diff --git "a/log/debug_0.log" "b/log/debug_0.log"
--- "a/log/debug_0.log"
+++ "b/log/debug_0.log"
@@ -48590,3 +48590,1009 @@ Use FP16 precision: False
 02/26/2022 15:44:38 - INFO - codeparrot_training - Step 47998: {'lr': 2.143058239704565e-06, 'samples': 24575488, 'steps': 47998, 'loss/train': 2.02597975730896}
 02/26/2022 15:44:41 - INFO - codeparrot_training - Step 47999: {'lr': 2.140920917251182e-06, 'samples': 24576000, 'steps': 47999, 'loss/train': 1.8916027545928955}
 02/26/2022 15:44:41 - INFO - codeparrot_training - Evaluating and saving model checkpoint
+02/26/2022 15:44:59 - WARNING - huggingface_hub.repository - Several commits (48) will be pushed upstream.
+02/26/2022 15:44:59 - WARNING - huggingface_hub.repository - The progress bars may be unreliable.
+02/26/2022 15:45:34 - WARNING - huggingface_hub.repository - To https://huggingface.co/ncoop57/multi-code-clippy
+   2e5c310..8e26ae5  floral-grass-11 -> floral-grass-11
+
+02/26/2022 15:45:43 - INFO - codeparrot_training - Step 48000: {'lr': 2.1387846565474044e-06, 'samples': 24576512, 'steps': 48000, 'loss/train': 0.4398959279060364}
+02/26/2022 15:45:46 - INFO - codeparrot_training - Step 48001: {'lr': 2.136649457602363e-06, 'samples': 24577024, 'steps': 48001, 'loss/train': 2.441908359527588}
+02/26/2022 15:45:52 - INFO - codeparrot_training - Step 48002: {'lr': 2.1345153204252175e-06, 'samples': 24577536, 'steps': 48002, 'loss/train': 1.3035922050476074}
+02/26/2022 15:45:55 - INFO - codeparrot_training - Step 48003: {'lr': 2.1323822450250997e-06, 'samples': 24578048, 'steps': 48003, 'loss/train': 2.239109516143799}
+02/26/2022 15:46:01 - INFO - codeparrot_training - Step 48004: {'lr': 2.1302502314111406e-06, 'samples': 24578560, 'steps': 48004, 'loss/train': 2.2362773418426514}
+02/26/2022 15:46:04 - INFO - codeparrot_training - Step 48005: {'lr': 2.128119279592472e-06, 'samples': 24579072, 'steps': 48005, 'loss/train': 0.5483190417289734}
+02/26/2022 15:46:10 - INFO - codeparrot_training - Step 48006: {'lr': 2.1259893895782533e-06, 'samples': 24579584, 'steps': 48006, 'loss/train': 1.788357138633728}
+02/26/2022 15:46:13 - INFO - codeparrot_training - Step 48007: {'lr': 2.1238605613775887e-06, 'samples': 24580096, 'steps': 48007, 'loss/train': 1.2798937559127808}
+02/26/2022 15:46:19 - INFO - codeparrot_training - Step 48008: {'lr': 2.121732794999581e-06, 'samples': 24580608, 'steps': 48008, 'loss/train': 1.9912223815917969}
+02/26/2022 15:46:22 - INFO - codeparrot_training - Step 48009: {'lr': 2.1196060904533632e-06, 'samples': 24581120, 'steps': 48009, 'loss/train': 1.3062742948532104}
+02/26/2022 15:46:29 - INFO - codeparrot_training - Step 48010: {'lr': 2.117480447748066e-06, 'samples': 24581632, 'steps': 48010, 'loss/train': 3.3507866859436035}
+02/26/2022 15:46:33 - INFO - codeparrot_training - Step 48011: {'lr': 2.1153558668927375e-06, 'samples': 24582144, 'steps': 48011, 'loss/train': 1.4072787761688232}
+02/26/2022 15:46:39 - INFO - codeparrot_training - Step 48012: {'lr': 2.1132323478965374e-06, 'samples': 24582656, 'steps': 48012, 'loss/train': 1.1705113649368286}
+02/26/2022 15:46:42 - INFO - codeparrot_training - Step 48013: {'lr': 2.1111098907685144e-06, 'samples': 24583168, 'steps': 48013, 'loss/train': 0.9358199238777161}
+02/26/2022 15:46:48 - INFO - codeparrot_training - Step 48014: {'lr': 2.1089884955177996e-06, 'samples': 24583680, 'steps': 48014, 'loss/train': 2.55591082572937}
+02/26/2022 15:46:51 - INFO - codeparrot_training - Step 48015: {'lr': 2.1068681621534693e-06, 'samples': 24584192, 'steps': 48015, 'loss/train': 2.250408411026001}
+02/26/2022 15:46:57 - INFO - codeparrot_training - Step 48016: {'lr': 2.1047488906845712e-06, 'samples': 24584704, 'steps': 48016, 'loss/train': 1.4187424182891846}
+02/26/2022 15:47:00 - INFO - codeparrot_training - Step 48017: {'lr': 2.102630681120238e-06, 'samples': 24585216, 'steps': 48017, 'loss/train': 1.6364868879318237}
+02/26/2022 15:47:06 - INFO - codeparrot_training - Step 48018: {'lr': 2.100513533469517e-06, 'samples': 24585728, 'steps': 48018, 'loss/train': 1.7462069988250732}
+02/26/2022 15:47:09 - INFO - codeparrot_training - Step 48019: {'lr': 2.098397447741485e-06, 'samples': 24586240, 'steps': 48019, 'loss/train': 1.7926084995269775}
+02/26/2022 15:47:15 - INFO - codeparrot_training - Step 48020: {'lr': 2.0962824239451893e-06, 'samples': 24586752, 'steps': 48020, 'loss/train': 1.2424486875534058}
+02/26/2022 15:47:18 - INFO - codeparrot_training - Step 48021: {'lr': 2.094168462089707e-06, 'samples': 24587264, 'steps': 48021, 'loss/train': 2.1140058040618896}
+02/26/2022 15:47:24 - INFO - codeparrot_training - Step 48022: {'lr': 2.0920555621840588e-06, 'samples': 24587776, 'steps': 48022, 'loss/train': 2.862093687057495}
+02/26/2022 15:47:28 - INFO - codeparrot_training - Step 48023: {'lr': 2.0899437242373754e-06, 'samples': 24588288, 'steps': 48023, 'loss/train': 2.104945182800293}
+02/26/2022 15:47:33 - INFO - codeparrot_training - Step 48024: {'lr': 2.0878329482586223e-06, 'samples': 24588800, 'steps': 48024, 'loss/train': 1.7665393352508545}
+02/26/2022 15:47:37 - INFO - codeparrot_training - Step 48025: {'lr': 2.0857232342568754e-06, 'samples': 24589312, 'steps': 48025, 'loss/train': 1.6297985315322876}
+02/26/2022 15:47:44 - INFO - codeparrot_training - Step 48026: {'lr': 2.083614582241156e-06, 'samples': 24589824, 'steps': 48026, 'loss/train': 1.2695255279541016}
+02/26/2022 15:47:47 - INFO - codeparrot_training - Step 48027: {'lr': 2.081506992220539e-06, 'samples': 24590336, 'steps': 48027, 'loss/train': 1.9487683773040771}
+02/26/2022 15:47:53 - INFO - codeparrot_training - Step 48028: {'lr': 2.07940046420399e-06, 'samples': 24590848, 'steps': 48028, 'loss/train': 1.2654321193695068}
+02/26/2022 15:47:57 - INFO - codeparrot_training - Step 48029: {'lr': 2.0772949982005853e-06, 'samples': 24591360, 'steps': 48029, 'loss/train': 1.6626259088516235}
+02/26/2022 15:48:02 - INFO - codeparrot_training - Step 48030: {'lr': 2.075190594219317e-06, 'samples': 24591872, 'steps': 48030, 'loss/train': 1.4501103162765503}
+02/26/2022 15:48:06 - INFO - codeparrot_training - Step 48031: {'lr': 2.073087252269207e-06, 'samples': 24592384, 'steps': 48031, 'loss/train': 1.3885493278503418}
+02/26/2022 15:48:11 - INFO - codeparrot_training - Step 48032: {'lr': 2.0709849723593023e-06, 'samples': 24592896, 'steps': 48032, 'loss/train': 1.429519772529602}
+02/26/2022 15:48:15 - INFO - codeparrot_training - Step 48033: {'lr': 2.0688837544985407e-06, 'samples': 24593408, 'steps': 48033, 'loss/train': 2.0424609184265137}
+02/26/2022 15:48:20 - INFO - codeparrot_training - Step 48034: {'lr': 2.066783598695943e-06, 'samples': 24593920, 'steps': 48034, 'loss/train': 1.3377666473388672}
+02/26/2022 15:48:24 - INFO - codeparrot_training - Step 48035: {'lr': 2.064684504960529e-06, 'samples': 24594432, 'steps': 48035, 'loss/train': 2.2199840545654297}
+02/26/2022 15:48:29 - INFO - codeparrot_training - Step 48036: {'lr': 2.0625864733012645e-06, 'samples': 24594944, 'steps': 48036, 'loss/train': 2.0081417560577393}
+02/26/2022 15:48:33 - INFO - codeparrot_training - Step 48037: {'lr': 2.06048950372717e-06, 'samples': 24595456, 'steps': 48037, 'loss/train': 1.405119776725769}
+02/26/2022 15:48:40 - INFO - codeparrot_training - Step 48038: {'lr': 2.05839359624721e-06, 'samples': 24595968, 'steps': 48038, 'loss/train': 1.1608302593231201}
+02/26/2022 15:48:43 - INFO - codeparrot_training - Step 48039: {'lr': 2.05629875087035e-06, 'samples': 24596480, 'steps': 48039, 'loss/train': 2.286485433578491}
+02/26/2022 15:48:49 - INFO - codeparrot_training - Step 48040: {'lr': 2.054204967605583e-06, 'samples': 24596992, 'steps': 48040, 'loss/train': 2.239549160003662}
+02/26/2022 15:48:52 - INFO - codeparrot_training - Step 48041: {'lr': 2.0521122464618735e-06, 'samples': 24597504, 'steps': 48041, 'loss/train': 1.8396660089492798}
+02/26/2022 15:48:58 - INFO - codeparrot_training - Step 48042: {'lr': 2.0500205874481593e-06, 'samples': 24598016, 'steps': 48042, 'loss/train': 1.7345612049102783}
+02/26/2022 15:49:01 - INFO - codeparrot_training - Step 48043: {'lr': 2.047929990573433e-06, 'samples': 24598528, 'steps': 48043, 'loss/train': 0.44698697328567505}
+02/26/2022 15:49:07 - INFO - codeparrot_training - Step 48044: {'lr': 2.0458404558466593e-06, 'samples': 24599040, 'steps': 48044, 'loss/train': 2.0572776794433594}
+02/26/2022 15:49:10 - INFO - codeparrot_training - Step 48045: {'lr': 2.0437519832767483e-06, 'samples': 24599552, 'steps': 48045, 'loss/train': 1.5602202415466309}
+02/26/2022 15:49:16 - INFO - codeparrot_training - Step 48046: {'lr': 2.0416645728726922e-06, 'samples': 24600064, 'steps': 48046, 'loss/train': 2.6575729846954346}
+02/26/2022 15:49:19 - INFO - codeparrot_training - Step 48047: {'lr': 2.0395782246433737e-06, 'samples': 24600576, 'steps': 48047, 'loss/train': 0.8888649344444275}
+02/26/2022 15:49:27 - INFO - codeparrot_training - Step 48048: {'lr': 2.0374929385978125e-06, 'samples': 24601088, 'steps': 48048, 'loss/train': 1.738717794418335}
+02/26/2022 15:49:30 - INFO - codeparrot_training - Step 48049: {'lr': 2.035408714744863e-06, 'samples': 24601600, 'steps': 48049, 'loss/train': 0.8226458430290222}
+02/26/2022 15:49:36 - INFO - codeparrot_training - Step 48050: {'lr': 2.0333255530934903e-06, 'samples': 24602112, 'steps': 48050, 'loss/train': 1.8816875219345093}
+02/26/2022 15:49:39 - INFO - codeparrot_training - Step 48051: {'lr': 2.031243453652604e-06, 'samples': 24602624, 'steps': 48051, 'loss/train': 2.0343282222747803}
+02/26/2022 15:49:45 - INFO - codeparrot_training - Step 48052: {'lr': 2.0291624164311686e-06, 'samples': 24603136, 'steps': 48052, 'loss/train': 1.7116533517837524}
+02/26/2022 15:49:48 - INFO - codeparrot_training - Step 48053: {'lr': 2.0270824414380107e-06, 'samples': 24603648, 'steps': 48053, 'loss/train': 1.1586358547210693}
+02/26/2022 15:49:54 - INFO - codeparrot_training - Step 48054: {'lr': 2.025003528682151e-06, 'samples': 24604160, 'steps': 48054, 'loss/train': 2.3007030487060547}
+02/26/2022 15:49:57 - INFO - codeparrot_training - Step 48055: {'lr': 2.022925678172388e-06, 'samples': 24604672, 'steps': 48055, 'loss/train': 2.2319235801696777}
+02/26/2022 15:50:03 - INFO - codeparrot_training - Step 48056: {'lr': 2.020848889917687e-06, 'samples': 24605184, 'steps': 48056, 'loss/train': 1.3865286111831665}
+02/26/2022 15:50:06 - INFO - codeparrot_training - Step 48057: {'lr': 2.018773163926901e-06, 'samples': 24605696, 'steps': 48057, 'loss/train': 1.8625080585479736}
+02/26/2022 15:50:14 - INFO - codeparrot_training - Step 48058: {'lr': 2.0166985002089967e-06, 'samples': 24606208, 'steps': 48058, 'loss/train': 1.929304599761963}
+02/26/2022 15:50:17 - INFO - codeparrot_training - Step 48059: {'lr': 2.014624898772771e-06, 'samples': 24606720, 'steps': 48059, 'loss/train': 1.9097601175308228}
+02/26/2022 15:50:23 - INFO - codeparrot_training - Step 48060: {'lr': 2.0125523596271623e-06, 'samples': 24607232, 'steps': 48060, 'loss/train': 1.6167644262313843}
+02/26/2022 15:50:26 - INFO - codeparrot_training - Step 48061: {'lr': 2.010480882781024e-06, 'samples': 24607744, 'steps': 48061, 'loss/train': 1.1160197257995605}
+02/26/2022 15:50:32 - INFO - codeparrot_training - Step 48062: {'lr': 2.0084104682432657e-06, 'samples': 24608256, 'steps': 48062, 'loss/train': 2.3234784603118896}
+02/26/2022 15:50:36 - INFO - codeparrot_training - Step 48063: {'lr': 2.0063411160227143e-06, 'samples': 24608768, 'steps': 48063, 'loss/train': 3.278146982192993}
+02/26/2022 15:50:41 - INFO - codeparrot_training - Step 48064: {'lr': 2.0042728261282504e-06, 'samples': 24609280, 'steps': 48064, 'loss/train': 1.760576605796814}
+02/26/2022 15:50:45 - INFO - codeparrot_training - Step 48065: {'lr': 2.002205598568729e-06, 'samples': 24609792, 'steps': 48065, 'loss/train': 1.1821355819702148}
+02/26/2022 15:50:50 - INFO - codeparrot_training - Step 48066: {'lr': 2.0001394333530322e-06, 'samples': 24610304, 'steps': 48066, 'loss/train': 0.029472487047314644}
+02/26/2022 15:50:54 - INFO - codeparrot_training - Step 48067: {'lr': 1.998074330489957e-06, 'samples': 24610816, 'steps': 48067, 'loss/train': 2.4676690101623535}
+02/26/2022 15:50:59 - INFO - codeparrot_training - Step 48068: {'lr': 1.9960102899884146e-06, 'samples': 24611328, 'steps': 48068, 'loss/train': 1.6439640522003174}
+02/26/2022 15:51:03 - INFO - codeparrot_training - Step 48069: {'lr': 1.9939473118571748e-06, 'samples': 24611840, 'steps': 48069, 'loss/train': 2.1548399925231934}
+02/26/2022 15:51:08 - INFO - codeparrot_training - Step 48070: {'lr': 1.9918853961051476e-06, 'samples': 24612352, 'steps': 48070, 'loss/train': 2.229962110519409}
+02/26/2022 15:51:12 - INFO - codeparrot_training - Step 48071: {'lr': 1.989824542741131e-06, 'samples': 24612864, 'steps': 48071, 'loss/train': 1.1238858699798584}
+02/26/2022 15:51:17 - INFO - codeparrot_training - Step 48072: {'lr': 1.9877647517739518e-06, 'samples': 24613376, 'steps': 48072, 'loss/train': 2.0451135635375977}
+02/26/2022 15:51:21 - INFO - codeparrot_training - Step 48073: {'lr': 1.985706023212408e-06, 'samples': 24613888, 'steps': 48073, 'loss/train': 0.42863568663597107}
+02/26/2022 15:51:28 - INFO - codeparrot_training - Step 48074: {'lr': 1.983648357065382e-06, 'samples': 24614400, 'steps': 48074, 'loss/train': 1.0816551446914673}
+02/26/2022 15:51:32 - INFO - codeparrot_training - Step 48075: {'lr': 1.981591753341644e-06, 'samples': 24614912, 'steps': 48075, 'loss/train': 2.251687526702881}
+02/26/2022 15:51:35 - INFO - codeparrot_training - Step 48076: {'lr': 1.9795362120499927e-06, 'samples': 24615424, 'steps': 48076, 'loss/train': 2.5053560733795166}
+02/26/2022 15:51:41 - INFO - codeparrot_training - Step 48077: {'lr': 1.9774817331992546e-06, 'samples': 24615936, 'steps': 48077, 'loss/train': 1.9417693614959717}
+02/26/2022 15:51:44 - INFO - codeparrot_training - Step 48078: {'lr': 1.9754283167982557e-06, 'samples': 24616448, 'steps': 48078, 'loss/train': 1.716273546218872}
+02/26/2022 15:51:50 - INFO - codeparrot_training - Step 48079: {'lr': 1.973375962855767e-06, 'samples': 24616960, 'steps': 48079, 'loss/train': 1.5255002975463867}
+02/26/2022 15:51:53 - INFO - codeparrot_training - Step 48080: {'lr': 1.9713246713805587e-06, 'samples': 24617472, 'steps': 48080, 'loss/train': 0.9694469571113586}
+02/26/2022 15:51:59 - INFO - codeparrot_training - Step 48081: {'lr': 1.96927444238143e-06, 'samples': 24617984, 'steps': 48081, 'loss/train': 1.755071997642517}
+02/26/2022 15:52:02 - INFO - codeparrot_training - Step 48082: {'lr': 1.9672252758671795e-06, 'samples': 24618496, 'steps': 48082, 'loss/train': 1.8848768472671509}
+02/26/2022 15:52:09 - INFO - codeparrot_training - Step 48083: {'lr': 1.9651771718465774e-06, 'samples': 24619008, 'steps': 48083, 'loss/train': 1.168839931488037}
+02/26/2022 15:52:13 - INFO - codeparrot_training - Step 48084: {'lr': 1.9631301303283944e-06, 'samples': 24619520, 'steps': 48084, 'loss/train': 1.6247429847717285}
+02/26/2022 15:52:18 - INFO - codeparrot_training - Step 48085: {'lr': 1.9610841513214015e-06, 'samples': 24620032, 'steps': 48085, 'loss/train': 2.532076835632324}
+02/26/2022 15:52:21 - INFO - codeparrot_training - Step 48086: {'lr': 1.95903923483437e-06, 'samples': 24620544, 'steps': 48086, 'loss/train': 1.5146517753601074}
+02/26/2022 15:52:27 - INFO - codeparrot_training - Step 48087: {'lr': 1.956995380876042e-06, 'samples': 24621056, 'steps': 48087, 'loss/train': 2.0462818145751953}
+02/26/2022 15:52:30 - INFO - codeparrot_training - Step 48088: {'lr': 1.954952589455189e-06, 'samples': 24621568, 'steps': 48088, 'loss/train': 0.31213557720184326}
+02/26/2022 15:52:36 - INFO - codeparrot_training - Step 48089: {'lr': 1.9529108605805534e-06, 'samples': 24622080, 'steps': 48089, 'loss/train': 0.6354958415031433}
+02/26/2022 15:52:39 - INFO - codeparrot_training - Step 48090: {'lr': 1.9508701942609064e-06, 'samples': 24622592, 'steps': 48090, 'loss/train': 1.7278589010238647}
+02/26/2022 15:52:45 - INFO - codeparrot_training - Step 48091: {'lr': 1.948830590504935e-06, 'samples': 24623104, 'steps': 48091, 'loss/train': 1.6710814237594604}
+02/26/2022 15:52:48 - INFO - codeparrot_training - Step 48092: {'lr': 1.946792049321411e-06, 'samples': 24623616, 'steps': 48092, 'loss/train': 1.0024529695510864}
+02/26/2022 15:52:54 - INFO - codeparrot_training - Step 48093: {'lr': 1.944754570719076e-06, 'samples': 24624128, 'steps': 48093, 'loss/train': 1.8819817304611206}
+02/26/2022 15:52:58 - INFO - codeparrot_training - Step 48094: {'lr': 1.942718154706646e-06, 'samples': 24624640, 'steps': 48094, 'loss/train': 2.11812424659729}
+02/26/2022 15:53:03 - INFO - codeparrot_training - Step 48095: {'lr': 1.9406828012928647e-06, 'samples': 24625152, 'steps': 48095, 'loss/train': 1.1509935855865479}
+02/26/2022 15:53:07 - INFO - codeparrot_training - Step 48096: {'lr': 1.9386485104863904e-06, 'samples': 24625664, 'steps': 48096, 'loss/train': 2.1935925483703613}
+02/26/2022 15:53:12 - INFO - codeparrot_training - Step 48097: {'lr': 1.9366152822959948e-06, 'samples': 24626176, 'steps': 48097, 'loss/train': 1.5797220468521118}
+02/26/2022 15:53:16 - INFO - codeparrot_training - Step 48098: {'lr': 1.934583116730365e-06, 'samples': 24626688, 'steps': 48098, 'loss/train': 2.4963557720184326}
+02/26/2022 15:53:21 - INFO - codeparrot_training - Step 48099: {'lr': 1.9325520137982166e-06, 'samples': 24627200, 'steps': 48099, 'loss/train': 0.6718520522117615}
+02/26/2022 15:53:27 - INFO - codeparrot_training - Step 48100: {'lr': 1.930521973508237e-06, 'samples': 24627712, 'steps': 48100, 'loss/train': 1.710747241973877}
+02/26/2022 15:53:30 - INFO - codeparrot_training - Step 48101: {'lr': 1.928492995869141e-06, 'samples': 24628224, 'steps': 48101, 'loss/train': 1.610666036605835}
+02/26/2022 15:53:36 - INFO - codeparrot_training - Step 48102: {'lr': 1.9264650808896167e-06, 'samples': 24628736, 'steps': 48102, 'loss/train': 1.7758033275604248}
+02/26/2022 15:53:39 - INFO - codeparrot_training - Step 48103: {'lr': 1.9244382285783515e-06, 'samples': 24629248, 'steps': 48103, 'loss/train': 0.4534592032432556}
+02/26/2022 15:53:43 - INFO - codeparrot_training - Step 48104: {'lr': 1.9224124389439767e-06, 'samples': 24629760, 'steps': 48104, 'loss/train': 1.828382134437561}
+02/26/2022 15:53:49 - INFO - codeparrot_training - Step 48105: {'lr': 1.920387711995236e-06, 'samples': 24630272, 'steps': 48105, 'loss/train': 1.3766778707504272}
+02/26/2022 15:53:53 - INFO - codeparrot_training - Step 48106: {'lr': 1.9183640477407894e-06, 'samples': 24630784, 'steps': 48106, 'loss/train': 0.21424119174480438}
+02/26/2022 15:53:58 - INFO - codeparrot_training - Step 48107: {'lr': 1.9163414461892957e-06, 'samples': 24631296, 'steps': 48107, 'loss/train': 0.5620800256729126}
+02/26/2022 15:54:02 - INFO - codeparrot_training - Step 48108: {'lr': 1.9143199073494154e-06, 'samples': 24631808, 'steps': 48108, 'loss/train': 2.1171562671661377}
+02/26/2022 15:54:07 - INFO - codeparrot_training - Step 48109: {'lr': 1.912299431229808e-06, 'samples': 24632320, 'steps': 48109, 'loss/train': 1.9083143472671509}
+02/26/2022 15:54:11 - INFO - codeparrot_training - Step 48110: {'lr': 1.9102800178391335e-06, 'samples': 24632832, 'steps': 48110, 'loss/train': 2.0866315364837646}
+02/26/2022 15:54:16 - INFO - codeparrot_training - Step 48111: {'lr': 1.9082616671860508e-06, 'samples': 24633344, 'steps': 48111, 'loss/train': 2.3169713020324707}
+02/26/2022 15:54:20 - INFO - codeparrot_training - Step 48112: {'lr': 1.9062443792791929e-06, 'samples': 24633856, 'steps': 48112, 'loss/train': 1.7829821109771729}
+02/26/2022 15:54:25 - INFO - codeparrot_training - Step 48113: {'lr': 1.9042281541272188e-06, 'samples': 24634368, 'steps': 48113, 'loss/train': 1.5835213661193848}
+02/26/2022 15:54:29 - INFO - codeparrot_training - Step 48114: {'lr': 1.9022129917387055e-06, 'samples': 24634880, 'steps': 48114, 'loss/train': 8.325510025024414}
+02/26/2022 15:54:35 - INFO - codeparrot_training - Step 48115: {'lr': 1.9001988921223957e-06, 'samples': 24635392, 'steps': 48115, 'loss/train': 2.0862507820129395}
+02/26/2022 15:54:38 - INFO - codeparrot_training - Step 48116: {'lr': 1.8981858552868104e-06, 'samples': 24635904, 'steps': 48116, 'loss/train': 1.3885626792907715}
+02/26/2022 15:54:44 - INFO - codeparrot_training - Step 48117: {'lr': 1.8961738812406372e-06, 'samples': 24636416, 'steps': 48117, 'loss/train': 1.9017176628112793}
+02/26/2022 15:54:49 - INFO - codeparrot_training - Step 48118: {'lr': 1.8941629699924522e-06, 'samples': 24636928, 'steps': 48118, 'loss/train': 1.3964951038360596}
+02/26/2022 15:54:53 - INFO - codeparrot_training - Step 48119: {'lr': 1.8921531215509157e-06, 'samples': 24637440, 'steps': 48119, 'loss/train': 0.062121689319610596}
