0506847500801086}
+01/27/2022 13:56:39 - INFO - codeparrot_training - Step 19148: {'lr': 0.000358395112283053, 'samples': 3676608, 'steps': 19148, 'loss/train': 0.8585214614868164}
+01/27/2022 13:56:43 - INFO - codeparrot_training - Step 19149: {'lr': 0.00035838036759345144, 'samples': 3676800, 'steps': 19149, 'loss/train': 0.6560810655355453}
+01/27/2022 13:56:46 - INFO - codeparrot_training - Step 19150: {'lr': 0.0003583656224395827, 'samples': 3676992, 'steps': 19150, 'loss/train': 0.5684812366962433}
+01/27/2022 13:56:49 - INFO - codeparrot_training - Step 19151: {'lr': 0.00035835087682151016, 'samples': 3677184, 'steps': 19151, 'loss/train': 0.7870864570140839}
+01/27/2022 13:56:52 - INFO - codeparrot_training - Step 19152: {'lr': 0.00035833613073929684, 'samples': 3677376, 'steps': 19152, 'loss/train': 0.6085349768400192}
+01/27/2022 13:56:55 - INFO - codeparrot_training - Step 19153: {'lr': 0.00035832138419300585, 'samples': 3677568, 'steps': 19153, 'loss/train': 0.8726634979248047}
+01/27/2022 13:56:58 - INFO - codeparrot_training - Step 19154: {'lr': 0.00035830663718270056, 'samples': 3677760, 'steps': 19154, 'loss/train': 0.6293101608753204}
+01/27/2022 13:57:01 - INFO - codeparrot_training - Step 19155: {'lr': 0.00035829188970844397, 'samples': 3677952, 'steps': 19155, 'loss/train': 0.9232025742530823}
+01/27/2022 13:57:04 - INFO - codeparrot_training - Step 19156: {'lr': 0.0003582771417702993, 'samples': 3678144, 'steps': 19156, 'loss/train': 0.07431205920875072}
+01/27/2022 13:57:09 - INFO - codeparrot_training - Step 19157: {'lr': 0.0003582623933683297, 'samples': 3678336, 'steps': 19157, 'loss/train': 0.8144190609455109}
+01/27/2022 13:57:13 - INFO - codeparrot_training - Step 19158: {'lr': 0.0003582476445025985, 'samples': 3678528, 'steps': 19158, 'loss/train': 0.48618923127651215}
+01/27/2022 13:57:16 - INFO - codeparrot_training - Step 19159: {'lr': 0.00035823289517316866, 'samples': 3678720, 'steps': 19159, 'loss/train': 1.1090679466724396}
+01/27/2022 13:57:19 - INFO - codeparrot_training - Step 19160: {'lr': 0.00035821814538010356, 'samples': 3678912, 'steps': 19160, 'loss/train': 0.6663577705621719}
+01/27/2022 13:57:22 - INFO - codeparrot_training - Step 19161: {'lr': 0.00035820339512346614, 'samples': 3679104, 'steps': 19161, 'loss/train': 0.750765323638916}
+01/27/2022 13:57:25 - INFO - codeparrot_training - Step 19162: {'lr': 0.0003581886444033199, 'samples': 3679296, 'steps': 19162, 'loss/train': 0.8415513038635254}
+01/27/2022 13:57:28 - INFO - codeparrot_training - Step 19163: {'lr': 0.00035817389321972777, 'samples': 3679488, 'steps': 19163, 'loss/train': 0.3187028244137764}
+01/27/2022 13:57:31 - INFO - codeparrot_training - Step 19164: {'lr': 0.000358159141572753, 'samples': 3679680, 'steps': 19164, 'loss/train': 0.7033755630254745}
+01/27/2022 13:57:35 - INFO - codeparrot_training - Step 19165: {'lr': 0.0003581443894624589, 'samples': 3679872, 'steps': 19165, 'loss/train': 0.9668006300926208}
+01/27/2022 13:57:39 - INFO - codeparrot_training - Step 19166: {'lr': 0.0003581296368889085, 'samples': 3680064, 'steps': 19166, 'loss/train': 0.46930794417858124}
+01/27/2022 13:57:42 - INFO - codeparrot_training - Step 19167: {'lr': 0.0003581148838521651, 'samples': 3680256, 'steps': 19167, 'loss/train': 1.0974068641662598}
+01/27/2022 13:57:45 - INFO - codeparrot_training - Step 19168: {'lr': 0.00035810013035229187, 'samples': 3680448, 'steps': 19168, 'loss/train': 0.6984552294015884}
+01/27/2022 13:57:48 - INFO - codeparrot_training - Step 19169: {'lr': 0.000358085376389352, 'samples': 3680640, 'steps': 19169, 'loss/train': 0.11412262916564941}
+01/27/2022 13:57:52 - INFO - codeparrot_training - Step 19170: {'lr': 0.0003580706219634087, 'samples': 3680832, 'steps': 19170, 'loss/train': 0.6542096883058548}
+01/27/2022 13:57:55 - INFO - codeparrot_training - Step 19171: {'lr': 0.0003580558670745252, 'samples': 3681024, 'steps': 19171, 'loss/train': 1.10946124792099}
+01/27/2022 13:57:58 - INFO - codeparrot_training - Step 19172: {'lr': 0.00035804111172276464, 'samples': 3681216, 'steps': 19172, 'loss/train': 0.8047996759414673}
+01/27/2022 13:58:01 - INFO - codeparrot_training - Step 19173: {'lr': 0.00035802635590819035, 'samples': 3681408, 'steps': 19173, 'loss/train': 0.5591379404067993}
+01/27/2022 13:58:04 - INFO - codeparrot_training - Step 19174: {'lr': 0.00035801159963086535, 'samples': 3681600, 'steps': 19174, 'loss/train': 0.8017979264259338}
+01/27/2022 13:58:09 - INFO - codeparrot_training - Step 19175: {'lr': 0.000357996842890853, 'samples': 3681792, 'steps': 19175, 'loss/train': 0.13271142169833183}
+01/27/2022 13:58:12 - INFO - codeparrot_training - Step 19176: {'lr': 0.00035798208568821647, 'samples': 3681984, 'steps': 19176, 'loss/train': 0.7266075611114502}
+01/27/2022 13:58:15 - INFO - codeparrot_training - Step 19177: {'lr': 0.00035796732802301895, 'samples': 3682176, 'steps': 19177, 'loss/train': 0.6272857189178467}
+01/27/2022 13:58:18 - INFO - codeparrot_training - Step 19178: {'lr': 0.00035795256989532367, 'samples': 3682368, 'steps': 19178, 'loss/train': 1.230138748884201}
+01/27/2022 13:58:21 - INFO - codeparrot_training - Step 19179: {'lr': 0.0003579378113051939, 'samples': 3682560, 'steps': 19179, 'loss/train': 0.8036463260650635}
+01/27/2022 13:58:24 - INFO - codeparrot_training - Step 19180: {'lr': 0.0003579230522526928, 'samples': 3682752, 'steps': 19180, 'loss/train': 0.882531076669693}
+01/27/2022 13:58:28 - INFO - codeparrot_training - Step 19181: {'lr': 0.00035790829273788356, 'samples': 3682944, 'steps': 19181, 'loss/train': 0.0790545791387558}
+01/27/2022 13:58:31 - INFO - codeparrot_training - Step 19182: {'lr': 0.0003578935327608295, 'samples': 3683136, 'steps': 19182, 'loss/train': 1.3187367618083954}
+01/27/2022 13:58:34 - INFO - codeparrot_training - Step 19183: {'lr': 0.00035787877232159384, 'samples': 3683328, 'steps': 19183, 'loss/train': 0.46655192971229553}
+01/27/2022 13:58:39 - INFO - codeparrot_training - Step 19184: {'lr': 0.00035786401142023975, 'samples': 3683520, 'steps': 19184, 'loss/train': 1.6295413970947266}
+01/27/2022 13:58:42 - INFO - codeparrot_training - Step 19185: {'lr': 0.0003578492500568304, 'samples': 3683712, 'steps': 19185, 'loss/train': 0.6635250449180603}
+01/27/2022 13:58:45 - INFO - codeparrot_training - Step 19186: {'lr': 0.00035783448823142926, 'samples': 3683904, 'steps': 19186, 'loss/train': 0.13666987046599388}
+01/27/2022 13:58:48 - INFO - codeparrot_training - Step 19187: {'lr': 0.00035781972594409937, 'samples': 3684096, 'steps': 19187, 'loss/train': 0.8109507858753204}
+01/27/2022 13:58:52 - INFO - codeparrot_training - Step 19188: {'lr': 0.000357804963194904, 'samples': 3684288, 'steps': 19188, 'loss/train': 0.7096500098705292}
+01/27/2022 13:58:55 - INFO - codeparrot_training - Step 19189: {'lr': 0.00035779019998390636, 'samples': 3684480, 'steps': 19189, 'loss/train': 0.947950154542923}
+01/27/2022 13:58:58 - INFO - codeparrot_training - Step 19190: {'lr': 0.00035777543631116977, 'samples': 3684672, 'steps': 19190, 'loss/train': 0.06473546102643013}
+01/27/2022 13:59:01 - INFO - codeparrot_training - Step 19191: {'lr': 0.00035776067217675744, 'samples': 3684864, 'steps': 19191, 'loss/train': 0.9052129089832306}
+01/27/2022 13:59:05 - INFO - codeparrot_training - Step 19192: {'lr': 0.0003577459075807326, 'samples': 3685056, 'steps': 19192, 'loss/train': 0.6763326823711395}
+01/27/2022 13:59:09 - INFO - codeparrot_training - Step 19193: {'lr': 0.00035773114252315844, 'samples': 3685248, 'steps': 19193, 'loss/train': 1.5266051888465881}
+01/27/2022 13:59:12 - INFO - codeparrot_training - Step 19194: {'lr': 0.0003577163770040984, 'samples': 3685440, 'steps': 19194, 'loss/train': 0.7138104289770126}
+01/27/2022 13:59:15 - INFO - codeparrot_training - Step 19195: {'lr': 0.00035770161102361553, 'samples': 3685632, 'steps': 19195, 'loss/train': 0.5702521204948425}
+01/27/2022 13:59:18 - INFO - codeparrot_training - Step 19196: {'lr': 0.0003576868445817732, 'samples': 3685824, 'steps': 19196, 'loss/train': 0.9198749363422394}
+01/27/2022 13:59:21 - INFO - codeparrot_training - Step 19197: {'lr': 0.00035767207767863453, 'samples': 3686016, 'steps': 19197, 'loss/train': 1.2762126624584198}
+01/27/2022 13:59:24 - INFO - codeparrot_training - Step 19198: {'lr': 0.00035765731031426294, 'samples': 3686208, 'steps': 19198, 'loss/train': 0.7993345856666565}
+01/27/2022 13:59:27 - INFO - codeparrot_training - Step 19199: {'lr': 0.0003576425424887216, 'samples': 3686400, 'steps': 19199, 'loss/train': 2.252220332622528}
+01/27/2022 13:59:31 - INFO - codeparrot_training - Step 19200: {'lr': 0.0003576277742020738, 'samples': 3686592, 'steps': 19200, 'loss/train': 1.743082344532013}
+01/27/2022 13:59:36 - INFO - codeparrot_training - Step 19201: {'lr': 0.0003576130054543828, 'samples': 3686784, 'steps': 19201, 'loss/train': 0.6246346235275269}
+01/27/2022 13:59:39 - INFO - codeparrot_training - Step 19202: {'lr': 0.00035759823624571184, 'samples': 3686976, 'steps': 19202, 'loss/train': 1.7783501744270325}
+01/27/2022 13:59:42 - INFO - codeparrot_training - Step 19203: {'lr': 0.00035758346657612417, 'samples': 3687168, 'steps': 19203, 'loss/train': 0.8300794064998627}
+01/27/2022 13:59:45 - INFO - codeparrot_training - Step 19204: {'lr': 0.00035756869644568314, 'samples': 3687360, 'steps': 19204, 'loss/train': 0.7453024238348007}
+01/27/2022 13:59:48 - INFO - codeparrot_training - Step 19205: {'lr': 0.00035755392585445195, 'samples': 3687552, 'steps': 19205, 'loss/train': 0.39335158467292786}
+01/27/2022 13:59:51 - INFO - codeparrot_training - Step 19206: {'lr': 0.0003575391548024939, 'samples': 3687744, 'steps': 19206, 'loss/train': 0.5866869986057281}
+01/27/2022 13:59:55 - INFO - codeparrot_training - Step 19207: {'lr': 0.00035752438328987224, 'samples': 3687936, 'steps': 19207, 'loss/train': 0.8152348101139069}
+01/27/2022 13:59:58 - INFO - codeparrot_training - Step 19208: {'lr': 0.00035750961131665034, 'samples': 3688128, 'steps': 19208, 'loss/train': 0.7222181260585785}
+01/27/2022 14:00:01 - INFO - codeparrot_training - Step 19209: {'lr': 0.0003574948388828913, 'samples': 3688320, 'steps': 19209, 'loss/train': 0.9057958424091339}
+01/27/2022 14:00:05 - INFO - codeparrot_training - Step 19210: {'lr': 0.0003574800659886586, 'samples': 3688512, 'steps': 19210, 'loss/train': 0.9648963510990143}
+01/27/2022 14:00:08 - INFO - codeparrot_training - Step 19211: {'lr': 0.0003574652926340153, 'samples': 3688704, 'steps': 19211, 'loss/train': 0.9687009751796722}
+01/27/2022 14:00:12 - INFO - codeparrot_training - Step 19212: {'lr': 0.0003574505188190249, 'samples': 3688896, 'steps': 19212, 'loss/train': 0.6907726228237152}
+01/27/2022 14:00:15 - INFO - codeparrot_training - Step 19213: {'lr': 0.0003574357445437506, 'samples': 3689088, 'steps': 19213, 'loss/train': 0.8353604078292847}
+01/27/2022 14:00:18 - INFO - codeparrot_training - Step 19214: {'lr': 0.0003574209698082556, 'samples': 3689280, 'steps': 19214, 'loss/train': 1.4373134672641754}
+01/27/2022 14:00:21 - INFO - codeparrot_training - Step 19215: {'lr': 0.0003574061946126034, 'samples': 3689472, 'steps': 19215, 'loss/train': 0.7857368588447571}
+01/27/2022 14:00:24 - INFO - codeparrot_training - Step 19216: {'lr': 0.0003573914189568571, 'samples': 3689664, 'steps': 19216, 'loss/train': 0.29886728525161743}
+01/27/2022 14:00:27 - INFO - codeparrot_training - Step 19217: {'lr': 0.0003573766428410801, 'samples': 3689856, 'steps': 19217, 'loss/train': 0.7149292677640915}
+01/27/2022 14:00:30 - INFO - codeparrot_training - Step 19218: {'lr': 0.00035736186626533556, 'samples': 3690048, 'steps': 19218, 'loss/train': 0.9177288115024567}
+01/27/2022 14:00:35 - INFO - codeparrot_training - Step 19219: {'lr': 0.00035734708922968695, 'samples': 3690240, 'steps': 19219, 'loss/train': 1.0800961554050446}
+01/27/2022 14:00:38 - INFO - codeparrot_training - Step 19220: {'lr': 0.0003573323117341975, 'samples': 3690432, 'steps': 19220, 'loss/train': 0.5670248419046402}
+01/27/2022 14:00:41 - INFO - codeparrot_training - Step 19221: {'lr': 0.0003573175337789305, 'samples': 3690624, 'steps': 19221, 'loss/train': 0.7898529767990112}
+01/27/2022 14:00:44 - INFO - codeparrot_training - Step 19222: {'lr': 0.0003573027553639492, 'samples': 3690816, 'steps': 19222, 'loss/train': 0.5092163532972336}
+01/27/2022 14:00:47 - INFO - codeparrot_training - Step 19223: {'lr': 0.0003572879764893171, 'samples': 3691008, 'steps': 19223, 'loss/train': 0.7751196920871735}
+01/27/2022 14:00:51 - INFO - codeparrot_training - Step 19224: {'lr': 0.0003572731971550973, 'samples': 3691200, 'steps': 19224, 'loss/train': 0.9035947322845459}
+01/27/2022 14:00:54 - INFO - codeparrot_training - Step 19225: {'lr': 0.0003572584173613532, 'samples': 3691392, 'steps': 19225, 'loss/train': 0.9621636271476746}
+01/27/2022 14:00:57 - INFO - codeparrot_training - Step 19226: {'lr': 0.00035724363710814807, 'samples': 3691584, 'steps': 19226, 'loss/train': 0.5303662419319153}
+01/27/2022 14:01:00 - INFO - codeparrot_training - Step 19227: {'lr': 0.00035722885639554526, 'samples': 3691776, 'steps': 19227, 'loss/train': 0.8980842232704163}
+01/27/2022 14:01:05 - INFO - codeparrot_training - Step 19228: {'lr': 0.0003572140752236081, 'samples': 3691968, 'steps': 19228, 'loss/train': 1.0414551794528961}
+01/27/2022 14:01:08 - INFO - codeparrot_training - Step 19229: {'lr': 0.0003571992935923999, 'samples': 3692160, 'steps': 19229, 'loss/train': 0.976617157459259}
+01/27/2022 14:01:11 - INFO - codeparrot_training - Step 19230: {'lr': 0.0003571845115019839, 'samples': 3692352, 'steps': 19230, 'loss/train': 1.2153810560703278}
+01/27/2022 14:01:14 - INFO - codeparrot_training - Step 19231: {'lr': 0.0003571697289524235, 'samples': 3692544, 'steps': 19231, 'loss/train': 1.0890796780586243}
+01/27/2022 14:01:18 - INFO - codeparrot_training - Step 19232: {'lr': 0.0003571549459437821, 'samples': 3692736, 'steps': 19232, 'loss/train': 1.042479693889618}
+01/27/2022 14:01:21 - INFO - codeparrot_training - Step 19233: {'lr': 0.0003571401624761229, 'samples': 3692928, 'steps': 19233, 'loss/train': 0.9542022049427032}
+01/27/2022 14:01:24 - INFO - codeparrot_training - Step 19234: {'lr': 0.0003571253785495092, 'samples': 3693120, 'steps': 19234, 'loss/train': 0.5999120771884918}
+01/27/2022 14:01:27 - INFO - codeparrot_training - Step 19235: {'lr': 0.00035711059416400446, 'samples': 3693312, 'steps': 19235, 'loss/train': 0.8575583696365356}
+01/27/2022 14:01:31 - INFO - codeparrot_training - Step 19236: {'lr': 0.0003570958093196719, 'samples': 3693504, 'steps': 19236, 'loss/train': 0.752792090177536}
+01/27/2022 14:01:35 - INFO - codeparrot_training - Step 19237: {'lr': 0.00035708102401657495, 'samples': 3693696, 'steps': 19237, 'loss/train': 1.0279844999313354}
+01/27/2022 14:01:38 - INFO - codeparrot_training - Step 19238: {'lr': 0.00035706623825477687, 'samples': 3693888, 'steps': 19238, 'loss/train': 0.9245600402355194}
+01/27/2022 14:01:41 - INFO - codeparrot_training - Step 19239: {'lr': 0.000357051452034341, 'samples': 3694080, 'steps': 19239, 'loss/train': 0.6621108949184418}
+01/27/2022 14:01:44 - INFO - codeparrot_training - Step 19240: {'lr': 0.00035703666535533076, 'samples': 3694272, 'steps': 19240, 'loss/train': 0.8972926139831543}
+01/27/2022 14:01:47 - INFO - codeparrot_training - Step 19241: {'lr': 0.0003570218782178094, 'samples': 3694464, 'steps': 19241, 'loss/train': 0.8641355931758881}
+01/27/2022 14:01:50 - INFO - codeparrot_training - Step 19242: {'lr': 0.0003570070906218403, 'samples': 3694656, 'steps': 19242, 'loss/train': 0.7012788355350494}
+01/27/2022 14:01:53 - INFO - codeparrot_training - Step 19243: {'lr': 0.00035699230256748684, 'samples': 3694848, 'steps': 19243, 'loss/train': 0.06638265028595924}
+01/27/2022 14:01:57 - INFO - codeparrot_training - Step 19244: {'lr': 0.0003569775140548122, 'samples': 3695040, 'steps': 19244, 'loss/train': 0.8867560029029846}
+01/27/2022 14:02:01 - INFO - codeparrot_training - Step 19245: {'lr': 0.00035696272508388, 'samples': 3695232, 'steps': 19245, 'loss/train': 0.8611036241054535}
+01/27/2022 14:02:04 - INFO - codeparrot_training - Step 19246: {'lr': 0.00035694793565475337, 'samples': 3695424, 'steps': 19246, 'loss/train': 1.4510469138622284}
+01/27/2022 14:02:07 - INFO - codeparrot_training - Step 19247: {'lr': 0.0003569331457674958, 'samples': 3695616, 'steps': 19247, 'loss/train': 0.9562502503395081}
+01/27/2022 14:02:11 - INFO - codeparrot_training - Step 19248: {'lr': 0.00035691835542217055, 'samples': 3695808, 'steps': 19248, 'loss/train': 1.1449478566646576}
+01/27/2022 14:02:14 - INFO - codeparrot_training - Step 19249: {'lr': 0.00035690356461884104, 'samples': 3696000, 'steps': 19249, 'loss/train': 0.7252838462591171}
+01/27/2022 14:02:17 - INFO - codeparrot_training - Step 19250: {'lr': 0.0003568887733575705, 'samples': 3696192, 'steps': 19250, 'loss/train': 0.8085598647594452}
+01/27/2022 14:02:20 - INFO - codeparrot_training - Step 19251: {'lr': 0.0003568739816384225, 'samples': 3696384, 'steps': 19251, 'loss/train': 0.8934873640537262}
+01/27/2022 14:02:23 - INFO - codeparrot_training - Step 19252: {'lr': 0.00035685918946146036, 'samples': 3696576, 'steps': 19252, 'loss/train': 0.5386114865541458}
+01/27/2022 14:02:26 - INFO - codeparrot_training - Step 19253: {'lr': 0.00035684439682674723, 'samples': 3696768, 'steps': 19253, 'loss/train': 0.9705954194068909}
