diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,128603 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 367350, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 5.444399074452157e-05, + "grad_norm": 117.16131591796875, + "learning_rate": 7.6e-07, + "loss": 8.636324310302735, + "step": 20 + }, + { + "epoch": 0.00010888798148904315, + "grad_norm": 54.38514709472656, + "learning_rate": 1.56e-06, + "loss": 7.934609985351562, + "step": 40 + }, + { + "epoch": 0.00016333197223356472, + "grad_norm": 23.872764587402344, + "learning_rate": 2.3600000000000003e-06, + "loss": 7.177521514892578, + "step": 60 + }, + { + "epoch": 0.0002177759629780863, + "grad_norm": 23.526960372924805, + "learning_rate": 3.1600000000000002e-06, + "loss": 6.544634246826172, + "step": 80 + }, + { + "epoch": 0.00027221995372260786, + "grad_norm": 23.14076805114746, + "learning_rate": 3.96e-06, + "loss": 5.889382171630859, + "step": 100 + }, + { + "epoch": 0.00032666394446712944, + "grad_norm": 19.954421997070312, + "learning_rate": 4.76e-06, + "loss": 5.29854736328125, + "step": 120 + }, + { + "epoch": 0.000381107935211651, + "grad_norm": 15.984484672546387, + "learning_rate": 5.560000000000001e-06, + "loss": 4.783774185180664, + "step": 140 + }, + { + "epoch": 0.0004355519259561726, + "grad_norm": 15.475462913513184, + "learning_rate": 6.360000000000001e-06, + "loss": 4.423848342895508, + "step": 160 + }, + { + "epoch": 0.0004899959167006942, + "grad_norm": 29.543569564819336, + "learning_rate": 7.16e-06, + "loss": 4.086998748779297, + "step": 180 + }, + { + "epoch": 0.0005444399074452157, + "grad_norm": 10.71650505065918, + "learning_rate": 7.960000000000002e-06, + "loss": 3.941355895996094, + "step": 200 + }, + { + "epoch": 0.0005988838981897374, + "grad_norm": 9.44576358795166, + "learning_rate": 8.76e-06, + "loss": 3.784453201293945, + "step": 220 + }, + { + "epoch": 0.0006533278889342589, + "grad_norm": 7.554831504821777, + "learning_rate": 9.56e-06, + "loss": 3.6747695922851564, + "step": 240 + }, + { + "epoch": 0.0007077718796787805, + "grad_norm": 7.4196906089782715, + "learning_rate": 1.036e-05, + "loss": 3.5718055725097657, + "step": 260 + }, + { + "epoch": 0.000762215870423302, + "grad_norm": 12.932816505432129, + "learning_rate": 1.1160000000000002e-05, + "loss": 3.5025100708007812, + "step": 280 + }, + { + "epoch": 0.0008166598611678236, + "grad_norm": 6.5054402351379395, + "learning_rate": 1.196e-05, + "loss": 3.4749221801757812, + "step": 300 + }, + { + "epoch": 0.0008711038519123452, + "grad_norm": 7.355905055999756, + "learning_rate": 1.2760000000000001e-05, + "loss": 3.419506072998047, + "step": 320 + }, + { + "epoch": 0.0009255478426568668, + "grad_norm": 7.590634346008301, + "learning_rate": 1.3560000000000002e-05, + "loss": 3.3350032806396483, + "step": 340 + }, + { + "epoch": 0.0009799918334013884, + "grad_norm": 6.259383201599121, + "learning_rate": 1.4360000000000001e-05, + "loss": 3.3104286193847656, + "step": 360 + }, + { + "epoch": 0.00103443582414591, + "grad_norm": 6.065864562988281, + "learning_rate": 1.516e-05, + "loss": 3.2756309509277344, + "step": 380 + }, + { + "epoch": 0.0010888798148904315, + "grad_norm": 5.130244731903076, + "learning_rate": 1.5960000000000003e-05, + "loss": 3.2485591888427736, + "step": 400 + }, + { + "epoch": 0.001143323805634953, + "grad_norm": 5.36083459854126, + "learning_rate": 1.6760000000000002e-05, + "loss": 3.2505489349365235, + "step": 420 + }, + { + "epoch": 0.0011977677963794747, + "grad_norm": 5.226541519165039, + "learning_rate": 1.756e-05, + "loss": 3.194860649108887, + "step": 440 + }, + { + "epoch": 0.0012522117871239962, + "grad_norm": 4.7790632247924805, + "learning_rate": 1.8360000000000004e-05, + "loss": 3.157110023498535, + "step": 460 + }, + { + "epoch": 0.0013066557778685177, + "grad_norm": 4.933673858642578, + "learning_rate": 1.916e-05, + "loss": 3.1444026947021486, + "step": 480 + }, + { + "epoch": 0.0013610997686130393, + "grad_norm": 8.902215957641602, + "learning_rate": 1.9960000000000002e-05, + "loss": 3.1174205780029296, + "step": 500 + }, + { + "epoch": 0.001415543759357561, + "grad_norm": 4.462031364440918, + "learning_rate": 1.9998964154286497e-05, + "loss": 3.076970672607422, + "step": 520 + }, + { + "epoch": 0.0014699877501020825, + "grad_norm": 5.526883602142334, + "learning_rate": 1.999787379037754e-05, + "loss": 3.0671226501464846, + "step": 540 + }, + { + "epoch": 0.001524431740846604, + "grad_norm": 4.8212714195251465, + "learning_rate": 1.9996783426468585e-05, + "loss": 3.0379480361938476, + "step": 560 + }, + { + "epoch": 0.0015788757315911256, + "grad_norm": 4.206490516662598, + "learning_rate": 1.999569306255963e-05, + "loss": 3.0166017532348635, + "step": 580 + }, + { + "epoch": 0.0016333197223356473, + "grad_norm": 6.3535380363464355, + "learning_rate": 1.9994602698650677e-05, + "loss": 3.020033073425293, + "step": 600 + }, + { + "epoch": 0.0016877637130801688, + "grad_norm": 4.848832130432129, + "learning_rate": 1.9993512334741722e-05, + "loss": 3.006058692932129, + "step": 620 + }, + { + "epoch": 0.0017422077038246903, + "grad_norm": 4.5470170974731445, + "learning_rate": 1.9992421970832768e-05, + "loss": 3.014019775390625, + "step": 640 + }, + { + "epoch": 0.0017966516945692118, + "grad_norm": 3.906399965286255, + "learning_rate": 1.999133160692381e-05, + "loss": 2.9718814849853517, + "step": 660 + }, + { + "epoch": 0.0018510956853137336, + "grad_norm": 3.718065023422241, + "learning_rate": 1.999024124301486e-05, + "loss": 2.955400276184082, + "step": 680 + }, + { + "epoch": 0.001905539676058255, + "grad_norm": 4.020778179168701, + "learning_rate": 1.9989150879105902e-05, + "loss": 2.956995391845703, + "step": 700 + }, + { + "epoch": 0.001959983666802777, + "grad_norm": 3.9321820735931396, + "learning_rate": 1.9988060515196948e-05, + "loss": 2.9358091354370117, + "step": 720 + }, + { + "epoch": 0.0020144276575472984, + "grad_norm": 4.821063041687012, + "learning_rate": 1.9986970151287994e-05, + "loss": 2.9159549713134765, + "step": 740 + }, + { + "epoch": 0.00206887164829182, + "grad_norm": 3.863384485244751, + "learning_rate": 1.998587978737904e-05, + "loss": 2.962903594970703, + "step": 760 + }, + { + "epoch": 0.0021233156390363414, + "grad_norm": 3.6936542987823486, + "learning_rate": 1.9984789423470085e-05, + "loss": 2.915324592590332, + "step": 780 + }, + { + "epoch": 0.002177759629780863, + "grad_norm": 5.6879072189331055, + "learning_rate": 1.998369905956113e-05, + "loss": 2.9138153076171873, + "step": 800 + }, + { + "epoch": 0.0022322036205253844, + "grad_norm": 3.51484751701355, + "learning_rate": 1.9982608695652174e-05, + "loss": 2.9062753677368165, + "step": 820 + }, + { + "epoch": 0.002286647611269906, + "grad_norm": 3.722736120223999, + "learning_rate": 1.9981518331743223e-05, + "loss": 2.8935712814331054, + "step": 840 + }, + { + "epoch": 0.0023410916020144275, + "grad_norm": 4.327465057373047, + "learning_rate": 1.9980427967834265e-05, + "loss": 2.899227523803711, + "step": 860 + }, + { + "epoch": 0.0023955355927589494, + "grad_norm": 3.812939405441284, + "learning_rate": 1.997933760392531e-05, + "loss": 2.9135507583618163, + "step": 880 + }, + { + "epoch": 0.002449979583503471, + "grad_norm": 3.5365281105041504, + "learning_rate": 1.9978247240016357e-05, + "loss": 2.900547981262207, + "step": 900 + }, + { + "epoch": 0.0025044235742479925, + "grad_norm": 3.4955570697784424, + "learning_rate": 1.9977156876107403e-05, + "loss": 2.8885566711425783, + "step": 920 + }, + { + "epoch": 0.002558867564992514, + "grad_norm": 3.5336105823516846, + "learning_rate": 1.997606651219845e-05, + "loss": 2.8555477142333983, + "step": 940 + }, + { + "epoch": 0.0026133115557370355, + "grad_norm": 3.5009891986846924, + "learning_rate": 1.9974976148289494e-05, + "loss": 2.8403093338012697, + "step": 960 + }, + { + "epoch": 0.002667755546481557, + "grad_norm": 5.099206924438477, + "learning_rate": 1.9973885784380537e-05, + "loss": 2.871915817260742, + "step": 980 + }, + { + "epoch": 0.0027221995372260785, + "grad_norm": 3.640634775161743, + "learning_rate": 1.9972795420471586e-05, + "loss": 2.8472322463989257, + "step": 1000 + }, + { + "epoch": 0.0027766435279706, + "grad_norm": 3.4642083644866943, + "learning_rate": 1.997170505656263e-05, + "loss": 2.8384647369384766, + "step": 1020 + }, + { + "epoch": 0.002831087518715122, + "grad_norm": 3.4871461391448975, + "learning_rate": 1.9970614692653674e-05, + "loss": 2.8672420501708986, + "step": 1040 + }, + { + "epoch": 0.0028855315094596435, + "grad_norm": 3.0806353092193604, + "learning_rate": 1.996952432874472e-05, + "loss": 2.8532951354980467, + "step": 1060 + }, + { + "epoch": 0.002939975500204165, + "grad_norm": 3.544351816177368, + "learning_rate": 1.9968433964835766e-05, + "loss": 2.863789939880371, + "step": 1080 + }, + { + "epoch": 0.0029944194909486866, + "grad_norm": 3.3323280811309814, + "learning_rate": 1.996734360092681e-05, + "loss": 2.835915184020996, + "step": 1100 + }, + { + "epoch": 0.003048863481693208, + "grad_norm": 3.1407883167266846, + "learning_rate": 1.9966253237017857e-05, + "loss": 2.8113595962524416, + "step": 1120 + }, + { + "epoch": 0.0031033074724377296, + "grad_norm": 3.416429281234741, + "learning_rate": 1.99651628731089e-05, + "loss": 2.8369508743286134, + "step": 1140 + }, + { + "epoch": 0.003157751463182251, + "grad_norm": 3.5840001106262207, + "learning_rate": 1.996407250919995e-05, + "loss": 2.8188098907470702, + "step": 1160 + }, + { + "epoch": 0.0032121954539267726, + "grad_norm": 3.420555830001831, + "learning_rate": 1.996298214529099e-05, + "loss": 2.8524944305419924, + "step": 1180 + }, + { + "epoch": 0.0032666394446712946, + "grad_norm": 3.6459848880767822, + "learning_rate": 1.996189178138204e-05, + "loss": 2.8092636108398437, + "step": 1200 + }, + { + "epoch": 0.003321083435415816, + "grad_norm": 3.4025509357452393, + "learning_rate": 1.9960801417473083e-05, + "loss": 2.7862497329711915, + "step": 1220 + }, + { + "epoch": 0.0033755274261603376, + "grad_norm": 3.4520034790039062, + "learning_rate": 1.995971105356413e-05, + "loss": 2.8034412384033205, + "step": 1240 + }, + { + "epoch": 0.003429971416904859, + "grad_norm": 3.321507453918457, + "learning_rate": 1.9958620689655175e-05, + "loss": 2.8086437225341796, + "step": 1260 + }, + { + "epoch": 0.0034844154076493807, + "grad_norm": 3.3660454750061035, + "learning_rate": 1.995753032574622e-05, + "loss": 2.8130802154541015, + "step": 1280 + }, + { + "epoch": 0.003538859398393902, + "grad_norm": 3.5532755851745605, + "learning_rate": 1.9956439961837263e-05, + "loss": 2.801799774169922, + "step": 1300 + }, + { + "epoch": 0.0035933033891384237, + "grad_norm": 3.828744649887085, + "learning_rate": 1.9955349597928312e-05, + "loss": 2.834650993347168, + "step": 1320 + }, + { + "epoch": 0.003647747379882945, + "grad_norm": 3.2432291507720947, + "learning_rate": 1.9954259234019355e-05, + "loss": 2.797213554382324, + "step": 1340 + }, + { + "epoch": 0.003702191370627467, + "grad_norm": 3.0557608604431152, + "learning_rate": 1.9953168870110404e-05, + "loss": 2.820281410217285, + "step": 1360 + }, + { + "epoch": 0.0037566353613719887, + "grad_norm": 3.2218637466430664, + "learning_rate": 1.9952078506201446e-05, + "loss": 2.807219314575195, + "step": 1380 + }, + { + "epoch": 0.00381107935211651, + "grad_norm": 3.2500696182250977, + "learning_rate": 1.9950988142292492e-05, + "loss": 2.783466339111328, + "step": 1400 + }, + { + "epoch": 0.0038655233428610317, + "grad_norm": 3.6877405643463135, + "learning_rate": 1.9949897778383538e-05, + "loss": 2.7857160568237305, + "step": 1420 + }, + { + "epoch": 0.003919967333605554, + "grad_norm": 2.9297385215759277, + "learning_rate": 1.9948807414474584e-05, + "loss": 2.774748611450195, + "step": 1440 + }, + { + "epoch": 0.003974411324350075, + "grad_norm": 3.2616729736328125, + "learning_rate": 1.9947717050565626e-05, + "loss": 2.7906484603881836, + "step": 1460 + }, + { + "epoch": 0.004028855315094597, + "grad_norm": 3.2573771476745605, + "learning_rate": 1.9946626686656675e-05, + "loss": 2.831247329711914, + "step": 1480 + }, + { + "epoch": 0.004083299305839118, + "grad_norm": 3.2660422325134277, + "learning_rate": 1.9945536322747718e-05, + "loss": 2.809173011779785, + "step": 1500 + }, + { + "epoch": 0.00413774329658364, + "grad_norm": 3.8030028343200684, + "learning_rate": 1.9944445958838763e-05, + "loss": 2.7827781677246093, + "step": 1520 + }, + { + "epoch": 0.004192187287328161, + "grad_norm": 2.8976891040802, + "learning_rate": 1.994335559492981e-05, + "loss": 2.758053016662598, + "step": 1540 + }, + { + "epoch": 0.004246631278072683, + "grad_norm": 2.8630518913269043, + "learning_rate": 1.9942265231020855e-05, + "loss": 2.801017189025879, + "step": 1560 + }, + { + "epoch": 0.004301075268817204, + "grad_norm": 4.468863487243652, + "learning_rate": 1.99411748671119e-05, + "loss": 2.777520179748535, + "step": 1580 + }, + { + "epoch": 0.004355519259561726, + "grad_norm": 2.856722116470337, + "learning_rate": 1.9940084503202947e-05, + "loss": 2.7686376571655273, + "step": 1600 + }, + { + "epoch": 0.004409963250306247, + "grad_norm": 3.1067957878112793, + "learning_rate": 1.993899413929399e-05, + "loss": 2.7694149017333984, + "step": 1620 + }, + { + "epoch": 0.004464407241050769, + "grad_norm": 12.470674514770508, + "learning_rate": 1.9937903775385038e-05, + "loss": 2.752527046203613, + "step": 1640 + }, + { + "epoch": 0.00451885123179529, + "grad_norm": 3.291011095046997, + "learning_rate": 1.993681341147608e-05, + "loss": 2.774333381652832, + "step": 1660 + }, + { + "epoch": 0.004573295222539812, + "grad_norm": 3.1459155082702637, + "learning_rate": 1.9935723047567127e-05, + "loss": 2.7757709503173826, + "step": 1680 + }, + { + "epoch": 0.004627739213284333, + "grad_norm": 2.9345884323120117, + "learning_rate": 1.9934632683658172e-05, + "loss": 2.7472454071044923, + "step": 1700 + }, + { + "epoch": 0.004682183204028855, + "grad_norm": 2.820209264755249, + "learning_rate": 1.9933542319749218e-05, + "loss": 2.807948112487793, + "step": 1720 + }, + { + "epoch": 0.004736627194773377, + "grad_norm": 4.761029243469238, + "learning_rate": 1.9932451955840264e-05, + "loss": 2.7764961242675783, + "step": 1740 + }, + { + "epoch": 0.004791071185517899, + "grad_norm": 3.0279619693756104, + "learning_rate": 1.993136159193131e-05, + "loss": 2.7987852096557617, + "step": 1760 + }, + { + "epoch": 0.00484551517626242, + "grad_norm": 2.9586145877838135, + "learning_rate": 1.9930271228022352e-05, + "loss": 2.755886268615723, + "step": 1780 + }, + { + "epoch": 0.004899959167006942, + "grad_norm": 2.820009231567383, + "learning_rate": 1.99291808641134e-05, + "loss": 2.747854232788086, + "step": 1800 + }, + { + "epoch": 0.004954403157751463, + "grad_norm": 3.0790109634399414, + "learning_rate": 1.9928090500204444e-05, + "loss": 2.7794673919677733, + "step": 1820 + }, + { + "epoch": 0.005008847148495985, + "grad_norm": 2.945971965789795, + "learning_rate": 1.992700013629549e-05, + "loss": 2.7311643600463866, + "step": 1840 + }, + { + "epoch": 0.005063291139240506, + "grad_norm": 3.013664960861206, + "learning_rate": 1.9925909772386535e-05, + "loss": 2.734986114501953, + "step": 1860 + }, + { + "epoch": 0.005117735129985028, + "grad_norm": 2.8899624347686768, + "learning_rate": 1.992481940847758e-05, + "loss": 2.7573686599731446, + "step": 1880 + }, + { + "epoch": 0.0051721791207295495, + "grad_norm": 2.7547717094421387, + "learning_rate": 1.9923729044568627e-05, + "loss": 2.762679672241211, + "step": 1900 + }, + { + "epoch": 0.005226623111474071, + "grad_norm": 3.4778029918670654, + "learning_rate": 1.9922638680659673e-05, + "loss": 2.736695671081543, + "step": 1920 + }, + { + "epoch": 0.0052810671022185925, + "grad_norm": 2.641951084136963, + "learning_rate": 1.9921548316750715e-05, + "loss": 2.7694528579711912, + "step": 1940 + }, + { + "epoch": 0.005335511092963114, + "grad_norm": 2.7980332374572754, + "learning_rate": 1.9920457952841764e-05, + "loss": 2.753557586669922, + "step": 1960 + }, + { + "epoch": 0.0053899550837076355, + "grad_norm": 2.8483400344848633, + "learning_rate": 1.9919367588932807e-05, + "loss": 2.7157588958740235, + "step": 1980 + }, + { + "epoch": 0.005444399074452157, + "grad_norm": 2.950073719024658, + "learning_rate": 1.9918277225023853e-05, + "loss": 2.7469385147094725, + "step": 2000 + }, + { + "epoch": 0.005498843065196679, + "grad_norm": 2.923618793487549, + "learning_rate": 1.99171868611149e-05, + "loss": 2.753184509277344, + "step": 2020 + }, + { + "epoch": 0.0055532870559412, + "grad_norm": 3.4532418251037598, + "learning_rate": 1.9916096497205944e-05, + "loss": 2.748311233520508, + "step": 2040 + }, + { + "epoch": 0.0056077310466857225, + "grad_norm": 2.733656167984009, + "learning_rate": 1.991500613329699e-05, + "loss": 2.73789119720459, + "step": 2060 + }, + { + "epoch": 0.005662175037430244, + "grad_norm": 2.850316286087036, + "learning_rate": 1.9913915769388036e-05, + "loss": 2.766009712219238, + "step": 2080 + }, + { + "epoch": 0.0057166190281747655, + "grad_norm": 2.922224998474121, + "learning_rate": 1.991282540547908e-05, + "loss": 2.7283721923828126, + "step": 2100 + }, + { + "epoch": 0.005771063018919287, + "grad_norm": 2.684157133102417, + "learning_rate": 1.9911735041570128e-05, + "loss": 2.7438804626464846, + "step": 2120 + }, + { + "epoch": 0.0058255070096638085, + "grad_norm": 2.715357542037964, + "learning_rate": 1.991064467766117e-05, + "loss": 2.747371864318848, + "step": 2140 + }, + { + "epoch": 0.00587995100040833, + "grad_norm": 2.6098525524139404, + "learning_rate": 1.9909554313752216e-05, + "loss": 2.727573585510254, + "step": 2160 + }, + { + "epoch": 0.005934394991152852, + "grad_norm": 2.8942079544067383, + "learning_rate": 1.990846394984326e-05, + "loss": 2.7574237823486327, + "step": 2180 + }, + { + "epoch": 0.005988838981897373, + "grad_norm": 2.73002028465271, + "learning_rate": 1.9907373585934307e-05, + "loss": 2.766430473327637, + "step": 2200 + }, + { + "epoch": 0.006043282972641895, + "grad_norm": 2.94378399848938, + "learning_rate": 1.9906283222025353e-05, + "loss": 2.707938385009766, + "step": 2220 + }, + { + "epoch": 0.006097726963386416, + "grad_norm": 2.715439796447754, + "learning_rate": 1.99051928581164e-05, + "loss": 2.736627388000488, + "step": 2240 + }, + { + "epoch": 0.006152170954130938, + "grad_norm": 2.7247486114501953, + "learning_rate": 1.990410249420744e-05, + "loss": 2.7497791290283202, + "step": 2260 + }, + { + "epoch": 0.006206614944875459, + "grad_norm": 2.6884191036224365, + "learning_rate": 1.990301213029849e-05, + "loss": 2.7588474273681642, + "step": 2280 + }, + { + "epoch": 0.006261058935619981, + "grad_norm": 2.8431036472320557, + "learning_rate": 1.9901921766389533e-05, + "loss": 2.7526409149169924, + "step": 2300 + }, + { + "epoch": 0.006315502926364502, + "grad_norm": 2.719444990158081, + "learning_rate": 1.990083140248058e-05, + "loss": 2.7284172058105467, + "step": 2320 + }, + { + "epoch": 0.006369946917109024, + "grad_norm": 2.8056387901306152, + "learning_rate": 1.9899741038571625e-05, + "loss": 2.724051284790039, + "step": 2340 + }, + { + "epoch": 0.006424390907853545, + "grad_norm": 2.6666743755340576, + "learning_rate": 1.989865067466267e-05, + "loss": 2.7156986236572265, + "step": 2360 + }, + { + "epoch": 0.006478834898598068, + "grad_norm": 2.6356003284454346, + "learning_rate": 1.9897560310753716e-05, + "loss": 2.749363899230957, + "step": 2380 + }, + { + "epoch": 0.006533278889342589, + "grad_norm": 3.164048910140991, + "learning_rate": 1.9896469946844762e-05, + "loss": 2.7273393630981446, + "step": 2400 + }, + { + "epoch": 0.006587722880087111, + "grad_norm": 2.6865322589874268, + "learning_rate": 1.9895379582935805e-05, + "loss": 2.7565874099731444, + "step": 2420 + }, + { + "epoch": 0.006642166870831632, + "grad_norm": 3.1781721115112305, + "learning_rate": 1.9894289219026854e-05, + "loss": 2.7477264404296875, + "step": 2440 + }, + { + "epoch": 0.006696610861576154, + "grad_norm": 2.60760498046875, + "learning_rate": 1.9893198855117896e-05, + "loss": 2.7420671463012694, + "step": 2460 + }, + { + "epoch": 0.006751054852320675, + "grad_norm": 2.8996617794036865, + "learning_rate": 1.9892108491208942e-05, + "loss": 2.74175968170166, + "step": 2480 + }, + { + "epoch": 0.006805498843065197, + "grad_norm": 3.066162347793579, + "learning_rate": 1.9891018127299988e-05, + "loss": 2.696652412414551, + "step": 2500 + }, + { + "epoch": 0.006859942833809718, + "grad_norm": 2.9869494438171387, + "learning_rate": 1.9889927763391034e-05, + "loss": 2.6968225479125976, + "step": 2520 + }, + { + "epoch": 0.00691438682455424, + "grad_norm": 2.6902618408203125, + "learning_rate": 1.988883739948208e-05, + "loss": 2.7206398010253907, + "step": 2540 + }, + { + "epoch": 0.006968830815298761, + "grad_norm": 2.953693151473999, + "learning_rate": 1.9887747035573125e-05, + "loss": 2.7302310943603514, + "step": 2560 + }, + { + "epoch": 0.007023274806043283, + "grad_norm": 2.766918182373047, + "learning_rate": 1.9886656671664168e-05, + "loss": 2.7148483276367186, + "step": 2580 + }, + { + "epoch": 0.007077718796787804, + "grad_norm": 2.668123960494995, + "learning_rate": 1.9885566307755217e-05, + "loss": 2.7188791275024413, + "step": 2600 + }, + { + "epoch": 0.007132162787532326, + "grad_norm": 3.7488512992858887, + "learning_rate": 1.988447594384626e-05, + "loss": 2.7228050231933594, + "step": 2620 + }, + { + "epoch": 0.007186606778276847, + "grad_norm": 2.684359550476074, + "learning_rate": 1.9883385579937305e-05, + "loss": 2.7217967987060545, + "step": 2640 + }, + { + "epoch": 0.007241050769021369, + "grad_norm": 2.7078211307525635, + "learning_rate": 1.988229521602835e-05, + "loss": 2.728895378112793, + "step": 2660 + }, + { + "epoch": 0.00729549475976589, + "grad_norm": 2.6795713901519775, + "learning_rate": 1.9881204852119397e-05, + "loss": 2.7307323455810546, + "step": 2680 + }, + { + "epoch": 0.007349938750510413, + "grad_norm": 4.549812316894531, + "learning_rate": 1.9880114488210442e-05, + "loss": 2.742350769042969, + "step": 2700 + }, + { + "epoch": 0.007404382741254934, + "grad_norm": 4.02097749710083, + "learning_rate": 1.9879024124301488e-05, + "loss": 2.6902820587158205, + "step": 2720 + }, + { + "epoch": 0.007458826731999456, + "grad_norm": 2.7239041328430176, + "learning_rate": 1.987793376039253e-05, + "loss": 2.7338016510009764, + "step": 2740 + }, + { + "epoch": 0.007513270722743977, + "grad_norm": 2.7732667922973633, + "learning_rate": 1.987684339648358e-05, + "loss": 2.7190664291381834, + "step": 2760 + }, + { + "epoch": 0.007567714713488499, + "grad_norm": 3.6507461071014404, + "learning_rate": 1.9875753032574622e-05, + "loss": 2.7067987442016603, + "step": 2780 + }, + { + "epoch": 0.00762215870423302, + "grad_norm": 2.623197078704834, + "learning_rate": 1.9874662668665668e-05, + "loss": 2.7266021728515626, + "step": 2800 + }, + { + "epoch": 0.007676602694977542, + "grad_norm": 4.129024505615234, + "learning_rate": 1.9873572304756714e-05, + "loss": 2.7163108825683593, + "step": 2820 + }, + { + "epoch": 0.007731046685722063, + "grad_norm": 2.5936689376831055, + "learning_rate": 1.987248194084776e-05, + "loss": 2.733190727233887, + "step": 2840 + }, + { + "epoch": 0.007785490676466585, + "grad_norm": 2.5424745082855225, + "learning_rate": 1.9871391576938806e-05, + "loss": 2.700815773010254, + "step": 2860 + }, + { + "epoch": 0.007839934667211107, + "grad_norm": 2.6111440658569336, + "learning_rate": 1.987030121302985e-05, + "loss": 2.730293846130371, + "step": 2880 + }, + { + "epoch": 0.007894378657955628, + "grad_norm": 2.4470760822296143, + "learning_rate": 1.9869210849120897e-05, + "loss": 2.727880668640137, + "step": 2900 + }, + { + "epoch": 0.00794882264870015, + "grad_norm": 2.6942691802978516, + "learning_rate": 1.9868120485211943e-05, + "loss": 2.6984811782836915, + "step": 2920 + }, + { + "epoch": 0.008003266639444671, + "grad_norm": 2.884506940841675, + "learning_rate": 1.9867030121302985e-05, + "loss": 2.7213018417358397, + "step": 2940 + }, + { + "epoch": 0.008057710630189193, + "grad_norm": 2.629615068435669, + "learning_rate": 1.986593975739403e-05, + "loss": 2.7468706130981446, + "step": 2960 + }, + { + "epoch": 0.008112154620933714, + "grad_norm": 2.7258057594299316, + "learning_rate": 1.9864849393485077e-05, + "loss": 2.732334327697754, + "step": 2980 + }, + { + "epoch": 0.008166598611678236, + "grad_norm": 2.651825189590454, + "learning_rate": 1.9863759029576123e-05, + "loss": 2.7368495941162108, + "step": 3000 + }, + { + "epoch": 0.008221042602422757, + "grad_norm": 2.703826665878296, + "learning_rate": 1.986266866566717e-05, + "loss": 2.7070632934570313, + "step": 3020 + }, + { + "epoch": 0.00827548659316728, + "grad_norm": 3.106436014175415, + "learning_rate": 1.9861578301758214e-05, + "loss": 2.706160545349121, + "step": 3040 + }, + { + "epoch": 0.0083299305839118, + "grad_norm": 2.733820915222168, + "learning_rate": 1.986048793784926e-05, + "loss": 2.7076595306396483, + "step": 3060 + }, + { + "epoch": 0.008384374574656323, + "grad_norm": 2.5774788856506348, + "learning_rate": 1.9859397573940306e-05, + "loss": 2.6943836212158203, + "step": 3080 + }, + { + "epoch": 0.008438818565400843, + "grad_norm": 3.041738986968994, + "learning_rate": 1.985830721003135e-05, + "loss": 2.694891166687012, + "step": 3100 + }, + { + "epoch": 0.008493262556145366, + "grad_norm": 2.5218238830566406, + "learning_rate": 1.9857216846122394e-05, + "loss": 2.703509521484375, + "step": 3120 + }, + { + "epoch": 0.008547706546889886, + "grad_norm": 2.668062210083008, + "learning_rate": 1.985612648221344e-05, + "loss": 2.7349502563476564, + "step": 3140 + }, + { + "epoch": 0.008602150537634409, + "grad_norm": 2.714092254638672, + "learning_rate": 1.9855036118304486e-05, + "loss": 2.6863821029663084, + "step": 3160 + }, + { + "epoch": 0.008656594528378931, + "grad_norm": 2.623169422149658, + "learning_rate": 1.9853945754395532e-05, + "loss": 2.6983869552612303, + "step": 3180 + }, + { + "epoch": 0.008711038519123452, + "grad_norm": 3.1775050163269043, + "learning_rate": 1.9852855390486578e-05, + "loss": 2.7267435073852537, + "step": 3200 + }, + { + "epoch": 0.008765482509867974, + "grad_norm": 2.4859671592712402, + "learning_rate": 1.9851765026577623e-05, + "loss": 2.7198673248291017, + "step": 3220 + }, + { + "epoch": 0.008819926500612495, + "grad_norm": 2.630014181137085, + "learning_rate": 1.985067466266867e-05, + "loss": 2.7183300018310548, + "step": 3240 + }, + { + "epoch": 0.008874370491357017, + "grad_norm": 2.606180191040039, + "learning_rate": 1.984958429875971e-05, + "loss": 2.654451370239258, + "step": 3260 + }, + { + "epoch": 0.008928814482101538, + "grad_norm": 2.4469919204711914, + "learning_rate": 1.9848493934850757e-05, + "loss": 2.6710556030273436, + "step": 3280 + }, + { + "epoch": 0.00898325847284606, + "grad_norm": 2.5037529468536377, + "learning_rate": 1.9847403570941803e-05, + "loss": 2.6792730331420898, + "step": 3300 + }, + { + "epoch": 0.00903770246359058, + "grad_norm": 2.512200355529785, + "learning_rate": 1.984631320703285e-05, + "loss": 2.6915969848632812, + "step": 3320 + }, + { + "epoch": 0.009092146454335103, + "grad_norm": 5.855691909790039, + "learning_rate": 1.9845222843123895e-05, + "loss": 2.6497316360473633, + "step": 3340 + }, + { + "epoch": 0.009146590445079624, + "grad_norm": 2.7562472820281982, + "learning_rate": 1.984413247921494e-05, + "loss": 2.7019357681274414, + "step": 3360 + }, + { + "epoch": 0.009201034435824146, + "grad_norm": 2.471576690673828, + "learning_rate": 1.9843042115305986e-05, + "loss": 2.6926822662353516, + "step": 3380 + }, + { + "epoch": 0.009255478426568667, + "grad_norm": 2.7435543537139893, + "learning_rate": 1.9841951751397032e-05, + "loss": 2.706832504272461, + "step": 3400 + }, + { + "epoch": 0.00930992241731319, + "grad_norm": 2.468756914138794, + "learning_rate": 1.9840861387488075e-05, + "loss": 2.683331108093262, + "step": 3420 + }, + { + "epoch": 0.00936436640805771, + "grad_norm": 2.473569631576538, + "learning_rate": 1.983977102357912e-05, + "loss": 2.6934810638427735, + "step": 3440 + }, + { + "epoch": 0.009418810398802232, + "grad_norm": 2.805030107498169, + "learning_rate": 1.9838680659670166e-05, + "loss": 2.680511474609375, + "step": 3460 + }, + { + "epoch": 0.009473254389546755, + "grad_norm": 2.7806742191314697, + "learning_rate": 1.9837590295761212e-05, + "loss": 2.7098464965820312, + "step": 3480 + }, + { + "epoch": 0.009527698380291275, + "grad_norm": 2.552508592605591, + "learning_rate": 1.9836499931852258e-05, + "loss": 2.679054069519043, + "step": 3500 + }, + { + "epoch": 0.009582142371035798, + "grad_norm": 2.4736580848693848, + "learning_rate": 1.9835409567943304e-05, + "loss": 2.6829864501953127, + "step": 3520 + }, + { + "epoch": 0.009636586361780318, + "grad_norm": 2.457491874694824, + "learning_rate": 1.983431920403435e-05, + "loss": 2.693918991088867, + "step": 3540 + }, + { + "epoch": 0.00969103035252484, + "grad_norm": 2.760754108428955, + "learning_rate": 1.9833228840125395e-05, + "loss": 2.702524185180664, + "step": 3560 + }, + { + "epoch": 0.009745474343269361, + "grad_norm": 2.524709463119507, + "learning_rate": 1.9832138476216438e-05, + "loss": 2.6728702545166017, + "step": 3580 + }, + { + "epoch": 0.009799918334013884, + "grad_norm": 2.689680576324463, + "learning_rate": 1.9831048112307484e-05, + "loss": 2.7208225250244142, + "step": 3600 + }, + { + "epoch": 0.009854362324758404, + "grad_norm": 2.8058600425720215, + "learning_rate": 1.982995774839853e-05, + "loss": 2.671524429321289, + "step": 3620 + }, + { + "epoch": 0.009908806315502927, + "grad_norm": 2.8456125259399414, + "learning_rate": 1.9828867384489575e-05, + "loss": 2.6663965225219726, + "step": 3640 + }, + { + "epoch": 0.009963250306247447, + "grad_norm": 2.536391258239746, + "learning_rate": 1.982777702058062e-05, + "loss": 2.675985336303711, + "step": 3660 + }, + { + "epoch": 0.01001769429699197, + "grad_norm": 2.8603169918060303, + "learning_rate": 1.9826686656671667e-05, + "loss": 2.670868682861328, + "step": 3680 + }, + { + "epoch": 0.01007213828773649, + "grad_norm": 2.336740255355835, + "learning_rate": 1.9825596292762713e-05, + "loss": 2.6830163955688477, + "step": 3700 + }, + { + "epoch": 0.010126582278481013, + "grad_norm": 2.561075210571289, + "learning_rate": 1.9824505928853755e-05, + "loss": 2.6841703414916993, + "step": 3720 + }, + { + "epoch": 0.010181026269225533, + "grad_norm": 2.464674711227417, + "learning_rate": 1.98234155649448e-05, + "loss": 2.6745004653930664, + "step": 3740 + }, + { + "epoch": 0.010235470259970056, + "grad_norm": 2.625389337539673, + "learning_rate": 1.9822325201035847e-05, + "loss": 2.688642120361328, + "step": 3760 + }, + { + "epoch": 0.010289914250714577, + "grad_norm": 13.385398864746094, + "learning_rate": 1.9821234837126892e-05, + "loss": 2.6866840362548827, + "step": 3780 + }, + { + "epoch": 0.010344358241459099, + "grad_norm": 2.459697723388672, + "learning_rate": 1.9820144473217938e-05, + "loss": 2.6920570373535155, + "step": 3800 + }, + { + "epoch": 0.010398802232203621, + "grad_norm": 2.6061456203460693, + "learning_rate": 1.9819054109308984e-05, + "loss": 2.649507522583008, + "step": 3820 + }, + { + "epoch": 0.010453246222948142, + "grad_norm": 2.533412218093872, + "learning_rate": 1.981796374540003e-05, + "loss": 2.7197525024414064, + "step": 3840 + }, + { + "epoch": 0.010507690213692664, + "grad_norm": 2.648465394973755, + "learning_rate": 1.9816873381491076e-05, + "loss": 2.678421211242676, + "step": 3860 + }, + { + "epoch": 0.010562134204437185, + "grad_norm": 2.558516263961792, + "learning_rate": 1.9815783017582118e-05, + "loss": 2.694631004333496, + "step": 3880 + }, + { + "epoch": 0.010616578195181707, + "grad_norm": 2.4478564262390137, + "learning_rate": 1.9814692653673164e-05, + "loss": 2.6588417053222657, + "step": 3900 + }, + { + "epoch": 0.010671022185926228, + "grad_norm": 2.4454894065856934, + "learning_rate": 1.981360228976421e-05, + "loss": 2.662460517883301, + "step": 3920 + }, + { + "epoch": 0.01072546617667075, + "grad_norm": 2.8146774768829346, + "learning_rate": 1.9812511925855256e-05, + "loss": 2.636836624145508, + "step": 3940 + }, + { + "epoch": 0.010779910167415271, + "grad_norm": 2.452869176864624, + "learning_rate": 1.98114215619463e-05, + "loss": 2.682052993774414, + "step": 3960 + }, + { + "epoch": 0.010834354158159793, + "grad_norm": 2.4153120517730713, + "learning_rate": 1.9810331198037347e-05, + "loss": 2.663201332092285, + "step": 3980 + }, + { + "epoch": 0.010888798148904314, + "grad_norm": 2.350930690765381, + "learning_rate": 1.9809240834128393e-05, + "loss": 2.685174560546875, + "step": 4000 + }, + { + "epoch": 0.010943242139648836, + "grad_norm": 2.6138248443603516, + "learning_rate": 1.980815047021944e-05, + "loss": 2.6392644882202148, + "step": 4020 + }, + { + "epoch": 0.010997686130393357, + "grad_norm": 2.5007264614105225, + "learning_rate": 1.980706010631048e-05, + "loss": 2.656688117980957, + "step": 4040 + }, + { + "epoch": 0.01105213012113788, + "grad_norm": 2.525615692138672, + "learning_rate": 1.9805969742401527e-05, + "loss": 2.673896598815918, + "step": 4060 + }, + { + "epoch": 0.0111065741118824, + "grad_norm": 2.5058887004852295, + "learning_rate": 1.9804879378492573e-05, + "loss": 2.6656415939331053, + "step": 4080 + }, + { + "epoch": 0.011161018102626923, + "grad_norm": 2.4749674797058105, + "learning_rate": 1.980378901458362e-05, + "loss": 2.6609350204467774, + "step": 4100 + }, + { + "epoch": 0.011215462093371445, + "grad_norm": 2.3915719985961914, + "learning_rate": 1.9802698650674664e-05, + "loss": 2.6453807830810545, + "step": 4120 + }, + { + "epoch": 0.011269906084115966, + "grad_norm": 2.612779140472412, + "learning_rate": 1.980160828676571e-05, + "loss": 2.665203094482422, + "step": 4140 + }, + { + "epoch": 0.011324350074860488, + "grad_norm": 2.4095332622528076, + "learning_rate": 1.9800517922856756e-05, + "loss": 2.702609634399414, + "step": 4160 + }, + { + "epoch": 0.011378794065605009, + "grad_norm": 2.420175790786743, + "learning_rate": 1.9799427558947802e-05, + "loss": 2.670509147644043, + "step": 4180 + }, + { + "epoch": 0.011433238056349531, + "grad_norm": 2.4343981742858887, + "learning_rate": 1.9798337195038844e-05, + "loss": 2.6790792465209963, + "step": 4200 + }, + { + "epoch": 0.011487682047094052, + "grad_norm": 2.5150997638702393, + "learning_rate": 1.979724683112989e-05, + "loss": 2.695829391479492, + "step": 4220 + }, + { + "epoch": 0.011542126037838574, + "grad_norm": 2.389597177505493, + "learning_rate": 1.9796156467220936e-05, + "loss": 2.6672079086303713, + "step": 4240 + }, + { + "epoch": 0.011596570028583095, + "grad_norm": 2.723372220993042, + "learning_rate": 1.979506610331198e-05, + "loss": 2.689576530456543, + "step": 4260 + }, + { + "epoch": 0.011651014019327617, + "grad_norm": 2.522329568862915, + "learning_rate": 1.9793975739403027e-05, + "loss": 2.668549156188965, + "step": 4280 + }, + { + "epoch": 0.011705458010072138, + "grad_norm": 2.5249950885772705, + "learning_rate": 1.9792885375494073e-05, + "loss": 2.707304763793945, + "step": 4300 + }, + { + "epoch": 0.01175990200081666, + "grad_norm": 2.7151007652282715, + "learning_rate": 1.979179501158512e-05, + "loss": 2.668537712097168, + "step": 4320 + }, + { + "epoch": 0.01181434599156118, + "grad_norm": 2.8399693965911865, + "learning_rate": 1.9790704647676165e-05, + "loss": 2.67172794342041, + "step": 4340 + }, + { + "epoch": 0.011868789982305703, + "grad_norm": 2.5547142028808594, + "learning_rate": 1.9789614283767207e-05, + "loss": 2.7082733154296874, + "step": 4360 + }, + { + "epoch": 0.011923233973050224, + "grad_norm": 2.599687337875366, + "learning_rate": 1.9788523919858253e-05, + "loss": 2.705434226989746, + "step": 4380 + }, + { + "epoch": 0.011977677963794746, + "grad_norm": 2.5729384422302246, + "learning_rate": 1.97874335559493e-05, + "loss": 2.679910087585449, + "step": 4400 + }, + { + "epoch": 0.012032121954539267, + "grad_norm": 2.475112199783325, + "learning_rate": 1.9786343192040345e-05, + "loss": 2.672877311706543, + "step": 4420 + }, + { + "epoch": 0.01208656594528379, + "grad_norm": 2.406857967376709, + "learning_rate": 1.978525282813139e-05, + "loss": 2.6597305297851563, + "step": 4440 + }, + { + "epoch": 0.012141009936028312, + "grad_norm": 4.045164108276367, + "learning_rate": 1.9784162464222436e-05, + "loss": 2.6829193115234373, + "step": 4460 + }, + { + "epoch": 0.012195453926772832, + "grad_norm": 2.681089162826538, + "learning_rate": 1.9783072100313482e-05, + "loss": 2.6908456802368166, + "step": 4480 + }, + { + "epoch": 0.012249897917517355, + "grad_norm": 2.5228965282440186, + "learning_rate": 1.9781981736404528e-05, + "loss": 2.6618459701538084, + "step": 4500 + }, + { + "epoch": 0.012304341908261875, + "grad_norm": 2.7162208557128906, + "learning_rate": 1.978089137249557e-05, + "loss": 2.6608577728271485, + "step": 4520 + }, + { + "epoch": 0.012358785899006398, + "grad_norm": 2.500537157058716, + "learning_rate": 1.9779801008586616e-05, + "loss": 2.639795684814453, + "step": 4540 + }, + { + "epoch": 0.012413229889750918, + "grad_norm": 2.479501485824585, + "learning_rate": 1.9778710644677662e-05, + "loss": 2.6594186782836915, + "step": 4560 + }, + { + "epoch": 0.01246767388049544, + "grad_norm": 2.4205260276794434, + "learning_rate": 1.9777620280768708e-05, + "loss": 2.701265335083008, + "step": 4580 + }, + { + "epoch": 0.012522117871239961, + "grad_norm": 2.5365347862243652, + "learning_rate": 1.9776529916859754e-05, + "loss": 2.669887161254883, + "step": 4600 + }, + { + "epoch": 0.012576561861984484, + "grad_norm": 2.2915585041046143, + "learning_rate": 1.97754395529508e-05, + "loss": 2.6659975051879883, + "step": 4620 + }, + { + "epoch": 0.012631005852729004, + "grad_norm": 2.386690139770508, + "learning_rate": 1.9774349189041845e-05, + "loss": 2.6535354614257813, + "step": 4640 + }, + { + "epoch": 0.012685449843473527, + "grad_norm": 2.562718152999878, + "learning_rate": 1.977325882513289e-05, + "loss": 2.641727066040039, + "step": 4660 + }, + { + "epoch": 0.012739893834218047, + "grad_norm": 2.4514882564544678, + "learning_rate": 1.9772168461223934e-05, + "loss": 2.671315383911133, + "step": 4680 + }, + { + "epoch": 0.01279433782496257, + "grad_norm": 2.6145870685577393, + "learning_rate": 1.977107809731498e-05, + "loss": 2.636921691894531, + "step": 4700 + }, + { + "epoch": 0.01284878181570709, + "grad_norm": 2.4395124912261963, + "learning_rate": 1.9769987733406025e-05, + "loss": 2.664874458312988, + "step": 4720 + }, + { + "epoch": 0.012903225806451613, + "grad_norm": 2.414395809173584, + "learning_rate": 1.976889736949707e-05, + "loss": 2.668491744995117, + "step": 4740 + }, + { + "epoch": 0.012957669797196135, + "grad_norm": 2.4952549934387207, + "learning_rate": 1.9767807005588117e-05, + "loss": 2.661863899230957, + "step": 4760 + }, + { + "epoch": 0.013012113787940656, + "grad_norm": 2.4593732357025146, + "learning_rate": 1.9766716641679163e-05, + "loss": 2.6559982299804688, + "step": 4780 + }, + { + "epoch": 0.013066557778685178, + "grad_norm": 2.3368165493011475, + "learning_rate": 1.976562627777021e-05, + "loss": 2.6896236419677733, + "step": 4800 + }, + { + "epoch": 0.013121001769429699, + "grad_norm": 2.4747846126556396, + "learning_rate": 1.9764535913861254e-05, + "loss": 2.6745018005371093, + "step": 4820 + }, + { + "epoch": 0.013175445760174221, + "grad_norm": 2.4989407062530518, + "learning_rate": 1.9763445549952297e-05, + "loss": 2.6713796615600587, + "step": 4840 + }, + { + "epoch": 0.013229889750918742, + "grad_norm": 2.7430295944213867, + "learning_rate": 1.9762355186043342e-05, + "loss": 2.6568876266479493, + "step": 4860 + }, + { + "epoch": 0.013284333741663264, + "grad_norm": 2.402122735977173, + "learning_rate": 1.9761264822134388e-05, + "loss": 2.6317983627319337, + "step": 4880 + }, + { + "epoch": 0.013338777732407785, + "grad_norm": 2.4256703853607178, + "learning_rate": 1.9760174458225434e-05, + "loss": 2.6708137512207033, + "step": 4900 + }, + { + "epoch": 0.013393221723152307, + "grad_norm": 2.4914021492004395, + "learning_rate": 1.975908409431648e-05, + "loss": 2.6753332138061525, + "step": 4920 + }, + { + "epoch": 0.013447665713896828, + "grad_norm": 2.3588850498199463, + "learning_rate": 1.9757993730407526e-05, + "loss": 2.6126220703125, + "step": 4940 + }, + { + "epoch": 0.01350210970464135, + "grad_norm": 2.6094789505004883, + "learning_rate": 1.975690336649857e-05, + "loss": 2.656305503845215, + "step": 4960 + }, + { + "epoch": 0.013556553695385871, + "grad_norm": 2.5412659645080566, + "learning_rate": 1.9755813002589617e-05, + "loss": 2.6837602615356446, + "step": 4980 + }, + { + "epoch": 0.013610997686130393, + "grad_norm": 2.3833460807800293, + "learning_rate": 1.975472263868066e-05, + "loss": 2.6479829788208007, + "step": 5000 + }, + { + "epoch": 0.013665441676874914, + "grad_norm": 2.3657350540161133, + "learning_rate": 1.9753632274771705e-05, + "loss": 2.655087471008301, + "step": 5020 + }, + { + "epoch": 0.013719885667619437, + "grad_norm": 2.4504036903381348, + "learning_rate": 1.975254191086275e-05, + "loss": 2.6527482986450197, + "step": 5040 + }, + { + "epoch": 0.013774329658363959, + "grad_norm": 2.5608763694763184, + "learning_rate": 1.9751451546953797e-05, + "loss": 2.688309097290039, + "step": 5060 + }, + { + "epoch": 0.01382877364910848, + "grad_norm": 2.4236209392547607, + "learning_rate": 1.9750361183044843e-05, + "loss": 2.6377464294433595, + "step": 5080 + }, + { + "epoch": 0.013883217639853002, + "grad_norm": 2.470017194747925, + "learning_rate": 1.974927081913589e-05, + "loss": 2.666164016723633, + "step": 5100 + }, + { + "epoch": 0.013937661630597523, + "grad_norm": 2.6672585010528564, + "learning_rate": 1.9748180455226935e-05, + "loss": 2.6708648681640623, + "step": 5120 + }, + { + "epoch": 0.013992105621342045, + "grad_norm": 4.076218128204346, + "learning_rate": 1.974709009131798e-05, + "loss": 2.6850564956665037, + "step": 5140 + }, + { + "epoch": 0.014046549612086566, + "grad_norm": 2.3578085899353027, + "learning_rate": 1.9745999727409023e-05, + "loss": 2.657058906555176, + "step": 5160 + }, + { + "epoch": 0.014100993602831088, + "grad_norm": 2.2923545837402344, + "learning_rate": 1.974490936350007e-05, + "loss": 2.65228157043457, + "step": 5180 + }, + { + "epoch": 0.014155437593575609, + "grad_norm": 2.480581760406494, + "learning_rate": 1.9743818999591114e-05, + "loss": 2.710798645019531, + "step": 5200 + }, + { + "epoch": 0.014209881584320131, + "grad_norm": 2.6026344299316406, + "learning_rate": 1.974272863568216e-05, + "loss": 2.698649024963379, + "step": 5220 + }, + { + "epoch": 0.014264325575064652, + "grad_norm": 2.635380268096924, + "learning_rate": 1.9741638271773206e-05, + "loss": 2.660744476318359, + "step": 5240 + }, + { + "epoch": 0.014318769565809174, + "grad_norm": 2.2739603519439697, + "learning_rate": 1.9740547907864252e-05, + "loss": 2.644626998901367, + "step": 5260 + }, + { + "epoch": 0.014373213556553695, + "grad_norm": 2.349048376083374, + "learning_rate": 1.9739457543955298e-05, + "loss": 2.6630300521850585, + "step": 5280 + }, + { + "epoch": 0.014427657547298217, + "grad_norm": 2.473942279815674, + "learning_rate": 1.9738367180046343e-05, + "loss": 2.639691162109375, + "step": 5300 + }, + { + "epoch": 0.014482101538042738, + "grad_norm": 2.3351175785064697, + "learning_rate": 1.9737276816137386e-05, + "loss": 2.6470735549926756, + "step": 5320 + }, + { + "epoch": 0.01453654552878726, + "grad_norm": 2.6672816276550293, + "learning_rate": 1.973618645222843e-05, + "loss": 2.6289093017578127, + "step": 5340 + }, + { + "epoch": 0.01459098951953178, + "grad_norm": 2.3488821983337402, + "learning_rate": 1.9735096088319477e-05, + "loss": 2.6542993545532227, + "step": 5360 + }, + { + "epoch": 0.014645433510276303, + "grad_norm": 2.379922866821289, + "learning_rate": 1.9734005724410523e-05, + "loss": 2.6681886672973634, + "step": 5380 + }, + { + "epoch": 0.014699877501020826, + "grad_norm": 2.3839337825775146, + "learning_rate": 1.973291536050157e-05, + "loss": 2.666107749938965, + "step": 5400 + }, + { + "epoch": 0.014754321491765346, + "grad_norm": 2.4457216262817383, + "learning_rate": 1.9731824996592615e-05, + "loss": 2.6472692489624023, + "step": 5420 + }, + { + "epoch": 0.014808765482509869, + "grad_norm": 2.3634235858917236, + "learning_rate": 1.973073463268366e-05, + "loss": 2.6199121475219727, + "step": 5440 + }, + { + "epoch": 0.01486320947325439, + "grad_norm": 2.459923267364502, + "learning_rate": 1.9729644268774707e-05, + "loss": 2.6652587890625, + "step": 5460 + }, + { + "epoch": 0.014917653463998912, + "grad_norm": 2.375122308731079, + "learning_rate": 1.972855390486575e-05, + "loss": 2.6186208724975586, + "step": 5480 + }, + { + "epoch": 0.014972097454743432, + "grad_norm": 2.529449701309204, + "learning_rate": 1.9727463540956795e-05, + "loss": 2.6216428756713865, + "step": 5500 + }, + { + "epoch": 0.015026541445487955, + "grad_norm": 2.5333502292633057, + "learning_rate": 1.972637317704784e-05, + "loss": 2.642997932434082, + "step": 5520 + }, + { + "epoch": 0.015080985436232475, + "grad_norm": 2.426917791366577, + "learning_rate": 1.9725282813138886e-05, + "loss": 2.6554901123046877, + "step": 5540 + }, + { + "epoch": 0.015135429426976998, + "grad_norm": 2.823945999145508, + "learning_rate": 1.9724192449229932e-05, + "loss": 2.6407358169555666, + "step": 5560 + }, + { + "epoch": 0.015189873417721518, + "grad_norm": 2.473532199859619, + "learning_rate": 1.9723102085320978e-05, + "loss": 2.6279083251953126, + "step": 5580 + }, + { + "epoch": 0.01524431740846604, + "grad_norm": 2.3855743408203125, + "learning_rate": 1.9722011721412024e-05, + "loss": 2.6696252822875977, + "step": 5600 + }, + { + "epoch": 0.015298761399210561, + "grad_norm": 2.432544469833374, + "learning_rate": 1.972092135750307e-05, + "loss": 2.6577415466308594, + "step": 5620 + }, + { + "epoch": 0.015353205389955084, + "grad_norm": 2.7873082160949707, + "learning_rate": 1.9719830993594112e-05, + "loss": 2.6334918975830077, + "step": 5640 + }, + { + "epoch": 0.015407649380699604, + "grad_norm": 2.587059259414673, + "learning_rate": 1.9718740629685158e-05, + "loss": 2.6455570220947267, + "step": 5660 + }, + { + "epoch": 0.015462093371444127, + "grad_norm": 2.42588210105896, + "learning_rate": 1.9717650265776204e-05, + "loss": 2.637130546569824, + "step": 5680 + }, + { + "epoch": 0.01551653736218865, + "grad_norm": 2.566836357116699, + "learning_rate": 1.971655990186725e-05, + "loss": 2.6631948471069338, + "step": 5700 + }, + { + "epoch": 0.01557098135293317, + "grad_norm": 2.3044493198394775, + "learning_rate": 1.9715469537958295e-05, + "loss": 2.663666915893555, + "step": 5720 + }, + { + "epoch": 0.015625425343677692, + "grad_norm": 2.434964895248413, + "learning_rate": 1.971437917404934e-05, + "loss": 2.6316015243530275, + "step": 5740 + }, + { + "epoch": 0.015679869334422215, + "grad_norm": 2.8035261631011963, + "learning_rate": 1.9713288810140387e-05, + "loss": 2.6184743881225585, + "step": 5760 + }, + { + "epoch": 0.015734313325166734, + "grad_norm": 2.7628750801086426, + "learning_rate": 1.9712198446231433e-05, + "loss": 2.645313835144043, + "step": 5780 + }, + { + "epoch": 0.015788757315911256, + "grad_norm": 2.391019582748413, + "learning_rate": 1.9711108082322475e-05, + "loss": 2.6488710403442384, + "step": 5800 + }, + { + "epoch": 0.01584320130665578, + "grad_norm": 2.3606908321380615, + "learning_rate": 1.971001771841352e-05, + "loss": 2.6262466430664064, + "step": 5820 + }, + { + "epoch": 0.0158976452974003, + "grad_norm": 2.27355694770813, + "learning_rate": 1.9708927354504567e-05, + "loss": 2.65002384185791, + "step": 5840 + }, + { + "epoch": 0.01595208928814482, + "grad_norm": 2.4118762016296387, + "learning_rate": 1.9707836990595613e-05, + "loss": 2.6143638610839846, + "step": 5860 + }, + { + "epoch": 0.016006533278889342, + "grad_norm": 2.3891425132751465, + "learning_rate": 1.970674662668666e-05, + "loss": 2.6332374572753907, + "step": 5880 + }, + { + "epoch": 0.016060977269633864, + "grad_norm": 2.345886468887329, + "learning_rate": 1.9705656262777704e-05, + "loss": 2.630428695678711, + "step": 5900 + }, + { + "epoch": 0.016115421260378387, + "grad_norm": 2.3462066650390625, + "learning_rate": 1.9704565898868747e-05, + "loss": 2.6632484436035155, + "step": 5920 + }, + { + "epoch": 0.016169865251122906, + "grad_norm": 2.3872761726379395, + "learning_rate": 1.9703475534959796e-05, + "loss": 2.6380992889404298, + "step": 5940 + }, + { + "epoch": 0.016224309241867428, + "grad_norm": 2.3115906715393066, + "learning_rate": 1.9702385171050838e-05, + "loss": 2.6308704376220704, + "step": 5960 + }, + { + "epoch": 0.01627875323261195, + "grad_norm": 2.363626718521118, + "learning_rate": 1.9701294807141884e-05, + "loss": 2.6708616256713866, + "step": 5980 + }, + { + "epoch": 0.016333197223356473, + "grad_norm": 2.4943482875823975, + "learning_rate": 1.970020444323293e-05, + "loss": 2.6479961395263674, + "step": 6000 + }, + { + "epoch": 0.016387641214100995, + "grad_norm": 2.547351360321045, + "learning_rate": 1.9699114079323976e-05, + "loss": 2.6357229232788084, + "step": 6020 + }, + { + "epoch": 0.016442085204845514, + "grad_norm": 2.4869441986083984, + "learning_rate": 1.969802371541502e-05, + "loss": 2.6585092544555664, + "step": 6040 + }, + { + "epoch": 0.016496529195590037, + "grad_norm": 2.543081045150757, + "learning_rate": 1.9696933351506067e-05, + "loss": 2.655788803100586, + "step": 6060 + }, + { + "epoch": 0.01655097318633456, + "grad_norm": 2.6503543853759766, + "learning_rate": 1.969584298759711e-05, + "loss": 2.6172870635986327, + "step": 6080 + }, + { + "epoch": 0.01660541717707908, + "grad_norm": 2.3932607173919678, + "learning_rate": 1.969475262368816e-05, + "loss": 2.6335994720458986, + "step": 6100 + }, + { + "epoch": 0.0166598611678236, + "grad_norm": 2.4672365188598633, + "learning_rate": 1.96936622597792e-05, + "loss": 2.646941375732422, + "step": 6120 + }, + { + "epoch": 0.016714305158568123, + "grad_norm": 2.375364303588867, + "learning_rate": 1.9692571895870247e-05, + "loss": 2.6365169525146483, + "step": 6140 + }, + { + "epoch": 0.016768749149312645, + "grad_norm": 2.299807071685791, + "learning_rate": 1.9691481531961293e-05, + "loss": 2.6452754974365233, + "step": 6160 + }, + { + "epoch": 0.016823193140057167, + "grad_norm": 2.3967385292053223, + "learning_rate": 1.969039116805234e-05, + "loss": 2.6403783798217773, + "step": 6180 + }, + { + "epoch": 0.016877637130801686, + "grad_norm": 2.314852237701416, + "learning_rate": 1.9689300804143385e-05, + "loss": 2.6374162673950194, + "step": 6200 + }, + { + "epoch": 0.01693208112154621, + "grad_norm": 2.338977575302124, + "learning_rate": 1.968821044023443e-05, + "loss": 2.6275949478149414, + "step": 6220 + }, + { + "epoch": 0.01698652511229073, + "grad_norm": 2.416416883468628, + "learning_rate": 1.9687120076325473e-05, + "loss": 2.6241783142089843, + "step": 6240 + }, + { + "epoch": 0.017040969103035254, + "grad_norm": 2.483301877975464, + "learning_rate": 1.9686029712416522e-05, + "loss": 2.635630989074707, + "step": 6260 + }, + { + "epoch": 0.017095413093779772, + "grad_norm": 2.815852642059326, + "learning_rate": 1.9684939348507564e-05, + "loss": 2.6139667510986326, + "step": 6280 + }, + { + "epoch": 0.017149857084524295, + "grad_norm": 4.550144672393799, + "learning_rate": 1.9683848984598614e-05, + "loss": 2.6419450759887697, + "step": 6300 + }, + { + "epoch": 0.017204301075268817, + "grad_norm": 2.569279909133911, + "learning_rate": 1.9682758620689656e-05, + "loss": 2.6522842407226563, + "step": 6320 + }, + { + "epoch": 0.01725874506601334, + "grad_norm": 2.4618735313415527, + "learning_rate": 1.9681668256780702e-05, + "loss": 2.6242042541503907, + "step": 6340 + }, + { + "epoch": 0.017313189056757862, + "grad_norm": 2.6002273559570312, + "learning_rate": 1.9680577892871748e-05, + "loss": 2.631578063964844, + "step": 6360 + }, + { + "epoch": 0.01736763304750238, + "grad_norm": 2.282216787338257, + "learning_rate": 1.9679487528962793e-05, + "loss": 2.615373229980469, + "step": 6380 + }, + { + "epoch": 0.017422077038246903, + "grad_norm": 2.43141770362854, + "learning_rate": 1.9678397165053836e-05, + "loss": 2.6303430557250977, + "step": 6400 + }, + { + "epoch": 0.017476521028991426, + "grad_norm": 2.34328556060791, + "learning_rate": 1.9677306801144885e-05, + "loss": 2.6521223068237303, + "step": 6420 + }, + { + "epoch": 0.017530965019735948, + "grad_norm": 2.231238842010498, + "learning_rate": 1.9676216437235927e-05, + "loss": 2.616868591308594, + "step": 6440 + }, + { + "epoch": 0.017585409010480467, + "grad_norm": 2.9436309337615967, + "learning_rate": 1.9675126073326977e-05, + "loss": 2.635344696044922, + "step": 6460 + }, + { + "epoch": 0.01763985300122499, + "grad_norm": 2.3164420127868652, + "learning_rate": 1.967403570941802e-05, + "loss": 2.6287681579589846, + "step": 6480 + }, + { + "epoch": 0.01769429699196951, + "grad_norm": 4.482106685638428, + "learning_rate": 1.9672945345509065e-05, + "loss": 2.609944534301758, + "step": 6500 + }, + { + "epoch": 0.017748740982714034, + "grad_norm": 2.418030261993408, + "learning_rate": 1.967185498160011e-05, + "loss": 2.633044624328613, + "step": 6520 + }, + { + "epoch": 0.017803184973458553, + "grad_norm": 2.291473627090454, + "learning_rate": 1.9670764617691157e-05, + "loss": 2.628916358947754, + "step": 6540 + }, + { + "epoch": 0.017857628964203075, + "grad_norm": 2.4323313236236572, + "learning_rate": 1.96696742537822e-05, + "loss": 2.6335525512695312, + "step": 6560 + }, + { + "epoch": 0.017912072954947598, + "grad_norm": 2.341857671737671, + "learning_rate": 1.9668583889873248e-05, + "loss": 2.6322233200073244, + "step": 6580 + }, + { + "epoch": 0.01796651694569212, + "grad_norm": 2.4729292392730713, + "learning_rate": 1.966749352596429e-05, + "loss": 2.623850631713867, + "step": 6600 + }, + { + "epoch": 0.01802096093643664, + "grad_norm": 2.488398551940918, + "learning_rate": 1.966640316205534e-05, + "loss": 2.6561439514160154, + "step": 6620 + }, + { + "epoch": 0.01807540492718116, + "grad_norm": 2.516674518585205, + "learning_rate": 1.9665312798146382e-05, + "loss": 2.6411602020263674, + "step": 6640 + }, + { + "epoch": 0.018129848917925684, + "grad_norm": 2.344259738922119, + "learning_rate": 1.9664222434237428e-05, + "loss": 2.6050344467163087, + "step": 6660 + }, + { + "epoch": 0.018184292908670206, + "grad_norm": 2.3555054664611816, + "learning_rate": 1.9663132070328474e-05, + "loss": 2.6517066955566406, + "step": 6680 + }, + { + "epoch": 0.01823873689941473, + "grad_norm": 2.6653072834014893, + "learning_rate": 1.966204170641952e-05, + "loss": 2.632643127441406, + "step": 6700 + }, + { + "epoch": 0.018293180890159248, + "grad_norm": 2.339796781539917, + "learning_rate": 1.9660951342510562e-05, + "loss": 2.63393497467041, + "step": 6720 + }, + { + "epoch": 0.01834762488090377, + "grad_norm": 2.695674180984497, + "learning_rate": 1.965986097860161e-05, + "loss": 2.639607238769531, + "step": 6740 + }, + { + "epoch": 0.018402068871648292, + "grad_norm": 2.6209170818328857, + "learning_rate": 1.9658770614692654e-05, + "loss": 2.6452619552612306, + "step": 6760 + }, + { + "epoch": 0.018456512862392815, + "grad_norm": 2.3691580295562744, + "learning_rate": 1.9657680250783703e-05, + "loss": 2.6699270248413085, + "step": 6780 + }, + { + "epoch": 0.018510956853137334, + "grad_norm": 2.281924247741699, + "learning_rate": 1.9656589886874745e-05, + "loss": 2.6153926849365234, + "step": 6800 + }, + { + "epoch": 0.018565400843881856, + "grad_norm": 2.2349154949188232, + "learning_rate": 1.965549952296579e-05, + "loss": 2.611507797241211, + "step": 6820 + }, + { + "epoch": 0.01861984483462638, + "grad_norm": 2.3102681636810303, + "learning_rate": 1.9654409159056837e-05, + "loss": 2.646292304992676, + "step": 6840 + }, + { + "epoch": 0.0186742888253709, + "grad_norm": 2.8698477745056152, + "learning_rate": 1.9653318795147883e-05, + "loss": 2.6299615859985352, + "step": 6860 + }, + { + "epoch": 0.01872873281611542, + "grad_norm": 2.2422637939453125, + "learning_rate": 1.9652228431238925e-05, + "loss": 2.63732795715332, + "step": 6880 + }, + { + "epoch": 0.018783176806859942, + "grad_norm": 2.1849448680877686, + "learning_rate": 1.9651138067329974e-05, + "loss": 2.6326242446899415, + "step": 6900 + }, + { + "epoch": 0.018837620797604464, + "grad_norm": 2.797301769256592, + "learning_rate": 1.9650047703421017e-05, + "loss": 2.654819869995117, + "step": 6920 + }, + { + "epoch": 0.018892064788348987, + "grad_norm": 2.2503116130828857, + "learning_rate": 1.9648957339512066e-05, + "loss": 2.6379823684692383, + "step": 6940 + }, + { + "epoch": 0.01894650877909351, + "grad_norm": 2.3441216945648193, + "learning_rate": 1.964786697560311e-05, + "loss": 2.6424726486206054, + "step": 6960 + }, + { + "epoch": 0.019000952769838028, + "grad_norm": 2.3958218097686768, + "learning_rate": 1.9646776611694154e-05, + "loss": 2.6547920227050783, + "step": 6980 + }, + { + "epoch": 0.01905539676058255, + "grad_norm": 2.624518871307373, + "learning_rate": 1.96456862477852e-05, + "loss": 2.6476823806762697, + "step": 7000 + }, + { + "epoch": 0.019109840751327073, + "grad_norm": 2.2461345195770264, + "learning_rate": 1.9644595883876246e-05, + "loss": 2.615263557434082, + "step": 7020 + }, + { + "epoch": 0.019164284742071595, + "grad_norm": 2.639742612838745, + "learning_rate": 1.9643505519967288e-05, + "loss": 2.641188049316406, + "step": 7040 + }, + { + "epoch": 0.019218728732816114, + "grad_norm": 2.3193306922912598, + "learning_rate": 1.9642415156058337e-05, + "loss": 2.6427698135375977, + "step": 7060 + }, + { + "epoch": 0.019273172723560637, + "grad_norm": 2.3581743240356445, + "learning_rate": 1.964132479214938e-05, + "loss": 2.6176355361938475, + "step": 7080 + }, + { + "epoch": 0.01932761671430516, + "grad_norm": 2.334656000137329, + "learning_rate": 1.964023442824043e-05, + "loss": 2.6335737228393556, + "step": 7100 + }, + { + "epoch": 0.01938206070504968, + "grad_norm": 2.3797013759613037, + "learning_rate": 1.963914406433147e-05, + "loss": 2.672977638244629, + "step": 7120 + }, + { + "epoch": 0.0194365046957942, + "grad_norm": 2.8761346340179443, + "learning_rate": 1.9638053700422517e-05, + "loss": 2.6477827072143554, + "step": 7140 + }, + { + "epoch": 0.019490948686538723, + "grad_norm": 2.2513325214385986, + "learning_rate": 1.9636963336513563e-05, + "loss": 2.6459928512573243, + "step": 7160 + }, + { + "epoch": 0.019545392677283245, + "grad_norm": 2.2563860416412354, + "learning_rate": 1.963587297260461e-05, + "loss": 2.662177085876465, + "step": 7180 + }, + { + "epoch": 0.019599836668027767, + "grad_norm": 2.429290771484375, + "learning_rate": 1.963478260869565e-05, + "loss": 2.627259063720703, + "step": 7200 + }, + { + "epoch": 0.019654280658772286, + "grad_norm": 2.317233085632324, + "learning_rate": 1.96336922447867e-05, + "loss": 2.600942039489746, + "step": 7220 + }, + { + "epoch": 0.01970872464951681, + "grad_norm": 2.2577219009399414, + "learning_rate": 1.9632601880877743e-05, + "loss": 2.6076692581176757, + "step": 7240 + }, + { + "epoch": 0.01976316864026133, + "grad_norm": 2.3724260330200195, + "learning_rate": 1.9631511516968792e-05, + "loss": 2.6431671142578126, + "step": 7260 + }, + { + "epoch": 0.019817612631005854, + "grad_norm": 2.2196528911590576, + "learning_rate": 1.9630421153059835e-05, + "loss": 2.639164924621582, + "step": 7280 + }, + { + "epoch": 0.019872056621750376, + "grad_norm": 2.3705344200134277, + "learning_rate": 1.962933078915088e-05, + "loss": 2.6664371490478516, + "step": 7300 + }, + { + "epoch": 0.019926500612494895, + "grad_norm": 2.302699565887451, + "learning_rate": 1.9628240425241926e-05, + "loss": 2.6341873168945313, + "step": 7320 + }, + { + "epoch": 0.019980944603239417, + "grad_norm": 2.5890071392059326, + "learning_rate": 1.9627150061332972e-05, + "loss": 2.6701412200927734, + "step": 7340 + }, + { + "epoch": 0.02003538859398394, + "grad_norm": 2.498607635498047, + "learning_rate": 1.9626059697424014e-05, + "loss": 2.6385772705078123, + "step": 7360 + }, + { + "epoch": 0.020089832584728462, + "grad_norm": 2.3113768100738525, + "learning_rate": 1.9624969333515064e-05, + "loss": 2.6423564910888673, + "step": 7380 + }, + { + "epoch": 0.02014427657547298, + "grad_norm": 2.3692991733551025, + "learning_rate": 1.9623878969606106e-05, + "loss": 2.647845649719238, + "step": 7400 + }, + { + "epoch": 0.020198720566217503, + "grad_norm": 2.3399300575256348, + "learning_rate": 1.9622788605697155e-05, + "loss": 2.6335628509521483, + "step": 7420 + }, + { + "epoch": 0.020253164556962026, + "grad_norm": 2.3926939964294434, + "learning_rate": 1.9621698241788198e-05, + "loss": 2.6461971282958983, + "step": 7440 + }, + { + "epoch": 0.020307608547706548, + "grad_norm": 2.336177349090576, + "learning_rate": 1.9620607877879243e-05, + "loss": 2.6322216033935546, + "step": 7460 + }, + { + "epoch": 0.020362052538451067, + "grad_norm": 2.289381504058838, + "learning_rate": 1.961951751397029e-05, + "loss": 2.6373619079589843, + "step": 7480 + }, + { + "epoch": 0.02041649652919559, + "grad_norm": 2.3485894203186035, + "learning_rate": 1.9618427150061335e-05, + "loss": 2.6251033782958983, + "step": 7500 + }, + { + "epoch": 0.020470940519940112, + "grad_norm": 2.532811164855957, + "learning_rate": 1.9617336786152377e-05, + "loss": 2.614103889465332, + "step": 7520 + }, + { + "epoch": 0.020525384510684634, + "grad_norm": 2.4177374839782715, + "learning_rate": 1.9616246422243427e-05, + "loss": 2.6297245025634766, + "step": 7540 + }, + { + "epoch": 0.020579828501429153, + "grad_norm": 2.391084671020508, + "learning_rate": 1.961515605833447e-05, + "loss": 2.625911903381348, + "step": 7560 + }, + { + "epoch": 0.020634272492173675, + "grad_norm": 2.3648695945739746, + "learning_rate": 1.9614065694425518e-05, + "loss": 2.6411840438842775, + "step": 7580 + }, + { + "epoch": 0.020688716482918198, + "grad_norm": 2.4067471027374268, + "learning_rate": 1.961297533051656e-05, + "loss": 2.646944999694824, + "step": 7600 + }, + { + "epoch": 0.02074316047366272, + "grad_norm": 2.403339147567749, + "learning_rate": 1.9611884966607606e-05, + "loss": 2.62078971862793, + "step": 7620 + }, + { + "epoch": 0.020797604464407243, + "grad_norm": 2.3990821838378906, + "learning_rate": 1.9610794602698652e-05, + "loss": 2.6024795532226563, + "step": 7640 + }, + { + "epoch": 0.02085204845515176, + "grad_norm": 2.6341054439544678, + "learning_rate": 1.9609704238789698e-05, + "loss": 2.6198297500610352, + "step": 7660 + }, + { + "epoch": 0.020906492445896284, + "grad_norm": 2.4670863151550293, + "learning_rate": 1.960861387488074e-05, + "loss": 2.629256248474121, + "step": 7680 + }, + { + "epoch": 0.020960936436640806, + "grad_norm": 2.3301124572753906, + "learning_rate": 1.960752351097179e-05, + "loss": 2.594441223144531, + "step": 7700 + }, + { + "epoch": 0.02101538042738533, + "grad_norm": 2.304774284362793, + "learning_rate": 1.9606433147062832e-05, + "loss": 2.6380380630493163, + "step": 7720 + }, + { + "epoch": 0.021069824418129848, + "grad_norm": 2.383925437927246, + "learning_rate": 1.960534278315388e-05, + "loss": 2.5886375427246096, + "step": 7740 + }, + { + "epoch": 0.02112426840887437, + "grad_norm": 2.317357063293457, + "learning_rate": 1.9604252419244924e-05, + "loss": 2.6117382049560547, + "step": 7760 + }, + { + "epoch": 0.021178712399618892, + "grad_norm": 2.4071249961853027, + "learning_rate": 1.960316205533597e-05, + "loss": 2.637472724914551, + "step": 7780 + }, + { + "epoch": 0.021233156390363415, + "grad_norm": 2.378729820251465, + "learning_rate": 1.9602071691427015e-05, + "loss": 2.642986297607422, + "step": 7800 + }, + { + "epoch": 0.021287600381107934, + "grad_norm": 2.3462395668029785, + "learning_rate": 1.960098132751806e-05, + "loss": 2.619418144226074, + "step": 7820 + }, + { + "epoch": 0.021342044371852456, + "grad_norm": 2.328014373779297, + "learning_rate": 1.9599890963609104e-05, + "loss": 2.655771255493164, + "step": 7840 + }, + { + "epoch": 0.02139648836259698, + "grad_norm": 3.381640672683716, + "learning_rate": 1.9598800599700153e-05, + "loss": 2.6407169342041015, + "step": 7860 + }, + { + "epoch": 0.0214509323533415, + "grad_norm": 2.366079092025757, + "learning_rate": 1.9597710235791195e-05, + "loss": 2.626661491394043, + "step": 7880 + }, + { + "epoch": 0.021505376344086023, + "grad_norm": 2.455803394317627, + "learning_rate": 1.9596619871882244e-05, + "loss": 2.644400978088379, + "step": 7900 + }, + { + "epoch": 0.021559820334830542, + "grad_norm": 2.285158634185791, + "learning_rate": 1.9595529507973287e-05, + "loss": 2.6488250732421874, + "step": 7920 + }, + { + "epoch": 0.021614264325575065, + "grad_norm": 2.4078145027160645, + "learning_rate": 1.9594439144064333e-05, + "loss": 2.6613616943359375, + "step": 7940 + }, + { + "epoch": 0.021668708316319587, + "grad_norm": 2.2150356769561768, + "learning_rate": 1.959334878015538e-05, + "loss": 2.6536890029907227, + "step": 7960 + }, + { + "epoch": 0.02172315230706411, + "grad_norm": 2.3267881870269775, + "learning_rate": 1.9592258416246424e-05, + "loss": 2.614603042602539, + "step": 7980 + }, + { + "epoch": 0.021777596297808628, + "grad_norm": 2.4170496463775635, + "learning_rate": 1.959116805233747e-05, + "loss": 2.6443449020385743, + "step": 8000 + }, + { + "epoch": 0.02183204028855315, + "grad_norm": 2.24118971824646, + "learning_rate": 1.9590077688428516e-05, + "loss": 2.650377082824707, + "step": 8020 + }, + { + "epoch": 0.021886484279297673, + "grad_norm": 2.6708850860595703, + "learning_rate": 1.958898732451956e-05, + "loss": 2.6127445220947267, + "step": 8040 + }, + { + "epoch": 0.021940928270042195, + "grad_norm": 2.312258005142212, + "learning_rate": 1.9587896960610608e-05, + "loss": 2.640095520019531, + "step": 8060 + }, + { + "epoch": 0.021995372260786714, + "grad_norm": 2.2890970706939697, + "learning_rate": 1.958680659670165e-05, + "loss": 2.638703155517578, + "step": 8080 + }, + { + "epoch": 0.022049816251531237, + "grad_norm": 2.240736961364746, + "learning_rate": 1.9585716232792696e-05, + "loss": 2.6037471771240233, + "step": 8100 + }, + { + "epoch": 0.02210426024227576, + "grad_norm": 3.4732680320739746, + "learning_rate": 1.958462586888374e-05, + "loss": 2.6117227554321287, + "step": 8120 + }, + { + "epoch": 0.02215870423302028, + "grad_norm": 2.3625242710113525, + "learning_rate": 1.9583535504974787e-05, + "loss": 2.6354246139526367, + "step": 8140 + }, + { + "epoch": 0.0222131482237648, + "grad_norm": 4.423672676086426, + "learning_rate": 1.9582445141065833e-05, + "loss": 2.614254951477051, + "step": 8160 + }, + { + "epoch": 0.022267592214509323, + "grad_norm": 2.354098081588745, + "learning_rate": 1.958135477715688e-05, + "loss": 2.6341033935546876, + "step": 8180 + }, + { + "epoch": 0.022322036205253845, + "grad_norm": 2.197171926498413, + "learning_rate": 1.958026441324792e-05, + "loss": 2.585676574707031, + "step": 8200 + }, + { + "epoch": 0.022376480195998368, + "grad_norm": 2.320507049560547, + "learning_rate": 1.957917404933897e-05, + "loss": 2.6232332229614257, + "step": 8220 + }, + { + "epoch": 0.02243092418674289, + "grad_norm": 2.299049139022827, + "learning_rate": 1.9578083685430013e-05, + "loss": 2.6251869201660156, + "step": 8240 + }, + { + "epoch": 0.02248536817748741, + "grad_norm": 2.306583881378174, + "learning_rate": 1.957699332152106e-05, + "loss": 2.6217281341552736, + "step": 8260 + }, + { + "epoch": 0.02253981216823193, + "grad_norm": 2.380938768386841, + "learning_rate": 1.9575902957612105e-05, + "loss": 2.6361759185791014, + "step": 8280 + }, + { + "epoch": 0.022594256158976454, + "grad_norm": 2.2601335048675537, + "learning_rate": 1.957481259370315e-05, + "loss": 2.6136028289794924, + "step": 8300 + }, + { + "epoch": 0.022648700149720976, + "grad_norm": 2.273740530014038, + "learning_rate": 1.9573722229794196e-05, + "loss": 2.666065788269043, + "step": 8320 + }, + { + "epoch": 0.022703144140465495, + "grad_norm": 2.3165953159332275, + "learning_rate": 1.9572631865885242e-05, + "loss": 2.600249099731445, + "step": 8340 + }, + { + "epoch": 0.022757588131210017, + "grad_norm": 2.303476572036743, + "learning_rate": 1.9571541501976284e-05, + "loss": 2.624444580078125, + "step": 8360 + }, + { + "epoch": 0.02281203212195454, + "grad_norm": 2.338021993637085, + "learning_rate": 1.9570451138067334e-05, + "loss": 2.6184654235839844, + "step": 8380 + }, + { + "epoch": 0.022866476112699062, + "grad_norm": 2.2363946437835693, + "learning_rate": 1.9569360774158376e-05, + "loss": 2.6572195053100587, + "step": 8400 + }, + { + "epoch": 0.02292092010344358, + "grad_norm": 2.259319305419922, + "learning_rate": 1.9568270410249422e-05, + "loss": 2.623958396911621, + "step": 8420 + }, + { + "epoch": 0.022975364094188103, + "grad_norm": 2.2682251930236816, + "learning_rate": 1.9567180046340468e-05, + "loss": 2.6325214385986326, + "step": 8440 + }, + { + "epoch": 0.023029808084932626, + "grad_norm": 2.2544076442718506, + "learning_rate": 1.9566089682431514e-05, + "loss": 2.6254024505615234, + "step": 8460 + }, + { + "epoch": 0.023084252075677148, + "grad_norm": 2.192348003387451, + "learning_rate": 1.956499931852256e-05, + "loss": 2.627546501159668, + "step": 8480 + }, + { + "epoch": 0.023138696066421667, + "grad_norm": 2.2892045974731445, + "learning_rate": 1.9563908954613605e-05, + "loss": 2.622854995727539, + "step": 8500 + }, + { + "epoch": 0.02319314005716619, + "grad_norm": 2.3182220458984375, + "learning_rate": 1.9562818590704648e-05, + "loss": 2.63751277923584, + "step": 8520 + }, + { + "epoch": 0.023247584047910712, + "grad_norm": 2.230804204940796, + "learning_rate": 1.9561728226795697e-05, + "loss": 2.6245073318481444, + "step": 8540 + }, + { + "epoch": 0.023302028038655234, + "grad_norm": 2.5811455249786377, + "learning_rate": 1.956063786288674e-05, + "loss": 2.6010272979736326, + "step": 8560 + }, + { + "epoch": 0.023356472029399757, + "grad_norm": 2.219494104385376, + "learning_rate": 1.9559547498977785e-05, + "loss": 2.624721145629883, + "step": 8580 + }, + { + "epoch": 0.023410916020144276, + "grad_norm": 2.276458978652954, + "learning_rate": 1.955845713506883e-05, + "loss": 2.636305236816406, + "step": 8600 + }, + { + "epoch": 0.023465360010888798, + "grad_norm": 2.335937261581421, + "learning_rate": 1.9557366771159877e-05, + "loss": 2.5898569107055662, + "step": 8620 + }, + { + "epoch": 0.02351980400163332, + "grad_norm": 2.2484588623046875, + "learning_rate": 1.9556276407250922e-05, + "loss": 2.5879602432250977, + "step": 8640 + }, + { + "epoch": 0.023574247992377843, + "grad_norm": 2.7979156970977783, + "learning_rate": 1.9555186043341968e-05, + "loss": 2.6012805938720702, + "step": 8660 + }, + { + "epoch": 0.02362869198312236, + "grad_norm": 2.245434284210205, + "learning_rate": 1.955409567943301e-05, + "loss": 2.6211736679077147, + "step": 8680 + }, + { + "epoch": 0.023683135973866884, + "grad_norm": 2.4291470050811768, + "learning_rate": 1.955300531552406e-05, + "loss": 2.6412399291992186, + "step": 8700 + }, + { + "epoch": 0.023737579964611406, + "grad_norm": 2.486508369445801, + "learning_rate": 1.9551914951615102e-05, + "loss": 2.5884931564331053, + "step": 8720 + }, + { + "epoch": 0.02379202395535593, + "grad_norm": 2.293489694595337, + "learning_rate": 1.9550824587706148e-05, + "loss": 2.6146230697631836, + "step": 8740 + }, + { + "epoch": 0.023846467946100448, + "grad_norm": 2.2903099060058594, + "learning_rate": 1.9549734223797194e-05, + "loss": 2.6132999420166017, + "step": 8760 + }, + { + "epoch": 0.02390091193684497, + "grad_norm": 2.288238763809204, + "learning_rate": 1.954864385988824e-05, + "loss": 2.621958923339844, + "step": 8780 + }, + { + "epoch": 0.023955355927589492, + "grad_norm": 2.2691304683685303, + "learning_rate": 1.9547553495979286e-05, + "loss": 2.5927408218383787, + "step": 8800 + }, + { + "epoch": 0.024009799918334015, + "grad_norm": 2.2664055824279785, + "learning_rate": 1.954646313207033e-05, + "loss": 2.6112545013427733, + "step": 8820 + }, + { + "epoch": 0.024064243909078534, + "grad_norm": 2.2946133613586426, + "learning_rate": 1.9545372768161374e-05, + "loss": 2.5841922760009766, + "step": 8840 + }, + { + "epoch": 0.024118687899823056, + "grad_norm": 2.287641763687134, + "learning_rate": 1.9544282404252423e-05, + "loss": 2.6105119705200197, + "step": 8860 + }, + { + "epoch": 0.02417313189056758, + "grad_norm": 2.536757469177246, + "learning_rate": 1.9543192040343465e-05, + "loss": 2.6605457305908202, + "step": 8880 + }, + { + "epoch": 0.0242275758813121, + "grad_norm": 2.14695143699646, + "learning_rate": 1.954210167643451e-05, + "loss": 2.589647674560547, + "step": 8900 + }, + { + "epoch": 0.024282019872056623, + "grad_norm": 2.2110941410064697, + "learning_rate": 1.9541011312525557e-05, + "loss": 2.6293182373046875, + "step": 8920 + }, + { + "epoch": 0.024336463862801142, + "grad_norm": 2.463189125061035, + "learning_rate": 1.9539920948616603e-05, + "loss": 2.6061384201049806, + "step": 8940 + }, + { + "epoch": 0.024390907853545665, + "grad_norm": 2.8322672843933105, + "learning_rate": 1.953883058470765e-05, + "loss": 2.6064870834350584, + "step": 8960 + }, + { + "epoch": 0.024445351844290187, + "grad_norm": 2.413684606552124, + "learning_rate": 1.9537740220798694e-05, + "loss": 2.607894515991211, + "step": 8980 + }, + { + "epoch": 0.02449979583503471, + "grad_norm": 2.336625337600708, + "learning_rate": 1.9536649856889737e-05, + "loss": 2.647352600097656, + "step": 9000 + }, + { + "epoch": 0.024554239825779228, + "grad_norm": 2.2014613151550293, + "learning_rate": 1.9535559492980786e-05, + "loss": 2.6162220001220704, + "step": 9020 + }, + { + "epoch": 0.02460868381652375, + "grad_norm": 2.348970651626587, + "learning_rate": 1.953446912907183e-05, + "loss": 2.6134311676025392, + "step": 9040 + }, + { + "epoch": 0.024663127807268273, + "grad_norm": 2.8069231510162354, + "learning_rate": 1.9533378765162874e-05, + "loss": 2.6337940216064455, + "step": 9060 + }, + { + "epoch": 0.024717571798012795, + "grad_norm": 2.242887496948242, + "learning_rate": 1.953228840125392e-05, + "loss": 2.632855987548828, + "step": 9080 + }, + { + "epoch": 0.024772015788757314, + "grad_norm": 2.205592393875122, + "learning_rate": 1.9531198037344966e-05, + "loss": 2.610125923156738, + "step": 9100 + }, + { + "epoch": 0.024826459779501837, + "grad_norm": 2.3043699264526367, + "learning_rate": 1.953010767343601e-05, + "loss": 2.625686454772949, + "step": 9120 + }, + { + "epoch": 0.02488090377024636, + "grad_norm": 2.728844404220581, + "learning_rate": 1.9529017309527057e-05, + "loss": 2.6040922164916993, + "step": 9140 + }, + { + "epoch": 0.02493534776099088, + "grad_norm": 2.2469406127929688, + "learning_rate": 1.95279269456181e-05, + "loss": 2.5896223068237303, + "step": 9160 + }, + { + "epoch": 0.024989791751735404, + "grad_norm": 2.351710557937622, + "learning_rate": 1.952683658170915e-05, + "loss": 2.6387964248657227, + "step": 9180 + }, + { + "epoch": 0.025044235742479923, + "grad_norm": 2.2811877727508545, + "learning_rate": 1.952574621780019e-05, + "loss": 2.630380058288574, + "step": 9200 + }, + { + "epoch": 0.025098679733224445, + "grad_norm": 2.378495216369629, + "learning_rate": 1.9524655853891237e-05, + "loss": 2.605019760131836, + "step": 9220 + }, + { + "epoch": 0.025153123723968968, + "grad_norm": 2.2248475551605225, + "learning_rate": 1.9523565489982283e-05, + "loss": 2.6159341812133787, + "step": 9240 + }, + { + "epoch": 0.02520756771471349, + "grad_norm": 2.2285501956939697, + "learning_rate": 1.952247512607333e-05, + "loss": 2.634505271911621, + "step": 9260 + }, + { + "epoch": 0.02526201170545801, + "grad_norm": 2.3579344749450684, + "learning_rate": 1.9521384762164375e-05, + "loss": 2.6416406631469727, + "step": 9280 + }, + { + "epoch": 0.02531645569620253, + "grad_norm": 2.197890520095825, + "learning_rate": 1.952029439825542e-05, + "loss": 2.6118698120117188, + "step": 9300 + }, + { + "epoch": 0.025370899686947054, + "grad_norm": 2.2764108180999756, + "learning_rate": 1.9519204034346463e-05, + "loss": 2.612049674987793, + "step": 9320 + }, + { + "epoch": 0.025425343677691576, + "grad_norm": 2.4146625995635986, + "learning_rate": 1.9518113670437512e-05, + "loss": 2.5929004669189455, + "step": 9340 + }, + { + "epoch": 0.025479787668436095, + "grad_norm": 2.3899855613708496, + "learning_rate": 1.9517023306528555e-05, + "loss": 2.6047882080078124, + "step": 9360 + }, + { + "epoch": 0.025534231659180617, + "grad_norm": 2.3715109825134277, + "learning_rate": 1.95159329426196e-05, + "loss": 2.6139591217041014, + "step": 9380 + }, + { + "epoch": 0.02558867564992514, + "grad_norm": 2.3378632068634033, + "learning_rate": 1.9514842578710646e-05, + "loss": 2.6192863464355467, + "step": 9400 + }, + { + "epoch": 0.025643119640669662, + "grad_norm": 2.2848310470581055, + "learning_rate": 1.9513752214801692e-05, + "loss": 2.5808616638183595, + "step": 9420 + }, + { + "epoch": 0.02569756363141418, + "grad_norm": 2.3653371334075928, + "learning_rate": 1.9512661850892738e-05, + "loss": 2.647669792175293, + "step": 9440 + }, + { + "epoch": 0.025752007622158703, + "grad_norm": 2.320772647857666, + "learning_rate": 1.9511571486983784e-05, + "loss": 2.5948068618774416, + "step": 9460 + }, + { + "epoch": 0.025806451612903226, + "grad_norm": 2.174755334854126, + "learning_rate": 1.9510481123074826e-05, + "loss": 2.6143543243408205, + "step": 9480 + }, + { + "epoch": 0.025860895603647748, + "grad_norm": 2.2083218097686768, + "learning_rate": 1.9509390759165875e-05, + "loss": 2.6226375579833983, + "step": 9500 + }, + { + "epoch": 0.02591533959439227, + "grad_norm": 2.38352370262146, + "learning_rate": 1.9508300395256918e-05, + "loss": 2.619731140136719, + "step": 9520 + }, + { + "epoch": 0.02596978358513679, + "grad_norm": 2.198093891143799, + "learning_rate": 1.9507210031347964e-05, + "loss": 2.6293367385864257, + "step": 9540 + }, + { + "epoch": 0.026024227575881312, + "grad_norm": 2.205353260040283, + "learning_rate": 1.950611966743901e-05, + "loss": 2.6266626358032226, + "step": 9560 + }, + { + "epoch": 0.026078671566625834, + "grad_norm": 2.24822998046875, + "learning_rate": 1.9505029303530055e-05, + "loss": 2.6256099700927735, + "step": 9580 + }, + { + "epoch": 0.026133115557370357, + "grad_norm": 2.4705026149749756, + "learning_rate": 1.95039389396211e-05, + "loss": 2.577090835571289, + "step": 9600 + }, + { + "epoch": 0.026187559548114876, + "grad_norm": 2.291149616241455, + "learning_rate": 1.9502848575712147e-05, + "loss": 2.6024797439575194, + "step": 9620 + }, + { + "epoch": 0.026242003538859398, + "grad_norm": 2.202991247177124, + "learning_rate": 1.950175821180319e-05, + "loss": 2.613430976867676, + "step": 9640 + }, + { + "epoch": 0.02629644752960392, + "grad_norm": 2.329167127609253, + "learning_rate": 1.950066784789424e-05, + "loss": 2.597941589355469, + "step": 9660 + }, + { + "epoch": 0.026350891520348443, + "grad_norm": 2.2481865882873535, + "learning_rate": 1.949957748398528e-05, + "loss": 2.6108423233032227, + "step": 9680 + }, + { + "epoch": 0.02640533551109296, + "grad_norm": 2.227247953414917, + "learning_rate": 1.9498487120076327e-05, + "loss": 2.619403266906738, + "step": 9700 + }, + { + "epoch": 0.026459779501837484, + "grad_norm": 2.3269782066345215, + "learning_rate": 1.9497396756167372e-05, + "loss": 2.6000907897949217, + "step": 9720 + }, + { + "epoch": 0.026514223492582006, + "grad_norm": 2.3307888507843018, + "learning_rate": 1.9496306392258418e-05, + "loss": 2.632893371582031, + "step": 9740 + }, + { + "epoch": 0.02656866748332653, + "grad_norm": 2.3236749172210693, + "learning_rate": 1.9495216028349464e-05, + "loss": 2.6280855178833007, + "step": 9760 + }, + { + "epoch": 0.026623111474071048, + "grad_norm": 2.4158287048339844, + "learning_rate": 1.949412566444051e-05, + "loss": 2.6431665420532227, + "step": 9780 + }, + { + "epoch": 0.02667755546481557, + "grad_norm": 2.416896104812622, + "learning_rate": 1.9493035300531552e-05, + "loss": 2.6015657424926757, + "step": 9800 + }, + { + "epoch": 0.026731999455560092, + "grad_norm": 2.3563292026519775, + "learning_rate": 1.94919449366226e-05, + "loss": 2.6156383514404298, + "step": 9820 + }, + { + "epoch": 0.026786443446304615, + "grad_norm": 2.312419891357422, + "learning_rate": 1.9490854572713644e-05, + "loss": 2.5764488220214843, + "step": 9840 + }, + { + "epoch": 0.026840887437049137, + "grad_norm": 2.2478036880493164, + "learning_rate": 1.948976420880469e-05, + "loss": 2.6238847732543946, + "step": 9860 + }, + { + "epoch": 0.026895331427793656, + "grad_norm": 2.409233331680298, + "learning_rate": 1.9488673844895735e-05, + "loss": 2.595559501647949, + "step": 9880 + }, + { + "epoch": 0.02694977541853818, + "grad_norm": 2.38667893409729, + "learning_rate": 1.948758348098678e-05, + "loss": 2.595134735107422, + "step": 9900 + }, + { + "epoch": 0.0270042194092827, + "grad_norm": 2.2159881591796875, + "learning_rate": 1.9486493117077827e-05, + "loss": 2.6372074127197265, + "step": 9920 + }, + { + "epoch": 0.027058663400027223, + "grad_norm": 2.3106236457824707, + "learning_rate": 1.9485402753168873e-05, + "loss": 2.6142337799072264, + "step": 9940 + }, + { + "epoch": 0.027113107390771742, + "grad_norm": 2.2233681678771973, + "learning_rate": 1.9484312389259915e-05, + "loss": 2.652903938293457, + "step": 9960 + }, + { + "epoch": 0.027167551381516265, + "grad_norm": 2.980315685272217, + "learning_rate": 1.9483222025350965e-05, + "loss": 2.5994218826293944, + "step": 9980 + }, + { + "epoch": 0.027221995372260787, + "grad_norm": 2.2768394947052, + "learning_rate": 1.9482131661442007e-05, + "loss": 2.6040555953979494, + "step": 10000 + }, + { + "epoch": 0.02727643936300531, + "grad_norm": 2.1772572994232178, + "learning_rate": 1.9481041297533053e-05, + "loss": 2.604224967956543, + "step": 10020 + }, + { + "epoch": 0.02733088335374983, + "grad_norm": 2.2202649116516113, + "learning_rate": 1.94799509336241e-05, + "loss": 2.567169189453125, + "step": 10040 + }, + { + "epoch": 0.02738532734449435, + "grad_norm": 2.4311416149139404, + "learning_rate": 1.9478860569715144e-05, + "loss": 2.634398078918457, + "step": 10060 + }, + { + "epoch": 0.027439771335238873, + "grad_norm": 2.2554166316986084, + "learning_rate": 1.947777020580619e-05, + "loss": 2.6026803970336916, + "step": 10080 + }, + { + "epoch": 0.027494215325983395, + "grad_norm": 2.2548506259918213, + "learning_rate": 1.9476679841897236e-05, + "loss": 2.6032436370849608, + "step": 10100 + }, + { + "epoch": 0.027548659316727918, + "grad_norm": 2.171844720840454, + "learning_rate": 1.947558947798828e-05, + "loss": 2.6244327545166017, + "step": 10120 + }, + { + "epoch": 0.027603103307472437, + "grad_norm": 2.211113691329956, + "learning_rate": 1.9474499114079328e-05, + "loss": 2.590160942077637, + "step": 10140 + }, + { + "epoch": 0.02765754729821696, + "grad_norm": 2.253783702850342, + "learning_rate": 1.947340875017037e-05, + "loss": 2.6217153549194334, + "step": 10160 + }, + { + "epoch": 0.02771199128896148, + "grad_norm": 2.2662389278411865, + "learning_rate": 1.9472318386261416e-05, + "loss": 2.6277486801147463, + "step": 10180 + }, + { + "epoch": 0.027766435279706004, + "grad_norm": 2.479433536529541, + "learning_rate": 1.947122802235246e-05, + "loss": 2.6133777618408205, + "step": 10200 + }, + { + "epoch": 0.027820879270450523, + "grad_norm": 2.3602194786071777, + "learning_rate": 1.9470137658443507e-05, + "loss": 2.664076042175293, + "step": 10220 + }, + { + "epoch": 0.027875323261195045, + "grad_norm": 2.2287609577178955, + "learning_rate": 1.9469047294534553e-05, + "loss": 2.60113468170166, + "step": 10240 + }, + { + "epoch": 0.027929767251939568, + "grad_norm": 2.975937604904175, + "learning_rate": 1.94679569306256e-05, + "loss": 2.6026798248291017, + "step": 10260 + }, + { + "epoch": 0.02798421124268409, + "grad_norm": 2.4937245845794678, + "learning_rate": 1.946686656671664e-05, + "loss": 2.6225103378295898, + "step": 10280 + }, + { + "epoch": 0.02803865523342861, + "grad_norm": 2.2510669231414795, + "learning_rate": 1.946577620280769e-05, + "loss": 2.5986349105834963, + "step": 10300 + }, + { + "epoch": 0.02809309922417313, + "grad_norm": 2.5198729038238525, + "learning_rate": 1.9464685838898733e-05, + "loss": 2.6127029418945313, + "step": 10320 + }, + { + "epoch": 0.028147543214917654, + "grad_norm": 2.247589588165283, + "learning_rate": 1.946359547498978e-05, + "loss": 2.6075767517089843, + "step": 10340 + }, + { + "epoch": 0.028201987205662176, + "grad_norm": 2.5265119075775146, + "learning_rate": 1.9462505111080825e-05, + "loss": 2.6152820587158203, + "step": 10360 + }, + { + "epoch": 0.028256431196406695, + "grad_norm": 2.310696601867676, + "learning_rate": 1.946141474717187e-05, + "loss": 2.6304676055908205, + "step": 10380 + }, + { + "epoch": 0.028310875187151217, + "grad_norm": 2.278599739074707, + "learning_rate": 1.9460324383262916e-05, + "loss": 2.623044967651367, + "step": 10400 + }, + { + "epoch": 0.02836531917789574, + "grad_norm": 2.2487664222717285, + "learning_rate": 1.9459234019353962e-05, + "loss": 2.5942670822143556, + "step": 10420 + }, + { + "epoch": 0.028419763168640262, + "grad_norm": 2.313417911529541, + "learning_rate": 1.9458143655445005e-05, + "loss": 2.6100910186767576, + "step": 10440 + }, + { + "epoch": 0.028474207159384785, + "grad_norm": 2.2188608646392822, + "learning_rate": 1.9457053291536054e-05, + "loss": 2.6193504333496094, + "step": 10460 + }, + { + "epoch": 0.028528651150129303, + "grad_norm": 2.5278682708740234, + "learning_rate": 1.9455962927627096e-05, + "loss": 2.6427980422973634, + "step": 10480 + }, + { + "epoch": 0.028583095140873826, + "grad_norm": 2.2812819480895996, + "learning_rate": 1.9454872563718142e-05, + "loss": 2.628111457824707, + "step": 10500 + }, + { + "epoch": 0.028637539131618348, + "grad_norm": 2.481032609939575, + "learning_rate": 1.9453782199809188e-05, + "loss": 2.6134586334228516, + "step": 10520 + }, + { + "epoch": 0.02869198312236287, + "grad_norm": 2.373839855194092, + "learning_rate": 1.9452691835900234e-05, + "loss": 2.6057050704956053, + "step": 10540 + }, + { + "epoch": 0.02874642711310739, + "grad_norm": 2.213454008102417, + "learning_rate": 1.945160147199128e-05, + "loss": 2.593522071838379, + "step": 10560 + }, + { + "epoch": 0.028800871103851912, + "grad_norm": 2.475458860397339, + "learning_rate": 1.9450511108082325e-05, + "loss": 2.610438919067383, + "step": 10580 + }, + { + "epoch": 0.028855315094596434, + "grad_norm": 2.2845780849456787, + "learning_rate": 1.9449420744173368e-05, + "loss": 2.633715057373047, + "step": 10600 + }, + { + "epoch": 0.028909759085340957, + "grad_norm": 2.201991558074951, + "learning_rate": 1.9448330380264417e-05, + "loss": 2.61590461730957, + "step": 10620 + }, + { + "epoch": 0.028964203076085476, + "grad_norm": 2.778975248336792, + "learning_rate": 1.944724001635546e-05, + "loss": 2.6128608703613283, + "step": 10640 + }, + { + "epoch": 0.029018647066829998, + "grad_norm": 2.193122148513794, + "learning_rate": 1.9446149652446505e-05, + "loss": 2.6094291687011717, + "step": 10660 + }, + { + "epoch": 0.02907309105757452, + "grad_norm": 2.363051176071167, + "learning_rate": 1.944505928853755e-05, + "loss": 2.5997007369995115, + "step": 10680 + }, + { + "epoch": 0.029127535048319043, + "grad_norm": 2.3052282333374023, + "learning_rate": 1.9443968924628597e-05, + "loss": 2.628156280517578, + "step": 10700 + }, + { + "epoch": 0.02918197903906356, + "grad_norm": 2.9887688159942627, + "learning_rate": 1.9442878560719643e-05, + "loss": 2.613487434387207, + "step": 10720 + }, + { + "epoch": 0.029236423029808084, + "grad_norm": 2.2481255531311035, + "learning_rate": 1.944178819681069e-05, + "loss": 2.604864501953125, + "step": 10740 + }, + { + "epoch": 0.029290867020552606, + "grad_norm": 2.2239491939544678, + "learning_rate": 1.944069783290173e-05, + "loss": 2.631897735595703, + "step": 10760 + }, + { + "epoch": 0.02934531101129713, + "grad_norm": 2.202369451522827, + "learning_rate": 1.943960746899278e-05, + "loss": 2.6106491088867188, + "step": 10780 + }, + { + "epoch": 0.02939975500204165, + "grad_norm": 2.2847602367401123, + "learning_rate": 1.9438517105083822e-05, + "loss": 2.602001762390137, + "step": 10800 + }, + { + "epoch": 0.02945419899278617, + "grad_norm": 2.5437426567077637, + "learning_rate": 1.9437426741174868e-05, + "loss": 2.5967294692993166, + "step": 10820 + }, + { + "epoch": 0.029508642983530693, + "grad_norm": 2.228459119796753, + "learning_rate": 1.9436336377265914e-05, + "loss": 2.6024120330810545, + "step": 10840 + }, + { + "epoch": 0.029563086974275215, + "grad_norm": 2.2122926712036133, + "learning_rate": 1.943524601335696e-05, + "loss": 2.596120262145996, + "step": 10860 + }, + { + "epoch": 0.029617530965019737, + "grad_norm": 2.2639241218566895, + "learning_rate": 1.9434155649448006e-05, + "loss": 2.6174684524536134, + "step": 10880 + }, + { + "epoch": 0.029671974955764256, + "grad_norm": 2.245857000350952, + "learning_rate": 1.943306528553905e-05, + "loss": 2.6096332550048826, + "step": 10900 + }, + { + "epoch": 0.02972641894650878, + "grad_norm": 2.2833709716796875, + "learning_rate": 1.9431974921630094e-05, + "loss": 2.61600284576416, + "step": 10920 + }, + { + "epoch": 0.0297808629372533, + "grad_norm": 2.3462467193603516, + "learning_rate": 1.943088455772114e-05, + "loss": 2.61390380859375, + "step": 10940 + }, + { + "epoch": 0.029835306927997823, + "grad_norm": 2.374663829803467, + "learning_rate": 1.9429794193812185e-05, + "loss": 2.6147518157958984, + "step": 10960 + }, + { + "epoch": 0.029889750918742342, + "grad_norm": 2.2471401691436768, + "learning_rate": 1.942870382990323e-05, + "loss": 2.596370506286621, + "step": 10980 + }, + { + "epoch": 0.029944194909486865, + "grad_norm": 2.628105401992798, + "learning_rate": 1.9427613465994277e-05, + "loss": 2.639068031311035, + "step": 11000 + }, + { + "epoch": 0.029998638900231387, + "grad_norm": 2.296719551086426, + "learning_rate": 1.9426523102085323e-05, + "loss": 2.590407943725586, + "step": 11020 + }, + { + "epoch": 0.03005308289097591, + "grad_norm": 2.389650344848633, + "learning_rate": 1.942543273817637e-05, + "loss": 2.6104379653930665, + "step": 11040 + }, + { + "epoch": 0.030107526881720432, + "grad_norm": 2.61991286277771, + "learning_rate": 1.9424342374267415e-05, + "loss": 2.6145986557006835, + "step": 11060 + }, + { + "epoch": 0.03016197087246495, + "grad_norm": 2.2053327560424805, + "learning_rate": 1.9423252010358457e-05, + "loss": 2.6139236450195313, + "step": 11080 + }, + { + "epoch": 0.030216414863209473, + "grad_norm": 2.267158031463623, + "learning_rate": 1.9422161646449503e-05, + "loss": 2.611237716674805, + "step": 11100 + }, + { + "epoch": 0.030270858853953996, + "grad_norm": 2.4401373863220215, + "learning_rate": 1.942107128254055e-05, + "loss": 2.624557876586914, + "step": 11120 + }, + { + "epoch": 0.030325302844698518, + "grad_norm": 2.320401191711426, + "learning_rate": 1.9419980918631594e-05, + "loss": 2.608122634887695, + "step": 11140 + }, + { + "epoch": 0.030379746835443037, + "grad_norm": 2.7719295024871826, + "learning_rate": 1.941889055472264e-05, + "loss": 2.595525550842285, + "step": 11160 + }, + { + "epoch": 0.03043419082618756, + "grad_norm": 2.1794517040252686, + "learning_rate": 1.9417800190813686e-05, + "loss": 2.5990726470947267, + "step": 11180 + }, + { + "epoch": 0.03048863481693208, + "grad_norm": 2.2948973178863525, + "learning_rate": 1.9416709826904732e-05, + "loss": 2.599068832397461, + "step": 11200 + }, + { + "epoch": 0.030543078807676604, + "grad_norm": 2.6655471324920654, + "learning_rate": 1.9415619462995778e-05, + "loss": 2.618877983093262, + "step": 11220 + }, + { + "epoch": 0.030597522798421123, + "grad_norm": 2.281316041946411, + "learning_rate": 1.941452909908682e-05, + "loss": 2.5868139266967773, + "step": 11240 + }, + { + "epoch": 0.030651966789165645, + "grad_norm": 2.3014135360717773, + "learning_rate": 1.9413438735177866e-05, + "loss": 2.5944145202636717, + "step": 11260 + }, + { + "epoch": 0.030706410779910168, + "grad_norm": 2.2994954586029053, + "learning_rate": 1.941234837126891e-05, + "loss": 2.5704828262329102, + "step": 11280 + }, + { + "epoch": 0.03076085477065469, + "grad_norm": 2.217416763305664, + "learning_rate": 1.9411258007359957e-05, + "loss": 2.6048458099365233, + "step": 11300 + }, + { + "epoch": 0.03081529876139921, + "grad_norm": 2.1608245372772217, + "learning_rate": 1.9410167643451003e-05, + "loss": 2.592626953125, + "step": 11320 + }, + { + "epoch": 0.03086974275214373, + "grad_norm": 2.053864002227783, + "learning_rate": 1.940907727954205e-05, + "loss": 2.588144302368164, + "step": 11340 + }, + { + "epoch": 0.030924186742888254, + "grad_norm": 2.2826783657073975, + "learning_rate": 1.9407986915633095e-05, + "loss": 2.6267004013061523, + "step": 11360 + }, + { + "epoch": 0.030978630733632776, + "grad_norm": 2.1896004676818848, + "learning_rate": 1.940689655172414e-05, + "loss": 2.606597328186035, + "step": 11380 + }, + { + "epoch": 0.0310330747243773, + "grad_norm": 2.2856056690216064, + "learning_rate": 1.9405806187815186e-05, + "loss": 2.63358211517334, + "step": 11400 + }, + { + "epoch": 0.031087518715121817, + "grad_norm": 2.5655386447906494, + "learning_rate": 1.940471582390623e-05, + "loss": 2.6092357635498047, + "step": 11420 + }, + { + "epoch": 0.03114196270586634, + "grad_norm": 2.2341880798339844, + "learning_rate": 1.9403625459997275e-05, + "loss": 2.607681083679199, + "step": 11440 + }, + { + "epoch": 0.031196406696610862, + "grad_norm": 2.392866849899292, + "learning_rate": 1.940253509608832e-05, + "loss": 2.601788139343262, + "step": 11460 + }, + { + "epoch": 0.031250850687355385, + "grad_norm": 2.1732280254364014, + "learning_rate": 1.9401444732179366e-05, + "loss": 2.632153129577637, + "step": 11480 + }, + { + "epoch": 0.03130529467809991, + "grad_norm": 2.220710515975952, + "learning_rate": 1.9400354368270412e-05, + "loss": 2.606007957458496, + "step": 11500 + }, + { + "epoch": 0.03135973866884443, + "grad_norm": 3.9981071949005127, + "learning_rate": 1.9399264004361458e-05, + "loss": 2.602276611328125, + "step": 11520 + }, + { + "epoch": 0.031414182659588945, + "grad_norm": 2.1235902309417725, + "learning_rate": 1.9398173640452504e-05, + "loss": 2.5934385299682616, + "step": 11540 + }, + { + "epoch": 0.03146862665033347, + "grad_norm": 2.5331437587738037, + "learning_rate": 1.939708327654355e-05, + "loss": 2.6236658096313477, + "step": 11560 + }, + { + "epoch": 0.03152307064107799, + "grad_norm": 2.343931198120117, + "learning_rate": 1.9395992912634592e-05, + "loss": 2.616116523742676, + "step": 11580 + }, + { + "epoch": 0.03157751463182251, + "grad_norm": 2.2508273124694824, + "learning_rate": 1.9394902548725638e-05, + "loss": 2.5664884567260744, + "step": 11600 + }, + { + "epoch": 0.031631958622567034, + "grad_norm": 2.242973804473877, + "learning_rate": 1.9393812184816684e-05, + "loss": 2.579962730407715, + "step": 11620 + }, + { + "epoch": 0.03168640261331156, + "grad_norm": 2.1931040287017822, + "learning_rate": 1.939272182090773e-05, + "loss": 2.5708749771118162, + "step": 11640 + }, + { + "epoch": 0.03174084660405608, + "grad_norm": 2.504610061645508, + "learning_rate": 1.9391631456998775e-05, + "loss": 2.581405830383301, + "step": 11660 + }, + { + "epoch": 0.0317952905948006, + "grad_norm": 2.3292934894561768, + "learning_rate": 1.939054109308982e-05, + "loss": 2.571659469604492, + "step": 11680 + }, + { + "epoch": 0.031849734585545124, + "grad_norm": 2.410827875137329, + "learning_rate": 1.9389450729180867e-05, + "loss": 2.607521629333496, + "step": 11700 + }, + { + "epoch": 0.03190417857628964, + "grad_norm": 2.5060527324676514, + "learning_rate": 1.9388360365271913e-05, + "loss": 2.596924591064453, + "step": 11720 + }, + { + "epoch": 0.03195862256703416, + "grad_norm": 2.1441824436187744, + "learning_rate": 1.9387270001362955e-05, + "loss": 2.6227542877197267, + "step": 11740 + }, + { + "epoch": 0.032013066557778684, + "grad_norm": 2.177985191345215, + "learning_rate": 1.9386179637454e-05, + "loss": 2.624140167236328, + "step": 11760 + }, + { + "epoch": 0.032067510548523206, + "grad_norm": 2.2635304927825928, + "learning_rate": 1.9385089273545047e-05, + "loss": 2.612191009521484, + "step": 11780 + }, + { + "epoch": 0.03212195453926773, + "grad_norm": 2.263814687728882, + "learning_rate": 1.9383998909636093e-05, + "loss": 2.6050628662109374, + "step": 11800 + }, + { + "epoch": 0.03217639853001225, + "grad_norm": 2.183809757232666, + "learning_rate": 1.938290854572714e-05, + "loss": 2.6046215057373048, + "step": 11820 + }, + { + "epoch": 0.032230842520756774, + "grad_norm": 2.409651517868042, + "learning_rate": 1.9381818181818184e-05, + "loss": 2.5868766784667967, + "step": 11840 + }, + { + "epoch": 0.032285286511501296, + "grad_norm": 2.2106456756591797, + "learning_rate": 1.938072781790923e-05, + "loss": 2.5890888214111327, + "step": 11860 + }, + { + "epoch": 0.03233973050224581, + "grad_norm": 2.1963446140289307, + "learning_rate": 1.9379637454000276e-05, + "loss": 2.5719627380371093, + "step": 11880 + }, + { + "epoch": 0.032394174492990334, + "grad_norm": 2.577387809753418, + "learning_rate": 1.9378547090091318e-05, + "loss": 2.625677299499512, + "step": 11900 + }, + { + "epoch": 0.032448618483734856, + "grad_norm": 2.2950944900512695, + "learning_rate": 1.9377456726182364e-05, + "loss": 2.6060428619384766, + "step": 11920 + }, + { + "epoch": 0.03250306247447938, + "grad_norm": 2.2031869888305664, + "learning_rate": 1.937636636227341e-05, + "loss": 2.629372406005859, + "step": 11940 + }, + { + "epoch": 0.0325575064652239, + "grad_norm": 2.293818473815918, + "learning_rate": 1.9375275998364456e-05, + "loss": 2.602303123474121, + "step": 11960 + }, + { + "epoch": 0.03261195045596842, + "grad_norm": 2.1802315711975098, + "learning_rate": 1.93741856344555e-05, + "loss": 2.589924430847168, + "step": 11980 + }, + { + "epoch": 0.032666394446712946, + "grad_norm": 2.5677030086517334, + "learning_rate": 1.9373095270546547e-05, + "loss": 2.5899173736572267, + "step": 12000 + }, + { + "epoch": 0.03272083843745747, + "grad_norm": 2.3577375411987305, + "learning_rate": 1.9372004906637593e-05, + "loss": 2.6216453552246093, + "step": 12020 + }, + { + "epoch": 0.03277528242820199, + "grad_norm": 2.249376058578491, + "learning_rate": 1.937091454272864e-05, + "loss": 2.598338317871094, + "step": 12040 + }, + { + "epoch": 0.032829726418946506, + "grad_norm": 2.187278985977173, + "learning_rate": 1.936982417881968e-05, + "loss": 2.6195676803588865, + "step": 12060 + }, + { + "epoch": 0.03288417040969103, + "grad_norm": 2.1724421977996826, + "learning_rate": 1.9368733814910727e-05, + "loss": 2.6129997253417967, + "step": 12080 + }, + { + "epoch": 0.03293861440043555, + "grad_norm": 2.7502713203430176, + "learning_rate": 1.9367643451001773e-05, + "loss": 2.607198143005371, + "step": 12100 + }, + { + "epoch": 0.03299305839118007, + "grad_norm": 2.4714481830596924, + "learning_rate": 1.936655308709282e-05, + "loss": 2.6316741943359374, + "step": 12120 + }, + { + "epoch": 0.033047502381924596, + "grad_norm": 2.1740002632141113, + "learning_rate": 1.9365462723183864e-05, + "loss": 2.6207231521606444, + "step": 12140 + }, + { + "epoch": 0.03310194637266912, + "grad_norm": 2.242650032043457, + "learning_rate": 1.936437235927491e-05, + "loss": 2.5994029998779298, + "step": 12160 + }, + { + "epoch": 0.03315639036341364, + "grad_norm": 2.193629026412964, + "learning_rate": 1.9363281995365956e-05, + "loss": 2.5998653411865233, + "step": 12180 + }, + { + "epoch": 0.03321083435415816, + "grad_norm": 2.2540314197540283, + "learning_rate": 1.9362191631457002e-05, + "loss": 2.592030334472656, + "step": 12200 + }, + { + "epoch": 0.03326527834490268, + "grad_norm": 2.517707586288452, + "learning_rate": 1.9361101267548044e-05, + "loss": 2.6234355926513673, + "step": 12220 + }, + { + "epoch": 0.0333197223356472, + "grad_norm": 2.222221851348877, + "learning_rate": 1.936001090363909e-05, + "loss": 2.590910530090332, + "step": 12240 + }, + { + "epoch": 0.03337416632639172, + "grad_norm": 2.398059844970703, + "learning_rate": 1.9358920539730136e-05, + "loss": 2.5929771423339845, + "step": 12260 + }, + { + "epoch": 0.033428610317136245, + "grad_norm": 2.3017189502716064, + "learning_rate": 1.9357830175821182e-05, + "loss": 2.582368278503418, + "step": 12280 + }, + { + "epoch": 0.03348305430788077, + "grad_norm": 2.361055612564087, + "learning_rate": 1.9356739811912228e-05, + "loss": 2.588355827331543, + "step": 12300 + }, + { + "epoch": 0.03353749829862529, + "grad_norm": 2.343076705932617, + "learning_rate": 1.9355649448003273e-05, + "loss": 2.57421932220459, + "step": 12320 + }, + { + "epoch": 0.03359194228936981, + "grad_norm": 2.2804629802703857, + "learning_rate": 1.935455908409432e-05, + "loss": 2.61047477722168, + "step": 12340 + }, + { + "epoch": 0.033646386280114335, + "grad_norm": 2.306739091873169, + "learning_rate": 1.9353468720185365e-05, + "loss": 2.584601402282715, + "step": 12360 + }, + { + "epoch": 0.03370083027085886, + "grad_norm": 2.8783671855926514, + "learning_rate": 1.9352378356276407e-05, + "loss": 2.5898038864135744, + "step": 12380 + }, + { + "epoch": 0.03375527426160337, + "grad_norm": 2.2005093097686768, + "learning_rate": 1.9351287992367453e-05, + "loss": 2.5994693756103517, + "step": 12400 + }, + { + "epoch": 0.033809718252347895, + "grad_norm": 2.445723533630371, + "learning_rate": 1.93501976284585e-05, + "loss": 2.598284912109375, + "step": 12420 + }, + { + "epoch": 0.03386416224309242, + "grad_norm": 2.2496254444122314, + "learning_rate": 1.9349107264549545e-05, + "loss": 2.5996971130371094, + "step": 12440 + }, + { + "epoch": 0.03391860623383694, + "grad_norm": 2.9585306644439697, + "learning_rate": 1.934801690064059e-05, + "loss": 2.5926883697509764, + "step": 12460 + }, + { + "epoch": 0.03397305022458146, + "grad_norm": 2.1332998275756836, + "learning_rate": 1.9346926536731636e-05, + "loss": 2.616838264465332, + "step": 12480 + }, + { + "epoch": 0.034027494215325985, + "grad_norm": 2.2180399894714355, + "learning_rate": 1.9345836172822682e-05, + "loss": 2.6033441543579103, + "step": 12500 + }, + { + "epoch": 0.03408193820607051, + "grad_norm": 2.60276460647583, + "learning_rate": 1.9344745808913728e-05, + "loss": 2.6019607543945313, + "step": 12520 + }, + { + "epoch": 0.03413638219681503, + "grad_norm": 2.1486451625823975, + "learning_rate": 1.934365544500477e-05, + "loss": 2.6030935287475585, + "step": 12540 + }, + { + "epoch": 0.034190826187559545, + "grad_norm": 2.2099010944366455, + "learning_rate": 1.9342565081095816e-05, + "loss": 2.6075748443603515, + "step": 12560 + }, + { + "epoch": 0.03424527017830407, + "grad_norm": 2.352964162826538, + "learning_rate": 1.9341474717186862e-05, + "loss": 2.6170394897460936, + "step": 12580 + }, + { + "epoch": 0.03429971416904859, + "grad_norm": 2.359194755554199, + "learning_rate": 1.9340384353277908e-05, + "loss": 2.648538017272949, + "step": 12600 + }, + { + "epoch": 0.03435415815979311, + "grad_norm": 2.4912943840026855, + "learning_rate": 1.9339293989368954e-05, + "loss": 2.609231948852539, + "step": 12620 + }, + { + "epoch": 0.034408602150537634, + "grad_norm": 2.4984476566314697, + "learning_rate": 1.933820362546e-05, + "loss": 2.6080677032470705, + "step": 12640 + }, + { + "epoch": 0.03446304614128216, + "grad_norm": 2.2989675998687744, + "learning_rate": 1.9337113261551045e-05, + "loss": 2.6360698699951173, + "step": 12660 + }, + { + "epoch": 0.03451749013202668, + "grad_norm": 2.2682995796203613, + "learning_rate": 1.933602289764209e-05, + "loss": 2.5993385314941406, + "step": 12680 + }, + { + "epoch": 0.0345719341227712, + "grad_norm": 2.184706926345825, + "learning_rate": 1.9334932533733134e-05, + "loss": 2.576816940307617, + "step": 12700 + }, + { + "epoch": 0.034626378113515724, + "grad_norm": 2.1859495639801025, + "learning_rate": 1.933384216982418e-05, + "loss": 2.592889404296875, + "step": 12720 + }, + { + "epoch": 0.03468082210426024, + "grad_norm": 2.264186382293701, + "learning_rate": 1.9332751805915225e-05, + "loss": 2.5945066452026366, + "step": 12740 + }, + { + "epoch": 0.03473526609500476, + "grad_norm": 2.2985849380493164, + "learning_rate": 1.933166144200627e-05, + "loss": 2.5900928497314455, + "step": 12760 + }, + { + "epoch": 0.034789710085749284, + "grad_norm": 2.397218942642212, + "learning_rate": 1.9330571078097317e-05, + "loss": 2.626373291015625, + "step": 12780 + }, + { + "epoch": 0.034844154076493807, + "grad_norm": 2.2779276371002197, + "learning_rate": 1.9329480714188363e-05, + "loss": 2.60040397644043, + "step": 12800 + }, + { + "epoch": 0.03489859806723833, + "grad_norm": 2.235010862350464, + "learning_rate": 1.932839035027941e-05, + "loss": 2.593388557434082, + "step": 12820 + }, + { + "epoch": 0.03495304205798285, + "grad_norm": 2.2086596488952637, + "learning_rate": 1.9327299986370454e-05, + "loss": 2.65124568939209, + "step": 12840 + }, + { + "epoch": 0.035007486048727374, + "grad_norm": 2.198188066482544, + "learning_rate": 1.9326209622461497e-05, + "loss": 2.6288515090942384, + "step": 12860 + }, + { + "epoch": 0.035061930039471896, + "grad_norm": 2.3627779483795166, + "learning_rate": 1.9325119258552542e-05, + "loss": 2.645145034790039, + "step": 12880 + }, + { + "epoch": 0.03511637403021641, + "grad_norm": 2.2160181999206543, + "learning_rate": 1.932402889464359e-05, + "loss": 2.5714384078979493, + "step": 12900 + }, + { + "epoch": 0.035170818020960934, + "grad_norm": 2.82326602935791, + "learning_rate": 1.9322938530734634e-05, + "loss": 2.5854894638061525, + "step": 12920 + }, + { + "epoch": 0.035225262011705456, + "grad_norm": 2.219057083129883, + "learning_rate": 1.932184816682568e-05, + "loss": 2.559368896484375, + "step": 12940 + }, + { + "epoch": 0.03527970600244998, + "grad_norm": 2.196040153503418, + "learning_rate": 1.9320757802916726e-05, + "loss": 2.598044776916504, + "step": 12960 + }, + { + "epoch": 0.0353341499931945, + "grad_norm": 2.317958354949951, + "learning_rate": 1.931966743900777e-05, + "loss": 2.60604248046875, + "step": 12980 + }, + { + "epoch": 0.03538859398393902, + "grad_norm": 2.256133794784546, + "learning_rate": 1.9318577075098817e-05, + "loss": 2.603437805175781, + "step": 13000 + }, + { + "epoch": 0.035443037974683546, + "grad_norm": 2.1859774589538574, + "learning_rate": 1.931748671118986e-05, + "loss": 2.6057735443115235, + "step": 13020 + }, + { + "epoch": 0.03549748196542807, + "grad_norm": 2.1746342182159424, + "learning_rate": 1.9316396347280906e-05, + "loss": 2.5744094848632812, + "step": 13040 + }, + { + "epoch": 0.03555192595617259, + "grad_norm": 2.2730014324188232, + "learning_rate": 1.931530598337195e-05, + "loss": 2.5953630447387694, + "step": 13060 + }, + { + "epoch": 0.035606369946917106, + "grad_norm": 2.2714266777038574, + "learning_rate": 1.9314215619462997e-05, + "loss": 2.630690002441406, + "step": 13080 + }, + { + "epoch": 0.03566081393766163, + "grad_norm": 2.212212562561035, + "learning_rate": 1.9313125255554043e-05, + "loss": 2.587653160095215, + "step": 13100 + }, + { + "epoch": 0.03571525792840615, + "grad_norm": 2.121459722518921, + "learning_rate": 1.931203489164509e-05, + "loss": 2.618377113342285, + "step": 13120 + }, + { + "epoch": 0.03576970191915067, + "grad_norm": 2.1144461631774902, + "learning_rate": 1.931094452773613e-05, + "loss": 2.5917951583862306, + "step": 13140 + }, + { + "epoch": 0.035824145909895196, + "grad_norm": 2.3325300216674805, + "learning_rate": 1.930985416382718e-05, + "loss": 2.5947765350341796, + "step": 13160 + }, + { + "epoch": 0.03587858990063972, + "grad_norm": 2.328878879547119, + "learning_rate": 1.9308763799918223e-05, + "loss": 2.5974668502807616, + "step": 13180 + }, + { + "epoch": 0.03593303389138424, + "grad_norm": 2.4149680137634277, + "learning_rate": 1.930767343600927e-05, + "loss": 2.6297622680664063, + "step": 13200 + }, + { + "epoch": 0.03598747788212876, + "grad_norm": 2.252086639404297, + "learning_rate": 1.9306583072100314e-05, + "loss": 2.6062686920166014, + "step": 13220 + }, + { + "epoch": 0.03604192187287328, + "grad_norm": 2.1081252098083496, + "learning_rate": 1.930549270819136e-05, + "loss": 2.568406105041504, + "step": 13240 + }, + { + "epoch": 0.0360963658636178, + "grad_norm": 2.3811819553375244, + "learning_rate": 1.9304402344282406e-05, + "loss": 2.620005989074707, + "step": 13260 + }, + { + "epoch": 0.03615080985436232, + "grad_norm": 2.3391976356506348, + "learning_rate": 1.9303311980373452e-05, + "loss": 2.600623893737793, + "step": 13280 + }, + { + "epoch": 0.036205253845106845, + "grad_norm": 2.2208878993988037, + "learning_rate": 1.9302221616464494e-05, + "loss": 2.626380729675293, + "step": 13300 + }, + { + "epoch": 0.03625969783585137, + "grad_norm": 2.2022109031677246, + "learning_rate": 1.9301131252555544e-05, + "loss": 2.5788021087646484, + "step": 13320 + }, + { + "epoch": 0.03631414182659589, + "grad_norm": 2.3260347843170166, + "learning_rate": 1.9300040888646586e-05, + "loss": 2.5829326629638674, + "step": 13340 + }, + { + "epoch": 0.03636858581734041, + "grad_norm": 4.056549549102783, + "learning_rate": 1.9298950524737632e-05, + "loss": 2.560036468505859, + "step": 13360 + }, + { + "epoch": 0.036423029808084935, + "grad_norm": 2.177025318145752, + "learning_rate": 1.9297860160828678e-05, + "loss": 2.588520050048828, + "step": 13380 + }, + { + "epoch": 0.03647747379882946, + "grad_norm": 2.192659616470337, + "learning_rate": 1.9296769796919723e-05, + "loss": 2.594629669189453, + "step": 13400 + }, + { + "epoch": 0.03653191778957397, + "grad_norm": 2.181628704071045, + "learning_rate": 1.929567943301077e-05, + "loss": 2.6255304336547853, + "step": 13420 + }, + { + "epoch": 0.036586361780318495, + "grad_norm": 2.19905424118042, + "learning_rate": 1.9294589069101815e-05, + "loss": 2.591668128967285, + "step": 13440 + }, + { + "epoch": 0.03664080577106302, + "grad_norm": 2.1873512268066406, + "learning_rate": 1.9293498705192857e-05, + "loss": 2.57061653137207, + "step": 13460 + }, + { + "epoch": 0.03669524976180754, + "grad_norm": 2.1762216091156006, + "learning_rate": 1.9292408341283907e-05, + "loss": 2.6127908706665037, + "step": 13480 + }, + { + "epoch": 0.03674969375255206, + "grad_norm": 2.2074742317199707, + "learning_rate": 1.929131797737495e-05, + "loss": 2.5829641342163088, + "step": 13500 + }, + { + "epoch": 0.036804137743296585, + "grad_norm": 2.1720542907714844, + "learning_rate": 1.9290227613465995e-05, + "loss": 2.616826629638672, + "step": 13520 + }, + { + "epoch": 0.03685858173404111, + "grad_norm": 2.129812002182007, + "learning_rate": 1.928913724955704e-05, + "loss": 2.6010320663452147, + "step": 13540 + }, + { + "epoch": 0.03691302572478563, + "grad_norm": 2.218517541885376, + "learning_rate": 1.9288046885648086e-05, + "loss": 2.6093687057495116, + "step": 13560 + }, + { + "epoch": 0.03696746971553015, + "grad_norm": 2.2423512935638428, + "learning_rate": 1.9286956521739132e-05, + "loss": 2.610659599304199, + "step": 13580 + }, + { + "epoch": 0.03702191370627467, + "grad_norm": 2.3195173740386963, + "learning_rate": 1.9285866157830178e-05, + "loss": 2.6014484405517577, + "step": 13600 + }, + { + "epoch": 0.03707635769701919, + "grad_norm": 2.303030252456665, + "learning_rate": 1.928477579392122e-05, + "loss": 2.611110877990723, + "step": 13620 + }, + { + "epoch": 0.03713080168776371, + "grad_norm": 2.1488394737243652, + "learning_rate": 1.928368543001227e-05, + "loss": 2.5966913223266603, + "step": 13640 + }, + { + "epoch": 0.037185245678508234, + "grad_norm": 2.2222042083740234, + "learning_rate": 1.9282595066103312e-05, + "loss": 2.621501922607422, + "step": 13660 + }, + { + "epoch": 0.03723968966925276, + "grad_norm": 2.2924907207489014, + "learning_rate": 1.9281504702194358e-05, + "loss": 2.5965131759643554, + "step": 13680 + }, + { + "epoch": 0.03729413365999728, + "grad_norm": 2.2723727226257324, + "learning_rate": 1.9280414338285404e-05, + "loss": 2.5976381301879883, + "step": 13700 + }, + { + "epoch": 0.0373485776507418, + "grad_norm": 2.1680164337158203, + "learning_rate": 1.927932397437645e-05, + "loss": 2.589723014831543, + "step": 13720 + }, + { + "epoch": 0.037403021641486324, + "grad_norm": 2.287033796310425, + "learning_rate": 1.9278233610467495e-05, + "loss": 2.579987144470215, + "step": 13740 + }, + { + "epoch": 0.03745746563223084, + "grad_norm": 2.2610325813293457, + "learning_rate": 1.927714324655854e-05, + "loss": 2.5845890045166016, + "step": 13760 + }, + { + "epoch": 0.03751190962297536, + "grad_norm": 2.208989381790161, + "learning_rate": 1.9276052882649584e-05, + "loss": 2.58333683013916, + "step": 13780 + }, + { + "epoch": 0.037566353613719884, + "grad_norm": 2.2655158042907715, + "learning_rate": 1.9274962518740633e-05, + "loss": 2.573101615905762, + "step": 13800 + }, + { + "epoch": 0.03762079760446441, + "grad_norm": 2.3803179264068604, + "learning_rate": 1.9273872154831675e-05, + "loss": 2.5859563827514647, + "step": 13820 + }, + { + "epoch": 0.03767524159520893, + "grad_norm": 2.3684701919555664, + "learning_rate": 1.927278179092272e-05, + "loss": 2.5960821151733398, + "step": 13840 + }, + { + "epoch": 0.03772968558595345, + "grad_norm": 2.305633783340454, + "learning_rate": 1.9271691427013767e-05, + "loss": 2.580807113647461, + "step": 13860 + }, + { + "epoch": 0.037784129576697974, + "grad_norm": 2.201977491378784, + "learning_rate": 1.9270601063104813e-05, + "loss": 2.6291948318481446, + "step": 13880 + }, + { + "epoch": 0.037838573567442496, + "grad_norm": 2.181976556777954, + "learning_rate": 1.926951069919586e-05, + "loss": 2.609023857116699, + "step": 13900 + }, + { + "epoch": 0.03789301755818702, + "grad_norm": 2.1703805923461914, + "learning_rate": 1.9268420335286904e-05, + "loss": 2.599422073364258, + "step": 13920 + }, + { + "epoch": 0.037947461548931534, + "grad_norm": 2.3908298015594482, + "learning_rate": 1.9267329971377947e-05, + "loss": 2.600595474243164, + "step": 13940 + }, + { + "epoch": 0.038001905539676056, + "grad_norm": 2.1507105827331543, + "learning_rate": 1.9266239607468996e-05, + "loss": 2.5894647598266602, + "step": 13960 + }, + { + "epoch": 0.03805634953042058, + "grad_norm": 2.462615489959717, + "learning_rate": 1.9265149243560038e-05, + "loss": 2.585793685913086, + "step": 13980 + }, + { + "epoch": 0.0381107935211651, + "grad_norm": 2.108099937438965, + "learning_rate": 1.9264058879651084e-05, + "loss": 2.5952630996704102, + "step": 14000 + }, + { + "epoch": 0.038165237511909623, + "grad_norm": 2.2694270610809326, + "learning_rate": 1.926296851574213e-05, + "loss": 2.618027687072754, + "step": 14020 + }, + { + "epoch": 0.038219681502654146, + "grad_norm": 2.293846607208252, + "learning_rate": 1.9261878151833176e-05, + "loss": 2.6340560913085938, + "step": 14040 + }, + { + "epoch": 0.03827412549339867, + "grad_norm": 2.3599603176116943, + "learning_rate": 1.926078778792422e-05, + "loss": 2.5795034408569335, + "step": 14060 + }, + { + "epoch": 0.03832856948414319, + "grad_norm": 2.3040692806243896, + "learning_rate": 1.9259697424015267e-05, + "loss": 2.5976686477661133, + "step": 14080 + }, + { + "epoch": 0.038383013474887706, + "grad_norm": 2.2360141277313232, + "learning_rate": 1.925860706010631e-05, + "loss": 2.6191888809204102, + "step": 14100 + }, + { + "epoch": 0.03843745746563223, + "grad_norm": 2.2985334396362305, + "learning_rate": 1.925751669619736e-05, + "loss": 2.5853851318359373, + "step": 14120 + }, + { + "epoch": 0.03849190145637675, + "grad_norm": 2.103585720062256, + "learning_rate": 1.92564263322884e-05, + "loss": 2.601578712463379, + "step": 14140 + }, + { + "epoch": 0.03854634544712127, + "grad_norm": 2.07915997505188, + "learning_rate": 1.9255335968379447e-05, + "loss": 2.5776769638061525, + "step": 14160 + }, + { + "epoch": 0.038600789437865796, + "grad_norm": 2.2168631553649902, + "learning_rate": 1.9254245604470493e-05, + "loss": 2.5955478668212892, + "step": 14180 + }, + { + "epoch": 0.03865523342861032, + "grad_norm": 2.1899657249450684, + "learning_rate": 1.925315524056154e-05, + "loss": 2.608894157409668, + "step": 14200 + }, + { + "epoch": 0.03870967741935484, + "grad_norm": 2.25978684425354, + "learning_rate": 1.9252064876652585e-05, + "loss": 2.58218936920166, + "step": 14220 + }, + { + "epoch": 0.03876412141009936, + "grad_norm": 2.2208733558654785, + "learning_rate": 1.925097451274363e-05, + "loss": 2.6084878921508787, + "step": 14240 + }, + { + "epoch": 0.038818565400843885, + "grad_norm": 2.1566355228424072, + "learning_rate": 1.9249884148834673e-05, + "loss": 2.5889970779418947, + "step": 14260 + }, + { + "epoch": 0.0388730093915884, + "grad_norm": 2.419577121734619, + "learning_rate": 1.9248793784925722e-05, + "loss": 2.5611400604248047, + "step": 14280 + }, + { + "epoch": 0.03892745338233292, + "grad_norm": 2.202677011489868, + "learning_rate": 1.9247703421016764e-05, + "loss": 2.574220657348633, + "step": 14300 + }, + { + "epoch": 0.038981897373077445, + "grad_norm": 3.5122132301330566, + "learning_rate": 1.924661305710781e-05, + "loss": 2.5897476196289064, + "step": 14320 + }, + { + "epoch": 0.03903634136382197, + "grad_norm": 2.3036599159240723, + "learning_rate": 1.9245522693198856e-05, + "loss": 2.5751338958740235, + "step": 14340 + }, + { + "epoch": 0.03909078535456649, + "grad_norm": 2.1909356117248535, + "learning_rate": 1.9244432329289902e-05, + "loss": 2.581009101867676, + "step": 14360 + }, + { + "epoch": 0.03914522934531101, + "grad_norm": 2.1909492015838623, + "learning_rate": 1.9243341965380948e-05, + "loss": 2.6093374252319337, + "step": 14380 + }, + { + "epoch": 0.039199673336055535, + "grad_norm": 2.2551167011260986, + "learning_rate": 1.9242251601471994e-05, + "loss": 2.575397491455078, + "step": 14400 + }, + { + "epoch": 0.03925411732680006, + "grad_norm": 2.1488301753997803, + "learning_rate": 1.9241161237563036e-05, + "loss": 2.5775897979736326, + "step": 14420 + }, + { + "epoch": 0.03930856131754457, + "grad_norm": 2.1247522830963135, + "learning_rate": 1.9240070873654085e-05, + "loss": 2.6067073822021483, + "step": 14440 + }, + { + "epoch": 0.039363005308289095, + "grad_norm": 2.356128454208374, + "learning_rate": 1.9238980509745128e-05, + "loss": 2.5828369140625, + "step": 14460 + }, + { + "epoch": 0.03941744929903362, + "grad_norm": 2.3274269104003906, + "learning_rate": 1.9237890145836173e-05, + "loss": 2.5775882720947267, + "step": 14480 + }, + { + "epoch": 0.03947189328977814, + "grad_norm": 2.5668179988861084, + "learning_rate": 1.923679978192722e-05, + "loss": 2.596177673339844, + "step": 14500 + }, + { + "epoch": 0.03952633728052266, + "grad_norm": 2.1803627014160156, + "learning_rate": 1.9235709418018265e-05, + "loss": 2.5785634994506834, + "step": 14520 + }, + { + "epoch": 0.039580781271267185, + "grad_norm": 2.3231213092803955, + "learning_rate": 1.923461905410931e-05, + "loss": 2.5728559494018555, + "step": 14540 + }, + { + "epoch": 0.03963522526201171, + "grad_norm": 2.3006341457366943, + "learning_rate": 1.9233528690200357e-05, + "loss": 2.59954776763916, + "step": 14560 + }, + { + "epoch": 0.03968966925275623, + "grad_norm": 2.1050302982330322, + "learning_rate": 1.92324383262914e-05, + "loss": 2.5905536651611327, + "step": 14580 + }, + { + "epoch": 0.03974411324350075, + "grad_norm": 2.1884829998016357, + "learning_rate": 1.9231347962382448e-05, + "loss": 2.5778059005737304, + "step": 14600 + }, + { + "epoch": 0.03979855723424527, + "grad_norm": 2.243831157684326, + "learning_rate": 1.923025759847349e-05, + "loss": 2.5931215286254883, + "step": 14620 + }, + { + "epoch": 0.03985300122498979, + "grad_norm": 2.172401189804077, + "learning_rate": 1.9229167234564536e-05, + "loss": 2.6015094757080077, + "step": 14640 + }, + { + "epoch": 0.03990744521573431, + "grad_norm": 2.2006192207336426, + "learning_rate": 1.9228076870655582e-05, + "loss": 2.615550422668457, + "step": 14660 + }, + { + "epoch": 0.039961889206478834, + "grad_norm": 2.9308266639709473, + "learning_rate": 1.9226986506746628e-05, + "loss": 2.576771545410156, + "step": 14680 + }, + { + "epoch": 0.04001633319722336, + "grad_norm": 2.193436861038208, + "learning_rate": 1.9225896142837674e-05, + "loss": 2.586319923400879, + "step": 14700 + }, + { + "epoch": 0.04007077718796788, + "grad_norm": 2.233414888381958, + "learning_rate": 1.922480577892872e-05, + "loss": 2.605754089355469, + "step": 14720 + }, + { + "epoch": 0.0401252211787124, + "grad_norm": 2.1257925033569336, + "learning_rate": 1.9223715415019762e-05, + "loss": 2.60345401763916, + "step": 14740 + }, + { + "epoch": 0.040179665169456924, + "grad_norm": 2.4096837043762207, + "learning_rate": 1.922262505111081e-05, + "loss": 2.612732696533203, + "step": 14760 + }, + { + "epoch": 0.04023410916020144, + "grad_norm": 2.2624075412750244, + "learning_rate": 1.9221534687201854e-05, + "loss": 2.592344284057617, + "step": 14780 + }, + { + "epoch": 0.04028855315094596, + "grad_norm": 2.4401369094848633, + "learning_rate": 1.9220444323292903e-05, + "loss": 2.58227424621582, + "step": 14800 + }, + { + "epoch": 0.040342997141690484, + "grad_norm": 2.637009859085083, + "learning_rate": 1.9219353959383945e-05, + "loss": 2.590806579589844, + "step": 14820 + }, + { + "epoch": 0.04039744113243501, + "grad_norm": 2.4115583896636963, + "learning_rate": 1.921826359547499e-05, + "loss": 2.5887250900268555, + "step": 14840 + }, + { + "epoch": 0.04045188512317953, + "grad_norm": 2.181079864501953, + "learning_rate": 1.9217173231566037e-05, + "loss": 2.570259666442871, + "step": 14860 + }, + { + "epoch": 0.04050632911392405, + "grad_norm": 2.1717402935028076, + "learning_rate": 1.9216082867657083e-05, + "loss": 2.602570152282715, + "step": 14880 + }, + { + "epoch": 0.040560773104668574, + "grad_norm": 2.1069936752319336, + "learning_rate": 1.9214992503748125e-05, + "loss": 2.582205390930176, + "step": 14900 + }, + { + "epoch": 0.040615217095413096, + "grad_norm": 2.100752115249634, + "learning_rate": 1.9213902139839174e-05, + "loss": 2.5542598724365235, + "step": 14920 + }, + { + "epoch": 0.04066966108615762, + "grad_norm": 11.856060981750488, + "learning_rate": 1.9212811775930217e-05, + "loss": 2.563450813293457, + "step": 14940 + }, + { + "epoch": 0.040724105076902134, + "grad_norm": 2.1760711669921875, + "learning_rate": 1.9211721412021266e-05, + "loss": 2.5964324951171873, + "step": 14960 + }, + { + "epoch": 0.040778549067646656, + "grad_norm": 2.3564441204071045, + "learning_rate": 1.921063104811231e-05, + "loss": 2.604144477844238, + "step": 14980 + }, + { + "epoch": 0.04083299305839118, + "grad_norm": 2.1759567260742188, + "learning_rate": 1.9209540684203354e-05, + "loss": 2.5887882232666017, + "step": 15000 + }, + { + "epoch": 0.0408874370491357, + "grad_norm": 2.208754301071167, + "learning_rate": 1.92084503202944e-05, + "loss": 2.5869232177734376, + "step": 15020 + }, + { + "epoch": 0.040941881039880224, + "grad_norm": 2.314180850982666, + "learning_rate": 1.9207359956385446e-05, + "loss": 2.593659782409668, + "step": 15040 + }, + { + "epoch": 0.040996325030624746, + "grad_norm": 3.4392638206481934, + "learning_rate": 1.9206269592476488e-05, + "loss": 2.610184097290039, + "step": 15060 + }, + { + "epoch": 0.04105076902136927, + "grad_norm": 2.1202101707458496, + "learning_rate": 1.9205179228567537e-05, + "loss": 2.5847291946411133, + "step": 15080 + }, + { + "epoch": 0.04110521301211379, + "grad_norm": 2.2497682571411133, + "learning_rate": 1.920408886465858e-05, + "loss": 2.5862173080444335, + "step": 15100 + }, + { + "epoch": 0.041159657002858306, + "grad_norm": 2.3930227756500244, + "learning_rate": 1.920299850074963e-05, + "loss": 2.5774843215942385, + "step": 15120 + }, + { + "epoch": 0.04121410099360283, + "grad_norm": 3.3780016899108887, + "learning_rate": 1.920190813684067e-05, + "loss": 2.584567642211914, + "step": 15140 + }, + { + "epoch": 0.04126854498434735, + "grad_norm": 2.3589959144592285, + "learning_rate": 1.9200817772931717e-05, + "loss": 2.5827873229980467, + "step": 15160 + }, + { + "epoch": 0.04132298897509187, + "grad_norm": 2.3598079681396484, + "learning_rate": 1.9199727409022763e-05, + "loss": 2.5959434509277344, + "step": 15180 + }, + { + "epoch": 0.041377432965836396, + "grad_norm": 2.2376341819763184, + "learning_rate": 1.919863704511381e-05, + "loss": 2.5629131317138674, + "step": 15200 + }, + { + "epoch": 0.04143187695658092, + "grad_norm": 2.1692638397216797, + "learning_rate": 1.919754668120485e-05, + "loss": 2.6320953369140625, + "step": 15220 + }, + { + "epoch": 0.04148632094732544, + "grad_norm": 2.169015645980835, + "learning_rate": 1.91964563172959e-05, + "loss": 2.6031469345092773, + "step": 15240 + }, + { + "epoch": 0.04154076493806996, + "grad_norm": 2.16092848777771, + "learning_rate": 1.9195365953386943e-05, + "loss": 2.57739143371582, + "step": 15260 + }, + { + "epoch": 0.041595208928814485, + "grad_norm": 2.2627830505371094, + "learning_rate": 1.9194275589477992e-05, + "loss": 2.6042604446411133, + "step": 15280 + }, + { + "epoch": 0.041649652919559, + "grad_norm": 2.1327598094940186, + "learning_rate": 1.9193185225569035e-05, + "loss": 2.6141605377197266, + "step": 15300 + }, + { + "epoch": 0.04170409691030352, + "grad_norm": 2.2010769844055176, + "learning_rate": 1.919209486166008e-05, + "loss": 2.577241325378418, + "step": 15320 + }, + { + "epoch": 0.041758540901048045, + "grad_norm": 2.257242202758789, + "learning_rate": 1.9191004497751126e-05, + "loss": 2.5666572570800783, + "step": 15340 + }, + { + "epoch": 0.04181298489179257, + "grad_norm": 2.360264778137207, + "learning_rate": 1.9189914133842172e-05, + "loss": 2.5867252349853516, + "step": 15360 + }, + { + "epoch": 0.04186742888253709, + "grad_norm": 2.2217345237731934, + "learning_rate": 1.9188823769933214e-05, + "loss": 2.5694965362548827, + "step": 15380 + }, + { + "epoch": 0.04192187287328161, + "grad_norm": 2.2553601264953613, + "learning_rate": 1.9187733406024264e-05, + "loss": 2.577853202819824, + "step": 15400 + }, + { + "epoch": 0.041976316864026135, + "grad_norm": 2.152657985687256, + "learning_rate": 1.9186643042115306e-05, + "loss": 2.6225955963134764, + "step": 15420 + }, + { + "epoch": 0.04203076085477066, + "grad_norm": 2.4909048080444336, + "learning_rate": 1.9185552678206355e-05, + "loss": 2.581648826599121, + "step": 15440 + }, + { + "epoch": 0.04208520484551517, + "grad_norm": 2.41308856010437, + "learning_rate": 1.9184462314297398e-05, + "loss": 2.597324180603027, + "step": 15460 + }, + { + "epoch": 0.042139648836259695, + "grad_norm": 2.2940421104431152, + "learning_rate": 1.9183371950388443e-05, + "loss": 2.567310905456543, + "step": 15480 + }, + { + "epoch": 0.04219409282700422, + "grad_norm": 2.124063730239868, + "learning_rate": 1.918228158647949e-05, + "loss": 2.584127426147461, + "step": 15500 + }, + { + "epoch": 0.04224853681774874, + "grad_norm": 2.3092808723449707, + "learning_rate": 1.9181191222570535e-05, + "loss": 2.5795854568481444, + "step": 15520 + }, + { + "epoch": 0.04230298080849326, + "grad_norm": 2.3351502418518066, + "learning_rate": 1.9180100858661578e-05, + "loss": 2.5939105987548827, + "step": 15540 + }, + { + "epoch": 0.042357424799237785, + "grad_norm": 2.2485125064849854, + "learning_rate": 1.9179010494752627e-05, + "loss": 2.5832170486450194, + "step": 15560 + }, + { + "epoch": 0.04241186878998231, + "grad_norm": 2.3726141452789307, + "learning_rate": 1.917792013084367e-05, + "loss": 2.5979684829711913, + "step": 15580 + }, + { + "epoch": 0.04246631278072683, + "grad_norm": 2.349501132965088, + "learning_rate": 1.917682976693472e-05, + "loss": 2.59901123046875, + "step": 15600 + }, + { + "epoch": 0.04252075677147135, + "grad_norm": 2.2040116786956787, + "learning_rate": 1.917573940302576e-05, + "loss": 2.5747749328613283, + "step": 15620 + }, + { + "epoch": 0.04257520076221587, + "grad_norm": 2.152039051055908, + "learning_rate": 1.9174649039116807e-05, + "loss": 2.5851364135742188, + "step": 15640 + }, + { + "epoch": 0.04262964475296039, + "grad_norm": 2.323735237121582, + "learning_rate": 1.9173558675207852e-05, + "loss": 2.6076385498046877, + "step": 15660 + }, + { + "epoch": 0.04268408874370491, + "grad_norm": 2.1378283500671387, + "learning_rate": 1.9172468311298898e-05, + "loss": 2.6037403106689454, + "step": 15680 + }, + { + "epoch": 0.042738532734449435, + "grad_norm": 2.1095237731933594, + "learning_rate": 1.917137794738994e-05, + "loss": 2.564078521728516, + "step": 15700 + }, + { + "epoch": 0.04279297672519396, + "grad_norm": 2.2157363891601562, + "learning_rate": 1.917028758348099e-05, + "loss": 2.578940010070801, + "step": 15720 + }, + { + "epoch": 0.04284742071593848, + "grad_norm": 2.308013916015625, + "learning_rate": 1.9169197219572032e-05, + "loss": 2.6262563705444335, + "step": 15740 + }, + { + "epoch": 0.042901864706683, + "grad_norm": 2.266671895980835, + "learning_rate": 1.916810685566308e-05, + "loss": 2.5845867156982423, + "step": 15760 + }, + { + "epoch": 0.042956308697427524, + "grad_norm": 2.1819963455200195, + "learning_rate": 1.9167016491754124e-05, + "loss": 2.594644546508789, + "step": 15780 + }, + { + "epoch": 0.043010752688172046, + "grad_norm": 2.141141891479492, + "learning_rate": 1.916592612784517e-05, + "loss": 2.587771987915039, + "step": 15800 + }, + { + "epoch": 0.04306519667891656, + "grad_norm": 2.341322660446167, + "learning_rate": 1.9164835763936215e-05, + "loss": 2.5882577896118164, + "step": 15820 + }, + { + "epoch": 0.043119640669661084, + "grad_norm": 2.320936441421509, + "learning_rate": 1.916374540002726e-05, + "loss": 2.6146726608276367, + "step": 15840 + }, + { + "epoch": 0.04317408466040561, + "grad_norm": 2.300889492034912, + "learning_rate": 1.9162655036118304e-05, + "loss": 2.5583425521850587, + "step": 15860 + }, + { + "epoch": 0.04322852865115013, + "grad_norm": 2.3180336952209473, + "learning_rate": 1.9161564672209353e-05, + "loss": 2.6048442840576174, + "step": 15880 + }, + { + "epoch": 0.04328297264189465, + "grad_norm": 2.6630349159240723, + "learning_rate": 1.9160474308300395e-05, + "loss": 2.558644676208496, + "step": 15900 + }, + { + "epoch": 0.043337416632639174, + "grad_norm": 2.282193660736084, + "learning_rate": 1.9159383944391445e-05, + "loss": 2.5489051818847654, + "step": 15920 + }, + { + "epoch": 0.043391860623383696, + "grad_norm": 2.423415422439575, + "learning_rate": 1.9158293580482487e-05, + "loss": 2.613843536376953, + "step": 15940 + }, + { + "epoch": 0.04344630461412822, + "grad_norm": 2.166344404220581, + "learning_rate": 1.9157203216573533e-05, + "loss": 2.589649963378906, + "step": 15960 + }, + { + "epoch": 0.043500748604872734, + "grad_norm": 2.222712278366089, + "learning_rate": 1.915611285266458e-05, + "loss": 2.5785234451293944, + "step": 15980 + }, + { + "epoch": 0.043555192595617256, + "grad_norm": 2.238750457763672, + "learning_rate": 1.9155022488755624e-05, + "loss": 2.611109161376953, + "step": 16000 + }, + { + "epoch": 0.04360963658636178, + "grad_norm": 2.1814517974853516, + "learning_rate": 1.9153932124846667e-05, + "loss": 2.584008979797363, + "step": 16020 + }, + { + "epoch": 0.0436640805771063, + "grad_norm": 2.272528886795044, + "learning_rate": 1.9152841760937716e-05, + "loss": 2.584987258911133, + "step": 16040 + }, + { + "epoch": 0.043718524567850824, + "grad_norm": 2.220470666885376, + "learning_rate": 1.915175139702876e-05, + "loss": 2.570693778991699, + "step": 16060 + }, + { + "epoch": 0.043772968558595346, + "grad_norm": 2.5187718868255615, + "learning_rate": 1.9150661033119808e-05, + "loss": 2.539967727661133, + "step": 16080 + }, + { + "epoch": 0.04382741254933987, + "grad_norm": 2.485111713409424, + "learning_rate": 1.914957066921085e-05, + "loss": 2.5969110488891602, + "step": 16100 + }, + { + "epoch": 0.04388185654008439, + "grad_norm": 2.29436993598938, + "learning_rate": 1.9148480305301896e-05, + "loss": 2.575216865539551, + "step": 16120 + }, + { + "epoch": 0.04393630053082891, + "grad_norm": 2.2628173828125, + "learning_rate": 1.914738994139294e-05, + "loss": 2.5838966369628906, + "step": 16140 + }, + { + "epoch": 0.04399074452157343, + "grad_norm": 2.9791014194488525, + "learning_rate": 1.9146299577483987e-05, + "loss": 2.572473335266113, + "step": 16160 + }, + { + "epoch": 0.04404518851231795, + "grad_norm": 2.173607587814331, + "learning_rate": 1.914520921357503e-05, + "loss": 2.580608367919922, + "step": 16180 + }, + { + "epoch": 0.04409963250306247, + "grad_norm": 2.12859845161438, + "learning_rate": 1.914411884966608e-05, + "loss": 2.5840484619140627, + "step": 16200 + }, + { + "epoch": 0.044154076493806996, + "grad_norm": 2.3020830154418945, + "learning_rate": 1.914302848575712e-05, + "loss": 2.572193908691406, + "step": 16220 + }, + { + "epoch": 0.04420852048455152, + "grad_norm": 2.2712297439575195, + "learning_rate": 1.914193812184817e-05, + "loss": 2.5721261978149412, + "step": 16240 + }, + { + "epoch": 0.04426296447529604, + "grad_norm": 2.1813642978668213, + "learning_rate": 1.9140847757939213e-05, + "loss": 2.58606014251709, + "step": 16260 + }, + { + "epoch": 0.04431740846604056, + "grad_norm": 2.2359092235565186, + "learning_rate": 1.913975739403026e-05, + "loss": 2.5781185150146486, + "step": 16280 + }, + { + "epoch": 0.044371852456785085, + "grad_norm": 2.285133123397827, + "learning_rate": 1.9138667030121305e-05, + "loss": 2.585922431945801, + "step": 16300 + }, + { + "epoch": 0.0444262964475296, + "grad_norm": 2.1099045276641846, + "learning_rate": 1.913757666621235e-05, + "loss": 2.5899188995361326, + "step": 16320 + }, + { + "epoch": 0.04448074043827412, + "grad_norm": 2.2325100898742676, + "learning_rate": 1.9136486302303393e-05, + "loss": 2.5832378387451174, + "step": 16340 + }, + { + "epoch": 0.044535184429018645, + "grad_norm": 2.2121284008026123, + "learning_rate": 1.9135395938394442e-05, + "loss": 2.559583282470703, + "step": 16360 + }, + { + "epoch": 0.04458962841976317, + "grad_norm": 2.1440107822418213, + "learning_rate": 1.9134305574485485e-05, + "loss": 2.5942110061645507, + "step": 16380 + }, + { + "epoch": 0.04464407241050769, + "grad_norm": 2.156012535095215, + "learning_rate": 1.9133215210576534e-05, + "loss": 2.5645328521728517, + "step": 16400 + }, + { + "epoch": 0.04469851640125221, + "grad_norm": 2.259136438369751, + "learning_rate": 1.9132124846667576e-05, + "loss": 2.588213348388672, + "step": 16420 + }, + { + "epoch": 0.044752960391996735, + "grad_norm": 2.223843574523926, + "learning_rate": 1.9131034482758622e-05, + "loss": 2.589293098449707, + "step": 16440 + }, + { + "epoch": 0.04480740438274126, + "grad_norm": 2.088223695755005, + "learning_rate": 1.9129944118849668e-05, + "loss": 2.588265228271484, + "step": 16460 + }, + { + "epoch": 0.04486184837348578, + "grad_norm": 2.1706345081329346, + "learning_rate": 1.9128853754940714e-05, + "loss": 2.5550104141235352, + "step": 16480 + }, + { + "epoch": 0.044916292364230295, + "grad_norm": 2.2188472747802734, + "learning_rate": 1.912776339103176e-05, + "loss": 2.5774126052856445, + "step": 16500 + }, + { + "epoch": 0.04497073635497482, + "grad_norm": 2.2660088539123535, + "learning_rate": 1.9126673027122805e-05, + "loss": 2.601150894165039, + "step": 16520 + }, + { + "epoch": 0.04502518034571934, + "grad_norm": 2.1214518547058105, + "learning_rate": 1.9125582663213848e-05, + "loss": 2.5900264739990235, + "step": 16540 + }, + { + "epoch": 0.04507962433646386, + "grad_norm": 2.208636522293091, + "learning_rate": 1.9124492299304897e-05, + "loss": 2.563441848754883, + "step": 16560 + }, + { + "epoch": 0.045134068327208385, + "grad_norm": 2.610219955444336, + "learning_rate": 1.912340193539594e-05, + "loss": 2.5838953018188477, + "step": 16580 + }, + { + "epoch": 0.04518851231795291, + "grad_norm": 2.267268180847168, + "learning_rate": 1.9122311571486985e-05, + "loss": 2.6006729125976564, + "step": 16600 + }, + { + "epoch": 0.04524295630869743, + "grad_norm": 2.26120662689209, + "learning_rate": 1.912122120757803e-05, + "loss": 2.57251033782959, + "step": 16620 + }, + { + "epoch": 0.04529740029944195, + "grad_norm": 2.220212697982788, + "learning_rate": 1.9120130843669077e-05, + "loss": 2.59658203125, + "step": 16640 + }, + { + "epoch": 0.04535184429018647, + "grad_norm": 2.1279518604278564, + "learning_rate": 1.9119040479760123e-05, + "loss": 2.6267595291137695, + "step": 16660 + }, + { + "epoch": 0.04540628828093099, + "grad_norm": 2.222945213317871, + "learning_rate": 1.911795011585117e-05, + "loss": 2.5712310791015627, + "step": 16680 + }, + { + "epoch": 0.04546073227167551, + "grad_norm": 2.4886975288391113, + "learning_rate": 1.911685975194221e-05, + "loss": 2.597377395629883, + "step": 16700 + }, + { + "epoch": 0.045515176262420035, + "grad_norm": 2.3089773654937744, + "learning_rate": 1.911576938803326e-05, + "loss": 2.630189895629883, + "step": 16720 + }, + { + "epoch": 0.04556962025316456, + "grad_norm": 2.1575584411621094, + "learning_rate": 1.9114679024124302e-05, + "loss": 2.5936519622802736, + "step": 16740 + }, + { + "epoch": 0.04562406424390908, + "grad_norm": 3.2180159091949463, + "learning_rate": 1.9113588660215348e-05, + "loss": 2.5940723419189453, + "step": 16760 + }, + { + "epoch": 0.0456785082346536, + "grad_norm": 2.2388148307800293, + "learning_rate": 1.9112498296306394e-05, + "loss": 2.5769901275634766, + "step": 16780 + }, + { + "epoch": 0.045732952225398124, + "grad_norm": 2.238835096359253, + "learning_rate": 1.911140793239744e-05, + "loss": 2.6126239776611326, + "step": 16800 + }, + { + "epoch": 0.045787396216142646, + "grad_norm": 2.3696765899658203, + "learning_rate": 1.9110317568488486e-05, + "loss": 2.5661027908325194, + "step": 16820 + }, + { + "epoch": 0.04584184020688716, + "grad_norm": 2.2033724784851074, + "learning_rate": 1.910922720457953e-05, + "loss": 2.5824920654296877, + "step": 16840 + }, + { + "epoch": 0.045896284197631684, + "grad_norm": 2.193445920944214, + "learning_rate": 1.9108136840670574e-05, + "loss": 2.5926427841186523, + "step": 16860 + }, + { + "epoch": 0.04595072818837621, + "grad_norm": 2.306762456893921, + "learning_rate": 1.9107046476761623e-05, + "loss": 2.601276969909668, + "step": 16880 + }, + { + "epoch": 0.04600517217912073, + "grad_norm": 2.1266231536865234, + "learning_rate": 1.9105956112852665e-05, + "loss": 2.616020393371582, + "step": 16900 + }, + { + "epoch": 0.04605961616986525, + "grad_norm": 5.2636799812316895, + "learning_rate": 1.910486574894371e-05, + "loss": 2.609017562866211, + "step": 16920 + }, + { + "epoch": 0.046114060160609774, + "grad_norm": 2.1099178791046143, + "learning_rate": 1.9103775385034757e-05, + "loss": 2.547172737121582, + "step": 16940 + }, + { + "epoch": 0.046168504151354296, + "grad_norm": 2.1146204471588135, + "learning_rate": 1.9102685021125803e-05, + "loss": 2.590086555480957, + "step": 16960 + }, + { + "epoch": 0.04622294814209882, + "grad_norm": 2.4054534435272217, + "learning_rate": 1.910159465721685e-05, + "loss": 2.5838062286376955, + "step": 16980 + }, + { + "epoch": 0.046277392132843334, + "grad_norm": 2.1598446369171143, + "learning_rate": 1.9100504293307894e-05, + "loss": 2.5547063827514647, + "step": 17000 + }, + { + "epoch": 0.046331836123587856, + "grad_norm": 2.2167463302612305, + "learning_rate": 1.9099413929398937e-05, + "loss": 2.5629716873168946, + "step": 17020 + }, + { + "epoch": 0.04638628011433238, + "grad_norm": 2.3005778789520264, + "learning_rate": 1.9098323565489986e-05, + "loss": 2.603608512878418, + "step": 17040 + }, + { + "epoch": 0.0464407241050769, + "grad_norm": 2.2303237915039062, + "learning_rate": 1.909723320158103e-05, + "loss": 2.575428771972656, + "step": 17060 + }, + { + "epoch": 0.046495168095821424, + "grad_norm": 2.2353732585906982, + "learning_rate": 1.9096142837672074e-05, + "loss": 2.5639022827148437, + "step": 17080 + }, + { + "epoch": 0.046549612086565946, + "grad_norm": 2.2622251510620117, + "learning_rate": 1.909505247376312e-05, + "loss": 2.601908874511719, + "step": 17100 + }, + { + "epoch": 0.04660405607731047, + "grad_norm": 2.160656452178955, + "learning_rate": 1.9093962109854166e-05, + "loss": 2.593302536010742, + "step": 17120 + }, + { + "epoch": 0.04665850006805499, + "grad_norm": 2.3510022163391113, + "learning_rate": 1.9092871745945212e-05, + "loss": 2.547384834289551, + "step": 17140 + }, + { + "epoch": 0.04671294405879951, + "grad_norm": 3.0294861793518066, + "learning_rate": 1.9091781382036258e-05, + "loss": 2.5735076904296874, + "step": 17160 + }, + { + "epoch": 0.04676738804954403, + "grad_norm": 2.197390079498291, + "learning_rate": 1.90906910181273e-05, + "loss": 2.615141487121582, + "step": 17180 + }, + { + "epoch": 0.04682183204028855, + "grad_norm": 2.1027867794036865, + "learning_rate": 1.908960065421835e-05, + "loss": 2.588683319091797, + "step": 17200 + }, + { + "epoch": 0.04687627603103307, + "grad_norm": 2.331399917602539, + "learning_rate": 1.908851029030939e-05, + "loss": 2.601495361328125, + "step": 17220 + }, + { + "epoch": 0.046930720021777596, + "grad_norm": 2.2198264598846436, + "learning_rate": 1.9087419926400437e-05, + "loss": 2.572374153137207, + "step": 17240 + }, + { + "epoch": 0.04698516401252212, + "grad_norm": 2.240039825439453, + "learning_rate": 1.9086329562491483e-05, + "loss": 2.586861419677734, + "step": 17260 + }, + { + "epoch": 0.04703960800326664, + "grad_norm": 2.2120540142059326, + "learning_rate": 1.908523919858253e-05, + "loss": 2.595490264892578, + "step": 17280 + }, + { + "epoch": 0.04709405199401116, + "grad_norm": 2.2792744636535645, + "learning_rate": 1.9084148834673575e-05, + "loss": 2.577165985107422, + "step": 17300 + }, + { + "epoch": 0.047148495984755685, + "grad_norm": 2.2037055492401123, + "learning_rate": 1.908305847076462e-05, + "loss": 2.582589530944824, + "step": 17320 + }, + { + "epoch": 0.0472029399755002, + "grad_norm": 2.1499695777893066, + "learning_rate": 1.9081968106855663e-05, + "loss": 2.592385673522949, + "step": 17340 + }, + { + "epoch": 0.04725738396624472, + "grad_norm": 2.340771198272705, + "learning_rate": 1.9080877742946712e-05, + "loss": 2.5715152740478517, + "step": 17360 + }, + { + "epoch": 0.047311827956989246, + "grad_norm": 2.1900620460510254, + "learning_rate": 1.9079787379037755e-05, + "loss": 2.5815711975097657, + "step": 17380 + }, + { + "epoch": 0.04736627194773377, + "grad_norm": 2.2830557823181152, + "learning_rate": 1.90786970151288e-05, + "loss": 2.587349510192871, + "step": 17400 + }, + { + "epoch": 0.04742071593847829, + "grad_norm": 2.233581304550171, + "learning_rate": 1.9077606651219846e-05, + "loss": 2.5850095748901367, + "step": 17420 + }, + { + "epoch": 0.04747515992922281, + "grad_norm": 2.118662118911743, + "learning_rate": 1.9076516287310892e-05, + "loss": 2.588897132873535, + "step": 17440 + }, + { + "epoch": 0.047529603919967335, + "grad_norm": 2.2965898513793945, + "learning_rate": 1.9075425923401938e-05, + "loss": 2.5486026763916017, + "step": 17460 + }, + { + "epoch": 0.04758404791071186, + "grad_norm": 3.5844781398773193, + "learning_rate": 1.9074335559492984e-05, + "loss": 2.5959236145019533, + "step": 17480 + }, + { + "epoch": 0.04763849190145638, + "grad_norm": 2.3912813663482666, + "learning_rate": 1.9073245195584026e-05, + "loss": 2.596241760253906, + "step": 17500 + }, + { + "epoch": 0.047692935892200895, + "grad_norm": 2.11598539352417, + "learning_rate": 1.9072154831675075e-05, + "loss": 2.6229976654052733, + "step": 17520 + }, + { + "epoch": 0.04774737988294542, + "grad_norm": 2.1170687675476074, + "learning_rate": 1.9071064467766118e-05, + "loss": 2.613518714904785, + "step": 17540 + }, + { + "epoch": 0.04780182387368994, + "grad_norm": 2.5269370079040527, + "learning_rate": 1.9069974103857164e-05, + "loss": 2.575615119934082, + "step": 17560 + }, + { + "epoch": 0.04785626786443446, + "grad_norm": 2.3671517372131348, + "learning_rate": 1.906888373994821e-05, + "loss": 2.5819570541381838, + "step": 17580 + }, + { + "epoch": 0.047910711855178985, + "grad_norm": 2.2735702991485596, + "learning_rate": 1.9067793376039255e-05, + "loss": 2.571992111206055, + "step": 17600 + }, + { + "epoch": 0.04796515584592351, + "grad_norm": 2.0790488719940186, + "learning_rate": 1.90667030121303e-05, + "loss": 2.6063364028930662, + "step": 17620 + }, + { + "epoch": 0.04801959983666803, + "grad_norm": 2.4133529663085938, + "learning_rate": 1.9065612648221347e-05, + "loss": 2.6106197357177736, + "step": 17640 + }, + { + "epoch": 0.04807404382741255, + "grad_norm": 2.224550724029541, + "learning_rate": 1.906452228431239e-05, + "loss": 2.610115623474121, + "step": 17660 + }, + { + "epoch": 0.04812848781815707, + "grad_norm": 2.138946533203125, + "learning_rate": 1.906343192040344e-05, + "loss": 2.557553291320801, + "step": 17680 + }, + { + "epoch": 0.04818293180890159, + "grad_norm": 2.1681342124938965, + "learning_rate": 1.906234155649448e-05, + "loss": 2.579910469055176, + "step": 17700 + }, + { + "epoch": 0.04823737579964611, + "grad_norm": 2.2647860050201416, + "learning_rate": 1.9061251192585527e-05, + "loss": 2.585797882080078, + "step": 17720 + }, + { + "epoch": 0.048291819790390635, + "grad_norm": 2.49924898147583, + "learning_rate": 1.9060160828676572e-05, + "loss": 2.582747459411621, + "step": 17740 + }, + { + "epoch": 0.04834626378113516, + "grad_norm": 2.2662618160247803, + "learning_rate": 1.9059070464767618e-05, + "loss": 2.5871536254882814, + "step": 17760 + }, + { + "epoch": 0.04840070777187968, + "grad_norm": 2.1592369079589844, + "learning_rate": 1.9057980100858664e-05, + "loss": 2.564101219177246, + "step": 17780 + }, + { + "epoch": 0.0484551517626242, + "grad_norm": 2.130392551422119, + "learning_rate": 1.905688973694971e-05, + "loss": 2.5587257385253905, + "step": 17800 + }, + { + "epoch": 0.048509595753368724, + "grad_norm": 2.169919729232788, + "learning_rate": 1.9055799373040752e-05, + "loss": 2.5724365234375, + "step": 17820 + }, + { + "epoch": 0.04856403974411325, + "grad_norm": 2.202577590942383, + "learning_rate": 1.90547090091318e-05, + "loss": 2.5564952850341798, + "step": 17840 + }, + { + "epoch": 0.04861848373485776, + "grad_norm": 2.437692403793335, + "learning_rate": 1.9053618645222844e-05, + "loss": 2.5894611358642576, + "step": 17860 + }, + { + "epoch": 0.048672927725602284, + "grad_norm": 2.168835401535034, + "learning_rate": 1.905252828131389e-05, + "loss": 2.5669918060302734, + "step": 17880 + }, + { + "epoch": 0.04872737171634681, + "grad_norm": 2.1957216262817383, + "learning_rate": 1.9051437917404936e-05, + "loss": 2.5753602981567383, + "step": 17900 + }, + { + "epoch": 0.04878181570709133, + "grad_norm": 2.1733152866363525, + "learning_rate": 1.905034755349598e-05, + "loss": 2.597913932800293, + "step": 17920 + }, + { + "epoch": 0.04883625969783585, + "grad_norm": 2.294877529144287, + "learning_rate": 1.9049257189587027e-05, + "loss": 2.559251594543457, + "step": 17940 + }, + { + "epoch": 0.048890703688580374, + "grad_norm": 2.170663595199585, + "learning_rate": 1.9048166825678073e-05, + "loss": 2.591808891296387, + "step": 17960 + }, + { + "epoch": 0.048945147679324896, + "grad_norm": 2.1552860736846924, + "learning_rate": 1.9047076461769115e-05, + "loss": 2.547669219970703, + "step": 17980 + }, + { + "epoch": 0.04899959167006942, + "grad_norm": 2.2407853603363037, + "learning_rate": 1.9045986097860165e-05, + "loss": 2.580365753173828, + "step": 18000 + }, + { + "epoch": 0.04905403566081394, + "grad_norm": 2.2348148822784424, + "learning_rate": 1.9044895733951207e-05, + "loss": 2.59734992980957, + "step": 18020 + }, + { + "epoch": 0.049108479651558457, + "grad_norm": 2.3038806915283203, + "learning_rate": 1.9043805370042253e-05, + "loss": 2.5691997528076174, + "step": 18040 + }, + { + "epoch": 0.04916292364230298, + "grad_norm": 2.1676409244537354, + "learning_rate": 1.90427150061333e-05, + "loss": 2.57359619140625, + "step": 18060 + }, + { + "epoch": 0.0492173676330475, + "grad_norm": 2.198269844055176, + "learning_rate": 1.9041624642224344e-05, + "loss": 2.5836347579956054, + "step": 18080 + }, + { + "epoch": 0.049271811623792024, + "grad_norm": 2.0829572677612305, + "learning_rate": 1.904053427831539e-05, + "loss": 2.5999448776245115, + "step": 18100 + }, + { + "epoch": 0.049326255614536546, + "grad_norm": 2.5146608352661133, + "learning_rate": 1.9039443914406436e-05, + "loss": 2.556977081298828, + "step": 18120 + }, + { + "epoch": 0.04938069960528107, + "grad_norm": 2.288431406021118, + "learning_rate": 1.903835355049748e-05, + "loss": 2.5611637115478514, + "step": 18140 + }, + { + "epoch": 0.04943514359602559, + "grad_norm": 2.1158134937286377, + "learning_rate": 1.9037263186588524e-05, + "loss": 2.5709402084350588, + "step": 18160 + }, + { + "epoch": 0.04948958758677011, + "grad_norm": 2.316092014312744, + "learning_rate": 1.903617282267957e-05, + "loss": 2.5677661895751953, + "step": 18180 + }, + { + "epoch": 0.04954403157751463, + "grad_norm": 2.3352603912353516, + "learning_rate": 1.9035082458770616e-05, + "loss": 2.579494285583496, + "step": 18200 + }, + { + "epoch": 0.04959847556825915, + "grad_norm": 2.2144827842712402, + "learning_rate": 1.9033992094861662e-05, + "loss": 2.5598535537719727, + "step": 18220 + }, + { + "epoch": 0.04965291955900367, + "grad_norm": 2.1000711917877197, + "learning_rate": 1.9032901730952708e-05, + "loss": 2.572928237915039, + "step": 18240 + }, + { + "epoch": 0.049707363549748196, + "grad_norm": 2.2878167629241943, + "learning_rate": 1.9031811367043753e-05, + "loss": 2.550933265686035, + "step": 18260 + }, + { + "epoch": 0.04976180754049272, + "grad_norm": 2.175947904586792, + "learning_rate": 1.90307210031348e-05, + "loss": 2.594309616088867, + "step": 18280 + }, + { + "epoch": 0.04981625153123724, + "grad_norm": 2.1495187282562256, + "learning_rate": 1.902963063922584e-05, + "loss": 2.5657108306884764, + "step": 18300 + }, + { + "epoch": 0.04987069552198176, + "grad_norm": 2.1366357803344727, + "learning_rate": 1.9028540275316887e-05, + "loss": 2.5850128173828124, + "step": 18320 + }, + { + "epoch": 0.049925139512726285, + "grad_norm": 2.3925421237945557, + "learning_rate": 1.9027449911407933e-05, + "loss": 2.580697441101074, + "step": 18340 + }, + { + "epoch": 0.04997958350347081, + "grad_norm": 2.2110228538513184, + "learning_rate": 1.902635954749898e-05, + "loss": 2.5833831787109376, + "step": 18360 + }, + { + "epoch": 0.05003402749421532, + "grad_norm": 2.202796459197998, + "learning_rate": 1.9025269183590025e-05, + "loss": 2.549668884277344, + "step": 18380 + }, + { + "epoch": 0.050088471484959846, + "grad_norm": 2.236145257949829, + "learning_rate": 1.902417881968107e-05, + "loss": 2.5771469116210937, + "step": 18400 + }, + { + "epoch": 0.05014291547570437, + "grad_norm": 2.3660871982574463, + "learning_rate": 1.9023088455772116e-05, + "loss": 2.5736555099487304, + "step": 18420 + }, + { + "epoch": 0.05019735946644889, + "grad_norm": 2.2815439701080322, + "learning_rate": 1.9021998091863162e-05, + "loss": 2.572745704650879, + "step": 18440 + }, + { + "epoch": 0.05025180345719341, + "grad_norm": 2.28481125831604, + "learning_rate": 1.9020907727954205e-05, + "loss": 2.579486846923828, + "step": 18460 + }, + { + "epoch": 0.050306247447937935, + "grad_norm": 3.973605155944824, + "learning_rate": 1.901981736404525e-05, + "loss": 2.542791748046875, + "step": 18480 + }, + { + "epoch": 0.05036069143868246, + "grad_norm": 2.3384673595428467, + "learning_rate": 1.9018727000136296e-05, + "loss": 2.5639713287353514, + "step": 18500 + }, + { + "epoch": 0.05041513542942698, + "grad_norm": 2.5233027935028076, + "learning_rate": 1.9017636636227342e-05, + "loss": 2.582438087463379, + "step": 18520 + }, + { + "epoch": 0.050469579420171495, + "grad_norm": 2.279411554336548, + "learning_rate": 1.9016546272318388e-05, + "loss": 2.5851343154907225, + "step": 18540 + }, + { + "epoch": 0.05052402341091602, + "grad_norm": 2.291764974594116, + "learning_rate": 1.9015455908409434e-05, + "loss": 2.5821605682373048, + "step": 18560 + }, + { + "epoch": 0.05057846740166054, + "grad_norm": 2.276137351989746, + "learning_rate": 1.901436554450048e-05, + "loss": 2.558827590942383, + "step": 18580 + }, + { + "epoch": 0.05063291139240506, + "grad_norm": 2.3308210372924805, + "learning_rate": 1.9013275180591525e-05, + "loss": 2.5833414077758787, + "step": 18600 + }, + { + "epoch": 0.050687355383149585, + "grad_norm": 2.3683059215545654, + "learning_rate": 1.9012184816682568e-05, + "loss": 2.5992849349975584, + "step": 18620 + }, + { + "epoch": 0.05074179937389411, + "grad_norm": 2.1625869274139404, + "learning_rate": 1.9011094452773614e-05, + "loss": 2.5698522567749023, + "step": 18640 + }, + { + "epoch": 0.05079624336463863, + "grad_norm": 2.240417003631592, + "learning_rate": 1.901000408886466e-05, + "loss": 2.575149917602539, + "step": 18660 + }, + { + "epoch": 0.05085068735538315, + "grad_norm": 2.1725316047668457, + "learning_rate": 1.9008913724955705e-05, + "loss": 2.5789955139160154, + "step": 18680 + }, + { + "epoch": 0.050905131346127674, + "grad_norm": 2.197335958480835, + "learning_rate": 1.900782336104675e-05, + "loss": 2.5972627639770507, + "step": 18700 + }, + { + "epoch": 0.05095957533687219, + "grad_norm": 2.2099251747131348, + "learning_rate": 1.9006732997137797e-05, + "loss": 2.562291145324707, + "step": 18720 + }, + { + "epoch": 0.05101401932761671, + "grad_norm": 2.203601837158203, + "learning_rate": 1.9005642633228843e-05, + "loss": 2.580080986022949, + "step": 18740 + }, + { + "epoch": 0.051068463318361235, + "grad_norm": 2.2345564365386963, + "learning_rate": 1.900455226931989e-05, + "loss": 2.5828651428222655, + "step": 18760 + }, + { + "epoch": 0.05112290730910576, + "grad_norm": 2.198194980621338, + "learning_rate": 1.900346190541093e-05, + "loss": 2.5983022689819335, + "step": 18780 + }, + { + "epoch": 0.05117735129985028, + "grad_norm": 2.161675214767456, + "learning_rate": 1.9002371541501977e-05, + "loss": 2.584638786315918, + "step": 18800 + }, + { + "epoch": 0.0512317952905948, + "grad_norm": 2.2392752170562744, + "learning_rate": 1.9001281177593022e-05, + "loss": 2.5690353393554686, + "step": 18820 + }, + { + "epoch": 0.051286239281339324, + "grad_norm": 2.1703803539276123, + "learning_rate": 1.9000190813684068e-05, + "loss": 2.5888555526733397, + "step": 18840 + }, + { + "epoch": 0.05134068327208385, + "grad_norm": 2.513322353363037, + "learning_rate": 1.8999100449775114e-05, + "loss": 2.6070838928222657, + "step": 18860 + }, + { + "epoch": 0.05139512726282836, + "grad_norm": 2.1538357734680176, + "learning_rate": 1.899801008586616e-05, + "loss": 2.567066955566406, + "step": 18880 + }, + { + "epoch": 0.051449571253572884, + "grad_norm": 2.328333616256714, + "learning_rate": 1.8996919721957206e-05, + "loss": 2.569324493408203, + "step": 18900 + }, + { + "epoch": 0.05150401524431741, + "grad_norm": 2.5017473697662354, + "learning_rate": 1.899582935804825e-05, + "loss": 2.5764896392822267, + "step": 18920 + }, + { + "epoch": 0.05155845923506193, + "grad_norm": 2.3962013721466064, + "learning_rate": 1.8994738994139294e-05, + "loss": 2.6015079498291014, + "step": 18940 + }, + { + "epoch": 0.05161290322580645, + "grad_norm": 2.0989603996276855, + "learning_rate": 1.899364863023034e-05, + "loss": 2.5920236587524412, + "step": 18960 + }, + { + "epoch": 0.051667347216550974, + "grad_norm": 2.1387953758239746, + "learning_rate": 1.8992558266321386e-05, + "loss": 2.575980377197266, + "step": 18980 + }, + { + "epoch": 0.051721791207295496, + "grad_norm": 4.224916934967041, + "learning_rate": 1.899146790241243e-05, + "loss": 2.570087432861328, + "step": 19000 + }, + { + "epoch": 0.05177623519804002, + "grad_norm": 2.1226375102996826, + "learning_rate": 1.8990377538503477e-05, + "loss": 2.59182071685791, + "step": 19020 + }, + { + "epoch": 0.05183067918878454, + "grad_norm": 2.2246365547180176, + "learning_rate": 1.8989287174594523e-05, + "loss": 2.5880435943603515, + "step": 19040 + }, + { + "epoch": 0.05188512317952906, + "grad_norm": 2.234271287918091, + "learning_rate": 1.898819681068557e-05, + "loss": 2.598816680908203, + "step": 19060 + }, + { + "epoch": 0.05193956717027358, + "grad_norm": 2.137120246887207, + "learning_rate": 1.8987106446776615e-05, + "loss": 2.5679447174072267, + "step": 19080 + }, + { + "epoch": 0.0519940111610181, + "grad_norm": 2.113203287124634, + "learning_rate": 1.8986016082867657e-05, + "loss": 2.6088703155517576, + "step": 19100 + }, + { + "epoch": 0.052048455151762624, + "grad_norm": 2.0395631790161133, + "learning_rate": 1.8984925718958703e-05, + "loss": 2.5508838653564454, + "step": 19120 + }, + { + "epoch": 0.052102899142507146, + "grad_norm": 2.2693984508514404, + "learning_rate": 1.898383535504975e-05, + "loss": 2.603200912475586, + "step": 19140 + }, + { + "epoch": 0.05215734313325167, + "grad_norm": 4.748312473297119, + "learning_rate": 1.8982744991140794e-05, + "loss": 2.556200408935547, + "step": 19160 + }, + { + "epoch": 0.05221178712399619, + "grad_norm": 2.0926761627197266, + "learning_rate": 1.898165462723184e-05, + "loss": 2.5582666397094727, + "step": 19180 + }, + { + "epoch": 0.05226623111474071, + "grad_norm": 2.1424410343170166, + "learning_rate": 1.8980564263322886e-05, + "loss": 2.5591854095458983, + "step": 19200 + }, + { + "epoch": 0.05232067510548523, + "grad_norm": 2.1880252361297607, + "learning_rate": 1.8979473899413932e-05, + "loss": 2.5892385482788085, + "step": 19220 + }, + { + "epoch": 0.05237511909622975, + "grad_norm": 2.2275619506835938, + "learning_rate": 1.8978383535504978e-05, + "loss": 2.598979377746582, + "step": 19240 + }, + { + "epoch": 0.052429563086974273, + "grad_norm": 2.1708579063415527, + "learning_rate": 1.897729317159602e-05, + "loss": 2.6250377655029298, + "step": 19260 + }, + { + "epoch": 0.052484007077718796, + "grad_norm": 2.182008743286133, + "learning_rate": 1.8976202807687066e-05, + "loss": 2.5788957595825197, + "step": 19280 + }, + { + "epoch": 0.05253845106846332, + "grad_norm": 2.273928165435791, + "learning_rate": 1.8975112443778112e-05, + "loss": 2.5782032012939453, + "step": 19300 + }, + { + "epoch": 0.05259289505920784, + "grad_norm": 2.1453709602355957, + "learning_rate": 1.8974022079869158e-05, + "loss": 2.5969921112060548, + "step": 19320 + }, + { + "epoch": 0.05264733904995236, + "grad_norm": 2.1415576934814453, + "learning_rate": 1.8972931715960203e-05, + "loss": 2.5815338134765624, + "step": 19340 + }, + { + "epoch": 0.052701783040696885, + "grad_norm": 2.291043758392334, + "learning_rate": 1.897184135205125e-05, + "loss": 2.6036624908447266, + "step": 19360 + }, + { + "epoch": 0.05275622703144141, + "grad_norm": 2.3288066387176514, + "learning_rate": 1.8970750988142295e-05, + "loss": 2.5714101791381836, + "step": 19380 + }, + { + "epoch": 0.05281067102218592, + "grad_norm": 2.1178462505340576, + "learning_rate": 1.896966062423334e-05, + "loss": 2.5567371368408205, + "step": 19400 + }, + { + "epoch": 0.052865115012930446, + "grad_norm": 2.148852825164795, + "learning_rate": 1.8968570260324383e-05, + "loss": 2.590232086181641, + "step": 19420 + }, + { + "epoch": 0.05291955900367497, + "grad_norm": 2.1317782402038574, + "learning_rate": 1.896747989641543e-05, + "loss": 2.572703170776367, + "step": 19440 + }, + { + "epoch": 0.05297400299441949, + "grad_norm": 2.312169075012207, + "learning_rate": 1.8966389532506475e-05, + "loss": 2.5602447509765627, + "step": 19460 + }, + { + "epoch": 0.05302844698516401, + "grad_norm": 2.2717373371124268, + "learning_rate": 1.896529916859752e-05, + "loss": 2.56848258972168, + "step": 19480 + }, + { + "epoch": 0.053082890975908535, + "grad_norm": 2.155869960784912, + "learning_rate": 1.8964208804688566e-05, + "loss": 2.589608001708984, + "step": 19500 + }, + { + "epoch": 0.05313733496665306, + "grad_norm": 2.1750447750091553, + "learning_rate": 1.8963118440779612e-05, + "loss": 2.6049123764038087, + "step": 19520 + }, + { + "epoch": 0.05319177895739758, + "grad_norm": 2.1206767559051514, + "learning_rate": 1.8962028076870658e-05, + "loss": 2.5902837753295898, + "step": 19540 + }, + { + "epoch": 0.053246222948142095, + "grad_norm": 2.190700054168701, + "learning_rate": 1.8960937712961704e-05, + "loss": 2.5992237091064454, + "step": 19560 + }, + { + "epoch": 0.05330066693888662, + "grad_norm": 2.3732011318206787, + "learning_rate": 1.8959847349052746e-05, + "loss": 2.6152063369750977, + "step": 19580 + }, + { + "epoch": 0.05335511092963114, + "grad_norm": 2.22076416015625, + "learning_rate": 1.8958756985143792e-05, + "loss": 2.6041521072387694, + "step": 19600 + }, + { + "epoch": 0.05340955492037566, + "grad_norm": 2.261955976486206, + "learning_rate": 1.8957666621234838e-05, + "loss": 2.564017486572266, + "step": 19620 + }, + { + "epoch": 0.053463998911120185, + "grad_norm": 2.2718262672424316, + "learning_rate": 1.8956576257325884e-05, + "loss": 2.575114059448242, + "step": 19640 + }, + { + "epoch": 0.05351844290186471, + "grad_norm": 2.1671833992004395, + "learning_rate": 1.895548589341693e-05, + "loss": 2.6043170928955077, + "step": 19660 + }, + { + "epoch": 0.05357288689260923, + "grad_norm": 2.116358995437622, + "learning_rate": 1.8954395529507975e-05, + "loss": 2.6006282806396483, + "step": 19680 + }, + { + "epoch": 0.05362733088335375, + "grad_norm": 2.254878520965576, + "learning_rate": 1.895330516559902e-05, + "loss": 2.5350852966308595, + "step": 19700 + }, + { + "epoch": 0.053681774874098274, + "grad_norm": 2.2230448722839355, + "learning_rate": 1.8952214801690067e-05, + "loss": 2.5568515777587892, + "step": 19720 + }, + { + "epoch": 0.05373621886484279, + "grad_norm": 2.4856536388397217, + "learning_rate": 1.8951124437781113e-05, + "loss": 2.578626251220703, + "step": 19740 + }, + { + "epoch": 0.05379066285558731, + "grad_norm": 2.060912847518921, + "learning_rate": 1.8950034073872155e-05, + "loss": 2.535233497619629, + "step": 19760 + }, + { + "epoch": 0.053845106846331835, + "grad_norm": 2.18770170211792, + "learning_rate": 1.89489437099632e-05, + "loss": 2.61221923828125, + "step": 19780 + }, + { + "epoch": 0.05389955083707636, + "grad_norm": 3.2052812576293945, + "learning_rate": 1.8947853346054247e-05, + "loss": 2.5758962631225586, + "step": 19800 + }, + { + "epoch": 0.05395399482782088, + "grad_norm": 3.528467893600464, + "learning_rate": 1.8946762982145293e-05, + "loss": 2.5878087997436525, + "step": 19820 + }, + { + "epoch": 0.0540084388185654, + "grad_norm": 2.0756704807281494, + "learning_rate": 1.894567261823634e-05, + "loss": 2.548657989501953, + "step": 19840 + }, + { + "epoch": 0.054062882809309924, + "grad_norm": 2.0741384029388428, + "learning_rate": 1.8944582254327384e-05, + "loss": 2.6073976516723634, + "step": 19860 + }, + { + "epoch": 0.05411732680005445, + "grad_norm": 2.09126877784729, + "learning_rate": 1.894349189041843e-05, + "loss": 2.6078733444213866, + "step": 19880 + }, + { + "epoch": 0.05417177079079897, + "grad_norm": 2.3375909328460693, + "learning_rate": 1.8942401526509476e-05, + "loss": 2.597652626037598, + "step": 19900 + }, + { + "epoch": 0.054226214781543484, + "grad_norm": 3.103196144104004, + "learning_rate": 1.8941311162600518e-05, + "loss": 2.6043624877929688, + "step": 19920 + }, + { + "epoch": 0.05428065877228801, + "grad_norm": 2.107133388519287, + "learning_rate": 1.8940220798691564e-05, + "loss": 2.5530370712280273, + "step": 19940 + }, + { + "epoch": 0.05433510276303253, + "grad_norm": 2.424520254135132, + "learning_rate": 1.893913043478261e-05, + "loss": 2.56463680267334, + "step": 19960 + }, + { + "epoch": 0.05438954675377705, + "grad_norm": 2.054795980453491, + "learning_rate": 1.8938040070873656e-05, + "loss": 2.582062911987305, + "step": 19980 + }, + { + "epoch": 0.054443990744521574, + "grad_norm": 2.1434266567230225, + "learning_rate": 1.89369497069647e-05, + "loss": 2.565500831604004, + "step": 20000 + }, + { + "epoch": 0.054498434735266096, + "grad_norm": 2.1476476192474365, + "learning_rate": 1.8935859343055747e-05, + "loss": 2.6140541076660155, + "step": 20020 + }, + { + "epoch": 0.05455287872601062, + "grad_norm": 2.171649932861328, + "learning_rate": 1.8934768979146793e-05, + "loss": 2.596476745605469, + "step": 20040 + }, + { + "epoch": 0.05460732271675514, + "grad_norm": 2.3643696308135986, + "learning_rate": 1.893367861523784e-05, + "loss": 2.5432861328125, + "step": 20060 + }, + { + "epoch": 0.05466176670749966, + "grad_norm": 2.5638654232025146, + "learning_rate": 1.893258825132888e-05, + "loss": 2.59295654296875, + "step": 20080 + }, + { + "epoch": 0.05471621069824418, + "grad_norm": 2.3370957374572754, + "learning_rate": 1.8931497887419927e-05, + "loss": 2.598040771484375, + "step": 20100 + }, + { + "epoch": 0.0547706546889887, + "grad_norm": 2.1154351234436035, + "learning_rate": 1.8930407523510973e-05, + "loss": 2.5648950576782226, + "step": 20120 + }, + { + "epoch": 0.054825098679733224, + "grad_norm": 2.1341800689697266, + "learning_rate": 1.892931715960202e-05, + "loss": 2.580570411682129, + "step": 20140 + }, + { + "epoch": 0.054879542670477746, + "grad_norm": 2.147289276123047, + "learning_rate": 1.8928226795693065e-05, + "loss": 2.5953561782836916, + "step": 20160 + }, + { + "epoch": 0.05493398666122227, + "grad_norm": 2.2639870643615723, + "learning_rate": 1.892713643178411e-05, + "loss": 2.5589984893798827, + "step": 20180 + }, + { + "epoch": 0.05498843065196679, + "grad_norm": 2.737220525741577, + "learning_rate": 1.8926046067875156e-05, + "loss": 2.5901647567749024, + "step": 20200 + }, + { + "epoch": 0.05504287464271131, + "grad_norm": 2.154000759124756, + "learning_rate": 1.8924955703966202e-05, + "loss": 2.603729820251465, + "step": 20220 + }, + { + "epoch": 0.055097318633455836, + "grad_norm": 2.2766308784484863, + "learning_rate": 1.8923865340057244e-05, + "loss": 2.593448448181152, + "step": 20240 + }, + { + "epoch": 0.05515176262420035, + "grad_norm": 2.1691813468933105, + "learning_rate": 1.892277497614829e-05, + "loss": 2.581180191040039, + "step": 20260 + }, + { + "epoch": 0.055206206614944874, + "grad_norm": 2.113574981689453, + "learning_rate": 1.8921684612239336e-05, + "loss": 2.5782241821289062, + "step": 20280 + }, + { + "epoch": 0.055260650605689396, + "grad_norm": 2.26521635055542, + "learning_rate": 1.8920594248330382e-05, + "loss": 2.5702726364135744, + "step": 20300 + }, + { + "epoch": 0.05531509459643392, + "grad_norm": 2.2626616954803467, + "learning_rate": 1.8919503884421428e-05, + "loss": 2.5842323303222656, + "step": 20320 + }, + { + "epoch": 0.05536953858717844, + "grad_norm": 2.195469856262207, + "learning_rate": 1.8918413520512473e-05, + "loss": 2.5708370208740234, + "step": 20340 + }, + { + "epoch": 0.05542398257792296, + "grad_norm": 2.186277151107788, + "learning_rate": 1.8917323156603516e-05, + "loss": 2.5686426162719727, + "step": 20360 + }, + { + "epoch": 0.055478426568667485, + "grad_norm": 2.1936655044555664, + "learning_rate": 1.8916232792694565e-05, + "loss": 2.583620071411133, + "step": 20380 + }, + { + "epoch": 0.05553287055941201, + "grad_norm": 2.200296401977539, + "learning_rate": 1.8915142428785608e-05, + "loss": 2.544207000732422, + "step": 20400 + }, + { + "epoch": 0.05558731455015652, + "grad_norm": 2.219076633453369, + "learning_rate": 1.8914052064876653e-05, + "loss": 2.573689079284668, + "step": 20420 + }, + { + "epoch": 0.055641758540901046, + "grad_norm": 2.894016981124878, + "learning_rate": 1.89129617009677e-05, + "loss": 2.5744409561157227, + "step": 20440 + }, + { + "epoch": 0.05569620253164557, + "grad_norm": 2.1397926807403564, + "learning_rate": 1.8911871337058745e-05, + "loss": 2.5953399658203127, + "step": 20460 + }, + { + "epoch": 0.05575064652239009, + "grad_norm": 2.199742317199707, + "learning_rate": 1.891078097314979e-05, + "loss": 2.5828277587890627, + "step": 20480 + }, + { + "epoch": 0.05580509051313461, + "grad_norm": 2.1946144104003906, + "learning_rate": 1.8909690609240837e-05, + "loss": 2.587192726135254, + "step": 20500 + }, + { + "epoch": 0.055859534503879135, + "grad_norm": 2.240936756134033, + "learning_rate": 1.890860024533188e-05, + "loss": 2.5841802597045898, + "step": 20520 + }, + { + "epoch": 0.05591397849462366, + "grad_norm": 2.125535011291504, + "learning_rate": 1.8907509881422928e-05, + "loss": 2.5633523941040037, + "step": 20540 + }, + { + "epoch": 0.05596842248536818, + "grad_norm": 2.1811635494232178, + "learning_rate": 1.890641951751397e-05, + "loss": 2.576185607910156, + "step": 20560 + }, + { + "epoch": 0.0560228664761127, + "grad_norm": 2.1812336444854736, + "learning_rate": 1.8905329153605016e-05, + "loss": 2.56775016784668, + "step": 20580 + }, + { + "epoch": 0.05607731046685722, + "grad_norm": 2.1448423862457275, + "learning_rate": 1.8904238789696062e-05, + "loss": 2.5915143966674803, + "step": 20600 + }, + { + "epoch": 0.05613175445760174, + "grad_norm": 2.1163270473480225, + "learning_rate": 1.8903148425787108e-05, + "loss": 2.5772966384887694, + "step": 20620 + }, + { + "epoch": 0.05618619844834626, + "grad_norm": 2.2243707180023193, + "learning_rate": 1.8902058061878154e-05, + "loss": 2.5976091384887696, + "step": 20640 + }, + { + "epoch": 0.056240642439090785, + "grad_norm": 2.02801775932312, + "learning_rate": 1.89009676979692e-05, + "loss": 2.5476095199584963, + "step": 20660 + }, + { + "epoch": 0.05629508642983531, + "grad_norm": 2.132317543029785, + "learning_rate": 1.8899877334060242e-05, + "loss": 2.548537826538086, + "step": 20680 + }, + { + "epoch": 0.05634953042057983, + "grad_norm": 2.1979992389678955, + "learning_rate": 1.889878697015129e-05, + "loss": 2.5320606231689453, + "step": 20700 + }, + { + "epoch": 0.05640397441132435, + "grad_norm": 2.4048068523406982, + "learning_rate": 1.8897696606242334e-05, + "loss": 2.5515443801879885, + "step": 20720 + }, + { + "epoch": 0.056458418402068875, + "grad_norm": 2.1291444301605225, + "learning_rate": 1.889660624233338e-05, + "loss": 2.5753122329711915, + "step": 20740 + }, + { + "epoch": 0.05651286239281339, + "grad_norm": 2.1127593517303467, + "learning_rate": 1.8895515878424425e-05, + "loss": 2.586741638183594, + "step": 20760 + }, + { + "epoch": 0.05656730638355791, + "grad_norm": 2.265756130218506, + "learning_rate": 1.889442551451547e-05, + "loss": 2.583378791809082, + "step": 20780 + }, + { + "epoch": 0.056621750374302435, + "grad_norm": 2.252526044845581, + "learning_rate": 1.8893335150606517e-05, + "loss": 2.569747734069824, + "step": 20800 + }, + { + "epoch": 0.05667619436504696, + "grad_norm": 2.2775661945343018, + "learning_rate": 1.8892244786697563e-05, + "loss": 2.569407272338867, + "step": 20820 + }, + { + "epoch": 0.05673063835579148, + "grad_norm": 2.4196853637695312, + "learning_rate": 1.8891154422788605e-05, + "loss": 2.561835289001465, + "step": 20840 + }, + { + "epoch": 0.056785082346536, + "grad_norm": 2.1052513122558594, + "learning_rate": 1.8890064058879654e-05, + "loss": 2.5571876525878907, + "step": 20860 + }, + { + "epoch": 0.056839526337280524, + "grad_norm": 2.3229868412017822, + "learning_rate": 1.8888973694970697e-05, + "loss": 2.581191062927246, + "step": 20880 + }, + { + "epoch": 0.05689397032802505, + "grad_norm": 2.080143451690674, + "learning_rate": 1.8887883331061743e-05, + "loss": 2.60589714050293, + "step": 20900 + }, + { + "epoch": 0.05694841431876957, + "grad_norm": 2.18637752532959, + "learning_rate": 1.888679296715279e-05, + "loss": 2.5734375, + "step": 20920 + }, + { + "epoch": 0.057002858309514085, + "grad_norm": 2.1905317306518555, + "learning_rate": 1.8885702603243834e-05, + "loss": 2.5716135025024416, + "step": 20940 + }, + { + "epoch": 0.05705730230025861, + "grad_norm": 2.4867501258850098, + "learning_rate": 1.888461223933488e-05, + "loss": 2.570295524597168, + "step": 20960 + }, + { + "epoch": 0.05711174629100313, + "grad_norm": 2.132891893386841, + "learning_rate": 1.8883521875425926e-05, + "loss": 2.5417015075683596, + "step": 20980 + }, + { + "epoch": 0.05716619028174765, + "grad_norm": 2.2856638431549072, + "learning_rate": 1.8882431511516968e-05, + "loss": 2.564146614074707, + "step": 21000 + }, + { + "epoch": 0.057220634272492174, + "grad_norm": 2.258615016937256, + "learning_rate": 1.8881341147608017e-05, + "loss": 2.566813659667969, + "step": 21020 + }, + { + "epoch": 0.057275078263236696, + "grad_norm": 2.2283573150634766, + "learning_rate": 1.888025078369906e-05, + "loss": 2.559792709350586, + "step": 21040 + }, + { + "epoch": 0.05732952225398122, + "grad_norm": 2.086956262588501, + "learning_rate": 1.8879160419790106e-05, + "loss": 2.570384979248047, + "step": 21060 + }, + { + "epoch": 0.05738396624472574, + "grad_norm": 2.3279855251312256, + "learning_rate": 1.887807005588115e-05, + "loss": 2.5963251113891603, + "step": 21080 + }, + { + "epoch": 0.05743841023547026, + "grad_norm": 2.375997304916382, + "learning_rate": 1.8876979691972197e-05, + "loss": 2.5772773742675783, + "step": 21100 + }, + { + "epoch": 0.05749285422621478, + "grad_norm": 2.3389904499053955, + "learning_rate": 1.8875889328063243e-05, + "loss": 2.561319923400879, + "step": 21120 + }, + { + "epoch": 0.0575472982169593, + "grad_norm": 2.281374216079712, + "learning_rate": 1.887479896415429e-05, + "loss": 2.564719009399414, + "step": 21140 + }, + { + "epoch": 0.057601742207703824, + "grad_norm": 2.162865400314331, + "learning_rate": 1.887370860024533e-05, + "loss": 2.5556644439697265, + "step": 21160 + }, + { + "epoch": 0.057656186198448346, + "grad_norm": 2.2047998905181885, + "learning_rate": 1.887261823633638e-05, + "loss": 2.5603769302368162, + "step": 21180 + }, + { + "epoch": 0.05771063018919287, + "grad_norm": 2.2331786155700684, + "learning_rate": 1.8871527872427423e-05, + "loss": 2.580649185180664, + "step": 21200 + }, + { + "epoch": 0.05776507417993739, + "grad_norm": 2.315486431121826, + "learning_rate": 1.887043750851847e-05, + "loss": 2.571198654174805, + "step": 21220 + }, + { + "epoch": 0.05781951817068191, + "grad_norm": 2.1689584255218506, + "learning_rate": 1.8869347144609515e-05, + "loss": 2.560248374938965, + "step": 21240 + }, + { + "epoch": 0.057873962161426436, + "grad_norm": 2.3297417163848877, + "learning_rate": 1.886825678070056e-05, + "loss": 2.5854997634887695, + "step": 21260 + }, + { + "epoch": 0.05792840615217095, + "grad_norm": 2.0845184326171875, + "learning_rate": 1.8867166416791606e-05, + "loss": 2.572097587585449, + "step": 21280 + }, + { + "epoch": 0.057982850142915474, + "grad_norm": 2.141206741333008, + "learning_rate": 1.8866076052882652e-05, + "loss": 2.580637741088867, + "step": 21300 + }, + { + "epoch": 0.058037294133659996, + "grad_norm": 2.197453737258911, + "learning_rate": 1.8864985688973694e-05, + "loss": 2.576548957824707, + "step": 21320 + }, + { + "epoch": 0.05809173812440452, + "grad_norm": 2.121647596359253, + "learning_rate": 1.8863895325064744e-05, + "loss": 2.571359634399414, + "step": 21340 + }, + { + "epoch": 0.05814618211514904, + "grad_norm": 2.1036458015441895, + "learning_rate": 1.8862804961155786e-05, + "loss": 2.582489776611328, + "step": 21360 + }, + { + "epoch": 0.05820062610589356, + "grad_norm": 2.22977876663208, + "learning_rate": 1.8861714597246832e-05, + "loss": 2.565116119384766, + "step": 21380 + }, + { + "epoch": 0.058255070096638085, + "grad_norm": 2.22660756111145, + "learning_rate": 1.8860624233337878e-05, + "loss": 2.610312652587891, + "step": 21400 + }, + { + "epoch": 0.05830951408738261, + "grad_norm": 2.1532413959503174, + "learning_rate": 1.8859533869428923e-05, + "loss": 2.5669940948486327, + "step": 21420 + }, + { + "epoch": 0.05836395807812712, + "grad_norm": 2.169442892074585, + "learning_rate": 1.885844350551997e-05, + "loss": 2.568611717224121, + "step": 21440 + }, + { + "epoch": 0.058418402068871646, + "grad_norm": 2.2518622875213623, + "learning_rate": 1.8857353141611015e-05, + "loss": 2.580371856689453, + "step": 21460 + }, + { + "epoch": 0.05847284605961617, + "grad_norm": 2.1761629581451416, + "learning_rate": 1.8856262777702058e-05, + "loss": 2.585295867919922, + "step": 21480 + }, + { + "epoch": 0.05852729005036069, + "grad_norm": 2.098457098007202, + "learning_rate": 1.8855172413793107e-05, + "loss": 2.548899269104004, + "step": 21500 + }, + { + "epoch": 0.05858173404110521, + "grad_norm": 2.151170492172241, + "learning_rate": 1.885408204988415e-05, + "loss": 2.582443618774414, + "step": 21520 + }, + { + "epoch": 0.058636178031849735, + "grad_norm": 2.2131905555725098, + "learning_rate": 1.8852991685975195e-05, + "loss": 2.563588523864746, + "step": 21540 + }, + { + "epoch": 0.05869062202259426, + "grad_norm": 2.235459566116333, + "learning_rate": 1.885190132206624e-05, + "loss": 2.5751951217651365, + "step": 21560 + }, + { + "epoch": 0.05874506601333878, + "grad_norm": 2.120368003845215, + "learning_rate": 1.8850810958157287e-05, + "loss": 2.595648193359375, + "step": 21580 + }, + { + "epoch": 0.0587995100040833, + "grad_norm": 2.107194662094116, + "learning_rate": 1.8849720594248332e-05, + "loss": 2.5588794708251954, + "step": 21600 + }, + { + "epoch": 0.05885395399482782, + "grad_norm": 2.131938934326172, + "learning_rate": 1.8848630230339378e-05, + "loss": 2.5940425872802733, + "step": 21620 + }, + { + "epoch": 0.05890839798557234, + "grad_norm": 2.340041160583496, + "learning_rate": 1.884753986643042e-05, + "loss": 2.578599548339844, + "step": 21640 + }, + { + "epoch": 0.05896284197631686, + "grad_norm": 2.252915620803833, + "learning_rate": 1.884644950252147e-05, + "loss": 2.557796669006348, + "step": 21660 + }, + { + "epoch": 0.059017285967061385, + "grad_norm": 2.229461908340454, + "learning_rate": 1.8845359138612512e-05, + "loss": 2.5672819137573244, + "step": 21680 + }, + { + "epoch": 0.05907172995780591, + "grad_norm": 2.428284168243408, + "learning_rate": 1.8844268774703558e-05, + "loss": 2.5734771728515624, + "step": 21700 + }, + { + "epoch": 0.05912617394855043, + "grad_norm": 2.290196180343628, + "learning_rate": 1.8843178410794604e-05, + "loss": 2.5324016571044923, + "step": 21720 + }, + { + "epoch": 0.05918061793929495, + "grad_norm": 2.2099475860595703, + "learning_rate": 1.884208804688565e-05, + "loss": 2.5673328399658204, + "step": 21740 + }, + { + "epoch": 0.059235061930039475, + "grad_norm": 2.1537859439849854, + "learning_rate": 1.8840997682976695e-05, + "loss": 2.559770393371582, + "step": 21760 + }, + { + "epoch": 0.05928950592078399, + "grad_norm": 2.1147372722625732, + "learning_rate": 1.883990731906774e-05, + "loss": 2.5733423233032227, + "step": 21780 + }, + { + "epoch": 0.05934394991152851, + "grad_norm": 2.127932071685791, + "learning_rate": 1.8838816955158784e-05, + "loss": 2.5811229705810548, + "step": 21800 + }, + { + "epoch": 0.059398393902273035, + "grad_norm": 2.371154308319092, + "learning_rate": 1.8837726591249833e-05, + "loss": 2.563626289367676, + "step": 21820 + }, + { + "epoch": 0.05945283789301756, + "grad_norm": 2.1663870811462402, + "learning_rate": 1.8836636227340875e-05, + "loss": 2.5963109970092773, + "step": 21840 + }, + { + "epoch": 0.05950728188376208, + "grad_norm": 2.276395559310913, + "learning_rate": 1.883554586343192e-05, + "loss": 2.53997745513916, + "step": 21860 + }, + { + "epoch": 0.0595617258745066, + "grad_norm": 2.261847972869873, + "learning_rate": 1.8834455499522967e-05, + "loss": 2.58294563293457, + "step": 21880 + }, + { + "epoch": 0.059616169865251124, + "grad_norm": 2.018873453140259, + "learning_rate": 1.8833365135614013e-05, + "loss": 2.562983512878418, + "step": 21900 + }, + { + "epoch": 0.05967061385599565, + "grad_norm": 3.4997031688690186, + "learning_rate": 1.883227477170506e-05, + "loss": 2.6048145294189453, + "step": 21920 + }, + { + "epoch": 0.05972505784674017, + "grad_norm": 2.2438459396362305, + "learning_rate": 1.8831184407796104e-05, + "loss": 2.5784168243408203, + "step": 21940 + }, + { + "epoch": 0.059779501837484685, + "grad_norm": 2.2181284427642822, + "learning_rate": 1.8830094043887147e-05, + "loss": 2.579705810546875, + "step": 21960 + }, + { + "epoch": 0.05983394582822921, + "grad_norm": 2.128181219100952, + "learning_rate": 1.8829003679978196e-05, + "loss": 2.56090087890625, + "step": 21980 + }, + { + "epoch": 0.05988838981897373, + "grad_norm": 2.0975699424743652, + "learning_rate": 1.882791331606924e-05, + "loss": 2.5722463607788084, + "step": 22000 + }, + { + "epoch": 0.05994283380971825, + "grad_norm": 2.1801252365112305, + "learning_rate": 1.8826822952160284e-05, + "loss": 2.5439163208007813, + "step": 22020 + }, + { + "epoch": 0.059997277800462774, + "grad_norm": 2.2119131088256836, + "learning_rate": 1.882573258825133e-05, + "loss": 2.5547388076782225, + "step": 22040 + }, + { + "epoch": 0.060051721791207296, + "grad_norm": 2.981874942779541, + "learning_rate": 1.8824642224342376e-05, + "loss": 2.5420085906982424, + "step": 22060 + }, + { + "epoch": 0.06010616578195182, + "grad_norm": 2.148658275604248, + "learning_rate": 1.882355186043342e-05, + "loss": 2.629714775085449, + "step": 22080 + }, + { + "epoch": 0.06016060977269634, + "grad_norm": 2.203197956085205, + "learning_rate": 1.8822461496524467e-05, + "loss": 2.5939327239990235, + "step": 22100 + }, + { + "epoch": 0.060215053763440864, + "grad_norm": 2.319988250732422, + "learning_rate": 1.882137113261551e-05, + "loss": 2.5300607681274414, + "step": 22120 + }, + { + "epoch": 0.06026949775418538, + "grad_norm": 2.0819380283355713, + "learning_rate": 1.882028076870656e-05, + "loss": 2.5556196212768554, + "step": 22140 + }, + { + "epoch": 0.0603239417449299, + "grad_norm": 2.2023189067840576, + "learning_rate": 1.88191904047976e-05, + "loss": 2.5943506240844725, + "step": 22160 + }, + { + "epoch": 0.060378385735674424, + "grad_norm": 2.2636468410491943, + "learning_rate": 1.8818100040888647e-05, + "loss": 2.571489143371582, + "step": 22180 + }, + { + "epoch": 0.060432829726418946, + "grad_norm": 2.1814188957214355, + "learning_rate": 1.8817009676979693e-05, + "loss": 2.5851165771484377, + "step": 22200 + }, + { + "epoch": 0.06048727371716347, + "grad_norm": 2.1284890174865723, + "learning_rate": 1.881591931307074e-05, + "loss": 2.631435012817383, + "step": 22220 + }, + { + "epoch": 0.06054171770790799, + "grad_norm": 2.0760672092437744, + "learning_rate": 1.8814828949161785e-05, + "loss": 2.59808349609375, + "step": 22240 + }, + { + "epoch": 0.06059616169865251, + "grad_norm": 2.1657984256744385, + "learning_rate": 1.881373858525283e-05, + "loss": 2.6034351348876954, + "step": 22260 + }, + { + "epoch": 0.060650605689397036, + "grad_norm": 2.1095216274261475, + "learning_rate": 1.8812648221343873e-05, + "loss": 2.5702945709228517, + "step": 22280 + }, + { + "epoch": 0.06070504968014155, + "grad_norm": 2.177677631378174, + "learning_rate": 1.8811557857434922e-05, + "loss": 2.588389778137207, + "step": 22300 + }, + { + "epoch": 0.060759493670886074, + "grad_norm": 2.1385509967803955, + "learning_rate": 1.8810467493525965e-05, + "loss": 2.59588737487793, + "step": 22320 + }, + { + "epoch": 0.060813937661630596, + "grad_norm": 2.198383092880249, + "learning_rate": 1.880937712961701e-05, + "loss": 2.5626379013061524, + "step": 22340 + }, + { + "epoch": 0.06086838165237512, + "grad_norm": 2.2024600505828857, + "learning_rate": 1.8808286765708056e-05, + "loss": 2.5634235382080077, + "step": 22360 + }, + { + "epoch": 0.06092282564311964, + "grad_norm": 2.273352861404419, + "learning_rate": 1.8807196401799102e-05, + "loss": 2.551395034790039, + "step": 22380 + }, + { + "epoch": 0.06097726963386416, + "grad_norm": 2.177323818206787, + "learning_rate": 1.8806106037890148e-05, + "loss": 2.5620115280151365, + "step": 22400 + }, + { + "epoch": 0.061031713624608686, + "grad_norm": 2.0506210327148438, + "learning_rate": 1.8805015673981194e-05, + "loss": 2.569753074645996, + "step": 22420 + }, + { + "epoch": 0.06108615761535321, + "grad_norm": 2.2280776500701904, + "learning_rate": 1.8803925310072236e-05, + "loss": 2.569076156616211, + "step": 22440 + }, + { + "epoch": 0.06114060160609773, + "grad_norm": 2.5864806175231934, + "learning_rate": 1.8802834946163285e-05, + "loss": 2.590236282348633, + "step": 22460 + }, + { + "epoch": 0.061195045596842246, + "grad_norm": 2.2058393955230713, + "learning_rate": 1.8801744582254328e-05, + "loss": 2.5658452987670897, + "step": 22480 + }, + { + "epoch": 0.06124948958758677, + "grad_norm": 2.141923189163208, + "learning_rate": 1.8800654218345373e-05, + "loss": 2.586460494995117, + "step": 22500 + }, + { + "epoch": 0.06130393357833129, + "grad_norm": 2.078402042388916, + "learning_rate": 1.879956385443642e-05, + "loss": 2.523159980773926, + "step": 22520 + }, + { + "epoch": 0.06135837756907581, + "grad_norm": 2.170888900756836, + "learning_rate": 1.8798473490527465e-05, + "loss": 2.580377769470215, + "step": 22540 + }, + { + "epoch": 0.061412821559820335, + "grad_norm": 2.278935194015503, + "learning_rate": 1.879738312661851e-05, + "loss": 2.5533985137939452, + "step": 22560 + }, + { + "epoch": 0.06146726555056486, + "grad_norm": 2.2832202911376953, + "learning_rate": 1.8796292762709557e-05, + "loss": 2.5727277755737306, + "step": 22580 + }, + { + "epoch": 0.06152170954130938, + "grad_norm": 2.1691935062408447, + "learning_rate": 1.87952023988006e-05, + "loss": 2.567647361755371, + "step": 22600 + }, + { + "epoch": 0.0615761535320539, + "grad_norm": 2.146658182144165, + "learning_rate": 1.8794112034891648e-05, + "loss": 2.5828680038452148, + "step": 22620 + }, + { + "epoch": 0.06163059752279842, + "grad_norm": 2.105332136154175, + "learning_rate": 1.879302167098269e-05, + "loss": 2.5639217376708983, + "step": 22640 + }, + { + "epoch": 0.06168504151354294, + "grad_norm": 2.1826860904693604, + "learning_rate": 1.8791931307073737e-05, + "loss": 2.5718088150024414, + "step": 22660 + }, + { + "epoch": 0.06173948550428746, + "grad_norm": 2.1564671993255615, + "learning_rate": 1.8790840943164782e-05, + "loss": 2.5833263397216797, + "step": 22680 + }, + { + "epoch": 0.061793929495031985, + "grad_norm": 2.2655069828033447, + "learning_rate": 1.8789750579255828e-05, + "loss": 2.5780195236206054, + "step": 22700 + }, + { + "epoch": 0.06184837348577651, + "grad_norm": 2.292395830154419, + "learning_rate": 1.8788660215346874e-05, + "loss": 2.5720510482788086, + "step": 22720 + }, + { + "epoch": 0.06190281747652103, + "grad_norm": 2.161766529083252, + "learning_rate": 1.878756985143792e-05, + "loss": 2.573247528076172, + "step": 22740 + }, + { + "epoch": 0.06195726146726555, + "grad_norm": 2.272834062576294, + "learning_rate": 1.8786479487528962e-05, + "loss": 2.5684577941894533, + "step": 22760 + }, + { + "epoch": 0.062011705458010075, + "grad_norm": 2.1289777755737305, + "learning_rate": 1.878538912362001e-05, + "loss": 2.572998046875, + "step": 22780 + }, + { + "epoch": 0.0620661494487546, + "grad_norm": 2.0414977073669434, + "learning_rate": 1.8784298759711054e-05, + "loss": 2.5560550689697266, + "step": 22800 + }, + { + "epoch": 0.06212059343949911, + "grad_norm": 2.1897387504577637, + "learning_rate": 1.87832083958021e-05, + "loss": 2.5544815063476562, + "step": 22820 + }, + { + "epoch": 0.062175037430243635, + "grad_norm": 2.4934940338134766, + "learning_rate": 1.8782118031893145e-05, + "loss": 2.5756114959716796, + "step": 22840 + }, + { + "epoch": 0.06222948142098816, + "grad_norm": 2.291771650314331, + "learning_rate": 1.878102766798419e-05, + "loss": 2.5481761932373046, + "step": 22860 + }, + { + "epoch": 0.06228392541173268, + "grad_norm": 2.429814338684082, + "learning_rate": 1.8779937304075237e-05, + "loss": 2.597581672668457, + "step": 22880 + }, + { + "epoch": 0.0623383694024772, + "grad_norm": 2.1260409355163574, + "learning_rate": 1.8778846940166283e-05, + "loss": 2.5894983291625975, + "step": 22900 + }, + { + "epoch": 0.062392813393221724, + "grad_norm": 3.3389828205108643, + "learning_rate": 1.8777756576257325e-05, + "loss": 2.5533321380615233, + "step": 22920 + }, + { + "epoch": 0.06244725738396625, + "grad_norm": 2.143951654434204, + "learning_rate": 1.8776666212348374e-05, + "loss": 2.5563636779785157, + "step": 22940 + }, + { + "epoch": 0.06250170137471077, + "grad_norm": 2.3694658279418945, + "learning_rate": 1.8775575848439417e-05, + "loss": 2.5689374923706056, + "step": 22960 + }, + { + "epoch": 0.06255614536545528, + "grad_norm": 2.2515852451324463, + "learning_rate": 1.8774485484530463e-05, + "loss": 2.5661096572875977, + "step": 22980 + }, + { + "epoch": 0.06261058935619981, + "grad_norm": 2.131169080734253, + "learning_rate": 1.877339512062151e-05, + "loss": 2.5409530639648437, + "step": 23000 + }, + { + "epoch": 0.06266503334694433, + "grad_norm": 2.2845702171325684, + "learning_rate": 1.8772304756712554e-05, + "loss": 2.568329429626465, + "step": 23020 + }, + { + "epoch": 0.06271947733768886, + "grad_norm": 2.118713855743408, + "learning_rate": 1.87712143928036e-05, + "loss": 2.552939605712891, + "step": 23040 + }, + { + "epoch": 0.06277392132843337, + "grad_norm": 2.1763381958007812, + "learning_rate": 1.8770124028894646e-05, + "loss": 2.561171531677246, + "step": 23060 + }, + { + "epoch": 0.06282836531917789, + "grad_norm": 2.1403985023498535, + "learning_rate": 1.876903366498569e-05, + "loss": 2.5584192276000977, + "step": 23080 + }, + { + "epoch": 0.06288280930992242, + "grad_norm": 2.3698768615722656, + "learning_rate": 1.8767943301076738e-05, + "loss": 2.5499744415283203, + "step": 23100 + }, + { + "epoch": 0.06293725330066693, + "grad_norm": 2.689005136489868, + "learning_rate": 1.876685293716778e-05, + "loss": 2.545399284362793, + "step": 23120 + }, + { + "epoch": 0.06299169729141146, + "grad_norm": 2.1375818252563477, + "learning_rate": 1.876576257325883e-05, + "loss": 2.568634796142578, + "step": 23140 + }, + { + "epoch": 0.06304614128215598, + "grad_norm": 2.192553758621216, + "learning_rate": 1.876467220934987e-05, + "loss": 2.5794105529785156, + "step": 23160 + }, + { + "epoch": 0.06310058527290051, + "grad_norm": 2.131232261657715, + "learning_rate": 1.8763581845440917e-05, + "loss": 2.5475910186767576, + "step": 23180 + }, + { + "epoch": 0.06315502926364502, + "grad_norm": 2.4473981857299805, + "learning_rate": 1.8762491481531963e-05, + "loss": 2.6115442276000977, + "step": 23200 + }, + { + "epoch": 0.06320947325438955, + "grad_norm": 2.1794073581695557, + "learning_rate": 1.876140111762301e-05, + "loss": 2.5825204849243164, + "step": 23220 + }, + { + "epoch": 0.06326391724513407, + "grad_norm": 2.223484516143799, + "learning_rate": 1.876031075371405e-05, + "loss": 2.5845890045166016, + "step": 23240 + }, + { + "epoch": 0.06331836123587858, + "grad_norm": 2.1942636966705322, + "learning_rate": 1.87592203898051e-05, + "loss": 2.563599395751953, + "step": 23260 + }, + { + "epoch": 0.06337280522662311, + "grad_norm": 2.2485973834991455, + "learning_rate": 1.8758130025896143e-05, + "loss": 2.5810115814208983, + "step": 23280 + }, + { + "epoch": 0.06342724921736763, + "grad_norm": 2.5409111976623535, + "learning_rate": 1.8757039661987192e-05, + "loss": 2.5442298889160155, + "step": 23300 + }, + { + "epoch": 0.06348169320811216, + "grad_norm": 2.1031577587127686, + "learning_rate": 1.8755949298078235e-05, + "loss": 2.5925378799438477, + "step": 23320 + }, + { + "epoch": 0.06353613719885667, + "grad_norm": 2.28236985206604, + "learning_rate": 1.875485893416928e-05, + "loss": 2.56027774810791, + "step": 23340 + }, + { + "epoch": 0.0635905811896012, + "grad_norm": 2.0485305786132812, + "learning_rate": 1.8753768570260326e-05, + "loss": 2.5546380996704103, + "step": 23360 + }, + { + "epoch": 0.06364502518034572, + "grad_norm": 2.023764133453369, + "learning_rate": 1.8752678206351372e-05, + "loss": 2.5651931762695312, + "step": 23380 + }, + { + "epoch": 0.06369946917109025, + "grad_norm": 2.120565176010132, + "learning_rate": 1.8751587842442415e-05, + "loss": 2.5452362060546876, + "step": 23400 + }, + { + "epoch": 0.06375391316183476, + "grad_norm": 2.165083169937134, + "learning_rate": 1.8750497478533464e-05, + "loss": 2.5577518463134767, + "step": 23420 + }, + { + "epoch": 0.06380835715257928, + "grad_norm": 2.852407932281494, + "learning_rate": 1.8749407114624506e-05, + "loss": 2.561536979675293, + "step": 23440 + }, + { + "epoch": 0.06386280114332381, + "grad_norm": 2.113361120223999, + "learning_rate": 1.8748316750715555e-05, + "loss": 2.576074981689453, + "step": 23460 + }, + { + "epoch": 0.06391724513406832, + "grad_norm": 2.1300981044769287, + "learning_rate": 1.8747226386806598e-05, + "loss": 2.547165107727051, + "step": 23480 + }, + { + "epoch": 0.06397168912481285, + "grad_norm": 2.1316051483154297, + "learning_rate": 1.8746136022897644e-05, + "loss": 2.5659589767456055, + "step": 23500 + }, + { + "epoch": 0.06402613311555737, + "grad_norm": 2.093492269515991, + "learning_rate": 1.874504565898869e-05, + "loss": 2.592864227294922, + "step": 23520 + }, + { + "epoch": 0.0640805771063019, + "grad_norm": 2.1390902996063232, + "learning_rate": 1.8743955295079735e-05, + "loss": 2.549394416809082, + "step": 23540 + }, + { + "epoch": 0.06413502109704641, + "grad_norm": 2.1320884227752686, + "learning_rate": 1.8742864931170778e-05, + "loss": 2.575642204284668, + "step": 23560 + }, + { + "epoch": 0.06418946508779093, + "grad_norm": 2.199669122695923, + "learning_rate": 1.8741774567261827e-05, + "loss": 2.5604740142822267, + "step": 23580 + }, + { + "epoch": 0.06424390907853546, + "grad_norm": 2.1227006912231445, + "learning_rate": 1.874068420335287e-05, + "loss": 2.5839696884155274, + "step": 23600 + }, + { + "epoch": 0.06429835306927997, + "grad_norm": 2.2663469314575195, + "learning_rate": 1.873959383944392e-05, + "loss": 2.5631561279296875, + "step": 23620 + }, + { + "epoch": 0.0643527970600245, + "grad_norm": 2.2390334606170654, + "learning_rate": 1.873850347553496e-05, + "loss": 2.5747037887573243, + "step": 23640 + }, + { + "epoch": 0.06440724105076902, + "grad_norm": 2.164612293243408, + "learning_rate": 1.8737413111626007e-05, + "loss": 2.560711669921875, + "step": 23660 + }, + { + "epoch": 0.06446168504151355, + "grad_norm": 2.1114633083343506, + "learning_rate": 1.8736322747717052e-05, + "loss": 2.572492218017578, + "step": 23680 + }, + { + "epoch": 0.06451612903225806, + "grad_norm": 2.115128993988037, + "learning_rate": 1.8735232383808098e-05, + "loss": 2.5556116104125977, + "step": 23700 + }, + { + "epoch": 0.06457057302300259, + "grad_norm": 2.3761909008026123, + "learning_rate": 1.873414201989914e-05, + "loss": 2.5782133102416993, + "step": 23720 + }, + { + "epoch": 0.06462501701374711, + "grad_norm": 2.2249057292938232, + "learning_rate": 1.873305165599019e-05, + "loss": 2.550052452087402, + "step": 23740 + }, + { + "epoch": 0.06467946100449162, + "grad_norm": 2.0904173851013184, + "learning_rate": 1.8731961292081232e-05, + "loss": 2.586536979675293, + "step": 23760 + }, + { + "epoch": 0.06473390499523615, + "grad_norm": 2.1248302459716797, + "learning_rate": 1.873087092817228e-05, + "loss": 2.5714605331420897, + "step": 23780 + }, + { + "epoch": 0.06478834898598067, + "grad_norm": 2.079730749130249, + "learning_rate": 1.8729780564263324e-05, + "loss": 2.565229034423828, + "step": 23800 + }, + { + "epoch": 0.0648427929767252, + "grad_norm": 2.2104172706604004, + "learning_rate": 1.872869020035437e-05, + "loss": 2.5754175186157227, + "step": 23820 + }, + { + "epoch": 0.06489723696746971, + "grad_norm": 2.1484272480010986, + "learning_rate": 1.8727599836445416e-05, + "loss": 2.5866348266601564, + "step": 23840 + }, + { + "epoch": 0.06495168095821424, + "grad_norm": 2.1394965648651123, + "learning_rate": 1.872650947253646e-05, + "loss": 2.5744440078735353, + "step": 23860 + }, + { + "epoch": 0.06500612494895876, + "grad_norm": 2.5086209774017334, + "learning_rate": 1.8725419108627504e-05, + "loss": 2.5744321823120115, + "step": 23880 + }, + { + "epoch": 0.06506056893970329, + "grad_norm": 2.1272196769714355, + "learning_rate": 1.8724328744718553e-05, + "loss": 2.5447437286376955, + "step": 23900 + }, + { + "epoch": 0.0651150129304478, + "grad_norm": 2.109093427658081, + "learning_rate": 1.8723238380809595e-05, + "loss": 2.5590049743652346, + "step": 23920 + }, + { + "epoch": 0.06516945692119232, + "grad_norm": 2.399324417114258, + "learning_rate": 1.8722148016900645e-05, + "loss": 2.546399688720703, + "step": 23940 + }, + { + "epoch": 0.06522390091193685, + "grad_norm": 2.1963305473327637, + "learning_rate": 1.8721057652991687e-05, + "loss": 2.5507646560668946, + "step": 23960 + }, + { + "epoch": 0.06527834490268136, + "grad_norm": 2.183244466781616, + "learning_rate": 1.8719967289082733e-05, + "loss": 2.5669061660766603, + "step": 23980 + }, + { + "epoch": 0.06533278889342589, + "grad_norm": 2.282819986343384, + "learning_rate": 1.871887692517378e-05, + "loss": 2.5618097305297853, + "step": 24000 + }, + { + "epoch": 0.06538723288417041, + "grad_norm": 2.2068307399749756, + "learning_rate": 1.8717786561264824e-05, + "loss": 2.581002044677734, + "step": 24020 + }, + { + "epoch": 0.06544167687491494, + "grad_norm": 2.1895618438720703, + "learning_rate": 1.8716696197355867e-05, + "loss": 2.566451835632324, + "step": 24040 + }, + { + "epoch": 0.06549612086565945, + "grad_norm": 2.5391201972961426, + "learning_rate": 1.8715605833446916e-05, + "loss": 2.5554697036743166, + "step": 24060 + }, + { + "epoch": 0.06555056485640398, + "grad_norm": 2.318371295928955, + "learning_rate": 1.871451546953796e-05, + "loss": 2.575593185424805, + "step": 24080 + }, + { + "epoch": 0.0656050088471485, + "grad_norm": 2.1644575595855713, + "learning_rate": 1.8713425105629008e-05, + "loss": 2.5636125564575196, + "step": 24100 + }, + { + "epoch": 0.06565945283789301, + "grad_norm": 2.1629581451416016, + "learning_rate": 1.871233474172005e-05, + "loss": 2.5557382583618162, + "step": 24120 + }, + { + "epoch": 0.06571389682863754, + "grad_norm": 2.0676839351654053, + "learning_rate": 1.8711244377811096e-05, + "loss": 2.5683584213256836, + "step": 24140 + }, + { + "epoch": 0.06576834081938206, + "grad_norm": 2.1639716625213623, + "learning_rate": 1.8710154013902142e-05, + "loss": 2.5551694869995116, + "step": 24160 + }, + { + "epoch": 0.06582278481012659, + "grad_norm": 2.286616563796997, + "learning_rate": 1.8709063649993188e-05, + "loss": 2.545287322998047, + "step": 24180 + }, + { + "epoch": 0.0658772288008711, + "grad_norm": 2.1462454795837402, + "learning_rate": 1.870797328608423e-05, + "loss": 2.5762720108032227, + "step": 24200 + }, + { + "epoch": 0.06593167279161563, + "grad_norm": 2.203075647354126, + "learning_rate": 1.870688292217528e-05, + "loss": 2.554981994628906, + "step": 24220 + }, + { + "epoch": 0.06598611678236015, + "grad_norm": 2.0736522674560547, + "learning_rate": 1.870579255826632e-05, + "loss": 2.524469184875488, + "step": 24240 + }, + { + "epoch": 0.06604056077310466, + "grad_norm": 2.153651237487793, + "learning_rate": 1.870470219435737e-05, + "loss": 2.6056039810180662, + "step": 24260 + }, + { + "epoch": 0.06609500476384919, + "grad_norm": 2.3104662895202637, + "learning_rate": 1.8703611830448413e-05, + "loss": 2.5521926879882812, + "step": 24280 + }, + { + "epoch": 0.0661494487545937, + "grad_norm": 2.5039281845092773, + "learning_rate": 1.870252146653946e-05, + "loss": 2.584322929382324, + "step": 24300 + }, + { + "epoch": 0.06620389274533824, + "grad_norm": 2.2385413646698, + "learning_rate": 1.8701431102630505e-05, + "loss": 2.5683053970336913, + "step": 24320 + }, + { + "epoch": 0.06625833673608275, + "grad_norm": 2.3483388423919678, + "learning_rate": 1.870034073872155e-05, + "loss": 2.5783920288085938, + "step": 24340 + }, + { + "epoch": 0.06631278072682728, + "grad_norm": 2.143033742904663, + "learning_rate": 1.8699250374812593e-05, + "loss": 2.604951858520508, + "step": 24360 + }, + { + "epoch": 0.0663672247175718, + "grad_norm": 2.2147023677825928, + "learning_rate": 1.8698160010903642e-05, + "loss": 2.5930288314819334, + "step": 24380 + }, + { + "epoch": 0.06642166870831633, + "grad_norm": 2.133896827697754, + "learning_rate": 1.8697069646994685e-05, + "loss": 2.594439697265625, + "step": 24400 + }, + { + "epoch": 0.06647611269906084, + "grad_norm": 2.0709919929504395, + "learning_rate": 1.8695979283085734e-05, + "loss": 2.5585702896118163, + "step": 24420 + }, + { + "epoch": 0.06653055668980536, + "grad_norm": 2.2579128742218018, + "learning_rate": 1.8694888919176776e-05, + "loss": 2.564336585998535, + "step": 24440 + }, + { + "epoch": 0.06658500068054989, + "grad_norm": 2.1465413570404053, + "learning_rate": 1.8693798555267822e-05, + "loss": 2.5723541259765623, + "step": 24460 + }, + { + "epoch": 0.0666394446712944, + "grad_norm": 2.184492826461792, + "learning_rate": 1.8692708191358868e-05, + "loss": 2.5815719604492187, + "step": 24480 + }, + { + "epoch": 0.06669388866203893, + "grad_norm": 2.0619094371795654, + "learning_rate": 1.8691617827449914e-05, + "loss": 2.536931800842285, + "step": 24500 + }, + { + "epoch": 0.06674833265278345, + "grad_norm": 2.166855812072754, + "learning_rate": 1.8690527463540956e-05, + "loss": 2.562894821166992, + "step": 24520 + }, + { + "epoch": 0.06680277664352798, + "grad_norm": 2.327969789505005, + "learning_rate": 1.8689437099632005e-05, + "loss": 2.55719051361084, + "step": 24540 + }, + { + "epoch": 0.06685722063427249, + "grad_norm": 2.0816290378570557, + "learning_rate": 1.8688346735723048e-05, + "loss": 2.5732343673706053, + "step": 24560 + }, + { + "epoch": 0.06691166462501702, + "grad_norm": 2.1310033798217773, + "learning_rate": 1.8687256371814097e-05, + "loss": 2.564645767211914, + "step": 24580 + }, + { + "epoch": 0.06696610861576154, + "grad_norm": 2.182833433151245, + "learning_rate": 1.868616600790514e-05, + "loss": 2.5692848205566405, + "step": 24600 + }, + { + "epoch": 0.06702055260650605, + "grad_norm": 2.1325323581695557, + "learning_rate": 1.8685075643996185e-05, + "loss": 2.5908021926879883, + "step": 24620 + }, + { + "epoch": 0.06707499659725058, + "grad_norm": 2.1355998516082764, + "learning_rate": 1.868398528008723e-05, + "loss": 2.5719930648803713, + "step": 24640 + }, + { + "epoch": 0.0671294405879951, + "grad_norm": 2.204057455062866, + "learning_rate": 1.8682894916178277e-05, + "loss": 2.575278091430664, + "step": 24660 + }, + { + "epoch": 0.06718388457873962, + "grad_norm": 2.2531814575195312, + "learning_rate": 1.868180455226932e-05, + "loss": 2.586996078491211, + "step": 24680 + }, + { + "epoch": 0.06723832856948414, + "grad_norm": 2.177381753921509, + "learning_rate": 1.868071418836037e-05, + "loss": 2.574261474609375, + "step": 24700 + }, + { + "epoch": 0.06729277256022867, + "grad_norm": 2.083916425704956, + "learning_rate": 1.867962382445141e-05, + "loss": 2.571217155456543, + "step": 24720 + }, + { + "epoch": 0.06734721655097319, + "grad_norm": 2.233010768890381, + "learning_rate": 1.867853346054246e-05, + "loss": 2.571368598937988, + "step": 24740 + }, + { + "epoch": 0.06740166054171771, + "grad_norm": 2.25661039352417, + "learning_rate": 1.8677443096633502e-05, + "loss": 2.534840202331543, + "step": 24760 + }, + { + "epoch": 0.06745610453246223, + "grad_norm": 2.220437526702881, + "learning_rate": 1.8676352732724548e-05, + "loss": 2.5927663803100587, + "step": 24780 + }, + { + "epoch": 0.06751054852320675, + "grad_norm": 2.619337558746338, + "learning_rate": 1.8675262368815594e-05, + "loss": 2.6027183532714844, + "step": 24800 + }, + { + "epoch": 0.06756499251395127, + "grad_norm": 2.3588523864746094, + "learning_rate": 1.867417200490664e-05, + "loss": 2.5803979873657226, + "step": 24820 + }, + { + "epoch": 0.06761943650469579, + "grad_norm": 2.3245444297790527, + "learning_rate": 1.8673081640997686e-05, + "loss": 2.5905950546264647, + "step": 24840 + }, + { + "epoch": 0.06767388049544032, + "grad_norm": 2.3534181118011475, + "learning_rate": 1.867199127708873e-05, + "loss": 2.584467315673828, + "step": 24860 + }, + { + "epoch": 0.06772832448618483, + "grad_norm": 2.1266238689422607, + "learning_rate": 1.8670900913179774e-05, + "loss": 2.5310600280761717, + "step": 24880 + }, + { + "epoch": 0.06778276847692936, + "grad_norm": 2.8633968830108643, + "learning_rate": 1.8669810549270823e-05, + "loss": 2.5697021484375, + "step": 24900 + }, + { + "epoch": 0.06783721246767388, + "grad_norm": 2.148674964904785, + "learning_rate": 1.8668720185361866e-05, + "loss": 2.5660301208496095, + "step": 24920 + }, + { + "epoch": 0.0678916564584184, + "grad_norm": 2.239762783050537, + "learning_rate": 1.866762982145291e-05, + "loss": 2.5729772567749025, + "step": 24940 + }, + { + "epoch": 0.06794610044916292, + "grad_norm": 2.2344536781311035, + "learning_rate": 1.8666539457543957e-05, + "loss": 2.555094909667969, + "step": 24960 + }, + { + "epoch": 0.06800054443990744, + "grad_norm": 2.32080340385437, + "learning_rate": 1.8665449093635003e-05, + "loss": 2.5628963470458985, + "step": 24980 + }, + { + "epoch": 0.06805498843065197, + "grad_norm": 2.1948790550231934, + "learning_rate": 1.866435872972605e-05, + "loss": 2.5665483474731445, + "step": 25000 + }, + { + "epoch": 0.06810943242139648, + "grad_norm": 2.193291664123535, + "learning_rate": 1.8663268365817095e-05, + "loss": 2.581097412109375, + "step": 25020 + }, + { + "epoch": 0.06816387641214101, + "grad_norm": 2.23465633392334, + "learning_rate": 1.8662178001908137e-05, + "loss": 2.5659902572631834, + "step": 25040 + }, + { + "epoch": 0.06821832040288553, + "grad_norm": 2.549323558807373, + "learning_rate": 1.8661087637999186e-05, + "loss": 2.5900379180908204, + "step": 25060 + }, + { + "epoch": 0.06827276439363006, + "grad_norm": 2.298152446746826, + "learning_rate": 1.865999727409023e-05, + "loss": 2.565587043762207, + "step": 25080 + }, + { + "epoch": 0.06832720838437457, + "grad_norm": 2.1283485889434814, + "learning_rate": 1.8658906910181274e-05, + "loss": 2.5579082489013674, + "step": 25100 + }, + { + "epoch": 0.06838165237511909, + "grad_norm": 2.2388651371002197, + "learning_rate": 1.865781654627232e-05, + "loss": 2.5593708038330076, + "step": 25120 + }, + { + "epoch": 0.06843609636586362, + "grad_norm": 2.2037346363067627, + "learning_rate": 1.8656726182363366e-05, + "loss": 2.562400054931641, + "step": 25140 + }, + { + "epoch": 0.06849054035660813, + "grad_norm": 2.2278993129730225, + "learning_rate": 1.8655635818454412e-05, + "loss": 2.5633876800537108, + "step": 25160 + }, + { + "epoch": 0.06854498434735266, + "grad_norm": 2.0880589485168457, + "learning_rate": 1.8654545454545458e-05, + "loss": 2.551900863647461, + "step": 25180 + }, + { + "epoch": 0.06859942833809718, + "grad_norm": 2.2957284450531006, + "learning_rate": 1.86534550906365e-05, + "loss": 2.551812171936035, + "step": 25200 + }, + { + "epoch": 0.06865387232884171, + "grad_norm": 2.1042160987854004, + "learning_rate": 1.865236472672755e-05, + "loss": 2.5749744415283202, + "step": 25220 + }, + { + "epoch": 0.06870831631958622, + "grad_norm": 2.121105909347534, + "learning_rate": 1.8651274362818592e-05, + "loss": 2.539056968688965, + "step": 25240 + }, + { + "epoch": 0.06876276031033075, + "grad_norm": 2.2286345958709717, + "learning_rate": 1.8650183998909638e-05, + "loss": 2.5979190826416017, + "step": 25260 + }, + { + "epoch": 0.06881720430107527, + "grad_norm": 2.0977368354797363, + "learning_rate": 1.8649093635000683e-05, + "loss": 2.581761932373047, + "step": 25280 + }, + { + "epoch": 0.06887164829181978, + "grad_norm": 2.435360908508301, + "learning_rate": 1.864800327109173e-05, + "loss": 2.5757369995117188, + "step": 25300 + }, + { + "epoch": 0.06892609228256431, + "grad_norm": 2.3741796016693115, + "learning_rate": 1.8646912907182775e-05, + "loss": 2.5701757431030274, + "step": 25320 + }, + { + "epoch": 0.06898053627330883, + "grad_norm": 2.105079174041748, + "learning_rate": 1.864582254327382e-05, + "loss": 2.5540632247924804, + "step": 25340 + }, + { + "epoch": 0.06903498026405336, + "grad_norm": 2.132053852081299, + "learning_rate": 1.8644732179364863e-05, + "loss": 2.5351499557495116, + "step": 25360 + }, + { + "epoch": 0.06908942425479787, + "grad_norm": 2.1040751934051514, + "learning_rate": 1.864364181545591e-05, + "loss": 2.5746910095214846, + "step": 25380 + }, + { + "epoch": 0.0691438682455424, + "grad_norm": 2.0860862731933594, + "learning_rate": 1.8642551451546955e-05, + "loss": 2.5837985992431642, + "step": 25400 + }, + { + "epoch": 0.06919831223628692, + "grad_norm": 3.7673866748809814, + "learning_rate": 1.8641461087638e-05, + "loss": 2.569639778137207, + "step": 25420 + }, + { + "epoch": 0.06925275622703145, + "grad_norm": 2.641902446746826, + "learning_rate": 1.8640370723729046e-05, + "loss": 2.561241340637207, + "step": 25440 + }, + { + "epoch": 0.06930720021777596, + "grad_norm": 2.1881258487701416, + "learning_rate": 1.8639280359820092e-05, + "loss": 2.585240364074707, + "step": 25460 + }, + { + "epoch": 0.06936164420852048, + "grad_norm": 2.3275671005249023, + "learning_rate": 1.8638189995911138e-05, + "loss": 2.579400634765625, + "step": 25480 + }, + { + "epoch": 0.06941608819926501, + "grad_norm": 2.1358842849731445, + "learning_rate": 1.8637099632002184e-05, + "loss": 2.545877456665039, + "step": 25500 + }, + { + "epoch": 0.06947053219000952, + "grad_norm": 2.1681110858917236, + "learning_rate": 1.8636009268093226e-05, + "loss": 2.5852741241455077, + "step": 25520 + }, + { + "epoch": 0.06952497618075405, + "grad_norm": 2.049018144607544, + "learning_rate": 1.8634918904184272e-05, + "loss": 2.54559211730957, + "step": 25540 + }, + { + "epoch": 0.06957942017149857, + "grad_norm": 2.216374397277832, + "learning_rate": 1.8633828540275318e-05, + "loss": 2.5515933990478517, + "step": 25560 + }, + { + "epoch": 0.0696338641622431, + "grad_norm": 2.3112564086914062, + "learning_rate": 1.8632738176366364e-05, + "loss": 2.5589773178100588, + "step": 25580 + }, + { + "epoch": 0.06968830815298761, + "grad_norm": 2.0679993629455566, + "learning_rate": 1.863164781245741e-05, + "loss": 2.5229564666748048, + "step": 25600 + }, + { + "epoch": 0.06974275214373214, + "grad_norm": 2.192862033843994, + "learning_rate": 1.8630557448548455e-05, + "loss": 2.528089714050293, + "step": 25620 + }, + { + "epoch": 0.06979719613447666, + "grad_norm": 2.1566121578216553, + "learning_rate": 1.86294670846395e-05, + "loss": 2.5649715423583985, + "step": 25640 + }, + { + "epoch": 0.06985164012522117, + "grad_norm": 2.146747589111328, + "learning_rate": 1.8628376720730547e-05, + "loss": 2.5535797119140624, + "step": 25660 + }, + { + "epoch": 0.0699060841159657, + "grad_norm": 2.1486175060272217, + "learning_rate": 1.862728635682159e-05, + "loss": 2.551877975463867, + "step": 25680 + }, + { + "epoch": 0.06996052810671022, + "grad_norm": 2.432758092880249, + "learning_rate": 1.8626195992912635e-05, + "loss": 2.575690269470215, + "step": 25700 + }, + { + "epoch": 0.07001497209745475, + "grad_norm": 2.3472673892974854, + "learning_rate": 1.862510562900368e-05, + "loss": 2.5339675903320313, + "step": 25720 + }, + { + "epoch": 0.07006941608819926, + "grad_norm": 2.171450138092041, + "learning_rate": 1.8624015265094727e-05, + "loss": 2.557364273071289, + "step": 25740 + }, + { + "epoch": 0.07012386007894379, + "grad_norm": 2.2930681705474854, + "learning_rate": 1.8622924901185773e-05, + "loss": 2.561827278137207, + "step": 25760 + }, + { + "epoch": 0.07017830406968831, + "grad_norm": 2.2271604537963867, + "learning_rate": 1.862183453727682e-05, + "loss": 2.547271728515625, + "step": 25780 + }, + { + "epoch": 0.07023274806043282, + "grad_norm": 2.1728322505950928, + "learning_rate": 1.8620744173367864e-05, + "loss": 2.573586273193359, + "step": 25800 + }, + { + "epoch": 0.07028719205117735, + "grad_norm": 2.5522689819335938, + "learning_rate": 1.861965380945891e-05, + "loss": 2.521917152404785, + "step": 25820 + }, + { + "epoch": 0.07034163604192187, + "grad_norm": 2.177201747894287, + "learning_rate": 1.8618563445549952e-05, + "loss": 2.54260139465332, + "step": 25840 + }, + { + "epoch": 0.0703960800326664, + "grad_norm": 2.3652162551879883, + "learning_rate": 1.8617473081640998e-05, + "loss": 2.555892562866211, + "step": 25860 + }, + { + "epoch": 0.07045052402341091, + "grad_norm": 2.3644940853118896, + "learning_rate": 1.8616382717732044e-05, + "loss": 2.545848846435547, + "step": 25880 + }, + { + "epoch": 0.07050496801415544, + "grad_norm": 2.1241328716278076, + "learning_rate": 1.861529235382309e-05, + "loss": 2.5265369415283203, + "step": 25900 + }, + { + "epoch": 0.07055941200489996, + "grad_norm": 2.196723461151123, + "learning_rate": 1.8614201989914136e-05, + "loss": 2.5414743423461914, + "step": 25920 + }, + { + "epoch": 0.07061385599564449, + "grad_norm": 2.219040632247925, + "learning_rate": 1.861311162600518e-05, + "loss": 2.568791961669922, + "step": 25940 + }, + { + "epoch": 0.070668299986389, + "grad_norm": 2.137838840484619, + "learning_rate": 1.8612021262096227e-05, + "loss": 2.566851997375488, + "step": 25960 + }, + { + "epoch": 0.07072274397713352, + "grad_norm": 2.262888193130493, + "learning_rate": 1.8610930898187273e-05, + "loss": 2.58046932220459, + "step": 25980 + }, + { + "epoch": 0.07077718796787805, + "grad_norm": 2.180440902709961, + "learning_rate": 1.8609840534278316e-05, + "loss": 2.5604694366455076, + "step": 26000 + }, + { + "epoch": 0.07083163195862256, + "grad_norm": 2.090799331665039, + "learning_rate": 1.860875017036936e-05, + "loss": 2.5538625717163086, + "step": 26020 + }, + { + "epoch": 0.07088607594936709, + "grad_norm": 2.152790069580078, + "learning_rate": 1.8607659806460407e-05, + "loss": 2.5794076919555664, + "step": 26040 + }, + { + "epoch": 0.07094051994011161, + "grad_norm": 2.3574483394622803, + "learning_rate": 1.8606569442551453e-05, + "loss": 2.565255546569824, + "step": 26060 + }, + { + "epoch": 0.07099496393085614, + "grad_norm": 2.119858741760254, + "learning_rate": 1.86054790786425e-05, + "loss": 2.5457786560058593, + "step": 26080 + }, + { + "epoch": 0.07104940792160065, + "grad_norm": 2.243922233581543, + "learning_rate": 1.8604388714733545e-05, + "loss": 2.5648513793945313, + "step": 26100 + }, + { + "epoch": 0.07110385191234518, + "grad_norm": 2.164456605911255, + "learning_rate": 1.860329835082459e-05, + "loss": 2.566106605529785, + "step": 26120 + }, + { + "epoch": 0.0711582959030897, + "grad_norm": 2.121318817138672, + "learning_rate": 1.8602207986915636e-05, + "loss": 2.5833269119262696, + "step": 26140 + }, + { + "epoch": 0.07121273989383421, + "grad_norm": 2.21048903465271, + "learning_rate": 1.860111762300668e-05, + "loss": 2.568721961975098, + "step": 26160 + }, + { + "epoch": 0.07126718388457874, + "grad_norm": 2.351146936416626, + "learning_rate": 1.8600027259097724e-05, + "loss": 2.592783737182617, + "step": 26180 + }, + { + "epoch": 0.07132162787532326, + "grad_norm": 2.1144115924835205, + "learning_rate": 1.859893689518877e-05, + "loss": 2.560613441467285, + "step": 26200 + }, + { + "epoch": 0.07137607186606779, + "grad_norm": 2.1317875385284424, + "learning_rate": 1.8597846531279816e-05, + "loss": 2.5517246246337892, + "step": 26220 + }, + { + "epoch": 0.0714305158568123, + "grad_norm": 2.158400535583496, + "learning_rate": 1.8596756167370862e-05, + "loss": 2.5602981567382814, + "step": 26240 + }, + { + "epoch": 0.07148495984755683, + "grad_norm": 2.1345250606536865, + "learning_rate": 1.8595665803461908e-05, + "loss": 2.5470903396606444, + "step": 26260 + }, + { + "epoch": 0.07153940383830135, + "grad_norm": 2.248173713684082, + "learning_rate": 1.8594575439552953e-05, + "loss": 2.554653549194336, + "step": 26280 + }, + { + "epoch": 0.07159384782904588, + "grad_norm": 2.4098801612854004, + "learning_rate": 1.8593485075644e-05, + "loss": 2.564887046813965, + "step": 26300 + }, + { + "epoch": 0.07164829181979039, + "grad_norm": 2.126270294189453, + "learning_rate": 1.859239471173504e-05, + "loss": 2.5639999389648436, + "step": 26320 + }, + { + "epoch": 0.0717027358105349, + "grad_norm": 2.1922264099121094, + "learning_rate": 1.8591304347826087e-05, + "loss": 2.59993839263916, + "step": 26340 + }, + { + "epoch": 0.07175717980127944, + "grad_norm": 2.573471784591675, + "learning_rate": 1.8590213983917133e-05, + "loss": 2.5586647033691405, + "step": 26360 + }, + { + "epoch": 0.07181162379202395, + "grad_norm": 2.118762731552124, + "learning_rate": 1.858912362000818e-05, + "loss": 2.5492115020751953, + "step": 26380 + }, + { + "epoch": 0.07186606778276848, + "grad_norm": 2.4225499629974365, + "learning_rate": 1.8588033256099225e-05, + "loss": 2.5394561767578123, + "step": 26400 + }, + { + "epoch": 0.071920511773513, + "grad_norm": 2.1848905086517334, + "learning_rate": 1.858694289219027e-05, + "loss": 2.5601661682128904, + "step": 26420 + }, + { + "epoch": 0.07197495576425753, + "grad_norm": 2.1984994411468506, + "learning_rate": 1.8585852528281317e-05, + "loss": 2.5787937164306642, + "step": 26440 + }, + { + "epoch": 0.07202939975500204, + "grad_norm": 2.0698726177215576, + "learning_rate": 1.8584762164372362e-05, + "loss": 2.574261474609375, + "step": 26460 + }, + { + "epoch": 0.07208384374574656, + "grad_norm": 2.1333367824554443, + "learning_rate": 1.8583671800463405e-05, + "loss": 2.5461393356323243, + "step": 26480 + }, + { + "epoch": 0.07213828773649109, + "grad_norm": 2.1410534381866455, + "learning_rate": 1.858258143655445e-05, + "loss": 2.562544822692871, + "step": 26500 + }, + { + "epoch": 0.0721927317272356, + "grad_norm": 2.2006912231445312, + "learning_rate": 1.8581491072645496e-05, + "loss": 2.5682296752929688, + "step": 26520 + }, + { + "epoch": 0.07224717571798013, + "grad_norm": 2.1920015811920166, + "learning_rate": 1.8580400708736542e-05, + "loss": 2.565677261352539, + "step": 26540 + }, + { + "epoch": 0.07230161970872465, + "grad_norm": 2.676314353942871, + "learning_rate": 1.8579310344827588e-05, + "loss": 2.587734413146973, + "step": 26560 + }, + { + "epoch": 0.07235606369946918, + "grad_norm": 2.1917266845703125, + "learning_rate": 1.8578219980918634e-05, + "loss": 2.587982749938965, + "step": 26580 + }, + { + "epoch": 0.07241050769021369, + "grad_norm": 6.714179515838623, + "learning_rate": 1.857712961700968e-05, + "loss": 2.5785089492797852, + "step": 26600 + }, + { + "epoch": 0.07246495168095822, + "grad_norm": 2.5550808906555176, + "learning_rate": 1.8576039253100725e-05, + "loss": 2.5261356353759767, + "step": 26620 + }, + { + "epoch": 0.07251939567170274, + "grad_norm": 2.0656585693359375, + "learning_rate": 1.8574948889191768e-05, + "loss": 2.56317253112793, + "step": 26640 + }, + { + "epoch": 0.07257383966244725, + "grad_norm": 2.75748348236084, + "learning_rate": 1.8573858525282814e-05, + "loss": 2.5525182723999023, + "step": 26660 + }, + { + "epoch": 0.07262828365319178, + "grad_norm": 2.1378307342529297, + "learning_rate": 1.857276816137386e-05, + "loss": 2.5898950576782225, + "step": 26680 + }, + { + "epoch": 0.0726827276439363, + "grad_norm": 2.182307004928589, + "learning_rate": 1.8571677797464905e-05, + "loss": 2.598641777038574, + "step": 26700 + }, + { + "epoch": 0.07273717163468083, + "grad_norm": 2.2080914974212646, + "learning_rate": 1.857058743355595e-05, + "loss": 2.5708644866943358, + "step": 26720 + }, + { + "epoch": 0.07279161562542534, + "grad_norm": 2.1089024543762207, + "learning_rate": 1.8569497069646997e-05, + "loss": 2.5537824630737305, + "step": 26740 + }, + { + "epoch": 0.07284605961616987, + "grad_norm": 2.148224115371704, + "learning_rate": 1.8568406705738043e-05, + "loss": 2.5452592849731444, + "step": 26760 + }, + { + "epoch": 0.07290050360691439, + "grad_norm": 2.0749094486236572, + "learning_rate": 1.856731634182909e-05, + "loss": 2.5681379318237303, + "step": 26780 + }, + { + "epoch": 0.07295494759765891, + "grad_norm": 2.297745704650879, + "learning_rate": 1.856622597792013e-05, + "loss": 2.5837543487548826, + "step": 26800 + }, + { + "epoch": 0.07300939158840343, + "grad_norm": 2.211413860321045, + "learning_rate": 1.8565135614011177e-05, + "loss": 2.554019546508789, + "step": 26820 + }, + { + "epoch": 0.07306383557914795, + "grad_norm": 2.1113507747650146, + "learning_rate": 1.8564045250102223e-05, + "loss": 2.5805942535400392, + "step": 26840 + }, + { + "epoch": 0.07311827956989247, + "grad_norm": 2.3622303009033203, + "learning_rate": 1.856295488619327e-05, + "loss": 2.5455163955688476, + "step": 26860 + }, + { + "epoch": 0.07317272356063699, + "grad_norm": 2.286278486251831, + "learning_rate": 1.8561864522284314e-05, + "loss": 2.5636629104614257, + "step": 26880 + }, + { + "epoch": 0.07322716755138152, + "grad_norm": 2.0617620944976807, + "learning_rate": 1.856077415837536e-05, + "loss": 2.5705575942993164, + "step": 26900 + }, + { + "epoch": 0.07328161154212604, + "grad_norm": 2.298422336578369, + "learning_rate": 1.8559683794466406e-05, + "loss": 2.5593055725097655, + "step": 26920 + }, + { + "epoch": 0.07333605553287056, + "grad_norm": 2.084522008895874, + "learning_rate": 1.855859343055745e-05, + "loss": 2.5686771392822267, + "step": 26940 + }, + { + "epoch": 0.07339049952361508, + "grad_norm": 2.1080591678619385, + "learning_rate": 1.8557503066648494e-05, + "loss": 2.557130241394043, + "step": 26960 + }, + { + "epoch": 0.07344494351435961, + "grad_norm": 2.2613234519958496, + "learning_rate": 1.855641270273954e-05, + "loss": 2.5245643615722657, + "step": 26980 + }, + { + "epoch": 0.07349938750510412, + "grad_norm": 2.0774178504943848, + "learning_rate": 1.8555322338830586e-05, + "loss": 2.545204734802246, + "step": 27000 + }, + { + "epoch": 0.07355383149584864, + "grad_norm": 2.150423765182495, + "learning_rate": 1.855423197492163e-05, + "loss": 2.5844186782836913, + "step": 27020 + }, + { + "epoch": 0.07360827548659317, + "grad_norm": 2.9108834266662598, + "learning_rate": 1.8553141611012677e-05, + "loss": 2.6031162261962892, + "step": 27040 + }, + { + "epoch": 0.07366271947733768, + "grad_norm": 2.36840558052063, + "learning_rate": 1.8552051247103723e-05, + "loss": 2.5726139068603517, + "step": 27060 + }, + { + "epoch": 0.07371716346808221, + "grad_norm": 2.132251262664795, + "learning_rate": 1.855096088319477e-05, + "loss": 2.5848392486572265, + "step": 27080 + }, + { + "epoch": 0.07377160745882673, + "grad_norm": 2.247891902923584, + "learning_rate": 1.8549870519285815e-05, + "loss": 2.5384254455566406, + "step": 27100 + }, + { + "epoch": 0.07382605144957126, + "grad_norm": 2.2491209506988525, + "learning_rate": 1.8548780155376857e-05, + "loss": 2.561193656921387, + "step": 27120 + }, + { + "epoch": 0.07388049544031577, + "grad_norm": 2.3306663036346436, + "learning_rate": 1.8547689791467903e-05, + "loss": 2.5383129119873047, + "step": 27140 + }, + { + "epoch": 0.0739349394310603, + "grad_norm": 2.3427562713623047, + "learning_rate": 1.854659942755895e-05, + "loss": 2.5425655364990236, + "step": 27160 + }, + { + "epoch": 0.07398938342180482, + "grad_norm": 2.1300172805786133, + "learning_rate": 1.8545509063649995e-05, + "loss": 2.559390640258789, + "step": 27180 + }, + { + "epoch": 0.07404382741254933, + "grad_norm": 2.1714584827423096, + "learning_rate": 1.854441869974104e-05, + "loss": 2.524231719970703, + "step": 27200 + }, + { + "epoch": 0.07409827140329386, + "grad_norm": 2.3332502841949463, + "learning_rate": 1.8543328335832086e-05, + "loss": 2.560656929016113, + "step": 27220 + }, + { + "epoch": 0.07415271539403838, + "grad_norm": 2.1450705528259277, + "learning_rate": 1.8542237971923132e-05, + "loss": 2.5593469619750975, + "step": 27240 + }, + { + "epoch": 0.07420715938478291, + "grad_norm": 2.2014570236206055, + "learning_rate": 1.8541147608014178e-05, + "loss": 2.5812744140625, + "step": 27260 + }, + { + "epoch": 0.07426160337552742, + "grad_norm": 2.251594305038452, + "learning_rate": 1.854005724410522e-05, + "loss": 2.5630393981933595, + "step": 27280 + }, + { + "epoch": 0.07431604736627195, + "grad_norm": 2.3776307106018066, + "learning_rate": 1.8538966880196266e-05, + "loss": 2.5790319442749023, + "step": 27300 + }, + { + "epoch": 0.07437049135701647, + "grad_norm": 2.089059352874756, + "learning_rate": 1.8537876516287312e-05, + "loss": 2.5782649993896483, + "step": 27320 + }, + { + "epoch": 0.07442493534776098, + "grad_norm": 2.157175302505493, + "learning_rate": 1.8536786152378358e-05, + "loss": 2.606978416442871, + "step": 27340 + }, + { + "epoch": 0.07447937933850551, + "grad_norm": 2.3065879344940186, + "learning_rate": 1.8535695788469403e-05, + "loss": 2.5629999160766603, + "step": 27360 + }, + { + "epoch": 0.07453382332925003, + "grad_norm": 2.094588041305542, + "learning_rate": 1.853460542456045e-05, + "loss": 2.519257736206055, + "step": 27380 + }, + { + "epoch": 0.07458826731999456, + "grad_norm": 2.138270139694214, + "learning_rate": 1.8533515060651495e-05, + "loss": 2.5799095153808596, + "step": 27400 + }, + { + "epoch": 0.07464271131073907, + "grad_norm": 2.1753785610198975, + "learning_rate": 1.853242469674254e-05, + "loss": 2.5445858001708985, + "step": 27420 + }, + { + "epoch": 0.0746971553014836, + "grad_norm": 2.066654682159424, + "learning_rate": 1.8531334332833583e-05, + "loss": 2.5438302993774413, + "step": 27440 + }, + { + "epoch": 0.07475159929222812, + "grad_norm": 2.2680301666259766, + "learning_rate": 1.853024396892463e-05, + "loss": 2.569886589050293, + "step": 27460 + }, + { + "epoch": 0.07480604328297265, + "grad_norm": 2.130418539047241, + "learning_rate": 1.8529153605015675e-05, + "loss": 2.5235090255737305, + "step": 27480 + }, + { + "epoch": 0.07486048727371716, + "grad_norm": 2.3482086658477783, + "learning_rate": 1.852806324110672e-05, + "loss": 2.572101593017578, + "step": 27500 + }, + { + "epoch": 0.07491493126446168, + "grad_norm": 2.16705322265625, + "learning_rate": 1.8526972877197767e-05, + "loss": 2.559983825683594, + "step": 27520 + }, + { + "epoch": 0.07496937525520621, + "grad_norm": 2.0715978145599365, + "learning_rate": 1.8525882513288812e-05, + "loss": 2.586870002746582, + "step": 27540 + }, + { + "epoch": 0.07502381924595072, + "grad_norm": 2.099414110183716, + "learning_rate": 1.8524792149379858e-05, + "loss": 2.5676816940307616, + "step": 27560 + }, + { + "epoch": 0.07507826323669525, + "grad_norm": 2.2181270122528076, + "learning_rate": 1.85237017854709e-05, + "loss": 2.5457231521606447, + "step": 27580 + }, + { + "epoch": 0.07513270722743977, + "grad_norm": 2.14526104927063, + "learning_rate": 1.8522611421561946e-05, + "loss": 2.5814653396606446, + "step": 27600 + }, + { + "epoch": 0.0751871512181843, + "grad_norm": 2.174107789993286, + "learning_rate": 1.8521521057652992e-05, + "loss": 2.55709285736084, + "step": 27620 + }, + { + "epoch": 0.07524159520892881, + "grad_norm": 2.106977701187134, + "learning_rate": 1.8520430693744038e-05, + "loss": 2.5244279861450196, + "step": 27640 + }, + { + "epoch": 0.07529603919967334, + "grad_norm": 2.1586313247680664, + "learning_rate": 1.8519340329835084e-05, + "loss": 2.569401741027832, + "step": 27660 + }, + { + "epoch": 0.07535048319041786, + "grad_norm": 2.358707904815674, + "learning_rate": 1.851824996592613e-05, + "loss": 2.552216720581055, + "step": 27680 + }, + { + "epoch": 0.07540492718116237, + "grad_norm": 2.208294630050659, + "learning_rate": 1.8517159602017175e-05, + "loss": 2.5654436111450196, + "step": 27700 + }, + { + "epoch": 0.0754593711719069, + "grad_norm": 2.1413750648498535, + "learning_rate": 1.851606923810822e-05, + "loss": 2.5533222198486327, + "step": 27720 + }, + { + "epoch": 0.07551381516265142, + "grad_norm": 2.1812713146209717, + "learning_rate": 1.8514978874199264e-05, + "loss": 2.5498889923095702, + "step": 27740 + }, + { + "epoch": 0.07556825915339595, + "grad_norm": 2.182156801223755, + "learning_rate": 1.851388851029031e-05, + "loss": 2.55142879486084, + "step": 27760 + }, + { + "epoch": 0.07562270314414046, + "grad_norm": 2.2660892009735107, + "learning_rate": 1.8512798146381355e-05, + "loss": 2.5705352783203126, + "step": 27780 + }, + { + "epoch": 0.07567714713488499, + "grad_norm": 2.0818521976470947, + "learning_rate": 1.85117077824724e-05, + "loss": 2.5490896224975588, + "step": 27800 + }, + { + "epoch": 0.07573159112562951, + "grad_norm": 2.098632335662842, + "learning_rate": 1.8510617418563447e-05, + "loss": 2.5556900024414064, + "step": 27820 + }, + { + "epoch": 0.07578603511637404, + "grad_norm": 2.192869186401367, + "learning_rate": 1.8509527054654493e-05, + "loss": 2.573437309265137, + "step": 27840 + }, + { + "epoch": 0.07584047910711855, + "grad_norm": 2.110872268676758, + "learning_rate": 1.850843669074554e-05, + "loss": 2.566316604614258, + "step": 27860 + }, + { + "epoch": 0.07589492309786307, + "grad_norm": 2.0738534927368164, + "learning_rate": 1.8507346326836584e-05, + "loss": 2.534881019592285, + "step": 27880 + }, + { + "epoch": 0.0759493670886076, + "grad_norm": 2.218609571456909, + "learning_rate": 1.8506255962927627e-05, + "loss": 2.5303874969482423, + "step": 27900 + }, + { + "epoch": 0.07600381107935211, + "grad_norm": 2.4635629653930664, + "learning_rate": 1.8505165599018673e-05, + "loss": 2.5612979888916017, + "step": 27920 + }, + { + "epoch": 0.07605825507009664, + "grad_norm": 2.1352953910827637, + "learning_rate": 1.850407523510972e-05, + "loss": 2.575644874572754, + "step": 27940 + }, + { + "epoch": 0.07611269906084116, + "grad_norm": 2.3145196437835693, + "learning_rate": 1.8502984871200764e-05, + "loss": 2.520529365539551, + "step": 27960 + }, + { + "epoch": 0.07616714305158569, + "grad_norm": 4.232629299163818, + "learning_rate": 1.850189450729181e-05, + "loss": 2.5636390686035155, + "step": 27980 + }, + { + "epoch": 0.0762215870423302, + "grad_norm": 2.21356463432312, + "learning_rate": 1.8500804143382856e-05, + "loss": 2.5539037704467775, + "step": 28000 + }, + { + "epoch": 0.07627603103307472, + "grad_norm": 2.1194536685943604, + "learning_rate": 1.84997137794739e-05, + "loss": 2.5583648681640625, + "step": 28020 + }, + { + "epoch": 0.07633047502381925, + "grad_norm": 2.1884076595306396, + "learning_rate": 1.8498623415564947e-05, + "loss": 2.582411193847656, + "step": 28040 + }, + { + "epoch": 0.07638491901456376, + "grad_norm": 2.1878864765167236, + "learning_rate": 1.849753305165599e-05, + "loss": 2.5516109466552734, + "step": 28060 + }, + { + "epoch": 0.07643936300530829, + "grad_norm": 2.1520237922668457, + "learning_rate": 1.8496442687747036e-05, + "loss": 2.5771244049072264, + "step": 28080 + }, + { + "epoch": 0.07649380699605281, + "grad_norm": 2.2343432903289795, + "learning_rate": 1.849535232383808e-05, + "loss": 2.56945743560791, + "step": 28100 + }, + { + "epoch": 0.07654825098679734, + "grad_norm": 2.104367256164551, + "learning_rate": 1.8494261959929127e-05, + "loss": 2.5818796157836914, + "step": 28120 + }, + { + "epoch": 0.07660269497754185, + "grad_norm": 2.16044282913208, + "learning_rate": 1.8493171596020173e-05, + "loss": 2.5577526092529297, + "step": 28140 + }, + { + "epoch": 0.07665713896828638, + "grad_norm": 2.2096340656280518, + "learning_rate": 1.849208123211122e-05, + "loss": 2.56512565612793, + "step": 28160 + }, + { + "epoch": 0.0767115829590309, + "grad_norm": 2.1783881187438965, + "learning_rate": 1.8490990868202265e-05, + "loss": 2.5604949951171876, + "step": 28180 + }, + { + "epoch": 0.07676602694977541, + "grad_norm": 19.91193389892578, + "learning_rate": 1.848990050429331e-05, + "loss": 2.5768409729003907, + "step": 28200 + }, + { + "epoch": 0.07682047094051994, + "grad_norm": 2.162404775619507, + "learning_rate": 1.8488810140384353e-05, + "loss": 2.5665027618408205, + "step": 28220 + }, + { + "epoch": 0.07687491493126446, + "grad_norm": 3.5002429485321045, + "learning_rate": 1.8487719776475402e-05, + "loss": 2.541372871398926, + "step": 28240 + }, + { + "epoch": 0.07692935892200899, + "grad_norm": 2.0772616863250732, + "learning_rate": 1.8486629412566445e-05, + "loss": 2.5635149002075197, + "step": 28260 + }, + { + "epoch": 0.0769838029127535, + "grad_norm": 2.1552505493164062, + "learning_rate": 1.848553904865749e-05, + "loss": 2.5262777328491213, + "step": 28280 + }, + { + "epoch": 0.07703824690349803, + "grad_norm": 2.2851436138153076, + "learning_rate": 1.8484448684748536e-05, + "loss": 2.541823959350586, + "step": 28300 + }, + { + "epoch": 0.07709269089424255, + "grad_norm": 2.02889084815979, + "learning_rate": 1.8483358320839582e-05, + "loss": 2.573161506652832, + "step": 28320 + }, + { + "epoch": 0.07714713488498708, + "grad_norm": 2.0398495197296143, + "learning_rate": 1.8482267956930628e-05, + "loss": 2.5492227554321287, + "step": 28340 + }, + { + "epoch": 0.07720157887573159, + "grad_norm": 2.217874050140381, + "learning_rate": 1.8481177593021674e-05, + "loss": 2.5681074142456053, + "step": 28360 + }, + { + "epoch": 0.0772560228664761, + "grad_norm": 2.182879686355591, + "learning_rate": 1.8480087229112716e-05, + "loss": 2.5416610717773436, + "step": 28380 + }, + { + "epoch": 0.07731046685722064, + "grad_norm": 2.289778232574463, + "learning_rate": 1.8478996865203765e-05, + "loss": 2.5294610977172853, + "step": 28400 + }, + { + "epoch": 0.07736491084796515, + "grad_norm": 2.520627737045288, + "learning_rate": 1.8477906501294808e-05, + "loss": 2.5263078689575194, + "step": 28420 + }, + { + "epoch": 0.07741935483870968, + "grad_norm": 2.347929000854492, + "learning_rate": 1.8476816137385853e-05, + "loss": 2.5553125381469726, + "step": 28440 + }, + { + "epoch": 0.0774737988294542, + "grad_norm": 2.1347994804382324, + "learning_rate": 1.84757257734769e-05, + "loss": 2.529450225830078, + "step": 28460 + }, + { + "epoch": 0.07752824282019873, + "grad_norm": 2.094590425491333, + "learning_rate": 1.8474635409567945e-05, + "loss": 2.5354305267333985, + "step": 28480 + }, + { + "epoch": 0.07758268681094324, + "grad_norm": 2.7854297161102295, + "learning_rate": 1.847354504565899e-05, + "loss": 2.554463768005371, + "step": 28500 + }, + { + "epoch": 0.07763713080168777, + "grad_norm": 2.312638521194458, + "learning_rate": 1.8472454681750037e-05, + "loss": 2.5796176910400392, + "step": 28520 + }, + { + "epoch": 0.07769157479243229, + "grad_norm": 2.0593533515930176, + "learning_rate": 1.847136431784108e-05, + "loss": 2.5510772705078124, + "step": 28540 + }, + { + "epoch": 0.0777460187831768, + "grad_norm": 2.153927803039551, + "learning_rate": 1.8470273953932128e-05, + "loss": 2.577483558654785, + "step": 28560 + }, + { + "epoch": 0.07780046277392133, + "grad_norm": 2.1587204933166504, + "learning_rate": 1.846918359002317e-05, + "loss": 2.5601015090942383, + "step": 28580 + }, + { + "epoch": 0.07785490676466585, + "grad_norm": 2.1574528217315674, + "learning_rate": 1.8468093226114217e-05, + "loss": 2.592529296875, + "step": 28600 + }, + { + "epoch": 0.07790935075541038, + "grad_norm": 2.1227645874023438, + "learning_rate": 1.8467002862205262e-05, + "loss": 2.576510429382324, + "step": 28620 + }, + { + "epoch": 0.07796379474615489, + "grad_norm": 2.2045459747314453, + "learning_rate": 1.8465912498296308e-05, + "loss": 2.5857122421264647, + "step": 28640 + }, + { + "epoch": 0.07801823873689942, + "grad_norm": 2.150580644607544, + "learning_rate": 1.8464822134387354e-05, + "loss": 2.550697135925293, + "step": 28660 + }, + { + "epoch": 0.07807268272764394, + "grad_norm": 2.1799564361572266, + "learning_rate": 1.84637317704784e-05, + "loss": 2.539801597595215, + "step": 28680 + }, + { + "epoch": 0.07812712671838845, + "grad_norm": 2.106555461883545, + "learning_rate": 1.8462641406569442e-05, + "loss": 2.584349250793457, + "step": 28700 + }, + { + "epoch": 0.07818157070913298, + "grad_norm": 2.1037638187408447, + "learning_rate": 1.846155104266049e-05, + "loss": 2.5631853103637696, + "step": 28720 + }, + { + "epoch": 0.0782360146998775, + "grad_norm": 2.252190351486206, + "learning_rate": 1.8460460678751534e-05, + "loss": 2.5738569259643556, + "step": 28740 + }, + { + "epoch": 0.07829045869062203, + "grad_norm": 2.525827407836914, + "learning_rate": 1.845937031484258e-05, + "loss": 2.5614904403686523, + "step": 28760 + }, + { + "epoch": 0.07834490268136654, + "grad_norm": 2.331411361694336, + "learning_rate": 1.8458279950933625e-05, + "loss": 2.5534744262695312, + "step": 28780 + }, + { + "epoch": 0.07839934667211107, + "grad_norm": 2.2775394916534424, + "learning_rate": 1.845718958702467e-05, + "loss": 2.5889572143554687, + "step": 28800 + }, + { + "epoch": 0.07845379066285559, + "grad_norm": 2.235701084136963, + "learning_rate": 1.8456099223115717e-05, + "loss": 2.5731647491455076, + "step": 28820 + }, + { + "epoch": 0.07850823465360011, + "grad_norm": 2.174215793609619, + "learning_rate": 1.8455008859206763e-05, + "loss": 2.580392837524414, + "step": 28840 + }, + { + "epoch": 0.07856267864434463, + "grad_norm": 2.11948823928833, + "learning_rate": 1.8453918495297805e-05, + "loss": 2.5596628189086914, + "step": 28860 + }, + { + "epoch": 0.07861712263508915, + "grad_norm": 2.273303985595703, + "learning_rate": 1.8452828131388854e-05, + "loss": 2.5753505706787108, + "step": 28880 + }, + { + "epoch": 0.07867156662583367, + "grad_norm": 2.3481879234313965, + "learning_rate": 1.8451737767479897e-05, + "loss": 2.571450614929199, + "step": 28900 + }, + { + "epoch": 0.07872601061657819, + "grad_norm": 2.1474227905273438, + "learning_rate": 1.8450647403570943e-05, + "loss": 2.577282524108887, + "step": 28920 + }, + { + "epoch": 0.07878045460732272, + "grad_norm": 2.134211778640747, + "learning_rate": 1.844955703966199e-05, + "loss": 2.5732534408569334, + "step": 28940 + }, + { + "epoch": 0.07883489859806724, + "grad_norm": 2.181583881378174, + "learning_rate": 1.8448466675753034e-05, + "loss": 2.5535280227661135, + "step": 28960 + }, + { + "epoch": 0.07888934258881176, + "grad_norm": 2.200026512145996, + "learning_rate": 1.844737631184408e-05, + "loss": 2.583490180969238, + "step": 28980 + }, + { + "epoch": 0.07894378657955628, + "grad_norm": 2.1251001358032227, + "learning_rate": 1.8446285947935126e-05, + "loss": 2.55838680267334, + "step": 29000 + }, + { + "epoch": 0.07899823057030081, + "grad_norm": 2.133291482925415, + "learning_rate": 1.844519558402617e-05, + "loss": 2.572433662414551, + "step": 29020 + }, + { + "epoch": 0.07905267456104532, + "grad_norm": 2.1832735538482666, + "learning_rate": 1.8444105220117218e-05, + "loss": 2.5677513122558593, + "step": 29040 + }, + { + "epoch": 0.07910711855178984, + "grad_norm": 2.2378079891204834, + "learning_rate": 1.844301485620826e-05, + "loss": 2.575445556640625, + "step": 29060 + }, + { + "epoch": 0.07916156254253437, + "grad_norm": 2.141896963119507, + "learning_rate": 1.8441924492299306e-05, + "loss": 2.5361595153808594, + "step": 29080 + }, + { + "epoch": 0.07921600653327888, + "grad_norm": 2.034019947052002, + "learning_rate": 1.844083412839035e-05, + "loss": 2.5610767364501954, + "step": 29100 + }, + { + "epoch": 0.07927045052402341, + "grad_norm": 2.1115612983703613, + "learning_rate": 1.8439743764481397e-05, + "loss": 2.567348098754883, + "step": 29120 + }, + { + "epoch": 0.07932489451476793, + "grad_norm": 2.0837583541870117, + "learning_rate": 1.8438653400572443e-05, + "loss": 2.5267875671386717, + "step": 29140 + }, + { + "epoch": 0.07937933850551246, + "grad_norm": 2.164400577545166, + "learning_rate": 1.843756303666349e-05, + "loss": 2.5678874969482424, + "step": 29160 + }, + { + "epoch": 0.07943378249625697, + "grad_norm": 2.3009912967681885, + "learning_rate": 1.843647267275453e-05, + "loss": 2.554843521118164, + "step": 29180 + }, + { + "epoch": 0.0794882264870015, + "grad_norm": 2.1644740104675293, + "learning_rate": 1.843538230884558e-05, + "loss": 2.5404693603515627, + "step": 29200 + }, + { + "epoch": 0.07954267047774602, + "grad_norm": 2.268188953399658, + "learning_rate": 1.8434291944936623e-05, + "loss": 2.560512161254883, + "step": 29220 + }, + { + "epoch": 0.07959711446849053, + "grad_norm": 2.5396223068237305, + "learning_rate": 1.843320158102767e-05, + "loss": 2.5408309936523437, + "step": 29240 + }, + { + "epoch": 0.07965155845923506, + "grad_norm": 2.169010877609253, + "learning_rate": 1.8432111217118715e-05, + "loss": 2.5576032638549804, + "step": 29260 + }, + { + "epoch": 0.07970600244997958, + "grad_norm": 2.2755086421966553, + "learning_rate": 1.843102085320976e-05, + "loss": 2.5899259567260744, + "step": 29280 + }, + { + "epoch": 0.07976044644072411, + "grad_norm": 2.106158971786499, + "learning_rate": 1.8429930489300806e-05, + "loss": 2.554769515991211, + "step": 29300 + }, + { + "epoch": 0.07981489043146862, + "grad_norm": 2.174595355987549, + "learning_rate": 1.8428840125391852e-05, + "loss": 2.578830909729004, + "step": 29320 + }, + { + "epoch": 0.07986933442221315, + "grad_norm": 2.0609819889068604, + "learning_rate": 1.8427749761482895e-05, + "loss": 2.5650381088256835, + "step": 29340 + }, + { + "epoch": 0.07992377841295767, + "grad_norm": 2.198760509490967, + "learning_rate": 1.8426659397573944e-05, + "loss": 2.5783802032470704, + "step": 29360 + }, + { + "epoch": 0.0799782224037022, + "grad_norm": 2.1206600666046143, + "learning_rate": 1.8425569033664986e-05, + "loss": 2.573335075378418, + "step": 29380 + }, + { + "epoch": 0.08003266639444671, + "grad_norm": 2.158616781234741, + "learning_rate": 1.8424478669756032e-05, + "loss": 2.5494049072265623, + "step": 29400 + }, + { + "epoch": 0.08008711038519123, + "grad_norm": 2.179838180541992, + "learning_rate": 1.8423388305847078e-05, + "loss": 2.543887710571289, + "step": 29420 + }, + { + "epoch": 0.08014155437593576, + "grad_norm": 2.459240198135376, + "learning_rate": 1.8422297941938124e-05, + "loss": 2.5921024322509765, + "step": 29440 + }, + { + "epoch": 0.08019599836668027, + "grad_norm": 2.1827471256256104, + "learning_rate": 1.842120757802917e-05, + "loss": 2.571135902404785, + "step": 29460 + }, + { + "epoch": 0.0802504423574248, + "grad_norm": 2.2822959423065186, + "learning_rate": 1.8420117214120215e-05, + "loss": 2.54454345703125, + "step": 29480 + }, + { + "epoch": 0.08030488634816932, + "grad_norm": 