diff --git "a/log/debug_0.log" "b/log/debug_0.log" --- "a/log/debug_0.log" +++ "b/log/debug_0.log" @@ -7517,3 +7517,2526 @@ Use FP16 precision: False 12/22/2021 14:51:53 - INFO - codeparrot_training - Step 7498: {'lr': 0.000473820366480841, 'samples': 3839488, 'steps': 7498, 'batch_loss/train': 0.8493719166144729} 12/22/2021 14:52:07 - INFO - codeparrot_training - Step 7499: {'lr': 0.00047381334354464945, 'samples': 3840000, 'steps': 7499, 'batch_loss/train': 0.9593390175141394} 12/22/2021 14:52:07 - INFO - codeparrot_training - Evaluating and saving model checkpoint +12/23/2021 13:47:06 - INFO - codeparrot_training - Distributed environment: TPU +Num processes: 8 +Process index: 0 +Local process index: 0 +Device: xla:1 +Use FP16 precision: False + +12/23/2021 13:47:06 - WARNING - huggingface_hub.repository - Revision `usual-cosmos-13` does not exist. Created and checked out branch `usual-cosmos-13`. +12/23/2021 13:47:06 - WARNING - huggingface_hub.repository - +12/23/2021 13:47:19 - WARNING - datasets.builder - Using custom data configuration lvwerra___codeparrot-clean-train-a1efdd1059bd841d +12/23/2021 13:47:20 - WARNING - datasets.builder - Using custom data configuration lvwerra___codeparrot-clean-valid-a800eb55c299abc0 +12/23/2021 13:50:01 - INFO - codeparrot_training - Distributed environment: TPU +Num processes: 8 +Process index: 0 +Local process index: 0 +Device: xla:1 +Use FP16 precision: False + +12/23/2021 13:50:01 - WARNING - huggingface_hub.repository - Revision `peach-smoke-14` does not exist. Created and checked out branch `peach-smoke-14`. +12/23/2021 13:50:01 - WARNING - huggingface_hub.repository - +12/23/2021 13:50:07 - WARNING - datasets.builder - Using custom data configuration lvwerra___codeparrot-clean-train-a1efdd1059bd841d +12/23/2021 13:50:08 - WARNING - datasets.builder - Using custom data configuration lvwerra___codeparrot-clean-valid-a800eb55c299abc0 +12/23/2021 15:28:29 - INFO - codeparrot_training - Step 5001: {'lr': 0.0004885148644702411, 'samples': 2561024, 'steps': 5001, 'batch_loss/train': 1.5026091635227203} +12/23/2021 15:29:47 - INFO - codeparrot_training - Step 5002: {'lr': 0.0004885101410877608, 'samples': 2561536, 'steps': 5002, 'batch_loss/train': 1.1865872200578451} +12/23/2021 15:31:11 - INFO - codeparrot_training - Step 5003: {'lr': 0.0004885054167570542, 'samples': 2562048, 'steps': 5003, 'batch_loss/train': 0.94274543505162} +12/23/2021 15:31:21 - INFO - codeparrot_training - Step 5004: {'lr': 0.0004885006914781403, 'samples': 2562560, 'steps': 5004, 'batch_loss/train': 0.9444990511983633} +12/23/2021 15:31:33 - INFO - codeparrot_training - Step 5005: {'lr': 0.0004884959652510374, 'samples': 2563072, 'steps': 5005, 'batch_loss/train': 0.937822412699461} +12/23/2021 15:31:44 - INFO - codeparrot_training - Step 5006: {'lr': 0.0004884912380757649, 'samples': 2563584, 'steps': 5006, 'batch_loss/train': 1.0078136511147022} +12/23/2021 15:31:55 - INFO - codeparrot_training - Step 5007: {'lr': 0.0004884865099523411, 'samples': 2564096, 'steps': 5007, 'batch_loss/train': 0.9146937504410744} +12/23/2021 15:32:05 - INFO - codeparrot_training - Step 5008: {'lr': 0.0004884817808807851, 'samples': 2564608, 'steps': 5008, 'batch_loss/train': 0.9582442473620176} +12/23/2021 15:32:18 - INFO - codeparrot_training - Step 5009: {'lr': 0.0004884770508611155, 'samples': 2565120, 'steps': 5009, 'batch_loss/train': 1.0170180341228843} +12/23/2021 15:32:29 - INFO - codeparrot_training - Step 5010: {'lr': 0.0004884723198933513, 'samples': 2565632, 'steps': 5010, 'batch_loss/train': 1.0913577722385526} +12/23/2021 15:32:39 - INFO - codeparrot_training - Step 5011: {'lr': 0.0004884675879775111, 'samples': 2566144, 'steps': 5011, 'batch_loss/train': 0.8828882509842515} +12/23/2021 15:32:51 - INFO - codeparrot_training - Step 5012: {'lr': 0.0004884628551136139, 'samples': 2566656, 'steps': 5012, 'batch_loss/train': 0.8071097414940596} +12/23/2021 15:33:02 - INFO - codeparrot_training - Step 5013: {'lr': 0.0004884581213016782, 'samples': 2567168, 'steps': 5013, 'batch_loss/train': 0.8648702101781964} +12/23/2021 15:33:13 - INFO - codeparrot_training - Step 5014: {'lr': 0.0004884533865417233, 'samples': 2567680, 'steps': 5014, 'batch_loss/train': 0.8629378015175462} +12/23/2021 15:33:25 - INFO - codeparrot_training - Step 5015: {'lr': 0.0004884486508337677, 'samples': 2568192, 'steps': 5015, 'batch_loss/train': 0.9607390277087688} +12/23/2021 15:33:35 - INFO - codeparrot_training - Step 5016: {'lr': 0.0004884439141778303, 'samples': 2568704, 'steps': 5016, 'batch_loss/train': 0.9778957339003682} +12/23/2021 15:33:46 - INFO - codeparrot_training - Step 5017: {'lr': 0.0004884391765739299, 'samples': 2569216, 'steps': 5017, 'batch_loss/train': 1.0375934960320592} +12/23/2021 15:33:59 - INFO - codeparrot_training - Step 5018: {'lr': 0.0004884344380220853, 'samples': 2569728, 'steps': 5018, 'batch_loss/train': 0.9520214977674186} +12/23/2021 15:34:09 - INFO - codeparrot_training - Step 5019: {'lr': 0.0004884296985223156, 'samples': 2570240, 'steps': 5019, 'batch_loss/train': 0.7895133283454925} +12/23/2021 15:34:20 - INFO - codeparrot_training - Step 5020: {'lr': 0.0004884249580746393, 'samples': 2570752, 'steps': 5020, 'batch_loss/train': 0.9972262429073453} +12/23/2021 15:34:30 - INFO - codeparrot_training - Step 5021: {'lr': 0.0004884202166790754, 'samples': 2571264, 'steps': 5021, 'batch_loss/train': 0.9639650443568826} +12/23/2021 15:34:42 - INFO - codeparrot_training - Step 5022: {'lr': 0.0004884154743356428, 'samples': 2571776, 'steps': 5022, 'batch_loss/train': 0.9555566008202732} +12/23/2021 15:34:53 - INFO - codeparrot_training - Step 5023: {'lr': 0.0004884107310443602, 'samples': 2572288, 'steps': 5023, 'batch_loss/train': 0.9700555251911283} +12/23/2021 15:35:04 - INFO - codeparrot_training - Step 5024: {'lr': 0.0004884059868052466, 'samples': 2572800, 'steps': 5024, 'batch_loss/train': 0.9354477424640208} +12/23/2021 15:35:17 - INFO - codeparrot_training - Step 5025: {'lr': 0.0004884012416183209, 'samples': 2573312, 'steps': 5025, 'batch_loss/train': 1.6362912580370903} +12/23/2021 15:35:28 - INFO - codeparrot_training - Step 5026: {'lr': 0.0004883964954836017, 'samples': 2573824, 'steps': 5026, 'batch_loss/train': 0.8716003783047199} +12/23/2021 15:35:38 - INFO - codeparrot_training - Step 5027: {'lr': 0.0004883917484011083, 'samples': 2574336, 'steps': 5027, 'batch_loss/train': 0.983111934736371} +12/23/2021 15:35:51 - INFO - codeparrot_training - Step 5028: {'lr': 0.0004883870003708591, 'samples': 2574848, 'steps': 5028, 'batch_loss/train': 0.972165203653276} +12/23/2021 15:36:01 - INFO - codeparrot_training - Step 5029: {'lr': 0.0004883822513928733, 'samples': 2575360, 'steps': 5029, 'batch_loss/train': 0.9556087236851454} +12/23/2021 15:36:12 - INFO - codeparrot_training - Step 5030: {'lr': 0.0004883775014671697, 'samples': 2575872, 'steps': 5030, 'batch_loss/train': 0.8103720187209547} +12/23/2021 15:36:23 - INFO - codeparrot_training - Step 5031: {'lr': 0.0004883727505937671, 'samples': 2576384, 'steps': 5031, 'batch_loss/train': 0.8694045953452587} +12/23/2021 15:36:35 - INFO - codeparrot_training - Step 5032: {'lr': 0.0004883679987726844, 'samples': 2576896, 'steps': 5032, 'batch_loss/train': 0.7771210488863289} +12/23/2021 15:36:45 - INFO - codeparrot_training - Step 5033: {'lr': 0.0004883632460039406, 'samples': 2577408, 'steps': 5033, 'batch_loss/train': 0.8132210718467832} +12/23/2021 15:36:56 - INFO - codeparrot_training - Step 5034: {'lr': 0.0004883584922875546, 'samples': 2577920, 'steps': 5034, 'batch_loss/train': 0.8654826935380697} +12/23/2021 15:37:09 - INFO - codeparrot_training - Step 5035: {'lr': 0.0004883537376235452, 'samples': 2578432, 'steps': 5035, 'batch_loss/train': 0.8686805320903659} +12/23/2021 15:37:19 - INFO - codeparrot_training - Step 5036: {'lr': 0.0004883489820119313, 'samples': 2578944, 'steps': 5036, 'batch_loss/train': 0.8766453023999929} +12/23/2021 15:37:30 - INFO - codeparrot_training - Step 5037: {'lr': 0.0004883442254527319, 'samples': 2579456, 'steps': 5037, 'batch_loss/train': 0.8979486878961325} +12/23/2021 15:37:42 - INFO - codeparrot_training - Step 5038: {'lr': 0.0004883394679459658, 'samples': 2579968, 'steps': 5038, 'batch_loss/train': 0.9537549670785666} +12/23/2021 15:37:52 - INFO - codeparrot_training - Step 5039: {'lr': 0.000488334709491652, 'samples': 2580480, 'steps': 5039, 'batch_loss/train': 0.9081472549587488} +12/23/2021 15:38:03 - INFO - codeparrot_training - Step 5040: {'lr': 0.0004883299500898094, 'samples': 2580992, 'steps': 5040, 'batch_loss/train': 0.8420660514384508} +12/23/2021 15:38:14 - INFO - codeparrot_training - Step 5041: {'lr': 0.0004883251897404569, 'samples': 2581504, 'steps': 5041, 'batch_loss/train': 0.8815709548071027} +12/23/2021 15:38:26 - INFO - codeparrot_training - Step 5042: {'lr': 0.0004883204284436134, 'samples': 2582016, 'steps': 5042, 'batch_loss/train': 0.9119223123416305} +12/23/2021 15:38:36 - INFO - codeparrot_training - Step 5043: {'lr': 0.0004883156661992979, 'samples': 2582528, 'steps': 5043, 'batch_loss/train': 0.8186895484104753} +12/23/2021 15:38:47 - INFO - codeparrot_training - Step 5044: {'lr': 0.0004883109030075293, 'samples': 2583040, 'steps': 5044, 'batch_loss/train': 0.8131888685747981} +12/23/2021 15:38:59 - INFO - codeparrot_training - Step 5045: {'lr': 0.0004883061388683264, 'samples': 2583552, 'steps': 5045, 'batch_loss/train': 0.8855598121881485} +12/23/2021 15:39:10 - INFO - codeparrot_training - Step 5046: {'lr': 0.0004883013737817085, 'samples': 2584064, 'steps': 5046, 'batch_loss/train': 0.8686516782036051} +12/23/2021 15:39:20 - INFO - codeparrot_training - Step 5047: {'lr': 0.0004882966077476941, 'samples': 2584576, 'steps': 5047, 'batch_loss/train': 0.8775115087628365} +12/23/2021 15:39:33 - INFO - codeparrot_training - Step 5048: {'lr': 0.0004882918407663025, 'samples': 2585088, 'steps': 5048, 'batch_loss/train': 0.9477125005796552} +12/23/2021 15:39:44 - INFO - codeparrot_training - Step 5049: {'lr': 0.0004882870728375525, 'samples': 2585600, 'steps': 5049, 'batch_loss/train': 0.9402937367558479} +12/23/2021 15:39:54 - INFO - codeparrot_training - Step 5050: {'lr': 0.000488282303961463, 'samples': 2586112, 'steps': 5050, 'batch_loss/train': 0.8628735924139619} +12/23/2021 15:40:06 - INFO - codeparrot_training - Step 5051: {'lr': 0.0004882775341380531, 'samples': 2586624, 'steps': 5051, 'batch_loss/train': 0.818708369974047} +12/23/2021 15:40:17 - INFO - codeparrot_training - Step 5052: {'lr': 0.0004882727633673417, 'samples': 2587136, 'steps': 5052, 'batch_loss/train': 0.7817599646514282} +12/23/2021 15:40:28 - INFO - codeparrot_training - Step 5053: {'lr': 0.0004882679916493478, 'samples': 2587648, 'steps': 5053, 'batch_loss/train': 0.9761519134044647} +12/23/2021 15:40:38 - INFO - codeparrot_training - Step 5054: {'lr': 0.0004882632189840902, 'samples': 2588160, 'steps': 5054, 'batch_loss/train': 0.9316751328296959} +12/23/2021 15:40:51 - INFO - codeparrot_training - Step 5055: {'lr': 0.00048825844537158807, 'samples': 2588672, 'steps': 5055, 'batch_loss/train': 0.6149575223680586} +12/23/2021 15:41:02 - INFO - codeparrot_training - Step 5056: {'lr': 0.00048825367081186034, 'samples': 2589184, 'steps': 5056, 'batch_loss/train': 0.7953188382089138} +12/23/2021 15:41:12 - INFO - codeparrot_training - Step 5057: {'lr': 0.000488248895304926, 'samples': 2589696, 'steps': 5057, 'batch_loss/train': 0.8482408225536346} +12/23/2021 15:41:25 - INFO - codeparrot_training - Step 5058: {'lr': 0.00048824411885080396, 'samples': 2590208, 'steps': 5058, 'batch_loss/train': 0.9526168508455157} +12/23/2021 15:41:35 - INFO - codeparrot_training - Step 5059: {'lr': 0.00048823934144951326, 'samples': 2590720, 'steps': 5059, 'batch_loss/train': 0.843859089538455} +12/23/2021 15:41:46 - INFO - codeparrot_training - Step 5060: {'lr': 0.000488234563101073, 'samples': 2591232, 'steps': 5060, 'batch_loss/train': 0.7656138380989432} +12/23/2021 15:41:58 - INFO - codeparrot_training - Step 5061: {'lr': 0.00048822978380550186, 'samples': 2591744, 'steps': 5061, 'batch_loss/train': 0.7750613241223618} +12/23/2021 15:42:09 - INFO - codeparrot_training - Step 5062: {'lr': 0.00048822500356281916, 'samples': 2592256, 'steps': 5062, 'batch_loss/train': 0.7809433155925944} +12/23/2021 15:42:19 - INFO - codeparrot_training - Step 5063: {'lr': 0.00048822022237304385, 'samples': 2592768, 'steps': 5063, 'batch_loss/train': 0.855440502637066} +12/23/2021 15:42:32 - INFO - codeparrot_training - Step 5064: {'lr': 0.0004882154402361949, 'samples': 2593280, 'steps': 5064, 'batch_loss/train': 0.9073802577331662} +12/23/2021 15:42:43 - INFO - codeparrot_training - Step 5065: {'lr': 0.0004882106571522912, 'samples': 2593792, 'steps': 5065, 'batch_loss/train': 0.7754645613022149} +12/23/2021 15:42:53 - INFO - codeparrot_training - Step 5066: {'lr': 0.00048820587312135193, 'samples': 2594304, 'steps': 5066, 'batch_loss/train': 0.8155799154192209} +12/23/2021 15:43:04 - INFO - codeparrot_training - Step 5067: {'lr': 0.000488201088143396, 'samples': 2594816, 'steps': 5067, 'batch_loss/train': 0.9725690968334675} +12/23/2021 15:43:16 - INFO - codeparrot_training - Step 5068: {'lr': 0.0004881963022184426, 'samples': 2595328, 'steps': 5068, 'batch_loss/train': 0.8566752867773175} +12/23/2021 15:43:27 - INFO - codeparrot_training - Step 5069: {'lr': 0.0004881915153465106, 'samples': 2595840, 'steps': 5069, 'batch_loss/train': 0.8825627989135683} +12/23/2021 15:43:37 - INFO - codeparrot_training - Step 5070: {'lr': 0.000488186727527619, 'samples': 2596352, 'steps': 5070, 'batch_loss/train': 0.9566769255325198} +12/23/2021 15:43:49 - INFO - codeparrot_training - Step 5071: {'lr': 0.00048818193876178697, 'samples': 2596864, 'steps': 5071, 'batch_loss/train': 0.8984131230972707} +12/23/2021 15:44:00 - INFO - codeparrot_training - Step 5072: {'lr': 0.00048817714904903344, 'samples': 2597376, 'steps': 5072, 'batch_loss/train': 0.8612307119183242} +12/23/2021 15:44:10 - INFO - codeparrot_training - Step 5073: {'lr': 0.00048817235838937757, 'samples': 2597888, 'steps': 5073, 'batch_loss/train': 0.9760541040450335} +12/23/2021 15:44:23 - INFO - codeparrot_training - Step 5074: {'lr': 0.00048816756678283824, 'samples': 2598400, 'steps': 5074, 'batch_loss/train': 0.87648830935359} +12/23/2021 15:44:34 - INFO - codeparrot_training - Step 5075: {'lr': 0.0004881627742294347, 'samples': 2598912, 'steps': 5075, 'batch_loss/train': 0.8510495238006115} +12/23/2021 15:44:44 - INFO - codeparrot_training - Step 5076: {'lr': 0.0004881579807291858, 'samples': 2599424, 'steps': 5076, 'batch_loss/train': 0.9107937561348081} +12/23/2021 15:44:56 - INFO - codeparrot_training - Step 5077: {'lr': 0.00048815318628211074, 'samples': 2599936, 'steps': 5077, 'batch_loss/train': 0.8791267406195402} +12/23/2021 15:45:07 - INFO - codeparrot_training - Step 5078: {'lr': 0.0004881483908882285, 'samples': 2600448, 'steps': 5078, 'batch_loss/train': 0.8774698209017515} +12/23/2021 15:45:18 - INFO - codeparrot_training - Step 5079: {'lr': 0.00048814359454755823, 'samples': 2600960, 'steps': 5079, 'batch_loss/train': 0.8795020994730294} +12/23/2021 15:45:28 - INFO - codeparrot_training - Step 5080: {'lr': 0.0004881387972601189, 'samples': 2601472, 'steps': 5080, 'batch_loss/train': 0.7977597746066749} +12/23/2021 15:45:41 - INFO - codeparrot_training - Step 5081: {'lr': 0.0004881339990259297, 'samples': 2601984, 'steps': 5081, 'batch_loss/train': 0.8704285835847259} +12/23/2021 15:45:51 - INFO - codeparrot_training - Step 5082: {'lr': 0.0004881291998450096, 'samples': 2602496, 'steps': 5082, 'batch_loss/train': 0.9047094183042645} +12/23/2021 15:46:02 - INFO - codeparrot_training - Step 5083: {'lr': 0.0004881243997173777, 'samples': 2603008, 'steps': 5083, 'batch_loss/train': 0.8659959118813276} +12/23/2021 15:46:15 - INFO - codeparrot_training - Step 5084: {'lr': 0.0004881195986430531, 'samples': 2603520, 'steps': 5084, 'batch_loss/train': 0.9663840988650918} +12/23/2021 15:46:25 - INFO - codeparrot_training - Step 5085: {'lr': 0.0004881147966220549, 'samples': 2604032, 'steps': 5085, 'batch_loss/train': 0.973696562461555} +12/23/2021 15:46:36 - INFO - codeparrot_training - Step 5086: {'lr': 0.0004881099936544021, 'samples': 2604544, 'steps': 5086, 'batch_loss/train': 0.8294914495199919} +12/23/2021 15:46:48 - INFO - codeparrot_training - Step 5087: {'lr': 0.000488105189740114, 'samples': 2605056, 'steps': 5087, 'batch_loss/train': 0.9522217009216547} +12/23/2021 15:46:58 - INFO - codeparrot_training - Step 5088: {'lr': 0.0004881003848792095, 'samples': 2605568, 'steps': 5088, 'batch_loss/train': 0.8545662397518754} +12/23/2021 15:47:09 - INFO - codeparrot_training - Step 5089: {'lr': 0.00048809557907170776, 'samples': 2606080, 'steps': 5089, 'batch_loss/train': 0.7805071684997529} +12/23/2021 15:47:20 - INFO - codeparrot_training - Step 5090: {'lr': 0.0004880907723176279, 'samples': 2606592, 'steps': 5090, 'batch_loss/train': 0.8963091094046831} +12/23/2021 15:47:31 - INFO - codeparrot_training - Step 5091: {'lr': 0.00048808596461698907, 'samples': 2607104, 'steps': 5091, 'batch_loss/train': 0.9134775875136256} +12/23/2021 15:47:42 - INFO - codeparrot_training - Step 5092: {'lr': 0.00048808115596981033, 'samples': 2607616, 'steps': 5092, 'batch_loss/train': 0.8666937302332371} +12/23/2021 15:47:53 - INFO - codeparrot_training - Step 5093: {'lr': 0.00048807634637611084, 'samples': 2608128, 'steps': 5093, 'batch_loss/train': 0.8661548839882016} +12/23/2021 15:48:05 - INFO - codeparrot_training - Step 5094: {'lr': 0.00048807153583590963, 'samples': 2608640, 'steps': 5094, 'batch_loss/train': 0.9108794671483338} +12/23/2021 15:48:16 - INFO - codeparrot_training - Step 5095: {'lr': 0.00048806672434922594, 'samples': 2609152, 'steps': 5095, 'batch_loss/train': 0.9526095055043697} +12/23/2021 15:48:27 - INFO - codeparrot_training - Step 5096: {'lr': 0.00048806191191607876, 'samples': 2609664, 'steps': 5096, 'batch_loss/train': 0.8360253497958183} +12/23/2021 15:48:39 - INFO - codeparrot_training - Step 5097: {'lr': 0.00048805709853648737, 'samples': 2610176, 'steps': 5097, 'batch_loss/train': 0.8948234813287854} +12/23/2021 15:48:49 - INFO - codeparrot_training - Step 5098: {'lr': 0.00048805228421047085, 'samples': 2610688, 'steps': 5098, 'batch_loss/train': 0.8901030803099275} +12/23/2021 15:49:00 - INFO - codeparrot_training - Step 5099: {'lr': 0.0004880474689380483, 'samples': 2611200, 'steps': 5099, 'batch_loss/train': 0.854511615820229} +12/23/2021 15:49:12 - INFO - codeparrot_training - Step 5100: {'lr': 0.0004880426527192389, 'samples': 2611712, 'steps': 5100, 'batch_loss/train': 0.8664661478251219} +12/23/2021 15:49:22 - INFO - codeparrot_training - Step 5101: {'lr': 0.00048803783555406184, 'samples': 2612224, 'steps': 5101, 'batch_loss/train': 0.859831496141851} +12/23/2021 15:49:33 - INFO - codeparrot_training - Step 5102: {'lr': 0.0004880330174425362, 'samples': 2612736, 'steps': 5102, 'batch_loss/train': 0.8280322453938425} +12/23/2021 15:49:44 - INFO - codeparrot_training - Step 5103: {'lr': 0.00048802819838468113, 'samples': 2613248, 'steps': 5103, 'batch_loss/train': 0.8251943215727806} +12/23/2021 15:49:56 - INFO - codeparrot_training - Step 5104: {'lr': 0.00048802337838051586, 'samples': 2613760, 'steps': 5104, 'batch_loss/train': 0.8112712358124554} +12/23/2021 15:50:07 - INFO - codeparrot_training - Step 5105: {'lr': 0.0004880185574300595, 'samples': 2614272, 'steps': 5105, 'batch_loss/train': 0.9277708940207958} +12/23/2021 15:50:18 - INFO - codeparrot_training - Step 5106: {'lr': 0.0004880137355333312, 'samples': 2614784, 'steps': 5106, 'batch_loss/train': 0.8675226448103786} +12/23/2021 15:50:30 - INFO - codeparrot_training - Step 5107: {'lr': 0.00048800891269035017, 'samples': 2615296, 'steps': 5107, 'batch_loss/train': 0.7895541829057038} +12/23/2021 15:50:40 - INFO - codeparrot_training - Step 5108: {'lr': 0.0004880040889011356, 'samples': 2615808, 'steps': 5108, 'batch_loss/train': 0.8319482561200857} +12/23/2021 15:50:51 - INFO - codeparrot_training - Step 5109: {'lr': 0.0004879992641657066, 'samples': 2616320, 'steps': 5109, 'batch_loss/train': 0.8801541225984693} +12/23/2021 15:51:04 - INFO - codeparrot_training - Step 5110: {'lr': 0.00048799443848408243, 'samples': 2616832, 'steps': 5110, 'batch_loss/train': 0.6825656982837245} +12/23/2021 15:51:14 - INFO - codeparrot_training - Step 5111: {'lr': 0.0004879896118562822, 'samples': 2617344, 'steps': 5111, 'batch_loss/train': 0.7790704960934818} +12/23/2021 15:51:25 - INFO - codeparrot_training - Step 5112: {'lr': 0.0004879847842823252, 'samples': 2617856, 'steps': 5112, 'batch_loss/train': 0.966109124943614} +12/23/2021 15:51:37 - INFO - codeparrot_training - Step 5113: {'lr': 0.0004879799557622305, 'samples': 2618368, 'steps': 5113, 'batch_loss/train': 0.901219037361443} +12/23/2021 15:51:47 - INFO - codeparrot_training - Step 5114: {'lr': 0.00048797512629601735, 'samples': 2618880, 'steps': 5114, 'batch_loss/train': 0.8336077006533742} +12/23/2021 15:51:58 - INFO - codeparrot_training - Step 5115: {'lr': 0.000487970295883705, 'samples': 2619392, 'steps': 5115, 'batch_loss/train': 0.8637348031625152} +12/23/2021 15:52:08 - INFO - codeparrot_training - Step 5116: {'lr': 0.00048796546452531256, 'samples': 2619904, 'steps': 5116, 'batch_loss/train': 0.9165717167779803} +12/23/2021 15:52:20 - INFO - codeparrot_training - Step 5117: {'lr': 0.0004879606322208593, 'samples': 2620416, 'steps': 5117, 'batch_loss/train': 0.8781238525407389} +12/23/2021 15:52:31 - INFO - codeparrot_training - Step 5118: {'lr': 0.0004879557989703644, 'samples': 2620928, 'steps': 5118, 'batch_loss/train': 0.8413781682029366} +12/23/2021 15:52:42 - INFO - codeparrot_training - Step 5119: {'lr': 0.00048795096477384714, 'samples': 2621440, 'steps': 5119, 'batch_loss/train': 0.8467745902016759} +12/23/2021 15:52:54 - INFO - codeparrot_training - Step 5120: {'lr': 0.0004879461296313267, 'samples': 2621952, 'steps': 5120, 'batch_loss/train': 0.83587427996099} +12/23/2021 15:53:05 - INFO - codeparrot_training - Step 5121: {'lr': 0.0004879412935428222, 'samples': 2622464, 'steps': 5121, 'batch_loss/train': 0.870086413808167} +12/23/2021 15:53:15 - INFO - codeparrot_training - Step 5122: {'lr': 0.00048793645650835305, 'samples': 2622976, 'steps': 5122, 'batch_loss/train': 0.6617107341298833} +12/23/2021 15:53:28 - INFO - codeparrot_training - Step 5123: {'lr': 0.00048793161852793833, 'samples': 2623488, 'steps': 5123, 'batch_loss/train': 0.9495914459694177} +12/23/2021 15:53:39 - INFO - codeparrot_training - Step 5124: {'lr': 0.0004879267796015974, 'samples': 2624000, 'steps': 5124, 'batch_loss/train': 0.7968164412304759} +12/23/2021 15:53:49 - INFO - codeparrot_training - Step 5125: {'lr': 0.00048792193972934934, 'samples': 2624512, 'steps': 5125, 'batch_loss/train': 0.8950514532625675} +12/23/2021 15:54:01 - INFO - codeparrot_training - Step 5126: {'lr': 0.0004879170989112135, 'samples': 2625024, 'steps': 5126, 'batch_loss/train': 0.8249944759299979} +12/23/2021 15:54:12 - INFO - codeparrot_training - Step 5127: {'lr': 0.00048791225714720914, 'samples': 2625536, 'steps': 5127, 'batch_loss/train': 0.985559078399092} +12/23/2021 15:54:23 - INFO - codeparrot_training - Step 5128: {'lr': 0.0004879074144373555, 'samples': 2626048, 'steps': 5128, 'batch_loss/train': 0.9424641523510218} +12/23/2021 15:54:33 - INFO - codeparrot_training - Step 5129: {'lr': 0.0004879025707816718, 'samples': 2626560, 'steps': 5129, 'batch_loss/train': 0.9210411766543984} +12/23/2021 15:54:45 - INFO - codeparrot_training - Step 5130: {'lr': 0.0004878977261801773, 'samples': 2627072, 'steps': 5130, 'batch_loss/train': 0.7952145347371697} +12/23/2021 15:54:56 - INFO - codeparrot_training - Step 5131: {'lr': 0.0004878928806328913, 'samples': 2627584, 'steps': 5131, 'batch_loss/train': 0.9276646012440324} +12/23/2021 15:55:06 - INFO - codeparrot_training - Step 5132: {'lr': 0.00048788803413983296, 'samples': 2628096, 'steps': 5132, 'batch_loss/train': 0.8722852931823581} +12/23/2021 15:55:19 - INFO - codeparrot_training - Step 5133: {'lr': 0.00048788318670102166, 'samples': 2628608, 'steps': 5133, 'batch_loss/train': 0.8821050124242902} +12/23/2021 15:55:30 - INFO - codeparrot_training - Step 5134: {'lr': 0.00048787833831647663, 'samples': 2629120, 'steps': 5134, 'batch_loss/train': 0.8455578377470374} +12/23/2021 15:55:40 - INFO - codeparrot_training - Step 5135: {'lr': 0.00048787348898621713, 'samples': 2629632, 'steps': 5135, 'batch_loss/train': 0.8344406075775623} +12/23/2021 15:55:52 - INFO - codeparrot_training - Step 5136: {'lr': 0.0004878686387102625, 'samples': 2630144, 'steps': 5136, 'batch_loss/train': 1.0528412610292435} +12/23/2021 15:56:03 - INFO - codeparrot_training - Step 5137: {'lr': 0.0004878637874886319, 'samples': 2630656, 'steps': 5137, 'batch_loss/train': 0.5806331451749429} +12/23/2021 15:56:13 - INFO - codeparrot_training - Step 5138: {'lr': 0.0004878589353213447, 'samples': 2631168, 'steps': 5138, 'batch_loss/train': 0.8641628930345178} +12/23/2021 15:56:24 - INFO - codeparrot_training - Step 5139: {'lr': 0.0004878540822084203, 'samples': 2631680, 'steps': 5139, 'batch_loss/train': 1.1372787691652775} +12/23/2021 15:56:38 - INFO - codeparrot_training - Step 5140: {'lr': 0.0004878492281498778, 'samples': 2632192, 'steps': 5140, 'batch_loss/train': 1.3216730738058686} +12/23/2021 15:56:48 - INFO - codeparrot_training - Step 5141: {'lr': 0.0004878443731457366, 'samples': 2632704, 'steps': 5141, 'batch_loss/train': 0.9198079258203506} +12/23/2021 15:56:59 - INFO - codeparrot_training - Step 5142: {'lr': 0.00048783951719601594, 'samples': 2633216, 'steps': 5142, 'batch_loss/train': 0.8868131330236793} +12/23/2021 15:57:11 - INFO - codeparrot_training - Step 5143: {'lr': 0.0004878346603007352, 'samples': 2633728, 'steps': 5143, 'batch_loss/train': 0.7653667835984379} +12/23/2021 15:57:22 - INFO - codeparrot_training - Step 5144: {'lr': 0.00048782980245991363, 'samples': 2634240, 'steps': 5144, 'batch_loss/train': 0.728238670155406} +12/23/2021 15:57:32 - INFO - codeparrot_training - Step 5145: {'lr': 0.00048782494367357053, 'samples': 2634752, 'steps': 5145, 'batch_loss/train': 1.771315807942301} +12/23/2021 15:57:45 - INFO - codeparrot_training - Step 5146: {'lr': 0.0004878200839417254, 'samples': 2635264, 'steps': 5146, 'batch_loss/train': 0.9261872041970491} +12/23/2021 15:57:55 - INFO - codeparrot_training - Step 5147: {'lr': 0.00048781522326439726, 'samples': 2635776, 'steps': 5147, 'batch_loss/train': 0.6632390524027869} +12/23/2021 15:58:06 - INFO - codeparrot_training - Step 5148: {'lr': 0.00048781036164160564, 'samples': 2636288, 'steps': 5148, 'batch_loss/train': 0.8320887368172407} +12/23/2021 15:58:18 - INFO - codeparrot_training - Step 5149: {'lr': 0.00048780549907336977, 'samples': 2636800, 'steps': 5149, 'batch_loss/train': 0.9134296369738877} +12/23/2021 15:58:29 - INFO - codeparrot_training - Step 5150: {'lr': 0.0004878006355597091, 'samples': 2637312, 'steps': 5150, 'batch_loss/train': 0.8401218093931675} +12/23/2021 15:58:40 - INFO - codeparrot_training - Step 5151: {'lr': 0.0004877957711006429, 'samples': 2637824, 'steps': 5151, 'batch_loss/train': 0.834359779022634} +12/23/2021 15:58:50 - INFO - codeparrot_training - Step 5152: {'lr': 0.00048779090569619045, 'samples': 2638336, 'steps': 5152, 'batch_loss/train': 0.9028393849730492} +12/23/2021 15:59:02 - INFO - codeparrot_training - Step 5153: {'lr': 0.00048778603934637125, 'samples': 2638848, 'steps': 5153, 'batch_loss/train': 0.8520970428362489} +12/23/2021 15:59:13 - INFO - codeparrot_training - Step 5154: {'lr': 0.0004877811720512044, 'samples': 2639360, 'steps': 5154, 'batch_loss/train': 0.8374455678276718} +12/23/2021 15:59:24 - INFO - codeparrot_training - Step 5155: {'lr': 0.0004877763038107095, 'samples': 2639872, 'steps': 5155, 'batch_loss/train': 0.7676745208445936} +12/23/2021 15:59:36 - INFO - codeparrot_training - Step 5156: {'lr': 0.0004877714346249057, 'samples': 2640384, 'steps': 5156, 'batch_loss/train': 0.902866561897099} +12/23/2021 15:59:46 - INFO - codeparrot_training - Step 5157: {'lr': 0.0004877665644938125, 'samples': 2640896, 'steps': 5157, 'batch_loss/train': 0.6161817027023062} +12/23/2021 15:59:57 - INFO - codeparrot_training - Step 5158: {'lr': 0.00048776169341744924, 'samples': 2641408, 'steps': 5158, 'batch_loss/train': 0.8791153328493237} +12/23/2021 16:00:10 - INFO - codeparrot_training - Step 5159: {'lr': 0.0004877568213958352, 'samples': 2641920, 'steps': 5159, 'batch_loss/train': 0.9370149979367852} +12/23/2021 16:00:20 - INFO - codeparrot_training - Step 5160: {'lr': 0.00048775194842898983, 'samples': 2642432, 'steps': 5160, 'batch_loss/train': 0.9828830366022885} +12/23/2021 16:00:31 - INFO - codeparrot_training - Step 5161: {'lr': 0.00048774707451693257, 'samples': 2642944, 'steps': 5161, 'batch_loss/train': 1.1311150202527642} +12/23/2021 16:00:43 - INFO - codeparrot_training - Step 5162: {'lr': 0.0004877421996596826, 'samples': 2643456, 'steps': 5162, 'batch_loss/train': 0.8588080340996385} +12/23/2021 16:00:54 - INFO - codeparrot_training - Step 5163: {'lr': 0.0004877373238572595, 'samples': 2643968, 'steps': 5163, 'batch_loss/train': 0.938912445679307} +12/23/2021 16:01:05 - INFO - codeparrot_training - Step 5164: {'lr': 0.0004877324471096825, 'samples': 2644480, 'steps': 5164, 'batch_loss/train': 0.7706322496524081} +12/23/2021 16:01:15 - INFO - codeparrot_training - Step 5165: {'lr': 0.0004877275694169711, 'samples': 2644992, 'steps': 5165, 'batch_loss/train': 0.8351135021075606} +12/23/2021 16:01:27 - INFO - codeparrot_training - Step 5166: {'lr': 0.00048772269077914466, 'samples': 2645504, 'steps': 5166, 'batch_loss/train': 1.6811448927037418} +12/23/2021 16:01:38 - INFO - codeparrot_training - Step 5167: {'lr': 0.0004877178111962225, 'samples': 2646016, 'steps': 5167, 'batch_loss/train': 0.9077048264443874} +12/23/2021 16:01:48 - INFO - codeparrot_training - Step 5168: {'lr': 0.0004877129306682241, 'samples': 2646528, 'steps': 5168, 'batch_loss/train': 0.984713482670486} +12/23/2021 16:02:01 - INFO - codeparrot_training - Step 5169: {'lr': 0.0004877080491951689, 'samples': 2647040, 'steps': 5169, 'batch_loss/train': 0.7918211440555751} +12/23/2021 16:02:12 - INFO - codeparrot_training - Step 5170: {'lr': 0.0004877031667770762, 'samples': 2647552, 'steps': 5170, 'batch_loss/train': 0.8449555281549692} +12/23/2021 16:02:22 - INFO - codeparrot_training - Step 5171: {'lr': 0.0004876982834139654, 'samples': 2648064, 'steps': 5171, 'batch_loss/train': 0.7173471720889211} +12/23/2021 16:02:34 - INFO - codeparrot_training - Step 5172: {'lr': 0.0004876933991058561, 'samples': 2648576, 'steps': 5172, 'batch_loss/train': 0.9032726273871958} +12/23/2021 16:02:45 - INFO - codeparrot_training - Step 5173: {'lr': 0.00048768851385276753, 'samples': 2649088, 'steps': 5173, 'batch_loss/train': 0.8930482282303274} +12/23/2021 16:02:56 - INFO - codeparrot_training - Step 5174: {'lr': 0.0004876836276547192, 'samples': 2649600, 'steps': 5174, 'batch_loss/train': 0.8985779462382197} +12/23/2021 16:03:06 - INFO - codeparrot_training - Step 5175: {'lr': 0.0004876787405117305, 'samples': 2650112, 'steps': 5175, 'batch_loss/train': 0.8352443668991327} +12/23/2021 16:03:18 - INFO - codeparrot_training - Step 5176: {'lr': 0.0004876738524238209, 'samples': 2650624, 'steps': 5176, 'batch_loss/train': 0.8773021437227726} +12/23/2021 16:03:29 - INFO - codeparrot_training - Step 5177: {'lr': 0.00048766896339100973, 'samples': 2651136, 'steps': 5177, 'batch_loss/train': 0.9391407128423452} +12/23/2021 16:03:39 - INFO - codeparrot_training - Step 5178: {'lr': 0.0004876640734133165, 'samples': 2651648, 'steps': 5178, 'batch_loss/train': 0.819476536475122} +12/23/2021 16:03:52 - INFO - codeparrot_training - Step 5179: {'lr': 0.0004876591824907608, 'samples': 2652160, 'steps': 5179, 'batch_loss/train': 0.9158071130514145} +12/23/2021 16:04:03 - INFO - codeparrot_training - Step 5180: {'lr': 0.00048765429062336176, 'samples': 2652672, 'steps': 5180, 'batch_loss/train': 0.8574013109318912} +12/23/2021 16:04:14 - INFO - codeparrot_training - Step 5181: {'lr': 0.0004876493978111391, 'samples': 2653184, 'steps': 5181, 'batch_loss/train': 0.883578764507547} +12/23/2021 16:04:26 - INFO - codeparrot_training - Step 5182: {'lr': 0.0004876445040541121, 'samples': 2653696, 'steps': 5182, 'batch_loss/train': 0.826340064406395} +12/23/2021 16:04:36 - INFO - codeparrot_training - Step 5183: {'lr': 0.00048763960935230033, 'samples': 2654208, 'steps': 5183, 'batch_loss/train': 0.9827705025672913} +12/23/2021 16:04:47 - INFO - codeparrot_training - Step 5184: {'lr': 0.0004876347137057232, 'samples': 2654720, 'steps': 5184, 'batch_loss/train': 0.7474596172105521} +12/23/2021 16:04:59 - INFO - codeparrot_training - Step 5185: {'lr': 0.0004876298171144001, 'samples': 2655232, 'steps': 5185, 'batch_loss/train': 0.8981895484030247} +12/23/2021 16:05:10 - INFO - codeparrot_training - Step 5186: {'lr': 0.0004876249195783507, 'samples': 2655744, 'steps': 5186, 'batch_loss/train': 0.8266614563763142} +12/23/2021 16:05:20 - INFO - codeparrot_training - Step 5187: {'lr': 0.0004876200210975943, 'samples': 2656256, 'steps': 5187, 'batch_loss/train': 0.5903100800933316} +12/23/2021 16:05:31 - INFO - codeparrot_training - Step 5188: {'lr': 0.0004876151216721504, 'samples': 2656768, 'steps': 5188, 'batch_loss/train': 0.8587072733789682} +12/23/2021 16:05:44 - INFO - codeparrot_training - Step 5189: {'lr': 0.00048761022130203846, 'samples': 2657280, 'steps': 5189, 'batch_loss/train': 0.8489085100591183} +12/23/2021 16:05:54 - INFO - codeparrot_training - Step 5190: {'lr': 0.00048760531998727805, 'samples': 2657792, 'steps': 5190, 'batch_loss/train': 0.8367858109995723} +12/23/2021 16:06:05 - INFO - codeparrot_training - Step 5191: {'lr': 0.00048760041772788856, 'samples': 2658304, 'steps': 5191, 'batch_loss/train': 0.9495682287961245} +12/23/2021 16:06:17 - INFO - codeparrot_training - Step 5192: {'lr': 0.00048759551452388954, 'samples': 2658816, 'steps': 5192, 'batch_loss/train': 0.869043055921793} +12/23/2021 16:06:27 - INFO - codeparrot_training - Step 5193: {'lr': 0.0004875906103753005, 'samples': 2659328, 'steps': 5193, 'batch_loss/train': 0.8486891235224903} +12/23/2021 16:06:38 - INFO - codeparrot_training - Step 5194: {'lr': 0.00048758570528214086, 'samples': 2659840, 'steps': 5194, 'batch_loss/train': 0.8906372943893075} +12/23/2021 16:06:51 - INFO - codeparrot_training - Step 5195: {'lr': 0.00048758079924443023, 'samples': 2660352, 'steps': 5195, 'batch_loss/train': 0.888426311314106} +12/23/2021 16:07:01 - INFO - codeparrot_training - Step 5196: {'lr': 0.000487575892262188, 'samples': 2660864, 'steps': 5196, 'batch_loss/train': 0.8461703220382333} +12/23/2021 16:07:12 - INFO - codeparrot_training - Step 5197: {'lr': 0.0004875709843354337, 'samples': 2661376, 'steps': 5197, 'batch_loss/train': 0.8402104526758194} +12/23/2021 16:07:23 - INFO - codeparrot_training - Step 5198: {'lr': 0.00048756607546418686, 'samples': 2661888, 'steps': 5198, 'batch_loss/train': 0.8352627651765943} +12/23/2021 16:07:35 - INFO - codeparrot_training - Step 5199: {'lr': 0.0004875611656484671, 'samples': 2662400, 'steps': 5199, 'batch_loss/train': 0.8864085986278951} +12/23/2021 16:07:45 - INFO - codeparrot_training - Step 5200: {'lr': 0.0004875562548882938, 'samples': 2662912, 'steps': 5200, 'batch_loss/train': 0.7915018824860454} +12/23/2021 16:07:56 - INFO - codeparrot_training - Step 5201: {'lr': 0.0004875513431836865, 'samples': 2663424, 'steps': 5201, 'batch_loss/train': 0.8539647553116083} +12/23/2021 16:08:08 - INFO - codeparrot_training - Step 5202: {'lr': 0.0004875464305346647, 'samples': 2663936, 'steps': 5202, 'batch_loss/train': 0.8072041100822389} +12/23/2021 16:08:18 - INFO - codeparrot_training - Step 5203: {'lr': 0.00048754151694124814, 'samples': 2664448, 'steps': 5203, 'batch_loss/train': 0.8064050707034767} +12/23/2021 16:08:29 - INFO - codeparrot_training - Step 5204: {'lr': 0.0004875366024034561, 'samples': 2664960, 'steps': 5204, 'batch_loss/train': 0.8362100617960095} +12/23/2021 16:08:41 - INFO - codeparrot_training - Step 5205: {'lr': 0.0004875316869213082, 'samples': 2665472, 'steps': 5205, 'batch_loss/train': 0.8530515935271978} +12/23/2021 16:08:52 - INFO - codeparrot_training - Step 5206: {'lr': 0.0004875267704948241, 'samples': 2665984, 'steps': 5206, 'batch_loss/train': 0.9809999521821737} +12/23/2021 16:09:02 - INFO - codeparrot_training - Step 5207: {'lr': 0.0004875218531240232, 'samples': 2666496, 'steps': 5207, 'batch_loss/train': 0.7495585037395358} +12/23/2021 16:09:15 - INFO - codeparrot_training - Step 5208: {'lr': 0.00048751693480892505, 'samples': 2667008, 'steps': 5208, 'batch_loss/train': 0.823962717782706} +12/23/2021 16:09:26 - INFO - codeparrot_training - Step 5209: {'lr': 0.0004875120155495493, 'samples': 2667520, 'steps': 5209, 'batch_loss/train': 0.8770559905096889} +12/23/2021 16:09:36 - INFO - codeparrot_training - Step 5210: {'lr': 0.00048750709534591544, 'samples': 2668032, 'steps': 5210, 'batch_loss/train': 0.9038038300350308} +12/23/2021 16:09:47 - INFO - codeparrot_training - Step 5211: {'lr': 0.00048750217419804304, 'samples': 2668544, 'steps': 5211, 'batch_loss/train': 0.8750148843973875} +12/23/2021 16:09:59 - INFO - codeparrot_training - Step 5212: {'lr': 0.0004874972521059517, 'samples': 2669056, 'steps': 5212, 'batch_loss/train': 0.8449505330063403} +12/23/2021 16:10:10 - INFO - codeparrot_training - Step 5213: {'lr': 0.00048749232906966097, 'samples': 2669568, 'steps': 5213, 'batch_loss/train': 0.8029340831562877} +12/23/2021 16:10:20 - INFO - codeparrot_training - Step 5214: {'lr': 0.00048748740508919037, 'samples': 2670080, 'steps': 5214, 'batch_loss/train': 0.8714582156389952} +12/23/2021 16:10:32 - INFO - codeparrot_training - Step 5215: {'lr': 0.00048748248016455955, 'samples': 2670592, 'steps': 5215, 'batch_loss/train': 0.7042688215151429} +12/23/2021 16:10:43 - INFO - codeparrot_training - Step 5216: {'lr': 0.000487477554295788, 'samples': 2671104, 'steps': 5216, 'batch_loss/train': 0.9037635866552591} +12/23/2021 16:10:53 - INFO - codeparrot_training - Step 5217: {'lr': 0.0004874726274828954, 'samples': 2671616, 'steps': 5217, 'batch_loss/train': 0.8789307737024501} +12/23/2021 16:11:06 - INFO - codeparrot_training - Step 5218: {'lr': 0.0004874676997259013, 'samples': 2672128, 'steps': 5218, 'batch_loss/train': 0.8022622941061854} +12/23/2021 16:11:17 - INFO - codeparrot_training - Step 5219: {'lr': 0.00048746277102482534, 'samples': 2672640, 'steps': 5219, 'batch_loss/train': 0.9200519239529967} +12/23/2021 16:11:27 - INFO - codeparrot_training - Step 5220: {'lr': 0.000487457841379687, 'samples': 2673152, 'steps': 5220, 'batch_loss/train': 0.8257863046601415} +12/23/2021 16:11:39 - INFO - codeparrot_training - Step 5221: {'lr': 0.00048745291079050604, 'samples': 2673664, 'steps': 5221, 'batch_loss/train': 1.340913669206202} +12/23/2021 16:11:50 - INFO - codeparrot_training - Step 5222: {'lr': 0.00048744797925730183, 'samples': 2674176, 'steps': 5222, 'batch_loss/train': 0.8200142844580114} +12/23/2021 16:12:01 - INFO - codeparrot_training - Step 5223: {'lr': 0.00048744304678009423, 'samples': 2674688, 'steps': 5223, 'batch_loss/train': 0.8258342994377017} +12/23/2021 16:12:11 - INFO - codeparrot_training - Step 5224: {'lr': 0.00048743811335890265, 'samples': 2675200, 'steps': 5224, 'batch_loss/train': 0.9213867979124188} +12/23/2021 16:12:24 - INFO - codeparrot_training - Step 5225: {'lr': 0.0004874331789937468, 'samples': 2675712, 'steps': 5225, 'batch_loss/train': 1.6109188366681337} +12/23/2021 16:12:35 - INFO - codeparrot_training - Step 5226: {'lr': 0.0004874282436846463, 'samples': 2676224, 'steps': 5226, 'batch_loss/train': 0.731355095282197} +12/23/2021 16:12:45 - INFO - codeparrot_training - Step 5227: {'lr': 0.0004874233074316207, 'samples': 2676736, 'steps': 5227, 'batch_loss/train': 0.7569799171760678} +12/23/2021 16:12:57 - INFO - codeparrot_training - Step 5228: {'lr': 0.00048741837023468974, 'samples': 2677248, 'steps': 5228, 'batch_loss/train': 0.8711100616492331} +12/23/2021 16:13:08 - INFO - codeparrot_training - Step 5229: {'lr': 0.00048741343209387296, 'samples': 2677760, 'steps': 5229, 'batch_loss/train': 0.8613164818380028} +12/23/2021 16:13:18 - INFO - codeparrot_training - Step 5230: {'lr': 0.00048740849300919, 'samples': 2678272, 'steps': 5230, 'batch_loss/train': 0.8373868661001325} +12/23/2021 16:13:31 - INFO - codeparrot_training - Step 5231: {'lr': 0.0004874035529806605, 'samples': 2678784, 'steps': 5231, 'batch_loss/train': 0.9138460620306432} +12/23/2021 16:13:41 - INFO - codeparrot_training - Step 5232: {'lr': 0.0004873986120083042, 'samples': 2679296, 'steps': 5232, 'batch_loss/train': 0.8925014492124319} +12/23/2021 16:13:52 - INFO - codeparrot_training - Step 5233: {'lr': 0.00048739367009214066, 'samples': 2679808, 'steps': 5233, 'batch_loss/train': 0.769422170240432} +12/23/2021 16:14:02 - INFO - codeparrot_training - Step 5234: {'lr': 0.0004873887272321894, 'samples': 2680320, 'steps': 5234, 'batch_loss/train': 0.7988339466974139} +12/23/2021 16:14:15 - INFO - codeparrot_training - Step 5235: {'lr': 0.00048738378342847035, 'samples': 2680832, 'steps': 5235, 'batch_loss/train': 0.8136687772348523} +12/23/2021 16:14:25 - INFO - codeparrot_training - Step 5236: {'lr': 0.00048737883868100286, 'samples': 2681344, 'steps': 5236, 'batch_loss/train': 0.8610944785177708} +12/23/2021 16:14:36 - INFO - codeparrot_training - Step 5237: {'lr': 0.0004873738929898068, 'samples': 2681856, 'steps': 5237, 'batch_loss/train': 0.8753723716363311} +12/23/2021 16:14:49 - INFO - codeparrot_training - Step 5238: {'lr': 0.0004873689463549018, 'samples': 2682368, 'steps': 5238, 'batch_loss/train': 0.830129380337894} +12/23/2021 16:14:59 - INFO - codeparrot_training - Step 5239: {'lr': 0.0004873639987763074, 'samples': 2682880, 'steps': 5239, 'batch_loss/train': 0.8405273575335741} +12/23/2021 16:15:10 - INFO - codeparrot_training - Step 5240: {'lr': 0.0004873590502540435, 'samples': 2683392, 'steps': 5240, 'batch_loss/train': 0.7600652379915118} +12/23/2021 16:15:22 - INFO - codeparrot_training - Step 5241: {'lr': 0.0004873541007881296, 'samples': 2683904, 'steps': 5241, 'batch_loss/train': 0.7708083838224411} +12/23/2021 16:15:33 - INFO - codeparrot_training - Step 5242: {'lr': 0.0004873491503785854, 'samples': 2684416, 'steps': 5242, 'batch_loss/train': 0.8405386693775654} +12/23/2021 16:15:43 - INFO - codeparrot_training - Step 5243: {'lr': 0.0004873441990254306, 'samples': 2684928, 'steps': 5243, 'batch_loss/train': 0.7605385719798505} +12/23/2021 16:15:54 - INFO - codeparrot_training - Step 5244: {'lr': 0.0004873392467286848, 'samples': 2685440, 'steps': 5244, 'batch_loss/train': 0.8143158247694373} +12/23/2021 16:16:06 - INFO - codeparrot_training - Step 5245: {'lr': 0.0004873342934883679, 'samples': 2685952, 'steps': 5245, 'batch_loss/train': 0.8865085495635867} +12/23/2021 16:16:17 - INFO - codeparrot_training - Step 5246: {'lr': 0.0004873293393044994, 'samples': 2686464, 'steps': 5246, 'batch_loss/train': 0.9576758984476328} +12/23/2021 16:16:27 - INFO - codeparrot_training - Step 5247: {'lr': 0.000487324384177099, 'samples': 2686976, 'steps': 5247, 'batch_loss/train': 0.8006468894891441} +12/23/2021 16:16:40 - INFO - codeparrot_training - Step 5248: {'lr': 0.00048731942810618657, 'samples': 2687488, 'steps': 5248, 'batch_loss/train': 0.913567965850234} +12/23/2021 16:16:51 - INFO - codeparrot_training - Step 5249: {'lr': 0.00048731447109178166, 'samples': 2688000, 'steps': 5249, 'batch_loss/train': 0.9248799169436097} +12/23/2021 16:17:01 - INFO - codeparrot_training - Step 5250: {'lr': 0.000487309513133904, 'samples': 2688512, 'steps': 5250, 'batch_loss/train': 0.8359837067546323} +12/23/2021 16:17:13 - INFO - codeparrot_training - Step 5251: {'lr': 0.00048730455423257334, 'samples': 2689024, 'steps': 5251, 'batch_loss/train': 0.9543391084298491} +12/23/2021 16:17:24 - INFO - codeparrot_training - Step 5252: {'lr': 0.00048729959438780933, 'samples': 2689536, 'steps': 5252, 'batch_loss/train': 0.6332688733236864} +12/23/2021 16:17:34 - INFO - codeparrot_training - Step 5253: {'lr': 0.00048729463359963184, 'samples': 2690048, 'steps': 5253, 'batch_loss/train': 0.8698622076772153} +12/23/2021 16:17:47 - INFO - codeparrot_training - Step 5254: {'lr': 0.0004872896718680604, 'samples': 2690560, 'steps': 5254, 'batch_loss/train': 0.8987905085086823} +12/23/2021 16:17:57 - INFO - codeparrot_training - Step 5255: {'lr': 0.00048728470919311487, 'samples': 2691072, 'steps': 5255, 'batch_loss/train': 0.7815705854445696} +12/23/2021 16:18:08 - INFO - codeparrot_training - Step 5256: {'lr': 0.0004872797455748149, 'samples': 2691584, 'steps': 5256, 'batch_loss/train': 0.9251176929101348} +12/23/2021 16:18:18 - INFO - codeparrot_training - Step 5257: {'lr': 0.00048727478101318023, 'samples': 2692096, 'steps': 5257, 'batch_loss/train': 0.8720349576324224} +12/23/2021 16:18:32 - INFO - codeparrot_training - Step 5258: {'lr': 0.0004872698155082307, 'samples': 2692608, 'steps': 5258, 'batch_loss/train': 0.8240997971734032} +12/23/2021 16:18:43 - INFO - codeparrot_training - Step 5259: {'lr': 0.0004872648490599859, 'samples': 2693120, 'steps': 5259, 'batch_loss/train': 0.8457135390490294} +12/23/2021 16:18:53 - INFO - codeparrot_training - Step 5260: {'lr': 0.0004872598816684657, 'samples': 2693632, 'steps': 5260, 'batch_loss/train': 0.7920829213690013} +12/23/2021 16:19:06 - INFO - codeparrot_training - Step 5261: {'lr': 0.00048725491333368977, 'samples': 2694144, 'steps': 5261, 'batch_loss/train': 0.9129653256386518} +12/23/2021 16:19:16 - INFO - codeparrot_training - Step 5262: {'lr': 0.00048724994405567785, 'samples': 2694656, 'steps': 5262, 'batch_loss/train': 0.7395861707627773} +12/23/2021 16:19:27 - INFO - codeparrot_training - Step 5263: {'lr': 0.0004872449738344498, 'samples': 2695168, 'steps': 5263, 'batch_loss/train': 0.8994851233437657} +12/23/2021 16:19:39 - INFO - codeparrot_training - Step 5264: {'lr': 0.0004872400026700253, 'samples': 2695680, 'steps': 5264, 'batch_loss/train': 0.8946950063109398} +12/23/2021 16:19:49 - INFO - codeparrot_training - Step 5265: {'lr': 0.0004872350305624241, 'samples': 2696192, 'steps': 5265, 'batch_loss/train': 0.9522080784663558} +12/23/2021 16:20:00 - INFO - codeparrot_training - Step 5266: {'lr': 0.000487230057511666, 'samples': 2696704, 'steps': 5266, 'batch_loss/train': 0.8744368813931942} +12/23/2021 16:20:11 - INFO - codeparrot_training - Step 5267: {'lr': 0.00048722508351777075, 'samples': 2697216, 'steps': 5267, 'batch_loss/train': 0.8643522392958403} +12/23/2021 16:20:25 - INFO - codeparrot_training - Step 5268: {'lr': 0.00048722010858075817, 'samples': 2697728, 'steps': 5268, 'batch_loss/train': 1.0017239544540644} +12/23/2021 16:20:35 - INFO - codeparrot_training - Step 5269: {'lr': 0.000487215132700648, 'samples': 2698240, 'steps': 5269, 'batch_loss/train': 0.788708905223757} +12/23/2021 16:20:46 - INFO - codeparrot_training - Step 5270: {'lr': 0.00048721015587746005, 'samples': 2698752, 'steps': 5270, 'batch_loss/train': 0.894611913478002} +12/23/2021 16:20:58 - INFO - codeparrot_training - Step 5271: {'lr': 0.0004872051781112141, 'samples': 2699264, 'steps': 5271, 'batch_loss/train': 0.7946038041263819} +12/23/2021 16:21:08 - INFO - codeparrot_training - Step 5272: {'lr': 0.00048720019940192983, 'samples': 2699776, 'steps': 5272, 'batch_loss/train': 0.9595844657160342} +12/23/2021 16:21:19 - INFO - codeparrot_training - Step 5273: {'lr': 0.00048719521974962724, 'samples': 2700288, 'steps': 5273, 'batch_loss/train': 0.6735741018783301} +12/23/2021 16:21:31 - INFO - codeparrot_training - Step 5274: {'lr': 0.000487190239154326, 'samples': 2700800, 'steps': 5274, 'batch_loss/train': 1.0999576188623905} +12/23/2021 16:21:42 - INFO - codeparrot_training - Step 5275: {'lr': 0.00048718525761604585, 'samples': 2701312, 'steps': 5275, 'batch_loss/train': 0.9683049526065588} +12/23/2021 16:21:53 - INFO - codeparrot_training - Step 5276: {'lr': 0.0004871802751348068, 'samples': 2701824, 'steps': 5276, 'batch_loss/train': 0.8910955227911472} +12/23/2021 16:22:03 - INFO - codeparrot_training - Step 5277: {'lr': 0.0004871752917106285, 'samples': 2702336, 'steps': 5277, 'batch_loss/train': 1.1133567597717047} +12/23/2021 16:22:16 - INFO - codeparrot_training - Step 5278: {'lr': 0.00048717030734353066, 'samples': 2702848, 'steps': 5278, 'batch_loss/train': 0.8769415374845266} +12/23/2021 16:22:27 - INFO - codeparrot_training - Step 5279: {'lr': 0.0004871653220335334, 'samples': 2703360, 'steps': 5279, 'batch_loss/train': 0.9150624638423324} +12/23/2021 16:22:38 - INFO - codeparrot_training - Step 5280: {'lr': 0.0004871603357806562, 'samples': 2703872, 'steps': 5280, 'batch_loss/train': 0.919238462112844} +12/23/2021 16:22:50 - INFO - codeparrot_training - Step 5281: {'lr': 0.00048715534858491914, 'samples': 2704384, 'steps': 5281, 'batch_loss/train': 0.898308320902288} +12/23/2021 16:23:01 - INFO - codeparrot_training - Step 5282: {'lr': 0.00048715036044634204, 'samples': 2704896, 'steps': 5282, 'batch_loss/train': 0.7492508511058986} +12/23/2021 16:23:11 - INFO - codeparrot_training - Step 5283: {'lr': 0.00048714537136494457, 'samples': 2705408, 'steps': 5283, 'batch_loss/train': 0.5116488956846297} +12/23/2021 16:23:22 - INFO - codeparrot_training - Step 5284: {'lr': 0.0004871403813407467, 'samples': 2705920, 'steps': 5284, 'batch_loss/train': 0.7132409383775666} +12/23/2021 16:23:35 - INFO - codeparrot_training - Step 5285: {'lr': 0.0004871353903737681, 'samples': 2706432, 'steps': 5285, 'batch_loss/train': 0.9177498687058687} +12/23/2021 16:23:46 - INFO - codeparrot_training - Step 5286: {'lr': 0.00048713039846402887, 'samples': 2706944, 'steps': 5286, 'batch_loss/train': 0.8387041185051203} +12/23/2021 16:23:57 - INFO - codeparrot_training - Step 5287: {'lr': 0.0004871254056115486, 'samples': 2707456, 'steps': 5287, 'batch_loss/train': 0.923771602101624} +12/23/2021 16:24:09 - INFO - codeparrot_training - Step 5288: {'lr': 0.00048712041181634726, 'samples': 2707968, 'steps': 5288, 'batch_loss/train': 0.758305405266583} +12/23/2021 16:24:19 - INFO - codeparrot_training - Step 5289: {'lr': 0.00048711541707844474, 'samples': 2708480, 'steps': 5289, 'batch_loss/train': 0.8637887629447505} +12/23/2021 16:24:30 - INFO - codeparrot_training - Step 5290: {'lr': 0.0004871104213978609, 'samples': 2708992, 'steps': 5290, 'batch_loss/train': 0.8502021729946136} +12/23/2021 16:24:42 - INFO - codeparrot_training - Step 5291: {'lr': 0.00048710542477461543, 'samples': 2709504, 'steps': 5291, 'batch_loss/train': 0.7943046921864152} +12/23/2021 16:24:52 - INFO - codeparrot_training - Step 5292: {'lr': 0.0004871004272087284, 'samples': 2710016, 'steps': 5292, 'batch_loss/train': 0.8912041252478957} +12/23/2021 16:25:03 - INFO - codeparrot_training - Step 5293: {'lr': 0.00048709542870021964, 'samples': 2710528, 'steps': 5293, 'batch_loss/train': 0.7331510013900697} +12/23/2021 16:25:16 - INFO - codeparrot_training - Step 5294: {'lr': 0.00048709042924910883, 'samples': 2711040, 'steps': 5294, 'batch_loss/train': 0.8696537343785167} +12/23/2021 16:25:27 - INFO - codeparrot_training - Step 5295: {'lr': 0.00048708542885541613, 'samples': 2711552, 'steps': 5295, 'batch_loss/train': 1.044104853644967} +12/23/2021 16:25:37 - INFO - codeparrot_training - Step 5296: {'lr': 0.00048708042751916127, 'samples': 2712064, 'steps': 5296, 'batch_loss/train': 0.9864169042557478} +12/23/2021 16:25:48 - INFO - codeparrot_training - Step 5297: {'lr': 0.0004870754252403641, 'samples': 2712576, 'steps': 5297, 'batch_loss/train': 0.8001178693957627} +12/23/2021 16:26:00 - INFO - codeparrot_training - Step 5298: {'lr': 0.0004870704220190446, 'samples': 2713088, 'steps': 5298, 'batch_loss/train': 0.8459446951746941} +12/23/2021 16:26:11 - INFO - codeparrot_training - Step 5299: {'lr': 0.00048706541785522255, 'samples': 2713600, 'steps': 5299, 'batch_loss/train': 0.8911434598267078} +12/23/2021 16:26:21 - INFO - codeparrot_training - Step 5300: {'lr': 0.000487060412748918, 'samples': 2714112, 'steps': 5300, 'batch_loss/train': 0.8026726664975286} +12/23/2021 16:26:34 - INFO - codeparrot_training - Step 5301: {'lr': 0.0004870554067001507, 'samples': 2714624, 'steps': 5301, 'batch_loss/train': 1.2333370428532362} +12/23/2021 16:26:45 - INFO - codeparrot_training - Step 5302: {'lr': 0.0004870503997089406, 'samples': 2715136, 'steps': 5302, 'batch_loss/train': 0.8824695469811559} +12/23/2021 16:26:55 - INFO - codeparrot_training - Step 5303: {'lr': 0.0004870453917753076, 'samples': 2715648, 'steps': 5303, 