+01/27/2022 14:02:31 - INFO - codeparrot_training - Step 19254: {'lr': 0.00035682960373434677, 'samples': 3696960, 'steps': 19254, 'loss/train': 0.42085012793540955}
+01/27/2022 14:02:34 - INFO - codeparrot_training - Step 19255: {'lr': 0.0003568148101843221, 'samples': 3697152, 'steps': 19255, 'loss/train': 0.851973295211792}
+01/27/2022 14:02:37 - INFO - codeparrot_training - Step 19256: {'lr': 0.0003568000161767368, 'samples': 3697344, 'steps': 19256, 'loss/train': 1.1983177363872528}
+01/27/2022 14:02:40 - INFO - codeparrot_training - Step 19257: {'lr': 0.00035678522171165406, 'samples': 3697536, 'steps': 19257, 'loss/train': 1.162205457687378}
+01/27/2022 14:02:43 - INFO - codeparrot_training - Step 19258: {'lr': 0.0003567704267891374, 'samples': 3697728, 'steps': 19258, 'loss/train': 1.6423010230064392}
+01/27/2022 14:02:46 - INFO - codeparrot_training - Step 19259: {'lr': 0.00035675563140925, 'samples': 3697920, 'steps': 19259, 'loss/train': 1.0004612803459167}
+01/27/2022 14:02:50 - INFO - codeparrot_training - Step 19260: {'lr': 0.00035674083557205553, 'samples': 3698112, 'steps': 19260, 'loss/train': 0.5786028653383255}
+01/27/2022 14:02:53 - INFO - codeparrot_training - Step 19261: {'lr': 0.00035672603927761716, 'samples': 3698304, 'steps': 19261, 'loss/train': 0.5756708979606628}
+01/27/2022 14:02:56 - INFO - codeparrot_training - Step 19262: {'lr': 0.0003567112425259984, 'samples': 3698496, 'steps': 19262, 'loss/train': 0.0594785213470459}
+01/27/2022 14:03:01 - INFO - codeparrot_training - Step 19263: {'lr': 0.00035669644531726244, 'samples': 3698688, 'steps': 19263, 'loss/train': 0.7229194790124893}
+01/27/2022 14:03:04 - INFO - codeparrot_training - Step 19264: {'lr': 0.00035668164765147284, 'samples': 3698880, 'steps': 19264, 'loss/train': 0.6371170431375504}
+01/27/2022 14:03:07 - INFO - codeparrot_training - Step 19265: {'lr': 0.00035666684952869295, 'samples': 3699072, 'steps': 19265, 'loss/train': 0.9050479531288147}
+01/27/2022 14:03:10 - INFO - codeparrot_training - Step 19266: {'lr': 0.00035665205094898613, 'samples': 3699264, 'steps': 19266, 'loss/train': 0.7332374900579453}
+01/27/2022 14:03:14 - INFO - codeparrot_training - Step 19267: {'lr': 0.0003566372519124158, 'samples': 3699456, 'steps': 19267, 'loss/train': 0.681949183344841}
+01/27/2022 14:03:17 - INFO - codeparrot_training - Step 19268: {'lr': 0.00035662245241904533, 'samples': 3699648, 'steps': 19268, 'loss/train': 0.8269691169261932}
+01/27/2022 14:03:20 - INFO - codeparrot_training - Step 19269: {'lr': 0.00035660765246893814, 'samples': 3699840, 'steps': 19269, 'loss/train': 0.60738904774189}
+01/27/2022 14:03:23 - INFO - codeparrot_training - Step 19270: {'lr': 0.00035659285206215757, 'samples': 3700032, 'steps': 19270, 'loss/train': 0.5091705769300461}
+01/27/2022 14:03:28 - INFO - codeparrot_training - Step 19271: {'lr': 0.0003565780511987672, 'samples': 3700224, 'steps': 19271, 'loss/train': 1.0282502174377441}
+01/27/2022 14:03:31 - INFO - codeparrot_training - Step 19272: {'lr': 0.00035656324987883015, 'samples': 3700416, 'steps': 19272, 'loss/train': 1.0346015989780426}
+01/27/2022 14:03:34 - INFO - codeparrot_training - Step 19273: {'lr': 0.00035654844810241004, 'samples': 3700608, 'steps': 19273, 'loss/train': 0.4737157076597214}
+01/27/2022 14:03:37 - INFO - codeparrot_training - Step 19274: {'lr': 0.0003565336458695702, 'samples': 3700800, 'steps': 19274, 'loss/train': 0.9879435002803802}
+01/27/2022 14:03:40 - INFO - codeparrot_training - Step 19275: {'lr': 0.0003565188431803741, 'samples': 3700992, 'steps': 19275, 'loss/train': 0.6706346869468689}
+01/27/2022 14:03:43 - INFO - codeparrot_training - Step 19276: {'lr': 0.0003565040400348851, 'samples': 3701184, 'steps': 19276, 'loss/train': 0.7232073247432709}
+01/27/2022 14:03:46 - INFO - codeparrot_training - Step 19277: {'lr': 0.0003564892364331665, 'samples': 3701376, 'steps': 19277, 'loss/train': 0.7853447198867798}
+01/27/2022 14:03:50 - INFO - codeparrot_training - Step 19278: {'lr': 0.0003564744323752819, 'samples': 3701568, 'steps': 19278, 'loss/train': 0.8192092180252075}
+01/27/2022 14:03:53 - INFO - codeparrot_training - Step 19279: {'lr': 0.00035645962786129464, 'samples': 3701760, 'steps': 19279, 'loss/train': 0.9530927538871765}
+01/27/2022 14:03:58 - INFO - codeparrot_training - Step 19280: {'lr': 0.0003564448228912682, 'samples': 3701952, 'steps': 19280, 'loss/train': 0.7135331332683563}
+01/27/2022 14:04:01 - INFO - codeparrot_training - Step 19281: {'lr': 0.00035643001746526586, 'samples': 3702144, 'steps': 19281, 'loss/train': 0.928640216588974}
+01/27/2022 14:04:04 - INFO - codeparrot_training - Step 19282: {'lr': 0.0003564152115833511, 'samples': 3702336, 'steps': 19282, 'loss/train': 0.7469267249107361}
+01/27/2022 14:04:07 - INFO - codeparrot_training - Step 19283: {'lr': 0.00035640040524558746, 'samples': 3702528, 'steps': 19283, 'loss/train': 1.099022001028061}
+01/27/2022 14:04:10 - INFO - codeparrot_training - Step 19284: {'lr': 0.00035638559845203817, 'samples': 3702720, 'steps': 19284, 'loss/train': 0.7202424556016922}
+01/27/2022 14:04:13 - INFO - codeparrot_training - Step 19285: {'lr': 0.00035637079120276683, 'samples': 3702912, 'steps': 19285, 'loss/train': 0.7995818853378296}
+01/27/2022 14:04:16 - INFO - codeparrot_training - Step 19286: {'lr': 0.00035635598349783676, 'samples': 3703104, 'steps': 19286, 'loss/train': 0.9302633106708527}
+01/27/2022 14:04:20 - INFO - codeparrot_training - Step 19287: {'lr': 0.0003563411753373115, 'samples': 3703296, 'steps': 19287, 'loss/train': 0.5404049456119537}
+01/27/2022 14:04:23 - INFO - codeparrot_training - Step 19288: {'lr': 0.0003563263667212543, 'samples': 3703488, 'steps': 19288, 'loss/train': 0.738874688744545}
+01/27/2022 14:04:27 - INFO - codeparrot_training - Step 19289: {'lr': 0.0003563115576497288, 'samples': 3703680, 'steps': 19289, 'loss/train': 0.7337576150894165}
+01/27/2022 14:04:31 - INFO - codeparrot_training - Step 19290: {'lr': 0.0003562967481227982, 'samples': 3703872, 'steps': 19290, 'loss/train': 0.8719587028026581}
+01/27/2022 14:04:34 - INFO - codeparrot_training - Step 19291: {'lr': 0.0003562819381405262, 'samples': 3704064, 'steps': 19291, 'loss/train': 0.7028544545173645}
+01/27/2022 14:04:37 - INFO - codeparrot_training - Step 19292: {'lr': 0.0003562671277029761, 'samples': 3704256, 'steps': 19292, 'loss/train': 0.9448243081569672}
+01/27/2022 14:04:40 - INFO - codeparrot_training - Step 19293: {'lr': 0.0003562523168102114, 'samples': 3704448, 'steps': 19293, 'loss/train': 1.2111946642398834}
+01/27/2022 14:04:43 - INFO - codeparrot_training - Step 19294: {'lr': 0.00035623750546229547, 'samples': 3704640, 'steps': 19294, 'loss/train': 0.8161003589630127}
+01/27/2022 14:04:46 - INFO - codeparrot_training - Step 19295: {'lr': 0.0003562226936592919, 'samples': 3704832, 'steps': 19295, 'loss/train': 0.4743112027645111}
+01/27/2022 14:04:49 - INFO - codeparrot_training - Step 19296: {'lr': 0.0003562078814012639, 'samples': 3705024, 'steps': 19296, 'loss/train': 0.761581152677536}
+01/27/2022 14:04:53 - INFO - codeparrot_training - Step 19297: {'lr': 0.00035619306868827516, 'samples': 3705216, 'steps': 19297, 'loss/train': 0.974837064743042}
+01/27/2022 14:04:57 - INFO - codeparrot_training - Step 19298: {'lr': 0.00035617825552038894, 'samples': 3705408, 'steps': 19298, 'loss/train': 0.29500626772642136}
+01/27/2022 14:05:00 - INFO - codeparrot_training - Step 19299: {'lr': 0.00035616344189766885, 'samples': 3705600, 'steps': 19299, 'loss/train': 1.1812503933906555}
+01/27/2022 14:05:03 - INFO - codeparrot_training - Step 19300: {'lr': 0.0003561486278201783, 'samples': 3705792, 'steps': 19300, 'loss/train': 0.9224915206432343}
+01/27/2022 14:05:06 - INFO - codeparrot_training - Step 19301: {'lr': 0.00035613381328798065, 'samples': 3705984, 'steps': 19301, 'loss/train': 0.7327959984540939}
+01/27/2022 14:05:10 - INFO - codeparrot_training - Step 19302: {'lr': 0.0003561189983011396, 'samples': 3706176, 'steps': 19302, 'loss/train': 0.5869664400815964}
+01/27/2022 14:05:13 - INFO - codeparrot_training - Step 19303: {'lr': 0.00035610418285971835, 'samples': 3706368, 'steps': 19303, 'loss/train': 1.345740169286728}
+01/27/2022 14:05:16 - INFO - codeparrot_training - Step 19304: {'lr': 0.00035608936696378046, 'samples': 3706560, 'steps': 19304, 'loss/train': 0.5483862161636353}
+01/27/2022 14:05:19 - INFO - codeparrot_training - Step 19305: {'lr': 0.00035607455061338947, 'samples': 3706752, 'steps': 19305, 'loss/train': 0.8167151212692261}
+01/27/2022 14:05:22 - INFO - codeparrot_training - Step 19306: {'lr': 0.0003560597338086088, 'samples': 3706944, 'steps': 19306, 'loss/train': 0.5385386198759079}
+01/27/2022 14:05:27 - INFO - codeparrot_training - Step 19307: {'lr': 0.0003560449165495018, 'samples': 3707136, 'steps': 19307, 'loss/train': 0.962944507598877}
+01/27/2022 14:05:31 - INFO - codeparrot_training - Step 19308: {'lr': 0.0003560300988361321, 'samples': 3707328, 'steps': 19308, 'loss/train': 0.6916294097900391}
+01/27/2022 14:05:34 - INFO - codeparrot_training - Step 19309: {'lr': 0.00035601528066856315, 'samples': 3707520, 'steps': 19309, 'loss/train': 0.6008533090353012}
+01/27/2022 14:05:37 - INFO - codeparrot_training - Step 19310: {'lr': 0.00035600046204685844, 'samples': 3707712, 'steps': 19310, 'loss/train': 0.4991789013147354}
+01/27/2022 14:05:40 - INFO - codeparrot_training - Step 19311: {'lr': 0.00035598564297108134, 'samples': 3707904, 'steps': 19311, 'loss/train': 0.12102025374770164}
+01/27/2022 14:05:43 - INFO - codeparrot_training - Step 19312: {'lr': 0.0003559708234412954, 'samples': 3708096, 'steps': 19312, 'loss/train': 0.56556236743927}
+01/27/2022 14:05:46 - INFO - codeparrot_training - Step 19313: {'lr': 0.00035595600345756414, 'samples': 3708288, 'steps': 19313, 'loss/train': 0.8253653347492218}
+01/27/2022 14:05:49 - INFO - codeparrot_training - Step 19314: {'lr': 0.00035594118301995095, 'samples': 3708480, 'steps': 19314, 'loss/train': 0.7145439237356186}
+01/27/2022 14:05:54 - INFO - codeparrot_training - Step 19315: {'lr': 0.00035592636212851945, 'samples': 3708672, 'steps': 19315, 'loss/train': 0.8095880448818207}
+01/27/2022 14:05:57 - INFO - codeparrot_training - Step 19316: {'lr': 0.000355911540783333, 'samples': 3708864, 'steps': 19316, 'loss/train': 1.10997873544693}
+01/27/2022 14:06:00 - INFO - codeparrot_training - Step 19317: {'lr': 0.00035589671898445517, 'samples': 3709056, 'steps': 19317, 'loss/train': 1.1824412047863007}
+01/27/2022 14:06:03 - INFO - codeparrot_training - Step 19318: {'lr': 0.0003558818967319494, 'samples': 3709248, 'steps': 19318, 'loss/train': 0.6474249958992004}
+01/27/2022 14:06:06 - INFO - codeparrot_training - Step 19319: {'lr': 0.0003558670740258792, 'samples': 3709440, 'steps': 19319, 'loss/train': 0.7056844532489777}
+01/27/2022 14:06:10 - INFO - codeparrot_training - Step 19320: {'lr': 0.00035585225086630807, 'samples': 3709632, 'steps': 19320, 'loss/train': 0.9286513030529022}
+01/27/2022 14:06:13 - INFO - codeparrot_training - Step 19321: {'lr': 0.00035583742725329954, 'samples': 3709824, 'steps': 19321, 'loss/train': 0.9308064579963684}
+01/27/2022 14:06:16 - INFO - codeparrot_training - Step 19322: {'lr': 0.0003558226031869171, 'samples': 3710016, 'steps': 19322, 'loss/train': 0.7210637927055359}
+01/27/2022 14:06:19 - INFO - codeparrot_training - Step 19323: {'lr': 0.00035580777866722415, 'samples': 3710208, 'steps': 19323, 'loss/train': 1.5588310360908508}
+01/27/2022 14:06:23 - INFO - codeparrot_training - Step 19324: {'lr': 0.00035579295369428425, 'samples': 3710400, 'steps': 19324, 'loss/train': 0.9282775819301605}
+01/27/2022 14:06:27 - INFO - codeparrot_training - Step 19325: {'lr': 0.000355778128268161, 'samples': 3710592, 'steps': 19325, 'loss/train': 1.2322884500026703}
+01/27/2022 14:06:30 - INFO - codeparrot_training - Step 19326: {'lr': 0.0003557633023889179, 'samples': 3710784, 'steps': 19326, 'loss/train': 0.6407614052295685}
+01/27/2022 14:06:33 - INFO - codeparrot_training - Step 19327: {'lr': 0.0003557484760566183, 'samples': 3710976, 'steps': 19327, 'loss/train': 0.5411854684352875}
+01/27/2022 14:06:36 - INFO - codeparrot_training - Step 19328: {'lr': 0.0003557336492713258, 'samples': 3711168, 'steps': 19328, 'loss/train': 1.4101194441318512}
+01/27/2022 14:06:39 - INFO - codeparrot_training - Step 19329: {'lr': 0.000355718822033104, 'samples': 3711360, 'steps': 19329, 'loss/train': 1.0383542776107788}
+01/27/2022 14:06:42 - INFO - codeparrot_training - Step 19330: {'lr': 0.0003557039943420163, 'samples': 3711552, 'steps': 19330, 'loss/train': 0.6315906196832657}
+01/27/2022 14:06:45 - INFO - codeparrot_training - Step 19331: {'lr': 0.00035568916619812624, 'samples': 3711744, 'steps': 19331, 'loss/train': 0.8763413429260254}
+01/27/2022 14:06:48 - INFO - codeparrot_training - Step 19332: {'lr': 0.00035567433760149737, 'samples': 3711936, 'steps': 19332, 'loss/train': 0.7371564209461212}
+01/27/2022 14:06:54 - INFO - codeparrot_training - Step 19333: {'lr': 0.0003556595085521931, 'samples': 3712128, 'steps': 19333, 'loss/train': 1.1591482758522034}
+01/27/2022 14:06:57 - INFO - codeparrot_training - Step 19334: {'lr': 0.0003556446790502772, 'samples': 3712320, 'steps': 19334, 'loss/train': 0.6855371296405792}
+01/27/2022 14:07:00 - INFO - codeparrot_training - Step 19335: {'lr': 0.00035562984909581297, 'samples': 3712512, 'steps': 19335, 'loss/train': 0.8909309506416321}
+01/27/2022 14:07:03 - INFO - codeparrot_training - Step 19336: {'lr': 0.0003556150186888639, 'samples': 3712704, 'steps': 19336, 'loss/train': 0.9689312875270844}
+01/27/2022 14:07:07 - INFO - codeparrot_training - Step 19337: {'lr': 0.00035560018782949384, 'samples': 3712896, 'steps': 19337, 'loss/train': 0.894442230463028}
+01/27/2022 14:07:10 - INFO - codeparrot_training - Step 19338: {'lr': 0.0003555853565177659, 'samples': 3713088, 'steps': 19338, 'loss/train': 0.8421327173709869}
+01/27/2022 14:07:13 - INFO - codeparrot_training - Step 19339: {'lr': 0.00035557052475374397, 'samples': 3713280, 'steps': 19339, 'loss/train': 0.7681375443935394}
+01/27/2022 14:07:16 - INFO - codeparrot_training - Step 19340: {'lr': 0.00035555569253749135, 'samples': 3713472, 'steps': 19340, 'loss/train': 0.35084690898656845}
+01/27/2022 14:07:19 - INFO - codeparrot_training - Step 19341: {'lr': 0.0003555408598690718, 'samples': 3713664, 'steps': 19341, 'loss/train': 0.4584037810564041}
+01/27/2022 14:07:24 - INFO - codeparrot_training - Step 19342: {'lr': 0.0003555260267485485, 'samples': 3713856, 'steps': 19342, 'loss/train': 1.71500962972641}
+01/27/2022 14:07:27 - INFO - codeparrot_training - Step 19343: {'lr': 0.00035551119317598533, 'samples': 3714048, 'steps': 19343, 'loss/train': 0.6020575314760208}
+01/27/2022 14:07:30 - INFO - codeparrot_training - Step 19344: {'lr': 0.00035549635915144574, 'samples': 3714240, 'steps': 19344, 'loss/train': 0.8192650973796844}
+01/27/2022 14:07:33 - INFO - codeparrot_training - Step 19345: {'lr': 0.0003554815246749932, 'samples': 3714432, 'steps': 19345, 'loss/train': 0.8531882464885712}
+01/27/2022 14:07:36 - INFO - codeparrot_training - Step 19346: {'lr': 0.00035546668974669127, 'samples': 3714624, 'steps': 19346, 'loss/train': 1.2701847553253174}
+01/27/2022 14:07:39 - INFO - codeparrot_training - Step 19347: {'lr': 0.00035545185436660357, 'samples': 3714816, 'steps': 19347, 'loss/train': 0.6478042155504227}
+01/27/2022 14:07:42 - INFO - codeparrot_training - Step 19348: {'lr': 0.00035543701853479366, 'samples': 3715008, 'steps': 19348, 'loss/train': 0.5539656579494476}
+01/27/2022 14:07:46 - INFO - codeparrot_training - Step 19349: {'lr': 0.00035542218225132497, 'samples': 3715200, 'steps': 19349, 'loss/train': 0.99610635638237}
+01/27/2022 14:07:49 - INFO - codeparrot_training - Step 19350: {'lr': 0.00035540734551626113, 'samples': 3715392, 'steps': 19350, 'loss/train': 0.44002296030521393}
+01/27/2022 14:07:53 - INFO - codeparrot_training - Step 19351: {'lr': 0.00035539250832966574, 'samples': 3715584, 'steps': 19351, 'loss/train': 1.3204123377799988}
+01/27/2022 14:07:56 - INFO - codeparrot_training - Step 19352: {'lr': 0.00035537767069160234, 'samples': 3715776, 'steps': 19352, 'loss/train': 1.2004095911979675}
+01/27/2022 14:08:00 - INFO - codeparrot_training - Step 19353: {'lr': 0.00035536283260213434, 'samples': 3715968, 'steps': 19353, 'loss/train': 1.0948330163955688}
+01/27/2022 14:08:03 - INFO - codeparrot_training - Step 19354: {'lr': 0.0003553479940613255, 'samples': 3716160, 'steps': 19354, 'loss/train': 0.7456142753362656}
+01/27/2022 14:08:06 - INFO - codeparrot_training - Step 19355: {'lr': 0.00035533315506923924, 'samples': 3716352, 'steps': 19355, 'loss/train': 0.7495635151863098}
+01/27/2022 14:08:09 - INFO - codeparrot_training - Step 19356: {'lr': 0.0003553183156259393, 'samples': 3716544, 'steps': 19356, 'loss/train': 0.9095952808856964}
+01/27/2022 14:08:12 - INFO - codeparrot_training - Step 19357: {'lr': 0.00035530347573148904, 'samples': 3716736, 'steps': 19357, 'loss/train': 0.6083195060491562}
+01/27/2022 14:08:15 - INFO - codeparrot_training - Step 19358: {'lr': 0.0003552886353859522, 'samples': 3716928, 'steps': 19358, 'loss/train': 0.3829289674758911}
+01/27/2022 14:08:20 - INFO - codeparrot_training - Step 19359: {'lr': 0.00035527379458939225, 'samples': 3717120, 'steps': 19359, 'loss/train': 0.7207385748624802}
+01/27/2022 14:08:24 - INFO - codeparrot_training - Step 19360: {'lr': 0.00035525895334187274, 'samples': 3717312, 'steps': 19360, 'loss/train': 1.5319958925247192}