3.2379019260406494, + "learning_rate": 1.8419026850211258e-05, + "loss": 2.605178642272949, + "step": 29500 + }, + { + "epoch": 0.08035933033891385, + "grad_norm": 2.150876998901367, + "learning_rate": 1.8417936486302307e-05, + "loss": 2.5758520126342774, + "step": 29520 + }, + { + "epoch": 0.08041377432965836, + "grad_norm": 2.801929473876953, + "learning_rate": 1.841684612239335e-05, + "loss": 2.550069808959961, + "step": 29540 + }, + { + "epoch": 0.08046821832040288, + "grad_norm": 2.0666592121124268, + "learning_rate": 1.8415755758484395e-05, + "loss": 2.545466995239258, + "step": 29560 + }, + { + "epoch": 0.08052266231114741, + "grad_norm": 3.233212947845459, + "learning_rate": 1.841466539457544e-05, + "loss": 2.5715999603271484, + "step": 29580 + }, + { + "epoch": 0.08057710630189192, + "grad_norm": 2.146275520324707, + "learning_rate": 1.8413575030666487e-05, + "loss": 2.5545631408691407, + "step": 29600 + }, + { + "epoch": 0.08063155029263645, + "grad_norm": 2.8792643547058105, + "learning_rate": 1.8412484666757532e-05, + "loss": 2.5767383575439453, + "step": 29620 + }, + { + "epoch": 0.08068599428338097, + "grad_norm": 2.312344551086426, + "learning_rate": 1.8411394302848578e-05, + "loss": 2.5556774139404297, + "step": 29640 + }, + { + "epoch": 0.0807404382741255, + "grad_norm": 2.135483741760254, + "learning_rate": 1.841030393893962e-05, + "loss": 2.581375312805176, + "step": 29660 + }, + { + "epoch": 0.08079488226487001, + "grad_norm": 2.3039259910583496, + "learning_rate": 1.840921357503067e-05, + "loss": 2.567531204223633, + "step": 29680 + }, + { + "epoch": 0.08084932625561454, + "grad_norm": 2.1072120666503906, + "learning_rate": 1.8408123211121712e-05, + "loss": 2.5631479263305663, + "step": 29700 + }, + { + "epoch": 0.08090377024635906, + "grad_norm": 2.3414971828460693, + "learning_rate": 1.8407032847212758e-05, + "loss": 2.543726348876953, + "step": 29720 + }, + { + "epoch": 0.08095821423710357, + "grad_norm": 2.1305272579193115, + "learning_rate": 1.8405942483303804e-05, + "loss": 2.5828168869018553, + "step": 29740 + }, + { + "epoch": 0.0810126582278481, + "grad_norm": 2.1304986476898193, + "learning_rate": 1.840485211939485e-05, + "loss": 2.5705965042114256, + "step": 29760 + }, + { + "epoch": 0.08106710221859262, + "grad_norm": 2.171783447265625, + "learning_rate": 1.8403761755485892e-05, + "loss": 2.5422348022460937, + "step": 29780 + }, + { + "epoch": 0.08112154620933715, + "grad_norm": 2.3127918243408203, + "learning_rate": 1.840267139157694e-05, + "loss": 2.572126007080078, + "step": 29800 + }, + { + "epoch": 0.08117599020008166, + "grad_norm": 2.2140884399414062, + "learning_rate": 1.8401581027667984e-05, + "loss": 2.5303300857543944, + "step": 29820 + }, + { + "epoch": 0.08123043419082619, + "grad_norm": 2.134960651397705, + "learning_rate": 1.8400490663759033e-05, + "loss": 2.5399803161621093, + "step": 29840 + }, + { + "epoch": 0.08128487818157071, + "grad_norm": 2.2027742862701416, + "learning_rate": 1.8399400299850075e-05, + "loss": 2.5856697082519533, + "step": 29860 + }, + { + "epoch": 0.08133932217231524, + "grad_norm": 2.0670974254608154, + "learning_rate": 1.839830993594112e-05, + "loss": 2.553943634033203, + "step": 29880 + }, + { + "epoch": 0.08139376616305975, + "grad_norm": 2.080112934112549, + "learning_rate": 1.8397219572032167e-05, + "loss": 2.5491125106811525, + "step": 29900 + }, + { + "epoch": 0.08144821015380427, + "grad_norm": 2.298460006713867, + "learning_rate": 1.8396129208123213e-05, + "loss": 2.554925727844238, + "step": 29920 + }, + { + "epoch": 0.0815026541445488, + "grad_norm": 2.103710174560547, + "learning_rate": 1.839503884421426e-05, + "loss": 2.587455940246582, + "step": 29940 + }, + { + "epoch": 0.08155709813529331, + "grad_norm": 2.111035108566284, + "learning_rate": 1.8393948480305304e-05, + "loss": 2.580238151550293, + "step": 29960 + }, + { + "epoch": 0.08161154212603784, + "grad_norm": 2.2416529655456543, + "learning_rate": 1.8392858116396347e-05, + "loss": 2.561504364013672, + "step": 29980 + }, + { + "epoch": 0.08166598611678236, + "grad_norm": 2.2516839504241943, + "learning_rate": 1.8391767752487396e-05, + "loss": 2.577525329589844, + "step": 30000 + }, + { + "epoch": 0.08172043010752689, + "grad_norm": 2.1874606609344482, + "learning_rate": 1.839067738857844e-05, + "loss": 2.553631591796875, + "step": 30020 + }, + { + "epoch": 0.0817748740982714, + "grad_norm": 2.097135066986084, + "learning_rate": 1.8389587024669484e-05, + "loss": 2.5533443450927735, + "step": 30040 + }, + { + "epoch": 0.08182931808901593, + "grad_norm": 2.058709144592285, + "learning_rate": 1.838849666076053e-05, + "loss": 2.55930233001709, + "step": 30060 + }, + { + "epoch": 0.08188376207976045, + "grad_norm": 2.141758441925049, + "learning_rate": 1.8387406296851576e-05, + "loss": 2.54139404296875, + "step": 30080 + }, + { + "epoch": 0.08193820607050496, + "grad_norm": 2.412738800048828, + "learning_rate": 1.8386315932942622e-05, + "loss": 2.556705093383789, + "step": 30100 + }, + { + "epoch": 0.08199265006124949, + "grad_norm": 2.0931103229522705, + "learning_rate": 1.8385225569033668e-05, + "loss": 2.5276056289672852, + "step": 30120 + }, + { + "epoch": 0.08204709405199401, + "grad_norm": 2.0413975715637207, + "learning_rate": 1.838413520512471e-05, + "loss": 2.572550582885742, + "step": 30140 + }, + { + "epoch": 0.08210153804273854, + "grad_norm": 2.21547532081604, + "learning_rate": 1.838304484121576e-05, + "loss": 2.579325485229492, + "step": 30160 + }, + { + "epoch": 0.08215598203348305, + "grad_norm": 2.1893692016601562, + "learning_rate": 1.83819544773068e-05, + "loss": 2.5486434936523437, + "step": 30180 + }, + { + "epoch": 0.08221042602422758, + "grad_norm": 2.133234977722168, + "learning_rate": 1.8380864113397847e-05, + "loss": 2.548659324645996, + "step": 30200 + }, + { + "epoch": 0.0822648700149721, + "grad_norm": 2.1883249282836914, + "learning_rate": 1.8379773749488893e-05, + "loss": 2.5526607513427733, + "step": 30220 + }, + { + "epoch": 0.08231931400571661, + "grad_norm": 2.199448585510254, + "learning_rate": 1.837868338557994e-05, + "loss": 2.5701927185058593, + "step": 30240 + }, + { + "epoch": 0.08237375799646114, + "grad_norm": 2.3185482025146484, + "learning_rate": 1.8377593021670985e-05, + "loss": 2.561257743835449, + "step": 30260 + }, + { + "epoch": 0.08242820198720566, + "grad_norm": 2.157356023788452, + "learning_rate": 1.837650265776203e-05, + "loss": 2.552052307128906, + "step": 30280 + }, + { + "epoch": 0.08248264597795019, + "grad_norm": 2.116755485534668, + "learning_rate": 1.8375412293853073e-05, + "loss": 2.5407794952392577, + "step": 30300 + }, + { + "epoch": 0.0825370899686947, + "grad_norm": 2.2307193279266357, + "learning_rate": 1.8374321929944122e-05, + "loss": 2.547009086608887, + "step": 30320 + }, + { + "epoch": 0.08259153395943923, + "grad_norm": 2.2323191165924072, + "learning_rate": 1.8373231566035165e-05, + "loss": 2.530338668823242, + "step": 30340 + }, + { + "epoch": 0.08264597795018375, + "grad_norm": 2.2182939052581787, + "learning_rate": 1.837214120212621e-05, + "loss": 2.5501522064208983, + "step": 30360 + }, + { + "epoch": 0.08270042194092828, + "grad_norm": 2.5120322704315186, + "learning_rate": 1.8371050838217256e-05, + "loss": 2.554744529724121, + "step": 30380 + }, + { + "epoch": 0.08275486593167279, + "grad_norm": 2.229562997817993, + "learning_rate": 1.8369960474308302e-05, + "loss": 2.5532283782958984, + "step": 30400 + }, + { + "epoch": 0.0828093099224173, + "grad_norm": 2.43152117729187, + "learning_rate": 1.8368870110399348e-05, + "loss": 2.5644371032714846, + "step": 30420 + }, + { + "epoch": 0.08286375391316184, + "grad_norm": 2.195586919784546, + "learning_rate": 1.8367779746490394e-05, + "loss": 2.5671463012695312, + "step": 30440 + }, + { + "epoch": 0.08291819790390635, + "grad_norm": 2.428392171859741, + "learning_rate": 1.8366689382581436e-05, + "loss": 2.555580711364746, + "step": 30460 + }, + { + "epoch": 0.08297264189465088, + "grad_norm": 2.1202478408813477, + "learning_rate": 1.8365599018672485e-05, + "loss": 2.5492136001586916, + "step": 30480 + }, + { + "epoch": 0.0830270858853954, + "grad_norm": 2.1184775829315186, + "learning_rate": 1.8364508654763528e-05, + "loss": 2.5782445907592773, + "step": 30500 + }, + { + "epoch": 0.08308152987613993, + "grad_norm": 2.141129970550537, + "learning_rate": 1.8363418290854574e-05, + "loss": 2.545554542541504, + "step": 30520 + }, + { + "epoch": 0.08313597386688444, + "grad_norm": 2.2110414505004883, + "learning_rate": 1.836232792694562e-05, + "loss": 2.5721343994140624, + "step": 30540 + }, + { + "epoch": 0.08319041785762897, + "grad_norm": 2.0826785564422607, + "learning_rate": 1.8361237563036665e-05, + "loss": 2.542689323425293, + "step": 30560 + }, + { + "epoch": 0.08324486184837349, + "grad_norm": 2.270921230316162, + "learning_rate": 1.836014719912771e-05, + "loss": 2.5808794021606447, + "step": 30580 + }, + { + "epoch": 0.083299305839118, + "grad_norm": 2.169328212738037, + "learning_rate": 1.8359056835218757e-05, + "loss": 2.5769536972045897, + "step": 30600 + }, + { + "epoch": 0.08335374982986253, + "grad_norm": 2.0694797039031982, + "learning_rate": 1.83579664713098e-05, + "loss": 2.5572015762329103, + "step": 30620 + }, + { + "epoch": 0.08340819382060705, + "grad_norm": 2.166994333267212, + "learning_rate": 1.835687610740085e-05, + "loss": 2.5759206771850587, + "step": 30640 + }, + { + "epoch": 0.08346263781135158, + "grad_norm": 2.1691441535949707, + "learning_rate": 1.835578574349189e-05, + "loss": 2.582478141784668, + "step": 30660 + }, + { + "epoch": 0.08351708180209609, + "grad_norm": 2.5496413707733154, + "learning_rate": 1.8354695379582937e-05, + "loss": 2.5536163330078123, + "step": 30680 + }, + { + "epoch": 0.08357152579284062, + "grad_norm": 2.2248713970184326, + "learning_rate": 1.8353605015673982e-05, + "loss": 2.5778715133666994, + "step": 30700 + }, + { + "epoch": 0.08362596978358514, + "grad_norm": 2.1783878803253174, + "learning_rate": 1.8352514651765028e-05, + "loss": 2.572823143005371, + "step": 30720 + }, + { + "epoch": 0.08368041377432966, + "grad_norm": 2.16725492477417, + "learning_rate": 1.8351424287856074e-05, + "loss": 2.527462959289551, + "step": 30740 + }, + { + "epoch": 0.08373485776507418, + "grad_norm": 2.3595104217529297, + "learning_rate": 1.835033392394712e-05, + "loss": 2.564670944213867, + "step": 30760 + }, + { + "epoch": 0.0837893017558187, + "grad_norm": 2.172717809677124, + "learning_rate": 1.8349243560038162e-05, + "loss": 2.562975311279297, + "step": 30780 + }, + { + "epoch": 0.08384374574656323, + "grad_norm": 2.499765634536743, + "learning_rate": 1.834815319612921e-05, + "loss": 2.600170707702637, + "step": 30800 + }, + { + "epoch": 0.08389818973730774, + "grad_norm": 2.3884823322296143, + "learning_rate": 1.8347062832220254e-05, + "loss": 2.5612993240356445, + "step": 30820 + }, + { + "epoch": 0.08395263372805227, + "grad_norm": 2.1352767944335938, + "learning_rate": 1.83459724683113e-05, + "loss": 2.5667787551879884, + "step": 30840 + }, + { + "epoch": 0.08400707771879679, + "grad_norm": 2.1010565757751465, + "learning_rate": 1.8344882104402346e-05, + "loss": 2.566105842590332, + "step": 30860 + }, + { + "epoch": 0.08406152170954131, + "grad_norm": 2.220996856689453, + "learning_rate": 1.834379174049339e-05, + "loss": 2.566444206237793, + "step": 30880 + }, + { + "epoch": 0.08411596570028583, + "grad_norm": 2.13452410697937, + "learning_rate": 1.8342701376584437e-05, + "loss": 2.5429288864135744, + "step": 30900 + }, + { + "epoch": 0.08417040969103035, + "grad_norm": 2.222231149673462, + "learning_rate": 1.8341611012675483e-05, + "loss": 2.54953556060791, + "step": 30920 + }, + { + "epoch": 0.08422485368177487, + "grad_norm": 2.1548352241516113, + "learning_rate": 1.8340520648766525e-05, + "loss": 2.5455453872680662, + "step": 30940 + }, + { + "epoch": 0.08427929767251939, + "grad_norm": 2.1819591522216797, + "learning_rate": 1.8339430284857575e-05, + "loss": 2.555816078186035, + "step": 30960 + }, + { + "epoch": 0.08433374166326392, + "grad_norm": 6.287555694580078, + "learning_rate": 1.8338339920948617e-05, + "loss": 2.5457937240600588, + "step": 30980 + }, + { + "epoch": 0.08438818565400844, + "grad_norm": 2.1017162799835205, + "learning_rate": 1.8337249557039663e-05, + "loss": 2.583380126953125, + "step": 31000 + }, + { + "epoch": 0.08444262964475296, + "grad_norm": 2.1579501628875732, + "learning_rate": 1.833615919313071e-05, + "loss": 2.5776477813720704, + "step": 31020 + }, + { + "epoch": 0.08449707363549748, + "grad_norm": 2.399909257888794, + "learning_rate": 1.8335068829221754e-05, + "loss": 2.565115547180176, + "step": 31040 + }, + { + "epoch": 0.08455151762624201, + "grad_norm": 2.14369797706604, + "learning_rate": 1.83339784653128e-05, + "loss": 2.5482364654541017, + "step": 31060 + }, + { + "epoch": 0.08460596161698652, + "grad_norm": 2.1204299926757812, + "learning_rate": 1.8332888101403846e-05, + "loss": 2.5684431076049803, + "step": 31080 + }, + { + "epoch": 0.08466040560773104, + "grad_norm": 2.1358816623687744, + "learning_rate": 1.833179773749489e-05, + "loss": 2.553989028930664, + "step": 31100 + }, + { + "epoch": 0.08471484959847557, + "grad_norm": 2.2612595558166504, + "learning_rate": 1.8330707373585938e-05, + "loss": 2.545004463195801, + "step": 31120 + }, + { + "epoch": 0.08476929358922008, + "grad_norm": 2.1850757598876953, + "learning_rate": 1.832961700967698e-05, + "loss": 2.564195251464844, + "step": 31140 + }, + { + "epoch": 0.08482373757996461, + "grad_norm": 2.183120012283325, + "learning_rate": 1.8328526645768026e-05, + "loss": 2.5782358169555666, + "step": 31160 + }, + { + "epoch": 0.08487818157070913, + "grad_norm": 2.1439294815063477, + "learning_rate": 1.832743628185907e-05, + "loss": 2.5694297790527343, + "step": 31180 + }, + { + "epoch": 0.08493262556145366, + "grad_norm": 2.1498501300811768, + "learning_rate": 1.8326345917950117e-05, + "loss": 2.5237457275390627, + "step": 31200 + }, + { + "epoch": 0.08498706955219817, + "grad_norm": 2.150940179824829, + "learning_rate": 1.8325255554041163e-05, + "loss": 2.5888084411621093, + "step": 31220 + }, + { + "epoch": 0.0850415135429427, + "grad_norm": 2.902275323867798, + "learning_rate": 1.832416519013221e-05, + "loss": 2.531839370727539, + "step": 31240 + }, + { + "epoch": 0.08509595753368722, + "grad_norm": 2.222759485244751, + "learning_rate": 1.832307482622325e-05, + "loss": 2.556698226928711, + "step": 31260 + }, + { + "epoch": 0.08515040152443173, + "grad_norm": 2.354145050048828, + "learning_rate": 1.83219844623143e-05, + "loss": 2.6024879455566405, + "step": 31280 + }, + { + "epoch": 0.08520484551517626, + "grad_norm": 2.131054401397705, + "learning_rate": 1.8320894098405343e-05, + "loss": 2.536322021484375, + "step": 31300 + }, + { + "epoch": 0.08525928950592078, + "grad_norm": 2.194857358932495, + "learning_rate": 1.831980373449639e-05, + "loss": 2.576128387451172, + "step": 31320 + }, + { + "epoch": 0.08531373349666531, + "grad_norm": 2.131690740585327, + "learning_rate": 1.8318713370587435e-05, + "loss": 2.5525259017944335, + "step": 31340 + }, + { + "epoch": 0.08536817748740982, + "grad_norm": 2.1102218627929688, + "learning_rate": 1.831762300667848e-05, + "loss": 2.5892847061157225, + "step": 31360 + }, + { + "epoch": 0.08542262147815435, + "grad_norm": 2.0493950843811035, + "learning_rate": 1.8316532642769526e-05, + "loss": 2.6003391265869142, + "step": 31380 + }, + { + "epoch": 0.08547706546889887, + "grad_norm": 3.9041054248809814, + "learning_rate": 1.8315442278860572e-05, + "loss": 2.554571533203125, + "step": 31400 + }, + { + "epoch": 0.0855315094596434, + "grad_norm": 2.0853421688079834, + "learning_rate": 1.8314351914951615e-05, + "loss": 2.602634239196777, + "step": 31420 + }, + { + "epoch": 0.08558595345038791, + "grad_norm": 2.174835681915283, + "learning_rate": 1.8313261551042664e-05, + "loss": 2.5555440902709963, + "step": 31440 + }, + { + "epoch": 0.08564039744113243, + "grad_norm": 2.116136074066162, + "learning_rate": 1.8312171187133706e-05, + "loss": 2.581534004211426, + "step": 31460 + }, + { + "epoch": 0.08569484143187696, + "grad_norm": 2.1198620796203613, + "learning_rate": 1.8311080823224752e-05, + "loss": 2.5474712371826174, + "step": 31480 + }, + { + "epoch": 0.08574928542262147, + "grad_norm": 2.1607253551483154, + "learning_rate": 1.8309990459315798e-05, + "loss": 2.5545221328735352, + "step": 31500 + }, + { + "epoch": 0.085803729413366, + "grad_norm": 2.2867183685302734, + "learning_rate": 1.8308900095406844e-05, + "loss": 2.5657520294189453, + "step": 31520 + }, + { + "epoch": 0.08585817340411052, + "grad_norm": 2.2273905277252197, + "learning_rate": 1.830780973149789e-05, + "loss": 2.539216995239258, + "step": 31540 + }, + { + "epoch": 0.08591261739485505, + "grad_norm": 2.2655723094940186, + "learning_rate": 1.8306719367588935e-05, + "loss": 2.574443244934082, + "step": 31560 + }, + { + "epoch": 0.08596706138559956, + "grad_norm": 2.1554641723632812, + "learning_rate": 1.8305629003679978e-05, + "loss": 2.5480026245117187, + "step": 31580 + }, + { + "epoch": 0.08602150537634409, + "grad_norm": 2.1415393352508545, + "learning_rate": 1.8304538639771027e-05, + "loss": 2.5576904296875, + "step": 31600 + }, + { + "epoch": 0.08607594936708861, + "grad_norm": 2.1536612510681152, + "learning_rate": 1.830344827586207e-05, + "loss": 2.544003486633301, + "step": 31620 + }, + { + "epoch": 0.08613039335783312, + "grad_norm": 2.220381021499634, + "learning_rate": 1.830235791195312e-05, + "loss": 2.572074317932129, + "step": 31640 + }, + { + "epoch": 0.08618483734857765, + "grad_norm": 2.1136534214019775, + "learning_rate": 1.830126754804416e-05, + "loss": 2.5918096542358398, + "step": 31660 + }, + { + "epoch": 0.08623928133932217, + "grad_norm": 2.2428078651428223, + "learning_rate": 1.8300177184135207e-05, + "loss": 2.5124324798583983, + "step": 31680 + }, + { + "epoch": 0.0862937253300667, + "grad_norm": 2.253265380859375, + "learning_rate": 1.8299086820226253e-05, + "loss": 2.564986991882324, + "step": 31700 + }, + { + "epoch": 0.08634816932081121, + "grad_norm": 2.2250847816467285, + "learning_rate": 1.82979964563173e-05, + "loss": 2.5593332290649413, + "step": 31720 + }, + { + "epoch": 0.08640261331155574, + "grad_norm": 2.163252830505371, + "learning_rate": 1.829690609240834e-05, + "loss": 2.581595611572266, + "step": 31740 + }, + { + "epoch": 0.08645705730230026, + "grad_norm": 2.6267032623291016, + "learning_rate": 1.829581572849939e-05, + "loss": 2.548160362243652, + "step": 31760 + }, + { + "epoch": 0.08651150129304477, + "grad_norm": 2.254728078842163, + "learning_rate": 1.8294725364590432e-05, + "loss": 2.551910972595215, + "step": 31780 + }, + { + "epoch": 0.0865659452837893, + "grad_norm": 8.47126579284668, + "learning_rate": 1.829363500068148e-05, + "loss": 2.60048770904541, + "step": 31800 + }, + { + "epoch": 0.08662038927453382, + "grad_norm": 3.2209689617156982, + "learning_rate": 1.8292544636772524e-05, + "loss": 2.593061828613281, + "step": 31820 + }, + { + "epoch": 0.08667483326527835, + "grad_norm": 12.447214126586914, + "learning_rate": 1.829145427286357e-05, + "loss": 2.5799636840820312, + "step": 31840 + }, + { + "epoch": 0.08672927725602286, + "grad_norm": 2.1954822540283203, + "learning_rate": 1.8290363908954616e-05, + "loss": 2.5702281951904298, + "step": 31860 + }, + { + "epoch": 0.08678372124676739, + "grad_norm": 2.4860928058624268, + "learning_rate": 1.828927354504566e-05, + "loss": 2.564305877685547, + "step": 31880 + }, + { + "epoch": 0.08683816523751191, + "grad_norm": 2.2635815143585205, + "learning_rate": 1.8288183181136704e-05, + "loss": 2.5446565628051756, + "step": 31900 + }, + { + "epoch": 0.08689260922825644, + "grad_norm": 2.1436879634857178, + "learning_rate": 1.8287092817227753e-05, + "loss": 2.5913568496704102, + "step": 31920 + }, + { + "epoch": 0.08694705321900095, + "grad_norm": 2.302140712738037, + "learning_rate": 1.8286002453318795e-05, + "loss": 2.564913749694824, + "step": 31940 + }, + { + "epoch": 0.08700149720974547, + "grad_norm": 2.2139248847961426, + "learning_rate": 1.8284912089409845e-05, + "loss": 2.569744682312012, + "step": 31960 + }, + { + "epoch": 0.08705594120049, + "grad_norm": 2.0754215717315674, + "learning_rate": 1.8283821725500887e-05, + "loss": 2.5671001434326173, + "step": 31980 + }, + { + "epoch": 0.08711038519123451, + "grad_norm": 2.021704912185669, + "learning_rate": 1.8282731361591933e-05, + "loss": 2.568072509765625, + "step": 32000 + }, + { + "epoch": 0.08716482918197904, + "grad_norm": 2.278270721435547, + "learning_rate": 1.828164099768298e-05, + "loss": 2.5424137115478516, + "step": 32020 + }, + { + "epoch": 0.08721927317272356, + "grad_norm": 2.4119338989257812, + "learning_rate": 1.8280550633774025e-05, + "loss": 2.555117607116699, + "step": 32040 + }, + { + "epoch": 0.08727371716346809, + "grad_norm": 2.2937910556793213, + "learning_rate": 1.8279460269865067e-05, + "loss": 2.5891048431396486, + "step": 32060 + }, + { + "epoch": 0.0873281611542126, + "grad_norm": 3.2070930004119873, + "learning_rate": 1.8278369905956116e-05, + "loss": 2.568084716796875, + "step": 32080 + }, + { + "epoch": 0.08738260514495713, + "grad_norm": 2.133183240890503, + "learning_rate": 1.827727954204716e-05, + "loss": 2.580074119567871, + "step": 32100 + }, + { + "epoch": 0.08743704913570165, + "grad_norm": 2.128298282623291, + "learning_rate": 1.8276189178138208e-05, + "loss": 2.5307106018066405, + "step": 32120 + }, + { + "epoch": 0.08749149312644616, + "grad_norm": 2.0860843658447266, + "learning_rate": 1.827509881422925e-05, + "loss": 2.5318181991577147, + "step": 32140 + }, + { + "epoch": 0.08754593711719069, + "grad_norm": 2.285076141357422, + "learning_rate": 1.8274008450320296e-05, + "loss": 2.5495094299316405, + "step": 32160 + }, + { + "epoch": 0.08760038110793521, + "grad_norm": 2.1587700843811035, + "learning_rate": 1.8272918086411342e-05, + "loss": 2.5826688766479493, + "step": 32180 + }, + { + "epoch": 0.08765482509867974, + "grad_norm": 5.327301025390625, + "learning_rate": 1.8271827722502388e-05, + "loss": 2.5591554641723633, + "step": 32200 + }, + { + "epoch": 0.08770926908942425, + "grad_norm": 2.187150239944458, + "learning_rate": 1.827073735859343e-05, + "loss": 2.54221248626709, + "step": 32220 + }, + { + "epoch": 0.08776371308016878, + "grad_norm": 2.3849709033966064, + "learning_rate": 1.826964699468448e-05, + "loss": 2.5656410217285157, + "step": 32240 + }, + { + "epoch": 0.0878181570709133, + "grad_norm": 2.0602304935455322, + "learning_rate": 1.826855663077552e-05, + "loss": 2.547947120666504, + "step": 32260 + }, + { + "epoch": 0.08787260106165783, + "grad_norm": 2.102842092514038, + "learning_rate": 1.826746626686657e-05, + "loss": 2.554482650756836, + "step": 32280 + }, + { + "epoch": 0.08792704505240234, + "grad_norm": 2.1271188259124756, + "learning_rate": 1.8266375902957613e-05, + "loss": 2.5739646911621095, + "step": 32300 + }, + { + "epoch": 0.08798148904314686, + "grad_norm": 2.185182571411133, + "learning_rate": 1.826528553904866e-05, + "loss": 2.5484142303466797, + "step": 32320 + }, + { + "epoch": 0.08803593303389139, + "grad_norm": 2.6403212547302246, + "learning_rate": 1.8264195175139705e-05, + "loss": 2.537436676025391, + "step": 32340 + }, + { + "epoch": 0.0880903770246359, + "grad_norm": 2.1677987575531006, + "learning_rate": 1.826310481123075e-05, + "loss": 2.5660091400146485, + "step": 32360 + }, + { + "epoch": 0.08814482101538043, + "grad_norm": 2.2372679710388184, + "learning_rate": 1.8262014447321793e-05, + "loss": 2.575030517578125, + "step": 32380 + }, + { + "epoch": 0.08819926500612495, + "grad_norm": 2.241821765899658, + "learning_rate": 1.8260924083412842e-05, + "loss": 2.5425785064697264, + "step": 32400 + }, + { + "epoch": 0.08825370899686948, + "grad_norm": 2.263479709625244, + "learning_rate": 1.8259833719503885e-05, + "loss": 2.556888961791992, + "step": 32420 + }, + { + "epoch": 0.08830815298761399, + "grad_norm": 2.2104198932647705, + "learning_rate": 1.8258743355594934e-05, + "loss": 2.552366828918457, + "step": 32440 + }, + { + "epoch": 0.0883625969783585, + "grad_norm": 2.231224775314331, + "learning_rate": 1.8257652991685976e-05, + "loss": 2.525859260559082, + "step": 32460 + }, + { + "epoch": 0.08841704096910304, + "grad_norm": 2.128645658493042, + "learning_rate": 1.8256562627777022e-05, + "loss": 2.570873260498047, + "step": 32480 + }, + { + "epoch": 0.08847148495984755, + "grad_norm": 2.042511224746704, + "learning_rate": 1.8255472263868068e-05, + "loss": 2.563824462890625, + "step": 32500 + }, + { + "epoch": 0.08852592895059208, + "grad_norm": 2.1186301708221436, + "learning_rate": 1.8254381899959114e-05, + "loss": 2.5608001708984376, + "step": 32520 + }, + { + "epoch": 0.0885803729413366, + "grad_norm": 2.22710919380188, + "learning_rate": 1.8253291536050156e-05, + "loss": 2.543324089050293, + "step": 32540 + }, + { + "epoch": 0.08863481693208113, + "grad_norm": 2.12424635887146, + "learning_rate": 1.8252201172141205e-05, + "loss": 2.5553256988525392, + "step": 32560 + }, + { + "epoch": 0.08868926092282564, + "grad_norm": 2.2764577865600586, + "learning_rate": 1.8251110808232248e-05, + "loss": 2.5712764739990233, + "step": 32580 + }, + { + "epoch": 0.08874370491357017, + "grad_norm": 2.3979897499084473, + "learning_rate": 1.8250020444323297e-05, + "loss": 2.558686065673828, + "step": 32600 + }, + { + "epoch": 0.08879814890431469, + "grad_norm": 2.2568399906158447, + "learning_rate": 1.824893008041434e-05, + "loss": 2.535037040710449, + "step": 32620 + }, + { + "epoch": 0.0888525928950592, + "grad_norm": 2.090498924255371, + "learning_rate": 1.8247839716505385e-05, + "loss": 2.569402313232422, + "step": 32640 + }, + { + "epoch": 0.08890703688580373, + "grad_norm": 2.6250152587890625, + "learning_rate": 1.824674935259643e-05, + "loss": 2.5693571090698244, + "step": 32660 + }, + { + "epoch": 0.08896148087654825, + "grad_norm": 2.1588456630706787, + "learning_rate": 1.8245658988687477e-05, + "loss": 2.5482826232910156, + "step": 32680 + }, + { + "epoch": 0.08901592486729278, + "grad_norm": 2.0658392906188965, + "learning_rate": 1.824456862477852e-05, + "loss": 2.531158447265625, + "step": 32700 + }, + { + "epoch": 0.08907036885803729, + "grad_norm": 2.2398412227630615, + "learning_rate": 1.824347826086957e-05, + "loss": 2.5204057693481445, + "step": 32720 + }, + { + "epoch": 0.08912481284878182, + "grad_norm": 2.1769258975982666, + "learning_rate": 1.824238789696061e-05, + "loss": 2.55578727722168, + "step": 32740 + }, + { + "epoch": 0.08917925683952634, + "grad_norm": 2.160175323486328, + "learning_rate": 1.8241297533051657e-05, + "loss": 2.544185829162598, + "step": 32760 + }, + { + "epoch": 0.08923370083027087, + "grad_norm": 2.205498456954956, + "learning_rate": 1.8240207169142703e-05, + "loss": 2.549402046203613, + "step": 32780 + }, + { + "epoch": 0.08928814482101538, + "grad_norm": 2.3562371730804443, + "learning_rate": 1.823911680523375e-05, + "loss": 2.560978317260742, + "step": 32800 + }, + { + "epoch": 0.0893425888117599, + "grad_norm": 2.2601664066314697, + "learning_rate": 1.8238026441324794e-05, + "loss": 2.5669181823730467, + "step": 32820 + }, + { + "epoch": 0.08939703280250443, + "grad_norm": 2.1903772354125977, + "learning_rate": 1.823693607741584e-05, + "loss": 2.555528450012207, + "step": 32840 + }, + { + "epoch": 0.08945147679324894, + "grad_norm": 2.2476601600646973, + "learning_rate": 1.8235845713506882e-05, + "loss": 2.5482295989990233, + "step": 32860 + }, + { + "epoch": 0.08950592078399347, + "grad_norm": 2.212416887283325, + "learning_rate": 1.823475534959793e-05, + "loss": 2.5767404556274416, + "step": 32880 + }, + { + "epoch": 0.08956036477473799, + "grad_norm": 2.157853126525879, + "learning_rate": 1.8233664985688974e-05, + "loss": 2.5436683654785157, + "step": 32900 + }, + { + "epoch": 0.08961480876548251, + "grad_norm": 2.1392993927001953, + "learning_rate": 1.823257462178002e-05, + "loss": 2.5486440658569336, + "step": 32920 + }, + { + "epoch": 0.08966925275622703, + "grad_norm": 2.1313726902008057, + "learning_rate": 1.8231484257871066e-05, + "loss": 2.5861417770385744, + "step": 32940 + }, + { + "epoch": 0.08972369674697156, + "grad_norm": 2.184736967086792, + "learning_rate": 1.823039389396211e-05, + "loss": 2.5645612716674804, + "step": 32960 + }, + { + "epoch": 0.08977814073771608, + "grad_norm": 2.1987078189849854, + "learning_rate": 1.8229303530053157e-05, + "loss": 2.569985198974609, + "step": 32980 + }, + { + "epoch": 0.08983258472846059, + "grad_norm": 2.090613842010498, + "learning_rate": 1.8228213166144203e-05, + "loss": 2.568836212158203, + "step": 33000 + }, + { + "epoch": 0.08988702871920512, + "grad_norm": 2.162984848022461, + "learning_rate": 1.8227122802235245e-05, + "loss": 2.5729063034057615, + "step": 33020 + }, + { + "epoch": 0.08994147270994964, + "grad_norm": 2.236701488494873, + "learning_rate": 1.8226032438326295e-05, + "loss": 2.5189537048339843, + "step": 33040 + }, + { + "epoch": 0.08999591670069416, + "grad_norm": 2.1353936195373535, + "learning_rate": 1.8224942074417337e-05, + "loss": 2.5406316757202148, + "step": 33060 + }, + { + "epoch": 0.09005036069143868, + "grad_norm": 2.4936883449554443, + "learning_rate": 1.8223851710508383e-05, + "loss": 2.554646301269531, + "step": 33080 + }, + { + "epoch": 0.09010480468218321, + "grad_norm": 2.1557226181030273, + "learning_rate": 1.822276134659943e-05, + "loss": 2.5438970565795898, + "step": 33100 + }, + { + "epoch": 0.09015924867292772, + "grad_norm": 2.1141650676727295, + "learning_rate": 1.8221670982690475e-05, + "loss": 2.5453655242919924, + "step": 33120 + }, + { + "epoch": 0.09021369266367224, + "grad_norm": 2.3960647583007812, + "learning_rate": 1.822058061878152e-05, + "loss": 2.5467405319213867, + "step": 33140 + }, + { + "epoch": 0.09026813665441677, + "grad_norm": 2.102341651916504, + "learning_rate": 1.8219490254872566e-05, + "loss": 2.542637825012207, + "step": 33160 + }, + { + "epoch": 0.09032258064516129, + "grad_norm": 2.137725353240967, + "learning_rate": 1.821839989096361e-05, + "loss": 2.5833551406860353, + "step": 33180 + }, + { + "epoch": 0.09037702463590581, + "grad_norm": 2.1211843490600586, + "learning_rate": 1.8217309527054658e-05, + "loss": 2.5763309478759764, + "step": 33200 + }, + { + "epoch": 0.09043146862665033, + "grad_norm": 2.1989634037017822, + "learning_rate": 1.82162191631457e-05, + "loss": 2.5582815170288087, + "step": 33220 + }, + { + "epoch": 0.09048591261739486, + "grad_norm": 3.7142622470855713, + "learning_rate": 1.8215128799236746e-05, + "loss": 2.551350212097168, + "step": 33240 + }, + { + "epoch": 0.09054035660813937, + "grad_norm": 2.062344789505005, + "learning_rate": 1.8214038435327792e-05, + "loss": 2.5675315856933594, + "step": 33260 + }, + { + "epoch": 0.0905948005988839, + "grad_norm": 2.1875216960906982, + "learning_rate": 1.8212948071418838e-05, + "loss": 2.5337963104248047, + "step": 33280 + }, + { + "epoch": 0.09064924458962842, + "grad_norm": 2.3395495414733887, + "learning_rate": 1.8211857707509883e-05, + "loss": 2.5549488067626953, + "step": 33300 + }, + { + "epoch": 0.09070368858037293, + "grad_norm": 2.120440721511841, + "learning_rate": 1.821076734360093e-05, + "loss": 2.603342628479004, + "step": 33320 + }, + { + "epoch": 0.09075813257111746, + "grad_norm": 2.0849153995513916, + "learning_rate": 1.8209676979691975e-05, + "loss": 2.5334239959716798, + "step": 33340 + }, + { + "epoch": 0.09081257656186198, + "grad_norm": 2.115413188934326, + "learning_rate": 1.820858661578302e-05, + "loss": 2.567546844482422, + "step": 33360 + }, + { + "epoch": 0.09086702055260651, + "grad_norm": 2.191404104232788, + "learning_rate": 1.8207496251874063e-05, + "loss": 2.542947769165039, + "step": 33380 + }, + { + "epoch": 0.09092146454335102, + "grad_norm": 2.1142044067382812, + "learning_rate": 1.820640588796511e-05, + "loss": 2.535161018371582, + "step": 33400 + }, + { + "epoch": 0.09097590853409555, + "grad_norm": 2.0874404907226562, + "learning_rate": 1.8205315524056155e-05, + "loss": 2.5429744720458984, + "step": 33420 + }, + { + "epoch": 0.09103035252484007, + "grad_norm": 2.1200814247131348, + "learning_rate": 1.82042251601472e-05, + "loss": 2.562647247314453, + "step": 33440 + }, + { + "epoch": 0.0910847965155846, + "grad_norm": 2.0716772079467773, + "learning_rate": 1.8203134796238246e-05, + "loss": 2.569462203979492, + "step": 33460 + }, + { + "epoch": 0.09113924050632911, + "grad_norm": 2.1919167041778564, + "learning_rate": 1.8202044432329292e-05, + "loss": 2.558748245239258, + "step": 33480 + }, + { + "epoch": 0.09119368449707363, + "grad_norm": 2.1556971073150635, + "learning_rate": 1.8200954068420338e-05, + "loss": 2.563668441772461, + "step": 33500 + }, + { + "epoch": 0.09124812848781816, + "grad_norm": 2.099142074584961, + "learning_rate": 1.8199863704511384e-05, + "loss": 2.5455272674560545, + "step": 33520 + }, + { + "epoch": 0.09130257247856267, + "grad_norm": 2.176589012145996, + "learning_rate": 1.8198773340602426e-05, + "loss": 2.5771509170532227, + "step": 33540 + }, + { + "epoch": 0.0913570164693072, + "grad_norm": 2.1533637046813965, + "learning_rate": 1.8197682976693472e-05, + "loss": 2.5530330657958986, + "step": 33560 + }, + { + "epoch": 0.09141146046005172, + "grad_norm": 2.3413479328155518, + "learning_rate": 1.8196592612784518e-05, + "loss": 2.5363080978393553, + "step": 33580 + }, + { + "epoch": 0.09146590445079625, + "grad_norm": 2.061610698699951, + "learning_rate": 1.8195502248875564e-05, + "loss": 2.5577295303344725, + "step": 33600 + }, + { + "epoch": 0.09152034844154076, + "grad_norm": 2.104210615158081, + "learning_rate": 1.819441188496661e-05, + "loss": 2.5649967193603516, + "step": 33620 + }, + { + "epoch": 0.09157479243228529, + "grad_norm": 2.069999933242798, + "learning_rate": 1.8193321521057655e-05, + "loss": 2.5537668228149415, + "step": 33640 + }, + { + "epoch": 0.09162923642302981, + "grad_norm": 2.252351760864258, + "learning_rate": 1.81922311571487e-05, + "loss": 2.5603023529052735, + "step": 33660 + }, + { + "epoch": 0.09168368041377432, + "grad_norm": 2.1027657985687256, + "learning_rate": 1.8191140793239747e-05, + "loss": 2.5430042266845705, + "step": 33680 + }, + { + "epoch": 0.09173812440451885, + "grad_norm": 2.2668721675872803, + "learning_rate": 1.819005042933079e-05, + "loss": 2.5427623748779298, + "step": 33700 + }, + { + "epoch": 0.09179256839526337, + "grad_norm": 2.128202438354492, + "learning_rate": 1.8188960065421835e-05, + "loss": 2.538819122314453, + "step": 33720 + }, + { + "epoch": 0.0918470123860079, + "grad_norm": 3.604358196258545, + "learning_rate": 1.818786970151288e-05, + "loss": 2.5411590576171874, + "step": 33740 + }, + { + "epoch": 0.09190145637675241, + "grad_norm": 2.175503730773926, + "learning_rate": 1.8186779337603927e-05, + "loss": 2.5417255401611327, + "step": 33760 + }, + { + "epoch": 0.09195590036749694, + "grad_norm": 2.072805166244507, + "learning_rate": 1.8185688973694973e-05, + "loss": 2.5663944244384767, + "step": 33780 + }, + { + "epoch": 0.09201034435824146, + "grad_norm": 2.1470868587493896, + "learning_rate": 1.818459860978602e-05, + "loss": 2.5607776641845703, + "step": 33800 + }, + { + "epoch": 0.09206478834898599, + "grad_norm": 2.1075351238250732, + "learning_rate": 1.8183508245877064e-05, + "loss": 2.545310401916504, + "step": 33820 + }, + { + "epoch": 0.0921192323397305, + "grad_norm": 2.4756946563720703, + "learning_rate": 1.818241788196811e-05, + "loss": 2.562882423400879, + "step": 33840 + }, + { + "epoch": 0.09217367633047502, + "grad_norm": 2.1131162643432617, + "learning_rate": 1.8181327518059153e-05, + "loss": 2.546308708190918, + "step": 33860 + }, + { + "epoch": 0.09222812032121955, + "grad_norm": 2.1554248332977295, + "learning_rate": 1.81802371541502e-05, + "loss": 2.5408231735229494, + "step": 33880 + }, + { + "epoch": 0.09228256431196406, + "grad_norm": 2.392078399658203, + "learning_rate": 1.8179146790241244e-05, + "loss": 2.5588645935058594, + "step": 33900 + }, + { + "epoch": 0.09233700830270859, + "grad_norm": 2.191007375717163, + "learning_rate": 1.817805642633229e-05, + "loss": 2.5547178268432615, + "step": 33920 + }, + { + "epoch": 0.09239145229345311, + "grad_norm": 2.0241713523864746, + "learning_rate": 1.8176966062423336e-05, + "loss": 2.5255910873413088, + "step": 33940 + }, + { + "epoch": 0.09244589628419764, + "grad_norm": 2.220369815826416, + "learning_rate": 1.817587569851438e-05, + "loss": 2.5425399780273437, + "step": 33960 + }, + { + "epoch": 0.09250034027494215, + "grad_norm": 2.1231191158294678, + "learning_rate": 1.8174785334605427e-05, + "loss": 2.5172613143920897, + "step": 33980 + }, + { + "epoch": 0.09255478426568667, + "grad_norm": 2.230025291442871, + "learning_rate": 1.8173694970696473e-05, + "loss": 2.54894905090332, + "step": 34000 + }, + { + "epoch": 0.0926092282564312, + "grad_norm": 2.157383918762207, + "learning_rate": 1.8172604606787516e-05, + "loss": 2.545298957824707, + "step": 34020 + }, + { + "epoch": 0.09266367224717571, + "grad_norm": 2.629673719406128, + "learning_rate": 1.817151424287856e-05, + "loss": 2.5486494064331056, + "step": 34040 + }, + { + "epoch": 0.09271811623792024, + "grad_norm": 2.116257667541504, + "learning_rate": 1.8170423878969607e-05, + "loss": 2.5669540405273437, + "step": 34060 + }, + { + "epoch": 0.09277256022866476, + "grad_norm": 2.186666488647461, + "learning_rate": 1.8169333515060653e-05, + "loss": 2.5741479873657225, + "step": 34080 + }, + { + "epoch": 0.09282700421940929, + "grad_norm": 2.2552154064178467, + "learning_rate": 1.81682431511517e-05, + "loss": 2.5647228240966795, + "step": 34100 + }, + { + "epoch": 0.0928814482101538, + "grad_norm": 2.280017375946045, + "learning_rate": 1.8167152787242745e-05, + "loss": 2.549831581115723, + "step": 34120 + }, + { + "epoch": 0.09293589220089833, + "grad_norm": 2.3307504653930664, + "learning_rate": 1.816606242333379e-05, + "loss": 2.5443363189697266, + "step": 34140 + }, + { + "epoch": 0.09299033619164285, + "grad_norm": 2.152290105819702, + "learning_rate": 1.8164972059424836e-05, + "loss": 2.5493844985961913, + "step": 34160 + }, + { + "epoch": 0.09304478018238736, + "grad_norm": 2.133167266845703, + "learning_rate": 1.816388169551588e-05, + "loss": 2.602106285095215, + "step": 34180 + }, + { + "epoch": 0.09309922417313189, + "grad_norm": 2.3015010356903076, + "learning_rate": 1.8162791331606924e-05, + "loss": 2.5609308242797852, + "step": 34200 + }, + { + "epoch": 0.09315366816387641, + "grad_norm": 2.4964656829833984, + "learning_rate": 1.816170096769797e-05, + "loss": 2.5788619995117186, + "step": 34220 + }, + { + "epoch": 0.09320811215462094, + "grad_norm": 2.1556308269500732, + "learning_rate": 1.8160610603789016e-05, + "loss": 2.551615905761719, + "step": 34240 + }, + { + "epoch": 0.09326255614536545, + "grad_norm": 2.082395315170288, + "learning_rate": 1.8159520239880062e-05, + "loss": 2.56622314453125, + "step": 34260 + }, + { + "epoch": 0.09331700013610998, + "grad_norm": 2.0837440490722656, + "learning_rate": 1.8158429875971108e-05, + "loss": 2.5252403259277343, + "step": 34280 + }, + { + "epoch": 0.0933714441268545, + "grad_norm": 2.1228225231170654, + "learning_rate": 1.8157339512062154e-05, + "loss": 2.583556365966797, + "step": 34300 + }, + { + "epoch": 0.09342588811759903, + "grad_norm": 2.2605676651000977, + "learning_rate": 1.81562491481532e-05, + "loss": 2.5572649002075196, + "step": 34320 + }, + { + "epoch": 0.09348033210834354, + "grad_norm": 3.8847362995147705, + "learning_rate": 1.8155158784244242e-05, + "loss": 2.5376918792724608, + "step": 34340 + }, + { + "epoch": 0.09353477609908806, + "grad_norm": 2.1282050609588623, + "learning_rate": 1.8154068420335288e-05, + "loss": 2.552316665649414, + "step": 34360 + }, + { + "epoch": 0.09358922008983259, + "grad_norm": 2.120448589324951, + "learning_rate": 1.8152978056426333e-05, + "loss": 2.5452043533325197, + "step": 34380 + }, + { + "epoch": 0.0936436640805771, + "grad_norm": 2.240276336669922, + "learning_rate": 1.815188769251738e-05, + "loss": 2.5536394119262695, + "step": 34400 + }, + { + "epoch": 0.09369810807132163, + "grad_norm": 2.121575355529785, + "learning_rate": 1.8150797328608425e-05, + "loss": 2.5761081695556642, + "step": 34420 + }, + { + "epoch": 0.09375255206206615, + "grad_norm": 2.2086048126220703, + "learning_rate": 1.814970696469947e-05, + "loss": 2.5415477752685547, + "step": 34440 + }, + { + "epoch": 0.09380699605281068, + "grad_norm": 2.121694803237915, + "learning_rate": 1.8148616600790517e-05, + "loss": 2.5612411499023438, + "step": 34460 + }, + { + "epoch": 0.09386144004355519, + "grad_norm": 2.2300093173980713, + "learning_rate": 1.8147526236881562e-05, + "loss": 2.534982109069824, + "step": 34480 + }, + { + "epoch": 0.09391588403429972, + "grad_norm": 2.14241099357605, + "learning_rate": 1.8146435872972605e-05, + "loss": 2.5579959869384767, + "step": 34500 + }, + { + "epoch": 0.09397032802504424, + "grad_norm": 2.15287184715271, + "learning_rate": 1.814534550906365e-05, + "loss": 2.556717872619629, + "step": 34520 + }, + { + "epoch": 0.09402477201578875, + "grad_norm": 2.1037511825561523, + "learning_rate": 1.8144255145154696e-05, + "loss": 2.567482566833496, + "step": 34540 + }, + { + "epoch": 0.09407921600653328, + "grad_norm": 2.143540620803833, + "learning_rate": 1.8143164781245742e-05, + "loss": 2.5421119689941407, + "step": 34560 + }, + { + "epoch": 0.0941336599972778, + "grad_norm": 2.163196325302124, + "learning_rate": 1.8142074417336788e-05, + "loss": 2.543865966796875, + "step": 34580 + }, + { + "epoch": 0.09418810398802233, + "grad_norm": 2.1565611362457275, + "learning_rate": 1.8140984053427834e-05, + "loss": 2.58160457611084, + "step": 34600 + }, + { + "epoch": 0.09424254797876684, + "grad_norm": 2.1960439682006836, + "learning_rate": 1.813989368951888e-05, + "loss": 2.549625587463379, + "step": 34620 + }, + { + "epoch": 0.09429699196951137, + "grad_norm": 2.2413582801818848, + "learning_rate": 1.8138803325609926e-05, + "loss": 2.5770973205566405, + "step": 34640 + }, + { + "epoch": 0.09435143596025589, + "grad_norm": 2.84073543548584, + "learning_rate": 1.8137712961700968e-05, + "loss": 2.548320770263672, + "step": 34660 + }, + { + "epoch": 0.0944058799510004, + "grad_norm": 2.2027738094329834, + "learning_rate": 1.8136622597792014e-05, + "loss": 2.5491222381591796, + "step": 34680 + }, + { + "epoch": 0.09446032394174493, + "grad_norm": 2.2260119915008545, + "learning_rate": 1.813553223388306e-05, + "loss": 2.540963554382324, + "step": 34700 + }, + { + "epoch": 0.09451476793248945, + "grad_norm": 2.1963446140289307, + "learning_rate": 1.8134441869974105e-05, + "loss": 2.5308624267578126, + "step": 34720 + }, + { + "epoch": 0.09456921192323398, + "grad_norm": 2.090190887451172, + "learning_rate": 1.813335150606515e-05, + "loss": 2.5244892120361326, + "step": 34740 + }, + { + "epoch": 0.09462365591397849, + "grad_norm": 2.126889944076538, + "learning_rate": 1.8132261142156197e-05, + "loss": 2.5211971282958983, + "step": 34760 + }, + { + "epoch": 0.09467809990472302, + "grad_norm": 2.1398086547851562, + "learning_rate": 1.8131170778247243e-05, + "loss": 2.5444290161132814, + "step": 34780 + }, + { + "epoch": 0.09473254389546754, + "grad_norm": 2.172711133956909, + "learning_rate": 1.8130080414338285e-05, + "loss": 2.545436477661133, + "step": 34800 + }, + { + "epoch": 0.09478698788621207, + "grad_norm": 2.115312099456787, + "learning_rate": 1.812899005042933e-05, + "loss": 2.5369558334350586, + "step": 34820 + }, + { + "epoch": 0.09484143187695658, + "grad_norm": 2.0631263256073, + "learning_rate": 1.8127899686520377e-05, + "loss": 2.5776256561279296, + "step": 34840 + }, + { + "epoch": 0.0948958758677011, + "grad_norm": 2.165421962738037, + "learning_rate": 1.8126809322611423e-05, + "loss": 2.5678070068359373, + "step": 34860 + }, + { + "epoch": 0.09495031985844563, + "grad_norm": 2.1522445678710938, + "learning_rate": 1.812571895870247e-05, + "loss": 2.5304737091064453, + "step": 34880 + }, + { + "epoch": 0.09500476384919014, + "grad_norm": 2.271676540374756, + "learning_rate": 1.8124628594793514e-05, + "loss": 2.5633914947509764, + "step": 34900 + }, + { + "epoch": 0.09505920783993467, + "grad_norm": 2.1127774715423584, + "learning_rate": 1.812353823088456e-05, + "loss": 2.531209182739258, + "step": 34920 + }, + { + "epoch": 0.09511365183067919, + "grad_norm": 2.1224892139434814, + "learning_rate": 1.8122447866975606e-05, + "loss": 2.5445930480957033, + "step": 34940 + }, + { + "epoch": 0.09516809582142371, + "grad_norm": 2.815861225128174, + "learning_rate": 1.812135750306665e-05, + "loss": 2.564535713195801, + "step": 34960 + }, + { + "epoch": 0.09522253981216823, + "grad_norm": 2.1927194595336914, + "learning_rate": 1.8120267139157694e-05, + "loss": 2.5454517364501954, + "step": 34980 + }, + { + "epoch": 0.09527698380291276, + "grad_norm": 2.23807954788208, + "learning_rate": 1.811917677524874e-05, + "loss": 2.551741027832031, + "step": 35000 + }, + { + "epoch": 0.09533142779365728, + "grad_norm": 3.1072704792022705, + "learning_rate": 1.8118086411339786e-05, + "loss": 2.5488510131835938, + "step": 35020 + }, + { + "epoch": 0.09538587178440179, + "grad_norm": 2.173774003982544, + "learning_rate": 1.811699604743083e-05, + "loss": 2.5396804809570312, + "step": 35040 + }, + { + "epoch": 0.09544031577514632, + "grad_norm": 2.1473135948181152, + "learning_rate": 1.8115905683521877e-05, + "loss": 2.526203918457031, + "step": 35060 + }, + { + "epoch": 0.09549475976589084, + "grad_norm": 5.6967058181762695, + "learning_rate": 1.8114815319612923e-05, + "loss": 2.553116226196289, + "step": 35080 + }, + { + "epoch": 0.09554920375663536, + "grad_norm": 2.1797683238983154, + "learning_rate": 1.811372495570397e-05, + "loss": 2.5421031951904296, + "step": 35100 + }, + { + "epoch": 0.09560364774737988, + "grad_norm": 2.129875421524048, + "learning_rate": 1.811263459179501e-05, + "loss": 2.5658027648925783, + "step": 35120 + }, + { + "epoch": 0.09565809173812441, + "grad_norm": 2.0516364574432373, + "learning_rate": 1.8111544227886057e-05, + "loss": 2.5501869201660154, + "step": 35140 + }, + { + "epoch": 0.09571253572886892, + "grad_norm": 2.222895860671997, + "learning_rate": 1.8110453863977103e-05, + "loss": 2.5638412475585937, + "step": 35160 + }, + { + "epoch": 0.09576697971961345, + "grad_norm": 2.1698062419891357, + "learning_rate": 1.810936350006815e-05, + "loss": 2.5458518981933596, + "step": 35180 + }, + { + "epoch": 0.09582142371035797, + "grad_norm": 2.0791497230529785, + "learning_rate": 1.8108273136159195e-05, + "loss": 2.5573589324951174, + "step": 35200 + }, + { + "epoch": 0.09587586770110249, + "grad_norm": 2.2149012088775635, + "learning_rate": 1.810718277225024e-05, + "loss": 2.5645050048828124, + "step": 35220 + }, + { + "epoch": 0.09593031169184701, + "grad_norm": 2.122610330581665, + "learning_rate": 1.8106092408341286e-05, + "loss": 2.542227554321289, + "step": 35240 + }, + { + "epoch": 0.09598475568259153, + "grad_norm": 2.0796656608581543, + "learning_rate": 1.8105002044432332e-05, + "loss": 2.5481731414794924, + "step": 35260 + }, + { + "epoch": 0.09603919967333606, + "grad_norm": 2.24880051612854, + "learning_rate": 1.8103911680523374e-05, + "loss": 2.5766983032226562, + "step": 35280 + }, + { + "epoch": 0.09609364366408057, + "grad_norm": 2.0986216068267822, + "learning_rate": 1.810282131661442e-05, + "loss": 2.5841712951660156, + "step": 35300 + }, + { + "epoch": 0.0961480876548251, + "grad_norm": 2.194265842437744, + "learning_rate": 1.8101730952705466e-05, + "loss": 2.557354736328125, + "step": 35320 + }, + { + "epoch": 0.09620253164556962, + "grad_norm": 2.281637191772461, + "learning_rate": 1.8100640588796512e-05, + "loss": 2.5661481857299804, + "step": 35340 + }, + { + "epoch": 0.09625697563631413, + "grad_norm": 2.1624886989593506, + "learning_rate": 1.8099550224887558e-05, + "loss": 2.5556404113769533, + "step": 35360 + }, + { + "epoch": 0.09631141962705866, + "grad_norm": 2.2903800010681152, + "learning_rate": 1.8098459860978604e-05, + "loss": 2.5312820434570313, + "step": 35380 + }, + { + "epoch": 0.09636586361780318, + "grad_norm": 2.16562819480896, + "learning_rate": 1.809736949706965e-05, + "loss": 2.566303253173828, + "step": 35400 + }, + { + "epoch": 0.09642030760854771, + "grad_norm": 2.184201717376709, + "learning_rate": 1.8096279133160695e-05, + "loss": 2.538151741027832, + "step": 35420 + }, + { + "epoch": 0.09647475159929222, + "grad_norm": 2.138658046722412, + "learning_rate": 1.8095188769251738e-05, + "loss": 2.5345199584960936, + "step": 35440 + }, + { + "epoch": 0.09652919559003675, + "grad_norm": 2.3361732959747314, + "learning_rate": 1.8094098405342783e-05, + "loss": 2.5599756240844727, + "step": 35460 + }, + { + "epoch": 0.09658363958078127, + "grad_norm": 2.1488540172576904, + "learning_rate": 1.809300804143383e-05, + "loss": 2.588279151916504, + "step": 35480 + }, + { + "epoch": 0.0966380835715258, + "grad_norm": 2.144557476043701, + "learning_rate": 1.8091917677524875e-05, + "loss": 2.592884063720703, + "step": 35500 + }, + { + "epoch": 0.09669252756227031, + "grad_norm": 2.163027763366699, + "learning_rate": 1.809082731361592e-05, + "loss": 2.551879119873047, + "step": 35520 + }, + { + "epoch": 0.09674697155301483, + "grad_norm": 2.427961826324463, + "learning_rate": 1.8089736949706967e-05, + "loss": 2.5758914947509766, + "step": 35540 + }, + { + "epoch": 0.09680141554375936, + "grad_norm": 2.116265296936035, + "learning_rate": 1.8088646585798012e-05, + "loss": 2.5425071716308594, + "step": 35560 + }, + { + "epoch": 0.09685585953450387, + "grad_norm": 2.2174649238586426, + "learning_rate": 1.8087556221889058e-05, + "loss": 2.5137205123901367, + "step": 35580 + }, + { + "epoch": 0.0969103035252484, + "grad_norm": 2.1454567909240723, + "learning_rate": 1.80864658579801e-05, + "loss": 2.5785776138305665, + "step": 35600 + }, + { + "epoch": 0.09696474751599292, + "grad_norm": 2.1463263034820557, + "learning_rate": 1.8085375494071146e-05, + "loss": 2.5655950546264648, + "step": 35620 + }, + { + "epoch": 0.09701919150673745, + "grad_norm": 2.1594717502593994, + "learning_rate": 1.8084285130162192e-05, + "loss": 2.5438133239746095, + "step": 35640 + }, + { + "epoch": 0.09707363549748196, + "grad_norm": 2.142590284347534, + "learning_rate": 1.8083194766253238e-05, + "loss": 2.5569318771362304, + "step": 35660 + }, + { + "epoch": 0.0971280794882265, + "grad_norm": 2.1129684448242188, + "learning_rate": 1.8082104402344284e-05, + "loss": 2.5767749786376952, + "step": 35680 + }, + { + "epoch": 0.09718252347897101, + "grad_norm": 2.4620280265808105, + "learning_rate": 1.808101403843533e-05, + "loss": 2.540994071960449, + "step": 35700 + }, + { + "epoch": 0.09723696746971552, + "grad_norm": 2.145508050918579, + "learning_rate": 1.8079923674526376e-05, + "loss": 2.556229019165039, + "step": 35720 + }, + { + "epoch": 0.09729141146046005, + "grad_norm": 2.197018623352051, + "learning_rate": 1.807883331061742e-05, + "loss": 2.518297576904297, + "step": 35740 + }, + { + "epoch": 0.09734585545120457, + "grad_norm": 2.2718381881713867, + "learning_rate": 1.8077742946708464e-05, + "loss": 2.5648523330688477, + "step": 35760 + }, + { + "epoch": 0.0974002994419491, + "grad_norm": 2.118774652481079, + "learning_rate": 1.807665258279951e-05, + "loss": 2.56771240234375, + "step": 35780 + }, + { + "epoch": 0.09745474343269361, + "grad_norm": 2.135432720184326, + "learning_rate": 1.8075562218890555e-05, + "loss": 2.5526859283447267, + "step": 35800 + }, + { + "epoch": 0.09750918742343814, + "grad_norm": 2.1618380546569824, + "learning_rate": 1.80744718549816e-05, + "loss": 2.5414907455444338, + "step": 35820 + }, + { + "epoch": 0.09756363141418266, + "grad_norm": 2.1826345920562744, + "learning_rate": 1.8073381491072647e-05, + "loss": 2.5328344345092773, + "step": 35840 + }, + { + "epoch": 0.09761807540492719, + "grad_norm": 2.0831291675567627, + "learning_rate": 1.8072291127163693e-05, + "loss": 2.573708724975586, + "step": 35860 + }, + { + "epoch": 0.0976725193956717, + "grad_norm": 2.2194225788116455, + "learning_rate": 1.807120076325474e-05, + "loss": 2.579145050048828, + "step": 35880 + }, + { + "epoch": 0.09772696338641622, + "grad_norm": 2.0795764923095703, + "learning_rate": 1.8070110399345784e-05, + "loss": 2.546634292602539, + "step": 35900 + }, + { + "epoch": 0.09778140737716075, + "grad_norm": 2.1551761627197266, + "learning_rate": 1.8069020035436827e-05, + "loss": 2.5553783416748046, + "step": 35920 + }, + { + "epoch": 0.09783585136790526, + "grad_norm": 2.2081596851348877, + "learning_rate": 1.8067929671527873e-05, + "loss": 2.5508844375610353, + "step": 35940 + }, + { + "epoch": 0.09789029535864979, + "grad_norm": 2.3719322681427, + "learning_rate": 1.806683930761892e-05, + "loss": 2.534630012512207, + "step": 35960 + }, + { + "epoch": 0.09794473934939431, + "grad_norm": 2.222127676010132, + "learning_rate": 1.8065748943709964e-05, + "loss": 2.5659963607788088, + "step": 35980 + }, + { + "epoch": 0.09799918334013884, + "grad_norm": 2.1607038974761963, + "learning_rate": 1.806465857980101e-05, + "loss": 2.546317481994629, + "step": 36000 + }, + { + "epoch": 0.09805362733088335, + "grad_norm": 2.4553122520446777, + "learning_rate": 1.8063568215892056e-05, + "loss": 2.5430978775024413, + "step": 36020 + }, + { + "epoch": 0.09810807132162788, + "grad_norm": 2.1295454502105713, + "learning_rate": 1.80624778519831e-05, + "loss": 2.5546812057495116, + "step": 36040 + }, + { + "epoch": 0.0981625153123724, + "grad_norm": 2.1460626125335693, + "learning_rate": 1.8061387488074147e-05, + "loss": 2.562511444091797, + "step": 36060 + }, + { + "epoch": 0.09821695930311691, + "grad_norm": 2.269110918045044, + "learning_rate": 1.806029712416519e-05, + "loss": 2.6042261123657227, + "step": 36080 + }, + { + "epoch": 0.09827140329386144, + "grad_norm": 2.131653070449829, + "learning_rate": 1.8059206760256236e-05, + "loss": 2.539511871337891, + "step": 36100 + }, + { + "epoch": 0.09832584728460596, + "grad_norm": 2.6612234115600586, + "learning_rate": 1.805811639634728e-05, + "loss": 2.517926788330078, + "step": 36120 + }, + { + "epoch": 0.09838029127535049, + "grad_norm": 3.022921323776245, + "learning_rate": 1.8057026032438327e-05, + "loss": 2.552167510986328, + "step": 36140 + }, + { + "epoch": 0.098434735266095, + "grad_norm": 2.0944879055023193, + "learning_rate": 1.8055935668529373e-05, + "loss": 2.5487110137939455, + "step": 36160 + }, + { + "epoch": 0.09848917925683953, + "grad_norm": 2.1155648231506348, + "learning_rate": 1.805484530462042e-05, + "loss": 2.5766218185424803, + "step": 36180 + }, + { + "epoch": 0.09854362324758405, + "grad_norm": 2.1215593814849854, + "learning_rate": 1.8053754940711465e-05, + "loss": 2.572320747375488, + "step": 36200 + }, + { + "epoch": 0.09859806723832856, + "grad_norm": 2.087834596633911, + "learning_rate": 1.805266457680251e-05, + "loss": 2.533958625793457, + "step": 36220 + }, + { + "epoch": 0.09865251122907309, + "grad_norm": 2.1750874519348145, + "learning_rate": 1.8051574212893553e-05, + "loss": 2.5400747299194335, + "step": 36240 + }, + { + "epoch": 0.09870695521981761, + "grad_norm": 3.346449613571167, + "learning_rate": 1.80504838489846e-05, + "loss": 2.5538375854492186, + "step": 36260 + }, + { + "epoch": 0.09876139921056214, + "grad_norm": 2.2274584770202637, + "learning_rate": 1.8049393485075645e-05, + "loss": 2.5408807754516602, + "step": 36280 + }, + { + "epoch": 0.09881584320130665, + "grad_norm": 2.0693423748016357, + "learning_rate": 1.804830312116669e-05, + "loss": 2.5409236907958985, + "step": 36300 + }, + { + "epoch": 0.09887028719205118, + "grad_norm": 2.0131795406341553, + "learning_rate": 1.8047212757257736e-05, + "loss": 2.555158805847168, + "step": 36320 + }, + { + "epoch": 0.0989247311827957, + "grad_norm": 2.0722858905792236, + "learning_rate": 1.8046122393348782e-05, + "loss": 2.5522159576416015, + "step": 36340 + }, + { + "epoch": 0.09897917517354023, + "grad_norm": 2.114981174468994, + "learning_rate": 1.8045032029439828e-05, + "loss": 2.562616157531738, + "step": 36360 + }, + { + "epoch": 0.09903361916428474, + "grad_norm": 2.297266960144043, + "learning_rate": 1.8043941665530874e-05, + "loss": 2.575385093688965, + "step": 36380 + }, + { + "epoch": 0.09908806315502926, + "grad_norm": 2.1166727542877197, + "learning_rate": 1.8042851301621916e-05, + "loss": 2.5529991149902345, + "step": 36400 + }, + { + "epoch": 0.09914250714577379, + "grad_norm": 3.253148317337036, + "learning_rate": 1.8041760937712962e-05, + "loss": 2.5082117080688477, + "step": 36420 + }, + { + "epoch": 0.0991969511365183, + "grad_norm": 2.186405658721924, + "learning_rate": 1.8040670573804008e-05, + "loss": 2.5462457656860353, + "step": 36440 + }, + { + "epoch": 0.09925139512726283, + "grad_norm": 6.786067485809326, + "learning_rate": 1.8039580209895054e-05, + "loss": 2.578941535949707, + "step": 36460 + }, + { + "epoch": 0.09930583911800735, + "grad_norm": 2.3471992015838623, + "learning_rate": 1.80384898459861e-05, + "loss": 2.5551380157470702, + "step": 36480 + }, + { + "epoch": 0.09936028310875188, + "grad_norm": 2.1627049446105957, + "learning_rate": 1.8037399482077145e-05, + "loss": 2.563740921020508, + "step": 36500 + }, + { + "epoch": 0.09941472709949639, + "grad_norm": 2.4986181259155273, + "learning_rate": 1.803630911816819e-05, + "loss": 2.540875244140625, + "step": 36520 + }, + { + "epoch": 0.09946917109024092, + "grad_norm": 2.109901189804077, + "learning_rate": 1.8035218754259237e-05, + "loss": 2.518585205078125, + "step": 36540 + }, + { + "epoch": 0.09952361508098544, + "grad_norm": 2.142338514328003, + "learning_rate": 1.803412839035028e-05, + "loss": 2.5578500747680666, + "step": 36560 + }, + { + "epoch": 0.09957805907172995, + "grad_norm": 2.3827593326568604, + "learning_rate": 1.8033038026441325e-05, + "loss": 2.564432907104492, + "step": 36580 + }, + { + "epoch": 0.09963250306247448, + "grad_norm": 2.1156704425811768, + "learning_rate": 1.803194766253237e-05, + "loss": 2.5755641937255858, + "step": 36600 + }, + { + "epoch": 0.099686947053219, + "grad_norm": 2.0849852561950684, + "learning_rate": 1.8030857298623417e-05, + "loss": 2.533620834350586, + "step": 36620 + }, + { + "epoch": 0.09974139104396353, + "grad_norm": 2.2998266220092773, + "learning_rate": 1.8029766934714462e-05, + "loss": 2.52593994140625, + "step": 36640 + }, + { + "epoch": 0.09979583503470804, + "grad_norm": 2.429089307785034, + "learning_rate": 1.8028676570805508e-05, + "loss": 2.5574644088745115, + "step": 36660 + }, + { + "epoch": 0.09985027902545257, + "grad_norm": 2.2308733463287354, + "learning_rate": 1.8027586206896554e-05, + "loss": 2.564302444458008, + "step": 36680 + }, + { + "epoch": 0.09990472301619709, + "grad_norm": 15.279032707214355, + "learning_rate": 1.80264958429876e-05, + "loss": 2.511400604248047, + "step": 36700 + }, + { + "epoch": 0.09995916700694162, + "grad_norm": 2.1003947257995605, + "learning_rate": 1.8025405479078642e-05, + "loss": 2.512722206115723, + "step": 36720 + }, + { + "epoch": 0.10001361099768613, + "grad_norm": 2.130995750427246, + "learning_rate": 1.802431511516969e-05, + "loss": 2.5457876205444334, + "step": 36740 + }, + { + "epoch": 0.10006805498843065, + "grad_norm": 2.207636594772339, + "learning_rate": 1.8023224751260734e-05, + "loss": 2.598371124267578, + "step": 36760 + }, + { + "epoch": 0.10012249897917518, + "grad_norm": 2.5180342197418213, + "learning_rate": 1.802213438735178e-05, + "loss": 2.5517898559570313, + "step": 36780 + }, + { + "epoch": 0.10017694296991969, + "grad_norm": 2.26369047164917, + "learning_rate": 1.8021044023442825e-05, + "loss": 2.534173583984375, + "step": 36800 + }, + { + "epoch": 0.10023138696066422, + "grad_norm": 2.1680307388305664, + "learning_rate": 1.801995365953387e-05, + "loss": 2.5643049240112306, + "step": 36820 + }, + { + "epoch": 0.10028583095140874, + "grad_norm": 2.1658754348754883, + "learning_rate": 1.8018863295624917e-05, + "loss": 2.4911949157714846, + "step": 36840 + }, + { + "epoch": 0.10034027494215327, + "grad_norm": 2.3736555576324463, + "learning_rate": 1.8017772931715963e-05, + "loss": 2.5156314849853514, + "step": 36860 + }, + { + "epoch": 0.10039471893289778, + "grad_norm": 2.1266562938690186, + "learning_rate": 1.8016682567807005e-05, + "loss": 2.561809539794922, + "step": 36880 + }, + { + "epoch": 0.1004491629236423, + "grad_norm": 2.092768430709839, + "learning_rate": 1.8015592203898055e-05, + "loss": 2.5636211395263673, + "step": 36900 + }, + { + "epoch": 0.10050360691438683, + "grad_norm": 2.199802875518799, + "learning_rate": 1.8014501839989097e-05, + "loss": 2.572914695739746, + "step": 36920 + }, + { + "epoch": 0.10055805090513134, + "grad_norm": 2.5835976600646973, + "learning_rate": 1.8013411476080143e-05, + "loss": 2.5351327896118163, + "step": 36940 + }, + { + "epoch": 0.10061249489587587, + "grad_norm": 2.1180472373962402, + "learning_rate": 1.801232111217119e-05, + "loss": 2.5456926345825197, + "step": 36960 + }, + { + "epoch": 0.10066693888662039, + "grad_norm": 2.3131842613220215, + "learning_rate": 1.8011230748262234e-05, + "loss": 2.5584264755249024, + "step": 36980 + }, + { + "epoch": 0.10072138287736492, + "grad_norm": 2.1786868572235107, + "learning_rate": 1.8010140384353277e-05, + "loss": 2.5270355224609373, + "step": 37000 + }, + { + "epoch": 0.10077582686810943, + "grad_norm": 2.1030616760253906, + "learning_rate": 1.8009050020444326e-05, + "loss": 2.545436668395996, + "step": 37020 + }, + { + "epoch": 0.10083027085885396, + "grad_norm": 2.4560306072235107, + "learning_rate": 1.800795965653537e-05, + "loss": 2.5670469284057615, + "step": 37040 + }, + { + "epoch": 0.10088471484959848, + "grad_norm": 2.1788370609283447, + "learning_rate": 1.8006869292626418e-05, + "loss": 2.5255035400390624, + "step": 37060 + }, + { + "epoch": 0.10093915884034299, + "grad_norm": 2.244699478149414, + "learning_rate": 1.800577892871746e-05, + "loss": 2.517462158203125, + "step": 37080 + }, + { + "epoch": 0.10099360283108752, + "grad_norm": 2.222896099090576, + "learning_rate": 1.8004688564808506e-05, + "loss": 2.5184867858886717, + "step": 37100 + }, + { + "epoch": 0.10104804682183204, + "grad_norm": 2.283137559890747, + "learning_rate": 1.800359820089955e-05, + "loss": 2.561541748046875, + "step": 37120 + }, + { + "epoch": 0.10110249081257656, + "grad_norm": 2.1690666675567627, + "learning_rate": 1.8002507836990597e-05, + "loss": 2.5595115661621093, + "step": 37140 + }, + { + "epoch": 0.10115693480332108, + "grad_norm": 2.1227924823760986, + "learning_rate": 1.800141747308164e-05, + "loss": 2.5373651504516603, + "step": 37160 + }, + { + "epoch": 0.10121137879406561, + "grad_norm": 2.0845160484313965, + "learning_rate": 1.800032710917269e-05, + "loss": 2.5393299102783202, + "step": 37180 + }, + { + "epoch": 0.10126582278481013, + "grad_norm": 2.1899638175964355, + "learning_rate": 1.799923674526373e-05, + "loss": 2.530426597595215, + "step": 37200 + }, + { + "epoch": 0.10132026677555465, + "grad_norm": 3.75636887550354, + "learning_rate": 1.799814638135478e-05, + "loss": 2.5380462646484374, + "step": 37220 + }, + { + "epoch": 0.10137471076629917, + "grad_norm": 2.2143359184265137, + "learning_rate": 1.7997056017445823e-05, + "loss": 2.513242721557617, + "step": 37240 + }, + { + "epoch": 0.10142915475704369, + "grad_norm": 2.2862284183502197, + "learning_rate": 1.799596565353687e-05, + "loss": 2.5899805068969726, + "step": 37260 + }, + { + "epoch": 0.10148359874778821, + "grad_norm": 2.172606945037842, + "learning_rate": 1.7994875289627915e-05, + "loss": 2.532843017578125, + "step": 37280 + }, + { + "epoch": 0.10153804273853273, + "grad_norm": 2.2176337242126465, + "learning_rate": 1.799378492571896e-05, + "loss": 2.5579395294189453, + "step": 37300 + }, + { + "epoch": 0.10159248672927726, + "grad_norm": 2.1729702949523926, + "learning_rate": 1.7992694561810003e-05, + "loss": 2.579197311401367, + "step": 37320 + }, + { + "epoch": 0.10164693072002177, + "grad_norm": 2.6819968223571777, + "learning_rate": 1.7991604197901052e-05, + "loss": 2.577828598022461, + "step": 37340 + }, + { + "epoch": 0.1017013747107663, + "grad_norm": 2.1354877948760986, + "learning_rate": 1.7990513833992095e-05, + "loss": 2.532953071594238, + "step": 37360 + }, + { + "epoch": 0.10175581870151082, + "grad_norm": 2.2292232513427734, + "learning_rate": 1.7989423470083144e-05, + "loss": 2.5678863525390625, + "step": 37380 + }, + { + "epoch": 0.10181026269225535, + "grad_norm": 2.0855958461761475, + "learning_rate": 1.7988333106174186e-05, + "loss": 2.545945930480957, + "step": 37400 + }, + { + "epoch": 0.10186470668299986, + "grad_norm": 3.515876293182373, + "learning_rate": 1.7987242742265232e-05, + "loss": 2.513616371154785, + "step": 37420 + }, + { + "epoch": 0.10191915067374438, + "grad_norm": 2.147745132446289, + "learning_rate": 1.7986152378356278e-05, + "loss": 2.5444278717041016, + "step": 37440 + }, + { + "epoch": 0.10197359466448891, + "grad_norm": 2.0699825286865234, + "learning_rate": 1.7985062014447324e-05, + "loss": 2.560342216491699, + "step": 37460 + }, + { + "epoch": 0.10202803865523342, + "grad_norm": 2.2418930530548096, + "learning_rate": 1.7983971650538366e-05, + "loss": 2.543430519104004, + "step": 37480 + }, + { + "epoch": 0.10208248264597795, + "grad_norm": 2.6466100215911865, + "learning_rate": 1.7982881286629415e-05, + "loss": 2.560530090332031, + "step": 37500 + }, + { + "epoch": 0.10213692663672247, + "grad_norm": 2.025939702987671, + "learning_rate": 1.7981790922720458e-05, + "loss": 2.5799785614013673, + "step": 37520 + }, + { + "epoch": 0.102191370627467, + "grad_norm": 2.3079721927642822, + "learning_rate": 1.7980700558811507e-05, + "loss": 2.5563076019287108, + "step": 37540 + }, + { + "epoch": 0.10224581461821151, + "grad_norm": 2.081698417663574, + "learning_rate": 1.797961019490255e-05, + "loss": 2.541326332092285, + "step": 37560 + }, + { + "epoch": 0.10230025860895604, + "grad_norm": 2.1253340244293213, + "learning_rate": 1.7978519830993595e-05, + "loss": 2.547895812988281, + "step": 37580 + }, + { + "epoch": 0.10235470259970056, + "grad_norm": 2.120762348175049, + "learning_rate": 1.797742946708464e-05, + "loss": 2.54422550201416, + "step": 37600 + }, + { + "epoch": 0.10240914659044507, + "grad_norm": 2.1092989444732666, + "learning_rate": 1.7976339103175687e-05, + "loss": 2.5279462814331053, + "step": 37620 + }, + { + "epoch": 0.1024635905811896, + "grad_norm": 2.1945934295654297, + "learning_rate": 1.797524873926673e-05, + "loss": 2.5361827850341796, + "step": 37640 + }, + { + "epoch": 0.10251803457193412, + "grad_norm": 2.154357671737671, + "learning_rate": 1.797415837535778e-05, + "loss": 2.526512336730957, + "step": 37660 + }, + { + "epoch": 0.10257247856267865, + "grad_norm": 2.0624818801879883, + "learning_rate": 1.797306801144882e-05, + "loss": 2.5533061981201173, + "step": 37680 + }, + { + "epoch": 0.10262692255342316, + "grad_norm": 2.2194719314575195, + "learning_rate": 1.797197764753987e-05, + "loss": 2.529607963562012, + "step": 37700 + }, + { + "epoch": 0.1026813665441677, + "grad_norm": 2.2000677585601807, + "learning_rate": 1.7970887283630912e-05, + "loss": 2.572311592102051, + "step": 37720 + }, + { + "epoch": 0.10273581053491221, + "grad_norm": 2.1650497913360596, + "learning_rate": 1.7969796919721958e-05, + "loss": 2.5708829879760744, + "step": 37740 + }, + { + "epoch": 0.10279025452565672, + "grad_norm": 2.233246088027954, + "learning_rate": 1.7968706555813004e-05, + "loss": 2.536641311645508, + "step": 37760 + }, + { + "epoch": 0.10284469851640125, + "grad_norm": 2.0520565509796143, + "learning_rate": 1.796761619190405e-05, + "loss": 2.588156318664551, + "step": 37780 + }, + { + "epoch": 0.10289914250714577, + "grad_norm": 2.2504043579101562, + "learning_rate": 1.7966525827995092e-05, + "loss": 2.5379098892211913, + "step": 37800 + }, + { + "epoch": 0.1029535864978903, + "grad_norm": 2.229914426803589, + "learning_rate": 1.796543546408614e-05, + "loss": 2.5769203186035154, + "step": 37820 + }, + { + "epoch": 0.10300803048863481, + "grad_norm": 2.0589306354522705, + "learning_rate": 1.7964345100177184e-05, + "loss": 2.5531074523925783, + "step": 37840 + }, + { + "epoch": 0.10306247447937934, + "grad_norm": 2.1145806312561035, + "learning_rate": 1.7963254736268233e-05, + "loss": 2.5540538787841798, + "step": 37860 + }, + { + "epoch": 0.10311691847012386, + "grad_norm": 2.06138277053833, + "learning_rate": 1.7962164372359275e-05, + "loss": 2.548496437072754, + "step": 37880 + }, + { + "epoch": 0.10317136246086839, + "grad_norm": 2.137213945388794, + "learning_rate": 1.796107400845032e-05, + "loss": 2.555440330505371, + "step": 37900 + }, + { + "epoch": 0.1032258064516129, + "grad_norm": 2.0707740783691406, + "learning_rate": 1.7959983644541367e-05, + "loss": 2.5520986557006835, + "step": 37920 + }, + { + "epoch": 0.10328025044235742, + "grad_norm": 2.109719753265381, + "learning_rate": 1.7958893280632413e-05, + "loss": 2.544598388671875, + "step": 37940 + }, + { + "epoch": 0.10333469443310195, + "grad_norm": 2.1682751178741455, + "learning_rate": 1.7957802916723455e-05, + "loss": 2.5536113739013673, + "step": 37960 + }, + { + "epoch": 0.10338913842384646, + "grad_norm": 2.0580453872680664, + "learning_rate": 1.7956712552814505e-05, + "loss": 2.5435062408447267, + "step": 37980 + }, + { + "epoch": 0.10344358241459099, + "grad_norm": 2.15277361869812, + "learning_rate": 1.7955622188905547e-05, + "loss": 2.513111114501953, + "step": 38000 + }, + { + "epoch": 0.10349802640533551, + "grad_norm": 2.120849132537842, + "learning_rate": 1.7954531824996596e-05, + "loss": 2.5464588165283204, + "step": 38020 + }, + { + "epoch": 0.10355247039608004, + "grad_norm": 2.1527488231658936, + "learning_rate": 1.795344146108764e-05, + "loss": 2.572945404052734, + "step": 38040 + }, + { + "epoch": 0.10360691438682455, + "grad_norm": 2.2880802154541016, + "learning_rate": 1.7952351097178684e-05, + "loss": 2.566145324707031, + "step": 38060 + }, + { + "epoch": 0.10366135837756908, + "grad_norm": 2.1622376441955566, + "learning_rate": 1.795126073326973e-05, + "loss": 2.5851112365722657, + "step": 38080 + }, + { + "epoch": 0.1037158023683136, + "grad_norm": 2.077587127685547, + "learning_rate": 1.7950170369360776e-05, + "loss": 2.5373409271240233, + "step": 38100 + }, + { + "epoch": 0.10377024635905811, + "grad_norm": 2.1375699043273926, + "learning_rate": 1.794908000545182e-05, + "loss": 2.555043411254883, + "step": 38120 + }, + { + "epoch": 0.10382469034980264, + "grad_norm": 2.2357261180877686, + "learning_rate": 1.7947989641542868e-05, + "loss": 2.553906059265137, + "step": 38140 + }, + { + "epoch": 0.10387913434054716, + "grad_norm": 2.675114393234253, + "learning_rate": 1.794689927763391e-05, + "loss": 2.559856986999512, + "step": 38160 + }, + { + "epoch": 0.10393357833129169, + "grad_norm": 2.3326573371887207, + "learning_rate": 1.794580891372496e-05, + "loss": 2.5469791412353517, + "step": 38180 + }, + { + "epoch": 0.1039880223220362, + "grad_norm": 2.0504190921783447, + "learning_rate": 1.7944718549816e-05, + "loss": 2.5687797546386717, + "step": 38200 + }, + { + "epoch": 0.10404246631278073, + "grad_norm": 2.1705515384674072, + "learning_rate": 1.7943628185907047e-05, + "loss": 2.5288896560668945, + "step": 38220 + }, + { + "epoch": 0.10409691030352525, + "grad_norm": 2.513148546218872, + "learning_rate": 1.7942537821998093e-05, + "loss": 2.5494625091552736, + "step": 38240 + }, + { + "epoch": 0.10415135429426978, + "grad_norm": 2.2672133445739746, + "learning_rate": 1.794144745808914e-05, + "loss": 2.583879089355469, + "step": 38260 + }, + { + "epoch": 0.10420579828501429, + "grad_norm": 2.163569927215576, + "learning_rate": 1.794035709418018e-05, + "loss": 2.5571401596069334, + "step": 38280 + }, + { + "epoch": 0.10426024227575881, + "grad_norm": 2.1288604736328125, + "learning_rate": 1.793926673027123e-05, + "loss": 2.5582584381103515, + "step": 38300 + }, + { + "epoch": 0.10431468626650334, + "grad_norm": 2.209597110748291, + "learning_rate": 1.7938176366362273e-05, + "loss": 2.553334617614746, + "step": 38320 + }, + { + "epoch": 0.10436913025724785, + "grad_norm": 2.2610714435577393, + "learning_rate": 1.7937086002453322e-05, + "loss": 2.538839912414551, + "step": 38340 + }, + { + "epoch": 0.10442357424799238, + "grad_norm": 2.237882137298584, + "learning_rate": 1.7935995638544365e-05, + "loss": 2.527474212646484, + "step": 38360 + }, + { + "epoch": 0.1044780182387369, + "grad_norm": 2.2397594451904297, + "learning_rate": 1.793490527463541e-05, + "loss": 2.524505043029785, + "step": 38380 + }, + { + "epoch": 0.10453246222948143, + "grad_norm": 2.109808921813965, + "learning_rate": 1.7933814910726456e-05, + "loss": 2.581486129760742, + "step": 38400 + }, + { + "epoch": 0.10458690622022594, + "grad_norm": 2.092906951904297, + "learning_rate": 1.7932724546817502e-05, + "loss": 2.538001823425293, + "step": 38420 + }, + { + "epoch": 0.10464135021097046, + "grad_norm": 2.254854917526245, + "learning_rate": 1.7931634182908548e-05, + "loss": 2.5606075286865235, + "step": 38440 + }, + { + "epoch": 0.10469579420171499, + "grad_norm": 2.1621782779693604, + "learning_rate": 1.7930543818999594e-05, + "loss": 2.5498775482177733, + "step": 38460 + }, + { + "epoch": 0.1047502381924595, + "grad_norm": 2.203192949295044, + "learning_rate": 1.7929453455090636e-05, + "loss": 2.5588525772094726, + "step": 38480 + }, + { + "epoch": 0.10480468218320403, + "grad_norm": 2.189850091934204, + "learning_rate": 1.7928363091181685e-05, + "loss": 2.5504722595214844, + "step": 38500 + }, + { + "epoch": 0.10485912617394855, + "grad_norm": 2.278193712234497, + "learning_rate": 1.7927272727272728e-05, + "loss": 2.5744773864746096, + "step": 38520 + }, + { + "epoch": 0.10491357016469308, + "grad_norm": 2.2094216346740723, + "learning_rate": 1.7926182363363774e-05, + "loss": 2.560259246826172, + "step": 38540 + }, + { + "epoch": 0.10496801415543759, + "grad_norm": 2.127765655517578, + "learning_rate": 1.792509199945482e-05, + "loss": 2.544881820678711, + "step": 38560 + }, + { + "epoch": 0.10502245814618212, + "grad_norm": 2.0801780223846436, + "learning_rate": 1.7924001635545865e-05, + "loss": 2.5169193267822267, + "step": 38580 + }, + { + "epoch": 0.10507690213692664, + "grad_norm": 2.2431507110595703, + "learning_rate": 1.792291127163691e-05, + "loss": 2.5425746917724608, + "step": 38600 + }, + { + "epoch": 0.10513134612767115, + "grad_norm": 2.540701389312744, + "learning_rate": 1.7921820907727957e-05, + "loss": 2.5400943756103516, + "step": 38620 + }, + { + "epoch": 0.10518579011841568, + "grad_norm": 2.1439900398254395, + "learning_rate": 1.7920730543819e-05, + "loss": 2.563876152038574, + "step": 38640 + }, + { + "epoch": 0.1052402341091602, + "grad_norm": 2.3114564418792725, + "learning_rate": 1.791964017991005e-05, + "loss": 2.5413646697998047, + "step": 38660 + }, + { + "epoch": 0.10529467809990473, + "grad_norm": 3.0981271266937256, + "learning_rate": 1.791854981600109e-05, + "loss": 2.548657035827637, + "step": 38680 + }, + { + "epoch": 0.10534912209064924, + "grad_norm": 2.0474472045898438, + "learning_rate": 1.7917459452092137e-05, + "loss": 2.525910568237305, + "step": 38700 + }, + { + "epoch": 0.10540356608139377, + "grad_norm": 2.11630916595459, + "learning_rate": 1.7916369088183183e-05, + "loss": 2.5638824462890626, + "step": 38720 + }, + { + "epoch": 0.10545801007213829, + "grad_norm": 2.1111979484558105, + "learning_rate": 1.791527872427423e-05, + "loss": 2.5164756774902344, + "step": 38740 + }, + { + "epoch": 0.10551245406288282, + "grad_norm": 2.056627035140991, + "learning_rate": 1.7914188360365274e-05, + "loss": 2.538258171081543, + "step": 38760 + }, + { + "epoch": 0.10556689805362733, + "grad_norm": 2.053391695022583, + "learning_rate": 1.791309799645632e-05, + "loss": 2.5266511917114256, + "step": 38780 + }, + { + "epoch": 0.10562134204437185, + "grad_norm": 2.165405511856079, + "learning_rate": 1.7912007632547362e-05, + "loss": 2.5662357330322267, + "step": 38800 + }, + { + "epoch": 0.10567578603511638, + "grad_norm": 2.1298935413360596, + "learning_rate": 1.791091726863841e-05, + "loss": 2.5414552688598633, + "step": 38820 + }, + { + "epoch": 0.10573023002586089, + "grad_norm": 2.142667055130005, + "learning_rate": 1.7909826904729454e-05, + "loss": 2.5636192321777345, + "step": 38840 + }, + { + "epoch": 0.10578467401660542, + "grad_norm": 2.2109203338623047, + "learning_rate": 1.79087365408205e-05, + "loss": 2.570294952392578, + "step": 38860 + }, + { + "epoch": 0.10583911800734994, + "grad_norm": 4.003000259399414, + "learning_rate": 1.7907646176911546e-05, + "loss": 2.5212621688842773, + "step": 38880 + }, + { + "epoch": 0.10589356199809447, + "grad_norm": 2.253497362136841, + "learning_rate": 1.790655581300259e-05, + "loss": 2.561587905883789, + "step": 38900 + }, + { + "epoch": 0.10594800598883898, + "grad_norm": 8.655647277832031, + "learning_rate": 1.7905465449093637e-05, + "loss": 2.5703807830810548, + "step": 38920 + }, + { + "epoch": 0.10600244997958351, + "grad_norm": 2.225534677505493, + "learning_rate": 1.7904375085184683e-05, + "loss": 2.568634033203125, + "step": 38940 + }, + { + "epoch": 0.10605689397032803, + "grad_norm": 2.0838091373443604, + "learning_rate": 1.7903284721275725e-05, + "loss": 2.5691783905029295, + "step": 38960 + }, + { + "epoch": 0.10611133796107254, + "grad_norm": 2.1985278129577637, + "learning_rate": 1.7902194357366775e-05, + "loss": 2.5426597595214844, + "step": 38980 + }, + { + "epoch": 0.10616578195181707, + "grad_norm": 2.0900118350982666, + "learning_rate": 1.7901103993457817e-05, + "loss": 2.571373176574707, + "step": 39000 + }, + { + "epoch": 0.10622022594256159, + "grad_norm": 2.0594162940979004, + "learning_rate": 1.7900013629548863e-05, + "loss": 2.555073928833008, + "step": 39020 + }, + { + "epoch": 0.10627466993330612, + "grad_norm": 2.115111827850342, + "learning_rate": 1.789892326563991e-05, + "loss": 2.5130855560302736, + "step": 39040 + }, + { + "epoch": 0.10632911392405063, + "grad_norm": 2.164984941482544, + "learning_rate": 1.7897832901730954e-05, + "loss": 2.5425899505615233, + "step": 39060 + }, + { + "epoch": 0.10638355791479516, + "grad_norm": 2.1035165786743164, + "learning_rate": 1.7896742537822e-05, + "loss": 2.546371078491211, + "step": 39080 + }, + { + "epoch": 0.10643800190553968, + "grad_norm": 2.2151894569396973, + "learning_rate": 1.7895652173913046e-05, + "loss": 2.568552780151367, + "step": 39100 + }, + { + "epoch": 0.10649244589628419, + "grad_norm": 2.2284233570098877, + "learning_rate": 1.789456181000409e-05, + "loss": 2.5236232757568358, + "step": 39120 + }, + { + "epoch": 0.10654688988702872, + "grad_norm": 2.1171085834503174, + "learning_rate": 1.7893471446095138e-05, + "loss": 2.5348674774169924, + "step": 39140 + }, + { + "epoch": 0.10660133387777324, + "grad_norm": 2.3126299381256104, + "learning_rate": 1.789238108218618e-05, + "loss": 2.517869758605957, + "step": 39160 + }, + { + "epoch": 0.10665577786851776, + "grad_norm": 2.446875810623169, + "learning_rate": 1.7891290718277226e-05, + "loss": 2.5654071807861327, + "step": 39180 + }, + { + "epoch": 0.10671022185926228, + "grad_norm": 2.2085790634155273, + "learning_rate": 1.7890200354368272e-05, + "loss": 2.5739532470703126, + "step": 39200 + }, + { + "epoch": 0.10676466585000681, + "grad_norm": 2.287297487258911, + "learning_rate": 1.7889109990459318e-05, + "loss": 2.5443098068237306, + "step": 39220 + }, + { + "epoch": 0.10681910984075133, + "grad_norm": 2.1552011966705322, + "learning_rate": 1.7888019626550363e-05, + "loss": 2.5520355224609377, + "step": 39240 + }, + { + "epoch": 0.10687355383149585, + "grad_norm": 2.1286768913269043, + "learning_rate": 1.788692926264141e-05, + "loss": 2.5486495971679686, + "step": 39260 + }, + { + "epoch": 0.10692799782224037, + "grad_norm": 2.1008222103118896, + "learning_rate": 1.788583889873245e-05, + "loss": 2.5546709060668946, + "step": 39280 + }, + { + "epoch": 0.10698244181298489, + "grad_norm": 2.1637649536132812, + "learning_rate": 1.78847485348235e-05, + "loss": 2.5440834045410154, + "step": 39300 + }, + { + "epoch": 0.10703688580372941, + "grad_norm": 2.086416244506836, + "learning_rate": 1.7883658170914543e-05, + "loss": 2.5619192123413086, + "step": 39320 + }, + { + "epoch": 0.10709132979447393, + "grad_norm": 2.1073620319366455, + "learning_rate": 1.788256780700559e-05, + "loss": 2.575174331665039, + "step": 39340 + }, + { + "epoch": 0.10714577378521846, + "grad_norm": 2.0976624488830566, + "learning_rate": 1.7881477443096635e-05, + "loss": 2.555156707763672, + "step": 39360 + }, + { + "epoch": 0.10720021777596297, + "grad_norm": 2.262937545776367, + "learning_rate": 1.788038707918768e-05, + "loss": 2.5101449966430662, + "step": 39380 + }, + { + "epoch": 0.1072546617667075, + "grad_norm": 2.1324124336242676, + "learning_rate": 1.7879296715278726e-05, + "loss": 2.5463594436645507, + "step": 39400 + }, + { + "epoch": 0.10730910575745202, + "grad_norm": 2.167602062225342, + "learning_rate": 1.7878206351369772e-05, + "loss": 2.5254899978637697, + "step": 39420 + }, + { + "epoch": 0.10736354974819655, + "grad_norm": 2.0813324451446533, + "learning_rate": 1.7877115987460815e-05, + "loss": 2.5613525390625, + "step": 39440 + }, + { + "epoch": 0.10741799373894106, + "grad_norm": 2.328359842300415, + "learning_rate": 1.7876025623551864e-05, + "loss": 2.5532415390014647, + "step": 39460 + }, + { + "epoch": 0.10747243772968558, + "grad_norm": 2.2112812995910645, + "learning_rate": 1.7874935259642906e-05, + "loss": 2.5556896209716795, + "step": 39480 + }, + { + "epoch": 0.10752688172043011, + "grad_norm": 2.3349826335906982, + "learning_rate": 1.7873844895733952e-05, + "loss": 2.542160987854004, + "step": 39500 + }, + { + "epoch": 0.10758132571117462, + "grad_norm": 2.0899786949157715, + "learning_rate": 1.7872754531824998e-05, + "loss": 2.558024024963379, + "step": 39520 + }, + { + "epoch": 0.10763576970191915, + "grad_norm": 2.4080424308776855, + "learning_rate": 1.7871664167916044e-05, + "loss": 2.5231800079345703, + "step": 39540 + }, + { + "epoch": 0.10769021369266367, + "grad_norm": 2.0958609580993652, + "learning_rate": 1.787057380400709e-05, + "loss": 2.5306806564331055, + "step": 39560 + }, + { + "epoch": 0.1077446576834082, + "grad_norm": 2.195272922515869, + "learning_rate": 1.7869483440098135e-05, + "loss": 2.55880184173584, + "step": 39580 + }, + { + "epoch": 0.10779910167415271, + "grad_norm": 2.127791404724121, + "learning_rate": 1.7868393076189178e-05, + "loss": 2.542574882507324, + "step": 39600 + }, + { + "epoch": 0.10785354566489724, + "grad_norm": 2.3423986434936523, + "learning_rate": 1.7867302712280227e-05, + "loss": 2.5718461990356447, + "step": 39620 + }, + { + "epoch": 0.10790798965564176, + "grad_norm": 2.2748048305511475, + "learning_rate": 1.786621234837127e-05, + "loss": 2.563517951965332, + "step": 39640 + }, + { + "epoch": 0.10796243364638627, + "grad_norm": 2.013371467590332, + "learning_rate": 1.7865121984462315e-05, + "loss": 2.547480010986328, + "step": 39660 + }, + { + "epoch": 0.1080168776371308, + "grad_norm": 2.175959825515747, + "learning_rate": 1.786403162055336e-05, + "loss": 2.563935470581055, + "step": 39680 + }, + { + "epoch": 0.10807132162787532, + "grad_norm": 2.2622785568237305, + "learning_rate": 1.7862941256644407e-05, + "loss": 2.536014938354492, + "step": 39700 + }, + { + "epoch": 0.10812576561861985, + "grad_norm": 2.191472291946411, + "learning_rate": 1.7861850892735453e-05, + "loss": 2.5500307083129883, + "step": 39720 + }, + { + "epoch": 0.10818020960936436, + "grad_norm": 2.3031702041625977, + "learning_rate": 1.78607605288265e-05, + "loss": 2.5620988845825194, + "step": 39740 + }, + { + "epoch": 0.1082346536001089, + "grad_norm": 2.149420738220215, + "learning_rate": 1.785967016491754e-05, + "loss": 2.5220245361328124, + "step": 39760 + }, + { + "epoch": 0.10828909759085341, + "grad_norm": 2.180468797683716, + "learning_rate": 1.785857980100859e-05, + "loss": 2.5670520782470705, + "step": 39780 + }, + { + "epoch": 0.10834354158159794, + "grad_norm": 2.0377886295318604, + "learning_rate": 1.7857489437099632e-05, + "loss": 2.4853668212890625, + "step": 39800 + }, + { + "epoch": 0.10839798557234245, + "grad_norm": 2.206907272338867, + "learning_rate": 1.7856399073190678e-05, + "loss": 2.544178581237793, + "step": 39820 + }, + { + "epoch": 0.10845242956308697, + "grad_norm": 1.9910589456558228, + "learning_rate": 1.7855308709281724e-05, + "loss": 2.5191923141479493, + "step": 39840 + }, + { + "epoch": 0.1085068735538315, + "grad_norm": 2.2230305671691895, + "learning_rate": 1.785421834537277e-05, + "loss": 2.5232875823974608, + "step": 39860 + }, + { + "epoch": 0.10856131754457601, + "grad_norm": 2.139108657836914, + "learning_rate": 1.7853127981463816e-05, + "loss": 2.5330604553222655, + "step": 39880 + }, + { + "epoch": 0.10861576153532054, + "grad_norm": 2.0533013343811035, + "learning_rate": 1.785203761755486e-05, + "loss": 2.5456466674804688, + "step": 39900 + }, + { + "epoch": 0.10867020552606506, + "grad_norm": 2.13923978805542, + "learning_rate": 1.7850947253645904e-05, + "loss": 2.5519737243652343, + "step": 39920 + }, + { + "epoch": 0.10872464951680959, + "grad_norm": 2.1114542484283447, + "learning_rate": 1.7849856889736953e-05, + "loss": 2.5448785781860352, + "step": 39940 + }, + { + "epoch": 0.1087790935075541, + "grad_norm": 2.1450517177581787, + "learning_rate": 1.7848766525827996e-05, + "loss": 2.551150894165039, + "step": 39960 + }, + { + "epoch": 0.10883353749829862, + "grad_norm": 2.1456432342529297, + "learning_rate": 1.784767616191904e-05, + "loss": 2.551975631713867, + "step": 39980 + }, + { + "epoch": 0.10888798148904315, + "grad_norm": 2.2001821994781494, + "learning_rate": 1.7846585798010087e-05, + "loss": 2.536049652099609, + "step": 40000 + }, + { + "epoch": 0.10894242547978766, + "grad_norm": 2.1204042434692383, + "learning_rate": 1.7845495434101133e-05, + "loss": 2.549427795410156, + "step": 40020 + }, + { + "epoch": 0.10899686947053219, + "grad_norm": 2.279510736465454, + "learning_rate": 1.784440507019218e-05, + "loss": 2.549898147583008, + "step": 40040 + }, + { + "epoch": 0.10905131346127671, + "grad_norm": 2.18314266204834, + "learning_rate": 1.7843314706283225e-05, + "loss": 2.5343847274780273, + "step": 40060 + }, + { + "epoch": 0.10910575745202124, + "grad_norm": 2.147700786590576, + "learning_rate": 1.7842224342374267e-05, + "loss": 2.531480407714844, + "step": 40080 + }, + { + "epoch": 0.10916020144276575, + "grad_norm": 2.0696773529052734, + "learning_rate": 1.7841133978465316e-05, + "loss": 2.5465511322021483, + "step": 40100 + }, + { + "epoch": 0.10921464543351028, + "grad_norm": 2.1015665531158447, + "learning_rate": 1.784004361455636e-05, + "loss": 2.5570072174072265, + "step": 40120 + }, + { + "epoch": 0.1092690894242548, + "grad_norm": 2.086550712585449, + "learning_rate": 1.7838953250647404e-05, + "loss": 2.5437665939331056, + "step": 40140 + }, + { + "epoch": 0.10932353341499931, + "grad_norm": 2.3157176971435547, + "learning_rate": 1.783786288673845e-05, + "loss": 2.552168083190918, + "step": 40160 + }, + { + "epoch": 0.10937797740574384, + "grad_norm": 2.1957032680511475, + "learning_rate": 1.7836772522829496e-05, + "loss": 2.574681282043457, + "step": 40180 + }, + { + "epoch": 0.10943242139648836, + "grad_norm": 2.107451915740967, + "learning_rate": 1.7835682158920542e-05, + "loss": 2.5448047637939455, + "step": 40200 + }, + { + "epoch": 0.10948686538723289, + "grad_norm": 2.10111403465271, + "learning_rate": 1.7834591795011588e-05, + "loss": 2.5717031478881838, + "step": 40220 + }, + { + "epoch": 0.1095413093779774, + "grad_norm": 2.162473678588867, + "learning_rate": 1.783350143110263e-05, + "loss": 2.5523494720458983, + "step": 40240 + }, + { + "epoch": 0.10959575336872193, + "grad_norm": 2.2701711654663086, + "learning_rate": 1.783241106719368e-05, + "loss": 2.542867660522461, + "step": 40260 + }, + { + "epoch": 0.10965019735946645, + "grad_norm": 2.2205281257629395, + "learning_rate": 1.7831320703284722e-05, + "loss": 2.527700996398926, + "step": 40280 + }, + { + "epoch": 0.10970464135021098, + "grad_norm": 2.1897764205932617, + "learning_rate": 1.7830230339375768e-05, + "loss": 2.5517480850219725, + "step": 40300 + }, + { + "epoch": 0.10975908534095549, + "grad_norm": 2.134894609451294, + "learning_rate": 1.7829139975466813e-05, + "loss": 2.524684715270996, + "step": 40320 + }, + { + "epoch": 0.10981352933170001, + "grad_norm": 2.0848586559295654, + "learning_rate": 1.782804961155786e-05, + "loss": 2.564030647277832, + "step": 40340 + }, + { + "epoch": 0.10986797332244454, + "grad_norm": 2.1668639183044434, + "learning_rate": 1.7826959247648905e-05, + "loss": 2.5827049255371093, + "step": 40360 + }, + { + "epoch": 0.10992241731318905, + "grad_norm": 2.1595096588134766, + "learning_rate": 1.782586888373995e-05, + "loss": 2.554102325439453, + "step": 40380 + }, + { + "epoch": 0.10997686130393358, + "grad_norm": 2.0007340908050537, + "learning_rate": 1.7824778519830993e-05, + "loss": 2.531597137451172, + "step": 40400 + }, + { + "epoch": 0.1100313052946781, + "grad_norm": 2.1288881301879883, + "learning_rate": 1.7823688155922042e-05, + "loss": 2.555001640319824, + "step": 40420 + }, + { + "epoch": 0.11008574928542263, + "grad_norm": 2.195211887359619, + "learning_rate": 1.7822597792013085e-05, + "loss": 2.5385223388671876, + "step": 40440 + }, + { + "epoch": 0.11014019327616714, + "grad_norm": 2.367387056350708, + "learning_rate": 1.782150742810413e-05, + "loss": 2.5673616409301756, + "step": 40460 + }, + { + "epoch": 0.11019463726691167, + "grad_norm": 2.2168755531311035, + "learning_rate": 1.7820417064195176e-05, + "loss": 2.548270034790039, + "step": 40480 + }, + { + "epoch": 0.11024908125765619, + "grad_norm": 2.1487808227539062, + "learning_rate": 1.7819326700286222e-05, + "loss": 2.531063461303711, + "step": 40500 + }, + { + "epoch": 0.1103035252484007, + "grad_norm": 2.2331502437591553, + "learning_rate": 1.7818236336377268e-05, + "loss": 2.5526609420776367, + "step": 40520 + }, + { + "epoch": 0.11035796923914523, + "grad_norm": 2.1282238960266113, + "learning_rate": 1.7817145972468314e-05, + "loss": 2.5413557052612306, + "step": 40540 + }, + { + "epoch": 0.11041241322988975, + "grad_norm": 2.047506332397461, + "learning_rate": 1.7816055608559356e-05, + "loss": 2.5589311599731444, + "step": 40560 + }, + { + "epoch": 0.11046685722063428, + "grad_norm": 2.136105537414551, + "learning_rate": 1.7814965244650406e-05, + "loss": 2.541647529602051, + "step": 40580 + }, + { + "epoch": 0.11052130121137879, + "grad_norm": 2.162637710571289, + "learning_rate": 1.7813874880741448e-05, + "loss": 2.573734474182129, + "step": 40600 + }, + { + "epoch": 0.11057574520212332, + "grad_norm": 2.270782709121704, + "learning_rate": 1.7812784516832494e-05, + "loss": 2.5155786514282226, + "step": 40620 + }, + { + "epoch": 0.11063018919286784, + "grad_norm": 2.3366212844848633, + "learning_rate": 1.781169415292354e-05, + "loss": 2.5392387390136717, + "step": 40640 + }, + { + "epoch": 0.11068463318361235, + "grad_norm": 2.017615556716919, + "learning_rate": 1.7810603789014585e-05, + "loss": 2.5416255950927735, + "step": 40660 + }, + { + "epoch": 0.11073907717435688, + "grad_norm": 2.1047632694244385, + "learning_rate": 1.780951342510563e-05, + "loss": 2.537862777709961, + "step": 40680 + }, + { + "epoch": 0.1107935211651014, + "grad_norm": 2.0962300300598145, + "learning_rate": 1.7808423061196677e-05, + "loss": 2.532161331176758, + "step": 40700 + }, + { + "epoch": 0.11084796515584593, + "grad_norm": 2.0985634326934814, + "learning_rate": 1.780733269728772e-05, + "loss": 2.5513212203979494, + "step": 40720 + }, + { + "epoch": 0.11090240914659044, + "grad_norm": 2.138345956802368, + "learning_rate": 1.780624233337877e-05, + "loss": 2.555580711364746, + "step": 40740 + }, + { + "epoch": 0.11095685313733497, + "grad_norm": 2.2042927742004395, + "learning_rate": 1.780515196946981e-05, + "loss": 2.52526741027832, + "step": 40760 + }, + { + "epoch": 0.11101129712807949, + "grad_norm": 2.268299102783203, + "learning_rate": 1.7804061605560857e-05, + "loss": 2.5748098373413084, + "step": 40780 + }, + { + "epoch": 0.11106574111882402, + "grad_norm": 2.2938168048858643, + "learning_rate": 1.7802971241651903e-05, + "loss": 2.5292551040649416, + "step": 40800 + }, + { + "epoch": 0.11112018510956853, + "grad_norm": 2.1671905517578125, + "learning_rate": 1.780188087774295e-05, + "loss": 2.561750793457031, + "step": 40820 + }, + { + "epoch": 0.11117462910031305, + "grad_norm": 2.148171901702881, + "learning_rate": 1.7800790513833994e-05, + "loss": 2.549598503112793, + "step": 40840 + }, + { + "epoch": 0.11122907309105758, + "grad_norm": 2.031028985977173, + "learning_rate": 1.779970014992504e-05, + "loss": 2.5358917236328127, + "step": 40860 + }, + { + "epoch": 0.11128351708180209, + "grad_norm": 2.0619113445281982, + "learning_rate": 1.7798609786016082e-05, + "loss": 2.552206611633301, + "step": 40880 + }, + { + "epoch": 0.11133796107254662, + "grad_norm": 2.2235214710235596, + "learning_rate": 1.779751942210713e-05, + "loss": 2.5573963165283202, + "step": 40900 + }, + { + "epoch": 0.11139240506329114, + "grad_norm": 2.3740718364715576, + "learning_rate": 1.7796429058198174e-05, + "loss": 2.5578210830688475, + "step": 40920 + }, + { + "epoch": 0.11144684905403567, + "grad_norm": 2.352169990539551, + "learning_rate": 1.779533869428922e-05, + "loss": 2.5507850646972656, + "step": 40940 + }, + { + "epoch": 0.11150129304478018, + "grad_norm": 2.4279823303222656, + "learning_rate": 1.7794248330380266e-05, + "loss": 2.5861358642578125, + "step": 40960 + }, + { + "epoch": 0.11155573703552471, + "grad_norm": 2.212115526199341, + "learning_rate": 1.779315796647131e-05, + "loss": 2.5332172393798826, + "step": 40980 + }, + { + "epoch": 0.11161018102626923, + "grad_norm": 2.079202890396118, + "learning_rate": 1.7792067602562357e-05, + "loss": 2.5467708587646483, + "step": 41000 + }, + { + "epoch": 0.11166462501701374, + "grad_norm": 2.143815279006958, + "learning_rate": 1.7790977238653403e-05, + "loss": 2.558207321166992, + "step": 41020 + }, + { + "epoch": 0.11171906900775827, + "grad_norm": 2.126553535461426, + "learning_rate": 1.7789886874744446e-05, + "loss": 2.569243621826172, + "step": 41040 + }, + { + "epoch": 0.11177351299850279, + "grad_norm": 25.000516891479492, + "learning_rate": 1.7788796510835495e-05, + "loss": 2.515280532836914, + "step": 41060 + }, + { + "epoch": 0.11182795698924732, + "grad_norm": 2.1331121921539307, + "learning_rate": 1.7787706146926537e-05, + "loss": 2.5403522491455077, + "step": 41080 + }, + { + "epoch": 0.11188240097999183, + "grad_norm": 2.4247148036956787, + "learning_rate": 1.7786615783017583e-05, + "loss": 2.5482301712036133, + "step": 41100 + }, + { + "epoch": 0.11193684497073636, + "grad_norm": 2.1559369564056396, + "learning_rate": 1.778552541910863e-05, + "loss": 2.5333189010620116, + "step": 41120 + }, + { + "epoch": 0.11199128896148088, + "grad_norm": 2.108821153640747, + "learning_rate": 1.7784435055199675e-05, + "loss": 2.5306943893432616, + "step": 41140 + }, + { + "epoch": 0.1120457329522254, + "grad_norm": 2.1797642707824707, + "learning_rate": 1.778334469129072e-05, + "loss": 2.5761110305786135, + "step": 41160 + }, + { + "epoch": 0.11210017694296992, + "grad_norm": 2.311356544494629, + "learning_rate": 1.7782254327381766e-05, + "loss": 2.562594413757324, + "step": 41180 + }, + { + "epoch": 0.11215462093371444, + "grad_norm": 2.1050493717193604, + "learning_rate": 1.778116396347281e-05, + "loss": 2.5520076751708984, + "step": 41200 + }, + { + "epoch": 0.11220906492445896, + "grad_norm": 2.2240331172943115, + "learning_rate": 1.7780073599563858e-05, + "loss": 2.558402442932129, + "step": 41220 + }, + { + "epoch": 0.11226350891520348, + "grad_norm": 2.97802472114563, + "learning_rate": 1.77789832356549e-05, + "loss": 2.538087844848633, + "step": 41240 + }, + { + "epoch": 0.11231795290594801, + "grad_norm": 2.4533193111419678, + "learning_rate": 1.7777892871745946e-05, + "loss": 2.5209766387939454, + "step": 41260 + }, + { + "epoch": 0.11237239689669253, + "grad_norm": 3.454923629760742, + "learning_rate": 1.7776802507836992e-05, + "loss": 2.548310470581055, + "step": 41280 + }, + { + "epoch": 0.11242684088743705, + "grad_norm": 2.3136658668518066, + "learning_rate": 1.7775712143928038e-05, + "loss": 2.573470687866211, + "step": 41300 + }, + { + "epoch": 0.11248128487818157, + "grad_norm": 2.261920213699341, + "learning_rate": 1.7774621780019084e-05, + "loss": 2.5709205627441407, + "step": 41320 + }, + { + "epoch": 0.11253572886892609, + "grad_norm": 2.1659786701202393, + "learning_rate": 1.777353141611013e-05, + "loss": 2.5670793533325194, + "step": 41340 + }, + { + "epoch": 0.11259017285967061, + "grad_norm": 4.298665523529053, + "learning_rate": 1.7772441052201172e-05, + "loss": 2.5658634185791014, + "step": 41360 + }, + { + "epoch": 0.11264461685041513, + "grad_norm": 2.175870656967163, + "learning_rate": 1.777135068829222e-05, + "loss": 2.5481441497802733, + "step": 41380 + }, + { + "epoch": 0.11269906084115966, + "grad_norm": 2.30899715423584, + "learning_rate": 1.7770260324383263e-05, + "loss": 2.5414192199707033, + "step": 41400 + }, + { + "epoch": 0.11275350483190417, + "grad_norm": 2.011826992034912, + "learning_rate": 1.776916996047431e-05, + "loss": 2.5441083908081055, + "step": 41420 + }, + { + "epoch": 0.1128079488226487, + "grad_norm": 2.274806499481201, + "learning_rate": 1.7768079596565355e-05, + "loss": 2.5519161224365234, + "step": 41440 + }, + { + "epoch": 0.11286239281339322, + "grad_norm": 2.0734221935272217, + "learning_rate": 1.77669892326564e-05, + "loss": 2.5684844970703127, + "step": 41460 + }, + { + "epoch": 0.11291683680413775, + "grad_norm": 2.3076884746551514, + "learning_rate": 1.7765898868747447e-05, + "loss": 2.5606548309326174, + "step": 41480 + }, + { + "epoch": 0.11297128079488226, + "grad_norm": 2.627164363861084, + "learning_rate": 1.7764808504838492e-05, + "loss": 2.577121925354004, + "step": 41500 + }, + { + "epoch": 0.11302572478562678, + "grad_norm": 2.0835118293762207, + "learning_rate": 1.7763718140929535e-05, + "loss": 2.5851768493652343, + "step": 41520 + }, + { + "epoch": 0.11308016877637131, + "grad_norm": 2.045599937438965, + "learning_rate": 1.7762627777020584e-05, + "loss": 2.540943908691406, + "step": 41540 + }, + { + "epoch": 0.11313461276711582, + "grad_norm": 2.175361156463623, + "learning_rate": 1.7761537413111626e-05, + "loss": 2.546260452270508, + "step": 41560 + }, + { + "epoch": 0.11318905675786035, + "grad_norm": 2.095844030380249, + "learning_rate": 1.7760447049202672e-05, + "loss": 2.5433990478515627, + "step": 41580 + }, + { + "epoch": 0.11324350074860487, + "grad_norm": 3.2889630794525146, + "learning_rate": 1.7759356685293718e-05, + "loss": 2.538486862182617, + "step": 41600 + }, + { + "epoch": 0.1132979447393494, + "grad_norm": 2.3407790660858154, + "learning_rate": 1.7758266321384764e-05, + "loss": 2.5353954315185545, + "step": 41620 + }, + { + "epoch": 0.11335238873009391, + "grad_norm": 2.101156711578369, + "learning_rate": 1.775717595747581e-05, + "loss": 2.5282413482666017, + "step": 41640 + }, + { + "epoch": 0.11340683272083844, + "grad_norm": 2.3127551078796387, + "learning_rate": 1.7756085593566855e-05, + "loss": 2.5510210037231444, + "step": 41660 + }, + { + "epoch": 0.11346127671158296, + "grad_norm": 2.146906852722168, + "learning_rate": 1.7754995229657898e-05, + "loss": 2.5414461135864257, + "step": 41680 + }, + { + "epoch": 0.11351572070232747, + "grad_norm": 2.284072160720825, + "learning_rate": 1.7753904865748947e-05, + "loss": 2.5494009017944337, + "step": 41700 + }, + { + "epoch": 0.113570164693072, + "grad_norm": 2.0927672386169434, + "learning_rate": 1.775281450183999e-05, + "loss": 2.5493343353271483, + "step": 41720 + }, + { + "epoch": 0.11362460868381652, + "grad_norm": 2.2968368530273438, + "learning_rate": 1.7751724137931035e-05, + "loss": 2.5174394607543946, + "step": 41740 + }, + { + "epoch": 0.11367905267456105, + "grad_norm": 2.2705821990966797, + "learning_rate": 1.775063377402208e-05, + "loss": 2.5572736740112303, + "step": 41760 + }, + { + "epoch": 0.11373349666530556, + "grad_norm": 2.3503551483154297, + "learning_rate": 1.7749543410113127e-05, + "loss": 2.5395280838012697, + "step": 41780 + }, + { + "epoch": 0.1137879406560501, + "grad_norm": 2.2100601196289062, + "learning_rate": 1.7748453046204173e-05, + "loss": 2.535105514526367, + "step": 41800 + }, + { + "epoch": 0.11384238464679461, + "grad_norm": 2.4397034645080566, + "learning_rate": 1.774736268229522e-05, + "loss": 2.554531478881836, + "step": 41820 + }, + { + "epoch": 0.11389682863753914, + "grad_norm": 2.1113533973693848, + "learning_rate": 1.7746272318386264e-05, + "loss": 2.5338768005371093, + "step": 41840 + }, + { + "epoch": 0.11395127262828365, + "grad_norm": 2.121143102645874, + "learning_rate": 1.774518195447731e-05, + "loss": 2.5443323135375975, + "step": 41860 + }, + { + "epoch": 0.11400571661902817, + "grad_norm": 2.469878673553467, + "learning_rate": 1.7744091590568353e-05, + "loss": 2.550006103515625, + "step": 41880 + }, + { + "epoch": 0.1140601606097727, + "grad_norm": 2.244673490524292, + "learning_rate": 1.77430012266594e-05, + "loss": 2.513001251220703, + "step": 41900 + }, + { + "epoch": 0.11411460460051721, + "grad_norm": 2.3717703819274902, + "learning_rate": 1.7741910862750444e-05, + "loss": 2.555767059326172, + "step": 41920 + }, + { + "epoch": 0.11416904859126174, + "grad_norm": 2.1155524253845215, + "learning_rate": 1.774082049884149e-05, + "loss": 2.5643571853637694, + "step": 41940 + }, + { + "epoch": 0.11422349258200626, + "grad_norm": 2.2179908752441406, + "learning_rate": 1.7739730134932536e-05, + "loss": 2.5582250595092773, + "step": 41960 + }, + { + "epoch": 0.11427793657275079, + "grad_norm": 2.2461211681365967, + "learning_rate": 1.773863977102358e-05, + "loss": 2.5527227401733397, + "step": 41980 + }, + { + "epoch": 0.1143323805634953, + "grad_norm": 2.344777822494507, + "learning_rate": 1.7737549407114627e-05, + "loss": 2.5643911361694336, + "step": 42000 + }, + { + "epoch": 0.11438682455423983, + "grad_norm": 2.2025599479675293, + "learning_rate": 1.7736459043205673e-05, + "loss": 2.5667644500732423, + "step": 42020 + }, + { + "epoch": 0.11444126854498435, + "grad_norm": 2.157344102859497, + "learning_rate": 1.7735368679296716e-05, + "loss": 2.557094955444336, + "step": 42040 + }, + { + "epoch": 0.11449571253572886, + "grad_norm": 2.2574660778045654, + "learning_rate": 1.773427831538776e-05, + "loss": 2.5341365814208983, + "step": 42060 + }, + { + "epoch": 0.11455015652647339, + "grad_norm": 2.325214385986328, + "learning_rate": 1.7733187951478807e-05, + "loss": 2.542080116271973, + "step": 42080 + }, + { + "epoch": 0.11460460051721791, + "grad_norm": 2.0265796184539795, + "learning_rate": 1.7732097587569853e-05, + "loss": 2.561308479309082, + "step": 42100 + }, + { + "epoch": 0.11465904450796244, + "grad_norm": 2.2383482456207275, + "learning_rate": 1.77310072236609e-05, + "loss": 2.543952751159668, + "step": 42120 + }, + { + "epoch": 0.11471348849870695, + "grad_norm": 2.441373825073242, + "learning_rate": 1.7729916859751945e-05, + "loss": 2.566787338256836, + "step": 42140 + }, + { + "epoch": 0.11476793248945148, + "grad_norm": 2.450249433517456, + "learning_rate": 1.772882649584299e-05, + "loss": 2.5949209213256834, + "step": 42160 + }, + { + "epoch": 0.114822376480196, + "grad_norm": 2.3520302772521973, + "learning_rate": 1.7727736131934033e-05, + "loss": 2.514897918701172, + "step": 42180 + }, + { + "epoch": 0.11487682047094051, + "grad_norm": 2.205185890197754, + "learning_rate": 1.772664576802508e-05, + "loss": 2.5436376571655273, + "step": 42200 + }, + { + "epoch": 0.11493126446168504, + "grad_norm": 5.695757865905762, + "learning_rate": 1.7725555404116125e-05, + "loss": 2.5557886123657227, + "step": 42220 + }, + { + "epoch": 0.11498570845242956, + "grad_norm": 16.172317504882812, + "learning_rate": 1.772446504020717e-05, + "loss": 2.519026184082031, + "step": 42240 + }, + { + "epoch": 0.11504015244317409, + "grad_norm": 2.226390838623047, + "learning_rate": 1.7723374676298216e-05, + "loss": 2.5521312713623048, + "step": 42260 + }, + { + "epoch": 0.1150945964339186, + "grad_norm": 2.431436777114868, + "learning_rate": 1.7722284312389262e-05, + "loss": 2.5261348724365233, + "step": 42280 + }, + { + "epoch": 0.11514904042466313, + "grad_norm": 2.2984535694122314, + "learning_rate": 1.7721193948480308e-05, + "loss": 2.556903266906738, + "step": 42300 + }, + { + "epoch": 0.11520348441540765, + "grad_norm": 2.203906297683716, + "learning_rate": 1.7720103584571354e-05, + "loss": 2.5590253829956056, + "step": 42320 + }, + { + "epoch": 0.11525792840615218, + "grad_norm": 2.09379243850708, + "learning_rate": 1.7719013220662396e-05, + "loss": 2.5615093231201174, + "step": 42340 + }, + { + "epoch": 0.11531237239689669, + "grad_norm": 2.4954848289489746, + "learning_rate": 1.7717922856753442e-05, + "loss": 2.5170930862426757, + "step": 42360 + }, + { + "epoch": 0.11536681638764121, + "grad_norm": 2.6483495235443115, + "learning_rate": 1.7716832492844488e-05, + "loss": 2.5467861175537108, + "step": 42380 + }, + { + "epoch": 0.11542126037838574, + "grad_norm": 2.1478776931762695, + "learning_rate": 1.7715742128935533e-05, + "loss": 2.560223197937012, + "step": 42400 + }, + { + "epoch": 0.11547570436913025, + "grad_norm": 2.2038209438323975, + "learning_rate": 1.771465176502658e-05, + "loss": 2.530961799621582, + "step": 42420 + }, + { + "epoch": 0.11553014835987478, + "grad_norm": 2.0812056064605713, + "learning_rate": 1.7713561401117625e-05, + "loss": 2.555230140686035, + "step": 42440 + }, + { + "epoch": 0.1155845923506193, + "grad_norm": 2.1490139961242676, + "learning_rate": 1.771247103720867e-05, + "loss": 2.5471176147460937, + "step": 42460 + }, + { + "epoch": 0.11563903634136383, + "grad_norm": 2.0845394134521484, + "learning_rate": 1.7711380673299717e-05, + "loss": 2.5480218887329102, + "step": 42480 + }, + { + "epoch": 0.11569348033210834, + "grad_norm": 2.178846597671509, + "learning_rate": 1.771029030939076e-05, + "loss": 2.5581764221191405, + "step": 42500 + }, + { + "epoch": 0.11574792432285287, + "grad_norm": 2.065422296524048, + "learning_rate": 1.7709199945481805e-05, + "loss": 2.5091650009155275, + "step": 42520 + }, + { + "epoch": 0.11580236831359739, + "grad_norm": 2.0643434524536133, + "learning_rate": 1.770810958157285e-05, + "loss": 2.5273759841918944, + "step": 42540 + }, + { + "epoch": 0.1158568123043419, + "grad_norm": 2.155809164047241, + "learning_rate": 1.7707019217663897e-05, + "loss": 2.5424264907836913, + "step": 42560 + }, + { + "epoch": 0.11591125629508643, + "grad_norm": 2.168099880218506, + "learning_rate": 1.7705928853754942e-05, + "loss": 2.5455062866210936, + "step": 42580 + }, + { + "epoch": 0.11596570028583095, + "grad_norm": 2.1999831199645996, + "learning_rate": 1.7704838489845988e-05, + "loss": 2.5372568130493165, + "step": 42600 + }, + { + "epoch": 0.11602014427657548, + "grad_norm": 2.3932888507843018, + "learning_rate": 1.7703748125937034e-05, + "loss": 2.5314662933349608, + "step": 42620 + }, + { + "epoch": 0.11607458826731999, + "grad_norm": 2.449885129928589, + "learning_rate": 1.770265776202808e-05, + "loss": 2.5433425903320312, + "step": 42640 + }, + { + "epoch": 0.11612903225806452, + "grad_norm": 2.094231128692627, + "learning_rate": 1.7701567398119122e-05, + "loss": 2.552061080932617, + "step": 42660 + }, + { + "epoch": 0.11618347624880904, + "grad_norm": 2.143378496170044, + "learning_rate": 1.7700477034210168e-05, + "loss": 2.5660421371459963, + "step": 42680 + }, + { + "epoch": 0.11623792023955357, + "grad_norm": 2.194634437561035, + "learning_rate": 1.7699386670301214e-05, + "loss": 2.559774971008301, + "step": 42700 + }, + { + "epoch": 0.11629236423029808, + "grad_norm": 2.29994535446167, + "learning_rate": 1.769829630639226e-05, + "loss": 2.5323244094848634, + "step": 42720 + }, + { + "epoch": 0.1163468082210426, + "grad_norm": 2.6048152446746826, + "learning_rate": 1.7697205942483305e-05, + "loss": 2.505926322937012, + "step": 42740 + }, + { + "epoch": 0.11640125221178713, + "grad_norm": 2.051499366760254, + "learning_rate": 1.769611557857435e-05, + "loss": 2.582682228088379, + "step": 42760 + }, + { + "epoch": 0.11645569620253164, + "grad_norm": 2.371623992919922, + "learning_rate": 1.7695025214665397e-05, + "loss": 2.5687320709228514, + "step": 42780 + }, + { + "epoch": 0.11651014019327617, + "grad_norm": 2.2780559062957764, + "learning_rate": 1.7693934850756443e-05, + "loss": 2.5570478439331055, + "step": 42800 + }, + { + "epoch": 0.11656458418402069, + "grad_norm": 2.072434425354004, + "learning_rate": 1.7692844486847485e-05, + "loss": 2.5161413192749023, + "step": 42820 + }, + { + "epoch": 0.11661902817476522, + "grad_norm": 2.839136838912964, + "learning_rate": 1.769175412293853e-05, + "loss": 2.5619503021240235, + "step": 42840 + }, + { + "epoch": 0.11667347216550973, + "grad_norm": 2.153491735458374, + "learning_rate": 1.7690663759029577e-05, + "loss": 2.5162422180175783, + "step": 42860 + }, + { + "epoch": 0.11672791615625425, + "grad_norm": 2.0951485633850098, + "learning_rate": 1.7689573395120623e-05, + "loss": 2.557778549194336, + "step": 42880 + }, + { + "epoch": 0.11678236014699878, + "grad_norm": 2.15541410446167, + "learning_rate": 1.768848303121167e-05, + "loss": 2.573769378662109, + "step": 42900 + }, + { + "epoch": 0.11683680413774329, + "grad_norm": 2.16461181640625, + "learning_rate": 1.7687392667302714e-05, + "loss": 2.5385000228881838, + "step": 42920 + }, + { + "epoch": 0.11689124812848782, + "grad_norm": 2.2671194076538086, + "learning_rate": 1.768630230339376e-05, + "loss": 2.536548042297363, + "step": 42940 + }, + { + "epoch": 0.11694569211923234, + "grad_norm": 2.2182445526123047, + "learning_rate": 1.7685211939484806e-05, + "loss": 2.5226934432983397, + "step": 42960 + }, + { + "epoch": 0.11700013610997687, + "grad_norm": 2.143728017807007, + "learning_rate": 1.768412157557585e-05, + "loss": 2.5431690216064453, + "step": 42980 + }, + { + "epoch": 0.11705458010072138, + "grad_norm": 2.0908899307250977, + "learning_rate": 1.7683031211666894e-05, + "loss": 2.5698143005371095, + "step": 43000 + }, + { + "epoch": 0.11710902409146591, + "grad_norm": 2.1053779125213623, + "learning_rate": 1.768194084775794e-05, + "loss": 2.5422582626342773, + "step": 43020 + }, + { + "epoch": 0.11716346808221043, + "grad_norm": 2.2049074172973633, + "learning_rate": 1.7680850483848986e-05, + "loss": 2.543682098388672, + "step": 43040 + }, + { + "epoch": 0.11721791207295494, + "grad_norm": 2.121757745742798, + "learning_rate": 1.767976011994003e-05, + "loss": 2.5506259918212892, + "step": 43060 + }, + { + "epoch": 0.11727235606369947, + "grad_norm": 2.2451021671295166, + "learning_rate": 1.7678669756031077e-05, + "loss": 2.545684814453125, + "step": 43080 + }, + { + "epoch": 0.11732680005444399, + "grad_norm": 2.1364924907684326, + "learning_rate": 1.7677579392122123e-05, + "loss": 2.5068376541137694, + "step": 43100 + }, + { + "epoch": 0.11738124404518852, + "grad_norm": 2.064337968826294, + "learning_rate": 1.767648902821317e-05, + "loss": 2.5549739837646483, + "step": 43120 + }, + { + "epoch": 0.11743568803593303, + "grad_norm": 2.113471508026123, + "learning_rate": 1.767539866430421e-05, + "loss": 2.5219221115112305, + "step": 43140 + }, + { + "epoch": 0.11749013202667756, + "grad_norm": 2.152401924133301, + "learning_rate": 1.7674308300395257e-05, + "loss": 2.5398090362548826, + "step": 43160 + }, + { + "epoch": 0.11754457601742208, + "grad_norm": 2.089707136154175, + "learning_rate": 1.7673217936486303e-05, + "loss": 2.544808006286621, + "step": 43180 + }, + { + "epoch": 0.1175990200081666, + "grad_norm": 2.2694482803344727, + "learning_rate": 1.767212757257735e-05, + "loss": 2.571455383300781, + "step": 43200 + }, + { + "epoch": 0.11765346399891112, + "grad_norm": 2.1565327644348145, + "learning_rate": 1.7671037208668395e-05, + "loss": 2.547476577758789, + "step": 43220 + }, + { + "epoch": 0.11770790798965564, + "grad_norm": 2.114063262939453, + "learning_rate": 1.766994684475944e-05, + "loss": 2.5153530120849608, + "step": 43240 + }, + { + "epoch": 0.11776235198040017, + "grad_norm": 2.1147947311401367, + "learning_rate": 1.7668856480850486e-05, + "loss": 2.581573486328125, + "step": 43260 + }, + { + "epoch": 0.11781679597114468, + "grad_norm": 2.337357521057129, + "learning_rate": 1.7667766116941532e-05, + "loss": 2.535326385498047, + "step": 43280 + }, + { + "epoch": 0.11787123996188921, + "grad_norm": 2.273089647293091, + "learning_rate": 1.7666675753032575e-05, + "loss": 2.5239961624145506, + "step": 43300 + }, + { + "epoch": 0.11792568395263373, + "grad_norm": 4.912265777587891, + "learning_rate": 1.766558538912362e-05, + "loss": 2.5311222076416016, + "step": 43320 + }, + { + "epoch": 0.11798012794337825, + "grad_norm": 2.321180820465088, + "learning_rate": 1.7664495025214666e-05, + "loss": 2.562734603881836, + "step": 43340 + }, + { + "epoch": 0.11803457193412277, + "grad_norm": 2.149627923965454, + "learning_rate": 1.7663404661305712e-05, + "loss": 2.5577232360839846, + "step": 43360 + }, + { + "epoch": 0.1180890159248673, + "grad_norm": 2.2531495094299316, + "learning_rate": 1.7662314297396758e-05, + "loss": 2.526377868652344, + "step": 43380 + }, + { + "epoch": 0.11814345991561181, + "grad_norm": 2.1825335025787354, + "learning_rate": 1.7661223933487804e-05, + "loss": 2.564112091064453, + "step": 43400 + }, + { + "epoch": 0.11819790390635633, + "grad_norm": 2.1173465251922607, + "learning_rate": 1.766013356957885e-05, + "loss": 2.5274938583374023, + "step": 43420 + }, + { + "epoch": 0.11825234789710086, + "grad_norm": 2.2250864505767822, + "learning_rate": 1.7659043205669895e-05, + "loss": 2.5507530212402343, + "step": 43440 + }, + { + "epoch": 0.11830679188784538, + "grad_norm": 2.101551055908203, + "learning_rate": 1.7657952841760938e-05, + "loss": 2.559111785888672, + "step": 43460 + }, + { + "epoch": 0.1183612358785899, + "grad_norm": 4.205621719360352, + "learning_rate": 1.7656862477851983e-05, + "loss": 2.5497045516967773, + "step": 43480 + }, + { + "epoch": 0.11841567986933442, + "grad_norm": 2.2332305908203125, + "learning_rate": 1.765577211394303e-05, + "loss": 2.552522659301758, + "step": 43500 + }, + { + "epoch": 0.11847012386007895, + "grad_norm": 2.0935287475585938, + "learning_rate": 1.7654681750034075e-05, + "loss": 2.5293510437011717, + "step": 43520 + }, + { + "epoch": 0.11852456785082346, + "grad_norm": 2.105924606323242, + "learning_rate": 1.765359138612512e-05, + "loss": 2.5455230712890624, + "step": 43540 + }, + { + "epoch": 0.11857901184156798, + "grad_norm": 2.1004538536071777, + "learning_rate": 1.7652501022216167e-05, + "loss": 2.524643135070801, + "step": 43560 + }, + { + "epoch": 0.11863345583231251, + "grad_norm": 2.1668167114257812, + "learning_rate": 1.7651410658307213e-05, + "loss": 2.538730812072754, + "step": 43580 + }, + { + "epoch": 0.11868789982305702, + "grad_norm": 2.1514172554016113, + "learning_rate": 1.765032029439826e-05, + "loss": 2.537051773071289, + "step": 43600 + }, + { + "epoch": 0.11874234381380155, + "grad_norm": 2.548018217086792, + "learning_rate": 1.76492299304893e-05, + "loss": 2.5432464599609377, + "step": 43620 + }, + { + "epoch": 0.11879678780454607, + "grad_norm": 2.106210231781006, + "learning_rate": 1.7648139566580347e-05, + "loss": 2.549130439758301, + "step": 43640 + }, + { + "epoch": 0.1188512317952906, + "grad_norm": 2.0906457901000977, + "learning_rate": 1.7647049202671392e-05, + "loss": 2.53143253326416, + "step": 43660 + }, + { + "epoch": 0.11890567578603511, + "grad_norm": 2.115119218826294, + "learning_rate": 1.7645958838762438e-05, + "loss": 2.4990764617919923, + "step": 43680 + }, + { + "epoch": 0.11896011977677964, + "grad_norm": 2.1360020637512207, + "learning_rate": 1.7644868474853484e-05, + "loss": 2.554409980773926, + "step": 43700 + }, + { + "epoch": 0.11901456376752416, + "grad_norm": 2.219923496246338, + "learning_rate": 1.764377811094453e-05, + "loss": 2.589762496948242, + "step": 43720 + }, + { + "epoch": 0.11906900775826867, + "grad_norm": 2.1493992805480957, + "learning_rate": 1.7642687747035576e-05, + "loss": 2.566301727294922, + "step": 43740 + }, + { + "epoch": 0.1191234517490132, + "grad_norm": 2.181527614593506, + "learning_rate": 1.764159738312662e-05, + "loss": 2.5278444290161133, + "step": 43760 + }, + { + "epoch": 0.11917789573975772, + "grad_norm": 2.1719706058502197, + "learning_rate": 1.7640507019217664e-05, + "loss": 2.5452903747558593, + "step": 43780 + }, + { + "epoch": 0.11923233973050225, + "grad_norm": 2.207106828689575, + "learning_rate": 1.763941665530871e-05, + "loss": 2.5521757125854494, + "step": 43800 + }, + { + "epoch": 0.11928678372124676, + "grad_norm": 2.2120678424835205, + "learning_rate": 1.7638326291399755e-05, + "loss": 2.5310083389282227, + "step": 43820 + }, + { + "epoch": 0.1193412277119913, + "grad_norm": 2.2307629585266113, + "learning_rate": 1.76372359274908e-05, + "loss": 2.5401052474975585, + "step": 43840 + }, + { + "epoch": 0.11939567170273581, + "grad_norm": 1.983925223350525, + "learning_rate": 1.7636145563581847e-05, + "loss": 2.536098289489746, + "step": 43860 + }, + { + "epoch": 0.11945011569348034, + "grad_norm": 2.0714333057403564, + "learning_rate": 1.7635055199672893e-05, + "loss": 2.5639238357543945, + "step": 43880 + }, + { + "epoch": 0.11950455968422485, + "grad_norm": 2.1242153644561768, + "learning_rate": 1.763396483576394e-05, + "loss": 2.5455265045166016, + "step": 43900 + }, + { + "epoch": 0.11955900367496937, + "grad_norm": 2.269913911819458, + "learning_rate": 1.7632874471854984e-05, + "loss": 2.5317008972167967, + "step": 43920 + }, + { + "epoch": 0.1196134476657139, + "grad_norm": 2.09926176071167, + "learning_rate": 1.7631784107946027e-05, + "loss": 2.5642818450927733, + "step": 43940 + }, + { + "epoch": 0.11966789165645841, + "grad_norm": 2.357940196990967, + "learning_rate": 1.7630693744037073e-05, + "loss": 2.53582878112793, + "step": 43960 + }, + { + "epoch": 0.11972233564720294, + "grad_norm": 3.2318737506866455, + "learning_rate": 1.762960338012812e-05, + "loss": 2.5372671127319335, + "step": 43980 + }, + { + "epoch": 0.11977677963794746, + "grad_norm": 2.198636054992676, + "learning_rate": 1.7628513016219164e-05, + "loss": 2.4957195281982423, + "step": 44000 + }, + { + "epoch": 0.11983122362869199, + "grad_norm": 2.356966972351074, + "learning_rate": 1.762742265231021e-05, + "loss": 2.51993408203125, + "step": 44020 + }, + { + "epoch": 0.1198856676194365, + "grad_norm": 2.40629506111145, + "learning_rate": 1.7626332288401256e-05, + "loss": 2.509454917907715, + "step": 44040 + }, + { + "epoch": 0.11994011161018103, + "grad_norm": 2.2629575729370117, + "learning_rate": 1.7625241924492302e-05, + "loss": 2.5338136672973635, + "step": 44060 + }, + { + "epoch": 0.11999455560092555, + "grad_norm": 2.2029006481170654, + "learning_rate": 1.7624151560583348e-05, + "loss": 2.5228288650512694, + "step": 44080 + }, + { + "epoch": 0.12004899959167006, + "grad_norm": 2.1815834045410156, + "learning_rate": 1.762306119667439e-05, + "loss": 2.572252082824707, + "step": 44100 + }, + { + "epoch": 0.12010344358241459, + "grad_norm": 2.1450326442718506, + "learning_rate": 1.7621970832765436e-05, + "loss": 2.544535446166992, + "step": 44120 + }, + { + "epoch": 0.12015788757315911, + "grad_norm": 2.260279893875122, + "learning_rate": 1.762088046885648e-05, + "loss": 2.561406898498535, + "step": 44140 + }, + { + "epoch": 0.12021233156390364, + "grad_norm": 2.17668080329895, + "learning_rate": 1.7619790104947527e-05, + "loss": 2.528061294555664, + "step": 44160 + }, + { + "epoch": 0.12026677555464815, + "grad_norm": 2.6495141983032227, + "learning_rate": 1.7618699741038573e-05, + "loss": 2.553346061706543, + "step": 44180 + }, + { + "epoch": 0.12032121954539268, + "grad_norm": 2.2001664638519287, + "learning_rate": 1.761760937712962e-05, + "loss": 2.5800424575805665, + "step": 44200 + }, + { + "epoch": 0.1203756635361372, + "grad_norm": 2.123779773712158, + "learning_rate": 1.7616519013220665e-05, + "loss": 2.521603012084961, + "step": 44220 + }, + { + "epoch": 0.12043010752688173, + "grad_norm": 2.24766206741333, + "learning_rate": 1.761542864931171e-05, + "loss": 2.51965446472168, + "step": 44240 + }, + { + "epoch": 0.12048455151762624, + "grad_norm": 2.537675380706787, + "learning_rate": 1.7614338285402753e-05, + "loss": 2.554439163208008, + "step": 44260 + }, + { + "epoch": 0.12053899550837076, + "grad_norm": 2.1037111282348633, + "learning_rate": 1.76132479214938e-05, + "loss": 2.5539873123168944, + "step": 44280 + }, + { + "epoch": 0.12059343949911529, + "grad_norm": 2.1299283504486084, + "learning_rate": 1.7612157557584845e-05, + "loss": 2.5494434356689455, + "step": 44300 + }, + { + "epoch": 0.1206478834898598, + "grad_norm": 2.301481008529663, + "learning_rate": 1.761106719367589e-05, + "loss": 2.57998046875, + "step": 44320 + }, + { + "epoch": 0.12070232748060433, + "grad_norm": 2.0772101879119873, + "learning_rate": 1.7609976829766936e-05, + "loss": 2.5376659393310548, + "step": 44340 + }, + { + "epoch": 0.12075677147134885, + "grad_norm": 2.0228023529052734, + "learning_rate": 1.7608886465857982e-05, + "loss": 2.5360536575317383, + "step": 44360 + }, + { + "epoch": 0.12081121546209338, + "grad_norm": 2.032440662384033, + "learning_rate": 1.7607796101949025e-05, + "loss": 2.5393152236938477, + "step": 44380 + }, + { + "epoch": 0.12086565945283789, + "grad_norm": 2.0449886322021484, + "learning_rate": 1.7606705738040074e-05, + "loss": 2.577625274658203, + "step": 44400 + }, + { + "epoch": 0.12092010344358241, + "grad_norm": 2.2448601722717285, + "learning_rate": 1.7605615374131116e-05, + "loss": 2.520222854614258, + "step": 44420 + }, + { + "epoch": 0.12097454743432694, + "grad_norm": 2.2083358764648438, + "learning_rate": 1.7604525010222162e-05, + "loss": 2.533464050292969, + "step": 44440 + }, + { + "epoch": 0.12102899142507145, + "grad_norm": 2.1813511848449707, + "learning_rate": 1.7603434646313208e-05, + "loss": 2.5679964065551757, + "step": 44460 + }, + { + "epoch": 0.12108343541581598, + "grad_norm": 2.273892879486084, + "learning_rate": 1.7602344282404254e-05, + "loss": 2.5293107986450196, + "step": 44480 + }, + { + "epoch": 0.1211378794065605, + "grad_norm": 2.110487699508667, + "learning_rate": 1.76012539184953e-05, + "loss": 2.5425107955932615, + "step": 44500 + }, + { + "epoch": 0.12119232339730503, + "grad_norm": 2.1104207038879395, + "learning_rate": 1.7600163554586345e-05, + "loss": 2.5287458419799806, + "step": 44520 + }, + { + "epoch": 0.12124676738804954, + "grad_norm": 2.1608731746673584, + "learning_rate": 1.7599073190677388e-05, + "loss": 2.5473913192749023, + "step": 44540 + }, + { + "epoch": 0.12130121137879407, + "grad_norm": 2.797501802444458, + "learning_rate": 1.7597982826768437e-05, + "loss": 2.5348657608032226, + "step": 44560 + }, + { + "epoch": 0.12135565536953859, + "grad_norm": 2.1526482105255127, + "learning_rate": 1.759689246285948e-05, + "loss": 2.5333412170410154, + "step": 44580 + }, + { + "epoch": 0.1214100993602831, + "grad_norm": 2.1748743057250977, + "learning_rate": 1.7595802098950525e-05, + "loss": 2.532379913330078, + "step": 44600 + }, + { + "epoch": 0.12146454335102763, + "grad_norm": 2.1615779399871826, + "learning_rate": 1.759471173504157e-05, + "loss": 2.5458795547485353, + "step": 44620 + }, + { + "epoch": 0.12151898734177215, + "grad_norm": 2.10370135307312, + "learning_rate": 1.7593621371132617e-05, + "loss": 2.534330368041992, + "step": 44640 + }, + { + "epoch": 0.12157343133251668, + "grad_norm": 2.181561231613159, + "learning_rate": 1.7592531007223662e-05, + "loss": 2.541735076904297, + "step": 44660 + }, + { + "epoch": 0.12162787532326119, + "grad_norm": 2.2809951305389404, + "learning_rate": 1.7591440643314708e-05, + "loss": 2.5410079956054688, + "step": 44680 + }, + { + "epoch": 0.12168231931400572, + "grad_norm": 2.23470139503479, + "learning_rate": 1.759035027940575e-05, + "loss": 2.578118896484375, + "step": 44700 + }, + { + "epoch": 0.12173676330475024, + "grad_norm": 2.162932872772217, + "learning_rate": 1.75892599154968e-05, + "loss": 2.548183631896973, + "step": 44720 + }, + { + "epoch": 0.12179120729549477, + "grad_norm": 2.257504940032959, + "learning_rate": 1.7588169551587842e-05, + "loss": 2.5426868438720702, + "step": 44740 + }, + { + "epoch": 0.12184565128623928, + "grad_norm": 2.2564597129821777, + "learning_rate": 1.7587079187678888e-05, + "loss": 2.5480369567871093, + "step": 44760 + }, + { + "epoch": 0.1219000952769838, + "grad_norm": 2.147704839706421, + "learning_rate": 1.7585988823769934e-05, + "loss": 2.527927780151367, + "step": 44780 + }, + { + "epoch": 0.12195453926772833, + "grad_norm": 2.3660504817962646, + "learning_rate": 1.758489845986098e-05, + "loss": 2.521548271179199, + "step": 44800 + }, + { + "epoch": 0.12200898325847284, + "grad_norm": 2.1646766662597656, + "learning_rate": 1.7583808095952026e-05, + "loss": 2.5316314697265625, + "step": 44820 + }, + { + "epoch": 0.12206342724921737, + "grad_norm": 2.0620174407958984, + "learning_rate": 1.758271773204307e-05, + "loss": 2.533623695373535, + "step": 44840 + }, + { + "epoch": 0.12211787123996189, + "grad_norm": 2.2284677028656006, + "learning_rate": 1.7581627368134114e-05, + "loss": 2.535648727416992, + "step": 44860 + }, + { + "epoch": 0.12217231523070642, + "grad_norm": 2.2565701007843018, + "learning_rate": 1.7580537004225163e-05, + "loss": 2.573968505859375, + "step": 44880 + }, + { + "epoch": 0.12222675922145093, + "grad_norm": 2.2286202907562256, + "learning_rate": 1.7579446640316205e-05, + "loss": 2.5072067260742186, + "step": 44900 + }, + { + "epoch": 0.12228120321219546, + "grad_norm": 3.1952617168426514, + "learning_rate": 1.757835627640725e-05, + "loss": 2.5294977188110352, + "step": 44920 + }, + { + "epoch": 0.12233564720293998, + "grad_norm": 2.40496826171875, + "learning_rate": 1.7577265912498297e-05, + "loss": 2.550784111022949, + "step": 44940 + }, + { + "epoch": 0.12239009119368449, + "grad_norm": 2.1519196033477783, + "learning_rate": 1.7576175548589343e-05, + "loss": 2.542201042175293, + "step": 44960 + }, + { + "epoch": 0.12244453518442902, + "grad_norm": 2.324496030807495, + "learning_rate": 1.757508518468039e-05, + "loss": 2.526931381225586, + "step": 44980 + }, + { + "epoch": 0.12249897917517354, + "grad_norm": 2.132888078689575, + "learning_rate": 1.7573994820771434e-05, + "loss": 2.536220169067383, + "step": 45000 + }, + { + "epoch": 0.12255342316591807, + "grad_norm": 2.189199924468994, + "learning_rate": 1.7572904456862477e-05, + "loss": 2.5450166702270507, + "step": 45020 + }, + { + "epoch": 0.12260786715666258, + "grad_norm": 2.083838701248169, + "learning_rate": 1.7571814092953526e-05, + "loss": 2.562422752380371, + "step": 45040 + }, + { + "epoch": 0.12266231114740711, + "grad_norm": 2.305612087249756, + "learning_rate": 1.757072372904457e-05, + "loss": 2.5590824127197265, + "step": 45060 + }, + { + "epoch": 0.12271675513815163, + "grad_norm": 2.146763324737549, + "learning_rate": 1.7569633365135614e-05, + "loss": 2.563155746459961, + "step": 45080 + }, + { + "epoch": 0.12277119912889614, + "grad_norm": 2.1139016151428223, + "learning_rate": 1.756854300122666e-05, + "loss": 2.5350671768188477, + "step": 45100 + }, + { + "epoch": 0.12282564311964067, + "grad_norm": 2.061697244644165, + "learning_rate": 1.7567452637317706e-05, + "loss": 2.518401336669922, + "step": 45120 + }, + { + "epoch": 0.12288008711038519, + "grad_norm": 2.196894407272339, + "learning_rate": 1.7566362273408752e-05, + "loss": 2.5380685806274412, + "step": 45140 + }, + { + "epoch": 0.12293453110112972, + "grad_norm": 2.1212687492370605, + "learning_rate": 1.7565271909499798e-05, + "loss": 2.5477340698242186, + "step": 45160 + }, + { + "epoch": 0.12298897509187423, + "grad_norm": 2.1486713886260986, + "learning_rate": 1.756418154559084e-05, + "loss": 2.529057502746582, + "step": 45180 + }, + { + "epoch": 0.12304341908261876, + "grad_norm": 2.1684675216674805, + "learning_rate": 1.756309118168189e-05, + "loss": 2.5409923553466798, + "step": 45200 + }, + { + "epoch": 0.12309786307336328, + "grad_norm": 2.0693159103393555, + "learning_rate": 1.756200081777293e-05, + "loss": 2.540645217895508, + "step": 45220 + }, + { + "epoch": 0.1231523070641078, + "grad_norm": 2.028236150741577, + "learning_rate": 1.756091045386398e-05, + "loss": 2.53021240234375, + "step": 45240 + }, + { + "epoch": 0.12320675105485232, + "grad_norm": 2.2070493698120117, + "learning_rate": 1.7559820089955023e-05, + "loss": 2.553508758544922, + "step": 45260 + }, + { + "epoch": 0.12326119504559684, + "grad_norm": 2.2139365673065186, + "learning_rate": 1.755872972604607e-05, + "loss": 2.537653350830078, + "step": 45280 + }, + { + "epoch": 0.12331563903634137, + "grad_norm": 2.1682963371276855, + "learning_rate": 1.7557639362137115e-05, + "loss": 2.533161163330078, + "step": 45300 + }, + { + "epoch": 0.12337008302708588, + "grad_norm": 2.2770423889160156, + "learning_rate": 1.755654899822816e-05, + "loss": 2.5489595413208006, + "step": 45320 + }, + { + "epoch": 0.12342452701783041, + "grad_norm": 2.1297872066497803, + "learning_rate": 1.7555458634319203e-05, + "loss": 2.547117233276367, + "step": 45340 + }, + { + "epoch": 0.12347897100857493, + "grad_norm": 2.1757326126098633, + "learning_rate": 1.7554368270410252e-05, + "loss": 2.5349542617797853, + "step": 45360 + }, + { + "epoch": 0.12353341499931945, + "grad_norm": 2.0877597332000732, + "learning_rate": 1.7553277906501295e-05, + "loss": 2.5121026992797852, + "step": 45380 + }, + { + "epoch": 0.12358785899006397, + "grad_norm": 2.1536643505096436, + "learning_rate": 1.7552187542592344e-05, + "loss": 2.563639450073242, + "step": 45400 + }, + { + "epoch": 0.1236423029808085, + "grad_norm": 2.10336971282959, + "learning_rate": 1.7551097178683386e-05, + "loss": 2.5438167572021486, + "step": 45420 + }, + { + "epoch": 0.12369674697155301, + "grad_norm": 2.2554562091827393, + "learning_rate": 1.7550006814774432e-05, + "loss": 2.5444107055664062, + "step": 45440 + }, + { + "epoch": 0.12375119096229753, + "grad_norm": 2.0556678771972656, + "learning_rate": 1.7548916450865478e-05, + "loss": 2.534837341308594, + "step": 45460 + }, + { + "epoch": 0.12380563495304206, + "grad_norm": 2.3333754539489746, + "learning_rate": 1.7547826086956524e-05, + "loss": 2.5414270401000976, + "step": 45480 + }, + { + "epoch": 0.12386007894378658, + "grad_norm": 2.125147581100464, + "learning_rate": 1.7546735723047566e-05, + "loss": 2.53747673034668, + "step": 45500 + }, + { + "epoch": 0.1239145229345311, + "grad_norm": 2.124734401702881, + "learning_rate": 1.7545645359138615e-05, + "loss": 2.560394287109375, + "step": 45520 + }, + { + "epoch": 0.12396896692527562, + "grad_norm": 2.0724902153015137, + "learning_rate": 1.7544554995229658e-05, + "loss": 2.5308759689331053, + "step": 45540 + }, + { + "epoch": 0.12402341091602015, + "grad_norm": 2.171808958053589, + "learning_rate": 1.7543464631320707e-05, + "loss": 2.5326828002929687, + "step": 45560 + }, + { + "epoch": 0.12407785490676466, + "grad_norm": 2.153528928756714, + "learning_rate": 1.754237426741175e-05, + "loss": 2.541112518310547, + "step": 45580 + }, + { + "epoch": 0.1241322988975092, + "grad_norm": 2.164377212524414, + "learning_rate": 1.7541283903502795e-05, + "loss": 2.555303764343262, + "step": 45600 + }, + { + "epoch": 0.12418674288825371, + "grad_norm": 2.0425899028778076, + "learning_rate": 1.754019353959384e-05, + "loss": 2.49407901763916, + "step": 45620 + }, + { + "epoch": 0.12424118687899822, + "grad_norm": 2.273261308670044, + "learning_rate": 1.7539103175684887e-05, + "loss": 2.5133209228515625, + "step": 45640 + }, + { + "epoch": 0.12429563086974275, + "grad_norm": 2.1106247901916504, + "learning_rate": 1.753801281177593e-05, + "loss": 2.5683109283447267, + "step": 45660 + }, + { + "epoch": 0.12435007486048727, + "grad_norm": 2.20625638961792, + "learning_rate": 1.753692244786698e-05, + "loss": 2.5313089370727537, + "step": 45680 + }, + { + "epoch": 0.1244045188512318, + "grad_norm": 2.2488415241241455, + "learning_rate": 1.753583208395802e-05, + "loss": 2.5235769271850588, + "step": 45700 + }, + { + "epoch": 0.12445896284197631, + "grad_norm": 2.130152940750122, + "learning_rate": 1.753474172004907e-05, + "loss": 2.5011199951171874, + "step": 45720 + }, + { + "epoch": 0.12451340683272084, + "grad_norm": 2.1991078853607178, + "learning_rate": 1.7533651356140112e-05, + "loss": 2.554117965698242, + "step": 45740 + }, + { + "epoch": 0.12456785082346536, + "grad_norm": 2.1404716968536377, + "learning_rate": 1.7532560992231158e-05, + "loss": 2.5149953842163084, + "step": 45760 + }, + { + "epoch": 0.12462229481420987, + "grad_norm": 2.110229969024658, + "learning_rate": 1.7531470628322204e-05, + "loss": 2.5374467849731444, + "step": 45780 + }, + { + "epoch": 0.1246767388049544, + "grad_norm": 2.137843608856201, + "learning_rate": 1.753038026441325e-05, + "loss": 2.5595474243164062, + "step": 45800 + }, + { + "epoch": 0.12473118279569892, + "grad_norm": 2.2119288444519043, + "learning_rate": 1.7529289900504292e-05, + "loss": 2.5628473281860353, + "step": 45820 + }, + { + "epoch": 0.12478562678644345, + "grad_norm": 2.141958236694336, + "learning_rate": 1.752819953659534e-05, + "loss": 2.5331321716308595, + "step": 45840 + }, + { + "epoch": 0.12484007077718796, + "grad_norm": 2.2624409198760986, + "learning_rate": 1.7527109172686384e-05, + "loss": 2.5671363830566407, + "step": 45860 + }, + { + "epoch": 0.1248945147679325, + "grad_norm": 2.0808370113372803, + "learning_rate": 1.7526018808777433e-05, + "loss": 2.523858070373535, + "step": 45880 + }, + { + "epoch": 0.12494895875867701, + "grad_norm": 2.2064106464385986, + "learning_rate": 1.7524928444868476e-05, + "loss": 2.552276039123535, + "step": 45900 + }, + { + "epoch": 0.12500340274942154, + "grad_norm": 2.1416356563568115, + "learning_rate": 1.752383808095952e-05, + "loss": 2.5575462341308595, + "step": 45920 + }, + { + "epoch": 0.12505784674016607, + "grad_norm": 2.1287853717803955, + "learning_rate": 1.7522747717050567e-05, + "loss": 2.534903907775879, + "step": 45940 + }, + { + "epoch": 0.12511229073091057, + "grad_norm": 2.2627458572387695, + "learning_rate": 1.7521657353141613e-05, + "loss": 2.5411565780639647, + "step": 45960 + }, + { + "epoch": 0.1251667347216551, + "grad_norm": 2.869173526763916, + "learning_rate": 1.7520566989232655e-05, + "loss": 2.5573144912719727, + "step": 45980 + }, + { + "epoch": 0.12522117871239963, + "grad_norm": 2.0836057662963867, + "learning_rate": 1.7519476625323705e-05, + "loss": 2.5292097091674806, + "step": 46000 + }, + { + "epoch": 0.12527562270314413, + "grad_norm": 2.220062255859375, + "learning_rate": 1.7518386261414747e-05, + "loss": 2.53585147857666, + "step": 46020 + }, + { + "epoch": 0.12533006669388866, + "grad_norm": 2.1444220542907715, + "learning_rate": 1.7517295897505796e-05, + "loss": 2.5679304122924806, + "step": 46040 + }, + { + "epoch": 0.1253845106846332, + "grad_norm": 2.194479465484619, + "learning_rate": 1.751620553359684e-05, + "loss": 2.538955307006836, + "step": 46060 + }, + { + "epoch": 0.12543895467537772, + "grad_norm": 2.1569743156433105, + "learning_rate": 1.7515115169687884e-05, + "loss": 2.5554738998413087, + "step": 46080 + }, + { + "epoch": 0.12549339866612222, + "grad_norm": 2.2509233951568604, + "learning_rate": 1.751402480577893e-05, + "loss": 2.539499282836914, + "step": 46100 + }, + { + "epoch": 0.12554784265686675, + "grad_norm": 2.1748321056365967, + "learning_rate": 1.7512934441869976e-05, + "loss": 2.5348514556884765, + "step": 46120 + }, + { + "epoch": 0.12560228664761128, + "grad_norm": 2.0451202392578125, + "learning_rate": 1.751184407796102e-05, + "loss": 2.5224397659301756, + "step": 46140 + }, + { + "epoch": 0.12565673063835578, + "grad_norm": 2.0818769931793213, + "learning_rate": 1.7510753714052068e-05, + "loss": 2.5396209716796876, + "step": 46160 + }, + { + "epoch": 0.1257111746291003, + "grad_norm": 2.2200188636779785, + "learning_rate": 1.750966335014311e-05, + "loss": 2.5774347305297853, + "step": 46180 + }, + { + "epoch": 0.12576561861984484, + "grad_norm": 2.152923345565796, + "learning_rate": 1.750857298623416e-05, + "loss": 2.5247135162353516, + "step": 46200 + }, + { + "epoch": 0.12582006261058937, + "grad_norm": 2.1327645778656006, + "learning_rate": 1.7507482622325202e-05, + "loss": 2.5378654479980467, + "step": 46220 + }, + { + "epoch": 0.12587450660133387, + "grad_norm": 2.1493351459503174, + "learning_rate": 1.7506392258416248e-05, + "loss": 2.5264062881469727, + "step": 46240 + }, + { + "epoch": 0.1259289505920784, + "grad_norm": 2.2062628269195557, + "learning_rate": 1.7505301894507293e-05, + "loss": 2.5581756591796876, + "step": 46260 + }, + { + "epoch": 0.12598339458282293, + "grad_norm": 2.3656504154205322, + "learning_rate": 1.750421153059834e-05, + "loss": 2.5474084854125976, + "step": 46280 + }, + { + "epoch": 0.12603783857356743, + "grad_norm": 2.2709200382232666, + "learning_rate": 1.750312116668938e-05, + "loss": 2.539765739440918, + "step": 46300 + }, + { + "epoch": 0.12609228256431196, + "grad_norm": 2.1665656566619873, + "learning_rate": 1.750203080278043e-05, + "loss": 2.5758899688720702, + "step": 46320 + }, + { + "epoch": 0.1261467265550565, + "grad_norm": 2.1596546173095703, + "learning_rate": 1.7500940438871473e-05, + "loss": 2.519619941711426, + "step": 46340 + }, + { + "epoch": 0.12620117054580102, + "grad_norm": 2.452171802520752, + "learning_rate": 1.7499850074962522e-05, + "loss": 2.5620742797851563, + "step": 46360 + }, + { + "epoch": 0.12625561453654552, + "grad_norm": 2.169741630554199, + "learning_rate": 1.7498759711053565e-05, + "loss": 2.534422492980957, + "step": 46380 + }, + { + "epoch": 0.12631005852729005, + "grad_norm": 2.1484296321868896, + "learning_rate": 1.749766934714461e-05, + "loss": 2.5467754364013673, + "step": 46400 + }, + { + "epoch": 0.12636450251803458, + "grad_norm": 2.2043840885162354, + "learning_rate": 1.7496578983235656e-05, + "loss": 2.5600629806518556, + "step": 46420 + }, + { + "epoch": 0.1264189465087791, + "grad_norm": 2.2285361289978027, + "learning_rate": 1.7495488619326702e-05, + "loss": 2.582737922668457, + "step": 46440 + }, + { + "epoch": 0.1264733904995236, + "grad_norm": 2.1727705001831055, + "learning_rate": 1.7494398255417745e-05, + "loss": 2.542888641357422, + "step": 46460 + }, + { + "epoch": 0.12652783449026814, + "grad_norm": 2.0757319927215576, + "learning_rate": 1.7493307891508794e-05, + "loss": 2.5512626647949217, + "step": 46480 + }, + { + "epoch": 0.12658227848101267, + "grad_norm": 2.3452539443969727, + "learning_rate": 1.7492217527599836e-05, + "loss": 2.5269317626953125, + "step": 46500 + }, + { + "epoch": 0.12663672247175717, + "grad_norm": 2.1339468955993652, + "learning_rate": 1.7491127163690885e-05, + "loss": 2.563302421569824, + "step": 46520 + }, + { + "epoch": 0.1266911664625017, + "grad_norm": 2.1137540340423584, + "learning_rate": 1.7490036799781928e-05, + "loss": 2.5366851806640627, + "step": 46540 + }, + { + "epoch": 0.12674561045324623, + "grad_norm": 2.0678021907806396, + "learning_rate": 1.7488946435872974e-05, + "loss": 2.540476608276367, + "step": 46560 + }, + { + "epoch": 0.12680005444399076, + "grad_norm": 2.119382858276367, + "learning_rate": 1.748785607196402e-05, + "loss": 2.519335174560547, + "step": 46580 + }, + { + "epoch": 0.12685449843473526, + "grad_norm": 2.0789449214935303, + "learning_rate": 1.7486765708055065e-05, + "loss": 2.544474411010742, + "step": 46600 + }, + { + "epoch": 0.1269089424254798, + "grad_norm": 2.1682441234588623, + "learning_rate": 1.7485675344146108e-05, + "loss": 2.5517906188964843, + "step": 46620 + }, + { + "epoch": 0.12696338641622432, + "grad_norm": 2.085383415222168, + "learning_rate": 1.7484584980237157e-05, + "loss": 2.5750404357910157, + "step": 46640 + }, + { + "epoch": 0.12701783040696882, + "grad_norm": 2.2337746620178223, + "learning_rate": 1.74834946163282e-05, + "loss": 2.515431785583496, + "step": 46660 + }, + { + "epoch": 0.12707227439771335, + "grad_norm": 2.210906982421875, + "learning_rate": 1.748240425241925e-05, + "loss": 2.5374746322631836, + "step": 46680 + }, + { + "epoch": 0.12712671838845788, + "grad_norm": 9.359243392944336, + "learning_rate": 1.748131388851029e-05, + "loss": 2.5596298217773437, + "step": 46700 + }, + { + "epoch": 0.1271811623792024, + "grad_norm": 2.2212860584259033, + "learning_rate": 1.7480223524601337e-05, + "loss": 2.5506559371948243, + "step": 46720 + }, + { + "epoch": 0.1272356063699469, + "grad_norm": 2.0696206092834473, + "learning_rate": 1.7479133160692383e-05, + "loss": 2.541576957702637, + "step": 46740 + }, + { + "epoch": 0.12729005036069144, + "grad_norm": 2.1563611030578613, + "learning_rate": 1.747804279678343e-05, + "loss": 2.533756065368652, + "step": 46760 + }, + { + "epoch": 0.12734449435143597, + "grad_norm": 2.165778875350952, + "learning_rate": 1.747695243287447e-05, + "loss": 2.5482059478759767, + "step": 46780 + }, + { + "epoch": 0.1273989383421805, + "grad_norm": 2.1130707263946533, + "learning_rate": 1.747586206896552e-05, + "loss": 2.552077865600586, + "step": 46800 + }, + { + "epoch": 0.127453382332925, + "grad_norm": 2.1278228759765625, + "learning_rate": 1.7474771705056562e-05, + "loss": 2.5225954055786133, + "step": 46820 + }, + { + "epoch": 0.12750782632366953, + "grad_norm": 2.047557830810547, + "learning_rate": 1.747368134114761e-05, + "loss": 2.5445882797241213, + "step": 46840 + }, + { + "epoch": 0.12756227031441406, + "grad_norm": 2.4154164791107178, + "learning_rate": 1.7472590977238654e-05, + "loss": 2.5445331573486327, + "step": 46860 + }, + { + "epoch": 0.12761671430515856, + "grad_norm": 2.146318197250366, + "learning_rate": 1.74715006133297e-05, + "loss": 2.5760826110839843, + "step": 46880 + }, + { + "epoch": 0.1276711582959031, + "grad_norm": 2.0905635356903076, + "learning_rate": 1.7470410249420746e-05, + "loss": 2.516803169250488, + "step": 46900 + }, + { + "epoch": 0.12772560228664762, + "grad_norm": 2.116541862487793, + "learning_rate": 1.746931988551179e-05, + "loss": 2.5287593841552733, + "step": 46920 + }, + { + "epoch": 0.12778004627739215, + "grad_norm": 2.173222780227661, + "learning_rate": 1.7468229521602837e-05, + "loss": 2.532147216796875, + "step": 46940 + }, + { + "epoch": 0.12783449026813665, + "grad_norm": 2.143005132675171, + "learning_rate": 1.7467139157693883e-05, + "loss": 2.4966714859008787, + "step": 46960 + }, + { + "epoch": 0.12788893425888118, + "grad_norm": 2.1177968978881836, + "learning_rate": 1.7466048793784926e-05, + "loss": 2.5540725708007814, + "step": 46980 + }, + { + "epoch": 0.1279433782496257, + "grad_norm": 2.3478939533233643, + "learning_rate": 1.7464958429875975e-05, + "loss": 2.540041923522949, + "step": 47000 + }, + { + "epoch": 0.1279978222403702, + "grad_norm": 2.2193446159362793, + "learning_rate": 1.7463868065967017e-05, + "loss": 2.527898406982422, + "step": 47020 + }, + { + "epoch": 0.12805226623111474, + "grad_norm": 2.178083896636963, + "learning_rate": 1.7462777702058063e-05, + "loss": 2.5533010482788088, + "step": 47040 + }, + { + "epoch": 0.12810671022185927, + "grad_norm": 2.196622371673584, + "learning_rate": 1.746168733814911e-05, + "loss": 2.551490592956543, + "step": 47060 + }, + { + "epoch": 0.1281611542126038, + "grad_norm": 2.1853506565093994, + "learning_rate": 1.7460596974240155e-05, + "loss": 2.5431407928466796, + "step": 47080 + }, + { + "epoch": 0.1282155982033483, + "grad_norm": 2.23626446723938, + "learning_rate": 1.74595066103312e-05, + "loss": 2.528164100646973, + "step": 47100 + }, + { + "epoch": 0.12827004219409283, + "grad_norm": 2.551429271697998, + "learning_rate": 1.7458416246422246e-05, + "loss": 2.5358972549438477, + "step": 47120 + }, + { + "epoch": 0.12832448618483736, + "grad_norm": 2.1331472396850586, + "learning_rate": 1.745732588251329e-05, + "loss": 2.5570091247558593, + "step": 47140 + }, + { + "epoch": 0.12837893017558186, + "grad_norm": 2.188416004180908, + "learning_rate": 1.7456235518604338e-05, + "loss": 2.551318550109863, + "step": 47160 + }, + { + "epoch": 0.1284333741663264, + "grad_norm": 2.255896806716919, + "learning_rate": 1.745514515469538e-05, + "loss": 2.540979194641113, + "step": 47180 + }, + { + "epoch": 0.12848781815707092, + "grad_norm": 2.2769274711608887, + "learning_rate": 1.7454054790786426e-05, + "loss": 2.563247871398926, + "step": 47200 + }, + { + "epoch": 0.12854226214781544, + "grad_norm": 2.043170213699341, + "learning_rate": 1.7452964426877472e-05, + "loss": 2.5445995330810547, + "step": 47220 + }, + { + "epoch": 0.12859670613855995, + "grad_norm": 2.1373605728149414, + "learning_rate": 1.7451874062968518e-05, + "loss": 2.551808166503906, + "step": 47240 + }, + { + "epoch": 0.12865115012930448, + "grad_norm": 2.1031718254089355, + "learning_rate": 1.7450783699059563e-05, + "loss": 2.5432270050048826, + "step": 47260 + }, + { + "epoch": 0.128705594120049, + "grad_norm": 2.4180376529693604, + "learning_rate": 1.744969333515061e-05, + "loss": 2.5365215301513673, + "step": 47280 + }, + { + "epoch": 0.12876003811079353, + "grad_norm": 2.2050981521606445, + "learning_rate": 1.7448602971241652e-05, + "loss": 2.575900840759277, + "step": 47300 + }, + { + "epoch": 0.12881448210153804, + "grad_norm": 2.2118303775787354, + "learning_rate": 1.74475126073327e-05, + "loss": 2.532234001159668, + "step": 47320 + }, + { + "epoch": 0.12886892609228257, + "grad_norm": 2.257340908050537, + "learning_rate": 1.7446422243423743e-05, + "loss": 2.525872230529785, + "step": 47340 + }, + { + "epoch": 0.1289233700830271, + "grad_norm": 2.185203790664673, + "learning_rate": 1.744533187951479e-05, + "loss": 2.5631526947021483, + "step": 47360 + }, + { + "epoch": 0.1289778140737716, + "grad_norm": 2.125795364379883, + "learning_rate": 1.7444241515605835e-05, + "loss": 2.5491458892822267, + "step": 47380 + }, + { + "epoch": 0.12903225806451613, + "grad_norm": 2.208866834640503, + "learning_rate": 1.744315115169688e-05, + "loss": 2.571201515197754, + "step": 47400 + }, + { + "epoch": 0.12908670205526065, + "grad_norm": 2.264326810836792, + "learning_rate": 1.7442060787787927e-05, + "loss": 2.555317687988281, + "step": 47420 + }, + { + "epoch": 0.12914114604600518, + "grad_norm": 2.0768256187438965, + "learning_rate": 1.7440970423878972e-05, + "loss": 2.5502222061157225, + "step": 47440 + }, + { + "epoch": 0.12919559003674969, + "grad_norm": 2.10144305229187, + "learning_rate": 1.7439880059970015e-05, + "loss": 2.565609169006348, + "step": 47460 + }, + { + "epoch": 0.12925003402749421, + "grad_norm": 2.137507915496826, + "learning_rate": 1.7438789696061064e-05, + "loss": 2.5310455322265626, + "step": 47480 + }, + { + "epoch": 0.12930447801823874, + "grad_norm": 2.0800602436065674, + "learning_rate": 1.7437699332152106e-05, + "loss": 2.5214805603027344, + "step": 47500 + }, + { + "epoch": 0.12935892200898325, + "grad_norm": 2.140188455581665, + "learning_rate": 1.7436608968243152e-05, + "loss": 2.5014368057250977, + "step": 47520 + }, + { + "epoch": 0.12941336599972778, + "grad_norm": 2.177884578704834, + "learning_rate": 1.7435518604334198e-05, + "loss": 2.5759674072265626, + "step": 47540 + }, + { + "epoch": 0.1294678099904723, + "grad_norm": 2.106264591217041, + "learning_rate": 1.7434428240425244e-05, + "loss": 2.5499767303466796, + "step": 47560 + }, + { + "epoch": 0.12952225398121683, + "grad_norm": 2.2741622924804688, + "learning_rate": 1.743333787651629e-05, + "loss": 2.548648643493652, + "step": 47580 + }, + { + "epoch": 0.12957669797196134, + "grad_norm": 2.033992052078247, + "learning_rate": 1.7432247512607335e-05, + "loss": 2.537437629699707, + "step": 47600 + }, + { + "epoch": 0.12963114196270586, + "grad_norm": 2.1089372634887695, + "learning_rate": 1.7431157148698378e-05, + "loss": 2.526949882507324, + "step": 47620 + }, + { + "epoch": 0.1296855859534504, + "grad_norm": 2.163144826889038, + "learning_rate": 1.7430066784789427e-05, + "loss": 2.546768379211426, + "step": 47640 + }, + { + "epoch": 0.1297400299441949, + "grad_norm": 2.206049919128418, + "learning_rate": 1.742897642088047e-05, + "loss": 2.5409219741821287, + "step": 47660 + }, + { + "epoch": 0.12979447393493942, + "grad_norm": 2.381179094314575, + "learning_rate": 1.7427886056971515e-05, + "loss": 2.5195356369018556, + "step": 47680 + }, + { + "epoch": 0.12984891792568395, + "grad_norm": 2.1656692028045654, + "learning_rate": 1.742679569306256e-05, + "loss": 2.556634712219238, + "step": 47700 + }, + { + "epoch": 0.12990336191642848, + "grad_norm": 2.1954150199890137, + "learning_rate": 1.7425705329153607e-05, + "loss": 2.5566436767578127, + "step": 47720 + }, + { + "epoch": 0.12995780590717299, + "grad_norm": 2.1760125160217285, + "learning_rate": 1.7424614965244653e-05, + "loss": 2.5636875152587892, + "step": 47740 + }, + { + "epoch": 0.13001224989791751, + "grad_norm": 2.193229913711548, + "learning_rate": 1.74235246013357e-05, + "loss": 2.55328369140625, + "step": 47760 + }, + { + "epoch": 0.13006669388866204, + "grad_norm": 2.2989933490753174, + "learning_rate": 1.742243423742674e-05, + "loss": 2.5404176712036133, + "step": 47780 + }, + { + "epoch": 0.13012113787940657, + "grad_norm": 2.303880453109741, + "learning_rate": 1.742134387351779e-05, + "loss": 2.5540672302246095, + "step": 47800 + }, + { + "epoch": 0.13017558187015107, + "grad_norm": 2.1820716857910156, + "learning_rate": 1.7420253509608833e-05, + "loss": 2.5338558197021483, + "step": 47820 + }, + { + "epoch": 0.1302300258608956, + "grad_norm": 2.0795388221740723, + "learning_rate": 1.741916314569988e-05, + "loss": 2.5186046600341796, + "step": 47840 + }, + { + "epoch": 0.13028446985164013, + "grad_norm": 2.3383660316467285, + "learning_rate": 1.7418072781790924e-05, + "loss": 2.5392494201660156, + "step": 47860 + }, + { + "epoch": 0.13033891384238463, + "grad_norm": 2.189809560775757, + "learning_rate": 1.741698241788197e-05, + "loss": 2.5173419952392577, + "step": 47880 + }, + { + "epoch": 0.13039335783312916, + "grad_norm": 2.60253643989563, + "learning_rate": 1.7415892053973016e-05, + "loss": 2.5444154739379883, + "step": 47900 + }, + { + "epoch": 0.1304478018238737, + "grad_norm": 2.0086209774017334, + "learning_rate": 1.741480169006406e-05, + "loss": 2.5045492172241213, + "step": 47920 + }, + { + "epoch": 0.13050224581461822, + "grad_norm": 3.998199462890625, + "learning_rate": 1.7413711326155104e-05, + "loss": 2.5406587600708006, + "step": 47940 + }, + { + "epoch": 0.13055668980536272, + "grad_norm": 2.1633222103118896, + "learning_rate": 1.7412620962246153e-05, + "loss": 2.522216033935547, + "step": 47960 + }, + { + "epoch": 0.13061113379610725, + "grad_norm": 2.1278998851776123, + "learning_rate": 1.7411530598337196e-05, + "loss": 2.5685117721557615, + "step": 47980 + }, + { + "epoch": 0.13066557778685178, + "grad_norm": 2.2278196811676025, + "learning_rate": 1.741044023442824e-05, + "loss": 2.532418632507324, + "step": 48000 + }, + { + "epoch": 0.13072002177759628, + "grad_norm": 4.6987504959106445, + "learning_rate": 1.7409349870519287e-05, + "loss": 2.5397161483764648, + "step": 48020 + }, + { + "epoch": 0.13077446576834081, + "grad_norm": 3.3786888122558594, + "learning_rate": 1.7408259506610333e-05, + "loss": 2.587442398071289, + "step": 48040 + }, + { + "epoch": 0.13082890975908534, + "grad_norm": 2.351865291595459, + "learning_rate": 1.740716914270138e-05, + "loss": 2.542751693725586, + "step": 48060 + }, + { + "epoch": 0.13088335374982987, + "grad_norm": 2.244884967803955, + "learning_rate": 1.7406078778792425e-05, + "loss": 2.5445632934570312, + "step": 48080 + }, + { + "epoch": 0.13093779774057437, + "grad_norm": 2.093197822570801, + "learning_rate": 1.7404988414883467e-05, + "loss": 2.5705169677734374, + "step": 48100 + }, + { + "epoch": 0.1309922417313189, + "grad_norm": 2.1090846061706543, + "learning_rate": 1.7403898050974516e-05, + "loss": 2.531101036071777, + "step": 48120 + }, + { + "epoch": 0.13104668572206343, + "grad_norm": 2.2044854164123535, + "learning_rate": 1.740280768706556e-05, + "loss": 2.5182525634765627, + "step": 48140 + }, + { + "epoch": 0.13110112971280796, + "grad_norm": 2.2574031352996826, + "learning_rate": 1.7401717323156605e-05, + "loss": 2.536306381225586, + "step": 48160 + }, + { + "epoch": 0.13115557370355246, + "grad_norm": 2.1566107273101807, + "learning_rate": 1.740062695924765e-05, + "loss": 2.5460010528564454, + "step": 48180 + }, + { + "epoch": 0.131210017694297, + "grad_norm": 2.235647201538086, + "learning_rate": 1.7399536595338696e-05, + "loss": 2.5759984970092775, + "step": 48200 + }, + { + "epoch": 0.13126446168504152, + "grad_norm": 2.142138719558716, + "learning_rate": 1.7398446231429742e-05, + "loss": 2.535896873474121, + "step": 48220 + }, + { + "epoch": 0.13131890567578602, + "grad_norm": 2.178906202316284, + "learning_rate": 1.7397355867520788e-05, + "loss": 2.5750267028808596, + "step": 48240 + }, + { + "epoch": 0.13137334966653055, + "grad_norm": 2.0431814193725586, + "learning_rate": 1.739626550361183e-05, + "loss": 2.553925323486328, + "step": 48260 + }, + { + "epoch": 0.13142779365727508, + "grad_norm": 2.300359010696411, + "learning_rate": 1.739517513970288e-05, + "loss": 2.545322799682617, + "step": 48280 + }, + { + "epoch": 0.1314822376480196, + "grad_norm": 2.133605480194092, + "learning_rate": 1.7394084775793922e-05, + "loss": 2.5733129501342775, + "step": 48300 + }, + { + "epoch": 0.1315366816387641, + "grad_norm": 2.1315038204193115, + "learning_rate": 1.7392994411884968e-05, + "loss": 2.536163902282715, + "step": 48320 + }, + { + "epoch": 0.13159112562950864, + "grad_norm": 2.0961132049560547, + "learning_rate": 1.7391904047976013e-05, + "loss": 2.520612907409668, + "step": 48340 + }, + { + "epoch": 0.13164556962025317, + "grad_norm": 2.074115037918091, + "learning_rate": 1.739081368406706e-05, + "loss": 2.532573699951172, + "step": 48360 + }, + { + "epoch": 0.13170001361099767, + "grad_norm": 2.1091737747192383, + "learning_rate": 1.7389723320158105e-05, + "loss": 2.5010948181152344, + "step": 48380 + }, + { + "epoch": 0.1317544576017422, + "grad_norm": 2.4262073040008545, + "learning_rate": 1.738863295624915e-05, + "loss": 2.544930839538574, + "step": 48400 + }, + { + "epoch": 0.13180890159248673, + "grad_norm": 2.221015214920044, + "learning_rate": 1.7387542592340193e-05, + "loss": 2.5371706008911135, + "step": 48420 + }, + { + "epoch": 0.13186334558323126, + "grad_norm": 2.243976593017578, + "learning_rate": 1.7386452228431243e-05, + "loss": 2.53526611328125, + "step": 48440 + }, + { + "epoch": 0.13191778957397576, + "grad_norm": 2.1296796798706055, + "learning_rate": 1.7385361864522285e-05, + "loss": 2.5588274002075195, + "step": 48460 + }, + { + "epoch": 0.1319722335647203, + "grad_norm": 2.082359552383423, + "learning_rate": 1.738427150061333e-05, + "loss": 2.50537052154541, + "step": 48480 + }, + { + "epoch": 0.13202667755546482, + "grad_norm": 2.0325043201446533, + "learning_rate": 1.7383181136704377e-05, + "loss": 2.530133819580078, + "step": 48500 + }, + { + "epoch": 0.13208112154620932, + "grad_norm": 2.1076278686523438, + "learning_rate": 1.7382090772795422e-05, + "loss": 2.526844024658203, + "step": 48520 + }, + { + "epoch": 0.13213556553695385, + "grad_norm": 2.13112735748291, + "learning_rate": 1.7381000408886468e-05, + "loss": 2.548111343383789, + "step": 48540 + }, + { + "epoch": 0.13219000952769838, + "grad_norm": 2.1924026012420654, + "learning_rate": 1.7379910044977514e-05, + "loss": 2.511417007446289, + "step": 48560 + }, + { + "epoch": 0.1322444535184429, + "grad_norm": 2.0422110557556152, + "learning_rate": 1.7378819681068556e-05, + "loss": 2.5524911880493164, + "step": 48580 + }, + { + "epoch": 0.1322988975091874, + "grad_norm": 2.262523889541626, + "learning_rate": 1.7377729317159606e-05, + "loss": 2.553946685791016, + "step": 48600 + }, + { + "epoch": 0.13235334149993194, + "grad_norm": 2.122945547103882, + "learning_rate": 1.7376638953250648e-05, + "loss": 2.5085731506347657, + "step": 48620 + }, + { + "epoch": 0.13240778549067647, + "grad_norm": 2.1371262073516846, + "learning_rate": 1.7375548589341694e-05, + "loss": 2.5336360931396484, + "step": 48640 + }, + { + "epoch": 0.132462229481421, + "grad_norm": 2.1306567192077637, + "learning_rate": 1.737445822543274e-05, + "loss": 2.518023872375488, + "step": 48660 + }, + { + "epoch": 0.1325166734721655, + "grad_norm": 2.1216564178466797, + "learning_rate": 1.7373367861523785e-05, + "loss": 2.554083824157715, + "step": 48680 + }, + { + "epoch": 0.13257111746291003, + "grad_norm": 2.165091037750244, + "learning_rate": 1.737227749761483e-05, + "loss": 2.526091766357422, + "step": 48700 + }, + { + "epoch": 0.13262556145365456, + "grad_norm": 2.1603572368621826, + "learning_rate": 1.7371187133705877e-05, + "loss": 2.5608802795410157, + "step": 48720 + }, + { + "epoch": 0.13268000544439906, + "grad_norm": 2.046146869659424, + "learning_rate": 1.737009676979692e-05, + "loss": 2.549506187438965, + "step": 48740 + }, + { + "epoch": 0.1327344494351436, + "grad_norm": 2.3105485439300537, + "learning_rate": 1.736900640588797e-05, + "loss": 2.5465824127197267, + "step": 48760 + }, + { + "epoch": 0.13278889342588812, + "grad_norm": 2.1997926235198975, + "learning_rate": 1.736791604197901e-05, + "loss": 2.53875675201416, + "step": 48780 + }, + { + "epoch": 0.13284333741663265, + "grad_norm": 2.066392660140991, + "learning_rate": 1.7366825678070057e-05, + "loss": 2.5287710189819337, + "step": 48800 + }, + { + "epoch": 0.13289778140737715, + "grad_norm": 2.2993979454040527, + "learning_rate": 1.7365735314161103e-05, + "loss": 2.5365734100341797, + "step": 48820 + }, + { + "epoch": 0.13295222539812168, + "grad_norm": 2.150189161300659, + "learning_rate": 1.736464495025215e-05, + "loss": 2.575897979736328, + "step": 48840 + }, + { + "epoch": 0.1330066693888662, + "grad_norm": 2.109114646911621, + "learning_rate": 1.7363554586343194e-05, + "loss": 2.52020263671875, + "step": 48860 + }, + { + "epoch": 0.1330611133796107, + "grad_norm": 2.07564115524292, + "learning_rate": 1.736246422243424e-05, + "loss": 2.534503364562988, + "step": 48880 + }, + { + "epoch": 0.13311555737035524, + "grad_norm": 2.1219263076782227, + "learning_rate": 1.7361373858525283e-05, + "loss": 2.5399906158447267, + "step": 48900 + }, + { + "epoch": 0.13317000136109977, + "grad_norm": 2.811833620071411, + "learning_rate": 1.7360283494616332e-05, + "loss": 2.532508659362793, + "step": 48920 + }, + { + "epoch": 0.1332244453518443, + "grad_norm": 2.202667713165283, + "learning_rate": 1.7359193130707374e-05, + "loss": 2.553944396972656, + "step": 48940 + }, + { + "epoch": 0.1332788893425888, + "grad_norm": 2.118929386138916, + "learning_rate": 1.735810276679842e-05, + "loss": 2.5208316802978517, + "step": 48960 + }, + { + "epoch": 0.13333333333333333, + "grad_norm": 2.145620584487915, + "learning_rate": 1.7357012402889466e-05, + "loss": 2.5382938385009766, + "step": 48980 + }, + { + "epoch": 0.13338777732407786, + "grad_norm": 2.228862762451172, + "learning_rate": 1.735592203898051e-05, + "loss": 2.5606706619262694, + "step": 49000 + }, + { + "epoch": 0.1334422213148224, + "grad_norm": 2.1228086948394775, + "learning_rate": 1.7354831675071557e-05, + "loss": 2.5248027801513673, + "step": 49020 + }, + { + "epoch": 0.1334966653055669, + "grad_norm": 2.1042604446411133, + "learning_rate": 1.7353741311162603e-05, + "loss": 2.517018508911133, + "step": 49040 + }, + { + "epoch": 0.13355110929631142, + "grad_norm": 2.024075746536255, + "learning_rate": 1.7352650947253646e-05, + "loss": 2.538751792907715, + "step": 49060 + }, + { + "epoch": 0.13360555328705595, + "grad_norm": 2.4124372005462646, + "learning_rate": 1.7351560583344695e-05, + "loss": 2.540721321105957, + "step": 49080 + }, + { + "epoch": 0.13365999727780045, + "grad_norm": 2.1796374320983887, + "learning_rate": 1.7350470219435737e-05, + "loss": 2.5248695373535157, + "step": 49100 + }, + { + "epoch": 0.13371444126854498, + "grad_norm": 2.230452060699463, + "learning_rate": 1.7349379855526783e-05, + "loss": 2.531893730163574, + "step": 49120 + }, + { + "epoch": 0.1337688852592895, + "grad_norm": 2.1498537063598633, + "learning_rate": 1.734828949161783e-05, + "loss": 2.572377014160156, + "step": 49140 + }, + { + "epoch": 0.13382332925003404, + "grad_norm": 2.2366631031036377, + "learning_rate": 1.7347199127708875e-05, + "loss": 2.5612102508544923, + "step": 49160 + }, + { + "epoch": 0.13387777324077854, + "grad_norm": 2.9077978134155273, + "learning_rate": 1.734610876379992e-05, + "loss": 2.56463737487793, + "step": 49180 + }, + { + "epoch": 0.13393221723152307, + "grad_norm": 2.203057050704956, + "learning_rate": 1.7345018399890966e-05, + "loss": 2.5567447662353517, + "step": 49200 + }, + { + "epoch": 0.1339866612222676, + "grad_norm": 2.224104404449463, + "learning_rate": 1.734392803598201e-05, + "loss": 2.5482486724853515, + "step": 49220 + }, + { + "epoch": 0.1340411052130121, + "grad_norm": 2.1435108184814453, + "learning_rate": 1.7342837672073058e-05, + "loss": 2.5759872436523437, + "step": 49240 + }, + { + "epoch": 0.13409554920375663, + "grad_norm": 2.179710626602173, + "learning_rate": 1.73417473081641e-05, + "loss": 2.5672275543212892, + "step": 49260 + }, + { + "epoch": 0.13414999319450116, + "grad_norm": 2.2066361904144287, + "learning_rate": 1.7340656944255146e-05, + "loss": 2.526367950439453, + "step": 49280 + }, + { + "epoch": 0.1342044371852457, + "grad_norm": 2.193265914916992, + "learning_rate": 1.7339566580346192e-05, + "loss": 2.53153133392334, + "step": 49300 + }, + { + "epoch": 0.1342588811759902, + "grad_norm": 2.101193904876709, + "learning_rate": 1.7338476216437238e-05, + "loss": 2.5121648788452147, + "step": 49320 + }, + { + "epoch": 0.13431332516673472, + "grad_norm": 2.117736339569092, + "learning_rate": 1.7337385852528284e-05, + "loss": 2.540042495727539, + "step": 49340 + }, + { + "epoch": 0.13436776915747925, + "grad_norm": 2.6432666778564453, + "learning_rate": 1.733629548861933e-05, + "loss": 2.533946228027344, + "step": 49360 + }, + { + "epoch": 0.13442221314822375, + "grad_norm": 2.237636089324951, + "learning_rate": 1.7335205124710372e-05, + "loss": 2.5192956924438477, + "step": 49380 + }, + { + "epoch": 0.13447665713896828, + "grad_norm": 2.2350189685821533, + "learning_rate": 1.7334114760801418e-05, + "loss": 2.543422317504883, + "step": 49400 + }, + { + "epoch": 0.1345311011297128, + "grad_norm": 2.105436086654663, + "learning_rate": 1.7333024396892463e-05, + "loss": 2.540359306335449, + "step": 49420 + }, + { + "epoch": 0.13458554512045734, + "grad_norm": 2.038468360900879, + "learning_rate": 1.733193403298351e-05, + "loss": 2.5477691650390626, + "step": 49440 + }, + { + "epoch": 0.13463998911120184, + "grad_norm": 3.3442139625549316, + "learning_rate": 1.7330843669074555e-05, + "loss": 2.5608142852783202, + "step": 49460 + }, + { + "epoch": 0.13469443310194637, + "grad_norm": 2.111243486404419, + "learning_rate": 1.73297533051656e-05, + "loss": 2.535293960571289, + "step": 49480 + }, + { + "epoch": 0.1347488770926909, + "grad_norm": 2.1815497875213623, + "learning_rate": 1.7328662941256647e-05, + "loss": 2.544963836669922, + "step": 49500 + }, + { + "epoch": 0.13480332108343543, + "grad_norm": 2.1694657802581787, + "learning_rate": 1.7327572577347692e-05, + "loss": 2.5716251373291015, + "step": 49520 + }, + { + "epoch": 0.13485776507417993, + "grad_norm": 2.3871519565582275, + "learning_rate": 1.7326482213438735e-05, + "loss": 2.536408233642578, + "step": 49540 + }, + { + "epoch": 0.13491220906492446, + "grad_norm": 2.177164077758789, + "learning_rate": 1.732539184952978e-05, + "loss": 2.536149787902832, + "step": 49560 + }, + { + "epoch": 0.134966653055669, + "grad_norm": 2.0888078212738037, + "learning_rate": 1.7324301485620827e-05, + "loss": 2.523006820678711, + "step": 49580 + }, + { + "epoch": 0.1350210970464135, + "grad_norm": 2.2066757678985596, + "learning_rate": 1.7323211121711872e-05, + "loss": 2.5164154052734373, + "step": 49600 + }, + { + "epoch": 0.13507554103715802, + "grad_norm": 2.14357328414917, + "learning_rate": 1.7322120757802918e-05, + "loss": 2.540791130065918, + "step": 49620 + }, + { + "epoch": 0.13512998502790255, + "grad_norm": 2.1994998455047607, + "learning_rate": 1.7321030393893964e-05, + "loss": 2.516806221008301, + "step": 49640 + }, + { + "epoch": 0.13518442901864708, + "grad_norm": 2.1234006881713867, + "learning_rate": 1.731994002998501e-05, + "loss": 2.540247917175293, + "step": 49660 + }, + { + "epoch": 0.13523887300939158, + "grad_norm": 2.085430383682251, + "learning_rate": 1.7318849666076056e-05, + "loss": 2.5324392318725586, + "step": 49680 + }, + { + "epoch": 0.1352933170001361, + "grad_norm": 2.1492857933044434, + "learning_rate": 1.7317759302167098e-05, + "loss": 2.537144088745117, + "step": 49700 + }, + { + "epoch": 0.13534776099088064, + "grad_norm": 2.1491663455963135, + "learning_rate": 1.7316668938258144e-05, + "loss": 2.5547439575195314, + "step": 49720 + }, + { + "epoch": 0.13540220498162514, + "grad_norm": 2.159968376159668, + "learning_rate": 1.731557857434919e-05, + "loss": 2.549546813964844, + "step": 49740 + }, + { + "epoch": 0.13545664897236967, + "grad_norm": 2.1603100299835205, + "learning_rate": 1.7314488210440235e-05, + "loss": 2.515609550476074, + "step": 49760 + }, + { + "epoch": 0.1355110929631142, + "grad_norm": 2.1404471397399902, + "learning_rate": 1.731339784653128e-05, + "loss": 2.507175064086914, + "step": 49780 + }, + { + "epoch": 0.13556553695385873, + "grad_norm": 2.0565855503082275, + "learning_rate": 1.7312307482622327e-05, + "loss": 2.5264102935791017, + "step": 49800 + }, + { + "epoch": 0.13561998094460323, + "grad_norm": 2.9253416061401367, + "learning_rate": 1.7311217118713373e-05, + "loss": 2.526899528503418, + "step": 49820 + }, + { + "epoch": 0.13567442493534776, + "grad_norm": 2.051680326461792, + "learning_rate": 1.731012675480442e-05, + "loss": 2.5245494842529297, + "step": 49840 + }, + { + "epoch": 0.1357288689260923, + "grad_norm": 2.1421749591827393, + "learning_rate": 1.730903639089546e-05, + "loss": 2.5561656951904297, + "step": 49860 + }, + { + "epoch": 0.1357833129168368, + "grad_norm": 2.1105635166168213, + "learning_rate": 1.7307946026986507e-05, + "loss": 2.5589242935180665, + "step": 49880 + }, + { + "epoch": 0.13583775690758132, + "grad_norm": 2.1964194774627686, + "learning_rate": 1.7306855663077553e-05, + "loss": 2.5513177871704102, + "step": 49900 + }, + { + "epoch": 0.13589220089832585, + "grad_norm": 2.155120849609375, + "learning_rate": 1.73057652991686e-05, + "loss": 2.522687530517578, + "step": 49920 + }, + { + "epoch": 0.13594664488907038, + "grad_norm": 2.2259178161621094, + "learning_rate": 1.7304674935259644e-05, + "loss": 2.55340461730957, + "step": 49940 + }, + { + "epoch": 0.13600108887981488, + "grad_norm": 2.107361316680908, + "learning_rate": 1.730358457135069e-05, + "loss": 2.5257213592529295, + "step": 49960 + }, + { + "epoch": 0.1360555328705594, + "grad_norm": 2.2603776454925537, + "learning_rate": 1.7302494207441736e-05, + "loss": 2.5381389617919923, + "step": 49980 + }, + { + "epoch": 0.13610997686130394, + "grad_norm": 2.1967039108276367, + "learning_rate": 1.7301403843532782e-05, + "loss": 2.5565889358520506, + "step": 50000 + }, + { + "epoch": 0.13616442085204847, + "grad_norm": 2.1936392784118652, + "learning_rate": 1.7300313479623824e-05, + "loss": 2.5423330307006835, + "step": 50020 + }, + { + "epoch": 0.13621886484279297, + "grad_norm": 2.095224618911743, + "learning_rate": 1.729922311571487e-05, + "loss": 2.5614269256591795, + "step": 50040 + }, + { + "epoch": 0.1362733088335375, + "grad_norm": 2.309277057647705, + "learning_rate": 1.7298132751805916e-05, + "loss": 2.5422426223754884, + "step": 50060 + }, + { + "epoch": 0.13632775282428203, + "grad_norm": 2.108131170272827, + "learning_rate": 1.729704238789696e-05, + "loss": 2.554815483093262, + "step": 50080 + }, + { + "epoch": 0.13638219681502653, + "grad_norm": 2.1262736320495605, + "learning_rate": 1.7295952023988007e-05, + "loss": 2.5167400360107424, + "step": 50100 + }, + { + "epoch": 0.13643664080577106, + "grad_norm": 2.1777827739715576, + "learning_rate": 1.7294861660079053e-05, + "loss": 2.522099494934082, + "step": 50120 + }, + { + "epoch": 0.1364910847965156, + "grad_norm": 2.14810848236084, + "learning_rate": 1.72937712961701e-05, + "loss": 2.5562536239624025, + "step": 50140 + }, + { + "epoch": 0.13654552878726012, + "grad_norm": 2.169658660888672, + "learning_rate": 1.7292680932261145e-05, + "loss": 2.52764949798584, + "step": 50160 + }, + { + "epoch": 0.13659997277800462, + "grad_norm": 2.180997133255005, + "learning_rate": 1.7291590568352187e-05, + "loss": 2.5239192962646486, + "step": 50180 + }, + { + "epoch": 0.13665441676874915, + "grad_norm": 2.163322925567627, + "learning_rate": 1.7290500204443233e-05, + "loss": 2.513571548461914, + "step": 50200 + }, + { + "epoch": 0.13670886075949368, + "grad_norm": 2.132305860519409, + "learning_rate": 1.728940984053428e-05, + "loss": 2.547587203979492, + "step": 50220 + }, + { + "epoch": 0.13676330475023818, + "grad_norm": 2.0805490016937256, + "learning_rate": 1.7288319476625325e-05, + "loss": 2.547662544250488, + "step": 50240 + }, + { + "epoch": 0.1368177487409827, + "grad_norm": 2.1318485736846924, + "learning_rate": 1.728722911271637e-05, + "loss": 2.5179519653320312, + "step": 50260 + }, + { + "epoch": 0.13687219273172724, + "grad_norm": 2.192206621170044, + "learning_rate": 1.7286138748807416e-05, + "loss": 2.556733322143555, + "step": 50280 + }, + { + "epoch": 0.13692663672247177, + "grad_norm": 2.4238133430480957, + "learning_rate": 1.7285048384898462e-05, + "loss": 2.517709159851074, + "step": 50300 + }, + { + "epoch": 0.13698108071321627, + "grad_norm": 2.3475863933563232, + "learning_rate": 1.7283958020989508e-05, + "loss": 2.521158981323242, + "step": 50320 + }, + { + "epoch": 0.1370355247039608, + "grad_norm": 2.1466262340545654, + "learning_rate": 1.7282867657080554e-05, + "loss": 2.536040115356445, + "step": 50340 + }, + { + "epoch": 0.13708996869470533, + "grad_norm": 2.1214592456817627, + "learning_rate": 1.7281777293171596e-05, + "loss": 2.565372848510742, + "step": 50360 + }, + { + "epoch": 0.13714441268544986, + "grad_norm": 2.1990373134613037, + "learning_rate": 1.7280686929262642e-05, + "loss": 2.531056022644043, + "step": 50380 + }, + { + "epoch": 0.13719885667619436, + "grad_norm": 2.0334665775299072, + "learning_rate": 1.7279596565353688e-05, + "loss": 2.542320442199707, + "step": 50400 + }, + { + "epoch": 0.1372533006669389, + "grad_norm": 2.3206942081451416, + "learning_rate": 1.7278506201444734e-05, + "loss": 2.554275131225586, + "step": 50420 + }, + { + "epoch": 0.13730774465768342, + "grad_norm": 2.1209256649017334, + "learning_rate": 1.727741583753578e-05, + "loss": 2.5253618240356444, + "step": 50440 + }, + { + "epoch": 0.13736218864842792, + "grad_norm": 2.1242966651916504, + "learning_rate": 1.7276325473626825e-05, + "loss": 2.538258934020996, + "step": 50460 + }, + { + "epoch": 0.13741663263917245, + "grad_norm": 2.1357083320617676, + "learning_rate": 1.727523510971787e-05, + "loss": 2.519656753540039, + "step": 50480 + }, + { + "epoch": 0.13747107662991698, + "grad_norm": 2.08039927482605, + "learning_rate": 1.7274144745808917e-05, + "loss": 2.5394567489624023, + "step": 50500 + }, + { + "epoch": 0.1375255206206615, + "grad_norm": 2.305264949798584, + "learning_rate": 1.727305438189996e-05, + "loss": 2.5349199295043947, + "step": 50520 + }, + { + "epoch": 0.137579964611406, + "grad_norm": 2.1043953895568848, + "learning_rate": 1.7271964017991005e-05, + "loss": 2.498118209838867, + "step": 50540 + }, + { + "epoch": 0.13763440860215054, + "grad_norm": 2.1456191539764404, + "learning_rate": 1.727087365408205e-05, + "loss": 2.529876136779785, + "step": 50560 + }, + { + "epoch": 0.13768885259289507, + "grad_norm": 2.0732064247131348, + "learning_rate": 1.7269783290173097e-05, + "loss": 2.540274238586426, + "step": 50580 + }, + { + "epoch": 0.13774329658363957, + "grad_norm": 2.419914484024048, + "learning_rate": 1.7268692926264142e-05, + "loss": 2.544773292541504, + "step": 50600 + }, + { + "epoch": 0.1377977405743841, + "grad_norm": 2.269648551940918, + "learning_rate": 1.7267602562355188e-05, + "loss": 2.527823638916016, + "step": 50620 + }, + { + "epoch": 0.13785218456512863, + "grad_norm": 2.09873628616333, + "learning_rate": 1.7266512198446234e-05, + "loss": 2.507652473449707, + "step": 50640 + }, + { + "epoch": 0.13790662855587316, + "grad_norm": 2.1635210514068604, + "learning_rate": 1.726542183453728e-05, + "loss": 2.52802734375, + "step": 50660 + }, + { + "epoch": 0.13796107254661766, + "grad_norm": 2.1816465854644775, + "learning_rate": 1.7264331470628322e-05, + "loss": 2.522372245788574, + "step": 50680 + }, + { + "epoch": 0.1380155165373622, + "grad_norm": 2.0643818378448486, + "learning_rate": 1.7263241106719368e-05, + "loss": 2.5203332901000977, + "step": 50700 + }, + { + "epoch": 0.13806996052810672, + "grad_norm": 2.1428332328796387, + "learning_rate": 1.7262150742810414e-05, + "loss": 2.4945741653442384, + "step": 50720 + }, + { + "epoch": 0.13812440451885122, + "grad_norm": 2.2084286212921143, + "learning_rate": 1.726106037890146e-05, + "loss": 2.5365976333618163, + "step": 50740 + }, + { + "epoch": 0.13817884850959575, + "grad_norm": 2.1432998180389404, + "learning_rate": 1.7259970014992506e-05, + "loss": 2.529557228088379, + "step": 50760 + }, + { + "epoch": 0.13823329250034028, + "grad_norm": 2.548997402191162, + "learning_rate": 1.725887965108355e-05, + "loss": 2.505398750305176, + "step": 50780 + }, + { + "epoch": 0.1382877364910848, + "grad_norm": 2.0750513076782227, + "learning_rate": 1.7257789287174597e-05, + "loss": 2.569177436828613, + "step": 50800 + }, + { + "epoch": 0.1383421804818293, + "grad_norm": 2.1720540523529053, + "learning_rate": 1.7256698923265643e-05, + "loss": 2.507685661315918, + "step": 50820 + }, + { + "epoch": 0.13839662447257384, + "grad_norm": 2.0622525215148926, + "learning_rate": 1.7255608559356685e-05, + "loss": 2.5485345840454103, + "step": 50840 + }, + { + "epoch": 0.13845106846331837, + "grad_norm": 2.09230899810791, + "learning_rate": 1.725451819544773e-05, + "loss": 2.5302867889404297, + "step": 50860 + }, + { + "epoch": 0.1385055124540629, + "grad_norm": 3.5381832122802734, + "learning_rate": 1.7253427831538777e-05, + "loss": 2.5172002792358397, + "step": 50880 + }, + { + "epoch": 0.1385599564448074, + "grad_norm": 2.1979787349700928, + "learning_rate": 1.7252337467629823e-05, + "loss": 2.523316192626953, + "step": 50900 + }, + { + "epoch": 0.13861440043555193, + "grad_norm": 2.0557708740234375, + "learning_rate": 1.725124710372087e-05, + "loss": 2.551894187927246, + "step": 50920 + }, + { + "epoch": 0.13866884442629646, + "grad_norm": 2.3378798961639404, + "learning_rate": 1.7250156739811914e-05, + "loss": 2.53994140625, + "step": 50940 + }, + { + "epoch": 0.13872328841704096, + "grad_norm": 2.155561923980713, + "learning_rate": 1.724906637590296e-05, + "loss": 2.5289920806884765, + "step": 50960 + }, + { + "epoch": 0.1387777324077855, + "grad_norm": 2.9705657958984375, + "learning_rate": 1.7247976011994006e-05, + "loss": 2.5195960998535156, + "step": 50980 + }, + { + "epoch": 0.13883217639853002, + "grad_norm": 2.053584098815918, + "learning_rate": 1.724688564808505e-05, + "loss": 2.523356246948242, + "step": 51000 + }, + { + "epoch": 0.13888662038927455, + "grad_norm": 2.171839714050293, + "learning_rate": 1.7245795284176094e-05, + "loss": 2.5637075424194338, + "step": 51020 + }, + { + "epoch": 0.13894106438001905, + "grad_norm": 2.2134764194488525, + "learning_rate": 1.724470492026714e-05, + "loss": 2.521371841430664, + "step": 51040 + }, + { + "epoch": 0.13899550837076358, + "grad_norm": 2.14593505859375, + "learning_rate": 1.7243614556358186e-05, + "loss": 2.574921798706055, + "step": 51060 + }, + { + "epoch": 0.1390499523615081, + "grad_norm": 2.0923807621002197, + "learning_rate": 1.7242524192449232e-05, + "loss": 2.5207515716552735, + "step": 51080 + }, + { + "epoch": 0.1391043963522526, + "grad_norm": 2.4564085006713867, + "learning_rate": 1.7241433828540278e-05, + "loss": 2.5390268325805665, + "step": 51100 + }, + { + "epoch": 0.13915884034299714, + "grad_norm": 2.1450002193450928, + "learning_rate": 1.7240343464631323e-05, + "loss": 2.540023994445801, + "step": 51120 + }, + { + "epoch": 0.13921328433374167, + "grad_norm": 2.3393218517303467, + "learning_rate": 1.723925310072237e-05, + "loss": 2.5112226486206053, + "step": 51140 + }, + { + "epoch": 0.1392677283244862, + "grad_norm": 2.1695775985717773, + "learning_rate": 1.723816273681341e-05, + "loss": 2.5550653457641603, + "step": 51160 + }, + { + "epoch": 0.1393221723152307, + "grad_norm": 2.5540449619293213, + "learning_rate": 1.7237072372904457e-05, + "loss": 2.551301383972168, + "step": 51180 + }, + { + "epoch": 0.13937661630597523, + "grad_norm": 2.088944673538208, + "learning_rate": 1.7235982008995503e-05, + "loss": 2.547719383239746, + "step": 51200 + }, + { + "epoch": 0.13943106029671976, + "grad_norm": 2.035911798477173, + "learning_rate": 1.723489164508655e-05, + "loss": 2.568440628051758, + "step": 51220 + }, + { + "epoch": 0.13948550428746428, + "grad_norm": 2.1519875526428223, + "learning_rate": 1.7233801281177595e-05, + "loss": 2.505954933166504, + "step": 51240 + }, + { + "epoch": 0.1395399482782088, + "grad_norm": 2.0803070068359375, + "learning_rate": 1.723271091726864e-05, + "loss": 2.5047897338867187, + "step": 51260 + }, + { + "epoch": 0.13959439226895332, + "grad_norm": 2.270724296569824, + "learning_rate": 1.7231620553359686e-05, + "loss": 2.559063720703125, + "step": 51280 + }, + { + "epoch": 0.13964883625969784, + "grad_norm": 2.2199764251708984, + "learning_rate": 1.7230530189450732e-05, + "loss": 2.5321273803710938, + "step": 51300 + }, + { + "epoch": 0.13970328025044235, + "grad_norm": 2.0982306003570557, + "learning_rate": 1.7229439825541775e-05, + "loss": 2.543201446533203, + "step": 51320 + }, + { + "epoch": 0.13975772424118688, + "grad_norm": 2.0673129558563232, + "learning_rate": 1.722834946163282e-05, + "loss": 2.5288854598999024, + "step": 51340 + }, + { + "epoch": 0.1398121682319314, + "grad_norm": 2.1300227642059326, + "learning_rate": 1.7227259097723866e-05, + "loss": 2.5388816833496093, + "step": 51360 + }, + { + "epoch": 0.13986661222267593, + "grad_norm": 2.205765962600708, + "learning_rate": 1.7226168733814912e-05, + "loss": 2.524849510192871, + "step": 51380 + }, + { + "epoch": 0.13992105621342044, + "grad_norm": 2.1072704792022705, + "learning_rate": 1.7225078369905958e-05, + "loss": 2.5448320388793944, + "step": 51400 + }, + { + "epoch": 0.13997550020416497, + "grad_norm": 2.179988145828247, + "learning_rate": 1.7223988005997004e-05, + "loss": 2.5376996994018555, + "step": 51420 + }, + { + "epoch": 0.1400299441949095, + "grad_norm": 2.288043975830078, + "learning_rate": 1.722289764208805e-05, + "loss": 2.504971504211426, + "step": 51440 + }, + { + "epoch": 0.140084388185654, + "grad_norm": 2.2355077266693115, + "learning_rate": 1.7221807278179095e-05, + "loss": 2.52618293762207, + "step": 51460 + }, + { + "epoch": 0.14013883217639853, + "grad_norm": 2.2438032627105713, + "learning_rate": 1.7220716914270138e-05, + "loss": 2.539295196533203, + "step": 51480 + }, + { + "epoch": 0.14019327616714305, + "grad_norm": 2.1272711753845215, + "learning_rate": 1.7219626550361184e-05, + "loss": 2.546504020690918, + "step": 51500 + }, + { + "epoch": 0.14024772015788758, + "grad_norm": 2.080585479736328, + "learning_rate": 1.721853618645223e-05, + "loss": 2.5413936614990233, + "step": 51520 + }, + { + "epoch": 0.14030216414863209, + "grad_norm": 2.194355010986328, + "learning_rate": 1.7217445822543275e-05, + "loss": 2.5262889862060547, + "step": 51540 + }, + { + "epoch": 0.14035660813937662, + "grad_norm": 2.0922691822052, + "learning_rate": 1.721635545863432e-05, + "loss": 2.510223960876465, + "step": 51560 + }, + { + "epoch": 0.14041105213012114, + "grad_norm": 2.2004871368408203, + "learning_rate": 1.7215265094725367e-05, + "loss": 2.538664436340332, + "step": 51580 + }, + { + "epoch": 0.14046549612086565, + "grad_norm": 2.43632435798645, + "learning_rate": 1.721417473081641e-05, + "loss": 2.5223888397216796, + "step": 51600 + }, + { + "epoch": 0.14051994011161018, + "grad_norm": 2.048722505569458, + "learning_rate": 1.721308436690746e-05, + "loss": 2.541136932373047, + "step": 51620 + }, + { + "epoch": 0.1405743841023547, + "grad_norm": 2.1262004375457764, + "learning_rate": 1.72119940029985e-05, + "loss": 2.5331308364868166, + "step": 51640 + }, + { + "epoch": 0.14062882809309923, + "grad_norm": 2.122004985809326, + "learning_rate": 1.7210903639089547e-05, + "loss": 2.536396598815918, + "step": 51660 + }, + { + "epoch": 0.14068327208384374, + "grad_norm": 2.491405725479126, + "learning_rate": 1.7209813275180592e-05, + "loss": 2.5442344665527346, + "step": 51680 + }, + { + "epoch": 0.14073771607458826, + "grad_norm": 2.088239908218384, + "learning_rate": 1.7208722911271638e-05, + "loss": 2.5402767181396486, + "step": 51700 + }, + { + "epoch": 0.1407921600653328, + "grad_norm": 2.0614075660705566, + "learning_rate": 1.7207632547362684e-05, + "loss": 2.5376949310302734, + "step": 51720 + }, + { + "epoch": 0.14084660405607732, + "grad_norm": 2.1506640911102295, + "learning_rate": 1.720654218345373e-05, + "loss": 2.5432884216308596, + "step": 51740 + }, + { + "epoch": 0.14090104804682183, + "grad_norm": 2.5515196323394775, + "learning_rate": 1.7205451819544772e-05, + "loss": 2.5582912445068358, + "step": 51760 + }, + { + "epoch": 0.14095549203756635, + "grad_norm": 2.0973899364471436, + "learning_rate": 1.720436145563582e-05, + "loss": 2.5457250595092775, + "step": 51780 + }, + { + "epoch": 0.14100993602831088, + "grad_norm": 2.236727237701416, + "learning_rate": 1.7203271091726864e-05, + "loss": 2.5414569854736326, + "step": 51800 + }, + { + "epoch": 0.14106438001905539, + "grad_norm": 2.1529242992401123, + "learning_rate": 1.720218072781791e-05, + "loss": 2.504323196411133, + "step": 51820 + }, + { + "epoch": 0.14111882400979991, + "grad_norm": 2.7989628314971924, + "learning_rate": 1.7201090363908956e-05, + "loss": 2.4989389419555663, + "step": 51840 + }, + { + "epoch": 0.14117326800054444, + "grad_norm": 2.222750663757324, + "learning_rate": 1.72e-05, + "loss": 2.5606508255004883, + "step": 51860 + }, + { + "epoch": 0.14122771199128897, + "grad_norm": 2.253650188446045, + "learning_rate": 1.7198909636091047e-05, + "loss": 2.521994400024414, + "step": 51880 + }, + { + "epoch": 0.14128215598203347, + "grad_norm": 2.2926747798919678, + "learning_rate": 1.7197819272182093e-05, + "loss": 2.563151168823242, + "step": 51900 + }, + { + "epoch": 0.141336599972778, + "grad_norm": 2.1523358821868896, + "learning_rate": 1.7196728908273135e-05, + "loss": 2.5245994567871093, + "step": 51920 + }, + { + "epoch": 0.14139104396352253, + "grad_norm": 2.2325732707977295, + "learning_rate": 1.7195638544364185e-05, + "loss": 2.5461399078369142, + "step": 51940 + }, + { + "epoch": 0.14144548795426704, + "grad_norm": 2.2017476558685303, + "learning_rate": 1.7194548180455227e-05, + "loss": 2.5288925170898438, + "step": 51960 + }, + { + "epoch": 0.14149993194501156, + "grad_norm": 2.1549668312072754, + "learning_rate": 1.7193457816546273e-05, + "loss": 2.5270551681518554, + "step": 51980 + }, + { + "epoch": 0.1415543759357561, + "grad_norm": 2.095977306365967, + "learning_rate": 1.719236745263732e-05, + "loss": 2.54632625579834, + "step": 52000 + }, + { + "epoch": 0.14160881992650062, + "grad_norm": 2.0901010036468506, + "learning_rate": 1.7191277088728364e-05, + "loss": 2.5173526763916017, + "step": 52020 + }, + { + "epoch": 0.14166326391724512, + "grad_norm": 2.1280765533447266, + "learning_rate": 1.719018672481941e-05, + "loss": 2.545742988586426, + "step": 52040 + }, + { + "epoch": 0.14171770790798965, + "grad_norm": 2.1708524227142334, + "learning_rate": 1.7189096360910456e-05, + "loss": 2.52618350982666, + "step": 52060 + }, + { + "epoch": 0.14177215189873418, + "grad_norm": 2.1224536895751953, + "learning_rate": 1.71880059970015e-05, + "loss": 2.518889808654785, + "step": 52080 + }, + { + "epoch": 0.1418265958894787, + "grad_norm": 2.1829912662506104, + "learning_rate": 1.7186915633092548e-05, + "loss": 2.518947792053223, + "step": 52100 + }, + { + "epoch": 0.14188103988022321, + "grad_norm": 2.110416889190674, + "learning_rate": 1.718582526918359e-05, + "loss": 2.5547155380249023, + "step": 52120 + }, + { + "epoch": 0.14193548387096774, + "grad_norm": 2.0930495262145996, + "learning_rate": 1.7184734905274636e-05, + "loss": 2.5511053085327147, + "step": 52140 + }, + { + "epoch": 0.14198992786171227, + "grad_norm": 2.0334949493408203, + "learning_rate": 1.7183644541365682e-05, + "loss": 2.531600570678711, + "step": 52160 + }, + { + "epoch": 0.14204437185245677, + "grad_norm": 2.177152156829834, + "learning_rate": 1.7182554177456728e-05, + "loss": 2.543073844909668, + "step": 52180 + }, + { + "epoch": 0.1420988158432013, + "grad_norm": 2.0686347484588623, + "learning_rate": 1.7181463813547773e-05, + "loss": 2.555716323852539, + "step": 52200 + }, + { + "epoch": 0.14215325983394583, + "grad_norm": 2.2355451583862305, + "learning_rate": 1.718037344963882e-05, + "loss": 2.5146379470825195, + "step": 52220 + }, + { + "epoch": 0.14220770382469036, + "grad_norm": 2.2102274894714355, + "learning_rate": 1.717928308572986e-05, + "loss": 2.5493154525756836, + "step": 52240 + }, + { + "epoch": 0.14226214781543486, + "grad_norm": 2.1166913509368896, + "learning_rate": 1.717819272182091e-05, + "loss": 2.53809814453125, + "step": 52260 + }, + { + "epoch": 0.1423165918061794, + "grad_norm": 2.1529791355133057, + "learning_rate": 1.7177102357911953e-05, + "loss": 2.5416738510131838, + "step": 52280 + }, + { + "epoch": 0.14237103579692392, + "grad_norm": 2.228912830352783, + "learning_rate": 1.7176011994003e-05, + "loss": 2.524318313598633, + "step": 52300 + }, + { + "epoch": 0.14242547978766842, + "grad_norm": 2.182502031326294, + "learning_rate": 1.7174921630094045e-05, + "loss": 2.5451820373535154, + "step": 52320 + }, + { + "epoch": 0.14247992377841295, + "grad_norm": 2.1205737590789795, + "learning_rate": 1.717383126618509e-05, + "loss": 2.5198080062866213, + "step": 52340 + }, + { + "epoch": 0.14253436776915748, + "grad_norm": 2.1504945755004883, + "learning_rate": 1.7172740902276136e-05, + "loss": 2.540089797973633, + "step": 52360 + }, + { + "epoch": 0.142588811759902, + "grad_norm": 2.0722193717956543, + "learning_rate": 1.7171650538367182e-05, + "loss": 2.4960336685180664, + "step": 52380 + }, + { + "epoch": 0.1426432557506465, + "grad_norm": 2.073298692703247, + "learning_rate": 1.7170560174458225e-05, + "loss": 2.5760005950927733, + "step": 52400 + }, + { + "epoch": 0.14269769974139104, + "grad_norm": 2.2880594730377197, + "learning_rate": 1.7169469810549274e-05, + "loss": 2.5225141525268553, + "step": 52420 + }, + { + "epoch": 0.14275214373213557, + "grad_norm": 2.0788886547088623, + "learning_rate": 1.7168379446640316e-05, + "loss": 2.526365852355957, + "step": 52440 + }, + { + "epoch": 0.14280658772288007, + "grad_norm": 2.107604503631592, + "learning_rate": 1.7167289082731362e-05, + "loss": 2.5423715591430662, + "step": 52460 + }, + { + "epoch": 0.1428610317136246, + "grad_norm": 2.281339406967163, + "learning_rate": 1.7166198718822408e-05, + "loss": 2.5346017837524415, + "step": 52480 + }, + { + "epoch": 0.14291547570436913, + "grad_norm": 2.15657377243042, + "learning_rate": 1.7165108354913454e-05, + "loss": 2.5360509872436525, + "step": 52500 + }, + { + "epoch": 0.14296991969511366, + "grad_norm": 2.20355486869812, + "learning_rate": 1.71640179910045e-05, + "loss": 2.562473106384277, + "step": 52520 + }, + { + "epoch": 0.14302436368585816, + "grad_norm": 2.176337242126465, + "learning_rate": 1.7162927627095545e-05, + "loss": 2.5318138122558596, + "step": 52540 + }, + { + "epoch": 0.1430788076766027, + "grad_norm": 2.166616201400757, + "learning_rate": 1.7161837263186588e-05, + "loss": 2.557282066345215, + "step": 52560 + }, + { + "epoch": 0.14313325166734722, + "grad_norm": 2.5397896766662598, + "learning_rate": 1.7160746899277637e-05, + "loss": 2.5324583053588867, + "step": 52580 + }, + { + "epoch": 0.14318769565809175, + "grad_norm": 2.2343053817749023, + "learning_rate": 1.715965653536868e-05, + "loss": 2.505377769470215, + "step": 52600 + }, + { + "epoch": 0.14324213964883625, + "grad_norm": 2.045095443725586, + "learning_rate": 1.7158566171459725e-05, + "loss": 2.5245357513427735, + "step": 52620 + }, + { + "epoch": 0.14329658363958078, + "grad_norm": 2.260411262512207, + "learning_rate": 1.715747580755077e-05, + "loss": 2.541285514831543, + "step": 52640 + }, + { + "epoch": 0.1433510276303253, + "grad_norm": 2.10176944732666, + "learning_rate": 1.7156385443641817e-05, + "loss": 2.556019401550293, + "step": 52660 + }, + { + "epoch": 0.1434054716210698, + "grad_norm": 2.0691845417022705, + "learning_rate": 1.7155295079732863e-05, + "loss": 2.5597135543823244, + "step": 52680 + }, + { + "epoch": 0.14345991561181434, + "grad_norm": 2.209774971008301, + "learning_rate": 1.715420471582391e-05, + "loss": 2.5112852096557616, + "step": 52700 + }, + { + "epoch": 0.14351435960255887, + "grad_norm": 1.9981441497802734, + "learning_rate": 1.715311435191495e-05, + "loss": 2.5601234436035156, + "step": 52720 + }, + { + "epoch": 0.1435688035933034, + "grad_norm": 2.1284494400024414, + "learning_rate": 1.7152023988006e-05, + "loss": 2.562088394165039, + "step": 52740 + }, + { + "epoch": 0.1436232475840479, + "grad_norm": 2.0545854568481445, + "learning_rate": 1.7150933624097042e-05, + "loss": 2.5255752563476563, + "step": 52760 + }, + { + "epoch": 0.14367769157479243, + "grad_norm": 2.1509807109832764, + "learning_rate": 1.7149843260188088e-05, + "loss": 2.5626943588256834, + "step": 52780 + }, + { + "epoch": 0.14373213556553696, + "grad_norm": 2.1402955055236816, + "learning_rate": 1.7148752896279134e-05, + "loss": 2.542271614074707, + "step": 52800 + }, + { + "epoch": 0.14378657955628146, + "grad_norm": 2.098141670227051, + "learning_rate": 1.714766253237018e-05, + "loss": 2.5542449951171875, + "step": 52820 + }, + { + "epoch": 0.143841023547026, + "grad_norm": 2.51033616065979, + "learning_rate": 1.7146572168461226e-05, + "loss": 2.5782276153564454, + "step": 52840 + }, + { + "epoch": 0.14389546753777052, + "grad_norm": 2.1883461475372314, + "learning_rate": 1.714548180455227e-05, + "loss": 2.525111198425293, + "step": 52860 + }, + { + "epoch": 0.14394991152851505, + "grad_norm": 2.122313976287842, + "learning_rate": 1.7144391440643314e-05, + "loss": 2.5482545852661134, + "step": 52880 + }, + { + "epoch": 0.14400435551925955, + "grad_norm": 2.11201810836792, + "learning_rate": 1.7143301076734363e-05, + "loss": 2.5564897537231444, + "step": 52900 + }, + { + "epoch": 0.14405879951000408, + "grad_norm": 2.1038689613342285, + "learning_rate": 1.7142210712825406e-05, + "loss": 2.551981544494629, + "step": 52920 + }, + { + "epoch": 0.1441132435007486, + "grad_norm": 2.080970287322998, + "learning_rate": 1.714112034891645e-05, + "loss": 2.5497644424438475, + "step": 52940 + }, + { + "epoch": 0.1441676874914931, + "grad_norm": 2.2879576683044434, + "learning_rate": 1.7140029985007497e-05, + "loss": 2.556519317626953, + "step": 52960 + }, + { + "epoch": 0.14422213148223764, + "grad_norm": 2.2461912631988525, + "learning_rate": 1.7138939621098543e-05, + "loss": 2.549017906188965, + "step": 52980 + }, + { + "epoch": 0.14427657547298217, + "grad_norm": 2.253251552581787, + "learning_rate": 1.713784925718959e-05, + "loss": 2.5537084579467773, + "step": 53000 + }, + { + "epoch": 0.1443310194637267, + "grad_norm": 2.119279384613037, + "learning_rate": 1.7136758893280635e-05, + "loss": 2.5586980819702148, + "step": 53020 + }, + { + "epoch": 0.1443854634544712, + "grad_norm": 2.10097074508667, + "learning_rate": 1.7135668529371677e-05, + "loss": 2.5542455673217774, + "step": 53040 + }, + { + "epoch": 0.14443990744521573, + "grad_norm": 3.524263381958008, + "learning_rate": 1.7134578165462726e-05, + "loss": 2.520603370666504, + "step": 53060 + }, + { + "epoch": 0.14449435143596026, + "grad_norm": 2.1362545490264893, + "learning_rate": 1.713348780155377e-05, + "loss": 2.5669607162475585, + "step": 53080 + }, + { + "epoch": 0.1445487954267048, + "grad_norm": 2.0195770263671875, + "learning_rate": 1.7132397437644814e-05, + "loss": 2.552469253540039, + "step": 53100 + }, + { + "epoch": 0.1446032394174493, + "grad_norm": 2.1651086807250977, + "learning_rate": 1.713130707373586e-05, + "loss": 2.553905487060547, + "step": 53120 + }, + { + "epoch": 0.14465768340819382, + "grad_norm": 2.0581440925598145, + "learning_rate": 1.7130216709826906e-05, + "loss": 2.548396873474121, + "step": 53140 + }, + { + "epoch": 0.14471212739893835, + "grad_norm": 2.2902793884277344, + "learning_rate": 1.7129126345917952e-05, + "loss": 2.5393659591674806, + "step": 53160 + }, + { + "epoch": 0.14476657138968285, + "grad_norm": 2.083461284637451, + "learning_rate": 1.7128035982008998e-05, + "loss": 2.5239734649658203, + "step": 53180 + }, + { + "epoch": 0.14482101538042738, + "grad_norm": 2.116476058959961, + "learning_rate": 1.712694561810004e-05, + "loss": 2.525436592102051, + "step": 53200 + }, + { + "epoch": 0.1448754593711719, + "grad_norm": 2.201559066772461, + "learning_rate": 1.712585525419109e-05, + "loss": 2.543721389770508, + "step": 53220 + }, + { + "epoch": 0.14492990336191644, + "grad_norm": 2.181180715560913, + "learning_rate": 1.712476489028213e-05, + "loss": 2.5301536560058593, + "step": 53240 + }, + { + "epoch": 0.14498434735266094, + "grad_norm": 2.226595878601074, + "learning_rate": 1.7123674526373177e-05, + "loss": 2.521097183227539, + "step": 53260 + }, + { + "epoch": 0.14503879134340547, + "grad_norm": 2.1392364501953125, + "learning_rate": 1.7122584162464223e-05, + "loss": 2.5463319778442384, + "step": 53280 + }, + { + "epoch": 0.14509323533415, + "grad_norm": 2.1342220306396484, + "learning_rate": 1.712149379855527e-05, + "loss": 2.531490516662598, + "step": 53300 + }, + { + "epoch": 0.1451476793248945, + "grad_norm": 2.088207244873047, + "learning_rate": 1.7120403434646315e-05, + "loss": 2.541626739501953, + "step": 53320 + }, + { + "epoch": 0.14520212331563903, + "grad_norm": 2.0920300483703613, + "learning_rate": 1.711931307073736e-05, + "loss": 2.5390037536621093, + "step": 53340 + }, + { + "epoch": 0.14525656730638356, + "grad_norm": 2.1037447452545166, + "learning_rate": 1.7118222706828403e-05, + "loss": 2.5333578109741213, + "step": 53360 + }, + { + "epoch": 0.1453110112971281, + "grad_norm": 2.13249135017395, + "learning_rate": 1.7117132342919452e-05, + "loss": 2.5386085510253906, + "step": 53380 + }, + { + "epoch": 0.1453654552878726, + "grad_norm": 2.1232402324676514, + "learning_rate": 1.7116041979010495e-05, + "loss": 2.563886260986328, + "step": 53400 + }, + { + "epoch": 0.14541989927861712, + "grad_norm": 2.116853952407837, + "learning_rate": 1.711495161510154e-05, + "loss": 2.5472341537475587, + "step": 53420 + }, + { + "epoch": 0.14547434326936165, + "grad_norm": 2.1759207248687744, + "learning_rate": 1.7113861251192586e-05, + "loss": 2.5612926483154297, + "step": 53440 + }, + { + "epoch": 0.14552878726010618, + "grad_norm": 2.109377861022949, + "learning_rate": 1.7112770887283632e-05, + "loss": 2.5026548385620115, + "step": 53460 + }, + { + "epoch": 0.14558323125085068, + "grad_norm": 2.5035901069641113, + "learning_rate": 1.7111680523374678e-05, + "loss": 2.551644515991211, + "step": 53480 + }, + { + "epoch": 0.1456376752415952, + "grad_norm": 2.0899577140808105, + "learning_rate": 1.7110590159465724e-05, + "loss": 2.5226932525634767, + "step": 53500 + }, + { + "epoch": 0.14569211923233974, + "grad_norm": 2.9105920791625977, + "learning_rate": 1.7109499795556766e-05, + "loss": 2.556803512573242, + "step": 53520 + }, + { + "epoch": 0.14574656322308424, + "grad_norm": 2.567746162414551, + "learning_rate": 1.7108409431647815e-05, + "loss": 2.5285308837890623, + "step": 53540 + }, + { + "epoch": 0.14580100721382877, + "grad_norm": 2.126544952392578, + "learning_rate": 1.7107319067738858e-05, + "loss": 2.5296884536743165, + "step": 53560 + }, + { + "epoch": 0.1458554512045733, + "grad_norm": 2.188490629196167, + "learning_rate": 1.7106228703829904e-05, + "loss": 2.5215911865234375, + "step": 53580 + }, + { + "epoch": 0.14590989519531783, + "grad_norm": 2.1738181114196777, + "learning_rate": 1.710513833992095e-05, + "loss": 2.520572471618652, + "step": 53600 + }, + { + "epoch": 0.14596433918606233, + "grad_norm": 2.2657549381256104, + "learning_rate": 1.7104047976011995e-05, + "loss": 2.50801944732666, + "step": 53620 + }, + { + "epoch": 0.14601878317680686, + "grad_norm": 2.394702434539795, + "learning_rate": 1.710295761210304e-05, + "loss": 2.5309471130371093, + "step": 53640 + }, + { + "epoch": 0.1460732271675514, + "grad_norm": 2.082538366317749, + "learning_rate": 1.7101867248194087e-05, + "loss": 2.5241157531738283, + "step": 53660 + }, + { + "epoch": 0.1461276711582959, + "grad_norm": 2.165623903274536, + "learning_rate": 1.710077688428513e-05, + "loss": 2.5369935989379884, + "step": 53680 + }, + { + "epoch": 0.14618211514904042, + "grad_norm": 2.1232032775878906, + "learning_rate": 1.709968652037618e-05, + "loss": 2.5125495910644533, + "step": 53700 + }, + { + "epoch": 0.14623655913978495, + "grad_norm": 2.152716636657715, + "learning_rate": 1.709859615646722e-05, + "loss": 2.5311023712158205, + "step": 53720 + }, + { + "epoch": 0.14629100313052948, + "grad_norm": 2.2040932178497314, + "learning_rate": 1.709750579255827e-05, + "loss": 2.5026422500610352, + "step": 53740 + }, + { + "epoch": 0.14634544712127398, + "grad_norm": 2.0828821659088135, + "learning_rate": 1.7096415428649313e-05, + "loss": 2.5322597503662108, + "step": 53760 + }, + { + "epoch": 0.1463998911120185, + "grad_norm": 2.191394090652466, + "learning_rate": 1.709532506474036e-05, + "loss": 2.566572380065918, + "step": 53780 + }, + { + "epoch": 0.14645433510276304, + "grad_norm": 2.1162736415863037, + "learning_rate": 1.7094234700831404e-05, + "loss": 2.545908546447754, + "step": 53800 + }, + { + "epoch": 0.14650877909350754, + "grad_norm": 2.1256120204925537, + "learning_rate": 1.709314433692245e-05, + "loss": 2.520437240600586, + "step": 53820 + }, + { + "epoch": 0.14656322308425207, + "grad_norm": 2.260012626647949, + "learning_rate": 1.7092053973013492e-05, + "loss": 2.5532253265380858, + "step": 53840 + }, + { + "epoch": 0.1466176670749966, + "grad_norm": 2.1104671955108643, + "learning_rate": 1.709096360910454e-05, + "loss": 2.530955505371094, + "step": 53860 + }, + { + "epoch": 0.14667211106574113, + "grad_norm": 2.6683733463287354, + "learning_rate": 1.7089873245195584e-05, + "loss": 2.5501453399658205, + "step": 53880 + }, + { + "epoch": 0.14672655505648563, + "grad_norm": 2.420952320098877, + "learning_rate": 1.7088782881286633e-05, + "loss": 2.5278186798095703, + "step": 53900 + }, + { + "epoch": 0.14678099904723016, + "grad_norm": 2.1027748584747314, + "learning_rate": 1.7087692517377676e-05, + "loss": 2.5382946014404295, + "step": 53920 + }, + { + "epoch": 0.1468354430379747, + "grad_norm": 2.2117021083831787, + "learning_rate": 1.708660215346872e-05, + "loss": 2.503372573852539, + "step": 53940 + }, + { + "epoch": 0.14688988702871922, + "grad_norm": 2.064241886138916, + "learning_rate": 1.7085511789559767e-05, + "loss": 2.538915824890137, + "step": 53960 + }, + { + "epoch": 0.14694433101946372, + "grad_norm": 3.4098618030548096, + "learning_rate": 1.7084421425650813e-05, + "loss": 2.5081247329711913, + "step": 53980 + }, + { + "epoch": 0.14699877501020825, + "grad_norm": 2.145413875579834, + "learning_rate": 1.7083331061741855e-05, + "loss": 2.5436573028564453, + "step": 54000 + }, + { + "epoch": 0.14705321900095278, + "grad_norm": 2.069911479949951, + "learning_rate": 1.7082240697832905e-05, + "loss": 2.5175771713256836, + "step": 54020 + }, + { + "epoch": 0.14710766299169728, + "grad_norm": 2.0883262157440186, + "learning_rate": 1.7081150333923947e-05, + "loss": 2.555440902709961, + "step": 54040 + }, + { + "epoch": 0.1471621069824418, + "grad_norm": 2.235701322555542, + "learning_rate": 1.7080059970014996e-05, + "loss": 2.559895706176758, + "step": 54060 + }, + { + "epoch": 0.14721655097318634, + "grad_norm": 2.0829014778137207, + "learning_rate": 1.707896960610604e-05, + "loss": 2.5479679107666016, + "step": 54080 + }, + { + "epoch": 0.14727099496393087, + "grad_norm": 2.10945987701416, + "learning_rate": 1.7077879242197085e-05, + "loss": 2.5421493530273436, + "step": 54100 + }, + { + "epoch": 0.14732543895467537, + "grad_norm": 2.3588788509368896, + "learning_rate": 1.707678887828813e-05, + "loss": 2.537469482421875, + "step": 54120 + }, + { + "epoch": 0.1473798829454199, + "grad_norm": 2.1536011695861816, + "learning_rate": 1.7075698514379176e-05, + "loss": 2.5266653060913087, + "step": 54140 + }, + { + "epoch": 0.14743432693616443, + "grad_norm": 2.1070168018341064, + "learning_rate": 1.707460815047022e-05, + "loss": 2.553432846069336, + "step": 54160 + }, + { + "epoch": 0.14748877092690893, + "grad_norm": 2.337131977081299, + "learning_rate": 1.7073517786561268e-05, + "loss": 2.5568702697753904, + "step": 54180 + }, + { + "epoch": 0.14754321491765346, + "grad_norm": 2.086080551147461, + "learning_rate": 1.707242742265231e-05, + "loss": 2.548203468322754, + "step": 54200 + }, + { + "epoch": 0.147597658908398, + "grad_norm": 2.2504494190216064, + "learning_rate": 1.707133705874336e-05, + "loss": 2.538418006896973, + "step": 54220 + }, + { + "epoch": 0.14765210289914252, + "grad_norm": 2.237316608428955, + "learning_rate": 1.7070246694834402e-05, + "loss": 2.5111614227294923, + "step": 54240 + }, + { + "epoch": 0.14770654688988702, + "grad_norm": 2.4956367015838623, + "learning_rate": 1.7069156330925448e-05, + "loss": 2.533810043334961, + "step": 54260 + }, + { + "epoch": 0.14776099088063155, + "grad_norm": 2.28088116645813, + "learning_rate": 1.7068065967016493e-05, + "loss": 2.5543596267700197, + "step": 54280 + }, + { + "epoch": 0.14781543487137608, + "grad_norm": 2.209055185317993, + "learning_rate": 1.706697560310754e-05, + "loss": 2.550743293762207, + "step": 54300 + }, + { + "epoch": 0.1478698788621206, + "grad_norm": 2.109701633453369, + "learning_rate": 1.706588523919858e-05, + "loss": 2.50576114654541, + "step": 54320 + }, + { + "epoch": 0.1479243228528651, + "grad_norm": 2.1766514778137207, + "learning_rate": 1.706479487528963e-05, + "loss": 2.5558908462524412, + "step": 54340 + }, + { + "epoch": 0.14797876684360964, + "grad_norm": 2.204470157623291, + "learning_rate": 1.7063704511380673e-05, + "loss": 2.524287986755371, + "step": 54360 + }, + { + "epoch": 0.14803321083435417, + "grad_norm": 2.322239398956299, + "learning_rate": 1.7062614147471722e-05, + "loss": 2.5464534759521484, + "step": 54380 + }, + { + "epoch": 0.14808765482509867, + "grad_norm": 2.1538054943084717, + "learning_rate": 1.7061523783562765e-05, + "loss": 2.5193244934082033, + "step": 54400 + }, + { + "epoch": 0.1481420988158432, + "grad_norm": 2.1265323162078857, + "learning_rate": 1.706043341965381e-05, + "loss": 2.5402027130126954, + "step": 54420 + }, + { + "epoch": 0.14819654280658773, + "grad_norm": 2.196110486984253, + "learning_rate": 1.7059343055744857e-05, + "loss": 2.521136474609375, + "step": 54440 + }, + { + "epoch": 0.14825098679733226, + "grad_norm": 2.0357847213745117, + "learning_rate": 1.7058252691835902e-05, + "loss": 2.53887882232666, + "step": 54460 + }, + { + "epoch": 0.14830543078807676, + "grad_norm": 2.180574655532837, + "learning_rate": 1.7057162327926945e-05, + "loss": 2.536471366882324, + "step": 54480 + }, + { + "epoch": 0.1483598747788213, + "grad_norm": 2.1204001903533936, + "learning_rate": 1.7056071964017994e-05, + "loss": 2.5093368530273437, + "step": 54500 + }, + { + "epoch": 0.14841431876956582, + "grad_norm": 2.1322574615478516, + "learning_rate": 1.7054981600109036e-05, + "loss": 2.5436546325683596, + "step": 54520 + }, + { + "epoch": 0.14846876276031032, + "grad_norm": 2.6461288928985596, + "learning_rate": 1.7053891236200086e-05, + "loss": 2.5497262954711912, + "step": 54540 + }, + { + "epoch": 0.14852320675105485, + "grad_norm": 2.1038994789123535, + "learning_rate": 1.7052800872291128e-05, + "loss": 2.5282039642333984, + "step": 54560 + }, + { + "epoch": 0.14857765074179938, + "grad_norm": 2.3503007888793945, + "learning_rate": 1.7051710508382174e-05, + "loss": 2.560028076171875, + "step": 54580 + }, + { + "epoch": 0.1486320947325439, + "grad_norm": 2.085221767425537, + "learning_rate": 1.705062014447322e-05, + "loss": 2.570255088806152, + "step": 54600 + }, + { + "epoch": 0.1486865387232884, + "grad_norm": 2.0600180625915527, + "learning_rate": 1.7049529780564265e-05, + "loss": 2.528207778930664, + "step": 54620 + }, + { + "epoch": 0.14874098271403294, + "grad_norm": 2.4003491401672363, + "learning_rate": 1.7048439416655308e-05, + "loss": 2.539195251464844, + "step": 54640 + }, + { + "epoch": 0.14879542670477747, + "grad_norm": 2.1335010528564453, + "learning_rate": 1.7047349052746357e-05, + "loss": 2.540371322631836, + "step": 54660 + }, + { + "epoch": 0.14884987069552197, + "grad_norm": 2.170994281768799, + "learning_rate": 1.70462586888374e-05, + "loss": 2.5194980621337892, + "step": 54680 + }, + { + "epoch": 0.1489043146862665, + "grad_norm": 2.055429220199585, + "learning_rate": 1.704516832492845e-05, + "loss": 2.540509796142578, + "step": 54700 + }, + { + "epoch": 0.14895875867701103, + "grad_norm": 2.1303820610046387, + "learning_rate": 1.704407796101949e-05, + "loss": 2.578803825378418, + "step": 54720 + }, + { + "epoch": 0.14901320266775556, + "grad_norm": 2.1728808879852295, + "learning_rate": 1.7042987597110537e-05, + "loss": 2.553484344482422, + "step": 54740 + }, + { + "epoch": 0.14906764665850006, + "grad_norm": 2.0854904651641846, + "learning_rate": 1.7041897233201583e-05, + "loss": 2.557915687561035, + "step": 54760 + }, + { + "epoch": 0.1491220906492446, + "grad_norm": 2.0174806118011475, + "learning_rate": 1.704080686929263e-05, + "loss": 2.5422231674194338, + "step": 54780 + }, + { + "epoch": 0.14917653463998912, + "grad_norm": 2.1004745960235596, + "learning_rate": 1.703971650538367e-05, + "loss": 2.5422975540161135, + "step": 54800 + }, + { + "epoch": 0.14923097863073365, + "grad_norm": 2.1277196407318115, + "learning_rate": 1.703862614147472e-05, + "loss": 2.520868682861328, + "step": 54820 + }, + { + "epoch": 0.14928542262147815, + "grad_norm": 2.200521469116211, + "learning_rate": 1.7037535777565763e-05, + "loss": 2.5409923553466798, + "step": 54840 + }, + { + "epoch": 0.14933986661222268, + "grad_norm": 2.1865944862365723, + "learning_rate": 1.7036445413656812e-05, + "loss": 2.5667787551879884, + "step": 54860 + }, + { + "epoch": 0.1493943106029672, + "grad_norm": 2.1276233196258545, + "learning_rate": 1.7035355049747854e-05, + "loss": 2.5286256790161135, + "step": 54880 + }, + { + "epoch": 0.1494487545937117, + "grad_norm": 2.193319082260132, + "learning_rate": 1.70342646858389e-05, + "loss": 2.523651123046875, + "step": 54900 + }, + { + "epoch": 0.14950319858445624, + "grad_norm": 2.1048097610473633, + "learning_rate": 1.7033174321929946e-05, + "loss": 2.5231756210327148, + "step": 54920 + }, + { + "epoch": 0.14955764257520077, + "grad_norm": 2.1642541885375977, + "learning_rate": 1.703208395802099e-05, + "loss": 2.5360090255737306, + "step": 54940 + }, + { + "epoch": 0.1496120865659453, + "grad_norm": 2.1340436935424805, + "learning_rate": 1.7030993594112034e-05, + "loss": 2.5385257720947267, + "step": 54960 + }, + { + "epoch": 0.1496665305566898, + "grad_norm": 2.090801239013672, + "learning_rate": 1.7029903230203083e-05, + "loss": 2.5044553756713865, + "step": 54980 + }, + { + "epoch": 0.14972097454743433, + "grad_norm": 2.108731508255005, + "learning_rate": 1.7028812866294126e-05, + "loss": 2.5298690795898438, + "step": 55000 + }, + { + "epoch": 0.14977541853817886, + "grad_norm": 2.068432331085205, + "learning_rate": 1.7027722502385175e-05, + "loss": 2.5325128555297853, + "step": 55020 + }, + { + "epoch": 0.14982986252892336, + "grad_norm": 2.206321954727173, + "learning_rate": 1.7026632138476217e-05, + "loss": 2.553960609436035, + "step": 55040 + }, + { + "epoch": 0.1498843065196679, + "grad_norm": 2.081554651260376, + "learning_rate": 1.7025541774567263e-05, + "loss": 2.5255455017089843, + "step": 55060 + }, + { + "epoch": 0.14993875051041242, + "grad_norm": 2.1027414798736572, + "learning_rate": 1.702445141065831e-05, + "loss": 2.5446645736694338, + "step": 55080 + }, + { + "epoch": 0.14999319450115695, + "grad_norm": 2.3112008571624756, + "learning_rate": 1.7023361046749355e-05, + "loss": 2.5107892990112304, + "step": 55100 + }, + { + "epoch": 0.15004763849190145, + "grad_norm": 2.2782983779907227, + "learning_rate": 1.7022270682840397e-05, + "loss": 2.5375297546386717, + "step": 55120 + }, + { + "epoch": 0.15010208248264598, + "grad_norm": 2.12104868888855, + "learning_rate": 1.7021180318931446e-05, + "loss": 2.515821647644043, + "step": 55140 + }, + { + "epoch": 0.1501565264733905, + "grad_norm": 2.106574535369873, + "learning_rate": 1.702008995502249e-05, + "loss": 2.5463693618774412, + "step": 55160 + }, + { + "epoch": 0.150210970464135, + "grad_norm": 2.3738811016082764, + "learning_rate": 1.7018999591113538e-05, + "loss": 2.529345703125, + "step": 55180 + }, + { + "epoch": 0.15026541445487954, + "grad_norm": 2.1271555423736572, + "learning_rate": 1.701790922720458e-05, + "loss": 2.538297653198242, + "step": 55200 + }, + { + "epoch": 0.15031985844562407, + "grad_norm": 2.0765750408172607, + "learning_rate": 1.7016818863295626e-05, + "loss": 2.549785614013672, + "step": 55220 + }, + { + "epoch": 0.1503743024363686, + "grad_norm": 2.1458165645599365, + "learning_rate": 1.7015728499386672e-05, + "loss": 2.525349426269531, + "step": 55240 + }, + { + "epoch": 0.1504287464271131, + "grad_norm": 2.0815389156341553, + "learning_rate": 1.7014638135477718e-05, + "loss": 2.5293169021606445, + "step": 55260 + }, + { + "epoch": 0.15048319041785763, + "grad_norm": 2.0464823246002197, + "learning_rate": 1.701354777156876e-05, + "loss": 2.5126495361328125, + "step": 55280 + }, + { + "epoch": 0.15053763440860216, + "grad_norm": 2.126920700073242, + "learning_rate": 1.701245740765981e-05, + "loss": 2.5351560592651365, + "step": 55300 + }, + { + "epoch": 0.15059207839934668, + "grad_norm": 2.3514866828918457, + "learning_rate": 1.7011367043750852e-05, + "loss": 2.5594789505004885, + "step": 55320 + }, + { + "epoch": 0.1506465223900912, + "grad_norm": 2.191021203994751, + "learning_rate": 1.70102766798419e-05, + "loss": 2.5349855422973633, + "step": 55340 + }, + { + "epoch": 0.15070096638083572, + "grad_norm": 2.0561940670013428, + "learning_rate": 1.7009186315932943e-05, + "loss": 2.5311283111572265, + "step": 55360 + }, + { + "epoch": 0.15075541037158025, + "grad_norm": 2.2588109970092773, + "learning_rate": 1.700809595202399e-05, + "loss": 2.5176605224609374, + "step": 55380 + }, + { + "epoch": 0.15080985436232475, + "grad_norm": 2.1366732120513916, + "learning_rate": 1.7007005588115035e-05, + "loss": 2.540685844421387, + "step": 55400 + }, + { + "epoch": 0.15086429835306928, + "grad_norm": 2.152092695236206, + "learning_rate": 1.700591522420608e-05, + "loss": 2.5213638305664063, + "step": 55420 + }, + { + "epoch": 0.1509187423438138, + "grad_norm": 2.131826639175415, + "learning_rate": 1.7004824860297127e-05, + "loss": 2.560260009765625, + "step": 55440 + }, + { + "epoch": 0.15097318633455833, + "grad_norm": 2.080052137374878, + "learning_rate": 1.7003734496388172e-05, + "loss": 2.545364189147949, + "step": 55460 + }, + { + "epoch": 0.15102763032530284, + "grad_norm": 2.193819761276245, + "learning_rate": 1.7002644132479215e-05, + "loss": 2.5546722412109375, + "step": 55480 + }, + { + "epoch": 0.15108207431604737, + "grad_norm": 2.051506519317627, + "learning_rate": 1.7001553768570264e-05, + "loss": 2.53033390045166, + "step": 55500 + }, + { + "epoch": 0.1511365183067919, + "grad_norm": 2.309523820877075, + "learning_rate": 1.7000463404661306e-05, + "loss": 2.508034515380859, + "step": 55520 + }, + { + "epoch": 0.1511909622975364, + "grad_norm": 2.1556568145751953, + "learning_rate": 1.6999373040752352e-05, + "loss": 2.5360851287841797, + "step": 55540 + }, + { + "epoch": 0.15124540628828093, + "grad_norm": 2.207348108291626, + "learning_rate": 1.6998282676843398e-05, + "loss": 2.5379049301147463, + "step": 55560 + }, + { + "epoch": 0.15129985027902546, + "grad_norm": 2.209583282470703, + "learning_rate": 1.6997192312934444e-05, + "loss": 2.5377985000610352, + "step": 55580 + }, + { + "epoch": 0.15135429426976998, + "grad_norm": 2.103703260421753, + "learning_rate": 1.699610194902549e-05, + "loss": 2.520722198486328, + "step": 55600 + }, + { + "epoch": 0.15140873826051449, + "grad_norm": 2.2732203006744385, + "learning_rate": 1.6995011585116536e-05, + "loss": 2.510940170288086, + "step": 55620 + }, + { + "epoch": 0.15146318225125902, + "grad_norm": 2.1051714420318604, + "learning_rate": 1.6993921221207578e-05, + "loss": 2.5546554565429687, + "step": 55640 + }, + { + "epoch": 0.15151762624200354, + "grad_norm": 2.125627279281616, + "learning_rate": 1.6992830857298627e-05, + "loss": 2.530165672302246, + "step": 55660 + }, + { + "epoch": 0.15157207023274807, + "grad_norm": 2.2206079959869385, + "learning_rate": 1.699174049338967e-05, + "loss": 2.5324682235717773, + "step": 55680 + }, + { + "epoch": 0.15162651422349258, + "grad_norm": 2.204050064086914, + "learning_rate": 1.6990650129480715e-05, + "loss": 2.5564918518066406, + "step": 55700 + }, + { + "epoch": 0.1516809582142371, + "grad_norm": 2.2783126831054688, + "learning_rate": 1.698955976557176e-05, + "loss": 2.510853576660156, + "step": 55720 + }, + { + "epoch": 0.15173540220498163, + "grad_norm": 2.1016294956207275, + "learning_rate": 1.6988469401662807e-05, + "loss": 2.488297462463379, + "step": 55740 + }, + { + "epoch": 0.15178984619572614, + "grad_norm": 2.1961510181427, + "learning_rate": 1.6987379037753853e-05, + "loss": 2.561857986450195, + "step": 55760 + }, + { + "epoch": 0.15184429018647067, + "grad_norm": 2.2981739044189453, + "learning_rate": 1.69862886738449e-05, + "loss": 2.541847801208496, + "step": 55780 + }, + { + "epoch": 0.1518987341772152, + "grad_norm": 3.527266263961792, + "learning_rate": 1.698519830993594e-05, + "loss": 2.5450443267822265, + "step": 55800 + }, + { + "epoch": 0.15195317816795972, + "grad_norm": 2.4038498401641846, + "learning_rate": 1.698410794602699e-05, + "loss": 2.5406162261962892, + "step": 55820 + }, + { + "epoch": 0.15200762215870423, + "grad_norm": 2.3482227325439453, + "learning_rate": 1.6983017582118033e-05, + "loss": 2.525999069213867, + "step": 55840 + }, + { + "epoch": 0.15206206614944875, + "grad_norm": 2.076782703399658, + "learning_rate": 1.698192721820908e-05, + "loss": 2.5128313064575196, + "step": 55860 + }, + { + "epoch": 0.15211651014019328, + "grad_norm": 2.159169912338257, + "learning_rate": 1.6980836854300124e-05, + "loss": 2.519985389709473, + "step": 55880 + }, + { + "epoch": 0.15217095413093779, + "grad_norm": 2.2910561561584473, + "learning_rate": 1.697974649039117e-05, + "loss": 2.561654281616211, + "step": 55900 + }, + { + "epoch": 0.15222539812168231, + "grad_norm": 2.1145639419555664, + "learning_rate": 1.6978656126482216e-05, + "loss": 2.552657890319824, + "step": 55920 + }, + { + "epoch": 0.15227984211242684, + "grad_norm": 2.0370383262634277, + "learning_rate": 1.6977565762573262e-05, + "loss": 2.532182502746582, + "step": 55940 + }, + { + "epoch": 0.15233428610317137, + "grad_norm": 2.214569330215454, + "learning_rate": 1.6976475398664304e-05, + "loss": 2.5573314666748046, + "step": 55960 + }, + { + "epoch": 0.15238873009391588, + "grad_norm": 2.1033618450164795, + "learning_rate": 1.6975385034755353e-05, + "loss": 2.5247165679931642, + "step": 55980 + }, + { + "epoch": 0.1524431740846604, + "grad_norm": 2.365717649459839, + "learning_rate": 1.6974294670846396e-05, + "loss": 2.5472970962524415, + "step": 56000 + }, + { + "epoch": 0.15249761807540493, + "grad_norm": 2.216967821121216, + "learning_rate": 1.697320430693744e-05, + "loss": 2.537884330749512, + "step": 56020 + }, + { + "epoch": 0.15255206206614944, + "grad_norm": 2.2918524742126465, + "learning_rate": 1.6972113943028487e-05, + "loss": 2.5130977630615234, + "step": 56040 + }, + { + "epoch": 0.15260650605689396, + "grad_norm": 2.141832113265991, + "learning_rate": 1.6971023579119533e-05, + "loss": 2.5485345840454103, + "step": 56060 + }, + { + "epoch": 0.1526609500476385, + "grad_norm": 2.6729681491851807, + "learning_rate": 1.696993321521058e-05, + "loss": 2.537478446960449, + "step": 56080 + }, + { + "epoch": 0.15271539403838302, + "grad_norm": 2.1741831302642822, + "learning_rate": 1.6968842851301625e-05, + "loss": 2.5489337921142576, + "step": 56100 + }, + { + "epoch": 0.15276983802912752, + "grad_norm": 2.136209726333618, + "learning_rate": 1.6967752487392667e-05, + "loss": 2.5460710525512695, + "step": 56120 + }, + { + "epoch": 0.15282428201987205, + "grad_norm": 2.1047146320343018, + "learning_rate": 1.6966662123483716e-05, + "loss": 2.546969413757324, + "step": 56140 + }, + { + "epoch": 0.15287872601061658, + "grad_norm": 2.118664264678955, + "learning_rate": 1.696557175957476e-05, + "loss": 2.529960060119629, + "step": 56160 + }, + { + "epoch": 0.1529331700013611, + "grad_norm": 2.171006202697754, + "learning_rate": 1.6964481395665805e-05, + "loss": 2.540279006958008, + "step": 56180 + }, + { + "epoch": 0.15298761399210561, + "grad_norm": 2.083918571472168, + "learning_rate": 1.696339103175685e-05, + "loss": 2.5407268524169924, + "step": 56200 + }, + { + "epoch": 0.15304205798285014, + "grad_norm": 2.1325623989105225, + "learning_rate": 1.6962300667847896e-05, + "loss": 2.5277950286865236, + "step": 56220 + }, + { + "epoch": 0.15309650197359467, + "grad_norm": 2.0525691509246826, + "learning_rate": 1.6961210303938942e-05, + "loss": 2.526158905029297, + "step": 56240 + }, + { + "epoch": 0.15315094596433917, + "grad_norm": 2.0732264518737793, + "learning_rate": 1.6960119940029988e-05, + "loss": 2.480418395996094, + "step": 56260 + }, + { + "epoch": 0.1532053899550837, + "grad_norm": 2.252467393875122, + "learning_rate": 1.695902957612103e-05, + "loss": 2.529491424560547, + "step": 56280 + }, + { + "epoch": 0.15325983394582823, + "grad_norm": 2.1089084148406982, + "learning_rate": 1.695793921221208e-05, + "loss": 2.5128189086914063, + "step": 56300 + }, + { + "epoch": 0.15331427793657276, + "grad_norm": 2.29122257232666, + "learning_rate": 1.6956848848303122e-05, + "loss": 2.547393226623535, + "step": 56320 + }, + { + "epoch": 0.15336872192731726, + "grad_norm": 2.244213581085205, + "learning_rate": 1.6955758484394168e-05, + "loss": 2.508877754211426, + "step": 56340 + }, + { + "epoch": 0.1534231659180618, + "grad_norm": 2.1763691902160645, + "learning_rate": 1.6954668120485214e-05, + "loss": 2.5306865692138674, + "step": 56360 + }, + { + "epoch": 0.15347760990880632, + "grad_norm": 2.885934352874756, + "learning_rate": 1.695357775657626e-05, + "loss": 2.545304870605469, + "step": 56380 + }, + { + "epoch": 0.15353205389955082, + "grad_norm": 2.1058449745178223, + "learning_rate": 1.6952487392667305e-05, + "loss": 2.5190151214599608, + "step": 56400 + }, + { + "epoch": 0.15358649789029535, + "grad_norm": 2.169600248336792, + "learning_rate": 1.695139702875835e-05, + "loss": 2.552135467529297, + "step": 56420 + }, + { + "epoch": 0.15364094188103988, + "grad_norm": 2.249969959259033, + "learning_rate": 1.6950306664849393e-05, + "loss": 2.5441051483154298, + "step": 56440 + }, + { + "epoch": 0.1536953858717844, + "grad_norm": 2.1780662536621094, + "learning_rate": 1.6949216300940443e-05, + "loss": 2.5505624771118165, + "step": 56460 + }, + { + "epoch": 0.15374982986252891, + "grad_norm": 2.128659248352051, + "learning_rate": 1.6948125937031485e-05, + "loss": 2.5418153762817384, + "step": 56480 + }, + { + "epoch": 0.15380427385327344, + "grad_norm": 2.3042991161346436, + "learning_rate": 1.694703557312253e-05, + "loss": 2.5653968811035157, + "step": 56500 + }, + { + "epoch": 0.15385871784401797, + "grad_norm": 2.0705771446228027, + "learning_rate": 1.6945945209213577e-05, + "loss": 2.539318084716797, + "step": 56520 + }, + { + "epoch": 0.1539131618347625, + "grad_norm": 2.009474039077759, + "learning_rate": 1.6944854845304622e-05, + "loss": 2.5468311309814453, + "step": 56540 + }, + { + "epoch": 0.153967605825507, + "grad_norm": 2.208528518676758, + "learning_rate": 1.6943764481395668e-05, + "loss": 2.564695358276367, + "step": 56560 + }, + { + "epoch": 0.15402204981625153, + "grad_norm": 2.4542524814605713, + "learning_rate": 1.6942674117486714e-05, + "loss": 2.525826644897461, + "step": 56580 + }, + { + "epoch": 0.15407649380699606, + "grad_norm": 2.043525218963623, + "learning_rate": 1.6941583753577756e-05, + "loss": 2.523974609375, + "step": 56600 + }, + { + "epoch": 0.15413093779774056, + "grad_norm": 2.293459177017212, + "learning_rate": 1.6940493389668802e-05, + "loss": 2.5107940673828124, + "step": 56620 + }, + { + "epoch": 0.1541853817884851, + "grad_norm": 3.4160337448120117, + "learning_rate": 1.6939403025759848e-05, + "loss": 2.5295557022094726, + "step": 56640 + }, + { + "epoch": 0.15423982577922962, + "grad_norm": 2.277552366256714, + "learning_rate": 1.6938312661850894e-05, + "loss": 2.552493858337402, + "step": 56660 + }, + { + "epoch": 0.15429426976997415, + "grad_norm": 2.2470543384552, + "learning_rate": 1.693722229794194e-05, + "loss": 2.5729228973388674, + "step": 56680 + }, + { + "epoch": 0.15434871376071865, + "grad_norm": 2.1494028568267822, + "learning_rate": 1.6936131934032986e-05, + "loss": 2.558858108520508, + "step": 56700 + }, + { + "epoch": 0.15440315775146318, + "grad_norm": 2.181889772415161, + "learning_rate": 1.693504157012403e-05, + "loss": 2.528067398071289, + "step": 56720 + }, + { + "epoch": 0.1544576017422077, + "grad_norm": 3.9364137649536133, + "learning_rate": 1.6933951206215077e-05, + "loss": 2.509506416320801, + "step": 56740 + }, + { + "epoch": 0.1545120457329522, + "grad_norm": 2.2344977855682373, + "learning_rate": 1.693286084230612e-05, + "loss": 2.521072196960449, + "step": 56760 + }, + { + "epoch": 0.15456648972369674, + "grad_norm": 2.0837433338165283, + "learning_rate": 1.6931770478397165e-05, + "loss": 2.5405904769897463, + "step": 56780 + }, + { + "epoch": 0.15462093371444127, + "grad_norm": 2.192119836807251, + "learning_rate": 1.693068011448821e-05, + "loss": 2.516325759887695, + "step": 56800 + }, + { + "epoch": 0.1546753777051858, + "grad_norm": 2.151730537414551, + "learning_rate": 1.6929589750579257e-05, + "loss": 2.548048210144043, + "step": 56820 + }, + { + "epoch": 0.1547298216959303, + "grad_norm": 2.1514132022857666, + "learning_rate": 1.6928499386670303e-05, + "loss": 2.537339782714844, + "step": 56840 + }, + { + "epoch": 0.15478426568667483, + "grad_norm": 2.1740047931671143, + "learning_rate": 1.692740902276135e-05, + "loss": 2.526084327697754, + "step": 56860 + }, + { + "epoch": 0.15483870967741936, + "grad_norm": 2.053633689880371, + "learning_rate": 1.6926318658852394e-05, + "loss": 2.5077632904052733, + "step": 56880 + }, + { + "epoch": 0.15489315366816386, + "grad_norm": 2.1899118423461914, + "learning_rate": 1.692522829494344e-05, + "loss": 2.5404754638671876, + "step": 56900 + }, + { + "epoch": 0.1549475976589084, + "grad_norm": 2.201960802078247, + "learning_rate": 1.6924137931034483e-05, + "loss": 2.537998008728027, + "step": 56920 + }, + { + "epoch": 0.15500204164965292, + "grad_norm": 2.192662477493286, + "learning_rate": 1.692304756712553e-05, + "loss": 2.5263654708862306, + "step": 56940 + }, + { + "epoch": 0.15505648564039745, + "grad_norm": 2.102816581726074, + "learning_rate": 1.6921957203216574e-05, + "loss": 2.505845832824707, + "step": 56960 + }, + { + "epoch": 0.15511092963114195, + "grad_norm": 2.20900297164917, + "learning_rate": 1.692086683930762e-05, + "loss": 2.5398746490478517, + "step": 56980 + }, + { + "epoch": 0.15516537362188648, + "grad_norm": 2.1066157817840576, + "learning_rate": 1.6919776475398666e-05, + "loss": 2.522246170043945, + "step": 57000 + }, + { + "epoch": 0.155219817612631, + "grad_norm": 2.216038942337036, + "learning_rate": 1.6918686111489712e-05, + "loss": 2.502796173095703, + "step": 57020 + }, + { + "epoch": 0.15527426160337554, + "grad_norm": 2.078583240509033, + "learning_rate": 1.6917595747580758e-05, + "loss": 2.5441825866699217, + "step": 57040 + }, + { + "epoch": 0.15532870559412004, + "grad_norm": 2.383822202682495, + "learning_rate": 1.6916505383671803e-05, + "loss": 2.5547176361083985, + "step": 57060 + }, + { + "epoch": 0.15538314958486457, + "grad_norm": 2.072566509246826, + "learning_rate": 1.6915415019762846e-05, + "loss": 2.5507469177246094, + "step": 57080 + }, + { + "epoch": 0.1554375935756091, + "grad_norm": 2.2191197872161865, + "learning_rate": 1.691432465585389e-05, + "loss": 2.5113597869873048, + "step": 57100 + }, + { + "epoch": 0.1554920375663536, + "grad_norm": 2.1486496925354004, + "learning_rate": 1.6913234291944937e-05, + "loss": 2.524210739135742, + "step": 57120 + }, + { + "epoch": 0.15554648155709813, + "grad_norm": 2.1760518550872803, + "learning_rate": 1.6912143928035983e-05, + "loss": 2.539645195007324, + "step": 57140 + }, + { + "epoch": 0.15560092554784266, + "grad_norm": 2.3715837001800537, + "learning_rate": 1.691105356412703e-05, + "loss": 2.5227306365966795, + "step": 57160 + }, + { + "epoch": 0.1556553695385872, + "grad_norm": 2.2420401573181152, + "learning_rate": 1.6909963200218075e-05, + "loss": 2.521051788330078, + "step": 57180 + }, + { + "epoch": 0.1557098135293317, + "grad_norm": 2.0396101474761963, + "learning_rate": 1.690887283630912e-05, + "loss": 2.5734699249267576, + "step": 57200 + }, + { + "epoch": 0.15576425752007622, + "grad_norm": 2.095059871673584, + "learning_rate": 1.6907782472400166e-05, + "loss": 2.552274703979492, + "step": 57220 + }, + { + "epoch": 0.15581870151082075, + "grad_norm": 2.101344585418701, + "learning_rate": 1.690669210849121e-05, + "loss": 2.543696975708008, + "step": 57240 + }, + { + "epoch": 0.15587314550156525, + "grad_norm": 2.1857762336730957, + "learning_rate": 1.6905601744582255e-05, + "loss": 2.5356975555419923, + "step": 57260 + }, + { + "epoch": 0.15592758949230978, + "grad_norm": 2.3165111541748047, + "learning_rate": 1.69045113806733e-05, + "loss": 2.530042839050293, + "step": 57280 + }, + { + "epoch": 0.1559820334830543, + "grad_norm": 2.1072421073913574, + "learning_rate": 1.6903421016764346e-05, + "loss": 2.5394565582275392, + "step": 57300 + }, + { + "epoch": 0.15603647747379884, + "grad_norm": 2.174537181854248, + "learning_rate": 1.6902330652855392e-05, + "loss": 2.523482894897461, + "step": 57320 + }, + { + "epoch": 0.15609092146454334, + "grad_norm": 2.230020046234131, + "learning_rate": 1.6901240288946438e-05, + "loss": 2.541861152648926, + "step": 57340 + }, + { + "epoch": 0.15614536545528787, + "grad_norm": 2.1523945331573486, + "learning_rate": 1.6900149925037484e-05, + "loss": 2.5283109664916994, + "step": 57360 + }, + { + "epoch": 0.1561998094460324, + "grad_norm": 3.787902355194092, + "learning_rate": 1.689905956112853e-05, + "loss": 2.545893669128418, + "step": 57380 + }, + { + "epoch": 0.1562542534367769, + "grad_norm": 2.2248587608337402, + "learning_rate": 1.6897969197219572e-05, + "loss": 2.560102081298828, + "step": 57400 + }, + { + "epoch": 0.15630869742752143, + "grad_norm": 2.282923460006714, + "learning_rate": 1.6896878833310618e-05, + "loss": 2.5114377975463866, + "step": 57420 + }, + { + "epoch": 0.15636314141826596, + "grad_norm": 2.213953971862793, + "learning_rate": 1.6895788469401664e-05, + "loss": 2.555810546875, + "step": 57440 + }, + { + "epoch": 0.1564175854090105, + "grad_norm": 2.084214210510254, + "learning_rate": 1.689469810549271e-05, + "loss": 2.530007743835449, + "step": 57460 + }, + { + "epoch": 0.156472029399755, + "grad_norm": 2.1619033813476562, + "learning_rate": 1.6893607741583755e-05, + "loss": 2.5298946380615233, + "step": 57480 + }, + { + "epoch": 0.15652647339049952, + "grad_norm": 2.26930570602417, + "learning_rate": 1.68925173776748e-05, + "loss": 2.5456634521484376, + "step": 57500 + }, + { + "epoch": 0.15658091738124405, + "grad_norm": 2.163874864578247, + "learning_rate": 1.6891427013765847e-05, + "loss": 2.521952819824219, + "step": 57520 + }, + { + "epoch": 0.15663536137198858, + "grad_norm": 2.1129958629608154, + "learning_rate": 1.6890336649856893e-05, + "loss": 2.518373489379883, + "step": 57540 + }, + { + "epoch": 0.15668980536273308, + "grad_norm": 2.1919162273406982, + "learning_rate": 1.6889246285947935e-05, + "loss": 2.512187385559082, + "step": 57560 + }, + { + "epoch": 0.1567442493534776, + "grad_norm": 2.092074155807495, + "learning_rate": 1.688815592203898e-05, + "loss": 2.5414058685302736, + "step": 57580 + }, + { + "epoch": 0.15679869334422214, + "grad_norm": 2.026899814605713, + "learning_rate": 1.6887065558130027e-05, + "loss": 2.5249767303466797, + "step": 57600 + }, + { + "epoch": 0.15685313733496664, + "grad_norm": 2.16996431350708, + "learning_rate": 1.6885975194221072e-05, + "loss": 2.513270950317383, + "step": 57620 + }, + { + "epoch": 0.15690758132571117, + "grad_norm": 2.089921236038208, + "learning_rate": 1.6884884830312118e-05, + "loss": 2.557298278808594, + "step": 57640 + }, + { + "epoch": 0.1569620253164557, + "grad_norm": 2.193930149078369, + "learning_rate": 1.6883794466403164e-05, + "loss": 2.5270149230957033, + "step": 57660 + }, + { + "epoch": 0.15701646930720023, + "grad_norm": 2.1742796897888184, + "learning_rate": 1.688270410249421e-05, + "loss": 2.550922393798828, + "step": 57680 + }, + { + "epoch": 0.15707091329794473, + "grad_norm": 2.0740017890930176, + "learning_rate": 1.6881613738585256e-05, + "loss": 2.5082164764404298, + "step": 57700 + }, + { + "epoch": 0.15712535728868926, + "grad_norm": 2.090914011001587, + "learning_rate": 1.6880523374676298e-05, + "loss": 2.5274799346923826, + "step": 57720 + }, + { + "epoch": 0.1571798012794338, + "grad_norm": 2.1639394760131836, + "learning_rate": 1.6879433010767344e-05, + "loss": 2.5246341705322264, + "step": 57740 + }, + { + "epoch": 0.1572342452701783, + "grad_norm": 2.1302177906036377, + "learning_rate": 1.687834264685839e-05, + "loss": 2.5390066146850585, + "step": 57760 + }, + { + "epoch": 0.15728868926092282, + "grad_norm": 3.09032940864563, + "learning_rate": 1.6877252282949436e-05, + "loss": 2.5110141754150392, + "step": 57780 + }, + { + "epoch": 0.15734313325166735, + "grad_norm": 2.0753729343414307, + "learning_rate": 1.687616191904048e-05, + "loss": 2.567436408996582, + "step": 57800 + }, + { + "epoch": 0.15739757724241188, + "grad_norm": 2.0857837200164795, + "learning_rate": 1.6875071555131527e-05, + "loss": 2.540340042114258, + "step": 57820 + }, + { + "epoch": 0.15745202123315638, + "grad_norm": 2.1297402381896973, + "learning_rate": 1.6873981191222573e-05, + "loss": 2.5060632705688475, + "step": 57840 + }, + { + "epoch": 0.1575064652239009, + "grad_norm": 2.091890573501587, + "learning_rate": 1.687289082731362e-05, + "loss": 2.5276700973510744, + "step": 57860 + }, + { + "epoch": 0.15756090921464544, + "grad_norm": 2.2968101501464844, + "learning_rate": 1.687180046340466e-05, + "loss": 2.5338497161865234, + "step": 57880 + }, + { + "epoch": 0.15761535320538997, + "grad_norm": 2.3494997024536133, + "learning_rate": 1.6870710099495707e-05, + "loss": 2.5521385192871096, + "step": 57900 + }, + { + "epoch": 0.15766979719613447, + "grad_norm": 2.1762871742248535, + "learning_rate": 1.6869619735586753e-05, + "loss": 2.517198181152344, + "step": 57920 + }, + { + "epoch": 0.157724241186879, + "grad_norm": 2.176905870437622, + "learning_rate": 1.68685293716778e-05, + "loss": 2.5713783264160157, + "step": 57940 + }, + { + "epoch": 0.15777868517762353, + "grad_norm": 2.0933456420898438, + "learning_rate": 1.6867439007768844e-05, + "loss": 2.5225400924682617, + "step": 57960 + }, + { + "epoch": 0.15783312916836803, + "grad_norm": 2.221723794937134, + "learning_rate": 1.686634864385989e-05, + "loss": 2.566448211669922, + "step": 57980 + }, + { + "epoch": 0.15788757315911256, + "grad_norm": 2.8937571048736572, + "learning_rate": 1.6865258279950936e-05, + "loss": 2.5156618118286134, + "step": 58000 + }, + { + "epoch": 0.1579420171498571, + "grad_norm": 2.158874034881592, + "learning_rate": 1.6864167916041982e-05, + "loss": 2.5292858123779296, + "step": 58020 + }, + { + "epoch": 0.15799646114060162, + "grad_norm": 2.0936365127563477, + "learning_rate": 1.6863077552133024e-05, + "loss": 2.539317512512207, + "step": 58040 + }, + { + "epoch": 0.15805090513134612, + "grad_norm": 2.16062593460083, + "learning_rate": 1.686198718822407e-05, + "loss": 2.5236278533935548, + "step": 58060 + }, + { + "epoch": 0.15810534912209065, + "grad_norm": 2.091580629348755, + "learning_rate": 1.6860896824315116e-05, + "loss": 2.558602714538574, + "step": 58080 + }, + { + "epoch": 0.15815979311283518, + "grad_norm": 2.063850164413452, + "learning_rate": 1.685980646040616e-05, + "loss": 2.572038269042969, + "step": 58100 + }, + { + "epoch": 0.15821423710357968, + "grad_norm": 2.0665791034698486, + "learning_rate": 1.6858716096497207e-05, + "loss": 2.542484092712402, + "step": 58120 + }, + { + "epoch": 0.1582686810943242, + "grad_norm": 2.061373472213745, + "learning_rate": 1.6857625732588253e-05, + "loss": 2.4945375442504885, + "step": 58140 + }, + { + "epoch": 0.15832312508506874, + "grad_norm": 2.0258431434631348, + "learning_rate": 1.68565353686793e-05, + "loss": 2.518861198425293, + "step": 58160 + }, + { + "epoch": 0.15837756907581327, + "grad_norm": 2.2714340686798096, + "learning_rate": 1.6855445004770345e-05, + "loss": 2.5638633728027345, + "step": 58180 + }, + { + "epoch": 0.15843201306655777, + "grad_norm": 2.031830310821533, + "learning_rate": 1.6854354640861387e-05, + "loss": 2.543252944946289, + "step": 58200 + }, + { + "epoch": 0.1584864570573023, + "grad_norm": 2.1109485626220703, + "learning_rate": 1.6853264276952433e-05, + "loss": 2.5333633422851562, + "step": 58220 + }, + { + "epoch": 0.15854090104804683, + "grad_norm": 2.7491369247436523, + "learning_rate": 1.685217391304348e-05, + "loss": 2.5323226928710936, + "step": 58240 + }, + { + "epoch": 0.15859534503879133, + "grad_norm": 2.200026512145996, + "learning_rate": 1.6851083549134525e-05, + "loss": 2.5280838012695312, + "step": 58260 + }, + { + "epoch": 0.15864978902953586, + "grad_norm": 2.2993664741516113, + "learning_rate": 1.684999318522557e-05, + "loss": 2.529017448425293, + "step": 58280 + }, + { + "epoch": 0.1587042330202804, + "grad_norm": 2.5626375675201416, + "learning_rate": 1.6848902821316616e-05, + "loss": 2.529058265686035, + "step": 58300 + }, + { + "epoch": 0.15875867701102492, + "grad_norm": 2.0648841857910156, + "learning_rate": 1.6847812457407662e-05, + "loss": 2.5228471755981445, + "step": 58320 + }, + { + "epoch": 0.15881312100176942, + "grad_norm": 2.08809494972229, + "learning_rate": 1.6846722093498708e-05, + "loss": 2.5351234436035157, + "step": 58340 + }, + { + "epoch": 0.15886756499251395, + "grad_norm": 2.317645311355591, + "learning_rate": 1.684563172958975e-05, + "loss": 2.573990249633789, + "step": 58360 + }, + { + "epoch": 0.15892200898325848, + "grad_norm": 2.061152696609497, + "learning_rate": 1.6844541365680796e-05, + "loss": 2.5230045318603516, + "step": 58380 + }, + { + "epoch": 0.158976452974003, + "grad_norm": 2.04959774017334, + "learning_rate": 1.6843451001771842e-05, + "loss": 2.5295993804931642, + "step": 58400 + }, + { + "epoch": 0.1590308969647475, + "grad_norm": 2.0630991458892822, + "learning_rate": 1.6842360637862888e-05, + "loss": 2.517447662353516, + "step": 58420 + }, + { + "epoch": 0.15908534095549204, + "grad_norm": 2.1286988258361816, + "learning_rate": 1.6841270273953934e-05, + "loss": 2.5473024368286135, + "step": 58440 + }, + { + "epoch": 0.15913978494623657, + "grad_norm": 2.1311583518981934, + "learning_rate": 1.684017991004498e-05, + "loss": 2.526260185241699, + "step": 58460 + }, + { + "epoch": 0.15919422893698107, + "grad_norm": 2.1179633140563965, + "learning_rate": 1.6839089546136025e-05, + "loss": 2.5204553604125977, + "step": 58480 + }, + { + "epoch": 0.1592486729277256, + "grad_norm": 2.1375160217285156, + "learning_rate": 1.683799918222707e-05, + "loss": 2.561386299133301, + "step": 58500 + }, + { + "epoch": 0.15930311691847013, + "grad_norm": 2.050184488296509, + "learning_rate": 1.6836908818318114e-05, + "loss": 2.5348995208740233, + "step": 58520 + }, + { + "epoch": 0.15935756090921466, + "grad_norm": 2.1383328437805176, + "learning_rate": 1.683581845440916e-05, + "loss": 2.5229360580444338, + "step": 58540 + }, + { + "epoch": 0.15941200489995916, + "grad_norm": 2.146054744720459, + "learning_rate": 1.6834728090500205e-05, + "loss": 2.5382551193237304, + "step": 58560 + }, + { + "epoch": 0.1594664488907037, + "grad_norm": 2.1250195503234863, + "learning_rate": 1.683363772659125e-05, + "loss": 2.527790641784668, + "step": 58580 + }, + { + "epoch": 0.15952089288144822, + "grad_norm": 2.134289503097534, + "learning_rate": 1.6832547362682297e-05, + "loss": 2.4985702514648436, + "step": 58600 + }, + { + "epoch": 0.15957533687219272, + "grad_norm": 2.3635101318359375, + "learning_rate": 1.6831456998773343e-05, + "loss": 2.503383827209473, + "step": 58620 + }, + { + "epoch": 0.15962978086293725, + "grad_norm": 2.216562032699585, + "learning_rate": 1.683036663486439e-05, + "loss": 2.523506736755371, + "step": 58640 + }, + { + "epoch": 0.15968422485368178, + "grad_norm": 2.1269690990448, + "learning_rate": 1.6829276270955434e-05, + "loss": 2.5175718307495116, + "step": 58660 + }, + { + "epoch": 0.1597386688444263, + "grad_norm": 2.03286075592041, + "learning_rate": 1.6828185907046477e-05, + "loss": 2.546217346191406, + "step": 58680 + }, + { + "epoch": 0.1597931128351708, + "grad_norm": 2.26318359375, + "learning_rate": 1.6827095543137522e-05, + "loss": 2.5466548919677736, + "step": 58700 + }, + { + "epoch": 0.15984755682591534, + "grad_norm": 2.1927316188812256, + "learning_rate": 1.6826005179228568e-05, + "loss": 2.510066795349121, + "step": 58720 + }, + { + "epoch": 0.15990200081665987, + "grad_norm": 2.9043538570404053, + "learning_rate": 1.6824914815319614e-05, + "loss": 2.542902374267578, + "step": 58740 + }, + { + "epoch": 0.1599564448074044, + "grad_norm": 2.181317090988159, + "learning_rate": 1.682382445141066e-05, + "loss": 2.5051458358764647, + "step": 58760 + }, + { + "epoch": 0.1600108887981489, + "grad_norm": 2.179403305053711, + "learning_rate": 1.6822734087501706e-05, + "loss": 2.5320659637451173, + "step": 58780 + }, + { + "epoch": 0.16006533278889343, + "grad_norm": 2.1121323108673096, + "learning_rate": 1.682164372359275e-05, + "loss": 2.5436607360839845, + "step": 58800 + }, + { + "epoch": 0.16011977677963796, + "grad_norm": 2.110603094100952, + "learning_rate": 1.6820553359683794e-05, + "loss": 2.4942880630493165, + "step": 58820 + }, + { + "epoch": 0.16017422077038246, + "grad_norm": 2.0819666385650635, + "learning_rate": 1.6819462995774843e-05, + "loss": 2.5331594467163088, + "step": 58840 + }, + { + "epoch": 0.160228664761127, + "grad_norm": 2.4833273887634277, + "learning_rate": 1.6818372631865885e-05, + "loss": 2.515485382080078, + "step": 58860 + }, + { + "epoch": 0.16028310875187152, + "grad_norm": 2.314976215362549, + "learning_rate": 1.681728226795693e-05, + "loss": 2.536971855163574, + "step": 58880 + }, + { + "epoch": 0.16033755274261605, + "grad_norm": 2.1736764907836914, + "learning_rate": 1.6816191904047977e-05, + "loss": 2.5085840225219727, + "step": 58900 + }, + { + "epoch": 0.16039199673336055, + "grad_norm": 2.2870166301727295, + "learning_rate": 1.6815101540139023e-05, + "loss": 2.527039337158203, + "step": 58920 + }, + { + "epoch": 0.16044644072410508, + "grad_norm": 2.2012834548950195, + "learning_rate": 1.681401117623007e-05, + "loss": 2.5519800186157227, + "step": 58940 + }, + { + "epoch": 0.1605008847148496, + "grad_norm": 2.1295766830444336, + "learning_rate": 1.6812920812321115e-05, + "loss": 2.509494400024414, + "step": 58960 + }, + { + "epoch": 0.1605553287055941, + "grad_norm": 2.096806764602661, + "learning_rate": 1.6811830448412157e-05, + "loss": 2.543512153625488, + "step": 58980 + }, + { + "epoch": 0.16060977269633864, + "grad_norm": 2.082092761993408, + "learning_rate": 1.6810740084503206e-05, + "loss": 2.5078243255615233, + "step": 59000 + }, + { + "epoch": 0.16066421668708317, + "grad_norm": 2.0420267581939697, + "learning_rate": 1.680964972059425e-05, + "loss": 