'batch_loss/train': 0.9643040746450424} +12/23/2021 16:27:07 - INFO - codeparrot_training - Step 5304: {'lr': 0.0004870403828992717, 'samples': 2716160, 'steps': 5304, 'batch_loss/train': 0.9376390869729221} +12/23/2021 16:27:18 - INFO - codeparrot_training - Step 5305: {'lr': 0.0004870353730808527, 'samples': 2716672, 'steps': 5305, 'batch_loss/train': 0.8772654398344457} +12/23/2021 16:27:28 - INFO - codeparrot_training - Step 5306: {'lr': 0.0004870303623200706, 'samples': 2717184, 'steps': 5306, 'batch_loss/train': 0.8874456090852618} +12/23/2021 16:27:39 - INFO - codeparrot_training - Step 5307: {'lr': 0.0004870253506169452, 'samples': 2717696, 'steps': 5307, 'batch_loss/train': 0.9026808366179466} +12/23/2021 16:27:52 - INFO - codeparrot_training - Step 5308: {'lr': 0.0004870203379714966, 'samples': 2718208, 'steps': 5308, 'batch_loss/train': 0.9405182553455234} +12/23/2021 16:28:03 - INFO - codeparrot_training - Step 5309: {'lr': 0.0004870153243837445, 'samples': 2718720, 'steps': 5309, 'batch_loss/train': 0.8305644830688834} +12/23/2021 16:28:13 - INFO - codeparrot_training - Step 5310: {'lr': 0.0004870103098537091, 'samples': 2719232, 'steps': 5310, 'batch_loss/train': 0.8557098507881165} +12/23/2021 16:28:25 - INFO - codeparrot_training - Step 5311: {'lr': 0.0004870052943814102, 'samples': 2719744, 'steps': 5311, 'batch_loss/train': 0.9230048917233944} +12/23/2021 16:28:36 - INFO - codeparrot_training - Step 5312: {'lr': 0.00048700027796686764, 'samples': 2720256, 'steps': 5312, 'batch_loss/train': 0.8038409533910453} +12/23/2021 16:28:47 - INFO - codeparrot_training - Step 5313: {'lr': 0.0004869952606101015, 'samples': 2720768, 'steps': 5313, 'batch_loss/train': 0.6772401641355827} +12/23/2021 16:28:59 - INFO - codeparrot_training - Step 5314: {'lr': 0.00048699024231113163, 'samples': 2721280, 'steps': 5314, 'batch_loss/train': 0.8588543841615319} +12/23/2021 16:29:09 - INFO - codeparrot_training - Step 5315: {'lr': 0.0004869852230699782, 'samples': 2721792, 'steps': 5315, 'batch_loss/train': 0.8439810443669558} +12/23/2021 16:29:20 - INFO - codeparrot_training - Step 5316: {'lr': 0.0004869802028866609, 'samples': 2722304, 'steps': 5316, 'batch_loss/train': 0.8028134508058429} +12/23/2021 16:29:33 - INFO - codeparrot_training - Step 5317: {'lr': 0.0004869751817611998, 'samples': 2722816, 'steps': 5317, 'batch_loss/train': 0.8499654736369848} +12/23/2021 16:29:44 - INFO - codeparrot_training - Step 5318: {'lr': 0.0004869701596936148, 'samples': 2723328, 'steps': 5318, 'batch_loss/train': 0.6854632245376706} +12/23/2021 16:29:54 - INFO - codeparrot_training - Step 5319: {'lr': 0.00048696513668392607, 'samples': 2723840, 'steps': 5319, 'batch_loss/train': 0.8734716540202498} +12/23/2021 16:30:05 - INFO - codeparrot_training - Step 5320: {'lr': 0.0004869601127321533, 'samples': 2724352, 'steps': 5320, 'batch_loss/train': 0.8590220487676561} +12/23/2021 16:30:17 - INFO - codeparrot_training - Step 5321: {'lr': 0.00048695508783831666, 'samples': 2724864, 'steps': 5321, 'batch_loss/train': 0.8713660016655922} +12/23/2021 16:30:27 - INFO - codeparrot_training - Step 5322: {'lr': 0.000486950062002436, 'samples': 2725376, 'steps': 5322, 'batch_loss/train': 0.7933802455663681} +12/23/2021 16:30:38 - INFO - codeparrot_training - Step 5323: {'lr': 0.0004869450352245314, 'samples': 2725888, 'steps': 5323, 'batch_loss/train': 0.8674264429137111} +12/23/2021 16:30:51 - INFO - codeparrot_training - Step 5324: {'lr': 0.0004869400075046228, 'samples': 2726400, 'steps': 5324, 'batch_loss/train': 0.8762605171650648} +12/23/2021 16:31:02 - INFO - codeparrot_training - Step 5325: {'lr': 0.0004869349788427302, 'samples': 2726912, 'steps': 5325, 'batch_loss/train': 0.983810419216752} +12/23/2021 16:31:12 - INFO - codeparrot_training - Step 5326: {'lr': 0.0004869299492388736, 'samples': 2727424, 'steps': 5326, 'batch_loss/train': 1.4499065298587084} +12/23/2021 16:31:24 - INFO - codeparrot_training - Step 5327: {'lr': 0.00048692491869307297, 'samples': 2727936, 'steps': 5327, 'batch_loss/train': 0.8516068027820438} +12/23/2021 16:31:35 - INFO - codeparrot_training - Step 5328: {'lr': 0.0004869198872053483, 'samples': 2728448, 'steps': 5328, 'batch_loss/train': 0.8848169082775712} +12/23/2021 16:31:45 - INFO - codeparrot_training - Step 5329: {'lr': 0.0004869148547757197, 'samples': 2728960, 'steps': 5329, 'batch_loss/train': 0.8424395131878555} +12/23/2021 16:31:56 - INFO - codeparrot_training - Step 5330: {'lr': 0.000486909821404207, 'samples': 2729472, 'steps': 5330, 'batch_loss/train': 0.9383692881092429} +12/23/2021 16:32:08 - INFO - codeparrot_training - Step 5331: {'lr': 0.00048690478709083036, 'samples': 2729984, 'steps': 5331, 'batch_loss/train': 0.8686944637447596} +12/23/2021 16:32:19 - INFO - codeparrot_training - Step 5332: {'lr': 0.0004868997518356098, 'samples': 2730496, 'steps': 5332, 'batch_loss/train': 0.8535953741520643} +12/23/2021 16:32:29 - INFO - codeparrot_training - Step 5333: {'lr': 0.00048689471563856515, 'samples': 2731008, 'steps': 5333, 'batch_loss/train': 0.7828864809125662} +12/23/2021 16:32:42 - INFO - codeparrot_training - Step 5334: {'lr': 0.00048688967849971654, 'samples': 2731520, 'steps': 5334, 'batch_loss/train': 0.8806298803538084} +12/23/2021 16:32:52 - INFO - codeparrot_training - Step 5335: {'lr': 0.00048688464041908407, 'samples': 2732032, 'steps': 5335, 'batch_loss/train': 0.7133628227747977} +12/23/2021 16:33:03 - INFO - codeparrot_training - Step 5336: {'lr': 0.0004868796013966877, 'samples': 2732544, 'steps': 5336, 'batch_loss/train': 0.8815459599718451} +12/23/2021 16:33:16 - INFO - codeparrot_training - Step 5337: {'lr': 0.0004868745614325475, 'samples': 2733056, 'steps': 5337, 'batch_loss/train': 0.8822910580784082} +12/23/2021 16:33:27 - INFO - codeparrot_training - Step 5338: {'lr': 0.0004868695205266835, 'samples': 2733568, 'steps': 5338, 'batch_loss/train': 0.9289957279106602} +12/23/2021 16:33:37 - INFO - codeparrot_training - Step 5339: {'lr': 0.00048686447867911576, 'samples': 2734080, 'steps': 5339, 'batch_loss/train': 0.7540243096882477} +12/23/2021 16:33:49 - INFO - codeparrot_training - Step 5340: {'lr': 0.00048685943588986426, 'samples': 2734592, 'steps': 5340, 'batch_loss/train': 0.9539390387944877} +12/23/2021 16:34:00 - INFO - codeparrot_training - Step 5341: {'lr': 0.000486854392158949, 'samples': 2735104, 'steps': 5341, 'batch_loss/train': 0.8150996305048466} +12/23/2021 16:34:10 - INFO - codeparrot_training - Step 5342: {'lr': 0.0004868493474863902, 'samples': 2735616, 'steps': 5342, 'batch_loss/train': 0.9241571687161922} +12/23/2021 16:34:21 - INFO - codeparrot_training - Step 5343: {'lr': 0.00048684430187220773, 'samples': 2736128, 'steps': 5343, 'batch_loss/train': 0.870395075995475} +12/23/2021 16:34:33 - INFO - codeparrot_training - Step 5344: {'lr': 0.0004868392553164218, 'samples': 2736640, 'steps': 5344, 'batch_loss/train': 0.9121706038713455} +12/23/2021 16:34:43 - INFO - codeparrot_training - Step 5345: {'lr': 0.0004868342078190524, 'samples': 2737152, 'steps': 5345, 'batch_loss/train': 0.9251508666202426} +12/23/2021 16:34:54 - INFO - codeparrot_training - Step 5346: {'lr': 0.00048682915938011965, 'samples': 2737664, 'steps': 5346, 'batch_loss/train': 0.8784493701532483} +12/23/2021 16:35:07 - INFO - codeparrot_training - Step 5347: {'lr': 0.00048682410999964363, 'samples': 2738176, 'steps': 5347, 'batch_loss/train': 0.8046607514843345} +12/23/2021 16:35:18 - INFO - codeparrot_training - Step 5348: {'lr': 0.00048681905967764427, 'samples': 2738688, 'steps': 5348, 'batch_loss/train': 0.8216858748346567} +12/23/2021 16:35:28 - INFO - codeparrot_training - Step 5349: {'lr': 0.0004868140084141418, 'samples': 2739200, 'steps': 5349, 'batch_loss/train': 0.7445935884024948} +12/23/2021 16:35:40 - INFO - codeparrot_training - Step 5350: {'lr': 0.0004868089562091562, 'samples': 2739712, 'steps': 5350, 'batch_loss/train': 0.8821705225855112} +12/23/2021 16:35:51 - INFO - codeparrot_training - Step 5351: {'lr': 0.00048680390306270774, 'samples': 2740224, 'steps': 5351, 'batch_loss/train': 0.9015653925016522} +12/23/2021 16:36:01 - INFO - codeparrot_training - Step 5352: {'lr': 0.00048679884897481627, 'samples': 2740736, 'steps': 5352, 'batch_loss/train': 0.816592394374311} +12/23/2021 16:36:12 - INFO - codeparrot_training - Step 5353: {'lr': 0.00048679379394550197, 'samples': 2741248, 'steps': 5353, 'batch_loss/train': 0.8538243742659688} +12/23/2021 16:36:25 - INFO - codeparrot_training - Step 5354: {'lr': 0.00048678873797478495, 'samples': 2741760, 'steps': 5354, 'batch_loss/train': 0.7908783396705985} +12/23/2021 16:36:36 - INFO - codeparrot_training - Step 5355: {'lr': 0.0004867836810626853, 'samples': 2742272, 'steps': 5355, 'batch_loss/train': 0.9454667512327433} +12/23/2021 16:36:46 - INFO - codeparrot_training - Step 5356: {'lr': 0.00048677862320922326, 'samples': 2742784, 'steps': 5356, 'batch_loss/train': 0.9546548193320632} +12/23/2021 16:36:58 - INFO - codeparrot_training - Step 5357: {'lr': 0.00048677356441441876, 'samples': 2743296, 'steps': 5357, 'batch_loss/train': 0.9479739340022206} +12/23/2021 16:37:09 - INFO - codeparrot_training - Step 5358: {'lr': 0.00048676850467829194, 'samples': 2743808, 'steps': 5358, 'batch_loss/train': 0.8547979686409235} +12/23/2021 16:37:20 - INFO - codeparrot_training - Step 5359: {'lr': 0.00048676344400086296, 'samples': 2744320, 'steps': 5359, 'batch_loss/train': 0.8723630853928626} +12/23/2021 16:37:32 - INFO - codeparrot_training - Step 5360: {'lr': 0.000486758382382152, 'samples': 2744832, 'steps': 5360, 'batch_loss/train': 0.8847291991114616} +12/23/2021 16:37:42 - INFO - codeparrot_training - Step 5361: {'lr': 0.000486753319822179, 'samples': 2745344, 'steps': 5361, 'batch_loss/train': 0.8829949535429478} +12/23/2021 16:37:53 - INFO - codeparrot_training - Step 5362: {'lr': 0.0004867482563209642, 'samples': 2745856, 'steps': 5362, 'batch_loss/train': 0.9154328349977732} +12/23/2021 16:38:03 - INFO - codeparrot_training - Step 5363: {'lr': 0.00048674319187852775, 'samples': 2746368, 'steps': 5363, 'batch_loss/train': 0.9211330087855458} +12/23/2021 16:38:17 - INFO - codeparrot_training - Step 5364: {'lr': 0.00048673812649488977, 'samples': 2746880, 'steps': 5364, 'batch_loss/train': 0.8377491421997547} +12/23/2021 16:38:28 - INFO - codeparrot_training - Step 5365: {'lr': 0.0004867330601700704, 'samples': 2747392, 'steps': 5365, 'batch_loss/train': 0.910118673928082} +12/23/2021 16:38:38 - INFO - codeparrot_training - Step 5366: {'lr': 0.00048672799290408973, 'samples': 2747904, 'steps': 5366, 'batch_loss/train': 0.8357130810618401} +12/23/2021 16:38:51 - INFO - codeparrot_training - Step 5367: {'lr': 0.000486722924696968, 'samples': 2748416, 'steps': 5367, 'batch_loss/train': 1.381870148703456} +12/23/2021 16:39:01 - INFO - codeparrot_training - Step 5368: {'lr': 0.0004867178555487253, 'samples': 2748928, 'steps': 5368, 'batch_loss/train': 0.8311923081055284} +12/23/2021 16:39:12 - INFO - codeparrot_training - Step 5369: {'lr': 0.00048671278545938177, 'samples': 2749440, 'steps': 5369, 'batch_loss/train': 0.8073940454050899} +12/23/2021 16:39:25 - INFO - codeparrot_training - Step 5370: {'lr': 0.0004867077144289576, 'samples': 2749952, 'steps': 5370, 'batch_loss/train': 0.6093897209502757} +12/23/2021 16:39:35 - INFO - codeparrot_training - Step 5371: {'lr': 0.00048670264245747284, 'samples': 2750464, 'steps': 5371, 'batch_loss/train': 0.8094451702199876} +12/23/2021 16:39:46 - INFO - codeparrot_training - Step 5372: {'lr': 0.0004866975695449478, 'samples': 2750976, 'steps': 5372, 'batch_loss/train': 0.7492595096118748} +12/23/2021 16:39:56 - INFO - codeparrot_training - Step 5373: {'lr': 0.00048669249569140263, 'samples': 2751488, 'steps': 5373, 'batch_loss/train': 0.8816837603226304} +12/23/2021 16:40:08 - INFO - codeparrot_training - Step 5374: {'lr': 0.0004866874208968575, 'samples': 2752000, 'steps': 5374, 'batch_loss/train': 0.8495887336321175} +12/23/2021 16:40:19 - INFO - codeparrot_training - Step 5375: {'lr': 0.00048668234516133246, 'samples': 2752512, 'steps': 5375, 'batch_loss/train': 0.9044250212609768} +12/23/2021 16:40:29 - INFO - codeparrot_training - Step 5376: {'lr': 0.00048667726848484783, 'samples': 2753024, 'steps': 5376, 'batch_loss/train': 0.874465717934072} +12/23/2021 16:40:43 - INFO - codeparrot_training - Step 5377: {'lr': 0.00048667219086742376, 'samples': 2753536, 'steps': 5377, 'batch_loss/train': 0.8486524485051632} +12/23/2021 16:40:54 - INFO - codeparrot_training - Step 5378: {'lr': 0.0004866671123090805, 'samples': 2754048, 'steps': 5378, 'batch_loss/train': 0.7860046112909913} +12/23/2021 16:41:04 - INFO - codeparrot_training - Step 5379: {'lr': 0.00048666203280983803, 'samples': 2754560, 'steps': 5379, 'batch_loss/train': 0.8727233214303851} +12/23/2021 16:41:16 - INFO - codeparrot_training - Step 5380: {'lr': 0.0004866569523697167, 'samples': 2755072, 'steps': 5380, 'batch_loss/train': 0.8357019661925733} +12/23/2021 16:41:27 - INFO - codeparrot_training - Step 5381: {'lr': 0.0004866518709887367, 'samples': 2755584, 'steps': 5381, 'batch_loss/train': 0.8919831872917712} +12/23/2021 16:41:37 - INFO - codeparrot_training - Step 5382: {'lr': 0.00048664678866691826, 'samples': 2756096, 'steps': 5382, 'batch_loss/train': 0.891495100222528} +12/23/2021 16:41:50 - INFO - codeparrot_training - Step 5383: {'lr': 0.0004866417054042815, 'samples': 2756608, 'steps': 5383, 'batch_loss/train': 0.8914495501667261} +12/23/2021 16:42:01 - INFO - codeparrot_training - Step 5384: {'lr': 0.00048663662120084666, 'samples': 2757120, 'steps': 5384, 'batch_loss/train': 0.8325051832944155} +12/23/2021 16:42:12 - INFO - codeparrot_training - Step 5385: {'lr': 0.00048663153605663404, 'samples': 2757632, 'steps': 5385, 'batch_loss/train': 0.8300542468205094} +12/23/2021 16:42:22 - INFO - codeparrot_training - Step 5386: {'lr': 0.0004866264499716637, 'samples': 2758144, 'steps': 5386, 'batch_loss/train': 0.8701753905043006} +12/23/2021 16:42:34 - INFO - codeparrot_training - Step 5387: {'lr': 0.0004866213629459559, 'samples': 2758656, 'steps': 5387, 'batch_loss/train': 0.8301691859960556} +12/23/2021 16:42:45 - INFO - codeparrot_training - Step 5388: {'lr': 0.00048661627497953105, 'samples': 2759168, 'steps': 5388, 'batch_loss/train': 0.8837770805694163} +12/23/2021 16:42:55 - INFO - codeparrot_training - Step 5389: {'lr': 0.0004866111860724092, 'samples': 2759680, 'steps': 5389, 'batch_loss/train': 0.8745271973311901} +12/23/2021 16:43:07 - INFO - codeparrot_training - Step 5390: {'lr': 0.00048660609622461047, 'samples': 2760192, 'steps': 5390, 'batch_loss/train': 0.867006110958755} +12/23/2021 16:43:18 - INFO - codeparrot_training - Step 5391: {'lr': 0.00048660100543615535, 'samples': 2760704, 'steps': 5391, 'batch_loss/train': 0.8336112545803189} +12/23/2021 16:43:29 - INFO - codeparrot_training - Step 5392: {'lr': 0.00048659591370706404, 'samples': 2761216, 'steps': 5392, 'batch_loss/train': 0.883025829680264} +12/23/2021 16:43:42 - INFO - codeparrot_training - Step 5393: {'lr': 0.00048659082103735655, 'samples': 2761728, 'steps': 5393, 'batch_loss/train': 0.7802399541251361} +12/23/2021 16:43:53 - INFO - codeparrot_training - Step 5394: {'lr': 0.0004865857274270534, 'samples': 2762240, 'steps': 5394, 'batch_loss/train': 0.8684125831350684} +12/23/2021 16:44:03 - INFO - codeparrot_training - Step 5395: {'lr': 0.0004865806328761747, 'samples': 2762752, 'steps': 5395, 'batch_loss/train': 0.844222305342555} +12/23/2021 16:44:14 - INFO - codeparrot_training - Step 5396: {'lr': 0.00048657553738474076, 'samples': 2763264, 'steps': 5396, 'batch_loss/train': 0.9278980467934161} +12/23/2021 16:44:26 - INFO - codeparrot_training - Step 5397: {'lr': 0.0004865704409527718, 'samples': 2763776, 'steps': 5397, 'batch_loss/train': 0.8894483372569084} +12/23/2021 16:44:37 - INFO - codeparrot_training - Step 5398: {'lr': 0.00048656534358028803, 'samples': 2764288, 'steps': 5398, 'batch_loss/train': 0.8530466109514236} +12/23/2021 16:44:47 - INFO - codeparrot_training - Step 5399: {'lr': 0.00048656024526730986, 'samples': 2764800, 'steps': 5399, 'batch_loss/train': 0.7062849182402715} +12/23/2021 16:45:01 - INFO - codeparrot_training - Step 5400: {'lr': 0.0004865551460138574, 'samples': 2765312, 'steps': 5400, 'batch_loss/train': 0.839220954105258} +12/23/2021 16:45:11 - INFO - codeparrot_training - Step 5401: {'lr': 0.0004865500458199511, 'samples': 2765824, 'steps': 5401, 'batch_loss/train': 0.8407015525735915} +12/23/2021 16:45:22 - INFO - codeparrot_training - Step 5402: {'lr': 0.00048654494468561104, 'samples': 2766336, 'steps': 5402, 'batch_loss/train': 0.7057989879976958} +12/23/2021 16:45:32 - INFO - codeparrot_training - Step 5403: {'lr': 0.00048653984261085763, 'samples': 2766848, 'steps': 5403, 'batch_loss/train': 0.7950535393320024} +12/23/2021 16:45:44 - INFO - codeparrot_training - Step 5404: {'lr': 0.00048653473959571117, 'samples': 2767360, 'steps': 5404, 'batch_loss/train': 0.8963530790060759} +12/23/2021 16:45:55 - INFO - codeparrot_training - Step 5405: {'lr': 0.0004865296356401918, 'samples': 2767872, 'steps': 5405, 'batch_loss/train': 0.8151414263993502} +12/23/2021 16:46:06 - INFO - codeparrot_training - Step 5406: {'lr': 0.00048652453074432, 'samples': 2768384, 'steps': 5406, 'batch_loss/train': 0.8070954130962491} +12/23/2021 16:46:18 - INFO - codeparrot_training - Step 5407: {'lr': 0.00048651942490811594, 'samples': 2768896, 'steps': 5407, 'batch_loss/train': 0.8110774857923388} +12/23/2021 16:46:28 - INFO - codeparrot_training - Step 5408: {'lr': 0.00048651431813159996, 'samples': 2769408, 'steps': 5408, 'batch_loss/train': 0.9703455241397023} +12/23/2021 16:46:39 - INFO - codeparrot_training - Step 5409: {'lr': 0.00048650921041479235, 'samples': 2769920, 'steps': 5409, 'batch_loss/train': 0.8798171691596508} +12/23/2021 16:46:51 - INFO - codeparrot_training - Step 5410: {'lr': 0.0004865041017577134, 'samples': 2770432, 'steps': 5410, 'batch_loss/train': 0.7630250910297036} +12/23/2021 16:47:02 - INFO - codeparrot_training - Step 5411: {'lr': 0.0004864989921603834, 'samples': 2770944, 'steps': 5411, 'batch_loss/train': 0.812802659580484} +12/23/2021 16:47:12 - INFO - codeparrot_training - Step 5412: {'lr': 0.0004864938816228228, 'samples': 2771456, 'steps': 5412, 'batch_loss/train': 0.6832397856051102} +12/23/2021 16:47:25 - INFO - codeparrot_training - Step 5413: {'lr': 0.00048648877014505177, 'samples': 2771968, 'steps': 5413, 'batch_loss/train': 0.8876504842191935} +12/23/2021 16:47:36 - INFO - codeparrot_training - Step 5414: {'lr': 0.0004864836577270907, 'samples': 2772480, 'steps': 5414, 'batch_loss/train': 0.8502401326550171} +12/23/2021 16:47:47 - INFO - codeparrot_training - Step 5415: {'lr': 0.0004864785443689599, 'samples': 2772992, 'steps': 5415, 'batch_loss/train': 0.9055342599749565} +12/23/2021 16:47:57 - INFO - codeparrot_training - Step 5416: {'lr': 0.00048647343007067966, 'samples': 2773504, 'steps': 5416, 'batch_loss/train': 0.9210329754278064} +12/23/2021 16:48:09 - INFO - codeparrot_training - Step 5417: {'lr': 0.00048646831483227043, 'samples': 2774016, 'steps': 5417, 'batch_loss/train': 0.8579631838947535} +12/23/2021 16:48:20 - INFO - codeparrot_training - Step 5418: {'lr': 0.00048646319865375243, 'samples': 2774528, 'steps': 5418, 'batch_loss/train': 0.8380297459661961} +12/23/2021 16:48:30 - INFO - codeparrot_training - Step 5419: {'lr': 0.00048645808153514604, 'samples': 2775040, 'steps': 5419, 'batch_loss/train': 0.8629515366628766} +12/23/2021 16:48:43 - INFO - codeparrot_training - Step 5420: {'lr': 0.00048645296347647163, 'samples': 2775552, 'steps': 5420, 'batch_loss/train': 0.7522060924675316} +12/23/2021 16:48:53 - INFO - codeparrot_training - Step 5421: {'lr': 0.00048644784447774943, 'samples': 2776064, 'steps': 5421, 'batch_loss/train': 1.2231357749551535} +12/23/2021 16:49:04 - INFO - codeparrot_training - Step 5422: {'lr': 0.00048644272453899994, 'samples': 2776576, 'steps': 5422, 'batch_loss/train': 0.8381822789087892} +12/23/2021 16:49:17 - INFO - codeparrot_training - Step 5423: {'lr': 0.0004864376036602435, 'samples': 2777088, 'steps': 5423, 'batch_loss/train': 0.8051809086464345} +12/23/2021 16:49:28 - INFO - codeparrot_training - Step 5424: {'lr': 0.00048643248184150037, 'samples': 2777600, 'steps': 5424, 'batch_loss/train': 0.9188210293650627} +12/23/2021 16:49:38 - INFO - codeparrot_training - Step 5425: {'lr': 0.0004864273590827909, 'samples': 2778112, 'steps': 5425, 'batch_loss/train': 0.8744485639035702} +12/23/2021 16:49:49 - INFO - codeparrot_training - Step 5426: {'lr': 0.0004864222353841356, 'samples': 2778624, 'steps': 5426, 'batch_loss/train': 0.9106654766947031} +12/23/2021 16:50:01 - INFO - codeparrot_training - Step 5427: {'lr': 0.0004864171107455547, 'samples': 2779136, 'steps': 5427, 'batch_loss/train': 0.958432037383318} +12/23/2021 16:50:12 - INFO - codeparrot_training - Step 5428: {'lr': 0.00048641198516706876, 'samples': 2779648, 'steps': 5428, 'batch_loss/train': 0.8499084641225636} +12/23/2021 16:50:22 - INFO - codeparrot_training - Step 5429: {'lr': 0.0004864068586486979, 'samples': 2780160, 'steps': 5429, 'batch_loss/train': 0.9809244927018881} +12/23/2021 16:50:35 - INFO - codeparrot_training - Step 5430: {'lr': 0.0004864017311904627, 'samples': 2780672, 'steps': 5430, 'batch_loss/train': 0.9190212339162827} +12/23/2021 16:50:45 - INFO - codeparrot_training - Step 5431: {'lr': 0.0004863966027923835, 'samples': 2781184, 'steps': 5431, 'batch_loss/train': 0.879918996244669} +12/23/2021 16:50:56 - INFO - codeparrot_training - Step 5432: {'lr': 0.00048639147345448067, 'samples': 2781696, 'steps': 5432, 'batch_loss/train': 0.8900058441795409} +12/23/2021 16:51:09 - INFO - codeparrot_training - Step 5433: {'lr': 0.00048638634317677454, 'samples': 2782208, 'steps': 5433, 'batch_loss/train': 0.9628579961135983} +12/23/2021 16:51:19 - INFO - codeparrot_training - Step 5434: {'lr': 0.00048638121195928555, 'samples': 2782720, 'steps': 5434, 'batch_loss/train': 0.9142745798453689} +12/23/2021 16:51:30 - INFO - codeparrot_training - Step 5435: {'lr': 0.00048637607980203413, 'samples': 2783232, 'steps': 5435, 'batch_loss/train': 1.0297857634723186} +12/23/2021 16:51:41 - INFO - codeparrot_training - Step 5436: {'lr': 0.0004863709467050407, 'samples': 2783744, 'steps': 5436, 'batch_loss/train': 0.785022190771997} +12/23/2021 16:51:53 - INFO - codeparrot_training - Step 5437: {'lr': 0.0004863658126683257, 'samples': 2784256, 'steps': 5437, 'batch_loss/train': 0.847307794727385} +12/23/2021 16:52:03 - INFO - codeparrot_training - Step 5438: {'lr': 0.0004863606776919094, 'samples': 2784768, 'steps': 5438, 'batch_loss/train': 0.7921666195616126} +12/23/2021 16:52:14 - INFO - codeparrot_training - Step 5439: {'lr': 0.0004863555417758123, 'samples': 2785280, 'steps': 5439, 'batch_loss/train': 0.7908816831186414} +12/23/2021 16:52:26 - INFO - codeparrot_training - Step 5440: {'lr': 0.00048635040492005477, 'samples': 2785792, 'steps': 5440, 'batch_loss/train': 0.7705679815262556} +12/23/2021 16:52:37 - INFO - codeparrot_training - Step 5441: {'lr': 0.0004863452671246573, 'samples': 2786304, 'steps': 5441, 'batch_loss/train': 0.8456028830260038} +12/23/2021 16:52:47 - INFO - codeparrot_training - Step 5442: {'lr': 0.00048634012838964037, 'samples': 2786816, 'steps': 5442, 'batch_loss/train': 0.8654043255373836} +12/23/2021 16:53:01 - INFO - codeparrot_training - Step 5443: {'lr': 0.00048633498871502424, 'samples': 2787328, 'steps': 5443, 'batch_loss/train': 0.7514548627659678} +12/23/2021 16:53:12 - INFO - codeparrot_training - Step 5444: {'lr': 0.0004863298481008295, 'samples': 2787840, 'steps': 5444, 'batch_loss/train': 0.763471735175699} +12/23/2021 16:53:22 - INFO - codeparrot_training - Step 5445: {'lr': 0.0004863247065470764, 'samples': 2788352, 'steps': 5445, 'batch_loss/train': 0.8640974806621671} +12/23/2021 16:53:33 - INFO - codeparrot_training - Step 5446: {'lr': 0.0004863195640537856, 'samples': 2788864, 'steps': 5446, 'batch_loss/train': 0.8433566894382238} +12/23/2021 16:53:45 - INFO - codeparrot_training - Step 5447: {'lr': 0.0004863144206209774, 'samples': 2789376, 'steps': 5447, 'batch_loss/train': 0.7550083799287677} +12/23/2021 16:53:55 - INFO - codeparrot_training - Step 5448: {'lr': 0.00048630927624867225, 'samples': 2789888, 'steps': 5448, 'batch_loss/train': 0.8935807971283793} +12/23/2021 16:54:06 - INFO - codeparrot_training - Step 5449: {'lr': 0.00048630413093689074, 'samples': 2790400, 'steps': 5449, 'batch_loss/train': 0.852018719073385} +12/23/2021 16:54:18 - INFO - codeparrot_training - Step 5450: {'lr': 0.0004862989846856532, 'samples': 2790912, 'steps': 5450, 'batch_loss/train': 0.7957274955697358} +12/23/2021 16:54:29 - INFO - codeparrot_training - Step 5451: {'lr': 0.0004862938374949801, 'samples': 2791424, 'steps': 5451, 'batch_loss/train': 0.9076069789007306} +12/23/2021 16:54:39 - INFO - codeparrot_training - Step 5452: {'lr': 0.00048628868936489197, 'samples': 2791936, 'steps': 5452, 'batch_loss/train': 0.9166996702551842} +12/23/2021 16:54:50 - INFO - codeparrot_training - Step 5453: {'lr': 0.00048628354029540917, 'samples': 2792448, 'steps': 5453, 'batch_loss/train': 0.908805008046329} +12/23/2021 16:55:03 - INFO - codeparrot_training - Step 5454: {'lr': 0.0004862783902865523, 'samples': 2792960, 'steps': 5454, 'batch_loss/train': 0.8838750598952174} +12/23/2021 16:55:14 - INFO - codeparrot_training - Step 5455: {'lr': 0.0004862732393383417, 'samples': 2793472, 'steps': 5455, 'batch_loss/train': 0.8492619628086686} +12/23/2021 16:55:24 - INFO - codeparrot_training - Step 5456: {'lr': 0.0004862680874507979, 'samples': 2793984, 'steps': 5456, 'batch_loss/train': 0.7578277804423124} +12/23/2021 16:55:36 - INFO - codeparrot_training - Step 5457: {'lr': 0.00048626293462394145, 'samples': 2794496, 'steps': 5457, 'batch_loss/train': 0.8438782002776861} +12/23/2021 16:55:47 - INFO - codeparrot_training - Step 5458: {'lr': 0.0004862577808577928, 'samples': 2795008, 'steps': 5458, 'batch_loss/train': 0.8565883748233318} +12/23/2021 16:55:58 - INFO - codeparrot_training - Step 5459: {'lr': 0.0004862526261523723, 'samples': 2795520, 'steps': 5459, 'batch_loss/train': 0.9030020087957382} +12/23/2021 16:56:11 - INFO - codeparrot_training - Step 5460: {'lr': 0.0004862474705077007, 'samples': 2796032, 'steps': 5460, 'batch_loss/train': 0.9739807085134089} +12/23/2021 16:56:21 - INFO - codeparrot_training - Step 5461: {'lr': 0.00048624231392379826, 'samples': 2796544, 'steps': 5461, 'batch_loss/train': 0.9053991818800569} +12/23/2021 16:56:32 - INFO - codeparrot_training - Step 5462: {'lr': 0.0004862371564006857, 'samples': 2797056, 'steps': 5462, 'batch_loss/train': 0.983168913051486} +12/23/2021 16:56:44 - INFO - codeparrot_training - Step 5463: {'lr': 0.0004862319979383833, 'samples': 2797568, 'steps': 5463, 'batch_loss/train': 0.9225147999823093} +12/23/2021 16:56:54 - INFO - codeparrot_training - Step 5464: {'lr': 0.0004862268385369117, 'samples': 2798080, 'steps': 5464, 'batch_loss/train': 0.8815447636879981} +12/23/2021 16:57:05 - INFO - codeparrot_training - Step 5465: {'lr': 0.0004862216781962914, 'samples': 2798592, 'steps': 5465, 'batch_loss/train': 0.7933923387899995} +12/23/2021 16:57:16 - INFO - codeparrot_training - Step 5466: {'lr': 0.0004862165169165429, 'samples': 2799104, 'steps': 5466, 'batch_loss/train': 0.9087414192035794} +12/23/2021 16:57:28 - INFO - codeparrot_training - Step 5467: {'lr': 0.00048621135469768676, 'samples': 2799616, 'steps': 5467, 'batch_loss/train': 0.8818594589829445} +12/23/2021 16:57:39 - INFO - codeparrot_training - Step 5468: {'lr': 0.0004862061915397433, 'samples': 2800128, 'steps': 5468, 'batch_loss/train': 0.9297546176239848} +12/23/2021 16:57:49 - INFO - codeparrot_training - Step 5469: {'lr': 0.00048620102744273343, 'samples': 2800640, 'steps': 5469, 'batch_loss/train': 0.8880697530694306} +12/23/2021 16:58:02 - INFO - codeparrot_training - Step 5470: {'lr': 0.00048619586240667735, 'samples': 2801152, 'steps': 5470, 'batch_loss/train': 0.8764357380568981} +12/23/2021 16:58:13 - INFO - codeparrot_training - Step 5471: {'lr': 0.00048619069643159567, 'samples': 2801664, 'steps': 5471, 'batch_loss/train': 0.9405564125627279} +12/23/2021 16:58:24 - INFO - codeparrot_training - Step 5472: {'lr': 0.00048618552951750903, 'samples': 2802176, 'steps': 5472, 'batch_loss/train': 0.8350269668735564} +12/23/2021 16:58:36 - INFO - codeparrot_training - Step 5473: {'lr': 0.0004861803616644379, 'samples': 2802688, 'steps': 5473, 'batch_loss/train': 0.8527972493320704} +12/23/2021 16:58:46 - INFO - codeparrot_training - Step 5474: {'lr': 0.0004861751928724028, 'samples': 2803200, 'steps': 5474, 'batch_loss/train': 0.8787162788212299} +12/23/2021 16:58:57 - INFO - codeparrot_training - Step 5475: {'lr': 0.00048617002314142434, 'samples': 2803712, 'steps': 5475, 'batch_loss/train': 0.8552487371489406} +12/23/2021 16:59:08 - INFO - codeparrot_training - Step 5476: {'lr': 0.000486164852471523, 'samples': 2804224, 'steps': 5476, 'batch_loss/train': 0.9269188800826669} +12/23/2021 16:59:20 - INFO - codeparrot_training - Step 5477: {'lr': 0.00048615968086271945, 'samples': 2804736, 'steps': 5477, 'batch_loss/train': 0.8380448473617435} +12/23/2021 16:59:30 - INFO - codeparrot_training - Step 5478: {'lr': 0.0004861545083150341, 'samples': 2805248, 'steps': 5478, 'batch_loss/train': 0.7759979721158743} +12/23/2021 16:59:41 - INFO - codeparrot_training - Step 5479: {'lr': 0.00048614933482848765, 'samples': 2805760, 'steps': 5479, 'batch_loss/train': 0.8907184517011046} +12/23/2021 16:59:53 - INFO - codeparrot_training - Step 5480: {'lr': 0.00048614416040310056, 'samples': 2806272, 'steps': 5480, 'batch_loss/train': 0.8324821889400482} +12/23/2021 17:00:03 - INFO - codeparrot_training - Step 5481: {'lr': 0.0004861389850388935, 'samples': 2806784, 'steps': 5481, 'batch_loss/train': 0.8123818365857005} +12/23/2021 17:00:14 - INFO - codeparrot_training - Step 5482: {'lr': 0.000486133808735887, 'samples': 2807296, 'steps': 5482, 'batch_loss/train': 0.9717820845544338} +12/23/2021 17:00:27 - INFO - codeparrot_training - Step 5483: {'lr': 0.0004861286314941016, 'samples': 2807808, 'steps': 5483, 'batch_loss/train': 0.8779890909790993} +12/23/2021 17:00:38 - INFO - codeparrot_training - Step 5484: {'lr': 0.0004861234533135579, 'samples': 2808320, 'steps': 5484, 'batch_loss/train': 0.8981337123550475} +12/23/2021 17:00:49 - INFO - codeparrot_training - Step 5485: {'lr': 0.00048611827419427656, 'samples': 2808832, 'steps': 5485, 'batch_loss/train': 0.9462939174845815} +12/23/2021 17:01:01 - INFO - codeparrot_training - Step 5486: {'lr': 0.00048611309413627813, 'samples': 2809344, 'steps': 5486, 'batch_loss/train': 0.8503827666863799} +12/23/2021 17:01:11 - INFO - codeparrot_training - Step 5487: {'lr': 0.00048610791313958315, 'samples': 2809856, 'steps': 5487, 'batch_loss/train': 0.8227664027363062} +12/23/2021 17:01:22 - INFO - codeparrot_training - Step 5488: {'lr': 0.0004861027312042122, 'samples': 2810368, 'steps': 5488, 'batch_loss/train': 0.8755974061787128} +12/23/2021 17:01:33 - INFO - codeparrot_training - Step 5489: {'lr': 0.00048609754833018606, 'samples': 2810880, 'steps': 5489, 'batch_loss/train': 0.8333402005955577} +12/23/2021 17:01:45 - INFO - codeparrot_training - Step 5490: {'lr': 0.00048609236451752514, 'samples': 2811392, 'steps': 5490, 'batch_loss/train': 0.8436583462171257} +12/23/2021 17:01:56 - INFO - codeparrot_training - Step 5491: {'lr': 0.0004860871797662502, 'samples': 2811904, 'steps': 5491, 'batch_loss/train': 0.9671539906412363} +12/23/2021 17:02:06 - INFO - codeparrot_training - Step 5492: {'lr': 0.00048608199407638173, 'samples': 2812416, 'steps': 5492, 'batch_loss/train': 0.593134640250355} +12/23/2021 17:02:19 - INFO - codeparrot_training - Step 5493: {'lr': 0.0004860768074479404, 'samples': 2812928, 'steps': 5493, 'batch_loss/train': 0.8570374399423599} +12/23/2021 17:02:30 - INFO - codeparrot_training - Step 5494: {'lr': 0.00048607161988094673, 'samples': 2813440, 'steps': 5494, 'batch_loss/train': 0.8861780175939202} +12/23/2021 17:02:40 - INFO - codeparrot_training - Step 5495: {'lr': 0.00048606643137542155, 'samples': 2813952, 'steps': 5495, 'batch_loss/train': 0.8693936923518777} +12/23/2021 17:02:53 - INFO - codeparrot_training - Step 5496: {'lr': 0.00048606124193138533, 'samples': 2814464, 'steps': 5496, 'batch_loss/train': 0.8396357074379921} +12/23/2021 17:03:03 - INFO - codeparrot_training - Step 5497: {'lr': 0.0004860560515488588, 'samples': 2814976, 'steps': 5497, 'batch_loss/train': 0.9266094658523798} +12/23/2021 17:03:14 - INFO - codeparrot_training - Step 5498: {'lr': 0.0004860508602278625, 'samples': 2815488, 'steps': 5498, 'batch_loss/train': 0.9052289724349976} +12/23/2021 17:03:24 - INFO - codeparrot_training - Step 5499: {'lr': 0.00048604566796841707, 'samples': 2816000, 'steps': 5499, 'batch_loss/train': 0.7189815999008715} +12/23/2021 17:03:38 - INFO - codeparrot_training - Step 5500: {'lr': 0.0004860404747705433, 'samples': 2816512, 'steps': 5500, 'batch_loss/train': 0.8741805134341121} +12/23/2021 17:03:48 - INFO - codeparrot_training - Step 5501: {'lr': 0.0004860352806342616, 'samples': 2817024, 'steps': 5501, 'batch_loss/train': 1.8579393615946174} +12/23/2021 17:03:59 - INFO - codeparrot_training - Step 5502: {'lr': 0.00048603008555959276, 'samples': 2817536, 'steps': 5502, 'batch_loss/train': 0.8429241038393229} +12/23/2021 17:04:11 - INFO - codeparrot_training - Step 5503: {'lr': 0.00048602488954655745, 'samples': 2818048, 'steps': 5503, 'batch_loss/train': 0.8866462865844369} +12/23/2021 17:04:22 - INFO - codeparrot_training - Step 5504: {'lr': 0.00048601969259517633, 'samples': 2818560, 'steps': 5504, 'batch_loss/train': 0.8725893534719944} +12/23/2021 17:04:32 - INFO - codeparrot_training - Step 5505: {'lr': 0.00048601449470547, 'samples': 2819072, 'steps': 5505, 'batch_loss/train': 0.8743888004682958} +12/23/2021 17:04:44 - INFO - codeparrot_training - Step 5506: {'lr': 0.00048600929587745917, 'samples': 2819584, 'steps': 5506, 'batch_loss/train': 0.9318708442151546} +12/23/2021 17:04:55 - INFO - codeparrot_training - Step 5507: {'lr': 0.0004860040961111645, 'samples': 2820096, 'steps': 5507, 'batch_loss/train': 0.8723258678801358} +12/23/2021 17:05:05 - INFO - codeparrot_training - Step 5508: {'lr': 0.0004859988954066067, 'samples': 2820608, 'steps': 5508, 'batch_loss/train': 1.011260747909546} +12/23/2021 17:05:16 - INFO - codeparrot_training - Step 5509: {'lr': 0.0004859936937638063, 'samples': 2821120, 'steps': 5509, 'batch_loss/train': 0.8856195937842131} +12/23/2021 17:05:29 - INFO - codeparrot_training - Step 5510: {'lr': 0.0004859884911827841, 'samples': 2821632, 'steps': 5510, 'batch_loss/train': 0.9002292226068676} +12/23/2021 17:05:40 - INFO - codeparrot_training - Step 5511: {'lr': 0.00048598328766356083, 'samples': 2822144, 'steps': 5511, 'batch_loss/train': 0.8228402028325945} +12/23/2021 17:05:50 - INFO - codeparrot_training - Step 5512: {'lr': 0.00048597808320615705, 'samples': 2822656, 'steps': 5512, 'batch_loss/train': 0.8222105987370014} +12/23/2021 17:06:03 - INFO - codeparrot_training - Step 5513: {'lr': 0.0004859728778105935, 'samples': 2823168, 'steps': 5513, 'batch_loss/train': 0.8975968901067972} +12/23/2021 17:06:13 - INFO - codeparrot_training - Step 5514: {'lr': 0.000485967671476891, 'samples': 2823680, 'steps': 5514, 'batch_loss/train': 0.8824088722467422} +12/23/2021 17:06:24 - INFO - codeparrot_training - Step 5515: {'lr': 0.00048596246420507005, 'samples': 2824192, 'steps': 5515, 'batch_loss/train': 0.8070039702579379} +12/23/2021 17:06:36 - INFO - codeparrot_training - Step 5516: {'lr': 0.00048595725599515137, 'samples': 2824704, 'steps': 5516, 'batch_loss/train': 0.8514657085761428} +12/23/2021 17:06:46 - INFO - codeparrot_training - Step 5517: {'lr': 0.0004859520468471559, 'samples': 2825216, 'steps': 5517, 'batch_loss/train': 0.8330312790349126} +12/23/2021 17:06:57 - INFO - codeparrot_training - Step 5518: {'lr': 0.00048594683676110405, 'samples': 2825728, 'steps': 5518, 'batch_loss/train': 0.8987758485600352} +12/23/2021 17:07:09 - INFO - codeparrot_training - Step 5519: {'lr': 0.0004859416257370167, 'samples': 2826240, 'steps': 5519, 'batch_loss/train': 0.819471298949793} +12/23/2021 17:07:20 - INFO - codeparrot_training - Step 5520: {'lr': 0.0004859364137749146, 'samples': 2826752, 'steps': 5520, 'batch_loss/train': 0.6587581299245358} +12/23/2021 17:07:30 - INFO - codeparrot_training - Step 5521: {'lr': 0.00048593120087481835, 'samples': 2827264, 'steps': 5521, 'batch_loss/train': 0.8435986842960119} +12/23/2021 17:07:41 - INFO - codeparrot_training - Step 5522: {'lr': 0.00048592598703674873, 'samples': 2827776, 'steps': 5522, 'batch_loss/train': 0.8170756348408759} +12/23/2021 17:07:54 - INFO - codeparrot_training - Step 5523: {'lr': 0.00048592077226072643, 'samples': 2828288, 'steps': 5523, 'batch_loss/train': 0.8291243680287153} +12/23/2021 17:08:04 - INFO - codeparrot_training - Step 5524: {'lr': 0.00048591555654677233, 'samples': 2828800, 'steps': 5524, 'batch_loss/train': 0.8320842143148184} +12/23/2021 17:08:15 - INFO - codeparrot_training - Step 5525: {'lr': 0.00048591033989490697, 'samples': 2829312, 'steps': 5525, 'batch_loss/train': 0.8551817871630192} +12/23/2021 17:08:27 - INFO - codeparrot_training - Step 5526: {'lr': 0.00048590512230515116, 'samples': 2829824, 'steps': 5526, 'batch_loss/train': 0.9437891258858144} +12/23/2021 17:08:38 - INFO - codeparrot_training - Step 5527: {'lr': 0.00048589990377752567, 'samples': 2830336, 'steps': 5527, 'batch_loss/train': 0.925607274286449} +12/23/2021 17:08:48 - INFO - codeparrot_training - Step 5528: {'lr': 0.0004858946843120513, 'samples': 2830848, 'steps': 5528, 'batch_loss/train': 0.9275646768510342} +12/23/2021 17:09:01 - INFO - codeparrot_training - Step 5529: {'lr': 0.00048588946390874864, 'samples': 2831360, 'steps': 5529, 'batch_loss/train': 0.8918994190171361} +12/23/2021 17:09:12 - INFO - codeparrot_training - Step 5530: {'lr': 0.0004858842425676386, 'samples': 2831872, 'steps': 5530, 'batch_loss/train': 0.8088418273255229} +12/23/2021 17:09:23 - INFO - codeparrot_training - Step 5531: {'lr': 0.00048587902028874187, 'samples': 2832384, 'steps': 5531, 'batch_loss/train': 0.8946389541961253} +12/23/2021 17:09:33 - INFO - codeparrot_training - Step 5532: {'lr': 0.00048587379707207924, 'samples': 2832896, 'steps': 5532, 'batch_loss/train': 0.8687834292650223} +12/23/2021 17:09:45 - INFO - codeparrot_training - Step 5533: {'lr': 0.0004858685729176714, 'samples': 2833408, 'steps': 5533, 'batch_loss/train': 1.0128828743472695} +12/23/2021 17:09:56 - INFO - codeparrot_training - Step 5534: {'lr': 0.0004858633478255392, 'samples': 2833920, 'steps': 5534, 'batch_loss/train': 0.892938076518476} +12/23/2021 17:10:06 - INFO - codeparrot_training - Step 5535: {'lr': 0.00048585812179570344, 'samples': 2834432, 'steps': 5535, 'batch_loss/train': 0.9575237291865051} +12/23/2021 17:10:18 - INFO - codeparrot_training - Step 5536: {'lr': 0.0004858528948281848, 'samples': 2834944, 'steps': 5536, 'batch_loss/train': 0.8917054496705532} +12/23/2021 17:10:29 - INFO - codeparrot_training - Step 5537: {'lr': 0.0004858476669230041, 'samples': 2835456, 'steps': 5537, 'batch_loss/train': 0.7965695176972076} +12/23/2021 17:10:39 - INFO - codeparrot_training - Step 5538: {'lr': 0.00048584243808018216, 'samples': 2835968, 'steps': 5538, 'batch_loss/train': 0.8610226305900142} +12/23/2021 17:10:52 - INFO - codeparrot_training - Step 5539: {'lr': 0.00048583720829973963, 'samples': 2836480, 'steps': 5539, 'batch_loss/train': 0.7794702122919261} +12/23/2021 17:11:03 - INFO - codeparrot_training - Step 5540: {'lr': 0.0004858319775816975, 'samples': 2836992, 'steps': 5540, 'batch_loss/train': 0.8852645913138986} +12/23/2021 17:11:14 - INFO - codeparrot_training - Step 5541: {'lr': 0.0004858267459260765, 'samples': 2837504, 'steps': 5541, 'batch_loss/train': 0.884495141915977} +12/23/2021 17:11:26 - INFO - codeparrot_training - Step 5542: {'lr': 0.00048582151333289734, 'samples': 2838016, 'steps': 5542, 'batch_loss/train': 0.9159451965242624} +12/23/2021 17:11:36 - INFO - codeparrot_training - Step 5543: {'lr': 0.0004858162798021809, 'samples': 2838528, 'steps': 5543, 'batch_loss/train': 0.8079344977159053} +12/23/2021 17:11:47 - INFO - codeparrot_training - Step 5544: {'lr': 0.000485811045333948, 'samples': 2839040, 'steps': 5544, 'batch_loss/train': 0.834439953789115} +12/23/2021 17:11:57 - INFO - codeparrot_training - Step 5545: {'lr': 0.00048580580992821945, 'samples': 2839552, 'steps': 5545, 'batch_loss/train': 0.8672593992669135} +12/23/2021 17:12:12 - INFO - codeparrot_training - Step 5546: {'lr': 0.00048580057358501595, 'samples': 2840064, 'steps': 5546, 'batch_loss/train': 0.8461138848215342} +12/23/2021 17:12:22 - INFO - codeparrot_training - Step 5547: {'lr': 0.00048579533630435846, 'samples': 2840576, 'steps': 5547, 'batch_loss/train': 0.7957617281936109} +12/23/2021 17:12:33 - INFO - codeparrot_training - Step 5548: {'lr': 0.00048579009808626773, 'samples': 2841088, 'steps': 5548, 'batch_loss/train': 1.1710157385095954} +12/23/2021 17:12:45 - INFO - codeparrot_training - Step 5549: {'lr': 0.00048578485893076463, 'samples': 2841600, 'steps': 5549, 'batch_loss/train': 0.8558984054252505} +12/23/2021 17:12:56 - INFO - codeparrot_training - Step 5550: {'lr': 0.00048577961883786993, 'samples': 2842112, 'steps': 5550, 'batch_loss/train': 0.973009247565642} +12/23/2021 17:13:06 - INFO - codeparrot_training - Step 5551: {'lr': 0.0004857743778076045, 'samples': 2842624, 'steps': 5551, 'batch_loss/train': 0.900227690115571} +12/23/2021 17:13:18 - INFO - codeparrot_training - Step 5552: {'lr': 0.00048576913583998915, 'samples': 2843136, 'steps': 5552, 'batch_loss/train': 0.8795542884618044} +12/23/2021 17:13:29 - INFO - codeparrot_training - Step 5553: {'lr': 0.00048576389293504476, 'samples': 2843648, 'steps': 5553, 'batch_loss/train': 0.9020623269025236} +12/23/2021 17:13:40 - INFO - codeparrot_training - Step 5554: {'lr': 0.0004857586490927921, 'samples': 2844160, 'steps': 5554, 'batch_loss/train': 0.8789241891354322} +12/23/2021 17:13:50 - INFO - codeparrot_training - Step 5555: {'lr': 0.0004857534043132521, 'samples': 2844672, 'steps': 5555, 'batch_loss/train': 0.8254860430024564} +12/23/2021 17:14:02 - INFO - codeparrot_training - Step 5556: {'lr': 0.0004857481585964456, 'samples': 2845184, 'steps': 5556, 'batch_loss/train': 0.7534028613008559} +12/23/2021 17:14:13 - INFO - codeparrot_training - Step 5557: {'lr': 0.0004857429119423935, 'samples': 2845696, 'steps': 5557, 'batch_loss/train': 0.8373604333028197} +12/23/2021 17:14:24 - INFO - codeparrot_training - Step 5558: {'lr': 0.00048573766435111647, 'samples': 2846208, 'steps': 5558, 'batch_loss/train': 0.866861255839467} +12/23/2021 17:14:37 - INFO - codeparrot_training - Step 5559: {'lr': 0.0004857324158226356, 'samples': 2846720, 'steps': 5559, 'batch_loss/train': 0.9144597514532506} +12/23/2021 17:14:47 - INFO - codeparrot_training - Step 5560: {'lr': 0.0004857271663569716, 'samples': 2847232, 'steps': 5560, 'batch_loss/train': 1.1073125498369336} +12/23/2021 17:14:58 - INFO - codeparrot_training - Step 5561: {'lr': 0.00048572191595414537, 'samples': 2847744, 'steps': 5561, 'batch_loss/train': 0.7938453843817115} +12/23/2021 17:15:10 - INFO - codeparrot_training - Step 5562: {'lr': 0.00048571666461417784, 'samples': 2848256, 'steps': 5562, 'batch_loss/train': 0.8941897638142109} +12/23/2021 17:15:21 - INFO - codeparrot_training - Step 5563: {'lr': 0.00048571141233708987, 'samples': 2848768, 'steps': 5563, 'batch_loss/train': 0.9739172831177711} +12/23/2021 17:15:31 - INFO - codeparrot_training - Step 5564: {'lr': 0.0004857061591229023, 'samples': 2849280, 'steps': 5564, 'batch_loss/train': 0.6936892472440377} +12/23/2021 17:15:42 - INFO - codeparrot_training - Step 5565: {'lr': 0.0004857009049716361, 'samples': 2849792, 'steps': 5565, 'batch_loss/train': 0.9611944258213043} +12/23/2021 17:15:54 - INFO - codeparrot_training - Step 5566: {'lr': 0.00048569564988331206, 'samples': 2850304, 'steps': 5566, 'batch_loss/train': 0.9998290464282036} +12/23/2021 17:16:04 - INFO - codeparrot_training - Step 5567: {'lr': 0.00048569039385795113, 'samples': 2850816, 'steps': 5567, 'batch_loss/train': 0.9147938899695873} +12/23/2021 17:16:15 - INFO - codeparrot_training - Step 5568: {'lr': 0.0004856851368955741, 'samples': 2851328, 'steps': 5568, 'batch_loss/train': 0.8900518424343318} +12/23/2021 17:16:29 - INFO - codeparrot_training - Step 5569: {'lr': 0.00048567987899620203, 'samples': 2851840, 'steps': 5569, 'batch_loss/train': 0.8980349758639932} +12/23/2021 17:16:40 - INFO - codeparrot_training - Step 5570: {'lr': 0.00048567462015985577, 'samples': 2852352, 'steps': 5570, 'batch_loss/train': 0.9185434896498919} +12/23/2021 17:16:50 - INFO - codeparrot_training - Step 5571: {'lr': 0.00048566936038655614, 'samples': 2852864, 'steps': 5571, 'batch_loss/train': 0.8423673161305487} +12/23/2021 17:17:04 - INFO - codeparrot_training - Step 5572: {'lr': 0.0004856640996763241, 'samples': 2853376, 'steps': 5572, 'batch_loss/train': 0.9444960216060281} +12/23/2021 17:17:14 - INFO - codeparrot_training - Step 5573: {'lr': 0.00048565883802918064, 'samples': 2853888, 'steps': 5573, 'batch_loss/train': 1.2567194746807218} +12/23/2021 17:17:25 - INFO - codeparrot_training - Step 5574: {'lr': 0.00048565357544514667, 'samples': 2854400, 'steps': 5574, 'batch_loss/train': 0.9332261020317674} +12/23/2021 17:17:35 - INFO - codeparrot_training - Step 5575: {'lr': 0.000485648311924243, 'samples': 2854912, 'steps': 5575, 'batch_loss/train': 0.859251391608268} +12/23/2021 17:17:49 - INFO - codeparrot_training - Step 5576: {'lr': 0.0004856430474664906, 'samples': 2855424, 'steps': 5576, 'batch_loss/train': 0.8427300970070064} +12/23/2021 17:18:00 - INFO - codeparrot_training - Step 5577: {'lr': 0.00048563778207191046, 'samples': 2855936, 'steps': 5577, 'batch_loss/train': 0.786781070753932} +12/23/2021 17:18:10 - INFO - codeparrot_training - Step 5578: {'lr': 0.0004856325157405235, 'samples': 2856448, 'steps': 5578, 'batch_loss/train': 0.7156032832572237} +12/23/2021 17:18:22 - INFO - codeparrot_training - Step 5579: {'lr': 0.0004856272484723506, 'samples': 2856960, 'steps': 5579, 'batch_loss/train': 0.7617356129921973} +12/23/2021 17:18:33 - INFO - codeparrot_training - Step 5580: {'lr': 0.00048562198026741264, 'samples': 2857472, 'steps': 5580, 'batch_loss/train': 0.9278225831221789} +12/23/2021 17:18:43 - INFO - codeparrot_training - Step 5581: {'lr': 0.0004856167111257308, 'samples': 2857984, 'steps': 5581, 'batch_loss/train': 0.8413434326648712} +12/23/2021 17:18:55 - INFO - codeparrot_training - Step 5582: {'lr': 0.0004856114410473258, 'samples': 2858496, 'steps': 5582, 'batch_loss/train': 0.852060136385262} +12/23/2021 17:19:06 - INFO - codeparrot_training - Step 5583: {'lr': 0.0004856061700322187, 'samples': 2859008, 'steps': 5583, 'batch_loss/train': 0.9400273836217821} +12/23/2021 17:19:17 - INFO - codeparrot_training - Step 5584: {'lr': 0.0004856008980804304, 'samples': 2859520, 'steps': 5584, 'batch_loss/train': 0.8032504841685295} +12/23/2021 17:19:27 - INFO - codeparrot_training - Step 5585: {'lr': 0.00048559562519198186, 'samples': 2860032, 'steps': 5585, 'batch_loss/train': 0.8981916531920433} +12/23/2021 17:19:40 - INFO - codeparrot_training - Step 5586: {'lr': 0.0004855903513668942, 'samples': 2860544, 'steps': 5586, 'batch_loss/train': 0.8717922540381551} +12/23/2021 17:19:51 - INFO - codeparrot_training - Step 5587: {'lr': 0.00048558507660518814, 'samples': 2861056, 'steps': 5587, 'batch_loss/train': 0.8536131763830781} +12/23/2021 17:20:02 - INFO - codeparrot_training - Step 5588: {'lr': 0.0004855798009068848, 'samples': 2861568, 'steps': 5588, 'batch_loss/train': 0.851682068197988} +12/23/2021 17:20:14 - INFO - codeparrot_training - Step 5589: {'lr': 0.0004855745242720051, 'samples': 2862080, 'steps': 5589, 'batch_loss/train': 0.9316445691511035} +12/23/2021 17:20:24 - INFO - codeparrot_training - Step 5590: {'lr': 0.0004855692467005701, 'samples': 2862592, 'steps': 5590, 'batch_loss/train': 0.8758534016087651} +12/23/2021 17:20:35 - INFO - codeparrot_training - Step 5591: {'lr': 0.0004855639681926007, 'samples': 2863104, 'steps': 5591, 'batch_loss/train': 0.8922669235616922} +12/23/2021 17:20:47 - INFO - codeparrot_training - Step 5592: {'lr': 0.0004855586887481179, 'samples': 2863616, 'steps': 5592, 'batch_loss/train': 0.7365989559330046} +12/23/2021 17:20:58 - INFO - codeparrot_training - Step 5593: {'lr': 0.0004855534083671427, 'samples': 2864128, 'steps': 5593, 'batch_loss/train': 0.9325183751061559} +12/23/2021 17:21:08 - INFO - codeparrot_training - Step 5594: {'lr': 0.0004855481270496961, 'samples': 2864640, 'steps': 5594, 'batch_loss/train': 0.8058261582627892} +12/23/2021 17:21:21 - INFO - codeparrot_training - Step 5595: {'lr': 0.000485542844795799, 'samples': 2865152, 'steps': 5595, 'batch_loss/train': 0.8849473455920815} +12/23/2021 17:21:32 - INFO - codeparrot_training - Step 5596: {'lr': 0.0004855375616054726, 'samples': 2865664, 'steps': 5596, 'batch_loss/train': 0.9457293432205915} +12/23/2021 17:21:43 - INFO - codeparrot_training - Step 5597: {'lr': 0.00048553227747873774, 'samples': 2866176, 'steps': 5597, 'batch_loss/train': 0.8494891335722059} +12/23/2021 17:21:53 - INFO - codeparrot_training - Step 5598: {'lr': 0.00048552699241561547, 'samples': 2866688, 'steps': 5598, 'batch_loss/train': 0.9992296798154712} +12/23/2021 17:22:05 - INFO - codeparrot_training - Step 5599: {'lr': 0.00048552170641612686, 'samples': 2867200, 'steps': 5599, 'batch_loss/train': 0.8889033505693078} +12/23/2021 17:22:16 - INFO - codeparrot_training - Step 5600: {'lr': 0.00048551641948029286, 'samples': 2867712, 'steps': 5600, 'batch_loss/train': 0.9205373004078865} +12/23/2021 17:22:26 - INFO - codeparrot_training - Step 5601: {'lr': 0.00048551113160813443, 'samples': 2868224, 'steps': 5601, 'batch_loss/train': 0.8006526557728648} +12/23/2021 17:22:39 - INFO - codeparrot_training - Step 5602: {'lr': 0.0004855058427996727, 'samples': 2868736, 'steps': 5602, 'batch_loss/train': 0.8697733879089355} +12/23/2021 17:22:49 - INFO - codeparrot_training - Step 5603: {'lr': 0.0004855005530549287, 'samples': 2869248, 'steps': 5603, 'batch_loss/train': 0.8155545084737241} +12/23/2021 17:23:00 - INFO - codeparrot_training - Step 5604: {'lr': 0.0004854952623739235, 'samples': 2869760, 'steps': 5604, 'batch_loss/train': 0.664637322188355} +12/23/2021 17:23:12 - INFO - codeparrot_training - Step 5605: {'lr': 0.00048548997075667787, 'samples': 2870272, 'steps': 5605, 'batch_loss/train': 0.8923755511641502} +12/23/2021 17:23:22 - INFO - codeparrot_training - Step 5606: {'lr': 0.00048548467820321316, 'samples': 2870784, 'steps': 5606, 'batch_loss/train': 0.8696077847853303} +12/23/2021 17:23:33 - INFO - codeparrot_training - Step 5607: {'lr': 0.0004854793847135502, 'samples': 2871296, 'steps': 5607, 'batch_loss/train': 0.8030636941548437} +12/23/2021 17:23:44 - INFO - codeparrot_training - Step 5608: {'lr': 0.00048547409028771026, 'samples': 2871808, 'steps': 5608, 'batch_loss/train': 0.7932616770267487} +12/23/2021 17:23:57 - INFO - codeparrot_training - Step 5609: {'lr': 0.00048546879492571414, 'samples': 2872320, 'steps': 5609, 'batch_loss/train': 0.9706066912040114} +12/23/2021 17:24:07 - INFO - codeparrot_training - Step 5610: {'lr': 0.000485463498627583, 'samples': 2872832, 'steps': 5610, 'batch_loss/train': 0.8683832297101617} +12/23/2021 17:24:18 - INFO - codeparrot_training - Step 5611: {'lr': 0.00048545820139333796, 'samples': 2873344, 'steps': 5611, 'batch_loss/train': 0.9176500802859664} +12/23/2021 17:24:30 - INFO - codeparrot_training - Step 5612: {'lr': 0.00048545290322299996, 'samples': 2873856, 'steps': 5612, 'batch_loss/train': 0.8297794726677239} +12/23/2021 17:24:41 - INFO - codeparrot_training - Step 5613: {'lr': 0.00048544760411659016, 'samples': 2874368, 'steps': 5613, 'batch_loss/train': 0.8236626544967294} +12/23/2021 17:24:51 - INFO - codeparrot_training - Step 5614: {'lr': 0.00048544230407412953, 'samples': 2874880, 'steps': 5614, 'batch_loss/train': 0.8302073460072279} +12/23/2021 17:25:04 - INFO - codeparrot_training - Step 5615: {'lr': 0.0004854370030956392, 'samples': 2875392, 'steps': 5615, 'batch_loss/train': 0.920360348187387} +12/23/2021 17:25:15 - INFO - codeparrot_training - Step 5616: {'lr': 0.0004854317011811403, 'samples': 2875904, 'steps': 5616, 'batch_loss/train': 0.8687647208571434} +12/23/2021 17:25:25 - INFO - codeparrot_training - Step 5617: {'lr': 0.00048542639833065385, 'samples': 2876416, 