+02/26/2022 15:54:58 - INFO - codeparrot_training - Step 48120: {'lr': 1.8901443359245762e-06, 'samples': 24637952, 'steps': 48120, 'loss/train': 1.5635122060775757}
+02/26/2022 15:55:02 - INFO - codeparrot_training - Step 48121: {'lr': 1.8881366131220934e-06, 'samples': 24638464, 'steps': 48121, 'loss/train': 2.294280529022217}
+02/26/2022 15:55:07 - INFO - codeparrot_training - Step 48122: {'lr': 1.8861299531520436e-06, 'samples': 24638976, 'steps': 48122, 'loss/train': 1.6215163469314575}
+02/26/2022 15:55:11 - INFO - codeparrot_training - Step 48123: {'lr': 1.8841243560230591e-06, 'samples': 24639488, 'steps': 48123, 'loss/train': 2.1305184364318848}
+02/26/2022 15:55:16 - INFO - codeparrot_training - Step 48124: {'lr': 1.8821198217436608e-06, 'samples': 24640000, 'steps': 48124, 'loss/train': 1.4907678365707397}
+02/26/2022 15:55:20 - INFO - codeparrot_training - Step 48125: {'lr': 1.8801163503225082e-06, 'samples': 24640512, 'steps': 48125, 'loss/train': 0.9205184578895569}
+02/26/2022 15:55:25 - INFO - codeparrot_training - Step 48126: {'lr': 1.8781139417681226e-06, 'samples': 24641024, 'steps': 48126, 'loss/train': 0.487720787525177}
+02/26/2022 15:55:29 - INFO - codeparrot_training - Step 48127: {'lr': 1.876112596089108e-06, 'samples': 24641536, 'steps': 48127, 'loss/train': 5.531866073608398}
+02/26/2022 15:55:34 - INFO - codeparrot_training - Step 48128: {'lr': 1.8741123132940685e-06, 'samples': 24642048, 'steps': 48128, 'loss/train': 1.314464807510376}
+02/26/2022 15:55:38 - INFO - codeparrot_training - Step 48129: {'lr': 1.8721130933915253e-06, 'samples': 24642560, 'steps': 48129, 'loss/train': 0.5421221256256104}
+02/26/2022 15:55:44 - INFO - codeparrot_training - Step 48130: {'lr': 1.870114936390055e-06, 'samples': 24643072, 'steps': 48130, 'loss/train': 1.758766531944275}
+02/26/2022 15:55:48 - INFO - codeparrot_training - Step 48131: {'lr': 1.8681178422982336e-06, 'samples': 24643584, 'steps': 48131, 'loss/train': 2.070037603378296}
+02/26/2022 15:55:53 - INFO - codeparrot_training - Step 48132: {'lr': 1.8661218111246102e-06, 'samples': 24644096, 'steps': 48132, 'loss/train': 1.9842796325683594}
+02/26/2022 15:55:57 - INFO - codeparrot_training - Step 48133: {'lr': 1.8641268428777057e-06, 'samples': 24644608, 'steps': 48133, 'loss/train': 1.1908373832702637}
+02/26/2022 15:56:03 - INFO - codeparrot_training - Step 48134: {'lr': 1.8621329375660968e-06, 'samples': 24645120, 'steps': 48134, 'loss/train': 1.1644662618637085}
+02/26/2022 15:56:06 - INFO - codeparrot_training - Step 48135: {'lr': 1.8601400951983316e-06, 'samples': 24645632, 'steps': 48135, 'loss/train': 2.710977077484131}
+02/26/2022 15:56:12 - INFO - codeparrot_training - Step 48136: {'lr': 1.8581483157829315e-06, 'samples': 24646144, 'steps': 48136, 'loss/train': 0.1924516260623932}
+02/26/2022 15:56:15 - INFO - codeparrot_training - Step 48137: {'lr': 1.8561575993284175e-06, 'samples': 24646656, 'steps': 48137, 'loss/train': 2.6672000885009766}
+02/26/2022 15:56:21 - INFO - codeparrot_training - Step 48138: {'lr': 1.854167945843338e-06, 'samples': 24647168, 'steps': 48138, 'loss/train': 1.7876639366149902}
+02/26/2022 15:56:24 - INFO - codeparrot_training - Step 48139: {'lr': 1.8521793553361865e-06, 'samples': 24647680, 'steps': 48139, 'loss/train': 2.317499876022339}
+02/26/2022 15:56:31 - INFO - codeparrot_training - Step 48140: {'lr': 1.8501918278155394e-06, 'samples': 24648192, 'steps': 48140, 'loss/train': 1.517192006111145}
+02/26/2022 15:56:34 - INFO - codeparrot_training - Step 48141: {'lr': 1.848205363289862e-06, 'samples': 24648704, 'steps': 48141, 'loss/train': 0.8325765132904053}
+02/26/2022 15:56:40 - INFO - codeparrot_training - Step 48142: {'lr': 1.8462199617676478e-06, 'samples': 24649216, 'steps': 48142, 'loss/train': 0.8042821884155273}
+02/26/2022 15:56:43 - INFO - codeparrot_training - Step 48143: {'lr': 1.8442356232574453e-06, 'samples': 24649728, 'steps': 48143, 'loss/train': 1.657462477684021}
+02/26/2022 15:56:48 - INFO - codeparrot_training - Step 48144: {'lr': 1.8422523477677477e-06, 'samples': 24650240, 'steps': 48144, 'loss/train': 0.9021700024604797}
+02/26/2022 15:56:52 - INFO - codeparrot_training - Step 48145: {'lr': 1.8402701353070483e-06, 'samples': 24650752, 'steps': 48145, 'loss/train': 1.8635834455490112}
+02/26/2022 15:56:58 - INFO - codeparrot_training - Step 48146: {'lr': 1.8382889858838403e-06, 'samples': 24651264, 'steps': 48146, 'loss/train': 2.2870888710021973}
+02/26/2022 15:57:01 - INFO - codeparrot_training - Step 48147: {'lr': 1.8363088995065614e-06, 'samples': 24651776, 'steps': 48147, 'loss/train': 2.040375232696533}
+02/26/2022 15:57:07 - INFO - codeparrot_training - Step 48148: {'lr': 1.8343298761837601e-06, 'samples': 24652288, 'steps': 48148, 'loss/train': 0.05891959369182587}
+02/26/2022 15:57:10 - INFO - codeparrot_training - Step 48149: {'lr': 1.8323519159239022e-06, 'samples': 24652800, 'steps': 48149, 'loss/train': 2.134617328643799}
+02/26/2022 15:57:16 - INFO - codeparrot_training - Step 48150: {'lr': 1.830375018735425e-06, 'samples': 24653312, 'steps': 48150, 'loss/train': 2.1146655082702637}
+02/26/2022 15:57:20 - INFO - codeparrot_training - Step 48151: {'lr': 1.8283991846268221e-06, 'samples': 24653824, 'steps': 48151, 'loss/train': 0.5072862505912781}
+02/26/2022 15:57:25 - INFO - codeparrot_training - Step 48152: {'lr': 1.826424413606559e-06, 'samples': 24654336, 'steps': 48152, 'loss/train': 1.6900368928909302}
+02/26/2022 15:57:29 - INFO - codeparrot_training - Step 48153: {'lr': 1.8244507056830729e-06, 'samples': 24654848, 'steps': 48153, 'loss/train': 2.5450944900512695}
+02/26/2022 15:57:34 - INFO - codeparrot_training - Step 48154: {'lr': 1.8224780608648572e-06, 'samples': 24655360, 'steps': 48154, 'loss/train': 1.4937857389450073}
+02/26/2022 15:57:38 - INFO - codeparrot_training - Step 48155: {'lr': 1.820506479160322e-06, 'samples': 24655872, 'steps': 48155, 'loss/train': 1.6144745349884033}
+02/26/2022 15:57:43 - INFO - codeparrot_training - Step 48156: {'lr': 1.8185359605779605e-06, 'samples': 24656384, 'steps': 48156, 'loss/train': 1.5675196647644043}
+02/26/2022 15:57:47 - INFO - codeparrot_training - Step 48157: {'lr': 1.8165665051261548e-06, 'samples': 24656896, 'steps': 48157, 'loss/train': 1.4855611324310303}
+02/26/2022 15:57:52 - INFO - codeparrot_training - Step 48158: {'lr': 1.8145981128133982e-06, 'samples': 24657408, 'steps': 48158, 'loss/train': 1.4053477048873901}
+02/26/2022 15:57:56 - INFO - codeparrot_training - Step 48159: {'lr': 1.812630783648045e-06, 'samples': 24657920, 'steps': 48159, 'loss/train': 1.4907281398773193}
+02/26/2022 15:58:02 - INFO - codeparrot_training - Step 48160: {'lr': 1.810664517638616e-06, 'samples': 24658432, 'steps': 48160, 'loss/train': 1.9743613004684448}
+02/26/2022 15:58:05 - INFO - codeparrot_training - Step 48161: {'lr': 1.8086993147934661e-06, 'samples': 24658944, 'steps': 48161, 'loss/train': 2.0596156120300293}
+02/26/2022 15:58:11 - INFO - codeparrot_training - Step 48162: {'lr': 1.8067351751210603e-06, 'samples': 24659456, 'steps': 48162, 'loss/train': 2.3782312870025635}
+02/26/2022 15:58:14 - INFO - codeparrot_training - Step 48163: {'lr': 1.8047720986298089e-06, 'samples': 24659968, 'steps': 48163, 'loss/train': 1.170705795288086}
+02/26/2022 15:58:20 - INFO - codeparrot_training - Step 48164: {'lr': 1.802810085328066e-06, 'samples': 24660480, 'steps': 48164, 'loss/train': 1.574622631072998}
+02/26/2022 15:58:23 - INFO - codeparrot_training - Step 48165: {'lr': 1.8008491352242696e-06, 'samples': 24660992, 'steps': 48165, 'loss/train': 2.717625379562378}
+02/26/2022 15:58:29 - INFO - codeparrot_training - Step 48166: {'lr': 1.798889248326857e-06, 'samples': 24661504, 'steps': 48166, 'loss/train': 1.588113784790039}
+02/26/2022 15:58:32 - INFO - codeparrot_training - Step 48167: {'lr': 1.796930424644183e-06, 'samples': 24662016, 'steps': 48167, 'loss/train': 2.857224225997925}
+02/26/2022 15:58:38 - INFO - codeparrot_training - Step 48168: {'lr': 1.7949726641846298e-06, 'samples': 24662528, 'steps': 48168, 'loss/train': 2.3585195541381836}
+02/26/2022 15:58:41 - INFO - codeparrot_training - Step 48169: {'lr': 1.7930159669566072e-06, 'samples': 24663040, 'steps': 48169, 'loss/train': 1.2491806745529175}
+02/26/2022 15:58:48 - INFO - codeparrot_training - Step 48170: {'lr': 1.7910603329684693e-06, 'samples': 24663552, 'steps': 48170, 'loss/train': 2.4087884426116943}
+02/26/2022 15:58:51 - INFO - codeparrot_training - Step 48171: {'lr': 1.7891057622286543e-06, 'samples': 24664064, 'steps': 48171, 'loss/train': 1.4979145526885986}
+02/26/2022 15:58:57 - INFO - codeparrot_training - Step 48172: {'lr': 1.787152254745461e-06, 'samples': 24664576, 'steps': 48172, 'loss/train': 1.2991231679916382}
+02/26/2022 15:59:00 - INFO - codeparrot_training - Step 48173: {'lr': 1.7851998105272992e-06, 'samples': 24665088, 'steps': 48173, 'loss/train': 0.7050226926803589}
+02/26/2022 15:59:06 - INFO - codeparrot_training - Step 48174: {'lr': 1.7832484295825236e-06, 'samples': 24665600, 'steps': 48174, 'loss/train': 1.8083151578903198}
+02/26/2022 15:59:09 - INFO - codeparrot_training - Step 48175: {'lr': 1.781298111919516e-06, 'samples': 24666112, 'steps': 48175, 'loss/train': 1.4875048398971558}
+02/26/2022 15:59:15 - INFO - codeparrot_training - Step 48176: {'lr': 1.7793488575466032e-06, 'samples': 24666624, 'steps': 48176, 'loss/train': 2.6878955364227295}
+02/26/2022 15:59:18 - INFO - codeparrot_training - Step 48177: {'lr': 1.777400666472112e-06, 'samples': 24667136, 'steps': 48177, 'loss/train': 2.1547327041625977}
+02/26/2022 15:59:24 - INFO - codeparrot_training - Step 48178: {'lr': 1.7754535387044246e-06, 'samples': 24667648, 'steps': 48178, 'loss/train': 1.327903389930725}
+02/26/2022 15:59:28 - INFO - codeparrot_training - Step 48179: {'lr': 1.7735074742518954e-06, 'samples': 24668160, 'steps': 48179, 'loss/train': 2.301586389541626}
+02/26/2022 15:59:31 - INFO - codeparrot_training - Step 48180: {'lr': 1.771562473122823e-06, 'samples': 24668672, 'steps': 48180, 'loss/train': 1.8878509998321533}
+02/26/2022 15:59:37 - INFO - codeparrot_training - Step 48181: {'lr': 1.7696185353255624e-06, 'samples': 24669184, 'steps': 48181, 'loss/train': 0.4387229084968567}
+02/26/2022 15:59:40 - INFO - codeparrot_training - Step 48182: {'lr': 1.7676756608684396e-06, 'samples': 24669696, 'steps': 48182, 'loss/train': 1.2409852743148804}
+02/26/2022 15:59:46 - INFO - codeparrot_training - Step 48183: {'lr': 1.7657338497597542e-06, 'samples': 24670208, 'steps': 48183, 'loss/train': 1.322680950164795}
+02/26/2022 15:59:49 - INFO - codeparrot_training - Step 48184: {'lr': 1.7637931020078602e-06, 'samples': 24670720, 'steps': 48184, 'loss/train': 1.5506795644760132}
+02/26/2022 15:59:55 - INFO - codeparrot_training - Step 48185: {'lr': 1.7618534176210288e-06, 'samples': 24671232, 'steps': 48185, 'loss/train': 1.3698604106903076}
+02/26/2022 16:00:01 - INFO - codeparrot_training - Step 48186: {'lr': 1.7599147966076146e-06, 'samples': 24671744, 'steps': 48186, 'loss/train': 1.5163310766220093}
+02/26/2022 16:00:04 - INFO - codeparrot_training - Step 48187: {'lr': 1.7579772389758886e-06, 'samples': 24672256, 'steps': 48187, 'loss/train': 1.2335726022720337}
+02/26/2022 16:00:10 - INFO - codeparrot_training - Step 48188: {'lr': 1.7560407447341497e-06, 'samples': 24672768, 'steps': 48188, 'loss/train': 1.682826280593872}
+02/26/2022 16:00:13 - INFO - codeparrot_training - Step 48189: {'lr': 1.7541053138907525e-06, 'samples': 24673280, 'steps': 48189, 'loss/train': 1.3601185083389282}
+02/26/2022 16:00:19 - INFO - codeparrot_training - Step 48190: {'lr': 1.7521709464539127e-06, 'samples': 24673792, 'steps': 48190, 'loss/train': 1.3802342414855957}
+02/26/2022 16:00:22 - INFO - codeparrot_training - Step 48191: {'lr': 1.750237642431929e-06, 'samples': 24674304, 'steps': 48191, 'loss/train': 0.38658061623573303}
+02/26/2022 16:00:28 - INFO - codeparrot_training - Step 48192: {'lr': 1.7483054018331002e-06, 'samples': 24674816, 'steps': 48192, 'loss/train': 1.097922921180725}
+02/26/2022 16:00:32 - INFO - codeparrot_training - Step 48193: {'lr': 1.7463742246657533e-06, 'samples': 24675328, 'steps': 48193, 'loss/train': 1.7961775064468384}
+02/26/2022 16:00:35 - INFO - codeparrot_training - Step 48194: {'lr': 1.7444441109380482e-06, 'samples': 24675840, 'steps': 48194, 'loss/train': 1.5271315574645996}
+02/26/2022 16:00:41 - INFO - codeparrot_training - Step 48195: {'lr': 1.7425150606583395e-06, 'samples': 24676352, 'steps': 48195, 'loss/train': 1.4275624752044678}
+02/26/2022 16:00:44 - INFO - codeparrot_training - Step 48196: {'lr': 1.7405870738348705e-06, 'samples': 24676864, 'steps': 48196, 'loss/train': 1.9668711423873901}
+02/26/2022 16:00:50 - INFO - codeparrot_training - Step 48197: {'lr': 1.7386601504758848e-06, 'samples': 24677376, 'steps': 48197, 'loss/train': 1.1421281099319458}
+02/26/2022 16:00:54 - INFO - codeparrot_training - Step 48198: {'lr': 1.7367342905896532e-06, 'samples': 24677888, 'steps': 48198, 'loss/train': 1.576008677482605}
+02/26/2022 16:00:59 - INFO - codeparrot_training - Step 48199: {'lr': 1.7348094941844194e-06, 'samples': 24678400, 'steps': 48199, 'loss/train': 1.7285265922546387}
+02/26/2022 16:01:03 - INFO - codeparrot_training - Step 48200: {'lr': 1.7328857612684267e-06, 'samples': 24678912, 'steps': 48200, 'loss/train': 1.9396990537643433}
+02/26/2022 16:01:08 - INFO - codeparrot_training - Step 48201: {'lr': 1.7309630918499187e-06, 'samples': 24679424, 'steps': 48201, 'loss/train': 1.681591272354126}
+02/26/2022 16:01:12 - INFO - codeparrot_training - Step 48202: {'lr': 1.7290414859371383e-06, 'samples': 24679936, 'steps': 48202, 'loss/train': 1.8649177551269531}
+02/26/2022 16:01:17 - INFO - codeparrot_training - Step 48203: {'lr': 1.7271209435383017e-06, 'samples': 24680448, 'steps': 48203, 'loss/train': 2.9391820430755615}
+02/26/2022 16:01:23 - INFO - codeparrot_training - Step 48204: {'lr': 1.7252014646616242e-06, 'samples': 24680960, 'steps': 48204, 'loss/train': 1.167412281036377}
+02/26/2022 16:01:26 - INFO - codeparrot_training - Step 48205: {'lr': 1.7232830493153774e-06, 'samples': 24681472, 'steps': 48205, 'loss/train': 2.2361602783203125}
+02/26/2022 16:01:32 - INFO - codeparrot_training - Step 48206: {'lr': 1.7213656975077485e-06, 'samples': 24681984, 'steps': 48206, 'loss/train': 0.8090372681617737}
+02/26/2022 16:01:36 - INFO - codeparrot_training - Step 48207: {'lr': 1.719449409246926e-06, 'samples': 24682496, 'steps': 48207, 'loss/train': 2.1550731658935547}
+02/26/2022 16:01:41 - INFO - codeparrot_training - Step 48208: {'lr': 1.717534184541153e-06, 'samples': 24683008, 'steps': 48208, 'loss/train': 2.4501538276672363}
+02/26/2022 16:01:45 - INFO - codeparrot_training - Step 48209: {'lr': 1.7156200233986453e-06, 'samples': 24683520, 'steps': 48209, 'loss/train': 1.6562236547470093}
+02/26/2022 16:01:50 - INFO - codeparrot_training - Step 48210: {'lr': 1.7137069258275629e-06, 'samples': 24684032, 'steps': 48210, 'loss/train': 2.0163137912750244}
+02/26/2022 16:01:54 - INFO - codeparrot_training - Step 48211: {'lr': 1.7117948918361493e-06, 'samples': 24684544, 'steps': 48211, 'loss/train': 0.8923929929733276}
+02/26/2022 16:01:59 - INFO - codeparrot_training - Step 48212: {'lr': 1.709883921432509e-06, 'samples': 24685056, 'steps': 48212, 'loss/train': 0.3566322326660156}
+02/26/2022 16:02:03 - INFO - codeparrot_training - Step 48213: {'lr': 1.707974014624941e-06, 'samples': 24685568, 'steps': 48213, 'loss/train': 1.1037888526916504}
+02/26/2022 16:02:08 - INFO - codeparrot_training - Step 48214: {'lr': 1.7060651714215503e-06, 'samples': 24686080, 'steps': 48214, 'loss/train': 2.328897476196289}
+02/26/2022 16:02:12 - INFO - codeparrot_training - Step 48215: {'lr': 1.7041573918305241e-06, 'samples': 24686592, 'steps': 48215, 'loss/train': 2.175065755844116}
+02/26/2022 16:02:17 - INFO - codeparrot_training - Step 48216: {'lr': 1.7022506758600509e-06, 'samples': 24687104, 'steps': 48216, 'loss/train': 1.1348466873168945}
+02/26/2022 16:02:21 - INFO - codeparrot_training - Step 48217: {'lr': 1.7003450235182904e-06, 'samples': 24687616, 'steps': 48217, 'loss/train': 1.7976552248001099}
+02/26/2022 16:02:27 - INFO - codeparrot_training - Step 48218: {'lr': 1.6984404348134308e-06, 'samples': 24688128, 'steps': 48218, 'loss/train': 1.7065542936325073}
+02/26/2022 16:02:30 - INFO - codeparrot_training - Step 48219: {'lr': 1.6965369097535766e-06, 'samples': 24688640, 'steps': 48219, 'loss/train': 1.7317028045654297}
+02/26/2022 16:02:36 - INFO - codeparrot_training - Step 48220: {'lr': 1.6946344483469155e-06, 'samples': 24689152, 'steps': 48220, 'loss/train': 1.4688822031021118}
+02/26/2022 16:02:39 - INFO - codeparrot_training - Step 48221: {'lr': 1.692733050601608e-06, 'samples': 24689664, 'steps': 48221, 'loss/train': 1.8245819807052612}
+02/26/2022 16:02:46 - INFO - codeparrot_training - Step 48222: {'lr': 1.6908327165257865e-06, 'samples': 24690176, 'steps': 48222, 'loss/train': 1.0730267763137817}
+02/26/2022 16:02:50 - INFO - codeparrot_training - Step 48223: {'lr': 1.6889334461275829e-06, 'samples': 24690688, 'steps': 48223, 'loss/train': 1.7888169288635254}
+02/26/2022 16:02:53 - INFO - codeparrot_training - Step 48224: {'lr': 1.6870352394151579e-06, 'samples': 24691200, 'steps': 48224, 'loss/train': 0.5856693387031555}
+02/26/2022 16:02:59 - INFO - codeparrot_training - Step 48225: {'lr': 1.6851380963966156e-06, 'samples': 24691712, 'steps': 48225, 'loss/train': 1.3649907112121582}
+02/26/2022 16:03:02 - INFO - codeparrot_training - Step 48226: {'lr': 1.6832420170800888e-06, 'samples': 24692224, 'steps': 48226, 'loss/train': 1.9108003377914429}
+02/26/2022 16:03:08 - INFO - codeparrot_training - Step 48227: {'lr': 1.681347001473682e-06, 'samples': 24692736, 'steps': 48227, 'loss/train': 2.4020004272460938}
+02/26/2022 16:03:11 - INFO - codeparrot_training - Step 48228: {'lr': 1.6794530495855831e-06, 'samples': 24693248, 'steps': 48228, 'loss/train': 0.6762791872024536}
+02/26/2022 16:03:17 - INFO - codeparrot_training - Step 48229: {'lr': 1.6775601614238134e-06, 'samples': 24693760, 'steps': 48229, 'loss/train': 2.834538459777832}
+02/26/2022 16:03:20 - INFO - codeparrot_training - Step 48230: {'lr': 1.675668336996533e-06, 'samples': 24694272, 'steps': 48230, 'loss/train': 2.110729217529297}
+02/26/2022 16:03:26 - INFO - codeparrot_training - Step 48231: {'lr': 1.6737775763118468e-06, 'samples': 24694784, 'steps': 48231, 'loss/train': 1.7509702444076538}
+02/26/2022 16:03:32 - INFO - codeparrot_training - Step 48232: {'lr': 1.671887879377859e-06, 'samples': 24695296, 'steps': 48232, 'loss/train': 0.037955548614263535}
+02/26/2022 16:03:36 - INFO - codeparrot_training - Step 48233: {'lr': 1.6699992462026193e-06, 'samples': 24695808, 'steps': 48233, 'loss/train': 0.7673493027687073}
+02/26/2022 16:03:41 - INFO - codeparrot_training - Step 48234: {'lr': 1.6681116767942871e-06, 'samples': 24696320, 'steps': 48234, 'loss/train': 1.8250377178192139}
+02/26/2022 16:03:44 - INFO - codeparrot_training - Step 48235: {'lr': 1.6662251711608846e-06, 'samples': 24696832, 'steps': 48235, 'loss/train': 2.45802640914917}
+02/26/2022 16:03:50 - INFO - codeparrot_training - Step 48236: {'lr': 1.6643397293105156e-06, 'samples': 24697344, 'steps': 48236, 'loss/train': 1.9320323467254639}
+02/26/2022 16:03:53 - INFO - codeparrot_training - Step 48237: {'lr': 1.6624553512512574e-06, 'samples': 24697856, 'steps': 48237, 'loss/train': 1.4622244834899902}
+02/26/2022 16:03:59 - INFO - codeparrot_training - Step 48238: {'lr': 1.6605720369912147e-06, 'samples': 24698368, 'steps': 48238, 'loss/train': 2.220015287399292}
+02/26/2022 16:04:02 - INFO - codeparrot_training - Step 48239: {'lr': 1.658689786538381e-06, 'samples': 24698880, 'steps': 48239, 'loss/train': 0.7419174909591675}
+02/26/2022 16:04:08 - INFO - codeparrot_training - Step 48240: {'lr': 1.6568085999008887e-06, 'samples': 24699392, 'steps': 48240, 'loss/train': 2.6873536109924316}
+02/26/2022 16:04:11 - INFO - codeparrot_training - Step 48241: {'lr': 1.6549284770867867e-06, 'samples': 24699904, 'steps': 48241, 'loss/train': 1.902570128440857}
+02/26/2022 16:04:18 - INFO - codeparrot_training - Step 48242: {'lr': 1.6530494181040968e-06, 'samples': 24700416, 'steps': 48242, 'loss/train': 2.783463954925537}
+02/26/2022 16:04:21 - INFO - codeparrot_training - Step 48243: {'lr': 1.6511714229608677e-06, 'samples': 24700928, 'steps': 48243, 'loss/train': 1.3428328037261963}
+02/26/2022 16:04:27 - INFO - codeparrot_training - Step 48244: {'lr': 1.6492944916651765e-06, 'samples': 24701440, 'steps': 48244, 'loss/train': 1.5959863662719727}