2.5218982696533203, + "step": 59020 + }, + { + "epoch": 0.1607186606778277, + "grad_norm": 2.2765462398529053, + "learning_rate": 1.6808559356685294e-05, + "loss": 2.5413702011108397, + "step": 59040 + }, + { + "epoch": 0.1607731046685722, + "grad_norm": 2.2167084217071533, + "learning_rate": 1.680746899277634e-05, + "loss": 2.5423641204833984, + "step": 59060 + }, + { + "epoch": 0.16082754865931673, + "grad_norm": 2.112936019897461, + "learning_rate": 1.6806378628867386e-05, + "loss": 2.5764856338500977, + "step": 59080 + }, + { + "epoch": 0.16088199265006126, + "grad_norm": 2.2672884464263916, + "learning_rate": 1.6805288264958432e-05, + "loss": 2.518332862854004, + "step": 59100 + }, + { + "epoch": 0.16093643664080576, + "grad_norm": 2.2316479682922363, + "learning_rate": 1.6804197901049478e-05, + "loss": 2.4914342880249025, + "step": 59120 + }, + { + "epoch": 0.1609908806315503, + "grad_norm": 2.205244779586792, + "learning_rate": 1.680310753714052e-05, + "loss": 2.4967552185058595, + "step": 59140 + }, + { + "epoch": 0.16104532462229482, + "grad_norm": 2.038928985595703, + "learning_rate": 1.680201717323157e-05, + "loss": 2.532392883300781, + "step": 59160 + }, + { + "epoch": 0.16109976861303935, + "grad_norm": 2.0241856575012207, + "learning_rate": 1.680092680932261e-05, + "loss": 2.557646369934082, + "step": 59180 + }, + { + "epoch": 0.16115421260378385, + "grad_norm": 4.563831329345703, + "learning_rate": 1.6799836445413657e-05, + "loss": 2.51684684753418, + "step": 59200 + }, + { + "epoch": 0.16120865659452838, + "grad_norm": 2.2267303466796875, + "learning_rate": 1.6798746081504703e-05, + "loss": 2.5242544174194337, + "step": 59220 + }, + { + "epoch": 0.1612631005852729, + "grad_norm": 2.0794548988342285, + "learning_rate": 1.679765571759575e-05, + "loss": 2.5241458892822264, + "step": 59240 + }, + { + "epoch": 0.16131754457601744, + "grad_norm": 2.1853837966918945, + "learning_rate": 1.6796565353686795e-05, + "loss": 2.5488475799560546, + "step": 59260 + }, + { + "epoch": 0.16137198856676194, + "grad_norm": 2.1210546493530273, + "learning_rate": 1.679547498977784e-05, + "loss": 2.5213476181030274, + "step": 59280 + }, + { + "epoch": 0.16142643255750647, + "grad_norm": 2.137037515640259, + "learning_rate": 1.6794384625868883e-05, + "loss": 2.5150693893432616, + "step": 59300 + }, + { + "epoch": 0.161480876548251, + "grad_norm": 2.17594838142395, + "learning_rate": 1.6793294261959932e-05, + "loss": 2.5302520751953126, + "step": 59320 + }, + { + "epoch": 0.1615353205389955, + "grad_norm": 2.2092020511627197, + "learning_rate": 1.6792203898050975e-05, + "loss": 2.5256677627563477, + "step": 59340 + }, + { + "epoch": 0.16158976452974003, + "grad_norm": 2.1643638610839844, + "learning_rate": 1.679111353414202e-05, + "loss": 2.555333709716797, + "step": 59360 + }, + { + "epoch": 0.16164420852048456, + "grad_norm": 2.1008007526397705, + "learning_rate": 1.6790023170233066e-05, + "loss": 2.5337665557861326, + "step": 59380 + }, + { + "epoch": 0.16169865251122909, + "grad_norm": 2.1434473991394043, + "learning_rate": 1.6788932806324112e-05, + "loss": 2.497768211364746, + "step": 59400 + }, + { + "epoch": 0.1617530965019736, + "grad_norm": 2.192255735397339, + "learning_rate": 1.6787842442415158e-05, + "loss": 2.5300310134887694, + "step": 59420 + }, + { + "epoch": 0.16180754049271812, + "grad_norm": 2.1200859546661377, + "learning_rate": 1.6786752078506204e-05, + "loss": 2.559305763244629, + "step": 59440 + }, + { + "epoch": 0.16186198448346265, + "grad_norm": 2.375041961669922, + "learning_rate": 1.6785661714597246e-05, + "loss": 2.552577590942383, + "step": 59460 + }, + { + "epoch": 0.16191642847420715, + "grad_norm": 2.1751585006713867, + "learning_rate": 1.6784571350688295e-05, + "loss": 2.576370048522949, + "step": 59480 + }, + { + "epoch": 0.16197087246495168, + "grad_norm": 2.3025763034820557, + "learning_rate": 1.6783480986779338e-05, + "loss": 2.5494382858276365, + "step": 59500 + }, + { + "epoch": 0.1620253164556962, + "grad_norm": 2.0630431175231934, + "learning_rate": 1.6782390622870384e-05, + "loss": 2.5469148635864256, + "step": 59520 + }, + { + "epoch": 0.16207976044644073, + "grad_norm": 2.2068324089050293, + "learning_rate": 1.678130025896143e-05, + "loss": 2.547720527648926, + "step": 59540 + }, + { + "epoch": 0.16213420443718524, + "grad_norm": 2.0944650173187256, + "learning_rate": 1.6780209895052475e-05, + "loss": 2.534218406677246, + "step": 59560 + }, + { + "epoch": 0.16218864842792977, + "grad_norm": 2.0981523990631104, + "learning_rate": 1.677911953114352e-05, + "loss": 2.525988006591797, + "step": 59580 + }, + { + "epoch": 0.1622430924186743, + "grad_norm": 2.3430721759796143, + "learning_rate": 1.6778029167234567e-05, + "loss": 2.5270896911621095, + "step": 59600 + }, + { + "epoch": 0.1622975364094188, + "grad_norm": 2.1622841358184814, + "learning_rate": 1.677693880332561e-05, + "loss": 2.5378915786743166, + "step": 59620 + }, + { + "epoch": 0.16235198040016333, + "grad_norm": 2.0971031188964844, + "learning_rate": 1.677584843941666e-05, + "loss": 2.558993911743164, + "step": 59640 + }, + { + "epoch": 0.16240642439090786, + "grad_norm": 2.131856918334961, + "learning_rate": 1.67747580755077e-05, + "loss": 2.5471080780029296, + "step": 59660 + }, + { + "epoch": 0.16246086838165238, + "grad_norm": 2.3163187503814697, + "learning_rate": 1.6773667711598747e-05, + "loss": 2.5038484573364257, + "step": 59680 + }, + { + "epoch": 0.1625153123723969, + "grad_norm": 2.136077880859375, + "learning_rate": 1.6772577347689793e-05, + "loss": 2.513311195373535, + "step": 59700 + }, + { + "epoch": 0.16256975636314142, + "grad_norm": 2.1880404949188232, + "learning_rate": 1.677148698378084e-05, + "loss": 2.5631587982177733, + "step": 59720 + }, + { + "epoch": 0.16262420035388594, + "grad_norm": 2.1539716720581055, + "learning_rate": 1.6770396619871884e-05, + "loss": 2.5126646041870115, + "step": 59740 + }, + { + "epoch": 0.16267864434463047, + "grad_norm": 2.100029468536377, + "learning_rate": 1.676930625596293e-05, + "loss": 2.546038818359375, + "step": 59760 + }, + { + "epoch": 0.16273308833537498, + "grad_norm": 2.783024787902832, + "learning_rate": 1.6768215892053972e-05, + "loss": 2.5250850677490235, + "step": 59780 + }, + { + "epoch": 0.1627875323261195, + "grad_norm": 2.9265878200531006, + "learning_rate": 1.676712552814502e-05, + "loss": 2.512374687194824, + "step": 59800 + }, + { + "epoch": 0.16284197631686403, + "grad_norm": 2.0773041248321533, + "learning_rate": 1.6766035164236064e-05, + "loss": 2.536484146118164, + "step": 59820 + }, + { + "epoch": 0.16289642030760854, + "grad_norm": 2.1648807525634766, + "learning_rate": 1.676494480032711e-05, + "loss": 2.5220699310302734, + "step": 59840 + }, + { + "epoch": 0.16295086429835307, + "grad_norm": 2.1838572025299072, + "learning_rate": 1.6763854436418156e-05, + "loss": 2.504376029968262, + "step": 59860 + }, + { + "epoch": 0.1630053082890976, + "grad_norm": 2.125988483428955, + "learning_rate": 1.67627640725092e-05, + "loss": 2.5309356689453124, + "step": 59880 + }, + { + "epoch": 0.16305975227984212, + "grad_norm": 2.9098591804504395, + "learning_rate": 1.6761673708600247e-05, + "loss": 2.5254493713378907, + "step": 59900 + }, + { + "epoch": 0.16311419627058663, + "grad_norm": 2.268754482269287, + "learning_rate": 1.6760583344691293e-05, + "loss": 2.531932258605957, + "step": 59920 + }, + { + "epoch": 0.16316864026133115, + "grad_norm": 2.83504056930542, + "learning_rate": 1.6759492980782335e-05, + "loss": 2.538380241394043, + "step": 59940 + }, + { + "epoch": 0.16322308425207568, + "grad_norm": 2.05875825881958, + "learning_rate": 1.6758402616873385e-05, + "loss": 2.5627490997314455, + "step": 59960 + }, + { + "epoch": 0.16327752824282019, + "grad_norm": 2.1331686973571777, + "learning_rate": 1.6757312252964427e-05, + "loss": 2.5182373046875, + "step": 59980 + }, + { + "epoch": 0.16333197223356472, + "grad_norm": 3.0059473514556885, + "learning_rate": 1.6756221889055473e-05, + "loss": 2.5426406860351562, + "step": 60000 + }, + { + "epoch": 0.16338641622430924, + "grad_norm": 2.1146349906921387, + "learning_rate": 1.675513152514652e-05, + "loss": 2.5417552947998048, + "step": 60020 + }, + { + "epoch": 0.16344086021505377, + "grad_norm": 2.120727062225342, + "learning_rate": 1.6754041161237565e-05, + "loss": 2.5535091400146483, + "step": 60040 + }, + { + "epoch": 0.16349530420579828, + "grad_norm": 2.1278233528137207, + "learning_rate": 1.675295079732861e-05, + "loss": 2.5548685073852537, + "step": 60060 + }, + { + "epoch": 0.1635497481965428, + "grad_norm": 2.0629167556762695, + "learning_rate": 1.6751860433419656e-05, + "loss": 2.4893749237060545, + "step": 60080 + }, + { + "epoch": 0.16360419218728733, + "grad_norm": 2.2762646675109863, + "learning_rate": 1.67507700695107e-05, + "loss": 2.503635025024414, + "step": 60100 + }, + { + "epoch": 0.16365863617803186, + "grad_norm": 2.269946575164795, + "learning_rate": 1.6749679705601748e-05, + "loss": 2.5418521881103517, + "step": 60120 + }, + { + "epoch": 0.16371308016877636, + "grad_norm": 2.238037586212158, + "learning_rate": 1.674858934169279e-05, + "loss": 2.548518943786621, + "step": 60140 + }, + { + "epoch": 0.1637675241595209, + "grad_norm": 2.09891414642334, + "learning_rate": 1.6747498977783836e-05, + "loss": 2.545430374145508, + "step": 60160 + }, + { + "epoch": 0.16382196815026542, + "grad_norm": 2.2138190269470215, + "learning_rate": 1.6746408613874882e-05, + "loss": 2.526405143737793, + "step": 60180 + }, + { + "epoch": 0.16387641214100993, + "grad_norm": 2.152268648147583, + "learning_rate": 1.6745318249965928e-05, + "loss": 2.5300809860229494, + "step": 60200 + }, + { + "epoch": 0.16393085613175445, + "grad_norm": 2.3959527015686035, + "learning_rate": 1.6744227886056973e-05, + "loss": 2.541533279418945, + "step": 60220 + }, + { + "epoch": 0.16398530012249898, + "grad_norm": 3.2566628456115723, + "learning_rate": 1.674313752214802e-05, + "loss": 2.51328182220459, + "step": 60240 + }, + { + "epoch": 0.1640397441132435, + "grad_norm": 2.085576057434082, + "learning_rate": 1.674204715823906e-05, + "loss": 2.527967071533203, + "step": 60260 + }, + { + "epoch": 0.16409418810398801, + "grad_norm": 2.3067307472229004, + "learning_rate": 1.674095679433011e-05, + "loss": 2.5032621383666993, + "step": 60280 + }, + { + "epoch": 0.16414863209473254, + "grad_norm": 2.5341262817382812, + "learning_rate": 1.6739866430421153e-05, + "loss": 2.5289291381835937, + "step": 60300 + }, + { + "epoch": 0.16420307608547707, + "grad_norm": 2.2566511631011963, + "learning_rate": 1.67387760665122e-05, + "loss": 2.5217796325683595, + "step": 60320 + }, + { + "epoch": 0.16425752007622157, + "grad_norm": 2.161198616027832, + "learning_rate": 1.6737685702603245e-05, + "loss": 2.520004653930664, + "step": 60340 + }, + { + "epoch": 0.1643119640669661, + "grad_norm": 2.165881872177124, + "learning_rate": 1.673659533869429e-05, + "loss": 2.54635066986084, + "step": 60360 + }, + { + "epoch": 0.16436640805771063, + "grad_norm": 2.1672067642211914, + "learning_rate": 1.6735504974785336e-05, + "loss": 2.5552343368530273, + "step": 60380 + }, + { + "epoch": 0.16442085204845516, + "grad_norm": 2.222656726837158, + "learning_rate": 1.6734414610876382e-05, + "loss": 2.547446441650391, + "step": 60400 + }, + { + "epoch": 0.16447529603919966, + "grad_norm": 2.21459698677063, + "learning_rate": 1.6733324246967425e-05, + "loss": 2.512654495239258, + "step": 60420 + }, + { + "epoch": 0.1645297400299442, + "grad_norm": 2.1714372634887695, + "learning_rate": 1.6732233883058474e-05, + "loss": 2.5555757522583007, + "step": 60440 + }, + { + "epoch": 0.16458418402068872, + "grad_norm": 3.538719654083252, + "learning_rate": 1.6731143519149516e-05, + "loss": 2.51492919921875, + "step": 60460 + }, + { + "epoch": 0.16463862801143322, + "grad_norm": 2.159299612045288, + "learning_rate": 1.6730053155240562e-05, + "loss": 2.497933006286621, + "step": 60480 + }, + { + "epoch": 0.16469307200217775, + "grad_norm": 2.110689878463745, + "learning_rate": 1.6728962791331608e-05, + "loss": 2.5348915100097655, + "step": 60500 + }, + { + "epoch": 0.16474751599292228, + "grad_norm": 2.0758073329925537, + "learning_rate": 1.6727872427422654e-05, + "loss": 2.5439102172851564, + "step": 60520 + }, + { + "epoch": 0.1648019599836668, + "grad_norm": 2.174318552017212, + "learning_rate": 1.67267820635137e-05, + "loss": 2.5345409393310545, + "step": 60540 + }, + { + "epoch": 0.16485640397441131, + "grad_norm": 2.134145736694336, + "learning_rate": 1.6725691699604745e-05, + "loss": 2.4994586944580077, + "step": 60560 + }, + { + "epoch": 0.16491084796515584, + "grad_norm": 2.2195518016815186, + "learning_rate": 1.6724601335695788e-05, + "loss": 2.5400373458862306, + "step": 60580 + }, + { + "epoch": 0.16496529195590037, + "grad_norm": 2.14821457862854, + "learning_rate": 1.6723510971786837e-05, + "loss": 2.542746162414551, + "step": 60600 + }, + { + "epoch": 0.1650197359466449, + "grad_norm": 2.4000940322875977, + "learning_rate": 1.672242060787788e-05, + "loss": 2.504446792602539, + "step": 60620 + }, + { + "epoch": 0.1650741799373894, + "grad_norm": 2.123568296432495, + "learning_rate": 1.6721330243968925e-05, + "loss": 2.48557186126709, + "step": 60640 + }, + { + "epoch": 0.16512862392813393, + "grad_norm": 2.114060878753662, + "learning_rate": 1.672023988005997e-05, + "loss": 2.551035690307617, + "step": 60660 + }, + { + "epoch": 0.16518306791887846, + "grad_norm": 2.1696090698242188, + "learning_rate": 1.6719149516151017e-05, + "loss": 2.5176525115966797, + "step": 60680 + }, + { + "epoch": 0.16523751190962296, + "grad_norm": 2.2700376510620117, + "learning_rate": 1.6718059152242063e-05, + "loss": 2.5411211013793946, + "step": 60700 + }, + { + "epoch": 0.1652919559003675, + "grad_norm": 2.062830686569214, + "learning_rate": 1.671696878833311e-05, + "loss": 2.523507308959961, + "step": 60720 + }, + { + "epoch": 0.16534639989111202, + "grad_norm": 2.0071215629577637, + "learning_rate": 1.671587842442415e-05, + "loss": 2.53291015625, + "step": 60740 + }, + { + "epoch": 0.16540084388185655, + "grad_norm": 2.076470136642456, + "learning_rate": 1.67147880605152e-05, + "loss": 2.543926811218262, + "step": 60760 + }, + { + "epoch": 0.16545528787260105, + "grad_norm": 2.284724473953247, + "learning_rate": 1.6713697696606243e-05, + "loss": 2.541183280944824, + "step": 60780 + }, + { + "epoch": 0.16550973186334558, + "grad_norm": 2.070897102355957, + "learning_rate": 1.671260733269729e-05, + "loss": 2.521127128601074, + "step": 60800 + }, + { + "epoch": 0.1655641758540901, + "grad_norm": 2.183929204940796, + "learning_rate": 1.6711516968788334e-05, + "loss": 2.5363185882568358, + "step": 60820 + }, + { + "epoch": 0.1656186198448346, + "grad_norm": 3.777548313140869, + "learning_rate": 1.671042660487938e-05, + "loss": 2.5152185440063475, + "step": 60840 + }, + { + "epoch": 0.16567306383557914, + "grad_norm": 2.9402072429656982, + "learning_rate": 1.6709336240970426e-05, + "loss": 2.5268362045288084, + "step": 60860 + }, + { + "epoch": 0.16572750782632367, + "grad_norm": 2.0458078384399414, + "learning_rate": 1.670824587706147e-05, + "loss": 2.5214344024658204, + "step": 60880 + }, + { + "epoch": 0.1657819518170682, + "grad_norm": 2.1919546127319336, + "learning_rate": 1.6707155513152514e-05, + "loss": 2.5394853591918944, + "step": 60900 + }, + { + "epoch": 0.1658363958078127, + "grad_norm": 2.2766153812408447, + "learning_rate": 1.6706065149243563e-05, + "loss": 2.542243003845215, + "step": 60920 + }, + { + "epoch": 0.16589083979855723, + "grad_norm": 2.1306333541870117, + "learning_rate": 1.6704974785334606e-05, + "loss": 2.5325450897216797, + "step": 60940 + }, + { + "epoch": 0.16594528378930176, + "grad_norm": 2.2290570735931396, + "learning_rate": 1.670388442142565e-05, + "loss": 2.5269983291625975, + "step": 60960 + }, + { + "epoch": 0.1659997277800463, + "grad_norm": 2.2989046573638916, + "learning_rate": 1.6702794057516697e-05, + "loss": 2.5670827865600585, + "step": 60980 + }, + { + "epoch": 0.1660541717707908, + "grad_norm": 2.208099365234375, + "learning_rate": 1.6701703693607743e-05, + "loss": 2.5012609481811525, + "step": 61000 + }, + { + "epoch": 0.16610861576153532, + "grad_norm": 2.153103828430176, + "learning_rate": 1.670061332969879e-05, + "loss": 2.527216339111328, + "step": 61020 + }, + { + "epoch": 0.16616305975227985, + "grad_norm": 2.1079108715057373, + "learning_rate": 1.6699522965789835e-05, + "loss": 2.539297103881836, + "step": 61040 + }, + { + "epoch": 0.16621750374302435, + "grad_norm": 2.155595541000366, + "learning_rate": 1.6698432601880877e-05, + "loss": 2.51525993347168, + "step": 61060 + }, + { + "epoch": 0.16627194773376888, + "grad_norm": 2.107593059539795, + "learning_rate": 1.6697342237971926e-05, + "loss": 2.534857749938965, + "step": 61080 + }, + { + "epoch": 0.1663263917245134, + "grad_norm": 2.078104019165039, + "learning_rate": 1.669625187406297e-05, + "loss": 2.534446907043457, + "step": 61100 + }, + { + "epoch": 0.16638083571525794, + "grad_norm": 2.1425883769989014, + "learning_rate": 1.6695161510154014e-05, + "loss": 2.5349184036254884, + "step": 61120 + }, + { + "epoch": 0.16643527970600244, + "grad_norm": 2.147444009780884, + "learning_rate": 1.669407114624506e-05, + "loss": 2.5377391815185546, + "step": 61140 + }, + { + "epoch": 0.16648972369674697, + "grad_norm": 2.1141138076782227, + "learning_rate": 1.6692980782336106e-05, + "loss": 2.556113624572754, + "step": 61160 + }, + { + "epoch": 0.1665441676874915, + "grad_norm": 2.1515464782714844, + "learning_rate": 1.6691890418427152e-05, + "loss": 2.540782928466797, + "step": 61180 + }, + { + "epoch": 0.166598611678236, + "grad_norm": 2.810175895690918, + "learning_rate": 1.6690800054518198e-05, + "loss": 2.51660099029541, + "step": 61200 + }, + { + "epoch": 0.16665305566898053, + "grad_norm": 2.1504952907562256, + "learning_rate": 1.668970969060924e-05, + "loss": 2.5133216857910154, + "step": 61220 + }, + { + "epoch": 0.16670749965972506, + "grad_norm": 2.247340440750122, + "learning_rate": 1.668861932670029e-05, + "loss": 2.5375692367553713, + "step": 61240 + }, + { + "epoch": 0.1667619436504696, + "grad_norm": 2.122303009033203, + "learning_rate": 1.6687528962791332e-05, + "loss": 2.4762123107910154, + "step": 61260 + }, + { + "epoch": 0.1668163876412141, + "grad_norm": 2.156012535095215, + "learning_rate": 1.6686438598882378e-05, + "loss": 2.5271692276000977, + "step": 61280 + }, + { + "epoch": 0.16687083163195862, + "grad_norm": 2.150991916656494, + "learning_rate": 1.6685348234973423e-05, + "loss": 2.5299129486083984, + "step": 61300 + }, + { + "epoch": 0.16692527562270315, + "grad_norm": 2.093238592147827, + "learning_rate": 1.668425787106447e-05, + "loss": 2.5421642303466796, + "step": 61320 + }, + { + "epoch": 0.16697971961344765, + "grad_norm": 2.1270127296447754, + "learning_rate": 1.6683167507155515e-05, + "loss": 2.5395273208618163, + "step": 61340 + }, + { + "epoch": 0.16703416360419218, + "grad_norm": 2.1463639736175537, + "learning_rate": 1.668207714324656e-05, + "loss": 2.5252969741821287, + "step": 61360 + }, + { + "epoch": 0.1670886075949367, + "grad_norm": 2.00319504737854, + "learning_rate": 1.6680986779337603e-05, + "loss": 2.5317752838134764, + "step": 61380 + }, + { + "epoch": 0.16714305158568124, + "grad_norm": 2.0161166191101074, + "learning_rate": 1.6679896415428652e-05, + "loss": 2.5236263275146484, + "step": 61400 + }, + { + "epoch": 0.16719749557642574, + "grad_norm": 14.201663970947266, + "learning_rate": 1.6678806051519695e-05, + "loss": 2.5592672348022463, + "step": 61420 + }, + { + "epoch": 0.16725193956717027, + "grad_norm": 2.186107635498047, + "learning_rate": 1.667771568761074e-05, + "loss": 2.5074012756347654, + "step": 61440 + }, + { + "epoch": 0.1673063835579148, + "grad_norm": 2.401510715484619, + "learning_rate": 1.6676625323701786e-05, + "loss": 2.5364274978637695, + "step": 61460 + }, + { + "epoch": 0.16736082754865933, + "grad_norm": 2.1867122650146484, + "learning_rate": 1.6675534959792832e-05, + "loss": 2.5617368698120115, + "step": 61480 + }, + { + "epoch": 0.16741527153940383, + "grad_norm": 2.1600208282470703, + "learning_rate": 1.6674444595883878e-05, + "loss": 2.5343130111694334, + "step": 61500 + }, + { + "epoch": 0.16746971553014836, + "grad_norm": 2.2028324604034424, + "learning_rate": 1.6673354231974924e-05, + "loss": 2.525233840942383, + "step": 61520 + }, + { + "epoch": 0.1675241595208929, + "grad_norm": 2.0929791927337646, + "learning_rate": 1.6672263868065966e-05, + "loss": 2.522884559631348, + "step": 61540 + }, + { + "epoch": 0.1675786035116374, + "grad_norm": 2.31990647315979, + "learning_rate": 1.6671173504157016e-05, + "loss": 2.5314727783203126, + "step": 61560 + }, + { + "epoch": 0.16763304750238192, + "grad_norm": 3.5957512855529785, + "learning_rate": 1.6670083140248058e-05, + "loss": 2.508609580993652, + "step": 61580 + }, + { + "epoch": 0.16768749149312645, + "grad_norm": 2.060431957244873, + "learning_rate": 1.6668992776339104e-05, + "loss": 2.5331350326538087, + "step": 61600 + }, + { + "epoch": 0.16774193548387098, + "grad_norm": 2.240873098373413, + "learning_rate": 1.666790241243015e-05, + "loss": 2.518323516845703, + "step": 61620 + }, + { + "epoch": 0.16779637947461548, + "grad_norm": 2.3210997581481934, + "learning_rate": 1.6666812048521195e-05, + "loss": 2.526697540283203, + "step": 61640 + }, + { + "epoch": 0.16785082346536, + "grad_norm": 2.188856601715088, + "learning_rate": 1.666572168461224e-05, + "loss": 2.5371078491210937, + "step": 61660 + }, + { + "epoch": 0.16790526745610454, + "grad_norm": 2.204845666885376, + "learning_rate": 1.6664631320703287e-05, + "loss": 2.5031597137451174, + "step": 61680 + }, + { + "epoch": 0.16795971144684904, + "grad_norm": 2.2690556049346924, + "learning_rate": 1.666354095679433e-05, + "loss": 2.5043075561523436, + "step": 61700 + }, + { + "epoch": 0.16801415543759357, + "grad_norm": 2.106783390045166, + "learning_rate": 1.666245059288538e-05, + "loss": 2.5126781463623047, + "step": 61720 + }, + { + "epoch": 0.1680685994283381, + "grad_norm": 2.140298843383789, + "learning_rate": 1.666136022897642e-05, + "loss": 2.5264780044555666, + "step": 61740 + }, + { + "epoch": 0.16812304341908263, + "grad_norm": 2.0293211936950684, + "learning_rate": 1.6660269865067467e-05, + "loss": 2.4877328872680664, + "step": 61760 + }, + { + "epoch": 0.16817748740982713, + "grad_norm": 2.121107339859009, + "learning_rate": 1.6659179501158513e-05, + "loss": 2.5388700485229494, + "step": 61780 + }, + { + "epoch": 0.16823193140057166, + "grad_norm": 2.1813454627990723, + "learning_rate": 1.665808913724956e-05, + "loss": 2.5099021911621096, + "step": 61800 + }, + { + "epoch": 0.1682863753913162, + "grad_norm": 2.0196211338043213, + "learning_rate": 1.6656998773340604e-05, + "loss": 2.5193550109863283, + "step": 61820 + }, + { + "epoch": 0.1683408193820607, + "grad_norm": 2.2178127765655518, + "learning_rate": 1.665590840943165e-05, + "loss": 2.5329753875732424, + "step": 61840 + }, + { + "epoch": 0.16839526337280522, + "grad_norm": 2.6539700031280518, + "learning_rate": 1.6654818045522692e-05, + "loss": 2.5175508499145507, + "step": 61860 + }, + { + "epoch": 0.16844970736354975, + "grad_norm": 2.047274351119995, + "learning_rate": 1.665372768161374e-05, + "loss": 2.495205879211426, + "step": 61880 + }, + { + "epoch": 0.16850415135429428, + "grad_norm": 2.2542366981506348, + "learning_rate": 1.6652637317704784e-05, + "loss": 2.5286108016967774, + "step": 61900 + }, + { + "epoch": 0.16855859534503878, + "grad_norm": 2.1918280124664307, + "learning_rate": 1.665154695379583e-05, + "loss": 2.5360134124755858, + "step": 61920 + }, + { + "epoch": 0.1686130393357833, + "grad_norm": 2.1245036125183105, + "learning_rate": 1.6650456589886876e-05, + "loss": 2.5173595428466795, + "step": 61940 + }, + { + "epoch": 0.16866748332652784, + "grad_norm": 2.0765974521636963, + "learning_rate": 1.664936622597792e-05, + "loss": 2.5078054428100587, + "step": 61960 + }, + { + "epoch": 0.16872192731727237, + "grad_norm": 2.1342813968658447, + "learning_rate": 1.6648275862068967e-05, + "loss": 2.508590888977051, + "step": 61980 + }, + { + "epoch": 0.16877637130801687, + "grad_norm": 2.15236496925354, + "learning_rate": 1.6647185498160013e-05, + "loss": 2.5543418884277345, + "step": 62000 + }, + { + "epoch": 0.1688308152987614, + "grad_norm": 2.236636161804199, + "learning_rate": 1.6646095134251056e-05, + "loss": 2.524974822998047, + "step": 62020 + }, + { + "epoch": 0.16888525928950593, + "grad_norm": 2.1372463703155518, + "learning_rate": 1.6645004770342105e-05, + "loss": 2.5703569412231446, + "step": 62040 + }, + { + "epoch": 0.16893970328025043, + "grad_norm": 2.49336314201355, + "learning_rate": 1.6643914406433147e-05, + "loss": 2.5469078063964843, + "step": 62060 + }, + { + "epoch": 0.16899414727099496, + "grad_norm": 2.139397382736206, + "learning_rate": 1.6642824042524193e-05, + "loss": 2.5421424865722657, + "step": 62080 + }, + { + "epoch": 0.1690485912617395, + "grad_norm": 2.2025485038757324, + "learning_rate": 1.664173367861524e-05, + "loss": 2.5177375793457033, + "step": 62100 + }, + { + "epoch": 0.16910303525248402, + "grad_norm": 2.1513445377349854, + "learning_rate": 1.6640643314706285e-05, + "loss": 2.5199001312255858, + "step": 62120 + }, + { + "epoch": 0.16915747924322852, + "grad_norm": 2.155477285385132, + "learning_rate": 1.663955295079733e-05, + "loss": 2.5376848220825194, + "step": 62140 + }, + { + "epoch": 0.16921192323397305, + "grad_norm": 2.134167194366455, + "learning_rate": 1.6638462586888376e-05, + "loss": 2.5673349380493162, + "step": 62160 + }, + { + "epoch": 0.16926636722471758, + "grad_norm": 2.1282241344451904, + "learning_rate": 1.663737222297942e-05, + "loss": 2.5205841064453125, + "step": 62180 + }, + { + "epoch": 0.16932081121546208, + "grad_norm": 2.2411322593688965, + "learning_rate": 1.6636281859070468e-05, + "loss": 2.526498794555664, + "step": 62200 + }, + { + "epoch": 0.1693752552062066, + "grad_norm": 2.4371848106384277, + "learning_rate": 1.663519149516151e-05, + "loss": 2.535456085205078, + "step": 62220 + }, + { + "epoch": 0.16942969919695114, + "grad_norm": 4.607028007507324, + "learning_rate": 1.663410113125256e-05, + "loss": 2.5165227890014648, + "step": 62240 + }, + { + "epoch": 0.16948414318769567, + "grad_norm": 2.223350763320923, + "learning_rate": 1.6633010767343602e-05, + "loss": 2.527704620361328, + "step": 62260 + }, + { + "epoch": 0.16953858717844017, + "grad_norm": 2.3385376930236816, + "learning_rate": 1.6631920403434648e-05, + "loss": 2.5558677673339845, + "step": 62280 + }, + { + "epoch": 0.1695930311691847, + "grad_norm": 2.199287176132202, + "learning_rate": 1.6630830039525694e-05, + "loss": 2.513274002075195, + "step": 62300 + }, + { + "epoch": 0.16964747515992923, + "grad_norm": 2.0961363315582275, + "learning_rate": 1.662973967561674e-05, + "loss": 2.5510303497314455, + "step": 62320 + }, + { + "epoch": 0.16970191915067376, + "grad_norm": 2.159196138381958, + "learning_rate": 1.6628649311707782e-05, + "loss": 2.55455322265625, + "step": 62340 + }, + { + "epoch": 0.16975636314141826, + "grad_norm": 2.0756616592407227, + "learning_rate": 1.662755894779883e-05, + "loss": 2.544492721557617, + "step": 62360 + }, + { + "epoch": 0.1698108071321628, + "grad_norm": 2.116427421569824, + "learning_rate": 1.6626468583889873e-05, + "loss": 2.5019357681274412, + "step": 62380 + }, + { + "epoch": 0.16986525112290732, + "grad_norm": 2.1866841316223145, + "learning_rate": 1.6625378219980923e-05, + "loss": 2.5041259765625, + "step": 62400 + }, + { + "epoch": 0.16991969511365182, + "grad_norm": 2.2052886486053467, + "learning_rate": 1.6624287856071965e-05, + "loss": 2.5387157440185546, + "step": 62420 + }, + { + "epoch": 0.16997413910439635, + "grad_norm": 2.0740785598754883, + "learning_rate": 1.662319749216301e-05, + "loss": 2.577713966369629, + "step": 62440 + }, + { + "epoch": 0.17002858309514088, + "grad_norm": 2.065912961959839, + "learning_rate": 1.6622107128254057e-05, + "loss": 2.5552753448486327, + "step": 62460 + }, + { + "epoch": 0.1700830270858854, + "grad_norm": 2.3623619079589844, + "learning_rate": 1.6621016764345102e-05, + "loss": 2.5581762313842775, + "step": 62480 + }, + { + "epoch": 0.1701374710766299, + "grad_norm": 1.991363286972046, + "learning_rate": 1.6619926400436145e-05, + "loss": 2.5279951095581055, + "step": 62500 + }, + { + "epoch": 0.17019191506737444, + "grad_norm": 2.210205316543579, + "learning_rate": 1.6618836036527194e-05, + "loss": 2.5455345153808593, + "step": 62520 + }, + { + "epoch": 0.17024635905811897, + "grad_norm": 2.313100576400757, + "learning_rate": 1.6617745672618236e-05, + "loss": 2.5465681076049806, + "step": 62540 + }, + { + "epoch": 0.17030080304886347, + "grad_norm": 2.1676089763641357, + "learning_rate": 1.6616655308709286e-05, + "loss": 2.526323699951172, + "step": 62560 + }, + { + "epoch": 0.170355247039608, + "grad_norm": 2.117124080657959, + "learning_rate": 1.6615564944800328e-05, + "loss": 2.528690719604492, + "step": 62580 + }, + { + "epoch": 0.17040969103035253, + "grad_norm": 2.125103712081909, + "learning_rate": 1.6614474580891374e-05, + "loss": 2.5146181106567385, + "step": 62600 + }, + { + "epoch": 0.17046413502109706, + "grad_norm": 2.1348557472229004, + "learning_rate": 1.661338421698242e-05, + "loss": 2.5431228637695313, + "step": 62620 + }, + { + "epoch": 0.17051857901184156, + "grad_norm": 2.0537261962890625, + "learning_rate": 1.6612293853073466e-05, + "loss": 2.519633483886719, + "step": 62640 + }, + { + "epoch": 0.1705730230025861, + "grad_norm": 2.2476933002471924, + "learning_rate": 1.6611203489164508e-05, + "loss": 2.521118927001953, + "step": 62660 + }, + { + "epoch": 0.17062746699333062, + "grad_norm": 2.063788890838623, + "learning_rate": 1.6610113125255557e-05, + "loss": 2.512138557434082, + "step": 62680 + }, + { + "epoch": 0.17068191098407512, + "grad_norm": 2.088043212890625, + "learning_rate": 1.66090227613466e-05, + "loss": 2.505668067932129, + "step": 62700 + }, + { + "epoch": 0.17073635497481965, + "grad_norm": 2.098395347595215, + "learning_rate": 1.660793239743765e-05, + "loss": 2.5001670837402346, + "step": 62720 + }, + { + "epoch": 0.17079079896556418, + "grad_norm": 2.318063259124756, + "learning_rate": 1.660684203352869e-05, + "loss": 2.55593318939209, + "step": 62740 + }, + { + "epoch": 0.1708452429563087, + "grad_norm": 2.374922275543213, + "learning_rate": 1.6605751669619737e-05, + "loss": 2.5373357772827148, + "step": 62760 + }, + { + "epoch": 0.1708996869470532, + "grad_norm": 2.39039945602417, + "learning_rate": 1.6604661305710783e-05, + "loss": 2.5422653198242187, + "step": 62780 + }, + { + "epoch": 0.17095413093779774, + "grad_norm": 2.0952517986297607, + "learning_rate": 1.660357094180183e-05, + "loss": 2.5244556427001954, + "step": 62800 + }, + { + "epoch": 0.17100857492854227, + "grad_norm": 2.1800262928009033, + "learning_rate": 1.660248057789287e-05, + "loss": 2.527425765991211, + "step": 62820 + }, + { + "epoch": 0.1710630189192868, + "grad_norm": 2.199526071548462, + "learning_rate": 1.660139021398392e-05, + "loss": 2.512772560119629, + "step": 62840 + }, + { + "epoch": 0.1711174629100313, + "grad_norm": 2.135460615158081, + "learning_rate": 1.6600299850074963e-05, + "loss": 2.534647750854492, + "step": 62860 + }, + { + "epoch": 0.17117190690077583, + "grad_norm": 2.151463747024536, + "learning_rate": 1.6599209486166012e-05, + "loss": 2.522428512573242, + "step": 62880 + }, + { + "epoch": 0.17122635089152036, + "grad_norm": 2.2231364250183105, + "learning_rate": 1.6598119122257054e-05, + "loss": 2.534769630432129, + "step": 62900 + }, + { + "epoch": 0.17128079488226486, + "grad_norm": 2.307779550552368, + "learning_rate": 1.65970287583481e-05, + "loss": 2.5169139862060548, + "step": 62920 + }, + { + "epoch": 0.1713352388730094, + "grad_norm": 2.1724281311035156, + "learning_rate": 1.6595938394439146e-05, + "loss": 2.5513553619384766, + "step": 62940 + }, + { + "epoch": 0.17138968286375392, + "grad_norm": 2.2166192531585693, + "learning_rate": 1.659484803053019e-05, + "loss": 2.5393457412719727, + "step": 62960 + }, + { + "epoch": 0.17144412685449845, + "grad_norm": 2.139261245727539, + "learning_rate": 1.6593757666621234e-05, + "loss": 2.528676223754883, + "step": 62980 + }, + { + "epoch": 0.17149857084524295, + "grad_norm": 2.0995383262634277, + "learning_rate": 1.6592667302712283e-05, + "loss": 2.543923187255859, + "step": 63000 + }, + { + "epoch": 0.17155301483598748, + "grad_norm": 2.1878395080566406, + "learning_rate": 1.6591576938803326e-05, + "loss": 2.5155811309814453, + "step": 63020 + }, + { + "epoch": 0.171607458826732, + "grad_norm": 2.10064959526062, + "learning_rate": 1.6590486574894375e-05, + "loss": 2.5211053848266602, + "step": 63040 + }, + { + "epoch": 0.1716619028174765, + "grad_norm": 2.217088222503662, + "learning_rate": 1.6589396210985417e-05, + "loss": 2.5271734237670898, + "step": 63060 + }, + { + "epoch": 0.17171634680822104, + "grad_norm": 2.219085216522217, + "learning_rate": 1.6588305847076463e-05, + "loss": 2.5362007141113283, + "step": 63080 + }, + { + "epoch": 0.17177079079896557, + "grad_norm": 2.1354238986968994, + "learning_rate": 1.658721548316751e-05, + "loss": 2.5159938812255858, + "step": 63100 + }, + { + "epoch": 0.1718252347897101, + "grad_norm": 2.0864086151123047, + "learning_rate": 1.6586125119258555e-05, + "loss": 2.516068458557129, + "step": 63120 + }, + { + "epoch": 0.1718796787804546, + "grad_norm": 2.2330572605133057, + "learning_rate": 1.6585034755349597e-05, + "loss": 2.512338638305664, + "step": 63140 + }, + { + "epoch": 0.17193412277119913, + "grad_norm": 2.1054341793060303, + "learning_rate": 1.6583944391440646e-05, + "loss": 2.5109405517578125, + "step": 63160 + }, + { + "epoch": 0.17198856676194366, + "grad_norm": 2.1700451374053955, + "learning_rate": 1.658285402753169e-05, + "loss": 2.5048513412475586, + "step": 63180 + }, + { + "epoch": 0.17204301075268819, + "grad_norm": 2.3696212768554688, + "learning_rate": 1.6581763663622738e-05, + "loss": 2.527309799194336, + "step": 63200 + }, + { + "epoch": 0.1720974547434327, + "grad_norm": 2.2792389392852783, + "learning_rate": 1.658067329971378e-05, + "loss": 2.539259338378906, + "step": 63220 + }, + { + "epoch": 0.17215189873417722, + "grad_norm": 2.4028353691101074, + "learning_rate": 1.6579582935804826e-05, + "loss": 2.542485809326172, + "step": 63240 + }, + { + "epoch": 0.17220634272492175, + "grad_norm": 2.5815935134887695, + "learning_rate": 1.6578492571895872e-05, + "loss": 2.487162971496582, + "step": 63260 + }, + { + "epoch": 0.17226078671566625, + "grad_norm": 2.397055149078369, + "learning_rate": 1.6577402207986918e-05, + "loss": 2.545796585083008, + "step": 63280 + }, + { + "epoch": 0.17231523070641078, + "grad_norm": 2.1782143115997314, + "learning_rate": 1.657631184407796e-05, + "loss": 2.5131731033325195, + "step": 63300 + }, + { + "epoch": 0.1723696746971553, + "grad_norm": 2.11417555809021, + "learning_rate": 1.657522148016901e-05, + "loss": 2.5320226669311525, + "step": 63320 + }, + { + "epoch": 0.17242411868789984, + "grad_norm": 2.136441707611084, + "learning_rate": 1.6574131116260052e-05, + "loss": 2.511879730224609, + "step": 63340 + }, + { + "epoch": 0.17247856267864434, + "grad_norm": 2.1429219245910645, + "learning_rate": 1.65730407523511e-05, + "loss": 2.509536552429199, + "step": 63360 + }, + { + "epoch": 0.17253300666938887, + "grad_norm": 2.1063332557678223, + "learning_rate": 1.6571950388442144e-05, + "loss": 2.5310745239257812, + "step": 63380 + }, + { + "epoch": 0.1725874506601334, + "grad_norm": 2.1145238876342773, + "learning_rate": 1.657086002453319e-05, + "loss": 2.5625423431396483, + "step": 63400 + }, + { + "epoch": 0.1726418946508779, + "grad_norm": 2.2286336421966553, + "learning_rate": 1.6569769660624235e-05, + "loss": 2.5267662048339843, + "step": 63420 + }, + { + "epoch": 0.17269633864162243, + "grad_norm": 2.2277262210845947, + "learning_rate": 1.656867929671528e-05, + "loss": 2.5489681243896483, + "step": 63440 + }, + { + "epoch": 0.17275078263236696, + "grad_norm": 2.232593059539795, + "learning_rate": 1.6567588932806323e-05, + "loss": 2.5533910751342774, + "step": 63460 + }, + { + "epoch": 0.17280522662311149, + "grad_norm": 2.1957521438598633, + "learning_rate": 1.6566498568897373e-05, + "loss": 2.50940055847168, + "step": 63480 + }, + { + "epoch": 0.172859670613856, + "grad_norm": 2.147383689880371, + "learning_rate": 1.6565408204988415e-05, + "loss": 2.5686504364013674, + "step": 63500 + }, + { + "epoch": 0.17291411460460052, + "grad_norm": 2.131673812866211, + "learning_rate": 1.6564317841079464e-05, + "loss": 2.5410768508911135, + "step": 63520 + }, + { + "epoch": 0.17296855859534505, + "grad_norm": 2.246596097946167, + "learning_rate": 1.6563227477170507e-05, + "loss": 2.536431312561035, + "step": 63540 + }, + { + "epoch": 0.17302300258608955, + "grad_norm": 2.143176555633545, + "learning_rate": 1.6562137113261552e-05, + "loss": 2.557330513000488, + "step": 63560 + }, + { + "epoch": 0.17307744657683408, + "grad_norm": 21.830411911010742, + "learning_rate": 1.6561046749352598e-05, + "loss": 2.5455360412597656, + "step": 63580 + }, + { + "epoch": 0.1731318905675786, + "grad_norm": 2.039825916290283, + "learning_rate": 1.6559956385443644e-05, + "loss": 2.5431949615478517, + "step": 63600 + }, + { + "epoch": 0.17318633455832314, + "grad_norm": 2.1572797298431396, + "learning_rate": 1.6558866021534686e-05, + "loss": 2.51627140045166, + "step": 63620 + }, + { + "epoch": 0.17324077854906764, + "grad_norm": 2.082399606704712, + "learning_rate": 1.6557775657625736e-05, + "loss": 2.4964962005615234, + "step": 63640 + }, + { + "epoch": 0.17329522253981217, + "grad_norm": 2.1182003021240234, + "learning_rate": 1.6556685293716778e-05, + "loss": 2.492158889770508, + "step": 63660 + }, + { + "epoch": 0.1733496665305567, + "grad_norm": 2.217388153076172, + "learning_rate": 1.6555594929807827e-05, + "loss": 2.5233911514282226, + "step": 63680 + }, + { + "epoch": 0.17340411052130122, + "grad_norm": 2.194948196411133, + "learning_rate": 1.655450456589887e-05, + "loss": 2.5222984313964845, + "step": 63700 + }, + { + "epoch": 0.17345855451204573, + "grad_norm": 2.085425853729248, + "learning_rate": 1.6553414201989915e-05, + "loss": 2.5369707107543946, + "step": 63720 + }, + { + "epoch": 0.17351299850279026, + "grad_norm": 2.1777679920196533, + "learning_rate": 1.655232383808096e-05, + "loss": 2.508889007568359, + "step": 63740 + }, + { + "epoch": 0.17356744249353478, + "grad_norm": 2.185948610305786, + "learning_rate": 1.6551233474172007e-05, + "loss": 2.52929573059082, + "step": 63760 + }, + { + "epoch": 0.1736218864842793, + "grad_norm": 2.1205391883850098, + "learning_rate": 1.655014311026305e-05, + "loss": 2.5768556594848633, + "step": 63780 + }, + { + "epoch": 0.17367633047502382, + "grad_norm": 2.0523648262023926, + "learning_rate": 1.65490527463541e-05, + "loss": 2.525034713745117, + "step": 63800 + }, + { + "epoch": 0.17373077446576835, + "grad_norm": 2.270650863647461, + "learning_rate": 1.654796238244514e-05, + "loss": 2.5165678024291993, + "step": 63820 + }, + { + "epoch": 0.17378521845651287, + "grad_norm": 2.1953721046447754, + "learning_rate": 1.6546872018536187e-05, + "loss": 2.5488203048706053, + "step": 63840 + }, + { + "epoch": 0.17383966244725738, + "grad_norm": 2.213557720184326, + "learning_rate": 1.6545781654627233e-05, + "loss": 2.5412132263183596, + "step": 63860 + }, + { + "epoch": 0.1738941064380019, + "grad_norm": 2.1876771450042725, + "learning_rate": 1.654469129071828e-05, + "loss": 2.5333383560180662, + "step": 63880 + }, + { + "epoch": 0.17394855042874643, + "grad_norm": 2.1803805828094482, + "learning_rate": 1.6543600926809324e-05, + "loss": 2.526073455810547, + "step": 63900 + }, + { + "epoch": 0.17400299441949094, + "grad_norm": 2.020641326904297, + "learning_rate": 1.654251056290037e-05, + "loss": 2.534478950500488, + "step": 63920 + }, + { + "epoch": 0.17405743841023547, + "grad_norm": 2.1367430686950684, + "learning_rate": 1.6541420198991416e-05, + "loss": 2.523224449157715, + "step": 63940 + }, + { + "epoch": 0.17411188240098, + "grad_norm": 2.3027970790863037, + "learning_rate": 1.6540329835082462e-05, + "loss": 2.5424211502075194, + "step": 63960 + }, + { + "epoch": 0.17416632639172452, + "grad_norm": 2.2035915851593018, + "learning_rate": 1.6539239471173504e-05, + "loss": 2.5652276992797853, + "step": 63980 + }, + { + "epoch": 0.17422077038246903, + "grad_norm": 3.3414154052734375, + "learning_rate": 1.653814910726455e-05, + "loss": 2.4989322662353515, + "step": 64000 + }, + { + "epoch": 0.17427521437321356, + "grad_norm": 2.259462356567383, + "learning_rate": 1.6537058743355596e-05, + "loss": 2.535392951965332, + "step": 64020 + }, + { + "epoch": 0.17432965836395808, + "grad_norm": 2.210552930831909, + "learning_rate": 1.653596837944664e-05, + "loss": 2.514975929260254, + "step": 64040 + }, + { + "epoch": 0.17438410235470259, + "grad_norm": 10.065014839172363, + "learning_rate": 1.6534878015537687e-05, + "loss": 2.5349626541137695, + "step": 64060 + }, + { + "epoch": 0.17443854634544712, + "grad_norm": 2.155895948410034, + "learning_rate": 1.6533787651628733e-05, + "loss": 2.541601371765137, + "step": 64080 + }, + { + "epoch": 0.17449299033619164, + "grad_norm": 3.1015830039978027, + "learning_rate": 1.653269728771978e-05, + "loss": 2.5516157150268555, + "step": 64100 + }, + { + "epoch": 0.17454743432693617, + "grad_norm": 2.5590903759002686, + "learning_rate": 1.6531606923810825e-05, + "loss": 2.5254800796508787, + "step": 64120 + }, + { + "epoch": 0.17460187831768068, + "grad_norm": 2.183504819869995, + "learning_rate": 1.6530516559901867e-05, + "loss": 2.5332271575927736, + "step": 64140 + }, + { + "epoch": 0.1746563223084252, + "grad_norm": 2.6527063846588135, + "learning_rate": 1.6529426195992913e-05, + "loss": 2.555986785888672, + "step": 64160 + }, + { + "epoch": 0.17471076629916973, + "grad_norm": 2.134610414505005, + "learning_rate": 1.652833583208396e-05, + "loss": 2.5243383407592774, + "step": 64180 + }, + { + "epoch": 0.17476521028991426, + "grad_norm": 2.1136343479156494, + "learning_rate": 1.6527245468175005e-05, + "loss": 2.555874061584473, + "step": 64200 + }, + { + "epoch": 0.17481965428065876, + "grad_norm": 2.1804072856903076, + "learning_rate": 1.652615510426605e-05, + "loss": 2.5474340438842775, + "step": 64220 + }, + { + "epoch": 0.1748740982714033, + "grad_norm": 2.1229631900787354, + "learning_rate": 1.6525064740357096e-05, + "loss": 2.5048147201538087, + "step": 64240 + }, + { + "epoch": 0.17492854226214782, + "grad_norm": 2.2586073875427246, + "learning_rate": 1.6523974376448142e-05, + "loss": 2.529627227783203, + "step": 64260 + }, + { + "epoch": 0.17498298625289233, + "grad_norm": 2.2385334968566895, + "learning_rate": 1.6522884012539188e-05, + "loss": 2.526688003540039, + "step": 64280 + }, + { + "epoch": 0.17503743024363685, + "grad_norm": 2.179994583129883, + "learning_rate": 1.652179364863023e-05, + "loss": 2.5224910736083985, + "step": 64300 + }, + { + "epoch": 0.17509187423438138, + "grad_norm": 2.207085132598877, + "learning_rate": 1.6520703284721276e-05, + "loss": 2.5399946212768554, + "step": 64320 + }, + { + "epoch": 0.1751463182251259, + "grad_norm": 2.083495855331421, + "learning_rate": 1.6519612920812322e-05, + "loss": 2.540165138244629, + "step": 64340 + }, + { + "epoch": 0.17520076221587041, + "grad_norm": 2.156376600265503, + "learning_rate": 1.6518522556903368e-05, + "loss": 2.53741569519043, + "step": 64360 + }, + { + "epoch": 0.17525520620661494, + "grad_norm": 2.2242531776428223, + "learning_rate": 1.6517432192994414e-05, + "loss": 2.531936836242676, + "step": 64380 + }, + { + "epoch": 0.17530965019735947, + "grad_norm": 2.5804245471954346, + "learning_rate": 1.651634182908546e-05, + "loss": 2.5015680313110353, + "step": 64400 + }, + { + "epoch": 0.17536409418810397, + "grad_norm": 2.4732093811035156, + "learning_rate": 1.6515251465176505e-05, + "loss": 2.503419876098633, + "step": 64420 + }, + { + "epoch": 0.1754185381788485, + "grad_norm": 2.0471558570861816, + "learning_rate": 1.651416110126755e-05, + "loss": 2.5342735290527343, + "step": 64440 + }, + { + "epoch": 0.17547298216959303, + "grad_norm": 2.3274240493774414, + "learning_rate": 1.6513070737358593e-05, + "loss": 2.537408638000488, + "step": 64460 + }, + { + "epoch": 0.17552742616033756, + "grad_norm": 2.246659755706787, + "learning_rate": 1.651198037344964e-05, + "loss": 2.558134841918945, + "step": 64480 + }, + { + "epoch": 0.17558187015108206, + "grad_norm": 2.21769642829895, + "learning_rate": 1.6510890009540685e-05, + "loss": 2.517192268371582, + "step": 64500 + }, + { + "epoch": 0.1756363141418266, + "grad_norm": 2.277353286743164, + "learning_rate": 1.650979964563173e-05, + "loss": 2.53431396484375, + "step": 64520 + }, + { + "epoch": 0.17569075813257112, + "grad_norm": 2.1311261653900146, + "learning_rate": 1.6508709281722777e-05, + "loss": 2.4863155364990233, + "step": 64540 + }, + { + "epoch": 0.17574520212331565, + "grad_norm": 2.076150417327881, + "learning_rate": 1.6507618917813823e-05, + "loss": 2.546453285217285, + "step": 64560 + }, + { + "epoch": 0.17579964611406015, + "grad_norm": 2.1231067180633545, + "learning_rate": 1.650652855390487e-05, + "loss": 2.5555116653442385, + "step": 64580 + }, + { + "epoch": 0.17585409010480468, + "grad_norm": 2.2426788806915283, + "learning_rate": 1.6505438189995914e-05, + "loss": 2.5683963775634764, + "step": 64600 + }, + { + "epoch": 0.1759085340955492, + "grad_norm": 2.0816280841827393, + "learning_rate": 1.6504347826086957e-05, + "loss": 2.5700651168823243, + "step": 64620 + }, + { + "epoch": 0.17596297808629371, + "grad_norm": 2.2405922412872314, + "learning_rate": 1.6503257462178002e-05, + "loss": 2.4971071243286134, + "step": 64640 + }, + { + "epoch": 0.17601742207703824, + "grad_norm": 2.07867431640625, + "learning_rate": 1.6502167098269048e-05, + "loss": 2.5321359634399414, + "step": 64660 + }, + { + "epoch": 0.17607186606778277, + "grad_norm": 2.6553051471710205, + "learning_rate": 1.6501076734360094e-05, + "loss": 2.5307445526123047, + "step": 64680 + }, + { + "epoch": 0.1761263100585273, + "grad_norm": 2.105212926864624, + "learning_rate": 1.649998637045114e-05, + "loss": 2.539122200012207, + "step": 64700 + }, + { + "epoch": 0.1761807540492718, + "grad_norm": 2.191957473754883, + "learning_rate": 1.6498896006542186e-05, + "loss": 2.5380401611328125, + "step": 64720 + }, + { + "epoch": 0.17623519804001633, + "grad_norm": 2.2840771675109863, + "learning_rate": 1.649780564263323e-05, + "loss": 2.5224960327148436, + "step": 64740 + }, + { + "epoch": 0.17628964203076086, + "grad_norm": 2.070115566253662, + "learning_rate": 1.6496715278724277e-05, + "loss": 2.5331382751464844, + "step": 64760 + }, + { + "epoch": 0.17634408602150536, + "grad_norm": 2.4697422981262207, + "learning_rate": 1.649562491481532e-05, + "loss": 2.5236190795898437, + "step": 64780 + }, + { + "epoch": 0.1763985300122499, + "grad_norm": 2.171046018600464, + "learning_rate": 1.6494534550906365e-05, + "loss": 2.573470878601074, + "step": 64800 + }, + { + "epoch": 0.17645297400299442, + "grad_norm": 2.3508379459381104, + "learning_rate": 1.649344418699741e-05, + "loss": 2.5210784912109374, + "step": 64820 + }, + { + "epoch": 0.17650741799373895, + "grad_norm": 2.3465054035186768, + "learning_rate": 1.6492353823088457e-05, + "loss": 2.523530387878418, + "step": 64840 + }, + { + "epoch": 0.17656186198448345, + "grad_norm": 2.1286253929138184, + "learning_rate": 1.6491263459179503e-05, + "loss": 2.5368824005126953, + "step": 64860 + }, + { + "epoch": 0.17661630597522798, + "grad_norm": 2.1119537353515625, + "learning_rate": 1.649017309527055e-05, + "loss": 2.5394575119018556, + "step": 64880 + }, + { + "epoch": 0.1766707499659725, + "grad_norm": 2.140129566192627, + "learning_rate": 1.6489082731361595e-05, + "loss": 2.5026649475097655, + "step": 64900 + }, + { + "epoch": 0.176725193956717, + "grad_norm": 2.0474863052368164, + "learning_rate": 1.648799236745264e-05, + "loss": 2.504299354553223, + "step": 64920 + }, + { + "epoch": 0.17677963794746154, + "grad_norm": 2.328782320022583, + "learning_rate": 1.6486902003543683e-05, + "loss": 2.519208526611328, + "step": 64940 + }, + { + "epoch": 0.17683408193820607, + "grad_norm": 2.0587878227233887, + "learning_rate": 1.648581163963473e-05, + "loss": 2.5315547943115235, + "step": 64960 + }, + { + "epoch": 0.1768885259289506, + "grad_norm": 2.202556848526001, + "learning_rate": 1.6484721275725774e-05, + "loss": 2.5356927871704102, + "step": 64980 + }, + { + "epoch": 0.1769429699196951, + "grad_norm": 2.1438920497894287, + "learning_rate": 1.648363091181682e-05, + "loss": 2.5166616439819336, + "step": 65000 + }, + { + "epoch": 0.17699741391043963, + "grad_norm": 2.419626474380493, + "learning_rate": 1.6482540547907866e-05, + "loss": 2.5350452423095704, + "step": 65020 + }, + { + "epoch": 0.17705185790118416, + "grad_norm": 2.1321046352386475, + "learning_rate": 1.6481450183998912e-05, + "loss": 2.5287010192871096, + "step": 65040 + }, + { + "epoch": 0.1771063018919287, + "grad_norm": 2.168868064880371, + "learning_rate": 1.6480359820089958e-05, + "loss": 2.5511501312255858, + "step": 65060 + }, + { + "epoch": 0.1771607458826732, + "grad_norm": 2.2862775325775146, + "learning_rate": 1.6479269456181003e-05, + "loss": 2.514273834228516, + "step": 65080 + }, + { + "epoch": 0.17721518987341772, + "grad_norm": 2.2148542404174805, + "learning_rate": 1.6478179092272046e-05, + "loss": 2.5024032592773438, + "step": 65100 + }, + { + "epoch": 0.17726963386416225, + "grad_norm": 2.026684045791626, + "learning_rate": 1.647708872836309e-05, + "loss": 2.5349519729614256, + "step": 65120 + }, + { + "epoch": 0.17732407785490675, + "grad_norm": 2.3359134197235107, + "learning_rate": 1.6475998364454137e-05, + "loss": 2.520768737792969, + "step": 65140 + }, + { + "epoch": 0.17737852184565128, + "grad_norm": 2.5894522666931152, + "learning_rate": 1.6474908000545183e-05, + "loss": 2.5259775161743163, + "step": 65160 + }, + { + "epoch": 0.1774329658363958, + "grad_norm": 2.2806591987609863, + "learning_rate": 1.647381763663623e-05, + "loss": 2.47955322265625, + "step": 65180 + }, + { + "epoch": 0.17748740982714034, + "grad_norm": 2.077819347381592, + "learning_rate": 1.6472727272727275e-05, + "loss": 2.5285552978515624, + "step": 65200 + }, + { + "epoch": 0.17754185381788484, + "grad_norm": 2.0977015495300293, + "learning_rate": 1.647163690881832e-05, + "loss": 2.5091907501220705, + "step": 65220 + }, + { + "epoch": 0.17759629780862937, + "grad_norm": 2.2454745769500732, + "learning_rate": 1.6470546544909366e-05, + "loss": 2.5079994201660156, + "step": 65240 + }, + { + "epoch": 0.1776507417993739, + "grad_norm": 2.0954439640045166, + "learning_rate": 1.646945618100041e-05, + "loss": 2.5094316482543944, + "step": 65260 + }, + { + "epoch": 0.1777051857901184, + "grad_norm": 2.104118585586548, + "learning_rate": 1.6468365817091455e-05, + "loss": 2.518482780456543, + "step": 65280 + }, + { + "epoch": 0.17775962978086293, + "grad_norm": 2.1758663654327393, + "learning_rate": 1.64672754531825e-05, + "loss": 2.539626884460449, + "step": 65300 + }, + { + "epoch": 0.17781407377160746, + "grad_norm": 2.1722397804260254, + "learning_rate": 1.6466185089273546e-05, + "loss": 2.5432159423828127, + "step": 65320 + }, + { + "epoch": 0.177868517762352, + "grad_norm": 2.2671375274658203, + "learning_rate": 1.6465094725364592e-05, + "loss": 2.5355953216552733, + "step": 65340 + }, + { + "epoch": 0.1779229617530965, + "grad_norm": 2.031062126159668, + "learning_rate": 1.6464004361455638e-05, + "loss": 2.532270812988281, + "step": 65360 + }, + { + "epoch": 0.17797740574384102, + "grad_norm": 2.2761552333831787, + "learning_rate": 1.6462913997546684e-05, + "loss": 2.519493293762207, + "step": 65380 + }, + { + "epoch": 0.17803184973458555, + "grad_norm": 2.2182137966156006, + "learning_rate": 1.646182363363773e-05, + "loss": 2.523579978942871, + "step": 65400 + }, + { + "epoch": 0.17808629372533008, + "grad_norm": 2.2194650173187256, + "learning_rate": 1.6460733269728772e-05, + "loss": 2.5549034118652343, + "step": 65420 + }, + { + "epoch": 0.17814073771607458, + "grad_norm": 2.245755672454834, + "learning_rate": 1.6459642905819818e-05, + "loss": 2.530739974975586, + "step": 65440 + }, + { + "epoch": 0.1781951817068191, + "grad_norm": 2.244786262512207, + "learning_rate": 1.6458552541910864e-05, + "loss": 2.5515113830566407, + "step": 65460 + }, + { + "epoch": 0.17824962569756364, + "grad_norm": 2.01299786567688, + "learning_rate": 1.645746217800191e-05, + "loss": 2.548950958251953, + "step": 65480 + }, + { + "epoch": 0.17830406968830814, + "grad_norm": 2.132111072540283, + "learning_rate": 1.6456371814092955e-05, + "loss": 2.5146217346191406, + "step": 65500 + }, + { + "epoch": 0.17835851367905267, + "grad_norm": 2.2515037059783936, + "learning_rate": 1.6455281450184e-05, + "loss": 2.546468734741211, + "step": 65520 + }, + { + "epoch": 0.1784129576697972, + "grad_norm": 2.056551933288574, + "learning_rate": 1.6454191086275047e-05, + "loss": 2.5210540771484373, + "step": 65540 + }, + { + "epoch": 0.17846740166054173, + "grad_norm": 2.2209200859069824, + "learning_rate": 1.6453100722366093e-05, + "loss": 2.519929313659668, + "step": 65560 + }, + { + "epoch": 0.17852184565128623, + "grad_norm": 2.166886329650879, + "learning_rate": 1.6452010358457135e-05, + "loss": 2.543574905395508, + "step": 65580 + }, + { + "epoch": 0.17857628964203076, + "grad_norm": 2.066669225692749, + "learning_rate": 1.645091999454818e-05, + "loss": 2.5016029357910154, + "step": 65600 + }, + { + "epoch": 0.1786307336327753, + "grad_norm": 2.23599910736084, + "learning_rate": 1.6449829630639227e-05, + "loss": 2.5272972106933596, + "step": 65620 + }, + { + "epoch": 0.1786851776235198, + "grad_norm": 2.0669498443603516, + "learning_rate": 1.6448739266730273e-05, + "loss": 2.511641502380371, + "step": 65640 + }, + { + "epoch": 0.17873962161426432, + "grad_norm": 2.1823501586914062, + "learning_rate": 1.644764890282132e-05, + "loss": 2.537789154052734, + "step": 65660 + }, + { + "epoch": 0.17879406560500885, + "grad_norm": 2.1537532806396484, + "learning_rate": 1.6446558538912364e-05, + "loss": 2.5216300964355467, + "step": 65680 + }, + { + "epoch": 0.17884850959575338, + "grad_norm": 2.1457064151763916, + "learning_rate": 1.644546817500341e-05, + "loss": 2.5489032745361326, + "step": 65700 + }, + { + "epoch": 0.17890295358649788, + "grad_norm": 2.1640334129333496, + "learning_rate": 1.6444377811094456e-05, + "loss": 2.535961151123047, + "step": 65720 + }, + { + "epoch": 0.1789573975772424, + "grad_norm": 2.1190617084503174, + "learning_rate": 1.6443287447185498e-05, + "loss": 2.5083600997924806, + "step": 65740 + }, + { + "epoch": 0.17901184156798694, + "grad_norm": 2.1849026679992676, + "learning_rate": 1.6442197083276544e-05, + "loss": 2.5541522979736326, + "step": 65760 + }, + { + "epoch": 0.17906628555873144, + "grad_norm": 2.1125001907348633, + "learning_rate": 1.644110671936759e-05, + "loss": 2.5313690185546873, + "step": 65780 + }, + { + "epoch": 0.17912072954947597, + "grad_norm": 2.1480915546417236, + "learning_rate": 1.6440016355458636e-05, + "loss": 2.519460105895996, + "step": 65800 + }, + { + "epoch": 0.1791751735402205, + "grad_norm": 2.1122562885284424, + "learning_rate": 1.643892599154968e-05, + "loss": 2.529817008972168, + "step": 65820 + }, + { + "epoch": 0.17922961753096503, + "grad_norm": 2.0772287845611572, + "learning_rate": 1.6437835627640727e-05, + "loss": 2.547233009338379, + "step": 65840 + }, + { + "epoch": 0.17928406152170953, + "grad_norm": 2.0878608226776123, + "learning_rate": 1.6436745263731773e-05, + "loss": 2.509295845031738, + "step": 65860 + }, + { + "epoch": 0.17933850551245406, + "grad_norm": 2.1875951290130615, + "learning_rate": 1.643565489982282e-05, + "loss": 2.493545150756836, + "step": 65880 + }, + { + "epoch": 0.1793929495031986, + "grad_norm": 2.220414400100708, + "learning_rate": 1.643456453591386e-05, + "loss": 2.545139694213867, + "step": 65900 + }, + { + "epoch": 0.17944739349394312, + "grad_norm": 2.134160280227661, + "learning_rate": 1.6433474172004907e-05, + "loss": 2.531596565246582, + "step": 65920 + }, + { + "epoch": 0.17950183748468762, + "grad_norm": 2.219066858291626, + "learning_rate": 1.6432383808095953e-05, + "loss": 2.4635923385620115, + "step": 65940 + }, + { + "epoch": 0.17955628147543215, + "grad_norm": 2.140270471572876, + "learning_rate": 1.6431293444187e-05, + "loss": 2.5177705764770506, + "step": 65960 + }, + { + "epoch": 0.17961072546617668, + "grad_norm": 2.084984302520752, + "learning_rate": 1.6430203080278044e-05, + "loss": 2.529393768310547, + "step": 65980 + }, + { + "epoch": 0.17966516945692118, + "grad_norm": 2.0450212955474854, + "learning_rate": 1.642911271636909e-05, + "loss": 2.531237030029297, + "step": 66000 + }, + { + "epoch": 0.1797196134476657, + "grad_norm": 2.6408333778381348, + "learning_rate": 1.6428022352460136e-05, + "loss": 2.542190170288086, + "step": 66020 + }, + { + "epoch": 0.17977405743841024, + "grad_norm": 2.187191963195801, + "learning_rate": 1.642693198855118e-05, + "loss": 2.546938896179199, + "step": 66040 + }, + { + "epoch": 0.17982850142915477, + "grad_norm": 2.340501308441162, + "learning_rate": 1.6425841624642224e-05, + "loss": 2.531366539001465, + "step": 66060 + }, + { + "epoch": 0.17988294541989927, + "grad_norm": 2.3927862644195557, + "learning_rate": 1.642475126073327e-05, + "loss": 2.5309276580810547, + "step": 66080 + }, + { + "epoch": 0.1799373894106438, + "grad_norm": 2.1839969158172607, + "learning_rate": 1.6423660896824316e-05, + "loss": 2.5277477264404298, + "step": 66100 + }, + { + "epoch": 0.17999183340138833, + "grad_norm": 2.174253463745117, + "learning_rate": 1.6422570532915362e-05, + "loss": 2.518735885620117, + "step": 66120 + }, + { + "epoch": 0.18004627739213283, + "grad_norm": 2.0519540309906006, + "learning_rate": 1.6421480169006408e-05, + "loss": 2.5184722900390626, + "step": 66140 + }, + { + "epoch": 0.18010072138287736, + "grad_norm": 2.108639717102051, + "learning_rate": 1.6420389805097453e-05, + "loss": 2.5099740982055665, + "step": 66160 + }, + { + "epoch": 0.1801551653736219, + "grad_norm": 2.113011121749878, + "learning_rate": 1.64192994411885e-05, + "loss": 2.513793182373047, + "step": 66180 + }, + { + "epoch": 0.18020960936436642, + "grad_norm": 2.1546337604522705, + "learning_rate": 1.641820907727954e-05, + "loss": 2.5225629806518555, + "step": 66200 + }, + { + "epoch": 0.18026405335511092, + "grad_norm": 2.1093857288360596, + "learning_rate": 1.6417118713370587e-05, + "loss": 2.545707130432129, + "step": 66220 + }, + { + "epoch": 0.18031849734585545, + "grad_norm": 2.1554770469665527, + "learning_rate": 1.6416028349461633e-05, + "loss": 2.5324411392211914, + "step": 66240 + }, + { + "epoch": 0.18037294133659998, + "grad_norm": 2.2486183643341064, + "learning_rate": 1.641493798555268e-05, + "loss": 2.5599912643432616, + "step": 66260 + }, + { + "epoch": 0.18042738532734448, + "grad_norm": 2.1491377353668213, + "learning_rate": 1.6413847621643725e-05, + "loss": 2.5132719039916993, + "step": 66280 + }, + { + "epoch": 0.180481829318089, + "grad_norm": 3.3256120681762695, + "learning_rate": 1.641275725773477e-05, + "loss": 2.5153871536254884, + "step": 66300 + }, + { + "epoch": 0.18053627330883354, + "grad_norm": 2.1295218467712402, + "learning_rate": 1.6411666893825816e-05, + "loss": 2.5180938720703123, + "step": 66320 + }, + { + "epoch": 0.18059071729957807, + "grad_norm": 2.4064104557037354, + "learning_rate": 1.6410576529916862e-05, + "loss": 2.533366584777832, + "step": 66340 + }, + { + "epoch": 0.18064516129032257, + "grad_norm": 2.1402535438537598, + "learning_rate": 1.6409486166007905e-05, + "loss": 2.52978630065918, + "step": 66360 + }, + { + "epoch": 0.1806996052810671, + "grad_norm": 2.1426031589508057, + "learning_rate": 1.640839580209895e-05, + "loss": 2.5298225402832033, + "step": 66380 + }, + { + "epoch": 0.18075404927181163, + "grad_norm": 2.0780797004699707, + "learning_rate": 1.6407305438189996e-05, + "loss": 2.508754539489746, + "step": 66400 + }, + { + "epoch": 0.18080849326255616, + "grad_norm": 2.178107738494873, + "learning_rate": 1.6406215074281042e-05, + "loss": 2.5085872650146483, + "step": 66420 + }, + { + "epoch": 0.18086293725330066, + "grad_norm": 2.1936874389648438, + "learning_rate": 1.6405124710372088e-05, + "loss": 2.499499702453613, + "step": 66440 + }, + { + "epoch": 0.1809173812440452, + "grad_norm": 2.326643466949463, + "learning_rate": 1.6404034346463134e-05, + "loss": 2.540932846069336, + "step": 66460 + }, + { + "epoch": 0.18097182523478972, + "grad_norm": 2.1827731132507324, + "learning_rate": 1.640294398255418e-05, + "loss": 2.495469856262207, + "step": 66480 + }, + { + "epoch": 0.18102626922553422, + "grad_norm": 2.167757272720337, + "learning_rate": 1.6401853618645225e-05, + "loss": 2.5244226455688477, + "step": 66500 + }, + { + "epoch": 0.18108071321627875, + "grad_norm": 2.1898090839385986, + "learning_rate": 1.6400763254736268e-05, + "loss": 2.5514404296875, + "step": 66520 + }, + { + "epoch": 0.18113515720702328, + "grad_norm": 2.129925012588501, + "learning_rate": 1.6399672890827314e-05, + "loss": 2.5771696090698244, + "step": 66540 + }, + { + "epoch": 0.1811896011977678, + "grad_norm": 2.1205122470855713, + "learning_rate": 1.639858252691836e-05, + "loss": 2.5099746704101564, + "step": 66560 + }, + { + "epoch": 0.1812440451885123, + "grad_norm": 2.1361498832702637, + "learning_rate": 1.6397492163009405e-05, + "loss": 2.5239336013793947, + "step": 66580 + }, + { + "epoch": 0.18129848917925684, + "grad_norm": 2.1784520149230957, + "learning_rate": 1.639640179910045e-05, + "loss": 2.524732780456543, + "step": 66600 + }, + { + "epoch": 0.18135293317000137, + "grad_norm": 2.115117073059082, + "learning_rate": 1.6395311435191497e-05, + "loss": 2.5059326171875, + "step": 66620 + }, + { + "epoch": 0.18140737716074587, + "grad_norm": 2.3083317279815674, + "learning_rate": 1.6394221071282543e-05, + "loss": 2.5300716400146483, + "step": 66640 + }, + { + "epoch": 0.1814618211514904, + "grad_norm": 2.198741912841797, + "learning_rate": 1.639313070737359e-05, + "loss": 2.522001266479492, + "step": 66660 + }, + { + "epoch": 0.18151626514223493, + "grad_norm": 2.0576207637786865, + "learning_rate": 1.639204034346463e-05, + "loss": 2.521762466430664, + "step": 66680 + }, + { + "epoch": 0.18157070913297946, + "grad_norm": 2.0679309368133545, + "learning_rate": 1.6390949979555677e-05, + "loss": 2.51778621673584, + "step": 66700 + }, + { + "epoch": 0.18162515312372396, + "grad_norm": 2.1649317741394043, + "learning_rate": 1.6389859615646722e-05, + "loss": 2.496381950378418, + "step": 66720 + }, + { + "epoch": 0.1816795971144685, + "grad_norm": 2.0893163681030273, + "learning_rate": 1.6388769251737768e-05, + "loss": 2.53619499206543, + "step": 66740 + }, + { + "epoch": 0.18173404110521302, + "grad_norm": 2.1422674655914307, + "learning_rate": 1.6387678887828814e-05, + "loss": 2.508144760131836, + "step": 66760 + }, + { + "epoch": 0.18178848509595755, + "grad_norm": 2.2214016914367676, + "learning_rate": 1.638658852391986e-05, + "loss": 2.5248151779174806, + "step": 66780 + }, + { + "epoch": 0.18184292908670205, + "grad_norm": 2.2869515419006348, + "learning_rate": 1.6385498160010906e-05, + "loss": 2.520731735229492, + "step": 66800 + }, + { + "epoch": 0.18189737307744658, + "grad_norm": 2.2001142501831055, + "learning_rate": 1.638440779610195e-05, + "loss": 2.527873992919922, + "step": 66820 + }, + { + "epoch": 0.1819518170681911, + "grad_norm": 2.075002908706665, + "learning_rate": 1.6383317432192994e-05, + "loss": 2.5126926422119142, + "step": 66840 + }, + { + "epoch": 0.1820062610589356, + "grad_norm": 2.2792391777038574, + "learning_rate": 1.638222706828404e-05, + "loss": 2.5373085021972654, + "step": 66860 + }, + { + "epoch": 0.18206070504968014, + "grad_norm": 2.1380105018615723, + "learning_rate": 1.6381136704375086e-05, + "loss": 2.5470481872558595, + "step": 66880 + }, + { + "epoch": 0.18211514904042467, + "grad_norm": 2.143643379211426, + "learning_rate": 1.638004634046613e-05, + "loss": 2.5478147506713866, + "step": 66900 + }, + { + "epoch": 0.1821695930311692, + "grad_norm": 2.228086233139038, + "learning_rate": 1.6378955976557177e-05, + "loss": 2.4827714920043946, + "step": 66920 + }, + { + "epoch": 0.1822240370219137, + "grad_norm": 2.1191444396972656, + "learning_rate": 1.6377865612648223e-05, + "loss": 2.5211790084838865, + "step": 66940 + }, + { + "epoch": 0.18227848101265823, + "grad_norm": 2.0661332607269287, + "learning_rate": 1.637677524873927e-05, + "loss": 2.5114864349365233, + "step": 66960 + }, + { + "epoch": 0.18233292500340276, + "grad_norm": 2.16839861869812, + "learning_rate": 1.6375684884830315e-05, + "loss": 2.508823013305664, + "step": 66980 + }, + { + "epoch": 0.18238736899414726, + "grad_norm": 2.1601359844207764, + "learning_rate": 1.6374594520921357e-05, + "loss": 2.5158227920532226, + "step": 67000 + }, + { + "epoch": 0.1824418129848918, + "grad_norm": 2.5151801109313965, + "learning_rate": 1.6373504157012403e-05, + "loss": 2.5640565872192385, + "step": 67020 + }, + { + "epoch": 0.18249625697563632, + "grad_norm": 2.1672439575195312, + "learning_rate": 1.637241379310345e-05, + "loss": 2.530617904663086, + "step": 67040 + }, + { + "epoch": 0.18255070096638085, + "grad_norm": 2.1267619132995605, + "learning_rate": 1.6371323429194494e-05, + "loss": 2.5444721221923827, + "step": 67060 + }, + { + "epoch": 0.18260514495712535, + "grad_norm": 2.2549777030944824, + "learning_rate": 1.637023306528554e-05, + "loss": 2.51273136138916, + "step": 67080 + }, + { + "epoch": 0.18265958894786988, + "grad_norm": 2.161698341369629, + "learning_rate": 1.6369142701376586e-05, + "loss": 2.4949432373046876, + "step": 67100 + }, + { + "epoch": 0.1827140329386144, + "grad_norm": 2.0837254524230957, + "learning_rate": 1.6368052337467632e-05, + "loss": 2.5275386810302733, + "step": 67120 + }, + { + "epoch": 0.1827684769293589, + "grad_norm": 2.1153342723846436, + "learning_rate": 1.6366961973558678e-05, + "loss": 2.5432682037353516, + "step": 67140 + }, + { + "epoch": 0.18282292092010344, + "grad_norm": 2.1066157817840576, + "learning_rate": 1.636587160964972e-05, + "loss": 2.5246458053588867, + "step": 67160 + }, + { + "epoch": 0.18287736491084797, + "grad_norm": 2.198275089263916, + "learning_rate": 1.636478124574077e-05, + "loss": 2.5097835540771483, + "step": 67180 + }, + { + "epoch": 0.1829318089015925, + "grad_norm": 2.235485076904297, + "learning_rate": 1.6363690881831812e-05, + "loss": 2.5271224975585938, + "step": 67200 + }, + { + "epoch": 0.182986252892337, + "grad_norm": 2.130725622177124, + "learning_rate": 1.6362600517922858e-05, + "loss": 2.558103561401367, + "step": 67220 + }, + { + "epoch": 0.18304069688308153, + "grad_norm": 2.352665424346924, + "learning_rate": 1.6361510154013903e-05, + "loss": 2.52242431640625, + "step": 67240 + }, + { + "epoch": 0.18309514087382606, + "grad_norm": 2.0746099948883057, + "learning_rate": 1.636041979010495e-05, + "loss": 2.5183067321777344, + "step": 67260 + }, + { + "epoch": 0.18314958486457059, + "grad_norm": 2.1039087772369385, + "learning_rate": 1.6359329426195995e-05, + "loss": 2.5266960144042967, + "step": 67280 + }, + { + "epoch": 0.1832040288553151, + "grad_norm": 2.1690800189971924, + "learning_rate": 1.635823906228704e-05, + "loss": 2.494482231140137, + "step": 67300 + }, + { + "epoch": 0.18325847284605962, + "grad_norm": 2.169593572616577, + "learning_rate": 1.6357148698378083e-05, + "loss": 2.5412948608398436, + "step": 67320 + }, + { + "epoch": 0.18331291683680415, + "grad_norm": 2.133121967315674, + "learning_rate": 1.6356058334469132e-05, + "loss": 2.543193244934082, + "step": 67340 + }, + { + "epoch": 0.18336736082754865, + "grad_norm": 2.139943838119507, + "learning_rate": 1.6354967970560175e-05, + "loss": 2.5199661254882812, + "step": 67360 + }, + { + "epoch": 0.18342180481829318, + "grad_norm": 2.0956923961639404, + "learning_rate": 1.635387760665122e-05, + "loss": 2.5407447814941406, + "step": 67380 + }, + { + "epoch": 0.1834762488090377, + "grad_norm": 2.2543811798095703, + "learning_rate": 1.6352787242742266e-05, + "loss": 2.5070564270019533, + "step": 67400 + }, + { + "epoch": 0.18353069279978224, + "grad_norm": 2.2327303886413574, + "learning_rate": 1.6351696878833312e-05, + "loss": 2.5394716262817383, + "step": 67420 + }, + { + "epoch": 0.18358513679052674, + "grad_norm": 2.1111581325531006, + "learning_rate": 1.6350606514924358e-05, + "loss": 2.542172431945801, + "step": 67440 + }, + { + "epoch": 0.18363958078127127, + "grad_norm": 2.2070653438568115, + "learning_rate": 1.6349516151015404e-05, + "loss": 2.50699405670166, + "step": 67460 + }, + { + "epoch": 0.1836940247720158, + "grad_norm": 2.008174419403076, + "learning_rate": 1.6348425787106446e-05, + "loss": 2.543416213989258, + "step": 67480 + }, + { + "epoch": 0.1837484687627603, + "grad_norm": 2.2757997512817383, + "learning_rate": 1.6347335423197495e-05, + "loss": 2.5161293029785154, + "step": 67500 + }, + { + "epoch": 0.18380291275350483, + "grad_norm": 2.1905126571655273, + "learning_rate": 1.6346245059288538e-05, + "loss": 2.5206487655639647, + "step": 67520 + }, + { + "epoch": 0.18385735674424936, + "grad_norm": 2.245192050933838, + "learning_rate": 1.6345154695379584e-05, + "loss": 2.4999080657958985, + "step": 67540 + }, + { + "epoch": 0.18391180073499389, + "grad_norm": 2.0324642658233643, + "learning_rate": 1.634406433147063e-05, + "loss": 2.5255897521972654, + "step": 67560 + }, + { + "epoch": 0.1839662447257384, + "grad_norm": 2.1643614768981934, + "learning_rate": 1.6342973967561675e-05, + "loss": 2.503865432739258, + "step": 67580 + }, + { + "epoch": 0.18402068871648292, + "grad_norm": 2.142939567565918, + "learning_rate": 1.634188360365272e-05, + "loss": 2.5275989532470704, + "step": 67600 + }, + { + "epoch": 0.18407513270722745, + "grad_norm": 2.2328124046325684, + "learning_rate": 1.6340793239743767e-05, + "loss": 2.528621864318848, + "step": 67620 + }, + { + "epoch": 0.18412957669797198, + "grad_norm": 2.1662135124206543, + "learning_rate": 1.633970287583481e-05, + "loss": 2.531506156921387, + "step": 67640 + }, + { + "epoch": 0.18418402068871648, + "grad_norm": 2.128754138946533, + "learning_rate": 1.633861251192586e-05, + "loss": 2.5348249435424806, + "step": 67660 + }, + { + "epoch": 0.184238464679461, + "grad_norm": 2.08144211769104, + "learning_rate": 1.63375221480169e-05, + "loss": 2.5138416290283203, + "step": 67680 + }, + { + "epoch": 0.18429290867020554, + "grad_norm": 2.0361998081207275, + "learning_rate": 1.6336431784107947e-05, + "loss": 2.515256881713867, + "step": 67700 + }, + { + "epoch": 0.18434735266095004, + "grad_norm": 2.2190425395965576, + "learning_rate": 1.6335341420198993e-05, + "loss": 2.5459291458129885, + "step": 67720 + }, + { + "epoch": 0.18440179665169457, + "grad_norm": 2.1779613494873047, + "learning_rate": 1.633425105629004e-05, + "loss": 2.5015148162841796, + "step": 67740 + }, + { + "epoch": 0.1844562406424391, + "grad_norm": 2.2372210025787354, + "learning_rate": 1.6333160692381084e-05, + "loss": 2.5246057510375977, + "step": 67760 + }, + { + "epoch": 0.18451068463318362, + "grad_norm": 2.1238503456115723, + "learning_rate": 1.633207032847213e-05, + "loss": 2.5232934951782227, + "step": 67780 + }, + { + "epoch": 0.18456512862392813, + "grad_norm": 2.083937168121338, + "learning_rate": 1.6330979964563172e-05, + "loss": 2.5562347412109374, + "step": 67800 + }, + { + "epoch": 0.18461957261467266, + "grad_norm": 2.096956253051758, + "learning_rate": 1.632988960065422e-05, + "loss": 2.5089101791381836, + "step": 67820 + }, + { + "epoch": 0.18467401660541718, + "grad_norm": 2.1923232078552246, + "learning_rate": 1.6328799236745264e-05, + "loss": 2.497927665710449, + "step": 67840 + }, + { + "epoch": 0.1847284605961617, + "grad_norm": 2.1215970516204834, + "learning_rate": 1.632770887283631e-05, + "loss": 2.5318702697753905, + "step": 67860 + }, + { + "epoch": 0.18478290458690622, + "grad_norm": 2.233356475830078, + "learning_rate": 1.6326618508927356e-05, + "loss": 2.5275384902954103, + "step": 67880 + }, + { + "epoch": 0.18483734857765075, + "grad_norm": 2.0641489028930664, + "learning_rate": 1.63255281450184e-05, + "loss": 2.532693862915039, + "step": 67900 + }, + { + "epoch": 0.18489179256839527, + "grad_norm": 2.5751092433929443, + "learning_rate": 1.6324437781109447e-05, + "loss": 2.505321502685547, + "step": 67920 + }, + { + "epoch": 0.18494623655913978, + "grad_norm": 2.0459625720977783, + "learning_rate": 1.6323347417200493e-05, + "loss": 2.5145296096801757, + "step": 67940 + }, + { + "epoch": 0.1850006805498843, + "grad_norm": 2.1430938243865967, + "learning_rate": 1.6322257053291536e-05, + "loss": 2.5181400299072267, + "step": 67960 + }, + { + "epoch": 0.18505512454062883, + "grad_norm": 2.143480062484741, + "learning_rate": 1.6321166689382585e-05, + "loss": 2.514093589782715, + "step": 67980 + }, + { + "epoch": 0.18510956853137334, + "grad_norm": 2.276453971862793, + "learning_rate": 1.6320076325473627e-05, + "loss": 2.529545211791992, + "step": 68000 + }, + { + "epoch": 0.18516401252211787, + "grad_norm": 2.135251045227051, + "learning_rate": 1.6318985961564673e-05, + "loss": 2.531406784057617, + "step": 68020 + }, + { + "epoch": 0.1852184565128624, + "grad_norm": 2.0906283855438232, + "learning_rate": 1.631789559765572e-05, + "loss": 2.547796440124512, + "step": 68040 + }, + { + "epoch": 0.18527290050360692, + "grad_norm": 2.1069424152374268, + "learning_rate": 1.6316805233746765e-05, + "loss": 2.523092269897461, + "step": 68060 + }, + { + "epoch": 0.18532734449435143, + "grad_norm": 2.29952335357666, + "learning_rate": 1.631571486983781e-05, + "loss": 2.5237070083618165, + "step": 68080 + }, + { + "epoch": 0.18538178848509596, + "grad_norm": 2.1076297760009766, + "learning_rate": 1.6314624505928856e-05, + "loss": 2.5422929763793944, + "step": 68100 + }, + { + "epoch": 0.18543623247584048, + "grad_norm": 2.082083225250244, + "learning_rate": 1.63135341420199e-05, + "loss": 2.5443450927734377, + "step": 68120 + }, + { + "epoch": 0.185490676466585, + "grad_norm": 2.214805841445923, + "learning_rate": 1.6312443778110948e-05, + "loss": 2.521524429321289, + "step": 68140 + }, + { + "epoch": 0.18554512045732952, + "grad_norm": 2.158055305480957, + "learning_rate": 1.631135341420199e-05, + "loss": 2.5119808197021483, + "step": 68160 + }, + { + "epoch": 0.18559956444807404, + "grad_norm": 2.7857651710510254, + "learning_rate": 1.6310263050293036e-05, + "loss": 2.518649673461914, + "step": 68180 + }, + { + "epoch": 0.18565400843881857, + "grad_norm": 2.431300163269043, + "learning_rate": 1.6309172686384082e-05, + "loss": 2.52249698638916, + "step": 68200 + }, + { + "epoch": 0.18570845242956308, + "grad_norm": 2.240355968475342, + "learning_rate": 1.6308082322475128e-05, + "loss": 2.5124244689941406, + "step": 68220 + }, + { + "epoch": 0.1857628964203076, + "grad_norm": 2.1849029064178467, + "learning_rate": 1.630699195856617e-05, + "loss": 2.5643890380859373, + "step": 68240 + }, + { + "epoch": 0.18581734041105213, + "grad_norm": 2.1234734058380127, + "learning_rate": 1.630590159465722e-05, + "loss": 2.5058757781982424, + "step": 68260 + }, + { + "epoch": 0.18587178440179666, + "grad_norm": 2.1058664321899414, + "learning_rate": 1.6304811230748262e-05, + "loss": 2.526914596557617, + "step": 68280 + }, + { + "epoch": 0.18592622839254117, + "grad_norm": 2.121602773666382, + "learning_rate": 1.630372086683931e-05, + "loss": 2.5465682983398437, + "step": 68300 + }, + { + "epoch": 0.1859806723832857, + "grad_norm": 2.1972742080688477, + "learning_rate": 1.6302630502930353e-05, + "loss": 2.5303714752197264, + "step": 68320 + }, + { + "epoch": 0.18603511637403022, + "grad_norm": 2.3225667476654053, + "learning_rate": 1.63015401390214e-05, + "loss": 2.5100852966308596, + "step": 68340 + }, + { + "epoch": 0.18608956036477473, + "grad_norm": 2.1360316276550293, + "learning_rate": 1.6300449775112445e-05, + "loss": 2.53985595703125, + "step": 68360 + }, + { + "epoch": 0.18614400435551925, + "grad_norm": 2.039764642715454, + "learning_rate": 1.629935941120349e-05, + "loss": 2.503043365478516, + "step": 68380 + }, + { + "epoch": 0.18619844834626378, + "grad_norm": 2.23779296875, + "learning_rate": 1.6298269047294533e-05, + "loss": 2.5537790298461913, + "step": 68400 + }, + { + "epoch": 0.1862528923370083, + "grad_norm": 2.160511016845703, + "learning_rate": 1.6297178683385582e-05, + "loss": 2.5587963104248046, + "step": 68420 + }, + { + "epoch": 0.18630733632775281, + "grad_norm": 2.2693729400634766, + "learning_rate": 1.6296088319476625e-05, + "loss": 2.5185165405273438, + "step": 68440 + }, + { + "epoch": 0.18636178031849734, + "grad_norm": 2.2417314052581787, + "learning_rate": 1.6294997955567674e-05, + "loss": 2.488820457458496, + "step": 68460 + }, + { + "epoch": 0.18641622430924187, + "grad_norm": 2.0563876628875732, + "learning_rate": 1.6293907591658716e-05, + "loss": 2.52825927734375, + "step": 68480 + }, + { + "epoch": 0.18647066829998638, + "grad_norm": 2.086116313934326, + "learning_rate": 1.6292817227749762e-05, + "loss": 2.506801414489746, + "step": 68500 + }, + { + "epoch": 0.1865251122907309, + "grad_norm": 5.817708492279053, + "learning_rate": 1.6291726863840808e-05, + "loss": 2.5373310089111327, + "step": 68520 + }, + { + "epoch": 0.18657955628147543, + "grad_norm": 2.109473943710327, + "learning_rate": 1.6290636499931854e-05, + "loss": 2.525677299499512, + "step": 68540 + }, + { + "epoch": 0.18663400027221996, + "grad_norm": 2.1041717529296875, + "learning_rate": 1.6289546136022896e-05, + "loss": 2.4899728775024412, + "step": 68560 + }, + { + "epoch": 0.18668844426296446, + "grad_norm": 2.1374478340148926, + "learning_rate": 1.6288455772113945e-05, + "loss": 2.536569595336914, + "step": 68580 + }, + { + "epoch": 0.186742888253709, + "grad_norm": 2.1550533771514893, + "learning_rate": 1.6287365408204988e-05, + "loss": 2.5713661193847654, + "step": 68600 + }, + { + "epoch": 0.18679733224445352, + "grad_norm": 2.04608154296875, + "learning_rate": 1.6286275044296037e-05, + "loss": 2.5248996734619142, + "step": 68620 + }, + { + "epoch": 0.18685177623519805, + "grad_norm": 2.2909140586853027, + "learning_rate": 1.628518468038708e-05, + "loss": 2.5162981033325194, + "step": 68640 + }, + { + "epoch": 0.18690622022594255, + "grad_norm": 2.224259614944458, + "learning_rate": 1.6284094316478125e-05, + "loss": 2.5307968139648436, + "step": 68660 + }, + { + "epoch": 0.18696066421668708, + "grad_norm": 2.096409320831299, + "learning_rate": 1.628300395256917e-05, + "loss": 2.570890426635742, + "step": 68680 + }, + { + "epoch": 0.1870151082074316, + "grad_norm": 2.081411838531494, + "learning_rate": 1.6281913588660217e-05, + "loss": 2.5488183975219725, + "step": 68700 + }, + { + "epoch": 0.18706955219817611, + "grad_norm": 2.1466071605682373, + "learning_rate": 1.628082322475126e-05, + "loss": 2.559827423095703, + "step": 68720 + }, + { + "epoch": 0.18712399618892064, + "grad_norm": 2.2841780185699463, + "learning_rate": 1.627973286084231e-05, + "loss": 2.5398941040039062, + "step": 68740 + }, + { + "epoch": 0.18717844017966517, + "grad_norm": 2.0929503440856934, + "learning_rate": 1.627864249693335e-05, + "loss": 2.5483724594116213, + "step": 68760 + }, + { + "epoch": 0.1872328841704097, + "grad_norm": 2.084181070327759, + "learning_rate": 1.62775521330244e-05, + "loss": 2.5195295333862306, + "step": 68780 + }, + { + "epoch": 0.1872873281611542, + "grad_norm": 2.1957998275756836, + "learning_rate": 1.6276461769115443e-05, + "loss": 2.536461067199707, + "step": 68800 + }, + { + "epoch": 0.18734177215189873, + "grad_norm": 2.4494059085845947, + "learning_rate": 1.627537140520649e-05, + "loss": 2.5305339813232424, + "step": 68820 + }, + { + "epoch": 0.18739621614264326, + "grad_norm": 2.2613558769226074, + "learning_rate": 1.6274281041297534e-05, + "loss": 2.5362388610839846, + "step": 68840 + }, + { + "epoch": 0.18745066013338776, + "grad_norm": 2.132770299911499, + "learning_rate": 1.627319067738858e-05, + "loss": 2.52332763671875, + "step": 68860 + }, + { + "epoch": 0.1875051041241323, + "grad_norm": 2.1556692123413086, + "learning_rate": 1.6272100313479626e-05, + "loss": 2.5265148162841795, + "step": 68880 + }, + { + "epoch": 0.18755954811487682, + "grad_norm": 2.1962082386016846, + "learning_rate": 1.627100994957067e-05, + "loss": 2.5365055084228514, + "step": 68900 + }, + { + "epoch": 0.18761399210562135, + "grad_norm": 2.645556926727295, + "learning_rate": 1.6269919585661714e-05, + "loss": 2.5532062530517576, + "step": 68920 + }, + { + "epoch": 0.18766843609636585, + "grad_norm": 2.1896228790283203, + "learning_rate": 1.6268829221752763e-05, + "loss": 2.52176513671875, + "step": 68940 + }, + { + "epoch": 0.18772288008711038, + "grad_norm": 2.3402693271636963, + "learning_rate": 1.6267738857843806e-05, + "loss": 2.517130661010742, + "step": 68960 + }, + { + "epoch": 0.1877773240778549, + "grad_norm": 2.1146254539489746, + "learning_rate": 1.626664849393485e-05, + "loss": 2.522248077392578, + "step": 68980 + }, + { + "epoch": 0.18783176806859944, + "grad_norm": 2.0980284214019775, + "learning_rate": 1.6265558130025897e-05, + "loss": 2.5304729461669924, + "step": 69000 + }, + { + "epoch": 0.18788621205934394, + "grad_norm": 2.205868721008301, + "learning_rate": 1.6264467766116943e-05, + "loss": 2.5200244903564455, + "step": 69020 + }, + { + "epoch": 0.18794065605008847, + "grad_norm": 2.126614809036255, + "learning_rate": 1.626337740220799e-05, + "loss": 2.528560829162598, + "step": 69040 + }, + { + "epoch": 0.187995100040833, + "grad_norm": 2.1935625076293945, + "learning_rate": 1.6262287038299035e-05, + "loss": 2.5218509674072265, + "step": 69060 + }, + { + "epoch": 0.1880495440315775, + "grad_norm": 2.3146169185638428, + "learning_rate": 1.6261196674390077e-05, + "loss": 2.5344518661499023, + "step": 69080 + }, + { + "epoch": 0.18810398802232203, + "grad_norm": 2.1765806674957275, + "learning_rate": 1.6260106310481126e-05, + "loss": 2.5622230529785157, + "step": 69100 + }, + { + "epoch": 0.18815843201306656, + "grad_norm": 2.198824644088745, + "learning_rate": 1.625901594657217e-05, + "loss": 2.512874221801758, + "step": 69120 + }, + { + "epoch": 0.1882128760038111, + "grad_norm": 3.7982828617095947, + "learning_rate": 1.6257925582663215e-05, + "loss": 2.5277666091918944, + "step": 69140 + }, + { + "epoch": 0.1882673199945556, + "grad_norm": 2.053093194961548, + "learning_rate": 1.625683521875426e-05, + "loss": 2.5205827713012696, + "step": 69160 + }, + { + "epoch": 0.18832176398530012, + "grad_norm": 2.2323665618896484, + "learning_rate": 1.6255744854845306e-05, + "loss": 2.520511817932129, + "step": 69180 + }, + { + "epoch": 0.18837620797604465, + "grad_norm": 2.248577833175659, + "learning_rate": 1.6254654490936352e-05, + "loss": 2.524465560913086, + "step": 69200 + }, + { + "epoch": 0.18843065196678915, + "grad_norm": 2.087261915206909, + "learning_rate": 1.6253564127027398e-05, + "loss": 2.5399944305419924, + "step": 69220 + }, + { + "epoch": 0.18848509595753368, + "grad_norm": 2.133530616760254, + "learning_rate": 1.625247376311844e-05, + "loss": 2.5067636489868166, + "step": 69240 + }, + { + "epoch": 0.1885395399482782, + "grad_norm": 2.100566864013672, + "learning_rate": 1.625138339920949e-05, + "loss": 2.534831237792969, + "step": 69260 + }, + { + "epoch": 0.18859398393902274, + "grad_norm": 2.0860376358032227, + "learning_rate": 1.6250293035300532e-05, + "loss": 2.5084659576416017, + "step": 69280 + }, + { + "epoch": 0.18864842792976724, + "grad_norm": 2.206327438354492, + "learning_rate": 1.6249202671391578e-05, + "loss": 2.532883071899414, + "step": 69300 + }, + { + "epoch": 0.18870287192051177, + "grad_norm": 2.0385918617248535, + "learning_rate": 1.6248112307482623e-05, + "loss": 2.5111522674560547, + "step": 69320 + }, + { + "epoch": 0.1887573159112563, + "grad_norm": 2.145409107208252, + "learning_rate": 1.624702194357367e-05, + "loss": 2.526347351074219, + "step": 69340 + }, + { + "epoch": 0.1888117599020008, + "grad_norm": 3.1044039726257324, + "learning_rate": 1.6245931579664715e-05, + "loss": 2.5269968032836916, + "step": 69360 + }, + { + "epoch": 0.18886620389274533, + "grad_norm": 2.3976569175720215, + "learning_rate": 1.624484121575576e-05, + "loss": 2.5072675704956056, + "step": 69380 + }, + { + "epoch": 0.18892064788348986, + "grad_norm": 2.2002274990081787, + "learning_rate": 1.6243750851846803e-05, + "loss": 2.540605354309082, + "step": 69400 + }, + { + "epoch": 0.1889750918742344, + "grad_norm": 2.578864574432373, + "learning_rate": 1.6242660487937853e-05, + "loss": 2.5217151641845703, + "step": 69420 + }, + { + "epoch": 0.1890295358649789, + "grad_norm": 2.105130910873413, + "learning_rate": 1.6241570124028895e-05, + "loss": 2.550076484680176, + "step": 69440 + }, + { + "epoch": 0.18908397985572342, + "grad_norm": 2.0327231884002686, + "learning_rate": 1.624047976011994e-05, + "loss": 2.5467390060424804, + "step": 69460 + }, + { + "epoch": 0.18913842384646795, + "grad_norm": 4.923985958099365, + "learning_rate": 1.6239389396210987e-05, + "loss": 2.5095565795898436, + "step": 69480 + }, + { + "epoch": 0.18919286783721248, + "grad_norm": 2.155526876449585, + "learning_rate": 1.6238299032302032e-05, + "loss": 2.4943048477172853, + "step": 69500 + }, + { + "epoch": 0.18924731182795698, + "grad_norm": 2.1458377838134766, + "learning_rate": 1.6237208668393078e-05, + "loss": 2.539040946960449, + "step": 69520 + }, + { + "epoch": 0.1893017558187015, + "grad_norm": 2.2880899906158447, + "learning_rate": 1.6236118304484124e-05, + "loss": 2.507915496826172, + "step": 69540 + }, + { + "epoch": 0.18935619980944604, + "grad_norm": 2.358097553253174, + "learning_rate": 1.6235027940575166e-05, + "loss": 2.507745552062988, + "step": 69560 + }, + { + "epoch": 0.18941064380019054, + "grad_norm": 2.105764150619507, + "learning_rate": 1.6233937576666216e-05, + "loss": 2.487460899353027, + "step": 69580 + }, + { + "epoch": 0.18946508779093507, + "grad_norm": 2.2492799758911133, + "learning_rate": 1.6232847212757258e-05, + "loss": 2.539149856567383, + "step": 69600 + }, + { + "epoch": 0.1895195317816796, + "grad_norm": 2.1829702854156494, + "learning_rate": 1.6231756848848304e-05, + "loss": 2.523136520385742, + "step": 69620 + }, + { + "epoch": 0.18957397577242413, + "grad_norm": 2.0865862369537354, + "learning_rate": 1.623066648493935e-05, + "loss": 2.5608469009399415, + "step": 69640 + }, + { + "epoch": 0.18962841976316863, + "grad_norm": 2.162458896636963, + "learning_rate": 1.6229576121030395e-05, + "loss": 2.5216142654418947, + "step": 69660 + }, + { + "epoch": 0.18968286375391316, + "grad_norm": 2.1031782627105713, + "learning_rate": 1.622848575712144e-05, + "loss": 2.512778472900391, + "step": 69680 + }, + { + "epoch": 0.1897373077446577, + "grad_norm": 2.0911946296691895, + "learning_rate": 1.6227395393212487e-05, + "loss": 2.562025260925293, + "step": 69700 + }, + { + "epoch": 0.1897917517354022, + "grad_norm": 2.112081527709961, + "learning_rate": 1.622630502930353e-05, + "loss": 2.5113222122192385, + "step": 69720 + }, + { + "epoch": 0.18984619572614672, + "grad_norm": 2.1431217193603516, + "learning_rate": 1.622521466539458e-05, + "loss": 2.5130359649658205, + "step": 69740 + }, + { + "epoch": 0.18990063971689125, + "grad_norm": 2.3996188640594482, + "learning_rate": 1.622412430148562e-05, + "loss": 2.5292015075683594, + "step": 69760 + }, + { + "epoch": 0.18995508370763578, + "grad_norm": 2.1681067943573, + "learning_rate": 1.6223033937576667e-05, + "loss": 2.5520145416259767, + "step": 69780 + }, + { + "epoch": 0.19000952769838028, + "grad_norm": 2.2459394931793213, + "learning_rate": 1.6221943573667713e-05, + "loss": 2.5449682235717774, + "step": 69800 + }, + { + "epoch": 0.1900639716891248, + "grad_norm": 2.146650552749634, + "learning_rate": 1.622085320975876e-05, + "loss": 2.524896812438965, + "step": 69820 + }, + { + "epoch": 0.19011841567986934, + "grad_norm": 3.76063871383667, + "learning_rate": 1.6219762845849804e-05, + "loss": 2.5494388580322265, + "step": 69840 + }, + { + "epoch": 0.19017285967061387, + "grad_norm": 2.064326524734497, + "learning_rate": 1.621867248194085e-05, + "loss": 2.5388599395751954, + "step": 69860 + }, + { + "epoch": 0.19022730366135837, + "grad_norm": 2.120466947555542, + "learning_rate": 1.6217582118031893e-05, + "loss": 2.5149208068847657, + "step": 69880 + }, + { + "epoch": 0.1902817476521029, + "grad_norm": 2.199180841445923, + "learning_rate": 1.6216491754122942e-05, + "loss": 2.555615234375, + "step": 69900 + }, + { + "epoch": 0.19033619164284743, + "grad_norm": 2.1527369022369385, + "learning_rate": 1.6215401390213984e-05, + "loss": 2.5320690155029295, + "step": 69920 + }, + { + "epoch": 0.19039063563359193, + "grad_norm": 2.246335506439209, + "learning_rate": 1.621431102630503e-05, + "loss": 2.5176010131835938, + "step": 69940 + }, + { + "epoch": 0.19044507962433646, + "grad_norm": 2.0648341178894043, + "learning_rate": 1.6213220662396076e-05, + "loss": 2.5067770004272463, + "step": 69960 + }, + { + "epoch": 0.190499523615081, + "grad_norm": 3.93833327293396, + "learning_rate": 1.621213029848712e-05, + "loss": 2.4775081634521485, + "step": 69980 + }, + { + "epoch": 0.19055396760582552, + "grad_norm": 2.135580539703369, + "learning_rate": 1.6211039934578167e-05, + "loss": 2.511947441101074, + "step": 70000 + }, + { + "epoch": 0.19060841159657002, + "grad_norm": 2.0327558517456055, + "learning_rate": 1.6209949570669213e-05, + "loss": 2.5032960891723635, + "step": 70020 + }, + { + "epoch": 0.19066285558731455, + "grad_norm": 2.185049295425415, + "learning_rate": 1.6208859206760256e-05, + "loss": 2.51212158203125, + "step": 70040 + }, + { + "epoch": 0.19071729957805908, + "grad_norm": 2.0915889739990234, + "learning_rate": 1.6207768842851305e-05, + "loss": 2.536433792114258, + "step": 70060 + }, + { + "epoch": 0.19077174356880358, + "grad_norm": 2.3366684913635254, + "learning_rate": 1.6206678478942347e-05, + "loss": 2.5051973342895506, + "step": 70080 + }, + { + "epoch": 0.1908261875595481, + "grad_norm": 2.128842353820801, + "learning_rate": 1.6205588115033393e-05, + "loss": 2.513718032836914, + "step": 70100 + }, + { + "epoch": 0.19088063155029264, + "grad_norm": 2.132270097732544, + "learning_rate": 1.620449775112444e-05, + "loss": 2.5171384811401367, + "step": 70120 + }, + { + "epoch": 0.19093507554103717, + "grad_norm": 2.072213888168335, + "learning_rate": 1.6203407387215485e-05, + "loss": 2.527007484436035, + "step": 70140 + }, + { + "epoch": 0.19098951953178167, + "grad_norm": 2.118497133255005, + "learning_rate": 1.620231702330653e-05, + "loss": 2.5303827285766602, + "step": 70160 + }, + { + "epoch": 0.1910439635225262, + "grad_norm": 2.101316213607788, + "learning_rate": 1.6201226659397576e-05, + "loss": 2.5135698318481445, + "step": 70180 + }, + { + "epoch": 0.19109840751327073, + "grad_norm": 2.140331268310547, + "learning_rate": 1.620013629548862e-05, + "loss": 2.538906478881836, + "step": 70200 + }, + { + "epoch": 0.19115285150401523, + "grad_norm": 2.1684584617614746, + "learning_rate": 1.6199045931579668e-05, + "loss": 2.5428464889526365, + "step": 70220 + }, + { + "epoch": 0.19120729549475976, + "grad_norm": 2.1731350421905518, + "learning_rate": 1.619795556767071e-05, + "loss": 2.5562101364135743, + "step": 70240 + }, + { + "epoch": 0.1912617394855043, + "grad_norm": 2.063603162765503, + "learning_rate": 1.6196865203761756e-05, + "loss": 2.5103384017944337, + "step": 70260 + }, + { + "epoch": 0.19131618347624882, + "grad_norm": 2.1832823753356934, + "learning_rate": 1.6195774839852802e-05, + "loss": 2.51601619720459, + "step": 70280 + }, + { + "epoch": 0.19137062746699332, + "grad_norm": 2.2126145362854004, + "learning_rate": 1.6194684475943848e-05, + "loss": 2.5132156372070313, + "step": 70300 + }, + { + "epoch": 0.19142507145773785, + "grad_norm": 2.236945629119873, + "learning_rate": 1.6193594112034894e-05, + "loss": 2.5213985443115234, + "step": 70320 + }, + { + "epoch": 0.19147951544848238, + "grad_norm": 2.004915714263916, + "learning_rate": 1.619250374812594e-05, + "loss": 2.516393280029297, + "step": 70340 + }, + { + "epoch": 0.1915339594392269, + "grad_norm": 2.118734359741211, + "learning_rate": 1.6191413384216982e-05, + "loss": 2.5054092407226562, + "step": 70360 + }, + { + "epoch": 0.1915884034299714, + "grad_norm": 2.0638046264648438, + "learning_rate": 1.619032302030803e-05, + "loss": 2.498320770263672, + "step": 70380 + }, + { + "epoch": 0.19164284742071594, + "grad_norm": 2.1239001750946045, + "learning_rate": 1.6189232656399073e-05, + "loss": 2.5299230575561524, + "step": 70400 + }, + { + "epoch": 0.19169729141146047, + "grad_norm": 2.8238673210144043, + "learning_rate": 1.618814229249012e-05, + "loss": 2.534267044067383, + "step": 70420 + }, + { + "epoch": 0.19175173540220497, + "grad_norm": 2.1380996704101562, + "learning_rate": 1.6187051928581165e-05, + "loss": 2.507009506225586, + "step": 70440 + }, + { + "epoch": 0.1918061793929495, + "grad_norm": 2.1127073764801025, + "learning_rate": 1.618596156467221e-05, + "loss": 2.5242191314697267, + "step": 70460 + }, + { + "epoch": 0.19186062338369403, + "grad_norm": 2.2920331954956055, + "learning_rate": 1.6184871200763257e-05, + "loss": 2.509250450134277, + "step": 70480 + }, + { + "epoch": 0.19191506737443856, + "grad_norm": 2.0667896270751953, + "learning_rate": 1.6183780836854303e-05, + "loss": 2.5437057495117186, + "step": 70500 + }, + { + "epoch": 0.19196951136518306, + "grad_norm": 2.224156379699707, + "learning_rate": 1.6182690472945345e-05, + "loss": 2.532526969909668, + "step": 70520 + }, + { + "epoch": 0.1920239553559276, + "grad_norm": 2.2583234310150146, + "learning_rate": 1.6181600109036394e-05, + "loss": 2.5390058517456056, + "step": 70540 + }, + { + "epoch": 0.19207839934667212, + "grad_norm": 2.11604642868042, + "learning_rate": 1.6180509745127437e-05, + "loss": 2.538857650756836, + "step": 70560 + }, + { + "epoch": 0.19213284333741662, + "grad_norm": 2.284940242767334, + "learning_rate": 1.6179419381218486e-05, + "loss": 2.497502899169922, + "step": 70580 + }, + { + "epoch": 0.19218728732816115, + "grad_norm": 2.1547470092773438, + "learning_rate": 1.6178329017309528e-05, + "loss": 2.5155912399291993, + "step": 70600 + }, + { + "epoch": 0.19224173131890568, + "grad_norm": 2.115727424621582, + "learning_rate": 1.6177238653400574e-05, + "loss": 2.5132808685302734, + "step": 70620 + }, + { + "epoch": 0.1922961753096502, + "grad_norm": 2.080122709274292, + "learning_rate": 1.617614828949162e-05, + "loss": 2.5388652801513674, + "step": 70640 + }, + { + "epoch": 0.1923506193003947, + "grad_norm": 2.0511314868927, + "learning_rate": 1.6175057925582666e-05, + "loss": 2.5003957748413086, + "step": 70660 + }, + { + "epoch": 0.19240506329113924, + "grad_norm": 2.2493510246276855, + "learning_rate": 1.6173967561673708e-05, + "loss": 2.5697898864746094, + "step": 70680 + }, + { + "epoch": 0.19245950728188377, + "grad_norm": 2.1730682849884033, + "learning_rate": 1.6172877197764757e-05, + "loss": 2.5121219635009764, + "step": 70700 + }, + { + "epoch": 0.19251395127262827, + "grad_norm": 2.0987610816955566, + "learning_rate": 1.61717868338558e-05, + "loss": 2.5031806945800783, + "step": 70720 + }, + { + "epoch": 0.1925683952633728, + "grad_norm": 2.145951271057129, + "learning_rate": 1.617069646994685e-05, + "loss": 2.52077579498291, + "step": 70740 + }, + { + "epoch": 0.19262283925411733, + "grad_norm": 2.0974133014678955, + "learning_rate": 1.616960610603789e-05, + "loss": 2.508327674865723, + "step": 70760 + }, + { + "epoch": 0.19267728324486186, + "grad_norm": 2.14070463180542, + "learning_rate": 1.6168515742128937e-05, + "loss": 2.5103851318359376, + "step": 70780 + }, + { + "epoch": 0.19273172723560636, + "grad_norm": 2.1276960372924805, + "learning_rate": 1.6167425378219983e-05, + "loss": 2.497260093688965, + "step": 70800 + }, + { + "epoch": 0.1927861712263509, + "grad_norm": 2.134361505508423, + "learning_rate": 1.616633501431103e-05, + "loss": 2.517618942260742, + "step": 70820 + }, + { + "epoch": 0.19284061521709542, + "grad_norm": 2.1337437629699707, + "learning_rate": 1.616524465040207e-05, + "loss": 2.5342884063720703, + "step": 70840 + }, + { + "epoch": 0.19289505920783995, + "grad_norm": 2.2036523818969727, + "learning_rate": 1.616415428649312e-05, + "loss": 2.550002670288086, + "step": 70860 + }, + { + "epoch": 0.19294950319858445, + "grad_norm": 2.1797499656677246, + "learning_rate": 1.6163063922584163e-05, + "loss": 2.5257238388061523, + "step": 70880 + }, + { + "epoch": 0.19300394718932898, + "grad_norm": 2.1219136714935303, + "learning_rate": 1.6161973558675212e-05, + "loss": 2.5330039978027346, + "step": 70900 + }, + { + "epoch": 0.1930583911800735, + "grad_norm": 2.377164125442505, + "learning_rate": 1.6160883194766254e-05, + "loss": 2.518006134033203, + "step": 70920 + }, + { + "epoch": 0.193112835170818, + "grad_norm": 2.1776041984558105, + "learning_rate": 1.61597928308573e-05, + "loss": 2.543589210510254, + "step": 70940 + }, + { + "epoch": 0.19316727916156254, + "grad_norm": 2.1213793754577637, + "learning_rate": 1.6158702466948346e-05, + "loss": 2.517097282409668, + "step": 70960 + }, + { + "epoch": 0.19322172315230707, + "grad_norm": 1.9956854581832886, + "learning_rate": 1.6157612103039392e-05, + "loss": 2.516242790222168, + "step": 70980 + }, + { + "epoch": 0.1932761671430516, + "grad_norm": 2.2345824241638184, + "learning_rate": 1.6156521739130434e-05, + "loss": 2.4918039321899412, + "step": 71000 + }, + { + "epoch": 0.1933306111337961, + "grad_norm": 2.205566883087158, + "learning_rate": 1.6155431375221483e-05, + "loss": 2.5059221267700194, + "step": 71020 + }, + { + "epoch": 0.19338505512454063, + "grad_norm": 2.268112897872925, + "learning_rate": 1.6154341011312526e-05, + "loss": 2.505150032043457, + "step": 71040 + }, + { + "epoch": 0.19343949911528516, + "grad_norm": 2.1810524463653564, + "learning_rate": 1.615325064740357e-05, + "loss": 2.5322687149047853, + "step": 71060 + }, + { + "epoch": 0.19349394310602966, + "grad_norm": 2.131727695465088, + "learning_rate": 1.6152160283494617e-05, + "loss": 2.5374277114868162, + "step": 71080 + }, + { + "epoch": 0.1935483870967742, + "grad_norm": 2.075683832168579, + "learning_rate": 1.6151069919585663e-05, + "loss": 2.5054956436157227, + "step": 71100 + }, + { + "epoch": 0.19360283108751872, + "grad_norm": 2.2109060287475586, + "learning_rate": 1.614997955567671e-05, + "loss": 2.5012628555297853, + "step": 71120 + }, + { + "epoch": 0.19365727507826325, + "grad_norm": 2.138535261154175, + "learning_rate": 1.6148889191767755e-05, + "loss": 2.5420835494995115, + "step": 71140 + }, + { + "epoch": 0.19371171906900775, + "grad_norm": 2.29781174659729, + "learning_rate": 1.6147798827858797e-05, + "loss": 2.5157346725463867, + "step": 71160 + }, + { + "epoch": 0.19376616305975228, + "grad_norm": 2.1472489833831787, + "learning_rate": 1.6146708463949846e-05, + "loss": 2.53594913482666, + "step": 71180 + }, + { + "epoch": 0.1938206070504968, + "grad_norm": 2.147152900695801, + "learning_rate": 1.614561810004089e-05, + "loss": 2.5280517578125, + "step": 71200 + }, + { + "epoch": 0.19387505104124134, + "grad_norm": 2.2412302494049072, + "learning_rate": 1.6144527736131935e-05, + "loss": 2.546041488647461, + "step": 71220 + }, + { + "epoch": 0.19392949503198584, + "grad_norm": 2.1812236309051514, + "learning_rate": 1.614343737222298e-05, + "loss": 2.556316375732422, + "step": 71240 + }, + { + "epoch": 0.19398393902273037, + "grad_norm": 2.060450792312622, + "learning_rate": 1.6142347008314026e-05, + "loss": 2.5318426132202148, + "step": 71260 + }, + { + "epoch": 0.1940383830134749, + "grad_norm": 2.1545870304107666, + "learning_rate": 1.6141256644405072e-05, + "loss": 2.522946357727051, + "step": 71280 + }, + { + "epoch": 0.1940928270042194, + "grad_norm": 2.0054776668548584, + "learning_rate": 1.6140166280496118e-05, + "loss": 2.5307075500488283, + "step": 71300 + }, + { + "epoch": 0.19414727099496393, + "grad_norm": 2.202615976333618, + "learning_rate": 1.613907591658716e-05, + "loss": 2.508612632751465, + "step": 71320 + }, + { + "epoch": 0.19420171498570846, + "grad_norm": 2.1331005096435547, + "learning_rate": 1.613798555267821e-05, + "loss": 2.5486860275268555, + "step": 71340 + }, + { + "epoch": 0.194256158976453, + "grad_norm": 2.145503520965576, + "learning_rate": 1.6136895188769252e-05, + "loss": 2.5307239532470702, + "step": 71360 + }, + { + "epoch": 0.1943106029671975, + "grad_norm": 2.26047420501709, + "learning_rate": 1.6135804824860298e-05, + "loss": 2.526230239868164, + "step": 71380 + }, + { + "epoch": 0.19436504695794202, + "grad_norm": 2.1099441051483154, + "learning_rate": 1.6134714460951344e-05, + "loss": 2.5418996810913086, + "step": 71400 + }, + { + "epoch": 0.19441949094868655, + "grad_norm": 2.094395399093628, + "learning_rate": 1.613362409704239e-05, + "loss": 2.5318111419677733, + "step": 71420 + }, + { + "epoch": 0.19447393493943105, + "grad_norm": 2.2066519260406494, + "learning_rate": 1.6132533733133435e-05, + "loss": 2.4992021560668944, + "step": 71440 + }, + { + "epoch": 0.19452837893017558, + "grad_norm": 2.1060750484466553, + "learning_rate": 1.613144336922448e-05, + "loss": 2.5159765243530274, + "step": 71460 + }, + { + "epoch": 0.1945828229209201, + "grad_norm": 2.1769392490386963, + "learning_rate": 1.6130353005315523e-05, + "loss": 2.5083141326904297, + "step": 71480 + }, + { + "epoch": 0.19463726691166464, + "grad_norm": 2.109954595565796, + "learning_rate": 1.6129262641406573e-05, + "loss": 2.5601335525512696, + "step": 71500 + }, + { + "epoch": 0.19469171090240914, + "grad_norm": 2.14113450050354, + "learning_rate": 1.6128172277497615e-05, + "loss": 2.4992557525634767, + "step": 71520 + }, + { + "epoch": 0.19474615489315367, + "grad_norm": 2.143296241760254, + "learning_rate": 1.612708191358866e-05, + "loss": 2.534730339050293, + "step": 71540 + }, + { + "epoch": 0.1948005988838982, + "grad_norm": 2.157243490219116, + "learning_rate": 1.6125991549679707e-05, + "loss": 2.4845596313476563, + "step": 71560 + }, + { + "epoch": 0.1948550428746427, + "grad_norm": 2.2046151161193848, + "learning_rate": 1.6124901185770752e-05, + "loss": 2.5422203063964846, + "step": 71580 + }, + { + "epoch": 0.19490948686538723, + "grad_norm": 2.1052513122558594, + "learning_rate": 1.6123810821861798e-05, + "loss": 2.539052200317383, + "step": 71600 + }, + { + "epoch": 0.19496393085613176, + "grad_norm": 2.090927839279175, + "learning_rate": 1.6122720457952844e-05, + "loss": 2.53250675201416, + "step": 71620 + }, + { + "epoch": 0.19501837484687629, + "grad_norm": 2.071936845779419, + "learning_rate": 1.6121630094043887e-05, + "loss": 2.529743194580078, + "step": 71640 + }, + { + "epoch": 0.1950728188376208, + "grad_norm": 2.188732862472534, + "learning_rate": 1.6120539730134936e-05, + "loss": 2.50386962890625, + "step": 71660 + }, + { + "epoch": 0.19512726282836532, + "grad_norm": 2.1381659507751465, + "learning_rate": 1.6119449366225978e-05, + "loss": 2.5209516525268554, + "step": 71680 + }, + { + "epoch": 0.19518170681910985, + "grad_norm": 2.1220643520355225, + "learning_rate": 1.6118359002317024e-05, + "loss": 2.5225372314453125, + "step": 71700 + }, + { + "epoch": 0.19523615080985438, + "grad_norm": 2.191488265991211, + "learning_rate": 1.611726863840807e-05, + "loss": 2.5066350936889648, + "step": 71720 + }, + { + "epoch": 0.19529059480059888, + "grad_norm": 2.1480085849761963, + "learning_rate": 1.6116178274499116e-05, + "loss": 2.523602294921875, + "step": 71740 + }, + { + "epoch": 0.1953450387913434, + "grad_norm": 2.070305824279785, + "learning_rate": 1.611508791059016e-05, + "loss": 2.5167604446411134, + "step": 71760 + }, + { + "epoch": 0.19539948278208794, + "grad_norm": 2.231902599334717, + "learning_rate": 1.6113997546681207e-05, + "loss": 2.497180938720703, + "step": 71780 + }, + { + "epoch": 0.19545392677283244, + "grad_norm": 2.0901036262512207, + "learning_rate": 1.611290718277225e-05, + "loss": 2.5181217193603516, + "step": 71800 + }, + { + "epoch": 0.19550837076357697, + "grad_norm": 2.316610097885132, + "learning_rate": 1.61118168188633e-05, + "loss": 2.5222421646118165, + "step": 71820 + }, + { + "epoch": 0.1955628147543215, + "grad_norm": 2.1659955978393555, + "learning_rate": 1.611072645495434e-05, + "loss": 2.5295940399169923, + "step": 71840 + }, + { + "epoch": 0.19561725874506602, + "grad_norm": 2.124467134475708, + "learning_rate": 1.6109636091045387e-05, + "loss": 2.5140058517456056, + "step": 71860 + }, + { + "epoch": 0.19567170273581053, + "grad_norm": 2.1102259159088135, + "learning_rate": 1.6108545727136433e-05, + "loss": 2.49129581451416, + "step": 71880 + }, + { + "epoch": 0.19572614672655506, + "grad_norm": 2.1382369995117188, + "learning_rate": 1.610745536322748e-05, + "loss": 2.5110795974731444, + "step": 71900 + }, + { + "epoch": 0.19578059071729959, + "grad_norm": 2.1132640838623047, + "learning_rate": 1.6106364999318524e-05, + "loss": 2.521668243408203, + "step": 71920 + }, + { + "epoch": 0.1958350347080441, + "grad_norm": 2.0856547355651855, + "learning_rate": 1.610527463540957e-05, + "loss": 2.5333110809326174, + "step": 71940 + }, + { + "epoch": 0.19588947869878862, + "grad_norm": 2.1786155700683594, + "learning_rate": 1.6104184271500613e-05, + "loss": 2.5288957595825194, + "step": 71960 + }, + { + "epoch": 0.19594392268953315, + "grad_norm": 2.158604621887207, + "learning_rate": 1.6103093907591662e-05, + "loss": 2.5122074127197265, + "step": 71980 + }, + { + "epoch": 0.19599836668027767, + "grad_norm": 2.146174907684326, + "learning_rate": 1.6102003543682704e-05, + "loss": 2.492033767700195, + "step": 72000 + }, + { + "epoch": 0.19605281067102218, + "grad_norm": 2.2732651233673096, + "learning_rate": 1.610091317977375e-05, + "loss": 2.5231252670288087, + "step": 72020 + }, + { + "epoch": 0.1961072546617667, + "grad_norm": 2.1916744709014893, + "learning_rate": 1.6099822815864796e-05, + "loss": 2.557240295410156, + "step": 72040 + }, + { + "epoch": 0.19616169865251123, + "grad_norm": 2.0724174976348877, + "learning_rate": 1.6098732451955842e-05, + "loss": 2.4771684646606444, + "step": 72060 + }, + { + "epoch": 0.19621614264325576, + "grad_norm": 2.097841262817383, + "learning_rate": 1.6097642088046888e-05, + "loss": 2.5294570922851562, + "step": 72080 + }, + { + "epoch": 0.19627058663400027, + "grad_norm": 2.2425315380096436, + "learning_rate": 1.6096551724137933e-05, + "loss": 2.5222198486328127, + "step": 72100 + }, + { + "epoch": 0.1963250306247448, + "grad_norm": 2.101127862930298, + "learning_rate": 1.6095461360228976e-05, + "loss": 2.5087846755981444, + "step": 72120 + }, + { + "epoch": 0.19637947461548932, + "grad_norm": 2.1805620193481445, + "learning_rate": 1.6094370996320025e-05, + "loss": 2.522703170776367, + "step": 72140 + }, + { + "epoch": 0.19643391860623383, + "grad_norm": 2.0996017456054688, + "learning_rate": 1.6093280632411067e-05, + "loss": 2.551276206970215, + "step": 72160 + }, + { + "epoch": 0.19648836259697836, + "grad_norm": 2.1236705780029297, + "learning_rate": 1.6092190268502113e-05, + "loss": 2.5282314300537108, + "step": 72180 + }, + { + "epoch": 0.19654280658772288, + "grad_norm": 2.238591432571411, + "learning_rate": 1.609109990459316e-05, + "loss": 2.528464508056641, + "step": 72200 + }, + { + "epoch": 0.19659725057846741, + "grad_norm": 2.1261751651763916, + "learning_rate": 1.6090009540684205e-05, + "loss": 2.5123886108398437, + "step": 72220 + }, + { + "epoch": 0.19665169456921192, + "grad_norm": 2.2138192653656006, + "learning_rate": 1.608891917677525e-05, + "loss": 2.5095645904541017, + "step": 72240 + }, + { + "epoch": 0.19670613855995644, + "grad_norm": 2.095834970474243, + "learning_rate": 1.6087828812866296e-05, + "loss": 2.5633913040161134, + "step": 72260 + }, + { + "epoch": 0.19676058255070097, + "grad_norm": 2.158944606781006, + "learning_rate": 1.6086738448957342e-05, + "loss": 2.520389747619629, + "step": 72280 + }, + { + "epoch": 0.19681502654144548, + "grad_norm": 2.2440237998962402, + "learning_rate": 1.6085648085048388e-05, + "loss": 2.488803482055664, + "step": 72300 + }, + { + "epoch": 0.19686947053219, + "grad_norm": 2.1090569496154785, + "learning_rate": 1.608455772113943e-05, + "loss": 2.5169734954833984, + "step": 72320 + }, + { + "epoch": 0.19692391452293453, + "grad_norm": 2.326796054840088, + "learning_rate": 1.6083467357230476e-05, + "loss": 2.5074081420898438, + "step": 72340 + }, + { + "epoch": 0.19697835851367906, + "grad_norm": 2.0802531242370605, + "learning_rate": 1.6082376993321522e-05, + "loss": 2.494256591796875, + "step": 72360 + }, + { + "epoch": 0.19703280250442357, + "grad_norm": 2.1004045009613037, + "learning_rate": 1.6081286629412568e-05, + "loss": 2.5268035888671876, + "step": 72380 + }, + { + "epoch": 0.1970872464951681, + "grad_norm": 2.0979979038238525, + "learning_rate": 1.6080196265503614e-05, + "loss": 2.511181640625, + "step": 72400 + }, + { + "epoch": 0.19714169048591262, + "grad_norm": 2.1426327228546143, + "learning_rate": 1.607910590159466e-05, + "loss": 2.503060722351074, + "step": 72420 + }, + { + "epoch": 0.19719613447665713, + "grad_norm": 2.1353960037231445, + "learning_rate": 1.6078015537685705e-05, + "loss": 2.5100263595581054, + "step": 72440 + }, + { + "epoch": 0.19725057846740165, + "grad_norm": 2.3747174739837646, + "learning_rate": 1.607692517377675e-05, + "loss": 2.5230987548828123, + "step": 72460 + }, + { + "epoch": 0.19730502245814618, + "grad_norm": 2.0734035968780518, + "learning_rate": 1.6075834809867794e-05, + "loss": 2.494968032836914, + "step": 72480 + }, + { + "epoch": 0.1973594664488907, + "grad_norm": 2.101391315460205, + "learning_rate": 1.607474444595884e-05, + "loss": 2.537153625488281, + "step": 72500 + }, + { + "epoch": 0.19741391043963522, + "grad_norm": 2.275059223175049, + "learning_rate": 1.6073654082049885e-05, + "loss": 2.4887182235717775, + "step": 72520 + }, + { + "epoch": 0.19746835443037974, + "grad_norm": 2.0399138927459717, + "learning_rate": 1.607256371814093e-05, + "loss": 2.5274288177490236, + "step": 72540 + }, + { + "epoch": 0.19752279842112427, + "grad_norm": 2.1245627403259277, + "learning_rate": 1.6071473354231977e-05, + "loss": 2.5241050720214844, + "step": 72560 + }, + { + "epoch": 0.1975772424118688, + "grad_norm": 2.104124069213867, + "learning_rate": 1.6070382990323023e-05, + "loss": 2.521304130554199, + "step": 72580 + }, + { + "epoch": 0.1976316864026133, + "grad_norm": 2.1506550312042236, + "learning_rate": 1.606929262641407e-05, + "loss": 2.5269855499267577, + "step": 72600 + }, + { + "epoch": 0.19768613039335783, + "grad_norm": 2.101036310195923, + "learning_rate": 1.6068202262505114e-05, + "loss": 2.496181869506836, + "step": 72620 + }, + { + "epoch": 0.19774057438410236, + "grad_norm": 2.1073410511016846, + "learning_rate": 1.6067111898596157e-05, + "loss": 2.5437448501586912, + "step": 72640 + }, + { + "epoch": 0.19779501837484686, + "grad_norm": 2.151951551437378, + "learning_rate": 1.6066021534687202e-05, + "loss": 2.5022708892822267, + "step": 72660 + }, + { + "epoch": 0.1978494623655914, + "grad_norm": 2.1038756370544434, + "learning_rate": 1.6064931170778248e-05, + "loss": 2.5003807067871096, + "step": 72680 + }, + { + "epoch": 0.19790390635633592, + "grad_norm": 2.149580478668213, + "learning_rate": 1.6063840806869294e-05, + "loss": 2.5118637084960938, + "step": 72700 + }, + { + "epoch": 0.19795835034708045, + "grad_norm": 2.1809327602386475, + "learning_rate": 1.606275044296034e-05, + "loss": 2.526309776306152, + "step": 72720 + }, + { + "epoch": 0.19801279433782495, + "grad_norm": 2.0688719749450684, + "learning_rate": 1.6061660079051386e-05, + "loss": 2.543120574951172, + "step": 72740 + }, + { + "epoch": 0.19806723832856948, + "grad_norm": 2.117119312286377, + "learning_rate": 1.606056971514243e-05, + "loss": 2.5622034072875977, + "step": 72760 + }, + { + "epoch": 0.198121682319314, + "grad_norm": 2.360407829284668, + "learning_rate": 1.6059479351233477e-05, + "loss": 2.515111541748047, + "step": 72780 + }, + { + "epoch": 0.19817612631005851, + "grad_norm": 2.4152212142944336, + "learning_rate": 1.605838898732452e-05, + "loss": 2.511082649230957, + "step": 72800 + }, + { + "epoch": 0.19823057030080304, + "grad_norm": 2.1640353202819824, + "learning_rate": 1.6057298623415566e-05, + "loss": 2.5141841888427736, + "step": 72820 + }, + { + "epoch": 0.19828501429154757, + "grad_norm": 2.122615337371826, + "learning_rate": 1.605620825950661e-05, + "loss": 2.520637321472168, + "step": 72840 + }, + { + "epoch": 0.1983394582822921, + "grad_norm": 2.190581798553467, + "learning_rate": 1.6055117895597657e-05, + "loss": 2.515619659423828, + "step": 72860 + }, + { + "epoch": 0.1983939022730366, + "grad_norm": 2.1471941471099854, + "learning_rate": 1.6054027531688703e-05, + "loss": 2.545846176147461, + "step": 72880 + }, + { + "epoch": 0.19844834626378113, + "grad_norm": 2.16701340675354, + "learning_rate": 1.605293716777975e-05, + "loss": 2.543422889709473, + "step": 72900 + }, + { + "epoch": 0.19850279025452566, + "grad_norm": 2.2832393646240234, + "learning_rate": 1.6051846803870795e-05, + "loss": 2.5229440689086915, + "step": 72920 + }, + { + "epoch": 0.1985572342452702, + "grad_norm": 2.0892224311828613, + "learning_rate": 1.605075643996184e-05, + "loss": 2.500750732421875, + "step": 72940 + }, + { + "epoch": 0.1986116782360147, + "grad_norm": 2.1637675762176514, + "learning_rate": 1.6049666076052883e-05, + "loss": 2.509083938598633, + "step": 72960 + }, + { + "epoch": 0.19866612222675922, + "grad_norm": 2.3781659603118896, + "learning_rate": 1.604857571214393e-05, + "loss": 2.5334821701049806, + "step": 72980 + }, + { + "epoch": 0.19872056621750375, + "grad_norm": 2.1305949687957764, + "learning_rate": 1.6047485348234974e-05, + "loss": 2.523465156555176, + "step": 73000 + }, + { + "epoch": 0.19877501020824825, + "grad_norm": 2.291196823120117, + "learning_rate": 1.604639498432602e-05, + "loss": 2.5095546722412108, + "step": 73020 + }, + { + "epoch": 0.19882945419899278, + "grad_norm": 2.1184773445129395, + "learning_rate": 1.6045304620417066e-05, + "loss": 2.523886299133301, + "step": 73040 + }, + { + "epoch": 0.1988838981897373, + "grad_norm": 2.1231563091278076, + "learning_rate": 1.6044214256508112e-05, + "loss": 2.531624412536621, + "step": 73060 + }, + { + "epoch": 0.19893834218048184, + "grad_norm": 2.1389102935791016, + "learning_rate": 1.6043123892599158e-05, + "loss": 2.5235652923583984, + "step": 73080 + }, + { + "epoch": 0.19899278617122634, + "grad_norm": 2.1890869140625, + "learning_rate": 1.6042033528690203e-05, + "loss": 2.5244043350219725, + "step": 73100 + }, + { + "epoch": 0.19904723016197087, + "grad_norm": 2.0929572582244873, + "learning_rate": 1.6040943164781246e-05, + "loss": 2.521678352355957, + "step": 73120 + }, + { + "epoch": 0.1991016741527154, + "grad_norm": 2.127127170562744, + "learning_rate": 1.6039852800872292e-05, + "loss": 2.5268768310546874, + "step": 73140 + }, + { + "epoch": 0.1991561181434599, + "grad_norm": 2.0641298294067383, + "learning_rate": 1.6038762436963338e-05, + "loss": 2.4970577239990233, + "step": 73160 + }, + { + "epoch": 0.19921056213420443, + "grad_norm": 2.2386789321899414, + "learning_rate": 1.6037672073054383e-05, + "loss": 2.5170618057250977, + "step": 73180 + }, + { + "epoch": 0.19926500612494896, + "grad_norm": 2.1136868000030518, + "learning_rate": 1.603658170914543e-05, + "loss": 2.5318008422851563, + "step": 73200 + }, + { + "epoch": 0.1993194501156935, + "grad_norm": 2.0996251106262207, + "learning_rate": 1.6035491345236475e-05, + "loss": 2.531292533874512, + "step": 73220 + }, + { + "epoch": 0.199373894106438, + "grad_norm": 2.059687852859497, + "learning_rate": 1.603440098132752e-05, + "loss": 2.525096321105957, + "step": 73240 + }, + { + "epoch": 0.19942833809718252, + "grad_norm": 2.1678619384765625, + "learning_rate": 1.6033310617418563e-05, + "loss": 2.526174545288086, + "step": 73260 + }, + { + "epoch": 0.19948278208792705, + "grad_norm": 2.1143057346343994, + "learning_rate": 1.603222025350961e-05, + "loss": 2.501502799987793, + "step": 73280 + }, + { + "epoch": 0.19953722607867155, + "grad_norm": 2.1518805027008057, + "learning_rate": 1.6031129889600655e-05, + "loss": 2.507135772705078, + "step": 73300 + }, + { + "epoch": 0.19959167006941608, + "grad_norm": 2.2485949993133545, + "learning_rate": 1.60300395256917e-05, + "loss": 2.531352233886719, + "step": 73320 + }, + { + "epoch": 0.1996461140601606, + "grad_norm": 2.1180548667907715, + "learning_rate": 1.6028949161782746e-05, + "loss": 2.5155033111572265, + "step": 73340 + }, + { + "epoch": 0.19970055805090514, + "grad_norm": 2.127732515335083, + "learning_rate": 1.6027858797873792e-05, + "loss": 2.4882551193237306, + "step": 73360 + }, + { + "epoch": 0.19975500204164964, + "grad_norm": 2.046571969985962, + "learning_rate": 1.6026768433964838e-05, + "loss": 2.53218994140625, + "step": 73380 + }, + { + "epoch": 0.19980944603239417, + "grad_norm": 2.118856430053711, + "learning_rate": 1.6025678070055884e-05, + "loss": 2.5414438247680664, + "step": 73400 + }, + { + "epoch": 0.1998638900231387, + "grad_norm": 2.131382465362549, + "learning_rate": 1.6024587706146926e-05, + "loss": 2.5297061920166017, + "step": 73420 + }, + { + "epoch": 0.19991833401388323, + "grad_norm": 2.04801869392395, + "learning_rate": 1.6023497342237972e-05, + "loss": 2.4954057693481446, + "step": 73440 + }, + { + "epoch": 0.19997277800462773, + "grad_norm": 2.0605862140655518, + "learning_rate": 1.6022406978329018e-05, + "loss": 2.521640586853027, + "step": 73460 + }, + { + "epoch": 0.20002722199537226, + "grad_norm": 2.020157814025879, + "learning_rate": 1.6021316614420064e-05, + "loss": 2.5091840744018556, + "step": 73480 + }, + { + "epoch": 0.2000816659861168, + "grad_norm": 2.2018682956695557, + "learning_rate": 1.602022625051111e-05, + "loss": 2.5167760848999023, + "step": 73500 + }, + { + "epoch": 0.2001361099768613, + "grad_norm": 2.2729361057281494, + "learning_rate": 1.6019135886602155e-05, + "loss": 2.5196712493896483, + "step": 73520 + }, + { + "epoch": 0.20019055396760582, + "grad_norm": 2.1069235801696777, + "learning_rate": 1.60180455226932e-05, + "loss": 2.5107736587524414, + "step": 73540 + }, + { + "epoch": 0.20024499795835035, + "grad_norm": 2.1913297176361084, + "learning_rate": 1.6016955158784247e-05, + "loss": 2.5328039169311523, + "step": 73560 + }, + { + "epoch": 0.20029944194909488, + "grad_norm": 2.1027214527130127, + "learning_rate": 1.601586479487529e-05, + "loss": 2.5021060943603515, + "step": 73580 + }, + { + "epoch": 0.20035388593983938, + "grad_norm": 2.1324286460876465, + "learning_rate": 1.6014774430966335e-05, + "loss": 2.5136043548583986, + "step": 73600 + }, + { + "epoch": 0.2004083299305839, + "grad_norm": 2.0813658237457275, + "learning_rate": 1.601368406705738e-05, + "loss": 2.5101491928100588, + "step": 73620 + }, + { + "epoch": 0.20046277392132844, + "grad_norm": 2.161609649658203, + "learning_rate": 1.6012593703148427e-05, + "loss": 2.5064208984375, + "step": 73640 + }, + { + "epoch": 0.20051721791207294, + "grad_norm": 2.3027899265289307, + "learning_rate": 1.6011503339239473e-05, + "loss": 2.4905420303344727, + "step": 73660 + }, + { + "epoch": 0.20057166190281747, + "grad_norm": 2.119462251663208, + "learning_rate": 1.601041297533052e-05, + "loss": 2.519088554382324, + "step": 73680 + }, + { + "epoch": 0.200626105893562, + "grad_norm": 2.035301923751831, + "learning_rate": 1.6009322611421564e-05, + "loss": 2.5398075103759767, + "step": 73700 + }, + { + "epoch": 0.20068054988430653, + "grad_norm": 2.264697790145874, + "learning_rate": 1.600823224751261e-05, + "loss": 2.516438293457031, + "step": 73720 + }, + { + "epoch": 0.20073499387505103, + "grad_norm": 2.6458804607391357, + "learning_rate": 1.6007141883603652e-05, + "loss": 2.529025840759277, + "step": 73740 + }, + { + "epoch": 0.20078943786579556, + "grad_norm": 2.1276822090148926, + "learning_rate": 1.6006051519694698e-05, + "loss": 2.5262773513793944, + "step": 73760 + }, + { + "epoch": 0.2008438818565401, + "grad_norm": 2.165679931640625, + "learning_rate": 1.6004961155785744e-05, + "loss": 2.5107933044433595, + "step": 73780 + }, + { + "epoch": 0.2008983258472846, + "grad_norm": 2.1657116413116455, + "learning_rate": 1.600387079187679e-05, + "loss": 2.5179983139038087, + "step": 73800 + }, + { + "epoch": 0.20095276983802912, + "grad_norm": 2.315786600112915, + "learning_rate": 1.6002780427967836e-05, + "loss": 2.51873836517334, + "step": 73820 + }, + { + "epoch": 0.20100721382877365, + "grad_norm": 2.046894073486328, + "learning_rate": 1.600169006405888e-05, + "loss": 2.5541990280151365, + "step": 73840 + }, + { + "epoch": 0.20106165781951818, + "grad_norm": 2.090688943862915, + "learning_rate": 1.6000599700149927e-05, + "loss": 2.517093467712402, + "step": 73860 + }, + { + "epoch": 0.20111610181026268, + "grad_norm": 2.121387243270874, + "learning_rate": 1.5999509336240973e-05, + "loss": 2.507880973815918, + "step": 73880 + }, + { + "epoch": 0.2011705458010072, + "grad_norm": 2.1314451694488525, + "learning_rate": 1.5998418972332016e-05, + "loss": 2.5496702194213867, + "step": 73900 + }, + { + "epoch": 0.20122498979175174, + "grad_norm": 2.140061616897583, + "learning_rate": 1.599732860842306e-05, + "loss": 2.5102264404296877, + "step": 73920 + }, + { + "epoch": 0.20127943378249627, + "grad_norm": 2.1146223545074463, + "learning_rate": 1.5996238244514107e-05, + "loss": 2.505282402038574, + "step": 73940 + }, + { + "epoch": 0.20133387777324077, + "grad_norm": 2.129812717437744, + "learning_rate": 1.5995147880605153e-05, + "loss": 2.5537744522094727, + "step": 73960 + }, + { + "epoch": 0.2013883217639853, + "grad_norm": 2.2314352989196777, + "learning_rate": 1.59940575166962e-05, + "loss": 2.5308355331420898, + "step": 73980 + }, + { + "epoch": 0.20144276575472983, + "grad_norm": 2.2029242515563965, + "learning_rate": 1.5992967152787245e-05, + "loss": 2.5614225387573244, + "step": 74000 + }, + { + "epoch": 0.20149720974547433, + "grad_norm": 2.055095911026001, + "learning_rate": 1.599187678887829e-05, + "loss": 2.528738784790039, + "step": 74020 + }, + { + "epoch": 0.20155165373621886, + "grad_norm": 2.112504482269287, + "learning_rate": 1.5990786424969336e-05, + "loss": 2.5162641525268556, + "step": 74040 + }, + { + "epoch": 0.2016060977269634, + "grad_norm": 3.1582720279693604, + "learning_rate": 1.598969606106038e-05, + "loss": 2.5303586959838866, + "step": 74060 + }, + { + "epoch": 0.20166054171770792, + "grad_norm": 2.1691384315490723, + "learning_rate": 1.5988605697151424e-05, + "loss": 2.550055503845215, + "step": 74080 + }, + { + "epoch": 0.20171498570845242, + "grad_norm": 2.125684976577759, + "learning_rate": 1.598751533324247e-05, + "loss": 2.535542106628418, + "step": 74100 + }, + { + "epoch": 0.20176942969919695, + "grad_norm": 2.2004690170288086, + "learning_rate": 1.5986424969333516e-05, + "loss": 2.501907539367676, + "step": 74120 + }, + { + "epoch": 0.20182387368994148, + "grad_norm": 2.169579267501831, + "learning_rate": 1.5985334605424562e-05, + "loss": 2.52701473236084, + "step": 74140 + }, + { + "epoch": 0.20187831768068598, + "grad_norm": 2.1060171127319336, + "learning_rate": 1.5984244241515608e-05, + "loss": 2.5191577911376952, + "step": 74160 + }, + { + "epoch": 0.2019327616714305, + "grad_norm": 2.2488651275634766, + "learning_rate": 1.5983153877606653e-05, + "loss": 2.535384750366211, + "step": 74180 + }, + { + "epoch": 0.20198720566217504, + "grad_norm": 2.258106231689453, + "learning_rate": 1.59820635136977e-05, + "loss": 2.539526176452637, + "step": 74200 + }, + { + "epoch": 0.20204164965291957, + "grad_norm": 2.0815999507904053, + "learning_rate": 1.5980973149788742e-05, + "loss": 2.5174617767333984, + "step": 74220 + }, + { + "epoch": 0.20209609364366407, + "grad_norm": 2.1783838272094727, + "learning_rate": 1.5979882785879788e-05, + "loss": 2.5318511962890624, + "step": 74240 + }, + { + "epoch": 0.2021505376344086, + "grad_norm": 2.125812530517578, + "learning_rate": 1.5978792421970833e-05, + "loss": 2.5242305755615235, + "step": 74260 + }, + { + "epoch": 0.20220498162515313, + "grad_norm": 2.139786720275879, + "learning_rate": 1.597770205806188e-05, + "loss": 2.5112796783447267, + "step": 74280 + }, + { + "epoch": 0.20225942561589766, + "grad_norm": 2.0860090255737305, + "learning_rate": 1.5976611694152925e-05, + "loss": 2.5023897171020506, + "step": 74300 + }, + { + "epoch": 0.20231386960664216, + "grad_norm": 2.0550429821014404, + "learning_rate": 1.597552133024397e-05, + "loss": 2.501087951660156, + "step": 74320 + }, + { + "epoch": 0.2023683135973867, + "grad_norm": 2.050138235092163, + "learning_rate": 1.5974430966335017e-05, + "loss": 2.5284719467163086, + "step": 74340 + }, + { + "epoch": 0.20242275758813122, + "grad_norm": 2.1236281394958496, + "learning_rate": 1.5973340602426062e-05, + "loss": 2.5027847290039062, + "step": 74360 + }, + { + "epoch": 0.20247720157887572, + "grad_norm": 2.0214455127716064, + "learning_rate": 1.5972250238517105e-05, + "loss": 2.5391326904296876, + "step": 74380 + }, + { + "epoch": 0.20253164556962025, + "grad_norm": 2.174123525619507, + "learning_rate": 1.597115987460815e-05, + "loss": 2.523392105102539, + "step": 74400 + }, + { + "epoch": 0.20258608956036478, + "grad_norm": 2.2178943157196045, + "learning_rate": 1.5970069510699196e-05, + "loss": 2.527180290222168, + "step": 74420 + }, + { + "epoch": 0.2026405335511093, + "grad_norm": 2.0683703422546387, + "learning_rate": 1.5968979146790242e-05, + "loss": 2.5365915298461914, + "step": 74440 + }, + { + "epoch": 0.2026949775418538, + "grad_norm": 2.0857837200164795, + "learning_rate": 1.5967888782881288e-05, + "loss": 2.4951562881469727, + "step": 74460 + }, + { + "epoch": 0.20274942153259834, + "grad_norm": 2.6625564098358154, + "learning_rate": 1.5966798418972334e-05, + "loss": 2.4957605361938477, + "step": 74480 + }, + { + "epoch": 0.20280386552334287, + "grad_norm": 2.074739694595337, + "learning_rate": 1.596570805506338e-05, + "loss": 2.5017488479614256, + "step": 74500 + }, + { + "epoch": 0.20285830951408737, + "grad_norm": 2.1191885471343994, + "learning_rate": 1.5964617691154425e-05, + "loss": 2.5173397064208984, + "step": 74520 + }, + { + "epoch": 0.2029127535048319, + "grad_norm": 2.2884883880615234, + "learning_rate": 1.5963527327245468e-05, + "loss": 2.5337623596191405, + "step": 74540 + }, + { + "epoch": 0.20296719749557643, + "grad_norm": 2.1568143367767334, + "learning_rate": 1.5962436963336514e-05, + "loss": 2.530877685546875, + "step": 74560 + }, + { + "epoch": 0.20302164148632096, + "grad_norm": 2.1611311435699463, + "learning_rate": 1.596134659942756e-05, + "loss": 2.507021903991699, + "step": 74580 + }, + { + "epoch": 0.20307608547706546, + "grad_norm": 2.047433376312256, + "learning_rate": 1.5960256235518605e-05, + "loss": 2.528774642944336, + "step": 74600 + }, + { + "epoch": 0.20313052946781, + "grad_norm": 2.4900782108306885, + "learning_rate": 1.595916587160965e-05, + "loss": 2.5641433715820314, + "step": 74620 + }, + { + "epoch": 0.20318497345855452, + "grad_norm": 2.095432996749878, + "learning_rate": 1.5958075507700697e-05, + "loss": 2.5260551452636717, + "step": 74640 + }, + { + "epoch": 0.20323941744929902, + "grad_norm": 2.1836373805999756, + "learning_rate": 1.5956985143791743e-05, + "loss": 2.5205955505371094, + "step": 74660 + }, + { + "epoch": 0.20329386144004355, + "grad_norm": 2.1199498176574707, + "learning_rate": 1.595589477988279e-05, + "loss": 2.5165342330932616, + "step": 74680 + }, + { + "epoch": 0.20334830543078808, + "grad_norm": 2.1828551292419434, + "learning_rate": 1.595480441597383e-05, + "loss": 2.5260955810546877, + "step": 74700 + }, + { + "epoch": 0.2034027494215326, + "grad_norm": 2.355286121368408, + "learning_rate": 1.5953714052064877e-05, + "loss": 2.5180084228515627, + "step": 74720 + }, + { + "epoch": 0.2034571934122771, + "grad_norm": 2.109220266342163, + "learning_rate": 1.5952623688155923e-05, + "loss": 2.508966255187988, + "step": 74740 + }, + { + "epoch": 0.20351163740302164, + "grad_norm": 2.0666773319244385, + "learning_rate": 1.595153332424697e-05, + "loss": 2.517831230163574, + "step": 74760 + }, + { + "epoch": 0.20356608139376617, + "grad_norm": 2.1694440841674805, + "learning_rate": 1.5950442960338014e-05, + "loss": 2.514942169189453, + "step": 74780 + }, + { + "epoch": 0.2036205253845107, + "grad_norm": 2.136550188064575, + "learning_rate": 1.594935259642906e-05, + "loss": 2.5584461212158205, + "step": 74800 + }, + { + "epoch": 0.2036749693752552, + "grad_norm": 2.1029202938079834, + "learning_rate": 1.5948262232520106e-05, + "loss": 2.5376913070678713, + "step": 74820 + }, + { + "epoch": 0.20372941336599973, + "grad_norm": 2.146688222885132, + "learning_rate": 1.594717186861115e-05, + "loss": 2.495050621032715, + "step": 74840 + }, + { + "epoch": 0.20378385735674426, + "grad_norm": 2.0704522132873535, + "learning_rate": 1.5946081504702194e-05, + "loss": 2.5558698654174803, + "step": 74860 + }, + { + "epoch": 0.20383830134748876, + "grad_norm": 2.136990547180176, + "learning_rate": 1.594499114079324e-05, + "loss": 2.527130699157715, + "step": 74880 + }, + { + "epoch": 0.2038927453382333, + "grad_norm": 2.2721805572509766, + "learning_rate": 1.5943900776884286e-05, + "loss": 2.531251907348633, + "step": 74900 + }, + { + "epoch": 0.20394718932897782, + "grad_norm": 2.1606104373931885, + "learning_rate": 1.594281041297533e-05, + "loss": 2.5119949340820313, + "step": 74920 + }, + { + "epoch": 0.20400163331972235, + "grad_norm": 2.193117141723633, + "learning_rate": 1.5941720049066377e-05, + "loss": 2.513337326049805, + "step": 74940 + }, + { + "epoch": 0.20405607731046685, + "grad_norm": 2.470456123352051, + "learning_rate": 1.5940629685157423e-05, + "loss": 2.5391530990600586, + "step": 74960 + }, + { + "epoch": 0.20411052130121138, + "grad_norm": 2.181241512298584, + "learning_rate": 1.593953932124847e-05, + "loss": 2.5272506713867187, + "step": 74980 + }, + { + "epoch": 0.2041649652919559, + "grad_norm": 2.533369541168213, + "learning_rate": 1.5938448957339515e-05, + "loss": 2.4917394638061525, + "step": 75000 + }, + { + "epoch": 0.2042194092827004, + "grad_norm": 2.0750067234039307, + "learning_rate": 1.5937358593430557e-05, + "loss": 2.496552276611328, + "step": 75020 + }, + { + "epoch": 0.20427385327344494, + "grad_norm": 2.0579042434692383, + "learning_rate": 1.5936268229521603e-05, + "loss": 2.499944877624512, + "step": 75040 + }, + { + "epoch": 0.20432829726418947, + "grad_norm": 2.1512298583984375, + "learning_rate": 1.593517786561265e-05, + "loss": 2.5171812057495115, + "step": 75060 + }, + { + "epoch": 0.204382741254934, + "grad_norm": 2.3493850231170654, + "learning_rate": 1.5934087501703695e-05, + "loss": 2.5352956771850588, + "step": 75080 + }, + { + "epoch": 0.2044371852456785, + "grad_norm": 2.827915668487549, + "learning_rate": 1.593299713779474e-05, + "loss": 2.5165288925170897, + "step": 75100 + }, + { + "epoch": 0.20449162923642303, + "grad_norm": 2.3184750080108643, + "learning_rate": 1.5931906773885786e-05, + "loss": 2.5147335052490236, + "step": 75120 + }, + { + "epoch": 0.20454607322716756, + "grad_norm": 2.100794792175293, + "learning_rate": 1.5930816409976832e-05, + "loss": 2.513186454772949, + "step": 75140 + }, + { + "epoch": 0.2046005172179121, + "grad_norm": 2.1607489585876465, + "learning_rate": 1.5929726046067878e-05, + "loss": 2.5101768493652346, + "step": 75160 + }, + { + "epoch": 0.2046549612086566, + "grad_norm": 2.149214029312134, + "learning_rate": 1.592863568215892e-05, + "loss": 2.5337141036987303, + "step": 75180 + }, + { + "epoch": 0.20470940519940112, + "grad_norm": 2.1229794025421143, + "learning_rate": 1.5927545318249966e-05, + "loss": 2.493845748901367, + "step": 75200 + }, + { + "epoch": 0.20476384919014565, + "grad_norm": 2.6039109230041504, + "learning_rate": 1.5926454954341012e-05, + "loss": 2.474003791809082, + "step": 75220 + }, + { + "epoch": 0.20481829318089015, + "grad_norm": 2.049010753631592, + "learning_rate": 1.5925364590432058e-05, + "loss": 2.5283552169799806, + "step": 75240 + }, + { + "epoch": 0.20487273717163468, + "grad_norm": 2.208636999130249, + "learning_rate": 1.5924274226523103e-05, + "loss": 2.513360023498535, + "step": 75260 + }, + { + "epoch": 0.2049271811623792, + "grad_norm": 2.200425624847412, + "learning_rate": 1.592318386261415e-05, + "loss": 2.558872032165527, + "step": 75280 + }, + { + "epoch": 0.20498162515312374, + "grad_norm": 2.0754683017730713, + "learning_rate": 1.5922093498705195e-05, + "loss": 2.5503374099731446, + "step": 75300 + }, + { + "epoch": 0.20503606914386824, + "grad_norm": 2.1635446548461914, + "learning_rate": 1.592100313479624e-05, + "loss": 2.538669204711914, + "step": 75320 + }, + { + "epoch": 0.20509051313461277, + "grad_norm": 1.9865108728408813, + "learning_rate": 1.5919912770887283e-05, + "loss": 2.519738960266113, + "step": 75340 + }, + { + "epoch": 0.2051449571253573, + "grad_norm": 2.2823662757873535, + "learning_rate": 1.591882240697833e-05, + "loss": 2.507820701599121, + "step": 75360 + }, + { + "epoch": 0.2051994011161018, + "grad_norm": 2.1196963787078857, + "learning_rate": 1.5917732043069375e-05, + "loss": 2.5153017044067383, + "step": 75380 + }, + { + "epoch": 0.20525384510684633, + "grad_norm": 2.1362318992614746, + "learning_rate": 1.591664167916042e-05, + "loss": 2.5365116119384767, + "step": 75400 + }, + { + "epoch": 0.20530828909759086, + "grad_norm": 2.142286777496338, + "learning_rate": 1.5915551315251467e-05, + "loss": 2.534989929199219, + "step": 75420 + }, + { + "epoch": 0.2053627330883354, + "grad_norm": 2.1451220512390137, + "learning_rate": 1.5914460951342512e-05, + "loss": 2.529769515991211, + "step": 75440 + }, + { + "epoch": 0.2054171770790799, + "grad_norm": 2.27357816696167, + "learning_rate": 1.5913370587433555e-05, + "loss": 2.5153966903686524, + "step": 75460 + }, + { + "epoch": 0.20547162106982442, + "grad_norm": 2.1569700241088867, + "learning_rate": 1.5912280223524604e-05, + "loss": 2.496315765380859, + "step": 75480 + }, + { + "epoch": 0.20552606506056895, + "grad_norm": 2.080913782119751, + "learning_rate": 1.5911189859615646e-05, + "loss": 2.521925163269043, + "step": 75500 + }, + { + "epoch": 0.20558050905131345, + "grad_norm": 2.1103274822235107, + "learning_rate": 1.5910099495706692e-05, + "loss": 2.513011169433594, + "step": 75520 + }, + { + "epoch": 0.20563495304205798, + "grad_norm": 2.042654514312744, + "learning_rate": 1.5909009131797738e-05, + "loss": 2.5390872955322266, + "step": 75540 + }, + { + "epoch": 0.2056893970328025, + "grad_norm": 2.2073349952697754, + "learning_rate": 1.5907918767888784e-05, + "loss": 2.5283021926879883, + "step": 75560 + }, + { + "epoch": 0.20574384102354704, + "grad_norm": 2.097007989883423, + "learning_rate": 1.590682840397983e-05, + "loss": 2.5114826202392577, + "step": 75580 + }, + { + "epoch": 0.20579828501429154, + "grad_norm": 2.2043511867523193, + "learning_rate": 1.5905738040070875e-05, + "loss": 2.4999696731567385, + "step": 75600 + }, + { + "epoch": 0.20585272900503607, + "grad_norm": 2.0886878967285156, + "learning_rate": 1.5904647676161918e-05, + "loss": 2.523691940307617, + "step": 75620 + }, + { + "epoch": 0.2059071729957806, + "grad_norm": 2.0791640281677246, + "learning_rate": 1.5903557312252967e-05, + "loss": 2.5247602462768555, + "step": 75640 + }, + { + "epoch": 0.20596161698652513, + "grad_norm": 2.07635760307312, + "learning_rate": 1.590246694834401e-05, + "loss": 2.511731719970703, + "step": 75660 + }, + { + "epoch": 0.20601606097726963, + "grad_norm": 2.02557373046875, + "learning_rate": 1.590137658443506e-05, + "loss": 2.5126983642578127, + "step": 75680 + }, + { + "epoch": 0.20607050496801416, + "grad_norm": 2.1802825927734375, + "learning_rate": 1.59002862205261e-05, + "loss": 2.524105262756348, + "step": 75700 + }, + { + "epoch": 0.20612494895875869, + "grad_norm": 2.1180734634399414, + "learning_rate": 1.5899195856617147e-05, + "loss": 2.5498897552490236, + "step": 75720 + }, + { + "epoch": 0.2061793929495032, + "grad_norm": 2.141172409057617, + "learning_rate": 1.5898105492708193e-05, + "loss": 2.5112131118774412, + "step": 75740 + }, + { + "epoch": 0.20623383694024772, + "grad_norm": 2.1972179412841797, + "learning_rate": 1.589701512879924e-05, + "loss": 2.518326759338379, + "step": 75760 + }, + { + "epoch": 0.20628828093099225, + "grad_norm": 2.122572898864746, + "learning_rate": 1.589592476489028e-05, + "loss": 2.5122175216674805, + "step": 75780 + }, + { + "epoch": 0.20634272492173678, + "grad_norm": 2.1243410110473633, + "learning_rate": 1.589483440098133e-05, + "loss": 2.531332778930664, + "step": 75800 + }, + { + "epoch": 0.20639716891248128, + "grad_norm": 2.1713647842407227, + "learning_rate": 1.5893744037072373e-05, + "loss": 2.52244873046875, + "step": 75820 + }, + { + "epoch": 0.2064516129032258, + "grad_norm": 2.2353451251983643, + "learning_rate": 1.5892653673163422e-05, + "loss": 2.5341928482055662, + "step": 75840 + }, + { + "epoch": 0.20650605689397034, + "grad_norm": 2.202221393585205, + "learning_rate": 1.5891563309254464e-05, + "loss": 2.527863693237305, + "step": 75860 + }, + { + "epoch": 0.20656050088471484, + "grad_norm": 2.1237854957580566, + "learning_rate": 1.589047294534551e-05, + "loss": 2.5251399993896486, + "step": 75880 + }, + { + "epoch": 0.20661494487545937, + "grad_norm": 2.091434955596924, + "learning_rate": 1.5889382581436556e-05, + "loss": 2.5250558853149414, + "step": 75900 + }, + { + "epoch": 0.2066693888662039, + "grad_norm": 2.093472719192505, + "learning_rate": 1.58882922175276e-05, + "loss": 2.5276792526245115, + "step": 75920 + }, + { + "epoch": 0.20672383285694843, + "grad_norm": 2.1587183475494385, + "learning_rate": 1.5887201853618644e-05, + "loss": 2.4985935211181642, + "step": 75940 + }, + { + "epoch": 0.20677827684769293, + "grad_norm": 2.0394108295440674, + "learning_rate": 1.5886111489709693e-05, + "loss": 2.517617034912109, + "step": 75960 + }, + { + "epoch": 0.20683272083843746, + "grad_norm": 2.1306581497192383, + "learning_rate": 1.5885021125800736e-05, + "loss": 2.524773597717285, + "step": 75980 + }, + { + "epoch": 0.20688716482918199, + "grad_norm": 2.2208356857299805, + "learning_rate": 1.5883930761891785e-05, + "loss": 2.495102310180664, + "step": 76000 + }, + { + "epoch": 0.2069416088199265, + "grad_norm": 2.037839889526367, + "learning_rate": 1.5882840397982827e-05, + "loss": 2.536330986022949, + "step": 76020 + }, + { + "epoch": 0.20699605281067102, + "grad_norm": 2.148653984069824, + "learning_rate": 1.5881750034073873e-05, + "loss": 2.5415939331054687, + "step": 76040 + }, + { + "epoch": 0.20705049680141555, + "grad_norm": 2.2260069847106934, + "learning_rate": 1.588065967016492e-05, + "loss": 2.493365478515625, + "step": 76060 + }, + { + "epoch": 0.20710494079216007, + "grad_norm": 2.219094753265381, + "learning_rate": 1.5879569306255965e-05, + "loss": 2.522081184387207, + "step": 76080 + }, + { + "epoch": 0.20715938478290458, + "grad_norm": 2.1568260192871094, + "learning_rate": 1.5878478942347007e-05, + "loss": 2.5125778198242186, + "step": 76100 + }, + { + "epoch": 0.2072138287736491, + "grad_norm": 2.1630806922912598, + "learning_rate": 1.5877388578438056e-05, + "loss": 2.5463748931884767, + "step": 76120 + }, + { + "epoch": 0.20726827276439364, + "grad_norm": 2.118774890899658, + "learning_rate": 1.58762982145291e-05, + "loss": 2.498028373718262, + "step": 76140 + }, + { + "epoch": 0.20732271675513816, + "grad_norm": 2.1981258392333984, + "learning_rate": 1.5875207850620148e-05, + "loss": 2.508875846862793, + "step": 76160 + }, + { + "epoch": 0.20737716074588267, + "grad_norm": 2.16019606590271, + "learning_rate": 1.587411748671119e-05, + "loss": 2.526045036315918, + "step": 76180 + }, + { + "epoch": 0.2074316047366272, + "grad_norm": 2.21859073638916, + "learning_rate": 1.5873027122802236e-05, + "loss": 2.516788291931152, + "step": 76200 + }, + { + "epoch": 0.20748604872737172, + "grad_norm": 2.325432538986206, + "learning_rate": 1.5871936758893282e-05, + "loss": 2.5482934951782226, + "step": 76220 + }, + { + "epoch": 0.20754049271811623, + "grad_norm": 2.1682236194610596, + "learning_rate": 1.5870846394984328e-05, + "loss": 2.5172941207885744, + "step": 76240 + }, + { + "epoch": 0.20759493670886076, + "grad_norm": 2.1687490940093994, + "learning_rate": 1.586975603107537e-05, + "loss": 2.50433464050293, + "step": 76260 + }, + { + "epoch": 0.20764938069960528, + "grad_norm": 2.133445978164673, + "learning_rate": 1.586866566716642e-05, + "loss": 2.5397499084472654, + "step": 76280 + }, + { + "epoch": 0.20770382469034981, + "grad_norm": 2.1511995792388916, + "learning_rate": 1.5867575303257462e-05, + "loss": 2.512333106994629, + "step": 76300 + }, + { + "epoch": 0.20775826868109432, + "grad_norm": 2.2545650005340576, + "learning_rate": 1.586648493934851e-05, + "loss": 2.5452943801879884, + "step": 76320 + }, + { + "epoch": 0.20781271267183885, + "grad_norm": 2.193068265914917, + "learning_rate": 1.5865394575439553e-05, + "loss": 2.5573009490966796, + "step": 76340 + }, + { + "epoch": 0.20786715666258337, + "grad_norm": 2.524693727493286, + "learning_rate": 1.58643042115306e-05, + "loss": 2.514090728759766, + "step": 76360 + }, + { + "epoch": 0.20792160065332788, + "grad_norm": 2.1432952880859375, + "learning_rate": 1.5863213847621645e-05, + "loss": 2.5214826583862306, + "step": 76380 + }, + { + "epoch": 0.2079760446440724, + "grad_norm": 2.0753326416015625, + "learning_rate": 1.586212348371269e-05, + "loss": 2.5088146209716795, + "step": 76400 + }, + { + "epoch": 0.20803048863481693, + "grad_norm": 2.1401736736297607, + "learning_rate": 1.5861033119803733e-05, + "loss": 2.519832420349121, + "step": 76420 + }, + { + "epoch": 0.20808493262556146, + "grad_norm": 2.112241268157959, + "learning_rate": 1.5859942755894782e-05, + "loss": 2.501139831542969, + "step": 76440 + }, + { + "epoch": 0.20813937661630597, + "grad_norm": 2.118422269821167, + "learning_rate": 1.5858852391985825e-05, + "loss": 2.509767532348633, + "step": 76460 + }, + { + "epoch": 0.2081938206070505, + "grad_norm": 2.2117233276367188, + "learning_rate": 1.5857762028076874e-05, + "loss": 2.530165672302246, + "step": 76480 + }, + { + "epoch": 0.20824826459779502, + "grad_norm": 2.0960991382598877, + "learning_rate": 1.5856671664167917e-05, + "loss": 2.5306005477905273, + "step": 76500 + }, + { + "epoch": 0.20830270858853955, + "grad_norm": 2.146122694015503, + "learning_rate": 1.5855581300258962e-05, + "loss": 2.5324729919433593, + "step": 76520 + }, + { + "epoch": 0.20835715257928406, + "grad_norm": 2.2513909339904785, + "learning_rate": 1.5854490936350008e-05, + "loss": 2.521592903137207, + "step": 76540 + }, + { + "epoch": 0.20841159657002858, + "grad_norm": 2.2584073543548584, + "learning_rate": 1.5853400572441054e-05, + "loss": 2.545452690124512, + "step": 76560 + }, + { + "epoch": 0.2084660405607731, + "grad_norm": 2.1543161869049072, + "learning_rate": 1.5852310208532096e-05, + "loss": 2.5063961029052733, + "step": 76580 + }, + { + "epoch": 0.20852048455151762, + "grad_norm": 2.1682162284851074, + "learning_rate": 1.5851219844623146e-05, + "loss": 2.5243160247802736, + "step": 76600 + }, + { + "epoch": 0.20857492854226214, + "grad_norm": 2.1052091121673584, + "learning_rate": 1.5850129480714188e-05, + "loss": 2.531351852416992, + "step": 76620 + }, + { + "epoch": 0.20862937253300667, + "grad_norm": 2.1128344535827637, + "learning_rate": 1.5849039116805237e-05, + "loss": 2.5186487197875977, + "step": 76640 + }, + { + "epoch": 0.2086838165237512, + "grad_norm": 2.1602869033813477, + "learning_rate": 1.584794875289628e-05, + "loss": 2.4909337997436523, + "step": 76660 + }, + { + "epoch": 0.2087382605144957, + "grad_norm": 2.1516246795654297, + "learning_rate": 1.5846858388987325e-05, + "loss": 2.527614212036133, + "step": 76680 + }, + { + "epoch": 0.20879270450524023, + "grad_norm": 2.137840747833252, + "learning_rate": 1.584576802507837e-05, + "loss": 2.526837921142578, + "step": 76700 + }, + { + "epoch": 0.20884714849598476, + "grad_norm": 2.0606837272644043, + "learning_rate": 1.5844677661169417e-05, + "loss": 2.488554573059082, + "step": 76720 + }, + { + "epoch": 0.20890159248672927, + "grad_norm": 2.1873772144317627, + "learning_rate": 1.584358729726046e-05, + "loss": 2.500718116760254, + "step": 76740 + }, + { + "epoch": 0.2089560364774738, + "grad_norm": 2.088970184326172, + "learning_rate": 1.584249693335151e-05, + "loss": 2.5715925216674806, + "step": 76760 + }, + { + "epoch": 0.20901048046821832, + "grad_norm": 2.046391725540161, + "learning_rate": 1.584140656944255e-05, + "loss": 2.5578868865966795, + "step": 76780 + }, + { + "epoch": 0.20906492445896285, + "grad_norm": 2.16328763961792, + "learning_rate": 1.58403162055336e-05, + "loss": 2.5285816192626953, + "step": 76800 + }, + { + "epoch": 0.20911936844970735, + "grad_norm": 2.306739091873169, + "learning_rate": 1.5839225841624643e-05, + "loss": 2.5230876922607424, + "step": 76820 + }, + { + "epoch": 0.20917381244045188, + "grad_norm": 2.064434766769409, + "learning_rate": 1.583813547771569e-05, + "loss": 2.5246538162231444, + "step": 76840 + }, + { + "epoch": 0.2092282564311964, + "grad_norm": 2.0734026432037354, + "learning_rate": 1.5837045113806734e-05, + "loss": 2.5105587005615235, + "step": 76860 + }, + { + "epoch": 0.20928270042194091, + "grad_norm": 2.106814384460449, + "learning_rate": 1.583595474989778e-05, + "loss": 2.491655158996582, + "step": 76880 + }, + { + "epoch": 0.20933714441268544, + "grad_norm": 2.18969464302063, + "learning_rate": 1.5834864385988823e-05, + "loss": 2.5024845123291017, + "step": 76900 + }, + { + "epoch": 0.20939158840342997, + "grad_norm": 2.108271837234497, + "learning_rate": 1.5833774022079872e-05, + "loss": 2.5261714935302733, + "step": 76920 + }, + { + "epoch": 0.2094460323941745, + "grad_norm": 2.412900686264038, + "learning_rate": 1.5832683658170914e-05, + "loss": 2.5315118789672852, + "step": 76940 + }, + { + "epoch": 0.209500476384919, + "grad_norm": 2.1424942016601562, + "learning_rate": 1.5831593294261963e-05, + "loss": 2.5475591659545898, + "step": 76960 + }, + { + "epoch": 0.20955492037566353, + "grad_norm": 2.1489996910095215, + "learning_rate": 1.5830502930353006e-05, + "loss": 2.50272274017334, + "step": 76980 + }, + { + "epoch": 0.20960936436640806, + "grad_norm": 2.0868098735809326, + "learning_rate": 1.582941256644405e-05, + "loss": 2.5342071533203123, + "step": 77000 + }, + { + "epoch": 0.2096638083571526, + "grad_norm": 2.106464147567749, + "learning_rate": 1.5828322202535097e-05, + "loss": 2.509832000732422, + "step": 77020 + }, + { + "epoch": 0.2097182523478971, + "grad_norm": 2.189394950866699, + "learning_rate": 1.5827231838626143e-05, + "loss": 2.511639404296875, + "step": 77040 + }, + { + "epoch": 0.20977269633864162, + "grad_norm": 2.262855291366577, + "learning_rate": 1.5826141474717186e-05, + "loss": 2.536171722412109, + "step": 77060 + }, + { + "epoch": 0.20982714032938615, + "grad_norm": 2.0705478191375732, + "learning_rate": 1.5825051110808235e-05, + "loss": 2.5060367584228516, + "step": 77080 + }, + { + "epoch": 0.20988158432013065, + "grad_norm": 2.090304374694824, + "learning_rate": 1.5823960746899277e-05, + "loss": 2.520005989074707, + "step": 77100 + }, + { + "epoch": 0.20993602831087518, + "grad_norm": 2.1439990997314453, + "learning_rate": 1.5822870382990326e-05, + "loss": 2.5157358169555666, + "step": 77120 + }, + { + "epoch": 0.2099904723016197, + "grad_norm": 2.116651773452759, + "learning_rate": 1.582178001908137e-05, + "loss": 2.5144805908203125, + "step": 77140 + }, + { + "epoch": 0.21004491629236424, + "grad_norm": 2.2211811542510986, + "learning_rate": 1.5820689655172415e-05, + "loss": 2.5272974014282226, + "step": 77160 + }, + { + "epoch": 0.21009936028310874, + "grad_norm": 2.2017478942871094, + "learning_rate": 1.581959929126346e-05, + "loss": 2.5197324752807617, + "step": 77180 + }, + { + "epoch": 0.21015380427385327, + "grad_norm": 2.1947171688079834, + "learning_rate": 1.5818508927354506e-05, + "loss": 2.541507339477539, + "step": 77200 + }, + { + "epoch": 0.2102082482645978, + "grad_norm": 2.118295907974243, + "learning_rate": 1.581741856344555e-05, + "loss": 2.508543586730957, + "step": 77220 + }, + { + "epoch": 0.2102626922553423, + "grad_norm": 2.006416082382202, + "learning_rate": 1.5816328199536598e-05, + "loss": 2.546124076843262, + "step": 77240 + }, + { + "epoch": 0.21031713624608683, + "grad_norm": 2.139326333999634, + "learning_rate": 1.581523783562764e-05, + "loss": 2.5013143539428713, + "step": 77260 + }, + { + "epoch": 0.21037158023683136, + "grad_norm": 2.2786643505096436, + "learning_rate": 1.581414747171869e-05, + "loss": 2.5045743942260743, + "step": 77280 + }, + { + "epoch": 0.2104260242275759, + "grad_norm": 2.2227563858032227, + "learning_rate": 1.5813057107809732e-05, + "loss": 2.5310348510742187, + "step": 77300 + }, + { + "epoch": 0.2104804682183204, + "grad_norm": 2.100957155227661, + "learning_rate": 1.5811966743900778e-05, + "loss": 2.522304725646973, + "step": 77320 + }, + { + "epoch": 0.21053491220906492, + "grad_norm": 2.5374741554260254, + "learning_rate": 1.5810876379991824e-05, + "loss": 2.5021156311035155, + "step": 77340 + }, + { + "epoch": 0.21058935619980945, + "grad_norm": 2.2654354572296143, + "learning_rate": 1.580978601608287e-05, + "loss": 2.5102506637573243, + "step": 77360 + }, + { + "epoch": 0.21064380019055398, + "grad_norm": 2.1161646842956543, + "learning_rate": 1.5808695652173915e-05, + "loss": 2.500759506225586, + "step": 77380 + }, + { + "epoch": 0.21069824418129848, + "grad_norm": 2.165992259979248, + "learning_rate": 1.580760528826496e-05, + "loss": 2.5223270416259767, + "step": 77400 + }, + { + "epoch": 0.210752688172043, + "grad_norm": 2.4409122467041016, + "learning_rate": 1.5806514924356003e-05, + "loss": 2.499435806274414, + "step": 77420 + }, + { + "epoch": 0.21080713216278754, + "grad_norm": 2.065412759780884, + "learning_rate": 1.5805424560447053e-05, + "loss": 2.510746192932129, + "step": 77440 + }, + { + "epoch": 0.21086157615353204, + "grad_norm": 2.1369354724884033, + "learning_rate": 1.5804334196538095e-05, + "loss": 2.526483917236328, + "step": 77460 + }, + { + "epoch": 0.21091602014427657, + "grad_norm": 2.262482166290283, + "learning_rate": 1.580324383262914e-05, + "loss": 2.513123321533203, + "step": 77480 + }, + { + "epoch": 0.2109704641350211, + "grad_norm": 2.049734592437744, + "learning_rate": 1.5802153468720187e-05, + "loss": 2.5273273468017576, + "step": 77500 + }, + { + "epoch": 0.21102490812576563, + "grad_norm": 2.2440145015716553, + "learning_rate": 1.5801063104811232e-05, + "loss": 2.509779357910156, + "step": 77520 + }, + { + "epoch": 0.21107935211651013, + "grad_norm": 2.1874516010284424, + "learning_rate": 1.5799972740902278e-05, + "loss": 2.525942802429199, + "step": 77540 + }, + { + "epoch": 0.21113379610725466, + "grad_norm": 2.146390438079834, + "learning_rate": 1.5798882376993324e-05, + "loss": 2.5039836883544924, + "step": 77560 + }, + { + "epoch": 0.2111882400979992, + "grad_norm": 2.2730979919433594, + "learning_rate": 1.5797792013084366e-05, + "loss": 2.5314592361450194, + "step": 77580 + }, + { + "epoch": 0.2112426840887437, + "grad_norm": 2.112809181213379, + "learning_rate": 1.5796701649175416e-05, + "loss": 2.5218523025512694, + "step": 77600 + }, + { + "epoch": 0.21129712807948822, + "grad_norm": 2.1576004028320312, + "learning_rate": 1.5795611285266458e-05, + "loss": 2.521183395385742, + "step": 77620 + }, + { + "epoch": 0.21135157207023275, + "grad_norm": 2.2096643447875977, + "learning_rate": 1.5794520921357504e-05, + "loss": 2.4880630493164064, + "step": 77640 + }, + { + "epoch": 0.21140601606097728, + "grad_norm": 2.0908515453338623, + "learning_rate": 1.579343055744855e-05, + "loss": 2.531048583984375, + "step": 77660 + }, + { + "epoch": 0.21146046005172178, + "grad_norm": 2.1807594299316406, + "learning_rate": 1.5792340193539596e-05, + "loss": 2.5335025787353516, + "step": 77680 + }, + { + "epoch": 0.2115149040424663, + "grad_norm": 2.119188070297241, + "learning_rate": 1.579124982963064e-05, + "loss": 2.5095998764038088, + "step": 77700 + }, + { + "epoch": 0.21156934803321084, + "grad_norm": 2.112464189529419, + "learning_rate": 1.5790159465721687e-05, + "loss": 2.5044410705566404, + "step": 77720 + }, + { + "epoch": 0.21162379202395534, + "grad_norm": 2.056090831756592, + "learning_rate": 1.578906910181273e-05, + "loss": 2.5233652114868166, + "step": 77740 + }, + { + "epoch": 0.21167823601469987, + "grad_norm": 2.1872239112854004, + "learning_rate": 1.578797873790378e-05, + "loss": 2.537296104431152, + "step": 77760 + }, + { + "epoch": 0.2117326800054444, + "grad_norm": 2.05757737159729, + "learning_rate": 1.578688837399482e-05, + "loss": 2.5209041595458985, + "step": 77780 + }, + { + "epoch": 0.21178712399618893, + "grad_norm": 2.2565276622772217, + "learning_rate": 1.5785798010085867e-05, + "loss": 2.5123260498046873, + "step": 77800 + }, + { + "epoch": 0.21184156798693343, + "grad_norm": 2.1597790718078613, + "learning_rate": 1.5784707646176913e-05, + "loss": 2.476272773742676, + "step": 77820 + }, + { + "epoch": 0.21189601197767796, + "grad_norm": 2.1527528762817383, + "learning_rate": 1.578361728226796e-05, + "loss": 2.4998435974121094, + "step": 77840 + }, + { + "epoch": 0.2119504559684225, + "grad_norm": 2.140474557876587, + "learning_rate": 1.5782526918359004e-05, + "loss": 2.533583641052246, + "step": 77860 + }, + { + "epoch": 0.21200489995916702, + "grad_norm": 2.110985040664673, + "learning_rate": 1.578143655445005e-05, + "loss": 2.5248411178588865, + "step": 77880 + }, + { + "epoch": 0.21205934394991152, + "grad_norm": 2.1114859580993652, + "learning_rate": 1.5780346190541093e-05, + "loss": 2.4971267700195314, + "step": 77900 + }, + { + "epoch": 0.21211378794065605, + "grad_norm": 2.0549957752227783, + "learning_rate": 1.5779255826632142e-05, + "loss": 2.493358612060547, + "step": 77920 + }, + { + "epoch": 0.21216823193140058, + "grad_norm": 1.9894686937332153, + "learning_rate": 1.5778165462723184e-05, + "loss": 2.5131252288818358, + "step": 77940 + }, + { + "epoch": 0.21222267592214508, + "grad_norm": 2.1368391513824463, + "learning_rate": 1.577707509881423e-05, + "loss": 2.5319921493530275, + "step": 77960 + }, + { + "epoch": 0.2122771199128896, + "grad_norm": 2.1141769886016846, + "learning_rate": 1.5775984734905276e-05, + "loss": 2.4760204315185548, + "step": 77980 + }, + { + "epoch": 0.21233156390363414, + "grad_norm": 2.0120463371276855, + "learning_rate": 1.5774894370996322e-05, + "loss": 2.540071487426758, + "step": 78000 + }, + { + "epoch": 0.21238600789437867, + "grad_norm": 2.1573915481567383, + "learning_rate": 1.5773804007087368e-05, + "loss": 2.523391342163086, + "step": 78020 + }, + { + "epoch": 0.21244045188512317, + "grad_norm": 2.0881264209747314, + "learning_rate": 1.5772713643178413e-05, + "loss": 2.5072555541992188, + "step": 78040 + }, + { + "epoch": 0.2124948958758677, + "grad_norm": 2.1477279663085938, + "learning_rate": 1.5771623279269456e-05, + "loss": 2.5294408798217773, + "step": 78060 + }, + { + "epoch": 0.21254933986661223, + "grad_norm": 2.2347633838653564, + "learning_rate": 1.5770532915360505e-05, + "loss": 2.5365610122680664, + "step": 78080 + }, + { + "epoch": 0.21260378385735673, + "grad_norm": 2.1740200519561768, + "learning_rate": 1.5769442551451547e-05, + "loss": 2.5080869674682615, + "step": 78100 + }, + { + "epoch": 0.21265822784810126, + "grad_norm": 2.21952486038208, + "learning_rate": 1.5768352187542593e-05, + "loss": 2.505309295654297, + "step": 78120 + }, + { + "epoch": 0.2127126718388458, + "grad_norm": 2.5783932209014893, + "learning_rate": 1.576726182363364e-05, + "loss": 2.5259845733642576, + "step": 78140 + }, + { + "epoch": 0.21276711582959032, + "grad_norm": 2.0939834117889404, + "learning_rate": 1.5766171459724685e-05, + "loss": 2.5419637680053713, + "step": 78160 + }, + { + "epoch": 0.21282155982033482, + "grad_norm": 2.04142689704895, + "learning_rate": 1.576508109581573e-05, + "loss": 2.5123958587646484, + "step": 78180 + }, + { + "epoch": 0.21287600381107935, + "grad_norm": 2.76177978515625, + "learning_rate": 1.5763990731906776e-05, + "loss": 2.5360385894775392, + "step": 78200 + }, + { + "epoch": 0.21293044780182388, + "grad_norm": 2.0938448905944824, + "learning_rate": 1.576290036799782e-05, + "loss": 2.497085189819336, + "step": 78220 + }, + { + "epoch": 0.21298489179256838, + "grad_norm": 2.0454912185668945, + "learning_rate": 1.5761810004088868e-05, + "loss": 2.514013671875, + "step": 78240 + }, + { + "epoch": 0.2130393357833129, + "grad_norm": 2.087343215942383, + "learning_rate": 1.576071964017991e-05, + "loss": 2.527816581726074, + "step": 78260 + }, + { + "epoch": 0.21309377977405744, + "grad_norm": 2.152017593383789, + "learning_rate": 1.5759629276270956e-05, + "loss": 2.5314729690551756, + "step": 78280 + }, + { + "epoch": 0.21314822376480197, + "grad_norm": 2.2431371212005615, + "learning_rate": 1.5758538912362002e-05, + "loss": 2.5179683685302736, + "step": 78300 + }, + { + "epoch": 0.21320266775554647, + "grad_norm": 2.028737783432007, + "learning_rate": 1.5757448548453048e-05, + "loss": 2.531698226928711, + "step": 78320 + }, + { + "epoch": 0.213257111746291, + "grad_norm": 2.284727096557617, + "learning_rate": 1.5756358184544094e-05, + "loss": 2.5342971801757814, + "step": 78340 + }, + { + "epoch": 0.21331155573703553, + "grad_norm": 2.0811984539031982, + "learning_rate": 1.575526782063514e-05, + "loss": 2.51761474609375, + "step": 78360 + }, + { + "epoch": 0.21336599972778006, + "grad_norm": 2.1303207874298096, + "learning_rate": 1.5754177456726182e-05, + "loss": 2.5209623336791993, + "step": 78380 + }, + { + "epoch": 0.21342044371852456, + "grad_norm": 2.0353498458862305, + "learning_rate": 1.575308709281723e-05, + "loss": 2.509588432312012, + "step": 78400 + }, + { + "epoch": 0.2134748877092691, + "grad_norm": 2.166388988494873, + "learning_rate": 1.5751996728908274e-05, + "loss": 2.518483352661133, + "step": 78420 + }, + { + "epoch": 0.21352933170001362, + "grad_norm": 2.4956471920013428, + "learning_rate": 1.575090636499932e-05, + "loss": 2.5245141983032227, + "step": 78440 + }, + { + "epoch": 0.21358377569075812, + "grad_norm": 2.0798096656799316, + "learning_rate": 1.5749816001090365e-05, + "loss": 2.5088897705078126, + "step": 78460 + }, + { + "epoch": 0.21363821968150265, + "grad_norm": 2.154766798019409, + "learning_rate": 1.574872563718141e-05, + "loss": 2.5134740829467774, + "step": 78480 + }, + { + "epoch": 0.21369266367224718, + "grad_norm": 2.1645750999450684, + "learning_rate": 1.5747635273272457e-05, + "loss": 2.4954952239990233, + "step": 78500 + }, + { + "epoch": 0.2137471076629917, + "grad_norm": 2.3189029693603516, + "learning_rate": 1.5746544909363503e-05, + "loss": 2.539710807800293, + "step": 78520 + }, + { + "epoch": 0.2138015516537362, + "grad_norm": 2.2702157497406006, + "learning_rate": 1.5745454545454545e-05, + "loss": 2.542556571960449, + "step": 78540 + }, + { + "epoch": 0.21385599564448074, + "grad_norm": 2.0705130100250244, + "learning_rate": 1.5744364181545594e-05, + "loss": 2.536543273925781, + "step": 78560 + }, + { + "epoch": 0.21391043963522527, + "grad_norm": 2.1757211685180664, + "learning_rate": 1.5743273817636637e-05, + "loss": 2.493263816833496, + "step": 78580 + }, + { + "epoch": 0.21396488362596977, + "grad_norm": 2.1714744567871094, + "learning_rate": 1.5742183453727682e-05, + "loss": 2.523751640319824, + "step": 78600 + }, + { + "epoch": 0.2140193276167143, + "grad_norm": 2.0750339031219482, + "learning_rate": 1.5741093089818728e-05, + "loss": 2.4970075607299806, + "step": 78620 + }, + { + "epoch": 0.21407377160745883, + "grad_norm": 2.1393420696258545, + "learning_rate": 1.5740002725909774e-05, + "loss": 2.5233108520507814, + "step": 78640 + }, + { + "epoch": 0.21412821559820336, + "grad_norm": 4.92030143737793, + "learning_rate": 1.573891236200082e-05, + "loss": 2.523697280883789, + "step": 78660 + }, + { + "epoch": 0.21418265958894786, + "grad_norm": 2.0677149295806885, + "learning_rate": 1.5737821998091866e-05, + "loss": 2.5060659408569337, + "step": 78680 + }, + { + "epoch": 0.2142371035796924, + "grad_norm": 2.0825695991516113, + "learning_rate": 1.5736731634182908e-05, + "loss": 2.522023391723633, + "step": 78700 + }, + { + "epoch": 0.21429154757043692, + "grad_norm": 2.262507438659668, + "learning_rate": 1.5735641270273957e-05, + "loss": 2.486974906921387, + "step": 78720 + }, + { + "epoch": 0.21434599156118145, + "grad_norm": 2.0624544620513916, + "learning_rate": 1.5734550906365e-05, + "loss": 2.5114862442016603, + "step": 78740 + }, + { + "epoch": 0.21440043555192595, + "grad_norm": 2.1744208335876465, + "learning_rate": 1.5733460542456046e-05, + "loss": 2.523947906494141, + "step": 78760 + }, + { + "epoch": 0.21445487954267048, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.573237017854709e-05, + "loss": 2.506320571899414, + "step": 78780 + }, + { + "epoch": 0.214509323533415, + "grad_norm": 2.084591865539551, + "learning_rate": 1.5731279814638137e-05, + "loss": 2.5356029510498046, + "step": 78800 + }, + { + "epoch": 0.2145637675241595, + "grad_norm": 2.1218769550323486, + "learning_rate": 1.5730189450729183e-05, + "loss": 2.52502326965332, + "step": 78820 + }, + { + "epoch": 0.21461821151490404, + "grad_norm": 2.194716691970825, + "learning_rate": 1.572909908682023e-05, + "loss": 2.5093626022338866, + "step": 78840 + }, + { + "epoch": 0.21467265550564857, + "grad_norm": 2.409330368041992, + "learning_rate": 1.572800872291127e-05, + "loss": 2.5128786087036135, + "step": 78860 + }, + { + "epoch": 0.2147270994963931, + "grad_norm": 2.22499942779541, + "learning_rate": 1.572691835900232e-05, + "loss": 2.4999292373657225, + "step": 78880 + }, + { + "epoch": 0.2147815434871376, + "grad_norm": 2.798003911972046, + "learning_rate": 1.5725827995093363e-05, + "loss": 2.519789123535156, + "step": 78900 + }, + { + "epoch": 0.21483598747788213, + "grad_norm": 2.194833517074585, + "learning_rate": 1.572473763118441e-05, + "loss": 2.538203239440918, + "step": 78920 + }, + { + "epoch": 0.21489043146862666, + "grad_norm": 2.1125612258911133, + "learning_rate": 1.5723647267275454e-05, + "loss": 2.516423225402832, + "step": 78940 + }, + { + "epoch": 0.21494487545937116, + "grad_norm": 2.3445627689361572, + "learning_rate": 1.57225569033665e-05, + "loss": 2.533234977722168, + "step": 78960 + }, + { + "epoch": 0.2149993194501157, + "grad_norm": 2.114109992980957, + "learning_rate": 1.5721466539457546e-05, + "loss": 2.5197011947631838, + "step": 78980 + }, + { + "epoch": 0.21505376344086022, + "grad_norm": 2.1237852573394775, + "learning_rate": 1.5720376175548592e-05, + "loss": 2.5414600372314453, + "step": 79000 + }, + { + "epoch": 0.21510820743160475, + "grad_norm": 2.0484542846679688, + "learning_rate": 1.5719285811639634e-05, + "loss": 2.528261184692383, + "step": 79020 + }, + { + "epoch": 0.21516265142234925, + "grad_norm": 2.167606830596924, + "learning_rate": 1.5718195447730683e-05, + "loss": 2.505496597290039, + "step": 79040 + }, + { + "epoch": 0.21521709541309378, + "grad_norm": 2.136054277420044, + "learning_rate": 1.5717105083821726e-05, + "loss": 2.5242305755615235, + "step": 79060 + }, + { + "epoch": 0.2152715394038383, + "grad_norm": 2.1355135440826416, + "learning_rate": 1.5716014719912772e-05, + "loss": 2.5000940322875977, + "step": 79080 + }, + { + "epoch": 0.2153259833945828, + "grad_norm": 2.1669514179229736, + "learning_rate": 1.5714924356003818e-05, + "loss": 2.56102294921875, + "step": 79100 + }, + { + "epoch": 0.21538042738532734, + "grad_norm": 2.1375763416290283, + "learning_rate": 1.5713833992094863e-05, + "loss": 2.4907779693603516, + "step": 79120 + }, + { + "epoch": 0.21543487137607187, + "grad_norm": 2.1280324459075928, + "learning_rate": 1.571274362818591e-05, + "loss": 2.5377424240112303, + "step": 79140 + }, + { + "epoch": 0.2154893153668164, + "grad_norm": 2.1790969371795654, + "learning_rate": 1.5711653264276955e-05, + "loss": 2.495547866821289, + "step": 79160 + }, + { + "epoch": 0.2155437593575609, + "grad_norm": 2.030488967895508, + "learning_rate": 1.5710562900367997e-05, + "loss": 2.5132017135620117, + "step": 79180 + }, + { + "epoch": 0.21559820334830543, + "grad_norm": 2.205991506576538, + "learning_rate": 1.5709472536459047e-05, + "loss": 2.4913448333740233, + "step": 79200 + }, + { + "epoch": 0.21565264733904996, + "grad_norm": 2.279571056365967, + "learning_rate": 1.570838217255009e-05, + "loss": 2.5120792388916016, + "step": 79220 + }, + { + "epoch": 0.2157070913297945, + "grad_norm": 2.1285600662231445, + "learning_rate": 1.5707291808641135e-05, + "loss": 2.526866912841797, + "step": 79240 + }, + { + "epoch": 0.215761535320539, + "grad_norm": 2.2057549953460693, + "learning_rate": 1.570620144473218e-05, + "loss": 2.5143125534057615, + "step": 79260 + }, + { + "epoch": 0.21581597931128352, + "grad_norm": 2.2520453929901123, + "learning_rate": 1.5705111080823226e-05, + "loss": 2.5132844924926756, + "step": 79280 + }, + { + "epoch": 0.21587042330202805, + "grad_norm": 2.114133834838867, + "learning_rate": 1.5704020716914272e-05, + "loss": 2.5329612731933593, + "step": 79300 + }, + { + "epoch": 0.21592486729277255, + "grad_norm": 2.2369046211242676, + "learning_rate": 1.5702930353005318e-05, + "loss": 2.539036178588867, + "step": 79320 + }, + { + "epoch": 0.21597931128351708, + "grad_norm": 2.2047317028045654, + "learning_rate": 1.570183998909636e-05, + "loss": 2.5167245864868164, + "step": 79340 + }, + { + "epoch": 0.2160337552742616, + "grad_norm": 2.0776121616363525, + "learning_rate": 1.570074962518741e-05, + "loss": 2.5038528442382812, + "step": 79360 + }, + { + "epoch": 0.21608819926500614, + "grad_norm": 2.1612417697906494, + "learning_rate": 1.5699659261278452e-05, + "loss": 2.4962722778320314, + "step": 79380 + }, + { + "epoch": 0.21614264325575064, + "grad_norm": 2.1410696506500244, + "learning_rate": 1.5698568897369498e-05, + "loss": 2.5382509231567383, + "step": 79400 + }, + { + "epoch": 0.21619708724649517, + "grad_norm": 2.2269160747528076, + "learning_rate": 1.5697478533460544e-05, + "loss": 2.507090377807617, + "step": 79420 + }, + { + "epoch": 0.2162515312372397, + "grad_norm": 2.0971202850341797, + "learning_rate": 1.569638816955159e-05, + "loss": 2.506302833557129, + "step": 79440 + }, + { + "epoch": 0.2163059752279842, + "grad_norm": 2.137660503387451, + "learning_rate": 1.5695297805642635e-05, + "loss": 2.5097000122070314, + "step": 79460 + }, + { + "epoch": 0.21636041921872873, + "grad_norm": 2.0438103675842285, + "learning_rate": 1.569420744173368e-05, + "loss": 2.5211877822875977, + "step": 79480 + }, + { + "epoch": 0.21641486320947326, + "grad_norm": 2.2127254009246826, + "learning_rate": 1.5693117077824724e-05, + "loss": 2.497832679748535, + "step": 79500 + }, + { + "epoch": 0.2164693072002178, + "grad_norm": 2.262817859649658, + "learning_rate": 1.5692026713915773e-05, + "loss": 2.5028802871704103, + "step": 79520 + }, + { + "epoch": 0.2165237511909623, + "grad_norm": 2.2145817279815674, + "learning_rate": 1.5690936350006815e-05, + "loss": 2.496740531921387, + "step": 79540 + }, + { + "epoch": 0.21657819518170682, + "grad_norm": 2.2005715370178223, + "learning_rate": 1.568984598609786e-05, + "loss": 2.497403144836426, + "step": 79560 + }, + { + "epoch": 0.21663263917245135, + "grad_norm": 2.2621893882751465, + "learning_rate": 1.5688755622188907e-05, + "loss": 2.515802192687988, + "step": 79580 + }, + { + "epoch": 0.21668708316319588, + "grad_norm": 2.050450325012207, + "learning_rate": 1.5687665258279953e-05, + "loss": 2.5266511917114256, + "step": 79600 + }, + { + "epoch": 0.21674152715394038, + "grad_norm": 2.1137733459472656, + "learning_rate": 1.5686574894371e-05, + "loss": 2.5270254135131838, + "step": 79620 + }, + { + "epoch": 0.2167959711446849, + "grad_norm": 2.217383623123169, + "learning_rate": 1.5685484530462044e-05, + "loss": 2.4961145401000975, + "step": 79640 + }, + { + "epoch": 0.21685041513542944, + "grad_norm": 2.139624834060669, + "learning_rate": 1.5684394166553087e-05, + "loss": 2.4958368301391602, + "step": 79660 + }, + { + "epoch": 0.21690485912617394, + "grad_norm": 2.3778791427612305, + "learning_rate": 1.5683303802644136e-05, + "loss": 2.528110122680664, + "step": 79680 + }, + { + "epoch": 0.21695930311691847, + "grad_norm": 2.1531882286071777, + "learning_rate": 1.5682213438735178e-05, + "loss": 2.5318981170654298, + "step": 79700 + }, + { + "epoch": 0.217013747107663, + "grad_norm": 2.1658284664154053, + "learning_rate": 1.5681123074826224e-05, + "loss": 2.5198925018310545, + "step": 79720 + }, + { + "epoch": 0.21706819109840753, + "grad_norm": 2.1848208904266357, + "learning_rate": 1.568003271091727e-05, + "loss": 2.5151578903198244, + "step": 79740 + }, + { + "epoch": 0.21712263508915203, + "grad_norm": 2.1375339031219482, + "learning_rate": 1.5678942347008316e-05, + "loss": 2.5184795379638674, + "step": 79760 + }, + { + "epoch": 0.21717707907989656, + "grad_norm": 3.3082199096679688, + "learning_rate": 1.567785198309936e-05, + "loss": 2.4972856521606444, + "step": 79780 + }, + { + "epoch": 0.21723152307064109, + "grad_norm": 2.1136882305145264, + "learning_rate": 1.5676761619190407e-05, + "loss": 2.542004203796387, + "step": 79800 + }, + { + "epoch": 0.2172859670613856, + "grad_norm": 2.1316165924072266, + "learning_rate": 1.567567125528145e-05, + "loss": 2.510283660888672, + "step": 79820 + }, + { + "epoch": 0.21734041105213012, + "grad_norm": 2.270200729370117, + "learning_rate": 1.56745808913725e-05, + "loss": 2.542050361633301, + "step": 79840 + }, + { + "epoch": 0.21739485504287465, + "grad_norm": 2.1216039657592773, + "learning_rate": 1.567349052746354e-05, + "loss": 2.4799625396728517, + "step": 79860 + }, + { + "epoch": 0.21744929903361918, + "grad_norm": 2.3233048915863037, + "learning_rate": 1.5672400163554587e-05, + "loss": 2.526564025878906, + "step": 79880 + }, + { + "epoch": 0.21750374302436368, + "grad_norm": 2.2128119468688965, + "learning_rate": 1.5671309799645633e-05, + "loss": 2.5146085739135744, + "step": 79900 + }, + { + "epoch": 0.2175581870151082, + "grad_norm": 2.0698728561401367, + "learning_rate": 1.567021943573668e-05, + "loss": 2.510046195983887, + "step": 79920 + }, + { + "epoch": 0.21761263100585274, + "grad_norm": 2.083660364151001, + "learning_rate": 1.5669129071827725e-05, + "loss": 2.5510446548461916, + "step": 79940 + }, + { + "epoch": 0.21766707499659724, + "grad_norm": 2.1252589225769043, + "learning_rate": 1.566803870791877e-05, + "loss": 2.503876876831055, + "step": 79960 + }, + { + "epoch": 0.21772151898734177, + "grad_norm": 2.042055130004883, + "learning_rate": 1.5666948344009813e-05, + "loss": 2.536336326599121, + "step": 79980 + }, + { + "epoch": 0.2177759629780863, + "grad_norm": 2.3864097595214844, + "learning_rate": 1.5665857980100862e-05, + "loss": 2.5322265625, + "step": 80000 + }, + { + "epoch": 0.21783040696883083, + "grad_norm": 2.1610488891601562, + "learning_rate": 1.5664767616191904e-05, + "loss": 2.553268814086914, + "step": 80020 + }, + { + "epoch": 0.21788485095957533, + "grad_norm": 2.4178597927093506, + "learning_rate": 1.566367725228295e-05, + "loss": 2.5147188186645506, + "step": 80040 + }, + { + "epoch": 0.21793929495031986, + "grad_norm": 2.1531288623809814, + "learning_rate": 1.5662586888373996e-05, + "loss": 2.557012939453125, + "step": 80060 + }, + { + "epoch": 0.21799373894106439, + "grad_norm": 2.109677791595459, + "learning_rate": 1.5661496524465042e-05, + "loss": 2.524235153198242, + "step": 80080 + }, + { + "epoch": 0.21804818293180891, + "grad_norm": 2.1541924476623535, + "learning_rate": 1.5660406160556088e-05, + "loss": 2.516292953491211, + "step": 80100 + }, + { + "epoch": 0.21810262692255342, + "grad_norm": 2.1326019763946533, + "learning_rate": 1.5659315796647133e-05, + "loss": 2.5298734664916993, + "step": 80120 + }, + { + "epoch": 0.21815707091329795, + "grad_norm": 2.1099231243133545, + "learning_rate": 1.5658225432738176e-05, + "loss": 2.525049018859863, + "step": 80140 + }, + { + "epoch": 0.21821151490404248, + "grad_norm": 2.1315860748291016, + "learning_rate": 1.5657135068829225e-05, + "loss": 2.5339143753051756, + "step": 80160 + }, + { + "epoch": 0.21826595889478698, + "grad_norm": 2.559943437576294, + "learning_rate": 1.5656044704920267e-05, + "loss": 2.5030723571777345, + "step": 80180 + }, + { + "epoch": 0.2183204028855315, + "grad_norm": 2.15580677986145, + "learning_rate": 1.5654954341011313e-05, + "loss": 2.4787715911865233, + "step": 80200 + }, + { + "epoch": 0.21837484687627604, + "grad_norm": 2.093804121017456, + "learning_rate": 1.565386397710236e-05, + "loss": 2.525823211669922, + "step": 80220 + }, + { + "epoch": 0.21842929086702056, + "grad_norm": 2.2540817260742188, + "learning_rate": 1.5652773613193405e-05, + "loss": 2.5220264434814452, + "step": 80240 + }, + { + "epoch": 0.21848373485776507, + "grad_norm": 2.220149040222168, + "learning_rate": 1.565168324928445e-05, + "loss": 2.537357521057129, + "step": 80260 + }, + { + "epoch": 0.2185381788485096, + "grad_norm": 2.275660991668701, + "learning_rate": 1.5650592885375497e-05, + "loss": 2.532044219970703, + "step": 80280 + }, + { + "epoch": 0.21859262283925412, + "grad_norm": 2.0482237339019775, + "learning_rate": 1.564950252146654e-05, + "loss": 2.5245840072631838, + "step": 80300 + }, + { + "epoch": 0.21864706682999863, + "grad_norm": 2.357417345046997, + "learning_rate": 1.5648412157557588e-05, + "loss": 2.5058752059936524, + "step": 80320 + }, + { + "epoch": 0.21870151082074316, + "grad_norm": 2.2107279300689697, + "learning_rate": 1.564732179364863e-05, + "loss": 2.5354665756225585, + "step": 80340 + }, + { + "epoch": 0.21875595481148769, + "grad_norm": 2.2597622871398926, + "learning_rate": 1.5646231429739676e-05, + "loss": 2.54580078125, + "step": 80360 + }, + { + "epoch": 0.21881039880223221, + "grad_norm": 2.0367443561553955, + "learning_rate": 1.5645141065830722e-05, + "loss": 2.508610725402832, + "step": 80380 + }, + { + "epoch": 0.21886484279297672, + "grad_norm": 2.1936092376708984, + "learning_rate": 1.5644050701921768e-05, + "loss": 2.526990509033203, + "step": 80400 + }, + { + "epoch": 0.21891928678372125, + "grad_norm": 2.1828784942626953, + "learning_rate": 1.5642960338012814e-05, + "loss": 2.5077182769775392, + "step": 80420 + }, + { + "epoch": 0.21897373077446577, + "grad_norm": 2.116558790206909, + "learning_rate": 1.564186997410386e-05, + "loss": 2.532755470275879, + "step": 80440 + }, + { + "epoch": 0.21902817476521028, + "grad_norm": 2.1271533966064453, + "learning_rate": 1.5640779610194902e-05, + "loss": 2.537251091003418, + "step": 80460 + }, + { + "epoch": 0.2190826187559548, + "grad_norm": 2.103339672088623, + "learning_rate": 1.5639689246285948e-05, + "loss": 2.524549674987793, + "step": 80480 + }, + { + "epoch": 0.21913706274669933, + "grad_norm": 2.134427785873413, + "learning_rate": 1.5638598882376994e-05, + "loss": 2.5482627868652346, + "step": 80500 + }, + { + "epoch": 0.21919150673744386, + "grad_norm": 2.296041488647461, + "learning_rate": 1.563750851846804e-05, + "loss": 2.530900001525879, + "step": 80520 + }, + { + "epoch": 0.21924595072818837, + "grad_norm": 2.2201716899871826, + "learning_rate": 1.5636418154559085e-05, + "loss": 2.538613128662109, + "step": 80540 + }, + { + "epoch": 0.2193003947189329, + "grad_norm": 2.100478172302246, + "learning_rate": 1.563532779065013e-05, + "loss": 2.507364845275879, + "step": 80560 + }, + { + "epoch": 0.21935483870967742, + "grad_norm": 2.1212315559387207, + "learning_rate": 1.5634237426741177e-05, + "loss": 2.5237154006958007, + "step": 80580 + }, + { + "epoch": 0.21940928270042195, + "grad_norm": 2.271599292755127, + "learning_rate": 1.5633147062832223e-05, + "loss": 2.5167518615722657, + "step": 80600 + }, + { + "epoch": 0.21946372669116646, + "grad_norm": 2.1539833545684814, + "learning_rate": 1.5632056698923265e-05, + "loss": 2.5186717987060545, + "step": 80620 + }, + { + "epoch": 0.21951817068191098, + "grad_norm": 2.128783702850342, + "learning_rate": 1.563096633501431e-05, + "loss": 2.5176462173461913, + "step": 80640 + }, + { + "epoch": 0.21957261467265551, + "grad_norm": 2.167884588241577, + "learning_rate": 1.5629875971105357e-05, + "loss": 2.530973434448242, + "step": 80660 + }, + { + "epoch": 0.21962705866340002, + "grad_norm": 2.15104079246521, + "learning_rate": 1.5628785607196403e-05, + "loss": 2.506561851501465, + "step": 80680 + }, + { + "epoch": 0.21968150265414454, + "grad_norm": 2.178234815597534, + "learning_rate": 1.562769524328745e-05, + "loss": 2.5508337020874023, + "step": 80700 + }, + { + "epoch": 0.21973594664488907, + "grad_norm": 2.166710615158081, + "learning_rate": 1.5626604879378494e-05, + "loss": 2.5337282180786134, + "step": 80720 + }, + { + "epoch": 0.2197903906356336, + "grad_norm": 2.197516441345215, + "learning_rate": 1.562551451546954e-05, + "loss": 2.4800939559936523, + "step": 80740 + }, + { + "epoch": 0.2198448346263781, + "grad_norm": 2.1046910285949707, + "learning_rate": 1.5624424151560586e-05, + "loss": 2.5401798248291017, + "step": 80760 + }, + { + "epoch": 0.21989927861712263, + "grad_norm": 2.1627559661865234, + "learning_rate": 1.562333378765163e-05, + "loss": 2.500405120849609, + "step": 80780 + }, + { + "epoch": 0.21995372260786716, + "grad_norm": 2.052217960357666, + "learning_rate": 1.5622243423742674e-05, + "loss": 2.5347736358642576, + "step": 80800 + }, + { + "epoch": 0.22000816659861167, + "grad_norm": 2.2170863151550293, + "learning_rate": 1.562115305983372e-05, + "loss": 2.529354476928711, + "step": 80820 + }, + { + "epoch": 0.2200626105893562, + "grad_norm": 2.216092109680176, + "learning_rate": 1.5620062695924766e-05, + "loss": 2.529708671569824, + "step": 80840 + }, + { + "epoch": 0.22011705458010072, + "grad_norm": 2.2421960830688477, + "learning_rate": 1.561897233201581e-05, + "loss": 2.5408477783203125, + "step": 80860 + }, + { + "epoch": 0.22017149857084525, + "grad_norm": 2.1084327697753906, + "learning_rate": 1.5617881968106857e-05, + "loss": 2.5254255294799806, + "step": 80880 + }, + { + "epoch": 0.22022594256158975, + "grad_norm": 2.1688613891601562, + "learning_rate": 1.5616791604197903e-05, + "loss": 2.5273458480834963, + "step": 80900 + }, + { + "epoch": 0.22028038655233428, + "grad_norm": 2.042738437652588, + "learning_rate": 1.561570124028895e-05, + "loss": 2.5233068466186523, + "step": 80920 + }, + { + "epoch": 0.2203348305430788, + "grad_norm": 2.1848809719085693, + "learning_rate": 1.5614610876379995e-05, + "loss": 2.5282947540283205, + "step": 80940 + }, + { + "epoch": 0.22038927453382334, + "grad_norm": 2.1000149250030518, + "learning_rate": 1.5613520512471037e-05, + "loss": 2.478535270690918, + "step": 80960 + }, + { + "epoch": 0.22044371852456784, + "grad_norm": 2.2060039043426514, + "learning_rate": 1.5612430148562083e-05, + "loss": 2.539739418029785, + "step": 80980 + }, + { + "epoch": 0.22049816251531237, + "grad_norm": 2.140516757965088, + "learning_rate": 1.561133978465313e-05, + "loss": 2.5187124252319335, + "step": 81000 + }, + { + "epoch": 0.2205526065060569, + "grad_norm": 2.1908650398254395, + "learning_rate": 1.5610249420744175e-05, + "loss": 2.536080741882324, + "step": 81020 + }, + { + "epoch": 0.2206070504968014, + "grad_norm": 2.1104767322540283, + "learning_rate": 1.560915905683522e-05, + "loss": 2.493904876708984, + "step": 81040 + }, + { + "epoch": 0.22066149448754593, + "grad_norm": 2.141505718231201, + "learning_rate": 1.5608068692926266e-05, + "loss": 2.506025695800781, + "step": 81060 + }, + { + "epoch": 0.22071593847829046, + "grad_norm": 2.103721857070923, + "learning_rate": 1.5606978329017312e-05, + "loss": 2.529403495788574, + "step": 81080 + }, + { + "epoch": 0.220770382469035, + "grad_norm": 2.4244325160980225, + "learning_rate": 1.5605887965108358e-05, + "loss": 2.515995979309082, + "step": 81100 + }, + { + "epoch": 0.2208248264597795, + "grad_norm": 2.1372108459472656, + "learning_rate": 1.56047976011994e-05, + "loss": 2.5139719009399415, + "step": 81120 + }, + { + "epoch": 0.22087927045052402, + "grad_norm": 2.17976975440979, + "learning_rate": 1.5603707237290446e-05, + "loss": 2.527039909362793, + "step": 81140 + }, + { + "epoch": 0.22093371444126855, + "grad_norm": 2.1266374588012695, + "learning_rate": 1.5602616873381492e-05, + "loss": 2.540962791442871, + "step": 81160 + }, + { + "epoch": 0.22098815843201305, + "grad_norm": 2.1192426681518555, + "learning_rate": 1.5601526509472538e-05, + "loss": 2.5244256973266603, + "step": 81180 + }, + { + "epoch": 0.22104260242275758, + "grad_norm": 2.152367353439331, + "learning_rate": 1.5600436145563583e-05, + "loss": 2.513475227355957, + "step": 81200 + }, + { + "epoch": 0.2210970464135021, + "grad_norm": 2.1258440017700195, + "learning_rate": 1.559934578165463e-05, + "loss": 2.5157026290893554, + "step": 81220 + }, + { + "epoch": 0.22115149040424664, + "grad_norm": 2.174116611480713, + "learning_rate": 1.5598255417745675e-05, + "loss": 2.5053388595581056, + "step": 81240 + }, + { + "epoch": 0.22120593439499114, + "grad_norm": 2.2362492084503174, + "learning_rate": 1.559716505383672e-05, + "loss": 2.5252735137939455, + "step": 81260 + }, + { + "epoch": 0.22126037838573567, + "grad_norm": 2.207479238510132, + "learning_rate": 1.5596074689927763e-05, + "loss": 2.5294557571411134, + "step": 81280 + }, + { + "epoch": 0.2213148223764802, + "grad_norm": 1.9935873746871948, + "learning_rate": 1.559498432601881e-05, + "loss": 2.525482177734375, + "step": 81300 + }, + { + "epoch": 0.2213692663672247, + "grad_norm": 2.235891103744507, + "learning_rate": 1.5593893962109855e-05, + "loss": 2.4926279067993162, + "step": 81320 + }, + { + "epoch": 0.22142371035796923, + "grad_norm": 2.126182794570923, + "learning_rate": 1.55928035982009e-05, + "loss": 2.5318878173828123, + "step": 81340 + }, + { + "epoch": 0.22147815434871376, + "grad_norm": 2.1402225494384766, + "learning_rate": 1.5591713234291947e-05, + "loss": 2.519215202331543, + "step": 81360 + }, + { + "epoch": 0.2215325983394583, + "grad_norm": 2.1612985134124756, + "learning_rate": 1.5590622870382992e-05, + "loss": 2.5099309921264648, + "step": 81380 + }, + { + "epoch": 0.2215870423302028, + "grad_norm": 2.1150577068328857, + "learning_rate": 1.5589532506474038e-05, + "loss": 2.4985782623291017, + "step": 81400 + }, + { + "epoch": 0.22164148632094732, + "grad_norm": 2.131103277206421, + "learning_rate": 1.5588442142565084e-05, + "loss": 2.526461601257324, + "step": 81420 + }, + { + "epoch": 0.22169593031169185, + "grad_norm": 2.8264317512512207, + "learning_rate": 1.5587351778656126e-05, + "loss": 2.4994930267333983, + "step": 81440 + }, + { + "epoch": 0.22175037430243638, + "grad_norm": 2.1958119869232178, + "learning_rate": 1.5586261414747172e-05, + "loss": 2.510288429260254, + "step": 81460 + }, + { + "epoch": 0.22180481829318088, + "grad_norm": 2.0532515048980713, + "learning_rate": 1.5585171050838218e-05, + "loss": 2.478800392150879, + "step": 81480 + }, + { + "epoch": 0.2218592622839254, + "grad_norm": 2.1620066165924072, + "learning_rate": 1.5584080686929264e-05, + "loss": 2.518246078491211, + "step": 81500 + }, + { + "epoch": 0.22191370627466994, + "grad_norm": 2.099398374557495, + "learning_rate": 1.558299032302031e-05, + "loss": 2.5388044357299804, + "step": 81520 + }, + { + "epoch": 0.22196815026541444, + "grad_norm": 2.1223509311676025, + "learning_rate": 1.5581899959111355e-05, + "loss": 2.5095804214477537, + "step": 81540 + }, + { + "epoch": 0.22202259425615897, + "grad_norm": 2.171562671661377, + "learning_rate": 1.55808095952024e-05, + "loss": 2.4887876510620117, + "step": 81560 + }, + { + "epoch": 0.2220770382469035, + "grad_norm": 2.3033783435821533, + "learning_rate": 1.5579719231293447e-05, + "loss": 2.50380802154541, + "step": 81580 + }, + { + "epoch": 0.22213148223764803, + "grad_norm": 2.0146234035491943, + "learning_rate": 1.557862886738449e-05, + "loss": 2.516248893737793, + "step": 81600 + }, + { + "epoch": 0.22218592622839253, + "grad_norm": 2.153663396835327, + "learning_rate": 1.5577538503475535e-05, + "loss": 2.5616411209106444, + "step": 81620 + }, + { + "epoch": 0.22224037021913706, + "grad_norm": 2.1227588653564453, + "learning_rate": 1.557644813956658e-05, + "loss": 2.492200469970703, + "step": 81640 + }, + { + "epoch": 0.2222948142098816, + "grad_norm": 2.1195473670959473, + "learning_rate": 1.5575357775657627e-05, + "loss": 2.5477584838867187, + "step": 81660 + }, + { + "epoch": 0.2223492582006261, + "grad_norm": 2.1061627864837646, + "learning_rate": 1.5574267411748673e-05, + "loss": 2.535223197937012, + "step": 81680 + }, + { + "epoch": 0.22240370219137062, + "grad_norm": 2.7347359657287598, + "learning_rate": 1.557317704783972e-05, + "loss": 2.5337125778198244, + "step": 81700 + }, + { + "epoch": 0.22245814618211515, + "grad_norm": 2.0972700119018555, + "learning_rate": 1.5572086683930764e-05, + "loss": 2.532636833190918, + "step": 81720 + }, + { + "epoch": 0.22251259017285968, + "grad_norm": 2.272225856781006, + "learning_rate": 1.557099632002181e-05, + "loss": 2.4970680236816407, + "step": 81740 + }, + { + "epoch": 0.22256703416360418, + "grad_norm": 2.12394642829895, + "learning_rate": 1.5569905956112853e-05, + "loss": 2.557442474365234, + "step": 81760 + }, + { + "epoch": 0.2226214781543487, + "grad_norm": 2.2327520847320557, + "learning_rate": 1.55688155922039e-05, + "loss": 2.5379878997802736, + "step": 81780 + }, + { + "epoch": 0.22267592214509324, + "grad_norm": 2.247077465057373, + "learning_rate": 1.5567725228294944e-05, + "loss": 2.5250267028808593, + "step": 81800 + }, + { + "epoch": 0.22273036613583777, + "grad_norm": 2.109722137451172, + "learning_rate": 1.556663486438599e-05, + "loss": 2.535160255432129, + "step": 81820 + }, + { + "epoch": 0.22278481012658227, + "grad_norm": 2.0986485481262207, + "learning_rate": 1.5565544500477036e-05, + "loss": 2.4547033309936523, + "step": 81840 + }, + { + "epoch": 0.2228392541173268, + "grad_norm": 2.065016746520996, + "learning_rate": 1.556445413656808e-05, + "loss": 2.526516151428223, + "step": 81860 + }, + { + "epoch": 0.22289369810807133, + "grad_norm": 2.977057933807373, + "learning_rate": 1.5563363772659127e-05, + "loss": 2.5299667358398437, + "step": 81880 + }, + { + "epoch": 0.22294814209881583, + "grad_norm": 2.267448902130127, + "learning_rate": 1.5562273408750173e-05, + "loss": 2.5266918182373046, + "step": 81900 + }, + { + "epoch": 0.22300258608956036, + "grad_norm": 2.2424569129943848, + "learning_rate": 1.5561183044841216e-05, + "loss": 2.530036735534668, + "step": 81920 + }, + { + "epoch": 0.2230570300803049, + "grad_norm": 2.1004443168640137, + "learning_rate": 1.556009268093226e-05, + "loss": 2.5223613739013673, + "step": 81940 + }, + { + "epoch": 0.22311147407104942, + "grad_norm": 2.1959757804870605, + "learning_rate": 1.5559002317023307e-05, + "loss": 2.5157955169677733, + "step": 81960 + }, + { + "epoch": 0.22316591806179392, + "grad_norm": 2.270277261734009, + "learning_rate": 1.5557911953114353e-05, + "loss": 2.487788200378418, + "step": 81980 + }, + { + "epoch": 0.22322036205253845, + "grad_norm": 2.1202619075775146, + "learning_rate": 1.55568215892054e-05, + "loss": 2.515464210510254, + "step": 82000 + }, + { + "epoch": 0.22327480604328298, + "grad_norm": 2.1704742908477783, + "learning_rate": 1.5555731225296445e-05, + "loss": 2.527575874328613, + "step": 82020 + }, + { + "epoch": 0.22332925003402748, + "grad_norm": 2.181070327758789, + "learning_rate": 1.555464086138749e-05, + "loss": 2.5433725357055663, + "step": 82040 + }, + { + "epoch": 0.223383694024772, + "grad_norm": 2.128645181655884, + "learning_rate": 1.5553550497478536e-05, + "loss": 2.508494758605957, + "step": 82060 + }, + { + "epoch": 0.22343813801551654, + "grad_norm": 2.1959726810455322, + "learning_rate": 1.555246013356958e-05, + "loss": 2.522114944458008, + "step": 82080 + }, + { + "epoch": 0.22349258200626107, + "grad_norm": 2.2042486667633057, + "learning_rate": 1.5551369769660625e-05, + "loss": 2.5455705642700197, + "step": 82100 + }, + { + "epoch": 0.22354702599700557, + "grad_norm": 2.1954476833343506, + "learning_rate": 1.555027940575167e-05, + "loss": 2.5275857925415037, + "step": 82120 + }, + { + "epoch": 0.2236014699877501, + "grad_norm": 2.1557939052581787, + "learning_rate": 1.5549189041842716e-05, + "loss": 2.526142120361328, + "step": 82140 + }, + { + "epoch": 0.22365591397849463, + "grad_norm": 2.3602232933044434, + "learning_rate": 1.5548098677933762e-05, + "loss": 2.503706169128418, + "step": 82160 + }, + { + "epoch": 0.22371035796923913, + "grad_norm": 2.148503541946411, + "learning_rate": 1.5547008314024808e-05, + "loss": 2.5331546783447267, + "step": 82180 + }, + { + "epoch": 0.22376480195998366, + "grad_norm": 2.204132556915283, + "learning_rate": 1.5545917950115854e-05, + "loss": 2.5394262313842773, + "step": 82200 + }, + { + "epoch": 0.2238192459507282, + "grad_norm": 2.117168426513672, + "learning_rate": 1.55448275862069e-05, + "loss": 2.531295967102051, + "step": 82220 + }, + { + "epoch": 0.22387368994147272, + "grad_norm": 2.2000396251678467, + "learning_rate": 1.5543737222297942e-05, + "loss": 2.5060611724853517, + "step": 82240 + }, + { + "epoch": 0.22392813393221722, + "grad_norm": 2.2649648189544678, + "learning_rate": 1.5542646858388988e-05, + "loss": 2.524626922607422, + "step": 82260 + }, + { + "epoch": 0.22398257792296175, + "grad_norm": 2.3429114818573, + "learning_rate": 1.5541556494480033e-05, + "loss": 2.521440887451172, + "step": 82280 + }, + { + "epoch": 0.22403702191370628, + "grad_norm": 2.0665628910064697, + "learning_rate": 1.554046613057108e-05, + "loss": 2.5335031509399415, + "step": 82300 + }, + { + "epoch": 0.2240914659044508, + "grad_norm": 2.0631752014160156, + "learning_rate": 1.5539375766662125e-05, + "loss": 2.5428117752075194, + "step": 82320 + }, + { + "epoch": 0.2241459098951953, + "grad_norm": 2.139164447784424, + "learning_rate": 1.553828540275317e-05, + "loss": 2.525852394104004, + "step": 82340 + }, + { + "epoch": 0.22420035388593984, + "grad_norm": 2.114609956741333, + "learning_rate": 1.5537195038844217e-05, + "loss": 2.5050745010375977, + "step": 82360 + }, + { + "epoch": 0.22425479787668437, + "grad_norm": 2.1515121459960938, + "learning_rate": 1.5536104674935262e-05, + "loss": 2.5298322677612304, + "step": 82380 + }, + { + "epoch": 0.22430924186742887, + "grad_norm": 2.1065549850463867, + "learning_rate": 1.5535014311026305e-05, + "loss": 2.509175109863281, + "step": 82400 + }, + { + "epoch": 0.2243636858581734, + "grad_norm": 2.0662217140197754, + "learning_rate": 1.553392394711735e-05, + "loss": 2.5108442306518555, + "step": 82420 + }, + { + "epoch": 0.22441812984891793, + "grad_norm": 2.3704984188079834, + "learning_rate": 1.5532833583208396e-05, + "loss": 2.493231201171875, + "step": 82440 + }, + { + "epoch": 0.22447257383966246, + "grad_norm": 2.077695846557617, + "learning_rate": 1.5531743219299442e-05, + "loss": 2.5124237060546877, + "step": 82460 + }, + { + "epoch": 0.22452701783040696, + "grad_norm": 2.147047996520996, + "learning_rate": 1.5530652855390488e-05, + "loss": 2.525053024291992, + "step": 82480 + }, + { + "epoch": 0.2245814618211515, + "grad_norm": 2.763442277908325, + "learning_rate": 1.5529562491481534e-05, + "loss": 2.512590026855469, + "step": 82500 + }, + { + "epoch": 0.22463590581189602, + "grad_norm": 2.121410608291626, + "learning_rate": 1.552847212757258e-05, + "loss": 2.481728935241699, + "step": 82520 + }, + { + "epoch": 0.22469034980264052, + "grad_norm": 2.344472646713257, + "learning_rate": 1.5527381763663626e-05, + "loss": 2.5260330200195313, + "step": 82540 + }, + { + "epoch": 0.22474479379338505, + "grad_norm": 2.1695780754089355, + "learning_rate": 1.5526291399754668e-05, + "loss": 2.521560287475586, + "step": 82560 + }, + { + "epoch": 0.22479923778412958, + "grad_norm": 2.1641411781311035, + "learning_rate": 1.5525201035845714e-05, + "loss": 2.53814754486084, + "step": 82580 + }, + { + "epoch": 0.2248536817748741, + "grad_norm": 2.183868408203125, + "learning_rate": 1.552411067193676e-05, + "loss": 2.5018239974975587, + "step": 82600 + }, + { + "epoch": 0.2249081257656186, + "grad_norm": 2.284677743911743, + "learning_rate": 1.5523020308027805e-05, + "loss": 2.5103240966796876, + "step": 82620 + }, + { + "epoch": 0.22496256975636314, + "grad_norm": 2.175452947616577, + "learning_rate": 1.552192994411885e-05, + "loss": 2.5422994613647463, + "step": 82640 + }, + { + "epoch": 0.22501701374710767, + "grad_norm": 2.109999895095825, + "learning_rate": 1.5520839580209897e-05, + "loss": 2.4959522247314454, + "step": 82660 + }, + { + "epoch": 0.22507145773785217, + "grad_norm": 2.197016716003418, + "learning_rate": 1.551974921630094e-05, + "loss": 2.535214424133301, + "step": 82680 + }, + { + "epoch": 0.2251259017285967, + "grad_norm": 2.166193723678589, + "learning_rate": 1.551865885239199e-05, + "loss": 2.536221504211426, + "step": 82700 + }, + { + "epoch": 0.22518034571934123, + "grad_norm": 2.1093597412109375, + "learning_rate": 1.551756848848303e-05, + "loss": 2.4994022369384767, + "step": 82720 + }, + { + "epoch": 0.22523478971008576, + "grad_norm": 2.0773136615753174, + "learning_rate": 1.5516478124574077e-05, + "loss": 2.4990877151489257, + "step": 82740 + }, + { + "epoch": 0.22528923370083026, + "grad_norm": 2.0785069465637207, + "learning_rate": 1.5515387760665123e-05, + "loss": 2.518459129333496, + "step": 82760 + }, + { + "epoch": 0.2253436776915748, + "grad_norm": 2.1129274368286133, + "learning_rate": 1.551429739675617e-05, + "loss": 2.496593475341797, + "step": 82780 + }, + { + "epoch": 0.22539812168231932, + "grad_norm": 2.120610237121582, + "learning_rate": 1.5513207032847214e-05, + "loss": 2.5043291091918944, + "step": 82800 + }, + { + "epoch": 0.22545256567306385, + "grad_norm": 2.1599392890930176, + "learning_rate": 1.551211666893826e-05, + "loss": 2.5233186721801757, + "step": 82820 + }, + { + "epoch": 0.22550700966380835, + "grad_norm": 2.173276662826538, + "learning_rate": 1.5511026305029303e-05, + "loss": 2.4951288223266603, + "step": 82840 + }, + { + "epoch": 0.22556145365455288, + "grad_norm": 2.0531654357910156, + "learning_rate": 1.5509935941120352e-05, + "loss": 2.5330909729003905, + "step": 82860 + }, + { + "epoch": 0.2256158976452974, + "grad_norm": 2.196868419647217, + "learning_rate": 1.5508845577211394e-05, + "loss": 2.5166906356811523, + "step": 82880 + }, + { + "epoch": 0.2256703416360419, + "grad_norm": 2.0296382904052734, + "learning_rate": 1.550775521330244e-05, + "loss": 2.508697509765625, + "step": 82900 + }, + { + "epoch": 0.22572478562678644, + "grad_norm": 2.030858278274536, + "learning_rate": 1.5506664849393486e-05, + "loss": 2.4954090118408203, + "step": 82920 + }, + { + "epoch": 0.22577922961753097, + "grad_norm": 2.150301218032837, + "learning_rate": 1.550557448548453e-05, + "loss": 2.5233360290527345, + "step": 82940 + }, + { + "epoch": 0.2258336736082755, + "grad_norm": 2.300541400909424, + "learning_rate": 1.5504484121575577e-05, + "loss": 2.5111122131347656, + "step": 82960 + }, + { + "epoch": 0.22588811759902, + "grad_norm": 2.2053909301757812, + "learning_rate": 1.5503393757666623e-05, + "loss": 2.5087661743164062, + "step": 82980 + }, + { + "epoch": 0.22594256158976453, + "grad_norm": 2.0808849334716797, + "learning_rate": 1.5502303393757666e-05, + "loss": 2.532505226135254, + "step": 83000 + }, + { + "epoch": 0.22599700558050906, + "grad_norm": 2.1767020225524902, + "learning_rate": 1.5501213029848715e-05, + "loss": 2.535739707946777, + "step": 83020 + }, + { + "epoch": 0.22605144957125356, + "grad_norm": 2.1782259941101074, + "learning_rate": 1.5500122665939757e-05, + "loss": 2.500509834289551, + "step": 83040 + }, + { + "epoch": 0.2261058935619981, + "grad_norm": 2.1368985176086426, + "learning_rate": 1.5499032302030803e-05, + "loss": 2.497061920166016, + "step": 83060 + }, + { + "epoch": 0.22616033755274262, + "grad_norm": 2.178107976913452, + "learning_rate": 1.549794193812185e-05, + "loss": 2.5268688201904297, + "step": 83080 + }, + { + "epoch": 0.22621478154348715, + "grad_norm": 2.1649105548858643, + "learning_rate": 1.5496851574212895e-05, + "loss": 2.544669532775879, + "step": 83100 + }, + { + "epoch": 0.22626922553423165, + "grad_norm": 2.4026360511779785, + "learning_rate": 1.549576121030394e-05, + "loss": 2.535491943359375, + "step": 83120 + }, + { + "epoch": 0.22632366952497618, + "grad_norm": 2.1993751525878906, + "learning_rate": 1.5494670846394986e-05, + "loss": 2.4826005935668944, + "step": 83140 + }, + { + "epoch": 0.2263781135157207, + "grad_norm": 2.0721359252929688, + "learning_rate": 1.549358048248603e-05, + "loss": 2.5313085556030273, + "step": 83160 + }, + { + "epoch": 0.22643255750646524, + "grad_norm": 2.2058568000793457, + "learning_rate": 1.5492490118577078e-05, + "loss": 2.5007793426513674, + "step": 83180 + }, + { + "epoch": 0.22648700149720974, + "grad_norm": 2.1510262489318848, + "learning_rate": 1.549139975466812e-05, + "loss": 2.5426919937133787, + "step": 83200 + }, + { + "epoch": 0.22654144548795427, + "grad_norm": 2.065234661102295, + "learning_rate": 1.5490309390759166e-05, + "loss": 2.4993852615356444, + "step": 83220 + }, + { + "epoch": 0.2265958894786988, + "grad_norm": 2.392896890640259, + "learning_rate": 1.5489219026850212e-05, + "loss": 2.539908027648926, + "step": 83240 + }, + { + "epoch": 0.2266503334694433, + "grad_norm": 2.1799964904785156, + "learning_rate": 1.5488128662941258e-05, + "loss": 2.516154670715332, + "step": 83260 + }, + { + "epoch": 0.22670477746018783, + "grad_norm": 2.1399059295654297, + "learning_rate": 1.5487038299032304e-05, + "loss": 2.5334638595581054, + "step": 83280 + }, + { + "epoch": 0.22675922145093236, + "grad_norm": 2.105072498321533, + "learning_rate": 1.548594793512335e-05, + "loss": 2.5084320068359376, + "step": 83300 + }, + { + "epoch": 0.2268136654416769, + "grad_norm": 2.2482714653015137, + "learning_rate": 1.5484857571214392e-05, + "loss": 2.5393339157104493, + "step": 83320 + }, + { + "epoch": 0.2268681094324214, + "grad_norm": 2.3402421474456787, + "learning_rate": 1.548376720730544e-05, + "loss": 2.527263069152832, + "step": 83340 + }, + { + "epoch": 0.22692255342316592, + "grad_norm": 2.1017303466796875, + "learning_rate": 1.5482676843396483e-05, + "loss": 2.499069595336914, + "step": 83360 + }, + { + "epoch": 0.22697699741391045, + "grad_norm": 2.258286476135254, + "learning_rate": 1.548158647948753e-05, + "loss": 2.5454288482666017, + "step": 83380 + }, + { + "epoch": 0.22703144140465495, + "grad_norm": 2.0574722290039062, + "learning_rate": 1.5480496115578575e-05, + "loss": 2.502461051940918, + "step": 83400 + }, + { + "epoch": 0.22708588539539948, + "grad_norm": 2.123807668685913, + "learning_rate": 1.547940575166962e-05, + "loss": 2.533389663696289, + "step": 83420 + }, + { + "epoch": 0.227140329386144, + "grad_norm": 2.2886605262756348, + "learning_rate": 1.5478315387760667e-05, + "loss": 2.503432846069336, + "step": 83440 + }, + { + "epoch": 0.22719477337688854, + "grad_norm": 2.237046241760254, + "learning_rate": 1.5477225023851712e-05, + "loss": 2.5429386138916015, + "step": 83460 + }, + { + "epoch": 0.22724921736763304, + "grad_norm": 2.5355775356292725, + "learning_rate": 1.5476134659942755e-05, + "loss": 2.5166563034057616, + "step": 83480 + }, + { + "epoch": 0.22730366135837757, + "grad_norm": 2.040393114089966, + "learning_rate": 1.5475044296033804e-05, + "loss": 2.506442832946777, + "step": 83500 + }, + { + "epoch": 0.2273581053491221, + "grad_norm": 2.2221503257751465, + "learning_rate": 1.5473953932124846e-05, + "loss": 2.4751100540161133, + "step": 83520 + }, + { + "epoch": 0.2274125493398666, + "grad_norm": 1.994104027748108, + "learning_rate": 1.5472863568215892e-05, + "loss": 2.5526071548461915, + "step": 83540 + }, + { + "epoch": 0.22746699333061113, + "grad_norm": 2.1730129718780518, + "learning_rate": 1.5471773204306938e-05, + "loss": 2.5144075393676757, + "step": 83560 + }, + { + "epoch": 0.22752143732135566, + "grad_norm": 2.246882915496826, + "learning_rate": 1.5470682840397984e-05, + "loss": 2.5375286102294923, + "step": 83580 + }, + { + "epoch": 0.2275758813121002, + "grad_norm": 2.183434009552002, + "learning_rate": 1.546959247648903e-05, + "loss": 2.526945877075195, + "step": 83600 + }, + { + "epoch": 0.2276303253028447, + "grad_norm": 2.176213264465332, + "learning_rate": 1.5468502112580076e-05, + "loss": 2.5243764877319337, + "step": 83620 + }, + { + "epoch": 0.22768476929358922, + "grad_norm": 2.5652003288269043, + "learning_rate": 1.5467411748671118e-05, + "loss": 2.542127799987793, + "step": 83640 + }, + { + "epoch": 0.22773921328433375, + "grad_norm": 2.0946497917175293, + "learning_rate": 1.5466321384762167e-05, + "loss": 2.5247655868530274, + "step": 83660 + }, + { + "epoch": 0.22779365727507828, + "grad_norm": 2.389598846435547, + "learning_rate": 1.546523102085321e-05, + "loss": 2.517954444885254, + "step": 83680 + }, + { + "epoch": 0.22784810126582278, + "grad_norm": 2.1078782081604004, + "learning_rate": 1.5464140656944255e-05, + "loss": 2.5061710357666014, + "step": 83700 + }, + { + "epoch": 0.2279025452565673, + "grad_norm": 2.1510720252990723, + "learning_rate": 1.54630502930353e-05, + "loss": 2.497239685058594, + "step": 83720 + }, + { + "epoch": 0.22795698924731184, + "grad_norm": 2.1747984886169434, + "learning_rate": 1.5461959929126347e-05, + "loss": 2.497853660583496, + "step": 83740 + }, + { + "epoch": 0.22801143323805634, + "grad_norm": 2.143885850906372, + "learning_rate": 1.5460869565217393e-05, + "loss": 2.537138748168945, + "step": 83760 + }, + { + "epoch": 0.22806587722880087, + "grad_norm": 2.258371114730835, + "learning_rate": 1.545977920130844e-05, + "loss": 2.505406379699707, + "step": 83780 + }, + { + "epoch": 0.2281203212195454, + "grad_norm": 2.2495365142822266, + "learning_rate": 1.545868883739948e-05, + "loss": 2.511842727661133, + "step": 83800 + }, + { + "epoch": 0.22817476521028993, + "grad_norm": 2.2251412868499756, + "learning_rate": 1.545759847349053e-05, + "loss": 2.4884963989257813, + "step": 83820 + }, + { + "epoch": 0.22822920920103443, + "grad_norm": 2.3328030109405518, + "learning_rate": 1.5456508109581573e-05, + "loss": 2.5048992156982424, + "step": 83840 + }, + { + "epoch": 0.22828365319177896, + "grad_norm": 2.1717100143432617, + "learning_rate": 1.545541774567262e-05, + "loss": 2.530280113220215, + "step": 83860 + }, + { + "epoch": 0.2283380971825235, + "grad_norm": 2.1161904335021973, + "learning_rate": 1.5454327381763664e-05, + "loss": 2.5180953979492187, + "step": 83880 + }, + { + "epoch": 0.228392541173268, + "grad_norm": 2.201936721801758, + "learning_rate": 1.545323701785471e-05, + "loss": 2.5086181640625, + "step": 83900 + }, + { + "epoch": 0.22844698516401252, + "grad_norm": 2.1227965354919434, + "learning_rate": 1.5452146653945756e-05, + "loss": 2.558370590209961, + "step": 83920 + }, + { + "epoch": 0.22850142915475705, + "grad_norm": 2.1157329082489014, + "learning_rate": 1.54510562900368e-05, + "loss": 2.5435422897338866, + "step": 83940 + }, + { + "epoch": 0.22855587314550158, + "grad_norm": 2.348206043243408, + "learning_rate": 1.5449965926127844e-05, + "loss": 2.5279329299926756, + "step": 83960 + }, + { + "epoch": 0.22861031713624608, + "grad_norm": 2.081409454345703, + "learning_rate": 1.5448875562218893e-05, + "loss": 2.4996997833251955, + "step": 83980 + }, + { + "epoch": 0.2286647611269906, + "grad_norm": 2.257620096206665, + "learning_rate": 1.5447785198309936e-05, + "loss": 2.515223503112793, + "step": 84000 + }, + { + "epoch": 0.22871920511773514, + "grad_norm": 2.1412742137908936, + "learning_rate": 1.544669483440098e-05, + "loss": 2.496454620361328, + "step": 84020 + }, + { + "epoch": 0.22877364910847967, + "grad_norm": 2.28544282913208, + "learning_rate": 1.5445604470492027e-05, + "loss": 2.5035261154174804, + "step": 84040 + }, + { + "epoch": 0.22882809309922417, + "grad_norm": 2.291642904281616, + "learning_rate": 1.5444514106583073e-05, + "loss": 2.5146581649780275, + "step": 84060 + }, + { + "epoch": 0.2288825370899687, + "grad_norm": 2.1397452354431152, + "learning_rate": 1.544342374267412e-05, + "loss": 2.5310253143310546, + "step": 84080 + }, + { + "epoch": 0.22893698108071323, + "grad_norm": 2.0742571353912354, + "learning_rate": 1.5442333378765165e-05, + "loss": 2.527352142333984, + "step": 84100 + }, + { + "epoch": 0.22899142507145773, + "grad_norm": 2.6492323875427246, + "learning_rate": 1.5441243014856207e-05, + "loss": 2.521988296508789, + "step": 84120 + }, + { + "epoch": 0.22904586906220226, + "grad_norm": 2.1093475818634033, + "learning_rate": 1.5440152650947256e-05, + "loss": 2.5455577850341795, + "step": 84140 + }, + { + "epoch": 0.22910031305294679, + "grad_norm": 2.1010494232177734, + "learning_rate": 1.54390622870383e-05, + "loss": 2.5072662353515627, + "step": 84160 + }, + { + "epoch": 0.22915475704369132, + "grad_norm": 2.1958537101745605, + "learning_rate": 1.5437971923129348e-05, + "loss": 2.528781700134277, + "step": 84180 + }, + { + "epoch": 0.22920920103443582, + "grad_norm": 2.3716464042663574, + "learning_rate": 1.543688155922039e-05, + "loss": 2.484036445617676, + "step": 84200 + }, + { + "epoch": 0.22926364502518035, + "grad_norm": 2.115520715713501, + "learning_rate": 1.5435791195311436e-05, + "loss": 2.4983346939086912, + "step": 84220 + }, + { + "epoch": 0.22931808901592488, + "grad_norm": 2.2541472911834717, + "learning_rate": 1.5434700831402482e-05, + "loss": 2.47418212890625, + "step": 84240 + }, + { + "epoch": 0.22937253300666938, + "grad_norm": 2.127256393432617, + "learning_rate": 1.5433610467493528e-05, + "loss": 2.5158626556396486, + "step": 84260 + }, + { + "epoch": 0.2294269769974139, + "grad_norm": 2.1372642517089844, + "learning_rate": 1.543252010358457e-05, + "loss": 2.4950820922851564, + "step": 84280 + }, + { + "epoch": 0.22948142098815844, + "grad_norm": 2.0955677032470703, + "learning_rate": 1.543142973967562e-05, + "loss": 2.5450868606567383, + "step": 84300 + }, + { + "epoch": 0.22953586497890296, + "grad_norm": 2.069096803665161, + "learning_rate": 1.5430339375766662e-05, + "loss": 2.510730743408203, + "step": 84320 + }, + { + "epoch": 0.22959030896964747, + "grad_norm": 2.423478126525879, + "learning_rate": 1.542924901185771e-05, + "loss": 2.5318098068237305, + "step": 84340 + }, + { + "epoch": 0.229644752960392, + "grad_norm": 2.113053560256958, + "learning_rate": 1.5428158647948754e-05, + "loss": 2.499500274658203, + "step": 84360 + }, + { + "epoch": 0.22969919695113653, + "grad_norm": 2.2245633602142334, + "learning_rate": 1.54270682840398e-05, + "loss": 2.538480758666992, + "step": 84380 + }, + { + "epoch": 0.22975364094188103, + "grad_norm": 2.064723253250122, + "learning_rate": 1.5425977920130845e-05, + "loss": 2.5447954177856444, + "step": 84400 + }, + { + "epoch": 0.22980808493262556, + "grad_norm": 2.0395350456237793, + "learning_rate": 1.542488755622189e-05, + "loss": 2.491291046142578, + "step": 84420 + }, + { + "epoch": 0.22986252892337009, + "grad_norm": 2.132218360900879, + "learning_rate": 1.5423797192312933e-05, + "loss": 2.5283090591430666, + "step": 84440 + }, + { + "epoch": 0.22991697291411461, + "grad_norm": 2.1558966636657715, + "learning_rate": 1.5422706828403983e-05, + "loss": 2.5091808319091795, + "step": 84460 + }, + { + "epoch": 0.22997141690485912, + "grad_norm": 2.062751054763794, + "learning_rate": 1.5421616464495025e-05, + "loss": 2.4974859237670897, + "step": 84480 + }, + { + "epoch": 0.23002586089560365, + "grad_norm": 2.291072130203247, + "learning_rate": 1.5420526100586074e-05, + "loss": 2.5000072479248048, + "step": 84500 + }, + { + "epoch": 0.23008030488634817, + "grad_norm": 2.198676586151123, + "learning_rate": 1.5419435736677117e-05, + "loss": 2.532333755493164, + "step": 84520 + }, + { + "epoch": 0.2301347488770927, + "grad_norm": 2.247995615005493, + "learning_rate": 1.5418345372768162e-05, + "loss": 2.50848503112793, + "step": 84540 + }, + { + "epoch": 0.2301891928678372, + "grad_norm": 2.0534451007843018, + "learning_rate": 1.5417255008859208e-05, + "loss": 2.52685661315918, + "step": 84560 + }, + { + "epoch": 0.23024363685858174, + "grad_norm": 2.073652982711792, + "learning_rate": 1.5416164644950254e-05, + "loss": 2.525966453552246, + "step": 84580 + }, + { + "epoch": 0.23029808084932626, + "grad_norm": 2.332763671875, + "learning_rate": 1.5415074281041296e-05, + "loss": 2.517487716674805, + "step": 84600 + }, + { + "epoch": 0.23035252484007077, + "grad_norm": 2.1871182918548584, + "learning_rate": 1.5413983917132346e-05, + "loss": 2.5045907974243162, + "step": 84620 + }, + { + "epoch": 0.2304069688308153, + "grad_norm": 2.110466718673706, + "learning_rate": 1.5412893553223388e-05, + "loss": 2.527999496459961, + "step": 84640 + }, + { + "epoch": 0.23046141282155982, + "grad_norm": 2.139220714569092, + "learning_rate": 1.5411803189314437e-05, + "loss": 2.524676704406738, + "step": 84660 + }, + { + "epoch": 0.23051585681230435, + "grad_norm": 2.158900737762451, + "learning_rate": 1.541071282540548e-05, + "loss": 2.5087989807128905, + "step": 84680 + }, + { + "epoch": 0.23057030080304886, + "grad_norm": 4.205978870391846, + "learning_rate": 1.5409622461496526e-05, + "loss": 2.524824333190918, + "step": 84700 + }, + { + "epoch": 0.23062474479379338, + "grad_norm": 2.1257455348968506, + "learning_rate": 1.540853209758757e-05, + "loss": 2.502097511291504, + "step": 84720 + }, + { + "epoch": 0.23067918878453791, + "grad_norm": 2.209200859069824, + "learning_rate": 1.5407441733678617e-05, + "loss": 2.531412124633789, + "step": 84740 + }, + { + "epoch": 0.23073363277528242, + "grad_norm": 2.038757562637329, + "learning_rate": 1.540635136976966e-05, + "loss": 2.4891313552856444, + "step": 84760 + }, + { + "epoch": 0.23078807676602694, + "grad_norm": 2.189107894897461, + "learning_rate": 1.540526100586071e-05, + "loss": 2.506149482727051, + "step": 84780 + }, + { + "epoch": 0.23084252075677147, + "grad_norm": 2.0809319019317627, + "learning_rate": 1.540417064195175e-05, + "loss": 2.5071186065673827, + "step": 84800 + }, + { + "epoch": 0.230896964747516, + "grad_norm": 2.2974226474761963, + "learning_rate": 1.54030802780428e-05, + "loss": 2.4898502349853517, + "step": 84820 + }, + { + "epoch": 0.2309514087382605, + "grad_norm": 2.164435386657715, + "learning_rate": 1.5401989914133843e-05, + "loss": 2.4886699676513673, + "step": 84840 + }, + { + "epoch": 0.23100585272900503, + "grad_norm": 2.165755033493042, + "learning_rate": 1.540089955022489e-05, + "loss": 2.5231142044067383, + "step": 84860 + }, + { + "epoch": 0.23106029671974956, + "grad_norm": 2.234952211380005, + "learning_rate": 1.5399809186315934e-05, + "loss": 2.505236053466797, + "step": 84880 + }, + { + "epoch": 0.23111474071049407, + "grad_norm": 2.0904595851898193, + "learning_rate": 1.539871882240698e-05, + "loss": 2.53845329284668, + "step": 84900 + }, + { + "epoch": 0.2311691847012386, + "grad_norm": 2.1597259044647217, + "learning_rate": 1.5397628458498023e-05, + "loss": 2.5346963882446287, + "step": 84920 + }, + { + "epoch": 0.23122362869198312, + "grad_norm": 2.3098745346069336, + "learning_rate": 1.5396538094589072e-05, + "loss": 2.52098388671875, + "step": 84940 + }, + { + "epoch": 0.23127807268272765, + "grad_norm": 2.0339791774749756, + "learning_rate": 1.5395447730680114e-05, + "loss": 2.488362503051758, + "step": 84960 + }, + { + "epoch": 0.23133251667347215, + "grad_norm": 2.149346113204956, + "learning_rate": 1.5394357366771163e-05, + "loss": 2.5444158554077148, + "step": 84980 + }, + { + "epoch": 0.23138696066421668, + "grad_norm": 2.1281039714813232, + "learning_rate": 1.5393267002862206e-05, + "loss": 2.514979934692383, + "step": 85000 + }, + { + "epoch": 0.2314414046549612, + "grad_norm": 2.1308839321136475, + "learning_rate": 1.539217663895325e-05, + "loss": 2.506264877319336, + "step": 85020 + }, + { + "epoch": 0.23149584864570574, + "grad_norm": 2.1020960807800293, + "learning_rate": 1.5391086275044297e-05, + "loss": 2.4894184112548827, + "step": 85040 + }, + { + "epoch": 0.23155029263645024, + "grad_norm": 2.1449363231658936, + "learning_rate": 1.5389995911135343e-05, + "loss": 2.5304664611816405, + "step": 85060 + }, + { + "epoch": 0.23160473662719477, + "grad_norm": 2.163412570953369, + "learning_rate": 1.5388905547226386e-05, + "loss": 2.532056999206543, + "step": 85080 + }, + { + "epoch": 0.2316591806179393, + "grad_norm": 2.1025009155273438, + "learning_rate": 1.5387815183317435e-05, + "loss": 2.555899238586426, + "step": 85100 + }, + { + "epoch": 0.2317136246086838, + "grad_norm": 2.376575231552124, + "learning_rate": 1.5386724819408477e-05, + "loss": 2.522031784057617, + "step": 85120 + }, + { + "epoch": 0.23176806859942833, + "grad_norm": 2.107602596282959, + "learning_rate": 1.5385634455499527e-05, + "loss": 2.5207283020019533, + "step": 85140 + }, + { + "epoch": 0.23182251259017286, + "grad_norm": 2.0595107078552246, + "learning_rate": 1.538454409159057e-05, + "loss": 2.5086151123046876, + "step": 85160 + }, + { + "epoch": 0.2318769565809174, + "grad_norm": 2.2959988117218018, + "learning_rate": 1.5383453727681615e-05, + "loss": 2.530730628967285, + "step": 85180 + }, + { + "epoch": 0.2319314005716619, + "grad_norm": 2.3920645713806152, + "learning_rate": 1.538236336377266e-05, + "loss": 2.505806732177734, + "step": 85200 + }, + { + "epoch": 0.23198584456240642, + "grad_norm": 2.190455198287964, + "learning_rate": 1.5381272999863706e-05, + "loss": 2.5067607879638674, + "step": 85220 + }, + { + "epoch": 0.23204028855315095, + "grad_norm": 2.2054872512817383, + "learning_rate": 1.538018263595475e-05, + "loss": 2.5369085311889648, + "step": 85240 + }, + { + "epoch": 0.23209473254389545, + "grad_norm": 2.0595195293426514, + "learning_rate": 1.5379092272045798e-05, + "loss": 2.5250158309936523, + "step": 85260 + }, + { + "epoch": 0.23214917653463998, + "grad_norm": 2.199991226196289, + "learning_rate": 1.537800190813684e-05, + "loss": 2.513136291503906, + "step": 85280 + }, + { + "epoch": 0.2322036205253845, + "grad_norm": 2.2159078121185303, + "learning_rate": 1.537691154422789e-05, + "loss": 2.4910375595092775, + "step": 85300 + }, + { + "epoch": 0.23225806451612904, + "grad_norm": 2.1970720291137695, + "learning_rate": 1.5375821180318932e-05, + "loss": 2.506759262084961, + "step": 85320 + }, + { + "epoch": 0.23231250850687354, + "grad_norm": 2.150485038757324, + "learning_rate": 1.5374730816409978e-05, + "loss": 2.5295791625976562, + "step": 85340 + }, + { + "epoch": 0.23236695249761807, + "grad_norm": 2.1552329063415527, + "learning_rate": 1.5373640452501024e-05, + "loss": 2.5161251068115233, + "step": 85360 + }, + { + "epoch": 0.2324213964883626, + "grad_norm": 2.2243685722351074, + "learning_rate": 1.537255008859207e-05, + "loss": 2.5112979888916014, + "step": 85380 + }, + { + "epoch": 0.23247584047910713, + "grad_norm": 2.155383348464966, + "learning_rate": 1.5371459724683112e-05, + "loss": 2.5098318099975585, + "step": 85400 + }, + { + "epoch": 0.23253028446985163, + "grad_norm": 2.3047542572021484, + "learning_rate": 1.537036936077416e-05, + "loss": 2.506123733520508, + "step": 85420 + }, + { + "epoch": 0.23258472846059616, + "grad_norm": 2.124516725540161, + "learning_rate": 1.5369278996865204e-05, + "loss": 2.552198791503906, + "step": 85440 + }, + { + "epoch": 0.2326391724513407, + "grad_norm": 2.137725591659546, + "learning_rate": 1.5368188632956253e-05, + "loss": 2.4832971572875975, + "step": 85460 + }, + { + "epoch": 0.2326936164420852, + "grad_norm": 2.1624295711517334, + "learning_rate": 1.5367098269047295e-05, + "loss": 2.547062873840332, + "step": 85480 + }, + { + "epoch": 0.23274806043282972, + "grad_norm": 2.182298421859741, + "learning_rate": 1.536600790513834e-05, + "loss": 2.48116455078125, + "step": 85500 + }, + { + "epoch": 0.23280250442357425, + "grad_norm": 2.2541775703430176, + "learning_rate": 1.5364917541229387e-05, + "loss": 2.527809715270996, + "step": 85520 + }, + { + "epoch": 0.23285694841431878, + "grad_norm": 2.0404927730560303, + "learning_rate": 1.5363827177320433e-05, + "loss": 2.490004539489746, + "step": 85540 + }, + { + "epoch": 0.23291139240506328, + "grad_norm": 2.200214385986328, + "learning_rate": 1.5362736813411475e-05, + "loss": 2.510689353942871, + "step": 85560 + }, + { + "epoch": 0.2329658363958078, + "grad_norm": 2.109954357147217, + "learning_rate": 1.5361646449502524e-05, + "loss": 2.497883415222168, + "step": 85580 + }, + { + "epoch": 0.23302028038655234, + "grad_norm": 2.0557806491851807, + "learning_rate": 1.5360556085593567e-05, + "loss": 2.528909683227539, + "step": 85600 + }, + { + "epoch": 0.23307472437729684, + "grad_norm": 2.3736817836761475, + "learning_rate": 1.5359465721684616e-05, + "loss": 2.526645469665527, + "step": 85620 + }, + { + "epoch": 0.23312916836804137, + "grad_norm": 2.130681037902832, + "learning_rate": 1.5358375357775658e-05, + "loss": 2.5365781784057617, + "step": 85640 + }, + { + "epoch": 0.2331836123587859, + "grad_norm": 2.0269031524658203, + "learning_rate": 1.5357284993866704e-05, + "loss": 2.5584362030029295, + "step": 85660 + }, + { + "epoch": 0.23323805634953043, + "grad_norm": 2.088742971420288, + "learning_rate": 1.535619462995775e-05, + "loss": 2.494722366333008, + "step": 85680 + }, + { + "epoch": 0.23329250034027493, + "grad_norm": 2.14597749710083, + "learning_rate": 1.5355104266048796e-05, + "loss": 2.5107166290283205, + "step": 85700 + }, + { + "epoch": 0.23334694433101946, + "grad_norm": 2.2887673377990723, + "learning_rate": 1.5354013902139838e-05, + "loss": 2.5232412338256838, + "step": 85720 + }, + { + "epoch": 0.233401388321764, + "grad_norm": 2.0445961952209473, + "learning_rate": 1.5352923538230887e-05, + "loss": 2.515950393676758, + "step": 85740 + }, + { + "epoch": 0.2334558323125085, + "grad_norm": 2.2146668434143066, + "learning_rate": 1.535183317432193e-05, + "loss": 2.532394790649414, + "step": 85760 + }, + { + "epoch": 0.23351027630325302, + "grad_norm": 2.1131513118743896, + "learning_rate": 1.535074281041298e-05, + "loss": 2.507149505615234, + "step": 85780 + }, + { + "epoch": 0.23356472029399755, + "grad_norm": 2.083374500274658, + "learning_rate": 1.534965244650402e-05, + "loss": 2.49414119720459, + "step": 85800 + }, + { + "epoch": 0.23361916428474208, + "grad_norm": 2.0581448078155518, + "learning_rate": 1.5348562082595067e-05, + "loss": 2.4957197189331053, + "step": 85820 + }, + { + "epoch": 0.23367360827548658, + "grad_norm": 2.2116856575012207, + "learning_rate": 1.5347471718686113e-05, + "loss": 2.5303813934326174, + "step": 85840 + }, + { + "epoch": 0.2337280522662311, + "grad_norm": 2.1349992752075195, + "learning_rate": 1.534638135477716e-05, + "loss": 2.5129074096679687, + "step": 85860 + }, + { + "epoch": 0.23378249625697564, + "grad_norm": 4.4109015464782715, + "learning_rate": 1.5345290990868205e-05, + "loss": 2.5059837341308593, + "step": 85880 + }, + { + "epoch": 0.23383694024772017, + "grad_norm": 2.6098880767822266, + "learning_rate": 1.534420062695925e-05, + "loss": 2.5056550979614256, + "step": 85900 + }, + { + "epoch": 0.23389138423846467, + "grad_norm": 2.172597646713257, + "learning_rate": 1.5343110263050293e-05, + "loss": 2.536260223388672, + "step": 85920 + }, + { + "epoch": 0.2339458282292092, + "grad_norm": 2.1593027114868164, + "learning_rate": 1.5342019899141342e-05, + "loss": 2.5139091491699217, + "step": 85940 + }, + { + "epoch": 0.23400027221995373, + "grad_norm": 2.135476589202881, + "learning_rate": 1.5340929535232384e-05, + "loss": 2.5033729553222654, + "step": 85960 + }, + { + "epoch": 0.23405471621069823, + "grad_norm": 2.134709119796753, + "learning_rate": 1.533983917132343e-05, + "loss": 2.506883239746094, + "step": 85980 + }, + { + "epoch": 0.23410916020144276, + "grad_norm": 2.00297474861145, + "learning_rate": 1.5338748807414476e-05, + "loss": 2.526509475708008, + "step": 86000 + }, + { + "epoch": 0.2341636041921873, + "grad_norm": 2.075549602508545, + "learning_rate": 1.5337658443505522e-05, + "loss": 2.5192245483398437, + "step": 86020 + }, + { + "epoch": 0.23421804818293182, + "grad_norm": 2.087507963180542, + "learning_rate": 1.5336568079596568e-05, + "loss": 2.549247169494629, + "step": 86040 + }, + { + "epoch": 0.23427249217367632, + "grad_norm": 2.0976462364196777, + "learning_rate": 1.5335477715687613e-05, + "loss": 2.537098693847656, + "step": 86060 + }, + { + "epoch": 0.23432693616442085, + "grad_norm": 2.1624810695648193, + "learning_rate": 1.5334387351778656e-05, + "loss": 2.518113708496094, + "step": 86080 + }, + { + "epoch": 0.23438138015516538, + "grad_norm": 2.0895795822143555, + "learning_rate": 1.5333296987869705e-05, + "loss": 2.489986038208008, + "step": 86100 + }, + { + "epoch": 0.23443582414590988, + "grad_norm": 2.113860607147217, + "learning_rate": 1.5332206623960747e-05, + "loss": 2.5062360763549805, + "step": 86120 + }, + { + "epoch": 0.2344902681366544, + "grad_norm": 2.0964760780334473, + "learning_rate": 1.5331116260051793e-05, + "loss": 2.512643241882324, + "step": 86140 + }, + { + "epoch": 0.23454471212739894, + "grad_norm": 2.1725544929504395, + "learning_rate": 1.533002589614284e-05, + "loss": 2.5331165313720705, + "step": 86160 + }, + { + "epoch": 0.23459915611814347, + "grad_norm": 2.285897970199585, + "learning_rate": 1.5328935532233885e-05, + "loss": 2.506003189086914, + "step": 86180 + }, + { + "epoch": 0.23465360010888797, + "grad_norm": 2.209895133972168, + "learning_rate": 1.532784516832493e-05, + "loss": 2.535617637634277, + "step": 86200 + }, + { + "epoch": 0.2347080440996325, + "grad_norm": 2.1064960956573486, + "learning_rate": 1.5326754804415977e-05, + "loss": 2.4972368240356446, + "step": 86220 + }, + { + "epoch": 0.23476248809037703, + "grad_norm": 2.1372568607330322, + "learning_rate": 1.532566444050702e-05, + "loss": 2.504205894470215, + "step": 86240 + }, + { + "epoch": 0.23481693208112156, + "grad_norm": 2.176480293273926, + "learning_rate": 1.5324574076598068e-05, + "loss": 2.5057937622070314, + "step": 86260 + }, + { + "epoch": 0.23487137607186606, + "grad_norm": 2.242323160171509, + "learning_rate": 1.532348371268911e-05, + "loss": 2.526830863952637, + "step": 86280 + }, + { + "epoch": 0.2349258200626106, + "grad_norm": 2.273097038269043, + "learning_rate": 1.5322393348780156e-05, + "loss": 2.5433841705322267, + "step": 86300 + }, + { + "epoch": 0.23498026405335512, + "grad_norm": 2.100496768951416, + "learning_rate": 1.5321302984871202e-05, + "loss": 2.5200122833251952, + "step": 86320 + }, + { + "epoch": 0.23503470804409962, + "grad_norm": 2.2939112186431885, + "learning_rate": 1.5320212620962248e-05, + "loss": 2.533681297302246, + "step": 86340 + }, + { + "epoch": 0.23508915203484415, + "grad_norm": 2.1503658294677734, + "learning_rate": 1.5319122257053294e-05, + "loss": 2.5066650390625, + "step": 86360 + }, + { + "epoch": 0.23514359602558868, + "grad_norm": 3.7909364700317383, + "learning_rate": 1.531803189314434e-05, + "loss": 2.511412811279297, + "step": 86380 + }, + { + "epoch": 0.2351980400163332, + "grad_norm": 2.1101129055023193, + "learning_rate": 1.5316941529235382e-05, + "loss": 2.523236846923828, + "step": 86400 + }, + { + "epoch": 0.2352524840070777, + "grad_norm": 2.132568359375, + "learning_rate": 1.531585116532643e-05, + "loss": 2.4973886489868162, + "step": 86420 + }, + { + "epoch": 0.23530692799782224, + "grad_norm": 3.1303818225860596, + "learning_rate": 1.5314760801417474e-05, + "loss": 2.5173791885375976, + "step": 86440 + }, + { + "epoch": 0.23536137198856677, + "grad_norm": 4.334115505218506, + "learning_rate": 1.531367043750852e-05, + "loss": 2.5066600799560548, + "step": 86460 + }, + { + "epoch": 0.23541581597931127, + "grad_norm": 2.1334738731384277, + "learning_rate": 1.5312580073599565e-05, + "loss": 2.5236879348754884, + "step": 86480 + }, + { + "epoch": 0.2354702599700558, + "grad_norm": 2.148465394973755, + "learning_rate": 1.531148970969061e-05, + "loss": 2.536732864379883, + "step": 86500 + }, + { + "epoch": 0.23552470396080033, + "grad_norm": 2.1843481063842773, + "learning_rate": 1.5310399345781657e-05, + "loss": 2.538481903076172, + "step": 86520 + }, + { + "epoch": 0.23557914795154486, + "grad_norm": 2.063269853591919, + "learning_rate": 1.5309308981872703e-05, + "loss": 2.5375720977783205, + "step": 86540 + }, + { + "epoch": 0.23563359194228936, + "grad_norm": 2.1877827644348145, + "learning_rate": 1.5308218617963745e-05, + "loss": 2.52911434173584, + "step": 86560 + }, + { + "epoch": 0.2356880359330339, + "grad_norm": 2.159990072250366, + "learning_rate": 1.5307128254054794e-05, + "loss": 2.519100570678711, + "step": 86580 + }, + { + "epoch": 0.23574247992377842, + "grad_norm": 2.14581561088562, + "learning_rate": 1.5306037890145837e-05, + "loss": 2.5194465637207033, + "step": 86600 + }, + { + "epoch": 0.23579692391452292, + "grad_norm": 2.1191160678863525, + "learning_rate": 1.5304947526236883e-05, + "loss": 2.509733200073242, + "step": 86620 + }, + { + "epoch": 0.23585136790526745, + "grad_norm": 2.143019199371338, + "learning_rate": 1.530385716232793e-05, + "loss": 2.5206674575805663, + "step": 86640 + }, + { + "epoch": 0.23590581189601198, + "grad_norm": 2.153050184249878, + "learning_rate": 1.5302766798418974e-05, + "loss": 2.510984992980957, + "step": 86660 + }, + { + "epoch": 0.2359602558867565, + "grad_norm": 2.1016900539398193, + "learning_rate": 1.530167643451002e-05, + "loss": 2.5300777435302733, + "step": 86680 + }, + { + "epoch": 0.236014699877501, + "grad_norm": 2.2362375259399414, + "learning_rate": 1.5300586070601066e-05, + "loss": 2.5069242477416993, + "step": 86700 + }, + { + "epoch": 0.23606914386824554, + "grad_norm": 2.1827447414398193, + "learning_rate": 1.5299495706692108e-05, + "loss": 2.5300668716430663, + "step": 86720 + }, + { + "epoch": 0.23612358785899007, + "grad_norm": 2.1719651222229004, + "learning_rate": 1.5298405342783157e-05, + "loss": 2.5174421310424804, + "step": 86740 + }, + { + "epoch": 0.2361780318497346, + "grad_norm": 2.0790765285491943, + "learning_rate": 1.52973149788742e-05, + "loss": 2.498744773864746, + "step": 86760 + }, + { + "epoch": 0.2362324758404791, + "grad_norm": 2.1251814365386963, + "learning_rate": 1.5296224614965246e-05, + "loss": 2.5352294921875, + "step": 86780 + }, + { + "epoch": 0.23628691983122363, + "grad_norm": 2.0856995582580566, + "learning_rate": 1.529513425105629e-05, + "loss": 2.505508232116699, + "step": 86800 + }, + { + "epoch": 0.23634136382196816, + "grad_norm": 2.072580337524414, + "learning_rate": 1.5294043887147337e-05, + "loss": 2.5292804718017576, + "step": 86820 + }, + { + "epoch": 0.23639580781271266, + "grad_norm": 2.068422555923462, + "learning_rate": 1.5292953523238383e-05, + "loss": 2.510381507873535, + "step": 86840 + }, + { + "epoch": 0.2364502518034572, + "grad_norm": 2.156902313232422, + "learning_rate": 1.529186315932943e-05, + "loss": 2.5310131072998048, + "step": 86860 + }, + { + "epoch": 0.23650469579420172, + "grad_norm": 2.040278911590576, + "learning_rate": 1.529077279542047e-05, + "loss": 2.509326934814453, + "step": 86880 + }, + { + "epoch": 0.23655913978494625, + "grad_norm": 2.077207565307617, + "learning_rate": 1.528968243151152e-05, + "loss": 2.5045425415039064, + "step": 86900 + }, + { + "epoch": 0.23661358377569075, + "grad_norm": 2.343207597732544, + "learning_rate": 1.5288592067602563e-05, + "loss": 2.493782615661621, + "step": 86920 + }, + { + "epoch": 0.23666802776643528, + "grad_norm": 2.1724448204040527, + "learning_rate": 1.528750170369361e-05, + "loss": 2.4786312103271486, + "step": 86940 + }, + { + "epoch": 0.2367224717571798, + "grad_norm": 2.329277515411377, + "learning_rate": 1.5286411339784655e-05, + "loss": 2.544844055175781, + "step": 86960 + }, + { + "epoch": 0.2367769157479243, + "grad_norm": 2.1053760051727295, + "learning_rate": 1.52853209758757e-05, + "loss": 2.522441101074219, + "step": 86980 + }, + { + "epoch": 0.23683135973866884, + "grad_norm": 2.2187299728393555, + "learning_rate": 1.5284230611966746e-05, + "loss": 2.5243066787719726, + "step": 87000 + }, + { + "epoch": 0.23688580372941337, + "grad_norm": 2.189626455307007, + "learning_rate": 1.5283140248057792e-05, + "loss": 2.50181884765625, + "step": 87020 + }, + { + "epoch": 0.2369402477201579, + "grad_norm": 2.3292911052703857, + "learning_rate": 1.5282049884148834e-05, + "loss": 2.52230281829834, + "step": 87040 + }, + { + "epoch": 0.2369946917109024, + "grad_norm": 2.193679094314575, + "learning_rate": 1.5280959520239884e-05, + "loss": 2.5512733459472656, + "step": 87060 + }, + { + "epoch": 0.23704913570164693, + "grad_norm": 2.7006125450134277, + "learning_rate": 1.5279869156330926e-05, + "loss": 2.5191497802734375, + "step": 87080 + }, + { + "epoch": 0.23710357969239146, + "grad_norm": 2.255837917327881, + "learning_rate": 1.5278778792421972e-05, + "loss": 2.528004455566406, + "step": 87100 + }, + { + "epoch": 0.23715802368313596, + "grad_norm": 2.0867602825164795, + "learning_rate": 1.5277688428513018e-05, + "loss": 2.503555679321289, + "step": 87120 + }, + { + "epoch": 0.2372124676738805, + "grad_norm": 2.5970287322998047, + "learning_rate": 1.5276598064604063e-05, + "loss": 2.5188068389892577, + "step": 87140 + }, + { + "epoch": 0.23726691166462502, + "grad_norm": 2.4700992107391357, + "learning_rate": 1.527550770069511e-05, + "loss": 2.4920591354370116, + "step": 87160 + }, + { + "epoch": 0.23732135565536955, + "grad_norm": 2.1485326290130615, + "learning_rate": 1.5274417336786155e-05, + "loss": 2.483774185180664, + "step": 87180 + }, + { + "epoch": 0.23737579964611405, + "grad_norm": 2.240893840789795, + "learning_rate": 1.5273326972877197e-05, + "loss": 2.5102148056030273, + "step": 87200 + }, + { + "epoch": 0.23743024363685858, + "grad_norm": 2.080296516418457, + "learning_rate": 1.5272236608968247e-05, + "loss": 2.5191581726074217, + "step": 87220 + }, + { + "epoch": 0.2374846876276031, + "grad_norm": 2.0750732421875, + "learning_rate": 1.527114624505929e-05, + "loss": 2.5401823043823244, + "step": 87240 + }, + { + "epoch": 0.23753913161834764, + "grad_norm": 2.0846312046051025, + "learning_rate": 1.5270055881150335e-05, + "loss": 2.507620620727539, + "step": 87260 + }, + { + "epoch": 0.23759357560909214, + "grad_norm": 2.1326797008514404, + "learning_rate": 1.526896551724138e-05, + "loss": 2.5410602569580076, + "step": 87280 + }, + { + "epoch": 0.23764801959983667, + "grad_norm": 2.0681238174438477, + "learning_rate": 1.5267875153332426e-05, + "loss": 2.5275833129882814, + "step": 87300 + }, + { + "epoch": 0.2377024635905812, + "grad_norm": 2.1998953819274902, + "learning_rate": 1.5266784789423472e-05, + "loss": 2.5141305923461914, + "step": 87320 + }, + { + "epoch": 0.2377569075813257, + "grad_norm": 2.097980499267578, + "learning_rate": 1.5265694425514518e-05, + "loss": 2.5004093170166017, + "step": 87340 + }, + { + "epoch": 0.23781135157207023, + "grad_norm": 2.1675047874450684, + "learning_rate": 1.526460406160556e-05, + "loss": 2.493112564086914, + "step": 87360 + }, + { + "epoch": 0.23786579556281476, + "grad_norm": 2.176769495010376, + "learning_rate": 1.526351369769661e-05, + "loss": 2.5059524536132813, + "step": 87380 + }, + { + "epoch": 0.2379202395535593, + "grad_norm": 2.151014804840088, + "learning_rate": 1.5262423333787652e-05, + "loss": 2.5363784790039063, + "step": 87400 + }, + { + "epoch": 0.2379746835443038, + "grad_norm": 2.171464681625366, + "learning_rate": 1.5261332969878698e-05, + "loss": 2.5312469482421873, + "step": 87420 + }, + { + "epoch": 0.23802912753504832, + "grad_norm": 2.0844860076904297, + "learning_rate": 1.5260242605969744e-05, + "loss": 2.5326511383056642, + "step": 87440 + }, + { + "epoch": 0.23808357152579285, + "grad_norm": 2.0792109966278076, + "learning_rate": 1.525915224206079e-05, + "loss": 2.5328224182128904, + "step": 87460 + }, + { + "epoch": 0.23813801551653735, + "grad_norm": 2.234074831008911, + "learning_rate": 1.5258061878151835e-05, + "loss": 2.52346248626709, + "step": 87480 + }, + { + "epoch": 0.23819245950728188, + "grad_norm": 2.207416534423828, + "learning_rate": 1.525697151424288e-05, + "loss": 2.5251432418823243, + "step": 87500 + }, + { + "epoch": 0.2382469034980264, + "grad_norm": 8.449379920959473, + "learning_rate": 1.5255881150333924e-05, + "loss": 2.5237472534179686, + "step": 87520 + }, + { + "epoch": 0.23830134748877094, + "grad_norm": 2.1142449378967285, + "learning_rate": 1.5254790786424971e-05, + "loss": 2.453261375427246, + "step": 87540 + }, + { + "epoch": 0.23835579147951544, + "grad_norm": 2.121983051300049, + "learning_rate": 1.5253700422516015e-05, + "loss": 2.5258928298950196, + "step": 87560 + }, + { + "epoch": 0.23841023547025997, + "grad_norm": 2.045785903930664, + "learning_rate": 1.5252610058607063e-05, + "loss": 2.518870735168457, + "step": 87580 + }, + { + "epoch": 0.2384646794610045, + "grad_norm": 2.3930928707122803, + "learning_rate": 1.5251519694698107e-05, + "loss": 2.507010650634766, + "step": 87600 + }, + { + "epoch": 0.23851912345174903, + "grad_norm": 2.282998561859131, + "learning_rate": 1.5250429330789151e-05, + "loss": 2.5166980743408205, + "step": 87620 + }, + { + "epoch": 0.23857356744249353, + "grad_norm": 2.0430755615234375, + "learning_rate": 1.5249338966880198e-05, + "loss": 2.501530647277832, + "step": 87640 + }, + { + "epoch": 0.23862801143323806, + "grad_norm": 2.230180263519287, + "learning_rate": 1.5248248602971243e-05, + "loss": 2.534821701049805, + "step": 87660 + }, + { + "epoch": 0.2386824554239826, + "grad_norm": 2.1149203777313232, + "learning_rate": 1.5247158239062287e-05, + "loss": 2.521976852416992, + "step": 87680 + }, + { + "epoch": 0.2387368994147271, + "grad_norm": 2.083728551864624, + "learning_rate": 1.5246067875153334e-05, + "loss": 2.5262166976928713, + "step": 87700 + }, + { + "epoch": 0.23879134340547162, + "grad_norm": 2.2056407928466797, + "learning_rate": 1.5244977511244378e-05, + "loss": 2.514803886413574, + "step": 87720 + }, + { + "epoch": 0.23884578739621615, + "grad_norm": 2.0985822677612305, + "learning_rate": 1.5243887147335426e-05, + "loss": 2.5420846939086914, + "step": 87740 + }, + { + "epoch": 0.23890023138696068, + "grad_norm": 2.115307092666626, + "learning_rate": 1.524279678342647e-05, + "loss": 2.501500129699707, + "step": 87760 + }, + { + "epoch": 0.23895467537770518, + "grad_norm": 2.144263744354248, + "learning_rate": 1.5241706419517514e-05, + "loss": 2.507814407348633, + "step": 87780 + }, + { + "epoch": 0.2390091193684497, + "grad_norm": 2.180598735809326, + "learning_rate": 1.5240616055608562e-05, + "loss": 2.4973546981811525, + "step": 87800 + }, + { + "epoch": 0.23906356335919424, + "grad_norm": 2.1070306301116943, + "learning_rate": 1.5239525691699606e-05, + "loss": 2.501473045349121, + "step": 87820 + }, + { + "epoch": 0.23911800734993874, + "grad_norm": 2.0906858444213867, + "learning_rate": 1.523843532779065e-05, + "loss": 2.5634199142456056, + "step": 87840 + }, + { + "epoch": 0.23917245134068327, + "grad_norm": 2.094641923904419, + "learning_rate": 1.5237344963881697e-05, + "loss": 2.5187870025634767, + "step": 87860 + }, + { + "epoch": 0.2392268953314278, + "grad_norm": 2.1339962482452393, + "learning_rate": 1.5236254599972741e-05, + "loss": 2.5238256454467773, + "step": 87880 + }, + { + "epoch": 0.23928133932217233, + "grad_norm": 2.187298059463501, + "learning_rate": 1.5235164236063789e-05, + "loss": 2.4773563385009765, + "step": 87900 + }, + { + "epoch": 0.23933578331291683, + "grad_norm": 2.090211868286133, + "learning_rate": 1.5234073872154833e-05, + "loss": 2.534241485595703, + "step": 87920 + }, + { + "epoch": 0.23939022730366136, + "grad_norm": 2.0751566886901855, + "learning_rate": 1.5232983508245877e-05, + "loss": 2.550953674316406, + "step": 87940 + }, + { + "epoch": 0.2394446712944059, + "grad_norm": 2.167123317718506, + "learning_rate": 1.5231893144336925e-05, + "loss": 2.516917419433594, + "step": 87960 + }, + { + "epoch": 0.2394991152851504, + "grad_norm": 2.1490094661712646, + "learning_rate": 1.5230802780427969e-05, + "loss": 2.5079240798950195, + "step": 87980 + }, + { + "epoch": 0.23955355927589492, + "grad_norm": 2.096280813217163, + "learning_rate": 1.5229712416519013e-05, + "loss": 2.5456005096435548, + "step": 88000 + }, + { + "epoch": 0.23960800326663945, + "grad_norm": 2.122464179992676, + "learning_rate": 1.522862205261006e-05, + "loss": 2.495442771911621, + "step": 88020 + }, + { + "epoch": 0.23966244725738398, + "grad_norm": 2.1411988735198975, + "learning_rate": 1.5227531688701104e-05, + "loss": 2.538204574584961, + "step": 88040 + }, + { + "epoch": 0.23971689124812848, + "grad_norm": 2.1028778553009033, + "learning_rate": 1.5226441324792152e-05, + "loss": 2.5294321060180662, + "step": 88060 + }, + { + "epoch": 0.239771335238873, + "grad_norm": 2.0464251041412354, + "learning_rate": 1.5225350960883196e-05, + "loss": 2.5345794677734377, + "step": 88080 + }, + { + "epoch": 0.23982577922961754, + "grad_norm": 2.2182939052581787, + "learning_rate": 1.522426059697424e-05, + "loss": 2.5048603057861327, + "step": 88100 + }, + { + "epoch": 0.23988022322036207, + "grad_norm": 2.178447961807251, + "learning_rate": 1.5223170233065288e-05, + "loss": 2.510395812988281, + "step": 88120 + }, + { + "epoch": 0.23993466721110657, + "grad_norm": 2.2710793018341064, + "learning_rate": 1.5222079869156332e-05, + "loss": 2.4943761825561523, + "step": 88140 + }, + { + "epoch": 0.2399891112018511, + "grad_norm": 2.1355228424072266, + "learning_rate": 1.5220989505247376e-05, + "loss": 2.5182701110839845, + "step": 88160 + }, + { + "epoch": 0.24004355519259563, + "grad_norm": 2.110018730163574, + "learning_rate": 1.5219899141338423e-05, + "loss": 2.5249759674072267, + "step": 88180 + }, + { + "epoch": 0.24009799918334013, + "grad_norm": 2.086801767349243, + "learning_rate": 1.5218808777429468e-05, + "loss": 2.486697769165039, + "step": 88200 + }, + { + "epoch": 0.24015244317408466, + "grad_norm": 2.1264705657958984, + "learning_rate": 1.5217718413520515e-05, + "loss": 2.4980663299560546, + "step": 88220 + }, + { + "epoch": 0.24020688716482919, + "grad_norm": 2.207369565963745, + "learning_rate": 1.521662804961156e-05, + "loss": 2.5438663482666017, + "step": 88240 + }, + { + "epoch": 0.24026133115557372, + "grad_norm": 2.175204277038574, + "learning_rate": 1.5215537685702603e-05, + "loss": 2.501069450378418, + "step": 88260 + }, + { + "epoch": 0.24031577514631822, + "grad_norm": 2.1685640811920166, + "learning_rate": 1.521444732179365e-05, + "loss": 2.514173698425293, + "step": 88280 + }, + { + "epoch": 0.24037021913706275, + "grad_norm": 2.091444492340088, + "learning_rate": 1.5213356957884695e-05, + "loss": 2.5080204010009766, + "step": 88300 + }, + { + "epoch": 0.24042466312780728, + "grad_norm": 2.2717151641845703, + "learning_rate": 1.5212266593975739e-05, + "loss": 2.5503870010375977, + "step": 88320 + }, + { + "epoch": 0.24047910711855178, + "grad_norm": 2.0836334228515625, + "learning_rate": 1.5211176230066787e-05, + "loss": 2.5288705825805664, + "step": 88340 + }, + { + "epoch": 0.2405335511092963, + "grad_norm": 2.1419270038604736, + "learning_rate": 1.521008586615783e-05, + "loss": 2.540593719482422, + "step": 88360 + }, + { + "epoch": 0.24058799510004084, + "grad_norm": 2.096191167831421, + "learning_rate": 1.5208995502248878e-05, + "loss": 2.5110803604125977, + "step": 88380 + }, + { + "epoch": 0.24064243909078536, + "grad_norm": 2.0474257469177246, + "learning_rate": 1.5207905138339922e-05, + "loss": 2.483652687072754, + "step": 88400 + }, + { + "epoch": 0.24069688308152987, + "grad_norm": 2.042910575866699, + "learning_rate": 1.5206814774430966e-05, + "loss": 2.5354284286499023, + "step": 88420 + }, + { + "epoch": 0.2407513270722744, + "grad_norm": 2.1612632274627686, + "learning_rate": 1.5205724410522014e-05, + "loss": 2.501001739501953, + "step": 88440 + }, + { + "epoch": 0.24080577106301893, + "grad_norm": 1.9916714429855347, + "learning_rate": 1.5204634046613058e-05, + "loss": 2.490492248535156, + "step": 88460 + }, + { + "epoch": 0.24086021505376345, + "grad_norm": 2.0634264945983887, + "learning_rate": 1.5203543682704102e-05, + "loss": 2.5136177062988283, + "step": 88480 + }, + { + "epoch": 0.24091465904450796, + "grad_norm": 2.2634711265563965, + "learning_rate": 1.520245331879515e-05, + "loss": 2.5067787170410156, + "step": 88500 + }, + { + "epoch": 0.24096910303525249, + "grad_norm": 2.1216378211975098, + "learning_rate": 1.5201362954886194e-05, + "loss": 2.513795852661133, + "step": 88520 + }, + { + "epoch": 0.24102354702599701, + "grad_norm": 2.0334672927856445, + "learning_rate": 1.5200272590977241e-05, + "loss": 2.5000810623168945, + "step": 88540 + }, + { + "epoch": 0.24107799101674152, + "grad_norm": 2.1263647079467773, + "learning_rate": 