+01/27/2022 14:08:27 - INFO - codeparrot_training - Step 19361: {'lr': 0.0003552441116434574, 'samples': 3717504, 'steps': 19361, 'loss/train': 0.3723587393760681}
+01/27/2022 14:08:30 - INFO - codeparrot_training - Step 19362: {'lr': 0.0003552292694942096, 'samples': 3717696, 'steps': 19362, 'loss/train': 0.5373742282390594}
+01/27/2022 14:08:33 - INFO - codeparrot_training - Step 19363: {'lr': 0.0003552144268941931, 'samples': 3717888, 'steps': 19363, 'loss/train': 0.9957906603813171}
+01/27/2022 14:08:36 - INFO - codeparrot_training - Step 19364: {'lr': 0.00035519958384347134, 'samples': 3718080, 'steps': 19364, 'loss/train': 1.0836081504821777}
+01/27/2022 14:08:39 - INFO - codeparrot_training - Step 19365: {'lr': 0.000355184740342108, 'samples': 3718272, 'steps': 19365, 'loss/train': 1.2811387181282043}
+01/27/2022 14:08:42 - INFO - codeparrot_training - Step 19366: {'lr': 0.00035516989639016664, 'samples': 3718464, 'steps': 19366, 'loss/train': 0.9179012775421143}
+01/27/2022 14:08:46 - INFO - codeparrot_training - Step 19367: {'lr': 0.00035515505198771086, 'samples': 3718656, 'steps': 19367, 'loss/train': 0.67164346575737}
+01/27/2022 14:08:50 - INFO - codeparrot_training - Step 19368: {'lr': 0.0003551402071348042, 'samples': 3718848, 'steps': 19368, 'loss/train': 0.37114492803812027}
+01/27/2022 14:08:53 - INFO - codeparrot_training - Step 19369: {'lr': 0.0003551253618315103, 'samples': 3719040, 'steps': 19369, 'loss/train': 0.5233320146799088}
+01/27/2022 14:08:56 - INFO - codeparrot_training - Step 19370: {'lr': 0.0003551105160778927, 'samples': 3719232, 'steps': 19370, 'loss/train': 1.061743140220642}
+01/27/2022 14:08:59 - INFO - codeparrot_training - Step 19371: {'lr': 0.000355095669874015, 'samples': 3719424, 'steps': 19371, 'loss/train': 0.7238346487283707}
+01/27/2022 14:09:03 - INFO - codeparrot_training - Step 19372: {'lr': 0.00035508082321994097, 'samples': 3719616, 'steps': 19372, 'loss/train': 0.7301110178232193}
+01/27/2022 14:09:06 - INFO - codeparrot_training - Step 19373: {'lr': 0.00035506597611573387, 'samples': 3719808, 'steps': 19373, 'loss/train': 0.8794837296009064}
+01/27/2022 14:09:09 - INFO - codeparrot_training - Step 19374: {'lr': 0.0003550511285614576, 'samples': 3720000, 'steps': 19374, 'loss/train': 1.0290948450565338}
+01/27/2022 14:09:12 - INFO - codeparrot_training - Step 19375: {'lr': 0.0003550362805571756, 'samples': 3720192, 'steps': 19375, 'loss/train': 0.5639881789684296}
+01/27/2022 14:09:15 - INFO - codeparrot_training - Step 19376: {'lr': 0.00035502143210295163, 'samples': 3720384, 'steps': 19376, 'loss/train': 0.2192763313651085}
+01/27/2022 14:09:20 - INFO - codeparrot_training - Step 19377: {'lr': 0.000355006583198849, 'samples': 3720576, 'steps': 19377, 'loss/train': 0.8092781603336334}
+01/27/2022 14:09:23 - INFO - codeparrot_training - Step 19378: {'lr': 0.00035499173384493174, 'samples': 3720768, 'steps': 19378, 'loss/train': 0.7538906335830688}
+01/27/2022 14:09:26 - INFO - codeparrot_training - Step 19379: {'lr': 0.00035497688404126306, 'samples': 3720960, 'steps': 19379, 'loss/train': 0.04677906073629856}
+01/27/2022 14:09:29 - INFO - codeparrot_training - Step 19380: {'lr': 0.00035496203378790683, 'samples': 3721152, 'steps': 19380, 'loss/train': 0.889223724603653}
+01/27/2022 14:09:32 - INFO - codeparrot_training - Step 19381: {'lr': 0.0003549471830849265, 'samples': 3721344, 'steps': 19381, 'loss/train': 1.0363776683807373}
+01/27/2022 14:09:35 - INFO - codeparrot_training - Step 19382: {'lr': 0.00035493233193238584, 'samples': 3721536, 'steps': 19382, 'loss/train': 0.8841041028499603}
+01/27/2022 14:09:39 - INFO - codeparrot_training - Step 19383: {'lr': 0.00035491748033034836, 'samples': 3721728, 'steps': 19383, 'loss/train': 1.0249505639076233}
+01/27/2022 14:09:42 - INFO - codeparrot_training - Step 19384: {'lr': 0.00035490262827887764, 'samples': 3721920, 'steps': 19384, 'loss/train': 0.9894567131996155}
+01/27/2022 14:09:45 - INFO - codeparrot_training - Step 19385: {'lr': 0.0003548877757780375, 'samples': 3722112, 'steps': 19385, 'loss/train': 1.098949134349823}
+01/27/2022 14:09:50 - INFO - codeparrot_training - Step 19386: {'lr': 0.00035487292282789136, 'samples': 3722304, 'steps': 19386, 'loss/train': 0.9469549655914307}
+01/27/2022 14:09:53 - INFO - codeparrot_training - Step 19387: {'lr': 0.000354858069428503, 'samples': 3722496, 'steps': 19387, 'loss/train': 1.0487899482250214}
+01/27/2022 14:09:57 - INFO - codeparrot_training - Step 19388: {'lr': 0.0003548432155799358, 'samples': 3722688, 'steps': 19388, 'loss/train': 1.1641011536121368}
+01/27/2022 14:10:00 - INFO - codeparrot_training - Step 19389: {'lr': 0.0003548283612822537, 'samples': 3722880, 'steps': 19389, 'loss/train': 1.4863089323043823}
+01/27/2022 14:10:03 - INFO - codeparrot_training - Step 19390: {'lr': 0.0003548135065355201, 'samples': 3723072, 'steps': 19390, 'loss/train': 1.583454430103302}
+01/27/2022 14:10:06 - INFO - codeparrot_training - Step 19391: {'lr': 0.0003547986513397988, 'samples': 3723264, 'steps': 19391, 'loss/train': 0.4516713470220566}
+01/27/2022 14:10:09 - INFO - codeparrot_training - Step 19392: {'lr': 0.0003547837956951533, 'samples': 3723456, 'steps': 19392, 'loss/train': 0.41571079194545746}
+01/27/2022 14:10:12 - INFO - codeparrot_training - Step 19393: {'lr': 0.00035476893960164734, 'samples': 3723648, 'steps': 19393, 'loss/train': 0.9315632879734039}
+01/27/2022 14:10:15 - INFO - codeparrot_training - Step 19394: {'lr': 0.00035475408305934444, 'samples': 3723840, 'steps': 19394, 'loss/train': 0.048429759219288826}
+01/27/2022 14:10:20 - INFO - codeparrot_training - Step 19395: {'lr': 0.0003547392260683084, 'samples': 3724032, 'steps': 19395, 'loss/train': 0.03219435643404722}
+01/27/2022 14:10:23 - INFO - codeparrot_training - Step 19396: {'lr': 0.0003547243686286027, 'samples': 3724224, 'steps': 19396, 'loss/train': 0.888474851846695}
+01/27/2022 14:10:26 - INFO - codeparrot_training - Step 19397: {'lr': 0.000354709510740291, 'samples': 3724416, 'steps': 19397, 'loss/train': 1.0052326619625092}
+01/27/2022 14:10:29 - INFO - codeparrot_training - Step 19398: {'lr': 0.0003546946524034371, 'samples': 3724608, 'steps': 19398, 'loss/train': 0.7054439932107925}
+01/27/2022 14:10:32 - INFO - codeparrot_training - Step 19399: {'lr': 0.00035467979361810455, 'samples': 3724800, 'steps': 19399, 'loss/train': 0.7228952050209045}
+01/27/2022 14:10:36 - INFO - codeparrot_training - Step 19400: {'lr': 0.00035466493438435703, 'samples': 3724992, 'steps': 19400, 'loss/train': 0.7085863351821899}
+01/27/2022 14:10:39 - INFO - codeparrot_training - Step 19401: {'lr': 0.00035465007470225813, 'samples': 3725184, 'steps': 19401, 'loss/train': 0.9048157632350922}
+01/27/2022 14:10:42 - INFO - codeparrot_training - Step 19402: {'lr': 0.0003546352145718715, 'samples': 3725376, 'steps': 19402, 'loss/train': 0.6143759787082672}
+01/27/2022 14:10:45 - INFO - codeparrot_training - Step 19403: {'lr': 0.0003546203539932609, 'samples': 3725568, 'steps': 19403, 'loss/train': 0.5851525515317917}
+01/27/2022 14:10:49 - INFO - codeparrot_training - Step 19404: {'lr': 0.0003546054929664899, 'samples': 3725760, 'steps': 19404, 'loss/train': 0.29351744055747986}
+01/27/2022 14:10:53 - INFO - codeparrot_training - Step 19405: {'lr': 0.0003545906314916222, 'samples': 3725952, 'steps': 19405, 'loss/train': 1.1539621353149414}
+01/27/2022 14:10:56 - INFO - codeparrot_training - Step 19406: {'lr': 0.00035457576956872145, 'samples': 3726144, 'steps': 19406, 'loss/train': 0.8700731098651886}
+01/27/2022 14:10:59 - INFO - codeparrot_training - Step 19407: {'lr': 0.00035456090719785126, 'samples': 3726336, 'steps': 19407, 'loss/train': 1.0811753869056702}
+01/27/2022 14:11:02 - INFO - codeparrot_training - Step 19408: {'lr': 0.00035454604437907536, 'samples': 3726528, 'steps': 19408, 'loss/train': 0.335936039686203}
+01/27/2022 14:11:05 - INFO - codeparrot_training - Step 19409: {'lr': 0.0003545311811124574, 'samples': 3726720, 'steps': 19409, 'loss/train': 1.6353033185005188}
+01/27/2022 14:11:08 - INFO - codeparrot_training - Step 19410: {'lr': 0.0003545163173980611, 'samples': 3726912, 'steps': 19410, 'loss/train': 1.1906725466251373}
+01/27/2022 14:11:11 - INFO - codeparrot_training - Step 19411: {'lr': 0.0003545014532359501, 'samples': 3727104, 'steps': 19411, 'loss/train': 0.9165949523448944}
+01/27/2022 14:11:16 - INFO - codeparrot_training - Step 19412: {'lr': 0.000354486588626188, 'samples': 3727296, 'steps': 19412, 'loss/train': 0.7860661447048187}
+01/27/2022 14:11:20 - INFO - codeparrot_training - Step 19413: {'lr': 0.0003544717235688385, 'samples': 3727488, 'steps': 19413, 'loss/train': 0.7623594403266907}
+01/27/2022 14:11:23 - INFO - codeparrot_training - Step 19414: {'lr': 0.00035445685806396543, 'samples': 3727680, 'steps': 19414, 'loss/train': 1.2226037979125977}
+01/27/2022 14:11:26 - INFO - codeparrot_training - Step 19415: {'lr': 0.00035444199211163226, 'samples': 3727872, 'steps': 19415, 'loss/train': 0.5843183845281601}
+01/27/2022 14:11:29 - INFO - codeparrot_training - Step 19416: {'lr': 0.0003544271257119028, 'samples': 3728064, 'steps': 19416, 'loss/train': 0.9046416878700256}
+01/27/2022 14:11:32 - INFO - codeparrot_training - Step 19417: {'lr': 0.00035441225886484066, 'samples': 3728256, 'steps': 19417, 'loss/train': 1.0531883239746094}
+01/27/2022 14:11:35 - INFO - codeparrot_training - Step 19418: {'lr': 0.0003543973915705095, 'samples': 3728448, 'steps': 19418, 'loss/train': 0.901370108127594}
+01/27/2022 14:11:38 - INFO - codeparrot_training - Step 19419: {'lr': 0.0003543825238289732, 'samples': 3728640, 'steps': 19419, 'loss/train': 0.3198922947049141}
+01/27/2022 14:11:42 - INFO - codeparrot_training - Step 19420: {'lr': 0.0003543676556402952, 'samples': 3728832, 'steps': 19420, 'loss/train': 1.1213891208171844}
+01/27/2022 14:11:46 - INFO - codeparrot_training - Step 19421: {'lr': 0.00035435278700453934, 'samples': 3729024, 'steps': 19421, 'loss/train': 0.723642960190773}
+01/27/2022 14:11:49 - INFO - codeparrot_training - Step 19422: {'lr': 0.0003543379179217693, 'samples': 3729216, 'steps': 19422, 'loss/train': 0.7832135260105133}
+01/27/2022 14:11:52 - INFO - codeparrot_training - Step 19423: {'lr': 0.0003543230483920487, 'samples': 3729408, 'steps': 19423, 'loss/train': 0.5043981224298477}
+01/27/2022 14:11:55 - INFO - codeparrot_training - Step 19424: {'lr': 0.0003543081784154414, 'samples': 3729600, 'steps': 19424, 'loss/train': 0.5785528421401978}
+01/27/2022 14:11:59 - INFO - codeparrot_training - Step 19425: {'lr': 0.00035429330799201085, 'samples': 3729792, 'steps': 19425, 'loss/train': 0.5320127159357071}
+01/27/2022 14:12:02 - INFO - codeparrot_training - Step 19426: {'lr': 0.00035427843712182097, 'samples': 3729984, 'steps': 19426, 'loss/train': 0.8111559748649597}
+01/27/2022 14:12:05 - INFO - codeparrot_training - Step 19427: {'lr': 0.0003542635658049353, 'samples': 3730176, 'steps': 19427, 'loss/train': 0.6566714644432068}
+01/27/2022 14:12:08 - INFO - codeparrot_training - Step 19428: {'lr': 0.0003542486940414177, 'samples': 3730368, 'steps': 19428, 'loss/train': 0.8711982071399689}
+01/27/2022 14:12:11 - INFO - codeparrot_training - Step 19429: {'lr': 0.00035423382183133174, 'samples': 3730560, 'steps': 19429, 'loss/train': 0.7827341258525848}
+01/27/2022 14:12:16 - INFO - codeparrot_training - Step 19430: {'lr': 0.00035421894917474125, 'samples': 3730752, 'steps': 19430, 'loss/train': 1.260263353586197}
+01/27/2022 14:12:20 - INFO - codeparrot_training - Step 19431: {'lr': 0.0003542040760717098, 'samples': 3730944, 'steps': 19431, 'loss/train': 0.8573777675628662}
+01/27/2022 14:12:23 - INFO - codeparrot_training - Step 19432: {'lr': 0.0003541892025223012, 'samples': 3731136, 'steps': 19432, 'loss/train': 1.0051724016666412}
+01/27/2022 14:12:26 - INFO - codeparrot_training - Step 19433: {'lr': 0.00035417432852657916, 'samples': 3731328, 'steps': 19433, 'loss/train': 0.9147574603557587}
+01/27/2022 14:12:29 - INFO - codeparrot_training - Step 19434: {'lr': 0.00035415945408460737, 'samples': 3731520, 'steps': 19434, 'loss/train': 0.9861118197441101}
+01/27/2022 14:12:32 - INFO - codeparrot_training - Step 19435: {'lr': 0.0003541445791964496, 'samples': 3731712, 'steps': 19435, 'loss/train': 1.1434222161769867}
+01/27/2022 14:12:35 - INFO - codeparrot_training - Step 19436: {'lr': 0.0003541297038621694, 'samples': 3731904, 'steps': 19436, 'loss/train': 0.8629707098007202}
+01/27/2022 14:12:38 - INFO - codeparrot_training - Step 19437: {'lr': 0.0003541148280818307, 'samples': 3732096, 'steps': 19437, 'loss/train': 1.7869332432746887}
+01/27/2022 14:12:41 - INFO - codeparrot_training - Step 19438: {'lr': 0.00035409995185549717, 'samples': 3732288, 'steps': 19438, 'loss/train': 0.6054085940122604}
+01/27/2022 14:12:46 - INFO - codeparrot_training - Step 19439: {'lr': 0.00035408507518323244, 'samples': 3732480, 'steps': 19439, 'loss/train': 0.9436999261379242}
+01/27/2022 14:12:49 - INFO - codeparrot_training - Step 19440: {'lr': 0.00035407019806510035, 'samples': 3732672, 'steps': 19440, 'loss/train': 0.4943774342536926}
+01/27/2022 14:12:52 - INFO - codeparrot_training - Step 19441: {'lr': 0.0003540553205011645, 'samples': 3732864, 'steps': 19441, 'loss/train': 0.7497723251581192}
+01/27/2022 14:12:55 - INFO - codeparrot_training - Step 19442: {'lr': 0.00035404044249148873, 'samples': 3733056, 'steps': 19442, 'loss/train': 0.6638262122869492}
+01/27/2022 14:12:58 - INFO - codeparrot_training - Step 19443: {'lr': 0.0003540255640361368, 'samples': 3733248, 'steps': 19443, 'loss/train': 0.9967392683029175}
+01/27/2022 14:13:01 - INFO - codeparrot_training - Step 19444: {'lr': 0.0003540106851351723, 'samples': 3733440, 'steps': 19444, 'loss/train': 0.8511635363101959}
+01/27/2022 14:13:05 - INFO - codeparrot_training - Step 19445: {'lr': 0.00035399580578865907, 'samples': 3733632, 'steps': 19445, 'loss/train': 1.1101444065570831}
+01/27/2022 14:13:08 - INFO - codeparrot_training - Step 19446: {'lr': 0.00035398092599666086, 'samples': 3733824, 'steps': 19446, 'loss/train': 1.0562118887901306}
+01/27/2022 14:13:11 - INFO - codeparrot_training - Step 19447: {'lr': 0.00035396604575924133, 'samples': 3734016, 'steps': 19447, 'loss/train': 0.7500544488430023}
+01/27/2022 14:13:15 - INFO - codeparrot_training - Step 19448: {'lr': 0.00035395116507646435, 'samples': 3734208, 'steps': 19448, 'loss/train': 2.2485328316688538}
+01/27/2022 14:13:19 - INFO - codeparrot_training - Step 19449: {'lr': 0.00035393628394839356, 'samples': 3734400, 'steps': 19449, 'loss/train': 0.9011058211326599}
+01/27/2022 14:13:22 - INFO - codeparrot_training - Step 19450: {'lr': 0.00035392140237509276, 'samples': 3734592, 'steps': 19450, 'loss/train': 0.2965891510248184}
+01/27/2022 14:13:25 - INFO - codeparrot_training - Step 19451: {'lr': 0.0003539065203566256, 'samples': 3734784, 'steps': 19451, 'loss/train': 0.7498652637004852}
+01/27/2022 14:13:28 - INFO - codeparrot_training - Step 19452: {'lr': 0.0003538916378930559, 'samples': 3734976, 'steps': 19452, 'loss/train': 0.9723486006259918}
+01/27/2022 14:13:31 - INFO - codeparrot_training - Step 19453: {'lr': 0.0003538767549844475, 'samples': 3735168, 'steps': 19453, 'loss/train': 0.9272070229053497}
+01/27/2022 14:13:34 - INFO - codeparrot_training - Step 19454: {'lr': 0.000353861871630864, 'samples': 3735360, 'steps': 19454, 'loss/train': 0.4280697852373123}
+01/27/2022 14:13:37 - INFO - codeparrot_training - Step 19455: {'lr': 0.00035384698783236923, 'samples': 3735552, 'steps': 19455, 'loss/train': 0.6031343042850494}
+01/27/2022 14:13:41 - INFO - codeparrot_training - Step 19456: {'lr': 0.00035383210358902695, 'samples': 3735744, 'steps': 19456, 'loss/train': 0.6464528292417526}
+01/27/2022 14:13:45 - INFO - codeparrot_training - Step 19457: {'lr': 0.0003538172189009009, 'samples': 3735936, 'steps': 19457, 'loss/train': 0.7536343932151794}
+01/27/2022 14:13:48 - INFO - codeparrot_training - Step 19458: {'lr': 0.00035380233376805487, 'samples': 3736128, 'steps': 19458, 'loss/train': 0.5618426352739334}
+01/27/2022 14:13:51 - INFO - codeparrot_training - Step 19459: {'lr': 0.00035378744819055264, 'samples': 3736320, 'steps': 19459, 'loss/train': 0.6686100214719772}
+01/27/2022 14:13:54 - INFO - codeparrot_training - Step 19460: {'lr': 0.0003537725621684578, 'samples': 3736512, 'steps': 19460, 'loss/train': 0.4414688050746918}
+01/27/2022 14:13:58 - INFO - codeparrot_training - Step 19461: {'lr': 0.0003537576757018344, 'samples': 3736704, 'steps': 19461, 'loss/train': 0.5290847718715668}
+01/27/2022 14:14:01 - INFO - codeparrot_training - Step 19462: {'lr': 0.0003537427887907459, 'samples': 3736896, 'steps': 19462, 'loss/train': 0.870970755815506}
+01/27/2022 14:14:04 - INFO - codeparrot_training - Step 19463: {'lr': 0.0003537279014352565, 'samples': 3737088, 'steps': 19463, 'loss/train': 0.6359722763299942}
+01/27/2022 14:14:07 - INFO - codeparrot_training - Step 19464: {'lr': 0.00035371301363542945, 'samples': 3737280, 'steps': 19464, 'loss/train': 0.44914716482162476}
+01/27/2022 14:14:12 - INFO - codeparrot_training - Step 19465: {'lr': 0.00035369812539132894, 'samples': 3737472, 'steps': 19465, 'loss/train': 0.6865744292736053}
+01/27/2022 14:14:15 - INFO - codeparrot_training - Step 19466: {'lr': 0.0003536832367030185, 'samples': 3737664, 'steps': 19466, 'loss/train': 0.6599008440971375}
+01/27/2022 14:14:18 - INFO - codeparrot_training - Step 19467: {'lr': 0.00035366834757056203, 'samples': 3737856, 'steps': 19467, 'loss/train': 0.8847377300262451}