+02/26/2022 16:04:30 - INFO - codeparrot_training - Step 48245: {'lr': 1.6474186242250445e-06, 'samples': 24701952, 'steps': 48245, 'loss/train': 1.367204189300537}
+02/26/2022 16:04:36 - INFO - codeparrot_training - Step 48246: {'lr': 1.6455438206485207e-06, 'samples': 24702464, 'steps': 48246, 'loss/train': 2.2631332874298096}
+02/26/2022 16:04:39 - INFO - codeparrot_training - Step 48247: {'lr': 1.6436700809435989e-06, 'samples': 24702976, 'steps': 48247, 'loss/train': 1.3153676986694336}
+02/26/2022 16:04:45 - INFO - codeparrot_training - Step 48248: {'lr': 1.641797405118356e-06, 'samples': 24703488, 'steps': 48248, 'loss/train': 1.4450786113739014}
+02/26/2022 16:04:48 - INFO - codeparrot_training - Step 48249: {'lr': 1.6399257931807855e-06, 'samples': 24704000, 'steps': 48249, 'loss/train': 1.1745692491531372}
+02/26/2022 16:04:54 - INFO - codeparrot_training - Step 48250: {'lr': 1.6380552451389085e-06, 'samples': 24704512, 'steps': 48250, 'loss/train': 2.138521671295166}
+02/26/2022 16:04:57 - INFO - codeparrot_training - Step 48251: {'lr': 1.6361857610007191e-06, 'samples': 24705024, 'steps': 48251, 'loss/train': 1.839398980140686}
+02/26/2022 16:05:03 - INFO - codeparrot_training - Step 48252: {'lr': 1.6343173407742662e-06, 'samples': 24705536, 'steps': 48252, 'loss/train': 2.284489393234253}
+02/26/2022 16:05:07 - INFO - codeparrot_training - Step 48253: {'lr': 1.6324499844675434e-06, 'samples': 24706048, 'steps': 48253, 'loss/train': 1.533995509147644}
+02/26/2022 16:05:12 - INFO - codeparrot_training - Step 48254: {'lr': 1.6305836920885442e-06, 'samples': 24706560, 'steps': 48254, 'loss/train': 0.5098047852516174}
+02/26/2022 16:05:16 - INFO - codeparrot_training - Step 48255: {'lr': 1.6287184636452345e-06, 'samples': 24707072, 'steps': 48255, 'loss/train': 2.63771915435791}
+02/26/2022 16:05:21 - INFO - codeparrot_training - Step 48256: {'lr': 1.6268542991456358e-06, 'samples': 24707584, 'steps': 48256, 'loss/train': 1.6032116413116455}
+02/26/2022 16:05:25 - INFO - codeparrot_training - Step 48257: {'lr': 1.624991198597714e-06, 'samples': 24708096, 'steps': 48257, 'loss/train': 1.1501610279083252}
+02/26/2022 16:05:30 - INFO - codeparrot_training - Step 48258: {'lr': 1.62312916200949e-06, 'samples': 24708608, 'steps': 48258, 'loss/train': 2.0278096199035645}
+02/26/2022 16:05:34 - INFO - codeparrot_training - Step 48259: {'lr': 1.6212681893888747e-06, 'samples': 24709120, 'steps': 48259, 'loss/train': 1.8049098253250122}
+02/26/2022 16:05:39 - INFO - codeparrot_training - Step 48260: {'lr': 1.619408280743917e-06, 'samples': 24709632, 'steps': 48260, 'loss/train': 0.5946051478385925}
+02/26/2022 16:05:42 - INFO - codeparrot_training - Step 48261: {'lr': 1.6175494360825272e-06, 'samples': 24710144, 'steps': 48261, 'loss/train': 1.7702735662460327}
+02/26/2022 16:05:48 - INFO - codeparrot_training - Step 48262: {'lr': 1.615691655412671e-06, 'samples': 24710656, 'steps': 48262, 'loss/train': 1.2659212350845337}
+02/26/2022 16:05:51 - INFO - codeparrot_training - Step 48263: {'lr': 1.6138349387423422e-06, 'samples': 24711168, 'steps': 48263, 'loss/train': 5.6597771644592285}
+02/26/2022 16:05:57 - INFO - codeparrot_training - Step 48264: {'lr': 1.6119792860794513e-06, 'samples': 24711680, 'steps': 48264, 'loss/train': 1.6914167404174805}
+02/26/2022 16:06:00 - INFO - codeparrot_training - Step 48265: {'lr': 1.6101246974319916e-06, 'samples': 24712192, 'steps': 48265, 'loss/train': 1.9811280965805054}
+02/26/2022 16:06:07 - INFO - codeparrot_training - Step 48266: {'lr': 1.6082711728078735e-06, 'samples': 24712704, 'steps': 48266, 'loss/train': 2.077152729034424}
+02/26/2022 16:06:11 - INFO - codeparrot_training - Step 48267: {'lr': 1.6064187122150353e-06, 'samples': 24713216, 'steps': 48267, 'loss/train': 1.2700115442276}
+02/26/2022 16:06:16 - INFO - codeparrot_training - Step 48268: {'lr': 1.6045673156614427e-06, 'samples': 24713728, 'steps': 48268, 'loss/train': 1.4722819328308105}
+02/26/2022 16:06:20 - INFO - codeparrot_training - Step 48269: {'lr': 1.602716983155006e-06, 'samples': 24714240, 'steps': 48269, 'loss/train': 1.5597320795059204}
+02/26/2022 16:06:25 - INFO - codeparrot_training - Step 48270: {'lr': 1.6008677147036633e-06, 'samples': 24714752, 'steps': 48270, 'loss/train': 2.409360885620117}
+02/26/2022 16:06:29 - INFO - codeparrot_training - Step 48271: {'lr': 1.5990195103153249e-06, 'samples': 24715264, 'steps': 48271, 'loss/train': 0.4325543940067291}
+02/26/2022 16:06:34 - INFO - codeparrot_training - Step 48272: {'lr': 1.5971723699979013e-06, 'samples': 24715776, 'steps': 48272, 'loss/train': 2.0530693531036377}
+02/26/2022 16:06:38 - INFO - codeparrot_training - Step 48273: {'lr': 1.5953262937593305e-06, 'samples': 24716288, 'steps': 48273, 'loss/train': 1.0317387580871582}
+02/26/2022 16:06:43 - INFO - codeparrot_training - Step 48274: {'lr': 1.5934812816074673e-06, 'samples': 24716800, 'steps': 48274, 'loss/train': 1.731128215789795}
+02/26/2022 16:06:47 - INFO - codeparrot_training - Step 48275: {'lr': 1.5916373335503054e-06, 'samples': 24717312, 'steps': 48275, 'loss/train': 1.4390676021575928}
+02/26/2022 16:06:52 - INFO - codeparrot_training - Step 48276: {'lr': 1.589794449595644e-06, 'samples': 24717824, 'steps': 48276, 'loss/train': 0.07702524960041046}
+02/26/2022 16:06:56 - INFO - codeparrot_training - Step 48277: {'lr': 1.587952629751449e-06, 'samples': 24718336, 'steps': 48277, 'loss/train': 1.5532214641571045}
+02/26/2022 16:07:02 - INFO - codeparrot_training - Step 48278: {'lr': 1.5861118740255753e-06, 'samples': 24718848, 'steps': 48278, 'loss/train': 1.0452167987823486}
+02/26/2022 16:07:06 - INFO - codeparrot_training - Step 48279: {'lr': 1.5842721824259332e-06, 'samples': 24719360, 'steps': 48279, 'loss/train': 1.191022276878357}
+02/26/2022 16:07:11 - INFO - codeparrot_training - Step 48280: {'lr': 1.5824335549603774e-06, 'samples': 24719872, 'steps': 48280, 'loss/train': 1.511879324913025}
+02/26/2022 16:07:15 - INFO - codeparrot_training - Step 48281: {'lr': 1.5805959916367906e-06, 'samples': 24720384, 'steps': 48281, 'loss/train': 1.860042929649353}
+02/26/2022 16:07:20 - INFO - codeparrot_training - Step 48282: {'lr': 1.5787594924630556e-06, 'samples': 24720896, 'steps': 48282, 'loss/train': 2.0549185276031494}
+02/26/2022 16:07:24 - INFO - codeparrot_training - Step 48283: {'lr': 1.5769240574470267e-06, 'samples': 24721408, 'steps': 48283, 'loss/train': 1.8378620147705078}
+02/26/2022 16:07:30 - INFO - codeparrot_training - Step 48284: {'lr': 1.5750896865965592e-06, 'samples': 24721920, 'steps': 48284, 'loss/train': 3.098766565322876}
+02/26/2022 16:07:33 - INFO - codeparrot_training - Step 48285: {'lr': 1.5732563799195632e-06, 'samples': 24722432, 'steps': 48285, 'loss/train': 0.4641614556312561}
+02/26/2022 16:07:39 - INFO - codeparrot_training - Step 48286: {'lr': 1.5714241374238103e-06, 'samples': 24722944, 'steps': 48286, 'loss/train': 2.4651243686676025}
+02/26/2022 16:07:42 - INFO - codeparrot_training - Step 48287: {'lr': 1.5695929591172109e-06, 'samples': 24723456, 'steps': 48287, 'loss/train': 1.2019466161727905}
+02/26/2022 16:07:47 - INFO - codeparrot_training - Step 48288: {'lr': 1.567762845007592e-06, 'samples': 24723968, 'steps': 48288, 'loss/train': 1.4710183143615723}
+02/26/2022 16:07:51 - INFO - codeparrot_training - Step 48289: {'lr': 1.5659337951028086e-06, 'samples': 24724480, 'steps': 48289, 'loss/train': 0.6994738578796387}
+02/26/2022 16:07:58 - INFO - codeparrot_training - Step 48290: {'lr': 1.564105809410632e-06, 'samples': 24724992, 'steps': 48290, 'loss/train': 1.680452823638916}
+02/26/2022 16:08:01 - INFO - codeparrot_training - Step 48291: {'lr': 1.5622788879389727e-06, 'samples': 24725504, 'steps': 48291, 'loss/train': 0.2028939127922058}
+02/26/2022 16:08:07 - INFO - codeparrot_training - Step 48292: {'lr': 1.560453030695602e-06, 'samples': 24726016, 'steps': 48292, 'loss/train': 2.0981016159057617}
+02/26/2022 16:08:10 - INFO - codeparrot_training - Step 48293: {'lr': 1.5586282376884032e-06, 'samples': 24726528, 'steps': 48293, 'loss/train': 0.9162625074386597}
+02/26/2022 16:08:16 - INFO - codeparrot_training - Step 48294: {'lr': 1.5568045089250916e-06, 'samples': 24727040, 'steps': 48294, 'loss/train': 1.4696649312973022}
+02/26/2022 16:08:19 - INFO - codeparrot_training - Step 48295: {'lr': 1.5549818444135777e-06, 'samples': 24727552, 'steps': 48295, 'loss/train': 1.9322148561477661}
+02/26/2022 16:08:25 - INFO - codeparrot_training - Step 48296: {'lr': 1.5531602441616332e-06, 'samples': 24728064, 'steps': 48296, 'loss/train': 2.042065382003784}
+02/26/2022 16:08:31 - INFO - codeparrot_training - Step 48297: {'lr': 1.5513397081770298e-06, 'samples': 24728576, 'steps': 48297, 'loss/train': 2.0719542503356934}
+02/26/2022 16:08:35 - INFO - codeparrot_training - Step 48298: {'lr': 1.549520236467622e-06, 'samples': 24729088, 'steps': 48298, 'loss/train': 1.8613502979278564}
+02/26/2022 16:08:40 - INFO - codeparrot_training - Step 48299: {'lr': 1.5477018290411815e-06, 'samples': 24729600, 'steps': 48299, 'loss/train': 1.3619030714035034}
+02/26/2022 16:08:44 - INFO - codeparrot_training - Step 48300: {'lr': 1.54588448590548e-06, 'samples': 24730112, 'steps': 48300, 'loss/train': 1.9936413764953613}
+02/26/2022 16:08:49 - INFO - codeparrot_training - Step 48301: {'lr': 1.5440682070683165e-06, 'samples': 24730624, 'steps': 48301, 'loss/train': 1.289435863494873}
+02/26/2022 16:08:53 - INFO - codeparrot_training - Step 48302: {'lr': 1.5422529925374907e-06, 'samples': 24731136, 'steps': 48302, 'loss/train': 0.029236581176519394}
+02/26/2022 16:08:58 - INFO - codeparrot_training - Step 48303: {'lr': 1.540438842320746e-06, 'samples': 24731648, 'steps': 48303, 'loss/train': 0.9637269377708435}
+02/26/2022 16:09:02 - INFO - codeparrot_training - Step 48304: {'lr': 1.5386257564258543e-06, 'samples': 24732160, 'steps': 48304, 'loss/train': 1.585838794708252}
+02/26/2022 16:09:07 - INFO - codeparrot_training - Step 48305: {'lr': 1.5368137348606148e-06, 'samples': 24732672, 'steps': 48305, 'loss/train': 2.0499584674835205}
+02/26/2022 16:09:11 - INFO - codeparrot_training - Step 48306: {'lr': 1.5350027776327435e-06, 'samples': 24733184, 'steps': 48306, 'loss/train': 1.1607283353805542}
+02/26/2022 16:09:16 - INFO - codeparrot_training - Step 48307: {'lr': 1.5331928847500399e-06, 'samples': 24733696, 'steps': 48307, 'loss/train': 1.7936731576919556}
+02/26/2022 16:09:20 - INFO - codeparrot_training - Step 48308: {'lr': 1.5313840562202475e-06, 'samples': 24734208, 'steps': 48308, 'loss/train': 1.611100673675537}
+02/26/2022 16:09:25 - INFO - codeparrot_training - Step 48309: {'lr': 1.5295762920511103e-06, 'samples': 24734720, 'steps': 48309, 'loss/train': 2.516822576522827}
+02/26/2022 16:09:29 - INFO - codeparrot_training - Step 48310: {'lr': 1.5277695922503442e-06, 'samples': 24735232, 'steps': 48310, 'loss/train': 1.8794735670089722}
+02/26/2022 16:09:34 - INFO - codeparrot_training - Step 48311: {'lr': 1.525963956825749e-06, 'samples': 24735744, 'steps': 48311, 'loss/train': 1.3278895616531372}
+02/26/2022 16:09:38 - INFO - codeparrot_training - Step 48312: {'lr': 1.5241593857850122e-06, 'samples': 24736256, 'steps': 48312, 'loss/train': 1.3245714902877808}
+02/26/2022 16:09:44 - INFO - codeparrot_training - Step 48313: {'lr': 1.5223558791358505e-06, 'samples': 24736768, 'steps': 48313, 'loss/train': 0.967051088809967}
+02/26/2022 16:09:48 - INFO - codeparrot_training - Step 48314: {'lr': 1.5205534368860352e-06, 'samples': 24737280, 'steps': 48314, 'loss/train': 1.2493215799331665}
+02/26/2022 16:09:53 - INFO - codeparrot_training - Step 48315: {'lr': 1.5187520590432823e-06, 'samples': 24737792, 'steps': 48315, 'loss/train': 1.688737392425537}
+02/26/2022 16:09:56 - INFO - codeparrot_training - Step 48316: {'lr': 1.5169517456152803e-06, 'samples': 24738304, 'steps': 48316, 'loss/train': 2.0536041259765625}
+02/26/2022 16:10:02 - INFO - codeparrot_training - Step 48317: {'lr': 1.515152496609745e-06, 'samples': 24738816, 'steps': 48317, 'loss/train': 1.2988651990890503}
+02/26/2022 16:10:05 - INFO - codeparrot_training - Step 48318: {'lr': 1.5133543120344207e-06, 'samples': 24739328, 'steps': 48318, 'loss/train': 1.5435714721679688}
+02/26/2022 16:10:11 - INFO - codeparrot_training - Step 48319: {'lr': 1.5115571918969396e-06, 'samples': 24739840, 'steps': 48319, 'loss/train': 1.995025396347046}
+02/26/2022 16:10:14 - INFO - codeparrot_training - Step 48320: {'lr': 1.5097611362051012e-06, 'samples': 24740352, 'steps': 48320, 'loss/train': 0.7486575841903687}
+02/26/2022 16:10:20 - INFO - codeparrot_training - Step 48321: {'lr': 1.5079661449664828e-06, 'samples': 24740864, 'steps': 48321, 'loss/train': 1.749605417251587}
+02/26/2022 16:10:23 - INFO - codeparrot_training - Step 48322: {'lr': 1.506172218188856e-06, 'samples': 24741376, 'steps': 48322, 'loss/train': 1.3769954442977905}
+02/26/2022 16:10:30 - INFO - codeparrot_training - Step 48323: {'lr': 1.5043793558798812e-06, 'samples': 24741888, 'steps': 48323, 'loss/train': 1.4700435400009155}
+02/26/2022 16:10:33 - INFO - codeparrot_training - Step 48324: {'lr': 1.5025875580472748e-06, 'samples': 24742400, 'steps': 48324, 'loss/train': 2.265261650085449}
+02/26/2022 16:10:39 - INFO - codeparrot_training - Step 48325: {'lr': 1.5007968246986137e-06, 'samples': 24742912, 'steps': 48325, 'loss/train': 3.0830962657928467}
+02/26/2022 16:10:42 - INFO - codeparrot_training - Step 48326: {'lr': 1.4990071558416695e-06, 'samples': 24743424, 'steps': 48326, 'loss/train': 1.3435145616531372}
+02/26/2022 16:10:48 - INFO - codeparrot_training - Step 48327: {'lr': 1.4972185514840752e-06, 'samples': 24743936, 'steps': 48327, 'loss/train': 0.5534811019897461}
+02/26/2022 16:10:51 - INFO - codeparrot_training - Step 48328: {'lr': 1.4954310116334913e-06, 'samples': 24744448, 'steps': 48328, 'loss/train': 1.7710227966308594}
+02/26/2022 16:10:57 - INFO - codeparrot_training - Step 48329: {'lr': 1.4936445362975504e-06, 'samples': 24744960, 'steps': 48329, 'loss/train': 1.5421433448791504}
+02/26/2022 16:11:00 - INFO - codeparrot_training - Step 48330: {'lr': 1.491859125483913e-06, 'samples': 24745472, 'steps': 48330, 'loss/train': 0.7247592210769653}
+02/26/2022 16:11:06 - INFO - codeparrot_training - Step 48331: {'lr': 1.4900747792002678e-06, 'samples': 24745984, 'steps': 48331, 'loss/train': 0.9581589102745056}
+02/26/2022 16:11:09 - INFO - codeparrot_training - Step 48332: {'lr': 1.4882914974542195e-06, 'samples': 24746496, 'steps': 48332, 'loss/train': 0.7496263980865479}
+02/26/2022 16:11:14 - INFO - codeparrot_training - Step 48333: {'lr': 1.4865092802534285e-06, 'samples': 24747008, 'steps': 48333, 'loss/train': 1.6404470205307007}
+02/26/2022 16:11:18 - INFO - codeparrot_training - Step 48334: {'lr': 1.484728127605528e-06, 'samples': 24747520, 'steps': 48334, 'loss/train': 1.3127260208129883}
+02/26/2022 16:11:24 - INFO - codeparrot_training - Step 48335: {'lr': 1.4829480395181228e-06, 'samples': 24748032, 'steps': 48335, 'loss/train': 1.2283155918121338}
+02/26/2022 16:11:28 - INFO - codeparrot_training - Step 48336: {'lr': 1.4811690159988457e-06, 'samples': 24748544, 'steps': 48336, 'loss/train': 1.6740950345993042}
+02/26/2022 16:11:33 - INFO - codeparrot_training - Step 48337: {'lr': 1.479391057055357e-06, 'samples': 24749056, 'steps': 48337, 'loss/train': 1.8351653814315796}
+02/26/2022 16:11:37 - INFO - codeparrot_training - Step 48338: {'lr': 1.4776141626952344e-06, 'samples': 24749568, 'steps': 48338, 'loss/train': 1.1003621816635132}
+02/26/2022 16:11:42 - INFO - codeparrot_training - Step 48339: {'lr': 1.4758383329260828e-06, 'samples': 24750080, 'steps': 48339, 'loss/train': 1.506985068321228}
+02/26/2022 16:11:46 - INFO - codeparrot_training - Step 48340: {'lr': 1.4740635677555348e-06, 'samples': 24750592, 'steps': 48340, 'loss/train': 1.9745590686798096}
+02/26/2022 16:11:51 - INFO - codeparrot_training - Step 48341: {'lr': 1.4722898671911678e-06, 'samples': 24751104, 'steps': 48341, 'loss/train': 1.2870476245880127}
+02/26/2022 16:11:55 - INFO - codeparrot_training - Step 48342: {'lr': 1.4705172312406146e-06, 'samples': 24751616, 'steps': 48342, 'loss/train': 1.9093905687332153}
+02/26/2022 16:12:00 - INFO - codeparrot_training - Step 48343: {'lr': 1.4687456599114245e-06, 'samples': 24752128, 'steps': 48343, 'loss/train': 2.1184017658233643}
+02/26/2022 16:12:03 - INFO - codeparrot_training - Step 48344: {'lr': 1.4669751532112308e-06, 'samples': 24752640, 'steps': 48344, 'loss/train': 1.351211667060852}
+02/26/2022 16:12:10 - INFO - codeparrot_training - Step 48345: {'lr': 1.4652057111476103e-06, 'samples': 24753152, 'steps': 48345, 'loss/train': 0.8802154660224915}
+02/26/2022 16:12:13 - INFO - codeparrot_training - Step 48346: {'lr': 1.4634373337281125e-06, 'samples': 24753664, 'steps': 48346, 'loss/train': 1.505128026008606}
+02/26/2022 16:12:19 - INFO - codeparrot_training - Step 48347: {'lr': 1.4616700209603151e-06, 'samples': 24754176, 'steps': 48347, 'loss/train': 2.3156750202178955}
+02/26/2022 16:12:22 - INFO - codeparrot_training - Step 48348: {'lr': 1.4599037728518228e-06, 'samples': 24754688, 'steps': 48348, 'loss/train': 2.5019710063934326}
+02/26/2022 16:12:28 - INFO - codeparrot_training - Step 48349: {'lr': 1.458138589410185e-06, 'samples': 24755200, 'steps': 48349, 'loss/train': 1.2664331197738647}
+02/26/2022 16:12:31 - INFO - codeparrot_training - Step 48350: {'lr': 1.4563744706429517e-06, 'samples': 24755712, 'steps': 48350, 'loss/train': 1.1959731578826904}
+02/26/2022 16:12:37 - INFO - codeparrot_training - Step 48351: {'lr': 1.4546114165576995e-06, 'samples': 24756224, 'steps': 48351, 'loss/train': 1.715791940689087}
+02/26/2022 16:12:40 - INFO - codeparrot_training - Step 48352: {'lr': 1.4528494271619507e-06, 'samples': 24756736, 'steps': 48352, 'loss/train': 2.287785768508911}
+02/26/2022 16:12:46 - INFO - codeparrot_training - Step 48353: {'lr': 1.4510885024632825e-06, 'samples': 24757248, 'steps': 48353, 'loss/train': 1.948127031326294}
+02/26/2022 16:12:49 - INFO - codeparrot_training - Step 48354: {'lr': 1.4493286424692441e-06, 'samples': 24757760, 'steps': 48354, 'loss/train': 1.8993802070617676}
+02/26/2022 16:12:54 - INFO - codeparrot_training - Step 48355: {'lr': 1.44756984718733e-06, 'samples': 24758272, 'steps': 48355, 'loss/train': 1.9417109489440918}
+02/26/2022 16:12:58 - INFO - codeparrot_training - Step 48356: {'lr': 1.445812116625117e-06, 'samples': 24758784, 'steps': 48356, 'loss/train': 1.736548662185669}
+02/26/2022 16:13:03 - INFO - codeparrot_training - Step 48357: {'lr': 1.4440554507901272e-06, 'samples': 24759296, 'steps': 48357, 'loss/train': 0.8327333331108093}
+02/26/2022 16:13:09 - INFO - codeparrot_training - Step 48358: {'lr': 1.4422998496898543e-06, 'samples': 24759808, 'steps': 48358, 'loss/train': 2.761749744415283}
+02/26/2022 16:13:12 - INFO - codeparrot_training - Step 48359: {'lr': 1.4405453133318757e-06, 'samples': 24760320, 'steps': 48359, 'loss/train': 0.676769495010376}
+02/26/2022 16:13:19 - INFO - codeparrot_training - Step 48360: {'lr': 1.4387918417236578e-06, 'samples': 24760832, 'steps': 48360, 'loss/train': 1.3783200979232788}
+02/26/2022 16:13:23 - INFO - codeparrot_training - Step 48361: {'lr': 1.4370394348727223e-06, 'samples': 24761344, 'steps': 48361, 'loss/train': 1.2662625312805176}
+02/26/2022 16:13:28 - INFO - codeparrot_training - Step 48362: {'lr': 1.435288092786563e-06, 'samples': 24761856, 'steps': 48362, 'loss/train': 2.250563859939575}
+02/26/2022 16:13:32 - INFO - codeparrot_training - Step 48363: {'lr': 1.4335378154727574e-06, 'samples': 24762368, 'steps': 48363, 'loss/train': 1.7857874631881714}
+02/26/2022 16:13:37 - INFO - codeparrot_training - Step 48364: {'lr': 1.4317886029387162e-06, 'samples': 24762880, 'steps': 48364, 'loss/train': 2.1841542720794678}
+02/26/2022 16:13:41 - INFO - codeparrot_training - Step 48365: {'lr': 1.430040455191961e-06, 'samples': 24763392, 'steps': 48365, 'loss/train': 1.6927770376205444}
+02/26/2022 16:13:46 - INFO - codeparrot_training - Step 48366: {'lr': 1.4282933722399583e-06, 'samples': 24763904, 'steps': 48366, 'loss/train': 0.5043333768844604}
+02/26/2022 16:13:50 - INFO - codeparrot_training - Step 48367: {'lr': 1.4265473540902574e-06, 'samples': 24764416, 'steps': 48367, 'loss/train': 1.06955087184906}
+02/26/2022 16:13:56 - INFO - codeparrot_training - Step 48368: {'lr': 1.424802400750269e-06, 'samples': 24764928, 'steps': 48368, 'loss/train': 1.1917561292648315}
+02/26/2022 16:14:00 - INFO - codeparrot_training - Step 48369: {'lr': 1.4230585122275152e-06, 'samples': 24765440, 'steps': 48369, 'loss/train': 1.95912504196167}
+02/26/2022 16:14:03 - INFO - codeparrot_training - Step 48370: {'lr': 1.4213156885294343e-06, 'samples': 24765952, 'steps': 48370, 'loss/train': 1.4929018020629883}