1.5199182227068285e-05, + "loss": 2.539175605773926, + "step": 88560 + }, + { + "epoch": 0.24113243500748605, + "grad_norm": 2.1341590881347656, + "learning_rate": 1.519809186315933e-05, + "loss": 2.5247919082641603, + "step": 88580 + }, + { + "epoch": 0.24118687899823057, + "grad_norm": 2.3715999126434326, + "learning_rate": 1.5197001499250377e-05, + "loss": 2.5548139572143556, + "step": 88600 + }, + { + "epoch": 0.2412413229889751, + "grad_norm": 2.143782377243042, + "learning_rate": 1.5195911135341421e-05, + "loss": 2.543574905395508, + "step": 88620 + }, + { + "epoch": 0.2412957669797196, + "grad_norm": 2.2064626216888428, + "learning_rate": 1.5194820771432465e-05, + "loss": 2.4917633056640627, + "step": 88640 + }, + { + "epoch": 0.24135021097046414, + "grad_norm": 2.0811541080474854, + "learning_rate": 1.5193730407523513e-05, + "loss": 2.536115837097168, + "step": 88660 + }, + { + "epoch": 0.24140465496120866, + "grad_norm": 2.1025612354278564, + "learning_rate": 1.5192640043614557e-05, + "loss": 2.526954460144043, + "step": 88680 + }, + { + "epoch": 0.24145909895195317, + "grad_norm": 2.042559862136841, + "learning_rate": 1.5191549679705604e-05, + "loss": 2.506852722167969, + "step": 88700 + }, + { + "epoch": 0.2415135429426977, + "grad_norm": 2.750521659851074, + "learning_rate": 1.5190459315796648e-05, + "loss": 2.5383094787597655, + "step": 88720 + }, + { + "epoch": 0.24156798693344222, + "grad_norm": 2.0491690635681152, + "learning_rate": 1.5189368951887693e-05, + "loss": 2.482341003417969, + "step": 88740 + }, + { + "epoch": 0.24162243092418675, + "grad_norm": 2.1170654296875, + "learning_rate": 1.518827858797874e-05, + "loss": 2.50430908203125, + "step": 88760 + }, + { + "epoch": 0.24167687491493126, + "grad_norm": 2.105512857437134, + "learning_rate": 1.5187188224069784e-05, + "loss": 2.555769157409668, + "step": 88780 + }, + { + "epoch": 0.24173131890567578, + "grad_norm": 2.1171042919158936, + "learning_rate": 1.5186097860160828e-05, + "loss": 2.5266036987304688, + "step": 88800 + }, + { + "epoch": 0.24178576289642031, + "grad_norm": 2.043659210205078, + "learning_rate": 1.5185007496251876e-05, + "loss": 2.5425418853759765, + "step": 88820 + }, + { + "epoch": 0.24184020688716482, + "grad_norm": 2.2198972702026367, + "learning_rate": 1.518391713234292e-05, + "loss": 2.5064659118652344, + "step": 88840 + }, + { + "epoch": 0.24189465087790935, + "grad_norm": 2.2151753902435303, + "learning_rate": 1.5182826768433967e-05, + "loss": 2.519717216491699, + "step": 88860 + }, + { + "epoch": 0.24194909486865387, + "grad_norm": 2.206430435180664, + "learning_rate": 1.5181736404525012e-05, + "loss": 2.507451820373535, + "step": 88880 + }, + { + "epoch": 0.2420035388593984, + "grad_norm": 2.229562520980835, + "learning_rate": 1.5180646040616056e-05, + "loss": 2.5377729415893553, + "step": 88900 + }, + { + "epoch": 0.2420579828501429, + "grad_norm": 2.1469178199768066, + "learning_rate": 1.5179555676707103e-05, + "loss": 2.540469169616699, + "step": 88920 + }, + { + "epoch": 0.24211242684088743, + "grad_norm": 2.2042925357818604, + "learning_rate": 1.5178465312798147e-05, + "loss": 2.533842086791992, + "step": 88940 + }, + { + "epoch": 0.24216687083163196, + "grad_norm": 2.18560528755188, + "learning_rate": 1.5177374948889191e-05, + "loss": 2.507604789733887, + "step": 88960 + }, + { + "epoch": 0.2422213148223765, + "grad_norm": 2.3544530868530273, + "learning_rate": 1.5176284584980239e-05, + "loss": 2.5284755706787108, + "step": 88980 + }, + { + "epoch": 0.242275758813121, + "grad_norm": 2.2975666522979736, + "learning_rate": 1.5175194221071283e-05, + "loss": 2.5122751235961913, + "step": 89000 + }, + { + "epoch": 0.24233020280386552, + "grad_norm": 2.1113760471343994, + "learning_rate": 1.517410385716233e-05, + "loss": 2.505504608154297, + "step": 89020 + }, + { + "epoch": 0.24238464679461005, + "grad_norm": 2.140944719314575, + "learning_rate": 1.5173013493253375e-05, + "loss": 2.5206079483032227, + "step": 89040 + }, + { + "epoch": 0.24243909078535456, + "grad_norm": 2.1664347648620605, + "learning_rate": 1.5171923129344419e-05, + "loss": 2.488739585876465, + "step": 89060 + }, + { + "epoch": 0.24249353477609908, + "grad_norm": 2.1089487075805664, + "learning_rate": 1.5170832765435466e-05, + "loss": 2.497848701477051, + "step": 89080 + }, + { + "epoch": 0.2425479787668436, + "grad_norm": 2.3352408409118652, + "learning_rate": 1.516974240152651e-05, + "loss": 2.494740104675293, + "step": 89100 + }, + { + "epoch": 0.24260242275758814, + "grad_norm": 2.1696858406066895, + "learning_rate": 1.5168652037617554e-05, + "loss": 2.5162485122680662, + "step": 89120 + }, + { + "epoch": 0.24265686674833264, + "grad_norm": 2.179377317428589, + "learning_rate": 1.5167561673708602e-05, + "loss": 2.4992416381835936, + "step": 89140 + }, + { + "epoch": 0.24271131073907717, + "grad_norm": 2.0856847763061523, + "learning_rate": 1.5166471309799646e-05, + "loss": 2.4925819396972657, + "step": 89160 + }, + { + "epoch": 0.2427657547298217, + "grad_norm": 2.2090654373168945, + "learning_rate": 1.5165380945890694e-05, + "loss": 2.534061813354492, + "step": 89180 + }, + { + "epoch": 0.2428201987205662, + "grad_norm": 2.1083507537841797, + "learning_rate": 1.5164290581981738e-05, + "loss": 2.5323625564575196, + "step": 89200 + }, + { + "epoch": 0.24287464271131073, + "grad_norm": 2.175015449523926, + "learning_rate": 1.5163200218072782e-05, + "loss": 2.5237249374389648, + "step": 89220 + }, + { + "epoch": 0.24292908670205526, + "grad_norm": 2.274742841720581, + "learning_rate": 1.516210985416383e-05, + "loss": 2.520938491821289, + "step": 89240 + }, + { + "epoch": 0.2429835306927998, + "grad_norm": 2.119974374771118, + "learning_rate": 1.5161019490254873e-05, + "loss": 2.5319746017456053, + "step": 89260 + }, + { + "epoch": 0.2430379746835443, + "grad_norm": 2.0623464584350586, + "learning_rate": 1.5159929126345921e-05, + "loss": 2.5426753997802733, + "step": 89280 + }, + { + "epoch": 0.24309241867428882, + "grad_norm": 2.1047511100769043, + "learning_rate": 1.5158838762436965e-05, + "loss": 2.4779497146606446, + "step": 89300 + }, + { + "epoch": 0.24314686266503335, + "grad_norm": 2.090703010559082, + "learning_rate": 1.515774839852801e-05, + "loss": 2.5145225524902344, + "step": 89320 + }, + { + "epoch": 0.24320130665577785, + "grad_norm": 2.350729465484619, + "learning_rate": 1.5156658034619057e-05, + "loss": 2.486292266845703, + "step": 89340 + }, + { + "epoch": 0.24325575064652238, + "grad_norm": 2.2967543601989746, + "learning_rate": 1.51555676707101e-05, + "loss": 2.5503057479858398, + "step": 89360 + }, + { + "epoch": 0.2433101946372669, + "grad_norm": 2.3368475437164307, + "learning_rate": 1.5154477306801145e-05, + "loss": 2.5113452911376952, + "step": 89380 + }, + { + "epoch": 0.24336463862801144, + "grad_norm": 2.652566909790039, + "learning_rate": 1.5153386942892192e-05, + "loss": 2.5001176834106444, + "step": 89400 + }, + { + "epoch": 0.24341908261875594, + "grad_norm": 2.133009195327759, + "learning_rate": 1.5152296578983237e-05, + "loss": 2.5372993469238283, + "step": 89420 + }, + { + "epoch": 0.24347352660950047, + "grad_norm": 2.8708603382110596, + "learning_rate": 1.5151206215074284e-05, + "loss": 2.52191219329834, + "step": 89440 + }, + { + "epoch": 0.243527970600245, + "grad_norm": 2.116776466369629, + "learning_rate": 1.5150115851165328e-05, + "loss": 2.512690544128418, + "step": 89460 + }, + { + "epoch": 0.24358241459098953, + "grad_norm": 2.123013973236084, + "learning_rate": 1.5149025487256372e-05, + "loss": 2.5144678115844727, + "step": 89480 + }, + { + "epoch": 0.24363685858173403, + "grad_norm": 2.1650636196136475, + "learning_rate": 1.514793512334742e-05, + "loss": 2.5257949829101562, + "step": 89500 + }, + { + "epoch": 0.24369130257247856, + "grad_norm": 2.2226128578186035, + "learning_rate": 1.5146844759438464e-05, + "loss": 2.5188997268676756, + "step": 89520 + }, + { + "epoch": 0.2437457465632231, + "grad_norm": 2.206953525543213, + "learning_rate": 1.5145754395529508e-05, + "loss": 2.5370059967041017, + "step": 89540 + }, + { + "epoch": 0.2438001905539676, + "grad_norm": 2.0524039268493652, + "learning_rate": 1.5144664031620555e-05, + "loss": 2.517597198486328, + "step": 89560 + }, + { + "epoch": 0.24385463454471212, + "grad_norm": 2.0692837238311768, + "learning_rate": 1.51435736677116e-05, + "loss": 2.492801475524902, + "step": 89580 + }, + { + "epoch": 0.24390907853545665, + "grad_norm": 2.0948381423950195, + "learning_rate": 1.5142483303802647e-05, + "loss": 2.541280174255371, + "step": 89600 + }, + { + "epoch": 0.24396352252620118, + "grad_norm": 2.1554765701293945, + "learning_rate": 1.5141392939893691e-05, + "loss": 2.5297407150268554, + "step": 89620 + }, + { + "epoch": 0.24401796651694568, + "grad_norm": 2.136643648147583, + "learning_rate": 1.5140302575984735e-05, + "loss": 2.5504339218139647, + "step": 89640 + }, + { + "epoch": 0.2440724105076902, + "grad_norm": 2.1368441581726074, + "learning_rate": 1.5139212212075783e-05, + "loss": 2.492941665649414, + "step": 89660 + }, + { + "epoch": 0.24412685449843474, + "grad_norm": 2.142453670501709, + "learning_rate": 1.5138121848166827e-05, + "loss": 2.5198028564453123, + "step": 89680 + }, + { + "epoch": 0.24418129848917924, + "grad_norm": 2.164273738861084, + "learning_rate": 1.5137031484257871e-05, + "loss": 2.4955562591552733, + "step": 89700 + }, + { + "epoch": 0.24423574247992377, + "grad_norm": 2.067206382751465, + "learning_rate": 1.5135941120348919e-05, + "loss": 2.501130294799805, + "step": 89720 + }, + { + "epoch": 0.2442901864706683, + "grad_norm": 2.1148533821105957, + "learning_rate": 1.5134850756439963e-05, + "loss": 2.5327289581298826, + "step": 89740 + }, + { + "epoch": 0.24434463046141283, + "grad_norm": 2.1034295558929443, + "learning_rate": 1.513376039253101e-05, + "loss": 2.516587829589844, + "step": 89760 + }, + { + "epoch": 0.24439907445215733, + "grad_norm": 2.226435661315918, + "learning_rate": 1.5132670028622054e-05, + "loss": 2.5316793441772463, + "step": 89780 + }, + { + "epoch": 0.24445351844290186, + "grad_norm": 2.0963313579559326, + "learning_rate": 1.5131579664713098e-05, + "loss": 2.5217784881591796, + "step": 89800 + }, + { + "epoch": 0.2445079624336464, + "grad_norm": 2.103234052658081, + "learning_rate": 1.5130489300804146e-05, + "loss": 2.5088714599609374, + "step": 89820 + }, + { + "epoch": 0.24456240642439092, + "grad_norm": 2.111478805541992, + "learning_rate": 1.512939893689519e-05, + "loss": 2.5640377044677733, + "step": 89840 + }, + { + "epoch": 0.24461685041513542, + "grad_norm": 2.206960439682007, + "learning_rate": 1.5128308572986234e-05, + "loss": 2.5318429946899412, + "step": 89860 + }, + { + "epoch": 0.24467129440587995, + "grad_norm": 2.0760245323181152, + "learning_rate": 1.5127218209077282e-05, + "loss": 2.518377494812012, + "step": 89880 + }, + { + "epoch": 0.24472573839662448, + "grad_norm": 2.1573078632354736, + "learning_rate": 1.5126127845168326e-05, + "loss": 2.550786590576172, + "step": 89900 + }, + { + "epoch": 0.24478018238736898, + "grad_norm": 2.1134026050567627, + "learning_rate": 1.5125037481259373e-05, + "loss": 2.5063350677490233, + "step": 89920 + }, + { + "epoch": 0.2448346263781135, + "grad_norm": 2.286442995071411, + "learning_rate": 1.5123947117350417e-05, + "loss": 2.4898775100708006, + "step": 89940 + }, + { + "epoch": 0.24488907036885804, + "grad_norm": 2.109009027481079, + "learning_rate": 1.5122856753441462e-05, + "loss": 2.5086391448974608, + "step": 89960 + }, + { + "epoch": 0.24494351435960257, + "grad_norm": 2.161017417907715, + "learning_rate": 1.5121766389532509e-05, + "loss": 2.5158077239990235, + "step": 89980 + }, + { + "epoch": 0.24499795835034707, + "grad_norm": 2.1611857414245605, + "learning_rate": 1.5120676025623553e-05, + "loss": 2.5304080963134767, + "step": 90000 + }, + { + "epoch": 0.2450524023410916, + "grad_norm": 2.1245810985565186, + "learning_rate": 1.5119585661714597e-05, + "loss": 2.5095426559448244, + "step": 90020 + }, + { + "epoch": 0.24510684633183613, + "grad_norm": 2.100689649581909, + "learning_rate": 1.5118495297805645e-05, + "loss": 2.514560508728027, + "step": 90040 + }, + { + "epoch": 0.24516129032258063, + "grad_norm": 2.1450607776641846, + "learning_rate": 1.5117404933896689e-05, + "loss": 2.4986047744750977, + "step": 90060 + }, + { + "epoch": 0.24521573431332516, + "grad_norm": 2.138916492462158, + "learning_rate": 1.5116314569987736e-05, + "loss": 2.5057708740234377, + "step": 90080 + }, + { + "epoch": 0.2452701783040697, + "grad_norm": 2.1963343620300293, + "learning_rate": 1.511522420607878e-05, + "loss": 2.5612466812133787, + "step": 90100 + }, + { + "epoch": 0.24532462229481422, + "grad_norm": 2.132944345474243, + "learning_rate": 1.5114133842169825e-05, + "loss": 2.516077423095703, + "step": 90120 + }, + { + "epoch": 0.24537906628555872, + "grad_norm": 2.423203945159912, + "learning_rate": 1.5113043478260872e-05, + "loss": 2.504952239990234, + "step": 90140 + }, + { + "epoch": 0.24543351027630325, + "grad_norm": 2.0985310077667236, + "learning_rate": 1.5111953114351916e-05, + "loss": 2.4961742401123046, + "step": 90160 + }, + { + "epoch": 0.24548795426704778, + "grad_norm": 2.0675439834594727, + "learning_rate": 1.511086275044296e-05, + "loss": 2.510536956787109, + "step": 90180 + }, + { + "epoch": 0.24554239825779228, + "grad_norm": 2.0676677227020264, + "learning_rate": 1.5109772386534008e-05, + "loss": 2.511058807373047, + "step": 90200 + }, + { + "epoch": 0.2455968422485368, + "grad_norm": 2.1419830322265625, + "learning_rate": 1.5108682022625052e-05, + "loss": 2.5286561965942385, + "step": 90220 + }, + { + "epoch": 0.24565128623928134, + "grad_norm": 2.1868953704833984, + "learning_rate": 1.51075916587161e-05, + "loss": 2.5076272964477537, + "step": 90240 + }, + { + "epoch": 0.24570573023002587, + "grad_norm": 2.1657228469848633, + "learning_rate": 1.5106501294807144e-05, + "loss": 2.526668167114258, + "step": 90260 + }, + { + "epoch": 0.24576017422077037, + "grad_norm": 2.193979024887085, + "learning_rate": 1.5105410930898188e-05, + "loss": 2.5032630920410157, + "step": 90280 + }, + { + "epoch": 0.2458146182115149, + "grad_norm": 2.0113794803619385, + "learning_rate": 1.5104320566989235e-05, + "loss": 2.5271560668945314, + "step": 90300 + }, + { + "epoch": 0.24586906220225943, + "grad_norm": 2.0821709632873535, + "learning_rate": 1.510323020308028e-05, + "loss": 2.517727279663086, + "step": 90320 + }, + { + "epoch": 0.24592350619300396, + "grad_norm": 2.2659125328063965, + "learning_rate": 1.5102139839171323e-05, + "loss": 2.5417518615722656, + "step": 90340 + }, + { + "epoch": 0.24597795018374846, + "grad_norm": 2.113555669784546, + "learning_rate": 1.5101049475262371e-05, + "loss": 2.514360809326172, + "step": 90360 + }, + { + "epoch": 0.246032394174493, + "grad_norm": 2.110125780105591, + "learning_rate": 1.5099959111353415e-05, + "loss": 2.529962921142578, + "step": 90380 + }, + { + "epoch": 0.24608683816523752, + "grad_norm": 2.0739753246307373, + "learning_rate": 1.5098868747444463e-05, + "loss": 2.4886919021606446, + "step": 90400 + }, + { + "epoch": 0.24614128215598202, + "grad_norm": 2.063555955886841, + "learning_rate": 1.5097778383535507e-05, + "loss": 2.493752098083496, + "step": 90420 + }, + { + "epoch": 0.24619572614672655, + "grad_norm": 2.192293405532837, + "learning_rate": 1.509668801962655e-05, + "loss": 2.5248716354370115, + "step": 90440 + }, + { + "epoch": 0.24625017013747108, + "grad_norm": 2.138265371322632, + "learning_rate": 1.5095597655717598e-05, + "loss": 2.5178808212280273, + "step": 90460 + }, + { + "epoch": 0.2463046141282156, + "grad_norm": 2.0938215255737305, + "learning_rate": 1.5094507291808642e-05, + "loss": 2.531142234802246, + "step": 90480 + }, + { + "epoch": 0.2463590581189601, + "grad_norm": 2.146296501159668, + "learning_rate": 1.5093416927899687e-05, + "loss": 2.5155624389648437, + "step": 90500 + }, + { + "epoch": 0.24641350210970464, + "grad_norm": 2.059997320175171, + "learning_rate": 1.5092326563990734e-05, + "loss": 2.5242382049560548, + "step": 90520 + }, + { + "epoch": 0.24646794610044917, + "grad_norm": 2.1442270278930664, + "learning_rate": 1.5091236200081778e-05, + "loss": 2.51070556640625, + "step": 90540 + }, + { + "epoch": 0.24652239009119367, + "grad_norm": 2.043739080429077, + "learning_rate": 1.5090145836172826e-05, + "loss": 2.489247703552246, + "step": 90560 + }, + { + "epoch": 0.2465768340819382, + "grad_norm": 2.089975595474243, + "learning_rate": 1.508905547226387e-05, + "loss": 2.513500213623047, + "step": 90580 + }, + { + "epoch": 0.24663127807268273, + "grad_norm": 2.3430981636047363, + "learning_rate": 1.5087965108354914e-05, + "loss": 2.5238208770751953, + "step": 90600 + }, + { + "epoch": 0.24668572206342726, + "grad_norm": 2.0794262886047363, + "learning_rate": 1.5086874744445961e-05, + "loss": 2.497884178161621, + "step": 90620 + }, + { + "epoch": 0.24674016605417176, + "grad_norm": 2.1537132263183594, + "learning_rate": 1.5085784380537005e-05, + "loss": 2.5024505615234376, + "step": 90640 + }, + { + "epoch": 0.2467946100449163, + "grad_norm": 2.079317808151245, + "learning_rate": 1.508469401662805e-05, + "loss": 2.546287727355957, + "step": 90660 + }, + { + "epoch": 0.24684905403566082, + "grad_norm": 2.379352331161499, + "learning_rate": 1.5083603652719097e-05, + "loss": 2.487615203857422, + "step": 90680 + }, + { + "epoch": 0.24690349802640535, + "grad_norm": 2.141120672225952, + "learning_rate": 1.5082513288810141e-05, + "loss": 2.526178741455078, + "step": 90700 + }, + { + "epoch": 0.24695794201714985, + "grad_norm": 2.135707378387451, + "learning_rate": 1.5081422924901189e-05, + "loss": 2.551692771911621, + "step": 90720 + }, + { + "epoch": 0.24701238600789438, + "grad_norm": 2.16257381439209, + "learning_rate": 1.5080332560992233e-05, + "loss": 2.524300384521484, + "step": 90740 + }, + { + "epoch": 0.2470668299986389, + "grad_norm": 2.1800193786621094, + "learning_rate": 1.5079242197083277e-05, + "loss": 2.540528106689453, + "step": 90760 + }, + { + "epoch": 0.2471212739893834, + "grad_norm": 2.109083652496338, + "learning_rate": 1.5078151833174324e-05, + "loss": 2.5229070663452147, + "step": 90780 + }, + { + "epoch": 0.24717571798012794, + "grad_norm": 2.0564608573913574, + "learning_rate": 1.5077061469265369e-05, + "loss": 2.490589904785156, + "step": 90800 + }, + { + "epoch": 0.24723016197087247, + "grad_norm": 2.3981282711029053, + "learning_rate": 1.5075971105356413e-05, + "loss": 2.5169918060302736, + "step": 90820 + }, + { + "epoch": 0.247284605961617, + "grad_norm": 2.283454656600952, + "learning_rate": 1.507488074144746e-05, + "loss": 2.520920753479004, + "step": 90840 + }, + { + "epoch": 0.2473390499523615, + "grad_norm": 2.136758327484131, + "learning_rate": 1.5073790377538504e-05, + "loss": 2.496215057373047, + "step": 90860 + }, + { + "epoch": 0.24739349394310603, + "grad_norm": 2.1665642261505127, + "learning_rate": 1.507270001362955e-05, + "loss": 2.514403533935547, + "step": 90880 + }, + { + "epoch": 0.24744793793385056, + "grad_norm": 2.132622003555298, + "learning_rate": 1.5071609649720596e-05, + "loss": 2.5098981857299805, + "step": 90900 + }, + { + "epoch": 0.24750238192459506, + "grad_norm": 2.182710886001587, + "learning_rate": 1.507051928581164e-05, + "loss": 2.499271011352539, + "step": 90920 + }, + { + "epoch": 0.2475568259153396, + "grad_norm": 2.2917075157165527, + "learning_rate": 1.5069428921902688e-05, + "loss": 2.5088924407958983, + "step": 90940 + }, + { + "epoch": 0.24761126990608412, + "grad_norm": 2.08325457572937, + "learning_rate": 1.5068338557993732e-05, + "loss": 2.4925174713134766, + "step": 90960 + }, + { + "epoch": 0.24766571389682865, + "grad_norm": 2.092160940170288, + "learning_rate": 1.5067248194084777e-05, + "loss": 2.4948328018188475, + "step": 90980 + }, + { + "epoch": 0.24772015788757315, + "grad_norm": 2.112668752670288, + "learning_rate": 1.5066157830175823e-05, + "loss": 2.526656913757324, + "step": 91000 + }, + { + "epoch": 0.24777460187831768, + "grad_norm": 2.115536689758301, + "learning_rate": 1.5065067466266867e-05, + "loss": 2.514077568054199, + "step": 91020 + }, + { + "epoch": 0.2478290458690622, + "grad_norm": 2.095891237258911, + "learning_rate": 1.5063977102357913e-05, + "loss": 2.5266395568847657, + "step": 91040 + }, + { + "epoch": 0.2478834898598067, + "grad_norm": 2.9241039752960205, + "learning_rate": 1.5062886738448959e-05, + "loss": 2.5309822082519533, + "step": 91060 + }, + { + "epoch": 0.24793793385055124, + "grad_norm": 2.095176935195923, + "learning_rate": 1.5061796374540003e-05, + "loss": 2.491827201843262, + "step": 91080 + }, + { + "epoch": 0.24799237784129577, + "grad_norm": 2.1050069332122803, + "learning_rate": 1.506070601063105e-05, + "loss": 2.4996992111206056, + "step": 91100 + }, + { + "epoch": 0.2480468218320403, + "grad_norm": 2.1686208248138428, + "learning_rate": 1.5059615646722095e-05, + "loss": 2.499745750427246, + "step": 91120 + }, + { + "epoch": 0.2481012658227848, + "grad_norm": 2.1723146438598633, + "learning_rate": 1.505852528281314e-05, + "loss": 2.51190128326416, + "step": 91140 + }, + { + "epoch": 0.24815570981352933, + "grad_norm": 2.1370174884796143, + "learning_rate": 1.5057434918904186e-05, + "loss": 2.4975559234619142, + "step": 91160 + }, + { + "epoch": 0.24821015380427386, + "grad_norm": 2.1618945598602295, + "learning_rate": 1.505634455499523e-05, + "loss": 2.4945453643798827, + "step": 91180 + }, + { + "epoch": 0.2482645977950184, + "grad_norm": 2.043947458267212, + "learning_rate": 1.5055254191086276e-05, + "loss": 2.525583267211914, + "step": 91200 + }, + { + "epoch": 0.2483190417857629, + "grad_norm": 2.129570484161377, + "learning_rate": 1.5054163827177322e-05, + "loss": 2.5083393096923827, + "step": 91220 + }, + { + "epoch": 0.24837348577650742, + "grad_norm": 2.2573747634887695, + "learning_rate": 1.5053073463268366e-05, + "loss": 2.4960277557373045, + "step": 91240 + }, + { + "epoch": 0.24842792976725195, + "grad_norm": 2.1481127738952637, + "learning_rate": 1.5051983099359414e-05, + "loss": 2.511062431335449, + "step": 91260 + }, + { + "epoch": 0.24848237375799645, + "grad_norm": 2.2213099002838135, + "learning_rate": 1.5050892735450458e-05, + "loss": 2.5268144607543945, + "step": 91280 + }, + { + "epoch": 0.24853681774874098, + "grad_norm": 2.2530014514923096, + "learning_rate": 1.5049802371541504e-05, + "loss": 2.509360504150391, + "step": 91300 + }, + { + "epoch": 0.2485912617394855, + "grad_norm": 2.09067964553833, + "learning_rate": 1.504871200763255e-05, + "loss": 2.473433494567871, + "step": 91320 + }, + { + "epoch": 0.24864570573023004, + "grad_norm": 2.2083652019500732, + "learning_rate": 1.5047621643723594e-05, + "loss": 2.49145450592041, + "step": 91340 + }, + { + "epoch": 0.24870014972097454, + "grad_norm": 2.1829211711883545, + "learning_rate": 1.504653127981464e-05, + "loss": 2.5092227935791014, + "step": 91360 + }, + { + "epoch": 0.24875459371171907, + "grad_norm": 2.095442533493042, + "learning_rate": 1.5045440915905685e-05, + "loss": 2.53173770904541, + "step": 91380 + }, + { + "epoch": 0.2488090377024636, + "grad_norm": 2.1725261211395264, + "learning_rate": 1.504435055199673e-05, + "loss": 2.5245525360107424, + "step": 91400 + }, + { + "epoch": 0.2488634816932081, + "grad_norm": 2.1503732204437256, + "learning_rate": 1.5043260188087777e-05, + "loss": 2.5318431854248047, + "step": 91420 + }, + { + "epoch": 0.24891792568395263, + "grad_norm": 2.092852830886841, + "learning_rate": 1.5042169824178821e-05, + "loss": 2.4995235443115233, + "step": 91440 + }, + { + "epoch": 0.24897236967469716, + "grad_norm": 2.0022575855255127, + "learning_rate": 1.5041079460269867e-05, + "loss": 2.535932922363281, + "step": 91460 + }, + { + "epoch": 0.2490268136654417, + "grad_norm": 2.155230760574341, + "learning_rate": 1.5039989096360913e-05, + "loss": 2.528537559509277, + "step": 91480 + }, + { + "epoch": 0.2490812576561862, + "grad_norm": 2.1839182376861572, + "learning_rate": 1.5038898732451957e-05, + "loss": 2.499323844909668, + "step": 91500 + }, + { + "epoch": 0.24913570164693072, + "grad_norm": 2.2511236667633057, + "learning_rate": 1.5037808368543002e-05, + "loss": 2.5108409881591798, + "step": 91520 + }, + { + "epoch": 0.24919014563767525, + "grad_norm": 1.9611742496490479, + "learning_rate": 1.5036718004634048e-05, + "loss": 2.5299909591674803, + "step": 91540 + }, + { + "epoch": 0.24924458962841975, + "grad_norm": 2.215461254119873, + "learning_rate": 1.5035627640725092e-05, + "loss": 2.5442161560058594, + "step": 91560 + }, + { + "epoch": 0.24929903361916428, + "grad_norm": 2.439068555831909, + "learning_rate": 1.503453727681614e-05, + "loss": 2.5098545074462892, + "step": 91580 + }, + { + "epoch": 0.2493534776099088, + "grad_norm": 2.168065309524536, + "learning_rate": 1.5033446912907184e-05, + "loss": 2.5295459747314455, + "step": 91600 + }, + { + "epoch": 0.24940792160065334, + "grad_norm": 2.2144932746887207, + "learning_rate": 1.503235654899823e-05, + "loss": 2.4982276916503907, + "step": 91620 + }, + { + "epoch": 0.24946236559139784, + "grad_norm": 2.148498296737671, + "learning_rate": 1.5031266185089276e-05, + "loss": 2.499031639099121, + "step": 91640 + }, + { + "epoch": 0.24951680958214237, + "grad_norm": 2.1817245483398438, + "learning_rate": 1.503017582118032e-05, + "loss": 2.522668647766113, + "step": 91660 + }, + { + "epoch": 0.2495712535728869, + "grad_norm": 2.258105754852295, + "learning_rate": 1.5029085457271366e-05, + "loss": 2.509684753417969, + "step": 91680 + }, + { + "epoch": 0.24962569756363143, + "grad_norm": 2.1983773708343506, + "learning_rate": 1.5027995093362411e-05, + "loss": 2.4917959213256835, + "step": 91700 + }, + { + "epoch": 0.24968014155437593, + "grad_norm": 2.0560989379882812, + "learning_rate": 1.5026904729453455e-05, + "loss": 2.5013830184936525, + "step": 91720 + }, + { + "epoch": 0.24973458554512046, + "grad_norm": 2.1181507110595703, + "learning_rate": 1.5025814365544503e-05, + "loss": 2.5314041137695313, + "step": 91740 + }, + { + "epoch": 0.249789029535865, + "grad_norm": 2.060760498046875, + "learning_rate": 1.5024724001635547e-05, + "loss": 2.4865114212036135, + "step": 91760 + }, + { + "epoch": 0.2498434735266095, + "grad_norm": 2.1006922721862793, + "learning_rate": 1.5023633637726593e-05, + "loss": 2.5052373886108397, + "step": 91780 + }, + { + "epoch": 0.24989791751735402, + "grad_norm": 2.0597500801086426, + "learning_rate": 1.5022543273817639e-05, + "loss": 2.4836572647094726, + "step": 91800 + }, + { + "epoch": 0.24995236150809855, + "grad_norm": 2.0669596195220947, + "learning_rate": 1.5021452909908683e-05, + "loss": 2.5366939544677733, + "step": 91820 + }, + { + "epoch": 0.2500068054988431, + "grad_norm": 2.2482011318206787, + "learning_rate": 1.5020362545999729e-05, + "loss": 2.520591926574707, + "step": 91840 + }, + { + "epoch": 0.2500612494895876, + "grad_norm": 2.1370644569396973, + "learning_rate": 1.5019272182090774e-05, + "loss": 2.52169246673584, + "step": 91860 + }, + { + "epoch": 0.25011569348033214, + "grad_norm": 2.220853567123413, + "learning_rate": 1.5018181818181819e-05, + "loss": 2.495357894897461, + "step": 91880 + }, + { + "epoch": 0.25017013747107664, + "grad_norm": 2.062633991241455, + "learning_rate": 1.5017091454272864e-05, + "loss": 2.505534553527832, + "step": 91900 + }, + { + "epoch": 0.25022458146182114, + "grad_norm": 2.412236213684082, + "learning_rate": 1.501600109036391e-05, + "loss": 2.4952259063720703, + "step": 91920 + }, + { + "epoch": 0.2502790254525657, + "grad_norm": 2.1578328609466553, + "learning_rate": 1.5014910726454956e-05, + "loss": 2.5140872955322267, + "step": 91940 + }, + { + "epoch": 0.2503334694433102, + "grad_norm": 2.1545865535736084, + "learning_rate": 1.5013820362546002e-05, + "loss": 2.5197689056396486, + "step": 91960 + }, + { + "epoch": 0.2503879134340547, + "grad_norm": 2.087090015411377, + "learning_rate": 1.5012729998637046e-05, + "loss": 2.5403024673461916, + "step": 91980 + }, + { + "epoch": 0.25044235742479926, + "grad_norm": 2.0930051803588867, + "learning_rate": 1.5011639634728092e-05, + "loss": 2.5053178787231447, + "step": 92000 + }, + { + "epoch": 0.25049680141554376, + "grad_norm": 2.110948324203491, + "learning_rate": 1.5010549270819138e-05, + "loss": 2.5153512954711914, + "step": 92020 + }, + { + "epoch": 0.25055124540628826, + "grad_norm": 2.164412498474121, + "learning_rate": 1.5009458906910182e-05, + "loss": 2.5292301177978516, + "step": 92040 + }, + { + "epoch": 0.2506056893970328, + "grad_norm": 2.268672227859497, + "learning_rate": 1.5008368543001227e-05, + "loss": 2.531870460510254, + "step": 92060 + }, + { + "epoch": 0.2506601333877773, + "grad_norm": 2.2276394367218018, + "learning_rate": 1.5007278179092273e-05, + "loss": 2.5060264587402346, + "step": 92080 + }, + { + "epoch": 0.2507145773785218, + "grad_norm": 2.05421781539917, + "learning_rate": 1.5006187815183319e-05, + "loss": 2.516712951660156, + "step": 92100 + }, + { + "epoch": 0.2507690213692664, + "grad_norm": 2.0363588333129883, + "learning_rate": 1.5005097451274365e-05, + "loss": 2.509097671508789, + "step": 92120 + }, + { + "epoch": 0.2508234653600109, + "grad_norm": 2.23820161819458, + "learning_rate": 1.5004007087365409e-05, + "loss": 2.502617073059082, + "step": 92140 + }, + { + "epoch": 0.25087790935075543, + "grad_norm": 2.127821683883667, + "learning_rate": 1.5002916723456455e-05, + "loss": 2.509027671813965, + "step": 92160 + }, + { + "epoch": 0.25093235334149994, + "grad_norm": 2.1917288303375244, + "learning_rate": 1.50018263595475e-05, + "loss": 2.4916568756103517, + "step": 92180 + }, + { + "epoch": 0.25098679733224444, + "grad_norm": 2.180908679962158, + "learning_rate": 1.5000735995638545e-05, + "loss": 2.515562057495117, + "step": 92200 + }, + { + "epoch": 0.251041241322989, + "grad_norm": 2.10880446434021, + "learning_rate": 1.499964563172959e-05, + "loss": 2.501008224487305, + "step": 92220 + }, + { + "epoch": 0.2510956853137335, + "grad_norm": 2.074723482131958, + "learning_rate": 1.4998555267820636e-05, + "loss": 2.5439695358276366, + "step": 92240 + }, + { + "epoch": 0.251150129304478, + "grad_norm": 2.0840554237365723, + "learning_rate": 1.4997464903911682e-05, + "loss": 2.5068286895751952, + "step": 92260 + }, + { + "epoch": 0.25120457329522256, + "grad_norm": 2.1495723724365234, + "learning_rate": 1.4996374540002728e-05, + "loss": 2.530460548400879, + "step": 92280 + }, + { + "epoch": 0.25125901728596706, + "grad_norm": 2.0685856342315674, + "learning_rate": 1.4995284176093772e-05, + "loss": 2.4818046569824217, + "step": 92300 + }, + { + "epoch": 0.25131346127671156, + "grad_norm": 2.173583507537842, + "learning_rate": 1.4994193812184818e-05, + "loss": 2.513485908508301, + "step": 92320 + }, + { + "epoch": 0.2513679052674561, + "grad_norm": 2.1438963413238525, + "learning_rate": 1.4993103448275864e-05, + "loss": 2.5218006134033204, + "step": 92340 + }, + { + "epoch": 0.2514223492582006, + "grad_norm": 2.1365597248077393, + "learning_rate": 1.4992013084366908e-05, + "loss": 2.4944215774536134, + "step": 92360 + }, + { + "epoch": 0.2514767932489452, + "grad_norm": 2.1351401805877686, + "learning_rate": 1.4990922720457954e-05, + "loss": 2.492892265319824, + "step": 92380 + }, + { + "epoch": 0.2515312372396897, + "grad_norm": 2.0746138095855713, + "learning_rate": 1.4989832356549e-05, + "loss": 2.5227603912353516, + "step": 92400 + }, + { + "epoch": 0.2515856812304342, + "grad_norm": 2.1396095752716064, + "learning_rate": 1.4988741992640045e-05, + "loss": 2.5073530197143556, + "step": 92420 + }, + { + "epoch": 0.25164012522117873, + "grad_norm": 2.0539894104003906, + "learning_rate": 1.4987651628731091e-05, + "loss": 2.473760223388672, + "step": 92440 + }, + { + "epoch": 0.25169456921192324, + "grad_norm": 2.413628578186035, + "learning_rate": 1.4986561264822135e-05, + "loss": 2.5166831970214845, + "step": 92460 + }, + { + "epoch": 0.25174901320266774, + "grad_norm": 2.0447046756744385, + "learning_rate": 1.4985470900913181e-05, + "loss": 2.500782012939453, + "step": 92480 + }, + { + "epoch": 0.2518034571934123, + "grad_norm": 2.1010305881500244, + "learning_rate": 1.4984380537004227e-05, + "loss": 2.4885232925415037, + "step": 92500 + }, + { + "epoch": 0.2518579011841568, + "grad_norm": 2.392120838165283, + "learning_rate": 1.4983290173095271e-05, + "loss": 2.533016014099121, + "step": 92520 + }, + { + "epoch": 0.2519123451749013, + "grad_norm": 2.196035146713257, + "learning_rate": 1.4982199809186317e-05, + "loss": 2.5027786254882813, + "step": 92540 + }, + { + "epoch": 0.25196678916564585, + "grad_norm": 2.1199722290039062, + "learning_rate": 1.4981109445277363e-05, + "loss": 2.4778076171875, + "step": 92560 + }, + { + "epoch": 0.25202123315639036, + "grad_norm": 2.1126186847686768, + "learning_rate": 1.4980019081368408e-05, + "loss": 2.5102413177490233, + "step": 92580 + }, + { + "epoch": 0.25207567714713486, + "grad_norm": 2.1722710132598877, + "learning_rate": 1.4978928717459454e-05, + "loss": 2.5094783782958983, + "step": 92600 + }, + { + "epoch": 0.2521301211378794, + "grad_norm": 2.24106764793396, + "learning_rate": 1.4977838353550498e-05, + "loss": 2.516971206665039, + "step": 92620 + }, + { + "epoch": 0.2521845651286239, + "grad_norm": 2.079728603363037, + "learning_rate": 1.4976747989641544e-05, + "loss": 2.515371894836426, + "step": 92640 + }, + { + "epoch": 0.2522390091193685, + "grad_norm": 2.203977108001709, + "learning_rate": 1.497565762573259e-05, + "loss": 2.5219951629638673, + "step": 92660 + }, + { + "epoch": 0.252293453110113, + "grad_norm": 2.6743884086608887, + "learning_rate": 1.4974567261823636e-05, + "loss": 2.5259729385375977, + "step": 92680 + }, + { + "epoch": 0.2523478971008575, + "grad_norm": 2.3427505493164062, + "learning_rate": 1.497347689791468e-05, + "loss": 2.5292289733886717, + "step": 92700 + }, + { + "epoch": 0.25240234109160203, + "grad_norm": 2.0900821685791016, + "learning_rate": 1.4972386534005726e-05, + "loss": 2.50854606628418, + "step": 92720 + }, + { + "epoch": 0.25245678508234654, + "grad_norm": 2.1898276805877686, + "learning_rate": 1.4971296170096771e-05, + "loss": 2.510354995727539, + "step": 92740 + }, + { + "epoch": 0.25251122907309104, + "grad_norm": 2.2141783237457275, + "learning_rate": 1.4970205806187817e-05, + "loss": 2.5267351150512694, + "step": 92760 + }, + { + "epoch": 0.2525656730638356, + "grad_norm": 2.21517014503479, + "learning_rate": 1.4969115442278861e-05, + "loss": 2.509586143493652, + "step": 92780 + }, + { + "epoch": 0.2526201170545801, + "grad_norm": 2.128455400466919, + "learning_rate": 1.4968025078369907e-05, + "loss": 2.4861124038696287, + "step": 92800 + }, + { + "epoch": 0.2526745610453246, + "grad_norm": 2.2427875995635986, + "learning_rate": 1.4966934714460953e-05, + "loss": 2.5226329803466796, + "step": 92820 + }, + { + "epoch": 0.25272900503606915, + "grad_norm": 2.2212884426116943, + "learning_rate": 1.4965844350551999e-05, + "loss": 2.5164113998413087, + "step": 92840 + }, + { + "epoch": 0.25278344902681366, + "grad_norm": 2.159898519515991, + "learning_rate": 1.4964753986643043e-05, + "loss": 2.541303253173828, + "step": 92860 + }, + { + "epoch": 0.2528378930175582, + "grad_norm": 2.177175998687744, + "learning_rate": 1.4963663622734089e-05, + "loss": 2.517203712463379, + "step": 92880 + }, + { + "epoch": 0.2528923370083027, + "grad_norm": 2.056551456451416, + "learning_rate": 1.4962573258825134e-05, + "loss": 2.5243951797485353, + "step": 92900 + }, + { + "epoch": 0.2529467809990472, + "grad_norm": 2.2197484970092773, + "learning_rate": 1.496148289491618e-05, + "loss": 2.507059860229492, + "step": 92920 + }, + { + "epoch": 0.2530012249897918, + "grad_norm": 2.159548044204712, + "learning_rate": 1.4960392531007224e-05, + "loss": 2.515131950378418, + "step": 92940 + }, + { + "epoch": 0.2530556689805363, + "grad_norm": 2.1185922622680664, + "learning_rate": 1.495930216709827e-05, + "loss": 2.5276803970336914, + "step": 92960 + }, + { + "epoch": 0.2531101129712808, + "grad_norm": 2.276968479156494, + "learning_rate": 1.4958211803189316e-05, + "loss": 2.514722442626953, + "step": 92980 + }, + { + "epoch": 0.25316455696202533, + "grad_norm": 2.1735174655914307, + "learning_rate": 1.4957121439280362e-05, + "loss": 2.5522884368896483, + "step": 93000 + }, + { + "epoch": 0.25321900095276983, + "grad_norm": 2.09358549118042, + "learning_rate": 1.4956031075371406e-05, + "loss": 2.538827896118164, + "step": 93020 + }, + { + "epoch": 0.25327344494351434, + "grad_norm": 2.180082082748413, + "learning_rate": 1.4954940711462452e-05, + "loss": 2.5281930923461915, + "step": 93040 + }, + { + "epoch": 0.2533278889342589, + "grad_norm": 2.1782848834991455, + "learning_rate": 1.4953850347553498e-05, + "loss": 2.4880868911743166, + "step": 93060 + }, + { + "epoch": 0.2533823329250034, + "grad_norm": 2.0005788803100586, + "learning_rate": 1.4952759983644542e-05, + "loss": 2.5020767211914063, + "step": 93080 + }, + { + "epoch": 0.2534367769157479, + "grad_norm": 2.1314244270324707, + "learning_rate": 1.4951669619735587e-05, + "loss": 2.537187194824219, + "step": 93100 + }, + { + "epoch": 0.25349122090649245, + "grad_norm": 2.250875234603882, + "learning_rate": 1.4950579255826633e-05, + "loss": 2.534898376464844, + "step": 93120 + }, + { + "epoch": 0.25354566489723696, + "grad_norm": 2.170055866241455, + "learning_rate": 1.4949488891917679e-05, + "loss": 2.5017576217651367, + "step": 93140 + }, + { + "epoch": 0.2536001088879815, + "grad_norm": 2.0929431915283203, + "learning_rate": 1.4948398528008725e-05, + "loss": 2.4849178314208986, + "step": 93160 + }, + { + "epoch": 0.253654552878726, + "grad_norm": 2.105581045150757, + "learning_rate": 1.4947308164099769e-05, + "loss": 2.4853799819946287, + "step": 93180 + }, + { + "epoch": 0.2537089968694705, + "grad_norm": 3.3690125942230225, + "learning_rate": 1.4946217800190815e-05, + "loss": 2.5155014038085937, + "step": 93200 + }, + { + "epoch": 0.2537634408602151, + "grad_norm": 2.1021976470947266, + "learning_rate": 1.494512743628186e-05, + "loss": 2.5016559600830077, + "step": 93220 + }, + { + "epoch": 0.2538178848509596, + "grad_norm": 2.189685821533203, + "learning_rate": 1.4944037072372905e-05, + "loss": 2.5423006057739257, + "step": 93240 + }, + { + "epoch": 0.2538723288417041, + "grad_norm": 2.1735806465148926, + "learning_rate": 1.494294670846395e-05, + "loss": 2.517268180847168, + "step": 93260 + }, + { + "epoch": 0.25392677283244863, + "grad_norm": 2.2640433311462402, + "learning_rate": 1.4941856344554996e-05, + "loss": 2.550157165527344, + "step": 93280 + }, + { + "epoch": 0.25398121682319313, + "grad_norm": 2.1078524589538574, + "learning_rate": 1.4940765980646042e-05, + "loss": 2.520288276672363, + "step": 93300 + }, + { + "epoch": 0.25403566081393764, + "grad_norm": 2.1397805213928223, + "learning_rate": 1.4939675616737088e-05, + "loss": 2.5366209030151365, + "step": 93320 + }, + { + "epoch": 0.2540901048046822, + "grad_norm": 2.060227155685425, + "learning_rate": 1.4938585252828132e-05, + "loss": 2.5016218185424806, + "step": 93340 + }, + { + "epoch": 0.2541445487954267, + "grad_norm": 2.2033047676086426, + "learning_rate": 1.4937494888919178e-05, + "loss": 2.4932689666748047, + "step": 93360 + }, + { + "epoch": 0.25419899278617125, + "grad_norm": 2.174675226211548, + "learning_rate": 1.4936404525010224e-05, + "loss": 2.4811368942260743, + "step": 93380 + }, + { + "epoch": 0.25425343677691575, + "grad_norm": 2.0946872234344482, + "learning_rate": 1.4935314161101268e-05, + "loss": 2.5353296279907225, + "step": 93400 + }, + { + "epoch": 0.25430788076766025, + "grad_norm": 2.1665260791778564, + "learning_rate": 1.4934223797192314e-05, + "loss": 2.496848487854004, + "step": 93420 + }, + { + "epoch": 0.2543623247584048, + "grad_norm": 2.211972951889038, + "learning_rate": 1.493313343328336e-05, + "loss": 2.518970489501953, + "step": 93440 + }, + { + "epoch": 0.2544167687491493, + "grad_norm": 2.549968719482422, + "learning_rate": 1.4932043069374405e-05, + "loss": 2.498686981201172, + "step": 93460 + }, + { + "epoch": 0.2544712127398938, + "grad_norm": 2.1810173988342285, + "learning_rate": 1.4930952705465451e-05, + "loss": 2.5426483154296875, + "step": 93480 + }, + { + "epoch": 0.25452565673063837, + "grad_norm": 2.091993570327759, + "learning_rate": 1.4929862341556495e-05, + "loss": 2.5204530715942384, + "step": 93500 + }, + { + "epoch": 0.2545801007213829, + "grad_norm": 2.11655855178833, + "learning_rate": 1.4928771977647541e-05, + "loss": 2.5011825561523438, + "step": 93520 + }, + { + "epoch": 0.2546345447121274, + "grad_norm": 2.0970520973205566, + "learning_rate": 1.4927681613738587e-05, + "loss": 2.51981258392334, + "step": 93540 + }, + { + "epoch": 0.25468898870287193, + "grad_norm": 2.2533786296844482, + "learning_rate": 1.4926591249829631e-05, + "loss": 2.5325870513916016, + "step": 93560 + }, + { + "epoch": 0.25474343269361643, + "grad_norm": 2.1500072479248047, + "learning_rate": 1.4925500885920677e-05, + "loss": 2.5497129440307615, + "step": 93580 + }, + { + "epoch": 0.254797876684361, + "grad_norm": 2.1301803588867188, + "learning_rate": 1.4924410522011723e-05, + "loss": 2.501534652709961, + "step": 93600 + }, + { + "epoch": 0.2548523206751055, + "grad_norm": 2.1828196048736572, + "learning_rate": 1.4923320158102768e-05, + "loss": 2.4943353652954103, + "step": 93620 + }, + { + "epoch": 0.25490676466585, + "grad_norm": 2.150847911834717, + "learning_rate": 1.4922229794193814e-05, + "loss": 2.5419254302978516, + "step": 93640 + }, + { + "epoch": 0.25496120865659455, + "grad_norm": 2.224656105041504, + "learning_rate": 1.4921139430284858e-05, + "loss": 2.5162492752075196, + "step": 93660 + }, + { + "epoch": 0.25501565264733905, + "grad_norm": 2.1062662601470947, + "learning_rate": 1.4920049066375904e-05, + "loss": 2.4936006546020506, + "step": 93680 + }, + { + "epoch": 0.25507009663808355, + "grad_norm": 2.137169599533081, + "learning_rate": 1.491895870246695e-05, + "loss": 2.5278324127197265, + "step": 93700 + }, + { + "epoch": 0.2551245406288281, + "grad_norm": 2.323225259780884, + "learning_rate": 1.4917868338557994e-05, + "loss": 2.540643310546875, + "step": 93720 + }, + { + "epoch": 0.2551789846195726, + "grad_norm": 2.2897090911865234, + "learning_rate": 1.491677797464904e-05, + "loss": 2.486084747314453, + "step": 93740 + }, + { + "epoch": 0.2552334286103171, + "grad_norm": 2.1248786449432373, + "learning_rate": 1.4915687610740086e-05, + "loss": 2.4869220733642576, + "step": 93760 + }, + { + "epoch": 0.25528787260106167, + "grad_norm": 2.164567232131958, + "learning_rate": 1.4914597246831131e-05, + "loss": 2.4963220596313476, + "step": 93780 + }, + { + "epoch": 0.2553423165918062, + "grad_norm": 2.117107629776001, + "learning_rate": 1.4913506882922177e-05, + "loss": 2.5267093658447264, + "step": 93800 + }, + { + "epoch": 0.2553967605825507, + "grad_norm": 2.1821813583374023, + "learning_rate": 1.4912416519013221e-05, + "loss": 2.4978052139282227, + "step": 93820 + }, + { + "epoch": 0.25545120457329523, + "grad_norm": 2.1440553665161133, + "learning_rate": 1.4911326155104267e-05, + "loss": 2.493012046813965, + "step": 93840 + }, + { + "epoch": 0.25550564856403973, + "grad_norm": 2.6692166328430176, + "learning_rate": 1.4910235791195313e-05, + "loss": 2.526677703857422, + "step": 93860 + }, + { + "epoch": 0.2555600925547843, + "grad_norm": 2.104203701019287, + "learning_rate": 1.4909145427286357e-05, + "loss": 2.505017852783203, + "step": 93880 + }, + { + "epoch": 0.2556145365455288, + "grad_norm": 2.0921998023986816, + "learning_rate": 1.4908055063377403e-05, + "loss": 2.509401321411133, + "step": 93900 + }, + { + "epoch": 0.2556689805362733, + "grad_norm": 2.2047665119171143, + "learning_rate": 1.4906964699468449e-05, + "loss": 2.533970069885254, + "step": 93920 + }, + { + "epoch": 0.25572342452701785, + "grad_norm": 3.331112861633301, + "learning_rate": 1.4905874335559495e-05, + "loss": 2.492984962463379, + "step": 93940 + }, + { + "epoch": 0.25577786851776235, + "grad_norm": 2.120258092880249, + "learning_rate": 1.490478397165054e-05, + "loss": 2.5054609298706056, + "step": 93960 + }, + { + "epoch": 0.25583231250850685, + "grad_norm": 2.038205146789551, + "learning_rate": 1.4903693607741584e-05, + "loss": 2.4849407196044924, + "step": 93980 + }, + { + "epoch": 0.2558867564992514, + "grad_norm": 2.198925733566284, + "learning_rate": 1.490260324383263e-05, + "loss": 2.522709274291992, + "step": 94000 + }, + { + "epoch": 0.2559412004899959, + "grad_norm": 2.236508369445801, + "learning_rate": 1.4901512879923676e-05, + "loss": 2.517752456665039, + "step": 94020 + }, + { + "epoch": 0.2559956444807404, + "grad_norm": 2.1140828132629395, + "learning_rate": 1.490042251601472e-05, + "loss": 2.5042644500732423, + "step": 94040 + }, + { + "epoch": 0.25605008847148497, + "grad_norm": 2.1283321380615234, + "learning_rate": 1.4899332152105766e-05, + "loss": 2.5114133834838865, + "step": 94060 + }, + { + "epoch": 0.2561045324622295, + "grad_norm": 2.102595090866089, + "learning_rate": 1.4898241788196812e-05, + "loss": 2.5468238830566405, + "step": 94080 + }, + { + "epoch": 0.25615897645297403, + "grad_norm": 2.539895534515381, + "learning_rate": 1.4897151424287856e-05, + "loss": 2.495160675048828, + "step": 94100 + }, + { + "epoch": 0.25621342044371853, + "grad_norm": 2.0498876571655273, + "learning_rate": 1.4896061060378903e-05, + "loss": 2.4968677520751954, + "step": 94120 + }, + { + "epoch": 0.25626786443446303, + "grad_norm": 2.0531082153320312, + "learning_rate": 1.4894970696469948e-05, + "loss": 2.5061288833618165, + "step": 94140 + }, + { + "epoch": 0.2563223084252076, + "grad_norm": 2.237785816192627, + "learning_rate": 1.4893880332560993e-05, + "loss": 2.514628791809082, + "step": 94160 + }, + { + "epoch": 0.2563767524159521, + "grad_norm": 2.04093861579895, + "learning_rate": 1.489278996865204e-05, + "loss": 2.5180870056152345, + "step": 94180 + }, + { + "epoch": 0.2564311964066966, + "grad_norm": 2.0437560081481934, + "learning_rate": 1.4891699604743083e-05, + "loss": 2.533846855163574, + "step": 94200 + }, + { + "epoch": 0.25648564039744115, + "grad_norm": 2.1903653144836426, + "learning_rate": 1.4890609240834129e-05, + "loss": 2.538465118408203, + "step": 94220 + }, + { + "epoch": 0.25654008438818565, + "grad_norm": 2.4152023792266846, + "learning_rate": 1.4889518876925175e-05, + "loss": 2.5205972671508787, + "step": 94240 + }, + { + "epoch": 0.25659452837893015, + "grad_norm": 2.365657091140747, + "learning_rate": 1.4888428513016219e-05, + "loss": 2.4681371688842773, + "step": 94260 + }, + { + "epoch": 0.2566489723696747, + "grad_norm": 2.2715647220611572, + "learning_rate": 1.4887338149107267e-05, + "loss": 2.5098155975341796, + "step": 94280 + }, + { + "epoch": 0.2567034163604192, + "grad_norm": 2.3393454551696777, + "learning_rate": 1.488624778519831e-05, + "loss": 2.515144920349121, + "step": 94300 + }, + { + "epoch": 0.2567578603511637, + "grad_norm": 2.2109766006469727, + "learning_rate": 1.4885157421289356e-05, + "loss": 2.5262172698974608, + "step": 94320 + }, + { + "epoch": 0.25681230434190827, + "grad_norm": 2.5819783210754395, + "learning_rate": 1.4884067057380402e-05, + "loss": 2.5218860626220705, + "step": 94340 + }, + { + "epoch": 0.2568667483326528, + "grad_norm": 2.31036639213562, + "learning_rate": 1.4882976693471446e-05, + "loss": 2.5319337844848633, + "step": 94360 + }, + { + "epoch": 0.25692119232339733, + "grad_norm": 5.167435169219971, + "learning_rate": 1.4881886329562494e-05, + "loss": 2.523164176940918, + "step": 94380 + }, + { + "epoch": 0.25697563631414183, + "grad_norm": 2.096463918685913, + "learning_rate": 1.4880795965653538e-05, + "loss": 2.4836021423339845, + "step": 94400 + }, + { + "epoch": 0.25703008030488633, + "grad_norm": 2.210212469100952, + "learning_rate": 1.4879705601744582e-05, + "loss": 2.514759063720703, + "step": 94420 + }, + { + "epoch": 0.2570845242956309, + "grad_norm": 2.151750326156616, + "learning_rate": 1.487861523783563e-05, + "loss": 2.5006752014160156, + "step": 94440 + }, + { + "epoch": 0.2571389682863754, + "grad_norm": 3.52060866355896, + "learning_rate": 1.4877524873926674e-05, + "loss": 2.4874753952026367, + "step": 94460 + }, + { + "epoch": 0.2571934122771199, + "grad_norm": 2.263481378555298, + "learning_rate": 1.487643451001772e-05, + "loss": 2.508363151550293, + "step": 94480 + }, + { + "epoch": 0.25724785626786445, + "grad_norm": 2.400479316711426, + "learning_rate": 1.4875344146108765e-05, + "loss": 2.516294479370117, + "step": 94500 + }, + { + "epoch": 0.25730230025860895, + "grad_norm": 2.144554376602173, + "learning_rate": 1.487425378219981e-05, + "loss": 2.48998966217041, + "step": 94520 + }, + { + "epoch": 0.25735674424935345, + "grad_norm": 2.085963726043701, + "learning_rate": 1.4873163418290857e-05, + "loss": 2.509836196899414, + "step": 94540 + }, + { + "epoch": 0.257411188240098, + "grad_norm": 2.139961004257202, + "learning_rate": 1.4872073054381901e-05, + "loss": 2.5190193176269533, + "step": 94560 + }, + { + "epoch": 0.2574656322308425, + "grad_norm": 2.0654826164245605, + "learning_rate": 1.4870982690472945e-05, + "loss": 2.5493555068969727, + "step": 94580 + }, + { + "epoch": 0.25752007622158707, + "grad_norm": 2.123553514480591, + "learning_rate": 1.4869892326563993e-05, + "loss": 2.5146858215332033, + "step": 94600 + }, + { + "epoch": 0.25757452021233157, + "grad_norm": 2.132044792175293, + "learning_rate": 1.4868801962655037e-05, + "loss": 2.541079521179199, + "step": 94620 + }, + { + "epoch": 0.25762896420307607, + "grad_norm": 3.118863821029663, + "learning_rate": 1.4867711598746083e-05, + "loss": 2.500472640991211, + "step": 94640 + }, + { + "epoch": 0.25768340819382063, + "grad_norm": 2.188540458679199, + "learning_rate": 1.4866621234837128e-05, + "loss": 2.468549346923828, + "step": 94660 + }, + { + "epoch": 0.25773785218456513, + "grad_norm": 2.2578561305999756, + "learning_rate": 1.4865530870928173e-05, + "loss": 2.509750556945801, + "step": 94680 + }, + { + "epoch": 0.25779229617530963, + "grad_norm": 2.2972872257232666, + "learning_rate": 1.486444050701922e-05, + "loss": 2.520713233947754, + "step": 94700 + }, + { + "epoch": 0.2578467401660542, + "grad_norm": 2.1555988788604736, + "learning_rate": 1.4863350143110264e-05, + "loss": 2.5035295486450195, + "step": 94720 + }, + { + "epoch": 0.2579011841567987, + "grad_norm": 2.1414146423339844, + "learning_rate": 1.4862259779201308e-05, + "loss": 2.5284191131591798, + "step": 94740 + }, + { + "epoch": 0.2579556281475432, + "grad_norm": 2.1368441581726074, + "learning_rate": 1.4861169415292356e-05, + "loss": 2.529792594909668, + "step": 94760 + }, + { + "epoch": 0.25801007213828775, + "grad_norm": 2.4573400020599365, + "learning_rate": 1.48600790513834e-05, + "loss": 2.4839744567871094, + "step": 94780 + }, + { + "epoch": 0.25806451612903225, + "grad_norm": 2.242121934890747, + "learning_rate": 1.4858988687474446e-05, + "loss": 2.512656593322754, + "step": 94800 + }, + { + "epoch": 0.25811896011977675, + "grad_norm": 2.128856897354126, + "learning_rate": 1.4857898323565492e-05, + "loss": 2.5170249938964844, + "step": 94820 + }, + { + "epoch": 0.2581734041105213, + "grad_norm": 2.1190245151519775, + "learning_rate": 1.4856807959656536e-05, + "loss": 2.488232421875, + "step": 94840 + }, + { + "epoch": 0.2582278481012658, + "grad_norm": 2.1795692443847656, + "learning_rate": 1.4855717595747583e-05, + "loss": 2.5076677322387697, + "step": 94860 + }, + { + "epoch": 0.25828229209201037, + "grad_norm": 2.098276376724243, + "learning_rate": 1.4854627231838627e-05, + "loss": 2.506919288635254, + "step": 94880 + }, + { + "epoch": 0.25833673608275487, + "grad_norm": 2.0484981536865234, + "learning_rate": 1.4853536867929671e-05, + "loss": 2.5294073104858397, + "step": 94900 + }, + { + "epoch": 0.25839118007349937, + "grad_norm": 2.253994941711426, + "learning_rate": 1.4852446504020719e-05, + "loss": 2.512893295288086, + "step": 94920 + }, + { + "epoch": 0.25844562406424393, + "grad_norm": 2.1415629386901855, + "learning_rate": 1.4851356140111763e-05, + "loss": 2.46673583984375, + "step": 94940 + }, + { + "epoch": 0.25850006805498843, + "grad_norm": 2.209989309310913, + "learning_rate": 1.4850265776202809e-05, + "loss": 2.5205476760864256, + "step": 94960 + }, + { + "epoch": 0.25855451204573293, + "grad_norm": 2.3956053256988525, + "learning_rate": 1.4849175412293855e-05, + "loss": 2.494091033935547, + "step": 94980 + }, + { + "epoch": 0.2586089560364775, + "grad_norm": 2.0439188480377197, + "learning_rate": 1.4848085048384899e-05, + "loss": 2.5410091400146486, + "step": 95000 + }, + { + "epoch": 0.258663400027222, + "grad_norm": 2.22347354888916, + "learning_rate": 1.4846994684475946e-05, + "loss": 2.511337471008301, + "step": 95020 + }, + { + "epoch": 0.2587178440179665, + "grad_norm": 2.226346731185913, + "learning_rate": 1.484590432056699e-05, + "loss": 2.4861919403076174, + "step": 95040 + }, + { + "epoch": 0.25877228800871105, + "grad_norm": 2.1042559146881104, + "learning_rate": 1.4844813956658034e-05, + "loss": 2.508806610107422, + "step": 95060 + }, + { + "epoch": 0.25882673199945555, + "grad_norm": 2.0563361644744873, + "learning_rate": 1.4843723592749082e-05, + "loss": 2.4874635696411134, + "step": 95080 + }, + { + "epoch": 0.2588811759902001, + "grad_norm": 2.1552531719207764, + "learning_rate": 1.4842633228840126e-05, + "loss": 2.5379125595092775, + "step": 95100 + }, + { + "epoch": 0.2589356199809446, + "grad_norm": 2.2694058418273926, + "learning_rate": 1.484154286493117e-05, + "loss": 2.5157749176025392, + "step": 95120 + }, + { + "epoch": 0.2589900639716891, + "grad_norm": 2.1102187633514404, + "learning_rate": 1.4840452501022218e-05, + "loss": 2.4945674896240235, + "step": 95140 + }, + { + "epoch": 0.25904450796243367, + "grad_norm": 2.255227565765381, + "learning_rate": 1.4839362137113262e-05, + "loss": 2.4836294174194338, + "step": 95160 + }, + { + "epoch": 0.25909895195317817, + "grad_norm": 2.383361577987671, + "learning_rate": 1.483827177320431e-05, + "loss": 2.4906049728393556, + "step": 95180 + }, + { + "epoch": 0.25915339594392267, + "grad_norm": 2.0868101119995117, + "learning_rate": 1.4837181409295353e-05, + "loss": 2.5220985412597656, + "step": 95200 + }, + { + "epoch": 0.25920783993466723, + "grad_norm": 2.1441774368286133, + "learning_rate": 1.4836091045386398e-05, + "loss": 2.5307756423950196, + "step": 95220 + }, + { + "epoch": 0.25926228392541173, + "grad_norm": 2.086928606033325, + "learning_rate": 1.4835000681477445e-05, + "loss": 2.5342037200927736, + "step": 95240 + }, + { + "epoch": 0.25931672791615623, + "grad_norm": 2.3251163959503174, + "learning_rate": 1.4833910317568489e-05, + "loss": 2.4823450088500976, + "step": 95260 + }, + { + "epoch": 0.2593711719069008, + "grad_norm": 2.1629176139831543, + "learning_rate": 1.4832819953659533e-05, + "loss": 2.503109169006348, + "step": 95280 + }, + { + "epoch": 0.2594256158976453, + "grad_norm": 2.162156820297241, + "learning_rate": 1.483172958975058e-05, + "loss": 2.5264656066894533, + "step": 95300 + }, + { + "epoch": 0.2594800598883898, + "grad_norm": 2.144041061401367, + "learning_rate": 1.4830639225841625e-05, + "loss": 2.489270782470703, + "step": 95320 + }, + { + "epoch": 0.25953450387913435, + "grad_norm": 2.1698544025421143, + "learning_rate": 1.4829548861932672e-05, + "loss": 2.5212160110473634, + "step": 95340 + }, + { + "epoch": 0.25958894786987885, + "grad_norm": 2.1364383697509766, + "learning_rate": 1.4828458498023717e-05, + "loss": 2.5201854705810547, + "step": 95360 + }, + { + "epoch": 0.2596433918606234, + "grad_norm": 2.2997968196868896, + "learning_rate": 1.482736813411476e-05, + "loss": 2.5093923568725587, + "step": 95380 + }, + { + "epoch": 0.2596978358513679, + "grad_norm": 2.2694876194000244, + "learning_rate": 1.4826277770205808e-05, + "loss": 2.5453058242797852, + "step": 95400 + }, + { + "epoch": 0.2597522798421124, + "grad_norm": 2.181464195251465, + "learning_rate": 1.4825187406296852e-05, + "loss": 2.5542322158813477, + "step": 95420 + }, + { + "epoch": 0.25980672383285697, + "grad_norm": 2.070835828781128, + "learning_rate": 1.4824097042387896e-05, + "loss": 2.5041357040405274, + "step": 95440 + }, + { + "epoch": 0.25986116782360147, + "grad_norm": 2.2150661945343018, + "learning_rate": 1.4823006678478944e-05, + "loss": 2.468563461303711, + "step": 95460 + }, + { + "epoch": 0.25991561181434597, + "grad_norm": 2.1401758193969727, + "learning_rate": 1.4821916314569988e-05, + "loss": 2.490087699890137, + "step": 95480 + }, + { + "epoch": 0.2599700558050905, + "grad_norm": 2.128775119781494, + "learning_rate": 1.4820825950661035e-05, + "loss": 2.5224431991577148, + "step": 95500 + }, + { + "epoch": 0.26002449979583503, + "grad_norm": 2.4150142669677734, + "learning_rate": 1.481973558675208e-05, + "loss": 2.5004674911499025, + "step": 95520 + }, + { + "epoch": 0.26007894378657953, + "grad_norm": 2.3657314777374268, + "learning_rate": 1.4818645222843124e-05, + "loss": 2.522560882568359, + "step": 95540 + }, + { + "epoch": 0.2601333877773241, + "grad_norm": 2.212491750717163, + "learning_rate": 1.4817554858934171e-05, + "loss": 2.522820472717285, + "step": 95560 + }, + { + "epoch": 0.2601878317680686, + "grad_norm": 2.214944362640381, + "learning_rate": 1.4816464495025215e-05, + "loss": 2.5430917739868164, + "step": 95580 + }, + { + "epoch": 0.26024227575881315, + "grad_norm": 2.126131772994995, + "learning_rate": 1.481537413111626e-05, + "loss": 2.4968774795532225, + "step": 95600 + }, + { + "epoch": 0.26029671974955765, + "grad_norm": 2.1573116779327393, + "learning_rate": 1.4814283767207307e-05, + "loss": 2.5198678970336914, + "step": 95620 + }, + { + "epoch": 0.26035116374030215, + "grad_norm": 2.074204206466675, + "learning_rate": 1.4813193403298351e-05, + "loss": 2.50341682434082, + "step": 95640 + }, + { + "epoch": 0.2604056077310467, + "grad_norm": 2.089972734451294, + "learning_rate": 1.4812103039389399e-05, + "loss": 2.501303863525391, + "step": 95660 + }, + { + "epoch": 0.2604600517217912, + "grad_norm": 2.1846530437469482, + "learning_rate": 1.4811012675480443e-05, + "loss": 2.513604736328125, + "step": 95680 + }, + { + "epoch": 0.2605144957125357, + "grad_norm": 2.1126718521118164, + "learning_rate": 1.4809922311571487e-05, + "loss": 2.541438674926758, + "step": 95700 + }, + { + "epoch": 0.26056893970328027, + "grad_norm": 2.020554542541504, + "learning_rate": 1.4808831947662534e-05, + "loss": 2.451371955871582, + "step": 95720 + }, + { + "epoch": 0.26062338369402477, + "grad_norm": 2.1115729808807373, + "learning_rate": 1.4807741583753578e-05, + "loss": 2.51656551361084, + "step": 95740 + }, + { + "epoch": 0.26067782768476927, + "grad_norm": 2.587578773498535, + "learning_rate": 1.4806651219844623e-05, + "loss": 2.5047861099243165, + "step": 95760 + }, + { + "epoch": 0.2607322716755138, + "grad_norm": 2.2279064655303955, + "learning_rate": 1.480556085593567e-05, + "loss": 2.5078853607177733, + "step": 95780 + }, + { + "epoch": 0.26078671566625833, + "grad_norm": 2.095621347427368, + "learning_rate": 1.4804470492026714e-05, + "loss": 2.516128158569336, + "step": 95800 + }, + { + "epoch": 0.2608411596570029, + "grad_norm": 2.1874773502349854, + "learning_rate": 1.4803380128117762e-05, + "loss": 2.4911340713500976, + "step": 95820 + }, + { + "epoch": 0.2608956036477474, + "grad_norm": 2.104555606842041, + "learning_rate": 1.4802289764208806e-05, + "loss": 2.5214242935180664, + "step": 95840 + }, + { + "epoch": 0.2609500476384919, + "grad_norm": 2.1211564540863037, + "learning_rate": 1.480119940029985e-05, + "loss": 2.5190959930419923, + "step": 95860 + }, + { + "epoch": 0.26100449162923645, + "grad_norm": 1.9985841512680054, + "learning_rate": 1.4800109036390897e-05, + "loss": 2.473496437072754, + "step": 95880 + }, + { + "epoch": 0.26105893561998095, + "grad_norm": 2.16851544380188, + "learning_rate": 1.4799018672481941e-05, + "loss": 2.5126657485961914, + "step": 95900 + }, + { + "epoch": 0.26111337961072545, + "grad_norm": 2.4638164043426514, + "learning_rate": 1.4797928308572986e-05, + "loss": 2.515640640258789, + "step": 95920 + }, + { + "epoch": 0.26116782360147, + "grad_norm": 2.222115993499756, + "learning_rate": 1.4796837944664033e-05, + "loss": 2.5021072387695313, + "step": 95940 + }, + { + "epoch": 0.2612222675922145, + "grad_norm": 2.223517894744873, + "learning_rate": 1.4795747580755077e-05, + "loss": 2.4833606719970702, + "step": 95960 + }, + { + "epoch": 0.261276711582959, + "grad_norm": 2.127859115600586, + "learning_rate": 1.4794657216846125e-05, + "loss": 2.516097640991211, + "step": 95980 + }, + { + "epoch": 0.26133115557370357, + "grad_norm": 2.2342894077301025, + "learning_rate": 1.4793566852937169e-05, + "loss": 2.488195610046387, + "step": 96000 + }, + { + "epoch": 0.26138559956444807, + "grad_norm": 2.2203056812286377, + "learning_rate": 1.4792476489028213e-05, + "loss": 2.5096563339233398, + "step": 96020 + }, + { + "epoch": 0.26144004355519257, + "grad_norm": 2.1136345863342285, + "learning_rate": 1.479138612511926e-05, + "loss": 2.5279590606689455, + "step": 96040 + }, + { + "epoch": 0.2614944875459371, + "grad_norm": 3.114624500274658, + "learning_rate": 1.4790295761210305e-05, + "loss": 2.496767044067383, + "step": 96060 + }, + { + "epoch": 0.26154893153668163, + "grad_norm": 2.1186275482177734, + "learning_rate": 1.4789205397301352e-05, + "loss": 2.554507827758789, + "step": 96080 + }, + { + "epoch": 0.2616033755274262, + "grad_norm": 2.161459445953369, + "learning_rate": 1.4788115033392396e-05, + "loss": 2.5032934188842773, + "step": 96100 + }, + { + "epoch": 0.2616578195181707, + "grad_norm": 2.1230013370513916, + "learning_rate": 1.478702466948344e-05, + "loss": 2.4977903366088867, + "step": 96120 + }, + { + "epoch": 0.2617122635089152, + "grad_norm": 2.151698589324951, + "learning_rate": 1.4785934305574488e-05, + "loss": 2.5401189804077147, + "step": 96140 + }, + { + "epoch": 0.26176670749965975, + "grad_norm": 2.534975528717041, + "learning_rate": 1.4784843941665532e-05, + "loss": 2.4881927490234377, + "step": 96160 + }, + { + "epoch": 0.26182115149040425, + "grad_norm": 2.2249324321746826, + "learning_rate": 1.4783753577756576e-05, + "loss": 2.480255889892578, + "step": 96180 + }, + { + "epoch": 0.26187559548114875, + "grad_norm": 2.123997449874878, + "learning_rate": 1.4782663213847624e-05, + "loss": 2.5061731338500977, + "step": 96200 + }, + { + "epoch": 0.2619300394718933, + "grad_norm": 2.2897531986236572, + "learning_rate": 1.4781572849938668e-05, + "loss": 2.48483829498291, + "step": 96220 + }, + { + "epoch": 0.2619844834626378, + "grad_norm": 2.156157970428467, + "learning_rate": 1.4780482486029715e-05, + "loss": 2.518441581726074, + "step": 96240 + }, + { + "epoch": 0.2620389274533823, + "grad_norm": 2.102342128753662, + "learning_rate": 1.477939212212076e-05, + "loss": 2.512181854248047, + "step": 96260 + }, + { + "epoch": 0.26209337144412687, + "grad_norm": 2.3925204277038574, + "learning_rate": 1.4778301758211803e-05, + "loss": 2.4997173309326173, + "step": 96280 + }, + { + "epoch": 0.26214781543487137, + "grad_norm": 2.4447922706604004, + "learning_rate": 1.4777211394302851e-05, + "loss": 2.521799850463867, + "step": 96300 + }, + { + "epoch": 0.2622022594256159, + "grad_norm": 2.163752317428589, + "learning_rate": 1.4776121030393895e-05, + "loss": 2.5073484420776366, + "step": 96320 + }, + { + "epoch": 0.2622567034163604, + "grad_norm": 2.0728118419647217, + "learning_rate": 1.4775030666484939e-05, + "loss": 2.533366012573242, + "step": 96340 + }, + { + "epoch": 0.2623111474071049, + "grad_norm": 2.212122678756714, + "learning_rate": 1.4773940302575987e-05, + "loss": 2.4809701919555662, + "step": 96360 + }, + { + "epoch": 0.2623655913978495, + "grad_norm": 2.1337878704071045, + "learning_rate": 1.477284993866703e-05, + "loss": 2.496371269226074, + "step": 96380 + }, + { + "epoch": 0.262420035388594, + "grad_norm": 2.1852481365203857, + "learning_rate": 1.4771759574758078e-05, + "loss": 2.5078659057617188, + "step": 96400 + }, + { + "epoch": 0.2624744793793385, + "grad_norm": 2.1753792762756348, + "learning_rate": 1.4770669210849122e-05, + "loss": 2.5380477905273438, + "step": 96420 + }, + { + "epoch": 0.26252892337008304, + "grad_norm": 2.0938034057617188, + "learning_rate": 1.4769578846940166e-05, + "loss": 2.5304357528686525, + "step": 96440 + }, + { + "epoch": 0.26258336736082755, + "grad_norm": 2.2589030265808105, + "learning_rate": 1.4768488483031214e-05, + "loss": 2.5149255752563477, + "step": 96460 + }, + { + "epoch": 0.26263781135157205, + "grad_norm": 2.2517600059509277, + "learning_rate": 1.4767398119122258e-05, + "loss": 2.510616683959961, + "step": 96480 + }, + { + "epoch": 0.2626922553423166, + "grad_norm": 2.111239433288574, + "learning_rate": 1.4766307755213302e-05, + "loss": 2.498382568359375, + "step": 96500 + }, + { + "epoch": 0.2627466993330611, + "grad_norm": 2.241828441619873, + "learning_rate": 1.476521739130435e-05, + "loss": 2.4808347702026365, + "step": 96520 + }, + { + "epoch": 0.2628011433238056, + "grad_norm": 2.058195114135742, + "learning_rate": 1.4764127027395394e-05, + "loss": 2.5212913513183595, + "step": 96540 + }, + { + "epoch": 0.26285558731455017, + "grad_norm": 2.302783966064453, + "learning_rate": 1.4763036663486441e-05, + "loss": 2.4889385223388674, + "step": 96560 + }, + { + "epoch": 0.26291003130529467, + "grad_norm": 2.1004152297973633, + "learning_rate": 1.4761946299577485e-05, + "loss": 2.5353450775146484, + "step": 96580 + }, + { + "epoch": 0.2629644752960392, + "grad_norm": 2.3982770442962646, + "learning_rate": 1.476085593566853e-05, + "loss": 2.5118009567260744, + "step": 96600 + }, + { + "epoch": 0.2630189192867837, + "grad_norm": 2.1683735847473145, + "learning_rate": 1.4759765571759577e-05, + "loss": 2.5015039443969727, + "step": 96620 + }, + { + "epoch": 0.2630733632775282, + "grad_norm": 2.193192958831787, + "learning_rate": 1.4758675207850621e-05, + "loss": 2.508672904968262, + "step": 96640 + }, + { + "epoch": 0.2631278072682728, + "grad_norm": 2.0871617794036865, + "learning_rate": 1.4757584843941665e-05, + "loss": 2.5201595306396483, + "step": 96660 + }, + { + "epoch": 0.2631822512590173, + "grad_norm": 2.1093387603759766, + "learning_rate": 1.4756494480032713e-05, + "loss": 2.533661460876465, + "step": 96680 + }, + { + "epoch": 0.2632366952497618, + "grad_norm": 2.5406293869018555, + "learning_rate": 1.4755404116123757e-05, + "loss": 2.539962577819824, + "step": 96700 + }, + { + "epoch": 0.26329113924050634, + "grad_norm": 2.161485195159912, + "learning_rate": 1.4754313752214804e-05, + "loss": 2.4977039337158202, + "step": 96720 + }, + { + "epoch": 0.26334558323125085, + "grad_norm": 2.1415703296661377, + "learning_rate": 1.4753223388305849e-05, + "loss": 2.5216915130615236, + "step": 96740 + }, + { + "epoch": 0.26340002722199535, + "grad_norm": 2.1135144233703613, + "learning_rate": 1.4752133024396893e-05, + "loss": 2.50766658782959, + "step": 96760 + }, + { + "epoch": 0.2634544712127399, + "grad_norm": 2.0009357929229736, + "learning_rate": 1.475104266048794e-05, + "loss": 2.504648780822754, + "step": 96780 + }, + { + "epoch": 0.2635089152034844, + "grad_norm": 2.060159206390381, + "learning_rate": 1.4749952296578984e-05, + "loss": 2.4834388732910155, + "step": 96800 + }, + { + "epoch": 0.26356335919422896, + "grad_norm": 2.240774631500244, + "learning_rate": 1.4748861932670028e-05, + "loss": 2.519709014892578, + "step": 96820 + }, + { + "epoch": 0.26361780318497346, + "grad_norm": 2.170107126235962, + "learning_rate": 1.4747771568761076e-05, + "loss": 2.5250581741333007, + "step": 96840 + }, + { + "epoch": 0.26367224717571797, + "grad_norm": 2.088809013366699, + "learning_rate": 1.474668120485212e-05, + "loss": 2.5456775665283202, + "step": 96860 + }, + { + "epoch": 0.2637266911664625, + "grad_norm": 2.176706314086914, + "learning_rate": 1.4745590840943168e-05, + "loss": 2.5345645904541017, + "step": 96880 + }, + { + "epoch": 0.263781135157207, + "grad_norm": 2.094202756881714, + "learning_rate": 1.4744500477034212e-05, + "loss": 2.5117446899414064, + "step": 96900 + }, + { + "epoch": 0.2638355791479515, + "grad_norm": 2.2886486053466797, + "learning_rate": 1.4743410113125256e-05, + "loss": 2.4984291076660154, + "step": 96920 + }, + { + "epoch": 0.2638900231386961, + "grad_norm": 2.1832668781280518, + "learning_rate": 1.4742319749216303e-05, + "loss": 2.486620330810547, + "step": 96940 + }, + { + "epoch": 0.2639444671294406, + "grad_norm": 2.193793296813965, + "learning_rate": 1.4741229385307347e-05, + "loss": 2.4966014862060546, + "step": 96960 + }, + { + "epoch": 0.2639989111201851, + "grad_norm": 2.031857967376709, + "learning_rate": 1.4740139021398391e-05, + "loss": 2.491903877258301, + "step": 96980 + }, + { + "epoch": 0.26405335511092964, + "grad_norm": 2.1060218811035156, + "learning_rate": 1.4739048657489439e-05, + "loss": 2.509645462036133, + "step": 97000 + }, + { + "epoch": 0.26410779910167415, + "grad_norm": 2.144242525100708, + "learning_rate": 1.4737958293580483e-05, + "loss": 2.5197385787963866, + "step": 97020 + }, + { + "epoch": 0.26416224309241865, + "grad_norm": 2.1158952713012695, + "learning_rate": 1.473686792967153e-05, + "loss": 2.527886390686035, + "step": 97040 + }, + { + "epoch": 0.2642166870831632, + "grad_norm": 2.08703875541687, + "learning_rate": 1.4735777565762575e-05, + "loss": 2.4765872955322266, + "step": 97060 + }, + { + "epoch": 0.2642711310739077, + "grad_norm": 2.228261709213257, + "learning_rate": 1.4734687201853619e-05, + "loss": 2.500438117980957, + "step": 97080 + }, + { + "epoch": 0.26432557506465226, + "grad_norm": 2.1477468013763428, + "learning_rate": 1.4733596837944666e-05, + "loss": 2.5333541870117187, + "step": 97100 + }, + { + "epoch": 0.26438001905539676, + "grad_norm": 2.1229848861694336, + "learning_rate": 1.473250647403571e-05, + "loss": 2.4816179275512695, + "step": 97120 + }, + { + "epoch": 0.26443446304614127, + "grad_norm": 2.2071337699890137, + "learning_rate": 1.4731416110126755e-05, + "loss": 2.5246591567993164, + "step": 97140 + }, + { + "epoch": 0.2644889070368858, + "grad_norm": 2.1608331203460693, + "learning_rate": 1.4730325746217802e-05, + "loss": 2.510728454589844, + "step": 97160 + }, + { + "epoch": 0.2645433510276303, + "grad_norm": 2.3233635425567627, + "learning_rate": 1.4729235382308846e-05, + "loss": 2.5056221008300783, + "step": 97180 + }, + { + "epoch": 0.2645977950183748, + "grad_norm": 2.1510379314422607, + "learning_rate": 1.4728145018399894e-05, + "loss": 2.5270664215087892, + "step": 97200 + }, + { + "epoch": 0.2646522390091194, + "grad_norm": 2.358851909637451, + "learning_rate": 1.4727054654490938e-05, + "loss": 2.502506446838379, + "step": 97220 + }, + { + "epoch": 0.2647066829998639, + "grad_norm": 2.177518844604492, + "learning_rate": 1.4725964290581982e-05, + "loss": 2.5025327682495115, + "step": 97240 + }, + { + "epoch": 0.2647611269906084, + "grad_norm": 2.166661024093628, + "learning_rate": 1.472487392667303e-05, + "loss": 2.523471450805664, + "step": 97260 + }, + { + "epoch": 0.26481557098135294, + "grad_norm": 2.1442909240722656, + "learning_rate": 1.4723783562764074e-05, + "loss": 2.48596134185791, + "step": 97280 + }, + { + "epoch": 0.26487001497209745, + "grad_norm": 2.092207670211792, + "learning_rate": 1.4722693198855118e-05, + "loss": 2.5536083221435546, + "step": 97300 + }, + { + "epoch": 0.264924458962842, + "grad_norm": 2.1963741779327393, + "learning_rate": 1.4721602834946165e-05, + "loss": 2.4903711318969726, + "step": 97320 + }, + { + "epoch": 0.2649789029535865, + "grad_norm": 2.193577289581299, + "learning_rate": 1.472051247103721e-05, + "loss": 2.5225862503051757, + "step": 97340 + }, + { + "epoch": 0.265033346944331, + "grad_norm": 2.21413254737854, + "learning_rate": 1.4719422107128257e-05, + "loss": 2.5154542922973633, + "step": 97360 + }, + { + "epoch": 0.26508779093507556, + "grad_norm": 2.1950299739837646, + "learning_rate": 1.4718331743219301e-05, + "loss": 2.5235897064208985, + "step": 97380 + }, + { + "epoch": 0.26514223492582006, + "grad_norm": 2.1122467517852783, + "learning_rate": 1.4717241379310345e-05, + "loss": 2.490077781677246, + "step": 97400 + }, + { + "epoch": 0.26519667891656457, + "grad_norm": 2.259026050567627, + "learning_rate": 1.4716151015401392e-05, + "loss": 2.507379913330078, + "step": 97420 + }, + { + "epoch": 0.2652511229073091, + "grad_norm": 2.1853060722351074, + "learning_rate": 1.4715060651492437e-05, + "loss": 2.496860694885254, + "step": 97440 + }, + { + "epoch": 0.2653055668980536, + "grad_norm": 2.194200277328491, + "learning_rate": 1.471397028758348e-05, + "loss": 2.5068374633789063, + "step": 97460 + }, + { + "epoch": 0.2653600108887981, + "grad_norm": 2.035963296890259, + "learning_rate": 1.4712879923674528e-05, + "loss": 2.5100231170654297, + "step": 97480 + }, + { + "epoch": 0.2654144548795427, + "grad_norm": 2.0141777992248535, + "learning_rate": 1.4711789559765572e-05, + "loss": 2.506264305114746, + "step": 97500 + }, + { + "epoch": 0.2654688988702872, + "grad_norm": 2.1910738945007324, + "learning_rate": 1.471069919585662e-05, + "loss": 2.4971364974975585, + "step": 97520 + }, + { + "epoch": 0.2655233428610317, + "grad_norm": 2.2375142574310303, + "learning_rate": 1.4709608831947664e-05, + "loss": 2.5435308456420898, + "step": 97540 + }, + { + "epoch": 0.26557778685177624, + "grad_norm": 2.268904685974121, + "learning_rate": 1.4708518468038708e-05, + "loss": 2.5224578857421873, + "step": 97560 + }, + { + "epoch": 0.26563223084252074, + "grad_norm": 2.5281875133514404, + "learning_rate": 1.4707428104129756e-05, + "loss": 2.503561019897461, + "step": 97580 + }, + { + "epoch": 0.2656866748332653, + "grad_norm": 2.0607190132141113, + "learning_rate": 1.47063377402208e-05, + "loss": 2.518318748474121, + "step": 97600 + }, + { + "epoch": 0.2657411188240098, + "grad_norm": 2.289013385772705, + "learning_rate": 1.4705247376311844e-05, + "loss": 2.5293300628662108, + "step": 97620 + }, + { + "epoch": 0.2657955628147543, + "grad_norm": 2.2040891647338867, + "learning_rate": 1.4704157012402891e-05, + "loss": 2.519044876098633, + "step": 97640 + }, + { + "epoch": 0.26585000680549886, + "grad_norm": 2.159043788909912, + "learning_rate": 1.4703066648493935e-05, + "loss": 2.5204063415527345, + "step": 97660 + }, + { + "epoch": 0.26590445079624336, + "grad_norm": 2.2785451412200928, + "learning_rate": 1.4701976284584983e-05, + "loss": 2.4833213806152346, + "step": 97680 + }, + { + "epoch": 0.26595889478698787, + "grad_norm": 3.8332231044769287, + "learning_rate": 1.4700885920676027e-05, + "loss": 2.491301345825195, + "step": 97700 + }, + { + "epoch": 0.2660133387777324, + "grad_norm": 2.7171576023101807, + "learning_rate": 1.4699795556767071e-05, + "loss": 2.484701728820801, + "step": 97720 + }, + { + "epoch": 0.2660677827684769, + "grad_norm": 2.2214341163635254, + "learning_rate": 1.4698705192858119e-05, + "loss": 2.5090755462646483, + "step": 97740 + }, + { + "epoch": 0.2661222267592214, + "grad_norm": 2.177319049835205, + "learning_rate": 1.4697614828949163e-05, + "loss": 2.4816612243652343, + "step": 97760 + }, + { + "epoch": 0.266176670749966, + "grad_norm": 2.1404523849487305, + "learning_rate": 1.469652446504021e-05, + "loss": 2.4951587677001954, + "step": 97780 + }, + { + "epoch": 0.2662311147407105, + "grad_norm": 2.1152987480163574, + "learning_rate": 1.4695434101131254e-05, + "loss": 2.506177520751953, + "step": 97800 + }, + { + "epoch": 0.26628555873145504, + "grad_norm": 2.107490301132202, + "learning_rate": 1.4694343737222299e-05, + "loss": 2.4876968383789064, + "step": 97820 + }, + { + "epoch": 0.26634000272219954, + "grad_norm": 2.1701223850250244, + "learning_rate": 1.4693253373313346e-05, + "loss": 2.5483985900878907, + "step": 97840 + }, + { + "epoch": 0.26639444671294404, + "grad_norm": 2.091485023498535, + "learning_rate": 1.469216300940439e-05, + "loss": 2.5250743865966796, + "step": 97860 + }, + { + "epoch": 0.2664488907036886, + "grad_norm": 2.1300437450408936, + "learning_rate": 1.4691072645495434e-05, + "loss": 2.532042694091797, + "step": 97880 + }, + { + "epoch": 0.2665033346944331, + "grad_norm": 2.187742233276367, + "learning_rate": 1.4689982281586482e-05, + "loss": 2.525736999511719, + "step": 97900 + }, + { + "epoch": 0.2665577786851776, + "grad_norm": 2.065608263015747, + "learning_rate": 1.4688891917677526e-05, + "loss": 2.4918424606323244, + "step": 97920 + }, + { + "epoch": 0.26661222267592216, + "grad_norm": 2.217653274536133, + "learning_rate": 1.4687801553768573e-05, + "loss": 2.5129161834716798, + "step": 97940 + }, + { + "epoch": 0.26666666666666666, + "grad_norm": 2.2810659408569336, + "learning_rate": 1.4686711189859617e-05, + "loss": 2.5131856918334963, + "step": 97960 + }, + { + "epoch": 0.26672111065741116, + "grad_norm": 2.229517936706543, + "learning_rate": 1.4685620825950662e-05, + "loss": 2.518198776245117, + "step": 97980 + }, + { + "epoch": 0.2667755546481557, + "grad_norm": 2.2608041763305664, + "learning_rate": 1.4684530462041709e-05, + "loss": 2.5192771911621095, + "step": 98000 + }, + { + "epoch": 0.2668299986389002, + "grad_norm": 2.262131929397583, + "learning_rate": 1.4683440098132753e-05, + "loss": 2.5018129348754883, + "step": 98020 + }, + { + "epoch": 0.2668844426296448, + "grad_norm": 2.0505452156066895, + "learning_rate": 1.4682349734223797e-05, + "loss": 2.498703956604004, + "step": 98040 + }, + { + "epoch": 0.2669388866203893, + "grad_norm": 2.135024309158325, + "learning_rate": 1.4681259370314845e-05, + "loss": 2.4919443130493164, + "step": 98060 + }, + { + "epoch": 0.2669933306111338, + "grad_norm": 2.1291487216949463, + "learning_rate": 1.4680169006405889e-05, + "loss": 2.5146581649780275, + "step": 98080 + }, + { + "epoch": 0.26704777460187834, + "grad_norm": 2.0515944957733154, + "learning_rate": 1.4679078642496935e-05, + "loss": 2.475507354736328, + "step": 98100 + }, + { + "epoch": 0.26710221859262284, + "grad_norm": 2.405493974685669, + "learning_rate": 1.467798827858798e-05, + "loss": 2.523290252685547, + "step": 98120 + }, + { + "epoch": 0.26715666258336734, + "grad_norm": 2.5205078125, + "learning_rate": 1.4676897914679025e-05, + "loss": 2.5257020950317384, + "step": 98140 + }, + { + "epoch": 0.2672111065741119, + "grad_norm": 2.038999319076538, + "learning_rate": 1.4675807550770072e-05, + "loss": 2.542085075378418, + "step": 98160 + }, + { + "epoch": 0.2672655505648564, + "grad_norm": 2.1552975177764893, + "learning_rate": 1.4674717186861116e-05, + "loss": 2.5213491439819338, + "step": 98180 + }, + { + "epoch": 0.2673199945556009, + "grad_norm": 2.10420298576355, + "learning_rate": 1.467362682295216e-05, + "loss": 2.5256128311157227, + "step": 98200 + }, + { + "epoch": 0.26737443854634546, + "grad_norm": 2.1567723751068115, + "learning_rate": 1.4672536459043208e-05, + "loss": 2.517792320251465, + "step": 98220 + }, + { + "epoch": 0.26742888253708996, + "grad_norm": 2.159670114517212, + "learning_rate": 1.4671446095134252e-05, + "loss": 2.4881683349609376, + "step": 98240 + }, + { + "epoch": 0.26748332652783446, + "grad_norm": 2.101219654083252, + "learning_rate": 1.4670355731225298e-05, + "loss": 2.518699836730957, + "step": 98260 + }, + { + "epoch": 0.267537770518579, + "grad_norm": 2.202754259109497, + "learning_rate": 1.4669265367316344e-05, + "loss": 2.5187503814697267, + "step": 98280 + }, + { + "epoch": 0.2675922145093235, + "grad_norm": 2.1994102001190186, + "learning_rate": 1.4668175003407388e-05, + "loss": 2.505843162536621, + "step": 98300 + }, + { + "epoch": 0.2676466585000681, + "grad_norm": 2.213944673538208, + "learning_rate": 1.4667084639498435e-05, + "loss": 2.510479736328125, + "step": 98320 + }, + { + "epoch": 0.2677011024908126, + "grad_norm": 2.2863776683807373, + "learning_rate": 1.466599427558948e-05, + "loss": 2.537653160095215, + "step": 98340 + }, + { + "epoch": 0.2677555464815571, + "grad_norm": 2.3358707427978516, + "learning_rate": 1.4664903911680524e-05, + "loss": 2.5230859756469726, + "step": 98360 + }, + { + "epoch": 0.26780999047230164, + "grad_norm": 2.089933395385742, + "learning_rate": 1.4663813547771571e-05, + "loss": 2.485587501525879, + "step": 98380 + }, + { + "epoch": 0.26786443446304614, + "grad_norm": 2.1950957775115967, + "learning_rate": 1.4662723183862615e-05, + "loss": 2.5253854751586915, + "step": 98400 + }, + { + "epoch": 0.26791887845379064, + "grad_norm": 2.213697671890259, + "learning_rate": 1.4661632819953661e-05, + "loss": 2.515915870666504, + "step": 98420 + }, + { + "epoch": 0.2679733224445352, + "grad_norm": 2.115419626235962, + "learning_rate": 1.4660542456044707e-05, + "loss": 2.5004913330078127, + "step": 98440 + }, + { + "epoch": 0.2680277664352797, + "grad_norm": 2.1684346199035645, + "learning_rate": 1.4659452092135751e-05, + "loss": 2.535763740539551, + "step": 98460 + }, + { + "epoch": 0.2680822104260242, + "grad_norm": 2.154116630554199, + "learning_rate": 1.4658361728226798e-05, + "loss": 2.4996225357055666, + "step": 98480 + }, + { + "epoch": 0.26813665441676876, + "grad_norm": 2.1974267959594727, + "learning_rate": 1.4657271364317842e-05, + "loss": 2.508875846862793, + "step": 98500 + }, + { + "epoch": 0.26819109840751326, + "grad_norm": 2.083944082260132, + "learning_rate": 1.4656181000408887e-05, + "loss": 2.498581314086914, + "step": 98520 + }, + { + "epoch": 0.2682455423982578, + "grad_norm": 2.2439637184143066, + "learning_rate": 1.4655090636499934e-05, + "loss": 2.4808250427246095, + "step": 98540 + }, + { + "epoch": 0.2682999863890023, + "grad_norm": 2.189481735229492, + "learning_rate": 1.4654000272590978e-05, + "loss": 2.4906589508056642, + "step": 98560 + }, + { + "epoch": 0.2683544303797468, + "grad_norm": 2.085118532180786, + "learning_rate": 1.4652909908682024e-05, + "loss": 2.5541406631469727, + "step": 98580 + }, + { + "epoch": 0.2684088743704914, + "grad_norm": 2.109198808670044, + "learning_rate": 1.465181954477307e-05, + "loss": 2.4884149551391603, + "step": 98600 + }, + { + "epoch": 0.2684633183612359, + "grad_norm": 2.1117732524871826, + "learning_rate": 1.4650729180864114e-05, + "loss": 2.5251306533813476, + "step": 98620 + }, + { + "epoch": 0.2685177623519804, + "grad_norm": 2.1274027824401855, + "learning_rate": 1.4649638816955161e-05, + "loss": 2.483651542663574, + "step": 98640 + }, + { + "epoch": 0.26857220634272494, + "grad_norm": 2.0832760334014893, + "learning_rate": 1.4648548453046206e-05, + "loss": 2.50985107421875, + "step": 98660 + }, + { + "epoch": 0.26862665033346944, + "grad_norm": 2.1048383712768555, + "learning_rate": 1.464745808913725e-05, + "loss": 2.4825878143310547, + "step": 98680 + }, + { + "epoch": 0.26868109432421394, + "grad_norm": 2.176877975463867, + "learning_rate": 1.4646367725228297e-05, + "loss": 2.5009185791015627, + "step": 98700 + }, + { + "epoch": 0.2687355383149585, + "grad_norm": 2.1500742435455322, + "learning_rate": 1.4645277361319341e-05, + "loss": 2.5014339447021485, + "step": 98720 + }, + { + "epoch": 0.268789982305703, + "grad_norm": 2.2583236694335938, + "learning_rate": 1.4644186997410387e-05, + "loss": 2.5130165100097654, + "step": 98740 + }, + { + "epoch": 0.2688444262964475, + "grad_norm": 2.2234878540039062, + "learning_rate": 1.4643096633501433e-05, + "loss": 2.522149658203125, + "step": 98760 + }, + { + "epoch": 0.26889887028719206, + "grad_norm": 2.124390125274658, + "learning_rate": 1.4642006269592477e-05, + "loss": 2.5449197769165037, + "step": 98780 + }, + { + "epoch": 0.26895331427793656, + "grad_norm": 2.677274227142334, + "learning_rate": 1.4640915905683525e-05, + "loss": 2.492312431335449, + "step": 98800 + }, + { + "epoch": 0.2690077582686811, + "grad_norm": 2.1565873622894287, + "learning_rate": 1.4639825541774569e-05, + "loss": 2.4916301727294923, + "step": 98820 + }, + { + "epoch": 0.2690622022594256, + "grad_norm": 2.2346408367156982, + "learning_rate": 1.4638735177865613e-05, + "loss": 2.5207807540893556, + "step": 98840 + }, + { + "epoch": 0.2691166462501701, + "grad_norm": 2.106719970703125, + "learning_rate": 1.463764481395666e-05, + "loss": 2.536482238769531, + "step": 98860 + }, + { + "epoch": 0.2691710902409147, + "grad_norm": 2.2173099517822266, + "learning_rate": 1.4636554450047704e-05, + "loss": 2.505009078979492, + "step": 98880 + }, + { + "epoch": 0.2692255342316592, + "grad_norm": 2.107348918914795, + "learning_rate": 1.463546408613875e-05, + "loss": 2.5163246154785157, + "step": 98900 + }, + { + "epoch": 0.2692799782224037, + "grad_norm": 2.104069709777832, + "learning_rate": 1.4634373722229796e-05, + "loss": 2.500605010986328, + "step": 98920 + }, + { + "epoch": 0.26933442221314824, + "grad_norm": 2.151437520980835, + "learning_rate": 1.463328335832084e-05, + "loss": 2.481230545043945, + "step": 98940 + }, + { + "epoch": 0.26938886620389274, + "grad_norm": 2.1065433025360107, + "learning_rate": 1.4632192994411888e-05, + "loss": 2.491745948791504, + "step": 98960 + }, + { + "epoch": 0.26944331019463724, + "grad_norm": 2.1560661792755127, + "learning_rate": 1.4631102630502932e-05, + "loss": 2.496672821044922, + "step": 98980 + }, + { + "epoch": 0.2694977541853818, + "grad_norm": 2.070753812789917, + "learning_rate": 1.4630012266593976e-05, + "loss": 2.501392364501953, + "step": 99000 + }, + { + "epoch": 0.2695521981761263, + "grad_norm": 2.2501726150512695, + "learning_rate": 1.4628921902685023e-05, + "loss": 2.5319330215454103, + "step": 99020 + }, + { + "epoch": 0.26960664216687086, + "grad_norm": 2.197327136993408, + "learning_rate": 1.4627831538776067e-05, + "loss": 2.519558334350586, + "step": 99040 + }, + { + "epoch": 0.26966108615761536, + "grad_norm": 2.1920087337493896, + "learning_rate": 1.4626741174867113e-05, + "loss": 2.503152084350586, + "step": 99060 + }, + { + "epoch": 0.26971553014835986, + "grad_norm": 2.1424341201782227, + "learning_rate": 1.4625650810958159e-05, + "loss": 2.4972904205322264, + "step": 99080 + }, + { + "epoch": 0.2697699741391044, + "grad_norm": 2.141186237335205, + "learning_rate": 1.4624560447049203e-05, + "loss": 2.499355506896973, + "step": 99100 + }, + { + "epoch": 0.2698244181298489, + "grad_norm": 2.0739428997039795, + "learning_rate": 1.4623470083140249e-05, + "loss": 2.5157697677612303, + "step": 99120 + }, + { + "epoch": 0.2698788621205934, + "grad_norm": 2.43076491355896, + "learning_rate": 1.4622379719231295e-05, + "loss": 2.518992233276367, + "step": 99140 + }, + { + "epoch": 0.269933306111338, + "grad_norm": 2.119086980819702, + "learning_rate": 1.4621289355322339e-05, + "loss": 2.504344367980957, + "step": 99160 + }, + { + "epoch": 0.2699877501020825, + "grad_norm": 2.230912208557129, + "learning_rate": 1.4620198991413386e-05, + "loss": 2.5076303482055664, + "step": 99180 + }, + { + "epoch": 0.270042194092827, + "grad_norm": 2.1619598865509033, + "learning_rate": 1.461910862750443e-05, + "loss": 2.4851613998413087, + "step": 99200 + }, + { + "epoch": 0.27009663808357154, + "grad_norm": 2.066709280014038, + "learning_rate": 1.4618018263595476e-05, + "loss": 2.502822685241699, + "step": 99220 + }, + { + "epoch": 0.27015108207431604, + "grad_norm": 2.1019210815429688, + "learning_rate": 1.4616927899686522e-05, + "loss": 2.507105255126953, + "step": 99240 + }, + { + "epoch": 0.27020552606506054, + "grad_norm": 2.2094104290008545, + "learning_rate": 1.4615837535777566e-05, + "loss": 2.506768798828125, + "step": 99260 + }, + { + "epoch": 0.2702599700558051, + "grad_norm": 2.13374924659729, + "learning_rate": 1.4614747171868612e-05, + "loss": 2.5393611907958986, + "step": 99280 + }, + { + "epoch": 0.2703144140465496, + "grad_norm": 2.225198745727539, + "learning_rate": 1.4613656807959658e-05, + "loss": 2.4908580780029297, + "step": 99300 + }, + { + "epoch": 0.27036885803729416, + "grad_norm": 2.176100969314575, + "learning_rate": 1.4612566444050702e-05, + "loss": 2.5322214126586915, + "step": 99320 + }, + { + "epoch": 0.27042330202803866, + "grad_norm": 2.150829792022705, + "learning_rate": 1.461147608014175e-05, + "loss": 2.4924192428588867, + "step": 99340 + }, + { + "epoch": 0.27047774601878316, + "grad_norm": 2.0585107803344727, + "learning_rate": 1.4610385716232794e-05, + "loss": 2.4825292587280274, + "step": 99360 + }, + { + "epoch": 0.2705321900095277, + "grad_norm": 2.230712652206421, + "learning_rate": 1.460929535232384e-05, + "loss": 2.513664436340332, + "step": 99380 + }, + { + "epoch": 0.2705866340002722, + "grad_norm": 2.3120789527893066, + "learning_rate": 1.4608204988414885e-05, + "loss": 2.4938333511352537, + "step": 99400 + }, + { + "epoch": 0.2706410779910167, + "grad_norm": 2.8407251834869385, + "learning_rate": 1.460711462450593e-05, + "loss": 2.5315147399902345, + "step": 99420 + }, + { + "epoch": 0.2706955219817613, + "grad_norm": 2.1977431774139404, + "learning_rate": 1.4606024260596975e-05, + "loss": 2.4943059921264648, + "step": 99440 + }, + { + "epoch": 0.2707499659725058, + "grad_norm": 2.1655900478363037, + "learning_rate": 1.4604933896688021e-05, + "loss": 2.5084476470947266, + "step": 99460 + }, + { + "epoch": 0.2708044099632503, + "grad_norm": 2.117265462875366, + "learning_rate": 1.4603843532779067e-05, + "loss": 2.5221343994140626, + "step": 99480 + }, + { + "epoch": 0.27085885395399484, + "grad_norm": 2.181835412979126, + "learning_rate": 1.4602753168870113e-05, + "loss": 2.521086502075195, + "step": 99500 + }, + { + "epoch": 0.27091329794473934, + "grad_norm": 2.122413158416748, + "learning_rate": 1.4601662804961157e-05, + "loss": 2.5096330642700195, + "step": 99520 + }, + { + "epoch": 0.2709677419354839, + "grad_norm": 2.0982463359832764, + "learning_rate": 1.4600572441052203e-05, + "loss": 2.5005167007446287, + "step": 99540 + }, + { + "epoch": 0.2710221859262284, + "grad_norm": 2.146561861038208, + "learning_rate": 1.4599482077143248e-05, + "loss": 2.5306596755981445, + "step": 99560 + }, + { + "epoch": 0.2710766299169729, + "grad_norm": 2.1604506969451904, + "learning_rate": 1.4598391713234292e-05, + "loss": 2.5059669494628904, + "step": 99580 + }, + { + "epoch": 0.27113107390771746, + "grad_norm": 2.1368942260742188, + "learning_rate": 1.4597301349325338e-05, + "loss": 2.508559799194336, + "step": 99600 + }, + { + "epoch": 0.27118551789846196, + "grad_norm": 2.152259111404419, + "learning_rate": 1.4596210985416384e-05, + "loss": 2.4765451431274412, + "step": 99620 + }, + { + "epoch": 0.27123996188920646, + "grad_norm": 2.129385471343994, + "learning_rate": 1.459512062150743e-05, + "loss": 2.5226926803588867, + "step": 99640 + }, + { + "epoch": 0.271294405879951, + "grad_norm": 2.210056781768799, + "learning_rate": 1.4594030257598476e-05, + "loss": 2.516920280456543, + "step": 99660 + }, + { + "epoch": 0.2713488498706955, + "grad_norm": 2.247333526611328, + "learning_rate": 1.459293989368952e-05, + "loss": 2.5091428756713867, + "step": 99680 + }, + { + "epoch": 0.27140329386144, + "grad_norm": 2.1469264030456543, + "learning_rate": 1.4591849529780566e-05, + "loss": 2.4935325622558593, + "step": 99700 + }, + { + "epoch": 0.2714577378521846, + "grad_norm": 2.0636606216430664, + "learning_rate": 1.4590759165871611e-05, + "loss": 2.5169225692749024, + "step": 99720 + }, + { + "epoch": 0.2715121818429291, + "grad_norm": 2.1345248222351074, + "learning_rate": 1.4589668801962656e-05, + "loss": 2.5134206771850587, + "step": 99740 + }, + { + "epoch": 0.2715666258336736, + "grad_norm": 2.4065065383911133, + "learning_rate": 1.4588578438053701e-05, + "loss": 2.4985368728637694, + "step": 99760 + }, + { + "epoch": 0.27162106982441814, + "grad_norm": 2.1775095462799072, + "learning_rate": 1.4587488074144747e-05, + "loss": 2.485602378845215, + "step": 99780 + }, + { + "epoch": 0.27167551381516264, + "grad_norm": 2.2339212894439697, + "learning_rate": 1.4586397710235793e-05, + "loss": 2.5056385040283202, + "step": 99800 + }, + { + "epoch": 0.2717299578059072, + "grad_norm": 2.077514886856079, + "learning_rate": 1.4585307346326839e-05, + "loss": 2.479844665527344, + "step": 99820 + }, + { + "epoch": 0.2717844017966517, + "grad_norm": 2.29414439201355, + "learning_rate": 1.4584216982417883e-05, + "loss": 2.521129608154297, + "step": 99840 + }, + { + "epoch": 0.2718388457873962, + "grad_norm": 2.182318687438965, + "learning_rate": 1.4583126618508929e-05, + "loss": 2.496173858642578, + "step": 99860 + }, + { + "epoch": 0.27189328977814076, + "grad_norm": 2.016119956970215, + "learning_rate": 1.4582036254599975e-05, + "loss": 2.5076930999755858, + "step": 99880 + }, + { + "epoch": 0.27194773376888526, + "grad_norm": 2.372340440750122, + "learning_rate": 1.4580945890691019e-05, + "loss": 2.5003503799438476, + "step": 99900 + }, + { + "epoch": 0.27200217775962976, + "grad_norm": 2.130960702896118, + "learning_rate": 1.4579855526782064e-05, + "loss": 2.5019905090332033, + "step": 99920 + }, + { + "epoch": 0.2720566217503743, + "grad_norm": 2.1401891708374023, + "learning_rate": 1.457876516287311e-05, + "loss": 2.5120113372802733, + "step": 99940 + }, + { + "epoch": 0.2721110657411188, + "grad_norm": 2.1421117782592773, + "learning_rate": 1.4577674798964156e-05, + "loss": 2.52056770324707, + "step": 99960 + }, + { + "epoch": 0.2721655097318633, + "grad_norm": 2.2227845191955566, + "learning_rate": 1.4576584435055202e-05, + "loss": 2.494709777832031, + "step": 99980 + }, + { + "epoch": 0.2722199537226079, + "grad_norm": 2.2083568572998047, + "learning_rate": 1.4575494071146246e-05, + "loss": 2.482089614868164, + "step": 100000 + }, + { + "epoch": 0.2722743977133524, + "grad_norm": 2.5236637592315674, + "learning_rate": 1.4574403707237292e-05, + "loss": 2.5167564392089843, + "step": 100020 + }, + { + "epoch": 0.27232884170409694, + "grad_norm": 2.1311633586883545, + "learning_rate": 1.4573313343328338e-05, + "loss": 2.524494934082031, + "step": 100040 + }, + { + "epoch": 0.27238328569484144, + "grad_norm": 2.122359037399292, + "learning_rate": 1.4572222979419382e-05, + "loss": 2.5191749572753905, + "step": 100060 + }, + { + "epoch": 0.27243772968558594, + "grad_norm": 2.425294876098633, + "learning_rate": 1.4571132615510428e-05, + "loss": 2.4995353698730467, + "step": 100080 + }, + { + "epoch": 0.2724921736763305, + "grad_norm": 2.134467363357544, + "learning_rate": 1.4570042251601473e-05, + "loss": 2.53640193939209, + "step": 100100 + }, + { + "epoch": 0.272546617667075, + "grad_norm": 2.094633102416992, + "learning_rate": 1.4568951887692519e-05, + "loss": 2.5102703094482424, + "step": 100120 + }, + { + "epoch": 0.2726010616578195, + "grad_norm": 2.2680704593658447, + "learning_rate": 1.4567861523783565e-05, + "loss": 2.531856727600098, + "step": 100140 + }, + { + "epoch": 0.27265550564856406, + "grad_norm": 2.2175910472869873, + "learning_rate": 1.4566771159874609e-05, + "loss": 2.508311080932617, + "step": 100160 + }, + { + "epoch": 0.27270994963930856, + "grad_norm": 2.094189405441284, + "learning_rate": 1.4565680795965655e-05, + "loss": 2.5181032180786134, + "step": 100180 + }, + { + "epoch": 0.27276439363005306, + "grad_norm": 2.170328140258789, + "learning_rate": 1.45645904320567e-05, + "loss": 2.481995964050293, + "step": 100200 + }, + { + "epoch": 0.2728188376207976, + "grad_norm": 2.0668303966522217, + "learning_rate": 1.4563500068147745e-05, + "loss": 2.4757633209228516, + "step": 100220 + }, + { + "epoch": 0.2728732816115421, + "grad_norm": 2.184065580368042, + "learning_rate": 1.456240970423879e-05, + "loss": 2.5215326309204102, + "step": 100240 + }, + { + "epoch": 0.2729277256022867, + "grad_norm": 2.099510669708252, + "learning_rate": 1.4561319340329836e-05, + "loss": 2.4912675857543944, + "step": 100260 + }, + { + "epoch": 0.2729821695930312, + "grad_norm": 2.232748031616211, + "learning_rate": 1.4560228976420882e-05, + "loss": 2.5009403228759766, + "step": 100280 + }, + { + "epoch": 0.2730366135837757, + "grad_norm": 2.2864129543304443, + "learning_rate": 1.4559138612511926e-05, + "loss": 2.519230270385742, + "step": 100300 + }, + { + "epoch": 0.27309105757452024, + "grad_norm": 2.101667881011963, + "learning_rate": 1.4558048248602972e-05, + "loss": 2.504242515563965, + "step": 100320 + }, + { + "epoch": 0.27314550156526474, + "grad_norm": 2.114422559738159, + "learning_rate": 1.4556957884694018e-05, + "loss": 2.5283535003662108, + "step": 100340 + }, + { + "epoch": 0.27319994555600924, + "grad_norm": 2.133458375930786, + "learning_rate": 1.4555867520785064e-05, + "loss": 2.5244653701782225, + "step": 100360 + }, + { + "epoch": 0.2732543895467538, + "grad_norm": 2.1588401794433594, + "learning_rate": 1.4554777156876108e-05, + "loss": 2.5488861083984373, + "step": 100380 + }, + { + "epoch": 0.2733088335374983, + "grad_norm": 2.1889076232910156, + "learning_rate": 1.4553686792967154e-05, + "loss": 2.5586318969726562, + "step": 100400 + }, + { + "epoch": 0.2733632775282428, + "grad_norm": 2.146371841430664, + "learning_rate": 1.45525964290582e-05, + "loss": 2.5046974182128907, + "step": 100420 + }, + { + "epoch": 0.27341772151898736, + "grad_norm": 2.098376512527466, + "learning_rate": 1.4551506065149245e-05, + "loss": 2.478895568847656, + "step": 100440 + }, + { + "epoch": 0.27347216550973186, + "grad_norm": 2.1435346603393555, + "learning_rate": 1.455041570124029e-05, + "loss": 2.5005640029907226, + "step": 100460 + }, + { + "epoch": 0.27352660950047636, + "grad_norm": 2.2039051055908203, + "learning_rate": 1.4549325337331335e-05, + "loss": 2.4963298797607423, + "step": 100480 + }, + { + "epoch": 0.2735810534912209, + "grad_norm": 2.1802845001220703, + "learning_rate": 1.4548234973422381e-05, + "loss": 2.5232688903808596, + "step": 100500 + }, + { + "epoch": 0.2736354974819654, + "grad_norm": 2.2333550453186035, + "learning_rate": 1.4547144609513427e-05, + "loss": 2.466781806945801, + "step": 100520 + }, + { + "epoch": 0.27368994147271, + "grad_norm": 2.1864206790924072, + "learning_rate": 1.4546054245604471e-05, + "loss": 2.5035350799560545, + "step": 100540 + }, + { + "epoch": 0.2737443854634545, + "grad_norm": 2.2006585597991943, + "learning_rate": 1.4544963881695517e-05, + "loss": 2.526285743713379, + "step": 100560 + }, + { + "epoch": 0.273798829454199, + "grad_norm": 2.1464483737945557, + "learning_rate": 1.4543873517786563e-05, + "loss": 2.501804161071777, + "step": 100580 + }, + { + "epoch": 0.27385327344494353, + "grad_norm": 2.122985363006592, + "learning_rate": 1.4542783153877608e-05, + "loss": 2.511653518676758, + "step": 100600 + }, + { + "epoch": 0.27390771743568804, + "grad_norm": 2.1144697666168213, + "learning_rate": 1.4541692789968653e-05, + "loss": 2.496887969970703, + "step": 100620 + }, + { + "epoch": 0.27396216142643254, + "grad_norm": 2.1571006774902344, + "learning_rate": 1.4540602426059698e-05, + "loss": 2.5047527313232423, + "step": 100640 + }, + { + "epoch": 0.2740166054171771, + "grad_norm": 2.06219482421875, + "learning_rate": 1.4539512062150744e-05, + "loss": 2.5506643295288085, + "step": 100660 + }, + { + "epoch": 0.2740710494079216, + "grad_norm": 2.1709063053131104, + "learning_rate": 1.453842169824179e-05, + "loss": 2.483958625793457, + "step": 100680 + }, + { + "epoch": 0.2741254933986661, + "grad_norm": 2.1947550773620605, + "learning_rate": 1.4537331334332834e-05, + "loss": 2.520336151123047, + "step": 100700 + }, + { + "epoch": 0.27417993738941066, + "grad_norm": 2.1915125846862793, + "learning_rate": 1.453624097042388e-05, + "loss": 2.494228553771973, + "step": 100720 + }, + { + "epoch": 0.27423438138015516, + "grad_norm": 2.192984104156494, + "learning_rate": 1.4535150606514926e-05, + "loss": 2.513895606994629, + "step": 100740 + }, + { + "epoch": 0.2742888253708997, + "grad_norm": 2.084656000137329, + "learning_rate": 1.4534060242605971e-05, + "loss": 2.491116523742676, + "step": 100760 + }, + { + "epoch": 0.2743432693616442, + "grad_norm": 2.2119545936584473, + "learning_rate": 1.4532969878697016e-05, + "loss": 2.4998428344726564, + "step": 100780 + }, + { + "epoch": 0.2743977133523887, + "grad_norm": 2.115386724472046, + "learning_rate": 1.4531879514788061e-05, + "loss": 2.4837223052978517, + "step": 100800 + }, + { + "epoch": 0.2744521573431333, + "grad_norm": 2.0893518924713135, + "learning_rate": 1.4530789150879107e-05, + "loss": 2.506013870239258, + "step": 100820 + }, + { + "epoch": 0.2745066013338778, + "grad_norm": 2.2020211219787598, + "learning_rate": 1.4529698786970153e-05, + "loss": 2.563308334350586, + "step": 100840 + }, + { + "epoch": 0.2745610453246223, + "grad_norm": 2.3245325088500977, + "learning_rate": 1.4528608423061197e-05, + "loss": 2.4945560455322267, + "step": 100860 + }, + { + "epoch": 0.27461548931536683, + "grad_norm": 2.1707613468170166, + "learning_rate": 1.4527518059152243e-05, + "loss": 2.498299980163574, + "step": 100880 + }, + { + "epoch": 0.27466993330611134, + "grad_norm": 2.140866279602051, + "learning_rate": 1.4526427695243289e-05, + "loss": 2.508391761779785, + "step": 100900 + }, + { + "epoch": 0.27472437729685584, + "grad_norm": 2.027219772338867, + "learning_rate": 1.4525337331334335e-05, + "loss": 2.5377517700195313, + "step": 100920 + }, + { + "epoch": 0.2747788212876004, + "grad_norm": 2.1487793922424316, + "learning_rate": 1.4524246967425379e-05, + "loss": 2.5110225677490234, + "step": 100940 + }, + { + "epoch": 0.2748332652783449, + "grad_norm": 2.238130807876587, + "learning_rate": 1.4523156603516424e-05, + "loss": 2.5048599243164062, + "step": 100960 + }, + { + "epoch": 0.2748877092690894, + "grad_norm": 2.2065088748931885, + "learning_rate": 1.452206623960747e-05, + "loss": 2.5019933700561525, + "step": 100980 + }, + { + "epoch": 0.27494215325983395, + "grad_norm": 2.025038003921509, + "learning_rate": 1.4520975875698516e-05, + "loss": 2.479499626159668, + "step": 101000 + }, + { + "epoch": 0.27499659725057846, + "grad_norm": 2.3811299800872803, + "learning_rate": 1.451988551178956e-05, + "loss": 2.4880117416381835, + "step": 101020 + }, + { + "epoch": 0.275051041241323, + "grad_norm": 2.0459377765655518, + "learning_rate": 1.4518795147880606e-05, + "loss": 2.5183305740356445, + "step": 101040 + }, + { + "epoch": 0.2751054852320675, + "grad_norm": 2.093477964401245, + "learning_rate": 1.4517704783971652e-05, + "loss": 2.525734710693359, + "step": 101060 + }, + { + "epoch": 0.275159929222812, + "grad_norm": 2.1636595726013184, + "learning_rate": 1.4516614420062698e-05, + "loss": 2.560188865661621, + "step": 101080 + }, + { + "epoch": 0.2752143732135566, + "grad_norm": 2.1281988620758057, + "learning_rate": 1.4515524056153742e-05, + "loss": 2.485516166687012, + "step": 101100 + }, + { + "epoch": 0.2752688172043011, + "grad_norm": 2.1443560123443604, + "learning_rate": 1.4514433692244788e-05, + "loss": 2.5038679122924803, + "step": 101120 + }, + { + "epoch": 0.2753232611950456, + "grad_norm": 2.1979637145996094, + "learning_rate": 1.4513343328335833e-05, + "loss": 2.502673530578613, + "step": 101140 + }, + { + "epoch": 0.27537770518579013, + "grad_norm": 2.1883318424224854, + "learning_rate": 1.451225296442688e-05, + "loss": 2.5196039199829103, + "step": 101160 + }, + { + "epoch": 0.27543214917653464, + "grad_norm": 2.2299675941467285, + "learning_rate": 1.4511162600517925e-05, + "loss": 2.515241813659668, + "step": 101180 + }, + { + "epoch": 0.27548659316727914, + "grad_norm": 2.172208309173584, + "learning_rate": 1.4510072236608969e-05, + "loss": 2.5161806106567384, + "step": 101200 + }, + { + "epoch": 0.2755410371580237, + "grad_norm": 2.1418237686157227, + "learning_rate": 1.4508981872700015e-05, + "loss": 2.49981575012207, + "step": 101220 + }, + { + "epoch": 0.2755954811487682, + "grad_norm": 2.109773635864258, + "learning_rate": 1.450789150879106e-05, + "loss": 2.5294015884399412, + "step": 101240 + }, + { + "epoch": 0.27564992513951275, + "grad_norm": 2.109886646270752, + "learning_rate": 1.4506801144882105e-05, + "loss": 2.4952922821044923, + "step": 101260 + }, + { + "epoch": 0.27570436913025725, + "grad_norm": 2.025862693786621, + "learning_rate": 1.450571078097315e-05, + "loss": 2.5358890533447265, + "step": 101280 + }, + { + "epoch": 0.27575881312100176, + "grad_norm": 2.12479829788208, + "learning_rate": 1.4504620417064196e-05, + "loss": 2.4961570739746093, + "step": 101300 + }, + { + "epoch": 0.2758132571117463, + "grad_norm": 2.222817897796631, + "learning_rate": 1.450353005315524e-05, + "loss": 2.5226078033447266, + "step": 101320 + }, + { + "epoch": 0.2758677011024908, + "grad_norm": 2.078322410583496, + "learning_rate": 1.4502439689246288e-05, + "loss": 2.50097770690918, + "step": 101340 + }, + { + "epoch": 0.2759221450932353, + "grad_norm": 2.1090290546417236, + "learning_rate": 1.4501349325337332e-05, + "loss": 2.5005895614624025, + "step": 101360 + }, + { + "epoch": 0.2759765890839799, + "grad_norm": 2.272287130355835, + "learning_rate": 1.4500258961428378e-05, + "loss": 2.498476028442383, + "step": 101380 + }, + { + "epoch": 0.2760310330747244, + "grad_norm": 2.203375816345215, + "learning_rate": 1.4499168597519424e-05, + "loss": 2.464717483520508, + "step": 101400 + }, + { + "epoch": 0.2760854770654689, + "grad_norm": 2.0754172801971436, + "learning_rate": 1.4498078233610468e-05, + "loss": 2.5145246505737306, + "step": 101420 + }, + { + "epoch": 0.27613992105621343, + "grad_norm": 2.1360232830047607, + "learning_rate": 1.4496987869701514e-05, + "loss": 2.4972789764404295, + "step": 101440 + }, + { + "epoch": 0.27619436504695793, + "grad_norm": 2.2281100749969482, + "learning_rate": 1.449589750579256e-05, + "loss": 2.4916719436645507, + "step": 101460 + }, + { + "epoch": 0.27624880903770244, + "grad_norm": 2.2103359699249268, + "learning_rate": 1.4494807141883604e-05, + "loss": 2.53443603515625, + "step": 101480 + }, + { + "epoch": 0.276303253028447, + "grad_norm": 2.098904609680176, + "learning_rate": 1.4493716777974651e-05, + "loss": 2.5017528533935547, + "step": 101500 + }, + { + "epoch": 0.2763576970191915, + "grad_norm": 2.1272871494293213, + "learning_rate": 1.4492626414065695e-05, + "loss": 2.483998489379883, + "step": 101520 + }, + { + "epoch": 0.27641214100993605, + "grad_norm": 2.266834020614624, + "learning_rate": 1.4491536050156741e-05, + "loss": 2.4809741973876953, + "step": 101540 + }, + { + "epoch": 0.27646658500068055, + "grad_norm": 2.1426146030426025, + "learning_rate": 1.4490445686247787e-05, + "loss": 2.49041862487793, + "step": 101560 + }, + { + "epoch": 0.27652102899142506, + "grad_norm": 2.1372272968292236, + "learning_rate": 1.4489355322338831e-05, + "loss": 2.4856740951538088, + "step": 101580 + }, + { + "epoch": 0.2765754729821696, + "grad_norm": 2.109548568725586, + "learning_rate": 1.4488264958429877e-05, + "loss": 2.481597328186035, + "step": 101600 + }, + { + "epoch": 0.2766299169729141, + "grad_norm": 2.1763997077941895, + "learning_rate": 1.4487174594520923e-05, + "loss": 2.4841909408569336, + "step": 101620 + }, + { + "epoch": 0.2766843609636586, + "grad_norm": 2.137441873550415, + "learning_rate": 1.4486084230611967e-05, + "loss": 2.445027160644531, + "step": 101640 + }, + { + "epoch": 0.2767388049544032, + "grad_norm": 2.072481632232666, + "learning_rate": 1.4484993866703014e-05, + "loss": 2.4778160095214843, + "step": 101660 + }, + { + "epoch": 0.2767932489451477, + "grad_norm": 2.100557804107666, + "learning_rate": 1.4483903502794058e-05, + "loss": 2.4947280883789062, + "step": 101680 + }, + { + "epoch": 0.2768476929358922, + "grad_norm": 2.203216075897217, + "learning_rate": 1.4482813138885104e-05, + "loss": 2.502336120605469, + "step": 101700 + }, + { + "epoch": 0.27690213692663673, + "grad_norm": 2.1917388439178467, + "learning_rate": 1.448172277497615e-05, + "loss": 2.506228446960449, + "step": 101720 + }, + { + "epoch": 0.27695658091738123, + "grad_norm": 2.0898351669311523, + "learning_rate": 1.4480632411067194e-05, + "loss": 2.506089782714844, + "step": 101740 + }, + { + "epoch": 0.2770110249081258, + "grad_norm": 2.127556800842285, + "learning_rate": 1.447954204715824e-05, + "loss": 2.498282051086426, + "step": 101760 + }, + { + "epoch": 0.2770654688988703, + "grad_norm": 2.2550888061523438, + "learning_rate": 1.4478451683249286e-05, + "loss": 2.5179828643798827, + "step": 101780 + }, + { + "epoch": 0.2771199128896148, + "grad_norm": 2.1110782623291016, + "learning_rate": 1.447736131934033e-05, + "loss": 2.5151769638061525, + "step": 101800 + }, + { + "epoch": 0.27717435688035935, + "grad_norm": 2.1845126152038574, + "learning_rate": 1.4476270955431377e-05, + "loss": 2.5206748962402346, + "step": 101820 + }, + { + "epoch": 0.27722880087110385, + "grad_norm": 2.5205183029174805, + "learning_rate": 1.4475180591522421e-05, + "loss": 2.5087793350219725, + "step": 101840 + }, + { + "epoch": 0.27728324486184835, + "grad_norm": 2.244807243347168, + "learning_rate": 1.4474090227613467e-05, + "loss": 2.497524642944336, + "step": 101860 + }, + { + "epoch": 0.2773376888525929, + "grad_norm": 2.074237108230591, + "learning_rate": 1.4472999863704513e-05, + "loss": 2.5433475494384767, + "step": 101880 + }, + { + "epoch": 0.2773921328433374, + "grad_norm": 2.108161449432373, + "learning_rate": 1.4471909499795557e-05, + "loss": 2.52828369140625, + "step": 101900 + }, + { + "epoch": 0.2774465768340819, + "grad_norm": 2.071347951889038, + "learning_rate": 1.4470819135886603e-05, + "loss": 2.514926528930664, + "step": 101920 + }, + { + "epoch": 0.27750102082482647, + "grad_norm": 2.0830435752868652, + "learning_rate": 1.4469728771977649e-05, + "loss": 2.5360132217407227, + "step": 101940 + }, + { + "epoch": 0.277555464815571, + "grad_norm": 2.198112964630127, + "learning_rate": 1.4468638408068693e-05, + "loss": 2.523976707458496, + "step": 101960 + }, + { + "epoch": 0.2776099088063155, + "grad_norm": 2.070540189743042, + "learning_rate": 1.446754804415974e-05, + "loss": 2.504409408569336, + "step": 101980 + }, + { + "epoch": 0.27766435279706003, + "grad_norm": 2.1845359802246094, + "learning_rate": 1.4466457680250785e-05, + "loss": 2.5131603240966798, + "step": 102000 + }, + { + "epoch": 0.27771879678780453, + "grad_norm": 2.280607223510742, + "learning_rate": 1.446536731634183e-05, + "loss": 2.4874380111694334, + "step": 102020 + }, + { + "epoch": 0.2777732407785491, + "grad_norm": 2.027252435684204, + "learning_rate": 1.4464276952432876e-05, + "loss": 2.4771961212158202, + "step": 102040 + }, + { + "epoch": 0.2778276847692936, + "grad_norm": 2.2236979007720947, + "learning_rate": 1.446318658852392e-05, + "loss": 2.500548553466797, + "step": 102060 + }, + { + "epoch": 0.2778821287600381, + "grad_norm": 2.0831656455993652, + "learning_rate": 1.4462096224614966e-05, + "loss": 2.509940528869629, + "step": 102080 + }, + { + "epoch": 0.27793657275078265, + "grad_norm": 2.2038843631744385, + "learning_rate": 1.4461005860706012e-05, + "loss": 2.5127351760864256, + "step": 102100 + }, + { + "epoch": 0.27799101674152715, + "grad_norm": 2.217829704284668, + "learning_rate": 1.4459915496797056e-05, + "loss": 2.5096561431884767, + "step": 102120 + }, + { + "epoch": 0.27804546073227165, + "grad_norm": 2.118424415588379, + "learning_rate": 1.4458825132888104e-05, + "loss": 2.5283742904663087, + "step": 102140 + }, + { + "epoch": 0.2780999047230162, + "grad_norm": 2.159438371658325, + "learning_rate": 1.4457734768979148e-05, + "loss": 2.5413335800170898, + "step": 102160 + }, + { + "epoch": 0.2781543487137607, + "grad_norm": 2.146301031112671, + "learning_rate": 1.4456644405070193e-05, + "loss": 2.505985450744629, + "step": 102180 + }, + { + "epoch": 0.2782087927045052, + "grad_norm": 2.4513766765594482, + "learning_rate": 1.445555404116124e-05, + "loss": 2.535029983520508, + "step": 102200 + }, + { + "epoch": 0.27826323669524977, + "grad_norm": 2.2267143726348877, + "learning_rate": 1.4454463677252283e-05, + "loss": 2.504252624511719, + "step": 102220 + }, + { + "epoch": 0.2783176806859943, + "grad_norm": 2.2246780395507812, + "learning_rate": 1.445337331334333e-05, + "loss": 2.5114748001098635, + "step": 102240 + }, + { + "epoch": 0.27837212467673883, + "grad_norm": 2.267303466796875, + "learning_rate": 1.4452282949434375e-05, + "loss": 2.4948963165283202, + "step": 102260 + }, + { + "epoch": 0.27842656866748333, + "grad_norm": 2.278331756591797, + "learning_rate": 1.4451192585525419e-05, + "loss": 2.5013065338134766, + "step": 102280 + }, + { + "epoch": 0.27848101265822783, + "grad_norm": 2.035330295562744, + "learning_rate": 1.4450102221616467e-05, + "loss": 2.51724910736084, + "step": 102300 + }, + { + "epoch": 0.2785354566489724, + "grad_norm": 2.3454396724700928, + "learning_rate": 1.444901185770751e-05, + "loss": 2.4842994689941404, + "step": 102320 + }, + { + "epoch": 0.2785899006397169, + "grad_norm": 2.0752201080322266, + "learning_rate": 1.4447921493798557e-05, + "loss": 2.52083683013916, + "step": 102340 + }, + { + "epoch": 0.2786443446304614, + "grad_norm": 2.04815936088562, + "learning_rate": 1.4446831129889602e-05, + "loss": 2.5209617614746094, + "step": 102360 + }, + { + "epoch": 0.27869878862120595, + "grad_norm": 2.496819019317627, + "learning_rate": 1.4445740765980646e-05, + "loss": 2.5256635665893556, + "step": 102380 + }, + { + "epoch": 0.27875323261195045, + "grad_norm": 2.28652286529541, + "learning_rate": 1.4444650402071692e-05, + "loss": 2.46361083984375, + "step": 102400 + }, + { + "epoch": 0.27880767660269495, + "grad_norm": 2.207435131072998, + "learning_rate": 1.4443560038162738e-05, + "loss": 2.518468475341797, + "step": 102420 + }, + { + "epoch": 0.2788621205934395, + "grad_norm": 2.1707518100738525, + "learning_rate": 1.4442469674253782e-05, + "loss": 2.492874336242676, + "step": 102440 + }, + { + "epoch": 0.278916564584184, + "grad_norm": 2.0830941200256348, + "learning_rate": 1.444137931034483e-05, + "loss": 2.526947784423828, + "step": 102460 + }, + { + "epoch": 0.27897100857492857, + "grad_norm": 2.369309186935425, + "learning_rate": 1.4440288946435874e-05, + "loss": 2.4972084045410154, + "step": 102480 + }, + { + "epoch": 0.27902545256567307, + "grad_norm": 2.2325356006622314, + "learning_rate": 1.4439198582526918e-05, + "loss": 2.4933496475219727, + "step": 102500 + }, + { + "epoch": 0.2790798965564176, + "grad_norm": 2.2198615074157715, + "learning_rate": 1.4438108218617965e-05, + "loss": 2.5161569595336912, + "step": 102520 + }, + { + "epoch": 0.27913434054716213, + "grad_norm": 2.113504409790039, + "learning_rate": 1.443701785470901e-05, + "loss": 2.5403440475463865, + "step": 102540 + }, + { + "epoch": 0.27918878453790663, + "grad_norm": 2.184149980545044, + "learning_rate": 1.4435927490800055e-05, + "loss": 2.508933448791504, + "step": 102560 + }, + { + "epoch": 0.27924322852865113, + "grad_norm": 2.025507926940918, + "learning_rate": 1.4434837126891101e-05, + "loss": 2.4712455749511717, + "step": 102580 + }, + { + "epoch": 0.2792976725193957, + "grad_norm": 2.0681400299072266, + "learning_rate": 1.4433746762982145e-05, + "loss": 2.5173612594604493, + "step": 102600 + }, + { + "epoch": 0.2793521165101402, + "grad_norm": 2.118278980255127, + "learning_rate": 1.4432656399073193e-05, + "loss": 2.556613540649414, + "step": 102620 + }, + { + "epoch": 0.2794065605008847, + "grad_norm": 2.248786211013794, + "learning_rate": 1.4431566035164237e-05, + "loss": 2.5065845489501952, + "step": 102640 + }, + { + "epoch": 0.27946100449162925, + "grad_norm": 2.1026015281677246, + "learning_rate": 1.4430475671255281e-05, + "loss": 2.519284248352051, + "step": 102660 + }, + { + "epoch": 0.27951544848237375, + "grad_norm": 2.3268673419952393, + "learning_rate": 1.4429385307346329e-05, + "loss": 2.5301252365112306, + "step": 102680 + }, + { + "epoch": 0.27956989247311825, + "grad_norm": 2.200023651123047, + "learning_rate": 1.4428294943437373e-05, + "loss": 2.500105285644531, + "step": 102700 + }, + { + "epoch": 0.2796243364638628, + "grad_norm": 2.11667799949646, + "learning_rate": 1.4427204579528418e-05, + "loss": 2.528072738647461, + "step": 102720 + }, + { + "epoch": 0.2796787804546073, + "grad_norm": 2.3201744556427, + "learning_rate": 1.4426114215619464e-05, + "loss": 2.524415969848633, + "step": 102740 + }, + { + "epoch": 0.27973322444535187, + "grad_norm": 2.130338430404663, + "learning_rate": 1.4425023851710508e-05, + "loss": 2.503168487548828, + "step": 102760 + }, + { + "epoch": 0.27978766843609637, + "grad_norm": 2.1053526401519775, + "learning_rate": 1.4423933487801556e-05, + "loss": 2.50903377532959, + "step": 102780 + }, + { + "epoch": 0.27984211242684087, + "grad_norm": 2.1587488651275635, + "learning_rate": 1.44228431238926e-05, + "loss": 2.529007148742676, + "step": 102800 + }, + { + "epoch": 0.27989655641758543, + "grad_norm": 2.157268524169922, + "learning_rate": 1.4421752759983644e-05, + "loss": 2.514204216003418, + "step": 102820 + }, + { + "epoch": 0.27995100040832993, + "grad_norm": 2.9768598079681396, + "learning_rate": 1.4420662396074692e-05, + "loss": 2.5018741607666017, + "step": 102840 + }, + { + "epoch": 0.28000544439907443, + "grad_norm": 2.1464526653289795, + "learning_rate": 1.4419572032165736e-05, + "loss": 2.50378475189209, + "step": 102860 + }, + { + "epoch": 0.280059888389819, + "grad_norm": 2.086848735809326, + "learning_rate": 1.4418481668256783e-05, + "loss": 2.525498962402344, + "step": 102880 + }, + { + "epoch": 0.2801143323805635, + "grad_norm": 2.223299264907837, + "learning_rate": 1.4417391304347827e-05, + "loss": 2.471820068359375, + "step": 102900 + }, + { + "epoch": 0.280168776371308, + "grad_norm": 2.149076223373413, + "learning_rate": 1.4416300940438871e-05, + "loss": 2.4953519821166994, + "step": 102920 + }, + { + "epoch": 0.28022322036205255, + "grad_norm": 2.1516997814178467, + "learning_rate": 1.4415210576529919e-05, + "loss": 2.490694999694824, + "step": 102940 + }, + { + "epoch": 0.28027766435279705, + "grad_norm": 2.250979423522949, + "learning_rate": 1.4414120212620963e-05, + "loss": 2.5229894638061525, + "step": 102960 + }, + { + "epoch": 0.2803321083435416, + "grad_norm": 2.1096725463867188, + "learning_rate": 1.4413029848712007e-05, + "loss": 2.460497283935547, + "step": 102980 + }, + { + "epoch": 0.2803865523342861, + "grad_norm": 2.215838670730591, + "learning_rate": 1.4411939484803055e-05, + "loss": 2.50350399017334, + "step": 103000 + }, + { + "epoch": 0.2804409963250306, + "grad_norm": 2.1605260372161865, + "learning_rate": 1.4410849120894099e-05, + "loss": 2.4822265625, + "step": 103020 + }, + { + "epoch": 0.28049544031577517, + "grad_norm": 2.1163265705108643, + "learning_rate": 1.4409758756985146e-05, + "loss": 2.5191858291625975, + "step": 103040 + }, + { + "epoch": 0.28054988430651967, + "grad_norm": 2.2605907917022705, + "learning_rate": 1.440866839307619e-05, + "loss": 2.4781801223754885, + "step": 103060 + }, + { + "epoch": 0.28060432829726417, + "grad_norm": 2.23230242729187, + "learning_rate": 1.4407578029167235e-05, + "loss": 2.515793037414551, + "step": 103080 + }, + { + "epoch": 0.28065877228800873, + "grad_norm": 2.07723069190979, + "learning_rate": 1.4406487665258282e-05, + "loss": 2.5098142623901367, + "step": 103100 + }, + { + "epoch": 0.28071321627875323, + "grad_norm": 2.724456787109375, + "learning_rate": 1.4405397301349326e-05, + "loss": 2.487743949890137, + "step": 103120 + }, + { + "epoch": 0.28076766026949773, + "grad_norm": 2.1757261753082275, + "learning_rate": 1.440430693744037e-05, + "loss": 2.4997596740722656, + "step": 103140 + }, + { + "epoch": 0.2808221042602423, + "grad_norm": 2.069183826446533, + "learning_rate": 1.4403216573531418e-05, + "loss": 2.5163536071777344, + "step": 103160 + }, + { + "epoch": 0.2808765482509868, + "grad_norm": 2.2107691764831543, + "learning_rate": 1.4402126209622462e-05, + "loss": 2.4951007843017576, + "step": 103180 + }, + { + "epoch": 0.2809309922417313, + "grad_norm": 2.027301549911499, + "learning_rate": 1.440103584571351e-05, + "loss": 2.506891441345215, + "step": 103200 + }, + { + "epoch": 0.28098543623247585, + "grad_norm": 2.179893732070923, + "learning_rate": 1.4399945481804554e-05, + "loss": 2.515647315979004, + "step": 103220 + }, + { + "epoch": 0.28103988022322035, + "grad_norm": 2.012820243835449, + "learning_rate": 1.4398855117895598e-05, + "loss": 2.5513893127441407, + "step": 103240 + }, + { + "epoch": 0.2810943242139649, + "grad_norm": 2.1470205783843994, + "learning_rate": 1.4397764753986645e-05, + "loss": 2.5141790390014647, + "step": 103260 + }, + { + "epoch": 0.2811487682047094, + "grad_norm": 2.151641368865967, + "learning_rate": 1.439667439007769e-05, + "loss": 2.5012012481689454, + "step": 103280 + }, + { + "epoch": 0.2812032121954539, + "grad_norm": 2.1183555126190186, + "learning_rate": 1.4395584026168733e-05, + "loss": 2.497914695739746, + "step": 103300 + }, + { + "epoch": 0.28125765618619847, + "grad_norm": 2.129371404647827, + "learning_rate": 1.4394493662259781e-05, + "loss": 2.5012182235717773, + "step": 103320 + }, + { + "epoch": 0.28131210017694297, + "grad_norm": 2.400459051132202, + "learning_rate": 1.4393403298350825e-05, + "loss": 2.48803825378418, + "step": 103340 + }, + { + "epoch": 0.28136654416768747, + "grad_norm": 2.1844780445098877, + "learning_rate": 1.4392312934441872e-05, + "loss": 2.4942743301391603, + "step": 103360 + }, + { + "epoch": 0.28142098815843203, + "grad_norm": 2.1683943271636963, + "learning_rate": 1.4391222570532917e-05, + "loss": 2.538191032409668, + "step": 103380 + }, + { + "epoch": 0.28147543214917653, + "grad_norm": 2.2001726627349854, + "learning_rate": 1.439013220662396e-05, + "loss": 2.5293342590332033, + "step": 103400 + }, + { + "epoch": 0.28152987613992103, + "grad_norm": 2.2783541679382324, + "learning_rate": 1.4389041842715008e-05, + "loss": 2.4458024978637694, + "step": 103420 + }, + { + "epoch": 0.2815843201306656, + "grad_norm": 2.1972670555114746, + "learning_rate": 1.4387951478806052e-05, + "loss": 2.5006099700927735, + "step": 103440 + }, + { + "epoch": 0.2816387641214101, + "grad_norm": 2.1696088314056396, + "learning_rate": 1.4386861114897096e-05, + "loss": 2.490127754211426, + "step": 103460 + }, + { + "epoch": 0.28169320811215465, + "grad_norm": 2.065941572189331, + "learning_rate": 1.4385770750988144e-05, + "loss": 2.510414886474609, + "step": 103480 + }, + { + "epoch": 0.28174765210289915, + "grad_norm": 2.2079696655273438, + "learning_rate": 1.4384680387079188e-05, + "loss": 2.519338607788086, + "step": 103500 + }, + { + "epoch": 0.28180209609364365, + "grad_norm": 2.262939453125, + "learning_rate": 1.4383590023170236e-05, + "loss": 2.525883102416992, + "step": 103520 + }, + { + "epoch": 0.2818565400843882, + "grad_norm": 2.1579408645629883, + "learning_rate": 1.438249965926128e-05, + "loss": 2.5115306854248045, + "step": 103540 + }, + { + "epoch": 0.2819109840751327, + "grad_norm": 2.07820725440979, + "learning_rate": 1.4381409295352324e-05, + "loss": 2.5116899490356444, + "step": 103560 + }, + { + "epoch": 0.2819654280658772, + "grad_norm": 2.3466427326202393, + "learning_rate": 1.4380318931443371e-05, + "loss": 2.485143280029297, + "step": 103580 + }, + { + "epoch": 0.28201987205662177, + "grad_norm": 2.1138274669647217, + "learning_rate": 1.4379228567534415e-05, + "loss": 2.5458850860595703, + "step": 103600 + }, + { + "epoch": 0.28207431604736627, + "grad_norm": 2.106508731842041, + "learning_rate": 1.437813820362546e-05, + "loss": 2.5171857833862306, + "step": 103620 + }, + { + "epoch": 0.28212876003811077, + "grad_norm": 2.1147990226745605, + "learning_rate": 1.4377047839716507e-05, + "loss": 2.4972507476806642, + "step": 103640 + }, + { + "epoch": 0.28218320402885533, + "grad_norm": 2.158798933029175, + "learning_rate": 1.4375957475807551e-05, + "loss": 2.5031806945800783, + "step": 103660 + }, + { + "epoch": 0.28223764801959983, + "grad_norm": 2.240565538406372, + "learning_rate": 1.4374867111898599e-05, + "loss": 2.500028610229492, + "step": 103680 + }, + { + "epoch": 0.28229209201034433, + "grad_norm": 2.5059518814086914, + "learning_rate": 1.4373776747989643e-05, + "loss": 2.512175750732422, + "step": 103700 + }, + { + "epoch": 0.2823465360010889, + "grad_norm": 2.1281614303588867, + "learning_rate": 1.4372686384080687e-05, + "loss": 2.5006723403930664, + "step": 103720 + }, + { + "epoch": 0.2824009799918334, + "grad_norm": 2.086984157562256, + "learning_rate": 1.4371596020171734e-05, + "loss": 2.487268829345703, + "step": 103740 + }, + { + "epoch": 0.28245542398257795, + "grad_norm": 2.163156032562256, + "learning_rate": 1.4370505656262778e-05, + "loss": 2.5338424682617187, + "step": 103760 + }, + { + "epoch": 0.28250986797332245, + "grad_norm": 2.123438596725464, + "learning_rate": 1.4369415292353823e-05, + "loss": 2.5240690231323244, + "step": 103780 + }, + { + "epoch": 0.28256431196406695, + "grad_norm": 2.164760112762451, + "learning_rate": 1.436832492844487e-05, + "loss": 2.4724803924560548, + "step": 103800 + }, + { + "epoch": 0.2826187559548115, + "grad_norm": 2.028644323348999, + "learning_rate": 1.4367234564535914e-05, + "loss": 2.5222286224365233, + "step": 103820 + }, + { + "epoch": 0.282673199945556, + "grad_norm": 2.1126606464385986, + "learning_rate": 1.4366144200626962e-05, + "loss": 2.513032150268555, + "step": 103840 + }, + { + "epoch": 0.2827276439363005, + "grad_norm": 2.168252944946289, + "learning_rate": 1.4365053836718006e-05, + "loss": 2.490066719055176, + "step": 103860 + }, + { + "epoch": 0.28278208792704507, + "grad_norm": 2.1292829513549805, + "learning_rate": 1.436396347280905e-05, + "loss": 2.4969455718994142, + "step": 103880 + }, + { + "epoch": 0.28283653191778957, + "grad_norm": 2.150448799133301, + "learning_rate": 1.4362873108900097e-05, + "loss": 2.4960872650146486, + "step": 103900 + }, + { + "epoch": 0.28289097590853407, + "grad_norm": 2.1616549491882324, + "learning_rate": 1.4361782744991142e-05, + "loss": 2.495427894592285, + "step": 103920 + }, + { + "epoch": 0.2829454198992786, + "grad_norm": 2.124967575073242, + "learning_rate": 1.4360692381082186e-05, + "loss": 2.5253524780273438, + "step": 103940 + }, + { + "epoch": 0.28299986389002313, + "grad_norm": 2.110626459121704, + "learning_rate": 1.4359602017173233e-05, + "loss": 2.492705535888672, + "step": 103960 + }, + { + "epoch": 0.2830543078807677, + "grad_norm": 2.1639888286590576, + "learning_rate": 1.4358511653264277e-05, + "loss": 2.5282947540283205, + "step": 103980 + }, + { + "epoch": 0.2831087518715122, + "grad_norm": 2.164738178253174, + "learning_rate": 1.4357421289355325e-05, + "loss": 2.5015937805175783, + "step": 104000 + }, + { + "epoch": 0.2831631958622567, + "grad_norm": 2.1343994140625, + "learning_rate": 1.4356330925446369e-05, + "loss": 2.4969011306762696, + "step": 104020 + }, + { + "epoch": 0.28321763985300125, + "grad_norm": 2.1978156566619873, + "learning_rate": 1.4355240561537413e-05, + "loss": 2.522835540771484, + "step": 104040 + }, + { + "epoch": 0.28327208384374575, + "grad_norm": 2.120347738265991, + "learning_rate": 1.435415019762846e-05, + "loss": 2.5348262786865234, + "step": 104060 + }, + { + "epoch": 0.28332652783449025, + "grad_norm": 2.2439968585968018, + "learning_rate": 1.4353059833719505e-05, + "loss": 2.4860769271850587, + "step": 104080 + }, + { + "epoch": 0.2833809718252348, + "grad_norm": 2.1958863735198975, + "learning_rate": 1.4351969469810549e-05, + "loss": 2.5135980606079102, + "step": 104100 + }, + { + "epoch": 0.2834354158159793, + "grad_norm": 2.086275339126587, + "learning_rate": 1.4350879105901596e-05, + "loss": 2.514809989929199, + "step": 104120 + }, + { + "epoch": 0.2834898598067238, + "grad_norm": 2.13014817237854, + "learning_rate": 1.434978874199264e-05, + "loss": 2.499584197998047, + "step": 104140 + }, + { + "epoch": 0.28354430379746837, + "grad_norm": 2.3981268405914307, + "learning_rate": 1.4348698378083688e-05, + "loss": 2.4977502822875977, + "step": 104160 + }, + { + "epoch": 0.28359874778821287, + "grad_norm": 2.124537944793701, + "learning_rate": 1.4347608014174732e-05, + "loss": 2.5249870300292967, + "step": 104180 + }, + { + "epoch": 0.2836531917789574, + "grad_norm": 3.219130754470825, + "learning_rate": 1.4346517650265776e-05, + "loss": 2.477892303466797, + "step": 104200 + }, + { + "epoch": 0.2837076357697019, + "grad_norm": 2.153069019317627, + "learning_rate": 1.4345427286356824e-05, + "loss": 2.537877655029297, + "step": 104220 + }, + { + "epoch": 0.28376207976044643, + "grad_norm": 2.1875710487365723, + "learning_rate": 1.4344336922447868e-05, + "loss": 2.512437629699707, + "step": 104240 + }, + { + "epoch": 0.283816523751191, + "grad_norm": 2.119006633758545, + "learning_rate": 1.4343246558538912e-05, + "loss": 2.499353218078613, + "step": 104260 + }, + { + "epoch": 0.2838709677419355, + "grad_norm": 2.174156665802002, + "learning_rate": 1.434215619462996e-05, + "loss": 2.501266288757324, + "step": 104280 + }, + { + "epoch": 0.28392541173268, + "grad_norm": 2.073348045349121, + "learning_rate": 1.4341065830721003e-05, + "loss": 2.479108428955078, + "step": 104300 + }, + { + "epoch": 0.28397985572342455, + "grad_norm": 2.169135093688965, + "learning_rate": 1.4339975466812051e-05, + "loss": 2.514565849304199, + "step": 104320 + }, + { + "epoch": 0.28403429971416905, + "grad_norm": 2.186509370803833, + "learning_rate": 1.4338885102903095e-05, + "loss": 2.4779731750488283, + "step": 104340 + }, + { + "epoch": 0.28408874370491355, + "grad_norm": 2.028609037399292, + "learning_rate": 1.433779473899414e-05, + "loss": 2.5089622497558595, + "step": 104360 + }, + { + "epoch": 0.2841431876956581, + "grad_norm": 2.347841739654541, + "learning_rate": 1.4336704375085187e-05, + "loss": 2.486622619628906, + "step": 104380 + }, + { + "epoch": 0.2841976316864026, + "grad_norm": 2.1674399375915527, + "learning_rate": 1.433561401117623e-05, + "loss": 2.490335464477539, + "step": 104400 + }, + { + "epoch": 0.2842520756771471, + "grad_norm": 2.1005032062530518, + "learning_rate": 1.4334523647267275e-05, + "loss": 2.513710784912109, + "step": 104420 + }, + { + "epoch": 0.28430651966789167, + "grad_norm": 2.1121671199798584, + "learning_rate": 1.4333433283358322e-05, + "loss": 2.5229665756225588, + "step": 104440 + }, + { + "epoch": 0.28436096365863617, + "grad_norm": 2.71250581741333, + "learning_rate": 1.4332342919449367e-05, + "loss": 2.5214595794677734, + "step": 104460 + }, + { + "epoch": 0.2844154076493807, + "grad_norm": 2.7898871898651123, + "learning_rate": 1.4331252555540414e-05, + "loss": 2.484892654418945, + "step": 104480 + }, + { + "epoch": 0.2844698516401252, + "grad_norm": 2.074429750442505, + "learning_rate": 1.4330162191631458e-05, + "loss": 2.5207763671875, + "step": 104500 + }, + { + "epoch": 0.28452429563086973, + "grad_norm": 2.093852996826172, + "learning_rate": 1.4329071827722502e-05, + "loss": 2.508980941772461, + "step": 104520 + }, + { + "epoch": 0.2845787396216143, + "grad_norm": 2.3391501903533936, + "learning_rate": 1.432798146381355e-05, + "loss": 2.4969194412231444, + "step": 104540 + }, + { + "epoch": 0.2846331836123588, + "grad_norm": 2.1681439876556396, + "learning_rate": 1.4326891099904594e-05, + "loss": 2.503584098815918, + "step": 104560 + }, + { + "epoch": 0.2846876276031033, + "grad_norm": 2.194530963897705, + "learning_rate": 1.4325800735995641e-05, + "loss": 2.5287353515625, + "step": 104580 + }, + { + "epoch": 0.28474207159384785, + "grad_norm": 2.1267237663269043, + "learning_rate": 1.4324710372086686e-05, + "loss": 2.5209281921386717, + "step": 104600 + }, + { + "epoch": 0.28479651558459235, + "grad_norm": 2.218616008758545, + "learning_rate": 1.432362000817773e-05, + "loss": 2.525385856628418, + "step": 104620 + }, + { + "epoch": 0.28485095957533685, + "grad_norm": 2.1670444011688232, + "learning_rate": 1.4322529644268777e-05, + "loss": 2.5127031326293947, + "step": 104640 + }, + { + "epoch": 0.2849054035660814, + "grad_norm": 2.0423545837402344, + "learning_rate": 1.4321439280359821e-05, + "loss": 2.489017677307129, + "step": 104660 + }, + { + "epoch": 0.2849598475568259, + "grad_norm": 2.0863454341888428, + "learning_rate": 1.4320348916450865e-05, + "loss": 2.5439781188964843, + "step": 104680 + }, + { + "epoch": 0.28501429154757046, + "grad_norm": 2.180250883102417, + "learning_rate": 1.4319258552541913e-05, + "loss": 2.5019014358520506, + "step": 104700 + }, + { + "epoch": 0.28506873553831497, + "grad_norm": 2.0653727054595947, + "learning_rate": 1.4318168188632957e-05, + "loss": 2.5206110000610353, + "step": 104720 + }, + { + "epoch": 0.28512317952905947, + "grad_norm": 2.31367564201355, + "learning_rate": 1.4317077824724005e-05, + "loss": 2.4909873962402345, + "step": 104740 + }, + { + "epoch": 0.285177623519804, + "grad_norm": 2.152118682861328, + "learning_rate": 1.4315987460815049e-05, + "loss": 2.499037170410156, + "step": 104760 + }, + { + "epoch": 0.2852320675105485, + "grad_norm": 2.0869932174682617, + "learning_rate": 1.4314897096906093e-05, + "loss": 2.4825502395629884, + "step": 104780 + }, + { + "epoch": 0.285286511501293, + "grad_norm": 2.166775703430176, + "learning_rate": 1.431380673299714e-05, + "loss": 2.4973817825317384, + "step": 104800 + }, + { + "epoch": 0.2853409554920376, + "grad_norm": 2.236069440841675, + "learning_rate": 1.4312716369088184e-05, + "loss": 2.5014556884765624, + "step": 104820 + }, + { + "epoch": 0.2853953994827821, + "grad_norm": 2.2129924297332764, + "learning_rate": 1.4311626005179228e-05, + "loss": 2.5216541290283203, + "step": 104840 + }, + { + "epoch": 0.2854498434735266, + "grad_norm": 2.0277364253997803, + "learning_rate": 1.4310535641270276e-05, + "loss": 2.521876907348633, + "step": 104860 + }, + { + "epoch": 0.28550428746427114, + "grad_norm": 2.1848878860473633, + "learning_rate": 1.430944527736132e-05, + "loss": 2.520441436767578, + "step": 104880 + }, + { + "epoch": 0.28555873145501565, + "grad_norm": 2.0674479007720947, + "learning_rate": 1.4308354913452368e-05, + "loss": 2.484822082519531, + "step": 104900 + }, + { + "epoch": 0.28561317544576015, + "grad_norm": 2.1046857833862305, + "learning_rate": 1.4307264549543412e-05, + "loss": 2.480461311340332, + "step": 104920 + }, + { + "epoch": 0.2856676194365047, + "grad_norm": 2.3019683361053467, + "learning_rate": 1.4306174185634456e-05, + "loss": 2.4976402282714845, + "step": 104940 + }, + { + "epoch": 0.2857220634272492, + "grad_norm": 2.161405563354492, + "learning_rate": 1.4305083821725503e-05, + "loss": 2.510872459411621, + "step": 104960 + }, + { + "epoch": 0.28577650741799376, + "grad_norm": 2.2030222415924072, + "learning_rate": 1.4303993457816547e-05, + "loss": 2.5244916915893554, + "step": 104980 + }, + { + "epoch": 0.28583095140873827, + "grad_norm": 2.211040735244751, + "learning_rate": 1.4302903093907592e-05, + "loss": 2.5172134399414063, + "step": 105000 + }, + { + "epoch": 0.28588539539948277, + "grad_norm": 2.227618932723999, + "learning_rate": 1.4301812729998639e-05, + "loss": 2.484000015258789, + "step": 105020 + }, + { + "epoch": 0.2859398393902273, + "grad_norm": 2.2562079429626465, + "learning_rate": 1.4300722366089683e-05, + "loss": 2.506279182434082, + "step": 105040 + }, + { + "epoch": 0.2859942833809718, + "grad_norm": 2.4886038303375244, + "learning_rate": 1.429963200218073e-05, + "loss": 2.5046144485473634, + "step": 105060 + }, + { + "epoch": 0.2860487273717163, + "grad_norm": 2.0634539127349854, + "learning_rate": 1.4298541638271775e-05, + "loss": 2.5007667541503906, + "step": 105080 + }, + { + "epoch": 0.2861031713624609, + "grad_norm": 2.21696138381958, + "learning_rate": 1.4297451274362819e-05, + "loss": 2.5151771545410155, + "step": 105100 + }, + { + "epoch": 0.2861576153532054, + "grad_norm": 2.575212240219116, + "learning_rate": 1.4296360910453866e-05, + "loss": 2.516979789733887, + "step": 105120 + }, + { + "epoch": 0.2862120593439499, + "grad_norm": 2.290692090988159, + "learning_rate": 1.429527054654491e-05, + "loss": 2.5018611907958985, + "step": 105140 + }, + { + "epoch": 0.28626650333469444, + "grad_norm": 2.127445697784424, + "learning_rate": 1.4294180182635955e-05, + "loss": 2.5041748046875, + "step": 105160 + }, + { + "epoch": 0.28632094732543895, + "grad_norm": 2.0647428035736084, + "learning_rate": 1.4293089818727002e-05, + "loss": 2.5228460311889647, + "step": 105180 + }, + { + "epoch": 0.2863753913161835, + "grad_norm": 2.1328015327453613, + "learning_rate": 1.4291999454818046e-05, + "loss": 2.543138122558594, + "step": 105200 + }, + { + "epoch": 0.286429835306928, + "grad_norm": 3.603645086288452, + "learning_rate": 1.4290909090909094e-05, + "loss": 2.5029060363769533, + "step": 105220 + }, + { + "epoch": 0.2864842792976725, + "grad_norm": 2.141800880432129, + "learning_rate": 1.4289818727000138e-05, + "loss": 2.5466640472412108, + "step": 105240 + }, + { + "epoch": 0.28653872328841706, + "grad_norm": 2.101351737976074, + "learning_rate": 1.4288728363091182e-05, + "loss": 2.5139774322509765, + "step": 105260 + }, + { + "epoch": 0.28659316727916156, + "grad_norm": 2.14274263381958, + "learning_rate": 1.428763799918223e-05, + "loss": 2.5045528411865234, + "step": 105280 + }, + { + "epoch": 0.28664761126990607, + "grad_norm": 2.1919987201690674, + "learning_rate": 1.4286547635273274e-05, + "loss": 2.5010921478271486, + "step": 105300 + }, + { + "epoch": 0.2867020552606506, + "grad_norm": 2.735875129699707, + "learning_rate": 1.4285457271364318e-05, + "loss": 2.5324506759643555, + "step": 105320 + }, + { + "epoch": 0.2867564992513951, + "grad_norm": 2.1380577087402344, + "learning_rate": 1.4284366907455365e-05, + "loss": 2.497011184692383, + "step": 105340 + }, + { + "epoch": 0.2868109432421396, + "grad_norm": 2.142319440841675, + "learning_rate": 1.428327654354641e-05, + "loss": 2.4919717788696287, + "step": 105360 + }, + { + "epoch": 0.2868653872328842, + "grad_norm": 2.152379274368286, + "learning_rate": 1.4282186179637457e-05, + "loss": 2.4984806060791014, + "step": 105380 + }, + { + "epoch": 0.2869198312236287, + "grad_norm": 2.165517568588257, + "learning_rate": 1.4281095815728501e-05, + "loss": 2.4887014389038087, + "step": 105400 + }, + { + "epoch": 0.2869742752143732, + "grad_norm": 2.6553614139556885, + "learning_rate": 1.4280005451819545e-05, + "loss": 2.5037761688232423, + "step": 105420 + }, + { + "epoch": 0.28702871920511774, + "grad_norm": 3.5494544506073, + "learning_rate": 1.4278915087910593e-05, + "loss": 2.5361968994140627, + "step": 105440 + }, + { + "epoch": 0.28708316319586225, + "grad_norm": 2.2335853576660156, + "learning_rate": 1.4277824724001637e-05, + "loss": 2.5358165740966796, + "step": 105460 + }, + { + "epoch": 0.2871376071866068, + "grad_norm": 2.9008843898773193, + "learning_rate": 1.427673436009268e-05, + "loss": 2.485246276855469, + "step": 105480 + }, + { + "epoch": 0.2871920511773513, + "grad_norm": 2.4311814308166504, + "learning_rate": 1.4275643996183728e-05, + "loss": 2.512295150756836, + "step": 105500 + }, + { + "epoch": 0.2872464951680958, + "grad_norm": 2.1666977405548096, + "learning_rate": 1.4274553632274772e-05, + "loss": 2.5168596267700196, + "step": 105520 + }, + { + "epoch": 0.28730093915884036, + "grad_norm": 2.038022994995117, + "learning_rate": 1.427346326836582e-05, + "loss": 2.5046623229980467, + "step": 105540 + }, + { + "epoch": 0.28735538314958486, + "grad_norm": 2.2935049533843994, + "learning_rate": 1.4272372904456864e-05, + "loss": 2.4730785369873045, + "step": 105560 + }, + { + "epoch": 0.28740982714032937, + "grad_norm": 2.183444023132324, + "learning_rate": 1.4271282540547908e-05, + "loss": 2.5124160766601564, + "step": 105580 + }, + { + "epoch": 0.2874642711310739, + "grad_norm": 2.085203170776367, + "learning_rate": 1.4270192176638956e-05, + "loss": 2.4935741424560547, + "step": 105600 + }, + { + "epoch": 0.2875187151218184, + "grad_norm": 2.1456401348114014, + "learning_rate": 1.426910181273e-05, + "loss": 2.4781425476074217, + "step": 105620 + }, + { + "epoch": 0.2875731591125629, + "grad_norm": 2.1393253803253174, + "learning_rate": 1.4268011448821044e-05, + "loss": 2.4757143020629884, + "step": 105640 + }, + { + "epoch": 0.2876276031033075, + "grad_norm": 2.1654274463653564, + "learning_rate": 1.4266921084912091e-05, + "loss": 2.507827377319336, + "step": 105660 + }, + { + "epoch": 0.287682047094052, + "grad_norm": 2.900887966156006, + "learning_rate": 1.4265830721003136e-05, + "loss": 2.536417007446289, + "step": 105680 + }, + { + "epoch": 0.28773649108479654, + "grad_norm": 2.2356927394866943, + "learning_rate": 1.4264740357094183e-05, + "loss": 2.547262763977051, + "step": 105700 + }, + { + "epoch": 0.28779093507554104, + "grad_norm": 2.2312815189361572, + "learning_rate": 1.4263649993185227e-05, + "loss": 2.5235143661499024, + "step": 105720 + }, + { + "epoch": 0.28784537906628554, + "grad_norm": 2.2536818981170654, + "learning_rate": 1.4262559629276271e-05, + "loss": 2.4945545196533203, + "step": 105740 + }, + { + "epoch": 0.2878998230570301, + "grad_norm": 2.2140238285064697, + "learning_rate": 1.4261469265367319e-05, + "loss": 2.5239242553710937, + "step": 105760 + }, + { + "epoch": 0.2879542670477746, + "grad_norm": 2.17596435546875, + "learning_rate": 1.4260378901458363e-05, + "loss": 2.529614067077637, + "step": 105780 + }, + { + "epoch": 0.2880087110385191, + "grad_norm": 2.1255240440368652, + "learning_rate": 1.4259288537549407e-05, + "loss": 2.4898120880126955, + "step": 105800 + }, + { + "epoch": 0.28806315502926366, + "grad_norm": 2.6339271068573, + "learning_rate": 1.4258198173640454e-05, + "loss": 2.552052688598633, + "step": 105820 + }, + { + "epoch": 0.28811759902000816, + "grad_norm": 2.0825421810150146, + "learning_rate": 1.4257107809731499e-05, + "loss": 2.513555908203125, + "step": 105840 + }, + { + "epoch": 0.28817204301075267, + "grad_norm": 2.3070826530456543, + "learning_rate": 1.4256017445822546e-05, + "loss": 2.474089813232422, + "step": 105860 + }, + { + "epoch": 0.2882264870014972, + "grad_norm": 2.0504136085510254, + "learning_rate": 1.425492708191359e-05, + "loss": 2.526809883117676, + "step": 105880 + }, + { + "epoch": 0.2882809309922417, + "grad_norm": 2.1246840953826904, + "learning_rate": 1.4253836718004634e-05, + "loss": 2.4967668533325194, + "step": 105900 + }, + { + "epoch": 0.2883353749829862, + "grad_norm": 2.162510633468628, + "learning_rate": 1.4252746354095682e-05, + "loss": 2.475497817993164, + "step": 105920 + }, + { + "epoch": 0.2883898189737308, + "grad_norm": 2.1472980976104736, + "learning_rate": 1.4251655990186726e-05, + "loss": 2.4962770462036135, + "step": 105940 + }, + { + "epoch": 0.2884442629644753, + "grad_norm": 2.106992244720459, + "learning_rate": 1.425056562627777e-05, + "loss": 2.542803955078125, + "step": 105960 + }, + { + "epoch": 0.28849870695521984, + "grad_norm": 2.1041436195373535, + "learning_rate": 1.4249475262368818e-05, + "loss": 2.4881967544555663, + "step": 105980 + }, + { + "epoch": 0.28855315094596434, + "grad_norm": 2.0340383052825928, + "learning_rate": 1.4248384898459862e-05, + "loss": 2.496595764160156, + "step": 106000 + }, + { + "epoch": 0.28860759493670884, + "grad_norm": 2.0586671829223633, + "learning_rate": 1.424729453455091e-05, + "loss": 2.513125991821289, + "step": 106020 + }, + { + "epoch": 0.2886620389274534, + "grad_norm": 2.061199903488159, + "learning_rate": 1.4246204170641953e-05, + "loss": 2.5372474670410154, + "step": 106040 + }, + { + "epoch": 0.2887164829181979, + "grad_norm": 2.238910675048828, + "learning_rate": 1.4245113806732997e-05, + "loss": 2.506912612915039, + "step": 106060 + }, + { + "epoch": 0.2887709269089424, + "grad_norm": 2.08559513092041, + "learning_rate": 1.4244023442824045e-05, + "loss": 2.4672224044799806, + "step": 106080 + }, + { + "epoch": 0.28882537089968696, + "grad_norm": 2.131401538848877, + "learning_rate": 1.4242933078915089e-05, + "loss": 2.5254194259643556, + "step": 106100 + }, + { + "epoch": 0.28887981489043146, + "grad_norm": 2.3558340072631836, + "learning_rate": 1.4241842715006133e-05, + "loss": 2.512032318115234, + "step": 106120 + }, + { + "epoch": 0.28893425888117596, + "grad_norm": 2.0459933280944824, + "learning_rate": 1.424075235109718e-05, + "loss": 2.495841217041016, + "step": 106140 + }, + { + "epoch": 0.2889887028719205, + "grad_norm": 2.110821008682251, + "learning_rate": 1.4239661987188225e-05, + "loss": 2.4812395095825197, + "step": 106160 + }, + { + "epoch": 0.289043146862665, + "grad_norm": 2.148758888244629, + "learning_rate": 1.4238571623279272e-05, + "loss": 2.485076904296875, + "step": 106180 + }, + { + "epoch": 0.2890975908534096, + "grad_norm": 2.088521957397461, + "learning_rate": 1.4237481259370316e-05, + "loss": 2.5196836471557615, + "step": 106200 + }, + { + "epoch": 0.2891520348441541, + "grad_norm": 2.28378963470459, + "learning_rate": 1.423639089546136e-05, + "loss": 2.5034200668334963, + "step": 106220 + }, + { + "epoch": 0.2892064788348986, + "grad_norm": 2.2264442443847656, + "learning_rate": 1.4235300531552408e-05, + "loss": 2.5105144500732424, + "step": 106240 + }, + { + "epoch": 0.28926092282564314, + "grad_norm": 2.2026937007904053, + "learning_rate": 1.4234210167643452e-05, + "loss": 2.546876907348633, + "step": 106260 + }, + { + "epoch": 0.28931536681638764, + "grad_norm": 2.1571624279022217, + "learning_rate": 1.4233119803734498e-05, + "loss": 2.479374885559082, + "step": 106280 + }, + { + "epoch": 0.28936981080713214, + "grad_norm": 2.1431403160095215, + "learning_rate": 1.4232029439825544e-05, + "loss": 2.4814737319946287, + "step": 106300 + }, + { + "epoch": 0.2894242547978767, + "grad_norm": 2.1155385971069336, + "learning_rate": 1.4230939075916588e-05, + "loss": 2.5240104675292967, + "step": 106320 + }, + { + "epoch": 0.2894786987886212, + "grad_norm": 2.1207709312438965, + "learning_rate": 1.4229848712007634e-05, + "loss": 2.5250911712646484, + "step": 106340 + }, + { + "epoch": 0.2895331427793657, + "grad_norm": 2.216200590133667, + "learning_rate": 1.422875834809868e-05, + "loss": 2.4976505279541015, + "step": 106360 + }, + { + "epoch": 0.28958758677011026, + "grad_norm": 2.003957509994507, + "learning_rate": 1.4227667984189724e-05, + "loss": 2.461402702331543, + "step": 106380 + }, + { + "epoch": 0.28964203076085476, + "grad_norm": 2.12910532951355, + "learning_rate": 1.4226577620280771e-05, + "loss": 2.5194046020507814, + "step": 106400 + }, + { + "epoch": 0.2896964747515993, + "grad_norm": 2.2074942588806152, + "learning_rate": 1.4225487256371815e-05, + "loss": 2.495466423034668, + "step": 106420 + }, + { + "epoch": 0.2897509187423438, + "grad_norm": 2.1734437942504883, + "learning_rate": 1.4224396892462861e-05, + "loss": 2.48050479888916, + "step": 106440 + }, + { + "epoch": 0.2898053627330883, + "grad_norm": 2.1491265296936035, + "learning_rate": 1.4223306528553907e-05, + "loss": 2.498724365234375, + "step": 106460 + }, + { + "epoch": 0.2898598067238329, + "grad_norm": 2.1480462551116943, + "learning_rate": 1.4222216164644951e-05, + "loss": 2.495051383972168, + "step": 106480 + }, + { + "epoch": 0.2899142507145774, + "grad_norm": 2.0864267349243164, + "learning_rate": 1.4221125800735997e-05, + "loss": 2.5111438751220705, + "step": 106500 + }, + { + "epoch": 0.2899686947053219, + "grad_norm": 2.109069347381592, + "learning_rate": 1.4220035436827043e-05, + "loss": 2.5087558746337892, + "step": 106520 + }, + { + "epoch": 0.29002313869606644, + "grad_norm": 2.188732624053955, + "learning_rate": 1.4218945072918087e-05, + "loss": 2.5207447052001952, + "step": 106540 + }, + { + "epoch": 0.29007758268681094, + "grad_norm": 2.1963727474212646, + "learning_rate": 1.4217854709009134e-05, + "loss": 2.515645217895508, + "step": 106560 + }, + { + "epoch": 0.29013202667755544, + "grad_norm": 2.1225926876068115, + "learning_rate": 1.4216764345100178e-05, + "loss": 2.4979278564453127, + "step": 106580 + }, + { + "epoch": 0.2901864706683, + "grad_norm": 2.102576971054077, + "learning_rate": 1.4215673981191224e-05, + "loss": 2.4983673095703125, + "step": 106600 + }, + { + "epoch": 0.2902409146590445, + "grad_norm": 2.1045644283294678, + "learning_rate": 1.421458361728227e-05, + "loss": 2.5240671157836916, + "step": 106620 + }, + { + "epoch": 0.290295358649789, + "grad_norm": 2.1541802883148193, + "learning_rate": 1.4213493253373314e-05, + "loss": 2.5103315353393554, + "step": 106640 + }, + { + "epoch": 0.29034980264053356, + "grad_norm": 2.1797194480895996, + "learning_rate": 1.421240288946436e-05, + "loss": 2.4848867416381837, + "step": 106660 + }, + { + "epoch": 0.29040424663127806, + "grad_norm": 2.1312530040740967, + "learning_rate": 1.4211312525555406e-05, + "loss": 2.5064023971557616, + "step": 106680 + }, + { + "epoch": 0.2904586906220226, + "grad_norm": 2.257638931274414, + "learning_rate": 1.421022216164645e-05, + "loss": 2.4706146240234377, + "step": 106700 + }, + { + "epoch": 0.2905131346127671, + "grad_norm": 2.273756265640259, + "learning_rate": 1.4209131797737497e-05, + "loss": 2.507844352722168, + "step": 106720 + }, + { + "epoch": 0.2905675786035116, + "grad_norm": 2.6203789710998535, + "learning_rate": 1.4208041433828541e-05, + "loss": 2.508839988708496, + "step": 106740 + }, + { + "epoch": 0.2906220225942562, + "grad_norm": 2.113325357437134, + "learning_rate": 1.4206951069919587e-05, + "loss": 2.504292678833008, + "step": 106760 + }, + { + "epoch": 0.2906764665850007, + "grad_norm": 2.2695727348327637, + "learning_rate": 1.4205860706010633e-05, + "loss": 2.474247932434082, + "step": 106780 + }, + { + "epoch": 0.2907309105757452, + "grad_norm": 3.5822982788085938, + "learning_rate": 1.4204770342101677e-05, + "loss": 2.4894588470458983, + "step": 106800 + }, + { + "epoch": 0.29078535456648974, + "grad_norm": 2.200371026992798, + "learning_rate": 1.4203679978192723e-05, + "loss": 2.48342342376709, + "step": 106820 + }, + { + "epoch": 0.29083979855723424, + "grad_norm": 2.2426609992980957, + "learning_rate": 1.4202589614283769e-05, + "loss": 2.4667827606201174, + "step": 106840 + }, + { + "epoch": 0.29089424254797874, + "grad_norm": 2.149946451187134, + "learning_rate": 1.4201499250374813e-05, + "loss": 2.4796308517456054, + "step": 106860 + }, + { + "epoch": 0.2909486865387233, + "grad_norm": 2.048546075820923, + "learning_rate": 1.420040888646586e-05, + "loss": 2.5114545822143555, + "step": 106880 + }, + { + "epoch": 0.2910031305294678, + "grad_norm": 2.1805710792541504, + "learning_rate": 1.4199318522556904e-05, + "loss": 2.505299949645996, + "step": 106900 + }, + { + "epoch": 0.29105757452021236, + "grad_norm": 2.17724871635437, + "learning_rate": 1.419822815864795e-05, + "loss": 2.512816619873047, + "step": 106920 + }, + { + "epoch": 0.29111201851095686, + "grad_norm": 2.1219232082366943, + "learning_rate": 1.4197137794738996e-05, + "loss": 2.5072343826293944, + "step": 106940 + }, + { + "epoch": 0.29116646250170136, + "grad_norm": 2.097792148590088, + "learning_rate": 1.419604743083004e-05, + "loss": 2.5293205261230467, + "step": 106960 + }, + { + "epoch": 0.2912209064924459, + "grad_norm": 2.1908812522888184, + "learning_rate": 1.4194957066921086e-05, + "loss": 2.4871307373046876, + "step": 106980 + }, + { + "epoch": 0.2912753504831904, + "grad_norm": 2.14115047454834, + "learning_rate": 1.4193866703012132e-05, + "loss": 2.512119674682617, + "step": 107000 + }, + { + "epoch": 0.2913297944739349, + "grad_norm": 2.363281011581421, + "learning_rate": 1.4192776339103176e-05, + "loss": 2.5111927032470702, + "step": 107020 + }, + { + "epoch": 0.2913842384646795, + "grad_norm": 2.1873881816864014, + "learning_rate": 1.4191685975194223e-05, + "loss": 2.552398681640625, + "step": 107040 + }, + { + "epoch": 0.291438682455424, + "grad_norm": 2.2506847381591797, + "learning_rate": 1.4190595611285268e-05, + "loss": 2.485360527038574, + "step": 107060 + }, + { + "epoch": 0.2914931264461685, + "grad_norm": 2.1384501457214355, + "learning_rate": 1.4189505247376313e-05, + "loss": 2.485521697998047, + "step": 107080 + }, + { + "epoch": 0.29154757043691304, + "grad_norm": 2.1735281944274902, + "learning_rate": 1.418841488346736e-05, + "loss": 2.4838523864746094, + "step": 107100 + }, + { + "epoch": 0.29160201442765754, + "grad_norm": 2.130815267562866, + "learning_rate": 1.4187324519558403e-05, + "loss": 2.5115068435668944, + "step": 107120 + }, + { + "epoch": 0.29165645841840204, + "grad_norm": 2.185173273086548, + "learning_rate": 1.4186234155649449e-05, + "loss": 2.507948112487793, + "step": 107140 + }, + { + "epoch": 0.2917109024091466, + "grad_norm": 2.2450685501098633, + "learning_rate": 1.4185143791740495e-05, + "loss": 2.4992448806762697, + "step": 107160 + }, + { + "epoch": 0.2917653463998911, + "grad_norm": 2.1143767833709717, + "learning_rate": 1.4184053427831539e-05, + "loss": 2.5099328994750976, + "step": 107180 + }, + { + "epoch": 0.29181979039063566, + "grad_norm": 3.5500192642211914, + "learning_rate": 1.4182963063922587e-05, + "loss": 2.4837215423583983, + "step": 107200 + }, + { + "epoch": 0.29187423438138016, + "grad_norm": 2.0287210941314697, + "learning_rate": 1.418187270001363e-05, + "loss": 2.5098058700561525, + "step": 107220 + }, + { + "epoch": 0.29192867837212466, + "grad_norm": 2.3885138034820557, + "learning_rate": 1.4180782336104676e-05, + "loss": 2.502375030517578, + "step": 107240 + }, + { + "epoch": 0.2919831223628692, + "grad_norm": 2.23630952835083, + "learning_rate": 1.4179691972195722e-05, + "loss": 2.4971189498901367, + "step": 107260 + }, + { + "epoch": 0.2920375663536137, + "grad_norm": 2.20406174659729, + "learning_rate": 1.4178601608286766e-05, + "loss": 2.545613098144531, + "step": 107280 + }, + { + "epoch": 0.2920920103443582, + "grad_norm": 2.126577138900757, + "learning_rate": 1.4177511244377812e-05, + "loss": 2.5269887924194334, + "step": 107300 + }, + { + "epoch": 0.2921464543351028, + "grad_norm": 2.346700668334961, + "learning_rate": 1.4176420880468858e-05, + "loss": 2.532230567932129, + "step": 107320 + }, + { + "epoch": 0.2922008983258473, + "grad_norm": 2.10420560836792, + "learning_rate": 1.4175330516559902e-05, + "loss": 2.5081153869628907, + "step": 107340 + }, + { + "epoch": 0.2922553423165918, + "grad_norm": 2.177809476852417, + "learning_rate": 1.417424015265095e-05, + "loss": 2.489972686767578, + "step": 107360 + }, + { + "epoch": 0.29230978630733634, + "grad_norm": 2.097430944442749, + "learning_rate": 1.4173149788741994e-05, + "loss": 2.4984668731689452, + "step": 107380 + }, + { + "epoch": 0.29236423029808084, + "grad_norm": 2.1818113327026367, + "learning_rate": 1.417205942483304e-05, + "loss": 2.478250503540039, + "step": 107400 + }, + { + "epoch": 0.2924186742888254, + "grad_norm": 2.212353229522705, + "learning_rate": 1.4170969060924085e-05, + "loss": 2.480216217041016, + "step": 107420 + }, + { + "epoch": 0.2924731182795699, + "grad_norm": 2.1060969829559326, + "learning_rate": 1.416987869701513e-05, + "loss": 2.5080739974975588, + "step": 107440 + }, + { + "epoch": 0.2925275622703144, + "grad_norm": 2.1978282928466797, + "learning_rate": 1.4168788333106175e-05, + "loss": 2.5257368087768555, + "step": 107460 + }, + { + "epoch": 0.29258200626105896, + "grad_norm": 2.056262731552124, + "learning_rate": 1.4167697969197221e-05, + "loss": 2.486477088928223, + "step": 107480 + }, + { + "epoch": 0.29263645025180346, + "grad_norm": 2.2049601078033447, + "learning_rate": 1.4166607605288265e-05, + "loss": 2.4791143417358397, + "step": 107500 + }, + { + "epoch": 0.29269089424254796, + "grad_norm": 2.1313507556915283, + "learning_rate": 1.4165517241379311e-05, + "loss": 2.505354118347168, + "step": 107520 + }, + { + "epoch": 0.2927453382332925, + "grad_norm": 2.1454434394836426, + "learning_rate": 1.4164426877470357e-05, + "loss": 2.478442573547363, + "step": 107540 + }, + { + "epoch": 0.292799782224037, + "grad_norm": 2.1288552284240723, + "learning_rate": 1.4163336513561403e-05, + "loss": 2.52022647857666, + "step": 107560 + }, + { + "epoch": 0.2928542262147815, + "grad_norm": 2.3615105152130127, + "learning_rate": 1.4162246149652448e-05, + "loss": 2.5094631195068358, + "step": 107580 + }, + { + "epoch": 0.2929086702055261, + "grad_norm": 2.1827430725097656, + "learning_rate": 1.4161155785743493e-05, + "loss": 2.4757577896118166, + "step": 107600 + }, + { + "epoch": 0.2929631141962706, + "grad_norm": 2.070945978164673, + "learning_rate": 1.4160065421834538e-05, + "loss": 2.5242183685302733, + "step": 107620 + }, + { + "epoch": 0.2930175581870151, + "grad_norm": 2.3136308193206787, + "learning_rate": 1.4158975057925584e-05, + "loss": 2.5018049240112306, + "step": 107640 + }, + { + "epoch": 0.29307200217775964, + "grad_norm": 2.178434371948242, + "learning_rate": 1.4157884694016628e-05, + "loss": 2.5199264526367187, + "step": 107660 + }, + { + "epoch": 0.29312644616850414, + "grad_norm": 2.136082649230957, + "learning_rate": 1.4156794330107674e-05, + "loss": 2.5065639495849608, + "step": 107680 + }, + { + "epoch": 0.2931808901592487, + "grad_norm": 2.080240488052368, + "learning_rate": 1.415570396619872e-05, + "loss": 2.5127025604248048, + "step": 107700 + }, + { + "epoch": 0.2932353341499932, + "grad_norm": 2.159236192703247, + "learning_rate": 1.4154613602289766e-05, + "loss": 2.514261817932129, + "step": 107720 + }, + { + "epoch": 0.2932897781407377, + "grad_norm": 2.183943033218384, + "learning_rate": 1.4153523238380812e-05, + "loss": 2.504832458496094, + "step": 107740 + }, + { + "epoch": 0.29334422213148226, + "grad_norm": 2.081944704055786, + "learning_rate": 1.4152432874471856e-05, + "loss": 2.5118614196777345, + "step": 107760 + }, + { + "epoch": 0.29339866612222676, + "grad_norm": 2.170804738998413, + "learning_rate": 1.4151342510562901e-05, + "loss": 2.4899688720703126, + "step": 107780 + }, + { + "epoch": 0.29345311011297126, + "grad_norm": 2.264352798461914, + "learning_rate": 1.4150252146653947e-05, + "loss": 2.469309425354004, + "step": 107800 + }, + { + "epoch": 0.2935075541037158, + "grad_norm": 2.2364144325256348, + "learning_rate": 1.4149161782744991e-05, + "loss": 2.515705680847168, + "step": 107820 + }, + { + "epoch": 0.2935619980944603, + "grad_norm": 2.1075472831726074, + "learning_rate": 1.4148071418836037e-05, + "loss": 2.4949161529541017, + "step": 107840 + }, + { + "epoch": 0.2936164420852048, + "grad_norm": 2.452122211456299, + "learning_rate": 1.4146981054927083e-05, + "loss": 2.5217754364013674, + "step": 107860 + }, + { + "epoch": 0.2936708860759494, + "grad_norm": 2.1070170402526855, + "learning_rate": 1.4145890691018129e-05, + "loss": 2.4870231628417967, + "step": 107880 + }, + { + "epoch": 0.2937253300666939, + "grad_norm": 2.073789119720459, + "learning_rate": 1.4144800327109175e-05, + "loss": 2.500413513183594, + "step": 107900 + }, + { + "epoch": 0.29377977405743844, + "grad_norm": 2.063553810119629, + "learning_rate": 1.4143709963200219e-05, + "loss": 2.5043357849121093, + "step": 107920 + }, + { + "epoch": 0.29383421804818294, + "grad_norm": 2.2291438579559326, + "learning_rate": 1.4142619599291265e-05, + "loss": 2.478589630126953, + "step": 107940 + }, + { + "epoch": 0.29388866203892744, + "grad_norm": 2.2450902462005615, + "learning_rate": 1.414152923538231e-05, + "loss": 2.4941566467285154, + "step": 107960 + }, + { + "epoch": 0.293943106029672, + "grad_norm": 2.063473701477051, + "learning_rate": 1.4140438871473356e-05, + "loss": 2.5058961868286134, + "step": 107980 + }, + { + "epoch": 0.2939975500204165, + "grad_norm": 2.242380380630493, + "learning_rate": 1.41393485075644e-05, + "loss": 2.50009822845459, + "step": 108000 + }, + { + "epoch": 0.294051994011161, + "grad_norm": 2.248302459716797, + "learning_rate": 1.4138258143655446e-05, + "loss": 2.50494327545166, + "step": 108020 + }, + { + "epoch": 0.29410643800190556, + "grad_norm": 2.198889970779419, + "learning_rate": 1.4137167779746492e-05, + "loss": 2.5054147720336912, + "step": 108040 + }, + { + "epoch": 0.29416088199265006, + "grad_norm": 2.132775068283081, + "learning_rate": 1.4136077415837538e-05, + "loss": 2.5130985260009764, + "step": 108060 + }, + { + "epoch": 0.29421532598339456, + "grad_norm": 2.1052067279815674, + "learning_rate": 1.4134987051928582e-05, + "loss": 2.4885704040527346, + "step": 108080 + }, + { + "epoch": 0.2942697699741391, + "grad_norm": 2.1484105587005615, + "learning_rate": 1.4133896688019628e-05, + "loss": 2.5436309814453124, + "step": 108100 + }, + { + "epoch": 0.2943242139648836, + "grad_norm": 2.1949868202209473, + "learning_rate": 1.4132806324110673e-05, + "loss": 2.488066482543945, + "step": 108120 + }, + { + "epoch": 0.2943786579556281, + "grad_norm": 2.108309268951416, + "learning_rate": 1.413171596020172e-05, + "loss": 2.4789220809936525, + "step": 108140 + }, + { + "epoch": 0.2944331019463727, + "grad_norm": 2.1168925762176514, + "learning_rate": 1.4130625596292763e-05, + "loss": 2.4789144515991213, + "step": 108160 + }, + { + "epoch": 0.2944875459371172, + "grad_norm": 2.259354829788208, + "learning_rate": 1.412953523238381e-05, + "loss": 2.4882791519165037, + "step": 108180 + }, + { + "epoch": 0.29454198992786174, + "grad_norm": 2.2643864154815674, + "learning_rate": 1.4128444868474855e-05, + "loss": 2.499311637878418, + "step": 108200 + }, + { + "epoch": 0.29459643391860624, + "grad_norm": 2.3689775466918945, + "learning_rate": 1.41273545045659e-05, + "loss": 2.51336784362793, + "step": 108220 + }, + { + "epoch": 0.29465087790935074, + "grad_norm": 2.1307971477508545, + "learning_rate": 1.4126264140656945e-05, + "loss": 2.478584098815918, + "step": 108240 + }, + { + "epoch": 0.2947053219000953, + "grad_norm": 2.1168878078460693, + "learning_rate": 1.412517377674799e-05, + "loss": 2.497874450683594, + "step": 108260 + }, + { + "epoch": 0.2947597658908398, + "grad_norm": 2.0419161319732666, + "learning_rate": 1.4124083412839037e-05, + "loss": 2.4994176864624023, + "step": 108280 + }, + { + "epoch": 0.2948142098815843, + "grad_norm": 2.2762632369995117, + "learning_rate": 1.4122993048930082e-05, + "loss": 2.4918142318725587, + "step": 108300 + }, + { + "epoch": 0.29486865387232886, + "grad_norm": 2.3398149013519287, + "learning_rate": 1.4121902685021126e-05, + "loss": 2.531114959716797, + "step": 108320 + }, + { + "epoch": 0.29492309786307336, + "grad_norm": 2.0693037509918213, + "learning_rate": 1.4120812321112172e-05, + "loss": 2.505397987365723, + "step": 108340 + }, + { + "epoch": 0.29497754185381786, + "grad_norm": 2.1440813541412354, + "learning_rate": 1.4119721957203218e-05, + "loss": 2.526715850830078, + "step": 108360 + }, + { + "epoch": 0.2950319858445624, + "grad_norm": 2.0471200942993164, + "learning_rate": 1.4118631593294264e-05, + "loss": 2.517678642272949, + "step": 108380 + }, + { + "epoch": 0.2950864298353069, + "grad_norm": 2.1810238361358643, + "learning_rate": 1.4117541229385308e-05, + "loss": 2.5052061080932617, + "step": 108400 + }, + { + "epoch": 0.2951408738260515, + "grad_norm": 2.3062667846679688, + "learning_rate": 1.4116450865476354e-05, + "loss": 2.4991920471191404, + "step": 108420 + }, + { + "epoch": 0.295195317816796, + "grad_norm": 2.740117311477661, + "learning_rate": 1.41153605015674e-05, + "loss": 2.4892839431762694, + "step": 108440 + }, + { + "epoch": 0.2952497618075405, + "grad_norm": 2.138181447982788, + "learning_rate": 1.4114270137658445e-05, + "loss": 2.514135551452637, + "step": 108460 + }, + { + "epoch": 0.29530420579828504, + "grad_norm": 2.1770100593566895, + "learning_rate": 1.411317977374949e-05, + "loss": 2.521273612976074, + "step": 108480 + }, + { + "epoch": 0.29535864978902954, + "grad_norm": 2.0993762016296387, + "learning_rate": 1.4112089409840535e-05, + "loss": 2.522274398803711, + "step": 108500 + }, + { + "epoch": 0.29541309377977404, + "grad_norm": 2.216829299926758, + "learning_rate": 1.4110999045931581e-05, + "loss": 2.4954790115356444, + "step": 108520 + }, + { + "epoch": 0.2954675377705186, + "grad_norm": 2.0700905323028564, + "learning_rate": 1.4109908682022625e-05, + "loss": 2.4904253005981447, + "step": 108540 + }, + { + "epoch": 0.2955219817612631, + "grad_norm": 2.429408311843872, + "learning_rate": 1.4108818318113671e-05, + "loss": 2.4982812881469725, + "step": 108560 + }, + { + "epoch": 0.2955764257520076, + "grad_norm": 2.3132975101470947, + "learning_rate": 1.4107727954204717e-05, + "loss": 2.5271963119506835, + "step": 108580 + }, + { + "epoch": 0.29563086974275216, + "grad_norm": 2.0922441482543945, + "learning_rate": 1.4106637590295763e-05, + "loss": 2.496413803100586, + "step": 108600 + }, + { + "epoch": 0.29568531373349666, + "grad_norm": 2.501600980758667, + "learning_rate": 1.4105547226386808e-05, + "loss": 2.494288444519043, + "step": 108620 + }, + { + "epoch": 0.2957397577242412, + "grad_norm": 2.1680784225463867, + "learning_rate": 1.4104456862477853e-05, + "loss": 2.486194610595703, + "step": 108640 + }, + { + "epoch": 0.2957942017149857, + "grad_norm": 2.3190648555755615, + "learning_rate": 1.4103366498568898e-05, + "loss": 2.5020456314086914, + "step": 108660 + }, + { + "epoch": 0.2958486457057302, + "grad_norm": 2.1162197589874268, + "learning_rate": 1.4102276134659944e-05, + "loss": 2.494325637817383, + "step": 108680 + }, + { + "epoch": 0.2959030896964748, + "grad_norm": 2.143122434616089, + "learning_rate": 1.4101185770750988e-05, + "loss": 2.5156349182128905, + "step": 108700 + }, + { + "epoch": 0.2959575336872193, + "grad_norm": 2.151216506958008, + "learning_rate": 1.4100095406842034e-05, + "loss": 2.5175601959228517, + "step": 108720 + }, + { + "epoch": 0.2960119776779638, + "grad_norm": 2.177722692489624, + "learning_rate": 1.409900504293308e-05, + "loss": 2.4969642639160154, + "step": 108740 + }, + { + "epoch": 0.29606642166870833, + "grad_norm": 2.1542327404022217, + "learning_rate": 1.4097914679024126e-05, + "loss": 2.4859556198120116, + "step": 108760 + }, + { + "epoch": 0.29612086565945284, + "grad_norm": 2.162764310836792, + "learning_rate": 1.4096824315115172e-05, + "loss": 2.5048580169677734, + "step": 108780 + }, + { + "epoch": 0.29617530965019734, + "grad_norm": 2.3143632411956787, + "learning_rate": 1.4095733951206216e-05, + "loss": 2.51352481842041, + "step": 108800 + }, + { + "epoch": 0.2962297536409419, + "grad_norm": 2.072533369064331, + "learning_rate": 1.4094643587297261e-05, + "loss": 2.507416534423828, + "step": 108820 + }, + { + "epoch": 0.2962841976316864, + "grad_norm": 2.120877504348755, + "learning_rate": 1.4093553223388307e-05, + "loss": 2.5108657836914063, + "step": 108840 + }, + { + "epoch": 0.2963386416224309, + "grad_norm": 2.1025099754333496, + "learning_rate": 1.4092462859479351e-05, + "loss": 2.530619812011719, + "step": 108860 + }, + { + "epoch": 0.29639308561317546, + "grad_norm": 2.099942922592163, + "learning_rate": 1.4091372495570397e-05, + "loss": 2.5018959045410156, + "step": 108880 + }, + { + "epoch": 0.29644752960391996, + "grad_norm": 2.2340471744537354, + "learning_rate": 1.4090282131661443e-05, + "loss": 2.5100175857543947, + "step": 108900 + }, + { + "epoch": 0.2965019735946645, + "grad_norm": 2.1207094192504883, + "learning_rate": 1.4089191767752489e-05, + "loss": 2.5301198959350586, + "step": 108920 + }, + { + "epoch": 0.296556417585409, + "grad_norm": 2.219562530517578, + "learning_rate": 1.4088101403843535e-05, + "loss": 2.487824630737305, + "step": 108940 + }, + { + "epoch": 0.2966108615761535, + "grad_norm": 2.1621623039245605, + "learning_rate": 1.4087011039934579e-05, + "loss": 2.5256250381469725, + "step": 108960 + }, + { + "epoch": 0.2966653055668981, + "grad_norm": 2.0769057273864746, + "learning_rate": 1.4085920676025625e-05, + "loss": 2.4709712982177736, + "step": 108980 + }, + { + "epoch": 0.2967197495576426, + "grad_norm": 2.1539552211761475, + "learning_rate": 1.408483031211667e-05, + "loss": 2.518607330322266, + "step": 109000 + }, + { + "epoch": 0.2967741935483871, + "grad_norm": 2.2090749740600586, + "learning_rate": 1.4083739948207715e-05, + "loss": 2.5185501098632814, + "step": 109020 + }, + { + "epoch": 0.29682863753913163, + "grad_norm": 2.1669633388519287, + "learning_rate": 1.408264958429876e-05, + "loss": 2.528852653503418, + "step": 109040 + }, + { + "epoch": 0.29688308152987614, + "grad_norm": 2.1140756607055664, + "learning_rate": 1.4081559220389806e-05, + "loss": 2.520512580871582, + "step": 109060 + }, + { + "epoch": 0.29693752552062064, + "grad_norm": 2.1356308460235596, + "learning_rate": 1.4080468856480852e-05, + "loss": 2.5104982376098635, + "step": 109080 + }, + { + "epoch": 0.2969919695113652, + "grad_norm": 2.0685746669769287, + "learning_rate": 1.4079378492571898e-05, + "loss": 2.513595199584961, + "step": 109100 + }, + { + "epoch": 0.2970464135021097, + "grad_norm": 2.0662190914154053, + "learning_rate": 1.4078288128662942e-05, + "loss": 2.4937095642089844, + "step": 109120 + }, + { + "epoch": 0.29710085749285425, + "grad_norm": 2.1589086055755615, + "learning_rate": 1.4077197764753988e-05, + "loss": 2.52929630279541, + "step": 109140 + }, + { + "epoch": 0.29715530148359875, + "grad_norm": 2.109232187271118, + "learning_rate": 1.4076107400845033e-05, + "loss": 2.492683982849121, + "step": 109160 + }, + { + "epoch": 0.29720974547434326, + "grad_norm": 2.166893720626831, + "learning_rate": 1.4075017036936078e-05, + "loss": 2.4791980743408204, + "step": 109180 + }, + { + "epoch": 0.2972641894650878, + "grad_norm": 2.084287405014038, + "learning_rate": 1.4073926673027123e-05, + "loss": 2.490152931213379, + "step": 109200 + }, + { + "epoch": 0.2973186334558323, + "grad_norm": 2.118809700012207, + "learning_rate": 1.407283630911817e-05, + "loss": 2.5080234527587892, + "step": 109220 + }, + { + "epoch": 0.2973730774465768, + "grad_norm": 2.2101986408233643, + "learning_rate": 1.4071745945209215e-05, + "loss": 2.5031816482543947, + "step": 109240 + }, + { + "epoch": 0.2974275214373214, + "grad_norm": 2.1783080101013184, + "learning_rate": 1.407065558130026e-05, + "loss": 2.5338911056518554, + "step": 109260 + }, + { + "epoch": 0.2974819654280659, + "grad_norm": 2.2446608543395996, + "learning_rate": 1.4069565217391305e-05, + "loss": 2.5253149032592774, + "step": 109280 + }, + { + "epoch": 0.2975364094188104, + "grad_norm": 2.0558013916015625, + "learning_rate": 1.406847485348235e-05, + "loss": 2.5091171264648438, + "step": 109300 + }, + { + "epoch": 0.29759085340955493, + "grad_norm": 2.122753620147705, + "learning_rate": 1.4067384489573397e-05, + "loss": 2.5187305450439452, + "step": 109320 + }, + { + "epoch": 0.29764529740029944, + "grad_norm": 2.1101419925689697, + "learning_rate": 1.406629412566444e-05, + "loss": 2.5147138595581056, + "step": 109340 + }, + { + "epoch": 0.29769974139104394, + "grad_norm": 2.0344297885894775, + "learning_rate": 1.4065203761755486e-05, + "loss": 2.4980478286743164, + "step": 109360 + }, + { + "epoch": 0.2977541853817885, + "grad_norm": 2.087738275527954, + "learning_rate": 1.4064113397846532e-05, + "loss": 2.520947265625, + "step": 109380 + }, + { + "epoch": 0.297808629372533, + "grad_norm": 2.151049852371216, + "learning_rate": 1.4063023033937578e-05, + "loss": 2.5354660034179686, + "step": 109400 + }, + { + "epoch": 0.29786307336327755, + "grad_norm": 2.210477352142334, + "learning_rate": 1.4061932670028624e-05, + "loss": 2.489082145690918, + "step": 109420 + }, + { + "epoch": 0.29791751735402205, + "grad_norm": 2.2232859134674072, + "learning_rate": 1.4060842306119668e-05, + "loss": 2.473220634460449, + "step": 109440 + }, + { + "epoch": 0.29797196134476656, + "grad_norm": 2.085840940475464, + "learning_rate": 1.4059751942210714e-05, + "loss": 2.492173194885254, + "step": 109460 + }, + { + "epoch": 0.2980264053355111, + "grad_norm": 2.130732297897339, + "learning_rate": 1.405866157830176e-05, + "loss": 2.5292301177978516, + "step": 109480 + }, + { + "epoch": 0.2980808493262556, + "grad_norm": 2.0980446338653564, + "learning_rate": 1.4057571214392804e-05, + "loss": 2.5070228576660156, + "step": 109500 + }, + { + "epoch": 0.2981352933170001, + "grad_norm": 2.868110418319702, + "learning_rate": 1.405648085048385e-05, + "loss": 2.505649375915527, + "step": 109520 + }, + { + "epoch": 0.2981897373077447, + "grad_norm": 2.112382411956787, + "learning_rate": 1.4055390486574895e-05, + "loss": 2.4781009674072267, + "step": 109540 + }, + { + "epoch": 0.2982441812984892, + "grad_norm": 2.1136481761932373, + "learning_rate": 1.4054300122665941e-05, + "loss": 2.498651123046875, + "step": 109560 + }, + { + "epoch": 0.2982986252892337, + "grad_norm": 2.165147066116333, + "learning_rate": 1.4053209758756987e-05, + "loss": 2.5179813385009764, + "step": 109580 + }, + { + "epoch": 0.29835306927997823, + "grad_norm": 2.1262564659118652, + "learning_rate": 1.4052119394848031e-05, + "loss": 2.4844451904296876, + "step": 109600 + }, + { + "epoch": 0.29840751327072274, + "grad_norm": 2.0903477668762207, + "learning_rate": 1.4051029030939077e-05, + "loss": 2.5330928802490233, + "step": 109620 + }, + { + "epoch": 0.2984619572614673, + "grad_norm": 2.2273144721984863, + "learning_rate": 1.4049938667030123e-05, + "loss": 2.5063318252563476, + "step": 109640 + }, + { + "epoch": 0.2985164012522118, + "grad_norm": 2.334998369216919, + "learning_rate": 1.4048848303121167e-05, + "loss": 2.5142288208007812, + "step": 109660 + }, + { + "epoch": 0.2985708452429563, + "grad_norm": 2.104832410812378, + "learning_rate": 1.4047757939212214e-05, + "loss": 2.5406377792358397, + "step": 109680 + }, + { + "epoch": 0.29862528923370085, + "grad_norm": 2.2365880012512207, + "learning_rate": 1.4046667575303258e-05, + "loss": 2.5326099395751953, + "step": 109700 + }, + { + "epoch": 0.29867973322444535, + "grad_norm": 2.148078680038452, + "learning_rate": 1.4045577211394303e-05, + "loss": 2.5289690017700197, + "step": 109720 + }, + { + "epoch": 0.29873417721518986, + "grad_norm": 2.030815839767456, + "learning_rate": 1.404448684748535e-05, + "loss": 2.5204463958740235, + "step": 109740 + }, + { + "epoch": 0.2987886212059344, + "grad_norm": 2.1339380741119385, + "learning_rate": 1.4043396483576394e-05, + "loss": 2.5138282775878906, + "step": 109760 + }, + { + "epoch": 0.2988430651966789, + "grad_norm": 2.0245556831359863, + "learning_rate": 1.404230611966744e-05, + "loss": 2.5255111694335937, + "step": 109780 + }, + { + "epoch": 0.2988975091874234, + "grad_norm": 2.1996681690216064, + "learning_rate": 1.4041215755758486e-05, + "loss": 2.498436164855957, + "step": 109800 + }, + { + "epoch": 0.298951953178168, + "grad_norm": 2.1467111110687256, + "learning_rate": 1.404012539184953e-05, + "loss": 2.498061752319336, + "step": 109820 + }, + { + "epoch": 0.2990063971689125, + "grad_norm": 2.1291418075561523, + "learning_rate": 1.4039035027940577e-05, + "loss": 2.504459571838379, + "step": 109840 + }, + { + "epoch": 0.299060841159657, + "grad_norm": 2.1010067462921143, + "learning_rate": 1.4037944664031622e-05, + "loss": 2.5042999267578123, + "step": 109860 + }, + { + "epoch": 0.29911528515040153, + "grad_norm": 2.135669469833374, + "learning_rate": 1.4036854300122666e-05, + "loss": 2.4990161895751952, + "step": 109880 + }, + { + "epoch": 0.29916972914114603, + "grad_norm": 2.1222195625305176, + "learning_rate": 1.4035763936213713e-05, + "loss": 2.5231056213378906, + "step": 109900 + }, + { + "epoch": 0.2992241731318906, + "grad_norm": 2.2156333923339844, + "learning_rate": 1.4034673572304757e-05, + "loss": 2.5458538055419924, + "step": 109920 + }, + { + "epoch": 0.2992786171226351, + "grad_norm": 2.2182466983795166, + "learning_rate": 1.4033583208395803e-05, + "loss": 2.493691635131836, + "step": 109940 + }, + { + "epoch": 0.2993330611133796, + "grad_norm": 2.552485466003418, + "learning_rate": 1.4032492844486849e-05, + "loss": 2.4907432556152345, + "step": 109960 + }, + { + "epoch": 0.29938750510412415, + "grad_norm": 2.0528764724731445, + "learning_rate": 1.4031402480577893e-05, + "loss": 2.4587936401367188, + "step": 109980 + }, + { + "epoch": 0.29944194909486865, + "grad_norm": 2.161949634552002, + "learning_rate": 1.403031211666894e-05, + "loss": 2.4863601684570313, + "step": 110000 + }, + { + "epoch": 0.29949639308561316, + "grad_norm": 2.2604634761810303, + "learning_rate": 1.4029221752759985e-05, + "loss": 2.510969543457031, + "step": 110020 + }, + { + "epoch": 0.2995508370763577, + "grad_norm": 2.2107861042022705, + "learning_rate": 1.4028131388851029e-05, + "loss": 2.505123519897461, + "step": 110040 + }, + { + "epoch": 0.2996052810671022, + "grad_norm": 2.6171247959136963, + "learning_rate": 1.4027041024942076e-05, + "loss": 2.5145654678344727, + "step": 110060 + }, + { + "epoch": 0.2996597250578467, + "grad_norm": 2.084285020828247, + "learning_rate": 1.402595066103312e-05, + "loss": 2.5151092529296877, + "step": 110080 + }, + { + "epoch": 0.2997141690485913, + "grad_norm": 3.0318386554718018, + "learning_rate": 1.4024860297124166e-05, + "loss": 2.522653007507324, + "step": 110100 + }, + { + "epoch": 0.2997686130393358, + "grad_norm": 2.101994752883911, + "learning_rate": 1.4023769933215212e-05, + "loss": 2.483918380737305, + "step": 110120 + }, + { + "epoch": 0.29982305703008033, + "grad_norm": 2.249098062515259, + "learning_rate": 1.4022679569306256e-05, + "loss": 2.5057872772216796, + "step": 110140 + }, + { + "epoch": 0.29987750102082483, + "grad_norm": 2.204695224761963, + "learning_rate": 1.4021589205397304e-05, + "loss": 2.4986480712890624, + "step": 110160 + }, + { + "epoch": 0.29993194501156933, + "grad_norm": 2.2698628902435303, + "learning_rate": 1.4020498841488348e-05, + "loss": 2.499949645996094, + "step": 110180 + }, + { + "epoch": 0.2999863890023139, + "grad_norm": 2.103623867034912, + "learning_rate": 1.4019408477579392e-05, + "loss": 2.5504005432128904, + "step": 110200 + }, + { + "epoch": 0.3000408329930584, + "grad_norm": 2.1073367595672607, + "learning_rate": 1.401831811367044e-05, + "loss": 2.5086997985839843, + "step": 110220 + }, + { + "epoch": 0.3000952769838029, + "grad_norm": 2.2222533226013184, + "learning_rate": 1.4017227749761483e-05, + "loss": 2.4736032485961914, + "step": 110240 + }, + { + "epoch": 0.30014972097454745, + "grad_norm": 2.108858823776245, + "learning_rate": 1.401613738585253e-05, + "loss": 2.490034484863281, + "step": 110260 + }, + { + "epoch": 0.30020416496529195, + "grad_norm": 2.058223009109497, + "learning_rate": 1.4015047021943575e-05, + "loss": 2.523149108886719, + "step": 110280 + }, + { + "epoch": 0.30025860895603645, + "grad_norm": 2.2319812774658203, + "learning_rate": 1.401395665803462e-05, + "loss": 2.521680450439453, + "step": 110300 + }, + { + "epoch": 0.300313052946781, + "grad_norm": 2.1420204639434814, + "learning_rate": 1.4012866294125667e-05, + "loss": 2.487852668762207, + "step": 110320 + }, + { + "epoch": 0.3003674969375255, + "grad_norm": 2.305285692214966, + "learning_rate": 1.401177593021671e-05, + "loss": 2.492740440368652, + "step": 110340 + }, + { + "epoch": 0.30042194092827, + "grad_norm": 2.254912853240967, + "learning_rate": 1.4010685566307755e-05, + "loss": 2.496837043762207, + "step": 110360 + }, + { + "epoch": 0.30047638491901457, + "grad_norm": 2.3078670501708984, + "learning_rate": 1.4009595202398802e-05, + "loss": 2.495756721496582, + "step": 110380 + }, + { + "epoch": 0.3005308289097591, + "grad_norm": 2.0896310806274414, + "learning_rate": 1.4008504838489847e-05, + "loss": 2.507769966125488, + "step": 110400 + }, + { + "epoch": 0.30058527290050363, + "grad_norm": 2.1555817127227783, + "learning_rate": 1.4007414474580892e-05, + "loss": 2.5260047912597656, + "step": 110420 + }, + { + "epoch": 0.30063971689124813, + "grad_norm": 2.1417593955993652, + "learning_rate": 1.4006324110671938e-05, + "loss": 2.500362205505371, + "step": 110440 + }, + { + "epoch": 0.30069416088199263, + "grad_norm": 2.1375927925109863, + "learning_rate": 1.4005233746762982e-05, + "loss": 2.5024816513061525, + "step": 110460 + }, + { + "epoch": 0.3007486048727372, + "grad_norm": 2.1732990741729736, + "learning_rate": 1.400414338285403e-05, + "loss": 2.4968095779418946, + "step": 110480 + }, + { + "epoch": 0.3008030488634817, + "grad_norm": 2.583188056945801, + "learning_rate": 1.4003053018945074e-05, + "loss": 2.5137557983398438, + "step": 110500 + }, + { + "epoch": 0.3008574928542262, + "grad_norm": 2.355466604232788, + "learning_rate": 1.4001962655036118e-05, + "loss": 2.48913459777832, + "step": 110520 + }, + { + "epoch": 0.30091193684497075, + "grad_norm": 2.1211955547332764, + "learning_rate": 1.4000872291127166e-05, + "loss": 2.501936149597168, + "step": 110540 + }, + { + "epoch": 0.30096638083571525, + "grad_norm": 2.242398977279663, + "learning_rate": 1.399978192721821e-05, + "loss": 2.483296775817871, + "step": 110560 + }, + { + "epoch": 0.30102082482645975, + "grad_norm": 2.1802937984466553, + "learning_rate": 1.3998691563309255e-05, + "loss": 2.533920669555664, + "step": 110580 + }, + { + "epoch": 0.3010752688172043, + "grad_norm": 2.1739423274993896, + "learning_rate": 1.3997601199400301e-05, + "loss": 2.553286552429199, + "step": 110600 + }, + { + "epoch": 0.3011297128079488, + "grad_norm": 2.0973079204559326, + "learning_rate": 1.3996510835491345e-05, + "loss": 2.5250682830810547, + "step": 110620 + }, + { + "epoch": 0.30118415679869337, + "grad_norm": 2.167114019393921, + "learning_rate": 1.3995420471582393e-05, + "loss": 2.555782699584961, + "step": 110640 + }, + { + "epoch": 0.30123860078943787, + "grad_norm": 2.1509745121002197, + "learning_rate": 1.3994330107673437e-05, + "loss": 2.521876907348633, + "step": 110660 + }, + { + "epoch": 0.3012930447801824, + "grad_norm": 2.2167677879333496, + "learning_rate": 1.3993239743764481e-05, + "loss": 2.443325424194336, + "step": 110680 + }, + { + "epoch": 0.30134748877092693, + "grad_norm": 2.1845133304595947, + "learning_rate": 1.3992149379855529e-05, + "loss": 2.5084285736083984, + "step": 110700 + }, + { + "epoch": 0.30140193276167143, + "grad_norm": 2.174440622329712, + "learning_rate": 1.3991059015946573e-05, + "loss": 2.507828712463379, + "step": 110720 + }, + { + "epoch": 0.30145637675241593, + "grad_norm": 2.139324188232422, + "learning_rate": 1.3989968652037617e-05, + "loss": 2.5017391204833985, + "step": 110740 + }, + { + "epoch": 0.3015108207431605, + "grad_norm": 2.065258264541626, + "learning_rate": 1.3988878288128664e-05, + "loss": 2.508102226257324, + "step": 110760 + }, + { + "epoch": 0.301565264733905, + "grad_norm": 2.4443869590759277, + "learning_rate": 1.3987787924219708e-05, + "loss": 2.5295787811279298, + "step": 110780 + }, + { + "epoch": 0.3016197087246495, + "grad_norm": 2.1933422088623047, + "learning_rate": 1.3986697560310756e-05, + "loss": 2.4952394485473635, + "step": 110800 + }, + { + "epoch": 0.30167415271539405, + "grad_norm": 2.1472110748291016, + "learning_rate": 1.39856071964018e-05, + "loss": 2.459664726257324, + "step": 110820 + }, + { + "epoch": 0.30172859670613855, + "grad_norm": 2.2176263332366943, + "learning_rate": 1.3984516832492844e-05, + "loss": 2.4863290786743164, + "step": 110840 + }, + { + "epoch": 0.3017830406968831, + "grad_norm": 2.1210038661956787, + "learning_rate": 1.3983426468583892e-05, + "loss": 2.495535469055176, + "step": 110860 + }, + { + "epoch": 0.3018374846876276, + "grad_norm": 2.1141514778137207, + "learning_rate": 1.3982336104674936e-05, + "loss": 2.504180908203125, + "step": 110880 + }, + { + "epoch": 0.3018919286783721, + "grad_norm": 2.23140287399292, + "learning_rate": 1.398124574076598e-05, + "loss": 2.5108470916748047, + "step": 110900 + }, + { + "epoch": 0.30194637266911667, + "grad_norm": 2.178985357284546, + "learning_rate": 1.3980155376857027e-05, + "loss": 2.4736446380615233, + "step": 110920 + }, + { + "epoch": 0.30200081665986117, + "grad_norm": 2.1160500049591064, + "learning_rate": 1.3979065012948072e-05, + "loss": 2.491840362548828, + "step": 110940 + }, + { + "epoch": 0.3020552606506057, + "grad_norm": 2.069453239440918, + "learning_rate": 1.3977974649039119e-05, + "loss": 2.5394781112670897, + "step": 110960 + }, + { + "epoch": 0.30210970464135023, + "grad_norm": 2.130690336227417, + "learning_rate": 1.3976884285130163e-05, + "loss": 2.5023326873779297, + "step": 110980 + }, + { + "epoch": 0.30216414863209473, + "grad_norm": 2.1456139087677, + "learning_rate": 1.3975793921221207e-05, + "loss": 2.5359363555908203, + "step": 111000 + }, + { + "epoch": 0.30221859262283923, + "grad_norm": 2.045478582382202, + "learning_rate": 1.3974703557312255e-05, + "loss": 2.4834005355834963, + "step": 111020 + }, + { + "epoch": 0.3022730366135838, + "grad_norm": 2.1104378700256348, + "learning_rate": 1.3973613193403299e-05, + "loss": 2.5268091201782226, + "step": 111040 + }, + { + "epoch": 0.3023274806043283, + "grad_norm": 2.118581533432007, + "learning_rate": 1.3972522829494343e-05, + "loss": 2.508917045593262, + "step": 111060 + }, + { + "epoch": 0.3023819245950728, + "grad_norm": 2.319145441055298, + "learning_rate": 1.397143246558539e-05, + "loss": 2.4939626693725585, + "step": 111080 + }, + { + "epoch": 0.30243636858581735, + "grad_norm": 2.107879400253296, + "learning_rate": 1.3970342101676435e-05, + "loss": 2.4942752838134767, + "step": 111100 + }, + { + "epoch": 0.30249081257656185, + "grad_norm": 2.1130521297454834, + "learning_rate": 1.3969251737767482e-05, + "loss": 2.512154960632324, + "step": 111120 + }, + { + "epoch": 0.3025452565673064, + "grad_norm": 2.201409339904785, + "learning_rate": 1.3968161373858526e-05, + "loss": 2.508260726928711, + "step": 111140 + }, + { + "epoch": 0.3025997005580509, + "grad_norm": 2.502194881439209, + "learning_rate": 1.396707100994957e-05, + "loss": 2.536717987060547, + "step": 111160 + }, + { + "epoch": 0.3026541445487954, + "grad_norm": 2.0941407680511475, + "learning_rate": 1.3965980646040618e-05, + "loss": 2.473469924926758, + "step": 111180 + }, + { + "epoch": 0.30270858853953997, + "grad_norm": 2.1464290618896484, + "learning_rate": 1.3964890282131662e-05, + "loss": 2.5142765045166016, + "step": 111200 + }, + { + "epoch": 0.30276303253028447, + "grad_norm": 2.073410987854004, + "learning_rate": 1.3963799918222706e-05, + "loss": 2.4781999588012695, + "step": 111220 + }, + { + "epoch": 0.30281747652102897, + "grad_norm": 2.1123886108398438, + "learning_rate": 1.3962709554313754e-05, + "loss": 2.484891891479492, + "step": 111240 + }, + { + "epoch": 0.30287192051177353, + "grad_norm": 2.0860097408294678, + "learning_rate": 1.3961619190404798e-05, + "loss": 2.4960508346557617, + "step": 111260 + }, + { + "epoch": 0.30292636450251803, + "grad_norm": 2.2494423389434814, + "learning_rate": 1.3960528826495845e-05, + "loss": 2.4934598922729494, + "step": 111280 + }, + { + "epoch": 0.30298080849326253, + "grad_norm": 2.2003910541534424, + "learning_rate": 1.395943846258689e-05, + "loss": 2.514560508728027, + "step": 111300 + }, + { + "epoch": 0.3030352524840071, + "grad_norm": 2.115057945251465, + "learning_rate": 1.3958348098677933e-05, + "loss": 2.4985986709594727, + "step": 111320 + }, + { + "epoch": 0.3030896964747516, + "grad_norm": 2.1732189655303955, + "learning_rate": 1.3957257734768981e-05, + "loss": 2.5025510787963867, + "step": 111340 + }, + { + "epoch": 0.30314414046549615, + "grad_norm": 2.1590170860290527, + "learning_rate": 1.3956167370860025e-05, + "loss": 2.5271848678588866, + "step": 111360 + }, + { + "epoch": 0.30319858445624065, + "grad_norm": 2.1354494094848633, + "learning_rate": 1.3955077006951073e-05, + "loss": 2.528974914550781, + "step": 111380 + }, + { + "epoch": 0.30325302844698515, + "grad_norm": 2.171643018722534, + "learning_rate": 1.3953986643042117e-05, + "loss": 2.4916757583618163, + "step": 111400 + }, + { + "epoch": 0.3033074724377297, + "grad_norm": 2.2082598209381104, + "learning_rate": 1.395289627913316e-05, + "loss": 2.512459564208984, + "step": 111420 + }, + { + "epoch": 0.3033619164284742, + "grad_norm": 2.2742724418640137, + "learning_rate": 1.3951805915224208e-05, + "loss": 2.484147834777832, + "step": 111440 + }, + { + "epoch": 0.3034163604192187, + "grad_norm": 2.091320276260376, + "learning_rate": 1.3950715551315252e-05, + "loss": 2.4911830902099608, + "step": 111460 + }, + { + "epoch": 0.30347080440996327, + "grad_norm": 2.0743629932403564, + "learning_rate": 1.3949625187406297e-05, + "loss": 2.4982231140136717, + "step": 111480 + }, + { + "epoch": 0.30352524840070777, + "grad_norm": 2.2096853256225586, + "learning_rate": 1.3948534823497344e-05, + "loss": 2.5055912017822264, + "step": 111500 + }, + { + "epoch": 0.30357969239145227, + "grad_norm": 2.2724523544311523, + "learning_rate": 1.3947444459588388e-05, + "loss": 2.490120697021484, + "step": 111520 + }, + { + "epoch": 0.30363413638219683, + "grad_norm": 2.2046313285827637, + "learning_rate": 1.3946354095679436e-05, + "loss": 2.489314842224121, + "step": 111540 + }, + { + "epoch": 0.30368858037294133, + "grad_norm": 2.1058783531188965, + "learning_rate": 1.394526373177048e-05, + "loss": 2.5148271560668944, + "step": 111560 + }, + { + "epoch": 0.30374302436368583, + "grad_norm": 2.164555788040161, + "learning_rate": 1.3944173367861524e-05, + "loss": 2.495625114440918, + "step": 111580 + }, + { + "epoch": 0.3037974683544304, + "grad_norm": 2.269015073776245, + "learning_rate": 1.3943083003952571e-05, + "loss": 2.5003355026245115, + "step": 111600 + }, + { + "epoch": 0.3038519123451749, + "grad_norm": 2.5567429065704346, + "learning_rate": 1.3941992640043615e-05, + "loss": 2.508005714416504, + "step": 111620 + }, + { + "epoch": 0.30390635633591945, + "grad_norm": 2.1621947288513184, + "learning_rate": 1.394090227613466e-05, + "loss": 2.502910614013672, + "step": 111640 + }, + { + "epoch": 0.30396080032666395, + "grad_norm": 2.1107852458953857, + "learning_rate": 1.3939811912225707e-05, + "loss": 2.511606979370117, + "step": 111660 + }, + { + "epoch": 0.30401524431740845, + "grad_norm": 2.148498296737671, + "learning_rate": 1.3938721548316751e-05, + "loss": 2.5308082580566404, + "step": 111680 + }, + { + "epoch": 0.304069688308153, + "grad_norm": 2.0536999702453613, + "learning_rate": 1.3937631184407799e-05, + "loss": 2.4828203201293944, + "step": 111700 + }, + { + "epoch": 0.3041241322988975, + "grad_norm": 2.122781276702881, + "learning_rate": 1.3936540820498843e-05, + "loss": 2.5068920135498045, + "step": 111720 + }, + { + "epoch": 0.304178576289642, + "grad_norm": 2.2375547885894775, + "learning_rate": 1.3935450456589887e-05, + "loss": 2.5092231750488283, + "step": 111740 + }, + { + "epoch": 0.30423302028038657, + "grad_norm": 2.1590166091918945, + "learning_rate": 1.3934360092680934e-05, + "loss": 2.51317138671875, + "step": 111760 + }, + { + "epoch": 0.30428746427113107, + "grad_norm": 2.274003267288208, + "learning_rate": 1.3933269728771979e-05, + "loss": 2.497697448730469, + "step": 111780 + }, + { + "epoch": 0.30434190826187557, + "grad_norm": 3.54301118850708, + "learning_rate": 1.3932179364863023e-05, + "loss": 2.4891845703125, + "step": 111800 + }, + { + "epoch": 0.30439635225262013, + "grad_norm": 2.1800732612609863, + "learning_rate": 1.393108900095407e-05, + "loss": 2.4988534927368162, + "step": 111820 + }, + { + "epoch": 0.30445079624336463, + "grad_norm": 2.130497455596924, + "learning_rate": 1.3929998637045114e-05, + "loss": 2.4883432388305664, + "step": 111840 + }, + { + "epoch": 0.3045052402341092, + "grad_norm": 2.215376377105713, + "learning_rate": 1.3928908273136162e-05, + "loss": 2.507029914855957, + "step": 111860 + }, + { + "epoch": 0.3045596842248537, + "grad_norm": 2.3174989223480225, + "learning_rate": 1.3927817909227206e-05, + "loss": 2.4958805084228515, + "step": 111880 + }, + { + "epoch": 0.3046141282155982, + "grad_norm": 2.5412373542785645, + "learning_rate": 1.392672754531825e-05, + "loss": 2.492368698120117, + "step": 111900 + }, + { + "epoch": 0.30466857220634275, + "grad_norm": 2.1275177001953125, + "learning_rate": 1.3925637181409298e-05, + "loss": 2.5005573272705077, + "step": 111920 + }, + { + "epoch": 0.30472301619708725, + "grad_norm": 3.070565700531006, + "learning_rate": 1.3924546817500342e-05, + "loss": 2.4969623565673826, + "step": 111940 + }, + { + "epoch": 0.30477746018783175, + "grad_norm": 2.2946934700012207, + "learning_rate": 1.3923456453591386e-05, + "loss": 2.4854427337646485, + "step": 111960 + }, + { + "epoch": 0.3048319041785763, + "grad_norm": 2.1335959434509277, + "learning_rate": 1.3922366089682433e-05, + "loss": 2.5000288009643556, + "step": 111980 + }, + { + "epoch": 0.3048863481693208, + "grad_norm": 2.064734697341919, + "learning_rate": 1.3921275725773477e-05, + "loss": 2.5037303924560548, + "step": 112000 + }, + { + "epoch": 0.3049407921600653, + "grad_norm": 2.1706125736236572, + "learning_rate": 1.3920185361864525e-05, + "loss": 2.509285736083984, + "step": 112020 + }, + { + "epoch": 0.30499523615080987, + "grad_norm": 2.209350347518921, + "learning_rate": 1.3919094997955569e-05, + "loss": 2.505946922302246, + "step": 112040 + }, + { + "epoch": 0.30504968014155437, + "grad_norm": 2.181584596633911, + "learning_rate": 1.3918004634046613e-05, + "loss": 2.4860612869262697, + "step": 112060 + }, + { + "epoch": 0.30510412413229887, + "grad_norm": 2.1816928386688232, + "learning_rate": 1.391691427013766e-05, + "loss": 2.4867563247680664, + "step": 112080 + }, + { + "epoch": 0.3051585681230434, + "grad_norm": 2.059222936630249, + "learning_rate": 1.3915823906228705e-05, + "loss": 2.492021369934082, + "step": 112100 + }, + { + "epoch": 0.30521301211378793, + "grad_norm": 2.204681873321533, + "learning_rate": 1.3914733542319749e-05, + "loss": 2.4796367645263673, + "step": 112120 + }, + { + "epoch": 0.3052674561045325, + "grad_norm": 2.1645686626434326, + "learning_rate": 1.3913643178410796e-05, + "loss": 2.471036911010742, + "step": 112140 + }, + { + "epoch": 0.305321900095277, + "grad_norm": 2.1402323246002197, + "learning_rate": 1.391255281450184e-05, + "loss": 2.469131278991699, + "step": 112160 + }, + { + "epoch": 0.3053763440860215, + "grad_norm": 2.090841293334961, + "learning_rate": 1.3911462450592888e-05, + "loss": 2.4750465393066405, + "step": 112180 + }, + { + "epoch": 0.30543078807676605, + "grad_norm": 2.1072092056274414, + "learning_rate": 1.3910372086683932e-05, + "loss": 2.4838979721069334, + "step": 112200 + }, + { + "epoch": 0.30548523206751055, + "grad_norm": 2.346930742263794, + "learning_rate": 1.3909281722774976e-05, + "loss": 2.4960796356201174, + "step": 112220 + }, + { + "epoch": 0.30553967605825505, + "grad_norm": 2.096489191055298, + "learning_rate": 1.3908191358866024e-05, + "loss": 2.4761648178100586, + "step": 112240 + }, + { + "epoch": 0.3055941200489996, + "grad_norm": 2.21927547454834, + "learning_rate": 1.3907100994957068e-05, + "loss": 2.5279672622680662, + "step": 112260 + }, + { + "epoch": 0.3056485640397441, + "grad_norm": 2.1911818981170654, + "learning_rate": 1.3906010631048112e-05, + "loss": 2.542032241821289, + "step": 112280 + }, + { + "epoch": 0.3057030080304886, + "grad_norm": 2.1556556224823, + "learning_rate": 1.390492026713916e-05, + "loss": 2.4890928268432617, + "step": 112300 + }, + { + "epoch": 0.30575745202123317, + "grad_norm": 2.104588270187378, + "learning_rate": 1.3903829903230204e-05, + "loss": 2.532362365722656, + "step": 112320 + }, + { + "epoch": 0.30581189601197767, + "grad_norm": 2.147900104522705, + "learning_rate": 1.3902739539321251e-05, + "loss": 2.4783233642578124, + "step": 112340 + }, + { + "epoch": 0.3058663400027222, + "grad_norm": 2.1513614654541016, + "learning_rate": 1.3901649175412295e-05, + "loss": 2.5022018432617186, + "step": 112360 + }, + { + "epoch": 0.3059207839934667, + "grad_norm": 2.2792301177978516, + "learning_rate": 1.390055881150334e-05, + "loss": 2.5232690811157226, + "step": 112380 + }, + { + "epoch": 0.30597522798421123, + "grad_norm": 2.1512770652770996, + "learning_rate": 1.3899468447594387e-05, + "loss": 2.5220735549926756, + "step": 112400 + }, + { + "epoch": 0.3060296719749558, + "grad_norm": 2.2475931644439697, + "learning_rate": 1.3898378083685431e-05, + "loss": 2.5041887283325197, + "step": 112420 + }, + { + "epoch": 0.3060841159657003, + "grad_norm": 2.1398186683654785, + "learning_rate": 1.3897287719776475e-05, + "loss": 2.485312080383301, + "step": 112440 + }, + { + "epoch": 0.3061385599564448, + "grad_norm": 2.1996493339538574, + "learning_rate": 1.3896197355867523e-05, + "loss": 2.523476791381836, + "step": 112460 + }, + { + "epoch": 0.30619300394718935, + "grad_norm": 2.0780699253082275, + "learning_rate": 1.3895106991958567e-05, + "loss": 2.5107257843017576, + "step": 112480 + }, + { + "epoch": 0.30624744793793385, + "grad_norm": 2.0908243656158447, + "learning_rate": 1.3894016628049614e-05, + "loss": 2.47827091217041, + "step": 112500 + }, + { + "epoch": 0.30630189192867835, + "grad_norm": 2.225828170776367, + "learning_rate": 1.3892926264140658e-05, + "loss": 2.5234796524047853, + "step": 112520 + }, + { + "epoch": 0.3063563359194229, + "grad_norm": 2.307814359664917, + "learning_rate": 1.3891835900231702e-05, + "loss": 2.479086685180664, + "step": 112540 + }, + { + "epoch": 0.3064107799101674, + "grad_norm": 2.111466646194458, + "learning_rate": 1.389074553632275e-05, + "loss": 2.509042739868164, + "step": 112560 + }, + { + "epoch": 0.3064652239009119, + "grad_norm": 2.2306861877441406, + "learning_rate": 1.3889655172413794e-05, + "loss": 2.514510726928711, + "step": 112580 + }, + { + "epoch": 0.30651966789165647, + "grad_norm": 2.212573766708374, + "learning_rate": 1.3888564808504838e-05, + "loss": 2.54622802734375, + "step": 112600 + }, + { + "epoch": 0.30657411188240097, + "grad_norm": 2.23922061920166, + "learning_rate": 1.3887474444595886e-05, + "loss": 2.499466896057129, + "step": 112620 + }, + { + "epoch": 0.3066285558731455, + "grad_norm": 2.026158094406128, + "learning_rate": 1.388638408068693e-05, + "loss": 2.5018022537231444, + "step": 112640 + }, + { + "epoch": 0.30668299986389, + "grad_norm": 2.136080741882324, + "learning_rate": 1.3885293716777977e-05, + "loss": 2.498231315612793, + "step": 112660 + }, + { + "epoch": 0.30673744385463453, + "grad_norm": 2.178473472595215, + "learning_rate": 1.3884203352869021e-05, + "loss": 2.482050895690918, + "step": 112680 + }, + { + "epoch": 0.3067918878453791, + "grad_norm": 2.091252565383911, + "learning_rate": 1.3883112988960065e-05, + "loss": 2.511245536804199, + "step": 112700 + }, + { + "epoch": 0.3068463318361236, + "grad_norm": 2.147052526473999, + "learning_rate": 1.3882022625051113e-05, + "loss": 2.5045726776123045, + "step": 112720 + }, + { + "epoch": 0.3069007758268681, + "grad_norm": 2.3241043090820312, + "learning_rate": 1.3880932261142157e-05, + "loss": 2.5033361434936525, + "step": 112740 + }, + { + "epoch": 0.30695521981761265, + "grad_norm": 2.3716373443603516, + "learning_rate": 1.3879841897233201e-05, + "loss": 2.504345703125, + "step": 112760 + }, + { + "epoch": 0.30700966380835715, + "grad_norm": 3.770040273666382, + "learning_rate": 1.3878751533324249e-05, + "loss": 2.5198766708374025, + "step": 112780 + }, + { + "epoch": 0.30706410779910165, + "grad_norm": 2.0781970024108887, + "learning_rate": 1.3877661169415293e-05, + "loss": 2.5151500701904297, + "step": 112800 + }, + { + "epoch": 0.3071185517898462, + "grad_norm": 2.1870288848876953, + "learning_rate": 1.387657080550634e-05, + "loss": 2.513905143737793, + "step": 112820 + }, + { + "epoch": 0.3071729957805907, + "grad_norm": 2.245954751968384, + "learning_rate": 1.3875480441597384e-05, + "loss": 2.5088212966918944, + "step": 112840 + }, + { + "epoch": 0.30722743977133526, + "grad_norm": 2.142467737197876, + "learning_rate": 1.3874390077688429e-05, + "loss": 2.4662006378173826, + "step": 112860 + }, + { + "epoch": 0.30728188376207977, + "grad_norm": 2.1307883262634277, + "learning_rate": 1.3873299713779476e-05, + "loss": 2.514019775390625, + "step": 112880 + }, + { + "epoch": 0.30733632775282427, + "grad_norm": 2.174480676651001, + "learning_rate": 1.387220934987052e-05, + "loss": 2.5065681457519533, + "step": 112900 + }, + { + "epoch": 0.3073907717435688, + "grad_norm": 2.0114593505859375, + "learning_rate": 1.3871118985961564e-05, + "loss": 2.518810844421387, + "step": 112920 + }, + { + "epoch": 0.3074452157343133, + "grad_norm": 2.0884628295898438, + "learning_rate": 1.3870028622052612e-05, + "loss": 2.475492477416992, + "step": 112940 + }, + { + "epoch": 0.30749965972505783, + "grad_norm": 2.2090065479278564, + "learning_rate": 1.3868938258143656e-05, + "loss": 2.4992475509643555, + "step": 112960 + }, + { + "epoch": 0.3075541037158024, + "grad_norm": 2.1092638969421387, + "learning_rate": 1.3867847894234703e-05, + "loss": 2.509899711608887, + "step": 112980 + }, + { + "epoch": 0.3076085477065469, + "grad_norm": 2.1545727252960205, + "learning_rate": 1.3866757530325748e-05, + "loss": 2.4931554794311523, + "step": 113000 + }, + { + "epoch": 0.3076629916972914, + "grad_norm": 2.1293087005615234, + "learning_rate": 1.3865667166416792e-05, + "loss": 2.506804847717285, + "step": 113020 + }, + { + "epoch": 0.30771743568803595, + "grad_norm": 2.07574725151062, + "learning_rate": 1.3864576802507839e-05, + "loss": 2.492583465576172, + "step": 113040 + }, + { + "epoch": 0.30777187967878045, + "grad_norm": 2.1798572540283203, + "learning_rate": 1.3863486438598883e-05, + "loss": 2.508957862854004, + "step": 113060 + }, + { + "epoch": 0.307826323669525, + "grad_norm": 2.7793185710906982, + "learning_rate": 1.386239607468993e-05, + "loss": 2.4897876739501954, + "step": 113080 + }, + { + "epoch": 0.3078807676602695, + "grad_norm": 2.062350273132324, + "learning_rate": 1.3861305710780975e-05, + "loss": 2.514078140258789, + "step": 113100 + }, + { + "epoch": 0.307935211651014, + "grad_norm": 2.257416248321533, + "learning_rate": 1.3860215346872019e-05, + "loss": 2.4686227798461915, + "step": 113120 + }, + { + "epoch": 0.30798965564175856, + "grad_norm": 2.1269919872283936, + "learning_rate": 1.3859124982963067e-05, + "loss": 2.5355833053588865, + "step": 113140 + }, + { + "epoch": 0.30804409963250307, + "grad_norm": 2.085801362991333, + "learning_rate": 1.385803461905411e-05, + "loss": 2.4655990600585938, + "step": 113160 + }, + { + "epoch": 0.30809854362324757, + "grad_norm": 2.1998353004455566, + "learning_rate": 1.3856944255145155e-05, + "loss": 2.505429267883301, + "step": 113180 + }, + { + "epoch": 0.3081529876139921, + "grad_norm": 2.1787750720977783, + "learning_rate": 1.3855853891236202e-05, + "loss": 2.48785457611084, + "step": 113200 + }, + { + "epoch": 0.3082074316047366, + "grad_norm": 2.210209608078003, + "learning_rate": 1.3854763527327246e-05, + "loss": 2.491733169555664, + "step": 113220 + }, + { + "epoch": 0.3082618755954811, + "grad_norm": 2.11076283454895, + "learning_rate": 1.3853673163418294e-05, + "loss": 2.481807327270508, + "step": 113240 + }, + { + "epoch": 0.3083163195862257, + "grad_norm": 2.077515125274658, + "learning_rate": 1.3852582799509338e-05, + "loss": 2.4987024307250976, + "step": 113260 + }, + { + "epoch": 0.3083707635769702, + "grad_norm": 2.0559253692626953, + "learning_rate": 1.3851492435600382e-05, + "loss": 2.484400177001953, + "step": 113280 + }, + { + "epoch": 0.3084252075677147, + "grad_norm": 2.2111964225769043, + "learning_rate": 1.385040207169143e-05, + "loss": 2.4872278213500976, + "step": 113300 + }, + { + "epoch": 0.30847965155845924, + "grad_norm": 2.085425615310669, + "learning_rate": 1.3849311707782474e-05, + "loss": 2.516221618652344, + "step": 113320 + }, + { + "epoch": 0.30853409554920375, + "grad_norm": 2.053281784057617, + "learning_rate": 1.3848221343873518e-05, + "loss": 2.5095895767211913, + "step": 113340 + }, + { + "epoch": 0.3085885395399483, + "grad_norm": 2.116875648498535, + "learning_rate": 1.3847130979964565e-05, + "loss": 2.4919830322265626, + "step": 113360 + }, + { + "epoch": 0.3086429835306928, + "grad_norm": 2.1987216472625732, + "learning_rate": 1.384604061605561e-05, + "loss": 2.5203073501586912, + "step": 113380 + }, + { + "epoch": 0.3086974275214373, + "grad_norm": 2.07631254196167, + "learning_rate": 1.3844950252146657e-05, + "loss": 2.503682327270508, + "step": 113400 + }, + { + "epoch": 0.30875187151218186, + "grad_norm": 2.1583569049835205, + "learning_rate": 1.3843859888237701e-05, + "loss": 2.4956262588500975, + "step": 113420 + }, + { + "epoch": 0.30880631550292637, + "grad_norm": 2.4048876762390137, + "learning_rate": 1.3842769524328745e-05, + "loss": 2.5163217544555665, + "step": 113440 + }, + { + "epoch": 0.30886075949367087, + "grad_norm": 2.2295267581939697, + "learning_rate": 1.3841679160419793e-05, + "loss": 2.521977424621582, + "step": 113460 + }, + { + "epoch": 0.3089152034844154, + "grad_norm": 2.100130558013916, + "learning_rate": 1.3840588796510837e-05, + "loss": 2.5115623474121094, + "step": 113480 + }, + { + "epoch": 0.3089696474751599, + "grad_norm": 2.2877237796783447, + "learning_rate": 1.3839498432601881e-05, + "loss": 2.484610939025879, + "step": 113500 + }, + { + "epoch": 0.3090240914659044, + "grad_norm": 2.239051580429077, + "learning_rate": 1.3838408068692928e-05, + "loss": 2.5053253173828125, + "step": 113520 + }, + { + "epoch": 0.309078535456649, + "grad_norm": 2.239800214767456, + "learning_rate": 1.3837317704783973e-05, + "loss": 2.5177804946899416, + "step": 113540 + }, + { + "epoch": 0.3091329794473935, + "grad_norm": 2.8182599544525146, + "learning_rate": 1.3836227340875018e-05, + "loss": 2.5350225448608397, + "step": 113560 + }, + { + "epoch": 0.30918742343813804, + "grad_norm": 2.099135398864746, + "learning_rate": 1.3835136976966064e-05, + "loss": 2.508898162841797, + "step": 113580 + }, + { + "epoch": 0.30924186742888254, + "grad_norm": 2.0601823329925537, + "learning_rate": 1.3834046613057108e-05, + "loss": 2.4859081268310548, + "step": 113600 + }, + { + "epoch": 0.30929631141962705, + "grad_norm": 2.1535301208496094, + "learning_rate": 1.3832956249148156e-05, + "loss": 2.502543067932129, + "step": 113620 + }, + { + "epoch": 0.3093507554103716, + "grad_norm": 2.113828182220459, + "learning_rate": 1.38318658852392e-05, + "loss": 2.4953096389770506, + "step": 113640 + }, + { + "epoch": 0.3094051994011161, + "grad_norm": 2.312788963317871, + "learning_rate": 1.3830775521330244e-05, + "loss": 2.5230886459350588, + "step": 113660 + }, + { + "epoch": 0.3094596433918606, + "grad_norm": 2.8685569763183594, + "learning_rate": 1.3829685157421291e-05, + "loss": 2.50040225982666, + "step": 113680 + }, + { + "epoch": 0.30951408738260516, + "grad_norm": 2.079092502593994, + "learning_rate": 1.3828594793512336e-05, + "loss": 2.4931396484375, + "step": 113700 + }, + { + "epoch": 0.30956853137334966, + "grad_norm": 2.5927696228027344, + "learning_rate": 1.3827504429603381e-05, + "loss": 2.482376289367676, + "step": 113720 + }, + { + "epoch": 0.30962297536409417, + "grad_norm": 2.0966062545776367, + "learning_rate": 1.3826414065694427e-05, + "loss": 2.496407890319824, + "step": 113740 + }, + { + "epoch": 0.3096774193548387, + "grad_norm": 2.2565805912017822, + "learning_rate": 1.3825323701785471e-05, + "loss": 2.512866401672363, + "step": 113760 + }, + { + "epoch": 0.3097318633455832, + "grad_norm": 2.680063486099243, + "learning_rate": 1.3824233337876519e-05, + "loss": 2.5240400314331053, + "step": 113780 + }, + { + "epoch": 0.3097863073363277, + "grad_norm": 2.2842631340026855, + "learning_rate": 1.3823142973967563e-05, + "loss": 2.539689636230469, + "step": 113800 + }, + { + "epoch": 0.3098407513270723, + "grad_norm": 2.117231845855713, + "learning_rate": 1.3822052610058607e-05, + "loss": 2.517568016052246, + "step": 113820 + }, + { + "epoch": 0.3098951953178168, + "grad_norm": 2.140674114227295, + "learning_rate": 1.3820962246149655e-05, + "loss": 2.5533472061157227, + "step": 113840 + }, + { + "epoch": 0.30994963930856134, + "grad_norm": 2.122422218322754, + "learning_rate": 1.3819871882240699e-05, + "loss": 2.5015148162841796, + "step": 113860 + }, + { + "epoch": 0.31000408329930584, + "grad_norm": 2.575655221939087, + "learning_rate": 1.3818781518331745e-05, + "loss": 2.5257654190063477, + "step": 113880 + }, + { + "epoch": 0.31005852729005035, + "grad_norm": 2.2028660774230957, + "learning_rate": 1.381769115442279e-05, + "loss": 2.4889854431152343, + "step": 113900 + }, + { + "epoch": 0.3101129712807949, + "grad_norm": 2.095369577407837, + "learning_rate": 1.3816600790513834e-05, + "loss": 2.5069265365600586, + "step": 113920 + }, + { + "epoch": 0.3101674152715394, + "grad_norm": 2.2407586574554443, + "learning_rate": 1.3815510426604882e-05, + "loss": 2.488093376159668, + "step": 113940 + }, + { + "epoch": 0.3102218592622839, + "grad_norm": 2.2057435512542725, + "learning_rate": 1.3814420062695926e-05, + "loss": 2.5147838592529297, + "step": 113960 + }, + { + "epoch": 0.31027630325302846, + "grad_norm": 2.2177817821502686, + "learning_rate": 1.381332969878697e-05, + "loss": 2.4967205047607424, + "step": 113980 + }, + { + "epoch": 0.31033074724377296, + "grad_norm": 2.1534645557403564, + "learning_rate": 1.3812239334878018e-05, + "loss": 2.499027633666992, + "step": 114000 + }, + { + "epoch": 0.31038519123451747, + "grad_norm": 2.0990631580352783, + "learning_rate": 1.3811148970969062e-05, + "loss": 2.4631797790527346, + "step": 114020 + }, + { + "epoch": 0.310439635225262, + "grad_norm": 2.105916738510132, + "learning_rate": 1.3810058607060108e-05, + "loss": 2.5041446685791016, + "step": 114040 + }, + { + "epoch": 0.3104940792160065, + "grad_norm": 2.1724228858947754, + "learning_rate": 1.3808968243151153e-05, + "loss": 2.4916446685791014, + "step": 114060 + }, + { + "epoch": 0.3105485232067511, + "grad_norm": 2.1340808868408203, + "learning_rate": 1.3807877879242198e-05, + "loss": 2.508651542663574, + "step": 114080 + }, + { + "epoch": 0.3106029671974956, + "grad_norm": 2.054638385772705, + "learning_rate": 1.3806787515333245e-05, + "loss": 2.521550941467285, + "step": 114100 + }, + { + "epoch": 0.3106574111882401, + "grad_norm": 2.1379523277282715, + "learning_rate": 1.3805697151424289e-05, + "loss": 2.509255218505859, + "step": 114120 + }, + { + "epoch": 0.31071185517898464, + "grad_norm": 2.175784111022949, + "learning_rate": 1.3804606787515333e-05, + "loss": 2.5018558502197266, + "step": 114140 + }, + { + "epoch": 0.31076629916972914, + "grad_norm": 2.040205955505371, + "learning_rate": 1.380351642360638e-05, + "loss": 2.5157855987548827, + "step": 114160 + }, + { + "epoch": 0.31082074316047364, + "grad_norm": 2.1720051765441895, + "learning_rate": 1.3802426059697425e-05, + "loss": 2.509522819519043, + "step": 114180 + }, + { + "epoch": 0.3108751871512182, + "grad_norm": 2.1842448711395264, + "learning_rate": 1.380133569578847e-05, + "loss": 2.511373519897461, + "step": 114200 + }, + { + "epoch": 0.3109296311419627, + "grad_norm": 1.9922086000442505, + "learning_rate": 1.3800245331879516e-05, + "loss": 2.4963871002197267, + "step": 114220 + }, + { + "epoch": 0.3109840751327072, + "grad_norm": 2.1986052989959717, + "learning_rate": 1.379915496797056e-05, + "loss": 2.524260139465332, + "step": 114240 + }, + { + "epoch": 0.31103851912345176, + "grad_norm": 2.2014880180358887, + "learning_rate": 1.3798064604061608e-05, + "loss": 2.501605987548828, + "step": 114260 + }, + { + "epoch": 0.31109296311419626, + "grad_norm": 2.1033599376678467, + "learning_rate": 1.3796974240152652e-05, + "loss": 2.500850296020508, + "step": 114280 + }, + { + "epoch": 0.31114740710494077, + "grad_norm": 2.158093214035034, + "learning_rate": 1.3795883876243696e-05, + "loss": 2.4961729049682617, + "step": 114300 + }, + { + "epoch": 0.3112018510956853, + "grad_norm": 2.1613142490386963, + "learning_rate": 1.3794793512334744e-05, + "loss": 2.532749366760254, + "step": 114320 + }, + { + "epoch": 0.3112562950864298, + "grad_norm": 2.6177804470062256, + "learning_rate": 1.3793703148425788e-05, + "loss": 2.493848991394043, + "step": 114340 + }, + { + "epoch": 0.3113107390771744, + "grad_norm": 2.2401039600372314, + "learning_rate": 1.3792612784516834e-05, + "loss": 2.5283786773681642, + "step": 114360 + }, + { + "epoch": 0.3113651830679189, + "grad_norm": 2.156094551086426, + "learning_rate": 1.379152242060788e-05, + "loss": 2.4973093032836915, + "step": 114380 + }, + { + "epoch": 0.3114196270586634, + "grad_norm": 2.0955121517181396, + "learning_rate": 1.3790432056698924e-05, + "loss": 2.481789398193359, + "step": 114400 + }, + { + "epoch": 0.31147407104940794, + "grad_norm": 2.105229377746582, + "learning_rate": 1.3789341692789971e-05, + "loss": 2.5001453399658202, + "step": 114420 + }, + { + "epoch": 0.31152851504015244, + "grad_norm": 2.099623203277588, + "learning_rate": 1.3788251328881015e-05, + "loss": 2.509937286376953, + "step": 114440 + }, + { + "epoch": 0.31158295903089694, + "grad_norm": 2.2546768188476562, + "learning_rate": 1.378716096497206e-05, + "loss": 2.49172306060791, + "step": 114460 + }, + { + "epoch": 0.3116374030216415, + "grad_norm": 2.0385525226593018, + "learning_rate": 1.3786070601063107e-05, + "loss": 2.5298370361328124, + "step": 114480 + }, + { + "epoch": 0.311691847012386, + "grad_norm": 2.1481640338897705, + "learning_rate": 1.3784980237154151e-05, + "loss": 2.4878261566162108, + "step": 114500 + }, + { + "epoch": 0.3117462910031305, + "grad_norm": 2.1740002632141113, + "learning_rate": 1.3783889873245197e-05, + "loss": 2.478916358947754, + "step": 114520 + }, + { + "epoch": 0.31180073499387506, + "grad_norm": 2.3178396224975586, + "learning_rate": 1.3782799509336243e-05, + "loss": 2.5387746810913088, + "step": 114540 + }, + { + "epoch": 0.31185517898461956, + "grad_norm": 2.087580919265747, + "learning_rate": 1.3781709145427287e-05, + "loss": 2.4721418380737306, + "step": 114560 + }, + { + "epoch": 0.3119096229753641, + "grad_norm": 2.1606624126434326, + "learning_rate": 1.3780618781518334e-05, + "loss": 2.51307373046875, + "step": 114580 + }, + { + "epoch": 0.3119640669661086, + "grad_norm": 2.163937568664551, + "learning_rate": 1.3779528417609378e-05, + "loss": 2.5103124618530273, + "step": 114600 + }, + { + "epoch": 0.3120185109568531, + "grad_norm": 2.102259397506714, + "learning_rate": 1.3778438053700423e-05, + "loss": 2.5229358673095703, + "step": 114620 + }, + { + "epoch": 0.3120729549475977, + "grad_norm": 2.216141939163208, + "learning_rate": 1.377734768979147e-05, + "loss": 2.4913446426391603, + "step": 114640 + }, + { + "epoch": 0.3121273989383422, + "grad_norm": 2.041494607925415, + "learning_rate": 1.3776257325882514e-05, + "loss": 2.51195068359375, + "step": 114660 + }, + { + "epoch": 0.3121818429290867, + "grad_norm": 2.24135684967041, + "learning_rate": 1.377516696197356e-05, + "loss": 2.511521339416504, + "step": 114680 + }, + { + "epoch": 0.31223628691983124, + "grad_norm": 2.1292428970336914, + "learning_rate": 1.3774076598064606e-05, + "loss": 2.538047218322754, + "step": 114700 + }, + { + "epoch": 0.31229073091057574, + "grad_norm": 2.1885194778442383, + "learning_rate": 1.377298623415565e-05, + "loss": 2.50368709564209, + "step": 114720 + }, + { + "epoch": 0.31234517490132024, + "grad_norm": 2.2053043842315674, + "learning_rate": 1.3771895870246696e-05, + "loss": 2.474806785583496, + "step": 114740 + }, + { + "epoch": 0.3123996188920648, + "grad_norm": 2.110393524169922, + "learning_rate": 1.3770805506337741e-05, + "loss": 2.4766706466674804, + "step": 114760 + }, + { + "epoch": 0.3124540628828093, + "grad_norm": 2.2864015102386475, + "learning_rate": 1.3769715142428787e-05, + "loss": 2.527064895629883, + "step": 114780 + }, + { + "epoch": 0.3125085068735538, + "grad_norm": 2.1641530990600586, + "learning_rate": 1.3768624778519833e-05, + "loss": 2.526238441467285, + "step": 114800 + }, + { + "epoch": 0.31256295086429836, + "grad_norm": 2.0795609951019287, + "learning_rate": 1.3767534414610877e-05, + "loss": 2.5035259246826174, + "step": 114820 + }, + { + "epoch": 0.31261739485504286, + "grad_norm": 2.1050477027893066, + "learning_rate": 1.3766444050701923e-05, + "loss": 2.502745246887207, + "step": 114840 + }, + { + "epoch": 0.3126718388457874, + "grad_norm": 2.2057816982269287, + "learning_rate": 1.3765353686792969e-05, + "loss": 2.5092864990234376, + "step": 114860 + }, + { + "epoch": 0.3127262828365319, + "grad_norm": 2.1028621196746826, + "learning_rate": 1.3764263322884013e-05, + "loss": 2.5104541778564453, + "step": 114880 + }, + { + "epoch": 0.3127807268272764, + "grad_norm": 3.1031975746154785, + "learning_rate": 1.3763172958975059e-05, + "loss": 2.5114501953125, + "step": 114900 + }, + { + "epoch": 0.312835170818021, + "grad_norm": 2.0938029289245605, + "learning_rate": 1.3762082595066105e-05, + "loss": 2.500048828125, + "step": 114920 + }, + { + "epoch": 0.3128896148087655, + "grad_norm": 2.0809173583984375, + "learning_rate": 1.376099223115715e-05, + "loss": 2.5146379470825195, + "step": 114940 + }, + { + "epoch": 0.31294405879951, + "grad_norm": 2.0946309566497803, + "learning_rate": 1.3759901867248196e-05, + "loss": 2.5091068267822267, + "step": 114960 + }, + { + "epoch": 0.31299850279025454, + "grad_norm": 2.118215560913086, + "learning_rate": 1.375881150333924e-05, + "loss": 2.4975889205932615, + "step": 114980 + }, + { + "epoch": 0.31305294678099904, + "grad_norm": 2.092942237854004, + "learning_rate": 1.3757721139430286e-05, + "loss": 2.529372215270996, + "step": 115000 + }, + { + "epoch": 0.31310739077174354, + "grad_norm": 2.0208001136779785, + "learning_rate": 1.3756630775521332e-05, + "loss": 2.5177459716796875, + "step": 115020 + }, + { + "epoch": 0.3131618347624881, + "grad_norm": 2.2803525924682617, + "learning_rate": 1.3755540411612376e-05, + "loss": 2.5028799057006834, + "step": 115040 + }, + { + "epoch": 0.3132162787532326, + "grad_norm": 2.145817995071411, + "learning_rate": 1.3754450047703422e-05, + "loss": 2.4829797744750977, + "step": 115060 + }, + { + "epoch": 0.31327072274397716, + "grad_norm": 2.092618703842163, + "learning_rate": 1.3753359683794468e-05, + "loss": 2.479542350769043, + "step": 115080 + }, + { + "epoch": 0.31332516673472166, + "grad_norm": 2.0821731090545654, + "learning_rate": 1.3752269319885513e-05, + "loss": 2.5221181869506837, + "step": 115100 + }, + { + "epoch": 0.31337961072546616, + "grad_norm": 2.2055299282073975, + "learning_rate": 1.375117895597656e-05, + "loss": 2.488081932067871, + "step": 115120 + }, + { + "epoch": 0.3134340547162107, + "grad_norm": 2.2051432132720947, + "learning_rate": 1.3750088592067603e-05, + "loss": 2.4418827056884767, + "step": 115140 + }, + { + "epoch": 0.3134884987069552, + "grad_norm": 2.1764252185821533, + "learning_rate": 1.374899822815865e-05, + "loss": 2.5035919189453124, + "step": 115160 + }, + { + "epoch": 0.3135429426976997, + "grad_norm": 2.2822253704071045, + "learning_rate": 1.3747907864249695e-05, + "loss": 2.494909477233887, + "step": 115180 + }, + { + "epoch": 0.3135973866884443, + "grad_norm": 2.031822681427002, + "learning_rate": 1.3746817500340739e-05, + "loss": 2.486175537109375, + "step": 115200 + }, + { + "epoch": 0.3136518306791888, + "grad_norm": 2.038198947906494, + "learning_rate": 1.3745727136431785e-05, + "loss": 2.4726486206054688, + "step": 115220 + }, + { + "epoch": 0.3137062746699333, + "grad_norm": 2.336487054824829, + "learning_rate": 1.374463677252283e-05, + "loss": 2.508363151550293, + "step": 115240 + }, + { + "epoch": 0.31376071866067784, + "grad_norm": 2.165107011795044, + "learning_rate": 1.3743546408613877e-05, + "loss": 2.4960342407226563, + "step": 115260 + }, + { + "epoch": 0.31381516265142234, + "grad_norm": 2.159085750579834, + "learning_rate": 1.3742456044704922e-05, + "loss": 2.5260446548461912, + "step": 115280 + }, + { + "epoch": 0.3138696066421669, + "grad_norm": 2.1640727519989014, + "learning_rate": 1.3741365680795966e-05, + "loss": 2.5119255065917967, + "step": 115300 + }, + { + "epoch": 0.3139240506329114, + "grad_norm": 2.144270181655884, + "learning_rate": 1.3740275316887012e-05, + "loss": 2.5056427001953123, + "step": 115320 + }, + { + "epoch": 0.3139784946236559, + "grad_norm": 2.139638662338257, + "learning_rate": 1.3739184952978058e-05, + "loss": 2.497130012512207, + "step": 115340 + }, + { + "epoch": 0.31403293861440046, + "grad_norm": 2.2793755531311035, + "learning_rate": 1.3738094589069102e-05, + "loss": 2.5263696670532227, + "step": 115360 + }, + { + "epoch": 0.31408738260514496, + "grad_norm": 2.140880823135376, + "learning_rate": 1.3737004225160148e-05, + "loss": 2.476099967956543, + "step": 115380 + }, + { + "epoch": 0.31414182659588946, + "grad_norm": 2.189626932144165, + "learning_rate": 1.3735913861251194e-05, + "loss": 2.4670394897460937, + "step": 115400 + }, + { + "epoch": 0.314196270586634, + "grad_norm": 2.1033976078033447, + "learning_rate": 1.373482349734224e-05, + "loss": 2.5011838912963866, + "step": 115420 + }, + { + "epoch": 0.3142507145773785, + "grad_norm": 2.138270854949951, + "learning_rate": 1.3733733133433285e-05, + "loss": 2.5080286026000977, + "step": 115440 + }, + { + "epoch": 0.314305158568123, + "grad_norm": 2.1645615100860596, + "learning_rate": 1.373264276952433e-05, + "loss": 2.498575973510742, + "step": 115460 + }, + { + "epoch": 0.3143596025588676, + "grad_norm": 2.120455265045166, + "learning_rate": 1.3731552405615375e-05, + "loss": 2.5037235260009765, + "step": 115480 + }, + { + "epoch": 0.3144140465496121, + "grad_norm": 2.0774850845336914, + "learning_rate": 1.3730462041706421e-05, + "loss": 2.501389503479004, + "step": 115500 + }, + { + "epoch": 0.3144684905403566, + "grad_norm": 2.177842617034912, + "learning_rate": 1.3729371677797465e-05, + "loss": 2.497692680358887, + "step": 115520 + }, + { + "epoch": 0.31452293453110114, + "grad_norm": 2.1373629570007324, + "learning_rate": 1.3728281313888511e-05, + "loss": 2.522882080078125, + "step": 115540 + }, + { + "epoch": 0.31457737852184564, + "grad_norm": 2.2497754096984863, + "learning_rate": 1.3727190949979557e-05, + "loss": 2.502815055847168, + "step": 115560 + }, + { + "epoch": 0.3146318225125902, + "grad_norm": 2.325895071029663, + "learning_rate": 1.3726100586070603e-05, + "loss": 2.5268049240112305, + "step": 115580 + }, + { + "epoch": 0.3146862665033347, + "grad_norm": 2.147273302078247, + "learning_rate": 1.3725010222161649e-05, + "loss": 2.499612236022949, + "step": 115600 + }, + { + "epoch": 0.3147407104940792, + "grad_norm": 2.2331669330596924, + "learning_rate": 1.3723919858252693e-05, + "loss": 2.5143280029296875, + "step": 115620 + }, + { + "epoch": 0.31479515448482376, + "grad_norm": 2.3315131664276123, + "learning_rate": 1.3722829494343738e-05, + "loss": 2.5000484466552733, + "step": 115640 + }, + { + "epoch": 0.31484959847556826, + "grad_norm": 2.2264297008514404, + "learning_rate": 1.3721739130434784e-05, + "loss": 2.4917835235595702, + "step": 115660 + }, + { + "epoch": 0.31490404246631276, + "grad_norm": 2.2007477283477783, + "learning_rate": 1.3720648766525828e-05, + "loss": 2.5353588104248046, + "step": 115680 + }, + { + "epoch": 0.3149584864570573, + "grad_norm": 2.2257049083709717, + "learning_rate": 1.3719558402616874e-05, + "loss": 2.4856855392456056, + "step": 115700 + }, + { + "epoch": 0.3150129304478018, + "grad_norm": 2.1906280517578125, + "learning_rate": 1.371846803870792e-05, + "loss": 2.485055923461914, + "step": 115720 + }, + { + "epoch": 0.3150673744385463, + "grad_norm": 2.2826480865478516, + "learning_rate": 1.3717377674798966e-05, + "loss": 2.477486801147461, + "step": 115740 + }, + { + "epoch": 0.3151218184292909, + "grad_norm": 2.273185968399048, + "learning_rate": 1.371628731089001e-05, + "loss": 2.483157157897949, + "step": 115760 + }, + { + "epoch": 0.3151762624200354, + "grad_norm": 2.169166326522827, + "learning_rate": 1.3715196946981056e-05, + "loss": 2.5024177551269533, + "step": 115780 + }, + { + "epoch": 0.31523070641077994, + "grad_norm": 2.0812385082244873, + "learning_rate": 1.3714106583072102e-05, + "loss": 2.4875514984130858, + "step": 115800 + }, + { + "epoch": 0.31528515040152444, + "grad_norm": 2.1557300090789795, + "learning_rate": 1.3713016219163147e-05, + "loss": 2.4885562896728515, + "step": 115820 + }, + { + "epoch": 0.31533959439226894, + "grad_norm": 2.2842466831207275, + "learning_rate": 1.3711925855254191e-05, + "loss": 2.504220199584961, + "step": 115840 + }, + { + "epoch": 0.3153940383830135, + "grad_norm": 2.130688190460205, + "learning_rate": 1.3710835491345237e-05, + "loss": 2.5208478927612306, + "step": 115860 + }, + { + "epoch": 0.315448482373758, + "grad_norm": 2.241117477416992, + "learning_rate": 1.3709745127436283e-05, + "loss": 2.5112173080444338, + "step": 115880 + }, + { + "epoch": 0.3155029263645025, + "grad_norm": 2.2178657054901123, + "learning_rate": 1.3708654763527329e-05, + "loss": 2.4883544921875, + "step": 115900 + }, + { + "epoch": 0.31555737035524706, + "grad_norm": 2.1592230796813965, + "learning_rate": 1.3707564399618373e-05, + "loss": 2.5043344497680664, + "step": 115920 + }, + { + "epoch": 0.31561181434599156, + "grad_norm": 2.123727321624756, + "learning_rate": 1.3706474035709419e-05, + "loss": 2.4968137741088867, + "step": 115940 + }, + { + "epoch": 0.31566625833673606, + "grad_norm": 2.1578876972198486, + "learning_rate": 1.3705383671800465e-05, + "loss": 2.5100065231323243, + "step": 115960 + }, + { + "epoch": 0.3157207023274806, + "grad_norm": 2.151536226272583, + "learning_rate": 1.370429330789151e-05, + "loss": 2.5109380722045898, + "step": 115980 + }, + { + "epoch": 0.3157751463182251, + "grad_norm": 2.1871964931488037, + "learning_rate": 1.3703202943982555e-05, + "loss": 2.487060546875, + "step": 116000 + }, + { + "epoch": 0.3158295903089696, + "grad_norm": 2.074012517929077, + "learning_rate": 1.37021125800736e-05, + "loss": 2.5142999649047852, + "step": 116020 + }, + { + "epoch": 0.3158840342997142, + "grad_norm": 2.1149425506591797, + "learning_rate": 1.3701022216164646e-05, + "loss": 2.4852914810180664, + "step": 116040 + }, + { + "epoch": 0.3159384782904587, + "grad_norm": 2.0388379096984863, + "learning_rate": 1.3699931852255692e-05, + "loss": 2.4580154418945312, + "step": 116060 + }, + { + "epoch": 0.31599292228120324, + "grad_norm": 2.1247823238372803, + "learning_rate": 1.3698841488346736e-05, + "loss": 2.513997268676758, + "step": 116080 + }, + { + "epoch": 0.31604736627194774, + "grad_norm": 2.0261988639831543, + "learning_rate": 1.3697751124437782e-05, + "loss": 2.5188507080078124, + "step": 116100 + }, + { + "epoch": 0.31610181026269224, + "grad_norm": 2.1799933910369873, + "learning_rate": 1.3696660760528828e-05, + "loss": 2.492543411254883, + "step": 116120 + }, + { + "epoch": 0.3161562542534368, + "grad_norm": 2.1413285732269287, + "learning_rate": 1.3695570396619874e-05, + "loss": 2.5134605407714843, + "step": 116140 + }, + { + "epoch": 0.3162106982441813, + "grad_norm": 2.185121774673462, + "learning_rate": 1.3694480032710918e-05, + "loss": 2.4997066497802733, + "step": 116160 + }, + { + "epoch": 0.3162651422349258, + "grad_norm": 2.106658935546875, + "learning_rate": 1.3693389668801963e-05, + "loss": 2.493520736694336, + "step": 116180 + }, + { + "epoch": 0.31631958622567036, + "grad_norm": 2.076584815979004, + "learning_rate": 1.369229930489301e-05, + "loss": 2.495746612548828, + "step": 116200 + }, + { + "epoch": 0.31637403021641486, + "grad_norm": 2.1203346252441406, + "learning_rate": 1.3691208940984055e-05, + "loss": 2.475300979614258, + "step": 116220 + }, + { + "epoch": 0.31642847420715936, + "grad_norm": 2.1164331436157227, + "learning_rate": 1.36901185770751e-05, + "loss": 2.479769706726074, + "step": 116240 + }, + { + "epoch": 0.3164829181979039, + "grad_norm": 2.2182607650756836, + "learning_rate": 1.3689028213166145e-05, + "loss": 2.4885433197021483, + "step": 116260 + }, + { + "epoch": 0.3165373621886484, + "grad_norm": 2.1180295944213867, + "learning_rate": 1.368793784925719e-05, + "loss": 2.512736511230469, + "step": 116280 + }, + { + "epoch": 0.316591806179393, + "grad_norm": 2.0929644107818604, + "learning_rate": 1.3686847485348237e-05, + "loss": 2.483786964416504, + "step": 116300 + }, + { + "epoch": 0.3166462501701375, + "grad_norm": 2.0550520420074463, + "learning_rate": 1.3685757121439282e-05, + "loss": 2.5022037506103514, + "step": 116320 + }, + { + "epoch": 0.316700694160882, + "grad_norm": 2.4590485095977783, + "learning_rate": 1.3684666757530327e-05, + "loss": 2.500203514099121, + "step": 116340 + }, + { + "epoch": 0.31675513815162654, + "grad_norm": 2.6664626598358154, + "learning_rate": 1.3683576393621372e-05, + "loss": 2.5048927307128905, + "step": 116360 + }, + { + "epoch": 0.31680958214237104, + "grad_norm": 2.1367924213409424, + "learning_rate": 1.3682486029712418e-05, + "loss": 2.5004568099975586, + "step": 116380 + }, + { + "epoch": 0.31686402613311554, + "grad_norm": 2.0501010417938232, + "learning_rate": 1.3681395665803462e-05, + "loss": 2.5173131942749025, + "step": 116400 + }, + { + "epoch": 0.3169184701238601, + "grad_norm": 2.189595937728882, + "learning_rate": 1.3680305301894508e-05, + "loss": 2.516595649719238, + "step": 116420 + }, + { + "epoch": 0.3169729141146046, + "grad_norm": 2.13114857673645, + "learning_rate": 1.3679214937985554e-05, + "loss": 2.528969383239746, + "step": 116440 + }, + { + "epoch": 0.3170273581053491, + "grad_norm": 2.1769161224365234, + "learning_rate": 1.36781245740766e-05, + "loss": 2.5299331665039064, + "step": 116460 + }, + { + "epoch": 0.31708180209609366, + "grad_norm": 3.482630729675293, + "learning_rate": 1.3677034210167645e-05, + "loss": 2.5021148681640626, + "step": 116480 + }, + { + "epoch": 0.31713624608683816, + "grad_norm": 2.193227529525757, + "learning_rate": 1.367594384625869e-05, + "loss": 2.5027923583984375, + "step": 116500 + }, + { + "epoch": 0.31719069007758266, + "grad_norm": 2.558729410171509, + "learning_rate": 1.3674853482349735e-05, + "loss": 2.4843610763549804, + "step": 116520 + }, + { + "epoch": 0.3172451340683272, + "grad_norm": 2.1997408866882324, + "learning_rate": 1.3673763118440781e-05, + "loss": 2.5111257553100588, + "step": 116540 + }, + { + "epoch": 0.3172995780590717, + "grad_norm": 2.2590134143829346, + "learning_rate": 1.3672672754531825e-05, + "loss": 2.4888147354125976, + "step": 116560 + }, + { + "epoch": 0.3173540220498163, + "grad_norm": 2.138443946838379, + "learning_rate": 1.3671582390622871e-05, + "loss": 2.519131088256836, + "step": 116580 + }, + { + "epoch": 0.3174084660405608, + "grad_norm": 2.2157530784606934, + "learning_rate": 1.3670492026713917e-05, + "loss": 2.5195249557495116, + "step": 116600 + }, + { + "epoch": 0.3174629100313053, + "grad_norm": 2.2291479110717773, + "learning_rate": 1.3669401662804963e-05, + "loss": 2.5088266372680663, + "step": 116620 + }, + { + "epoch": 0.31751735402204984, + "grad_norm": 2.0657992362976074, + "learning_rate": 1.3668311298896009e-05, + "loss": 2.495823287963867, + "step": 116640 + }, + { + "epoch": 0.31757179801279434, + "grad_norm": 2.099072217941284, + "learning_rate": 1.3667220934987053e-05, + "loss": 2.4847270965576174, + "step": 116660 + }, + { + "epoch": 0.31762624200353884, + "grad_norm": 2.184901714324951, + "learning_rate": 1.3666130571078098e-05, + "loss": 2.5020545959472655, + "step": 116680 + }, + { + "epoch": 0.3176806859942834, + "grad_norm": 2.3093249797821045, + "learning_rate": 1.3665040207169144e-05, + "loss": 2.4989603042602537, + "step": 116700 + }, + { + "epoch": 0.3177351299850279, + "grad_norm": 2.09401798248291, + "learning_rate": 1.3663949843260188e-05, + "loss": 2.5181880950927735, + "step": 116720 + }, + { + "epoch": 0.3177895739757724, + "grad_norm": 2.0917105674743652, + "learning_rate": 1.3662859479351234e-05, + "loss": 2.505847358703613, + "step": 116740 + }, + { + "epoch": 0.31784401796651696, + "grad_norm": 2.197563648223877, + "learning_rate": 1.366176911544228e-05, + "loss": 2.508871078491211, + "step": 116760 + }, + { + "epoch": 0.31789846195726146, + "grad_norm": 2.1569647789001465, + "learning_rate": 1.3660678751533326e-05, + "loss": 2.513523483276367, + "step": 116780 + }, + { + "epoch": 0.317952905948006, + "grad_norm": 2.1151764392852783, + "learning_rate": 1.3659588387624372e-05, + "loss": 2.4842185974121094, + "step": 116800 + }, + { + "epoch": 0.3180073499387505, + "grad_norm": 2.101332664489746, + "learning_rate": 1.3658498023715416e-05, + "loss": 2.485873222351074, + "step": 116820 + }, + { + "epoch": 0.318061793929495, + "grad_norm": 2.092128276824951, + "learning_rate": 1.3657407659806462e-05, + "loss": 2.473368835449219, + "step": 116840 + }, + { + "epoch": 0.3181162379202396, + "grad_norm": 2.238401174545288, + "learning_rate": 1.3656317295897507e-05, + "loss": 2.5072324752807615, + "step": 116860 + }, + { + "epoch": 0.3181706819109841, + "grad_norm": 2.0387256145477295, + "learning_rate": 1.3655226931988552e-05, + "loss": 2.4966142654418944, + "step": 116880 + }, + { + "epoch": 0.3182251259017286, + "grad_norm": 2.114067792892456, + "learning_rate": 1.3654136568079597e-05, + "loss": 2.484404754638672, + "step": 116900 + }, + { + "epoch": 0.31827956989247314, + "grad_norm": 2.079709768295288, + "learning_rate": 1.3653046204170643e-05, + "loss": 2.459677314758301, + "step": 116920 + }, + { + "epoch": 0.31833401388321764, + "grad_norm": 2.1678080558776855, + "learning_rate": 1.3651955840261687e-05, + "loss": 2.5013099670410157, + "step": 116940 + }, + { + "epoch": 0.31838845787396214, + "grad_norm": 2.1394999027252197, + "learning_rate": 1.3650865476352735e-05, + "loss": 2.523228073120117, + "step": 116960 + }, + { + "epoch": 0.3184429018647067, + "grad_norm": 2.1599488258361816, + "learning_rate": 1.3649775112443779e-05, + "loss": 2.529387283325195, + "step": 116980 + }, + { + "epoch": 0.3184973458554512, + "grad_norm": 2.0932962894439697, + "learning_rate": 1.3648684748534825e-05, + "loss": 2.5151657104492187, + "step": 117000 + }, + { + "epoch": 0.3185517898461957, + "grad_norm": 2.126814842224121, + "learning_rate": 1.364759438462587e-05, + "loss": 2.497310256958008, + "step": 117020 + }, + { + "epoch": 0.31860623383694026, + "grad_norm": 2.276381254196167, + "learning_rate": 1.3646504020716915e-05, + "loss": 2.4976823806762694, + "step": 117040 + }, + { + "epoch": 0.31866067782768476, + "grad_norm": 2.0984461307525635, + "learning_rate": 1.364541365680796e-05, + "loss": 2.4929269790649413, + "step": 117060 + }, + { + "epoch": 0.3187151218184293, + "grad_norm": 2.077679395675659, + "learning_rate": 1.3644323292899006e-05, + "loss": 2.4946128845214846, + "step": 117080 + }, + { + "epoch": 0.3187695658091738, + "grad_norm": 2.1362054347991943, + "learning_rate": 1.364323292899005e-05, + "loss": 2.483063888549805, + "step": 117100 + }, + { + "epoch": 0.3188240097999183, + "grad_norm": 2.2507572174072266, + "learning_rate": 1.3642142565081098e-05, + "loss": 2.506245994567871, + "step": 117120 + }, + { + "epoch": 0.3188784537906629, + "grad_norm": 2.1200335025787354, + "learning_rate": 1.3641052201172142e-05, + "loss": 2.4900079727172852, + "step": 117140 + }, + { + "epoch": 0.3189328977814074, + "grad_norm": 2.161473035812378, + "learning_rate": 1.3639961837263188e-05, + "loss": 2.4995582580566404, + "step": 117160 + }, + { + "epoch": 0.3189873417721519, + "grad_norm": 2.122847080230713, + "learning_rate": 1.3638871473354234e-05, + "loss": 2.495783805847168, + "step": 117180 + }, + { + "epoch": 0.31904178576289643, + "grad_norm": 2.125203847885132, + "learning_rate": 1.3637781109445278e-05, + "loss": 2.5010473251342775, + "step": 117200 + }, + { + "epoch": 0.31909622975364094, + "grad_norm": 2.161719560623169, + "learning_rate": 1.3636690745536323e-05, + "loss": 2.501437187194824, + "step": 117220 + }, + { + "epoch": 0.31915067374438544, + "grad_norm": 2.4069530963897705, + "learning_rate": 1.363560038162737e-05, + "loss": 2.5269874572753905, + "step": 117240 + }, + { + "epoch": 0.31920511773513, + "grad_norm": 2.123732089996338, + "learning_rate": 1.3634510017718413e-05, + "loss": 2.5000476837158203, + "step": 117260 + }, + { + "epoch": 0.3192595617258745, + "grad_norm": 2.3164384365081787, + "learning_rate": 1.3633419653809461e-05, + "loss": 2.5316864013671876, + "step": 117280 + }, + { + "epoch": 0.31931400571661905, + "grad_norm": 2.1972761154174805, + "learning_rate": 1.3632329289900505e-05, + "loss": 2.504292678833008, + "step": 117300 + }, + { + "epoch": 0.31936844970736356, + "grad_norm": 2.139171838760376, + "learning_rate": 1.363123892599155e-05, + "loss": 2.478034782409668, + "step": 117320 + }, + { + "epoch": 0.31942289369810806, + "grad_norm": 2.0342838764190674, + "learning_rate": 1.3630148562082597e-05, + "loss": 2.47427921295166, + "step": 117340 + }, + { + "epoch": 0.3194773376888526, + "grad_norm": 2.175900936126709, + "learning_rate": 1.362905819817364e-05, + "loss": 2.5025728225708006, + "step": 117360 + }, + { + "epoch": 0.3195317816795971, + "grad_norm": 2.2518625259399414, + "learning_rate": 1.3627967834264687e-05, + "loss": 2.505170440673828, + "step": 117380 + }, + { + "epoch": 0.3195862256703416, + "grad_norm": 2.168297529220581, + "learning_rate": 1.3626877470355732e-05, + "loss": 2.515735626220703, + "step": 117400 + }, + { + "epoch": 0.3196406696610862, + "grad_norm": 2.2265522480010986, + "learning_rate": 1.3625787106446776e-05, + "loss": 2.5087894439697265, + "step": 117420 + }, + { + "epoch": 0.3196951136518307, + "grad_norm": 2.1571643352508545, + "learning_rate": 1.3624696742537824e-05, + "loss": 2.516057777404785, + "step": 117440 + }, + { + "epoch": 0.3197495576425752, + "grad_norm": 2.117255926132202, + "learning_rate": 1.3623606378628868e-05, + "loss": 2.5070491790771485, + "step": 117460 + }, + { + "epoch": 0.31980400163331973, + "grad_norm": 2.153151512145996, + "learning_rate": 1.3622516014719914e-05, + "loss": 2.501953887939453, + "step": 117480 + }, + { + "epoch": 0.31985844562406424, + "grad_norm": 2.2791664600372314, + "learning_rate": 1.362142565081096e-05, + "loss": 2.4947893142700197, + "step": 117500 + }, + { + "epoch": 0.3199128896148088, + "grad_norm": 2.0817136764526367, + "learning_rate": 1.3620335286902004e-05, + "loss": 2.502641296386719, + "step": 117520 + }, + { + "epoch": 0.3199673336055533, + "grad_norm": 2.2453224658966064, + "learning_rate": 1.361924492299305e-05, + "loss": 2.5023513793945313, + "step": 117540 + }, + { + "epoch": 0.3200217775962978, + "grad_norm": 2.062014102935791, + "learning_rate": 1.3618154559084095e-05, + "loss": 2.49930419921875, + "step": 117560 + }, + { + "epoch": 0.32007622158704235, + "grad_norm": 2.160963296890259, + "learning_rate": 1.361706419517514e-05, + "loss": 2.525313377380371, + "step": 117580 + }, + { + "epoch": 0.32013066557778685, + "grad_norm": 2.16556453704834, + "learning_rate": 1.3615973831266187e-05, + "loss": 2.458750534057617, + "step": 117600 + }, + { + "epoch": 0.32018510956853136, + "grad_norm": 2.100958824157715, + "learning_rate": 1.3614883467357231e-05, + "loss": 2.5268096923828125, + "step": 117620 + }, + { + "epoch": 0.3202395535592759, + "grad_norm": 2.119382858276367, + "learning_rate": 1.3613793103448277e-05, + "loss": 2.512453079223633, + "step": 117640 + }, + { + "epoch": 0.3202939975500204, + "grad_norm": 2.234332323074341, + "learning_rate": 1.3612702739539323e-05, + "loss": 2.4911405563354494, + "step": 117660 + }, + { + "epoch": 0.3203484415407649, + "grad_norm": 2.0644707679748535, + "learning_rate": 1.3611612375630367e-05, + "loss": 2.4922391891479494, + "step": 117680 + }, + { + "epoch": 0.3204028855315095, + "grad_norm": 2.187610626220703, + "learning_rate": 1.3610522011721413e-05, + "loss": 2.5077627182006834, + "step": 117700 + }, + { + "epoch": 0.320457329522254, + "grad_norm": 2.14052677154541, + "learning_rate": 1.3609431647812459e-05, + "loss": 2.505929374694824, + "step": 117720 + }, + { + "epoch": 0.3205117735129985, + "grad_norm": 2.155557155609131, + "learning_rate": 1.3608341283903503e-05, + "loss": 2.5228614807128906, + "step": 117740 + }, + { + "epoch": 0.32056621750374303, + "grad_norm": 2.0682692527770996, + "learning_rate": 1.360725091999455e-05, + "loss": 2.4891061782836914, + "step": 117760 + }, + { + "epoch": 0.32062066149448754, + "grad_norm": 2.06718111038208, + "learning_rate": 1.3606160556085594e-05, + "loss": 2.5254184722900392, + "step": 117780 + }, + { + "epoch": 0.3206751054852321, + "grad_norm": 2.144204616546631, + "learning_rate": 1.360507019217664e-05, + "loss": 2.496623229980469, + "step": 117800 + }, + { + "epoch": 0.3207295494759766, + "grad_norm": 2.875966787338257, + "learning_rate": 1.3603979828267686e-05, + "loss": 2.5291006088256838, + "step": 117820 + }, + { + "epoch": 0.3207839934667211, + "grad_norm": 2.1093857288360596, + "learning_rate": 1.360288946435873e-05, + "loss": 2.4607948303222655, + "step": 117840 + }, + { + "epoch": 0.32083843745746565, + "grad_norm": 2.2198026180267334, + "learning_rate": 1.3601799100449776e-05, + "loss": 2.5224588394165037, + "step": 117860 + }, + { + "epoch": 0.32089288144821015, + "grad_norm": 2.2107739448547363, + "learning_rate": 1.3600708736540822e-05, + "loss": 2.5255081176757814, + "step": 117880 + }, + { + "epoch": 0.32094732543895466, + "grad_norm": 2.580570697784424, + "learning_rate": 1.3599618372631866e-05, + "loss": 2.506851005554199, + "step": 117900 + }, + { + "epoch": 0.3210017694296992, + "grad_norm": 2.1490321159362793, + "learning_rate": 1.3598528008722913e-05, + "loss": 2.5294971466064453, + "step": 117920 + }, + { + "epoch": 0.3210562134204437, + "grad_norm": 2.139840841293335, + "learning_rate": 1.3597437644813957e-05, + "loss": 2.5183544158935547, + "step": 117940 + }, + { + "epoch": 0.3211106574111882, + "grad_norm": 2.139434814453125, + "learning_rate": 1.3596347280905001e-05, + "loss": 2.5095733642578124, + "step": 117960 + }, + { + "epoch": 0.3211651014019328, + "grad_norm": 2.149243116378784, + "learning_rate": 1.3595256916996049e-05, + "loss": 2.5114002227783203, + "step": 117980 + }, + { + "epoch": 0.3212195453926773, + "grad_norm": 2.1055266857147217, + "learning_rate": 1.3594166553087093e-05, + "loss": 2.518703079223633, + "step": 118000 + }, + { + "epoch": 0.32127398938342183, + "grad_norm": 