+01/27/2022 14:14:22 - INFO - codeparrot_training - Step 19468: {'lr': 0.0003536534579940233, 'samples': 3738048, 'steps': 19468, 'loss/train': 0.9295083582401276}
+01/27/2022 14:14:25 - INFO - codeparrot_training - Step 19469: {'lr': 0.0003536385679734659, 'samples': 3738240, 'steps': 19469, 'loss/train': 0.9306831657886505}
+01/27/2022 14:14:28 - INFO - codeparrot_training - Step 19470: {'lr': 0.000353623677508954, 'samples': 3738432, 'steps': 19470, 'loss/train': 0.776753157377243}
+01/27/2022 14:14:31 - INFO - codeparrot_training - Step 19471: {'lr': 0.00035360878660055107, 'samples': 3738624, 'steps': 19471, 'loss/train': 0.8190747499465942}
+01/27/2022 14:14:34 - INFO - codeparrot_training - Step 19472: {'lr': 0.0003535938952483211, 'samples': 3738816, 'steps': 19472, 'loss/train': 0.5463986992835999}
+01/27/2022 14:14:37 - INFO - codeparrot_training - Step 19473: {'lr': 0.00035357900345232764, 'samples': 3739008, 'steps': 19473, 'loss/train': 0.2329401969909668}
+01/27/2022 14:14:42 - INFO - codeparrot_training - Step 19474: {'lr': 0.0003535641112126347, 'samples': 3739200, 'steps': 19474, 'loss/train': 0.9679434299468994}
+01/27/2022 14:14:45 - INFO - codeparrot_training - Step 19475: {'lr': 0.00035354921852930596, 'samples': 3739392, 'steps': 19475, 'loss/train': 0.7869312465190887}
+01/27/2022 14:14:48 - INFO - codeparrot_training - Step 19476: {'lr': 0.0003535343254024053, 'samples': 3739584, 'steps': 19476, 'loss/train': 0.6084536612033844}
+01/27/2022 14:14:51 - INFO - codeparrot_training - Step 19477: {'lr': 0.00035351943183199643, 'samples': 3739776, 'steps': 19477, 'loss/train': 0.5082223266363144}
+01/27/2022 14:14:54 - INFO - codeparrot_training - Step 19478: {'lr': 0.0003535045378181432, 'samples': 3739968, 'steps': 19478, 'loss/train': 0.9243129193782806}
+01/27/2022 14:14:57 - INFO - codeparrot_training - Step 19479: {'lr': 0.0003534896433609093, 'samples': 3740160, 'steps': 19479, 'loss/train': 0.5868056863546371}
+01/27/2022 14:15:01 - INFO - codeparrot_training - Step 19480: {'lr': 0.0003534747484603587, 'samples': 3740352, 'steps': 19480, 'loss/train': 0.6190521419048309}
+01/27/2022 14:15:04 - INFO - codeparrot_training - Step 19481: {'lr': 0.00035345985311655513, 'samples': 3740544, 'steps': 19481, 'loss/train': 0.8864107131958008}
+01/27/2022 14:15:07 - INFO - codeparrot_training - Step 19482: {'lr': 0.0003534449573295624, 'samples': 3740736, 'steps': 19482, 'loss/train': 0.6944574415683746}
+01/27/2022 14:15:11 - INFO - codeparrot_training - Step 19483: {'lr': 0.0003534300610994444, 'samples': 3740928, 'steps': 19483, 'loss/train': 1.1989716589450836}
+01/27/2022 14:15:15 - INFO - codeparrot_training - Step 19484: {'lr': 0.00035341516442626475, 'samples': 3741120, 'steps': 19484, 'loss/train': 0.6869207918643951}
+01/27/2022 14:15:18 - INFO - codeparrot_training - Step 19485: {'lr': 0.00035340026731008745, 'samples': 3741312, 'steps': 19485, 'loss/train': 0.7209068387746811}
+01/27/2022 14:15:21 - INFO - codeparrot_training - Step 19486: {'lr': 0.0003533853697509762, 'samples': 3741504, 'steps': 19486, 'loss/train': 0.647133395075798}
+01/27/2022 14:15:24 - INFO - codeparrot_training - Step 19487: {'lr': 0.0003533704717489949, 'samples': 3741696, 'steps': 19487, 'loss/train': 0.8474768400192261}
+01/27/2022 14:15:27 - INFO - codeparrot_training - Step 19488: {'lr': 0.00035335557330420724, 'samples': 3741888, 'steps': 19488, 'loss/train': 1.3505685925483704}
+01/27/2022 14:15:30 - INFO - codeparrot_training - Step 19489: {'lr': 0.0003533406744166772, 'samples': 3742080, 'steps': 19489, 'loss/train': 0.4493958503007889}
+01/27/2022 14:15:33 - INFO - codeparrot_training - Step 19490: {'lr': 0.00035332577508646846, 'samples': 3742272, 'steps': 19490, 'loss/train': 1.5691915154457092}
+01/27/2022 14:15:37 - INFO - codeparrot_training - Step 19491: {'lr': 0.000353310875313645, 'samples': 3742464, 'steps': 19491, 'loss/train': 0.7112170904874802}
+01/27/2022 14:15:42 - INFO - codeparrot_training - Step 19492: {'lr': 0.00035329597509827046, 'samples': 3742656, 'steps': 19492, 'loss/train': 0.9918762445449829}
+01/27/2022 14:15:45 - INFO - codeparrot_training - Step 19493: {'lr': 0.00035328107444040875, 'samples': 3742848, 'steps': 19493, 'loss/train': 0.675085186958313}
+01/27/2022 14:15:48 - INFO - codeparrot_training - Step 19494: {'lr': 0.0003532661733401238, 'samples': 3743040, 'steps': 19494, 'loss/train': 0.8099081218242645}
+01/27/2022 14:15:51 - INFO - codeparrot_training - Step 19495: {'lr': 0.0003532512717974793, 'samples': 3743232, 'steps': 19495, 'loss/train': 1.0101793706417084}
+01/27/2022 14:15:54 - INFO - codeparrot_training - Step 19496: {'lr': 0.00035323636981253914, 'samples': 3743424, 'steps': 19496, 'loss/train': 1.2552567422389984}
+01/27/2022 14:15:57 - INFO - codeparrot_training - Step 19497: {'lr': 0.00035322146738536714, 'samples': 3743616, 'steps': 19497, 'loss/train': 1.2592334747314453}
+01/27/2022 14:16:01 - INFO - codeparrot_training - Step 19498: {'lr': 0.00035320656451602725, 'samples': 3743808, 'steps': 19498, 'loss/train': 0.569161981344223}
+01/27/2022 14:16:04 - INFO - codeparrot_training - Step 19499: {'lr': 0.00035319166120458315, 'samples': 3744000, 'steps': 19499, 'loss/train': 0.6892805099487305}
+01/27/2022 14:16:07 - INFO - codeparrot_training - Step 19500: {'lr': 0.00035317675745109866, 'samples': 3744192, 'steps': 19500, 'loss/train': 0.7300190180540085}
+01/27/2022 14:16:11 - INFO - codeparrot_training - Step 19501: {'lr': 0.0003531618532556378, 'samples': 3744384, 'steps': 19501, 'loss/train': 0.789417564868927}
+01/27/2022 14:16:14 - INFO - codeparrot_training - Step 19502: {'lr': 0.00035314694861826427, 'samples': 3744576, 'steps': 19502, 'loss/train': 0.624591663479805}
+01/27/2022 14:16:18 - INFO - codeparrot_training - Step 19503: {'lr': 0.00035313204353904203, 'samples': 3744768, 'steps': 19503, 'loss/train': 0.8117698431015015}
+01/27/2022 14:16:21 - INFO - codeparrot_training - Step 19504: {'lr': 0.0003531171380180348, 'samples': 3744960, 'steps': 19504, 'loss/train': 0.7787246704101562}
+01/27/2022 14:16:24 - INFO - codeparrot_training - Step 19505: {'lr': 0.00035310223205530655, 'samples': 3745152, 'steps': 19505, 'loss/train': 0.6784188151359558}
+01/27/2022 14:16:27 - INFO - codeparrot_training - Step 19506: {'lr': 0.00035308732565092097, 'samples': 3745344, 'steps': 19506, 'loss/train': 0.5127600878477097}
+01/27/2022 14:16:30 - INFO - codeparrot_training - Step 19507: {'lr': 0.0003530724188049421, 'samples': 3745536, 'steps': 19507, 'loss/train': 0.5809745639562607}
+01/27/2022 14:16:33 - INFO - codeparrot_training - Step 19508: {'lr': 0.0003530575115174337, 'samples': 3745728, 'steps': 19508, 'loss/train': 0.7990142405033112}
+01/27/2022 14:16:36 - INFO - codeparrot_training - Step 19509: {'lr': 0.0003530426037884597, 'samples': 3745920, 'steps': 19509, 'loss/train': 0.6677492558956146}
+01/27/2022 14:16:41 - INFO - codeparrot_training - Step 19510: {'lr': 0.00035302769561808387, 'samples': 3746112, 'steps': 19510, 'loss/train': 1.0593432784080505}
+01/27/2022 14:16:45 - INFO - codeparrot_training - Step 19511: {'lr': 0.00035301278700637005, 'samples': 3746304, 'steps': 19511, 'loss/train': 1.0085486769676208}
+01/27/2022 14:16:48 - INFO - codeparrot_training - Step 19512: {'lr': 0.0003529978779533822, 'samples': 3746496, 'steps': 19512, 'loss/train': 0.5501835197210312}
+01/27/2022 14:16:51 - INFO - codeparrot_training - Step 19513: {'lr': 0.0003529829684591842, 'samples': 3746688, 'steps': 19513, 'loss/train': 1.438300609588623}
+01/27/2022 14:16:54 - INFO - codeparrot_training - Step 19514: {'lr': 0.0003529680585238398, 'samples': 3746880, 'steps': 19514, 'loss/train': 0.9882878065109253}
+01/27/2022 14:16:57 - INFO - codeparrot_training - Step 19515: {'lr': 0.000352953148147413, 'samples': 3747072, 'steps': 19515, 'loss/train': 0.9454243183135986}
+01/27/2022 14:17:00 - INFO - codeparrot_training - Step 19516: {'lr': 0.0003529382373299675, 'samples': 3747264, 'steps': 19516, 'loss/train': 1.0225533843040466}
+01/27/2022 14:17:03 - INFO - codeparrot_training - Step 19517: {'lr': 0.0003529233260715673, 'samples': 3747456, 'steps': 19517, 'loss/train': 1.2328218519687653}
+01/27/2022 14:17:07 - INFO - codeparrot_training - Step 19518: {'lr': 0.00035290841437227627, 'samples': 3747648, 'steps': 19518, 'loss/train': 0.6179350912570953}
+01/27/2022 14:17:11 - INFO - codeparrot_training - Step 19519: {'lr': 0.00035289350223215827, 'samples': 3747840, 'steps': 19519, 'loss/train': 1.329677939414978}
+01/27/2022 14:17:14 - INFO - codeparrot_training - Step 19520: {'lr': 0.00035287858965127723, 'samples': 3748032, 'steps': 19520, 'loss/train': 1.0303166806697845}
+01/27/2022 14:17:18 - INFO - codeparrot_training - Step 19521: {'lr': 0.0003528636766296969, 'samples': 3748224, 'steps': 19521, 'loss/train': 1.1714334189891815}
+01/27/2022 14:17:21 - INFO - codeparrot_training - Step 19522: {'lr': 0.0003528487631674813, 'samples': 3748416, 'steps': 19522, 'loss/train': 0.9679166972637177}
+01/27/2022 14:17:24 - INFO - codeparrot_training - Step 19523: {'lr': 0.00035283384926469426, 'samples': 3748608, 'steps': 19523, 'loss/train': 1.054144710302353}
+01/27/2022 14:17:27 - INFO - codeparrot_training - Step 19524: {'lr': 0.00035281893492139965, 'samples': 3748800, 'steps': 19524, 'loss/train': 0.9691188633441925}
+01/27/2022 14:17:30 - INFO - codeparrot_training - Step 19525: {'lr': 0.0003528040201376613, 'samples': 3748992, 'steps': 19525, 'loss/train': 0.911686509847641}
+01/27/2022 14:17:33 - INFO - codeparrot_training - Step 19526: {'lr': 0.00035278910491354335, 'samples': 3749184, 'steps': 19526, 'loss/train': 0.8264159560203552}
+01/27/2022 14:17:38 - INFO - codeparrot_training - Step 19527: {'lr': 0.0003527741892491093, 'samples': 3749376, 'steps': 19527, 'loss/train': 0.9428652226924896}
+01/27/2022 14:17:41 - INFO - codeparrot_training - Step 19528: {'lr': 0.00035275927314442344, 'samples': 3749568, 'steps': 19528, 'loss/train': 0.30977898091077805}
+01/27/2022 14:17:44 - INFO - codeparrot_training - Step 19529: {'lr': 0.00035274435659954936, 'samples': 3749760, 'steps': 19529, 'loss/train': 0.9585222601890564}
+01/27/2022 14:17:47 - INFO - codeparrot_training - Step 19530: {'lr': 0.00035272943961455106, 'samples': 3749952, 'steps': 19530, 'loss/train': 0.7067334651947021}
+01/27/2022 14:17:51 - INFO - codeparrot_training - Step 19531: {'lr': 0.00035271452218949256, 'samples': 3750144, 'steps': 19531, 'loss/train': 1.2381550669670105}
+01/27/2022 14:17:54 - INFO - codeparrot_training - Step 19532: {'lr': 0.0003526996043244376, 'samples': 3750336, 'steps': 19532, 'loss/train': 0.9682613611221313}
+01/27/2022 14:17:57 - INFO - codeparrot_training - Step 19533: {'lr': 0.0003526846860194502, 'samples': 3750528, 'steps': 19533, 'loss/train': 0.8318165838718414}
+01/27/2022 14:18:00 - INFO - codeparrot_training - Step 19534: {'lr': 0.0003526697672745942, 'samples': 3750720, 'steps': 19534, 'loss/train': 1.1263751685619354}
+01/27/2022 14:18:03 - INFO - codeparrot_training - Step 19535: {'lr': 0.0003526548480899335, 'samples': 3750912, 'steps': 19535, 'loss/train': 0.9373343288898468}
+01/27/2022 14:18:06 - INFO - codeparrot_training - Step 19536: {'lr': 0.000352639928465532, 'samples': 3751104, 'steps': 19536, 'loss/train': 0.4764983654022217}
+01/27/2022 14:18:12 - INFO - codeparrot_training - Step 19537: {'lr': 0.00035262500840145365, 'samples': 3751296, 'steps': 19537, 'loss/train': 0.6796870976686478}
+01/27/2022 14:18:15 - INFO - codeparrot_training - Step 19538: {'lr': 0.00035261008789776237, 'samples': 3751488, 'steps': 19538, 'loss/train': 0.7539314031600952}
+01/27/2022 14:18:18 - INFO - codeparrot_training - Step 19539: {'lr': 0.0003525951669545221, 'samples': 3751680, 'steps': 19539, 'loss/train': 1.122983068227768}
+01/27/2022 14:18:21 - INFO - codeparrot_training - Step 19540: {'lr': 0.00035258024557179664, 'samples': 3751872, 'steps': 19540, 'loss/train': 0.07513032294809818}
+01/27/2022 14:18:25 - INFO - codeparrot_training - Step 19541: {'lr': 0.00035256532374965, 'samples': 3752064, 'steps': 19541, 'loss/train': 0.4791265279054642}
+01/27/2022 14:18:28 - INFO - codeparrot_training - Step 19542: {'lr': 0.0003525504014881461, 'samples': 3752256, 'steps': 19542, 'loss/train': 1.0117066204547882}
+01/27/2022 14:18:31 - INFO - codeparrot_training - Step 19543: {'lr': 0.0003525354787873488, 'samples': 3752448, 'steps': 19543, 'loss/train': 0.06842857226729393}
+01/27/2022 14:18:34 - INFO - codeparrot_training - Step 19544: {'lr': 0.0003525205556473221, 'samples': 3752640, 'steps': 19544, 'loss/train': 0.6727123260498047}
+01/27/2022 14:18:37 - INFO - codeparrot_training - Step 19545: {'lr': 0.0003525056320681299, 'samples': 3752832, 'steps': 19545, 'loss/train': 0.8313365578651428}
+01/27/2022 14:18:42 - INFO - codeparrot_training - Step 19546: {'lr': 0.00035249070804983616, 'samples': 3753024, 'steps': 19546, 'loss/train': 0.6173986047506332}
+01/27/2022 14:18:45 - INFO - codeparrot_training - Step 19547: {'lr': 0.00035247578359250473, 'samples': 3753216, 'steps': 19547, 'loss/train': 0.7741822600364685}
+01/27/2022 14:18:48 - INFO - codeparrot_training - Step 19548: {'lr': 0.0003524608586961996, 'samples': 3753408, 'steps': 19548, 'loss/train': 1.317454844713211}
+01/27/2022 14:18:51 - INFO - codeparrot_training - Step 19549: {'lr': 0.00035244593336098464, 'samples': 3753600, 'steps': 19549, 'loss/train': 0.08971647173166275}
+01/27/2022 14:18:54 - INFO - codeparrot_training - Step 19550: {'lr': 0.0003524310075869239, 'samples': 3753792, 'steps': 19550, 'loss/train': 1.1492847204208374}
+01/27/2022 14:18:57 - INFO - codeparrot_training - Step 19551: {'lr': 0.0003524160813740812, 'samples': 3753984, 'steps': 19551, 'loss/train': 1.0013016164302826}
+01/27/2022 14:19:00 - INFO - codeparrot_training - Step 19552: {'lr': 0.00035240115472252056, 'samples': 3754176, 'steps': 19552, 'loss/train': 1.1041552126407623}
+01/27/2022 14:19:04 - INFO - codeparrot_training - Step 19553: {'lr': 0.00035238622763230574, 'samples': 3754368, 'steps': 19553, 'loss/train': 0.5667447298765182}
+01/27/2022 14:19:08 - INFO - codeparrot_training - Step 19554: {'lr': 0.000352371300103501, 'samples': 3754560, 'steps': 19554, 'loss/train': 0.7588256299495697}
+01/27/2022 14:19:11 - INFO - codeparrot_training - Step 19555: {'lr': 0.0003523563721361701, 'samples': 3754752, 'steps': 19555, 'loss/train': 0.2501487135887146}
+01/27/2022 14:19:14 - INFO - codeparrot_training - Step 19556: {'lr': 0.0003523414437303769, 'samples': 3754944, 'steps': 19556, 'loss/train': 0.9237551987171173}
+01/27/2022 14:19:17 - INFO - codeparrot_training - Step 19557: {'lr': 0.00035232651488618564, 'samples': 3755136, 'steps': 19557, 'loss/train': 0.9538207948207855}
+01/27/2022 14:19:21 - INFO - codeparrot_training - Step 19558: {'lr': 0.00035231158560365987, 'samples': 3755328, 'steps': 19558, 'loss/train': 0.8491564393043518}
+01/27/2022 14:19:24 - INFO - codeparrot_training - Step 19559: {'lr': 0.0003522966558828639, 'samples': 3755520, 'steps': 19559, 'loss/train': 1.0922627449035645}
+01/27/2022 14:19:27 - INFO - codeparrot_training - Step 19560: {'lr': 0.00035228172572386146, 'samples': 3755712, 'steps': 19560, 'loss/train': 1.1534889936447144}
+01/27/2022 14:19:30 - INFO - codeparrot_training - Step 19561: {'lr': 0.00035226679512671664, 'samples': 3755904, 'steps': 19561, 'loss/train': 0.5551798045635223}
+01/27/2022 14:19:33 - INFO - codeparrot_training - Step 19562: {'lr': 0.00035225186409149333, 'samples': 3756096, 'steps': 19562, 'loss/train': 0.7213871330022812}
+01/27/2022 14:19:38 - INFO - codeparrot_training - Step 19563: {'lr': 0.00035223693261825554, 'samples': 3756288, 'steps': 19563, 'loss/train': 0.9676121771335602}
+01/27/2022 14:19:41 - INFO - codeparrot_training - Step 19564: {'lr': 0.0003522220007070671, 'samples': 3756480, 'steps': 19564, 'loss/train': 0.7174867987632751}
+01/27/2022 14:19:45 - INFO - codeparrot_training - Step 19565: {'lr': 0.0003522070683579921, 'samples': 3756672, 'steps': 19565, 'loss/train': 1.0418184399604797}
+01/27/2022 14:19:48 - INFO - codeparrot_training - Step 19566: {'lr': 0.00035219213557109453, 'samples': 3756864, 'steps': 19566, 'loss/train': 0.6766509711742401}
+01/27/2022 14:19:51 - INFO - codeparrot_training - Step 19567: {'lr': 0.00035217720234643823, 'samples': 3757056, 'steps': 19567, 'loss/train': 0.871608316898346}
+01/27/2022 14:19:54 - INFO - codeparrot_training - Step 19568: {'lr': 0.0003521622686840873, 'samples': 3757248, 'steps': 19568, 'loss/train': 1.3245534896850586}
+01/27/2022 14:19:57 - INFO - codeparrot_training - Step 19569: {'lr': 0.00035214733458410557, 'samples': 3757440, 'steps': 19569, 'loss/train': 0.09151941910386086}
+01/27/2022 14:20:00 - INFO - codeparrot_training - Step 19570: {'lr': 0.00035213240004655714, 'samples': 3757632, 'steps': 19570, 'loss/train': 0.33582763373851776}
+01/27/2022 14:20:03 - INFO - codeparrot_training - Step 19571: {'lr': 0.000352117465071506, 'samples': 3757824, 'steps': 19571, 'loss/train': 0.3808223605155945}
+01/27/2022 14:20:08 - INFO - codeparrot_training - Step 19572: {'lr': 0.000352102529659016, 'samples': 3758016, 'steps': 19572, 'loss/train': 0.4737963527441025}
+01/27/2022 14:20:11 - INFO - codeparrot_training - Step 19573: {'lr': 0.00035208759380915116, 'samples': 3758208, 'steps': 19573, 'loss/train': 0.290145568549633}