+02/26/2022 16:14:09 - INFO - codeparrot_training - Step 48371: {'lr': 1.4195739296635202e-06, 'samples': 24766464, 'steps': 48371, 'loss/train': 2.3770599365234375}
+02/26/2022 16:14:13 - INFO - codeparrot_training - Step 48372: {'lr': 1.4178332356372114e-06, 'samples': 24766976, 'steps': 48372, 'loss/train': 2.1477420330047607}
+02/26/2022 16:14:18 - INFO - codeparrot_training - Step 48373: {'lr': 1.4160936064579744e-06, 'samples': 24767488, 'steps': 48373, 'loss/train': 0.5264931321144104}
+02/26/2022 16:14:22 - INFO - codeparrot_training - Step 48374: {'lr': 1.4143550421332196e-06, 'samples': 24768000, 'steps': 48374, 'loss/train': 1.6225192546844482}
+02/26/2022 16:14:27 - INFO - codeparrot_training - Step 48375: {'lr': 1.412617542670469e-06, 'samples': 24768512, 'steps': 48375, 'loss/train': 1.0946015119552612}
+02/26/2022 16:14:31 - INFO - codeparrot_training - Step 48376: {'lr': 1.4108811080771333e-06, 'samples': 24769024, 'steps': 48376, 'loss/train': 2.090301513671875}
+02/26/2022 16:14:36 - INFO - codeparrot_training - Step 48377: {'lr': 1.409145738360651e-06, 'samples': 24769536, 'steps': 48377, 'loss/train': 1.943251371383667}
+02/26/2022 16:14:40 - INFO - codeparrot_training - Step 48378: {'lr': 1.4074114335284327e-06, 'samples': 24770048, 'steps': 48378, 'loss/train': 2.4265153408050537}
+02/26/2022 16:14:45 - INFO - codeparrot_training - Step 48379: {'lr': 1.4056781935879448e-06, 'samples': 24770560, 'steps': 48379, 'loss/train': 1.4212602376937866}
+02/26/2022 16:14:48 - INFO - codeparrot_training - Step 48380: {'lr': 1.4039460185465703e-06, 'samples': 24771072, 'steps': 48380, 'loss/train': 0.567535400390625}
+02/26/2022 16:14:55 - INFO - codeparrot_training - Step 48381: {'lr': 1.4022149084117753e-06, 'samples': 24771584, 'steps': 48381, 'loss/train': 2.491757869720459}
+02/26/2022 16:14:58 - INFO - codeparrot_training - Step 48382: {'lr': 1.4004848631909428e-06, 'samples': 24772096, 'steps': 48382, 'loss/train': 0.42956382036209106}
+02/26/2022 16:15:04 - INFO - codeparrot_training - Step 48383: {'lr': 1.3987558828914837e-06, 'samples': 24772608, 'steps': 48383, 'loss/train': 1.173322081565857}
+02/26/2022 16:15:07 - INFO - codeparrot_training - Step 48384: {'lr': 1.3970279675208364e-06, 'samples': 24773120, 'steps': 48384, 'loss/train': 0.4197263717651367}
+02/26/2022 16:15:13 - INFO - codeparrot_training - Step 48385: {'lr': 1.395301117086356e-06, 'samples': 24773632, 'steps': 48385, 'loss/train': 2.993868350982666}
+02/26/2022 16:15:16 - INFO - codeparrot_training - Step 48386: {'lr': 1.3935753315954814e-06, 'samples': 24774144, 'steps': 48386, 'loss/train': 1.747635841369629}
+02/26/2022 16:15:22 - INFO - codeparrot_training - Step 48387: {'lr': 1.391850611055595e-06, 'samples': 24774656, 'steps': 48387, 'loss/train': 3.0021746158599854}
+02/26/2022 16:15:25 - INFO - codeparrot_training - Step 48388: {'lr': 1.3901269554740525e-06, 'samples': 24775168, 'steps': 48388, 'loss/train': 1.4070488214492798}
+02/26/2022 16:15:31 - INFO - codeparrot_training - Step 48389: {'lr': 1.3884043648582922e-06, 'samples': 24775680, 'steps': 48389, 'loss/train': 2.9738705158233643}
+02/26/2022 16:15:34 - INFO - codeparrot_training - Step 48390: {'lr': 1.3866828392156416e-06, 'samples': 24776192, 'steps': 48390, 'loss/train': 1.5986583232879639}
+02/26/2022 16:15:40 - INFO - codeparrot_training - Step 48391: {'lr': 1.3849623785535114e-06, 'samples': 24776704, 'steps': 48391, 'loss/train': 1.838472604751587}
+02/26/2022 16:15:44 - INFO - codeparrot_training - Step 48392: {'lr': 1.383242982879257e-06, 'samples': 24777216, 'steps': 48392, 'loss/train': 2.0443947315216064}
+02/26/2022 16:15:49 - INFO - codeparrot_training - Step 48393: {'lr': 1.3815246522002334e-06, 'samples': 24777728, 'steps': 48393, 'loss/train': 1.369342565536499}
+02/26/2022 16:15:53 - INFO - codeparrot_training - Step 48394: {'lr': 1.3798073865238236e-06, 'samples': 24778240, 'steps': 48394, 'loss/train': 1.599927544593811}
+02/26/2022 16:15:58 - INFO - codeparrot_training - Step 48395: {'lr': 1.378091185857383e-06, 'samples': 24778752, 'steps': 48395, 'loss/train': 1.2066359519958496}
+02/26/2022 16:16:02 - INFO - codeparrot_training - Step 48396: {'lr': 1.376376050208239e-06, 'samples': 24779264, 'steps': 48396, 'loss/train': 0.7172000408172607}
+02/26/2022 16:16:07 - INFO - codeparrot_training - Step 48397: {'lr': 1.3746619795837467e-06, 'samples': 24779776, 'steps': 48397, 'loss/train': 2.030796527862549}
+02/26/2022 16:16:11 - INFO - codeparrot_training - Step 48398: {'lr': 1.3729489739912893e-06, 'samples': 24780288, 'steps': 48398, 'loss/train': 1.5205879211425781}
+02/26/2022 16:16:16 - INFO - codeparrot_training - Step 48399: {'lr': 1.3712370334381385e-06, 'samples': 24780800, 'steps': 48399, 'loss/train': 1.164409875869751}
+02/26/2022 16:16:20 - INFO - codeparrot_training - Step 48400: {'lr': 1.3695261579316775e-06, 'samples': 24781312, 'steps': 48400, 'loss/train': 3.031507968902588}
+02/26/2022 16:16:25 - INFO - codeparrot_training - Step 48401: {'lr': 1.367816347479206e-06, 'samples': 24781824, 'steps': 48401, 'loss/train': 2.8563151359558105}
+02/26/2022 16:16:29 - INFO - codeparrot_training - Step 48402: {'lr': 1.3661076020880514e-06, 'samples': 24782336, 'steps': 48402, 'loss/train': 1.5785893201828003}
+02/26/2022 16:16:34 - INFO - codeparrot_training - Step 48403: {'lr': 1.364399921765569e-06, 'samples': 24782848, 'steps': 48403, 'loss/train': 1.141401767730713}
+02/26/2022 16:16:38 - INFO - codeparrot_training - Step 48404: {'lr': 1.3626933065190306e-06, 'samples': 24783360, 'steps': 48404, 'loss/train': 2.185882091522217}
+02/26/2022 16:16:43 - INFO - codeparrot_training - Step 48405: {'lr': 1.360987756355736e-06, 'samples': 24783872, 'steps': 48405, 'loss/train': 0.8031267523765564}
+02/26/2022 16:16:47 - INFO - codeparrot_training - Step 48406: {'lr': 1.3592832712830405e-06, 'samples': 24784384, 'steps': 48406, 'loss/train': 1.9465123414993286}
+02/26/2022 16:16:53 - INFO - codeparrot_training - Step 48407: {'lr': 1.3575798513082438e-06, 'samples': 24784896, 'steps': 48407, 'loss/train': 1.2222914695739746}
+02/26/2022 16:16:56 - INFO - codeparrot_training - Step 48408: {'lr': 1.3558774964386177e-06, 'samples': 24785408, 'steps': 48408, 'loss/train': 0.8967334032058716}
+02/26/2022 16:17:02 - INFO - codeparrot_training - Step 48409: {'lr': 1.3541762066814346e-06, 'samples': 24785920, 'steps': 48409, 'loss/train': 0.7138988971710205}
+02/26/2022 16:17:05 - INFO - codeparrot_training - Step 48410: {'lr': 1.3524759820440212e-06, 'samples': 24786432, 'steps': 48410, 'loss/train': 1.5887726545333862}
+02/26/2022 16:17:11 - INFO - codeparrot_training - Step 48411: {'lr': 1.3507768225336503e-06, 'samples': 24786944, 'steps': 48411, 'loss/train': 2.3518736362457275}
+02/26/2022 16:17:14 - INFO - codeparrot_training - Step 48412: {'lr': 1.3490787281575933e-06, 'samples': 24787456, 'steps': 48412, 'loss/train': 0.9374799728393555}
+02/26/2022 16:17:20 - INFO - codeparrot_training - Step 48413: {'lr': 1.3473816989230947e-06, 'samples': 24787968, 'steps': 48413, 'loss/train': 1.8472918272018433}
+02/26/2022 16:17:24 - INFO - codeparrot_training - Step 48414: {'lr': 1.3456857348374818e-06, 'samples': 24788480, 'steps': 48414, 'loss/train': 2.2600040435791016}
+02/26/2022 16:17:29 - INFO - codeparrot_training - Step 48415: {'lr': 1.3439908359080266e-06, 'samples': 24788992, 'steps': 48415, 'loss/train': 2.49967360496521}
+02/26/2022 16:17:33 - INFO - codeparrot_training - Step 48416: {'lr': 1.3422970021419178e-06, 'samples': 24789504, 'steps': 48416, 'loss/train': 0.8635879158973694}
+02/26/2022 16:17:39 - INFO - codeparrot_training - Step 48417: {'lr': 1.3406042335464552e-06, 'samples': 24790016, 'steps': 48417, 'loss/train': 2.140446901321411}
+02/26/2022 16:17:42 - INFO - codeparrot_training - Step 48418: {'lr': 1.3389125301289107e-06, 'samples': 24790528, 'steps': 48418, 'loss/train': 0.9869257807731628}
+02/26/2022 16:17:48 - INFO - codeparrot_training - Step 48419: {'lr': 1.337221891896473e-06, 'samples': 24791040, 'steps': 48419, 'loss/train': 1.6491167545318604}
+02/26/2022 16:17:51 - INFO - codeparrot_training - Step 48420: {'lr': 1.3355323188564417e-06, 'samples': 24791552, 'steps': 48420, 'loss/train': 3.9106357097625732}
+02/26/2022 16:17:57 - INFO - codeparrot_training - Step 48421: {'lr': 1.3338438110160056e-06, 'samples': 24792064, 'steps': 48421, 'loss/train': 1.005301594734192}
+02/26/2022 16:18:00 - INFO - codeparrot_training - Step 48422: {'lr': 1.3321563683824367e-06, 'samples': 24792576, 'steps': 48422, 'loss/train': 1.6604828834533691}
+02/26/2022 16:18:06 - INFO - codeparrot_training - Step 48423: {'lr': 1.3304699909629513e-06, 'samples': 24793088, 'steps': 48423, 'loss/train': 1.4465149641036987}
+02/26/2022 16:18:09 - INFO - codeparrot_training - Step 48424: {'lr': 1.3287846787647939e-06, 'samples': 24793600, 'steps': 48424, 'loss/train': 1.853302240371704}
+02/26/2022 16:18:15 - INFO - codeparrot_training - Step 48425: {'lr': 1.3271004317951252e-06, 'samples': 24794112, 'steps': 48425, 'loss/train': 1.9257960319519043}
+02/26/2022 16:18:18 - INFO - codeparrot_training - Step 48426: {'lr': 1.3254172500612171e-06, 'samples': 24794624, 'steps': 48426, 'loss/train': 0.797614336013794}
+02/26/2022 16:18:25 - INFO - codeparrot_training - Step 48427: {'lr': 1.3237351335702308e-06, 'samples': 24795136, 'steps': 48427, 'loss/train': 1.9741054773330688}
+02/26/2022 16:18:28 - INFO - codeparrot_training - Step 48428: {'lr': 1.3220540823294104e-06, 'samples': 24795648, 'steps': 48428, 'loss/train': 1.6475262641906738}
+02/26/2022 16:18:34 - INFO - codeparrot_training - Step 48429: {'lr': 1.320374096345972e-06, 'samples': 24796160, 'steps': 48429, 'loss/train': 1.7777506113052368}
+02/26/2022 16:18:37 - INFO - codeparrot_training - Step 48430: {'lr': 1.3186951756270493e-06, 'samples': 24796672, 'steps': 48430, 'loss/train': 1.9330613613128662}
+02/26/2022 16:18:43 - INFO - codeparrot_training - Step 48431: {'lr': 1.3170173201798863e-06, 'samples': 24797184, 'steps': 48431, 'loss/train': 1.565766453742981}
+02/26/2022 16:18:46 - INFO - codeparrot_training - Step 48432: {'lr': 1.3153405300116717e-06, 'samples': 24797696, 'steps': 48432, 'loss/train': 1.7955403327941895}
+02/26/2022 16:18:52 - INFO - codeparrot_training - Step 48433: {'lr': 1.3136648051295663e-06, 'samples': 24798208, 'steps': 48433, 'loss/train': 1.0911246538162231}
+02/26/2022 16:18:55 - INFO - codeparrot_training - Step 48434: {'lr': 1.3119901455407313e-06, 'samples': 24798720, 'steps': 48434, 'loss/train': 1.308014988899231}
+02/26/2022 16:19:01 - INFO - codeparrot_training - Step 48435: {'lr': 1.310316551252383e-06, 'samples': 24799232, 'steps': 48435, 'loss/train': 1.3888943195343018}
+02/26/2022 16:19:04 - INFO - codeparrot_training - Step 48436: {'lr': 1.3086440222716546e-06, 'samples': 24799744, 'steps': 48436, 'loss/train': 1.4703361988067627}
+02/26/2022 16:19:10 - INFO - codeparrot_training - Step 48437: {'lr': 1.3069725586057623e-06, 'samples': 24800256, 'steps': 48437, 'loss/train': 1.1862928867340088}
+02/26/2022 16:19:14 - INFO - codeparrot_training - Step 48438: {'lr': 1.305302160261812e-06, 'samples': 24800768, 'steps': 48438, 'loss/train': 0.6966497898101807}
+02/26/2022 16:19:19 - INFO - codeparrot_training - Step 48439: {'lr': 1.3036328272469644e-06, 'samples': 24801280, 'steps': 48439, 'loss/train': 1.1118990182876587}
+02/26/2022 16:19:23 - INFO - codeparrot_training - Step 48440: {'lr': 1.3019645595683804e-06, 'samples': 24801792, 'steps': 48440, 'loss/train': 1.9786264896392822}
+02/26/2022 16:19:28 - INFO - codeparrot_training - Step 48441: {'lr': 1.3002973572332211e-06, 'samples': 24802304, 'steps': 48441, 'loss/train': 1.5097253322601318}
+02/26/2022 16:19:32 - INFO - codeparrot_training - Step 48442: {'lr': 1.2986312202486195e-06, 'samples': 24802816, 'steps': 48442, 'loss/train': 0.760098397731781}
+02/26/2022 16:19:37 - INFO - codeparrot_training - Step 48443: {'lr': 1.296966148621709e-06, 'samples': 24803328, 'steps': 48443, 'loss/train': 0.6391351819038391}
+02/26/2022 16:19:41 - INFO - codeparrot_training - Step 48444: {'lr': 1.2953021423596223e-06, 'samples': 24803840, 'steps': 48444, 'loss/train': 2.5851023197174072}
+02/26/2022 16:19:46 - INFO - codeparrot_training - Step 48445: {'lr': 1.293639201469493e-06, 'samples': 24804352, 'steps': 48445, 'loss/train': 2.5467782020568848}
+02/26/2022 16:19:50 - INFO - codeparrot_training - Step 48446: {'lr': 1.2919773259584266e-06, 'samples': 24804864, 'steps': 48446, 'loss/train': 1.2552155256271362}
+02/26/2022 16:19:55 - INFO - codeparrot_training - Step 48447: {'lr': 1.2903165158335838e-06, 'samples': 24805376, 'steps': 48447, 'loss/train': 1.3458584547042847}
+02/26/2022 16:19:59 - INFO - codeparrot_training - Step 48448: {'lr': 1.2886567711020148e-06, 'samples': 24805888, 'steps': 48448, 'loss/train': 0.8545482754707336}
+02/26/2022 16:20:04 - INFO - codeparrot_training - Step 48449: {'lr': 1.2869980917708801e-06, 'samples': 24806400, 'steps': 48449, 'loss/train': 1.8371524810791016}
+02/26/2022 16:20:08 - INFO - codeparrot_training - Step 48450: {'lr': 1.2853404778472856e-06, 'samples': 24806912, 'steps': 48450, 'loss/train': 0.0968320369720459}
+02/26/2022 16:20:13 - INFO - codeparrot_training - Step 48451: {'lr': 1.2836839293383085e-06, 'samples': 24807424, 'steps': 48451, 'loss/train': 2.960578203201294}
+02/26/2022 16:20:17 - INFO - codeparrot_training - Step 48452: {'lr': 1.2820284462510267e-06, 'samples': 24807936, 'steps': 48452, 'loss/train': 1.594126582145691}
+02/26/2022 16:20:23 - INFO - codeparrot_training - Step 48453: {'lr': 1.2803740285926013e-06, 'samples': 24808448, 'steps': 48453, 'loss/train': 1.706684947013855}
+02/26/2022 16:20:26 - INFO - codeparrot_training - Step 48454: {'lr': 1.2787206763700265e-06, 'samples': 24808960, 'steps': 48454, 'loss/train': 2.9540576934814453}
+02/26/2022 16:20:32 - INFO - codeparrot_training - Step 48455: {'lr': 1.277068389590491e-06, 'samples': 24809472, 'steps': 48455, 'loss/train': 1.9281927347183228}
+02/26/2022 16:20:35 - INFO - codeparrot_training - Step 48456: {'lr': 1.2754171682609617e-06, 'samples': 24809984, 'steps': 48456, 'loss/train': 2.2091524600982666}
+02/26/2022 16:20:41 - INFO - codeparrot_training - Step 48457: {'lr': 1.2737670123885992e-06, 'samples': 24810496, 'steps': 48457, 'loss/train': 1.5281059741973877}
+02/26/2022 16:20:44 - INFO - codeparrot_training - Step 48458: {'lr': 1.2721179219804258e-06, 'samples': 24811008, 'steps': 48458, 'loss/train': 1.6805046796798706}
+02/26/2022 16:20:50 - INFO - codeparrot_training - Step 48459: {'lr': 1.270469897043547e-06, 'samples': 24811520, 'steps': 48459, 'loss/train': 2.421518564224243}
+02/26/2022 16:20:53 - INFO - codeparrot_training - Step 48460: {'lr': 1.2688229375849568e-06, 'samples': 24812032, 'steps': 48460, 'loss/train': 1.8304585218429565}
+02/26/2022 16:20:59 - INFO - codeparrot_training - Step 48461: {'lr': 1.2671770436117613e-06, 'samples': 24812544, 'steps': 48461, 'loss/train': 0.7550714015960693}
+02/26/2022 16:21:02 - INFO - codeparrot_training - Step 48462: {'lr': 1.26553221513101e-06, 'samples': 24813056, 'steps': 48462, 'loss/train': 1.0804290771484375}
+02/26/2022 16:21:09 - INFO - codeparrot_training - Step 48463: {'lr': 1.263888452149725e-06, 'samples': 24813568, 'steps': 48463, 'loss/train': 1.0991851091384888}
+02/26/2022 16:21:12 - INFO - codeparrot_training - Step 48464: {'lr': 1.2622457546749566e-06, 'samples': 24814080, 'steps': 48464, 'loss/train': 1.117771029472351}
+02/26/2022 16:21:18 - INFO - codeparrot_training - Step 48465: {'lr': 1.2606041227137544e-06, 'samples': 24814592, 'steps': 48465, 'loss/train': 1.8303498029708862}
+02/26/2022 16:21:21 - INFO - codeparrot_training - Step 48466: {'lr': 1.2589635562731405e-06, 'samples': 24815104, 'steps': 48466, 'loss/train': 1.2512751817703247}
+02/26/2022 16:21:27 - INFO - codeparrot_training - Step 48467: {'lr': 1.2573240553601374e-06, 'samples': 24815616, 'steps': 48467, 'loss/train': 1.6498616933822632}
+02/26/2022 16:21:30 - INFO - codeparrot_training - Step 48468: {'lr': 1.2556856199817668e-06, 'samples': 24816128, 'steps': 48468, 'loss/train': 2.1129143238067627}
+02/26/2022 16:21:36 - INFO - codeparrot_training - Step 48469: {'lr': 1.254048250145079e-06, 'samples': 24816640, 'steps': 48469, 'loss/train': 1.7093323469161987}
+02/26/2022 16:21:39 - INFO - codeparrot_training - Step 48470: {'lr': 1.2524119458570126e-06, 'samples': 24817152, 'steps': 48470, 'loss/train': 2.3321712017059326}
+02/26/2022 16:21:45 - INFO - codeparrot_training - Step 48471: {'lr': 1.2507767071246734e-06, 'samples': 24817664, 'steps': 48471, 'loss/train': 1.8468883037567139}
+02/26/2022 16:21:48 - INFO - codeparrot_training - Step 48472: {'lr': 1.249142533955e-06, 'samples': 24818176, 'steps': 48472, 'loss/train': 1.765306830406189}
+02/26/2022 16:21:54 - INFO - codeparrot_training - Step 48473: {'lr': 1.2475094263550423e-06, 'samples': 24818688, 'steps': 48473, 'loss/train': 1.2802129983901978}
+02/26/2022 16:21:58 - INFO - codeparrot_training - Step 48474: {'lr': 1.2458773843317118e-06, 'samples': 24819200, 'steps': 48474, 'loss/train': 1.5911240577697754}
+02/26/2022 16:22:03 - INFO - codeparrot_training - Step 48475: {'lr': 1.2442464078920855e-06, 'samples': 24819712, 'steps': 48475, 'loss/train': 0.43828141689300537}
+02/26/2022 16:22:07 - INFO - codeparrot_training - Step 48476: {'lr': 1.242616497043131e-06, 'samples': 24820224, 'steps': 48476, 'loss/train': 2.0459420680999756}
+02/26/2022 16:22:12 - INFO - codeparrot_training - Step 48477: {'lr': 1.2409876517918139e-06, 'samples': 24820736, 'steps': 48477, 'loss/train': 1.2960195541381836}
+02/26/2022 16:22:16 - INFO - codeparrot_training - Step 48478: {'lr': 1.2393598721451017e-06, 'samples': 24821248, 'steps': 48478, 'loss/train': 2.3047893047332764}
+02/26/2022 16:22:22 - INFO - codeparrot_training - Step 48479: {'lr': 1.2377331581099883e-06, 'samples': 24821760, 'steps': 48479, 'loss/train': 0.30409908294677734}
+02/26/2022 16:22:25 - INFO - codeparrot_training - Step 48480: {'lr': 1.2361075096934404e-06, 'samples': 24822272, 'steps': 48480, 'loss/train': 1.5186514854431152}
+02/26/2022 16:22:31 - INFO - codeparrot_training - Step 48481: {'lr': 1.2344829269023972e-06, 'samples': 24822784, 'steps': 48481, 'loss/train': 0.8618290424346924}
+02/26/2022 16:22:34 - INFO - codeparrot_training - Step 48482: {'lr': 1.2328594097438528e-06, 'samples': 24823296, 'steps': 48482, 'loss/train': 2.224228858947754}
+02/26/2022 16:22:40 - INFO - codeparrot_training - Step 48483: {'lr': 1.2312369582247462e-06, 'samples': 24823808, 'steps': 48483, 'loss/train': 1.2652878761291504}
+02/26/2022 16:22:43 - INFO - codeparrot_training - Step 48484: {'lr': 1.2296155723520163e-06, 'samples': 24824320, 'steps': 48484, 'loss/train': 2.494157314300537}
+02/26/2022 16:22:49 - INFO - codeparrot_training - Step 48485: {'lr': 1.2279952521326298e-06, 'samples': 24824832, 'steps': 48485, 'loss/train': 1.5959936380386353}
+02/26/2022 16:22:53 - INFO - codeparrot_training - Step 48486: {'lr': 1.2263759975735257e-06, 'samples': 24825344, 'steps': 48486, 'loss/train': 1.6957417726516724}
+02/26/2022 16:22:58 - INFO - codeparrot_training - Step 48487: {'lr': 1.2247578086816148e-06, 'samples': 24825856, 'steps': 48487, 'loss/train': 1.5053510665893555}
+02/26/2022 16:23:01 - INFO - codeparrot_training - Step 48488: {'lr': 1.223140685463864e-06, 'samples': 24826368, 'steps': 48488, 'loss/train': 1.4984582662582397}
+02/26/2022 16:23:07 - INFO - codeparrot_training - Step 48489: {'lr': 1.2215246279271563e-06, 'samples': 24826880, 'steps': 48489, 'loss/train': 2.164133071899414}
+02/26/2022 16:23:11 - INFO - codeparrot_training - Step 48490: {'lr': 1.219909636078459e-06, 'samples': 24827392, 'steps': 48490, 'loss/train': 1.4851778745651245}
+02/26/2022 16:23:16 - INFO - codeparrot_training - Step 48491: {'lr': 1.2182957099246828e-06, 'samples': 24827904, 'steps': 48491, 'loss/train': 2.2463409900665283}
+02/26/2022 16:23:20 - INFO - codeparrot_training - Step 48492: {'lr': 1.2166828494727112e-06, 'samples': 24828416, 'steps': 48492, 'loss/train': 1.7992091178894043}
+02/26/2022 16:23:25 - INFO - codeparrot_training - Step 48493: {'lr': 1.215071054729483e-06, 'samples': 24828928, 'steps': 48493, 'loss/train': 2.279757499694824}
+02/26/2022 16:23:29 - INFO - codeparrot_training - Step 48494: {'lr': 1.2134603257018817e-06, 'samples': 24829440, 'steps': 48494, 'loss/train': 2.6845078468322754}
+02/26/2022 16:23:34 - INFO - codeparrot_training - Step 48495: {'lr': 1.2118506623968184e-06, 'samples': 24829952, 'steps': 48495, 'loss/train': 2.1672542095184326}