'steps': 5617, 'batch_loss/train': 0.7807512022554874} +12/23/2021 17:25:37 - INFO - codeparrot_training - Step 5618: {'lr': 0.0004854210945442009, 'samples': 2876928, 'steps': 5618, 'batch_loss/train': 0.8431373322382569} +12/23/2021 17:25:48 - INFO - codeparrot_training - Step 5619: {'lr': 0.0004854157898218026, 'samples': 2877440, 'steps': 5619, 'batch_loss/train': 0.8802944794297218} +12/23/2021 17:25:59 - INFO - codeparrot_training - Step 5620: {'lr': 0.00048541048416348, 'samples': 2877952, 'steps': 5620, 'batch_loss/train': 0.9604625422507524} +12/23/2021 17:26:09 - INFO - codeparrot_training - Step 5621: {'lr': 0.0004854051775692542, 'samples': 2878464, 'steps': 5621, 'batch_loss/train': 0.8999402206391096} +12/23/2021 17:26:21 - INFO - codeparrot_training - Step 5622: {'lr': 0.0004853998700391463, 'samples': 2878976, 'steps': 5622, 'batch_loss/train': 0.7686800982337445} +12/23/2021 17:26:32 - INFO - codeparrot_training - Step 5623: {'lr': 0.0004853945615731774, 'samples': 2879488, 'steps': 5623, 'batch_loss/train': 0.8545526005327702} +12/23/2021 17:26:42 - INFO - codeparrot_training - Step 5624: {'lr': 0.00048538925217136855, 'samples': 2880000, 'steps': 5624, 'batch_loss/train': 0.8539441218599677} +12/23/2021 17:26:56 - INFO - codeparrot_training - Step 5625: {'lr': 0.000485383941833741, 'samples': 2880512, 'steps': 5625, 'batch_loss/train': 0.8568756757304072} +12/23/2021 17:27:06 - INFO - codeparrot_training - Step 5626: {'lr': 0.00048537863056031573, 'samples': 2881024, 'steps': 5626, 'batch_loss/train': 0.792392217554152} +12/23/2021 17:27:17 - INFO - codeparrot_training - Step 5627: {'lr': 0.00048537331835111385, 'samples': 2881536, 'steps': 5627, 'batch_loss/train': 0.7625353699550033} +12/23/2021 17:27:29 - INFO - codeparrot_training - Step 5628: {'lr': 0.00048536800520615665, 'samples': 2882048, 'steps': 5628, 'batch_loss/train': 0.9085093894973397} +12/23/2021 17:27:39 - INFO - codeparrot_training - Step 5629: {'lr': 0.00048536269112546505, 'samples': 2882560, 'steps': 5629, 'batch_loss/train': 0.8547924719750881} +12/23/2021 17:27:50 - INFO - codeparrot_training - Step 5630: {'lr': 0.0004853573761090603, 'samples': 2883072, 'steps': 5630, 'batch_loss/train': 0.8494716705754399} +12/23/2021 17:28:01 - INFO - codeparrot_training - Step 5631: {'lr': 0.00048535206015696343, 'samples': 2883584, 'steps': 5631, 'batch_loss/train': 0.877021761611104} +12/23/2021 17:28:13 - INFO - codeparrot_training - Step 5632: {'lr': 0.0004853467432691957, 'samples': 2884096, 'steps': 5632, 'batch_loss/train': 0.947802297770977} +12/23/2021 17:28:23 - INFO - codeparrot_training - Step 5633: {'lr': 0.00048534142544577814, 'samples': 2884608, 'steps': 5633, 'batch_loss/train': 0.8717921879142523} +12/23/2021 17:28:34 - INFO - codeparrot_training - Step 5634: {'lr': 0.00048533610668673193, 'samples': 2885120, 'steps': 5634, 'batch_loss/train': 0.9233399308286607} +12/23/2021 17:28:47 - INFO - codeparrot_training - Step 5635: {'lr': 0.00048533078699207824, 'samples': 2885632, 'steps': 5635, 'batch_loss/train': 0.8021317915990949} +12/23/2021 17:28:58 - INFO - codeparrot_training - Step 5636: {'lr': 0.00048532546636183825, 'samples': 2886144, 'steps': 5636, 'batch_loss/train': 0.9622331839054823} +12/23/2021 17:29:08 - INFO - codeparrot_training - Step 5637: {'lr': 0.000485320144796033, 'samples': 2886656, 'steps': 5637, 'batch_loss/train': 0.8594139166816603} +12/23/2021 17:29:21 - INFO - codeparrot_training - Step 5638: {'lr': 0.0004853148222946837, 'samples': 2887168, 'steps': 5638, 'batch_loss/train': 0.8732613250613213} +12/23/2021 17:29:31 - INFO - codeparrot_training - Step 5639: {'lr': 0.0004853094988578115, 'samples': 2887680, 'steps': 5639, 'batch_loss/train': 0.8512482673395425} +12/23/2021 17:29:42 - INFO - codeparrot_training - Step 5640: {'lr': 0.00048530417448543764, 'samples': 2888192, 'steps': 5640, 'batch_loss/train': 0.773575053550303} +12/23/2021 17:29:54 - INFO - codeparrot_training - Step 5641: {'lr': 0.0004852988491775832, 'samples': 2888704, 'steps': 5641, 'batch_loss/train': 0.7572517129592597} +12/23/2021 17:30:04 - INFO - codeparrot_training - Step 5642: {'lr': 0.0004852935229342694, 'samples': 2889216, 'steps': 5642, 'batch_loss/train': 0.8855923749506474} +12/23/2021 17:30:15 - INFO - codeparrot_training - Step 5643: {'lr': 0.0004852881957555174, 'samples': 2889728, 'steps': 5643, 'batch_loss/train': 0.8325124559924006} +12/23/2021 17:30:26 - INFO - codeparrot_training - Step 5644: {'lr': 0.00048528286764134833, 'samples': 2890240, 'steps': 5644, 'batch_loss/train': 0.6692821539472789} +12/23/2021 17:30:39 - INFO - codeparrot_training - Step 5645: {'lr': 0.00048527753859178347, 'samples': 2890752, 'steps': 5645, 'batch_loss/train': 0.8705287212505937} +12/23/2021 17:30:49 - INFO - codeparrot_training - Step 5646: {'lr': 0.0004852722086068439, 'samples': 2891264, 'steps': 5646, 'batch_loss/train': 0.9769562250003219} +12/23/2021 17:31:00 - INFO - codeparrot_training - Step 5647: {'lr': 0.00048526687768655086, 'samples': 2891776, 'steps': 5647, 'batch_loss/train': 0.7590757622383535} +12/23/2021 17:31:12 - INFO - codeparrot_training - Step 5648: {'lr': 0.0004852615458309256, 'samples': 2892288, 'steps': 5648, 'batch_loss/train': 0.947014482691884} +12/23/2021 17:31:23 - INFO - codeparrot_training - Step 5649: {'lr': 0.00048525621303998923, 'samples': 2892800, 'steps': 5649, 'batch_loss/train': 0.9017032599076629} +12/23/2021 17:31:33 - INFO - codeparrot_training - Step 5650: {'lr': 0.000485250879313763, 'samples': 2893312, 'steps': 5650, 'batch_loss/train': 0.706723022274673} +12/23/2021 17:31:44 - INFO - codeparrot_training - Step 5651: {'lr': 0.0004852455446522681, 'samples': 2893824, 'steps': 5651, 'batch_loss/train': 0.7976221679709852} +12/23/2021 17:31:56 - INFO - codeparrot_training - Step 5652: {'lr': 0.0004852402090555258, 'samples': 2894336, 'steps': 5652, 'batch_loss/train': 0.8768794368952513} +12/23/2021 17:32:07 - INFO - codeparrot_training - Step 5653: {'lr': 0.0004852348725235571, 'samples': 2894848, 'steps': 5653, 'batch_loss/train': 0.8027380462735891} +12/23/2021 17:32:17 - INFO - codeparrot_training - Step 5654: {'lr': 0.0004852295350563836, 'samples': 2895360, 'steps': 5654, 'batch_loss/train': 0.8682599854655564} +12/23/2021 17:32:30 - INFO - codeparrot_training - Step 5655: {'lr': 0.00048522419665402605, 'samples': 2895872, 'steps': 5655, 'batch_loss/train': 0.8331684852018952} +12/23/2021 17:32:41 - INFO - codeparrot_training - Step 5656: {'lr': 0.0004852188573165061, 'samples': 2896384, 'steps': 5656, 'batch_loss/train': 0.7908188076689839} +12/23/2021 17:32:52 - INFO - codeparrot_training - Step 5657: {'lr': 0.00048521351704384466, 'samples': 2896896, 'steps': 5657, 'batch_loss/train': 0.8206956987269223} +12/23/2021 17:33:04 - INFO - codeparrot_training - Step 5658: {'lr': 0.00048520817583606315, 'samples': 2897408, 'steps': 5658, 'batch_loss/train': 0.8490797565318644} +12/23/2021 17:33:15 - INFO - codeparrot_training - Step 5659: {'lr': 0.0004852028336931827, 'samples': 2897920, 'steps': 5659, 'batch_loss/train': 1.2442417982965708} +12/23/2021 17:33:25 - INFO - codeparrot_training - Step 5660: {'lr': 0.00048519749061522454, 'samples': 2898432, 'steps': 5660, 'batch_loss/train': 0.8903687437996268} +12/23/2021 17:33:36 - INFO - codeparrot_training - Step 5661: {'lr': 0.0004851921466022101, 'samples': 2898944, 'steps': 5661, 'batch_loss/train': 0.9303020723164082} +12/23/2021 17:33:49 - INFO - codeparrot_training - Step 5662: {'lr': 0.0004851868016541604, 'samples': 2899456, 'steps': 5662, 'batch_loss/train': 0.6762329021003097} +12/23/2021 17:34:00 - INFO - codeparrot_training - Step 5663: {'lr': 0.00048518145577109686, 'samples': 2899968, 'steps': 5663, 'batch_loss/train': 0.8446913557127118} +12/23/2021 17:34:10 - INFO - codeparrot_training - Step 5664: {'lr': 0.0004851761089530406, 'samples': 2900480, 'steps': 5664, 'batch_loss/train': 0.8422333234921098} +12/23/2021 17:34:22 - INFO - codeparrot_training - Step 5665: {'lr': 0.0004851707612000129, 'samples': 2900992, 'steps': 5665, 'batch_loss/train': 0.8478340557776392} +12/23/2021 17:34:33 - INFO - codeparrot_training - Step 5666: {'lr': 0.00048516541251203513, 'samples': 2901504, 'steps': 5666, 'batch_loss/train': 0.7539586089551449} +12/23/2021 17:34:44 - INFO - codeparrot_training - Step 5667: {'lr': 0.0004851600628891284, 'samples': 2902016, 'steps': 5667, 'batch_loss/train': 0.9275765102356672} +12/23/2021 17:34:56 - INFO - codeparrot_training - Step 5668: {'lr': 0.00048515471233131406, 'samples': 2902528, 'steps': 5668, 'batch_loss/train': 0.8563982918858528} +12/23/2021 17:35:06 - INFO - codeparrot_training - Step 5669: {'lr': 0.00048514936083861337, 'samples': 2903040, 'steps': 5669, 'batch_loss/train': 0.8024991089478135} +12/23/2021 17:35:17 - INFO - codeparrot_training - Step 5670: {'lr': 0.00048514400841104764, 'samples': 2903552, 'steps': 5670, 'batch_loss/train': 0.8589194258674979} +12/23/2021 17:35:28 - INFO - codeparrot_training - Step 5671: {'lr': 0.0004851386550486382, 'samples': 2904064, 'steps': 5671, 'batch_loss/train': 1.0097386883571744} +12/23/2021 17:35:40 - INFO - codeparrot_training - Step 5672: {'lr': 0.00048513330075140615, 'samples': 2904576, 'steps': 5672, 'batch_loss/train': 0.7166034178808331} +12/23/2021 17:35:50 - INFO - codeparrot_training - Step 5673: {'lr': 0.0004851279455193729, 'samples': 2905088, 'steps': 5673, 'batch_loss/train': 0.8665933618322015} +12/23/2021 17:36:01 - INFO - codeparrot_training - Step 5674: {'lr': 0.0004851225893525597, 'samples': 2905600, 'steps': 5674, 'batch_loss/train': 0.9810452135279775} +12/23/2021 17:36:14 - INFO - codeparrot_training - Step 5675: {'lr': 0.00048511723225098783, 'samples': 2906112, 'steps': 5675, 'batch_loss/train': 0.8630236145108938} +12/23/2021 17:36:25 - INFO - codeparrot_training - Step 5676: {'lr': 0.0004851118742146787, 'samples': 2906624, 'steps': 5676, 'batch_loss/train': 0.8741940199397504} +12/23/2021 17:36:35 - INFO - codeparrot_training - Step 5677: {'lr': 0.00048510651524365353, 'samples': 2907136, 'steps': 5677, 'batch_loss/train': 0.9369575306773186} +12/23/2021 17:36:47 - INFO - codeparrot_training - Step 5678: {'lr': 0.0004851011553379336, 'samples': 2907648, 'steps': 5678, 'batch_loss/train': 0.9262470761314034} +12/23/2021 17:36:58 - INFO - codeparrot_training - Step 5679: {'lr': 0.00048509579449754024, 'samples': 2908160, 'steps': 5679, 'batch_loss/train': 0.8292013071477413} +12/23/2021 17:37:08 - INFO - codeparrot_training - Step 5680: {'lr': 0.00048509043272249475, 'samples': 2908672, 'steps': 5680, 'batch_loss/train': 0.9375350645277649} +12/23/2021 17:37:19 - INFO - codeparrot_training - Step 5681: {'lr': 0.0004850850700128185, 'samples': 2909184, 'steps': 5681, 'batch_loss/train': 0.8607015833258629} +12/23/2021 17:37:31 - INFO - codeparrot_training - Step 5682: {'lr': 0.0004850797063685328, 'samples': 2909696, 'steps': 5682, 'batch_loss/train': 0.7911530383862555} +12/23/2021 17:37:42 - INFO - codeparrot_training - Step 5683: {'lr': 0.0004850743417896589, 'samples': 2910208, 'steps': 5683, 'batch_loss/train': 0.9002758674323559} +12/23/2021 17:37:52 - INFO - codeparrot_training - Step 5684: {'lr': 0.00048506897627621825, 'samples': 2910720, 'steps': 5684, 'batch_loss/train': 0.8946467079222202} +12/23/2021 17:38:05 - INFO - codeparrot_training - Step 5685: {'lr': 0.00048506360982823206, 'samples': 2911232, 'steps': 5685, 'batch_loss/train': 0.9298830665647984} +12/23/2021 17:38:16 - INFO - codeparrot_training - Step 5686: {'lr': 0.0004850582424457217, 'samples': 2911744, 'steps': 5686, 'batch_loss/train': 0.8906488846987486} +12/23/2021 17:38:27 - INFO - codeparrot_training - Step 5687: {'lr': 0.0004850528741287085, 'samples': 2912256, 'steps': 5687, 'batch_loss/train': 0.728834193199873} +12/23/2021 17:38:40 - INFO - codeparrot_training - Step 5688: {'lr': 0.0004850475048772139, 'samples': 2912768, 'steps': 5688, 'batch_loss/train': 0.8610173701308668} +12/23/2021 17:38:50 - INFO - codeparrot_training - Step 5689: {'lr': 0.0004850421346912591, 'samples': 2913280, 'steps': 5689, 'batch_loss/train': 0.490997152402997} +12/23/2021 17:39:01 - INFO - codeparrot_training - Step 5690: {'lr': 0.00048503676357086555, 'samples': 2913792, 'steps': 5690, 'batch_loss/train': 0.9332099710591137} +12/23/2021 17:39:11 - INFO - codeparrot_training - Step 5691: {'lr': 0.0004850313915160546, 'samples': 2914304, 'steps': 5691, 'batch_loss/train': 1.530093121342361} +12/23/2021 17:39:25 - INFO - codeparrot_training - Step 5692: {'lr': 0.0004850260185268476, 'samples': 2914816, 'steps': 5692, 'batch_loss/train': 0.7996871937066317} +12/23/2021 17:39:35 - INFO - codeparrot_training - Step 5693: {'lr': 0.0004850206446032658, 'samples': 2915328, 'steps': 5693, 'batch_loss/train': 0.80346107063815} +12/23/2021 17:39:46 - INFO - codeparrot_training - Step 5694: {'lr': 0.0004850152697453307, 'samples': 2915840, 'steps': 5694, 'batch_loss/train': 0.8043468984542415} +12/23/2021 17:39:58 - INFO - codeparrot_training - Step 5695: {'lr': 0.0004850098939530636, 'samples': 2916352, 'steps': 5695, 'batch_loss/train': 0.977509181946516} +12/23/2021 17:40:09 - INFO - codeparrot_training - Step 5696: {'lr': 0.00048500451722648596, 'samples': 2916864, 'steps': 5696, 'batch_loss/train': 0.9594697924330831} +12/23/2021 17:40:19 - INFO - codeparrot_training - Step 5697: {'lr': 0.0004849991395656191, 'samples': 2917376, 'steps': 5697, 'batch_loss/train': 0.9153572823852301} +12/23/2021 17:40:34 - INFO - codeparrot_training - Step 5698: {'lr': 0.0004849937609704843, 'samples': 2917888, 'steps': 5698, 'batch_loss/train': 0.921175523661077} +12/23/2021 17:40:44 - INFO - codeparrot_training - Step 5699: {'lr': 0.00048498838144110315, 'samples': 2918400, 'steps': 5699, 'batch_loss/train': 0.8596014480572194} +12/23/2021 17:40:55 - INFO - codeparrot_training - Step 5700: {'lr': 0.00048498300097749687, 'samples': 2918912, 'steps': 5700, 'batch_loss/train': 0.8912921277806163} +12/23/2021 17:41:05 - INFO - codeparrot_training - Step 5701: {'lr': 0.0004849776195796869, 'samples': 2919424, 'steps': 5701, 'batch_loss/train': 0.7450002958066761} +12/23/2021 17:41:17 - INFO - codeparrot_training - Step 5702: {'lr': 0.0004849722372476947, 'samples': 2919936, 'steps': 5702, 'batch_loss/train': 0.8302383599802852} +12/23/2021 17:41:28 - INFO - codeparrot_training - Step 5703: {'lr': 0.0004849668539815415, 'samples': 2920448, 'steps': 5703, 'batch_loss/train': 0.8983387500047684} +12/23/2021 17:41:39 - INFO - codeparrot_training - Step 5704: {'lr': 0.0004849614697812489, 'samples': 2920960, 'steps': 5704, 'batch_loss/train': 0.8286627158522606} +12/23/2021 17:41:51 - INFO - codeparrot_training - Step 5705: {'lr': 0.00048495608464683813, 'samples': 2921472, 'steps': 5705, 'batch_loss/train': 0.8180871140211821} +12/23/2021 17:42:01 - INFO - codeparrot_training - Step 5706: {'lr': 0.00048495069857833076, 'samples': 2921984, 'steps': 5706, 'batch_loss/train': 0.9020958906039596} +12/23/2021 17:42:12 - INFO - codeparrot_training - Step 5707: {'lr': 0.0004849453115757481, 'samples': 2922496, 'steps': 5707, 'batch_loss/train': 0.8186493096873164} +12/23/2021 17:42:26 - INFO - codeparrot_training - Step 5708: {'lr': 0.0004849399236391115, 'samples': 2923008, 'steps': 5708, 'batch_loss/train': 0.8741459120064974} +12/23/2021 17:42:36 - INFO - codeparrot_training - Step 5709: {'lr': 0.00048493453476844265, 'samples': 2923520, 'steps': 5709, 'batch_loss/train': 0.9111541910097003} +12/23/2021 17:42:47 - INFO - codeparrot_training - Step 5710: {'lr': 0.0004849291449637626, 'samples': 2924032, 'steps': 5710, 'batch_loss/train': 0.8488597963005304} +12/23/2021 17:42:58 - INFO - codeparrot_training - Step 5711: {'lr': 0.00048492375422509314, 'samples': 2924544, 'steps': 5711, 'batch_loss/train': 0.8909217910841107} +12/23/2021 17:43:10 - INFO - codeparrot_training - Step 5712: {'lr': 0.0004849183625524555, 'samples': 2925056, 'steps': 5712, 'batch_loss/train': 0.8746017012745142} +12/23/2021 17:43:21 - INFO - codeparrot_training - Step 5713: {'lr': 0.00048491296994587107, 'samples': 2925568, 'steps': 5713, 'batch_loss/train': 0.8949397504329681} +12/23/2021 17:43:31 - INFO - codeparrot_training - Step 5714: {'lr': 0.00048490757640536145, 'samples': 2926080, 'steps': 5714, 'batch_loss/train': 0.5550980230327696} +12/23/2021 17:43:43 - INFO - codeparrot_training - Step 5715: {'lr': 0.00048490218193094806, 'samples': 2926592, 'steps': 5715, 'batch_loss/train': 0.8172649592161179} +12/23/2021 17:43:54 - INFO - codeparrot_training - Step 5716: {'lr': 0.00048489678652265214, 'samples': 2927104, 'steps': 5716, 'batch_loss/train': 0.8432203438133001} +12/23/2021 17:44:04 - INFO - codeparrot_training - Step 5717: {'lr': 0.0004848913901804954, 'samples': 2927616, 'steps': 5717, 'batch_loss/train': 0.878210287541151} +12/23/2021 17:44:18 - INFO - codeparrot_training - Step 5718: {'lr': 0.00048488599290449913, 'samples': 2928128, 'steps': 5718, 'batch_loss/train': 0.8559604017063975} +12/23/2021 17:44:28 - INFO - codeparrot_training - Step 5719: {'lr': 0.00048488059469468486, 'samples': 2928640, 'steps': 5719, 'batch_loss/train': 0.8142470642924309} +12/23/2021 17:44:39 - INFO - codeparrot_training - Step 5720: {'lr': 0.00048487519555107404, 'samples': 2929152, 'steps': 5720, 'batch_loss/train': 0.8788981977850199} +12/23/2021 17:44:49 - INFO - codeparrot_training - Step 5721: {'lr': 0.00048486979547368814, 'samples': 2929664, 'steps': 5721, 'batch_loss/train': 0.853825356811285} +12/23/2021 17:45:01 - INFO - codeparrot_training - Step 5722: {'lr': 0.0004848643944625486, 'samples': 2930176, 'steps': 5722, 'batch_loss/train': 0.9267551824450493} +12/23/2021 17:45:12 - INFO - codeparrot_training - Step 5723: {'lr': 0.0004848589925176769, 'samples': 2930688, 'steps': 5723, 'batch_loss/train': 0.8458045292645693} +12/23/2021 17:45:23 - INFO - codeparrot_training - Step 5724: {'lr': 0.00048485358963909454, 'samples': 2931200, 'steps': 5724, 'batch_loss/train': 0.8388996729627252} +12/23/2021 17:45:36 - INFO - codeparrot_training - Step 5725: {'lr': 0.000484848185826823, 'samples': 2931712, 'steps': 5725, 'batch_loss/train': 0.887157610617578} +12/23/2021 17:45:47 - INFO - codeparrot_training - Step 5726: {'lr': 0.0004848427810808837, 'samples': 2932224, 'steps': 5726, 'batch_loss/train': 0.9172366484999657} +12/23/2021 17:45:58 - INFO - codeparrot_training - Step 5727: {'lr': 0.0004848373754012982, 'samples': 2932736, 'steps': 5727, 'batch_loss/train': 0.8495680810883641} +12/23/2021 17:46:10 - INFO - codeparrot_training - Step 5728: {'lr': 0.000484831968788088, 'samples': 2933248, 'steps': 5728, 'batch_loss/train': 0.8288836134597659} +12/23/2021 17:46:20 - INFO - codeparrot_training - Step 5729: {'lr': 0.0004848265612412745, 'samples': 2933760, 'steps': 5729, 'batch_loss/train': 0.6694669981952757} +12/23/2021 17:46:31 - INFO - codeparrot_training - Step 5730: {'lr': 0.0004848211527608793, 'samples': 2934272, 'steps': 5730, 'batch_loss/train': 0.89100047852844} +12/23/2021 17:46:43 - INFO - codeparrot_training - Step 5731: {'lr': 0.0004848157433469239, 'samples': 2934784, 'steps': 5731, 'batch_loss/train': 0.8551372173242271} +12/23/2021 17:46:53 - INFO - codeparrot_training - Step 5732: {'lr': 0.0004848103329994297, 'samples': 2935296, 'steps': 5732, 'batch_loss/train': 0.8487052330747247} +12/23/2021 17:47:04 - INFO - codeparrot_training - Step 5733: {'lr': 0.00048480492171841825, 'samples': 2935808, 'steps': 5733, 'batch_loss/train': 0.8736532917246222} +12/23/2021 17:47:15 - INFO - codeparrot_training - Step 5734: {'lr': 0.0004847995095039112, 'samples': 2936320, 'steps': 5734, 'batch_loss/train': 0.7677104147151113} +12/23/2021 17:47:27 - INFO - codeparrot_training - Step 5735: {'lr': 0.00048479409635592983, 'samples': 2936832, 'steps': 5735, 'batch_loss/train': 0.833510048687458} +12/23/2021 17:47:37 - INFO - codeparrot_training - Step 5736: {'lr': 0.00048478868227449585, 'samples': 2937344, 'steps': 5736, 'batch_loss/train': 0.8808179637417197} +12/23/2021 17:47:48 - INFO - codeparrot_training - Step 5737: {'lr': 0.00048478326725963074, 'samples': 2937856, 'steps': 5737, 'batch_loss/train': 0.9663164839148521} +12/23/2021 17:48:02 - INFO - codeparrot_training - Step 5738: {'lr': 0.000484777851311356, 'samples': 2938368, 'steps': 5738, 'batch_loss/train': 0.8004668047651649} +12/23/2021 17:48:12 - INFO - codeparrot_training - Step 5739: {'lr': 0.00048477243442969313, 'samples': 2938880, 'steps': 5739, 'batch_loss/train': 0.8829970816150308} +12/23/2021 17:48:23 - INFO - codeparrot_training - Step 5740: {'lr': 0.0004847670166146638, 'samples': 2939392, 'steps': 5740, 'batch_loss/train': 0.7511804001405835} +12/23/2021 17:48:35 - INFO - codeparrot_training - Step 5741: {'lr': 0.00048476159786628936, 'samples': 2939904, 'steps': 5741, 'batch_loss/train': 0.6992355352267623} +12/23/2021 17:48:46 - INFO - codeparrot_training - Step 5742: {'lr': 0.0004847561781845915, 'samples': 2940416, 'steps': 5742, 'batch_loss/train': 0.9154954534024} +12/23/2021 17:48:56 - INFO - codeparrot_training - Step 5743: {'lr': 0.0004847507575695917, 'samples': 2940928, 'steps': 5743, 'batch_loss/train': 0.8479732144623995} +12/23/2021 17:49:07 - INFO - codeparrot_training - Step 5744: {'lr': 0.00048474533602131155, 'samples': 2941440, 'steps': 5744, 'batch_loss/train': 0.8439125348813832} +12/23/2021 17:49:19 - INFO - codeparrot_training - Step 5745: {'lr': 0.0004847399135397725, 'samples': 2941952, 'steps': 5745, 'batch_loss/train': 0.8634787406772375} +12/23/2021 17:49:30 - INFO - codeparrot_training - Step 5746: {'lr': 0.00048473449012499626, 'samples': 2942464, 'steps': 5746, 'batch_loss/train': 0.8475349908694625} +12/23/2021 17:49:40 - INFO - codeparrot_training - Step 5747: {'lr': 0.0004847290657770044, 'samples': 2942976, 'steps': 5747, 'batch_loss/train': 0.8270867080427706} +12/23/2021 17:49:54 - INFO - codeparrot_training - Step 5748: {'lr': 0.0004847236404958182, 'samples': 2943488, 'steps': 5748, 'batch_loss/train': 0.7840717742219567} +12/23/2021 17:50:05 - INFO - codeparrot_training - Step 5749: {'lr': 0.0004847182142814596, 'samples': 2944000, 'steps': 5749, 'batch_loss/train': 0.7996553983539343} +12/23/2021 17:50:15 - INFO - codeparrot_training - Step 5750: {'lr': 0.00048471278713394997, 'samples': 2944512, 'steps': 5750, 'batch_loss/train': 2.1070206351578236} +12/23/2021 17:50:27 - INFO - codeparrot_training - Step 5751: {'lr': 0.00048470735905331087, 'samples': 2945024, 'steps': 5751, 'batch_loss/train': 0.9943737210705876} +12/23/2021 17:50:38 - INFO - codeparrot_training - Step 5752: {'lr': 0.00048470193003956395, 'samples': 2945536, 'steps': 5752, 'batch_loss/train': 0.8111037453636527} +12/23/2021 17:50:48 - INFO - codeparrot_training - Step 5753: {'lr': 0.0004846965000927308, 'samples': 2946048, 'steps': 5753, 'batch_loss/train': 0.7603561477735639} +12/23/2021 17:50:59 - INFO - codeparrot_training - Step 5754: {'lr': 0.000484691069212833, 'samples': 2946560, 'steps': 5754, 'batch_loss/train': 0.8025361662730575} +12/23/2021 17:51:12 - INFO - codeparrot_training - Step 5755: {'lr': 0.00048468563739989214, 'samples': 2947072, 'steps': 5755, 'batch_loss/train': 0.8946326747536659} +12/23/2021 17:51:23 - INFO - codeparrot_training - Step 5756: {'lr': 0.0004846802046539297, 'samples': 2947584, 'steps': 5756, 'batch_loss/train': 0.8663171147927642} +12/23/2021 17:51:33 - INFO - codeparrot_training - Step 5757: {'lr': 0.0004846747709749675, 'samples': 2948096, 'steps': 5757, 'batch_loss/train': 0.8758942447602749} +12/23/2021 17:51:45 - INFO - codeparrot_training - Step 5758: {'lr': 0.0004846693363630269, 'samples': 2948608, 'steps': 5758, 'batch_loss/train': 0.8662596447393298} +12/23/2021 17:51:56 - INFO - codeparrot_training - Step 5759: {'lr': 0.0004846639008181297, 'samples': 2949120, 'steps': 5759, 'batch_loss/train': 0.6931184604763985} +12/23/2021 17:52:06 - INFO - codeparrot_training - Step 5760: {'lr': 0.0004846584643402975, 'samples': 2949632, 'steps': 5760, 'batch_loss/train': 0.8460764815099537} +12/23/2021 17:52:19 - INFO - codeparrot_training - Step 5761: {'lr': 0.0004846530269295517, 'samples': 2950144, 'steps': 5761, 'batch_loss/train': 0.9372547529637814} +12/23/2021 17:52:29 - INFO - codeparrot_training - Step 5762: {'lr': 0.00048464758858591416, 'samples': 2950656, 'steps': 5762, 'batch_loss/train': 0.8251945469528437} +12/23/2021 17:52:40 - INFO - codeparrot_training - Step 5763: {'lr': 0.00048464214930940644, 'samples': 2951168, 'steps': 5763, 'batch_loss/train': 0.8841389818117023} +12/23/2021 17:52:50 - INFO - codeparrot_training - Step 5764: {'lr': 0.00048463670910005004, 'samples': 2951680, 'steps': 5764, 'batch_loss/train': 0.8269897382706404} +12/23/2021 17:53:04 - INFO - codeparrot_training - Step 5765: {'lr': 0.00048463126795786674, 'samples': 2952192, 'steps': 5765, 'batch_loss/train': 0.8345305370166898} +12/23/2021 17:53:15 - INFO - codeparrot_training - Step 5766: {'lr': 0.00048462582588287807, 'samples': 2952704, 'steps': 5766, 'batch_loss/train': 0.8199377274140716} +12/23/2021 17:53:25 - INFO - codeparrot_training - Step 5767: {'lr': 0.0004846203828751057, 'samples': 2953216, 'steps': 5767, 'batch_loss/train': 0.8685622564516962} +12/23/2021 17:53:38 - INFO - codeparrot_training - Step 5768: {'lr': 0.00048461493893457133, 'samples': 2953728, 'steps': 5768, 'batch_loss/train': 0.8400259651243687} +12/23/2021 17:53:48 - INFO - codeparrot_training - Step 5769: {'lr': 0.0004846094940612965, 'samples': 2954240, 'steps': 5769, 'batch_loss/train': 0.8218321073800325} +12/23/2021 17:53:59 - INFO - codeparrot_training - Step 5770: {'lr': 0.00048460404825530294, 'samples': 2954752, 'steps': 5770, 'batch_loss/train': 0.9200469302013516} +12/23/2021 17:54:11 - INFO - codeparrot_training - Step 5771: {'lr': 0.0004845986015166123, 'samples': 2955264, 'steps': 5771, 'batch_loss/train': 0.8687302637845278} +12/23/2021 17:54:22 - INFO - codeparrot_training - Step 5772: {'lr': 0.0004845931538452462, 'samples': 2955776, 'steps': 5772, 'batch_loss/train': 0.7901577027514577} +12/23/2021 17:54:32 - INFO - codeparrot_training - Step 5773: {'lr': 0.0004845877052412262, 'samples': 2956288, 'steps': 5773, 'batch_loss/train': 0.751711543649435} +12/23/2021 17:54:44 - INFO - codeparrot_training - Step 5774: {'lr': 0.0004845822557045741, 'samples': 2956800, 'steps': 5774, 'batch_loss/train': 0.9358236128464341} +12/23/2021 17:54:55 - INFO - codeparrot_training - Step 5775: {'lr': 0.00048457680523531155, 'samples': 2957312, 'steps': 5775, 'batch_loss/train': 0.8457398852333426} +12/23/2021 17:55:05 - INFO - codeparrot_training - Step 5776: {'lr': 0.0004845713538334602, 'samples': 2957824, 'steps': 5776, 'batch_loss/train': 0.8078504307195544} +12/23/2021 17:55:16 - INFO - codeparrot_training - Step 5777: {'lr': 0.00048456590149904166, 'samples': 2958336, 'steps': 5777, 'batch_loss/train': 0.8546206876635551} +12/23/2021 17:55:30 - INFO - codeparrot_training - Step 5778: {'lr': 0.0004845604482320778, 'samples': 2958848, 'steps': 5778, 'batch_loss/train': 0.865051562897861} +12/23/2021 17:55:41 - INFO - codeparrot_training - Step 5779: {'lr': 0.00048455499403259007, 'samples': 2959360, 'steps': 5779, 'batch_loss/train': 0.8522650999948382} +12/23/2021 17:55:51 - INFO - codeparrot_training - Step 5780: {'lr': 0.0004845495389006002, 'samples': 2959872, 'steps': 5780, 'batch_loss/train': 0.8406418971717358} +12/23/2021 17:56:03 - INFO - codeparrot_training - Step 5781: {'lr': 0.00048454408283613, 'samples': 2960384, 'steps': 5781, 'batch_loss/train': 0.8683620989322662} +12/23/2021 17:56:14 - INFO - codeparrot_training - Step 5782: {'lr': 0.00048453862583920106, 'samples': 2960896, 'steps': 5782, 'batch_loss/train': 0.8906156211160123} +12/23/2021 17:56:24 - INFO - codeparrot_training - Step 5783: {'lr': 0.00048453316790983517, 'samples': 2961408, 'steps': 5783, 'batch_loss/train': 0.9435799550265074} +12/23/2021 17:56:38 - INFO - codeparrot_training - Step 5784: {'lr': 0.0004845277090480539, 'samples': 2961920, 'steps': 5784, 'batch_loss/train': 0.7738277893513441} +12/23/2021 17:56:49 - INFO - codeparrot_training - Step 5785: {'lr': 0.000484522249253879, 'samples': 2962432, 'steps': 5785, 'batch_loss/train': 0.8862414290197194} +12/23/2021 17:56:59 - INFO - codeparrot_training - Step 5786: {'lr': 0.0004845167885273322, 'samples': 2962944, 'steps': 5786, 'batch_loss/train': 0.9296642746776342} +12/23/2021 17:57:10 - INFO - codeparrot_training - Step 5787: {'lr': 0.0004845113268684351, 'samples': 2963456, 'steps': 5787, 'batch_loss/train': 0.7956486102193594} +12/23/2021 17:57:22 - INFO - codeparrot_training - Step 5788: {'lr': 0.0004845058642772097, 'samples': 2963968, 'steps': 5788, 'batch_loss/train': 0.8849366893991828} +12/23/2021 17:57:33 - INFO - codeparrot_training - Step 5789: {'lr': 0.0004845004007536774, 'samples': 2964480, 'steps': 5789, 'batch_loss/train': 0.8426930643618107} +12/23/2021 17:57:43 - INFO - codeparrot_training - Step 5790: {'lr': 0.0004844949362978601, 'samples': 2964992, 'steps': 5790, 'batch_loss/train': 0.6134842461906374} +12/23/2021 17:57:55 - INFO - codeparrot_training - Step 5791: {'lr': 0.00048448947090977944, 'samples': 2965504, 'steps': 5791, 'batch_loss/train': 0.844102589180693} +12/23/2021 17:58:06 - INFO - codeparrot_training - Step 5792: {'lr': 0.0004844840045894573, 'samples': 2966016, 'steps': 5792, 'batch_loss/train': 0.962216068059206} +12/23/2021 17:58:17 - INFO - codeparrot_training - Step 5793: {'lr': 0.00048447853733691514, 'samples': 2966528, 'steps': 5793, 'batch_loss/train': 0.7510251980274916} +12/23/2021 17:58:31 - INFO - codeparrot_training - Step 5794: {'lr': 0.00048447306915217495, 'samples': 2967040, 'steps': 5794, 'batch_loss/train': 0.7968386858701706} +12/23/2021 17:58:41 - INFO - codeparrot_training - Step 5795: {'lr': 0.0004844676000352584, 'samples': 2967552, 'steps': 5795, 'batch_loss/train': 0.9431507186964154} +12/23/2021 17:58:52 - INFO - codeparrot_training - Step 5796: {'lr': 0.0004844621299861871, 'samples': 2968064, 'steps': 5796, 'batch_loss/train': 0.8042679475620389} +12/23/2021 17:59:02 - INFO - codeparrot_training - Step 5797: {'lr': 0.00048445665900498293, 'samples': 2968576, 'steps': 5797, 'batch_loss/train': 0.8323586732149124} +12/23/2021 17:59:14 - INFO - codeparrot_training - Step 5798: {'lr': 0.0004844511870916676, 'samples': 2969088, 'steps': 5798, 'batch_loss/train': 0.8965103412047029} +12/23/2021 17:59:25 - INFO - codeparrot_training - Step 5799: {'lr': 0.00048444571424626295, 'samples': 2969600, 'steps': 5799, 'batch_loss/train': 0.8718155124224722} +12/23/2021 17:59:35 - INFO - codeparrot_training - Step 5800: {'lr': 0.0004844402404687906, 'samples': 2970112, 'steps': 5800, 'batch_loss/train': 0.7887903396040201} +12/23/2021 17:59:49 - INFO - codeparrot_training - Step 5801: {'lr': 0.00048443476575927235, 'samples': 2970624, 'steps': 5801, 'batch_loss/train': 0.8299019364640117} +12/23/2021 18:00:00 - INFO - codeparrot_training - Step 5802: {'lr': 0.00048442929011773006, 'samples': 2971136, 'steps': 5802, 'batch_loss/train': 0.8887974796816707} +12/23/2021 18:00:10 - INFO - codeparrot_training - Step 5803: {'lr': 0.0004844238135441854, 'samples': 2971648, 'steps': 5803, 'batch_loss/train': 0.8258345681242645} +12/23/2021 18:00:22 - INFO - codeparrot_training - Step 5804: {'lr': 0.00048441833603866014, 'samples': 2972160, 'steps': 5804, 'batch_loss/train': 0.8969822460785508} +12/23/2021 18:00:33 - INFO - codeparrot_training - Step 5805: {'lr': 0.00048441285760117615, 'samples': 2972672, 'steps': 5805, 'batch_loss/train': 0.8619542452506721} +12/23/2021 18:00:44 - INFO - codeparrot_training - Step 5806: {'lr': 0.0004844073782317552, 'samples': 2973184, 'steps': 5806, 'batch_loss/train': 0.9738603178411722} +12/23/2021 18:00:56 - INFO - codeparrot_training - Step 5807: {'lr': 0.0004844018979304189, 'samples': 2973696, 'steps': 5807, 'batch_loss/train': 0.8271306455135345} +12/23/2021 18:01:06 - INFO - codeparrot_training - Step 5808: {'lr': 0.0004843964166971892, 'samples': 2974208, 'steps': 5808, 'batch_loss/train': 0.9077390963211656} +12/23/2021 18:01:17 - INFO - codeparrot_training - Step 5809: {'lr': 0.0004843909345320878, 'samples': 2974720, 'steps': 5809, 'batch_loss/train': 0.8486774093471467} +12/23/2021 18:01:27 - INFO - codeparrot_training - Step 5810: {'lr': 0.00048438545143513666, 'samples': 2975232, 'steps': 5810, 'batch_loss/train': 1.0926573490723968} +12/23/2021 18:01:40 - INFO - codeparrot_training - Step 5811: {'lr': 0.0004843799674063574, 'samples': 2975744, 'steps': 5811, 'batch_loss/train': 0.8336267080157995} +12/23/2021 18:01:50 - INFO - codeparrot_training - Step 5812: {'lr': 0.00048437448244577184, 'samples': 2976256, 'steps': 5812, 'batch_loss/train': 0.8439084365963936} +12/23/2021 18:02:01 - INFO - codeparrot_training - Step 5813: {'lr': 0.0004843689965534019, 'samples': 2976768, 'steps': 5813, 'batch_loss/train': 0.8429314438253641} +12/23/2021 18:02:15 - INFO - codeparrot_training - Step 5814: {'lr': 0.00048436350972926934, 'samples': 2977280, 'steps': 5814, 'batch_loss/train': 0.8312474051490426} +12/23/2021 18:02:26 - INFO - codeparrot_training - Step 5815: {'lr': 0.000484358021973396, 'samples': 2977792, 'steps': 5815, 'batch_loss/train': 0.9268621187657118} +12/23/2021 18:02:36 - INFO - codeparrot_training - Step 5816: {'lr': 0.00048435253328580355, 'samples': 2978304, 'steps': 5816, 'batch_loss/train': 0.8216818887740374} +12/23/2021 18:02:48 - INFO - codeparrot_training - Step 5817: {'lr': 0.0004843470436665139, 'samples': 2978816, 'steps': 5817, 'batch_loss/train': 0.7420337898656726} +12/23/2021 18:02:59 - INFO - codeparrot_training - Step 5818: {'lr': 0.00048434155311554897, 'samples': 2979328, 'steps': 5818, 'batch_loss/train': 0.8367470325902104} +12/23/2021 18:03:09 - INFO - codeparrot_training - Step 5819: {'lr': 0.0004843360616329305, 'samples': 2979840, 'steps': 5819, 'batch_loss/train': 0.7958736615255475} +12/23/2021 18:03:20 - INFO - codeparrot_training - Step 5820: {'lr': 0.0004843305692186804, 'samples': 2980352, 'steps': 5820, 'batch_loss/train': 0.8476417800411582} +12/23/2021 18:03:32 - INFO - codeparrot_training - Step 5821: {'lr': 0.00048432507587282026, 'samples': 2980864, 'steps': 5821, 'batch_loss/train': 0.9120648400858045} +12/23/2021 18:03:43 - INFO - codeparrot_training - Step 5822: {'lr': 0.00048431958159537216, 'samples': 2981376, 'steps': 5822, 'batch_loss/train': 0.9358613155782223} +12/23/2021 18:03:53 - INFO - codeparrot_training - Step 5823: {'lr': 0.00048431408638635793, 'samples': 2981888, 'steps': 5823, 'batch_loss/train': 0.7689091609790921} +12/23/2021 18:04:07 - INFO - codeparrot_training - Step 5824: {'lr': 0.0004843085902457993, 'samples': 2982400, 'steps': 5824, 'batch_loss/train': 0.891480048187077} +12/23/2021 18:04:17 - INFO - codeparrot_training - Step 5825: {'lr': 0.00048430309317371825, 'samples': 2982912, 'steps': 5825, 'batch_loss/train': 0.8775230934843421} +12/23/2021 18:04:28 - INFO - codeparrot_training - Step 5826: {'lr': 0.0004842975951701366, 'samples': 2983424, 'steps': 5826, 'batch_loss/train': 0.8653912174049765} +12/23/2021 18:04:40 - INFO - codeparrot_training - Step 5827: {'lr': 0.0004842920962350761, 'samples': 2983936, 'steps': 5827, 'batch_loss/train': 0.8171444504987448} +12/23/2021 18:04:51 - INFO - codeparrot_training - Step 5828: {'lr': 0.0004842865963685587, 'samples': 2984448, 'steps': 5828, 'batch_loss/train': 0.8398151290602982} +12/23/2021 18:05:01 - INFO - codeparrot_training - Step 5829: {'lr': 0.0004842810955706063, 'samples': 2984960, 'steps': 5829, 'batch_loss/train': 0.8316138354130089} +12/23/2021 18:05:15 - INFO - codeparrot_training - Step 5830: {'lr': 0.00048427559384124066, 'samples': 2985472, 'steps': 5830, 'batch_loss/train': 0.9397130000870675} +12/23/2021 18:05:26 - INFO - codeparrot_training - Step 5831: {'lr': 0.0004842700911804838, 'samples': 2985984, 'steps': 5831, 'batch_loss/train': 0.7919710800051689} +12/23/2021 18:05:36 - INFO - codeparrot_training - Step 5832: {'lr': 0.00048426458758835746, 'samples': 2986496, 'steps': 5832, 'batch_loss/train': 0.880448171403259} +12/23/2021 18:05:47 - INFO - codeparrot_training - Step 5833: {'lr': 0.0004842590830648836, 'samples': 2987008, 'steps': 5833, 'batch_loss/train': 0.6716298459796235} +12/23/2021 18:05:59 - INFO - codeparrot_training - Step 5834: {'lr': 0.00048425357761008403, 'samples': 2987520, 'steps': 5834, 'batch_loss/train': 0.8312158337794244} +12/23/2021 18:06:09 - INFO - codeparrot_training - Step 5835: {'lr': 0.0004842480712239807, 'samples': 2988032, 'steps': 5835, 'batch_loss/train': 0.8833014294505119} +12/23/2021 18:06:20 - INFO - codeparrot_training - Step 5836: {'lr': 0.00048424256390659553, 'samples': 2988544, 'steps': 5836, 'batch_loss/train': 0.8235817458480597} +12/23/2021 18:06:32 - INFO - codeparrot_training - Step 5837: {'lr': 0.0004842370556579504, 'samples': 2989056, 'steps': 5837, 'batch_loss/train': 0.7049318433273584} +12/23/2021 18:06:42 - INFO - codeparrot_training - Step 5838: {'lr': 0.0004842315464780671, 'samples': 2989568, 'steps': 5838, 'batch_loss/train': 0.9805558985099196} +12/23/2021 18:06:53 - INFO - codeparrot_training - Step 5839: {'lr': 0.00048422603636696763, 'samples': 2990080, 'steps': 5839, 'batch_loss/train': 0.9872951870784163} +12/23/2021 18:07:08 - INFO - codeparrot_training - Step 5840: {'lr': 0.00048422052532467395, 'samples': 2990592, 'steps': 5840, 'batch_loss/train': 1.0052135940641165} +12/23/2021 18:07:19 - INFO - codeparrot_training - Step 5841: {'lr': 0.00048421501335120775, 'samples': 2991104, 'steps': 5841, 'batch_loss/train': 3.9218246303498745} +12/23/2021 18:07:29 - INFO - codeparrot_training - Step 5842: {'lr': 0.0004842095004465912, 'samples': 2991616, 'steps': 5842, 'batch_loss/train': 0.8735242439433932} +12/23/2021 18:07:40 - INFO - codeparrot_training - Step 5843: {'lr': 0.00048420398661084613, 'samples': 2992128, 'steps': 5843, 'batch_loss/train': 0.886126521974802} +12/23/2021 18:07:52 - INFO - codeparrot_training - Step 5844: {'lr': 0.00048419847184399446, 'samples': 2992640, 'steps': 5844, 'batch_loss/train': 0.9197756946086884} +12/23/2021 18:08:02 - INFO - codeparrot_training - Step 5845: {'lr': 0.00048419295614605804, 'samples': 2993152, 'steps': 5845, 'batch_loss/train': 0.9648728920146823} +12/23/2021 18:08:13 - INFO - codeparrot_training - Step 5846: {'lr': 0.0004841874395170588, 'samples': 2993664, 'steps': 5846, 'batch_loss/train': 0.875336735509336} +12/23/2021 18:08:25 - INFO - codeparrot_training - Step 5847: {'lr': 0.00048418192195701884, 'samples': 2994176, 'steps': 5847, 'batch_loss/train': 0.7077372325584292} +12/23/2021 18:08:36 - INFO - codeparrot_training - Step 5848: {'lr': 0.0004841764034659599, 'samples': 2994688, 'steps': 5848, 'batch_loss/train': 0.6185529751237482} +12/23/2021 18:08:46 - INFO - codeparrot_training - Step 5849: {'lr': 0.000484170884043904, 'samples': 2995200, 'steps': 5849, 'batch_loss/train': 0.7704075709916651} +12/23/2021 18:08:59 - INFO - codeparrot_training - Step 5850: {'lr': 0.00048416536369087315, 'samples': 2995712, 'steps': 5850, 'batch_loss/train': 0.8455570274963975} +12/23/2021 18:09:09 - INFO - codeparrot_training - Step 5851: {'lr': 0.00048415984240688916, 'samples': 2996224, 'steps': 5851, 'batch_loss/train': 0.8689518561586738} +12/23/2021 18:09:20 - INFO - codeparrot_training - Step 5852: {'lr': 0.00048415432019197416, 'samples': 2996736, 'steps': 5852, 'batch_loss/train': 0.8580337492749095} +12/23/2021 18:09:31 - INFO - codeparrot_training - Step 5853: {'lr': 0.0004841487970461499, 'samples': 2997248, 'steps': 5853, 'batch_loss/train': 0.8797200359404087} +12/23/2021 18:09:44 - INFO - codeparrot_training - Step 5854: {'lr': 0.0004841432729694384, 'samples': 2997760, 'steps': 5854, 'batch_loss/train': 0.8501955345273018} +12/23/2021 18:09:55 - INFO - codeparrot_training - Step 5855: {'lr': 0.0004841377479618617, 'samples': 2998272, 'steps': 5855, 'batch_loss/train': 0.8610369497910142} +12/23/2021 18:10:05 - INFO - codeparrot_training - Step 5856: {'lr': 0.00048413222202344175, 'samples': 2998784, 'steps': 5856, 'batch_loss/train': 0.8796071424148977} +12/23/2021 18:10:18 - INFO - codeparrot_training - Step 5857: {'lr': 0.0004841266951542005, 'samples': 2999296, 'steps': 5857, 'batch_loss/train': 0.73899145424366} +12/23/2021 18:10:28 - INFO - codeparrot_training - Step 5858: {'lr': 0.0004841211673541599, 'samples': 2999808, 'steps': 5858, 'batch_loss/train': 0.7185219149105251} +12/23/2021 18:10:39 - INFO - codeparrot_training - Step 5859: {'lr': 0.00048411563862334195, 'samples': 3000320, 'steps': 5859, 'batch_loss/train': 0.6403381248237565} +12/23/2021 18:10:51 - INFO - codeparrot_training - Step 5860: {'lr': 0.00048411010896176853, 'samples': 3000832, 'steps': 5860, 'batch_loss/train': 0.7100799726322293} +12/23/2021 18:11:02 - INFO - codeparrot_training - Step 5861: {'lr': 0.0004841045783694618, 'samples': 3001344, 'steps': 5861, 'batch_loss/train': 0.8244436588138342} +12/23/2021 18:11:12 - INFO - codeparrot_training - Step 5862: {'lr': 0.0004840990468464437, 'samples': 3001856, 'steps': 5862, 'batch_loss/train': 0.7207987507572398} +12/23/2021 18:11:26 - INFO - codeparrot_training - Step 5863: {'lr': 0.0004840935143927361, 'samples': 3002368, 'steps': 5863, 'batch_loss/train': 0.926836606580764} +12/23/2021 18:11:36 - INFO - codeparrot_training - Step 5864: {'lr': 0.00048408798100836104, 'samples': 3002880, 'steps': 5864, 'batch_loss/train': 0.7815721521619707} +12/23/2021 18:11:47 - INFO - codeparrot_training - Step 5865: {'lr': 0.0004840824466933407, 'samples': 3003392, 'steps': 5865, 'batch_loss/train': 0.868167910259217} +12/23/2021 18:11:58 - INFO - codeparrot_training - Step 5866: {'lr': 0.00048407691144769684, 'samples': 3003904, 'steps': 5866, 'batch_loss/train': 0.810678671579808} +12/23/2021 18:12:10 - INFO - codeparrot_training - Step 5867: {'lr': 0.00048407137527145154, 'samples': 3004416, 'steps': 5867, 'batch_loss/train': 0.648551924386993} +12/23/2021 18:12:20 - INFO - codeparrot_training - Step 5868: {'lr': 0.0004840658381646269, 'samples': 3004928, 'steps': 5868, 'batch_loss/train': 0.8872129004448652} +12/23/2021 18:12:31 - INFO - codeparrot_training - Step 5869: {'lr': 0.0004840603001272448, 'samples': 3005440, 'steps': 5869, 'batch_loss/train': 0.7358945580199361} +12/23/2021 18:12:45 - INFO - codeparrot_training - Step 5870: {'lr': 0.00048405476115932734, 'samples': 3005952, 'steps': 5870, 'batch_loss/train': 0.7914250721223652} +12/23/2021 18:12:55 - INFO - codeparrot_training - Step 5871: {'lr': 0.00048404922126089655, 'samples': 3006464, 'steps': 5871, 'batch_loss/train': 0.9335980229079723} +12/23/2021 18:13:06 - INFO - codeparrot_training - Step 5872: {'lr': 0.00048404368043197446, 'samples': 3006976, 'steps': 5872, 'batch_loss/train': 0.7655609147623181} +12/23/2021 18:13:18 - INFO - codeparrot_training - Step 5873: {'lr': 0.0004840381386725831, 'samples': 3007488, 'steps': 5873, 'batch_loss/train': 0.8234609849750996} +12/23/2021 18:13:28 - INFO - codeparrot_training - Step 5874: {'lr': 0.00048403259598274435, 'samples': 3008000, 'steps': 5874, 'batch_loss/train': 0.8895758548751473} +12/23/2021 18:13:39 - INFO - codeparrot_training - Step 5875: {'lr': 0.0004840270523624804, 'samples': 3008512, 'steps': 5875, 'batch_loss/train': 0.795543879037723} +12/23/2021 18:13:49 - INFO - codeparrot_training - Step 5876: {'lr': 0.00048402150781181333, 'samples': 3009024, 'steps': 5876, 'batch_loss/train': 0.8219110975041986} +12/23/2021 18:14:02 - INFO - codeparrot_training - Step 5877: {'lr': 0.00048401596233076506, 'samples': 3009536, 'steps': 5877, 'batch_loss/train': 0.8186393054202199} +12/23/2021 18:14:12 - INFO - codeparrot_training - Step 5878: {'lr': 0.00048401041591935765, 'samples': 3010048, 'steps': 5878, 'batch_loss/train': 0.8694507135078311} +12/23/2021 18:14:23 - INFO - codeparrot_training - Step 5879: {'lr': 0.00048400486857761316, 'samples': 3010560, 'steps': 5879, 'batch_loss/train': 0.9220739644952118} +12/23/2021 18:14:36 - INFO - codeparrot_training - Step 5880: {'lr': 0.00048399932030555377, 'samples': 3011072, 'steps': 5880, 'batch_loss/train': 0.8920849887654185} +12/23/2021 18:14:47 - INFO - codeparrot_training - Step 5881: {'lr': 0.0004839937711032014, 'samples': 3011584, 'steps': 5881, 'batch_loss/train': 0.8300386248156428} +12/23/2021 18:14:57 - INFO - codeparrot_training - Step 5882: {'lr': 0.00048398822097057813, 'samples': 3012096, 'steps': 5882, 'batch_loss/train': 0.8608750870334916} +12/23/2021 18:15:10 - INFO - codeparrot_training - Step 5883: {'lr': 0.000483982669907706, 'samples': 3012608, 'steps': 5883, 'batch_loss/train': 0.5857427216833457} +12/23/2021 18:15:21 - INFO - codeparrot_training - Step 5884: {'lr': 0.0004839771179146072, 'samples': 3013120, 'steps': 5884, 'batch_loss/train': 0.8927373439073563} +12/23/2021 18:15:31 - INFO - codeparrot_training - Step 5885: {'lr': 0.0004839715649913037, 'samples': 3013632, 'steps': 5885, 'batch_loss/train': 0.9143444001674652} +12/23/2021 18:15:42 - INFO - codeparrot_training - Step 5886: {'lr': 0.0004839660111378176, 'samples': 3014144, 'steps': 5886, 'batch_loss/train': 1.088699326151982} +12/23/2021 18:15:54 - INFO - codeparrot_training - Step 5887: {'lr': 0.000483960456354171, 'samples': 3014656, 'steps': 5887, 'batch_loss/train': 0.8673433605581522} +12/23/2021 18:16:04 - INFO - codeparrot_training - Step 5888: {'lr': 0.00048395490064038593, 'samples': 3015168, 'steps': 5888, 'batch_loss/train': 0.802117079962045} +12/23/2021 18:16:15 - INFO - codeparrot_training - Step 5889: {'lr': 0.00048394934399648463, 'samples': 3015680, 'steps': 5889, 'batch_loss/train': 0.7741905804723501} +12/23/2021 18:16:27 - INFO - codeparrot_training - Step 5890: {'lr': 0.00048394378642248895, 'samples': 3016192, 'steps': 5890, 'batch_loss/train': 0.8539569918066263} +12/23/2021 18:16:38 - INFO - codeparrot_training - Step 5891: {'lr': 0.0004839382279184211, 'samples': 3016704, 'steps': 5891, 'batch_loss/train': 0.8418521154671907} +12/23/2021 18:16:48 - INFO - codeparrot_training - Step 5892: {'lr': 0.00048393266848430325, 'samples': 3017216, 'steps': 5892, 'batch_loss/train': 1.0218718964606524} +12/23/2021 18:17:02 - INFO - codeparrot_training - Step 5893: {'lr': 0.0004839271081201574, 'samples': 3017728, 'steps': 5893, 'batch_loss/train': 0.7163914463017136} +12/23/2021 18:17:13 - INFO - codeparrot_training - Step 5894: {'lr': 0.0004839215468260057, 'samples': 3018240, 'steps': 5894, 'batch_loss/train': 0.7738532070070505} +12/23/2021 18:17:23 - INFO - codeparrot_training - Step 5895: {'lr': 0.00048391598460187027, 'samples': 3018752, 'steps': 5895, 'batch_loss/train': 0.8983491808176041} +12/23/2021 18:17:34 - INFO - codeparrot_training - Step 5896: {'lr': 0.0004839104214477732, 'samples': 3019264, 'steps': 5896, 'batch_loss/train': 0.7808998366817832} +12/23/2021 18:17:46 - INFO - codeparrot_training - Step 5897: {'lr': 0.00048390485736373665, 'samples': 3019776, 'steps': 5897, 'batch_loss/train': 0.8678877027705312} +12/23/2021 18:17:57 - INFO - codeparrot_training - Step 5898: {'lr': 0.00048389929234978266, 'samples': 3020288, 'steps': 5898, 'batch_loss/train': 0.6519565869821236} +12/23/2021 18:18:07 - INFO - codeparrot_training - Step 5899: {'lr': 0.00048389372640593345, 'samples': 3020800, 'steps': 5899, 'batch_loss/train': 0.8964795849751681} +12/23/2021 18:18:21 - INFO - codeparrot_training - Step 5900: {'lr': 0.000483888159532211, 'samples': 3021312, 'steps': 5900, 'batch_loss/train': 0.7515653483569622} +12/23/2021 18:18:31 - INFO - codeparrot_training - Step 5901: {'lr': 0.00048388259172863765, 'samples': 3021824, 'steps': 5901, 'batch_loss/train': 0.8825102504342794} +12/23/2021 18:18:42 - INFO - codeparrot_training - Step 5902: {'lr': 0.0004838770229952354, 'samples': 3022336, 'steps': 5902, 'batch_loss/train': 0.7940139025449753} +12/23/2021 18:18:54 - INFO - codeparrot_training - Step 5903: {'lr': 0.0004838714533320263, 'samples': 3022848, 'steps': 5903, 'batch_loss/train': 0.9552051154896617} +12/23/2021 18:19:04 - INFO - codeparrot_training - Step 5904: {'lr': 0.00048386588273903274, 'samples': 3023360, 'steps': 5904, 'batch_loss/train': 0.8947020824998617} +12/23/2021 18:19:15 - INFO - codeparrot_training - Step 5905: {'lr': 0.0004838603112162767, 'samples': 3023872, 'steps': 5905, 'batch_loss/train': 0.7900797035545111} +12/23/2021 18:19:26 - INFO - codeparrot_training - Step 5906: {'lr': 0.0004838547387637804, 'samples': 3024384, 'steps': 5906, 'batch_loss/train': 0.8998640021309257} +12/23/2021 18:19:38 - INFO - codeparrot_training - Step 5907: {'lr': 0.00048384916538156587, 'samples': 3024896, 'steps': 5907, 'batch_loss/train': 0.8650648687034845} +12/23/2021 18:19:48 - INFO - codeparrot_training - Step 5908: {'lr': 0.0004838435910696555, 'samples': 3025408, 'steps': 5908, 'batch_loss/train': 0.7423968308139592} +12/23/2021 18:19:59 - INFO - codeparrot_training - Step 5909: {'lr': 0.00048383801582807117, 'samples': 3025920, 'steps': 5909, 'batch_loss/train': 0.6298225386999547} +12/23/2021 18:20:12 - INFO - codeparrot_training - Step 5910: {'lr': 0.0004838324396568352, 'samples': 3026432, 'steps': 5910, 'batch_loss/train': 0.9363216264173388} +12/23/2021 18:20:23 - INFO - codeparrot_training - Step 5911: {'lr': 0.00048382686255596976, 'samples': 3026944, 'steps': 5911, 'batch_loss/train': 0.935461837798357} +12/23/2021 18:20:33 - INFO - codeparrot_training - Step 5912: {'lr': 0.000483821284525497, 'samples': 3027456, 'steps': 5912, 'batch_loss/train': 0.7423773258924484} +12/23/2021 18:20:45 - INFO - codeparrot_training - Step 5913: {'lr': 0.0004838157055654392, 'samples': 3027968, 'steps': 5913, 'batch_loss/train': 0.8543748846277595} +12/23/2021 18:20:56 - INFO - codeparrot_training - Step 5914: {'lr': 0.00048381012567581827, 'samples': 3028480, 'steps': 5914, 'batch_loss/train': 0.8287508133798838} +12/23/2021 18:21:07 - INFO - codeparrot_training - Step 5915: {'lr': 0.0004838045448566567, 'samples': 3028992, 'steps': 5915, 'batch_loss/train': 0.7635943321511149} +12/23/2021 18:21:17 - INFO - codeparrot_training - Step 5916: {'lr': 0.0004837989631079764, 'samples': 3029504, 'steps': 5916, 'batch_loss/train': 0.9069278221577406} +12/23/2021 18:21:29 - INFO - codeparrot_training - Step 5917: {'lr': 0.00048379338042979985, 'samples': 3030016, 'steps': 5917, 'batch_loss/train': 0.8368039270862937} +12/23/2021 18:21:40 - INFO - codeparrot_training - Step 5918: {'lr': 0.000483787796822149, 'samples': 3030528, 'steps': 5918, 'batch_loss/train': 0.8643777701072395} +12/23/2021 18:21:50 - INFO - codeparrot_training - Step 5919: {'lr': 0.00048378221228504624, 'samples': 3031040, 'steps': 5919, 'batch_loss/train': 0.890282591804862} +12/23/2021 18:22:02 - INFO - codeparrot_training - Step 5920: {'lr': 0.00048377662681851365, 'samples': 3031552, 'steps': 5920, 'batch_loss/train': 0.8312801364809275} +12/23/2021 18:22:13 - INFO - codeparrot_training - Step 5921: {'lr': 0.00048377104042257336, 'samples': 3032064, 'steps': 5921, 'batch_loss/train': 0.8321523237973452} +12/23/2021 18:22:23 - INFO - codeparrot_training - Step 5922: {'lr': 0.0004837654530972478, 'samples': 3032576, 'steps': 5922, 'batch_loss/train': 0.7769948198692873} +12/23/2021 18:22:37 - INFO - codeparrot_training - Step 5923: {'lr': 0.000483759864842559, 'samples': 3033088, 'steps': 5923, 'batch_loss/train': 0.8089629737660289} +12/23/2021 18:22:48 - INFO - codeparrot_training - Step 5924: {'lr': 0.00048375427565852924, 'samples': 3033600, 'steps': 5924, 'batch_loss/train': 0.8336866507306695} +12/23/2021 18:22:58 - INFO - codeparrot_training - Step 5925: {'lr': 0.00048374868554518083, 'samples': 3034112, 'steps': 5925, 'batch_loss/train': 0.8991731656715274} +12/23/2021 18:23:10 - INFO - codeparrot_training - Step 5926: {'lr': 0.0004837430945025358, 'samples': 3034624, 'steps': 5926, 'batch_loss/train': 0.8068174128420651} +12/23/2021 18:23:21 - INFO - codeparrot_training - Step 5927: {'lr': 0.00048373750253061655, 'samples': 3035136, 'steps': 5927, 'batch_loss/train': 0.8222529571503401} +12/23/2021 18:23:32 - INFO - codeparrot_training - Step 5928: {'lr': 0.00048373190962944517, 'samples': 3035648, 'steps': 5928, 'batch_loss/train': 0.9597984892316163} +12/23/2021 18:23:42 - INFO - codeparrot_training - Step 5929: {'lr': 0.000483726315799044, 'samples': 3036160, 'steps': 5929, 'batch_loss/train': 0.696131584700197} +12/23/2021 18:23:56 - INFO - codeparrot_training - Step 5930: {'lr': 0.00048372072103943534, 'samples': 3036672, 'steps': 5930, 'batch_loss/train': 0.8417520001530647} +12/23/2021 18:24:07 - INFO - codeparrot_training - Step 5931: {'lr': 0.00048371512535064123, 'samples': 3037184, 