+01/27/2022 14:20:14 - INFO - codeparrot_training - Step 19574: {'lr': 0.0003520726575219756, 'samples': 3758400, 'steps': 19574, 'loss/train': 0.8872269988059998}
+01/27/2022 14:20:17 - INFO - codeparrot_training - Step 19575: {'lr': 0.00035205772079755307, 'samples': 3758592, 'steps': 19575, 'loss/train': 0.8332065045833588}
+01/27/2022 14:20:21 - INFO - codeparrot_training - Step 19576: {'lr': 0.0003520427836359477, 'samples': 3758784, 'steps': 19576, 'loss/train': 1.0248441696166992}
+01/27/2022 14:20:24 - INFO - codeparrot_training - Step 19577: {'lr': 0.00035202784603722344, 'samples': 3758976, 'steps': 19577, 'loss/train': 0.5091615915298462}
+01/27/2022 14:20:27 - INFO - codeparrot_training - Step 19578: {'lr': 0.00035201290800144423, 'samples': 3759168, 'steps': 19578, 'loss/train': 0.643872857093811}
+01/27/2022 14:20:30 - INFO - codeparrot_training - Step 19579: {'lr': 0.00035199796952867425, 'samples': 3759360, 'steps': 19579, 'loss/train': 0.9586743414402008}
+01/27/2022 14:20:33 - INFO - codeparrot_training - Step 19580: {'lr': 0.0003519830306189773, 'samples': 3759552, 'steps': 19580, 'loss/train': 0.26058726757764816}
+01/27/2022 14:20:38 - INFO - codeparrot_training - Step 19581: {'lr': 0.0003519680912724174, 'samples': 3759744, 'steps': 19581, 'loss/train': 1.2268230020999908}
+01/27/2022 14:20:41 - INFO - codeparrot_training - Step 19582: {'lr': 0.0003519531514890586, 'samples': 3759936, 'steps': 19582, 'loss/train': 0.8443522453308105}
+01/27/2022 14:20:44 - INFO - codeparrot_training - Step 19583: {'lr': 0.00035193821126896493, 'samples': 3760128, 'steps': 19583, 'loss/train': 1.0709995329380035}
+01/27/2022 14:20:47 - INFO - codeparrot_training - Step 19584: {'lr': 0.0003519232706122002, 'samples': 3760320, 'steps': 19584, 'loss/train': 0.7127667814493179}
+01/27/2022 14:20:50 - INFO - codeparrot_training - Step 19585: {'lr': 0.0003519083295188287, 'samples': 3760512, 'steps': 19585, 'loss/train': 0.9748610258102417}
+01/27/2022 14:20:53 - INFO - codeparrot_training - Step 19586: {'lr': 0.0003518933879889142, 'samples': 3760704, 'steps': 19586, 'loss/train': 1.021315723657608}
+01/27/2022 14:20:57 - INFO - codeparrot_training - Step 19587: {'lr': 0.00035187844602252075, 'samples': 3760896, 'steps': 19587, 'loss/train': 1.0221087634563446}
+01/27/2022 14:21:00 - INFO - codeparrot_training - Step 19588: {'lr': 0.0003518635036197124, 'samples': 3761088, 'steps': 19588, 'loss/train': 0.5292531251907349}
+01/27/2022 14:21:05 - INFO - codeparrot_training - Step 19589: {'lr': 0.00035184856078055316, 'samples': 3761280, 'steps': 19589, 'loss/train': 0.9478439390659332}
+01/27/2022 14:21:08 - INFO - codeparrot_training - Step 19590: {'lr': 0.000351833617505107, 'samples': 3761472, 'steps': 19590, 'loss/train': 1.375247597694397}
+01/27/2022 14:21:11 - INFO - codeparrot_training - Step 19591: {'lr': 0.000351818673793438, 'samples': 3761664, 'steps': 19591, 'loss/train': 1.0605688691139221}
+01/27/2022 14:21:14 - INFO - codeparrot_training - Step 19592: {'lr': 0.00035180372964561013, 'samples': 3761856, 'steps': 19592, 'loss/train': 0.6668794602155685}
+01/27/2022 14:21:17 - INFO - codeparrot_training - Step 19593: {'lr': 0.00035178878506168733, 'samples': 3762048, 'steps': 19593, 'loss/train': 0.903924822807312}
+01/27/2022 14:21:21 - INFO - codeparrot_training - Step 19594: {'lr': 0.0003517738400417338, 'samples': 3762240, 'steps': 19594, 'loss/train': 0.8968283236026764}
+01/27/2022 14:21:24 - INFO - codeparrot_training - Step 19595: {'lr': 0.0003517588945858134, 'samples': 3762432, 'steps': 19595, 'loss/train': 0.9576274752616882}
+01/27/2022 14:21:27 - INFO - codeparrot_training - Step 19596: {'lr': 0.00035174394869399024, 'samples': 3762624, 'steps': 19596, 'loss/train': 0.9424448311328888}
+01/27/2022 14:21:30 - INFO - codeparrot_training - Step 19597: {'lr': 0.0003517290023663283, 'samples': 3762816, 'steps': 19597, 'loss/train': 0.550383523106575}
+01/27/2022 14:21:34 - INFO - codeparrot_training - Step 19598: {'lr': 0.0003517140556028916, 'samples': 3763008, 'steps': 19598, 'loss/train': 0.9588927626609802}
+01/27/2022 14:21:38 - INFO - codeparrot_training - Step 19599: {'lr': 0.0003516991084037442, 'samples': 3763200, 'steps': 19599, 'loss/train': 0.6846987158060074}
+01/27/2022 14:21:41 - INFO - codeparrot_training - Step 19600: {'lr': 0.0003516841607689501, 'samples': 3763392, 'steps': 19600, 'loss/train': 0.46922729909420013}
+01/27/2022 14:21:44 - INFO - codeparrot_training - Step 19601: {'lr': 0.0003516692126985733, 'samples': 3763584, 'steps': 19601, 'loss/train': 0.8294418454170227}
+01/27/2022 14:21:47 - INFO - codeparrot_training - Step 19602: {'lr': 0.00035165426419267795, 'samples': 3763776, 'steps': 19602, 'loss/train': 0.8391192555427551}
+01/27/2022 14:21:50 - INFO - codeparrot_training - Step 19603: {'lr': 0.00035163931525132797, 'samples': 3763968, 'steps': 19603, 'loss/train': 1.14288729429245}
+01/27/2022 14:21:53 - INFO - codeparrot_training - Step 19604: {'lr': 0.00035162436587458744, 'samples': 3764160, 'steps': 19604, 'loss/train': 0.576606810092926}
+01/27/2022 14:21:57 - INFO - codeparrot_training - Step 19605: {'lr': 0.00035160941606252044, 'samples': 3764352, 'steps': 19605, 'loss/train': 0.43760450184345245}
+01/27/2022 14:22:00 - INFO - codeparrot_training - Step 19606: {'lr': 0.00035159446581519094, 'samples': 3764544, 'steps': 19606, 'loss/train': 0.6055115461349487}
+01/27/2022 14:22:04 - INFO - codeparrot_training - Step 19607: {'lr': 0.000351579515132663, 'samples': 3764736, 'steps': 19607, 'loss/train': 0.830065906047821}
+01/27/2022 14:22:07 - INFO - codeparrot_training - Step 19608: {'lr': 0.0003515645640150006, 'samples': 3764928, 'steps': 19608, 'loss/train': 0.7837049961090088}
+01/27/2022 14:22:11 - INFO - codeparrot_training - Step 19609: {'lr': 0.000351549612462268, 'samples': 3765120, 'steps': 19609, 'loss/train': 0.6141153573989868}
+01/27/2022 14:22:14 - INFO - codeparrot_training - Step 19610: {'lr': 0.00035153466047452904, 'samples': 3765312, 'steps': 19610, 'loss/train': 0.6394770294427872}
+01/27/2022 14:22:17 - INFO - codeparrot_training - Step 19611: {'lr': 0.00035151970805184785, 'samples': 3765504, 'steps': 19611, 'loss/train': 1.1314073503017426}
+01/27/2022 14:22:20 - INFO - codeparrot_training - Step 19612: {'lr': 0.00035150475519428844, 'samples': 3765696, 'steps': 19612, 'loss/train': 0.6149888187646866}
+01/27/2022 14:22:23 - INFO - codeparrot_training - Step 19613: {'lr': 0.00035148980190191485, 'samples': 3765888, 'steps': 19613, 'loss/train': 0.31140265613794327}
+01/27/2022 14:22:26 - INFO - codeparrot_training - Step 19614: {'lr': 0.00035147484817479126, 'samples': 3766080, 'steps': 19614, 'loss/train': 0.6419171690940857}
+01/27/2022 14:22:29 - INFO - codeparrot_training - Step 19615: {'lr': 0.00035145989401298163, 'samples': 3766272, 'steps': 19615, 'loss/train': 0.6391653567552567}
+01/27/2022 14:22:35 - INFO - codeparrot_training - Step 19616: {'lr': 0.00035144493941655, 'samples': 3766464, 'steps': 19616, 'loss/train': 0.7438561320304871}
+01/27/2022 14:22:38 - INFO - codeparrot_training - Step 19617: {'lr': 0.0003514299843855605, 'samples': 3766656, 'steps': 19617, 'loss/train': 1.1477394998073578}
+01/27/2022 14:22:41 - INFO - codeparrot_training - Step 19618: {'lr': 0.0003514150289200771, 'samples': 3766848, 'steps': 19618, 'loss/train': 0.6374634951353073}
+01/27/2022 14:22:44 - INFO - codeparrot_training - Step 19619: {'lr': 0.00035140007302016395, 'samples': 3767040, 'steps': 19619, 'loss/train': 2.2873435020446777}
+01/27/2022 14:22:47 - INFO - codeparrot_training - Step 19620: {'lr': 0.0003513851166858851, 'samples': 3767232, 'steps': 19620, 'loss/train': 0.842313140630722}
+01/27/2022 14:22:50 - INFO - codeparrot_training - Step 19621: {'lr': 0.0003513701599173046, 'samples': 3767424, 'steps': 19621, 'loss/train': 0.6839834600687027}
+01/27/2022 14:22:54 - INFO - codeparrot_training - Step 19622: {'lr': 0.0003513552027144865, 'samples': 3767616, 'steps': 19622, 'loss/train': 0.7687686681747437}
+01/27/2022 14:22:57 - INFO - codeparrot_training - Step 19623: {'lr': 0.00035134024507749487, 'samples': 3767808, 'steps': 19623, 'loss/train': 0.6801070868968964}
+01/27/2022 14:23:00 - INFO - codeparrot_training - Step 19624: {'lr': 0.0003513252870063939, 'samples': 3768000, 'steps': 19624, 'loss/train': 0.9064651429653168}
+01/27/2022 14:23:04 - INFO - codeparrot_training - Step 19625: {'lr': 0.00035131032850124744, 'samples': 3768192, 'steps': 19625, 'loss/train': 0.6030409634113312}
+01/27/2022 14:23:07 - INFO - codeparrot_training - Step 19626: {'lr': 0.0003512953695621198, 'samples': 3768384, 'steps': 19626, 'loss/train': 0.5850992649793625}
+01/27/2022 14:23:11 - INFO - codeparrot_training - Step 19627: {'lr': 0.00035128041018907496, 'samples': 3768576, 'steps': 19627, 'loss/train': 0.488021582365036}
+01/27/2022 14:23:14 - INFO - codeparrot_training - Step 19628: {'lr': 0.000351265450382177, 'samples': 3768768, 'steps': 19628, 'loss/train': 1.0064581632614136}
+01/27/2022 14:23:17 - INFO - codeparrot_training - Step 19629: {'lr': 0.00035125049014148995, 'samples': 3768960, 'steps': 19629, 'loss/train': 1.1173312067985535}
+01/27/2022 14:23:20 - INFO - codeparrot_training - Step 19630: {'lr': 0.000351235529467078, 'samples': 3769152, 'steps': 19630, 'loss/train': 0.805592805147171}
+01/27/2022 14:23:23 - INFO - codeparrot_training - Step 19631: {'lr': 0.00035122056835900517, 'samples': 3769344, 'steps': 19631, 'loss/train': 0.7188971936702728}
+01/27/2022 14:23:26 - INFO - codeparrot_training - Step 19632: {'lr': 0.0003512056068173356, 'samples': 3769536, 'steps': 19632, 'loss/train': 0.08489197120070457}
+01/27/2022 14:23:30 - INFO - codeparrot_training - Step 19633: {'lr': 0.00035119064484213333, 'samples': 3769728, 'steps': 19633, 'loss/train': 1.3246707022190094}
+01/27/2022 14:23:35 - INFO - codeparrot_training - Step 19634: {'lr': 0.0003511756824334625, 'samples': 3769920, 'steps': 19634, 'loss/train': 0.9014390408992767}
+01/27/2022 14:23:38 - INFO - codeparrot_training - Step 19635: {'lr': 0.0003511607195913872, 'samples': 3770112, 'steps': 19635, 'loss/train': 1.1014546751976013}
+01/27/2022 14:23:41 - INFO - codeparrot_training - Step 19636: {'lr': 0.00035114575631597136, 'samples': 3770304, 'steps': 19636, 'loss/train': 0.7470315545797348}
+01/27/2022 14:23:44 - INFO - codeparrot_training - Step 19637: {'lr': 0.0003511307926072793, 'samples': 3770496, 'steps': 19637, 'loss/train': 0.6994226574897766}
+01/27/2022 14:23:47 - INFO - codeparrot_training - Step 19638: {'lr': 0.00035111582846537507, 'samples': 3770688, 'steps': 19638, 'loss/train': 1.1055066883563995}
+01/27/2022 14:23:51 - INFO - codeparrot_training - Step 19639: {'lr': 0.00035110086389032264, 'samples': 3770880, 'steps': 19639, 'loss/train': 1.0539009869098663}
+01/27/2022 14:23:54 - INFO - codeparrot_training - Step 19640: {'lr': 0.0003510858988821863, 'samples': 3771072, 'steps': 19640, 'loss/train': 0.570220559835434}
+01/27/2022 14:23:57 - INFO - codeparrot_training - Step 19641: {'lr': 0.00035107093344103, 'samples': 3771264, 'steps': 19641, 'loss/train': 0.7219975143671036}
+01/27/2022 14:24:00 - INFO - codeparrot_training - Step 19642: {'lr': 0.000351055967566918, 'samples': 3771456, 'steps': 19642, 'loss/train': 0.6689979135990143}
+01/27/2022 14:24:05 - INFO - codeparrot_training - Step 19643: {'lr': 0.0003510410012599142, 'samples': 3771648, 'steps': 19643, 'loss/train': 0.824083149433136}
+01/27/2022 14:24:08 - INFO - codeparrot_training - Step 19644: {'lr': 0.00035102603452008297, 'samples': 3771840, 'steps': 19644, 'loss/train': 0.6641299724578857}
+01/27/2022 14:24:11 - INFO - codeparrot_training - Step 19645: {'lr': 0.00035101106734748824, 'samples': 3772032, 'steps': 19645, 'loss/train': 0.9158144295215607}
+01/27/2022 14:24:14 - INFO - codeparrot_training - Step 19646: {'lr': 0.00035099609974219417, 'samples': 3772224, 'steps': 19646, 'loss/train': 0.8583366572856903}
+01/27/2022 14:24:17 - INFO - codeparrot_training - Step 19647: {'lr': 0.00035098113170426484, 'samples': 3772416, 'steps': 19647, 'loss/train': 0.7262599915266037}
+01/27/2022 14:24:20 - INFO - codeparrot_training - Step 19648: {'lr': 0.0003509661632337645, 'samples': 3772608, 'steps': 19648, 'loss/train': 0.8183291852474213}
+01/27/2022 14:24:23 - INFO - codeparrot_training - Step 19649: {'lr': 0.00035095119433075706, 'samples': 3772800, 'steps': 19649, 'loss/train': 0.4222690165042877}
+01/27/2022 14:24:27 - INFO - codeparrot_training - Step 19650: {'lr': 0.00035093622499530677, 'samples': 3772992, 'steps': 19650, 'loss/train': 0.7574697732925415}
+01/27/2022 14:24:31 - INFO - codeparrot_training - Step 19651: {'lr': 0.0003509212552274778, 'samples': 3773184, 'steps': 19651, 'loss/train': 0.7400950938463211}
+01/27/2022 14:24:34 - INFO - codeparrot_training - Step 19652: {'lr': 0.0003509062850273342, 'samples': 3773376, 'steps': 19652, 'loss/train': 1.6594296097755432}
+01/27/2022 14:24:37 - INFO - codeparrot_training - Step 19653: {'lr': 0.00035089131439494013, 'samples': 3773568, 'steps': 19653, 'loss/train': 0.4765952378511429}
+01/27/2022 14:24:41 - INFO - codeparrot_training - Step 19654: {'lr': 0.00035087634333035966, 'samples': 3773760, 'steps': 19654, 'loss/train': 1.2637836635112762}
+01/27/2022 14:24:44 - INFO - codeparrot_training - Step 19655: {'lr': 0.00035086137183365707, 'samples': 3773952, 'steps': 19655, 'loss/train': 0.4617803692817688}
+01/27/2022 14:24:47 - INFO - codeparrot_training - Step 19656: {'lr': 0.0003508463999048963, 'samples': 3774144, 'steps': 19656, 'loss/train': 0.7626776397228241}
+01/27/2022 14:24:50 - INFO - codeparrot_training - Step 19657: {'lr': 0.0003508314275441416, 'samples': 3774336, 'steps': 19657, 'loss/train': 1.0602639019489288}
+01/27/2022 14:24:53 - INFO - codeparrot_training - Step 19658: {'lr': 0.0003508164547514571, 'samples': 3774528, 'steps': 19658, 'loss/train': 1.7879526615142822}
+01/27/2022 14:24:56 - INFO - codeparrot_training - Step 19659: {'lr': 0.00035080148152690687, 'samples': 3774720, 'steps': 19659, 'loss/train': 0.6451916545629501}
+01/27/2022 14:25:01 - INFO - codeparrot_training - Step 19660: {'lr': 0.0003507865078705551, 'samples': 3774912, 'steps': 19660, 'loss/train': 0.7923869490623474}
+01/27/2022 14:25:04 - INFO - codeparrot_training - Step 19661: {'lr': 0.00035077153378246603, 'samples': 3775104, 'steps': 19661, 'loss/train': 0.5544196218252182}
+01/27/2022 14:25:07 - INFO - codeparrot_training - Step 19662: {'lr': 0.0003507565592627036, 'samples': 3775296, 'steps': 19662, 'loss/train': 0.688909113407135}
+01/27/2022 14:25:11 - INFO - codeparrot_training - Step 19663: {'lr': 0.0003507415843113321, 'samples': 3775488, 'steps': 19663, 'loss/train': 0.9452697336673737}
+01/27/2022 14:25:14 - INFO - codeparrot_training - Step 19664: {'lr': 0.00035072660892841566, 'samples': 3775680, 'steps': 19664, 'loss/train': 1.1842007339000702}
+01/27/2022 14:25:17 - INFO - codeparrot_training - Step 19665: {'lr': 0.0003507116331140184, 'samples': 3775872, 'steps': 19665, 'loss/train': 1.4287302196025848}
+01/27/2022 14:25:20 - INFO - codeparrot_training - Step 19666: {'lr': 0.00035069665686820453, 'samples': 3776064, 'steps': 19666, 'loss/train': 0.9068146347999573}
+01/27/2022 14:25:23 - INFO - codeparrot_training - Step 19667: {'lr': 0.0003506816801910381, 'samples': 3776256, 'steps': 19667, 'loss/train': 0.8850484192371368}
+01/27/2022 14:25:26 - INFO - codeparrot_training - Step 19668: {'lr': 0.00035066670308258333, 'samples': 3776448, 'steps': 19668, 'loss/train': 1.507181167602539}
+01/27/2022 14:25:31 - INFO - codeparrot_training - Step 19669: {'lr': 0.00035065172554290435, 'samples': 3776640, 'steps': 19669, 'loss/train': 0.9139318764209747}
+01/27/2022 14:25:35 - INFO - codeparrot_training - Step 19670: {'lr': 0.0003506367475720654, 'samples': 3776832, 'steps': 19670, 'loss/train': 0.31293419748544693}
+01/27/2022 14:25:38 - INFO - codeparrot_training - Step 19671: {'lr': 0.0003506217691701305, 'samples': 3777024, 'steps': 19671, 'loss/train': 0.3486611545085907}
+01/27/2022 14:25:41 - INFO - codeparrot_training - Step 19672: {'lr': 0.000350606790337164, 'samples': 3777216, 'steps': 19672, 'loss/train': 0.8262868523597717}
+01/27/2022 14:25:44 - INFO - codeparrot_training - Step 19673: {'lr': 0.00035059181107322977, 'samples': 3777408, 'steps': 19673, 'loss/train': 0.7132445275783539}
+01/27/2022 14:25:47 - INFO - codeparrot_training - Step 19674: {'lr': 0.00035057683137839236, 'samples': 3777600, 'steps': 19674, 'loss/train': 0.8232385218143463}
+01/27/2022 14:25:50 - INFO - codeparrot_training - Step 19675: {'lr': 0.00035056185125271566, 'samples': 3777792, 'steps': 19675, 'loss/train': 0.9783961772918701}
+01/27/2022 14:25:54 - INFO - codeparrot_training - Step 19676: {'lr': 0.0003505468706962639, 'samples': 3777984, 'steps': 19676, 'loss/train': 0.8206941783428192}
+01/27/2022 14:25:57 - INFO - codeparrot_training - Step 19677: {'lr': 0.0003505318897091013, 'samples': 3778176, 'steps': 19677, 'loss/train': 0.6649615466594696}
+01/27/2022 14:26:01 - INFO - codeparrot_training - Step 19678: {'lr': 0.000350516908291292, 'samples': 3778368, 'steps': 19678, 'loss/train': 0.8174855411052704}
+01/27/2022 14:26:05 - INFO - codeparrot_training - Step 19679: {'lr': 0.00035050192644290023, 'samples': 3778560, 'steps': 19679, 'loss/train': 2.681984603404999}
+01/27/2022 14:26:08 - INFO - codeparrot_training - Step 19680: {'lr': 0.00035048694416399005, 'samples': 3778752, 'steps': 19680, 'loss/train': 1.0002517104148865}