+02/26/2022 16:23:38 - INFO - codeparrot_training - Step 48496: {'lr': 1.2102420648212041e-06, 'samples': 24830464, 'steps': 48496, 'loss/train': 1.6132885217666626}
+02/26/2022 16:23:43 - INFO - codeparrot_training - Step 48497: {'lr': 1.2086345329818949e-06, 'samples': 24830976, 'steps': 48497, 'loss/train': 2.910344362258911}
+02/26/2022 16:23:47 - INFO - codeparrot_training - Step 48498: {'lr': 1.2070280668858292e-06, 'samples': 24831488, 'steps': 48498, 'loss/train': 1.2739242315292358}
+02/26/2022 16:23:53 - INFO - codeparrot_training - Step 48499: {'lr': 1.205422666539807e-06, 'samples': 24832000, 'steps': 48499, 'loss/train': 1.6384413242340088}
+02/26/2022 16:23:57 - INFO - codeparrot_training - Step 48500: {'lr': 1.2038183319507957e-06, 'samples': 24832512, 'steps': 48500, 'loss/train': 2.0628714561462402}
+02/26/2022 16:24:02 - INFO - codeparrot_training - Step 48501: {'lr': 1.2022150631255947e-06, 'samples': 24833024, 'steps': 48501, 'loss/train': 2.5776493549346924}
+02/26/2022 16:24:06 - INFO - codeparrot_training - Step 48502: {'lr': 1.2006128600711152e-06, 'samples': 24833536, 'steps': 48502, 'loss/train': 1.632768154144287}
+02/26/2022 16:24:11 - INFO - codeparrot_training - Step 48503: {'lr': 1.1990117227941855e-06, 'samples': 24834048, 'steps': 48503, 'loss/train': 1.8911515474319458}
+02/26/2022 16:24:14 - INFO - codeparrot_training - Step 48504: {'lr': 1.1974116513017163e-06, 'samples': 24834560, 'steps': 48504, 'loss/train': 1.5433056354522705}
+02/26/2022 16:24:20 - INFO - codeparrot_training - Step 48505: {'lr': 1.1958126456005081e-06, 'samples': 24835072, 'steps': 48505, 'loss/train': 1.962333083152771}
+02/26/2022 16:24:24 - INFO - codeparrot_training - Step 48506: {'lr': 1.1942147056974717e-06, 'samples': 24835584, 'steps': 48506, 'loss/train': 1.8714172840118408}
+02/26/2022 16:24:29 - INFO - codeparrot_training - Step 48507: {'lr': 1.1926178315993796e-06, 'samples': 24836096, 'steps': 48507, 'loss/train': 1.2034095525741577}
+02/26/2022 16:24:33 - INFO - codeparrot_training - Step 48508: {'lr': 1.1910220233131153e-06, 'samples': 24836608, 'steps': 48508, 'loss/train': 0.48684197664260864}
+02/26/2022 16:24:39 - INFO - codeparrot_training - Step 48509: {'lr': 1.1894272808455064e-06, 'samples': 24837120, 'steps': 48509, 'loss/train': 1.3360004425048828}
+02/26/2022 16:24:42 - INFO - codeparrot_training - Step 48510: {'lr': 1.187833604203381e-06, 'samples': 24837632, 'steps': 48510, 'loss/train': 1.5148578882217407}
+02/26/2022 16:24:48 - INFO - codeparrot_training - Step 48511: {'lr': 1.1862409933935947e-06, 'samples': 24838144, 'steps': 48511, 'loss/train': 1.8555047512054443}
+02/26/2022 16:24:51 - INFO - codeparrot_training - Step 48512: {'lr': 1.1846494484229198e-06, 'samples': 24838656, 'steps': 48512, 'loss/train': 1.1072171926498413}
+02/26/2022 16:24:57 - INFO - codeparrot_training - Step 48513: {'lr': 1.1830589692982118e-06, 'samples': 24839168, 'steps': 48513, 'loss/train': 1.6367236375808716}
+02/26/2022 16:25:00 - INFO - codeparrot_training - Step 48514: {'lr': 1.181469556026271e-06, 'samples': 24839680, 'steps': 48514, 'loss/train': 1.1613576412200928}
+02/26/2022 16:25:06 - INFO - codeparrot_training - Step 48515: {'lr': 1.1798812086138977e-06, 'samples': 24840192, 'steps': 48515, 'loss/train': 1.9894667863845825}
+02/26/2022 16:25:09 - INFO - codeparrot_training - Step 48516: {'lr': 1.1782939270678917e-06, 'samples': 24840704, 'steps': 48516, 'loss/train': 1.737265944480896}
+02/26/2022 16:25:15 - INFO - codeparrot_training - Step 48517: {'lr': 1.1767077113950807e-06, 'samples': 24841216, 'steps': 48517, 'loss/train': 0.4749722182750702}
+02/26/2022 16:25:18 - INFO - codeparrot_training - Step 48518: {'lr': 1.1751225616022376e-06, 'samples': 24841728, 'steps': 48518, 'loss/train': 1.4081003665924072}
+02/26/2022 16:25:24 - INFO - codeparrot_training - Step 48519: {'lr': 1.1735384776961622e-06, 'samples': 24842240, 'steps': 48519, 'loss/train': 2.072645425796509}
+02/26/2022 16:25:28 - INFO - codeparrot_training - Step 48520: {'lr': 1.1719554596836545e-06, 'samples': 24842752, 'steps': 48520, 'loss/train': 1.3582074642181396}
+02/26/2022 16:25:33 - INFO - codeparrot_training - Step 48521: {'lr': 1.1703735075714594e-06, 'samples': 24843264, 'steps': 48521, 'loss/train': 1.6174136400222778}
+02/26/2022 16:25:37 - INFO - codeparrot_training - Step 48522: {'lr': 1.1687926213663769e-06, 'samples': 24843776, 'steps': 48522, 'loss/train': 1.0460364818572998}
+02/26/2022 16:25:42 - INFO - codeparrot_training - Step 48523: {'lr': 1.1672128010751515e-06, 'samples': 24844288, 'steps': 48523, 'loss/train': 2.1129560470581055}
+02/26/2022 16:25:46 - INFO - codeparrot_training - Step 48524: {'lr': 1.165634046704611e-06, 'samples': 24844800, 'steps': 48524, 'loss/train': 1.6464053392410278}
+02/26/2022 16:25:51 - INFO - codeparrot_training - Step 48525: {'lr': 1.164056358261445e-06, 'samples': 24845312, 'steps': 48525, 'loss/train': 2.392225980758667}
+02/26/2022 16:25:55 - INFO - codeparrot_training - Step 48526: {'lr': 1.162479735752453e-06, 'samples': 24845824, 'steps': 48526, 'loss/train': 1.4088304042816162}
+02/26/2022 16:26:00 - INFO - codeparrot_training - Step 48527: {'lr': 1.1609041791844077e-06, 'samples': 24846336, 'steps': 48527, 'loss/train': 2.3873565196990967}
+02/26/2022 16:26:03 - INFO - codeparrot_training - Step 48528: {'lr': 1.1593296885640259e-06, 'samples': 24846848, 'steps': 48528, 'loss/train': 1.412457823753357}
+02/26/2022 16:26:09 - INFO - codeparrot_training - Step 48529: {'lr': 1.157756263898052e-06, 'samples': 24847360, 'steps': 48529, 'loss/train': 1.4757179021835327}
+02/26/2022 16:26:12 - INFO - codeparrot_training - Step 48530: {'lr': 1.156183905193231e-06, 'samples': 24847872, 'steps': 48530, 'loss/train': 2.0022785663604736}
+02/26/2022 16:26:19 - INFO - codeparrot_training - Step 48531: {'lr': 1.1546126124563072e-06, 'samples': 24848384, 'steps': 48531, 'loss/train': 1.9121297597885132}
+02/26/2022 16:26:23 - INFO - codeparrot_training - Step 48532: {'lr': 1.1530423856939975e-06, 'samples': 24848896, 'steps': 48532, 'loss/train': 0.6375647187232971}
+02/26/2022 16:26:28 - INFO - codeparrot_training - Step 48533: {'lr': 1.1514732249130466e-06, 'samples': 24849408, 'steps': 48533, 'loss/train': 1.8892377614974976}
+02/26/2022 16:26:32 - INFO - codeparrot_training - Step 48534: {'lr': 1.1499051301201713e-06, 'samples': 24849920, 'steps': 48534, 'loss/train': 1.2111845016479492}
+02/26/2022 16:26:37 - INFO - codeparrot_training - Step 48535: {'lr': 1.1483381013220607e-06, 'samples': 24850432, 'steps': 48535, 'loss/train': 1.4634150266647339}
+02/26/2022 16:26:41 - INFO - codeparrot_training - Step 48536: {'lr': 1.1467721385254593e-06, 'samples': 24850944, 'steps': 48536, 'loss/train': 0.2927151620388031}
+02/26/2022 16:26:46 - INFO - codeparrot_training - Step 48537: {'lr': 1.145207241737084e-06, 'samples': 24851456, 'steps': 48537, 'loss/train': 2.428645133972168}
+02/26/2022 16:26:50 - INFO - codeparrot_training - Step 48538: {'lr': 1.1436434109635962e-06, 'samples': 24851968, 'steps': 48538, 'loss/train': 1.378011703491211}
+02/26/2022 16:26:55 - INFO - codeparrot_training - Step 48539: {'lr': 1.1420806462117406e-06, 'samples': 24852480, 'steps': 48539, 'loss/train': 1.6328016519546509}
+02/26/2022 16:26:59 - INFO - codeparrot_training - Step 48540: {'lr': 1.1405189474881783e-06, 'samples': 24852992, 'steps': 48540, 'loss/train': 1.3970850706100464}
+02/26/2022 16:27:05 - INFO - codeparrot_training - Step 48541: {'lr': 1.1389583147996262e-06, 'samples': 24853504, 'steps': 48541, 'loss/train': 0.06693976372480392}
+02/26/2022 16:27:09 - INFO - codeparrot_training - Step 48542: {'lr': 1.1373987481527459e-06, 'samples': 24854016, 'steps': 48542, 'loss/train': 0.04737776517868042}
+02/26/2022 16:27:14 - INFO - codeparrot_training - Step 48543: {'lr': 1.1358402475542261e-06, 'samples': 24854528, 'steps': 48543, 'loss/train': 0.8508061170578003}
+02/26/2022 16:27:18 - INFO - codeparrot_training - Step 48544: {'lr': 1.1342828130107286e-06, 'samples': 24855040, 'steps': 48544, 'loss/train': 2.413389205932617}
+02/26/2022 16:27:23 - INFO - codeparrot_training - Step 48545: {'lr': 1.132726444528942e-06, 'samples': 24855552, 'steps': 48545, 'loss/train': 1.0434762239456177}
+02/26/2022 16:27:27 - INFO - codeparrot_training - Step 48546: {'lr': 1.1311711421155556e-06, 'samples': 24856064, 'steps': 48546, 'loss/train': 1.6834280490875244}
+02/26/2022 16:27:32 - INFO - codeparrot_training - Step 48547: {'lr': 1.1296169057771756e-06, 'samples': 24856576, 'steps': 48547, 'loss/train': 2.1511762142181396}
+02/26/2022 16:27:36 - INFO - codeparrot_training - Step 48548: {'lr': 1.1280637355205182e-06, 'samples': 24857088, 'steps': 48548, 'loss/train': 1.6215265989303589}
+02/26/2022 16:27:41 - INFO - codeparrot_training - Step 48549: {'lr': 1.1265116313521894e-06, 'samples': 24857600, 'steps': 48549, 'loss/train': 1.106916069984436}
+02/26/2022 16:27:45 - INFO - codeparrot_training - Step 48550: {'lr': 1.124960593278851e-06, 'samples': 24858112, 'steps': 48550, 'loss/train': 1.8584131002426147}
+02/26/2022 16:27:51 - INFO - codeparrot_training - Step 48551: {'lr': 1.1234106213071915e-06, 'samples': 24858624, 'steps': 48551, 'loss/train': 1.3429499864578247}
+02/26/2022 16:27:54 - INFO - codeparrot_training - Step 48552: {'lr': 1.1218617154437893e-06, 'samples': 24859136, 'steps': 48552, 'loss/train': 0.07465500384569168}
+02/26/2022 16:28:00 - INFO - codeparrot_training - Step 48553: {'lr': 1.1203138756953058e-06, 'samples': 24859648, 'steps': 48553, 'loss/train': 1.2795872688293457}
+02/26/2022 16:28:03 - INFO - codeparrot_training - Step 48554: {'lr': 1.1187671020683743e-06, 'samples': 24860160, 'steps': 48554, 'loss/train': 1.0946496725082397}
+02/26/2022 16:28:09 - INFO - codeparrot_training - Step 48555: {'lr': 1.1172213945696286e-06, 'samples': 24860672, 'steps': 48555, 'loss/train': 1.3925013542175293}
+02/26/2022 16:28:13 - INFO - codeparrot_training - Step 48556: {'lr': 1.1156767532056467e-06, 'samples': 24861184, 'steps': 48556, 'loss/train': 2.6532912254333496}
+02/26/2022 16:28:18 - INFO - codeparrot_training - Step 48557: {'lr': 1.11413317798309e-06, 'samples': 24861696, 'steps': 48557, 'loss/train': 1.9645737409591675}
+02/26/2022 16:28:22 - INFO - codeparrot_training - Step 48558: {'lr': 1.1125906689085641e-06, 'samples': 24862208, 'steps': 48558, 'loss/train': 1.743770718574524}
+02/26/2022 16:28:27 - INFO - codeparrot_training - Step 48559: {'lr': 1.111049225988675e-06, 'samples': 24862720, 'steps': 48559, 'loss/train': 2.0174946784973145}
+02/26/2022 16:28:31 - INFO - codeparrot_training - Step 48560: {'lr': 1.109508849230001e-06, 'samples': 24863232, 'steps': 48560, 'loss/train': 1.5074763298034668}
+02/26/2022 16:28:36 - INFO - codeparrot_training - Step 48561: {'lr': 1.1079695386391754e-06, 'samples': 24863744, 'steps': 48561, 'loss/train': 1.159696102142334}
+02/26/2022 16:28:40 - INFO - codeparrot_training - Step 48562: {'lr': 1.1064312942227484e-06, 'samples': 24864256, 'steps': 48562, 'loss/train': 1.5916857719421387}
+02/26/2022 16:28:45 - INFO - codeparrot_training - Step 48563: {'lr': 1.1048941159873538e-06, 'samples': 24864768, 'steps': 48563, 'loss/train': 1.5484447479248047}
+02/26/2022 16:28:48 - INFO - codeparrot_training - Step 48564: {'lr': 1.1033580039395697e-06, 'samples': 24865280, 'steps': 48564, 'loss/train': 2.149129867553711}
+02/26/2022 16:28:55 - INFO - codeparrot_training - Step 48565: {'lr': 1.101822958085974e-06, 'samples': 24865792, 'steps': 48565, 'loss/train': 1.097700595855713}
+02/26/2022 16:28:58 - INFO - codeparrot_training - Step 48566: {'lr': 1.100288978433117e-06, 'samples': 24866304, 'steps': 48566, 'loss/train': 1.9912561178207397}
+02/26/2022 16:29:04 - INFO - codeparrot_training - Step 48567: {'lr': 1.0987560649875771e-06, 'samples': 24866816, 'steps': 48567, 'loss/train': 1.572648286819458}
+02/26/2022 16:29:07 - INFO - codeparrot_training - Step 48568: {'lr': 1.0972242177559599e-06, 'samples': 24867328, 'steps': 48568, 'loss/train': 1.4035009145736694}
+02/26/2022 16:29:13 - INFO - codeparrot_training - Step 48569: {'lr': 1.0956934367447601e-06, 'samples': 24867840, 'steps': 48569, 'loss/train': 3.298931121826172}
+02/26/2022 16:29:16 - INFO - codeparrot_training - Step 48570: {'lr': 1.0941637219606114e-06, 'samples': 24868352, 'steps': 48570, 'loss/train': 0.9593859314918518}
+02/26/2022 16:29:20 - INFO - codeparrot_training - Step 48571: {'lr': 1.0926350734100088e-06, 'samples': 24868864, 'steps': 48571, 'loss/train': 2.4536402225494385}
+02/26/2022 16:29:25 - INFO - codeparrot_training - Step 48572: {'lr': 1.09110749109953e-06, 'samples': 24869376, 'steps': 48572, 'loss/train': 2.2696402072906494}
+02/26/2022 16:29:29 - INFO - codeparrot_training - Step 48573: {'lr': 1.0895809750356978e-06, 'samples': 24869888, 'steps': 48573, 'loss/train': 1.0938876867294312}
+02/26/2022 16:29:34 - INFO - codeparrot_training - Step 48574: {'lr': 1.0880555252250623e-06, 'samples': 24870400, 'steps': 48574, 'loss/train': 0.8961057066917419}
+02/26/2022 16:29:38 - INFO - codeparrot_training - Step 48575: {'lr': 1.0865311416741742e-06, 'samples': 24870912, 'steps': 48575, 'loss/train': 2.8931827545166016}
+02/26/2022 16:29:43 - INFO - codeparrot_training - Step 48576: {'lr': 1.0850078243895278e-06, 'samples': 24871424, 'steps': 48576, 'loss/train': 1.0553991794586182}
+02/26/2022 16:29:47 - INFO - codeparrot_training - Step 48577: {'lr': 1.0834855733777017e-06, 'samples': 24871936, 'steps': 48577, 'loss/train': 1.8819741010665894}
+02/26/2022 16:29:53 - INFO - codeparrot_training - Step 48578: {'lr': 1.0819643886451625e-06, 'samples': 24872448, 'steps': 48578, 'loss/train': 1.4161831140518188}
+02/26/2022 16:29:57 - INFO - codeparrot_training - Step 48579: {'lr': 1.0804442701984329e-06, 'samples': 24872960, 'steps': 48579, 'loss/train': 1.3909871578216553}
+02/26/2022 16:30:02 - INFO - codeparrot_training - Step 48580: {'lr': 1.0789252180440635e-06, 'samples': 24873472, 'steps': 48580, 'loss/train': 1.8278183937072754}
+02/26/2022 16:30:06 - INFO - codeparrot_training - Step 48581: {'lr': 1.077407232188521e-06, 'samples': 24873984, 'steps': 48581, 'loss/train': 1.3987740278244019}
+02/26/2022 16:30:11 - INFO - codeparrot_training - Step 48582: {'lr': 1.075890312638328e-06, 'samples': 24874496, 'steps': 48582, 'loss/train': 0.39701828360557556}
+02/26/2022 16:30:15 - INFO - codeparrot_training - Step 48583: {'lr': 1.0743744593999794e-06, 'samples': 24875008, 'steps': 48583, 'loss/train': 2.290710210800171}
+02/26/2022 16:30:20 - INFO - codeparrot_training - Step 48584: {'lr': 1.07285967247997e-06, 'samples': 24875520, 'steps': 48584, 'loss/train': 1.753527283668518}
+02/26/2022 16:30:24 - INFO - codeparrot_training - Step 48585: {'lr': 1.0713459518847667e-06, 'samples': 24876032, 'steps': 48585, 'loss/train': 1.1332815885543823}
+02/26/2022 16:30:30 - INFO - codeparrot_training - Step 48586: {'lr': 1.0698332976209201e-06, 'samples': 24876544, 'steps': 48586, 'loss/train': 2.501163959503174}
+02/26/2022 16:30:33 - INFO - codeparrot_training - Step 48587: {'lr': 1.0683217096948138e-06, 'samples': 24877056, 'steps': 48587, 'loss/train': 2.2284014225006104}
+02/26/2022 16:30:39 - INFO - codeparrot_training - Step 48588: {'lr': 1.0668111881129982e-06, 'samples': 24877568, 'steps': 48588, 'loss/train': 2.13496470451355}
+02/26/2022 16:30:42 - INFO - codeparrot_training - Step 48589: {'lr': 1.0653017328819126e-06, 'samples': 24878080, 'steps': 48589, 'loss/train': 1.9204634428024292}
+02/26/2022 16:30:48 - INFO - codeparrot_training - Step 48590: {'lr': 1.0637933440080516e-06, 'samples': 24878592, 'steps': 48590, 'loss/train': 1.3442003726959229}
+02/26/2022 16:30:51 - INFO - codeparrot_training - Step 48591: {'lr': 1.0622860214978269e-06, 'samples': 24879104, 'steps': 48591, 'loss/train': 1.264943242073059}
+02/26/2022 16:30:57 - INFO - codeparrot_training - Step 48592: {'lr': 1.0607797653577334e-06, 'samples': 24879616, 'steps': 48592, 'loss/train': 1.5720057487487793}
+02/26/2022 16:31:00 - INFO - codeparrot_training - Step 48593: {'lr': 1.0592745755941823e-06, 'samples': 24880128, 'steps': 48593, 'loss/train': 1.7504733800888062}
+02/26/2022 16:31:05 - INFO - codeparrot_training - Step 48594: {'lr': 1.0577704522136967e-06, 'samples': 24880640, 'steps': 48594, 'loss/train': 1.8449431657791138}
+02/26/2022 16:31:09 - INFO - codeparrot_training - Step 48595: {'lr': 1.056267395222632e-06, 'samples': 24881152, 'steps': 48595, 'loss/train': 1.8730652332305908}
+02/26/2022 16:31:16 - INFO - codeparrot_training - Step 48596: {'lr': 1.0547654046275114e-06, 'samples': 24881664, 'steps': 48596, 'loss/train': 1.0368913412094116}
+02/26/2022 16:31:19 - INFO - codeparrot_training - Step 48597: {'lr': 1.0532644804346902e-06, 'samples': 24882176, 'steps': 48597, 'loss/train': 1.398057222366333}
+02/26/2022 16:31:25 - INFO - codeparrot_training - Step 48598: {'lr': 1.0517646226506638e-06, 'samples': 24882688, 'steps': 48598, 'loss/train': 1.4415875673294067}
+02/26/2022 16:31:30 - INFO - codeparrot_training - Step 48599: {'lr': 1.0502658312818159e-06, 'samples': 24883200, 'steps': 48599, 'loss/train': 1.8884658813476562}
+02/26/2022 16:31:34 - INFO - codeparrot_training - Step 48600: {'lr': 1.0487681063345856e-06, 'samples': 24883712, 'steps': 48600, 'loss/train': 2.6032891273498535}
+02/26/2022 16:31:40 - INFO - codeparrot_training - Step 48601: {'lr': 1.0472714478153567e-06, 'samples': 24884224, 'steps': 48601, 'loss/train': 2.009718656539917}
+02/26/2022 16:31:43 - INFO - codeparrot_training - Step 48602: {'lr': 1.0457758557305687e-06, 'samples': 24884736, 'steps': 48602, 'loss/train': 2.674697160720825}
+02/26/2022 16:31:49 - INFO - codeparrot_training - Step 48603: {'lr': 1.0442813300866605e-06, 'samples': 24885248, 'steps': 48603, 'loss/train': 1.2860231399536133}
+02/26/2022 16:31:52 - INFO - codeparrot_training - Step 48604: {'lr': 1.0427878708899608e-06, 'samples': 24885760, 'steps': 48604, 'loss/train': 0.46338459849357605}
+02/26/2022 16:31:56 - INFO - codeparrot_training - Step 48605: {'lr': 1.0412954781469363e-06, 'samples': 24886272, 'steps': 48605, 'loss/train': 0.6531505584716797}
+02/26/2022 16:32:02 - INFO - codeparrot_training - Step 48606: {'lr': 1.0398041518639157e-06, 'samples': 24886784, 'steps': 48606, 'loss/train': 1.8503721952438354}
+02/26/2022 16:32:06 - INFO - codeparrot_training - Step 48607: {'lr': 1.0383138920473378e-06, 'samples': 24887296, 'steps': 48607, 'loss/train': 1.4208728075027466}
+02/26/2022 16:32:11 - INFO - codeparrot_training - Step 48608: {'lr': 1.0368246987035868e-06, 'samples': 24887808, 'steps': 48608, 'loss/train': 0.08011391758918762}
+02/26/2022 16:32:15 - INFO - codeparrot_training - Step 48609: {'lr': 1.0353365718390184e-06, 'samples': 24888320, 'steps': 48609, 'loss/train': 1.9577525854110718}
+02/26/2022 16:32:20 - INFO - codeparrot_training - Step 48610: {'lr': 1.0338495114599887e-06, 'samples': 24888832, 'steps': 48610, 'loss/train': 0.777114987373352}
+02/26/2022 16:32:24 - INFO - codeparrot_training - Step 48611: {'lr': 1.0323635175729095e-06, 'samples': 24889344, 'steps': 48611, 'loss/train': 1.9848923683166504}
+02/26/2022 16:32:29 - INFO - codeparrot_training - Step 48612: {'lr': 1.030878590184109e-06, 'samples': 24889856, 'steps': 48612, 'loss/train': 1.8646644353866577}
+02/26/2022 16:32:33 - INFO - codeparrot_training - Step 48613: {'lr': 1.0293947292999984e-06, 'samples': 24890368, 'steps': 48613, 'loss/train': 2.247981309890747}
+02/26/2022 16:32:38 - INFO - codeparrot_training - Step 48614: {'lr': 1.0279119349268783e-06, 'samples': 24890880, 'steps': 48614, 'loss/train': 1.999135971069336}
+02/26/2022 16:32:42 - INFO - codeparrot_training - Step 48615: {'lr': 1.0264302070711328e-06, 'samples': 24891392, 'steps': 48615, 'loss/train': 1.5612785816192627}
+02/26/2022 16:32:47 - INFO - codeparrot_training - Step 48616: {'lr': 1.0249495457391177e-06, 'samples': 24891904, 'steps': 48616, 'loss/train': 2.289363384246826}
+02/26/2022 16:32:51 - INFO - codeparrot_training - Step 48617: {'lr': 1.0234699509371336e-06, 'samples': 24892416, 'steps': 48617, 'loss/train': 2.0629048347473145}
+02/26/2022 16:32:58 - INFO - codeparrot_training - Step 48618: {'lr': 1.0219914226715643e-06, 'samples': 24892928, 'steps': 48618, 'loss/train': 2.570551872253418}
+02/26/2022 16:33:03 - INFO - codeparrot_training - Step 48619: {'lr': 1.0205139609487102e-06, 'samples': 24893440, 'steps': 48619, 'loss/train': 1.9808770418167114}
+02/26/2022 16:33:07 - INFO - codeparrot_training - Step 48620: {'lr': 1.0190375657749273e-06, 'samples': 24893952, 'steps': 48620, 'loss/train': 1.8617417812347412}
+02/26/2022 16:33:13 - INFO - codeparrot_training - Step 48621: {'lr': 1.017562237156544e-06, 'samples': 24894464, 'steps': 48621, 'loss/train': 1.5349764823913574}