'steps': 5931, 'batch_loss/train': 1.3293260484933853} +12/23/2021 18:24:17 - INFO - codeparrot_training - Step 5932: {'lr': 0.00048370952873268407, 'samples': 3037696, 'steps': 5932, 'batch_loss/train': 0.8406922807916999} +12/23/2021 18:24:29 - INFO - codeparrot_training - Step 5933: {'lr': 0.0004837039311855861, 'samples': 3038208, 'steps': 5933, 'batch_loss/train': 0.7413743166252971} +12/23/2021 18:24:40 - INFO - codeparrot_training - Step 5934: {'lr': 0.00048369833270936946, 'samples': 3038720, 'steps': 5934, 'batch_loss/train': 0.8279759557917714} +12/23/2021 18:24:51 - INFO - codeparrot_training - Step 5935: {'lr': 0.0004836927333040566, 'samples': 3039232, 'steps': 5935, 'batch_loss/train': 0.8022311292588711} +12/23/2021 18:25:01 - INFO - codeparrot_training - Step 5936: {'lr': 0.0004836871329696696, 'samples': 3039744, 'steps': 5936, 'batch_loss/train': 0.6844540894962847} +12/23/2021 18:25:13 - INFO - codeparrot_training - Step 5937: {'lr': 0.0004836815317062308, 'samples': 3040256, 'steps': 5937, 'batch_loss/train': 0.8421481447294354} +12/23/2021 18:25:24 - INFO - codeparrot_training - Step 5938: {'lr': 0.0004836759295137625, 'samples': 3040768, 'steps': 5938, 'batch_loss/train': 0.727612632792443} +12/23/2021 18:25:35 - INFO - codeparrot_training - Step 5939: {'lr': 0.0004836703263922869, 'samples': 3041280, 'steps': 5939, 'batch_loss/train': 0.8579008253291249} +12/23/2021 18:25:48 - INFO - codeparrot_training - Step 5940: {'lr': 0.00048366472234182635, 'samples': 3041792, 'steps': 5940, 'batch_loss/train': 0.8456213781610131} +12/23/2021 18:25:59 - INFO - codeparrot_training - Step 5941: {'lr': 0.00048365911736240305, 'samples': 3042304, 'steps': 5941, 'batch_loss/train': 0.9164325152523816} +12/23/2021 18:26:10 - INFO - codeparrot_training - Step 5942: {'lr': 0.00048365351145403934, 'samples': 3042816, 'steps': 5942, 'batch_loss/train': 0.9445583280175924} +12/23/2021 18:26:22 - INFO - codeparrot_training - Step 5943: {'lr': 0.00048364790461675745, 'samples': 3043328, 'steps': 5943, 'batch_loss/train': 1.048178169876337} +12/23/2021 18:26:32 - INFO - codeparrot_training - Step 5944: {'lr': 0.0004836422968505798, 'samples': 3043840, 'steps': 5944, 'batch_loss/train': 0.9658558396622539} +12/23/2021 18:26:43 - INFO - codeparrot_training - Step 5945: {'lr': 0.0004836366881555285, 'samples': 3044352, 'steps': 5945, 'batch_loss/train': 0.8396646352484822} +12/23/2021 18:26:55 - INFO - codeparrot_training - Step 5946: {'lr': 0.000483631078531626, 'samples': 3044864, 'steps': 5946, 'batch_loss/train': 0.8205954618752003} +12/23/2021 18:27:05 - INFO - codeparrot_training - Step 5947: {'lr': 0.00048362546797889463, 'samples': 3045376, 'steps': 5947, 'batch_loss/train': 0.911721358075738} +12/23/2021 18:27:16 - INFO - codeparrot_training - Step 5948: {'lr': 0.0004836198564973565, 'samples': 3045888, 'steps': 5948, 'batch_loss/train': 1.050477771088481} +12/23/2021 18:27:27 - INFO - codeparrot_training - Step 5949: {'lr': 0.00048361424408703405, 'samples': 3046400, 'steps': 5949, 'batch_loss/train': 0.9376128036528826} +12/23/2021 18:27:40 - INFO - codeparrot_training - Step 5950: {'lr': 0.0004836086307479496, 'samples': 3046912, 'steps': 5950, 'batch_loss/train': 0.8373533766716719} +12/23/2021 18:27:50 - INFO - codeparrot_training - Step 5951: {'lr': 0.00048360301648012546, 'samples': 3047424, 'steps': 5951, 'batch_loss/train': 0.839091713540256} +12/23/2021 18:28:01 - INFO - codeparrot_training - Step 5952: {'lr': 0.0004835974012835839, 'samples': 3047936, 'steps': 5952, 'batch_loss/train': 0.9065551529638469} +12/23/2021 18:28:13 - INFO - codeparrot_training - Step 5953: {'lr': 0.0004835917851583472, 'samples': 3048448, 'steps': 5953, 'batch_loss/train': 0.7569553982466459} +12/23/2021 18:28:23 - INFO - codeparrot_training - Step 5954: {'lr': 0.0004835861681044379, 'samples': 3048960, 'steps': 5954, 'batch_loss/train': 0.9055330436676741} +12/23/2021 18:28:34 - INFO - codeparrot_training - Step 5955: {'lr': 0.0004835805501218781, 'samples': 3049472, 'steps': 5955, 'batch_loss/train': 0.8092591767199337} +12/23/2021 18:28:46 - INFO - codeparrot_training - Step 5956: {'lr': 0.0004835749312106903, 'samples': 3049984, 'steps': 5956, 'batch_loss/train': 0.8787575841415673} +12/23/2021 18:28:57 - INFO - codeparrot_training - Step 5957: {'lr': 0.0004835693113708968, 'samples': 3050496, 'steps': 5957, 'batch_loss/train': 0.879173609893769} +12/23/2021 18:29:07 - INFO - codeparrot_training - Step 5958: {'lr': 0.0004835636906025198, 'samples': 3051008, 'steps': 5958, 'batch_loss/train': 0.8264527887804434} +12/23/2021 18:29:18 - INFO - codeparrot_training - Step 5959: {'lr': 0.00048355806890558174, 'samples': 3051520, 'steps': 5959, 'batch_loss/train': 0.9518850147724152} +12/23/2021 18:29:31 - INFO - codeparrot_training - Step 5960: {'lr': 0.0004835524462801051, 'samples': 3052032, 'steps': 5960, 'batch_loss/train': 0.8356580315157771} +12/23/2021 18:29:42 - INFO - codeparrot_training - Step 5961: {'lr': 0.00048354682272611207, 'samples': 3052544, 'steps': 5961, 'batch_loss/train': 0.826870153658092} +12/23/2021 18:29:53 - INFO - codeparrot_training - Step 5962: {'lr': 0.000483541198243625, 'samples': 3053056, 'steps': 5962, 'batch_loss/train': 0.8062297524884343} +12/23/2021 18:30:05 - INFO - codeparrot_training - Step 5963: {'lr': 0.00048353557283266634, 'samples': 3053568, 'steps': 5963, 'batch_loss/train': 0.8778781611472368} +12/23/2021 18:30:15 - INFO - codeparrot_training - Step 5964: {'lr': 0.0004835299464932585, 'samples': 3054080, 'steps': 5964, 'batch_loss/train': 0.8918313686735928} +12/23/2021 18:30:26 - INFO - codeparrot_training - Step 5965: {'lr': 0.00048352431922542375, 'samples': 3054592, 'steps': 5965, 'batch_loss/train': 0.8362298272550106} +12/23/2021 18:30:38 - INFO - codeparrot_training - Step 5966: {'lr': 0.0004835186910291844, 'samples': 3055104, 'steps': 5966, 'batch_loss/train': 0.8554074121639132} +12/23/2021 18:30:48 - INFO - codeparrot_training - Step 5967: {'lr': 0.0004835130619045629, 'samples': 3055616, 'steps': 5967, 'batch_loss/train': 1.0016941986978054} +12/23/2021 18:30:59 - INFO - codeparrot_training - Step 5968: {'lr': 0.0004835074318515817, 'samples': 3056128, 'steps': 5968, 'batch_loss/train': 0.7218209733255208} +12/23/2021 18:31:12 - INFO - codeparrot_training - Step 5969: {'lr': 0.0004835018008702631, 'samples': 3056640, 'steps': 5969, 'batch_loss/train': 0.8479747083038092} +12/23/2021 18:31:23 - INFO - codeparrot_training - Step 5970: {'lr': 0.00048349616896062954, 'samples': 3057152, 'steps': 5970, 'batch_loss/train': 0.8528989123879} +12/23/2021 18:31:33 - INFO - codeparrot_training - Step 5971: {'lr': 0.00048349053612270335, 'samples': 3057664, 'steps': 5971, 'batch_loss/train': 0.8675800366327167} +12/23/2021 18:31:44 - INFO - codeparrot_training - Step 5972: {'lr': 0.000483484902356507, 'samples': 3058176, 'steps': 5972, 'batch_loss/train': 0.9167897319421172} +12/23/2021 18:31:56 - INFO - codeparrot_training - Step 5973: {'lr': 0.00048347926766206274, 'samples': 3058688, 'steps': 5973, 'batch_loss/train': 0.8887089462950826} +12/23/2021 18:32:07 - INFO - codeparrot_training - Step 5974: {'lr': 0.0004834736320393931, 'samples': 3059200, 'steps': 5974, 'batch_loss/train': 0.8711890717968345} +12/23/2021 18:32:17 - INFO - codeparrot_training - Step 5975: {'lr': 0.0004834679954885205, 'samples': 3059712, 'steps': 5975, 'batch_loss/train': 0.7936429511755705} +12/23/2021 18:32:29 - INFO - codeparrot_training - Step 5976: {'lr': 0.0004834623580094673, 'samples': 3060224, 'steps': 5976, 'batch_loss/train': 0.8866534028202295} +12/23/2021 18:32:40 - INFO - codeparrot_training - Step 5977: {'lr': 0.0004834567196022559, 'samples': 3060736, 'steps': 5977, 'batch_loss/train': 0.7090451908297837} +12/23/2021 18:32:50 - INFO - codeparrot_training - Step 5978: {'lr': 0.0004834510802669087, 'samples': 3061248, 'steps': 5978, 'batch_loss/train': 0.6976567292585969} +12/23/2021 18:33:03 - INFO - codeparrot_training - Step 5979: {'lr': 0.0004834454400034482, 'samples': 3061760, 'steps': 5979, 'batch_loss/train': 0.7815812593325973} +12/23/2021 18:33:14 - INFO - codeparrot_training - Step 5980: {'lr': 0.00048343979881189673, 'samples': 3062272, 'steps': 5980, 'batch_loss/train': 0.8052142797969282} +12/23/2021 18:33:25 - INFO - codeparrot_training - Step 5981: {'lr': 0.0004834341566922768, 'samples': 3062784, 'steps': 5981, 'batch_loss/train': 0.850577131845057} +12/23/2021 18:33:35 - INFO - codeparrot_training - Step 5982: {'lr': 0.00048342851364461083, 'samples': 3063296, 'steps': 5982, 'batch_loss/train': 0.9440705347806215} +12/23/2021 18:33:47 - INFO - codeparrot_training - Step 5983: {'lr': 0.0004834228696689212, 'samples': 3063808, 'steps': 5983, 'batch_loss/train': 0.8709281287156045} +12/23/2021 18:33:58 - INFO - codeparrot_training - Step 5984: {'lr': 0.00048341722476523034, 'samples': 3064320, 'steps': 5984, 'batch_loss/train': 0.8309761034324765} +12/23/2021 18:34:09 - INFO - codeparrot_training - Step 5985: {'lr': 0.0004834115789335608, 'samples': 3064832, 'steps': 5985, 'batch_loss/train': 0.7751219645142555} +12/23/2021 18:34:22 - INFO - codeparrot_training - Step 5986: {'lr': 0.0004834059321739349, 'samples': 3065344, 'steps': 5986, 'batch_loss/train': 0.9069859487935901} +12/23/2021 18:34:33 - INFO - codeparrot_training - Step 5987: {'lr': 0.0004834002844863752, 'samples': 3065856, 'steps': 5987, 'batch_loss/train': 0.9089980474673212} +12/23/2021 18:34:43 - INFO - codeparrot_training - Step 5988: {'lr': 0.00048339463587090405, 'samples': 3066368, 'steps': 5988, 'batch_loss/train': 0.8730621905997396} +12/23/2021 18:34:55 - INFO - codeparrot_training - Step 5989: {'lr': 0.0004833889863275439, 'samples': 3066880, 'steps': 5989, 'batch_loss/train': 0.8102513086050749} +12/23/2021 18:35:06 - INFO - codeparrot_training - Step 5990: {'lr': 0.00048338333585631736, 'samples': 3067392, 'steps': 5990, 'batch_loss/train': 0.8804748326074332} +12/23/2021 18:35:16 - INFO - codeparrot_training - Step 5991: {'lr': 0.00048337768445724674, 'samples': 3067904, 'steps': 5991, 'batch_loss/train': 0.8145453683100641} +12/23/2021 18:35:29 - INFO - codeparrot_training - Step 5992: {'lr': 0.0004833720321303546, 'samples': 3068416, 'steps': 5992, 'batch_loss/train': 0.8424277566373348} +12/23/2021 18:35:39 - INFO - codeparrot_training - Step 5993: {'lr': 0.00048336637887566337, 'samples': 3068928, 'steps': 5993, 'batch_loss/train': 0.7449813117273152} +12/23/2021 18:35:50 - INFO - codeparrot_training - Step 5994: {'lr': 0.00048336072469319554, 'samples': 3069440, 'steps': 5994, 'batch_loss/train': 0.841082124505192} +12/23/2021 18:36:01 - INFO - codeparrot_training - Step 5995: {'lr': 0.0004833550695829736, 'samples': 3069952, 'steps': 5995, 'batch_loss/train': 0.6587314223870635} +12/23/2021 18:36:13 - INFO - codeparrot_training - Step 5996: {'lr': 0.00048334941354501995, 'samples': 3070464, 'steps': 5996, 'batch_loss/train': 0.8242392148822546} +12/23/2021 18:36:23 - INFO - codeparrot_training - Step 5997: {'lr': 0.00048334375657935717, 'samples': 3070976, 'steps': 5997, 'batch_loss/train': 0.8484153663739562} +12/23/2021 18:36:34 - INFO - codeparrot_training - Step 5998: {'lr': 0.00048333809868600776, 'samples': 3071488, 'steps': 5998, 'batch_loss/train': 0.7519930629059672} +12/23/2021 18:36:47 - INFO - codeparrot_training - Step 5999: {'lr': 0.00048333243986499415, 'samples': 3072000, 'steps': 5999, 'batch_loss/train': 0.8782752612605691} +12/23/2021 18:36:57 - INFO - codeparrot_training - Step 6000: {'lr': 0.00048332678011633884, 'samples': 3072512, 'steps': 6000, 'batch_loss/train': 0.8264210755005479} +12/23/2021 18:37:08 - INFO - codeparrot_training - Step 6001: {'lr': 0.0004833211194400644, 'samples': 3073024, 'steps': 6001, 'batch_loss/train': 0.9253470338881016} +12/23/2021 18:37:20 - INFO - codeparrot_training - Step 6002: {'lr': 0.0004833154578361932, 'samples': 3073536, 'steps': 6002, 'batch_loss/train': 0.8473845296539366} +12/23/2021 18:37:31 - INFO - codeparrot_training - Step 6003: {'lr': 0.00048330979530474797, 'samples': 3074048, 'steps': 6003, 'batch_loss/train': 0.9023696146905422} +12/23/2021 18:37:41 - INFO - codeparrot_training - Step 6004: {'lr': 0.000483304131845751, 'samples': 3074560, 'steps': 6004, 'batch_loss/train': 0.9237347561866045} +12/23/2021 18:37:52 - INFO - codeparrot_training - Step 6005: {'lr': 0.00048329846745922487, 'samples': 3075072, 'steps': 6005, 'batch_loss/train': 0.7790967528708279} +12/23/2021 18:38:04 - INFO - codeparrot_training - Step 6006: {'lr': 0.00048329280214519225, 'samples': 3075584, 'steps': 6006, 'batch_loss/train': 0.8019183361902833} +12/23/2021 18:38:14 - INFO - codeparrot_training - Step 6007: {'lr': 0.0004832871359036754, 'samples': 3076096, 'steps': 6007, 'batch_loss/train': 0.8217471595853567} +12/23/2021 18:38:25 - INFO - codeparrot_training - Step 6008: {'lr': 0.000483281468734697, 'samples': 3076608, 'steps': 6008, 'batch_loss/train': 0.877291375072673} +12/23/2021 18:38:38 - INFO - codeparrot_training - Step 6009: {'lr': 0.0004832758006382796, 'samples': 3077120, 'steps': 6009, 'batch_loss/train': 0.8736789496615529} +12/23/2021 18:38:49 - INFO - codeparrot_training - Step 6010: {'lr': 0.00048327013161444574, 'samples': 3077632, 'steps': 6010, 'batch_loss/train': 0.9237018302083015} +12/23/2021 18:38:59 - INFO - codeparrot_training - Step 6011: {'lr': 0.00048326446166321787, 'samples': 3078144, 'steps': 6011, 'batch_loss/train': 0.699217552319169} +12/23/2021 18:39:11 - INFO - codeparrot_training - Step 6012: {'lr': 0.00048325879078461853, 'samples': 3078656, 'steps': 6012, 'batch_loss/train': 0.8516709730029106} +12/23/2021 18:39:22 - INFO - codeparrot_training - Step 6013: {'lr': 0.0004832531189786704, 'samples': 3079168, 'steps': 6013, 'batch_loss/train': 0.7765497546643019} +12/23/2021 18:39:33 - INFO - codeparrot_training - Step 6014: {'lr': 0.00048324744624539585, 'samples': 3079680, 'steps': 6014, 'batch_loss/train': 0.8897068416699767} +12/23/2021 18:39:47 - INFO - codeparrot_training - Step 6015: {'lr': 0.0004832417725848176, 'samples': 3080192, 'steps': 6015, 'batch_loss/train': 0.9273698022589087} +12/23/2021 18:39:58 - INFO - codeparrot_training - Step 6016: {'lr': 0.0004832360979969581, 'samples': 3080704, 'steps': 6016, 'batch_loss/train': 0.9752066088840365} +12/23/2021 18:40:08 - INFO - codeparrot_training - Step 6017: {'lr': 0.00048323042248184, 'samples': 3081216, 'steps': 6017, 'batch_loss/train': 0.7916126633062959} +12/23/2021 18:40:19 - INFO - codeparrot_training - Step 6018: {'lr': 0.00048322474603948576, 'samples': 3081728, 'steps': 6018, 'batch_loss/train': 0.8713384144939482} +12/23/2021 18:40:31 - INFO - codeparrot_training - Step 6019: {'lr': 0.00048321906866991796, 'samples': 3082240, 'steps': 6019, 'batch_loss/train': 0.7927553588524461} +12/23/2021 18:40:41 - INFO - codeparrot_training - Step 6020: {'lr': 0.0004832133903731593, 'samples': 3082752, 'steps': 6020, 'batch_loss/train': 0.8820639187470078} +12/23/2021 18:40:52 - INFO - codeparrot_training - Step 6021: {'lr': 0.0004832077111492321, 'samples': 3083264, 'steps': 6021, 'batch_loss/train': 0.8043414205312729} +12/23/2021 18:41:04 - INFO - codeparrot_training - Step 6022: {'lr': 0.0004832020309981592, 'samples': 3083776, 'steps': 6022, 'batch_loss/train': 0.9307437846437097} +12/23/2021 18:41:15 - INFO - codeparrot_training - Step 6023: {'lr': 0.0004831963499199631, 'samples': 3084288, 'steps': 6023, 'batch_loss/train': 0.5668099818285555} +12/23/2021 18:41:25 - INFO - codeparrot_training - Step 6024: {'lr': 0.00048319066791466627, 'samples': 3084800, 'steps': 6024, 'batch_loss/train': 0.8870122330263257} +12/23/2021 18:41:39 - INFO - codeparrot_training - Step 6025: {'lr': 0.0004831849849822915, 'samples': 3085312, 'steps': 6025, 'batch_loss/train': 0.8729191739112139} +12/23/2021 18:41:50 - INFO - codeparrot_training - Step 6026: {'lr': 0.00048317930112286116, 'samples': 3085824, 'steps': 6026, 'batch_loss/train': 0.7843036269769073} +12/23/2021 18:42:00 - INFO - codeparrot_training - Step 6027: {'lr': 0.00048317361633639803, 'samples': 3086336, 'steps': 6027, 'batch_loss/train': 0.9210655381903052} +12/23/2021 18:42:11 - INFO - codeparrot_training - Step 6028: {'lr': 0.00048316793062292465, 'samples': 3086848, 'steps': 6028, 'batch_loss/train': 0.8638545237481594} +12/23/2021 18:42:23 - INFO - codeparrot_training - Step 6029: {'lr': 0.00048316224398246366, 'samples': 3087360, 'steps': 6029, 'batch_loss/train': 0.8182992837391794} +12/23/2021 18:42:34 - INFO - codeparrot_training - Step 6030: {'lr': 0.0004831565564150375, 'samples': 3087872, 'steps': 6030, 'batch_loss/train': 0.7776999892666936} +12/23/2021 18:42:44 - INFO - codeparrot_training - Step 6031: {'lr': 0.00048315086792066904, 'samples': 3088384, 'steps': 6031, 'batch_loss/train': 0.9893842488527298} +12/23/2021 18:42:56 - INFO - codeparrot_training - Step 6032: {'lr': 0.00048314517849938073, 'samples': 3088896, 'steps': 6032, 'batch_loss/train': 1.040649883914739} +12/23/2021 18:43:07 - INFO - codeparrot_training - Step 6033: {'lr': 0.0004831394881511952, 'samples': 3089408, 'steps': 6033, 'batch_loss/train': 0.7766550513915718} +12/23/2021 18:43:17 - INFO - codeparrot_training - Step 6034: {'lr': 0.00048313379687613513, 'samples': 3089920, 'steps': 6034, 'batch_loss/train': 0.8496537888422608} +12/23/2021 18:43:30 - INFO - codeparrot_training - Step 6035: {'lr': 0.000483128104674223, 'samples': 3090432, 'steps': 6035, 'batch_loss/train': 0.8899904228746891} +12/23/2021 18:43:40 - INFO - codeparrot_training - Step 6036: {'lr': 0.00048312241154548167, 'samples': 3090944, 'steps': 6036, 'batch_loss/train': 0.9509349009022117} +12/23/2021 18:43:51 - INFO - codeparrot_training - Step 6037: {'lr': 0.00048311671748993364, 'samples': 3091456, 'steps': 6037, 'batch_loss/train': 0.8607864184305072} +12/23/2021 18:44:05 - INFO - codeparrot_training - Step 6038: {'lr': 0.0004831110225076015, 'samples': 3091968, 'steps': 6038, 'batch_loss/train': 0.811508153565228} +12/23/2021 18:44:15 - INFO - codeparrot_training - Step 6039: {'lr': 0.000483105326598508, 'samples': 3092480, 'steps': 6039, 'batch_loss/train': 1.3041536659002304} +12/23/2021 18:44:26 - INFO - codeparrot_training - Step 6040: {'lr': 0.00048309962976267573, 'samples': 3092992, 'steps': 6040, 'batch_loss/train': 0.8852015174925327} +12/23/2021 18:44:37 - INFO - codeparrot_training - Step 6041: {'lr': 0.00048309393200012733, 'samples': 3093504, 'steps': 6041, 'batch_loss/train': 0.8187704272568226} +12/23/2021 18:44:49 - INFO - codeparrot_training - Step 6042: {'lr': 0.00048308823331088554, 'samples': 3094016, 'steps': 6042, 'batch_loss/train': 0.7864492959342897} +12/23/2021 18:45:00 - INFO - codeparrot_training - Step 6043: {'lr': 0.0004830825336949729, 'samples': 3094528, 'steps': 6043, 'batch_loss/train': 0.7567867940524593} +12/23/2021 18:45:10 - INFO - codeparrot_training - Step 6044: {'lr': 0.00048307683315241206, 'samples': 3095040, 'steps': 6044, 'batch_loss/train': 0.5998000521212816} +12/23/2021 18:45:22 - INFO - codeparrot_training - Step 6045: {'lr': 0.0004830711316832258, 'samples': 3095552, 'steps': 6045, 'batch_loss/train': 0.8776691320817918} +12/23/2021 18:45:33 - INFO - codeparrot_training - Step 6046: {'lr': 0.0004830654292874367, 'samples': 3096064, 'steps': 6046, 'batch_loss/train': 0.7344814697280526} +12/23/2021 18:45:43 - INFO - codeparrot_training - Step 6047: {'lr': 0.00048305972596506745, 'samples': 3096576, 'steps': 6047, 'batch_loss/train': 0.8988330420106649} +12/23/2021 18:45:57 - INFO - codeparrot_training - Step 6048: {'lr': 0.0004830540217161408, 'samples': 3097088, 'steps': 6048, 'batch_loss/train': 0.9210238931700587} +12/23/2021 18:46:08 - INFO - codeparrot_training - Step 6049: {'lr': 0.0004830483165406793, 'samples': 3097600, 'steps': 6049, 'batch_loss/train': 0.9225013954564929} +12/23/2021 18:46:18 - INFO - codeparrot_training - Step 6050: {'lr': 0.0004830426104387057, 'samples': 3098112, 'steps': 6050, 'batch_loss/train': 0.8733643908053637} +12/23/2021 18:46:30 - INFO - codeparrot_training - Step 6051: {'lr': 0.00048303690341024267, 'samples': 3098624, 'steps': 6051, 'batch_loss/train': 0.8344813836738467} +12/23/2021 18:46:41 - INFO - codeparrot_training - Step 6052: {'lr': 0.0004830311954553129, 'samples': 3099136, 'steps': 6052, 'batch_loss/train': 0.8877637218683958} +12/23/2021 18:46:51 - INFO - codeparrot_training - Step 6053: {'lr': 0.00048302548657393906, 'samples': 3099648, 'steps': 6053, 'batch_loss/train': 0.8854377190582454} +12/23/2021 18:47:02 - INFO - codeparrot_training - Step 6054: {'lr': 0.00048301977676614387, 'samples': 3100160, 'steps': 6054, 'batch_loss/train': 0.8647752702236176} +12/23/2021 18:47:16 - INFO - codeparrot_training - Step 6055: {'lr': 0.00048301406603195007, 'samples': 3100672, 'steps': 6055, 'batch_loss/train': 0.8764647338539362} +12/23/2021 18:47:26 - INFO - codeparrot_training - Step 6056: {'lr': 0.0004830083543713804, 'samples': 3101184, 'steps': 6056, 'batch_loss/train': 0.7129918681457639} +12/23/2021 18:47:37 - INFO - codeparrot_training - Step 6057: {'lr': 0.00048300264178445735, 'samples': 3101696, 'steps': 6057, 'batch_loss/train': 0.8662923723459244} +12/23/2021 18:47:49 - INFO - codeparrot_training - Step 6058: {'lr': 0.00048299692827120385, 'samples': 3102208, 'steps': 6058, 'batch_loss/train': 0.791578525211662} +12/23/2021 18:47:59 - INFO - codeparrot_training - Step 6059: {'lr': 0.0004829912138316425, 'samples': 3102720, 'steps': 6059, 'batch_loss/train': 0.8699720883741975} +12/23/2021 18:48:10 - INFO - codeparrot_training - Step 6060: {'lr': 0.0004829854984657961, 'samples': 3103232, 'steps': 6060, 'batch_loss/train': 0.8421018337830901} +12/23/2021 18:48:22 - INFO - codeparrot_training - Step 6061: {'lr': 0.00048297978217368734, 'samples': 3103744, 'steps': 6061, 'batch_loss/train': 0.8643220970407128} +12/23/2021 18:48:32 - INFO - codeparrot_training - Step 6062: {'lr': 0.0004829740649553389, 'samples': 3104256, 'steps': 6062, 'batch_loss/train': 0.8845199160277843} +12/23/2021 18:48:43 - INFO - codeparrot_training - Step 6063: {'lr': 0.00048296834681077353, 'samples': 3104768, 'steps': 6063, 'batch_loss/train': 0.8419940415769815} +12/23/2021 18:48:54 - INFO - codeparrot_training - Step 6064: {'lr': 0.00048296262774001397, 'samples': 3105280, 'steps': 6064, 'batch_loss/train': 0.839435251429677} +12/23/2021 18:49:07 - INFO - codeparrot_training - Step 6065: {'lr': 0.000482956907743083, 'samples': 3105792, 'steps': 6065, 'batch_loss/train': 0.888203009031713} +12/23/2021 18:49:18 - INFO - codeparrot_training - Step 6066: {'lr': 0.0004829511868200034, 'samples': 3106304, 'steps': 6066, 'batch_loss/train': 0.8083533113822341} +12/23/2021 18:49:28 - INFO - codeparrot_training - Step 6067: {'lr': 0.00048294546497079774, 'samples': 3106816, 'steps': 6067, 'batch_loss/train': 0.7899560732766986} +12/23/2021 18:49:40 - INFO - codeparrot_training - Step 6068: {'lr': 0.00048293974219548887, 'samples': 3107328, 'steps': 6068, 'batch_loss/train': 0.7578352666459978} +12/23/2021 18:49:51 - INFO - codeparrot_training - Step 6069: {'lr': 0.00048293401849409956, 'samples': 3107840, 'steps': 6069, 'batch_loss/train': 0.8833907786756754} +12/23/2021 18:50:01 - INFO - codeparrot_training - Step 6070: {'lr': 0.00048292829386665256, 'samples': 3108352, 'steps': 6070, 'batch_loss/train': 0.9024988077580929} +12/23/2021 18:50:13 - INFO - codeparrot_training - Step 6071: {'lr': 0.0004829225683131706, 'samples': 3108864, 'steps': 6071, 'batch_loss/train': 0.8155947374179959} +12/23/2021 18:50:24 - INFO - codeparrot_training - Step 6072: {'lr': 0.00048291684183367645, 'samples': 3109376, 'steps': 6072, 'batch_loss/train': 0.9466386893764138} +12/23/2021 18:50:35 - INFO - codeparrot_training - Step 6073: {'lr': 0.0004829111144281929, 'samples': 3109888, 'steps': 6073, 'batch_loss/train': 0.9996325615793467} +12/23/2021 18:50:46 - INFO - codeparrot_training - Step 6074: {'lr': 0.0004829053860967427, 'samples': 3110400, 'steps': 6074, 'batch_loss/train': 0.8356164740398526} +12/23/2021 18:50:57 - INFO - codeparrot_training - Step 6075: {'lr': 0.00048289965683934855, 'samples': 3110912, 'steps': 6075, 'batch_loss/train': 0.8024642560631037} +12/23/2021 18:51:08 - INFO - codeparrot_training - Step 6076: {'lr': 0.0004828939266560334, 'samples': 3111424, 'steps': 6076, 'batch_loss/train': 0.8086444847285748} +12/23/2021 18:51:18 - INFO - codeparrot_training - Step 6077: {'lr': 0.00048288819554681995, 'samples': 3111936, 'steps': 6077, 'batch_loss/train': 0.7802216736599803} +12/23/2021 18:51:32 - INFO - codeparrot_training - Step 6078: {'lr': 0.0004828824635117309, 'samples': 3112448, 'steps': 6078, 'batch_loss/train': 0.8514965893700719} +12/23/2021 18:51:43 - INFO - codeparrot_training - Step 6079: {'lr': 0.00048287673055078916, 'samples': 3112960, 'steps': 6079, 'batch_loss/train': 0.8371637957170606} +12/23/2021 18:51:53 - INFO - codeparrot_training - Step 6080: {'lr': 0.0004828709966640175, 'samples': 3113472, 'steps': 6080, 'batch_loss/train': 0.8800743115134537} +12/23/2021 18:52:06 - INFO - codeparrot_training - Step 6081: {'lr': 0.00048286526185143856, 'samples': 3113984, 'steps': 6081, 'batch_loss/train': 0.8473503924906254} +12/23/2021 18:52:16 - INFO - codeparrot_training - Step 6082: {'lr': 0.00048285952611307534, 'samples': 3114496, 'steps': 6082, 'batch_loss/train': 0.757990549900569} +12/23/2021 18:52:27 - INFO - codeparrot_training - Step 6083: {'lr': 0.00048285378944895054, 'samples': 3115008, 'steps': 6083, 'batch_loss/train': 0.6985275216866285} +12/23/2021 18:52:41 - INFO - codeparrot_training - Step 6084: {'lr': 0.00048284805185908703, 'samples': 3115520, 'steps': 6084, 'batch_loss/train': 0.4915030673146248} +12/23/2021 18:52:51 - INFO - codeparrot_training - Step 6085: {'lr': 0.0004828423133435076, 'samples': 3116032, 'steps': 6085, 'batch_loss/train': 0.6555395472096279} +12/23/2021 18:53:02 - INFO - codeparrot_training - Step 6086: {'lr': 0.00048283657390223503, 'samples': 3116544, 'steps': 6086, 'batch_loss/train': 0.9301524758338928} +12/23/2021 18:53:13 - INFO - codeparrot_training - Step 6087: {'lr': 0.0004828308335352921, 'samples': 3117056, 'steps': 6087, 'batch_loss/train': 0.9040381808299571} +12/23/2021 18:53:25 - INFO - codeparrot_training - Step 6088: {'lr': 0.00048282509224270174, 'samples': 3117568, 'steps': 6088, 'batch_loss/train': 0.8505504108034074} +12/23/2021 18:53:36 - INFO - codeparrot_training - Step 6089: {'lr': 0.0004828193500244867, 'samples': 3118080, 'steps': 6089, 'batch_loss/train': 0.8892074674367905} +12/23/2021 18:53:47 - INFO - codeparrot_training - Step 6090: {'lr': 0.0004828136068806699, 'samples': 3118592, 'steps': 6090, 'batch_loss/train': 1.108687499538064} +12/23/2021 18:53:59 - INFO - codeparrot_training - Step 6091: {'lr': 0.0004828078628112741, 'samples': 3119104, 'steps': 6091, 'batch_loss/train': 0.8654631478711963} +12/23/2021 18:54:09 - INFO - codeparrot_training - Step 6092: {'lr': 0.00048280211781632213, 'samples': 3119616, 'steps': 6092, 'batch_loss/train': 0.8887252797139809} +12/23/2021 18:54:20 - INFO - codeparrot_training - Step 6093: {'lr': 0.0004827963718958368, 'samples': 3120128, 'steps': 6093, 'batch_loss/train': 0.8193308799527586} +12/23/2021 18:54:30 - INFO - codeparrot_training - Step 6094: {'lr': 0.00048279062504984107, 'samples': 3120640, 'steps': 6094, 'batch_loss/train': 0.8812017049640417} +12/23/2021 18:54:43 - INFO - codeparrot_training - Step 6095: {'lr': 0.00048278487727835764, 'samples': 3121152, 'steps': 6095, 'batch_loss/train': 1.2335653752088547} +12/23/2021 18:54:54 - INFO - codeparrot_training - Step 6096: {'lr': 0.00048277912858140943, 'samples': 3121664, 'steps': 6096, 'batch_loss/train': 0.8355279108509421} +12/23/2021 18:55:05 - INFO - codeparrot_training - Step 6097: {'lr': 0.0004827733789590193, 'samples': 3122176, 'steps': 6097, 'batch_loss/train': 0.8016488822177052} +12/23/2021 18:55:17 - INFO - codeparrot_training - Step 6098: {'lr': 0.00048276762841121016, 'samples': 3122688, 'steps': 6098, 'batch_loss/train': 0.8629789846017957} +12/23/2021 18:55:27 - INFO - codeparrot_training - Step 6099: {'lr': 0.00048276187693800475, 'samples': 3123200, 'steps': 6099, 'batch_loss/train': 0.9461807813495398} +12/23/2021 18:55:38 - INFO - codeparrot_training - Step 6100: {'lr': 0.00048275612453942606, 'samples': 3123712, 'steps': 6100, 'batch_loss/train': 0.9055876852944493} +12/23/2021 18:55:50 - INFO - codeparrot_training - Step 6101: {'lr': 0.0004827503712154969, 'samples': 3124224, 'steps': 6101, 'batch_loss/train': 0.8732046624645591} +12/23/2021 18:56:00 - INFO - codeparrot_training - Step 6102: {'lr': 0.00048274461696624015, 'samples': 3124736, 'steps': 6102, 'batch_loss/train': 0.9314009435474873} +12/23/2021 18:56:11 - INFO - codeparrot_training - Step 6103: {'lr': 0.0004827388617916787, 'samples': 3125248, 'steps': 6103, 'batch_loss/train': 0.9585136533714831} +12/23/2021 18:56:23 - INFO - codeparrot_training - Step 6104: {'lr': 0.0004827331056918354, 'samples': 3125760, 'steps': 6104, 'batch_loss/train': 0.8485535969957709} +12/23/2021 18:56:33 - INFO - codeparrot_training - Step 6105: {'lr': 0.0004827273486667331, 'samples': 3126272, 'steps': 6105, 'batch_loss/train': 0.8591582207009196} +12/23/2021 18:56:44 - INFO - codeparrot_training - Step 6106: {'lr': 0.0004827215907163948, 'samples': 3126784, 'steps': 6106, 'batch_loss/train': 0.9051511958241463} +12/23/2021 18:56:55 - INFO - codeparrot_training - Step 6107: {'lr': 0.0004827158318408432, 'samples': 3127296, 'steps': 6107, 'batch_loss/train': 0.8840127671137452} +12/23/2021 18:57:08 - INFO - codeparrot_training - Step 6108: {'lr': 0.0004827100720401015, 'samples': 3127808, 'steps': 6108, 'batch_loss/train': 0.8397765811532736} +12/23/2021 18:57:18 - INFO - codeparrot_training - Step 6109: {'lr': 0.00048270431131419234, 'samples': 3128320, 'steps': 6109, 'batch_loss/train': 0.7868944401852787} +12/23/2021 18:57:29 - INFO - codeparrot_training - Step 6110: {'lr': 0.0004826985496631386, 'samples': 3128832, 'steps': 6110, 'batch_loss/train': 0.8444827385246754} +12/23/2021 18:57:41 - INFO - codeparrot_training - Step 6111: {'lr': 0.00048269278708696354, 'samples': 3129344, 'steps': 6111, 'batch_loss/train': 0.8832928659394383} +12/23/2021 18:57:52 - INFO - codeparrot_training - Step 6112: {'lr': 0.0004826870235856896, 'samples': 3129856, 'steps': 6112, 'batch_loss/train': 0.8001764630898833} +12/23/2021 18:58:02 - INFO - codeparrot_training - Step 6113: {'lr': 0.00048268125915934003, 'samples': 3130368, 'steps': 6113, 'batch_loss/train': 0.9402580093592405} +12/23/2021 18:58:16 - INFO - codeparrot_training - Step 6114: {'lr': 0.00048267549380793753, 'samples': 3130880, 'steps': 6114, 'batch_loss/train': 0.8214766839519143} +12/23/2021 18:58:27 - INFO - codeparrot_training - Step 6115: {'lr': 0.00048266972753150517, 'samples': 3131392, 'steps': 6115, 'batch_loss/train': 0.950416692532599} +12/23/2021 18:58:37 - INFO - codeparrot_training - Step 6116: {'lr': 0.0004826639603300659, 'samples': 3131904, 'steps': 6116, 'batch_loss/train': 0.7947676749899983} +12/23/2021 18:58:48 - INFO - codeparrot_training - Step 6117: {'lr': 0.0004826581922036425, 'samples': 3132416, 'steps': 6117, 'batch_loss/train': 0.900772487744689} +12/23/2021 18:59:00 - INFO - codeparrot_training - Step 6118: {'lr': 0.000482652423152258, 'samples': 3132928, 'steps': 6118, 'batch_loss/train': 0.8240331513807178} +12/23/2021 18:59:11 - INFO - codeparrot_training - Step 6119: {'lr': 0.00048264665317593525, 'samples': 3133440, 'steps': 6119, 'batch_loss/train': 0.6882699513807893} +12/23/2021 18:59:21 - INFO - codeparrot_training - Step 6120: {'lr': 0.0004826408822746973, 'samples': 3133952, 'steps': 6120, 'batch_loss/train': 0.9271773518994451} +12/23/2021 18:59:33 - INFO - codeparrot_training - Step 6121: {'lr': 0.00048263511044856704, 'samples': 3134464, 'steps': 6121, 'batch_loss/train': 0.921166536398232} +12/23/2021 18:59:44 - INFO - codeparrot_training - Step 6122: {'lr': 0.00048262933769756744, 'samples': 3134976, 'steps': 6122, 'batch_loss/train': 0.8769117183983326} +12/23/2021 18:59:55 - INFO - codeparrot_training - Step 6123: {'lr': 0.0004826235640217214, 'samples': 3135488, 'steps': 6123, 'batch_loss/train': 1.2360077584162354} +12/23/2021 19:00:09 - INFO - codeparrot_training - Step 6124: {'lr': 0.00048261778942105193, 'samples': 3136000, 'steps': 6124, 'batch_loss/train': 0.9147136111278087} +12/23/2021 19:00:19 - INFO - codeparrot_training - Step 6125: {'lr': 0.00048261201389558194, 'samples': 3136512, 'steps': 6125, 'batch_loss/train': 0.6462871782714501} +12/23/2021 19:00:30 - INFO - codeparrot_training - Step 6126: {'lr': 0.00048260623744533445, 'samples': 3137024, 'steps': 6126, 'batch_loss/train': 0.891575706191361} +12/23/2021 19:00:42 - INFO - codeparrot_training - Step 6127: {'lr': 0.0004826004600703324, 'samples': 3137536, 'steps': 6127, 'batch_loss/train': 0.7820113841444254} +12/23/2021 19:00:52 - INFO - codeparrot_training - Step 6128: {'lr': 0.00048259468177059867, 'samples': 3138048, 'steps': 6128, 'batch_loss/train': 0.8805674090981483} +12/23/2021 19:01:03 - INFO - codeparrot_training - Step 6129: {'lr': 0.0004825889025461564, 'samples': 3138560, 'steps': 6129, 'batch_loss/train': 1.1016242085024714} +12/23/2021 19:01:14 - INFO - codeparrot_training - Step 6130: {'lr': 0.0004825831223970284, 'samples': 3139072, 'steps': 6130, 'batch_loss/train': 0.8293384676799178} +12/23/2021 19:01:27 - INFO - codeparrot_training - Step 6131: {'lr': 0.0004825773413232377, 'samples': 3139584, 'steps': 6131, 'batch_loss/train': 0.8244412890635431} +12/23/2021 19:01:37 - INFO - codeparrot_training - Step 6132: {'lr': 0.0004825715593248074, 'samples': 3140096, 'steps': 6132, 'batch_loss/train': 0.8453193414025009} +12/23/2021 19:01:48 - INFO - codeparrot_training - Step 6133: {'lr': 0.00048256577640176033, 'samples': 3140608, 'steps': 6133, 'batch_loss/train': 0.7297395975328982} +12/23/2021 19:02:00 - INFO - codeparrot_training - Step 6134: {'lr': 0.0004825599925541196, 'samples': 3141120, 'steps': 6134, 'batch_loss/train': 0.8003038046881557} +12/23/2021 19:02:11 - INFO - codeparrot_training - Step 6135: {'lr': 0.0004825542077819081, 'samples': 3141632, 'steps': 6135, 'batch_loss/train': 0.8453643801622093} +12/23/2021 19:02:21 - INFO - codeparrot_training - Step 6136: {'lr': 0.0004825484220851489, 'samples': 3142144, 'steps': 6136, 'batch_loss/train': 0.9247198291122913} +12/23/2021 19:02:32 - INFO - codeparrot_training - Step 6137: {'lr': 0.000482542635463865, 'samples': 3142656, 'steps': 6137, 'batch_loss/train': 0.693823928013444} +12/23/2021 19:02:44 - INFO - codeparrot_training - Step 6138: {'lr': 0.0004825368479180794, 'samples': 3143168, 'steps': 6138, 'batch_loss/train': 0.8098215004429221} +12/23/2021 19:02:55 - INFO - codeparrot_training - Step 6139: {'lr': 0.00048253105944781507, 'samples': 3143680, 'steps': 6139, 'batch_loss/train': 0.8460358530282974} +12/23/2021 19:03:05 - INFO - codeparrot_training - Step 6140: {'lr': 0.000482525270053095, 'samples': 3144192, 'steps': 6140, 'batch_loss/train': 0.7398361524101347} +12/23/2021 19:03:17 - INFO - codeparrot_training - Step 6141: {'lr': 0.0004825194797339423, 'samples': 3144704, 'steps': 6141, 'batch_loss/train': 0.8418246213113889} +12/23/2021 19:03:28 - INFO - codeparrot_training - Step 6142: {'lr': 0.00048251368849037995, 'samples': 3145216, 'steps': 6142, 'batch_loss/train': 0.7992138676345348} +12/23/2021 19:03:38 - INFO - codeparrot_training - Step 6143: {'lr': 0.00048250789632243096, 'samples': 3145728, 'steps': 6143, 'batch_loss/train': 0.8477866742759943} +12/23/2021 19:03:52 - INFO - codeparrot_training - Step 6144: {'lr': 0.00048250210323011835, 'samples': 3146240, 'steps': 6144, 'batch_loss/train': 0.8197430316358805} +12/23/2021 19:04:02 - INFO - codeparrot_training - Step 6145: {'lr': 0.0004824963092134652, 'samples': 3146752, 'steps': 6145, 'batch_loss/train': 0.8445977009832859} +12/23/2021 19:04:13 - INFO - codeparrot_training - Step 6146: {'lr': 0.00048249051427249446, 'samples': 3147264, 'steps': 6146, 'batch_loss/train': 0.8739229273051023} +12/23/2021 19:04:26 - INFO - codeparrot_training - Step 6147: {'lr': 0.0004824847184072293, 'samples': 3147776, 'steps': 6147, 'batch_loss/train': 0.9471801407635212} +12/23/2021 19:04:37 - INFO - codeparrot_training - Step 6148: {'lr': 0.00048247892161769264, 'samples': 3148288, 'steps': 6148, 'batch_loss/train': 1.0071319108828902} +12/23/2021 19:04:48 - INFO - codeparrot_training - Step 6149: {'lr': 0.0004824731239039075, 'samples': 3148800, 'steps': 6149, 'batch_loss/train': 0.8772745300084352} +12/23/2021 19:04:58 - INFO - codeparrot_training - Step 6150: {'lr': 0.0004824673252658971, 'samples': 3149312, 'steps': 6150, 'batch_loss/train': 0.8093972927890718} +12/23/2021 19:05:10 - INFO - codeparrot_training - Step 6151: {'lr': 0.0004824615257036844, 'samples': 3149824, 'steps': 6151, 'batch_loss/train': 0.8091893712989986} +12/23/2021 19:05:21 - INFO - codeparrot_training - Step 6152: {'lr': 0.0004824557252172924, 'samples': 3150336, 'steps': 6152, 'batch_loss/train': 0.8403333653695881} +12/23/2021 19:05:31 - INFO - codeparrot_training - Step 6153: {'lr': 0.0004824499238067442, 'samples': 3150848, 'steps': 6153, 'batch_loss/train': 0.8612623130902648} +12/23/2021 19:05:43 - INFO - codeparrot_training - Step 6154: {'lr': 0.00048244412147206283, 'samples': 3151360, 'steps': 6154, 'batch_loss/train': 0.8586272064130753} +12/23/2021 19:05:54 - INFO - codeparrot_training - Step 6155: {'lr': 0.0004824383182132715, 'samples': 3151872, 'steps': 6155, 'batch_loss/train': 0.7778337458148599} +12/23/2021 19:06:04 - INFO - codeparrot_training - Step 6156: {'lr': 0.0004824325140303931, 'samples': 3152384, 'steps': 6156, 'batch_loss/train': 0.8357659606263041} +12/23/2021 19:06:17 - INFO - codeparrot_training - Step 6157: {'lr': 0.00048242670892345086, 'samples': 3152896, 'steps': 6157, 'batch_loss/train': 0.8548043286427855} +12/23/2021 19:06:28 - INFO - codeparrot_training - Step 6158: {'lr': 0.0004824209028924678, 'samples': 3153408, 'steps': 6158, 'batch_loss/train': 0.910292711108923} +12/23/2021 19:06:38 - INFO - codeparrot_training - Step 6159: {'lr': 0.0004824150959374669, 'samples': 3153920, 'steps': 6159, 'batch_loss/train': 0.9178377948701382} +12/23/2021 19:06:49 - INFO - codeparrot_training - Step 6160: {'lr': 0.00048240928805847133, 'samples': 3154432, 'steps': 6160, 'batch_loss/train': 0.9256278686225414} +12/23/2021 19:07:01 - INFO - codeparrot_training - Step 6161: {'lr': 0.0004824034792555042, 'samples': 3154944, 'steps': 6161, 'batch_loss/train': 0.7469811677001417} +12/23/2021 19:07:12 - INFO - codeparrot_training - Step 6162: {'lr': 0.0004823976695285887, 'samples': 3155456, 'steps': 6162, 'batch_loss/train': 0.7979245013557374} +12/23/2021 19:07:22 - INFO - codeparrot_training - Step 6163: {'lr': 0.0004823918588777477, 'samples': 3155968, 'steps': 6163, 'batch_loss/train': 0.7927864962257445} +12/23/2021 19:07:35 - INFO - codeparrot_training - Step 6164: {'lr': 0.00048238604730300446, 'samples': 3156480, 'steps': 6164, 'batch_loss/train': 0.9374984344467521} +12/23/2021 19:07:46 - INFO - codeparrot_training - Step 6165: {'lr': 0.00048238023480438196, 'samples': 3156992, 'steps': 6165, 'batch_loss/train': 0.9147194372490048} +12/23/2021 19:07:56 - INFO - codeparrot_training - Step 6166: {'lr': 0.00048237442138190353, 'samples': 3157504, 'steps': 6166, 'batch_loss/train': 0.8917863848619163} +12/23/2021 19:08:09 - INFO - codeparrot_training - Step 6167: {'lr': 0.00048236860703559204, 'samples': 3158016, 'steps': 6167, 'batch_loss/train': 0.8740701312199235} +12/23/2021 19:08:19 - INFO - codeparrot_training - Step 6168: {'lr': 0.00048236279176547075, 'samples': 3158528, 'steps': 6168, 'batch_loss/train': 0.8546557240188122} +12/23/2021 19:08:30 - INFO - codeparrot_training - Step 6169: {'lr': 0.00048235697557156274, 'samples': 3159040, 'steps': 6169, 'batch_loss/train': 1.0232348684221506} +12/23/2021 19:08:42 - INFO - codeparrot_training - Step 6170: {'lr': 0.0004823511584538911, 'samples': 3159552, 'steps': 6170, 'batch_loss/train': 0.7492199530825019} +12/23/2021 19:08:53 - INFO - codeparrot_training - Step 6171: {'lr': 0.00048234534041247904, 'samples': 3160064, 'steps': 6171, 'batch_loss/train': 0.9214504533447325} +12/23/2021 19:09:03 - INFO - codeparrot_training - Step 6172: {'lr': 0.0004823395214473496, 'samples': 3160576, 'steps': 6172, 'batch_loss/train': 0.7399221435189247} +12/23/2021 19:09:14 - INFO - codeparrot_training - Step 6173: {'lr': 0.00048233370155852595, 'samples': 3161088, 'steps': 6173, 'batch_loss/train': 1.5138627802953124} +12/23/2021 19:09:28 - INFO - codeparrot_training - Step 6174: {'lr': 0.0004823278807460312, 'samples': 3161600, 'steps': 6174, 'batch_loss/train': 0.8278568810783327} +12/23/2021 19:09:38 - INFO - codeparrot_training - Step 6175: {'lr': 0.0004823220590098886, 'samples': 3162112, 'steps': 6175, 'batch_loss/train': 0.8829247537069023} +12/23/2021 19:09:49 - INFO - codeparrot_training - Step 6176: {'lr': 0.0004823162363501211, 'samples': 3162624, 'steps': 6176, 'batch_loss/train': 1.058266320032999} +12/23/2021 19:10:01 - INFO - codeparrot_training - Step 6177: {'lr': 0.00048231041276675205, 'samples': 3163136, 'steps': 6177, 'batch_loss/train': 0.708885476924479} +12/23/2021 19:10:11 - INFO - codeparrot_training - Step 6178: {'lr': 0.0004823045882598045, 'samples': 3163648, 'steps': 6178, 'batch_loss/train': 0.8257036246359348} +12/23/2021 19:10:22 - INFO - codeparrot_training - Step 6179: {'lr': 0.0004822987628293016, 'samples': 3164160, 'steps': 6179, 'batch_loss/train': 0.9502774062566459} +12/23/2021 19:10:35 - INFO - codeparrot_training - Step 6180: {'lr': 0.0004822929364752665, 'samples': 3164672, 'steps': 6180, 'batch_loss/train': 0.7205281648784876} +12/23/2021 19:10:46 - INFO - codeparrot_training - Step 6181: {'lr': 0.00048228710919772245, 'samples': 3165184, 'steps': 6181, 'batch_loss/train': 0.785894088447094} +12/23/2021 19:10:56 - INFO - codeparrot_training - Step 6182: {'lr': 0.00048228128099669253, 'samples': 3165696, 'steps': 6182, 'batch_loss/train': 0.9273154772818089} +12/23/2021 19:11:07 - INFO - codeparrot_training - Step 6183: {'lr': 0.00048227545187220003, 'samples': 3166208, 'steps': 6183, 'batch_loss/train': 0.848409584723413} +12/23/2021 19:11:19 - INFO - codeparrot_training - Step 6184: {'lr': 0.0004822696218242679, 'samples': 3166720, 'steps': 6184, 'batch_loss/train': 0.9004194997251034} +12/23/2021 19:11:29 - INFO - codeparrot_training - Step 6185: {'lr': 0.0004822637908529196, 'samples': 3167232, 'steps': 6185, 'batch_loss/train': 0.9522846639156342} +12/23/2021 19:11:40 - INFO - codeparrot_training - Step 6186: {'lr': 0.0004822579589581781, 'samples': 3167744, 'steps': 6186, 'batch_loss/train': 0.8746916735544801} +12/23/2021 19:11:52 - INFO - codeparrot_training - Step 6187: {'lr': 0.0004822521261400666, 'samples': 3168256, 'steps': 6187, 'batch_loss/train': 0.8010245035402477} +12/23/2021 19:12:03 - INFO - codeparrot_training - Step 6188: {'lr': 0.0004822462923986084, 'samples': 3168768, 'steps': 6188, 'batch_loss/train': 0.8500561127439141} +12/23/2021 19:12:13 - INFO - codeparrot_training - Step 6189: {'lr': 0.00048224045773382664, 'samples': 3169280, 'steps': 6189, 'batch_loss/train': 0.8987848963588476} +12/23/2021 19:12:25 - INFO - codeparrot_training - Step 6190: {'lr': 0.00048223462214574454, 'samples': 3169792, 'steps': 6190, 'batch_loss/train': 0.6886999923735857} +12/23/2021 19:12:36 - INFO - codeparrot_training - Step 6191: {'lr': 0.0004822287856343852, 'samples': 3170304, 'steps': 6191, 'batch_loss/train': 0.8146905489265919} +12/23/2021 19:12:46 - INFO - codeparrot_training - Step 6192: {'lr': 0.00048222294819977193, 'samples': 3170816, 'steps': 6192, 'batch_loss/train': 0.8736959975212812} +12/23/2021 19:13:00 - INFO - codeparrot_training - Step 6193: {'lr': 0.0004822171098419279, 'samples': 3171328, 'steps': 6193, 'batch_loss/train': 0.8078755624592304} +12/23/2021 19:13:10 - INFO - codeparrot_training - Step 6194: {'lr': 0.00048221127056087624, 'samples': 3171840, 'steps': 6194, 'batch_loss/train': 0.9787458833307028} +12/23/2021 19:13:21 - INFO - codeparrot_training - Step 6195: {'lr': 0.00048220543035664033, 'samples': 3172352, 'steps': 6195, 'batch_loss/train': 0.9517631372436881} +12/23/2021 19:13:31 - INFO - codeparrot_training - Step 6196: {'lr': 0.00048219958922924336, 'samples': 3172864, 'steps': 6196, 'batch_loss/train': 0.8827500715851784} +12/23/2021 19:13:43 - INFO - codeparrot_training - Step 6197: {'lr': 0.0004821937471787084, 'samples': 3173376, 'steps': 6197, 'batch_loss/train': 0.8455467880703509} +12/23/2021 19:13:54 - INFO - codeparrot_training - Step 6198: {'lr': 0.00048218790420505883, 'samples': 3173888, 'steps': 6198, 'batch_loss/train': 0.845529867336154} +12/23/2021 19:14:05 - INFO - codeparrot_training - Step 6199: {'lr': 0.0004821820603083178, 'samples': 3174400, 'steps': 6199, 'batch_loss/train': 0.9171948381699622} +12/23/2021 19:14:17 - INFO - codeparrot_training - Step 6200: {'lr': 0.00048217621548850856, 'samples': 3174912, 'steps': 6200, 'batch_loss/train': 0.8182698942255229} +12/23/2021 19:14:27 - INFO - codeparrot_training - Step 6201: {'lr': 0.00048217036974565437, 'samples': 3175424, 'steps': 6201, 'batch_loss/train': 0.7989611849188805} +12/23/2021 19:14:38 - INFO - codeparrot_training - Step 6202: {'lr': 0.0004821645230797784, 'samples': 3175936, 'steps': 6202, 'batch_loss/train': 0.76821602974087} +12/23/2021 19:14:51 - INFO - codeparrot_training - Step 6203: {'lr': 0.00048215867549090404, 'samples': 3176448, 'steps': 6203, 'batch_loss/train': 0.9126338651403785} +12/23/2021 19:15:01 - INFO - codeparrot_training - Step 6204: {'lr': 0.00048215282697905437, 'samples': 3176960, 'steps': 6204, 'batch_loss/train': 0.8932211957871914} +12/23/2021 19:15:12 - INFO - codeparrot_training - Step 6205: {'lr': 0.00048214697754425277, 'samples': 3177472, 'steps': 6205, 'batch_loss/train': 0.801686423830688} +12/23/2021 19:15:23 - INFO - codeparrot_training - Step 6206: {'lr': 0.0004821411271865225, 'samples': 3177984, 'steps': 6206, 'batch_loss/train': 0.8449810985475779} +12/23/2021 19:15:35 - INFO - codeparrot_training - Step 6207: {'lr': 0.00048213527590588664, 'samples': 3178496, 'steps': 6207, 'batch_loss/train': 0.8430665256455541} +12/23/2021 19:15:45 - INFO - codeparrot_training - Step 6208: {'lr': 0.0004821294237023686, 'samples': 3179008, 'steps': 6208, 'batch_loss/train': 0.7703283014707267} +12/23/2021 19:15:56 - INFO - codeparrot_training - Step 6209: {'lr': 0.00048212357057599166, 'samples': 3179520, 'steps': 6209, 'batch_loss/train': 0.8773196246474981} +12/23/2021 19:16:09 - INFO - codeparrot_training - Step 6210: {'lr': 0.00048211771652677905, 'samples': 3180032, 'steps': 6210, 'batch_loss/train': 0.7646572799421847} +12/23/2021 19:16:20 - INFO - codeparrot_training - Step 6211: {'lr': 0.00048211186155475394, 'samples': 3180544, 'steps': 6211, 'batch_loss/train': 0.7790250760735944} +12/23/2021 19:16:30 - INFO - codeparrot_training - Step 6212: {'lr': 0.0004821060056599398, 'samples': 3181056, 'steps': 6212, 'batch_loss/train': 0.8670127233490348} +12/23/2021 19:16:43 - INFO - codeparrot_training - Step 6213: {'lr': 0.0004821001488423599, 'samples': 3181568, 'steps': 6213, 'batch_loss/train': 0.8039958495646715} +12/23/2021 19:16:53 - INFO - codeparrot_training - Step 6214: {'lr': 0.0004820942911020373, 'samples': 3182080, 'steps': 6214, 'batch_loss/train': 0.8576817661523819} +12/23/2021 19:17:04 - INFO - codeparrot_training - Step 6215: {'lr': 0.00048208843243899546, 'samples': 3182592, 'steps': 6215, 'batch_loss/train': 0.8986051958054304} +12/23/2021 19:17:14 - INFO - codeparrot_training - Step 6216: {'lr': 0.0004820825728532577, 'samples': 3183104, 'steps': 6216, 'batch_loss/train': 0.6305037504062057} +12/23/2021 19:17:26 - INFO - codeparrot_training - Step 6217: {'lr': 0.00048207671234484724, 'samples': 3183616, 'steps': 6217, 'batch_loss/train': 0.8251875410787761} +12/23/2021 19:17:37 - INFO - codeparrot_training - Step 6218: {'lr': 0.0004820708509137874, 'samples': 3184128, 'steps': 6218, 'batch_loss/train': 0.8745332164689898} +12/23/2021 19:17:48 - INFO - codeparrot_training - Step 6219: {'lr': 0.0004820649885601015, 'samples': 3184640, 'steps': 6219, 'batch_loss/train': 0.7438611974939704} +12/23/2021 19:18:00 - INFO - codeparrot_training - Step 6220: {'lr': 0.0004820591252838128, 'samples': 3185152, 'steps': 6220, 'batch_loss/train': 0.8181492444127798} +12/23/2021 19:18:10 - INFO - codeparrot_training - Step 6221: {'lr': 0.00048205326108494465, 'samples': 3185664, 'steps': 6221, 'batch_loss/train': 0.7528134528547525} +12/23/2021 19:18:21 - INFO - codeparrot_training - Step 6222: {'lr': 0.00048204739596352037, 'samples': 3186176, 'steps': 6222, 'batch_loss/train': 0.9202347565442324} +12/23/2021 19:18:33 - INFO - codeparrot_training - Step 6223: {'lr': 0.00048204152991956326, 'samples': 3186688, 'steps': 6223, 'batch_loss/train': 0.7890992667526007} +12/23/2021 19:18:43 - INFO - codeparrot_training - Step 6224: {'lr': 0.0004820356629530966, 'samples': 3187200, 'steps': 6224, 'batch_loss/train': 0.8971185488626361} +12/23/2021 19:18:54 - INFO - codeparrot_training - Step 6225: {'lr': 0.0004820297950641438, 'samples': 3187712, 'steps': 6225, 'batch_loss/train': 0.8682784158736467} +12/23/2021 19:19:07 - INFO - codeparrot_training - Step 6226: {'lr': 0.00048202392625272815, 'samples': 3188224, 'steps': 6226, 'batch_loss/train': 0.9077216796576977} +12/23/2021 19:19:18 - INFO - codeparrot_training - Step 6227: {'lr': 0.000482018056518873, 'samples': 3188736, 'steps': 6227, 'batch_loss/train': 1.0404776018112898} +12/23/2021 19:19:28 - INFO - codeparrot_training - Step 6228: {'lr': 0.0004820121858626016, 'samples': 3189248, 'steps': 6228, 'batch_loss/train': 0.7974925003945827} +12/23/2021 19:19:39 - INFO - codeparrot_training - Step 6229: {'lr': 0.00048200631428393735, 'samples': 3189760, 'steps': 6229, 'batch_loss/train': 0.8182296128943563} +12/23/2021 19:19:51 - INFO - codeparrot_training - Step 6230: {'lr': 0.00048200044178290367, 'samples': 3190272, 'steps': 6230, 'batch_loss/train': 0.8441724951844662} +12/23/2021 19:20:01 - INFO - codeparrot_training - Step 6231: {'lr': 0.00048199456835952374, 'samples': 3190784, 'steps': 6231, 'batch_loss/train': 0.8412117380648851} +12/23/2021 19:20:12 - INFO - codeparrot_training - Step 6232: {'lr': 0.00048198869401382106, 'samples': 3191296, 'steps': 6232, 'batch_loss/train': 0.8549791490659118} +12/23/2021 19:20:25 - INFO - codeparrot_training - Step 6233: {'lr': 0.00048198281874581897, 'samples': 3191808, 'steps': 6233, 'batch_loss/train': 0.8353495132178068} +12/23/2021 19:20:36 - INFO - codeparrot_training - Step 6234: {'lr': 0.0004819769425555407, 'samples': 3192320, 'steps': 6234, 'batch_loss/train': 0.8882966991513968} +12/23/2021 19:20:46 - INFO - codeparrot_training - Step 6235: {'lr': 0.0004819710654430097, 'samples': 3192832, 'steps': 6235, 'batch_loss/train': 0.8255678373388946} +12/23/2021 19:20:59 - INFO - codeparrot_training - Step 6236: {'lr': 0.0004819651874082494, 'samples': 3193344, 'steps': 6236, 'batch_loss/train': 0.7989011937752366} +12/23/2021 19:21:09 - INFO - codeparrot_training - Step 6237: {'lr': 0.0004819593084512831, 'samples': 3193856, 'steps': 6237, 'batch_loss/train': 0.7307513374835253} +12/23/2021 19:21:20 - INFO - codeparrot_training - Step 6238: {'lr': 0.0004819534285721341, 'samples': 3194368, 'steps': 6238, 'batch_loss/train': 0.8647325607016683} +12/23/2021 19:21:30 - INFO - codeparrot_training - Step 6239: {'lr': 0.00048194754777082597, 'samples': 3194880, 'steps': 6239, 'batch_loss/train': 1.0070551447570324} +12/23/2021 19:21:42 - INFO - codeparrot_training - Step 6240: {'lr': 0.00048194166604738185, 'samples': 3195392, 'steps': 6240, 'batch_loss/train': 0.9316873487550765} +12/23/2021 19:21:53 - INFO - codeparrot_training - Step 6241: {'lr': 0.0004819357834018253, 'samples': 3195904, 'steps': 6241, 'batch_loss/train': 0.8423882224597037} +12/23/2021 19:22:04 - INFO - codeparrot_training - Step 6242: {'lr': 0.00048192989983417963, 'samples': 3196416, 'steps': 6242, 'batch_loss/train': 0.8540492197498679} +12/23/2021 19:22:17 - INFO - codeparrot_training - Step 6243: {'lr': 0.00048192401534446827, 'samples': 3196928, 'steps': 6243, 'batch_loss/train': 0.8456971254199743} +12/23/2021 19:22:28 - INFO - codeparrot_training - Step 6244: {'lr': 0.00048191812993271456, 'samples': 3197440, 'steps': 6244, 'batch_loss/train': 0.842575398273766} +12/23/2021 19:22:38 - INFO - codeparrot_training - Step 6245: {'lr': 0.00048191224359894194, 