+01/27/2022 14:26:11 - INFO - codeparrot_training - Step 19681: {'lr': 0.0003504719614546258, 'samples': 3778944, 'steps': 19681, 'loss/train': 1.0692269504070282}
+01/27/2022 14:26:14 - INFO - codeparrot_training - Step 19682: {'lr': 0.00035045697831487146, 'samples': 3779136, 'steps': 19682, 'loss/train': 0.4680267423391342}
+01/27/2022 14:26:17 - INFO - codeparrot_training - Step 19683: {'lr': 0.00035044199474479137, 'samples': 3779328, 'steps': 19683, 'loss/train': 0.9153154492378235}
+01/27/2022 14:26:20 - INFO - codeparrot_training - Step 19684: {'lr': 0.00035042701074444965, 'samples': 3779520, 'steps': 19684, 'loss/train': 0.2726229205727577}
+01/27/2022 14:26:23 - INFO - codeparrot_training - Step 19685: {'lr': 0.00035041202631391056, 'samples': 3779712, 'steps': 19685, 'loss/train': 0.49869905412197113}
+01/27/2022 14:26:28 - INFO - codeparrot_training - Step 19686: {'lr': 0.0003503970414532382, 'samples': 3779904, 'steps': 19686, 'loss/train': 1.0594726502895355}
+01/27/2022 14:26:31 - INFO - codeparrot_training - Step 19687: {'lr': 0.00035038205616249674, 'samples': 3780096, 'steps': 19687, 'loss/train': 0.8030432760715485}
+01/27/2022 14:26:34 - INFO - codeparrot_training - Step 19688: {'lr': 0.00035036707044175055, 'samples': 3780288, 'steps': 19688, 'loss/train': 1.0118459165096283}
+01/27/2022 14:26:37 - INFO - codeparrot_training - Step 19689: {'lr': 0.00035035208429106356, 'samples': 3780480, 'steps': 19689, 'loss/train': 0.562382698059082}
+01/27/2022 14:26:41 - INFO - codeparrot_training - Step 19690: {'lr': 0.0003503370977105002, 'samples': 3780672, 'steps': 19690, 'loss/train': 0.6421509236097336}
+01/27/2022 14:26:44 - INFO - codeparrot_training - Step 19691: {'lr': 0.00035032211070012455, 'samples': 3780864, 'steps': 19691, 'loss/train': 0.6968682557344437}
+01/27/2022 14:26:47 - INFO - codeparrot_training - Step 19692: {'lr': 0.00035030712326000084, 'samples': 3781056, 'steps': 19692, 'loss/train': 0.33236968517303467}
+01/27/2022 14:26:50 - INFO - codeparrot_training - Step 19693: {'lr': 0.00035029213539019324, 'samples': 3781248, 'steps': 19693, 'loss/train': 0.8013027906417847}
+01/27/2022 14:26:53 - INFO - codeparrot_training - Step 19694: {'lr': 0.000350277147090766, 'samples': 3781440, 'steps': 19694, 'loss/train': 0.8997577428817749}
+01/27/2022 14:26:58 - INFO - codeparrot_training - Step 19695: {'lr': 0.0003502621583617833, 'samples': 3781632, 'steps': 19695, 'loss/train': 0.5657816827297211}
+01/27/2022 14:27:02 - INFO - codeparrot_training - Step 19696: {'lr': 0.00035024716920330933, 'samples': 3781824, 'steps': 19696, 'loss/train': 1.1507024466991425}
+01/27/2022 14:27:05 - INFO - codeparrot_training - Step 19697: {'lr': 0.0003502321796154084, 'samples': 3782016, 'steps': 19697, 'loss/train': 0.9584307074546814}
+01/27/2022 14:27:08 - INFO - codeparrot_training - Step 19698: {'lr': 0.00035021718959814453, 'samples': 3782208, 'steps': 19698, 'loss/train': 0.9946712851524353}
+01/27/2022 14:27:11 - INFO - codeparrot_training - Step 19699: {'lr': 0.0003502021991515821, 'samples': 3782400, 'steps': 19699, 'loss/train': 0.9093162417411804}
+01/27/2022 14:27:14 - INFO - codeparrot_training - Step 19700: {'lr': 0.0003501872082757852, 'samples': 3782592, 'steps': 19700, 'loss/train': 0.6505986303091049}
+01/27/2022 14:27:17 - INFO - codeparrot_training - Step 19701: {'lr': 0.00035017221697081826, 'samples': 3782784, 'steps': 19701, 'loss/train': 1.020242840051651}
+01/27/2022 14:27:20 - INFO - codeparrot_training - Step 19702: {'lr': 0.0003501572252367452, 'samples': 3782976, 'steps': 19702, 'loss/train': 0.8630232810974121}
+01/27/2022 14:27:24 - INFO - codeparrot_training - Step 19703: {'lr': 0.00035014223307363045, 'samples': 3783168, 'steps': 19703, 'loss/train': 1.0528368651866913}
+01/27/2022 14:27:28 - INFO - codeparrot_training - Step 19704: {'lr': 0.0003501272404815382, 'samples': 3783360, 'steps': 19704, 'loss/train': 0.18638339638710022}
+01/27/2022 14:27:31 - INFO - codeparrot_training - Step 19705: {'lr': 0.0003501122474605326, 'samples': 3783552, 'steps': 19705, 'loss/train': 1.0772386193275452}
+01/27/2022 14:27:34 - INFO - codeparrot_training - Step 19706: {'lr': 0.00035009725401067795, 'samples': 3783744, 'steps': 19706, 'loss/train': 0.7274583578109741}
+01/27/2022 14:27:37 - INFO - codeparrot_training - Step 19707: {'lr': 0.00035008226013203845, 'samples': 3783936, 'steps': 19707, 'loss/train': 0.8535030484199524}
+01/27/2022 14:27:41 - INFO - codeparrot_training - Step 19708: {'lr': 0.0003500672658246783, 'samples': 3784128, 'steps': 19708, 'loss/train': 0.825296938419342}
+01/27/2022 14:27:44 - INFO - codeparrot_training - Step 19709: {'lr': 0.0003500522710886618, 'samples': 3784320, 'steps': 19709, 'loss/train': 0.45952141284942627}
+01/27/2022 14:27:47 - INFO - codeparrot_training - Step 19710: {'lr': 0.0003500372759240531, 'samples': 3784512, 'steps': 19710, 'loss/train': 0.6975622773170471}
+01/27/2022 14:27:50 - INFO - codeparrot_training - Step 19711: {'lr': 0.00035002228033091643, 'samples': 3784704, 'steps': 19711, 'loss/train': 0.628915473818779}
+01/27/2022 14:27:53 - INFO - codeparrot_training - Step 19712: {'lr': 0.00035000728430931616, 'samples': 3784896, 'steps': 19712, 'loss/train': 0.5489722788333893}
+01/27/2022 14:27:58 - INFO - codeparrot_training - Step 19713: {'lr': 0.00034999228785931644, 'samples': 3785088, 'steps': 19713, 'loss/train': 0.34397464245557785}
+01/27/2022 14:28:01 - INFO - codeparrot_training - Step 19714: {'lr': 0.0003499772909809815, 'samples': 3785280, 'steps': 19714, 'loss/train': 0.8667342066764832}
+01/27/2022 14:28:04 - INFO - codeparrot_training - Step 19715: {'lr': 0.0003499622936743756, 'samples': 3785472, 'steps': 19715, 'loss/train': 0.7769384980201721}
+01/27/2022 14:28:07 - INFO - codeparrot_training - Step 19716: {'lr': 0.0003499472959395629, 'samples': 3785664, 'steps': 19716, 'loss/train': 0.6309301257133484}
+01/27/2022 14:28:10 - INFO - codeparrot_training - Step 19717: {'lr': 0.00034993229777660785, 'samples': 3785856, 'steps': 19717, 'loss/train': 0.5563713759183884}
+01/27/2022 14:28:13 - INFO - codeparrot_training - Step 19718: {'lr': 0.0003499172991855744, 'samples': 3786048, 'steps': 19718, 'loss/train': 0.8798454701900482}
+01/27/2022 14:28:16 - INFO - codeparrot_training - Step 19719: {'lr': 0.00034990230016652713, 'samples': 3786240, 'steps': 19719, 'loss/train': 1.0315495133399963}
+01/27/2022 14:28:20 - INFO - codeparrot_training - Step 19720: {'lr': 0.00034988730071953, 'samples': 3786432, 'steps': 19720, 'loss/train': 1.1913085877895355}
+01/27/2022 14:28:25 - INFO - codeparrot_training - Step 19721: {'lr': 0.0003498723008446475, 'samples': 3786624, 'steps': 19721, 'loss/train': 0.9347498416900635}
+01/27/2022 14:28:28 - INFO - codeparrot_training - Step 19722: {'lr': 0.0003498573005419437, 'samples': 3786816, 'steps': 19722, 'loss/train': 0.6913055330514908}
+01/27/2022 14:28:31 - INFO - codeparrot_training - Step 19723: {'lr': 0.000349842299811483, 'samples': 3787008, 'steps': 19723, 'loss/train': 1.0747533738613129}
+01/27/2022 14:28:34 - INFO - codeparrot_training - Step 19724: {'lr': 0.00034982729865332953, 'samples': 3787200, 'steps': 19724, 'loss/train': 0.7712821662425995}
+01/27/2022 14:28:37 - INFO - codeparrot_training - Step 19725: {'lr': 0.00034981229706754755, 'samples': 3787392, 'steps': 19725, 'loss/train': 0.7185169905424118}
+01/27/2022 14:28:40 - INFO - codeparrot_training - Step 19726: {'lr': 0.0003497972950542015, 'samples': 3787584, 'steps': 19726, 'loss/train': 1.0540443062782288}
+01/27/2022 14:28:44 - INFO - codeparrot_training - Step 19727: {'lr': 0.0003497822926133555, 'samples': 3787776, 'steps': 19727, 'loss/train': 1.0555404424667358}
+01/27/2022 14:28:47 - INFO - codeparrot_training - Step 19728: {'lr': 0.0003497672897450738, 'samples': 3787968, 'steps': 19728, 'loss/train': 0.7982284426689148}
+01/27/2022 14:28:50 - INFO - codeparrot_training - Step 19729: {'lr': 0.0003497522864494208, 'samples': 3788160, 'steps': 19729, 'loss/train': 0.838407576084137}
+01/27/2022 14:28:54 - INFO - codeparrot_training - Step 19730: {'lr': 0.0003497372827264606, 'samples': 3788352, 'steps': 19730, 'loss/train': 0.5037300288677216}
+01/27/2022 14:28:57 - INFO - codeparrot_training - Step 19731: {'lr': 0.0003497222785762576, 'samples': 3788544, 'steps': 19731, 'loss/train': 1.130009114742279}
+01/27/2022 14:29:01 - INFO - codeparrot_training - Step 19732: {'lr': 0.000349707273998876, 'samples': 3788736, 'steps': 19732, 'loss/train': 0.8464545607566833}
+01/27/2022 14:29:04 - INFO - codeparrot_training - Step 19733: {'lr': 0.00034969226899438013, 'samples': 3788928, 'steps': 19733, 'loss/train': 0.9235986471176147}
+01/27/2022 14:29:07 - INFO - codeparrot_training - Step 19734: {'lr': 0.00034967726356283416, 'samples': 3789120, 'steps': 19734, 'loss/train': 0.8261380791664124}
+01/27/2022 14:29:10 - INFO - codeparrot_training - Step 19735: {'lr': 0.00034966225770430244, 'samples': 3789312, 'steps': 19735, 'loss/train': 1.0145914256572723}
+01/27/2022 14:29:13 - INFO - codeparrot_training - Step 19736: {'lr': 0.00034964725141884936, 'samples': 3789504, 'steps': 19736, 'loss/train': 0.6910649389028549}
+01/27/2022 14:29:16 - INFO - codeparrot_training - Step 19737: {'lr': 0.000349632244706539, 'samples': 3789696, 'steps': 19737, 'loss/train': 1.1405586004257202}
+01/27/2022 14:29:19 - INFO - codeparrot_training - Step 19738: {'lr': 0.0003496172375674358, 'samples': 3789888, 'steps': 19738, 'loss/train': 1.1956051290035248}
+01/27/2022 14:29:25 - INFO - codeparrot_training - Step 19739: {'lr': 0.0003496022300016039, 'samples': 3790080, 'steps': 19739, 'loss/train': 0.6953363567590714}
+01/27/2022 14:29:28 - INFO - codeparrot_training - Step 19740: {'lr': 0.00034958722200910777, 'samples': 3790272, 'steps': 19740, 'loss/train': 0.7392996400594711}
+01/27/2022 14:29:31 - INFO - codeparrot_training - Step 19741: {'lr': 0.00034957221359001154, 'samples': 3790464, 'steps': 19741, 'loss/train': 0.6584598273038864}
+01/27/2022 14:29:34 - INFO - codeparrot_training - Step 19742: {'lr': 0.0003495572047443796, 'samples': 3790656, 'steps': 19742, 'loss/train': 1.2632261216640472}
+01/27/2022 14:29:37 - INFO - codeparrot_training - Step 19743: {'lr': 0.0003495421954722762, 'samples': 3790848, 'steps': 19743, 'loss/train': 1.089457243680954}
+01/27/2022 14:29:40 - INFO - codeparrot_training - Step 19744: {'lr': 0.0003495271857737657, 'samples': 3791040, 'steps': 19744, 'loss/train': 1.009457141160965}
+01/27/2022 14:29:44 - INFO - codeparrot_training - Step 19745: {'lr': 0.00034951217564891226, 'samples': 3791232, 'steps': 19745, 'loss/train': 0.8250389993190765}
+01/27/2022 14:29:47 - INFO - codeparrot_training - Step 19746: {'lr': 0.0003494971650977802, 'samples': 3791424, 'steps': 19746, 'loss/train': 0.9431262910366058}
+01/27/2022 14:29:50 - INFO - codeparrot_training - Step 19747: {'lr': 0.00034948215412043405, 'samples': 3791616, 'steps': 19747, 'loss/train': 0.04144127760082483}
+01/27/2022 14:29:54 - INFO - codeparrot_training - Step 19748: {'lr': 0.00034946714271693783, 'samples': 3791808, 'steps': 19748, 'loss/train': 0.5094870775938034}
+01/27/2022 14:29:57 - INFO - codeparrot_training - Step 19749: {'lr': 0.00034945213088735595, 'samples': 3792000, 'steps': 19749, 'loss/train': 0.9024822413921356}
+01/27/2022 14:30:01 - INFO - codeparrot_training - Step 19750: {'lr': 0.00034943711863175277, 'samples': 3792192, 'steps': 19750, 'loss/train': 1.1553838849067688}
+01/27/2022 14:30:04 - INFO - codeparrot_training - Step 19751: {'lr': 0.0003494221059501925, 'samples': 3792384, 'steps': 19751, 'loss/train': 0.8642564713954926}
+01/27/2022 14:30:07 - INFO - codeparrot_training - Step 19752: {'lr': 0.0003494070928427395, 'samples': 3792576, 'steps': 19752, 'loss/train': 0.4574185609817505}
+01/27/2022 14:30:10 - INFO - codeparrot_training - Step 19753: {'lr': 0.00034939207930945816, 'samples': 3792768, 'steps': 19753, 'loss/train': 0.922255128622055}
+01/27/2022 14:30:13 - INFO - codeparrot_training - Step 19754: {'lr': 0.00034937706535041263, 'samples': 3792960, 'steps': 19754, 'loss/train': 0.7591567039489746}
+01/27/2022 14:30:16 - INFO - codeparrot_training - Step 19755: {'lr': 0.0003493620509656674, 'samples': 3793152, 'steps': 19755, 'loss/train': 0.8058052361011505}
+01/27/2022 14:30:21 - INFO - codeparrot_training - Step 19756: {'lr': 0.00034934703615528657, 'samples': 3793344, 'steps': 19756, 'loss/train': 0.38734039664268494}
+01/27/2022 14:30:24 - INFO - codeparrot_training - Step 19757: {'lr': 0.0003493320209193347, 'samples': 3793536, 'steps': 19757, 'loss/train': 0.9241312444210052}
+01/27/2022 14:30:27 - INFO - codeparrot_training - Step 19758: {'lr': 0.0003493170052578759, 'samples': 3793728, 'steps': 19758, 'loss/train': 0.6229194402694702}
+01/27/2022 14:30:30 - INFO - codeparrot_training - Step 19759: {'lr': 0.00034930198917097467, 'samples': 3793920, 'steps': 19759, 'loss/train': 0.8520568013191223}
+01/27/2022 14:30:33 - INFO - codeparrot_training - Step 19760: {'lr': 0.00034928697265869515, 'samples': 3794112, 'steps': 19760, 'loss/train': 0.8213016986846924}
+01/27/2022 14:30:36 - INFO - codeparrot_training - Step 19761: {'lr': 0.0003492719557211018, 'samples': 3794304, 'steps': 19761, 'loss/train': 0.7622648477554321}
+01/27/2022 14:30:39 - INFO - codeparrot_training - Step 19762: {'lr': 0.0003492569383582589, 'samples': 3794496, 'steps': 19762, 'loss/train': 0.8662754595279694}
+01/27/2022 14:30:43 - INFO - codeparrot_training - Step 19763: {'lr': 0.0003492419205702309, 'samples': 3794688, 'steps': 19763, 'loss/train': 1.099509447813034}
+01/27/2022 14:30:46 - INFO - codeparrot_training - Step 19764: {'lr': 0.00034922690235708197, 'samples': 3794880, 'steps': 19764, 'loss/train': 0.713667020201683}
+01/27/2022 14:30:50 - INFO - codeparrot_training - Step 19765: {'lr': 0.0003492118837188765, 'samples': 3795072, 'steps': 19765, 'loss/train': 0.7121306955814362}
+01/27/2022 14:30:53 - INFO - codeparrot_training - Step 19766: {'lr': 0.0003491968646556788, 'samples': 3795264, 'steps': 19766, 'loss/train': 1.0644871294498444}
+01/27/2022 14:30:57 - INFO - codeparrot_training - Step 19767: {'lr': 0.00034918184516755324, 'samples': 3795456, 'steps': 19767, 'loss/train': 0.6692989468574524}
+01/27/2022 14:31:00 - INFO - codeparrot_training - Step 19768: {'lr': 0.00034916682525456416, 'samples': 3795648, 'steps': 19768, 'loss/train': 0.6098275780677795}
+01/27/2022 14:31:03 - INFO - codeparrot_training - Step 19769: {'lr': 0.00034915180491677583, 'samples': 3795840, 'steps': 19769, 'loss/train': 0.24129514396190643}
+01/27/2022 14:31:06 - INFO - codeparrot_training - Step 19770: {'lr': 0.00034913678415425276, 'samples': 3796032, 'steps': 19770, 'loss/train': 1.0328033566474915}
+01/27/2022 14:31:09 - INFO - codeparrot_training - Step 19771: {'lr': 0.00034912176296705903, 'samples': 3796224, 'steps': 19771, 'loss/train': 1.1578963994979858}
+01/27/2022 14:31:12 - INFO - codeparrot_training - Step 19772: {'lr': 0.00034910674135525926, 'samples': 3796416, 'steps': 19772, 'loss/train': 0.6762576699256897}
+01/27/2022 14:31:15 - INFO - codeparrot_training - Step 19773: {'lr': 0.0003490917193189177, 'samples': 3796608, 'steps': 19773, 'loss/train': 0.5437828302383423}
+01/27/2022 14:31:21 - INFO - codeparrot_training - Step 19774: {'lr': 0.00034907669685809855, 'samples': 3796800, 'steps': 19774, 'loss/train': 1.2111424505710602}
+01/27/2022 14:31:24 - INFO - codeparrot_training - Step 19775: {'lr': 0.0003490616739728664, 'samples': 3796992, 'steps': 19775, 'loss/train': 0.6455262154340744}
+01/27/2022 14:31:27 - INFO - codeparrot_training - Step 19776: {'lr': 0.00034904665066328545, 'samples': 3797184, 'steps': 19776, 'loss/train': 0.32741039246320724}
+01/27/2022 14:31:30 - INFO - codeparrot_training - Step 19777: {'lr': 0.00034903162692942013, 'samples': 3797376, 'steps': 19777, 'loss/train': 0.9657024443149567}
+01/27/2022 14:31:33 - INFO - codeparrot_training - Step 19778: {'lr': 0.0003490166027713348, 'samples': 3797568, 'steps': 19778, 'loss/train': 0.9155499637126923}
+01/27/2022 14:31:37 - INFO - codeparrot_training - Step 19779: {'lr': 0.0003490015781890937, 'samples': 3797760, 'steps': 19779, 'loss/train': 1.005632221698761}
+01/27/2022 14:31:40 - INFO - codeparrot_training - Step 19780: {'lr': 0.00034898655318276134, 'samples': 3797952, 'steps': 19780, 'loss/train': 0.7851841449737549}
+01/27/2022 14:31:43 - INFO - codeparrot_training - Step 19781: {'lr': 0.00034897152775240206, 'samples': 3798144, 'steps': 19781, 'loss/train': 1.0701836049556732}
+01/27/2022 14:31:46 - INFO - codeparrot_training - Step 19782: {'lr': 0.00034895650189808005, 'samples': 3798336, 'steps': 19782, 'loss/train': 0.7451811879873276}
+01/27/2022 14:31:50 - INFO - codeparrot_training - Step 19783: {'lr': 0.0003489414756198598, 'samples': 3798528, 'steps': 19783, 'loss/train': 0.8199529945850372}
+01/27/2022 14:31:54 - INFO - codeparrot_training - Step 19784: {'lr': 0.0003489264489178058, 'samples': 3798720, 'steps': 19784, 'loss/train': 0.7409288138151169}
+01/27/2022 14:31:57 - INFO - codeparrot_training - Step 19785: {'lr': 0.0003489114217919823, 'samples': 3798912, 'steps': 19785, 'loss/train': 1.0582976639270782}
+01/27/2022 14:32:00 - INFO - codeparrot_training - Step 19786: {'lr': 0.00034889639424245364, 'samples': 3799104, 'steps': 19786, 'loss/train': 0.7628023624420166}