+02/26/2022 16:33:16 - INFO - codeparrot_training - Step 48622: {'lr': 1.016087975099833e-06, 'samples': 24894976, 'steps': 48622, 'loss/train': 5.161903381347656}
+02/26/2022 16:33:22 - INFO - codeparrot_training - Step 48623: {'lr': 1.0146147796111505e-06, 'samples': 24895488, 'steps': 48623, 'loss/train': 1.5258817672729492}
+02/26/2022 16:33:25 - INFO - codeparrot_training - Step 48624: {'lr': 1.0131426506967689e-06, 'samples': 24896000, 'steps': 48624, 'loss/train': 1.4922313690185547}
+02/26/2022 16:33:31 - INFO - codeparrot_training - Step 48625: {'lr': 1.0116715883630723e-06, 'samples': 24896512, 'steps': 48625, 'loss/train': 0.6071093678474426}
+02/26/2022 16:33:34 - INFO - codeparrot_training - Step 48626: {'lr': 1.0102015926162778e-06, 'samples': 24897024, 'steps': 48626, 'loss/train': 0.1705748289823532}
+02/26/2022 16:33:40 - INFO - codeparrot_training - Step 48627: {'lr': 1.0087326634627415e-06, 'samples': 24897536, 'steps': 48627, 'loss/train': 0.19413359463214874}
+02/26/2022 16:33:44 - INFO - codeparrot_training - Step 48628: {'lr': 1.0072648009087083e-06, 'samples': 24898048, 'steps': 48628, 'loss/train': 2.0149431228637695}
+02/26/2022 16:33:50 - INFO - codeparrot_training - Step 48629: {'lr': 1.0057980049605065e-06, 'samples': 24898560, 'steps': 48629, 'loss/train': 1.406548261642456}
+02/26/2022 16:33:53 - INFO - codeparrot_training - Step 48630: {'lr': 1.0043322756243812e-06, 'samples': 24899072, 'steps': 48630, 'loss/train': 1.3164587020874023}
+02/26/2022 16:33:57 - INFO - codeparrot_training - Step 48631: {'lr': 1.0028676129066327e-06, 'samples': 24899584, 'steps': 48631, 'loss/train': 1.6917215585708618}
+02/26/2022 16:34:02 - INFO - codeparrot_training - Step 48632: {'lr': 1.001404016813534e-06, 'samples': 24900096, 'steps': 48632, 'loss/train': 0.969963014125824}
+02/26/2022 16:34:06 - INFO - codeparrot_training - Step 48633: {'lr': 9.999414873513857e-07, 'samples': 24900608, 'steps': 48633, 'loss/train': 1.9355295896530151}
+02/26/2022 16:34:11 - INFO - codeparrot_training - Step 48634: {'lr': 9.984800245264047e-07, 'samples': 24901120, 'steps': 48634, 'loss/train': 3.2074432373046875}
+02/26/2022 16:34:17 - INFO - codeparrot_training - Step 48635: {'lr': 9.97019628344864e-07, 'samples': 24901632, 'steps': 48635, 'loss/train': 2.0678586959838867}
+02/26/2022 16:34:20 - INFO - codeparrot_training - Step 48636: {'lr': 9.955602988130085e-07, 'samples': 24902144, 'steps': 48636, 'loss/train': 2.0958006381988525}
+02/26/2022 16:34:26 - INFO - codeparrot_training - Step 48637: {'lr': 9.94102035937111e-07, 'samples': 24902656, 'steps': 48637, 'loss/train': 1.2168176174163818}
+02/26/2022 16:34:29 - INFO - codeparrot_training - Step 48638: {'lr': 9.926448397234167e-07, 'samples': 24903168, 'steps': 48638, 'loss/train': 1.7434332370758057}
+02/26/2022 16:34:35 - INFO - codeparrot_training - Step 48639: {'lr': 9.911887101781703e-07, 'samples': 24903680, 'steps': 48639, 'loss/train': 2.728332996368408}
+02/26/2022 16:34:38 - INFO - codeparrot_training - Step 48640: {'lr': 9.897336473076167e-07, 'samples': 24904192, 'steps': 48640, 'loss/train': 1.2537264823913574}
+02/26/2022 16:34:42 - INFO - codeparrot_training - Step 48641: {'lr': 9.882796511179459e-07, 'samples': 24904704, 'steps': 48641, 'loss/train': 0.8997842669487}
+02/26/2022 16:34:47 - INFO - codeparrot_training - Step 48642: {'lr': 9.868267216154303e-07, 'samples': 24905216, 'steps': 48642, 'loss/train': 1.4228382110595703}
+02/26/2022 16:34:51 - INFO - codeparrot_training - Step 48643: {'lr': 9.853748588062873e-07, 'samples': 24905728, 'steps': 48643, 'loss/train': 1.6537553071975708}
+02/26/2022 16:34:57 - INFO - codeparrot_training - Step 48644: {'lr': 9.83924062696706e-07, 'samples': 24906240, 'steps': 48644, 'loss/train': 3.461794376373291}
+02/26/2022 16:35:02 - INFO - codeparrot_training - Step 48645: {'lr': 9.82474333292932e-07, 'samples': 24906752, 'steps': 48645, 'loss/train': 2.364292621612549}
+02/26/2022 16:35:06 - INFO - codeparrot_training - Step 48646: {'lr': 9.810256706011545e-07, 'samples': 24907264, 'steps': 48646, 'loss/train': 1.7181931734085083}
+02/26/2022 16:35:11 - INFO - codeparrot_training - Step 48647: {'lr': 9.795780746275906e-07, 'samples': 24907776, 'steps': 48647, 'loss/train': 1.9545812606811523}
+02/26/2022 16:35:15 - INFO - codeparrot_training - Step 48648: {'lr': 9.78131545378458e-07, 'samples': 24908288, 'steps': 48648, 'loss/train': 0.8487032055854797}
+02/26/2022 16:35:21 - INFO - codeparrot_training - Step 48649: {'lr': 9.766860828599178e-07, 'samples': 24908800, 'steps': 48649, 'loss/train': 2.215221405029297}
+02/26/2022 16:35:24 - INFO - codeparrot_training - Step 48650: {'lr': 9.752416870782156e-07, 'samples': 24909312, 'steps': 48650, 'loss/train': 0.03904295712709427}
+02/26/2022 16:35:29 - INFO - codeparrot_training - Step 48651: {'lr': 9.73798358039485e-07, 'samples': 24909824, 'steps': 48651, 'loss/train': 1.9457398653030396}
+02/26/2022 16:35:33 - INFO - codeparrot_training - Step 48652: {'lr': 9.723560957499156e-07, 'samples': 24910336, 'steps': 48652, 'loss/train': 1.3193747997283936}
+02/26/2022 16:35:39 - INFO - codeparrot_training - Step 48653: {'lr': 9.709149002156969e-07, 'samples': 24910848, 'steps': 48653, 'loss/train': 2.0491995811462402}
+02/26/2022 16:35:43 - INFO - codeparrot_training - Step 48654: {'lr': 9.694747714430186e-07, 'samples': 24911360, 'steps': 48654, 'loss/train': 1.517682671546936}
+02/26/2022 16:35:48 - INFO - codeparrot_training - Step 48655: {'lr': 9.680357094380144e-07, 'samples': 24911872, 'steps': 48655, 'loss/train': 2.345839500427246}
+02/26/2022 16:35:51 - INFO - codeparrot_training - Step 48656: {'lr': 9.665977142068738e-07, 'samples': 24912384, 'steps': 48656, 'loss/train': 0.9726991057395935}
+02/26/2022 16:35:57 - INFO - codeparrot_training - Step 48657: {'lr': 9.651607857557586e-07, 'samples': 24912896, 'steps': 48657, 'loss/train': 0.8982393145561218}
+02/26/2022 16:36:01 - INFO - codeparrot_training - Step 48658: {'lr': 9.637249240908307e-07, 'samples': 24913408, 'steps': 48658, 'loss/train': 1.6993069648742676}
+02/26/2022 16:36:06 - INFO - codeparrot_training - Step 48659: {'lr': 9.622901292181963e-07, 'samples': 24913920, 'steps': 48659, 'loss/train': 1.3577886819839478}
+02/26/2022 16:36:09 - INFO - codeparrot_training - Step 48660: {'lr': 9.608564011440724e-07, 'samples': 24914432, 'steps': 48660, 'loss/train': 1.2054692506790161}
+02/26/2022 16:36:15 - INFO - codeparrot_training - Step 48661: {'lr': 9.594237398745375e-07, 'samples': 24914944, 'steps': 48661, 'loss/train': 2.3172712326049805}
+02/26/2022 16:36:18 - INFO - codeparrot_training - Step 48662: {'lr': 9.579921454157536e-07, 'samples': 24915456, 'steps': 48662, 'loss/train': 0.8853294253349304}
+02/26/2022 16:36:25 - INFO - codeparrot_training - Step 48663: {'lr': 9.565616177738545e-07, 'samples': 24915968, 'steps': 48663, 'loss/train': 2.126934289932251}
+02/26/2022 16:36:28 - INFO - codeparrot_training - Step 48664: {'lr': 9.551321569549743e-07, 'samples': 24916480, 'steps': 48664, 'loss/train': 1.4586352109909058}
+02/26/2022 16:36:34 - INFO - codeparrot_training - Step 48665: {'lr': 9.537037629651912e-07, 'samples': 24916992, 'steps': 48665, 'loss/train': 0.7465572953224182}
+02/26/2022 16:36:37 - INFO - codeparrot_training - Step 48666: {'lr': 9.52276435810695e-07, 'samples': 24917504, 'steps': 48666, 'loss/train': 2.002891778945923}
+02/26/2022 16:36:43 - INFO - codeparrot_training - Step 48667: {'lr': 9.508501754975363e-07, 'samples': 24918016, 'steps': 48667, 'loss/train': 1.3721352815628052}
+02/26/2022 16:36:47 - INFO - codeparrot_training - Step 48668: {'lr': 9.494249820318768e-07, 'samples': 24918528, 'steps': 48668, 'loss/train': 2.2423346042633057}
+02/26/2022 16:36:50 - INFO - codeparrot_training - Step 48669: {'lr': 9.480008554197949e-07, 'samples': 24919040, 'steps': 48669, 'loss/train': 2.2046985626220703}
+02/26/2022 16:36:56 - INFO - codeparrot_training - Step 48670: {'lr': 9.46577795667397e-07, 'samples': 24919552, 'steps': 48670, 'loss/train': 1.5289809703826904}
+02/26/2022 16:36:59 - INFO - codeparrot_training - Step 48671: {'lr': 9.451558027807617e-07, 'samples': 24920064, 'steps': 48671, 'loss/train': 1.7409809827804565}
+02/26/2022 16:37:05 - INFO - codeparrot_training - Step 48672: {'lr': 9.437348767659948e-07, 'samples': 24920576, 'steps': 48672, 'loss/train': 1.8067082166671753}
+02/26/2022 16:37:08 - INFO - codeparrot_training - Step 48673: {'lr': 9.423150176292028e-07, 'samples': 24921088, 'steps': 48673, 'loss/train': 1.1441411972045898}
+02/26/2022 16:37:15 - INFO - codeparrot_training - Step 48674: {'lr': 9.408962253764364e-07, 'samples': 24921600, 'steps': 48674, 'loss/train': 0.058101993054151535}
+02/26/2022 16:37:19 - INFO - codeparrot_training - Step 48675: {'lr': 9.394785000138018e-07, 'samples': 24922112, 'steps': 48675, 'loss/train': 1.3000520467758179}
+02/26/2022 16:37:24 - INFO - codeparrot_training - Step 48676: {'lr': 9.380618415473219e-07, 'samples': 24922624, 'steps': 48676, 'loss/train': 2.370760440826416}
+02/26/2022 16:37:28 - INFO - codeparrot_training - Step 48677: {'lr': 9.366462499831307e-07, 'samples': 24923136, 'steps': 48677, 'loss/train': 1.69815194606781}
+02/26/2022 16:37:33 - INFO - codeparrot_training - Step 48678: {'lr': 9.352317253272513e-07, 'samples': 24923648, 'steps': 48678, 'loss/train': 1.0883640050888062}
+02/26/2022 16:37:37 - INFO - codeparrot_training - Step 48679: {'lr': 9.338182675857343e-07, 'samples': 24924160, 'steps': 48679, 'loss/train': 1.762582540512085}
+02/26/2022 16:37:42 - INFO - codeparrot_training - Step 48680: {'lr': 9.324058767646859e-07, 'samples': 24924672, 'steps': 48680, 'loss/train': 1.7224711179733276}
+02/26/2022 16:37:46 - INFO - codeparrot_training - Step 48681: {'lr': 9.309945528701014e-07, 'samples': 24925184, 'steps': 48681, 'loss/train': 0.5291641354560852}
+02/26/2022 16:37:51 - INFO - codeparrot_training - Step 48682: {'lr': 9.295842959080592e-07, 'samples': 24925696, 'steps': 48682, 'loss/train': 1.3491417169570923}
+02/26/2022 16:37:55 - INFO - codeparrot_training - Step 48683: {'lr': 9.281751058845822e-07, 'samples': 24926208, 'steps': 48683, 'loss/train': 2.165400981903076}
+02/26/2022 16:38:00 - INFO - codeparrot_training - Step 48684: {'lr': 9.267669828056934e-07, 'samples': 24926720, 'steps': 48684, 'loss/train': 1.8449147939682007}
+02/26/2022 16:38:04 - INFO - codeparrot_training - Step 48685: {'lr': 9.253599266774715e-07, 'samples': 24927232, 'steps': 48685, 'loss/train': 1.1529393196105957}
+02/26/2022 16:38:09 - INFO - codeparrot_training - Step 48686: {'lr': 9.239539375059113e-07, 'samples': 24927744, 'steps': 48686, 'loss/train': 3.8546957969665527}
+02/26/2022 16:38:12 - INFO - codeparrot_training - Step 48687: {'lr': 9.225490152970361e-07, 'samples': 24928256, 'steps': 48687, 'loss/train': 1.87710702419281}
+02/26/2022 16:38:18 - INFO - codeparrot_training - Step 48688: {'lr': 9.211451600568966e-07, 'samples': 24928768, 'steps': 48688, 'loss/train': 1.5914483070373535}
+02/26/2022 16:38:21 - INFO - codeparrot_training - Step 48689: {'lr': 9.197423717914321e-07, 'samples': 24929280, 'steps': 48689, 'loss/train': 2.135009765625}
+02/26/2022 16:38:28 - INFO - codeparrot_training - Step 48690: {'lr': 9.183406505067494e-07, 'samples': 24929792, 'steps': 48690, 'loss/train': 1.0253850221633911}
+02/26/2022 16:38:31 - INFO - codeparrot_training - Step 48691: {'lr': 9.169399962087599e-07, 'samples': 24930304, 'steps': 48691, 'loss/train': 0.8639960885047913}
+02/26/2022 16:38:37 - INFO - codeparrot_training - Step 48692: {'lr': 9.155404089035424e-07, 'samples': 24930816, 'steps': 48692, 'loss/train': 1.39597749710083}
+02/26/2022 16:38:41 - INFO - codeparrot_training - Step 48693: {'lr': 9.141418885970365e-07, 'samples': 24931328, 'steps': 48693, 'loss/train': 1.156754732131958}
+02/26/2022 16:38:46 - INFO - codeparrot_training - Step 48694: {'lr': 9.127444352952375e-07, 'samples': 24931840, 'steps': 48694, 'loss/train': 0.9060541987419128}
+02/26/2022 16:38:49 - INFO - codeparrot_training - Step 48695: {'lr': 9.113480490041681e-07, 'samples': 24932352, 'steps': 48695, 'loss/train': 1.4301373958587646}
+02/26/2022 16:38:55 - INFO - codeparrot_training - Step 48696: {'lr': 9.099527297297961e-07, 'samples': 24932864, 'steps': 48696, 'loss/train': 3.0084667205810547}
+02/26/2022 16:38:58 - INFO - codeparrot_training - Step 48697: {'lr': 9.085584774780886e-07, 'samples': 24933376, 'steps': 48697, 'loss/train': 0.3221356272697449}
+02/26/2022 16:39:04 - INFO - codeparrot_training - Step 48698: {'lr': 9.071652922550133e-07, 'samples': 24933888, 'steps': 48698, 'loss/train': 0.17614074051380157}
+02/26/2022 16:39:07 - INFO - codeparrot_training - Step 48699: {'lr': 9.057731740665376e-07, 'samples': 24934400, 'steps': 48699, 'loss/train': 1.8399994373321533}
+02/26/2022 16:39:14 - INFO - codeparrot_training - Step 48700: {'lr': 9.043821229186566e-07, 'samples': 24934912, 'steps': 48700, 'loss/train': 0.6145563125610352}
+02/26/2022 16:39:17 - INFO - codeparrot_training - Step 48701: {'lr': 9.029921388172823e-07, 'samples': 24935424, 'steps': 48701, 'loss/train': 1.7535086870193481}
+02/26/2022 16:39:23 - INFO - codeparrot_training - Step 48702: {'lr': 9.0160322176841e-07, 'samples': 24935936, 'steps': 48702, 'loss/train': 1.829583764076233}
+02/26/2022 16:39:26 - INFO - codeparrot_training - Step 48703: {'lr': 9.002153717779515e-07, 'samples': 24936448, 'steps': 48703, 'loss/train': 1.8609371185302734}
+02/26/2022 16:39:32 - INFO - codeparrot_training - Step 48704: {'lr': 8.988285888519021e-07, 'samples': 24936960, 'steps': 48704, 'loss/train': 1.3831982612609863}
+02/26/2022 16:39:35 - INFO - codeparrot_training - Step 48705: {'lr': 8.97442872996146e-07, 'samples': 24937472, 'steps': 48705, 'loss/train': 2.1711933612823486}
+02/26/2022 16:39:41 - INFO - codeparrot_training - Step 48706: {'lr': 8.960582242166504e-07, 'samples': 24937984, 'steps': 48706, 'loss/train': 2.2444212436676025}
+02/26/2022 16:39:44 - INFO - codeparrot_training - Step 48707: {'lr': 8.946746425193553e-07, 'samples': 24938496, 'steps': 48707, 'loss/train': 1.9515472650527954}
+02/26/2022 16:39:50 - INFO - codeparrot_training - Step 48708: {'lr': 8.932921279101725e-07, 'samples': 24939008, 'steps': 48708, 'loss/train': 0.18177679181098938}
+02/26/2022 16:39:53 - INFO - codeparrot_training - Step 48709: {'lr': 8.919106803950139e-07, 'samples': 24939520, 'steps': 48709, 'loss/train': 1.665985345840454}
+02/26/2022 16:39:59 - INFO - codeparrot_training - Step 48710: {'lr': 8.905302999798192e-07, 'samples': 24940032, 'steps': 48710, 'loss/train': 0.06281256675720215}
+02/26/2022 16:40:03 - INFO - codeparrot_training - Step 48711: {'lr': 8.891509866705006e-07, 'samples': 24940544, 'steps': 48711, 'loss/train': 1.3541584014892578}
+02/26/2022 16:40:08 - INFO - codeparrot_training - Step 48712: {'lr': 8.877727404729419e-07, 'samples': 24941056, 'steps': 48712, 'loss/train': 2.0485923290252686}
+02/26/2022 16:40:12 - INFO - codeparrot_training - Step 48713: {'lr': 8.863955613930553e-07, 'samples': 24941568, 'steps': 48713, 'loss/train': 2.7983922958374023}
+02/26/2022 16:40:17 - INFO - codeparrot_training - Step 48714: {'lr': 8.850194494367525e-07, 'samples': 24942080, 'steps': 48714, 'loss/train': 0.6717244386672974}
+02/26/2022 16:40:21 - INFO - codeparrot_training - Step 48715: {'lr': 8.836444046099457e-07, 'samples': 24942592, 'steps': 48715, 'loss/train': 0.8583987355232239}
+02/26/2022 16:40:26 - INFO - codeparrot_training - Step 48716: {'lr': 8.822704269184633e-07, 'samples': 24943104, 'steps': 48716, 'loss/train': 2.296792507171631}
+02/26/2022 16:40:30 - INFO - codeparrot_training - Step 48717: {'lr': 8.808975163682731e-07, 'samples': 24943616, 'steps': 48717, 'loss/train': 1.149545431137085}
+02/26/2022 16:40:35 - INFO - codeparrot_training - Step 48718: {'lr': 8.795256729651757e-07, 'samples': 24944128, 'steps': 48718, 'loss/train': 1.2793878316879272}
+02/26/2022 16:40:39 - INFO - codeparrot_training - Step 48719: {'lr': 8.781548967151109e-07, 'samples': 24944640, 'steps': 48719, 'loss/train': 1.8845187425613403}
+02/26/2022 16:40:45 - INFO - codeparrot_training - Step 48720: {'lr': 8.767851876239075e-07, 'samples': 24945152, 'steps': 48720, 'loss/train': 1.430665373802185}
+02/26/2022 16:40:48 - INFO - codeparrot_training - Step 48721: {'lr': 8.754165456974772e-07, 'samples': 24945664, 'steps': 48721, 'loss/train': 1.5877289772033691}
+02/26/2022 16:40:54 - INFO - codeparrot_training - Step 48722: {'lr': 8.740489709416211e-07, 'samples': 24946176, 'steps': 48722, 'loss/train': 1.6608823537826538}
+02/26/2022 16:40:57 - INFO - codeparrot_training - Step 48723: {'lr': 8.726824633622788e-07, 'samples': 24946688, 'steps': 48723, 'loss/train': 1.69318425655365}
+02/26/2022 16:41:03 - INFO - codeparrot_training - Step 48724: {'lr': 8.713170229652234e-07, 'samples': 24947200, 'steps': 48724, 'loss/train': 1.438439130783081}
+02/26/2022 16:41:07 - INFO - codeparrot_training - Step 48725: {'lr': 8.699526497563392e-07, 'samples': 24947712, 'steps': 48725, 'loss/train': 1.9111677408218384}
+02/26/2022 16:41:13 - INFO - codeparrot_training - Step 48726: {'lr': 8.685893437414827e-07, 'samples': 24948224, 'steps': 48726, 'loss/train': 2.1296932697296143}
+02/26/2022 16:41:16 - INFO - codeparrot_training - Step 48727: {'lr': 8.672271049264824e-07, 'samples': 24948736, 'steps': 48727, 'loss/train': 1.0653467178344727}
+02/26/2022 16:41:22 - INFO - codeparrot_training - Step 48728: {'lr': 8.658659333171392e-07, 'samples': 24949248, 'steps': 48728, 'loss/train': 1.6033753156661987}
+02/26/2022 16:41:25 - INFO - codeparrot_training - Step 48729: {'lr': 8.645058289193652e-07, 'samples': 24949760, 'steps': 48729, 'loss/train': 0.7593342661857605}
+02/26/2022 16:41:31 - INFO - codeparrot_training - Step 48730: {'lr': 8.631467917389058e-07, 'samples': 24950272, 'steps': 48730, 'loss/train': 1.277632713317871}
+02/26/2022 16:41:34 - INFO - codeparrot_training - Step 48731: {'lr': 8.61788821781645e-07, 'samples': 24950784, 'steps': 48731, 'loss/train': 1.8686374425888062}
+02/26/2022 16:41:40 - INFO - codeparrot_training - Step 48732: {'lr': 8.604319190533283e-07, 'samples': 24951296, 'steps': 48732, 'loss/train': 1.7932742834091187}
+02/26/2022 16:41:43 - INFO - codeparrot_training - Step 48733: {'lr': 8.590760835598399e-07, 'samples': 24951808, 'steps': 48733, 'loss/train': 1.4419950246810913}
+02/26/2022 16:41:49 - INFO - codeparrot_training - Step 48734: {'lr': 8.577213153069807e-07, 'samples': 24952320, 'steps': 48734, 'loss/train': 1.365775465965271}
+02/26/2022 16:41:52 - INFO - codeparrot_training - Step 48735: {'lr': 8.563676143004962e-07, 'samples': 24952832, 'steps': 48735, 'loss/train': 0.5562288165092468}
+02/26/2022 16:41:59 - INFO - codeparrot_training - Step 48736: {'lr': 8.550149805462148e-07, 'samples': 24953344, 'steps': 48736, 'loss/train': 2.045912504196167}
+02/26/2022 16:42:02 - INFO - codeparrot_training - Step 48737: {'lr': 8.536634140499656e-07, 'samples': 24953856, 'steps': 48737, 'loss/train': 1.2927608489990234}
+02/26/2022 16:42:08 - INFO - codeparrot_training - Step 48738: {'lr': 8.523129148174935e-07, 'samples': 24954368, 'steps': 48738, 'loss/train': 0.04306618869304657}
+02/26/2022 16:42:11 - INFO - codeparrot_training - Step 48739: {'lr': 8.509634828545998e-07, 'samples': 24954880, 'steps': 48739, 'loss/train': 2.117922306060791}
+02/26/2022 16:42:17 - INFO - codeparrot_training - Step 48740: {'lr': 8.496151181670853e-07, 'samples': 24955392, 'steps': 48740, 'loss/train': 1.6562367677688599}
+02/26/2022 16:42:20 - INFO - codeparrot_training - Step 48741: {'lr': 8.482678207606675e-07, 'samples': 24955904, 'steps': 48741, 'loss/train': 0.9660260081291199}
+02/26/2022 16:42:26 - INFO - codeparrot_training - Step 48742: {'lr': 8.469215906411753e-07, 'samples': 24956416, 'steps': 48742, 'loss/train': 1.2809741497039795}
+02/26/2022 16:42:29 - INFO - codeparrot_training - Step 48743: {'lr': 8.45576427814354e-07, 'samples': 24956928, 'steps': 48743, 'loss/train': 1.1946102380752563}
+02/26/2022 16:42:35 - INFO - codeparrot_training - Step 48744: {'lr': 8.442323322859768e-07, 'samples': 24957440, 'steps': 48744, 'loss/train': 2.6889636516571045}
+02/26/2022 16:42:38 - INFO - codeparrot_training - Step 48745: {'lr': 8.428893040617891e-07, 'samples': 24957952, 'steps': 48745, 'loss/train': 0.8061151504516602}
+02/26/2022 16:42:44 - INFO - codeparrot_training - Step 48746: {'lr': 8.415473431475362e-07, 'samples': 24958464, 'steps': 48746, 'loss/train': 0.03204713016748428}
+02/26/2022 16:42:48 - INFO - codeparrot_training - Step 48747: {'lr': 8.402064495489914e-07, 'samples': 24958976, 'steps': 48747, 'loss/train': 2.025420904159546}