'samples': 3197952, 'steps': 6245, 'batch_loss/train': 0.906162385828793} +12/23/2021 19:22:51 - INFO - codeparrot_training - Step 6246: {'lr': 0.0004819063563431738, 'samples': 3198464, 'steps': 6246, 'batch_loss/train': 0.7923900242894888} +12/23/2021 19:23:01 - INFO - codeparrot_training - Step 6247: {'lr': 0.00048190046816543367, 'samples': 3198976, 'steps': 6247, 'batch_loss/train': 0.8844696078449488} +12/23/2021 19:23:12 - INFO - codeparrot_training - Step 6248: {'lr': 0.00048189457906574475, 'samples': 3199488, 'steps': 6248, 'batch_loss/train': 0.4925309788668528} +12/23/2021 19:23:25 - INFO - codeparrot_training - Step 6249: {'lr': 0.00048188868904413054, 'samples': 3200000, 'steps': 6249, 'batch_loss/train': 0.8247741032391787} +12/23/2021 19:23:36 - INFO - codeparrot_training - Step 6250: {'lr': 0.0004818827981006144, 'samples': 3200512, 'steps': 6250, 'batch_loss/train': 0.9985879638697952} +12/23/2021 19:23:46 - INFO - codeparrot_training - Step 6251: {'lr': 0.0004818769062352199, 'samples': 3201024, 'steps': 6251, 'batch_loss/train': 2.35749838873744} +12/23/2021 19:23:57 - INFO - codeparrot_training - Step 6252: {'lr': 0.00048187101344797037, 'samples': 3201536, 'steps': 6252, 'batch_loss/train': 0.8339326102286577} +12/23/2021 19:24:09 - INFO - codeparrot_training - Step 6253: {'lr': 0.0004818651197388892, 'samples': 3202048, 'steps': 6253, 'batch_loss/train': 0.922753281891346} +12/23/2021 19:24:20 - INFO - codeparrot_training - Step 6254: {'lr': 0.00048185922510799985, 'samples': 3202560, 'steps': 6254, 'batch_loss/train': 0.8321730103343725} +12/23/2021 19:24:30 - INFO - codeparrot_training - Step 6255: {'lr': 0.0004818533295553258, 'samples': 3203072, 'steps': 6255, 'batch_loss/train': 0.7096857642754912} +12/23/2021 19:24:42 - INFO - codeparrot_training - Step 6256: {'lr': 0.00048184743308089047, 'samples': 3203584, 'steps': 6256, 'batch_loss/train': 0.8968213042244315} +12/23/2021 19:24:53 - INFO - codeparrot_training - Step 6257: {'lr': 0.0004818415356847173, 'samples': 3204096, 'steps': 6257, 'batch_loss/train': 1.0182666124310344} +12/23/2021 19:25:03 - INFO - codeparrot_training - Step 6258: {'lr': 0.0004818356373668297, 'samples': 3204608, 'steps': 6258, 'batch_loss/train': 0.8146989271044731} +12/23/2021 19:25:14 - INFO - codeparrot_training - Step 6259: {'lr': 0.00048182973812725113, 'samples': 3205120, 'steps': 6259, 'batch_loss/train': 1.046090287156403} +12/23/2021 19:25:26 - INFO - codeparrot_training - Step 6260: {'lr': 0.00048182383796600505, 'samples': 3205632, 'steps': 6260, 'batch_loss/train': 0.8464795826002955} +12/23/2021 19:25:37 - INFO - codeparrot_training - Step 6261: {'lr': 0.0004818179368831149, 'samples': 3206144, 'steps': 6261, 'batch_loss/train': 0.8145375489257276} +12/23/2021 19:25:47 - INFO - codeparrot_training - Step 6262: {'lr': 0.00048181203487860426, 'samples': 3206656, 'steps': 6262, 'batch_loss/train': 0.8262369148433208} +12/23/2021 19:26:01 - INFO - codeparrot_training - Step 6263: {'lr': 0.00048180613195249644, 'samples': 3207168, 'steps': 6263, 'batch_loss/train': 0.8696820484474301} +12/23/2021 19:26:11 - INFO - codeparrot_training - Step 6264: {'lr': 0.00048180022810481493, 'samples': 3207680, 'steps': 6264, 'batch_loss/train': 0.7509583665523678} +12/23/2021 19:26:22 - INFO - codeparrot_training - Step 6265: {'lr': 0.0004817943233355833, 'samples': 3208192, 'steps': 6265, 'batch_loss/train': 0.894938675686717} +12/23/2021 19:26:34 - INFO - codeparrot_training - Step 6266: {'lr': 0.00048178841764482495, 'samples': 3208704, 'steps': 6266, 'batch_loss/train': 0.7860732432454824} +12/23/2021 19:26:44 - INFO - codeparrot_training - Step 6267: {'lr': 0.0004817825110325633, 'samples': 3209216, 'steps': 6267, 'batch_loss/train': 0.9254091447219253} +12/23/2021 19:26:55 - INFO - codeparrot_training - Step 6268: {'lr': 0.00048177660349882205, 'samples': 3209728, 'steps': 6268, 'batch_loss/train': 0.854687936604023} +12/23/2021 19:27:07 - INFO - codeparrot_training - Step 6269: {'lr': 0.0004817706950436244, 'samples': 3210240, 'steps': 6269, 'batch_loss/train': 0.898751325905323} +12/23/2021 19:27:18 - INFO - codeparrot_training - Step 6270: {'lr': 0.00048176478566699406, 'samples': 3210752, 'steps': 6270, 'batch_loss/train': 0.8186372700147331} +12/23/2021 19:27:28 - INFO - codeparrot_training - Step 6271: {'lr': 0.0004817588753689544, 'samples': 3211264, 'steps': 6271, 'batch_loss/train': 0.7120343195274472} +12/23/2021 19:27:39 - INFO - codeparrot_training - Step 6272: {'lr': 0.00048175296414952894, 'samples': 3211776, 'steps': 6272, 'batch_loss/train': 0.7893921909853816} +12/23/2021 19:27:52 - INFO - codeparrot_training - Step 6273: {'lr': 0.0004817470520087413, 'samples': 3212288, 'steps': 6273, 'batch_loss/train': 0.8019486330449581} +12/23/2021 19:28:03 - INFO - codeparrot_training - Step 6274: {'lr': 0.0004817411389466148, 'samples': 3212800, 'steps': 6274, 'batch_loss/train': 0.8419052688404918} +12/23/2021 19:28:13 - INFO - codeparrot_training - Step 6275: {'lr': 0.00048173522496317304, 'samples': 3213312, 'steps': 6275, 'batch_loss/train': 0.8861635765060782} +12/23/2021 19:28:25 - INFO - codeparrot_training - Step 6276: {'lr': 0.00048172931005843956, 'samples': 3213824, 'steps': 6276, 'batch_loss/train': 0.865821173414588} +12/23/2021 19:28:36 - INFO - codeparrot_training - Step 6277: {'lr': 0.00048172339423243786, 'samples': 3214336, 'steps': 6277, 'batch_loss/train': 0.7708904966711998} +12/23/2021 19:28:47 - INFO - codeparrot_training - Step 6278: {'lr': 0.00048171747748519136, 'samples': 3214848, 'steps': 6278, 'batch_loss/train': 0.8452870985493064} +12/23/2021 19:29:00 - INFO - codeparrot_training - Step 6279: {'lr': 0.00048171155981672375, 'samples': 3215360, 'steps': 6279, 'batch_loss/train': 0.9331196639686823} +12/23/2021 19:29:10 - INFO - codeparrot_training - Step 6280: {'lr': 0.0004817056412270584, 'samples': 3215872, 'steps': 6280, 'batch_loss/train': 0.8746827887371182} +12/23/2021 19:29:21 - INFO - codeparrot_training - Step 6281: {'lr': 0.00048169972171621894, 'samples': 3216384, 'steps': 6281, 'batch_loss/train': 0.9949339097365737} +12/23/2021 19:29:32 - INFO - codeparrot_training - Step 6282: {'lr': 0.0004816938012842289, 'samples': 3216896, 'steps': 6282, 'batch_loss/train': 0.8937361128628254} +12/23/2021 19:29:44 - INFO - codeparrot_training - Step 6283: {'lr': 0.00048168787993111176, 'samples': 3217408, 'steps': 6283, 'batch_loss/train': 0.8732508295215666} +12/23/2021 19:29:54 - INFO - codeparrot_training - Step 6284: {'lr': 0.00048168195765689104, 'samples': 3217920, 'steps': 6284, 'batch_loss/train': 0.8593590329401195} +12/23/2021 19:30:05 - INFO - codeparrot_training - Step 6285: {'lr': 0.0004816760344615904, 'samples': 3218432, 'steps': 6285, 'batch_loss/train': 1.6732439696788788} +12/23/2021 19:30:17 - INFO - codeparrot_training - Step 6286: {'lr': 0.00048167011034523335, 'samples': 3218944, 'steps': 6286, 'batch_loss/train': 0.8306072591803968} +12/23/2021 19:30:28 - INFO - codeparrot_training - Step 6287: {'lr': 0.0004816641853078433, 'samples': 3219456, 'steps': 6287, 'batch_loss/train': 1.2401721561327577} +12/23/2021 19:30:39 - INFO - codeparrot_training - Step 6288: {'lr': 0.0004816582593494441, 'samples': 3219968, 'steps': 6288, 'batch_loss/train': 0.7845822980161756} +12/23/2021 19:30:53 - INFO - codeparrot_training - Step 6289: {'lr': 0.000481652332470059, 'samples': 3220480, 'steps': 6289, 'batch_loss/train': 0.8633116045966744} +12/23/2021 19:31:03 - INFO - codeparrot_training - Step 6290: {'lr': 0.00048164640466971166, 'samples': 3220992, 'steps': 6290, 'batch_loss/train': 0.9027934828773141} +12/23/2021 19:31:14 - INFO - codeparrot_training - Step 6291: {'lr': 0.0004816404759484258, 'samples': 3221504, 'steps': 6291, 'batch_loss/train': 0.7266103513538837} +12/23/2021 19:31:24 - INFO - codeparrot_training - Step 6292: {'lr': 0.0004816345463062248, 'samples': 3222016, 'steps': 6292, 'batch_loss/train': 0.8378871174063534} +12/23/2021 19:31:37 - INFO - codeparrot_training - Step 6293: {'lr': 0.0004816286157431323, 'samples': 3222528, 'steps': 6293, 'batch_loss/train': 0.8284159526228905} +12/23/2021 19:31:47 - INFO - codeparrot_training - Step 6294: {'lr': 0.00048162268425917197, 'samples': 3223040, 'steps': 6294, 'batch_loss/train': 0.9377811551094055} +12/23/2021 19:31:58 - INFO - codeparrot_training - Step 6295: {'lr': 0.0004816167518543672, 'samples': 3223552, 'steps': 6295, 'batch_loss/train': 0.7822399965953082} +12/23/2021 19:32:10 - INFO - codeparrot_training - Step 6296: {'lr': 0.00048161081852874174, 'samples': 3224064, 'steps': 6296, 'batch_loss/train': 1.0336824841797352} +12/23/2021 19:32:20 - INFO - codeparrot_training - Step 6297: {'lr': 0.0004816048842823191, 'samples': 3224576, 'steps': 6297, 'batch_loss/train': 0.8062356067821383} +12/23/2021 19:32:31 - INFO - codeparrot_training - Step 6298: {'lr': 0.0004815989491151229, 'samples': 3225088, 'steps': 6298, 'batch_loss/train': 0.8574892738834023} +12/23/2021 19:32:44 - INFO - codeparrot_training - Step 6299: {'lr': 0.00048159301302717666, 'samples': 3225600, 'steps': 6299, 'batch_loss/train': 0.8124414128251374} +12/23/2021 19:32:54 - INFO - codeparrot_training - Step 6300: {'lr': 0.00048158707601850416, 'samples': 3226112, 'steps': 6300, 'batch_loss/train': 0.8690347336232662} +12/23/2021 19:33:05 - INFO - codeparrot_training - Step 6301: {'lr': 0.00048158113808912886, 'samples': 3226624, 'steps': 6301, 'batch_loss/train': 0.9247024366632104} +12/23/2021 19:33:17 - INFO - codeparrot_training - Step 6302: {'lr': 0.00048157519923907433, 'samples': 3227136, 'steps': 6302, 'batch_loss/train': 0.881863534450531} +12/23/2021 19:33:28 - INFO - codeparrot_training - Step 6303: {'lr': 0.00048156925946836427, 'samples': 3227648, 'steps': 6303, 'batch_loss/train': 0.8179659713059664} +12/23/2021 19:33:38 - INFO - codeparrot_training - Step 6304: {'lr': 0.0004815633187770223, 'samples': 3228160, 'steps': 6304, 'batch_loss/train': 0.9069101735949516} +12/23/2021 19:33:49 - INFO - codeparrot_training - Step 6305: {'lr': 0.000481557377165072, 'samples': 3228672, 'steps': 6305, 'batch_loss/train': 0.8446366367861629} +12/23/2021 19:34:01 - INFO - codeparrot_training - Step 6306: {'lr': 0.000481551434632537, 'samples': 3229184, 'steps': 6306, 'batch_loss/train': 0.8532740147784352} +12/23/2021 19:34:11 - INFO - codeparrot_training - Step 6307: {'lr': 0.0004815454911794409, 'samples': 3229696, 'steps': 6307, 'batch_loss/train': 0.9394666934385896} +12/23/2021 19:34:22 - INFO - codeparrot_training - Step 6308: {'lr': 0.00048153954680580735, 'samples': 3230208, 'steps': 6308, 'batch_loss/train': 0.8495687246322632} +12/23/2021 19:34:35 - INFO - codeparrot_training - Step 6309: {'lr': 0.00048153360151166006, 'samples': 3230720, 'steps': 6309, 'batch_loss/train': 0.8641701713204384} +12/23/2021 19:34:46 - INFO - codeparrot_training - Step 6310: {'lr': 0.0004815276552970226, 'samples': 3231232, 'steps': 6310, 'batch_loss/train': 0.7383868074975908} +12/23/2021 19:34:56 - INFO - codeparrot_training - Step 6311: {'lr': 0.00048152170816191854, 'samples': 3231744, 'steps': 6311, 'batch_loss/train': 0.8469790294766426} +12/23/2021 19:35:08 - INFO - codeparrot_training - Step 6312: {'lr': 0.0004815157601063715, 'samples': 3232256, 'steps': 6312, 'batch_loss/train': 0.7395428996533155} +12/23/2021 19:35:19 - INFO - codeparrot_training - Step 6313: {'lr': 0.0004815098111304054, 'samples': 3232768, 'steps': 6313, 'batch_loss/train': 0.9501498807221651} +12/23/2021 19:35:30 - INFO - codeparrot_training - Step 6314: {'lr': 0.0004815038612340436, 'samples': 3233280, 'steps': 6314, 'batch_loss/train': 0.7551151113584638} +12/23/2021 19:35:40 - INFO - codeparrot_training - Step 6315: {'lr': 0.0004814979104173099, 'samples': 3233792, 'steps': 6315, 'batch_loss/train': 0.7866106517612934} +12/23/2021 19:35:52 - INFO - codeparrot_training - Step 6316: {'lr': 0.0004814919586802279, 'samples': 3234304, 'steps': 6316, 'batch_loss/train': 0.8321128427051008} +12/23/2021 19:36:03 - INFO - codeparrot_training - Step 6317: {'lr': 0.0004814860060228213, 'samples': 3234816, 'steps': 6317, 'batch_loss/train': 0.7506366181187332} +12/23/2021 19:36:14 - INFO - codeparrot_training - Step 6318: {'lr': 0.00048148005244511374, 'samples': 3235328, 'steps': 6318, 'batch_loss/train': 0.8028223887085915} +12/23/2021 19:36:27 - INFO - codeparrot_training - Step 6319: {'lr': 0.00048147409794712883, 'samples': 3235840, 'steps': 6319, 'batch_loss/train': 0.9844986703246832} +12/23/2021 19:36:38 - INFO - codeparrot_training - Step 6320: {'lr': 0.00048146814252889036, 'samples': 3236352, 'steps': 6320, 'batch_loss/train': 0.8473936170339584} +12/23/2021 19:36:49 - INFO - codeparrot_training - Step 6321: {'lr': 0.00048146218619042196, 'samples': 3236864, 'steps': 6321, 'batch_loss/train': 0.9189420435577631} +12/23/2021 19:37:01 - INFO - codeparrot_training - Step 6322: {'lr': 0.0004814562289317473, 'samples': 3237376, 'steps': 6322, 'batch_loss/train': 0.8597549072001129} +12/23/2021 19:37:11 - INFO - codeparrot_training - Step 6323: {'lr': 0.0004814502707528901, 'samples': 3237888, 'steps': 6323, 'batch_loss/train': 0.6611430912744254} +12/23/2021 19:37:22 - INFO - codeparrot_training - Step 6324: {'lr': 0.00048144431165387397, 'samples': 3238400, 'steps': 6324, 'batch_loss/train': 0.8822643496096134} +12/23/2021 19:37:34 - INFO - codeparrot_training - Step 6325: {'lr': 0.00048143835163472264, 'samples': 3238912, 'steps': 6325, 'batch_loss/train': 0.9055756200104952} +12/23/2021 19:37:45 - INFO - codeparrot_training - Step 6326: {'lr': 0.0004814323906954599, 'samples': 3239424, 'steps': 6326, 'batch_loss/train': 0.7927223173901439} +12/23/2021 19:37:55 - INFO - codeparrot_training - Step 6327: {'lr': 0.0004814264288361093, 'samples': 3239936, 'steps': 6327, 'batch_loss/train': 0.777548614423722} +12/23/2021 19:38:06 - INFO - codeparrot_training - Step 6328: {'lr': 0.0004814204660566945, 'samples': 3240448, 'steps': 6328, 'batch_loss/train': 0.8562621530145407} +12/23/2021 19:38:19 - INFO - codeparrot_training - Step 6329: {'lr': 0.00048141450235723946, 'samples': 3240960, 'steps': 6329, 'batch_loss/train': 0.7070676367729902} +12/23/2021 19:38:30 - INFO - codeparrot_training - Step 6330: {'lr': 0.0004814085377377677, 'samples': 3241472, 'steps': 6330, 'batch_loss/train': 0.8096684413030744} +12/23/2021 19:38:40 - INFO - codeparrot_training - Step 6331: {'lr': 0.0004814025721983029, 'samples': 3241984, 'steps': 6331, 'batch_loss/train': 0.9125736569985747} +12/23/2021 19:38:53 - INFO - codeparrot_training - Step 6332: {'lr': 0.0004813966057388689, 'samples': 3242496, 'steps': 6332, 'batch_loss/train': 0.8385059824213386} +12/23/2021 19:39:03 - INFO - codeparrot_training - Step 6333: {'lr': 0.0004813906383594894, 'samples': 3243008, 'steps': 6333, 'batch_loss/train': 0.8188949963077903} +12/23/2021 19:39:14 - INFO - codeparrot_training - Step 6334: {'lr': 0.00048138467006018806, 'samples': 3243520, 'steps': 6334, 'batch_loss/train': 0.7544340868480504} +12/23/2021 19:39:27 - INFO - codeparrot_training - Step 6335: {'lr': 0.0004813787008409886, 'samples': 3244032, 'steps': 6335, 'batch_loss/train': 0.8732138620689511} +12/23/2021 19:39:37 - INFO - codeparrot_training - Step 6336: {'lr': 0.00048137273070191487, 'samples': 3244544, 'steps': 6336, 'batch_loss/train': 0.8448209017515182} +12/23/2021 19:39:48 - INFO - codeparrot_training - Step 6337: {'lr': 0.0004813667596429905, 'samples': 3245056, 'steps': 6337, 'batch_loss/train': 0.911784770898521} +12/23/2021 19:39:59 - INFO - codeparrot_training - Step 6338: {'lr': 0.0004813607876642392, 'samples': 3245568, 'steps': 6338, 'batch_loss/train': 0.7801407835213467} +12/23/2021 19:40:11 - INFO - codeparrot_training - Step 6339: {'lr': 0.00048135481476568474, 'samples': 3246080, 'steps': 6339, 'batch_loss/train': 0.7718469509854913} +12/23/2021 19:40:22 - INFO - codeparrot_training - Step 6340: {'lr': 0.00048134884094735097, 'samples': 3246592, 'steps': 6340, 'batch_loss/train': 0.9036385612562299} +12/23/2021 19:40:32 - INFO - codeparrot_training - Step 6341: {'lr': 0.0004813428662092616, 'samples': 3247104, 'steps': 6341, 'batch_loss/train': 0.9393473770469427} +12/23/2021 19:40:44 - INFO - codeparrot_training - Step 6342: {'lr': 0.0004813368905514403, 'samples': 3247616, 'steps': 6342, 'batch_loss/train': 0.8473982876166701} +12/23/2021 19:40:55 - INFO - codeparrot_training - Step 6343: {'lr': 0.0004813309139739108, 'samples': 3248128, 'steps': 6343, 'batch_loss/train': 0.9001483237370849} +12/23/2021 19:41:05 - INFO - codeparrot_training - Step 6344: {'lr': 0.00048132493647669707, 'samples': 3248640, 'steps': 6344, 'batch_loss/train': 0.8117022826336324} +12/23/2021 19:41:18 - INFO - codeparrot_training - Step 6345: {'lr': 0.0004813189580598226, 'samples': 3249152, 'steps': 6345, 'batch_loss/train': 0.8036365965381265} +12/23/2021 19:41:28 - INFO - codeparrot_training - Step 6346: {'lr': 0.00048131297872331145, 'samples': 3249664, 'steps': 6346, 'batch_loss/train': 0.8008499867282808} +12/23/2021 19:41:39 - INFO - codeparrot_training - Step 6347: {'lr': 0.0004813069984671871, 'samples': 3250176, 'steps': 6347, 'batch_loss/train': 0.8498573061078787} +12/23/2021 19:41:49 - INFO - codeparrot_training - Step 6348: {'lr': 0.00048130101729147354, 'samples': 3250688, 'steps': 6348, 'batch_loss/train': 0.8696333570405841} +12/23/2021 19:42:03 - INFO - codeparrot_training - Step 6349: {'lr': 0.00048129503519619445, 'samples': 3251200, 'steps': 6349, 'batch_loss/train': 0.6134173478931189} +12/23/2021 19:42:14 - INFO - codeparrot_training - Step 6350: {'lr': 0.0004812890521813736, 'samples': 3251712, 'steps': 6350, 'batch_loss/train': 0.856201246380806} +12/23/2021 19:42:24 - INFO - codeparrot_training - Step 6351: {'lr': 0.0004812830682470348, 'samples': 3252224, 'steps': 6351, 'batch_loss/train': 1.0737401600927114} +12/23/2021 19:42:36 - INFO - codeparrot_training - Step 6352: {'lr': 0.00048127708339320196, 'samples': 3252736, 'steps': 6352, 'batch_loss/train': 0.7982528447173536} +12/23/2021 19:42:47 - INFO - codeparrot_training - Step 6353: {'lr': 0.0004812710976198987, 'samples': 3253248, 'steps': 6353, 'batch_loss/train': 0.8610861953347921} +12/23/2021 19:42:57 - INFO - codeparrot_training - Step 6354: {'lr': 0.00048126511092714885, 'samples': 3253760, 'steps': 6354, 'batch_loss/train': 0.8516823714599013} +12/23/2021 19:43:09 - INFO - codeparrot_training - Step 6355: {'lr': 0.0004812591233149762, 'samples': 3254272, 'steps': 6355, 'batch_loss/train': 0.8582667214795947} +12/23/2021 19:43:20 - INFO - codeparrot_training - Step 6356: {'lr': 0.0004812531347834047, 'samples': 3254784, 'steps': 6356, 'batch_loss/train': 0.8837211281061172} +12/23/2021 19:43:31 - INFO - codeparrot_training - Step 6357: {'lr': 0.00048124714533245806, 'samples': 3255296, 'steps': 6357, 'batch_loss/train': 0.8715837579220533} +12/23/2021 19:43:44 - INFO - codeparrot_training - Step 6358: {'lr': 0.00048124115496216006, 'samples': 3255808, 'steps': 6358, 'batch_loss/train': 0.7893350338563323} +12/23/2021 19:43:55 - INFO - codeparrot_training - Step 6359: {'lr': 0.00048123516367253445, 'samples': 3256320, 'steps': 6359, 'batch_loss/train': 1.5292960852384567} +12/23/2021 19:44:05 - INFO - codeparrot_training - Step 6360: {'lr': 0.0004812291714636052, 'samples': 3256832, 'steps': 6360, 'batch_loss/train': 0.9796842550858855} +12/23/2021 19:44:16 - INFO - codeparrot_training - Step 6361: {'lr': 0.00048122317833539606, 'samples': 3257344, 'steps': 6361, 'batch_loss/train': 0.8317240439355373} +12/23/2021 19:44:28 - INFO - codeparrot_training - Step 6362: {'lr': 0.00048121718428793093, 'samples': 3257856, 'steps': 6362, 'batch_loss/train': 0.903604437597096} +12/23/2021 19:44:39 - INFO - codeparrot_training - Step 6363: {'lr': 0.00048121118932123355, 'samples': 3258368, 'steps': 6363, 'batch_loss/train': 0.6846163356676698} +12/23/2021 19:44:49 - INFO - codeparrot_training - Step 6364: {'lr': 0.00048120519343532777, 'samples': 3258880, 'steps': 6364, 'batch_loss/train': 0.6292945020832121} +12/23/2021 19:45:03 - INFO - codeparrot_training - Step 6365: {'lr': 0.0004811991966302374, 'samples': 3259392, 'steps': 6365, 'batch_loss/train': 0.7843211852014065} +12/23/2021 19:45:13 - INFO - codeparrot_training - Step 6366: {'lr': 0.0004811931989059864, 'samples': 3259904, 'steps': 6366, 'batch_loss/train': 0.8552224021404982} +12/23/2021 19:45:24 - INFO - codeparrot_training - Step 6367: {'lr': 0.0004811872002625985, 'samples': 3260416, 'steps': 6367, 'batch_loss/train': 0.8568976307287812} +12/23/2021 19:45:36 - INFO - codeparrot_training - Step 6368: {'lr': 0.00048118120070009756, 'samples': 3260928, 'steps': 6368, 'batch_loss/train': 1.033564972691238} +12/23/2021 19:45:46 - INFO - codeparrot_training - Step 6369: {'lr': 0.00048117520021850754, 'samples': 3261440, 'steps': 6369, 'batch_loss/train': 0.7953883782029152} +12/23/2021 19:45:57 - INFO - codeparrot_training - Step 6370: {'lr': 0.00048116919881785214, 'samples': 3261952, 'steps': 6370, 'batch_loss/train': 0.8229173682630062} +12/23/2021 19:46:07 - INFO - codeparrot_training - Step 6371: {'lr': 0.0004811631964981553, 'samples': 3262464, 'steps': 6371, 'batch_loss/train': 0.9379030726850033} +12/23/2021 19:46:19 - INFO - codeparrot_training - Step 6372: {'lr': 0.00048115719325944085, 'samples': 3262976, 'steps': 6372, 'batch_loss/train': 0.8385634506121278} +12/23/2021 19:46:30 - INFO - codeparrot_training - Step 6373: {'lr': 0.00048115118910173273, 'samples': 3263488, 'steps': 6373, 'batch_loss/train': 0.9522645147517323} +12/23/2021 19:46:41 - INFO - codeparrot_training - Step 6374: {'lr': 0.00048114518402505473, 'samples': 3264000, 'steps': 6374, 'batch_loss/train': 0.626209256472066} +12/23/2021 19:46:54 - INFO - codeparrot_training - Step 6375: {'lr': 0.0004811391780294308, 'samples': 3264512, 'steps': 6375, 'batch_loss/train': 0.8362092841416597} +12/23/2021 19:47:05 - INFO - codeparrot_training - Step 6376: {'lr': 0.00048113317111488476, 'samples': 3265024, 'steps': 6376, 'batch_loss/train': 0.8504860168322921} +12/23/2021 19:47:15 - INFO - codeparrot_training - Step 6377: {'lr': 0.0004811271632814405, 'samples': 3265536, 'steps': 6377, 'batch_loss/train': 0.8129426334053278} +12/23/2021 19:47:27 - INFO - codeparrot_training - Step 6378: {'lr': 0.00048112115452912186, 'samples': 3266048, 'steps': 6378, 'batch_loss/train': 0.6898706387728453} +12/23/2021 19:47:38 - INFO - codeparrot_training - Step 6379: {'lr': 0.00048111514485795284, 'samples': 3266560, 'steps': 6379, 'batch_loss/train': 0.8930934621021152} +12/23/2021 19:47:49 - INFO - codeparrot_training - Step 6380: {'lr': 0.00048110913426795723, 'samples': 3267072, 'steps': 6380, 'batch_loss/train': 0.8650483749806881} +12/23/2021 19:48:01 - INFO - codeparrot_training - Step 6381: {'lr': 0.00048110312275915894, 'samples': 3267584, 'steps': 6381, 'batch_loss/train': 0.8397275903262198} +12/23/2021 19:48:12 - INFO - codeparrot_training - Step 6382: {'lr': 0.00048109711033158197, 'samples': 3268096, 'steps': 6382, 'batch_loss/train': 0.8087675860151649} +12/23/2021 19:48:22 - INFO - codeparrot_training - Step 6383: {'lr': 0.00048109109698525013, 'samples': 3268608, 'steps': 6383, 'batch_loss/train': 0.8483970323577523} +12/23/2021 19:48:33 - INFO - codeparrot_training - Step 6384: {'lr': 0.0004810850827201873, 'samples': 3269120, 'steps': 6384, 'batch_loss/train': 0.8381582163274288} +12/23/2021 19:48:45 - INFO - codeparrot_training - Step 6385: {'lr': 0.00048107906753641746, 'samples': 3269632, 'steps': 6385, 'batch_loss/train': 0.8905774354934692} +12/23/2021 19:48:55 - INFO - codeparrot_training - Step 6386: {'lr': 0.0004810730514339645, 'samples': 3270144, 'steps': 6386, 'batch_loss/train': 0.8074898542836308} +12/23/2021 19:49:06 - INFO - codeparrot_training - Step 6387: {'lr': 0.0004810670344128524, 'samples': 3270656, 'steps': 6387, 'batch_loss/train': 0.7808846835978329} +12/23/2021 19:49:19 - INFO - codeparrot_training - Step 6388: {'lr': 0.0004810610164731049, 'samples': 3271168, 'steps': 6388, 'batch_loss/train': 0.9271870595403016} +12/23/2021 19:49:30 - INFO - codeparrot_training - Step 6389: {'lr': 0.00048105499761474615, 'samples': 3271680, 'steps': 6389, 'batch_loss/train': 0.9958838375750929} +12/23/2021 19:49:40 - INFO - codeparrot_training - Step 6390: {'lr': 0.00048104897783779986, 'samples': 3272192, 'steps': 6390, 'batch_loss/train': 0.8702499251812696} +12/23/2021 19:49:52 - INFO - codeparrot_training - Step 6391: {'lr': 0.00048104295714229017, 'samples': 3272704, 'steps': 6391, 'batch_loss/train': 0.853831073269248} +12/23/2021 19:50:03 - INFO - codeparrot_training - Step 6392: {'lr': 0.00048103693552824093, 'samples': 3273216, 'steps': 6392, 'batch_loss/train': 0.795758131891489} +12/23/2021 19:50:14 - INFO - codeparrot_training - Step 6393: {'lr': 0.000481030912995676, 'samples': 3273728, 'steps': 6393, 'batch_loss/train': 0.804118241649121} +12/23/2021 19:50:24 - INFO - codeparrot_training - Step 6394: {'lr': 0.0004810248895446194, 'samples': 3274240, 'steps': 6394, 'batch_loss/train': 0.8954105116426945} +12/23/2021 19:50:37 - INFO - codeparrot_training - Step 6395: {'lr': 0.0004810188651750951, 'samples': 3274752, 'steps': 6395, 'batch_loss/train': 0.7883667070418596} +12/23/2021 19:50:48 - INFO - codeparrot_training - Step 6396: {'lr': 0.000481012839887127, 'samples': 3275264, 'steps': 6396, 'batch_loss/train': 0.8737477893009782} +12/23/2021 19:50:59 - INFO - codeparrot_training - Step 6397: {'lr': 0.0004810068136807391, 'samples': 3275776, 'steps': 6397, 'batch_loss/train': 0.8072757786139846} +12/23/2021 19:51:11 - INFO - codeparrot_training - Step 6398: {'lr': 0.0004810007865559554, 'samples': 3276288, 'steps': 6398, 'batch_loss/train': 0.7836789852008224} +12/23/2021 19:51:22 - INFO - codeparrot_training - Step 6399: {'lr': 0.0004809947585127996, 'samples': 3276800, 'steps': 6399, 'batch_loss/train': 0.8123418122995645} +12/23/2021 19:51:32 - INFO - codeparrot_training - Step 6400: {'lr': 0.000480988729551296, 'samples': 3277312, 'steps': 6400, 'batch_loss/train': 0.9266685266047716} +12/23/2021 19:51:44 - INFO - codeparrot_training - Step 6401: {'lr': 0.00048098269967146846, 'samples': 3277824, 'steps': 6401, 'batch_loss/train': 0.8726921994239092} +12/23/2021 19:51:55 - INFO - codeparrot_training - Step 6402: {'lr': 0.00048097666887334087, 'samples': 3278336, 'steps': 6402, 'batch_loss/train': 0.7403618663083762} +12/23/2021 19:52:06 - INFO - codeparrot_training - Step 6403: {'lr': 0.00048097063715693723, 'samples': 3278848, 'steps': 6403, 'batch_loss/train': 0.85459856223315} +12/23/2021 19:52:16 - INFO - codeparrot_training - Step 6404: {'lr': 0.0004809646045222816, 'samples': 3279360, 'steps': 6404, 'batch_loss/train': 0.8324901051819324} +12/23/2021 19:52:30 - INFO - codeparrot_training - Step 6405: {'lr': 0.00048095857096939787, 'samples': 3279872, 'steps': 6405, 'batch_loss/train': 0.7237504990771413} +12/23/2021 19:52:41 - INFO - codeparrot_training - Step 6406: {'lr': 0.00048095253649831006, 'samples': 3280384, 'steps': 6406, 'batch_loss/train': 0.8302901855204254} +12/23/2021 19:52:51 - INFO - codeparrot_training - Step 6407: {'lr': 0.0004809465011090423, 'samples': 3280896, 'steps': 6407, 'batch_loss/train': 0.8410794641822577} +12/23/2021 19:53:03 - INFO - codeparrot_training - Step 6408: {'lr': 0.00048094046480161827, 'samples': 3281408, 'steps': 6408, 'batch_loss/train': 0.8613712405785918} +12/23/2021 19:53:14 - INFO - codeparrot_training - Step 6409: {'lr': 0.0004809344275760622, 'samples': 3281920, 'steps': 6409, 'batch_loss/train': 0.7484542764723301} +12/23/2021 19:53:25 - INFO - codeparrot_training - Step 6410: {'lr': 0.00048092838943239815, 'samples': 3282432, 'steps': 6410, 'batch_loss/train': 1.0185464266687632} +12/23/2021 19:53:39 - INFO - codeparrot_training - Step 6411: {'lr': 0.0004809223503706499, 'samples': 3282944, 'steps': 6411, 'batch_loss/train': 0.7482278631068766} +12/23/2021 19:53:50 - INFO - codeparrot_training - Step 6412: {'lr': 0.0004809163103908416, 'samples': 3283456, 'steps': 6412, 'batch_loss/train': 0.8691043246071786} +12/23/2021 19:54:01 - INFO - codeparrot_training - Step 6413: {'lr': 0.00048091026949299734, 'samples': 3283968, 'steps': 6413, 'batch_loss/train': 1.037929973565042} +12/23/2021 19:54:11 - INFO - codeparrot_training - Step 6414: {'lr': 0.0004809042276771409, 'samples': 3284480, 'steps': 6414, 'batch_loss/train': 1.1975951215717942} +12/23/2021 19:54:24 - INFO - codeparrot_training - Step 6415: {'lr': 0.0004808981849432965, 'samples': 3284992, 'steps': 6415, 'batch_loss/train': 0.9099480397999287} +12/23/2021 19:54:34 - INFO - codeparrot_training - Step 6416: {'lr': 0.00048089214129148813, 'samples': 3285504, 'steps': 6416, 'batch_loss/train': 0.8592549653258175} +12/23/2021 19:54:45 - INFO - codeparrot_training - Step 6417: {'lr': 0.00048088609672173975, 'samples': 3286016, 'steps': 6417, 'batch_loss/train': 0.9314098749309778} +12/23/2021 19:54:57 - INFO - codeparrot_training - Step 6418: {'lr': 0.0004808800512340754, 'samples': 3286528, 'steps': 6418, 'batch_loss/train': 0.9605526095256209} +12/23/2021 19:55:08 - INFO - codeparrot_training - Step 6419: {'lr': 0.00048087400482851915, 'samples': 3287040, 'steps': 6419, 'batch_loss/train': 0.851240792311728} +12/23/2021 19:55:18 - INFO - codeparrot_training - Step 6420: {'lr': 0.0004808679575050951, 'samples': 3287552, 'steps': 6420, 'batch_loss/train': 0.9608183884993196} +12/23/2021 19:55:30 - INFO - codeparrot_training - Step 6421: {'lr': 0.00048086190926382717, 'samples': 3288064, 'steps': 6421, 'batch_loss/train': 0.838118314743042} +12/23/2021 19:55:41 - INFO - codeparrot_training - Step 6422: {'lr': 0.00048085586010473947, 'samples': 3288576, 'steps': 6422, 'batch_loss/train': 0.8038072532508522} +12/23/2021 19:55:52 - INFO - codeparrot_training - Step 6423: {'lr': 0.00048084981002785603, 'samples': 3289088, 'steps': 6423, 'batch_loss/train': 0.8141986755654216} +12/23/2021 19:56:02 - INFO - codeparrot_training - Step 6424: {'lr': 0.00048084375903320093, 'samples': 3289600, 'steps': 6424, 'batch_loss/train': 1.060533693060279} +12/23/2021 19:56:16 - INFO - codeparrot_training - Step 6425: {'lr': 0.00048083770712079824, 'samples': 3290112, 'steps': 6425, 'batch_loss/train': 1.7615102054551244} +12/23/2021 19:56:26 - INFO - codeparrot_training - Step 6426: {'lr': 0.00048083165429067186, 'samples': 3290624, 'steps': 6426, 'batch_loss/train': 0.9591216673143208} +12/23/2021 19:56:37 - INFO - codeparrot_training - Step 6427: {'lr': 0.00048082560054284613, 'samples': 3291136, 'steps': 6427, 'batch_loss/train': 0.8680540444329381} +12/23/2021 19:56:49 - INFO - codeparrot_training - Step 6428: {'lr': 0.0004808195458773449, 'samples': 3291648, 'steps': 6428, 'batch_loss/train': 0.8220078824087977} +12/23/2021 19:56:59 - INFO - codeparrot_training - Step 6429: {'lr': 0.0004808134902941923, 'samples': 3292160, 'steps': 6429, 'batch_loss/train': 0.8071011849679053} +12/23/2021 19:57:10 - INFO - codeparrot_training - Step 6430: {'lr': 0.0004808074337934124, 'samples': 3292672, 'steps': 6430, 'batch_loss/train': 0.7421411306131631} +12/23/2021 19:57:23 - INFO - codeparrot_training - Step 6431: {'lr': 0.00048080137637502933, 'samples': 3293184, 'steps': 6431, 'batch_loss/train': 0.9479916235432029} +12/23/2021 19:57:34 - INFO - codeparrot_training - Step 6432: {'lr': 0.00048079531803906715, 'samples': 3293696, 'steps': 6432, 'batch_loss/train': 0.8733512340113521} +12/23/2021 19:57:44 - INFO - codeparrot_training - Step 6433: {'lr': 0.0004807892587855499, 'samples': 3294208, 'steps': 6433, 'batch_loss/train': 1.2029640022665262} +12/23/2021 19:57:55 - INFO - codeparrot_training - Step 6434: {'lr': 0.0004807831986145017, 'samples': 3294720, 'steps': 6434, 'batch_loss/train': 0.9388163797557354} +12/23/2021 19:58:08 - INFO - codeparrot_training - Step 6435: {'lr': 0.0004807771375259466, 'samples': 3295232, 'steps': 6435, 'batch_loss/train': 0.8262115097604692} +12/23/2021 19:58:19 - INFO - codeparrot_training - Step 6436: {'lr': 0.00048077107551990886, 'samples': 3295744, 'steps': 6436, 'batch_loss/train': 0.9198131035082042} +12/23/2021 19:58:29 - INFO - codeparrot_training - Step 6437: {'lr': 0.0004807650125964124, 'samples': 3296256, 'steps': 6437, 'batch_loss/train': 0.9057351322844625} +12/23/2021 19:58:41 - INFO - codeparrot_training - Step 6438: {'lr': 0.0004807589487554813, 'samples': 3296768, 'steps': 6438, 'batch_loss/train': 0.7224897118285298} +12/23/2021 19:58:52 - INFO - codeparrot_training - Step 6439: {'lr': 0.00048075288399713985, 'samples': 3297280, 'steps': 6439, 'batch_loss/train': 0.8056897735223174} +12/23/2021 19:59:03 - INFO - codeparrot_training - Step 6440: {'lr': 0.00048074681832141204, 'samples': 3297792, 'steps': 6440, 'batch_loss/train': 0.9764278046786785} +12/23/2021 19:59:16 - INFO - codeparrot_training - Step 6441: {'lr': 0.000480740751728322, 'samples': 3298304, 'steps': 6441, 'batch_loss/train': 0.8987559946253896} +12/23/2021 19:59:26 - INFO - codeparrot_training - Step 6442: {'lr': 0.00048073468421789385, 'samples': 3298816, 'steps': 6442, 'batch_loss/train': 0.7083995998837054} +12/23/2021 19:59:37 - INFO - codeparrot_training - Step 6443: {'lr': 0.0004807286157901517, 'samples': 3299328, 'steps': 6443, 'batch_loss/train': 0.961805148050189} +12/23/2021 19:59:47 - INFO - codeparrot_training - Step 6444: {'lr': 0.0004807225464451197, 'samples': 3299840, 'steps': 6444, 'batch_loss/train': 0.8678195807151496} +12/23/2021 20:00:00 - INFO - codeparrot_training - Step 6445: {'lr': 0.00048071647618282197, 'samples': 3300352, 'steps': 6445, 'batch_loss/train': 0.8939205892384052} +12/23/2021 20:00:10 - INFO - codeparrot_training - Step 6446: {'lr': 0.0004807104050032827, 'samples': 3300864, 'steps': 6446, 'batch_loss/train': 1.3441665079444647} +12/23/2021 20:00:21 - INFO - codeparrot_training - Step 6447: {'lr': 0.00048070433290652595, 'samples': 3301376, 'steps': 6447, 'batch_loss/train': 0.8329836772754788} +12/23/2021 20:00:33 - INFO - codeparrot_training - Step 6448: {'lr': 0.0004806982598925759, 'samples': 3301888, 'steps': 6448, 'batch_loss/train': 0.8244279325008392} +12/23/2021 20:00:43 - INFO - codeparrot_training - Step 6449: {'lr': 0.00048069218596145664, 'samples': 3302400, 'steps': 6449, 'batch_loss/train': 0.9494886007159948} +12/23/2021 20:00:54 - INFO - codeparrot_training - Step 6450: {'lr': 0.0004806861111131924, 'samples': 3302912, 'steps': 6450, 'batch_loss/train': 0.8874971494078636} +12/23/2021 20:01:07 - INFO - codeparrot_training - Step 6451: {'lr': 0.00048068003534780724, 'samples': 3303424, 'steps': 6451, 'batch_loss/train': 0.8556239377940074} +12/23/2021 20:01:18 - INFO - codeparrot_training - Step 6452: {'lr': 0.0004806739586653254, 'samples': 3303936, 'steps': 6452, 'batch_loss/train': 0.9241299037821591} +12/23/2021 20:01:29 - INFO - codeparrot_training - Step 6453: {'lr': 0.0004806678810657709, 'samples': 3304448, 'steps': 6453, 'batch_loss/train': 1.1470076469704509} +12/23/2021 20:01:39 - INFO - codeparrot_training - Step 6454: {'lr': 0.00048066180254916803, 'samples': 3304960, 'steps': 6454, 'batch_loss/train': 0.766250197775662} +12/23/2021 20:01:51 - INFO - codeparrot_training - Step 6455: {'lr': 0.000480655723115541, 'samples': 3305472, 'steps': 6455, 'batch_loss/train': 0.8772042475175112} +12/23/2021 20:02:02 - INFO - codeparrot_training - Step 6456: {'lr': 0.00048064964276491387, 'samples': 3305984, 'steps': 6456, 'batch_loss/train': 0.8490691867191344} +12/23/2021 20:02:13 - INFO - codeparrot_training - Step 6457: {'lr': 0.0004806435614973109, 'samples': 3306496, 'steps': 6457, 'batch_loss/train': 0.9350399544928223} +12/23/2021 20:02:25 - INFO - codeparrot_training - Step 6458: {'lr': 0.00048063747931275616, 'samples': 3307008, 'steps': 6458, 'batch_loss/train': 0.9145379476249218} +12/23/2021 20:02:35 - INFO - codeparrot_training - Step 6459: {'lr': 0.0004806313962112739, 'samples': 3307520, 'steps': 6459, 'batch_loss/train': 0.8328520357608795} +12/23/2021 20:02:46 - INFO - codeparrot_training - Step 6460: {'lr': 0.0004806253121928883, 'samples': 3308032, 'steps': 6460, 'batch_loss/train': 0.8851549779064953} +12/23/2021 20:02:58 - INFO - codeparrot_training - Step 6461: {'lr': 0.00048061922725762353, 'samples': 3308544, 'steps': 6461, 'batch_loss/train': 0.8435468666721135} +12/23/2021 20:03:09 - INFO - codeparrot_training - Step 6462: {'lr': 0.00048061314140550375, 'samples': 3309056, 'steps': 6462, 'batch_loss/train': 0.895835685543716} +12/23/2021 20:03:19 - INFO - codeparrot_training - Step 6463: {'lr': 0.00048060705463655323, 'samples': 3309568, 'steps': 6463, 'batch_loss/train': 0.9102049637585878} +12/23/2021 20:03:30 - INFO - codeparrot_training - Step 6464: {'lr': 0.00048060096695079615, 'samples': 3310080, 'steps': 6464, 'batch_loss/train': 0.8298476664349437} +12/23/2021 20:03:43 - INFO - codeparrot_training - Step 6465: {'lr': 0.0004805948783482567, 'samples': 3310592, 'steps': 6465, 'batch_loss/train': 0.7796417763456702} +12/23/2021 20:03:53 - INFO - codeparrot_training - Step 6466: {'lr': 0.0004805887888289591, 'samples': 3311104, 'steps': 6466, 'batch_loss/train': 0.8460877658799291} +12/23/2021 20:04:04 - INFO - codeparrot_training - Step 6467: {'lr': 0.00048058269839292743, 'samples': 3311616, 'steps': 6467, 'batch_loss/train': 0.8286208184435964} +12/23/2021 20:04:16 - INFO - codeparrot_training - Step 6468: {'lr': 0.0004805766070401861, 'samples': 3312128, 'steps': 6468, 'batch_loss/train': 0.9626922030001879} +12/23/2021 20:04:27 - INFO - codeparrot_training - Step 6469: {'lr': 0.0004805705147707593, 'samples': 3312640, 'steps': 6469, 'batch_loss/train': 0.8472865018993616} +12/23/2021 20:04:37 - INFO - codeparrot_training - Step 6470: {'lr': 0.0004805644215846712, 'samples': 3313152, 'steps': 6470, 'batch_loss/train': 0.8984522046521306} +12/23/2021 20:04:49 - INFO - codeparrot_training - Step 6471: {'lr': 0.0004805583274819459, 'samples': 3313664, 'steps': 6471, 'batch_loss/train': 0.8957743154605851} +12/23/2021 20:05:00 - INFO - codeparrot_training - Step 6472: {'lr': 0.00048055223246260784, 'samples': 3314176, 'steps': 6472, 'batch_loss/train': 0.8871749863028526} +12/23/2021 20:05:10 - INFO - codeparrot_training - Step 6473: {'lr': 0.00048054613652668106, 'samples': 3314688, 'steps': 6473, 'batch_loss/train': 0.7900642659515142} +12/23/2021 20:05:24 - INFO - codeparrot_training - Step 6474: {'lr': 0.00048054003967419, 'samples': 3315200, 'steps': 6474, 'batch_loss/train': 0.906851259060204} +12/23/2021 20:05:34 - INFO - codeparrot_training - Step 6475: {'lr': 0.0004805339419051587, 'samples': 3315712, 'steps': 6475, 'batch_loss/train': 0.7048701220192015} +12/23/2021 20:05:45 - INFO - codeparrot_training - Step 6476: {'lr': 0.0004805278432196115, 'samples': 3316224, 'steps': 6476, 'batch_loss/train': 0.8619322348386049} +12/23/2021 20:05:55 - INFO - codeparrot_training - Step 6477: {'lr': 0.0004805217436175727, 'samples': 3316736, 'steps': 6477, 'batch_loss/train': 0.8694862369447947} +12/23/2021 20:06:08 - INFO - codeparrot_training - Step 6478: {'lr': 0.0004805156430990665, 'samples': 3317248, 'steps': 6478, 'batch_loss/train': 1.5792609071359038} +12/23/2021 20:06:18 - INFO - codeparrot_training - Step 6479: {'lr': 0.0004805095416641171, 'samples': 3317760, 'steps': 6479, 'batch_loss/train': 0.8769527189433575} +12/23/2021 20:06:29 - INFO - codeparrot_training - Step 6480: {'lr': 0.00048050343931274874, 'samples': 3318272, 'steps': 6480, 'batch_loss/train': 0.9341939566656947} +12/23/2021 20:06:42 - INFO - codeparrot_training - Step 6481: {'lr': 0.0004804973360449858, 'samples': 3318784, 'steps': 6481, 'batch_loss/train': 0.821267444640398} +12/23/2021 20:06:53 - INFO - codeparrot_training - Step 6482: {'lr': 0.0004804912318608525, 'samples': 3319296, 'steps': 6482, 'batch_loss/train': 0.840239395853132} +12/23/2021 20:07:03 - INFO - codeparrot_training - Step 6483: {'lr': 0.0004804851267603731, 'samples': 3319808, 'steps': 6483, 'batch_loss/train': 0.9293851079419255} +12/23/2021 20:07:15 - INFO - codeparrot_training - Step 6484: {'lr': 0.00048047902074357186, 'samples': 3320320, 'steps': 6484, 'batch_loss/train': 0.9276780383661389} +12/23/2021 20:07:26 - INFO - codeparrot_training - Step 6485: {'lr': 0.000480472913810473, 'samples': 3320832, 'steps': 6485, 'batch_loss/train': 0.7623352063819766} +12/23/2021 20:07:36 - INFO - codeparrot_training - Step 6486: {'lr': 0.00048046680596110095, 'samples': 3321344, 'steps': 6486, 'batch_loss/train': 0.90085812471807} +12/23/2021 20:07:47 - INFO - codeparrot_training - Step 6487: {'lr': 0.0004804606971954798, 'samples': 3321856, 'steps': 6487, 'batch_loss/train': 0.806686520576477} +12/23/2021 20:07:59 - INFO - codeparrot_training - Step 6488: {'lr': 0.00048045458751363405, 'samples': 3322368, 'steps': 6488, 'batch_loss/train': 0.8182236813008785} +12/23/2021 20:08:10 - INFO - codeparrot_training - Step 6489: {'lr': 0.00048044847691558777, 'samples': 3322880, 'steps': 6489, 'batch_loss/train': 0.706020301906392} +12/23/2021 20:08:20 - INFO - codeparrot_training - Step 6490: {'lr': 0.00048044236540136545, 'samples': 3323392, 'steps': 6490, 'batch_loss/train': 0.8961585154756904} +12/23/2021 20:08:33 - INFO - codeparrot_training - Step 6491: {'lr': 0.0004804362529709912, 'samples': 3323904, 'steps': 6491, 'batch_loss/train': 0.908109886571765} +12/23/2021 20:08:44 - INFO - codeparrot_training - Step 6492: {'lr': 0.00048043013962448944, 'samples': 3324416, 'steps': 6492, 'batch_loss/train': 0.8798798006027937} +12/23/2021 20:08:55 - INFO - codeparrot_training - Step 6493: {'lr': 0.0004804240253618845, 'samples': 3324928, 'steps': 6493, 'batch_loss/train': 0.7110456082737073} +12/23/2021 20:09:07 - INFO - codeparrot_training - Step 6494: {'lr': 0.0004804179101832006, 'samples': 3325440, 'steps': 6494, 'batch_loss/train': 0.7935129031538963} +12/23/2021 20:09:17 - INFO - codeparrot_training - Step 6495: {'lr': 0.0004804117940884621, 'samples': 3325952, 'steps': 6495, 'batch_loss/train': 0.9848827430978417} +12/23/2021 20:09:28 - INFO - codeparrot_training - Step 6496: {'lr': 0.00048040567707769333, 'samples': 3326464, 'steps': 6496, 'batch_loss/train': 0.8262265999801457} +12/23/2021 20:09:40 - INFO - codeparrot_training - Step 6497: {'lr': 0.0004803995591509186, 'samples': 3326976, 'steps': 6497, 'batch_loss/train': 0.8847430376335979} +12/23/2021 20:09:50 - INFO - codeparrot_training - Step 6498: {'lr': 0.0004803934403081622, 'samples': 3327488, 'steps': 6498, 'batch_loss/train': 0.8682459238916636} +12/23/2021 20:10:01 - INFO - codeparrot_training - Step 6499: {'lr': 0.00048038732054944846, 'samples': 3328000, 'steps': 6499, 'batch_loss/train': 0.8376254271715879} +12/23/2021 20:10:12 - INFO - codeparrot_training - Step 6500: {'lr': 0.0004803811998748017, 'samples': 3328512, 'steps': 6500, 'batch_loss/train': 0.8831671602092683} +12/23/2021 20:10:24 - INFO - codeparrot_training - Step 6501: {'lr': 0.00048037507828424633, 'samples': 3329024, 'steps': 6501, 'batch_loss/train': 0.6934801011811942} +12/23/2021 20:10:34 - INFO - codeparrot_training - Step 6502: {'lr': 0.00048036895577780657, 'samples': 3329536, 'steps': 6502, 'batch_loss/train': 0.9600462084636092} +12/23/2021 20:10:45 - INFO - codeparrot_training - Step 6503: {'lr': 0.0004803628323555069, 'samples': 3330048, 'steps': 6503, 'batch_loss/train': 0.7408625164534897} +12/23/2021 20:10:58 - INFO - codeparrot_training - Step 6504: {'lr': 0.00048035670801737154, 'samples': 3330560, 'steps': 6504, 'batch_loss/train': 0.7881877408362925} +12/23/2021 20:11:09 - INFO - codeparrot_training - Step 6505: {'lr': 0.0004803505827634249, 'samples': 3331072, 'steps': 6505, 'batch_loss/train': 0.8209440105129033} +12/23/2021 20:11:19 - INFO - codeparrot_training - Step 6506: {'lr': 0.00048034445659369136, 'samples': 3331584, 'steps': 6506, 'batch_loss/train': 0.8336921203881502} +12/23/2021 20:11:31 - INFO - codeparrot_training - Step 6507: {'lr': 0.0004803383295081952, 'samples': 3332096, 'steps': 6507, 'batch_loss/train': 0.8553354847244918} +12/23/2021 20:11:42 - INFO - codeparrot_training - Step 6508: {'lr': 0.00048033220150696084, 'samples': 3332608, 'steps': 6508, 'batch_loss/train': 0.8041099812835455} +12/23/2021 20:11:53 - INFO - codeparrot_training - Step 6509: {'lr': 0.0004803260725900126, 'samples': 3333120, 'steps': 6509, 'batch_loss/train': 0.8419149741530418} +12/23/2021 20:12:03 - INFO - codeparrot_training - Step 6510: {'lr': 0.0004803199427573748, 'samples': 3333632, 'steps': 6510, 'batch_loss/train': 0.8743484714068472} +12/23/2021 20:12:17 - INFO - codeparrot_training - Step 6511: {'lr': 0.00048031381200907197, 'samples': 3334144, 'steps': 6511, 'batch_loss/train': 0.8302447162568569} +12/23/2021 20:12:27 - INFO - codeparrot_training - Step 6512: {'lr': 0.00048030768034512835, 'samples': 3334656, 'steps': 6512, 'batch_loss/train': 0.800785016734153} +12/23/2021 20:12:38 - INFO - codeparrot_training - Step 6513: {'lr': 0.00048030154776556834, 'samples': 3335168, 'steps': 6513, 'batch_loss/train': 0.8584978580474854} +12/23/2021 20:12:50 - INFO - codeparrot_training - Step 6514: {'lr': 0.00048029541427041636, 'samples': 3335680, 'steps': 6514, 'batch_loss/train': 0.8323653806000948} +12/23/2021 20:13:01 - INFO - codeparrot_training - Step 6515: {'lr': 0.00048028927985969673, 'samples': 3336192, 'steps': 6515, 'batch_loss/train': 0.8048906493932009} +12/23/2021 20:13:11 - INFO - codeparrot_training - Step 6516: {'lr': 0.0004802831445334339, 'samples': 3336704, 'steps': 6516, 'batch_loss/train': 0.9010712541639805} +12/23/2021 20:13:23 - INFO - codeparrot_training - Step 6517: {'lr': 0.00048027700829165223, 'samples': 3337216, 'steps': 6517, 'batch_loss/train': 0.8179630199447274} +12/23/2021 20:13:34 - INFO - codeparrot_training - Step 6518: {'lr': 0.00048027087113437615, 'samples': 3337728, 'steps': 6518, 'batch_loss/train': 0.8379490403458476} +12/23/2021 20:13:44 - INFO - codeparrot_training - Step 6519: {'lr': 0.00048026473306163, 'samples': 3338240, 'steps': 6519, 'batch_loss/train': 0.8069698652252555} +12/23/2021 20:13:58 - INFO - codeparrot_training - Step 6520: {'lr': 0.00048025859407343824, 'samples': 3338752, 'steps': 6520, 'batch_loss/train': 0.8801242681220174} +12/23/2021 20:14:09 - INFO - codeparrot_training - Step 6521: {'lr': 0.0004802524541698252, 'samples': 3339264, 'steps': 6521, 'batch_loss/train': 0.73272276786156} +12/23/2021 20:14:20 - INFO - codeparrot_training - Step 6522: {'lr': 0.00048024631335081544, 'samples': 3339776, 'steps': 6522, 'batch_loss/train': 0.744210752658546} +12/23/2021 20:14:30 - INFO - codeparrot_training - Step 6523: {'lr': 0.0004802401716164332, 'samples': 3340288, 'steps': 6523, 'batch_loss/train': 0.8441927218809724} +12/23/2021 20:14:43 - INFO - codeparrot_training - Step 6524: {'lr': 0.00048023402896670296, 'samples': 3340800, 'steps': 6524, 'batch_loss/train': 0.8280469486489892} +12/23/2021 20:14:53 - INFO - codeparrot_training - Step 6525: {'lr': 0.00048022788540164925, 'samples': 3341312, 'steps': 6525, 'batch_loss/train': 0.9143739575520158} +12/23/2021 20:15:04 - INFO - codeparrot_training - Step 6526: {'lr': 0.00048022174092129634, 'samples': 3341824, 'steps': 6526, 'batch_loss/train': 0.7089239874621853} +12/23/2021 20:15:17 - INFO - codeparrot_training - Step 6527: {'lr': 0.00048021559552566875, 'samples': 3342336, 'steps': 6527, 'batch_loss/train': 0.9276117165572941} +12/23/2021 20:15:28 - INFO - codeparrot_training - Step 6528: {'lr': 0.0004802094492147908, 'samples': 3342848, 'steps': 6528, 'batch_loss/train': 0.8338212631642818} +12/23/2021 20:15:38 - INFO - codeparrot_training - Step 6529: {'lr': 0.00048020330198868705, 'samples': 3343360, 'steps': 6529, 'batch_loss/train': 0.9346329420804977} +12/23/2021 20:15:50 - INFO - codeparrot_training - Step 6530: {'lr': 0.0004801971538473819, 'samples': 3343872, 'steps': 6530, 'batch_loss/train': 0.8599731419235468} +12/23/2021 20:16:01 - INFO - codeparrot_training - Step 6531: {'lr': 0.0004801910047908997, 'samples': 3344384, 'steps': 6531, 'batch_loss/train': 0.8909636307507753} +12/23/2021 20:16:11 - INFO - codeparrot_training - Step 6532: {'lr': 0.0004801848548192651, 'samples': 3344896, 'steps': 6532, 'batch_loss/train': 0.8314579529687762} +12/23/2021 20:16:23 - INFO - codeparrot_training - Step 6533: {'lr': 0.0004801787039325024, 'samples': 3345408, 'steps': 6533, 'batch_loss/train': 0.8733090781606734} +12/23/2021 20:16:34 - INFO - codeparrot_training - Step 6534: {'lr': 0.00048017255213063603, 'samples': 3345920, 'steps': 6534, 'batch_loss/train': 0.8336505237966776} +12/23/2021 20:16:45 - INFO - codeparrot_training - Step 6535: {'lr': 0.00048016639941369057, 'samples': 3346432, 'steps': 6535, 'batch_loss/train': 0.767832423793152} +12/23/2021 20:16:55 - INFO - codeparrot_training - Step 6536: {'lr': 0.00048016024578169045, 'samples': 3346944, 'steps': 6536, 'batch_loss/train': 0.9003973612561822} +12/23/2021 20:17:07 - INFO - codeparrot_training - Step 6537: {'lr': 0.00048015409123466, 'samples': 3347456, 'steps': 6537, 'batch_loss/train': 0.940321097150445} +12/23/2021 20:17:18 - INFO - codeparrot_training - Step 6538: {'lr': 0.0004801479357726238, 'samples': 3347968, 'steps': 6538, 'batch_loss/train': 0.8167865313589573} +12/23/2021 20:17:29 - INFO - codeparrot_training - Step 6539: {'lr': 0.00048014177939560633, 'samples': 3348480, 'steps': 6539, 'batch_loss/train': 0.9539633998647332} +12/23/2021 20:17:42 - INFO - codeparrot_training - Step 6540: {'lr': 0.00048013562210363203, 'samples': 3348992, 'steps': 6540, 'batch_loss/train': 0.8450488159433007} +12/23/2021 20:17:53 - INFO - codeparrot_training - Step 6541: {'lr': 0.0004801294638967254, 'samples': 3349504, 'steps': 6541, 'batch_loss/train': 0.6666249760892242} +12/23/2021 20:18:03 - INFO - codeparrot_training - Step 6542: {'lr': 0.0004801233047749109, 'samples': 3350016, 'steps': 6542, 'batch_loss/train': 0.8120054844766855} +12/23/2021 20:18:15 - INFO - codeparrot_training - Step 6543: {'lr': 0.0004801171447382131, 'samples': 3350528, 'steps': 6543, 'batch_loss/train': 0.6139815181377344} +12/23/2021 20:18:26 - INFO - codeparrot_training - Step 6544: {'lr': 0.0004801109837866564, 'samples': 3351040, 'steps': 6544, 'batch_loss/train': 0.8561935196630657} +12/23/2021 20:18:36 - INFO - codeparrot_training - Step 6545: {'lr': 0.00048010482192026527, 'samples': 3351552, 'steps': 6545, 'batch_loss/train': 0.6940255202935077} +12/23/2021 20:18:47 - INFO - codeparrot_training - Step 6546: {'lr': 0.0004800986591390643, 'samples': 3352064, 'steps': 6546, 'batch_loss/train': 0.7759471326135099} +12/23/2021 20:18:59 - INFO - codeparrot_training - Step 6547: {'lr': 0.0004800924954430779, 'samples': 3352576, 'steps': 6547, 'batch_loss/train': 0.860518092289567} +12/23/2021 20:19:10 - INFO - codeparrot_training - Step 6548: {'lr': 0.00048008633083233067, 'samples': 3353088, 'steps': 6548, 'batch_loss/train': 0.8524567922577262} +12/23/2021 20:19:20 - INFO - codeparrot_training - Step 6549: {'lr': 0.000480080165306847, 'samples': 3353600, 'steps': 6549, 'batch_loss/train': 0.7964440425857902} +12/23/2021 20:19:34 - INFO - codeparrot_training - Step 6550: {'lr': 0.00048007399886665147, 'samples': 3354112, 'steps': 6550, 'batch_loss/train': 0.7383102020248771} +12/23/2021 20:19:44 - INFO - codeparrot_training - Step 6551: {'lr': 0.00048006783151176866, 'samples': 3354624, 'steps': 6551, 'batch_loss/train': 0.8843648051843047} +12/23/2021 20:19:55 - INFO - codeparrot_training - Step 6552: {'lr': 0.000480061663242223, 'samples': 3355136, 'steps': 6552, 'batch_loss/train': 0.8106037890538573} +12/23/2021 20:20:07 - INFO - codeparrot_training - Step 6553: {'lr': 0.00048005549405803896, 'samples': 3355648, 'steps': 6553, 'batch_loss/train': 0.8211616163607687} +12/23/2021 20:20:18 - INFO - codeparrot_training - Step 6554: {'lr': 0.0004800493239592412, 'samples': 3356160, 'steps': 6554, 'batch_loss/train': 0.8528864551335573} +12/23/2021 20:20:28 - INFO - codeparrot_training - Step 6555: {'lr': 0.0004800431529458542, 'samples': 3356672, 'steps': 6555, 'batch_loss/train': 0.8219797797501087} +12/23/2021 20:20:39 - INFO - codeparrot_training - Step 6556: {'lr': 0.0004800369810179025, 'samples': 3357184, 'steps': 6556, 'batch_loss/train': 0.8364271381869912} +12/23/2021 20:20:51 - INFO - codeparrot_training - Step 6557: {'lr': 0.00048003080817541054, 'samples': 3357696, 'steps': 6557, 'batch_loss/train': 0.7052570275263861} +12/23/2021 20:21:02 - INFO - codeparrot_training - Step 6558: {'lr': 0.0004800246344184031, 'samples': 3358208, 'steps': 6558, 'batch_loss/train': 0.9026262436527759} +12/23/2021 20:21:12 - INFO - codeparrot_training - Step 6559: {'lr': 