2.098660707473755, + "learning_rate": 1.359307618917814e-05, + "loss": 2.4953330993652343, + "step": 118020 + }, + { + "epoch": 0.32132843337416633, + "grad_norm": 2.2420244216918945, + "learning_rate": 1.3591985825269185e-05, + "loss": 2.4889875411987306, + "step": 118040 + }, + { + "epoch": 0.32138287736491084, + "grad_norm": 2.187696695327759, + "learning_rate": 1.3590895461360229e-05, + "loss": 2.5089670181274415, + "step": 118060 + }, + { + "epoch": 0.3214373213556554, + "grad_norm": 2.1360297203063965, + "learning_rate": 1.3589805097451276e-05, + "loss": 2.522548484802246, + "step": 118080 + }, + { + "epoch": 0.3214917653463999, + "grad_norm": 2.240306854248047, + "learning_rate": 1.358871473354232e-05, + "loss": 2.5179140090942385, + "step": 118100 + }, + { + "epoch": 0.3215462093371444, + "grad_norm": 2.0625123977661133, + "learning_rate": 1.3587624369633365e-05, + "loss": 2.4929082870483397, + "step": 118120 + }, + { + "epoch": 0.32160065332788895, + "grad_norm": 2.118220806121826, + "learning_rate": 1.3586534005724412e-05, + "loss": 2.485493469238281, + "step": 118140 + }, + { + "epoch": 0.32165509731863345, + "grad_norm": 2.225165843963623, + "learning_rate": 1.3585443641815456e-05, + "loss": 2.513661575317383, + "step": 118160 + }, + { + "epoch": 0.32170954130937796, + "grad_norm": 2.092916965484619, + "learning_rate": 1.3584353277906504e-05, + "loss": 2.477817153930664, + "step": 118180 + }, + { + "epoch": 0.3217639853001225, + "grad_norm": 2.0835049152374268, + "learning_rate": 1.3583262913997548e-05, + "loss": 2.4997226715087892, + "step": 118200 + }, + { + "epoch": 0.321818429290867, + "grad_norm": 2.109102249145508, + "learning_rate": 1.3582172550088592e-05, + "loss": 2.521352767944336, + "step": 118220 + }, + { + "epoch": 0.3218728732816115, + "grad_norm": 2.1098222732543945, + "learning_rate": 1.358108218617964e-05, + "loss": 2.5064424514770507, + "step": 118240 + }, + { + "epoch": 0.3219273172723561, + "grad_norm": 2.2195394039154053, + "learning_rate": 1.3579991822270684e-05, + "loss": 2.4898534774780274, + "step": 118260 + }, + { + "epoch": 0.3219817612631006, + "grad_norm": 2.1047656536102295, + "learning_rate": 1.3578901458361728e-05, + "loss": 2.507845878601074, + "step": 118280 + }, + { + "epoch": 0.32203620525384513, + "grad_norm": 2.162142038345337, + "learning_rate": 1.3577811094452775e-05, + "loss": 2.5272100448608397, + "step": 118300 + }, + { + "epoch": 0.32209064924458963, + "grad_norm": 2.1288743019104004, + "learning_rate": 1.357672073054382e-05, + "loss": 2.516522216796875, + "step": 118320 + }, + { + "epoch": 0.32214509323533413, + "grad_norm": 2.078038454055786, + "learning_rate": 1.3575630366634867e-05, + "loss": 2.529213523864746, + "step": 118340 + }, + { + "epoch": 0.3221995372260787, + "grad_norm": 2.12166166305542, + "learning_rate": 1.3574540002725911e-05, + "loss": 2.5073064804077148, + "step": 118360 + }, + { + "epoch": 0.3222539812168232, + "grad_norm": 2.277883529663086, + "learning_rate": 1.3573449638816955e-05, + "loss": 2.4909082412719727, + "step": 118380 + }, + { + "epoch": 0.3223084252075677, + "grad_norm": 3.735929012298584, + "learning_rate": 1.3572359274908003e-05, + "loss": 2.508626937866211, + "step": 118400 + }, + { + "epoch": 0.32236286919831225, + "grad_norm": 2.3361313343048096, + "learning_rate": 1.3571268910999047e-05, + "loss": 2.4868932723999024, + "step": 118420 + }, + { + "epoch": 0.32241731318905675, + "grad_norm": 2.1148669719696045, + "learning_rate": 1.357017854709009e-05, + "loss": 2.497017669677734, + "step": 118440 + }, + { + "epoch": 0.32247175717980125, + "grad_norm": 2.2114570140838623, + "learning_rate": 1.3569088183181138e-05, + "loss": 2.52294979095459, + "step": 118460 + }, + { + "epoch": 0.3225262011705458, + "grad_norm": 2.216857433319092, + "learning_rate": 1.3567997819272182e-05, + "loss": 2.508920669555664, + "step": 118480 + }, + { + "epoch": 0.3225806451612903, + "grad_norm": 2.160459518432617, + "learning_rate": 1.356690745536323e-05, + "loss": 2.519358253479004, + "step": 118500 + }, + { + "epoch": 0.32263508915203487, + "grad_norm": 2.175762176513672, + "learning_rate": 1.3565817091454274e-05, + "loss": 2.4996482849121096, + "step": 118520 + }, + { + "epoch": 0.3226895331427794, + "grad_norm": 2.137749433517456, + "learning_rate": 1.3564726727545318e-05, + "loss": 2.4890275955200196, + "step": 118540 + }, + { + "epoch": 0.3227439771335239, + "grad_norm": 2.203723192214966, + "learning_rate": 1.3563636363636366e-05, + "loss": 2.4726749420166017, + "step": 118560 + }, + { + "epoch": 0.32279842112426843, + "grad_norm": 2.188958168029785, + "learning_rate": 1.356254599972741e-05, + "loss": 2.5014719009399413, + "step": 118580 + }, + { + "epoch": 0.32285286511501293, + "grad_norm": 2.2240943908691406, + "learning_rate": 1.3561455635818454e-05, + "loss": 2.4882068634033203, + "step": 118600 + }, + { + "epoch": 0.32290730910575743, + "grad_norm": 2.046078681945801, + "learning_rate": 1.3560365271909501e-05, + "loss": 2.4930984497070314, + "step": 118620 + }, + { + "epoch": 0.322961753096502, + "grad_norm": 2.218593120574951, + "learning_rate": 1.3559274908000545e-05, + "loss": 2.5086761474609376, + "step": 118640 + }, + { + "epoch": 0.3230161970872465, + "grad_norm": 2.1112539768218994, + "learning_rate": 1.3558184544091593e-05, + "loss": 2.4969388961791994, + "step": 118660 + }, + { + "epoch": 0.323070641077991, + "grad_norm": 2.312835693359375, + "learning_rate": 1.3557094180182637e-05, + "loss": 2.5286516189575194, + "step": 118680 + }, + { + "epoch": 0.32312508506873555, + "grad_norm": 2.3019864559173584, + "learning_rate": 1.3556003816273681e-05, + "loss": 2.5153913497924805, + "step": 118700 + }, + { + "epoch": 0.32317952905948005, + "grad_norm": 2.164262294769287, + "learning_rate": 1.3554913452364729e-05, + "loss": 2.5255062103271486, + "step": 118720 + }, + { + "epoch": 0.32323397305022455, + "grad_norm": 2.1528401374816895, + "learning_rate": 1.3553823088455773e-05, + "loss": 2.492348289489746, + "step": 118740 + }, + { + "epoch": 0.3232884170409691, + "grad_norm": 2.101773500442505, + "learning_rate": 1.3552732724546817e-05, + "loss": 2.5130020141601563, + "step": 118760 + }, + { + "epoch": 0.3233428610317136, + "grad_norm": 2.0933284759521484, + "learning_rate": 1.3551642360637864e-05, + "loss": 2.519557571411133, + "step": 118780 + }, + { + "epoch": 0.32339730502245817, + "grad_norm": 2.0985984802246094, + "learning_rate": 1.3550551996728909e-05, + "loss": 2.4973051071166994, + "step": 118800 + }, + { + "epoch": 0.32345174901320267, + "grad_norm": 2.1324639320373535, + "learning_rate": 1.3549461632819956e-05, + "loss": 2.484493637084961, + "step": 118820 + }, + { + "epoch": 0.3235061930039472, + "grad_norm": 2.19240665435791, + "learning_rate": 1.3548371268911e-05, + "loss": 2.4957324981689455, + "step": 118840 + }, + { + "epoch": 0.32356063699469173, + "grad_norm": 2.1168324947357178, + "learning_rate": 1.3547280905002044e-05, + "loss": 2.499069023132324, + "step": 118860 + }, + { + "epoch": 0.32361508098543623, + "grad_norm": 2.143883228302002, + "learning_rate": 1.3546190541093092e-05, + "loss": 2.4844783782958983, + "step": 118880 + }, + { + "epoch": 0.32366952497618073, + "grad_norm": 2.098994731903076, + "learning_rate": 1.3545100177184136e-05, + "loss": 2.491476821899414, + "step": 118900 + }, + { + "epoch": 0.3237239689669253, + "grad_norm": 2.1127912998199463, + "learning_rate": 1.354400981327518e-05, + "loss": 2.49536018371582, + "step": 118920 + }, + { + "epoch": 0.3237784129576698, + "grad_norm": 2.1007349491119385, + "learning_rate": 1.3542919449366228e-05, + "loss": 2.4821170806884765, + "step": 118940 + }, + { + "epoch": 0.3238328569484143, + "grad_norm": 2.3687617778778076, + "learning_rate": 1.3541829085457272e-05, + "loss": 2.503174591064453, + "step": 118960 + }, + { + "epoch": 0.32388730093915885, + "grad_norm": 2.101637601852417, + "learning_rate": 1.3540738721548319e-05, + "loss": 2.5207683563232424, + "step": 118980 + }, + { + "epoch": 0.32394174492990335, + "grad_norm": 2.1586363315582275, + "learning_rate": 1.3539648357639363e-05, + "loss": 2.4872257232666017, + "step": 119000 + }, + { + "epoch": 0.3239961889206479, + "grad_norm": 2.1658966541290283, + "learning_rate": 1.3538557993730407e-05, + "loss": 2.498103141784668, + "step": 119020 + }, + { + "epoch": 0.3240506329113924, + "grad_norm": 2.3670504093170166, + "learning_rate": 1.3537467629821455e-05, + "loss": 2.5058055877685548, + "step": 119040 + }, + { + "epoch": 0.3241050769021369, + "grad_norm": 2.063136100769043, + "learning_rate": 1.3536377265912499e-05, + "loss": 2.495920753479004, + "step": 119060 + }, + { + "epoch": 0.32415952089288147, + "grad_norm": 2.1690969467163086, + "learning_rate": 1.3535286902003543e-05, + "loss": 2.5294206619262694, + "step": 119080 + }, + { + "epoch": 0.32421396488362597, + "grad_norm": 2.5998528003692627, + "learning_rate": 1.353419653809459e-05, + "loss": 2.5015098571777346, + "step": 119100 + }, + { + "epoch": 0.3242684088743705, + "grad_norm": 2.065741777420044, + "learning_rate": 1.3533106174185635e-05, + "loss": 2.4863859176635743, + "step": 119120 + }, + { + "epoch": 0.32432285286511503, + "grad_norm": 2.572817325592041, + "learning_rate": 1.3532015810276682e-05, + "loss": 2.5140085220336914, + "step": 119140 + }, + { + "epoch": 0.32437729685585953, + "grad_norm": 2.1859631538391113, + "learning_rate": 1.3530925446367726e-05, + "loss": 2.5098957061767577, + "step": 119160 + }, + { + "epoch": 0.32443174084660403, + "grad_norm": 2.3607723712921143, + "learning_rate": 1.352983508245877e-05, + "loss": 2.5313148498535156, + "step": 119180 + }, + { + "epoch": 0.3244861848373486, + "grad_norm": 3.725543975830078, + "learning_rate": 1.3528744718549818e-05, + "loss": 2.491327095031738, + "step": 119200 + }, + { + "epoch": 0.3245406288280931, + "grad_norm": 2.1235740184783936, + "learning_rate": 1.3527654354640862e-05, + "loss": 2.5112092971801756, + "step": 119220 + }, + { + "epoch": 0.3245950728188376, + "grad_norm": 2.2112550735473633, + "learning_rate": 1.3526563990731906e-05, + "loss": 2.4861387252807616, + "step": 119240 + }, + { + "epoch": 0.32464951680958215, + "grad_norm": 2.155329704284668, + "learning_rate": 1.3525473626822954e-05, + "loss": 2.467201805114746, + "step": 119260 + }, + { + "epoch": 0.32470396080032665, + "grad_norm": 2.136469602584839, + "learning_rate": 1.3524383262913998e-05, + "loss": 2.5000722885131834, + "step": 119280 + }, + { + "epoch": 0.3247584047910712, + "grad_norm": 2.168976306915283, + "learning_rate": 1.3523292899005045e-05, + "loss": 2.495248794555664, + "step": 119300 + }, + { + "epoch": 0.3248128487818157, + "grad_norm": 2.482516288757324, + "learning_rate": 1.352220253509609e-05, + "loss": 2.5019485473632814, + "step": 119320 + }, + { + "epoch": 0.3248672927725602, + "grad_norm": 2.1389994621276855, + "learning_rate": 1.3521112171187134e-05, + "loss": 2.501530075073242, + "step": 119340 + }, + { + "epoch": 0.32492173676330477, + "grad_norm": 2.068795680999756, + "learning_rate": 1.3520021807278181e-05, + "loss": 2.4779836654663088, + "step": 119360 + }, + { + "epoch": 0.32497618075404927, + "grad_norm": 2.27731990814209, + "learning_rate": 1.3518931443369225e-05, + "loss": 2.5131624221801756, + "step": 119380 + }, + { + "epoch": 0.3250306247447938, + "grad_norm": 2.2023942470550537, + "learning_rate": 1.351784107946027e-05, + "loss": 2.5220310211181642, + "step": 119400 + }, + { + "epoch": 0.32508506873553833, + "grad_norm": 2.2010538578033447, + "learning_rate": 1.3516750715551317e-05, + "loss": 2.5023670196533203, + "step": 119420 + }, + { + "epoch": 0.32513951272628283, + "grad_norm": 2.0964784622192383, + "learning_rate": 1.3515660351642361e-05, + "loss": 2.482356834411621, + "step": 119440 + }, + { + "epoch": 0.32519395671702733, + "grad_norm": 2.2143635749816895, + "learning_rate": 1.3514569987733408e-05, + "loss": 2.499798393249512, + "step": 119460 + }, + { + "epoch": 0.3252484007077719, + "grad_norm": 2.1324901580810547, + "learning_rate": 1.3513479623824452e-05, + "loss": 2.5009246826171876, + "step": 119480 + }, + { + "epoch": 0.3253028446985164, + "grad_norm": 2.120584726333618, + "learning_rate": 1.3512389259915497e-05, + "loss": 2.510644721984863, + "step": 119500 + }, + { + "epoch": 0.32535728868926095, + "grad_norm": 2.0840930938720703, + "learning_rate": 1.3511298896006544e-05, + "loss": 2.5277873992919924, + "step": 119520 + }, + { + "epoch": 0.32541173268000545, + "grad_norm": 2.456263303756714, + "learning_rate": 1.3510208532097588e-05, + "loss": 2.509909248352051, + "step": 119540 + }, + { + "epoch": 0.32546617667074995, + "grad_norm": 2.0989444255828857, + "learning_rate": 1.3509118168188632e-05, + "loss": 2.500900459289551, + "step": 119560 + }, + { + "epoch": 0.3255206206614945, + "grad_norm": 2.0978236198425293, + "learning_rate": 1.350802780427968e-05, + "loss": 2.4909095764160156, + "step": 119580 + }, + { + "epoch": 0.325575064652239, + "grad_norm": 2.169023275375366, + "learning_rate": 1.3506937440370724e-05, + "loss": 2.5064844131469726, + "step": 119600 + }, + { + "epoch": 0.3256295086429835, + "grad_norm": 2.189643383026123, + "learning_rate": 1.3505847076461771e-05, + "loss": 2.4995401382446287, + "step": 119620 + }, + { + "epoch": 0.32568395263372807, + "grad_norm": 2.083204984664917, + "learning_rate": 1.3504756712552816e-05, + "loss": 2.517769432067871, + "step": 119640 + }, + { + "epoch": 0.32573839662447257, + "grad_norm": 2.3290133476257324, + "learning_rate": 1.350366634864386e-05, + "loss": 2.5227182388305662, + "step": 119660 + }, + { + "epoch": 0.32579284061521707, + "grad_norm": 2.1567530632019043, + "learning_rate": 1.3502575984734907e-05, + "loss": 2.490125846862793, + "step": 119680 + }, + { + "epoch": 0.32584728460596163, + "grad_norm": 2.3628904819488525, + "learning_rate": 1.3501485620825951e-05, + "loss": 2.5044326782226562, + "step": 119700 + }, + { + "epoch": 0.32590172859670613, + "grad_norm": 2.13010835647583, + "learning_rate": 1.3500395256916999e-05, + "loss": 2.5074607849121096, + "step": 119720 + }, + { + "epoch": 0.3259561725874507, + "grad_norm": 2.1426498889923096, + "learning_rate": 1.3499304893008043e-05, + "loss": 2.4928632736206056, + "step": 119740 + }, + { + "epoch": 0.3260106165781952, + "grad_norm": 2.058107614517212, + "learning_rate": 1.3498214529099087e-05, + "loss": 2.486636924743652, + "step": 119760 + }, + { + "epoch": 0.3260650605689397, + "grad_norm": 2.1975901126861572, + "learning_rate": 1.3497124165190135e-05, + "loss": 2.5094884872436523, + "step": 119780 + }, + { + "epoch": 0.32611950455968425, + "grad_norm": 2.3833138942718506, + "learning_rate": 1.3496033801281179e-05, + "loss": 2.5103733062744142, + "step": 119800 + }, + { + "epoch": 0.32617394855042875, + "grad_norm": 2.1442675590515137, + "learning_rate": 1.3494943437372223e-05, + "loss": 2.500404167175293, + "step": 119820 + }, + { + "epoch": 0.32622839254117325, + "grad_norm": 2.197098731994629, + "learning_rate": 1.349385307346327e-05, + "loss": 2.4793397903442385, + "step": 119840 + }, + { + "epoch": 0.3262828365319178, + "grad_norm": 2.1481761932373047, + "learning_rate": 1.3492762709554314e-05, + "loss": 2.4448740005493166, + "step": 119860 + }, + { + "epoch": 0.3263372805226623, + "grad_norm": 2.1533799171447754, + "learning_rate": 1.3491672345645362e-05, + "loss": 2.4823238372802736, + "step": 119880 + }, + { + "epoch": 0.3263917245134068, + "grad_norm": 2.274122953414917, + "learning_rate": 1.3490581981736406e-05, + "loss": 2.4917470932006838, + "step": 119900 + }, + { + "epoch": 0.32644616850415137, + "grad_norm": 2.190281867980957, + "learning_rate": 1.348949161782745e-05, + "loss": 2.487660026550293, + "step": 119920 + }, + { + "epoch": 0.32650061249489587, + "grad_norm": 2.254565477371216, + "learning_rate": 1.3488401253918498e-05, + "loss": 2.52786865234375, + "step": 119940 + }, + { + "epoch": 0.32655505648564037, + "grad_norm": 2.224947690963745, + "learning_rate": 1.3487310890009542e-05, + "loss": 2.5248422622680664, + "step": 119960 + }, + { + "epoch": 0.32660950047638493, + "grad_norm": 2.2336063385009766, + "learning_rate": 1.3486220526100586e-05, + "loss": 2.503876304626465, + "step": 119980 + }, + { + "epoch": 0.32666394446712943, + "grad_norm": 2.2523767948150635, + "learning_rate": 1.3485130162191633e-05, + "loss": 2.4679718017578125, + "step": 120000 + }, + { + "epoch": 0.326718388457874, + "grad_norm": 2.2228994369506836, + "learning_rate": 1.3484039798282677e-05, + "loss": 2.503855323791504, + "step": 120020 + }, + { + "epoch": 0.3267728324486185, + "grad_norm": 2.1362504959106445, + "learning_rate": 1.3482949434373725e-05, + "loss": 2.4855918884277344, + "step": 120040 + }, + { + "epoch": 0.326827276439363, + "grad_norm": 2.352267026901245, + "learning_rate": 1.3481859070464769e-05, + "loss": 2.5015232086181642, + "step": 120060 + }, + { + "epoch": 0.32688172043010755, + "grad_norm": 2.1450905799865723, + "learning_rate": 1.3480768706555813e-05, + "loss": 2.4757640838623045, + "step": 120080 + }, + { + "epoch": 0.32693616442085205, + "grad_norm": 2.1115429401397705, + "learning_rate": 1.347967834264686e-05, + "loss": 2.5117645263671875, + "step": 120100 + }, + { + "epoch": 0.32699060841159655, + "grad_norm": 2.1280810832977295, + "learning_rate": 1.3478587978737905e-05, + "loss": 2.509712028503418, + "step": 120120 + }, + { + "epoch": 0.3270450524023411, + "grad_norm": 2.147082567214966, + "learning_rate": 1.3477497614828949e-05, + "loss": 2.52188720703125, + "step": 120140 + }, + { + "epoch": 0.3270994963930856, + "grad_norm": 2.188546657562256, + "learning_rate": 1.3476407250919996e-05, + "loss": 2.4915443420410157, + "step": 120160 + }, + { + "epoch": 0.3271539403838301, + "grad_norm": 2.199735164642334, + "learning_rate": 1.347531688701104e-05, + "loss": 2.5042818069458006, + "step": 120180 + }, + { + "epoch": 0.32720838437457467, + "grad_norm": 2.1466355323791504, + "learning_rate": 1.3474226523102088e-05, + "loss": 2.516199493408203, + "step": 120200 + }, + { + "epoch": 0.32726282836531917, + "grad_norm": 2.158205270767212, + "learning_rate": 1.3473136159193132e-05, + "loss": 2.4914648056030275, + "step": 120220 + }, + { + "epoch": 0.3273172723560637, + "grad_norm": 2.0909583568573, + "learning_rate": 1.3472045795284176e-05, + "loss": 2.494062614440918, + "step": 120240 + }, + { + "epoch": 0.32737171634680823, + "grad_norm": 2.3421378135681152, + "learning_rate": 1.3470955431375224e-05, + "loss": 2.527840232849121, + "step": 120260 + }, + { + "epoch": 0.32742616033755273, + "grad_norm": 2.3940422534942627, + "learning_rate": 1.3469865067466268e-05, + "loss": 2.4799915313720704, + "step": 120280 + }, + { + "epoch": 0.3274806043282973, + "grad_norm": 2.1653895378112793, + "learning_rate": 1.3468774703557312e-05, + "loss": 2.519598960876465, + "step": 120300 + }, + { + "epoch": 0.3275350483190418, + "grad_norm": 2.1871674060821533, + "learning_rate": 1.346768433964836e-05, + "loss": 2.520265579223633, + "step": 120320 + }, + { + "epoch": 0.3275894923097863, + "grad_norm": 2.1353602409362793, + "learning_rate": 1.3466593975739404e-05, + "loss": 2.5107418060302735, + "step": 120340 + }, + { + "epoch": 0.32764393630053085, + "grad_norm": 2.450554609298706, + "learning_rate": 1.3465503611830451e-05, + "loss": 2.5071468353271484, + "step": 120360 + }, + { + "epoch": 0.32769838029127535, + "grad_norm": 2.1097328662872314, + "learning_rate": 1.3464413247921495e-05, + "loss": 2.5329521179199217, + "step": 120380 + }, + { + "epoch": 0.32775282428201985, + "grad_norm": 13.48741626739502, + "learning_rate": 1.346332288401254e-05, + "loss": 2.492152214050293, + "step": 120400 + }, + { + "epoch": 0.3278072682727644, + "grad_norm": 2.254925012588501, + "learning_rate": 1.3462232520103587e-05, + "loss": 2.528284454345703, + "step": 120420 + }, + { + "epoch": 0.3278617122635089, + "grad_norm": 2.479999542236328, + "learning_rate": 1.3461142156194631e-05, + "loss": 2.5217144012451174, + "step": 120440 + }, + { + "epoch": 0.3279161562542534, + "grad_norm": 2.272597312927246, + "learning_rate": 1.3460051792285675e-05, + "loss": 2.4919437408447265, + "step": 120460 + }, + { + "epoch": 0.32797060024499797, + "grad_norm": 2.1275241374969482, + "learning_rate": 1.3458961428376723e-05, + "loss": 2.5145170211791994, + "step": 120480 + }, + { + "epoch": 0.32802504423574247, + "grad_norm": 2.132164716720581, + "learning_rate": 1.3457871064467767e-05, + "loss": 2.5058353424072264, + "step": 120500 + }, + { + "epoch": 0.328079488226487, + "grad_norm": 2.253439426422119, + "learning_rate": 1.3456780700558814e-05, + "loss": 2.501479911804199, + "step": 120520 + }, + { + "epoch": 0.3281339322172315, + "grad_norm": 2.1590464115142822, + "learning_rate": 1.3455690336649858e-05, + "loss": 2.4974428176879884, + "step": 120540 + }, + { + "epoch": 0.32818837620797603, + "grad_norm": 2.087240695953369, + "learning_rate": 1.3454599972740902e-05, + "loss": 2.480768013000488, + "step": 120560 + }, + { + "epoch": 0.3282428201987206, + "grad_norm": 2.140200138092041, + "learning_rate": 1.345350960883195e-05, + "loss": 2.5150272369384767, + "step": 120580 + }, + { + "epoch": 0.3282972641894651, + "grad_norm": 2.170106887817383, + "learning_rate": 1.3452419244922994e-05, + "loss": 2.5063196182250977, + "step": 120600 + }, + { + "epoch": 0.3283517081802096, + "grad_norm": 2.1915183067321777, + "learning_rate": 1.3451328881014038e-05, + "loss": 2.499808502197266, + "step": 120620 + }, + { + "epoch": 0.32840615217095415, + "grad_norm": 2.1891660690307617, + "learning_rate": 1.3450238517105086e-05, + "loss": 2.50579891204834, + "step": 120640 + }, + { + "epoch": 0.32846059616169865, + "grad_norm": 2.1612985134124756, + "learning_rate": 1.344914815319613e-05, + "loss": 2.488071250915527, + "step": 120660 + }, + { + "epoch": 0.32851504015244315, + "grad_norm": 2.1794941425323486, + "learning_rate": 1.3448057789287177e-05, + "loss": 2.509182929992676, + "step": 120680 + }, + { + "epoch": 0.3285694841431877, + "grad_norm": 2.126723051071167, + "learning_rate": 1.3446967425378221e-05, + "loss": 2.4855812072753904, + "step": 120700 + }, + { + "epoch": 0.3286239281339322, + "grad_norm": 2.052490711212158, + "learning_rate": 1.3445877061469266e-05, + "loss": 2.4857322692871096, + "step": 120720 + }, + { + "epoch": 0.32867837212467677, + "grad_norm": 2.0796618461608887, + "learning_rate": 1.3444786697560313e-05, + "loss": 2.510847473144531, + "step": 120740 + }, + { + "epoch": 0.32873281611542127, + "grad_norm": 6.822824001312256, + "learning_rate": 1.3443696333651357e-05, + "loss": 2.51845703125, + "step": 120760 + }, + { + "epoch": 0.32878726010616577, + "grad_norm": 2.0530612468719482, + "learning_rate": 1.3442605969742401e-05, + "loss": 2.4793699264526365, + "step": 120780 + }, + { + "epoch": 0.3288417040969103, + "grad_norm": 2.218430757522583, + "learning_rate": 1.3441515605833449e-05, + "loss": 2.4897882461547853, + "step": 120800 + }, + { + "epoch": 0.3288961480876548, + "grad_norm": 2.096837282180786, + "learning_rate": 1.3440425241924493e-05, + "loss": 2.5047693252563477, + "step": 120820 + }, + { + "epoch": 0.32895059207839933, + "grad_norm": 2.3610193729400635, + "learning_rate": 1.343933487801554e-05, + "loss": 2.4701217651367187, + "step": 120840 + }, + { + "epoch": 0.3290050360691439, + "grad_norm": 2.1431431770324707, + "learning_rate": 1.3438244514106585e-05, + "loss": 2.497734260559082, + "step": 120860 + }, + { + "epoch": 0.3290594800598884, + "grad_norm": 2.1965107917785645, + "learning_rate": 1.3437154150197629e-05, + "loss": 2.4948715209960937, + "step": 120880 + }, + { + "epoch": 0.3291139240506329, + "grad_norm": 2.0693459510803223, + "learning_rate": 1.3436063786288676e-05, + "loss": 2.485119438171387, + "step": 120900 + }, + { + "epoch": 0.32916836804137745, + "grad_norm": 2.1681973934173584, + "learning_rate": 1.343497342237972e-05, + "loss": 2.5058780670166017, + "step": 120920 + }, + { + "epoch": 0.32922281203212195, + "grad_norm": 2.1488637924194336, + "learning_rate": 1.3433883058470764e-05, + "loss": 2.510940933227539, + "step": 120940 + }, + { + "epoch": 0.32927725602286645, + "grad_norm": 2.1930999755859375, + "learning_rate": 1.3432792694561812e-05, + "loss": 2.521571922302246, + "step": 120960 + }, + { + "epoch": 0.329331700013611, + "grad_norm": 2.182818651199341, + "learning_rate": 1.3431702330652856e-05, + "loss": 2.503452682495117, + "step": 120980 + }, + { + "epoch": 0.3293861440043555, + "grad_norm": 2.130953073501587, + "learning_rate": 1.3430611966743904e-05, + "loss": 2.477859115600586, + "step": 121000 + }, + { + "epoch": 0.32944058799510006, + "grad_norm": 2.095353364944458, + "learning_rate": 1.3429521602834948e-05, + "loss": 2.499315071105957, + "step": 121020 + }, + { + "epoch": 0.32949503198584457, + "grad_norm": 2.1832141876220703, + "learning_rate": 1.3428431238925992e-05, + "loss": 2.487808609008789, + "step": 121040 + }, + { + "epoch": 0.32954947597658907, + "grad_norm": 2.647752046585083, + "learning_rate": 1.342734087501704e-05, + "loss": 2.497214698791504, + "step": 121060 + }, + { + "epoch": 0.3296039199673336, + "grad_norm": 2.0606024265289307, + "learning_rate": 1.3426250511108083e-05, + "loss": 2.4707881927490236, + "step": 121080 + }, + { + "epoch": 0.3296583639580781, + "grad_norm": 2.1418020725250244, + "learning_rate": 1.3425160147199127e-05, + "loss": 2.496290397644043, + "step": 121100 + }, + { + "epoch": 0.32971280794882263, + "grad_norm": 2.2376914024353027, + "learning_rate": 1.3424069783290175e-05, + "loss": 2.478612518310547, + "step": 121120 + }, + { + "epoch": 0.3297672519395672, + "grad_norm": 2.1428394317626953, + "learning_rate": 1.3422979419381219e-05, + "loss": 2.4992340087890623, + "step": 121140 + }, + { + "epoch": 0.3298216959303117, + "grad_norm": 2.1116995811462402, + "learning_rate": 1.3421889055472267e-05, + "loss": 2.50780029296875, + "step": 121160 + }, + { + "epoch": 0.3298761399210562, + "grad_norm": 2.7084248065948486, + "learning_rate": 1.342079869156331e-05, + "loss": 2.502909851074219, + "step": 121180 + }, + { + "epoch": 0.32993058391180075, + "grad_norm": 2.1754419803619385, + "learning_rate": 1.3419708327654355e-05, + "loss": 2.5123363494873048, + "step": 121200 + }, + { + "epoch": 0.32998502790254525, + "grad_norm": 2.20961332321167, + "learning_rate": 1.3418617963745402e-05, + "loss": 2.488876724243164, + "step": 121220 + }, + { + "epoch": 0.3300394718932898, + "grad_norm": 2.1727263927459717, + "learning_rate": 1.3417527599836446e-05, + "loss": 2.5150381088256837, + "step": 121240 + }, + { + "epoch": 0.3300939158840343, + "grad_norm": 2.1424593925476074, + "learning_rate": 1.341643723592749e-05, + "loss": 2.466709518432617, + "step": 121260 + }, + { + "epoch": 0.3301483598747788, + "grad_norm": 2.1425211429595947, + "learning_rate": 1.3415346872018538e-05, + "loss": 2.4984619140625, + "step": 121280 + }, + { + "epoch": 0.33020280386552336, + "grad_norm": 2.1563522815704346, + "learning_rate": 1.3414256508109582e-05, + "loss": 2.5056280136108398, + "step": 121300 + }, + { + "epoch": 0.33025724785626787, + "grad_norm": 2.1271471977233887, + "learning_rate": 1.341316614420063e-05, + "loss": 2.4975379943847655, + "step": 121320 + }, + { + "epoch": 0.33031169184701237, + "grad_norm": 2.095684766769409, + "learning_rate": 1.3412075780291674e-05, + "loss": 2.494437026977539, + "step": 121340 + }, + { + "epoch": 0.3303661358377569, + "grad_norm": 2.1234254837036133, + "learning_rate": 1.3410985416382718e-05, + "loss": 2.540948486328125, + "step": 121360 + }, + { + "epoch": 0.3304205798285014, + "grad_norm": 2.298919200897217, + "learning_rate": 1.3409895052473765e-05, + "loss": 2.5155420303344727, + "step": 121380 + }, + { + "epoch": 0.3304750238192459, + "grad_norm": 3.1157476902008057, + "learning_rate": 1.340880468856481e-05, + "loss": 2.4965686798095703, + "step": 121400 + }, + { + "epoch": 0.3305294678099905, + "grad_norm": 2.0614755153656006, + "learning_rate": 1.3407714324655855e-05, + "loss": 2.4909019470214844, + "step": 121420 + }, + { + "epoch": 0.330583911800735, + "grad_norm": 2.103416681289673, + "learning_rate": 1.3406623960746901e-05, + "loss": 2.535167121887207, + "step": 121440 + }, + { + "epoch": 0.3306383557914795, + "grad_norm": 2.2418205738067627, + "learning_rate": 1.3405533596837945e-05, + "loss": 2.510609245300293, + "step": 121460 + }, + { + "epoch": 0.33069279978222405, + "grad_norm": 2.096442699432373, + "learning_rate": 1.3404443232928993e-05, + "loss": 2.481074333190918, + "step": 121480 + }, + { + "epoch": 0.33074724377296855, + "grad_norm": 2.180271863937378, + "learning_rate": 1.3403352869020037e-05, + "loss": 2.489410400390625, + "step": 121500 + }, + { + "epoch": 0.3308016877637131, + "grad_norm": 2.114969253540039, + "learning_rate": 1.3402262505111081e-05, + "loss": 2.527559280395508, + "step": 121520 + }, + { + "epoch": 0.3308561317544576, + "grad_norm": 2.19576358795166, + "learning_rate": 1.3401172141202128e-05, + "loss": 2.51053409576416, + "step": 121540 + }, + { + "epoch": 0.3309105757452021, + "grad_norm": 2.1459221839904785, + "learning_rate": 1.3400081777293173e-05, + "loss": 2.502473831176758, + "step": 121560 + }, + { + "epoch": 0.33096501973594666, + "grad_norm": 2.245842456817627, + "learning_rate": 1.3398991413384218e-05, + "loss": 2.526492881774902, + "step": 121580 + }, + { + "epoch": 0.33101946372669117, + "grad_norm": 2.1592001914978027, + "learning_rate": 1.3397901049475264e-05, + "loss": 2.47800350189209, + "step": 121600 + }, + { + "epoch": 0.33107390771743567, + "grad_norm": 2.4166383743286133, + "learning_rate": 1.3396810685566308e-05, + "loss": 2.455204200744629, + "step": 121620 + }, + { + "epoch": 0.3311283517081802, + "grad_norm": 2.0962657928466797, + "learning_rate": 1.3395720321657356e-05, + "loss": 2.4923648834228516, + "step": 121640 + }, + { + "epoch": 0.3311827956989247, + "grad_norm": 2.1713669300079346, + "learning_rate": 1.33946299577484e-05, + "loss": 2.492353820800781, + "step": 121660 + }, + { + "epoch": 0.3312372396896692, + "grad_norm": 2.050718307495117, + "learning_rate": 1.3393539593839444e-05, + "loss": 2.510294532775879, + "step": 121680 + }, + { + "epoch": 0.3312916836804138, + "grad_norm": 2.192640781402588, + "learning_rate": 1.3392449229930492e-05, + "loss": 2.481650733947754, + "step": 121700 + }, + { + "epoch": 0.3313461276711583, + "grad_norm": 2.187696933746338, + "learning_rate": 1.3391358866021536e-05, + "loss": 2.5066593170166014, + "step": 121720 + }, + { + "epoch": 0.33140057166190284, + "grad_norm": 2.1172070503234863, + "learning_rate": 1.3390268502112582e-05, + "loss": 2.513731575012207, + "step": 121740 + }, + { + "epoch": 0.33145501565264734, + "grad_norm": 2.0154964923858643, + "learning_rate": 1.3389178138203627e-05, + "loss": 2.510021781921387, + "step": 121760 + }, + { + "epoch": 0.33150945964339185, + "grad_norm": 2.000915288925171, + "learning_rate": 1.3388087774294671e-05, + "loss": 2.507603645324707, + "step": 121780 + }, + { + "epoch": 0.3315639036341364, + "grad_norm": 2.1516082286834717, + "learning_rate": 1.3386997410385719e-05, + "loss": 2.5033897399902343, + "step": 121800 + }, + { + "epoch": 0.3316183476248809, + "grad_norm": 2.026869773864746, + "learning_rate": 1.3385907046476763e-05, + "loss": 2.5092544555664062, + "step": 121820 + }, + { + "epoch": 0.3316727916156254, + "grad_norm": 2.074017286300659, + "learning_rate": 1.3384816682567807e-05, + "loss": 2.498648452758789, + "step": 121840 + }, + { + "epoch": 0.33172723560636996, + "grad_norm": 2.2066285610198975, + "learning_rate": 1.3383726318658855e-05, + "loss": 2.4763633728027346, + "step": 121860 + }, + { + "epoch": 0.33178167959711446, + "grad_norm": 2.2499985694885254, + "learning_rate": 1.3382635954749899e-05, + "loss": 2.486695098876953, + "step": 121880 + }, + { + "epoch": 0.33183612358785897, + "grad_norm": 2.1366772651672363, + "learning_rate": 1.3381545590840945e-05, + "loss": 2.498469352722168, + "step": 121900 + }, + { + "epoch": 0.3318905675786035, + "grad_norm": 2.4325802326202393, + "learning_rate": 1.338045522693199e-05, + "loss": 2.480428123474121, + "step": 121920 + }, + { + "epoch": 0.331945011569348, + "grad_norm": 2.1531026363372803, + "learning_rate": 1.3379364863023035e-05, + "loss": 2.52390193939209, + "step": 121940 + }, + { + "epoch": 0.3319994555600926, + "grad_norm": 2.066744565963745, + "learning_rate": 1.337827449911408e-05, + "loss": 2.5327228546142577, + "step": 121960 + }, + { + "epoch": 0.3320538995508371, + "grad_norm": 2.1214845180511475, + "learning_rate": 1.3377184135205126e-05, + "loss": 2.5016727447509766, + "step": 121980 + }, + { + "epoch": 0.3321083435415816, + "grad_norm": 2.078883409500122, + "learning_rate": 1.337609377129617e-05, + "loss": 2.480670928955078, + "step": 122000 + }, + { + "epoch": 0.33216278753232614, + "grad_norm": 2.132380962371826, + "learning_rate": 1.3375003407387218e-05, + "loss": 2.5115060806274414, + "step": 122020 + }, + { + "epoch": 0.33221723152307064, + "grad_norm": 2.1734468936920166, + "learning_rate": 1.3373913043478262e-05, + "loss": 2.5498125076293947, + "step": 122040 + }, + { + "epoch": 0.33227167551381515, + "grad_norm": 2.2910239696502686, + "learning_rate": 1.3372822679569308e-05, + "loss": 2.493988609313965, + "step": 122060 + }, + { + "epoch": 0.3323261195045597, + "grad_norm": 2.4894206523895264, + "learning_rate": 1.3371732315660353e-05, + "loss": 2.492569351196289, + "step": 122080 + }, + { + "epoch": 0.3323805634953042, + "grad_norm": 2.187946319580078, + "learning_rate": 1.3370641951751398e-05, + "loss": 2.4875255584716798, + "step": 122100 + }, + { + "epoch": 0.3324350074860487, + "grad_norm": 5.937261581420898, + "learning_rate": 1.3369551587842443e-05, + "loss": 2.4949983596801757, + "step": 122120 + }, + { + "epoch": 0.33248945147679326, + "grad_norm": 2.217695713043213, + "learning_rate": 1.336846122393349e-05, + "loss": 2.5085248947143555, + "step": 122140 + }, + { + "epoch": 0.33254389546753776, + "grad_norm": 2.1612751483917236, + "learning_rate": 1.3367370860024533e-05, + "loss": 2.5191013336181642, + "step": 122160 + }, + { + "epoch": 0.33259833945828227, + "grad_norm": 2.1079976558685303, + "learning_rate": 1.336628049611558e-05, + "loss": 2.4806528091430664, + "step": 122180 + }, + { + "epoch": 0.3326527834490268, + "grad_norm": 2.1539931297302246, + "learning_rate": 1.3365190132206625e-05, + "loss": 2.486923408508301, + "step": 122200 + }, + { + "epoch": 0.3327072274397713, + "grad_norm": 2.201049566268921, + "learning_rate": 1.336409976829767e-05, + "loss": 2.5123868942260743, + "step": 122220 + }, + { + "epoch": 0.3327616714305159, + "grad_norm": 2.208726644515991, + "learning_rate": 1.3363009404388717e-05, + "loss": 2.5072038650512694, + "step": 122240 + }, + { + "epoch": 0.3328161154212604, + "grad_norm": 2.127920150756836, + "learning_rate": 1.336191904047976e-05, + "loss": 2.4783632278442385, + "step": 122260 + }, + { + "epoch": 0.3328705594120049, + "grad_norm": 2.3370985984802246, + "learning_rate": 1.3360828676570806e-05, + "loss": 2.519562530517578, + "step": 122280 + }, + { + "epoch": 0.33292500340274944, + "grad_norm": 2.120659828186035, + "learning_rate": 1.3359738312661852e-05, + "loss": 2.5060003280639647, + "step": 122300 + }, + { + "epoch": 0.33297944739349394, + "grad_norm": 2.211174249649048, + "learning_rate": 1.3358647948752896e-05, + "loss": 2.4827733993530274, + "step": 122320 + }, + { + "epoch": 0.33303389138423845, + "grad_norm": 2.1446378231048584, + "learning_rate": 1.3357557584843944e-05, + "loss": 2.5126569747924803, + "step": 122340 + }, + { + "epoch": 0.333088335374983, + "grad_norm": 2.174543857574463, + "learning_rate": 1.3356467220934988e-05, + "loss": 2.4808225631713867, + "step": 122360 + }, + { + "epoch": 0.3331427793657275, + "grad_norm": 2.126439332962036, + "learning_rate": 1.3355376857026034e-05, + "loss": 2.491580581665039, + "step": 122380 + }, + { + "epoch": 0.333197223356472, + "grad_norm": 2.2943673133850098, + "learning_rate": 1.335428649311708e-05, + "loss": 2.4673025131225588, + "step": 122400 + }, + { + "epoch": 0.33325166734721656, + "grad_norm": 2.0589547157287598, + "learning_rate": 1.3353196129208124e-05, + "loss": 2.479053497314453, + "step": 122420 + }, + { + "epoch": 0.33330611133796106, + "grad_norm": 2.1359329223632812, + "learning_rate": 1.335210576529917e-05, + "loss": 2.4919483184814455, + "step": 122440 + }, + { + "epoch": 0.3333605553287056, + "grad_norm": 2.3174445629119873, + "learning_rate": 1.3351015401390215e-05, + "loss": 2.4910823822021486, + "step": 122460 + }, + { + "epoch": 0.3334149993194501, + "grad_norm": 2.287792682647705, + "learning_rate": 1.334992503748126e-05, + "loss": 2.469812202453613, + "step": 122480 + }, + { + "epoch": 0.3334694433101946, + "grad_norm": 2.0545318126678467, + "learning_rate": 1.3348834673572307e-05, + "loss": 2.4856245040893556, + "step": 122500 + }, + { + "epoch": 0.3335238873009392, + "grad_norm": 2.1865642070770264, + "learning_rate": 1.3347744309663351e-05, + "loss": 2.4906763076782226, + "step": 122520 + }, + { + "epoch": 0.3335783312916837, + "grad_norm": 2.1610021591186523, + "learning_rate": 1.3346653945754397e-05, + "loss": 2.5098167419433595, + "step": 122540 + }, + { + "epoch": 0.3336327752824282, + "grad_norm": 2.112555503845215, + "learning_rate": 1.3345563581845443e-05, + "loss": 2.5004310607910156, + "step": 122560 + }, + { + "epoch": 0.33368721927317274, + "grad_norm": 2.4782724380493164, + "learning_rate": 1.3344473217936487e-05, + "loss": 2.506394958496094, + "step": 122580 + }, + { + "epoch": 0.33374166326391724, + "grad_norm": 3.6179990768432617, + "learning_rate": 1.3343382854027533e-05, + "loss": 2.5080886840820313, + "step": 122600 + }, + { + "epoch": 0.33379610725466174, + "grad_norm": 2.1541755199432373, + "learning_rate": 1.3342292490118578e-05, + "loss": 2.4801307678222657, + "step": 122620 + }, + { + "epoch": 0.3338505512454063, + "grad_norm": 2.2281455993652344, + "learning_rate": 1.3341202126209623e-05, + "loss": 2.487341117858887, + "step": 122640 + }, + { + "epoch": 0.3339049952361508, + "grad_norm": 2.1644034385681152, + "learning_rate": 1.334011176230067e-05, + "loss": 2.5102746963500975, + "step": 122660 + }, + { + "epoch": 0.3339594392268953, + "grad_norm": 2.148998498916626, + "learning_rate": 1.3339021398391714e-05, + "loss": 2.5221689224243162, + "step": 122680 + }, + { + "epoch": 0.33401388321763986, + "grad_norm": 3.4742159843444824, + "learning_rate": 1.333793103448276e-05, + "loss": 2.484485626220703, + "step": 122700 + }, + { + "epoch": 0.33406832720838436, + "grad_norm": 2.177849769592285, + "learning_rate": 1.3336840670573806e-05, + "loss": 2.4944957733154296, + "step": 122720 + }, + { + "epoch": 0.3341227711991289, + "grad_norm": 2.1689295768737793, + "learning_rate": 1.333575030666485e-05, + "loss": 2.4985727310180663, + "step": 122740 + }, + { + "epoch": 0.3341772151898734, + "grad_norm": 2.182605504989624, + "learning_rate": 1.3334659942755896e-05, + "loss": 2.498438262939453, + "step": 122760 + }, + { + "epoch": 0.3342316591806179, + "grad_norm": 2.233051061630249, + "learning_rate": 1.3333569578846942e-05, + "loss": 2.4925151824951173, + "step": 122780 + }, + { + "epoch": 0.3342861031713625, + "grad_norm": 2.15786075592041, + "learning_rate": 1.3332479214937986e-05, + "loss": 2.490203857421875, + "step": 122800 + }, + { + "epoch": 0.334340547162107, + "grad_norm": 2.159956455230713, + "learning_rate": 1.3331388851029033e-05, + "loss": 2.479772758483887, + "step": 122820 + }, + { + "epoch": 0.3343949911528515, + "grad_norm": 2.1819427013397217, + "learning_rate": 1.3330298487120077e-05, + "loss": 2.4907110214233397, + "step": 122840 + }, + { + "epoch": 0.33444943514359604, + "grad_norm": 2.1296565532684326, + "learning_rate": 1.3329208123211123e-05, + "loss": 2.495486259460449, + "step": 122860 + }, + { + "epoch": 0.33450387913434054, + "grad_norm": 2.217468500137329, + "learning_rate": 1.3328117759302169e-05, + "loss": 2.488175392150879, + "step": 122880 + }, + { + "epoch": 0.33455832312508504, + "grad_norm": 2.110076904296875, + "learning_rate": 1.3327027395393213e-05, + "loss": 2.4655878067016603, + "step": 122900 + }, + { + "epoch": 0.3346127671158296, + "grad_norm": 2.147684335708618, + "learning_rate": 1.3325937031484259e-05, + "loss": 2.4936195373535157, + "step": 122920 + }, + { + "epoch": 0.3346672111065741, + "grad_norm": 2.187870979309082, + "learning_rate": 1.3324846667575305e-05, + "loss": 2.5123363494873048, + "step": 122940 + }, + { + "epoch": 0.33472165509731866, + "grad_norm": 2.048137664794922, + "learning_rate": 1.3323756303666349e-05, + "loss": 2.5178348541259767, + "step": 122960 + }, + { + "epoch": 0.33477609908806316, + "grad_norm": 2.1290712356567383, + "learning_rate": 1.3322665939757395e-05, + "loss": 2.5052576065063477, + "step": 122980 + }, + { + "epoch": 0.33483054307880766, + "grad_norm": 2.21589732170105, + "learning_rate": 1.332157557584844e-05, + "loss": 2.5110942840576174, + "step": 123000 + }, + { + "epoch": 0.3348849870695522, + "grad_norm": 2.2295193672180176, + "learning_rate": 1.3320485211939486e-05, + "loss": 2.520947074890137, + "step": 123020 + }, + { + "epoch": 0.3349394310602967, + "grad_norm": 2.1672463417053223, + "learning_rate": 1.3319394848030532e-05, + "loss": 2.4653377532958984, + "step": 123040 + }, + { + "epoch": 0.3349938750510412, + "grad_norm": 2.1192283630371094, + "learning_rate": 1.3318304484121576e-05, + "loss": 2.4869110107421877, + "step": 123060 + }, + { + "epoch": 0.3350483190417858, + "grad_norm": 2.0985515117645264, + "learning_rate": 1.3317214120212622e-05, + "loss": 2.494535446166992, + "step": 123080 + }, + { + "epoch": 0.3351027630325303, + "grad_norm": 2.1191041469573975, + "learning_rate": 1.3316123756303668e-05, + "loss": 2.4715269088745115, + "step": 123100 + }, + { + "epoch": 0.3351572070232748, + "grad_norm": 2.1107842922210693, + "learning_rate": 1.3315033392394714e-05, + "loss": 2.5179121017456056, + "step": 123120 + }, + { + "epoch": 0.33521165101401934, + "grad_norm": 2.6312172412872314, + "learning_rate": 1.3313943028485758e-05, + "loss": 2.4621191024780273, + "step": 123140 + }, + { + "epoch": 0.33526609500476384, + "grad_norm": 2.9215855598449707, + "learning_rate": 1.3312852664576803e-05, + "loss": 2.5172348022460938, + "step": 123160 + }, + { + "epoch": 0.33532053899550834, + "grad_norm": 2.2108752727508545, + "learning_rate": 1.331176230066785e-05, + "loss": 2.4985170364379883, + "step": 123180 + }, + { + "epoch": 0.3353749829862529, + "grad_norm": 2.127443552017212, + "learning_rate": 1.3310671936758895e-05, + "loss": 2.515330696105957, + "step": 123200 + }, + { + "epoch": 0.3354294269769974, + "grad_norm": 2.0536513328552246, + "learning_rate": 1.330958157284994e-05, + "loss": 2.533213424682617, + "step": 123220 + }, + { + "epoch": 0.33548387096774196, + "grad_norm": 2.281528949737549, + "learning_rate": 1.3308491208940985e-05, + "loss": 2.5162891387939452, + "step": 123240 + }, + { + "epoch": 0.33553831495848646, + "grad_norm": 2.062988519668579, + "learning_rate": 1.330740084503203e-05, + "loss": 2.4971542358398438, + "step": 123260 + }, + { + "epoch": 0.33559275894923096, + "grad_norm": 2.1333906650543213, + "learning_rate": 1.3306310481123077e-05, + "loss": 2.4934665679931642, + "step": 123280 + }, + { + "epoch": 0.3356472029399755, + "grad_norm": 2.1027188301086426, + "learning_rate": 1.330522011721412e-05, + "loss": 2.4837440490722655, + "step": 123300 + }, + { + "epoch": 0.33570164693072, + "grad_norm": 2.1387600898742676, + "learning_rate": 1.3304129753305167e-05, + "loss": 2.5011178970336916, + "step": 123320 + }, + { + "epoch": 0.3357560909214645, + "grad_norm": 2.171677350997925, + "learning_rate": 1.3303039389396212e-05, + "loss": 2.5356109619140623, + "step": 123340 + }, + { + "epoch": 0.3358105349122091, + "grad_norm": 2.2010996341705322, + "learning_rate": 1.3301949025487258e-05, + "loss": 2.5064813613891603, + "step": 123360 + }, + { + "epoch": 0.3358649789029536, + "grad_norm": 2.2059435844421387, + "learning_rate": 1.3300858661578302e-05, + "loss": 2.5284833908081055, + "step": 123380 + }, + { + "epoch": 0.3359194228936981, + "grad_norm": 2.1911494731903076, + "learning_rate": 1.3299768297669348e-05, + "loss": 2.481036376953125, + "step": 123400 + }, + { + "epoch": 0.33597386688444264, + "grad_norm": 2.1343092918395996, + "learning_rate": 1.3298677933760394e-05, + "loss": 2.510474967956543, + "step": 123420 + }, + { + "epoch": 0.33602831087518714, + "grad_norm": 2.1797876358032227, + "learning_rate": 1.329758756985144e-05, + "loss": 2.494423675537109, + "step": 123440 + }, + { + "epoch": 0.3360827548659317, + "grad_norm": 2.0188887119293213, + "learning_rate": 1.3296497205942484e-05, + "loss": 2.493934440612793, + "step": 123460 + }, + { + "epoch": 0.3361371988566762, + "grad_norm": 2.1325905323028564, + "learning_rate": 1.329540684203353e-05, + "loss": 2.50228271484375, + "step": 123480 + }, + { + "epoch": 0.3361916428474207, + "grad_norm": 2.13478684425354, + "learning_rate": 1.3294316478124575e-05, + "loss": 2.4907964706420898, + "step": 123500 + }, + { + "epoch": 0.33624608683816526, + "grad_norm": 2.196514129638672, + "learning_rate": 1.3293226114215621e-05, + "loss": 2.4863142013549804, + "step": 123520 + }, + { + "epoch": 0.33630053082890976, + "grad_norm": 2.1383347511291504, + "learning_rate": 1.3292135750306665e-05, + "loss": 2.465806579589844, + "step": 123540 + }, + { + "epoch": 0.33635497481965426, + "grad_norm": 2.101846218109131, + "learning_rate": 1.3291045386397711e-05, + "loss": 2.4954833984375, + "step": 123560 + }, + { + "epoch": 0.3364094188103988, + "grad_norm": 2.1248717308044434, + "learning_rate": 1.3289955022488757e-05, + "loss": 2.4851097106933593, + "step": 123580 + }, + { + "epoch": 0.3364638628011433, + "grad_norm": 2.134201765060425, + "learning_rate": 1.3288864658579803e-05, + "loss": 2.4841840744018553, + "step": 123600 + }, + { + "epoch": 0.3365183067918878, + "grad_norm": 2.1189072132110596, + "learning_rate": 1.3287774294670847e-05, + "loss": 2.528579902648926, + "step": 123620 + }, + { + "epoch": 0.3365727507826324, + "grad_norm": 2.1938788890838623, + "learning_rate": 1.3286683930761893e-05, + "loss": 2.5048118591308595, + "step": 123640 + }, + { + "epoch": 0.3366271947733769, + "grad_norm": 2.08158278465271, + "learning_rate": 1.3285593566852939e-05, + "loss": 2.515591049194336, + "step": 123660 + }, + { + "epoch": 0.3366816387641214, + "grad_norm": 2.268558979034424, + "learning_rate": 1.3284503202943984e-05, + "loss": 2.5008010864257812, + "step": 123680 + }, + { + "epoch": 0.33673608275486594, + "grad_norm": 2.081061601638794, + "learning_rate": 1.3283412839035028e-05, + "loss": 2.4898910522460938, + "step": 123700 + }, + { + "epoch": 0.33679052674561044, + "grad_norm": 2.144117832183838, + "learning_rate": 1.3282322475126074e-05, + "loss": 2.493634796142578, + "step": 123720 + }, + { + "epoch": 0.336844970736355, + "grad_norm": 2.556129217147827, + "learning_rate": 1.328123211121712e-05, + "loss": 2.5188623428344727, + "step": 123740 + }, + { + "epoch": 0.3368994147270995, + "grad_norm": 2.193004846572876, + "learning_rate": 1.3280141747308166e-05, + "loss": 2.5143070220947266, + "step": 123760 + }, + { + "epoch": 0.336953858717844, + "grad_norm": 2.0735552310943604, + "learning_rate": 1.327905138339921e-05, + "loss": 2.470348358154297, + "step": 123780 + }, + { + "epoch": 0.33700830270858856, + "grad_norm": 2.1747682094573975, + "learning_rate": 1.3277961019490256e-05, + "loss": 2.4767807006835936, + "step": 123800 + }, + { + "epoch": 0.33706274669933306, + "grad_norm": 2.1526358127593994, + "learning_rate": 1.3276870655581302e-05, + "loss": 2.4896093368530274, + "step": 123820 + }, + { + "epoch": 0.33711719069007756, + "grad_norm": 2.1647286415100098, + "learning_rate": 1.3275780291672347e-05, + "loss": 2.4958589553833006, + "step": 123840 + }, + { + "epoch": 0.3371716346808221, + "grad_norm": 2.180436611175537, + "learning_rate": 1.3274689927763392e-05, + "loss": 2.4681753158569335, + "step": 123860 + }, + { + "epoch": 0.3372260786715666, + "grad_norm": 2.202482223510742, + "learning_rate": 1.3273599563854437e-05, + "loss": 2.478010559082031, + "step": 123880 + }, + { + "epoch": 0.3372805226623111, + "grad_norm": 2.127933979034424, + "learning_rate": 1.3272509199945483e-05, + "loss": 2.471124267578125, + "step": 123900 + }, + { + "epoch": 0.3373349666530557, + "grad_norm": 2.9437766075134277, + "learning_rate": 1.3271418836036529e-05, + "loss": 2.475577163696289, + "step": 123920 + }, + { + "epoch": 0.3373894106438002, + "grad_norm": 2.0974104404449463, + "learning_rate": 1.3270328472127573e-05, + "loss": 2.4871450424194337, + "step": 123940 + }, + { + "epoch": 0.33744385463454474, + "grad_norm": 2.0545849800109863, + "learning_rate": 1.3269238108218619e-05, + "loss": 2.507870674133301, + "step": 123960 + }, + { + "epoch": 0.33749829862528924, + "grad_norm": 2.3196873664855957, + "learning_rate": 1.3268147744309665e-05, + "loss": 2.476169776916504, + "step": 123980 + }, + { + "epoch": 0.33755274261603374, + "grad_norm": 2.5215611457824707, + "learning_rate": 1.326705738040071e-05, + "loss": 2.485591697692871, + "step": 124000 + }, + { + "epoch": 0.3376071866067783, + "grad_norm": 2.158139705657959, + "learning_rate": 1.3265967016491755e-05, + "loss": 2.4788209915161135, + "step": 124020 + }, + { + "epoch": 0.3376616305975228, + "grad_norm": 2.029977798461914, + "learning_rate": 1.32648766525828e-05, + "loss": 2.5143505096435548, + "step": 124040 + }, + { + "epoch": 0.3377160745882673, + "grad_norm": 3.2869198322296143, + "learning_rate": 1.3263786288673846e-05, + "loss": 2.513194274902344, + "step": 124060 + }, + { + "epoch": 0.33777051857901186, + "grad_norm": 2.1442601680755615, + "learning_rate": 1.3262695924764892e-05, + "loss": 2.5016368865966796, + "step": 124080 + }, + { + "epoch": 0.33782496256975636, + "grad_norm": 2.1535048484802246, + "learning_rate": 1.3261605560855936e-05, + "loss": 2.4933080673217773, + "step": 124100 + }, + { + "epoch": 0.33787940656050086, + "grad_norm": 2.136235475540161, + "learning_rate": 1.3260515196946982e-05, + "loss": 2.4815418243408205, + "step": 124120 + }, + { + "epoch": 0.3379338505512454, + "grad_norm": 2.093992233276367, + "learning_rate": 1.3259424833038028e-05, + "loss": 2.501388740539551, + "step": 124140 + }, + { + "epoch": 0.3379882945419899, + "grad_norm": 2.557337999343872, + "learning_rate": 1.3258334469129072e-05, + "loss": 2.5285966873168944, + "step": 124160 + }, + { + "epoch": 0.3380427385327345, + "grad_norm": 2.1718151569366455, + "learning_rate": 1.3257244105220118e-05, + "loss": 2.5062225341796873, + "step": 124180 + }, + { + "epoch": 0.338097182523479, + "grad_norm": 2.1256871223449707, + "learning_rate": 1.3256153741311164e-05, + "loss": 2.4990875244140627, + "step": 124200 + }, + { + "epoch": 0.3381516265142235, + "grad_norm": 2.070241928100586, + "learning_rate": 1.325506337740221e-05, + "loss": 2.5166126251220704, + "step": 124220 + }, + { + "epoch": 0.33820607050496804, + "grad_norm": 2.192976951599121, + "learning_rate": 1.3253973013493255e-05, + "loss": 2.4986557006835937, + "step": 124240 + }, + { + "epoch": 0.33826051449571254, + "grad_norm": 2.1087934970855713, + "learning_rate": 1.32528826495843e-05, + "loss": 2.5102518081665037, + "step": 124260 + }, + { + "epoch": 0.33831495848645704, + "grad_norm": 2.772005558013916, + "learning_rate": 1.3251792285675345e-05, + "loss": 2.510026550292969, + "step": 124280 + }, + { + "epoch": 0.3383694024772016, + "grad_norm": 2.1212077140808105, + "learning_rate": 1.3250701921766391e-05, + "loss": 2.5059141159057616, + "step": 124300 + }, + { + "epoch": 0.3384238464679461, + "grad_norm": 2.0728394985198975, + "learning_rate": 1.3249611557857435e-05, + "loss": 2.5071908950805666, + "step": 124320 + }, + { + "epoch": 0.3384782904586906, + "grad_norm": 2.1095659732818604, + "learning_rate": 1.324852119394848e-05, + "loss": 2.481209564208984, + "step": 124340 + }, + { + "epoch": 0.33853273444943516, + "grad_norm": 2.2926859855651855, + "learning_rate": 1.3247430830039527e-05, + "loss": 2.513774871826172, + "step": 124360 + }, + { + "epoch": 0.33858717844017966, + "grad_norm": 2.2007992267608643, + "learning_rate": 1.3246340466130572e-05, + "loss": 2.443976402282715, + "step": 124380 + }, + { + "epoch": 0.33864162243092416, + "grad_norm": 2.159773588180542, + "learning_rate": 1.3245250102221618e-05, + "loss": 2.54205436706543, + "step": 124400 + }, + { + "epoch": 0.3386960664216687, + "grad_norm": 2.209738254547119, + "learning_rate": 1.3244159738312662e-05, + "loss": 2.488681983947754, + "step": 124420 + }, + { + "epoch": 0.3387505104124132, + "grad_norm": 2.273545980453491, + "learning_rate": 1.3243069374403708e-05, + "loss": 2.505777931213379, + "step": 124440 + }, + { + "epoch": 0.3388049544031578, + "grad_norm": 2.204929828643799, + "learning_rate": 1.3241979010494754e-05, + "loss": 2.5121795654296877, + "step": 124460 + }, + { + "epoch": 0.3388593983939023, + "grad_norm": 2.262814998626709, + "learning_rate": 1.3240888646585798e-05, + "loss": 2.495692825317383, + "step": 124480 + }, + { + "epoch": 0.3389138423846468, + "grad_norm": 2.064587354660034, + "learning_rate": 1.3239798282676844e-05, + "loss": 2.5153921127319334, + "step": 124500 + }, + { + "epoch": 0.33896828637539134, + "grad_norm": 2.0976157188415527, + "learning_rate": 1.323870791876789e-05, + "loss": 2.463353729248047, + "step": 124520 + }, + { + "epoch": 0.33902273036613584, + "grad_norm": 2.3375189304351807, + "learning_rate": 1.3237617554858936e-05, + "loss": 2.4673471450805664, + "step": 124540 + }, + { + "epoch": 0.33907717435688034, + "grad_norm": 2.272433042526245, + "learning_rate": 1.3236527190949981e-05, + "loss": 2.536631393432617, + "step": 124560 + }, + { + "epoch": 0.3391316183476249, + "grad_norm": 2.1782596111297607, + "learning_rate": 1.3235436827041025e-05, + "loss": 2.482925033569336, + "step": 124580 + }, + { + "epoch": 0.3391860623383694, + "grad_norm": 2.13798189163208, + "learning_rate": 1.3234346463132071e-05, + "loss": 2.5462011337280273, + "step": 124600 + }, + { + "epoch": 0.3392405063291139, + "grad_norm": 2.134608030319214, + "learning_rate": 1.3233256099223117e-05, + "loss": 2.5326108932495117, + "step": 124620 + }, + { + "epoch": 0.33929495031985846, + "grad_norm": 2.512446165084839, + "learning_rate": 1.3232165735314161e-05, + "loss": 2.5027372360229494, + "step": 124640 + }, + { + "epoch": 0.33934939431060296, + "grad_norm": 2.1814801692962646, + "learning_rate": 1.3231075371405207e-05, + "loss": 2.4852375030517577, + "step": 124660 + }, + { + "epoch": 0.3394038383013475, + "grad_norm": 2.238424777984619, + "learning_rate": 1.3229985007496253e-05, + "loss": 2.489977264404297, + "step": 124680 + }, + { + "epoch": 0.339458282292092, + "grad_norm": 2.2248594760894775, + "learning_rate": 1.3228894643587299e-05, + "loss": 2.4863153457641602, + "step": 124700 + }, + { + "epoch": 0.3395127262828365, + "grad_norm": 2.149578332901001, + "learning_rate": 1.3227804279678344e-05, + "loss": 2.4922006607055662, + "step": 124720 + }, + { + "epoch": 0.3395671702735811, + "grad_norm": 2.2673332691192627, + "learning_rate": 1.3226713915769389e-05, + "loss": 2.481486511230469, + "step": 124740 + }, + { + "epoch": 0.3396216142643256, + "grad_norm": 2.0445494651794434, + "learning_rate": 1.3225623551860434e-05, + "loss": 2.532049369812012, + "step": 124760 + }, + { + "epoch": 0.3396760582550701, + "grad_norm": 2.0951619148254395, + "learning_rate": 1.322453318795148e-05, + "loss": 2.488226127624512, + "step": 124780 + }, + { + "epoch": 0.33973050224581464, + "grad_norm": 2.1317648887634277, + "learning_rate": 1.3223442824042524e-05, + "loss": 2.4730655670166017, + "step": 124800 + }, + { + "epoch": 0.33978494623655914, + "grad_norm": 4.285908222198486, + "learning_rate": 1.3222352460133572e-05, + "loss": 2.510947036743164, + "step": 124820 + }, + { + "epoch": 0.33983939022730364, + "grad_norm": 2.0791590213775635, + "learning_rate": 1.3221262096224616e-05, + "loss": 2.4906213760375975, + "step": 124840 + }, + { + "epoch": 0.3398938342180482, + "grad_norm": 2.1859726905822754, + "learning_rate": 1.3220171732315662e-05, + "loss": 2.5182754516601564, + "step": 124860 + }, + { + "epoch": 0.3399482782087927, + "grad_norm": 2.204308271408081, + "learning_rate": 1.3219081368406707e-05, + "loss": 2.5300565719604493, + "step": 124880 + }, + { + "epoch": 0.3400027221995372, + "grad_norm": 2.302579641342163, + "learning_rate": 1.3217991004497752e-05, + "loss": 2.489861488342285, + "step": 124900 + }, + { + "epoch": 0.34005716619028176, + "grad_norm": 2.196263551712036, + "learning_rate": 1.3216900640588797e-05, + "loss": 2.4857410430908202, + "step": 124920 + }, + { + "epoch": 0.34011161018102626, + "grad_norm": 2.123013734817505, + "learning_rate": 1.3215810276679843e-05, + "loss": 2.496860313415527, + "step": 124940 + }, + { + "epoch": 0.3401660541717708, + "grad_norm": 2.118157148361206, + "learning_rate": 1.3214719912770887e-05, + "loss": 2.4927513122558596, + "step": 124960 + }, + { + "epoch": 0.3402204981625153, + "grad_norm": 2.0884852409362793, + "learning_rate": 1.3213629548861935e-05, + "loss": 2.489518737792969, + "step": 124980 + }, + { + "epoch": 0.3402749421532598, + "grad_norm": 2.106214761734009, + "learning_rate": 1.3212539184952979e-05, + "loss": 2.467898941040039, + "step": 125000 + }, + { + "epoch": 0.3403293861440044, + "grad_norm": 2.275961399078369, + "learning_rate": 1.3211448821044025e-05, + "loss": 2.50323543548584, + "step": 125020 + }, + { + "epoch": 0.3403838301347489, + "grad_norm": 2.1129603385925293, + "learning_rate": 1.321035845713507e-05, + "loss": 2.48693962097168, + "step": 125040 + }, + { + "epoch": 0.3404382741254934, + "grad_norm": 2.2672221660614014, + "learning_rate": 1.3209268093226115e-05, + "loss": 2.5428543090820312, + "step": 125060 + }, + { + "epoch": 0.34049271811623794, + "grad_norm": 2.255235195159912, + "learning_rate": 1.320817772931716e-05, + "loss": 2.483054542541504, + "step": 125080 + }, + { + "epoch": 0.34054716210698244, + "grad_norm": 2.0317177772521973, + "learning_rate": 1.3207087365408206e-05, + "loss": 2.486362838745117, + "step": 125100 + }, + { + "epoch": 0.34060160609772694, + "grad_norm": 3.159538984298706, + "learning_rate": 1.320599700149925e-05, + "loss": 2.483806037902832, + "step": 125120 + }, + { + "epoch": 0.3406560500884715, + "grad_norm": 2.1404480934143066, + "learning_rate": 1.3204906637590298e-05, + "loss": 2.507346534729004, + "step": 125140 + }, + { + "epoch": 0.340710494079216, + "grad_norm": 2.1219120025634766, + "learning_rate": 1.3203816273681342e-05, + "loss": 2.4782175064086913, + "step": 125160 + }, + { + "epoch": 0.34076493806996055, + "grad_norm": 2.283376455307007, + "learning_rate": 1.3202725909772386e-05, + "loss": 2.5018463134765625, + "step": 125180 + }, + { + "epoch": 0.34081938206070506, + "grad_norm": 2.1622469425201416, + "learning_rate": 1.3201635545863434e-05, + "loss": 2.5146203994750977, + "step": 125200 + }, + { + "epoch": 0.34087382605144956, + "grad_norm": 2.286648988723755, + "learning_rate": 1.3200545181954478e-05, + "loss": 2.513237953186035, + "step": 125220 + }, + { + "epoch": 0.3409282700421941, + "grad_norm": 2.2074060440063477, + "learning_rate": 1.3199454818045524e-05, + "loss": 2.484902191162109, + "step": 125240 + }, + { + "epoch": 0.3409827140329386, + "grad_norm": 2.148206949234009, + "learning_rate": 1.319836445413657e-05, + "loss": 2.4995038986206053, + "step": 125260 + }, + { + "epoch": 0.3410371580236831, + "grad_norm": 2.1857082843780518, + "learning_rate": 1.3197274090227614e-05, + "loss": 2.5098764419555666, + "step": 125280 + }, + { + "epoch": 0.3410916020144277, + "grad_norm": 2.2439773082733154, + "learning_rate": 1.3196183726318661e-05, + "loss": 2.5155302047729493, + "step": 125300 + }, + { + "epoch": 0.3411460460051722, + "grad_norm": 2.087874412536621, + "learning_rate": 1.3195093362409705e-05, + "loss": 2.512694549560547, + "step": 125320 + }, + { + "epoch": 0.3412004899959167, + "grad_norm": 2.3384597301483154, + "learning_rate": 1.319400299850075e-05, + "loss": 2.527881622314453, + "step": 125340 + }, + { + "epoch": 0.34125493398666124, + "grad_norm": 2.1782994270324707, + "learning_rate": 1.3192912634591797e-05, + "loss": 2.4776437759399412, + "step": 125360 + }, + { + "epoch": 0.34130937797740574, + "grad_norm": 2.0733447074890137, + "learning_rate": 1.3191822270682841e-05, + "loss": 2.490351104736328, + "step": 125380 + }, + { + "epoch": 0.34136382196815024, + "grad_norm": 2.1116926670074463, + "learning_rate": 1.3190731906773887e-05, + "loss": 2.508293342590332, + "step": 125400 + }, + { + "epoch": 0.3414182659588948, + "grad_norm": 2.516091823577881, + "learning_rate": 1.3189641542864932e-05, + "loss": 2.5160638809204103, + "step": 125420 + }, + { + "epoch": 0.3414727099496393, + "grad_norm": 2.3792002201080322, + "learning_rate": 1.3188551178955977e-05, + "loss": 2.5003469467163084, + "step": 125440 + }, + { + "epoch": 0.34152715394038385, + "grad_norm": 2.2236876487731934, + "learning_rate": 1.3187460815047024e-05, + "loss": 2.5255096435546873, + "step": 125460 + }, + { + "epoch": 0.34158159793112836, + "grad_norm": 2.1888320446014404, + "learning_rate": 1.3186370451138068e-05, + "loss": 2.47713623046875, + "step": 125480 + }, + { + "epoch": 0.34163604192187286, + "grad_norm": 2.5106186866760254, + "learning_rate": 1.3185280087229112e-05, + "loss": 2.496138000488281, + "step": 125500 + }, + { + "epoch": 0.3416904859126174, + "grad_norm": 2.040928602218628, + "learning_rate": 1.318418972332016e-05, + "loss": 2.5109718322753904, + "step": 125520 + }, + { + "epoch": 0.3417449299033619, + "grad_norm": 2.1206135749816895, + "learning_rate": 1.3183099359411204e-05, + "loss": 2.495720100402832, + "step": 125540 + }, + { + "epoch": 0.3417993738941064, + "grad_norm": 2.1747560501098633, + "learning_rate": 1.318200899550225e-05, + "loss": 2.505108642578125, + "step": 125560 + }, + { + "epoch": 0.341853817884851, + "grad_norm": 2.1944658756256104, + "learning_rate": 1.3180918631593296e-05, + "loss": 2.5012174606323243, + "step": 125580 + }, + { + "epoch": 0.3419082618755955, + "grad_norm": 2.2867305278778076, + "learning_rate": 1.317982826768434e-05, + "loss": 2.491897392272949, + "step": 125600 + }, + { + "epoch": 0.34196270586634, + "grad_norm": 2.1366934776306152, + "learning_rate": 1.3178737903775387e-05, + "loss": 2.5076696395874025, + "step": 125620 + }, + { + "epoch": 0.34201714985708453, + "grad_norm": 2.147082805633545, + "learning_rate": 1.3177647539866431e-05, + "loss": 2.4784154891967773, + "step": 125640 + }, + { + "epoch": 0.34207159384782904, + "grad_norm": 2.3550734519958496, + "learning_rate": 1.3176557175957475e-05, + "loss": 2.480421447753906, + "step": 125660 + }, + { + "epoch": 0.3421260378385736, + "grad_norm": 2.1428751945495605, + "learning_rate": 1.3175466812048523e-05, + "loss": 2.4784393310546875, + "step": 125680 + }, + { + "epoch": 0.3421804818293181, + "grad_norm": 2.1148412227630615, + "learning_rate": 1.3174376448139567e-05, + "loss": 2.5117555618286134, + "step": 125700 + }, + { + "epoch": 0.3422349258200626, + "grad_norm": 2.1181437969207764, + "learning_rate": 1.3173286084230613e-05, + "loss": 2.51988525390625, + "step": 125720 + }, + { + "epoch": 0.34228936981080715, + "grad_norm": 2.0177485942840576, + "learning_rate": 1.3172195720321659e-05, + "loss": 2.4893320083618162, + "step": 125740 + }, + { + "epoch": 0.34234381380155166, + "grad_norm": 2.1846771240234375, + "learning_rate": 1.3171105356412703e-05, + "loss": 2.4778324127197267, + "step": 125760 + }, + { + "epoch": 0.34239825779229616, + "grad_norm": 2.212921142578125, + "learning_rate": 1.317001499250375e-05, + "loss": 2.508160400390625, + "step": 125780 + }, + { + "epoch": 0.3424527017830407, + "grad_norm": 2.1377053260803223, + "learning_rate": 1.3168924628594794e-05, + "loss": 2.4972980499267576, + "step": 125800 + }, + { + "epoch": 0.3425071457737852, + "grad_norm": 2.1376471519470215, + "learning_rate": 1.3167834264685838e-05, + "loss": 2.498265266418457, + "step": 125820 + }, + { + "epoch": 0.3425615897645297, + "grad_norm": 2.131563186645508, + "learning_rate": 1.3166743900776886e-05, + "loss": 2.516915512084961, + "step": 125840 + }, + { + "epoch": 0.3426160337552743, + "grad_norm": 2.3541157245635986, + "learning_rate": 1.316565353686793e-05, + "loss": 2.514470672607422, + "step": 125860 + }, + { + "epoch": 0.3426704777460188, + "grad_norm": 2.1698036193847656, + "learning_rate": 1.3164563172958976e-05, + "loss": 2.4768495559692383, + "step": 125880 + }, + { + "epoch": 0.3427249217367633, + "grad_norm": 2.0493152141571045, + "learning_rate": 1.3163472809050022e-05, + "loss": 2.502823257446289, + "step": 125900 + }, + { + "epoch": 0.34277936572750783, + "grad_norm": 2.217540740966797, + "learning_rate": 1.3162382445141066e-05, + "loss": 2.4857000350952148, + "step": 125920 + }, + { + "epoch": 0.34283380971825234, + "grad_norm": 2.179154634475708, + "learning_rate": 1.3161292081232113e-05, + "loss": 2.5148000717163086, + "step": 125940 + }, + { + "epoch": 0.3428882537089969, + "grad_norm": 2.108551502227783, + "learning_rate": 1.3160201717323157e-05, + "loss": 2.519683265686035, + "step": 125960 + }, + { + "epoch": 0.3429426976997414, + "grad_norm": 2.164721965789795, + "learning_rate": 1.3159111353414202e-05, + "loss": 2.5114376068115236, + "step": 125980 + }, + { + "epoch": 0.3429971416904859, + "grad_norm": 2.099806785583496, + "learning_rate": 1.3158020989505249e-05, + "loss": 2.5140193939208983, + "step": 126000 + }, + { + "epoch": 0.34305158568123045, + "grad_norm": 2.2446751594543457, + "learning_rate": 1.3156930625596293e-05, + "loss": 2.504446601867676, + "step": 126020 + }, + { + "epoch": 0.34310602967197495, + "grad_norm": 2.1749820709228516, + "learning_rate": 1.3155840261687339e-05, + "loss": 2.489010238647461, + "step": 126040 + }, + { + "epoch": 0.34316047366271946, + "grad_norm": 2.2242724895477295, + "learning_rate": 1.3154749897778385e-05, + "loss": 2.485668182373047, + "step": 126060 + }, + { + "epoch": 0.343214917653464, + "grad_norm": 2.0784695148468018, + "learning_rate": 1.3153659533869429e-05, + "loss": 2.5105527877807616, + "step": 126080 + }, + { + "epoch": 0.3432693616442085, + "grad_norm": 2.1396870613098145, + "learning_rate": 1.3152569169960476e-05, + "loss": 2.500462532043457, + "step": 126100 + }, + { + "epoch": 0.343323805634953, + "grad_norm": 2.247570037841797, + "learning_rate": 1.315147880605152e-05, + "loss": 2.481942558288574, + "step": 126120 + }, + { + "epoch": 0.3433782496256976, + "grad_norm": 2.1818506717681885, + "learning_rate": 1.3150388442142565e-05, + "loss": 2.50142765045166, + "step": 126140 + }, + { + "epoch": 0.3434326936164421, + "grad_norm": 2.1398191452026367, + "learning_rate": 1.3149298078233612e-05, + "loss": 2.480317306518555, + "step": 126160 + }, + { + "epoch": 0.34348713760718663, + "grad_norm": 2.098024845123291, + "learning_rate": 1.3148207714324656e-05, + "loss": 2.487238311767578, + "step": 126180 + }, + { + "epoch": 0.34354158159793113, + "grad_norm": 2.19372820854187, + "learning_rate": 1.3147117350415702e-05, + "loss": 2.502048873901367, + "step": 126200 + }, + { + "epoch": 0.34359602558867564, + "grad_norm": 2.0573506355285645, + "learning_rate": 1.3146026986506748e-05, + "loss": 2.5016204833984377, + "step": 126220 + }, + { + "epoch": 0.3436504695794202, + "grad_norm": 2.1623871326446533, + "learning_rate": 1.3144936622597792e-05, + "loss": 2.502132606506348, + "step": 126240 + }, + { + "epoch": 0.3437049135701647, + "grad_norm": 2.13899827003479, + "learning_rate": 1.314384625868884e-05, + "loss": 2.5023077011108397, + "step": 126260 + }, + { + "epoch": 0.3437593575609092, + "grad_norm": 2.082369089126587, + "learning_rate": 1.3142755894779884e-05, + "loss": 2.490297317504883, + "step": 126280 + }, + { + "epoch": 0.34381380155165375, + "grad_norm": 2.05511736869812, + "learning_rate": 1.3141665530870928e-05, + "loss": 2.5083740234375, + "step": 126300 + }, + { + "epoch": 0.34386824554239825, + "grad_norm": 2.0849103927612305, + "learning_rate": 1.3140575166961975e-05, + "loss": 2.4771074295043944, + "step": 126320 + }, + { + "epoch": 0.34392268953314276, + "grad_norm": 2.1763224601745605, + "learning_rate": 1.313948480305302e-05, + "loss": 2.497361946105957, + "step": 126340 + }, + { + "epoch": 0.3439771335238873, + "grad_norm": 2.206627368927002, + "learning_rate": 1.3138394439144063e-05, + "loss": 2.4948890686035154, + "step": 126360 + }, + { + "epoch": 0.3440315775146318, + "grad_norm": 2.319735288619995, + "learning_rate": 1.3137304075235111e-05, + "loss": 2.50424861907959, + "step": 126380 + }, + { + "epoch": 0.34408602150537637, + "grad_norm": 2.062228202819824, + "learning_rate": 1.3136213711326155e-05, + "loss": 2.493322563171387, + "step": 126400 + }, + { + "epoch": 0.3441404654961209, + "grad_norm": 2.1700940132141113, + "learning_rate": 1.3135123347417203e-05, + "loss": 2.5012285232543947, + "step": 126420 + }, + { + "epoch": 0.3441949094868654, + "grad_norm": 2.0832931995391846, + "learning_rate": 1.3134032983508247e-05, + "loss": 2.4923833847045898, + "step": 126440 + }, + { + "epoch": 0.34424935347760993, + "grad_norm": 2.1579878330230713, + "learning_rate": 1.313294261959929e-05, + "loss": 2.516825485229492, + "step": 126460 + }, + { + "epoch": 0.34430379746835443, + "grad_norm": 2.2047996520996094, + "learning_rate": 1.3131852255690338e-05, + "loss": 2.4814964294433595, + "step": 126480 + }, + { + "epoch": 0.34435824145909893, + "grad_norm": 2.2244362831115723, + "learning_rate": 1.3130761891781382e-05, + "loss": 2.5094804763793945, + "step": 126500 + }, + { + "epoch": 0.3444126854498435, + "grad_norm": 2.137310266494751, + "learning_rate": 1.312967152787243e-05, + "loss": 2.5045730590820314, + "step": 126520 + }, + { + "epoch": 0.344467129440588, + "grad_norm": 2.182129383087158, + "learning_rate": 1.3128581163963474e-05, + "loss": 2.4925607681274413, + "step": 126540 + }, + { + "epoch": 0.3445215734313325, + "grad_norm": 2.128488302230835, + "learning_rate": 1.3127490800054518e-05, + "loss": 2.4966350555419923, + "step": 126560 + }, + { + "epoch": 0.34457601742207705, + "grad_norm": 2.346958637237549, + "learning_rate": 1.3126400436145566e-05, + "loss": 2.490437316894531, + "step": 126580 + }, + { + "epoch": 0.34463046141282155, + "grad_norm": 2.083566904067993, + "learning_rate": 1.312531007223661e-05, + "loss": 2.492141532897949, + "step": 126600 + }, + { + "epoch": 0.34468490540356606, + "grad_norm": 2.174729585647583, + "learning_rate": 1.3124219708327654e-05, + "loss": 2.4918724060058595, + "step": 126620 + }, + { + "epoch": 0.3447393493943106, + "grad_norm": 2.2170917987823486, + "learning_rate": 1.3123129344418701e-05, + "loss": 2.5085260391235353, + "step": 126640 + }, + { + "epoch": 0.3447937933850551, + "grad_norm": 2.1747260093688965, + "learning_rate": 1.3122038980509746e-05, + "loss": 2.4964082717895506, + "step": 126660 + }, + { + "epoch": 0.34484823737579967, + "grad_norm": 2.109773635864258, + "learning_rate": 1.3120948616600793e-05, + "loss": 2.4913375854492186, + "step": 126680 + }, + { + "epoch": 0.3449026813665442, + "grad_norm": 2.143611192703247, + "learning_rate": 1.3119858252691837e-05, + "loss": 2.4961292266845705, + "step": 126700 + }, + { + "epoch": 0.3449571253572887, + "grad_norm": 2.2013072967529297, + "learning_rate": 1.3118767888782881e-05, + "loss": 2.4997806549072266, + "step": 126720 + }, + { + "epoch": 0.34501156934803323, + "grad_norm": 2.1279826164245605, + "learning_rate": 1.3117677524873929e-05, + "loss": 2.467571258544922, + "step": 126740 + }, + { + "epoch": 0.34506601333877773, + "grad_norm": 2.274247169494629, + "learning_rate": 1.3116587160964973e-05, + "loss": 2.485329818725586, + "step": 126760 + }, + { + "epoch": 0.34512045732952223, + "grad_norm": 2.2193093299865723, + "learning_rate": 1.3115496797056017e-05, + "loss": 2.491042137145996, + "step": 126780 + }, + { + "epoch": 0.3451749013202668, + "grad_norm": 2.212153434753418, + "learning_rate": 1.3114406433147065e-05, + "loss": 2.5139867782592775, + "step": 126800 + }, + { + "epoch": 0.3452293453110113, + "grad_norm": 2.1470019817352295, + "learning_rate": 1.3113316069238109e-05, + "loss": 2.486281394958496, + "step": 126820 + }, + { + "epoch": 0.3452837893017558, + "grad_norm": 2.062110424041748, + "learning_rate": 1.3112225705329156e-05, + "loss": 2.5232311248779298, + "step": 126840 + }, + { + "epoch": 0.34533823329250035, + "grad_norm": 2.1308913230895996, + "learning_rate": 1.31111353414202e-05, + "loss": 2.471990966796875, + "step": 126860 + }, + { + "epoch": 0.34539267728324485, + "grad_norm": 2.2585902214050293, + "learning_rate": 1.3110044977511244e-05, + "loss": 2.499542999267578, + "step": 126880 + }, + { + "epoch": 0.3454471212739894, + "grad_norm": 2.154552698135376, + "learning_rate": 1.3108954613602292e-05, + "loss": 2.501463508605957, + "step": 126900 + }, + { + "epoch": 0.3455015652647339, + "grad_norm": 2.4413363933563232, + "learning_rate": 1.3107864249693336e-05, + "loss": 2.4838342666625977, + "step": 126920 + }, + { + "epoch": 0.3455560092554784, + "grad_norm": 2.12753963470459, + "learning_rate": 1.310677388578438e-05, + "loss": 2.4996501922607424, + "step": 126940 + }, + { + "epoch": 0.34561045324622297, + "grad_norm": 2.180020570755005, + "learning_rate": 1.3105683521875428e-05, + "loss": 2.49395809173584, + "step": 126960 + }, + { + "epoch": 0.34566489723696747, + "grad_norm": 2.138458013534546, + "learning_rate": 1.3104593157966472e-05, + "loss": 2.4966121673583985, + "step": 126980 + }, + { + "epoch": 0.345719341227712, + "grad_norm": 2.0801377296447754, + "learning_rate": 1.310350279405752e-05, + "loss": 2.5187387466430664, + "step": 127000 + }, + { + "epoch": 0.34577378521845653, + "grad_norm": 2.1552462577819824, + "learning_rate": 1.3102412430148563e-05, + "loss": 2.493898391723633, + "step": 127020 + }, + { + "epoch": 0.34582822920920103, + "grad_norm": 2.178881883621216, + "learning_rate": 1.3101322066239607e-05, + "loss": 2.459905433654785, + "step": 127040 + }, + { + "epoch": 0.34588267319994553, + "grad_norm": 1.9945164918899536, + "learning_rate": 1.3100231702330655e-05, + "loss": 2.4991003036499024, + "step": 127060 + }, + { + "epoch": 0.3459371171906901, + "grad_norm": 2.2616374492645264, + "learning_rate": 1.3099141338421699e-05, + "loss": 2.4788122177124023, + "step": 127080 + }, + { + "epoch": 0.3459915611814346, + "grad_norm": 2.256577491760254, + "learning_rate": 1.3098050974512743e-05, + "loss": 2.522252655029297, + "step": 127100 + }, + { + "epoch": 0.3460460051721791, + "grad_norm": 2.230156660079956, + "learning_rate": 1.309696061060379e-05, + "loss": 2.4746204376220704, + "step": 127120 + }, + { + "epoch": 0.34610044916292365, + "grad_norm": 2.127652406692505, + "learning_rate": 1.3095870246694835e-05, + "loss": 2.502954292297363, + "step": 127140 + }, + { + "epoch": 0.34615489315366815, + "grad_norm": 2.5338690280914307, + "learning_rate": 1.3094779882785882e-05, + "loss": 2.505704879760742, + "step": 127160 + }, + { + "epoch": 0.3462093371444127, + "grad_norm": 2.2152700424194336, + "learning_rate": 1.3093689518876926e-05, + "loss": 2.4987945556640625, + "step": 127180 + }, + { + "epoch": 0.3462637811351572, + "grad_norm": 2.324394702911377, + "learning_rate": 1.309259915496797e-05, + "loss": 2.4772621154785157, + "step": 127200 + }, + { + "epoch": 0.3463182251259017, + "grad_norm": 2.1924424171447754, + "learning_rate": 1.3091508791059018e-05, + "loss": 2.504496765136719, + "step": 127220 + }, + { + "epoch": 0.34637266911664627, + "grad_norm": 2.9495668411254883, + "learning_rate": 1.3090418427150062e-05, + "loss": 2.506149673461914, + "step": 127240 + }, + { + "epoch": 0.34642711310739077, + "grad_norm": 2.0600712299346924, + "learning_rate": 1.3089328063241106e-05, + "loss": 2.5172056198120116, + "step": 127260 + }, + { + "epoch": 0.3464815570981353, + "grad_norm": 2.1559009552001953, + "learning_rate": 1.3088237699332154e-05, + "loss": 2.506076431274414, + "step": 127280 + }, + { + "epoch": 0.34653600108887983, + "grad_norm": 2.0669333934783936, + "learning_rate": 1.3087147335423198e-05, + "loss": 2.5000511169433595, + "step": 127300 + }, + { + "epoch": 0.34659044507962433, + "grad_norm": 2.193889617919922, + "learning_rate": 1.3086056971514245e-05, + "loss": 2.4821340560913088, + "step": 127320 + }, + { + "epoch": 0.34664488907036883, + "grad_norm": 2.1054601669311523, + "learning_rate": 1.308496660760529e-05, + "loss": 2.483208656311035, + "step": 127340 + }, + { + "epoch": 0.3466993330611134, + "grad_norm": 2.244908094406128, + "learning_rate": 1.3083876243696334e-05, + "loss": 2.5095104217529296, + "step": 127360 + }, + { + "epoch": 0.3467537770518579, + "grad_norm": 2.1011993885040283, + "learning_rate": 1.3082785879787381e-05, + "loss": 2.4842304229736327, + "step": 127380 + }, + { + "epoch": 0.34680822104260245, + "grad_norm": 2.1349034309387207, + "learning_rate": 1.3081695515878425e-05, + "loss": 2.5150529861450197, + "step": 127400 + }, + { + "epoch": 0.34686266503334695, + "grad_norm": 2.151432752609253, + "learning_rate": 1.308060515196947e-05, + "loss": 2.502089500427246, + "step": 127420 + }, + { + "epoch": 0.34691710902409145, + "grad_norm": 2.1494946479797363, + "learning_rate": 1.3079514788060517e-05, + "loss": 2.4927574157714845, + "step": 127440 + }, + { + "epoch": 0.346971553014836, + "grad_norm": 2.319545030593872, + "learning_rate": 1.3078424424151561e-05, + "loss": 2.50894832611084, + "step": 127460 + }, + { + "epoch": 0.3470259970055805, + "grad_norm": 2.1901352405548096, + "learning_rate": 1.3077334060242608e-05, + "loss": 2.499711799621582, + "step": 127480 + }, + { + "epoch": 0.347080440996325, + "grad_norm": 2.179060220718384, + "learning_rate": 1.3076243696333653e-05, + "loss": 2.489302062988281, + "step": 127500 + }, + { + "epoch": 0.34713488498706957, + "grad_norm": 2.215736150741577, + "learning_rate": 1.3075153332424697e-05, + "loss": 2.4923847198486326, + "step": 127520 + }, + { + "epoch": 0.34718932897781407, + "grad_norm": 2.1269209384918213, + "learning_rate": 1.3074062968515744e-05, + "loss": 2.5235162734985352, + "step": 127540 + }, + { + "epoch": 0.3472437729685586, + "grad_norm": 2.1663169860839844, + "learning_rate": 1.3072972604606788e-05, + "loss": 2.503071403503418, + "step": 127560 + }, + { + "epoch": 0.34729821695930313, + "grad_norm": 2.2333521842956543, + "learning_rate": 1.3071882240697832e-05, + "loss": 2.5048639297485353, + "step": 127580 + }, + { + "epoch": 0.34735266095004763, + "grad_norm": 2.0980637073516846, + "learning_rate": 1.307079187678888e-05, + "loss": 2.4857282638549805, + "step": 127600 + }, + { + "epoch": 0.34740710494079213, + "grad_norm": 2.1331629753112793, + "learning_rate": 1.3069701512879924e-05, + "loss": 2.5116239547729493, + "step": 127620 + }, + { + "epoch": 0.3474615489315367, + "grad_norm": 2.145406484603882, + "learning_rate": 1.3068611148970972e-05, + "loss": 2.475689697265625, + "step": 127640 + }, + { + "epoch": 0.3475159929222812, + "grad_norm": 2.169802188873291, + "learning_rate": 1.3067520785062016e-05, + "loss": 2.4978967666625977, + "step": 127660 + }, + { + "epoch": 0.34757043691302575, + "grad_norm": 3.0958750247955322, + "learning_rate": 1.306643042115306e-05, + "loss": 2.492715835571289, + "step": 127680 + }, + { + "epoch": 0.34762488090377025, + "grad_norm": 2.1191296577453613, + "learning_rate": 1.3065340057244107e-05, + "loss": 2.5111709594726563, + "step": 127700 + }, + { + "epoch": 0.34767932489451475, + "grad_norm": 2.120159387588501, + "learning_rate": 1.3064249693335151e-05, + "loss": 2.5131868362426757, + "step": 127720 + }, + { + "epoch": 0.3477337688852593, + "grad_norm": 2.097201347351074, + "learning_rate": 1.3063159329426196e-05, + "loss": 2.507297897338867, + "step": 127740 + }, + { + "epoch": 0.3477882128760038, + "grad_norm": 2.136622190475464, + "learning_rate": 1.3062068965517243e-05, + "loss": 2.4886884689331055, + "step": 127760 + }, + { + "epoch": 0.3478426568667483, + "grad_norm": 2.226268768310547, + "learning_rate": 1.3060978601608287e-05, + "loss": 2.5003145217895506, + "step": 127780 + }, + { + "epoch": 0.34789710085749287, + "grad_norm": 2.266815662384033, + "learning_rate": 1.3059888237699335e-05, + "loss": 2.5011152267456054, + "step": 127800 + }, + { + "epoch": 0.34795154484823737, + "grad_norm": 2.307767152786255, + "learning_rate": 1.3058797873790379e-05, + "loss": 2.5175031661987304, + "step": 127820 + }, + { + "epoch": 0.3480059888389819, + "grad_norm": 2.3765323162078857, + "learning_rate": 1.3057707509881423e-05, + "loss": 2.4910003662109377, + "step": 127840 + }, + { + "epoch": 0.34806043282972643, + "grad_norm": 2.4291045665740967, + "learning_rate": 1.305661714597247e-05, + "loss": 2.5103023529052733, + "step": 127860 + }, + { + "epoch": 0.34811487682047093, + "grad_norm": 2.210430383682251, + "learning_rate": 1.3055526782063514e-05, + "loss": 2.50689754486084, + "step": 127880 + }, + { + "epoch": 0.3481693208112155, + "grad_norm": 2.131692886352539, + "learning_rate": 1.3054436418154559e-05, + "loss": 2.503743362426758, + "step": 127900 + }, + { + "epoch": 0.34822376480196, + "grad_norm": 2.3093137741088867, + "learning_rate": 1.3053346054245606e-05, + "loss": 2.4786136627197264, + "step": 127920 + }, + { + "epoch": 0.3482782087927045, + "grad_norm": 2.1383302211761475, + "learning_rate": 1.305225569033665e-05, + "loss": 2.4876964569091795, + "step": 127940 + }, + { + "epoch": 0.34833265278344905, + "grad_norm": 2.303574562072754, + "learning_rate": 1.3051165326427698e-05, + "loss": 2.5217527389526366, + "step": 127960 + }, + { + "epoch": 0.34838709677419355, + "grad_norm": 2.1393802165985107, + "learning_rate": 1.3050074962518742e-05, + "loss": 2.487570571899414, + "step": 127980 + }, + { + "epoch": 0.34844154076493805, + "grad_norm": 2.1683287620544434, + "learning_rate": 1.3048984598609786e-05, + "loss": 2.5088932037353517, + "step": 128000 + }, + { + "epoch": 0.3484959847556826, + "grad_norm": 2.140662908554077, + "learning_rate": 1.3047894234700833e-05, + "loss": 2.5112260818481444, + "step": 128020 + }, + { + "epoch": 0.3485504287464271, + "grad_norm": 2.2381038665771484, + "learning_rate": 1.3046803870791878e-05, + "loss": 2.4772552490234374, + "step": 128040 + }, + { + "epoch": 0.3486048727371716, + "grad_norm": 2.0801584720611572, + "learning_rate": 1.3045713506882922e-05, + "loss": 2.5026405334472654, + "step": 128060 + }, + { + "epoch": 0.34865931672791617, + "grad_norm": 2.4670307636260986, + "learning_rate": 1.304462314297397e-05, + "loss": 2.50368595123291, + "step": 128080 + }, + { + "epoch": 0.34871376071866067, + "grad_norm": 2.1698811054229736, + "learning_rate": 1.3043532779065013e-05, + "loss": 2.5104887008666994, + "step": 128100 + }, + { + "epoch": 0.34876820470940517, + "grad_norm": 2.158474922180176, + "learning_rate": 1.304244241515606e-05, + "loss": 2.50643253326416, + "step": 128120 + }, + { + "epoch": 0.34882264870014973, + "grad_norm": 2.1802027225494385, + "learning_rate": 1.3041352051247105e-05, + "loss": 2.480080795288086, + "step": 128140 + }, + { + "epoch": 0.34887709269089423, + "grad_norm": 2.1280364990234375, + "learning_rate": 1.3040261687338149e-05, + "loss": 2.527796745300293, + "step": 128160 + }, + { + "epoch": 0.3489315366816388, + "grad_norm": 2.1669797897338867, + "learning_rate": 1.3039171323429197e-05, + "loss": 2.4699119567871093, + "step": 128180 + }, + { + "epoch": 0.3489859806723833, + "grad_norm": 2.19069766998291, + "learning_rate": 1.303808095952024e-05, + "loss": 2.472138214111328, + "step": 128200 + }, + { + "epoch": 0.3490404246631278, + "grad_norm": 2.2318673133850098, + "learning_rate": 1.3036990595611288e-05, + "loss": 2.4964435577392576, + "step": 128220 + }, + { + "epoch": 0.34909486865387235, + "grad_norm": 2.2181270122528076, + "learning_rate": 1.3035900231702332e-05, + "loss": 2.5219884872436524, + "step": 128240 + }, + { + "epoch": 0.34914931264461685, + "grad_norm": 2.1951022148132324, + "learning_rate": 1.3034809867793376e-05, + "loss": 2.522673797607422, + "step": 128260 + }, + { + "epoch": 0.34920375663536135, + "grad_norm": 2.24298357963562, + "learning_rate": 1.3033719503884424e-05, + "loss": 2.4970300674438475, + "step": 128280 + }, + { + "epoch": 0.3492582006261059, + "grad_norm": 2.034193754196167, + "learning_rate": 1.3032629139975468e-05, + "loss": 2.492824745178223, + "step": 128300 + }, + { + "epoch": 0.3493126446168504, + "grad_norm": 2.1301426887512207, + "learning_rate": 1.3031538776066512e-05, + "loss": 2.4867532730102537, + "step": 128320 + }, + { + "epoch": 0.3493670886075949, + "grad_norm": 2.154405117034912, + "learning_rate": 1.303044841215756e-05, + "loss": 2.5163646697998048, + "step": 128340 + }, + { + "epoch": 0.34942153259833947, + "grad_norm": 2.1206741333007812, + "learning_rate": 1.3029358048248604e-05, + "loss": 2.5053239822387696, + "step": 128360 + }, + { + "epoch": 0.34947597658908397, + "grad_norm": 2.175009250640869, + "learning_rate": 1.3028267684339651e-05, + "loss": 2.4591344833374023, + "step": 128380 + }, + { + "epoch": 0.3495304205798285, + "grad_norm": 2.0843265056610107, + "learning_rate": 1.3027177320430695e-05, + "loss": 2.4643035888671876, + "step": 128400 + }, + { + "epoch": 0.34958486457057303, + "grad_norm": 2.1661527156829834, + "learning_rate": 1.302608695652174e-05, + "loss": 2.4852027893066406, + "step": 128420 + }, + { + "epoch": 0.34963930856131753, + "grad_norm": 2.27459979057312, + "learning_rate": 1.3024996592612787e-05, + "loss": 2.5228357315063477, + "step": 128440 + }, + { + "epoch": 0.3496937525520621, + "grad_norm": 2.2528514862060547, + "learning_rate": 1.3023906228703831e-05, + "loss": 2.5537370681762694, + "step": 128460 + }, + { + "epoch": 0.3497481965428066, + "grad_norm": 2.141064405441284, + "learning_rate": 1.3022815864794875e-05, + "loss": 2.4867300033569335, + "step": 128480 + }, + { + "epoch": 0.3498026405335511, + "grad_norm": 2.1531248092651367, + "learning_rate": 1.3021725500885923e-05, + "loss": 2.503964424133301, + "step": 128500 + }, + { + "epoch": 0.34985708452429565, + "grad_norm": 2.2232370376586914, + "learning_rate": 1.3020635136976967e-05, + "loss": 2.471622276306152, + "step": 128520 + }, + { + "epoch": 0.34991152851504015, + "grad_norm": 2.163084030151367, + "learning_rate": 1.3019544773068014e-05, + "loss": 2.439350891113281, + "step": 128540 + }, + { + "epoch": 0.34996597250578465, + "grad_norm": 2.153409004211426, + "learning_rate": 1.3018454409159058e-05, + "loss": 2.513503074645996, + "step": 128560 + }, + { + "epoch": 0.3500204164965292, + "grad_norm": 2.0538666248321533, + "learning_rate": 1.3017364045250103e-05, + "loss": 2.477710151672363, + "step": 128580 + }, + { + "epoch": 0.3500748604872737, + "grad_norm": 2.1918530464172363, + "learning_rate": 1.301627368134115e-05, + "loss": 2.4722158432006838, + "step": 128600 + }, + { + "epoch": 0.35012930447801827, + "grad_norm": 2.195563554763794, + "learning_rate": 1.3015183317432194e-05, + "loss": 2.5036598205566407, + "step": 128620 + }, + { + "epoch": 0.35018374846876277, + "grad_norm": 2.4410383701324463, + "learning_rate": 1.3014092953523238e-05, + "loss": 2.5117040634155274, + "step": 128640 + }, + { + "epoch": 0.35023819245950727, + "grad_norm": 2.0842125415802, + "learning_rate": 1.3013002589614286e-05, + "loss": 2.5064205169677733, + "step": 128660 + }, + { + "epoch": 0.3502926364502518, + "grad_norm": 1.9893519878387451, + "learning_rate": 1.301191222570533e-05, + "loss": 2.4949373245239257, + "step": 128680 + }, + { + "epoch": 0.35034708044099633, + "grad_norm": 2.106966733932495, + "learning_rate": 1.3010821861796377e-05, + "loss": 2.505618667602539, + "step": 128700 + }, + { + "epoch": 0.35040152443174083, + "grad_norm": 2.1084232330322266, + "learning_rate": 1.3009731497887422e-05, + "loss": 2.5355932235717775, + "step": 128720 + }, + { + "epoch": 0.3504559684224854, + "grad_norm": 2.203657388687134, + "learning_rate": 1.3008641133978466e-05, + "loss": 2.496349906921387, + "step": 128740 + }, + { + "epoch": 0.3505104124132299, + "grad_norm": 2.7448513507843018, + "learning_rate": 1.3007550770069513e-05, + "loss": 2.503176689147949, + "step": 128760 + }, + { + "epoch": 0.3505648564039744, + "grad_norm": 2.2292892932891846, + "learning_rate": 1.3006460406160557e-05, + "loss": 2.5244823455810548, + "step": 128780 + }, + { + "epoch": 0.35061930039471895, + "grad_norm": 2.226527214050293, + "learning_rate": 1.3005370042251601e-05, + "loss": 2.484709548950195, + "step": 128800 + }, + { + "epoch": 0.35067374438546345, + "grad_norm": 2.1153950691223145, + "learning_rate": 1.3004279678342649e-05, + "loss": 2.5407993316650392, + "step": 128820 + }, + { + "epoch": 0.35072818837620795, + "grad_norm": 2.0935192108154297, + "learning_rate": 1.3003189314433693e-05, + "loss": 2.4913652420043944, + "step": 128840 + }, + { + "epoch": 0.3507826323669525, + "grad_norm": 2.1113553047180176, + "learning_rate": 1.300209895052474e-05, + "loss": 2.492996025085449, + "step": 128860 + }, + { + "epoch": 0.350837076357697, + "grad_norm": 2.3240184783935547, + "learning_rate": 1.3001008586615785e-05, + "loss": 2.505877685546875, + "step": 128880 + }, + { + "epoch": 0.35089152034844157, + "grad_norm": 2.133977174758911, + "learning_rate": 1.2999918222706829e-05, + "loss": 2.4609790802001954, + "step": 128900 + }, + { + "epoch": 0.35094596433918607, + "grad_norm": 2.1254513263702393, + "learning_rate": 1.2998827858797876e-05, + "loss": 2.5284517288208006, + "step": 128920 + }, + { + "epoch": 0.35100040832993057, + "grad_norm": 2.262235164642334, + "learning_rate": 1.299773749488892e-05, + "loss": 2.476242446899414, + "step": 128940 + }, + { + "epoch": 0.3510548523206751, + "grad_norm": 2.0512826442718506, + "learning_rate": 1.2996647130979964e-05, + "loss": 2.484401893615723, + "step": 128960 + }, + { + "epoch": 0.3511092963114196, + "grad_norm": 2.1964335441589355, + "learning_rate": 1.2995556767071012e-05, + "loss": 2.508231544494629, + "step": 128980 + }, + { + "epoch": 0.35116374030216413, + "grad_norm": 2.207646369934082, + "learning_rate": 1.2994466403162056e-05, + "loss": 2.457352638244629, + "step": 129000 + }, + { + "epoch": 0.3512181842929087, + "grad_norm": 2.230471611022949, + "learning_rate": 1.2993376039253104e-05, + "loss": 2.5153154373168944, + "step": 129020 + }, + { + "epoch": 0.3512726282836532, + "grad_norm": 2.0841784477233887, + "learning_rate": 1.2992285675344148e-05, + "loss": 2.5288633346557616, + "step": 129040 + }, + { + "epoch": 0.3513270722743977, + "grad_norm": 3.960549831390381, + "learning_rate": 1.2991195311435192e-05, + "loss": 2.5043601989746094, + "step": 129060 + }, + { + "epoch": 0.35138151626514225, + "grad_norm": 2.0605080127716064, + "learning_rate": 1.299010494752624e-05, + "loss": 2.5228904724121093, + "step": 129080 + }, + { + "epoch": 0.35143596025588675, + "grad_norm": 2.5008912086486816, + "learning_rate": 1.2989014583617283e-05, + "loss": 2.4975332260131835, + "step": 129100 + }, + { + "epoch": 0.3514904042466313, + "grad_norm": 2.1165993213653564, + "learning_rate": 1.2987924219708328e-05, + "loss": 2.4920589447021486, + "step": 129120 + }, + { + "epoch": 0.3515448482373758, + "grad_norm": 2.0531184673309326, + "learning_rate": 1.2986833855799375e-05, + "loss": 2.5221729278564453, + "step": 129140 + }, + { + "epoch": 0.3515992922281203, + "grad_norm": 2.211463212966919, + "learning_rate": 1.298574349189042e-05, + "loss": 2.5020606994628904, + "step": 129160 + }, + { + "epoch": 0.35165373621886487, + "grad_norm": 2.2116661071777344, + "learning_rate": 1.2984653127981465e-05, + "loss": 2.5036396026611327, + "step": 129180 + }, + { + "epoch": 0.35170818020960937, + "grad_norm": 2.177968978881836, + "learning_rate": 1.298356276407251e-05, + "loss": 2.4653244018554688, + "step": 129200 + }, + { + "epoch": 0.35176262420035387, + "grad_norm": 2.046349048614502, + "learning_rate": 1.2982472400163555e-05, + "loss": 2.50162410736084, + "step": 129220 + }, + { + "epoch": 0.3518170681910984, + "grad_norm": 2.065345525741577, + "learning_rate": 1.2981382036254602e-05, + "loss": 2.5321624755859373, + "step": 129240 + }, + { + "epoch": 0.3518715121818429, + "grad_norm": 2.23498797416687, + "learning_rate": 1.2980291672345647e-05, + "loss": 2.48993034362793, + "step": 129260 + }, + { + "epoch": 0.35192595617258743, + "grad_norm": 2.2009997367858887, + "learning_rate": 1.297920130843669e-05, + "loss": 2.491263008117676, + "step": 129280 + }, + { + "epoch": 0.351980400163332, + "grad_norm": 2.1127474308013916, + "learning_rate": 1.2978110944527738e-05, + "loss": 2.477631759643555, + "step": 129300 + }, + { + "epoch": 0.3520348441540765, + "grad_norm": 2.1313765048980713, + "learning_rate": 1.2977020580618782e-05, + "loss": 2.4880294799804688, + "step": 129320 + }, + { + "epoch": 0.352089288144821, + "grad_norm": 2.1538405418395996, + "learning_rate": 1.2975930216709828e-05, + "loss": 2.501535987854004, + "step": 129340 + }, + { + "epoch": 0.35214373213556555, + "grad_norm": 2.1319453716278076, + "learning_rate": 1.2974839852800874e-05, + "loss": 2.453628349304199, + "step": 129360 + }, + { + "epoch": 0.35219817612631005, + "grad_norm": 2.090435266494751, + "learning_rate": 1.2973749488891918e-05, + "loss": 2.490713882446289, + "step": 129380 + }, + { + "epoch": 0.3522526201170546, + "grad_norm": 2.201486825942993, + "learning_rate": 1.2972659124982965e-05, + "loss": 2.500767707824707, + "step": 129400 + }, + { + "epoch": 0.3523070641077991, + "grad_norm": 2.287044048309326, + "learning_rate": 1.297156876107401e-05, + "loss": 2.48944091796875, + "step": 129420 + }, + { + "epoch": 0.3523615080985436, + "grad_norm": 2.0845987796783447, + "learning_rate": 1.2970478397165054e-05, + "loss": 2.503849220275879, + "step": 129440 + }, + { + "epoch": 0.35241595208928816, + "grad_norm": 2.41343092918396, + "learning_rate": 1.2969388033256101e-05, + "loss": 2.481779098510742, + "step": 129460 + }, + { + "epoch": 0.35247039608003267, + "grad_norm": 2.039947986602783, + "learning_rate": 1.2968297669347145e-05, + "loss": 2.4887231826782226, + "step": 129480 + }, + { + "epoch": 0.35252484007077717, + "grad_norm": 2.12199330329895, + "learning_rate": 1.2967207305438191e-05, + "loss": 2.514542007446289, + "step": 129500 + }, + { + "epoch": 0.3525792840615217, + "grad_norm": 2.0446135997772217, + "learning_rate": 1.2966116941529237e-05, + "loss": 2.478994369506836, + "step": 129520 + }, + { + "epoch": 0.3526337280522662, + "grad_norm": 2.140368938446045, + "learning_rate": 1.2965026577620281e-05, + "loss": 2.4760385513305665, + "step": 129540 + }, + { + "epoch": 0.35268817204301073, + "grad_norm": 2.121229410171509, + "learning_rate": 1.2963936213711329e-05, + "loss": 2.491132926940918, + "step": 129560 + }, + { + "epoch": 0.3527426160337553, + "grad_norm": 2.331197738647461, + "learning_rate": 1.2962845849802373e-05, + "loss": 2.49686336517334, + "step": 129580 + }, + { + "epoch": 0.3527970600244998, + "grad_norm": 2.13427472114563, + "learning_rate": 1.2961755485893417e-05, + "loss": 2.5018211364746095, + "step": 129600 + }, + { + "epoch": 0.35285150401524434, + "grad_norm": 2.344407796859741, + "learning_rate": 1.2960665121984464e-05, + "loss": 2.5039581298828124, + "step": 129620 + }, + { + "epoch": 0.35290594800598885, + "grad_norm": 2.117920160293579, + "learning_rate": 1.2959574758075508e-05, + "loss": 2.4844457626342775, + "step": 129640 + }, + { + "epoch": 0.35296039199673335, + "grad_norm": 2.255225658416748, + "learning_rate": 1.2958484394166554e-05, + "loss": 2.5031299591064453, + "step": 129660 + }, + { + "epoch": 0.3530148359874779, + "grad_norm": 2.1408421993255615, + "learning_rate": 1.29573940302576e-05, + "loss": 2.504359245300293, + "step": 129680 + }, + { + "epoch": 0.3530692799782224, + "grad_norm": 2.2069320678710938, + "learning_rate": 1.2956303666348644e-05, + "loss": 2.466919708251953, + "step": 129700 + }, + { + "epoch": 0.3531237239689669, + "grad_norm": 2.1161582469940186, + "learning_rate": 1.2955213302439692e-05, + "loss": 2.4410720825195313, + "step": 129720 + }, + { + "epoch": 0.35317816795971146, + "grad_norm": 2.1826484203338623, + "learning_rate": 1.2954122938530736e-05, + "loss": 2.492471122741699, + "step": 129740 + }, + { + "epoch": 0.35323261195045597, + "grad_norm": 2.2276694774627686, + "learning_rate": 1.295303257462178e-05, + "loss": 2.471967887878418, + "step": 129760 + }, + { + "epoch": 0.35328705594120047, + "grad_norm": 2.0605485439300537, + "learning_rate": 1.2951942210712827e-05, + "loss": 2.476084899902344, + "step": 129780 + }, + { + "epoch": 0.353341499931945, + "grad_norm": 2.156836986541748, + "learning_rate": 1.2950851846803872e-05, + "loss": 2.497318458557129, + "step": 129800 + }, + { + "epoch": 0.3533959439226895, + "grad_norm": 2.136871814727783, + "learning_rate": 1.2949761482894917e-05, + "loss": 2.4788293838500977, + "step": 129820 + }, + { + "epoch": 0.353450387913434, + "grad_norm": 2.1400294303894043, + "learning_rate": 1.2948671118985963e-05, + "loss": 2.517666816711426, + "step": 129840 + }, + { + "epoch": 0.3535048319041786, + "grad_norm": 2.1908886432647705, + "learning_rate": 1.2947580755077007e-05, + "loss": 2.448228454589844, + "step": 129860 + }, + { + "epoch": 0.3535592758949231, + "grad_norm": 2.319521427154541, + "learning_rate": 1.2946490391168055e-05, + "loss": 2.4633148193359373, + "step": 129880 + }, + { + "epoch": 0.35361371988566764, + "grad_norm": 2.475581407546997, + "learning_rate": 1.2945400027259099e-05, + "loss": 2.5256784439086912, + "step": 129900 + }, + { + "epoch": 0.35366816387641214, + "grad_norm": 2.4795985221862793, + "learning_rate": 1.2944309663350145e-05, + "loss": 2.523224449157715, + "step": 129920 + }, + { + "epoch": 0.35372260786715665, + "grad_norm": 2.103764295578003, + "learning_rate": 1.294321929944119e-05, + "loss": 2.5042545318603517, + "step": 129940 + }, + { + "epoch": 0.3537770518579012, + "grad_norm": 2.292801856994629, + "learning_rate": 1.2942128935532235e-05, + "loss": 2.4893417358398438, + "step": 129960 + }, + { + "epoch": 0.3538314958486457, + "grad_norm": 2.1665358543395996, + "learning_rate": 1.294103857162328e-05, + "loss": 2.5170482635498046, + "step": 129980 + }, + { + "epoch": 0.3538859398393902, + "grad_norm": 2.227318525314331, + "learning_rate": 1.2939948207714326e-05, + "loss": 2.479742240905762, + "step": 130000 + }, + { + "epoch": 0.35394038383013476, + "grad_norm": 2.071133613586426, + "learning_rate": 1.293885784380537e-05, + "loss": 2.5024087905883787, + "step": 130020 + }, + { + "epoch": 0.35399482782087927, + "grad_norm": 2.089411497116089, + "learning_rate": 1.2937767479896418e-05, + "loss": 2.5170671463012697, + "step": 130040 + }, + { + "epoch": 0.35404927181162377, + "grad_norm": 2.141098976135254, + "learning_rate": 1.2936677115987462e-05, + "loss": 2.496212387084961, + "step": 130060 + }, + { + "epoch": 0.3541037158023683, + "grad_norm": 2.085062265396118, + "learning_rate": 1.2935586752078508e-05, + "loss": 2.5199617385864257, + "step": 130080 + }, + { + "epoch": 0.3541581597931128, + "grad_norm": 2.226404905319214, + "learning_rate": 1.2934496388169554e-05, + "loss": 2.5023223876953127, + "step": 130100 + }, + { + "epoch": 0.3542126037838574, + "grad_norm": 2.291853666305542, + "learning_rate": 1.2933406024260598e-05, + "loss": 2.520332908630371, + "step": 130120 + }, + { + "epoch": 0.3542670477746019, + "grad_norm": 2.0653579235076904, + "learning_rate": 1.2932315660351643e-05, + "loss": 2.477193260192871, + "step": 130140 + }, + { + "epoch": 0.3543214917653464, + "grad_norm": 2.204219102859497, + "learning_rate": 1.293122529644269e-05, + "loss": 2.5110967636108397, + "step": 130160 + }, + { + "epoch": 0.35437593575609094, + "grad_norm": 2.2681825160980225, + "learning_rate": 1.2930134932533733e-05, + "loss": 2.4962026596069338, + "step": 130180 + }, + { + "epoch": 0.35443037974683544, + "grad_norm": 2.1497368812561035, + "learning_rate": 1.2929044568624781e-05, + "loss": 2.530269813537598, + "step": 130200 + }, + { + "epoch": 0.35448482373757995, + "grad_norm": 2.204010248184204, + "learning_rate": 1.2927954204715825e-05, + "loss": 2.484643745422363, + "step": 130220 + }, + { + "epoch": 0.3545392677283245, + "grad_norm": 2.086667776107788, + "learning_rate": 1.2926863840806871e-05, + "loss": 2.50440673828125, + "step": 130240 + }, + { + "epoch": 0.354593711719069, + "grad_norm": 2.1734445095062256, + "learning_rate": 1.2925773476897917e-05, + "loss": 2.488394927978516, + "step": 130260 + }, + { + "epoch": 0.3546481557098135, + "grad_norm": 2.051034688949585, + "learning_rate": 1.292468311298896e-05, + "loss": 2.4810657501220703, + "step": 130280 + }, + { + "epoch": 0.35470259970055806, + "grad_norm": 2.1884710788726807, + "learning_rate": 1.2923592749080007e-05, + "loss": 2.511125373840332, + "step": 130300 + }, + { + "epoch": 0.35475704369130256, + "grad_norm": 2.150648355484009, + "learning_rate": 1.2922502385171052e-05, + "loss": 2.480168342590332, + "step": 130320 + }, + { + "epoch": 0.35481148768204707, + "grad_norm": 2.3609960079193115, + "learning_rate": 1.2921412021262097e-05, + "loss": 2.4959157943725585, + "step": 130340 + }, + { + "epoch": 0.3548659316727916, + "grad_norm": 2.150500774383545, + "learning_rate": 1.2920321657353142e-05, + "loss": 2.5089359283447266, + "step": 130360 + }, + { + "epoch": 0.3549203756635361, + "grad_norm": 2.369276523590088, + "learning_rate": 1.2919231293444188e-05, + "loss": 2.4907203674316407, + "step": 130380 + }, + { + "epoch": 0.3549748196542807, + "grad_norm": 2.1231892108917236, + "learning_rate": 1.2918140929535234e-05, + "loss": 2.483426094055176, + "step": 130400 + }, + { + "epoch": 0.3550292636450252, + "grad_norm": 2.2137601375579834, + "learning_rate": 1.291705056562628e-05, + "loss": 2.530171203613281, + "step": 130420 + }, + { + "epoch": 0.3550837076357697, + "grad_norm": 2.1065876483917236, + "learning_rate": 1.2915960201717324e-05, + "loss": 2.479661750793457, + "step": 130440 + }, + { + "epoch": 0.35513815162651424, + "grad_norm": 2.259934902191162, + "learning_rate": 1.291486983780837e-05, + "loss": 2.498350715637207, + "step": 130460 + }, + { + "epoch": 0.35519259561725874, + "grad_norm": 2.1283042430877686, + "learning_rate": 1.2913779473899415e-05, + "loss": 2.5065196990966796, + "step": 130480 + }, + { + "epoch": 0.35524703960800325, + "grad_norm": 2.3143887519836426, + "learning_rate": 1.291268910999046e-05, + "loss": 2.486530303955078, + "step": 130500 + }, + { + "epoch": 0.3553014835987478, + "grad_norm": 2.2378485202789307, + "learning_rate": 1.2911598746081505e-05, + "loss": 2.4642105102539062, + "step": 130520 + }, + { + "epoch": 0.3553559275894923, + "grad_norm": 2.4517133235931396, + "learning_rate": 1.2910508382172551e-05, + "loss": 2.500164031982422, + "step": 130540 + }, + { + "epoch": 0.3554103715802368, + "grad_norm": 2.063513994216919, + "learning_rate": 1.2909418018263597e-05, + "loss": 2.4893831253051757, + "step": 130560 + }, + { + "epoch": 0.35546481557098136, + "grad_norm": 2.3242099285125732, + "learning_rate": 1.2908327654354643e-05, + "loss": 2.5209428787231447, + "step": 130580 + }, + { + "epoch": 0.35551925956172586, + "grad_norm": 2.172379970550537, + "learning_rate": 1.2907237290445687e-05, + "loss": 2.502865219116211, + "step": 130600 + }, + { + "epoch": 0.3555737035524704, + "grad_norm": 2.2116708755493164, + "learning_rate": 1.2906146926536733e-05, + "loss": 2.5089557647705076, + "step": 130620 + }, + { + "epoch": 0.3556281475432149, + "grad_norm": 2.1464571952819824, + "learning_rate": 1.2905056562627779e-05, + "loss": 2.502817153930664, + "step": 130640 + }, + { + "epoch": 0.3556825915339594, + "grad_norm": 3.3874993324279785, + "learning_rate": 1.2903966198718823e-05, + "loss": 2.5165678024291993, + "step": 130660 + }, + { + "epoch": 0.355737035524704, + "grad_norm": 2.129315137863159, + "learning_rate": 1.2902875834809868e-05, + "loss": 2.4968576431274414, + "step": 130680 + }, + { + "epoch": 0.3557914795154485, + "grad_norm": 2.2623419761657715, + "learning_rate": 1.2901785470900914e-05, + "loss": 2.4935556411743165, + "step": 130700 + }, + { + "epoch": 0.355845923506193, + "grad_norm": 2.0981993675231934, + "learning_rate": 1.290069510699196e-05, + "loss": 2.5021509170532226, + "step": 130720 + }, + { + "epoch": 0.35590036749693754, + "grad_norm": 2.08591628074646, + "learning_rate": 1.2899604743083006e-05, + "loss": 2.4825166702270507, + "step": 130740 + }, + { + "epoch": 0.35595481148768204, + "grad_norm": 2.092782497406006, + "learning_rate": 1.289851437917405e-05, + "loss": 2.482205390930176, + "step": 130760 + }, + { + "epoch": 0.35600925547842655, + "grad_norm": 2.097674608230591, + "learning_rate": 1.2897424015265096e-05, + "loss": 2.5051769256591796, + "step": 130780 + }, + { + "epoch": 0.3560636994691711, + "grad_norm": 2.122797727584839, + "learning_rate": 1.2896333651356142e-05, + "loss": 2.488125038146973, + "step": 130800 + }, + { + "epoch": 0.3561181434599156, + "grad_norm": 2.1034963130950928, + "learning_rate": 1.2895243287447186e-05, + "loss": 2.5056432723999023, + "step": 130820 + }, + { + "epoch": 0.35617258745066016, + "grad_norm": 2.09018874168396, + "learning_rate": 1.2894152923538232e-05, + "loss": 2.50426025390625, + "step": 130840 + }, + { + "epoch": 0.35622703144140466, + "grad_norm": 2.162839889526367, + "learning_rate": 1.2893062559629277e-05, + "loss": 2.5104846954345703, + "step": 130860 + }, + { + "epoch": 0.35628147543214916, + "grad_norm": 2.103412389755249, + "learning_rate": 1.2891972195720323e-05, + "loss": 2.5124465942382814, + "step": 130880 + }, + { + "epoch": 0.3563359194228937, + "grad_norm": 2.242417812347412, + "learning_rate": 1.2890881831811369e-05, + "loss": 2.4784473419189452, + "step": 130900 + }, + { + "epoch": 0.3563903634136382, + "grad_norm": 2.217951536178589, + "learning_rate": 1.2889791467902413e-05, + "loss": 2.474344253540039, + "step": 130920 + }, + { + "epoch": 0.3564448074043827, + "grad_norm": 2.168336868286133, + "learning_rate": 1.2888701103993459e-05, + "loss": 2.4897905349731446, + "step": 130940 + }, + { + "epoch": 0.3564992513951273, + "grad_norm": 3.7163209915161133, + "learning_rate": 1.2887610740084505e-05, + "loss": 2.505392074584961, + "step": 130960 + }, + { + "epoch": 0.3565536953858718, + "grad_norm": 2.0939576625823975, + "learning_rate": 1.2886520376175549e-05, + "loss": 2.4904409408569337, + "step": 130980 + }, + { + "epoch": 0.3566081393766163, + "grad_norm": 2.2262394428253174, + "learning_rate": 1.2885430012266595e-05, + "loss": 2.4978706359863283, + "step": 131000 + }, + { + "epoch": 0.35666258336736084, + "grad_norm": 2.1236751079559326, + "learning_rate": 1.288433964835764e-05, + "loss": 2.506982421875, + "step": 131020 + }, + { + "epoch": 0.35671702735810534, + "grad_norm": 2.0994081497192383, + "learning_rate": 1.2883249284448686e-05, + "loss": 2.522600555419922, + "step": 131040 + }, + { + "epoch": 0.35677147134884984, + "grad_norm": 2.279160976409912, + "learning_rate": 1.2882158920539732e-05, + "loss": 2.4838140487670897, + "step": 131060 + }, + { + "epoch": 0.3568259153395944, + "grad_norm": 3.492009401321411, + "learning_rate": 1.2881068556630776e-05, + "loss": 2.509539031982422, + "step": 131080 + }, + { + "epoch": 0.3568803593303389, + "grad_norm": 2.2811496257781982, + "learning_rate": 1.2879978192721822e-05, + "loss": 2.4868974685668945, + "step": 131100 + }, + { + "epoch": 0.35693480332108346, + "grad_norm": 2.2408254146575928, + "learning_rate": 1.2878887828812868e-05, + "loss": 2.496263122558594, + "step": 131120 + }, + { + "epoch": 0.35698924731182796, + "grad_norm": 2.099853277206421, + "learning_rate": 1.2877797464903912e-05, + "loss": 2.4991498947143556, + "step": 131140 + }, + { + "epoch": 0.35704369130257246, + "grad_norm": 2.3170666694641113, + "learning_rate": 1.2876707100994958e-05, + "loss": 2.5165185928344727, + "step": 131160 + }, + { + "epoch": 0.357098135293317, + "grad_norm": 2.3776841163635254, + "learning_rate": 1.2875616737086004e-05, + "loss": 2.4924320220947265, + "step": 131180 + }, + { + "epoch": 0.3571525792840615, + "grad_norm": 2.1573476791381836, + "learning_rate": 1.287452637317705e-05, + "loss": 2.4994266510009764, + "step": 131200 + }, + { + "epoch": 0.357207023274806, + "grad_norm": 2.2550177574157715, + "learning_rate": 1.2873436009268095e-05, + "loss": 2.501661682128906, + "step": 131220 + }, + { + "epoch": 0.3572614672655506, + "grad_norm": 2.0366852283477783, + "learning_rate": 1.287234564535914e-05, + "loss": 2.5181751251220703, + "step": 131240 + }, + { + "epoch": 0.3573159112562951, + "grad_norm": 2.105825424194336, + "learning_rate": 1.2871255281450185e-05, + "loss": 2.4995214462280275, + "step": 131260 + }, + { + "epoch": 0.3573703552470396, + "grad_norm": 2.247058629989624, + "learning_rate": 1.2870164917541231e-05, + "loss": 2.4535579681396484, + "step": 131280 + }, + { + "epoch": 0.35742479923778414, + "grad_norm": 2.23508620262146, + "learning_rate": 1.2869074553632275e-05, + "loss": 2.488367462158203, + "step": 131300 + }, + { + "epoch": 0.35747924322852864, + "grad_norm": 2.171717643737793, + "learning_rate": 1.286798418972332e-05, + "loss": 2.4912788391113283, + "step": 131320 + }, + { + "epoch": 0.3575336872192732, + "grad_norm": 2.1891674995422363, + "learning_rate": 1.2866893825814367e-05, + "loss": 2.4727060317993166, + "step": 131340 + }, + { + "epoch": 0.3575881312100177, + "grad_norm": 2.0563859939575195, + "learning_rate": 1.2865803461905412e-05, + "loss": 2.5138689041137696, + "step": 131360 + }, + { + "epoch": 0.3576425752007622, + "grad_norm": 2.1351234912872314, + "learning_rate": 1.2864713097996457e-05, + "loss": 2.4869974136352537, + "step": 131380 + }, + { + "epoch": 0.35769701919150676, + "grad_norm": 2.232386350631714, + "learning_rate": 1.2863622734087502e-05, + "loss": 2.5309547424316405, + "step": 131400 + }, + { + "epoch": 0.35775146318225126, + "grad_norm": 2.418158531188965, + "learning_rate": 1.2862532370178548e-05, + "loss": 2.509067916870117, + "step": 131420 + }, + { + "epoch": 0.35780590717299576, + "grad_norm": 2.1564555168151855, + "learning_rate": 1.2861442006269594e-05, + "loss": 2.4970977783203123, + "step": 131440 + }, + { + "epoch": 0.3578603511637403, + "grad_norm": 2.0678508281707764, + "learning_rate": 1.2860351642360638e-05, + "loss": 2.492644500732422, + "step": 131460 + }, + { + "epoch": 0.3579147951544848, + "grad_norm": 2.2743263244628906, + "learning_rate": 1.2859261278451684e-05, + "loss": 2.529087448120117, + "step": 131480 + }, + { + "epoch": 0.3579692391452293, + "grad_norm": 2.107753038406372, + "learning_rate": 1.285817091454273e-05, + "loss": 2.4785947799682617, + "step": 131500 + }, + { + "epoch": 0.3580236831359739, + "grad_norm": 2.0544021129608154, + "learning_rate": 1.2857080550633776e-05, + "loss": 2.4433732986450196, + "step": 131520 + }, + { + "epoch": 0.3580781271267184, + "grad_norm": 2.3075239658355713, + "learning_rate": 1.285599018672482e-05, + "loss": 2.4821638107299804, + "step": 131540 + }, + { + "epoch": 0.3581325711174629, + "grad_norm": 2.0471808910369873, + "learning_rate": 1.2854899822815865e-05, + "loss": 2.4943565368652343, + "step": 131560 + }, + { + "epoch": 0.35818701510820744, + "grad_norm": 2.219444751739502, + "learning_rate": 1.2853809458906911e-05, + "loss": 2.494792938232422, + "step": 131580 + }, + { + "epoch": 0.35824145909895194, + "grad_norm": 2.282949209213257, + "learning_rate": 1.2852719094997957e-05, + "loss": 2.509268951416016, + "step": 131600 + }, + { + "epoch": 0.3582959030896965, + "grad_norm": 2.482642889022827, + "learning_rate": 1.2851628731089003e-05, + "loss": 2.521650505065918, + "step": 131620 + }, + { + "epoch": 0.358350347080441, + "grad_norm": 2.1213693618774414, + "learning_rate": 1.2850538367180047e-05, + "loss": 2.4667118072509764, + "step": 131640 + }, + { + "epoch": 0.3584047910711855, + "grad_norm": 2.3132457733154297, + "learning_rate": 1.2849448003271093e-05, + "loss": 2.4986379623413084, + "step": 131660 + }, + { + "epoch": 0.35845923506193006, + "grad_norm": 2.250490427017212, + "learning_rate": 1.2848357639362139e-05, + "loss": 2.5096969604492188, + "step": 131680 + }, + { + "epoch": 0.35851367905267456, + "grad_norm": 2.041191577911377, + "learning_rate": 1.2847267275453183e-05, + "loss": 2.4904172897338865, + "step": 131700 + }, + { + "epoch": 0.35856812304341906, + "grad_norm": 2.369194746017456, + "learning_rate": 1.2846176911544229e-05, + "loss": 2.4714250564575195, + "step": 131720 + }, + { + "epoch": 0.3586225670341636, + "grad_norm": 2.300426721572876, + "learning_rate": 1.2845086547635274e-05, + "loss": 2.490497016906738, + "step": 131740 + }, + { + "epoch": 0.3586770110249081, + "grad_norm": 2.335881471633911, + "learning_rate": 1.284399618372632e-05, + "loss": 2.520852279663086, + "step": 131760 + }, + { + "epoch": 0.3587314550156526, + "grad_norm": 3.7960145473480225, + "learning_rate": 1.2842905819817366e-05, + "loss": 2.4673368453979494, + "step": 131780 + }, + { + "epoch": 0.3587858990063972, + "grad_norm": 2.696305751800537, + "learning_rate": 1.284181545590841e-05, + "loss": 2.4737836837768556, + "step": 131800 + }, + { + "epoch": 0.3588403429971417, + "grad_norm": 2.6798501014709473, + "learning_rate": 1.2840725091999456e-05, + "loss": 2.498764991760254, + "step": 131820 + }, + { + "epoch": 0.35889478698788624, + "grad_norm": 2.218410015106201, + "learning_rate": 1.2839634728090502e-05, + "loss": 2.496624755859375, + "step": 131840 + }, + { + "epoch": 0.35894923097863074, + "grad_norm": 2.2623984813690186, + "learning_rate": 1.2838544364181546e-05, + "loss": 2.4786727905273436, + "step": 131860 + }, + { + "epoch": 0.35900367496937524, + "grad_norm": 2.0935773849487305, + "learning_rate": 1.2837454000272592e-05, + "loss": 2.5431539535522463, + "step": 131880 + }, + { + "epoch": 0.3590581189601198, + "grad_norm": 2.012429714202881, + "learning_rate": 1.2836363636363637e-05, + "loss": 2.480046844482422, + "step": 131900 + }, + { + "epoch": 0.3591125629508643, + "grad_norm": 2.175137519836426, + "learning_rate": 1.2835273272454683e-05, + "loss": 2.476251792907715, + "step": 131920 + }, + { + "epoch": 0.3591670069416088, + "grad_norm": 2.2316513061523438, + "learning_rate": 1.2834182908545729e-05, + "loss": 2.5082071304321287, + "step": 131940 + }, + { + "epoch": 0.35922145093235336, + "grad_norm": 2.337855100631714, + "learning_rate": 1.2833092544636773e-05, + "loss": 2.464889717102051, + "step": 131960 + }, + { + "epoch": 0.35927589492309786, + "grad_norm": 2.121889352798462, + "learning_rate": 1.2832002180727819e-05, + "loss": 2.4626874923706055, + "step": 131980 + }, + { + "epoch": 0.35933033891384236, + "grad_norm": 2.138411045074463, + "learning_rate": 1.2830911816818865e-05, + "loss": 2.5074750900268556, + "step": 132000 + }, + { + "epoch": 0.3593847829045869, + "grad_norm": 2.2520527839660645, + "learning_rate": 1.2829821452909909e-05, + "loss": 2.4997182846069337, + "step": 132020 + }, + { + "epoch": 0.3594392268953314, + "grad_norm": 2.5698609352111816, + "learning_rate": 1.2828731089000955e-05, + "loss": 2.4861982345581053, + "step": 132040 + }, + { + "epoch": 0.3594936708860759, + "grad_norm": 4.9664106369018555, + "learning_rate": 1.2827640725092e-05, + "loss": 2.5092609405517576, + "step": 132060 + }, + { + "epoch": 0.3595481148768205, + "grad_norm": 2.1820976734161377, + "learning_rate": 1.2826550361183046e-05, + "loss": 2.4929763793945314, + "step": 132080 + }, + { + "epoch": 0.359602558867565, + "grad_norm": 2.137921094894409, + "learning_rate": 1.2825459997274092e-05, + "loss": 2.498104476928711, + "step": 132100 + }, + { + "epoch": 0.35965700285830954, + "grad_norm": 2.1796035766601562, + "learning_rate": 1.2824369633365136e-05, + "loss": 2.4715997695922853, + "step": 132120 + }, + { + "epoch": 0.35971144684905404, + "grad_norm": 2.2553675174713135, + "learning_rate": 1.2823279269456182e-05, + "loss": 2.524959182739258, + "step": 132140 + }, + { + "epoch": 0.35976589083979854, + "grad_norm": 2.142568349838257, + "learning_rate": 1.2822188905547228e-05, + "loss": 2.496431350708008, + "step": 132160 + }, + { + "epoch": 0.3598203348305431, + "grad_norm": 2.1578404903411865, + "learning_rate": 1.2821098541638272e-05, + "loss": 2.5043006896972657, + "step": 132180 + }, + { + "epoch": 0.3598747788212876, + "grad_norm": 2.146859884262085, + "learning_rate": 1.2820008177729318e-05, + "loss": 2.4702659606933595, + "step": 132200 + }, + { + "epoch": 0.3599292228120321, + "grad_norm": 2.1435344219207764, + "learning_rate": 1.2818917813820364e-05, + "loss": 2.5064250946044924, + "step": 132220 + }, + { + "epoch": 0.35998366680277666, + "grad_norm": 2.9386982917785645, + "learning_rate": 1.281782744991141e-05, + "loss": 2.4883737564086914, + "step": 132240 + }, + { + "epoch": 0.36003811079352116, + "grad_norm": 2.2399520874023438, + "learning_rate": 1.2816737086002455e-05, + "loss": 2.4894840240478517, + "step": 132260 + }, + { + "epoch": 0.36009255478426566, + "grad_norm": 2.088796377182007, + "learning_rate": 1.28156467220935e-05, + "loss": 2.5081640243530274, + "step": 132280 + }, + { + "epoch": 0.3601469987750102, + "grad_norm": 2.2783093452453613, + "learning_rate": 1.2814556358184545e-05, + "loss": 2.537446975708008, + "step": 132300 + }, + { + "epoch": 0.3602014427657547, + "grad_norm": 2.183469772338867, + "learning_rate": 1.2813465994275591e-05, + "loss": 2.5130029678344727, + "step": 132320 + }, + { + "epoch": 0.3602558867564993, + "grad_norm": 2.1136221885681152, + "learning_rate": 1.2812375630366635e-05, + "loss": 2.506278419494629, + "step": 132340 + }, + { + "epoch": 0.3603103307472438, + "grad_norm": 2.7726919651031494, + "learning_rate": 1.2811285266457681e-05, + "loss": 2.4817028045654297, + "step": 132360 + }, + { + "epoch": 0.3603647747379883, + "grad_norm": 2.0920467376708984, + "learning_rate": 1.2810194902548727e-05, + "loss": 2.52120418548584, + "step": 132380 + }, + { + "epoch": 0.36041921872873284, + "grad_norm": 2.107175588607788, + "learning_rate": 1.2809104538639773e-05, + "loss": 2.4924690246582033, + "step": 132400 + }, + { + "epoch": 0.36047366271947734, + "grad_norm": 2.0877814292907715, + "learning_rate": 1.2808014174730818e-05, + "loss": 2.4994373321533203, + "step": 132420 + }, + { + "epoch": 0.36052810671022184, + "grad_norm": 2.220104217529297, + "learning_rate": 1.2806923810821862e-05, + "loss": 2.4972118377685546, + "step": 132440 + }, + { + "epoch": 0.3605825507009664, + "grad_norm": 2.2244386672973633, + "learning_rate": 1.2805833446912908e-05, + "loss": 2.4765512466430666, + "step": 132460 + }, + { + "epoch": 0.3606369946917109, + "grad_norm": 2.1558191776275635, + "learning_rate": 1.2804743083003954e-05, + "loss": 2.505250358581543, + "step": 132480 + }, + { + "epoch": 0.3606914386824554, + "grad_norm": 2.755676031112671, + "learning_rate": 1.2803652719094998e-05, + "loss": 2.5075782775878905, + "step": 132500 + }, + { + "epoch": 0.36074588267319996, + "grad_norm": 2.1264874935150146, + "learning_rate": 1.2802562355186044e-05, + "loss": 2.4880655288696287, + "step": 132520 + }, + { + "epoch": 0.36080032666394446, + "grad_norm": 2.1806015968322754, + "learning_rate": 1.280147199127709e-05, + "loss": 2.4926130294799806, + "step": 132540 + }, + { + "epoch": 0.36085477065468896, + "grad_norm": 2.1956329345703125, + "learning_rate": 1.2800381627368134e-05, + "loss": 2.5261850357055664, + "step": 132560 + }, + { + "epoch": 0.3609092146454335, + "grad_norm": 2.0995497703552246, + "learning_rate": 1.2799291263459181e-05, + "loss": 2.462843322753906, + "step": 132580 + }, + { + "epoch": 0.360963658636178, + "grad_norm": 2.091571092605591, + "learning_rate": 1.2798200899550226e-05, + "loss": 2.4998476028442385, + "step": 132600 + }, + { + "epoch": 0.3610181026269226, + "grad_norm": 2.1615493297576904, + "learning_rate": 1.2797110535641271e-05, + "loss": 2.507242965698242, + "step": 132620 + }, + { + "epoch": 0.3610725466176671, + "grad_norm": 2.1707828044891357, + "learning_rate": 1.2796020171732317e-05, + "loss": 2.5081533432006835, + "step": 132640 + }, + { + "epoch": 0.3611269906084116, + "grad_norm": 2.1705007553100586, + "learning_rate": 1.2794929807823361e-05, + "loss": 2.4953861236572266, + "step": 132660 + }, + { + "epoch": 0.36118143459915614, + "grad_norm": 2.257392406463623, + "learning_rate": 1.2793839443914407e-05, + "loss": 2.4808422088623048, + "step": 132680 + }, + { + "epoch": 0.36123587858990064, + "grad_norm": 2.09881329536438, + "learning_rate": 1.2792749080005453e-05, + "loss": 2.4970304489135744, + "step": 132700 + }, + { + "epoch": 0.36129032258064514, + "grad_norm": 2.1874794960021973, + "learning_rate": 1.2791658716096497e-05, + "loss": 2.5358470916748046, + "step": 132720 + }, + { + "epoch": 0.3613447665713897, + "grad_norm": 2.1401267051696777, + "learning_rate": 1.2790568352187544e-05, + "loss": 2.4788169860839844, + "step": 132740 + }, + { + "epoch": 0.3613992105621342, + "grad_norm": 2.092005491256714, + "learning_rate": 1.2789477988278589e-05, + "loss": 2.506173324584961, + "step": 132760 + }, + { + "epoch": 0.3614536545528787, + "grad_norm": 2.1330270767211914, + "learning_rate": 1.2788387624369634e-05, + "loss": 2.502847099304199, + "step": 132780 + }, + { + "epoch": 0.36150809854362326, + "grad_norm": 2.152409076690674, + "learning_rate": 1.278729726046068e-05, + "loss": 2.5036943435668944, + "step": 132800 + }, + { + "epoch": 0.36156254253436776, + "grad_norm": 2.1962149143218994, + "learning_rate": 1.2786206896551724e-05, + "loss": 2.5038131713867187, + "step": 132820 + }, + { + "epoch": 0.3616169865251123, + "grad_norm": 2.2002274990081787, + "learning_rate": 1.278511653264277e-05, + "loss": 2.499689483642578, + "step": 132840 + }, + { + "epoch": 0.3616714305158568, + "grad_norm": 2.060401439666748, + "learning_rate": 1.2784026168733816e-05, + "loss": 2.487620544433594, + "step": 132860 + }, + { + "epoch": 0.3617258745066013, + "grad_norm": 2.0904059410095215, + "learning_rate": 1.278293580482486e-05, + "loss": 2.5083740234375, + "step": 132880 + }, + { + "epoch": 0.3617803184973459, + "grad_norm": 2.1228277683258057, + "learning_rate": 1.2781845440915908e-05, + "loss": 2.543710136413574, + "step": 132900 + }, + { + "epoch": 0.3618347624880904, + "grad_norm": 2.1423423290252686, + "learning_rate": 1.2780755077006952e-05, + "loss": 2.5172666549682616, + "step": 132920 + }, + { + "epoch": 0.3618892064788349, + "grad_norm": 2.180921792984009, + "learning_rate": 1.2779664713097997e-05, + "loss": 2.4814891815185547, + "step": 132940 + }, + { + "epoch": 0.36194365046957944, + "grad_norm": 2.1026201248168945, + "learning_rate": 1.2778574349189043e-05, + "loss": 2.496881294250488, + "step": 132960 + }, + { + "epoch": 0.36199809446032394, + "grad_norm": 2.4304940700531006, + "learning_rate": 1.2777483985280087e-05, + "loss": 2.4980117797851564, + "step": 132980 + }, + { + "epoch": 0.36205253845106844, + "grad_norm": 2.1735687255859375, + "learning_rate": 1.2776393621371133e-05, + "loss": 2.5019805908203123, + "step": 133000 + }, + { + "epoch": 0.362106982441813, + "grad_norm": 2.4323863983154297, + "learning_rate": 1.2775303257462179e-05, + "loss": 2.48742790222168, + "step": 133020 + }, + { + "epoch": 0.3621614264325575, + "grad_norm": 2.0969595909118652, + "learning_rate": 1.2774212893553223e-05, + "loss": 2.5121822357177734, + "step": 133040 + }, + { + "epoch": 0.36221587042330206, + "grad_norm": 2.1175076961517334, + "learning_rate": 1.277312252964427e-05, + "loss": 2.517389488220215, + "step": 133060 + }, + { + "epoch": 0.36227031441404656, + "grad_norm": 2.125549077987671, + "learning_rate": 1.2772032165735315e-05, + "loss": 2.4883657455444337, + "step": 133080 + }, + { + "epoch": 0.36232475840479106, + "grad_norm": 2.1980514526367188, + "learning_rate": 1.277094180182636e-05, + "loss": 2.4842508316040037, + "step": 133100 + }, + { + "epoch": 0.3623792023955356, + "grad_norm": 3.9038119316101074, + "learning_rate": 1.2769851437917406e-05, + "loss": 2.4881282806396485, + "step": 133120 + }, + { + "epoch": 0.3624336463862801, + "grad_norm": 2.12774658203125, + "learning_rate": 1.276876107400845e-05, + "loss": 2.4837066650390627, + "step": 133140 + }, + { + "epoch": 0.3624880903770246, + "grad_norm": 2.0942745208740234, + "learning_rate": 1.2767670710099496e-05, + "loss": 2.5018138885498047, + "step": 133160 + }, + { + "epoch": 0.3625425343677692, + "grad_norm": 2.109518527984619, + "learning_rate": 1.2766580346190542e-05, + "loss": 2.5236745834350587, + "step": 133180 + }, + { + "epoch": 0.3625969783585137, + "grad_norm": 2.2033400535583496, + "learning_rate": 1.2765489982281586e-05, + "loss": 2.5191776275634767, + "step": 133200 + }, + { + "epoch": 0.3626514223492582, + "grad_norm": 2.159372329711914, + "learning_rate": 1.2764399618372634e-05, + "loss": 2.4957206726074217, + "step": 133220 + }, + { + "epoch": 0.36270586634000274, + "grad_norm": 2.2879602909088135, + "learning_rate": 1.2763309254463678e-05, + "loss": 2.5144363403320313, + "step": 133240 + }, + { + "epoch": 0.36276031033074724, + "grad_norm": 2.2345645427703857, + "learning_rate": 1.2762218890554724e-05, + "loss": 2.482941818237305, + "step": 133260 + }, + { + "epoch": 0.36281475432149174, + "grad_norm": 2.3452093601226807, + "learning_rate": 1.276112852664577e-05, + "loss": 2.5200653076171875, + "step": 133280 + }, + { + "epoch": 0.3628691983122363, + "grad_norm": 2.163506269454956, + "learning_rate": 1.2760038162736814e-05, + "loss": 2.4891103744506835, + "step": 133300 + }, + { + "epoch": 0.3629236423029808, + "grad_norm": 2.104597330093384, + "learning_rate": 1.2758947798827861e-05, + "loss": 2.5068836212158203, + "step": 133320 + }, + { + "epoch": 0.36297808629372535, + "grad_norm": 2.1378824710845947, + "learning_rate": 1.2757857434918905e-05, + "loss": 2.465675163269043, + "step": 133340 + }, + { + "epoch": 0.36303253028446986, + "grad_norm": 2.177988290786743, + "learning_rate": 1.275676707100995e-05, + "loss": 2.490254211425781, + "step": 133360 + }, + { + "epoch": 0.36308697427521436, + "grad_norm": 2.4652669429779053, + "learning_rate": 1.2755676707100997e-05, + "loss": 2.5067848205566405, + "step": 133380 + }, + { + "epoch": 0.3631414182659589, + "grad_norm": 2.293083906173706, + "learning_rate": 1.2754586343192041e-05, + "loss": 2.493925666809082, + "step": 133400 + }, + { + "epoch": 0.3631958622567034, + "grad_norm": 2.1307287216186523, + "learning_rate": 1.2753495979283087e-05, + "loss": 2.471676826477051, + "step": 133420 + }, + { + "epoch": 0.3632503062474479, + "grad_norm": 2.1974358558654785, + "learning_rate": 1.2752405615374133e-05, + "loss": 2.515768623352051, + "step": 133440 + }, + { + "epoch": 0.3633047502381925, + "grad_norm": 2.0927276611328125, + "learning_rate": 1.2751315251465177e-05, + "loss": 2.4931509017944338, + "step": 133460 + }, + { + "epoch": 0.363359194228937, + "grad_norm": 2.0540521144866943, + "learning_rate": 1.2750224887556224e-05, + "loss": 2.4876729965209963, + "step": 133480 + }, + { + "epoch": 0.3634136382196815, + "grad_norm": 2.864015579223633, + "learning_rate": 1.2749134523647268e-05, + "loss": 2.5056968688964845, + "step": 133500 + }, + { + "epoch": 0.36346808221042604, + "grad_norm": 2.101531744003296, + "learning_rate": 1.2748044159738312e-05, + "loss": 2.5059574127197264, + "step": 133520 + }, + { + "epoch": 0.36352252620117054, + "grad_norm": 2.1492676734924316, + "learning_rate": 1.274695379582936e-05, + "loss": 2.4906732559204103, + "step": 133540 + }, + { + "epoch": 0.3635769701919151, + "grad_norm": 2.0619661808013916, + "learning_rate": 1.2745863431920404e-05, + "loss": 2.480690765380859, + "step": 133560 + }, + { + "epoch": 0.3636314141826596, + "grad_norm": 2.186091661453247, + "learning_rate": 1.2744773068011448e-05, + "loss": 2.4554187774658205, + "step": 133580 + }, + { + "epoch": 0.3636858581734041, + "grad_norm": 2.148224353790283, + "learning_rate": 1.2743682704102496e-05, + "loss": 2.5174484252929688, + "step": 133600 + }, + { + "epoch": 0.36374030216414865, + "grad_norm": 2.2461538314819336, + "learning_rate": 1.274259234019354e-05, + "loss": 2.499434471130371, + "step": 133620 + }, + { + "epoch": 0.36379474615489316, + "grad_norm": 2.154341459274292, + "learning_rate": 1.2741501976284587e-05, + "loss": 2.4906558990478516, + "step": 133640 + }, + { + "epoch": 0.36384919014563766, + "grad_norm": 2.203188896179199, + "learning_rate": 1.2740411612375631e-05, + "loss": 2.4798803329467773, + "step": 133660 + }, + { + "epoch": 0.3639036341363822, + "grad_norm": 2.259868621826172, + "learning_rate": 1.2739321248466675e-05, + "loss": 2.5036336898803713, + "step": 133680 + }, + { + "epoch": 0.3639580781271267, + "grad_norm": 2.1352391242980957, + "learning_rate": 1.2738230884557723e-05, + "loss": 2.5009990692138673, + "step": 133700 + }, + { + "epoch": 0.3640125221178712, + "grad_norm": 2.432521343231201, + "learning_rate": 1.2737140520648767e-05, + "loss": 2.518321418762207, + "step": 133720 + }, + { + "epoch": 0.3640669661086158, + "grad_norm": 2.2737932205200195, + "learning_rate": 1.2736050156739811e-05, + "loss": 2.512497138977051, + "step": 133740 + }, + { + "epoch": 0.3641214100993603, + "grad_norm": 2.179896593093872, + "learning_rate": 1.2734959792830859e-05, + "loss": 2.4759925842285155, + "step": 133760 + }, + { + "epoch": 0.3641758540901048, + "grad_norm": 2.2550036907196045, + "learning_rate": 1.2733869428921903e-05, + "loss": 2.483336639404297, + "step": 133780 + }, + { + "epoch": 0.36423029808084934, + "grad_norm": 2.034179925918579, + "learning_rate": 1.273277906501295e-05, + "loss": 2.5445798873901366, + "step": 133800 + }, + { + "epoch": 0.36428474207159384, + "grad_norm": 2.124701499938965, + "learning_rate": 1.2731688701103994e-05, + "loss": 2.523076629638672, + "step": 133820 + }, + { + "epoch": 0.3643391860623384, + "grad_norm": 2.136411666870117, + "learning_rate": 1.2730598337195039e-05, + "loss": 2.4988733291625977, + "step": 133840 + }, + { + "epoch": 0.3643936300530829, + "grad_norm": 2.2501823902130127, + "learning_rate": 1.2729507973286086e-05, + "loss": 2.505929374694824, + "step": 133860 + }, + { + "epoch": 0.3644480740438274, + "grad_norm": 2.1231627464294434, + "learning_rate": 1.272841760937713e-05, + "loss": 2.4689661026000977, + "step": 133880 + }, + { + "epoch": 0.36450251803457195, + "grad_norm": 2.1174607276916504, + "learning_rate": 1.2727327245468174e-05, + "loss": 2.476275825500488, + "step": 133900 + }, + { + "epoch": 0.36455696202531646, + "grad_norm": 2.101085662841797, + "learning_rate": 1.2726236881559222e-05, + "loss": 2.4924768447875976, + "step": 133920 + }, + { + "epoch": 0.36461140601606096, + "grad_norm": 2.1196069717407227, + "learning_rate": 1.2725146517650266e-05, + "loss": 2.5058908462524414, + "step": 133940 + }, + { + "epoch": 0.3646658500068055, + "grad_norm": 2.2885618209838867, + "learning_rate": 1.2724056153741313e-05, + "loss": 2.5032651901245115, + "step": 133960 + }, + { + "epoch": 0.36472029399755, + "grad_norm": 2.138157844543457, + "learning_rate": 1.2722965789832358e-05, + "loss": 2.46301326751709, + "step": 133980 + }, + { + "epoch": 0.3647747379882945, + "grad_norm": 2.2124409675598145, + "learning_rate": 1.2721875425923402e-05, + "loss": 2.509273147583008, + "step": 134000 + }, + { + "epoch": 0.3648291819790391, + "grad_norm": 2.231471061706543, + "learning_rate": 1.272078506201445e-05, + "loss": 2.5290456771850587, + "step": 134020 + }, + { + "epoch": 0.3648836259697836, + "grad_norm": 2.488241672515869, + "learning_rate": 1.2719694698105493e-05, + "loss": 2.5058931350708007, + "step": 134040 + }, + { + "epoch": 0.36493806996052813, + "grad_norm": 2.195297956466675, + "learning_rate": 1.2718604334196537e-05, + "loss": 2.5137189865112304, + "step": 134060 + }, + { + "epoch": 0.36499251395127263, + "grad_norm": 2.1801254749298096, + "learning_rate": 1.2717513970287585e-05, + "loss": 2.4833194732666017, + "step": 134080 + }, + { + "epoch": 0.36504695794201714, + "grad_norm": 2.224893569946289, + "learning_rate": 1.2716423606378629e-05, + "loss": 2.500419807434082, + "step": 134100 + }, + { + "epoch": 0.3651014019327617, + "grad_norm": 2.4016332626342773, + "learning_rate": 1.2715333242469677e-05, + "loss": 2.4731273651123047, + "step": 134120 + }, + { + "epoch": 0.3651558459235062, + "grad_norm": 2.2935070991516113, + "learning_rate": 1.271424287856072e-05, + "loss": 2.495537757873535, + "step": 134140 + }, + { + "epoch": 0.3652102899142507, + "grad_norm": 2.212911367416382, + "learning_rate": 1.2713152514651765e-05, + "loss": 2.49706916809082, + "step": 134160 + }, + { + "epoch": 0.36526473390499525, + "grad_norm": 2.245572090148926, + "learning_rate": 1.2712062150742812e-05, + "loss": 2.484086036682129, + "step": 134180 + }, + { + "epoch": 0.36531917789573976, + "grad_norm": 2.0912914276123047, + "learning_rate": 1.2710971786833856e-05, + "loss": 2.495344543457031, + "step": 134200 + }, + { + "epoch": 0.36537362188648426, + "grad_norm": 2.162656784057617, + "learning_rate": 1.27098814229249e-05, + "loss": 2.5012784957885743, + "step": 134220 + }, + { + "epoch": 0.3654280658772288, + "grad_norm": 2.1313605308532715, + "learning_rate": 1.2708791059015948e-05, + "loss": 2.486721992492676, + "step": 134240 + }, + { + "epoch": 0.3654825098679733, + "grad_norm": 3.026827096939087, + "learning_rate": 1.2707700695106992e-05, + "loss": 2.4671356201171877, + "step": 134260 + }, + { + "epoch": 0.3655369538587178, + "grad_norm": 2.1034743785858154, + "learning_rate": 1.270661033119804e-05, + "loss": 2.492789459228516, + "step": 134280 + }, + { + "epoch": 0.3655913978494624, + "grad_norm": 2.138166904449463, + "learning_rate": 1.2705519967289084e-05, + "loss": 2.487455368041992, + "step": 134300 + }, + { + "epoch": 0.3656458418402069, + "grad_norm": 2.1056535243988037, + "learning_rate": 1.2704429603380128e-05, + "loss": 2.48391056060791, + "step": 134320 + }, + { + "epoch": 0.36570028583095143, + "grad_norm": 2.206141948699951, + "learning_rate": 1.2703339239471175e-05, + "loss": 2.480434036254883, + "step": 134340 + }, + { + "epoch": 0.36575472982169593, + "grad_norm": 2.1781561374664307, + "learning_rate": 1.270224887556222e-05, + "loss": 2.5139186859130858, + "step": 134360 + }, + { + "epoch": 0.36580917381244044, + "grad_norm": 2.422483205795288, + "learning_rate": 1.2701158511653264e-05, + "loss": 2.513515663146973, + "step": 134380 + }, + { + "epoch": 0.365863617803185, + "grad_norm": 2.1013519763946533, + "learning_rate": 1.2700068147744311e-05, + "loss": 2.5222393035888673, + "step": 134400 + }, + { + "epoch": 0.3659180617939295, + "grad_norm": 2.21832275390625, + "learning_rate": 1.2698977783835355e-05, + "loss": 2.4967723846435548, + "step": 134420 + }, + { + "epoch": 0.365972505784674, + "grad_norm": 2.5060038566589355, + "learning_rate": 1.2697887419926403e-05, + "loss": 2.5053905487060546, + "step": 134440 + }, + { + "epoch": 0.36602694977541855, + "grad_norm": 2.218419075012207, + "learning_rate": 1.2696797056017447e-05, + "loss": 2.497446632385254, + "step": 134460 + }, + { + "epoch": 0.36608139376616305, + "grad_norm": 2.1307125091552734, + "learning_rate": 1.2695706692108491e-05, + "loss": 2.5013025283813475, + "step": 134480 + }, + { + "epoch": 0.36613583775690756, + "grad_norm": 2.1670684814453125, + "learning_rate": 1.2694616328199538e-05, + "loss": 2.509551239013672, + "step": 134500 + }, + { + "epoch": 0.3661902817476521, + "grad_norm": 2.1283223628997803, + "learning_rate": 1.2693525964290583e-05, + "loss": 2.4712779998779295, + "step": 134520 + }, + { + "epoch": 0.3662447257383966, + "grad_norm": 2.0937979221343994, + "learning_rate": 1.2692435600381627e-05, + "loss": 2.4840791702270506, + "step": 134540 + }, + { + "epoch": 0.36629916972914117, + "grad_norm": 2.074146270751953, + "learning_rate": 1.2691345236472674e-05, + "loss": 2.5182672500610352, + "step": 134560 + }, + { + "epoch": 0.3663536137198857, + "grad_norm": 2.1128990650177, + "learning_rate": 1.2690254872563718e-05, + "loss": 2.491483688354492, + "step": 134580 + }, + { + "epoch": 0.3664080577106302, + "grad_norm": 2.245863676071167, + "learning_rate": 1.2689164508654766e-05, + "loss": 2.532155227661133, + "step": 134600 + }, + { + "epoch": 0.36646250170137473, + "grad_norm": 2.1301233768463135, + "learning_rate": 1.268807414474581e-05, + "loss": 2.5008066177368162, + "step": 134620 + }, + { + "epoch": 0.36651694569211923, + "grad_norm": 2.152427911758423, + "learning_rate": 1.2686983780836854e-05, + "loss": 2.518301582336426, + "step": 134640 + }, + { + "epoch": 0.36657138968286374, + "grad_norm": 2.1856861114501953, + "learning_rate": 1.2685893416927902e-05, + "loss": 2.514242935180664, + "step": 134660 + }, + { + "epoch": 0.3666258336736083, + "grad_norm": 2.1977438926696777, + "learning_rate": 1.2684803053018946e-05, + "loss": 2.474105453491211, + "step": 134680 + }, + { + "epoch": 0.3666802776643528, + "grad_norm": 2.0866329669952393, + "learning_rate": 1.268371268910999e-05, + "loss": 2.500775146484375, + "step": 134700 + }, + { + "epoch": 0.3667347216550973, + "grad_norm": 2.14302921295166, + "learning_rate": 1.2682622325201037e-05, + "loss": 2.5169902801513673, + "step": 134720 + }, + { + "epoch": 0.36678916564584185, + "grad_norm": 2.154711961746216, + "learning_rate": 1.2681531961292081e-05, + "loss": 2.5259941101074217, + "step": 134740 + }, + { + "epoch": 0.36684360963658635, + "grad_norm": 2.1185126304626465, + "learning_rate": 1.2680441597383129e-05, + "loss": 2.494559669494629, + "step": 134760 + }, + { + "epoch": 0.36689805362733086, + "grad_norm": 2.138367176055908, + "learning_rate": 1.2679351233474173e-05, + "loss": 2.5023597717285155, + "step": 134780 + }, + { + "epoch": 0.3669524976180754, + "grad_norm": 2.2024242877960205, + "learning_rate": 1.2678260869565217e-05, + "loss": 2.5355329513549805, + "step": 134800 + }, + { + "epoch": 0.3670069416088199, + "grad_norm": 2.110651969909668, + "learning_rate": 1.2677170505656265e-05, + "loss": 2.465671157836914, + "step": 134820 + }, + { + "epoch": 0.36706138559956447, + "grad_norm": 2.1732351779937744, + "learning_rate": 1.2676080141747309e-05, + "loss": 2.5034515380859377, + "step": 134840 + }, + { + "epoch": 0.367115829590309, + "grad_norm": 2.0500388145446777, + "learning_rate": 1.2674989777838353e-05, + "loss": 2.515898323059082, + "step": 134860 + }, + { + "epoch": 0.3671702735810535, + "grad_norm": 2.1842939853668213, + "learning_rate": 1.26738994139294e-05, + "loss": 2.5014848709106445, + "step": 134880 + }, + { + "epoch": 0.36722471757179803, + "grad_norm": 2.266526222229004, + "learning_rate": 1.2672809050020444e-05, + "loss": 2.515818786621094, + "step": 134900 + }, + { + "epoch": 0.36727916156254253, + "grad_norm": 2.1537771224975586, + "learning_rate": 1.2671718686111492e-05, + "loss": 2.4968425750732424, + "step": 134920 + }, + { + "epoch": 0.36733360555328703, + "grad_norm": 2.119896173477173, + "learning_rate": 1.2670628322202536e-05, + "loss": 2.5038497924804686, + "step": 134940 + }, + { + "epoch": 0.3673880495440316, + "grad_norm": 2.148798942565918, + "learning_rate": 1.266953795829358e-05, + "loss": 2.489773750305176, + "step": 134960 + }, + { + "epoch": 0.3674424935347761, + "grad_norm": 2.268364429473877, + "learning_rate": 1.2668447594384628e-05, + "loss": 2.4781864166259764, + "step": 134980 + }, + { + "epoch": 0.3674969375255206, + "grad_norm": 2.097703218460083, + "learning_rate": 1.2667357230475672e-05, + "loss": 2.484794807434082, + "step": 135000 + }, + { + "epoch": 0.36755138151626515, + "grad_norm": 2.1328673362731934, + "learning_rate": 1.266626686656672e-05, + "loss": 2.463663673400879, + "step": 135020 + }, + { + "epoch": 0.36760582550700965, + "grad_norm": 3.4699320793151855, + "learning_rate": 1.2665176502657763e-05, + "loss": 2.5081945419311524, + "step": 135040 + }, + { + "epoch": 0.3676602694977542, + "grad_norm": 2.135913610458374, + "learning_rate": 1.2664086138748808e-05, + "loss": 2.4913406372070312, + "step": 135060 + }, + { + "epoch": 0.3677147134884987, + "grad_norm": 2.017455816268921, + "learning_rate": 1.2662995774839855e-05, + "loss": 2.5078369140625, + "step": 135080 + }, + { + "epoch": 0.3677691574792432, + "grad_norm": 2.124715805053711, + "learning_rate": 1.2661905410930899e-05, + "loss": 2.5406803131103515, + "step": 135100 + }, + { + "epoch": 0.36782360146998777, + "grad_norm": 2.279003381729126, + "learning_rate": 1.2660815047021943e-05, + "loss": 2.5040733337402346, + "step": 135120 + }, + { + "epoch": 0.3678780454607323, + "grad_norm": 2.1112120151519775, + "learning_rate": 1.265972468311299e-05, + "loss": 2.4797037124633787, + "step": 135140 + }, + { + "epoch": 0.3679324894514768, + "grad_norm": 2.140484094619751, + "learning_rate": 1.2658634319204035e-05, + "loss": 2.5005760192871094, + "step": 135160 + }, + { + "epoch": 0.36798693344222133, + "grad_norm": 2.1572673320770264, + "learning_rate": 1.2657543955295082e-05, + "loss": 2.5048107147216796, + "step": 135180 + }, + { + "epoch": 0.36804137743296583, + "grad_norm": 2.095618963241577, + "learning_rate": 1.2656453591386127e-05, + "loss": 2.5009946823120117, + "step": 135200 + }, + { + "epoch": 0.36809582142371033, + "grad_norm": 2.336329698562622, + "learning_rate": 1.265536322747717e-05, + "loss": 2.506614112854004, + "step": 135220 + }, + { + "epoch": 0.3681502654144549, + "grad_norm": 2.182786703109741, + "learning_rate": 1.2654272863568218e-05, + "loss": 2.503514862060547, + "step": 135240 + }, + { + "epoch": 0.3682047094051994, + "grad_norm": 2.2003014087677, + "learning_rate": 1.2653182499659262e-05, + "loss": 2.4958751678466795, + "step": 135260 + }, + { + "epoch": 0.36825915339594395, + "grad_norm": 2.136162757873535, + "learning_rate": 1.2652092135750306e-05, + "loss": 2.510301971435547, + "step": 135280 + }, + { + "epoch": 0.36831359738668845, + "grad_norm": 2.140456438064575, + "learning_rate": 1.2651001771841354e-05, + "loss": 2.492928886413574, + "step": 135300 + }, + { + "epoch": 0.36836804137743295, + "grad_norm": 2.1328892707824707, + "learning_rate": 1.2649911407932398e-05, + "loss": 2.5079187393188476, + "step": 135320 + }, + { + "epoch": 0.3684224853681775, + "grad_norm": 2.1988236904144287, + "learning_rate": 1.2648821044023445e-05, + "loss": 2.522018051147461, + "step": 135340 + }, + { + "epoch": 0.368476929358922, + "grad_norm": 2.213731050491333, + "learning_rate": 1.264773068011449e-05, + "loss": 2.5096967697143553, + "step": 135360 + }, + { + "epoch": 0.3685313733496665, + "grad_norm": 2.198676586151123, + "learning_rate": 1.2646640316205534e-05, + "loss": 2.4981733322143556, + "step": 135380 + }, + { + "epoch": 0.36858581734041107, + "grad_norm": 2.1302082538604736, + "learning_rate": 1.2645549952296581e-05, + "loss": 2.495234489440918, + "step": 135400 + }, + { + "epoch": 0.36864026133115557, + "grad_norm": 2.247565269470215, + "learning_rate": 1.2644459588387625e-05, + "loss": 2.493043899536133, + "step": 135420 + }, + { + "epoch": 0.3686947053219001, + "grad_norm": 2.2374205589294434, + "learning_rate": 1.264336922447867e-05, + "loss": 2.51413631439209, + "step": 135440 + }, + { + "epoch": 0.36874914931264463, + "grad_norm": 2.1502463817596436, + "learning_rate": 1.2642278860569717e-05, + "loss": 2.4953119277954103, + "step": 135460 + }, + { + "epoch": 0.36880359330338913, + "grad_norm": 2.153592109680176, + "learning_rate": 1.2641188496660761e-05, + "loss": 2.4886938095092774, + "step": 135480 + }, + { + "epoch": 0.36885803729413363, + "grad_norm": 2.1093969345092773, + "learning_rate": 1.2640098132751809e-05, + "loss": 2.4937667846679688, + "step": 135500 + }, + { + "epoch": 0.3689124812848782, + "grad_norm": 2.1922388076782227, + "learning_rate": 1.2639007768842853e-05, + "loss": 2.5192127227783203, + "step": 135520 + }, + { + "epoch": 0.3689669252756227, + "grad_norm": 2.5254719257354736, + "learning_rate": 1.2637917404933897e-05, + "loss": 2.4693355560302734, + "step": 135540 + }, + { + "epoch": 0.36902136926636725, + "grad_norm": 2.111093282699585, + "learning_rate": 1.2636827041024944e-05, + "loss": 2.497829818725586, + "step": 135560 + }, + { + "epoch": 0.36907581325711175, + "grad_norm": 2.1833479404449463, + "learning_rate": 1.2635736677115988e-05, + "loss": 2.503599166870117, + "step": 135580 + }, + { + "epoch": 0.36913025724785625, + "grad_norm": 2.167307138442993, + "learning_rate": 1.2634646313207033e-05, + "loss": 2.5008392333984375, + "step": 135600 + }, + { + "epoch": 0.3691847012386008, + "grad_norm": 2.100308418273926, + "learning_rate": 1.263355594929808e-05, + "loss": 2.5190378189086915, + "step": 135620 + }, + { + "epoch": 0.3692391452293453, + "grad_norm": 2.0417327880859375, + "learning_rate": 1.2632465585389124e-05, + "loss": 2.5253746032714846, + "step": 135640 + }, + { + "epoch": 0.3692935892200898, + "grad_norm": 2.201122760772705, + "learning_rate": 1.2631375221480172e-05, + "loss": 2.470860481262207, + "step": 135660 + }, + { + "epoch": 0.36934803321083437, + "grad_norm": 2.224412202835083, + "learning_rate": 1.2630284857571216e-05, + "loss": 2.4928005218505858, + "step": 135680 + }, + { + "epoch": 0.36940247720157887, + "grad_norm": 2.123836040496826, + "learning_rate": 1.262919449366226e-05, + "loss": 2.506494140625, + "step": 135700 + }, + { + "epoch": 0.3694569211923234, + "grad_norm": 2.0871386528015137, + "learning_rate": 1.2628104129753307e-05, + "loss": 2.4891288757324217, + "step": 135720 + }, + { + "epoch": 0.36951136518306793, + "grad_norm": 2.0831971168518066, + "learning_rate": 1.2627013765844351e-05, + "loss": 2.4749937057495117, + "step": 135740 + }, + { + "epoch": 0.36956580917381243, + "grad_norm": 2.4519119262695312, + "learning_rate": 1.2625923401935396e-05, + "loss": 2.490242767333984, + "step": 135760 + }, + { + "epoch": 0.369620253164557, + "grad_norm": 2.0982184410095215, + "learning_rate": 1.2624833038026443e-05, + "loss": 2.4832061767578124, + "step": 135780 + }, + { + "epoch": 0.3696746971553015, + "grad_norm": 2.1089868545532227, + "learning_rate": 1.2623742674117487e-05, + "loss": 2.5118167877197264, + "step": 135800 + }, + { + "epoch": 0.369729141146046, + "grad_norm": 2.122296094894409, + "learning_rate": 1.2622652310208535e-05, + "loss": 2.4919883728027346, + "step": 135820 + }, + { + "epoch": 0.36978358513679055, + "grad_norm": 2.111389636993408, + "learning_rate": 1.2621561946299579e-05, + "loss": 2.483042526245117, + "step": 135840 + }, + { + "epoch": 0.36983802912753505, + "grad_norm": 2.273568868637085, + "learning_rate": 1.2620471582390623e-05, + "loss": 2.5385814666748048, + "step": 135860 + }, + { + "epoch": 0.36989247311827955, + "grad_norm": 2.1546247005462646, + "learning_rate": 1.261938121848167e-05, + "loss": 2.486474609375, + "step": 135880 + }, + { + "epoch": 0.3699469171090241, + "grad_norm": 2.060880184173584, + "learning_rate": 1.2618290854572715e-05, + "loss": 2.4891935348510743, + "step": 135900 + }, + { + "epoch": 0.3700013610997686, + "grad_norm": 2.1891660690307617, + "learning_rate": 1.2617200490663759e-05, + "loss": 2.502062225341797, + "step": 135920 + }, + { + "epoch": 0.3700558050905131, + "grad_norm": 2.5768203735351562, + "learning_rate": 1.2616110126754806e-05, + "loss": 2.4671817779541017, + "step": 135940 + }, + { + "epoch": 0.37011024908125767, + "grad_norm": 2.228100061416626, + "learning_rate": 1.261501976284585e-05, + "loss": 2.5241283416748046, + "step": 135960 + }, + { + "epoch": 0.37016469307200217, + "grad_norm": 2.1194560527801514, + "learning_rate": 1.2613929398936898e-05, + "loss": 2.4494083404541014, + "step": 135980 + }, + { + "epoch": 0.3702191370627467, + "grad_norm": 2.2228620052337646, + "learning_rate": 1.2612839035027942e-05, + "loss": 2.5218927383422853, + "step": 136000 + }, + { + "epoch": 0.37027358105349123, + "grad_norm": 2.412209987640381, + "learning_rate": 1.2611748671118986e-05, + "loss": 2.4601308822631838, + "step": 136020 + }, + { + "epoch": 0.37032802504423573, + "grad_norm": 2.2604057788848877, + "learning_rate": 1.2610658307210034e-05, + "loss": 2.4835046768188476, + "step": 136040 + }, + { + "epoch": 0.3703824690349803, + "grad_norm": 2.277616262435913, + "learning_rate": 1.2609567943301078e-05, + "loss": 2.4671504974365233, + "step": 136060 + }, + { + "epoch": 0.3704369130257248, + "grad_norm": 2.2508018016815186, + "learning_rate": 1.2608477579392122e-05, + "loss": 2.4933162689208985, + "step": 136080 + }, + { + "epoch": 0.3704913570164693, + "grad_norm": 2.1405463218688965, + "learning_rate": 1.260738721548317e-05, + "loss": 2.5147846221923826, + "step": 136100 + }, + { + "epoch": 0.37054580100721385, + "grad_norm": 2.0709824562072754, + "learning_rate": 1.2606296851574213e-05, + "loss": 2.490681457519531, + "step": 136120 + }, + { + "epoch": 0.37060024499795835, + "grad_norm": 2.171168327331543, + "learning_rate": 1.2605206487665261e-05, + "loss": 2.4667470932006834, + "step": 136140 + }, + { + "epoch": 0.37065468898870285, + "grad_norm": 2.182337999343872, + "learning_rate": 1.2604116123756305e-05, + "loss": 2.464019775390625, + "step": 136160 + }, + { + "epoch": 0.3707091329794474, + "grad_norm": 2.2272279262542725, + "learning_rate": 1.2603025759847349e-05, + "loss": 2.474535369873047, + "step": 136180 + }, + { + "epoch": 0.3707635769701919, + "grad_norm": 2.1619060039520264, + "learning_rate": 1.2601935395938397e-05, + "loss": 2.522762489318848, + "step": 136200 + }, + { + "epoch": 0.3708180209609364, + "grad_norm": 2.098752498626709, + "learning_rate": 1.260084503202944e-05, + "loss": 2.4834556579589844, + "step": 136220 + }, + { + "epoch": 0.37087246495168097, + "grad_norm": 2.1435234546661377, + "learning_rate": 1.2599754668120485e-05, + "loss": 2.499302291870117, + "step": 136240 + }, + { + "epoch": 0.37092690894242547, + "grad_norm": 2.28383469581604, + "learning_rate": 1.2598664304211532e-05, + "loss": 2.489543151855469, + "step": 136260 + }, + { + "epoch": 0.37098135293317, + "grad_norm": 2.0821890830993652, + "learning_rate": 1.2597573940302576e-05, + "loss": 2.5135250091552734, + "step": 136280 + }, + { + "epoch": 0.37103579692391453, + "grad_norm": 2.108987808227539, + "learning_rate": 1.2596483576393624e-05, + "loss": 2.5103477478027343, + "step": 136300 + }, + { + "epoch": 0.37109024091465903, + "grad_norm": 2.1459343433380127, + "learning_rate": 1.2595393212484668e-05, + "loss": 2.4994686126708983, + "step": 136320 + }, + { + "epoch": 0.3711446849054036, + "grad_norm": 2.2923500537872314, + "learning_rate": 1.2594302848575712e-05, + "loss": 2.5170997619628905, + "step": 136340 + }, + { + "epoch": 0.3711991288961481, + "grad_norm": 2.0994784832000732, + "learning_rate": 1.259321248466676e-05, + "loss": 2.4851558685302733, + "step": 136360 + }, + { + "epoch": 0.3712535728868926, + "grad_norm": 2.2296183109283447, + "learning_rate": 1.2592122120757804e-05, + "loss": 2.507951545715332, + "step": 136380 + }, + { + "epoch": 0.37130801687763715, + "grad_norm": 3.2704389095306396, + "learning_rate": 1.2591031756848848e-05, + "loss": 2.494548797607422, + "step": 136400 + }, + { + "epoch": 0.37136246086838165, + "grad_norm": 2.218505382537842, + "learning_rate": 1.2589941392939895e-05, + "loss": 2.494091606140137, + "step": 136420 + }, + { + "epoch": 0.37141690485912615, + "grad_norm": 2.2980337142944336, + "learning_rate": 1.258885102903094e-05, + "loss": 2.508752632141113, + "step": 136440 + }, + { + "epoch": 0.3714713488498707, + "grad_norm": 2.911170482635498, + "learning_rate": 1.2587760665121987e-05, + "loss": 2.521864318847656, + "step": 136460 + }, + { + "epoch": 0.3715257928406152, + "grad_norm": 2.2131690979003906, + "learning_rate": 1.2586670301213031e-05, + "loss": 2.4970157623291014, + "step": 136480 + }, + { + "epoch": 0.3715802368313597, + "grad_norm": 2.3100104331970215, + "learning_rate": 1.2585579937304075e-05, + "loss": 2.5193124771118165, + "step": 136500 + }, + { + "epoch": 0.37163468082210427, + "grad_norm": 2.2049670219421387, + "learning_rate": 1.2584489573395123e-05, + "loss": 2.4675973892211913, + "step": 136520 + }, + { + "epoch": 0.37168912481284877, + "grad_norm": 2.3060479164123535, + "learning_rate": 1.2583399209486167e-05, + "loss": 2.496468925476074, + "step": 136540 + }, + { + "epoch": 0.3717435688035933, + "grad_norm": 2.15895676612854, + "learning_rate": 1.2582308845577211e-05, + "loss": 2.502028465270996, + "step": 136560 + }, + { + "epoch": 0.37179801279433783, + "grad_norm": 10.004064559936523, + "learning_rate": 1.2581218481668259e-05, + "loss": 2.5079307556152344, + "step": 136580 + }, + { + "epoch": 0.37185245678508233, + "grad_norm": 2.104491710662842, + "learning_rate": 1.2580128117759303e-05, + "loss": 2.4844268798828124, + "step": 136600 + }, + { + "epoch": 0.3719069007758269, + "grad_norm": 2.1553142070770264, + "learning_rate": 1.257903775385035e-05, + "loss": 2.484117317199707, + "step": 136620 + }, + { + "epoch": 0.3719613447665714, + "grad_norm": 2.135061502456665, + "learning_rate": 1.2577947389941394e-05, + "loss": 2.4908233642578126, + "step": 136640 + }, + { + "epoch": 0.3720157887573159, + "grad_norm": 3.3049046993255615, + "learning_rate": 1.2576857026032438e-05, + "loss": 2.4916757583618163, + "step": 136660 + }, + { + "epoch": 0.37207023274806045, + "grad_norm": 5.565262317657471, + "learning_rate": 1.2575766662123486e-05, + "loss": 2.5086187362670898, + "step": 136680 + }, + { + "epoch": 0.37212467673880495, + "grad_norm": 2.082831382751465, + "learning_rate": 1.257467629821453e-05, + "loss": 2.519326591491699, + "step": 136700 + }, + { + "epoch": 0.37217912072954945, + "grad_norm": 2.3369486331939697, + "learning_rate": 1.2573585934305576e-05, + "loss": 2.476823425292969, + "step": 136720 + }, + { + "epoch": 0.372233564720294, + "grad_norm": 2.0434019565582275, + "learning_rate": 1.2572495570396622e-05, + "loss": 2.4981374740600586, + "step": 136740 + }, + { + "epoch": 0.3722880087110385, + "grad_norm": 2.2131643295288086, + "learning_rate": 1.2571405206487666e-05, + "loss": 2.4927173614501954, + "step": 136760 + }, + { + "epoch": 0.37234245270178307, + "grad_norm": 2.176572561264038, + "learning_rate": 1.2570314842578713e-05, + "loss": 2.5069194793701173, + "step": 136780 + }, + { + "epoch": 0.37239689669252757, + "grad_norm": 2.189157724380493, + "learning_rate": 1.2569224478669757e-05, + "loss": 2.495773124694824, + "step": 136800 + }, + { + "epoch": 0.37245134068327207, + "grad_norm": 2.1310575008392334, + "learning_rate": 1.2568134114760801e-05, + "loss": 2.490342712402344, + "step": 136820 + }, + { + "epoch": 0.3725057846740166, + "grad_norm": 2.1257872581481934, + "learning_rate": 1.2567043750851849e-05, + "loss": 2.518673896789551, + "step": 136840 + }, + { + "epoch": 0.37256022866476113, + "grad_norm": 2.1248626708984375, + "learning_rate": 1.2565953386942893e-05, + "loss": 2.5060182571411134, + "step": 136860 + }, + { + "epoch": 0.37261467265550563, + "grad_norm": 2.2628097534179688, + "learning_rate": 1.2564863023033939e-05, + "loss": 2.5106863021850585, + "step": 136880 + }, + { + "epoch": 0.3726691166462502, + "grad_norm": 32.63116455078125, + "learning_rate": 1.2563772659124985e-05, + "loss": 2.4771406173706056, + "step": 136900 + }, + { + "epoch": 0.3727235606369947, + "grad_norm": 2.1819324493408203, + "learning_rate": 1.2562682295216029e-05, + "loss": 2.5285259246826173, + "step": 136920 + }, + { + "epoch": 0.3727780046277392, + "grad_norm": 2.15240478515625, + "learning_rate": 1.2561591931307076e-05, + "loss": 2.5017467498779298, + "step": 136940 + }, + { + "epoch": 0.37283244861848375, + "grad_norm": 2.1544904708862305, + "learning_rate": 1.256050156739812e-05, + "loss": 2.4981842041015625, + "step": 136960 + }, + { + "epoch": 0.37288689260922825, + "grad_norm": 2.075880289077759, + "learning_rate": 1.2559411203489165e-05, + "loss": 2.483830451965332, + "step": 136980 + }, + { + "epoch": 0.37294133659997275, + "grad_norm": 2.156273126602173, + "learning_rate": 1.2558320839580212e-05, + "loss": 2.5040256500244142, + "step": 137000 + }, + { + "epoch": 0.3729957805907173, + "grad_norm": 2.0609381198883057, + "learning_rate": 1.2557230475671256e-05, + "loss": 2.447041893005371, + "step": 137020 + }, + { + "epoch": 0.3730502245814618, + "grad_norm": 2.2507855892181396, + "learning_rate": 1.2556140111762302e-05, + "loss": 2.4844572067260744, + "step": 137040 + }, + { + "epoch": 0.37310466857220637, + "grad_norm": 2.0731239318847656, + "learning_rate": 1.2555049747853348e-05, + "loss": 2.4875749588012694, + "step": 137060 + }, + { + "epoch": 0.37315911256295087, + "grad_norm": 2.0902976989746094, + "learning_rate": 1.2553959383944392e-05, + "loss": 2.5173831939697267, + "step": 137080 + }, + { + "epoch": 0.37321355655369537, + "grad_norm": 2.1774046421051025, + "learning_rate": 1.255286902003544e-05, + "loss": 2.496555519104004, + "step": 137100 + }, + { + "epoch": 0.3732680005444399, + "grad_norm": 2.5025527477264404, + "learning_rate": 1.2551778656126484e-05, + "loss": 2.479597282409668, + "step": 137120 + }, + { + "epoch": 0.37332244453518443, + "grad_norm": 3.488029956817627, + "learning_rate": 1.2550688292217528e-05, + "loss": 2.5359434127807616, + "step": 137140 + }, + { + "epoch": 0.37337688852592893, + "grad_norm": 2.1369428634643555, + "learning_rate": 1.2549597928308575e-05, + "loss": 2.504151153564453, + "step": 137160 + }, + { + "epoch": 0.3734313325166735, + "grad_norm": 2.1963083744049072, + "learning_rate": 1.254850756439962e-05, + "loss": 2.4840007781982423, + "step": 137180 + }, + { + "epoch": 0.373485776507418, + "grad_norm": 2.1613776683807373, + "learning_rate": 1.2547417200490665e-05, + "loss": 2.463800621032715, + "step": 137200 + }, + { + "epoch": 0.3735402204981625, + "grad_norm": 2.078697681427002, + "learning_rate": 1.2546326836581711e-05, + "loss": 2.512717628479004, + "step": 137220 + }, + { + "epoch": 0.37359466448890705, + "grad_norm": 2.251498222351074, + "learning_rate": 1.2545236472672755e-05, + "loss": 2.5358907699584963, + "step": 137240 + }, + { + "epoch": 0.37364910847965155, + "grad_norm": 2.1428871154785156, + "learning_rate": 1.2544146108763802e-05, + "loss": 2.470816230773926, + "step": 137260 + }, + { + "epoch": 0.3737035524703961, + "grad_norm": 2.2126078605651855, + "learning_rate": 1.2543055744854847e-05, + "loss": 2.503183937072754, + "step": 137280 + }, + { + "epoch": 0.3737579964611406, + "grad_norm": 2.1747100353240967, + "learning_rate": 1.254196538094589e-05, + "loss": 2.507106399536133, + "step": 137300 + }, + { + "epoch": 0.3738124404518851, + "grad_norm": 2.311462163925171, + "learning_rate": 1.2540875017036938e-05, + "loss": 2.481354331970215, + "step": 137320 + }, + { + "epoch": 0.37386688444262967, + "grad_norm": 2.222813367843628, + "learning_rate": 1.2539784653127982e-05, + "loss": 2.5038278579711912, + "step": 137340 + }, + { + "epoch": 0.37392132843337417, + "grad_norm": 2.0681533813476562, + "learning_rate": 1.2538694289219028e-05, + "loss": 2.503701400756836, + "step": 137360 + }, + { + "epoch": 0.37397577242411867, + "grad_norm": 2.2573647499084473, + "learning_rate": 1.2537603925310074e-05, + "loss": 2.4884971618652343, + "step": 137380 + }, + { + "epoch": 0.3740302164148632, + "grad_norm": 2.169743776321411, + "learning_rate": 1.2536513561401118e-05, + "loss": 2.476373291015625, + "step": 137400 + }, + { + "epoch": 0.3740846604056077, + "grad_norm": 2.060629367828369, + "learning_rate": 1.2535423197492166e-05, + "loss": 2.471958351135254, + "step": 137420 + }, + { + "epoch": 0.37413910439635223, + "grad_norm": 2.1782054901123047, + "learning_rate": 1.253433283358321e-05, + "loss": 2.499462890625, + "step": 137440 + }, + { + "epoch": 0.3741935483870968, + "grad_norm": 3.9296300411224365, + "learning_rate": 1.2533242469674254e-05, + "loss": 2.4898870468139647, + "step": 137460 + }, + { + "epoch": 0.3742479923778413, + "grad_norm": 2.5814483165740967, + "learning_rate": 1.2532152105765301e-05, + "loss": 2.5060056686401366, + "step": 137480 + }, + { + "epoch": 0.37430243636858584, + "grad_norm": 2.3630993366241455, + "learning_rate": 1.2531061741856345e-05, + "loss": 2.5305755615234373, + "step": 137500 + }, + { + "epoch": 0.37435688035933035, + "grad_norm": 2.1075470447540283, + "learning_rate": 1.2529971377947391e-05, + "loss": 2.48549747467041, + "step": 137520 + }, + { + "epoch": 0.37441132435007485, + "grad_norm": 3.2869372367858887, + "learning_rate": 1.2528881014038437e-05, + "loss": 2.493510055541992, + "step": 137540 + }, + { + "epoch": 0.3744657683408194, + "grad_norm": 2.185455083847046, + "learning_rate": 1.2527790650129481e-05, + "loss": 2.496729278564453, + "step": 137560 + }, + { + "epoch": 0.3745202123315639, + "grad_norm": 2.263584613800049, + "learning_rate": 1.2526700286220527e-05, + "loss": 2.5358856201171873, + "step": 137580 + }, + { + "epoch": 0.3745746563223084, + "grad_norm": 2.1320786476135254, + "learning_rate": 1.2525609922311573e-05, + "loss": 2.493887519836426, + "step": 137600 + }, + { + "epoch": 0.37462910031305297, + "grad_norm": 2.084716320037842, + "learning_rate": 1.2524519558402617e-05, + "loss": 2.4749309539794924, + "step": 137620 + }, + { + "epoch": 0.37468354430379747, + "grad_norm": 2.5568227767944336, + "learning_rate": 1.2523429194493664e-05, + "loss": 2.5200891494750977, + "step": 137640 + }, + { + "epoch": 0.37473798829454197, + "grad_norm": 2.1542487144470215, + "learning_rate": 1.2522338830584709e-05, + "loss": 2.5352027893066404, + "step": 137660 + }, + { + "epoch": 0.3747924322852865, + "grad_norm": 2.255117654800415, + "learning_rate": 1.2521248466675754e-05, + "loss": 2.5159645080566406, + "step": 137680 + }, + { + "epoch": 0.374846876276031, + "grad_norm": 2.1862895488739014, + "learning_rate": 1.25201581027668e-05, + "loss": 2.480930137634277, + "step": 137700 + }, + { + "epoch": 0.37490132026677553, + "grad_norm": 2.543907642364502, + "learning_rate": 1.2519067738857844e-05, + "loss": 2.5042800903320312, + "step": 137720 + }, + { + "epoch": 0.3749557642575201, + "grad_norm": 2.176342010498047, + "learning_rate": 1.251797737494889e-05, + "loss": 2.5081430435180665, + "step": 137740 + }, + { + "epoch": 0.3750102082482646, + "grad_norm": 2.4723408222198486, + "learning_rate": 1.2516887011039936e-05, + "loss": 2.490533447265625, + "step": 137760 + }, + { + "epoch": 0.37506465223900914, + "grad_norm": 2.111248016357422, + "learning_rate": 1.251579664713098e-05, + "loss": 2.4810930252075196, + "step": 137780 + }, + { + "epoch": 0.37511909622975365, + "grad_norm": 2.143303155899048, + "learning_rate": 1.2514706283222027e-05, + "loss": 2.4865964889526366, + "step": 137800 + }, + { + "epoch": 0.37517354022049815, + "grad_norm": 2.2375333309173584, + "learning_rate": 1.2513615919313072e-05, + "loss": 2.4878726959228517, + "step": 137820 + }, + { + "epoch": 0.3752279842112427, + "grad_norm": 2.081977128982544, + "learning_rate": 1.2512525555404117e-05, + "loss": 2.488467788696289, + "step": 137840 + }, + { + "epoch": 0.3752824282019872, + "grad_norm": 2.275705337524414, + "learning_rate": 1.2511435191495163e-05, + "loss": 2.5130386352539062, + "step": 137860 + }, + { + "epoch": 0.3753368721927317, + "grad_norm": 2.118396520614624, + "learning_rate": 1.2510344827586207e-05, + "loss": 2.467844772338867, + "step": 137880 + }, + { + "epoch": 0.37539131618347626, + "grad_norm": 2.315580129623413, + "learning_rate": 1.2509254463677253e-05, + "loss": 2.4924760818481446, + "step": 137900 + }, + { + "epoch": 0.37544576017422077, + "grad_norm": 2.287785768508911, + "learning_rate": 1.2508164099768299e-05, + "loss": 2.5089269638061524, + "step": 137920 + }, + { + "epoch": 0.37550020416496527, + "grad_norm": 2.2458760738372803, + "learning_rate": 1.2507073735859343e-05, + "loss": 2.5095996856689453, + "step": 137940 + }, + { + "epoch": 0.3755546481557098, + "grad_norm": 2.1415772438049316, + "learning_rate": 1.250598337195039e-05, + "loss": 2.4916622161865236, + "step": 137960 + }, + { + "epoch": 0.3756090921464543, + "grad_norm": 2.266895294189453, + "learning_rate": 1.2504893008041435e-05, + "loss": 2.524361991882324, + "step": 137980 + }, + { + "epoch": 0.3756635361371989, + "grad_norm": 2.400080680847168, + "learning_rate": 1.250380264413248e-05, + "loss": 2.4926401138305665, + "step": 138000 + }, + { + "epoch": 0.3757179801279434, + "grad_norm": 2.1431498527526855, + "learning_rate": 1.2502712280223526e-05, + "loss": 2.507124328613281, + "step": 138020 + }, + { + "epoch": 0.3757724241186879, + "grad_norm": 2.3284201622009277, + "learning_rate": 1.250162191631457e-05, + "loss": 2.4868932723999024, + "step": 138040 + }, + { + "epoch": 0.37582686810943244, + "grad_norm": 2.0967578887939453, + "learning_rate": 1.2500531552405616e-05, + "loss": 2.49664192199707, + "step": 138060 + }, + { + "epoch": 0.37588131210017695, + "grad_norm": 2.119920015335083, + "learning_rate": 1.2499441188496662e-05, + "loss": 2.4593406677246095, + "step": 138080 + }, + { + "epoch": 0.37593575609092145, + "grad_norm": 2.2873623371124268, + "learning_rate": 1.2498350824587706e-05, + "loss": 2.4886735916137694, + "step": 138100 + }, + { + "epoch": 0.375990200081666, + "grad_norm": 2.134207010269165, + "learning_rate": 1.2497260460678754e-05, + "loss": 2.515518569946289, + "step": 138120 + }, + { + "epoch": 0.3760446440724105, + "grad_norm": 2.2048850059509277, + "learning_rate": 1.2496170096769798e-05, + "loss": 2.5103660583496095, + "step": 138140 + }, + { + "epoch": 0.376099088063155, + "grad_norm": 2.2154006958007812, + "learning_rate": 1.2495079732860844e-05, + "loss": 2.4761846542358397, + "step": 138160 + }, + { + "epoch": 0.37615353205389956, + "grad_norm": 2.223236560821533, + "learning_rate": 1.249398936895189e-05, + "loss": 2.5066715240478517, + "step": 138180 + }, + { + "epoch": 0.37620797604464407, + "grad_norm": 2.1930043697357178, + "learning_rate": 1.2492899005042934e-05, + "loss": 2.5024505615234376, + "step": 138200 + }, + { + "epoch": 0.37626242003538857, + "grad_norm": 2.2274632453918457, + "learning_rate": 1.249180864113398e-05, + "loss": 2.507821273803711, + "step": 138220 + }, + { + "epoch": 0.3763168640261331, + "grad_norm": 2.110257387161255, + "learning_rate": 1.2490718277225025e-05, + "loss": 2.5088565826416014, + "step": 138240 + }, + { + "epoch": 0.3763713080168776, + "grad_norm": 2.253288507461548, + "learning_rate": 1.248962791331607e-05, + "loss": 2.482803726196289, + "step": 138260 + }, + { + "epoch": 0.3764257520076222, + "grad_norm": 2.230429172515869, + "learning_rate": 1.2488537549407117e-05, + "loss": 2.497767448425293, + "step": 138280 + }, + { + "epoch": 0.3764801959983667, + "grad_norm": 2.1690990924835205, + "learning_rate": 1.2487447185498161e-05, + "loss": 2.506244659423828, + "step": 138300 + }, + { + "epoch": 0.3765346399891112, + "grad_norm": 2.278759479522705, + "learning_rate": 1.2486356821589207e-05, + "loss": 2.4658464431762694, + "step": 138320 + }, + { + "epoch": 0.37658908397985574, + "grad_norm": 2.130643367767334, + "learning_rate": 1.2485266457680252e-05, + "loss": 2.481740379333496, + "step": 138340 + }, + { + "epoch": 0.37664352797060024, + "grad_norm": 2.182792901992798, + "learning_rate": 1.2484176093771297e-05, + "loss": 2.480455780029297, + "step": 138360 + }, + { + "epoch": 0.37669797196134475, + "grad_norm": 2.1521503925323486, + "learning_rate": 1.2483085729862342e-05, + "loss": 2.5090381622314455, + "step": 138380 + }, + { + "epoch": 0.3767524159520893, + "grad_norm": 2.0635552406311035, + "learning_rate": 1.2481995365953388e-05, + "loss": 2.4948659896850587, + "step": 138400 + }, + { + "epoch": 0.3768068599428338, + "grad_norm": 2.3183960914611816, + "learning_rate": 1.2480905002044434e-05, + "loss": 2.484027862548828, + "step": 138420 + }, + { + "epoch": 0.3768613039335783, + "grad_norm": 2.1282129287719727, + "learning_rate": 1.247981463813548e-05, + "loss": 2.5010862350463867, + "step": 138440 + }, + { + "epoch": 0.37691574792432286, + "grad_norm": 3.2539751529693604, + "learning_rate": 1.2478724274226524e-05, + "loss": 2.4883134841918944, + "step": 138460 + }, + { + "epoch": 0.37697019191506737, + "grad_norm": 2.4238321781158447, + "learning_rate": 1.247763391031757e-05, + "loss": 2.498612403869629, + "step": 138480 + }, + { + "epoch": 0.3770246359058119, + "grad_norm": 2.127497434616089, + "learning_rate": 1.2476543546408616e-05, + "loss": 2.493094635009766, + "step": 138500 + }, + { + "epoch": 0.3770790798965564, + "grad_norm": 2.131279706954956, + "learning_rate": 1.247545318249966e-05, + "loss": 2.4888942718505858, + "step": 138520 + }, + { + "epoch": 0.3771335238873009, + "grad_norm": 2.110128402709961, + "learning_rate": 1.2474362818590705e-05, + "loss": 2.4876476287841798, + "step": 138540 + }, + { + "epoch": 0.3771879678780455, + "grad_norm": 2.0619773864746094, + "learning_rate": 1.2473272454681751e-05, + "loss": 2.4674827575683596, + "step": 138560 + }, + { + "epoch": 0.37724241186879, + "grad_norm": 2.152973175048828, + "learning_rate": 1.2472182090772797e-05, + "loss": 2.4897199630737306, + "step": 138580 + }, + { + "epoch": 0.3772968558595345, + "grad_norm": 2.186272621154785, + "learning_rate": 1.2471091726863841e-05, + "loss": 2.5125667572021486, + "step": 138600 + }, + { + "epoch": 0.37735129985027904, + "grad_norm": 2.1210546493530273, + "learning_rate": 1.2470001362954887e-05, + "loss": 2.496372032165527, + "step": 138620 + }, + { + "epoch": 0.37740574384102354, + "grad_norm": 2.3132970333099365, + "learning_rate": 1.2468910999045933e-05, + "loss": 2.5084775924682616, + "step": 138640 + }, + { + "epoch": 0.37746018783176805, + "grad_norm": 2.1782565116882324, + "learning_rate": 1.2467820635136979e-05, + "loss": 2.5085046768188475, + "step": 138660 + }, + { + "epoch": 0.3775146318225126, + "grad_norm": 2.235544443130493, + "learning_rate": 1.2466730271228023e-05, + "loss": 2.4895248413085938, + "step": 138680 + }, + { + "epoch": 0.3775690758132571, + "grad_norm": 2.173469305038452, + "learning_rate": 1.2465639907319069e-05, + "loss": 2.500691795349121, + "step": 138700 + }, + { + "epoch": 0.3776235198040016, + "grad_norm": 2.1880056858062744, + "learning_rate": 1.2464549543410114e-05, + "loss": 2.492647171020508, + "step": 138720 + }, + { + "epoch": 0.37767796379474616, + "grad_norm": 2.143094778060913, + "learning_rate": 1.246345917950116e-05, + "loss": 2.4864879608154298, + "step": 138740 + }, + { + "epoch": 0.37773240778549066, + "grad_norm": 2.182417631149292, + "learning_rate": 1.2462368815592204e-05, + "loss": 2.498994827270508, + "step": 138760 + }, + { + "epoch": 0.3777868517762352, + "grad_norm": 2.1985254287719727, + "learning_rate": 1.246127845168325e-05, + "loss": 2.4854267120361326, + "step": 138780 + }, + { + "epoch": 0.3778412957669797, + "grad_norm": 1.9866889715194702, + "learning_rate": 1.2460188087774296e-05, + "loss": 2.508163642883301, + "step": 138800 + }, + { + "epoch": 0.3778957397577242, + "grad_norm": 2.172072172164917, + "learning_rate": 1.2459097723865342e-05, + "loss": 2.4708019256591798, + "step": 138820 + }, + { + "epoch": 0.3779501837484688, + "grad_norm": 2.231166362762451, + "learning_rate": 1.2458007359956386e-05, + "loss": 2.4940200805664063, + "step": 138840 + }, + { + "epoch": 0.3780046277392133, + "grad_norm": 2.267235279083252, + "learning_rate": 1.2456916996047432e-05, + "loss": 2.503931427001953, + "step": 138860 + }, + { + "epoch": 0.3780590717299578, + "grad_norm": 2.194577693939209, + "learning_rate": 1.2455826632138477e-05, + "loss": 2.4925020217895506, + "step": 138880 + }, + { + "epoch": 0.37811351572070234, + "grad_norm": 2.24126935005188, + "learning_rate": 1.2454736268229523e-05, + "loss": 2.503011703491211, + "step": 138900 + }, + { + "epoch": 0.37816795971144684, + "grad_norm": 2.156345844268799, + "learning_rate": 1.2453645904320567e-05, + "loss": 2.4704771041870117, + "step": 138920 + }, + { + "epoch": 0.37822240370219135, + "grad_norm": 2.2379825115203857, + "learning_rate": 1.2452555540411613e-05, + "loss": 2.482827377319336, + "step": 138940 + }, + { + "epoch": 0.3782768476929359, + "grad_norm": 2.703679084777832, + "learning_rate": 1.2451465176502659e-05, + "loss": 2.5048669815063476, + "step": 138960 + }, + { + "epoch": 0.3783312916836804, + "grad_norm": 2.083709955215454, + "learning_rate": 1.2450374812593705e-05, + "loss": 2.4922931671142576, + "step": 138980 + }, + { + "epoch": 0.37838573567442496, + "grad_norm": 2.3696694374084473, + "learning_rate": 1.2449284448684749e-05, + "loss": 2.491895866394043, + "step": 139000 + }, + { + "epoch": 0.37844017966516946, + "grad_norm": 2.163235902786255, + "learning_rate": 1.2448194084775795e-05, + "loss": 2.482617950439453, + "step": 139020 + }, + { + "epoch": 0.37849462365591396, + "grad_norm": 2.3660647869110107, + "learning_rate": 1.244710372086684e-05, + "loss": 2.480949592590332, + "step": 139040 + }, + { + "epoch": 0.3785490676466585, + "grad_norm": 2.658414363861084, + "learning_rate": 1.2446013356957886e-05, + "loss": 2.4765798568725588, + "step": 139060 + }, + { + "epoch": 0.378603511637403, + "grad_norm": 2.1168830394744873, + "learning_rate": 1.244492299304893e-05, + "loss": 2.487593078613281, + "step": 139080 + }, + { + "epoch": 0.3786579556281475, + "grad_norm": 2.0717430114746094, + "learning_rate": 1.2443832629139976e-05, + "loss": 2.460537338256836, + "step": 139100 + }, + { + "epoch": 0.3787123996188921, + "grad_norm": 2.8297958374023438, + "learning_rate": 1.2442742265231022e-05, + "loss": 2.4737617492675783, + "step": 139120 + }, + { + "epoch": 0.3787668436096366, + "grad_norm": 2.2445595264434814, + "learning_rate": 1.2441651901322068e-05, + "loss": 2.514369583129883, + "step": 139140 + }, + { + "epoch": 0.3788212876003811, + "grad_norm": 2.6247897148132324, + "learning_rate": 1.2440561537413112e-05, + "loss": 2.4583192825317384, + "step": 139160 + }, + { + "epoch": 0.37887573159112564, + "grad_norm": 2.0730032920837402, + "learning_rate": 1.2439471173504158e-05, + "loss": 2.4990562438964843, + "step": 139180 + }, + { + "epoch": 0.37893017558187014, + "grad_norm": 2.1597795486450195, + "learning_rate": 1.2438380809595204e-05, + "loss": 2.457056427001953, + "step": 139200 + }, + { + "epoch": 0.37898461957261464, + "grad_norm": 2.1472930908203125, + "learning_rate": 1.243729044568625e-05, + "loss": 2.512388229370117, + "step": 139220 + }, + { + "epoch": 0.3790390635633592, + "grad_norm": 2.0589957237243652, + "learning_rate": 1.2436200081777294e-05, + "loss": 2.478033447265625, + "step": 139240 + }, + { + "epoch": 0.3790935075541037, + "grad_norm": 2.1975507736206055, + "learning_rate": 1.243510971786834e-05, + "loss": 2.488907814025879, + "step": 139260 + }, + { + "epoch": 0.37914795154484826, + "grad_norm": 2.406536817550659, + "learning_rate": 1.2434019353959385e-05, + "loss": 2.4885274887084963, + "step": 139280 + }, + { + "epoch": 0.37920239553559276, + "grad_norm": 2.26485538482666, + "learning_rate": 1.2432928990050431e-05, + "loss": 2.5059471130371094, + "step": 139300 + }, + { + "epoch": 0.37925683952633726, + "grad_norm": 2.3585360050201416, + "learning_rate": 1.2431838626141475e-05, + "loss": 2.50960750579834, + "step": 139320 + }, + { + "epoch": 0.3793112835170818, + "grad_norm": 2.1879653930664062, + "learning_rate": 1.2430748262232521e-05, + "loss": 2.4745702743530273, + "step": 139340 + }, + { + "epoch": 0.3793657275078263, + "grad_norm": 2.0865318775177, + "learning_rate": 1.2429657898323567e-05, + "loss": 2.509158134460449, + "step": 139360 + }, + { + "epoch": 0.3794201714985708, + "grad_norm": 2.1180315017700195, + "learning_rate": 1.2428567534414613e-05, + "loss": 2.500698471069336, + "step": 139380 + }, + { + "epoch": 0.3794746154893154, + "grad_norm": 2.3502559661865234, + "learning_rate": 1.2427477170505657e-05, + "loss": 2.522044372558594, + "step": 139400 + }, + { + "epoch": 0.3795290594800599, + "grad_norm": 2.1521236896514893, + "learning_rate": 1.2426386806596702e-05, + "loss": 2.4818231582641603, + "step": 139420 + }, + { + "epoch": 0.3795835034708044, + "grad_norm": 2.0975069999694824, + "learning_rate": 1.2425296442687748e-05, + "loss": 2.4790508270263674, + "step": 139440 + }, + { + "epoch": 0.37963794746154894, + "grad_norm": 2.1280200481414795, + "learning_rate": 1.2424206078778794e-05, + "loss": 2.4858646392822266, + "step": 139460 + }, + { + "epoch": 0.37969239145229344, + "grad_norm": 2.183476209640503, + "learning_rate": 1.2423115714869838e-05, + "loss": 2.4688447952270507, + "step": 139480 + }, + { + "epoch": 0.379746835443038, + "grad_norm": 2.2734720706939697, + "learning_rate": 1.2422025350960884e-05, + "loss": 2.505522537231445, + "step": 139500 + }, + { + "epoch": 0.3798012794337825, + "grad_norm": 2.1644372940063477, + "learning_rate": 1.242093498705193e-05, + "loss": 2.4559709548950197, + "step": 139520 + }, + { + "epoch": 0.379855723424527, + "grad_norm": 2.097238540649414, + "learning_rate": 1.2419844623142976e-05, + "loss": 2.4818326950073244, + "step": 139540 + }, + { + "epoch": 0.37991016741527156, + "grad_norm": 2.2278196811676025, + "learning_rate": 1.241875425923402e-05, + "loss": 2.469603157043457, + "step": 139560 + }, + { + "epoch": 0.37996461140601606, + "grad_norm": 2.4452264308929443, + "learning_rate": 1.2417663895325066e-05, + "loss": 2.480426025390625, + "step": 139580 + }, + { + "epoch": 0.38001905539676056, + "grad_norm": 2.114993095397949, + "learning_rate": 1.2416573531416111e-05, + "loss": 2.4861064910888673, + "step": 139600 + }, + { + "epoch": 0.3800734993875051, + "grad_norm": 2.132652759552002, + "learning_rate": 1.2415483167507157e-05, + "loss": 2.480510711669922, + "step": 139620 + }, + { + "epoch": 0.3801279433782496, + "grad_norm": 2.602903127670288, + "learning_rate": 1.2414392803598201e-05, + "loss": 2.462441825866699, + "step": 139640 + }, + { + "epoch": 0.3801823873689941, + "grad_norm": 2.092829465866089, + "learning_rate": 1.2413302439689247e-05, + "loss": 2.477809715270996, + "step": 139660 + }, + { + "epoch": 0.3802368313597387, + "grad_norm": 2.126237392425537, + "learning_rate": 1.2412212075780293e-05, + "loss": 2.499650573730469, + "step": 139680 + }, + { + "epoch": 0.3802912753504832, + "grad_norm": 2.2062623500823975, + "learning_rate": 1.2411121711871339e-05, + "loss": 2.50463924407959, + "step": 139700 + }, + { + "epoch": 0.38034571934122774, + "grad_norm": 2.1791985034942627, + "learning_rate": 1.2410031347962383e-05, + "loss": 2.473119354248047, + "step": 139720 + }, + { + "epoch": 0.38040016333197224, + "grad_norm": 2.2005093097686768, + "learning_rate": 1.2408940984053429e-05, + "loss": 2.498669242858887, + "step": 139740 + }, + { + "epoch": 0.38045460732271674, + "grad_norm": 2.265960693359375, + "learning_rate": 1.2407850620144474e-05, + "loss": 2.499639320373535, + "step": 139760 + }, + { + "epoch": 0.3805090513134613, + "grad_norm": 2.1255502700805664, + "learning_rate": 1.2406760256235519e-05, + "loss": 2.492232322692871, + "step": 139780 + }, + { + "epoch": 0.3805634953042058, + "grad_norm": 2.1796560287475586, + "learning_rate": 1.2405669892326564e-05, + "loss": 2.4950481414794923, + "step": 139800 + }, + { + "epoch": 0.3806179392949503, + "grad_norm": 2.098005533218384, + "learning_rate": 1.240457952841761e-05, + "loss": 2.4830970764160156, + "step": 139820 + }, + { + "epoch": 0.38067238328569486, + "grad_norm": 2.0830657482147217, + "learning_rate": 1.2403489164508656e-05, + "loss": 2.4858449935913085, + "step": 139840 + }, + { + "epoch": 0.38072682727643936, + "grad_norm": 2.2652060985565186, + "learning_rate": 1.2402398800599702e-05, + "loss": 2.5297929763793947, + "step": 139860 + }, + { + "epoch": 0.38078127126718386, + "grad_norm": 2.060765266418457, + "learning_rate": 1.2401308436690746e-05, + "loss": 2.4768787384033204, + "step": 139880 + }, + { + "epoch": 0.3808357152579284, + "grad_norm": 2.1390256881713867, + "learning_rate": 1.2400218072781792e-05, + "loss": 2.4992385864257813, + "step": 139900 + }, + { + "epoch": 0.3808901592486729, + "grad_norm": 2.101658821105957, + "learning_rate": 1.2399127708872838e-05, + "loss": 2.548704719543457, + "step": 139920 + }, + { + "epoch": 0.3809446032394174, + "grad_norm": 2.1439645290374756, + "learning_rate": 1.2398037344963882e-05, + "loss": 2.515268898010254, + "step": 139940 + }, + { + "epoch": 0.380999047230162, + "grad_norm": 2.0888845920562744, + "learning_rate": 1.2396946981054927e-05, + "loss": 2.4958066940307617, + "step": 139960 + }, + { + "epoch": 0.3810534912209065, + "grad_norm": 2.1655945777893066, + "learning_rate": 1.2395856617145973e-05, + "loss": 2.473809814453125, + "step": 139980 + }, + { + "epoch": 0.38110793521165104, + "grad_norm": 2.0614497661590576, + "learning_rate": 1.2394766253237019e-05, + "loss": 2.498116302490234, + "step": 140000 + }, + { + "epoch": 0.38116237920239554, + "grad_norm": 2.168736219406128, + "learning_rate": 1.2393675889328065e-05, + "loss": 2.520002746582031, + "step": 140020 + }, + { + "epoch": 0.38121682319314004, + "grad_norm": 2.157097101211548, + "learning_rate": 1.2392585525419109e-05, + "loss": 2.4766748428344725, + "step": 140040 + }, + { + "epoch": 0.3812712671838846, + "grad_norm": 2.364471197128296, + "learning_rate": 1.2391495161510155e-05, + "loss": 2.499907875061035, + "step": 140060 + }, + { + "epoch": 0.3813257111746291, + "grad_norm": 2.045931577682495, + "learning_rate": 1.23904047976012e-05, + "loss": 2.485547637939453, + "step": 140080 + }, + { + "epoch": 0.3813801551653736, + "grad_norm": 2.151900291442871, + "learning_rate": 1.2389314433692245e-05, + "loss": 2.476857376098633, + "step": 140100 + }, + { + "epoch": 0.38143459915611816, + "grad_norm": 2.1744837760925293, + "learning_rate": 1.2388224069783292e-05, + "loss": 2.4933067321777345, + "step": 140120 + }, + { + "epoch": 0.38148904314686266, + "grad_norm": 2.173400640487671, + "learning_rate": 1.2387133705874336e-05, + "loss": 2.539068412780762, + "step": 140140 + }, + { + "epoch": 0.38154348713760716, + "grad_norm": 2.2126548290252686, + "learning_rate": 1.2386043341965382e-05, + "loss": 2.516169548034668, + "step": 140160 + }, + { + "epoch": 0.3815979311283517, + "grad_norm": 2.5245814323425293, + "learning_rate": 1.2384952978056428e-05, + "loss": 2.4821407318115236, + "step": 140180 + }, + { + "epoch": 0.3816523751190962, + "grad_norm": 2.2082645893096924, + "learning_rate": 1.2383862614147472e-05, + "loss": 2.5099782943725586, + "step": 140200 + }, + { + "epoch": 0.3817068191098408, + "grad_norm": 2.3000855445861816, + "learning_rate": 1.2382772250238518e-05, + "loss": 2.4440526962280273, + "step": 140220 + }, + { + "epoch": 0.3817612631005853, + "grad_norm": 2.363441228866577, + "learning_rate": 1.2381681886329564e-05, + "loss": 2.453042221069336, + "step": 140240 + }, + { + "epoch": 0.3818157070913298, + "grad_norm": 2.969774007797241, + "learning_rate": 1.2380591522420608e-05, + "loss": 2.5119340896606444, + "step": 140260 + }, + { + "epoch": 0.38187015108207434, + "grad_norm": 2.1264359951019287, + "learning_rate": 1.2379501158511655e-05, + "loss": 2.4959362030029295, + "step": 140280 + }, + { + "epoch": 0.38192459507281884, + "grad_norm": 2.08573842048645, + "learning_rate": 1.23784107946027e-05, + "loss": 2.514854621887207, + "step": 140300 + }, + { + "epoch": 0.38197903906356334, + "grad_norm": 2.1220052242279053, + "learning_rate": 1.2377320430693745e-05, + "loss": 2.513307571411133, + "step": 140320 + }, + { + "epoch": 0.3820334830543079, + "grad_norm": 2.115847110748291, + "learning_rate": 1.2376230066784791e-05, + "loss": 2.4744598388671877, + "step": 140340 + }, + { + "epoch": 0.3820879270450524, + "grad_norm": 2.0808732509613037, + "learning_rate": 1.2375139702875835e-05, + "loss": 2.5046796798706055, + "step": 140360 + }, + { + "epoch": 0.3821423710357969, + "grad_norm": 2.306244134902954, + "learning_rate": 1.2374049338966881e-05, + "loss": 2.511289596557617, + "step": 140380 + }, + { + "epoch": 0.38219681502654146, + "grad_norm": 2.0692639350891113, + "learning_rate": 1.2372958975057927e-05, + "loss": 2.491343116760254, + "step": 140400 + }, + { + "epoch": 0.38225125901728596, + "grad_norm": 2.214164972305298, + "learning_rate": 1.2371868611148971e-05, + "loss": 2.489803695678711, + "step": 140420 + }, + { + "epoch": 0.38230570300803046, + "grad_norm": 2.2209558486938477, + "learning_rate": 1.2370778247240018e-05, + "loss": 2.50262393951416, + "step": 140440 + }, + { + "epoch": 0.382360146998775, + "grad_norm": 2.1812074184417725, + "learning_rate": 1.2369687883331063e-05, + "loss": 2.495022201538086, + "step": 140460 + }, + { + "epoch": 0.3824145909895195, + "grad_norm": 2.1410562992095947, + "learning_rate": 1.2368597519422108e-05, + "loss": 2.4960960388183593, + "step": 140480 + }, + { + "epoch": 0.3824690349802641, + "grad_norm": 2.1550133228302, + "learning_rate": 1.2367507155513154e-05, + "loss": 2.4866743087768555, + "step": 140500 + }, + { + "epoch": 0.3825234789710086, + "grad_norm": 2.4158270359039307, + "learning_rate": 1.2366416791604198e-05, + "loss": 2.465468978881836, + "step": 140520 + }, + { + "epoch": 0.3825779229617531, + "grad_norm": 2.065239667892456, + "learning_rate": 1.2365326427695244e-05, + "loss": 2.502547836303711, + "step": 140540 + }, + { + "epoch": 0.38263236695249764, + "grad_norm": 2.272944927215576, + "learning_rate": 1.236423606378629e-05, + "loss": 2.506198501586914, + "step": 140560 + }, + { + "epoch": 0.38268681094324214, + "grad_norm": 2.262517213821411, + "learning_rate": 1.2363145699877334e-05, + "loss": 2.4974992752075194, + "step": 140580 + }, + { + "epoch": 0.38274125493398664, + "grad_norm": 2.237475633621216, + "learning_rate": 1.2362055335968381e-05, + "loss": 2.4854917526245117, + "step": 140600 + }, + { + "epoch": 0.3827956989247312, + "grad_norm": 2.247840166091919, + "learning_rate": 1.2360964972059426e-05, + "loss": 2.4923465728759764, + "step": 140620 + }, + { + "epoch": 0.3828501429154757, + "grad_norm": 2.3762619495391846, + "learning_rate": 1.2359874608150471e-05, + "loss": 2.47678165435791, + "step": 140640 + }, + { + "epoch": 0.3829045869062202, + "grad_norm": 2.1376521587371826, + "learning_rate": 1.2358784244241517e-05, + "loss": 2.519077491760254, + "step": 140660 + }, + { + "epoch": 0.38295903089696476, + "grad_norm": 2.1255221366882324, + "learning_rate": 1.2357693880332561e-05, + "loss": 2.4826244354248046, + "step": 140680 + }, + { + "epoch": 0.38301347488770926, + "grad_norm": 2.084845781326294, + "learning_rate": 1.2356603516423607e-05, + "loss": 2.4758867263793944, + "step": 140700 + }, + { + "epoch": 0.3830679188784538, + "grad_norm": 2.110989809036255, + "learning_rate": 1.2355513152514653e-05, + "loss": 2.5150394439697266, + "step": 140720 + }, + { + "epoch": 0.3831223628691983, + "grad_norm": 2.085131883621216, + "learning_rate": 1.2354422788605697e-05, + "loss": 2.4798553466796873, + "step": 140740 + }, + { + "epoch": 0.3831768068599428, + "grad_norm": 2.2924764156341553, + "learning_rate": 1.2353332424696745e-05, + "loss": 2.499064064025879, + "step": 140760 + }, + { + "epoch": 0.3832312508506874, + "grad_norm": 2.1854684352874756, + "learning_rate": 1.2352242060787789e-05, + "loss": 2.4846019744873047, + "step": 140780 + }, + { + "epoch": 0.3832856948414319, + "grad_norm": 2.2124829292297363, + "learning_rate": 1.2351151696878833e-05, + "loss": 2.4863727569580076, + "step": 140800 + }, + { + "epoch": 0.3833401388321764, + "grad_norm": 2.228330135345459, + "learning_rate": 1.235006133296988e-05, + "loss": 2.4812541961669923, + "step": 140820 + }, + { + "epoch": 0.38339458282292094, + "grad_norm": 2.2257304191589355, + "learning_rate": 1.2348970969060924e-05, + "loss": 2.489228439331055, + "step": 140840 + }, + { + "epoch": 0.38344902681366544, + "grad_norm": 2.1600780487060547, + "learning_rate": 1.234788060515197e-05, + "loss": 2.5153427124023438, + "step": 140860 + }, + { + "epoch": 0.38350347080440994, + "grad_norm": 2.2194480895996094, + "learning_rate": 1.2346790241243016e-05, + "loss": 2.4983726501464845, + "step": 140880 + }, + { + "epoch": 0.3835579147951545, + "grad_norm": 2.1034088134765625, + "learning_rate": 1.234569987733406e-05, + "loss": 2.4980249404907227, + "step": 140900 + }, + { + "epoch": 0.383612358785899, + "grad_norm": 2.0878942012786865, + "learning_rate": 1.2344609513425108e-05, + "loss": 2.516330528259277, + "step": 140920 + }, + { + "epoch": 0.3836668027766435, + "grad_norm": 2.1971583366394043, + "learning_rate": 1.2343519149516152e-05, + "loss": 2.495846176147461, + "step": 140940 + }, + { + "epoch": 0.38372124676738806, + "grad_norm": 2.215334177017212, + "learning_rate": 1.2342428785607196e-05, + "loss": 2.493487548828125, + "step": 140960 + }, + { + "epoch": 0.38377569075813256, + "grad_norm": 2.1223716735839844, + "learning_rate": 1.2341338421698243e-05, + "loss": 2.4829229354858398, + "step": 140980 + }, + { + "epoch": 0.3838301347488771, + "grad_norm": 2.0959935188293457, + "learning_rate": 1.2340248057789288e-05, + "loss": 2.482141876220703, + "step": 141000 + }, + { + "epoch": 0.3838845787396216, + "grad_norm": 2.1395740509033203, + "learning_rate": 1.2339157693880333e-05, + "loss": 2.507109260559082, + "step": 141020 + }, + { + "epoch": 0.3839390227303661, + "grad_norm": 2.2184195518493652, + "learning_rate": 1.2338067329971379e-05, + "loss": 2.4970523834228517, + "step": 141040 + }, + { + "epoch": 0.3839934667211107, + "grad_norm": 2.1938369274139404, + "learning_rate": 1.2336976966062423e-05, + "loss": 2.48903751373291, + "step": 141060 + }, + { + "epoch": 0.3840479107118552, + "grad_norm": 2.1319644451141357, + "learning_rate": 1.233588660215347e-05, + "loss": 2.521879959106445, + "step": 141080 + }, + { + "epoch": 0.3841023547025997, + "grad_norm": 2.160985231399536, + "learning_rate": 1.2334796238244515e-05, + "loss": 2.509016990661621, + "step": 141100 + }, + { + "epoch": 0.38415679869334424, + "grad_norm": 2.566493272781372, + "learning_rate": 1.2333705874335559e-05, + "loss": 2.4944040298461916, + "step": 141120 + }, + { + "epoch": 0.38421124268408874, + "grad_norm": 2.195507526397705, + "learning_rate": 1.2332615510426606e-05, + "loss": 2.492296600341797, + "step": 141140 + }, + { + "epoch": 0.38426568667483324, + "grad_norm": 2.14656400680542, + "learning_rate": 1.233152514651765e-05, + "loss": 2.48046760559082, + "step": 141160 + }, + { + "epoch": 0.3843201306655778, + "grad_norm": 2.137570858001709, + "learning_rate": 1.2330434782608696e-05, + "loss": 2.534876823425293, + "step": 141180 + }, + { + "epoch": 0.3843745746563223, + "grad_norm": 2.33520245552063, + "learning_rate": 1.2329344418699742e-05, + "loss": 2.523433876037598, + "step": 141200 + }, + { + "epoch": 0.38442901864706686, + "grad_norm": 2.1861729621887207, + "learning_rate": 1.2328254054790786e-05, + "loss": 2.5037567138671877, + "step": 141220 + }, + { + "epoch": 0.38448346263781136, + "grad_norm": 2.146425485610962, + "learning_rate": 1.2327163690881834e-05, + "loss": 2.4868600845336912, + "step": 141240 + }, + { + "epoch": 0.38453790662855586, + "grad_norm": 2.4797682762145996, + "learning_rate": 1.2326073326972878e-05, + "loss": 2.4809261322021485, + "step": 141260 + }, + { + "epoch": 0.3845923506193004, + "grad_norm": 2.051896572113037, + "learning_rate": 1.2324982963063922e-05, + "loss": 2.4826107025146484, + "step": 141280 + }, + { + "epoch": 0.3846467946100449, + "grad_norm": 2.24841570854187, + "learning_rate": 1.232389259915497e-05, + "loss": 2.446695899963379, + "step": 141300 + }, + { + "epoch": 0.3847012386007894, + "grad_norm": 2.1600565910339355, + "learning_rate": 1.2322802235246014e-05, + "loss": 2.503672218322754, + "step": 141320 + }, + { + "epoch": 0.384755682591534, + "grad_norm": 2.3618133068084717, + "learning_rate": 1.232171187133706e-05, + "loss": 2.4871082305908203, + "step": 141340 + }, + { + "epoch": 0.3848101265822785, + "grad_norm": 2.308011054992676, + "learning_rate": 1.2320621507428105e-05, + "loss": 2.4857648849487304, + "step": 141360 + }, + { + "epoch": 0.384864570573023, + "grad_norm": 2.2234673500061035, + "learning_rate": 1.231953114351915e-05, + "loss": 2.487593650817871, + "step": 141380 + }, + { + "epoch": 0.38491901456376754, + "grad_norm": 2.0952911376953125, + "learning_rate": 1.2318440779610197e-05, + "loss": 2.469906234741211, + "step": 141400 + }, + { + "epoch": 0.38497345855451204, + "grad_norm": 2.4589686393737793, + "learning_rate": 1.2317350415701241e-05, + "loss": 2.474497413635254, + "step": 141420 + }, + { + "epoch": 0.38502790254525654, + "grad_norm": 2.0675454139709473, + "learning_rate": 1.2316260051792285e-05, + "loss": 2.5127870559692385, + "step": 141440 + }, + { + "epoch": 0.3850823465360011, + "grad_norm": 2.3390257358551025, + "learning_rate": 1.2315169687883333e-05, + "loss": 2.5185102462768554, + "step": 141460 + }, + { + "epoch": 0.3851367905267456, + "grad_norm": 2.1785409450531006, + "learning_rate": 1.2314079323974377e-05, + "loss": 2.5005346298217774, + "step": 141480 + }, + { + "epoch": 0.38519123451749016, + "grad_norm": 2.1410770416259766, + "learning_rate": 1.2312988960065423e-05, + "loss": 2.4892282485961914, + "step": 141500 + }, + { + "epoch": 0.38524567850823466, + "grad_norm": 2.677346706390381, + "learning_rate": 1.2311898596156468e-05, + "loss": 2.506515312194824, + "step": 141520 + }, + { + "epoch": 0.38530012249897916, + "grad_norm": 2.4995815753936768, + "learning_rate": 1.2310808232247512e-05, + "loss": 2.50331974029541, + "step": 141540 + }, + { + "epoch": 0.3853545664897237, + "grad_norm": 2.289538621902466, + "learning_rate": 1.230971786833856e-05, + "loss": 2.452079010009766, + "step": 141560 + }, + { + "epoch": 0.3854090104804682, + "grad_norm": 2.1448657512664795, + "learning_rate": 1.2308627504429604e-05, + "loss": 2.5097539901733397, + "step": 141580 + }, + { + "epoch": 0.3854634544712127, + "grad_norm": 2.0409750938415527, + "learning_rate": 1.2307537140520648e-05, + "loss": 2.5312013626098633, + "step": 141600 + }, + { + "epoch": 0.3855178984619573, + "grad_norm": 2.1129465103149414, + "learning_rate": 1.2306446776611696e-05, + "loss": 2.485280990600586, + "step": 141620 + }, + { + "epoch": 0.3855723424527018, + "grad_norm": 2.1113760471343994, + "learning_rate": 1.230535641270274e-05, + "loss": 2.470819091796875, + "step": 141640 + }, + { + "epoch": 0.3856267864434463, + "grad_norm": 2.2001938819885254, + "learning_rate": 1.2304266048793786e-05, + "loss": 2.518092155456543, + "step": 141660 + }, + { + "epoch": 0.38568123043419084, + "grad_norm": 2.0669798851013184, + "learning_rate": 1.2303175684884831e-05, + "loss": 2.4940561294555663, + "step": 141680 + }, + { + "epoch": 0.38573567442493534, + "grad_norm": 2.1302428245544434, + "learning_rate": 1.2302085320975876e-05, + "loss": 2.5015613555908205, + "step": 141700 + }, + { + "epoch": 0.3857901184156799, + "grad_norm": 2.2279233932495117, + "learning_rate": 1.2300994957066923e-05, + "loss": 2.5116832733154295, + "step": 141720 + }, + { + "epoch": 0.3858445624064244, + "grad_norm": 2.094848871231079, + "learning_rate": 1.2299904593157967e-05, + "loss": 2.4903263092041015, + "step": 141740 + }, + { + "epoch": 0.3858990063971689, + "grad_norm": 2.1704866886138916, + "learning_rate": 1.2298814229249011e-05, + "loss": 2.5229095458984374, + "step": 141760 + }, + { + "epoch": 0.38595345038791345, + "grad_norm": 2.138312816619873, + "learning_rate": 1.2297723865340059e-05, + "loss": 2.485068511962891, + "step": 141780 + }, + { + "epoch": 0.38600789437865796, + "grad_norm": 2.1095848083496094, + "learning_rate": 1.2296633501431103e-05, + "loss": 2.512617301940918, + "step": 141800 + }, + { + "epoch": 0.38606233836940246, + "grad_norm": 2.244271755218506, + "learning_rate": 1.229554313752215e-05, + "loss": 2.468514633178711, + "step": 141820 + }, + { + "epoch": 0.386116782360147, + "grad_norm": 2.1298670768737793, + "learning_rate": 1.2294452773613195e-05, + "loss": 2.4886343002319338, + "step": 141840 + }, + { + "epoch": 0.3861712263508915, + "grad_norm": 2.124762773513794, + "learning_rate": 1.2293362409704239e-05, + "loss": 2.4815128326416014, + "step": 141860 + }, + { + "epoch": 0.386225670341636, + "grad_norm": 2.3281450271606445, + "learning_rate": 1.2292272045795286e-05, + "loss": 2.501374435424805, + "step": 141880 + }, + { + "epoch": 0.3862801143323806, + "grad_norm": 2.117144823074341, + "learning_rate": 1.229118168188633e-05, + "loss": 2.531477355957031, + "step": 141900 + }, + { + "epoch": 0.3863345583231251, + "grad_norm": 2.199712038040161, + "learning_rate": 1.2290091317977374e-05, + "loss": 2.489467811584473, + "step": 141920 + }, + { + "epoch": 0.38638900231386963, + "grad_norm": 2.2113566398620605, + "learning_rate": 1.2289000954068422e-05, + "loss": 2.503632736206055, + "step": 141940 + }, + { + "epoch": 0.38644344630461414, + "grad_norm": 2.165642261505127, + "learning_rate": 1.2287910590159466e-05, + "loss": 2.497791290283203, + "step": 141960 + }, + { + "epoch": 0.38649789029535864, + "grad_norm": 2.129840135574341, + "learning_rate": 1.2286820226250514e-05, + "loss": 2.4855113983154298, + "step": 141980 + }, + { + "epoch": 0.3865523342861032, + "grad_norm": 2.103104591369629, + "learning_rate": 1.2285729862341558e-05, + "loss": 2.5218700408935546, + "step": 142000 + }, + { + "epoch": 0.3866067782768477, + "grad_norm": 2.1075708866119385, + "learning_rate": 1.2284639498432602e-05, + "loss": 2.472026824951172, + "step": 142020 + }, + { + "epoch": 0.3866612222675922, + "grad_norm": 2.1267411708831787, + "learning_rate": 1.228354913452365e-05, + "loss": 2.4832008361816404, + "step": 142040 + }, + { + "epoch": 0.38671566625833675, + "grad_norm": 2.197953462600708, + "learning_rate": 1.2282458770614693e-05, + "loss": 2.519835090637207, + "step": 142060 + }, + { + "epoch": 0.38677011024908126, + "grad_norm": 2.1562893390655518, + "learning_rate": 1.2281368406705737e-05, + "loss": 2.4832252502441405, + "step": 142080 + }, + { + "epoch": 0.38682455423982576, + "grad_norm": 2.195223569869995, + "learning_rate": 1.2280278042796785e-05, + "loss": 2.479889678955078, + "step": 142100 + }, + { + "epoch": 0.3868789982305703, + "grad_norm": 2.1271135807037354, + "learning_rate": 1.2279187678887829e-05, + "loss": 2.4961305618286134, + "step": 142120 + }, + { + "epoch": 0.3869334422213148, + "grad_norm": 2.1933517456054688, + "learning_rate": 1.2278097314978877e-05, + "loss": 2.520233726501465, + "step": 142140 + }, + { + "epoch": 0.3869878862120593, + "grad_norm": 2.1794018745422363, + "learning_rate": 1.227700695106992e-05, + "loss": 2.492152976989746, + "step": 142160 + }, + { + "epoch": 0.3870423302028039, + "grad_norm": 2.1409754753112793, + "learning_rate": 1.2275916587160965e-05, + "loss": 2.4881940841674806, + "step": 142180 + }, + { + "epoch": 0.3870967741935484, + "grad_norm": 2.0921876430511475, + "learning_rate": 1.2274826223252012e-05, + "loss": 2.46893253326416, + "step": 142200 + }, + { + "epoch": 0.38715121818429293, + "grad_norm": 2.15923810005188, + "learning_rate": 1.2273735859343056e-05, + "loss": 2.487030792236328, + "step": 142220 + }, + { + "epoch": 0.38720566217503744, + "grad_norm": 2.211732864379883, + "learning_rate": 1.22726454954341e-05, + "loss": 2.490289306640625, + "step": 142240 + }, + { + "epoch": 0.38726010616578194, + "grad_norm": 2.0587565898895264, + "learning_rate": 1.2271555131525148e-05, + "loss": 2.5475770950317385, + "step": 142260 + }, + { + "epoch": 0.3873145501565265, + "grad_norm": 2.0692636966705322, + "learning_rate": 1.2270464767616192e-05, + "loss": 2.485312271118164, + "step": 142280 + }, + { + "epoch": 0.387368994147271, + "grad_norm": 2.1923649311065674, + "learning_rate": 1.226937440370724e-05, + "loss": 2.4640268325805663, + "step": 142300 + }, + { + "epoch": 0.3874234381380155, + "grad_norm": 4.514732837677002, + "learning_rate": 1.2268284039798284e-05, + "loss": 2.468617248535156, + "step": 142320 + }, + { + "epoch": 0.38747788212876005, + "grad_norm": 2.2211172580718994, + "learning_rate": 1.2267193675889328e-05, + "loss": 2.4919717788696287, + "step": 142340 + }, + { + "epoch": 0.38753232611950456, + "grad_norm": 5.008827209472656, + "learning_rate": 1.2266103311980375e-05, + "loss": 2.4889247894287108, + "step": 142360 + }, + { + "epoch": 0.38758677011024906, + "grad_norm": 2.215826988220215, + "learning_rate": 1.226501294807142e-05, + "loss": 2.5077762603759766, + "step": 142380 + }, + { + "epoch": 0.3876412141009936, + "grad_norm": 3.671781063079834, + "learning_rate": 1.2263922584162464e-05, + "loss": 2.4744169235229494, + "step": 142400 + }, + { + "epoch": 0.3876956580917381, + "grad_norm": 2.1549232006073, + "learning_rate": 1.2262832220253511e-05, + "loss": 2.481758689880371, + "step": 142420 + }, + { + "epoch": 0.3877501020824827, + "grad_norm": 2.2378015518188477, + "learning_rate": 1.2261741856344555e-05, + "loss": 2.4875858306884764, + "step": 142440 + }, + { + "epoch": 0.3878045460732272, + "grad_norm": 2.1928629875183105, + "learning_rate": 1.2260651492435603e-05, + "loss": 2.507202911376953, + "step": 142460 + }, + { + "epoch": 0.3878589900639717, + "grad_norm": 2.1792850494384766, + "learning_rate": 1.2259561128526647e-05, + "loss": 2.4890092849731444, + "step": 142480 + }, + { + "epoch": 0.38791343405471623, + "grad_norm": 2.114487409591675, + "learning_rate": 1.2258470764617691e-05, + "loss": 2.500248908996582, + "step": 142500 + }, + { + "epoch": 0.38796787804546073, + "grad_norm": 2.1132612228393555, + "learning_rate": 1.2257380400708739e-05, + "loss": 2.4682249069213866, + "step": 142520 + }, + { + "epoch": 0.38802232203620524, + "grad_norm": 2.1505396366119385, + "learning_rate": 1.2256290036799783e-05, + "loss": 2.4993085861206055, + "step": 142540 + }, + { + "epoch": 0.3880767660269498, + "grad_norm": 2.155880928039551, + "learning_rate": 1.2255199672890827e-05, + "loss": 2.4736055374145507, + "step": 142560 + }, + { + "epoch": 0.3881312100176943, + "grad_norm": 2.257795572280884, + "learning_rate": 1.2254109308981874e-05, + "loss": 2.4771493911743163, + "step": 142580 + }, + { + "epoch": 0.3881856540084388, + "grad_norm": 2.0516748428344727, + "learning_rate": 1.2253018945072918e-05, + "loss": 2.4907758712768553, + "step": 142600 + }, + { + "epoch": 0.38824009799918335, + "grad_norm": 2.1832847595214844, + "learning_rate": 1.2251928581163966e-05, + "loss": 2.4831001281738283, + "step": 142620 + }, + { + "epoch": 0.38829454198992785, + "grad_norm": 2.16975736618042, + "learning_rate": 1.225083821725501e-05, + "loss": 2.510794830322266, + "step": 142640 + }, + { + "epoch": 0.38834898598067236, + "grad_norm": 2.5651166439056396, + "learning_rate": 1.2249747853346054e-05, + "loss": 2.5033552169799806, + "step": 142660 + }, + { + "epoch": 0.3884034299714169, + "grad_norm": 2.1222944259643555, + "learning_rate": 1.2248657489437102e-05, + "loss": 2.4980167388916015, + "step": 142680 + }, + { + "epoch": 0.3884578739621614, + "grad_norm": 2.627607583999634, + "learning_rate": 1.2247567125528146e-05, + "loss": 2.4861907958984375, + "step": 142700 + }, + { + "epoch": 0.388512317952906, + "grad_norm": 2.2947027683258057, + "learning_rate": 1.224647676161919e-05, + "loss": 2.4914180755615236, + "step": 142720 + }, + { + "epoch": 0.3885667619436505, + "grad_norm": 2.0998148918151855, + "learning_rate": 1.2245386397710237e-05, + "loss": 2.4714975357055664, + "step": 142740 + }, + { + "epoch": 0.388621205934395, + "grad_norm": 2.206089496612549, + "learning_rate": 1.2244296033801281e-05, + "loss": 2.4895267486572266, + "step": 142760 + }, + { + "epoch": 0.38867564992513953, + "grad_norm": 2.1822872161865234, + "learning_rate": 1.2243205669892329e-05, + "loss": 2.4929534912109377, + "step": 142780 + }, + { + "epoch": 0.38873009391588403, + "grad_norm": 2.152040719985962, + "learning_rate": 1.2242115305983373e-05, + "loss": 2.508617401123047, + "step": 142800 + }, + { + "epoch": 0.38878453790662854, + "grad_norm": 2.111116647720337, + "learning_rate": 1.2241024942074417e-05, + "loss": 2.5021587371826173, + "step": 142820 + }, + { + "epoch": 0.3888389818973731, + "grad_norm": 2.192352294921875, + "learning_rate": 1.2239934578165465e-05, + "loss": 2.496343803405762, + "step": 142840 + }, + { + "epoch": 0.3888934258881176, + "grad_norm": 2.0788421630859375, + "learning_rate": 1.2238844214256509e-05, + "loss": 2.48618221282959, + "step": 142860 + }, + { + "epoch": 0.3889478698788621, + "grad_norm": 2.1775755882263184, + "learning_rate": 1.2237753850347553e-05, + "loss": 2.5004276275634765, + "step": 142880 + }, + { + "epoch": 0.38900231386960665, + "grad_norm": 2.1505062580108643, + "learning_rate": 1.22366634864386e-05, + "loss": 2.499015235900879, + "step": 142900 + }, + { + "epoch": 0.38905675786035115, + "grad_norm": 2.570119857788086, + "learning_rate": 1.2235573122529645e-05, + "loss": 2.5178913116455077, + "step": 142920 + }, + { + "epoch": 0.3891112018510957, + "grad_norm": 2.1427836418151855, + "learning_rate": 1.2234482758620692e-05, + "loss": 2.4805181503295897, + "step": 142940 + }, + { + "epoch": 0.3891656458418402, + "grad_norm": 2.113156795501709, + "learning_rate": 1.2233392394711736e-05, + "loss": 2.481919288635254, + "step": 142960 + }, + { + "epoch": 0.3892200898325847, + "grad_norm": 2.239677906036377, + "learning_rate": 1.223230203080278e-05, + "loss": 2.507853698730469, + "step": 142980 + }, + { + "epoch": 0.38927453382332927, + "grad_norm": 2.102306365966797, + "learning_rate": 1.2231211666893828e-05, + "loss": 2.492658805847168, + "step": 143000 + }, + { + "epoch": 0.3893289778140738, + "grad_norm": 2.1555335521698, + "learning_rate": 1.2230121302984872e-05, + "loss": 2.4997322082519533, + "step": 143020 + }, + { + "epoch": 0.3893834218048183, + "grad_norm": 2.14572811126709, + "learning_rate": 1.2229030939075916e-05, + "loss": 2.471004676818848, + "step": 143040 + }, + { + "epoch": 0.38943786579556283, + "grad_norm": 2.2655913829803467, + "learning_rate": 1.2227940575166964e-05, + "loss": 2.492629623413086, + "step": 143060 + }, + { + "epoch": 0.38949230978630733, + "grad_norm": 2.111428737640381, + "learning_rate": 1.2226850211258008e-05, + "loss": 2.534157943725586, + "step": 143080 + }, + { + "epoch": 0.38954675377705184, + "grad_norm": 2.126861095428467, + "learning_rate": 1.2225759847349055e-05, + "loss": 2.4520648956298827, + "step": 143100 + }, + { + "epoch": 0.3896011977677964, + "grad_norm": 2.4630751609802246, + "learning_rate": 1.22246694834401e-05, + "loss": 2.4879459381103515, + "step": 143120 + }, + { + "epoch": 0.3896556417585409, + "grad_norm": 2.2184948921203613, + "learning_rate": 1.2223579119531143e-05, + "loss": 2.489033317565918, + "step": 143140 + }, + { + "epoch": 0.3897100857492854, + "grad_norm": 2.0322155952453613, + "learning_rate": 1.2222488755622191e-05, + "loss": 2.4984624862670897, + "step": 143160 + }, + { + "epoch": 0.38976452974002995, + "grad_norm": 2.1879968643188477, + "learning_rate": 1.2221398391713235e-05, + "loss": 2.499950981140137, + "step": 143180 + }, + { + "epoch": 0.38981897373077445, + "grad_norm": 2.1832470893859863, + "learning_rate": 1.2220308027804279e-05, + "loss": 2.4979570388793944, + "step": 143200 + }, + { + "epoch": 0.389873417721519, + "grad_norm": 2.061176300048828, + "learning_rate": 1.2219217663895327e-05, + "loss": 2.525920295715332, + "step": 143220 + }, + { + "epoch": 0.3899278617122635, + "grad_norm": 2.1632068157196045, + "learning_rate": 1.221812729998637e-05, + "loss": 2.4760066986083986, + "step": 143240 + }, + { + "epoch": 0.389982305703008, + "grad_norm": 2.145117998123169, + "learning_rate": 1.2217036936077418e-05, + "loss": 2.4932876586914063, + "step": 143260 + }, + { + "epoch": 0.39003674969375257, + "grad_norm": 2.060105085372925, + "learning_rate": 1.2215946572168462e-05, + "loss": 2.5016809463500977, + "step": 143280 + }, + { + "epoch": 0.3900911936844971, + "grad_norm": 2.4147021770477295, + "learning_rate": 1.2214856208259506e-05, + "loss": 2.4866851806640624, + "step": 143300 + }, + { + "epoch": 0.3901456376752416, + "grad_norm": 2.1495048999786377, + "learning_rate": 1.2213765844350554e-05, + "loss": 2.489425468444824, + "step": 143320 + }, + { + "epoch": 0.39020008166598613, + "grad_norm": 8.605986595153809, + "learning_rate": 1.2212675480441598e-05, + "loss": 2.5029773712158203, + "step": 143340 + }, + { + "epoch": 0.39025452565673063, + "grad_norm": 2.4173808097839355, + "learning_rate": 1.2211585116532642e-05, + "loss": 2.5032100677490234, + "step": 143360 + }, + { + "epoch": 0.39030896964747513, + "grad_norm": 2.232084274291992, + "learning_rate": 1.221049475262369e-05, + "loss": 2.509193992614746, + "step": 143380 + }, + { + "epoch": 0.3903634136382197, + "grad_norm": 2.263230323791504, + "learning_rate": 1.2209404388714734e-05, + "loss": 2.521345329284668, + "step": 143400 + }, + { + "epoch": 0.3904178576289642, + "grad_norm": 2.111051559448242, + "learning_rate": 1.2208314024805781e-05, + "loss": 2.474875831604004, + "step": 143420 + }, + { + "epoch": 0.39047230161970875, + "grad_norm": 2.1508948802948, + "learning_rate": 1.2207223660896825e-05, + "loss": 2.4813928604125977, + "step": 143440 + }, + { + "epoch": 0.39052674561045325, + "grad_norm": 2.402379274368286, + "learning_rate": 1.220613329698787e-05, + "loss": 2.5186485290527343, + "step": 143460 + }, + { + "epoch": 0.39058118960119775, + "grad_norm": 2.745248794555664, + "learning_rate": 1.2205042933078917e-05, + "loss": 2.5362741470336916, + "step": 143480 + }, + { + "epoch": 0.3906356335919423, + "grad_norm": 2.306565523147583, + "learning_rate": 1.2203952569169961e-05, + "loss": 2.4966787338256835, + "step": 143500 + }, + { + "epoch": 0.3906900775826868, + "grad_norm": 2.121882915496826, + "learning_rate": 1.2202862205261009e-05, + "loss": 2.5128686904907225, + "step": 143520 + }, + { + "epoch": 0.3907445215734313, + "grad_norm": 2.2568085193634033, + "learning_rate": 1.2201771841352053e-05, + "loss": 2.5249929428100586, + "step": 143540 + }, + { + "epoch": 0.39079896556417587, + "grad_norm": 2.141298770904541, + "learning_rate": 1.2200681477443097e-05, + "loss": 2.4736265182495116, + "step": 143560 + }, + { + "epoch": 0.3908534095549204, + "grad_norm": 2.210634708404541, + "learning_rate": 1.2199591113534144e-05, + "loss": 2.5080102920532226, + "step": 143580 + }, + { + "epoch": 0.3909078535456649, + "grad_norm": 2.208017110824585, + "learning_rate": 1.2198500749625188e-05, + "loss": 2.479915809631348, + "step": 143600 + }, + { + "epoch": 0.39096229753640943, + "grad_norm": 2.0772008895874023, + "learning_rate": 1.2197410385716233e-05, + "loss": 2.49200553894043, + "step": 143620 + }, + { + "epoch": 0.39101674152715393, + "grad_norm": 2.1977250576019287, + "learning_rate": 1.219632002180728e-05, + "loss": 2.5074018478393554, + "step": 143640 + }, + { + "epoch": 0.3910711855178985, + "grad_norm": 2.3517885208129883, + "learning_rate": 1.2195229657898324e-05, + "loss": 2.50161075592041, + "step": 143660 + }, + { + "epoch": 0.391125629508643, + "grad_norm": 2.3201985359191895, + "learning_rate": 1.2194139293989372e-05, + "loss": 2.500489044189453, + "step": 143680 + }, + { + "epoch": 0.3911800734993875, + "grad_norm": 3.0183587074279785, + "learning_rate": 1.2193048930080416e-05, + "loss": 2.488172721862793, + "step": 143700 + }, + { + "epoch": 0.39123451749013205, + "grad_norm": 2.095446825027466, + "learning_rate": 1.219195856617146e-05, + "loss": 2.464985656738281, + "step": 143720 + }, + { + "epoch": 0.39128896148087655, + "grad_norm": 2.171452522277832, + "learning_rate": 1.2190868202262507e-05, + "loss": 2.4920026779174806, + "step": 143740 + }, + { + "epoch": 0.39134340547162105, + "grad_norm": 2.2205147743225098, + "learning_rate": 1.2189777838353552e-05, + "loss": 2.492312431335449, + "step": 143760 + }, + { + "epoch": 0.3913978494623656, + "grad_norm": 2.208735466003418, + "learning_rate": 1.2188687474444596e-05, + "loss": 2.4797641754150392, + "step": 143780 + }, + { + "epoch": 0.3914522934531101, + "grad_norm": 2.1105566024780273, + "learning_rate": 1.2187597110535643e-05, + "loss": 2.4789287567138674, + "step": 143800 + }, + { + "epoch": 0.3915067374438546, + "grad_norm": 2.0470054149627686, + "learning_rate": 1.2186506746626687e-05, + "loss": 2.4941919326782225, + "step": 143820 + }, + { + "epoch": 0.39156118143459917, + "grad_norm": 2.1387557983398438, + "learning_rate": 1.2185416382717735e-05, + "loss": 2.5158084869384765, + "step": 143840 + }, + { + "epoch": 0.39161562542534367, + "grad_norm": 2.1840877532958984, + "learning_rate": 1.2184326018808779e-05, + "loss": 2.494395065307617, + "step": 143860 + }, + { + "epoch": 0.3916700694160882, + "grad_norm": 2.0779802799224854, + "learning_rate": 1.2183235654899823e-05, + "loss": 2.5025182723999024, + "step": 143880 + }, + { + "epoch": 0.39172451340683273, + "grad_norm": 2.5468571186065674, + "learning_rate": 1.218214529099087e-05, + "loss": 2.505259704589844, + "step": 143900 + }, + { + "epoch": 0.39177895739757723, + "grad_norm": 2.119696617126465, + "learning_rate": 1.2181054927081915e-05, + "loss": 2.4889875411987306, + "step": 143920 + }, + { + "epoch": 0.3918334013883218, + "grad_norm": 2.144334316253662, + "learning_rate": 1.2179964563172959e-05, + "loss": 2.4790096282958984, + "step": 143940 + }, + { + "epoch": 0.3918878453790663, + "grad_norm": 2.1474130153656006, + "learning_rate": 1.2178874199264006e-05, + "loss": 2.5032154083251954, + "step": 143960 + }, + { + "epoch": 0.3919422893698108, + "grad_norm": 2.118858814239502, + "learning_rate": 1.217778383535505e-05, + "loss": 2.5045881271362305, + "step": 143980 + }, + { + "epoch": 0.39199673336055535, + "grad_norm": 3.1448047161102295, + "learning_rate": 1.2176693471446098e-05, + "loss": 2.460299491882324, + "step": 144000 + }, + { + "epoch": 0.39205117735129985, + "grad_norm": 2.2550723552703857, + "learning_rate": 1.2175603107537142e-05, + "loss": 2.519321250915527, + "step": 144020 + }, + { + "epoch": 0.39210562134204435, + "grad_norm": 2.2351014614105225, + "learning_rate": 1.2174512743628186e-05, + "loss": 2.5138708114624024, + "step": 144040 + }, + { + "epoch": 0.3921600653327889, + "grad_norm": 2.223036527633667, + "learning_rate": 1.2173422379719234e-05, + "loss": 2.4737550735473635, + "step": 144060 + }, + { + "epoch": 0.3922145093235334, + "grad_norm": 2.0767829418182373, + "learning_rate": 1.2172332015810278e-05, + "loss": 2.4705364227294924, + "step": 144080 + }, + { + "epoch": 0.3922689533142779, + "grad_norm": 2.125229597091675, + "learning_rate": 1.2171241651901322e-05, + "loss": 2.479528045654297, + "step": 144100 + }, + { + "epoch": 0.39232339730502247, + "grad_norm": 2.2632007598876953, + "learning_rate": 1.217015128799237e-05, + "loss": 2.4851879119873046, + "step": 144120 + }, + { + "epoch": 0.39237784129576697, + "grad_norm": 2.174041271209717, + "learning_rate": 1.2169060924083413e-05, + "loss": 2.4773077011108398, + "step": 144140 + }, + { + "epoch": 0.39243228528651153, + "grad_norm": 2.146024703979492, + "learning_rate": 1.2167970560174461e-05, + "loss": 2.5076087951660155, + "step": 144160 + }, + { + "epoch": 0.39248672927725603, + "grad_norm": 2.7995331287384033, + "learning_rate": 1.2166880196265505e-05, + "loss": 2.498065376281738, + "step": 144180 + }, + { + "epoch": 0.39254117326800053, + "grad_norm": 2.348876714706421, + "learning_rate": 1.216578983235655e-05, + "loss": 2.4636257171630858, + "step": 144200 + }, + { + "epoch": 0.3925956172587451, + "grad_norm": 2.1424710750579834, + "learning_rate": 1.2164699468447597e-05, + "loss": 2.492258071899414, + "step": 144220 + }, + { + "epoch": 0.3926500612494896, + "grad_norm": 2.146416187286377, + "learning_rate": 1.216360910453864e-05, + "loss": 2.5044294357299806, + "step": 144240 + }, + { + "epoch": 0.3927045052402341, + "grad_norm": 2.1238760948181152, + "learning_rate": 1.2162518740629685e-05, + "loss": 2.518048095703125, + "step": 144260 + }, + { + "epoch": 0.39275894923097865, + "grad_norm": 2.1623904705047607, + "learning_rate": 1.2161428376720732e-05, + "loss": 2.450493812561035, + "step": 144280 + }, + { + "epoch": 0.39281339322172315, + "grad_norm": 2.2213897705078125, + "learning_rate": 1.2160338012811777e-05, + "loss": 2.474256134033203, + "step": 144300 + }, + { + "epoch": 0.39286783721246765, + "grad_norm": 2.337479829788208, + "learning_rate": 1.2159247648902824e-05, + "loss": 2.485762596130371, + "step": 144320 + }, + { + "epoch": 0.3929222812032122, + "grad_norm": 2.5371181964874268, + "learning_rate": 1.2158157284993868e-05, + "loss": 2.4926076889038087, + "step": 144340 + }, + { + "epoch": 0.3929767251939567, + "grad_norm": 2.192487955093384, + "learning_rate": 1.2157066921084912e-05, + "loss": 2.483180618286133, + "step": 144360 + }, + { + "epoch": 0.3930311691847012, + "grad_norm": 3.3094394207000732, + "learning_rate": 1.215597655717596e-05, + "loss": 2.4847177505493163, + "step": 144380 + }, + { + "epoch": 0.39308561317544577, + "grad_norm": 2.207512378692627, + "learning_rate": 1.2154886193267004e-05, + "loss": 2.491455078125, + "step": 144400 + }, + { + "epoch": 0.39314005716619027, + "grad_norm": 2.063809871673584, + "learning_rate": 1.2153795829358048e-05, + "loss": 2.477769470214844, + "step": 144420 + }, + { + "epoch": 0.39319450115693483, + "grad_norm": 2.3881847858428955, + "learning_rate": 1.2152705465449096e-05, + "loss": 2.5236026763916017, + "step": 144440 + }, + { + "epoch": 0.39324894514767933, + "grad_norm": 2.41764760017395, + "learning_rate": 1.215161510154014e-05, + "loss": 2.482106018066406, + "step": 144460 + }, + { + "epoch": 0.39330338913842383, + "grad_norm": 2.1001157760620117, + "learning_rate": 1.2150524737631187e-05, + "loss": 2.4945343017578123, + "step": 144480 + }, + { + "epoch": 0.3933578331291684, + "grad_norm": 2.156102180480957, + "learning_rate": 1.2149434373722231e-05, + "loss": 2.5030439376831053, + "step": 144500 + }, + { + "epoch": 0.3934122771199129, + "grad_norm": 2.2174715995788574, + "learning_rate": 1.2148344009813275e-05, + "loss": 2.507817840576172, + "step": 144520 + }, + { + "epoch": 0.3934667211106574, + "grad_norm": 2.0347938537597656, + "learning_rate": 1.2147253645904323e-05, + "loss": 2.4884098052978514, + "step": 144540 + }, + { + "epoch": 0.39352116510140195, + "grad_norm": 2.2187628746032715, + "learning_rate": 1.2146163281995367e-05, + "loss": 2.4914493560791016, + "step": 144560 + }, + { + "epoch": 0.39357560909214645, + "grad_norm": 2.221794843673706, + "learning_rate": 1.2145072918086411e-05, + "loss": 2.506939697265625, + "step": 144580 + }, + { + "epoch": 0.39363005308289095, + "grad_norm": 2.1009109020233154, + "learning_rate": 1.2143982554177459e-05, + "loss": 2.5020673751831053, + "step": 144600 + }, + { + "epoch": 0.3936844970736355, + "grad_norm": 2.1705737113952637, + "learning_rate": 1.2142892190268503e-05, + "loss": 2.446030044555664, + "step": 144620 + }, + { + "epoch": 0.39373894106438, + "grad_norm": 2.1349878311157227, + "learning_rate": 1.214180182635955e-05, + "loss": 2.5017044067382814, + "step": 144640 + }, + { + "epoch": 0.39379338505512457, + "grad_norm": 2.2893707752227783, + "learning_rate": 1.2140711462450594e-05, + "loss": 2.5305347442626953, + "step": 144660 + }, + { + "epoch": 0.39384782904586907, + "grad_norm": 2.9642457962036133, + "learning_rate": 1.2139621098541638e-05, + "loss": 2.5109270095825194, + "step": 144680 + }, + { + "epoch": 0.39390227303661357, + "grad_norm": 2.2198028564453125, + "learning_rate": 1.2138530734632686e-05, + "loss": 2.52431697845459, + "step": 144700 + }, + { + "epoch": 0.3939567170273581, + "grad_norm": 2.1965856552124023, + "learning_rate": 1.213744037072373e-05, + "loss": 2.493977737426758, + "step": 144720 + }, + { + "epoch": 0.39401116101810263, + "grad_norm": 4.359067440032959, + "learning_rate": 1.2136350006814774e-05, + "loss": 2.510847282409668, + "step": 144740 + }, + { + "epoch": 0.39406560500884713, + "grad_norm": 2.0599138736724854, + "learning_rate": 1.2135259642905822e-05, + "loss": 2.499287223815918, + "step": 144760 + }, + { + "epoch": 0.3941200489995917, + "grad_norm": 2.323305130004883, + "learning_rate": 1.2134169278996866e-05, + "loss": 2.5153339385986326, + "step": 144780 + }, + { + "epoch": 0.3941744929903362, + "grad_norm": 2.225370407104492, + "learning_rate": 1.2133078915087912e-05, + "loss": 2.480253791809082, + "step": 144800 + }, + { + "epoch": 0.3942289369810807, + "grad_norm": 2.097839117050171, + "learning_rate": 1.2131988551178957e-05, + "loss": 2.477962112426758, + "step": 144820 + }, + { + "epoch": 0.39428338097182525, + "grad_norm": 2.436978816986084, + "learning_rate": 1.2130898187270002e-05, + "loss": 2.4829032897949217, + "step": 144840 + }, + { + "epoch": 0.39433782496256975, + "grad_norm": 2.148982524871826, + "learning_rate": 1.2129807823361049e-05, + "loss": 2.501984214782715, + "step": 144860 + }, + { + "epoch": 0.39439226895331425, + "grad_norm": 2.1906697750091553, + "learning_rate": 1.2128717459452093e-05, + "loss": 2.4909616470336915, + "step": 144880 + }, + { + "epoch": 0.3944467129440588, + "grad_norm": 2.1746788024902344, + "learning_rate": 1.2127627095543137e-05, + "loss": 2.5243431091308595, + "step": 144900 + }, + { + "epoch": 0.3945011569348033, + "grad_norm": 2.1607511043548584, + "learning_rate": 1.2126536731634185e-05, + "loss": 2.476394271850586, + "step": 144920 + }, + { + "epoch": 0.39455560092554787, + "grad_norm": 2.115367889404297, + "learning_rate": 1.2125446367725229e-05, + "loss": 2.4955286026000976, + "step": 144940 + }, + { + "epoch": 0.39461004491629237, + "grad_norm": 2.2980270385742188, + "learning_rate": 1.2124356003816275e-05, + "loss": 2.536796760559082, + "step": 144960 + }, + { + "epoch": 0.39466448890703687, + "grad_norm": 2.25602388381958, + "learning_rate": 1.212326563990732e-05, + "loss": 2.5254695892333983, + "step": 144980 + }, + { + "epoch": 0.3947189328977814, + "grad_norm": 2.1632473468780518, + "learning_rate": 1.2122175275998365e-05, + "loss": 2.4973499298095705, + "step": 145000 + }, + { + "epoch": 0.39477337688852593, + "grad_norm": 2.2673866748809814, + "learning_rate": 1.2121084912089412e-05, + "loss": 2.476664733886719, + "step": 145020 + }, + { + "epoch": 0.39482782087927043, + "grad_norm": 2.1086502075195312, + "learning_rate": 1.2119994548180456e-05, + "loss": 2.4767913818359375, + "step": 145040 + }, + { + "epoch": 0.394882264870015, + "grad_norm": 2.068617343902588, + "learning_rate": 1.21189041842715e-05, + "loss": 2.4827463150024416, + "step": 145060 + }, + { + "epoch": 0.3949367088607595, + "grad_norm": 2.188645362854004, + "learning_rate": 1.2117813820362548e-05, + "loss": 2.489198684692383, + "step": 145080 + }, + { + "epoch": 0.394991152851504, + "grad_norm": 2.1744110584259033, + "learning_rate": 1.2116723456453592e-05, + "loss": 2.4708559036254885, + "step": 145100 + }, + { + "epoch": 0.39504559684224855, + "grad_norm": 2.066619634628296, + "learning_rate": 1.2115633092544638e-05, + "loss": 2.4928192138671874, + "step": 145120 + }, + { + "epoch": 0.39510004083299305, + "grad_norm": 2.1422672271728516, + "learning_rate": 1.2114542728635684e-05, + "loss": 2.4784807205200194, + "step": 145140 + }, + { + "epoch": 0.3951544848237376, + "grad_norm": 2.181006669998169, + "learning_rate": 1.2113452364726728e-05, + "loss": 2.498662757873535, + "step": 145160 + }, + { + "epoch": 0.3952089288144821, + "grad_norm": 2.248619556427002, + "learning_rate": 1.2112362000817775e-05, + "loss": 2.4701236724853515, + "step": 145180 + }, + { + "epoch": 0.3952633728052266, + "grad_norm": 2.2160120010375977, + "learning_rate": 1.211127163690882e-05, + "loss": 2.4830739974975584, + "step": 145200 + }, + { + "epoch": 0.39531781679597117, + "grad_norm": 2.2146427631378174, + "learning_rate": 1.2110181272999865e-05, + "loss": 2.5007116317749025, + "step": 145220 + }, + { + "epoch": 0.39537226078671567, + "grad_norm": 2.0213592052459717, + "learning_rate": 1.2109090909090911e-05, + "loss": 2.4995281219482424, + "step": 145240 + }, + { + "epoch": 0.39542670477746017, + "grad_norm": 2.091359853744507, + "learning_rate": 1.2108000545181955e-05, + "loss": 2.4977903366088867, + "step": 145260 + }, + { + "epoch": 0.3954811487682047, + "grad_norm": 2.097761631011963, + "learning_rate": 1.2106910181273001e-05, + "loss": 2.5093292236328124, + "step": 145280 + }, + { + "epoch": 0.39553559275894923, + "grad_norm": 2.6207327842712402, + "learning_rate": 1.2105819817364047e-05, + "loss": 2.4919261932373047, + "step": 145300 + }, + { + "epoch": 0.39559003674969373, + "grad_norm": 2.251000165939331, + "learning_rate": 1.210472945345509e-05, + "loss": 2.525754928588867, + "step": 145320 + }, + { + "epoch": 0.3956444807404383, + "grad_norm": 2.1584665775299072, + "learning_rate": 1.2103639089546138e-05, + "loss": 2.487440490722656, + "step": 145340 + }, + { + "epoch": 0.3956989247311828, + "grad_norm": 2.1005542278289795, + "learning_rate": 1.2102548725637182e-05, + "loss": 2.5103023529052733, + "step": 145360 + }, + { + "epoch": 0.3957533687219273, + "grad_norm": 2.2617642879486084, + "learning_rate": 1.2101458361728228e-05, + "loss": 2.5149152755737303, + "step": 145380 + }, + { + "epoch": 0.39580781271267185, + "grad_norm": 2.1444435119628906, + "learning_rate": 1.2100367997819274e-05, + "loss": 2.4901552200317383, + "step": 145400 + }, + { + "epoch": 0.39586225670341635, + "grad_norm": 2.172032117843628, + "learning_rate": 1.2099277633910318e-05, + "loss": 2.4742671966552736, + "step": 145420 + }, + { + "epoch": 0.3959167006941609, + "grad_norm": 2.1974120140075684, + "learning_rate": 1.2098187270001364e-05, + "loss": 2.497245216369629, + "step": 145440 + }, + { + "epoch": 0.3959711446849054, + "grad_norm": 2.8608710765838623, + "learning_rate": 1.209709690609241e-05, + "loss": 2.501508331298828, + "step": 145460 + }, + { + "epoch": 0.3960255886756499, + "grad_norm": 2.2742865085601807, + "learning_rate": 1.2096006542183454e-05, + "loss": 2.4857194900512694, + "step": 145480 + }, + { + "epoch": 0.39608003266639447, + "grad_norm": 2.2561914920806885, + "learning_rate": 1.2094916178274501e-05, + "loss": 2.489921760559082, + "step": 145500 + }, + { + "epoch": 0.39613447665713897, + "grad_norm": 3.023381471633911, + "learning_rate": 1.2093825814365546e-05, + "loss": 2.492848205566406, + "step": 145520 + }, + { + "epoch": 0.39618892064788347, + "grad_norm": 2.3423233032226562, + "learning_rate": 1.2092735450456591e-05, + "loss": 2.5068248748779296, + "step": 145540 + }, + { + "epoch": 0.396243364638628, + "grad_norm": 2.1511197090148926, + "learning_rate": 1.2091645086547637e-05, + "loss": 2.491425895690918, + "step": 145560 + }, + { + "epoch": 0.3962978086293725, + "grad_norm": 2.095489501953125, + "learning_rate": 1.2090554722638681e-05, + "loss": 2.5154392242431642, + "step": 145580 + }, + { + "epoch": 0.39635225262011703, + "grad_norm": 2.2106640338897705, + "learning_rate": 1.2089464358729727e-05, + "loss": 2.438930320739746, + "step": 145600 + }, + { + "epoch": 0.3964066966108616, + "grad_norm": 2.8368866443634033, + "learning_rate": 1.2088373994820773e-05, + "loss": 2.5000516891479494, + "step": 145620 + }, + { + "epoch": 0.3964611406016061, + "grad_norm": 2.150322437286377, + "learning_rate": 1.2087283630911817e-05, + "loss": 2.4801679611206056, + "step": 145640 + }, + { + "epoch": 0.39651558459235065, + "grad_norm": 2.206772565841675, + "learning_rate": 1.2086193267002864e-05, + "loss": 2.5050966262817385, + "step": 145660 + }, + { + "epoch": 0.39657002858309515, + "grad_norm": 2.1225521564483643, + "learning_rate": 1.2085102903093909e-05, + "loss": 2.5168004989624024, + "step": 145680 + }, + { + "epoch": 0.39662447257383965, + "grad_norm": 2.0542304515838623, + "learning_rate": 1.2084012539184954e-05, + "loss": 2.4954317092895506, + "step": 145700 + }, + { + "epoch": 0.3966789165645842, + "grad_norm": 2.309872627258301, + "learning_rate": 1.2082922175276e-05, + "loss": 2.507085609436035, + "step": 145720 + }, + { + "epoch": 0.3967333605553287, + "grad_norm": 2.157416582107544, + "learning_rate": 1.2081831811367044e-05, + "loss": 2.5307369232177734, + "step": 145740 + }, + { + "epoch": 0.3967878045460732, + "grad_norm": 2.5149319171905518, + "learning_rate": 1.208074144745809e-05, + "loss": 2.4869483947753905, + "step": 145760 + }, + { + "epoch": 0.39684224853681777, + "grad_norm": 2.1912548542022705, + "learning_rate": 1.2079651083549136e-05, + "loss": 2.4955263137817383, + "step": 145780 + }, + { + "epoch": 0.39689669252756227, + "grad_norm": 3.1394288539886475, + "learning_rate": 1.207856071964018e-05, + "loss": 2.483481025695801, + "step": 145800 + }, + { + "epoch": 0.39695113651830677, + "grad_norm": 2.1641831398010254, + "learning_rate": 1.2077470355731226e-05, + "loss": 2.4515718460083007, + "step": 145820 + }, + { + "epoch": 0.3970055805090513, + "grad_norm": 2.1337616443634033, + "learning_rate": 1.2076379991822272e-05, + "loss": 2.489574432373047, + "step": 145840 + }, + { + "epoch": 0.3970600244997958, + "grad_norm": 2.1311593055725098, + "learning_rate": 1.2075289627913318e-05, + "loss": 2.4820123672485352, + "step": 145860 + }, + { + "epoch": 0.3971144684905404, + "grad_norm": 2.182166337966919, + "learning_rate": 1.2074199264004363e-05, + "loss": 2.485869216918945, + "step": 145880 + }, + { + "epoch": 0.3971689124812849, + "grad_norm": 2.1379806995391846, + "learning_rate": 1.2073108900095407e-05, + "loss": 2.4620792388916017, + "step": 145900 + }, + { + "epoch": 0.3972233564720294, + "grad_norm": 2.077925682067871, + "learning_rate": 1.2072018536186453e-05, + "loss": 2.516383171081543, + "step": 145920 + }, + { + "epoch": 0.39727780046277394, + "grad_norm": 2.1412458419799805, + "learning_rate": 1.2070928172277499e-05, + "loss": 2.492899703979492, + "step": 145940 + }, + { + "epoch": 0.39733224445351845, + "grad_norm": 2.0723676681518555, + "learning_rate": 1.2069837808368543e-05, + "loss": 2.5104724884033205, + "step": 145960 + }, + { + "epoch": 0.39738668844426295, + "grad_norm": 2.2366886138916016, + "learning_rate": 1.2068747444459589e-05, + "loss": 2.5039934158325194, + "step": 145980 + }, + { + "epoch": 0.3974411324350075, + "grad_norm": 2.0761027336120605, + "learning_rate": 1.2067657080550635e-05, + "loss": 2.4992095947265627, + "step": 146000 + }, + { + "epoch": 0.397495576425752, + "grad_norm": 2.0744218826293945, + "learning_rate": 1.206656671664168e-05, + "loss": 2.480034828186035, + "step": 146020 + }, + { + "epoch": 0.3975500204164965, + "grad_norm": 2.3112852573394775, + "learning_rate": 1.2065476352732726e-05, + "loss": 2.4809070587158204, + "step": 146040 + }, + { + "epoch": 0.39760446440724106, + "grad_norm": 2.2205381393432617, + "learning_rate": 1.206438598882377e-05, + "loss": 2.4989898681640623, + "step": 146060 + }, + { + "epoch": 0.39765890839798557, + "grad_norm": 2.257880926132202, + "learning_rate": 1.2063295624914816e-05, + "loss": 2.4847503662109376, + "step": 146080 + }, + { + "epoch": 0.39771335238873007, + "grad_norm": 2.1563057899475098, + "learning_rate": 1.2062205261005862e-05, + "loss": 2.5301586151123048, + "step": 146100 + }, + { + "epoch": 0.3977677963794746, + "grad_norm": 2.0654194355010986, + "learning_rate": 1.2061114897096906e-05, + "loss": 2.4966997146606444, + "step": 146120 + }, + { + "epoch": 0.3978222403702191, + "grad_norm": 2.2614402770996094, + "learning_rate": 1.2060024533187952e-05, + "loss": 2.476763153076172, + "step": 146140 + }, + { + "epoch": 0.3978766843609637, + "grad_norm": 2.1077544689178467, + "learning_rate": 1.2058934169278998e-05, + "loss": 2.4586069107055666, + "step": 146160 + }, + { + "epoch": 0.3979311283517082, + "grad_norm": 2.1692960262298584, + "learning_rate": 1.2057843805370044e-05, + "loss": 2.5204206466674806, + "step": 146180 + }, + { + "epoch": 0.3979855723424527, + "grad_norm": 2.122950315475464, + "learning_rate": 1.205675344146109e-05, + "loss": 2.4781150817871094, + "step": 146200 + }, + { + "epoch": 0.39804001633319724, + "grad_norm": 2.1906187534332275, + "learning_rate": 1.2055663077552134e-05, + "loss": 2.507012939453125, + "step": 146220 + }, + { + "epoch": 0.39809446032394175, + "grad_norm": 2.3846049308776855, + "learning_rate": 1.205457271364318e-05, + "loss": 2.4534069061279298, + "step": 146240 + }, + { + "epoch": 0.39814890431468625, + "grad_norm": 2.149662971496582, + "learning_rate": 1.2053482349734225e-05, + "loss": 2.4999265670776367, + "step": 146260 + }, + { + "epoch": 0.3982033483054308, + "grad_norm": 2.3248798847198486, + "learning_rate": 1.205239198582527e-05, + "loss": 2.4975534439086915, + "step": 146280 + }, + { + "epoch": 0.3982577922961753, + "grad_norm": 2.62652325630188, + "learning_rate": 1.2051301621916315e-05, + "loss": 2.4985372543334963, + "step": 146300 + }, + { + "epoch": 0.3983122362869198, + "grad_norm": 2.170149803161621, + "learning_rate": 1.2050211258007361e-05, + "loss": 2.5227951049804687, + "step": 146320 + }, + { + "epoch": 0.39836668027766436, + "grad_norm": 2.150800943374634, + "learning_rate": 1.2049120894098407e-05, + "loss": 2.4956621170043944, + "step": 146340 + }, + { + "epoch": 0.39842112426840887, + "grad_norm": 2.079681873321533, + "learning_rate": 1.2048030530189453e-05, + "loss": 2.4790681838989257, + "step": 146360 + }, + { + "epoch": 0.3984755682591534, + "grad_norm": 2.4281675815582275, + "learning_rate": 1.2046940166280497e-05, + "loss": 2.4996999740600585, + "step": 146380 + }, + { + "epoch": 0.3985300122498979, + "grad_norm": 2.470224618911743, + "learning_rate": 1.2045849802371542e-05, + "loss": 2.5033857345581056, + "step": 146400 + }, + { + "epoch": 0.3985844562406424, + "grad_norm": 2.175447940826416, + "learning_rate": 1.2044759438462588e-05, + "loss": 2.469499206542969, + "step": 146420 + }, + { + "epoch": 0.398638900231387, + "grad_norm": 2.0400290489196777, + "learning_rate": 1.2043669074553632e-05, + "loss": 2.473945426940918, + "step": 146440 + }, + { + "epoch": 0.3986933442221315, + "grad_norm": 2.283310651779175, + "learning_rate": 1.2042578710644678e-05, + "loss": 2.496226692199707, + "step": 146460 + }, + { + "epoch": 0.398747788212876, + "grad_norm": 2.3689780235290527, + "learning_rate": 1.2041488346735724e-05, + "loss": 2.526222991943359, + "step": 146480 + }, + { + "epoch": 0.39880223220362054, + "grad_norm": 2.1791839599609375, + "learning_rate": 1.204039798282677e-05, + "loss": 2.4982488632202147, + "step": 146500 + }, + { + "epoch": 0.39885667619436505, + "grad_norm": 2.1500978469848633, + "learning_rate": 1.2039307618917816e-05, + "loss": 2.4997787475585938, + "step": 146520 + }, + { + "epoch": 0.39891112018510955, + "grad_norm": 2.192981243133545, + "learning_rate": 1.203821725500886e-05, + "loss": 2.50359992980957, + "step": 146540 + }, + { + "epoch": 0.3989655641758541, + "grad_norm": 2.262730121612549, + "learning_rate": 1.2037126891099906e-05, + "loss": 2.4678932189941407, + "step": 146560 + }, + { + "epoch": 0.3990200081665986, + "grad_norm": 2.116469621658325, + "learning_rate": 1.2036036527190951e-05, + "loss": 2.461879920959473, + "step": 146580 + }, + { + "epoch": 0.3990744521573431, + "grad_norm": 2.201913833618164, + "learning_rate": 1.2034946163281996e-05, + "loss": 2.4604280471801756, + "step": 146600 + }, + { + "epoch": 0.39912889614808766, + "grad_norm": 2.1274874210357666, + "learning_rate": 1.2033855799373041e-05, + "loss": 2.4864126205444337, + "step": 146620 + }, + { + "epoch": 0.39918334013883217, + "grad_norm": 2.052466630935669, + "learning_rate": 1.2032765435464087e-05, + "loss": 2.488380241394043, + "step": 146640 + }, + { + "epoch": 0.3992377841295767, + "grad_norm": 2.1469743251800537, + "learning_rate": 1.2031675071555133e-05, + "loss": 2.521512985229492, + "step": 146660 + }, + { + "epoch": 0.3992922281203212, + "grad_norm": 2.165842056274414, + "learning_rate": 1.2030584707646179e-05, + "loss": 2.509779167175293, + "step": 146680 + }, + { + "epoch": 0.3993466721110657, + "grad_norm": 2.0816650390625, + "learning_rate": 1.2029494343737223e-05, + "loss": 2.4909341812133787, + "step": 146700 + }, + { + "epoch": 0.3994011161018103, + "grad_norm": 2.0506272315979004, + "learning_rate": 1.2028403979828269e-05, + "loss": 2.490904426574707, + "step": 146720 + }, + { + "epoch": 0.3994555600925548, + "grad_norm": 2.0753157138824463, + "learning_rate": 1.2027313615919314e-05, + "loss": 2.4891897201538087, + "step": 146740 + }, + { + "epoch": 0.3995100040832993, + "grad_norm": 2.4140820503234863, + "learning_rate": 1.2026223252010359e-05, + "loss": 2.5007673263549806, + "step": 146760 + }, + { + "epoch": 0.39956444807404384, + "grad_norm": 2.238896131515503, + "learning_rate": 1.2025132888101404e-05, + "loss": 2.456522560119629, + "step": 146780 + }, + { + "epoch": 0.39961889206478834, + "grad_norm": 2.1221091747283936, + "learning_rate": 1.202404252419245e-05, + "loss": 2.521109962463379, + "step": 146800 + }, + { + "epoch": 0.39967333605553285, + "grad_norm": 2.2264633178710938, + "learning_rate": 1.2022952160283496e-05, + "loss": 2.487528419494629, + "step": 146820 + }, + { + "epoch": 0.3997277800462774, + "grad_norm": 2.0758254528045654, + "learning_rate": 1.2021861796374542e-05, + "loss": 2.4767332077026367, + "step": 146840 + }, + { + "epoch": 0.3997822240370219, + "grad_norm": 2.215590715408325, + "learning_rate": 1.2020771432465586e-05, + "loss": 2.491460418701172, + "step": 146860 + }, + { + "epoch": 0.39983666802776646, + "grad_norm": 2.1549477577209473, + "learning_rate": 1.2019681068556632e-05, + "loss": 2.5092756271362306, + "step": 146880 + }, + { + "epoch": 0.39989111201851096, + "grad_norm": 2.0612785816192627, + "learning_rate": 1.2018590704647678e-05, + "loss": 2.505086326599121, + "step": 146900 + }, + { + "epoch": 0.39994555600925547, + "grad_norm": 2.3393757343292236, + "learning_rate": 1.2017500340738723e-05, + "loss": 2.508549118041992, + "step": 146920 + }, + { + "epoch": 0.4, + "grad_norm": 2.3330845832824707, + "learning_rate": 1.2016409976829767e-05, + "loss": 2.503195571899414, + "step": 146940 + }, + { + "epoch": 0.4000544439907445, + "grad_norm": 2.1766903400421143, + "learning_rate": 1.2015319612920813e-05, + "loss": 2.4997381210327148, + "step": 146960 + }, + { + "epoch": 0.400108887981489, + "grad_norm": 3.780147075653076, + "learning_rate": 1.2014229249011859e-05, + "loss": 2.4928218841552736, + "step": 146980 + }, + { + "epoch": 0.4001633319722336, + "grad_norm": 2.4168429374694824, + "learning_rate": 1.2013138885102903e-05, + "loss": 2.512074661254883, + "step": 147000 + }, + { + "epoch": 0.4002177759629781, + "grad_norm": 2.375068187713623, + "learning_rate": 1.2012048521193949e-05, + "loss": 2.514390754699707, + "step": 147020 + }, + { + "epoch": 0.4002722199537226, + "grad_norm": 2.130037784576416, + "learning_rate": 1.2010958157284995e-05, + "loss": 2.5042680740356444, + "step": 147040 + }, + { + "epoch": 0.40032666394446714, + "grad_norm": 2.788369655609131, + "learning_rate": 1.200986779337604e-05, + "loss": 2.491278648376465, + "step": 147060 + }, + { + "epoch": 0.40038110793521164, + "grad_norm": 2.1305930614471436, + "learning_rate": 1.2008777429467086e-05, + "loss": 2.5053583145141602, + "step": 147080 + }, + { + "epoch": 0.40043555192595615, + "grad_norm": 2.281494140625, + "learning_rate": 1.200768706555813e-05, + "loss": 2.501162528991699, + "step": 147100 + }, + { + "epoch": 0.4004899959167007, + "grad_norm": 6.0172438621521, + "learning_rate": 1.2006596701649176e-05, + "loss": 2.496577262878418, + "step": 147120 + }, + { + "epoch": 0.4005444399074452, + "grad_norm": 2.194733142852783, + "learning_rate": 1.2005506337740222e-05, + "loss": 2.5217700958251954, + "step": 147140 + }, + { + "epoch": 0.40059888389818976, + "grad_norm": 2.154444456100464, + "learning_rate": 1.2004415973831266e-05, + "loss": 2.478424072265625, + "step": 147160 + }, + { + "epoch": 0.40065332788893426, + "grad_norm": 2.04231858253479, + "learning_rate": 1.2003325609922312e-05, + "loss": 2.471587562561035, + "step": 147180 + }, + { + "epoch": 0.40070777187967876, + "grad_norm": 2.2072877883911133, + "learning_rate": 1.2002235246013358e-05, + "loss": 2.4735443115234377, + "step": 147200 + }, + { + "epoch": 0.4007622158704233, + "grad_norm": 2.111689567565918, + "learning_rate": 1.2001144882104404e-05, + "loss": 2.475267791748047, + "step": 147220 + }, + { + "epoch": 0.4008166598611678, + "grad_norm": 2.1971137523651123, + "learning_rate": 1.200005451819545e-05, + "loss": 2.5082752227783205, + "step": 147240 + }, + { + "epoch": 0.4008711038519123, + "grad_norm": 2.121173143386841, + "learning_rate": 1.1998964154286494e-05, + "loss": 2.5115407943725585, + "step": 147260 + }, + { + "epoch": 0.4009255478426569, + "grad_norm": 2.1954703330993652, + "learning_rate": 1.199787379037754e-05, + "loss": 2.508531951904297, + "step": 147280 + }, + { + "epoch": 0.4009799918334014, + "grad_norm": 2.1578941345214844, + "learning_rate": 1.1996783426468585e-05, + "loss": 2.499569892883301, + "step": 147300 + }, + { + "epoch": 0.4010344358241459, + "grad_norm": 2.5858137607574463, + "learning_rate": 1.199569306255963e-05, + "loss": 2.4722951889038085, + "step": 147320 + }, + { + "epoch": 0.40108887981489044, + "grad_norm": 2.571688413619995, + "learning_rate": 1.1994602698650675e-05, + "loss": 2.5149585723876955, + "step": 147340 + }, + { + "epoch": 0.40114332380563494, + "grad_norm": 2.2311818599700928, + "learning_rate": 1.1993512334741721e-05, + "loss": 2.5085710525512694, + "step": 147360 + }, + { + "epoch": 0.4011977677963795, + "grad_norm": 2.143810510635376, + "learning_rate": 1.1992421970832767e-05, + "loss": 2.4734630584716797, + "step": 147380 + }, + { + "epoch": 0.401252211787124, + "grad_norm": 