+01/27/2022 14:32:03 - INFO - codeparrot_training - Step 19787: {'lr': 0.00034888136626928427, 'samples': 3799296, 'steps': 19787, 'loss/train': 0.3121131956577301}
+01/27/2022 14:32:06 - INFO - codeparrot_training - Step 19788: {'lr': 0.00034886633787253846, 'samples': 3799488, 'steps': 19788, 'loss/train': 0.9505481421947479}
+01/27/2022 14:32:09 - INFO - codeparrot_training - Step 19789: {'lr': 0.00034885130905228063, 'samples': 3799680, 'steps': 19789, 'loss/train': 0.3030192330479622}
+01/27/2022 14:32:13 - INFO - codeparrot_training - Step 19790: {'lr': 0.00034883627980857523, 'samples': 3799872, 'steps': 19790, 'loss/train': 0.8661814033985138}
+01/27/2022 14:32:16 - INFO - codeparrot_training - Step 19791: {'lr': 0.0003488212501414866, 'samples': 3800064, 'steps': 19791, 'loss/train': 1.0952494740486145}
+01/27/2022 14:32:20 - INFO - codeparrot_training - Step 19792: {'lr': 0.00034880622005107915, 'samples': 3800256, 'steps': 19792, 'loss/train': 0.7606033980846405}
+01/27/2022 14:32:23 - INFO - codeparrot_training - Step 19793: {'lr': 0.00034879118953741716, 'samples': 3800448, 'steps': 19793, 'loss/train': 0.7775324285030365}
+01/27/2022 14:32:26 - INFO - codeparrot_training - Step 19794: {'lr': 0.0003487761586005651, 'samples': 3800640, 'steps': 19794, 'loss/train': 1.2764904499053955}
+01/27/2022 14:32:30 - INFO - codeparrot_training - Step 19795: {'lr': 0.0003487611272405873, 'samples': 3800832, 'steps': 19795, 'loss/train': 1.1090977191925049}
+01/27/2022 14:32:33 - INFO - codeparrot_training - Step 19796: {'lr': 0.00034874609545754826, 'samples': 3801024, 'steps': 19796, 'loss/train': 1.0921995341777802}
+01/27/2022 14:32:36 - INFO - codeparrot_training - Step 19797: {'lr': 0.00034873106325151234, 'samples': 3801216, 'steps': 19797, 'loss/train': 1.1231814622879028}
+01/27/2022 14:32:39 - INFO - codeparrot_training - Step 19798: {'lr': 0.0003487160306225438, 'samples': 3801408, 'steps': 19798, 'loss/train': 0.7688643336296082}
+01/27/2022 14:32:42 - INFO - codeparrot_training - Step 19799: {'lr': 0.00034870099757070716, 'samples': 3801600, 'steps': 19799, 'loss/train': 0.9608098268508911}
+01/27/2022 14:32:45 - INFO - codeparrot_training - Step 19800: {'lr': 0.0003486859640960668, 'samples': 3801792, 'steps': 19800, 'loss/train': 0.746026486158371}
+01/27/2022 14:32:50 - INFO - codeparrot_training - Step 19801: {'lr': 0.0003486709301986871, 'samples': 3801984, 'steps': 19801, 'loss/train': 0.5828608721494675}
+01/27/2022 14:32:54 - INFO - codeparrot_training - Step 19802: {'lr': 0.00034865589587863247, 'samples': 3802176, 'steps': 19802, 'loss/train': 0.6232021898031235}
+01/27/2022 14:32:57 - INFO - codeparrot_training - Step 19803: {'lr': 0.0003486408611359673, 'samples': 3802368, 'steps': 19803, 'loss/train': 1.091566264629364}
+01/27/2022 14:33:00 - INFO - codeparrot_training - Step 19804: {'lr': 0.00034862582597075607, 'samples': 3802560, 'steps': 19804, 'loss/train': 0.9706878662109375}
+01/27/2022 14:33:03 - INFO - codeparrot_training - Step 19805: {'lr': 0.00034861079038306305, 'samples': 3802752, 'steps': 19805, 'loss/train': 0.5561365485191345}
+01/27/2022 14:33:06 - INFO - codeparrot_training - Step 19806: {'lr': 0.00034859575437295277, 'samples': 3802944, 'steps': 19806, 'loss/train': 0.7854202687740326}
+01/27/2022 14:33:09 - INFO - codeparrot_training - Step 19807: {'lr': 0.00034858071794048953, 'samples': 3803136, 'steps': 19807, 'loss/train': 0.6779172420501709}
+01/27/2022 14:33:13 - INFO - codeparrot_training - Step 19808: {'lr': 0.0003485656810857378, 'samples': 3803328, 'steps': 19808, 'loss/train': 1.0081038773059845}
+01/27/2022 14:33:16 - INFO - codeparrot_training - Step 19809: {'lr': 0.00034855064380876193, 'samples': 3803520, 'steps': 19809, 'loss/train': 0.939114511013031}
+01/27/2022 14:33:20 - INFO - codeparrot_training - Step 19810: {'lr': 0.00034853560610962654, 'samples': 3803712, 'steps': 19810, 'loss/train': 1.1696589589118958}
+01/27/2022 14:33:23 - INFO - codeparrot_training - Step 19811: {'lr': 0.0003485205679883958, 'samples': 3803904, 'steps': 19811, 'loss/train': 0.5680818110704422}
+01/27/2022 14:33:27 - INFO - codeparrot_training - Step 19812: {'lr': 0.00034850552944513426, 'samples': 3804096, 'steps': 19812, 'loss/train': 0.7155122458934784}
+01/27/2022 14:33:30 - INFO - codeparrot_training - Step 19813: {'lr': 0.00034849049047990633, 'samples': 3804288, 'steps': 19813, 'loss/train': 0.8984583020210266}
+01/27/2022 14:33:33 - INFO - codeparrot_training - Step 19814: {'lr': 0.0003484754510927764, 'samples': 3804480, 'steps': 19814, 'loss/train': 0.9706480801105499}
+01/27/2022 14:33:36 - INFO - codeparrot_training - Step 19815: {'lr': 0.00034846041128380886, 'samples': 3804672, 'steps': 19815, 'loss/train': 0.35307933390140533}
+01/27/2022 14:33:39 - INFO - codeparrot_training - Step 19816: {'lr': 0.0003484453710530682, 'samples': 3804864, 'steps': 19816, 'loss/train': 0.6552969664335251}
+01/27/2022 14:33:42 - INFO - codeparrot_training - Step 19817: {'lr': 0.0003484303304006189, 'samples': 3805056, 'steps': 19817, 'loss/train': 1.062898188829422}
+01/27/2022 14:33:47 - INFO - codeparrot_training - Step 19818: {'lr': 0.0003484152893265253, 'samples': 3805248, 'steps': 19818, 'loss/train': 0.7321252673864365}
+01/27/2022 14:33:50 - INFO - codeparrot_training - Step 19819: {'lr': 0.00034840024783085177, 'samples': 3805440, 'steps': 19819, 'loss/train': 0.2373429536819458}
+01/27/2022 14:33:54 - INFO - codeparrot_training - Step 19820: {'lr': 0.00034838520591366285, 'samples': 3805632, 'steps': 19820, 'loss/train': 0.671404168009758}
+01/27/2022 14:33:57 - INFO - codeparrot_training - Step 19821: {'lr': 0.00034837016357502297, 'samples': 3805824, 'steps': 19821, 'loss/train': 0.9229432046413422}
+01/27/2022 14:34:00 - INFO - codeparrot_training - Step 19822: {'lr': 0.0003483551208149965, 'samples': 3806016, 'steps': 19822, 'loss/train': 0.8015734255313873}
+01/27/2022 14:34:03 - INFO - codeparrot_training - Step 19823: {'lr': 0.00034834007763364803, 'samples': 3806208, 'steps': 19823, 'loss/train': 0.7144668996334076}
+01/27/2022 14:34:06 - INFO - codeparrot_training - Step 19824: {'lr': 0.00034832503403104184, 'samples': 3806400, 'steps': 19824, 'loss/train': 1.0925495624542236}
+01/27/2022 14:34:09 - INFO - codeparrot_training - Step 19825: {'lr': 0.00034830999000724246, 'samples': 3806592, 'steps': 19825, 'loss/train': 1.4910094141960144}
+01/27/2022 14:34:12 - INFO - codeparrot_training - Step 19826: {'lr': 0.00034829494556231423, 'samples': 3806784, 'steps': 19826, 'loss/train': 0.6962500959634781}
+01/27/2022 14:34:17 - INFO - codeparrot_training - Step 19827: {'lr': 0.00034827990069632173, 'samples': 3806976, 'steps': 19827, 'loss/train': 0.8716259300708771}
+01/27/2022 14:34:20 - INFO - codeparrot_training - Step 19828: {'lr': 0.0003482648554093293, 'samples': 3807168, 'steps': 19828, 'loss/train': 0.6793643832206726}
+01/27/2022 14:34:23 - INFO - codeparrot_training - Step 19829: {'lr': 0.0003482498097014015, 'samples': 3807360, 'steps': 19829, 'loss/train': 0.8299587070941925}
+01/27/2022 14:34:26 - INFO - codeparrot_training - Step 19830: {'lr': 0.0003482347635726026, 'samples': 3807552, 'steps': 19830, 'loss/train': 0.5277935117483139}
+01/27/2022 14:34:29 - INFO - codeparrot_training - Step 19831: {'lr': 0.00034821971702299716, 'samples': 3807744, 'steps': 19831, 'loss/train': 0.5589883625507355}
+01/27/2022 14:34:33 - INFO - codeparrot_training - Step 19832: {'lr': 0.0003482046700526498, 'samples': 3807936, 'steps': 19832, 'loss/train': 0.97634357213974}
+01/27/2022 14:34:36 - INFO - codeparrot_training - Step 19833: {'lr': 0.0003481896226616246, 'samples': 3808128, 'steps': 19833, 'loss/train': 0.9469732940196991}
+01/27/2022 14:34:39 - INFO - codeparrot_training - Step 19834: {'lr': 0.00034817457484998644, 'samples': 3808320, 'steps': 19834, 'loss/train': 0.7027046531438828}
+01/27/2022 14:34:42 - INFO - codeparrot_training - Step 19835: {'lr': 0.00034815952661779946, 'samples': 3808512, 'steps': 19835, 'loss/train': 0.35637688636779785}
+01/27/2022 14:34:46 - INFO - codeparrot_training - Step 19836: {'lr': 0.00034814447796512824, 'samples': 3808704, 'steps': 19836, 'loss/train': 0.7840908765792847}
+01/27/2022 14:34:50 - INFO - codeparrot_training - Step 19837: {'lr': 0.0003481294288920373, 'samples': 3808896, 'steps': 19837, 'loss/train': 0.5219949781894684}
+01/27/2022 14:34:53 - INFO - codeparrot_training - Step 19838: {'lr': 0.00034811437939859106, 'samples': 3809088, 'steps': 19838, 'loss/train': 0.68085877597332}
+01/27/2022 14:34:56 - INFO - codeparrot_training - Step 19839: {'lr': 0.0003480993294848539, 'samples': 3809280, 'steps': 19839, 'loss/train': 0.947620153427124}
+01/27/2022 14:34:59 - INFO - codeparrot_training - Step 19840: {'lr': 0.00034808427915089036, 'samples': 3809472, 'steps': 19840, 'loss/train': 0.7603780031204224}
+01/27/2022 14:35:02 - INFO - codeparrot_training - Step 19841: {'lr': 0.000348069228396765, 'samples': 3809664, 'steps': 19841, 'loss/train': 0.9401345551013947}
+01/27/2022 14:35:05 - INFO - codeparrot_training - Step 19842: {'lr': 0.00034805417722254213, 'samples': 3809856, 'steps': 19842, 'loss/train': 1.4752970337867737}
+01/27/2022 14:35:08 - INFO - codeparrot_training - Step 19843: {'lr': 0.00034803912562828633, 'samples': 3810048, 'steps': 19843, 'loss/train': 0.5910725891590118}
+01/27/2022 14:35:12 - INFO - codeparrot_training - Step 19844: {'lr': 0.000348024073614062, 'samples': 3810240, 'steps': 19844, 'loss/train': 0.8059833347797394}
+01/27/2022 14:35:17 - INFO - codeparrot_training - Step 19845: {'lr': 0.0003480090211799337, 'samples': 3810432, 'steps': 19845, 'loss/train': 0.9462956786155701}
+01/27/2022 14:35:20 - INFO - codeparrot_training - Step 19846: {'lr': 0.0003479939683259659, 'samples': 3810624, 'steps': 19846, 'loss/train': 0.695938915014267}
+01/27/2022 14:35:23 - INFO - codeparrot_training - Step 19847: {'lr': 0.000347978915052223, 'samples': 3810816, 'steps': 19847, 'loss/train': 0.8767204284667969}
+01/27/2022 14:35:26 - INFO - codeparrot_training - Step 19848: {'lr': 0.0003479638613587696, 'samples': 3811008, 'steps': 19848, 'loss/train': 0.9209830462932587}
+01/27/2022 14:35:30 - INFO - codeparrot_training - Step 19849: {'lr': 0.0003479488072456701, 'samples': 3811200, 'steps': 19849, 'loss/train': 1.2466685771942139}
+01/27/2022 14:35:33 - INFO - codeparrot_training - Step 19850: {'lr': 0.000347933752712989, 'samples': 3811392, 'steps': 19850, 'loss/train': 0.8900529742240906}
+01/27/2022 14:35:36 - INFO - codeparrot_training - Step 19851: {'lr': 0.00034791869776079084, 'samples': 3811584, 'steps': 19851, 'loss/train': 0.8086373805999756}
+01/27/2022 14:35:39 - INFO - codeparrot_training - Step 19852: {'lr': 0.00034790364238914003, 'samples': 3811776, 'steps': 19852, 'loss/train': 0.3686172589659691}
+01/27/2022 14:35:42 - INFO - codeparrot_training - Step 19853: {'lr': 0.0003478885865981011, 'samples': 3811968, 'steps': 19853, 'loss/train': 1.3671565353870392}
+01/27/2022 14:35:47 - INFO - codeparrot_training - Step 19854: {'lr': 0.0003478735303877386, 'samples': 3812160, 'steps': 19854, 'loss/train': 0.965486615896225}
+01/27/2022 14:35:50 - INFO - codeparrot_training - Step 19855: {'lr': 0.0003478584737581169, 'samples': 3812352, 'steps': 19855, 'loss/train': 1.2551003694534302}
+01/27/2022 14:35:53 - INFO - codeparrot_training - Step 19856: {'lr': 0.00034784341670930066, 'samples': 3812544, 'steps': 19856, 'loss/train': 0.9367102682590485}
+01/27/2022 14:35:56 - INFO - codeparrot_training - Step 19857: {'lr': 0.00034782835924135417, 'samples': 3812736, 'steps': 19857, 'loss/train': 0.5664292126893997}
+01/27/2022 14:35:59 - INFO - codeparrot_training - Step 19858: {'lr': 0.0003478133013543422, 'samples': 3812928, 'steps': 19858, 'loss/train': 0.8036853969097137}
+01/27/2022 14:36:02 - INFO - codeparrot_training - Step 19859: {'lr': 0.000347798243048329, 'samples': 3813120, 'steps': 19859, 'loss/train': 1.3768320679664612}
+01/27/2022 14:36:05 - INFO - codeparrot_training - Step 19860: {'lr': 0.00034778318432337926, 'samples': 3813312, 'steps': 19860, 'loss/train': 0.6209570914506912}
+01/27/2022 14:36:09 - INFO - codeparrot_training - Step 19861: {'lr': 0.0003477681251795573, 'samples': 3813504, 'steps': 19861, 'loss/train': 0.7041314989328384}
+01/27/2022 14:36:12 - INFO - codeparrot_training - Step 19862: {'lr': 0.0003477530656169278, 'samples': 3813696, 'steps': 19862, 'loss/train': 0.7967999875545502}
+01/27/2022 14:36:16 - INFO - codeparrot_training - Step 19863: {'lr': 0.00034773800563555517, 'samples': 3813888, 'steps': 19863, 'loss/train': 0.6996750980615616}
+01/27/2022 14:36:20 - INFO - codeparrot_training - Step 19864: {'lr': 0.0003477229452355041, 'samples': 3814080, 'steps': 19864, 'loss/train': 0.2579038739204407}
+01/27/2022 14:36:23 - INFO - codeparrot_training - Step 19865: {'lr': 0.00034770788441683875, 'samples': 3814272, 'steps': 19865, 'loss/train': 0.5895550847053528}
+01/27/2022 14:36:26 - INFO - codeparrot_training - Step 19866: {'lr': 0.00034769282317962405, 'samples': 3814464, 'steps': 19866, 'loss/train': 0.7407797276973724}
+01/27/2022 14:36:29 - INFO - codeparrot_training - Step 19867: {'lr': 0.00034767776152392417, 'samples': 3814656, 'steps': 19867, 'loss/train': 0.8345270454883575}
+01/27/2022 14:36:32 - INFO - codeparrot_training - Step 19868: {'lr': 0.0003476626994498038, 'samples': 3814848, 'steps': 19868, 'loss/train': 0.6220634579658508}
+01/27/2022 14:36:35 - INFO - codeparrot_training - Step 19869: {'lr': 0.00034764763695732746, 'samples': 3815040, 'steps': 19869, 'loss/train': 0.6131460964679718}
+01/27/2022 14:36:38 - INFO - codeparrot_training - Step 19870: {'lr': 0.0003476325740465597, 'samples': 3815232, 'steps': 19870, 'loss/train': 1.0365108847618103}
+01/27/2022 14:36:42 - INFO - codeparrot_training - Step 19871: {'lr': 0.0003476175107175649, 'samples': 3815424, 'steps': 19871, 'loss/train': 0.8843781352043152}
+01/27/2022 14:36:46 - INFO - codeparrot_training - Step 19872: {'lr': 0.00034760244697040776, 'samples': 3815616, 'steps': 19872, 'loss/train': 0.3942389488220215}
+01/27/2022 14:36:49 - INFO - codeparrot_training - Step 19873: {'lr': 0.00034758738280515265, 'samples': 3815808, 'steps': 19873, 'loss/train': 0.47521868348121643}
+01/27/2022 14:36:52 - INFO - codeparrot_training - Step 19874: {'lr': 0.00034757231822186426, 'samples': 3816000, 'steps': 19874, 'loss/train': 1.0749592781066895}
+01/27/2022 14:36:56 - INFO - codeparrot_training - Step 19875: {'lr': 0.00034755725322060705, 'samples': 3816192, 'steps': 19875, 'loss/train': 1.6232061982154846}
+01/27/2022 14:36:59 - INFO - codeparrot_training - Step 19876: {'lr': 0.00034754218780144546, 'samples': 3816384, 'steps': 19876, 'loss/train': 1.0200988948345184}
+01/27/2022 14:37:02 - INFO - codeparrot_training - Step 19877: {'lr': 0.00034752712196444417, 'samples': 3816576, 'steps': 19877, 'loss/train': 0.9826299548149109}
+01/27/2022 14:37:05 - INFO - codeparrot_training - Step 19878: {'lr': 0.00034751205570966764, 'samples': 3816768, 'steps': 19878, 'loss/train': 1.147928148508072}
+01/27/2022 14:37:08 - INFO - codeparrot_training - Step 19879: {'lr': 0.0003474969890371805, 'samples': 3816960, 'steps': 19879, 'loss/train': 0.38178683817386627}
+01/27/2022 14:37:11 - INFO - codeparrot_training - Step 19880: {'lr': 0.0003474819219470471, 'samples': 3817152, 'steps': 19880, 'loss/train': 1.418780654668808}
+01/27/2022 14:37:16 - INFO - codeparrot_training - Step 19881: {'lr': 0.0003474668544393321, 'samples': 3817344, 'steps': 19881, 'loss/train': 0.6610980033874512}
+01/27/2022 14:37:19 - INFO - codeparrot_training - Step 19882: {'lr': 0.00034745178651410014, 'samples': 3817536, 'steps': 19882, 'loss/train': 0.7202821969985962}
+01/27/2022 14:37:23 - INFO - codeparrot_training - Step 19883: {'lr': 0.0003474367181714156, 'samples': 3817728, 'steps': 19883, 'loss/train': 0.9308326542377472}
+01/27/2022 14:37:26 - INFO - codeparrot_training - Step 19884: {'lr': 0.0003474216494113431, 'samples': 3817920, 'steps': 19884, 'loss/train': 0.4929964542388916}
+01/27/2022 14:37:29 - INFO - codeparrot_training - Step 19885: {'lr': 0.00034740658023394723, 'samples': 3818112, 'steps': 19885, 'loss/train': 1.4231217205524445}
+01/27/2022 14:37:32 - INFO - codeparrot_training - Step 19886: {'lr': 0.0003473915106392925, 'samples': 3818304, 'steps': 19886, 'loss/train': 0.44526542723178864}
+01/27/2022 14:37:35 - INFO - codeparrot_training - Step 19887: {'lr': 0.00034737644062744343, 'samples': 3818496, 'steps': 19887, 'loss/train': 0.8696772158145905}
+01/27/2022 14:37:38 - INFO - codeparrot_training - Step 19888: {'lr': 0.0003473613701984646, 'samples': 3818688, 'steps': 19888, 'loss/train': 0.7480353266000748}
+01/27/2022 14:37:41 - INFO - codeparrot_training - Step 19889: {'lr': 0.0003473462993524206, 'samples': 3818880, 'steps': 19889, 'loss/train': 0.46953508257865906}
+01/27/2022 14:37:46 - INFO - codeparrot_training - Step 19890: {'lr': 0.000347331228089376, 'samples': 3819072, 'steps': 19890, 'loss/train': 1.0592038035392761}
+01/27/2022 14:37:49 - INFO - codeparrot_training - Step 19891: {'lr': 0.0003473161564093953, 'samples': 3819264, 'steps': 19891, 'loss/train': 2.0833654403686523}
+01/27/2022 14:37:52 - INFO - codeparrot_training - Step 19892: {'lr': 0.0003473010843125431, 'samples': 3819456, 'steps': 19892, 'loss/train': 1.1808803379535675}
+01/27/2022 14:37:56 - INFO - codeparrot_training - Step 19893: {'lr': 0.00034728601179888395, 'samples': 3819648, 'steps': 19893, 'loss/train': 0.675475001335144}