0.0004800184597469044, 'samples': 3358720, 'steps': 6559, 'batch_loss/train': 0.885263055562973} +12/23/2021 20:21:26 - INFO - codeparrot_training - Step 6560: {'lr': 0.00048001228416093923, 'samples': 3359232, 'steps': 6560, 'batch_loss/train': 0.8901649378240108} +12/23/2021 20:21:36 - INFO - codeparrot_training - Step 6561: {'lr': 0.000480006107660532, 'samples': 3359744, 'steps': 6561, 'batch_loss/train': 0.7668513956014067} +12/23/2021 20:21:47 - INFO - codeparrot_training - Step 6562: {'lr': 0.00047999993024570745, 'samples': 3360256, 'steps': 6562, 'batch_loss/train': 0.9234152613207698} +12/23/2021 20:21:59 - INFO - codeparrot_training - Step 6563: {'lr': 0.00047999375191648997, 'samples': 3360768, 'steps': 6563, 'batch_loss/train': 0.8032295210286975} +12/23/2021 20:22:10 - INFO - codeparrot_training - Step 6564: {'lr': 0.0004799875726729042, 'samples': 3361280, 'steps': 6564, 'batch_loss/train': 0.8076360421255231} +12/23/2021 20:22:20 - INFO - codeparrot_training - Step 6565: {'lr': 0.00047998139251497455, 'samples': 3361792, 'steps': 6565, 'batch_loss/train': 0.8898304784670472} +12/23/2021 20:22:33 - INFO - codeparrot_training - Step 6566: {'lr': 0.00047997521144272585, 'samples': 3362304, 'steps': 6566, 'batch_loss/train': 0.8550694966688752} +12/23/2021 20:22:44 - INFO - codeparrot_training - Step 6567: {'lr': 0.0004799690294561825, 'samples': 3362816, 'steps': 6567, 'batch_loss/train': 0.8138840948231518} +12/23/2021 20:22:55 - INFO - codeparrot_training - Step 6568: {'lr': 0.00047996284655536916, 'samples': 3363328, 'steps': 6568, 'batch_loss/train': 0.9386024735867977} +12/23/2021 20:23:05 - INFO - codeparrot_training - Step 6569: {'lr': 0.00047995666274031035, 'samples': 3363840, 'steps': 6569, 'batch_loss/train': 0.8730572648346424} +12/23/2021 20:23:18 - INFO - codeparrot_training - Step 6570: {'lr': 0.00047995047801103064, 'samples': 3364352, 'steps': 6570, 'batch_loss/train': 0.7664410825818777} +12/23/2021 20:23:28 - INFO - codeparrot_training - Step 6571: {'lr': 0.00047994429236755475, 'samples': 3364864, 'steps': 6571, 'batch_loss/train': 0.8345467457547784} +12/23/2021 20:23:39 - INFO - codeparrot_training - Step 6572: {'lr': 0.0004799381058099071, 'samples': 3365376, 'steps': 6572, 'batch_loss/train': 0.7103086861316115} +12/23/2021 20:23:51 - INFO - codeparrot_training - Step 6573: {'lr': 0.0004799319183381124, 'samples': 3365888, 'steps': 6573, 'batch_loss/train': 0.7529295559506863} +12/23/2021 20:24:02 - INFO - codeparrot_training - Step 6574: {'lr': 0.0004799257299521952, 'samples': 3366400, 'steps': 6574, 'batch_loss/train': 0.8248808747157454} +12/23/2021 20:24:12 - INFO - codeparrot_training - Step 6575: {'lr': 0.00047991954065218013, 'samples': 3366912, 'steps': 6575, 'batch_loss/train': 0.8490822650492191} +12/23/2021 20:24:24 - INFO - codeparrot_training - Step 6576: {'lr': 0.0004799133504380917, 'samples': 3367424, 'steps': 6576, 'batch_loss/train': 0.8851858330890536} +12/23/2021 20:24:35 - INFO - codeparrot_training - Step 6577: {'lr': 0.00047990715930995467, 'samples': 3367936, 'steps': 6577, 'batch_loss/train': 0.8618127414956689} +12/23/2021 20:24:46 - INFO - codeparrot_training - Step 6578: {'lr': 0.0004799009672677936, 'samples': 3368448, 'steps': 6578, 'batch_loss/train': 0.8040322391316295} +12/23/2021 20:24:56 - INFO - codeparrot_training - Step 6579: {'lr': 0.000479894774311633, 'samples': 3368960, 'steps': 6579, 'batch_loss/train': 0.896579273045063} +12/23/2021 20:25:10 - INFO - codeparrot_training - Step 6580: {'lr': 0.0004798885804414977, 'samples': 3369472, 'steps': 6580, 'batch_loss/train': 0.7979774139821529} +12/23/2021 20:25:21 - INFO - codeparrot_training - Step 6581: {'lr': 0.00047988238565741214, 'samples': 3369984, 'steps': 6581, 'batch_loss/train': 0.7537045758217573} +12/23/2021 20:25:31 - INFO - codeparrot_training - Step 6582: {'lr': 0.00047987618995940103, 'samples': 3370496, 'steps': 6582, 'batch_loss/train': 0.8839086559601128} +12/23/2021 20:25:43 - INFO - codeparrot_training - Step 6583: {'lr': 0.00047986999334748896, 'samples': 3371008, 'steps': 6583, 'batch_loss/train': 0.7279384811408818} +12/23/2021 20:25:54 - INFO - codeparrot_training - Step 6584: {'lr': 0.00047986379582170066, 'samples': 3371520, 'steps': 6584, 'batch_loss/train': 0.856566933915019} +12/23/2021 20:26:05 - INFO - codeparrot_training - Step 6585: {'lr': 0.00047985759738206063, 'samples': 3372032, 'steps': 6585, 'batch_loss/train': 0.8632339239120483} +12/23/2021 20:26:17 - INFO - codeparrot_training - Step 6586: {'lr': 0.00047985139802859364, 'samples': 3372544, 'steps': 6586, 'batch_loss/train': 0.8242329321801662} +12/23/2021 20:26:27 - INFO - codeparrot_training - Step 6587: {'lr': 0.00047984519776132425, 'samples': 3373056, 'steps': 6587, 'batch_loss/train': 0.81181943975389} +12/23/2021 20:26:38 - INFO - codeparrot_training - Step 6588: {'lr': 0.00047983899658027713, 'samples': 3373568, 'steps': 6588, 'batch_loss/train': 0.9180567217990756} +12/23/2021 20:26:51 - INFO - codeparrot_training - Step 6589: {'lr': 0.00047983279448547694, 'samples': 3374080, 'steps': 6589, 'batch_loss/train': 0.9406764656305313} +12/23/2021 20:27:02 - INFO - codeparrot_training - Step 6590: {'lr': 0.00047982659147694835, 'samples': 3374592, 'steps': 6590, 'batch_loss/train': 0.8980073183774948} +12/23/2021 20:27:12 - INFO - codeparrot_training - Step 6591: {'lr': 0.000479820387554716, 'samples': 3375104, 'steps': 6591, 'batch_loss/train': 0.8851090921089053} +12/23/2021 20:27:23 - INFO - codeparrot_training - Step 6592: {'lr': 0.0004798141827188046, 'samples': 3375616, 'steps': 6592, 'batch_loss/train': 0.833463029935956} +12/23/2021 20:27:35 - INFO - codeparrot_training - Step 6593: {'lr': 0.00047980797696923875, 'samples': 3376128, 'steps': 6593, 'batch_loss/train': 0.888979397714138} +12/23/2021 20:27:45 - INFO - codeparrot_training - Step 6594: {'lr': 0.0004798017703060432, 'samples': 3376640, 'steps': 6594, 'batch_loss/train': 0.871117239817977} +12/23/2021 20:27:56 - INFO - codeparrot_training - Step 6595: {'lr': 0.0004797955627292425, 'samples': 3377152, 'steps': 6595, 'batch_loss/train': 0.7689546751789749} +12/23/2021 20:28:10 - INFO - codeparrot_training - Step 6596: {'lr': 0.0004797893542388615, 'samples': 3377664, 'steps': 6596, 'batch_loss/train': 0.8592088455334306} +12/23/2021 20:28:20 - INFO - codeparrot_training - Step 6597: {'lr': 0.0004797831448349248, 'samples': 3378176, 'steps': 6597, 'batch_loss/train': 0.9597657714039087} +12/23/2021 20:28:31 - INFO - codeparrot_training - Step 6598: {'lr': 0.000479776934517457, 'samples': 3378688, 'steps': 6598, 'batch_loss/train': 0.8436875697225332} +12/23/2021 20:28:43 - INFO - codeparrot_training - Step 6599: {'lr': 0.0004797707232864829, 'samples': 3379200, 'steps': 6599, 'batch_loss/train': 0.805052412673831} +12/23/2021 20:28:53 - INFO - codeparrot_training - Step 6600: {'lr': 0.0004797645111420272, 'samples': 3379712, 'steps': 6600, 'batch_loss/train': 0.7700484776869416} +12/23/2021 20:29:04 - INFO - codeparrot_training - Step 6601: {'lr': 0.0004797582980841145, 'samples': 3380224, 'steps': 6601, 'batch_loss/train': 0.7565264399163425} +12/23/2021 20:29:14 - INFO - codeparrot_training - Step 6602: {'lr': 0.0004797520841127696, 'samples': 3380736, 'steps': 6602, 'batch_loss/train': 0.7521341005340219} +12/23/2021 20:29:27 - INFO - codeparrot_training - Step 6603: {'lr': 0.0004797458692280171, 'samples': 3381248, 'steps': 6603, 'batch_loss/train': 0.8128621224313974} +12/23/2021 20:29:37 - INFO - codeparrot_training - Step 6604: {'lr': 0.00047973965342988184, 'samples': 3381760, 'steps': 6604, 'batch_loss/train': 0.85841272957623} +12/23/2021 20:29:48 - INFO - codeparrot_training - Step 6605: {'lr': 0.0004797334367183885, 'samples': 3382272, 'steps': 6605, 'batch_loss/train': 0.7509722039103508} +12/23/2021 20:30:01 - INFO - codeparrot_training - Step 6606: {'lr': 0.00047972721909356165, 'samples': 3382784, 'steps': 6606, 'batch_loss/train': 0.8125764739234} +12/23/2021 20:30:12 - INFO - codeparrot_training - Step 6607: {'lr': 0.0004797210005554262, 'samples': 3383296, 'steps': 6607, 'batch_loss/train': 0.828918632119894} +12/23/2021 20:30:22 - INFO - codeparrot_training - Step 6608: {'lr': 0.0004797147811040068, 'samples': 3383808, 'steps': 6608, 'batch_loss/train': 0.8678867779672146} +12/23/2021 20:30:34 - INFO - codeparrot_training - Step 6609: {'lr': 0.0004797085607393281, 'samples': 3384320, 'steps': 6609, 'batch_loss/train': 1.0170717069413513} +12/23/2021 20:30:45 - INFO - codeparrot_training - Step 6610: {'lr': 0.00047970233946141496, 'samples': 3384832, 'steps': 6610, 'batch_loss/train': 0.8456258634105325} +12/23/2021 20:30:56 - INFO - codeparrot_training - Step 6611: {'lr': 0.000479696117270292, 'samples': 3385344, 'steps': 6611, 'batch_loss/train': 0.9679913921281695} +12/23/2021 20:31:06 - INFO - codeparrot_training - Step 6612: {'lr': 0.00047968989416598403, 'samples': 3385856, 'steps': 6612, 'batch_loss/train': 0.8808631477877498} +12/23/2021 20:31:20 - INFO - codeparrot_training - Step 6613: {'lr': 0.0004796836701485158, 'samples': 3386368, 'steps': 6613, 'batch_loss/train': 0.8501312183216214} +12/23/2021 20:31:31 - INFO - codeparrot_training - Step 6614: {'lr': 0.00047967744521791193, 'samples': 3386880, 'steps': 6614, 'batch_loss/train': 0.8296098364517093} +12/23/2021 20:31:41 - INFO - codeparrot_training - Step 6615: {'lr': 0.00047967121937419734, 'samples': 3387392, 'steps': 6615, 'batch_loss/train': 0.769038790720515} +12/23/2021 20:31:53 - INFO - codeparrot_training - Step 6616: {'lr': 0.00047966499261739673, 'samples': 3387904, 'steps': 6616, 'batch_loss/train': 0.9692985601723194} +12/23/2021 20:32:04 - INFO - codeparrot_training - Step 6617: {'lr': 0.0004796587649475347, 'samples': 3388416, 'steps': 6617, 'batch_loss/train': 0.8823113003745675} +12/23/2021 20:32:15 - INFO - codeparrot_training - Step 6618: {'lr': 0.00047965253636463625, 'samples': 3388928, 'steps': 6618, 'batch_loss/train': 0.9336572419852018} +12/23/2021 20:32:27 - INFO - codeparrot_training - Step 6619: {'lr': 0.00047964630686872604, 'samples': 3389440, 'steps': 6619, 'batch_loss/train': 0.8475145776756108} +12/23/2021 20:32:38 - INFO - codeparrot_training - Step 6620: {'lr': 0.0004796400764598288, 'samples': 3389952, 'steps': 6620, 'batch_loss/train': 0.7922201426699758} +12/23/2021 20:32:48 - INFO - codeparrot_training - Step 6621: {'lr': 0.00047963384513796924, 'samples': 3390464, 'steps': 6621, 'batch_loss/train': 0.9108885796740651} +12/23/2021 20:33:01 - INFO - codeparrot_training - Step 6622: {'lr': 0.0004796276129031723, 'samples': 3390976, 'steps': 6622, 'batch_loss/train': 0.8728324109688401} +12/23/2021 20:33:12 - INFO - codeparrot_training - Step 6623: {'lr': 0.0004796213797554626, 'samples': 3391488, 'steps': 6623, 'batch_loss/train': 0.824275222257711} +12/23/2021 20:33:23 - INFO - codeparrot_training - Step 6624: {'lr': 0.0004796151456948651, 'samples': 3392000, 'steps': 6624, 'batch_loss/train': 0.9303242638707161} +12/23/2021 20:33:35 - INFO - codeparrot_training - Step 6625: {'lr': 0.00047960891072140445, 'samples': 3392512, 'steps': 6625, 'batch_loss/train': 0.9236344583332539} +12/23/2021 20:33:45 - INFO - codeparrot_training - Step 6626: {'lr': 0.0004796026748351054, 'samples': 3393024, 'steps': 6626, 'batch_loss/train': 0.926085464656353} +12/23/2021 20:33:56 - INFO - codeparrot_training - Step 6627: {'lr': 0.0004795964380359929, 'samples': 3393536, 'steps': 6627, 'batch_loss/train': 0.946737531106919} +12/23/2021 20:34:07 - INFO - codeparrot_training - Step 6628: {'lr': 0.00047959020032409163, 'samples': 3394048, 'steps': 6628, 'batch_loss/train': 0.8760807253420353} +12/23/2021 20:34:19 - INFO - codeparrot_training - Step 6629: {'lr': 0.00047958396169942637, 'samples': 3394560, 'steps': 6629, 'batch_loss/train': 0.896924139931798} +12/23/2021 20:34:30 - INFO - codeparrot_training - Step 6630: {'lr': 0.000479577722162022, 'samples': 3395072, 'steps': 6630, 'batch_loss/train': 0.5449301741900854} +12/23/2021 20:34:40 - INFO - codeparrot_training - Step 6631: {'lr': 0.0004795714817119033, 'samples': 3395584, 'steps': 6631, 'batch_loss/train': 0.7550488223787397} +12/23/2021 20:34:53 - INFO - codeparrot_training - Step 6632: {'lr': 0.00047956524034909506, 'samples': 3396096, 'steps': 6632, 'batch_loss/train': 0.763871056959033} +12/23/2021 20:35:03 - INFO - codeparrot_training - Step 6633: {'lr': 0.0004795589980736221, 'samples': 3396608, 'steps': 6633, 'batch_loss/train': 0.8935642391443253} +12/23/2021 20:35:14 - INFO - codeparrot_training - Step 6634: {'lr': 0.0004795527548855092, 'samples': 3397120, 'steps': 6634, 'batch_loss/train': 0.8226279886439443} +12/23/2021 20:35:27 - INFO - codeparrot_training - Step 6635: {'lr': 0.0004795465107847813, 'samples': 3397632, 'steps': 6635, 'batch_loss/train': 0.7025929940864444} +12/23/2021 20:35:38 - INFO - codeparrot_training - Step 6636: {'lr': 0.00047954026577146314, 'samples': 3398144, 'steps': 6636, 'batch_loss/train': 0.9043339593335986} +12/23/2021 20:35:48 - INFO - codeparrot_training - Step 6637: {'lr': 0.00047953401984557955, 'samples': 3398656, 'steps': 6637, 'batch_loss/train': 0.9101201212033629} +12/23/2021 20:35:59 - INFO - codeparrot_training - Step 6638: {'lr': 0.00047952777300715535, 'samples': 3399168, 'steps': 6638, 'batch_loss/train': 0.8260428783250973} +12/23/2021 20:36:11 - INFO - codeparrot_training - Step 6639: {'lr': 0.0004795215252562153, 'samples': 3399680, 'steps': 6639, 'batch_loss/train': 0.8529995121061802} +12/23/2021 20:36:21 - INFO - codeparrot_training - Step 6640: {'lr': 0.00047951527659278445, 'samples': 3400192, 'steps': 6640, 'batch_loss/train': 0.8848206102848053} +12/23/2021 20:36:32 - INFO - codeparrot_training - Step 6641: {'lr': 0.00047950902701688746, 'samples': 3400704, 'steps': 6641, 'batch_loss/train': 0.8876137929037213} +12/23/2021 20:36:45 - INFO - codeparrot_training - Step 6642: {'lr': 0.0004795027765285493, 'samples': 3401216, 'steps': 6642, 'batch_loss/train': 0.8540183550212532} +12/23/2021 20:36:56 - INFO - codeparrot_training - Step 6643: {'lr': 0.00047949652512779476, 'samples': 3401728, 'steps': 6643, 'batch_loss/train': 0.8294060833286494} +12/23/2021 20:37:07 - INFO - codeparrot_training - Step 6644: {'lr': 0.00047949027281464864, 'samples': 3402240, 'steps': 6644, 'batch_loss/train': 0.7701177913695574} +12/23/2021 20:37:19 - INFO - codeparrot_training - Step 6645: {'lr': 0.0004794840195891358, 'samples': 3402752, 'steps': 6645, 'batch_loss/train': 0.7635275437496603} +12/23/2021 20:37:29 - INFO - codeparrot_training - Step 6646: {'lr': 0.00047947776545128126, 'samples': 3403264, 'steps': 6646, 'batch_loss/train': 0.8984386101365089} +12/23/2021 20:37:40 - INFO - codeparrot_training - Step 6647: {'lr': 0.0004794715104011097, 'samples': 3403776, 'steps': 6647, 'batch_loss/train': 0.8621068382635713} +12/23/2021 20:37:51 - INFO - codeparrot_training - Step 6648: {'lr': 0.0004794652544386461, 'samples': 3404288, 'steps': 6648, 'batch_loss/train': 0.7232420991640538} +12/23/2021 20:38:03 - INFO - codeparrot_training - Step 6649: {'lr': 0.00047945899756391525, 'samples': 3404800, 'steps': 6649, 'batch_loss/train': 0.9124392122030258} +12/23/2021 20:38:13 - INFO - codeparrot_training - Step 6650: {'lr': 0.0004794527397769421, 'samples': 3405312, 'steps': 6650, 'batch_loss/train': 0.8317047453019768} +12/23/2021 20:38:24 - INFO - codeparrot_training - Step 6651: {'lr': 0.00047944648107775144, 'samples': 3405824, 'steps': 6651, 'batch_loss/train': 0.8011546034831554} +12/23/2021 20:38:38 - INFO - codeparrot_training - Step 6652: {'lr': 0.0004794402214663682, 'samples': 3406336, 'steps': 6652, 'batch_loss/train': 0.9000873919576406} +12/23/2021 20:38:48 - INFO - codeparrot_training - Step 6653: {'lr': 0.00047943396094281737, 'samples': 3406848, 'steps': 6653, 'batch_loss/train': 0.8845077259466052} +12/23/2021 20:38:59 - INFO - codeparrot_training - Step 6654: {'lr': 0.0004794276995071237, 'samples': 3407360, 'steps': 6654, 'batch_loss/train': 0.9161198073998094} +12/23/2021 20:39:11 - INFO - codeparrot_training - Step 6655: {'lr': 0.0004794214371593121, 'samples': 3407872, 'steps': 6655, 'batch_loss/train': 0.7794443755410612} +12/23/2021 20:39:22 - INFO - codeparrot_training - Step 6656: {'lr': 0.00047941517389940753, 'samples': 3408384, 'steps': 6656, 'batch_loss/train': 0.7517812326550484} +12/23/2021 20:39:32 - INFO - codeparrot_training - Step 6657: {'lr': 0.00047940890972743484, 'samples': 3408896, 'steps': 6657, 'batch_loss/train': 0.8103777937358245} +12/23/2021 20:39:44 - INFO - codeparrot_training - Step 6658: {'lr': 0.00047940264464341897, 'samples': 3409408, 'steps': 6658, 'batch_loss/train': 0.7194143342785537} +12/23/2021 20:39:55 - INFO - codeparrot_training - Step 6659: {'lr': 0.0004793963786473847, 'samples': 3409920, 'steps': 6659, 'batch_loss/train': 0.8112652800045907} +12/23/2021 20:40:05 - INFO - codeparrot_training - Step 6660: {'lr': 0.0004793901117393572, 'samples': 3410432, 'steps': 6660, 'batch_loss/train': 0.7887753872200847} +12/23/2021 20:40:16 - INFO - codeparrot_training - Step 6661: {'lr': 0.0004793838439193611, 'samples': 3410944, 'steps': 6661, 'batch_loss/train': 0.8202629266306758} +12/23/2021 20:40:30 - INFO - codeparrot_training - Step 6662: {'lr': 0.0004793775751874215, 'samples': 3411456, 'steps': 6662, 'batch_loss/train': 1.1202405570074916} +12/23/2021 20:40:40 - INFO - codeparrot_training - Step 6663: {'lr': 0.0004793713055435633, 'samples': 3411968, 'steps': 6663, 'batch_loss/train': 0.8954945714212954} +12/23/2021 20:40:51 - INFO - codeparrot_training - Step 6664: {'lr': 0.00047936503498781136, 'samples': 3412480, 'steps': 6664, 'batch_loss/train': 0.8334438726305962} +12/23/2021 20:41:03 - INFO - codeparrot_training - Step 6665: {'lr': 0.0004793587635201906, 'samples': 3412992, 'steps': 6665, 'batch_loss/train': 0.8577756038866937} +12/23/2021 20:41:14 - INFO - codeparrot_training - Step 6666: {'lr': 0.0004793524911407261, 'samples': 3413504, 'steps': 6666, 'batch_loss/train': 0.9338244497776031} +12/23/2021 20:41:24 - INFO - codeparrot_training - Step 6667: {'lr': 0.00047934621784944266, 'samples': 3414016, 'steps': 6667, 'batch_loss/train': 0.7857520310208201} +12/23/2021 20:41:38 - INFO - codeparrot_training - Step 6668: {'lr': 0.0004793399436463652, 'samples': 3414528, 'steps': 6668, 'batch_loss/train': 0.8022098513320088} +12/23/2021 20:41:49 - INFO - codeparrot_training - Step 6669: {'lr': 0.00047933366853151874, 'samples': 3415040, 'steps': 6669, 'batch_loss/train': 0.9003030983731151} +12/23/2021 20:41:59 - INFO - codeparrot_training - Step 6670: {'lr': 0.00047932739250492823, 'samples': 3415552, 'steps': 6670, 'batch_loss/train': 0.8780073677189648} +12/23/2021 20:42:10 - INFO - codeparrot_training - Step 6671: {'lr': 0.0004793211155666186, 'samples': 3416064, 'steps': 6671, 'batch_loss/train': 0.8394334698095918} +12/23/2021 20:42:22 - INFO - codeparrot_training - Step 6672: {'lr': 0.00047931483771661477, 'samples': 3416576, 'steps': 6672, 'batch_loss/train': 0.9474397907033563} +12/23/2021 20:42:32 - INFO - codeparrot_training - Step 6673: {'lr': 0.0004793085589549418, 'samples': 3417088, 'steps': 6673, 'batch_loss/train': 0.8269268833100796} +12/23/2021 20:42:43 - INFO - codeparrot_training - Step 6674: {'lr': 0.0004793022792816245, 'samples': 3417600, 'steps': 6674, 'batch_loss/train': 0.8975230623036623} +12/23/2021 20:42:55 - INFO - codeparrot_training - Step 6675: {'lr': 0.000479295998696688, 'samples': 3418112, 'steps': 6675, 'batch_loss/train': 0.9879155559465289} +12/23/2021 20:43:06 - INFO - codeparrot_training - Step 6676: {'lr': 0.0004792897172001571, 'samples': 3418624, 'steps': 6676, 'batch_loss/train': 0.8829214414581656} +12/23/2021 20:43:16 - INFO - codeparrot_training - Step 6677: {'lr': 0.00047928343479205696, 'samples': 3419136, 'steps': 6677, 'batch_loss/train': 0.9424074152484536} +12/23/2021 20:43:28 - INFO - codeparrot_training - Step 6678: {'lr': 0.00047927715147241237, 'samples': 3419648, 'steps': 6678, 'batch_loss/train': 0.9066180679947138} +12/23/2021 20:43:39 - INFO - codeparrot_training - Step 6679: {'lr': 0.00047927086724124844, 'samples': 3420160, 'steps': 6679, 'batch_loss/train': 0.7890382120385766} +12/23/2021 20:43:49 - INFO - codeparrot_training - Step 6680: {'lr': 0.00047926458209859017, 'samples': 3420672, 'steps': 6680, 'batch_loss/train': 1.655387505888939} +12/23/2021 20:44:00 - INFO - codeparrot_training - Step 6681: {'lr': 0.00047925829604446237, 'samples': 3421184, 'steps': 6681, 'batch_loss/train': 0.809243296040222} +12/23/2021 20:44:14 - INFO - codeparrot_training - Step 6682: {'lr': 0.0004792520090788902, 'samples': 3421696, 'steps': 6682, 'batch_loss/train': 0.9528067009523511} +12/23/2021 20:44:25 - INFO - codeparrot_training - Step 6683: {'lr': 0.0004792457212018986, 'samples': 3422208, 'steps': 6683, 'batch_loss/train': 0.8920184532180429} +12/23/2021 20:44:35 - INFO - codeparrot_training - Step 6684: {'lr': 0.0004792394324135126, 'samples': 3422720, 'steps': 6684, 'batch_loss/train': 0.8260881472378969} +12/23/2021 20:44:47 - INFO - codeparrot_training - Step 6685: {'lr': 0.0004792331427137572, 'samples': 3423232, 'steps': 6685, 'batch_loss/train': 0.8085690503939986} +12/23/2021 20:44:58 - INFO - codeparrot_training - Step 6686: {'lr': 0.0004792268521026573, 'samples': 3423744, 'steps': 6686, 'batch_loss/train': 0.78311303909868} +12/23/2021 20:45:08 - INFO - codeparrot_training - Step 6687: {'lr': 0.000479220560580238, 'samples': 3424256, 'steps': 6687, 'batch_loss/train': 0.859293001703918} +12/23/2021 20:45:20 - INFO - codeparrot_training - Step 6688: {'lr': 0.00047921426814652435, 'samples': 3424768, 'steps': 6688, 'batch_loss/train': 0.8593164384365082} +12/23/2021 20:45:31 - INFO - codeparrot_training - Step 6689: {'lr': 0.0004792079748015413, 'samples': 3425280, 'steps': 6689, 'batch_loss/train': 0.9048238396644592} +12/23/2021 20:45:42 - INFO - codeparrot_training - Step 6690: {'lr': 0.0004792016805453139, 'samples': 3425792, 'steps': 6690, 'batch_loss/train': 0.9100741157308221} +12/23/2021 20:45:52 - INFO - codeparrot_training - Step 6691: {'lr': 0.0004791953853778671, 'samples': 3426304, 'steps': 6691, 'batch_loss/train': 0.8635801672935486} +12/23/2021 20:46:06 - INFO - codeparrot_training - Step 6692: {'lr': 0.000479189089299226, 'samples': 3426816, 'steps': 6692, 'batch_loss/train': 0.9278983566910028} +12/23/2021 20:46:16 - INFO - codeparrot_training - Step 6693: {'lr': 0.00047918279230941565, 'samples': 3427328, 'steps': 6693, 'batch_loss/train': 0.8910356042906642} +12/23/2021 20:46:27 - INFO - codeparrot_training - Step 6694: {'lr': 0.00047917649440846105, 'samples': 3427840, 'steps': 6694, 'batch_loss/train': 0.7119110734201968} +12/23/2021 20:46:39 - INFO - codeparrot_training - Step 6695: {'lr': 0.00047917019559638724, 'samples': 3428352, 'steps': 6695, 'batch_loss/train': 0.8642592700198293} +12/23/2021 20:46:50 - INFO - codeparrot_training - Step 6696: {'lr': 0.0004791638958732193, 'samples': 3428864, 'steps': 6696, 'batch_loss/train': 0.8524750911165029} +12/23/2021 20:47:00 - INFO - codeparrot_training - Step 6697: {'lr': 0.00047915759523898215, 'samples': 3429376, 'steps': 6697, 'batch_loss/train': 0.835584718734026} +12/23/2021 20:47:13 - INFO - codeparrot_training - Step 6698: {'lr': 0.000479151293693701, 'samples': 3429888, 'steps': 6698, 'batch_loss/train': 0.7797718772199005} +12/23/2021 20:47:24 - INFO - codeparrot_training - Step 6699: {'lr': 0.00047914499123740084, 'samples': 3430400, 'steps': 6699, 'batch_loss/train': 0.8321781340055168} +12/23/2021 20:47:35 - INFO - codeparrot_training - Step 6700: {'lr': 0.00047913868787010673, 'samples': 3430912, 'steps': 6700, 'batch_loss/train': 0.982667782343924} +12/23/2021 20:47:45 - INFO - codeparrot_training - Step 6701: {'lr': 0.0004791323835918436, 'samples': 3431424, 'steps': 6701, 'batch_loss/train': 0.8421556460671127} +12/23/2021 20:47:57 - INFO - codeparrot_training - Step 6702: {'lr': 0.0004791260784026368, 'samples': 3431936, 'steps': 6702, 'batch_loss/train': 0.8852901887148619} +12/23/2021 20:48:08 - INFO - codeparrot_training - Step 6703: {'lr': 0.0004791197723025111, 'samples': 3432448, 'steps': 6703, 'batch_loss/train': 0.8241078401915729} +12/23/2021 20:48:19 - INFO - codeparrot_training - Step 6704: {'lr': 0.0004791134652914917, 'samples': 3432960, 'steps': 6704, 'batch_loss/train': 0.7714668652042747} +12/23/2021 20:48:31 - INFO - codeparrot_training - Step 6705: {'lr': 0.0004791071573696038, 'samples': 3433472, 'steps': 6705, 'batch_loss/train': 0.7445695642381907} +12/23/2021 20:48:42 - INFO - codeparrot_training - Step 6706: {'lr': 0.0004791008485368722, 'samples': 3433984, 'steps': 6706, 'batch_loss/train': 0.9269942231476307} +12/23/2021 20:48:52 - INFO - codeparrot_training - Step 6707: {'lr': 0.00047909453879332214, 'samples': 3434496, 'steps': 6707, 'batch_loss/train': 0.730744605883956} +12/23/2021 20:49:05 - INFO - codeparrot_training - Step 6708: {'lr': 0.0004790882281389788, 'samples': 3435008, 'steps': 6708, 'batch_loss/train': 1.6285644238814712} +12/23/2021 20:49:16 - INFO - codeparrot_training - Step 6709: {'lr': 0.00047908191657386706, 'samples': 3435520, 'steps': 6709, 'batch_loss/train': 0.8003596235066652} +12/23/2021 20:49:26 - INFO - codeparrot_training - Step 6710: {'lr': 0.00047907560409801215, 'samples': 3436032, 'steps': 6710, 'batch_loss/train': 0.8086223844438791} +12/23/2021 20:49:37 - INFO - codeparrot_training - Step 6711: {'lr': 0.0004790692907114392, 'samples': 3436544, 'steps': 6711, 'batch_loss/train': 0.8057316560298204} +12/23/2021 20:49:49 - INFO - codeparrot_training - Step 6712: {'lr': 0.0004790629764141732, 'samples': 3437056, 'steps': 6712, 'batch_loss/train': 0.9071357706561685} +12/23/2021 20:50:00 - INFO - codeparrot_training - Step 6713: {'lr': 0.00047905666120623923, 'samples': 3437568, 'steps': 6713, 'batch_loss/train': 0.8905982195865363} +12/23/2021 20:50:11 - INFO - codeparrot_training - Step 6714: {'lr': 0.00047905034508766256, 'samples': 3438080, 'steps': 6714, 'batch_loss/train': 0.8337859492748976} +12/23/2021 20:50:23 - INFO - codeparrot_training - Step 6715: {'lr': 0.0004790440280584681, 'samples': 3438592, 'steps': 6715, 'batch_loss/train': 0.8412412768229842} +12/23/2021 20:50:33 - INFO - codeparrot_training - Step 6716: {'lr': 0.0004790377101186811, 'samples': 3439104, 'steps': 6716, 'batch_loss/train': 0.8075630851089954} +12/23/2021 20:50:44 - INFO - codeparrot_training - Step 6717: {'lr': 0.00047903139126832675, 'samples': 3439616, 'steps': 6717, 'batch_loss/train': 0.9009091155603528} +12/23/2021 20:50:56 - INFO - codeparrot_training - Step 6718: {'lr': 0.00047902507150743, 'samples': 3440128, 'steps': 6718, 'batch_loss/train': 0.7275892862817273} +12/23/2021 20:51:07 - INFO - codeparrot_training - Step 6719: {'lr': 0.00047901875083601603, 'samples': 3440640, 'steps': 6719, 'batch_loss/train': 0.8669413924217224} +12/23/2021 20:51:17 - INFO - codeparrot_training - Step 6720: {'lr': 0.00047901242925411003, 'samples': 3441152, 'steps': 6720, 'batch_loss/train': 0.7736031590029597} +12/23/2021 20:51:28 - INFO - codeparrot_training - Step 6721: {'lr': 0.000479006106761737, 'samples': 3441664, 'steps': 6721, 'batch_loss/train': 0.7984777381643653} +12/23/2021 20:51:41 - INFO - codeparrot_training - Step 6722: {'lr': 0.00047899978335892224, 'samples': 3442176, 'steps': 6722, 'batch_loss/train': 0.8448220891878009} +12/23/2021 20:51:52 - INFO - codeparrot_training - Step 6723: {'lr': 0.00047899345904569077, 'samples': 3442688, 'steps': 6723, 'batch_loss/train': 0.7651079595088959} +12/23/2021 20:52:02 - INFO - codeparrot_training - Step 6724: {'lr': 0.0004789871338220677, 'samples': 3443200, 'steps': 6724, 'batch_loss/train': 0.7777837186586112} +12/23/2021 20:52:14 - INFO - codeparrot_training - Step 6725: {'lr': 0.00047898080768807836, 'samples': 3443712, 'steps': 6725, 'batch_loss/train': 0.8633901653811336} +12/23/2021 20:52:25 - INFO - codeparrot_training - Step 6726: {'lr': 0.0004789744806437477, 'samples': 3444224, 'steps': 6726, 'batch_loss/train': 0.7009223294444382} +12/23/2021 20:52:35 - INFO - codeparrot_training - Step 6727: {'lr': 0.0004789681526891011, 'samples': 3444736, 'steps': 6727, 'batch_loss/train': 0.8816190073266625} +12/23/2021 20:52:49 - INFO - codeparrot_training - Step 6728: {'lr': 0.0004789618238241634, 'samples': 3445248, 'steps': 6728, 'batch_loss/train': 0.8198448829352856} +12/23/2021 20:53:00 - INFO - codeparrot_training - Step 6729: {'lr': 0.0004789554940489601, 'samples': 3445760, 'steps': 6729, 'batch_loss/train': 1.0520649617537856} +12/23/2021 20:53:10 - INFO - codeparrot_training - Step 6730: {'lr': 0.0004789491633635161, 'samples': 3446272, 'steps': 6730, 'batch_loss/train': 0.8346819328144193} +12/23/2021 20:53:23 - INFO - codeparrot_training - Step 6731: {'lr': 0.00047894283176785674, 'samples': 3446784, 'steps': 6731, 'batch_loss/train': 0.8742049774155021} +12/23/2021 20:53:33 - INFO - codeparrot_training - Step 6732: {'lr': 0.0004789364992620071, 'samples': 3447296, 'steps': 6732, 'batch_loss/train': 1.6327752680517733} +12/23/2021 20:53:44 - INFO - codeparrot_training - Step 6733: {'lr': 0.00047893016584599235, 'samples': 3447808, 'steps': 6733, 'batch_loss/train': 0.7721715164370835} +12/23/2021 20:53:55 - INFO - codeparrot_training - Step 6734: {'lr': 0.00047892383151983776, 'samples': 3448320, 'steps': 6734, 'batch_loss/train': 0.7850738032720983} +12/23/2021 20:54:07 - INFO - codeparrot_training - Step 6735: {'lr': 0.0004789174962835684, 'samples': 3448832, 'steps': 6735, 'batch_loss/train': 0.8947492763400078} +12/23/2021 20:54:18 - INFO - codeparrot_training - Step 6736: {'lr': 0.0004789111601372096, 'samples': 3449344, 'steps': 6736, 'batch_loss/train': 0.9636008236557245} +12/23/2021 20:54:28 - INFO - codeparrot_training - Step 6737: {'lr': 0.0004789048230807863, 'samples': 3449856, 'steps': 6737, 'batch_loss/train': 0.8673527650535107} +12/23/2021 20:54:41 - INFO - codeparrot_training - Step 6738: {'lr': 0.000478898485114324, 'samples': 3450368, 'steps': 6738, 'batch_loss/train': 0.6305848130141385} +12/23/2021 20:54:52 - INFO - codeparrot_training - Step 6739: {'lr': 0.0004788921462378477, 'samples': 3450880, 'steps': 6739, 'batch_loss/train': 0.9109242632985115} +12/23/2021 20:55:02 - INFO - codeparrot_training - Step 6740: {'lr': 0.0004788858064513827, 'samples': 3451392, 'steps': 6740, 'batch_loss/train': 0.7358981855213642} +12/23/2021 20:55:15 - INFO - codeparrot_training - Step 6741: {'lr': 0.00047887946575495406, 'samples': 3451904, 'steps': 6741, 'batch_loss/train': 1.0021455520763993} +12/23/2021 20:55:25 - INFO - codeparrot_training - Step 6742: {'lr': 0.00047887312414858713, 'samples': 3452416, 'steps': 6742, 'batch_loss/train': 0.8687023911625147} +12/23/2021 20:55:36 - INFO - codeparrot_training - Step 6743: {'lr': 0.0004788667816323071, 'samples': 3452928, 'steps': 6743, 'batch_loss/train': 1.0248452108353376} +12/23/2021 20:55:46 - INFO - codeparrot_training - Step 6744: {'lr': 0.00047886043820613913, 'samples': 3453440, 'steps': 6744, 'batch_loss/train': 0.8421798916533589} +12/23/2021 20:55:58 - INFO - codeparrot_training - Step 6745: {'lr': 0.0004788540938701085, 'samples': 3453952, 'steps': 6745, 'batch_loss/train': 0.9436667738482356} +12/23/2021 20:56:09 - INFO - codeparrot_training - Step 6746: {'lr': 0.0004788477486242404, 'samples': 3454464, 'steps': 6746, 'batch_loss/train': 0.86714340839535} +12/23/2021 20:56:20 - INFO - codeparrot_training - Step 6747: {'lr': 0.00047884140246856014, 'samples': 3454976, 'steps': 6747, 'batch_loss/train': 0.9236221835017204} +12/23/2021 20:56:33 - INFO - codeparrot_training - Step 6748: {'lr': 0.0004788350554030927, 'samples': 3455488, 'steps': 6748, 'batch_loss/train': 0.9481103112921119} +12/23/2021 20:56:43 - INFO - codeparrot_training - Step 6749: {'lr': 0.0004788287074278636, 'samples': 3456000, 'steps': 6749, 'batch_loss/train': 0.8813913967460394} +12/23/2021 20:56:54 - INFO - codeparrot_training - Step 6750: {'lr': 0.0004788223585428979, 'samples': 3456512, 'steps': 6750, 'batch_loss/train': 0.8395000696182251} +12/23/2021 20:57:06 - INFO - codeparrot_training - Step 6751: {'lr': 0.00047881600874822097, 'samples': 3457024, 'steps': 6751, 'batch_loss/train': 0.8946070903912187} +12/23/2021 20:57:17 - INFO - codeparrot_training - Step 6752: {'lr': 0.00047880965804385797, 'samples': 3457536, 'steps': 6752, 'batch_loss/train': 0.9072881620377302} +12/23/2021 20:57:27 - INFO - codeparrot_training - Step 6753: {'lr': 0.0004788033064298342, 'samples': 3458048, 'steps': 6753, 'batch_loss/train': 0.8117232285439968} +12/23/2021 20:57:38 - INFO - codeparrot_training - Step 6754: {'lr': 0.00047879695390617473, 'samples': 3458560, 'steps': 6754, 'batch_loss/train': 0.8055873294360936} +12/23/2021 20:57:51 - INFO - codeparrot_training - Step 6755: {'lr': 0.00047879060047290506, 'samples': 3459072, 'steps': 6755, 'batch_loss/train': 0.8941704761236906} +12/23/2021 20:58:01 - INFO - codeparrot_training - Step 6756: {'lr': 0.0004787842461300503, 'samples': 3459584, 'steps': 6756, 'batch_loss/train': 0.9083206569775939} +12/23/2021 20:58:12 - INFO - codeparrot_training - Step 6757: {'lr': 0.0004787778908776359, 'samples': 3460096, 'steps': 6757, 'batch_loss/train': 0.7625836357474327} +12/23/2021 20:58:25 - INFO - codeparrot_training - Step 6758: {'lr': 0.0004787715347156869, 'samples': 3460608, 'steps': 6758, 'batch_loss/train': 0.8206991897895932} +12/23/2021 20:58:36 - INFO - codeparrot_training - Step 6759: {'lr': 0.0004787651776442287, 'samples': 3461120, 'steps': 6759, 'batch_loss/train': 0.9092725683003664} +12/23/2021 20:58:47 - INFO - codeparrot_training - Step 6760: {'lr': 0.00047875881966328644, 'samples': 3461632, 'steps': 6760, 'batch_loss/train': 0.9716774942353368} +12/23/2021 20:58:58 - INFO - codeparrot_training - Step 6761: {'lr': 0.00047875246077288554, 'samples': 3462144, 'steps': 6761, 'batch_loss/train': 0.8823424505535513} +12/23/2021 20:59:09 - INFO - codeparrot_training - Step 6762: {'lr': 0.0004787461009730512, 'samples': 3462656, 'steps': 6762, 'batch_loss/train': 0.8596898671239614} +12/23/2021 20:59:20 - INFO - codeparrot_training - Step 6763: {'lr': 0.00047873974026380875, 'samples': 3463168, 'steps': 6763, 'batch_loss/train': 0.7984058540314436} +12/23/2021 20:59:30 - INFO - codeparrot_training - Step 6764: {'lr': 0.0004787333786451835, 'samples': 3463680, 'steps': 6764, 'batch_loss/train': 0.7721211379393935} +12/23/2021 20:59:42 - INFO - codeparrot_training - Step 6765: {'lr': 0.0004787270161172007, 'samples': 3464192, 'steps': 6765, 'batch_loss/train': 0.8740882845595479} +12/23/2021 20:59:53 - INFO - codeparrot_training - Step 6766: {'lr': 0.00047872065267988564, 'samples': 3464704, 'steps': 6766, 'batch_loss/train': 0.8015094147995114} +12/23/2021 21:00:04 - INFO - codeparrot_training - Step 6767: {'lr': 0.0004787142883332636, 'samples': 3465216, 'steps': 6767, 'batch_loss/train': 0.8679456375539303} +12/23/2021 21:00:17 - INFO - codeparrot_training - Step 6768: {'lr': 0.00047870792307735987, 'samples': 3465728, 'steps': 6768, 'batch_loss/train': 0.9637270905077457} +12/23/2021 21:00:28 - INFO - codeparrot_training - Step 6769: {'lr': 0.00047870155691219984, 'samples': 3466240, 'steps': 6769, 'batch_loss/train': 0.7353827222250402} +12/23/2021 21:00:38 - INFO - codeparrot_training - Step 6770: {'lr': 0.0004786951898378088, 'samples': 3466752, 'steps': 6770, 'batch_loss/train': 0.7076111203059554} +12/23/2021 21:00:50 - INFO - codeparrot_training - Step 6771: {'lr': 0.000478688821854212, 'samples': 3467264, 'steps': 6771, 'batch_loss/train': 0.9769663270562887} +12/23/2021 21:01:01 - INFO - codeparrot_training - Step 6772: {'lr': 0.0004786824529614349, 'samples': 3467776, 'steps': 6772, 'batch_loss/train': 0.9124030778184533} +12/23/2021 21:01:12 - INFO - codeparrot_training - Step 6773: {'lr': 0.0004786760831595025, 'samples': 3468288, 'steps': 6773, 'batch_loss/train': 0.8351344226393849} +12/23/2021 21:01:24 - INFO - codeparrot_training - Step 6774: {'lr': 0.0004786697124484405, 'samples': 3468800, 'steps': 6774, 'batch_loss/train': 0.8159699346870184} +12/23/2021 21:01:34 - INFO - codeparrot_training - Step 6775: {'lr': 0.000478663340828274, 'samples': 3469312, 'steps': 6775, 'batch_loss/train': 0.9002412529662251} +12/23/2021 21:01:45 - INFO - codeparrot_training - Step 6776: {'lr': 0.0004786569682990285, 'samples': 3469824, 'steps': 6776, 'batch_loss/train': 0.8576286644674838} +12/23/2021 21:01:55 - INFO - codeparrot_training - Step 6777: {'lr': 0.00047865059486072915, 'samples': 3470336, 'steps': 6777, 'batch_loss/train': 0.8671825360506773} +12/23/2021 21:02:09 - INFO - codeparrot_training - Step 6778: {'lr': 0.00047864422051340134, 'samples': 3470848, 'steps': 6778, 'batch_loss/train': 0.8635653387755156} +12/23/2021 21:02:19 - INFO - codeparrot_training - Step 6779: {'lr': 0.0004786378452570704, 'samples': 3471360, 'steps': 6779, 'batch_loss/train': 0.8978708039503545} +12/23/2021 21:02:30 - INFO - codeparrot_training - Step 6780: {'lr': 0.0004786314690917619, 'samples': 3471872, 'steps': 6780, 'batch_loss/train': 0.8652812903746963} +12/23/2021 21:02:42 - INFO - codeparrot_training - Step 6781: {'lr': 0.00047862509201750087, 'samples': 3472384, 'steps': 6781, 'batch_loss/train': 0.7703868071548641} +12/23/2021 21:02:52 - INFO - codeparrot_training - Step 6782: {'lr': 0.00047861871403431284, 'samples': 3472896, 'steps': 6782, 'batch_loss/train': 0.8925214176997542} +12/23/2021 21:03:03 - INFO - codeparrot_training - Step 6783: {'lr': 0.0004786123351422231, 'samples': 3473408, 'steps': 6783, 'batch_loss/train': 0.888842673972249} +12/23/2021 21:03:16 - INFO - codeparrot_training - Step 6784: {'lr': 0.00047860595534125706, 'samples': 3473920, 'steps': 6784, 'batch_loss/train': 0.9059709962457418} +12/23/2021 21:03:27 - INFO - codeparrot_training - Step 6785: {'lr': 0.0004785995746314401, 'samples': 3474432, 'steps': 6785, 'batch_loss/train': 0.8947339355945587} +12/23/2021 21:03:37 - INFO - codeparrot_training - Step 6786: {'lr': 0.00047859319301279745, 'samples': 3474944, 'steps': 6786, 'batch_loss/train': 0.854473166167736} +12/23/2021 21:03:48 - INFO - codeparrot_training - Step 6787: {'lr': 0.00047858681048535464, 'samples': 3475456, 'steps': 6787, 'batch_loss/train': 0.8533919500187039} +12/23/2021 21:04:00 - INFO - codeparrot_training - Step 6788: {'lr': 0.000478580427049137, 'samples': 3475968, 'steps': 6788, 'batch_loss/train': 0.9810123527422547} +12/23/2021 21:04:11 - INFO - codeparrot_training - Step 6789: {'lr': 0.00047857404270416985, 'samples': 3476480, 'steps': 6789, 'batch_loss/train': 0.8880990985780954} +12/23/2021 21:04:21 - INFO - codeparrot_training - Step 6790: {'lr': 0.0004785676574504786, 'samples': 3476992, 'steps': 6790, 'batch_loss/train': 0.8497190391644835} +12/23/2021 21:04:33 - INFO - codeparrot_training - Step 6791: {'lr': 0.0004785612712880887, 'samples': 3477504, 'steps': 6791, 'batch_loss/train': 0.8764082421548665} +12/23/2021 21:04:44 - INFO - codeparrot_training - Step 6792: {'lr': 0.0004785548842170254, 'samples': 3478016, 'steps': 6792, 'batch_loss/train': 0.8846822972409427} +12/23/2021 21:04:55 - INFO - codeparrot_training - Step 6793: {'lr': 0.0004785484962373143, 'samples': 3478528, 'steps': 6793, 'batch_loss/train': 1.060983944684267} +12/23/2021 21:05:07 - INFO - codeparrot_training - Step 6794: {'lr': 0.0004785421073489806, 'samples': 3479040, 'steps': 6794, 'batch_loss/train': 0.7613816782832146} +12/23/2021 21:05:17 - INFO - codeparrot_training - Step 6795: {'lr': 0.00047853571755204977, 'samples': 3479552, 'steps': 6795, 'batch_loss/train': 0.8572881268337369} +12/23/2021 21:05:28 - INFO - codeparrot_training - Step 6796: {'lr': 0.0004785293268465473, 'samples': 3480064, 'steps': 6796, 'batch_loss/train': 0.880180150270462} +12/23/2021 21:05:39 - INFO - codeparrot_training - Step 6797: {'lr': 0.0004785229352324984, 'samples': 3480576, 'steps': 6797, 'batch_loss/train': 0.8172860054764897} +12/23/2021 21:05:52 - INFO - codeparrot_training - Step 6798: {'lr': 0.0004785165427099286, 'samples': 3481088, 'steps': 6798, 'batch_loss/train': 0.8280820790678263} +12/23/2021 21:06:03 - INFO - codeparrot_training - Step 6799: {'lr': 0.0004785101492788634, 'samples': 3481600, 'steps': 6799, 'batch_loss/train': 0.907324830070138} +12/23/2021 21:06:13 - INFO - codeparrot_training - Step 6800: {'lr': 0.0004785037549393281, 'samples': 3482112, 'steps': 6800, 'batch_loss/train': 0.7586327223107219} +12/23/2021 21:06:25 - INFO - codeparrot_training - Step 6801: {'lr': 0.0004784973596913481, 'samples': 3482624, 'steps': 6801, 'batch_loss/train': 0.9165416769683361} +12/23/2021 21:06:36 - INFO - codeparrot_training - Step 6802: {'lr': 0.0004784909635349489, 'samples': 3483136, 'steps': 6802, 'batch_loss/train': 0.884535956196487} +12/23/2021 21:06:46 - INFO - codeparrot_training - Step 6803: {'lr': 0.000478484566470156, 'samples': 3483648, 'steps': 6803, 'batch_loss/train': 0.8315873481333256} +12/23/2021 21:06:58 - INFO - codeparrot_training - Step 6804: {'lr': 0.0004784781684969947, 'samples': 3484160, 'steps': 6804, 'batch_loss/train': 0.9270602986216545} +12/23/2021 21:07:09 - INFO - codeparrot_training - Step 6805: {'lr': 0.0004784717696154903, 'samples': 3484672, 'steps': 6805, 'batch_loss/train': 0.8351595769636333} +12/23/2021 21:07:19 - INFO - codeparrot_training - Step 6806: {'lr': 0.0004784653698256686, 'samples': 3485184, 'steps': 6806, 'batch_loss/train': 0.8327388130128384} +12/23/2021 21:07:30 - INFO - codeparrot_training - Step 6807: {'lr': 0.00047845896912755476, 'samples': 3485696, 'steps': 6807, 'batch_loss/train': 0.7512462181039155} +12/23/2021 21:07:43 - INFO - codeparrot_training - Step 6808: {'lr': 0.0004784525675211744, 'samples': 3486208, 'steps': 6808, 'batch_loss/train': 0.9116589166224003} +12/23/2021 21:07:54 - INFO - codeparrot_training - Step 6809: {'lr': 0.0004784461650065528, 'samples': 3486720, 'steps': 6809, 'batch_loss/train': 0.9086630614474416} +12/23/2021 21:08:04 - INFO - codeparrot_training - Step 6810: {'lr': 0.0004784397615837156, 'samples': 3487232, 'steps': 6810, 'batch_loss/train': 0.9504419374279678} +12/23/2021 21:08:17 - INFO - codeparrot_training - Step 6811: {'lr': 0.0004784333572526881, 'samples': 3487744, 'steps': 6811, 'batch_loss/train': 0.8582804100587964} +12/23/2021 21:08:27 - INFO - codeparrot_training - Step 6812: {'lr': 0.0004784269520134958, 'samples': 3488256, 'steps': 6812, 'batch_loss/train': 0.7246524328365922} +12/23/2021 21:08:38 - INFO - codeparrot_training - Step 6813: {'lr': 0.00047842054586616424, 'samples': 3488768, 'steps': 6813, 'batch_loss/train': 0.9155134812463075} +12/23/2021 21:08:51 - INFO - codeparrot_training - Step 6814: {'lr': 0.0004784141388107188, 'samples': 3489280, 'steps': 6814, 'batch_loss/train': 0.7727695470675826} +12/23/2021 21:09:02 - INFO - codeparrot_training - Step 6815: {'lr': 0.00047840773084718505, 'samples': 3489792, 'steps': 6815, 'batch_loss/train': 0.9048065398819745} +12/23/2021 21:09:12 - INFO - codeparrot_training - Step 6816: {'lr': 0.0004784013219755884, 'samples': 3490304, 'steps': 6816, 'batch_loss/train': 0.8989137066528201} +12/23/2021 21:09:23 - INFO - codeparrot_training - Step 6817: {'lr': 0.00047839491219595434, 'samples': 3490816, 'steps': 6817, 'batch_loss/train': 0.8360137981362641} +12/23/2021 21:09:35 - INFO - codeparrot_training - Step 6818: {'lr': 0.00047838850150830824, 'samples': 3491328, 'steps': 6818, 'batch_loss/train': 0.7997170654125512} +12/23/2021 21:09:45 - INFO - codeparrot_training - Step 6819: {'lr': 0.0004783820899126758, 'samples': 3491840, 'steps': 6819, 'batch_loss/train': 0.8546549053862691} +12/23/2021 21:09:56 - INFO - codeparrot_training - Step 6820: {'lr': 0.0004783756774090824, 'samples': 3492352, 'steps': 6820, 'batch_loss/train': 0.8413745132274926} +12/23/2021 21:10:08 - INFO - codeparrot_training - Step 6821: {'lr': 0.0004783692639975535, 'samples': 3492864, 'steps': 6821, 'batch_loss/train': 0.9242864809930325} +12/23/2021 21:10:19 - INFO - codeparrot_training - Step 6822: {'lr': 0.0004783628496781146, 'samples': 3493376, 'steps': 6822, 'batch_loss/train': 0.7881619688123465} +12/23/2021 21:10:29 - INFO - codeparrot_training - Step 6823: {'lr': 0.0004783564344507913, 'samples': 3493888, 'steps': 6823, 'batch_loss/train': 0.8427679110318422} +12/23/2021 21:10:42 - INFO - codeparrot_training - Step 6824: {'lr': 0.000478350018315609, 'samples': 3494400, 'steps': 6824, 'batch_loss/train': 0.9608896728605032} +12/23/2021 21:10:53 - INFO - codeparrot_training - Step 6825: {'lr': 0.0004783436012725933, 'samples': 3494912, 'steps': 6825, 'batch_loss/train': 0.8207084964960814} +12/23/2021 21:11:04 - INFO - codeparrot_training - Step 6826: {'lr': 0.00047833718332176967, 'samples': 3495424, 'steps': 6826, 'batch_loss/train': 0.8036062577739358} +12/23/2021 21:11:16 - INFO - codeparrot_training - Step 6827: {'lr': 0.0004783307644631635, 'samples': 3495936, 'steps': 6827, 'batch_loss/train': 0.8489548121578991} +12/23/2021 21:11:27 - INFO - codeparrot_training - Step 6828: {'lr': 0.00047832434469680053, 'samples': 3496448, 'steps': 6828, 'batch_loss/train': 0.8474846272729337} +12/23/2021 21:11:37 - INFO - codeparrot_training - Step 6829: {'lr': 0.0004783179240227062, 'samples': 3496960, 'steps': 6829, 'batch_loss/train': 0.8332756608724594} +12/23/2021 21:11:48 - INFO - codeparrot_training - Step 6830: {'lr': 0.0004783115024409059, 'samples': 3497472, 'steps': 6830, 'batch_loss/train': 0.9006317872554064} +12/23/2021 21:12:00 - INFO - codeparrot_training - Step 6831: {'lr': 0.00047830507995142536, 'samples': 3497984, 'steps': 6831, 'batch_loss/train': 1.1406070683151484} +12/23/2021 21:12:10 - INFO - codeparrot_training - Step 6832: {'lr': 0.00047829865655429004, 'samples': 3498496, 'steps': 6832, 'batch_loss/train': 0.7750192121602595} +12/23/2021 21:12:21 - INFO - codeparrot_training - Step 6833: {'lr': 0.00047829223224952545, 'samples': 3499008, 'steps': 6833, 'batch_loss/train': 0.8441748600453138} +12/23/2021 21:12:33 - INFO - codeparrot_training - Step 6834: {'lr': 0.00047828580703715705, 'samples': 3499520, 'steps': 6834, 'batch_loss/train': 0.8680101870559156} +12/23/2021 21:12:44 - INFO - codeparrot_training - Step 6835: {'lr': 0.00047827938091721055, 'samples': 3500032, 'steps': 6835, 'batch_loss/train': 0.8360354145988822} +12/23/2021 21:12:54 - INFO - codeparrot_training - Step 6836: {'lr': 0.0004782729538897115, 'samples': 3500544, 'steps': 6836, 'batch_loss/train': 0.8310494599863887} +12/23/2021 21:13:07 - INFO - codeparrot_training - Step 6837: {'lr': 0.00047826652595468527, 'samples': 3501056, 'steps': 6837, 'batch_loss/train': 0.7814680333249271} +12/23/2021 21:13:18 - INFO - codeparrot_training - Step 6838: {'lr': 0.00047826009711215755, 'samples': 3501568, 'steps': 6838, 'batch_loss/train': 0.7618309729732573} +12/23/2021 21:13:28 - INFO - codeparrot_training - Step 6839: {'lr': 0.0004782536673621539, 'samples': 3502080, 'steps': 6839, 'batch_loss/train': 0.8725486984476447} +12/23/2021 21:13:39 - INFO - codeparrot_training - Step 6840: {'lr': 0.00047824723670469984, 'samples': 3502592, 'steps': 6840, 'batch_loss/train': 0.8565727192908525} +12/23/2021 21:13:51 - INFO - codeparrot_training - Step 6841: {'lr': 0.0004782408051398209, 'samples': 3503104, 'steps': 6841, 'batch_loss/train': 0.910105497110635} +12/23/2021 21:14:02 - INFO - codeparrot_training - Step 6842: {'lr': 0.0004782343726675428, 'samples': 3503616, 'steps': 6842, 'batch_loss/train': 0.8011477636173368} +12/23/2021 21:14:12 - INFO - codeparrot_training - Step 6843: {'lr': 0.0004782279392878909, 'samples': 3504128, 'steps': 6843, 'batch_loss/train': 0.8945971010252833} +12/23/2021 21:14:26 - INFO - codeparrot_training - Step 6844: {'lr': 0.000478221505000891, 'samples': 3504640, 'steps': 6844, 'batch_loss/train': 0.8794456906616688} +12/23/2021 21:14:37 - INFO - codeparrot_training - Step 6845: {'lr': 0.0004782150698065685, 'samples': 3505152, 'steps': 6845, 'batch_loss/train': 1.2586653204634786} +12/23/2021 21:14:47 - INFO - codeparrot_training - Step 6846: {'lr': 0.00047820863370494907, 'samples': 3505664, 'steps': 6846, 'batch_loss/train': 0.7950726374983788} +12/23/2021 21:14:59 - INFO - codeparrot_training - Step 6847: {'lr': 0.00047820219669605826, 'samples': 3506176, 'steps': 6847, 'batch_loss/train': 0.896015883423388} +12/23/2021 21:15:10 - INFO - codeparrot_training - Step 6848: {'lr': 0.0004781957587799217, 'samples': 3506688, 'steps': 6848, 'batch_loss/train': 0.8924163747578859} +12/23/2021 21:15:21 - INFO - codeparrot_training - Step 6849: {'lr': 0.00047818931995656494, 'samples': 3507200, 'steps': 6849, 'batch_loss/train': 0.7634015390649438} +12/23/2021 21:15:31 - INFO - codeparrot_training - Step 6850: {'lr': 0.0004781828802260136, 'samples': 3507712, 'steps': 6850, 'batch_loss/train': 0.7579085901379585} +12/23/2021 21:15:43 - INFO - codeparrot_training - Step 6851: {'lr': 0.00047817643958829336, 'samples': 3508224, 'steps': 6851, 'batch_loss/train': 0.8636228758841753} +12/23/2021 21:15:54 - INFO - codeparrot_training - Step 6852: {'lr': 0.00047816999804342977, 'samples': 3508736, 'steps': 6852, 'batch_loss/train': 0.9168087225407362} +12/23/2021 21:16:04 - INFO - codeparrot_training - Step 6853: {'lr': 0.0004781635555914483, 'samples': 3509248, 'steps': 6853, 'batch_loss/train': 0.8644823590293527} +12/23/2021 21:16:18 - INFO - codeparrot_training - Step 6854: {'lr': 0.0004781571122323748, 'samples': 3509760, 'steps': 6854, 'batch_loss/train': 0.8179903533309698} +12/23/2021 21:16:28 - INFO - codeparrot_training - Step 6855: {'lr': 0.0004781506679662346, 'samples': 3510272, 'steps': 6855, 'batch_loss/train': 0.9181501027196646} +12/23/2021 21:16:39 - INFO - codeparrot_training - Step 6856: {'lr': 0.0004781442227930537, 'samples': 3510784, 'steps': 6856, 'batch_loss/train': 0.8546944269910455} +12/23/2021 21:16:51 - INFO - codeparrot_training - Step 6857: {'lr': 0.0004781377767128574, 'samples': 3511296, 'steps': 6857, 'batch_loss/train': 0.8204962192103267} +12/23/2021 21:17:01 - INFO - codeparrot_training - Step 6858: {'lr': 0.00047813132972567143, 'samples': 3511808, 'steps': 6858, 'batch_loss/train': 0.8043594378978014} +12/23/2021 21:17:12 - INFO - codeparrot_training - Step 6859: {'lr': 0.00047812488183152145, 'samples': 3512320, 'steps': 6859, 'batch_loss/train': 1.026729678735137} +12/23/2021 21:17:23 - INFO - codeparrot_training - Step 6860: {'lr': 0.0004781184330304331, 'samples': 3512832, 'steps': 6860, 'batch_loss/train': 0.8216695245355368} +12/23/2021 21:17:36 - INFO - codeparrot_training - Step 6861: {'lr': 0.000478111983322432, 'samples': 3513344, 'steps': 6861, 'batch_loss/train': 0.8799852337688208} +12/23/2021 21:17:46 - INFO - codeparrot_training - Step 6862: {'lr': 0.00047810553270754376, 'samples': 3513856, 'steps': 6862, 'batch_loss/train': 0.8247129013761878} +12/23/2021 21:17:57 - INFO - codeparrot_training - Step 6863: {'lr': 0.00047809908118579404, 'samples': 3514368, 'steps': 6863, 'batch_loss/train': 0.8491076836362481} +12/23/2021 21:18:09 - INFO - codeparrot_training - Step 6864: {'lr': 0.00047809262875720863, 'samples': 3514880, 'steps': 6864, 'batch_loss/train': 0.8476924812421203} +12/23/2021 21:18:19 - INFO - codeparrot_training - Step 6865: {'lr': 0.0004780861754218129, 'samples': 3515392, 'steps': 6865, 'batch_loss/train': 0.8137873122468591} +12/23/2021 21:18:30 - INFO - codeparrot_training - Step 6866: {'lr': 0.0004780797211796327, 'samples': 3515904, 'steps': 6866, 'batch_loss/train': 0.8919210112653673} +12/23/2021 21:18:42 - INFO - codeparrot_training - Step 6867: {'lr': 0.0004780732660306936, 'samples': 3516416, 'steps': 6867, 'batch_loss/train': 0.844324897043407} +12/23/2021 21:18:53 - INFO - codeparrot_training - Step 6868: {'lr': 0.0004780668099750214, 'samples': 3516928, 'steps': 6868, 'batch_loss/train': 1.444854022236541} +12/23/2021 21:19:03 - INFO - codeparrot_training - Step 6869: {'lr': 0.0004780603530126416, 'samples': 3517440, 'steps': 6869, 'batch_loss/train': 0.8653070954605937} +12/23/2021 21:19:15 - INFO - codeparrot_training - Step 6870: {'lr': 0.00047805389514358, 'samples': 3517952, 'steps': 6870, 'batch_loss/train': 0.8950120368972421} +12/23/2021 21:19:26 - INFO - codeparrot_training - Step 6871: {'lr': 0.00047804743636786225, 'samples': 3518464, 'steps': 6871, 'batch_loss/train': 0.7418174126651138} +12/23/2021 21:19:37 - INFO - codeparrot_training - Step 6872: {'lr': 0.00047804097668551393, 'samples': 3518976, 'steps': 6872, 'batch_loss/train': 0.800217701587826} +12/23/2021 21:19:47 - INFO - codeparrot_training - Step 6873: {'lr': 