2.2066519260406494, + "learning_rate": 1.1991331606923813e-05, + "loss": 2.513461112976074, + "step": 147400 + }, + { + "epoch": 0.4013066557778685, + "grad_norm": 2.1134510040283203, + "learning_rate": 1.1990241243014857e-05, + "loss": 2.4959503173828126, + "step": 147420 + }, + { + "epoch": 0.40136109976861306, + "grad_norm": 2.1726150512695312, + "learning_rate": 1.1989150879105903e-05, + "loss": 2.4994197845458985, + "step": 147440 + }, + { + "epoch": 0.40141554375935756, + "grad_norm": 2.790297031402588, + "learning_rate": 1.1988060515196948e-05, + "loss": 2.505229949951172, + "step": 147460 + }, + { + "epoch": 0.40146998775010206, + "grad_norm": 2.2008397579193115, + "learning_rate": 1.1986970151287992e-05, + "loss": 2.512795066833496, + "step": 147480 + }, + { + "epoch": 0.4015244317408466, + "grad_norm": 2.114633798599243, + "learning_rate": 1.1985879787379038e-05, + "loss": 2.4841522216796874, + "step": 147500 + }, + { + "epoch": 0.4015788757315911, + "grad_norm": 2.233675956726074, + "learning_rate": 1.1984789423470084e-05, + "loss": 2.495746040344238, + "step": 147520 + }, + { + "epoch": 0.4016333197223356, + "grad_norm": 2.1064319610595703, + "learning_rate": 1.198369905956113e-05, + "loss": 2.523575019836426, + "step": 147540 + }, + { + "epoch": 0.4016877637130802, + "grad_norm": 2.03930401802063, + "learning_rate": 1.1982608695652176e-05, + "loss": 2.507707977294922, + "step": 147560 + }, + { + "epoch": 0.4017422077038247, + "grad_norm": 2.132765531539917, + "learning_rate": 1.198151833174322e-05, + "loss": 2.5258419036865236, + "step": 147580 + }, + { + "epoch": 0.4017966516945692, + "grad_norm": 2.2053349018096924, + "learning_rate": 1.1980427967834266e-05, + "loss": 2.5134254455566407, + "step": 147600 + }, + { + "epoch": 0.40185109568531374, + "grad_norm": 2.2566077709198, + "learning_rate": 1.1979337603925311e-05, + "loss": 2.492550086975098, + "step": 147620 + }, + { + "epoch": 0.40190553967605824, + "grad_norm": 2.10915207862854, + "learning_rate": 1.1978247240016356e-05, + "loss": 2.4966104507446287, + "step": 147640 + }, + { + "epoch": 0.4019599836668028, + "grad_norm": 2.23691725730896, + "learning_rate": 1.1977156876107401e-05, + "loss": 2.477585029602051, + "step": 147660 + }, + { + "epoch": 0.4020144276575473, + "grad_norm": 2.295933246612549, + "learning_rate": 1.1976066512198447e-05, + "loss": 2.520421600341797, + "step": 147680 + }, + { + "epoch": 0.4020688716482918, + "grad_norm": 2.1362414360046387, + "learning_rate": 1.1974976148289493e-05, + "loss": 2.4939571380615235, + "step": 147700 + }, + { + "epoch": 0.40212331563903636, + "grad_norm": 2.1321334838867188, + "learning_rate": 1.1973885784380539e-05, + "loss": 2.47537841796875, + "step": 147720 + }, + { + "epoch": 0.40217775962978086, + "grad_norm": 2.194899559020996, + "learning_rate": 1.1972795420471583e-05, + "loss": 2.481622314453125, + "step": 147740 + }, + { + "epoch": 0.40223220362052536, + "grad_norm": 2.22615122795105, + "learning_rate": 1.1971705056562629e-05, + "loss": 2.524500846862793, + "step": 147760 + }, + { + "epoch": 0.4022866476112699, + "grad_norm": 2.2680797576904297, + "learning_rate": 1.1970614692653675e-05, + "loss": 2.4689495086669924, + "step": 147780 + }, + { + "epoch": 0.4023410916020144, + "grad_norm": 2.2039875984191895, + "learning_rate": 1.1969524328744719e-05, + "loss": 2.497943305969238, + "step": 147800 + }, + { + "epoch": 0.4023955355927589, + "grad_norm": 2.092790126800537, + "learning_rate": 1.1968433964835764e-05, + "loss": 2.5266290664672852, + "step": 147820 + }, + { + "epoch": 0.4024499795835035, + "grad_norm": 2.1150951385498047, + "learning_rate": 1.196734360092681e-05, + "loss": 2.4909526824951174, + "step": 147840 + }, + { + "epoch": 0.402504423574248, + "grad_norm": 2.092332363128662, + "learning_rate": 1.1966253237017856e-05, + "loss": 2.509419059753418, + "step": 147860 + }, + { + "epoch": 0.40255886756499254, + "grad_norm": 2.083380699157715, + "learning_rate": 1.1965162873108902e-05, + "loss": 2.4856433868408203, + "step": 147880 + }, + { + "epoch": 0.40261331155573704, + "grad_norm": 2.1630067825317383, + "learning_rate": 1.1964072509199946e-05, + "loss": 2.498367691040039, + "step": 147900 + }, + { + "epoch": 0.40266775554648154, + "grad_norm": 2.1611554622650146, + "learning_rate": 1.1962982145290992e-05, + "loss": 2.4636817932128907, + "step": 147920 + }, + { + "epoch": 0.4027221995372261, + "grad_norm": 2.124614953994751, + "learning_rate": 1.1961891781382038e-05, + "loss": 2.489299011230469, + "step": 147940 + }, + { + "epoch": 0.4027766435279706, + "grad_norm": 2.1569998264312744, + "learning_rate": 1.1960801417473082e-05, + "loss": 2.4919490814208984, + "step": 147960 + }, + { + "epoch": 0.4028310875187151, + "grad_norm": 2.010406494140625, + "learning_rate": 1.1959711053564128e-05, + "loss": 2.4921398162841797, + "step": 147980 + }, + { + "epoch": 0.40288553150945966, + "grad_norm": 2.0889999866485596, + "learning_rate": 1.1958620689655173e-05, + "loss": 2.511949920654297, + "step": 148000 + }, + { + "epoch": 0.40293997550020416, + "grad_norm": 2.0792269706726074, + "learning_rate": 1.1957530325746217e-05, + "loss": 2.50982551574707, + "step": 148020 + }, + { + "epoch": 0.40299441949094866, + "grad_norm": 2.278549909591675, + "learning_rate": 1.1956439961837265e-05, + "loss": 2.513262939453125, + "step": 148040 + }, + { + "epoch": 0.4030488634816932, + "grad_norm": 2.0957603454589844, + "learning_rate": 1.1955349597928309e-05, + "loss": 2.4867143630981445, + "step": 148060 + }, + { + "epoch": 0.4031033074724377, + "grad_norm": 2.0716404914855957, + "learning_rate": 1.1954259234019355e-05, + "loss": 2.497635269165039, + "step": 148080 + }, + { + "epoch": 0.4031577514631823, + "grad_norm": 2.0845987796783447, + "learning_rate": 1.19531688701104e-05, + "loss": 2.4814624786376953, + "step": 148100 + }, + { + "epoch": 0.4032121954539268, + "grad_norm": 2.195817470550537, + "learning_rate": 1.1952078506201445e-05, + "loss": 2.4955074310302736, + "step": 148120 + }, + { + "epoch": 0.4032666394446713, + "grad_norm": 2.066669225692749, + "learning_rate": 1.195098814229249e-05, + "loss": 2.466214179992676, + "step": 148140 + }, + { + "epoch": 0.40332108343541584, + "grad_norm": 2.207899332046509, + "learning_rate": 1.1949897778383536e-05, + "loss": 2.523600196838379, + "step": 148160 + }, + { + "epoch": 0.40337552742616034, + "grad_norm": 2.1307919025421143, + "learning_rate": 1.194880741447458e-05, + "loss": 2.4935121536254883, + "step": 148180 + }, + { + "epoch": 0.40342997141690484, + "grad_norm": 2.1036152839660645, + "learning_rate": 1.1947717050565628e-05, + "loss": 2.4925701141357424, + "step": 148200 + }, + { + "epoch": 0.4034844154076494, + "grad_norm": 2.0191354751586914, + "learning_rate": 1.1946626686656672e-05, + "loss": 2.4612796783447264, + "step": 148220 + }, + { + "epoch": 0.4035388593983939, + "grad_norm": 2.1477222442626953, + "learning_rate": 1.1945536322747718e-05, + "loss": 2.5059865951538085, + "step": 148240 + }, + { + "epoch": 0.4035933033891384, + "grad_norm": 2.1604859828948975, + "learning_rate": 1.1944445958838764e-05, + "loss": 2.501598358154297, + "step": 148260 + }, + { + "epoch": 0.40364774737988296, + "grad_norm": 2.0302844047546387, + "learning_rate": 1.1943355594929808e-05, + "loss": 2.500174140930176, + "step": 148280 + }, + { + "epoch": 0.40370219137062746, + "grad_norm": 2.112452983856201, + "learning_rate": 1.1942265231020854e-05, + "loss": 2.4925714492797852, + "step": 148300 + }, + { + "epoch": 0.40375663536137196, + "grad_norm": 2.2336513996124268, + "learning_rate": 1.19411748671119e-05, + "loss": 2.46822452545166, + "step": 148320 + }, + { + "epoch": 0.4038110793521165, + "grad_norm": 2.3704166412353516, + "learning_rate": 1.1940084503202944e-05, + "loss": 2.495968246459961, + "step": 148340 + }, + { + "epoch": 0.403865523342861, + "grad_norm": 2.1513710021972656, + "learning_rate": 1.1938994139293991e-05, + "loss": 2.5064006805419923, + "step": 148360 + }, + { + "epoch": 0.4039199673336056, + "grad_norm": 2.146073579788208, + "learning_rate": 1.1937903775385035e-05, + "loss": 2.496182441711426, + "step": 148380 + }, + { + "epoch": 0.4039744113243501, + "grad_norm": 2.1071484088897705, + "learning_rate": 1.1936813411476081e-05, + "loss": 2.4947364807128904, + "step": 148400 + }, + { + "epoch": 0.4040288553150946, + "grad_norm": 2.1205153465270996, + "learning_rate": 1.1935723047567127e-05, + "loss": 2.504198455810547, + "step": 148420 + }, + { + "epoch": 0.40408329930583914, + "grad_norm": 2.132354736328125, + "learning_rate": 1.1934632683658171e-05, + "loss": 2.471484565734863, + "step": 148440 + }, + { + "epoch": 0.40413774329658364, + "grad_norm": 2.2131519317626953, + "learning_rate": 1.1933542319749217e-05, + "loss": 2.4851198196411133, + "step": 148460 + }, + { + "epoch": 0.40419218728732814, + "grad_norm": 2.1192662715911865, + "learning_rate": 1.1932451955840263e-05, + "loss": 2.4971866607666016, + "step": 148480 + }, + { + "epoch": 0.4042466312780727, + "grad_norm": 2.233372926712036, + "learning_rate": 1.1931361591931307e-05, + "loss": 2.4973997116088866, + "step": 148500 + }, + { + "epoch": 0.4043010752688172, + "grad_norm": 2.245269775390625, + "learning_rate": 1.1930271228022354e-05, + "loss": 2.461055374145508, + "step": 148520 + }, + { + "epoch": 0.4043555192595617, + "grad_norm": 2.24235463142395, + "learning_rate": 1.1929180864113398e-05, + "loss": 2.4973621368408203, + "step": 148540 + }, + { + "epoch": 0.40440996325030626, + "grad_norm": 2.098529577255249, + "learning_rate": 1.1928090500204444e-05, + "loss": 2.496041488647461, + "step": 148560 + }, + { + "epoch": 0.40446440724105076, + "grad_norm": 2.0618338584899902, + "learning_rate": 1.192700013629549e-05, + "loss": 2.49925537109375, + "step": 148580 + }, + { + "epoch": 0.4045188512317953, + "grad_norm": 2.078673839569092, + "learning_rate": 1.1925909772386534e-05, + "loss": 2.4884222030639647, + "step": 148600 + }, + { + "epoch": 0.4045732952225398, + "grad_norm": 2.024505853652954, + "learning_rate": 1.1924819408477582e-05, + "loss": 2.5012432098388673, + "step": 148620 + }, + { + "epoch": 0.4046277392132843, + "grad_norm": 2.2439770698547363, + "learning_rate": 1.1923729044568626e-05, + "loss": 2.4742948532104494, + "step": 148640 + }, + { + "epoch": 0.4046821832040289, + "grad_norm": 2.217888355255127, + "learning_rate": 1.192263868065967e-05, + "loss": 2.4893299102783204, + "step": 148660 + }, + { + "epoch": 0.4047366271947734, + "grad_norm": 2.128157377243042, + "learning_rate": 1.1921548316750717e-05, + "loss": 2.5124866485595705, + "step": 148680 + }, + { + "epoch": 0.4047910711855179, + "grad_norm": 2.0762147903442383, + "learning_rate": 1.1920457952841761e-05, + "loss": 2.5053180694580077, + "step": 148700 + }, + { + "epoch": 0.40484551517626244, + "grad_norm": 2.0254976749420166, + "learning_rate": 1.1919367588932807e-05, + "loss": 2.4734123229980467, + "step": 148720 + }, + { + "epoch": 0.40489995916700694, + "grad_norm": 2.232856273651123, + "learning_rate": 1.1918277225023853e-05, + "loss": 2.5064634323120116, + "step": 148740 + }, + { + "epoch": 0.40495440315775144, + "grad_norm": 2.2278106212615967, + "learning_rate": 1.1917186861114897e-05, + "loss": 2.49630069732666, + "step": 148760 + }, + { + "epoch": 0.405008847148496, + "grad_norm": 2.0475046634674072, + "learning_rate": 1.1916096497205945e-05, + "loss": 2.5026628494262697, + "step": 148780 + }, + { + "epoch": 0.4050632911392405, + "grad_norm": 2.2180986404418945, + "learning_rate": 1.1915006133296989e-05, + "loss": 2.486777114868164, + "step": 148800 + }, + { + "epoch": 0.405117735129985, + "grad_norm": 2.2251663208007812, + "learning_rate": 1.1913915769388033e-05, + "loss": 2.481431007385254, + "step": 148820 + }, + { + "epoch": 0.40517217912072956, + "grad_norm": 2.139660358428955, + "learning_rate": 1.191282540547908e-05, + "loss": 2.4928823471069337, + "step": 148840 + }, + { + "epoch": 0.40522662311147406, + "grad_norm": 2.0741236209869385, + "learning_rate": 1.1911735041570125e-05, + "loss": 2.5086074829101563, + "step": 148860 + }, + { + "epoch": 0.4052810671022186, + "grad_norm": 2.072805643081665, + "learning_rate": 1.191064467766117e-05, + "loss": 2.4938753128051756, + "step": 148880 + }, + { + "epoch": 0.4053355110929631, + "grad_norm": 2.153092861175537, + "learning_rate": 1.1909554313752216e-05, + "loss": 2.4743452072143555, + "step": 148900 + }, + { + "epoch": 0.4053899550837076, + "grad_norm": 2.190823554992676, + "learning_rate": 1.190846394984326e-05, + "loss": 2.4969581604003905, + "step": 148920 + }, + { + "epoch": 0.4054443990744522, + "grad_norm": 2.390486478805542, + "learning_rate": 1.1907373585934308e-05, + "loss": 2.483320426940918, + "step": 148940 + }, + { + "epoch": 0.4054988430651967, + "grad_norm": 2.1867432594299316, + "learning_rate": 1.1906283222025352e-05, + "loss": 2.4866289138793944, + "step": 148960 + }, + { + "epoch": 0.4055532870559412, + "grad_norm": 2.084864616394043, + "learning_rate": 1.1905192858116396e-05, + "loss": 2.4968944549560548, + "step": 148980 + }, + { + "epoch": 0.40560773104668574, + "grad_norm": 2.1704294681549072, + "learning_rate": 1.1904102494207443e-05, + "loss": 2.503229331970215, + "step": 149000 + }, + { + "epoch": 0.40566217503743024, + "grad_norm": 2.1137304306030273, + "learning_rate": 1.1903012130298488e-05, + "loss": 2.505674362182617, + "step": 149020 + }, + { + "epoch": 0.40571661902817474, + "grad_norm": 2.840080499649048, + "learning_rate": 1.1901921766389533e-05, + "loss": 2.4963560104370117, + "step": 149040 + }, + { + "epoch": 0.4057710630189193, + "grad_norm": 2.2445170879364014, + "learning_rate": 1.190083140248058e-05, + "loss": 2.4958391189575195, + "step": 149060 + }, + { + "epoch": 0.4058255070096638, + "grad_norm": 2.307274580001831, + "learning_rate": 1.1899741038571623e-05, + "loss": 2.5036270141601564, + "step": 149080 + }, + { + "epoch": 0.40587995100040836, + "grad_norm": 2.217041492462158, + "learning_rate": 1.189865067466267e-05, + "loss": 2.4961929321289062, + "step": 149100 + }, + { + "epoch": 0.40593439499115286, + "grad_norm": 2.1454787254333496, + "learning_rate": 1.1897560310753715e-05, + "loss": 2.459568977355957, + "step": 149120 + }, + { + "epoch": 0.40598883898189736, + "grad_norm": 2.3253121376037598, + "learning_rate": 1.1896469946844759e-05, + "loss": 2.4979789733886717, + "step": 149140 + }, + { + "epoch": 0.4060432829726419, + "grad_norm": 2.229702949523926, + "learning_rate": 1.1895379582935807e-05, + "loss": 2.4958892822265626, + "step": 149160 + }, + { + "epoch": 0.4060977269633864, + "grad_norm": 2.1249959468841553, + "learning_rate": 1.189428921902685e-05, + "loss": 2.4708126068115233, + "step": 149180 + }, + { + "epoch": 0.4061521709541309, + "grad_norm": 2.2962119579315186, + "learning_rate": 1.1893198855117895e-05, + "loss": 2.5014602661132814, + "step": 149200 + }, + { + "epoch": 0.4062066149448755, + "grad_norm": 2.087458372116089, + "learning_rate": 1.1892108491208942e-05, + "loss": 2.478141975402832, + "step": 149220 + }, + { + "epoch": 0.40626105893562, + "grad_norm": 2.2081551551818848, + "learning_rate": 1.1891018127299986e-05, + "loss": 2.4984073638916016, + "step": 149240 + }, + { + "epoch": 0.4063155029263645, + "grad_norm": 2.121151924133301, + "learning_rate": 1.1889927763391034e-05, + "loss": 2.483901786804199, + "step": 149260 + }, + { + "epoch": 0.40636994691710904, + "grad_norm": 2.2071828842163086, + "learning_rate": 1.1888837399482078e-05, + "loss": 2.491510772705078, + "step": 149280 + }, + { + "epoch": 0.40642439090785354, + "grad_norm": 2.2388060092926025, + "learning_rate": 1.1887747035573122e-05, + "loss": 2.5064165115356447, + "step": 149300 + }, + { + "epoch": 0.40647883489859804, + "grad_norm": 2.135239362716675, + "learning_rate": 1.188665667166417e-05, + "loss": 2.527817153930664, + "step": 149320 + }, + { + "epoch": 0.4065332788893426, + "grad_norm": 2.30683970451355, + "learning_rate": 1.1885566307755214e-05, + "loss": 2.487464714050293, + "step": 149340 + }, + { + "epoch": 0.4065877228800871, + "grad_norm": 2.1433801651000977, + "learning_rate": 1.1884475943846258e-05, + "loss": 2.4820934295654298, + "step": 149360 + }, + { + "epoch": 0.40664216687083166, + "grad_norm": 2.089348077774048, + "learning_rate": 1.1883385579937305e-05, + "loss": 2.4881494522094725, + "step": 149380 + }, + { + "epoch": 0.40669661086157616, + "grad_norm": 2.047738552093506, + "learning_rate": 1.188229521602835e-05, + "loss": 2.482357406616211, + "step": 149400 + }, + { + "epoch": 0.40675105485232066, + "grad_norm": 2.280886650085449, + "learning_rate": 1.1881204852119397e-05, + "loss": 2.5138113021850588, + "step": 149420 + }, + { + "epoch": 0.4068054988430652, + "grad_norm": 2.082559585571289, + "learning_rate": 1.1880114488210441e-05, + "loss": 2.4965112686157225, + "step": 149440 + }, + { + "epoch": 0.4068599428338097, + "grad_norm": 2.1267282962799072, + "learning_rate": 1.1879024124301485e-05, + "loss": 2.505130386352539, + "step": 149460 + }, + { + "epoch": 0.4069143868245542, + "grad_norm": 2.0871543884277344, + "learning_rate": 1.1877933760392533e-05, + "loss": 2.471793365478516, + "step": 149480 + }, + { + "epoch": 0.4069688308152988, + "grad_norm": 2.5111074447631836, + "learning_rate": 1.1876843396483577e-05, + "loss": 2.49029426574707, + "step": 149500 + }, + { + "epoch": 0.4070232748060433, + "grad_norm": 2.1400084495544434, + "learning_rate": 1.1875753032574621e-05, + "loss": 2.4748886108398436, + "step": 149520 + }, + { + "epoch": 0.4070777187967878, + "grad_norm": 2.294743061065674, + "learning_rate": 1.1874662668665668e-05, + "loss": 2.497021293640137, + "step": 149540 + }, + { + "epoch": 0.40713216278753234, + "grad_norm": 2.160478353500366, + "learning_rate": 1.1873572304756713e-05, + "loss": 2.5233224868774413, + "step": 149560 + }, + { + "epoch": 0.40718660677827684, + "grad_norm": 2.196772813796997, + "learning_rate": 1.187248194084776e-05, + "loss": 2.483871841430664, + "step": 149580 + }, + { + "epoch": 0.4072410507690214, + "grad_norm": 2.122824192047119, + "learning_rate": 1.1871391576938804e-05, + "loss": 2.517470359802246, + "step": 149600 + }, + { + "epoch": 0.4072954947597659, + "grad_norm": 2.373896598815918, + "learning_rate": 1.1870301213029848e-05, + "loss": 2.511700439453125, + "step": 149620 + }, + { + "epoch": 0.4073499387505104, + "grad_norm": 2.1237571239471436, + "learning_rate": 1.1869210849120896e-05, + "loss": 2.489198112487793, + "step": 149640 + }, + { + "epoch": 0.40740438274125496, + "grad_norm": 2.14836049079895, + "learning_rate": 1.186812048521194e-05, + "loss": 2.51599178314209, + "step": 149660 + }, + { + "epoch": 0.40745882673199946, + "grad_norm": 2.221803903579712, + "learning_rate": 1.1867030121302984e-05, + "loss": 2.516558074951172, + "step": 149680 + }, + { + "epoch": 0.40751327072274396, + "grad_norm": 2.2384321689605713, + "learning_rate": 1.1865939757394032e-05, + "loss": 2.4898420333862306, + "step": 149700 + }, + { + "epoch": 0.4075677147134885, + "grad_norm": 2.1164472103118896, + "learning_rate": 1.1864849393485076e-05, + "loss": 2.4903215408325194, + "step": 149720 + }, + { + "epoch": 0.407622158704233, + "grad_norm": 2.249311685562134, + "learning_rate": 1.1863759029576123e-05, + "loss": 2.4725914001464844, + "step": 149740 + }, + { + "epoch": 0.4076766026949775, + "grad_norm": 4.685268402099609, + "learning_rate": 1.1862668665667167e-05, + "loss": 2.4953914642333985, + "step": 149760 + }, + { + "epoch": 0.4077310466857221, + "grad_norm": 2.0256803035736084, + "learning_rate": 1.1861578301758211e-05, + "loss": 2.500440788269043, + "step": 149780 + }, + { + "epoch": 0.4077854906764666, + "grad_norm": 2.2508246898651123, + "learning_rate": 1.1860487937849259e-05, + "loss": 2.5121288299560547, + "step": 149800 + }, + { + "epoch": 0.4078399346672111, + "grad_norm": 2.084827184677124, + "learning_rate": 1.1859397573940303e-05, + "loss": 2.5239273071289063, + "step": 149820 + }, + { + "epoch": 0.40789437865795564, + "grad_norm": 2.153506278991699, + "learning_rate": 1.1858307210031347e-05, + "loss": 2.493457794189453, + "step": 149840 + }, + { + "epoch": 0.40794882264870014, + "grad_norm": 2.1667497158050537, + "learning_rate": 1.1857216846122395e-05, + "loss": 2.5054698944091798, + "step": 149860 + }, + { + "epoch": 0.4080032666394447, + "grad_norm": 2.2543344497680664, + "learning_rate": 1.1856126482213439e-05, + "loss": 2.4813051223754883, + "step": 149880 + }, + { + "epoch": 0.4080577106301892, + "grad_norm": 2.2107324600219727, + "learning_rate": 1.1855036118304486e-05, + "loss": 2.488038444519043, + "step": 149900 + }, + { + "epoch": 0.4081121546209337, + "grad_norm": 2.1594295501708984, + "learning_rate": 1.185394575439553e-05, + "loss": 2.5105003356933593, + "step": 149920 + }, + { + "epoch": 0.40816659861167826, + "grad_norm": 2.2957260608673096, + "learning_rate": 1.1852855390486574e-05, + "loss": 2.5203889846801757, + "step": 149940 + }, + { + "epoch": 0.40822104260242276, + "grad_norm": 2.2518715858459473, + "learning_rate": 1.1851765026577622e-05, + "loss": 2.50552978515625, + "step": 149960 + }, + { + "epoch": 0.40827548659316726, + "grad_norm": 2.1419413089752197, + "learning_rate": 1.1850674662668666e-05, + "loss": 2.489211082458496, + "step": 149980 + }, + { + "epoch": 0.4083299305839118, + "grad_norm": 2.2432219982147217, + "learning_rate": 1.184958429875971e-05, + "loss": 2.5013694763183594, + "step": 150000 + }, + { + "epoch": 0.4083843745746563, + "grad_norm": 2.1506645679473877, + "learning_rate": 1.1848493934850758e-05, + "loss": 2.4980228424072264, + "step": 150020 + }, + { + "epoch": 0.4084388185654008, + "grad_norm": 2.142571449279785, + "learning_rate": 1.1847403570941802e-05, + "loss": 2.496059608459473, + "step": 150040 + }, + { + "epoch": 0.4084932625561454, + "grad_norm": 2.1338844299316406, + "learning_rate": 1.184631320703285e-05, + "loss": 2.491474151611328, + "step": 150060 + }, + { + "epoch": 0.4085477065468899, + "grad_norm": 2.1666343212127686, + "learning_rate": 1.1845222843123893e-05, + "loss": 2.5254329681396483, + "step": 150080 + }, + { + "epoch": 0.40860215053763443, + "grad_norm": 2.040761947631836, + "learning_rate": 1.1844132479214938e-05, + "loss": 2.500520706176758, + "step": 150100 + }, + { + "epoch": 0.40865659452837894, + "grad_norm": 2.1883766651153564, + "learning_rate": 1.1843042115305985e-05, + "loss": 2.4989553451538087, + "step": 150120 + }, + { + "epoch": 0.40871103851912344, + "grad_norm": 2.0341947078704834, + "learning_rate": 1.184195175139703e-05, + "loss": 2.5251110076904295, + "step": 150140 + }, + { + "epoch": 0.408765482509868, + "grad_norm": 2.142642021179199, + "learning_rate": 1.1840861387488073e-05, + "loss": 2.4858797073364256, + "step": 150160 + }, + { + "epoch": 0.4088199265006125, + "grad_norm": 2.1749088764190674, + "learning_rate": 1.183977102357912e-05, + "loss": 2.4892688751220704, + "step": 150180 + }, + { + "epoch": 0.408874370491357, + "grad_norm": 2.188934326171875, + "learning_rate": 1.1838680659670165e-05, + "loss": 2.5060941696166994, + "step": 150200 + }, + { + "epoch": 0.40892881448210155, + "grad_norm": 2.1214699745178223, + "learning_rate": 1.1837590295761212e-05, + "loss": 2.4894901275634767, + "step": 150220 + }, + { + "epoch": 0.40898325847284606, + "grad_norm": 2.13197660446167, + "learning_rate": 1.1836499931852257e-05, + "loss": 2.478862190246582, + "step": 150240 + }, + { + "epoch": 0.40903770246359056, + "grad_norm": 2.1750950813293457, + "learning_rate": 1.18354095679433e-05, + "loss": 2.48553352355957, + "step": 150260 + }, + { + "epoch": 0.4090921464543351, + "grad_norm": 2.2272088527679443, + "learning_rate": 1.1834319204034348e-05, + "loss": 2.5085128784179687, + "step": 150280 + }, + { + "epoch": 0.4091465904450796, + "grad_norm": 2.5506999492645264, + "learning_rate": 1.1833228840125392e-05, + "loss": 2.5017932891845702, + "step": 150300 + }, + { + "epoch": 0.4092010344358242, + "grad_norm": 2.0620248317718506, + "learning_rate": 1.183213847621644e-05, + "loss": 2.4912038803100587, + "step": 150320 + }, + { + "epoch": 0.4092554784265687, + "grad_norm": 2.182295799255371, + "learning_rate": 1.1831048112307484e-05, + "loss": 2.4767200469970705, + "step": 150340 + }, + { + "epoch": 0.4093099224173132, + "grad_norm": 2.1887564659118652, + "learning_rate": 1.1829957748398528e-05, + "loss": 2.489859962463379, + "step": 150360 + }, + { + "epoch": 0.40936436640805773, + "grad_norm": 2.2492477893829346, + "learning_rate": 1.1828867384489576e-05, + "loss": 2.469379425048828, + "step": 150380 + }, + { + "epoch": 0.40941881039880224, + "grad_norm": 2.1119818687438965, + "learning_rate": 1.182777702058062e-05, + "loss": 2.4890506744384764, + "step": 150400 + }, + { + "epoch": 0.40947325438954674, + "grad_norm": 2.2311761379241943, + "learning_rate": 1.1826686656671664e-05, + "loss": 2.478832817077637, + "step": 150420 + }, + { + "epoch": 0.4095276983802913, + "grad_norm": 2.1430821418762207, + "learning_rate": 1.1825596292762711e-05, + "loss": 2.488016128540039, + "step": 150440 + }, + { + "epoch": 0.4095821423710358, + "grad_norm": 2.079524517059326, + "learning_rate": 1.1824505928853755e-05, + "loss": 2.497275161743164, + "step": 150460 + }, + { + "epoch": 0.4096365863617803, + "grad_norm": 2.21376371383667, + "learning_rate": 1.1823415564944803e-05, + "loss": 2.506847953796387, + "step": 150480 + }, + { + "epoch": 0.40969103035252485, + "grad_norm": 2.13061261177063, + "learning_rate": 1.1822325201035847e-05, + "loss": 2.525925064086914, + "step": 150500 + }, + { + "epoch": 0.40974547434326936, + "grad_norm": 2.1550939083099365, + "learning_rate": 1.1821234837126891e-05, + "loss": 2.487702560424805, + "step": 150520 + }, + { + "epoch": 0.40979991833401386, + "grad_norm": 2.058016300201416, + "learning_rate": 1.1820144473217939e-05, + "loss": 2.4768291473388673, + "step": 150540 + }, + { + "epoch": 0.4098543623247584, + "grad_norm": 2.151353359222412, + "learning_rate": 1.1819054109308983e-05, + "loss": 2.47572021484375, + "step": 150560 + }, + { + "epoch": 0.4099088063155029, + "grad_norm": 2.0778040885925293, + "learning_rate": 1.1817963745400027e-05, + "loss": 2.4855812072753904, + "step": 150580 + }, + { + "epoch": 0.4099632503062475, + "grad_norm": 2.21309494972229, + "learning_rate": 1.1816873381491074e-05, + "loss": 2.5029096603393555, + "step": 150600 + }, + { + "epoch": 0.410017694296992, + "grad_norm": 2.2121691703796387, + "learning_rate": 1.1815783017582118e-05, + "loss": 2.498426628112793, + "step": 150620 + }, + { + "epoch": 0.4100721382877365, + "grad_norm": 2.177528142929077, + "learning_rate": 1.1814692653673166e-05, + "loss": 2.5018964767456056, + "step": 150640 + }, + { + "epoch": 0.41012658227848103, + "grad_norm": 2.0969960689544678, + "learning_rate": 1.181360228976421e-05, + "loss": 2.479370880126953, + "step": 150660 + }, + { + "epoch": 0.41018102626922553, + "grad_norm": 2.2059600353240967, + "learning_rate": 1.1812511925855254e-05, + "loss": 2.4814258575439454, + "step": 150680 + }, + { + "epoch": 0.41023547025997004, + "grad_norm": 2.215879440307617, + "learning_rate": 1.1811421561946302e-05, + "loss": 2.5145795822143553, + "step": 150700 + }, + { + "epoch": 0.4102899142507146, + "grad_norm": 2.1273837089538574, + "learning_rate": 1.1810331198037346e-05, + "loss": 2.5032623291015623, + "step": 150720 + }, + { + "epoch": 0.4103443582414591, + "grad_norm": 2.1916615962982178, + "learning_rate": 1.180924083412839e-05, + "loss": 2.5167139053344725, + "step": 150740 + }, + { + "epoch": 0.4103988022322036, + "grad_norm": 2.095510721206665, + "learning_rate": 1.1808150470219437e-05, + "loss": 2.5203474044799803, + "step": 150760 + }, + { + "epoch": 0.41045324622294815, + "grad_norm": 2.1704084873199463, + "learning_rate": 1.1807060106310482e-05, + "loss": 2.493103790283203, + "step": 150780 + }, + { + "epoch": 0.41050769021369266, + "grad_norm": 2.138988494873047, + "learning_rate": 1.1805969742401529e-05, + "loss": 2.4896781921386717, + "step": 150800 + }, + { + "epoch": 0.4105621342044372, + "grad_norm": 2.0614843368530273, + "learning_rate": 1.1804879378492573e-05, + "loss": 2.4608179092407227, + "step": 150820 + }, + { + "epoch": 0.4106165781951817, + "grad_norm": 2.2196407318115234, + "learning_rate": 1.1803789014583617e-05, + "loss": 2.482716941833496, + "step": 150840 + }, + { + "epoch": 0.4106710221859262, + "grad_norm": 2.1836493015289307, + "learning_rate": 1.1802698650674665e-05, + "loss": 2.465943717956543, + "step": 150860 + }, + { + "epoch": 0.4107254661766708, + "grad_norm": 2.121070384979248, + "learning_rate": 1.1801608286765709e-05, + "loss": 2.486100959777832, + "step": 150880 + }, + { + "epoch": 0.4107799101674153, + "grad_norm": 2.1474485397338867, + "learning_rate": 1.1800517922856753e-05, + "loss": 2.500361442565918, + "step": 150900 + }, + { + "epoch": 0.4108343541581598, + "grad_norm": 2.152963876724243, + "learning_rate": 1.17994275589478e-05, + "loss": 2.4602060317993164, + "step": 150920 + }, + { + "epoch": 0.41088879814890433, + "grad_norm": 2.1420087814331055, + "learning_rate": 1.1798337195038845e-05, + "loss": 2.508449172973633, + "step": 150940 + }, + { + "epoch": 0.41094324213964883, + "grad_norm": 2.2197790145874023, + "learning_rate": 1.1797246831129892e-05, + "loss": 2.4794332504272463, + "step": 150960 + }, + { + "epoch": 0.41099768613039334, + "grad_norm": 2.170975685119629, + "learning_rate": 1.1796156467220936e-05, + "loss": 2.5114349365234374, + "step": 150980 + }, + { + "epoch": 0.4110521301211379, + "grad_norm": 2.006072759628296, + "learning_rate": 1.179506610331198e-05, + "loss": 2.4790054321289063, + "step": 151000 + }, + { + "epoch": 0.4111065741118824, + "grad_norm": 2.2103958129882812, + "learning_rate": 1.1793975739403028e-05, + "loss": 2.499155616760254, + "step": 151020 + }, + { + "epoch": 0.4111610181026269, + "grad_norm": 2.1420116424560547, + "learning_rate": 1.1792885375494072e-05, + "loss": 2.486545753479004, + "step": 151040 + }, + { + "epoch": 0.41121546209337145, + "grad_norm": 2.0958762168884277, + "learning_rate": 1.1791795011585116e-05, + "loss": 2.463441848754883, + "step": 151060 + }, + { + "epoch": 0.41126990608411595, + "grad_norm": 2.0429461002349854, + "learning_rate": 1.1790704647676164e-05, + "loss": 2.5071605682373046, + "step": 151080 + }, + { + "epoch": 0.4113243500748605, + "grad_norm": 2.675778865814209, + "learning_rate": 1.1789614283767208e-05, + "loss": 2.50094051361084, + "step": 151100 + }, + { + "epoch": 0.411378794065605, + "grad_norm": 2.0561752319335938, + "learning_rate": 1.1788523919858255e-05, + "loss": 2.4782684326171873, + "step": 151120 + }, + { + "epoch": 0.4114332380563495, + "grad_norm": 2.6143858432769775, + "learning_rate": 1.17874335559493e-05, + "loss": 2.5063304901123047, + "step": 151140 + }, + { + "epoch": 0.41148768204709407, + "grad_norm": 2.130770206451416, + "learning_rate": 1.1786343192040343e-05, + "loss": 2.4813087463378904, + "step": 151160 + }, + { + "epoch": 0.4115421260378386, + "grad_norm": 2.2264416217803955, + "learning_rate": 1.1785252828131391e-05, + "loss": 2.487537956237793, + "step": 151180 + }, + { + "epoch": 0.4115965700285831, + "grad_norm": 2.0724029541015625, + "learning_rate": 1.1784162464222435e-05, + "loss": 2.4984588623046875, + "step": 151200 + }, + { + "epoch": 0.41165101401932763, + "grad_norm": 2.150385856628418, + "learning_rate": 1.178307210031348e-05, + "loss": 2.483225440979004, + "step": 151220 + }, + { + "epoch": 0.41170545801007213, + "grad_norm": 2.1976187229156494, + "learning_rate": 1.1781981736404527e-05, + "loss": 2.4831329345703126, + "step": 151240 + }, + { + "epoch": 0.41175990200081664, + "grad_norm": 2.23612904548645, + "learning_rate": 1.178089137249557e-05, + "loss": 2.48953800201416, + "step": 151260 + }, + { + "epoch": 0.4118143459915612, + "grad_norm": 2.228867530822754, + "learning_rate": 1.1779801008586618e-05, + "loss": 2.506752777099609, + "step": 151280 + }, + { + "epoch": 0.4118687899823057, + "grad_norm": 2.187387228012085, + "learning_rate": 1.1778710644677662e-05, + "loss": 2.483225631713867, + "step": 151300 + }, + { + "epoch": 0.41192323397305025, + "grad_norm": 2.0865490436553955, + "learning_rate": 1.1777620280768707e-05, + "loss": 2.4949296951293944, + "step": 151320 + }, + { + "epoch": 0.41197767796379475, + "grad_norm": 2.133751630783081, + "learning_rate": 1.1776529916859754e-05, + "loss": 2.4972728729248046, + "step": 151340 + }, + { + "epoch": 0.41203212195453925, + "grad_norm": 2.166567087173462, + "learning_rate": 1.1775439552950798e-05, + "loss": 2.5091251373291015, + "step": 151360 + }, + { + "epoch": 0.4120865659452838, + "grad_norm": 2.2956809997558594, + "learning_rate": 1.1774349189041842e-05, + "loss": 2.501433181762695, + "step": 151380 + }, + { + "epoch": 0.4121410099360283, + "grad_norm": 2.0843710899353027, + "learning_rate": 1.177325882513289e-05, + "loss": 2.494776153564453, + "step": 151400 + }, + { + "epoch": 0.4121954539267728, + "grad_norm": 2.2538387775421143, + "learning_rate": 1.1772168461223934e-05, + "loss": 2.4822689056396485, + "step": 151420 + }, + { + "epoch": 0.41224989791751737, + "grad_norm": 2.218397378921509, + "learning_rate": 1.1771078097314981e-05, + "loss": 2.517545700073242, + "step": 151440 + }, + { + "epoch": 0.4123043419082619, + "grad_norm": 2.238651990890503, + "learning_rate": 1.1769987733406025e-05, + "loss": 2.44732551574707, + "step": 151460 + }, + { + "epoch": 0.4123587858990064, + "grad_norm": 2.0913214683532715, + "learning_rate": 1.176889736949707e-05, + "loss": 2.490170478820801, + "step": 151480 + }, + { + "epoch": 0.41241322988975093, + "grad_norm": 2.126631736755371, + "learning_rate": 1.1767807005588117e-05, + "loss": 2.4948644638061523, + "step": 151500 + }, + { + "epoch": 0.41246767388049543, + "grad_norm": 2.193870782852173, + "learning_rate": 1.1766716641679161e-05, + "loss": 2.4867271423339843, + "step": 151520 + }, + { + "epoch": 0.41252211787123994, + "grad_norm": 2.242382049560547, + "learning_rate": 1.1765626277770205e-05, + "loss": 2.494994354248047, + "step": 151540 + }, + { + "epoch": 0.4125765618619845, + "grad_norm": 2.142036199569702, + "learning_rate": 1.1764535913861253e-05, + "loss": 2.494972801208496, + "step": 151560 + }, + { + "epoch": 0.412631005852729, + "grad_norm": 2.0807669162750244, + "learning_rate": 1.1763445549952297e-05, + "loss": 2.4973867416381834, + "step": 151580 + }, + { + "epoch": 0.41268544984347355, + "grad_norm": 2.2451980113983154, + "learning_rate": 1.1762355186043344e-05, + "loss": 2.4874187469482423, + "step": 151600 + }, + { + "epoch": 0.41273989383421805, + "grad_norm": 2.1060640811920166, + "learning_rate": 1.1761264822134389e-05, + "loss": 2.5066518783569336, + "step": 151620 + }, + { + "epoch": 0.41279433782496255, + "grad_norm": 2.1646525859832764, + "learning_rate": 1.1760174458225433e-05, + "loss": 2.4755659103393555, + "step": 151640 + }, + { + "epoch": 0.4128487818157071, + "grad_norm": 2.0648176670074463, + "learning_rate": 1.175908409431648e-05, + "loss": 2.5035852432250976, + "step": 151660 + }, + { + "epoch": 0.4129032258064516, + "grad_norm": 2.0650172233581543, + "learning_rate": 1.1757993730407524e-05, + "loss": 2.508430290222168, + "step": 151680 + }, + { + "epoch": 0.4129576697971961, + "grad_norm": 2.185987710952759, + "learning_rate": 1.1756903366498568e-05, + "loss": 2.4964225769042967, + "step": 151700 + }, + { + "epoch": 0.41301211378794067, + "grad_norm": 2.096982717514038, + "learning_rate": 1.1755813002589616e-05, + "loss": 2.4744247436523437, + "step": 151720 + }, + { + "epoch": 0.4130665577786852, + "grad_norm": 2.628889322280884, + "learning_rate": 1.175472263868066e-05, + "loss": 2.506752777099609, + "step": 151740 + }, + { + "epoch": 0.4131210017694297, + "grad_norm": 2.203460216522217, + "learning_rate": 1.1753632274771708e-05, + "loss": 2.511275863647461, + "step": 151760 + }, + { + "epoch": 0.41317544576017423, + "grad_norm": 2.157529354095459, + "learning_rate": 1.1752541910862752e-05, + "loss": 2.466481018066406, + "step": 151780 + }, + { + "epoch": 0.41322988975091873, + "grad_norm": 2.2116916179656982, + "learning_rate": 1.1751451546953796e-05, + "loss": 2.508243179321289, + "step": 151800 + }, + { + "epoch": 0.4132843337416633, + "grad_norm": 2.1452324390411377, + "learning_rate": 1.1750361183044843e-05, + "loss": 2.5055587768554686, + "step": 151820 + }, + { + "epoch": 0.4133387777324078, + "grad_norm": 2.246870994567871, + "learning_rate": 1.1749270819135887e-05, + "loss": 2.490144157409668, + "step": 151840 + }, + { + "epoch": 0.4133932217231523, + "grad_norm": 2.3786566257476807, + "learning_rate": 1.1748180455226932e-05, + "loss": 2.4933904647827148, + "step": 151860 + }, + { + "epoch": 0.41344766571389685, + "grad_norm": 2.2025680541992188, + "learning_rate": 1.1747090091317979e-05, + "loss": 2.4765962600708007, + "step": 151880 + }, + { + "epoch": 0.41350210970464135, + "grad_norm": 2.122021198272705, + "learning_rate": 1.1745999727409023e-05, + "loss": 2.5088266372680663, + "step": 151900 + }, + { + "epoch": 0.41355655369538585, + "grad_norm": 2.2412149906158447, + "learning_rate": 1.174490936350007e-05, + "loss": 2.5085220336914062, + "step": 151920 + }, + { + "epoch": 0.4136109976861304, + "grad_norm": 2.2024614810943604, + "learning_rate": 1.1743818999591115e-05, + "loss": 2.4844865798950195, + "step": 151940 + }, + { + "epoch": 0.4136654416768749, + "grad_norm": 2.0929067134857178, + "learning_rate": 1.1742728635682159e-05, + "loss": 2.467184829711914, + "step": 151960 + }, + { + "epoch": 0.4137198856676194, + "grad_norm": 2.1956799030303955, + "learning_rate": 1.1741638271773206e-05, + "loss": 2.4883277893066404, + "step": 151980 + }, + { + "epoch": 0.41377432965836397, + "grad_norm": 2.289928436279297, + "learning_rate": 1.174054790786425e-05, + "loss": 2.498947334289551, + "step": 152000 + }, + { + "epoch": 0.4138287736491085, + "grad_norm": 2.056549072265625, + "learning_rate": 1.1739457543955296e-05, + "loss": 2.4935855865478516, + "step": 152020 + }, + { + "epoch": 0.413883217639853, + "grad_norm": 2.149329423904419, + "learning_rate": 1.1738367180046342e-05, + "loss": 2.5131778717041016, + "step": 152040 + }, + { + "epoch": 0.41393766163059753, + "grad_norm": 2.2599878311157227, + "learning_rate": 1.1737276816137386e-05, + "loss": 2.4780532836914064, + "step": 152060 + }, + { + "epoch": 0.41399210562134203, + "grad_norm": 2.0127055644989014, + "learning_rate": 1.1736186452228434e-05, + "loss": 2.479777526855469, + "step": 152080 + }, + { + "epoch": 0.4140465496120866, + "grad_norm": 2.083742141723633, + "learning_rate": 1.1735096088319478e-05, + "loss": 2.5142499923706056, + "step": 152100 + }, + { + "epoch": 0.4141009936028311, + "grad_norm": 2.248208999633789, + "learning_rate": 1.1734005724410522e-05, + "loss": 2.485939598083496, + "step": 152120 + }, + { + "epoch": 0.4141554375935756, + "grad_norm": 2.232851982116699, + "learning_rate": 1.173291536050157e-05, + "loss": 2.5148284912109373, + "step": 152140 + }, + { + "epoch": 0.41420988158432015, + "grad_norm": 2.173609972000122, + "learning_rate": 1.1731824996592614e-05, + "loss": 2.5141389846801756, + "step": 152160 + }, + { + "epoch": 0.41426432557506465, + "grad_norm": 2.2122066020965576, + "learning_rate": 1.173073463268366e-05, + "loss": 2.4889726638793945, + "step": 152180 + }, + { + "epoch": 0.41431876956580915, + "grad_norm": 2.0658583641052246, + "learning_rate": 1.1729644268774705e-05, + "loss": 2.478662872314453, + "step": 152200 + }, + { + "epoch": 0.4143732135565537, + "grad_norm": 2.169642448425293, + "learning_rate": 1.172855390486575e-05, + "loss": 2.49643611907959, + "step": 152220 + }, + { + "epoch": 0.4144276575472982, + "grad_norm": 2.342275381088257, + "learning_rate": 1.1727463540956797e-05, + "loss": 2.4861705780029295, + "step": 152240 + }, + { + "epoch": 0.4144821015380427, + "grad_norm": 2.2120614051818848, + "learning_rate": 1.1726373177047841e-05, + "loss": 2.506696891784668, + "step": 152260 + }, + { + "epoch": 0.41453654552878727, + "grad_norm": 2.5499725341796875, + "learning_rate": 1.1725282813138885e-05, + "loss": 2.4987258911132812, + "step": 152280 + }, + { + "epoch": 0.41459098951953177, + "grad_norm": 2.1305227279663086, + "learning_rate": 1.1724192449229933e-05, + "loss": 2.5152029037475585, + "step": 152300 + }, + { + "epoch": 0.41464543351027633, + "grad_norm": 2.0957722663879395, + "learning_rate": 1.1723102085320977e-05, + "loss": 2.5027713775634766, + "step": 152320 + }, + { + "epoch": 0.41469987750102083, + "grad_norm": 2.3435189723968506, + "learning_rate": 1.1722011721412022e-05, + "loss": 2.488251876831055, + "step": 152340 + }, + { + "epoch": 0.41475432149176533, + "grad_norm": 2.3377151489257812, + "learning_rate": 1.1720921357503068e-05, + "loss": 2.4694425582885744, + "step": 152360 + }, + { + "epoch": 0.4148087654825099, + "grad_norm": 2.085278272628784, + "learning_rate": 1.1719830993594112e-05, + "loss": 2.475999450683594, + "step": 152380 + }, + { + "epoch": 0.4148632094732544, + "grad_norm": 2.06420636177063, + "learning_rate": 1.171874062968516e-05, + "loss": 2.491413879394531, + "step": 152400 + }, + { + "epoch": 0.4149176534639989, + "grad_norm": 2.1944336891174316, + "learning_rate": 1.1717650265776204e-05, + "loss": 2.486222839355469, + "step": 152420 + }, + { + "epoch": 0.41497209745474345, + "grad_norm": 2.1123178005218506, + "learning_rate": 1.1716559901867248e-05, + "loss": 2.4896942138671876, + "step": 152440 + }, + { + "epoch": 0.41502654144548795, + "grad_norm": 2.1572978496551514, + "learning_rate": 1.1715469537958296e-05, + "loss": 2.489125061035156, + "step": 152460 + }, + { + "epoch": 0.41508098543623245, + "grad_norm": 2.162846326828003, + "learning_rate": 1.171437917404934e-05, + "loss": 2.484630012512207, + "step": 152480 + }, + { + "epoch": 0.415135429426977, + "grad_norm": 2.2476367950439453, + "learning_rate": 1.1713288810140386e-05, + "loss": 2.5114383697509766, + "step": 152500 + }, + { + "epoch": 0.4151898734177215, + "grad_norm": 2.1417298316955566, + "learning_rate": 1.1712198446231431e-05, + "loss": 2.5175045013427733, + "step": 152520 + }, + { + "epoch": 0.41524431740846607, + "grad_norm": 3.1991422176361084, + "learning_rate": 1.1711108082322475e-05, + "loss": 2.4996286392211915, + "step": 152540 + }, + { + "epoch": 0.41529876139921057, + "grad_norm": 2.228618860244751, + "learning_rate": 1.1710017718413523e-05, + "loss": 2.5094930648803713, + "step": 152560 + }, + { + "epoch": 0.41535320538995507, + "grad_norm": 2.0977354049682617, + "learning_rate": 1.1708927354504567e-05, + "loss": 2.480498123168945, + "step": 152580 + }, + { + "epoch": 0.41540764938069963, + "grad_norm": 2.195916175842285, + "learning_rate": 1.1707836990595611e-05, + "loss": 2.511490058898926, + "step": 152600 + }, + { + "epoch": 0.41546209337144413, + "grad_norm": 2.4965662956237793, + "learning_rate": 1.1706746626686659e-05, + "loss": 2.499538040161133, + "step": 152620 + }, + { + "epoch": 0.41551653736218863, + "grad_norm": 2.152832269668579, + "learning_rate": 1.1705656262777703e-05, + "loss": 2.5402128219604494, + "step": 152640 + }, + { + "epoch": 0.4155709813529332, + "grad_norm": 2.154085636138916, + "learning_rate": 1.1704565898868749e-05, + "loss": 2.471877861022949, + "step": 152660 + }, + { + "epoch": 0.4156254253436777, + "grad_norm": 2.170502185821533, + "learning_rate": 1.1703475534959794e-05, + "loss": 2.481765937805176, + "step": 152680 + }, + { + "epoch": 0.4156798693344222, + "grad_norm": 2.4356002807617188, + "learning_rate": 1.1702385171050839e-05, + "loss": 2.50082950592041, + "step": 152700 + }, + { + "epoch": 0.41573431332516675, + "grad_norm": 2.200852155685425, + "learning_rate": 1.1701294807141886e-05, + "loss": 2.511994171142578, + "step": 152720 + }, + { + "epoch": 0.41578875731591125, + "grad_norm": 2.2308366298675537, + "learning_rate": 1.170020444323293e-05, + "loss": 2.505307197570801, + "step": 152740 + }, + { + "epoch": 0.41584320130665575, + "grad_norm": 2.2121989727020264, + "learning_rate": 1.1699114079323974e-05, + "loss": 2.519135856628418, + "step": 152760 + }, + { + "epoch": 0.4158976452974003, + "grad_norm": 2.1710824966430664, + "learning_rate": 1.1698023715415022e-05, + "loss": 2.4732627868652344, + "step": 152780 + }, + { + "epoch": 0.4159520892881448, + "grad_norm": 2.163501501083374, + "learning_rate": 1.1696933351506066e-05, + "loss": 2.469739532470703, + "step": 152800 + }, + { + "epoch": 0.41600653327888937, + "grad_norm": 2.2888717651367188, + "learning_rate": 1.1695842987597112e-05, + "loss": 2.4951900482177733, + "step": 152820 + }, + { + "epoch": 0.41606097726963387, + "grad_norm": 2.138204336166382, + "learning_rate": 1.1694752623688158e-05, + "loss": 2.5046112060546877, + "step": 152840 + }, + { + "epoch": 0.41611542126037837, + "grad_norm": 2.140392780303955, + "learning_rate": 1.1693662259779202e-05, + "loss": 2.491603469848633, + "step": 152860 + }, + { + "epoch": 0.41616986525112293, + "grad_norm": 2.191232681274414, + "learning_rate": 1.1692571895870249e-05, + "loss": 2.5150184631347656, + "step": 152880 + }, + { + "epoch": 0.41622430924186743, + "grad_norm": 2.0721161365509033, + "learning_rate": 1.1691481531961293e-05, + "loss": 2.468581962585449, + "step": 152900 + }, + { + "epoch": 0.41627875323261193, + "grad_norm": 2.119623899459839, + "learning_rate": 1.1690391168052337e-05, + "loss": 2.4951688766479494, + "step": 152920 + }, + { + "epoch": 0.4163331972233565, + "grad_norm": 2.0411787033081055, + "learning_rate": 1.1689300804143385e-05, + "loss": 2.458195114135742, + "step": 152940 + }, + { + "epoch": 0.416387641214101, + "grad_norm": 2.0824365615844727, + "learning_rate": 1.1688210440234429e-05, + "loss": 2.515107536315918, + "step": 152960 + }, + { + "epoch": 0.4164420852048455, + "grad_norm": 2.1364667415618896, + "learning_rate": 1.1687120076325475e-05, + "loss": 2.4610530853271486, + "step": 152980 + }, + { + "epoch": 0.41649652919559005, + "grad_norm": 2.29968523979187, + "learning_rate": 1.168602971241652e-05, + "loss": 2.4950225830078123, + "step": 153000 + }, + { + "epoch": 0.41655097318633455, + "grad_norm": 2.1348228454589844, + "learning_rate": 1.1684939348507565e-05, + "loss": 2.505482864379883, + "step": 153020 + }, + { + "epoch": 0.4166054171770791, + "grad_norm": 2.0534260272979736, + "learning_rate": 1.168384898459861e-05, + "loss": 2.498996353149414, + "step": 153040 + }, + { + "epoch": 0.4166598611678236, + "grad_norm": 2.2645156383514404, + "learning_rate": 1.1682758620689656e-05, + "loss": 2.483547592163086, + "step": 153060 + }, + { + "epoch": 0.4167143051585681, + "grad_norm": 2.2462308406829834, + "learning_rate": 1.16816682567807e-05, + "loss": 2.4588428497314454, + "step": 153080 + }, + { + "epoch": 0.41676874914931267, + "grad_norm": 2.216859817504883, + "learning_rate": 1.1680577892871748e-05, + "loss": 2.4958654403686524, + "step": 153100 + }, + { + "epoch": 0.41682319314005717, + "grad_norm": 2.0985827445983887, + "learning_rate": 1.1679487528962792e-05, + "loss": 2.4600839614868164, + "step": 153120 + }, + { + "epoch": 0.41687763713080167, + "grad_norm": 2.1504063606262207, + "learning_rate": 1.1678397165053838e-05, + "loss": 2.488781547546387, + "step": 153140 + }, + { + "epoch": 0.4169320811215462, + "grad_norm": 2.055359363555908, + "learning_rate": 1.1677306801144884e-05, + "loss": 2.5096342086791994, + "step": 153160 + }, + { + "epoch": 0.41698652511229073, + "grad_norm": 2.0668957233428955, + "learning_rate": 1.1676216437235928e-05, + "loss": 2.4886260986328126, + "step": 153180 + }, + { + "epoch": 0.41704096910303523, + "grad_norm": 2.0433435440063477, + "learning_rate": 1.1675126073326974e-05, + "loss": 2.5077842712402343, + "step": 153200 + }, + { + "epoch": 0.4170954130937798, + "grad_norm": 2.2635746002197266, + "learning_rate": 1.167403570941802e-05, + "loss": 2.4929800033569336, + "step": 153220 + }, + { + "epoch": 0.4171498570845243, + "grad_norm": 2.31607985496521, + "learning_rate": 1.1672945345509064e-05, + "loss": 2.51671142578125, + "step": 153240 + }, + { + "epoch": 0.4172043010752688, + "grad_norm": 2.165215015411377, + "learning_rate": 1.1671854981600111e-05, + "loss": 2.496302032470703, + "step": 153260 + }, + { + "epoch": 0.41725874506601335, + "grad_norm": 2.1869819164276123, + "learning_rate": 1.1670764617691155e-05, + "loss": 2.478242874145508, + "step": 153280 + }, + { + "epoch": 0.41731318905675785, + "grad_norm": 2.127497673034668, + "learning_rate": 1.1669674253782201e-05, + "loss": 2.515002632141113, + "step": 153300 + }, + { + "epoch": 0.4173676330475024, + "grad_norm": 2.1773743629455566, + "learning_rate": 1.1668583889873247e-05, + "loss": 2.473893165588379, + "step": 153320 + }, + { + "epoch": 0.4174220770382469, + "grad_norm": 2.206271171569824, + "learning_rate": 1.1667493525964291e-05, + "loss": 2.4664352416992186, + "step": 153340 + }, + { + "epoch": 0.4174765210289914, + "grad_norm": 2.1899051666259766, + "learning_rate": 1.1666403162055337e-05, + "loss": 2.529731559753418, + "step": 153360 + }, + { + "epoch": 0.41753096501973597, + "grad_norm": 2.2478723526000977, + "learning_rate": 1.1665312798146383e-05, + "loss": 2.4895578384399415, + "step": 153380 + }, + { + "epoch": 0.41758540901048047, + "grad_norm": 2.9196033477783203, + "learning_rate": 1.1664222434237427e-05, + "loss": 2.4951251983642577, + "step": 153400 + }, + { + "epoch": 0.41763985300122497, + "grad_norm": 2.6733710765838623, + "learning_rate": 1.1663132070328474e-05, + "loss": 2.4704986572265626, + "step": 153420 + }, + { + "epoch": 0.4176942969919695, + "grad_norm": 2.1099987030029297, + "learning_rate": 1.1662041706419518e-05, + "loss": 2.494866371154785, + "step": 153440 + }, + { + "epoch": 0.41774874098271403, + "grad_norm": 2.1670327186584473, + "learning_rate": 1.1660951342510564e-05, + "loss": 2.4871225357055664, + "step": 153460 + }, + { + "epoch": 0.41780318497345853, + "grad_norm": 2.1424388885498047, + "learning_rate": 1.165986097860161e-05, + "loss": 2.494273376464844, + "step": 153480 + }, + { + "epoch": 0.4178576289642031, + "grad_norm": 2.072059392929077, + "learning_rate": 1.1658770614692654e-05, + "loss": 2.4929874420166014, + "step": 153500 + }, + { + "epoch": 0.4179120729549476, + "grad_norm": 2.0092077255249023, + "learning_rate": 1.16576802507837e-05, + "loss": 2.501616096496582, + "step": 153520 + }, + { + "epoch": 0.41796651694569215, + "grad_norm": 2.7178900241851807, + "learning_rate": 1.1656589886874746e-05, + "loss": 2.4802026748657227, + "step": 153540 + }, + { + "epoch": 0.41802096093643665, + "grad_norm": 2.0531463623046875, + "learning_rate": 1.165549952296579e-05, + "loss": 2.540814208984375, + "step": 153560 + }, + { + "epoch": 0.41807540492718115, + "grad_norm": 2.2686498165130615, + "learning_rate": 1.1654409159056837e-05, + "loss": 2.51744384765625, + "step": 153580 + }, + { + "epoch": 0.4181298489179257, + "grad_norm": 2.1475136280059814, + "learning_rate": 1.1653318795147881e-05, + "loss": 2.4747797012329102, + "step": 153600 + }, + { + "epoch": 0.4181842929086702, + "grad_norm": 2.2690603733062744, + "learning_rate": 1.1652228431238927e-05, + "loss": 2.5035673141479493, + "step": 153620 + }, + { + "epoch": 0.4182387368994147, + "grad_norm": 2.525132656097412, + "learning_rate": 1.1651138067329973e-05, + "loss": 2.4911884307861327, + "step": 153640 + }, + { + "epoch": 0.41829318089015927, + "grad_norm": 2.0896902084350586, + "learning_rate": 1.1650047703421017e-05, + "loss": 2.4887033462524415, + "step": 153660 + }, + { + "epoch": 0.41834762488090377, + "grad_norm": 2.2898902893066406, + "learning_rate": 1.1648957339512063e-05, + "loss": 2.476206588745117, + "step": 153680 + }, + { + "epoch": 0.41840206887164827, + "grad_norm": 2.137094736099243, + "learning_rate": 1.1647866975603109e-05, + "loss": 2.500883865356445, + "step": 153700 + }, + { + "epoch": 0.4184565128623928, + "grad_norm": 2.164341449737549, + "learning_rate": 1.1646776611694155e-05, + "loss": 2.505929183959961, + "step": 153720 + }, + { + "epoch": 0.41851095685313733, + "grad_norm": 2.0892386436462402, + "learning_rate": 1.16456862477852e-05, + "loss": 2.4895166397094726, + "step": 153740 + }, + { + "epoch": 0.41856540084388183, + "grad_norm": 2.1121108531951904, + "learning_rate": 1.1644595883876244e-05, + "loss": 2.503096580505371, + "step": 153760 + }, + { + "epoch": 0.4186198448346264, + "grad_norm": 2.1591899394989014, + "learning_rate": 1.164350551996729e-05, + "loss": 2.493307113647461, + "step": 153780 + }, + { + "epoch": 0.4186742888253709, + "grad_norm": 2.091733694076538, + "learning_rate": 1.1642415156058336e-05, + "loss": 2.519868850708008, + "step": 153800 + }, + { + "epoch": 0.41872873281611545, + "grad_norm": 2.042053461074829, + "learning_rate": 1.164132479214938e-05, + "loss": 2.489668273925781, + "step": 153820 + }, + { + "epoch": 0.41878317680685995, + "grad_norm": 2.1157467365264893, + "learning_rate": 1.1640234428240426e-05, + "loss": 2.48546142578125, + "step": 153840 + }, + { + "epoch": 0.41883762079760445, + "grad_norm": 2.321258068084717, + "learning_rate": 1.1639144064331472e-05, + "loss": 2.4942739486694334, + "step": 153860 + }, + { + "epoch": 0.418892064788349, + "grad_norm": 2.10569429397583, + "learning_rate": 1.1638053700422518e-05, + "loss": 2.4531993865966797, + "step": 153880 + }, + { + "epoch": 0.4189465087790935, + "grad_norm": 2.273425579071045, + "learning_rate": 1.1636963336513563e-05, + "loss": 2.488649368286133, + "step": 153900 + }, + { + "epoch": 0.419000952769838, + "grad_norm": 2.0997226238250732, + "learning_rate": 1.1635872972604608e-05, + "loss": 2.5148059844970705, + "step": 153920 + }, + { + "epoch": 0.41905539676058257, + "grad_norm": 2.176391363143921, + "learning_rate": 1.1634782608695653e-05, + "loss": 2.4592927932739257, + "step": 153940 + }, + { + "epoch": 0.41910984075132707, + "grad_norm": 2.155466079711914, + "learning_rate": 1.1633692244786699e-05, + "loss": 2.5025009155273437, + "step": 153960 + }, + { + "epoch": 0.41916428474207157, + "grad_norm": 2.3300390243530273, + "learning_rate": 1.1632601880877743e-05, + "loss": 2.46256160736084, + "step": 153980 + }, + { + "epoch": 0.4192187287328161, + "grad_norm": 2.1451690196990967, + "learning_rate": 1.1631511516968789e-05, + "loss": 2.5152679443359376, + "step": 154000 + }, + { + "epoch": 0.4192731727235606, + "grad_norm": 2.2003865242004395, + "learning_rate": 1.1630421153059835e-05, + "loss": 2.516012191772461, + "step": 154020 + }, + { + "epoch": 0.4193276167143052, + "grad_norm": 2.19905161857605, + "learning_rate": 1.162933078915088e-05, + "loss": 2.504568099975586, + "step": 154040 + }, + { + "epoch": 0.4193820607050497, + "grad_norm": 2.2495765686035156, + "learning_rate": 1.1628240425241926e-05, + "loss": 2.4888992309570312, + "step": 154060 + }, + { + "epoch": 0.4194365046957942, + "grad_norm": 2.1085028648376465, + "learning_rate": 1.162715006133297e-05, + "loss": 2.464519500732422, + "step": 154080 + }, + { + "epoch": 0.41949094868653874, + "grad_norm": 2.1545214653015137, + "learning_rate": 1.1626059697424016e-05, + "loss": 2.4885635375976562, + "step": 154100 + }, + { + "epoch": 0.41954539267728325, + "grad_norm": 2.0902111530303955, + "learning_rate": 1.1624969333515062e-05, + "loss": 2.4784692764282226, + "step": 154120 + }, + { + "epoch": 0.41959983666802775, + "grad_norm": 2.1104695796966553, + "learning_rate": 1.1623878969606106e-05, + "loss": 2.4949602127075194, + "step": 154140 + }, + { + "epoch": 0.4196542806587723, + "grad_norm": 2.1665172576904297, + "learning_rate": 1.1622788605697152e-05, + "loss": 2.508984184265137, + "step": 154160 + }, + { + "epoch": 0.4197087246495168, + "grad_norm": 2.1775975227355957, + "learning_rate": 1.1621698241788198e-05, + "loss": 2.501078414916992, + "step": 154180 + }, + { + "epoch": 0.4197631686402613, + "grad_norm": 2.1560776233673096, + "learning_rate": 1.1620607877879244e-05, + "loss": 2.4737617492675783, + "step": 154200 + }, + { + "epoch": 0.41981761263100587, + "grad_norm": 2.1058802604675293, + "learning_rate": 1.1619517513970288e-05, + "loss": 2.4932878494262694, + "step": 154220 + }, + { + "epoch": 0.41987205662175037, + "grad_norm": 2.2233288288116455, + "learning_rate": 1.1618427150061334e-05, + "loss": 2.5042469024658205, + "step": 154240 + }, + { + "epoch": 0.41992650061249487, + "grad_norm": 2.1146774291992188, + "learning_rate": 1.161733678615238e-05, + "loss": 2.4836505889892577, + "step": 154260 + }, + { + "epoch": 0.4199809446032394, + "grad_norm": 2.2774388790130615, + "learning_rate": 1.1616246422243425e-05, + "loss": 2.5036289215087892, + "step": 154280 + }, + { + "epoch": 0.4200353885939839, + "grad_norm": 2.160870313644409, + "learning_rate": 1.161515605833447e-05, + "loss": 2.479629325866699, + "step": 154300 + }, + { + "epoch": 0.4200898325847285, + "grad_norm": 2.103208541870117, + "learning_rate": 1.1614065694425515e-05, + "loss": 2.4999109268188477, + "step": 154320 + }, + { + "epoch": 0.420144276575473, + "grad_norm": 2.13564395904541, + "learning_rate": 1.1612975330516561e-05, + "loss": 2.4724752426147463, + "step": 154340 + }, + { + "epoch": 0.4201987205662175, + "grad_norm": 2.1278722286224365, + "learning_rate": 1.1611884966607607e-05, + "loss": 2.4860113143920897, + "step": 154360 + }, + { + "epoch": 0.42025316455696204, + "grad_norm": 2.2254650592803955, + "learning_rate": 1.1610794602698651e-05, + "loss": 2.510558319091797, + "step": 154380 + }, + { + "epoch": 0.42030760854770655, + "grad_norm": 2.150798797607422, + "learning_rate": 1.1609704238789697e-05, + "loss": 2.4926156997680664, + "step": 154400 + }, + { + "epoch": 0.42036205253845105, + "grad_norm": 2.220181465148926, + "learning_rate": 1.1608613874880743e-05, + "loss": 2.4829174041748048, + "step": 154420 + }, + { + "epoch": 0.4204164965291956, + "grad_norm": 2.1635048389434814, + "learning_rate": 1.1607523510971788e-05, + "loss": 2.486357307434082, + "step": 154440 + }, + { + "epoch": 0.4204709405199401, + "grad_norm": 2.1821460723876953, + "learning_rate": 1.1606433147062833e-05, + "loss": 2.468400001525879, + "step": 154460 + }, + { + "epoch": 0.4205253845106846, + "grad_norm": 2.1118292808532715, + "learning_rate": 1.1605342783153878e-05, + "loss": 2.5168556213378905, + "step": 154480 + }, + { + "epoch": 0.42057982850142916, + "grad_norm": 2.04707670211792, + "learning_rate": 1.1604252419244924e-05, + "loss": 2.486295700073242, + "step": 154500 + }, + { + "epoch": 0.42063427249217367, + "grad_norm": 2.2302539348602295, + "learning_rate": 1.160316205533597e-05, + "loss": 2.475107765197754, + "step": 154520 + }, + { + "epoch": 0.4206887164829182, + "grad_norm": 2.1718709468841553, + "learning_rate": 1.1602071691427014e-05, + "loss": 2.511627769470215, + "step": 154540 + }, + { + "epoch": 0.4207431604736627, + "grad_norm": 2.0563549995422363, + "learning_rate": 1.160098132751806e-05, + "loss": 2.4878244400024414, + "step": 154560 + }, + { + "epoch": 0.4207976044644072, + "grad_norm": 2.4063804149627686, + "learning_rate": 1.1599890963609106e-05, + "loss": 2.476771354675293, + "step": 154580 + }, + { + "epoch": 0.4208520484551518, + "grad_norm": 2.084301233291626, + "learning_rate": 1.1598800599700151e-05, + "loss": 2.497724723815918, + "step": 154600 + }, + { + "epoch": 0.4209064924458963, + "grad_norm": 2.086655855178833, + "learning_rate": 1.1597710235791196e-05, + "loss": 2.514323616027832, + "step": 154620 + }, + { + "epoch": 0.4209609364366408, + "grad_norm": 2.1490426063537598, + "learning_rate": 1.1596619871882241e-05, + "loss": 2.5071155548095705, + "step": 154640 + }, + { + "epoch": 0.42101538042738534, + "grad_norm": 2.1014623641967773, + "learning_rate": 1.1595529507973287e-05, + "loss": 2.4913267135620116, + "step": 154660 + }, + { + "epoch": 0.42106982441812985, + "grad_norm": 2.278451442718506, + "learning_rate": 1.1594439144064333e-05, + "loss": 2.4290821075439455, + "step": 154680 + }, + { + "epoch": 0.42112426840887435, + "grad_norm": 2.077684164047241, + "learning_rate": 1.1593348780155377e-05, + "loss": 2.4851263046264647, + "step": 154700 + }, + { + "epoch": 0.4211787123996189, + "grad_norm": 2.21368670463562, + "learning_rate": 1.1592258416246423e-05, + "loss": 2.482859802246094, + "step": 154720 + }, + { + "epoch": 0.4212331563903634, + "grad_norm": 2.245361089706421, + "learning_rate": 1.1591168052337469e-05, + "loss": 2.4737571716308593, + "step": 154740 + }, + { + "epoch": 0.42128760038110796, + "grad_norm": 2.3601760864257812, + "learning_rate": 1.1590077688428515e-05, + "loss": 2.482637405395508, + "step": 154760 + }, + { + "epoch": 0.42134204437185246, + "grad_norm": 2.1323082447052, + "learning_rate": 1.1588987324519559e-05, + "loss": 2.495559310913086, + "step": 154780 + }, + { + "epoch": 0.42139648836259697, + "grad_norm": 2.3231470584869385, + "learning_rate": 1.1587896960610604e-05, + "loss": 2.5054113388061525, + "step": 154800 + }, + { + "epoch": 0.4214509323533415, + "grad_norm": 2.365267038345337, + "learning_rate": 1.158680659670165e-05, + "loss": 2.5030025482177733, + "step": 154820 + }, + { + "epoch": 0.421505376344086, + "grad_norm": 2.1582818031311035, + "learning_rate": 1.1585716232792696e-05, + "loss": 2.5070152282714844, + "step": 154840 + }, + { + "epoch": 0.4215598203348305, + "grad_norm": 2.4366674423217773, + "learning_rate": 1.158462586888374e-05, + "loss": 2.5013628005981445, + "step": 154860 + }, + { + "epoch": 0.4216142643255751, + "grad_norm": 2.071704626083374, + "learning_rate": 1.1583535504974786e-05, + "loss": 2.459848976135254, + "step": 154880 + }, + { + "epoch": 0.4216687083163196, + "grad_norm": 2.224139928817749, + "learning_rate": 1.1582445141065832e-05, + "loss": 2.499724006652832, + "step": 154900 + }, + { + "epoch": 0.4217231523070641, + "grad_norm": 2.107853412628174, + "learning_rate": 1.1581354777156878e-05, + "loss": 2.4601694107055665, + "step": 154920 + }, + { + "epoch": 0.42177759629780864, + "grad_norm": 2.204930305480957, + "learning_rate": 1.1580264413247922e-05, + "loss": 2.5024368286132814, + "step": 154940 + }, + { + "epoch": 0.42183204028855315, + "grad_norm": 2.183413505554199, + "learning_rate": 1.1579174049338968e-05, + "loss": 2.4699689865112306, + "step": 154960 + }, + { + "epoch": 0.42188648427929765, + "grad_norm": 2.3343024253845215, + "learning_rate": 1.1578083685430013e-05, + "loss": 2.5198486328125, + "step": 154980 + }, + { + "epoch": 0.4219409282700422, + "grad_norm": 2.414024829864502, + "learning_rate": 1.157699332152106e-05, + "loss": 2.525467300415039, + "step": 155000 + }, + { + "epoch": 0.4219953722607867, + "grad_norm": 2.0188419818878174, + "learning_rate": 1.1575902957612103e-05, + "loss": 2.5010732650756835, + "step": 155020 + }, + { + "epoch": 0.42204981625153126, + "grad_norm": 2.143143653869629, + "learning_rate": 1.1574812593703149e-05, + "loss": 2.4673799514770507, + "step": 155040 + }, + { + "epoch": 0.42210426024227576, + "grad_norm": 2.139263868331909, + "learning_rate": 1.1573722229794195e-05, + "loss": 2.5177536010742188, + "step": 155060 + }, + { + "epoch": 0.42215870423302027, + "grad_norm": 2.1473653316497803, + "learning_rate": 1.157263186588524e-05, + "loss": 2.5008445739746095, + "step": 155080 + }, + { + "epoch": 0.4222131482237648, + "grad_norm": 2.204444646835327, + "learning_rate": 1.1571541501976285e-05, + "loss": 2.507358741760254, + "step": 155100 + }, + { + "epoch": 0.4222675922145093, + "grad_norm": 2.1252522468566895, + "learning_rate": 1.157045113806733e-05, + "loss": 2.503427505493164, + "step": 155120 + }, + { + "epoch": 0.4223220362052538, + "grad_norm": 2.2165162563323975, + "learning_rate": 1.1569360774158376e-05, + "loss": 2.463628578186035, + "step": 155140 + }, + { + "epoch": 0.4223764801959984, + "grad_norm": 2.064941883087158, + "learning_rate": 1.1568270410249422e-05, + "loss": 2.5092859268188477, + "step": 155160 + }, + { + "epoch": 0.4224309241867429, + "grad_norm": 2.1290574073791504, + "learning_rate": 1.1567180046340466e-05, + "loss": 2.4950439453125, + "step": 155180 + }, + { + "epoch": 0.4224853681774874, + "grad_norm": 2.385284900665283, + "learning_rate": 1.1566089682431512e-05, + "loss": 2.499935340881348, + "step": 155200 + }, + { + "epoch": 0.42253981216823194, + "grad_norm": 2.1985268592834473, + "learning_rate": 1.1564999318522558e-05, + "loss": 2.483190155029297, + "step": 155220 + }, + { + "epoch": 0.42259425615897644, + "grad_norm": 2.1544883251190186, + "learning_rate": 1.1563908954613602e-05, + "loss": 2.4776493072509767, + "step": 155240 + }, + { + "epoch": 0.422648700149721, + "grad_norm": 2.1587975025177, + "learning_rate": 1.1562818590704648e-05, + "loss": 2.4900386810302733, + "step": 155260 + }, + { + "epoch": 0.4227031441404655, + "grad_norm": 2.0881426334381104, + "learning_rate": 1.1561728226795694e-05, + "loss": 2.4747901916503907, + "step": 155280 + }, + { + "epoch": 0.42275758813121, + "grad_norm": 2.1050429344177246, + "learning_rate": 1.156063786288674e-05, + "loss": 2.4915176391601563, + "step": 155300 + }, + { + "epoch": 0.42281203212195456, + "grad_norm": 2.1450932025909424, + "learning_rate": 1.1559547498977785e-05, + "loss": 2.4747669219970705, + "step": 155320 + }, + { + "epoch": 0.42286647611269906, + "grad_norm": 2.29890513420105, + "learning_rate": 1.155845713506883e-05, + "loss": 2.483862113952637, + "step": 155340 + }, + { + "epoch": 0.42292092010344357, + "grad_norm": 2.106128215789795, + "learning_rate": 1.1557366771159875e-05, + "loss": 2.5056161880493164, + "step": 155360 + }, + { + "epoch": 0.4229753640941881, + "grad_norm": 2.1077699661254883, + "learning_rate": 1.1556276407250921e-05, + "loss": 2.5146738052368165, + "step": 155380 + }, + { + "epoch": 0.4230298080849326, + "grad_norm": 2.253640651702881, + "learning_rate": 1.1555186043341965e-05, + "loss": 2.483918380737305, + "step": 155400 + }, + { + "epoch": 0.4230842520756771, + "grad_norm": 2.2492666244506836, + "learning_rate": 1.1554095679433013e-05, + "loss": 2.5059558868408205, + "step": 155420 + }, + { + "epoch": 0.4231386960664217, + "grad_norm": 2.1780412197113037, + "learning_rate": 1.1553005315524057e-05, + "loss": 2.5126873016357423, + "step": 155440 + }, + { + "epoch": 0.4231931400571662, + "grad_norm": 2.1087708473205566, + "learning_rate": 1.1551914951615103e-05, + "loss": 2.4560373306274412, + "step": 155460 + }, + { + "epoch": 0.4232475840479107, + "grad_norm": 2.1621482372283936, + "learning_rate": 1.1550824587706148e-05, + "loss": 2.4888689041137697, + "step": 155480 + }, + { + "epoch": 0.42330202803865524, + "grad_norm": 2.125279664993286, + "learning_rate": 1.1549734223797193e-05, + "loss": 2.4794795989990233, + "step": 155500 + }, + { + "epoch": 0.42335647202939974, + "grad_norm": 2.2324395179748535, + "learning_rate": 1.1548643859888238e-05, + "loss": 2.492657470703125, + "step": 155520 + }, + { + "epoch": 0.4234109160201443, + "grad_norm": 1.994541883468628, + "learning_rate": 1.1547553495979284e-05, + "loss": 2.4710386276245115, + "step": 155540 + }, + { + "epoch": 0.4234653600108888, + "grad_norm": 2.1731839179992676, + "learning_rate": 1.1546463132070328e-05, + "loss": 2.4785398483276366, + "step": 155560 + }, + { + "epoch": 0.4235198040016333, + "grad_norm": 2.118649482727051, + "learning_rate": 1.1545372768161376e-05, + "loss": 2.4778648376464845, + "step": 155580 + }, + { + "epoch": 0.42357424799237786, + "grad_norm": 2.140504837036133, + "learning_rate": 1.154428240425242e-05, + "loss": 2.5138505935668944, + "step": 155600 + }, + { + "epoch": 0.42362869198312236, + "grad_norm": 2.1510703563690186, + "learning_rate": 1.1543192040343466e-05, + "loss": 2.497018051147461, + "step": 155620 + }, + { + "epoch": 0.42368313597386686, + "grad_norm": 2.168665647506714, + "learning_rate": 1.1542101676434512e-05, + "loss": 2.5195234298706053, + "step": 155640 + }, + { + "epoch": 0.4237375799646114, + "grad_norm": 2.373462677001953, + "learning_rate": 1.1541011312525556e-05, + "loss": 2.508165740966797, + "step": 155660 + }, + { + "epoch": 0.4237920239553559, + "grad_norm": 2.147095203399658, + "learning_rate": 1.1539920948616601e-05, + "loss": 2.4658437728881837, + "step": 155680 + }, + { + "epoch": 0.4238464679461004, + "grad_norm": 2.1074790954589844, + "learning_rate": 1.1538830584707647e-05, + "loss": 2.511565399169922, + "step": 155700 + }, + { + "epoch": 0.423900911936845, + "grad_norm": 2.2254631519317627, + "learning_rate": 1.1537740220798691e-05, + "loss": 2.502905082702637, + "step": 155720 + }, + { + "epoch": 0.4239553559275895, + "grad_norm": 2.111626625061035, + "learning_rate": 1.1536649856889739e-05, + "loss": 2.4767183303833007, + "step": 155740 + }, + { + "epoch": 0.42400979991833404, + "grad_norm": 2.1776294708251953, + "learning_rate": 1.1535559492980783e-05, + "loss": 2.5053890228271483, + "step": 155760 + }, + { + "epoch": 0.42406424390907854, + "grad_norm": 2.0914838314056396, + "learning_rate": 1.1534469129071829e-05, + "loss": 2.4878562927246093, + "step": 155780 + }, + { + "epoch": 0.42411868789982304, + "grad_norm": 2.3123233318328857, + "learning_rate": 1.1533378765162875e-05, + "loss": 2.466925621032715, + "step": 155800 + }, + { + "epoch": 0.4241731318905676, + "grad_norm": 2.0797669887542725, + "learning_rate": 1.1532288401253919e-05, + "loss": 2.461601448059082, + "step": 155820 + }, + { + "epoch": 0.4242275758813121, + "grad_norm": 2.237653970718384, + "learning_rate": 1.1531198037344965e-05, + "loss": 2.4656135559082033, + "step": 155840 + }, + { + "epoch": 0.4242820198720566, + "grad_norm": 2.1437623500823975, + "learning_rate": 1.153010767343601e-05, + "loss": 2.4767812728881835, + "step": 155860 + }, + { + "epoch": 0.42433646386280116, + "grad_norm": 2.032850503921509, + "learning_rate": 1.1529017309527054e-05, + "loss": 2.475576400756836, + "step": 155880 + }, + { + "epoch": 0.42439090785354566, + "grad_norm": 2.101215362548828, + "learning_rate": 1.1527926945618102e-05, + "loss": 2.4880178451538084, + "step": 155900 + }, + { + "epoch": 0.42444535184429016, + "grad_norm": 2.2030816078186035, + "learning_rate": 1.1526836581709146e-05, + "loss": 2.4886009216308596, + "step": 155920 + }, + { + "epoch": 0.4244997958350347, + "grad_norm": 2.1285958290100098, + "learning_rate": 1.1525746217800192e-05, + "loss": 2.511761474609375, + "step": 155940 + }, + { + "epoch": 0.4245542398257792, + "grad_norm": 2.2668564319610596, + "learning_rate": 1.1524655853891238e-05, + "loss": 2.469369888305664, + "step": 155960 + }, + { + "epoch": 0.4246086838165237, + "grad_norm": 2.1417293548583984, + "learning_rate": 1.1523565489982282e-05, + "loss": 2.4806711196899416, + "step": 155980 + }, + { + "epoch": 0.4246631278072683, + "grad_norm": 2.4660725593566895, + "learning_rate": 1.1522475126073328e-05, + "loss": 2.5006032943725587, + "step": 156000 + }, + { + "epoch": 0.4247175717980128, + "grad_norm": 2.25288724899292, + "learning_rate": 1.1521384762164373e-05, + "loss": 2.4898656845092773, + "step": 156020 + }, + { + "epoch": 0.42477201578875734, + "grad_norm": 2.1031525135040283, + "learning_rate": 1.1520294398255418e-05, + "loss": 2.499335289001465, + "step": 156040 + }, + { + "epoch": 0.42482645977950184, + "grad_norm": 2.201645612716675, + "learning_rate": 1.1519204034346465e-05, + "loss": 2.524640655517578, + "step": 156060 + }, + { + "epoch": 0.42488090377024634, + "grad_norm": 2.1158432960510254, + "learning_rate": 1.151811367043751e-05, + "loss": 2.503041648864746, + "step": 156080 + }, + { + "epoch": 0.4249353477609909, + "grad_norm": 2.1965208053588867, + "learning_rate": 1.1517023306528555e-05, + "loss": 2.4606693267822264, + "step": 156100 + }, + { + "epoch": 0.4249897917517354, + "grad_norm": 2.2334516048431396, + "learning_rate": 1.15159329426196e-05, + "loss": 2.495084762573242, + "step": 156120 + }, + { + "epoch": 0.4250442357424799, + "grad_norm": 2.228023052215576, + "learning_rate": 1.1514842578710645e-05, + "loss": 2.5070522308349608, + "step": 156140 + }, + { + "epoch": 0.42509867973322446, + "grad_norm": 2.1632418632507324, + "learning_rate": 1.151375221480169e-05, + "loss": 2.4958099365234374, + "step": 156160 + }, + { + "epoch": 0.42515312372396896, + "grad_norm": 2.1069767475128174, + "learning_rate": 1.1512661850892737e-05, + "loss": 2.500440788269043, + "step": 156180 + }, + { + "epoch": 0.42520756771471346, + "grad_norm": 2.6428427696228027, + "learning_rate": 1.151157148698378e-05, + "loss": 2.4970149993896484, + "step": 156200 + }, + { + "epoch": 0.425262011705458, + "grad_norm": 2.1324191093444824, + "learning_rate": 1.1510481123074828e-05, + "loss": 2.5066049575805662, + "step": 156220 + }, + { + "epoch": 0.4253164556962025, + "grad_norm": 2.2743170261383057, + "learning_rate": 1.1509390759165872e-05, + "loss": 2.4865673065185545, + "step": 156240 + }, + { + "epoch": 0.4253708996869471, + "grad_norm": 2.205216646194458, + "learning_rate": 1.1508300395256918e-05, + "loss": 2.4751142501831054, + "step": 156260 + }, + { + "epoch": 0.4254253436776916, + "grad_norm": 2.571422815322876, + "learning_rate": 1.1507210031347964e-05, + "loss": 2.5031524658203126, + "step": 156280 + }, + { + "epoch": 0.4254797876684361, + "grad_norm": 2.2539360523223877, + "learning_rate": 1.1506119667439008e-05, + "loss": 2.474857711791992, + "step": 156300 + }, + { + "epoch": 0.42553423165918064, + "grad_norm": 2.20465087890625, + "learning_rate": 1.1505029303530054e-05, + "loss": 2.516607475280762, + "step": 156320 + }, + { + "epoch": 0.42558867564992514, + "grad_norm": 2.1094136238098145, + "learning_rate": 1.15039389396211e-05, + "loss": 2.469923210144043, + "step": 156340 + }, + { + "epoch": 0.42564311964066964, + "grad_norm": 2.1099143028259277, + "learning_rate": 1.1502848575712144e-05, + "loss": 2.4984375, + "step": 156360 + }, + { + "epoch": 0.4256975636314142, + "grad_norm": 2.3120861053466797, + "learning_rate": 1.1501758211803191e-05, + "loss": 2.4939754486083983, + "step": 156380 + }, + { + "epoch": 0.4257520076221587, + "grad_norm": 2.1810355186462402, + "learning_rate": 1.1500667847894235e-05, + "loss": 2.4844390869140627, + "step": 156400 + }, + { + "epoch": 0.4258064516129032, + "grad_norm": 2.1593682765960693, + "learning_rate": 1.149957748398528e-05, + "loss": 2.4949684143066406, + "step": 156420 + }, + { + "epoch": 0.42586089560364776, + "grad_norm": 2.120736598968506, + "learning_rate": 1.1498487120076327e-05, + "loss": 2.481001853942871, + "step": 156440 + }, + { + "epoch": 0.42591533959439226, + "grad_norm": 2.1624112129211426, + "learning_rate": 1.1497396756167371e-05, + "loss": 2.506938934326172, + "step": 156460 + }, + { + "epoch": 0.42596978358513676, + "grad_norm": 2.1738433837890625, + "learning_rate": 1.1496306392258417e-05, + "loss": 2.523612403869629, + "step": 156480 + }, + { + "epoch": 0.4260242275758813, + "grad_norm": 2.4546947479248047, + "learning_rate": 1.1495216028349463e-05, + "loss": 2.4696239471435546, + "step": 156500 + }, + { + "epoch": 0.4260786715666258, + "grad_norm": 2.078015089035034, + "learning_rate": 1.1494125664440507e-05, + "loss": 2.475973129272461, + "step": 156520 + }, + { + "epoch": 0.4261331155573704, + "grad_norm": 2.1417391300201416, + "learning_rate": 1.1493035300531554e-05, + "loss": 2.477634811401367, + "step": 156540 + }, + { + "epoch": 0.4261875595481149, + "grad_norm": 2.7936019897460938, + "learning_rate": 1.1491944936622598e-05, + "loss": 2.4921804428100587, + "step": 156560 + }, + { + "epoch": 0.4262420035388594, + "grad_norm": 2.3025736808776855, + "learning_rate": 1.1490854572713643e-05, + "loss": 2.5031816482543947, + "step": 156580 + }, + { + "epoch": 0.42629644752960394, + "grad_norm": 2.081057071685791, + "learning_rate": 1.148976420880469e-05, + "loss": 2.474669075012207, + "step": 156600 + }, + { + "epoch": 0.42635089152034844, + "grad_norm": 2.1447651386260986, + "learning_rate": 1.1488673844895734e-05, + "loss": 2.483512306213379, + "step": 156620 + }, + { + "epoch": 0.42640533551109294, + "grad_norm": 2.081115484237671, + "learning_rate": 1.148758348098678e-05, + "loss": 2.47891845703125, + "step": 156640 + }, + { + "epoch": 0.4264597795018375, + "grad_norm": 2.1511170864105225, + "learning_rate": 1.1486493117077826e-05, + "loss": 2.497416114807129, + "step": 156660 + }, + { + "epoch": 0.426514223492582, + "grad_norm": 2.2130048274993896, + "learning_rate": 1.148540275316887e-05, + "loss": 2.478743553161621, + "step": 156680 + }, + { + "epoch": 0.4265686674833265, + "grad_norm": 2.811872959136963, + "learning_rate": 1.1484312389259917e-05, + "loss": 2.508755683898926, + "step": 156700 + }, + { + "epoch": 0.42662311147407106, + "grad_norm": 2.1674869060516357, + "learning_rate": 1.1483222025350962e-05, + "loss": 2.482666015625, + "step": 156720 + }, + { + "epoch": 0.42667755546481556, + "grad_norm": 2.1392743587493896, + "learning_rate": 1.1482131661442006e-05, + "loss": 2.4756717681884766, + "step": 156740 + }, + { + "epoch": 0.4267319994555601, + "grad_norm": 2.072892904281616, + "learning_rate": 1.1481041297533053e-05, + "loss": 2.490447235107422, + "step": 156760 + }, + { + "epoch": 0.4267864434463046, + "grad_norm": 2.2837421894073486, + "learning_rate": 1.1479950933624097e-05, + "loss": 2.478011894226074, + "step": 156780 + }, + { + "epoch": 0.4268408874370491, + "grad_norm": 2.2151730060577393, + "learning_rate": 1.1478860569715143e-05, + "loss": 2.495575523376465, + "step": 156800 + }, + { + "epoch": 0.4268953314277937, + "grad_norm": 2.500500202178955, + "learning_rate": 1.1477770205806189e-05, + "loss": 2.487746810913086, + "step": 156820 + }, + { + "epoch": 0.4269497754185382, + "grad_norm": 2.076368808746338, + "learning_rate": 1.1476679841897233e-05, + "loss": 2.4552438735961912, + "step": 156840 + }, + { + "epoch": 0.4270042194092827, + "grad_norm": 2.2058141231536865, + "learning_rate": 1.147558947798828e-05, + "loss": 2.4904556274414062, + "step": 156860 + }, + { + "epoch": 0.42705866340002724, + "grad_norm": 2.215214490890503, + "learning_rate": 1.1474499114079325e-05, + "loss": 2.480366516113281, + "step": 156880 + }, + { + "epoch": 0.42711310739077174, + "grad_norm": 2.1978650093078613, + "learning_rate": 1.1473408750170369e-05, + "loss": 2.5122365951538086, + "step": 156900 + }, + { + "epoch": 0.42716755138151624, + "grad_norm": 2.1957781314849854, + "learning_rate": 1.1472318386261416e-05, + "loss": 2.5170236587524415, + "step": 156920 + }, + { + "epoch": 0.4272219953722608, + "grad_norm": 2.238903760910034, + "learning_rate": 1.147122802235246e-05, + "loss": 2.4406200408935548, + "step": 156940 + }, + { + "epoch": 0.4272764393630053, + "grad_norm": 2.5351123809814453, + "learning_rate": 1.1470137658443506e-05, + "loss": 2.4975648880004884, + "step": 156960 + }, + { + "epoch": 0.42733088335374986, + "grad_norm": 3.821087121963501, + "learning_rate": 1.1469047294534552e-05, + "loss": 2.5000116348266603, + "step": 156980 + }, + { + "epoch": 0.42738532734449436, + "grad_norm": 3.1013169288635254, + "learning_rate": 1.1467956930625596e-05, + "loss": 2.492624282836914, + "step": 157000 + }, + { + "epoch": 0.42743977133523886, + "grad_norm": 2.1499574184417725, + "learning_rate": 1.1466866566716644e-05, + "loss": 2.474365234375, + "step": 157020 + }, + { + "epoch": 0.4274942153259834, + "grad_norm": 2.2646403312683105, + "learning_rate": 1.1465776202807688e-05, + "loss": 2.5048826217651365, + "step": 157040 + }, + { + "epoch": 0.4275486593167279, + "grad_norm": 2.107133626937866, + "learning_rate": 1.1464685838898732e-05, + "loss": 2.4896106719970703, + "step": 157060 + }, + { + "epoch": 0.4276031033074724, + "grad_norm": 2.4268460273742676, + "learning_rate": 1.146359547498978e-05, + "loss": 2.4491451263427733, + "step": 157080 + }, + { + "epoch": 0.427657547298217, + "grad_norm": 2.0223913192749023, + "learning_rate": 1.1462505111080823e-05, + "loss": 2.475972366333008, + "step": 157100 + }, + { + "epoch": 0.4277119912889615, + "grad_norm": 2.2464873790740967, + "learning_rate": 1.1461414747171871e-05, + "loss": 2.508785438537598, + "step": 157120 + }, + { + "epoch": 0.427766435279706, + "grad_norm": 2.1771299839019775, + "learning_rate": 1.1460324383262915e-05, + "loss": 2.480320930480957, + "step": 157140 + }, + { + "epoch": 0.42782087927045054, + "grad_norm": 4.322911739349365, + "learning_rate": 1.1459234019353959e-05, + "loss": 2.485733985900879, + "step": 157160 + }, + { + "epoch": 0.42787532326119504, + "grad_norm": 2.16788387298584, + "learning_rate": 1.1458143655445007e-05, + "loss": 2.4833234786987304, + "step": 157180 + }, + { + "epoch": 0.42792976725193954, + "grad_norm": 2.121278762817383, + "learning_rate": 1.145705329153605e-05, + "loss": 2.4800945281982423, + "step": 157200 + }, + { + "epoch": 0.4279842112426841, + "grad_norm": 2.0645601749420166, + "learning_rate": 1.1455962927627095e-05, + "loss": 2.5022680282592775, + "step": 157220 + }, + { + "epoch": 0.4280386552334286, + "grad_norm": 2.124861478805542, + "learning_rate": 1.1454872563718142e-05, + "loss": 2.4983552932739257, + "step": 157240 + }, + { + "epoch": 0.42809309922417316, + "grad_norm": 2.2496254444122314, + "learning_rate": 1.1453782199809187e-05, + "loss": 2.496807098388672, + "step": 157260 + }, + { + "epoch": 0.42814754321491766, + "grad_norm": 2.017658233642578, + "learning_rate": 1.1452691835900234e-05, + "loss": 2.4962400436401366, + "step": 157280 + }, + { + "epoch": 0.42820198720566216, + "grad_norm": 2.253472089767456, + "learning_rate": 1.1451601471991278e-05, + "loss": 2.4774646759033203, + "step": 157300 + }, + { + "epoch": 0.4282564311964067, + "grad_norm": 2.0635318756103516, + "learning_rate": 1.1450511108082322e-05, + "loss": 2.5317102432250977, + "step": 157320 + }, + { + "epoch": 0.4283108751871512, + "grad_norm": 2.1724746227264404, + "learning_rate": 1.144942074417337e-05, + "loss": 2.5040618896484377, + "step": 157340 + }, + { + "epoch": 0.4283653191778957, + "grad_norm": 2.231058359146118, + "learning_rate": 1.1448330380264414e-05, + "loss": 2.4989446640014648, + "step": 157360 + }, + { + "epoch": 0.4284197631686403, + "grad_norm": 2.1744048595428467, + "learning_rate": 1.1447240016355458e-05, + "loss": 2.504988670349121, + "step": 157380 + }, + { + "epoch": 0.4284742071593848, + "grad_norm": 2.399549722671509, + "learning_rate": 1.1446149652446505e-05, + "loss": 2.499651336669922, + "step": 157400 + }, + { + "epoch": 0.4285286511501293, + "grad_norm": 2.115018844604492, + "learning_rate": 1.144505928853755e-05, + "loss": 2.4708837509155273, + "step": 157420 + }, + { + "epoch": 0.42858309514087384, + "grad_norm": 2.431851863861084, + "learning_rate": 1.1443968924628597e-05, + "loss": 2.515063667297363, + "step": 157440 + }, + { + "epoch": 0.42863753913161834, + "grad_norm": 2.116395950317383, + "learning_rate": 1.1442878560719641e-05, + "loss": 2.483852195739746, + "step": 157460 + }, + { + "epoch": 0.4286919831223629, + "grad_norm": 2.239318609237671, + "learning_rate": 1.1441788196810685e-05, + "loss": 2.4865373611450194, + "step": 157480 + }, + { + "epoch": 0.4287464271131074, + "grad_norm": 2.793046712875366, + "learning_rate": 1.1440697832901733e-05, + "loss": 2.4758974075317384, + "step": 157500 + }, + { + "epoch": 0.4288008711038519, + "grad_norm": 2.286654233932495, + "learning_rate": 1.1439607468992777e-05, + "loss": 2.4998199462890627, + "step": 157520 + }, + { + "epoch": 0.42885531509459646, + "grad_norm": 2.1486690044403076, + "learning_rate": 1.1438517105083821e-05, + "loss": 2.528722381591797, + "step": 157540 + }, + { + "epoch": 0.42890975908534096, + "grad_norm": 2.225609064102173, + "learning_rate": 1.1437426741174869e-05, + "loss": 2.499673843383789, + "step": 157560 + }, + { + "epoch": 0.42896420307608546, + "grad_norm": 2.1211843490600586, + "learning_rate": 1.1436336377265913e-05, + "loss": 2.5163536071777344, + "step": 157580 + }, + { + "epoch": 0.42901864706683, + "grad_norm": 2.155043363571167, + "learning_rate": 1.143524601335696e-05, + "loss": 2.4893035888671875, + "step": 157600 + }, + { + "epoch": 0.4290730910575745, + "grad_norm": 2.092075824737549, + "learning_rate": 1.1434155649448004e-05, + "loss": 2.4669193267822265, + "step": 157620 + }, + { + "epoch": 0.429127535048319, + "grad_norm": 6.56940221786499, + "learning_rate": 1.1433065285539048e-05, + "loss": 2.489856147766113, + "step": 157640 + }, + { + "epoch": 0.4291819790390636, + "grad_norm": 2.194951057434082, + "learning_rate": 1.1431974921630096e-05, + "loss": 2.4840511322021483, + "step": 157660 + }, + { + "epoch": 0.4292364230298081, + "grad_norm": 2.2298004627227783, + "learning_rate": 1.143088455772114e-05, + "loss": 2.5154144287109377, + "step": 157680 + }, + { + "epoch": 0.4292908670205526, + "grad_norm": 2.1600282192230225, + "learning_rate": 1.1429794193812184e-05, + "loss": 2.4993492126464845, + "step": 157700 + }, + { + "epoch": 0.42934531101129714, + "grad_norm": 2.0982887744903564, + "learning_rate": 1.1428703829903232e-05, + "loss": 2.4923307418823244, + "step": 157720 + }, + { + "epoch": 0.42939975500204164, + "grad_norm": 2.0914230346679688, + "learning_rate": 1.1427613465994276e-05, + "loss": 2.4781282424926756, + "step": 157740 + }, + { + "epoch": 0.4294541989927862, + "grad_norm": 2.4307479858398438, + "learning_rate": 1.1426523102085323e-05, + "loss": 2.4977880477905274, + "step": 157760 + }, + { + "epoch": 0.4295086429835307, + "grad_norm": 2.265505790710449, + "learning_rate": 1.1425432738176367e-05, + "loss": 2.422154426574707, + "step": 157780 + }, + { + "epoch": 0.4295630869742752, + "grad_norm": 2.0046513080596924, + "learning_rate": 1.1424342374267411e-05, + "loss": 2.5258983612060546, + "step": 157800 + }, + { + "epoch": 0.42961753096501976, + "grad_norm": 2.128394842147827, + "learning_rate": 1.1423252010358459e-05, + "loss": 2.4929574966430663, + "step": 157820 + }, + { + "epoch": 0.42967197495576426, + "grad_norm": 2.0722804069519043, + "learning_rate": 1.1422161646449503e-05, + "loss": 2.488363265991211, + "step": 157840 + }, + { + "epoch": 0.42972641894650876, + "grad_norm": 2.199389934539795, + "learning_rate": 1.1421071282540547e-05, + "loss": 2.4855730056762697, + "step": 157860 + }, + { + "epoch": 0.4297808629372533, + "grad_norm": 2.1894216537475586, + "learning_rate": 1.1419980918631595e-05, + "loss": 2.49495849609375, + "step": 157880 + }, + { + "epoch": 0.4298353069279978, + "grad_norm": 2.1890594959259033, + "learning_rate": 1.1418890554722639e-05, + "loss": 2.495749092102051, + "step": 157900 + }, + { + "epoch": 0.4298897509187423, + "grad_norm": 2.250363826751709, + "learning_rate": 1.1417800190813686e-05, + "loss": 2.482843780517578, + "step": 157920 + }, + { + "epoch": 0.4299441949094869, + "grad_norm": 2.1961402893066406, + "learning_rate": 1.141670982690473e-05, + "loss": 2.510470962524414, + "step": 157940 + }, + { + "epoch": 0.4299986389002314, + "grad_norm": 2.213974714279175, + "learning_rate": 1.1415619462995775e-05, + "loss": 2.495649719238281, + "step": 157960 + }, + { + "epoch": 0.43005308289097594, + "grad_norm": 2.314204454421997, + "learning_rate": 1.1414529099086822e-05, + "loss": 2.485476493835449, + "step": 157980 + }, + { + "epoch": 0.43010752688172044, + "grad_norm": 2.174509286880493, + "learning_rate": 1.1413438735177866e-05, + "loss": 2.517679214477539, + "step": 158000 + }, + { + "epoch": 0.43016197087246494, + "grad_norm": 2.221931219100952, + "learning_rate": 1.141234837126891e-05, + "loss": 2.458515930175781, + "step": 158020 + }, + { + "epoch": 0.4302164148632095, + "grad_norm": 2.2760355472564697, + "learning_rate": 1.1411258007359958e-05, + "loss": 2.4788223266601563, + "step": 158040 + }, + { + "epoch": 0.430270858853954, + "grad_norm": 2.163184404373169, + "learning_rate": 1.1410167643451002e-05, + "loss": 2.4852209091186523, + "step": 158060 + }, + { + "epoch": 0.4303253028446985, + "grad_norm": 2.2358343601226807, + "learning_rate": 1.140907727954205e-05, + "loss": 2.463149833679199, + "step": 158080 + }, + { + "epoch": 0.43037974683544306, + "grad_norm": 2.163548469543457, + "learning_rate": 1.1407986915633094e-05, + "loss": 2.450914764404297, + "step": 158100 + }, + { + "epoch": 0.43043419082618756, + "grad_norm": 2.293095827102661, + "learning_rate": 1.1406896551724138e-05, + "loss": 2.473088264465332, + "step": 158120 + }, + { + "epoch": 0.43048863481693206, + "grad_norm": 2.136812210083008, + "learning_rate": 1.1405806187815185e-05, + "loss": 2.515398406982422, + "step": 158140 + }, + { + "epoch": 0.4305430788076766, + "grad_norm": 2.217395544052124, + "learning_rate": 1.140471582390623e-05, + "loss": 2.5150390625, + "step": 158160 + }, + { + "epoch": 0.4305975227984211, + "grad_norm": 2.123013734817505, + "learning_rate": 1.1403625459997273e-05, + "loss": 2.4927610397338866, + "step": 158180 + }, + { + "epoch": 0.4306519667891656, + "grad_norm": 2.1104018688201904, + "learning_rate": 1.1402535096088321e-05, + "loss": 2.455678176879883, + "step": 158200 + }, + { + "epoch": 0.4307064107799102, + "grad_norm": 2.28128981590271, + "learning_rate": 1.1401444732179365e-05, + "loss": 2.472056198120117, + "step": 158220 + }, + { + "epoch": 0.4307608547706547, + "grad_norm": 2.2519185543060303, + "learning_rate": 1.1400354368270413e-05, + "loss": 2.513665962219238, + "step": 158240 + }, + { + "epoch": 0.43081529876139923, + "grad_norm": 2.126688003540039, + "learning_rate": 1.1399264004361457e-05, + "loss": 2.511431121826172, + "step": 158260 + }, + { + "epoch": 0.43086974275214374, + "grad_norm": 2.106651782989502, + "learning_rate": 1.13981736404525e-05, + "loss": 2.4843914031982424, + "step": 158280 + }, + { + "epoch": 0.43092418674288824, + "grad_norm": 2.0562326908111572, + "learning_rate": 1.1397083276543548e-05, + "loss": 2.4950592041015627, + "step": 158300 + }, + { + "epoch": 0.4309786307336328, + "grad_norm": 2.1020333766937256, + "learning_rate": 1.1395992912634592e-05, + "loss": 2.4924814224243166, + "step": 158320 + }, + { + "epoch": 0.4310330747243773, + "grad_norm": 2.1447110176086426, + "learning_rate": 1.1394902548725636e-05, + "loss": 2.4615734100341795, + "step": 158340 + }, + { + "epoch": 0.4310875187151218, + "grad_norm": 2.2528133392333984, + "learning_rate": 1.1393812184816684e-05, + "loss": 2.5026363372802733, + "step": 158360 + }, + { + "epoch": 0.43114196270586636, + "grad_norm": 2.5693368911743164, + "learning_rate": 1.1392721820907728e-05, + "loss": 2.4813312530517577, + "step": 158380 + }, + { + "epoch": 0.43119640669661086, + "grad_norm": 2.1538450717926025, + "learning_rate": 1.1391631456998776e-05, + "loss": 2.502998924255371, + "step": 158400 + }, + { + "epoch": 0.43125085068735536, + "grad_norm": 2.135004997253418, + "learning_rate": 1.139054109308982e-05, + "loss": 2.5262943267822267, + "step": 158420 + }, + { + "epoch": 0.4313052946780999, + "grad_norm": 2.1348166465759277, + "learning_rate": 1.1389450729180864e-05, + "loss": 2.4735198974609376, + "step": 158440 + }, + { + "epoch": 0.4313597386688444, + "grad_norm": 2.0663928985595703, + "learning_rate": 1.1388360365271911e-05, + "loss": 2.495522117614746, + "step": 158460 + }, + { + "epoch": 0.431414182659589, + "grad_norm": 2.1034977436065674, + "learning_rate": 1.1387270001362955e-05, + "loss": 2.4646739959716797, + "step": 158480 + }, + { + "epoch": 0.4314686266503335, + "grad_norm": 2.136899948120117, + "learning_rate": 1.1386179637454e-05, + "loss": 2.4973690032958986, + "step": 158500 + }, + { + "epoch": 0.431523070641078, + "grad_norm": 2.1811561584472656, + "learning_rate": 1.1385089273545047e-05, + "loss": 2.488460922241211, + "step": 158520 + }, + { + "epoch": 0.43157751463182253, + "grad_norm": 2.287851572036743, + "learning_rate": 1.1383998909636091e-05, + "loss": 2.5008638381958006, + "step": 158540 + }, + { + "epoch": 0.43163195862256704, + "grad_norm": 2.0858054161071777, + "learning_rate": 1.1382908545727139e-05, + "loss": 2.4926328659057617, + "step": 158560 + }, + { + "epoch": 0.43168640261331154, + "grad_norm": 2.241292953491211, + "learning_rate": 1.1381818181818183e-05, + "loss": 2.4614080429077148, + "step": 158580 + }, + { + "epoch": 0.4317408466040561, + "grad_norm": 2.1492843627929688, + "learning_rate": 1.1380727817909227e-05, + "loss": 2.493235397338867, + "step": 158600 + }, + { + "epoch": 0.4317952905948006, + "grad_norm": 2.367809295654297, + "learning_rate": 1.1379637454000274e-05, + "loss": 2.509945297241211, + "step": 158620 + }, + { + "epoch": 0.4318497345855451, + "grad_norm": 2.2011985778808594, + "learning_rate": 1.1378547090091319e-05, + "loss": 2.4991386413574217, + "step": 158640 + }, + { + "epoch": 0.43190417857628965, + "grad_norm": 2.2731130123138428, + "learning_rate": 1.1377456726182363e-05, + "loss": 2.481618881225586, + "step": 158660 + }, + { + "epoch": 0.43195862256703416, + "grad_norm": 2.111321449279785, + "learning_rate": 1.137636636227341e-05, + "loss": 2.507145309448242, + "step": 158680 + }, + { + "epoch": 0.43201306655777866, + "grad_norm": 2.1435046195983887, + "learning_rate": 1.1375275998364454e-05, + "loss": 2.487934684753418, + "step": 158700 + }, + { + "epoch": 0.4320675105485232, + "grad_norm": 2.126516819000244, + "learning_rate": 1.1374185634455502e-05, + "loss": 2.5025068283081056, + "step": 158720 + }, + { + "epoch": 0.4321219545392677, + "grad_norm": 2.748525381088257, + "learning_rate": 1.1373095270546546e-05, + "loss": 2.4614179611206053, + "step": 158740 + }, + { + "epoch": 0.4321763985300123, + "grad_norm": 2.1338775157928467, + "learning_rate": 1.137200490663759e-05, + "loss": 2.5204288482666017, + "step": 158760 + }, + { + "epoch": 0.4322308425207568, + "grad_norm": 2.109926700592041, + "learning_rate": 1.1370914542728638e-05, + "loss": 2.5135074615478517, + "step": 158780 + }, + { + "epoch": 0.4322852865115013, + "grad_norm": 2.179535388946533, + "learning_rate": 1.1369824178819682e-05, + "loss": 2.489653205871582, + "step": 158800 + }, + { + "epoch": 0.43233973050224583, + "grad_norm": 2.0768415927886963, + "learning_rate": 1.1368733814910729e-05, + "loss": 2.502249336242676, + "step": 158820 + }, + { + "epoch": 0.43239417449299034, + "grad_norm": 2.329068422317505, + "learning_rate": 1.1367643451001773e-05, + "loss": 2.4899391174316405, + "step": 158840 + }, + { + "epoch": 0.43244861848373484, + "grad_norm": 2.165601968765259, + "learning_rate": 1.1366553087092817e-05, + "loss": 2.4918615341186525, + "step": 158860 + }, + { + "epoch": 0.4325030624744794, + "grad_norm": 2.2278060913085938, + "learning_rate": 1.1365462723183865e-05, + "loss": 2.4859800338745117, + "step": 158880 + }, + { + "epoch": 0.4325575064652239, + "grad_norm": 2.186858892440796, + "learning_rate": 1.1364372359274909e-05, + "loss": 2.468840980529785, + "step": 158900 + }, + { + "epoch": 0.4326119504559684, + "grad_norm": 2.247884511947632, + "learning_rate": 1.1363281995365953e-05, + "loss": 2.4745338439941404, + "step": 158920 + }, + { + "epoch": 0.43266639444671295, + "grad_norm": 2.150442600250244, + "learning_rate": 1.1362191631457e-05, + "loss": 2.5122570037841796, + "step": 158940 + }, + { + "epoch": 0.43272083843745746, + "grad_norm": 2.1050190925598145, + "learning_rate": 1.1361101267548045e-05, + "loss": 2.5043283462524415, + "step": 158960 + }, + { + "epoch": 0.432775282428202, + "grad_norm": 2.141294479370117, + "learning_rate": 1.1360010903639092e-05, + "loss": 2.5217006683349608, + "step": 158980 + }, + { + "epoch": 0.4328297264189465, + "grad_norm": 2.106250524520874, + "learning_rate": 1.1358920539730136e-05, + "loss": 2.465514373779297, + "step": 159000 + }, + { + "epoch": 0.432884170409691, + "grad_norm": 2.173741102218628, + "learning_rate": 1.135783017582118e-05, + "loss": 2.4990697860717774, + "step": 159020 + }, + { + "epoch": 0.4329386144004356, + "grad_norm": 2.6162643432617188, + "learning_rate": 1.1356739811912228e-05, + "loss": 2.5147361755371094, + "step": 159040 + }, + { + "epoch": 0.4329930583911801, + "grad_norm": 2.2366793155670166, + "learning_rate": 1.1355649448003272e-05, + "loss": 2.5140796661376954, + "step": 159060 + }, + { + "epoch": 0.4330475023819246, + "grad_norm": 2.2193689346313477, + "learning_rate": 1.1354559084094316e-05, + "loss": 2.4890989303588866, + "step": 159080 + }, + { + "epoch": 0.43310194637266913, + "grad_norm": 2.0457189083099365, + "learning_rate": 1.1353468720185364e-05, + "loss": 2.4834667205810548, + "step": 159100 + }, + { + "epoch": 0.43315639036341363, + "grad_norm": 2.1604719161987305, + "learning_rate": 1.1352378356276408e-05, + "loss": 2.4956146240234376, + "step": 159120 + }, + { + "epoch": 0.43321083435415814, + "grad_norm": 2.220081090927124, + "learning_rate": 1.1351287992367455e-05, + "loss": 2.5213907241821287, + "step": 159140 + }, + { + "epoch": 0.4332652783449027, + "grad_norm": 2.1223175525665283, + "learning_rate": 1.13501976284585e-05, + "loss": 2.488618278503418, + "step": 159160 + }, + { + "epoch": 0.4333197223356472, + "grad_norm": 2.42989182472229, + "learning_rate": 1.1349107264549544e-05, + "loss": 2.509867858886719, + "step": 159180 + }, + { + "epoch": 0.43337416632639175, + "grad_norm": 2.0677437782287598, + "learning_rate": 1.1348016900640591e-05, + "loss": 2.504931640625, + "step": 159200 + }, + { + "epoch": 0.43342861031713625, + "grad_norm": 2.2278647422790527, + "learning_rate": 1.1346926536731635e-05, + "loss": 2.492384910583496, + "step": 159220 + }, + { + "epoch": 0.43348305430788076, + "grad_norm": 2.080503463745117, + "learning_rate": 1.134583617282268e-05, + "loss": 2.4817127227783202, + "step": 159240 + }, + { + "epoch": 0.4335374982986253, + "grad_norm": 2.1143500804901123, + "learning_rate": 1.1344745808913727e-05, + "loss": 2.491969680786133, + "step": 159260 + }, + { + "epoch": 0.4335919422893698, + "grad_norm": 2.161651849746704, + "learning_rate": 1.1343655445004771e-05, + "loss": 2.5031511306762697, + "step": 159280 + }, + { + "epoch": 0.4336463862801143, + "grad_norm": 2.0472218990325928, + "learning_rate": 1.1342565081095818e-05, + "loss": 2.4972442626953124, + "step": 159300 + }, + { + "epoch": 0.4337008302708589, + "grad_norm": 2.225674629211426, + "learning_rate": 1.1341474717186862e-05, + "loss": 2.5106040954589846, + "step": 159320 + }, + { + "epoch": 0.4337552742616034, + "grad_norm": 2.1827878952026367, + "learning_rate": 1.1340384353277907e-05, + "loss": 2.4816761016845703, + "step": 159340 + }, + { + "epoch": 0.4338097182523479, + "grad_norm": 2.203350305557251, + "learning_rate": 1.1339293989368954e-05, + "loss": 2.4893011093139648, + "step": 159360 + }, + { + "epoch": 0.43386416224309243, + "grad_norm": 2.2066454887390137, + "learning_rate": 1.1338203625459998e-05, + "loss": 2.4759140014648438, + "step": 159380 + }, + { + "epoch": 0.43391860623383693, + "grad_norm": 2.112713575363159, + "learning_rate": 1.1337113261551042e-05, + "loss": 2.477926826477051, + "step": 159400 + }, + { + "epoch": 0.43397305022458144, + "grad_norm": 2.147571086883545, + "learning_rate": 1.133602289764209e-05, + "loss": 2.464390754699707, + "step": 159420 + }, + { + "epoch": 0.434027494215326, + "grad_norm": 2.2218754291534424, + "learning_rate": 1.1334932533733134e-05, + "loss": 2.5027301788330076, + "step": 159440 + }, + { + "epoch": 0.4340819382060705, + "grad_norm": 2.3957958221435547, + "learning_rate": 1.1333842169824181e-05, + "loss": 2.4848302841186523, + "step": 159460 + }, + { + "epoch": 0.43413638219681505, + "grad_norm": 2.1112043857574463, + "learning_rate": 1.1332751805915226e-05, + "loss": 2.4803232192993163, + "step": 159480 + }, + { + "epoch": 0.43419082618755955, + "grad_norm": 2.2261812686920166, + "learning_rate": 1.133166144200627e-05, + "loss": 2.4676427841186523, + "step": 159500 + }, + { + "epoch": 0.43424527017830405, + "grad_norm": 2.1903493404388428, + "learning_rate": 1.1330571078097317e-05, + "loss": 2.4620227813720703, + "step": 159520 + }, + { + "epoch": 0.4342997141690486, + "grad_norm": 2.2981064319610596, + "learning_rate": 1.1329480714188361e-05, + "loss": 2.495198440551758, + "step": 159540 + }, + { + "epoch": 0.4343541581597931, + "grad_norm": 2.2798595428466797, + "learning_rate": 1.1328390350279405e-05, + "loss": 2.502900314331055, + "step": 159560 + }, + { + "epoch": 0.4344086021505376, + "grad_norm": 2.232269287109375, + "learning_rate": 1.1327299986370453e-05, + "loss": 2.493956184387207, + "step": 159580 + }, + { + "epoch": 0.43446304614128217, + "grad_norm": 2.197153091430664, + "learning_rate": 1.1326209622461497e-05, + "loss": 2.483063316345215, + "step": 159600 + }, + { + "epoch": 0.4345174901320267, + "grad_norm": 2.091092586517334, + "learning_rate": 1.1325119258552545e-05, + "loss": 2.4690475463867188, + "step": 159620 + }, + { + "epoch": 0.4345719341227712, + "grad_norm": 2.1273181438446045, + "learning_rate": 1.1324028894643589e-05, + "loss": 2.47259635925293, + "step": 159640 + }, + { + "epoch": 0.43462637811351573, + "grad_norm": 2.234471559524536, + "learning_rate": 1.1322938530734633e-05, + "loss": 2.5130102157592775, + "step": 159660 + }, + { + "epoch": 0.43468082210426023, + "grad_norm": 2.0385501384735107, + "learning_rate": 1.132184816682568e-05, + "loss": 2.505122756958008, + "step": 159680 + }, + { + "epoch": 0.4347352660950048, + "grad_norm": 3.489593744277954, + "learning_rate": 1.1320757802916724e-05, + "loss": 2.4861608505249024, + "step": 159700 + }, + { + "epoch": 0.4347897100857493, + "grad_norm": 2.125333070755005, + "learning_rate": 1.1319667439007769e-05, + "loss": 2.5051212310791016, + "step": 159720 + }, + { + "epoch": 0.4348441540764938, + "grad_norm": 2.1980092525482178, + "learning_rate": 1.1318577075098816e-05, + "loss": 2.4817705154418945, + "step": 159740 + }, + { + "epoch": 0.43489859806723835, + "grad_norm": 2.0302488803863525, + "learning_rate": 1.131748671118986e-05, + "loss": 2.4762134552001953, + "step": 159760 + }, + { + "epoch": 0.43495304205798285, + "grad_norm": 2.1116795539855957, + "learning_rate": 1.1316396347280908e-05, + "loss": 2.5042041778564452, + "step": 159780 + }, + { + "epoch": 0.43500748604872735, + "grad_norm": 2.296872615814209, + "learning_rate": 1.1315305983371952e-05, + "loss": 2.4777254104614257, + "step": 159800 + }, + { + "epoch": 0.4350619300394719, + "grad_norm": 2.0861470699310303, + "learning_rate": 1.1314215619462996e-05, + "loss": 2.4609302520751952, + "step": 159820 + }, + { + "epoch": 0.4351163740302164, + "grad_norm": 2.173023223876953, + "learning_rate": 1.1313125255554043e-05, + "loss": 2.504397392272949, + "step": 159840 + }, + { + "epoch": 0.4351708180209609, + "grad_norm": 2.1729683876037598, + "learning_rate": 1.1312034891645087e-05, + "loss": 2.495430755615234, + "step": 159860 + }, + { + "epoch": 0.43522526201170547, + "grad_norm": 2.107975482940674, + "learning_rate": 1.1310944527736132e-05, + "loss": 2.479270172119141, + "step": 159880 + }, + { + "epoch": 0.43527970600245, + "grad_norm": 2.1566874980926514, + "learning_rate": 1.1309854163827179e-05, + "loss": 2.5081710815429688, + "step": 159900 + }, + { + "epoch": 0.4353341499931945, + "grad_norm": 2.216338872909546, + "learning_rate": 1.1308763799918223e-05, + "loss": 2.4867826461791993, + "step": 159920 + }, + { + "epoch": 0.43538859398393903, + "grad_norm": 2.2181670665740967, + "learning_rate": 1.130767343600927e-05, + "loss": 2.491602897644043, + "step": 159940 + }, + { + "epoch": 0.43544303797468353, + "grad_norm": 2.1998395919799805, + "learning_rate": 1.1306583072100315e-05, + "loss": 2.4646108627319334, + "step": 159960 + }, + { + "epoch": 0.4354974819654281, + "grad_norm": 2.1738131046295166, + "learning_rate": 1.1305492708191359e-05, + "loss": 2.514189910888672, + "step": 159980 + }, + { + "epoch": 0.4355519259561726, + "grad_norm": 2.074479818344116, + "learning_rate": 1.1304402344282406e-05, + "loss": 2.5021736145019533, + "step": 160000 + }, + { + "epoch": 0.4356063699469171, + "grad_norm": 2.210984468460083, + "learning_rate": 1.130331198037345e-05, + "loss": 2.495810127258301, + "step": 160020 + }, + { + "epoch": 0.43566081393766165, + "grad_norm": 2.1778135299682617, + "learning_rate": 1.1302221616464495e-05, + "loss": 2.5206069946289062, + "step": 160040 + }, + { + "epoch": 0.43571525792840615, + "grad_norm": 2.021679401397705, + "learning_rate": 1.1301131252555542e-05, + "loss": 2.4973064422607423, + "step": 160060 + }, + { + "epoch": 0.43576970191915065, + "grad_norm": 2.112868309020996, + "learning_rate": 1.1300040888646586e-05, + "loss": 2.481869125366211, + "step": 160080 + }, + { + "epoch": 0.4358241459098952, + "grad_norm": 2.2606306076049805, + "learning_rate": 1.1298950524737634e-05, + "loss": 2.4912752151489257, + "step": 160100 + }, + { + "epoch": 0.4358785899006397, + "grad_norm": 2.2336158752441406, + "learning_rate": 1.1297860160828678e-05, + "loss": 2.506151580810547, + "step": 160120 + }, + { + "epoch": 0.4359330338913842, + "grad_norm": 2.246091842651367, + "learning_rate": 1.1296769796919722e-05, + "loss": 2.477412986755371, + "step": 160140 + }, + { + "epoch": 0.43598747788212877, + "grad_norm": 2.067394733428955, + "learning_rate": 1.129567943301077e-05, + "loss": 2.503101348876953, + "step": 160160 + }, + { + "epoch": 0.4360419218728733, + "grad_norm": 2.1074111461639404, + "learning_rate": 1.1294589069101814e-05, + "loss": 2.4846324920654297, + "step": 160180 + }, + { + "epoch": 0.43609636586361783, + "grad_norm": 2.2519516944885254, + "learning_rate": 1.1293498705192858e-05, + "loss": 2.496114158630371, + "step": 160200 + }, + { + "epoch": 0.43615080985436233, + "grad_norm": 2.0644500255584717, + "learning_rate": 1.1292408341283905e-05, + "loss": 2.4764144897460936, + "step": 160220 + }, + { + "epoch": 0.43620525384510683, + "grad_norm": 2.3991591930389404, + "learning_rate": 1.129131797737495e-05, + "loss": 2.50850772857666, + "step": 160240 + }, + { + "epoch": 0.4362596978358514, + "grad_norm": 2.247809648513794, + "learning_rate": 1.1290227613465997e-05, + "loss": 2.467013359069824, + "step": 160260 + }, + { + "epoch": 0.4363141418265959, + "grad_norm": 2.464421510696411, + "learning_rate": 1.1289137249557041e-05, + "loss": 2.4767478942871093, + "step": 160280 + }, + { + "epoch": 0.4363685858173404, + "grad_norm": 2.077920913696289, + "learning_rate": 1.1288046885648085e-05, + "loss": 2.4889339447021483, + "step": 160300 + }, + { + "epoch": 0.43642302980808495, + "grad_norm": 2.192641019821167, + "learning_rate": 1.1286956521739133e-05, + "loss": 2.506583404541016, + "step": 160320 + }, + { + "epoch": 0.43647747379882945, + "grad_norm": 2.1838386058807373, + "learning_rate": 1.1285866157830177e-05, + "loss": 2.5017303466796874, + "step": 160340 + }, + { + "epoch": 0.43653191778957395, + "grad_norm": 3.0302584171295166, + "learning_rate": 1.1284775793921221e-05, + "loss": 2.4798200607299803, + "step": 160360 + }, + { + "epoch": 0.4365863617803185, + "grad_norm": 2.1648147106170654, + "learning_rate": 1.1283685430012268e-05, + "loss": 2.488508605957031, + "step": 160380 + }, + { + "epoch": 0.436640805771063, + "grad_norm": 2.1546530723571777, + "learning_rate": 1.1282595066103312e-05, + "loss": 2.4784267425537108, + "step": 160400 + }, + { + "epoch": 0.4366952497618075, + "grad_norm": 2.2946763038635254, + "learning_rate": 1.1281504702194358e-05, + "loss": 2.4860544204711914, + "step": 160420 + }, + { + "epoch": 0.43674969375255207, + "grad_norm": 2.2273340225219727, + "learning_rate": 1.1280414338285404e-05, + "loss": 2.4715288162231444, + "step": 160440 + }, + { + "epoch": 0.43680413774329657, + "grad_norm": 2.1282975673675537, + "learning_rate": 1.1279323974376448e-05, + "loss": 2.494350624084473, + "step": 160460 + }, + { + "epoch": 0.43685858173404113, + "grad_norm": 2.345170736312866, + "learning_rate": 1.1278233610467496e-05, + "loss": 2.53485107421875, + "step": 160480 + }, + { + "epoch": 0.43691302572478563, + "grad_norm": 2.139608144760132, + "learning_rate": 1.127714324655854e-05, + "loss": 2.486911964416504, + "step": 160500 + }, + { + "epoch": 0.43696746971553013, + "grad_norm": 2.0755090713500977, + "learning_rate": 1.1276052882649586e-05, + "loss": 2.478571319580078, + "step": 160520 + }, + { + "epoch": 0.4370219137062747, + "grad_norm": 2.1440603733062744, + "learning_rate": 1.1274962518740631e-05, + "loss": 2.4497575759887695, + "step": 160540 + }, + { + "epoch": 0.4370763576970192, + "grad_norm": 2.1363918781280518, + "learning_rate": 1.1273872154831676e-05, + "loss": 2.459183120727539, + "step": 160560 + }, + { + "epoch": 0.4371308016877637, + "grad_norm": 2.1107559204101562, + "learning_rate": 1.1272781790922721e-05, + "loss": 2.4949260711669923, + "step": 160580 + }, + { + "epoch": 0.43718524567850825, + "grad_norm": 2.154468297958374, + "learning_rate": 1.1271691427013767e-05, + "loss": 2.464082717895508, + "step": 160600 + }, + { + "epoch": 0.43723968966925275, + "grad_norm": 2.060265064239502, + "learning_rate": 1.1270601063104811e-05, + "loss": 2.4730024337768555, + "step": 160620 + }, + { + "epoch": 0.43729413365999725, + "grad_norm": 2.0369763374328613, + "learning_rate": 1.1269510699195859e-05, + "loss": 2.491996192932129, + "step": 160640 + }, + { + "epoch": 0.4373485776507418, + "grad_norm": 2.066741704940796, + "learning_rate": 1.1268420335286903e-05, + "loss": 2.5102615356445312, + "step": 160660 + }, + { + "epoch": 0.4374030216414863, + "grad_norm": 2.1240344047546387, + "learning_rate": 1.1267329971377949e-05, + "loss": 2.486189842224121, + "step": 160680 + }, + { + "epoch": 0.43745746563223087, + "grad_norm": 2.0679221153259277, + "learning_rate": 1.1266239607468995e-05, + "loss": 2.48380069732666, + "step": 160700 + }, + { + "epoch": 0.43751190962297537, + "grad_norm": 2.289252996444702, + "learning_rate": 1.1265149243560039e-05, + "loss": 2.511602783203125, + "step": 160720 + }, + { + "epoch": 0.43756635361371987, + "grad_norm": 2.3792426586151123, + "learning_rate": 1.1264058879651084e-05, + "loss": 2.521629524230957, + "step": 160740 + }, + { + "epoch": 0.43762079760446443, + "grad_norm": 2.1523077487945557, + "learning_rate": 1.126296851574213e-05, + "loss": 2.465669059753418, + "step": 160760 + }, + { + "epoch": 0.43767524159520893, + "grad_norm": 2.2306582927703857, + "learning_rate": 1.1261878151833174e-05, + "loss": 2.493271255493164, + "step": 160780 + }, + { + "epoch": 0.43772968558595343, + "grad_norm": 2.1349892616271973, + "learning_rate": 1.1260787787924222e-05, + "loss": 2.5245309829711915, + "step": 160800 + }, + { + "epoch": 0.437784129576698, + "grad_norm": 2.1532554626464844, + "learning_rate": 1.1259697424015266e-05, + "loss": 2.490315055847168, + "step": 160820 + }, + { + "epoch": 0.4378385735674425, + "grad_norm": 2.148638963699341, + "learning_rate": 1.1258607060106312e-05, + "loss": 2.5104461669921876, + "step": 160840 + }, + { + "epoch": 0.437893017558187, + "grad_norm": 2.079467296600342, + "learning_rate": 1.1257516696197358e-05, + "loss": 2.48834228515625, + "step": 160860 + }, + { + "epoch": 0.43794746154893155, + "grad_norm": 2.3269081115722656, + "learning_rate": 1.1256426332288402e-05, + "loss": 2.499139595031738, + "step": 160880 + }, + { + "epoch": 0.43800190553967605, + "grad_norm": 2.183224678039551, + "learning_rate": 1.1255335968379448e-05, + "loss": 2.4958362579345703, + "step": 160900 + }, + { + "epoch": 0.43805634953042055, + "grad_norm": 2.1401991844177246, + "learning_rate": 1.1254245604470493e-05, + "loss": 2.505860137939453, + "step": 160920 + }, + { + "epoch": 0.4381107935211651, + "grad_norm": 2.1130874156951904, + "learning_rate": 1.1253155240561537e-05, + "loss": 2.5100944519042967, + "step": 160940 + }, + { + "epoch": 0.4381652375119096, + "grad_norm": 2.097675085067749, + "learning_rate": 1.1252064876652585e-05, + "loss": 2.494695281982422, + "step": 160960 + }, + { + "epoch": 0.43821968150265417, + "grad_norm": 2.0912601947784424, + "learning_rate": 1.1250974512743629e-05, + "loss": 2.515104293823242, + "step": 160980 + }, + { + "epoch": 0.43827412549339867, + "grad_norm": 2.218971014022827, + "learning_rate": 1.1249884148834675e-05, + "loss": 2.4837787628173826, + "step": 161000 + }, + { + "epoch": 0.43832856948414317, + "grad_norm": 2.03367280960083, + "learning_rate": 1.124879378492572e-05, + "loss": 2.495780181884766, + "step": 161020 + }, + { + "epoch": 0.43838301347488773, + "grad_norm": 2.152123212814331, + "learning_rate": 1.1247703421016765e-05, + "loss": 2.467277717590332, + "step": 161040 + }, + { + "epoch": 0.43843745746563223, + "grad_norm": 2.1796717643737793, + "learning_rate": 1.124661305710781e-05, + "loss": 2.489462471008301, + "step": 161060 + }, + { + "epoch": 0.43849190145637673, + "grad_norm": 2.2023539543151855, + "learning_rate": 1.1245522693198856e-05, + "loss": 2.5208648681640624, + "step": 161080 + }, + { + "epoch": 0.4385463454471213, + "grad_norm": 2.132016658782959, + "learning_rate": 1.12444323292899e-05, + "loss": 2.513546180725098, + "step": 161100 + }, + { + "epoch": 0.4386007894378658, + "grad_norm": 2.0251851081848145, + "learning_rate": 1.1243341965380948e-05, + "loss": 2.4720340728759767, + "step": 161120 + }, + { + "epoch": 0.4386552334286103, + "grad_norm": 2.2201356887817383, + "learning_rate": 1.1242251601471992e-05, + "loss": 2.4722394943237305, + "step": 161140 + }, + { + "epoch": 0.43870967741935485, + "grad_norm": 2.187556028366089, + "learning_rate": 1.1241161237563038e-05, + "loss": 2.4978059768676757, + "step": 161160 + }, + { + "epoch": 0.43876412141009935, + "grad_norm": 2.216644287109375, + "learning_rate": 1.1240070873654084e-05, + "loss": 2.4786855697631838, + "step": 161180 + }, + { + "epoch": 0.4388185654008439, + "grad_norm": 2.4570772647857666, + "learning_rate": 1.1238980509745128e-05, + "loss": 2.5105186462402345, + "step": 161200 + }, + { + "epoch": 0.4388730093915884, + "grad_norm": 2.2399425506591797, + "learning_rate": 1.1237890145836174e-05, + "loss": 2.453704261779785, + "step": 161220 + }, + { + "epoch": 0.4389274533823329, + "grad_norm": 2.1246936321258545, + "learning_rate": 1.123679978192722e-05, + "loss": 2.5480636596679687, + "step": 161240 + }, + { + "epoch": 0.43898189737307747, + "grad_norm": 2.3008477687835693, + "learning_rate": 1.1235709418018264e-05, + "loss": 2.5286178588867188, + "step": 161260 + }, + { + "epoch": 0.43903634136382197, + "grad_norm": 2.3290722370147705, + "learning_rate": 1.1234619054109311e-05, + "loss": 2.4885818481445314, + "step": 161280 + }, + { + "epoch": 0.43909078535456647, + "grad_norm": 2.182244300842285, + "learning_rate": 1.1233528690200355e-05, + "loss": 2.488946723937988, + "step": 161300 + }, + { + "epoch": 0.43914522934531103, + "grad_norm": 2.1532535552978516, + "learning_rate": 1.1232438326291401e-05, + "loss": 2.4880434036254884, + "step": 161320 + }, + { + "epoch": 0.43919967333605553, + "grad_norm": 2.383685350418091, + "learning_rate": 1.1231347962382447e-05, + "loss": 2.494454002380371, + "step": 161340 + }, + { + "epoch": 0.43925411732680003, + "grad_norm": 2.147036552429199, + "learning_rate": 1.1230257598473491e-05, + "loss": 2.517790412902832, + "step": 161360 + }, + { + "epoch": 0.4393085613175446, + "grad_norm": 2.305778741836548, + "learning_rate": 1.1229167234564537e-05, + "loss": 2.496278762817383, + "step": 161380 + }, + { + "epoch": 0.4393630053082891, + "grad_norm": 2.056647539138794, + "learning_rate": 1.1228076870655583e-05, + "loss": 2.5327646255493166, + "step": 161400 + }, + { + "epoch": 0.43941744929903365, + "grad_norm": 2.1873161792755127, + "learning_rate": 1.1226986506746627e-05, + "loss": 2.4690074920654297, + "step": 161420 + }, + { + "epoch": 0.43947189328977815, + "grad_norm": 2.2400057315826416, + "learning_rate": 1.1225896142837673e-05, + "loss": 2.4896728515625, + "step": 161440 + }, + { + "epoch": 0.43952633728052265, + "grad_norm": 2.1946516036987305, + "learning_rate": 1.1224805778928718e-05, + "loss": 2.50906925201416, + "step": 161460 + }, + { + "epoch": 0.4395807812712672, + "grad_norm": 2.155564308166504, + "learning_rate": 1.1223715415019764e-05, + "loss": 2.501277732849121, + "step": 161480 + }, + { + "epoch": 0.4396352252620117, + "grad_norm": 2.0162813663482666, + "learning_rate": 1.122262505111081e-05, + "loss": 2.5069751739501953, + "step": 161500 + }, + { + "epoch": 0.4396896692527562, + "grad_norm": 2.2247629165649414, + "learning_rate": 1.1221534687201854e-05, + "loss": 2.5079458236694334, + "step": 161520 + }, + { + "epoch": 0.43974411324350077, + "grad_norm": 2.1234965324401855, + "learning_rate": 1.12204443232929e-05, + "loss": 2.499087905883789, + "step": 161540 + }, + { + "epoch": 0.43979855723424527, + "grad_norm": 2.2457916736602783, + "learning_rate": 1.1219353959383946e-05, + "loss": 2.4845937728881835, + "step": 161560 + }, + { + "epoch": 0.43985300122498977, + "grad_norm": 2.469404458999634, + "learning_rate": 1.121826359547499e-05, + "loss": 2.506141471862793, + "step": 161580 + }, + { + "epoch": 0.4399074452157343, + "grad_norm": 2.1861987113952637, + "learning_rate": 1.1217173231566036e-05, + "loss": 2.488209342956543, + "step": 161600 + }, + { + "epoch": 0.43996188920647883, + "grad_norm": 2.1221649646759033, + "learning_rate": 1.1216082867657081e-05, + "loss": 2.498849868774414, + "step": 161620 + }, + { + "epoch": 0.44001633319722333, + "grad_norm": 2.236861228942871, + "learning_rate": 1.1214992503748127e-05, + "loss": 2.478973960876465, + "step": 161640 + }, + { + "epoch": 0.4400707771879679, + "grad_norm": 2.1072745323181152, + "learning_rate": 1.1213902139839173e-05, + "loss": 2.4432292938232423, + "step": 161660 + }, + { + "epoch": 0.4401252211787124, + "grad_norm": 2.2941858768463135, + "learning_rate": 1.1212811775930217e-05, + "loss": 2.459722900390625, + "step": 161680 + }, + { + "epoch": 0.44017966516945695, + "grad_norm": 2.1727285385131836, + "learning_rate": 1.1211721412021263e-05, + "loss": 2.475299835205078, + "step": 161700 + }, + { + "epoch": 0.44023410916020145, + "grad_norm": 2.0836780071258545, + "learning_rate": 1.1210631048112309e-05, + "loss": 2.4520498275756837, + "step": 161720 + }, + { + "epoch": 0.44028855315094595, + "grad_norm": 2.145320177078247, + "learning_rate": 1.1209540684203353e-05, + "loss": 2.4561161041259765, + "step": 161740 + }, + { + "epoch": 0.4403429971416905, + "grad_norm": 2.1241328716278076, + "learning_rate": 1.1208450320294399e-05, + "loss": 2.4890281677246096, + "step": 161760 + }, + { + "epoch": 0.440397441132435, + "grad_norm": 2.121783494949341, + "learning_rate": 1.1207359956385445e-05, + "loss": 2.483450508117676, + "step": 161780 + }, + { + "epoch": 0.4404518851231795, + "grad_norm": 3.169316530227661, + "learning_rate": 1.120626959247649e-05, + "loss": 2.4866153717041017, + "step": 161800 + }, + { + "epoch": 0.44050632911392407, + "grad_norm": 2.133765459060669, + "learning_rate": 1.1205179228567536e-05, + "loss": 2.436493492126465, + "step": 161820 + }, + { + "epoch": 0.44056077310466857, + "grad_norm": 2.1303467750549316, + "learning_rate": 1.120408886465858e-05, + "loss": 2.484604072570801, + "step": 161840 + }, + { + "epoch": 0.44061521709541307, + "grad_norm": 2.221280813217163, + "learning_rate": 1.1202998500749626e-05, + "loss": 2.5074846267700197, + "step": 161860 + }, + { + "epoch": 0.4406696610861576, + "grad_norm": 2.158421754837036, + "learning_rate": 1.1201908136840672e-05, + "loss": 2.4875049591064453, + "step": 161880 + }, + { + "epoch": 0.44072410507690213, + "grad_norm": 2.1589629650115967, + "learning_rate": 1.1200817772931716e-05, + "loss": 2.483915328979492, + "step": 161900 + }, + { + "epoch": 0.4407785490676467, + "grad_norm": 2.0968177318573, + "learning_rate": 1.1199727409022762e-05, + "loss": 2.491790771484375, + "step": 161920 + }, + { + "epoch": 0.4408329930583912, + "grad_norm": 2.3504793643951416, + "learning_rate": 1.1198637045113808e-05, + "loss": 2.497681427001953, + "step": 161940 + }, + { + "epoch": 0.4408874370491357, + "grad_norm": 2.186239004135132, + "learning_rate": 1.1197546681204853e-05, + "loss": 2.473776626586914, + "step": 161960 + }, + { + "epoch": 0.44094188103988025, + "grad_norm": 2.1976723670959473, + "learning_rate": 1.11964563172959e-05, + "loss": 2.480131721496582, + "step": 161980 + }, + { + "epoch": 0.44099632503062475, + "grad_norm": 2.0758206844329834, + "learning_rate": 1.1195365953386943e-05, + "loss": 2.5295658111572266, + "step": 162000 + }, + { + "epoch": 0.44105076902136925, + "grad_norm": 2.180476665496826, + "learning_rate": 1.1194275589477989e-05, + "loss": 2.491983985900879, + "step": 162020 + }, + { + "epoch": 0.4411052130121138, + "grad_norm": 3.68941330909729, + "learning_rate": 1.1193185225569035e-05, + "loss": 2.4520357131958006, + "step": 162040 + }, + { + "epoch": 0.4411596570028583, + "grad_norm": 2.222081422805786, + "learning_rate": 1.1192094861660079e-05, + "loss": 2.5229114532470702, + "step": 162060 + }, + { + "epoch": 0.4412141009936028, + "grad_norm": 2.342698812484741, + "learning_rate": 1.1191004497751125e-05, + "loss": 2.4623960494995116, + "step": 162080 + }, + { + "epoch": 0.44126854498434737, + "grad_norm": 2.15789532661438, + "learning_rate": 1.118991413384217e-05, + "loss": 2.4856422424316404, + "step": 162100 + }, + { + "epoch": 0.44132298897509187, + "grad_norm": 2.2960357666015625, + "learning_rate": 1.1188823769933216e-05, + "loss": 2.496502494812012, + "step": 162120 + }, + { + "epoch": 0.44137743296583637, + "grad_norm": 2.2520220279693604, + "learning_rate": 1.1187733406024262e-05, + "loss": 2.490559196472168, + "step": 162140 + }, + { + "epoch": 0.4414318769565809, + "grad_norm": 2.377230644226074, + "learning_rate": 1.1186643042115306e-05, + "loss": 2.503891372680664, + "step": 162160 + }, + { + "epoch": 0.44148632094732543, + "grad_norm": 2.2185816764831543, + "learning_rate": 1.1185552678206352e-05, + "loss": 2.468430519104004, + "step": 162180 + }, + { + "epoch": 0.44154076493807, + "grad_norm": 2.1779229640960693, + "learning_rate": 1.1184462314297398e-05, + "loss": 2.5145254135131836, + "step": 162200 + }, + { + "epoch": 0.4415952089288145, + "grad_norm": 2.169428586959839, + "learning_rate": 1.1183371950388444e-05, + "loss": 2.4681112289428713, + "step": 162220 + }, + { + "epoch": 0.441649652919559, + "grad_norm": 2.110477924346924, + "learning_rate": 1.1182281586479488e-05, + "loss": 2.489320182800293, + "step": 162240 + }, + { + "epoch": 0.44170409691030355, + "grad_norm": 2.115403652191162, + "learning_rate": 1.1181191222570534e-05, + "loss": 2.505497360229492, + "step": 162260 + }, + { + "epoch": 0.44175854090104805, + "grad_norm": 2.2568254470825195, + "learning_rate": 1.118010085866158e-05, + "loss": 2.482432556152344, + "step": 162280 + }, + { + "epoch": 0.44181298489179255, + "grad_norm": 2.2014307975769043, + "learning_rate": 1.1179010494752625e-05, + "loss": 2.481014060974121, + "step": 162300 + }, + { + "epoch": 0.4418674288825371, + "grad_norm": 2.1581952571868896, + "learning_rate": 1.117792013084367e-05, + "loss": 2.4876161575317384, + "step": 162320 + }, + { + "epoch": 0.4419218728732816, + "grad_norm": 2.2075746059417725, + "learning_rate": 1.1176829766934715e-05, + "loss": 2.454772186279297, + "step": 162340 + }, + { + "epoch": 0.4419763168640261, + "grad_norm": 2.0823612213134766, + "learning_rate": 1.1175739403025761e-05, + "loss": 2.459603500366211, + "step": 162360 + }, + { + "epoch": 0.44203076085477067, + "grad_norm": 2.100947618484497, + "learning_rate": 1.1174649039116807e-05, + "loss": 2.48614501953125, + "step": 162380 + }, + { + "epoch": 0.44208520484551517, + "grad_norm": 2.035250186920166, + "learning_rate": 1.1173558675207851e-05, + "loss": 2.480098342895508, + "step": 162400 + }, + { + "epoch": 0.4421396488362597, + "grad_norm": 2.2406890392303467, + "learning_rate": 1.1172468311298897e-05, + "loss": 2.4722618103027343, + "step": 162420 + }, + { + "epoch": 0.4421940928270042, + "grad_norm": 2.8034558296203613, + "learning_rate": 1.1171377947389943e-05, + "loss": 2.4931116104125977, + "step": 162440 + }, + { + "epoch": 0.4422485368177487, + "grad_norm": 2.0901756286621094, + "learning_rate": 1.1170287583480987e-05, + "loss": 2.4939027786254884, + "step": 162460 + }, + { + "epoch": 0.4423029808084933, + "grad_norm": 2.3244848251342773, + "learning_rate": 1.1169197219572033e-05, + "loss": 2.509120750427246, + "step": 162480 + }, + { + "epoch": 0.4423574247992378, + "grad_norm": 2.230862617492676, + "learning_rate": 1.1168106855663078e-05, + "loss": 2.478653144836426, + "step": 162500 + }, + { + "epoch": 0.4424118687899823, + "grad_norm": 2.3566250801086426, + "learning_rate": 1.1167016491754124e-05, + "loss": 2.4760189056396484, + "step": 162520 + }, + { + "epoch": 0.44246631278072684, + "grad_norm": 2.1645679473876953, + "learning_rate": 1.116592612784517e-05, + "loss": 2.5025398254394533, + "step": 162540 + }, + { + "epoch": 0.44252075677147135, + "grad_norm": 2.229580879211426, + "learning_rate": 1.1164835763936214e-05, + "loss": 2.498312568664551, + "step": 162560 + }, + { + "epoch": 0.44257520076221585, + "grad_norm": 2.1737060546875, + "learning_rate": 1.116374540002726e-05, + "loss": 2.488530731201172, + "step": 162580 + }, + { + "epoch": 0.4426296447529604, + "grad_norm": 2.038771152496338, + "learning_rate": 1.1162655036118306e-05, + "loss": 2.489912414550781, + "step": 162600 + }, + { + "epoch": 0.4426840887437049, + "grad_norm": 2.1949074268341064, + "learning_rate": 1.116156467220935e-05, + "loss": 2.4954355239868162, + "step": 162620 + }, + { + "epoch": 0.4427385327344494, + "grad_norm": 2.2738754749298096, + "learning_rate": 1.1160474308300396e-05, + "loss": 2.469446563720703, + "step": 162640 + }, + { + "epoch": 0.44279297672519397, + "grad_norm": 2.04388689994812, + "learning_rate": 1.1159383944391441e-05, + "loss": 2.4921489715576173, + "step": 162660 + }, + { + "epoch": 0.44284742071593847, + "grad_norm": 2.0772202014923096, + "learning_rate": 1.1158293580482487e-05, + "loss": 2.480204200744629, + "step": 162680 + }, + { + "epoch": 0.442901864706683, + "grad_norm": 2.2847437858581543, + "learning_rate": 1.1157203216573533e-05, + "loss": 2.4844383239746093, + "step": 162700 + }, + { + "epoch": 0.4429563086974275, + "grad_norm": 2.2587318420410156, + "learning_rate": 1.1156112852664577e-05, + "loss": 2.4651086807250975, + "step": 162720 + }, + { + "epoch": 0.443010752688172, + "grad_norm": 2.130218982696533, + "learning_rate": 1.1155022488755623e-05, + "loss": 2.459336280822754, + "step": 162740 + }, + { + "epoch": 0.4430651966789166, + "grad_norm": 2.5711493492126465, + "learning_rate": 1.1153932124846669e-05, + "loss": 2.490548324584961, + "step": 162760 + }, + { + "epoch": 0.4431196406696611, + "grad_norm": 2.211967945098877, + "learning_rate": 1.1152841760937713e-05, + "loss": 2.485696792602539, + "step": 162780 + }, + { + "epoch": 0.4431740846604056, + "grad_norm": 2.141185998916626, + "learning_rate": 1.1151751397028759e-05, + "loss": 2.494390678405762, + "step": 162800 + }, + { + "epoch": 0.44322852865115014, + "grad_norm": 2.283886671066284, + "learning_rate": 1.1150661033119805e-05, + "loss": 2.485952377319336, + "step": 162820 + }, + { + "epoch": 0.44328297264189465, + "grad_norm": 2.4424428939819336, + "learning_rate": 1.114957066921085e-05, + "loss": 2.492901611328125, + "step": 162840 + }, + { + "epoch": 0.44333741663263915, + "grad_norm": 2.1008307933807373, + "learning_rate": 1.1148480305301896e-05, + "loss": 2.4812458038330076, + "step": 162860 + }, + { + "epoch": 0.4433918606233837, + "grad_norm": 2.084416389465332, + "learning_rate": 1.114738994139294e-05, + "loss": 2.516794776916504, + "step": 162880 + }, + { + "epoch": 0.4434463046141282, + "grad_norm": 2.145580530166626, + "learning_rate": 1.1146299577483986e-05, + "loss": 2.472391128540039, + "step": 162900 + }, + { + "epoch": 0.44350074860487276, + "grad_norm": 2.117245674133301, + "learning_rate": 1.1145209213575032e-05, + "loss": 2.4781455993652344, + "step": 162920 + }, + { + "epoch": 0.44355519259561726, + "grad_norm": 2.149646282196045, + "learning_rate": 1.1144118849666076e-05, + "loss": 2.526658630371094, + "step": 162940 + }, + { + "epoch": 0.44360963658636177, + "grad_norm": 2.203385591506958, + "learning_rate": 1.1143028485757122e-05, + "loss": 2.4716989517211916, + "step": 162960 + }, + { + "epoch": 0.4436640805771063, + "grad_norm": 2.2552733421325684, + "learning_rate": 1.1141938121848168e-05, + "loss": 2.4976497650146485, + "step": 162980 + }, + { + "epoch": 0.4437185245678508, + "grad_norm": 2.2140767574310303, + "learning_rate": 1.1140847757939213e-05, + "loss": 2.4658576965332033, + "step": 163000 + }, + { + "epoch": 0.4437729685585953, + "grad_norm": 2.2447669506073, + "learning_rate": 1.113975739403026e-05, + "loss": 2.475630760192871, + "step": 163020 + }, + { + "epoch": 0.4438274125493399, + "grad_norm": 2.271031379699707, + "learning_rate": 1.1138667030121303e-05, + "loss": 2.480838394165039, + "step": 163040 + }, + { + "epoch": 0.4438818565400844, + "grad_norm": 2.5352752208709717, + "learning_rate": 1.113757666621235e-05, + "loss": 2.4778913497924804, + "step": 163060 + }, + { + "epoch": 0.4439363005308289, + "grad_norm": 2.08475923538208, + "learning_rate": 1.1136486302303395e-05, + "loss": 2.4475000381469725, + "step": 163080 + }, + { + "epoch": 0.44399074452157344, + "grad_norm": 2.150697946548462, + "learning_rate": 1.1135395938394439e-05, + "loss": 2.463603973388672, + "step": 163100 + }, + { + "epoch": 0.44404518851231795, + "grad_norm": 2.5519254207611084, + "learning_rate": 1.1134305574485485e-05, + "loss": 2.472316360473633, + "step": 163120 + }, + { + "epoch": 0.44409963250306245, + "grad_norm": 2.120394229888916, + "learning_rate": 1.113321521057653e-05, + "loss": 2.4730010986328126, + "step": 163140 + }, + { + "epoch": 0.444154076493807, + "grad_norm": 2.1635141372680664, + "learning_rate": 1.1132124846667577e-05, + "loss": 2.499821662902832, + "step": 163160 + }, + { + "epoch": 0.4442085204845515, + "grad_norm": 2.1001954078674316, + "learning_rate": 1.1131034482758622e-05, + "loss": 2.504830551147461, + "step": 163180 + }, + { + "epoch": 0.44426296447529606, + "grad_norm": 2.2441422939300537, + "learning_rate": 1.1129944118849666e-05, + "loss": 2.4763423919677736, + "step": 163200 + }, + { + "epoch": 0.44431740846604056, + "grad_norm": 2.208700180053711, + "learning_rate": 1.1128853754940712e-05, + "loss": 2.491974639892578, + "step": 163220 + }, + { + "epoch": 0.44437185245678507, + "grad_norm": 2.0954909324645996, + "learning_rate": 1.1127763391031758e-05, + "loss": 2.4983388900756838, + "step": 163240 + }, + { + "epoch": 0.4444262964475296, + "grad_norm": 2.159623384475708, + "learning_rate": 1.1126673027122802e-05, + "loss": 2.4837902069091795, + "step": 163260 + }, + { + "epoch": 0.4444807404382741, + "grad_norm": 2.0877788066864014, + "learning_rate": 1.1125582663213848e-05, + "loss": 2.462177848815918, + "step": 163280 + }, + { + "epoch": 0.4445351844290186, + "grad_norm": 2.119166612625122, + "learning_rate": 1.1124492299304894e-05, + "loss": 2.47731990814209, + "step": 163300 + }, + { + "epoch": 0.4445896284197632, + "grad_norm": 2.1428372859954834, + "learning_rate": 1.112340193539594e-05, + "loss": 2.5060964584350587, + "step": 163320 + }, + { + "epoch": 0.4446440724105077, + "grad_norm": 2.204280138015747, + "learning_rate": 1.1122311571486985e-05, + "loss": 2.4953052520751955, + "step": 163340 + }, + { + "epoch": 0.4446985164012522, + "grad_norm": 2.1777827739715576, + "learning_rate": 1.112122120757803e-05, + "loss": 2.475131607055664, + "step": 163360 + }, + { + "epoch": 0.44475296039199674, + "grad_norm": 2.2443368434906006, + "learning_rate": 1.1120130843669075e-05, + "loss": 2.4674793243408204, + "step": 163380 + }, + { + "epoch": 0.44480740438274124, + "grad_norm": 2.1954376697540283, + "learning_rate": 1.1119040479760121e-05, + "loss": 2.4906524658203124, + "step": 163400 + }, + { + "epoch": 0.4448618483734858, + "grad_norm": 2.1101012229919434, + "learning_rate": 1.1117950115851165e-05, + "loss": 2.501709747314453, + "step": 163420 + }, + { + "epoch": 0.4449162923642303, + "grad_norm": 2.259401559829712, + "learning_rate": 1.1116859751942211e-05, + "loss": 2.494815444946289, + "step": 163440 + }, + { + "epoch": 0.4449707363549748, + "grad_norm": 2.1269824504852295, + "learning_rate": 1.1115769388033257e-05, + "loss": 2.521311950683594, + "step": 163460 + }, + { + "epoch": 0.44502518034571936, + "grad_norm": 2.1054811477661133, + "learning_rate": 1.1114679024124303e-05, + "loss": 2.5078624725341796, + "step": 163480 + }, + { + "epoch": 0.44507962433646386, + "grad_norm": 2.201819896697998, + "learning_rate": 1.1113588660215349e-05, + "loss": 2.4868131637573243, + "step": 163500 + }, + { + "epoch": 0.44513406832720837, + "grad_norm": 2.1366751194000244, + "learning_rate": 1.1112498296306393e-05, + "loss": 2.511859893798828, + "step": 163520 + }, + { + "epoch": 0.4451885123179529, + "grad_norm": 2.2485079765319824, + "learning_rate": 1.1111407932397438e-05, + "loss": 2.504754066467285, + "step": 163540 + }, + { + "epoch": 0.4452429563086974, + "grad_norm": 2.1880156993865967, + "learning_rate": 1.1110317568488484e-05, + "loss": 2.493015480041504, + "step": 163560 + }, + { + "epoch": 0.4452974002994419, + "grad_norm": 2.1282920837402344, + "learning_rate": 1.1109227204579528e-05, + "loss": 2.5070516586303713, + "step": 163580 + }, + { + "epoch": 0.4453518442901865, + "grad_norm": 2.148207902908325, + "learning_rate": 1.1108136840670574e-05, + "loss": 2.492605781555176, + "step": 163600 + }, + { + "epoch": 0.445406288280931, + "grad_norm": 2.031280040740967, + "learning_rate": 1.110704647676162e-05, + "loss": 2.4693891525268556, + "step": 163620 + }, + { + "epoch": 0.44546073227167554, + "grad_norm": 2.5585415363311768, + "learning_rate": 1.1105956112852664e-05, + "loss": 2.4846818923950194, + "step": 163640 + }, + { + "epoch": 0.44551517626242004, + "grad_norm": 2.37070631980896, + "learning_rate": 1.1104865748943712e-05, + "loss": 2.485918426513672, + "step": 163660 + }, + { + "epoch": 0.44556962025316454, + "grad_norm": 2.140155553817749, + "learning_rate": 1.1103775385034756e-05, + "loss": 2.4568321228027346, + "step": 163680 + }, + { + "epoch": 0.4456240642439091, + "grad_norm": 2.1767842769622803, + "learning_rate": 1.1102685021125802e-05, + "loss": 2.484866142272949, + "step": 163700 + }, + { + "epoch": 0.4456785082346536, + "grad_norm": 2.159773111343384, + "learning_rate": 1.1101594657216847e-05, + "loss": 2.5202714920043947, + "step": 163720 + }, + { + "epoch": 0.4457329522253981, + "grad_norm": 3.931851863861084, + "learning_rate": 1.1100504293307891e-05, + "loss": 2.489597702026367, + "step": 163740 + }, + { + "epoch": 0.44578739621614266, + "grad_norm": 2.2225325107574463, + "learning_rate": 1.1099413929398937e-05, + "loss": 2.470235824584961, + "step": 163760 + }, + { + "epoch": 0.44584184020688716, + "grad_norm": 2.171888589859009, + "learning_rate": 1.1098323565489983e-05, + "loss": 2.5087217330932616, + "step": 163780 + }, + { + "epoch": 0.44589628419763166, + "grad_norm": 2.1091084480285645, + "learning_rate": 1.1097233201581027e-05, + "loss": 2.4717878341674804, + "step": 163800 + }, + { + "epoch": 0.4459507281883762, + "grad_norm": 2.149909734725952, + "learning_rate": 1.1096142837672075e-05, + "loss": 2.4940345764160154, + "step": 163820 + }, + { + "epoch": 0.4460051721791207, + "grad_norm": 2.4715943336486816, + "learning_rate": 1.1095052473763119e-05, + "loss": 2.486497688293457, + "step": 163840 + }, + { + "epoch": 0.4460596161698652, + "grad_norm": 2.061583995819092, + "learning_rate": 1.1093962109854165e-05, + "loss": 2.480812644958496, + "step": 163860 + }, + { + "epoch": 0.4461140601606098, + "grad_norm": 2.1877315044403076, + "learning_rate": 1.109287174594521e-05, + "loss": 2.493373680114746, + "step": 163880 + }, + { + "epoch": 0.4461685041513543, + "grad_norm": 2.223052978515625, + "learning_rate": 1.1091781382036255e-05, + "loss": 2.503744888305664, + "step": 163900 + }, + { + "epoch": 0.44622294814209884, + "grad_norm": 2.043591260910034, + "learning_rate": 1.1090691018127302e-05, + "loss": 2.514682388305664, + "step": 163920 + }, + { + "epoch": 0.44627739213284334, + "grad_norm": 2.1348674297332764, + "learning_rate": 1.1089600654218346e-05, + "loss": 2.515286445617676, + "step": 163940 + }, + { + "epoch": 0.44633183612358784, + "grad_norm": 2.293578624725342, + "learning_rate": 1.108851029030939e-05, + "loss": 2.505343437194824, + "step": 163960 + }, + { + "epoch": 0.4463862801143324, + "grad_norm": 2.1935300827026367, + "learning_rate": 1.1087419926400438e-05, + "loss": 2.4677173614501955, + "step": 163980 + }, + { + "epoch": 0.4464407241050769, + "grad_norm": 2.1267943382263184, + "learning_rate": 1.1086329562491482e-05, + "loss": 2.4791399002075196, + "step": 164000 + }, + { + "epoch": 0.4464951680958214, + "grad_norm": 2.161811113357544, + "learning_rate": 1.1085239198582528e-05, + "loss": 2.4723207473754885, + "step": 164020 + }, + { + "epoch": 0.44654961208656596, + "grad_norm": 2.144871950149536, + "learning_rate": 1.1084148834673574e-05, + "loss": 2.47037353515625, + "step": 164040 + }, + { + "epoch": 0.44660405607731046, + "grad_norm": 2.1319172382354736, + "learning_rate": 1.1083058470764618e-05, + "loss": 2.4861675262451173, + "step": 164060 + }, + { + "epoch": 0.44665850006805496, + "grad_norm": 2.209307909011841, + "learning_rate": 1.1081968106855665e-05, + "loss": 2.4958728790283202, + "step": 164080 + }, + { + "epoch": 0.4467129440587995, + "grad_norm": 2.1202690601348877, + "learning_rate": 1.108087774294671e-05, + "loss": 2.5138315200805663, + "step": 164100 + }, + { + "epoch": 0.446767388049544, + "grad_norm": 2.112752676010132, + "learning_rate": 1.1079787379037753e-05, + "loss": 2.492331695556641, + "step": 164120 + }, + { + "epoch": 0.4468218320402886, + "grad_norm": 2.112837076187134, + "learning_rate": 1.1078697015128801e-05, + "loss": 2.5011301040649414, + "step": 164140 + }, + { + "epoch": 0.4468762760310331, + "grad_norm": 2.223966121673584, + "learning_rate": 1.1077606651219845e-05, + "loss": 2.477145767211914, + "step": 164160 + }, + { + "epoch": 0.4469307200217776, + "grad_norm": 2.1003432273864746, + "learning_rate": 1.107651628731089e-05, + "loss": 2.506929397583008, + "step": 164180 + }, + { + "epoch": 0.44698516401252214, + "grad_norm": 2.1207568645477295, + "learning_rate": 1.1075425923401937e-05, + "loss": 2.4935569763183594, + "step": 164200 + }, + { + "epoch": 0.44703960800326664, + "grad_norm": 2.3612687587738037, + "learning_rate": 1.107433555949298e-05, + "loss": 2.498245429992676, + "step": 164220 + }, + { + "epoch": 0.44709405199401114, + "grad_norm": 2.524519205093384, + "learning_rate": 1.1073245195584028e-05, + "loss": 2.495545768737793, + "step": 164240 + }, + { + "epoch": 0.4471484959847557, + "grad_norm": 2.0850419998168945, + "learning_rate": 1.1072154831675072e-05, + "loss": 2.4971580505371094, + "step": 164260 + }, + { + "epoch": 0.4472029399755002, + "grad_norm": 2.1061551570892334, + "learning_rate": 1.1071064467766116e-05, + "loss": 2.4627300262451173, + "step": 164280 + }, + { + "epoch": 0.4472573839662447, + "grad_norm": 3.0486600399017334, + "learning_rate": 1.1069974103857164e-05, + "loss": 2.5171392440795897, + "step": 164300 + }, + { + "epoch": 0.44731182795698926, + "grad_norm": 2.2105419635772705, + "learning_rate": 1.1068883739948208e-05, + "loss": 2.475353240966797, + "step": 164320 + }, + { + "epoch": 0.44736627194773376, + "grad_norm": 2.0979626178741455, + "learning_rate": 1.1067793376039254e-05, + "loss": 2.4601251602172853, + "step": 164340 + }, + { + "epoch": 0.44742071593847826, + "grad_norm": 2.0843191146850586, + "learning_rate": 1.10667030121303e-05, + "loss": 2.474717140197754, + "step": 164360 + }, + { + "epoch": 0.4474751599292228, + "grad_norm": 2.431586980819702, + "learning_rate": 1.1065612648221344e-05, + "loss": 2.5076263427734373, + "step": 164380 + }, + { + "epoch": 0.4475296039199673, + "grad_norm": 2.1293013095855713, + "learning_rate": 1.1064522284312391e-05, + "loss": 2.4900293350219727, + "step": 164400 + }, + { + "epoch": 0.4475840479107119, + "grad_norm": 2.1651623249053955, + "learning_rate": 1.1063431920403435e-05, + "loss": 2.478449821472168, + "step": 164420 + }, + { + "epoch": 0.4476384919014564, + "grad_norm": 2.2381339073181152, + "learning_rate": 1.106234155649448e-05, + "loss": 2.5086175918579103, + "step": 164440 + }, + { + "epoch": 0.4476929358922009, + "grad_norm": 2.1554598808288574, + "learning_rate": 1.1061251192585527e-05, + "loss": 2.4997543334960937, + "step": 164460 + }, + { + "epoch": 0.44774737988294544, + "grad_norm": 2.1370136737823486, + "learning_rate": 1.1060160828676571e-05, + "loss": 2.4578149795532225, + "step": 164480 + }, + { + "epoch": 0.44780182387368994, + "grad_norm": 2.116398572921753, + "learning_rate": 1.1059070464767617e-05, + "loss": 2.471917915344238, + "step": 164500 + }, + { + "epoch": 0.44785626786443444, + "grad_norm": 2.1904208660125732, + "learning_rate": 1.1057980100858663e-05, + "loss": 2.4718151092529297, + "step": 164520 + }, + { + "epoch": 0.447910711855179, + "grad_norm": 2.471710443496704, + "learning_rate": 1.1056889736949707e-05, + "loss": 2.4727930068969726, + "step": 164540 + }, + { + "epoch": 0.4479651558459235, + "grad_norm": 2.14967679977417, + "learning_rate": 1.1055799373040754e-05, + "loss": 2.462248420715332, + "step": 164560 + }, + { + "epoch": 0.448019599836668, + "grad_norm": 2.2205276489257812, + "learning_rate": 1.1054709009131799e-05, + "loss": 2.5141119003295898, + "step": 164580 + }, + { + "epoch": 0.44807404382741256, + "grad_norm": 2.20843505859375, + "learning_rate": 1.1053618645222843e-05, + "loss": 2.473776435852051, + "step": 164600 + }, + { + "epoch": 0.44812848781815706, + "grad_norm": 2.230809450149536, + "learning_rate": 1.105252828131389e-05, + "loss": 2.5046682357788086, + "step": 164620 + }, + { + "epoch": 0.4481829318089016, + "grad_norm": 2.5847787857055664, + "learning_rate": 1.1051437917404934e-05, + "loss": 2.5117639541625976, + "step": 164640 + }, + { + "epoch": 0.4482373757996461, + "grad_norm": 2.1071338653564453, + "learning_rate": 1.1050347553495978e-05, + "loss": 2.5220945358276365, + "step": 164660 + }, + { + "epoch": 0.4482918197903906, + "grad_norm": 2.2112207412719727, + "learning_rate": 1.1049257189587026e-05, + "loss": 2.4763505935668944, + "step": 164680 + }, + { + "epoch": 0.4483462637811352, + "grad_norm": 2.1531822681427, + "learning_rate": 1.104816682567807e-05, + "loss": 2.480596351623535, + "step": 164700 + }, + { + "epoch": 0.4484007077718797, + "grad_norm": 2.1331117153167725, + "learning_rate": 1.1047076461769117e-05, + "loss": 2.4960796356201174, + "step": 164720 + }, + { + "epoch": 0.4484551517626242, + "grad_norm": 2.1275267601013184, + "learning_rate": 1.1045986097860162e-05, + "loss": 2.4796043395996095, + "step": 164740 + }, + { + "epoch": 0.44850959575336874, + "grad_norm": 2.1006641387939453, + "learning_rate": 1.1044895733951206e-05, + "loss": 2.5045976638793945, + "step": 164760 + }, + { + "epoch": 0.44856403974411324, + "grad_norm": 2.1300764083862305, + "learning_rate": 1.1043805370042253e-05, + "loss": 2.5113916397094727, + "step": 164780 + }, + { + "epoch": 0.44861848373485774, + "grad_norm": 2.0949041843414307, + "learning_rate": 1.1042715006133297e-05, + "loss": 2.5120355606079103, + "step": 164800 + }, + { + "epoch": 0.4486729277256023, + "grad_norm": 2.1465694904327393, + "learning_rate": 1.1041624642224341e-05, + "loss": 2.5065704345703126, + "step": 164820 + }, + { + "epoch": 0.4487273717163468, + "grad_norm": 2.0912559032440186, + "learning_rate": 1.1040534278315389e-05, + "loss": 2.5002580642700196, + "step": 164840 + }, + { + "epoch": 0.4487818157070913, + "grad_norm": 2.1370158195495605, + "learning_rate": 1.1039443914406433e-05, + "loss": 2.4529230117797853, + "step": 164860 + }, + { + "epoch": 0.44883625969783586, + "grad_norm": 2.150235652923584, + "learning_rate": 1.103835355049748e-05, + "loss": 2.4568496704101563, + "step": 164880 + }, + { + "epoch": 0.44889070368858036, + "grad_norm": 3.3296804428100586, + "learning_rate": 1.1037263186588525e-05, + "loss": 2.493611717224121, + "step": 164900 + }, + { + "epoch": 0.4489451476793249, + "grad_norm": 2.1157734394073486, + "learning_rate": 1.1036172822679569e-05, + "loss": 2.508649444580078, + "step": 164920 + }, + { + "epoch": 0.4489995916700694, + "grad_norm": 2.2314236164093018, + "learning_rate": 1.1035082458770616e-05, + "loss": 2.445867919921875, + "step": 164940 + }, + { + "epoch": 0.4490540356608139, + "grad_norm": 2.5549216270446777, + "learning_rate": 1.103399209486166e-05, + "loss": 2.469913673400879, + "step": 164960 + }, + { + "epoch": 0.4491084796515585, + "grad_norm": 2.7416181564331055, + "learning_rate": 1.1032901730952705e-05, + "loss": 2.4671604156494142, + "step": 164980 + }, + { + "epoch": 0.449162923642303, + "grad_norm": 2.1238017082214355, + "learning_rate": 1.1031811367043752e-05, + "loss": 2.483952522277832, + "step": 165000 + }, + { + "epoch": 0.4492173676330475, + "grad_norm": 2.132336139678955, + "learning_rate": 1.1030721003134796e-05, + "loss": 2.4832481384277343, + "step": 165020 + }, + { + "epoch": 0.44927181162379204, + "grad_norm": 2.3279144763946533, + "learning_rate": 1.1029630639225844e-05, + "loss": 2.5079111099243163, + "step": 165040 + }, + { + "epoch": 0.44932625561453654, + "grad_norm": 2.0177366733551025, + "learning_rate": 1.1028540275316888e-05, + "loss": 2.5074846267700197, + "step": 165060 + }, + { + "epoch": 0.44938069960528104, + "grad_norm": 2.200605630874634, + "learning_rate": 1.1027449911407932e-05, + "loss": 2.4798486709594725, + "step": 165080 + }, + { + "epoch": 0.4494351435960256, + "grad_norm": 2.1245386600494385, + "learning_rate": 1.102635954749898e-05, + "loss": 2.469228744506836, + "step": 165100 + }, + { + "epoch": 0.4494895875867701, + "grad_norm": 2.1348814964294434, + "learning_rate": 1.1025269183590024e-05, + "loss": 2.4919830322265626, + "step": 165120 + }, + { + "epoch": 0.44954403157751466, + "grad_norm": 2.179089307785034, + "learning_rate": 1.1024178819681068e-05, + "loss": 2.490158271789551, + "step": 165140 + }, + { + "epoch": 0.44959847556825916, + "grad_norm": 2.118894338607788, + "learning_rate": 1.1023088455772115e-05, + "loss": 2.4850425720214844, + "step": 165160 + }, + { + "epoch": 0.44965291955900366, + "grad_norm": 2.1772046089172363, + "learning_rate": 1.102199809186316e-05, + "loss": 2.457440948486328, + "step": 165180 + }, + { + "epoch": 0.4497073635497482, + "grad_norm": 2.0877881050109863, + "learning_rate": 1.1020907727954207e-05, + "loss": 2.502370262145996, + "step": 165200 + }, + { + "epoch": 0.4497618075404927, + "grad_norm": 2.1411614418029785, + "learning_rate": 1.1019817364045251e-05, + "loss": 2.4594598770141602, + "step": 165220 + }, + { + "epoch": 0.4498162515312372, + "grad_norm": 2.1479651927948, + "learning_rate": 1.1018727000136295e-05, + "loss": 2.4750045776367187, + "step": 165240 + }, + { + "epoch": 0.4498706955219818, + "grad_norm": 2.2307238578796387, + "learning_rate": 1.1017636636227342e-05, + "loss": 2.4792758941650392, + "step": 165260 + }, + { + "epoch": 0.4499251395127263, + "grad_norm": 2.602041244506836, + "learning_rate": 1.1016546272318387e-05, + "loss": 2.498283576965332, + "step": 165280 + }, + { + "epoch": 0.4499795835034708, + "grad_norm": 2.2287402153015137, + "learning_rate": 1.101545590840943e-05, + "loss": 2.50675106048584, + "step": 165300 + }, + { + "epoch": 0.45003402749421534, + "grad_norm": 2.1060361862182617, + "learning_rate": 1.1014365544500478e-05, + "loss": 2.5016416549682616, + "step": 165320 + }, + { + "epoch": 0.45008847148495984, + "grad_norm": 2.1603217124938965, + "learning_rate": 1.1013275180591522e-05, + "loss": 2.4871906280517577, + "step": 165340 + }, + { + "epoch": 0.45014291547570434, + "grad_norm": 2.112347364425659, + "learning_rate": 1.101218481668257e-05, + "loss": 2.489642333984375, + "step": 165360 + }, + { + "epoch": 0.4501973594664489, + "grad_norm": 2.255797863006592, + "learning_rate": 1.1011094452773614e-05, + "loss": 2.4992984771728515, + "step": 165380 + }, + { + "epoch": 0.4502518034571934, + "grad_norm": 2.191991090774536, + "learning_rate": 1.1010004088864658e-05, + "loss": 2.492504692077637, + "step": 165400 + }, + { + "epoch": 0.45030624744793796, + "grad_norm": 2.2219045162200928, + "learning_rate": 1.1008913724955706e-05, + "loss": 2.4944831848144533, + "step": 165420 + }, + { + "epoch": 0.45036069143868246, + "grad_norm": 2.185720920562744, + "learning_rate": 1.100782336104675e-05, + "loss": 2.4857763290405273, + "step": 165440 + }, + { + "epoch": 0.45041513542942696, + "grad_norm": 2.1422271728515625, + "learning_rate": 1.1006732997137797e-05, + "loss": 2.465838623046875, + "step": 165460 + }, + { + "epoch": 0.4504695794201715, + "grad_norm": 2.3698713779449463, + "learning_rate": 1.1005642633228841e-05, + "loss": 2.483811950683594, + "step": 165480 + }, + { + "epoch": 0.450524023410916, + "grad_norm": 2.1829497814178467, + "learning_rate": 1.1004552269319885e-05, + "loss": 2.482029342651367, + "step": 165500 + }, + { + "epoch": 0.4505784674016605, + "grad_norm": 2.1063156127929688, + "learning_rate": 1.1003461905410933e-05, + "loss": 2.482064437866211, + "step": 165520 + }, + { + "epoch": 0.4506329113924051, + "grad_norm": 2.1099822521209717, + "learning_rate": 1.1002371541501977e-05, + "loss": 2.5052249908447264, + "step": 165540 + }, + { + "epoch": 0.4506873553831496, + "grad_norm": 2.62833309173584, + "learning_rate": 1.1001281177593021e-05, + "loss": 2.4742446899414063, + "step": 165560 + }, + { + "epoch": 0.4507417993738941, + "grad_norm": 2.1539456844329834, + "learning_rate": 1.1000190813684069e-05, + "loss": 2.4842973709106446, + "step": 165580 + }, + { + "epoch": 0.45079624336463864, + "grad_norm": 2.3932242393493652, + "learning_rate": 1.0999100449775113e-05, + "loss": 2.4940025329589846, + "step": 165600 + }, + { + "epoch": 0.45085068735538314, + "grad_norm": 2.280177593231201, + "learning_rate": 1.099801008586616e-05, + "loss": 2.4711481094360352, + "step": 165620 + }, + { + "epoch": 0.4509051313461277, + "grad_norm": 2.136192798614502, + "learning_rate": 1.0996919721957204e-05, + "loss": 2.4926528930664062, + "step": 165640 + }, + { + "epoch": 0.4509595753368722, + "grad_norm": 2.136733293533325, + "learning_rate": 1.0995829358048248e-05, + "loss": 2.466598129272461, + "step": 165660 + }, + { + "epoch": 0.4510140193276167, + "grad_norm": 2.1010193824768066, + "learning_rate": 1.0994738994139296e-05, + "loss": 2.455994415283203, + "step": 165680 + }, + { + "epoch": 0.45106846331836126, + "grad_norm": 2.1102261543273926, + "learning_rate": 1.099364863023034e-05, + "loss": 2.4945159912109376, + "step": 165700 + }, + { + "epoch": 0.45112290730910576, + "grad_norm": 3.0537843704223633, + "learning_rate": 1.0992558266321384e-05, + "loss": 2.465869140625, + "step": 165720 + }, + { + "epoch": 0.45117735129985026, + "grad_norm": 2.1575636863708496, + "learning_rate": 1.0991467902412432e-05, + "loss": 2.4796045303344725, + "step": 165740 + }, + { + "epoch": 0.4512317952905948, + "grad_norm": 2.36702561378479, + "learning_rate": 1.0990377538503476e-05, + "loss": 2.4920278549194337, + "step": 165760 + }, + { + "epoch": 0.4512862392813393, + "grad_norm": 2.4204933643341064, + "learning_rate": 1.0989287174594523e-05, + "loss": 2.5212884902954102, + "step": 165780 + }, + { + "epoch": 0.4513406832720838, + "grad_norm": 2.1611921787261963, + "learning_rate": 1.0988196810685567e-05, + "loss": 2.499879837036133, + "step": 165800 + }, + { + "epoch": 0.4513951272628284, + "grad_norm": 2.186525344848633, + "learning_rate": 1.0987106446776612e-05, + "loss": 2.4766513824462892, + "step": 165820 + }, + { + "epoch": 0.4514495712535729, + "grad_norm": 2.144007682800293, + "learning_rate": 1.0986016082867659e-05, + "loss": 2.4442014694213867, + "step": 165840 + }, + { + "epoch": 0.45150401524431744, + "grad_norm": 2.19244647026062, + "learning_rate": 1.0984925718958703e-05, + "loss": 2.4552120208740233, + "step": 165860 + }, + { + "epoch": 0.45155845923506194, + "grad_norm": 2.251699209213257, + "learning_rate": 1.0983835355049747e-05, + "loss": 2.5082231521606446, + "step": 165880 + }, + { + "epoch": 0.45161290322580644, + "grad_norm": 2.427847385406494, + "learning_rate": 1.0982744991140795e-05, + "loss": 2.478253173828125, + "step": 165900 + }, + { + "epoch": 0.451667347216551, + "grad_norm": 2.1736817359924316, + "learning_rate": 1.0981654627231839e-05, + "loss": 2.4760725021362306, + "step": 165920 + }, + { + "epoch": 0.4517217912072955, + "grad_norm": 2.0899932384490967, + "learning_rate": 1.0980564263322886e-05, + "loss": 2.4743980407714843, + "step": 165940 + }, + { + "epoch": 0.45177623519804, + "grad_norm": 2.07517147064209, + "learning_rate": 1.097947389941393e-05, + "loss": 2.476127815246582, + "step": 165960 + }, + { + "epoch": 0.45183067918878456, + "grad_norm": 2.1683690547943115, + "learning_rate": 1.0978383535504975e-05, + "loss": 2.500909614562988, + "step": 165980 + }, + { + "epoch": 0.45188512317952906, + "grad_norm": 2.2237348556518555, + "learning_rate": 1.0977293171596022e-05, + "loss": 2.520049476623535, + "step": 166000 + }, + { + "epoch": 0.45193956717027356, + "grad_norm": 2.322482109069824, + "learning_rate": 1.0976202807687066e-05, + "loss": 2.5304006576538085, + "step": 166020 + }, + { + "epoch": 0.4519940111610181, + "grad_norm": 2.2403364181518555, + "learning_rate": 1.097511244377811e-05, + "loss": 2.485678291320801, + "step": 166040 + }, + { + "epoch": 0.4520484551517626, + "grad_norm": 2.0854380130767822, + "learning_rate": 1.0974022079869158e-05, + "loss": 2.4857500076293944, + "step": 166060 + }, + { + "epoch": 0.4521028991425071, + "grad_norm": 2.013531446456909, + "learning_rate": 1.0972931715960202e-05, + "loss": 2.5267013549804687, + "step": 166080 + }, + { + "epoch": 0.4521573431332517, + "grad_norm": 2.277491331100464, + "learning_rate": 1.097184135205125e-05, + "loss": 2.4886924743652346, + "step": 166100 + }, + { + "epoch": 0.4522117871239962, + "grad_norm": 2.0707781314849854, + "learning_rate": 1.0970750988142294e-05, + "loss": 2.484427261352539, + "step": 166120 + }, + { + "epoch": 0.45226623111474074, + "grad_norm": 2.5392541885375977, + "learning_rate": 1.0969660624233338e-05, + "loss": 2.484658432006836, + "step": 166140 + }, + { + "epoch": 0.45232067510548524, + "grad_norm": 2.2053065299987793, + "learning_rate": 1.0968570260324385e-05, + "loss": 2.5206649780273436, + "step": 166160 + }, + { + "epoch": 0.45237511909622974, + "grad_norm": 2.090540885925293, + "learning_rate": 1.096747989641543e-05, + "loss": 2.503073310852051, + "step": 166180 + }, + { + "epoch": 0.4524295630869743, + "grad_norm": 2.1106300354003906, + "learning_rate": 1.0966389532506473e-05, + "loss": 2.4847936630249023, + "step": 166200 + }, + { + "epoch": 0.4524840070777188, + "grad_norm": 2.099935293197632, + "learning_rate": 1.0965299168597521e-05, + "loss": 2.4711025238037108, + "step": 166220 + }, + { + "epoch": 0.4525384510684633, + "grad_norm": 2.183816909790039, + "learning_rate": 1.0964208804688565e-05, + "loss": 2.4729497909545897, + "step": 166240 + }, + { + "epoch": 0.45259289505920786, + "grad_norm": 2.0841445922851562, + "learning_rate": 1.0963118440779613e-05, + "loss": 2.443478584289551, + "step": 166260 + }, + { + "epoch": 0.45264733904995236, + "grad_norm": 2.456144332885742, + "learning_rate": 1.0962028076870657e-05, + "loss": 2.4584577560424803, + "step": 166280 + }, + { + "epoch": 0.45270178304069686, + "grad_norm": 2.122072696685791, + "learning_rate": 1.09609377129617e-05, + "loss": 2.474264907836914, + "step": 166300 + }, + { + "epoch": 0.4527562270314414, + "grad_norm": 2.1724560260772705, + "learning_rate": 1.0959847349052748e-05, + "loss": 2.455343818664551, + "step": 166320 + }, + { + "epoch": 0.4528106710221859, + "grad_norm": 2.147639513015747, + "learning_rate": 1.0958756985143792e-05, + "loss": 2.483000564575195, + "step": 166340 + }, + { + "epoch": 0.4528651150129305, + "grad_norm": 2.1053473949432373, + "learning_rate": 1.0957666621234837e-05, + "loss": 2.4876962661743165, + "step": 166360 + }, + { + "epoch": 0.452919559003675, + "grad_norm": 2.125300168991089, + "learning_rate": 1.0956576257325884e-05, + "loss": 2.4884706497192384, + "step": 166380 + }, + { + "epoch": 0.4529740029944195, + "grad_norm": 3.2731316089630127, + "learning_rate": 1.0955485893416928e-05, + "loss": 2.49729061126709, + "step": 166400 + }, + { + "epoch": 0.45302844698516403, + "grad_norm": 2.079376220703125, + "learning_rate": 1.0954395529507976e-05, + "loss": 2.5002622604370117, + "step": 166420 + }, + { + "epoch": 0.45308289097590854, + "grad_norm": 2.150723457336426, + "learning_rate": 1.095330516559902e-05, + "loss": 2.4937885284423826, + "step": 166440 + }, + { + "epoch": 0.45313733496665304, + "grad_norm": 2.2101309299468994, + "learning_rate": 1.0952214801690064e-05, + "loss": 2.5057209014892576, + "step": 166460 + }, + { + "epoch": 0.4531917789573976, + "grad_norm": 2.115544319152832, + "learning_rate": 1.0951124437781111e-05, + "loss": 2.5068822860717774, + "step": 166480 + }, + { + "epoch": 0.4532462229481421, + "grad_norm": 2.27441143989563, + "learning_rate": 1.0950034073872156e-05, + "loss": 2.524286460876465, + "step": 166500 + }, + { + "epoch": 0.4533006669388866, + "grad_norm": 2.162101984024048, + "learning_rate": 1.09489437099632e-05, + "loss": 2.5023258209228514, + "step": 166520 + }, + { + "epoch": 0.45335511092963116, + "grad_norm": 2.233009099960327, + "learning_rate": 1.0947853346054247e-05, + "loss": 2.4915115356445314, + "step": 166540 + }, + { + "epoch": 0.45340955492037566, + "grad_norm": 2.152496814727783, + "learning_rate": 1.0946762982145291e-05, + "loss": 2.479391098022461, + "step": 166560 + }, + { + "epoch": 0.45346399891112016, + "grad_norm": 2.035004138946533, + "learning_rate": 1.0945672618236339e-05, + "loss": 2.4422740936279297, + "step": 166580 + }, + { + "epoch": 0.4535184429018647, + "grad_norm": 2.116816759109497, + "learning_rate": 1.0944582254327383e-05, + "loss": 2.4933427810668944, + "step": 166600 + }, + { + "epoch": 0.4535728868926092, + "grad_norm": 2.0715696811676025, + "learning_rate": 1.0943491890418427e-05, + "loss": 2.4716165542602537, + "step": 166620 + }, + { + "epoch": 0.4536273308833538, + "grad_norm": 2.140897274017334, + "learning_rate": 1.0942401526509475e-05, + "loss": 2.4751859664916993, + "step": 166640 + }, + { + "epoch": 0.4536817748740983, + "grad_norm": 2.192500591278076, + "learning_rate": 1.0941311162600519e-05, + "loss": 2.478057861328125, + "step": 166660 + }, + { + "epoch": 0.4537362188648428, + "grad_norm": 2.221896171569824, + "learning_rate": 1.0940220798691563e-05, + "loss": 2.4733795166015624, + "step": 166680 + }, + { + "epoch": 0.45379066285558733, + "grad_norm": 2.227059841156006, + "learning_rate": 1.093913043478261e-05, + "loss": 2.4991924285888674, + "step": 166700 + }, + { + "epoch": 0.45384510684633184, + "grad_norm": 2.169461488723755, + "learning_rate": 1.0938040070873654e-05, + "loss": 2.480143165588379, + "step": 166720 + }, + { + "epoch": 0.45389955083707634, + "grad_norm": 2.916771650314331, + "learning_rate": 1.0936949706964702e-05, + "loss": 2.4752641677856446, + "step": 166740 + }, + { + "epoch": 0.4539539948278209, + "grad_norm": 2.164911985397339, + "learning_rate": 1.0935859343055746e-05, + "loss": 2.438816452026367, + "step": 166760 + }, + { + "epoch": 0.4540084388185654, + "grad_norm": 2.191546678543091, + "learning_rate": 1.093476897914679e-05, + "loss": 2.494159126281738, + "step": 166780 + }, + { + "epoch": 0.4540628828093099, + "grad_norm": 2.2368810176849365, + "learning_rate": 1.0933678615237838e-05, + "loss": 2.4936527252197265, + "step": 166800 + }, + { + "epoch": 0.45411732680005445, + "grad_norm": 2.2447590827941895, + "learning_rate": 1.0932588251328882e-05, + "loss": 2.4864858627319335, + "step": 166820 + }, + { + "epoch": 0.45417177079079896, + "grad_norm": 2.230257987976074, + "learning_rate": 1.0931497887419926e-05, + "loss": 2.4851884841918945, + "step": 166840 + }, + { + "epoch": 0.4542262147815435, + "grad_norm": 3.5688939094543457, + "learning_rate": 1.0930407523510973e-05, + "loss": 2.4950660705566405, + "step": 166860 + }, + { + "epoch": 0.454280658772288, + "grad_norm": 2.1325457096099854, + "learning_rate": 1.0929317159602017e-05, + "loss": 2.476361846923828, + "step": 166880 + }, + { + "epoch": 0.4543351027630325, + "grad_norm": 2.096536636352539, + "learning_rate": 1.0928226795693065e-05, + "loss": 2.4780963897705077, + "step": 166900 + }, + { + "epoch": 0.4543895467537771, + "grad_norm": 2.157529830932617, + "learning_rate": 1.0927136431784109e-05, + "loss": 2.5481210708618165, + "step": 166920 + }, + { + "epoch": 0.4544439907445216, + "grad_norm": 2.5715255737304688, + "learning_rate": 1.0926046067875153e-05, + "loss": 2.468621826171875, + "step": 166940 + }, + { + "epoch": 0.4544984347352661, + "grad_norm": 2.2069952487945557, + "learning_rate": 1.09249557039662e-05, + "loss": 2.514811134338379, + "step": 166960 + }, + { + "epoch": 0.45455287872601063, + "grad_norm": 2.1753790378570557, + "learning_rate": 1.0923865340057245e-05, + "loss": 2.489836883544922, + "step": 166980 + }, + { + "epoch": 0.45460732271675514, + "grad_norm": 2.166356086730957, + "learning_rate": 1.0922774976148289e-05, + "loss": 2.489754867553711, + "step": 167000 + }, + { + "epoch": 0.45466176670749964, + "grad_norm": 2.1473300457000732, + "learning_rate": 1.0921684612239336e-05, + "loss": 2.478611183166504, + "step": 167020 + }, + { + "epoch": 0.4547162106982442, + "grad_norm": 2.1662654876708984, + "learning_rate": 1.092059424833038e-05, + "loss": 2.464616394042969, + "step": 167040 + }, + { + "epoch": 0.4547706546889887, + "grad_norm": 2.132978916168213, + "learning_rate": 1.0919503884421428e-05, + "loss": 2.47462158203125, + "step": 167060 + }, + { + "epoch": 0.4548250986797332, + "grad_norm": 2.2341887950897217, + "learning_rate": 1.0918413520512472e-05, + "loss": 2.471208953857422, + "step": 167080 + }, + { + "epoch": 0.45487954267047775, + "grad_norm": 2.1845157146453857, + "learning_rate": 1.0917323156603516e-05, + "loss": 2.495004081726074, + "step": 167100 + }, + { + "epoch": 0.45493398666122226, + "grad_norm": 2.2562193870544434, + "learning_rate": 1.0916232792694564e-05, + "loss": 2.4689508438110352, + "step": 167120 + }, + { + "epoch": 0.4549884306519668, + "grad_norm": 2.2884886264801025, + "learning_rate": 1.0915142428785608e-05, + "loss": 2.5012353897094726, + "step": 167140 + }, + { + "epoch": 0.4550428746427113, + "grad_norm": 2.1382853984832764, + "learning_rate": 1.0914052064876655e-05, + "loss": 2.499453544616699, + "step": 167160 + }, + { + "epoch": 0.4550973186334558, + "grad_norm": 2.330045700073242, + "learning_rate": 1.09129617009677e-05, + "loss": 2.5011260986328123, + "step": 167180 + }, + { + "epoch": 0.4551517626242004, + "grad_norm": 2.122546672821045, + "learning_rate": 1.0911871337058744e-05, + "loss": 2.4655071258544923, + "step": 167200 + }, + { + "epoch": 0.4552062066149449, + "grad_norm": 2.2134358882904053, + "learning_rate": 1.0910780973149791e-05, + "loss": 2.5077726364135744, + "step": 167220 + }, + { + "epoch": 0.4552606506056894, + "grad_norm": 2.1153550148010254, + "learning_rate": 1.0909690609240835e-05, + "loss": 2.4735107421875, + "step": 167240 + }, + { + "epoch": 0.45531509459643393, + "grad_norm": 2.904261350631714, + "learning_rate": 1.090860024533188e-05, + "loss": 2.4639123916625976, + "step": 167260 + }, + { + "epoch": 0.45536953858717844, + "grad_norm": 2.1426281929016113, + "learning_rate": 1.0907509881422927e-05, + "loss": 2.5036205291748046, + "step": 167280 + }, + { + "epoch": 0.45542398257792294, + "grad_norm": 2.2441468238830566, + "learning_rate": 1.0906419517513971e-05, + "loss": 2.4758636474609377, + "step": 167300 + }, + { + "epoch": 0.4554784265686675, + "grad_norm": 2.2070670127868652, + "learning_rate": 1.0905329153605018e-05, + "loss": 2.516861152648926, + "step": 167320 + }, + { + "epoch": 0.455532870559412, + "grad_norm": 2.2449429035186768, + "learning_rate": 1.0904238789696063e-05, + "loss": 2.5012096405029296, + "step": 167340 + }, + { + "epoch": 0.45558731455015655, + "grad_norm": 2.203561782836914, + "learning_rate": 1.0903148425787107e-05, + "loss": 2.4807676315307616, + "step": 167360 + }, + { + "epoch": 0.45564175854090105, + "grad_norm": 2.0121867656707764, + "learning_rate": 1.0902058061878154e-05, + "loss": 2.4778060913085938, + "step": 167380 + }, + { + "epoch": 0.45569620253164556, + "grad_norm": 2.087123394012451, + "learning_rate": 1.0900967697969198e-05, + "loss": 2.4717279434204102, + "step": 167400 + }, + { + "epoch": 0.4557506465223901, + "grad_norm": 2.2016537189483643, + "learning_rate": 1.0899877334060242e-05, + "loss": 2.4786426544189455, + "step": 167420 + }, + { + "epoch": 0.4558050905131346, + "grad_norm": 2.3210902214050293, + "learning_rate": 1.089878697015129e-05, + "loss": 2.4845855712890623, + "step": 167440 + }, + { + "epoch": 0.4558595345038791, + "grad_norm": 2.218626022338867, + "learning_rate": 1.0897696606242334e-05, + "loss": 2.4718971252441406, + "step": 167460 + }, + { + "epoch": 0.4559139784946237, + "grad_norm": 2.205052375793457, + "learning_rate": 1.0896606242333382e-05, + "loss": 2.483730125427246, + "step": 167480 + }, + { + "epoch": 0.4559684224853682, + "grad_norm": 3.5563466548919678, + "learning_rate": 1.0895515878424426e-05, + "loss": 2.465424156188965, + "step": 167500 + }, + { + "epoch": 0.4560228664761127, + "grad_norm": 2.177341938018799, + "learning_rate": 1.089442551451547e-05, + "loss": 2.4814180374145507, + "step": 167520 + }, + { + "epoch": 0.45607731046685723, + "grad_norm": 2.2443971633911133, + "learning_rate": 1.0893335150606517e-05, + "loss": 2.500741386413574, + "step": 167540 + }, + { + "epoch": 0.45613175445760173, + "grad_norm": 2.20225191116333, + "learning_rate": 1.0892244786697561e-05, + "loss": 2.496577262878418, + "step": 167560 + }, + { + "epoch": 0.45618619844834624, + "grad_norm": 2.1742756366729736, + "learning_rate": 1.0891154422788606e-05, + "loss": 2.4620388031005858, + "step": 167580 + }, + { + "epoch": 0.4562406424390908, + "grad_norm": 2.166271209716797, + "learning_rate": 1.0890064058879653e-05, + "loss": 2.4606746673583983, + "step": 167600 + }, + { + "epoch": 0.4562950864298353, + "grad_norm": 2.134289264678955, + "learning_rate": 1.0888973694970697e-05, + "loss": 2.4977514266967775, + "step": 167620 + }, + { + "epoch": 0.45634953042057985, + "grad_norm": 2.1248483657836914, + "learning_rate": 1.0887883331061743e-05, + "loss": 2.4853076934814453, + "step": 167640 + }, + { + "epoch": 0.45640397441132435, + "grad_norm": 2.22279691696167, + "learning_rate": 1.0886792967152789e-05, + "loss": 2.4671628952026365, + "step": 167660 + }, + { + "epoch": 0.45645841840206886, + "grad_norm": 2.0997536182403564, + "learning_rate": 1.0885702603243833e-05, + "loss": 2.4815988540649414, + "step": 167680 + }, + { + "epoch": 0.4565128623928134, + "grad_norm": 2.147636651992798, + "learning_rate": 1.088461223933488e-05, + "loss": 2.463917541503906, + "step": 167700 + }, + { + "epoch": 0.4565673063835579, + "grad_norm": 2.126696825027466, + "learning_rate": 1.0883521875425924e-05, + "loss": 2.458944892883301, + "step": 167720 + }, + { + "epoch": 0.4566217503743024, + "grad_norm": 2.145632743835449, + "learning_rate": 1.0882431511516969e-05, + "loss": 2.461225128173828, + "step": 167740 + }, + { + "epoch": 0.456676194365047, + "grad_norm": 2.1389758586883545, + "learning_rate": 1.0881341147608016e-05, + "loss": 2.5062265396118164, + "step": 167760 + }, + { + "epoch": 0.4567306383557915, + "grad_norm": 2.0997767448425293, + "learning_rate": 1.088025078369906e-05, + "loss": 2.484519195556641, + "step": 167780 + }, + { + "epoch": 0.456785082346536, + "grad_norm": 2.3304457664489746, + "learning_rate": 1.0879160419790106e-05, + "loss": 2.51772403717041, + "step": 167800 + }, + { + "epoch": 0.45683952633728053, + "grad_norm": 2.1653130054473877, + "learning_rate": 1.0878070055881152e-05, + "loss": 2.47863712310791, + "step": 167820 + }, + { + "epoch": 0.45689397032802503, + "grad_norm": 2.2534053325653076, + "learning_rate": 1.0876979691972196e-05, + "loss": 2.458901596069336, + "step": 167840 + }, + { + "epoch": 0.4569484143187696, + "grad_norm": 2.270189046859741, + "learning_rate": 1.0875889328063243e-05, + "loss": 2.4966638565063475, + "step": 167860 + }, + { + "epoch": 0.4570028583095141, + "grad_norm": 2.4315574169158936, + "learning_rate": 1.0874798964154288e-05, + "loss": 2.469136428833008, + "step": 167880 + }, + { + "epoch": 0.4570573023002586, + "grad_norm": 2.231111526489258, + "learning_rate": 1.0873708600245332e-05, + "loss": 2.4892593383789063, + "step": 167900 + }, + { + "epoch": 0.45711174629100315, + "grad_norm": 2.177093267440796, + "learning_rate": 1.087261823633638e-05, + "loss": 2.503326988220215, + "step": 167920 + }, + { + "epoch": 0.45716619028174765, + "grad_norm": 2.0803239345550537, + "learning_rate": 1.0871527872427423e-05, + "loss": 2.4820070266723633, + "step": 167940 + }, + { + "epoch": 0.45722063427249215, + "grad_norm": 2.113713264465332, + "learning_rate": 1.0870437508518469e-05, + "loss": 2.4901155471801757, + "step": 167960 + }, + { + "epoch": 0.4572750782632367, + "grad_norm": 2.0717031955718994, + "learning_rate": 1.0869347144609515e-05, + "loss": 2.4914989471435547, + "step": 167980 + }, + { + "epoch": 0.4573295222539812, + "grad_norm": 2.1789064407348633, + "learning_rate": 1.0868256780700559e-05, + "loss": 2.513597106933594, + "step": 168000 + }, + { + "epoch": 0.4573839662447257, + "grad_norm": 2.164116144180298, + "learning_rate": 1.0867166416791607e-05, + "loss": 2.5087865829467773, + "step": 168020 + }, + { + "epoch": 0.45743841023547027, + "grad_norm": 2.1581978797912598, + "learning_rate": 1.086607605288265e-05, + "loss": 2.517975616455078, + "step": 168040 + }, + { + "epoch": 0.4574928542262148, + "grad_norm": 2.09454607963562, + "learning_rate": 1.0864985688973695e-05, + "loss": 2.475987434387207, + "step": 168060 + }, + { + "epoch": 0.45754729821695933, + "grad_norm": 2.3227107524871826, + "learning_rate": 1.0863895325064742e-05, + "loss": 2.487639617919922, + "step": 168080 + }, + { + "epoch": 0.45760174220770383, + "grad_norm": 2.151620626449585, + "learning_rate": 1.0862804961155786e-05, + "loss": 2.493421173095703, + "step": 168100 + }, + { + "epoch": 0.45765618619844833, + "grad_norm": 2.1945552825927734, + "learning_rate": 1.0861714597246832e-05, + "loss": 2.5188085556030275, + "step": 168120 + }, + { + "epoch": 0.4577106301891929, + "grad_norm": 2.313439130783081, + "learning_rate": 1.0860624233337878e-05, + "loss": 2.4886348724365233, + "step": 168140 + }, + { + "epoch": 0.4577650741799374, + "grad_norm": 2.109018325805664, + "learning_rate": 1.0859533869428922e-05, + "loss": 2.4648488998413085, + "step": 168160 + }, + { + "epoch": 0.4578195181706819, + "grad_norm": 2.23113751411438, + "learning_rate": 1.085844350551997e-05, + "loss": 2.4734188079833985, + "step": 168180 + }, + { + "epoch": 0.45787396216142645, + "grad_norm": 2.2584738731384277, + "learning_rate": 1.0857353141611014e-05, + "loss": 2.460968780517578, + "step": 168200 + }, + { + "epoch": 0.45792840615217095, + "grad_norm": 2.1016764640808105, + "learning_rate": 1.0856262777702058e-05, + "loss": 2.504189872741699, + "step": 168220 + }, + { + "epoch": 0.45798285014291545, + "grad_norm": 2.2530314922332764, + "learning_rate": 1.0855172413793105e-05, + "loss": 2.49277286529541, + "step": 168240 + }, + { + "epoch": 0.45803729413366, + "grad_norm": 2.163874387741089, + "learning_rate": 1.085408204988415e-05, + "loss": 2.498113250732422, + "step": 168260 + }, + { + "epoch": 0.4580917381244045, + "grad_norm": 2.204810380935669, + "learning_rate": 1.0852991685975195e-05, + "loss": 2.490361976623535, + "step": 168280 + }, + { + "epoch": 0.458146182115149, + "grad_norm": 2.1644275188446045, + "learning_rate": 1.0851901322066241e-05, + "loss": 2.4999950408935545, + "step": 168300 + }, + { + "epoch": 0.45820062610589357, + "grad_norm": 2.157702922821045, + "learning_rate": 1.0850810958157285e-05, + "loss": 2.4942758560180662, + "step": 168320 + }, + { + "epoch": 0.4582550700966381, + "grad_norm": 2.2473702430725098, + "learning_rate": 1.0849720594248333e-05, + "loss": 2.5030467987060545, + "step": 168340 + }, + { + "epoch": 0.45830951408738263, + "grad_norm": 2.438164710998535, + "learning_rate": 1.0848630230339377e-05, + "loss": 2.4789522171020506, + "step": 168360 + }, + { + "epoch": 0.45836395807812713, + "grad_norm": 2.0563669204711914, + "learning_rate": 1.0847539866430421e-05, + "loss": 2.49810905456543, + "step": 168380 + }, + { + "epoch": 0.45841840206887163, + "grad_norm": 2.289726495742798, + "learning_rate": 1.0846449502521468e-05, + "loss": 2.4866172790527346, + "step": 168400 + }, + { + "epoch": 0.4584728460596162, + "grad_norm": 2.121730089187622, + "learning_rate": 1.0845359138612513e-05, + "loss": 2.4643640518188477, + "step": 168420 + }, + { + "epoch": 0.4585272900503607, + "grad_norm": 2.1169850826263428, + "learning_rate": 1.0844268774703558e-05, + "loss": 2.5041858673095705, + "step": 168440 + }, + { + "epoch": 0.4585817340411052, + "grad_norm": 2.3157389163970947, + "learning_rate": 1.0843178410794604e-05, + "loss": 2.5144596099853516, + "step": 168460 + }, + { + "epoch": 0.45863617803184975, + "grad_norm": 2.120133876800537, + "learning_rate": 1.0842088046885648e-05, + "loss": 2.442667579650879, + "step": 168480 + }, + { + "epoch": 0.45869062202259425, + "grad_norm": 2.061565399169922, + "learning_rate": 1.0840997682976696e-05, + "loss": 2.4645124435424806, + "step": 168500 + }, + { + "epoch": 0.45874506601333875, + "grad_norm": 2.075340986251831, + "learning_rate": 1.083990731906774e-05, + "loss": 2.4532503128051757, + "step": 168520 + }, + { + "epoch": 0.4587995100040833, + "grad_norm": 2.2282280921936035, + "learning_rate": 1.0838816955158784e-05, + "loss": 2.484598731994629, + "step": 168540 + }, + { + "epoch": 0.4588539539948278, + "grad_norm": 2.137260675430298, + "learning_rate": 1.0837726591249832e-05, + "loss": 2.4784162521362303, + "step": 168560 + }, + { + "epoch": 0.45890839798557237, + "grad_norm": 2.1744115352630615, + "learning_rate": 1.0836636227340876e-05, + "loss": 2.4839014053344726, + "step": 168580 + }, + { + "epoch": 0.45896284197631687, + "grad_norm": 2.1867446899414062, + "learning_rate": 1.0835545863431921e-05, + "loss": 2.496309280395508, + "step": 168600 + }, + { + "epoch": 0.4590172859670614, + "grad_norm": 2.201099157333374, + "learning_rate": 1.0834455499522967e-05, + "loss": 2.5001182556152344, + "step": 168620 + }, + { + "epoch": 0.45907172995780593, + "grad_norm": 2.4317243099212646, + "learning_rate": 1.0833365135614011e-05, + "loss": 2.490625762939453, + "step": 168640 + }, + { + "epoch": 0.45912617394855043, + "grad_norm": 2.24676513671875, + "learning_rate": 1.0832274771705057e-05, + "loss": 2.438681411743164, + "step": 168660 + }, + { + "epoch": 0.45918061793929493, + "grad_norm": 2.238598346710205, + "learning_rate": 1.0831184407796103e-05, + "loss": 2.4723228454589843, + "step": 168680 + }, + { + "epoch": 0.4592350619300395, + "grad_norm": 2.114241361618042, + "learning_rate": 1.0830094043887147e-05, + "loss": 2.4570735931396483, + "step": 168700 + }, + { + "epoch": 0.459289505920784, + "grad_norm": 2.0780067443847656, + "learning_rate": 1.0829003679978195e-05, + "loss": 2.5042816162109376, + "step": 168720 + }, + { + "epoch": 0.4593439499115285, + "grad_norm": 2.099752187728882, + "learning_rate": 1.0827913316069239e-05, + "loss": 2.4896902084350585, + "step": 168740 + }, + { + "epoch": 0.45939839390227305, + "grad_norm": 2.841641902923584, + "learning_rate": 1.0826822952160285e-05, + "loss": 2.484110641479492, + "step": 168760 + }, + { + "epoch": 0.45945283789301755, + "grad_norm": 2.221170663833618, + "learning_rate": 1.082573258825133e-05, + "loss": 2.460721015930176, + "step": 168780 + }, + { + "epoch": 0.45950728188376205, + "grad_norm": 2.2352709770202637, + "learning_rate": 1.0824642224342374e-05, + "loss": 2.504267692565918, + "step": 168800 + }, + { + "epoch": 0.4595617258745066, + "grad_norm": 2.13904070854187, + "learning_rate": 1.082355186043342e-05, + "loss": 2.479730796813965, + "step": 168820 + }, + { + "epoch": 0.4596161698652511, + "grad_norm": 2.0873658657073975, + "learning_rate": 1.0822461496524466e-05, + "loss": 2.496296501159668, + "step": 168840 + }, + { + "epoch": 0.45967061385599567, + "grad_norm": 2.207737445831299, + "learning_rate": 1.0821371132615512e-05, + "loss": 2.4814279556274412, + "step": 168860 + }, + { + "epoch": 0.45972505784674017, + "grad_norm": 2.504572629928589, + "learning_rate": 1.0820280768706558e-05, + "loss": 2.4885936737060548, + "step": 168880 + }, + { + "epoch": 0.45977950183748467, + "grad_norm": 2.1369011402130127, + "learning_rate": 1.0819190404797602e-05, + "loss": 2.477919006347656, + "step": 168900 + }, + { + "epoch": 0.45983394582822923, + "grad_norm": 2.110668659210205, + "learning_rate": 1.0818100040888648e-05, + "loss": 2.4948326110839845, + "step": 168920 + }, + { + "epoch": 0.45988838981897373, + "grad_norm": 2.279761552810669, + "learning_rate": 1.0817009676979693e-05, + "loss": 2.513802909851074, + "step": 168940 + }, + { + "epoch": 0.45994283380971823, + "grad_norm": 2.156743049621582, + "learning_rate": 1.0815919313070738e-05, + "loss": 2.4958284378051756, + "step": 168960 + }, + { + "epoch": 0.4599972778004628, + "grad_norm": 2.198329448699951, + "learning_rate": 1.0814828949161783e-05, + "loss": 2.495876121520996, + "step": 168980 + }, + { + "epoch": 0.4600517217912073, + "grad_norm": 2.314591884613037, + "learning_rate": 1.081373858525283e-05, + "loss": 2.4778535842895506, + "step": 169000 + }, + { + "epoch": 0.4601061657819518, + "grad_norm": 2.254861354827881, + "learning_rate": 1.0812648221343875e-05, + "loss": 2.4863956451416014, + "step": 169020 + }, + { + "epoch": 0.46016060977269635, + "grad_norm": 2.190359592437744, + "learning_rate": 1.081155785743492e-05, + "loss": 2.4807918548583983, + "step": 169040 + }, + { + "epoch": 0.46021505376344085, + "grad_norm": 2.270272970199585, + "learning_rate": 1.0810467493525965e-05, + "loss": 2.4664690017700197, + "step": 169060 + }, + { + "epoch": 0.4602694977541854, + "grad_norm": 2.1350595951080322, + "learning_rate": 1.080937712961701e-05, + "loss": 2.4741397857666017, + "step": 169080 + }, + { + "epoch": 0.4603239417449299, + "grad_norm": 2.1204397678375244, + "learning_rate": 1.0808286765708057e-05, + "loss": 2.4827314376831056, + "step": 169100 + }, + { + "epoch": 0.4603783857356744, + "grad_norm": 2.132460355758667, + "learning_rate": 1.08071964017991e-05, + "loss": 2.482879066467285, + "step": 169120 + }, + { + "epoch": 0.46043282972641897, + "grad_norm": 2.913285255432129, + "learning_rate": 1.0806106037890146e-05, + "loss": 2.492320442199707, + "step": 169140 + }, + { + "epoch": 0.46048727371716347, + "grad_norm": 2.2299160957336426, + "learning_rate": 1.0805015673981192e-05, + "loss": 2.473530387878418, + "step": 169160 + }, + { + "epoch": 0.46054171770790797, + "grad_norm": 2.2754087448120117, + "learning_rate": 1.0803925310072238e-05, + "loss": 2.4982418060302733, + "step": 169180 + }, + { + "epoch": 0.46059616169865253, + "grad_norm": 2.6096813678741455, + "learning_rate": 1.0802834946163284e-05, + "loss": 2.486964797973633, + "step": 169200 + }, + { + "epoch": 0.46065060568939703, + "grad_norm": 2.419111728668213, + "learning_rate": 1.0801744582254328e-05, + "loss": 2.4700595855712892, + "step": 169220 + }, + { + "epoch": 0.46070504968014153, + "grad_norm": 2.1969969272613525, + "learning_rate": 1.0800654218345374e-05, + "loss": 2.5136739730834963, + "step": 169240 + }, + { + "epoch": 0.4607594936708861, + "grad_norm": 2.1478803157806396, + "learning_rate": 1.079956385443642e-05, + "loss": 2.508563232421875, + "step": 169260 + }, + { + "epoch": 0.4608139376616306, + "grad_norm": 2.147397756576538, + "learning_rate": 1.0798473490527464e-05, + "loss": 2.481501007080078, + "step": 169280 + }, + { + "epoch": 0.4608683816523751, + "grad_norm": 2.1393954753875732, + "learning_rate": 1.079738312661851e-05, + "loss": 2.4791597366333007, + "step": 169300 + }, + { + "epoch": 0.46092282564311965, + "grad_norm": 2.4164485931396484, + "learning_rate": 1.0796292762709555e-05, + "loss": 2.490243911743164, + "step": 169320 + }, + { + "epoch": 0.46097726963386415, + "grad_norm": 2.1830129623413086, + "learning_rate": 1.0795202398800601e-05, + "loss": 2.4853187561035157, + "step": 169340 + }, + { + "epoch": 0.4610317136246087, + "grad_norm": 2.157463788986206, + "learning_rate": 1.0794112034891647e-05, + "loss": 2.5100992202758787, + "step": 169360 + }, + { + "epoch": 0.4610861576153532, + "grad_norm": 2.1391334533691406, + "learning_rate": 1.0793021670982691e-05, + "loss": 2.5164445877075194, + "step": 169380 + }, + { + "epoch": 0.4611406016060977, + "grad_norm": 2.4327304363250732, + "learning_rate": 1.0791931307073737e-05, + "loss": 2.457286262512207, + "step": 169400 + }, + { + "epoch": 0.46119504559684227, + "grad_norm": 2.1122498512268066, + "learning_rate": 1.0790840943164783e-05, + "loss": 2.4921543121337892, + "step": 169420 + }, + { + "epoch": 0.46124948958758677, + "grad_norm": 2.1437621116638184, + "learning_rate": 1.0789750579255827e-05, + "loss": 2.4638952255249023, + "step": 169440 + }, + { + "epoch": 0.46130393357833127, + "grad_norm": 2.1505417823791504, + "learning_rate": 1.0788660215346873e-05, + "loss": 2.4751398086547853, + "step": 169460 + }, + { + "epoch": 0.46135837756907583, + "grad_norm": 2.1326284408569336, + "learning_rate": 1.0787569851437918e-05, + "loss": 2.497789764404297, + "step": 169480 + }, + { + "epoch": 0.46141282155982033, + "grad_norm": 2.1464688777923584, + "learning_rate": 1.0786479487528964e-05, + "loss": 2.4709379196166994, + "step": 169500 + }, + { + "epoch": 0.46146726555056483, + "grad_norm": 2.293449878692627, + "learning_rate": 1.078538912362001e-05, + "loss": 2.486599159240723, + "step": 169520 + }, + { + "epoch": 0.4615217095413094, + "grad_norm": 2.264880895614624, + "learning_rate": 1.0784298759711054e-05, + "loss": 2.492595100402832, + "step": 169540 + }, + { + "epoch": 0.4615761535320539, + "grad_norm": 2.18350887298584, + "learning_rate": 1.07832083958021e-05, + "loss": 2.5010066986083985, + "step": 169560 + }, + { + "epoch": 0.46163059752279845, + "grad_norm": 2.121561288833618, + "learning_rate": 1.0782118031893146e-05, + "loss": 2.459453582763672, + "step": 169580 + }, + { + "epoch": 0.46168504151354295, + "grad_norm": 2.231008529663086, + "learning_rate": 1.078102766798419e-05, + "loss": 2.506612777709961, + "step": 169600 + }, + { + "epoch": 0.46173948550428745, + "grad_norm": 2.076007604598999, + "learning_rate": 1.0779937304075236e-05, + "loss": 2.4569766998291014, + "step": 169620 + }, + { + "epoch": 0.461793929495032, + "grad_norm": 2.1125099658966064, + "learning_rate": 1.0778846940166282e-05, + "loss": 2.4686832427978516, + "step": 169640 + }, + { + "epoch": 0.4618483734857765, + "grad_norm": 2.111194610595703, + "learning_rate": 1.0777756576257327e-05, + "loss": 2.480632019042969, + "step": 169660 + }, + { + "epoch": 0.461902817476521, + "grad_norm": 2.572097063064575, + "learning_rate": 1.0776666212348373e-05, + "loss": 2.491611671447754, + "step": 169680 + }, + { + "epoch": 0.46195726146726557, + "grad_norm": 2.2184805870056152, + "learning_rate": 1.0775575848439417e-05, + "loss": 2.4921403884887696, + "step": 169700 + }, + { + "epoch": 0.46201170545801007, + "grad_norm": 2.306889772415161, + "learning_rate": 1.0774485484530463e-05, + "loss": 2.461886405944824, + "step": 169720 + }, + { + "epoch": 0.46206614944875457, + "grad_norm": 2.174758195877075, + "learning_rate": 1.0773395120621509e-05, + "loss": 2.4732534408569338, + "step": 169740 + }, + { + "epoch": 0.4621205934394991, + "grad_norm": 2.1213314533233643, + "learning_rate": 1.0772304756712553e-05, + "loss": 2.4713369369506837, + "step": 169760 + }, + { + "epoch": 0.46217503743024363, + "grad_norm": 2.4402856826782227, + "learning_rate": 1.0771214392803599e-05, + "loss": 2.4777959823608398, + "step": 169780 + }, + { + "epoch": 0.46222948142098813, + "grad_norm": 2.165807008743286, + "learning_rate": 1.0770124028894645e-05, + "loss": 2.5131591796875, + "step": 169800 + }, + { + "epoch": 0.4622839254117327, + "grad_norm": 2.1132354736328125, + "learning_rate": 1.076903366498569e-05, + "loss": 2.4936845779418944, + "step": 169820 + }, + { + "epoch": 0.4623383694024772, + "grad_norm": 2.107004165649414, + "learning_rate": 1.0767943301076735e-05, + "loss": 2.48571662902832, + "step": 169840 + }, + { + "epoch": 0.46239281339322175, + "grad_norm": 2.1391165256500244, + "learning_rate": 1.076685293716778e-05, + "loss": 2.4737060546875, + "step": 169860 + }, + { + "epoch": 0.46244725738396625, + "grad_norm": 2.2022016048431396, + "learning_rate": 1.0765762573258826e-05, + "loss": 2.4799478530883787, + "step": 169880 + }, + { + "epoch": 0.46250170137471075, + "grad_norm": 2.1199588775634766, + "learning_rate": 1.0764672209349872e-05, + "loss": 2.4817455291748045, + "step": 169900 + }, + { + "epoch": 0.4625561453654553, + "grad_norm": 2.3696067333221436, + "learning_rate": 1.0763581845440916e-05, + "loss": 2.49869327545166, + "step": 169920 + }, + { + "epoch": 0.4626105893561998, + "grad_norm": 2.1429028511047363, + "learning_rate": 1.0762491481531962e-05, + "loss": 2.4935684204101562, + "step": 169940 + }, + { + "epoch": 0.4626650333469443, + "grad_norm": 2.2192296981811523, + "learning_rate": 1.0761401117623008e-05, + "loss": 2.4631114959716798, + "step": 169960 + }, + { + "epoch": 0.46271947733768887, + "grad_norm": 2.074890375137329, + "learning_rate": 1.0760310753714053e-05, + "loss": 2.486598587036133, + "step": 169980 + }, + { + "epoch": 0.46277392132843337, + "grad_norm": 2.222118854522705, + "learning_rate": 1.0759220389805098e-05, + "loss": 2.4796510696411134, + "step": 170000 + }, + { + "epoch": 0.46282836531917787, + "grad_norm": 2.2252426147460938, + "learning_rate": 1.0758130025896143e-05, + "loss": 2.48128719329834, + "step": 170020 + }, + { + "epoch": 0.4628828093099224, + "grad_norm": 2.10487961769104, + "learning_rate": 1.075703966198719e-05, + "loss": 2.473830986022949, + "step": 170040 + }, + { + "epoch": 0.46293725330066693, + "grad_norm": 2.2234373092651367, + "learning_rate": 1.0755949298078235e-05, + "loss": 2.4688907623291017, + "step": 170060 + }, + { + "epoch": 0.4629916972914115, + "grad_norm": 2.0647685527801514, + "learning_rate": 1.075485893416928e-05, + "loss": 2.461118698120117, + "step": 170080 + }, + { + "epoch": 0.463046141282156, + "grad_norm": 2.2529428005218506, + "learning_rate": 1.0753768570260325e-05, + "loss": 2.4892562866210937, + "step": 170100 + }, + { + "epoch": 0.4631005852729005, + "grad_norm": 2.1851677894592285, + "learning_rate": 1.075267820635137e-05, + "loss": 2.4858821868896483, + "step": 170120 + }, + { + "epoch": 0.46315502926364505, + "grad_norm": 2.2531914710998535, + "learning_rate": 1.0751587842442417e-05, + "loss": 2.4902111053466798, + "step": 170140 + }, + { + "epoch": 0.46320947325438955, + "grad_norm": 2.130250930786133, + "learning_rate": 1.075049747853346e-05, + "loss": 2.4809101104736326, + "step": 170160 + }, + { + "epoch": 0.46326391724513405, + "grad_norm": 2.1589157581329346, + "learning_rate": 1.0749407114624507e-05, + "loss": 2.4506963729858398, + "step": 170180 + }, + { + "epoch": 0.4633183612358786, + "grad_norm": 2.0104687213897705, + "learning_rate": 1.0748316750715552e-05, + "loss": 2.491603660583496, + "step": 170200 + }, + { + "epoch": 0.4633728052266231, + "grad_norm": 2.1235086917877197, + "learning_rate": 1.0747226386806598e-05, + "loss": 2.4880701065063477, + "step": 170220 + }, + { + "epoch": 0.4634272492173676, + "grad_norm": 2.1067698001861572, + "learning_rate": 1.0746136022897642e-05, + "loss": 2.4706775665283205, + "step": 170240 + }, + { + "epoch": 0.46348169320811217, + "grad_norm": 2.0988686084747314, + "learning_rate": 1.0745045658988688e-05, + "loss": 2.5060714721679687, + "step": 170260 + }, + { + "epoch": 0.46353613719885667, + "grad_norm": 2.1261162757873535, + "learning_rate": 1.0743955295079734e-05, + "loss": 2.4913877487182616, + "step": 170280 + }, + { + "epoch": 0.4635905811896012, + "grad_norm": 2.980536937713623, + "learning_rate": 1.074286493117078e-05, + "loss": 2.475872039794922, + "step": 170300 + }, + { + "epoch": 0.4636450251803457, + "grad_norm": 2.155101776123047, + "learning_rate": 1.0741774567261824e-05, + "loss": 2.463627052307129, + "step": 170320 + }, + { + "epoch": 0.46369946917109023, + "grad_norm": 2.167825698852539, + "learning_rate": 1.074068420335287e-05, + "loss": 2.4839799880981444, + "step": 170340 + }, + { + "epoch": 0.4637539131618348, + "grad_norm": 2.203510046005249, + "learning_rate": 1.0739593839443915e-05, + "loss": 2.5392364501953124, + "step": 170360 + }, + { + "epoch": 0.4638083571525793, + "grad_norm": 2.2054638862609863, + "learning_rate": 1.0738503475534961e-05, + "loss": 2.5172039031982423, + "step": 170380 + }, + { + "epoch": 0.4638628011433238, + "grad_norm": 2.3325247764587402, + "learning_rate": 1.0737413111626005e-05, + "loss": 2.5133459091186525, + "step": 170400 + }, + { + "epoch": 0.46391724513406835, + "grad_norm": 2.1111717224121094, + "learning_rate": 1.0736322747717051e-05, + "loss": 2.4865745544433593, + "step": 170420 + }, + { + "epoch": 0.46397168912481285, + "grad_norm": 2.1489617824554443, + "learning_rate": 1.0735232383808097e-05, + "loss": 2.4643653869628905, + "step": 170440 + }, + { + "epoch": 0.46402613311555735, + "grad_norm": 2.220738649368286, + "learning_rate": 1.0734142019899143e-05, + "loss": 2.4863283157348635, + "step": 170460 + }, + { + "epoch": 0.4640805771063019, + "grad_norm": 2.389033555984497, + "learning_rate": 1.0733051655990187e-05, + "loss": 2.4665584564208984, + "step": 170480 + }, + { + "epoch": 0.4641350210970464, + "grad_norm": 2.1474411487579346, + "learning_rate": 1.0731961292081233e-05, + "loss": 2.477033996582031, + "step": 170500 + }, + { + "epoch": 0.4641894650877909, + "grad_norm": 2.2092087268829346, + "learning_rate": 1.0730870928172278e-05, + "loss": 2.4727109909057616, + "step": 170520 + }, + { + "epoch": 0.46424390907853547, + "grad_norm": 2.297344446182251, + "learning_rate": 1.0729780564263324e-05, + "loss": 2.486833953857422, + "step": 170540 + }, + { + "epoch": 0.46429835306927997, + "grad_norm": 2.2178008556365967, + "learning_rate": 1.072869020035437e-05, + "loss": 2.46020393371582, + "step": 170560 + }, + { + "epoch": 0.4643527970600245, + "grad_norm": 2.1094932556152344, + "learning_rate": 1.0727599836445414e-05, + "loss": 2.4916881561279296, + "step": 170580 + }, + { + "epoch": 0.464407241050769, + "grad_norm": 2.195709228515625, + "learning_rate": 1.072650947253646e-05, + "loss": 2.4988468170166014, + "step": 170600 + }, + { + "epoch": 0.46446168504151353, + "grad_norm": 2.1466495990753174, + "learning_rate": 1.0725419108627506e-05, + "loss": 2.4825860977172853, + "step": 170620 + }, + { + "epoch": 0.4645161290322581, + "grad_norm": 2.308899402618408, + "learning_rate": 1.072432874471855e-05, + "loss": 2.4547542572021483, + "step": 170640 + }, + { + "epoch": 0.4645705730230026, + "grad_norm": 2.1545968055725098, + "learning_rate": 1.0723238380809596e-05, + "loss": 2.483476257324219, + "step": 170660 + }, + { + "epoch": 0.4646250170137471, + "grad_norm": 2.2347824573516846, + "learning_rate": 1.0722148016900642e-05, + "loss": 2.4999534606933596, + "step": 170680 + }, + { + "epoch": 0.46467946100449165, + "grad_norm": 2.199592351913452, + "learning_rate": 1.0721057652991687e-05, + "loss": 2.492742919921875, + "step": 170700 + }, + { + "epoch": 0.46473390499523615, + "grad_norm": 2.265310525894165, + "learning_rate": 1.0719967289082733e-05, + "loss": 2.4604719161987303, + "step": 170720 + }, + { + "epoch": 0.46478834898598065, + "grad_norm": 2.1282806396484375, + "learning_rate": 1.0718876925173777e-05, + "loss": 2.4968706130981446, + "step": 170740 + }, + { + "epoch": 0.4648427929767252, + "grad_norm": 2.1413469314575195, + "learning_rate": 1.0717786561264823e-05, + "loss": 2.496260070800781, + "step": 170760 + }, + { + "epoch": 0.4648972369674697, + "grad_norm": 2.213369131088257, + "learning_rate": 1.0716696197355869e-05, + "loss": 2.466230010986328, + "step": 170780 + }, + { + "epoch": 0.46495168095821426, + "grad_norm": 2.0661158561706543, + "learning_rate": 1.0715605833446913e-05, + "loss": 2.488602066040039, + "step": 170800 + }, + { + "epoch": 0.46500612494895877, + "grad_norm": 2.148198366165161, + "learning_rate": 1.0714515469537959e-05, + "loss": 2.4779895782470702, + "step": 170820 + }, + { + "epoch": 0.46506056893970327, + "grad_norm": 2.212320566177368, + "learning_rate": 1.0713425105629005e-05, + "loss": 2.4811481475830077, + "step": 170840 + }, + { + "epoch": 0.4651150129304478, + "grad_norm": 2.1449084281921387, + "learning_rate": 1.0712334741720049e-05, + "loss": 2.4905960083007814, + "step": 170860 + }, + { + "epoch": 0.4651694569211923, + "grad_norm": 2.123241662979126, + "learning_rate": 1.0711244377811096e-05, + "loss": 2.4746896743774416, + "step": 170880 + }, + { + "epoch": 0.4652239009119368, + "grad_norm": 2.1946370601654053, + "learning_rate": 1.071015401390214e-05, + "loss": 2.485397148132324, + "step": 170900 + }, + { + "epoch": 0.4652783449026814, + "grad_norm": 2.1161065101623535, + "learning_rate": 1.0709063649993186e-05, + "loss": 2.462514877319336, + "step": 170920 + }, + { + "epoch": 0.4653327888934259, + "grad_norm": 2.194892168045044, + "learning_rate": 1.0707973286084232e-05, + "loss": 2.466323471069336, + "step": 170940 + }, + { + "epoch": 0.4653872328841704, + "grad_norm": 2.2842564582824707, + "learning_rate": 1.0706882922175276e-05, + "loss": 2.4854494094848634, + "step": 170960 + }, + { + "epoch": 0.46544167687491494, + "grad_norm": 2.1308295726776123, + "learning_rate": 1.0705792558266322e-05, + "loss": 2.5007869720458986, + "step": 170980 + }, + { + "epoch": 0.46549612086565945, + "grad_norm": 2.1999990940093994, + "learning_rate": 1.0704702194357368e-05, + "loss": 2.497418212890625, + "step": 171000 + }, + { + "epoch": 0.46555056485640395, + "grad_norm": 2.1978979110717773, + "learning_rate": 1.0703611830448412e-05, + "loss": 2.4947994232177733, + "step": 171020 + }, + { + "epoch": 0.4656050088471485, + "grad_norm": 2.1773183345794678, + "learning_rate": 1.070252146653946e-05, + "loss": 2.5092193603515627, + "step": 171040 + }, + { + "epoch": 0.465659452837893, + "grad_norm": 2.2387118339538574, + "learning_rate": 1.0701431102630503e-05, + "loss": 2.505093574523926, + "step": 171060 + }, + { + "epoch": 0.46571389682863756, + "grad_norm": 2.2804744243621826, + "learning_rate": 1.070034073872155e-05, + "loss": 2.5074945449829102, + "step": 171080 + }, + { + "epoch": 0.46576834081938207, + "grad_norm": 2.1450693607330322, + "learning_rate": 1.0699250374812595e-05, + "loss": 2.456259536743164, + "step": 171100 + }, + { + "epoch": 0.46582278481012657, + "grad_norm": 2.322789430618286, + "learning_rate": 1.069816001090364e-05, + "loss": 2.4696640014648437, + "step": 171120 + }, + { + "epoch": 0.4658772288008711, + "grad_norm": 2.253563165664673, + "learning_rate": 1.0697069646994685e-05, + "loss": 2.4896429061889647, + "step": 171140 + }, + { + "epoch": 0.4659316727916156, + "grad_norm": 2.066602945327759, + "learning_rate": 1.069597928308573e-05, + "loss": 2.4644292831420898, + "step": 171160 + }, + { + "epoch": 0.4659861167823601, + "grad_norm": 2.665534734725952, + "learning_rate": 1.0694888919176775e-05, + "loss": 2.483717918395996, + "step": 171180 + }, + { + "epoch": 0.4660405607731047, + "grad_norm": 2.22866153717041, + "learning_rate": 1.0693798555267822e-05, + "loss": 2.5116601943969727, + "step": 171200 + }, + { + "epoch": 0.4660950047638492, + "grad_norm": 2.139431953430176, + "learning_rate": 1.0692708191358867e-05, + "loss": 2.4924211502075195, + "step": 171220 + }, + { + "epoch": 0.4661494487545937, + "grad_norm": 2.2506766319274902, + "learning_rate": 1.0691617827449912e-05, + "loss": 2.49752197265625, + "step": 171240 + }, + { + "epoch": 0.46620389274533824, + "grad_norm": 2.0764806270599365, + "learning_rate": 1.0690527463540958e-05, + "loss": 2.4753210067749025, + "step": 171260 + }, + { + "epoch": 0.46625833673608275, + "grad_norm": 2.2049875259399414, + "learning_rate": 1.0689437099632002e-05, + "loss": 2.4670555114746096, + "step": 171280 + }, + { + "epoch": 0.4663127807268273, + "grad_norm": 2.180023670196533, + "learning_rate": 1.0688346735723048e-05, + "loss": 2.4958574295043947, + "step": 171300 + }, + { + "epoch": 0.4663672247175718, + "grad_norm": 2.330763816833496, + "learning_rate": 1.0687256371814094e-05, + "loss": 2.4684015274047852, + "step": 171320 + }, + { + "epoch": 0.4664216687083163, + "grad_norm": 2.14396595954895, + "learning_rate": 1.0686166007905138e-05, + "loss": 2.4855918884277344, + "step": 171340 + }, + { + "epoch": 0.46647611269906086, + "grad_norm": 2.172992467880249, + "learning_rate": 1.0685075643996186e-05, + "loss": 2.4691848754882812, + "step": 171360 + }, + { + "epoch": 0.46653055668980536, + "grad_norm": 2.100405216217041, + "learning_rate": 1.068398528008723e-05, + "loss": 2.499769592285156, + "step": 171380 + }, + { + "epoch": 0.46658500068054987, + "grad_norm": 2.14792799949646, + "learning_rate": 1.0682894916178275e-05, + "loss": 2.4980167388916015, + "step": 171400 + }, + { + "epoch": 0.4666394446712944, + "grad_norm": 2.1419131755828857, + "learning_rate": 1.0681804552269321e-05, + "loss": 2.4935985565185548, + "step": 171420 + }, + { + "epoch": 0.4666938886620389, + "grad_norm": 2.093740224838257, + "learning_rate": 1.0680714188360365e-05, + "loss": 2.480511474609375, + "step": 171440 + }, + { + "epoch": 0.4667483326527834, + "grad_norm": 2.211416006088257, + "learning_rate": 1.0679623824451411e-05, + "loss": 2.468441390991211, + "step": 171460 + }, + { + "epoch": 0.466802776643528, + "grad_norm": 2.3096842765808105, + "learning_rate": 1.0678533460542457e-05, + "loss": 2.506231117248535, + "step": 171480 + }, + { + "epoch": 0.4668572206342725, + "grad_norm": 2.1993134021759033, + "learning_rate": 1.0677443096633501e-05, + "loss": 2.489790344238281, + "step": 171500 + }, + { + "epoch": 0.466911664625017, + "grad_norm": 2.2472639083862305, + "learning_rate": 1.0676352732724549e-05, + "loss": 2.481598472595215, + "step": 171520 + }, + { + "epoch": 0.46696610861576154, + "grad_norm": 2.1449625492095947, + "learning_rate": 1.0675262368815593e-05, + "loss": 2.49000244140625, + "step": 171540 + }, + { + "epoch": 0.46702055260650605, + "grad_norm": 2.374549388885498, + "learning_rate": 1.0674172004906639e-05, + "loss": 2.4567115783691404, + "step": 171560 + }, + { + "epoch": 0.4670749965972506, + "grad_norm": 2.054697036743164, + "learning_rate": 1.0673081640997684e-05, + "loss": 2.485733222961426, + "step": 171580 + }, + { + "epoch": 0.4671294405879951, + "grad_norm": 2.309272050857544, + "learning_rate": 1.0671991277088728e-05, + "loss": 2.4907438278198244, + "step": 171600 + }, + { + "epoch": 0.4671838845787396, + "grad_norm": 2.073045253753662, + "learning_rate": 1.0670900913179774e-05, + "loss": 2.5069355010986327, + "step": 171620 + }, + { + "epoch": 0.46723832856948416, + "grad_norm": 2.2900025844573975, + "learning_rate": 1.066981054927082e-05, + "loss": 2.510921096801758, + "step": 171640 + }, + { + "epoch": 0.46729277256022866, + "grad_norm": 2.2805769443511963, + "learning_rate": 1.0668720185361864e-05, + "loss": 2.490864372253418, + "step": 171660 + }, + { + "epoch": 0.46734721655097317, + "grad_norm": 2.097734212875366, + "learning_rate": 1.0667629821452912e-05, + "loss": 2.480005645751953, + "step": 171680 + }, + { + "epoch": 0.4674016605417177, + "grad_norm": 2.142568349838257, + "learning_rate": 1.0666539457543956e-05, + "loss": 2.496059799194336, + "step": 171700 + }, + { + "epoch": 0.4674561045324622, + "grad_norm": 2.5109238624572754, + "learning_rate": 1.0665449093635002e-05, + "loss": 2.5012062072753904, + "step": 171720 + }, + { + "epoch": 0.4675105485232067, + "grad_norm": 5.887131690979004, + "learning_rate": 1.0664358729726047e-05, + "loss": 2.498223304748535, + "step": 171740 + }, + { + "epoch": 0.4675649925139513, + "grad_norm": 2.2119760513305664, + "learning_rate": 1.0663268365817092e-05, + "loss": 2.4858556747436524, + "step": 171760 + }, + { + "epoch": 0.4676194365046958, + "grad_norm": 2.3931143283843994, + "learning_rate": 1.0662178001908137e-05, + "loss": 2.4753097534179687, + "step": 171780 + }, + { + "epoch": 0.46767388049544034, + "grad_norm": 2.5939993858337402, + "learning_rate": 1.0661087637999183e-05, + "loss": 2.499074172973633, + "step": 171800 + }, + { + "epoch": 0.46772832448618484, + "grad_norm": 2.1282992362976074, + "learning_rate": 1.0659997274090227e-05, + "loss": 2.5232730865478517, + "step": 171820 + }, + { + "epoch": 0.46778276847692934, + "grad_norm": 2.2474095821380615, + "learning_rate": 1.0658906910181275e-05, + "loss": 2.4480239868164064, + "step": 171840 + }, + { + "epoch": 0.4678372124676739, + "grad_norm": 2.198737621307373, + "learning_rate": 1.0657816546272319e-05, + "loss": 2.4679283142089843, + "step": 171860 + }, + { + "epoch": 0.4678916564584184, + "grad_norm": 2.168530225753784, + "learning_rate": 1.0656726182363365e-05, + "loss": 2.4857139587402344, + "step": 171880 + }, + { + "epoch": 0.4679461004491629, + "grad_norm": 2.128817081451416, + "learning_rate": 1.065563581845441e-05, + "loss": 2.4625247955322265, + "step": 171900 + }, + { + "epoch": 0.46800054443990746, + "grad_norm": 2.071216583251953, + "learning_rate": 1.0654545454545455e-05, + "loss": 2.4936336517333983, + "step": 171920 + }, + { + "epoch": 0.46805498843065196, + "grad_norm": 2.223939895629883, + "learning_rate": 1.06534550906365e-05, + "loss": 2.486143684387207, + "step": 171940 + }, + { + "epoch": 0.46810943242139647, + "grad_norm": 2.1104984283447266, + "learning_rate": 1.0652364726727546e-05, + "loss": 2.485041618347168, + "step": 171960 + }, + { + "epoch": 0.468163876412141, + "grad_norm": 2.1155028343200684, + "learning_rate": 1.065127436281859e-05, + "loss": 2.4579254150390626, + "step": 171980 + }, + { + "epoch": 0.4682183204028855, + "grad_norm": 2.3143293857574463, + "learning_rate": 1.0650183998909638e-05, + "loss": 2.5145896911621093, + "step": 172000 + }, + { + "epoch": 0.46827276439363, + "grad_norm": 2.247551918029785, + "learning_rate": 1.0649093635000682e-05, + "loss": 2.4589677810668946, + "step": 172020 + }, + { + "epoch": 0.4683272083843746, + "grad_norm": 2.180497884750366, + "learning_rate": 1.0648003271091726e-05, + "loss": 2.4584468841552733, + "step": 172040 + }, + { + "epoch": 0.4683816523751191, + "grad_norm": 2.136936664581299, + "learning_rate": 1.0646912907182774e-05, + "loss": 2.457938003540039, + "step": 172060 + }, + { + "epoch": 0.46843609636586364, + "grad_norm": 2.4730725288391113, + "learning_rate": 1.0645822543273818e-05, + "loss": 2.4879695892333986, + "step": 172080 + }, + { + "epoch": 0.46849054035660814, + "grad_norm": 2.133291006088257, + "learning_rate": 1.0644732179364864e-05, + "loss": 2.484001350402832, + "step": 172100 + }, + { + "epoch": 0.46854498434735264, + "grad_norm": 2.304226875305176, + "learning_rate": 1.064364181545591e-05, + "loss": 2.48547477722168, + "step": 172120 + }, + { + "epoch": 0.4685994283380972, + "grad_norm": 2.363945960998535, + "learning_rate": 1.0642551451546953e-05, + "loss": 2.486141395568848, + "step": 172140 + }, + { + "epoch": 0.4686538723288417, + "grad_norm": 2.125856876373291, + "learning_rate": 1.0641461087638001e-05, + "loss": 2.4832246780395506, + "step": 172160 + }, + { + "epoch": 0.4687083163195862, + "grad_norm": 2.188157320022583, + "learning_rate": 1.0640370723729045e-05, + "loss": 2.4337472915649414, + "step": 172180 + }, + { + "epoch": 0.46876276031033076, + "grad_norm": 2.2619082927703857, + "learning_rate": 1.063928035982009e-05, + "loss": 2.476526069641113, + "step": 172200 + }, + { + "epoch": 0.46881720430107526, + "grad_norm": 2.478865623474121, + "learning_rate": 1.0638189995911137e-05, + "loss": 2.4941534042358398, + "step": 172220 + }, + { + "epoch": 0.46887164829181976, + "grad_norm": 2.1269185543060303, + "learning_rate": 1.063709963200218e-05, + "loss": 2.469106674194336, + "step": 172240 + }, + { + "epoch": 0.4689260922825643, + "grad_norm": 2.0954842567443848, + "learning_rate": 1.0636009268093228e-05, + "loss": 2.495567512512207, + "step": 172260 + }, + { + "epoch": 0.4689805362733088, + "grad_norm": 2.214362382888794, + "learning_rate": 1.0634918904184272e-05, + "loss": 2.470087242126465, + "step": 172280 + }, + { + "epoch": 0.4690349802640534, + "grad_norm": 2.140902042388916, + "learning_rate": 1.0633828540275317e-05, + "loss": 2.495945930480957, + "step": 172300 + }, + { + "epoch": 0.4690894242547979, + "grad_norm": 2.1177945137023926, + "learning_rate": 1.0632738176366364e-05, + "loss": 2.4597530364990234, + "step": 172320 + }, + { + "epoch": 0.4691438682455424, + "grad_norm": 2.176464557647705, + "learning_rate": 1.0631647812457408e-05, + "loss": 2.4980138778686523, + "step": 172340 + }, + { + "epoch": 0.46919831223628694, + "grad_norm": 2.1330227851867676, + "learning_rate": 1.0630557448548452e-05, + "loss": 2.515296745300293, + "step": 172360 + }, + { + "epoch": 0.46925275622703144, + "grad_norm": 2.1115920543670654, + "learning_rate": 1.06294670846395e-05, + "loss": 2.4734445571899415, + "step": 172380 + }, + { + "epoch": 0.46930720021777594, + "grad_norm": 2.8990495204925537, + "learning_rate": 1.0628376720730544e-05, + "loss": 2.4602920532226564, + "step": 172400 + }, + { + "epoch": 0.4693616442085205, + "grad_norm": 2.165583848953247, + "learning_rate": 1.0627286356821591e-05, + "loss": 2.481680488586426, + "step": 172420 + }, + { + "epoch": 0.469416088199265, + "grad_norm": 2.7818336486816406, + "learning_rate": 1.0626195992912636e-05, + "loss": 2.49293212890625, + "step": 172440 + }, + { + "epoch": 0.4694705321900095, + "grad_norm": 2.114314079284668, + "learning_rate": 1.062510562900368e-05, + "loss": 2.499658966064453, + "step": 172460 + }, + { + "epoch": 0.46952497618075406, + "grad_norm": 2.2273902893066406, + "learning_rate": 1.0624015265094727e-05, + "loss": 2.4695301055908203, + "step": 172480 + }, + { + "epoch": 0.46957942017149856, + "grad_norm": 2.147050142288208, + "learning_rate": 1.0622924901185771e-05, + "loss": 2.500374984741211, + "step": 172500 + }, + { + "epoch": 0.4696338641622431, + "grad_norm": 2.0994138717651367, + "learning_rate": 1.0621834537276815e-05, + "loss": 2.503457450866699, + "step": 172520 + }, + { + "epoch": 0.4696883081529876, + "grad_norm": 2.1962780952453613, + "learning_rate": 1.0620744173367863e-05, + "loss": 2.476124572753906, + "step": 172540 + }, + { + "epoch": 0.4697427521437321, + "grad_norm": 2.252504825592041, + "learning_rate": 1.0619653809458907e-05, + "loss": 2.4981128692626955, + "step": 172560 + }, + { + "epoch": 0.4697971961344767, + "grad_norm": 2.211182117462158, + "learning_rate": 1.0618563445549954e-05, + "loss": 2.4738903045654297, + "step": 172580 + }, + { + "epoch": 0.4698516401252212, + "grad_norm": 2.3087196350097656, + "learning_rate": 1.0617473081640999e-05, + "loss": 2.4610876083374023, + "step": 172600 + }, + { + "epoch": 0.4699060841159657, + "grad_norm": 2.402946949005127, + "learning_rate": 1.0616382717732043e-05, + "loss": 2.4621303558349608, + "step": 172620 + }, + { + "epoch": 0.46996052810671024, + "grad_norm": 2.2169158458709717, + "learning_rate": 1.061529235382309e-05, + "loss": 2.469877815246582, + "step": 172640 + }, + { + "epoch": 0.47001497209745474, + "grad_norm": 2.188267946243286, + "learning_rate": 1.0614201989914134e-05, + "loss": 2.538979721069336, + "step": 172660 + }, + { + "epoch": 0.47006941608819924, + "grad_norm": 2.1824283599853516, + "learning_rate": 1.0613111626005178e-05, + "loss": 2.478013801574707, + "step": 172680 + }, + { + "epoch": 0.4701238600789438, + "grad_norm": 2.438570261001587, + "learning_rate": 1.0612021262096226e-05, + "loss": 2.4669532775878906, + "step": 172700 + }, + { + "epoch": 0.4701783040696883, + "grad_norm": 2.2505600452423096, + "learning_rate": 1.061093089818727e-05, + "loss": 2.5043817520141602, + "step": 172720 + }, + { + "epoch": 0.4702327480604328, + "grad_norm": 2.0570638179779053, + "learning_rate": 1.0609840534278318e-05, + "loss": 2.5044931411743163, + "step": 172740 + }, + { + "epoch": 0.47028719205117736, + "grad_norm": 2.280571699142456, + "learning_rate": 1.0608750170369362e-05, + "loss": 2.4691888809204103, + "step": 172760 + }, + { + "epoch": 0.47034163604192186, + "grad_norm": 2.1559767723083496, + "learning_rate": 1.0607659806460406e-05, + "loss": 2.454785919189453, + "step": 172780 + }, + { + "epoch": 0.4703960800326664, + "grad_norm": 2.1513543128967285, + "learning_rate": 1.0606569442551453e-05, + "loss": 2.4937070846557616, + "step": 172800 + }, + { + "epoch": 0.4704505240234109, + "grad_norm": 2.132533311843872, + "learning_rate": 1.0605479078642497e-05, + "loss": 2.482695388793945, + "step": 172820 + }, + { + "epoch": 0.4705049680141554, + "grad_norm": 2.2158102989196777, + "learning_rate": 1.0604388714733542e-05, + "loss": 2.4729005813598635, + "step": 172840 + }, + { + "epoch": 0.4705594120049, + "grad_norm": 2.424760341644287, + "learning_rate": 1.0603298350824589e-05, + "loss": 2.4944509506225585, + "step": 172860 + }, + { + "epoch": 0.4706138559956445, + "grad_norm": 2.160473585128784, + "learning_rate": 1.0602207986915633e-05, + "loss": 2.4645557403564453, + "step": 172880 + }, + { + "epoch": 0.470668299986389, + "grad_norm": 2.142786979675293, + "learning_rate": 1.060111762300668e-05, + "loss": 2.485915756225586, + "step": 172900 + }, + { + "epoch": 0.47072274397713354, + "grad_norm": 2.148340940475464, + "learning_rate": 1.0600027259097725e-05, + "loss": 2.5006946563720702, + "step": 172920 + }, + { + "epoch": 0.47077718796787804, + "grad_norm": 2.153377056121826, + "learning_rate": 1.0598936895188769e-05, + "loss": 2.4617687225341798, + "step": 172940 + }, + { + "epoch": 0.47083163195862254, + "grad_norm": 2.1668124198913574, + "learning_rate": 1.0597846531279816e-05, + "loss": 2.4830738067626954, + "step": 172960 + }, + { + "epoch": 0.4708860759493671, + "grad_norm": 2.0642287731170654, + "learning_rate": 1.059675616737086e-05, + "loss": 2.4681962966918944, + "step": 172980 + }, + { + "epoch": 0.4709405199401116, + "grad_norm": 2.1172401905059814, + "learning_rate": 1.0595665803461905e-05, + "loss": 2.4859407424926756, + "step": 173000 + }, + { + "epoch": 0.47099496393085616, + "grad_norm": 2.234095811843872, + "learning_rate": 1.0594575439552952e-05, + "loss": 2.4630733489990235, + "step": 173020 + }, + { + "epoch": 0.47104940792160066, + "grad_norm": 2.4197514057159424, + "learning_rate": 1.0593485075643996e-05, + "loss": 2.505380630493164, + "step": 173040 + }, + { + "epoch": 0.47110385191234516, + "grad_norm": 2.1154801845550537, + "learning_rate": 1.0592394711735044e-05, + "loss": 2.5084230422973635, + "step": 173060 + }, + { + "epoch": 0.4711582959030897, + "grad_norm": 2.244734764099121, + "learning_rate": 1.0591304347826088e-05, + "loss": 2.476333808898926, + "step": 173080 + }, + { + "epoch": 0.4712127398938342, + "grad_norm": 2.1457087993621826, + "learning_rate": 1.0590213983917132e-05, + "loss": 2.482019805908203, + "step": 173100 + }, + { + "epoch": 0.4712671838845787, + "grad_norm": 2.120206832885742, + "learning_rate": 1.058912362000818e-05, + "loss": 2.4711151123046875, + "step": 173120 + }, + { + "epoch": 0.4713216278753233, + "grad_norm": 3.200869560241699, + "learning_rate": 1.0588033256099224e-05, + "loss": 2.4958208084106444, + "step": 173140 + }, + { + "epoch": 0.4713760718660678, + "grad_norm": 2.1815474033355713, + "learning_rate": 1.0586942892190268e-05, + "loss": 2.477149772644043, + "step": 173160 + }, + { + "epoch": 0.4714305158568123, + "grad_norm": 2.1396660804748535, + "learning_rate": 1.0585852528281315e-05, + "loss": 2.478099060058594, + "step": 173180 + }, + { + "epoch": 0.47148495984755684, + "grad_norm": 3.096222400665283, + "learning_rate": 1.058476216437236e-05, + "loss": 2.4821353912353517, + "step": 173200 + }, + { + "epoch": 0.47153940383830134, + "grad_norm": 2.1656460762023926, + "learning_rate": 1.0583671800463407e-05, + "loss": 2.4767873764038084, + "step": 173220 + }, + { + "epoch": 0.47159384782904584, + "grad_norm": 2.1393394470214844, + "learning_rate": 1.0582581436554451e-05, + "loss": 2.5043235778808595, + "step": 173240 + }, + { + "epoch": 0.4716482918197904, + "grad_norm": 2.145854949951172, + "learning_rate": 1.0581491072645495e-05, + "loss": 2.4803272247314454, + "step": 173260 + }, + { + "epoch": 0.4717027358105349, + "grad_norm": 2.237321138381958, + "learning_rate": 1.0580400708736543e-05, + "loss": 2.465689849853516, + "step": 173280 + }, + { + "epoch": 0.47175717980127946, + "grad_norm": 2.14426589012146, + "learning_rate": 1.0579310344827587e-05, + "loss": 2.4866037368774414, + "step": 173300 + }, + { + "epoch": 0.47181162379202396, + "grad_norm": 2.0603654384613037, + "learning_rate": 1.057821998091863e-05, + "loss": 2.504337120056152, + "step": 173320 + }, + { + "epoch": 0.47186606778276846, + "grad_norm": 2.16756272315979, + "learning_rate": 1.0577129617009678e-05, + "loss": 2.4716796875, + "step": 173340 + }, + { + "epoch": 0.471920511773513, + "grad_norm": 2.126950979232788, + "learning_rate": 1.0576039253100722e-05, + "loss": 2.46597957611084, + "step": 173360 + }, + { + "epoch": 0.4719749557642575, + "grad_norm": 2.0656096935272217, + "learning_rate": 1.057494888919177e-05, + "loss": 2.4986360549926756, + "step": 173380 + }, + { + "epoch": 0.472029399755002, + "grad_norm": 2.1683454513549805, + "learning_rate": 1.0573858525282814e-05, + "loss": 2.532014274597168, + "step": 173400 + }, + { + "epoch": 0.4720838437457466, + "grad_norm": 2.1592187881469727, + "learning_rate": 1.0572768161373858e-05, + "loss": 2.496809387207031, + "step": 173420 + }, + { + "epoch": 0.4721382877364911, + "grad_norm": 2.1381072998046875, + "learning_rate": 1.0571677797464906e-05, + "loss": 2.5039443969726562, + "step": 173440 + }, + { + "epoch": 0.4721927317272356, + "grad_norm": 2.219911813735962, + "learning_rate": 1.057058743355595e-05, + "loss": 2.5021995544433593, + "step": 173460 + }, + { + "epoch": 0.47224717571798014, + "grad_norm": 2.2688260078430176, + "learning_rate": 1.0569497069646994e-05, + "loss": 2.478455352783203, + "step": 173480 + }, + { + "epoch": 0.47230161970872464, + "grad_norm": 2.3535242080688477, + "learning_rate": 1.0568406705738041e-05, + "loss": 2.4933658599853517, + "step": 173500 + }, + { + "epoch": 0.4723560636994692, + "grad_norm": 2.097151279449463, + "learning_rate": 1.0567316341829085e-05, + "loss": 2.4726823806762694, + "step": 173520 + }, + { + "epoch": 0.4724105076902137, + "grad_norm": 2.405939817428589, + "learning_rate": 1.0566225977920133e-05, + "loss": 2.5060806274414062, + "step": 173540 + }, + { + "epoch": 0.4724649516809582, + "grad_norm": 2.2531228065490723, + "learning_rate": 1.0565135614011177e-05, + "loss": 2.528472137451172, + "step": 173560 + }, + { + "epoch": 0.47251939567170276, + "grad_norm": 2.441396951675415, + "learning_rate": 1.0564045250102221e-05, + "loss": 2.501799774169922, + "step": 173580 + }, + { + "epoch": 0.47257383966244726, + "grad_norm": 2.0863802433013916, + "learning_rate": 1.0562954886193269e-05, + "loss": 2.501200294494629, + "step": 173600 + }, + { + "epoch": 0.47262828365319176, + "grad_norm": 2.0982847213745117, + "learning_rate": 1.0561864522284313e-05, + "loss": 2.5035175323486327, + "step": 173620 + }, + { + "epoch": 0.4726827276439363, + "grad_norm": 2.155410051345825, + "learning_rate": 1.0560774158375357e-05, + "loss": 2.494804573059082, + "step": 173640 + }, + { + "epoch": 0.4727371716346808, + "grad_norm": 2.1538851261138916, + "learning_rate": 1.0559683794466404e-05, + "loss": 2.5319122314453124, + "step": 173660 + }, + { + "epoch": 0.4727916156254253, + "grad_norm": 2.1114840507507324, + "learning_rate": 1.0558593430557449e-05, + "loss": 2.5056196212768556, + "step": 173680 + }, + { + "epoch": 0.4728460596161699, + "grad_norm": 2.427060127258301, + "learning_rate": 1.0557503066648496e-05, + "loss": 2.5080240249633787, + "step": 173700 + }, + { + "epoch": 0.4729005036069144, + "grad_norm": 2.27934193611145, + "learning_rate": 1.055641270273954e-05, + "loss": 2.4939525604248045, + "step": 173720 + }, + { + "epoch": 0.4729549475976589, + "grad_norm": 2.144298791885376, + "learning_rate": 1.0555322338830584e-05, + "loss": 2.4806869506835936, + "step": 173740 + }, + { + "epoch": 0.47300939158840344, + "grad_norm": 2.3779499530792236, + "learning_rate": 1.0554231974921632e-05, + "loss": 2.45552978515625, + "step": 173760 + }, + { + "epoch": 0.47306383557914794, + "grad_norm": 2.183056116104126, + "learning_rate": 1.0553141611012676e-05, + "loss": 2.483262062072754, + "step": 173780 + }, + { + "epoch": 0.4731182795698925, + "grad_norm": 2.155791997909546, + "learning_rate": 1.055205124710372e-05, + "loss": 2.486013412475586, + "step": 173800 + }, + { + "epoch": 0.473172723560637, + "grad_norm": 2.133434534072876, + "learning_rate": 1.0550960883194768e-05, + "loss": 2.4697925567626955, + "step": 173820 + }, + { + "epoch": 0.4732271675513815, + "grad_norm": 2.179567813873291, + "learning_rate": 1.0549870519285812e-05, + "loss": 2.481615447998047, + "step": 173840 + }, + { + "epoch": 0.47328161154212606, + "grad_norm": 2.048769474029541, + "learning_rate": 1.054878015537686e-05, + "loss": 2.4678054809570313, + "step": 173860 + }, + { + "epoch": 0.47333605553287056, + "grad_norm": 2.2942962646484375, + "learning_rate": 1.0547689791467903e-05, + "loss": 2.489059066772461, + "step": 173880 + }, + { + "epoch": 0.47339049952361506, + "grad_norm": 2.0777978897094727, + "learning_rate": 1.0546599427558947e-05, + "loss": 2.4487781524658203, + "step": 173900 + }, + { + "epoch": 0.4734449435143596, + "grad_norm": 2.123486280441284, + "learning_rate": 1.0545509063649995e-05, + "loss": 2.497852897644043, + "step": 173920 + }, + { + "epoch": 0.4734993875051041, + "grad_norm": 2.107454538345337, + "learning_rate": 1.0544418699741039e-05, + "loss": 2.5131385803222654, + "step": 173940 + }, + { + "epoch": 0.4735538314958486, + "grad_norm": 2.2010202407836914, + "learning_rate": 1.0543328335832087e-05, + "loss": 2.4860645294189454, + "step": 173960 + }, + { + "epoch": 0.4736082754865932, + "grad_norm": 2.199766159057617, + "learning_rate": 1.054223797192313e-05, + "loss": 2.4404361724853514, + "step": 173980 + }, + { + "epoch": 0.4736627194773377, + "grad_norm": 2.1206417083740234, + "learning_rate": 1.0541147608014175e-05, + "loss": 2.4965707778930666, + "step": 174000 + }, + { + "epoch": 0.47371716346808224, + "grad_norm": 2.1103689670562744, + "learning_rate": 1.0540057244105222e-05, + "loss": 2.4727041244506838, + "step": 174020 + }, + { + "epoch": 0.47377160745882674, + "grad_norm": 2.2497661113739014, + "learning_rate": 1.0538966880196266e-05, + "loss": 2.4642868041992188, + "step": 174040 + }, + { + "epoch": 0.47382605144957124, + "grad_norm": 2.2988390922546387, + "learning_rate": 1.053787651628731e-05, + "loss": 2.510454559326172, + "step": 174060 + }, + { + "epoch": 0.4738804954403158, + "grad_norm": 2.2141129970550537, + "learning_rate": 1.0536786152378358e-05, + "loss": 2.4849712371826174, + "step": 174080 + }, + { + "epoch": 0.4739349394310603, + "grad_norm": 2.139742612838745, + "learning_rate": 1.0535695788469402e-05, + "loss": 2.503170204162598, + "step": 174100 + }, + { + "epoch": 0.4739893834218048, + "grad_norm": 2.1615824699401855, + "learning_rate": 1.053460542456045e-05, + "loss": 2.4784231185913086, + "step": 174120 + }, + { + "epoch": 0.47404382741254936, + "grad_norm": 2.1464433670043945, + "learning_rate": 1.0533515060651494e-05, + "loss": 2.452011299133301, + "step": 174140 + }, + { + "epoch": 0.47409827140329386, + "grad_norm": 2.251615285873413, + "learning_rate": 1.0532424696742538e-05, + "loss": 2.4929922103881834, + "step": 174160 + }, + { + "epoch": 0.47415271539403836, + "grad_norm": 2.0844926834106445, + "learning_rate": 1.0531334332833585e-05, + "loss": 2.497114372253418, + "step": 174180 + }, + { + "epoch": 0.4742071593847829, + "grad_norm": 2.0644161701202393, + "learning_rate": 1.053024396892463e-05, + "loss": 2.477749252319336, + "step": 174200 + }, + { + "epoch": 0.4742616033755274, + "grad_norm": 2.1543173789978027, + "learning_rate": 1.0529153605015674e-05, + "loss": 2.472533416748047, + "step": 174220 + }, + { + "epoch": 0.4743160473662719, + "grad_norm": 2.120818614959717, + "learning_rate": 1.0528063241106721e-05, + "loss": 2.457450103759766, + "step": 174240 + }, + { + "epoch": 0.4743704913570165, + "grad_norm": 2.094975233078003, + "learning_rate": 1.0526972877197765e-05, + "loss": 2.502273941040039, + "step": 174260 + }, + { + "epoch": 0.474424935347761, + "grad_norm": 2.1445696353912354, + "learning_rate": 1.0525882513288813e-05, + "loss": 2.445389747619629, + "step": 174280 + }, + { + "epoch": 0.47447937933850554, + "grad_norm": 2.157083749771118, + "learning_rate": 1.0524792149379857e-05, + "loss": 2.4760229110717775, + "step": 174300 + }, + { + "epoch": 0.47453382332925004, + "grad_norm": 2.247894048690796, + "learning_rate": 1.0523701785470901e-05, + "loss": 2.4925439834594725, + "step": 174320 + }, + { + "epoch": 0.47458826731999454, + "grad_norm": 2.047121286392212, + "learning_rate": 1.0522611421561948e-05, + "loss": 2.475833702087402, + "step": 174340 + }, + { + "epoch": 0.4746427113107391, + "grad_norm": 2.162255048751831, + "learning_rate": 1.0521521057652993e-05, + "loss": 2.522647666931152, + "step": 174360 + }, + { + "epoch": 0.4746971553014836, + "grad_norm": 2.1291310787200928, + "learning_rate": 1.0520430693744037e-05, + "loss": 2.5103775024414063, + "step": 174380 + }, + { + "epoch": 0.4747515992922281, + "grad_norm": 2.5257132053375244, + "learning_rate": 1.0519340329835084e-05, + "loss": 2.493316078186035, + "step": 174400 + }, + { + "epoch": 0.47480604328297266, + "grad_norm": 2.151893138885498, + "learning_rate": 1.0518249965926128e-05, + "loss": 2.480508804321289, + "step": 174420 + }, + { + "epoch": 0.47486048727371716, + "grad_norm": 2.118936538696289, + "learning_rate": 1.0517159602017176e-05, + "loss": 2.4677732467651365, + "step": 174440 + }, + { + "epoch": 0.47491493126446166, + "grad_norm": 2.1233363151550293, + "learning_rate": 1.051606923810822e-05, + "loss": 2.4536764144897463, + "step": 174460 + }, + { + "epoch": 0.4749693752552062, + "grad_norm": 2.1560802459716797, + "learning_rate": 1.0514978874199264e-05, + "loss": 2.515012741088867, + "step": 174480 + }, + { + "epoch": 0.4750238192459507, + "grad_norm": 2.2381081581115723, + "learning_rate": 1.0513888510290312e-05, + "loss": 2.488272857666016, + "step": 174500 + }, + { + "epoch": 0.4750782632366953, + "grad_norm": 2.201648712158203, + "learning_rate": 1.0512798146381356e-05, + "loss": 2.503982925415039, + "step": 174520 + }, + { + "epoch": 0.4751327072274398, + "grad_norm": 2.1231212615966797, + "learning_rate": 1.05117077824724e-05, + "loss": 2.5063474655151365, + "step": 174540 + }, + { + "epoch": 0.4751871512181843, + "grad_norm": 2.2453577518463135, + "learning_rate": 1.0510617418563447e-05, + "loss": 2.483380126953125, + "step": 174560 + }, + { + "epoch": 0.47524159520892884, + "grad_norm": 2.05771803855896, + "learning_rate": 1.0509527054654491e-05, + "loss": 2.487690734863281, + "step": 174580 + }, + { + "epoch": 0.47529603919967334, + "grad_norm": 2.225994825363159, + "learning_rate": 1.0508436690745539e-05, + "loss": 2.4745162963867187, + "step": 174600 + }, + { + "epoch": 0.47535048319041784, + "grad_norm": 2.16176438331604, + "learning_rate": 1.0507346326836583e-05, + "loss": 2.5001909255981447, + "step": 174620 + }, + { + "epoch": 0.4754049271811624, + "grad_norm": 2.1886706352233887, + "learning_rate": 1.0506255962927627e-05, + "loss": 2.502167510986328, + "step": 174640 + }, + { + "epoch": 0.4754593711719069, + "grad_norm": 2.117375135421753, + "learning_rate": 1.0505165599018675e-05, + "loss": 2.482483673095703, + "step": 174660 + }, + { + "epoch": 0.4755138151626514, + "grad_norm": 2.1528284549713135, + "learning_rate": 1.0504075235109719e-05, + "loss": 2.512246513366699, + "step": 174680 + }, + { + "epoch": 0.47556825915339596, + "grad_norm": 2.138468027114868, + "learning_rate": 1.0502984871200763e-05, + "loss": 2.465526008605957, + "step": 174700 + }, + { + "epoch": 0.47562270314414046, + "grad_norm": 2.118025541305542, + "learning_rate": 1.050189450729181e-05, + "loss": 2.5037858963012694, + "step": 174720 + }, + { + "epoch": 0.475677147134885, + "grad_norm": 2.136446475982666, + "learning_rate": 1.0500804143382854e-05, + "loss": 2.4788042068481446, + "step": 174740 + }, + { + "epoch": 0.4757315911256295, + "grad_norm": 2.1017417907714844, + "learning_rate": 1.0499713779473902e-05, + "loss": 2.4624372482299806, + "step": 174760 + }, + { + "epoch": 0.475786035116374, + "grad_norm": 2.145479202270508, + "learning_rate": 1.0498623415564946e-05, + "loss": 2.4845863342285157, + "step": 174780 + }, + { + "epoch": 0.4758404791071186, + "grad_norm": 2.1994969844818115, + "learning_rate": 1.049753305165599e-05, + "loss": 2.4769870758056642, + "step": 174800 + }, + { + "epoch": 0.4758949230978631, + "grad_norm": 2.324394464492798, + "learning_rate": 1.0496442687747038e-05, + "loss": 2.486342430114746, + "step": 174820 + }, + { + "epoch": 0.4759493670886076, + "grad_norm": 2.087714433670044, + "learning_rate": 1.0495352323838082e-05, + "loss": 2.488326072692871, + "step": 174840 + }, + { + "epoch": 0.47600381107935213, + "grad_norm": 2.175909996032715, + "learning_rate": 1.0494261959929126e-05, + "loss": 2.495418357849121, + "step": 174860 + }, + { + "epoch": 0.47605825507009664, + "grad_norm": 2.1518807411193848, + "learning_rate": 1.0493171596020173e-05, + "loss": 2.476731872558594, + "step": 174880 + }, + { + "epoch": 0.47611269906084114, + "grad_norm": 2.2074666023254395, + "learning_rate": 1.0492081232111218e-05, + "loss": 2.4667530059814453, + "step": 174900 + }, + { + "epoch": 0.4761671430515857, + "grad_norm": 2.1077799797058105, + "learning_rate": 1.0490990868202265e-05, + "loss": 2.493870735168457, + "step": 174920 + }, + { + "epoch": 0.4762215870423302, + "grad_norm": 2.211275100708008, + "learning_rate": 1.0489900504293309e-05, + "loss": 2.500574493408203, + "step": 174940 + }, + { + "epoch": 0.4762760310330747, + "grad_norm": 2.0926434993743896, + "learning_rate": 1.0488810140384353e-05, + "loss": 2.472840690612793, + "step": 174960 + }, + { + "epoch": 0.47633047502381926, + "grad_norm": 2.055400848388672, + "learning_rate": 1.04877197764754e-05, + "loss": 2.487257957458496, + "step": 174980 + }, + { + "epoch": 0.47638491901456376, + "grad_norm": 2.1104979515075684, + "learning_rate": 1.0486629412566445e-05, + "loss": 2.492660331726074, + "step": 175000 + }, + { + "epoch": 0.4764393630053083, + "grad_norm": 2.1879830360412598, + "learning_rate": 1.0485539048657489e-05, + "loss": 2.4713979721069337, + "step": 175020 + }, + { + "epoch": 0.4764938069960528, + "grad_norm": 2.0976998805999756, + "learning_rate": 1.0484448684748537e-05, + "loss": 2.4729339599609377, + "step": 175040 + }, + { + "epoch": 0.4765482509867973, + "grad_norm": 2.088291645050049, + "learning_rate": 1.048335832083958e-05, + "loss": 2.478182792663574, + "step": 175060 + }, + { + "epoch": 0.4766026949775419, + "grad_norm": 2.1276094913482666, + "learning_rate": 1.0482267956930628e-05, + "loss": 2.49930419921875, + "step": 175080 + }, + { + "epoch": 0.4766571389682864, + "grad_norm": 2.0410754680633545, + "learning_rate": 1.0481177593021672e-05, + "loss": 2.4756143569946287, + "step": 175100 + }, + { + "epoch": 0.4767115829590309, + "grad_norm": 2.1646454334259033, + "learning_rate": 1.0480087229112716e-05, + "loss": 2.4720407485961915, + "step": 175120 + }, + { + "epoch": 0.47676602694977543, + "grad_norm": 2.1581814289093018, + "learning_rate": 1.0478996865203764e-05, + "loss": 2.4890878677368162, + "step": 175140 + }, + { + "epoch": 0.47682047094051994, + "grad_norm": 2.3435184955596924, + "learning_rate": 1.0477906501294808e-05, + "loss": 2.5250267028808593, + "step": 175160 + }, + { + "epoch": 0.47687491493126444, + "grad_norm": 4.570929050445557, + "learning_rate": 1.0476816137385852e-05, + "loss": 2.5195159912109375, + "step": 175180 + }, + { + "epoch": 0.476929358922009, + "grad_norm": 2.213181495666504, + "learning_rate": 1.04757257734769e-05, + "loss": 2.4997669219970704, + "step": 175200 + }, + { + "epoch": 0.4769838029127535, + "grad_norm": 2.0961408615112305, + "learning_rate": 1.0474635409567944e-05, + "loss": 2.5138954162597655, + "step": 175220 + }, + { + "epoch": 0.47703824690349805, + "grad_norm": 2.1393067836761475, + "learning_rate": 1.0473545045658991e-05, + "loss": 2.469990348815918, + "step": 175240 + }, + { + "epoch": 0.47709269089424255, + "grad_norm": 2.164112091064453, + "learning_rate": 1.0472454681750035e-05, + "loss": 2.473787307739258, + "step": 175260 + }, + { + "epoch": 0.47714713488498706, + "grad_norm": 2.12349271774292, + "learning_rate": 1.047136431784108e-05, + "loss": 2.4750844955444338, + "step": 175280 + }, + { + "epoch": 0.4772015788757316, + "grad_norm": 2.204338550567627, + "learning_rate": 1.0470273953932127e-05, + "loss": 2.471011734008789, + "step": 175300 + }, + { + "epoch": 0.4772560228664761, + "grad_norm": 2.2547008991241455, + "learning_rate": 1.0469183590023171e-05, + "loss": 2.482114791870117, + "step": 175320 + }, + { + "epoch": 0.4773104668572206, + "grad_norm": 2.321666717529297, + "learning_rate": 1.0468093226114215e-05, + "loss": 2.4988632202148438, + "step": 175340 + }, + { + "epoch": 0.4773649108479652, + "grad_norm": 2.1389200687408447, + "learning_rate": 1.0467002862205263e-05, + "loss": 2.50653018951416, + "step": 175360 + }, + { + "epoch": 0.4774193548387097, + "grad_norm": 2.1475937366485596, + "learning_rate": 1.0465912498296307e-05, + "loss": 2.450051116943359, + "step": 175380 + }, + { + "epoch": 0.4774737988294542, + "grad_norm": 2.3916308879852295, + "learning_rate": 1.0464822134387354e-05, + "loss": 2.5012977600097654, + "step": 175400 + }, + { + "epoch": 0.47752824282019873, + "grad_norm": 2.10250186920166, + "learning_rate": 1.0463731770478398e-05, + "loss": 2.492610549926758, + "step": 175420 + }, + { + "epoch": 0.47758268681094324, + "grad_norm": 2.1623682975769043, + "learning_rate": 1.0462641406569443e-05, + "loss": 2.4648258209228517, + "step": 175440 + }, + { + "epoch": 0.47763713080168774, + "grad_norm": 2.1064999103546143, + "learning_rate": 1.046155104266049e-05, + "loss": 2.452323341369629, + "step": 175460 + }, + { + "epoch": 0.4776915747924323, + "grad_norm": 2.1836609840393066, + "learning_rate": 1.0460460678751534e-05, + "loss": 2.486054039001465, + "step": 175480 + }, + { + "epoch": 0.4777460187831768, + "grad_norm": 2.1246187686920166, + "learning_rate": 1.0459370314842578e-05, + "loss": 2.4983739852905273, + "step": 175500 + }, + { + "epoch": 0.47780046277392135, + "grad_norm": 2.2262983322143555, + "learning_rate": 1.0458279950933626e-05, + "loss": 2.4703378677368164, + "step": 175520 + }, + { + "epoch": 0.47785490676466585, + "grad_norm": 2.1338627338409424, + "learning_rate": 1.045718958702467e-05, + "loss": 2.472259521484375, + "step": 175540 + }, + { + "epoch": 0.47790935075541036, + "grad_norm": 2.147048234939575, + "learning_rate": 1.0456099223115717e-05, + "loss": 2.472285842895508, + "step": 175560 + }, + { + "epoch": 0.4779637947461549, + "grad_norm": 2.165844202041626, + "learning_rate": 1.0455008859206761e-05, + "loss": 2.50192928314209, + "step": 175580 + }, + { + "epoch": 0.4780182387368994, + "grad_norm": 2.122818946838379, + "learning_rate": 1.0453918495297806e-05, + "loss": 2.4704826354980467, + "step": 175600 + }, + { + "epoch": 0.4780726827276439, + "grad_norm": 2.0520753860473633, + "learning_rate": 1.0452828131388853e-05, + "loss": 2.4835304260253905, + "step": 175620 + }, + { + "epoch": 0.4781271267183885, + "grad_norm": 2.1750786304473877, + "learning_rate": 1.0451737767479897e-05, + "loss": 2.511781692504883, + "step": 175640 + }, + { + "epoch": 0.478181570709133, + "grad_norm": 2.1559669971466064, + "learning_rate": 1.0450647403570943e-05, + "loss": 2.500575065612793, + "step": 175660 + }, + { + "epoch": 0.4782360146998775, + "grad_norm": 2.222180128097534, + "learning_rate": 1.0449557039661989e-05, + "loss": 2.491831970214844, + "step": 175680 + }, + { + "epoch": 0.47829045869062203, + "grad_norm": 2.1918532848358154, + "learning_rate": 1.0448466675753033e-05, + "loss": 2.4827720642089846, + "step": 175700 + }, + { + "epoch": 0.47834490268136654, + "grad_norm": 2.1760330200195312, + "learning_rate": 1.044737631184408e-05, + "loss": 2.4855607986450194, + "step": 175720 + }, + { + "epoch": 0.4783993466721111, + "grad_norm": 2.15199613571167, + "learning_rate": 1.0446285947935125e-05, + "loss": 2.4553478240966795, + "step": 175740 + }, + { + "epoch": 0.4784537906628556, + "grad_norm": 2.187744617462158, + "learning_rate": 1.0445195584026169e-05, + "loss": 2.456290054321289, + "step": 175760 + }, + { + "epoch": 0.4785082346536001, + "grad_norm": 2.1345832347869873, + "learning_rate": 1.0444105220117216e-05, + "loss": 2.5228303909301757, + "step": 175780 + }, + { + "epoch": 0.47856267864434465, + "grad_norm": 2.1902668476104736, + "learning_rate": 1.044301485620826e-05, + "loss": 2.4967185974121096, + "step": 175800 + }, + { + "epoch": 0.47861712263508915, + "grad_norm": 2.1817121505737305, + "learning_rate": 1.0441924492299306e-05, + "loss": 2.4966218948364256, + "step": 175820 + }, + { + "epoch": 0.47867156662583366, + "grad_norm": 2.1483490467071533, + "learning_rate": 1.0440834128390352e-05, + "loss": 2.473261070251465, + "step": 175840 + }, + { + "epoch": 0.4787260106165782, + "grad_norm": 2.0600929260253906, + "learning_rate": 1.0439743764481396e-05, + "loss": 2.4831180572509766, + "step": 175860 + }, + { + "epoch": 0.4787804546073227, + "grad_norm": 2.1743857860565186, + "learning_rate": 1.0438653400572442e-05, + "loss": 2.5003263473510744, + "step": 175880 + }, + { + "epoch": 0.4788348985980672, + "grad_norm": 2.157170534133911, + "learning_rate": 1.0437563036663488e-05, + "loss": 2.471275520324707, + "step": 175900 + }, + { + "epoch": 0.4788893425888118, + "grad_norm": 2.171712875366211, + "learning_rate": 1.0436472672754532e-05, + "loss": 2.476955604553223, + "step": 175920 + }, + { + "epoch": 0.4789437865795563, + "grad_norm": 2.170663595199585, + "learning_rate": 1.043538230884558e-05, + "loss": 2.4852836608886717, + "step": 175940 + }, + { + "epoch": 0.4789982305703008, + "grad_norm": 2.0792975425720215, + "learning_rate": 1.0434291944936623e-05, + "loss": 2.4998809814453127, + "step": 175960 + }, + { + "epoch": 0.47905267456104533, + "grad_norm": 2.0555386543273926, + "learning_rate": 1.043320158102767e-05, + "loss": 2.4866140365600584, + "step": 175980 + }, + { + "epoch": 0.47910711855178983, + "grad_norm": 2.139273166656494, + "learning_rate": 1.0432111217118715e-05, + "loss": 2.4700740814208983, + "step": 176000 + }, + { + "epoch": 0.4791615625425344, + "grad_norm": 2.223884344100952, + "learning_rate": 1.0431020853209759e-05, + "loss": 2.4894376754760743, + "step": 176020 + }, + { + "epoch": 0.4792160065332789, + "grad_norm": 2.004641532897949, + "learning_rate": 1.0429930489300805e-05, + "loss": 2.486089897155762, + "step": 176040 + }, + { + "epoch": 0.4792704505240234, + "grad_norm": 2.0507354736328125, + "learning_rate": 1.042884012539185e-05, + "loss": 2.4790048599243164, + "step": 176060 + }, + { + "epoch": 0.47932489451476795, + "grad_norm": 2.2020583152770996, + "learning_rate": 1.0427749761482895e-05, + "loss": 2.5012941360473633, + "step": 176080 + }, + { + "epoch": 0.47937933850551245, + "grad_norm": 2.137364149093628, + "learning_rate": 1.0426659397573942e-05, + "loss": 2.487580108642578, + "step": 176100 + }, + { + "epoch": 0.47943378249625695, + "grad_norm": 2.0993075370788574, + "learning_rate": 1.0425569033664986e-05, + "loss": 2.4923580169677733, + "step": 176120 + }, + { + "epoch": 0.4794882264870015, + "grad_norm": 2.193256378173828, + "learning_rate": 1.0424478669756032e-05, + "loss": 2.4635448455810547, + "step": 176140 + }, + { + "epoch": 0.479542670477746, + "grad_norm": 2.145397424697876, + "learning_rate": 1.0423388305847078e-05, + "loss": 2.4962160110473635, + "step": 176160 + }, + { + "epoch": 0.4795971144684905, + "grad_norm": 2.1665000915527344, + "learning_rate": 1.0422297941938122e-05, + "loss": 2.4904548645019533, + "step": 176180 + }, + { + "epoch": 0.4796515584592351, + "grad_norm": 2.318434238433838, + "learning_rate": 1.0421207578029168e-05, + "loss": 2.4829032897949217, + "step": 176200 + }, + { + "epoch": 0.4797060024499796, + "grad_norm": 2.0796778202056885, + "learning_rate": 1.0420117214120214e-05, + "loss": 2.4839962005615233, + "step": 176220 + }, + { + "epoch": 0.47976044644072413, + "grad_norm": 2.1159908771514893, + "learning_rate": 1.0419026850211258e-05, + "loss": 2.4734130859375, + "step": 176240 + }, + { + "epoch": 0.47981489043146863, + "grad_norm": 2.1033968925476074, + "learning_rate": 1.0417936486302305e-05, + "loss": 2.451716423034668, + "step": 176260 + }, + { + "epoch": 0.47986933442221313, + "grad_norm": 2.107614517211914, + "learning_rate": 1.041684612239335e-05, + "loss": 2.5032506942749024, + "step": 176280 + }, + { + "epoch": 0.4799237784129577, + "grad_norm": 5.242280960083008, + "learning_rate": 1.0415755758484395e-05, + "loss": 2.468992805480957, + "step": 176300 + }, + { + "epoch": 0.4799782224037022, + "grad_norm": 2.185234785079956, + "learning_rate": 1.0414665394575441e-05, + "loss": 2.484984588623047, + "step": 176320 + }, + { + "epoch": 0.4800326663944467, + "grad_norm": 2.1127402782440186, + "learning_rate": 1.0413575030666485e-05, + "loss": 2.463356590270996, + "step": 176340 + }, + { + "epoch": 0.48008711038519125, + "grad_norm": 2.542522668838501, + "learning_rate": 1.0412484666757531e-05, + "loss": 2.515228843688965, + "step": 176360 + }, + { + "epoch": 0.48014155437593575, + "grad_norm": 2.150916576385498, + "learning_rate": 1.0411394302848577e-05, + "loss": 2.4919971466064452, + "step": 176380 + }, + { + "epoch": 0.48019599836668025, + "grad_norm": 2.1680221557617188, + "learning_rate": 1.0410303938939621e-05, + "loss": 2.459700584411621, + "step": 176400 + }, + { + "epoch": 0.4802504423574248, + "grad_norm": 2.1676676273345947, + "learning_rate": 1.0409213575030669e-05, + "loss": 2.4851621627807616, + "step": 176420 + }, + { + "epoch": 0.4803048863481693, + "grad_norm": 3.099013090133667, + "learning_rate": 1.0408123211121713e-05, + "loss": 2.4895004272460937, + "step": 176440 + }, + { + "epoch": 0.4803593303389138, + "grad_norm": 2.2181074619293213, + "learning_rate": 1.0407032847212758e-05, + "loss": 2.4960132598876954, + "step": 176460 + }, + { + "epoch": 0.48041377432965837, + "grad_norm": 2.1341280937194824, + "learning_rate": 1.0405942483303804e-05, + "loss": 2.4932811737060545, + "step": 176480 + }, + { + "epoch": 0.4804682183204029, + "grad_norm": 2.1524693965911865, + "learning_rate": 1.0404852119394848e-05, + "loss": 2.4572702407836915, + "step": 176500 + }, + { + "epoch": 0.48052266231114743, + "grad_norm": 2.1345314979553223, + "learning_rate": 1.0403761755485894e-05, + "loss": 2.463230323791504, + "step": 176520 + }, + { + "epoch": 0.48057710630189193, + "grad_norm": 2.208035945892334, + "learning_rate": 1.040267139157694e-05, + "loss": 2.4890859603881834, + "step": 176540 + }, + { + "epoch": 0.48063155029263643, + "grad_norm": 2.2308812141418457, + "learning_rate": 1.0401581027667984e-05, + "loss": 2.5300729751586912, + "step": 176560 + }, + { + "epoch": 0.480685994283381, + "grad_norm": 2.2291133403778076, + "learning_rate": 1.0400490663759032e-05, + "loss": 2.4759552001953127, + "step": 176580 + }, + { + "epoch": 0.4807404382741255, + "grad_norm": 2.091484785079956, + "learning_rate": 1.0399400299850076e-05, + "loss": 2.47415771484375, + "step": 176600 + }, + { + "epoch": 0.48079488226487, + "grad_norm": 2.1260452270507812, + "learning_rate": 1.0398309935941122e-05, + "loss": 2.4752729415893553, + "step": 176620 + }, + { + "epoch": 0.48084932625561455, + "grad_norm": 2.148674249649048, + "learning_rate": 1.0397219572032167e-05, + "loss": 2.524863052368164, + "step": 176640 + }, + { + "epoch": 0.48090377024635905, + "grad_norm": 2.196622848510742, + "learning_rate": 1.0396129208123211e-05, + "loss": 2.4780706405639648, + "step": 176660 + }, + { + "epoch": 0.48095821423710355, + "grad_norm": 2.238312244415283, + "learning_rate": 1.0395038844214257e-05, + "loss": 2.4850568771362305, + "step": 176680 + }, + { + "epoch": 0.4810126582278481, + "grad_norm": 2.2080280780792236, + "learning_rate": 1.0393948480305303e-05, + "loss": 2.49188289642334, + "step": 176700 + }, + { + "epoch": 0.4810671022185926, + "grad_norm": 2.1522507667541504, + "learning_rate": 1.0392858116396347e-05, + "loss": 2.4751955032348634, + "step": 176720 + }, + { + "epoch": 0.48112154620933717, + "grad_norm": 2.1108357906341553, + "learning_rate": 1.0391767752487395e-05, + "loss": 2.5402835845947265, + "step": 176740 + }, + { + "epoch": 0.48117599020008167, + "grad_norm": 2.089698553085327, + "learning_rate": 1.0390677388578439e-05, + "loss": 2.464666557312012, + "step": 176760 + }, + { + "epoch": 0.4812304341908262, + "grad_norm": 2.1599438190460205, + "learning_rate": 1.0389587024669485e-05, + "loss": 2.473633575439453, + "step": 