+01/27/2022 14:37:59 - INFO - codeparrot_training - Step 19894: {'lr': 0.00034727093886848236, 'samples': 3819840, 'steps': 19894, 'loss/train': 0.6707288771867752}
+01/27/2022 14:38:02 - INFO - codeparrot_training - Step 19895: {'lr': 0.000347255865521403, 'samples': 3820032, 'steps': 19895, 'loss/train': 0.8673428893089294}
+01/27/2022 14:38:05 - INFO - codeparrot_training - Step 19896: {'lr': 0.0003472407917577104, 'samples': 3820224, 'steps': 19896, 'loss/train': 0.6748683303594589}
+01/27/2022 14:38:08 - INFO - codeparrot_training - Step 19897: {'lr': 0.0003472257175774691, 'samples': 3820416, 'steps': 19897, 'loss/train': 0.7604217231273651}
+01/27/2022 14:38:11 - INFO - codeparrot_training - Step 19898: {'lr': 0.00034721064298074366, 'samples': 3820608, 'steps': 19898, 'loss/train': 0.8023392856121063}
+01/27/2022 14:38:16 - INFO - codeparrot_training - Step 19899: {'lr': 0.0003471955679675988, 'samples': 3820800, 'steps': 19899, 'loss/train': 1.036697655916214}
+01/27/2022 14:38:19 - INFO - codeparrot_training - Step 19900: {'lr': 0.0003471804925380989, 'samples': 3820992, 'steps': 19900, 'loss/train': 0.6142769604921341}
+01/27/2022 14:38:22 - INFO - codeparrot_training - Step 19901: {'lr': 0.0003471654166923087, 'samples': 3821184, 'steps': 19901, 'loss/train': 1.0707724392414093}
+01/27/2022 14:38:25 - INFO - codeparrot_training - Step 19902: {'lr': 0.00034715034043029263, 'samples': 3821376, 'steps': 19902, 'loss/train': 0.8523239493370056}
+01/27/2022 14:38:28 - INFO - codeparrot_training - Step 19903: {'lr': 0.00034713526375211546, 'samples': 3821568, 'steps': 19903, 'loss/train': 0.8656088411808014}
+01/27/2022 14:38:31 - INFO - codeparrot_training - Step 19904: {'lr': 0.00034712018665784155, 'samples': 3821760, 'steps': 19904, 'loss/train': 0.9913526773452759}
+01/27/2022 14:38:34 - INFO - codeparrot_training - Step 19905: {'lr': 0.00034710510914753563, 'samples': 3821952, 'steps': 19905, 'loss/train': 1.073285311460495}
+01/27/2022 14:38:38 - INFO - codeparrot_training - Step 19906: {'lr': 0.00034709003122126227, 'samples': 3822144, 'steps': 19906, 'loss/train': 0.926423192024231}
+01/27/2022 14:38:41 - INFO - codeparrot_training - Step 19907: {'lr': 0.000347074952879086, 'samples': 3822336, 'steps': 19907, 'loss/train': 0.7737300395965576}
+01/27/2022 14:38:47 - INFO - codeparrot_training - Step 19908: {'lr': 0.0003470598741210715, 'samples': 3822528, 'steps': 19908, 'loss/train': 0.7037508934736252}
+01/27/2022 14:38:51 - INFO - codeparrot_training - Step 19909: {'lr': 0.00034704479494728337, 'samples': 3822720, 'steps': 19909, 'loss/train': 0.7650665938854218}
+01/27/2022 14:38:54 - INFO - codeparrot_training - Step 19910: {'lr': 0.00034702971535778614, 'samples': 3822912, 'steps': 19910, 'loss/train': 0.7969690561294556}
+01/27/2022 14:38:57 - INFO - codeparrot_training - Step 19911: {'lr': 0.00034701463535264434, 'samples': 3823104, 'steps': 19911, 'loss/train': 0.6664220988750458}
+01/27/2022 14:39:00 - INFO - codeparrot_training - Step 19912: {'lr': 0.0003469995549319227, 'samples': 3823296, 'steps': 19912, 'loss/train': 0.9040322005748749}
+01/27/2022 14:39:03 - INFO - codeparrot_training - Step 19913: {'lr': 0.0003469844740956858, 'samples': 3823488, 'steps': 19913, 'loss/train': 0.5811881124973297}
+01/27/2022 14:39:06 - INFO - codeparrot_training - Step 19914: {'lr': 0.0003469693928439982, 'samples': 3823680, 'steps': 19914, 'loss/train': 0.20418034493923187}
+01/27/2022 14:39:09 - INFO - codeparrot_training - Step 19915: {'lr': 0.00034695431117692446, 'samples': 3823872, 'steps': 19915, 'loss/train': 1.048704832792282}
+01/27/2022 14:39:13 - INFO - codeparrot_training - Step 19916: {'lr': 0.0003469392290945292, 'samples': 3824064, 'steps': 19916, 'loss/train': 0.6516337841749191}
+01/27/2022 14:39:18 - INFO - codeparrot_training - Step 19917: {'lr': 0.00034692414659687714, 'samples': 3824256, 'steps': 19917, 'loss/train': 0.6595115661621094}
+01/27/2022 14:39:21 - INFO - codeparrot_training - Step 19918: {'lr': 0.0003469090636840328, 'samples': 3824448, 'steps': 19918, 'loss/train': 0.7805255055427551}
+01/27/2022 14:39:24 - INFO - codeparrot_training - Step 19919: {'lr': 0.0003468939803560608, 'samples': 3824640, 'steps': 19919, 'loss/train': 1.2523106038570404}
+01/27/2022 14:39:27 - INFO - codeparrot_training - Step 19920: {'lr': 0.00034687889661302575, 'samples': 3824832, 'steps': 19920, 'loss/train': 0.5439706742763519}
+01/27/2022 14:39:30 - INFO - codeparrot_training - Step 19921: {'lr': 0.0003468638124549923, 'samples': 3825024, 'steps': 19921, 'loss/train': 0.6259864121675491}
+01/27/2022 14:39:34 - INFO - codeparrot_training - Step 19922: {'lr': 0.00034684872788202497, 'samples': 3825216, 'steps': 19922, 'loss/train': 1.7887920141220093}
+01/27/2022 14:39:37 - INFO - codeparrot_training - Step 19923: {'lr': 0.0003468336428941885, 'samples': 3825408, 'steps': 19923, 'loss/train': 0.6546574383974075}
+01/27/2022 14:39:40 - INFO - codeparrot_training - Step 19924: {'lr': 0.00034681855749154743, 'samples': 3825600, 'steps': 19924, 'loss/train': 5.663194298744202}
+01/27/2022 14:39:43 - INFO - codeparrot_training - Step 19925: {'lr': 0.00034680347167416643, 'samples': 3825792, 'steps': 19925, 'loss/train': 0.6536028832197189}
+01/27/2022 14:39:46 - INFO - codeparrot_training - Step 19926: {'lr': 0.00034678838544211003, 'samples': 3825984, 'steps': 19926, 'loss/train': 1.7352635264396667}
+01/27/2022 14:39:52 - INFO - codeparrot_training - Step 19927: {'lr': 0.000346773298795443, 'samples': 3826176, 'steps': 19927, 'loss/train': 0.7690545916557312}
+01/27/2022 14:39:55 - INFO - codeparrot_training - Step 19928: {'lr': 0.00034675821173422983, 'samples': 3826368, 'steps': 19928, 'loss/train': 0.8367191255092621}
+01/27/2022 14:39:59 - INFO - codeparrot_training - Step 19929: {'lr': 0.0003467431242585352, 'samples': 3826560, 'steps': 19929, 'loss/train': 0.5961487144231796}
+01/27/2022 14:40:02 - INFO - codeparrot_training - Step 19930: {'lr': 0.0003467280363684238, 'samples': 3826752, 'steps': 19930, 'loss/train': 0.8861354291439056}
+01/27/2022 14:40:05 - INFO - codeparrot_training - Step 19931: {'lr': 0.0003467129480639601, 'samples': 3826944, 'steps': 19931, 'loss/train': 0.9080999493598938}
+01/27/2022 14:40:08 - INFO - codeparrot_training - Step 19932: {'lr': 0.000346697859345209, 'samples': 3827136, 'steps': 19932, 'loss/train': 0.6633761376142502}
+01/27/2022 14:40:11 - INFO - codeparrot_training - Step 19933: {'lr': 0.00034668277021223493, 'samples': 3827328, 'steps': 19933, 'loss/train': 0.6035650223493576}
+01/27/2022 14:40:14 - INFO - codeparrot_training - Step 19934: {'lr': 0.0003466676806651025, 'samples': 3827520, 'steps': 19934, 'loss/train': 0.8759147822856903}
+01/27/2022 14:40:17 - INFO - codeparrot_training - Step 19935: {'lr': 0.0003466525907038765, 'samples': 3827712, 'steps': 19935, 'loss/train': 0.8005030453205109}
+01/27/2022 14:40:22 - INFO - codeparrot_training - Step 19936: {'lr': 0.0003466375003286214, 'samples': 3827904, 'steps': 19936, 'loss/train': 0.9488546848297119}
+01/27/2022 14:40:25 - INFO - codeparrot_training - Step 19937: {'lr': 0.00034662240953940205, 'samples': 3828096, 'steps': 19937, 'loss/train': 0.6040018200874329}
+01/27/2022 14:40:28 - INFO - codeparrot_training - Step 19938: {'lr': 0.0003466073183362829, 'samples': 3828288, 'steps': 19938, 'loss/train': 0.8734752237796783}
+01/27/2022 14:40:31 - INFO - codeparrot_training - Step 19939: {'lr': 0.00034659222671932865, 'samples': 3828480, 'steps': 19939, 'loss/train': 0.7465503215789795}
+01/27/2022 14:40:35 - INFO - codeparrot_training - Step 19940: {'lr': 0.000346577134688604, 'samples': 3828672, 'steps': 19940, 'loss/train': 0.6718389987945557}
+01/27/2022 14:40:38 - INFO - codeparrot_training - Step 19941: {'lr': 0.0003465620422441737, 'samples': 3828864, 'steps': 19941, 'loss/train': 0.7507781982421875}
+01/27/2022 14:40:41 - INFO - codeparrot_training - Step 19942: {'lr': 0.00034654694938610205, 'samples': 3829056, 'steps': 19942, 'loss/train': 0.6716022044420242}
+01/27/2022 14:40:44 - INFO - codeparrot_training - Step 19943: {'lr': 0.00034653185611445403, 'samples': 3829248, 'steps': 19943, 'loss/train': 0.8509013056755066}
+01/27/2022 14:40:47 - INFO - codeparrot_training - Step 19944: {'lr': 0.0003465167624292942, 'samples': 3829440, 'steps': 19944, 'loss/train': 1.0307892858982086}
+01/27/2022 14:40:52 - INFO - codeparrot_training - Step 19945: {'lr': 0.0003465016683306872, 'samples': 3829632, 'steps': 19945, 'loss/train': 0.9924068748950958}
+01/27/2022 14:40:55 - INFO - codeparrot_training - Step 19946: {'lr': 0.0003464865738186977, 'samples': 3829824, 'steps': 19946, 'loss/train': 1.3496181070804596}
+01/27/2022 14:40:58 - INFO - codeparrot_training - Step 19947: {'lr': 0.0003464714788933904, 'samples': 3830016, 'steps': 19947, 'loss/train': 0.8892586827278137}
+01/27/2022 14:41:01 - INFO - codeparrot_training - Step 19948: {'lr': 0.0003464563835548298, 'samples': 3830208, 'steps': 19948, 'loss/train': 0.852963924407959}
+01/27/2022 14:41:05 - INFO - codeparrot_training - Step 19949: {'lr': 0.0003464412878030808, 'samples': 3830400, 'steps': 19949, 'loss/train': 0.6909165680408478}
+01/27/2022 14:41:08 - INFO - codeparrot_training - Step 19950: {'lr': 0.0003464261916382079, 'samples': 3830592, 'steps': 19950, 'loss/train': 0.6429280936717987}
+01/27/2022 14:41:11 - INFO - codeparrot_training - Step 19951: {'lr': 0.0003464110950602758, 'samples': 3830784, 'steps': 19951, 'loss/train': 1.2538221180438995}
+01/27/2022 14:41:14 - INFO - codeparrot_training - Step 19952: {'lr': 0.00034639599806934917, 'samples': 3830976, 'steps': 19952, 'loss/train': 0.952578216791153}
+01/27/2022 14:41:21 - INFO - codeparrot_training - Step 19953: {'lr': 0.0003463809006654927, 'samples': 3831168, 'steps': 19953, 'loss/train': 2.022986948490143}
+01/27/2022 14:41:24 - INFO - codeparrot_training - Step 19954: {'lr': 0.0003463658028487711, 'samples': 3831360, 'steps': 19954, 'loss/train': 1.1705638468265533}
+01/27/2022 14:41:27 - INFO - codeparrot_training - Step 19955: {'lr': 0.0003463507046192489, 'samples': 3831552, 'steps': 19955, 'loss/train': 1.3308018743991852}
+01/27/2022 14:41:30 - INFO - codeparrot_training - Step 19956: {'lr': 0.000346335605976991, 'samples': 3831744, 'steps': 19956, 'loss/train': 1.0911744832992554}
+01/27/2022 14:41:33 - INFO - codeparrot_training - Step 19957: {'lr': 0.00034632050692206184, 'samples': 3831936, 'steps': 19957, 'loss/train': 1.1510403156280518}
+01/27/2022 14:41:36 - INFO - codeparrot_training - Step 19958: {'lr': 0.0003463054074545263, 'samples': 3832128, 'steps': 19958, 'loss/train': 0.3634873554110527}
+01/27/2022 14:41:39 - INFO - codeparrot_training - Step 19959: {'lr': 0.00034629030757444895, 'samples': 3832320, 'steps': 19959, 'loss/train': 1.19656303524971}
+01/27/2022 14:41:43 - INFO - codeparrot_training - Step 19960: {'lr': 0.00034627520728189456, 'samples': 3832512, 'steps': 19960, 'loss/train': 1.403536319732666}
+01/27/2022 14:41:46 - INFO - codeparrot_training - Step 19961: {'lr': 0.00034626010657692766, 'samples': 3832704, 'steps': 19961, 'loss/train': 0.6550229340791702}
+01/27/2022 14:41:50 - INFO - codeparrot_training - Step 19962: {'lr': 0.0003462450054596131, 'samples': 3832896, 'steps': 19962, 'loss/train': 0.9628220200538635}
+01/27/2022 14:41:53 - INFO - codeparrot_training - Step 19963: {'lr': 0.0003462299039300154, 'samples': 3833088, 'steps': 19963, 'loss/train': 0.489440381526947}
+01/27/2022 14:41:56 - INFO - codeparrot_training - Step 19964: {'lr': 0.0003462148019881994, 'samples': 3833280, 'steps': 19964, 'loss/train': 0.25204624235630035}
+01/27/2022 14:42:00 - INFO - codeparrot_training - Step 19965: {'lr': 0.0003461996996342298, 'samples': 3833472, 'steps': 19965, 'loss/train': 0.6812572628259659}
+01/27/2022 14:42:03 - INFO - codeparrot_training - Step 19966: {'lr': 0.00034618459686817113, 'samples': 3833664, 'steps': 19966, 'loss/train': 1.0840181708335876}
+01/27/2022 14:42:06 - INFO - codeparrot_training - Step 19967: {'lr': 0.0003461694936900883, 'samples': 3833856, 'steps': 19967, 'loss/train': 0.9728666245937347}
+01/27/2022 14:42:09 - INFO - codeparrot_training - Step 19968: {'lr': 0.0003461543901000458, 'samples': 3834048, 'steps': 19968, 'loss/train': 0.6055312156677246}
+01/27/2022 14:42:12 - INFO - codeparrot_training - Step 19969: {'lr': 0.00034613928609810845, 'samples': 3834240, 'steps': 19969, 'loss/train': 0.8488118648529053}
+01/27/2022 14:42:15 - INFO - codeparrot_training - Step 19970: {'lr': 0.0003461241816843409, 'samples': 3834432, 'steps': 19970, 'loss/train': 0.9936572313308716}
+01/27/2022 14:42:22 - INFO - codeparrot_training - Step 19971: {'lr': 0.00034610907685880794, 'samples': 3834624, 'steps': 19971, 'loss/train': 1.0092098414897919}
+01/27/2022 14:42:25 - INFO - codeparrot_training - Step 19972: {'lr': 0.00034609397162157417, 'samples': 3834816, 'steps': 19972, 'loss/train': 1.8019208908081055}
+01/27/2022 14:42:28 - INFO - codeparrot_training - Step 19973: {'lr': 0.0003460788659727044, 'samples': 3835008, 'steps': 19973, 'loss/train': 0.7596907317638397}
+01/27/2022 14:42:32 - INFO - codeparrot_training - Step 19974: {'lr': 0.0003460637599122632, 'samples': 3835200, 'steps': 19974, 'loss/train': 0.411023274064064}
+01/27/2022 14:42:35 - INFO - codeparrot_training - Step 19975: {'lr': 0.0003460486534403154, 'samples': 3835392, 'steps': 19975, 'loss/train': 0.6755359321832657}
+01/27/2022 14:42:38 - INFO - codeparrot_training - Step 19976: {'lr': 0.0003460335465569256, 'samples': 3835584, 'steps': 19976, 'loss/train': 1.1234873235225677}
+01/27/2022 14:42:41 - INFO - codeparrot_training - Step 19977: {'lr': 0.0003460184392621587, 'samples': 3835776, 'steps': 19977, 'loss/train': 0.568824514746666}
+01/27/2022 14:42:44 - INFO - codeparrot_training - Step 19978: {'lr': 0.0003460033315560792, 'samples': 3835968, 'steps': 19978, 'loss/train': 0.6737221330404282}
+01/27/2022 14:42:47 - INFO - codeparrot_training - Step 19979: {'lr': 0.00034598822343875197, 'samples': 3836160, 'steps': 19979, 'loss/train': 0.7464563548564911}
+01/27/2022 14:42:50 - INFO - codeparrot_training - Step 19980: {'lr': 0.0003459731149102417, 'samples': 3836352, 'steps': 19980, 'loss/train': 2.0262778401374817}
+01/27/2022 14:42:55 - INFO - codeparrot_training - Step 19981: {'lr': 0.000345958005970613, 'samples': 3836544, 'steps': 19981, 'loss/train': 0.46031682193279266}
+01/27/2022 14:42:58 - INFO - codeparrot_training - Step 19982: {'lr': 0.0003459428966199307, 'samples': 3836736, 'steps': 19982, 'loss/train': 0.20641037821769714}
+01/27/2022 14:43:01 - INFO - codeparrot_training - Step 19983: {'lr': 0.0003459277868582595, 'samples': 3836928, 'steps': 19983, 'loss/train': 0.9296490848064423}
+01/27/2022 14:43:04 - INFO - codeparrot_training - Step 19984: {'lr': 0.00034591267668566413, 'samples': 3837120, 'steps': 19984, 'loss/train': 0.833126038312912}
+01/27/2022 14:43:07 - INFO - codeparrot_training - Step 19985: {'lr': 0.00034589756610220923, 'samples': 3837312, 'steps': 19985, 'loss/train': 0.53171606361866}
+01/27/2022 14:43:11 - INFO - codeparrot_training - Step 19986: {'lr': 0.0003458824551079597, 'samples': 3837504, 'steps': 19986, 'loss/train': 1.2326968610286713}
+01/27/2022 14:43:14 - INFO - codeparrot_training - Step 19987: {'lr': 0.00034586734370298017, 'samples': 3837696, 'steps': 19987, 'loss/train': 0.5999709069728851}
+01/27/2022 14:43:17 - INFO - codeparrot_training - Step 19988: {'lr': 0.00034585223188733535, 'samples': 3837888, 'steps': 19988, 'loss/train': 0.8908984065055847}
+01/27/2022 14:43:20 - INFO - codeparrot_training - Step 19989: {'lr': 0.00034583711966109005, 'samples': 3838080, 'steps': 19989, 'loss/train': 0.7726276516914368}
+01/27/2022 14:43:25 - INFO - codeparrot_training - Step 19990: {'lr': 0.0003458220070243089, 'samples': 3838272, 'steps': 19990, 'loss/train': 0.8756560385227203}
+01/27/2022 14:43:28 - INFO - codeparrot_training - Step 19991: {'lr': 0.0003458068939770567, 'samples': 3838464, 'steps': 19991, 'loss/train': 0.6107423007488251}
+01/27/2022 14:43:31 - INFO - codeparrot_training - Step 19992: {'lr': 0.00034579178051939827, 'samples': 3838656, 'steps': 19992, 'loss/train': 0.4919268786907196}
+01/27/2022 14:43:34 - INFO - codeparrot_training - Step 19993: {'lr': 0.00034577666665139815, 'samples': 3838848, 'steps': 19993, 'loss/train': 1.175744354724884}
+01/27/2022 14:43:37 - INFO - codeparrot_training - Step 19994: {'lr': 0.0003457615523731213, 'samples': 3839040, 'steps': 19994, 'loss/train': 0.7821950018405914}
+01/27/2022 14:43:40 - INFO - codeparrot_training - Step 19995: {'lr': 0.00034574643768463237, 'samples': 3839232, 'steps': 19995, 'loss/train': 0.9034610688686371}
+01/27/2022 14:43:44 - INFO - codeparrot_training - Step 19996: {'lr': 0.00034573132258599606, 'samples': 3839424, 'steps': 19996, 'loss/train': 0.6069746017456055}
+01/27/2022 14:43:47 - INFO - codeparrot_training - Step 19997: {'lr': 0.00034571620707727713, 'samples': 3839616, 'steps': 19997, 'loss/train': 0.9488583505153656}
+01/27/2022 14:43:53 - INFO - codeparrot_training - Step 19998: {'lr': 0.0003457010911585404, 'samples': 3839808, 'steps': 19998, 'loss/train': 1.1606356501579285}
+01/27/2022 14:43:56 - INFO - codeparrot_training - Step 19999: {'lr': 0.00034568597482985067, 'samples': 3840000, 'steps': 19999, 'loss/train': 0.9583664238452911}
+01/27/2022 14:43:56 - INFO - codeparrot_training - Evaluating and saving model checkpoint