+02/26/2022 16:42:53 - INFO - codeparrot_training - Step 48748: {'lr': 8.388666232718722e-07, 'samples': 24959488, 'steps': 48748, 'loss/train': 2.071467638015747}
+02/26/2022 16:42:57 - INFO - codeparrot_training - Step 48749: {'lr': 8.375278643219519e-07, 'samples': 24960000, 'steps': 48749, 'loss/train': 1.2462078332901}
+02/26/2022 16:43:02 - INFO - codeparrot_training - Step 48750: {'lr': 8.361901727049204e-07, 'samples': 24960512, 'steps': 48750, 'loss/train': 1.4476274251937866}
+02/26/2022 16:43:06 - INFO - codeparrot_training - Step 48751: {'lr': 8.348535484265508e-07, 'samples': 24961024, 'steps': 48751, 'loss/train': 2.1303017139434814}
+02/26/2022 16:43:11 - INFO - codeparrot_training - Step 48752: {'lr': 8.335179914925328e-07, 'samples': 24961536, 'steps': 48752, 'loss/train': 2.0322623252868652}
+02/26/2022 16:43:15 - INFO - codeparrot_training - Step 48753: {'lr': 8.321835019086122e-07, 'samples': 24962048, 'steps': 48753, 'loss/train': 2.335151195526123}
+02/26/2022 16:43:20 - INFO - codeparrot_training - Step 48754: {'lr': 8.308500796805063e-07, 'samples': 24962560, 'steps': 48754, 'loss/train': 1.8535524606704712}
+02/26/2022 16:43:24 - INFO - codeparrot_training - Step 48755: {'lr': 8.295177248139052e-07, 'samples': 24963072, 'steps': 48755, 'loss/train': 2.127371072769165}
+02/26/2022 16:43:30 - INFO - codeparrot_training - Step 48756: {'lr': 8.281864373145265e-07, 'samples': 24963584, 'steps': 48756, 'loss/train': 2.833411455154419}
+02/26/2022 16:43:33 - INFO - codeparrot_training - Step 48757: {'lr': 8.268562171880878e-07, 'samples': 24964096, 'steps': 48757, 'loss/train': 1.9080238342285156}
+02/26/2022 16:43:39 - INFO - codeparrot_training - Step 48758: {'lr': 8.25527064440279e-07, 'samples': 24964608, 'steps': 48758, 'loss/train': 1.6851392984390259}
+02/26/2022 16:43:42 - INFO - codeparrot_training - Step 48759: {'lr': 8.241989790767623e-07, 'samples': 24965120, 'steps': 48759, 'loss/train': 2.0645151138305664}
+02/26/2022 16:43:48 - INFO - codeparrot_training - Step 48760: {'lr': 8.228719611032831e-07, 'samples': 24965632, 'steps': 48760, 'loss/train': 1.9589732885360718}
+02/26/2022 16:43:51 - INFO - codeparrot_training - Step 48761: {'lr': 8.215460105255035e-07, 'samples': 24966144, 'steps': 48761, 'loss/train': 2.447303056716919}
+02/26/2022 16:43:57 - INFO - codeparrot_training - Step 48762: {'lr': 8.202211273490857e-07, 'samples': 24966656, 'steps': 48762, 'loss/train': 2.395120143890381}
+02/26/2022 16:44:01 - INFO - codeparrot_training - Step 48763: {'lr': 8.188973115796916e-07, 'samples': 24967168, 'steps': 48763, 'loss/train': 1.0769151449203491}
+02/26/2022 16:44:07 - INFO - codeparrot_training - Step 48764: {'lr': 8.175745632230669e-07, 'samples': 24967680, 'steps': 48764, 'loss/train': 1.6376525163650513}
+02/26/2022 16:44:10 - INFO - codeparrot_training - Step 48765: {'lr': 8.162528822847904e-07, 'samples': 24968192, 'steps': 48765, 'loss/train': 1.5445570945739746}
+02/26/2022 16:44:14 - INFO - codeparrot_training - Step 48766: {'lr': 8.149322687705796e-07, 'samples': 24968704, 'steps': 48766, 'loss/train': 1.4840246438980103}
+02/26/2022 16:44:19 - INFO - codeparrot_training - Step 48767: {'lr': 8.13612722686069e-07, 'samples': 24969216, 'steps': 48767, 'loss/train': 0.3326415717601776}
+02/26/2022 16:44:23 - INFO - codeparrot_training - Step 48768: {'lr': 8.122942440369208e-07, 'samples': 24969728, 'steps': 48768, 'loss/train': 1.6522136926651}
+02/26/2022 16:44:28 - INFO - codeparrot_training - Step 48769: {'lr': 8.109768328287692e-07, 'samples': 24970240, 'steps': 48769, 'loss/train': 1.6850029230117798}
+02/26/2022 16:44:32 - INFO - codeparrot_training - Step 48770: {'lr': 8.096604890672765e-07, 'samples': 24970752, 'steps': 48770, 'loss/train': 2.3310251235961914}
+02/26/2022 16:44:37 - INFO - codeparrot_training - Step 48771: {'lr': 8.083452127580493e-07, 'samples': 24971264, 'steps': 48771, 'loss/train': 0.7881247997283936}
+02/26/2022 16:44:41 - INFO - codeparrot_training - Step 48772: {'lr': 8.070310039067774e-07, 'samples': 24971776, 'steps': 48772, 'loss/train': 1.2660707235336304}
+02/26/2022 16:44:47 - INFO - codeparrot_training - Step 48773: {'lr': 8.057178625190397e-07, 'samples': 24972288, 'steps': 48773, 'loss/train': 2.4484686851501465}
+02/26/2022 16:44:50 - INFO - codeparrot_training - Step 48774: {'lr': 8.044057886004708e-07, 'samples': 24972800, 'steps': 48774, 'loss/train': 1.5282375812530518}
+02/26/2022 16:44:56 - INFO - codeparrot_training - Step 48775: {'lr': 8.030947821567047e-07, 'samples': 24973312, 'steps': 48775, 'loss/train': 2.041247844696045}
+02/26/2022 16:45:00 - INFO - codeparrot_training - Step 48776: {'lr': 8.017848431933484e-07, 'samples': 24973824, 'steps': 48776, 'loss/train': 1.362694263458252}
+02/26/2022 16:45:05 - INFO - codeparrot_training - Step 48777: {'lr': 8.004759717160359e-07, 'samples': 24974336, 'steps': 48777, 'loss/train': 1.7827330827713013}
+02/26/2022 16:45:08 - INFO - codeparrot_training - Step 48778: {'lr': 7.991681677303186e-07, 'samples': 24974848, 'steps': 48778, 'loss/train': 1.5937379598617554}
+02/26/2022 16:45:14 - INFO - codeparrot_training - Step 48779: {'lr': 7.978614312418586e-07, 'samples': 24975360, 'steps': 48779, 'loss/train': 1.102156639099121}
+02/26/2022 16:45:17 - INFO - codeparrot_training - Step 48780: {'lr': 7.965557622562347e-07, 'samples': 24975872, 'steps': 48780, 'loss/train': 1.416162133216858}
+02/26/2022 16:45:23 - INFO - codeparrot_training - Step 48781: {'lr': 7.952511607790258e-07, 'samples': 24976384, 'steps': 48781, 'loss/train': 1.525479793548584}
+02/26/2022 16:45:26 - INFO - codeparrot_training - Step 48782: {'lr': 7.939476268158385e-07, 'samples': 24976896, 'steps': 48782, 'loss/train': 1.8329294919967651}
+02/26/2022 16:45:33 - INFO - codeparrot_training - Step 48783: {'lr': 7.92645160372224e-07, 'samples': 24977408, 'steps': 48783, 'loss/train': 1.46579110622406}
+02/26/2022 16:45:36 - INFO - codeparrot_training - Step 48784: {'lr': 7.913437614538166e-07, 'samples': 24977920, 'steps': 48784, 'loss/train': 1.6328544616699219}
+02/26/2022 16:45:42 - INFO - codeparrot_training - Step 48785: {'lr': 7.900434300661397e-07, 'samples': 24978432, 'steps': 48785, 'loss/train': 1.7565546035766602}
+02/26/2022 16:45:45 - INFO - codeparrot_training - Step 48786: {'lr': 7.887441662147998e-07, 'samples': 24978944, 'steps': 48786, 'loss/train': 1.3826297521591187}
+02/26/2022 16:45:51 - INFO - codeparrot_training - Step 48787: {'lr': 7.874459699053204e-07, 'samples': 24979456, 'steps': 48787, 'loss/train': 0.5662800669670105}
+02/26/2022 16:45:54 - INFO - codeparrot_training - Step 48788: {'lr': 7.861488411433082e-07, 'samples': 24979968, 'steps': 48788, 'loss/train': 1.8822400569915771}
+02/26/2022 16:46:00 - INFO - codeparrot_training - Step 48789: {'lr': 7.848527799343141e-07, 'samples': 24980480, 'steps': 48789, 'loss/train': 1.7091679573059082}
+02/26/2022 16:46:03 - INFO - codeparrot_training - Step 48790: {'lr': 7.835577862838339e-07, 'samples': 24980992, 'steps': 48790, 'loss/train': 1.2195299863815308}
+02/26/2022 16:46:09 - INFO - codeparrot_training - Step 48791: {'lr': 7.822638601975018e-07, 'samples': 24981504, 'steps': 48791, 'loss/train': 1.2716575860977173}
+02/26/2022 16:46:12 - INFO - codeparrot_training - Step 48792: {'lr': 7.809710016807859e-07, 'samples': 24982016, 'steps': 48792, 'loss/train': 2.415796995162964}
+02/26/2022 16:46:18 - INFO - codeparrot_training - Step 48793: {'lr': 7.796792107392647e-07, 'samples': 24982528, 'steps': 48793, 'loss/train': 2.1237080097198486}
+02/26/2022 16:46:22 - INFO - codeparrot_training - Step 48794: {'lr': 7.783884873784619e-07, 'samples': 24983040, 'steps': 48794, 'loss/train': 2.117501974105835}
+02/26/2022 16:46:27 - INFO - codeparrot_training - Step 48795: {'lr': 7.770988316039008e-07, 'samples': 24983552, 'steps': 48795, 'loss/train': 2.6740622520446777}
+02/26/2022 16:46:31 - INFO - codeparrot_training - Step 48796: {'lr': 7.758102434211045e-07, 'samples': 24984064, 'steps': 48796, 'loss/train': 3.1281399726867676}
+02/26/2022 16:46:36 - INFO - codeparrot_training - Step 48797: {'lr': 7.745227228356244e-07, 'samples': 24984576, 'steps': 48797, 'loss/train': 1.4416662454605103}
+02/26/2022 16:46:40 - INFO - codeparrot_training - Step 48798: {'lr': 7.732362698529283e-07, 'samples': 24985088, 'steps': 48798, 'loss/train': 1.584259033203125}
+02/26/2022 16:46:45 - INFO - codeparrot_training - Step 48799: {'lr': 7.719508844785672e-07, 'samples': 24985600, 'steps': 48799, 'loss/train': 1.4350366592407227}
+02/26/2022 16:46:49 - INFO - codeparrot_training - Step 48800: {'lr': 7.70666566718009e-07, 'samples': 24986112, 'steps': 48800, 'loss/train': 1.67714262008667}
+02/26/2022 16:46:54 - INFO - codeparrot_training - Step 48801: {'lr': 7.693833165767772e-07, 'samples': 24986624, 'steps': 48801, 'loss/train': 1.9618850946426392}
+02/26/2022 16:46:58 - INFO - codeparrot_training - Step 48802: {'lr': 7.681011340603949e-07, 'samples': 24987136, 'steps': 48802, 'loss/train': 0.9408183693885803}
+02/26/2022 16:47:03 - INFO - codeparrot_training - Step 48803: {'lr': 7.668200191743024e-07, 'samples': 24987648, 'steps': 48803, 'loss/train': 1.1661893129348755}
+02/26/2022 16:47:07 - INFO - codeparrot_training - Step 48804: {'lr': 7.65539971924023e-07, 'samples': 24988160, 'steps': 48804, 'loss/train': 1.5404794216156006}
+02/26/2022 16:47:13 - INFO - codeparrot_training - Step 48805: {'lr': 7.642609923150246e-07, 'samples': 24988672, 'steps': 48805, 'loss/train': 1.5036308765411377}
+02/26/2022 16:47:16 - INFO - codeparrot_training - Step 48806: {'lr': 7.629830803527749e-07, 'samples': 24989184, 'steps': 48806, 'loss/train': 1.780240535736084}
+02/26/2022 16:47:22 - INFO - codeparrot_training - Step 48807: {'lr': 7.617062360427973e-07, 'samples': 24989696, 'steps': 48807, 'loss/train': 1.278171420097351}
+02/26/2022 16:47:25 - INFO - codeparrot_training - Step 48808: {'lr': 7.604304593905042e-07, 'samples': 24990208, 'steps': 48808, 'loss/train': 1.5116682052612305}
+02/26/2022 16:47:31 - INFO - codeparrot_training - Step 48809: {'lr': 7.591557504013913e-07, 'samples': 24990720, 'steps': 48809, 'loss/train': 1.6538307666778564}
+02/26/2022 16:47:35 - INFO - codeparrot_training - Step 48810: {'lr': 7.578821090809263e-07, 'samples': 24991232, 'steps': 48810, 'loss/train': 1.1046043634414673}
+02/26/2022 16:47:40 - INFO - codeparrot_training - Step 48811: {'lr': 7.566095354345215e-07, 'samples': 24991744, 'steps': 48811, 'loss/train': 0.9560911059379578}
+02/26/2022 16:47:44 - INFO - codeparrot_training - Step 48812: {'lr': 7.553380294676726e-07, 'samples': 24992256, 'steps': 48812, 'loss/train': 1.5707350969314575}
+02/26/2022 16:47:49 - INFO - codeparrot_training - Step 48813: {'lr': 7.540675911858197e-07, 'samples': 24992768, 'steps': 48813, 'loss/train': 1.8504838943481445}
+02/26/2022 16:47:53 - INFO - codeparrot_training - Step 48814: {'lr': 7.527982205944028e-07, 'samples': 24993280, 'steps': 48814, 'loss/train': 1.5718964338302612}
+02/26/2022 16:47:59 - INFO - codeparrot_training - Step 48815: {'lr': 7.515299176988344e-07, 'samples': 24993792, 'steps': 48815, 'loss/train': 1.8446481227874756}
+02/26/2022 16:48:02 - INFO - codeparrot_training - Step 48816: {'lr': 7.502626825045544e-07, 'samples': 24994304, 'steps': 48816, 'loss/train': 2.2348520755767822}
+02/26/2022 16:48:08 - INFO - codeparrot_training - Step 48817: {'lr': 7.489965150170586e-07, 'samples': 24994816, 'steps': 48817, 'loss/train': 1.857997179031372}
+02/26/2022 16:48:11 - INFO - codeparrot_training - Step 48818: {'lr': 7.477314152416481e-07, 'samples': 24995328, 'steps': 48818, 'loss/train': 2.0076744556427}
+02/26/2022 16:48:17 - INFO - codeparrot_training - Step 48819: {'lr': 7.464673831838464e-07, 'samples': 24995840, 'steps': 48819, 'loss/train': 1.8949207067489624}
+02/26/2022 16:48:20 - INFO - codeparrot_training - Step 48820: {'lr': 7.452044188490381e-07, 'samples': 24996352, 'steps': 48820, 'loss/train': 2.1093645095825195}
+02/26/2022 16:48:26 - INFO - codeparrot_training - Step 48821: {'lr': 7.439425222426077e-07, 'samples': 24996864, 'steps': 48821, 'loss/train': 1.644471526145935}
+02/26/2022 16:48:29 - INFO - codeparrot_training - Step 48822: {'lr': 7.426816933699954e-07, 'samples': 24997376, 'steps': 48822, 'loss/train': 1.4608527421951294}
+02/26/2022 16:48:35 - INFO - codeparrot_training - Step 48823: {'lr': 7.414219322365578e-07, 'samples': 24997888, 'steps': 48823, 'loss/train': 2.208585262298584}
+02/26/2022 16:48:38 - INFO - codeparrot_training - Step 48824: {'lr': 7.401632388477631e-07, 'samples': 24998400, 'steps': 48824, 'loss/train': 2.027339220046997}
+02/26/2022 16:48:44 - INFO - codeparrot_training - Step 48825: {'lr': 7.3890561320894e-07, 'samples': 24998912, 'steps': 48825, 'loss/train': 1.1359057426452637}
+02/26/2022 16:48:47 - INFO - codeparrot_training - Step 48826: {'lr': 7.376490553254734e-07, 'samples': 24999424, 'steps': 48826, 'loss/train': 0.6023389101028442}
+02/26/2022 16:48:53 - INFO - codeparrot_training - Step 48827: {'lr': 7.363935652028031e-07, 'samples': 24999936, 'steps': 48827, 'loss/train': 0.6397705078125}
+02/26/2022 16:48:56 - INFO - codeparrot_training - Step 48828: {'lr': 7.351391428462584e-07, 'samples': 25000448, 'steps': 48828, 'loss/train': 0.9860379099845886}
+02/26/2022 16:49:03 - INFO - codeparrot_training - Step 48829: {'lr': 7.338857882612238e-07, 'samples': 25000960, 'steps': 48829, 'loss/train': 0.6927613019943237}
+02/26/2022 16:49:06 - INFO - codeparrot_training - Step 48830: {'lr': 7.326335014530561e-07, 'samples': 25001472, 'steps': 48830, 'loss/train': 1.9804426431655884}
+02/26/2022 16:49:12 - INFO - codeparrot_training - Step 48831: {'lr': 7.313822824271676e-07, 'samples': 25001984, 'steps': 48831, 'loss/train': 1.6332658529281616}
+02/26/2022 16:49:15 - INFO - codeparrot_training - Step 48832: {'lr': 7.301321311888876e-07, 'samples': 25002496, 'steps': 48832, 'loss/train': 1.8073056936264038}
+02/26/2022 16:49:21 - INFO - codeparrot_training - Step 48833: {'lr': 7.288830477435448e-07, 'samples': 25003008, 'steps': 48833, 'loss/train': 1.422527551651001}
+02/26/2022 16:49:24 - INFO - codeparrot_training - Step 48834: {'lr': 7.276350320965241e-07, 'samples': 25003520, 'steps': 48834, 'loss/train': 2.2832090854644775}
+02/26/2022 16:49:30 - INFO - codeparrot_training - Step 48835: {'lr': 7.263880842531545e-07, 'samples': 25004032, 'steps': 48835, 'loss/train': 0.46577489376068115}
+02/26/2022 16:49:33 - INFO - codeparrot_training - Step 48836: {'lr': 7.251422042187927e-07, 'samples': 25004544, 'steps': 48836, 'loss/train': 1.4138360023498535}
+02/26/2022 16:49:39 - INFO - codeparrot_training - Step 48837: {'lr': 7.238973919987679e-07, 'samples': 25005056, 'steps': 48837, 'loss/train': 0.9904511570930481}
+02/26/2022 16:49:42 - INFO - codeparrot_training - Step 48838: {'lr': 7.226536475984092e-07, 'samples': 25005568, 'steps': 48838, 'loss/train': 2.4837419986724854}
+02/26/2022 16:49:49 - INFO - codeparrot_training - Step 48839: {'lr': 7.214109710230454e-07, 'samples': 25006080, 'steps': 48839, 'loss/train': 1.3925635814666748}
+02/26/2022 16:49:52 - INFO - codeparrot_training - Step 48840: {'lr': 7.201693622780059e-07, 'samples': 25006592, 'steps': 48840, 'loss/train': 1.4889143705368042}
+02/26/2022 16:49:58 - INFO - codeparrot_training - Step 48841: {'lr': 7.189288213686196e-07, 'samples': 25007104, 'steps': 48841, 'loss/train': 2.4012787342071533}
+02/26/2022 16:50:01 - INFO - codeparrot_training - Step 48842: {'lr': 7.176893483001601e-07, 'samples': 25007616, 'steps': 48842, 'loss/train': 1.550831913948059}
+02/26/2022 16:50:07 - INFO - codeparrot_training - Step 48843: {'lr': 7.164509430779842e-07, 'samples': 25008128, 'steps': 48843, 'loss/train': 0.19090385735034943}
+02/26/2022 16:50:10 - INFO - codeparrot_training - Step 48844: {'lr': 7.152136057073655e-07, 'samples': 25008640, 'steps': 48844, 'loss/train': 2.9373362064361572}
+02/26/2022 16:50:16 - INFO - codeparrot_training - Step 48845: {'lr': 7.139773361936053e-07, 'samples': 25009152, 'steps': 48845, 'loss/train': 2.0581512451171875}
+02/26/2022 16:50:19 - INFO - codeparrot_training - Step 48846: {'lr': 7.127421345420326e-07, 'samples': 25009664, 'steps': 48846, 'loss/train': 1.744335412979126}
+02/26/2022 16:50:25 - INFO - codeparrot_training - Step 48847: {'lr': 7.11508000757921e-07, 'samples': 25010176, 'steps': 48847, 'loss/train': 2.596433639526367}
+02/26/2022 16:50:28 - INFO - codeparrot_training - Step 48848: {'lr': 7.102749348465165e-07, 'samples': 25010688, 'steps': 48848, 'loss/train': 1.710936427116394}
+02/26/2022 16:50:34 - INFO - codeparrot_training - Step 48849: {'lr': 7.090429368131479e-07, 'samples': 25011200, 'steps': 48849, 'loss/train': 2.4011363983154297}
+02/26/2022 16:50:37 - INFO - codeparrot_training - Step 48850: {'lr': 7.078120066630889e-07, 'samples': 25011712, 'steps': 48850, 'loss/train': 1.4815024137496948}
+02/26/2022 16:50:44 - INFO - codeparrot_training - Step 48851: {'lr': 7.065821444016129e-07, 'samples': 25012224, 'steps': 48851, 'loss/train': 2.5250051021575928}
+02/26/2022 16:50:47 - INFO - codeparrot_training - Step 48852: {'lr': 7.053533500339659e-07, 'samples': 25012736, 'steps': 48852, 'loss/train': 2.3422412872314453}
+02/26/2022 16:50:52 - INFO - codeparrot_training - Step 48853: {'lr': 7.041256235654214e-07, 'samples': 25013248, 'steps': 48853, 'loss/train': 2.815161943435669}
+02/26/2022 16:50:56 - INFO - codeparrot_training - Step 48854: {'lr': 7.028989650012529e-07, 'samples': 25013760, 'steps': 48854, 'loss/train': 1.9013689756393433}
+02/26/2022 16:51:01 - INFO - codeparrot_training - Step 48855: {'lr': 7.016733743466786e-07, 'samples': 25014272, 'steps': 48855, 'loss/train': 1.5986576080322266}
+02/26/2022 16:51:05 - INFO - codeparrot_training - Step 48856: {'lr': 7.004488516070273e-07, 'samples': 25014784, 'steps': 48856, 'loss/train': 2.037203788757324}
+02/26/2022 16:51:10 - INFO - codeparrot_training - Step 48857: {'lr': 6.99225396787434e-07, 'samples': 25015296, 'steps': 48857, 'loss/train': 1.5300078392028809}
+02/26/2022 16:51:14 - INFO - codeparrot_training - Step 48858: {'lr': 6.980030098932277e-07, 'samples': 25015808, 'steps': 48858, 'loss/train': 1.9404606819152832}
+02/26/2022 16:51:20 - INFO - codeparrot_training - Step 48859: {'lr': 6.967816909295988e-07, 'samples': 25016320, 'steps': 48859, 'loss/train': 0.05252516269683838}
+02/26/2022 16:51:23 - INFO - codeparrot_training - Step 48860: {'lr': 6.955614399018207e-07, 'samples': 25016832, 'steps': 48860, 'loss/train': 2.270653486251831}
+02/26/2022 16:51:29 - INFO - codeparrot_training - Step 48861: {'lr': 6.943422568150836e-07, 'samples': 25017344, 'steps': 48861, 'loss/train': 1.713470697402954}
+02/26/2022 16:51:33 - INFO - codeparrot_training - Step 48862: {'lr': 6.931241416746059e-07, 'samples': 25017856, 'steps': 48862, 'loss/train': 1.769823431968689}
+02/26/2022 16:51:38 - INFO - codeparrot_training - Step 48863: {'lr': 6.919070944856332e-07, 'samples': 25018368, 'steps': 48863, 'loss/train': 0.2431567758321762}
+02/26/2022 16:51:42 - INFO - codeparrot_training - Step 48864: {'lr': 6.906911152533557e-07, 'samples': 25018880, 'steps': 48864, 'loss/train': 1.2163974046707153}
+02/26/2022 16:51:47 - INFO - codeparrot_training - Step 48865: {'lr': 6.894762039830193e-07, 'samples': 25019392, 'steps': 48865, 'loss/train': 1.2048203945159912}
+02/26/2022 16:51:51 - INFO - codeparrot_training - Step 48866: {'lr': 6.882623606797867e-07, 'samples': 25019904, 'steps': 48866, 'loss/train': 2.3501455783843994}
+02/26/2022 16:51:56 - INFO - codeparrot_training - Step 48867: {'lr': 6.870495853488756e-07, 'samples': 25020416, 'steps': 48867, 'loss/train': 1.6358106136322021}
+02/26/2022 16:52:00 - INFO - codeparrot_training - Step 48868: {'lr': 6.858378779954488e-07, 'samples': 25020928, 'steps': 48868, 'loss/train': 1.8266350030899048}
+02/26/2022 16:52:05 - INFO - codeparrot_training - Step 48869: {'lr': 6.84627238624752e-07, 'samples': 25021440, 'steps': 48869, 'loss/train': 1.988757610321045}
+02/26/2022 16:52:09 - INFO - codeparrot_training - Step 48870: {'lr': 6.834176672419479e-07, 'samples': 25021952, 'steps': 48870, 'loss/train': 2.291236639022827}
+02/26/2022 16:52:15 - INFO - codeparrot_training - Step 48871: {'lr': 6.822091638522265e-07, 'samples': 25022464, 'steps': 48871, 'loss/train': 1.7195982933044434}
+02/26/2022 16:52:19 - INFO - codeparrot_training - Step 48872: {'lr': 6.810017284607229e-07, 'samples': 25022976, 'steps': 48872, 'loss/train': 2.53993558883667}
+02/26/2022 16:52:24 - INFO - codeparrot_training - Step 48873: {'lr': 6.797953610726548e-07, 'samples': 25023488, 'steps': 48873, 'loss/train': 0.134123295545578}