176780 + }, + { + "epoch": 0.48128487818157073, + "grad_norm": 2.125783681869507, + "learning_rate": 1.038849666076053e-05, + "loss": 2.5032960891723635, + "step": 176800 + }, + { + "epoch": 0.48133932217231523, + "grad_norm": 2.182882308959961, + "learning_rate": 1.0387406296851575e-05, + "loss": 2.483532524108887, + "step": 176820 + }, + { + "epoch": 0.48139376616305973, + "grad_norm": 2.165597677230835, + "learning_rate": 1.038631593294262e-05, + "loss": 2.5221765518188475, + "step": 176840 + }, + { + "epoch": 0.4814482101538043, + "grad_norm": 2.1856203079223633, + "learning_rate": 1.0385225569033666e-05, + "loss": 2.489095115661621, + "step": 176860 + }, + { + "epoch": 0.4815026541445488, + "grad_norm": 2.3280279636383057, + "learning_rate": 1.038413520512471e-05, + "loss": 2.472446060180664, + "step": 176880 + }, + { + "epoch": 0.4815570981352933, + "grad_norm": 2.086796760559082, + "learning_rate": 1.0383044841215758e-05, + "loss": 2.473803901672363, + "step": 176900 + }, + { + "epoch": 0.48161154212603785, + "grad_norm": 2.143281936645508, + "learning_rate": 1.0381954477306802e-05, + "loss": 2.50994930267334, + "step": 176920 + }, + { + "epoch": 0.48166598611678235, + "grad_norm": 2.316141128540039, + "learning_rate": 1.0380864113397848e-05, + "loss": 2.439892387390137, + "step": 176940 + }, + { + "epoch": 0.4817204301075269, + "grad_norm": 2.1089529991149902, + "learning_rate": 1.0379773749488894e-05, + "loss": 2.4652246475219726, + "step": 176960 + }, + { + "epoch": 0.4817748740982714, + "grad_norm": 6.314028263092041, + "learning_rate": 1.0378683385579938e-05, + "loss": 2.494102668762207, + "step": 176980 + }, + { + "epoch": 0.4818293180890159, + "grad_norm": 2.164196252822876, + "learning_rate": 1.0377593021670983e-05, + "loss": 2.4682544708251952, + "step": 177000 + }, + { + "epoch": 0.48188376207976047, + "grad_norm": 2.1305060386657715, + "learning_rate": 1.037650265776203e-05, + "loss": 2.4566574096679688, + "step": 177020 + }, + { + "epoch": 0.48193820607050497, + "grad_norm": 2.303065538406372, + "learning_rate": 1.0375412293853073e-05, + "loss": 2.491122817993164, + "step": 177040 + }, + { + "epoch": 0.4819926500612495, + "grad_norm": 2.1641082763671875, + "learning_rate": 1.037432192994412e-05, + "loss": 2.4817075729370117, + "step": 177060 + }, + { + "epoch": 0.48204709405199403, + "grad_norm": 2.447190999984741, + "learning_rate": 1.0373231566035165e-05, + "loss": 2.4894052505493165, + "step": 177080 + }, + { + "epoch": 0.48210153804273853, + "grad_norm": 2.2142910957336426, + "learning_rate": 1.037214120212621e-05, + "loss": 2.4939029693603514, + "step": 177100 + }, + { + "epoch": 0.48215598203348303, + "grad_norm": 2.136747360229492, + "learning_rate": 1.0371050838217257e-05, + "loss": 2.469803047180176, + "step": 177120 + }, + { + "epoch": 0.4822104260242276, + "grad_norm": 2.2186880111694336, + "learning_rate": 1.03699604743083e-05, + "loss": 2.488578033447266, + "step": 177140 + }, + { + "epoch": 0.4822648700149721, + "grad_norm": 2.3365678787231445, + "learning_rate": 1.0368870110399347e-05, + "loss": 2.506956672668457, + "step": 177160 + }, + { + "epoch": 0.4823193140057166, + "grad_norm": 2.129319190979004, + "learning_rate": 1.0367779746490392e-05, + "loss": 2.4820722579956054, + "step": 177180 + }, + { + "epoch": 0.48237375799646115, + "grad_norm": 4.118745803833008, + "learning_rate": 1.0366689382581436e-05, + "loss": 2.4978870391845702, + "step": 177200 + }, + { + "epoch": 0.48242820198720565, + "grad_norm": 2.1787314414978027, + "learning_rate": 1.0365599018672482e-05, + "loss": 2.4807390213012694, + "step": 177220 + }, + { + "epoch": 0.4824826459779502, + "grad_norm": 2.281254291534424, + "learning_rate": 1.0364508654763528e-05, + "loss": 2.5069063186645506, + "step": 177240 + }, + { + "epoch": 0.4825370899686947, + "grad_norm": 2.168109178543091, + "learning_rate": 1.0363418290854574e-05, + "loss": 2.5267629623413086, + "step": 177260 + }, + { + "epoch": 0.4825915339594392, + "grad_norm": 2.192913770675659, + "learning_rate": 1.036232792694562e-05, + "loss": 2.4896751403808595, + "step": 177280 + }, + { + "epoch": 0.48264597795018377, + "grad_norm": 2.1798031330108643, + "learning_rate": 1.0361237563036664e-05, + "loss": 2.479417419433594, + "step": 177300 + }, + { + "epoch": 0.48270042194092827, + "grad_norm": 2.1905179023742676, + "learning_rate": 1.036014719912771e-05, + "loss": 2.495871353149414, + "step": 177320 + }, + { + "epoch": 0.48275486593167277, + "grad_norm": 2.284369945526123, + "learning_rate": 1.0359056835218755e-05, + "loss": 2.501560401916504, + "step": 177340 + }, + { + "epoch": 0.48280930992241733, + "grad_norm": 2.1588988304138184, + "learning_rate": 1.0357966471309801e-05, + "loss": 2.4764741897583007, + "step": 177360 + }, + { + "epoch": 0.48286375391316183, + "grad_norm": 2.2487363815307617, + "learning_rate": 1.0356876107400845e-05, + "loss": 2.456001091003418, + "step": 177380 + }, + { + "epoch": 0.48291819790390633, + "grad_norm": 2.2624197006225586, + "learning_rate": 1.0355785743491891e-05, + "loss": 2.489030647277832, + "step": 177400 + }, + { + "epoch": 0.4829726418946509, + "grad_norm": 2.2502424716949463, + "learning_rate": 1.0354695379582937e-05, + "loss": 2.5113574981689455, + "step": 177420 + }, + { + "epoch": 0.4830270858853954, + "grad_norm": 2.3342325687408447, + "learning_rate": 1.0353605015673983e-05, + "loss": 2.4795862197875977, + "step": 177440 + }, + { + "epoch": 0.48308152987613995, + "grad_norm": 2.1120808124542236, + "learning_rate": 1.0352514651765027e-05, + "loss": 2.4871803283691407, + "step": 177460 + }, + { + "epoch": 0.48313597386688445, + "grad_norm": 2.2059719562530518, + "learning_rate": 1.0351424287856073e-05, + "loss": 2.4730766296386717, + "step": 177480 + }, + { + "epoch": 0.48319041785762895, + "grad_norm": 2.109449863433838, + "learning_rate": 1.0350333923947119e-05, + "loss": 2.509486961364746, + "step": 177500 + }, + { + "epoch": 0.4832448618483735, + "grad_norm": 2.0087528228759766, + "learning_rate": 1.0349243560038164e-05, + "loss": 2.5075376510620115, + "step": 177520 + }, + { + "epoch": 0.483299305839118, + "grad_norm": 2.1650233268737793, + "learning_rate": 1.0348153196129208e-05, + "loss": 2.487061309814453, + "step": 177540 + }, + { + "epoch": 0.4833537498298625, + "grad_norm": 2.24345326423645, + "learning_rate": 1.0347062832220254e-05, + "loss": 2.522890090942383, + "step": 177560 + }, + { + "epoch": 0.48340819382060707, + "grad_norm": 2.2582314014434814, + "learning_rate": 1.03459724683113e-05, + "loss": 2.465150833129883, + "step": 177580 + }, + { + "epoch": 0.48346263781135157, + "grad_norm": 2.261751890182495, + "learning_rate": 1.0344882104402346e-05, + "loss": 2.477768135070801, + "step": 177600 + }, + { + "epoch": 0.48351708180209607, + "grad_norm": 2.1261255741119385, + "learning_rate": 1.034379174049339e-05, + "loss": 2.482954978942871, + "step": 177620 + }, + { + "epoch": 0.48357152579284063, + "grad_norm": 2.109992265701294, + "learning_rate": 1.0342701376584436e-05, + "loss": 2.4739429473876955, + "step": 177640 + }, + { + "epoch": 0.48362596978358513, + "grad_norm": 2.262190580368042, + "learning_rate": 1.0341611012675482e-05, + "loss": 2.4798460006713867, + "step": 177660 + }, + { + "epoch": 0.48368041377432963, + "grad_norm": 2.1078972816467285, + "learning_rate": 1.0340520648766527e-05, + "loss": 2.503721809387207, + "step": 177680 + }, + { + "epoch": 0.4837348577650742, + "grad_norm": 2.246535062789917, + "learning_rate": 1.0339430284857572e-05, + "loss": 2.464083671569824, + "step": 177700 + }, + { + "epoch": 0.4837893017558187, + "grad_norm": 2.1838810443878174, + "learning_rate": 1.0338339920948617e-05, + "loss": 2.4679872512817385, + "step": 177720 + }, + { + "epoch": 0.48384374574656325, + "grad_norm": 2.0749385356903076, + "learning_rate": 1.0337249557039663e-05, + "loss": 2.4868873596191405, + "step": 177740 + }, + { + "epoch": 0.48389818973730775, + "grad_norm": 2.2317495346069336, + "learning_rate": 1.0336159193130709e-05, + "loss": 2.5054243087768553, + "step": 177760 + }, + { + "epoch": 0.48395263372805225, + "grad_norm": 2.0329644680023193, + "learning_rate": 1.0335068829221753e-05, + "loss": 2.5074281692504883, + "step": 177780 + }, + { + "epoch": 0.4840070777187968, + "grad_norm": 2.255889415740967, + "learning_rate": 1.0333978465312799e-05, + "loss": 2.4893442153930665, + "step": 177800 + }, + { + "epoch": 0.4840615217095413, + "grad_norm": 2.1559290885925293, + "learning_rate": 1.0332888101403845e-05, + "loss": 2.460005760192871, + "step": 177820 + }, + { + "epoch": 0.4841159657002858, + "grad_norm": 2.197798728942871, + "learning_rate": 1.033179773749489e-05, + "loss": 2.4878219604492187, + "step": 177840 + }, + { + "epoch": 0.48417040969103037, + "grad_norm": 2.223702907562256, + "learning_rate": 1.0330707373585935e-05, + "loss": 2.494667625427246, + "step": 177860 + }, + { + "epoch": 0.48422485368177487, + "grad_norm": 2.1491007804870605, + "learning_rate": 1.032961700967698e-05, + "loss": 2.4967830657958983, + "step": 177880 + }, + { + "epoch": 0.48427929767251937, + "grad_norm": 2.1109702587127686, + "learning_rate": 1.0328526645768026e-05, + "loss": 2.4270145416259767, + "step": 177900 + }, + { + "epoch": 0.48433374166326393, + "grad_norm": 2.2635231018066406, + "learning_rate": 1.0327436281859072e-05, + "loss": 2.479775619506836, + "step": 177920 + }, + { + "epoch": 0.48438818565400843, + "grad_norm": 2.181370258331299, + "learning_rate": 1.0326345917950116e-05, + "loss": 2.473422050476074, + "step": 177940 + }, + { + "epoch": 0.484442629644753, + "grad_norm": 2.201096296310425, + "learning_rate": 1.0325255554041162e-05, + "loss": 2.491885757446289, + "step": 177960 + }, + { + "epoch": 0.4844970736354975, + "grad_norm": 2.068002700805664, + "learning_rate": 1.0324165190132208e-05, + "loss": 2.502594757080078, + "step": 177980 + }, + { + "epoch": 0.484551517626242, + "grad_norm": 2.2723867893218994, + "learning_rate": 1.0323074826223254e-05, + "loss": 2.482920265197754, + "step": 178000 + }, + { + "epoch": 0.48460596161698655, + "grad_norm": 2.355438232421875, + "learning_rate": 1.0321984462314298e-05, + "loss": 2.5041885375976562, + "step": 178020 + }, + { + "epoch": 0.48466040560773105, + "grad_norm": 2.2063114643096924, + "learning_rate": 1.0320894098405344e-05, + "loss": 2.4575395584106445, + "step": 178040 + }, + { + "epoch": 0.48471484959847555, + "grad_norm": 2.0413410663604736, + "learning_rate": 1.031980373449639e-05, + "loss": 2.4615127563476564, + "step": 178060 + }, + { + "epoch": 0.4847692935892201, + "grad_norm": 2.179103136062622, + "learning_rate": 1.0318713370587433e-05, + "loss": 2.490219306945801, + "step": 178080 + }, + { + "epoch": 0.4848237375799646, + "grad_norm": 2.2566781044006348, + "learning_rate": 1.031762300667848e-05, + "loss": 2.4847192764282227, + "step": 178100 + }, + { + "epoch": 0.4848781815707091, + "grad_norm": 2.2425975799560547, + "learning_rate": 1.0316532642769525e-05, + "loss": 2.4713592529296875, + "step": 178120 + }, + { + "epoch": 0.48493262556145367, + "grad_norm": 2.171160936355591, + "learning_rate": 1.0315442278860571e-05, + "loss": 2.4582088470458983, + "step": 178140 + }, + { + "epoch": 0.48498706955219817, + "grad_norm": 3.4534242153167725, + "learning_rate": 1.0314351914951617e-05, + "loss": 2.472154998779297, + "step": 178160 + }, + { + "epoch": 0.48504151354294267, + "grad_norm": 2.249629497528076, + "learning_rate": 1.031326155104266e-05, + "loss": 2.4657009124755858, + "step": 178180 + }, + { + "epoch": 0.4850959575336872, + "grad_norm": 2.255493402481079, + "learning_rate": 1.0312171187133707e-05, + "loss": 2.4995420455932615, + "step": 178200 + }, + { + "epoch": 0.48515040152443173, + "grad_norm": 2.2097887992858887, + "learning_rate": 1.0311080823224752e-05, + "loss": 2.4750852584838867, + "step": 178220 + }, + { + "epoch": 0.4852048455151763, + "grad_norm": 2.2242259979248047, + "learning_rate": 1.0309990459315797e-05, + "loss": 2.4823720932006834, + "step": 178240 + }, + { + "epoch": 0.4852592895059208, + "grad_norm": 2.1630806922912598, + "learning_rate": 1.0308900095406842e-05, + "loss": 2.4570085525512697, + "step": 178260 + }, + { + "epoch": 0.4853137334966653, + "grad_norm": 2.2562901973724365, + "learning_rate": 1.0307809731497888e-05, + "loss": 2.4747812271118166, + "step": 178280 + }, + { + "epoch": 0.48536817748740985, + "grad_norm": 2.213158130645752, + "learning_rate": 1.0306719367588934e-05, + "loss": 2.497593307495117, + "step": 178300 + }, + { + "epoch": 0.48542262147815435, + "grad_norm": 2.148486852645874, + "learning_rate": 1.030562900367998e-05, + "loss": 2.496062660217285, + "step": 178320 + }, + { + "epoch": 0.48547706546889885, + "grad_norm": 2.1625306606292725, + "learning_rate": 1.0304538639771024e-05, + "loss": 2.5124675750732424, + "step": 178340 + }, + { + "epoch": 0.4855315094596434, + "grad_norm": 2.4201414585113525, + "learning_rate": 1.030344827586207e-05, + "loss": 2.5089700698852537, + "step": 178360 + }, + { + "epoch": 0.4855859534503879, + "grad_norm": 2.1582107543945312, + "learning_rate": 1.0302357911953115e-05, + "loss": 2.499492073059082, + "step": 178380 + }, + { + "epoch": 0.4856403974411324, + "grad_norm": 2.0690321922302246, + "learning_rate": 1.030126754804416e-05, + "loss": 2.479568099975586, + "step": 178400 + }, + { + "epoch": 0.48569484143187697, + "grad_norm": 2.1973369121551514, + "learning_rate": 1.0300177184135205e-05, + "loss": 2.494541549682617, + "step": 178420 + }, + { + "epoch": 0.48574928542262147, + "grad_norm": 2.175670623779297, + "learning_rate": 1.0299086820226251e-05, + "loss": 2.492298889160156, + "step": 178440 + }, + { + "epoch": 0.485803729413366, + "grad_norm": 2.11663556098938, + "learning_rate": 1.0297996456317297e-05, + "loss": 2.5162057876586914, + "step": 178460 + }, + { + "epoch": 0.4858581734041105, + "grad_norm": 2.435704231262207, + "learning_rate": 1.0296906092408343e-05, + "loss": 2.4552711486816405, + "step": 178480 + }, + { + "epoch": 0.48591261739485503, + "grad_norm": 2.161816358566284, + "learning_rate": 1.0295815728499387e-05, + "loss": 2.496244430541992, + "step": 178500 + }, + { + "epoch": 0.4859670613855996, + "grad_norm": 2.1530258655548096, + "learning_rate": 1.0294725364590433e-05, + "loss": 2.50732479095459, + "step": 178520 + }, + { + "epoch": 0.4860215053763441, + "grad_norm": 2.196718692779541, + "learning_rate": 1.0293635000681479e-05, + "loss": 2.46455078125, + "step": 178540 + }, + { + "epoch": 0.4860759493670886, + "grad_norm": 2.110401153564453, + "learning_rate": 1.0292544636772523e-05, + "loss": 2.482403373718262, + "step": 178560 + }, + { + "epoch": 0.48613039335783315, + "grad_norm": 2.178279399871826, + "learning_rate": 1.0291454272863569e-05, + "loss": 2.4520362854003905, + "step": 178580 + }, + { + "epoch": 0.48618483734857765, + "grad_norm": 2.3919894695281982, + "learning_rate": 1.0290363908954614e-05, + "loss": 2.485400390625, + "step": 178600 + }, + { + "epoch": 0.48623928133932215, + "grad_norm": 2.1386561393737793, + "learning_rate": 1.028927354504566e-05, + "loss": 2.4907276153564455, + "step": 178620 + }, + { + "epoch": 0.4862937253300667, + "grad_norm": 2.263439655303955, + "learning_rate": 1.0288183181136706e-05, + "loss": 2.4943803787231444, + "step": 178640 + }, + { + "epoch": 0.4863481693208112, + "grad_norm": 2.1264350414276123, + "learning_rate": 1.028709281722775e-05, + "loss": 2.4747335433959963, + "step": 178660 + }, + { + "epoch": 0.4864026133115557, + "grad_norm": 2.151343584060669, + "learning_rate": 1.0286002453318796e-05, + "loss": 2.474223518371582, + "step": 178680 + }, + { + "epoch": 0.48645705730230027, + "grad_norm": 2.220127820968628, + "learning_rate": 1.0284912089409842e-05, + "loss": 2.478466796875, + "step": 178700 + }, + { + "epoch": 0.48651150129304477, + "grad_norm": 2.0621418952941895, + "learning_rate": 1.0283821725500886e-05, + "loss": 2.499357986450195, + "step": 178720 + }, + { + "epoch": 0.4865659452837893, + "grad_norm": 2.136108636856079, + "learning_rate": 1.0282731361591932e-05, + "loss": 2.484673500061035, + "step": 178740 + }, + { + "epoch": 0.4866203892745338, + "grad_norm": 2.1128478050231934, + "learning_rate": 1.0281640997682977e-05, + "loss": 2.4885839462280273, + "step": 178760 + }, + { + "epoch": 0.48667483326527833, + "grad_norm": 2.17336106300354, + "learning_rate": 1.0280550633774023e-05, + "loss": 2.4741851806640627, + "step": 178780 + }, + { + "epoch": 0.4867292772560229, + "grad_norm": 2.228273630142212, + "learning_rate": 1.0279460269865069e-05, + "loss": 2.4949111938476562, + "step": 178800 + }, + { + "epoch": 0.4867837212467674, + "grad_norm": 3.018731117248535, + "learning_rate": 1.0278369905956113e-05, + "loss": 2.4673894882202148, + "step": 178820 + }, + { + "epoch": 0.4868381652375119, + "grad_norm": 2.1288440227508545, + "learning_rate": 1.0277279542047159e-05, + "loss": 2.4919105529785157, + "step": 178840 + }, + { + "epoch": 0.48689260922825645, + "grad_norm": 2.330493927001953, + "learning_rate": 1.0276189178138205e-05, + "loss": 2.4792428970336915, + "step": 178860 + }, + { + "epoch": 0.48694705321900095, + "grad_norm": 2.0744597911834717, + "learning_rate": 1.0275098814229249e-05, + "loss": 2.4507925033569338, + "step": 178880 + }, + { + "epoch": 0.48700149720974545, + "grad_norm": 2.0936601161956787, + "learning_rate": 1.0274008450320295e-05, + "loss": 2.464984321594238, + "step": 178900 + }, + { + "epoch": 0.48705594120049, + "grad_norm": 2.220377206802368, + "learning_rate": 1.027291808641134e-05, + "loss": 2.4721115112304686, + "step": 178920 + }, + { + "epoch": 0.4871103851912345, + "grad_norm": 2.274746894836426, + "learning_rate": 1.0271827722502386e-05, + "loss": 2.4774524688720705, + "step": 178940 + }, + { + "epoch": 0.48716482918197906, + "grad_norm": 2.3424136638641357, + "learning_rate": 1.0270737358593432e-05, + "loss": 2.4612735748291015, + "step": 178960 + }, + { + "epoch": 0.48721927317272357, + "grad_norm": 2.0636556148529053, + "learning_rate": 1.0269646994684476e-05, + "loss": 2.5176675796508787, + "step": 178980 + }, + { + "epoch": 0.48727371716346807, + "grad_norm": 2.174424648284912, + "learning_rate": 1.0268556630775522e-05, + "loss": 2.513448143005371, + "step": 179000 + }, + { + "epoch": 0.4873281611542126, + "grad_norm": 2.83005690574646, + "learning_rate": 1.0267466266866568e-05, + "loss": 2.4697452545166017, + "step": 179020 + }, + { + "epoch": 0.4873826051449571, + "grad_norm": 2.142590284347534, + "learning_rate": 1.0266375902957612e-05, + "loss": 2.5042816162109376, + "step": 179040 + }, + { + "epoch": 0.4874370491357016, + "grad_norm": 2.217291831970215, + "learning_rate": 1.026528553904866e-05, + "loss": 2.4831077575683596, + "step": 179060 + }, + { + "epoch": 0.4874914931264462, + "grad_norm": 2.1581544876098633, + "learning_rate": 1.0264195175139704e-05, + "loss": 2.4725061416625977, + "step": 179080 + }, + { + "epoch": 0.4875459371171907, + "grad_norm": 2.1486659049987793, + "learning_rate": 1.026310481123075e-05, + "loss": 2.4791946411132812, + "step": 179100 + }, + { + "epoch": 0.4876003811079352, + "grad_norm": 2.1698334217071533, + "learning_rate": 1.0262014447321795e-05, + "loss": 2.472451400756836, + "step": 179120 + }, + { + "epoch": 0.48765482509867975, + "grad_norm": 2.183328151702881, + "learning_rate": 1.026092408341284e-05, + "loss": 2.513555717468262, + "step": 179140 + }, + { + "epoch": 0.48770926908942425, + "grad_norm": 2.178067922592163, + "learning_rate": 1.0259833719503885e-05, + "loss": 2.481931686401367, + "step": 179160 + }, + { + "epoch": 0.4877637130801688, + "grad_norm": 2.1047539710998535, + "learning_rate": 1.0258743355594931e-05, + "loss": 2.48187313079834, + "step": 179180 + }, + { + "epoch": 0.4878181570709133, + "grad_norm": 2.093588352203369, + "learning_rate": 1.0257652991685975e-05, + "loss": 2.45742130279541, + "step": 179200 + }, + { + "epoch": 0.4878726010616578, + "grad_norm": 2.4664244651794434, + "learning_rate": 1.0256562627777023e-05, + "loss": 2.5101417541503905, + "step": 179220 + }, + { + "epoch": 0.48792704505240236, + "grad_norm": 2.3083996772766113, + "learning_rate": 1.0255472263868067e-05, + "loss": 2.4613153457641603, + "step": 179240 + }, + { + "epoch": 0.48798148904314687, + "grad_norm": 2.1621832847595215, + "learning_rate": 1.025438189995911e-05, + "loss": 2.495927429199219, + "step": 179260 + }, + { + "epoch": 0.48803593303389137, + "grad_norm": 2.214519500732422, + "learning_rate": 1.0253291536050158e-05, + "loss": 2.5002166748046877, + "step": 179280 + }, + { + "epoch": 0.4880903770246359, + "grad_norm": 2.315765380859375, + "learning_rate": 1.0252201172141202e-05, + "loss": 2.49404354095459, + "step": 179300 + }, + { + "epoch": 0.4881448210153804, + "grad_norm": 2.2805209159851074, + "learning_rate": 1.0251110808232248e-05, + "loss": 2.480070877075195, + "step": 179320 + }, + { + "epoch": 0.4881992650061249, + "grad_norm": 2.1774094104766846, + "learning_rate": 1.0250020444323294e-05, + "loss": 2.4923213958740233, + "step": 179340 + }, + { + "epoch": 0.4882537089968695, + "grad_norm": 2.143235445022583, + "learning_rate": 1.0248930080414338e-05, + "loss": 2.477685546875, + "step": 179360 + }, + { + "epoch": 0.488308152987614, + "grad_norm": 2.1528992652893066, + "learning_rate": 1.0247839716505386e-05, + "loss": 2.4942346572875977, + "step": 179380 + }, + { + "epoch": 0.4883625969783585, + "grad_norm": 2.32572865486145, + "learning_rate": 1.024674935259643e-05, + "loss": 2.465155029296875, + "step": 179400 + }, + { + "epoch": 0.48841704096910304, + "grad_norm": 2.182948112487793, + "learning_rate": 1.0245658988687474e-05, + "loss": 2.488755226135254, + "step": 179420 + }, + { + "epoch": 0.48847148495984755, + "grad_norm": 2.164632797241211, + "learning_rate": 1.0244568624778521e-05, + "loss": 2.5101268768310545, + "step": 179440 + }, + { + "epoch": 0.4885259289505921, + "grad_norm": 14.57091999053955, + "learning_rate": 1.0243478260869565e-05, + "loss": 2.494211196899414, + "step": 179460 + }, + { + "epoch": 0.4885803729413366, + "grad_norm": 2.1788086891174316, + "learning_rate": 1.0242387896960611e-05, + "loss": 2.480103302001953, + "step": 179480 + }, + { + "epoch": 0.4886348169320811, + "grad_norm": 2.0249879360198975, + "learning_rate": 1.0241297533051657e-05, + "loss": 2.5049047470092773, + "step": 179500 + }, + { + "epoch": 0.48868926092282566, + "grad_norm": 2.3158700466156006, + "learning_rate": 1.0240207169142701e-05, + "loss": 2.516195869445801, + "step": 179520 + }, + { + "epoch": 0.48874370491357016, + "grad_norm": 2.3565800189971924, + "learning_rate": 1.0239116805233749e-05, + "loss": 2.476878356933594, + "step": 179540 + }, + { + "epoch": 0.48879814890431467, + "grad_norm": 2.1628708839416504, + "learning_rate": 1.0238026441324793e-05, + "loss": 2.464474678039551, + "step": 179560 + }, + { + "epoch": 0.4888525928950592, + "grad_norm": 2.129223585128784, + "learning_rate": 1.0236936077415837e-05, + "loss": 2.4620487213134767, + "step": 179580 + }, + { + "epoch": 0.4889070368858037, + "grad_norm": 2.1696345806121826, + "learning_rate": 1.0235845713506884e-05, + "loss": 2.4727943420410154, + "step": 179600 + }, + { + "epoch": 0.4889614808765482, + "grad_norm": 2.2433834075927734, + "learning_rate": 1.0234755349597929e-05, + "loss": 2.4755477905273438, + "step": 179620 + }, + { + "epoch": 0.4890159248672928, + "grad_norm": 2.077772378921509, + "learning_rate": 1.0233664985688974e-05, + "loss": 2.5084693908691404, + "step": 179640 + }, + { + "epoch": 0.4890703688580373, + "grad_norm": 2.1269757747650146, + "learning_rate": 1.023257462178002e-05, + "loss": 2.5116466522216796, + "step": 179660 + }, + { + "epoch": 0.48912481284878184, + "grad_norm": 2.263157367706299, + "learning_rate": 1.0231484257871064e-05, + "loss": 2.480451202392578, + "step": 179680 + }, + { + "epoch": 0.48917925683952634, + "grad_norm": 2.2117321491241455, + "learning_rate": 1.0230393893962112e-05, + "loss": 2.4847124099731444, + "step": 179700 + }, + { + "epoch": 0.48923370083027085, + "grad_norm": 2.245532989501953, + "learning_rate": 1.0229303530053156e-05, + "loss": 2.50109748840332, + "step": 179720 + }, + { + "epoch": 0.4892881448210154, + "grad_norm": 2.137601375579834, + "learning_rate": 1.02282131661442e-05, + "loss": 2.487106132507324, + "step": 179740 + }, + { + "epoch": 0.4893425888117599, + "grad_norm": 2.1184258460998535, + "learning_rate": 1.0227122802235248e-05, + "loss": 2.4908504486083984, + "step": 179760 + }, + { + "epoch": 0.4893970328025044, + "grad_norm": 2.249293327331543, + "learning_rate": 1.0226032438326292e-05, + "loss": 2.481021690368652, + "step": 179780 + }, + { + "epoch": 0.48945147679324896, + "grad_norm": 2.133697748184204, + "learning_rate": 1.0224942074417337e-05, + "loss": 2.4834314346313477, + "step": 179800 + }, + { + "epoch": 0.48950592078399346, + "grad_norm": 2.1517996788024902, + "learning_rate": 1.0223851710508383e-05, + "loss": 2.4727359771728517, + "step": 179820 + }, + { + "epoch": 0.48956036477473797, + "grad_norm": 2.215226411819458, + "learning_rate": 1.0222761346599427e-05, + "loss": 2.478152847290039, + "step": 179840 + }, + { + "epoch": 0.4896148087654825, + "grad_norm": 2.123727798461914, + "learning_rate": 1.0221670982690475e-05, + "loss": 2.5017873764038088, + "step": 179860 + }, + { + "epoch": 0.489669252756227, + "grad_norm": 2.128162145614624, + "learning_rate": 1.0220580618781519e-05, + "loss": 2.4619083404541016, + "step": 179880 + }, + { + "epoch": 0.4897236967469715, + "grad_norm": 2.109302520751953, + "learning_rate": 1.0219490254872563e-05, + "loss": 2.505947303771973, + "step": 179900 + }, + { + "epoch": 0.4897781407377161, + "grad_norm": 2.1204442977905273, + "learning_rate": 1.021839989096361e-05, + "loss": 2.469870758056641, + "step": 179920 + }, + { + "epoch": 0.4898325847284606, + "grad_norm": 2.094578266143799, + "learning_rate": 1.0217309527054655e-05, + "loss": 2.4740819931030273, + "step": 179940 + }, + { + "epoch": 0.48988702871920514, + "grad_norm": 2.142014741897583, + "learning_rate": 1.02162191631457e-05, + "loss": 2.4711076736450197, + "step": 179960 + }, + { + "epoch": 0.48994147270994964, + "grad_norm": 2.3947973251342773, + "learning_rate": 1.0215128799236746e-05, + "loss": 2.48736515045166, + "step": 179980 + }, + { + "epoch": 0.48999591670069415, + "grad_norm": 2.249494791030884, + "learning_rate": 1.021403843532779e-05, + "loss": 2.470488357543945, + "step": 180000 + }, + { + "epoch": 0.4900503606914387, + "grad_norm": 2.2337403297424316, + "learning_rate": 1.0212948071418838e-05, + "loss": 2.480472183227539, + "step": 180020 + }, + { + "epoch": 0.4901048046821832, + "grad_norm": 2.2265515327453613, + "learning_rate": 1.0211857707509882e-05, + "loss": 2.4819547653198244, + "step": 180040 + }, + { + "epoch": 0.4901592486729277, + "grad_norm": 2.277212619781494, + "learning_rate": 1.0210767343600926e-05, + "loss": 2.4869436264038085, + "step": 180060 + }, + { + "epoch": 0.49021369266367226, + "grad_norm": 2.2181715965270996, + "learning_rate": 1.0209676979691974e-05, + "loss": 2.447381782531738, + "step": 180080 + }, + { + "epoch": 0.49026813665441676, + "grad_norm": 2.136974573135376, + "learning_rate": 1.0208586615783018e-05, + "loss": 2.5151172637939454, + "step": 180100 + }, + { + "epoch": 0.49032258064516127, + "grad_norm": 2.1494343280792236, + "learning_rate": 1.0207496251874064e-05, + "loss": 2.4931678771972656, + "step": 180120 + }, + { + "epoch": 0.4903770246359058, + "grad_norm": 2.210019111633301, + "learning_rate": 1.020640588796511e-05, + "loss": 2.4662458419799806, + "step": 180140 + }, + { + "epoch": 0.4904314686266503, + "grad_norm": 2.193908452987671, + "learning_rate": 1.0205315524056154e-05, + "loss": 2.4840089797973635, + "step": 180160 + }, + { + "epoch": 0.4904859126173949, + "grad_norm": 2.300227403640747, + "learning_rate": 1.0204225160147201e-05, + "loss": 2.458680534362793, + "step": 180180 + }, + { + "epoch": 0.4905403566081394, + "grad_norm": 2.2403059005737305, + "learning_rate": 1.0203134796238245e-05, + "loss": 2.4457975387573243, + "step": 180200 + }, + { + "epoch": 0.4905948005988839, + "grad_norm": 2.3445026874542236, + "learning_rate": 1.020204443232929e-05, + "loss": 2.4846954345703125, + "step": 180220 + }, + { + "epoch": 0.49064924458962844, + "grad_norm": 2.15437912940979, + "learning_rate": 1.0200954068420337e-05, + "loss": 2.490540313720703, + "step": 180240 + }, + { + "epoch": 0.49070368858037294, + "grad_norm": 2.205244779586792, + "learning_rate": 1.0199863704511381e-05, + "loss": 2.468221092224121, + "step": 180260 + }, + { + "epoch": 0.49075813257111744, + "grad_norm": 2.1693382263183594, + "learning_rate": 1.0198773340602425e-05, + "loss": 2.48550968170166, + "step": 180280 + }, + { + "epoch": 0.490812576561862, + "grad_norm": 2.1723434925079346, + "learning_rate": 1.0197682976693473e-05, + "loss": 2.492557907104492, + "step": 180300 + }, + { + "epoch": 0.4908670205526065, + "grad_norm": 2.152186155319214, + "learning_rate": 1.0196592612784517e-05, + "loss": 2.4526784896850584, + "step": 180320 + }, + { + "epoch": 0.490921464543351, + "grad_norm": 2.0393524169921875, + "learning_rate": 1.0195502248875564e-05, + "loss": 2.495102119445801, + "step": 180340 + }, + { + "epoch": 0.49097590853409556, + "grad_norm": 2.2292416095733643, + "learning_rate": 1.0194411884966608e-05, + "loss": 2.481821632385254, + "step": 180360 + }, + { + "epoch": 0.49103035252484006, + "grad_norm": 2.410252332687378, + "learning_rate": 1.0193321521057652e-05, + "loss": 2.503162956237793, + "step": 180380 + }, + { + "epoch": 0.49108479651558457, + "grad_norm": 2.119185447692871, + "learning_rate": 1.01922311571487e-05, + "loss": 2.505481719970703, + "step": 180400 + }, + { + "epoch": 0.4911392405063291, + "grad_norm": 2.1466352939605713, + "learning_rate": 1.0191140793239744e-05, + "loss": 2.4537206649780274, + "step": 180420 + }, + { + "epoch": 0.4911936844970736, + "grad_norm": 2.1263504028320312, + "learning_rate": 1.0190050429330788e-05, + "loss": 2.4655927658081054, + "step": 180440 + }, + { + "epoch": 0.4912481284878182, + "grad_norm": 2.006366491317749, + "learning_rate": 1.0188960065421836e-05, + "loss": 2.4819007873535157, + "step": 180460 + }, + { + "epoch": 0.4913025724785627, + "grad_norm": 2.2371890544891357, + "learning_rate": 1.018786970151288e-05, + "loss": 2.4766923904418947, + "step": 180480 + }, + { + "epoch": 0.4913570164693072, + "grad_norm": 2.1145100593566895, + "learning_rate": 1.0186779337603927e-05, + "loss": 2.4585697174072267, + "step": 180500 + }, + { + "epoch": 0.49141146046005174, + "grad_norm": 2.1712918281555176, + "learning_rate": 1.0185688973694971e-05, + "loss": 2.497841644287109, + "step": 180520 + }, + { + "epoch": 0.49146590445079624, + "grad_norm": 2.1120505332946777, + "learning_rate": 1.0184598609786015e-05, + "loss": 2.456732177734375, + "step": 180540 + }, + { + "epoch": 0.49152034844154074, + "grad_norm": 2.108386278152466, + "learning_rate": 1.0183508245877063e-05, + "loss": 2.4971221923828124, + "step": 180560 + }, + { + "epoch": 0.4915747924322853, + "grad_norm": 2.30208683013916, + "learning_rate": 1.0182417881968107e-05, + "loss": 2.4965457916259766, + "step": 180580 + }, + { + "epoch": 0.4916292364230298, + "grad_norm": 2.134965181350708, + "learning_rate": 1.0181327518059151e-05, + "loss": 2.4984222412109376, + "step": 180600 + }, + { + "epoch": 0.4916836804137743, + "grad_norm": 2.377598524093628, + "learning_rate": 1.0180237154150199e-05, + "loss": 2.48946533203125, + "step": 180620 + }, + { + "epoch": 0.49173812440451886, + "grad_norm": 2.255754232406616, + "learning_rate": 1.0179146790241243e-05, + "loss": 2.46295166015625, + "step": 180640 + }, + { + "epoch": 0.49179256839526336, + "grad_norm": 2.093275785446167, + "learning_rate": 1.017805642633229e-05, + "loss": 2.499093246459961, + "step": 180660 + }, + { + "epoch": 0.4918470123860079, + "grad_norm": 2.9686057567596436, + "learning_rate": 1.0176966062423334e-05, + "loss": 2.475145721435547, + "step": 180680 + }, + { + "epoch": 0.4919014563767524, + "grad_norm": 2.079634189605713, + "learning_rate": 1.0175875698514379e-05, + "loss": 2.500334930419922, + "step": 180700 + }, + { + "epoch": 0.4919559003674969, + "grad_norm": 2.157799482345581, + "learning_rate": 1.0174785334605426e-05, + "loss": 2.4887237548828125, + "step": 180720 + }, + { + "epoch": 0.4920103443582415, + "grad_norm": 2.058119773864746, + "learning_rate": 1.017369497069647e-05, + "loss": 2.5015453338623046, + "step": 180740 + }, + { + "epoch": 0.492064788348986, + "grad_norm": 2.047293186187744, + "learning_rate": 1.0172604606787518e-05, + "loss": 2.5001611709594727, + "step": 180760 + }, + { + "epoch": 0.4921192323397305, + "grad_norm": 2.048813581466675, + "learning_rate": 1.0171514242878562e-05, + "loss": 2.4836093902587892, + "step": 180780 + }, + { + "epoch": 0.49217367633047504, + "grad_norm": 2.0689280033111572, + "learning_rate": 1.0170423878969606e-05, + "loss": 2.494029235839844, + "step": 180800 + }, + { + "epoch": 0.49222812032121954, + "grad_norm": 2.137054204940796, + "learning_rate": 1.0169333515060653e-05, + "loss": 2.502662467956543, + "step": 180820 + }, + { + "epoch": 0.49228256431196404, + "grad_norm": 2.279355764389038, + "learning_rate": 1.0168243151151698e-05, + "loss": 2.4852371215820312, + "step": 180840 + }, + { + "epoch": 0.4923370083027086, + "grad_norm": 2.0879712104797363, + "learning_rate": 1.0167152787242742e-05, + "loss": 2.4906423568725584, + "step": 180860 + }, + { + "epoch": 0.4923914522934531, + "grad_norm": 2.0889382362365723, + "learning_rate": 1.0166062423333789e-05, + "loss": 2.477243423461914, + "step": 180880 + }, + { + "epoch": 0.4924458962841976, + "grad_norm": 2.1701509952545166, + "learning_rate": 1.0164972059424833e-05, + "loss": 2.4885114669799804, + "step": 180900 + }, + { + "epoch": 0.49250034027494216, + "grad_norm": 2.1017203330993652, + "learning_rate": 1.016388169551588e-05, + "loss": 2.491880416870117, + "step": 180920 + }, + { + "epoch": 0.49255478426568666, + "grad_norm": 2.0771644115448, + "learning_rate": 1.0162791331606925e-05, + "loss": 2.4940235137939455, + "step": 180940 + }, + { + "epoch": 0.4926092282564312, + "grad_norm": 2.189389705657959, + "learning_rate": 1.0161700967697969e-05, + "loss": 2.4938840866088867, + "step": 180960 + }, + { + "epoch": 0.4926636722471757, + "grad_norm": 19.028722763061523, + "learning_rate": 1.0160610603789016e-05, + "loss": 2.4956775665283204, + "step": 180980 + }, + { + "epoch": 0.4927181162379202, + "grad_norm": 2.1982688903808594, + "learning_rate": 1.015952023988006e-05, + "loss": 2.463019371032715, + "step": 181000 + }, + { + "epoch": 0.4927725602286648, + "grad_norm": 2.1173605918884277, + "learning_rate": 1.0158429875971105e-05, + "loss": 2.4635190963745117, + "step": 181020 + }, + { + "epoch": 0.4928270042194093, + "grad_norm": 2.146181583404541, + "learning_rate": 1.0157339512062152e-05, + "loss": 2.4962202072143556, + "step": 181040 + }, + { + "epoch": 0.4928814482101538, + "grad_norm": 2.204907178878784, + "learning_rate": 1.0156249148153196e-05, + "loss": 2.4662853240966798, + "step": 181060 + }, + { + "epoch": 0.49293589220089834, + "grad_norm": 2.216907501220703, + "learning_rate": 1.0155158784244244e-05, + "loss": 2.4775257110595703, + "step": 181080 + }, + { + "epoch": 0.49299033619164284, + "grad_norm": 2.269582986831665, + "learning_rate": 1.0154068420335288e-05, + "loss": 2.498610496520996, + "step": 181100 + }, + { + "epoch": 0.49304478018238734, + "grad_norm": 2.250253438949585, + "learning_rate": 1.0152978056426332e-05, + "loss": 2.4784656524658204, + "step": 181120 + }, + { + "epoch": 0.4930992241731319, + "grad_norm": 2.4292781352996826, + "learning_rate": 1.015188769251738e-05, + "loss": 2.4588342666625977, + "step": 181140 + }, + { + "epoch": 0.4931536681638764, + "grad_norm": 2.0863754749298096, + "learning_rate": 1.0150797328608424e-05, + "loss": 2.4937156677246093, + "step": 181160 + }, + { + "epoch": 0.49320811215462096, + "grad_norm": 2.271681547164917, + "learning_rate": 1.0149706964699468e-05, + "loss": 2.4709182739257813, + "step": 181180 + }, + { + "epoch": 0.49326255614536546, + "grad_norm": 2.127703905105591, + "learning_rate": 1.0148616600790515e-05, + "loss": 2.463987350463867, + "step": 181200 + }, + { + "epoch": 0.49331700013610996, + "grad_norm": 2.1581079959869385, + "learning_rate": 1.014752623688156e-05, + "loss": 2.4829666137695314, + "step": 181220 + }, + { + "epoch": 0.4933714441268545, + "grad_norm": 2.168447971343994, + "learning_rate": 1.0146435872972607e-05, + "loss": 2.4857198715209963, + "step": 181240 + }, + { + "epoch": 0.493425888117599, + "grad_norm": 2.0639750957489014, + "learning_rate": 1.0145345509063651e-05, + "loss": 2.4893705368041994, + "step": 181260 + }, + { + "epoch": 0.4934803321083435, + "grad_norm": 2.274399757385254, + "learning_rate": 1.0144255145154695e-05, + "loss": 2.4616445541381835, + "step": 181280 + }, + { + "epoch": 0.4935347760990881, + "grad_norm": 2.324864387512207, + "learning_rate": 1.0143164781245743e-05, + "loss": 2.480852508544922, + "step": 181300 + }, + { + "epoch": 0.4935892200898326, + "grad_norm": 2.199209690093994, + "learning_rate": 1.0142074417336787e-05, + "loss": 2.4693187713623046, + "step": 181320 + }, + { + "epoch": 0.4936436640805771, + "grad_norm": 2.136441230773926, + "learning_rate": 1.0140984053427831e-05, + "loss": 2.488554763793945, + "step": 181340 + }, + { + "epoch": 0.49369810807132164, + "grad_norm": 2.245450258255005, + "learning_rate": 1.0139893689518878e-05, + "loss": 2.4883800506591798, + "step": 181360 + }, + { + "epoch": 0.49375255206206614, + "grad_norm": 2.3344688415527344, + "learning_rate": 1.0138803325609922e-05, + "loss": 2.461495018005371, + "step": 181380 + }, + { + "epoch": 0.4938069960528107, + "grad_norm": 2.282567024230957, + "learning_rate": 1.013771296170097e-05, + "loss": 2.5055572509765627, + "step": 181400 + }, + { + "epoch": 0.4938614400435552, + "grad_norm": 2.205970048904419, + "learning_rate": 1.0136622597792014e-05, + "loss": 2.4375679016113283, + "step": 181420 + }, + { + "epoch": 0.4939158840342997, + "grad_norm": 2.1503047943115234, + "learning_rate": 1.0135532233883058e-05, + "loss": 2.474064826965332, + "step": 181440 + }, + { + "epoch": 0.49397032802504426, + "grad_norm": 2.174105405807495, + "learning_rate": 1.0134441869974106e-05, + "loss": 2.4863088607788084, + "step": 181460 + }, + { + "epoch": 0.49402477201578876, + "grad_norm": 2.2087204456329346, + "learning_rate": 1.013335150606515e-05, + "loss": 2.499372863769531, + "step": 181480 + }, + { + "epoch": 0.49407921600653326, + "grad_norm": 2.2148501873016357, + "learning_rate": 1.0132261142156194e-05, + "loss": 2.481688690185547, + "step": 181500 + }, + { + "epoch": 0.4941336599972778, + "grad_norm": 2.1824023723602295, + "learning_rate": 1.0131170778247241e-05, + "loss": 2.4854171752929686, + "step": 181520 + }, + { + "epoch": 0.4941881039880223, + "grad_norm": 2.162287473678589, + "learning_rate": 1.0130080414338286e-05, + "loss": 2.493543815612793, + "step": 181540 + }, + { + "epoch": 0.4942425479787668, + "grad_norm": 2.404994010925293, + "learning_rate": 1.0128990050429333e-05, + "loss": 2.495564079284668, + "step": 181560 + }, + { + "epoch": 0.4942969919695114, + "grad_norm": 2.0781359672546387, + "learning_rate": 1.0127899686520377e-05, + "loss": 2.496782112121582, + "step": 181580 + }, + { + "epoch": 0.4943514359602559, + "grad_norm": 2.1478893756866455, + "learning_rate": 1.0126809322611421e-05, + "loss": 2.5158241271972654, + "step": 181600 + }, + { + "epoch": 0.4944058799510004, + "grad_norm": 2.1489288806915283, + "learning_rate": 1.0125718958702469e-05, + "loss": 2.503513717651367, + "step": 181620 + }, + { + "epoch": 0.49446032394174494, + "grad_norm": 2.4590189456939697, + "learning_rate": 1.0124628594793513e-05, + "loss": 2.4816944122314455, + "step": 181640 + }, + { + "epoch": 0.49451476793248944, + "grad_norm": 2.1490368843078613, + "learning_rate": 1.0123538230884557e-05, + "loss": 2.5049991607666016, + "step": 181660 + }, + { + "epoch": 0.494569211923234, + "grad_norm": 2.210773468017578, + "learning_rate": 1.0122447866975605e-05, + "loss": 2.4864212036132813, + "step": 181680 + }, + { + "epoch": 0.4946236559139785, + "grad_norm": 2.3477280139923096, + "learning_rate": 1.0121357503066649e-05, + "loss": 2.4661323547363283, + "step": 181700 + }, + { + "epoch": 0.494678099904723, + "grad_norm": 2.303499460220337, + "learning_rate": 1.0120267139157696e-05, + "loss": 2.482582855224609, + "step": 181720 + }, + { + "epoch": 0.49473254389546756, + "grad_norm": 2.1686832904815674, + "learning_rate": 1.011917677524874e-05, + "loss": 2.476191520690918, + "step": 181740 + }, + { + "epoch": 0.49478698788621206, + "grad_norm": 2.170776605606079, + "learning_rate": 1.0118086411339784e-05, + "loss": 2.4686546325683594, + "step": 181760 + }, + { + "epoch": 0.49484143187695656, + "grad_norm": 2.144974946975708, + "learning_rate": 1.0116996047430832e-05, + "loss": 2.48433780670166, + "step": 181780 + }, + { + "epoch": 0.4948958758677011, + "grad_norm": 2.2384049892425537, + "learning_rate": 1.0115905683521876e-05, + "loss": 2.476211738586426, + "step": 181800 + }, + { + "epoch": 0.4949503198584456, + "grad_norm": 2.817249059677124, + "learning_rate": 1.011481531961292e-05, + "loss": 2.5038562774658204, + "step": 181820 + }, + { + "epoch": 0.4950047638491901, + "grad_norm": 6.463140964508057, + "learning_rate": 1.0113724955703968e-05, + "loss": 2.510825347900391, + "step": 181840 + }, + { + "epoch": 0.4950592078399347, + "grad_norm": 2.1232640743255615, + "learning_rate": 1.0112634591795012e-05, + "loss": 2.4588716506958006, + "step": 181860 + }, + { + "epoch": 0.4951136518306792, + "grad_norm": 2.1095666885375977, + "learning_rate": 1.011154422788606e-05, + "loss": 2.494172477722168, + "step": 181880 + }, + { + "epoch": 0.49516809582142374, + "grad_norm": 2.147200345993042, + "learning_rate": 1.0110453863977103e-05, + "loss": 2.4825159072875977, + "step": 181900 + }, + { + "epoch": 0.49522253981216824, + "grad_norm": 2.2295494079589844, + "learning_rate": 1.0109363500068147e-05, + "loss": 2.4475753784179686, + "step": 181920 + }, + { + "epoch": 0.49527698380291274, + "grad_norm": 2.180992603302002, + "learning_rate": 1.0108273136159195e-05, + "loss": 2.4650224685668944, + "step": 181940 + }, + { + "epoch": 0.4953314277936573, + "grad_norm": 2.187967538833618, + "learning_rate": 1.0107182772250239e-05, + "loss": 2.489883232116699, + "step": 181960 + }, + { + "epoch": 0.4953858717844018, + "grad_norm": 2.224247932434082, + "learning_rate": 1.0106092408341283e-05, + "loss": 2.4993688583374025, + "step": 181980 + }, + { + "epoch": 0.4954403157751463, + "grad_norm": 2.279508352279663, + "learning_rate": 1.010500204443233e-05, + "loss": 2.4669055938720703, + "step": 182000 + }, + { + "epoch": 0.49549475976589086, + "grad_norm": 2.1033759117126465, + "learning_rate": 1.0103911680523375e-05, + "loss": 2.489415740966797, + "step": 182020 + }, + { + "epoch": 0.49554920375663536, + "grad_norm": 2.0385348796844482, + "learning_rate": 1.0102821316614422e-05, + "loss": 2.4718246459960938, + "step": 182040 + }, + { + "epoch": 0.49560364774737986, + "grad_norm": 2.180462121963501, + "learning_rate": 1.0101730952705466e-05, + "loss": 2.4855175018310547, + "step": 182060 + }, + { + "epoch": 0.4956580917381244, + "grad_norm": 2.3284895420074463, + "learning_rate": 1.010064058879651e-05, + "loss": 2.473341178894043, + "step": 182080 + }, + { + "epoch": 0.4957125357288689, + "grad_norm": 2.266860246658325, + "learning_rate": 1.0099550224887558e-05, + "loss": 2.4734596252441405, + "step": 182100 + }, + { + "epoch": 0.4957669797196134, + "grad_norm": 2.157426595687866, + "learning_rate": 1.0098459860978602e-05, + "loss": 2.4845758438110352, + "step": 182120 + }, + { + "epoch": 0.495821423710358, + "grad_norm": 2.263066291809082, + "learning_rate": 1.0097369497069646e-05, + "loss": 2.458553695678711, + "step": 182140 + }, + { + "epoch": 0.4958758677011025, + "grad_norm": 2.2095813751220703, + "learning_rate": 1.0096279133160694e-05, + "loss": 2.4660884857177736, + "step": 182160 + }, + { + "epoch": 0.49593031169184704, + "grad_norm": 2.120065689086914, + "learning_rate": 1.0095188769251738e-05, + "loss": 2.4817066192626953, + "step": 182180 + }, + { + "epoch": 0.49598475568259154, + "grad_norm": 2.207923412322998, + "learning_rate": 1.0094098405342785e-05, + "loss": 2.530994987487793, + "step": 182200 + }, + { + "epoch": 0.49603919967333604, + "grad_norm": 2.5186100006103516, + "learning_rate": 1.009300804143383e-05, + "loss": 2.481151580810547, + "step": 182220 + }, + { + "epoch": 0.4960936436640806, + "grad_norm": 2.1865336894989014, + "learning_rate": 1.0091917677524874e-05, + "loss": 2.491623306274414, + "step": 182240 + }, + { + "epoch": 0.4961480876548251, + "grad_norm": 2.1776821613311768, + "learning_rate": 1.0090827313615921e-05, + "loss": 2.4765914916992187, + "step": 182260 + }, + { + "epoch": 0.4962025316455696, + "grad_norm": 2.1100358963012695, + "learning_rate": 1.0089736949706965e-05, + "loss": 2.490127944946289, + "step": 182280 + }, + { + "epoch": 0.49625697563631416, + "grad_norm": 2.1115753650665283, + "learning_rate": 1.008864658579801e-05, + "loss": 2.4716630935668946, + "step": 182300 + }, + { + "epoch": 0.49631141962705866, + "grad_norm": 2.166918992996216, + "learning_rate": 1.0087556221889057e-05, + "loss": 2.5001657485961912, + "step": 182320 + }, + { + "epoch": 0.49636586361780316, + "grad_norm": 2.182896614074707, + "learning_rate": 1.0086465857980101e-05, + "loss": 2.4991865158081055, + "step": 182340 + }, + { + "epoch": 0.4964203076085477, + "grad_norm": 2.0742576122283936, + "learning_rate": 1.0085375494071149e-05, + "loss": 2.4761770248413084, + "step": 182360 + }, + { + "epoch": 0.4964747515992922, + "grad_norm": 2.1441879272460938, + "learning_rate": 1.0084285130162193e-05, + "loss": 2.499607276916504, + "step": 182380 + }, + { + "epoch": 0.4965291955900368, + "grad_norm": 2.1217877864837646, + "learning_rate": 1.0083194766253237e-05, + "loss": 2.4549026489257812, + "step": 182400 + }, + { + "epoch": 0.4965836395807813, + "grad_norm": 2.259999990463257, + "learning_rate": 1.0082104402344284e-05, + "loss": 2.4785221099853514, + "step": 182420 + }, + { + "epoch": 0.4966380835715258, + "grad_norm": 2.42993426322937, + "learning_rate": 1.0081014038435328e-05, + "loss": 2.4568634033203125, + "step": 182440 + }, + { + "epoch": 0.49669252756227034, + "grad_norm": 2.2224667072296143, + "learning_rate": 1.0079923674526376e-05, + "loss": 2.4630393981933594, + "step": 182460 + }, + { + "epoch": 0.49674697155301484, + "grad_norm": 2.2012739181518555, + "learning_rate": 1.007883331061742e-05, + "loss": 2.5010923385620116, + "step": 182480 + }, + { + "epoch": 0.49680141554375934, + "grad_norm": 2.1230521202087402, + "learning_rate": 1.0077742946708464e-05, + "loss": 2.5043006896972657, + "step": 182500 + }, + { + "epoch": 0.4968558595345039, + "grad_norm": 2.5949816703796387, + "learning_rate": 1.0076652582799512e-05, + "loss": 2.4860485076904295, + "step": 182520 + }, + { + "epoch": 0.4969103035252484, + "grad_norm": 2.1631410121917725, + "learning_rate": 1.0075562218890556e-05, + "loss": 2.500771713256836, + "step": 182540 + }, + { + "epoch": 0.4969647475159929, + "grad_norm": 2.0792973041534424, + "learning_rate": 1.00744718549816e-05, + "loss": 2.4587202072143555, + "step": 182560 + }, + { + "epoch": 0.49701919150673746, + "grad_norm": 2.1118884086608887, + "learning_rate": 1.0073381491072647e-05, + "loss": 2.5114368438720702, + "step": 182580 + }, + { + "epoch": 0.49707363549748196, + "grad_norm": 2.2370636463165283, + "learning_rate": 1.0072291127163691e-05, + "loss": 2.4931207656860352, + "step": 182600 + }, + { + "epoch": 0.49712807948822646, + "grad_norm": 2.108736753463745, + "learning_rate": 1.0071200763254739e-05, + "loss": 2.491742706298828, + "step": 182620 + }, + { + "epoch": 0.497182523478971, + "grad_norm": 2.295773506164551, + "learning_rate": 1.0070110399345783e-05, + "loss": 2.4816919326782227, + "step": 182640 + }, + { + "epoch": 0.4972369674697155, + "grad_norm": 2.16381573677063, + "learning_rate": 1.0069020035436827e-05, + "loss": 2.4887508392333983, + "step": 182660 + }, + { + "epoch": 0.4972914114604601, + "grad_norm": 2.776590347290039, + "learning_rate": 1.0067929671527875e-05, + "loss": 2.463978576660156, + "step": 182680 + }, + { + "epoch": 0.4973458554512046, + "grad_norm": 2.2105038166046143, + "learning_rate": 1.0066839307618919e-05, + "loss": 2.507169723510742, + "step": 182700 + }, + { + "epoch": 0.4974002994419491, + "grad_norm": 2.096139907836914, + "learning_rate": 1.0065748943709963e-05, + "loss": 2.482242965698242, + "step": 182720 + }, + { + "epoch": 0.49745474343269364, + "grad_norm": 2.0773000717163086, + "learning_rate": 1.006465857980101e-05, + "loss": 2.5090429306030275, + "step": 182740 + }, + { + "epoch": 0.49750918742343814, + "grad_norm": 2.0292398929595947, + "learning_rate": 1.0063568215892055e-05, + "loss": 2.4606960296630858, + "step": 182760 + }, + { + "epoch": 0.49756363141418264, + "grad_norm": 2.138044595718384, + "learning_rate": 1.0062477851983102e-05, + "loss": 2.4624866485595702, + "step": 182780 + }, + { + "epoch": 0.4976180754049272, + "grad_norm": 2.3084774017333984, + "learning_rate": 1.0061387488074146e-05, + "loss": 2.514186477661133, + "step": 182800 + }, + { + "epoch": 0.4976725193956717, + "grad_norm": 2.180131435394287, + "learning_rate": 1.006029712416519e-05, + "loss": 2.4747886657714844, + "step": 182820 + }, + { + "epoch": 0.4977269633864162, + "grad_norm": 2.123163938522339, + "learning_rate": 1.0059206760256238e-05, + "loss": 2.4835302352905275, + "step": 182840 + }, + { + "epoch": 0.49778140737716076, + "grad_norm": 2.290443181991577, + "learning_rate": 1.0058116396347282e-05, + "loss": 2.494742584228516, + "step": 182860 + }, + { + "epoch": 0.49783585136790526, + "grad_norm": 2.267777681350708, + "learning_rate": 1.0057026032438326e-05, + "loss": 2.497635841369629, + "step": 182880 + }, + { + "epoch": 0.4978902953586498, + "grad_norm": 2.236647367477417, + "learning_rate": 1.0055935668529374e-05, + "loss": 2.464802551269531, + "step": 182900 + }, + { + "epoch": 0.4979447393493943, + "grad_norm": 2.3198156356811523, + "learning_rate": 1.0054845304620418e-05, + "loss": 2.4802667617797853, + "step": 182920 + }, + { + "epoch": 0.4979991833401388, + "grad_norm": 2.174229621887207, + "learning_rate": 1.0053754940711465e-05, + "loss": 2.497387504577637, + "step": 182940 + }, + { + "epoch": 0.4980536273308834, + "grad_norm": 2.2879552841186523, + "learning_rate": 1.005266457680251e-05, + "loss": 2.490254211425781, + "step": 182960 + }, + { + "epoch": 0.4981080713216279, + "grad_norm": 2.080174684524536, + "learning_rate": 1.0051574212893553e-05, + "loss": 2.4958803176879885, + "step": 182980 + }, + { + "epoch": 0.4981625153123724, + "grad_norm": 2.186615467071533, + "learning_rate": 1.0050483848984601e-05, + "loss": 2.503276252746582, + "step": 183000 + }, + { + "epoch": 0.49821695930311694, + "grad_norm": 2.1412923336029053, + "learning_rate": 1.0049393485075645e-05, + "loss": 2.471845817565918, + "step": 183020 + }, + { + "epoch": 0.49827140329386144, + "grad_norm": 2.163706064224243, + "learning_rate": 1.0048303121166689e-05, + "loss": 2.468374252319336, + "step": 183040 + }, + { + "epoch": 0.49832584728460594, + "grad_norm": 2.164839506149292, + "learning_rate": 1.0047212757257737e-05, + "loss": 2.4989591598510743, + "step": 183060 + }, + { + "epoch": 0.4983802912753505, + "grad_norm": 2.171161651611328, + "learning_rate": 1.004612239334878e-05, + "loss": 2.4776081085205077, + "step": 183080 + }, + { + "epoch": 0.498434735266095, + "grad_norm": 2.3374571800231934, + "learning_rate": 1.0045032029439827e-05, + "loss": 2.4630741119384765, + "step": 183100 + }, + { + "epoch": 0.4984891792568395, + "grad_norm": 2.1896421909332275, + "learning_rate": 1.0043941665530872e-05, + "loss": 2.491933250427246, + "step": 183120 + }, + { + "epoch": 0.49854362324758406, + "grad_norm": 2.0926876068115234, + "learning_rate": 1.0042851301621916e-05, + "loss": 2.447175216674805, + "step": 183140 + }, + { + "epoch": 0.49859806723832856, + "grad_norm": 2.1640405654907227, + "learning_rate": 1.0041760937712964e-05, + "loss": 2.4867666244506834, + "step": 183160 + }, + { + "epoch": 0.4986525112290731, + "grad_norm": 2.157578229904175, + "learning_rate": 1.0040670573804008e-05, + "loss": 2.5210041046142577, + "step": 183180 + }, + { + "epoch": 0.4987069552198176, + "grad_norm": 4.217861175537109, + "learning_rate": 1.0039580209895052e-05, + "loss": 2.468572807312012, + "step": 183200 + }, + { + "epoch": 0.4987613992105621, + "grad_norm": 2.227419853210449, + "learning_rate": 1.00384898459861e-05, + "loss": 2.476663017272949, + "step": 183220 + }, + { + "epoch": 0.4988158432013067, + "grad_norm": 2.1362905502319336, + "learning_rate": 1.0037399482077144e-05, + "loss": 2.5036958694458007, + "step": 183240 + }, + { + "epoch": 0.4988702871920512, + "grad_norm": 2.146789073944092, + "learning_rate": 1.003630911816819e-05, + "loss": 2.4601112365722657, + "step": 183260 + }, + { + "epoch": 0.4989247311827957, + "grad_norm": 2.095839738845825, + "learning_rate": 1.0035218754259235e-05, + "loss": 2.4681652069091795, + "step": 183280 + }, + { + "epoch": 0.49897917517354023, + "grad_norm": 2.4008891582489014, + "learning_rate": 1.003412839035028e-05, + "loss": 2.462377166748047, + "step": 183300 + }, + { + "epoch": 0.49903361916428474, + "grad_norm": 2.2521214485168457, + "learning_rate": 1.0033038026441327e-05, + "loss": 2.4588296890258787, + "step": 183320 + }, + { + "epoch": 0.49908806315502924, + "grad_norm": 2.202227830886841, + "learning_rate": 1.0031947662532371e-05, + "loss": 2.472798156738281, + "step": 183340 + }, + { + "epoch": 0.4991425071457738, + "grad_norm": 2.1580562591552734, + "learning_rate": 1.0030857298623415e-05, + "loss": 2.5030582427978514, + "step": 183360 + }, + { + "epoch": 0.4991969511365183, + "grad_norm": 2.1759390830993652, + "learning_rate": 1.0029766934714463e-05, + "loss": 2.5029323577880858, + "step": 183380 + }, + { + "epoch": 0.49925139512726285, + "grad_norm": 2.080820322036743, + "learning_rate": 1.0028676570805507e-05, + "loss": 2.5082401275634765, + "step": 183400 + }, + { + "epoch": 0.49930583911800736, + "grad_norm": 2.161487579345703, + "learning_rate": 1.0027586206896553e-05, + "loss": 2.4817657470703125, + "step": 183420 + }, + { + "epoch": 0.49936028310875186, + "grad_norm": 2.198869228363037, + "learning_rate": 1.0026495842987598e-05, + "loss": 2.476414680480957, + "step": 183440 + }, + { + "epoch": 0.4994147270994964, + "grad_norm": 2.106473922729492, + "learning_rate": 1.0025405479078643e-05, + "loss": 2.463934326171875, + "step": 183460 + }, + { + "epoch": 0.4994691710902409, + "grad_norm": 2.5156307220458984, + "learning_rate": 1.002431511516969e-05, + "loss": 2.4885467529296874, + "step": 183480 + }, + { + "epoch": 0.4995236150809854, + "grad_norm": 2.113022565841675, + "learning_rate": 1.0023224751260734e-05, + "loss": 2.5233480453491213, + "step": 183500 + }, + { + "epoch": 0.49957805907173, + "grad_norm": 2.6014297008514404, + "learning_rate": 1.0022134387351778e-05, + "loss": 2.4848796844482424, + "step": 183520 + }, + { + "epoch": 0.4996325030624745, + "grad_norm": 2.2259206771850586, + "learning_rate": 1.0021044023442826e-05, + "loss": 2.491571235656738, + "step": 183540 + }, + { + "epoch": 0.499686947053219, + "grad_norm": 2.1986052989959717, + "learning_rate": 1.001995365953387e-05, + "loss": 2.485920524597168, + "step": 183560 + }, + { + "epoch": 0.49974139104396353, + "grad_norm": 2.1449201107025146, + "learning_rate": 1.0018863295624916e-05, + "loss": 2.4680273056030275, + "step": 183580 + }, + { + "epoch": 0.49979583503470804, + "grad_norm": 2.186558246612549, + "learning_rate": 1.0017772931715962e-05, + "loss": 2.4784481048583986, + "step": 183600 + }, + { + "epoch": 0.4998502790254526, + "grad_norm": 2.27898907661438, + "learning_rate": 1.0016682567807006e-05, + "loss": 2.5169084548950194, + "step": 183620 + }, + { + "epoch": 0.4999047230161971, + "grad_norm": 2.0332775115966797, + "learning_rate": 1.0015592203898053e-05, + "loss": 2.4407943725585937, + "step": 183640 + }, + { + "epoch": 0.4999591670069416, + "grad_norm": 2.088862180709839, + "learning_rate": 1.0014501839989097e-05, + "loss": 2.4753528594970704, + "step": 183660 + }, + { + "epoch": 0.5000136109976862, + "grad_norm": 2.199650764465332, + "learning_rate": 1.0013411476080141e-05, + "loss": 2.4899246215820314, + "step": 183680 + }, + { + "epoch": 0.5000680549884307, + "grad_norm": 2.320793628692627, + "learning_rate": 1.0012321112171189e-05, + "loss": 2.4965984344482424, + "step": 183700 + }, + { + "epoch": 0.5001224989791752, + "grad_norm": 2.1942331790924072, + "learning_rate": 1.0011230748262233e-05, + "loss": 2.458812713623047, + "step": 183720 + }, + { + "epoch": 0.5001769429699197, + "grad_norm": 2.74253511428833, + "learning_rate": 1.0010140384353279e-05, + "loss": 2.470114898681641, + "step": 183740 + }, + { + "epoch": 0.5002313869606643, + "grad_norm": 2.0966994762420654, + "learning_rate": 1.0009050020444325e-05, + "loss": 2.4624794006347654, + "step": 183760 + }, + { + "epoch": 0.5002858309514088, + "grad_norm": 2.1472926139831543, + "learning_rate": 1.0007959656535369e-05, + "loss": 2.4990203857421873, + "step": 183780 + }, + { + "epoch": 0.5003402749421533, + "grad_norm": 2.064840078353882, + "learning_rate": 1.0006869292626416e-05, + "loss": 2.5157089233398438, + "step": 183800 + }, + { + "epoch": 0.5003947189328978, + "grad_norm": 2.1077961921691895, + "learning_rate": 1.000577892871746e-05, + "loss": 2.4858510971069334, + "step": 183820 + }, + { + "epoch": 0.5004491629236423, + "grad_norm": 2.1818623542785645, + "learning_rate": 1.0004688564808505e-05, + "loss": 2.4624460220336912, + "step": 183840 + }, + { + "epoch": 0.5005036069143868, + "grad_norm": 2.1804914474487305, + "learning_rate": 1.0003598200899552e-05, + "loss": 2.4740148544311524, + "step": 183860 + }, + { + "epoch": 0.5005580509051314, + "grad_norm": 2.7867581844329834, + "learning_rate": 1.0002507836990596e-05, + "loss": 2.5210723876953125, + "step": 183880 + }, + { + "epoch": 0.5006124948958759, + "grad_norm": 3.1705455780029297, + "learning_rate": 1.0001417473081642e-05, + "loss": 2.4590217590332033, + "step": 183900 + }, + { + "epoch": 0.5006669388866204, + "grad_norm": 2.185575008392334, + "learning_rate": 1.0000327109172688e-05, + "loss": 2.5095073699951174, + "step": 183920 + }, + { + "epoch": 0.5007213828773649, + "grad_norm": 2.2027034759521484, + "learning_rate": 9.999236745263734e-06, + "loss": 2.494873809814453, + "step": 183940 + }, + { + "epoch": 0.5007758268681094, + "grad_norm": 2.0619654655456543, + "learning_rate": 9.998146381354778e-06, + "loss": 2.5001739501953124, + "step": 183960 + }, + { + "epoch": 0.5008302708588539, + "grad_norm": 2.158191442489624, + "learning_rate": 9.997056017445823e-06, + "loss": 2.4570972442626955, + "step": 183980 + }, + { + "epoch": 0.5008847148495985, + "grad_norm": 2.199748992919922, + "learning_rate": 9.99596565353687e-06, + "loss": 2.4627700805664063, + "step": 184000 + }, + { + "epoch": 0.500939158840343, + "grad_norm": 2.0673022270202637, + "learning_rate": 9.994875289627915e-06, + "loss": 2.5013736724853515, + "step": 184020 + }, + { + "epoch": 0.5009936028310875, + "grad_norm": 2.239067792892456, + "learning_rate": 9.99378492571896e-06, + "loss": 2.498378562927246, + "step": 184040 + }, + { + "epoch": 0.501048046821832, + "grad_norm": 2.1815154552459717, + "learning_rate": 9.992694561810005e-06, + "loss": 2.4717727661132813, + "step": 184060 + }, + { + "epoch": 0.5011024908125765, + "grad_norm": 2.181187152862549, + "learning_rate": 9.99160419790105e-06, + "loss": 2.454503631591797, + "step": 184080 + }, + { + "epoch": 0.5011569348033211, + "grad_norm": 2.120248556137085, + "learning_rate": 9.990513833992097e-06, + "loss": 2.4814939498901367, + "step": 184100 + }, + { + "epoch": 0.5012113787940656, + "grad_norm": 2.149763345718384, + "learning_rate": 9.989423470083142e-06, + "loss": 2.4556453704833983, + "step": 184120 + }, + { + "epoch": 0.5012658227848101, + "grad_norm": 2.1431238651275635, + "learning_rate": 9.988333106174187e-06, + "loss": 2.4875802993774414, + "step": 184140 + }, + { + "epoch": 0.5013202667755546, + "grad_norm": 2.2396903038024902, + "learning_rate": 9.987242742265232e-06, + "loss": 2.5052978515625, + "step": 184160 + }, + { + "epoch": 0.5013747107662991, + "grad_norm": 2.0890111923217773, + "learning_rate": 9.986152378356278e-06, + "loss": 2.463174247741699, + "step": 184180 + }, + { + "epoch": 0.5014291547570436, + "grad_norm": 2.1379997730255127, + "learning_rate": 9.985062014447322e-06, + "loss": 2.4894439697265627, + "step": 184200 + }, + { + "epoch": 0.5014835987477883, + "grad_norm": 2.100818634033203, + "learning_rate": 9.983971650538368e-06, + "loss": 2.494979476928711, + "step": 184220 + }, + { + "epoch": 0.5015380427385328, + "grad_norm": 2.1280033588409424, + "learning_rate": 9.982881286629414e-06, + "loss": 2.4668577194213865, + "step": 184240 + }, + { + "epoch": 0.5015924867292773, + "grad_norm": 2.07491135597229, + "learning_rate": 9.98179092272046e-06, + "loss": 2.492757797241211, + "step": 184260 + }, + { + "epoch": 0.5016469307200218, + "grad_norm": 2.109374523162842, + "learning_rate": 9.980700558811504e-06, + "loss": 2.46755428314209, + "step": 184280 + }, + { + "epoch": 0.5017013747107663, + "grad_norm": 2.060025453567505, + "learning_rate": 9.97961019490255e-06, + "loss": 2.462630844116211, + "step": 184300 + }, + { + "epoch": 0.5017558187015109, + "grad_norm": 2.1155614852905273, + "learning_rate": 9.978519830993595e-06, + "loss": 2.4859331130981444, + "step": 184320 + }, + { + "epoch": 0.5018102626922554, + "grad_norm": 2.0537240505218506, + "learning_rate": 9.977429467084641e-06, + "loss": 2.5076871871948243, + "step": 184340 + }, + { + "epoch": 0.5018647066829999, + "grad_norm": 2.138519525527954, + "learning_rate": 9.976339103175685e-06, + "loss": 2.486651802062988, + "step": 184360 + }, + { + "epoch": 0.5019191506737444, + "grad_norm": 2.130229949951172, + "learning_rate": 9.975248739266731e-06, + "loss": 2.501448059082031, + "step": 184380 + }, + { + "epoch": 0.5019735946644889, + "grad_norm": 2.197927474975586, + "learning_rate": 9.974158375357777e-06, + "loss": 2.4761589050292967, + "step": 184400 + }, + { + "epoch": 0.5020280386552334, + "grad_norm": 2.2375409603118896, + "learning_rate": 9.973068011448823e-06, + "loss": 2.4591922760009766, + "step": 184420 + }, + { + "epoch": 0.502082482645978, + "grad_norm": 2.157869815826416, + "learning_rate": 9.971977647539867e-06, + "loss": 2.4862857818603517, + "step": 184440 + }, + { + "epoch": 0.5021369266367225, + "grad_norm": 2.491943836212158, + "learning_rate": 9.970887283630913e-06, + "loss": 2.48348331451416, + "step": 184460 + }, + { + "epoch": 0.502191370627467, + "grad_norm": 2.158142566680908, + "learning_rate": 9.969796919721959e-06, + "loss": 2.4868106842041016, + "step": 184480 + }, + { + "epoch": 0.5022458146182115, + "grad_norm": 2.118462324142456, + "learning_rate": 9.968706555813004e-06, + "loss": 2.490556526184082, + "step": 184500 + }, + { + "epoch": 0.502300258608956, + "grad_norm": 2.1358542442321777, + "learning_rate": 9.967616191904048e-06, + "loss": 2.5037654876708983, + "step": 184520 + }, + { + "epoch": 0.5023547025997006, + "grad_norm": 2.1840946674346924, + "learning_rate": 9.966525827995094e-06, + "loss": 2.4677019119262695, + "step": 184540 + }, + { + "epoch": 0.5024091465904451, + "grad_norm": 2.0929477214813232, + "learning_rate": 9.96543546408614e-06, + "loss": 2.4912202835083006, + "step": 184560 + }, + { + "epoch": 0.5024635905811896, + "grad_norm": 2.143479347229004, + "learning_rate": 9.964345100177186e-06, + "loss": 2.4669517517089843, + "step": 184580 + }, + { + "epoch": 0.5025180345719341, + "grad_norm": 2.1211166381835938, + "learning_rate": 9.96325473626823e-06, + "loss": 2.485671615600586, + "step": 184600 + }, + { + "epoch": 0.5025724785626786, + "grad_norm": 2.164423942565918, + "learning_rate": 9.962164372359276e-06, + "loss": 2.464317512512207, + "step": 184620 + }, + { + "epoch": 0.5026269225534231, + "grad_norm": 2.1606414318084717, + "learning_rate": 9.961074008450322e-06, + "loss": 2.488423156738281, + "step": 184640 + }, + { + "epoch": 0.5026813665441677, + "grad_norm": 2.1398375034332275, + "learning_rate": 9.959983644541367e-06, + "loss": 2.4706293106079102, + "step": 184660 + }, + { + "epoch": 0.5027358105349122, + "grad_norm": 2.4068946838378906, + "learning_rate": 9.958893280632412e-06, + "loss": 2.449069595336914, + "step": 184680 + }, + { + "epoch": 0.5027902545256567, + "grad_norm": 2.4581046104431152, + "learning_rate": 9.957802916723457e-06, + "loss": 2.512983703613281, + "step": 184700 + }, + { + "epoch": 0.5028446985164012, + "grad_norm": 2.0687456130981445, + "learning_rate": 9.956712552814503e-06, + "loss": 2.495012092590332, + "step": 184720 + }, + { + "epoch": 0.5028991425071457, + "grad_norm": 2.1410021781921387, + "learning_rate": 9.955622188905549e-06, + "loss": 2.4467788696289063, + "step": 184740 + }, + { + "epoch": 0.5029535864978903, + "grad_norm": 2.242689609527588, + "learning_rate": 9.954531824996593e-06, + "loss": 2.4759296417236327, + "step": 184760 + }, + { + "epoch": 0.5030080304886348, + "grad_norm": 2.0445594787597656, + "learning_rate": 9.953441461087639e-06, + "loss": 2.461440658569336, + "step": 184780 + }, + { + "epoch": 0.5030624744793794, + "grad_norm": 2.1667165756225586, + "learning_rate": 9.952351097178685e-06, + "loss": 2.4781620025634767, + "step": 184800 + }, + { + "epoch": 0.5031169184701239, + "grad_norm": 2.1274073123931885, + "learning_rate": 9.95126073326973e-06, + "loss": 2.456834411621094, + "step": 184820 + }, + { + "epoch": 0.5031713624608684, + "grad_norm": 2.236628532409668, + "learning_rate": 9.950170369360775e-06, + "loss": 2.498655319213867, + "step": 184840 + }, + { + "epoch": 0.5032258064516129, + "grad_norm": 2.112529754638672, + "learning_rate": 9.94908000545182e-06, + "loss": 2.531208038330078, + "step": 184860 + }, + { + "epoch": 0.5032802504423575, + "grad_norm": 2.232593059539795, + "learning_rate": 9.947989641542866e-06, + "loss": 2.502592658996582, + "step": 184880 + }, + { + "epoch": 0.503334694433102, + "grad_norm": 2.1429359912872314, + "learning_rate": 9.946899277633912e-06, + "loss": 2.450412559509277, + "step": 184900 + }, + { + "epoch": 0.5033891384238465, + "grad_norm": 2.0419130325317383, + "learning_rate": 9.945808913724956e-06, + "loss": 2.5000715255737305, + "step": 184920 + }, + { + "epoch": 0.503443582414591, + "grad_norm": 2.232085943222046, + "learning_rate": 9.944718549816002e-06, + "loss": 2.4885616302490234, + "step": 184940 + }, + { + "epoch": 0.5034980264053355, + "grad_norm": 2.102576494216919, + "learning_rate": 9.943628185907048e-06, + "loss": 2.495726203918457, + "step": 184960 + }, + { + "epoch": 0.5035524703960801, + "grad_norm": 2.300309181213379, + "learning_rate": 9.942537821998094e-06, + "loss": 2.458152961730957, + "step": 184980 + }, + { + "epoch": 0.5036069143868246, + "grad_norm": 2.3137121200561523, + "learning_rate": 9.941447458089138e-06, + "loss": 2.5052494049072265, + "step": 185000 + }, + { + "epoch": 0.5036613583775691, + "grad_norm": 2.226203203201294, + "learning_rate": 9.940357094180184e-06, + "loss": 2.47880859375, + "step": 185020 + }, + { + "epoch": 0.5037158023683136, + "grad_norm": 3.3453257083892822, + "learning_rate": 9.93926673027123e-06, + "loss": 2.488750457763672, + "step": 185040 + }, + { + "epoch": 0.5037702463590581, + "grad_norm": 2.1464390754699707, + "learning_rate": 9.938176366362275e-06, + "loss": 2.480289649963379, + "step": 185060 + }, + { + "epoch": 0.5038246903498026, + "grad_norm": 2.1998908519744873, + "learning_rate": 9.93708600245332e-06, + "loss": 2.4861417770385743, + "step": 185080 + }, + { + "epoch": 0.5038791343405472, + "grad_norm": 2.111652135848999, + "learning_rate": 9.935995638544365e-06, + "loss": 2.4628929138183593, + "step": 185100 + }, + { + "epoch": 0.5039335783312917, + "grad_norm": 2.192619562149048, + "learning_rate": 9.934905274635411e-06, + "loss": 2.494565010070801, + "step": 185120 + }, + { + "epoch": 0.5039880223220362, + "grad_norm": 2.12092661857605, + "learning_rate": 9.933814910726457e-06, + "loss": 2.497538375854492, + "step": 185140 + }, + { + "epoch": 0.5040424663127807, + "grad_norm": 2.176778554916382, + "learning_rate": 9.9327245468175e-06, + "loss": 2.4748302459716798, + "step": 185160 + }, + { + "epoch": 0.5040969103035252, + "grad_norm": 2.143402338027954, + "learning_rate": 9.931634182908547e-06, + "loss": 2.4751781463623046, + "step": 185180 + }, + { + "epoch": 0.5041513542942697, + "grad_norm": 2.100630044937134, + "learning_rate": 9.930543818999592e-06, + "loss": 2.4873836517333983, + "step": 185200 + }, + { + "epoch": 0.5042057982850143, + "grad_norm": 2.2505321502685547, + "learning_rate": 9.929453455090638e-06, + "loss": 2.481706428527832, + "step": 185220 + }, + { + "epoch": 0.5042602422757588, + "grad_norm": 2.7873995304107666, + "learning_rate": 9.928363091181682e-06, + "loss": 2.500733184814453, + "step": 185240 + }, + { + "epoch": 0.5043146862665033, + "grad_norm": 2.1558680534362793, + "learning_rate": 9.927272727272728e-06, + "loss": 2.4761529922485352, + "step": 185260 + }, + { + "epoch": 0.5043691302572478, + "grad_norm": 2.1628928184509277, + "learning_rate": 9.926182363363774e-06, + "loss": 2.4888416290283204, + "step": 185280 + }, + { + "epoch": 0.5044235742479923, + "grad_norm": 2.1904587745666504, + "learning_rate": 9.925091999454818e-06, + "loss": 2.484037971496582, + "step": 185300 + }, + { + "epoch": 0.504478018238737, + "grad_norm": 2.213967800140381, + "learning_rate": 9.924001635545864e-06, + "loss": 2.4918485641479493, + "step": 185320 + }, + { + "epoch": 0.5045324622294814, + "grad_norm": 2.0740740299224854, + "learning_rate": 9.92291127163691e-06, + "loss": 2.4962299346923826, + "step": 185340 + }, + { + "epoch": 0.504586906220226, + "grad_norm": 4.189365863800049, + "learning_rate": 9.921820907727956e-06, + "loss": 2.4712570190429686, + "step": 185360 + }, + { + "epoch": 0.5046413502109705, + "grad_norm": 2.104133129119873, + "learning_rate": 9.920730543819e-06, + "loss": 2.4983592987060548, + "step": 185380 + }, + { + "epoch": 0.504695794201715, + "grad_norm": 2.1374454498291016, + "learning_rate": 9.919640179910045e-06, + "loss": 2.4805917739868164, + "step": 185400 + }, + { + "epoch": 0.5047502381924595, + "grad_norm": 2.1128807067871094, + "learning_rate": 9.918549816001091e-06, + "loss": 2.4885833740234373, + "step": 185420 + }, + { + "epoch": 0.5048046821832041, + "grad_norm": 2.2474446296691895, + "learning_rate": 9.917459452092137e-06, + "loss": 2.488362503051758, + "step": 185440 + }, + { + "epoch": 0.5048591261739486, + "grad_norm": 2.1689014434814453, + "learning_rate": 9.916369088183181e-06, + "loss": 2.4814031600952147, + "step": 185460 + }, + { + "epoch": 0.5049135701646931, + "grad_norm": 2.296215534210205, + "learning_rate": 9.915278724274227e-06, + "loss": 2.4784440994262695, + "step": 185480 + }, + { + "epoch": 0.5049680141554376, + "grad_norm": 2.153886318206787, + "learning_rate": 9.914188360365273e-06, + "loss": 2.4701290130615234, + "step": 185500 + }, + { + "epoch": 0.5050224581461821, + "grad_norm": 2.0354697704315186, + "learning_rate": 9.913097996456319e-06, + "loss": 2.481778144836426, + "step": 185520 + }, + { + "epoch": 0.5050769021369267, + "grad_norm": 2.048154830932617, + "learning_rate": 9.912007632547363e-06, + "loss": 2.500388336181641, + "step": 185540 + }, + { + "epoch": 0.5051313461276712, + "grad_norm": 2.0704329013824463, + "learning_rate": 9.910917268638409e-06, + "loss": 2.507397270202637, + "step": 185560 + }, + { + "epoch": 0.5051857901184157, + "grad_norm": 2.1092352867126465, + "learning_rate": 9.909826904729454e-06, + "loss": 2.478534698486328, + "step": 185580 + }, + { + "epoch": 0.5052402341091602, + "grad_norm": 2.2821240425109863, + "learning_rate": 9.9087365408205e-06, + "loss": 2.443779182434082, + "step": 185600 + }, + { + "epoch": 0.5052946780999047, + "grad_norm": 2.7114648818969727, + "learning_rate": 9.907646176911544e-06, + "loss": 2.447579002380371, + "step": 185620 + }, + { + "epoch": 0.5053491220906492, + "grad_norm": 2.1395962238311768, + "learning_rate": 9.90655581300259e-06, + "loss": 2.496380615234375, + "step": 185640 + }, + { + "epoch": 0.5054035660813938, + "grad_norm": 2.0731372833251953, + "learning_rate": 9.905465449093636e-06, + "loss": 2.4988746643066406, + "step": 185660 + }, + { + "epoch": 0.5054580100721383, + "grad_norm": 2.135990858078003, + "learning_rate": 9.904375085184682e-06, + "loss": 2.5008823394775392, + "step": 185680 + }, + { + "epoch": 0.5055124540628828, + "grad_norm": 2.4280881881713867, + "learning_rate": 9.903284721275726e-06, + "loss": 2.4910640716552734, + "step": 185700 + }, + { + "epoch": 0.5055668980536273, + "grad_norm": 2.148585557937622, + "learning_rate": 9.902194357366772e-06, + "loss": 2.5066459655761717, + "step": 185720 + }, + { + "epoch": 0.5056213420443718, + "grad_norm": 2.2278146743774414, + "learning_rate": 9.901103993457817e-06, + "loss": 2.472169303894043, + "step": 185740 + }, + { + "epoch": 0.5056757860351164, + "grad_norm": 2.1349942684173584, + "learning_rate": 9.900013629548863e-06, + "loss": 2.4975296020507813, + "step": 185760 + }, + { + "epoch": 0.5057302300258609, + "grad_norm": 2.1432604789733887, + "learning_rate": 9.898923265639907e-06, + "loss": 2.4861305236816404, + "step": 185780 + }, + { + "epoch": 0.5057846740166054, + "grad_norm": 2.0781447887420654, + "learning_rate": 9.897832901730953e-06, + "loss": 2.4806707382202147, + "step": 185800 + }, + { + "epoch": 0.5058391180073499, + "grad_norm": 2.1904091835021973, + "learning_rate": 9.896742537821999e-06, + "loss": 2.4928964614868163, + "step": 185820 + }, + { + "epoch": 0.5058935619980944, + "grad_norm": 2.164152145385742, + "learning_rate": 9.895652173913045e-06, + "loss": 2.4689184188842774, + "step": 185840 + }, + { + "epoch": 0.5059480059888389, + "grad_norm": 2.20324969291687, + "learning_rate": 9.894561810004089e-06, + "loss": 2.481284332275391, + "step": 185860 + }, + { + "epoch": 0.5060024499795835, + "grad_norm": 2.300929069519043, + "learning_rate": 9.893471446095135e-06, + "loss": 2.462899589538574, + "step": 185880 + }, + { + "epoch": 0.506056893970328, + "grad_norm": 2.17030930519104, + "learning_rate": 9.89238108218618e-06, + "loss": 2.491585540771484, + "step": 185900 + }, + { + "epoch": 0.5061113379610725, + "grad_norm": 2.1863765716552734, + "learning_rate": 9.891290718277226e-06, + "loss": 2.4845014572143556, + "step": 185920 + }, + { + "epoch": 0.506165781951817, + "grad_norm": 2.0791289806365967, + "learning_rate": 9.89020035436827e-06, + "loss": 2.4511734008789063, + "step": 185940 + }, + { + "epoch": 0.5062202259425616, + "grad_norm": 2.1652305126190186, + "learning_rate": 9.889109990459316e-06, + "loss": 2.4901636123657225, + "step": 185960 + }, + { + "epoch": 0.5062746699333062, + "grad_norm": 2.098015069961548, + "learning_rate": 9.888019626550362e-06, + "loss": 2.492101860046387, + "step": 185980 + }, + { + "epoch": 0.5063291139240507, + "grad_norm": 2.1685538291931152, + "learning_rate": 9.886929262641408e-06, + "loss": 2.492233085632324, + "step": 186000 + }, + { + "epoch": 0.5063835579147952, + "grad_norm": 2.2080295085906982, + "learning_rate": 9.885838898732452e-06, + "loss": 2.4487674713134764, + "step": 186020 + }, + { + "epoch": 0.5064380019055397, + "grad_norm": 2.13836932182312, + "learning_rate": 9.884748534823498e-06, + "loss": 2.505327606201172, + "step": 186040 + }, + { + "epoch": 0.5064924458962842, + "grad_norm": 2.139744997024536, + "learning_rate": 9.883658170914544e-06, + "loss": 2.4597404479980467, + "step": 186060 + }, + { + "epoch": 0.5065468898870287, + "grad_norm": 2.179255247116089, + "learning_rate": 9.88256780700559e-06, + "loss": 2.482710838317871, + "step": 186080 + }, + { + "epoch": 0.5066013338777733, + "grad_norm": 2.451721668243408, + "learning_rate": 9.881477443096634e-06, + "loss": 2.516783905029297, + "step": 186100 + }, + { + "epoch": 0.5066557778685178, + "grad_norm": 8.967924118041992, + "learning_rate": 9.88038707918768e-06, + "loss": 2.4880191802978517, + "step": 186120 + }, + { + "epoch": 0.5067102218592623, + "grad_norm": 2.262861728668213, + "learning_rate": 9.879296715278725e-06, + "loss": 2.486958312988281, + "step": 186140 + }, + { + "epoch": 0.5067646658500068, + "grad_norm": 2.2060680389404297, + "learning_rate": 9.878206351369771e-06, + "loss": 2.486790084838867, + "step": 186160 + }, + { + "epoch": 0.5068191098407513, + "grad_norm": 2.128450632095337, + "learning_rate": 9.877115987460815e-06, + "loss": 2.4987720489501952, + "step": 186180 + }, + { + "epoch": 0.5068735538314958, + "grad_norm": 2.1062190532684326, + "learning_rate": 9.876025623551861e-06, + "loss": 2.461574172973633, + "step": 186200 + }, + { + "epoch": 0.5069279978222404, + "grad_norm": 2.2295081615448, + "learning_rate": 9.874935259642907e-06, + "loss": 2.505276679992676, + "step": 186220 + }, + { + "epoch": 0.5069824418129849, + "grad_norm": 2.130950927734375, + "learning_rate": 9.873844895733952e-06, + "loss": 2.4678857803344725, + "step": 186240 + }, + { + "epoch": 0.5070368858037294, + "grad_norm": 4.568525314331055, + "learning_rate": 9.872754531824997e-06, + "loss": 2.4536260604858398, + "step": 186260 + }, + { + "epoch": 0.5070913297944739, + "grad_norm": 2.142265558242798, + "learning_rate": 9.871664167916042e-06, + "loss": 2.4661144256591796, + "step": 186280 + }, + { + "epoch": 0.5071457737852184, + "grad_norm": 2.127964973449707, + "learning_rate": 9.870573804007088e-06, + "loss": 2.4738080978393553, + "step": 186300 + }, + { + "epoch": 0.507200217775963, + "grad_norm": 2.194885015487671, + "learning_rate": 9.869483440098134e-06, + "loss": 2.5239049911499025, + "step": 186320 + }, + { + "epoch": 0.5072546617667075, + "grad_norm": 2.251798629760742, + "learning_rate": 9.868393076189178e-06, + "loss": 2.4788726806640624, + "step": 186340 + }, + { + "epoch": 0.507309105757452, + "grad_norm": 2.081699848175049, + "learning_rate": 9.867302712280224e-06, + "loss": 2.447335624694824, + "step": 186360 + }, + { + "epoch": 0.5073635497481965, + "grad_norm": 2.101118564605713, + "learning_rate": 9.86621234837127e-06, + "loss": 2.4722713470458983, + "step": 186380 + }, + { + "epoch": 0.507417993738941, + "grad_norm": 2.175921678543091, + "learning_rate": 9.865121984462314e-06, + "loss": 2.515153503417969, + "step": 186400 + }, + { + "epoch": 0.5074724377296855, + "grad_norm": 2.1252565383911133, + "learning_rate": 9.86403162055336e-06, + "loss": 2.4982231140136717, + "step": 186420 + }, + { + "epoch": 0.5075268817204301, + "grad_norm": 2.2312216758728027, + "learning_rate": 9.862941256644406e-06, + "loss": 2.4941707611083985, + "step": 186440 + }, + { + "epoch": 0.5075813257111746, + "grad_norm": 2.097083330154419, + "learning_rate": 9.861850892735451e-06, + "loss": 2.4855138778686525, + "step": 186460 + }, + { + "epoch": 0.5076357697019191, + "grad_norm": 2.129960536956787, + "learning_rate": 9.860760528826495e-06, + "loss": 2.511344337463379, + "step": 186480 + }, + { + "epoch": 0.5076902136926636, + "grad_norm": 2.2091727256774902, + "learning_rate": 9.859670164917541e-06, + "loss": 2.4495519638061523, + "step": 186500 + }, + { + "epoch": 0.5077446576834082, + "grad_norm": 2.134902238845825, + "learning_rate": 9.858579801008587e-06, + "loss": 2.4854429244995115, + "step": 186520 + }, + { + "epoch": 0.5077991016741528, + "grad_norm": 2.1366889476776123, + "learning_rate": 9.857489437099633e-06, + "loss": 2.4999298095703124, + "step": 186540 + }, + { + "epoch": 0.5078535456648973, + "grad_norm": 2.186077356338501, + "learning_rate": 9.856399073190677e-06, + "loss": 2.486313819885254, + "step": 186560 + }, + { + "epoch": 0.5079079896556418, + "grad_norm": 2.1892669200897217, + "learning_rate": 9.855308709281723e-06, + "loss": 2.505245590209961, + "step": 186580 + }, + { + "epoch": 0.5079624336463863, + "grad_norm": 2.0762407779693604, + "learning_rate": 9.854218345372769e-06, + "loss": 2.478842926025391, + "step": 186600 + }, + { + "epoch": 0.5080168776371308, + "grad_norm": 2.2386562824249268, + "learning_rate": 9.853127981463814e-06, + "loss": 2.492212677001953, + "step": 186620 + }, + { + "epoch": 0.5080713216278753, + "grad_norm": 2.3221888542175293, + "learning_rate": 9.852037617554859e-06, + "loss": 2.4985015869140623, + "step": 186640 + }, + { + "epoch": 0.5081257656186199, + "grad_norm": 2.0813357830047607, + "learning_rate": 9.850947253645904e-06, + "loss": 2.4596681594848633, + "step": 186660 + }, + { + "epoch": 0.5081802096093644, + "grad_norm": 2.0837268829345703, + "learning_rate": 9.84985688973695e-06, + "loss": 2.4900959014892576, + "step": 186680 + }, + { + "epoch": 0.5082346536001089, + "grad_norm": 2.205601930618286, + "learning_rate": 9.848766525827996e-06, + "loss": 2.4842531204223635, + "step": 186700 + }, + { + "epoch": 0.5082890975908534, + "grad_norm": 2.124908685684204, + "learning_rate": 9.84767616191904e-06, + "loss": 2.50721435546875, + "step": 186720 + }, + { + "epoch": 0.5083435415815979, + "grad_norm": 2.0649263858795166, + "learning_rate": 9.846585798010086e-06, + "loss": 2.4945405960083007, + "step": 186740 + }, + { + "epoch": 0.5083979855723425, + "grad_norm": 2.364466667175293, + "learning_rate": 9.845495434101132e-06, + "loss": 2.469399070739746, + "step": 186760 + }, + { + "epoch": 0.508452429563087, + "grad_norm": 2.1814568042755127, + "learning_rate": 9.844405070192177e-06, + "loss": 2.497334289550781, + "step": 186780 + }, + { + "epoch": 0.5085068735538315, + "grad_norm": 2.4921834468841553, + "learning_rate": 9.843314706283222e-06, + "loss": 2.484260368347168, + "step": 186800 + }, + { + "epoch": 0.508561317544576, + "grad_norm": 2.118398427963257, + "learning_rate": 9.842224342374267e-06, + "loss": 2.4723106384277345, + "step": 186820 + }, + { + "epoch": 0.5086157615353205, + "grad_norm": 2.1561527252197266, + "learning_rate": 9.841133978465313e-06, + "loss": 2.503916358947754, + "step": 186840 + }, + { + "epoch": 0.508670205526065, + "grad_norm": 2.281292676925659, + "learning_rate": 9.840043614556359e-06, + "loss": 2.4911861419677734, + "step": 186860 + }, + { + "epoch": 0.5087246495168096, + "grad_norm": 2.1029810905456543, + "learning_rate": 9.838953250647403e-06, + "loss": 2.464656448364258, + "step": 186880 + }, + { + "epoch": 0.5087790935075541, + "grad_norm": 2.1175129413604736, + "learning_rate": 9.837862886738449e-06, + "loss": 2.500332069396973, + "step": 186900 + }, + { + "epoch": 0.5088335374982986, + "grad_norm": 2.2530388832092285, + "learning_rate": 9.836772522829495e-06, + "loss": 2.521631622314453, + "step": 186920 + }, + { + "epoch": 0.5088879814890431, + "grad_norm": 2.037410259246826, + "learning_rate": 9.83568215892054e-06, + "loss": 2.4854135513305664, + "step": 186940 + }, + { + "epoch": 0.5089424254797876, + "grad_norm": 2.1818675994873047, + "learning_rate": 9.834591795011585e-06, + "loss": 2.469869613647461, + "step": 186960 + }, + { + "epoch": 0.5089968694705322, + "grad_norm": 2.186673164367676, + "learning_rate": 9.83350143110263e-06, + "loss": 2.462118148803711, + "step": 186980 + }, + { + "epoch": 0.5090513134612767, + "grad_norm": 2.1534063816070557, + "learning_rate": 9.832411067193676e-06, + "loss": 2.4650447845458983, + "step": 187000 + }, + { + "epoch": 0.5091057574520212, + "grad_norm": 2.241849660873413, + "learning_rate": 9.831320703284722e-06, + "loss": 2.486835479736328, + "step": 187020 + }, + { + "epoch": 0.5091602014427657, + "grad_norm": 2.0821800231933594, + "learning_rate": 9.830230339375766e-06, + "loss": 2.4362346649169924, + "step": 187040 + }, + { + "epoch": 0.5092146454335102, + "grad_norm": 2.2371559143066406, + "learning_rate": 9.829139975466812e-06, + "loss": 2.4694360733032226, + "step": 187060 + }, + { + "epoch": 0.5092690894242548, + "grad_norm": 2.2172915935516357, + "learning_rate": 9.828049611557858e-06, + "loss": 2.4487865447998045, + "step": 187080 + }, + { + "epoch": 0.5093235334149994, + "grad_norm": 2.1449553966522217, + "learning_rate": 9.826959247648904e-06, + "loss": 2.4394296646118163, + "step": 187100 + }, + { + "epoch": 0.5093779774057439, + "grad_norm": 2.1027932167053223, + "learning_rate": 9.825868883739948e-06, + "loss": 2.4758974075317384, + "step": 187120 + }, + { + "epoch": 0.5094324213964884, + "grad_norm": 2.187051773071289, + "learning_rate": 9.824778519830994e-06, + "loss": 2.4990997314453125, + "step": 187140 + }, + { + "epoch": 0.5094868653872329, + "grad_norm": 2.114523410797119, + "learning_rate": 9.82368815592204e-06, + "loss": 2.442812919616699, + "step": 187160 + }, + { + "epoch": 0.5095413093779774, + "grad_norm": 2.259842872619629, + "learning_rate": 9.822597792013085e-06, + "loss": 2.4764413833618164, + "step": 187180 + }, + { + "epoch": 0.509595753368722, + "grad_norm": 2.1345367431640625, + "learning_rate": 9.82150742810413e-06, + "loss": 2.4293270111083984, + "step": 187200 + }, + { + "epoch": 0.5096501973594665, + "grad_norm": 2.3711981773376465, + "learning_rate": 9.820417064195175e-06, + "loss": 2.4877450942993162, + "step": 187220 + }, + { + "epoch": 0.509704641350211, + "grad_norm": 2.1700026988983154, + "learning_rate": 9.819326700286221e-06, + "loss": 2.483145332336426, + "step": 187240 + }, + { + "epoch": 0.5097590853409555, + "grad_norm": 2.06642746925354, + "learning_rate": 9.818236336377267e-06, + "loss": 2.5317266464233397, + "step": 187260 + }, + { + "epoch": 0.5098135293317, + "grad_norm": 2.163576602935791, + "learning_rate": 9.817145972468311e-06, + "loss": 2.484200668334961, + "step": 187280 + }, + { + "epoch": 0.5098679733224445, + "grad_norm": 2.2328429222106934, + "learning_rate": 9.816055608559357e-06, + "loss": 2.4901695251464844, + "step": 187300 + }, + { + "epoch": 0.5099224173131891, + "grad_norm": 2.273700475692749, + "learning_rate": 9.814965244650402e-06, + "loss": 2.465257453918457, + "step": 187320 + }, + { + "epoch": 0.5099768613039336, + "grad_norm": 2.3629698753356934, + "learning_rate": 9.813874880741448e-06, + "loss": 2.4594070434570314, + "step": 187340 + }, + { + "epoch": 0.5100313052946781, + "grad_norm": 2.039821147918701, + "learning_rate": 9.812784516832492e-06, + "loss": 2.480612945556641, + "step": 187360 + }, + { + "epoch": 0.5100857492854226, + "grad_norm": 2.0914466381073, + "learning_rate": 9.811694152923538e-06, + "loss": 2.474275588989258, + "step": 187380 + }, + { + "epoch": 0.5101401932761671, + "grad_norm": 2.2020280361175537, + "learning_rate": 9.810603789014584e-06, + "loss": 2.4985734939575197, + "step": 187400 + }, + { + "epoch": 0.5101946372669116, + "grad_norm": 2.1879796981811523, + "learning_rate": 9.80951342510563e-06, + "loss": 2.4981584548950195, + "step": 187420 + }, + { + "epoch": 0.5102490812576562, + "grad_norm": 2.3574697971343994, + "learning_rate": 9.808423061196674e-06, + "loss": 2.47867431640625, + "step": 187440 + }, + { + "epoch": 0.5103035252484007, + "grad_norm": 2.1854217052459717, + "learning_rate": 9.80733269728772e-06, + "loss": 2.4789239883422853, + "step": 187460 + }, + { + "epoch": 0.5103579692391452, + "grad_norm": 2.322016954421997, + "learning_rate": 9.806242333378766e-06, + "loss": 2.478228759765625, + "step": 187480 + }, + { + "epoch": 0.5104124132298897, + "grad_norm": 2.2629177570343018, + "learning_rate": 9.805151969469811e-06, + "loss": 2.476701354980469, + "step": 187500 + }, + { + "epoch": 0.5104668572206342, + "grad_norm": 2.5157968997955322, + "learning_rate": 9.804061605560857e-06, + "loss": 2.4642810821533203, + "step": 187520 + }, + { + "epoch": 0.5105213012113788, + "grad_norm": 2.1089589595794678, + "learning_rate": 9.802971241651901e-06, + "loss": 2.4543214797973634, + "step": 187540 + }, + { + "epoch": 0.5105757452021233, + "grad_norm": 2.261826992034912, + "learning_rate": 9.801880877742947e-06, + "loss": 2.481661796569824, + "step": 187560 + }, + { + "epoch": 0.5106301891928678, + "grad_norm": 2.118276596069336, + "learning_rate": 9.800790513833993e-06, + "loss": 2.4652299880981445, + "step": 187580 + }, + { + "epoch": 0.5106846331836123, + "grad_norm": 2.3139772415161133, + "learning_rate": 9.799700149925039e-06, + "loss": 2.4502105712890625, + "step": 187600 + }, + { + "epoch": 0.5107390771743568, + "grad_norm": 2.428715467453003, + "learning_rate": 9.798609786016083e-06, + "loss": 2.5318126678466797, + "step": 187620 + }, + { + "epoch": 0.5107935211651013, + "grad_norm": 2.3083760738372803, + "learning_rate": 9.797519422107129e-06, + "loss": 2.462466812133789, + "step": 187640 + }, + { + "epoch": 0.510847965155846, + "grad_norm": 2.2184455394744873, + "learning_rate": 9.796429058198174e-06, + "loss": 2.4928871154785157, + "step": 187660 + }, + { + "epoch": 0.5109024091465905, + "grad_norm": 2.1643712520599365, + "learning_rate": 9.79533869428922e-06, + "loss": 2.5026811599731444, + "step": 187680 + }, + { + "epoch": 0.510956853137335, + "grad_norm": 2.1848137378692627, + "learning_rate": 9.794248330380264e-06, + "loss": 2.4508975982666015, + "step": 187700 + }, + { + "epoch": 0.5110112971280795, + "grad_norm": 2.4098093509674072, + "learning_rate": 9.79315796647131e-06, + "loss": 2.4543331146240233, + "step": 187720 + }, + { + "epoch": 0.511065741118824, + "grad_norm": 2.228724956512451, + "learning_rate": 9.792067602562356e-06, + "loss": 2.5111597061157225, + "step": 187740 + }, + { + "epoch": 0.5111201851095686, + "grad_norm": 2.094755172729492, + "learning_rate": 9.790977238653402e-06, + "loss": 2.478143501281738, + "step": 187760 + }, + { + "epoch": 0.5111746291003131, + "grad_norm": 2.053556442260742, + "learning_rate": 9.789886874744446e-06, + "loss": 2.4740053176879884, + "step": 187780 + }, + { + "epoch": 0.5112290730910576, + "grad_norm": 2.1809396743774414, + "learning_rate": 9.788796510835492e-06, + "loss": 2.4632499694824217, + "step": 187800 + }, + { + "epoch": 0.5112835170818021, + "grad_norm": 2.2164647579193115, + "learning_rate": 9.787706146926538e-06, + "loss": 2.475328826904297, + "step": 187820 + }, + { + "epoch": 0.5113379610725466, + "grad_norm": 2.3975911140441895, + "learning_rate": 9.786615783017583e-06, + "loss": 2.483383560180664, + "step": 187840 + }, + { + "epoch": 0.5113924050632911, + "grad_norm": 2.069035768508911, + "learning_rate": 9.785525419108627e-06, + "loss": 2.4389974594116213, + "step": 187860 + }, + { + "epoch": 0.5114468490540357, + "grad_norm": 2.199920654296875, + "learning_rate": 9.784435055199673e-06, + "loss": 2.498091697692871, + "step": 187880 + }, + { + "epoch": 0.5115012930447802, + "grad_norm": 2.193959951400757, + "learning_rate": 9.783344691290719e-06, + "loss": 2.488515853881836, + "step": 187900 + }, + { + "epoch": 0.5115557370355247, + "grad_norm": 2.1761386394500732, + "learning_rate": 9.782254327381765e-06, + "loss": 2.477163887023926, + "step": 187920 + }, + { + "epoch": 0.5116101810262692, + "grad_norm": 2.136206865310669, + "learning_rate": 9.781163963472809e-06, + "loss": 2.467350959777832, + "step": 187940 + }, + { + "epoch": 0.5116646250170137, + "grad_norm": 2.1832778453826904, + "learning_rate": 9.780073599563855e-06, + "loss": 2.4747268676757814, + "step": 187960 + }, + { + "epoch": 0.5117190690077583, + "grad_norm": 2.2094569206237793, + "learning_rate": 9.7789832356549e-06, + "loss": 2.4537160873413084, + "step": 187980 + }, + { + "epoch": 0.5117735129985028, + "grad_norm": 2.1398282051086426, + "learning_rate": 9.777892871745946e-06, + "loss": 2.4933677673339845, + "step": 188000 + }, + { + "epoch": 0.5118279569892473, + "grad_norm": 2.108473062515259, + "learning_rate": 9.77680250783699e-06, + "loss": 2.4588550567626952, + "step": 188020 + }, + { + "epoch": 0.5118824009799918, + "grad_norm": 2.179044008255005, + "learning_rate": 9.775712143928036e-06, + "loss": 2.4992441177368163, + "step": 188040 + }, + { + "epoch": 0.5119368449707363, + "grad_norm": 2.2670342922210693, + "learning_rate": 9.774621780019082e-06, + "loss": 2.4796751022338865, + "step": 188060 + }, + { + "epoch": 0.5119912889614808, + "grad_norm": 2.1655168533325195, + "learning_rate": 9.773531416110128e-06, + "loss": 2.5029375076293947, + "step": 188080 + }, + { + "epoch": 0.5120457329522254, + "grad_norm": 2.1981680393218994, + "learning_rate": 9.772441052201172e-06, + "loss": 2.4804319381713866, + "step": 188100 + }, + { + "epoch": 0.5121001769429699, + "grad_norm": 2.1589138507843018, + "learning_rate": 9.771350688292218e-06, + "loss": 2.4646934509277343, + "step": 188120 + }, + { + "epoch": 0.5121546209337144, + "grad_norm": 2.176344394683838, + "learning_rate": 9.770260324383264e-06, + "loss": 2.487287902832031, + "step": 188140 + }, + { + "epoch": 0.512209064924459, + "grad_norm": 2.123802900314331, + "learning_rate": 9.76916996047431e-06, + "loss": 2.4748760223388673, + "step": 188160 + }, + { + "epoch": 0.5122635089152034, + "grad_norm": 2.3078577518463135, + "learning_rate": 9.768079596565354e-06, + "loss": 2.4595596313476564, + "step": 188180 + }, + { + "epoch": 0.5123179529059481, + "grad_norm": 2.1368463039398193, + "learning_rate": 9.7669892326564e-06, + "loss": 2.4955986022949217, + "step": 188200 + }, + { + "epoch": 0.5123723968966926, + "grad_norm": 2.224682092666626, + "learning_rate": 9.765898868747445e-06, + "loss": 2.532529830932617, + "step": 188220 + }, + { + "epoch": 0.5124268408874371, + "grad_norm": 2.2571308612823486, + "learning_rate": 9.764808504838491e-06, + "loss": 2.4882102966308595, + "step": 188240 + }, + { + "epoch": 0.5124812848781816, + "grad_norm": 2.0804035663604736, + "learning_rate": 9.763718140929535e-06, + "loss": 2.484083557128906, + "step": 188260 + }, + { + "epoch": 0.5125357288689261, + "grad_norm": 2.99584698677063, + "learning_rate": 9.762627777020581e-06, + "loss": 2.4900001525878905, + "step": 188280 + }, + { + "epoch": 0.5125901728596706, + "grad_norm": 2.06084942817688, + "learning_rate": 9.761537413111627e-06, + "loss": 2.4892526626586915, + "step": 188300 + }, + { + "epoch": 0.5126446168504152, + "grad_norm": 2.0988073348999023, + "learning_rate": 9.760447049202673e-06, + "loss": 2.477306365966797, + "step": 188320 + }, + { + "epoch": 0.5126990608411597, + "grad_norm": 2.211899757385254, + "learning_rate": 9.759356685293717e-06, + "loss": 2.450870895385742, + "step": 188340 + }, + { + "epoch": 0.5127535048319042, + "grad_norm": 2.0858395099639893, + "learning_rate": 9.758266321384763e-06, + "loss": 2.4870004653930664, + "step": 188360 + }, + { + "epoch": 0.5128079488226487, + "grad_norm": 2.1521103382110596, + "learning_rate": 9.757175957475808e-06, + "loss": 2.4608692169189452, + "step": 188380 + }, + { + "epoch": 0.5128623928133932, + "grad_norm": 2.2046687602996826, + "learning_rate": 9.756085593566854e-06, + "loss": 2.484829902648926, + "step": 188400 + }, + { + "epoch": 0.5129168368041377, + "grad_norm": 2.080127477645874, + "learning_rate": 9.754995229657898e-06, + "loss": 2.479985237121582, + "step": 188420 + }, + { + "epoch": 0.5129712807948823, + "grad_norm": 2.1237926483154297, + "learning_rate": 9.753904865748944e-06, + "loss": 2.477619171142578, + "step": 188440 + }, + { + "epoch": 0.5130257247856268, + "grad_norm": 2.1623215675354004, + "learning_rate": 9.75281450183999e-06, + "loss": 2.465007019042969, + "step": 188460 + }, + { + "epoch": 0.5130801687763713, + "grad_norm": 2.2599713802337646, + "learning_rate": 9.751724137931036e-06, + "loss": 2.5047687530517577, + "step": 188480 + }, + { + "epoch": 0.5131346127671158, + "grad_norm": 2.205198287963867, + "learning_rate": 9.75063377402208e-06, + "loss": 2.4596878051757813, + "step": 188500 + }, + { + "epoch": 0.5131890567578603, + "grad_norm": 4.41204309463501, + "learning_rate": 9.749543410113126e-06, + "loss": 2.470267677307129, + "step": 188520 + }, + { + "epoch": 0.5132435007486049, + "grad_norm": 2.2681586742401123, + "learning_rate": 9.748453046204171e-06, + "loss": 2.4634897232055666, + "step": 188540 + }, + { + "epoch": 0.5132979447393494, + "grad_norm": 2.2383501529693604, + "learning_rate": 9.747362682295217e-06, + "loss": 2.4396291732788087, + "step": 188560 + }, + { + "epoch": 0.5133523887300939, + "grad_norm": 2.143324613571167, + "learning_rate": 9.746272318386261e-06, + "loss": 2.4788276672363283, + "step": 188580 + }, + { + "epoch": 0.5134068327208384, + "grad_norm": 2.058866262435913, + "learning_rate": 9.745181954477307e-06, + "loss": 2.511068916320801, + "step": 188600 + }, + { + "epoch": 0.5134612767115829, + "grad_norm": 2.1248910427093506, + "learning_rate": 9.744091590568353e-06, + "loss": 2.462211799621582, + "step": 188620 + }, + { + "epoch": 0.5135157207023274, + "grad_norm": 2.1054790019989014, + "learning_rate": 9.743001226659399e-06, + "loss": 2.4713443756103515, + "step": 188640 + }, + { + "epoch": 0.513570164693072, + "grad_norm": 2.128237247467041, + "learning_rate": 9.741910862750443e-06, + "loss": 2.468368148803711, + "step": 188660 + }, + { + "epoch": 0.5136246086838165, + "grad_norm": 2.1454930305480957, + "learning_rate": 9.740820498841489e-06, + "loss": 2.4971309661865235, + "step": 188680 + }, + { + "epoch": 0.513679052674561, + "grad_norm": 2.3887197971343994, + "learning_rate": 9.739730134932535e-06, + "loss": 2.461648941040039, + "step": 188700 + }, + { + "epoch": 0.5137334966653055, + "grad_norm": 2.2210235595703125, + "learning_rate": 9.73863977102358e-06, + "loss": 2.4787445068359375, + "step": 188720 + }, + { + "epoch": 0.51378794065605, + "grad_norm": 2.145885705947876, + "learning_rate": 9.737549407114624e-06, + "loss": 2.4709295272827148, + "step": 188740 + }, + { + "epoch": 0.5138423846467947, + "grad_norm": 2.258925437927246, + "learning_rate": 9.73645904320567e-06, + "loss": 2.477198600769043, + "step": 188760 + }, + { + "epoch": 0.5138968286375392, + "grad_norm": 2.0607354640960693, + "learning_rate": 9.735368679296716e-06, + "loss": 2.446755790710449, + "step": 188780 + }, + { + "epoch": 0.5139512726282837, + "grad_norm": 2.2441155910491943, + "learning_rate": 9.734278315387762e-06, + "loss": 2.4875022888183596, + "step": 188800 + }, + { + "epoch": 0.5140057166190282, + "grad_norm": 2.1283063888549805, + "learning_rate": 9.733187951478806e-06, + "loss": 2.499015998840332, + "step": 188820 + }, + { + "epoch": 0.5140601606097727, + "grad_norm": 2.149109363555908, + "learning_rate": 9.732097587569852e-06, + "loss": 2.457316017150879, + "step": 188840 + }, + { + "epoch": 0.5141146046005172, + "grad_norm": 2.202385425567627, + "learning_rate": 9.731007223660898e-06, + "loss": 2.4721221923828125, + "step": 188860 + }, + { + "epoch": 0.5141690485912618, + "grad_norm": 2.176272392272949, + "learning_rate": 9.729916859751943e-06, + "loss": 2.5151674270629885, + "step": 188880 + }, + { + "epoch": 0.5142234925820063, + "grad_norm": 2.4589145183563232, + "learning_rate": 9.728826495842988e-06, + "loss": 2.4697628021240234, + "step": 188900 + }, + { + "epoch": 0.5142779365727508, + "grad_norm": 2.173776865005493, + "learning_rate": 9.727736131934033e-06, + "loss": 2.460612678527832, + "step": 188920 + }, + { + "epoch": 0.5143323805634953, + "grad_norm": 2.1538028717041016, + "learning_rate": 9.726645768025079e-06, + "loss": 2.4779029846191407, + "step": 188940 + }, + { + "epoch": 0.5143868245542398, + "grad_norm": 2.2267568111419678, + "learning_rate": 9.725555404116125e-06, + "loss": 2.488968849182129, + "step": 188960 + }, + { + "epoch": 0.5144412685449844, + "grad_norm": 2.1194722652435303, + "learning_rate": 9.724465040207169e-06, + "loss": 2.478628158569336, + "step": 188980 + }, + { + "epoch": 0.5144957125357289, + "grad_norm": 2.14538311958313, + "learning_rate": 9.723374676298215e-06, + "loss": 2.470195770263672, + "step": 189000 + }, + { + "epoch": 0.5145501565264734, + "grad_norm": 2.0727381706237793, + "learning_rate": 9.72228431238926e-06, + "loss": 2.4787660598754884, + "step": 189020 + }, + { + "epoch": 0.5146046005172179, + "grad_norm": 2.1373510360717773, + "learning_rate": 9.721193948480306e-06, + "loss": 2.4893749237060545, + "step": 189040 + }, + { + "epoch": 0.5146590445079624, + "grad_norm": 2.1452198028564453, + "learning_rate": 9.72010358457135e-06, + "loss": 2.505074691772461, + "step": 189060 + }, + { + "epoch": 0.5147134884987069, + "grad_norm": 2.1419007778167725, + "learning_rate": 9.719013220662396e-06, + "loss": 2.473812294006348, + "step": 189080 + }, + { + "epoch": 0.5147679324894515, + "grad_norm": 2.342529296875, + "learning_rate": 9.717922856753442e-06, + "loss": 2.4825246810913084, + "step": 189100 + }, + { + "epoch": 0.514822376480196, + "grad_norm": 2.1525685787200928, + "learning_rate": 9.716832492844488e-06, + "loss": 2.445587921142578, + "step": 189120 + }, + { + "epoch": 0.5148768204709405, + "grad_norm": 2.334627389907837, + "learning_rate": 9.715742128935532e-06, + "loss": 2.500849723815918, + "step": 189140 + }, + { + "epoch": 0.514931264461685, + "grad_norm": 2.130474805831909, + "learning_rate": 9.714651765026578e-06, + "loss": 2.462619400024414, + "step": 189160 + }, + { + "epoch": 0.5149857084524295, + "grad_norm": 58.22713088989258, + "learning_rate": 9.713561401117624e-06, + "loss": 2.492750358581543, + "step": 189180 + }, + { + "epoch": 0.5150401524431741, + "grad_norm": 2.151792526245117, + "learning_rate": 9.71247103720867e-06, + "loss": 2.4946374893188477, + "step": 189200 + }, + { + "epoch": 0.5150945964339186, + "grad_norm": 3.6465938091278076, + "learning_rate": 9.711380673299715e-06, + "loss": 2.499865913391113, + "step": 189220 + }, + { + "epoch": 0.5151490404246631, + "grad_norm": 2.2601754665374756, + "learning_rate": 9.71029030939076e-06, + "loss": 2.468802261352539, + "step": 189240 + }, + { + "epoch": 0.5152034844154076, + "grad_norm": 2.130270004272461, + "learning_rate": 9.709199945481805e-06, + "loss": 2.4855218887329102, + "step": 189260 + }, + { + "epoch": 0.5152579284061521, + "grad_norm": 2.239997625350952, + "learning_rate": 9.708109581572851e-06, + "loss": 2.48472900390625, + "step": 189280 + }, + { + "epoch": 0.5153123723968966, + "grad_norm": 2.4006919860839844, + "learning_rate": 9.707019217663897e-06, + "loss": 2.471373176574707, + "step": 189300 + }, + { + "epoch": 0.5153668163876413, + "grad_norm": 2.098102331161499, + "learning_rate": 9.705928853754941e-06, + "loss": 2.470250701904297, + "step": 189320 + }, + { + "epoch": 0.5154212603783858, + "grad_norm": 2.2084782123565674, + "learning_rate": 9.704838489845987e-06, + "loss": 2.4591463088989256, + "step": 189340 + }, + { + "epoch": 0.5154757043691303, + "grad_norm": 2.200075626373291, + "learning_rate": 9.703748125937033e-06, + "loss": 2.4804834365844726, + "step": 189360 + }, + { + "epoch": 0.5155301483598748, + "grad_norm": 2.3432650566101074, + "learning_rate": 9.702657762028078e-06, + "loss": 2.5185070037841797, + "step": 189380 + }, + { + "epoch": 0.5155845923506193, + "grad_norm": 2.131369113922119, + "learning_rate": 9.701567398119123e-06, + "loss": 2.521994972229004, + "step": 189400 + }, + { + "epoch": 0.5156390363413639, + "grad_norm": 2.1619367599487305, + "learning_rate": 9.700477034210168e-06, + "loss": 2.503293037414551, + "step": 189420 + }, + { + "epoch": 0.5156934803321084, + "grad_norm": 2.156003475189209, + "learning_rate": 9.699386670301214e-06, + "loss": 2.465842437744141, + "step": 189440 + }, + { + "epoch": 0.5157479243228529, + "grad_norm": 2.245361328125, + "learning_rate": 9.69829630639226e-06, + "loss": 2.4679805755615236, + "step": 189460 + }, + { + "epoch": 0.5158023683135974, + "grad_norm": 2.1062495708465576, + "learning_rate": 9.697205942483304e-06, + "loss": 2.471622848510742, + "step": 189480 + }, + { + "epoch": 0.5158568123043419, + "grad_norm": 2.1623823642730713, + "learning_rate": 9.69611557857435e-06, + "loss": 2.4543670654296874, + "step": 189500 + }, + { + "epoch": 0.5159112562950864, + "grad_norm": 2.1501407623291016, + "learning_rate": 9.695025214665396e-06, + "loss": 2.5142375946044924, + "step": 189520 + }, + { + "epoch": 0.515965700285831, + "grad_norm": 2.2541887760162354, + "learning_rate": 9.693934850756442e-06, + "loss": 2.5036134719848633, + "step": 189540 + }, + { + "epoch": 0.5160201442765755, + "grad_norm": 2.213625431060791, + "learning_rate": 9.692844486847486e-06, + "loss": 2.4828819274902343, + "step": 189560 + }, + { + "epoch": 0.51607458826732, + "grad_norm": 2.160658597946167, + "learning_rate": 9.691754122938531e-06, + "loss": 2.4954734802246095, + "step": 189580 + }, + { + "epoch": 0.5161290322580645, + "grad_norm": 2.1364622116088867, + "learning_rate": 9.690663759029577e-06, + "loss": 2.486007881164551, + "step": 189600 + }, + { + "epoch": 0.516183476248809, + "grad_norm": 2.4800641536712646, + "learning_rate": 9.689573395120623e-06, + "loss": 2.4683698654174804, + "step": 189620 + }, + { + "epoch": 0.5162379202395535, + "grad_norm": 3.581740617752075, + "learning_rate": 9.688483031211667e-06, + "loss": 2.484444427490234, + "step": 189640 + }, + { + "epoch": 0.5162923642302981, + "grad_norm": 2.144054412841797, + "learning_rate": 9.687392667302713e-06, + "loss": 2.4714332580566407, + "step": 189660 + }, + { + "epoch": 0.5163468082210426, + "grad_norm": 2.322683095932007, + "learning_rate": 9.686302303393759e-06, + "loss": 2.4591306686401366, + "step": 189680 + }, + { + "epoch": 0.5164012522117871, + "grad_norm": 2.18135404586792, + "learning_rate": 9.685211939484805e-06, + "loss": 2.4619123458862306, + "step": 189700 + }, + { + "epoch": 0.5164556962025316, + "grad_norm": 2.2727062702178955, + "learning_rate": 9.684121575575849e-06, + "loss": 2.5242095947265626, + "step": 189720 + }, + { + "epoch": 0.5165101401932761, + "grad_norm": 2.1613717079162598, + "learning_rate": 9.683031211666895e-06, + "loss": 2.482453536987305, + "step": 189740 + }, + { + "epoch": 0.5165645841840207, + "grad_norm": 2.259465456008911, + "learning_rate": 9.68194084775794e-06, + "loss": 2.4830169677734375, + "step": 189760 + }, + { + "epoch": 0.5166190281747652, + "grad_norm": 2.0911524295806885, + "learning_rate": 9.680850483848986e-06, + "loss": 2.485546684265137, + "step": 189780 + }, + { + "epoch": 0.5166734721655097, + "grad_norm": 3.310393810272217, + "learning_rate": 9.67976011994003e-06, + "loss": 2.4804481506347655, + "step": 189800 + }, + { + "epoch": 0.5167279161562542, + "grad_norm": 2.4117238521575928, + "learning_rate": 9.678669756031076e-06, + "loss": 2.4635089874267577, + "step": 189820 + }, + { + "epoch": 0.5167823601469987, + "grad_norm": 2.143775463104248, + "learning_rate": 9.677579392122122e-06, + "loss": 2.493886375427246, + "step": 189840 + }, + { + "epoch": 0.5168368041377432, + "grad_norm": 2.065293073654175, + "learning_rate": 9.676489028213168e-06, + "loss": 2.464089584350586, + "step": 189860 + }, + { + "epoch": 0.5168912481284879, + "grad_norm": 2.241147756576538, + "learning_rate": 9.675398664304212e-06, + "loss": 2.459375762939453, + "step": 189880 + }, + { + "epoch": 0.5169456921192324, + "grad_norm": 2.202890396118164, + "learning_rate": 9.674308300395258e-06, + "loss": 2.487396240234375, + "step": 189900 + }, + { + "epoch": 0.5170001361099769, + "grad_norm": 2.2016828060150146, + "learning_rate": 9.673217936486303e-06, + "loss": 2.4811737060546877, + "step": 189920 + }, + { + "epoch": 0.5170545801007214, + "grad_norm": 2.1987459659576416, + "learning_rate": 9.67212757257735e-06, + "loss": 2.4807844161987305, + "step": 189940 + }, + { + "epoch": 0.5171090240914659, + "grad_norm": 2.139716863632202, + "learning_rate": 9.671037208668393e-06, + "loss": 2.502754783630371, + "step": 189960 + }, + { + "epoch": 0.5171634680822105, + "grad_norm": 2.158970594406128, + "learning_rate": 9.66994684475944e-06, + "loss": 2.4728864669799804, + "step": 189980 + }, + { + "epoch": 0.517217912072955, + "grad_norm": 2.1620943546295166, + "learning_rate": 9.668856480850485e-06, + "loss": 2.506363868713379, + "step": 190000 + }, + { + "epoch": 0.5172723560636995, + "grad_norm": 2.1751182079315186, + "learning_rate": 9.66776611694153e-06, + "loss": 2.516373634338379, + "step": 190020 + }, + { + "epoch": 0.517326800054444, + "grad_norm": 2.3548994064331055, + "learning_rate": 9.666675753032575e-06, + "loss": 2.4663885116577147, + "step": 190040 + }, + { + "epoch": 0.5173812440451885, + "grad_norm": 2.770960569381714, + "learning_rate": 9.66558538912362e-06, + "loss": 2.498759651184082, + "step": 190060 + }, + { + "epoch": 0.517435688035933, + "grad_norm": 2.2353219985961914, + "learning_rate": 9.664495025214667e-06, + "loss": 2.4765159606933596, + "step": 190080 + }, + { + "epoch": 0.5174901320266776, + "grad_norm": 2.31668758392334, + "learning_rate": 9.663404661305712e-06, + "loss": 2.4851615905761717, + "step": 190100 + }, + { + "epoch": 0.5175445760174221, + "grad_norm": 2.281447410583496, + "learning_rate": 9.662314297396756e-06, + "loss": 2.4609519958496096, + "step": 190120 + }, + { + "epoch": 0.5175990200081666, + "grad_norm": 2.0678229331970215, + "learning_rate": 9.661223933487802e-06, + "loss": 2.486367416381836, + "step": 190140 + }, + { + "epoch": 0.5176534639989111, + "grad_norm": 2.0738906860351562, + "learning_rate": 9.660133569578848e-06, + "loss": 2.44980411529541, + "step": 190160 + }, + { + "epoch": 0.5177079079896556, + "grad_norm": 2.1168062686920166, + "learning_rate": 9.659043205669894e-06, + "loss": 2.489178276062012, + "step": 190180 + }, + { + "epoch": 0.5177623519804002, + "grad_norm": 2.1585021018981934, + "learning_rate": 9.657952841760938e-06, + "loss": 2.5028923034667967, + "step": 190200 + }, + { + "epoch": 0.5178167959711447, + "grad_norm": 2.212557315826416, + "learning_rate": 9.656862477851984e-06, + "loss": 2.47985897064209, + "step": 190220 + }, + { + "epoch": 0.5178712399618892, + "grad_norm": 2.2170543670654297, + "learning_rate": 9.65577211394303e-06, + "loss": 2.452260971069336, + "step": 190240 + }, + { + "epoch": 0.5179256839526337, + "grad_norm": 2.2210700511932373, + "learning_rate": 9.654681750034075e-06, + "loss": 2.490283966064453, + "step": 190260 + }, + { + "epoch": 0.5179801279433782, + "grad_norm": 2.1507515907287598, + "learning_rate": 9.65359138612512e-06, + "loss": 2.478727912902832, + "step": 190280 + }, + { + "epoch": 0.5180345719341227, + "grad_norm": 2.2662789821624756, + "learning_rate": 9.652501022216165e-06, + "loss": 2.47121524810791, + "step": 190300 + }, + { + "epoch": 0.5180890159248673, + "grad_norm": 2.2286086082458496, + "learning_rate": 9.651410658307211e-06, + "loss": 2.5118124008178713, + "step": 190320 + }, + { + "epoch": 0.5181434599156118, + "grad_norm": 2.1512887477874756, + "learning_rate": 9.650320294398257e-06, + "loss": 2.5033357620239256, + "step": 190340 + }, + { + "epoch": 0.5181979039063563, + "grad_norm": 2.071197032928467, + "learning_rate": 9.649229930489301e-06, + "loss": 2.4545799255371095, + "step": 190360 + }, + { + "epoch": 0.5182523478971008, + "grad_norm": 2.161045551300049, + "learning_rate": 9.648139566580347e-06, + "loss": 2.4733306884765627, + "step": 190380 + }, + { + "epoch": 0.5183067918878453, + "grad_norm": 2.141692876815796, + "learning_rate": 9.647049202671393e-06, + "loss": 2.4890188217163085, + "step": 190400 + }, + { + "epoch": 0.51836123587859, + "grad_norm": 4.392356872558594, + "learning_rate": 9.645958838762439e-06, + "loss": 2.4576993942260743, + "step": 190420 + }, + { + "epoch": 0.5184156798693345, + "grad_norm": 2.216547727584839, + "learning_rate": 9.644868474853483e-06, + "loss": 2.502482032775879, + "step": 190440 + }, + { + "epoch": 0.518470123860079, + "grad_norm": 2.2570056915283203, + "learning_rate": 9.643778110944528e-06, + "loss": 2.5154338836669923, + "step": 190460 + }, + { + "epoch": 0.5185245678508235, + "grad_norm": 2.1154470443725586, + "learning_rate": 9.642687747035574e-06, + "loss": 2.4614782333374023, + "step": 190480 + }, + { + "epoch": 0.518579011841568, + "grad_norm": 2.074302911758423, + "learning_rate": 9.64159738312662e-06, + "loss": 2.4842046737670898, + "step": 190500 + }, + { + "epoch": 0.5186334558323125, + "grad_norm": 2.2176737785339355, + "learning_rate": 9.640507019217664e-06, + "loss": 2.4994480133056642, + "step": 190520 + }, + { + "epoch": 0.5186878998230571, + "grad_norm": 2.187485933303833, + "learning_rate": 9.63941665530871e-06, + "loss": 2.482345962524414, + "step": 190540 + }, + { + "epoch": 0.5187423438138016, + "grad_norm": 2.358560085296631, + "learning_rate": 9.638326291399756e-06, + "loss": 2.470402717590332, + "step": 190560 + }, + { + "epoch": 0.5187967878045461, + "grad_norm": 2.326902151107788, + "learning_rate": 9.637235927490802e-06, + "loss": 2.47869758605957, + "step": 190580 + }, + { + "epoch": 0.5188512317952906, + "grad_norm": 2.301785707473755, + "learning_rate": 9.636145563581846e-06, + "loss": 2.473514938354492, + "step": 190600 + }, + { + "epoch": 0.5189056757860351, + "grad_norm": 2.378739356994629, + "learning_rate": 9.635055199672892e-06, + "loss": 2.476267433166504, + "step": 190620 + }, + { + "epoch": 0.5189601197767796, + "grad_norm": 2.0760881900787354, + "learning_rate": 9.633964835763937e-06, + "loss": 2.4659143447875977, + "step": 190640 + }, + { + "epoch": 0.5190145637675242, + "grad_norm": 12.098588943481445, + "learning_rate": 9.632874471854983e-06, + "loss": 2.4711071014404298, + "step": 190660 + }, + { + "epoch": 0.5190690077582687, + "grad_norm": 2.158700942993164, + "learning_rate": 9.631784107946027e-06, + "loss": 2.459160804748535, + "step": 190680 + }, + { + "epoch": 0.5191234517490132, + "grad_norm": 2.287768840789795, + "learning_rate": 9.630693744037073e-06, + "loss": 2.4903127670288088, + "step": 190700 + }, + { + "epoch": 0.5191778957397577, + "grad_norm": 8.194973945617676, + "learning_rate": 9.629603380128119e-06, + "loss": 2.4805503845214845, + "step": 190720 + }, + { + "epoch": 0.5192323397305022, + "grad_norm": 2.203489303588867, + "learning_rate": 9.628513016219165e-06, + "loss": 2.4666984558105467, + "step": 190740 + }, + { + "epoch": 0.5192867837212468, + "grad_norm": 2.0778279304504395, + "learning_rate": 9.627422652310209e-06, + "loss": 2.479121780395508, + "step": 190760 + }, + { + "epoch": 0.5193412277119913, + "grad_norm": 2.247143030166626, + "learning_rate": 9.626332288401255e-06, + "loss": 2.4815389633178713, + "step": 190780 + }, + { + "epoch": 0.5193956717027358, + "grad_norm": 2.1243371963500977, + "learning_rate": 9.6252419244923e-06, + "loss": 2.497203826904297, + "step": 190800 + }, + { + "epoch": 0.5194501156934803, + "grad_norm": 2.1815028190612793, + "learning_rate": 9.624151560583346e-06, + "loss": 2.4704025268554686, + "step": 190820 + }, + { + "epoch": 0.5195045596842248, + "grad_norm": 2.2540197372436523, + "learning_rate": 9.62306119667439e-06, + "loss": 2.457703971862793, + "step": 190840 + }, + { + "epoch": 0.5195590036749693, + "grad_norm": 2.2019340991973877, + "learning_rate": 9.621970832765436e-06, + "loss": 2.477819633483887, + "step": 190860 + }, + { + "epoch": 0.5196134476657139, + "grad_norm": 2.1222407817840576, + "learning_rate": 9.620880468856482e-06, + "loss": 2.4936506271362306, + "step": 190880 + }, + { + "epoch": 0.5196678916564584, + "grad_norm": 2.128096103668213, + "learning_rate": 9.619790104947528e-06, + "loss": 2.4938125610351562, + "step": 190900 + }, + { + "epoch": 0.5197223356472029, + "grad_norm": 2.3888702392578125, + "learning_rate": 9.618699741038574e-06, + "loss": 2.4595640182495115, + "step": 190920 + }, + { + "epoch": 0.5197767796379474, + "grad_norm": 3.583671808242798, + "learning_rate": 9.617609377129618e-06, + "loss": 2.5181325912475585, + "step": 190940 + }, + { + "epoch": 0.5198312236286919, + "grad_norm": 2.136054754257202, + "learning_rate": 9.616519013220664e-06, + "loss": 2.4732854843139647, + "step": 190960 + }, + { + "epoch": 0.5198856676194366, + "grad_norm": 2.2552130222320557, + "learning_rate": 9.61542864931171e-06, + "loss": 2.4651123046875, + "step": 190980 + }, + { + "epoch": 0.519940111610181, + "grad_norm": 2.246077299118042, + "learning_rate": 9.614338285402755e-06, + "loss": 2.5074539184570312, + "step": 191000 + }, + { + "epoch": 0.5199945556009256, + "grad_norm": 2.298464775085449, + "learning_rate": 9.6132479214938e-06, + "loss": 2.4426855087280273, + "step": 191020 + }, + { + "epoch": 0.5200489995916701, + "grad_norm": 2.1552233695983887, + "learning_rate": 9.612157557584845e-06, + "loss": 2.4973112106323243, + "step": 191040 + }, + { + "epoch": 0.5201034435824146, + "grad_norm": 2.2127525806427, + "learning_rate": 9.611067193675891e-06, + "loss": 2.466603660583496, + "step": 191060 + }, + { + "epoch": 0.5201578875731591, + "grad_norm": 2.3006091117858887, + "learning_rate": 9.609976829766937e-06, + "loss": 2.505802536010742, + "step": 191080 + }, + { + "epoch": 0.5202123315639037, + "grad_norm": 2.282984972000122, + "learning_rate": 9.60888646585798e-06, + "loss": 2.476399230957031, + "step": 191100 + }, + { + "epoch": 0.5202667755546482, + "grad_norm": 2.1582257747650146, + "learning_rate": 9.607796101949027e-06, + "loss": 2.4704723358154297, + "step": 191120 + }, + { + "epoch": 0.5203212195453927, + "grad_norm": 2.2487833499908447, + "learning_rate": 9.606705738040072e-06, + "loss": 2.4732046127319336, + "step": 191140 + }, + { + "epoch": 0.5203756635361372, + "grad_norm": 2.161975383758545, + "learning_rate": 9.605615374131118e-06, + "loss": 2.4814807891845705, + "step": 191160 + }, + { + "epoch": 0.5204301075268817, + "grad_norm": 2.156928062438965, + "learning_rate": 9.604525010222162e-06, + "loss": 2.485692024230957, + "step": 191180 + }, + { + "epoch": 0.5204845515176263, + "grad_norm": 2.153409242630005, + "learning_rate": 9.603434646313208e-06, + "loss": 2.473654365539551, + "step": 191200 + }, + { + "epoch": 0.5205389955083708, + "grad_norm": 2.2692043781280518, + "learning_rate": 9.602344282404254e-06, + "loss": 2.4593276977539062, + "step": 191220 + }, + { + "epoch": 0.5205934394991153, + "grad_norm": 2.1146655082702637, + "learning_rate": 9.6012539184953e-06, + "loss": 2.4837068557739257, + "step": 191240 + }, + { + "epoch": 0.5206478834898598, + "grad_norm": 2.025911569595337, + "learning_rate": 9.600163554586344e-06, + "loss": 2.4687734603881837, + "step": 191260 + }, + { + "epoch": 0.5207023274806043, + "grad_norm": 2.2783491611480713, + "learning_rate": 9.59907319067739e-06, + "loss": 2.4843482971191406, + "step": 191280 + }, + { + "epoch": 0.5207567714713488, + "grad_norm": 2.1114444732666016, + "learning_rate": 9.597982826768435e-06, + "loss": 2.48815975189209, + "step": 191300 + }, + { + "epoch": 0.5208112154620934, + "grad_norm": 2.106044054031372, + "learning_rate": 9.596892462859481e-06, + "loss": 2.4751426696777346, + "step": 191320 + }, + { + "epoch": 0.5208656594528379, + "grad_norm": 2.1410584449768066, + "learning_rate": 9.595802098950525e-06, + "loss": 2.4835058212280274, + "step": 191340 + }, + { + "epoch": 0.5209201034435824, + "grad_norm": 2.2101662158966064, + "learning_rate": 9.594711735041571e-06, + "loss": 2.5140146255493163, + "step": 191360 + }, + { + "epoch": 0.5209745474343269, + "grad_norm": 2.2735135555267334, + "learning_rate": 9.593621371132617e-06, + "loss": 2.4595312118530273, + "step": 191380 + }, + { + "epoch": 0.5210289914250714, + "grad_norm": 2.2350263595581055, + "learning_rate": 9.592531007223663e-06, + "loss": 2.5205427169799806, + "step": 191400 + }, + { + "epoch": 0.521083435415816, + "grad_norm": 2.445981025695801, + "learning_rate": 9.591440643314707e-06, + "loss": 2.467093086242676, + "step": 191420 + }, + { + "epoch": 0.5211378794065605, + "grad_norm": 2.2261362075805664, + "learning_rate": 9.590350279405753e-06, + "loss": 2.4719215393066407, + "step": 191440 + }, + { + "epoch": 0.521192323397305, + "grad_norm": 2.3254852294921875, + "learning_rate": 9.589259915496799e-06, + "loss": 2.474934387207031, + "step": 191460 + }, + { + "epoch": 0.5212467673880495, + "grad_norm": 2.095316171646118, + "learning_rate": 9.588169551587844e-06, + "loss": 2.471324348449707, + "step": 191480 + }, + { + "epoch": 0.521301211378794, + "grad_norm": 2.0885021686553955, + "learning_rate": 9.587079187678889e-06, + "loss": 2.496341514587402, + "step": 191500 + }, + { + "epoch": 0.5213556553695385, + "grad_norm": 2.3394622802734375, + "learning_rate": 9.585988823769934e-06, + "loss": 2.454458999633789, + "step": 191520 + }, + { + "epoch": 0.5214100993602832, + "grad_norm": 2.205836534500122, + "learning_rate": 9.58489845986098e-06, + "loss": 2.506218147277832, + "step": 191540 + }, + { + "epoch": 0.5214645433510277, + "grad_norm": 2.0546681880950928, + "learning_rate": 9.583808095952026e-06, + "loss": 2.48995361328125, + "step": 191560 + }, + { + "epoch": 0.5215189873417722, + "grad_norm": 2.1459851264953613, + "learning_rate": 9.58271773204307e-06, + "loss": 2.4681577682495117, + "step": 191580 + }, + { + "epoch": 0.5215734313325167, + "grad_norm": 2.3478002548217773, + "learning_rate": 9.581627368134116e-06, + "loss": 2.4731233596801756, + "step": 191600 + }, + { + "epoch": 0.5216278753232612, + "grad_norm": 2.14660906791687, + "learning_rate": 9.580537004225162e-06, + "loss": 2.4811494827270506, + "step": 191620 + }, + { + "epoch": 0.5216823193140058, + "grad_norm": 2.108900547027588, + "learning_rate": 9.579446640316207e-06, + "loss": 2.490035629272461, + "step": 191640 + }, + { + "epoch": 0.5217367633047503, + "grad_norm": 2.1107430458068848, + "learning_rate": 9.578356276407252e-06, + "loss": 2.492254447937012, + "step": 191660 + }, + { + "epoch": 0.5217912072954948, + "grad_norm": 2.104482889175415, + "learning_rate": 9.577265912498297e-06, + "loss": 2.5192089080810547, + "step": 191680 + }, + { + "epoch": 0.5218456512862393, + "grad_norm": 2.2279770374298096, + "learning_rate": 9.576175548589343e-06, + "loss": 2.4957931518554686, + "step": 191700 + }, + { + "epoch": 0.5219000952769838, + "grad_norm": 2.1567258834838867, + "learning_rate": 9.575085184680389e-06, + "loss": 2.47824649810791, + "step": 191720 + }, + { + "epoch": 0.5219545392677283, + "grad_norm": 2.190924882888794, + "learning_rate": 9.573994820771433e-06, + "loss": 2.4916494369506834, + "step": 191740 + }, + { + "epoch": 0.5220089832584729, + "grad_norm": 2.136465311050415, + "learning_rate": 9.572904456862479e-06, + "loss": 2.5058000564575194, + "step": 191760 + }, + { + "epoch": 0.5220634272492174, + "grad_norm": 2.190746784210205, + "learning_rate": 9.571814092953525e-06, + "loss": 2.480027770996094, + "step": 191780 + }, + { + "epoch": 0.5221178712399619, + "grad_norm": 2.2433054447174072, + "learning_rate": 9.57072372904457e-06, + "loss": 2.4717483520507812, + "step": 191800 + }, + { + "epoch": 0.5221723152307064, + "grad_norm": 2.081953763961792, + "learning_rate": 9.569633365135615e-06, + "loss": 2.466839408874512, + "step": 191820 + }, + { + "epoch": 0.5222267592214509, + "grad_norm": 2.3072285652160645, + "learning_rate": 9.56854300122666e-06, + "loss": 2.4761993408203127, + "step": 191840 + }, + { + "epoch": 0.5222812032121954, + "grad_norm": 2.1529510021209717, + "learning_rate": 9.567452637317706e-06, + "loss": 2.4604711532592773, + "step": 191860 + }, + { + "epoch": 0.52233564720294, + "grad_norm": 2.114575147628784, + "learning_rate": 9.566362273408752e-06, + "loss": 2.4704198837280273, + "step": 191880 + }, + { + "epoch": 0.5223900911936845, + "grad_norm": 2.146742105484009, + "learning_rate": 9.565271909499796e-06, + "loss": 2.460545539855957, + "step": 191900 + }, + { + "epoch": 0.522444535184429, + "grad_norm": 2.224424362182617, + "learning_rate": 9.564181545590842e-06, + "loss": 2.455318069458008, + "step": 191920 + }, + { + "epoch": 0.5224989791751735, + "grad_norm": 2.113945484161377, + "learning_rate": 9.563091181681888e-06, + "loss": 2.5005071640014647, + "step": 191940 + }, + { + "epoch": 0.522553423165918, + "grad_norm": 2.50825834274292, + "learning_rate": 9.562000817772934e-06, + "loss": 2.4618316650390626, + "step": 191960 + }, + { + "epoch": 0.5226078671566626, + "grad_norm": 2.0771703720092773, + "learning_rate": 9.560910453863978e-06, + "loss": 2.485141372680664, + "step": 191980 + }, + { + "epoch": 0.5226623111474071, + "grad_norm": 2.1918797492980957, + "learning_rate": 9.559820089955024e-06, + "loss": 2.482059097290039, + "step": 192000 + }, + { + "epoch": 0.5227167551381516, + "grad_norm": 2.3125030994415283, + "learning_rate": 9.55872972604607e-06, + "loss": 2.493561935424805, + "step": 192020 + }, + { + "epoch": 0.5227711991288961, + "grad_norm": 2.0893688201904297, + "learning_rate": 9.557639362137115e-06, + "loss": 2.499803161621094, + "step": 192040 + }, + { + "epoch": 0.5228256431196406, + "grad_norm": 2.1170856952667236, + "learning_rate": 9.55654899822816e-06, + "loss": 2.4849222183227537, + "step": 192060 + }, + { + "epoch": 0.5228800871103851, + "grad_norm": 2.4148173332214355, + "learning_rate": 9.555458634319205e-06, + "loss": 2.456159973144531, + "step": 192080 + }, + { + "epoch": 0.5229345311011298, + "grad_norm": 2.0390501022338867, + "learning_rate": 9.554368270410251e-06, + "loss": 2.5200431823730467, + "step": 192100 + }, + { + "epoch": 0.5229889750918743, + "grad_norm": 2.242577314376831, + "learning_rate": 9.553277906501297e-06, + "loss": 2.4492204666137694, + "step": 192120 + }, + { + "epoch": 0.5230434190826188, + "grad_norm": 2.178363800048828, + "learning_rate": 9.552187542592341e-06, + "loss": 2.474406051635742, + "step": 192140 + }, + { + "epoch": 0.5230978630733633, + "grad_norm": 2.16919207572937, + "learning_rate": 9.551097178683387e-06, + "loss": 2.4671865463256837, + "step": 192160 + }, + { + "epoch": 0.5231523070641078, + "grad_norm": 2.1828460693359375, + "learning_rate": 9.550006814774432e-06, + "loss": 2.4626581192016603, + "step": 192180 + }, + { + "epoch": 0.5232067510548524, + "grad_norm": 2.1924095153808594, + "learning_rate": 9.548916450865478e-06, + "loss": 2.430710220336914, + "step": 192200 + }, + { + "epoch": 0.5232611950455969, + "grad_norm": 2.1828324794769287, + "learning_rate": 9.547826086956522e-06, + "loss": 2.477616882324219, + "step": 192220 + }, + { + "epoch": 0.5233156390363414, + "grad_norm": 2.2124252319335938, + "learning_rate": 9.546735723047568e-06, + "loss": 2.49896297454834, + "step": 192240 + }, + { + "epoch": 0.5233700830270859, + "grad_norm": 2.0754384994506836, + "learning_rate": 9.545645359138614e-06, + "loss": 2.452711296081543, + "step": 192260 + }, + { + "epoch": 0.5234245270178304, + "grad_norm": 2.225569248199463, + "learning_rate": 9.54455499522966e-06, + "loss": 2.4907787322998045, + "step": 192280 + }, + { + "epoch": 0.5234789710085749, + "grad_norm": 2.1378257274627686, + "learning_rate": 9.543464631320704e-06, + "loss": 2.4924589157104493, + "step": 192300 + }, + { + "epoch": 0.5235334149993195, + "grad_norm": 2.0955350399017334, + "learning_rate": 9.54237426741175e-06, + "loss": 2.483642578125, + "step": 192320 + }, + { + "epoch": 0.523587858990064, + "grad_norm": 2.389411687850952, + "learning_rate": 9.541283903502796e-06, + "loss": 2.4711139678955076, + "step": 192340 + }, + { + "epoch": 0.5236423029808085, + "grad_norm": 2.1206114292144775, + "learning_rate": 9.540193539593841e-06, + "loss": 2.490650177001953, + "step": 192360 + }, + { + "epoch": 0.523696746971553, + "grad_norm": 2.0679571628570557, + "learning_rate": 9.539103175684885e-06, + "loss": 2.483095169067383, + "step": 192380 + }, + { + "epoch": 0.5237511909622975, + "grad_norm": 2.324681282043457, + "learning_rate": 9.538012811775931e-06, + "loss": 2.480465888977051, + "step": 192400 + }, + { + "epoch": 0.5238056349530421, + "grad_norm": 2.3794326782226562, + "learning_rate": 9.536922447866977e-06, + "loss": 2.4768245697021483, + "step": 192420 + }, + { + "epoch": 0.5238600789437866, + "grad_norm": 2.088343620300293, + "learning_rate": 9.535832083958023e-06, + "loss": 2.44595947265625, + "step": 192440 + }, + { + "epoch": 0.5239145229345311, + "grad_norm": 2.191488742828369, + "learning_rate": 9.534741720049067e-06, + "loss": 2.53302001953125, + "step": 192460 + }, + { + "epoch": 0.5239689669252756, + "grad_norm": 2.4157304763793945, + "learning_rate": 9.533651356140113e-06, + "loss": 2.4715606689453127, + "step": 192480 + }, + { + "epoch": 0.5240234109160201, + "grad_norm": 2.225168466567993, + "learning_rate": 9.532560992231159e-06, + "loss": 2.4867580413818358, + "step": 192500 + }, + { + "epoch": 0.5240778549067646, + "grad_norm": 2.3957934379577637, + "learning_rate": 9.531470628322203e-06, + "loss": 2.472583770751953, + "step": 192520 + }, + { + "epoch": 0.5241322988975092, + "grad_norm": 2.224583625793457, + "learning_rate": 9.530380264413249e-06, + "loss": 2.445479393005371, + "step": 192540 + }, + { + "epoch": 0.5241867428882537, + "grad_norm": 2.166316270828247, + "learning_rate": 9.529289900504294e-06, + "loss": 2.497662353515625, + "step": 192560 + }, + { + "epoch": 0.5242411868789982, + "grad_norm": 2.0950849056243896, + "learning_rate": 9.52819953659534e-06, + "loss": 2.512499237060547, + "step": 192580 + }, + { + "epoch": 0.5242956308697427, + "grad_norm": 2.1860997676849365, + "learning_rate": 9.527109172686384e-06, + "loss": 2.4478532791137697, + "step": 192600 + }, + { + "epoch": 0.5243500748604872, + "grad_norm": 2.232146978378296, + "learning_rate": 9.52601880877743e-06, + "loss": 2.513784980773926, + "step": 192620 + }, + { + "epoch": 0.5244045188512318, + "grad_norm": 2.688563823699951, + "learning_rate": 9.524928444868476e-06, + "loss": 2.477700042724609, + "step": 192640 + }, + { + "epoch": 0.5244589628419764, + "grad_norm": 2.1172502040863037, + "learning_rate": 9.523838080959522e-06, + "loss": 2.4811986923217773, + "step": 192660 + }, + { + "epoch": 0.5245134068327209, + "grad_norm": 2.210655450820923, + "learning_rate": 9.522747717050566e-06, + "loss": 2.4899024963378906, + "step": 192680 + }, + { + "epoch": 0.5245678508234654, + "grad_norm": 2.1751036643981934, + "learning_rate": 9.521657353141612e-06, + "loss": 2.4843597412109375, + "step": 192700 + }, + { + "epoch": 0.5246222948142099, + "grad_norm": 2.2795052528381348, + "learning_rate": 9.520566989232657e-06, + "loss": 2.4568292617797853, + "step": 192720 + }, + { + "epoch": 0.5246767388049544, + "grad_norm": 2.0856926441192627, + "learning_rate": 9.519476625323703e-06, + "loss": 2.453089141845703, + "step": 192740 + }, + { + "epoch": 0.524731182795699, + "grad_norm": 2.2092339992523193, + "learning_rate": 9.518386261414747e-06, + "loss": 2.457438278198242, + "step": 192760 + }, + { + "epoch": 0.5247856267864435, + "grad_norm": 2.372326374053955, + "learning_rate": 9.517295897505793e-06, + "loss": 2.5036115646362305, + "step": 192780 + }, + { + "epoch": 0.524840070777188, + "grad_norm": 2.203890323638916, + "learning_rate": 9.516205533596839e-06, + "loss": 2.5235090255737305, + "step": 192800 + }, + { + "epoch": 0.5248945147679325, + "grad_norm": 2.2030820846557617, + "learning_rate": 9.515115169687885e-06, + "loss": 2.47458438873291, + "step": 192820 + }, + { + "epoch": 0.524948958758677, + "grad_norm": 2.210662603378296, + "learning_rate": 9.514024805778929e-06, + "loss": 2.47315673828125, + "step": 192840 + }, + { + "epoch": 0.5250034027494215, + "grad_norm": 2.1291086673736572, + "learning_rate": 9.512934441869975e-06, + "loss": 2.4907758712768553, + "step": 192860 + }, + { + "epoch": 0.5250578467401661, + "grad_norm": 2.104398250579834, + "learning_rate": 9.51184407796102e-06, + "loss": 2.4836658477783202, + "step": 192880 + }, + { + "epoch": 0.5251122907309106, + "grad_norm": 2.77946138381958, + "learning_rate": 9.510753714052066e-06, + "loss": 2.4942426681518555, + "step": 192900 + }, + { + "epoch": 0.5251667347216551, + "grad_norm": 2.1459298133850098, + "learning_rate": 9.50966335014311e-06, + "loss": 2.5176010131835938, + "step": 192920 + }, + { + "epoch": 0.5252211787123996, + "grad_norm": 2.1750681400299072, + "learning_rate": 9.508572986234156e-06, + "loss": 2.4870512008666994, + "step": 192940 + }, + { + "epoch": 0.5252756227031441, + "grad_norm": 2.2379367351531982, + "learning_rate": 9.507482622325202e-06, + "loss": 2.489158821105957, + "step": 192960 + }, + { + "epoch": 0.5253300666938887, + "grad_norm": 2.199005603790283, + "learning_rate": 9.506392258416248e-06, + "loss": 2.5123823165893553, + "step": 192980 + }, + { + "epoch": 0.5253845106846332, + "grad_norm": 2.1813461780548096, + "learning_rate": 9.505301894507292e-06, + "loss": 2.477034568786621, + "step": 193000 + }, + { + "epoch": 0.5254389546753777, + "grad_norm": 2.1962685585021973, + "learning_rate": 9.504211530598338e-06, + "loss": 2.4711416244506834, + "step": 193020 + }, + { + "epoch": 0.5254933986661222, + "grad_norm": 2.1894478797912598, + "learning_rate": 9.503121166689384e-06, + "loss": 2.4784473419189452, + "step": 193040 + }, + { + "epoch": 0.5255478426568667, + "grad_norm": 2.0397565364837646, + "learning_rate": 9.50203080278043e-06, + "loss": 2.4913309097290037, + "step": 193060 + }, + { + "epoch": 0.5256022866476112, + "grad_norm": 2.2116200923919678, + "learning_rate": 9.500940438871474e-06, + "loss": 2.4668670654296876, + "step": 193080 + }, + { + "epoch": 0.5256567306383558, + "grad_norm": 2.2867636680603027, + "learning_rate": 9.49985007496252e-06, + "loss": 2.465729904174805, + "step": 193100 + }, + { + "epoch": 0.5257111746291003, + "grad_norm": 3.2343075275421143, + "learning_rate": 9.498759711053565e-06, + "loss": 2.472027587890625, + "step": 193120 + }, + { + "epoch": 0.5257656186198448, + "grad_norm": 2.4736082553863525, + "learning_rate": 9.497669347144611e-06, + "loss": 2.4836183547973634, + "step": 193140 + }, + { + "epoch": 0.5258200626105893, + "grad_norm": 2.071770191192627, + "learning_rate": 9.496578983235655e-06, + "loss": 2.455097198486328, + "step": 193160 + }, + { + "epoch": 0.5258745066013338, + "grad_norm": 2.6437132358551025, + "learning_rate": 9.495488619326701e-06, + "loss": 2.4654460906982423, + "step": 193180 + }, + { + "epoch": 0.5259289505920784, + "grad_norm": 2.2023370265960693, + "learning_rate": 9.494398255417747e-06, + "loss": 2.4955793380737306, + "step": 193200 + }, + { + "epoch": 0.525983394582823, + "grad_norm": 2.2689504623413086, + "learning_rate": 9.493307891508793e-06, + "loss": 2.4894309997558595, + "step": 193220 + }, + { + "epoch": 0.5260378385735675, + "grad_norm": 2.1379473209381104, + "learning_rate": 9.492217527599837e-06, + "loss": 2.4803773880004885, + "step": 193240 + }, + { + "epoch": 0.526092282564312, + "grad_norm": 2.244691848754883, + "learning_rate": 9.491127163690882e-06, + "loss": 2.473739242553711, + "step": 193260 + }, + { + "epoch": 0.5261467265550565, + "grad_norm": 2.170013904571533, + "learning_rate": 9.490036799781928e-06, + "loss": 2.4963539123535154, + "step": 193280 + }, + { + "epoch": 0.526201170545801, + "grad_norm": 2.2768282890319824, + "learning_rate": 9.488946435872974e-06, + "loss": 2.4796070098876952, + "step": 193300 + }, + { + "epoch": 0.5262556145365456, + "grad_norm": 2.142001152038574, + "learning_rate": 9.487856071964018e-06, + "loss": 2.467072105407715, + "step": 193320 + }, + { + "epoch": 0.5263100585272901, + "grad_norm": 2.1727397441864014, + "learning_rate": 9.486765708055064e-06, + "loss": 2.478495979309082, + "step": 193340 + }, + { + "epoch": 0.5263645025180346, + "grad_norm": 2.266134023666382, + "learning_rate": 9.48567534414611e-06, + "loss": 2.4869094848632813, + "step": 193360 + }, + { + "epoch": 0.5264189465087791, + "grad_norm": 2.136664628982544, + "learning_rate": 9.484584980237156e-06, + "loss": 2.504175567626953, + "step": 193380 + }, + { + "epoch": 0.5264733904995236, + "grad_norm": 2.141667366027832, + "learning_rate": 9.4834946163282e-06, + "loss": 2.484453010559082, + "step": 193400 + }, + { + "epoch": 0.5265278344902682, + "grad_norm": 2.1159956455230713, + "learning_rate": 9.482404252419246e-06, + "loss": 2.488303375244141, + "step": 193420 + }, + { + "epoch": 0.5265822784810127, + "grad_norm": 2.1212105751037598, + "learning_rate": 9.481313888510291e-06, + "loss": 2.47934627532959, + "step": 193440 + }, + { + "epoch": 0.5266367224717572, + "grad_norm": 2.2420477867126465, + "learning_rate": 9.480223524601337e-06, + "loss": 2.4754180908203125, + "step": 193460 + }, + { + "epoch": 0.5266911664625017, + "grad_norm": 2.1984071731567383, + "learning_rate": 9.479133160692381e-06, + "loss": 2.4450490951538084, + "step": 193480 + }, + { + "epoch": 0.5267456104532462, + "grad_norm": 2.290303945541382, + "learning_rate": 9.478042796783427e-06, + "loss": 2.489297866821289, + "step": 193500 + }, + { + "epoch": 0.5268000544439907, + "grad_norm": 2.297577142715454, + "learning_rate": 9.476952432874473e-06, + "loss": 2.4800542831420898, + "step": 193520 + }, + { + "epoch": 0.5268544984347353, + "grad_norm": 2.3641772270202637, + "learning_rate": 9.475862068965519e-06, + "loss": 2.4693359375, + "step": 193540 + }, + { + "epoch": 0.5269089424254798, + "grad_norm": 2.1001627445220947, + "learning_rate": 9.474771705056563e-06, + "loss": 2.500955581665039, + "step": 193560 + }, + { + "epoch": 0.5269633864162243, + "grad_norm": 2.1456806659698486, + "learning_rate": 9.473681341147609e-06, + "loss": 2.455274963378906, + "step": 193580 + }, + { + "epoch": 0.5270178304069688, + "grad_norm": 2.2091784477233887, + "learning_rate": 9.472590977238654e-06, + "loss": 2.4744796752929688, + "step": 193600 + }, + { + "epoch": 0.5270722743977133, + "grad_norm": 2.2033700942993164, + "learning_rate": 9.471500613329699e-06, + "loss": 2.480164337158203, + "step": 193620 + }, + { + "epoch": 0.5271267183884579, + "grad_norm": 2.242706537246704, + "learning_rate": 9.470410249420744e-06, + "loss": 2.481049919128418, + "step": 193640 + }, + { + "epoch": 0.5271811623792024, + "grad_norm": 2.1900694370269775, + "learning_rate": 9.46931988551179e-06, + "loss": 2.4867120742797852, + "step": 193660 + }, + { + "epoch": 0.5272356063699469, + "grad_norm": 2.1915194988250732, + "learning_rate": 9.468229521602836e-06, + "loss": 2.490436553955078, + "step": 193680 + }, + { + "epoch": 0.5272900503606914, + "grad_norm": 2.108947992324829, + "learning_rate": 9.46713915769388e-06, + "loss": 2.477804756164551, + "step": 193700 + }, + { + "epoch": 0.5273444943514359, + "grad_norm": 2.1111702919006348, + "learning_rate": 9.466048793784926e-06, + "loss": 2.430927276611328, + "step": 193720 + }, + { + "epoch": 0.5273989383421804, + "grad_norm": 2.155345916748047, + "learning_rate": 9.464958429875972e-06, + "loss": 2.5110752105712892, + "step": 193740 + }, + { + "epoch": 0.527453382332925, + "grad_norm": 2.4688961505889893, + "learning_rate": 9.463868065967018e-06, + "loss": 2.4752641677856446, + "step": 193760 + }, + { + "epoch": 0.5275078263236695, + "grad_norm": 2.45413875579834, + "learning_rate": 9.462777702058062e-06, + "loss": 2.4736164093017576, + "step": 193780 + }, + { + "epoch": 0.527562270314414, + "grad_norm": 2.0979690551757812, + "learning_rate": 9.461687338149107e-06, + "loss": 2.471671485900879, + "step": 193800 + }, + { + "epoch": 0.5276167143051586, + "grad_norm": 2.130286693572998, + "learning_rate": 9.460596974240153e-06, + "loss": 2.482736015319824, + "step": 193820 + }, + { + "epoch": 0.527671158295903, + "grad_norm": 2.1722538471221924, + "learning_rate": 9.459506610331199e-06, + "loss": 2.472222900390625, + "step": 193840 + }, + { + "epoch": 0.5277256022866477, + "grad_norm": 2.199930429458618, + "learning_rate": 9.458416246422243e-06, + "loss": 2.4583599090576174, + "step": 193860 + }, + { + "epoch": 0.5277800462773922, + "grad_norm": 2.1522057056427, + "learning_rate": 9.457325882513289e-06, + "loss": 2.456828498840332, + "step": 193880 + }, + { + "epoch": 0.5278344902681367, + "grad_norm": 2.141223907470703, + "learning_rate": 9.456235518604335e-06, + "loss": 2.47088737487793, + "step": 193900 + }, + { + "epoch": 0.5278889342588812, + "grad_norm": 2.0688817501068115, + "learning_rate": 9.45514515469538e-06, + "loss": 2.4375001907348635, + "step": 193920 + }, + { + "epoch": 0.5279433782496257, + "grad_norm": 2.0868539810180664, + "learning_rate": 9.454054790786425e-06, + "loss": 2.484823226928711, + "step": 193940 + }, + { + "epoch": 0.5279978222403702, + "grad_norm": 2.140256643295288, + "learning_rate": 9.45296442687747e-06, + "loss": 2.4594980239868165, + "step": 193960 + }, + { + "epoch": 0.5280522662311148, + "grad_norm": 2.104135513305664, + "learning_rate": 9.451874062968516e-06, + "loss": 2.4605390548706056, + "step": 193980 + }, + { + "epoch": 0.5281067102218593, + "grad_norm": 2.1703133583068848, + "learning_rate": 9.450783699059562e-06, + "loss": 2.4653480529785154, + "step": 194000 + }, + { + "epoch": 0.5281611542126038, + "grad_norm": 2.1816985607147217, + "learning_rate": 9.449693335150606e-06, + "loss": 2.476837921142578, + "step": 194020 + }, + { + "epoch": 0.5282155982033483, + "grad_norm": 2.268599510192871, + "learning_rate": 9.448602971241652e-06, + "loss": 2.4755420684814453, + "step": 194040 + }, + { + "epoch": 0.5282700421940928, + "grad_norm": 2.1837754249572754, + "learning_rate": 9.447512607332698e-06, + "loss": 2.488779830932617, + "step": 194060 + }, + { + "epoch": 0.5283244861848373, + "grad_norm": 2.3815362453460693, + "learning_rate": 9.446422243423744e-06, + "loss": 2.4798868179321287, + "step": 194080 + }, + { + "epoch": 0.5283789301755819, + "grad_norm": 2.3534095287323, + "learning_rate": 9.445331879514788e-06, + "loss": 2.4935176849365233, + "step": 194100 + }, + { + "epoch": 0.5284333741663264, + "grad_norm": 2.246838092803955, + "learning_rate": 9.444241515605834e-06, + "loss": 2.454924201965332, + "step": 194120 + }, + { + "epoch": 0.5284878181570709, + "grad_norm": 2.0594046115875244, + "learning_rate": 9.44315115169688e-06, + "loss": 2.507088279724121, + "step": 194140 + }, + { + "epoch": 0.5285422621478154, + "grad_norm": 2.1625235080718994, + "learning_rate": 9.442060787787925e-06, + "loss": 2.4634145736694335, + "step": 194160 + }, + { + "epoch": 0.5285967061385599, + "grad_norm": 2.1089680194854736, + "learning_rate": 9.44097042387897e-06, + "loss": 2.4548242568969725, + "step": 194180 + }, + { + "epoch": 0.5286511501293045, + "grad_norm": 2.1982269287109375, + "learning_rate": 9.439880059970015e-06, + "loss": 2.4513145446777345, + "step": 194200 + }, + { + "epoch": 0.528705594120049, + "grad_norm": 2.0824899673461914, + "learning_rate": 9.438789696061061e-06, + "loss": 2.4910961151123048, + "step": 194220 + }, + { + "epoch": 0.5287600381107935, + "grad_norm": 2.1906495094299316, + "learning_rate": 9.437699332152107e-06, + "loss": 2.4968809127807616, + "step": 194240 + }, + { + "epoch": 0.528814482101538, + "grad_norm": 2.165966749191284, + "learning_rate": 9.436608968243151e-06, + "loss": 2.478463554382324, + "step": 194260 + }, + { + "epoch": 0.5288689260922825, + "grad_norm": 2.166231632232666, + "learning_rate": 9.435518604334197e-06, + "loss": 2.4820171356201173, + "step": 194280 + }, + { + "epoch": 0.528923370083027, + "grad_norm": 2.245604991912842, + "learning_rate": 9.434428240425243e-06, + "loss": 2.4978401184082033, + "step": 194300 + }, + { + "epoch": 0.5289778140737716, + "grad_norm": 2.1186625957489014, + "learning_rate": 9.433337876516288e-06, + "loss": 2.5016910552978517, + "step": 194320 + }, + { + "epoch": 0.5290322580645161, + "grad_norm": 2.2252423763275146, + "learning_rate": 9.432247512607332e-06, + "loss": 2.482887840270996, + "step": 194340 + }, + { + "epoch": 0.5290867020552606, + "grad_norm": 2.2328298091888428, + "learning_rate": 9.431157148698378e-06, + "loss": 2.499444770812988, + "step": 194360 + }, + { + "epoch": 0.5291411460460052, + "grad_norm": 2.132962465286255, + "learning_rate": 9.430066784789424e-06, + "loss": 2.517884063720703, + "step": 194380 + }, + { + "epoch": 0.5291955900367497, + "grad_norm": 2.0504345893859863, + "learning_rate": 9.42897642088047e-06, + "loss": 2.4724655151367188, + "step": 194400 + }, + { + "epoch": 0.5292500340274943, + "grad_norm": 2.1416094303131104, + "learning_rate": 9.427886056971514e-06, + "loss": 2.4818429946899414, + "step": 194420 + }, + { + "epoch": 0.5293044780182388, + "grad_norm": 2.0950522422790527, + "learning_rate": 9.42679569306256e-06, + "loss": 2.5027738571166993, + "step": 194440 + }, + { + "epoch": 0.5293589220089833, + "grad_norm": 2.2422211170196533, + "learning_rate": 9.425705329153606e-06, + "loss": 2.497617721557617, + "step": 194460 + }, + { + "epoch": 0.5294133659997278, + "grad_norm": 2.1876511573791504, + "learning_rate": 9.424614965244651e-06, + "loss": 2.505925750732422, + "step": 194480 + }, + { + "epoch": 0.5294678099904723, + "grad_norm": 2.444763422012329, + "learning_rate": 9.423524601335696e-06, + "loss": 2.486105537414551, + "step": 194500 + }, + { + "epoch": 0.5295222539812168, + "grad_norm": 2.197006940841675, + "learning_rate": 9.422434237426741e-06, + "loss": 2.486981010437012, + "step": 194520 + }, + { + "epoch": 0.5295766979719614, + "grad_norm": 6.318744659423828, + "learning_rate": 9.421343873517787e-06, + "loss": 2.4957534790039064, + "step": 194540 + }, + { + "epoch": 0.5296311419627059, + "grad_norm": 2.3212530612945557, + "learning_rate": 9.420253509608833e-06, + "loss": 2.485183334350586, + "step": 194560 + }, + { + "epoch": 0.5296855859534504, + "grad_norm": 2.3012919425964355, + "learning_rate": 9.419163145699877e-06, + "loss": 2.510222816467285, + "step": 194580 + }, + { + "epoch": 0.5297400299441949, + "grad_norm": 2.184326171875, + "learning_rate": 9.418072781790923e-06, + "loss": 2.4761398315429686, + "step": 194600 + }, + { + "epoch": 0.5297944739349394, + "grad_norm": 2.6466944217681885, + "learning_rate": 9.416982417881969e-06, + "loss": 2.4887163162231447, + "step": 194620 + }, + { + "epoch": 0.529848917925684, + "grad_norm": 2.121223211288452, + "learning_rate": 9.415892053973014e-06, + "loss": 2.511785316467285, + "step": 194640 + }, + { + "epoch": 0.5299033619164285, + "grad_norm": 2.159771203994751, + "learning_rate": 9.414801690064059e-06, + "loss": 2.461905288696289, + "step": 194660 + }, + { + "epoch": 0.529957805907173, + "grad_norm": 2.1864726543426514, + "learning_rate": 9.413711326155104e-06, + "loss": 2.471476173400879, + "step": 194680 + }, + { + "epoch": 0.5300122498979175, + "grad_norm": 2.224487066268921, + "learning_rate": 9.41262096224615e-06, + "loss": 2.461297607421875, + "step": 194700 + }, + { + "epoch": 0.530066693888662, + "grad_norm": 2.1986708641052246, + "learning_rate": 9.411530598337196e-06, + "loss": 2.467473602294922, + "step": 194720 + }, + { + "epoch": 0.5301211378794065, + "grad_norm": 2.169539213180542, + "learning_rate": 9.41044023442824e-06, + "loss": 2.493214416503906, + "step": 194740 + }, + { + "epoch": 0.5301755818701511, + "grad_norm": 2.469871997833252, + "learning_rate": 9.409349870519286e-06, + "loss": 2.464056968688965, + "step": 194760 + }, + { + "epoch": 0.5302300258608956, + "grad_norm": 2.1293652057647705, + "learning_rate": 9.408259506610332e-06, + "loss": 2.476388359069824, + "step": 194780 + }, + { + "epoch": 0.5302844698516401, + "grad_norm": 2.211794853210449, + "learning_rate": 9.407169142701378e-06, + "loss": 2.4607370376586912, + "step": 194800 + }, + { + "epoch": 0.5303389138423846, + "grad_norm": 2.198697566986084, + "learning_rate": 9.406078778792422e-06, + "loss": 2.4467153549194336, + "step": 194820 + }, + { + "epoch": 0.5303933578331291, + "grad_norm": 2.083435297012329, + "learning_rate": 9.404988414883467e-06, + "loss": 2.5009075164794923, + "step": 194840 + }, + { + "epoch": 0.5304478018238737, + "grad_norm": 2.217115640640259, + "learning_rate": 9.403898050974513e-06, + "loss": 2.4710294723510744, + "step": 194860 + }, + { + "epoch": 0.5305022458146182, + "grad_norm": 2.0967564582824707, + "learning_rate": 9.402807687065559e-06, + "loss": 2.4614181518554688, + "step": 194880 + }, + { + "epoch": 0.5305566898053627, + "grad_norm": 2.130162000656128, + "learning_rate": 9.401717323156603e-06, + "loss": 2.488314628601074, + "step": 194900 + }, + { + "epoch": 0.5306111337961072, + "grad_norm": 2.2213993072509766, + "learning_rate": 9.400626959247649e-06, + "loss": 2.4914594650268556, + "step": 194920 + }, + { + "epoch": 0.5306655777868517, + "grad_norm": 2.2835617065429688, + "learning_rate": 9.399536595338695e-06, + "loss": 2.4751806259155273, + "step": 194940 + }, + { + "epoch": 0.5307200217775963, + "grad_norm": 2.1262998580932617, + "learning_rate": 9.39844623142974e-06, + "loss": 2.4874414443969726, + "step": 194960 + }, + { + "epoch": 0.5307744657683409, + "grad_norm": 2.0967061519622803, + "learning_rate": 9.397355867520785e-06, + "loss": 2.4779943466186523, + "step": 194980 + }, + { + "epoch": 0.5308289097590854, + "grad_norm": 2.2863378524780273, + "learning_rate": 9.39626550361183e-06, + "loss": 2.4980932235717774, + "step": 195000 + }, + { + "epoch": 0.5308833537498299, + "grad_norm": 2.111168384552002, + "learning_rate": 9.395175139702876e-06, + "loss": 2.4986139297485352, + "step": 195020 + }, + { + "epoch": 0.5309377977405744, + "grad_norm": 6.943628311157227, + "learning_rate": 9.394084775793922e-06, + "loss": 2.4876113891601563, + "step": 195040 + }, + { + "epoch": 0.5309922417313189, + "grad_norm": 2.4981539249420166, + "learning_rate": 9.392994411884966e-06, + "loss": 2.4735134124755858, + "step": 195060 + }, + { + "epoch": 0.5310466857220634, + "grad_norm": 2.2181503772735596, + "learning_rate": 9.391904047976012e-06, + "loss": 2.490326690673828, + "step": 195080 + }, + { + "epoch": 0.531101129712808, + "grad_norm": 2.239420175552368, + "learning_rate": 9.390813684067058e-06, + "loss": 2.4560016632080077, + "step": 195100 + }, + { + "epoch": 0.5311555737035525, + "grad_norm": 2.16935658454895, + "learning_rate": 9.389723320158104e-06, + "loss": 2.4858325958251952, + "step": 195120 + }, + { + "epoch": 0.531210017694297, + "grad_norm": 2.203209638595581, + "learning_rate": 9.388632956249148e-06, + "loss": 2.513211250305176, + "step": 195140 + }, + { + "epoch": 0.5312644616850415, + "grad_norm": 2.2169792652130127, + "learning_rate": 9.387542592340194e-06, + "loss": 2.4849929809570312, + "step": 195160 + }, + { + "epoch": 0.531318905675786, + "grad_norm": 2.095930337905884, + "learning_rate": 9.38645222843124e-06, + "loss": 2.4590887069702148, + "step": 195180 + }, + { + "epoch": 0.5313733496665306, + "grad_norm": 2.3487913608551025, + "learning_rate": 9.385361864522285e-06, + "loss": 2.4871612548828126, + "step": 195200 + }, + { + "epoch": 0.5314277936572751, + "grad_norm": 2.1247479915618896, + "learning_rate": 9.38427150061333e-06, + "loss": 2.462204360961914, + "step": 195220 + }, + { + "epoch": 0.5314822376480196, + "grad_norm": 2.1543643474578857, + "learning_rate": 9.383181136704375e-06, + "loss": 2.472188377380371, + "step": 195240 + }, + { + "epoch": 0.5315366816387641, + "grad_norm": 2.2091307640075684, + "learning_rate": 9.382090772795421e-06, + "loss": 2.4737720489501953, + "step": 195260 + }, + { + "epoch": 0.5315911256295086, + "grad_norm": 2.335062026977539, + "learning_rate": 9.381000408886467e-06, + "loss": 2.485050582885742, + "step": 195280 + }, + { + "epoch": 0.5316455696202531, + "grad_norm": 2.329773426055908, + "learning_rate": 9.379910044977511e-06, + "loss": 2.495412063598633, + "step": 195300 + }, + { + "epoch": 0.5317000136109977, + "grad_norm": 2.1126675605773926, + "learning_rate": 9.378819681068557e-06, + "loss": 2.4927627563476564, + "step": 195320 + }, + { + "epoch": 0.5317544576017422, + "grad_norm": 2.1323797702789307, + "learning_rate": 9.377729317159603e-06, + "loss": 2.466217803955078, + "step": 195340 + }, + { + "epoch": 0.5318089015924867, + "grad_norm": 2.207080841064453, + "learning_rate": 9.376638953250648e-06, + "loss": 2.509040641784668, + "step": 195360 + }, + { + "epoch": 0.5318633455832312, + "grad_norm": 2.1427040100097656, + "learning_rate": 9.375548589341692e-06, + "loss": 2.4692285537719725, + "step": 195380 + }, + { + "epoch": 0.5319177895739757, + "grad_norm": 2.1223835945129395, + "learning_rate": 9.374458225432738e-06, + "loss": 2.5013904571533203, + "step": 195400 + }, + { + "epoch": 0.5319722335647203, + "grad_norm": 2.3840975761413574, + "learning_rate": 9.373367861523784e-06, + "loss": 2.451228141784668, + "step": 195420 + }, + { + "epoch": 0.5320266775554648, + "grad_norm": 2.2857933044433594, + "learning_rate": 9.37227749761483e-06, + "loss": 2.468377685546875, + "step": 195440 + }, + { + "epoch": 0.5320811215462093, + "grad_norm": 2.0545032024383545, + "learning_rate": 9.371187133705874e-06, + "loss": 2.5157100677490236, + "step": 195460 + }, + { + "epoch": 0.5321355655369538, + "grad_norm": 2.25392484664917, + "learning_rate": 9.37009676979692e-06, + "loss": 2.4844146728515626, + "step": 195480 + }, + { + "epoch": 0.5321900095276983, + "grad_norm": 2.6108458042144775, + "learning_rate": 9.369006405887966e-06, + "loss": 2.4825000762939453, + "step": 195500 + }, + { + "epoch": 0.5322444535184429, + "grad_norm": 2.456599712371826, + "learning_rate": 9.367916041979011e-06, + "loss": 2.480487251281738, + "step": 195520 + }, + { + "epoch": 0.5322988975091875, + "grad_norm": 2.214940071105957, + "learning_rate": 9.366825678070056e-06, + "loss": 2.4745407104492188, + "step": 195540 + }, + { + "epoch": 0.532353341499932, + "grad_norm": 2.1924970149993896, + "learning_rate": 9.365735314161101e-06, + "loss": 2.4817584991455077, + "step": 195560 + }, + { + "epoch": 0.5324077854906765, + "grad_norm": 2.4358623027801514, + "learning_rate": 9.364644950252147e-06, + "loss": 2.469718170166016, + "step": 195580 + }, + { + "epoch": 0.532462229481421, + "grad_norm": 2.0827736854553223, + "learning_rate": 9.363554586343193e-06, + "loss": 2.477268028259277, + "step": 195600 + }, + { + "epoch": 0.5325166734721655, + "grad_norm": 2.109889030456543, + "learning_rate": 9.362464222434237e-06, + "loss": 2.4884666442871093, + "step": 195620 + }, + { + "epoch": 0.5325711174629101, + "grad_norm": 2.0849616527557373, + "learning_rate": 9.361373858525283e-06, + "loss": 2.4944034576416017, + "step": 195640 + }, + { + "epoch": 0.5326255614536546, + "grad_norm": 2.1927120685577393, + "learning_rate": 9.360283494616329e-06, + "loss": 2.5093032836914064, + "step": 195660 + }, + { + "epoch": 0.5326800054443991, + "grad_norm": 2.1379566192626953, + "learning_rate": 9.359193130707375e-06, + "loss": 2.462143325805664, + "step": 195680 + }, + { + "epoch": 0.5327344494351436, + "grad_norm": 2.15187931060791, + "learning_rate": 9.358102766798419e-06, + "loss": 2.458053779602051, + "step": 195700 + }, + { + "epoch": 0.5327888934258881, + "grad_norm": 2.171926259994507, + "learning_rate": 9.357012402889464e-06, + "loss": 2.4839468002319336, + "step": 195720 + }, + { + "epoch": 0.5328433374166326, + "grad_norm": 2.4100067615509033, + "learning_rate": 9.35592203898051e-06, + "loss": 2.482866096496582, + "step": 195740 + }, + { + "epoch": 0.5328977814073772, + "grad_norm": 2.2426958084106445, + "learning_rate": 9.354831675071556e-06, + "loss": 2.4646800994873046, + "step": 195760 + }, + { + "epoch": 0.5329522253981217, + "grad_norm": 2.2230424880981445, + "learning_rate": 9.3537413111626e-06, + "loss": 2.471294975280762, + "step": 195780 + }, + { + "epoch": 0.5330066693888662, + "grad_norm": 2.135413646697998, + "learning_rate": 9.352650947253646e-06, + "loss": 2.497972297668457, + "step": 195800 + }, + { + "epoch": 0.5330611133796107, + "grad_norm": 2.1234161853790283, + "learning_rate": 9.351560583344692e-06, + "loss": 2.504561996459961, + "step": 195820 + }, + { + "epoch": 0.5331155573703552, + "grad_norm": 2.222576141357422, + "learning_rate": 9.350470219435738e-06, + "loss": 2.4861331939697267, + "step": 195840 + }, + { + "epoch": 0.5331700013610998, + "grad_norm": 2.0676403045654297, + "learning_rate": 9.349379855526782e-06, + "loss": 2.45739803314209, + "step": 195860 + }, + { + "epoch": 0.5332244453518443, + "grad_norm": 2.138585329055786, + "learning_rate": 9.348289491617828e-06, + "loss": 2.477334403991699, + "step": 195880 + }, + { + "epoch": 0.5332788893425888, + "grad_norm": 2.210597515106201, + "learning_rate": 9.347199127708873e-06, + "loss": 2.4803239822387697, + "step": 195900 + }, + { + "epoch": 0.5333333333333333, + "grad_norm": 2.1563830375671387, + "learning_rate": 9.34610876379992e-06, + "loss": 2.4974185943603517, + "step": 195920 + }, + { + "epoch": 0.5333877773240778, + "grad_norm": 2.264312267303467, + "learning_rate": 9.345018399890963e-06, + "loss": 2.482126808166504, + "step": 195940 + }, + { + "epoch": 0.5334422213148223, + "grad_norm": 2.240823984146118, + "learning_rate": 9.343928035982009e-06, + "loss": 2.4775590896606445, + "step": 195960 + }, + { + "epoch": 0.5334966653055669, + "grad_norm": 2.1792984008789062, + "learning_rate": 9.342837672073055e-06, + "loss": 2.4419010162353514, + "step": 195980 + }, + { + "epoch": 0.5335511092963114, + "grad_norm": 2.192216634750366, + "learning_rate": 9.3417473081641e-06, + "loss": 2.511854553222656, + "step": 196000 + }, + { + "epoch": 0.533605553287056, + "grad_norm": 2.107328414916992, + "learning_rate": 9.340656944255147e-06, + "loss": 2.5159019470214843, + "step": 196020 + }, + { + "epoch": 0.5336599972778004, + "grad_norm": 2.1600894927978516, + "learning_rate": 9.33956658034619e-06, + "loss": 2.4781211853027343, + "step": 196040 + }, + { + "epoch": 0.533714441268545, + "grad_norm": 2.206658124923706, + "learning_rate": 9.338476216437236e-06, + "loss": 2.449156951904297, + "step": 196060 + }, + { + "epoch": 0.5337688852592896, + "grad_norm": 2.1803555488586426, + "learning_rate": 9.337385852528282e-06, + "loss": 2.507104682922363, + "step": 196080 + }, + { + "epoch": 0.5338233292500341, + "grad_norm": 2.1993613243103027, + "learning_rate": 9.336295488619328e-06, + "loss": 2.4901605606079102, + "step": 196100 + }, + { + "epoch": 0.5338777732407786, + "grad_norm": 2.0369536876678467, + "learning_rate": 9.335205124710372e-06, + "loss": 2.4884166717529297, + "step": 196120 + }, + { + "epoch": 0.5339322172315231, + "grad_norm": 2.05301833152771, + "learning_rate": 9.334114760801418e-06, + "loss": 2.491854667663574, + "step": 196140 + }, + { + "epoch": 0.5339866612222676, + "grad_norm": 2.1981921195983887, + "learning_rate": 9.333024396892464e-06, + "loss": 2.4634910583496095, + "step": 196160 + }, + { + "epoch": 0.5340411052130121, + "grad_norm": 2.214251756668091, + "learning_rate": 9.33193403298351e-06, + "loss": 2.447260856628418, + "step": 196180 + }, + { + "epoch": 0.5340955492037567, + "grad_norm": 2.158789873123169, + "learning_rate": 9.330843669074554e-06, + "loss": 2.4571306228637697, + "step": 196200 + }, + { + "epoch": 0.5341499931945012, + "grad_norm": 2.3678767681121826, + "learning_rate": 9.3297533051656e-06, + "loss": 2.4730152130126952, + "step": 196220 + }, + { + "epoch": 0.5342044371852457, + "grad_norm": 2.1882028579711914, + "learning_rate": 9.328662941256645e-06, + "loss": 2.512447738647461, + "step": 196240 + }, + { + "epoch": 0.5342588811759902, + "grad_norm": 2.1777267456054688, + "learning_rate": 9.327572577347691e-06, + "loss": 2.4974302291870116, + "step": 196260 + }, + { + "epoch": 0.5343133251667347, + "grad_norm": 2.3676891326904297, + "learning_rate": 9.326482213438735e-06, + "loss": 2.4914325714111327, + "step": 196280 + }, + { + "epoch": 0.5343677691574792, + "grad_norm": 2.1975576877593994, + "learning_rate": 9.325391849529781e-06, + "loss": 2.4810962677001953, + "step": 196300 + }, + { + "epoch": 0.5344222131482238, + "grad_norm": 2.1574795246124268, + "learning_rate": 9.324301485620827e-06, + "loss": 2.47976131439209, + "step": 196320 + }, + { + "epoch": 0.5344766571389683, + "grad_norm": 2.3260059356689453, + "learning_rate": 9.323211121711873e-06, + "loss": 2.478347396850586, + "step": 196340 + }, + { + "epoch": 0.5345311011297128, + "grad_norm": 2.0995194911956787, + "learning_rate": 9.322120757802917e-06, + "loss": 2.4930984497070314, + "step": 196360 + }, + { + "epoch": 0.5345855451204573, + "grad_norm": 2.1359496116638184, + "learning_rate": 9.321030393893963e-06, + "loss": 2.474625015258789, + "step": 196380 + }, + { + "epoch": 0.5346399891112018, + "grad_norm": 2.160212755203247, + "learning_rate": 9.319940029985008e-06, + "loss": 2.467816162109375, + "step": 196400 + }, + { + "epoch": 0.5346944331019464, + "grad_norm": 2.2098917961120605, + "learning_rate": 9.318849666076054e-06, + "loss": 2.4808855056762695, + "step": 196420 + }, + { + "epoch": 0.5347488770926909, + "grad_norm": 2.1488564014434814, + "learning_rate": 9.317759302167098e-06, + "loss": 2.4839223861694335, + "step": 196440 + }, + { + "epoch": 0.5348033210834354, + "grad_norm": 2.3043019771575928, + "learning_rate": 9.316668938258144e-06, + "loss": 2.50085391998291, + "step": 196460 + }, + { + "epoch": 0.5348577650741799, + "grad_norm": 2.4311764240264893, + "learning_rate": 9.31557857434919e-06, + "loss": 2.461321258544922, + "step": 196480 + }, + { + "epoch": 0.5349122090649244, + "grad_norm": 2.055628776550293, + "learning_rate": 9.314488210440236e-06, + "loss": 2.4661083221435547, + "step": 196500 + }, + { + "epoch": 0.5349666530556689, + "grad_norm": 2.305063009262085, + "learning_rate": 9.31339784653128e-06, + "loss": 2.503376007080078, + "step": 196520 + }, + { + "epoch": 0.5350210970464135, + "grad_norm": 7.6535539627075195, + "learning_rate": 9.312307482622326e-06, + "loss": 2.4563133239746096, + "step": 196540 + }, + { + "epoch": 0.535075541037158, + "grad_norm": 2.3587474822998047, + "learning_rate": 9.311217118713372e-06, + "loss": 2.479804801940918, + "step": 196560 + }, + { + "epoch": 0.5351299850279025, + "grad_norm": 2.496830940246582, + "learning_rate": 9.310126754804417e-06, + "loss": 2.473537635803223, + "step": 196580 + }, + { + "epoch": 0.535184429018647, + "grad_norm": 2.2092175483703613, + "learning_rate": 9.309036390895461e-06, + "loss": 2.4957860946655273, + "step": 196600 + }, + { + "epoch": 0.5352388730093915, + "grad_norm": 2.524714708328247, + "learning_rate": 9.307946026986507e-06, + "loss": 2.4685131072998048, + "step": 196620 + }, + { + "epoch": 0.5352933170001362, + "grad_norm": 2.1896190643310547, + "learning_rate": 9.306855663077553e-06, + "loss": 2.4642093658447264, + "step": 196640 + }, + { + "epoch": 0.5353477609908807, + "grad_norm": 2.214512825012207, + "learning_rate": 9.305765299168599e-06, + "loss": 2.4960647583007813, + "step": 196660 + }, + { + "epoch": 0.5354022049816252, + "grad_norm": 2.167344331741333, + "learning_rate": 9.304674935259643e-06, + "loss": 2.454814910888672, + "step": 196680 + }, + { + "epoch": 0.5354566489723697, + "grad_norm": 2.1574251651763916, + "learning_rate": 9.303584571350689e-06, + "loss": 2.4568012237548826, + "step": 196700 + }, + { + "epoch": 0.5355110929631142, + "grad_norm": 2.670001983642578, + "learning_rate": 9.302494207441735e-06, + "loss": 2.471152114868164, + "step": 196720 + }, + { + "epoch": 0.5355655369538587, + "grad_norm": 2.123997926712036, + "learning_rate": 9.30140384353278e-06, + "loss": 2.4858928680419923, + "step": 196740 + }, + { + "epoch": 0.5356199809446033, + "grad_norm": 2.3767035007476807, + "learning_rate": 9.300313479623825e-06, + "loss": 2.5062599182128906, + "step": 196760 + }, + { + "epoch": 0.5356744249353478, + "grad_norm": 2.1453335285186768, + "learning_rate": 9.29922311571487e-06, + "loss": 2.4670793533325197, + "step": 196780 + }, + { + "epoch": 0.5357288689260923, + "grad_norm": 2.16554594039917, + "learning_rate": 9.298132751805916e-06, + "loss": 2.4569957733154295, + "step": 196800 + }, + { + "epoch": 0.5357833129168368, + "grad_norm": 2.3450214862823486, + "learning_rate": 9.297042387896962e-06, + "loss": 2.4970281600952147, + "step": 196820 + }, + { + "epoch": 0.5358377569075813, + "grad_norm": 2.143063545227051, + "learning_rate": 9.295952023988006e-06, + "loss": 2.5146900177001954, + "step": 196840 + }, + { + "epoch": 0.5358922008983259, + "grad_norm": 2.0642805099487305, + "learning_rate": 9.294861660079052e-06, + "loss": 2.439415168762207, + "step": 196860 + }, + { + "epoch": 0.5359466448890704, + "grad_norm": 2.2621042728424072, + "learning_rate": 9.293771296170098e-06, + "loss": 2.5044715881347654, + "step": 196880 + }, + { + "epoch": 0.5360010888798149, + "grad_norm": 2.0906434059143066, + "learning_rate": 9.292680932261143e-06, + "loss": 2.4612249374389648, + "step": 196900 + }, + { + "epoch": 0.5360555328705594, + "grad_norm": 2.111804723739624, + "learning_rate": 9.291590568352188e-06, + "loss": 2.483677291870117, + "step": 196920 + }, + { + "epoch": 0.5361099768613039, + "grad_norm": 2.2014994621276855, + "learning_rate": 9.290500204443233e-06, + "loss": 2.479658508300781, + "step": 196940 + }, + { + "epoch": 0.5361644208520484, + "grad_norm": 2.1164448261260986, + "learning_rate": 9.28940984053428e-06, + "loss": 2.4941923141479494, + "step": 196960 + }, + { + "epoch": 0.536218864842793, + "grad_norm": 2.2042222023010254, + "learning_rate": 9.288319476625325e-06, + "loss": 2.470726203918457, + "step": 196980 + }, + { + "epoch": 0.5362733088335375, + "grad_norm": 2.285616159439087, + "learning_rate": 9.287229112716369e-06, + "loss": 2.4687408447265624, + "step": 197000 + }, + { + "epoch": 0.536327752824282, + "grad_norm": 2.1583611965179443, + "learning_rate": 9.286138748807415e-06, + "loss": 2.4744552612304687, + "step": 197020 + }, + { + "epoch": 0.5363821968150265, + "grad_norm": 2.095489978790283, + "learning_rate": 9.28504838489846e-06, + "loss": 2.5022357940673827, + "step": 197040 + }, + { + "epoch": 0.536436640805771, + "grad_norm": 2.228425979614258, + "learning_rate": 9.283958020989507e-06, + "loss": 2.4835865020751955, + "step": 197060 + }, + { + "epoch": 0.5364910847965156, + "grad_norm": 2.1637654304504395, + "learning_rate": 9.28286765708055e-06, + "loss": 2.478627586364746, + "step": 197080 + }, + { + "epoch": 0.5365455287872601, + "grad_norm": 2.2287142276763916, + "learning_rate": 9.281777293171597e-06, + "loss": 2.479435157775879, + "step": 197100 + }, + { + "epoch": 0.5365999727780046, + "grad_norm": 2.1277291774749756, + "learning_rate": 9.280686929262642e-06, + "loss": 2.52526798248291, + "step": 197120 + }, + { + "epoch": 0.5366544167687491, + "grad_norm": 2.0948736667633057, + "learning_rate": 9.279596565353688e-06, + "loss": 2.437642288208008, + "step": 197140 + }, + { + "epoch": 0.5367088607594936, + "grad_norm": 2.096627712249756, + "learning_rate": 9.278506201444732e-06, + "loss": 2.4824478149414064, + "step": 197160 + }, + { + "epoch": 0.5367633047502381, + "grad_norm": 2.1645331382751465, + "learning_rate": 9.277415837535778e-06, + "loss": 2.4875823974609377, + "step": 197180 + }, + { + "epoch": 0.5368177487409828, + "grad_norm": 2.1478536128997803, + "learning_rate": 9.276325473626824e-06, + "loss": 2.491998481750488, + "step": 197200 + }, + { + "epoch": 0.5368721927317273, + "grad_norm": 2.257337808609009, + "learning_rate": 9.27523510971787e-06, + "loss": 2.4654157638549803, + "step": 197220 + }, + { + "epoch": 0.5369266367224718, + "grad_norm": 2.151050567626953, + "learning_rate": 9.274144745808914e-06, + "loss": 2.4610315322875977, + "step": 197240 + }, + { + "epoch": 0.5369810807132163, + "grad_norm": 2.29292631149292, + "learning_rate": 9.27305438189996e-06, + "loss": 2.5197343826293945, + "step": 197260 + }, + { + "epoch": 0.5370355247039608, + "grad_norm": 2.1698334217071533, + "learning_rate": 9.271964017991005e-06, + "loss": 2.4991804122924806, + "step": 197280 + }, + { + "epoch": 0.5370899686947053, + "grad_norm": 2.1265976428985596, + "learning_rate": 9.270873654082051e-06, + "loss": 2.457418441772461, + "step": 197300 + }, + { + "epoch": 0.5371444126854499, + "grad_norm": 2.13446044921875, + "learning_rate": 9.269783290173095e-06, + "loss": 2.4532827377319335, + "step": 197320 + }, + { + "epoch": 0.5371988566761944, + "grad_norm": 2.1835944652557373, + "learning_rate": 9.268692926264141e-06, + "loss": 2.4821193695068358, + "step": 197340 + }, + { + "epoch": 0.5372533006669389, + "grad_norm": 2.1776490211486816, + "learning_rate": 9.267602562355187e-06, + "loss": 2.479483795166016, + "step": 197360 + }, + { + "epoch": 0.5373077446576834, + "grad_norm": 2.070195436477661, + "learning_rate": 9.266512198446233e-06, + "loss": 2.4508689880371093, + "step": 197380 + }, + { + "epoch": 0.5373621886484279, + "grad_norm": 2.1256449222564697, + "learning_rate": 9.265421834537277e-06, + "loss": 2.465227508544922, + "step": 197400 + }, + { + "epoch": 0.5374166326391725, + "grad_norm": 2.0812859535217285, + "learning_rate": 9.264331470628323e-06, + "loss": 2.4870670318603514, + "step": 197420 + }, + { + "epoch": 0.537471076629917, + "grad_norm": 2.184400796890259, + "learning_rate": 9.263241106719368e-06, + "loss": 2.4718271255493165, + "step": 197440 + }, + { + "epoch": 0.5375255206206615, + "grad_norm": 2.128150463104248, + "learning_rate": 9.262150742810414e-06, + "loss": 2.4753332138061523, + "step": 197460 + }, + { + "epoch": 0.537579964611406, + "grad_norm": 2.2266957759857178, + "learning_rate": 9.261060378901458e-06, + "loss": 2.4715490341186523, + "step": 197480 + }, + { + "epoch": 0.5376344086021505, + "grad_norm": 2.110578775405884, + "learning_rate": 9.259970014992504e-06, + "loss": 2.4714210510253904, + "step": 197500 + }, + { + "epoch": 0.537688852592895, + "grad_norm": 2.2487127780914307, + "learning_rate": 9.25887965108355e-06, + "loss": 2.4995695114135743, + "step": 197520 + }, + { + "epoch": 0.5377432965836396, + "grad_norm": 2.214057683944702, + "learning_rate": 9.257789287174596e-06, + "loss": 2.4928461074829102, + "step": 197540 + }, + { + "epoch": 0.5377977405743841, + "grad_norm": 2.196361780166626, + "learning_rate": 9.25669892326564e-06, + "loss": 2.482244300842285, + "step": 197560 + }, + { + "epoch": 0.5378521845651286, + "grad_norm": 2.2461612224578857, + "learning_rate": 9.255608559356686e-06, + "loss": 2.4608701705932616, + "step": 197580 + }, + { + "epoch": 0.5379066285558731, + "grad_norm": 2.1306331157684326, + "learning_rate": 9.254518195447732e-06, + "loss": 2.4648618698120117, + "step": 197600 + }, + { + "epoch": 0.5379610725466176, + "grad_norm": 2.0976901054382324, + "learning_rate": 9.253427831538777e-06, + "loss": 2.4839384078979494, + "step": 197620 + }, + { + "epoch": 0.5380155165373622, + "grad_norm": 2.273219347000122, + "learning_rate": 9.252337467629821e-06, + "loss": 2.4801637649536135, + "step": 197640 + }, + { + "epoch": 0.5380699605281067, + "grad_norm": 2.153831720352173, + "learning_rate": 9.251247103720867e-06, + "loss": 2.5051958084106447, + "step": 197660 + }, + { + "epoch": 0.5381244045188512, + "grad_norm": 2.3590543270111084, + "learning_rate": 9.250156739811913e-06, + "loss": 2.485249710083008, + "step": 197680 + }, + { + "epoch": 0.5381788485095957, + "grad_norm": 2.2710371017456055, + "learning_rate": 9.249066375902959e-06, + "loss": 2.473863983154297, + "step": 197700 + }, + { + "epoch": 0.5382332925003402, + "grad_norm": 2.226571798324585, + "learning_rate": 9.247976011994005e-06, + "loss": 2.5198410034179686, + "step": 197720 + }, + { + "epoch": 0.5382877364910847, + "grad_norm": 2.1791939735412598, + "learning_rate": 9.246885648085049e-06, + "loss": 2.4847923278808595, + "step": 197740 + }, + { + "epoch": 0.5383421804818294, + "grad_norm": 2.1571526527404785, + "learning_rate": 9.245795284176095e-06, + "loss": 2.474259376525879, + "step": 197760 + }, + { + "epoch": 0.5383966244725739, + "grad_norm": 2.2519869804382324, + "learning_rate": 9.24470492026714e-06, + "loss": 2.4824941635131834, + "step": 197780 + }, + { + "epoch": 0.5384510684633184, + "grad_norm": 2.1312317848205566, + "learning_rate": 9.243614556358186e-06, + "loss": 2.5039743423461913, + "step": 197800 + }, + { + "epoch": 0.5385055124540629, + "grad_norm": 2.1796858310699463, + "learning_rate": 9.24252419244923e-06, + "loss": 2.456707763671875, + "step": 197820 + }, + { + "epoch": 0.5385599564448074, + "grad_norm": 2.108231544494629, + "learning_rate": 9.241433828540276e-06, + "loss": 2.483053207397461, + "step": 197840 + }, + { + "epoch": 0.538614400435552, + "grad_norm": 2.1198296546936035, + "learning_rate": 9.240343464631322e-06, + "loss": 2.498604965209961, + "step": 197860 + }, + { + "epoch": 0.5386688444262965, + "grad_norm": 2.417454481124878, + "learning_rate": 9.239253100722368e-06, + "loss": 2.483064651489258, + "step": 197880 + }, + { + "epoch": 0.538723288417041, + "grad_norm": 2.1420488357543945, + "learning_rate": 9.238162736813412e-06, + "loss": 2.431308364868164, + "step": 197900 + }, + { + "epoch": 0.5387777324077855, + "grad_norm": 2.544686794281006, + "learning_rate": 9.237072372904458e-06, + "loss": 2.4708793640136717, + "step": 197920 + }, + { + "epoch": 0.53883217639853, + "grad_norm": 2.223557710647583, + "learning_rate": 9.235982008995504e-06, + "loss": 2.4607120513916017, + "step": 197940 + }, + { + "epoch": 0.5388866203892745, + "grad_norm": 2.1795814037323, + "learning_rate": 9.23489164508655e-06, + "loss": 2.463184356689453, + "step": 197960 + }, + { + "epoch": 0.5389410643800191, + "grad_norm": 2.217935800552368, + "learning_rate": 9.233801281177593e-06, + "loss": 2.486438751220703, + "step": 197980 + }, + { + "epoch": 0.5389955083707636, + "grad_norm": 2.1449074745178223, + "learning_rate": 9.23271091726864e-06, + "loss": 2.46435546875, + "step": 198000 + }, + { + "epoch": 0.5390499523615081, + "grad_norm": 2.132739305496216, + "learning_rate": 9.231620553359685e-06, + "loss": 2.4675792694091796, + "step": 198020 + }, + { + "epoch": 0.5391043963522526, + "grad_norm": 2.1626076698303223, + "learning_rate": 9.230530189450731e-06, + "loss": 2.4776538848876952, + "step": 198040 + }, + { + "epoch": 0.5391588403429971, + "grad_norm": 2.182359218597412, + "learning_rate": 9.229439825541775e-06, + "loss": 2.4713159561157227, + "step": 198060 + }, + { + "epoch": 0.5392132843337417, + "grad_norm": 2.193135976791382, + "learning_rate": 9.22834946163282e-06, + "loss": 2.4795726776123046, + "step": 198080 + }, + { + "epoch": 0.5392677283244862, + "grad_norm": 2.0776102542877197, + "learning_rate": 9.227259097723867e-06, + "loss": 2.473821258544922, + "step": 198100 + }, + { + "epoch": 0.5393221723152307, + "grad_norm": 2.2735891342163086, + "learning_rate": 9.226168733814912e-06, + "loss": 2.4892467498779296, + "step": 198120 + }, + { + "epoch": 0.5393766163059752, + "grad_norm": 2.1658525466918945, + "learning_rate": 9.225078369905957e-06, + "loss": 2.4700489044189453, + "step": 198140 + }, + { + "epoch": 0.5394310602967197, + "grad_norm": 2.094888687133789, + "learning_rate": 9.223988005997002e-06, + "loss": 2.476594352722168, + "step": 198160 + }, + { + "epoch": 0.5394855042874642, + "grad_norm": 2.1216487884521484, + "learning_rate": 9.222897642088048e-06, + "loss": 2.452309226989746, + "step": 198180 + }, + { + "epoch": 0.5395399482782088, + "grad_norm": 2.1271705627441406, + "learning_rate": 9.221807278179094e-06, + "loss": 2.497278594970703, + "step": 198200 + }, + { + "epoch": 0.5395943922689533, + "grad_norm": 2.2349674701690674, + "learning_rate": 9.220716914270138e-06, + "loss": 2.502994728088379, + "step": 198220 + }, + { + "epoch": 0.5396488362596978, + "grad_norm": 2.357821464538574, + "learning_rate": 9.219626550361184e-06, + "loss": 2.4745508193969727, + "step": 198240 + }, + { + "epoch": 0.5397032802504423, + "grad_norm": 2.262495994567871, + "learning_rate": 9.21853618645223e-06, + "loss": 2.4783340454101563, + "step": 198260 + }, + { + "epoch": 0.5397577242411868, + "grad_norm": 2.3752434253692627, + "learning_rate": 9.217445822543276e-06, + "loss": 2.4816429138183596, + "step": 198280 + }, + { + "epoch": 0.5398121682319315, + "grad_norm": 2.087960958480835, + "learning_rate": 9.21635545863432e-06, + "loss": 2.440878486633301, + "step": 198300 + }, + { + "epoch": 0.539866612222676, + "grad_norm": 2.146609306335449, + "learning_rate": 9.215265094725365e-06, + "loss": 2.468989372253418, + "step": 198320 + }, + { + "epoch": 0.5399210562134205, + "grad_norm": 2.083780288696289, + "learning_rate": 9.214174730816411e-06, + "loss": 2.4713083267211915, + "step": 198340 + }, + { + "epoch": 0.539975500204165, + "grad_norm": 2.4181368350982666, + "learning_rate": 9.213084366907457e-06, + "loss": 2.4688432693481444, + "step": 198360 + }, + { + "epoch": 0.5400299441949095, + "grad_norm": 2.288707733154297, + "learning_rate": 9.211994002998501e-06, + "loss": 2.4712839126586914, + "step": 198380 + }, + { + "epoch": 0.540084388185654, + "grad_norm": 2.1875205039978027, + "learning_rate": 9.210903639089547e-06, + "loss": 2.4857900619506834, + "step": 198400 + }, + { + "epoch": 0.5401388321763986, + "grad_norm": 2.167010545730591, + "learning_rate": 9.209813275180593e-06, + "loss": 2.4491973876953126, + "step": 198420 + }, + { + "epoch": 0.5401932761671431, + "grad_norm": 2.0674569606781006, + "learning_rate": 9.208722911271639e-06, + "loss": 2.502151870727539, + "step": 198440 + }, + { + "epoch": 0.5402477201578876, + "grad_norm": 2.113464117050171, + "learning_rate": 9.207632547362683e-06, + "loss": 2.4930999755859373, + "step": 198460 + }, + { + "epoch": 0.5403021641486321, + "grad_norm": 2.156841993331909, + "learning_rate": 9.206542183453729e-06, + "loss": 2.474185752868652, + "step": 198480 + }, + { + "epoch": 0.5403566081393766, + "grad_norm": 2.169590711593628, + "learning_rate": 9.205451819544774e-06, + "loss": 2.4880039215087892, + "step": 198500 + }, + { + "epoch": 0.5404110521301211, + "grad_norm": 2.086921453475952, + "learning_rate": 9.20436145563582e-06, + "loss": 2.4606948852539063, + "step": 198520 + }, + { + "epoch": 0.5404654961208657, + "grad_norm": 2.583996534347534, + "learning_rate": 9.203271091726864e-06, + "loss": 2.43810977935791, + "step": 198540 + }, + { + "epoch": 0.5405199401116102, + "grad_norm": 2.177623987197876, + "learning_rate": 9.20218072781791e-06, + "loss": 2.4869094848632813, + "step": 198560 + }, + { + "epoch": 0.5405743841023547, + "grad_norm": 2.083714246749878, + "learning_rate": 9.201090363908956e-06, + "loss": 2.514454460144043, + "step": 198580 + }, + { + "epoch": 0.5406288280930992, + "grad_norm": 2.1534106731414795, + "learning_rate": 9.200000000000002e-06, + "loss": 2.445358657836914, + "step": 198600 + }, + { + "epoch": 0.5406832720838437, + "grad_norm": 2.0915491580963135, + "learning_rate": 9.198909636091046e-06, + "loss": 2.483418083190918, + "step": 198620 + }, + { + "epoch": 0.5407377160745883, + "grad_norm": 2.1299326419830322, + "learning_rate": 9.197819272182092e-06, + "loss": 2.4431262969970704, + "step": 198640 + }, + { + "epoch": 0.5407921600653328, + "grad_norm": 2.1865758895874023, + "learning_rate": 9.196728908273137e-06, + "loss": 2.497090530395508, + "step": 198660 + }, + { + "epoch": 0.5408466040560773, + "grad_norm": 3.038944721221924, + "learning_rate": 9.195638544364183e-06, + "loss": 2.487657356262207, + "step": 198680 + }, + { + "epoch": 0.5409010480468218, + "grad_norm": 2.202909231185913, + "learning_rate": 9.194548180455227e-06, + "loss": 2.4729557037353516, + "step": 198700 + }, + { + "epoch": 0.5409554920375663, + "grad_norm": 2.238271713256836, + "learning_rate": 9.193457816546273e-06, + "loss": 2.4971643447875977, + "step": 198720 + }, + { + "epoch": 0.5410099360283108, + "grad_norm": 2.139132022857666, + "learning_rate": 9.192367452637319e-06, + "loss": 2.470871162414551, + "step": 198740 + }, + { + "epoch": 0.5410643800190554, + "grad_norm": 3.0445728302001953, + "learning_rate": 9.191277088728365e-06, + "loss": 2.5120025634765626, + "step": 198760 + }, + { + "epoch": 0.5411188240097999, + "grad_norm": 2.1125190258026123, + "learning_rate": 9.190186724819409e-06, + "loss": 2.476920318603516, + "step": 198780 + }, + { + "epoch": 0.5411732680005444, + "grad_norm": 2.137112855911255, + "learning_rate": 9.189096360910455e-06, + "loss": 2.491690254211426, + "step": 198800 + }, + { + "epoch": 0.5412277119912889, + "grad_norm": 2.174560308456421, + "learning_rate": 9.1880059970015e-06, + "loss": 2.4782987594604493, + "step": 198820 + }, + { + "epoch": 0.5412821559820334, + "grad_norm": 2.0911343097686768, + "learning_rate": 9.186915633092546e-06, + "loss": 2.4646684646606447, + "step": 198840 + }, + { + "epoch": 0.541336599972778, + "grad_norm": 2.219088077545166, + "learning_rate": 9.18582526918359e-06, + "loss": 2.4527801513671874, + "step": 198860 + }, + { + "epoch": 0.5413910439635226, + "grad_norm": 2.1600327491760254, + "learning_rate": 9.184734905274636e-06, + "loss": 2.4838180541992188, + "step": 198880 + }, + { + "epoch": 0.5414454879542671, + "grad_norm": 2.1862289905548096, + "learning_rate": 9.183644541365682e-06, + "loss": 2.464191436767578, + "step": 198900 + }, + { + "epoch": 0.5414999319450116, + "grad_norm": 2.132970094680786, + "learning_rate": 9.182554177456728e-06, + "loss": 2.485690689086914, + "step": 198920 + }, + { + "epoch": 0.5415543759357561, + "grad_norm": 2.0914556980133057, + "learning_rate": 9.181463813547772e-06, + "loss": 2.481044387817383, + "step": 198940 + }, + { + "epoch": 0.5416088199265006, + "grad_norm": 2.1810622215270996, + "learning_rate": 9.180373449638818e-06, + "loss": 2.4844860076904296, + "step": 198960 + }, + { + "epoch": 0.5416632639172452, + "grad_norm": 2.0277161598205566, + "learning_rate": 9.179283085729864e-06, + "loss": 2.4593263626098634, + "step": 198980 + }, + { + "epoch": 0.5417177079079897, + "grad_norm": 2.1381568908691406, + "learning_rate": 9.17819272182091e-06, + "loss": 2.476953887939453, + "step": 199000 + }, + { + "epoch": 0.5417721518987342, + "grad_norm": 2.281172752380371, + "learning_rate": 9.177102357911954e-06, + "loss": 2.5047927856445313, + "step": 199020 + }, + { + "epoch": 0.5418265958894787, + "grad_norm": 2.20947003364563, + "learning_rate": 9.176011994003e-06, + "loss": 2.48874568939209, + "step": 199040 + }, + { + "epoch": 0.5418810398802232, + "grad_norm": 2.170417547225952, + "learning_rate": 9.174921630094045e-06, + "loss": 2.4622875213623048, + "step": 199060 + }, + { + "epoch": 0.5419354838709678, + "grad_norm": 2.1337647438049316, + "learning_rate": 9.173831266185091e-06, + "loss": 2.4700441360473633, + "step": 199080 + }, + { + "epoch": 0.5419899278617123, + "grad_norm": 2.089811086654663, + "learning_rate": 9.172740902276135e-06, + "loss": 2.483695220947266, + "step": 199100 + }, + { + "epoch": 0.5420443718524568, + "grad_norm": 2.1890218257904053, + "learning_rate": 9.171650538367181e-06, + "loss": 2.4422281265258787, + "step": 199120 + }, + { + "epoch": 0.5420988158432013, + "grad_norm": 2.1962733268737793, + "learning_rate": 9.170560174458227e-06, + "loss": 2.465467071533203, + "step": 199140 + }, + { + "epoch": 0.5421532598339458, + "grad_norm": 2.211042881011963, + "learning_rate": 9.169469810549273e-06, + "loss": 2.4783435821533204, + "step": 199160 + }, + { + "epoch": 0.5422077038246903, + "grad_norm": 2.091505765914917, + "learning_rate": 9.168379446640317e-06, + "loss": 2.4849681854248047, + "step": 199180 + }, + { + "epoch": 0.5422621478154349, + "grad_norm": 2.241366386413574, + "learning_rate": 9.167289082731362e-06, + "loss": 2.4600360870361326, + "step": 199200 + }, + { + "epoch": 0.5423165918061794, + "grad_norm": 2.184983491897583, + "learning_rate": 9.166198718822408e-06, + "loss": 2.511264419555664, + "step": 199220 + }, + { + "epoch": 0.5423710357969239, + "grad_norm": 2.2668662071228027, + "learning_rate": 9.165108354913454e-06, + "loss": 2.47503547668457, + "step": 199240 + }, + { + "epoch": 0.5424254797876684, + "grad_norm": 2.0644445419311523, + "learning_rate": 9.164017991004498e-06, + "loss": 2.4598567962646483, + "step": 199260 + }, + { + "epoch": 0.5424799237784129, + "grad_norm": 2.1470980644226074, + "learning_rate": 9.162927627095544e-06, + "loss": 2.480845832824707, + "step": 199280 + }, + { + "epoch": 0.5425343677691575, + "grad_norm": 2.2046210765838623, + "learning_rate": 9.16183726318659e-06, + "loss": 2.4616798400878905, + "step": 199300 + }, + { + "epoch": 0.542588811759902, + "grad_norm": 2.0962231159210205, + "learning_rate": 9.160746899277636e-06, + "loss": 2.4495094299316404, + "step": 199320 + }, + { + "epoch": 0.5426432557506465, + "grad_norm": 2.118109941482544, + "learning_rate": 9.15965653536868e-06, + "loss": 2.4820734024047852, + "step": 199340 + }, + { + "epoch": 0.542697699741391, + "grad_norm": 2.1749637126922607, + "learning_rate": 9.158566171459726e-06, + "loss": 2.4756425857543944, + "step": 199360 + }, + { + "epoch": 0.5427521437321355, + "grad_norm": 2.100442886352539, + "learning_rate": 9.157475807550771e-06, + "loss": 2.476828384399414, + "step": 199380 + }, + { + "epoch": 0.54280658772288, + "grad_norm": 2.1982083320617676, + "learning_rate": 9.156385443641817e-06, + "loss": 2.487474060058594, + "step": 199400 + }, + { + "epoch": 0.5428610317136247, + "grad_norm": 2.1373085975646973, + "learning_rate": 9.155295079732863e-06, + "loss": 2.502354049682617, + "step": 199420 + }, + { + "epoch": 0.5429154757043692, + "grad_norm": 2.1707427501678467, + "learning_rate": 9.154204715823907e-06, + "loss": 2.458617401123047, + "step": 199440 + }, + { + "epoch": 0.5429699196951137, + "grad_norm": 2.122924566268921, + "learning_rate": 9.153114351914953e-06, + "loss": 2.4935922622680664, + "step": 199460 + }, + { + "epoch": 0.5430243636858582, + "grad_norm": 2.2079901695251465, + "learning_rate": 9.152023988005999e-06, + "loss": 2.478952407836914, + "step": 199480 + }, + { + "epoch": 0.5430788076766027, + "grad_norm": 2.12019419670105, + "learning_rate": 9.150933624097044e-06, + "loss": 2.493720054626465, + "step": 199500 + }, + { + "epoch": 0.5431332516673472, + "grad_norm": 2.103563070297241, + "learning_rate": 9.149843260188089e-06, + "loss": 2.481230545043945, + "step": 199520 + }, + { + "epoch": 0.5431876956580918, + "grad_norm": 2.068338394165039, + "learning_rate": 9.148752896279134e-06, + "loss": 2.4447166442871096, + "step": 199540 + }, + { + "epoch": 0.5432421396488363, + "grad_norm": 2.209892511367798, + "learning_rate": 9.14766253237018e-06, + "loss": 2.4689226150512695, + "step": 199560 + }, + { + "epoch": 0.5432965836395808, + "grad_norm": 2.310542583465576, + "learning_rate": 9.146572168461226e-06, + "loss": 2.4471668243408202, + "step": 199580 + }, + { + "epoch": 0.5433510276303253, + "grad_norm": 2.0379650592803955, + "learning_rate": 9.14548180455227e-06, + "loss": 2.5084095001220703, + "step": 199600 + }, + { + "epoch": 0.5434054716210698, + "grad_norm": 2.1967294216156006, + "learning_rate": 9.144391440643316e-06, + "loss": 2.4924278259277344, + "step": 199620 + }, + { + "epoch": 0.5434599156118144, + "grad_norm": 2.2111730575561523, + "learning_rate": 9.143301076734362e-06, + "loss": 2.4378000259399415, + "step": 199640 + }, + { + "epoch": 0.5435143596025589, + "grad_norm": 2.1291232109069824, + "learning_rate": 9.142210712825408e-06, + "loss": 2.4824546813964843, + "step": 199660 + }, + { + "epoch": 0.5435688035933034, + "grad_norm": 2.25704026222229, + "learning_rate": 9.141120348916452e-06, + "loss": 2.472955894470215, + "step": 199680 + }, + { + "epoch": 0.5436232475840479, + "grad_norm": 2.1122562885284424, + "learning_rate": 9.140029985007497e-06, + "loss": 2.4885786056518553, + "step": 199700 + }, + { + "epoch": 0.5436776915747924, + "grad_norm": 2.1948928833007812, + "learning_rate": 9.138939621098543e-06, + "loss": 2.486337661743164, + "step": 199720 + }, + { + "epoch": 0.5437321355655369, + "grad_norm": 2.0988049507141113, + "learning_rate": 9.137849257189589e-06, + "loss": 2.496235466003418, + "step": 199740 + }, + { + "epoch": 0.5437865795562815, + "grad_norm": 2.156019449234009, + "learning_rate": 9.136758893280633e-06, + "loss": 2.496688652038574, + "step": 199760 + }, + { + "epoch": 0.543841023547026, + "grad_norm": 2.164482355117798, + "learning_rate": 9.135668529371679e-06, + "loss": 2.466463088989258, + "step": 199780 + }, + { + "epoch": 0.5438954675377705, + "grad_norm": 2.2733023166656494, + "learning_rate": 9.134578165462725e-06, + "loss": 2.4583593368530274, + "step": 199800 + }, + { + "epoch": 0.543949911528515, + "grad_norm": 2.2236220836639404, + "learning_rate": 9.133487801553769e-06, + "loss": 2.4741231918334963, + "step": 199820 + }, + { + "epoch": 0.5440043555192595, + "grad_norm": 2.100961208343506, + "learning_rate": 9.132397437644815e-06, + "loss": 2.511841583251953, + "step": 199840 + }, + { + "epoch": 0.5440587995100041, + "grad_norm": 2.132761240005493, + "learning_rate": 9.13130707373586e-06, + "loss": 2.490169906616211, + "step": 199860 + }, + { + "epoch": 0.5441132435007486, + "grad_norm": 2.1784896850585938, + "learning_rate": 9.130216709826906e-06, + "loss": 2.502704620361328, + "step": 199880 + }, + { + "epoch": 0.5441676874914931, + "grad_norm": 2.9122211933135986, + "learning_rate": 9.12912634591795e-06, + "loss": 2.489254379272461, + "step": 199900 + }, + { + "epoch": 0.5442221314822376, + "grad_norm": 2.116405725479126, + "learning_rate": 9.128035982008996e-06, + "loss": 2.480159378051758, + "step": 199920 + }, + { + "epoch": 0.5442765754729821, + "grad_norm": 2.386247158050537, + "learning_rate": 9.126945618100042e-06, + "loss": 2.458065223693848, + "step": 199940 + }, + { + "epoch": 0.5443310194637266, + "grad_norm": 2.1992170810699463, + "learning_rate": 9.125855254191088e-06, + "loss": 2.475248908996582, + "step": 199960 + }, + { + "epoch": 0.5443854634544713, + "grad_norm": 2.087937116622925, + "learning_rate": 9.124764890282132e-06, + "loss": 2.474134826660156, + "step": 199980 + }, + { + "epoch": 0.5444399074452158, + "grad_norm": 2.473175287246704, + "learning_rate": 9.123674526373178e-06, + "loss": 2.4999698638916015, + "step": 200000 + }, + { + "epoch": 0.5444943514359603, + "grad_norm": 2.1292378902435303, + "learning_rate": 9.122584162464224e-06, + "loss": 2.4831331253051756, + "step": 200020 + }, + { + "epoch": 0.5445487954267048, + "grad_norm": 2.1298482418060303, + "learning_rate": 9.12149379855527e-06, + "loss": 2.4825281143188476, + "step": 200040 + }, + { + "epoch": 0.5446032394174493, + "grad_norm": 2.223388671875, + "learning_rate": 9.120403434646314e-06, + "loss": 2.46752815246582, + "step": 200060 + }, + { + "epoch": 0.5446576834081939, + "grad_norm": 2.212489604949951, + "learning_rate": 9.11931307073736e-06, + "loss": 2.457680320739746, + "step": 200080 + }, + { + "epoch": 0.5447121273989384, + "grad_norm": 2.337108612060547, + "learning_rate": 9.118222706828405e-06, + "loss": 2.4669168472290037, + "step": 200100 + }, + { + "epoch": 0.5447665713896829, + "grad_norm": 2.148270606994629, + "learning_rate": 9.117132342919451e-06, + "loss": 2.447254753112793, + "step": 200120 + }, + { + "epoch": 0.5448210153804274, + "grad_norm": 2.07391357421875, + "learning_rate": 9.116041979010495e-06, + "loss": 2.4812042236328127, + "step": 200140 + }, + { + "epoch": 0.5448754593711719, + "grad_norm": 2.2485618591308594, + "learning_rate": 9.114951615101541e-06, + "loss": 2.442626953125, + "step": 200160 + }, + { + "epoch": 0.5449299033619164, + "grad_norm": 2.201359510421753, + "learning_rate": 9.113861251192587e-06, + "loss": 2.4711761474609375, + "step": 200180 + }, + { + "epoch": 0.544984347352661, + "grad_norm": 2.0859291553497314, + "learning_rate": 9.112770887283633e-06, + "loss": 2.4785894393920898, + "step": 200200 + }, + { + "epoch": 0.5450387913434055, + "grad_norm": 2.135713577270508, + "learning_rate": 9.111680523374677e-06, + "loss": 2.4796689987182616, + "step": 200220 + }, + { + "epoch": 0.54509323533415, + "grad_norm": 2.1105754375457764, + "learning_rate": 9.110590159465722e-06, + "loss": 2.4717334747314452, + "step": 200240 + }, + { + "epoch": 0.5451476793248945, + "grad_norm": 2.2556376457214355, + "learning_rate": 9.109499795556768e-06, + "loss": 2.485057067871094, + "step": 200260 + }, + { + "epoch": 0.545202123315639, + "grad_norm": 2.161376714706421, + "learning_rate": 9.108409431647814e-06, + "loss": 2.50516357421875, + "step": 200280 + }, + { + "epoch": 0.5452565673063836, + "grad_norm": 2.1054294109344482, + "learning_rate": 9.107319067738858e-06, + "loss": 2.471241760253906, + "step": 200300 + }, + { + "epoch": 0.5453110112971281, + "grad_norm": 2.2057642936706543, + "learning_rate": 9.106228703829904e-06, + "loss": 2.5046001434326173, + "step": 200320 + }, + { + "epoch": 0.5453654552878726, + "grad_norm": 2.1674840450286865, + "learning_rate": 9.10513833992095e-06, + "loss": 2.467513656616211, + "step": 200340 + }, + { + "epoch": 0.5454198992786171, + "grad_norm": 2.2622499465942383, + "learning_rate": 9.104047976011996e-06, + "loss": 2.4870214462280273, + "step": 200360 + }, + { + "epoch": 0.5454743432693616, + "grad_norm": 2.1384565830230713, + "learning_rate": 9.10295761210304e-06, + "loss": 2.4633960723876953, + "step": 200380 + }, + { + "epoch": 0.5455287872601061, + "grad_norm": 2.1395609378814697, + "learning_rate": 9.101867248194086e-06, + "loss": 2.467650032043457, + "step": 200400 + }, + { + "epoch": 0.5455832312508507, + "grad_norm": 2.1566224098205566, + "learning_rate": 9.100776884285131e-06, + "loss": 2.4658246994018556, + "step": 200420 + }, + { + "epoch": 0.5456376752415952, + "grad_norm": 2.3641550540924072, + "learning_rate": 9.099686520376177e-06, + "loss": 2.461699676513672, + "step": 200440 + }, + { + "epoch": 0.5456921192323397, + "grad_norm": 2.1250407695770264, + "learning_rate": 9.098596156467221e-06, + "loss": 2.4811721801757813, + "step": 200460 + }, + { + "epoch": 0.5457465632230842, + "grad_norm": 2.0654523372650146, + "learning_rate": 9.097505792558267e-06, + "loss": 2.483543586730957, + "step": 200480 + }, + { + "epoch": 0.5458010072138287, + "grad_norm": 2.1656813621520996, + "learning_rate": 9.096415428649313e-06, + "loss": 2.476084899902344, + "step": 200500 + }, + { + "epoch": 0.5458554512045733, + "grad_norm": 2.144728899002075, + "learning_rate": 9.095325064740359e-06, + "loss": 2.4857946395874024, + "step": 200520 + }, + { + "epoch": 0.5459098951953179, + "grad_norm": 2.1700870990753174, + "learning_rate": 9.094234700831403e-06, + "loss": 2.487436294555664, + "step": 200540 + }, + { + "epoch": 0.5459643391860624, + "grad_norm": 2.1576709747314453, + "learning_rate": 9.093144336922449e-06, + "loss": 2.483352470397949, + "step": 200560 + }, + { + "epoch": 0.5460187831768069, + "grad_norm": 2.2061047554016113, + "learning_rate": 9.092053973013494e-06, + "loss": 2.4837488174438476, + "step": 200580 + }, + { + "epoch": 0.5460732271675514, + "grad_norm": 2.2001819610595703, + "learning_rate": 9.09096360910454e-06, + "loss": 2.4707973480224608, + "step": 200600 + }, + { + "epoch": 0.5461276711582959, + "grad_norm": 2.090472459793091, + "learning_rate": 9.089873245195584e-06, + "loss": 2.498360252380371, + "step": 200620 + }, + { + "epoch": 0.5461821151490405, + "grad_norm": 2.217965602874756, + "learning_rate": 9.08878288128663e-06, + "loss": 2.489723014831543, + "step": 200640 + }, + { + "epoch": 0.546236559139785, + "grad_norm": 2.092301845550537, + "learning_rate": 9.087692517377676e-06, + "loss": 2.5021364212036135, + "step": 200660 + }, + { + "epoch": 0.5462910031305295, + "grad_norm": 3.465853452682495, + "learning_rate": 9.086602153468722e-06, + "loss": 2.4839542388916014, + "step": 200680 + }, + { + "epoch": 0.546345447121274, + "grad_norm": 2.2392289638519287, + "learning_rate": 9.085511789559766e-06, + "loss": 2.5120044708251954, + "step": 200700 + }, + { + "epoch": 0.5463998911120185, + "grad_norm": 2.141632080078125, + "learning_rate": 9.084421425650812e-06, + "loss": 2.4708213806152344, + "step": 200720 + }, + { + "epoch": 0.546454335102763, + "grad_norm": 2.1714348793029785, + "learning_rate": 9.083331061741858e-06, + "loss": 2.4823732376098633, + "step": 200740 + }, + { + "epoch": 0.5465087790935076, + "grad_norm": 2.1449050903320312, + "learning_rate": 9.082240697832903e-06, + "loss": 2.4913990020751955, + "step": 200760 + }, + { + "epoch": 0.5465632230842521, + "grad_norm": 2.0771892070770264, + "learning_rate": 9.081150333923947e-06, + "loss": 2.4939523696899415, + "step": 200780 + }, + { + "epoch": 0.5466176670749966, + "grad_norm": 2.2567269802093506, + "learning_rate": 9.080059970014993e-06, + "loss": 2.4706853866577148, + "step": 200800 + }, + { + "epoch": 0.5466721110657411, + "grad_norm": 2.272705554962158, + "learning_rate": 9.078969606106039e-06, + "loss": 2.4849884033203127, + "step": 200820 + }, + { + "epoch": 0.5467265550564856, + "grad_norm": 2.2279133796691895, + "learning_rate": 9.077879242197083e-06, + "loss": 2.4593072891235352, + "step": 200840 + }, + { + "epoch": 0.5467809990472302, + "grad_norm": 2.1745316982269287, + "learning_rate": 9.076788878288129e-06, + "loss": 2.484682083129883, + "step": 200860 + }, + { + "epoch": 0.5468354430379747, + "grad_norm": 2.721220016479492, + "learning_rate": 9.075698514379175e-06, + "loss": 2.487155532836914, + "step": 200880 + }, + { + "epoch": 0.5468898870287192, + "grad_norm": 2.2580626010894775, + "learning_rate": 9.07460815047022e-06, + "loss": 2.5291576385498047, + "step": 200900 + }, + { + "epoch": 0.5469443310194637, + "grad_norm": 2.221043348312378, + "learning_rate": 9.073517786561265e-06, + "loss": 2.459359359741211, + "step": 200920 + }, + { + "epoch": 0.5469987750102082, + "grad_norm": 2.2105391025543213, + "learning_rate": 9.07242742265231e-06, + "loss": 2.4734737396240236, + "step": 200940 + }, + { + "epoch": 0.5470532190009527, + "grad_norm": 2.2567636966705322, + "learning_rate": 9.071337058743356e-06, + "loss": 2.462491989135742, + "step": 200960 + }, + { + "epoch": 0.5471076629916973, + "grad_norm": 2.8279783725738525, + "learning_rate": 9.070246694834402e-06, + "loss": 2.4867076873779297, + "step": 200980 + }, + { + "epoch": 0.5471621069824418, + "grad_norm": 2.382298469543457, + "learning_rate": 9.069156330925446e-06, + "loss": 2.4606250762939452, + "step": 201000 + }, + { + "epoch": 0.5472165509731863, + "grad_norm": 2.1432018280029297, + "learning_rate": 9.068065967016492e-06, + "loss": 2.5152462005615233, + "step": 201020 + }, + { + "epoch": 0.5472709949639308, + "grad_norm": 2.211300849914551, + "learning_rate": 9.066975603107538e-06, + "loss": 2.43807315826416, + "step": 201040 + }, + { + "epoch": 0.5473254389546753, + "grad_norm": 2.1345601081848145, + "learning_rate": 9.065885239198584e-06, + "loss": 2.4904436111450194, + "step": 201060 + }, + { + "epoch": 0.54737988294542, + "grad_norm": 2.382944107055664, + "learning_rate": 9.064794875289628e-06, + "loss": 2.4821807861328127, + "step": 201080 + }, + { + "epoch": 0.5474343269361645, + "grad_norm": 7.68849515914917, + "learning_rate": 9.063704511380674e-06, + "loss": 2.5064706802368164, + "step": 201100 + }, + { + "epoch": 0.547488770926909, + "grad_norm": 2.2197580337524414, + "learning_rate": 9.06261414747172e-06, + "loss": 2.476993942260742, + "step": 201120 + }, + { + "epoch": 0.5475432149176535, + "grad_norm": 2.117232084274292, + "learning_rate": 9.061523783562765e-06, + "loss": 2.454767417907715, + "step": 201140 + }, + { + "epoch": 0.547597658908398, + "grad_norm": 2.320615291595459, + "learning_rate": 9.06043341965381e-06, + "loss": 2.4768604278564452, + "step": 201160 + }, + { + "epoch": 0.5476521028991425, + "grad_norm": 2.3101019859313965, + "learning_rate": 9.059343055744855e-06, + "loss": 2.4745695114135744, + "step": 201180 + }, + { + "epoch": 0.5477065468898871, + "grad_norm": 2.3971831798553467, + "learning_rate": 9.058252691835901e-06, + "loss": 2.4587734222412108, + "step": 201200 + }, + { + "epoch": 0.5477609908806316, + "grad_norm": 2.5995326042175293, + "learning_rate": 9.057162327926947e-06, + "loss": 2.4931798934936524, + "step": 201220 + }, + { + "epoch": 0.5478154348713761, + "grad_norm": 2.118086338043213, + "learning_rate": 9.056071964017991e-06, + "loss": 2.463735580444336, + "step": 201240 + }, + { + "epoch": 0.5478698788621206, + "grad_norm": 2.156322479248047, + "learning_rate": 9.054981600109037e-06, + "loss": 2.490085220336914, + "step": 201260 + }, + { + "epoch": 0.5479243228528651, + "grad_norm": 2.423398017883301, + "learning_rate": 9.053891236200083e-06, + "loss": 2.475657081604004, + "step": 201280 + }, + { + "epoch": 0.5479787668436097, + "grad_norm": 2.5696263313293457, + "learning_rate": 9.052800872291128e-06, + "loss": 2.4728878021240233, + "step": 201300 + }, + { + "epoch": 0.5480332108343542, + "grad_norm": 2.2241291999816895, + "learning_rate": 9.051710508382172e-06, + "loss": 2.4787166595458983, + "step": 201320 + }, + { + "epoch": 0.5480876548250987, + "grad_norm": 2.160325765609741, + "learning_rate": 9.050620144473218e-06, + "loss": 2.4879966735839845, + "step": 201340 + }, + { + "epoch": 0.5481420988158432, + "grad_norm": 2.3877694606781006, + "learning_rate": 9.049529780564264e-06, + "loss": 2.4648565292358398, + "step": 201360 + }, + { + "epoch": 0.5481965428065877, + "grad_norm": 2.230652332305908, + "learning_rate": 9.04843941665531e-06, + "loss": 2.4926300048828125, + "step": 201380 + }, + { + "epoch": 0.5482509867973322, + "grad_norm": 2.1874287128448486, + "learning_rate": 9.047349052746354e-06, + "loss": 2.4918354034423826, + "step": 201400 + }, + { + "epoch": 0.5483054307880768, + "grad_norm": 2.171478509902954, + "learning_rate": 9.0462586888374e-06, + "loss": 2.5180917739868165, + "step": 201420 + }, + { + "epoch": 0.5483598747788213, + "grad_norm": 2.353055477142334, + "learning_rate": 9.045168324928446e-06, + "loss": 2.484891891479492, + "step": 201440 + }, + { + "epoch": 0.5484143187695658, + "grad_norm": 2.2500288486480713, + "learning_rate": 9.044077961019491e-06, + "loss": 2.476352500915527, + "step": 201460 + }, + { + "epoch": 0.5484687627603103, + "grad_norm": 2.1647961139678955, + "learning_rate": 9.042987597110536e-06, + "loss": 2.491529655456543, + "step": 201480 + }, + { + "epoch": 0.5485232067510548, + "grad_norm": 2.309145450592041, + "learning_rate": 9.041897233201581e-06, + "loss": 2.4902271270751952, + "step": 201500 + }, + { + "epoch": 0.5485776507417994, + "grad_norm": 2.169182777404785, + "learning_rate": 9.040806869292627e-06, + "loss": 2.5251691818237303, + "step": 201520 + }, + { + "epoch": 0.5486320947325439, + "grad_norm": 2.2436165809631348, + "learning_rate": 9.039716505383673e-06, + "loss": 2.494868278503418, + "step": 201540 + }, + { + "epoch": 0.5486865387232884, + "grad_norm": 2.1744635105133057, + "learning_rate": 9.038626141474717e-06, + "loss": 2.4907079696655274, + "step": 201560 + }, + { + "epoch": 0.5487409827140329, + "grad_norm": 2.103254556655884, + "learning_rate": 9.037535777565763e-06, + "loss": 2.474094200134277, + "step": 201580 + }, + { + "epoch": 0.5487954267047774, + "grad_norm": 2.0726683139801025, + "learning_rate": 9.036445413656809e-06, + "loss": 2.468012046813965, + "step": 201600 + }, + { + "epoch": 0.5488498706955219, + "grad_norm": 2.2324748039245605, + "learning_rate": 9.035355049747855e-06, + "loss": 2.484035110473633, + "step": 201620 + }, + { + "epoch": 0.5489043146862665, + "grad_norm": 2.0898830890655518, + "learning_rate": 9.034264685838899e-06, + "loss": 2.4872617721557617, + "step": 201640 + }, + { + "epoch": 0.548958758677011, + "grad_norm": 2.2283055782318115, + "learning_rate": 9.033174321929944e-06, + "loss": 2.5139019012451174, + "step": 201660 + }, + { + "epoch": 0.5490132026677556, + "grad_norm": 2.1399166584014893, + "learning_rate": 9.03208395802099e-06, + "loss": 2.4657661437988283, + "step": 201680 + }, + { + "epoch": 0.5490676466585, + "grad_norm": 2.0726542472839355, + "learning_rate": 9.030993594112036e-06, + "loss": 2.493551826477051, + "step": 201700 + }, + { + "epoch": 0.5491220906492446, + "grad_norm": 2.149412155151367, + "learning_rate": 9.02990323020308e-06, + "loss": 2.4710086822509765, + "step": 201720 + }, + { + "epoch": 0.5491765346399891, + "grad_norm": 2.1264686584472656, + "learning_rate": 9.028812866294126e-06, + "loss": 2.445081329345703, + "step": 201740 + }, + { + "epoch": 0.5492309786307337, + "grad_norm": 2.2501630783081055, + "learning_rate": 9.027722502385172e-06, + "loss": 2.4776344299316406, + "step": 201760 + }, + { + "epoch": 0.5492854226214782, + "grad_norm": 2.097477912902832, + "learning_rate": 9.026632138476218e-06, + "loss": 2.4534950256347656, + "step": 201780 + }, + { + "epoch": 0.5493398666122227, + "grad_norm": 2.161346197128296, + "learning_rate": 9.025541774567262e-06, + "loss": 2.5065502166748046, + "step": 201800 + }, + { + "epoch": 0.5493943106029672, + "grad_norm": 2.207071542739868, + "learning_rate": 9.024451410658308e-06, + "loss": 2.486652946472168, + "step": 201820 + }, + { + "epoch": 0.5494487545937117, + "grad_norm": 2.0976431369781494, + "learning_rate": 9.023361046749353e-06, + "loss": 2.4491460800170897, + "step": 201840 + }, + { + "epoch": 0.5495031985844563, + "grad_norm": 2.088245153427124, + "learning_rate": 9.022270682840399e-06, + "loss": 2.4868478775024414, + "step": 201860 + }, + { + "epoch": 0.5495576425752008, + "grad_norm": 2.298670530319214, + "learning_rate": 9.021180318931443e-06, + "loss": 2.4680442810058594, + "step": 201880 + }, + { + "epoch": 0.5496120865659453, + "grad_norm": 2.056830883026123, + "learning_rate": 9.020089955022489e-06, + "loss": 2.438889503479004, + "step": 201900 + }, + { + "epoch": 0.5496665305566898, + "grad_norm": 2.289214611053467, + "learning_rate": 9.018999591113535e-06, + "loss": 2.4651004791259767, + "step": 201920 + }, + { + "epoch": 0.5497209745474343, + "grad_norm": 2.1915321350097656, + "learning_rate": 9.017909227204579e-06, + "loss": 2.4605451583862306, + "step": 201940 + }, + { + "epoch": 0.5497754185381788, + "grad_norm": 2.1080334186553955, + "learning_rate": 9.016818863295625e-06, + "loss": 2.4816375732421876, + "step": 201960 + }, + { + "epoch": 0.5498298625289234, + "grad_norm": 2.07271671295166, + "learning_rate": 9.01572849938667e-06, + "loss": 2.4563777923583983, + "step": 201980 + }, + { + "epoch": 0.5498843065196679, + "grad_norm": 2.1145541667938232, + "learning_rate": 9.014638135477716e-06, + "loss": 2.5001502990722657, + "step": 202000 + }, + { + "epoch": 0.5499387505104124, + "grad_norm": 2.170552968978882, + "learning_rate": 9.01354777156876e-06, + "loss": 2.469315528869629, + "step": 202020 + }, + { + "epoch": 0.5499931945011569, + "grad_norm": 2.1828715801239014, + "learning_rate": 9.012457407659806e-06, + "loss": 2.451231575012207, + "step": 202040 + }, + { + "epoch": 0.5500476384919014, + "grad_norm": 2.132934331893921, + "learning_rate": 9.011367043750852e-06, + "loss": 2.5065479278564453, + "step": 202060 + }, + { + "epoch": 0.550102082482646, + "grad_norm": 2.225233554840088, + "learning_rate": 9.010276679841898e-06, + "loss": 2.4969385147094725, + "step": 202080 + }, + { + "epoch": 0.5501565264733905, + "grad_norm": 2.192584991455078, + "learning_rate": 9.009186315932942e-06, + "loss": 2.431395721435547, + "step": 202100 + }, + { + "epoch": 0.550210970464135, + "grad_norm": 2.1254823207855225, + "learning_rate": 9.008095952023988e-06, + "loss": 2.504818344116211, + "step": 202120 + }, + { + "epoch": 0.5502654144548795, + "grad_norm": 2.1424481868743896, + "learning_rate": 9.007005588115034e-06, + "loss": 2.5019676208496096, + "step": 202140 + }, + { + "epoch": 0.550319858445624, + "grad_norm": 2.6560633182525635, + "learning_rate": 9.00591522420608e-06, + "loss": 2.477878189086914, + "step": 202160 + }, + { + "epoch": 0.5503743024363685, + "grad_norm": 2.168159008026123, + "learning_rate": 9.004824860297124e-06, + "loss": 2.502821159362793, + "step": 202180 + }, + { + "epoch": 0.5504287464271131, + "grad_norm": 2.149871349334717, + "learning_rate": 9.00373449638817e-06, + "loss": 2.494915771484375, + "step": 202200 + }, + { + "epoch": 0.5504831904178576, + "grad_norm": 2.032313346862793, + "learning_rate": 9.002644132479215e-06, + "loss": 2.469584846496582, + "step": 202220 + }, + { + "epoch": 0.5505376344086022, + "grad_norm": 2.412081003189087, + "learning_rate": 9.001553768570261e-06, + "loss": 2.495124626159668, + "step": 202240 + }, + { + "epoch": 0.5505920783993467, + "grad_norm": 2.076181411743164, + "learning_rate": 9.000463404661305e-06, + "loss": 2.49735164642334, + "step": 202260 + }, + { + "epoch": 0.5506465223900912, + "grad_norm": 2.5979161262512207, + "learning_rate": 8.999373040752351e-06, + "loss": 2.4918731689453124, + "step": 202280 + }, + { + "epoch": 0.5507009663808358, + "grad_norm": 2.1218979358673096, + "learning_rate": 8.998282676843397e-06, + "loss": 2.4463796615600586, + "step": 202300 + }, + { + "epoch": 0.5507554103715803, + "grad_norm": 2.2155356407165527, + "learning_rate": 8.997192312934443e-06, + "loss": 2.4659231185913084, + "step": 202320 + }, + { + "epoch": 0.5508098543623248, + "grad_norm": 2.204263925552368, + "learning_rate": 8.996101949025487e-06, + "loss": 2.459675407409668, + "step": 202340 + }, + { + "epoch": 0.5508642983530693, + "grad_norm": 2.263364315032959, + "learning_rate": 8.995011585116533e-06, + "loss": 2.4741889953613283, + "step": 202360 + }, + { + "epoch": 0.5509187423438138, + "grad_norm": 2.4417948722839355, + "learning_rate": 8.993921221207578e-06, + "loss": 2.4772640228271485, + "step": 202380 + }, + { + "epoch": 0.5509731863345583, + "grad_norm": 2.1105473041534424, + "learning_rate": 8.992830857298624e-06, + "loss": 2.4814594268798826, + "step": 202400 + }, + { + "epoch": 0.5510276303253029, + "grad_norm": 2.247776508331299, + "learning_rate": 8.991740493389668e-06, + "loss": 2.493184280395508, + "step": 202420 + }, + { + "epoch": 0.5510820743160474, + "grad_norm": 2.101545572280884, + "learning_rate": 8.990650129480714e-06, + "loss": 2.4701988220214846, + "step": 202440 + }, + { + "epoch": 0.5511365183067919, + "grad_norm": 2.0952956676483154, + "learning_rate": 8.98955976557176e-06, + "loss": 2.4799173355102537, + "step": 202460 + }, + { + "epoch": 0.5511909622975364, + "grad_norm": 2.1162281036376953, + "learning_rate": 8.988469401662806e-06, + "loss": 2.4791040420532227, + "step": 202480 + }, + { + "epoch": 0.5512454062882809, + "grad_norm": 2.2230112552642822, + "learning_rate": 8.98737903775385e-06, + "loss": 2.477753257751465, + "step": 202500 + }, + { + "epoch": 0.5512998502790255, + "grad_norm": 2.26279878616333, + "learning_rate": 8.986288673844896e-06, + "loss": 2.4872869491577148, + "step": 202520 + }, + { + "epoch": 0.55135429426977, + "grad_norm": 2.0904383659362793, + "learning_rate": 8.985198309935941e-06, + "loss": 2.4789360046386717, + "step": 202540 + }, + { + "epoch": 0.5514087382605145, + "grad_norm": 2.0961015224456787, + "learning_rate": 8.984107946026987e-06, + "loss": 2.454377365112305, + "step": 202560 + }, + { + "epoch": 0.551463182251259, + "grad_norm": 2.174109935760498, + "learning_rate": 8.983017582118031e-06, + "loss": 2.4663394927978515, + "step": 202580 + }, + { + "epoch": 0.5515176262420035, + "grad_norm": 2.0844907760620117, + "learning_rate": 8.981927218209077e-06, + "loss": 2.4731340408325195, + "step": 202600 + }, + { + "epoch": 0.551572070232748, + "grad_norm": 2.1616926193237305, + "learning_rate": 8.980836854300123e-06, + "loss": 2.4955562591552733, + "step": 202620 + }, + { + "epoch": 0.5516265142234926, + "grad_norm": 2.259906530380249, + "learning_rate": 8.979746490391169e-06, + "loss": 2.4794052124023436, + "step": 202640 + }, + { + "epoch": 0.5516809582142371, + "grad_norm": 2.3446764945983887, + "learning_rate": 8.978656126482213e-06, + "loss": 2.5044330596923827, + "step": 202660 + }, + { + "epoch": 0.5517354022049816, + "grad_norm": 2.2831637859344482, + "learning_rate": 8.977565762573259e-06, + "loss": 2.4577030181884765, + "step": 202680 + }, + { + "epoch": 0.5517898461957261, + "grad_norm": 2.1165308952331543, + "learning_rate": 8.976475398664304e-06, + "loss": 2.4776689529418947, + "step": 202700 + }, + { + "epoch": 0.5518442901864706, + "grad_norm": 2.276139736175537, + "learning_rate": 8.97538503475535e-06, + "loss": 2.513003921508789, + "step": 202720 + }, + { + "epoch": 0.5518987341772152, + "grad_norm": 2.185914993286133, + "learning_rate": 8.974294670846396e-06, + "loss": 2.4903940200805663, + "step": 202740 + }, + { + "epoch": 0.5519531781679597, + "grad_norm": 2.085463047027588, + "learning_rate": 8.97320430693744e-06, + "loss": 2.472393608093262, + "step": 202760 + }, + { + "epoch": 0.5520076221587042, + "grad_norm": 2.1945436000823975, + "learning_rate": 8.972113943028486e-06, + "loss": 2.4909622192382814, + "step": 202780 + }, + { + "epoch": 0.5520620661494487, + "grad_norm": 2.19244122505188, + "learning_rate": 8.971023579119532e-06, + "loss": 2.4883115768432615, + "step": 202800 + }, + { + "epoch": 0.5521165101401933, + "grad_norm": 2.356229066848755, + "learning_rate": 8.969933215210578e-06, + "loss": 2.489012336730957, + "step": 202820 + }, + { + "epoch": 0.5521709541309378, + "grad_norm": 2.2668354511260986, + "learning_rate": 8.968842851301622e-06, + "loss": 2.481622314453125, + "step": 202840 + }, + { + "epoch": 0.5522253981216824, + "grad_norm": 2.2337851524353027, + "learning_rate": 8.967752487392668e-06, + "loss": 2.459554100036621, + "step": 202860 + }, + { + "epoch": 0.5522798421124269, + "grad_norm": 2.1988725662231445, + "learning_rate": 8.966662123483713e-06, + "loss": 2.490176963806152, + "step": 202880 + }, + { + "epoch": 0.5523342861031714, + "grad_norm": 2.2484776973724365, + "learning_rate": 8.96557175957476e-06, + "loss": 2.4672386169433596, + "step": 202900 + }, + { + "epoch": 0.5523887300939159, + "grad_norm": 2.3142051696777344, + "learning_rate": 8.964481395665803e-06, + "loss": 2.487045097351074, + "step": 202920 + }, + { + "epoch": 0.5524431740846604, + "grad_norm": 2.130674362182617, + "learning_rate": 8.963391031756849e-06, + "loss": 2.4779731750488283, + "step": 202940 + }, + { + "epoch": 0.5524976180754049, + "grad_norm": 2.1215362548828125, + "learning_rate": 8.962300667847895e-06, + "loss": 2.5027204513549806, + "step": 202960 + }, + { + "epoch": 0.5525520620661495, + "grad_norm": 2.173013687133789, + "learning_rate": 8.96121030393894e-06, + "loss": 2.462060737609863, + "step": 202980 + }, + { + "epoch": 0.552606506056894, + "grad_norm": 3.5013883113861084, + "learning_rate": 8.960119940029985e-06, + "loss": 2.498061752319336, + "step": 203000 + }, + { + "epoch": 0.5526609500476385, + "grad_norm": 2.083833694458008, + "learning_rate": 8.95902957612103e-06, + "loss": 2.515129280090332, + "step": 203020 + }, + { + "epoch": 0.552715394038383, + "grad_norm": 2.2290327548980713, + "learning_rate": 8.957939212212076e-06, + "loss": 2.4605817794799805, + "step": 203040 + }, + { + "epoch": 0.5527698380291275, + "grad_norm": 2.0732228755950928, + "learning_rate": 8.956848848303122e-06, + "loss": 2.4941415786743164, + "step": 203060 + }, + { + "epoch": 0.5528242820198721, + "grad_norm": 2.161454916000366, + "learning_rate": 8.955758484394166e-06, + "loss": 2.4902664184570313, + "step": 203080 + }, + { + "epoch": 0.5528787260106166, + "grad_norm": 2.172663927078247, + "learning_rate": 8.954668120485212e-06, + "loss": 2.4917022705078127, + "step": 203100 + }, + { + "epoch": 0.5529331700013611, + "grad_norm": 2.277937412261963, + "learning_rate": 8.953577756576258e-06, + "loss": 2.5070690155029296, + "step": 203120 + }, + { + "epoch": 0.5529876139921056, + "grad_norm": 2.2353906631469727, + "learning_rate": 8.952487392667304e-06, + "loss": 2.4387792587280273, + "step": 203140 + }, + { + "epoch": 0.5530420579828501, + "grad_norm": 2.11308217048645, + "learning_rate": 8.951397028758348e-06, + "loss": 2.4441497802734373, + "step": 203160 + }, + { + "epoch": 0.5530965019735946, + "grad_norm": 2.074960947036743, + "learning_rate": 8.950306664849394e-06, + "loss": 2.45529727935791, + "step": 203180 + }, + { + "epoch": 0.5531509459643392, + "grad_norm": 2.118391513824463, + "learning_rate": 8.94921630094044e-06, + "loss": 2.47357063293457, + "step": 203200 + }, + { + "epoch": 0.5532053899550837, + "grad_norm": 2.169019937515259, + "learning_rate": 8.948125937031485e-06, + "loss": 2.4387773513793944, + "step": 203220 + }, + { + "epoch": 0.5532598339458282, + "grad_norm": 2.167154550552368, + "learning_rate": 8.94703557312253e-06, + "loss": 2.4692907333374023, + "step": 203240 + }, + { + "epoch": 0.5533142779365727, + "grad_norm": 2.1133663654327393, + "learning_rate": 8.945945209213575e-06, + "loss": 2.496755599975586, + "step": 203260 + }, + { + "epoch": 0.5533687219273172, + "grad_norm": 2.1306607723236084, + "learning_rate": 8.944854845304621e-06, + "loss": 2.4389411926269533, + "step": 203280 + }, + { + "epoch": 0.5534231659180618, + "grad_norm": 2.0953497886657715, + "learning_rate": 8.943764481395667e-06, + "loss": 2.506917953491211, + "step": 203300 + }, + { + "epoch": 0.5534776099088063, + "grad_norm": 2.181239604949951, + "learning_rate": 8.942674117486711e-06, + "loss": 2.479930114746094, + "step": 203320 + }, + { + "epoch": 0.5535320538995508, + "grad_norm": 2.121396780014038, + "learning_rate": 8.941583753577757e-06, + "loss": 2.483387565612793, + "step": 203340 + }, + { + "epoch": 0.5535864978902953, + "grad_norm": 2.072035789489746, + "learning_rate": 8.940493389668803e-06, + "loss": 2.4637998580932616, + "step": 203360 + }, + { + "epoch": 0.5536409418810398, + "grad_norm": 2.125657558441162, + "learning_rate": 8.939403025759848e-06, + "loss": 2.4779983520507813, + "step": 203380 + }, + { + "epoch": 0.5536953858717844, + "grad_norm": 2.164050817489624, + "learning_rate": 8.938312661850893e-06, + "loss": 2.4881792068481445, + "step": 203400 + }, + { + "epoch": 0.553749829862529, + "grad_norm": 2.1385648250579834, + "learning_rate": 8.937222297941938e-06, + "loss": 2.4518909454345703, + "step": 203420 + }, + { + "epoch": 0.5538042738532735, + "grad_norm": 2.1407930850982666, + "learning_rate": 8.936131934032984e-06, + "loss": 2.4520858764648437, + "step": 203440 + }, + { + "epoch": 0.553858717844018, + "grad_norm": 2.6113927364349365, + "learning_rate": 8.93504157012403e-06, + "loss": 2.4681686401367187, + "step": 203460 + }, + { + "epoch": 0.5539131618347625, + "grad_norm": 2.115185022354126, + "learning_rate": 8.933951206215074e-06, + "loss": 2.4569015502929688, + "step": 203480 + }, + { + "epoch": 0.553967605825507, + "grad_norm": 2.0967156887054443, + "learning_rate": 8.93286084230612e-06, + "loss": 2.4708045959472655, + "step": 203500 + }, + { + "epoch": 0.5540220498162516, + "grad_norm": 2.162114143371582, + "learning_rate": 8.931770478397166e-06, + "loss": 2.458660888671875, + "step": 203520 + }, + { + "epoch": 0.5540764938069961, + "grad_norm": 2.2254624366760254, + "learning_rate": 8.930680114488212e-06, + "loss": 2.477203369140625, + "step": 203540 + }, + { + "epoch": 0.5541309377977406, + "grad_norm": 2.2454912662506104, + "learning_rate": 8.929589750579256e-06, + "loss": 2.4583736419677735, + "step": 203560 + }, + { + "epoch": 0.5541853817884851, + "grad_norm": 2.372718334197998, + "learning_rate": 8.928499386670301e-06, + "loss": 2.4851972579956056, + "step": 203580 + }, + { + "epoch": 0.5542398257792296, + "grad_norm": 3.101578712463379, + "learning_rate": 8.927409022761347e-06, + "loss": 2.4810626983642576, + "step": 203600 + }, + { + "epoch": 0.5542942697699741, + "grad_norm": 2.398406505584717, + "learning_rate": 8.926318658852393e-06, + "loss": 2.4841604232788086, + "step": 203620 + }, + { + "epoch": 0.5543487137607187, + "grad_norm": 2.15574049949646, + "learning_rate": 8.925228294943437e-06, + "loss": 2.478473663330078, + "step": 203640 + }, + { + "epoch": 0.5544031577514632, + "grad_norm": 2.81644868850708, + "learning_rate": 8.924137931034483e-06, + "loss": 2.479954147338867, + "step": 203660 + }, + { + "epoch": 0.5544576017422077, + "grad_norm": 2.1715407371520996, + "learning_rate": 8.923047567125529e-06, + "loss": 2.477369499206543, + "step": 203680 + }, + { + "epoch": 0.5545120457329522, + "grad_norm": 2.134026050567627, + "learning_rate": 8.921957203216575e-06, + "loss": 2.4392799377441405, + "step": 203700 + }, + { + "epoch": 0.5545664897236967, + "grad_norm": 2.199234962463379, + "learning_rate": 8.920866839307619e-06, + "loss": 2.4519176483154297, + "step": 203720 + }, + { + "epoch": 0.5546209337144413, + "grad_norm": 2.237717628479004, + "learning_rate": 8.919776475398665e-06, + "loss": 2.4918233871459963, + "step": 203740 + }, + { + "epoch": 0.5546753777051858, + "grad_norm": 2.1924898624420166, + "learning_rate": 8.91868611148971e-06, + "loss": 2.509159469604492, + "step": 203760 + }, + { + "epoch": 0.5547298216959303, + "grad_norm": 2.1511621475219727, + "learning_rate": 8.917595747580756e-06, + "loss": 2.499594879150391, + "step": 203780 + }, + { + "epoch": 0.5547842656866748, + "grad_norm": 2.339184284210205, + "learning_rate": 8.9165053836718e-06, + "loss": 2.4916122436523436, + "step": 203800 + }, + { + "epoch": 0.5548387096774193, + "grad_norm": 2.245359420776367, + "learning_rate": 8.915415019762846e-06, + "loss": 2.483549880981445, + "step": 203820 + }, + { + "epoch": 0.5548931536681638, + "grad_norm": 2.4257211685180664, + "learning_rate": 8.914324655853892e-06, + "loss": 2.4336322784423827, + "step": 203840 + }, + { + "epoch": 0.5549475976589084, + "grad_norm": 2.1815972328186035, + "learning_rate": 8.913234291944938e-06, + "loss": 2.4609704971313477, + "step": 203860 + }, + { + "epoch": 0.5550020416496529, + "grad_norm": 2.040316104888916, + "learning_rate": 8.912143928035982e-06, + "loss": 2.5029743194580076, + "step": 203880 + }, + { + "epoch": 0.5550564856403974, + "grad_norm": 2.11503005027771, + "learning_rate": 8.911053564127028e-06, + "loss": 2.4729787826538088, + "step": 203900 + }, + { + "epoch": 0.555110929631142, + "grad_norm": 2.1971287727355957, + "learning_rate": 8.909963200218073e-06, + "loss": 2.4788631439208983, + "step": 203920 + }, + { + "epoch": 0.5551653736218864, + "grad_norm": 2.19362735748291, + "learning_rate": 8.90887283630912e-06, + "loss": 2.511060905456543, + "step": 203940 + }, + { + "epoch": 0.555219817612631, + "grad_norm": 2.187885046005249, + "learning_rate": 8.907782472400163e-06, + "loss": 2.4725278854370116, + "step": 203960 + }, + { + "epoch": 0.5552742616033756, + "grad_norm": 2.3025665283203125, + "learning_rate": 8.90669210849121e-06, + "loss": 2.476871109008789, + "step": 203980 + }, + { + "epoch": 0.5553287055941201, + "grad_norm": 2.0903213024139404, + "learning_rate": 8.905601744582255e-06, + "loss": 2.4536725997924806, + "step": 204000 + }, + { + "epoch": 0.5553831495848646, + "grad_norm": 2.0980381965637207, + "learning_rate": 8.9045113806733e-06, + "loss": 2.4928606033325194, + "step": 204020 + }, + { + "epoch": 0.5554375935756091, + "grad_norm": 2.0723628997802734, + "learning_rate": 8.903421016764345e-06, + "loss": 2.4938514709472654, + "step": 204040 + }, + { + "epoch": 0.5554920375663536, + "grad_norm": 2.105193853378296, + "learning_rate": 8.90233065285539e-06, + "loss": 2.4710079193115235, + "step": 204060 + }, + { + "epoch": 0.5555464815570982, + "grad_norm": 2.1338462829589844, + "learning_rate": 8.901240288946437e-06, + "loss": 2.482053756713867, + "step": 204080 + }, + { + "epoch": 0.5556009255478427, + "grad_norm": 2.170167922973633, + "learning_rate": 8.900149925037482e-06, + "loss": 2.5056827545166014, + "step": 204100 + }, + { + "epoch": 0.5556553695385872, + "grad_norm": 2.2822465896606445, + "learning_rate": 8.899059561128526e-06, + "loss": 2.518378257751465, + "step": 204120 + }, + { + "epoch": 0.5557098135293317, + "grad_norm": 2.179598093032837, + "learning_rate": 8.897969197219572e-06, + "loss": 2.4651689529418945, + "step": 204140 + }, + { + "epoch": 0.5557642575200762, + "grad_norm": 2.2132651805877686, + "learning_rate": 8.896878833310618e-06, + "loss": 2.4714981079101563, + "step": 204160 + }, + { + "epoch": 0.5558187015108207, + "grad_norm": 2.10438871383667, + "learning_rate": 8.895788469401664e-06, + "loss": 2.490620994567871, + "step": 204180 + }, + { + "epoch": 0.5558731455015653, + "grad_norm": 2.075526237487793, + "learning_rate": 8.894698105492708e-06, + "loss": 2.475617027282715, + "step": 204200 + }, + { + "epoch": 0.5559275894923098, + "grad_norm": 2.1247940063476562, + "learning_rate": 8.893607741583754e-06, + "loss": 2.4813974380493162, + "step": 204220 + }, + { + "epoch": 0.5559820334830543, + "grad_norm": 2.1365292072296143, + "learning_rate": 8.8925173776748e-06, + "loss": 2.435871124267578, + "step": 204240 + }, + { + "epoch": 0.5560364774737988, + "grad_norm": 2.0793445110321045, + "learning_rate": 8.891427013765845e-06, + "loss": 2.4916683197021485, + "step": 204260 + }, + { + "epoch": 0.5560909214645433, + "grad_norm": 2.168886184692383, + "learning_rate": 8.89033664985689e-06, + "loss": 2.478036880493164, + "step": 204280 + }, + { + "epoch": 0.5561453654552879, + "grad_norm": 2.095029592514038, + "learning_rate": 8.889246285947935e-06, + "loss": 2.4580575942993166, + "step": 204300 + }, + { + "epoch": 0.5561998094460324, + "grad_norm": 2.4712259769439697, + "learning_rate": 8.888155922038981e-06, + "loss": 2.455666923522949, + "step": 204320 + }, + { + "epoch": 0.5562542534367769, + "grad_norm": 2.2274441719055176, + "learning_rate": 8.887065558130027e-06, + "loss": 2.499444770812988, + "step": 204340 + }, + { + "epoch": 0.5563086974275214, + "grad_norm": 2.183133840560913, + "learning_rate": 8.885975194221071e-06, + "loss": 2.5041440963745116, + "step": 204360 + }, + { + "epoch": 0.5563631414182659, + "grad_norm": 2.2210190296173096, + "learning_rate": 8.884884830312117e-06, + "loss": 2.474798393249512, + "step": 204380 + }, + { + "epoch": 0.5564175854090104, + "grad_norm": 2.457197666168213, + "learning_rate": 8.883794466403163e-06, + "loss": 2.4701351165771483, + "step": 204400 + }, + { + "epoch": 0.556472029399755, + "grad_norm": 2.1662003993988037, + "learning_rate": 8.882704102494209e-06, + "loss": 2.458882522583008, + "step": 204420 + }, + { + "epoch": 0.5565264733904995, + "grad_norm": 2.9248361587524414, + "learning_rate": 8.881613738585254e-06, + "loss": 2.4612035751342773, + "step": 204440 + }, + { + "epoch": 0.556580917381244, + "grad_norm": 2.1192524433135986, + "learning_rate": 8.880523374676298e-06, + "loss": 2.454080581665039, + "step": 204460 + }, + { + "epoch": 0.5566353613719885, + "grad_norm": 2.236384630203247, + "learning_rate": 8.879433010767344e-06, + "loss": 2.4755270004272463, + "step": 204480 + }, + { + "epoch": 0.556689805362733, + "grad_norm": 2.1192514896392822, + "learning_rate": 8.87834264685839e-06, + "loss": 2.4489423751831056, + "step": 204500 + }, + { + "epoch": 0.5567442493534777, + "grad_norm": 2.206407308578491, + "learning_rate": 8.877252282949436e-06, + "loss": 2.475859832763672, + "step": 204520 + }, + { + "epoch": 0.5567986933442222, + "grad_norm": 2.1851840019226074, + "learning_rate": 8.87616191904048e-06, + "loss": 2.477204513549805, + "step": 204540 + }, + { + "epoch": 0.5568531373349667, + "grad_norm": 2.1493074893951416, + "learning_rate": 8.875071555131526e-06, + "loss": 2.5092514038085936, + "step": 204560 + }, + { + "epoch": 0.5569075813257112, + "grad_norm": 2.0738606452941895, + "learning_rate": 8.873981191222572e-06, + "loss": 2.500336456298828, + "step": 204580 + }, + { + "epoch": 0.5569620253164557, + "grad_norm": 2.220425605773926, + "learning_rate": 8.872890827313617e-06, + "loss": 2.4995706558227537, + "step": 204600 + }, + { + "epoch": 0.5570164693072002, + "grad_norm": 2.168983221054077, + "learning_rate": 8.871800463404662e-06, + "loss": 2.4960163116455076, + "step": 204620 + }, + { + "epoch": 0.5570709132979448, + "grad_norm": 2.2672765254974365, + "learning_rate": 8.870710099495707e-06, + "loss": 2.4550460815429687, + "step": 204640 + }, + { + "epoch": 0.5571253572886893, + "grad_norm": 2.326416015625, + "learning_rate": 8.869619735586753e-06, + "loss": 2.4644283294677733, + "step": 204660 + }, + { + "epoch": 0.5571798012794338, + "grad_norm": 2.1565747261047363, + "learning_rate": 8.868529371677799e-06, + "loss": 2.4754411697387697, + "step": 204680 + }, + { + "epoch": 0.5572342452701783, + "grad_norm": 2.1106908321380615, + "learning_rate": 8.867439007768843e-06, + "loss": 2.477412223815918, + "step": 204700 + }, + { + "epoch": 0.5572886892609228, + "grad_norm": 2.131204605102539, + "learning_rate": 8.866348643859889e-06, + "loss": 2.4946962356567384, + "step": 204720 + }, + { + "epoch": 0.5573431332516674, + "grad_norm": 2.2259793281555176, + "learning_rate": 8.865258279950935e-06, + "loss": 2.466546058654785, + "step": 204740 + }, + { + "epoch": 0.5573975772424119, + "grad_norm": 2.143177032470703, + "learning_rate": 8.86416791604198e-06, + "loss": 2.4739540100097654, + "step": 204760 + }, + { + "epoch": 0.5574520212331564, + "grad_norm": 2.1105294227600098, + "learning_rate": 8.863077552133025e-06, + "loss": 2.5070367813110352, + "step": 204780 + }, + { + "epoch": 0.5575064652239009, + "grad_norm": 2.2454681396484375, + "learning_rate": 8.86198718822407e-06, + "loss": 2.4671113967895506, + "step": 204800 + }, + { + "epoch": 0.5575609092146454, + "grad_norm": 2.136608123779297, + "learning_rate": 8.860896824315116e-06, + "loss": 2.4816072463989256, + "step": 204820 + }, + { + "epoch": 0.5576153532053899, + "grad_norm": 2.1049702167510986, + "learning_rate": 8.859806460406162e-06, + "loss": 2.487032890319824, + "step": 204840 + }, + { + "epoch": 0.5576697971961345, + "grad_norm": 2.1831891536712646, + "learning_rate": 8.858716096497206e-06, + "loss": 2.4795841217041015, + "step": 204860 + }, + { + "epoch": 0.557724241186879, + "grad_norm": 2.126171350479126, + "learning_rate": 8.857625732588252e-06, + "loss": 2.474522018432617, + "step": 204880 + }, + { + "epoch": 0.5577786851776235, + "grad_norm": 2.0545167922973633, + "learning_rate": 8.856535368679298e-06, + "loss": 2.4461942672729493, + "step": 204900 + }, + { + "epoch": 0.557833129168368, + "grad_norm": 2.152757167816162, + "learning_rate": 8.855445004770344e-06, + "loss": 2.4513187408447266, + "step": 204920 + }, + { + "epoch": 0.5578875731591125, + "grad_norm": 2.1671628952026367, + "learning_rate": 8.854354640861388e-06, + "loss": 2.4889888763427734, + "step": 204940 + }, + { + "epoch": 0.5579420171498571, + "grad_norm": 2.01902437210083, + "learning_rate": 8.853264276952434e-06, + "loss": 2.489866828918457, + "step": 204960 + }, + { + "epoch": 0.5579964611406016, + "grad_norm": 2.1730642318725586, + "learning_rate": 8.85217391304348e-06, + "loss": 2.465269088745117, + "step": 204980 + }, + { + "epoch": 0.5580509051313461, + "grad_norm": 2.23219895362854, + "learning_rate": 8.851083549134525e-06, + "loss": 2.4902151107788084, + "step": 205000 + }, + { + "epoch": 0.5581053491220906, + "grad_norm": 2.2539119720458984, + "learning_rate": 8.84999318522557e-06, + "loss": 2.498325729370117, + "step": 205020 + }, + { + "epoch": 0.5581597931128351, + "grad_norm": 2.184483051300049, + "learning_rate": 8.848902821316615e-06, + "loss": 2.473447799682617, + "step": 205040 + }, + { + "epoch": 0.5582142371035796, + "grad_norm": 2.0747556686401367, + "learning_rate": 8.847812457407661e-06, + "loss": 2.496215057373047, + "step": 205060 + }, + { + "epoch": 0.5582686810943243, + "grad_norm": 2.173062801361084, + "learning_rate": 8.846722093498707e-06, + "loss": 2.4891616821289064, + "step": 205080 + }, + { + "epoch": 0.5583231250850688, + "grad_norm": 2.2119195461273193, + "learning_rate": 8.84563172958975e-06, + "loss": 2.4944080352783202, + "step": 205100 + }, + { + "epoch": 0.5583775690758133, + "grad_norm": 2.172637462615967, + "learning_rate": 8.844541365680797e-06, + "loss": 2.4976264953613283, + "step": 205120 + }, + { + "epoch": 0.5584320130665578, + "grad_norm": 2.090131998062134, + "learning_rate": 8.843451001771842e-06, + "loss": 2.4714054107666015, + "step": 205140 + }, + { + "epoch": 0.5584864570573023, + "grad_norm": 2.8806023597717285, + "learning_rate": 8.842360637862888e-06, + "loss": 2.487230682373047, + "step": 205160 + }, + { + "epoch": 0.5585409010480468, + "grad_norm": 2.0289790630340576, + "learning_rate": 8.841270273953932e-06, + "loss": 2.498055076599121, + "step": 205180 + }, + { + "epoch": 0.5585953450387914, + "grad_norm": 2.0377144813537598, + "learning_rate": 8.840179910044978e-06, + "loss": 2.4607851028442385, + "step": 205200 + }, + { + "epoch": 0.5586497890295359, + "grad_norm": 2.1141605377197266, + "learning_rate": 8.839089546136024e-06, + "loss": 2.4729856491088866, + "step": 205220 + }, + { + "epoch": 0.5587042330202804, + "grad_norm": 2.3563270568847656, + "learning_rate": 8.83799918222707e-06, + "loss": 2.49859504699707, + "step": 205240 + }, + { + "epoch": 0.5587586770110249, + "grad_norm": 3.0240695476531982, + "learning_rate": 8.836908818318114e-06, + "loss": 2.4392879486083983, + "step": 205260 + }, + { + "epoch": 0.5588131210017694, + "grad_norm": 2.0980050563812256, + "learning_rate": 8.83581845440916e-06, + "loss": 2.469900131225586, + "step": 205280 + }, + { + "epoch": 0.558867564992514, + "grad_norm": 2.387397050857544, + "learning_rate": 8.834728090500205e-06, + "loss": 2.481847953796387, + "step": 205300 + }, + { + "epoch": 0.5589220089832585, + "grad_norm": 2.0675137042999268, + "learning_rate": 8.833637726591251e-06, + "loss": 2.4969085693359374, + "step": 205320 + }, + { + "epoch": 0.558976452974003, + "grad_norm": 2.2003607749938965, + "learning_rate": 8.832547362682295e-06, + "loss": 2.4812686920166014, + "step": 205340 + }, + { + "epoch": 0.5590308969647475, + "grad_norm": 2.1567542552948, + "learning_rate": 8.831456998773341e-06, + "loss": 2.4658851623535156, + "step": 205360 + }, + { + "epoch": 0.559085340955492, + "grad_norm": 2.1337084770202637, + "learning_rate": 8.830366634864387e-06, + "loss": 2.518826484680176, + "step": 205380 + }, + { + "epoch": 0.5591397849462365, + "grad_norm": 2.142648458480835, + "learning_rate": 8.829276270955433e-06, + "loss": 2.4776279449462892, + "step": 205400 + }, + { + "epoch": 0.5591942289369811, + "grad_norm": 2.832487106323242, + "learning_rate": 8.828185907046477e-06, + "loss": 2.4796445846557615, + "step": 205420 + }, + { + "epoch": 0.5592486729277256, + "grad_norm": 2.28886342048645, + "learning_rate": 8.827095543137523e-06, + "loss": 2.4643924713134764, + "step": 205440 + }, + { + "epoch": 0.5593031169184701, + "grad_norm": 2.155360698699951, + "learning_rate": 8.826005179228569e-06, + "loss": 2.4629919052124025, + "step": 205460 + }, + { + "epoch": 0.5593575609092146, + "grad_norm": 2.23405122756958, + "learning_rate": 8.824914815319614e-06, + "loss": 2.4610925674438477, + "step": 205480 + }, + { + "epoch": 0.5594120048999591, + "grad_norm": 2.3170275688171387, + "learning_rate": 8.823824451410658e-06, + "loss": 2.515785026550293, + "step": 205500 + }, + { + "epoch": 0.5594664488907037, + "grad_norm": 2.2744686603546143, + "learning_rate": 8.822734087501704e-06, + "loss": 2.5096216201782227, + "step": 205520 + }, + { + "epoch": 0.5595208928814482, + "grad_norm": 2.1531765460968018, + "learning_rate": 8.82164372359275e-06, + "loss": 2.4997802734375, + "step": 205540 + }, + { + "epoch": 0.5595753368721927, + "grad_norm": 2.1029765605926514, + "learning_rate": 8.820553359683796e-06, + "loss": 2.4660953521728515, + "step": 205560 + }, + { + "epoch": 0.5596297808629372, + "grad_norm": 2.1569559574127197, + "learning_rate": 8.81946299577484e-06, + "loss": 2.4817060470581054, + "step": 205580 + }, + { + "epoch": 0.5596842248536817, + "grad_norm": 2.1586556434631348, + "learning_rate": 8.818372631865886e-06, + "loss": 2.4695234298706055, + "step": 205600 + }, + { + "epoch": 0.5597386688444262, + "grad_norm": 2.289233684539795, + "learning_rate": 8.817282267956932e-06, + "loss": 2.47241268157959, + "step": 205620 + }, + { + "epoch": 0.5597931128351709, + "grad_norm": 2.134225606918335, + "learning_rate": 8.816191904047977e-06, + "loss": 2.477123260498047, + "step": 205640 + }, + { + "epoch": 0.5598475568259154, + "grad_norm": 2.1719069480895996, + "learning_rate": 8.815101540139022e-06, + "loss": 2.4634634017944337, + "step": 205660 + }, + { + "epoch": 0.5599020008166599, + "grad_norm": 2.1222989559173584, + "learning_rate": 8.814011176230067e-06, + "loss": 2.449661445617676, + "step": 205680 + }, + { + "epoch": 0.5599564448074044, + "grad_norm": 2.2420403957366943, + "learning_rate": 8.812920812321113e-06, + "loss": 2.459610176086426, + "step": 205700 + }, + { + "epoch": 0.5600108887981489, + "grad_norm": 2.2857096195220947, + "learning_rate": 8.811830448412159e-06, + "loss": 2.4685409545898436, + "step": 205720 + }, + { + "epoch": 0.5600653327888935, + "grad_norm": 2.0747532844543457, + "learning_rate": 8.810740084503203e-06, + "loss": 2.4914524078369142, + "step": 205740 + }, + { + "epoch": 0.560119776779638, + "grad_norm": 2.2144320011138916, + "learning_rate": 8.809649720594249e-06, + "loss": 2.4976600646972655, + "step": 205760 + }, + { + "epoch": 0.5601742207703825, + "grad_norm": 2.046192169189453, + "learning_rate": 8.808559356685295e-06, + "loss": 2.4926429748535157, + "step": 205780 + }, + { + "epoch": 0.560228664761127, + "grad_norm": 2.135226249694824, + "learning_rate": 8.80746899277634e-06, + "loss": 2.450931358337402, + "step": 205800 + }, + { + "epoch": 0.5602831087518715, + "grad_norm": 2.1319639682769775, + "learning_rate": 8.806378628867385e-06, + "loss": 2.4835638046264648, + "step": 205820 + }, + { + "epoch": 0.560337552742616, + "grad_norm": 2.2096729278564453, + "learning_rate": 8.80528826495843e-06, + "loss": 2.4726736068725588, + "step": 205840 + }, + { + "epoch": 0.5603919967333606, + "grad_norm": 2.231461763381958, + "learning_rate": 8.804197901049476e-06, + "loss": 2.472150993347168, + "step": 205860 + }, + { + "epoch": 0.5604464407241051, + "grad_norm": 2.672186851501465, + "learning_rate": 8.803107537140522e-06, + "loss": 2.466723823547363, + "step": 205880 + }, + { + "epoch": 0.5605008847148496, + "grad_norm": 2.2475156784057617, + "learning_rate": 8.802017173231566e-06, + "loss": 2.4780323028564455, + "step": 205900 + }, + { + "epoch": 0.5605553287055941, + "grad_norm": 2.126171350479126, + "learning_rate": 8.800926809322612e-06, + "loss": 2.4617288589477537, + "step": 205920 + }, + { + "epoch": 0.5606097726963386, + "grad_norm": 2.2286901473999023, + "learning_rate": 8.799836445413658e-06, + "loss": 2.496141815185547, + "step": 205940 + }, + { + "epoch": 0.5606642166870832, + "grad_norm": 2.090010643005371, + "learning_rate": 8.798746081504704e-06, + "loss": 2.4580812454223633, + "step": 205960 + }, + { + "epoch": 0.5607186606778277, + "grad_norm": 2.1471166610717773, + "learning_rate": 8.797655717595748e-06, + "loss": 2.477531623840332, + "step": 205980 + }, + { + "epoch": 0.5607731046685722, + "grad_norm": 2.412814140319824, + "learning_rate": 8.796565353686794e-06, + "loss": 2.464201736450195, + "step": 206000 + }, + { + "epoch": 0.5608275486593167, + "grad_norm": 2.1536805629730225, + "learning_rate": 8.79547498977784e-06, + "loss": 2.4753438949584963, + "step": 206020 + }, + { + "epoch": 0.5608819926500612, + "grad_norm": 2.478691339492798, + "learning_rate": 8.794384625868885e-06, + "loss": 2.4331085205078127, + "step": 206040 + }, + { + "epoch": 0.5609364366408057, + "grad_norm": 2.173203945159912, + "learning_rate": 8.79329426195993e-06, + "loss": 2.4969413757324217, + "step": 206060 + }, + { + "epoch": 0.5609908806315503, + "grad_norm": 2.187009572982788, + "learning_rate": 8.792203898050975e-06, + "loss": 2.4884407043457033, + "step": 206080 + }, + { + "epoch": 0.5610453246222948, + "grad_norm": 2.0752978324890137, + "learning_rate": 8.791113534142021e-06, + "loss": 2.4638572692871095, + "step": 206100 + }, + { + "epoch": 0.5610997686130393, + "grad_norm": 4.087273120880127, + "learning_rate": 8.790023170233067e-06, + "loss": 2.458195114135742, + "step": 206120 + }, + { + "epoch": 0.5611542126037838, + "grad_norm": 2.1505110263824463, + "learning_rate": 8.788932806324113e-06, + "loss": 2.497139549255371, + "step": 206140 + }, + { + "epoch": 0.5612086565945283, + "grad_norm": 2.244908094406128, + "learning_rate": 8.787842442415157e-06, + "loss": 2.4912948608398438, + "step": 206160 + }, + { + "epoch": 0.561263100585273, + "grad_norm": 2.131310224533081, + "learning_rate": 8.786752078506202e-06, + "loss": 2.4554534912109376, + "step": 206180 + }, + { + "epoch": 0.5613175445760175, + "grad_norm": 2.2003159523010254, + "learning_rate": 8.785661714597248e-06, + "loss": 2.4532398223876952, + "step": 206200 + }, + { + "epoch": 0.561371988566762, + "grad_norm": 2.235668659210205, + "learning_rate": 8.784571350688294e-06, + "loss": 2.460346221923828, + "step": 206220 + }, + { + "epoch": 0.5614264325575065, + "grad_norm": 2.208418846130371, + "learning_rate": 8.783480986779338e-06, + "loss": 2.487508201599121, + "step": 206240 + }, + { + "epoch": 0.561480876548251, + "grad_norm": 2.260146141052246, + "learning_rate": 8.782390622870384e-06, + "loss": 2.441777801513672, + "step": 206260 + }, + { + "epoch": 0.5615353205389955, + "grad_norm": 2.153214693069458, + "learning_rate": 8.78130025896143e-06, + "loss": 2.4759788513183594, + "step": 206280 + }, + { + "epoch": 0.5615897645297401, + "grad_norm": 2.238044023513794, + "learning_rate": 8.780209895052476e-06, + "loss": 2.4689327239990235, + "step": 206300 + }, + { + "epoch": 0.5616442085204846, + "grad_norm": 2.88576340675354, + "learning_rate": 8.77911953114352e-06, + "loss": 2.5102056503295898, + "step": 206320 + }, + { + "epoch": 0.5616986525112291, + "grad_norm": 2.3563597202301025, + "learning_rate": 8.778029167234566e-06, + "loss": 2.455719566345215, + "step": 206340 + }, + { + "epoch": 0.5617530965019736, + "grad_norm": 2.2329049110412598, + "learning_rate": 8.776938803325611e-06, + "loss": 2.4822803497314454, + "step": 206360 + }, + { + "epoch": 0.5618075404927181, + "grad_norm": 2.1653072834014893, + "learning_rate": 8.775848439416657e-06, + "loss": 