0.00047803451609656084, 'samples': 3519488, 'steps': 6873, 'batch_loss/train': 0.9161938698962331} +12/23/2021 21:20:01 - INFO - codeparrot_training - Step 6874: {'lr': 0.0004780280546010286, 'samples': 3520000, 'steps': 6874, 'batch_loss/train': 0.8701260765083134} +12/23/2021 21:20:12 - INFO - codeparrot_training - Step 6875: {'lr': 0.0004780215921989429, 'samples': 3520512, 'steps': 6875, 'batch_loss/train': 0.8756373533979058} +12/23/2021 21:20:22 - INFO - codeparrot_training - Step 6876: {'lr': 0.0004780151288903295, 'samples': 3521024, 'steps': 6876, 'batch_loss/train': 0.9333141464740038} +12/23/2021 21:20:34 - INFO - codeparrot_training - Step 6877: {'lr': 0.000478008664675214, 'samples': 3521536, 'steps': 6877, 'batch_loss/train': 0.8050722442567348} +12/23/2021 21:20:45 - INFO - codeparrot_training - Step 6878: {'lr': 0.0004780021995536221, 'samples': 3522048, 'steps': 6878, 'batch_loss/train': 0.773861629422754} +12/23/2021 21:20:55 - INFO - codeparrot_training - Step 6879: {'lr': 0.0004779957335255796, 'samples': 3522560, 'steps': 6879, 'batch_loss/train': 0.7964892415329814} +12/23/2021 21:21:07 - INFO - codeparrot_training - Step 6880: {'lr': 0.0004779892665911122, 'samples': 3523072, 'steps': 6880, 'batch_loss/train': 0.8098608902655542} +12/23/2021 21:21:18 - INFO - codeparrot_training - Step 6881: {'lr': 0.0004779827987502455, 'samples': 3523584, 'steps': 6881, 'batch_loss/train': 0.8478281069546938} +12/23/2021 21:21:29 - INFO - codeparrot_training - Step 6882: {'lr': 0.00047797633000300525, 'samples': 3524096, 'steps': 6882, 'batch_loss/train': 0.8418682394549251} +12/23/2021 21:21:39 - INFO - codeparrot_training - Step 6883: {'lr': 0.00047796986034941727, 'samples': 3524608, 'steps': 6883, 'batch_loss/train': 0.795907424762845} +12/23/2021 21:21:52 - INFO - codeparrot_training - Step 6884: {'lr': 0.00047796338978950707, 'samples': 3525120, 'steps': 6884, 'batch_loss/train': 0.8303869580850005} +12/23/2021 21:22:03 - INFO - codeparrot_training - Step 6885: {'lr': 0.0004779569183233006, 'samples': 3525632, 'steps': 6885, 'batch_loss/train': 0.9259656025096774} +12/23/2021 21:22:14 - INFO - codeparrot_training - Step 6886: {'lr': 0.0004779504459508235, 'samples': 3526144, 'steps': 6886, 'batch_loss/train': 0.7305252196965739} +12/23/2021 21:22:26 - INFO - codeparrot_training - Step 6887: {'lr': 0.0004779439726721015, 'samples': 3526656, 'steps': 6887, 'batch_loss/train': 0.8586296564899385} +12/23/2021 21:22:36 - INFO - codeparrot_training - Step 6888: {'lr': 0.0004779374984871603, 'samples': 3527168, 'steps': 6888, 'batch_loss/train': 0.8818472079001367} +12/23/2021 21:22:47 - INFO - codeparrot_training - Step 6889: {'lr': 0.0004779310233960257, 'samples': 3527680, 'steps': 6889, 'batch_loss/train': 0.9410274475812912} +12/23/2021 21:22:59 - INFO - codeparrot_training - Step 6890: {'lr': 0.0004779245473987234, 'samples': 3528192, 'steps': 6890, 'batch_loss/train': 0.8087455299682915} +12/23/2021 21:23:09 - INFO - codeparrot_training - Step 6891: {'lr': 0.0004779180704952792, 'samples': 3528704, 'steps': 6891, 'batch_loss/train': 0.9342662808485329} +12/23/2021 21:23:20 - INFO - codeparrot_training - Step 6892: {'lr': 0.0004779115926857188, 'samples': 3529216, 'steps': 6892, 'batch_loss/train': 0.9096384616568685} +12/23/2021 21:23:33 - INFO - codeparrot_training - Step 6893: {'lr': 0.0004779051139700679, 'samples': 3529728, 'steps': 6893, 'batch_loss/train': 0.850265197455883} +12/23/2021 21:23:44 - INFO - codeparrot_training - Step 6894: {'lr': 0.00047789863434835237, 'samples': 3530240, 'steps': 6894, 'batch_loss/train': 0.7936879615299404} +12/23/2021 21:23:55 - INFO - codeparrot_training - Step 6895: {'lr': 0.0004778921538205979, 'samples': 3530752, 'steps': 6895, 'batch_loss/train': 0.7766846921294928} +12/23/2021 21:24:05 - INFO - codeparrot_training - Step 6896: {'lr': 0.0004778856723868303, 'samples': 3531264, 'steps': 6896, 'batch_loss/train': 0.9429766647517681} +12/23/2021 21:24:17 - INFO - codeparrot_training - Step 6897: {'lr': 0.0004778791900470753, 'samples': 3531776, 'steps': 6897, 'batch_loss/train': 0.7262771585956216} +12/23/2021 21:24:28 - INFO - codeparrot_training - Step 6898: {'lr': 0.00047787270680135866, 'samples': 3532288, 'steps': 6898, 'batch_loss/train': 0.8736706208437681} +12/23/2021 21:24:38 - INFO - codeparrot_training - Step 6899: {'lr': 0.00047786622264970626, 'samples': 3532800, 'steps': 6899, 'batch_loss/train': 0.8382629537954926} +12/23/2021 21:24:52 - INFO - codeparrot_training - Step 6900: {'lr': 0.0004778597375921437, 'samples': 3533312, 'steps': 6900, 'batch_loss/train': 0.887857224792242} +12/23/2021 21:25:02 - INFO - codeparrot_training - Step 6901: {'lr': 0.00047785325162869686, 'samples': 3533824, 'steps': 6901, 'batch_loss/train': 0.8788446914404631} +12/23/2021 21:25:13 - INFO - codeparrot_training - Step 6902: {'lr': 0.0004778467647593916, 'samples': 3534336, 'steps': 6902, 'batch_loss/train': 0.9211486061103642} +12/23/2021 21:25:25 - INFO - codeparrot_training - Step 6903: {'lr': 0.00047784027698425356, 'samples': 3534848, 'steps': 6903, 'batch_loss/train': 0.8855870626866817} +12/23/2021 21:25:36 - INFO - codeparrot_training - Step 6904: {'lr': 0.0004778337883033087, 'samples': 3535360, 'steps': 6904, 'batch_loss/train': 0.9640071485191584} +12/23/2021 21:25:46 - INFO - codeparrot_training - Step 6905: {'lr': 0.0004778272987165827, 'samples': 3535872, 'steps': 6905, 'batch_loss/train': 0.8632942088879645} +12/23/2021 21:25:57 - INFO - codeparrot_training - Step 6906: {'lr': 0.00047782080822410136, 'samples': 3536384, 'steps': 6906, 'batch_loss/train': 0.916967460885644} +12/23/2021 21:26:09 - INFO - codeparrot_training - Step 6907: {'lr': 0.0004778143168258905, 'samples': 3536896, 'steps': 6907, 'batch_loss/train': 0.7660887343809009} +12/23/2021 21:26:20 - INFO - codeparrot_training - Step 6908: {'lr': 0.0004778078245219759, 'samples': 3537408, 'steps': 6908, 'batch_loss/train': 0.8793529821559787} +12/23/2021 21:26:30 - INFO - codeparrot_training - Step 6909: {'lr': 0.0004778013313123835, 'samples': 3537920, 'steps': 6909, 'batch_loss/train': 0.8436913387849927} +12/23/2021 21:26:43 - INFO - codeparrot_training - Step 6910: {'lr': 0.0004777948371971389, 'samples': 3538432, 'steps': 6910, 'batch_loss/train': 0.910773865878582} +12/23/2021 21:26:53 - INFO - codeparrot_training - Step 6911: {'lr': 0.00047778834217626816, 'samples': 3538944, 'steps': 6911, 'batch_loss/train': 0.8641600804403424} +12/23/2021 21:27:04 - INFO - codeparrot_training - Step 6912: {'lr': 0.00047778184624979695, 'samples': 3539456, 'steps': 6912, 'batch_loss/train': 0.8594919866882265} +12/23/2021 21:27:17 - INFO - codeparrot_training - Step 6913: {'lr': 0.000477775349417751, 'samples': 3539968, 'steps': 6913, 'batch_loss/train': 0.7160270513268188} +12/23/2021 21:27:27 - INFO - codeparrot_training - Step 6914: {'lr': 0.0004777688516801564, 'samples': 3540480, 'steps': 6914, 'batch_loss/train': 0.8863203013315797} +12/23/2021 21:27:38 - INFO - codeparrot_training - Step 6915: {'lr': 0.00047776235303703885, 'samples': 3540992, 'steps': 6915, 'batch_loss/train': 0.8603785252198577} +12/23/2021 21:27:49 - INFO - codeparrot_training - Step 6916: {'lr': 0.0004777558534884242, 'samples': 3541504, 'steps': 6916, 'batch_loss/train': 0.873324510641396} +12/23/2021 21:28:01 - INFO - codeparrot_training - Step 6917: {'lr': 0.0004777493530343382, 'samples': 3542016, 'steps': 6917, 'batch_loss/train': 0.8190066455863416} +12/23/2021 21:28:11 - INFO - codeparrot_training - Step 6918: {'lr': 0.0004777428516748068, 'samples': 3542528, 'steps': 6918, 'batch_loss/train': 0.877871772274375} +12/23/2021 21:28:22 - INFO - codeparrot_training - Step 6919: {'lr': 0.0004777363494098558, 'samples': 3543040, 'steps': 6919, 'batch_loss/train': 0.8326048064045608} +12/23/2021 21:28:34 - INFO - codeparrot_training - Step 6920: {'lr': 0.0004777298462395111, 'samples': 3543552, 'steps': 6920, 'batch_loss/train': 0.8132760592270643} +12/23/2021 21:28:45 - INFO - codeparrot_training - Step 6921: {'lr': 0.00047772334216379843, 'samples': 3544064, 'steps': 6921, 'batch_loss/train': 0.7483230878133327} +12/23/2021 21:28:55 - INFO - codeparrot_training - Step 6922: {'lr': 0.00047771683718274386, 'samples': 3544576, 'steps': 6922, 'batch_loss/train': 0.7688084384426475} +12/23/2021 21:29:08 - INFO - codeparrot_training - Step 6923: {'lr': 0.00047771033129637304, 'samples': 3545088, 'steps': 6923, 'batch_loss/train': 0.804033050313592} +12/23/2021 21:29:19 - INFO - codeparrot_training - Step 6924: {'lr': 0.00047770382450471196, 'samples': 3545600, 'steps': 6924, 'batch_loss/train': 0.824066323461011} +12/23/2021 21:29:30 - INFO - codeparrot_training - Step 6925: {'lr': 0.00047769731680778646, 'samples': 3546112, 'steps': 6925, 'batch_loss/train': 0.7487220182083547} +12/23/2021 21:29:42 - INFO - codeparrot_training - Step 6926: {'lr': 0.0004776908082056224, 'samples': 3546624, 'steps': 6926, 'batch_loss/train': 0.8334716958925128} +12/23/2021 21:29:52 - INFO - codeparrot_training - Step 6927: {'lr': 0.0004776842986982456, 'samples': 3547136, 'steps': 6927, 'batch_loss/train': 0.8945160759612918} +12/23/2021 21:30:03 - INFO - codeparrot_training - Step 6928: {'lr': 0.00047767778828568204, 'samples': 3547648, 'steps': 6928, 'batch_loss/train': 0.887838161084801} +12/23/2021 21:30:13 - INFO - codeparrot_training - Step 6929: {'lr': 0.00047767127696795756, 'samples': 3548160, 'steps': 6929, 'batch_loss/train': 0.7850989541038871} +12/23/2021 21:30:27 - INFO - codeparrot_training - Step 6930: {'lr': 0.00047766476474509807, 'samples': 3548672, 'steps': 6930, 'batch_loss/train': 0.8520954325795174} +12/23/2021 21:30:37 - INFO - codeparrot_training - Step 6931: {'lr': 0.00047765825161712937, 'samples': 3549184, 'steps': 6931, 'batch_loss/train': 0.7806204040534794} +12/23/2021 21:30:48 - INFO - codeparrot_training - Step 6932: {'lr': 0.0004776517375840775, 'samples': 3549696, 'steps': 6932, 'batch_loss/train': 0.8299152031540871} +12/23/2021 21:31:00 - INFO - codeparrot_training - Step 6933: {'lr': 0.00047764522264596824, 'samples': 3550208, 'steps': 6933, 'batch_loss/train': 0.8329054163768888} +12/23/2021 21:31:10 - INFO - codeparrot_training - Step 6934: {'lr': 0.00047763870680282754, 'samples': 3550720, 'steps': 6934, 'batch_loss/train': 0.7841111859306693} +12/23/2021 21:31:21 - INFO - codeparrot_training - Step 6935: {'lr': 0.00047763219005468127, 'samples': 3551232, 'steps': 6935, 'batch_loss/train': 0.7907864074222744} +12/23/2021 21:31:33 - INFO - codeparrot_training - Step 6936: {'lr': 0.0004776256724015554, 'samples': 3551744, 'steps': 6936, 'batch_loss/train': 0.8823614921420813} +12/23/2021 21:31:44 - INFO - codeparrot_training - Step 6937: {'lr': 0.00047761915384347586, 'samples': 3552256, 'steps': 6937, 'batch_loss/train': 0.8969839821802452} +12/23/2021 21:31:54 - INFO - codeparrot_training - Step 6938: {'lr': 0.00047761263438046844, 'samples': 3552768, 'steps': 6938, 'batch_loss/train': 0.8140365155413747} +12/23/2021 21:32:05 - INFO - codeparrot_training - Step 6939: {'lr': 0.0004776061140125591, 'samples': 3553280, 'steps': 6939, 'batch_loss/train': 0.921349816955626} +12/23/2021 21:32:18 - INFO - codeparrot_training - Step 6940: {'lr': 0.0004775995927397738, 'samples': 3553792, 'steps': 6940, 'batch_loss/train': 0.7398234708234668} +12/23/2021 21:32:29 - INFO - codeparrot_training - Step 6941: {'lr': 0.0004775930705621385, 'samples': 3554304, 'steps': 6941, 'batch_loss/train': 0.6969168032519519} +12/23/2021 21:32:40 - INFO - codeparrot_training - Step 6942: {'lr': 0.0004775865474796791, 'samples': 3554816, 'steps': 6942, 'batch_loss/train': 0.8029583496972919} +12/23/2021 21:32:52 - INFO - codeparrot_training - Step 6943: {'lr': 0.00047758002349242146, 'samples': 3555328, 'steps': 6943, 'batch_loss/train': 0.873280230909586} +12/23/2021 21:33:02 - INFO - codeparrot_training - Step 6944: {'lr': 0.0004775734986003916, 'samples': 3555840, 'steps': 6944, 'batch_loss/train': 0.8082992830313742} +12/23/2021 21:33:13 - INFO - codeparrot_training - Step 6945: {'lr': 0.0004775669728036155, 'samples': 3556352, 'steps': 6945, 'batch_loss/train': 0.9000324052758515} +12/23/2021 21:33:25 - INFO - codeparrot_training - Step 6946: {'lr': 0.000477560446102119, 'samples': 3556864, 'steps': 6946, 'batch_loss/train': 0.7769851819612086} +12/23/2021 21:33:36 - INFO - codeparrot_training - Step 6947: {'lr': 0.000477553918495928, 'samples': 3557376, 'steps': 6947, 'batch_loss/train': 0.8370891949161887} +12/23/2021 21:33:46 - INFO - codeparrot_training - Step 6948: {'lr': 0.00047754738998506864, 'samples': 3557888, 'steps': 6948, 'batch_loss/train': 0.7971552750095725} +12/23/2021 21:33:59 - INFO - codeparrot_training - Step 6949: {'lr': 0.0004775408605695668, 'samples': 3558400, 'steps': 6949, 'batch_loss/train': 0.7799787297844887} +12/23/2021 21:34:10 - INFO - codeparrot_training - Step 6950: {'lr': 0.0004775343302494484, 'samples': 3558912, 'steps': 6950, 'batch_loss/train': 0.7929456057026982} +12/23/2021 21:34:20 - INFO - codeparrot_training - Step 6951: {'lr': 0.00047752779902473933, 'samples': 3559424, 'steps': 6951, 'batch_loss/train': 0.8636689987033606} +12/23/2021 21:34:31 - INFO - codeparrot_training - Step 6952: {'lr': 0.0004775212668954657, 'samples': 3559936, 'steps': 6952, 'batch_loss/train': 0.8237831378355622} +12/23/2021 21:34:43 - INFO - codeparrot_training - Step 6953: {'lr': 0.0004775147338616534, 'samples': 3560448, 'steps': 6953, 'batch_loss/train': 0.7537208134308457} +12/23/2021 21:34:54 - INFO - codeparrot_training - Step 6954: {'lr': 0.00047750819992332846, 'samples': 3560960, 'steps': 6954, 'batch_loss/train': 0.6149218869395554} +12/23/2021 21:35:04 - INFO - codeparrot_training - Step 6955: {'lr': 0.0004775016650805168, 'samples': 3561472, 'steps': 6955, 'batch_loss/train': 0.6823960365727544} +12/23/2021 21:35:16 - INFO - codeparrot_training - Step 6956: {'lr': 0.00047749512933324445, 'samples': 3561984, 'steps': 6956, 'batch_loss/train': 0.8558639781549573} +12/23/2021 21:35:27 - INFO - codeparrot_training - Step 6957: {'lr': 0.00047748859268153735, 'samples': 3562496, 'steps': 6957, 'batch_loss/train': 0.8584126867353916} +12/23/2021 21:35:37 - INFO - codeparrot_training - Step 6958: {'lr': 0.0004774820551254214, 'samples': 3563008, 'steps': 6958, 'batch_loss/train': 0.7832226548343897} +12/23/2021 21:35:48 - INFO - codeparrot_training - Step 6959: {'lr': 0.00047747551666492286, 'samples': 3563520, 'steps': 6959, 'batch_loss/train': 0.9911059439182281} +12/23/2021 21:36:01 - INFO - codeparrot_training - Step 6960: {'lr': 0.00047746897730006747, 'samples': 3564032, 'steps': 6960, 'batch_loss/train': 1.0335938390344381} +12/23/2021 21:36:12 - INFO - codeparrot_training - Step 6961: {'lr': 0.0004774624370308813, 'samples': 3564544, 'steps': 6961, 'batch_loss/train': 0.7864361939718947} +12/23/2021 21:36:22 - INFO - codeparrot_training - Step 6962: {'lr': 0.0004774558958573904, 'samples': 3565056, 'steps': 6962, 'batch_loss/train': 0.8587045464664698} +12/23/2021 21:36:34 - INFO - codeparrot_training - Step 6963: {'lr': 0.00047744935377962074, 'samples': 3565568, 'steps': 6963, 'batch_loss/train': 0.9856368647888303} +12/23/2021 21:36:45 - INFO - codeparrot_training - Step 6964: {'lr': 0.00047744281079759835, 'samples': 3566080, 'steps': 6964, 'batch_loss/train': 0.8497234676033258} +12/23/2021 21:36:56 - INFO - codeparrot_training - Step 6965: {'lr': 0.00047743626691134917, 'samples': 3566592, 'steps': 6965, 'batch_loss/train': 0.7528545372188091} +12/23/2021 21:37:08 - INFO - codeparrot_training - Step 6966: {'lr': 0.0004774297221208993, 'samples': 3567104, 'steps': 6966, 'batch_loss/train': 0.6819566730409861} +12/23/2021 21:37:18 - INFO - codeparrot_training - Step 6967: {'lr': 0.0004774231764262747, 'samples': 3567616, 'steps': 6967, 'batch_loss/train': 0.9301007464528084} +12/23/2021 21:37:29 - INFO - codeparrot_training - Step 6968: {'lr': 0.0004774166298275015, 'samples': 3568128, 'steps': 6968, 'batch_loss/train': 0.8708732137456536} +12/23/2021 21:37:39 - INFO - codeparrot_training - Step 6969: {'lr': 0.00047741008232460557, 'samples': 3568640, 'steps': 6969, 'batch_loss/train': 0.8726643212139606} +12/23/2021 21:37:53 - INFO - codeparrot_training - Step 6970: {'lr': 0.000477403533917613, 'samples': 3569152, 'steps': 6970, 'batch_loss/train': 0.7577850488014519} +12/23/2021 21:38:04 - INFO - codeparrot_training - Step 6971: {'lr': 0.00047739698460654993, 'samples': 3569664, 'steps': 6971, 'batch_loss/train': 0.9218739485368133} +12/23/2021 21:38:14 - INFO - codeparrot_training - Step 6972: {'lr': 0.0004773904343914423, 'samples': 3570176, 'steps': 6972, 'batch_loss/train': 0.7965298276394606} +12/23/2021 21:38:26 - INFO - codeparrot_training - Step 6973: {'lr': 0.00047738388327231613, 'samples': 3570688, 'steps': 6973, 'batch_loss/train': 0.5234253676608205} +12/23/2021 21:38:37 - INFO - codeparrot_training - Step 6974: {'lr': 0.0004773773312491975, 'samples': 3571200, 'steps': 6974, 'batch_loss/train': 0.7423238670453429} +12/23/2021 21:38:48 - INFO - codeparrot_training - Step 6975: {'lr': 0.0004773707783221125, 'samples': 3571712, 'steps': 6975, 'batch_loss/train': 0.8767646825872362} +12/23/2021 21:38:58 - INFO - codeparrot_training - Step 6976: {'lr': 0.00047736422449108705, 'samples': 3572224, 'steps': 6976, 'batch_loss/train': 0.8286002082750201} +12/23/2021 21:39:10 - INFO - codeparrot_training - Step 6977: {'lr': 0.0004773576697561474, 'samples': 3572736, 'steps': 6977, 'batch_loss/train': 0.9640879780054092} +12/23/2021 21:39:21 - INFO - codeparrot_training - Step 6978: {'lr': 0.00047735111411731937, 'samples': 3573248, 'steps': 6978, 'batch_loss/train': 0.8873682417906821} +12/23/2021 21:39:32 - INFO - codeparrot_training - Step 6979: {'lr': 0.00047734455757462925, 'samples': 3573760, 'steps': 6979, 'batch_loss/train': 0.8106065536849201} +12/23/2021 21:39:44 - INFO - codeparrot_training - Step 6980: {'lr': 0.00047733800012810304, 'samples': 3574272, 'steps': 6980, 'batch_loss/train': 0.9439310887828469} +12/23/2021 21:39:54 - INFO - codeparrot_training - Step 6981: {'lr': 0.00047733144177776666, 'samples': 3574784, 'steps': 6981, 'batch_loss/train': 0.9083540197461843} +12/23/2021 21:40:05 - INFO - codeparrot_training - Step 6982: {'lr': 0.00047732488252364634, 'samples': 3575296, 'steps': 6982, 'batch_loss/train': 0.857195420190692} +12/23/2021 21:40:18 - INFO - codeparrot_training - Step 6983: {'lr': 0.0004773183223657682, 'samples': 3575808, 'steps': 6983, 'batch_loss/train': 0.9363523237407207} +12/23/2021 21:40:29 - INFO - codeparrot_training - Step 6984: {'lr': 0.00047731176130415823, 'samples': 3576320, 'steps': 6984, 'batch_loss/train': 0.831555612385273} +12/23/2021 21:40:39 - INFO - codeparrot_training - Step 6985: {'lr': 0.00047730519933884244, 'samples': 3576832, 'steps': 6985, 'batch_loss/train': 1.1711747404187918} +12/23/2021 21:40:50 - INFO - codeparrot_training - Step 6986: {'lr': 0.00047729863646984706, 'samples': 3577344, 'steps': 6986, 'batch_loss/train': 0.9158955286256969} +12/23/2021 21:41:02 - INFO - codeparrot_training - Step 6987: {'lr': 0.0004772920726971981, 'samples': 3577856, 'steps': 6987, 'batch_loss/train': 0.7504816681612283} +12/23/2021 21:41:13 - INFO - codeparrot_training - Step 6988: {'lr': 0.00047728550802092175, 'samples': 3578368, 'steps': 6988, 'batch_loss/train': 0.8788252212107182} +12/23/2021 21:41:23 - INFO - codeparrot_training - Step 6989: {'lr': 0.00047727894244104397, 'samples': 3578880, 'steps': 6989, 'batch_loss/train': 0.8937845444306731} +12/23/2021 21:41:36 - INFO - codeparrot_training - Step 6990: {'lr': 0.000477272375957591, 'samples': 3579392, 'steps': 6990, 'batch_loss/train': 0.8343823002651334} +12/23/2021 21:41:47 - INFO - codeparrot_training - Step 6991: {'lr': 0.0004772658085705888, 'samples': 3579904, 'steps': 6991, 'batch_loss/train': 0.9020528383553028} +12/23/2021 21:41:57 - INFO - codeparrot_training - Step 6992: {'lr': 0.0004772592402800636, 'samples': 3580416, 'steps': 6992, 'batch_loss/train': 0.8797792373225093} +12/23/2021 21:42:09 - INFO - codeparrot_training - Step 6993: {'lr': 0.00047725267108604145, 'samples': 3580928, 'steps': 6993, 'batch_loss/train': 0.7949721990153193} +12/23/2021 21:42:20 - INFO - codeparrot_training - Step 6994: {'lr': 0.0004772461009885485, 'samples': 3581440, 'steps': 6994, 'batch_loss/train': 0.9329234268516302} +12/23/2021 21:42:30 - INFO - codeparrot_training - Step 6995: {'lr': 0.0004772395299876109, 'samples': 3581952, 'steps': 6995, 'batch_loss/train': 0.8202592865563929} +12/23/2021 21:42:41 - INFO - codeparrot_training - Step 6996: {'lr': 0.0004772329580832546, 'samples': 3582464, 'steps': 6996, 'batch_loss/train': 0.8843750730156898} +12/23/2021 21:42:53 - INFO - codeparrot_training - Step 6997: {'lr': 0.00047722638527550597, 'samples': 3582976, 'steps': 6997, 'batch_loss/train': 0.8224238567054272} +12/23/2021 21:43:04 - INFO - codeparrot_training - Step 6998: {'lr': 0.00047721981156439097, 'samples': 3583488, 'steps': 6998, 'batch_loss/train': 0.7706106407567859} +12/23/2021 21:43:14 - INFO - codeparrot_training - Step 6999: {'lr': 0.0004772132369499359, 'samples': 3584000, 'steps': 6999, 'batch_loss/train': 0.9770363448187709} +12/23/2021 21:43:28 - INFO - codeparrot_training - Step 7000: {'lr': 0.00047720666143216667, 'samples': 3584512, 'steps': 7000, 'batch_loss/train': 0.8627406032755971} +12/23/2021 21:43:38 - INFO - codeparrot_training - Step 7001: {'lr': 0.00047720008501110964, 'samples': 3585024, 'steps': 7001, 'batch_loss/train': 0.8209858741611242} +12/23/2021 21:43:49 - INFO - codeparrot_training - Step 7002: {'lr': 0.0004771935076867908, 'samples': 3585536, 'steps': 7002, 'batch_loss/train': 0.8554926086217165} +12/23/2021 21:44:01 - INFO - codeparrot_training - Step 7003: {'lr': 0.00047718692945923635, 'samples': 3586048, 'steps': 7003, 'batch_loss/train': 0.8171973074786365} +12/23/2021 21:44:12 - INFO - codeparrot_training - Step 7004: {'lr': 0.0004771803503284725, 'samples': 3586560, 'steps': 7004, 'batch_loss/train': 0.8709346745163202} +12/23/2021 21:44:22 - INFO - codeparrot_training - Step 7005: {'lr': 0.0004771737702945253, 'samples': 3587072, 'steps': 7005, 'batch_loss/train': 0.7434025150723755} +12/23/2021 21:44:33 - INFO - codeparrot_training - Step 7006: {'lr': 0.000477167189357421, 'samples': 3587584, 'steps': 7006, 'batch_loss/train': 0.8018251610919833} +12/23/2021 21:44:46 - INFO - codeparrot_training - Step 7007: {'lr': 0.0004771606075171857, 'samples': 3588096, 'steps': 7007, 'batch_loss/train': 0.8249971307814121} +12/23/2021 21:44:57 - INFO - codeparrot_training - Step 7008: {'lr': 0.00047715402477384566, 'samples': 3588608, 'steps': 7008, 'batch_loss/train': 1.0606078249402344} +12/23/2021 21:45:07 - INFO - codeparrot_training - Step 7009: {'lr': 0.00047714744112742694, 'samples': 3589120, 'steps': 7009, 'batch_loss/train': 0.9145002886652946} +12/23/2021 21:45:20 - INFO - codeparrot_training - Step 7010: {'lr': 0.0004771408565779558, 'samples': 3589632, 'steps': 7010, 'batch_loss/train': 0.7489120806567371} +12/23/2021 21:45:30 - INFO - codeparrot_training - Step 7011: {'lr': 0.0004771342711254584, 'samples': 3590144, 'steps': 7011, 'batch_loss/train': 0.7681552087888122} +12/23/2021 21:45:41 - INFO - codeparrot_training - Step 7012: {'lr': 0.00047712768476996083, 'samples': 3590656, 'steps': 7012, 'batch_loss/train': 0.8108293209224939} +12/23/2021 21:45:54 - INFO - codeparrot_training - Step 7013: {'lr': 0.00047712109751148934, 'samples': 3591168, 'steps': 7013, 'batch_loss/train': 0.833675223402679} +12/23/2021 21:46:05 - INFO - codeparrot_training - Step 7014: {'lr': 0.00047711450935007015, 'samples': 3591680, 'steps': 7014, 'batch_loss/train': 0.9611263908445835} +12/23/2021 21:46:15 - INFO - codeparrot_training - Step 7015: {'lr': 0.0004771079202857294, 'samples': 3592192, 'steps': 7015, 'batch_loss/train': 0.8538799770176411} +12/23/2021 21:46:26 - INFO - codeparrot_training - Step 7016: {'lr': 0.0004771013303184933, 'samples': 3592704, 'steps': 7016, 'batch_loss/train': 1.3463701736181974} +12/23/2021 21:46:38 - INFO - codeparrot_training - Step 7017: {'lr': 0.00047709473944838816, 'samples': 3593216, 'steps': 7017, 'batch_loss/train': 0.9446089905686677} +12/23/2021 21:46:48 - INFO - codeparrot_training - Step 7018: {'lr': 0.00047708814767544004, 'samples': 3593728, 'steps': 7018, 'batch_loss/train': 0.8777445056475699} +12/23/2021 21:46:59 - INFO - codeparrot_training - Step 7019: {'lr': 0.0004770815549996752, 'samples': 3594240, 'steps': 7019, 'batch_loss/train': 0.7572078624507412} +12/23/2021 21:47:12 - INFO - codeparrot_training - Step 7020: {'lr': 0.00047707496142111977, 'samples': 3594752, 'steps': 7020, 'batch_loss/train': 0.8707859227433801} +12/23/2021 21:47:23 - INFO - codeparrot_training - Step 7021: {'lr': 0.00047706836693980003, 'samples': 3595264, 'steps': 7021, 'batch_loss/train': 0.9137544594705105} +12/23/2021 21:47:33 - INFO - codeparrot_training - Step 7022: {'lr': 0.0004770617715557423, 'samples': 3595776, 'steps': 7022, 'batch_loss/train': 0.8091956032440066} +12/23/2021 21:47:44 - INFO - codeparrot_training - Step 7023: {'lr': 0.00047705517526897267, 'samples': 3596288, 'steps': 7023, 'batch_loss/train': 0.8850630563683808} +12/23/2021 21:47:56 - INFO - codeparrot_training - Step 7024: {'lr': 0.00047704857807951735, 'samples': 3596800, 'steps': 7024, 'batch_loss/train': 0.7992569357156754} +12/23/2021 21:48:06 - INFO - codeparrot_training - Step 7025: {'lr': 0.00047704197998740263, 'samples': 3597312, 'steps': 7025, 'batch_loss/train': 0.7226830534636974} +12/23/2021 21:48:17 - INFO - codeparrot_training - Step 7026: {'lr': 0.00047703538099265475, 'samples': 3597824, 'steps': 7026, 'batch_loss/train': 0.7723088413476944} +12/23/2021 21:48:29 - INFO - codeparrot_training - Step 7027: {'lr': 0.0004770287810952999, 'samples': 3598336, 'steps': 7027, 'batch_loss/train': 0.8270145659334958} +12/23/2021 21:48:39 - INFO - codeparrot_training - Step 7028: {'lr': 0.0004770221802953644, 'samples': 3598848, 'steps': 7028, 'batch_loss/train': 0.7483779487665743} +12/23/2021 21:48:50 - INFO - codeparrot_training - Step 7029: {'lr': 0.00047701557859287436, 'samples': 3599360, 'steps': 7029, 'batch_loss/train': 0.8419028399512172} +12/23/2021 21:49:03 - INFO - codeparrot_training - Step 7030: {'lr': 0.0004770089759878561, 'samples': 3599872, 'steps': 7030, 'batch_loss/train': 0.8473116201348603} +12/23/2021 21:49:14 - INFO - codeparrot_training - Step 7031: {'lr': 0.0004770023724803358, 'samples': 3600384, 'steps': 7031, 'batch_loss/train': 0.8210955592803657} +12/23/2021 21:49:25 - INFO - codeparrot_training - Step 7032: {'lr': 0.00047699576807033994, 'samples': 3600896, 'steps': 7032, 'batch_loss/train': 0.9150357120670378} +12/23/2021 21:49:35 - INFO - codeparrot_training - Step 7033: {'lr': 0.0004769891627578945, 'samples': 3601408, 'steps': 7033, 'batch_loss/train': 0.7812175392173231} +12/23/2021 21:49:47 - INFO - codeparrot_training - Step 7034: {'lr': 0.0004769825565430259, 'samples': 3601920, 'steps': 7034, 'batch_loss/train': 0.7894929046742618} +12/23/2021 21:49:58 - INFO - codeparrot_training - Step 7035: {'lr': 0.0004769759494257604, 'samples': 3602432, 'steps': 7035, 'batch_loss/train': 0.821730013936758} +12/23/2021 21:50:08 - INFO - codeparrot_training - Step 7036: {'lr': 0.0004769693414061241, 'samples': 3602944, 'steps': 7036, 'batch_loss/train': 0.8367009349167347} +12/23/2021 21:50:22 - INFO - codeparrot_training - Step 7037: {'lr': 0.0004769627324841436, 'samples': 3603456, 'steps': 7037, 'batch_loss/train': 1.021661525592208} +12/23/2021 21:50:32 - INFO - codeparrot_training - Step 7038: {'lr': 0.0004769561226598448, 'samples': 3603968, 'steps': 7038, 'batch_loss/train': 0.8633162369951606} +12/23/2021 21:50:43 - INFO - codeparrot_training - Step 7039: {'lr': 0.0004769495119332542, 'samples': 3604480, 'steps': 7039, 'batch_loss/train': 0.8557260381057858} +12/23/2021 21:50:55 - INFO - codeparrot_training - Step 7040: {'lr': 0.00047694290030439804, 'samples': 3604992, 'steps': 7040, 'batch_loss/train': 0.8030759415123612} +12/23/2021 21:51:05 - INFO - codeparrot_training - Step 7041: {'lr': 0.00047693628777330254, 'samples': 3605504, 'steps': 7041, 'batch_loss/train': 0.7809805022552609} +12/23/2021 21:51:16 - INFO - codeparrot_training - Step 7042: {'lr': 0.0004769296743399941, 'samples': 3606016, 'steps': 7042, 'batch_loss/train': 0.8357144435867667} +12/23/2021 21:51:28 - INFO - codeparrot_training - Step 7043: {'lr': 0.000476923060004499, 'samples': 3606528, 'steps': 7043, 'batch_loss/train': 0.9061266211792827} +12/23/2021 21:51:39 - INFO - codeparrot_training - Step 7044: {'lr': 0.0004769164447668435, 'samples': 3607040, 'steps': 7044, 'batch_loss/train': 0.8316645794548094} +12/23/2021 21:51:49 - INFO - codeparrot_training - Step 7045: {'lr': 0.0004769098286270538, 'samples': 3607552, 'steps': 7045, 'batch_loss/train': 0.7856781715527177} +12/23/2021 21:52:00 - INFO - codeparrot_training - Step 7046: {'lr': 0.0004769032115851564, 'samples': 3608064, 'steps': 7046, 'batch_loss/train': 0.8048768024891615} +12/23/2021 21:52:13 - INFO - codeparrot_training - Step 7047: {'lr': 0.00047689659364117743, 'samples': 3608576, 'steps': 7047, 'batch_loss/train': 0.8409268818795681} +12/23/2021 21:52:23 - INFO - codeparrot_training - Step 7048: {'lr': 0.00047688997479514335, 'samples': 3609088, 'steps': 7048, 'batch_loss/train': 0.8756671724840999} +12/23/2021 21:52:34 - INFO - codeparrot_training - Step 7049: {'lr': 0.0004768833550470804, 'samples': 3609600, 'steps': 7049, 'batch_loss/train': 0.8765383241698146} +12/23/2021 21:52:46 - INFO - codeparrot_training - Step 7050: {'lr': 0.0004768767343970148, 'samples': 3610112, 'steps': 7050, 'batch_loss/train': 0.8027260899543762} +12/23/2021 21:52:56 - INFO - codeparrot_training - Step 7051: {'lr': 0.00047687011284497306, 'samples': 3610624, 'steps': 7051, 'batch_loss/train': 0.8468901868909597} +12/23/2021 21:53:07 - INFO - codeparrot_training - Step 7052: {'lr': 0.0004768634903909813, 'samples': 3611136, 'steps': 7052, 'batch_loss/train': 0.8151417942717671} +12/23/2021 21:53:21 - INFO - codeparrot_training - Step 7053: {'lr': 0.0004768568670350661, 'samples': 3611648, 'steps': 7053, 'batch_loss/train': 0.8688656468875706} +12/23/2021 21:53:31 - INFO - codeparrot_training - Step 7054: {'lr': 0.00047685024277725356, 'samples': 3612160, 'steps': 7054, 'batch_loss/train': 0.7908484635408968} +12/23/2021 21:53:42 - INFO - codeparrot_training - Step 7055: {'lr': 0.00047684361761757017, 'samples': 3612672, 'steps': 7055, 'batch_loss/train': 0.9079984035342932} +12/23/2021 21:53:53 - INFO - codeparrot_training - Step 7056: {'lr': 0.0004768369915560422, 'samples': 3613184, 'steps': 7056, 'batch_loss/train': 0.8892711522057652} +12/23/2021 21:54:05 - INFO - codeparrot_training - Step 7057: {'lr': 0.000476830364592696, 'samples': 3613696, 'steps': 7057, 'batch_loss/train': 0.8204667172394693} +12/23/2021 21:54:15 - INFO - codeparrot_training - Step 7058: {'lr': 0.00047682373672755784, 'samples': 3614208, 'steps': 7058, 'batch_loss/train': 0.7685574372299016} +12/23/2021 21:54:26 - INFO - codeparrot_training - Step 7059: {'lr': 0.00047681710796065424, 'samples': 3614720, 'steps': 7059, 'batch_loss/train': 0.8492756919004023} +12/23/2021 21:54:38 - INFO - codeparrot_training - Step 7060: {'lr': 0.00047681047829201145, 'samples': 3615232, 'steps': 7060, 'batch_loss/train': 0.8275305279530585} +12/23/2021 21:54:49 - INFO - codeparrot_training - Step 7061: {'lr': 0.0004768038477216559, 'samples': 3615744, 'steps': 7061, 'batch_loss/train': 0.8560773991048336} +12/23/2021 21:55:00 - INFO - codeparrot_training - Step 7062: {'lr': 0.00047679721624961375, 'samples': 3616256, 'steps': 7062, 'batch_loss/train': 0.8078741421923041} +12/23/2021 21:55:10 - INFO - codeparrot_training - Step 7063: {'lr': 0.0004767905838759116, 'samples': 3616768, 'steps': 7063, 'batch_loss/train': 0.859865877777338} +12/23/2021 21:55:22 - INFO - codeparrot_training - Step 7064: {'lr': 0.0004767839506005757, 'samples': 3617280, 'steps': 7064, 'batch_loss/train': 0.9333210922777653} +12/23/2021 21:55:33 - INFO - codeparrot_training - Step 7065: {'lr': 0.0004767773164236324, 'samples': 3617792, 'steps': 7065, 'batch_loss/train': 0.8756831269711256} +12/23/2021 21:55:43 - INFO - codeparrot_training - Step 7066: {'lr': 0.00047677068134510816, 'samples': 3618304, 'steps': 7066, 'batch_loss/train': 0.9581021247431636} +12/23/2021 21:55:57 - INFO - codeparrot_training - Step 7067: {'lr': 0.00047676404536502924, 'samples': 3618816, 'steps': 7067, 'batch_loss/train': 0.8781484963838011} +12/23/2021 21:56:07 - INFO - codeparrot_training - Step 7068: {'lr': 0.0004767574084834222, 'samples': 3619328, 'steps': 7068, 'batch_loss/train': 0.8242837665602565} +12/23/2021 21:56:18 - INFO - codeparrot_training - Step 7069: {'lr': 0.0004767507707003133, 'samples': 3619840, 'steps': 7069, 'batch_loss/train': 0.8424425749108195} +12/23/2021 21:56:30 - INFO - codeparrot_training - Step 7070: {'lr': 0.000476744132015729, 'samples': 3620352, 'steps': 7070, 'batch_loss/train': 0.9247558452188969} +12/23/2021 21:56:41 - INFO - codeparrot_training - Step 7071: {'lr': 0.00047673749242969557, 'samples': 3620864, 'steps': 7071, 'batch_loss/train': 0.7969437669962645} +12/23/2021 21:56:52 - INFO - codeparrot_training - Step 7072: {'lr': 0.0004767308519422395, 'samples': 3621376, 'steps': 7072, 'batch_loss/train': 0.8529840242117643} +12/23/2021 21:57:02 - INFO - codeparrot_training - Step 7073: {'lr': 0.00047672421055338716, 'samples': 3621888, 'steps': 7073, 'batch_loss/train': 0.6819092989899218} +12/23/2021 21:57:15 - INFO - codeparrot_training - Step 7074: {'lr': 0.00047671756826316497, 'samples': 3622400, 'steps': 7074, 'batch_loss/train': 0.8025955725461245} +12/23/2021 21:57:26 - INFO - codeparrot_training - Step 7075: {'lr': 0.0004767109250715994, 'samples': 3622912, 'steps': 7075, 'batch_loss/train': 0.9294234002009034} +12/23/2021 21:57:36 - INFO - codeparrot_training - Step 7076: {'lr': 0.00047670428097871663, 'samples': 3623424, 'steps': 7076, 'batch_loss/train': 0.7387437426950783} +12/23/2021 21:57:50 - INFO - codeparrot_training - Step 7077: {'lr': 0.0004766976359845434, 'samples': 3623936, 'steps': 7077, 'batch_loss/train': 0.5355094410479069} +12/23/2021 21:58:01 - INFO - codeparrot_training - Step 7078: {'lr': 0.00047669099008910596, 'samples': 3624448, 'steps': 7078, 'batch_loss/train': 0.8686794368550181} +12/23/2021 21:58:12 - INFO - codeparrot_training - Step 7079: {'lr': 0.0004766843432924307, 'samples': 3624960, 'steps': 7079, 'batch_loss/train': 0.9436307149007916} +12/23/2021 21:58:22 - INFO - codeparrot_training - Step 7080: {'lr': 0.00047667769559454404, 'samples': 3625472, 'steps': 7080, 'batch_loss/train': 0.8577733309939504} +12/23/2021 21:58:34 - INFO - codeparrot_training - Step 7081: {'lr': 0.00047667104699547247, 'samples': 3625984, 'steps': 7081, 'batch_loss/train': 0.9396647168323398} +12/23/2021 21:58:45 - INFO - codeparrot_training - Step 7082: {'lr': 0.00047666439749524247, 'samples': 3626496, 'steps': 7082, 'batch_loss/train': 0.8304682312300429} +12/23/2021 21:58:55 - INFO - codeparrot_training - Step 7083: {'lr': 0.0004766577470938803, 'samples': 3627008, 'steps': 7083, 'batch_loss/train': 0.8625599388033152} +12/23/2021 21:59:09 - INFO - codeparrot_training - Step 7084: {'lr': 0.00047665109579141264, 'samples': 3627520, 'steps': 7084, 'batch_loss/train': 0.9053611978888512} +12/23/2021 21:59:20 - INFO - codeparrot_training - Step 7085: {'lr': 0.00047664444358786574, 'samples': 3628032, 'steps': 7085, 'batch_loss/train': 0.8335259025916457} +12/23/2021 21:59:30 - INFO - codeparrot_training - Step 7086: {'lr': 0.00047663779048326605, 'samples': 3628544, 'steps': 7086, 'batch_loss/train': 1.16911512427032} +12/23/2021 21:59:41 - INFO - codeparrot_training - Step 7087: {'lr': 0.00047663113647764015, 'samples': 3629056, 'steps': 7087, 'batch_loss/train': 0.8802017355337739} +12/23/2021 21:59:53 - INFO - codeparrot_training - Step 7088: {'lr': 0.00047662448157101444, 'samples': 3629568, 'steps': 7088, 'batch_loss/train': 1.0103995138779283} +12/23/2021 22:00:03 - INFO - codeparrot_training - Step 7089: {'lr': 0.0004766178257634153, 'samples': 3630080, 'steps': 7089, 'batch_loss/train': 0.8452618420124054} +12/23/2021 22:00:14 - INFO - codeparrot_training - Step 7090: {'lr': 0.00047661116905486933, 'samples': 3630592, 'steps': 7090, 'batch_loss/train': 0.8492038112599403} +12/23/2021 22:00:26 - INFO - codeparrot_training - Step 7091: {'lr': 0.00047660451144540284, 'samples': 3631104, 'steps': 7091, 'batch_loss/train': 0.8572692908346653} +12/23/2021 22:00:37 - INFO - codeparrot_training - Step 7092: {'lr': 0.0004765978529350424, 'samples': 3631616, 'steps': 7092, 'batch_loss/train': 0.9819591864943504} +12/23/2021 22:00:47 - INFO - codeparrot_training - Step 7093: {'lr': 0.00047659119352381455, 'samples': 3632128, 'steps': 7093, 'batch_loss/train': 0.9078832312952727} +12/23/2021 22:01:00 - INFO - codeparrot_training - Step 7094: {'lr': 0.00047658453321174554, 'samples': 3632640, 'steps': 7094, 'batch_loss/train': 0.8477963795885444} +12/23/2021 22:01:11 - INFO - codeparrot_training - Step 7095: {'lr': 0.00047657787199886206, 'samples': 3633152, 'steps': 7095, 'batch_loss/train': 0.9292211914435029} +12/23/2021 22:01:21 - INFO - codeparrot_training - Step 7096: {'lr': 0.0004765712098851904, 'samples': 3633664, 'steps': 7096, 'batch_loss/train': 0.8953883415088058} +12/23/2021 22:01:33 - INFO - codeparrot_training - Step 7097: {'lr': 0.0004765645468707573, 'samples': 3634176, 'steps': 7097, 'batch_loss/train': 0.9221909549087286} +12/23/2021 22:01:44 - INFO - codeparrot_training - Step 7098: {'lr': 0.000476557882955589, 'samples': 3634688, 'steps': 7098, 'batch_loss/train': 0.8659280855208635} +12/23/2021 22:01:55 - INFO - codeparrot_training - Step 7099: {'lr': 0.0004765512181397122, 'samples': 3635200, 'steps': 7099, 'batch_loss/train': 0.8665328733623028} +12/23/2021 22:02:05 - INFO - codeparrot_training - Step 7100: {'lr': 0.00047654455242315325, 'samples': 3635712, 'steps': 7100, 'batch_loss/train': 0.9241405609063804} +12/23/2021 22:02:17 - INFO - codeparrot_training - Step 7101: {'lr': 0.0004765378858059387, 'samples': 3636224, 'steps': 7101, 'batch_loss/train': 0.9064558884128928} +12/23/2021 22:02:28 - INFO - codeparrot_training - Step 7102: {'lr': 0.0004765312182880951, 'samples': 3636736, 'steps': 7102, 'batch_loss/train': 0.8637956436723471} +12/23/2021 22:02:38 - INFO - codeparrot_training - Step 7103: {'lr': 0.0004765245498696488, 'samples': 3637248, 'steps': 7103, 'batch_loss/train': 0.9026485257782042} +12/23/2021 22:02:50 - INFO - codeparrot_training - Step 7104: {'lr': 0.0004765178805506265, 'samples': 3637760, 'steps': 7104, 'batch_loss/train': 0.9055257830768824} +12/23/2021 22:03:01 - INFO - codeparrot_training - Step 7105: {'lr': 0.0004765112103310546, 'samples': 3638272, 'steps': 7105, 'batch_loss/train': 1.0557616939768195} +12/23/2021 22:03:12 - INFO - codeparrot_training - Step 7106: {'lr': 0.0004765045392109597, 'samples': 3638784, 'steps': 7106, 'batch_loss/train': 0.8269264586269855} +12/23/2021 22:03:25 - INFO - codeparrot_training - Step 7107: {'lr': 0.0004764978671903683, 'samples': 3639296, 'steps': 7107, 'batch_loss/train': 0.9484722539782524} +12/23/2021 22:03:36 - INFO - codeparrot_training - Step 7108: {'lr': 0.00047649119426930686, 'samples': 3639808, 'steps': 7108, 'batch_loss/train': 0.890125778503716} +12/23/2021 22:03:47 - INFO - codeparrot_training - Step 7109: {'lr': 0.00047648452044780193, 'samples': 3640320, 'steps': 7109, 'batch_loss/train': 0.8303958363831043} +12/23/2021 22:03:57 - INFO - codeparrot_training - Step 7110: {'lr': 0.00047647784572588006, 'samples': 3640832, 'steps': 7110, 'batch_loss/train': 0.9781421292573214} +12/23/2021 22:04:09 - INFO - codeparrot_training - Step 7111: {'lr': 0.00047647117010356785, 'samples': 3641344, 'steps': 7111, 'batch_loss/train': 0.8151944568380713} +12/23/2021 22:04:20 - INFO - codeparrot_training - Step 7112: {'lr': 0.0004764644935808917, 'samples': 3641856, 'steps': 7112, 'batch_loss/train': 0.9074508361518383} +12/23/2021 22:04:31 - INFO - codeparrot_training - Step 7113: {'lr': 0.0004764578161578783, 'samples': 3642368, 'steps': 7113, 'batch_loss/train': 0.7770725693553686} +12/23/2021 22:04:45 - INFO - codeparrot_training - Step 7114: {'lr': 0.0004764511378345541, 'samples': 3642880, 'steps': 7114, 'batch_loss/train': 0.8988014319911599} +12/23/2021 22:04:55 - INFO - codeparrot_training - Step 7115: {'lr': 0.0004764444586109457, 'samples': 3643392, 'steps': 7115, 'batch_loss/train': 0.7124287937767804} +12/23/2021 22:05:06 - INFO - codeparrot_training - Step 7116: {'lr': 0.00047643777848707967, 'samples': 3643904, 'steps': 7116, 'batch_loss/train': 1.0039802081882954} +12/23/2021 22:05:18 - INFO - codeparrot_training - Step 7117: {'lr': 0.00047643109746298243, 'samples': 3644416, 'steps': 7117, 'batch_loss/train': 0.8921778431395069} +12/23/2021 22:05:28 - INFO - codeparrot_training - Step 7118: {'lr': 0.0004764244155386807, 'samples': 3644928, 'steps': 7118, 'batch_loss/train': 0.8831130862236023} +12/23/2021 22:05:39 - INFO - codeparrot_training - Step 7119: {'lr': 0.000476417732714201, 'samples': 3645440, 'steps': 7119, 'batch_loss/train': 0.7204559100791812} +12/23/2021 22:05:50 - INFO - codeparrot_training - Step 7120: {'lr': 0.0004764110489895699, 'samples': 3645952, 'steps': 7120, 'batch_loss/train': 0.9098513480275869} +12/23/2021 22:06:02 - INFO - codeparrot_training - Step 7121: {'lr': 0.0004764043643648139, 'samples': 3646464, 'steps': 7121, 'batch_loss/train': 0.8997217696160078} +12/23/2021 22:06:12 - INFO - codeparrot_training - Step 7122: {'lr': 0.00047639767883995965, 'samples': 3646976, 'steps': 7122, 'batch_loss/train': 0.8732679598033428} +12/23/2021 22:06:23 - INFO - codeparrot_training - Step 7123: {'lr': 0.00047639099241503376, 'samples': 3647488, 'steps': 7123, 'batch_loss/train': 0.8913336666300893} +12/23/2021 22:06:36 - INFO - codeparrot_training - Step 7124: {'lr': 0.0004763843050900627, 'samples': 3648000, 'steps': 7124, 'batch_loss/train': 0.8023087093606591} +12/23/2021 22:06:47 - INFO - codeparrot_training - Step 7125: {'lr': 0.0004763776168650732, 'samples': 3648512, 'steps': 7125, 'batch_loss/train': 0.5725774568272755} +12/23/2021 22:06:57 - INFO - codeparrot_training - Step 7126: {'lr': 0.0004763709277400917, 'samples': 3649024, 'steps': 7126, 'batch_loss/train': 0.8983312845230103} +12/23/2021 22:07:09 - INFO - codeparrot_training - Step 7127: {'lr': 0.00047636423771514494, 'samples': 3649536, 'steps': 7127, 'batch_loss/train': 0.8601238057017326} +12/23/2021 22:07:20 - INFO - codeparrot_training - Step 7128: {'lr': 0.0004763575467902594, 'samples': 3650048, 'steps': 7128, 'batch_loss/train': 0.9003271451219916} +12/23/2021 22:07:31 - INFO - codeparrot_training - Step 7129: {'lr': 0.0004763508549654617, 'samples': 3650560, 'steps': 7129, 'batch_loss/train': 0.8236901350319386} +12/23/2021 22:07:41 - INFO - codeparrot_training - Step 7130: {'lr': 0.0004763441622407785, 'samples': 3651072, 'steps': 7130, 'batch_loss/train': 0.9429804398678243} +12/23/2021 22:07:53 - INFO - codeparrot_training - Step 7131: {'lr': 0.0004763374686162364, 'samples': 3651584, 'steps': 7131, 'batch_loss/train': 0.7423534032423049} +12/23/2021 22:08:04 - INFO - codeparrot_training - Step 7132: {'lr': 0.00047633077409186186, 'samples': 3652096, 'steps': 7132, 'batch_loss/train': 0.9489136971533298} +12/23/2021 22:08:15 - INFO - codeparrot_training - Step 7133: {'lr': 0.00047632407866768177, 'samples': 3652608, 'steps': 7133, 'batch_loss/train': 0.7162669689860195} +12/23/2021 22:08:28 - INFO - codeparrot_training - Step 7134: {'lr': 0.0004763173823437226, 'samples': 3653120, 'steps': 7134, 'batch_loss/train': 0.8643149416893721} +12/23/2021 22:08:38 - INFO - codeparrot_training - Step 7135: {'lr': 0.0004763106851200109, 'samples': 3653632, 'steps': 7135, 'batch_loss/train': 0.9786082170903683} +12/23/2021 22:08:49 - INFO - codeparrot_training - Step 7136: {'lr': 0.00047630398699657336, 'samples': 3654144, 'steps': 7136, 'batch_loss/train': 0.830652441829443} +12/23/2021 22:09:01 - INFO - codeparrot_training - Step 7137: {'lr': 0.00047629728797343664, 'samples': 3654656, 'steps': 7137, 'batch_loss/train': 0.9153794758021832} +12/23/2021 22:09:11 - INFO - codeparrot_training - Step 7138: {'lr': 0.0004762905880506273, 'samples': 3655168, 'steps': 7138, 'batch_loss/train': 0.8882739930413663} +12/23/2021 22:09:22 - INFO - codeparrot_training - Step 7139: {'lr': 0.00047628388722817204, 'samples': 3655680, 'steps': 7139, 'batch_loss/train': 0.7658907491713762} +12/23/2021 22:09:33 - INFO - codeparrot_training - Step 7140: {'lr': 0.00047627718550609756, 'samples': 3656192, 'steps': 7140, 'batch_loss/train': 0.732025052420795} +12/23/2021 22:09:45 - INFO - codeparrot_training - Step 7141: {'lr': 0.0004762704828844304, 'samples': 3656704, 'steps': 7141, 'batch_loss/train': 0.894548729993403} +12/23/2021 22:09:55 - INFO - codeparrot_training - Step 7142: {'lr': 0.00047626377936319723, 'samples': 3657216, 'steps': 7142, 'batch_loss/train': 0.8301494484767318} +12/23/2021 22:10:06 - INFO - codeparrot_training - Step 7143: {'lr': 0.0004762570749424247, 'samples': 3657728, 'steps': 7143, 'batch_loss/train': 0.8316847011446953} +12/23/2021 22:10:18 - INFO - codeparrot_training - Step 7144: {'lr': 0.00047625036962213954, 'samples': 3658240, 'steps': 7144, 'batch_loss/train': 0.9406375009566545} +12/23/2021 22:10:28 - INFO - codeparrot_training - Step 7145: {'lr': 0.00047624366340236825, 'samples': 3658752, 'steps': 7145, 'batch_loss/train': 0.9711662102490664} +12/23/2021 22:10:39 - INFO - codeparrot_training - Step 7146: {'lr': 0.0004762369562831376, 'samples': 3659264, 'steps': 7146, 'batch_loss/train': 0.8923220029100776} +12/23/2021 22:10:53 - INFO - codeparrot_training - Step 7147: {'lr': 0.00047623024826447426, 'samples': 3659776, 'steps': 7147, 'batch_loss/train': 0.8995286580175161} +12/23/2021 22:11:03 - INFO - codeparrot_training - Step 7148: {'lr': 0.00047622353934640484, 'samples': 3660288, 'steps': 7148, 'batch_loss/train': 0.8688778700307012} +12/23/2021 22:11:14 - INFO - codeparrot_training - Step 7149: {'lr': 0.00047621682952895606, 'samples': 3660800, 'steps': 7149, 'batch_loss/train': 0.9132317891344428} +12/23/2021 22:11:26 - INFO - codeparrot_training - Step 7150: {'lr': 0.0004762101188121546, 'samples': 3661312, 'steps': 7150, 'batch_loss/train': 0.8631434831768274} +12/23/2021 22:11:37 - INFO - codeparrot_training - Step 7151: {'lr': 0.0004762034071960271, 'samples': 3661824, 'steps': 7151, 'batch_loss/train': 0.9237594418227673} +12/23/2021 22:11:47 - INFO - codeparrot_training - Step 7152: {'lr': 0.0004761966946806002, 'samples': 3662336, 'steps': 7152, 'batch_loss/train': 0.8708985168486834} +12/23/2021 22:11:58 - INFO - codeparrot_training - Step 7153: {'lr': 0.00047618998126590075, 'samples': 3662848, 'steps': 7153, 'batch_loss/train': 0.8869612235575914} +12/23/2021 22:12:12 - INFO - codeparrot_training - Step 7154: {'lr': 0.00047618326695195536, 'samples': 3663360, 'steps': 7154, 'batch_loss/train': 0.8105840273201466} +12/23/2021 22:12:22 - INFO - codeparrot_training - Step 7155: {'lr': 0.0004761765517387906, 'samples': 3663872, 'steps': 7155, 'batch_loss/train': 0.8454761682078242} +12/23/2021 22:12:33 - INFO - codeparrot_training - Step 7156: {'lr': 0.00047616983562643337, 'samples': 3664384, 'steps': 7156, 'batch_loss/train': 0.8192092105746269} +12/23/2021 22:12:45 - INFO - codeparrot_training - Step 7157: {'lr': 0.0004761631186149102, 'samples': 3664896, 'steps': 7157, 'batch_loss/train': 0.8437828877940774} +12/23/2021 22:12:56 - INFO - codeparrot_training - Step 7158: {'lr': 0.00047615640070424787, 'samples': 3665408, 'steps': 7158, 'batch_loss/train': 0.8436084389686584} +12/23/2021 22:13:06 - INFO - codeparrot_training - Step 7159: {'lr': 0.0004761496818944731, 'samples': 3665920, 'steps': 7159, 'batch_loss/train': 0.8661539917811751} +12/23/2021 22:13:18 - INFO - codeparrot_training - Step 7160: {'lr': 0.0004761429621856126, 'samples': 3666432, 'steps': 7160, 'batch_loss/train': 0.9259837463032454} +12/23/2021 22:13:29 - INFO - codeparrot_training - Step 7161: {'lr': 0.00047613624157769307, 'samples': 3666944, 'steps': 7161, 'batch_loss/train': 0.9372528344392776} +12/23/2021 22:13:39 - INFO - codeparrot_training - Step 7162: {'lr': 0.00047612952007074124, 'samples': 3667456, 'steps': 7162, 'batch_loss/train': 0.7784742321819067} +12/23/2021 22:13:50 - INFO - codeparrot_training - Step 7163: {'lr': 0.00047612279766478377, 'samples': 3667968, 'steps': 7163, 'batch_loss/train': 0.8833864014595747} +12/23/2021 22:14:03 - INFO - codeparrot_training - Step 7164: {'lr': 0.0004761160743598475, 'samples': 3668480, 'steps': 7164, 'batch_loss/train': 0.8476778566837311} +12/23/2021 22:14:14 - INFO - codeparrot_training - Step 7165: {'lr': 0.00047610935015595903, 'samples': 3668992, 'steps': 7165, 'batch_loss/train': 0.9163713799789548} +12/23/2021 22:14:24 - INFO - codeparrot_training - Step 7166: {'lr': 0.0004761026250531452, 'samples': 3669504, 'steps': 7166, 'batch_loss/train': 0.7988437680760399} +12/23/2021 22:14:36 - INFO - codeparrot_training - Step 7167: {'lr': 0.0004760958990514327, 'samples': 3670016, 'steps': 7167, 'batch_loss/train': 0.8144973344169557} +12/23/2021 22:14:47 - INFO - codeparrot_training - Step 7168: {'lr': 0.0004760891721508483, 'samples': 3670528, 'steps': 7168, 'batch_loss/train': 0.8629741882905364} +12/23/2021 22:14:57 - INFO - codeparrot_training - Step 7169: {'lr': 0.0004760824443514187, 'samples': 3671040, 'steps': 7169, 'batch_loss/train': 0.8266983684152365} +12/23/2021 22:15:11 - INFO - codeparrot_training - Step 7170: {'lr': 0.00047607571565317063, 'samples': 3671552, 'steps': 7170, 'batch_loss/train': 0.9463236704468727} +12/23/2021 22:15:22 - INFO - codeparrot_training - Step 7171: {'lr': 0.0004760689860561309, 'samples': 3672064, 'steps': 7171, 'batch_loss/train': 0.7686031833291054} +12/23/2021 22:15:32 - INFO - codeparrot_training - Step 7172: {'lr': 0.0004760622555603263, 'samples': 3672576, 'steps': 7172, 'batch_loss/train': 0.8118474070215598} +12/23/2021 22:15:44 - INFO - codeparrot_training - Step 7173: {'lr': 0.00047605552416578345, 'samples': 3673088, 'steps': 7173, 'batch_loss/train': 0.8701425045728683} +12/23/2021 22:15:55 - INFO - codeparrot_training - Step 7174: {'lr': 0.00047604879187252927, 'samples': 3673600, 'steps': 7174, 'batch_loss/train': 0.8993316092528403} +12/23/2021 22:16:06 - INFO - codeparrot_training - Step 7175: {'lr': 0.00047604205868059037, 'samples': 3674112, 'steps': 7175, 'batch_loss/train': 1.172151891514659} +12/23/2021 22:16:16 - INFO - codeparrot_training - Step 7176: {'lr': 0.00047603532458999364, 'samples': 3674624, 'steps': 7176, 'batch_loss/train': 0.8207385744899511} +12/23/2021 22:16:28 - INFO - codeparrot_training - Step 7177: {'lr': 0.00047602858960076576, 'samples': 3675136, 'steps': 7177, 'batch_loss/train': 0.8318614037707448} +12/23/2021 22:16:39 - INFO - codeparrot_training - Step 7178: {'lr': 0.0004760218537129336, 'samples': 3675648, 'steps': 7178, 'batch_loss/train': 1.1730604637414217} +12/23/2021 22:16:50 - INFO - codeparrot_training - Step 7179: {'lr': 0.0004760151169265239, 'samples': 3676160, 'steps': 7179, 'batch_loss/train': 0.8400692827999592} +12/23/2021 22:17:02 - INFO - codeparrot_training - Step 7180: {'lr': 0.00047600837924156343, 'samples': 3676672, 'steps': 7180, 'batch_loss/train': 0.8645874678622931} +12/23/2021 22:17:12 - INFO - codeparrot_training - Step 7181: {'lr': 0.0004760016406580789, 'samples': 3677184, 'steps': 7181, 'batch_loss/train': 0.8704931819811463} +12/23/2021 22:17:23 - INFO - codeparrot_training - Step 7182: {'lr': 0.00047599490117609727, 'samples': 3677696, 'steps': 7182, 'batch_loss/train': 0.8840293912217021} +12/23/2021 22:17:33 - INFO - codeparrot_training - Step 7183: {'lr': 0.00047598816079564524, 'samples': 3678208, 'steps': 7183, 'batch_loss/train': 0.6847890492063016} +12/23/2021 22:17:47 - INFO - codeparrot_training - Step 7184: {'lr': 0.00047598141951674957, 'samples': 3678720, 'steps': 7184, 'batch_loss/train': 0.7825206872075796} +12/23/2021 22:17:57 - INFO - codeparrot_training - Step 7185: {'lr': 0.0004759746773394372, 'samples': 3679232, 'steps': 7185, 'batch_loss/train': 0.874156286008656} +12/23/2021 22:18:08 - INFO - codeparrot_training - Step 7186: {'lr': 0.0004759679342637347, 'samples': 3679744, 'steps': 7186, 'batch_loss/train': 0.8914982099086046} +12/23/2021 22:18:20 - INFO - codeparrot_training - Step 7187: {'lr': 0.0004759611902896691, 'samples': 3680256, 'steps': 7187, 'batch_loss/train': 0.9215790024027228} +12/23/2021 22:18:30 - INFO - codeparrot_training - Step 7188: {'lr': 0.0004759544454172671, 'samples': 3680768, 'steps': 7188, 'batch_loss/train': 0.8413446652702987} +12/23/2021 22:18:41 - INFO - codeparrot_training - Step 7189: {'lr': 0.0004759476996465556, 'samples': 3681280, 'steps': 7189, 'batch_loss/train': 0.8209305707132444} +12/23/2021 22:18:53 - INFO - codeparrot_training - Step 7190: {'lr': 0.00047594095297756123, 'samples': 3681792, 'steps': 7190, 'batch_loss/train': 0.6359039228409529} +12/23/2021 22:19:04 - INFO - codeparrot_training - Step 7191: {'lr': 0.00047593420541031106, 'samples': 3682304, 'steps': 7191, 'batch_loss/train': 0.8565322319045663} +12/23/2021 22:19:14 - INFO - codeparrot_training - Step 7192: {'lr': 0.00047592745694483176, 'samples': 3682816, 'steps': 7192, 'batch_loss/train': 0.9060135236941278} +12/23/2021 22:19:25 - INFO - codeparrot_training - Step 7193: {'lr': 0.0004759207075811502, 'samples': 3683328, 'steps': 7193, 'batch_loss/train': 1.0113025628961623} +12/23/2021 22:19:38 - INFO - codeparrot_training - Step 7194: {'lr': 0.00047591395731929324, 'samples': 3683840, 'steps': 7194, 'batch_loss/train': 0.8120838310569525} +12/23/2021 22:19:49 - INFO - codeparrot_training - Step 7195: {'lr': 0.00047590720615928763, 'samples': 3684352, 'steps': 7195, 'batch_loss/train': 0.8515205564908683} +12/23/2021 22:19:59 - INFO - codeparrot_training - Step 7196: {'lr': 0.0004759004541011603, 'samples': 3684864, 'steps': 7196, 'batch_loss/train': 0.7362040920415893} +12/23/2021 22:20:11 - INFO - codeparrot_training - Step 7197: {'lr': 0.000475893701144938, 'samples': 3685376, 'steps': 7197, 'batch_loss/train': 0.8056928256992251} +12/23/2021 22:20:22 - INFO - codeparrot_training - Step 7198: {'lr': 0.00047588694729064774, 'samples': 3685888, 'steps': 7198, 'batch_loss/train': 0.8515034643933177} +12/23/2021 22:20:32 - INFO - codeparrot_training - Step 7199: {'lr': 0.00047588019253831627, 'samples': 3686400, 'steps': 7199, 'batch_loss/train': 1.0531422086060047} +12/23/2021 22:20:46 - INFO - codeparrot_training - Step 7200: {'lr': 0.0004758734368879704, 'samples': 3686912, 'steps': 7200, 'batch_loss/train': 0.8940094448626041} +12/23/2021 22:20:57 - INFO - codeparrot_training - Step 7201: {'lr': 0.000475866680339637, 'samples': 3687424, 'steps': 7201, 'batch_loss/train': 0.9129518223926425} +12/23/2021 22:21:07 - INFO - codeparrot_training - Step 7202: {'lr': 0.000475859922893343, 'samples': 3687936, 'steps': 7202, 'batch_loss/train': 0.9280796656385064} +12/23/2021 22:21:19 - INFO - codeparrot_training - Step 7203: {'lr': 0.0004758531645491152, 'samples': 3688448, 'steps': 7203, 'batch_loss/train': 0.938473570626229} +12/23/2021 22:21:30 - INFO - codeparrot_training - Step 7204: {'lr': 0.00047584640530698056, 'samples': 3688960, 'steps': 7204, 'batch_loss/train': 0.8720258586108685} +12/23/2021 22:21:41 - INFO - codeparrot_training - Step 7205: {'lr': 0.00047583964516696584, 'samples': 3689472, 'steps': 7205, 'batch_loss/train': 0.8892642483115196} +12/23/2021 22:21:51 - INFO - codeparrot_training - Step 7206: {'lr': 0.000475832884129098, 'samples': 3689984, 'steps': 7206, 'batch_loss/train': 0.7553557874634862} +12/23/2021 22:22:03 - INFO - codeparrot_training - Step 7207: {'lr': 0.0004758261221934038, 'samples': 3690496, 'steps': 7207, 'batch_loss/train': 0.8047151919454336} +12/23/2021 22:22:14 - INFO - codeparrot_training - Step 7208: {'lr': 0.00047581935935991026, 'samples': 3691008, 'steps': 7208, 'batch_loss/train': 0.8520024251192808} +12/23/2021 22:22:24 - INFO - codeparrot_training - Step 7209: {'lr': 0.0004758125956286442, 'samples': 3691520, 'steps': 7209, 'batch_loss/train': 0.8737692488357425} +12/23/2021 22:22:38 - INFO - codeparrot_training - Step 7210: {'lr': 0.00047580583099963254, 'samples': 3692032, 'steps': 7210, 'batch_loss/train': 0.8845615470781922} +12/23/2021 22:22:49 - INFO - codeparrot_training - Step 7211: {'lr': 0.00047579906547290217, 'samples': 3692544, 'steps': 7211, 'batch_loss/train': 0.8318544521462172} +12/23/2021 22:23:00 - INFO - codeparrot_training - Step 7212: {'lr': 0.00047579229904848, 'samples': 3693056, 'steps': 7212, 'batch_loss/train': 0.9091790029779077} +12/23/2021 22:23:12 - INFO - codeparrot_training - Step 7213: {'lr': 0.00047578553172639284, 'samples': 3693568, 'steps': 7213, 'batch_loss/train': 0.7906980393454432} +12/23/2021 22:23:22 - INFO - codeparrot_training - Step 7214: {'lr': 0.00047577876350666766, 'samples': 3694080, 'steps': 7214, 'batch_loss/train': 0.9186754077672958} +12/23/2021 22:23:33 - INFO - codeparrot_training - Step 7215: {'lr': 0.0004757719943893314, 'samples': 3694592, 'steps': 7215, 'batch_loss/train': 0.799704628996551} +12/23/2021 22:23:43 - INFO - codeparrot_training - Step 7216: {'lr': 0.00047576522437441093, 'samples': 3695104, 'steps': 7216, 'batch_loss/train': 0.8131472829263657} +12/23/2021 22:23:55 - INFO - codeparrot_training - Step 7217: {'lr': 0.00047575845346193316, 'samples': 3695616, 'steps': 7217, 'batch_loss/train': 0.9083469267934561} +12/23/2021 22:24:06 - INFO - codeparrot_training - Step 7218: {'lr': 0.00047575168165192507, 'samples': 3696128, 'steps': 7218, 'batch_loss/train': 0.7571154278703034} +12/23/2021 22:24:17 - INFO - codeparrot_training - Step 7219: {'lr': 0.00047574490894441345, 'samples': 3696640, 'steps': 7219, 'batch_loss/train': 0.8386464649811387} +12/23/2021 22:24:29 - INFO - codeparrot_training - Step 7220: {'lr': 0.00047573813533942534, 'samples': 3697152, 'steps': 7220, 'batch_loss/train': 0.9241423960775137} +12/23/2021 22:24:39 - INFO - codeparrot_training - Step 7221: {'lr': 0.00047573136083698764, 'samples': 3697664, 'steps': 7221, 'batch_loss/train': 0.8815926527604461} +12/23/2021 22:24:50 - INFO - codeparrot_training - Step 7222: {'lr': 0.0004757245854371274, 'samples': 3698176, 'steps': 7222, 'batch_loss/train': 0.826761425472796} +12/23/2021 22:25:04 - INFO - codeparrot_training - Step 7223: {'lr': 0.0004757178091398713, 'samples': 3698688, 'steps': 7223, 'batch_loss/train': 0.7886532364645973} +12/23/2021 22:25:14 - INFO - codeparrot_training - Step 7224: {'lr': 0.00047571103194524644, 'samples': 3699200, 'steps': 7224, 'batch_loss/train': 0.9342003371566534} +12/23/2021 22:25:25 - INFO - codeparrot_training - Step 7225: {'lr': 0.00047570425385327973, 'samples': 3699712, 'steps': 7225, 'batch_loss/train': 0.8868701467290521} +12/23/2021 22:25:37 - INFO - codeparrot_training - Step 7226: {'lr': 0.0004756974748639982, 'samples': 3700224, 'steps': 7226, 'batch_loss/train': 0.8471612092107534} +12/23/2021 22:25:48 - INFO - codeparrot_training - Step 7227: {'lr': 0.00047569069497742865, 'samples': 3700736, 'steps': 7227, 'batch_loss/train': 0.7988603012636304} +12/23/2021 22:25:58 - INFO - codeparrot_training - Step 7228: {'lr': 0.0004756839141935982, 'samples': 3701248, 'steps': 7228, 'batch_loss/train': 0.8808815320953727} +12/23/2021 22:26:09 - INFO - codeparrot_training - Step 7229: {'lr': 0.00047567713251253364, 'samples': 3701760, 'steps': 7229, 'batch_loss/train': 0.8638949180021882} +12/23/2021 22:26:23 - INFO - codeparrot_training - Step 7230: {'lr': 0.00047567034993426205, 'samples': 3702272, 'steps': 7230, 'batch_loss/train': 0.8415845250710845} +12/23/2021 22:26:33 - INFO - codeparrot_training - Step 7231: {'lr': 0.0004756635664588104, 'samples': 3702784, 'steps': 7231, 'batch_loss/train': 0.8353385077789426} +12/23/2021 22:26:44 - INFO - codeparrot_training - Step 7232: {'lr': 0.00047565678208620556, 'samples': 3703296, 'steps': 7232, 'batch_loss/train': 0.8616458000615239} +12/23/2021 22:26:56 - INFO - codeparrot_training - Step 7233: {'lr': 0.0004756499968164746, 'samples': 3703808, 'steps': 7233, 'batch_loss/train': 0.8378886980935931} +12/23/2021 22:27:06 - INFO - codeparrot_training - Step 7234: {'lr': 0.00047564321064964436, 'samples': 3704320, 'steps': 7234, 'batch_loss/train': 0.8354263212531805} +12/23/2021 22:27:17 - INFO - codeparrot_training - Step 7235: {'lr': 0.000475636423585742, 'samples': 3704832, 'steps': 7235, 'batch_loss/train': 0.8735621180385351} +12/23/2021 22:27:29 - INFO - codeparrot_training - Step 7236: {'lr': 0.0004756296356247944, 'samples': 3705344, 'steps': 7236, 'batch_loss/train': 0.9141302180942148} +12/23/2021 22:27:40 - INFO - codeparrot_training - Step 7237: {'lr': 0.00047562284676682856, 'samples': 3705856, 'steps': 7237, 'batch_loss/train': 0.9488301025703549} +12/23/2021 22:27:50 - INFO - codeparrot_training - Step 7238: {'lr': 0.0004756160570118715, 'samples': 3706368, 'steps': 7238, 'batch_loss/train': 0.8907468193210661} +12/23/2021 22:28:01 - INFO - codeparrot_training - Step 7239: {'lr': 0.00047560926635995015, 'samples': 3706880, 'steps': 7239, 'batch_loss/train': 0.8251990715507418} +12/23/2021 22:28:14 - INFO - codeparrot_training - Step 7240: {'lr': 0.00047560247481109153, 'samples': 3707392, 'steps': 7240, 'batch_loss/train': 0.867845562286675} +12/23/2021 22:28:25 - INFO - codeparrot_training - Step 7241: {'lr': 0.0004755956823653227, 'samples': 3707904, 'steps': 7241, 'batch_loss/train': 0.7553672262001783} +12/23/2021 22:28:35 - INFO - codeparrot_training - Step 7242: {'lr': 0.0004755888890226706, 'samples': 3708416, 'steps': 7242, 'batch_loss/train': 0.8331583528779447} +12/23/2021 22:28:47 - INFO - codeparrot_training - Step 7243: {'lr': 0.0004755820947831623, 'samples': 3708928, 'steps': 7243, 'batch_loss/train': 0.9076865054666996} +12/23/2021 22:28:58 - INFO - codeparrot_training - Step 7244: {'lr': 0.00047557529964682466, 'samples': 3709440, 'steps': 7244, 'batch_loss/train': 0.8868251238018274} +12/23/2021 22:29:09 - INFO - codeparrot_training - Step 7245: {'lr': 0.0004755685036136849, 'samples': 3709952, 'steps': 7245, 'batch_loss/train': 0.8523754719644785} +12/23/2021 22:29:20 - INFO - codeparrot_training - Step 7246: {'lr': 0.0004755617066837699, 'samples': 3710464, 'steps': 7246, 'batch_loss/train': 0.8600508235394955} +12/23/2021 22:29:31 - INFO - codeparrot_training - Step 7247: {'lr': 0.0004755549088571067, 'samples': 3710976, 'steps': 7247, 'batch_loss/train': 0.8314569815993309} +12/23/2021 22:29:42 - INFO - codeparrot_training - Step 7248: {'lr': 0.0004755481101337224, 'samples': 3711488, 'steps': 7248, 'batch_loss/train': 0.8691320922225714} +12/23/2021 22:29:52 - INFO - codeparrot_training - Step 7249: {'lr': 0.0004755413105136439, 'samples': 3712000, 'steps': 7249, 'batch_loss/train': 0.8862196244299412} +12/23/2021 22:30:06 - INFO - codeparrot_training - Step 7250: {'lr': 0.0004755345099968984, 'samples': 3712512, 'steps': 7250, 'batch_loss/train': 0.9124447191134095} +12/23/2021 22:30:17 - INFO - codeparrot_training - Step 7251: {'lr': 0.0004755277085835128, 'samples': 3713024, 'steps': 7251, 'batch_loss/train': 0.9372757142409682} +12/23/2021 22:30:27 - INFO - codeparrot_training - Step 7252: {'lr': 0.0004755209062735142, 'samples': 3713536, 'steps': 7252, 'batch_loss/train': 0.8732603499665856} +12/23/2021 22:30:39 - INFO - codeparrot_training - Step 7253: {'lr': 0.0004755141030669297, 'samples': 3714048, 'steps': 7253, 'batch_loss/train': 1.0275812204927206} +12/23/2021 22:30:50 - INFO - codeparrot_training - Step 7254: {'lr': 0.0004755072989637862, 'samples': 3714560, 'steps': 7254, 'batch_loss/train': 0.827024769037962} +12/23/2021 22:31:01 - INFO - codeparrot_training - Step 7255: {'lr': 0.0004755004939641108, 'samples': 3715072, 'steps': 7255, 'batch_loss/train': 0.7855765447020531} +12/23/2021 22:31:13 - INFO - codeparrot_training - Step 7256: {'lr': 0.0004754936880679306, 'samples': 3715584, 'steps': 7256, 'batch_loss/train': 0.8148875897750258} +12/23/2021 22:31:23 - INFO - codeparrot_training - Step 7257: {'lr': 0.00047548688127527265, 'samples': 3716096, 'steps': 7257, 'batch_loss/train': 0.7509629909764044} +12/23/2021 22:31:34 - INFO - codeparrot_training - Step 7258: {'lr': 0.00047548007358616404, 'samples': 3716608, 'steps': 7258, 'batch_loss/train': 0.8803666736930609} +12/23/2021 22:31:48 - INFO - codeparrot_training - Step 7259: {'lr': 0.0004754732650006317, 'samples': 3717120, 'steps': 7259, 'batch_loss/train': 0.9197321804240346} +12/23/2021 22:31:58 - INFO - codeparrot_training - Step 7260: {'lr': 0.0004754664555187029, 'samples': 3717632, 'steps': 7260, 'batch_loss/train': 0.8115100637078285} +12/23/2021 22:32:09 - INFO - codeparrot_training - Step 7261: {'lr': 0.0004754596451404045, 'samples': 3718144, 'steps': 7261, 'batch_loss/train': 0.8382074618712068} +12/23/2021 22:32:20 - INFO - codeparrot_training - Step 7262: {'lr': 0.0004754528338657638, 'samples': 3718656, 'steps': 7262, 'batch_loss/train': 0.7726742569357157} +12/23/2021 22:32:32 - INFO - codeparrot_training - Step 7263: {'lr': 0.00047544602169480765, 'samples': 3719168, 'steps': 7263, 'batch_loss/train': 0.8978134477511048} +12/23/2021 22:32:42 - INFO - codeparrot_training - Step 7264: {'lr': 0.00047543920862756333, 'samples': 3719680, 'steps': 7264, 'batch_loss/train': 0.853446327149868} +12/23/2021 22:32:53 - INFO - codeparrot_training - Step 7265: {'lr': 0.00047543239466405786, 'samples': 3720192, 'steps': 7265, 'batch_loss/train': 0.7265933325979859} +12/23/2021 22:33:05 - INFO - codeparrot_training - Step 7266: {'lr': 0.00047542557980431824, 'samples': 3720704, 'steps': 7266, 'batch_loss/train': 0.8611368429847062} +12/23/2021 22:33:15 - INFO - codeparrot_training - Step 7267: {'lr': 0.0004754187640483716, 'samples': 3721216, 'steps': 7267, 'batch_loss/train': 0.7807685635052621} +12/23/2021 22:33:26 - INFO - codeparrot_training - Step 7268: {'lr': 0.00047541194739624513, 'samples': 3721728, 'steps': 7268, 'batch_loss/train': 0.8692255239002407} +12/23/2021 22:33:40 - INFO - codeparrot_training - Step 7269: {'lr': 0.00047540512984796585, 'samples': 3722240, 'steps': 7269, 'batch_loss/train': 0.7750208945944905} +12/23/2021 22:33:50 - INFO - codeparrot_training - Step 7270: {'lr': 0.0004753983114035609, 'samples': 3722752, 'steps': 7270, 'batch_loss/train': 0.8954297536984086} +12/23/2021 22:34:01 - INFO - codeparrot_training - Step 7271: {'lr': 0.0004753914920630573, 'samples': 3723264, 'steps': 7271, 'batch_loss/train': 0.873485129326582} +12/23/2021 22:34:12 - INFO - codeparrot_training - Step 7272: {'lr': 0.00047538467182648233, 'samples': 3723776, 'steps': 7272, 'batch_loss/train': 0.8681098688393831} +12/23/2021 22:34:24 - INFO - codeparrot_training - Step 7273: {'lr': 0.000475377850693863, 'samples': 3724288, 'steps': 7273, 'batch_loss/train': 0.7788551856065169} +12/23/2021 22:34:34 - INFO - codeparrot_training - Step 7274: {'lr': 0.0004753710286652264, 'samples': 3724800, 'steps': 7274, 'batch_loss/train': 0.8729011369869113} +12/23/2021 22:34:45 - INFO - codeparrot_training - Step 7275: {'lr': 0.00047536420574059967, 'samples': 3725312, 'steps': 7275, 'batch_loss/train': 0.8426068043336272} +12/23/2021 22:34:57 - INFO - codeparrot_training - Step 7276: {'lr': 0.00047535738192001, 'samples': 3725824, 'steps': 7276, 'batch_loss/train': 0.784079541452229} +12/23/2021 22:35:07 - INFO - codeparrot_training - Step 7277: {'lr': 0.0004753505572034844, 'samples': 3726336, 'steps': 7277, 'batch_loss/train': 0.9100342178717256} +12/23/2021 22:35:18 - INFO - codeparrot_training - Step 7278: {'lr': 0.00047534373159105014, 'samples': 3726848, 'steps': 7278, 'batch_loss/train': 0.8450402459129691} +12/23/2021 22:35:32 - INFO - codeparrot_training - Step 7279: {'lr': 0.00047533690508273427, 'samples': 3727360, 'steps': 7279, 'batch_loss/train': 0.8396212072111666} +12/23/2021 22:35:42 - INFO - codeparrot_training - Step 7280: {'lr': 0.0004753300776785639, 'samples': 3727872, 'steps': 7280, 'batch_loss/train': 0.8835293436422944} +12/23/2021 22:35:53 - INFO - codeparrot_training - Step 7281: {'lr': 0.0004753232493785663, 'samples': 3728384, 'steps': 7281, 'batch_loss/train': 0.9276621453464031} +12/23/2021 22:36:03 - INFO - codeparrot_training - Step 7282: {'lr': 0.00047531642018276845, 'samples': 3728896, 'steps': 7282, 'batch_loss/train': 0.8553156089037657} +12/23/2021 22:36:15 - INFO - codeparrot_training - Step 7283: {'lr': 0.0004753095900911977, 'samples': 3729408, 'steps': 7283, 'batch_loss/train': 0.798710091970861} +12/23/2021 22:36:26 - INFO - codeparrot_training - Step 7284: {'lr': 0.000475302759103881, 'samples': 3729920, 'steps': 7284, 'batch_loss/train': 0.9327971581369638} +12/23/2021 22:36:37 - INFO - codeparrot_training - Step 7285: {'lr': 0.0004752959272208457, 'samples': 3730432, 'steps': 7285, 'batch_loss/train': 0.772613943554461} +12/23/2021 22:36:51 - INFO - codeparrot_training - Step 7286: {'lr': 0.0004752890944421188, 'samples': 3730944, 'steps': 7286, 'batch_loss/train': 0.7839570548385382} +12/23/2021 22:37:01 - INFO - codeparrot_training - Step 7287: {'lr': 0.0004752822607677275, 'samples': 3731456, 'steps': 7287, 'batch_loss/train': 0.8701863288879395} +12/23/2021 22:37:12 - INFO - codeparrot_training - Step 7288: {'lr': 0.00047527542619769903, 'samples': 3731968, 'steps': 7288, 'batch_loss/train': 0.863541450118646} +12/23/2021 22:37:24 - INFO - codeparrot_training - Step 7289: {'lr': 0.00047526859073206053, 'samples': 3732480, 'steps': 7289, 'batch_loss/train': 1.0777589045464993} +12/23/2021 22:37:34 - INFO - codeparrot_training - Step 7290: {'lr': 0.0004752617543708391, 'samples': 3732992, 'steps': 7290, 'batch_loss/train': 0.7905243104323745} +12/23/2021 22:37:45 - INFO - codeparrot_training - Step 7291: {'lr': 0.0004752549171140621, 'samples': 3733504, 'steps': 7291, 'batch_loss/train': 0.855791718699038} +12/23/2021 22:37:56 - INFO - codeparrot_training - Step 7292: {'lr': 0.00047524807896175646, 'samples': 3734016, 'steps': 7292, 'batch_loss/train': 0.8326421505771577} +12/23/2021 22:38:08 - INFO - codeparrot_training - Step 7293: {'lr': 0.00047524123991394965, 'samples': 3734528, 'steps': 7293, 'batch_loss/train': 0.8702390557155013} +12/23/2021 22:38:18 - INFO - codeparrot_training - Step 7294: {'lr': 0.0004752343999706686, 'samples': 3735040, 'steps': 7294, 'batch_loss/train': 0.8785776263102889} +12/23/2021 22:38:29 - INFO - codeparrot_training - Step 7295: {'lr': 0.0004752275591319408, 'samples': 3735552, 'steps': 7295, 'batch_loss/train': 0.7786138607189059} +12/23/2021 22:38:41 - INFO - codeparrot_training - Step 7296: {'lr': 0.0004752207173977932, 'samples': 3736064, 'steps': 7296, 'batch_loss/train': 0.8927205875515938} +12/23/2021 22:38:51 - INFO - codeparrot_training - Step 7297: {'lr': 0.00047521387476825294, 'samples': 3736576, 'steps': 7297, 'batch_loss/train': 0.8611820712685585} +12/23/2021 22:39:02 - INFO - codeparrot_training - Step 7298: {'lr': 0.00047520703124334743, 'samples': 3737088, 'steps': 7298, 'batch_loss/train': 0.8873681016266346} +12/23/2021 22:39:15 - INFO - codeparrot_training - Step 7299: {'lr': 0.00047520018682310387, 'samples': 3737600, 'steps': 7299, 'batch_loss/train': 0.815628195181489} +12/23/2021 22:39:26 - INFO - codeparrot_training - Step 7300: {'lr': 0.0004751933415075493, 'samples': 3738112, 'steps': 7300, 'batch_loss/train': 0.8931346135213971} +12/23/2021 22:39:36 - INFO - codeparrot_training - Step 7301: {'lr': 0.00047518649529671103, 'samples': 3738624, 'steps': 7301, 'batch_loss/train': 0.9324156614020467} +12/23/2021 22:39:48 - INFO - codeparrot_training - Step 7302: {'lr': 0.0004751796481906163, 'samples': 3739136, 'steps': 7302, 'batch_loss/train': 0.7534533790312707} +12/23/2021 22:39:59 - INFO - codeparrot_training - Step 7303: {'lr': 0.0004751728001892924, 'samples': 3739648, 'steps': 7303, 'batch_loss/train': 0.8913266519084573} +12/23/2021 22:40:09 - INFO - codeparrot_training - Step 7304: {'lr': 0.0004751659512927664, 'samples': 3740160, 'steps': 7304, 'batch_loss/train': 0.8806086229160428} +12/23/2021 22:40:20 - INFO - codeparrot_training - Step 7305: {'lr': 0.00047515910150106554, 'samples': 3740672, 'steps': 7305, 'batch_loss/train': 0.8239913210272789} +12/23/2021 22:40:32 - INFO - codeparrot_training - Step 7306: {'lr': 0.0004751522508142172, 'samples': 3741184, 'steps': 7306, 'batch_loss/train': 0.8270023395307362} +12/23/2021 22:40:42 - INFO - codeparrot_training - Step 7307: {'lr': 0.00047514539923224856, 'samples': 3741696, 'steps': 7307, 'batch_loss/train': 0.7447771745501086} +12/23/2021 22:40:53 - INFO - codeparrot_training - Step 7308: {'lr': 0.0004751385467551867, 'samples': 3742208, 'steps': 7308, 'batch_loss/train': 1.0470271725207567} +12/23/2021 22:41:07 - INFO - codeparrot_training - Step 7309: {'lr': 0.0004751316933830591, 'samples': 3742720, 'steps': 7309, 'batch_loss/train': 0.9937335019931197} +12/23/2021 22:41:18 - INFO - codeparrot_training - Step 7310: {'lr': 0.00047512483911589276, 'samples': 3743232, 'steps': 7310, 'batch_loss/train': 0.8160195662640035} +12/23/2021 22:41:28 - INFO - codeparrot_training - Step 7311: {'lr': 0.0004751179839537152, 'samples': 3743744, 'steps': 7311, 'batch_loss/train': 0.8612446011975408} +12/23/2021 22:41:39 - INFO - codeparrot_training - Step 7312: {'lr': 0.00047511112789655344, 'samples': 3744256, 'steps': 7312, 'batch_loss/train': 0.8350908271968365} +12/23/2021 22:41:51 - INFO - codeparrot_training - Step 7313: {'lr': 0.00047510427094443487, 'samples': 3744768, 'steps': 7313, 'batch_loss/train': 0.749801623634994} +12/23/2021 22:42:01 - INFO - codeparrot_training - Step 7314: {'lr': 0.00047509741309738676, 'samples': 3745280, 'steps': 7314, 'batch_loss/train': 0.8744569020345807} +12/23/2021 22:42:12 - INFO - codeparrot_training - Step 7315: {'lr': 0.00047509055435543624, 'samples': 3745792, 'steps': 7315, 'batch_loss/train': 0.8569806087762117} +12/23/2021 22:42:26 - INFO - codeparrot_training - Step 7316: {'lr': 0.00047508369471861066, 'samples': 3746304, 'steps': 7316, 'batch_loss/train': 0.8492662720382214} +12/23/2021 22:42:36 - INFO - codeparrot_training - Step 7317: {'lr': 0.00047507683418693736, 'samples': 3746816, 'steps': 7317, 'batch_loss/train': 0.8124135341495275} +12/23/2021 22:42:47 - INFO - codeparrot_training - Step 7318: {'lr': 0.0004750699727604436, 'samples': 3747328, 'steps': 7318, 'batch_loss/train': 0.8426740779541433} +12/23/2021 22:42:59 - INFO - codeparrot_training - Step 7319: {'lr': 0.00047506311043915653, 'samples': 3747840, 'steps': 7319, 'batch_loss/train': 0.837029417976737} +12/23/2021 22:43:09 - INFO - codeparrot_training - Step 7320: {'lr': 0.00047505624722310355, 'samples': 3748352, 'steps': 7320, 'batch_loss/train': 0.8152125794440508} +12/23/2021 22:43:20 - INFO - codeparrot_training - Step 7321: {'lr': 0.0004750493831123119, 'samples': 3748864, 'steps': 7321, 'batch_loss/train': 0.8973736839834601} +12/23/2021 22:43:32 - INFO - codeparrot_training - Step 7322: {'lr': 0.0004750425181068089, 'samples': 3749376, 'steps': 7322, 'batch_loss/train': 0.8462751721963286} +12/23/2021 22:43:43 - INFO - codeparrot_training - Step 7323: {'lr': 0.0004750356522066218, 'samples': 3749888, 'steps': 7323, 'batch_loss/train': 0.8495282789226621} +12/23/2021 22:43:53 - INFO - codeparrot_training - Step 7324: {'lr': 0.00047502878541177783, 'samples': 3750400, 'steps': 7324, 'batch_loss/train': 0.8889597551897168} +12/23/2021 22:44:04 - INFO - codeparrot_training - Step 7325: {'lr': 0.0004750219177223045, 'samples': 3750912, 'steps': 7325, 'batch_loss/train': 0.9577281167730689} +12/23/2021 22:44:18 - INFO - codeparrot_training - Step 7326: {'lr': 0.000475015049138229, 'samples': 3751424, 'steps': 7326, 'batch_loss/train': 0.8170597348362207} +12/23/2021 22:44:29 - INFO - codeparrot_training - Step 7327: {'lr': 0.0004750081796595785, 'samples': 3751936, 'steps': 7327, 'batch_loss/train': 0.6163395744515583} +12/23/2021 22:44:39 - INFO - codeparrot_training - Step 7328: {'lr': 0.00047500130928638054, 'samples': 3752448, 'steps': 7328, 'batch_loss/train': 0.6613188716582954} +12/23/2021 22:44:51 - INFO - codeparrot_training - Step 7329: {'lr': 0.0004749944380186623, 'samples': 3752960, 'steps': 7329, 'batch_loss/train': 0.8462913623079658} +12/23/2021 22:45:02 - INFO - codeparrot_training - Step 7330: {'lr': 0.0004749875658564511, 'samples': 3753472, 'steps': 7330, 'batch_loss/train': 0.7175891302758828} +12/23/2021 22:45:12 - INFO - codeparrot_training - Step 7331: {'lr': 0.0004749806927997744, 'samples': 3753984, 'steps': 7331, 'batch_loss/train': 0.85547827440314} +12/23/2021 22:45:23 - INFO - codeparrot_training - Step 7332: {'lr': 0.0004749738188486593, 'samples': 3754496, 'steps': 7332, 'batch_loss/train': 0.861702342517674} +12/23/2021 22:45:35 - INFO - codeparrot_training - Step 7333: {'lr': 0.0004749669440031333, 'samples': 3755008, 'steps': 7333, 'batch_loss/train': 0.9068438038229942} +12/23/2021 22:45:46 - INFO - codeparrot_training - Step 7334: {'lr': 0.0004749600682632237, 'samples': 3755520, 'steps': 7334, 'batch_loss/train': 0.8125765137374401} +12/23/2021 22:45:56 - INFO - codeparrot_training - Step 7335: {'lr': 0.00047495319162895773, 'samples': 3756032, 'steps': 7335, 'batch_loss/train': 0.8958035344257951} +12/23/2021 22:46:10 - INFO - codeparrot_training - Step 7336: {'lr': 0.0004749463141003628, 'samples': 3756544, 'steps': 7336, 'batch_loss/train': 0.8079745285212994} +12/23/2021 22:46:21 - INFO - codeparrot_training - Step 7337: {'lr': 0.00047493943567746634, 'samples': 3757056, 'steps': 7337, 'batch_loss/train': 0.8728948291391134} +12/23/2021 22:46:31 - INFO - codeparrot_training - Step 7338: {'lr': 0.00047493255636029563, 'samples': 3757568, 'steps': 7338, 'batch_loss/train': 0.7914179407525808} +12/23/2021 22:46:44 - INFO - codeparrot_training - Step 7339: {'lr': 0.00047492567614887795, 'samples': 3758080, 'steps': 7339, 'batch_loss/train': 0.8991232896223664} +12/23/2021 22:46:54 - INFO - codeparrot_training - Step 7340: {'lr': 0.0004749187950432407, 'samples': 3758592, 'steps': 7340, 'batch_loss/train': 0.8722152232658118} +12/23/2021 22:47:05 - INFO - codeparrot_training - Step 7341: {'lr': 0.0004749119130434113, 'samples': 3759104, 'steps': 7341, 'batch_loss/train': 0.8289101738482714} +12/23/2021 22:47:17 - INFO - codeparrot_training - Step 7342: {'lr': 0.000474905030149417, 'samples': 3759616, 'steps': 7342, 'batch_loss/train': 0.7520562279969454} +12/23/2021 22:47:28 - INFO - codeparrot_training - Step 7343: {'lr': 0.00047489814636128525, 'samples': 3760128, 'steps': 7343, 'batch_loss/train': 0.7980201272293925} +12/23/2021 22:47:38 - INFO - codeparrot_training - Step 7344: {'lr': 0.0004748912616790434, 'samples': 3760640, 'steps': 7344, 'batch_loss/train': 0.8371398281306028} +12/23/2021 22:47:49 - INFO - codeparrot_training - Step 7345: {'lr': 0.0004748843761027188, 'samples': 3761152, 'steps': 7345, 'batch_loss/train': 0.9171497728675604} +12/23/2021 22:48:03 - INFO - codeparrot_training - Step 7346: {'lr': 0.0004748774896323388, 'samples': 3761664, 'steps': 7346, 'batch_loss/train': 0.8294218447990716} +12/23/2021 22:48:13 - INFO - codeparrot_training - Step 7347: {'lr': 0.0004748706022679309, 'samples': 3762176, 'steps': 7347, 'batch_loss/train': 0.9093356309458613} +12/23/2021 22:48:24 - INFO - codeparrot_training - Step 7348: {'lr': 0.0004748637140095223, 'samples': 3762688, 'steps': 7348, 'batch_loss/train': 0.8720782767049968} +12/23/2021 22:48:36 - INFO - codeparrot_training - Step 7349: {'lr': 0.00047485682485714055, 'samples': 3763200, 'steps': 7349, 'batch_loss/train': 0.7351636462844908} +12/23/2021 22:48:46 - INFO - codeparrot_training - Step 7350: {'lr': 0.0004748499348108129, 'samples': 3763712, 'steps': 7350, 'batch_loss/train': 0.7433497526217252} +12/23/2021 22:48:57 - INFO - codeparrot_training - Step 7351: {'lr': 0.00047484304387056687, 'samples': 3764224, 'steps': 7351, 'batch_loss/train': 0.9673277325928211} +12/23/2021 22:49:09 - INFO - codeparrot_training - Step 7352: {'lr': 0.0004748361520364297, 'samples': 3764736, 'steps': 7352, 'batch_loss/train': 0.8353835688903928} +12/23/2021 22:49:20 - INFO - codeparrot_training - Step 7353: {'lr': 0.000474829259308429, 'samples': 3765248, 'steps': 7353, 'batch_loss/train': 0.813460536301136} +12/23/2021 22:49:30 - INFO - codeparrot_training - Step 7354: {'lr': 0.00047482236568659197, 'samples': 3765760, 'steps': 7354, 'batch_loss/train': 0.8580814236775041} +12/23/2021 22:49:41 - INFO - codeparrot_training - Step 7355: {'lr': 0.00047481547117094613, 'samples': 3766272, 'steps': 7355, 'batch_loss/train': 0.8070556549355388} +12/23/2021 22:49:55 - INFO - codeparrot_training - Step 7356: {'lr': 0.00047480857576151893, 'samples': 3766784, 'steps': 7356, 'batch_loss/train': 0.8729828046634793} +12/23/2021 22:50:05 - INFO - codeparrot_training - Step 7357: {'lr': 0.00047480167945833764, 'samples': 3767296, 'steps': 7357, 'batch_loss/train': 0.8907235842198133} +12/23/2021 22:50:16 - INFO - codeparrot_training - Step 7358: {'lr': 0.0004747947822614298, 'samples': 3767808, 'steps': 7358, 'batch_loss/train': 0.9513157475739717} +12/23/2021 22:50:28 - INFO - codeparrot_training - Step 7359: {'lr': 0.00047478788417082277, 'samples': 3768320, 'steps': 7359, 'batch_loss/train': 0.873375347815454} +12/23/2021 22:50:39 - INFO - codeparrot_training - Step 7360: {'lr': 0.00047478098518654395, 'samples': 3768832, 'steps': 7360, 'batch_loss/train': 0.8406261480413377} +12/23/2021 22:50:49 - INFO - codeparrot_training - Step 7361: {'lr': 0.0004747740853086208, 'samples': 3769344, 'steps': 7361, 'batch_loss/train': 0.8101363596506417} +12/23/2021 22:51:02 - INFO - codeparrot_training - Step 7362: {'lr': 0.0004747671845370808, 'samples': 3769856, 'steps': 7362, 'batch_loss/train': 0.8539978386834264} +12/23/2021 22:51:13 - INFO - codeparrot_training - Step 7363: {'lr': 0.0004747602828719514, 'samples': 3770368, 'steps': 7363, 'batch_loss/train': 0.8443191330879927} +12/23/2021 22:51:23 - INFO - codeparrot_training - Step 7364: {'lr': 0.0004747533803132599, 'samples': 3770880, 'steps': 7364, 'batch_loss/train': 0.8802364291623235} +12/23/2021 22:51:34 - INFO - codeparrot_training - Step 7365: {'lr': 0.0004747464768610338, 'samples': 3771392, 'steps': 7365, 'batch_loss/train': 0.9542759573087096} +12/23/2021 22:51:46 - INFO - codeparrot_training - Step 7366: {'lr': 0.00047473957251530064, 'samples': 3771904, 'steps': 7366, 'batch_loss/train': 0.8764359252527356} +12/23/2021 22:51:57 - INFO - codeparrot_training - Step 7367: {'lr': 0.00047473266727608775, 'samples': 3772416, 'steps': 7367, 'batch_loss/train': 0.9336402269545943} +12/23/2021 22:52:07 - INFO - codeparrot_training - Step 7368: {'lr': 0.00047472576114342266, 'samples': 3772928, 'steps': 7368, 'batch_loss/train': 0.7198241604492068} +12/23/2021 22:52:20 - INFO - codeparrot_training - Step 7369: {'lr': 0.0004747188541173328, 'samples': 3773440, 'steps': 7369, 'batch_loss/train': 0.970033866353333} +12/23/2021 22:52:30 - INFO - codeparrot_training - Step 7370: {'lr': 0.0004747119461978456, 'samples': 3773952, 'steps': 7370, 'batch_loss/train': 0.9685698198154569} +12/23/2021 22:52:41 - INFO - codeparrot_training - Step 7371: {'lr': 0.00047470503738498857, 'samples': 3774464, 'steps': 7371, 'batch_loss/train': 0.8583505055867136} +12/23/2021 22:52:53 - INFO - codeparrot_training - Step 7372: {'lr': 0.00047469812767878916, 'samples': 3774976, 'steps': 7372, 'batch_loss/train': 0.862472883425653} +12/23/2021 22:53:04 - INFO - codeparrot_training - Step 7373: {'lr': 0.0004746912170792748, 'samples': 3775488, 'steps': 7373, 'batch_loss/train': 0.91508549451828} +12/23/2021 22:53:14 - INFO - codeparrot_training - Step 7374: {'lr': 0.000474684305586473, 'samples': 3776000, 'steps': 7374, 'batch_loss/train': 0.9404789693653584} +12/23/2021 22:53:25 - INFO - codeparrot_training - Step 7375: {'lr': 0.0004746773932004113, 'samples': 3776512, 'steps': 7375, 'batch_loss/train': 0.9061777601018548} +12/23/2021 22:53:38 - INFO - codeparrot_training - Step 7376: {'lr': 0.0004746704799211171, 'samples': 3777024, 'steps': 7376, 'batch_loss/train': 0.7811345458030701} +12/23/2021 22:53:48 - INFO - codeparrot_training - Step 7377: {'lr': 0.00047466356574861793, 'samples': 3777536, 'steps': 7377, 'batch_loss/train': 0.8822394805029035} +12/23/2021 22:53:59 - INFO - codeparrot_training - Step 7378: {'lr': 0.0004746566506829412, 'samples': 3778048, 'steps': 7378, 'batch_loss/train': 0.8272259030491114} +12/23/2021 22:54:11 - INFO - codeparrot_training - Step 7379: {'lr': 0.00047464973472411445, 'samples': 3778560, 'steps': 7379, 'batch_loss/train': 0.9604299340862781} +12/23/2021 22:54:22 - INFO - codeparrot_training - Step 7380: {'lr': 0.0004746428178721652, 'samples': 3779072, 'steps': 7380, 'batch_loss/train': 0.8207512549124658} +12/23/2021 22:54:33 - INFO - codeparrot_training - Step 7381: {'lr': 0.0004746359001271209, 'samples': 3779584, 'steps': 7381, 'batch_loss/train': 0.8157329778186977} +12/23/2021 22:54:45 - INFO - codeparrot_training - Step 7382: {'lr': 0.00047462898148900915, 'samples': 3780096, 'steps': 7382, 'batch_loss/train': 0.8143916169647127} +12/23/2021 22:54:55 - INFO - codeparrot_training - Step 7383: {'lr': 0.0004746220619578573, 'samples': 3780608, 'steps': 7383, 'batch_loss/train': 0.7699949784437194} +12/23/2021 22:55:06 - INFO - codeparrot_training - Step 7384: {'lr': 0.00047461514153369303, 'samples': 3781120, 'steps': 7384, 'batch_loss/train': 0.743561731884256} +12/23/2021 22:55:19 - INFO - codeparrot_training - Step 7385: {'lr': 0.0004746082202165437, 'samples': 3781632, 'steps': 7385, 'batch_loss/train': 0.8535814071074128} +12/23/2021 22:55:30 - INFO - codeparrot_training - Step 7386: {'lr': 0.0004746012980064369, 'samples': 3782144, 'steps': 7386, 'batch_loss/train': 0.8486644122749567} +12/23/2021 22:55:41 - INFO - codeparrot_training - Step 7387: {'lr': 0.00047459437490340016, 'samples': 3782656, 'steps': 7387, 'batch_loss/train': 0.8407515708822757} +12/23/2021 22:55:51 - INFO - codeparrot_training - Step 7388: {'lr': 0.000474587450907461, 'samples': 3783168, 'steps': 7388, 'batch_loss/train': 0.8715252494439483} +12/23/2021 22:56:04 - INFO - codeparrot_training - Step 7389: {'lr': 0.00047458052601864693, 'samples': 3783680, 'steps': 7389, 'batch_loss/train': 0.9482160545885563} +12/23/2021 22:56:14 - INFO - codeparrot_training - Step 7390: {'lr': 0.0004745736002369855, 'samples': 3784192, 'steps': 7390, 'batch_loss/train': 0.8773426171392202} +12/23/2021 22:56:25 - INFO - codeparrot_training - Step 7391: {'lr': 0.00047456667356250416, 'samples': 3784704, 'steps': 7391, 'batch_loss/train': 0.7511476268991828} +12/23/2021 22:56:37 - INFO - codeparrot_training - Step 7392: {'lr': 0.0004745597459952306, 'samples': 3785216, 'steps': 7392, 'batch_loss/train': 0.7284593405202031} +12/23/2021 22:56:47 - INFO - codeparrot_training - Step 7393: {'lr': 0.0004745528175351923, 'samples': 3785728, 'steps': 7393, 'batch_loss/train': 0.8732496672309935} +12/23/2021 22:56:58 - INFO - codeparrot_training - Step 7394: {'lr': 0.0004745458881824167, 'samples': 3786240, 'steps': 7394, 'batch_loss/train': 0.7921279240399599} +12/23/2021 22:57:11 - INFO - codeparrot_training - Step 7395: {'lr': 0.0004745389579369315, 'samples': 3786752, 'steps': 7395, 'batch_loss/train': 0.9539952492341399} +12/23/2021 22:57:22 - INFO - codeparrot_training - Step 7396: {'lr': 0.00047453202679876417, 'samples': 3787264, 'steps': 7396, 'batch_loss/train': 0.9378670882433653} +12/23/2021 22:57:33 - INFO - codeparrot_training - Step 7397: {'lr': 0.00047452509476794225, 'samples': 3787776, 'steps': 7397, 'batch_loss/train': 0.8647222593426704} +12/23/2021 22:57:45 - INFO - codeparrot_training - Step 7398: {'lr': 0.00047451816184449337, 'samples': 3788288, 'steps': 7398, 'batch_loss/train': 0.7493531173095107} +12/23/2021 22:57:55 - INFO - codeparrot_training - Step 7399: {'lr': 0.00047451122802844503, 'samples': 3788800, 'steps': 7399, 'batch_loss/train': 0.7978656971827149} +12/23/2021 22:58:06 - INFO - codeparrot_training - Step 7400: {'lr': 0.0004745042933198248, 'samples': 3789312, 'steps': 7400, 'batch_loss/train': 0.8925178768113256} +12/23/2021 22:58:17 - INFO - codeparrot_training - Step 7401: {'lr': 0.0004744973577186603, 'samples': 3789824, 'steps': 7401, 'batch_loss/train': 0.8168132922146469} +12/23/2021 22:58:30 - INFO - codeparrot_training - Step 7402: {'lr': 0.0004744904212249791, 'samples': 3790336, 'steps': 7402, 'batch_loss/train': 0.8160746006760746} +12/23/2021 22:58:41 - INFO - codeparrot_training - Step 7403: {'lr': 0.0004744834838388087, 'samples': 3790848, 'steps': 7403, 'batch_loss/train': 1.204260278493166} +12/23/2021 22:58:51 - INFO - codeparrot_training - Step 7404: {'lr': 0.0004744765455601767, 'samples': 3791360, 'steps': 7404, 'batch_loss/train': 0.8363184873014688} +12/23/2021 22:59:03 - INFO - codeparrot_training - Step 7405: {'lr': 0.00047446960638911084, 'samples': 3791872, 'steps': 7405, 'batch_loss/train': 0.711411414318718} +12/23/2021 22:59:14 - INFO - codeparrot_training - Step 7406: {'lr': 0.00047446266632563853, 'samples': 3792384, 'steps': 7406, 'batch_loss/train': 0.8252337276935577} +12/23/2021 22:59:24 - INFO - codeparrot_training - Step 7407: {'lr': 0.00047445572536978733, 'samples': 3792896, 'steps': 7407, 'batch_loss/train': 0.8637401573359966} +12/23/2021 22:59:36 - INFO - codeparrot_training - Step 7408: {'lr': 0.000474448783521585, 'samples': 3793408, 'steps': 7408, 'batch_loss/train': 0.9115225546993315} +12/23/2021 22:59:47 - INFO - codeparrot_training - Step 7409: {'lr': 0.000474441840781059, 'samples': 3793920, 'steps': 7409, 'batch_loss/train': 0.8117505181580782} +12/23/2021 22:59:58 - INFO - codeparrot_training - Step 7410: {'lr': 0.0004744348971482371, 'samples': 3794432, 'steps': 7410, 'batch_loss/train': 0.754987636115402} +12/23/2021 23:00:08 - INFO - codeparrot_training - Step 7411: {'lr': 0.00047442795262314663, 'samples': 3794944, 'steps': 7411, 'batch_loss/train': 0.7899832371622324} +12/23/2021 23:00:20 - INFO - codeparrot_training - Step 7412: {'lr': 0.0004744210072058154, 'samples': 3795456, 'steps': 7412, 'batch_loss/train': 0.84541339520365} +12/23/2021 23:00:31 - INFO - codeparrot_training - Step 7413: {'lr': 0.00047441406089627103, 'samples': 3795968, 'steps': 7413, 'batch_loss/train': 0.8436379255726933} +12/23/2021 23:00:42 - INFO - codeparrot_training - Step 7414: {'lr': 0.0004744071136945411, 'samples': 3796480, 'steps': 7414, 'batch_loss/train': 0.9956261776387691} +12/23/2021 23:00:55 - INFO - codeparrot_training - Step 7415: {'lr': 0.00047440016560065324, 'samples': 3796992, 'steps': 7415, 'batch_loss/train': 0.7731343805789948} +12/23/2021 23:01:05 - INFO - codeparrot_training - Step 7416: {'lr': 0.00047439321661463496, 'samples': 3797504, 'steps': 7416, 'batch_loss/train': 0.8430437203496695} +12/23/2021 23:01:16 - INFO - codeparrot_training - Step 7417: {'lr': 0.0004743862667365141, 'samples': 3798016, 'steps': 7417, 'batch_loss/train': 0.792008189484477} +12/23/2021 23:01:28 - INFO - codeparrot_training - Step 7418: {'lr': 0.0004743793159663181, 'samples': 3798528, 'steps': 7418, 'batch_loss/train': 0.8757421653717756} +12/23/2021 23:01:38 - INFO - codeparrot_training - Step 7419: {'lr': 0.0004743723643040746, 'samples': 3799040, 'steps': 7419, 'batch_loss/train': 0.829989917576313} +12/23/2021 23:01:49 - INFO - codeparrot_training - Step 7420: {'lr': 0.0004743654117498114, 'samples': 3799552, 'steps': 7420, 'batch_loss/train': 0.755330110900104} +12/23/2021 23:02:00 - INFO - codeparrot_training - Step 7421: {'lr': 0.00047435845830355594, 'samples': 3800064, 'steps': 7421, 'batch_loss/train': 0.8273093365132809} +12/23/2021 23:02:12 - INFO - codeparrot_training - Step 7422: {'lr': 0.0004743515039653361, 'samples': 3800576, 'steps': 7422, 'batch_loss/train': 0.829190026037395} +12/23/2021 23:02:22 - INFO - codeparrot_training - Step 7423: {'lr': 0.0004743445487351793, 'samples': 3801088, 'steps': 7423, 'batch_loss/train': 0.7878340044990182} +12/23/2021 23:02:33 - INFO - codeparrot_training - Step 7424: {'lr': 0.0004743375926131133, 'samples': 3801600, 'steps': 7424, 'batch_loss/train': 0.8892932906746864} +12/23/2021 23:02:46 - INFO - codeparrot_training - Step 7425: {'lr': 0.0004743306355991657, 'samples': 3802112, 'steps': 7425, 'batch_loss/train': 0.8961858318652958} +12/23/2021 23:02:57 - INFO - codeparrot_training - Step 7426: {'lr': 0.0004743236776933643, 'samples': 3802624, 'steps': 7426, 'batch_loss/train': 0.9107004450634122} +12/23/2021 23:03:08 - INFO - codeparrot_training - Step 7427: {'lr': 0.0004743167188957366, 'samples': 3803136, 'steps': 7427, 'batch_loss/train': 0.9011791367083788} +12/23/2021 23:03:20 - INFO - codeparrot_training - Step 7428: {'lr': 0.00047430975920631025, 'samples': 3803648, 'steps': 7428, 'batch_loss/train': 0.9647365035489202} +12/23/2021 23:03:31 - INFO - codeparrot_training - Step 7429: {'lr': 0.00047430279862511305, 'samples': 3804160, 'steps': 7429, 'batch_loss/train': 0.9068957790732384} +12/23/2021 23:03:41 - INFO - codeparrot_training - Step 7430: {'lr': 0.0004742958371521727, 'samples': 3804672, 'steps': 7430, 'batch_loss/train': 0.7712791804224253} +12/23/2021 23:03:52 - INFO - codeparrot_training - Step 7431: {'lr': 0.0004742888747875167, 'samples': 3805184, 'steps': 7431, 'batch_loss/train': 0.8925664955750108} +12/23/2021 23:04:05 - INFO - codeparrot_training - Step 7432: {'lr': 0.0004742819115311728, 'samples': 3805696, 'steps': 7432, 'batch_loss/train': 0.7760916077531874} +12/23/2021 23:04:15 - INFO - codeparrot_training - Step 7433: {'lr': 0.0004742749473831688, 'samples': 3806208, 'steps': 7433, 'batch_loss/train': 0.9407868925482035} +12/23/2021 23:04:26 - INFO - codeparrot_training - Step 7434: {'lr': 0.0004742679823435322, 'samples': 3806720, 'steps': 7434, 'batch_loss/train': 0.8700456977821887} +12/23/2021 23:04:38 - INFO - codeparrot_training - Step 7435: {'lr': 0.00047426101641229087, 'samples': 3807232, 'steps': 7435, 'batch_loss/train': 0.9327864646911621} +12/23/2021 23:04:49 - INFO - codeparrot_training - Step 7436: {'lr': 0.00047425404958947236, 'samples': 3807744, 'steps': 7436, 'batch_loss/train': 0.7882370674051344} +12/23/2021 23:04:59 - INFO - codeparrot_training - Step 7437: {'lr': 0.0004742470818751045, 'samples': 3808256, 'steps': 7437, 'batch_loss/train': 0.9563119374215603} +12/23/2021 23:05:11 - INFO - codeparrot_training - Step 7438: {'lr': 0.00047424011326921477, 'samples': 3808768, 'steps': 7438, 'batch_loss/train': 0.8778168261051178} +12/23/2021 23:05:22 - INFO - codeparrot_training - Step 7439: {'lr': 0.00047423314377183115, 'samples': 3809280, 'steps': 7439, 'batch_loss/train': 0.6723882332444191} +12/23/2021 23:05:32 - INFO - codeparrot_training - Step 7440: {'lr': 0.0004742261733829811, 'samples': 3809792, 'steps': 7440, 'batch_loss/train': 0.895679852925241} +12/23/2021 23:05:43 - INFO - codeparrot_training - Step 7441: {'lr': 0.00047421920210269255, 'samples': 3810304, 'steps': 7441, 'batch_loss/train': 0.8850605143234134} +12/23/2021 23:05:57 - INFO - codeparrot_training - Step 7442: {'lr': 0.00047421222993099303, 'samples': 3810816, 'steps': 7442, 'batch_loss/train': 0.8709240341559052} +12/23/2021 23:06:07 - INFO - codeparrot_training - Step 7443: {'lr': 0.0004742052568679104, 'samples': 3811328, 'steps': 7443, 'batch_loss/train': 0.9331055302172899} +12/23/2021 23:06:18 - INFO - codeparrot_training - Step 7444: {'lr': 0.00047419828291347233, 'samples': 3811840, 'steps': 7444, 'batch_loss/train': 0.7031475922558457} +12/23/2021 23:06:30 - INFO - codeparrot_training - Step 7445: {'lr': 0.0004741913080677065, 'samples': 3812352, 'steps': 7445, 'batch_loss/train': 0.999621968716383} +12/23/2021 23:06:40 - INFO - codeparrot_training - Step 7446: {'lr': 0.0004741843323306407, 'samples': 3812864, 'steps': 7446, 'batch_loss/train': 0.9387432718649507} +12/23/2021 23:06:51 - INFO - codeparrot_training - Step 7447: {'lr': 0.0004741773557023027, 'samples': 3813376, 'steps': 7447, 'batch_loss/train': 0.9573125680908561} +12/23/2021 23:07:03 - INFO - codeparrot_training - Step 7448: {'lr': 0.0004741703781827201, 'samples': 3813888, 'steps': 7448, 'batch_loss/train': 0.8253097645938396} +12/23/2021 23:07:14 - INFO - codeparrot_training - Step 7449: {'lr': 0.0004741633997719208, 'samples': 3814400, 'steps': 7449, 'batch_loss/train': 0.6635673111304641} +12/23/2021 23:07:24 - INFO - codeparrot_training - Step 7450: {'lr': 0.00047415642046993236, 'samples': 3814912, 'steps': 7450, 'batch_loss/train': 0.8421465288847685} +12/23/2021 23:07:35 - INFO - codeparrot_training - Step 7451: {'lr': 0.00047414944027678274, 'samples': 3815424, 'steps': 7451, 'batch_loss/train': 0.8995887152850628} +12/23/2021 23:07:47 - INFO - codeparrot_training - Step 7452: {'lr': 0.0004741424591924995, 'samples': 3815936, 'steps': 7452, 'batch_loss/train': 0.828726933337748} +12/23/2021 23:07:58 - INFO - codeparrot_training - Step 7453: {'lr': 0.0004741354772171105, 'samples': 3816448, 'steps': 7453, 'batch_loss/train': 0.9722810578532517} +12/23/2021 23:08:08 - INFO - codeparrot_training - Step 7454: {'lr': 0.00047412849435064347, 'samples': 3816960, 'steps': 7454, 'batch_loss/train': 0.9699595207348466} +12/23/2021 23:08:21 - INFO - codeparrot_training - Step 7455: {'lr': 0.00047412151059312614, 'samples': 3817472, 'steps': 7455, 'batch_loss/train': 0.9019744899123907} +12/23/2021 23:08:32 - INFO - codeparrot_training - Step 7456: {'lr': 0.00047411452594458637, 'samples': 3817984, 'steps': 7456, 'batch_loss/train': 0.7708200737833977} +12/23/2021 23:08:42 - INFO - codeparrot_training - Step 7457: {'lr': 0.0004741075404050519, 'samples': 3818496, 'steps': 7457, 'batch_loss/train': 0.8560521313920617} +12/23/2021 23:08:54 - INFO - codeparrot_training - Step 7458: {'lr': 0.0004741005539745504, 'samples': 3819008, 'steps': 7458, 'batch_loss/train': 0.7005306915380061} +12/23/2021 23:09:05 - INFO - codeparrot_training - Step 7459: {'lr': 0.0004740935666531097, 'samples': 3819520, 'steps': 7459, 'batch_loss/train': 1.1265014135278761} +12/23/2021 23:09:16 - INFO - codeparrot_training - Step 7460: {'lr': 0.00047408657844075767, 'samples': 3820032, 'steps': 7460, 'batch_loss/train': 0.9465006170794368} +12/23/2021 23:09:26 - INFO - codeparrot_training - Step 7461: {'lr': 0.00047407958933752195, 'samples': 3820544, 'steps': 7461, 'batch_loss/train': 0.7627403682563454} +12/23/2021 23:09:38 - INFO - codeparrot_training - Step 7462: {'lr': 0.0004740725993434304, 'samples': 3821056, 'steps': 7462, 'batch_loss/train': 0.8775930041447282} +12/23/2021 23:09:49 - INFO - codeparrot_training - Step 7463: {'lr': 0.00047406560845851076, 'samples': 3821568, 'steps': 7463, 'batch_loss/train': 0.9164081336930394} +12/23/2021 23:09:59 - INFO - codeparrot_training - Step 7464: {'lr': 0.00047405861668279095, 'samples': 3822080, 'steps': 7464, 'batch_loss/train': 0.8048156923614442} +12/23/2021 23:10:13 - INFO - codeparrot_training - Step 7465: {'lr': 0.00047405162401629867, 'samples': 3822592, 'steps': 7465, 'batch_loss/train': 0.8128450093790889} +12/23/2021 23:10:24 - INFO - codeparrot_training - Step 7466: {'lr': 0.00047404463045906166, 'samples': 3823104, 'steps': 7466, 'batch_loss/train': 0.8695209436118603} +12/23/2021 23:10:34 - INFO - codeparrot_training - Step 7467: {'lr': 0.00047403763601110784, 'samples': 3823616, 'steps': 7467, 'batch_loss/train': 0.7706090891733766} +12/23/2021 23:10:47 - INFO - codeparrot_training - Step 7468: {'lr': 0.00047403064067246496, 'samples': 3824128, 'steps': 7468, 'batch_loss/train': 0.7665291712619364} +12/23/2021 23:10:58 - INFO - codeparrot_training - Step 7469: {'lr': 0.0004740236444431609, 'samples': 3824640, 'steps': 7469, 'batch_loss/train': 0.8064218247309327} +12/23/2021 23:11:08 - INFO - codeparrot_training - Step 7470: {'lr': 0.00047401664732322335, 'samples': 3825152, 'steps': 7470, 'batch_loss/train': 0.842583785764873} +12/23/2021 23:11:19 - INFO - codeparrot_training - Step 7471: {'lr': 0.0004740096493126802, 'samples': 3825664, 'steps': 7471, 'batch_loss/train': 2.2097465582191944} +12/23/2021 23:11:32 - INFO - codeparrot_training - Step 7472: {'lr': 0.0004740026504115593, 'samples': 3826176, 'steps': 7472, 'batch_loss/train': 1.2286911169067025} +12/23/2021 23:11:43 - INFO - codeparrot_training - Step 7473: {'lr': 0.0004739956506198885, 'samples': 3826688, 'steps': 7473, 'batch_loss/train': 0.8320282474160194} +12/23/2021 23:11:53 - INFO - codeparrot_training - Step 7474: {'lr': 0.00047398864993769546, 'samples': 3827200, 'steps': 7474, 'batch_loss/train': 0.8919558590278029} +12/23/2021 23:12:04 - INFO - codeparrot_training - Step 7475: {'lr': 0.00047398164836500817, 'samples': 3827712, 'steps': 7475, 'batch_loss/train': 0.8428762156981975} +12/23/2021 23:12:16 - INFO - codeparrot_training - Step 7476: {'lr': 0.0004739746459018544, 'samples': 3828224, 'steps': 7476, 'batch_loss/train': 0.8776055378839374} +12/23/2021 23:12:26 - INFO - codeparrot_training - Step 7477: {'lr': 0.00047396764254826204, 'samples': 3828736, 'steps': 7477, 'batch_loss/train': 0.8241258775815368} +12/23/2021 23:12:37 - INFO - codeparrot_training - Step 7478: {'lr': 0.00047396063830425884, 'samples': 3829248, 'steps': 7478, 'batch_loss/train': 0.8864692654460669} +12/23/2021 23:12:49 - INFO - codeparrot_training - Step 7479: {'lr': 0.00047395363316987275, 'samples': 3829760, 'steps': 7479, 'batch_loss/train': 0.9412556127645075} +12/23/2021 23:13:00 - INFO - codeparrot_training - Step 7480: {'lr': 0.0004739466271451316, 'samples': 3830272, 'steps': 7480, 'batch_loss/train': 0.8506463719531894} +12/23/2021 23:13:10 - INFO - codeparrot_training - Step 7481: {'lr': 0.0004739396202300632, 'samples': 3830784, 'steps': 7481, 'batch_loss/train': 0.7088622117880732} +12/23/2021 23:13:24 - INFO - codeparrot_training - Step 7482: {'lr': 0.0004739326124246954, 'samples': 3831296, 'steps': 7482, 'batch_loss/train': 0.8813962955027819} +12/23/2021 23:13:34 - INFO - codeparrot_training - Step 7483: {'lr': 0.00047392560372905615, 'samples': 3831808, 'steps': 7483, 'batch_loss/train': 0.905711667612195} +12/23/2021 23:13:45 - INFO - codeparrot_training - Step 7484: {'lr': 0.0004739185941431732, 'samples': 3832320, 'steps': 7484, 'batch_loss/train': 0.9354615481570363} +12/23/2021 23:13:57 - INFO - codeparrot_training - Step 7485: {'lr': 0.00047391158366707445, 'samples': 3832832, 'steps': 7485, 'batch_loss/train': 0.8573918910697103} +12/23/2021 23:14:07 - INFO - codeparrot_training - Step 7486: {'lr': 0.0004739045723007878, 'samples': 3833344, 'steps': 7486, 'batch_loss/train': 0.8435942968353629} +12/23/2021 23:14:18 - INFO - codeparrot_training - Step 7487: {'lr': 0.0004738975600443411, 'samples': 3833856, 'steps': 7487, 'batch_loss/train': 0.8321838597767055} +12/23/2021 23:14:28 - INFO - codeparrot_training - Step 7488: {'lr': 0.00047389054689776235, 'samples': 3834368, 'steps': 7488, 'batch_loss/train': 0.9326350353658199} +12/23/2021 23:14:42 - INFO - codeparrot_training - Step 7489: {'lr': 0.00047388353286107924, 'samples': 3834880, 'steps': 7489, 'batch_loss/train': 0.8962859711609781} +12/23/2021 23:14:53 - INFO - codeparrot_training - Step 7490: {'lr': 0.00047387651793431976, 'samples': 3835392, 'steps': 7490, 'batch_loss/train': 0.7623732752399519} +12/23/2021 23:15:03 - INFO - codeparrot_training - Step 7491: {'lr': 0.00047386950211751166, 'samples': 3835904, 'steps': 7491, 'batch_loss/train': 0.8511980762705207} +12/23/2021 23:15:15 - INFO - codeparrot_training - Step 7492: {'lr': 0.0004738624854106831, 'samples': 3836416, 'steps': 7492, 'batch_loss/train': 1.2422330037225038} +12/23/2021 23:15:26 - INFO - codeparrot_training - Step 7493: {'lr': 0.0004738554678138618, 'samples': 3836928, 'steps': 7493, 'batch_loss/train': 0.8317286828532815} +12/23/2021 23:15:37 - INFO - codeparrot_training - Step 7494: {'lr': 0.0004738484493270756, 'samples': 3837440, 'steps': 7494, 'batch_loss/train': 0.8770252224057913} +12/23/2021 23:15:47 - INFO - codeparrot_training - Step 7495: {'lr': 0.0004738414299503526, 'samples': 3837952, 'steps': 7495, 'batch_loss/train': 0.8537774831056595} +12/23/2021 23:15:59 - INFO - codeparrot_training - Step 7496: {'lr': 0.0004738344096837205, 'samples': 3838464, 'steps': 7496, 'batch_loss/train': 0.859804212115705} +12/23/2021 23:16:10 - INFO - codeparrot_training - Step 7497: {'lr': 0.0004738273885272074, 'samples': 3838976, 'steps': 7497, 'batch_loss/train': 0.7818895909003913} +12/23/2021 23:16:20 - INFO - codeparrot_training - Step 7498: {'lr': 0.000473820366480841, 'samples': 3839488, 'steps': 7498, 'batch_loss/train': 0.863894612994045} +12/23/2021 23:16:33 - INFO - codeparrot_training - Step 7499: {'lr': 0.00047381334354464945, 'samples': 3840000, 'steps': 7499, 'batch_loss/train': 0.9735000836662948} +12/23/2021 23:16:33 - INFO - codeparrot_training - Evaluating and saving model checkpoint +12/23/2021 23:20:08 - INFO - codeparrot_training - Step 7500: {'loss/eval': 0.8546667695045471, 'perplexity': 2.350590944290161}