+02/26/2022 16:52:27 - INFO - codeparrot_training - Step 48874: {'lr': 6.785900616931851e-07, 'samples': 25024000, 'steps': 48874, 'loss/train': 1.010966420173645}
+02/26/2022 16:52:33 - INFO - codeparrot_training - Step 48875: {'lr': 6.773858303274482e-07, 'samples': 25024512, 'steps': 48875, 'loss/train': 1.0960313081741333}
+02/26/2022 16:52:36 - INFO - codeparrot_training - Step 48876: {'lr': 6.761826669806348e-07, 'samples': 25025024, 'steps': 48876, 'loss/train': 1.8134791851043701}
+02/26/2022 16:52:42 - INFO - codeparrot_training - Step 48877: {'lr': 6.749805716578794e-07, 'samples': 25025536, 'steps': 48877, 'loss/train': 2.035311222076416}
+02/26/2022 16:52:45 - INFO - codeparrot_training - Step 48878: {'lr': 6.737795443643446e-07, 'samples': 25026048, 'steps': 48878, 'loss/train': 2.643656015396118}
+02/26/2022 16:52:51 - INFO - codeparrot_training - Step 48879: {'lr': 6.725795851051653e-07, 'samples': 25026560, 'steps': 48879, 'loss/train': 1.8357720375061035}
+02/26/2022 16:52:54 - INFO - codeparrot_training - Step 48880: {'lr': 6.713806938854761e-07, 'samples': 25027072, 'steps': 48880, 'loss/train': 1.4059455394744873}
+02/26/2022 16:53:00 - INFO - codeparrot_training - Step 48881: {'lr': 6.701828707104396e-07, 'samples': 25027584, 'steps': 48881, 'loss/train': 2.1597094535827637}
+02/26/2022 16:53:03 - INFO - codeparrot_training - Step 48882: {'lr': 6.689861155851351e-07, 'samples': 25028096, 'steps': 48882, 'loss/train': 0.33912619948387146}
+02/26/2022 16:53:09 - INFO - codeparrot_training - Step 48883: {'lr': 6.677904285147529e-07, 'samples': 25028608, 'steps': 48883, 'loss/train': 1.812813401222229}
+02/26/2022 16:53:12 - INFO - codeparrot_training - Step 48884: {'lr': 6.665958095043723e-07, 'samples': 25029120, 'steps': 48884, 'loss/train': 1.9970989227294922}
+02/26/2022 16:53:18 - INFO - codeparrot_training - Step 48885: {'lr': 6.65402258559128e-07, 'samples': 25029632, 'steps': 48885, 'loss/train': 1.2113577127456665}
+02/26/2022 16:53:21 - INFO - codeparrot_training - Step 48886: {'lr': 6.64209775684127e-07, 'samples': 25030144, 'steps': 48886, 'loss/train': 1.9954582452774048}
+02/26/2022 16:53:28 - INFO - codeparrot_training - Step 48887: {'lr': 6.630183608844764e-07, 'samples': 25030656, 'steps': 48887, 'loss/train': 2.0644354820251465}
+02/26/2022 16:53:31 - INFO - codeparrot_training - Step 48888: {'lr': 6.618280141652833e-07, 'samples': 25031168, 'steps': 48888, 'loss/train': 1.4872416257858276}
+02/26/2022 16:53:36 - INFO - codeparrot_training - Step 48889: {'lr': 6.606387355316545e-07, 'samples': 25031680, 'steps': 48889, 'loss/train': 1.973291277885437}
+02/26/2022 16:53:40 - INFO - codeparrot_training - Step 48890: {'lr': 6.594505249886695e-07, 'samples': 25032192, 'steps': 48890, 'loss/train': 2.922168254852295}
+02/26/2022 16:53:46 - INFO - codeparrot_training - Step 48891: {'lr': 6.582633825414075e-07, 'samples': 25032704, 'steps': 48891, 'loss/train': 0.972525417804718}
+02/26/2022 16:53:49 - INFO - codeparrot_training - Step 48892: {'lr': 6.570773081950033e-07, 'samples': 25033216, 'steps': 48892, 'loss/train': 1.3254616260528564}
+02/26/2022 16:53:55 - INFO - codeparrot_training - Step 48893: {'lr': 6.558923019544804e-07, 'samples': 25033728, 'steps': 48893, 'loss/train': 1.7696346044540405}
+02/26/2022 16:53:58 - INFO - codeparrot_training - Step 48894: {'lr': 6.547083638249185e-07, 'samples': 25034240, 'steps': 48894, 'loss/train': 1.839529037475586}
+02/26/2022 16:54:04 - INFO - codeparrot_training - Step 48895: {'lr': 6.535254938114521e-07, 'samples': 25034752, 'steps': 48895, 'loss/train': 2.6115753650665283}
+02/26/2022 16:54:07 - INFO - codeparrot_training - Step 48896: {'lr': 6.523436919190773e-07, 'samples': 25035264, 'steps': 48896, 'loss/train': 1.5751363039016724}
+02/26/2022 16:54:13 - INFO - codeparrot_training - Step 48897: {'lr': 6.511629581529011e-07, 'samples': 25035776, 'steps': 48897, 'loss/train': 1.649093747138977}
+02/26/2022 16:54:17 - INFO - codeparrot_training - Step 48898: {'lr': 6.499832925179749e-07, 'samples': 25036288, 'steps': 48898, 'loss/train': 1.4541383981704712}
+02/26/2022 16:54:22 - INFO - codeparrot_training - Step 48899: {'lr': 6.488046950193228e-07, 'samples': 25036800, 'steps': 48899, 'loss/train': 1.5391324758529663}
+02/26/2022 16:54:26 - INFO - codeparrot_training - Step 48900: {'lr': 6.476271656620236e-07, 'samples': 25037312, 'steps': 48900, 'loss/train': 2.4783318042755127}
+02/26/2022 16:54:31 - INFO - codeparrot_training - Step 48901: {'lr': 6.464507044511015e-07, 'samples': 25037824, 'steps': 48901, 'loss/train': 1.8874729871749878}
+02/26/2022 16:54:35 - INFO - codeparrot_training - Step 48902: {'lr': 6.452753113916077e-07, 'samples': 25038336, 'steps': 48902, 'loss/train': 2.030632972717285}
+02/26/2022 16:54:40 - INFO - codeparrot_training - Step 48903: {'lr': 6.441009864885938e-07, 'samples': 25038848, 'steps': 48903, 'loss/train': 1.3733640909194946}
+02/26/2022 16:54:44 - INFO - codeparrot_training - Step 48904: {'lr': 6.42927729747056e-07, 'samples': 25039360, 'steps': 48904, 'loss/train': 0.734083354473114}
+02/26/2022 16:54:49 - INFO - codeparrot_training - Step 48905: {'lr': 6.417555411720454e-07, 'samples': 25039872, 'steps': 48905, 'loss/train': 1.6773123741149902}
+02/26/2022 16:54:53 - INFO - codeparrot_training - Step 48906: {'lr': 6.405844207685585e-07, 'samples': 25040384, 'steps': 48906, 'loss/train': 1.7317454814910889}
+02/26/2022 16:54:59 - INFO - codeparrot_training - Step 48907: {'lr': 6.394143685416187e-07, 'samples': 25040896, 'steps': 48907, 'loss/train': 1.48530113697052}
+02/26/2022 16:55:02 - INFO - codeparrot_training - Step 48908: {'lr': 6.382453844962776e-07, 'samples': 25041408, 'steps': 48908, 'loss/train': 1.9938817024230957}
+02/26/2022 16:55:08 - INFO - codeparrot_training - Step 48909: {'lr': 6.370774686375036e-07, 'samples': 25041920, 'steps': 48909, 'loss/train': 1.591796636581421}
+02/26/2022 16:55:11 - INFO - codeparrot_training - Step 48910: {'lr': 6.359106209702926e-07, 'samples': 25042432, 'steps': 48910, 'loss/train': 2.0834133625030518}
+02/26/2022 16:55:17 - INFO - codeparrot_training - Step 48911: {'lr': 6.347448414996682e-07, 'samples': 25042944, 'steps': 48911, 'loss/train': 1.879281997680664}
+02/26/2022 16:55:20 - INFO - codeparrot_training - Step 48912: {'lr': 6.335801302306265e-07, 'samples': 25043456, 'steps': 48912, 'loss/train': 1.8462363481521606}
+02/26/2022 16:55:26 - INFO - codeparrot_training - Step 48913: {'lr': 6.324164871681359e-07, 'samples': 25043968, 'steps': 48913, 'loss/train': 1.70614492893219}
+02/26/2022 16:55:29 - INFO - codeparrot_training - Step 48914: {'lr': 6.312539123171646e-07, 'samples': 25044480, 'steps': 48914, 'loss/train': 1.4532690048217773}
+02/26/2022 16:55:35 - INFO - codeparrot_training - Step 48915: {'lr': 6.300924056827362e-07, 'samples': 25044992, 'steps': 48915, 'loss/train': 2.6289873123168945}
+02/26/2022 16:55:38 - INFO - codeparrot_training - Step 48916: {'lr': 6.28931967269819e-07, 'samples': 25045504, 'steps': 48916, 'loss/train': 2.073599100112915}
+02/26/2022 16:55:44 - INFO - codeparrot_training - Step 48917: {'lr': 6.277725970833815e-07, 'samples': 25046016, 'steps': 48917, 'loss/train': 1.506932258605957}
+02/26/2022 16:55:48 - INFO - codeparrot_training - Step 48918: {'lr': 6.266142951283638e-07, 'samples': 25046528, 'steps': 48918, 'loss/train': 1.5356069803237915}
+02/26/2022 16:55:53 - INFO - codeparrot_training - Step 48919: {'lr': 6.254570614097344e-07, 'samples': 25047040, 'steps': 48919, 'loss/train': 0.13488370180130005}
+02/26/2022 16:55:57 - INFO - codeparrot_training - Step 48920: {'lr': 6.243008959324892e-07, 'samples': 25047552, 'steps': 48920, 'loss/train': 0.3730163276195526}
+02/26/2022 16:56:02 - INFO - codeparrot_training - Step 48921: {'lr': 6.23145798701541e-07, 'samples': 25048064, 'steps': 48921, 'loss/train': 1.9878838062286377}
+02/26/2022 16:56:06 - INFO - codeparrot_training - Step 48922: {'lr': 6.219917697218303e-07, 'samples': 25048576, 'steps': 48922, 'loss/train': 1.1903020143508911}
+02/26/2022 16:56:11 - INFO - codeparrot_training - Step 48923: {'lr': 6.20838808998353e-07, 'samples': 25049088, 'steps': 48923, 'loss/train': 1.5491689443588257}
+02/26/2022 16:56:15 - INFO - codeparrot_training - Step 48924: {'lr': 6.196869165359941e-07, 'samples': 25049600, 'steps': 48924, 'loss/train': 0.41483357548713684}
+02/26/2022 16:56:20 - INFO - codeparrot_training - Step 48925: {'lr': 6.185360923396943e-07, 'samples': 25050112, 'steps': 48925, 'loss/train': 1.0842278003692627}
+02/26/2022 16:56:24 - INFO - codeparrot_training - Step 48926: {'lr': 6.173863364144494e-07, 'samples': 25050624, 'steps': 48926, 'loss/train': 1.283677101135254}
+02/26/2022 16:56:29 - INFO - codeparrot_training - Step 48927: {'lr': 6.162376487650889e-07, 'samples': 25051136, 'steps': 48927, 'loss/train': 2.1205806732177734}
+02/26/2022 16:56:33 - INFO - codeparrot_training - Step 48928: {'lr': 6.150900293966089e-07, 'samples': 25051648, 'steps': 48928, 'loss/train': 2.7395167350769043}
+02/26/2022 16:56:38 - INFO - codeparrot_training - Step 48929: {'lr': 6.139434783138664e-07, 'samples': 25052160, 'steps': 48929, 'loss/train': 2.0299055576324463}
+02/26/2022 16:56:42 - INFO - codeparrot_training - Step 48930: {'lr': 6.127979955218299e-07, 'samples': 25052672, 'steps': 48930, 'loss/train': 0.6684316396713257}
+02/26/2022 16:56:47 - INFO - codeparrot_training - Step 48931: {'lr': 6.116535810253565e-07, 'samples': 25053184, 'steps': 48931, 'loss/train': 0.9637073278427124}
+02/26/2022 16:56:53 - INFO - codeparrot_training - Step 48932: {'lr': 6.10510234829359e-07, 'samples': 25053696, 'steps': 48932, 'loss/train': 1.3618954420089722}
+02/26/2022 16:56:57 - INFO - codeparrot_training - Step 48933: {'lr': 6.0936795693875e-07, 'samples': 25054208, 'steps': 48933, 'loss/train': 1.6464715003967285}
+02/26/2022 16:57:02 - INFO - codeparrot_training - Step 48934: {'lr': 6.082267473584424e-07, 'samples': 25054720, 'steps': 48934, 'loss/train': 2.2851829528808594}
+02/26/2022 16:57:06 - INFO - codeparrot_training - Step 48935: {'lr': 6.070866060932934e-07, 'samples': 25055232, 'steps': 48935, 'loss/train': 1.779329538345337}
+02/26/2022 16:57:11 - INFO - codeparrot_training - Step 48936: {'lr': 6.059475331481878e-07, 'samples': 25055744, 'steps': 48936, 'loss/train': 2.0829570293426514}
+02/26/2022 16:57:15 - INFO - codeparrot_training - Step 48937: {'lr': 6.048095285280109e-07, 'samples': 25056256, 'steps': 48937, 'loss/train': 0.2626766860485077}
+02/26/2022 16:57:20 - INFO - codeparrot_training - Step 48938: {'lr': 6.036725922376196e-07, 'samples': 25056768, 'steps': 48938, 'loss/train': 1.6045479774475098}
+02/26/2022 16:57:24 - INFO - codeparrot_training - Step 48939: {'lr': 6.025367242819268e-07, 'samples': 25057280, 'steps': 48939, 'loss/train': 1.910038709640503}
+02/26/2022 16:57:29 - INFO - codeparrot_training - Step 48940: {'lr': 6.014019246657898e-07, 'samples': 25057792, 'steps': 48940, 'loss/train': 1.8424007892608643}
+02/26/2022 16:57:33 - INFO - codeparrot_training - Step 48941: {'lr': 6.002681933940102e-07, 'samples': 25058304, 'steps': 48941, 'loss/train': 0.9923708438873291}
+02/26/2022 16:57:39 - INFO - codeparrot_training - Step 48942: {'lr': 5.991355304715285e-07, 'samples': 25058816, 'steps': 48942, 'loss/train': 3.176571846008301}
+02/26/2022 16:57:42 - INFO - codeparrot_training - Step 48943: {'lr': 5.980039359031186e-07, 'samples': 25059328, 'steps': 48943, 'loss/train': 0.3473702073097229}
+02/26/2022 16:57:48 - INFO - codeparrot_training - Step 48944: {'lr': 5.968734096936935e-07, 'samples': 25059840, 'steps': 48944, 'loss/train': 0.9416552186012268}
+02/26/2022 16:57:51 - INFO - codeparrot_training - Step 48945: {'lr': 5.957439518480545e-07, 'samples': 25060352, 'steps': 48945, 'loss/train': 1.5790685415267944}
+02/26/2022 16:57:57 - INFO - codeparrot_training - Step 48946: {'lr': 5.946155623710592e-07, 'samples': 25060864, 'steps': 48946, 'loss/train': 1.6193063259124756}
+02/26/2022 16:58:00 - INFO - codeparrot_training - Step 48947: {'lr': 5.934882412675646e-07, 'samples': 25061376, 'steps': 48947, 'loss/train': 2.1959645748138428}
+02/26/2022 16:58:06 - INFO - codeparrot_training - Step 48948: {'lr': 5.923619885423448e-07, 'samples': 25061888, 'steps': 48948, 'loss/train': 1.487581729888916}
+02/26/2022 16:58:09 - INFO - codeparrot_training - Step 48949: {'lr': 5.912368042002292e-07, 'samples': 25062400, 'steps': 48949, 'loss/train': 1.0387383699417114}
+02/26/2022 16:58:15 - INFO - codeparrot_training - Step 48950: {'lr': 5.90112688246075e-07, 'samples': 25062912, 'steps': 48950, 'loss/train': 1.5394786596298218}
+02/26/2022 16:58:18 - INFO - codeparrot_training - Step 48951: {'lr': 5.889896406846839e-07, 'samples': 25063424, 'steps': 48951, 'loss/train': 2.4579970836639404}
+02/26/2022 16:58:25 - INFO - codeparrot_training - Step 48952: {'lr': 5.878676615208856e-07, 'samples': 25063936, 'steps': 48952, 'loss/train': 1.8241870403289795}
+02/26/2022 16:58:28 - INFO - codeparrot_training - Step 48953: {'lr': 5.867467507594259e-07, 'samples': 25064448, 'steps': 48953, 'loss/train': 0.7845878601074219}
+02/26/2022 16:58:34 - INFO - codeparrot_training - Step 48954: {'lr': 5.856269084051624e-07, 'samples': 25064960, 'steps': 48954, 'loss/train': 0.5075451731681824}
+02/26/2022 16:58:37 - INFO - codeparrot_training - Step 48955: {'lr': 5.845081344628967e-07, 'samples': 25065472, 'steps': 48955, 'loss/train': 3.1926262378692627}
+02/26/2022 16:58:42 - INFO - codeparrot_training - Step 48956: {'lr': 5.833904289373748e-07, 'samples': 25065984, 'steps': 48956, 'loss/train': 2.351822853088379}
+02/26/2022 16:58:46 - INFO - codeparrot_training - Step 48957: {'lr': 5.822737918333987e-07, 'samples': 25066496, 'steps': 48957, 'loss/train': 1.7050217390060425}
+02/26/2022 16:58:52 - INFO - codeparrot_training - Step 48958: {'lr': 5.811582231557977e-07, 'samples': 25067008, 'steps': 48958, 'loss/train': 1.4442594051361084}
+02/26/2022 16:58:55 - INFO - codeparrot_training - Step 48959: {'lr': 5.800437229092903e-07, 'samples': 25067520, 'steps': 48959, 'loss/train': 1.4676461219787598}
+02/26/2022 16:59:01 - INFO - codeparrot_training - Step 48960: {'lr': 5.789302910986782e-07, 'samples': 25068032, 'steps': 48960, 'loss/train': 1.693637728691101}
+02/26/2022 16:59:04 - INFO - codeparrot_training - Step 48961: {'lr': 5.778179277287354e-07, 'samples': 25068544, 'steps': 48961, 'loss/train': 3.154338836669922}
+02/26/2022 16:59:10 - INFO - codeparrot_training - Step 48962: {'lr': 5.767066328042081e-07, 'samples': 25069056, 'steps': 48962, 'loss/train': 1.5748413801193237}
+02/26/2022 16:59:13 - INFO - codeparrot_training - Step 48963: {'lr': 5.75596406329898e-07, 'samples': 25069568, 'steps': 48963, 'loss/train': 1.761905550956726}
+02/26/2022 16:59:19 - INFO - codeparrot_training - Step 48964: {'lr': 5.744872483104957e-07, 'samples': 25070080, 'steps': 48964, 'loss/train': 1.4534127712249756}
+02/26/2022 16:59:23 - INFO - codeparrot_training - Step 48965: {'lr': 5.73379158750803e-07, 'samples': 25070592, 'steps': 48965, 'loss/train': 1.1781994104385376}
+02/26/2022 16:59:28 - INFO - codeparrot_training - Step 48966: {'lr': 5.722721376555662e-07, 'samples': 25071104, 'steps': 48966, 'loss/train': 1.6667585372924805}
+02/26/2022 16:59:32 - INFO - codeparrot_training - Step 48967: {'lr': 5.711661850295036e-07, 'samples': 25071616, 'steps': 48967, 'loss/train': 1.5644289255142212}
+02/26/2022 16:59:37 - INFO - codeparrot_training - Step 48968: {'lr': 5.700613008773336e-07, 'samples': 25072128, 'steps': 48968, 'loss/train': 2.4919207096099854}
+02/26/2022 16:59:41 - INFO - codeparrot_training - Step 48969: {'lr': 5.689574852038582e-07, 'samples': 25072640, 'steps': 48969, 'loss/train': 1.1718039512634277}
+02/26/2022 16:59:46 - INFO - codeparrot_training - Step 48970: {'lr': 5.6785473801374e-07, 'samples': 25073152, 'steps': 48970, 'loss/train': 1.782148838043213}
+02/26/2022 16:59:50 - INFO - codeparrot_training - Step 48971: {'lr': 5.667530593117531e-07, 'samples': 25073664, 'steps': 48971, 'loss/train': 1.9602621793746948}
+02/26/2022 16:59:55 - INFO - codeparrot_training - Step 48972: {'lr': 5.656524491025883e-07, 'samples': 25074176, 'steps': 48972, 'loss/train': 1.281722903251648}
+02/26/2022 16:59:59 - INFO - codeparrot_training - Step 48973: {'lr': 5.645529073909361e-07, 'samples': 25074688, 'steps': 48973, 'loss/train': 2.163677453994751}
+02/26/2022 17:00:04 - INFO - codeparrot_training - Step 48974: {'lr': 5.634544341815706e-07, 'samples': 25075200, 'steps': 48974, 'loss/train': 0.9127210974693298}
+02/26/2022 17:00:08 - INFO - codeparrot_training - Step 48975: {'lr': 5.623570294791269e-07, 'samples': 25075712, 'steps': 48975, 'loss/train': 2.605700731277466}
+02/26/2022 17:00:13 - INFO - codeparrot_training - Step 48976: {'lr': 5.612606932883513e-07, 'samples': 25076224, 'steps': 48976, 'loss/train': 1.877730131149292}
+02/26/2022 17:00:17 - INFO - codeparrot_training - Step 48977: {'lr': 5.601654256139621e-07, 'samples': 25076736, 'steps': 48977, 'loss/train': 1.9526714086532593}
+02/26/2022 17:00:23 - INFO - codeparrot_training - Step 48978: {'lr': 5.590712264605668e-07, 'samples': 25077248, 'steps': 48978, 'loss/train': 2.089413642883301}
+02/26/2022 17:00:26 - INFO - codeparrot_training - Step 48979: {'lr': 5.579780958329394e-07, 'samples': 25077760, 'steps': 48979, 'loss/train': 1.099550724029541}
+02/26/2022 17:00:32 - INFO - codeparrot_training - Step 48980: {'lr': 5.568860337357151e-07, 'samples': 25078272, 'steps': 48980, 'loss/train': 1.7878739833831787}
+02/26/2022 17:00:36 - INFO - codeparrot_training - Step 48981: {'lr': 5.557950401735845e-07, 'samples': 25078784, 'steps': 48981, 'loss/train': 1.6897863149642944}
+02/26/2022 17:00:41 - INFO - codeparrot_training - Step 48982: {'lr': 5.547051151512384e-07, 'samples': 25079296, 'steps': 48982, 'loss/train': 1.6523233652114868}
+02/26/2022 17:00:45 - INFO - codeparrot_training - Step 48983: {'lr': 5.536162586733118e-07, 'samples': 25079808, 'steps': 48983, 'loss/train': 2.362736701965332}
+02/26/2022 17:00:50 - INFO - codeparrot_training - Step 48984: {'lr': 5.525284707444678e-07, 'samples': 25080320, 'steps': 48984, 'loss/train': 0.5700541734695435}
+02/26/2022 17:00:54 - INFO - codeparrot_training - Step 48985: {'lr': 5.51441751369397e-07, 'samples': 25080832, 'steps': 48985, 'loss/train': 1.792810320854187}
+02/26/2022 17:00:59 - INFO - codeparrot_training - Step 48986: {'lr': 5.503561005527347e-07, 'samples': 25081344, 'steps': 48986, 'loss/train': 1.5757895708084106}
+02/26/2022 17:01:03 - INFO - codeparrot_training - Step 48987: {'lr': 5.492715182991714e-07, 'samples': 25081856, 'steps': 48987, 'loss/train': 1.5620430707931519}
+02/26/2022 17:01:08 - INFO - codeparrot_training - Step 48988: {'lr': 5.48188004613287e-07, 'samples': 25082368, 'steps': 48988, 'loss/train': 1.836152195930481}
+02/26/2022 17:01:12 - INFO - codeparrot_training - Step 48989: {'lr': 5.471055594997443e-07, 'samples': 25082880, 'steps': 48989, 'loss/train': 0.47212541103363037}
+02/26/2022 17:01:18 - INFO - codeparrot_training - Step 48990: {'lr': 5.460241829632063e-07, 'samples': 25083392, 'steps': 48990, 'loss/train': 1.2685925960540771}
+02/26/2022 17:01:21 - INFO - codeparrot_training - Step 48991: {'lr': 5.44943875008308e-07, 'samples': 25083904, 'steps': 48991, 'loss/train': 1.3224447965621948}
+02/26/2022 17:01:27 - INFO - codeparrot_training - Step 48992: {'lr': 5.438646356396293e-07, 'samples': 25084416, 'steps': 48992, 'loss/train': 1.1389464139938354}
+02/26/2022 17:01:30 - INFO - codeparrot_training - Step 48993: {'lr': 5.427864648618608e-07, 'samples': 25084928, 'steps': 48993, 'loss/train': 1.6031832695007324}
+02/26/2022 17:01:36 - INFO - codeparrot_training - Step 48994: {'lr': 5.417093626795544e-07, 'samples': 25085440, 'steps': 48994, 'loss/train': 2.1603238582611084}
+02/26/2022 17:01:39 - INFO - codeparrot_training - Step 48995: {'lr': 5.406333290973453e-07, 'samples': 25085952, 'steps': 48995, 'loss/train': 1.7539637088775635}
+02/26/2022 17:01:45 - INFO - codeparrot_training - Step 48996: {'lr': 5.395583641198687e-07, 'samples': 25086464, 'steps': 48996, 'loss/train': 2.132857322692871}
+02/26/2022 17:01:48 - INFO - codeparrot_training - Step 48997: {'lr': 5.384844677517043e-07, 'samples': 25086976, 'steps': 48997, 'loss/train': 1.8419115543365479}
+02/26/2022 17:01:54 - INFO - codeparrot_training - Step 48998: {'lr': 5.374116399974871e-07, 'samples': 25087488, 'steps': 48998, 'loss/train': 2.223346471786499}
+02/26/2022 17:01:57 - INFO - codeparrot_training - Step 48999: {'lr': 5.363398808617414e-07, 'samples': 25088000, 'steps': 48999, 'loss/train': 1.8959709405899048}
+02/26/2022 17:01:57 - INFO - codeparrot_training - Evaluating and saving model checkpoint