2.4923261642456054, + "step": 206380 + }, + { + "epoch": 0.5618619844834626, + "grad_norm": 2.215909242630005, + "learning_rate": 8.774758075507701e-06, + "loss": 2.491739273071289, + "step": 206400 + }, + { + "epoch": 0.5619164284742072, + "grad_norm": 2.2836194038391113, + "learning_rate": 8.773667711598747e-06, + "loss": 2.5060325622558595, + "step": 206420 + }, + { + "epoch": 0.5619708724649517, + "grad_norm": 2.4153504371643066, + "learning_rate": 8.772577347689793e-06, + "loss": 2.4771284103393554, + "step": 206440 + }, + { + "epoch": 0.5620253164556962, + "grad_norm": 2.145934581756592, + "learning_rate": 8.771486983780839e-06, + "loss": 2.500415229797363, + "step": 206460 + }, + { + "epoch": 0.5620797604464407, + "grad_norm": 2.1109778881073, + "learning_rate": 8.770396619871883e-06, + "loss": 2.4716136932373045, + "step": 206480 + }, + { + "epoch": 0.5621342044371852, + "grad_norm": 2.194415807723999, + "learning_rate": 8.769306255962929e-06, + "loss": 2.4805070877075197, + "step": 206500 + }, + { + "epoch": 0.5621886484279298, + "grad_norm": 2.1283271312713623, + "learning_rate": 8.768215892053974e-06, + "loss": 2.4781749725341795, + "step": 206520 + }, + { + "epoch": 0.5622430924186743, + "grad_norm": 2.202096462249756, + "learning_rate": 8.76712552814502e-06, + "loss": 2.4803350448608397, + "step": 206540 + }, + { + "epoch": 0.5622975364094188, + "grad_norm": 2.1576881408691406, + "learning_rate": 8.766035164236064e-06, + "loss": 2.4693883895874023, + "step": 206560 + }, + { + "epoch": 0.5623519804001633, + "grad_norm": 2.1481363773345947, + "learning_rate": 8.76494480032711e-06, + "loss": 2.4780040740966798, + "step": 206580 + }, + { + "epoch": 0.5624064243909078, + "grad_norm": 2.117121934890747, + "learning_rate": 8.763854436418156e-06, + "loss": 2.497404861450195, + "step": 206600 + }, + { + "epoch": 0.5624608683816523, + "grad_norm": 2.2246148586273193, + "learning_rate": 8.762764072509202e-06, + "loss": 2.471072959899902, + "step": 206620 + }, + { + "epoch": 0.5625153123723969, + "grad_norm": 2.2254154682159424, + "learning_rate": 8.761673708600246e-06, + "loss": 2.4751258850097657, + "step": 206640 + }, + { + "epoch": 0.5625697563631414, + "grad_norm": 2.6604602336883545, + "learning_rate": 8.760583344691292e-06, + "loss": 2.460069465637207, + "step": 206660 + }, + { + "epoch": 0.5626242003538859, + "grad_norm": 2.245602607727051, + "learning_rate": 8.759492980782338e-06, + "loss": 2.4575931549072267, + "step": 206680 + }, + { + "epoch": 0.5626786443446304, + "grad_norm": 2.149832248687744, + "learning_rate": 8.758402616873383e-06, + "loss": 2.4737104415893554, + "step": 206700 + }, + { + "epoch": 0.5627330883353749, + "grad_norm": 5.278063774108887, + "learning_rate": 8.757312252964427e-06, + "loss": 2.4671730041503905, + "step": 206720 + }, + { + "epoch": 0.5627875323261196, + "grad_norm": 2.230053663253784, + "learning_rate": 8.756221889055473e-06, + "loss": 2.4581539154052736, + "step": 206740 + }, + { + "epoch": 0.5628419763168641, + "grad_norm": 2.1338062286376953, + "learning_rate": 8.755131525146519e-06, + "loss": 2.447794532775879, + "step": 206760 + }, + { + "epoch": 0.5628964203076086, + "grad_norm": 2.1876628398895264, + "learning_rate": 8.754041161237565e-06, + "loss": 2.498048210144043, + "step": 206780 + }, + { + "epoch": 0.5629508642983531, + "grad_norm": 2.1652610301971436, + "learning_rate": 8.752950797328609e-06, + "loss": 2.4609912872314452, + "step": 206800 + }, + { + "epoch": 0.5630053082890976, + "grad_norm": 2.1414601802825928, + "learning_rate": 8.751860433419655e-06, + "loss": 2.4545434951782226, + "step": 206820 + }, + { + "epoch": 0.5630597522798421, + "grad_norm": 2.1656243801116943, + "learning_rate": 8.7507700695107e-06, + "loss": 2.5104564666748046, + "step": 206840 + }, + { + "epoch": 0.5631141962705867, + "grad_norm": 2.3228237628936768, + "learning_rate": 8.749679705601746e-06, + "loss": 2.510831832885742, + "step": 206860 + }, + { + "epoch": 0.5631686402613312, + "grad_norm": 2.217541217803955, + "learning_rate": 8.74858934169279e-06, + "loss": 2.4560052871704103, + "step": 206880 + }, + { + "epoch": 0.5632230842520757, + "grad_norm": 2.1427218914031982, + "learning_rate": 8.747498977783836e-06, + "loss": 2.4750656127929687, + "step": 206900 + }, + { + "epoch": 0.5632775282428202, + "grad_norm": 2.0505728721618652, + "learning_rate": 8.746408613874882e-06, + "loss": 2.482983207702637, + "step": 206920 + }, + { + "epoch": 0.5633319722335647, + "grad_norm": 2.153841972351074, + "learning_rate": 8.745318249965928e-06, + "loss": 2.5003292083740236, + "step": 206940 + }, + { + "epoch": 0.5633864162243093, + "grad_norm": 2.1749205589294434, + "learning_rate": 8.744227886056972e-06, + "loss": 2.4994281768798827, + "step": 206960 + }, + { + "epoch": 0.5634408602150538, + "grad_norm": 2.1417813301086426, + "learning_rate": 8.743137522148018e-06, + "loss": 2.47159423828125, + "step": 206980 + }, + { + "epoch": 0.5634953042057983, + "grad_norm": 2.093587636947632, + "learning_rate": 8.742047158239064e-06, + "loss": 2.478391647338867, + "step": 207000 + }, + { + "epoch": 0.5635497481965428, + "grad_norm": 2.1214818954467773, + "learning_rate": 8.74095679433011e-06, + "loss": 2.4671497344970703, + "step": 207020 + }, + { + "epoch": 0.5636041921872873, + "grad_norm": 2.1347644329071045, + "learning_rate": 8.739866430421154e-06, + "loss": 2.4745786666870115, + "step": 207040 + }, + { + "epoch": 0.5636586361780318, + "grad_norm": 2.1566989421844482, + "learning_rate": 8.7387760665122e-06, + "loss": 2.5057090759277343, + "step": 207060 + }, + { + "epoch": 0.5637130801687764, + "grad_norm": 2.214787483215332, + "learning_rate": 8.737685702603245e-06, + "loss": 2.488943099975586, + "step": 207080 + }, + { + "epoch": 0.5637675241595209, + "grad_norm": 2.0887279510498047, + "learning_rate": 8.736595338694291e-06, + "loss": 2.493874168395996, + "step": 207100 + }, + { + "epoch": 0.5638219681502654, + "grad_norm": 2.102240562438965, + "learning_rate": 8.735504974785335e-06, + "loss": 2.471295166015625, + "step": 207120 + }, + { + "epoch": 0.5638764121410099, + "grad_norm": 2.285106897354126, + "learning_rate": 8.734414610876381e-06, + "loss": 2.463081169128418, + "step": 207140 + }, + { + "epoch": 0.5639308561317544, + "grad_norm": 2.087507963180542, + "learning_rate": 8.733324246967427e-06, + "loss": 2.4965888977050783, + "step": 207160 + }, + { + "epoch": 0.563985300122499, + "grad_norm": 2.1140899658203125, + "learning_rate": 8.732233883058473e-06, + "loss": 2.462085151672363, + "step": 207180 + }, + { + "epoch": 0.5640397441132435, + "grad_norm": 2.2767064571380615, + "learning_rate": 8.731143519149517e-06, + "loss": 2.4759572982788085, + "step": 207200 + }, + { + "epoch": 0.564094188103988, + "grad_norm": 2.3046939373016357, + "learning_rate": 8.730053155240563e-06, + "loss": 2.477713394165039, + "step": 207220 + }, + { + "epoch": 0.5641486320947325, + "grad_norm": 2.376112937927246, + "learning_rate": 8.728962791331608e-06, + "loss": 2.513058662414551, + "step": 207240 + }, + { + "epoch": 0.564203076085477, + "grad_norm": 2.1690561771392822, + "learning_rate": 8.727872427422654e-06, + "loss": 2.5089651107788087, + "step": 207260 + }, + { + "epoch": 0.5642575200762215, + "grad_norm": 2.162177801132202, + "learning_rate": 8.726782063513698e-06, + "loss": 2.459664726257324, + "step": 207280 + }, + { + "epoch": 0.5643119640669662, + "grad_norm": 2.0947225093841553, + "learning_rate": 8.725691699604744e-06, + "loss": 2.4855924606323243, + "step": 207300 + }, + { + "epoch": 0.5643664080577107, + "grad_norm": 2.1063358783721924, + "learning_rate": 8.72460133569579e-06, + "loss": 2.474051856994629, + "step": 207320 + }, + { + "epoch": 0.5644208520484552, + "grad_norm": 2.3260657787323, + "learning_rate": 8.723510971786836e-06, + "loss": 2.526301383972168, + "step": 207340 + }, + { + "epoch": 0.5644752960391997, + "grad_norm": 2.224581003189087, + "learning_rate": 8.72242060787788e-06, + "loss": 2.46478328704834, + "step": 207360 + }, + { + "epoch": 0.5645297400299442, + "grad_norm": 2.161876678466797, + "learning_rate": 8.721330243968926e-06, + "loss": 2.4627052307128907, + "step": 207380 + }, + { + "epoch": 0.5645841840206887, + "grad_norm": 2.084200859069824, + "learning_rate": 8.720239880059971e-06, + "loss": 2.4584985733032227, + "step": 207400 + }, + { + "epoch": 0.5646386280114333, + "grad_norm": 2.2089788913726807, + "learning_rate": 8.719149516151017e-06, + "loss": 2.506117820739746, + "step": 207420 + }, + { + "epoch": 0.5646930720021778, + "grad_norm": 2.2427542209625244, + "learning_rate": 8.718059152242061e-06, + "loss": 2.462841033935547, + "step": 207440 + }, + { + "epoch": 0.5647475159929223, + "grad_norm": 2.1556553840637207, + "learning_rate": 8.716968788333107e-06, + "loss": 2.492728424072266, + "step": 207460 + }, + { + "epoch": 0.5648019599836668, + "grad_norm": 2.131216287612915, + "learning_rate": 8.715878424424153e-06, + "loss": 2.479152297973633, + "step": 207480 + }, + { + "epoch": 0.5648564039744113, + "grad_norm": 2.0753109455108643, + "learning_rate": 8.714788060515199e-06, + "loss": 2.46951847076416, + "step": 207500 + }, + { + "epoch": 0.5649108479651559, + "grad_norm": 2.126720666885376, + "learning_rate": 8.713697696606243e-06, + "loss": 2.4647804260253907, + "step": 207520 + }, + { + "epoch": 0.5649652919559004, + "grad_norm": 2.2952945232391357, + "learning_rate": 8.712607332697289e-06, + "loss": 2.49340934753418, + "step": 207540 + }, + { + "epoch": 0.5650197359466449, + "grad_norm": 2.4483301639556885, + "learning_rate": 8.711516968788334e-06, + "loss": 2.4923080444335937, + "step": 207560 + }, + { + "epoch": 0.5650741799373894, + "grad_norm": 2.1914336681365967, + "learning_rate": 8.71042660487938e-06, + "loss": 2.4767173767089843, + "step": 207580 + }, + { + "epoch": 0.5651286239281339, + "grad_norm": 2.209388494491577, + "learning_rate": 8.709336240970424e-06, + "loss": 2.4602497100830076, + "step": 207600 + }, + { + "epoch": 0.5651830679188784, + "grad_norm": 2.0813534259796143, + "learning_rate": 8.70824587706147e-06, + "loss": 2.4794063568115234, + "step": 207620 + }, + { + "epoch": 0.565237511909623, + "grad_norm": 2.272142171859741, + "learning_rate": 8.707155513152516e-06, + "loss": 2.495387649536133, + "step": 207640 + }, + { + "epoch": 0.5652919559003675, + "grad_norm": 2.063163995742798, + "learning_rate": 8.706065149243562e-06, + "loss": 2.4623489379882812, + "step": 207660 + }, + { + "epoch": 0.565346399891112, + "grad_norm": 2.3643102645874023, + "learning_rate": 8.704974785334606e-06, + "loss": 2.489810562133789, + "step": 207680 + }, + { + "epoch": 0.5654008438818565, + "grad_norm": 2.081181526184082, + "learning_rate": 8.703884421425652e-06, + "loss": 2.4875467300415037, + "step": 207700 + }, + { + "epoch": 0.565455287872601, + "grad_norm": 2.076930522918701, + "learning_rate": 8.702794057516698e-06, + "loss": 2.4720203399658205, + "step": 207720 + }, + { + "epoch": 0.5655097318633456, + "grad_norm": 2.218766212463379, + "learning_rate": 8.701703693607743e-06, + "loss": 2.490492248535156, + "step": 207740 + }, + { + "epoch": 0.5655641758540901, + "grad_norm": 2.560940742492676, + "learning_rate": 8.700613329698788e-06, + "loss": 2.47460994720459, + "step": 207760 + }, + { + "epoch": 0.5656186198448346, + "grad_norm": 2.240870952606201, + "learning_rate": 8.699522965789833e-06, + "loss": 2.488620948791504, + "step": 207780 + }, + { + "epoch": 0.5656730638355791, + "grad_norm": 2.099425792694092, + "learning_rate": 8.698432601880879e-06, + "loss": 2.4515283584594725, + "step": 207800 + }, + { + "epoch": 0.5657275078263236, + "grad_norm": 2.1570370197296143, + "learning_rate": 8.697342237971925e-06, + "loss": 2.455870819091797, + "step": 207820 + }, + { + "epoch": 0.5657819518170681, + "grad_norm": 2.226599931716919, + "learning_rate": 8.696251874062969e-06, + "loss": 2.506656837463379, + "step": 207840 + }, + { + "epoch": 0.5658363958078128, + "grad_norm": 2.982893228530884, + "learning_rate": 8.695161510154015e-06, + "loss": 2.514307403564453, + "step": 207860 + }, + { + "epoch": 0.5658908397985573, + "grad_norm": 2.2485361099243164, + "learning_rate": 8.69407114624506e-06, + "loss": 2.493879699707031, + "step": 207880 + }, + { + "epoch": 0.5659452837893018, + "grad_norm": 2.105914354324341, + "learning_rate": 8.692980782336106e-06, + "loss": 2.4520145416259767, + "step": 207900 + }, + { + "epoch": 0.5659997277800463, + "grad_norm": 2.209378480911255, + "learning_rate": 8.69189041842715e-06, + "loss": 2.487746810913086, + "step": 207920 + }, + { + "epoch": 0.5660541717707908, + "grad_norm": 2.061560869216919, + "learning_rate": 8.690800054518196e-06, + "loss": 2.4602981567382813, + "step": 207940 + }, + { + "epoch": 0.5661086157615354, + "grad_norm": 2.151791572570801, + "learning_rate": 8.689709690609242e-06, + "loss": 2.457431602478027, + "step": 207960 + }, + { + "epoch": 0.5661630597522799, + "grad_norm": 2.1359686851501465, + "learning_rate": 8.688619326700288e-06, + "loss": 2.468695068359375, + "step": 207980 + }, + { + "epoch": 0.5662175037430244, + "grad_norm": 2.0979671478271484, + "learning_rate": 8.687528962791332e-06, + "loss": 2.4582359313964846, + "step": 208000 + }, + { + "epoch": 0.5662719477337689, + "grad_norm": 2.3038923740386963, + "learning_rate": 8.686438598882378e-06, + "loss": 2.4524341583251954, + "step": 208020 + }, + { + "epoch": 0.5663263917245134, + "grad_norm": 2.7280640602111816, + "learning_rate": 8.685348234973424e-06, + "loss": 2.461737060546875, + "step": 208040 + }, + { + "epoch": 0.5663808357152579, + "grad_norm": 2.2899887561798096, + "learning_rate": 8.68425787106447e-06, + "loss": 2.479755401611328, + "step": 208060 + }, + { + "epoch": 0.5664352797060025, + "grad_norm": 2.5293219089508057, + "learning_rate": 8.683167507155514e-06, + "loss": 2.4918710708618166, + "step": 208080 + }, + { + "epoch": 0.566489723696747, + "grad_norm": 2.092914581298828, + "learning_rate": 8.68207714324656e-06, + "loss": 2.480558395385742, + "step": 208100 + }, + { + "epoch": 0.5665441676874915, + "grad_norm": 2.131615400314331, + "learning_rate": 8.680986779337605e-06, + "loss": 2.502383804321289, + "step": 208120 + }, + { + "epoch": 0.566598611678236, + "grad_norm": 2.158344268798828, + "learning_rate": 8.67989641542865e-06, + "loss": 2.487959861755371, + "step": 208140 + }, + { + "epoch": 0.5666530556689805, + "grad_norm": 2.7311770915985107, + "learning_rate": 8.678806051519695e-06, + "loss": 2.487253189086914, + "step": 208160 + }, + { + "epoch": 0.5667074996597251, + "grad_norm": 2.1547529697418213, + "learning_rate": 8.677715687610741e-06, + "loss": 2.462836265563965, + "step": 208180 + }, + { + "epoch": 0.5667619436504696, + "grad_norm": 2.2153000831604004, + "learning_rate": 8.676625323701787e-06, + "loss": 2.483433723449707, + "step": 208200 + }, + { + "epoch": 0.5668163876412141, + "grad_norm": 2.240337371826172, + "learning_rate": 8.675534959792831e-06, + "loss": 2.4633834838867186, + "step": 208220 + }, + { + "epoch": 0.5668708316319586, + "grad_norm": 2.1724843978881836, + "learning_rate": 8.674444595883877e-06, + "loss": 2.4660518646240233, + "step": 208240 + }, + { + "epoch": 0.5669252756227031, + "grad_norm": 2.189091682434082, + "learning_rate": 8.673354231974923e-06, + "loss": 2.477117347717285, + "step": 208260 + }, + { + "epoch": 0.5669797196134476, + "grad_norm": 2.116997718811035, + "learning_rate": 8.672263868065968e-06, + "loss": 2.5055257797241213, + "step": 208280 + }, + { + "epoch": 0.5670341636041922, + "grad_norm": 2.133812189102173, + "learning_rate": 8.671173504157012e-06, + "loss": 2.474839210510254, + "step": 208300 + }, + { + "epoch": 0.5670886075949367, + "grad_norm": 2.092289924621582, + "learning_rate": 8.670083140248058e-06, + "loss": 2.49124698638916, + "step": 208320 + }, + { + "epoch": 0.5671430515856812, + "grad_norm": 2.216899871826172, + "learning_rate": 8.668992776339104e-06, + "loss": 2.5040401458740233, + "step": 208340 + }, + { + "epoch": 0.5671974955764257, + "grad_norm": 2.7304654121398926, + "learning_rate": 8.66790241243015e-06, + "loss": 2.48812255859375, + "step": 208360 + }, + { + "epoch": 0.5672519395671702, + "grad_norm": 2.2013542652130127, + "learning_rate": 8.666812048521194e-06, + "loss": 2.4977502822875977, + "step": 208380 + }, + { + "epoch": 0.5673063835579149, + "grad_norm": 2.5508244037628174, + "learning_rate": 8.66572168461224e-06, + "loss": 2.4657703399658204, + "step": 208400 + }, + { + "epoch": 0.5673608275486594, + "grad_norm": 2.0627248287200928, + "learning_rate": 8.664631320703286e-06, + "loss": 2.4737661361694334, + "step": 208420 + }, + { + "epoch": 0.5674152715394039, + "grad_norm": 2.1294105052948, + "learning_rate": 8.663540956794331e-06, + "loss": 2.4654890060424806, + "step": 208440 + }, + { + "epoch": 0.5674697155301484, + "grad_norm": 2.1313893795013428, + "learning_rate": 8.662450592885376e-06, + "loss": 2.460636520385742, + "step": 208460 + }, + { + "epoch": 0.5675241595208929, + "grad_norm": 2.103440761566162, + "learning_rate": 8.661360228976421e-06, + "loss": 2.479915428161621, + "step": 208480 + }, + { + "epoch": 0.5675786035116374, + "grad_norm": 2.2821738719940186, + "learning_rate": 8.660269865067467e-06, + "loss": 2.4798862457275392, + "step": 208500 + }, + { + "epoch": 0.567633047502382, + "grad_norm": 2.085094928741455, + "learning_rate": 8.659179501158513e-06, + "loss": 2.4843639373779296, + "step": 208520 + }, + { + "epoch": 0.5676874914931265, + "grad_norm": 2.228132486343384, + "learning_rate": 8.658089137249557e-06, + "loss": 2.479333686828613, + "step": 208540 + }, + { + "epoch": 0.567741935483871, + "grad_norm": 2.166077136993408, + "learning_rate": 8.656998773340603e-06, + "loss": 2.4517208099365235, + "step": 208560 + }, + { + "epoch": 0.5677963794746155, + "grad_norm": 2.190108299255371, + "learning_rate": 8.655908409431649e-06, + "loss": 2.464289093017578, + "step": 208580 + }, + { + "epoch": 0.56785082346536, + "grad_norm": 2.1110544204711914, + "learning_rate": 8.654818045522695e-06, + "loss": 2.491143989562988, + "step": 208600 + }, + { + "epoch": 0.5679052674561045, + "grad_norm": 2.0641047954559326, + "learning_rate": 8.653727681613739e-06, + "loss": 2.4856639862060548, + "step": 208620 + }, + { + "epoch": 0.5679597114468491, + "grad_norm": 2.3631060123443604, + "learning_rate": 8.652637317704784e-06, + "loss": 2.47772159576416, + "step": 208640 + }, + { + "epoch": 0.5680141554375936, + "grad_norm": 2.1855406761169434, + "learning_rate": 8.65154695379583e-06, + "loss": 2.477712631225586, + "step": 208660 + }, + { + "epoch": 0.5680685994283381, + "grad_norm": 2.2434170246124268, + "learning_rate": 8.650456589886876e-06, + "loss": 2.505177307128906, + "step": 208680 + }, + { + "epoch": 0.5681230434190826, + "grad_norm": 2.302035093307495, + "learning_rate": 8.64936622597792e-06, + "loss": 2.4618976593017576, + "step": 208700 + }, + { + "epoch": 0.5681774874098271, + "grad_norm": 2.1419336795806885, + "learning_rate": 8.648275862068966e-06, + "loss": 2.4729545593261717, + "step": 208720 + }, + { + "epoch": 0.5682319314005717, + "grad_norm": 2.1600558757781982, + "learning_rate": 8.647185498160012e-06, + "loss": 2.490397071838379, + "step": 208740 + }, + { + "epoch": 0.5682863753913162, + "grad_norm": 2.2116026878356934, + "learning_rate": 8.646095134251058e-06, + "loss": 2.481294631958008, + "step": 208760 + }, + { + "epoch": 0.5683408193820607, + "grad_norm": 2.287061929702759, + "learning_rate": 8.645004770342102e-06, + "loss": 2.504364013671875, + "step": 208780 + }, + { + "epoch": 0.5683952633728052, + "grad_norm": 2.1778101921081543, + "learning_rate": 8.643914406433148e-06, + "loss": 2.477887725830078, + "step": 208800 + }, + { + "epoch": 0.5684497073635497, + "grad_norm": 2.153118371963501, + "learning_rate": 8.642824042524193e-06, + "loss": 2.5021455764770506, + "step": 208820 + }, + { + "epoch": 0.5685041513542942, + "grad_norm": 2.121647596359253, + "learning_rate": 8.64173367861524e-06, + "loss": 2.467807960510254, + "step": 208840 + }, + { + "epoch": 0.5685585953450388, + "grad_norm": 2.0789692401885986, + "learning_rate": 8.640643314706283e-06, + "loss": 2.5169002532958986, + "step": 208860 + }, + { + "epoch": 0.5686130393357833, + "grad_norm": 2.11728835105896, + "learning_rate": 8.639552950797329e-06, + "loss": 2.4213188171386717, + "step": 208880 + }, + { + "epoch": 0.5686674833265278, + "grad_norm": 2.793760299682617, + "learning_rate": 8.638462586888375e-06, + "loss": 2.4726768493652345, + "step": 208900 + }, + { + "epoch": 0.5687219273172723, + "grad_norm": 2.1686782836914062, + "learning_rate": 8.63737222297942e-06, + "loss": 2.459234619140625, + "step": 208920 + }, + { + "epoch": 0.5687763713080168, + "grad_norm": 2.054892063140869, + "learning_rate": 8.636281859070465e-06, + "loss": 2.478021812438965, + "step": 208940 + }, + { + "epoch": 0.5688308152987614, + "grad_norm": 2.149897575378418, + "learning_rate": 8.63519149516151e-06, + "loss": 2.4780149459838867, + "step": 208960 + }, + { + "epoch": 0.568885259289506, + "grad_norm": 2.487902879714966, + "learning_rate": 8.634101131252556e-06, + "loss": 2.4571746826171874, + "step": 208980 + }, + { + "epoch": 0.5689397032802505, + "grad_norm": 2.1845245361328125, + "learning_rate": 8.633010767343602e-06, + "loss": 2.489995574951172, + "step": 209000 + }, + { + "epoch": 0.568994147270995, + "grad_norm": 2.194610118865967, + "learning_rate": 8.631920403434646e-06, + "loss": 2.4852922439575194, + "step": 209020 + }, + { + "epoch": 0.5690485912617395, + "grad_norm": 2.193457841873169, + "learning_rate": 8.630830039525692e-06, + "loss": 2.5087236404418944, + "step": 209040 + }, + { + "epoch": 0.569103035252484, + "grad_norm": 2.2179365158081055, + "learning_rate": 8.629739675616738e-06, + "loss": 2.497519302368164, + "step": 209060 + }, + { + "epoch": 0.5691574792432286, + "grad_norm": 2.085315227508545, + "learning_rate": 8.628649311707784e-06, + "loss": 2.499690818786621, + "step": 209080 + }, + { + "epoch": 0.5692119232339731, + "grad_norm": 2.1837620735168457, + "learning_rate": 8.627558947798828e-06, + "loss": 2.508159065246582, + "step": 209100 + }, + { + "epoch": 0.5692663672247176, + "grad_norm": 2.112093687057495, + "learning_rate": 8.626468583889874e-06, + "loss": 2.4664344787597656, + "step": 209120 + }, + { + "epoch": 0.5693208112154621, + "grad_norm": 2.140378713607788, + "learning_rate": 8.62537821998092e-06, + "loss": 2.4812578201293944, + "step": 209140 + }, + { + "epoch": 0.5693752552062066, + "grad_norm": 2.133849859237671, + "learning_rate": 8.624287856071965e-06, + "loss": 2.4807096481323243, + "step": 209160 + }, + { + "epoch": 0.5694296991969512, + "grad_norm": 2.1517364978790283, + "learning_rate": 8.62319749216301e-06, + "loss": 2.46288948059082, + "step": 209180 + }, + { + "epoch": 0.5694841431876957, + "grad_norm": 2.2738678455352783, + "learning_rate": 8.622107128254055e-06, + "loss": 2.487552452087402, + "step": 209200 + }, + { + "epoch": 0.5695385871784402, + "grad_norm": 2.1779367923736572, + "learning_rate": 8.621016764345101e-06, + "loss": 2.5077098846435546, + "step": 209220 + }, + { + "epoch": 0.5695930311691847, + "grad_norm": 2.168259859085083, + "learning_rate": 8.619926400436145e-06, + "loss": 2.452815628051758, + "step": 209240 + }, + { + "epoch": 0.5696474751599292, + "grad_norm": 2.0523252487182617, + "learning_rate": 8.618836036527191e-06, + "loss": 2.44683837890625, + "step": 209260 + }, + { + "epoch": 0.5697019191506737, + "grad_norm": 2.1475701332092285, + "learning_rate": 8.617745672618237e-06, + "loss": 2.4850957870483397, + "step": 209280 + }, + { + "epoch": 0.5697563631414183, + "grad_norm": 2.1458330154418945, + "learning_rate": 8.616655308709283e-06, + "loss": 2.471235656738281, + "step": 209300 + }, + { + "epoch": 0.5698108071321628, + "grad_norm": 2.11899733543396, + "learning_rate": 8.615564944800327e-06, + "loss": 2.472775459289551, + "step": 209320 + }, + { + "epoch": 0.5698652511229073, + "grad_norm": 2.3194797039031982, + "learning_rate": 8.614474580891373e-06, + "loss": 2.4408058166503905, + "step": 209340 + }, + { + "epoch": 0.5699196951136518, + "grad_norm": 2.059519052505493, + "learning_rate": 8.613384216982418e-06, + "loss": 2.476716423034668, + "step": 209360 + }, + { + "epoch": 0.5699741391043963, + "grad_norm": 2.141324758529663, + "learning_rate": 8.612293853073464e-06, + "loss": 2.4684770584106444, + "step": 209380 + }, + { + "epoch": 0.5700285830951409, + "grad_norm": 2.2004852294921875, + "learning_rate": 8.611203489164508e-06, + "loss": 2.4473493576049803, + "step": 209400 + }, + { + "epoch": 0.5700830270858854, + "grad_norm": 2.145501136779785, + "learning_rate": 8.610113125255554e-06, + "loss": 2.484471893310547, + "step": 209420 + }, + { + "epoch": 0.5701374710766299, + "grad_norm": 2.3181252479553223, + "learning_rate": 8.6090227613466e-06, + "loss": 2.4597646713256838, + "step": 209440 + }, + { + "epoch": 0.5701919150673744, + "grad_norm": 2.044562816619873, + "learning_rate": 8.607932397437646e-06, + "loss": 2.485958290100098, + "step": 209460 + }, + { + "epoch": 0.5702463590581189, + "grad_norm": 2.166288375854492, + "learning_rate": 8.60684203352869e-06, + "loss": 2.475512886047363, + "step": 209480 + }, + { + "epoch": 0.5703008030488634, + "grad_norm": 2.229349136352539, + "learning_rate": 8.605751669619736e-06, + "loss": 2.4412689208984375, + "step": 209500 + }, + { + "epoch": 0.570355247039608, + "grad_norm": 2.126575231552124, + "learning_rate": 8.604661305710781e-06, + "loss": 2.46612548828125, + "step": 209520 + }, + { + "epoch": 0.5704096910303526, + "grad_norm": 2.181417226791382, + "learning_rate": 8.603570941801827e-06, + "loss": 2.4695707321166993, + "step": 209540 + }, + { + "epoch": 0.570464135021097, + "grad_norm": 2.1410980224609375, + "learning_rate": 8.602480577892871e-06, + "loss": 2.453866195678711, + "step": 209560 + }, + { + "epoch": 0.5705185790118416, + "grad_norm": 2.1002633571624756, + "learning_rate": 8.601390213983917e-06, + "loss": 2.4857126235961915, + "step": 209580 + }, + { + "epoch": 0.570573023002586, + "grad_norm": 2.354419469833374, + "learning_rate": 8.600299850074963e-06, + "loss": 2.4502639770507812, + "step": 209600 + }, + { + "epoch": 0.5706274669933306, + "grad_norm": 2.2879703044891357, + "learning_rate": 8.599209486166009e-06, + "loss": 2.4529701232910157, + "step": 209620 + }, + { + "epoch": 0.5706819109840752, + "grad_norm": 2.110015630722046, + "learning_rate": 8.598119122257053e-06, + "loss": 2.469868850708008, + "step": 209640 + }, + { + "epoch": 0.5707363549748197, + "grad_norm": 2.2565250396728516, + "learning_rate": 8.597028758348099e-06, + "loss": 2.475502777099609, + "step": 209660 + }, + { + "epoch": 0.5707907989655642, + "grad_norm": 2.039788246154785, + "learning_rate": 8.595938394439145e-06, + "loss": 2.4693475723266602, + "step": 209680 + }, + { + "epoch": 0.5708452429563087, + "grad_norm": 2.2144670486450195, + "learning_rate": 8.59484803053019e-06, + "loss": 2.515750503540039, + "step": 209700 + }, + { + "epoch": 0.5708996869470532, + "grad_norm": 2.1699211597442627, + "learning_rate": 8.593757666621234e-06, + "loss": 2.4645978927612306, + "step": 209720 + }, + { + "epoch": 0.5709541309377978, + "grad_norm": 2.2320024967193604, + "learning_rate": 8.59266730271228e-06, + "loss": 2.4571630477905275, + "step": 209740 + }, + { + "epoch": 0.5710085749285423, + "grad_norm": 2.1901800632476807, + "learning_rate": 8.591576938803326e-06, + "loss": 2.468727874755859, + "step": 209760 + }, + { + "epoch": 0.5710630189192868, + "grad_norm": 2.27870512008667, + "learning_rate": 8.590486574894372e-06, + "loss": 2.4634456634521484, + "step": 209780 + }, + { + "epoch": 0.5711174629100313, + "grad_norm": 3.7863926887512207, + "learning_rate": 8.589396210985416e-06, + "loss": 2.483007049560547, + "step": 209800 + }, + { + "epoch": 0.5711719069007758, + "grad_norm": 2.073983907699585, + "learning_rate": 8.588305847076462e-06, + "loss": 2.476667594909668, + "step": 209820 + }, + { + "epoch": 0.5712263508915203, + "grad_norm": 2.1807291507720947, + "learning_rate": 8.587215483167508e-06, + "loss": 2.484781837463379, + "step": 209840 + }, + { + "epoch": 0.5712807948822649, + "grad_norm": 2.206986665725708, + "learning_rate": 8.586125119258553e-06, + "loss": 2.448569488525391, + "step": 209860 + }, + { + "epoch": 0.5713352388730094, + "grad_norm": 2.2765398025512695, + "learning_rate": 8.585034755349598e-06, + "loss": 2.4777345657348633, + "step": 209880 + }, + { + "epoch": 0.5713896828637539, + "grad_norm": 2.2322161197662354, + "learning_rate": 8.583944391440643e-06, + "loss": 2.454829788208008, + "step": 209900 + }, + { + "epoch": 0.5714441268544984, + "grad_norm": 2.0877013206481934, + "learning_rate": 8.58285402753169e-06, + "loss": 2.4652706146240235, + "step": 209920 + }, + { + "epoch": 0.5714985708452429, + "grad_norm": 2.295199155807495, + "learning_rate": 8.581763663622735e-06, + "loss": 2.479751396179199, + "step": 209940 + }, + { + "epoch": 0.5715530148359875, + "grad_norm": 2.1573667526245117, + "learning_rate": 8.580673299713779e-06, + "loss": 2.46435432434082, + "step": 209960 + }, + { + "epoch": 0.571607458826732, + "grad_norm": 2.17260479927063, + "learning_rate": 8.579582935804825e-06, + "loss": 2.4717229843139648, + "step": 209980 + }, + { + "epoch": 0.5716619028174765, + "grad_norm": 2.128824472427368, + "learning_rate": 8.57849257189587e-06, + "loss": 2.4712839126586914, + "step": 210000 + }, + { + "epoch": 0.571716346808221, + "grad_norm": 2.171499490737915, + "learning_rate": 8.577402207986917e-06, + "loss": 2.4654613494873048, + "step": 210020 + }, + { + "epoch": 0.5717707907989655, + "grad_norm": 2.1209092140197754, + "learning_rate": 8.57631184407796e-06, + "loss": 2.463038444519043, + "step": 210040 + }, + { + "epoch": 0.57182523478971, + "grad_norm": 2.2064311504364014, + "learning_rate": 8.575221480169006e-06, + "loss": 2.4781520843505858, + "step": 210060 + }, + { + "epoch": 0.5718796787804546, + "grad_norm": 2.369795322418213, + "learning_rate": 8.574131116260052e-06, + "loss": 2.49166259765625, + "step": 210080 + }, + { + "epoch": 0.5719341227711991, + "grad_norm": 2.1776111125946045, + "learning_rate": 8.573040752351098e-06, + "loss": 2.487049102783203, + "step": 210100 + }, + { + "epoch": 0.5719885667619437, + "grad_norm": 2.067626476287842, + "learning_rate": 8.571950388442142e-06, + "loss": 2.4588287353515623, + "step": 210120 + }, + { + "epoch": 0.5720430107526882, + "grad_norm": 2.216500997543335, + "learning_rate": 8.570860024533188e-06, + "loss": 2.4726661682128905, + "step": 210140 + }, + { + "epoch": 0.5720974547434327, + "grad_norm": 2.1165387630462646, + "learning_rate": 8.569769660624234e-06, + "loss": 2.4671390533447264, + "step": 210160 + }, + { + "epoch": 0.5721518987341773, + "grad_norm": 2.0705997943878174, + "learning_rate": 8.56867929671528e-06, + "loss": 2.4912261962890625, + "step": 210180 + }, + { + "epoch": 0.5722063427249218, + "grad_norm": 2.277181386947632, + "learning_rate": 8.567588932806324e-06, + "loss": 2.4362112045288087, + "step": 210200 + }, + { + "epoch": 0.5722607867156663, + "grad_norm": 2.133615493774414, + "learning_rate": 8.56649856889737e-06, + "loss": 2.491664695739746, + "step": 210220 + }, + { + "epoch": 0.5723152307064108, + "grad_norm": 2.2645046710968018, + "learning_rate": 8.565408204988415e-06, + "loss": 2.445372200012207, + "step": 210240 + }, + { + "epoch": 0.5723696746971553, + "grad_norm": 2.1891446113586426, + "learning_rate": 8.564317841079461e-06, + "loss": 2.4649871826171874, + "step": 210260 + }, + { + "epoch": 0.5724241186878998, + "grad_norm": 2.255675792694092, + "learning_rate": 8.563227477170505e-06, + "loss": 2.4829607009887695, + "step": 210280 + }, + { + "epoch": 0.5724785626786444, + "grad_norm": 2.0606284141540527, + "learning_rate": 8.562137113261551e-06, + "loss": 2.502083587646484, + "step": 210300 + }, + { + "epoch": 0.5725330066693889, + "grad_norm": 2.6207692623138428, + "learning_rate": 8.561046749352597e-06, + "loss": 2.5066593170166014, + "step": 210320 + }, + { + "epoch": 0.5725874506601334, + "grad_norm": 2.1000216007232666, + "learning_rate": 8.559956385443643e-06, + "loss": 2.4664146423339846, + "step": 210340 + }, + { + "epoch": 0.5726418946508779, + "grad_norm": 2.1639387607574463, + "learning_rate": 8.558866021534687e-06, + "loss": 2.5124322891235353, + "step": 210360 + }, + { + "epoch": 0.5726963386416224, + "grad_norm": 2.3571436405181885, + "learning_rate": 8.557775657625733e-06, + "loss": 2.4681381225585937, + "step": 210380 + }, + { + "epoch": 0.572750782632367, + "grad_norm": 2.1986355781555176, + "learning_rate": 8.556685293716778e-06, + "loss": 2.5144416809082033, + "step": 210400 + }, + { + "epoch": 0.5728052266231115, + "grad_norm": 2.167090654373169, + "learning_rate": 8.555594929807824e-06, + "loss": 2.491256904602051, + "step": 210420 + }, + { + "epoch": 0.572859670613856, + "grad_norm": 2.173896312713623, + "learning_rate": 8.554504565898868e-06, + "loss": 2.489066505432129, + "step": 210440 + }, + { + "epoch": 0.5729141146046005, + "grad_norm": 2.2456037998199463, + "learning_rate": 8.553414201989914e-06, + "loss": 2.4690717697143554, + "step": 210460 + }, + { + "epoch": 0.572968558595345, + "grad_norm": 2.146627187728882, + "learning_rate": 8.55232383808096e-06, + "loss": 2.485749053955078, + "step": 210480 + }, + { + "epoch": 0.5730230025860895, + "grad_norm": 2.4322361946105957, + "learning_rate": 8.551233474172006e-06, + "loss": 2.475505065917969, + "step": 210500 + }, + { + "epoch": 0.5730774465768341, + "grad_norm": 2.2690067291259766, + "learning_rate": 8.55014311026305e-06, + "loss": 2.4527820587158202, + "step": 210520 + }, + { + "epoch": 0.5731318905675786, + "grad_norm": 2.075814962387085, + "learning_rate": 8.549052746354096e-06, + "loss": 2.471669578552246, + "step": 210540 + }, + { + "epoch": 0.5731863345583231, + "grad_norm": 2.1100518703460693, + "learning_rate": 8.547962382445142e-06, + "loss": 2.4838935852050783, + "step": 210560 + }, + { + "epoch": 0.5732407785490676, + "grad_norm": 2.256120443344116, + "learning_rate": 8.546872018536187e-06, + "loss": 2.482301139831543, + "step": 210580 + }, + { + "epoch": 0.5732952225398121, + "grad_norm": 2.1860311031341553, + "learning_rate": 8.545781654627231e-06, + "loss": 2.475393295288086, + "step": 210600 + }, + { + "epoch": 0.5733496665305567, + "grad_norm": 2.108724594116211, + "learning_rate": 8.544691290718277e-06, + "loss": 2.4567438125610352, + "step": 210620 + }, + { + "epoch": 0.5734041105213012, + "grad_norm": 2.1232879161834717, + "learning_rate": 8.543600926809323e-06, + "loss": 2.4557174682617187, + "step": 210640 + }, + { + "epoch": 0.5734585545120457, + "grad_norm": 2.1230177879333496, + "learning_rate": 8.542510562900369e-06, + "loss": 2.483673667907715, + "step": 210660 + }, + { + "epoch": 0.5735129985027902, + "grad_norm": 2.09651517868042, + "learning_rate": 8.541420198991413e-06, + "loss": 2.482925224304199, + "step": 210680 + }, + { + "epoch": 0.5735674424935348, + "grad_norm": 2.1746811866760254, + "learning_rate": 8.540329835082459e-06, + "loss": 2.4453611373901367, + "step": 210700 + }, + { + "epoch": 0.5736218864842793, + "grad_norm": 2.208533525466919, + "learning_rate": 8.539239471173505e-06, + "loss": 2.4930404663085937, + "step": 210720 + }, + { + "epoch": 0.5736763304750239, + "grad_norm": 2.1712663173675537, + "learning_rate": 8.53814910726455e-06, + "loss": 2.4714990615844727, + "step": 210740 + }, + { + "epoch": 0.5737307744657684, + "grad_norm": 2.137685775756836, + "learning_rate": 8.537058743355595e-06, + "loss": 2.467709541320801, + "step": 210760 + }, + { + "epoch": 0.5737852184565129, + "grad_norm": 2.151916742324829, + "learning_rate": 8.53596837944664e-06, + "loss": 2.4951622009277346, + "step": 210780 + }, + { + "epoch": 0.5738396624472574, + "grad_norm": 2.2052130699157715, + "learning_rate": 8.534878015537686e-06, + "loss": 2.4585744857788088, + "step": 210800 + }, + { + "epoch": 0.5738941064380019, + "grad_norm": 2.2241439819335938, + "learning_rate": 8.533787651628732e-06, + "loss": 2.478644371032715, + "step": 210820 + }, + { + "epoch": 0.5739485504287464, + "grad_norm": 2.1628010272979736, + "learning_rate": 8.532697287719776e-06, + "loss": 2.482126808166504, + "step": 210840 + }, + { + "epoch": 0.574002994419491, + "grad_norm": 2.2673590183258057, + "learning_rate": 8.531606923810822e-06, + "loss": 2.4913143157958983, + "step": 210860 + }, + { + "epoch": 0.5740574384102355, + "grad_norm": 2.2196273803710938, + "learning_rate": 8.530516559901868e-06, + "loss": 2.467586135864258, + "step": 210880 + }, + { + "epoch": 0.57411188240098, + "grad_norm": 2.207235097885132, + "learning_rate": 8.529426195992913e-06, + "loss": 2.446846771240234, + "step": 210900 + }, + { + "epoch": 0.5741663263917245, + "grad_norm": 2.130431652069092, + "learning_rate": 8.528335832083958e-06, + "loss": 2.468713569641113, + "step": 210920 + }, + { + "epoch": 0.574220770382469, + "grad_norm": 2.1721739768981934, + "learning_rate": 8.527245468175003e-06, + "loss": 2.497249221801758, + "step": 210940 + }, + { + "epoch": 0.5742752143732136, + "grad_norm": 2.1258740425109863, + "learning_rate": 8.52615510426605e-06, + "loss": 2.476283645629883, + "step": 210960 + }, + { + "epoch": 0.5743296583639581, + "grad_norm": 2.1628918647766113, + "learning_rate": 8.525064740357095e-06, + "loss": 2.456061363220215, + "step": 210980 + }, + { + "epoch": 0.5743841023547026, + "grad_norm": 2.1301376819610596, + "learning_rate": 8.523974376448139e-06, + "loss": 2.4852460861206054, + "step": 211000 + }, + { + "epoch": 0.5744385463454471, + "grad_norm": 2.23067569732666, + "learning_rate": 8.522884012539185e-06, + "loss": 2.4816261291503907, + "step": 211020 + }, + { + "epoch": 0.5744929903361916, + "grad_norm": 2.297428846359253, + "learning_rate": 8.52179364863023e-06, + "loss": 2.4751426696777346, + "step": 211040 + }, + { + "epoch": 0.5745474343269361, + "grad_norm": 2.140369415283203, + "learning_rate": 8.520703284721277e-06, + "loss": 2.473395347595215, + "step": 211060 + }, + { + "epoch": 0.5746018783176807, + "grad_norm": 2.542170286178589, + "learning_rate": 8.51961292081232e-06, + "loss": 2.4737737655639647, + "step": 211080 + }, + { + "epoch": 0.5746563223084252, + "grad_norm": 2.1612112522125244, + "learning_rate": 8.518522556903366e-06, + "loss": 2.4751306533813477, + "step": 211100 + }, + { + "epoch": 0.5747107662991697, + "grad_norm": 2.0998573303222656, + "learning_rate": 8.517432192994412e-06, + "loss": 2.470619773864746, + "step": 211120 + }, + { + "epoch": 0.5747652102899142, + "grad_norm": 2.2881662845611572, + "learning_rate": 8.516341829085458e-06, + "loss": 2.4502422332763674, + "step": 211140 + }, + { + "epoch": 0.5748196542806587, + "grad_norm": 2.1071979999542236, + "learning_rate": 8.515251465176502e-06, + "loss": 2.4632349014282227, + "step": 211160 + }, + { + "epoch": 0.5748740982714033, + "grad_norm": 2.2581422328948975, + "learning_rate": 8.514161101267548e-06, + "loss": 2.4877668380737306, + "step": 211180 + }, + { + "epoch": 0.5749285422621478, + "grad_norm": 2.050196886062622, + "learning_rate": 8.513070737358594e-06, + "loss": 2.4640804290771485, + "step": 211200 + }, + { + "epoch": 0.5749829862528923, + "grad_norm": 2.3990731239318848, + "learning_rate": 8.51198037344964e-06, + "loss": 2.476536178588867, + "step": 211220 + }, + { + "epoch": 0.5750374302436368, + "grad_norm": 2.0933380126953125, + "learning_rate": 8.510890009540685e-06, + "loss": 2.4843994140625, + "step": 211240 + }, + { + "epoch": 0.5750918742343814, + "grad_norm": 2.0956456661224365, + "learning_rate": 8.50979964563173e-06, + "loss": 2.4717784881591798, + "step": 211260 + }, + { + "epoch": 0.5751463182251259, + "grad_norm": 2.1929283142089844, + "learning_rate": 8.508709281722775e-06, + "loss": 2.4793806076049805, + "step": 211280 + }, + { + "epoch": 0.5752007622158705, + "grad_norm": 2.2977135181427, + "learning_rate": 8.507618917813821e-06, + "loss": 2.4654142379760744, + "step": 211300 + }, + { + "epoch": 0.575255206206615, + "grad_norm": 3.1532998085021973, + "learning_rate": 8.506528553904867e-06, + "loss": 2.485447883605957, + "step": 211320 + }, + { + "epoch": 0.5753096501973595, + "grad_norm": 2.293297529220581, + "learning_rate": 8.505438189995911e-06, + "loss": 2.488237190246582, + "step": 211340 + }, + { + "epoch": 0.575364094188104, + "grad_norm": 2.1911683082580566, + "learning_rate": 8.504347826086957e-06, + "loss": 2.5029062271118163, + "step": 211360 + }, + { + "epoch": 0.5754185381788485, + "grad_norm": 2.239891529083252, + "learning_rate": 8.503257462178003e-06, + "loss": 2.4666934967041017, + "step": 211380 + }, + { + "epoch": 0.5754729821695931, + "grad_norm": 2.1360743045806885, + "learning_rate": 8.502167098269049e-06, + "loss": 2.4712347030639648, + "step": 211400 + }, + { + "epoch": 0.5755274261603376, + "grad_norm": 2.049247980117798, + "learning_rate": 8.501076734360093e-06, + "loss": 2.4400522232055666, + "step": 211420 + }, + { + "epoch": 0.5755818701510821, + "grad_norm": 2.157862663269043, + "learning_rate": 8.499986370451138e-06, + "loss": 2.4440423965454103, + "step": 211440 + }, + { + "epoch": 0.5756363141418266, + "grad_norm": 2.1710665225982666, + "learning_rate": 8.498896006542184e-06, + "loss": 2.495737075805664, + "step": 211460 + }, + { + "epoch": 0.5756907581325711, + "grad_norm": 2.1599245071411133, + "learning_rate": 8.49780564263323e-06, + "loss": 2.485829162597656, + "step": 211480 + }, + { + "epoch": 0.5757452021233156, + "grad_norm": 2.149709939956665, + "learning_rate": 8.496715278724274e-06, + "loss": 2.5226743698120115, + "step": 211500 + }, + { + "epoch": 0.5757996461140602, + "grad_norm": 2.139010429382324, + "learning_rate": 8.49562491481532e-06, + "loss": 2.4907798767089844, + "step": 211520 + }, + { + "epoch": 0.5758540901048047, + "grad_norm": 2.084552764892578, + "learning_rate": 8.494534550906366e-06, + "loss": 2.4745410919189452, + "step": 211540 + }, + { + "epoch": 0.5759085340955492, + "grad_norm": 2.2240233421325684, + "learning_rate": 8.493444186997412e-06, + "loss": 2.4955989837646486, + "step": 211560 + }, + { + "epoch": 0.5759629780862937, + "grad_norm": 2.2609095573425293, + "learning_rate": 8.492353823088456e-06, + "loss": 2.4871902465820312, + "step": 211580 + }, + { + "epoch": 0.5760174220770382, + "grad_norm": 2.049638271331787, + "learning_rate": 8.491263459179502e-06, + "loss": 2.445250701904297, + "step": 211600 + }, + { + "epoch": 0.5760718660677828, + "grad_norm": 2.2143402099609375, + "learning_rate": 8.490173095270547e-06, + "loss": 2.42819766998291, + "step": 211620 + }, + { + "epoch": 0.5761263100585273, + "grad_norm": 2.310669422149658, + "learning_rate": 8.489082731361593e-06, + "loss": 2.478995513916016, + "step": 211640 + }, + { + "epoch": 0.5761807540492718, + "grad_norm": 2.1653289794921875, + "learning_rate": 8.487992367452637e-06, + "loss": 2.4461917877197266, + "step": 211660 + }, + { + "epoch": 0.5762351980400163, + "grad_norm": 2.1251115798950195, + "learning_rate": 8.486902003543683e-06, + "loss": 2.48168830871582, + "step": 211680 + }, + { + "epoch": 0.5762896420307608, + "grad_norm": 2.094696044921875, + "learning_rate": 8.485811639634729e-06, + "loss": 2.4839221954345705, + "step": 211700 + }, + { + "epoch": 0.5763440860215053, + "grad_norm": 2.127819061279297, + "learning_rate": 8.484721275725775e-06, + "loss": 2.4667036056518556, + "step": 211720 + }, + { + "epoch": 0.5763985300122499, + "grad_norm": 2.3198163509368896, + "learning_rate": 8.483630911816819e-06, + "loss": 2.5024147033691406, + "step": 211740 + }, + { + "epoch": 0.5764529740029944, + "grad_norm": 2.496479034423828, + "learning_rate": 8.482540547907865e-06, + "loss": 2.469075393676758, + "step": 211760 + }, + { + "epoch": 0.576507417993739, + "grad_norm": 2.2156896591186523, + "learning_rate": 8.48145018399891e-06, + "loss": 2.489366912841797, + "step": 211780 + }, + { + "epoch": 0.5765618619844834, + "grad_norm": 2.165409803390503, + "learning_rate": 8.480359820089956e-06, + "loss": 2.4936599731445312, + "step": 211800 + }, + { + "epoch": 0.576616305975228, + "grad_norm": 2.121434450149536, + "learning_rate": 8.479269456181e-06, + "loss": 2.475050163269043, + "step": 211820 + }, + { + "epoch": 0.5766707499659725, + "grad_norm": 2.2637784481048584, + "learning_rate": 8.478179092272046e-06, + "loss": 2.4650331497192384, + "step": 211840 + }, + { + "epoch": 0.5767251939567171, + "grad_norm": 2.1861190795898438, + "learning_rate": 8.477088728363092e-06, + "loss": 2.4930961608886717, + "step": 211860 + }, + { + "epoch": 0.5767796379474616, + "grad_norm": 2.1630895137786865, + "learning_rate": 8.475998364454138e-06, + "loss": 2.483149528503418, + "step": 211880 + }, + { + "epoch": 0.5768340819382061, + "grad_norm": 2.160501003265381, + "learning_rate": 8.474908000545182e-06, + "loss": 2.4781450271606444, + "step": 211900 + }, + { + "epoch": 0.5768885259289506, + "grad_norm": 2.2417526245117188, + "learning_rate": 8.473817636636228e-06, + "loss": 2.5072683334350585, + "step": 211920 + }, + { + "epoch": 0.5769429699196951, + "grad_norm": 2.26920485496521, + "learning_rate": 8.472727272727274e-06, + "loss": 2.483776092529297, + "step": 211940 + }, + { + "epoch": 0.5769974139104397, + "grad_norm": 2.184656858444214, + "learning_rate": 8.47163690881832e-06, + "loss": 2.504824638366699, + "step": 211960 + }, + { + "epoch": 0.5770518579011842, + "grad_norm": 2.0488805770874023, + "learning_rate": 8.470546544909363e-06, + "loss": 2.498163032531738, + "step": 211980 + }, + { + "epoch": 0.5771063018919287, + "grad_norm": 3.2719411849975586, + "learning_rate": 8.46945618100041e-06, + "loss": 2.4734500885009765, + "step": 212000 + }, + { + "epoch": 0.5771607458826732, + "grad_norm": 2.2498714923858643, + "learning_rate": 8.468365817091455e-06, + "loss": 2.497673225402832, + "step": 212020 + }, + { + "epoch": 0.5772151898734177, + "grad_norm": 5.08745813369751, + "learning_rate": 8.467275453182501e-06, + "loss": 2.4619413375854493, + "step": 212040 + }, + { + "epoch": 0.5772696338641622, + "grad_norm": 2.282231092453003, + "learning_rate": 8.466185089273545e-06, + "loss": 2.509195899963379, + "step": 212060 + }, + { + "epoch": 0.5773240778549068, + "grad_norm": 2.142878770828247, + "learning_rate": 8.46509472536459e-06, + "loss": 2.463388442993164, + "step": 212080 + }, + { + "epoch": 0.5773785218456513, + "grad_norm": 2.1542625427246094, + "learning_rate": 8.464004361455637e-06, + "loss": 2.468816375732422, + "step": 212100 + }, + { + "epoch": 0.5774329658363958, + "grad_norm": 2.2034449577331543, + "learning_rate": 8.462913997546682e-06, + "loss": 2.5218883514404298, + "step": 212120 + }, + { + "epoch": 0.5774874098271403, + "grad_norm": 2.2303311824798584, + "learning_rate": 8.461823633637727e-06, + "loss": 2.493875503540039, + "step": 212140 + }, + { + "epoch": 0.5775418538178848, + "grad_norm": 2.079341411590576, + "learning_rate": 8.460733269728772e-06, + "loss": 2.4728420257568358, + "step": 212160 + }, + { + "epoch": 0.5775962978086294, + "grad_norm": 2.2358450889587402, + "learning_rate": 8.459642905819818e-06, + "loss": 2.46895751953125, + "step": 212180 + }, + { + "epoch": 0.5776507417993739, + "grad_norm": 2.363898992538452, + "learning_rate": 8.458552541910864e-06, + "loss": 2.4854000091552733, + "step": 212200 + }, + { + "epoch": 0.5777051857901184, + "grad_norm": 3.786612033843994, + "learning_rate": 8.457462178001908e-06, + "loss": 2.482920265197754, + "step": 212220 + }, + { + "epoch": 0.5777596297808629, + "grad_norm": 2.1801695823669434, + "learning_rate": 8.456371814092954e-06, + "loss": 2.454224967956543, + "step": 212240 + }, + { + "epoch": 0.5778140737716074, + "grad_norm": 2.3794093132019043, + "learning_rate": 8.455281450184e-06, + "loss": 2.479245185852051, + "step": 212260 + }, + { + "epoch": 0.5778685177623519, + "grad_norm": 2.1631181240081787, + "learning_rate": 8.454191086275046e-06, + "loss": 2.485502243041992, + "step": 212280 + }, + { + "epoch": 0.5779229617530965, + "grad_norm": 2.0976221561431885, + "learning_rate": 8.45310072236609e-06, + "loss": 2.476222610473633, + "step": 212300 + }, + { + "epoch": 0.577977405743841, + "grad_norm": 2.182518720626831, + "learning_rate": 8.452010358457135e-06, + "loss": 2.488548469543457, + "step": 212320 + }, + { + "epoch": 0.5780318497345855, + "grad_norm": 2.191999673843384, + "learning_rate": 8.450919994548181e-06, + "loss": 2.4637548446655275, + "step": 212340 + }, + { + "epoch": 0.57808629372533, + "grad_norm": 3.3354246616363525, + "learning_rate": 8.449829630639227e-06, + "loss": 2.4737308502197264, + "step": 212360 + }, + { + "epoch": 0.5781407377160745, + "grad_norm": 2.130857229232788, + "learning_rate": 8.448739266730271e-06, + "loss": 2.47268009185791, + "step": 212380 + }, + { + "epoch": 0.5781951817068192, + "grad_norm": 2.1902718544006348, + "learning_rate": 8.447648902821317e-06, + "loss": 2.4353389739990234, + "step": 212400 + }, + { + "epoch": 0.5782496256975637, + "grad_norm": 2.3814449310302734, + "learning_rate": 8.446558538912363e-06, + "loss": 2.4862443923950197, + "step": 212420 + }, + { + "epoch": 0.5783040696883082, + "grad_norm": 2.177501678466797, + "learning_rate": 8.445468175003409e-06, + "loss": 2.477946662902832, + "step": 212440 + }, + { + "epoch": 0.5783585136790527, + "grad_norm": 2.138031244277954, + "learning_rate": 8.444377811094453e-06, + "loss": 2.437849426269531, + "step": 212460 + }, + { + "epoch": 0.5784129576697972, + "grad_norm": 2.2618815898895264, + "learning_rate": 8.443287447185499e-06, + "loss": 2.4822710037231444, + "step": 212480 + }, + { + "epoch": 0.5784674016605417, + "grad_norm": 2.2000746726989746, + "learning_rate": 8.442197083276544e-06, + "loss": 2.473007392883301, + "step": 212500 + }, + { + "epoch": 0.5785218456512863, + "grad_norm": 2.2234649658203125, + "learning_rate": 8.44110671936759e-06, + "loss": 2.492331886291504, + "step": 212520 + }, + { + "epoch": 0.5785762896420308, + "grad_norm": 2.237678050994873, + "learning_rate": 8.440016355458634e-06, + "loss": 2.4937107086181642, + "step": 212540 + }, + { + "epoch": 0.5786307336327753, + "grad_norm": 2.258436918258667, + "learning_rate": 8.43892599154968e-06, + "loss": 2.457220458984375, + "step": 212560 + }, + { + "epoch": 0.5786851776235198, + "grad_norm": 2.20119571685791, + "learning_rate": 8.437835627640726e-06, + "loss": 2.460145950317383, + "step": 212580 + }, + { + "epoch": 0.5787396216142643, + "grad_norm": 2.1107242107391357, + "learning_rate": 8.436745263731772e-06, + "loss": 2.477433776855469, + "step": 212600 + }, + { + "epoch": 0.5787940656050089, + "grad_norm": 2.246316909790039, + "learning_rate": 8.435654899822816e-06, + "loss": 2.445663070678711, + "step": 212620 + }, + { + "epoch": 0.5788485095957534, + "grad_norm": 2.172663688659668, + "learning_rate": 8.434564535913862e-06, + "loss": 2.449302864074707, + "step": 212640 + }, + { + "epoch": 0.5789029535864979, + "grad_norm": 2.1891913414001465, + "learning_rate": 8.433474172004907e-06, + "loss": 2.465021514892578, + "step": 212660 + }, + { + "epoch": 0.5789573975772424, + "grad_norm": 2.272740602493286, + "learning_rate": 8.432383808095953e-06, + "loss": 2.4803699493408202, + "step": 212680 + }, + { + "epoch": 0.5790118415679869, + "grad_norm": 2.1927597522735596, + "learning_rate": 8.431293444186997e-06, + "loss": 2.4789642333984374, + "step": 212700 + }, + { + "epoch": 0.5790662855587314, + "grad_norm": 2.265793561935425, + "learning_rate": 8.430203080278043e-06, + "loss": 2.4904548645019533, + "step": 212720 + }, + { + "epoch": 0.579120729549476, + "grad_norm": 2.197880744934082, + "learning_rate": 8.429112716369089e-06, + "loss": 2.4521549224853514, + "step": 212740 + }, + { + "epoch": 0.5791751735402205, + "grad_norm": 2.124009370803833, + "learning_rate": 8.428022352460135e-06, + "loss": 2.493107223510742, + "step": 212760 + }, + { + "epoch": 0.579229617530965, + "grad_norm": 2.15022349357605, + "learning_rate": 8.426931988551179e-06, + "loss": 2.4777088165283203, + "step": 212780 + }, + { + "epoch": 0.5792840615217095, + "grad_norm": 2.162477970123291, + "learning_rate": 8.425841624642225e-06, + "loss": 2.4478179931640627, + "step": 212800 + }, + { + "epoch": 0.579338505512454, + "grad_norm": 2.1165125370025635, + "learning_rate": 8.42475126073327e-06, + "loss": 2.480653190612793, + "step": 212820 + }, + { + "epoch": 0.5793929495031986, + "grad_norm": 2.0369622707366943, + "learning_rate": 8.423660896824316e-06, + "loss": 2.435382843017578, + "step": 212840 + }, + { + "epoch": 0.5794473934939431, + "grad_norm": 2.136378765106201, + "learning_rate": 8.42257053291536e-06, + "loss": 2.4874101638793946, + "step": 212860 + }, + { + "epoch": 0.5795018374846876, + "grad_norm": 2.090885877609253, + "learning_rate": 8.421480169006406e-06, + "loss": 2.4480922698974608, + "step": 212880 + }, + { + "epoch": 0.5795562814754321, + "grad_norm": 2.500274658203125, + "learning_rate": 8.420389805097452e-06, + "loss": 2.4659215927124025, + "step": 212900 + }, + { + "epoch": 0.5796107254661766, + "grad_norm": 2.1604163646698, + "learning_rate": 8.419299441188498e-06, + "loss": 2.489747428894043, + "step": 212920 + }, + { + "epoch": 0.5796651694569211, + "grad_norm": 5.092579364776611, + "learning_rate": 8.418209077279544e-06, + "loss": 2.4606632232666015, + "step": 212940 + }, + { + "epoch": 0.5797196134476658, + "grad_norm": 2.318708896636963, + "learning_rate": 8.417118713370588e-06, + "loss": 2.4528356552124024, + "step": 212960 + }, + { + "epoch": 0.5797740574384103, + "grad_norm": 2.20752215385437, + "learning_rate": 8.416028349461634e-06, + "loss": 2.4482378005981444, + "step": 212980 + }, + { + "epoch": 0.5798285014291548, + "grad_norm": 2.248140811920166, + "learning_rate": 8.41493798555268e-06, + "loss": 2.4839534759521484, + "step": 213000 + }, + { + "epoch": 0.5798829454198993, + "grad_norm": 2.2865936756134033, + "learning_rate": 8.413847621643725e-06, + "loss": 2.479389762878418, + "step": 213020 + }, + { + "epoch": 0.5799373894106438, + "grad_norm": 2.115396022796631, + "learning_rate": 8.41275725773477e-06, + "loss": 2.477038764953613, + "step": 213040 + }, + { + "epoch": 0.5799918334013883, + "grad_norm": 2.217877149581909, + "learning_rate": 8.411666893825815e-06, + "loss": 2.4931922912597657, + "step": 213060 + }, + { + "epoch": 0.5800462773921329, + "grad_norm": 2.1373350620269775, + "learning_rate": 8.410576529916861e-06, + "loss": 2.477690887451172, + "step": 213080 + }, + { + "epoch": 0.5801007213828774, + "grad_norm": 2.1753957271575928, + "learning_rate": 8.409486166007907e-06, + "loss": 2.461927795410156, + "step": 213100 + }, + { + "epoch": 0.5801551653736219, + "grad_norm": 2.2311551570892334, + "learning_rate": 8.408395802098951e-06, + "loss": 2.4538156509399416, + "step": 213120 + }, + { + "epoch": 0.5802096093643664, + "grad_norm": 2.7938730716705322, + "learning_rate": 8.407305438189997e-06, + "loss": 2.489955520629883, + "step": 213140 + }, + { + "epoch": 0.5802640533551109, + "grad_norm": 2.173092842102051, + "learning_rate": 8.406215074281042e-06, + "loss": 2.459459686279297, + "step": 213160 + }, + { + "epoch": 0.5803184973458555, + "grad_norm": 2.153792381286621, + "learning_rate": 8.405124710372088e-06, + "loss": 2.4837636947631836, + "step": 213180 + }, + { + "epoch": 0.5803729413366, + "grad_norm": 2.328317403793335, + "learning_rate": 8.404034346463132e-06, + "loss": 2.468007278442383, + "step": 213200 + }, + { + "epoch": 0.5804273853273445, + "grad_norm": 2.2378435134887695, + "learning_rate": 8.402943982554178e-06, + "loss": 2.4552597045898437, + "step": 213220 + }, + { + "epoch": 0.580481829318089, + "grad_norm": 2.063615560531616, + "learning_rate": 8.401853618645224e-06, + "loss": 2.478756141662598, + "step": 213240 + }, + { + "epoch": 0.5805362733088335, + "grad_norm": 2.236095905303955, + "learning_rate": 8.40076325473627e-06, + "loss": 2.4907655715942383, + "step": 213260 + }, + { + "epoch": 0.580590717299578, + "grad_norm": 2.3265788555145264, + "learning_rate": 8.399672890827314e-06, + "loss": 2.4781909942626954, + "step": 213280 + }, + { + "epoch": 0.5806451612903226, + "grad_norm": 2.154949903488159, + "learning_rate": 8.39858252691836e-06, + "loss": 2.499011421203613, + "step": 213300 + }, + { + "epoch": 0.5806996052810671, + "grad_norm": 2.1048531532287598, + "learning_rate": 8.397492163009406e-06, + "loss": 2.473529815673828, + "step": 213320 + }, + { + "epoch": 0.5807540492718116, + "grad_norm": 2.135871171951294, + "learning_rate": 8.396401799100451e-06, + "loss": 2.5045778274536135, + "step": 213340 + }, + { + "epoch": 0.5808084932625561, + "grad_norm": 2.2057929039001465, + "learning_rate": 8.395311435191495e-06, + "loss": 2.473976707458496, + "step": 213360 + }, + { + "epoch": 0.5808629372533006, + "grad_norm": 2.13211727142334, + "learning_rate": 8.394221071282541e-06, + "loss": 2.455174446105957, + "step": 213380 + }, + { + "epoch": 0.5809173812440452, + "grad_norm": 2.512810230255127, + "learning_rate": 8.393130707373587e-06, + "loss": 2.486669158935547, + "step": 213400 + }, + { + "epoch": 0.5809718252347897, + "grad_norm": 2.178340196609497, + "learning_rate": 8.392040343464633e-06, + "loss": 2.464534568786621, + "step": 213420 + }, + { + "epoch": 0.5810262692255342, + "grad_norm": 2.1640257835388184, + "learning_rate": 8.390949979555677e-06, + "loss": 2.486103820800781, + "step": 213440 + }, + { + "epoch": 0.5810807132162787, + "grad_norm": 2.171067237854004, + "learning_rate": 8.389859615646723e-06, + "loss": 2.4717708587646485, + "step": 213460 + }, + { + "epoch": 0.5811351572070232, + "grad_norm": 2.2262344360351562, + "learning_rate": 8.388769251737769e-06, + "loss": 2.4608089447021486, + "step": 213480 + }, + { + "epoch": 0.5811896011977677, + "grad_norm": 2.2269043922424316, + "learning_rate": 8.387678887828814e-06, + "loss": 2.460430908203125, + "step": 213500 + }, + { + "epoch": 0.5812440451885124, + "grad_norm": 2.0883262157440186, + "learning_rate": 8.386588523919859e-06, + "loss": 2.458479309082031, + "step": 213520 + }, + { + "epoch": 0.5812984891792569, + "grad_norm": 2.276658773422241, + "learning_rate": 8.385498160010904e-06, + "loss": 2.479482078552246, + "step": 213540 + }, + { + "epoch": 0.5813529331700014, + "grad_norm": 2.2086269855499268, + "learning_rate": 8.38440779610195e-06, + "loss": 2.4936986923217774, + "step": 213560 + }, + { + "epoch": 0.5814073771607459, + "grad_norm": 2.140873908996582, + "learning_rate": 8.383317432192996e-06, + "loss": 2.457340621948242, + "step": 213580 + }, + { + "epoch": 0.5814618211514904, + "grad_norm": 2.0706303119659424, + "learning_rate": 8.38222706828404e-06, + "loss": 2.4757226943969726, + "step": 213600 + }, + { + "epoch": 0.581516265142235, + "grad_norm": 2.4482600688934326, + "learning_rate": 8.381136704375086e-06, + "loss": 2.4832233428955077, + "step": 213620 + }, + { + "epoch": 0.5815707091329795, + "grad_norm": 2.144693374633789, + "learning_rate": 8.380046340466132e-06, + "loss": 2.471384620666504, + "step": 213640 + }, + { + "epoch": 0.581625153123724, + "grad_norm": 2.2192599773406982, + "learning_rate": 8.378955976557178e-06, + "loss": 2.4911497116088865, + "step": 213660 + }, + { + "epoch": 0.5816795971144685, + "grad_norm": 2.0855209827423096, + "learning_rate": 8.377865612648222e-06, + "loss": 2.4741704940795897, + "step": 213680 + }, + { + "epoch": 0.581734041105213, + "grad_norm": 2.1299843788146973, + "learning_rate": 8.376775248739267e-06, + "loss": 2.492059898376465, + "step": 213700 + }, + { + "epoch": 0.5817884850959575, + "grad_norm": 2.1313960552215576, + "learning_rate": 8.375684884830313e-06, + "loss": 2.49554500579834, + "step": 213720 + }, + { + "epoch": 0.5818429290867021, + "grad_norm": 2.1576650142669678, + "learning_rate": 8.374594520921359e-06, + "loss": 2.4642917633056642, + "step": 213740 + }, + { + "epoch": 0.5818973730774466, + "grad_norm": 2.0678586959838867, + "learning_rate": 8.373504157012403e-06, + "loss": 2.453311729431152, + "step": 213760 + }, + { + "epoch": 0.5819518170681911, + "grad_norm": 2.181908130645752, + "learning_rate": 8.372413793103449e-06, + "loss": 2.499542999267578, + "step": 213780 + }, + { + "epoch": 0.5820062610589356, + "grad_norm": 2.2059903144836426, + "learning_rate": 8.371323429194495e-06, + "loss": 2.483533477783203, + "step": 213800 + }, + { + "epoch": 0.5820607050496801, + "grad_norm": 2.304675817489624, + "learning_rate": 8.37023306528554e-06, + "loss": 2.4781970977783203, + "step": 213820 + }, + { + "epoch": 0.5821151490404247, + "grad_norm": 2.332961082458496, + "learning_rate": 8.369142701376585e-06, + "loss": 2.4981454849243163, + "step": 213840 + }, + { + "epoch": 0.5821695930311692, + "grad_norm": 2.4047579765319824, + "learning_rate": 8.36805233746763e-06, + "loss": 2.5083524703979494, + "step": 213860 + }, + { + "epoch": 0.5822240370219137, + "grad_norm": 2.1688709259033203, + "learning_rate": 8.366961973558676e-06, + "loss": 2.4762348175048827, + "step": 213880 + }, + { + "epoch": 0.5822784810126582, + "grad_norm": 3.522038221359253, + "learning_rate": 8.365871609649722e-06, + "loss": 2.4696041107177735, + "step": 213900 + }, + { + "epoch": 0.5823329250034027, + "grad_norm": 2.092294692993164, + "learning_rate": 8.364781245740766e-06, + "loss": 2.4713735580444336, + "step": 213920 + }, + { + "epoch": 0.5823873689941472, + "grad_norm": 2.114931106567383, + "learning_rate": 8.363690881831812e-06, + "loss": 2.4761024475097657, + "step": 213940 + }, + { + "epoch": 0.5824418129848918, + "grad_norm": 2.23007869720459, + "learning_rate": 8.362600517922858e-06, + "loss": 2.4529346466064452, + "step": 213960 + }, + { + "epoch": 0.5824962569756363, + "grad_norm": 2.1378934383392334, + "learning_rate": 8.361510154013904e-06, + "loss": 2.475996398925781, + "step": 213980 + }, + { + "epoch": 0.5825507009663808, + "grad_norm": 2.5539791584014893, + "learning_rate": 8.360419790104948e-06, + "loss": 2.4903364181518555, + "step": 214000 + }, + { + "epoch": 0.5826051449571253, + "grad_norm": 2.130121946334839, + "learning_rate": 8.359329426195994e-06, + "loss": 2.4747522354125975, + "step": 214020 + }, + { + "epoch": 0.5826595889478698, + "grad_norm": 3.8612940311431885, + "learning_rate": 8.35823906228704e-06, + "loss": 2.4781671524047852, + "step": 214040 + }, + { + "epoch": 0.5827140329386143, + "grad_norm": 2.1667747497558594, + "learning_rate": 8.357148698378085e-06, + "loss": 2.482509803771973, + "step": 214060 + }, + { + "epoch": 0.582768476929359, + "grad_norm": 2.1945877075195312, + "learning_rate": 8.35605833446913e-06, + "loss": 2.4895095825195312, + "step": 214080 + }, + { + "epoch": 0.5828229209201035, + "grad_norm": 2.1381020545959473, + "learning_rate": 8.354967970560175e-06, + "loss": 2.47597599029541, + "step": 214100 + }, + { + "epoch": 0.582877364910848, + "grad_norm": 2.2027475833892822, + "learning_rate": 8.353877606651221e-06, + "loss": 2.521007537841797, + "step": 214120 + }, + { + "epoch": 0.5829318089015925, + "grad_norm": 2.0839474201202393, + "learning_rate": 8.352787242742267e-06, + "loss": 2.463793182373047, + "step": 214140 + }, + { + "epoch": 0.582986252892337, + "grad_norm": 2.159852981567383, + "learning_rate": 8.351696878833311e-06, + "loss": 2.496548652648926, + "step": 214160 + }, + { + "epoch": 0.5830406968830816, + "grad_norm": 2.1711764335632324, + "learning_rate": 8.350606514924357e-06, + "loss": 2.4803960800170897, + "step": 214180 + }, + { + "epoch": 0.5830951408738261, + "grad_norm": 2.1819844245910645, + "learning_rate": 8.349516151015403e-06, + "loss": 2.4807497024536134, + "step": 214200 + }, + { + "epoch": 0.5831495848645706, + "grad_norm": 2.147618532180786, + "learning_rate": 8.348425787106448e-06, + "loss": 2.4900150299072266, + "step": 214220 + }, + { + "epoch": 0.5832040288553151, + "grad_norm": 2.087698459625244, + "learning_rate": 8.347335423197492e-06, + "loss": 2.4927047729492187, + "step": 214240 + }, + { + "epoch": 0.5832584728460596, + "grad_norm": 2.183596611022949, + "learning_rate": 8.346245059288538e-06, + "loss": 2.4765438079833983, + "step": 214260 + }, + { + "epoch": 0.5833129168368041, + "grad_norm": 2.2878198623657227, + "learning_rate": 8.345154695379584e-06, + "loss": 2.472840690612793, + "step": 214280 + }, + { + "epoch": 0.5833673608275487, + "grad_norm": 2.478722095489502, + "learning_rate": 8.34406433147063e-06, + "loss": 2.4804201126098633, + "step": 214300 + }, + { + "epoch": 0.5834218048182932, + "grad_norm": 2.0778377056121826, + "learning_rate": 8.342973967561674e-06, + "loss": 2.4611846923828127, + "step": 214320 + }, + { + "epoch": 0.5834762488090377, + "grad_norm": 2.091315269470215, + "learning_rate": 8.34188360365272e-06, + "loss": 2.474131202697754, + "step": 214340 + }, + { + "epoch": 0.5835306927997822, + "grad_norm": 2.1303417682647705, + "learning_rate": 8.340793239743766e-06, + "loss": 2.485376739501953, + "step": 214360 + }, + { + "epoch": 0.5835851367905267, + "grad_norm": 2.144019365310669, + "learning_rate": 8.339702875834811e-06, + "loss": 2.5003429412841798, + "step": 214380 + }, + { + "epoch": 0.5836395807812713, + "grad_norm": 2.0913562774658203, + "learning_rate": 8.338612511925856e-06, + "loss": 2.4900068283081054, + "step": 214400 + }, + { + "epoch": 0.5836940247720158, + "grad_norm": 2.1680004596710205, + "learning_rate": 8.337522148016901e-06, + "loss": 2.4781850814819335, + "step": 214420 + }, + { + "epoch": 0.5837484687627603, + "grad_norm": 2.2272164821624756, + "learning_rate": 8.336431784107947e-06, + "loss": 2.483296203613281, + "step": 214440 + }, + { + "epoch": 0.5838029127535048, + "grad_norm": 2.2823386192321777, + "learning_rate": 8.335341420198993e-06, + "loss": 2.4676889419555663, + "step": 214460 + }, + { + "epoch": 0.5838573567442493, + "grad_norm": 2.0819942951202393, + "learning_rate": 8.334251056290037e-06, + "loss": 2.4641923904418945, + "step": 214480 + }, + { + "epoch": 0.5839118007349938, + "grad_norm": 2.180734157562256, + "learning_rate": 8.333160692381083e-06, + "loss": 2.4759071350097654, + "step": 214500 + }, + { + "epoch": 0.5839662447257384, + "grad_norm": 2.133877992630005, + "learning_rate": 8.332070328472129e-06, + "loss": 2.4743520736694338, + "step": 214520 + }, + { + "epoch": 0.5840206887164829, + "grad_norm": 2.0822157859802246, + "learning_rate": 8.330979964563175e-06, + "loss": 2.479060173034668, + "step": 214540 + }, + { + "epoch": 0.5840751327072274, + "grad_norm": 2.5442280769348145, + "learning_rate": 8.329889600654219e-06, + "loss": 2.4891761779785155, + "step": 214560 + }, + { + "epoch": 0.5841295766979719, + "grad_norm": 2.164656639099121, + "learning_rate": 8.328799236745264e-06, + "loss": 2.471495246887207, + "step": 214580 + }, + { + "epoch": 0.5841840206887164, + "grad_norm": 2.3403639793395996, + "learning_rate": 8.32770887283631e-06, + "loss": 2.4587453842163085, + "step": 214600 + }, + { + "epoch": 0.5842384646794611, + "grad_norm": 2.135836362838745, + "learning_rate": 8.326618508927356e-06, + "loss": 2.4539228439331056, + "step": 214620 + }, + { + "epoch": 0.5842929086702056, + "grad_norm": 2.169499397277832, + "learning_rate": 8.325528145018402e-06, + "loss": 2.4835227966308593, + "step": 214640 + }, + { + "epoch": 0.5843473526609501, + "grad_norm": 2.1240854263305664, + "learning_rate": 8.324437781109446e-06, + "loss": 2.4976539611816406, + "step": 214660 + }, + { + "epoch": 0.5844017966516946, + "grad_norm": 2.108167886734009, + "learning_rate": 8.323347417200492e-06, + "loss": 2.4491891860961914, + "step": 214680 + }, + { + "epoch": 0.5844562406424391, + "grad_norm": 2.194814920425415, + "learning_rate": 8.322257053291538e-06, + "loss": 2.461271286010742, + "step": 214700 + }, + { + "epoch": 0.5845106846331836, + "grad_norm": 2.0947959423065186, + "learning_rate": 8.321166689382583e-06, + "loss": 2.472900390625, + "step": 214720 + }, + { + "epoch": 0.5845651286239282, + "grad_norm": 2.856060743331909, + "learning_rate": 8.320076325473628e-06, + "loss": 2.4416009902954103, + "step": 214740 + }, + { + "epoch": 0.5846195726146727, + "grad_norm": 2.1444621086120605, + "learning_rate": 8.318985961564673e-06, + "loss": 2.4428611755371095, + "step": 214760 + }, + { + "epoch": 0.5846740166054172, + "grad_norm": 2.171473741531372, + "learning_rate": 8.31789559765572e-06, + "loss": 2.4585800170898438, + "step": 214780 + }, + { + "epoch": 0.5847284605961617, + "grad_norm": 2.1876296997070312, + "learning_rate": 8.316805233746765e-06, + "loss": 2.47751407623291, + "step": 214800 + }, + { + "epoch": 0.5847829045869062, + "grad_norm": 2.2541608810424805, + "learning_rate": 8.315714869837809e-06, + "loss": 2.4819713592529298, + "step": 214820 + }, + { + "epoch": 0.5848373485776508, + "grad_norm": 2.114701271057129, + "learning_rate": 8.314624505928855e-06, + "loss": 2.481823539733887, + "step": 214840 + }, + { + "epoch": 0.5848917925683953, + "grad_norm": 2.1517105102539062, + "learning_rate": 8.3135341420199e-06, + "loss": 2.4910755157470703, + "step": 214860 + }, + { + "epoch": 0.5849462365591398, + "grad_norm": 2.129730224609375, + "learning_rate": 8.312443778110947e-06, + "loss": 2.482345199584961, + "step": 214880 + }, + { + "epoch": 0.5850006805498843, + "grad_norm": 2.1669516563415527, + "learning_rate": 8.31135341420199e-06, + "loss": 2.4761064529418944, + "step": 214900 + }, + { + "epoch": 0.5850551245406288, + "grad_norm": 2.1947638988494873, + "learning_rate": 8.310263050293036e-06, + "loss": 2.4517818450927735, + "step": 214920 + }, + { + "epoch": 0.5851095685313733, + "grad_norm": 2.2752597332000732, + "learning_rate": 8.309172686384082e-06, + "loss": 2.450226593017578, + "step": 214940 + }, + { + "epoch": 0.5851640125221179, + "grad_norm": 2.1059987545013428, + "learning_rate": 8.308082322475128e-06, + "loss": 2.4625646591186525, + "step": 214960 + }, + { + "epoch": 0.5852184565128624, + "grad_norm": 2.207221508026123, + "learning_rate": 8.306991958566172e-06, + "loss": 2.4791534423828123, + "step": 214980 + }, + { + "epoch": 0.5852729005036069, + "grad_norm": 2.188992500305176, + "learning_rate": 8.305901594657218e-06, + "loss": 2.4841083526611327, + "step": 215000 + }, + { + "epoch": 0.5853273444943514, + "grad_norm": 2.127964735031128, + "learning_rate": 8.304811230748264e-06, + "loss": 2.4584323883056642, + "step": 215020 + }, + { + "epoch": 0.5853817884850959, + "grad_norm": 2.128418445587158, + "learning_rate": 8.30372086683931e-06, + "loss": 2.4915809631347656, + "step": 215040 + }, + { + "epoch": 0.5854362324758405, + "grad_norm": 2.100059747695923, + "learning_rate": 8.302630502930354e-06, + "loss": 2.487415313720703, + "step": 215060 + }, + { + "epoch": 0.585490676466585, + "grad_norm": 2.063680648803711, + "learning_rate": 8.3015401390214e-06, + "loss": 2.491472625732422, + "step": 215080 + }, + { + "epoch": 0.5855451204573295, + "grad_norm": 2.0631136894226074, + "learning_rate": 8.300449775112445e-06, + "loss": 2.4652021408081053, + "step": 215100 + }, + { + "epoch": 0.585599564448074, + "grad_norm": 2.1508328914642334, + "learning_rate": 8.299359411203491e-06, + "loss": 2.4622859954833984, + "step": 215120 + }, + { + "epoch": 0.5856540084388185, + "grad_norm": 2.1932358741760254, + "learning_rate": 8.298269047294535e-06, + "loss": 2.458283615112305, + "step": 215140 + }, + { + "epoch": 0.585708452429563, + "grad_norm": 2.1555898189544678, + "learning_rate": 8.297178683385581e-06, + "loss": 2.4457420349121093, + "step": 215160 + }, + { + "epoch": 0.5857628964203077, + "grad_norm": 2.1970677375793457, + "learning_rate": 8.296088319476627e-06, + "loss": 2.477674102783203, + "step": 215180 + }, + { + "epoch": 0.5858173404110522, + "grad_norm": 2.2988100051879883, + "learning_rate": 8.294997955567673e-06, + "loss": 2.490081024169922, + "step": 215200 + }, + { + "epoch": 0.5858717844017967, + "grad_norm": 2.2082748413085938, + "learning_rate": 8.293907591658717e-06, + "loss": 2.4729001998901365, + "step": 215220 + }, + { + "epoch": 0.5859262283925412, + "grad_norm": 2.09075927734375, + "learning_rate": 8.292817227749763e-06, + "loss": 2.4716764450073243, + "step": 215240 + }, + { + "epoch": 0.5859806723832857, + "grad_norm": 2.4050285816192627, + "learning_rate": 8.291726863840808e-06, + "loss": 2.4746606826782225, + "step": 215260 + }, + { + "epoch": 0.5860351163740302, + "grad_norm": 2.1193459033966064, + "learning_rate": 8.290636499931854e-06, + "loss": 2.4867982864379883, + "step": 215280 + }, + { + "epoch": 0.5860895603647748, + "grad_norm": 2.19866943359375, + "learning_rate": 8.289546136022898e-06, + "loss": 2.467900276184082, + "step": 215300 + }, + { + "epoch": 0.5861440043555193, + "grad_norm": 2.2857697010040283, + "learning_rate": 8.288455772113944e-06, + "loss": 2.487742233276367, + "step": 215320 + }, + { + "epoch": 0.5861984483462638, + "grad_norm": 2.277480363845825, + "learning_rate": 8.28736540820499e-06, + "loss": 2.4915435791015623, + "step": 215340 + }, + { + "epoch": 0.5862528923370083, + "grad_norm": 2.243332624435425, + "learning_rate": 8.286275044296034e-06, + "loss": 2.4820583343505858, + "step": 215360 + }, + { + "epoch": 0.5863073363277528, + "grad_norm": 2.0533103942871094, + "learning_rate": 8.28518468038708e-06, + "loss": 2.469523620605469, + "step": 215380 + }, + { + "epoch": 0.5863617803184974, + "grad_norm": 2.407412528991699, + "learning_rate": 8.284094316478126e-06, + "loss": 2.4877012252807615, + "step": 215400 + }, + { + "epoch": 0.5864162243092419, + "grad_norm": 2.3047358989715576, + "learning_rate": 8.283003952569171e-06, + "loss": 2.4505401611328126, + "step": 215420 + }, + { + "epoch": 0.5864706682999864, + "grad_norm": 2.1231729984283447, + "learning_rate": 8.281913588660216e-06, + "loss": 2.4772056579589843, + "step": 215440 + }, + { + "epoch": 0.5865251122907309, + "grad_norm": 2.12677264213562, + "learning_rate": 8.280823224751261e-06, + "loss": 2.464030075073242, + "step": 215460 + }, + { + "epoch": 0.5865795562814754, + "grad_norm": 2.1677284240722656, + "learning_rate": 8.279732860842307e-06, + "loss": 2.454604911804199, + "step": 215480 + }, + { + "epoch": 0.5866340002722199, + "grad_norm": 2.0725760459899902, + "learning_rate": 8.278642496933353e-06, + "loss": 2.4895807266235352, + "step": 215500 + }, + { + "epoch": 0.5866884442629645, + "grad_norm": 2.414114475250244, + "learning_rate": 8.277552133024397e-06, + "loss": 2.4876081466674806, + "step": 215520 + }, + { + "epoch": 0.586742888253709, + "grad_norm": 2.179133653640747, + "learning_rate": 8.276461769115443e-06, + "loss": 2.4618772506713866, + "step": 215540 + }, + { + "epoch": 0.5867973322444535, + "grad_norm": 2.190154552459717, + "learning_rate": 8.275371405206489e-06, + "loss": 2.4931243896484374, + "step": 215560 + }, + { + "epoch": 0.586851776235198, + "grad_norm": 2.1368656158447266, + "learning_rate": 8.274281041297535e-06, + "loss": 2.4623945236206053, + "step": 215580 + }, + { + "epoch": 0.5869062202259425, + "grad_norm": 2.1507456302642822, + "learning_rate": 8.273190677388579e-06, + "loss": 2.505557060241699, + "step": 215600 + }, + { + "epoch": 0.5869606642166871, + "grad_norm": 2.1154885292053223, + "learning_rate": 8.272100313479625e-06, + "loss": 2.447328567504883, + "step": 215620 + }, + { + "epoch": 0.5870151082074316, + "grad_norm": 2.183616876602173, + "learning_rate": 8.27100994957067e-06, + "loss": 2.473341369628906, + "step": 215640 + }, + { + "epoch": 0.5870695521981761, + "grad_norm": 2.077892303466797, + "learning_rate": 8.269919585661716e-06, + "loss": 2.486439895629883, + "step": 215660 + }, + { + "epoch": 0.5871239961889206, + "grad_norm": 2.2350099086761475, + "learning_rate": 8.26882922175276e-06, + "loss": 2.4424251556396483, + "step": 215680 + }, + { + "epoch": 0.5871784401796651, + "grad_norm": 2.1382644176483154, + "learning_rate": 8.267738857843806e-06, + "loss": 2.4823024749755858, + "step": 215700 + }, + { + "epoch": 0.5872328841704096, + "grad_norm": 2.181952714920044, + "learning_rate": 8.266648493934852e-06, + "loss": 2.4916257858276367, + "step": 215720 + }, + { + "epoch": 0.5872873281611543, + "grad_norm": 2.2277469635009766, + "learning_rate": 8.265558130025898e-06, + "loss": 2.4875665664672852, + "step": 215740 + }, + { + "epoch": 0.5873417721518988, + "grad_norm": 2.1033334732055664, + "learning_rate": 8.264467766116942e-06, + "loss": 2.5014017105102537, + "step": 215760 + }, + { + "epoch": 0.5873962161426433, + "grad_norm": 2.2917821407318115, + "learning_rate": 8.263377402207988e-06, + "loss": 2.4668012619018556, + "step": 215780 + }, + { + "epoch": 0.5874506601333878, + "grad_norm": 2.2666807174682617, + "learning_rate": 8.262287038299033e-06, + "loss": 2.5182016372680662, + "step": 215800 + }, + { + "epoch": 0.5875051041241323, + "grad_norm": 2.1231002807617188, + "learning_rate": 8.26119667439008e-06, + "loss": 2.4515981674194336, + "step": 215820 + }, + { + "epoch": 0.5875595481148769, + "grad_norm": 2.228879690170288, + "learning_rate": 8.260106310481123e-06, + "loss": 2.4592927932739257, + "step": 215840 + }, + { + "epoch": 0.5876139921056214, + "grad_norm": 2.149174690246582, + "learning_rate": 8.259015946572169e-06, + "loss": 2.478495216369629, + "step": 215860 + }, + { + "epoch": 0.5876684360963659, + "grad_norm": 2.1821448802948, + "learning_rate": 8.257925582663215e-06, + "loss": 2.4744972229003905, + "step": 215880 + }, + { + "epoch": 0.5877228800871104, + "grad_norm": 2.173574209213257, + "learning_rate": 8.25683521875426e-06, + "loss": 2.461722183227539, + "step": 215900 + }, + { + "epoch": 0.5877773240778549, + "grad_norm": 2.1786255836486816, + "learning_rate": 8.255744854845305e-06, + "loss": 2.4903358459472655, + "step": 215920 + }, + { + "epoch": 0.5878317680685994, + "grad_norm": 2.1526644229888916, + "learning_rate": 8.25465449093635e-06, + "loss": 2.4847557067871096, + "step": 215940 + }, + { + "epoch": 0.587886212059344, + "grad_norm": 2.1224610805511475, + "learning_rate": 8.253564127027396e-06, + "loss": 2.467436981201172, + "step": 215960 + }, + { + "epoch": 0.5879406560500885, + "grad_norm": 2.0840399265289307, + "learning_rate": 8.252473763118442e-06, + "loss": 2.491125297546387, + "step": 215980 + }, + { + "epoch": 0.587995100040833, + "grad_norm": 2.1806862354278564, + "learning_rate": 8.251383399209486e-06, + "loss": 2.4922292709350584, + "step": 216000 + }, + { + "epoch": 0.5880495440315775, + "grad_norm": 2.101860761642456, + "learning_rate": 8.250293035300532e-06, + "loss": 2.498818588256836, + "step": 216020 + }, + { + "epoch": 0.588103988022322, + "grad_norm": 2.1135973930358887, + "learning_rate": 8.249202671391578e-06, + "loss": 2.4712400436401367, + "step": 216040 + }, + { + "epoch": 0.5881584320130666, + "grad_norm": 2.063612937927246, + "learning_rate": 8.248112307482624e-06, + "loss": 2.45143985748291, + "step": 216060 + }, + { + "epoch": 0.5882128760038111, + "grad_norm": 2.1269938945770264, + "learning_rate": 8.247021943573668e-06, + "loss": 2.488166618347168, + "step": 216080 + }, + { + "epoch": 0.5882673199945556, + "grad_norm": 2.1305692195892334, + "learning_rate": 8.245931579664714e-06, + "loss": 2.5012014389038084, + "step": 216100 + }, + { + "epoch": 0.5883217639853001, + "grad_norm": 2.113238573074341, + "learning_rate": 8.24484121575576e-06, + "loss": 2.4716188430786135, + "step": 216120 + }, + { + "epoch": 0.5883762079760446, + "grad_norm": 2.085029125213623, + "learning_rate": 8.243750851846805e-06, + "loss": 2.4815750122070312, + "step": 216140 + }, + { + "epoch": 0.5884306519667891, + "grad_norm": 2.114969253540039, + "learning_rate": 8.24266048793785e-06, + "loss": 2.469661331176758, + "step": 216160 + }, + { + "epoch": 0.5884850959575337, + "grad_norm": 2.095522165298462, + "learning_rate": 8.241570124028895e-06, + "loss": 2.4721294403076173, + "step": 216180 + }, + { + "epoch": 0.5885395399482782, + "grad_norm": 2.237086057662964, + "learning_rate": 8.240479760119941e-06, + "loss": 2.5134424209594726, + "step": 216200 + }, + { + "epoch": 0.5885939839390227, + "grad_norm": 2.1323153972625732, + "learning_rate": 8.239389396210987e-06, + "loss": 2.46911563873291, + "step": 216220 + }, + { + "epoch": 0.5886484279297672, + "grad_norm": 2.1114373207092285, + "learning_rate": 8.238299032302031e-06, + "loss": 2.5099065780639647, + "step": 216240 + }, + { + "epoch": 0.5887028719205117, + "grad_norm": 2.1696841716766357, + "learning_rate": 8.237208668393077e-06, + "loss": 2.4977962493896486, + "step": 216260 + }, + { + "epoch": 0.5887573159112562, + "grad_norm": 2.1900932788848877, + "learning_rate": 8.236118304484123e-06, + "loss": 2.4743146896362305, + "step": 216280 + }, + { + "epoch": 0.5888117599020009, + "grad_norm": 2.165904998779297, + "learning_rate": 8.235027940575168e-06, + "loss": 2.4586135864257814, + "step": 216300 + }, + { + "epoch": 0.5888662038927454, + "grad_norm": 2.3272714614868164, + "learning_rate": 8.233937576666213e-06, + "loss": 2.467806816101074, + "step": 216320 + }, + { + "epoch": 0.5889206478834899, + "grad_norm": 2.0690293312072754, + "learning_rate": 8.232847212757258e-06, + "loss": 2.4814990997314452, + "step": 216340 + }, + { + "epoch": 0.5889750918742344, + "grad_norm": 2.086118698120117, + "learning_rate": 8.231756848848304e-06, + "loss": 2.4628992080688477, + "step": 216360 + }, + { + "epoch": 0.5890295358649789, + "grad_norm": 2.1454782485961914, + "learning_rate": 8.23066648493935e-06, + "loss": 2.432832145690918, + "step": 216380 + }, + { + "epoch": 0.5890839798557235, + "grad_norm": 2.2131290435791016, + "learning_rate": 8.229576121030394e-06, + "loss": 2.4863372802734376, + "step": 216400 + }, + { + "epoch": 0.589138423846468, + "grad_norm": 2.1426124572753906, + "learning_rate": 8.22848575712144e-06, + "loss": 2.4880243301391602, + "step": 216420 + }, + { + "epoch": 0.5891928678372125, + "grad_norm": 2.165955066680908, + "learning_rate": 8.227395393212486e-06, + "loss": 2.49533748626709, + "step": 216440 + }, + { + "epoch": 0.589247311827957, + "grad_norm": 2.337977409362793, + "learning_rate": 8.22630502930353e-06, + "loss": 2.4617795944213867, + "step": 216460 + }, + { + "epoch": 0.5893017558187015, + "grad_norm": 2.2528090476989746, + "learning_rate": 8.225214665394576e-06, + "loss": 2.470388984680176, + "step": 216480 + }, + { + "epoch": 0.589356199809446, + "grad_norm": 2.084782838821411, + "learning_rate": 8.224124301485621e-06, + "loss": 2.4618568420410156, + "step": 216500 + }, + { + "epoch": 0.5894106438001906, + "grad_norm": 2.126702308654785, + "learning_rate": 8.223033937576667e-06, + "loss": 2.4561147689819336, + "step": 216520 + }, + { + "epoch": 0.5894650877909351, + "grad_norm": 2.2693629264831543, + "learning_rate": 8.221943573667711e-06, + "loss": 2.4711635589599608, + "step": 216540 + }, + { + "epoch": 0.5895195317816796, + "grad_norm": 2.231799602508545, + "learning_rate": 8.220853209758757e-06, + "loss": 2.5113462448120116, + "step": 216560 + }, + { + "epoch": 0.5895739757724241, + "grad_norm": 2.129127025604248, + "learning_rate": 8.219762845849803e-06, + "loss": 2.4831287384033205, + "step": 216580 + }, + { + "epoch": 0.5896284197631686, + "grad_norm": 2.2655603885650635, + "learning_rate": 8.218672481940849e-06, + "loss": 2.4529464721679686, + "step": 216600 + }, + { + "epoch": 0.5896828637539132, + "grad_norm": 2.142815589904785, + "learning_rate": 8.217582118031893e-06, + "loss": 2.494071197509766, + "step": 216620 + }, + { + "epoch": 0.5897373077446577, + "grad_norm": 2.121079444885254, + "learning_rate": 8.216491754122939e-06, + "loss": 2.4654035568237305, + "step": 216640 + }, + { + "epoch": 0.5897917517354022, + "grad_norm": 2.2586424350738525, + "learning_rate": 8.215401390213985e-06, + "loss": 2.4836252212524412, + "step": 216660 + }, + { + "epoch": 0.5898461957261467, + "grad_norm": 2.07490611076355, + "learning_rate": 8.21431102630503e-06, + "loss": 2.456072998046875, + "step": 216680 + }, + { + "epoch": 0.5899006397168912, + "grad_norm": 2.283705472946167, + "learning_rate": 8.213220662396074e-06, + "loss": 2.485101509094238, + "step": 216700 + }, + { + "epoch": 0.5899550837076357, + "grad_norm": 2.100749969482422, + "learning_rate": 8.21213029848712e-06, + "loss": 2.456799125671387, + "step": 216720 + }, + { + "epoch": 0.5900095276983803, + "grad_norm": 2.3250787258148193, + "learning_rate": 8.211039934578166e-06, + "loss": 2.4598533630371096, + "step": 216740 + }, + { + "epoch": 0.5900639716891248, + "grad_norm": 2.169616222381592, + "learning_rate": 8.209949570669212e-06, + "loss": 2.4905363082885743, + "step": 216760 + }, + { + "epoch": 0.5901184156798693, + "grad_norm": 2.1205716133117676, + "learning_rate": 8.208859206760256e-06, + "loss": 2.4739097595214843, + "step": 216780 + }, + { + "epoch": 0.5901728596706138, + "grad_norm": 2.1718852519989014, + "learning_rate": 8.207768842851302e-06, + "loss": 2.4776830673217773, + "step": 216800 + }, + { + "epoch": 0.5902273036613583, + "grad_norm": 2.284281015396118, + "learning_rate": 8.206678478942348e-06, + "loss": 2.480491828918457, + "step": 216820 + }, + { + "epoch": 0.590281747652103, + "grad_norm": 2.1209356784820557, + "learning_rate": 8.205588115033393e-06, + "loss": 2.5118263244628904, + "step": 216840 + }, + { + "epoch": 0.5903361916428475, + "grad_norm": 2.2497668266296387, + "learning_rate": 8.204497751124438e-06, + "loss": 2.475245475769043, + "step": 216860 + }, + { + "epoch": 0.590390635633592, + "grad_norm": 2.0802688598632812, + "learning_rate": 8.203407387215483e-06, + "loss": 2.4781206130981444, + "step": 216880 + }, + { + "epoch": 0.5904450796243365, + "grad_norm": 2.2033212184906006, + "learning_rate": 8.20231702330653e-06, + "loss": 2.4604673385620117, + "step": 216900 + }, + { + "epoch": 0.590499523615081, + "grad_norm": 2.307016611099243, + "learning_rate": 8.201226659397575e-06, + "loss": 2.4906099319458006, + "step": 216920 + }, + { + "epoch": 0.5905539676058255, + "grad_norm": 2.318693161010742, + "learning_rate": 8.200136295488619e-06, + "loss": 2.4585493087768553, + "step": 216940 + }, + { + "epoch": 0.5906084115965701, + "grad_norm": 2.286259412765503, + "learning_rate": 8.199045931579665e-06, + "loss": 2.4694957733154297, + "step": 216960 + }, + { + "epoch": 0.5906628555873146, + "grad_norm": 2.195833444595337, + "learning_rate": 8.19795556767071e-06, + "loss": 2.450814628601074, + "step": 216980 + }, + { + "epoch": 0.5907172995780591, + "grad_norm": 2.1276121139526367, + "learning_rate": 8.196865203761757e-06, + "loss": 2.4506912231445312, + "step": 217000 + }, + { + "epoch": 0.5907717435688036, + "grad_norm": 2.1731479167938232, + "learning_rate": 8.1957748398528e-06, + "loss": 2.471074867248535, + "step": 217020 + }, + { + "epoch": 0.5908261875595481, + "grad_norm": 2.214794158935547, + "learning_rate": 8.194684475943846e-06, + "loss": 2.501359558105469, + "step": 217040 + }, + { + "epoch": 0.5908806315502927, + "grad_norm": 2.154249429702759, + "learning_rate": 8.193594112034892e-06, + "loss": 2.437612533569336, + "step": 217060 + }, + { + "epoch": 0.5909350755410372, + "grad_norm": 2.159273386001587, + "learning_rate": 8.192503748125938e-06, + "loss": 2.4717283248901367, + "step": 217080 + }, + { + "epoch": 0.5909895195317817, + "grad_norm": 2.089019298553467, + "learning_rate": 8.191413384216982e-06, + "loss": 2.4614866256713865, + "step": 217100 + }, + { + "epoch": 0.5910439635225262, + "grad_norm": 2.1914689540863037, + "learning_rate": 8.190323020308028e-06, + "loss": 2.45946044921875, + "step": 217120 + }, + { + "epoch": 0.5910984075132707, + "grad_norm": 2.0721638202667236, + "learning_rate": 8.189232656399074e-06, + "loss": 2.4800703048706056, + "step": 217140 + }, + { + "epoch": 0.5911528515040152, + "grad_norm": 2.1020095348358154, + "learning_rate": 8.18814229249012e-06, + "loss": 2.4920515060424804, + "step": 217160 + }, + { + "epoch": 0.5912072954947598, + "grad_norm": 2.1071510314941406, + "learning_rate": 8.187051928581164e-06, + "loss": 2.461027145385742, + "step": 217180 + }, + { + "epoch": 0.5912617394855043, + "grad_norm": 2.2939987182617188, + "learning_rate": 8.18596156467221e-06, + "loss": 2.483852767944336, + "step": 217200 + }, + { + "epoch": 0.5913161834762488, + "grad_norm": 2.2152438163757324, + "learning_rate": 8.184871200763255e-06, + "loss": 2.4658763885498045, + "step": 217220 + }, + { + "epoch": 0.5913706274669933, + "grad_norm": 2.1408653259277344, + "learning_rate": 8.183780836854301e-06, + "loss": 2.5044246673583985, + "step": 217240 + }, + { + "epoch": 0.5914250714577378, + "grad_norm": 2.09879994392395, + "learning_rate": 8.182690472945345e-06, + "loss": 2.4680532455444335, + "step": 217260 + }, + { + "epoch": 0.5914795154484824, + "grad_norm": 2.1524834632873535, + "learning_rate": 8.181600109036391e-06, + "loss": 2.4919052124023438, + "step": 217280 + }, + { + "epoch": 0.5915339594392269, + "grad_norm": 2.4568686485290527, + "learning_rate": 8.180509745127437e-06, + "loss": 2.4463178634643556, + "step": 217300 + }, + { + "epoch": 0.5915884034299714, + "grad_norm": 2.4895665645599365, + "learning_rate": 8.179419381218483e-06, + "loss": 2.4936609268188477, + "step": 217320 + }, + { + "epoch": 0.5916428474207159, + "grad_norm": 2.224135398864746, + "learning_rate": 8.178329017309527e-06, + "loss": 2.4840007781982423, + "step": 217340 + }, + { + "epoch": 0.5916972914114604, + "grad_norm": 2.0975840091705322, + "learning_rate": 8.177238653400573e-06, + "loss": 2.4980491638183593, + "step": 217360 + }, + { + "epoch": 0.5917517354022049, + "grad_norm": 2.1962687969207764, + "learning_rate": 8.176148289491618e-06, + "loss": 2.46655387878418, + "step": 217380 + }, + { + "epoch": 0.5918061793929495, + "grad_norm": 2.3701367378234863, + "learning_rate": 8.175057925582664e-06, + "loss": 2.502548408508301, + "step": 217400 + }, + { + "epoch": 0.591860623383694, + "grad_norm": 2.1409811973571777, + "learning_rate": 8.173967561673708e-06, + "loss": 2.478425216674805, + "step": 217420 + }, + { + "epoch": 0.5919150673744386, + "grad_norm": 2.2104296684265137, + "learning_rate": 8.172877197764754e-06, + "loss": 2.4679452896118166, + "step": 217440 + }, + { + "epoch": 0.591969511365183, + "grad_norm": 2.106236457824707, + "learning_rate": 8.1717868338558e-06, + "loss": 2.5153236389160156, + "step": 217460 + }, + { + "epoch": 0.5920239553559276, + "grad_norm": 2.1061911582946777, + "learning_rate": 8.170696469946846e-06, + "loss": 2.46085262298584, + "step": 217480 + }, + { + "epoch": 0.5920783993466721, + "grad_norm": 2.107120990753174, + "learning_rate": 8.16960610603789e-06, + "loss": 2.4597246170043947, + "step": 217500 + }, + { + "epoch": 0.5921328433374167, + "grad_norm": 2.3414196968078613, + "learning_rate": 8.168515742128936e-06, + "loss": 2.4564985275268554, + "step": 217520 + }, + { + "epoch": 0.5921872873281612, + "grad_norm": 2.176755428314209, + "learning_rate": 8.167425378219982e-06, + "loss": 2.4686357498168947, + "step": 217540 + }, + { + "epoch": 0.5922417313189057, + "grad_norm": 2.041886329650879, + "learning_rate": 8.166335014311026e-06, + "loss": 2.4556854248046873, + "step": 217560 + }, + { + "epoch": 0.5922961753096502, + "grad_norm": 2.3008205890655518, + "learning_rate": 8.165244650402071e-06, + "loss": 2.476040840148926, + "step": 217580 + }, + { + "epoch": 0.5923506193003947, + "grad_norm": 2.1798341274261475, + "learning_rate": 8.164154286493117e-06, + "loss": 2.4717273712158203, + "step": 217600 + }, + { + "epoch": 0.5924050632911393, + "grad_norm": 2.235863208770752, + "learning_rate": 8.163063922584163e-06, + "loss": 2.4666265487670898, + "step": 217620 + }, + { + "epoch": 0.5924595072818838, + "grad_norm": 2.065615653991699, + "learning_rate": 8.161973558675207e-06, + "loss": 2.453627014160156, + "step": 217640 + }, + { + "epoch": 0.5925139512726283, + "grad_norm": 2.0337414741516113, + "learning_rate": 8.160883194766253e-06, + "loss": 2.494005584716797, + "step": 217660 + }, + { + "epoch": 0.5925683952633728, + "grad_norm": 2.078735113143921, + "learning_rate": 8.159792830857299e-06, + "loss": 2.473862648010254, + "step": 217680 + }, + { + "epoch": 0.5926228392541173, + "grad_norm": 2.2153286933898926, + "learning_rate": 8.158702466948345e-06, + "loss": 2.469241142272949, + "step": 217700 + }, + { + "epoch": 0.5926772832448618, + "grad_norm": 2.090198040008545, + "learning_rate": 8.157612103039389e-06, + "loss": 2.4912464141845705, + "step": 217720 + }, + { + "epoch": 0.5927317272356064, + "grad_norm": 2.124727964401245, + "learning_rate": 8.156521739130435e-06, + "loss": 2.4959617614746095, + "step": 217740 + }, + { + "epoch": 0.5927861712263509, + "grad_norm": 2.543193817138672, + "learning_rate": 8.15543137522148e-06, + "loss": 2.4781213760375977, + "step": 217760 + }, + { + "epoch": 0.5928406152170954, + "grad_norm": 2.131837844848633, + "learning_rate": 8.154341011312526e-06, + "loss": 2.451436996459961, + "step": 217780 + }, + { + "epoch": 0.5928950592078399, + "grad_norm": 2.369882345199585, + "learning_rate": 8.15325064740357e-06, + "loss": 2.467333221435547, + "step": 217800 + }, + { + "epoch": 0.5929495031985844, + "grad_norm": 2.152560234069824, + "learning_rate": 8.152160283494616e-06, + "loss": 2.472537040710449, + "step": 217820 + }, + { + "epoch": 0.593003947189329, + "grad_norm": 2.154686212539673, + "learning_rate": 8.151069919585662e-06, + "loss": 2.4734458923339844, + "step": 217840 + }, + { + "epoch": 0.5930583911800735, + "grad_norm": 2.256990671157837, + "learning_rate": 8.149979555676708e-06, + "loss": 2.4767847061157227, + "step": 217860 + }, + { + "epoch": 0.593112835170818, + "grad_norm": 2.1287567615509033, + "learning_rate": 8.148889191767752e-06, + "loss": 2.4478368759155273, + "step": 217880 + }, + { + "epoch": 0.5931672791615625, + "grad_norm": 2.1668918132781982, + "learning_rate": 8.147798827858798e-06, + "loss": 2.4491094589233398, + "step": 217900 + }, + { + "epoch": 0.593221723152307, + "grad_norm": 2.163724660873413, + "learning_rate": 8.146708463949843e-06, + "loss": 2.4600341796875, + "step": 217920 + }, + { + "epoch": 0.5932761671430515, + "grad_norm": 2.1635384559631348, + "learning_rate": 8.14561810004089e-06, + "loss": 2.4624826431274416, + "step": 217940 + }, + { + "epoch": 0.5933306111337961, + "grad_norm": 2.2004470825195312, + "learning_rate": 8.144527736131933e-06, + "loss": 2.4569082260131836, + "step": 217960 + }, + { + "epoch": 0.5933850551245407, + "grad_norm": 2.174482822418213, + "learning_rate": 8.14343737222298e-06, + "loss": 2.4974695205688477, + "step": 217980 + }, + { + "epoch": 0.5934394991152852, + "grad_norm": 2.150481939315796, + "learning_rate": 8.142347008314025e-06, + "loss": 2.476754379272461, + "step": 218000 + }, + { + "epoch": 0.5934939431060297, + "grad_norm": 2.0899319648742676, + "learning_rate": 8.14125664440507e-06, + "loss": 2.475808334350586, + "step": 218020 + }, + { + "epoch": 0.5935483870967742, + "grad_norm": 2.486340045928955, + "learning_rate": 8.140166280496117e-06, + "loss": 2.4838680267333983, + "step": 218040 + }, + { + "epoch": 0.5936028310875188, + "grad_norm": 2.0665628910064697, + "learning_rate": 8.13907591658716e-06, + "loss": 2.4588043212890627, + "step": 218060 + }, + { + "epoch": 0.5936572750782633, + "grad_norm": 2.1457102298736572, + "learning_rate": 8.137985552678207e-06, + "loss": 2.4680448532104493, + "step": 218080 + }, + { + "epoch": 0.5937117190690078, + "grad_norm": 2.134225606918335, + "learning_rate": 8.136895188769252e-06, + "loss": 2.4702342987060546, + "step": 218100 + }, + { + "epoch": 0.5937661630597523, + "grad_norm": 2.104156732559204, + "learning_rate": 8.135804824860298e-06, + "loss": 2.484881782531738, + "step": 218120 + }, + { + "epoch": 0.5938206070504968, + "grad_norm": 2.1977131366729736, + "learning_rate": 8.134714460951342e-06, + "loss": 2.4897626876831054, + "step": 218140 + }, + { + "epoch": 0.5938750510412413, + "grad_norm": 2.189688205718994, + "learning_rate": 8.133624097042388e-06, + "loss": 2.437734603881836, + "step": 218160 + }, + { + "epoch": 0.5939294950319859, + "grad_norm": 2.113797187805176, + "learning_rate": 8.132533733133434e-06, + "loss": 2.479808044433594, + "step": 218180 + }, + { + "epoch": 0.5939839390227304, + "grad_norm": 2.233002185821533, + "learning_rate": 8.13144336922448e-06, + "loss": 2.4848121643066405, + "step": 218200 + }, + { + "epoch": 0.5940383830134749, + "grad_norm": 2.329958915710449, + "learning_rate": 8.130353005315524e-06, + "loss": 2.469046783447266, + "step": 218220 + }, + { + "epoch": 0.5940928270042194, + "grad_norm": 2.196788787841797, + "learning_rate": 8.12926264140657e-06, + "loss": 2.4532678604125975, + "step": 218240 + }, + { + "epoch": 0.5941472709949639, + "grad_norm": 2.176988363265991, + "learning_rate": 8.128172277497615e-06, + "loss": 2.459280014038086, + "step": 218260 + }, + { + "epoch": 0.5942017149857085, + "grad_norm": 2.3697352409362793, + "learning_rate": 8.127081913588661e-06, + "loss": 2.470853805541992, + "step": 218280 + }, + { + "epoch": 0.594256158976453, + "grad_norm": 2.298840045928955, + "learning_rate": 8.125991549679705e-06, + "loss": 2.477523422241211, + "step": 218300 + }, + { + "epoch": 0.5943106029671975, + "grad_norm": 2.101813316345215, + "learning_rate": 8.124901185770751e-06, + "loss": 2.4894426345825194, + "step": 218320 + }, + { + "epoch": 0.594365046957942, + "grad_norm": 2.080925226211548, + "learning_rate": 8.123810821861797e-06, + "loss": 2.479775047302246, + "step": 218340 + }, + { + "epoch": 0.5944194909486865, + "grad_norm": 2.280243396759033, + "learning_rate": 8.122720457952843e-06, + "loss": 2.468532943725586, + "step": 218360 + }, + { + "epoch": 0.594473934939431, + "grad_norm": 2.1682963371276855, + "learning_rate": 8.121630094043887e-06, + "loss": 2.447544288635254, + "step": 218380 + }, + { + "epoch": 0.5945283789301756, + "grad_norm": 2.207131862640381, + "learning_rate": 8.120539730134933e-06, + "loss": 2.4543155670166015, + "step": 218400 + }, + { + "epoch": 0.5945828229209201, + "grad_norm": 2.2483882904052734, + "learning_rate": 8.119449366225979e-06, + "loss": 2.5048830032348635, + "step": 218420 + }, + { + "epoch": 0.5946372669116646, + "grad_norm": 2.147325277328491, + "learning_rate": 8.118359002317024e-06, + "loss": 2.4557987213134767, + "step": 218440 + }, + { + "epoch": 0.5946917109024091, + "grad_norm": 2.3937251567840576, + "learning_rate": 8.117268638408068e-06, + "loss": 2.5017187118530275, + "step": 218460 + }, + { + "epoch": 0.5947461548931536, + "grad_norm": 2.0280911922454834, + "learning_rate": 8.116178274499114e-06, + "loss": 2.4875871658325197, + "step": 218480 + }, + { + "epoch": 0.5948005988838981, + "grad_norm": 2.252700090408325, + "learning_rate": 8.11508791059016e-06, + "loss": 2.472140312194824, + "step": 218500 + }, + { + "epoch": 0.5948550428746427, + "grad_norm": 2.193455934524536, + "learning_rate": 8.113997546681206e-06, + "loss": 2.4634366989135743, + "step": 218520 + }, + { + "epoch": 0.5949094868653872, + "grad_norm": 2.157155990600586, + "learning_rate": 8.11290718277225e-06, + "loss": 2.494125747680664, + "step": 218540 + }, + { + "epoch": 0.5949639308561318, + "grad_norm": 2.1736655235290527, + "learning_rate": 8.111816818863296e-06, + "loss": 2.4716302871704103, + "step": 218560 + }, + { + "epoch": 0.5950183748468763, + "grad_norm": 2.185420513153076, + "learning_rate": 8.110726454954342e-06, + "loss": 2.475352096557617, + "step": 218580 + }, + { + "epoch": 0.5950728188376208, + "grad_norm": 2.157942056655884, + "learning_rate": 8.109636091045387e-06, + "loss": 2.4899850845336915, + "step": 218600 + }, + { + "epoch": 0.5951272628283654, + "grad_norm": 2.3369178771972656, + "learning_rate": 8.108545727136432e-06, + "loss": 2.4737880706787108, + "step": 218620 + }, + { + "epoch": 0.5951817068191099, + "grad_norm": 2.1069302558898926, + "learning_rate": 8.107455363227477e-06, + "loss": 2.4398202896118164, + "step": 218640 + }, + { + "epoch": 0.5952361508098544, + "grad_norm": 2.5434205532073975, + "learning_rate": 8.106364999318523e-06, + "loss": 2.4820463180541994, + "step": 218660 + }, + { + "epoch": 0.5952905948005989, + "grad_norm": 2.2628254890441895, + "learning_rate": 8.105274635409569e-06, + "loss": 2.5079198837280274, + "step": 218680 + }, + { + "epoch": 0.5953450387913434, + "grad_norm": 2.154600143432617, + "learning_rate": 8.104184271500613e-06, + "loss": 2.4391555786132812, + "step": 218700 + }, + { + "epoch": 0.5953994827820879, + "grad_norm": 2.1703388690948486, + "learning_rate": 8.103093907591659e-06, + "loss": 2.4668588638305664, + "step": 218720 + }, + { + "epoch": 0.5954539267728325, + "grad_norm": 2.3117806911468506, + "learning_rate": 8.102003543682705e-06, + "loss": 2.4949108123779298, + "step": 218740 + }, + { + "epoch": 0.595508370763577, + "grad_norm": 2.769573450088501, + "learning_rate": 8.10091317977375e-06, + "loss": 2.484870719909668, + "step": 218760 + }, + { + "epoch": 0.5955628147543215, + "grad_norm": 2.184654951095581, + "learning_rate": 8.099822815864795e-06, + "loss": 2.475513458251953, + "step": 218780 + }, + { + "epoch": 0.595617258745066, + "grad_norm": 2.1884546279907227, + "learning_rate": 8.09873245195584e-06, + "loss": 2.4699018478393553, + "step": 218800 + }, + { + "epoch": 0.5956717027358105, + "grad_norm": 2.1952943801879883, + "learning_rate": 8.097642088046886e-06, + "loss": 2.477128028869629, + "step": 218820 + }, + { + "epoch": 0.5957261467265551, + "grad_norm": 2.4077653884887695, + "learning_rate": 8.096551724137932e-06, + "loss": 2.468856430053711, + "step": 218840 + }, + { + "epoch": 0.5957805907172996, + "grad_norm": 2.4551467895507812, + "learning_rate": 8.095461360228976e-06, + "loss": 2.4508495330810547, + "step": 218860 + }, + { + "epoch": 0.5958350347080441, + "grad_norm": 2.164339303970337, + "learning_rate": 8.094370996320022e-06, + "loss": 2.4448062896728517, + "step": 218880 + }, + { + "epoch": 0.5958894786987886, + "grad_norm": 2.2143025398254395, + "learning_rate": 8.093280632411068e-06, + "loss": 2.459735107421875, + "step": 218900 + }, + { + "epoch": 0.5959439226895331, + "grad_norm": 2.081871271133423, + "learning_rate": 8.092190268502114e-06, + "loss": 2.5101245880126952, + "step": 218920 + }, + { + "epoch": 0.5959983666802776, + "grad_norm": 2.1374564170837402, + "learning_rate": 8.091099904593158e-06, + "loss": 2.4723861694335936, + "step": 218940 + }, + { + "epoch": 0.5960528106710222, + "grad_norm": 2.088373899459839, + "learning_rate": 8.090009540684203e-06, + "loss": 2.462031936645508, + "step": 218960 + }, + { + "epoch": 0.5961072546617667, + "grad_norm": 2.134477376937866, + "learning_rate": 8.08891917677525e-06, + "loss": 2.473693084716797, + "step": 218980 + }, + { + "epoch": 0.5961616986525112, + "grad_norm": 2.2166621685028076, + "learning_rate": 8.087828812866295e-06, + "loss": 2.487503433227539, + "step": 219000 + }, + { + "epoch": 0.5962161426432557, + "grad_norm": 2.187328338623047, + "learning_rate": 8.08673844895734e-06, + "loss": 2.503853416442871, + "step": 219020 + }, + { + "epoch": 0.5962705866340002, + "grad_norm": 2.105696678161621, + "learning_rate": 8.085648085048385e-06, + "loss": 2.484465980529785, + "step": 219040 + }, + { + "epoch": 0.5963250306247448, + "grad_norm": 2.1854710578918457, + "learning_rate": 8.08455772113943e-06, + "loss": 2.4538095474243162, + "step": 219060 + }, + { + "epoch": 0.5963794746154893, + "grad_norm": 2.710571765899658, + "learning_rate": 8.083467357230477e-06, + "loss": 2.4705921173095704, + "step": 219080 + }, + { + "epoch": 0.5964339186062338, + "grad_norm": 2.1219987869262695, + "learning_rate": 8.08237699332152e-06, + "loss": 2.483857536315918, + "step": 219100 + }, + { + "epoch": 0.5964883625969783, + "grad_norm": 2.908561944961548, + "learning_rate": 8.081286629412567e-06, + "loss": 2.4922719955444337, + "step": 219120 + }, + { + "epoch": 0.5965428065877229, + "grad_norm": 2.260305881500244, + "learning_rate": 8.080196265503612e-06, + "loss": 2.483135986328125, + "step": 219140 + }, + { + "epoch": 0.5965972505784674, + "grad_norm": 2.1341471672058105, + "learning_rate": 8.079105901594658e-06, + "loss": 2.4838930130004884, + "step": 219160 + }, + { + "epoch": 0.596651694569212, + "grad_norm": 2.1311845779418945, + "learning_rate": 8.078015537685702e-06, + "loss": 2.459311676025391, + "step": 219180 + }, + { + "epoch": 0.5967061385599565, + "grad_norm": 2.173304319381714, + "learning_rate": 8.076925173776748e-06, + "loss": 2.4827693939208983, + "step": 219200 + }, + { + "epoch": 0.596760582550701, + "grad_norm": 2.141842842102051, + "learning_rate": 8.075834809867794e-06, + "loss": 2.453985595703125, + "step": 219220 + }, + { + "epoch": 0.5968150265414455, + "grad_norm": 2.189391851425171, + "learning_rate": 8.07474444595884e-06, + "loss": 2.475797653198242, + "step": 219240 + }, + { + "epoch": 0.59686947053219, + "grad_norm": 2.6250007152557373, + "learning_rate": 8.073654082049884e-06, + "loss": 2.4698759078979493, + "step": 219260 + }, + { + "epoch": 0.5969239145229346, + "grad_norm": 2.2201263904571533, + "learning_rate": 8.07256371814093e-06, + "loss": 2.4812631607055664, + "step": 219280 + }, + { + "epoch": 0.5969783585136791, + "grad_norm": 2.2128660678863525, + "learning_rate": 8.071473354231975e-06, + "loss": 2.457776641845703, + "step": 219300 + }, + { + "epoch": 0.5970328025044236, + "grad_norm": 2.2537589073181152, + "learning_rate": 8.070382990323021e-06, + "loss": 2.459208106994629, + "step": 219320 + }, + { + "epoch": 0.5970872464951681, + "grad_norm": 2.226433038711548, + "learning_rate": 8.069292626414065e-06, + "loss": 2.4719623565673827, + "step": 219340 + }, + { + "epoch": 0.5971416904859126, + "grad_norm": 2.361985683441162, + "learning_rate": 8.068202262505111e-06, + "loss": 2.4555776596069334, + "step": 219360 + }, + { + "epoch": 0.5971961344766571, + "grad_norm": 2.111013650894165, + "learning_rate": 8.067111898596157e-06, + "loss": 2.499590301513672, + "step": 219380 + }, + { + "epoch": 0.5972505784674017, + "grad_norm": 2.188948154449463, + "learning_rate": 8.066021534687203e-06, + "loss": 2.442766571044922, + "step": 219400 + }, + { + "epoch": 0.5973050224581462, + "grad_norm": 2.1980063915252686, + "learning_rate": 8.064931170778247e-06, + "loss": 2.465536117553711, + "step": 219420 + }, + { + "epoch": 0.5973594664488907, + "grad_norm": 110.28980255126953, + "learning_rate": 8.063840806869293e-06, + "loss": 2.4917036056518556, + "step": 219440 + }, + { + "epoch": 0.5974139104396352, + "grad_norm": 2.1832234859466553, + "learning_rate": 8.062750442960339e-06, + "loss": 2.446877861022949, + "step": 219460 + }, + { + "epoch": 0.5974683544303797, + "grad_norm": 2.1031992435455322, + "learning_rate": 8.061660079051384e-06, + "loss": 2.4638431549072264, + "step": 219480 + }, + { + "epoch": 0.5975227984211243, + "grad_norm": 2.1079485416412354, + "learning_rate": 8.060569715142428e-06, + "loss": 2.4891616821289064, + "step": 219500 + }, + { + "epoch": 0.5975772424118688, + "grad_norm": 2.1089181900024414, + "learning_rate": 8.059479351233474e-06, + "loss": 2.4791746139526367, + "step": 219520 + }, + { + "epoch": 0.5976316864026133, + "grad_norm": 2.181751251220703, + "learning_rate": 8.05838898732452e-06, + "loss": 2.480748176574707, + "step": 219540 + }, + { + "epoch": 0.5976861303933578, + "grad_norm": 2.154898166656494, + "learning_rate": 8.057298623415566e-06, + "loss": 2.460776519775391, + "step": 219560 + }, + { + "epoch": 0.5977405743841023, + "grad_norm": 2.13852596282959, + "learning_rate": 8.05620825950661e-06, + "loss": 2.489647102355957, + "step": 219580 + }, + { + "epoch": 0.5977950183748468, + "grad_norm": 2.471036434173584, + "learning_rate": 8.055117895597656e-06, + "loss": 2.456144332885742, + "step": 219600 + }, + { + "epoch": 0.5978494623655914, + "grad_norm": 2.108710289001465, + "learning_rate": 8.054027531688702e-06, + "loss": 2.511061668395996, + "step": 219620 + }, + { + "epoch": 0.597903906356336, + "grad_norm": 2.170044422149658, + "learning_rate": 8.052937167779747e-06, + "loss": 2.455297660827637, + "step": 219640 + }, + { + "epoch": 0.5979583503470804, + "grad_norm": 2.092397689819336, + "learning_rate": 8.051846803870792e-06, + "loss": 2.4732730865478514, + "step": 219660 + }, + { + "epoch": 0.598012794337825, + "grad_norm": 2.220632791519165, + "learning_rate": 8.050756439961837e-06, + "loss": 2.478692817687988, + "step": 219680 + }, + { + "epoch": 0.5980672383285695, + "grad_norm": 2.151604652404785, + "learning_rate": 8.049666076052883e-06, + "loss": 2.45261173248291, + "step": 219700 + }, + { + "epoch": 0.598121682319314, + "grad_norm": 2.103862762451172, + "learning_rate": 8.048575712143929e-06, + "loss": 2.4979274749755858, + "step": 219720 + }, + { + "epoch": 0.5981761263100586, + "grad_norm": 2.188863515853882, + "learning_rate": 8.047485348234975e-06, + "loss": 2.4589099884033203, + "step": 219740 + }, + { + "epoch": 0.5982305703008031, + "grad_norm": 2.138280153274536, + "learning_rate": 8.046394984326019e-06, + "loss": 2.4607200622558594, + "step": 219760 + }, + { + "epoch": 0.5982850142915476, + "grad_norm": 2.1167080402374268, + "learning_rate": 8.045304620417065e-06, + "loss": 2.4920124053955077, + "step": 219780 + }, + { + "epoch": 0.5983394582822921, + "grad_norm": 2.1456966400146484, + "learning_rate": 8.04421425650811e-06, + "loss": 2.462480354309082, + "step": 219800 + }, + { + "epoch": 0.5983939022730366, + "grad_norm": 2.230834722518921, + "learning_rate": 8.043123892599156e-06, + "loss": 2.4305648803710938, + "step": 219820 + }, + { + "epoch": 0.5984483462637812, + "grad_norm": 2.130262851715088, + "learning_rate": 8.0420335286902e-06, + "loss": 2.4645177841186525, + "step": 219840 + }, + { + "epoch": 0.5985027902545257, + "grad_norm": 2.0675711631774902, + "learning_rate": 8.040943164781246e-06, + "loss": 2.4630334854125975, + "step": 219860 + }, + { + "epoch": 0.5985572342452702, + "grad_norm": 2.1969995498657227, + "learning_rate": 8.039852800872292e-06, + "loss": 2.477596473693848, + "step": 219880 + }, + { + "epoch": 0.5986116782360147, + "grad_norm": 2.2321245670318604, + "learning_rate": 8.038762436963338e-06, + "loss": 2.4976116180419923, + "step": 219900 + }, + { + "epoch": 0.5986661222267592, + "grad_norm": 2.1168203353881836, + "learning_rate": 8.037672073054382e-06, + "loss": 2.4576395034790037, + "step": 219920 + }, + { + "epoch": 0.5987205662175037, + "grad_norm": 2.1433491706848145, + "learning_rate": 8.036581709145428e-06, + "loss": 2.481031799316406, + "step": 219940 + }, + { + "epoch": 0.5987750102082483, + "grad_norm": 2.209536075592041, + "learning_rate": 8.035491345236474e-06, + "loss": 2.5279407501220703, + "step": 219960 + }, + { + "epoch": 0.5988294541989928, + "grad_norm": 2.1987359523773193, + "learning_rate": 8.03440098132752e-06, + "loss": 2.4596736907958983, + "step": 219980 + }, + { + "epoch": 0.5988838981897373, + "grad_norm": 2.1525025367736816, + "learning_rate": 8.033310617418564e-06, + "loss": 2.4649370193481444, + "step": 220000 + }, + { + "epoch": 0.5989383421804818, + "grad_norm": 2.2015321254730225, + "learning_rate": 8.03222025350961e-06, + "loss": 2.4687957763671875, + "step": 220020 + }, + { + "epoch": 0.5989927861712263, + "grad_norm": 2.183795928955078, + "learning_rate": 8.031129889600655e-06, + "loss": 2.4577545166015624, + "step": 220040 + }, + { + "epoch": 0.5990472301619709, + "grad_norm": 2.044708251953125, + "learning_rate": 8.030039525691701e-06, + "loss": 2.440840148925781, + "step": 220060 + }, + { + "epoch": 0.5991016741527154, + "grad_norm": 2.093154191970825, + "learning_rate": 8.028949161782745e-06, + "loss": 2.4822649002075194, + "step": 220080 + }, + { + "epoch": 0.5991561181434599, + "grad_norm": 2.2190842628479004, + "learning_rate": 8.027858797873791e-06, + "loss": 2.4661556243896485, + "step": 220100 + }, + { + "epoch": 0.5992105621342044, + "grad_norm": 2.228483200073242, + "learning_rate": 8.026768433964837e-06, + "loss": 2.4684169769287108, + "step": 220120 + }, + { + "epoch": 0.5992650061249489, + "grad_norm": 2.259779453277588, + "learning_rate": 8.025678070055883e-06, + "loss": 2.490193176269531, + "step": 220140 + }, + { + "epoch": 0.5993194501156934, + "grad_norm": 2.2967848777770996, + "learning_rate": 8.024587706146927e-06, + "loss": 2.453240966796875, + "step": 220160 + }, + { + "epoch": 0.599373894106438, + "grad_norm": 2.1578738689422607, + "learning_rate": 8.023497342237972e-06, + "loss": 2.4560214996337892, + "step": 220180 + }, + { + "epoch": 0.5994283380971825, + "grad_norm": 2.091529607772827, + "learning_rate": 8.022406978329018e-06, + "loss": 2.465550422668457, + "step": 220200 + }, + { + "epoch": 0.599482782087927, + "grad_norm": 2.1080105304718018, + "learning_rate": 8.021316614420064e-06, + "loss": 2.461472511291504, + "step": 220220 + }, + { + "epoch": 0.5995372260786715, + "grad_norm": 2.115679979324341, + "learning_rate": 8.020226250511108e-06, + "loss": 2.482778549194336, + "step": 220240 + }, + { + "epoch": 0.599591670069416, + "grad_norm": 2.0956640243530273, + "learning_rate": 8.019135886602154e-06, + "loss": 2.5172481536865234, + "step": 220260 + }, + { + "epoch": 0.5996461140601607, + "grad_norm": 2.2145698070526123, + "learning_rate": 8.0180455226932e-06, + "loss": 2.466878128051758, + "step": 220280 + }, + { + "epoch": 0.5997005580509052, + "grad_norm": 2.134549379348755, + "learning_rate": 8.016955158784246e-06, + "loss": 2.461012268066406, + "step": 220300 + }, + { + "epoch": 0.5997550020416497, + "grad_norm": 2.1705901622772217, + "learning_rate": 8.01586479487529e-06, + "loss": 2.4630306243896483, + "step": 220320 + }, + { + "epoch": 0.5998094460323942, + "grad_norm": 3.496246576309204, + "learning_rate": 8.014774430966336e-06, + "loss": 2.4777320861816405, + "step": 220340 + }, + { + "epoch": 0.5998638900231387, + "grad_norm": 2.4120051860809326, + "learning_rate": 8.013684067057381e-06, + "loss": 2.466420364379883, + "step": 220360 + }, + { + "epoch": 0.5999183340138832, + "grad_norm": 2.160374402999878, + "learning_rate": 8.012593703148427e-06, + "loss": 2.5015029907226562, + "step": 220380 + }, + { + "epoch": 0.5999727780046278, + "grad_norm": 2.1573903560638428, + "learning_rate": 8.011503339239471e-06, + "loss": 2.441590690612793, + "step": 220400 + }, + { + "epoch": 0.6000272219953723, + "grad_norm": 2.3243510723114014, + "learning_rate": 8.010412975330517e-06, + "loss": 2.4452289581298827, + "step": 220420 + }, + { + "epoch": 0.6000816659861168, + "grad_norm": 2.094184398651123, + "learning_rate": 8.009322611421563e-06, + "loss": 2.4484718322753904, + "step": 220440 + }, + { + "epoch": 0.6001361099768613, + "grad_norm": 2.267970561981201, + "learning_rate": 8.008232247512609e-06, + "loss": 2.461750030517578, + "step": 220460 + }, + { + "epoch": 0.6001905539676058, + "grad_norm": 2.089205265045166, + "learning_rate": 8.007141883603653e-06, + "loss": 2.4851314544677736, + "step": 220480 + }, + { + "epoch": 0.6002449979583504, + "grad_norm": 2.1584959030151367, + "learning_rate": 8.006051519694699e-06, + "loss": 2.470408630371094, + "step": 220500 + }, + { + "epoch": 0.6002994419490949, + "grad_norm": 2.2614097595214844, + "learning_rate": 8.004961155785744e-06, + "loss": 2.4579057693481445, + "step": 220520 + }, + { + "epoch": 0.6003538859398394, + "grad_norm": 2.1228432655334473, + "learning_rate": 8.00387079187679e-06, + "loss": 2.4581695556640626, + "step": 220540 + }, + { + "epoch": 0.6004083299305839, + "grad_norm": 2.1553821563720703, + "learning_rate": 8.002780427967834e-06, + "loss": 2.4987842559814455, + "step": 220560 + }, + { + "epoch": 0.6004627739213284, + "grad_norm": 2.168963670730591, + "learning_rate": 8.00169006405888e-06, + "loss": 2.4350141525268554, + "step": 220580 + }, + { + "epoch": 0.6005172179120729, + "grad_norm": 2.2084126472473145, + "learning_rate": 8.000599700149926e-06, + "loss": 2.460590362548828, + "step": 220600 + }, + { + "epoch": 0.6005716619028175, + "grad_norm": 2.1623191833496094, + "learning_rate": 7.999509336240972e-06, + "loss": 2.429291915893555, + "step": 220620 + }, + { + "epoch": 0.600626105893562, + "grad_norm": 2.058717966079712, + "learning_rate": 7.998418972332016e-06, + "loss": 2.4864133834838866, + "step": 220640 + }, + { + "epoch": 0.6006805498843065, + "grad_norm": 2.251619815826416, + "learning_rate": 7.997328608423062e-06, + "loss": 2.463117790222168, + "step": 220660 + }, + { + "epoch": 0.600734993875051, + "grad_norm": 2.438985824584961, + "learning_rate": 7.996238244514108e-06, + "loss": 2.4868534088134764, + "step": 220680 + }, + { + "epoch": 0.6007894378657955, + "grad_norm": 2.2112679481506348, + "learning_rate": 7.995147880605153e-06, + "loss": 2.465641975402832, + "step": 220700 + }, + { + "epoch": 0.60084388185654, + "grad_norm": 2.172475576400757, + "learning_rate": 7.994057516696197e-06, + "loss": 2.4832923889160154, + "step": 220720 + }, + { + "epoch": 0.6008983258472846, + "grad_norm": 2.113792657852173, + "learning_rate": 7.992967152787243e-06, + "loss": 2.46505069732666, + "step": 220740 + }, + { + "epoch": 0.6009527698380291, + "grad_norm": 2.192596435546875, + "learning_rate": 7.991876788878289e-06, + "loss": 2.4751792907714845, + "step": 220760 + }, + { + "epoch": 0.6010072138287736, + "grad_norm": 2.1901779174804688, + "learning_rate": 7.990786424969335e-06, + "loss": 2.4561880111694334, + "step": 220780 + }, + { + "epoch": 0.6010616578195181, + "grad_norm": 2.418722629547119, + "learning_rate": 7.989696061060379e-06, + "loss": 2.448636054992676, + "step": 220800 + }, + { + "epoch": 0.6011161018102626, + "grad_norm": 2.3357019424438477, + "learning_rate": 7.988605697151425e-06, + "loss": 2.478296089172363, + "step": 220820 + }, + { + "epoch": 0.6011705458010073, + "grad_norm": 2.100841760635376, + "learning_rate": 7.98751533324247e-06, + "loss": 2.485124969482422, + "step": 220840 + }, + { + "epoch": 0.6012249897917518, + "grad_norm": 2.2429094314575195, + "learning_rate": 7.986424969333516e-06, + "loss": 2.491372489929199, + "step": 220860 + }, + { + "epoch": 0.6012794337824963, + "grad_norm": 2.1153810024261475, + "learning_rate": 7.98533460542456e-06, + "loss": 2.459260940551758, + "step": 220880 + }, + { + "epoch": 0.6013338777732408, + "grad_norm": 2.1701056957244873, + "learning_rate": 7.984244241515606e-06, + "loss": 2.46456241607666, + "step": 220900 + }, + { + "epoch": 0.6013883217639853, + "grad_norm": 2.163383722305298, + "learning_rate": 7.983153877606652e-06, + "loss": 2.4647918701171876, + "step": 220920 + }, + { + "epoch": 0.6014427657547298, + "grad_norm": 2.111525774002075, + "learning_rate": 7.982063513697698e-06, + "loss": 2.5066957473754883, + "step": 220940 + }, + { + "epoch": 0.6014972097454744, + "grad_norm": 2.217855453491211, + "learning_rate": 7.980973149788742e-06, + "loss": 2.453520393371582, + "step": 220960 + }, + { + "epoch": 0.6015516537362189, + "grad_norm": 2.246790885925293, + "learning_rate": 7.979882785879788e-06, + "loss": 2.4664382934570312, + "step": 220980 + }, + { + "epoch": 0.6016060977269634, + "grad_norm": 2.2487690448760986, + "learning_rate": 7.978792421970834e-06, + "loss": 2.459224319458008, + "step": 221000 + }, + { + "epoch": 0.6016605417177079, + "grad_norm": 2.237595558166504, + "learning_rate": 7.97770205806188e-06, + "loss": 2.4577932357788086, + "step": 221020 + }, + { + "epoch": 0.6017149857084524, + "grad_norm": 2.1543729305267334, + "learning_rate": 7.976611694152924e-06, + "loss": 2.477660369873047, + "step": 221040 + }, + { + "epoch": 0.601769429699197, + "grad_norm": 2.2587599754333496, + "learning_rate": 7.97552133024397e-06, + "loss": 2.534804344177246, + "step": 221060 + }, + { + "epoch": 0.6018238736899415, + "grad_norm": 2.1329762935638428, + "learning_rate": 7.974430966335015e-06, + "loss": 2.4921758651733397, + "step": 221080 + }, + { + "epoch": 0.601878317680686, + "grad_norm": 2.1949431896209717, + "learning_rate": 7.973340602426061e-06, + "loss": 2.455002212524414, + "step": 221100 + }, + { + "epoch": 0.6019327616714305, + "grad_norm": 2.1920065879821777, + "learning_rate": 7.972250238517105e-06, + "loss": 2.481490707397461, + "step": 221120 + }, + { + "epoch": 0.601987205662175, + "grad_norm": 2.258969783782959, + "learning_rate": 7.971159874608151e-06, + "loss": 2.4500789642333984, + "step": 221140 + }, + { + "epoch": 0.6020416496529195, + "grad_norm": 2.220506429672241, + "learning_rate": 7.970069510699197e-06, + "loss": 2.4720592498779297, + "step": 221160 + }, + { + "epoch": 0.6020960936436641, + "grad_norm": 2.1913397312164307, + "learning_rate": 7.968979146790243e-06, + "loss": 2.464290237426758, + "step": 221180 + }, + { + "epoch": 0.6021505376344086, + "grad_norm": 2.2494871616363525, + "learning_rate": 7.967888782881287e-06, + "loss": 2.4493507385253905, + "step": 221200 + }, + { + "epoch": 0.6022049816251531, + "grad_norm": 2.1631722450256348, + "learning_rate": 7.966798418972333e-06, + "loss": 2.481511688232422, + "step": 221220 + }, + { + "epoch": 0.6022594256158976, + "grad_norm": 2.1169872283935547, + "learning_rate": 7.965708055063378e-06, + "loss": 2.4788379669189453, + "step": 221240 + }, + { + "epoch": 0.6023138696066421, + "grad_norm": 2.35062313079834, + "learning_rate": 7.964617691154424e-06, + "loss": 2.4688844680786133, + "step": 221260 + }, + { + "epoch": 0.6023683135973867, + "grad_norm": 2.2366223335266113, + "learning_rate": 7.963527327245468e-06, + "loss": 2.4578466415405273, + "step": 221280 + }, + { + "epoch": 0.6024227575881312, + "grad_norm": 2.3608977794647217, + "learning_rate": 7.962436963336514e-06, + "loss": 2.4635108947753905, + "step": 221300 + }, + { + "epoch": 0.6024772015788757, + "grad_norm": 2.223738431930542, + "learning_rate": 7.96134659942756e-06, + "loss": 2.484007453918457, + "step": 221320 + }, + { + "epoch": 0.6025316455696202, + "grad_norm": 2.452014923095703, + "learning_rate": 7.960256235518606e-06, + "loss": 2.483995819091797, + "step": 221340 + }, + { + "epoch": 0.6025860895603647, + "grad_norm": 2.1118814945220947, + "learning_rate": 7.95916587160965e-06, + "loss": 2.4713863372802733, + "step": 221360 + }, + { + "epoch": 0.6026405335511092, + "grad_norm": 2.094883918762207, + "learning_rate": 7.958075507700696e-06, + "loss": 2.4780168533325195, + "step": 221380 + }, + { + "epoch": 0.6026949775418539, + "grad_norm": 2.161729574203491, + "learning_rate": 7.956985143791741e-06, + "loss": 2.496307373046875, + "step": 221400 + }, + { + "epoch": 0.6027494215325984, + "grad_norm": 2.184847354888916, + "learning_rate": 7.955894779882787e-06, + "loss": 2.4387210845947265, + "step": 221420 + }, + { + "epoch": 0.6028038655233429, + "grad_norm": 2.364000082015991, + "learning_rate": 7.954804415973833e-06, + "loss": 2.4655935287475588, + "step": 221440 + }, + { + "epoch": 0.6028583095140874, + "grad_norm": 2.118638277053833, + "learning_rate": 7.953714052064877e-06, + "loss": 2.481228256225586, + "step": 221460 + }, + { + "epoch": 0.6029127535048319, + "grad_norm": 2.0991721153259277, + "learning_rate": 7.952623688155923e-06, + "loss": 2.469173049926758, + "step": 221480 + }, + { + "epoch": 0.6029671974955765, + "grad_norm": 2.1516730785369873, + "learning_rate": 7.951533324246969e-06, + "loss": 2.4817100524902345, + "step": 221500 + }, + { + "epoch": 0.603021641486321, + "grad_norm": 2.333371877670288, + "learning_rate": 7.950442960338015e-06, + "loss": 2.506794738769531, + "step": 221520 + }, + { + "epoch": 0.6030760854770655, + "grad_norm": 2.1986608505249023, + "learning_rate": 7.949352596429059e-06, + "loss": 2.4459068298339846, + "step": 221540 + }, + { + "epoch": 0.60313052946781, + "grad_norm": 2.223379373550415, + "learning_rate": 7.948262232520104e-06, + "loss": 2.5136966705322266, + "step": 221560 + }, + { + "epoch": 0.6031849734585545, + "grad_norm": 2.101457118988037, + "learning_rate": 7.94717186861115e-06, + "loss": 2.462224578857422, + "step": 221580 + }, + { + "epoch": 0.603239417449299, + "grad_norm": 2.19728946685791, + "learning_rate": 7.946081504702196e-06, + "loss": 2.462485122680664, + "step": 221600 + }, + { + "epoch": 0.6032938614400436, + "grad_norm": 2.024204730987549, + "learning_rate": 7.94499114079324e-06, + "loss": 2.470384979248047, + "step": 221620 + }, + { + "epoch": 0.6033483054307881, + "grad_norm": 2.120910167694092, + "learning_rate": 7.943900776884286e-06, + "loss": 2.465725898742676, + "step": 221640 + }, + { + "epoch": 0.6034027494215326, + "grad_norm": 2.069004535675049, + "learning_rate": 7.942810412975332e-06, + "loss": 2.4771678924560545, + "step": 221660 + }, + { + "epoch": 0.6034571934122771, + "grad_norm": 2.400251626968384, + "learning_rate": 7.941720049066378e-06, + "loss": 2.4782329559326173, + "step": 221680 + }, + { + "epoch": 0.6035116374030216, + "grad_norm": 2.0891776084899902, + "learning_rate": 7.940629685157422e-06, + "loss": 2.4348840713500977, + "step": 221700 + }, + { + "epoch": 0.6035660813937662, + "grad_norm": 2.14665150642395, + "learning_rate": 7.939539321248468e-06, + "loss": 2.480943298339844, + "step": 221720 + }, + { + "epoch": 0.6036205253845107, + "grad_norm": 2.1687824726104736, + "learning_rate": 7.938448957339513e-06, + "loss": 2.463576889038086, + "step": 221740 + }, + { + "epoch": 0.6036749693752552, + "grad_norm": 2.2393012046813965, + "learning_rate": 7.93735859343056e-06, + "loss": 2.471253204345703, + "step": 221760 + }, + { + "epoch": 0.6037294133659997, + "grad_norm": 2.1324923038482666, + "learning_rate": 7.936268229521603e-06, + "loss": 2.447624397277832, + "step": 221780 + }, + { + "epoch": 0.6037838573567442, + "grad_norm": 2.346616268157959, + "learning_rate": 7.935177865612649e-06, + "loss": 2.4662010192871096, + "step": 221800 + }, + { + "epoch": 0.6038383013474887, + "grad_norm": 2.1601457595825195, + "learning_rate": 7.934087501703695e-06, + "loss": 2.488700103759766, + "step": 221820 + }, + { + "epoch": 0.6038927453382333, + "grad_norm": 2.1001689434051514, + "learning_rate": 7.93299713779474e-06, + "loss": 2.454136276245117, + "step": 221840 + }, + { + "epoch": 0.6039471893289778, + "grad_norm": 2.1819355487823486, + "learning_rate": 7.931906773885785e-06, + "loss": 2.492924690246582, + "step": 221860 + }, + { + "epoch": 0.6040016333197223, + "grad_norm": 2.2500433921813965, + "learning_rate": 7.93081640997683e-06, + "loss": 2.4624353408813477, + "step": 221880 + }, + { + "epoch": 0.6040560773104668, + "grad_norm": 2.0904886722564697, + "learning_rate": 7.929726046067876e-06, + "loss": 2.4554601669311524, + "step": 221900 + }, + { + "epoch": 0.6041105213012113, + "grad_norm": 2.1157517433166504, + "learning_rate": 7.928635682158922e-06, + "loss": 2.469906234741211, + "step": 221920 + }, + { + "epoch": 0.6041649652919558, + "grad_norm": 2.1404871940612793, + "learning_rate": 7.927545318249966e-06, + "loss": 2.459551239013672, + "step": 221940 + }, + { + "epoch": 0.6042194092827005, + "grad_norm": 2.1172688007354736, + "learning_rate": 7.926454954341012e-06, + "loss": 2.4847450256347656, + "step": 221960 + }, + { + "epoch": 0.604273853273445, + "grad_norm": 2.1267189979553223, + "learning_rate": 7.925364590432058e-06, + "loss": 2.460490417480469, + "step": 221980 + }, + { + "epoch": 0.6043282972641895, + "grad_norm": 2.3254268169403076, + "learning_rate": 7.924274226523104e-06, + "loss": 2.4636005401611327, + "step": 222000 + }, + { + "epoch": 0.604382741254934, + "grad_norm": 2.5086846351623535, + "learning_rate": 7.923183862614148e-06, + "loss": 2.450597953796387, + "step": 222020 + }, + { + "epoch": 0.6044371852456785, + "grad_norm": 2.2201321125030518, + "learning_rate": 7.922093498705194e-06, + "loss": 2.490070915222168, + "step": 222040 + }, + { + "epoch": 0.6044916292364231, + "grad_norm": 2.229814291000366, + "learning_rate": 7.92100313479624e-06, + "loss": 2.470807647705078, + "step": 222060 + }, + { + "epoch": 0.6045460732271676, + "grad_norm": 2.1572141647338867, + "learning_rate": 7.919912770887285e-06, + "loss": 2.4716272354125977, + "step": 222080 + }, + { + "epoch": 0.6046005172179121, + "grad_norm": 2.257309913635254, + "learning_rate": 7.91882240697833e-06, + "loss": 2.4650882720947265, + "step": 222100 + }, + { + "epoch": 0.6046549612086566, + "grad_norm": 2.242492437362671, + "learning_rate": 7.917732043069375e-06, + "loss": 2.494426727294922, + "step": 222120 + }, + { + "epoch": 0.6047094051994011, + "grad_norm": 2.064016819000244, + "learning_rate": 7.916641679160421e-06, + "loss": 2.461551284790039, + "step": 222140 + }, + { + "epoch": 0.6047638491901456, + "grad_norm": 2.0851638317108154, + "learning_rate": 7.915551315251467e-06, + "loss": 2.4743810653686524, + "step": 222160 + }, + { + "epoch": 0.6048182931808902, + "grad_norm": 2.355219602584839, + "learning_rate": 7.914460951342511e-06, + "loss": 2.4662879943847655, + "step": 222180 + }, + { + "epoch": 0.6048727371716347, + "grad_norm": 2.2263998985290527, + "learning_rate": 7.913370587433557e-06, + "loss": 2.4782712936401365, + "step": 222200 + }, + { + "epoch": 0.6049271811623792, + "grad_norm": 2.0668880939483643, + "learning_rate": 7.912280223524603e-06, + "loss": 2.4918107986450195, + "step": 222220 + }, + { + "epoch": 0.6049816251531237, + "grad_norm": 2.1377079486846924, + "learning_rate": 7.911189859615648e-06, + "loss": 2.486089897155762, + "step": 222240 + }, + { + "epoch": 0.6050360691438682, + "grad_norm": 2.2467360496520996, + "learning_rate": 7.910099495706693e-06, + "loss": 2.4516790390014647, + "step": 222260 + }, + { + "epoch": 0.6050905131346128, + "grad_norm": 2.0743093490600586, + "learning_rate": 7.909009131797738e-06, + "loss": 2.487314224243164, + "step": 222280 + }, + { + "epoch": 0.6051449571253573, + "grad_norm": 2.166987895965576, + "learning_rate": 7.907918767888784e-06, + "loss": 2.463580322265625, + "step": 222300 + }, + { + "epoch": 0.6051994011161018, + "grad_norm": 2.1821470260620117, + "learning_rate": 7.90682840397983e-06, + "loss": 2.4845859527587892, + "step": 222320 + }, + { + "epoch": 0.6052538451068463, + "grad_norm": 2.555211305618286, + "learning_rate": 7.905738040070874e-06, + "loss": 2.486616325378418, + "step": 222340 + }, + { + "epoch": 0.6053082890975908, + "grad_norm": 2.223479747772217, + "learning_rate": 7.90464767616192e-06, + "loss": 2.4609424591064455, + "step": 222360 + }, + { + "epoch": 0.6053627330883353, + "grad_norm": 2.321829080581665, + "learning_rate": 7.903557312252966e-06, + "loss": 2.4575736999511717, + "step": 222380 + }, + { + "epoch": 0.6054171770790799, + "grad_norm": 2.1522884368896484, + "learning_rate": 7.902466948344012e-06, + "loss": 2.4646501541137695, + "step": 222400 + }, + { + "epoch": 0.6054716210698244, + "grad_norm": 2.1832828521728516, + "learning_rate": 7.901376584435056e-06, + "loss": 2.4640613555908204, + "step": 222420 + }, + { + "epoch": 0.6055260650605689, + "grad_norm": 2.7568001747131348, + "learning_rate": 7.900286220526101e-06, + "loss": 2.467926788330078, + "step": 222440 + }, + { + "epoch": 0.6055805090513134, + "grad_norm": 2.182234287261963, + "learning_rate": 7.899195856617147e-06, + "loss": 2.477615547180176, + "step": 222460 + }, + { + "epoch": 0.6056349530420579, + "grad_norm": 2.1452929973602295, + "learning_rate": 7.898105492708193e-06, + "loss": 2.472731590270996, + "step": 222480 + }, + { + "epoch": 0.6056893970328026, + "grad_norm": 2.1867175102233887, + "learning_rate": 7.897015128799237e-06, + "loss": 2.4706489562988283, + "step": 222500 + }, + { + "epoch": 0.6057438410235471, + "grad_norm": 2.187406063079834, + "learning_rate": 7.895924764890283e-06, + "loss": 2.4766029357910155, + "step": 222520 + }, + { + "epoch": 0.6057982850142916, + "grad_norm": 2.1522936820983887, + "learning_rate": 7.894834400981329e-06, + "loss": 2.4417655944824217, + "step": 222540 + }, + { + "epoch": 0.6058527290050361, + "grad_norm": 2.105215549468994, + "learning_rate": 7.893744037072375e-06, + "loss": 2.4764934539794923, + "step": 222560 + }, + { + "epoch": 0.6059071729957806, + "grad_norm": 2.112295150756836, + "learning_rate": 7.892653673163419e-06, + "loss": 2.4937549591064454, + "step": 222580 + }, + { + "epoch": 0.6059616169865251, + "grad_norm": 2.2256956100463867, + "learning_rate": 7.891563309254465e-06, + "loss": 2.5043718338012697, + "step": 222600 + }, + { + "epoch": 0.6060160609772697, + "grad_norm": 2.2267327308654785, + "learning_rate": 7.89047294534551e-06, + "loss": 2.489817810058594, + "step": 222620 + }, + { + "epoch": 0.6060705049680142, + "grad_norm": 2.1610429286956787, + "learning_rate": 7.889382581436556e-06, + "loss": 2.475812530517578, + "step": 222640 + }, + { + "epoch": 0.6061249489587587, + "grad_norm": 2.2100236415863037, + "learning_rate": 7.8882922175276e-06, + "loss": 2.486033630371094, + "step": 222660 + }, + { + "epoch": 0.6061793929495032, + "grad_norm": 2.203259229660034, + "learning_rate": 7.887201853618646e-06, + "loss": 2.4547906875610352, + "step": 222680 + }, + { + "epoch": 0.6062338369402477, + "grad_norm": 2.191587209701538, + "learning_rate": 7.886111489709692e-06, + "loss": 2.4939334869384764, + "step": 222700 + }, + { + "epoch": 0.6062882809309923, + "grad_norm": 2.0846917629241943, + "learning_rate": 7.885021125800738e-06, + "loss": 2.4920045852661135, + "step": 222720 + }, + { + "epoch": 0.6063427249217368, + "grad_norm": 2.1493866443634033, + "learning_rate": 7.883930761891782e-06, + "loss": 2.4509675979614256, + "step": 222740 + }, + { + "epoch": 0.6063971689124813, + "grad_norm": 2.12823224067688, + "learning_rate": 7.882840397982828e-06, + "loss": 2.4454700469970705, + "step": 222760 + }, + { + "epoch": 0.6064516129032258, + "grad_norm": 2.133129835128784, + "learning_rate": 7.881750034073873e-06, + "loss": 2.448303985595703, + "step": 222780 + }, + { + "epoch": 0.6065060568939703, + "grad_norm": 2.2168209552764893, + "learning_rate": 7.88065967016492e-06, + "loss": 2.4253387451171875, + "step": 222800 + }, + { + "epoch": 0.6065605008847148, + "grad_norm": 2.2325174808502197, + "learning_rate": 7.879569306255963e-06, + "loss": 2.5106231689453127, + "step": 222820 + }, + { + "epoch": 0.6066149448754594, + "grad_norm": 2.1258037090301514, + "learning_rate": 7.87847894234701e-06, + "loss": 2.4705669403076174, + "step": 222840 + }, + { + "epoch": 0.6066693888662039, + "grad_norm": 2.135828733444214, + "learning_rate": 7.877388578438055e-06, + "loss": 2.474246025085449, + "step": 222860 + }, + { + "epoch": 0.6067238328569484, + "grad_norm": 2.0845260620117188, + "learning_rate": 7.8762982145291e-06, + "loss": 2.4397556304931642, + "step": 222880 + }, + { + "epoch": 0.6067782768476929, + "grad_norm": 2.1952428817749023, + "learning_rate": 7.875207850620145e-06, + "loss": 2.475810432434082, + "step": 222900 + }, + { + "epoch": 0.6068327208384374, + "grad_norm": 2.192697763442993, + "learning_rate": 7.87411748671119e-06, + "loss": 2.445903205871582, + "step": 222920 + }, + { + "epoch": 0.6068871648291819, + "grad_norm": 2.690929651260376, + "learning_rate": 7.873027122802237e-06, + "loss": 2.4590314865112304, + "step": 222940 + }, + { + "epoch": 0.6069416088199265, + "grad_norm": 2.142432451248169, + "learning_rate": 7.871936758893282e-06, + "loss": 2.462020492553711, + "step": 222960 + }, + { + "epoch": 0.606996052810671, + "grad_norm": 2.3070271015167236, + "learning_rate": 7.870846394984326e-06, + "loss": 2.4747520446777345, + "step": 222980 + }, + { + "epoch": 0.6070504968014155, + "grad_norm": 2.1543447971343994, + "learning_rate": 7.869756031075372e-06, + "loss": 2.4941055297851564, + "step": 223000 + }, + { + "epoch": 0.60710494079216, + "grad_norm": 2.1601884365081787, + "learning_rate": 7.868665667166418e-06, + "loss": 2.446006202697754, + "step": 223020 + }, + { + "epoch": 0.6071593847829045, + "grad_norm": 2.2986602783203125, + "learning_rate": 7.867575303257464e-06, + "loss": 2.492819595336914, + "step": 223040 + }, + { + "epoch": 0.6072138287736492, + "grad_norm": 2.1611685752868652, + "learning_rate": 7.866484939348508e-06, + "loss": 2.4922508239746093, + "step": 223060 + }, + { + "epoch": 0.6072682727643937, + "grad_norm": 2.2341904640197754, + "learning_rate": 7.865394575439554e-06, + "loss": 2.4791688919067383, + "step": 223080 + }, + { + "epoch": 0.6073227167551382, + "grad_norm": 2.1248645782470703, + "learning_rate": 7.8643042115306e-06, + "loss": 2.4492326736450196, + "step": 223100 + }, + { + "epoch": 0.6073771607458827, + "grad_norm": 2.1149978637695312, + "learning_rate": 7.863213847621645e-06, + "loss": 2.4419506072998045, + "step": 223120 + }, + { + "epoch": 0.6074316047366272, + "grad_norm": 2.109445095062256, + "learning_rate": 7.86212348371269e-06, + "loss": 2.4555871963500975, + "step": 223140 + }, + { + "epoch": 0.6074860487273717, + "grad_norm": 2.11094069480896, + "learning_rate": 7.861033119803735e-06, + "loss": 2.482512283325195, + "step": 223160 + }, + { + "epoch": 0.6075404927181163, + "grad_norm": 2.1890904903411865, + "learning_rate": 7.859942755894781e-06, + "loss": 2.472622871398926, + "step": 223180 + }, + { + "epoch": 0.6075949367088608, + "grad_norm": 2.1848952770233154, + "learning_rate": 7.858852391985827e-06, + "loss": 2.442806434631348, + "step": 223200 + }, + { + "epoch": 0.6076493806996053, + "grad_norm": 2.0791919231414795, + "learning_rate": 7.857762028076871e-06, + "loss": 2.4925514221191407, + "step": 223220 + }, + { + "epoch": 0.6077038246903498, + "grad_norm": 2.181973695755005, + "learning_rate": 7.856671664167917e-06, + "loss": 2.4564586639404298, + "step": 223240 + }, + { + "epoch": 0.6077582686810943, + "grad_norm": 2.409839630126953, + "learning_rate": 7.855581300258963e-06, + "loss": 2.461988639831543, + "step": 223260 + }, + { + "epoch": 0.6078127126718389, + "grad_norm": 2.058379650115967, + "learning_rate": 7.854490936350008e-06, + "loss": 2.4494213104248046, + "step": 223280 + }, + { + "epoch": 0.6078671566625834, + "grad_norm": 2.1824419498443604, + "learning_rate": 7.853400572441053e-06, + "loss": 2.437472343444824, + "step": 223300 + }, + { + "epoch": 0.6079216006533279, + "grad_norm": 2.1568403244018555, + "learning_rate": 7.852310208532098e-06, + "loss": 2.4796409606933594, + "step": 223320 + }, + { + "epoch": 0.6079760446440724, + "grad_norm": 2.2145798206329346, + "learning_rate": 7.851219844623144e-06, + "loss": 2.491303634643555, + "step": 223340 + }, + { + "epoch": 0.6080304886348169, + "grad_norm": 2.3073878288269043, + "learning_rate": 7.85012948071419e-06, + "loss": 2.450844383239746, + "step": 223360 + }, + { + "epoch": 0.6080849326255614, + "grad_norm": 2.1000654697418213, + "learning_rate": 7.849039116805234e-06, + "loss": 2.478940010070801, + "step": 223380 + }, + { + "epoch": 0.608139376616306, + "grad_norm": 2.29252290725708, + "learning_rate": 7.84794875289628e-06, + "loss": 2.466110420227051, + "step": 223400 + }, + { + "epoch": 0.6081938206070505, + "grad_norm": 2.218189239501953, + "learning_rate": 7.846858388987326e-06, + "loss": 2.4970766067504884, + "step": 223420 + }, + { + "epoch": 0.608248264597795, + "grad_norm": 2.3449623584747314, + "learning_rate": 7.845768025078372e-06, + "loss": 2.476134490966797, + "step": 223440 + }, + { + "epoch": 0.6083027085885395, + "grad_norm": 2.124859094619751, + "learning_rate": 7.844677661169416e-06, + "loss": 2.4523433685302733, + "step": 223460 + }, + { + "epoch": 0.608357152579284, + "grad_norm": 2.4867732524871826, + "learning_rate": 7.843587297260462e-06, + "loss": 2.472865676879883, + "step": 223480 + }, + { + "epoch": 0.6084115965700286, + "grad_norm": 2.1912214756011963, + "learning_rate": 7.842496933351507e-06, + "loss": 2.48321533203125, + "step": 223500 + }, + { + "epoch": 0.6084660405607731, + "grad_norm": 2.1848227977752686, + "learning_rate": 7.841406569442553e-06, + "loss": 2.4644168853759765, + "step": 223520 + }, + { + "epoch": 0.6085204845515176, + "grad_norm": 2.2369070053100586, + "learning_rate": 7.840316205533597e-06, + "loss": 2.459812545776367, + "step": 223540 + }, + { + "epoch": 0.6085749285422621, + "grad_norm": 2.3872923851013184, + "learning_rate": 7.839225841624643e-06, + "loss": 2.493771553039551, + "step": 223560 + }, + { + "epoch": 0.6086293725330066, + "grad_norm": 2.2137179374694824, + "learning_rate": 7.838135477715689e-06, + "loss": 2.487427520751953, + "step": 223580 + }, + { + "epoch": 0.6086838165237511, + "grad_norm": 2.1798408031463623, + "learning_rate": 7.837045113806735e-06, + "loss": 2.4939847946166993, + "step": 223600 + }, + { + "epoch": 0.6087382605144958, + "grad_norm": 2.5910842418670654, + "learning_rate": 7.835954749897779e-06, + "loss": 2.4529258728027346, + "step": 223620 + }, + { + "epoch": 0.6087927045052403, + "grad_norm": 2.061023712158203, + "learning_rate": 7.834864385988825e-06, + "loss": 2.4673076629638673, + "step": 223640 + }, + { + "epoch": 0.6088471484959848, + "grad_norm": 2.1468756198883057, + "learning_rate": 7.83377402207987e-06, + "loss": 2.4831661224365233, + "step": 223660 + }, + { + "epoch": 0.6089015924867293, + "grad_norm": 2.1341567039489746, + "learning_rate": 7.832683658170915e-06, + "loss": 2.4792963027954102, + "step": 223680 + }, + { + "epoch": 0.6089560364774738, + "grad_norm": 2.2390873432159424, + "learning_rate": 7.83159329426196e-06, + "loss": 2.4879276275634767, + "step": 223700 + }, + { + "epoch": 0.6090104804682184, + "grad_norm": 2.1772186756134033, + "learning_rate": 7.830502930353006e-06, + "loss": 2.475955009460449, + "step": 223720 + }, + { + "epoch": 0.6090649244589629, + "grad_norm": 2.2291817665100098, + "learning_rate": 7.829412566444052e-06, + "loss": 2.5075281143188475, + "step": 223740 + }, + { + "epoch": 0.6091193684497074, + "grad_norm": 2.2937140464782715, + "learning_rate": 7.828322202535096e-06, + "loss": 2.4367748260498048, + "step": 223760 + }, + { + "epoch": 0.6091738124404519, + "grad_norm": 2.353788137435913, + "learning_rate": 7.827231838626142e-06, + "loss": 2.4775957107543944, + "step": 223780 + }, + { + "epoch": 0.6092282564311964, + "grad_norm": 2.2023823261260986, + "learning_rate": 7.826141474717188e-06, + "loss": 2.4576580047607424, + "step": 223800 + }, + { + "epoch": 0.6092827004219409, + "grad_norm": 2.2959954738616943, + "learning_rate": 7.825051110808233e-06, + "loss": 2.477440071105957, + "step": 223820 + }, + { + "epoch": 0.6093371444126855, + "grad_norm": 2.1214377880096436, + "learning_rate": 7.823960746899278e-06, + "loss": 2.4734729766845702, + "step": 223840 + }, + { + "epoch": 0.60939158840343, + "grad_norm": 2.16626238822937, + "learning_rate": 7.822870382990323e-06, + "loss": 2.4747238159179688, + "step": 223860 + }, + { + "epoch": 0.6094460323941745, + "grad_norm": 2.1152331829071045, + "learning_rate": 7.82178001908137e-06, + "loss": 2.493580627441406, + "step": 223880 + }, + { + "epoch": 0.609500476384919, + "grad_norm": 2.2506909370422363, + "learning_rate": 7.820689655172415e-06, + "loss": 2.4914894104003906, + "step": 223900 + }, + { + "epoch": 0.6095549203756635, + "grad_norm": 2.207934617996216, + "learning_rate": 7.819599291263459e-06, + "loss": 2.471666145324707, + "step": 223920 + }, + { + "epoch": 0.6096093643664081, + "grad_norm": 2.1321024894714355, + "learning_rate": 7.818508927354505e-06, + "loss": 2.4607927322387697, + "step": 223940 + }, + { + "epoch": 0.6096638083571526, + "grad_norm": 4.872452259063721, + "learning_rate": 7.81741856344555e-06, + "loss": 2.484231948852539, + "step": 223960 + }, + { + "epoch": 0.6097182523478971, + "grad_norm": 2.203403949737549, + "learning_rate": 7.816328199536597e-06, + "loss": 2.4521564483642577, + "step": 223980 + }, + { + "epoch": 0.6097726963386416, + "grad_norm": 2.139486074447632, + "learning_rate": 7.81523783562764e-06, + "loss": 2.465823173522949, + "step": 224000 + }, + { + "epoch": 0.6098271403293861, + "grad_norm": 2.2597854137420654, + "learning_rate": 7.814147471718686e-06, + "loss": 2.4711097717285155, + "step": 224020 + }, + { + "epoch": 0.6098815843201306, + "grad_norm": 2.2247440814971924, + "learning_rate": 7.813057107809732e-06, + "loss": 2.4576183319091798, + "step": 224040 + }, + { + "epoch": 0.6099360283108752, + "grad_norm": 2.2618911266326904, + "learning_rate": 7.811966743900778e-06, + "loss": 2.4589256286621093, + "step": 224060 + }, + { + "epoch": 0.6099904723016197, + "grad_norm": 2.17769718170166, + "learning_rate": 7.810876379991822e-06, + "loss": 2.486241912841797, + "step": 224080 + }, + { + "epoch": 0.6100449162923642, + "grad_norm": 2.2069551944732666, + "learning_rate": 7.809786016082868e-06, + "loss": 2.491847801208496, + "step": 224100 + }, + { + "epoch": 0.6100993602831087, + "grad_norm": 2.5334126949310303, + "learning_rate": 7.808695652173914e-06, + "loss": 2.4735279083251953, + "step": 224120 + }, + { + "epoch": 0.6101538042738532, + "grad_norm": 2.183129072189331, + "learning_rate": 7.80760528826496e-06, + "loss": 2.491618537902832, + "step": 224140 + }, + { + "epoch": 0.6102082482645977, + "grad_norm": 2.3631680011749268, + "learning_rate": 7.806514924356004e-06, + "loss": 2.4642559051513673, + "step": 224160 + }, + { + "epoch": 0.6102626922553424, + "grad_norm": 2.895702362060547, + "learning_rate": 7.80542456044705e-06, + "loss": 2.4680059432983397, + "step": 224180 + }, + { + "epoch": 0.6103171362460869, + "grad_norm": 2.2793796062469482, + "learning_rate": 7.804334196538095e-06, + "loss": 2.5042856216430662, + "step": 224200 + }, + { + "epoch": 0.6103715802368314, + "grad_norm": 2.265199899673462, + "learning_rate": 7.803243832629141e-06, + "loss": 2.4852325439453127, + "step": 224220 + }, + { + "epoch": 0.6104260242275759, + "grad_norm": 2.144879102706909, + "learning_rate": 7.802153468720185e-06, + "loss": 2.4665767669677736, + "step": 224240 + }, + { + "epoch": 0.6104804682183204, + "grad_norm": 2.143683910369873, + "learning_rate": 7.801063104811231e-06, + "loss": 2.471227836608887, + "step": 224260 + }, + { + "epoch": 0.610534912209065, + "grad_norm": 2.272841215133667, + "learning_rate": 7.799972740902277e-06, + "loss": 2.463485336303711, + "step": 224280 + }, + { + "epoch": 0.6105893561998095, + "grad_norm": 4.477794647216797, + "learning_rate": 7.798882376993323e-06, + "loss": 2.484019088745117, + "step": 224300 + }, + { + "epoch": 0.610643800190554, + "grad_norm": 2.0675578117370605, + "learning_rate": 7.797792013084367e-06, + "loss": 2.4687936782836912, + "step": 224320 + }, + { + "epoch": 0.6106982441812985, + "grad_norm": 2.155505895614624, + "learning_rate": 7.796701649175413e-06, + "loss": 2.471436882019043, + "step": 224340 + }, + { + "epoch": 0.610752688172043, + "grad_norm": 1.996190071105957, + "learning_rate": 7.795611285266458e-06, + "loss": 2.474060821533203, + "step": 224360 + }, + { + "epoch": 0.6108071321627875, + "grad_norm": 2.0875627994537354, + "learning_rate": 7.794520921357504e-06, + "loss": 2.47119255065918, + "step": 224380 + }, + { + "epoch": 0.6108615761535321, + "grad_norm": 2.2217822074890137, + "learning_rate": 7.793430557448548e-06, + "loss": 2.4478809356689455, + "step": 224400 + }, + { + "epoch": 0.6109160201442766, + "grad_norm": 2.248220920562744, + "learning_rate": 7.792340193539594e-06, + "loss": 2.447036552429199, + "step": 224420 + }, + { + "epoch": 0.6109704641350211, + "grad_norm": 2.1820099353790283, + "learning_rate": 7.79124982963064e-06, + "loss": 2.4733367919921876, + "step": 224440 + }, + { + "epoch": 0.6110249081257656, + "grad_norm": 2.1574881076812744, + "learning_rate": 7.790159465721686e-06, + "loss": 2.4590600967407226, + "step": 224460 + }, + { + "epoch": 0.6110793521165101, + "grad_norm": 2.131915330886841, + "learning_rate": 7.78906910181273e-06, + "loss": 2.4782896041870117, + "step": 224480 + }, + { + "epoch": 0.6111337961072547, + "grad_norm": 2.1401233673095703, + "learning_rate": 7.787978737903776e-06, + "loss": 2.465581512451172, + "step": 224500 + }, + { + "epoch": 0.6111882400979992, + "grad_norm": 2.1351425647735596, + "learning_rate": 7.786888373994822e-06, + "loss": 2.4922637939453125, + "step": 224520 + }, + { + "epoch": 0.6112426840887437, + "grad_norm": 2.150796413421631, + "learning_rate": 7.785798010085867e-06, + "loss": 2.4440250396728516, + "step": 224540 + }, + { + "epoch": 0.6112971280794882, + "grad_norm": 2.4606645107269287, + "learning_rate": 7.784707646176911e-06, + "loss": 2.4617156982421875, + "step": 224560 + }, + { + "epoch": 0.6113515720702327, + "grad_norm": 2.2339046001434326, + "learning_rate": 7.783617282267957e-06, + "loss": 2.478105926513672, + "step": 224580 + }, + { + "epoch": 0.6114060160609772, + "grad_norm": 2.1363203525543213, + "learning_rate": 7.782526918359003e-06, + "loss": 2.4818639755249023, + "step": 224600 + }, + { + "epoch": 0.6114604600517218, + "grad_norm": 2.1917386054992676, + "learning_rate": 7.781436554450049e-06, + "loss": 2.5005617141723633, + "step": 224620 + }, + { + "epoch": 0.6115149040424663, + "grad_norm": 2.133479118347168, + "learning_rate": 7.780346190541093e-06, + "loss": 2.4947792053222657, + "step": 224640 + }, + { + "epoch": 0.6115693480332108, + "grad_norm": 2.1446423530578613, + "learning_rate": 7.779255826632139e-06, + "loss": 2.4736087799072264, + "step": 224660 + }, + { + "epoch": 0.6116237920239553, + "grad_norm": 2.1312358379364014, + "learning_rate": 7.778165462723185e-06, + "loss": 2.4857322692871096, + "step": 224680 + }, + { + "epoch": 0.6116782360146998, + "grad_norm": 2.208040475845337, + "learning_rate": 7.77707509881423e-06, + "loss": 2.4648937225341796, + "step": 224700 + }, + { + "epoch": 0.6117326800054445, + "grad_norm": 2.1576972007751465, + "learning_rate": 7.775984734905275e-06, + "loss": 2.462260055541992, + "step": 224720 + }, + { + "epoch": 0.611787123996189, + "grad_norm": 2.134035587310791, + "learning_rate": 7.77489437099632e-06, + "loss": 2.490130805969238, + "step": 224740 + }, + { + "epoch": 0.6118415679869335, + "grad_norm": 2.1553659439086914, + "learning_rate": 7.773804007087366e-06, + "loss": 2.4909910202026366, + "step": 224760 + }, + { + "epoch": 0.611896011977678, + "grad_norm": 2.2187399864196777, + "learning_rate": 7.77271364317841e-06, + "loss": 2.480984687805176, + "step": 224780 + }, + { + "epoch": 0.6119504559684225, + "grad_norm": 2.4025089740753174, + "learning_rate": 7.771623279269456e-06, + "loss": 2.476413536071777, + "step": 224800 + }, + { + "epoch": 0.612004899959167, + "grad_norm": 2.2154948711395264, + "learning_rate": 7.770532915360502e-06, + "loss": 2.480061340332031, + "step": 224820 + }, + { + "epoch": 0.6120593439499116, + "grad_norm": 2.1555893421173096, + "learning_rate": 7.769442551451548e-06, + "loss": 2.5112300872802735, + "step": 224840 + }, + { + "epoch": 0.6121137879406561, + "grad_norm": 2.1473610401153564, + "learning_rate": 7.768352187542592e-06, + "loss": 2.4718000411987306, + "step": 224860 + }, + { + "epoch": 0.6121682319314006, + "grad_norm": 3.803757905960083, + "learning_rate": 7.767261823633638e-06, + "loss": 2.4550664901733397, + "step": 224880 + }, + { + "epoch": 0.6122226759221451, + "grad_norm": 2.110734462738037, + "learning_rate": 7.766171459724683e-06, + "loss": 2.4472570419311523, + "step": 224900 + }, + { + "epoch": 0.6122771199128896, + "grad_norm": 2.246932029724121, + "learning_rate": 7.76508109581573e-06, + "loss": 2.4948745727539063, + "step": 224920 + }, + { + "epoch": 0.6123315639036342, + "grad_norm": 2.231738805770874, + "learning_rate": 7.763990731906773e-06, + "loss": 2.48012752532959, + "step": 224940 + }, + { + "epoch": 0.6123860078943787, + "grad_norm": 2.1181821823120117, + "learning_rate": 7.76290036799782e-06, + "loss": 2.462144660949707, + "step": 224960 + }, + { + "epoch": 0.6124404518851232, + "grad_norm": 2.1471354961395264, + "learning_rate": 7.761810004088865e-06, + "loss": 2.4463390350341796, + "step": 224980 + }, + { + "epoch": 0.6124948958758677, + "grad_norm": 2.5153398513793945, + "learning_rate": 7.76071964017991e-06, + "loss": 2.5020639419555666, + "step": 225000 + }, + { + "epoch": 0.6125493398666122, + "grad_norm": 2.097787857055664, + "learning_rate": 7.759629276270955e-06, + "loss": 2.4718563079833986, + "step": 225020 + }, + { + "epoch": 0.6126037838573567, + "grad_norm": 2.0792603492736816, + "learning_rate": 7.758538912362e-06, + "loss": 2.498855209350586, + "step": 225040 + }, + { + "epoch": 0.6126582278481013, + "grad_norm": 2.1890029907226562, + "learning_rate": 7.757448548453047e-06, + "loss": 2.47009162902832, + "step": 225060 + }, + { + "epoch": 0.6127126718388458, + "grad_norm": 2.135355234146118, + "learning_rate": 7.756358184544092e-06, + "loss": 2.4890308380126953, + "step": 225080 + }, + { + "epoch": 0.6127671158295903, + "grad_norm": 2.2523844242095947, + "learning_rate": 7.755267820635136e-06, + "loss": 2.4494297027587892, + "step": 225100 + }, + { + "epoch": 0.6128215598203348, + "grad_norm": 2.7200520038604736, + "learning_rate": 7.754177456726182e-06, + "loss": 2.483180618286133, + "step": 225120 + }, + { + "epoch": 0.6128760038110793, + "grad_norm": 2.1209511756896973, + "learning_rate": 7.753087092817228e-06, + "loss": 2.453068161010742, + "step": 225140 + }, + { + "epoch": 0.6129304478018238, + "grad_norm": 2.1464896202087402, + "learning_rate": 7.751996728908274e-06, + "loss": 2.4312082290649415, + "step": 225160 + }, + { + "epoch": 0.6129848917925684, + "grad_norm": 2.077603340148926, + "learning_rate": 7.750906364999318e-06, + "loss": 2.471705436706543, + "step": 225180 + }, + { + "epoch": 0.6130393357833129, + "grad_norm": 2.2446303367614746, + "learning_rate": 7.749816001090364e-06, + "loss": 2.4452770233154295, + "step": 225200 + }, + { + "epoch": 0.6130937797740574, + "grad_norm": 2.1925363540649414, + "learning_rate": 7.74872563718141e-06, + "loss": 2.483261489868164, + "step": 225220 + }, + { + "epoch": 0.6131482237648019, + "grad_norm": 2.1676242351531982, + "learning_rate": 7.747635273272455e-06, + "loss": 2.4665603637695312, + "step": 225240 + }, + { + "epoch": 0.6132026677555464, + "grad_norm": 2.229776382446289, + "learning_rate": 7.7465449093635e-06, + "loss": 2.4522953033447266, + "step": 225260 + }, + { + "epoch": 0.613257111746291, + "grad_norm": 2.3892054557800293, + "learning_rate": 7.745454545454545e-06, + "loss": 2.479221725463867, + "step": 225280 + }, + { + "epoch": 0.6133115557370356, + "grad_norm": 2.2367100715637207, + "learning_rate": 7.744364181545591e-06, + "loss": 2.463210678100586, + "step": 225300 + }, + { + "epoch": 0.61336599972778, + "grad_norm": 2.1308341026306152, + "learning_rate": 7.743273817636637e-06, + "loss": 2.463906669616699, + "step": 225320 + }, + { + "epoch": 0.6134204437185246, + "grad_norm": 2.1376616954803467, + "learning_rate": 7.742183453727681e-06, + "loss": 2.488918113708496, + "step": 225340 + }, + { + "epoch": 0.6134748877092691, + "grad_norm": 2.1886885166168213, + "learning_rate": 7.741093089818727e-06, + "loss": 2.4595783233642576, + "step": 225360 + }, + { + "epoch": 0.6135293317000136, + "grad_norm": 2.394298553466797, + "learning_rate": 7.740002725909773e-06, + "loss": 2.488683891296387, + "step": 225380 + }, + { + "epoch": 0.6135837756907582, + "grad_norm": 2.1609551906585693, + "learning_rate": 7.738912362000819e-06, + "loss": 2.492963409423828, + "step": 225400 + }, + { + "epoch": 0.6136382196815027, + "grad_norm": 2.2141754627227783, + "learning_rate": 7.737821998091863e-06, + "loss": 2.481395721435547, + "step": 225420 + }, + { + "epoch": 0.6136926636722472, + "grad_norm": 2.1391475200653076, + "learning_rate": 7.736731634182908e-06, + "loss": 2.4823688507080077, + "step": 225440 + }, + { + "epoch": 0.6137471076629917, + "grad_norm": 2.1987645626068115, + "learning_rate": 7.735641270273954e-06, + "loss": 2.4642967224121093, + "step": 225460 + }, + { + "epoch": 0.6138015516537362, + "grad_norm": 2.1326684951782227, + "learning_rate": 7.734550906365e-06, + "loss": 2.4376264572143556, + "step": 225480 + }, + { + "epoch": 0.6138559956444808, + "grad_norm": 2.222597360610962, + "learning_rate": 7.733460542456044e-06, + "loss": 2.481040191650391, + "step": 225500 + }, + { + "epoch": 0.6139104396352253, + "grad_norm": 2.274613857269287, + "learning_rate": 7.73237017854709e-06, + "loss": 2.47729606628418, + "step": 225520 + }, + { + "epoch": 0.6139648836259698, + "grad_norm": 2.114983558654785, + "learning_rate": 7.731279814638136e-06, + "loss": 2.477946662902832, + "step": 225540 + }, + { + "epoch": 0.6140193276167143, + "grad_norm": 2.158256769180298, + "learning_rate": 7.730189450729182e-06, + "loss": 2.481380271911621, + "step": 225560 + }, + { + "epoch": 0.6140737716074588, + "grad_norm": 2.1471712589263916, + "learning_rate": 7.729099086820226e-06, + "loss": 2.4336734771728517, + "step": 225580 + }, + { + "epoch": 0.6141282155982033, + "grad_norm": 2.306638717651367, + "learning_rate": 7.728008722911272e-06, + "loss": 2.472268295288086, + "step": 225600 + }, + { + "epoch": 0.6141826595889479, + "grad_norm": 2.2025842666625977, + "learning_rate": 7.726918359002317e-06, + "loss": 2.4624996185302734, + "step": 225620 + }, + { + "epoch": 0.6142371035796924, + "grad_norm": 2.209172010421753, + "learning_rate": 7.725827995093363e-06, + "loss": 2.4675941467285156, + "step": 225640 + }, + { + "epoch": 0.6142915475704369, + "grad_norm": 2.149550199508667, + "learning_rate": 7.724737631184407e-06, + "loss": 2.4477481842041016, + "step": 225660 + }, + { + "epoch": 0.6143459915611814, + "grad_norm": 2.0840978622436523, + "learning_rate": 7.723647267275453e-06, + "loss": 2.481346321105957, + "step": 225680 + }, + { + "epoch": 0.6144004355519259, + "grad_norm": 2.1259443759918213, + "learning_rate": 7.722556903366499e-06, + "loss": 2.499241065979004, + "step": 225700 + }, + { + "epoch": 0.6144548795426705, + "grad_norm": 2.2194416522979736, + "learning_rate": 7.721466539457545e-06, + "loss": 2.479704666137695, + "step": 225720 + }, + { + "epoch": 0.614509323533415, + "grad_norm": 2.2852344512939453, + "learning_rate": 7.720376175548589e-06, + "loss": 2.478633689880371, + "step": 225740 + }, + { + "epoch": 0.6145637675241595, + "grad_norm": 2.3318495750427246, + "learning_rate": 7.719285811639635e-06, + "loss": 2.5017091751098635, + "step": 225760 + }, + { + "epoch": 0.614618211514904, + "grad_norm": 2.2403783798217773, + "learning_rate": 7.71819544773068e-06, + "loss": 2.4697498321533202, + "step": 225780 + }, + { + "epoch": 0.6146726555056485, + "grad_norm": 2.1120946407318115, + "learning_rate": 7.717105083821726e-06, + "loss": 2.496552276611328, + "step": 225800 + }, + { + "epoch": 0.614727099496393, + "grad_norm": 2.086580276489258, + "learning_rate": 7.71601471991277e-06, + "loss": 2.4564069747924804, + "step": 225820 + }, + { + "epoch": 0.6147815434871376, + "grad_norm": 2.3934569358825684, + "learning_rate": 7.714924356003816e-06, + "loss": 2.4796634674072267, + "step": 225840 + }, + { + "epoch": 0.6148359874778822, + "grad_norm": 2.103473424911499, + "learning_rate": 7.713833992094862e-06, + "loss": 2.436646842956543, + "step": 225860 + }, + { + "epoch": 0.6148904314686267, + "grad_norm": 2.242119312286377, + "learning_rate": 7.712743628185908e-06, + "loss": 2.4713651657104494, + "step": 225880 + }, + { + "epoch": 0.6149448754593712, + "grad_norm": 2.194777727127075, + "learning_rate": 7.711653264276952e-06, + "loss": 2.4909551620483397, + "step": 225900 + }, + { + "epoch": 0.6149993194501157, + "grad_norm": 2.12687611579895, + "learning_rate": 7.710562900367998e-06, + "loss": 2.5039962768554687, + "step": 225920 + }, + { + "epoch": 0.6150537634408603, + "grad_norm": 2.1846730709075928, + "learning_rate": 7.709472536459044e-06, + "loss": 2.4737260818481444, + "step": 225940 + }, + { + "epoch": 0.6151082074316048, + "grad_norm": 2.0419044494628906, + "learning_rate": 7.70838217255009e-06, + "loss": 2.4489635467529296, + "step": 225960 + }, + { + "epoch": 0.6151626514223493, + "grad_norm": 2.174814462661743, + "learning_rate": 7.707291808641133e-06, + "loss": 2.453561210632324, + "step": 225980 + }, + { + "epoch": 0.6152170954130938, + "grad_norm": 2.1629621982574463, + "learning_rate": 7.70620144473218e-06, + "loss": 2.47528190612793, + "step": 226000 + }, + { + "epoch": 0.6152715394038383, + "grad_norm": 2.515092611312866, + "learning_rate": 7.705111080823225e-06, + "loss": 2.4896812438964844, + "step": 226020 + }, + { + "epoch": 0.6153259833945828, + "grad_norm": 2.253448724746704, + "learning_rate": 7.704020716914271e-06, + "loss": 2.4855491638183596, + "step": 226040 + }, + { + "epoch": 0.6153804273853274, + "grad_norm": 3.060427188873291, + "learning_rate": 7.702930353005315e-06, + "loss": 2.48803825378418, + "step": 226060 + }, + { + "epoch": 0.6154348713760719, + "grad_norm": 2.2737877368927, + "learning_rate": 7.70183998909636e-06, + "loss": 2.4913015365600586, + "step": 226080 + }, + { + "epoch": 0.6154893153668164, + "grad_norm": 2.1889164447784424, + "learning_rate": 7.700749625187407e-06, + "loss": 2.483389663696289, + "step": 226100 + }, + { + "epoch": 0.6155437593575609, + "grad_norm": 2.415982723236084, + "learning_rate": 7.699659261278452e-06, + "loss": 2.443062973022461, + "step": 226120 + }, + { + "epoch": 0.6155982033483054, + "grad_norm": 2.177039384841919, + "learning_rate": 7.698568897369497e-06, + "loss": 2.489582633972168, + "step": 226140 + }, + { + "epoch": 0.61565264733905, + "grad_norm": 2.1775546073913574, + "learning_rate": 7.697478533460542e-06, + "loss": 2.4835988998413088, + "step": 226160 + }, + { + "epoch": 0.6157070913297945, + "grad_norm": 2.2626254558563232, + "learning_rate": 7.696388169551588e-06, + "loss": 2.487538528442383, + "step": 226180 + }, + { + "epoch": 0.615761535320539, + "grad_norm": 2.1778628826141357, + "learning_rate": 7.695297805642634e-06, + "loss": 2.480410575866699, + "step": 226200 + }, + { + "epoch": 0.6158159793112835, + "grad_norm": 2.085510015487671, + "learning_rate": 7.694207441733678e-06, + "loss": 2.4712167739868165, + "step": 226220 + }, + { + "epoch": 0.615870423302028, + "grad_norm": 2.148949146270752, + "learning_rate": 7.693117077824724e-06, + "loss": 2.481361198425293, + "step": 226240 + }, + { + "epoch": 0.6159248672927725, + "grad_norm": 2.152941942214966, + "learning_rate": 7.69202671391577e-06, + "loss": 2.4744522094726564, + "step": 226260 + }, + { + "epoch": 0.6159793112835171, + "grad_norm": 2.1196513175964355, + "learning_rate": 7.690936350006816e-06, + "loss": 2.4878658294677733, + "step": 226280 + }, + { + "epoch": 0.6160337552742616, + "grad_norm": 2.359743118286133, + "learning_rate": 7.68984598609786e-06, + "loss": 2.4817970275878904, + "step": 226300 + }, + { + "epoch": 0.6160881992650061, + "grad_norm": 2.226276159286499, + "learning_rate": 7.688755622188905e-06, + "loss": 2.4658172607421873, + "step": 226320 + }, + { + "epoch": 0.6161426432557506, + "grad_norm": 2.1133737564086914, + "learning_rate": 7.687665258279951e-06, + "loss": 2.473730278015137, + "step": 226340 + }, + { + "epoch": 0.6161970872464951, + "grad_norm": 2.444335460662842, + "learning_rate": 7.686574894370997e-06, + "loss": 2.4825260162353517, + "step": 226360 + }, + { + "epoch": 0.6162515312372396, + "grad_norm": 2.1734769344329834, + "learning_rate": 7.685484530462041e-06, + "loss": 2.4827741622924804, + "step": 226380 + }, + { + "epoch": 0.6163059752279842, + "grad_norm": 2.110685348510742, + "learning_rate": 7.684394166553087e-06, + "loss": 2.464876174926758, + "step": 226400 + }, + { + "epoch": 0.6163604192187288, + "grad_norm": 2.1315665245056152, + "learning_rate": 7.683303802644133e-06, + "loss": 2.433616828918457, + "step": 226420 + }, + { + "epoch": 0.6164148632094733, + "grad_norm": 2.0742733478546143, + "learning_rate": 7.682213438735179e-06, + "loss": 2.4568910598754883, + "step": 226440 + }, + { + "epoch": 0.6164693072002178, + "grad_norm": 2.0526766777038574, + "learning_rate": 7.681123074826224e-06, + "loss": 2.5064889907836916, + "step": 226460 + }, + { + "epoch": 0.6165237511909623, + "grad_norm": 2.246596336364746, + "learning_rate": 7.680032710917269e-06, + "loss": 2.4690053939819334, + "step": 226480 + }, + { + "epoch": 0.6165781951817069, + "grad_norm": 2.2183356285095215, + "learning_rate": 7.678942347008314e-06, + "loss": 2.4637107849121094, + "step": 226500 + }, + { + "epoch": 0.6166326391724514, + "grad_norm": 2.1598732471466064, + "learning_rate": 7.67785198309936e-06, + "loss": 2.4827144622802733, + "step": 226520 + }, + { + "epoch": 0.6166870831631959, + "grad_norm": 2.184962749481201, + "learning_rate": 7.676761619190406e-06, + "loss": 2.4808090209960936, + "step": 226540 + }, + { + "epoch": 0.6167415271539404, + "grad_norm": 2.218841791152954, + "learning_rate": 7.67567125528145e-06, + "loss": 2.499820518493652, + "step": 226560 + }, + { + "epoch": 0.6167959711446849, + "grad_norm": 2.2285170555114746, + "learning_rate": 7.674580891372496e-06, + "loss": 2.481549835205078, + "step": 226580 + }, + { + "epoch": 0.6168504151354294, + "grad_norm": 2.231088161468506, + "learning_rate": 7.673490527463542e-06, + "loss": 2.501888465881348, + "step": 226600 + }, + { + "epoch": 0.616904859126174, + "grad_norm": 2.6265344619750977, + "learning_rate": 7.672400163554587e-06, + "loss": 2.4473697662353517, + "step": 226620 + }, + { + "epoch": 0.6169593031169185, + "grad_norm": 2.218045949935913, + "learning_rate": 7.671309799645632e-06, + "loss": 2.4565364837646486, + "step": 226640 + }, + { + "epoch": 0.617013747107663, + "grad_norm": 2.310533046722412, + "learning_rate": 7.670219435736677e-06, + "loss": 2.462628936767578, + "step": 226660 + }, + { + "epoch": 0.6170681910984075, + "grad_norm": 2.2226054668426514, + "learning_rate": 7.669129071827723e-06, + "loss": 2.4646902084350586, + "step": 226680 + }, + { + "epoch": 0.617122635089152, + "grad_norm": 2.307807683944702, + "learning_rate": 7.668038707918769e-06, + "loss": 2.4769115447998047, + "step": 226700 + }, + { + "epoch": 0.6171770790798966, + "grad_norm": 2.1742773056030273, + "learning_rate": 7.666948344009813e-06, + "loss": 2.4711078643798827, + "step": 226720 + }, + { + "epoch": 0.6172315230706411, + "grad_norm": 2.4690968990325928, + "learning_rate": 7.665857980100859e-06, + "loss": 2.49448299407959, + "step": 226740 + }, + { + "epoch": 0.6172859670613856, + "grad_norm": 2.514390230178833, + "learning_rate": 7.664767616191905e-06, + "loss": 2.447383499145508, + "step": 226760 + }, + { + "epoch": 0.6173404110521301, + "grad_norm": 2.1020472049713135, + "learning_rate": 7.66367725228295e-06, + "loss": 2.490876388549805, + "step": 226780 + }, + { + "epoch": 0.6173948550428746, + "grad_norm": 2.2035739421844482, + "learning_rate": 7.662586888373995e-06, + "loss": 2.488460350036621, + "step": 226800 + }, + { + "epoch": 0.6174492990336191, + "grad_norm": 2.2458062171936035, + "learning_rate": 7.66149652446504e-06, + "loss": 2.4586448669433594, + "step": 226820 + }, + { + "epoch": 0.6175037430243637, + "grad_norm": 2.17423939704895, + "learning_rate": 7.660406160556086e-06, + "loss": 2.4546142578125, + "step": 226840 + }, + { + "epoch": 0.6175581870151082, + "grad_norm": 2.1609787940979004, + "learning_rate": 7.659315796647132e-06, + "loss": 2.4626205444335936, + "step": 226860 + }, + { + "epoch": 0.6176126310058527, + "grad_norm": 2.13081431388855, + "learning_rate": 7.658225432738176e-06, + "loss": 2.51144962310791, + "step": 226880 + }, + { + "epoch": 0.6176670749965972, + "grad_norm": 2.1950855255126953, + "learning_rate": 7.657135068829222e-06, + "loss": 2.4624235153198244, + "step": 226900 + }, + { + "epoch": 0.6177215189873417, + "grad_norm": 2.097736120223999, + "learning_rate": 7.656044704920268e-06, + "loss": 2.4799901962280275, + "step": 226920 + }, + { + "epoch": 0.6177759629780863, + "grad_norm": 2.146028518676758, + "learning_rate": 7.654954341011314e-06, + "loss": 2.448647117614746, + "step": 226940 + }, + { + "epoch": 0.6178304069688308, + "grad_norm": 2.825493574142456, + "learning_rate": 7.653863977102358e-06, + "loss": 2.4576086044311523, + "step": 226960 + }, + { + "epoch": 0.6178848509595753, + "grad_norm": 2.1866579055786133, + "learning_rate": 7.652773613193404e-06, + "loss": 2.460439109802246, + "step": 226980 + }, + { + "epoch": 0.6179392949503199, + "grad_norm": 2.126828193664551, + "learning_rate": 7.65168324928445e-06, + "loss": 2.4962432861328123, + "step": 227000 + }, + { + "epoch": 0.6179937389410644, + "grad_norm": 2.215714931488037, + "learning_rate": 7.650592885375495e-06, + "loss": 2.4816261291503907, + "step": 227020 + }, + { + "epoch": 0.6180481829318089, + "grad_norm": 2.456787586212158, + "learning_rate": 7.64950252146654e-06, + "loss": 2.4672508239746094, + "step": 227040 + }, + { + "epoch": 0.6181026269225535, + "grad_norm": 2.2140345573425293, + "learning_rate": 7.648412157557585e-06, + "loss": 2.4741209030151365, + "step": 227060 + }, + { + "epoch": 0.618157070913298, + "grad_norm": 2.2456371784210205, + "learning_rate": 7.647321793648631e-06, + "loss": 2.488323974609375, + "step": 227080 + }, + { + "epoch": 0.6182115149040425, + "grad_norm": 2.1800343990325928, + "learning_rate": 7.646231429739677e-06, + "loss": 2.443629837036133, + "step": 227100 + }, + { + "epoch": 0.618265958894787, + "grad_norm": 2.102992296218872, + "learning_rate": 7.645141065830721e-06, + "loss": 2.465907669067383, + "step": 227120 + }, + { + "epoch": 0.6183204028855315, + "grad_norm": 2.304960250854492, + "learning_rate": 7.644050701921767e-06, + "loss": 2.4658336639404297, + "step": 227140 + }, + { + "epoch": 0.6183748468762761, + "grad_norm": 2.1091556549072266, + "learning_rate": 7.642960338012812e-06, + "loss": 2.476775550842285, + "step": 227160 + }, + { + "epoch": 0.6184292908670206, + "grad_norm": 2.133321523666382, + "learning_rate": 7.641869974103858e-06, + "loss": 2.4857309341430662, + "step": 227180 + }, + { + "epoch": 0.6184837348577651, + "grad_norm": 2.2143850326538086, + "learning_rate": 7.640779610194902e-06, + "loss": 2.4638999938964843, + "step": 227200 + }, + { + "epoch": 0.6185381788485096, + "grad_norm": 2.1521530151367188, + "learning_rate": 7.639689246285948e-06, + "loss": 2.487919044494629, + "step": 227220 + }, + { + "epoch": 0.6185926228392541, + "grad_norm": 2.220296621322632, + "learning_rate": 7.638598882376994e-06, + "loss": 2.4590694427490236, + "step": 227240 + }, + { + "epoch": 0.6186470668299986, + "grad_norm": 2.0122251510620117, + "learning_rate": 7.63750851846804e-06, + "loss": 2.4645896911621095, + "step": 227260 + }, + { + "epoch": 0.6187015108207432, + "grad_norm": 2.265620470046997, + "learning_rate": 7.636418154559084e-06, + "loss": 2.474802017211914, + "step": 227280 + }, + { + "epoch": 0.6187559548114877, + "grad_norm": 2.2099788188934326, + "learning_rate": 7.63532779065013e-06, + "loss": 2.465859603881836, + "step": 227300 + }, + { + "epoch": 0.6188103988022322, + "grad_norm": 2.2169432640075684, + "learning_rate": 7.634237426741176e-06, + "loss": 2.4598169326782227, + "step": 227320 + }, + { + "epoch": 0.6188648427929767, + "grad_norm": 2.2538342475891113, + "learning_rate": 7.633147062832221e-06, + "loss": 2.480426216125488, + "step": 227340 + }, + { + "epoch": 0.6189192867837212, + "grad_norm": 2.116527795791626, + "learning_rate": 7.632056698923265e-06, + "loss": 2.4738311767578125, + "step": 227360 + }, + { + "epoch": 0.6189737307744657, + "grad_norm": 2.2582991123199463, + "learning_rate": 7.630966335014311e-06, + "loss": 2.4998748779296873, + "step": 227380 + }, + { + "epoch": 0.6190281747652103, + "grad_norm": 2.1164374351501465, + "learning_rate": 7.629875971105357e-06, + "loss": 2.449471855163574, + "step": 227400 + }, + { + "epoch": 0.6190826187559548, + "grad_norm": 2.6693239212036133, + "learning_rate": 7.628785607196403e-06, + "loss": 2.4942228317260744, + "step": 227420 + }, + { + "epoch": 0.6191370627466993, + "grad_norm": 2.2500619888305664, + "learning_rate": 7.627695243287447e-06, + "loss": 2.488674545288086, + "step": 227440 + }, + { + "epoch": 0.6191915067374438, + "grad_norm": 2.1224021911621094, + "learning_rate": 7.626604879378493e-06, + "loss": 2.4303155899047852, + "step": 227460 + }, + { + "epoch": 0.6192459507281883, + "grad_norm": 2.1781177520751953, + "learning_rate": 7.625514515469539e-06, + "loss": 2.459257698059082, + "step": 227480 + }, + { + "epoch": 0.619300394718933, + "grad_norm": 2.9265754222869873, + "learning_rate": 7.6244241515605845e-06, + "loss": 2.459926986694336, + "step": 227500 + }, + { + "epoch": 0.6193548387096774, + "grad_norm": 2.107530117034912, + "learning_rate": 7.6233337876516286e-06, + "loss": 2.4883827209472655, + "step": 227520 + }, + { + "epoch": 0.619409282700422, + "grad_norm": 2.136533498764038, + "learning_rate": 7.622243423742674e-06, + "loss": 2.4650140762329102, + "step": 227540 + }, + { + "epoch": 0.6194637266911664, + "grad_norm": 2.7331154346466064, + "learning_rate": 7.62115305983372e-06, + "loss": 2.4513900756835936, + "step": 227560 + }, + { + "epoch": 0.619518170681911, + "grad_norm": 2.3706514835357666, + "learning_rate": 7.620062695924766e-06, + "loss": 2.472591018676758, + "step": 227580 + }, + { + "epoch": 0.6195726146726555, + "grad_norm": 2.1746232509613037, + "learning_rate": 7.61897233201581e-06, + "loss": 2.4579675674438475, + "step": 227600 + }, + { + "epoch": 0.6196270586634001, + "grad_norm": 2.150618076324463, + "learning_rate": 7.617881968106856e-06, + "loss": 2.4613380432128906, + "step": 227620 + }, + { + "epoch": 0.6196815026541446, + "grad_norm": 2.1861793994903564, + "learning_rate": 7.616791604197902e-06, + "loss": 2.48095703125, + "step": 227640 + }, + { + "epoch": 0.6197359466448891, + "grad_norm": 2.101374626159668, + "learning_rate": 7.6157012402889475e-06, + "loss": 2.467572021484375, + "step": 227660 + }, + { + "epoch": 0.6197903906356336, + "grad_norm": 2.123462677001953, + "learning_rate": 7.614610876379992e-06, + "loss": 2.4746831893920898, + "step": 227680 + }, + { + "epoch": 0.6198448346263781, + "grad_norm": 2.1416499614715576, + "learning_rate": 7.6135205124710375e-06, + "loss": 2.4932601928710936, + "step": 227700 + }, + { + "epoch": 0.6198992786171227, + "grad_norm": 2.1941819190979004, + "learning_rate": 7.612430148562083e-06, + "loss": 2.454051208496094, + "step": 227720 + }, + { + "epoch": 0.6199537226078672, + "grad_norm": 2.256016731262207, + "learning_rate": 7.611339784653129e-06, + "loss": 2.4778701782226564, + "step": 227740 + }, + { + "epoch": 0.6200081665986117, + "grad_norm": 2.109104871749878, + "learning_rate": 7.610249420744173e-06, + "loss": 2.477067565917969, + "step": 227760 + }, + { + "epoch": 0.6200626105893562, + "grad_norm": 2.2145581245422363, + "learning_rate": 7.609159056835219e-06, + "loss": 2.448879432678223, + "step": 227780 + }, + { + "epoch": 0.6201170545801007, + "grad_norm": 2.1808760166168213, + "learning_rate": 7.608068692926265e-06, + "loss": 2.4741596221923827, + "step": 227800 + }, + { + "epoch": 0.6201714985708452, + "grad_norm": 2.340757131576538, + "learning_rate": 7.606978329017311e-06, + "loss": 2.4831005096435548, + "step": 227820 + }, + { + "epoch": 0.6202259425615898, + "grad_norm": 2.238795518875122, + "learning_rate": 7.605887965108355e-06, + "loss": 2.4685386657714843, + "step": 227840 + }, + { + "epoch": 0.6202803865523343, + "grad_norm": 2.093332052230835, + "learning_rate": 7.6047976011994006e-06, + "loss": 2.4802093505859375, + "step": 227860 + }, + { + "epoch": 0.6203348305430788, + "grad_norm": 2.1868529319763184, + "learning_rate": 7.603707237290446e-06, + "loss": 2.486197280883789, + "step": 227880 + }, + { + "epoch": 0.6203892745338233, + "grad_norm": 2.152945041656494, + "learning_rate": 7.602616873381492e-06, + "loss": 2.4435714721679687, + "step": 227900 + }, + { + "epoch": 0.6204437185245678, + "grad_norm": 2.0899887084960938, + "learning_rate": 7.601526509472536e-06, + "loss": 2.4469320297241213, + "step": 227920 + }, + { + "epoch": 0.6204981625153124, + "grad_norm": 2.5238192081451416, + "learning_rate": 7.600436145563582e-06, + "loss": 2.478878211975098, + "step": 227940 + }, + { + "epoch": 0.6205526065060569, + "grad_norm": 2.157953977584839, + "learning_rate": 7.599345781654628e-06, + "loss": 2.4857656478881838, + "step": 227960 + }, + { + "epoch": 0.6206070504968014, + "grad_norm": 2.2516214847564697, + "learning_rate": 7.598255417745674e-06, + "loss": 2.4877647399902343, + "step": 227980 + }, + { + "epoch": 0.6206614944875459, + "grad_norm": 2.1914212703704834, + "learning_rate": 7.597165053836718e-06, + "loss": 2.480575752258301, + "step": 228000 + }, + { + "epoch": 0.6207159384782904, + "grad_norm": 3.1988468170166016, + "learning_rate": 7.596074689927764e-06, + "loss": 2.4572015762329102, + "step": 228020 + }, + { + "epoch": 0.6207703824690349, + "grad_norm": 2.5070133209228516, + "learning_rate": 7.5949843260188094e-06, + "loss": 2.4637128829956056, + "step": 228040 + }, + { + "epoch": 0.6208248264597795, + "grad_norm": 2.1023571491241455, + "learning_rate": 7.593893962109855e-06, + "loss": 2.4699275970458983, + "step": 228060 + }, + { + "epoch": 0.620879270450524, + "grad_norm": 2.1055731773376465, + "learning_rate": 7.592803598200899e-06, + "loss": 2.4819662094116213, + "step": 228080 + }, + { + "epoch": 0.6209337144412685, + "grad_norm": 2.134483814239502, + "learning_rate": 7.591713234291945e-06, + "loss": 2.4948505401611327, + "step": 228100 + }, + { + "epoch": 0.620988158432013, + "grad_norm": 2.2809996604919434, + "learning_rate": 7.590622870382991e-06, + "loss": 2.512569618225098, + "step": 228120 + }, + { + "epoch": 0.6210426024227576, + "grad_norm": 2.2853872776031494, + "learning_rate": 7.589532506474037e-06, + "loss": 2.489715003967285, + "step": 228140 + }, + { + "epoch": 0.6210970464135022, + "grad_norm": 2.1001412868499756, + "learning_rate": 7.588442142565083e-06, + "loss": 2.4831804275512694, + "step": 228160 + }, + { + "epoch": 0.6211514904042467, + "grad_norm": 2.2545456886291504, + "learning_rate": 7.587351778656127e-06, + "loss": 2.5007806777954102, + "step": 228180 + }, + { + "epoch": 0.6212059343949912, + "grad_norm": 2.1609694957733154, + "learning_rate": 7.5862614147471725e-06, + "loss": 2.4809053421020506, + "step": 228200 + }, + { + "epoch": 0.6212603783857357, + "grad_norm": 2.3005545139312744, + "learning_rate": 7.585171050838218e-06, + "loss": 2.485990524291992, + "step": 228220 + }, + { + "epoch": 0.6213148223764802, + "grad_norm": 2.1718180179595947, + "learning_rate": 7.584080686929264e-06, + "loss": 2.4684371948242188, + "step": 228240 + }, + { + "epoch": 0.6213692663672247, + "grad_norm": 2.1241981983184814, + "learning_rate": 7.582990323020308e-06, + "loss": 2.437818717956543, + "step": 228260 + }, + { + "epoch": 0.6214237103579693, + "grad_norm": 2.088829278945923, + "learning_rate": 7.581899959111354e-06, + "loss": 2.474499320983887, + "step": 228280 + }, + { + "epoch": 0.6214781543487138, + "grad_norm": 2.206294059753418, + "learning_rate": 7.5808095952024e-06, + "loss": 2.4913673400878906, + "step": 228300 + }, + { + "epoch": 0.6215325983394583, + "grad_norm": 2.3137311935424805, + "learning_rate": 7.579719231293446e-06, + "loss": 2.4574779510498046, + "step": 228320 + }, + { + "epoch": 0.6215870423302028, + "grad_norm": 2.2468936443328857, + "learning_rate": 7.57862886738449e-06, + "loss": 2.4690601348876955, + "step": 228340 + }, + { + "epoch": 0.6216414863209473, + "grad_norm": 2.0692684650421143, + "learning_rate": 7.577538503475536e-06, + "loss": 2.4719539642333985, + "step": 228360 + }, + { + "epoch": 0.6216959303116919, + "grad_norm": 2.1998307704925537, + "learning_rate": 7.5764481395665814e-06, + "loss": 2.4711624145507813, + "step": 228380 + }, + { + "epoch": 0.6217503743024364, + "grad_norm": 2.2007803916931152, + "learning_rate": 7.575357775657627e-06, + "loss": 2.4821868896484376, + "step": 228400 + }, + { + "epoch": 0.6218048182931809, + "grad_norm": 2.0524826049804688, + "learning_rate": 7.574267411748671e-06, + "loss": 2.494528961181641, + "step": 228420 + }, + { + "epoch": 0.6218592622839254, + "grad_norm": 2.1657705307006836, + "learning_rate": 7.573177047839717e-06, + "loss": 2.4677757263183593, + "step": 228440 + }, + { + "epoch": 0.6219137062746699, + "grad_norm": 2.2106313705444336, + "learning_rate": 7.572086683930763e-06, + "loss": 2.477797508239746, + "step": 228460 + }, + { + "epoch": 0.6219681502654144, + "grad_norm": 2.063401222229004, + "learning_rate": 7.570996320021809e-06, + "loss": 2.4949689865112306, + "step": 228480 + }, + { + "epoch": 0.622022594256159, + "grad_norm": 2.053786516189575, + "learning_rate": 7.569905956112853e-06, + "loss": 2.484849739074707, + "step": 228500 + }, + { + "epoch": 0.6220770382469035, + "grad_norm": 2.2846686840057373, + "learning_rate": 7.568815592203899e-06, + "loss": 2.4752487182617187, + "step": 228520 + }, + { + "epoch": 0.622131482237648, + "grad_norm": 2.1772077083587646, + "learning_rate": 7.5677252282949445e-06, + "loss": 2.458662414550781, + "step": 228540 + }, + { + "epoch": 0.6221859262283925, + "grad_norm": 2.1398966312408447, + "learning_rate": 7.56663486438599e-06, + "loss": 2.4585947036743163, + "step": 228560 + }, + { + "epoch": 0.622240370219137, + "grad_norm": 2.2989463806152344, + "learning_rate": 7.5655445004770344e-06, + "loss": 2.466768264770508, + "step": 228580 + }, + { + "epoch": 0.6222948142098815, + "grad_norm": 2.152177095413208, + "learning_rate": 7.56445413656808e-06, + "loss": 2.4593029022216797, + "step": 228600 + }, + { + "epoch": 0.6223492582006261, + "grad_norm": 2.288113594055176, + "learning_rate": 7.563363772659126e-06, + "loss": 2.458011817932129, + "step": 228620 + }, + { + "epoch": 0.6224037021913706, + "grad_norm": 2.175004243850708, + "learning_rate": 7.562273408750172e-06, + "loss": 2.464383125305176, + "step": 228640 + }, + { + "epoch": 0.6224581461821151, + "grad_norm": 2.1338984966278076, + "learning_rate": 7.561183044841216e-06, + "loss": 2.4936687469482424, + "step": 228660 + }, + { + "epoch": 0.6225125901728596, + "grad_norm": 2.116960048675537, + "learning_rate": 7.560092680932262e-06, + "loss": 2.4933971405029296, + "step": 228680 + }, + { + "epoch": 0.6225670341636041, + "grad_norm": 2.1627869606018066, + "learning_rate": 7.559002317023308e-06, + "loss": 2.4629859924316406, + "step": 228700 + }, + { + "epoch": 0.6226214781543488, + "grad_norm": 2.0771632194519043, + "learning_rate": 7.557911953114353e-06, + "loss": 2.4599935531616213, + "step": 228720 + }, + { + "epoch": 0.6226759221450933, + "grad_norm": 2.2013373374938965, + "learning_rate": 7.5568215892053975e-06, + "loss": 2.4767406463623045, + "step": 228740 + }, + { + "epoch": 0.6227303661358378, + "grad_norm": 2.079143762588501, + "learning_rate": 7.555731225296443e-06, + "loss": 2.4973440170288086, + "step": 228760 + }, + { + "epoch": 0.6227848101265823, + "grad_norm": 2.2440879344940186, + "learning_rate": 7.554640861387489e-06, + "loss": 2.486433982849121, + "step": 228780 + }, + { + "epoch": 0.6228392541173268, + "grad_norm": 2.114293098449707, + "learning_rate": 7.553550497478535e-06, + "loss": 2.494557189941406, + "step": 228800 + }, + { + "epoch": 0.6228936981080713, + "grad_norm": 2.1696603298187256, + "learning_rate": 7.552460133569579e-06, + "loss": 2.456584930419922, + "step": 228820 + }, + { + "epoch": 0.6229481420988159, + "grad_norm": 2.0615549087524414, + "learning_rate": 7.551369769660625e-06, + "loss": 2.5076482772827147, + "step": 228840 + }, + { + "epoch": 0.6230025860895604, + "grad_norm": 2.1806466579437256, + "learning_rate": 7.550279405751671e-06, + "loss": 2.4487482070922852, + "step": 228860 + }, + { + "epoch": 0.6230570300803049, + "grad_norm": 2.132864475250244, + "learning_rate": 7.549189041842716e-06, + "loss": 2.484320068359375, + "step": 228880 + }, + { + "epoch": 0.6231114740710494, + "grad_norm": 2.1981730461120605, + "learning_rate": 7.548098677933761e-06, + "loss": 2.488617706298828, + "step": 228900 + }, + { + "epoch": 0.6231659180617939, + "grad_norm": 2.2209222316741943, + "learning_rate": 7.547008314024806e-06, + "loss": 2.460471534729004, + "step": 228920 + }, + { + "epoch": 0.6232203620525385, + "grad_norm": 2.0749495029449463, + "learning_rate": 7.545917950115852e-06, + "loss": 2.467448616027832, + "step": 228940 + }, + { + "epoch": 0.623274806043283, + "grad_norm": 2.5094032287597656, + "learning_rate": 7.544827586206897e-06, + "loss": 2.4744823455810545, + "step": 228960 + }, + { + "epoch": 0.6233292500340275, + "grad_norm": 2.1446280479431152, + "learning_rate": 7.543737222297942e-06, + "loss": 2.470302772521973, + "step": 228980 + }, + { + "epoch": 0.623383694024772, + "grad_norm": 2.0651803016662598, + "learning_rate": 7.542646858388988e-06, + "loss": 2.486033058166504, + "step": 229000 + }, + { + "epoch": 0.6234381380155165, + "grad_norm": 2.0593299865722656, + "learning_rate": 7.541556494480034e-06, + "loss": 2.471576690673828, + "step": 229020 + }, + { + "epoch": 0.623492582006261, + "grad_norm": 2.1560819149017334, + "learning_rate": 7.540466130571079e-06, + "loss": 2.4621532440185545, + "step": 229040 + }, + { + "epoch": 0.6235470259970056, + "grad_norm": 2.209747552871704, + "learning_rate": 7.539375766662124e-06, + "loss": 2.4633480072021485, + "step": 229060 + }, + { + "epoch": 0.6236014699877501, + "grad_norm": 2.19922137260437, + "learning_rate": 7.5382854027531695e-06, + "loss": 2.459077072143555, + "step": 229080 + }, + { + "epoch": 0.6236559139784946, + "grad_norm": 2.174464702606201, + "learning_rate": 7.537195038844215e-06, + "loss": 2.46398983001709, + "step": 229100 + }, + { + "epoch": 0.6237103579692391, + "grad_norm": 2.06182861328125, + "learning_rate": 7.53610467493526e-06, + "loss": 2.4586822509765627, + "step": 229120 + }, + { + "epoch": 0.6237648019599836, + "grad_norm": 2.3282341957092285, + "learning_rate": 7.535014311026305e-06, + "loss": 2.481538009643555, + "step": 229140 + }, + { + "epoch": 0.6238192459507282, + "grad_norm": 2.116852283477783, + "learning_rate": 7.533923947117351e-06, + "loss": 2.4679035186767577, + "step": 229160 + }, + { + "epoch": 0.6238736899414727, + "grad_norm": 2.217656373977661, + "learning_rate": 7.532833583208397e-06, + "loss": 2.471934509277344, + "step": 229180 + }, + { + "epoch": 0.6239281339322172, + "grad_norm": 2.1275277137756348, + "learning_rate": 7.531743219299442e-06, + "loss": 2.4695878982543946, + "step": 229200 + }, + { + "epoch": 0.6239825779229617, + "grad_norm": 2.2039682865142822, + "learning_rate": 7.530652855390487e-06, + "loss": 2.4593143463134766, + "step": 229220 + }, + { + "epoch": 0.6240370219137062, + "grad_norm": 2.1216349601745605, + "learning_rate": 7.529562491481533e-06, + "loss": 2.472566032409668, + "step": 229240 + }, + { + "epoch": 0.6240914659044507, + "grad_norm": 2.1036887168884277, + "learning_rate": 7.528472127572578e-06, + "loss": 2.4886226654052734, + "step": 229260 + }, + { + "epoch": 0.6241459098951954, + "grad_norm": 2.2371692657470703, + "learning_rate": 7.527381763663623e-06, + "loss": 2.4752870559692384, + "step": 229280 + }, + { + "epoch": 0.6242003538859399, + "grad_norm": 2.1738288402557373, + "learning_rate": 7.526291399754668e-06, + "loss": 2.4539308547973633, + "step": 229300 + }, + { + "epoch": 0.6242547978766844, + "grad_norm": 2.1843388080596924, + "learning_rate": 7.525201035845714e-06, + "loss": 2.461543846130371, + "step": 229320 + }, + { + "epoch": 0.6243092418674289, + "grad_norm": 2.2851040363311768, + "learning_rate": 7.52411067193676e-06, + "loss": 2.4521728515625, + "step": 229340 + }, + { + "epoch": 0.6243636858581734, + "grad_norm": 2.1795215606689453, + "learning_rate": 7.523020308027805e-06, + "loss": 2.4454730987548827, + "step": 229360 + }, + { + "epoch": 0.624418129848918, + "grad_norm": 2.1361587047576904, + "learning_rate": 7.52192994411885e-06, + "loss": 2.4452213287353515, + "step": 229380 + }, + { + "epoch": 0.6244725738396625, + "grad_norm": 2.214980125427246, + "learning_rate": 7.520839580209896e-06, + "loss": 2.465191650390625, + "step": 229400 + }, + { + "epoch": 0.624527017830407, + "grad_norm": 2.282719373703003, + "learning_rate": 7.5197492163009415e-06, + "loss": 2.448017883300781, + "step": 229420 + }, + { + "epoch": 0.6245814618211515, + "grad_norm": 2.118536949157715, + "learning_rate": 7.5186588523919864e-06, + "loss": 2.4410703659057615, + "step": 229440 + }, + { + "epoch": 0.624635905811896, + "grad_norm": 2.1468825340270996, + "learning_rate": 7.517568488483031e-06, + "loss": 2.506937789916992, + "step": 229460 + }, + { + "epoch": 0.6246903498026405, + "grad_norm": 2.362804412841797, + "learning_rate": 7.516478124574077e-06, + "loss": 2.4683696746826174, + "step": 229480 + }, + { + "epoch": 0.6247447937933851, + "grad_norm": 2.1486639976501465, + "learning_rate": 7.515387760665123e-06, + "loss": 2.474043083190918, + "step": 229500 + }, + { + "epoch": 0.6247992377841296, + "grad_norm": 2.196563720703125, + "learning_rate": 7.514297396756168e-06, + "loss": 2.4785703659057616, + "step": 229520 + }, + { + "epoch": 0.6248536817748741, + "grad_norm": 2.150219678878784, + "learning_rate": 7.513207032847213e-06, + "loss": 2.4606149673461912, + "step": 229540 + }, + { + "epoch": 0.6249081257656186, + "grad_norm": 2.1927802562713623, + "learning_rate": 7.512116668938259e-06, + "loss": 2.457206153869629, + "step": 229560 + }, + { + "epoch": 0.6249625697563631, + "grad_norm": 6.460984706878662, + "learning_rate": 7.5110263050293046e-06, + "loss": 2.450749969482422, + "step": 229580 + }, + { + "epoch": 0.6250170137471076, + "grad_norm": 2.184659004211426, + "learning_rate": 7.5099359411203495e-06, + "loss": 2.4406856536865233, + "step": 229600 + }, + { + "epoch": 0.6250714577378522, + "grad_norm": 2.178602695465088, + "learning_rate": 7.5088455772113945e-06, + "loss": 2.49209041595459, + "step": 229620 + }, + { + "epoch": 0.6251259017285967, + "grad_norm": 2.480909585952759, + "learning_rate": 7.50775521330244e-06, + "loss": 2.490546226501465, + "step": 229640 + }, + { + "epoch": 0.6251803457193412, + "grad_norm": 2.1444382667541504, + "learning_rate": 7.506664849393486e-06, + "loss": 2.4609485626220704, + "step": 229660 + }, + { + "epoch": 0.6252347897100857, + "grad_norm": 2.1949548721313477, + "learning_rate": 7.505574485484531e-06, + "loss": 2.4860448837280273, + "step": 229680 + }, + { + "epoch": 0.6252892337008302, + "grad_norm": 2.1188313961029053, + "learning_rate": 7.504484121575576e-06, + "loss": 2.480899620056152, + "step": 229700 + }, + { + "epoch": 0.6253436776915748, + "grad_norm": 2.1625845432281494, + "learning_rate": 7.503393757666622e-06, + "loss": 2.5004817962646486, + "step": 229720 + }, + { + "epoch": 0.6253981216823193, + "grad_norm": 2.171377658843994, + "learning_rate": 7.502303393757668e-06, + "loss": 2.487691116333008, + "step": 229740 + }, + { + "epoch": 0.6254525656730638, + "grad_norm": 2.186976671218872, + "learning_rate": 7.501213029848713e-06, + "loss": 2.492814826965332, + "step": 229760 + }, + { + "epoch": 0.6255070096638083, + "grad_norm": 2.2057206630706787, + "learning_rate": 7.500122665939758e-06, + "loss": 2.4549673080444334, + "step": 229780 + }, + { + "epoch": 0.6255614536545528, + "grad_norm": 2.2973408699035645, + "learning_rate": 7.499032302030803e-06, + "loss": 2.483590316772461, + "step": 229800 + }, + { + "epoch": 0.6256158976452973, + "grad_norm": 2.2232656478881836, + "learning_rate": 7.497941938121849e-06, + "loss": 2.492728042602539, + "step": 229820 + }, + { + "epoch": 0.625670341636042, + "grad_norm": 2.073017120361328, + "learning_rate": 7.496851574212894e-06, + "loss": 2.490154266357422, + "step": 229840 + }, + { + "epoch": 0.6257247856267865, + "grad_norm": 2.102863311767578, + "learning_rate": 7.49576121030394e-06, + "loss": 2.440353775024414, + "step": 229860 + }, + { + "epoch": 0.625779229617531, + "grad_norm": 2.218244791030884, + "learning_rate": 7.494670846394985e-06, + "loss": 2.4981010437011717, + "step": 229880 + }, + { + "epoch": 0.6258336736082755, + "grad_norm": 2.295116424560547, + "learning_rate": 7.493580482486031e-06, + "loss": 2.4614057540893555, + "step": 229900 + }, + { + "epoch": 0.62588811759902, + "grad_norm": 2.186541795730591, + "learning_rate": 7.492490118577076e-06, + "loss": 2.4605918884277345, + "step": 229920 + }, + { + "epoch": 0.6259425615897646, + "grad_norm": 2.125831365585327, + "learning_rate": 7.4913997546681215e-06, + "loss": 2.4661155700683595, + "step": 229940 + }, + { + "epoch": 0.6259970055805091, + "grad_norm": 2.6744604110717773, + "learning_rate": 7.4903093907591665e-06, + "loss": 2.4840938568115236, + "step": 229960 + }, + { + "epoch": 0.6260514495712536, + "grad_norm": 2.2345285415649414, + "learning_rate": 7.4892190268502114e-06, + "loss": 2.473202133178711, + "step": 229980 + }, + { + "epoch": 0.6261058935619981, + "grad_norm": 2.2330079078674316, + "learning_rate": 7.488128662941257e-06, + "loss": 2.5070941925048826, + "step": 230000 + }, + { + "epoch": 0.6261603375527426, + "grad_norm": 2.134011745452881, + "learning_rate": 7.487038299032303e-06, + "loss": 2.464596176147461, + "step": 230020 + }, + { + "epoch": 0.6262147815434871, + "grad_norm": 2.2170443534851074, + "learning_rate": 7.485947935123348e-06, + "loss": 2.4792596817016603, + "step": 230040 + }, + { + "epoch": 0.6262692255342317, + "grad_norm": 2.1778483390808105, + "learning_rate": 7.484857571214393e-06, + "loss": 2.491066551208496, + "step": 230060 + }, + { + "epoch": 0.6263236695249762, + "grad_norm": 2.282414674758911, + "learning_rate": 7.483767207305439e-06, + "loss": 2.4716697692871095, + "step": 230080 + }, + { + "epoch": 0.6263781135157207, + "grad_norm": 2.174992561340332, + "learning_rate": 7.482676843396485e-06, + "loss": 2.5052005767822267, + "step": 230100 + }, + { + "epoch": 0.6264325575064652, + "grad_norm": 2.1469533443450928, + "learning_rate": 7.4815864794875296e-06, + "loss": 2.481916809082031, + "step": 230120 + }, + { + "epoch": 0.6264870014972097, + "grad_norm": 2.1143829822540283, + "learning_rate": 7.4804961155785745e-06, + "loss": 2.4674072265625, + "step": 230140 + }, + { + "epoch": 0.6265414454879543, + "grad_norm": 2.4752285480499268, + "learning_rate": 7.47940575166962e-06, + "loss": 2.4833450317382812, + "step": 230160 + }, + { + "epoch": 0.6265958894786988, + "grad_norm": 2.2965028285980225, + "learning_rate": 7.478315387760666e-06, + "loss": 2.4766969680786133, + "step": 230180 + }, + { + "epoch": 0.6266503334694433, + "grad_norm": 2.1427130699157715, + "learning_rate": 7.477225023851711e-06, + "loss": 2.4568872451782227, + "step": 230200 + }, + { + "epoch": 0.6267047774601878, + "grad_norm": 2.169015645980835, + "learning_rate": 7.476134659942756e-06, + "loss": 2.439173698425293, + "step": 230220 + }, + { + "epoch": 0.6267592214509323, + "grad_norm": 2.353116273880005, + "learning_rate": 7.475044296033802e-06, + "loss": 2.491575241088867, + "step": 230240 + }, + { + "epoch": 0.6268136654416768, + "grad_norm": 2.205767869949341, + "learning_rate": 7.473953932124848e-06, + "loss": 2.4829547882080076, + "step": 230260 + }, + { + "epoch": 0.6268681094324214, + "grad_norm": 2.1880548000335693, + "learning_rate": 7.472863568215893e-06, + "loss": 2.462065887451172, + "step": 230280 + }, + { + "epoch": 0.6269225534231659, + "grad_norm": 2.271059989929199, + "learning_rate": 7.471773204306938e-06, + "loss": 2.4811786651611327, + "step": 230300 + }, + { + "epoch": 0.6269769974139104, + "grad_norm": 2.1344971656799316, + "learning_rate": 7.470682840397983e-06, + "loss": 2.501170349121094, + "step": 230320 + }, + { + "epoch": 0.6270314414046549, + "grad_norm": 2.253282070159912, + "learning_rate": 7.469592476489029e-06, + "loss": 2.48148250579834, + "step": 230340 + }, + { + "epoch": 0.6270858853953994, + "grad_norm": 2.196054220199585, + "learning_rate": 7.468502112580074e-06, + "loss": 2.4721303939819337, + "step": 230360 + }, + { + "epoch": 0.6271403293861441, + "grad_norm": 2.3273262977600098, + "learning_rate": 7.467411748671119e-06, + "loss": 2.4643884658813477, + "step": 230380 + }, + { + "epoch": 0.6271947733768886, + "grad_norm": 2.096087694168091, + "learning_rate": 7.466321384762165e-06, + "loss": 2.448300361633301, + "step": 230400 + }, + { + "epoch": 0.6272492173676331, + "grad_norm": 2.1618988513946533, + "learning_rate": 7.465231020853211e-06, + "loss": 2.472824478149414, + "step": 230420 + }, + { + "epoch": 0.6273036613583776, + "grad_norm": 2.1790168285369873, + "learning_rate": 7.464140656944256e-06, + "loss": 2.465321159362793, + "step": 230440 + }, + { + "epoch": 0.6273581053491221, + "grad_norm": 2.2204372882843018, + "learning_rate": 7.463050293035301e-06, + "loss": 2.4762599945068358, + "step": 230460 + }, + { + "epoch": 0.6274125493398666, + "grad_norm": 2.403873920440674, + "learning_rate": 7.4619599291263465e-06, + "loss": 2.4611751556396486, + "step": 230480 + }, + { + "epoch": 0.6274669933306112, + "grad_norm": 2.109057664871216, + "learning_rate": 7.460869565217392e-06, + "loss": 2.4377445220947265, + "step": 230500 + }, + { + "epoch": 0.6275214373213557, + "grad_norm": 2.1576759815216064, + "learning_rate": 7.459779201308437e-06, + "loss": 2.439243698120117, + "step": 230520 + }, + { + "epoch": 0.6275758813121002, + "grad_norm": 2.2721588611602783, + "learning_rate": 7.458688837399482e-06, + "loss": 2.441938018798828, + "step": 230540 + }, + { + "epoch": 0.6276303253028447, + "grad_norm": 2.1101932525634766, + "learning_rate": 7.457598473490528e-06, + "loss": 2.4790239334106445, + "step": 230560 + }, + { + "epoch": 0.6276847692935892, + "grad_norm": 2.0992817878723145, + "learning_rate": 7.456508109581574e-06, + "loss": 2.4935998916625977, + "step": 230580 + }, + { + "epoch": 0.6277392132843338, + "grad_norm": 2.095459222793579, + "learning_rate": 7.455417745672619e-06, + "loss": 2.485729217529297, + "step": 230600 + }, + { + "epoch": 0.6277936572750783, + "grad_norm": 2.1502950191497803, + "learning_rate": 7.454327381763664e-06, + "loss": 2.4863683700561525, + "step": 230620 + }, + { + "epoch": 0.6278481012658228, + "grad_norm": 2.1705305576324463, + "learning_rate": 7.45323701785471e-06, + "loss": 2.4707387924194335, + "step": 230640 + }, + { + "epoch": 0.6279025452565673, + "grad_norm": 2.463115930557251, + "learning_rate": 7.452146653945755e-06, + "loss": 2.466107940673828, + "step": 230660 + }, + { + "epoch": 0.6279569892473118, + "grad_norm": 2.1631104946136475, + "learning_rate": 7.4510562900368e-06, + "loss": 2.4649696350097656, + "step": 230680 + }, + { + "epoch": 0.6280114332380563, + "grad_norm": 2.0976288318634033, + "learning_rate": 7.449965926127845e-06, + "loss": 2.4781909942626954, + "step": 230700 + }, + { + "epoch": 0.6280658772288009, + "grad_norm": 2.139030694961548, + "learning_rate": 7.448875562218891e-06, + "loss": 2.4642541885375975, + "step": 230720 + }, + { + "epoch": 0.6281203212195454, + "grad_norm": 2.1787753105163574, + "learning_rate": 7.447785198309937e-06, + "loss": 2.4520822525024415, + "step": 230740 + }, + { + "epoch": 0.6281747652102899, + "grad_norm": 2.2834064960479736, + "learning_rate": 7.446694834400982e-06, + "loss": 2.5051937103271484, + "step": 230760 + }, + { + "epoch": 0.6282292092010344, + "grad_norm": 2.225154161453247, + "learning_rate": 7.445604470492027e-06, + "loss": 2.478511619567871, + "step": 230780 + }, + { + "epoch": 0.6282836531917789, + "grad_norm": 2.3053596019744873, + "learning_rate": 7.444514106583073e-06, + "loss": 2.455240821838379, + "step": 230800 + }, + { + "epoch": 0.6283380971825234, + "grad_norm": 2.366610288619995, + "learning_rate": 7.4434237426741185e-06, + "loss": 2.480540084838867, + "step": 230820 + }, + { + "epoch": 0.628392541173268, + "grad_norm": 2.1513428688049316, + "learning_rate": 7.4423333787651634e-06, + "loss": 2.450700569152832, + "step": 230840 + }, + { + "epoch": 0.6284469851640125, + "grad_norm": 2.1573052406311035, + "learning_rate": 7.441243014856208e-06, + "loss": 2.5003610610961915, + "step": 230860 + }, + { + "epoch": 0.628501429154757, + "grad_norm": 2.1127631664276123, + "learning_rate": 7.440152650947254e-06, + "loss": 2.4808280944824217, + "step": 230880 + }, + { + "epoch": 0.6285558731455015, + "grad_norm": 2.1559388637542725, + "learning_rate": 7.4390622870383e-06, + "loss": 2.4681770324707033, + "step": 230900 + }, + { + "epoch": 0.628610317136246, + "grad_norm": 2.1865108013153076, + "learning_rate": 7.437971923129345e-06, + "loss": 2.4627525329589846, + "step": 230920 + }, + { + "epoch": 0.6286647611269907, + "grad_norm": 2.1992852687835693, + "learning_rate": 7.43688155922039e-06, + "loss": 2.4619016647338867, + "step": 230940 + }, + { + "epoch": 0.6287192051177352, + "grad_norm": 2.114311456680298, + "learning_rate": 7.435791195311436e-06, + "loss": 2.4665294647216798, + "step": 230960 + }, + { + "epoch": 0.6287736491084797, + "grad_norm": 2.1472389698028564, + "learning_rate": 7.4347008314024816e-06, + "loss": 2.477443504333496, + "step": 230980 + }, + { + "epoch": 0.6288280930992242, + "grad_norm": 2.2998833656311035, + "learning_rate": 7.433610467493526e-06, + "loss": 2.4697572708129885, + "step": 231000 + }, + { + "epoch": 0.6288825370899687, + "grad_norm": 2.1880970001220703, + "learning_rate": 7.4325201035845715e-06, + "loss": 2.475995635986328, + "step": 231020 + }, + { + "epoch": 0.6289369810807132, + "grad_norm": 2.1593878269195557, + "learning_rate": 7.431429739675617e-06, + "loss": 2.5111261367797852, + "step": 231040 + }, + { + "epoch": 0.6289914250714578, + "grad_norm": 2.1119163036346436, + "learning_rate": 7.430339375766663e-06, + "loss": 2.4649110794067384, + "step": 231060 + }, + { + "epoch": 0.6290458690622023, + "grad_norm": 2.0850870609283447, + "learning_rate": 7.429249011857707e-06, + "loss": 2.4869089126586914, + "step": 231080 + }, + { + "epoch": 0.6291003130529468, + "grad_norm": 2.342102527618408, + "learning_rate": 7.428158647948753e-06, + "loss": 2.468288612365723, + "step": 231100 + }, + { + "epoch": 0.6291547570436913, + "grad_norm": 2.2324492931365967, + "learning_rate": 7.427068284039799e-06, + "loss": 2.4840354919433594, + "step": 231120 + }, + { + "epoch": 0.6292092010344358, + "grad_norm": 2.212416410446167, + "learning_rate": 7.425977920130845e-06, + "loss": 2.4854499816894533, + "step": 231140 + }, + { + "epoch": 0.6292636450251804, + "grad_norm": 2.3143653869628906, + "learning_rate": 7.424887556221889e-06, + "loss": 2.50531063079834, + "step": 231160 + }, + { + "epoch": 0.6293180890159249, + "grad_norm": 2.2811717987060547, + "learning_rate": 7.423797192312935e-06, + "loss": 2.5140764236450197, + "step": 231180 + }, + { + "epoch": 0.6293725330066694, + "grad_norm": 2.092599391937256, + "learning_rate": 7.42270682840398e-06, + "loss": 2.4661582946777343, + "step": 231200 + }, + { + "epoch": 0.6294269769974139, + "grad_norm": 2.1679954528808594, + "learning_rate": 7.421616464495026e-06, + "loss": 2.4704568862915037, + "step": 231220 + }, + { + "epoch": 0.6294814209881584, + "grad_norm": 2.1530354022979736, + "learning_rate": 7.42052610058607e-06, + "loss": 2.442657470703125, + "step": 231240 + }, + { + "epoch": 0.6295358649789029, + "grad_norm": 2.051983118057251, + "learning_rate": 7.419435736677116e-06, + "loss": 2.4418523788452147, + "step": 231260 + }, + { + "epoch": 0.6295903089696475, + "grad_norm": 2.084463596343994, + "learning_rate": 7.418345372768162e-06, + "loss": 2.445748138427734, + "step": 231280 + }, + { + "epoch": 0.629644752960392, + "grad_norm": 2.2253081798553467, + "learning_rate": 7.417255008859208e-06, + "loss": 2.4393049240112306, + "step": 231300 + }, + { + "epoch": 0.6296991969511365, + "grad_norm": 2.1866679191589355, + "learning_rate": 7.416164644950252e-06, + "loss": 2.452156639099121, + "step": 231320 + }, + { + "epoch": 0.629753640941881, + "grad_norm": 2.08870530128479, + "learning_rate": 7.415074281041298e-06, + "loss": 2.4513877868652343, + "step": 231340 + }, + { + "epoch": 0.6298080849326255, + "grad_norm": 2.38177490234375, + "learning_rate": 7.4139839171323435e-06, + "loss": 2.4862859725952147, + "step": 231360 + }, + { + "epoch": 0.6298625289233701, + "grad_norm": 2.123295307159424, + "learning_rate": 7.412893553223389e-06, + "loss": 2.4799592971801756, + "step": 231380 + }, + { + "epoch": 0.6299169729141146, + "grad_norm": 2.1823067665100098, + "learning_rate": 7.411803189314433e-06, + "loss": 2.4918306350708006, + "step": 231400 + }, + { + "epoch": 0.6299714169048591, + "grad_norm": 2.209136724472046, + "learning_rate": 7.410712825405479e-06, + "loss": 2.452661895751953, + "step": 231420 + }, + { + "epoch": 0.6300258608956036, + "grad_norm": 3.4855170249938965, + "learning_rate": 7.409622461496525e-06, + "loss": 2.502712059020996, + "step": 231440 + }, + { + "epoch": 0.6300803048863481, + "grad_norm": 2.1061594486236572, + "learning_rate": 7.408532097587571e-06, + "loss": 2.491042137145996, + "step": 231460 + }, + { + "epoch": 0.6301347488770926, + "grad_norm": 2.2100512981414795, + "learning_rate": 7.407441733678615e-06, + "loss": 2.455971336364746, + "step": 231480 + }, + { + "epoch": 0.6301891928678373, + "grad_norm": 2.2823269367218018, + "learning_rate": 7.406351369769661e-06, + "loss": 2.4571401596069338, + "step": 231500 + }, + { + "epoch": 0.6302436368585818, + "grad_norm": 2.2535359859466553, + "learning_rate": 7.4052610058607066e-06, + "loss": 2.448294830322266, + "step": 231520 + }, + { + "epoch": 0.6302980808493263, + "grad_norm": 2.2111058235168457, + "learning_rate": 7.404170641951752e-06, + "loss": 2.4608154296875, + "step": 231540 + }, + { + "epoch": 0.6303525248400708, + "grad_norm": 2.2171695232391357, + "learning_rate": 7.403080278042798e-06, + "loss": 2.443109130859375, + "step": 231560 + }, + { + "epoch": 0.6304069688308153, + "grad_norm": 2.0813541412353516, + "learning_rate": 7.401989914133842e-06, + "loss": 2.4869224548339846, + "step": 231580 + }, + { + "epoch": 0.6304614128215599, + "grad_norm": 2.1122798919677734, + "learning_rate": 7.400899550224888e-06, + "loss": 2.458930015563965, + "step": 231600 + }, + { + "epoch": 0.6305158568123044, + "grad_norm": 2.0970418453216553, + "learning_rate": 7.399809186315934e-06, + "loss": 2.49575252532959, + "step": 231620 + }, + { + "epoch": 0.6305703008030489, + "grad_norm": 2.2496330738067627, + "learning_rate": 7.39871882240698e-06, + "loss": 2.490610694885254, + "step": 231640 + }, + { + "epoch": 0.6306247447937934, + "grad_norm": 2.3544652462005615, + "learning_rate": 7.397628458498024e-06, + "loss": 2.4764373779296873, + "step": 231660 + }, + { + "epoch": 0.6306791887845379, + "grad_norm": 2.1299309730529785, + "learning_rate": 7.39653809458907e-06, + "loss": 2.4713369369506837, + "step": 231680 + }, + { + "epoch": 0.6307336327752824, + "grad_norm": 2.2889723777770996, + "learning_rate": 7.3954477306801155e-06, + "loss": 2.4823610305786135, + "step": 231700 + }, + { + "epoch": 0.630788076766027, + "grad_norm": 2.2659778594970703, + "learning_rate": 7.394357366771161e-06, + "loss": 2.4513769149780273, + "step": 231720 + }, + { + "epoch": 0.6308425207567715, + "grad_norm": 2.209625482559204, + "learning_rate": 7.393267002862205e-06, + "loss": 2.4779170989990233, + "step": 231740 + }, + { + "epoch": 0.630896964747516, + "grad_norm": 2.1283316612243652, + "learning_rate": 7.392176638953251e-06, + "loss": 2.456756591796875, + "step": 231760 + }, + { + "epoch": 0.6309514087382605, + "grad_norm": 2.2515852451324463, + "learning_rate": 7.391086275044297e-06, + "loss": 2.502766227722168, + "step": 231780 + }, + { + "epoch": 0.631005852729005, + "grad_norm": 2.1546382904052734, + "learning_rate": 7.389995911135343e-06, + "loss": 2.4528322219848633, + "step": 231800 + }, + { + "epoch": 0.6310602967197495, + "grad_norm": 2.1267006397247314, + "learning_rate": 7.388905547226387e-06, + "loss": 2.4606704711914062, + "step": 231820 + }, + { + "epoch": 0.6311147407104941, + "grad_norm": 2.1464104652404785, + "learning_rate": 7.387815183317433e-06, + "loss": 2.4937000274658203, + "step": 231840 + }, + { + "epoch": 0.6311691847012386, + "grad_norm": 2.199707269668579, + "learning_rate": 7.3867248194084785e-06, + "loss": 2.4895418167114256, + "step": 231860 + }, + { + "epoch": 0.6312236286919831, + "grad_norm": 2.083197593688965, + "learning_rate": 7.385634455499524e-06, + "loss": 2.4618820190429687, + "step": 231880 + }, + { + "epoch": 0.6312780726827276, + "grad_norm": 2.1113080978393555, + "learning_rate": 7.3845440915905685e-06, + "loss": 2.4514276504516603, + "step": 231900 + }, + { + "epoch": 0.6313325166734721, + "grad_norm": 2.139272451400757, + "learning_rate": 7.383453727681614e-06, + "loss": 2.4713472366333007, + "step": 231920 + }, + { + "epoch": 0.6313869606642167, + "grad_norm": 2.0411274433135986, + "learning_rate": 7.38236336377266e-06, + "loss": 2.4769765853881838, + "step": 231940 + }, + { + "epoch": 0.6314414046549612, + "grad_norm": 2.4527902603149414, + "learning_rate": 7.381272999863706e-06, + "loss": 2.4682979583740234, + "step": 231960 + }, + { + "epoch": 0.6314958486457057, + "grad_norm": 2.0689706802368164, + "learning_rate": 7.38018263595475e-06, + "loss": 2.46240291595459, + "step": 231980 + }, + { + "epoch": 0.6315502926364502, + "grad_norm": 2.131676197052002, + "learning_rate": 7.379092272045796e-06, + "loss": 2.486629867553711, + "step": 232000 + }, + { + "epoch": 0.6316047366271947, + "grad_norm": 2.160598039627075, + "learning_rate": 7.378001908136842e-06, + "loss": 2.458632469177246, + "step": 232020 + }, + { + "epoch": 0.6316591806179392, + "grad_norm": 2.0861284732818604, + "learning_rate": 7.3769115442278874e-06, + "loss": 2.4488807678222657, + "step": 232040 + }, + { + "epoch": 0.6317136246086839, + "grad_norm": 2.3050520420074463, + "learning_rate": 7.3758211803189315e-06, + "loss": 2.477975082397461, + "step": 232060 + }, + { + "epoch": 0.6317680685994284, + "grad_norm": 2.4183318614959717, + "learning_rate": 7.374730816409977e-06, + "loss": 2.4648632049560546, + "step": 232080 + }, + { + "epoch": 0.6318225125901729, + "grad_norm": 2.258422613143921, + "learning_rate": 7.373640452501023e-06, + "loss": 2.5102014541625977, + "step": 232100 + }, + { + "epoch": 0.6318769565809174, + "grad_norm": 2.181696891784668, + "learning_rate": 7.372550088592069e-06, + "loss": 2.4962568283081055, + "step": 232120 + }, + { + "epoch": 0.6319314005716619, + "grad_norm": 2.29848575592041, + "learning_rate": 7.371459724683113e-06, + "loss": 2.48099365234375, + "step": 232140 + }, + { + "epoch": 0.6319858445624065, + "grad_norm": 2.1758036613464355, + "learning_rate": 7.370369360774159e-06, + "loss": 2.438762664794922, + "step": 232160 + }, + { + "epoch": 0.632040288553151, + "grad_norm": 2.1411120891571045, + "learning_rate": 7.369278996865205e-06, + "loss": 2.480533981323242, + "step": 232180 + }, + { + "epoch": 0.6320947325438955, + "grad_norm": 2.218579053878784, + "learning_rate": 7.3681886329562505e-06, + "loss": 2.456968879699707, + "step": 232200 + }, + { + "epoch": 0.63214917653464, + "grad_norm": 2.186202049255371, + "learning_rate": 7.367098269047295e-06, + "loss": 2.4741889953613283, + "step": 232220 + }, + { + "epoch": 0.6322036205253845, + "grad_norm": 2.6079463958740234, + "learning_rate": 7.3660079051383404e-06, + "loss": 2.4685617446899415, + "step": 232240 + }, + { + "epoch": 0.632258064516129, + "grad_norm": 2.2829103469848633, + "learning_rate": 7.364917541229386e-06, + "loss": 2.4609209060668946, + "step": 232260 + }, + { + "epoch": 0.6323125085068736, + "grad_norm": 2.137511730194092, + "learning_rate": 7.363827177320432e-06, + "loss": 2.4456718444824217, + "step": 232280 + }, + { + "epoch": 0.6323669524976181, + "grad_norm": 2.324251174926758, + "learning_rate": 7.362736813411476e-06, + "loss": 2.4668224334716795, + "step": 232300 + }, + { + "epoch": 0.6324213964883626, + "grad_norm": 2.1678075790405273, + "learning_rate": 7.361646449502522e-06, + "loss": 2.4730016708374025, + "step": 232320 + }, + { + "epoch": 0.6324758404791071, + "grad_norm": 2.1764135360717773, + "learning_rate": 7.360556085593568e-06, + "loss": 2.4851234436035154, + "step": 232340 + }, + { + "epoch": 0.6325302844698516, + "grad_norm": 2.2466487884521484, + "learning_rate": 7.359465721684614e-06, + "loss": 2.4631231307983397, + "step": 232360 + }, + { + "epoch": 0.6325847284605962, + "grad_norm": 2.0773913860321045, + "learning_rate": 7.358375357775658e-06, + "loss": 2.460976982116699, + "step": 232380 + }, + { + "epoch": 0.6326391724513407, + "grad_norm": 2.1430106163024902, + "learning_rate": 7.3572849938667035e-06, + "loss": 2.4771310806274416, + "step": 232400 + }, + { + "epoch": 0.6326936164420852, + "grad_norm": 2.0805726051330566, + "learning_rate": 7.356194629957749e-06, + "loss": 2.4583600997924804, + "step": 232420 + }, + { + "epoch": 0.6327480604328297, + "grad_norm": 2.1463093757629395, + "learning_rate": 7.355104266048795e-06, + "loss": 2.4632118225097654, + "step": 232440 + }, + { + "epoch": 0.6328025044235742, + "grad_norm": 2.1417737007141113, + "learning_rate": 7.354013902139839e-06, + "loss": 2.473201560974121, + "step": 232460 + }, + { + "epoch": 0.6328569484143187, + "grad_norm": 2.1120824813842773, + "learning_rate": 7.352923538230885e-06, + "loss": 2.46513671875, + "step": 232480 + }, + { + "epoch": 0.6329113924050633, + "grad_norm": 2.3791260719299316, + "learning_rate": 7.351833174321931e-06, + "loss": 2.4609451293945312, + "step": 232500 + }, + { + "epoch": 0.6329658363958078, + "grad_norm": 2.3060550689697266, + "learning_rate": 7.350742810412977e-06, + "loss": 2.4775060653686523, + "step": 232520 + }, + { + "epoch": 0.6330202803865523, + "grad_norm": 2.2641844749450684, + "learning_rate": 7.349652446504021e-06, + "loss": 2.4859554290771486, + "step": 232540 + }, + { + "epoch": 0.6330747243772968, + "grad_norm": 2.1515142917633057, + "learning_rate": 7.348562082595067e-06, + "loss": 2.4950923919677734, + "step": 232560 + }, + { + "epoch": 0.6331291683680413, + "grad_norm": 2.120300531387329, + "learning_rate": 7.347471718686112e-06, + "loss": 2.464063262939453, + "step": 232580 + }, + { + "epoch": 0.633183612358786, + "grad_norm": 2.1926372051239014, + "learning_rate": 7.346381354777158e-06, + "loss": 2.467543601989746, + "step": 232600 + }, + { + "epoch": 0.6332380563495305, + "grad_norm": 2.1861910820007324, + "learning_rate": 7.345290990868202e-06, + "loss": 2.460537910461426, + "step": 232620 + }, + { + "epoch": 0.633292500340275, + "grad_norm": 2.151179790496826, + "learning_rate": 7.344200626959248e-06, + "loss": 2.4427864074707033, + "step": 232640 + }, + { + "epoch": 0.6333469443310195, + "grad_norm": 2.18312668800354, + "learning_rate": 7.343110263050294e-06, + "loss": 2.4769271850585937, + "step": 232660 + }, + { + "epoch": 0.633401388321764, + "grad_norm": 2.1490063667297363, + "learning_rate": 7.34201989914134e-06, + "loss": 2.4644903182983398, + "step": 232680 + }, + { + "epoch": 0.6334558323125085, + "grad_norm": 2.289438247680664, + "learning_rate": 7.340929535232384e-06, + "loss": 2.4571674346923826, + "step": 232700 + }, + { + "epoch": 0.6335102763032531, + "grad_norm": 2.230766773223877, + "learning_rate": 7.33983917132343e-06, + "loss": 2.477680778503418, + "step": 232720 + }, + { + "epoch": 0.6335647202939976, + "grad_norm": 2.279792308807373, + "learning_rate": 7.3387488074144755e-06, + "loss": 2.4939403533935547, + "step": 232740 + }, + { + "epoch": 0.6336191642847421, + "grad_norm": 2.180769681930542, + "learning_rate": 7.337658443505521e-06, + "loss": 2.483569526672363, + "step": 232760 + }, + { + "epoch": 0.6336736082754866, + "grad_norm": 2.2728543281555176, + "learning_rate": 7.3365680795965654e-06, + "loss": 2.4658493041992187, + "step": 232780 + }, + { + "epoch": 0.6337280522662311, + "grad_norm": 2.134606122970581, + "learning_rate": 7.335477715687611e-06, + "loss": 2.4618814468383787, + "step": 232800 + }, + { + "epoch": 0.6337824962569757, + "grad_norm": 2.1301238536834717, + "learning_rate": 7.334387351778657e-06, + "loss": 2.4992979049682615, + "step": 232820 + }, + { + "epoch": 0.6338369402477202, + "grad_norm": 2.1904919147491455, + "learning_rate": 7.333296987869703e-06, + "loss": 2.4401113510131838, + "step": 232840 + }, + { + "epoch": 0.6338913842384647, + "grad_norm": 2.4123282432556152, + "learning_rate": 7.332206623960747e-06, + "loss": 2.463507080078125, + "step": 232860 + }, + { + "epoch": 0.6339458282292092, + "grad_norm": 2.1634016036987305, + "learning_rate": 7.331116260051793e-06, + "loss": 2.472990608215332, + "step": 232880 + }, + { + "epoch": 0.6340002722199537, + "grad_norm": 2.139641284942627, + "learning_rate": 7.330025896142839e-06, + "loss": 2.4784591674804686, + "step": 232900 + }, + { + "epoch": 0.6340547162106982, + "grad_norm": 2.7402238845825195, + "learning_rate": 7.328935532233884e-06, + "loss": 2.4761877059936523, + "step": 232920 + }, + { + "epoch": 0.6341091602014428, + "grad_norm": 2.249288320541382, + "learning_rate": 7.3278451683249285e-06, + "loss": 2.4584529876708983, + "step": 232940 + }, + { + "epoch": 0.6341636041921873, + "grad_norm": 2.1603095531463623, + "learning_rate": 7.326754804415974e-06, + "loss": 2.463863182067871, + "step": 232960 + }, + { + "epoch": 0.6342180481829318, + "grad_norm": 2.283684015274048, + "learning_rate": 7.32566444050702e-06, + "loss": 2.437420463562012, + "step": 232980 + }, + { + "epoch": 0.6342724921736763, + "grad_norm": 2.2497267723083496, + "learning_rate": 7.324574076598065e-06, + "loss": 2.4786224365234375, + "step": 233000 + }, + { + "epoch": 0.6343269361644208, + "grad_norm": 2.2491402626037598, + "learning_rate": 7.32348371268911e-06, + "loss": 2.4877029418945313, + "step": 233020 + }, + { + "epoch": 0.6343813801551653, + "grad_norm": 2.2543880939483643, + "learning_rate": 7.322393348780156e-06, + "loss": 2.4886383056640624, + "step": 233040 + }, + { + "epoch": 0.6344358241459099, + "grad_norm": 2.1362853050231934, + "learning_rate": 7.321302984871202e-06, + "loss": 2.4564630508422853, + "step": 233060 + }, + { + "epoch": 0.6344902681366544, + "grad_norm": 2.1237518787384033, + "learning_rate": 7.320212620962247e-06, + "loss": 2.464190864562988, + "step": 233080 + }, + { + "epoch": 0.6345447121273989, + "grad_norm": 2.2105393409729004, + "learning_rate": 7.319122257053292e-06, + "loss": 2.480572509765625, + "step": 233100 + }, + { + "epoch": 0.6345991561181434, + "grad_norm": 2.1662328243255615, + "learning_rate": 7.318031893144337e-06, + "loss": 2.459813117980957, + "step": 233120 + }, + { + "epoch": 0.6346536001088879, + "grad_norm": 2.3763129711151123, + "learning_rate": 7.316941529235383e-06, + "loss": 2.4937252044677733, + "step": 233140 + }, + { + "epoch": 0.6347080440996326, + "grad_norm": 2.137587785720825, + "learning_rate": 7.315851165326428e-06, + "loss": 2.472374153137207, + "step": 233160 + }, + { + "epoch": 0.634762488090377, + "grad_norm": 6.41962194442749, + "learning_rate": 7.314760801417473e-06, + "loss": 2.4629226684570313, + "step": 233180 + }, + { + "epoch": 0.6348169320811216, + "grad_norm": 2.0628485679626465, + "learning_rate": 7.313670437508519e-06, + "loss": 2.4732328414916993, + "step": 233200 + }, + { + "epoch": 0.6348713760718661, + "grad_norm": 2.254298210144043, + "learning_rate": 7.312580073599565e-06, + "loss": 2.502476119995117, + "step": 233220 + }, + { + "epoch": 0.6349258200626106, + "grad_norm": 2.144526243209839, + "learning_rate": 7.31148970969061e-06, + "loss": 2.480246162414551, + "step": 233240 + }, + { + "epoch": 0.6349802640533551, + "grad_norm": 2.1903133392333984, + "learning_rate": 7.3103993457816555e-06, + "loss": 2.471338081359863, + "step": 233260 + }, + { + "epoch": 0.6350347080440997, + "grad_norm": 2.2878479957580566, + "learning_rate": 7.3093089818727005e-06, + "loss": 2.4763755798339844, + "step": 233280 + }, + { + "epoch": 0.6350891520348442, + "grad_norm": 2.1923274993896484, + "learning_rate": 7.308218617963746e-06, + "loss": 2.4602283477783202, + "step": 233300 + }, + { + "epoch": 0.6351435960255887, + "grad_norm": 2.2569751739501953, + "learning_rate": 7.307128254054791e-06, + "loss": 2.456675338745117, + "step": 233320 + }, + { + "epoch": 0.6351980400163332, + "grad_norm": 2.330660343170166, + "learning_rate": 7.306037890145837e-06, + "loss": 2.475747489929199, + "step": 233340 + }, + { + "epoch": 0.6352524840070777, + "grad_norm": 2.2442538738250732, + "learning_rate": 7.304947526236882e-06, + "loss": 2.461071014404297, + "step": 233360 + }, + { + "epoch": 0.6353069279978223, + "grad_norm": 2.187638521194458, + "learning_rate": 7.303857162327928e-06, + "loss": 2.4490789413452148, + "step": 233380 + }, + { + "epoch": 0.6353613719885668, + "grad_norm": 2.157493829727173, + "learning_rate": 7.302766798418973e-06, + "loss": 2.4806921005249025, + "step": 233400 + }, + { + "epoch": 0.6354158159793113, + "grad_norm": 2.138662099838257, + "learning_rate": 7.301676434510019e-06, + "loss": 2.4834213256835938, + "step": 233420 + }, + { + "epoch": 0.6354702599700558, + "grad_norm": 2.223679542541504, + "learning_rate": 7.300586070601064e-06, + "loss": 2.474139595031738, + "step": 233440 + }, + { + "epoch": 0.6355247039608003, + "grad_norm": 2.2618179321289062, + "learning_rate": 7.299495706692109e-06, + "loss": 2.5099021911621096, + "step": 233460 + }, + { + "epoch": 0.6355791479515448, + "grad_norm": 2.2132062911987305, + "learning_rate": 7.298405342783154e-06, + "loss": 2.459452819824219, + "step": 233480 + }, + { + "epoch": 0.6356335919422894, + "grad_norm": 2.1486947536468506, + "learning_rate": 7.2973149788742e-06, + "loss": 2.499319839477539, + "step": 233500 + }, + { + "epoch": 0.6356880359330339, + "grad_norm": 2.1858713626861572, + "learning_rate": 7.296224614965245e-06, + "loss": 2.487554931640625, + "step": 233520 + }, + { + "epoch": 0.6357424799237784, + "grad_norm": 2.2641403675079346, + "learning_rate": 7.295134251056291e-06, + "loss": 2.471929359436035, + "step": 233540 + }, + { + "epoch": 0.6357969239145229, + "grad_norm": 2.397585868835449, + "learning_rate": 7.294043887147336e-06, + "loss": 2.457940673828125, + "step": 233560 + }, + { + "epoch": 0.6358513679052674, + "grad_norm": 2.3013851642608643, + "learning_rate": 7.292953523238382e-06, + "loss": 2.4765041351318358, + "step": 233580 + }, + { + "epoch": 0.635905811896012, + "grad_norm": 4.435417652130127, + "learning_rate": 7.291863159329427e-06, + "loss": 2.460725212097168, + "step": 233600 + }, + { + "epoch": 0.6359602558867565, + "grad_norm": 2.2702441215515137, + "learning_rate": 7.2907727954204725e-06, + "loss": 2.474605369567871, + "step": 233620 + }, + { + "epoch": 0.636014699877501, + "grad_norm": 2.1623854637145996, + "learning_rate": 7.2896824315115174e-06, + "loss": 2.5135129928588866, + "step": 233640 + }, + { + "epoch": 0.6360691438682455, + "grad_norm": 2.189439296722412, + "learning_rate": 7.288592067602563e-06, + "loss": 2.4941301345825195, + "step": 233660 + }, + { + "epoch": 0.63612358785899, + "grad_norm": 2.2011852264404297, + "learning_rate": 7.287501703693608e-06, + "loss": 2.455637741088867, + "step": 233680 + }, + { + "epoch": 0.6361780318497345, + "grad_norm": 2.4481499195098877, + "learning_rate": 7.286411339784654e-06, + "loss": 2.450983238220215, + "step": 233700 + }, + { + "epoch": 0.6362324758404792, + "grad_norm": 2.313387155532837, + "learning_rate": 7.285320975875699e-06, + "loss": 2.4252058029174806, + "step": 233720 + }, + { + "epoch": 0.6362869198312237, + "grad_norm": 2.204374313354492, + "learning_rate": 7.284230611966745e-06, + "loss": 2.4730113983154296, + "step": 233740 + }, + { + "epoch": 0.6363413638219682, + "grad_norm": 2.189502716064453, + "learning_rate": 7.28314024805779e-06, + "loss": 2.488167572021484, + "step": 233760 + }, + { + "epoch": 0.6363958078127127, + "grad_norm": 2.189424753189087, + "learning_rate": 7.2820498841488356e-06, + "loss": 2.4765758514404297, + "step": 233780 + }, + { + "epoch": 0.6364502518034572, + "grad_norm": 2.27142333984375, + "learning_rate": 7.2809595202398805e-06, + "loss": 2.463631439208984, + "step": 233800 + }, + { + "epoch": 0.6365046957942018, + "grad_norm": 2.3439066410064697, + "learning_rate": 7.279869156330926e-06, + "loss": 2.428102493286133, + "step": 233820 + }, + { + "epoch": 0.6365591397849463, + "grad_norm": 2.114086151123047, + "learning_rate": 7.278778792421971e-06, + "loss": 2.4717199325561525, + "step": 233840 + }, + { + "epoch": 0.6366135837756908, + "grad_norm": 2.0379526615142822, + "learning_rate": 7.277688428513017e-06, + "loss": 2.4619760513305664, + "step": 233860 + }, + { + "epoch": 0.6366680277664353, + "grad_norm": 2.187589406967163, + "learning_rate": 7.276598064604062e-06, + "loss": 2.4624202728271483, + "step": 233880 + }, + { + "epoch": 0.6367224717571798, + "grad_norm": 2.2348577976226807, + "learning_rate": 7.275507700695108e-06, + "loss": 2.4458173751831054, + "step": 233900 + }, + { + "epoch": 0.6367769157479243, + "grad_norm": 2.291635513305664, + "learning_rate": 7.274417336786153e-06, + "loss": 2.4960359573364257, + "step": 233920 + }, + { + "epoch": 0.6368313597386689, + "grad_norm": 2.1364991664886475, + "learning_rate": 7.273326972877199e-06, + "loss": 2.4523138046264648, + "step": 233940 + }, + { + "epoch": 0.6368858037294134, + "grad_norm": 3.0857536792755127, + "learning_rate": 7.272236608968244e-06, + "loss": 2.4914337158203126, + "step": 233960 + }, + { + "epoch": 0.6369402477201579, + "grad_norm": 2.1676275730133057, + "learning_rate": 7.271146245059289e-06, + "loss": 2.5120780944824217, + "step": 233980 + }, + { + "epoch": 0.6369946917109024, + "grad_norm": 2.1631152629852295, + "learning_rate": 7.270055881150334e-06, + "loss": 2.496783447265625, + "step": 234000 + }, + { + "epoch": 0.6370491357016469, + "grad_norm": 2.250565528869629, + "learning_rate": 7.26896551724138e-06, + "loss": 2.510526657104492, + "step": 234020 + }, + { + "epoch": 0.6371035796923914, + "grad_norm": 2.1574766635894775, + "learning_rate": 7.267875153332425e-06, + "loss": 2.4788875579833984, + "step": 234040 + }, + { + "epoch": 0.637158023683136, + "grad_norm": 2.130068302154541, + "learning_rate": 7.266784789423471e-06, + "loss": 2.480763244628906, + "step": 234060 + }, + { + "epoch": 0.6372124676738805, + "grad_norm": 2.412374496459961, + "learning_rate": 7.265694425514516e-06, + "loss": 2.46658992767334, + "step": 234080 + }, + { + "epoch": 0.637266911664625, + "grad_norm": 2.1759417057037354, + "learning_rate": 7.264604061605561e-06, + "loss": 2.4586448669433594, + "step": 234100 + }, + { + "epoch": 0.6373213556553695, + "grad_norm": 2.062390089035034, + "learning_rate": 7.263513697696607e-06, + "loss": 2.4298080444335937, + "step": 234120 + }, + { + "epoch": 0.637375799646114, + "grad_norm": 2.2269277572631836, + "learning_rate": 7.2624233337876525e-06, + "loss": 2.4757667541503907, + "step": 234140 + }, + { + "epoch": 0.6374302436368586, + "grad_norm": 2.157503843307495, + "learning_rate": 7.2613329698786975e-06, + "loss": 2.4815799713134767, + "step": 234160 + }, + { + "epoch": 0.6374846876276031, + "grad_norm": 2.1344923973083496, + "learning_rate": 7.2602426059697424e-06, + "loss": 2.439779853820801, + "step": 234180 + }, + { + "epoch": 0.6375391316183476, + "grad_norm": 2.1242682933807373, + "learning_rate": 7.259152242060788e-06, + "loss": 2.501336860656738, + "step": 234200 + }, + { + "epoch": 0.6375935756090921, + "grad_norm": 2.2725136280059814, + "learning_rate": 7.258061878151834e-06, + "loss": 2.458746910095215, + "step": 234220 + }, + { + "epoch": 0.6376480195998366, + "grad_norm": 3.372762680053711, + "learning_rate": 7.256971514242879e-06, + "loss": 2.4726804733276366, + "step": 234240 + }, + { + "epoch": 0.6377024635905811, + "grad_norm": 2.204782009124756, + "learning_rate": 7.255881150333924e-06, + "loss": 2.4757293701171874, + "step": 234260 + }, + { + "epoch": 0.6377569075813257, + "grad_norm": 2.313370704650879, + "learning_rate": 7.25479078642497e-06, + "loss": 2.4608938217163088, + "step": 234280 + }, + { + "epoch": 0.6378113515720703, + "grad_norm": 2.172058343887329, + "learning_rate": 7.253700422516016e-06, + "loss": 2.4650314331054686, + "step": 234300 + }, + { + "epoch": 0.6378657955628148, + "grad_norm": 2.252436399459839, + "learning_rate": 7.2526100586070606e-06, + "loss": 2.46359920501709, + "step": 234320 + }, + { + "epoch": 0.6379202395535593, + "grad_norm": 2.156148910522461, + "learning_rate": 7.2515196946981055e-06, + "loss": 2.4631834030151367, + "step": 234340 + }, + { + "epoch": 0.6379746835443038, + "grad_norm": 2.1517364978790283, + "learning_rate": 7.250429330789151e-06, + "loss": 2.514124298095703, + "step": 234360 + }, + { + "epoch": 0.6380291275350484, + "grad_norm": 2.127488374710083, + "learning_rate": 7.249338966880197e-06, + "loss": 2.4855451583862305, + "step": 234380 + }, + { + "epoch": 0.6380835715257929, + "grad_norm": 2.167083978652954, + "learning_rate": 7.248248602971242e-06, + "loss": 2.466341018676758, + "step": 234400 + }, + { + "epoch": 0.6381380155165374, + "grad_norm": 2.173137903213501, + "learning_rate": 7.247158239062287e-06, + "loss": 2.4526905059814452, + "step": 234420 + }, + { + "epoch": 0.6381924595072819, + "grad_norm": 3.1352453231811523, + "learning_rate": 7.246067875153333e-06, + "loss": 2.4647504806518556, + "step": 234440 + }, + { + "epoch": 0.6382469034980264, + "grad_norm": 2.1943001747131348, + "learning_rate": 7.244977511244379e-06, + "loss": 2.4818742752075194, + "step": 234460 + }, + { + "epoch": 0.6383013474887709, + "grad_norm": 2.2232699394226074, + "learning_rate": 7.243887147335424e-06, + "loss": 2.4582660675048826, + "step": 234480 + }, + { + "epoch": 0.6383557914795155, + "grad_norm": 2.66072154045105, + "learning_rate": 7.242796783426469e-06, + "loss": 2.4655630111694338, + "step": 234500 + }, + { + "epoch": 0.63841023547026, + "grad_norm": 2.0377187728881836, + "learning_rate": 7.241706419517514e-06, + "loss": 2.480280303955078, + "step": 234520 + }, + { + "epoch": 0.6384646794610045, + "grad_norm": 2.164726972579956, + "learning_rate": 7.24061605560856e-06, + "loss": 2.4716997146606445, + "step": 234540 + }, + { + "epoch": 0.638519123451749, + "grad_norm": 2.072502374649048, + "learning_rate": 7.239525691699605e-06, + "loss": 2.4454582214355467, + "step": 234560 + }, + { + "epoch": 0.6385735674424935, + "grad_norm": 2.0833752155303955, + "learning_rate": 7.23843532779065e-06, + "loss": 2.4805179595947267, + "step": 234580 + }, + { + "epoch": 0.6386280114332381, + "grad_norm": 2.204651355743408, + "learning_rate": 7.237344963881696e-06, + "loss": 2.4674530029296875, + "step": 234600 + }, + { + "epoch": 0.6386824554239826, + "grad_norm": 2.6012964248657227, + "learning_rate": 7.236254599972742e-06, + "loss": 2.4852039337158205, + "step": 234620 + }, + { + "epoch": 0.6387368994147271, + "grad_norm": 2.202223300933838, + "learning_rate": 7.235164236063787e-06, + "loss": 2.454402542114258, + "step": 234640 + }, + { + "epoch": 0.6387913434054716, + "grad_norm": 2.1002771854400635, + "learning_rate": 7.234073872154832e-06, + "loss": 2.4402450561523437, + "step": 234660 + }, + { + "epoch": 0.6388457873962161, + "grad_norm": 2.1316893100738525, + "learning_rate": 7.2329835082458775e-06, + "loss": 2.454781150817871, + "step": 234680 + }, + { + "epoch": 0.6389002313869606, + "grad_norm": 2.0586376190185547, + "learning_rate": 7.231893144336923e-06, + "loss": 2.472788619995117, + "step": 234700 + }, + { + "epoch": 0.6389546753777052, + "grad_norm": 2.264098644256592, + "learning_rate": 7.230802780427968e-06, + "loss": 2.447805404663086, + "step": 234720 + }, + { + "epoch": 0.6390091193684497, + "grad_norm": 2.0659101009368896, + "learning_rate": 7.229712416519013e-06, + "loss": 2.466122055053711, + "step": 234740 + }, + { + "epoch": 0.6390635633591942, + "grad_norm": 2.27833890914917, + "learning_rate": 7.228622052610059e-06, + "loss": 2.475859451293945, + "step": 234760 + }, + { + "epoch": 0.6391180073499387, + "grad_norm": 2.6599721908569336, + "learning_rate": 7.227531688701105e-06, + "loss": 2.4766279220581056, + "step": 234780 + }, + { + "epoch": 0.6391724513406832, + "grad_norm": 2.2762763500213623, + "learning_rate": 7.22644132479215e-06, + "loss": 2.472251129150391, + "step": 234800 + }, + { + "epoch": 0.6392268953314278, + "grad_norm": 2.2160515785217285, + "learning_rate": 7.225350960883195e-06, + "loss": 2.498038101196289, + "step": 234820 + }, + { + "epoch": 0.6392813393221723, + "grad_norm": 2.245351791381836, + "learning_rate": 7.224260596974241e-06, + "loss": 2.4886920928955076, + "step": 234840 + }, + { + "epoch": 0.6393357833129168, + "grad_norm": 2.103966474533081, + "learning_rate": 7.223170233065286e-06, + "loss": 2.4905792236328126, + "step": 234860 + }, + { + "epoch": 0.6393902273036614, + "grad_norm": 2.2602312564849854, + "learning_rate": 7.222079869156331e-06, + "loss": 2.4959487915039062, + "step": 234880 + }, + { + "epoch": 0.6394446712944059, + "grad_norm": 2.199260711669922, + "learning_rate": 7.220989505247376e-06, + "loss": 2.4776411056518555, + "step": 234900 + }, + { + "epoch": 0.6394991152851504, + "grad_norm": 2.222496271133423, + "learning_rate": 7.219899141338422e-06, + "loss": 2.4433572769165037, + "step": 234920 + }, + { + "epoch": 0.639553559275895, + "grad_norm": 2.7973225116729736, + "learning_rate": 7.218808777429468e-06, + "loss": 2.491324043273926, + "step": 234940 + }, + { + "epoch": 0.6396080032666395, + "grad_norm": 2.34419322013855, + "learning_rate": 7.217718413520514e-06, + "loss": 2.452411651611328, + "step": 234960 + }, + { + "epoch": 0.639662447257384, + "grad_norm": 2.1269445419311523, + "learning_rate": 7.216628049611558e-06, + "loss": 2.4537994384765627, + "step": 234980 + }, + { + "epoch": 0.6397168912481285, + "grad_norm": 2.306279182434082, + "learning_rate": 7.215537685702604e-06, + "loss": 2.492029571533203, + "step": 235000 + }, + { + "epoch": 0.639771335238873, + "grad_norm": 2.0638113021850586, + "learning_rate": 7.2144473217936495e-06, + "loss": 2.492141914367676, + "step": 235020 + }, + { + "epoch": 0.6398257792296176, + "grad_norm": 2.2949795722961426, + "learning_rate": 7.213356957884695e-06, + "loss": 2.4928104400634767, + "step": 235040 + }, + { + "epoch": 0.6398802232203621, + "grad_norm": 3.6638705730438232, + "learning_rate": 7.212266593975739e-06, + "loss": 2.476505470275879, + "step": 235060 + }, + { + "epoch": 0.6399346672111066, + "grad_norm": 2.343036651611328, + "learning_rate": 7.211176230066785e-06, + "loss": 2.4573953628540037, + "step": 235080 + }, + { + "epoch": 0.6399891112018511, + "grad_norm": 2.105103015899658, + "learning_rate": 7.210085866157831e-06, + "loss": 2.4923370361328123, + "step": 235100 + }, + { + "epoch": 0.6400435551925956, + "grad_norm": 2.1941208839416504, + "learning_rate": 7.208995502248877e-06, + "loss": 2.502205467224121, + "step": 235120 + }, + { + "epoch": 0.6400979991833401, + "grad_norm": 2.196589946746826, + "learning_rate": 7.207905138339921e-06, + "loss": 2.462814712524414, + "step": 235140 + }, + { + "epoch": 0.6401524431740847, + "grad_norm": 2.1649985313415527, + "learning_rate": 7.206814774430967e-06, + "loss": 2.4659635543823244, + "step": 235160 + }, + { + "epoch": 0.6402068871648292, + "grad_norm": 2.1761250495910645, + "learning_rate": 7.2057244105220126e-06, + "loss": 2.4586551666259764, + "step": 235180 + }, + { + "epoch": 0.6402613311555737, + "grad_norm": 2.14766001701355, + "learning_rate": 7.204634046613058e-06, + "loss": 2.500840950012207, + "step": 235200 + }, + { + "epoch": 0.6403157751463182, + "grad_norm": 2.1370060443878174, + "learning_rate": 7.2035436827041025e-06, + "loss": 2.4794422149658204, + "step": 235220 + }, + { + "epoch": 0.6403702191370627, + "grad_norm": 2.119729995727539, + "learning_rate": 7.202453318795148e-06, + "loss": 2.483206367492676, + "step": 235240 + }, + { + "epoch": 0.6404246631278072, + "grad_norm": 2.1082417964935303, + "learning_rate": 7.201362954886194e-06, + "loss": 2.476937675476074, + "step": 235260 + }, + { + "epoch": 0.6404791071185518, + "grad_norm": 2.1442670822143555, + "learning_rate": 7.20027259097724e-06, + "loss": 2.495026779174805, + "step": 235280 + }, + { + "epoch": 0.6405335511092963, + "grad_norm": 2.172773838043213, + "learning_rate": 7.199182227068284e-06, + "loss": 2.472117805480957, + "step": 235300 + }, + { + "epoch": 0.6405879951000408, + "grad_norm": 2.105126142501831, + "learning_rate": 7.19809186315933e-06, + "loss": 2.4453338623046874, + "step": 235320 + }, + { + "epoch": 0.6406424390907853, + "grad_norm": 2.187943458557129, + "learning_rate": 7.197001499250376e-06, + "loss": 2.474718475341797, + "step": 235340 + }, + { + "epoch": 0.6406968830815298, + "grad_norm": 2.5733108520507812, + "learning_rate": 7.1959111353414215e-06, + "loss": 2.4723094940185546, + "step": 235360 + }, + { + "epoch": 0.6407513270722744, + "grad_norm": 2.2735373973846436, + "learning_rate": 7.1948207714324656e-06, + "loss": 2.490440559387207, + "step": 235380 + }, + { + "epoch": 0.640805771063019, + "grad_norm": 2.23471999168396, + "learning_rate": 7.193730407523511e-06, + "loss": 2.4529075622558594, + "step": 235400 + }, + { + "epoch": 0.6408602150537634, + "grad_norm": 2.124643564224243, + "learning_rate": 7.192640043614557e-06, + "loss": 2.4905488967895506, + "step": 235420 + }, + { + "epoch": 0.640914659044508, + "grad_norm": 2.2762300968170166, + "learning_rate": 7.191549679705603e-06, + "loss": 2.4615768432617187, + "step": 235440 + }, + { + "epoch": 0.6409691030352525, + "grad_norm": 2.3073034286499023, + "learning_rate": 7.190459315796647e-06, + "loss": 2.477199935913086, + "step": 235460 + }, + { + "epoch": 0.641023547025997, + "grad_norm": 2.164905071258545, + "learning_rate": 7.189368951887693e-06, + "loss": 2.448549270629883, + "step": 235480 + }, + { + "epoch": 0.6410779910167416, + "grad_norm": 2.254484176635742, + "learning_rate": 7.188278587978739e-06, + "loss": 2.470841979980469, + "step": 235500 + }, + { + "epoch": 0.6411324350074861, + "grad_norm": 2.1875295639038086, + "learning_rate": 7.1871882240697845e-06, + "loss": 2.448562431335449, + "step": 235520 + }, + { + "epoch": 0.6411868789982306, + "grad_norm": 2.190530776977539, + "learning_rate": 7.186097860160829e-06, + "loss": 2.472535514831543, + "step": 235540 + }, + { + "epoch": 0.6412413229889751, + "grad_norm": 2.1364359855651855, + "learning_rate": 7.1850074962518745e-06, + "loss": 2.463412284851074, + "step": 235560 + }, + { + "epoch": 0.6412957669797196, + "grad_norm": 2.418764591217041, + "learning_rate": 7.18391713234292e-06, + "loss": 2.488288688659668, + "step": 235580 + }, + { + "epoch": 0.6413502109704642, + "grad_norm": 2.1508355140686035, + "learning_rate": 7.182826768433966e-06, + "loss": 2.4671346664428713, + "step": 235600 + }, + { + "epoch": 0.6414046549612087, + "grad_norm": 2.34928035736084, + "learning_rate": 7.18173640452501e-06, + "loss": 2.446238327026367, + "step": 235620 + }, + { + "epoch": 0.6414590989519532, + "grad_norm": 2.6567189693450928, + "learning_rate": 7.180646040616056e-06, + "loss": 2.460715675354004, + "step": 235640 + }, + { + "epoch": 0.6415135429426977, + "grad_norm": 2.138920307159424, + "learning_rate": 7.179555676707102e-06, + "loss": 2.4718017578125, + "step": 235660 + }, + { + "epoch": 0.6415679869334422, + "grad_norm": 2.2078421115875244, + "learning_rate": 7.178465312798148e-06, + "loss": 2.461760902404785, + "step": 235680 + }, + { + "epoch": 0.6416224309241867, + "grad_norm": 2.227240562438965, + "learning_rate": 7.177374948889192e-06, + "loss": 2.478034019470215, + "step": 235700 + }, + { + "epoch": 0.6416768749149313, + "grad_norm": 2.6050796508789062, + "learning_rate": 7.1762845849802376e-06, + "loss": 2.5095048904418946, + "step": 235720 + }, + { + "epoch": 0.6417313189056758, + "grad_norm": 2.1285247802734375, + "learning_rate": 7.175194221071283e-06, + "loss": 2.4984739303588865, + "step": 235740 + }, + { + "epoch": 0.6417857628964203, + "grad_norm": 2.1966981887817383, + "learning_rate": 7.174103857162329e-06, + "loss": 2.474544334411621, + "step": 235760 + }, + { + "epoch": 0.6418402068871648, + "grad_norm": 2.172991991043091, + "learning_rate": 7.173013493253373e-06, + "loss": 2.48907527923584, + "step": 235780 + }, + { + "epoch": 0.6418946508779093, + "grad_norm": 2.196575880050659, + "learning_rate": 7.171923129344419e-06, + "loss": 2.4562551498413088, + "step": 235800 + }, + { + "epoch": 0.6419490948686539, + "grad_norm": 2.200812578201294, + "learning_rate": 7.170832765435465e-06, + "loss": 2.4691640853881838, + "step": 235820 + }, + { + "epoch": 0.6420035388593984, + "grad_norm": 2.3104796409606934, + "learning_rate": 7.169742401526511e-06, + "loss": 2.4854351043701173, + "step": 235840 + }, + { + "epoch": 0.6420579828501429, + "grad_norm": 2.1577858924865723, + "learning_rate": 7.168652037617555e-06, + "loss": 2.47582893371582, + "step": 235860 + }, + { + "epoch": 0.6421124268408874, + "grad_norm": 2.246929407119751, + "learning_rate": 7.167561673708601e-06, + "loss": 2.4851909637451173, + "step": 235880 + }, + { + "epoch": 0.6421668708316319, + "grad_norm": 3.4285471439361572, + "learning_rate": 7.1664713097996464e-06, + "loss": 2.4733938217163085, + "step": 235900 + }, + { + "epoch": 0.6422213148223764, + "grad_norm": 2.1904804706573486, + "learning_rate": 7.165380945890692e-06, + "loss": 2.4642587661743165, + "step": 235920 + }, + { + "epoch": 0.642275758813121, + "grad_norm": 2.203850030899048, + "learning_rate": 7.164290581981736e-06, + "loss": 2.4604530334472656, + "step": 235940 + }, + { + "epoch": 0.6423302028038655, + "grad_norm": 2.2250349521636963, + "learning_rate": 7.163200218072782e-06, + "loss": 2.505519676208496, + "step": 235960 + }, + { + "epoch": 0.64238464679461, + "grad_norm": 2.195763111114502, + "learning_rate": 7.162109854163828e-06, + "loss": 2.4895668029785156, + "step": 235980 + }, + { + "epoch": 0.6424390907853545, + "grad_norm": 2.2782766819000244, + "learning_rate": 7.161019490254874e-06, + "loss": 2.4750749588012697, + "step": 236000 + }, + { + "epoch": 0.642493534776099, + "grad_norm": 2.5678904056549072, + "learning_rate": 7.159929126345918e-06, + "loss": 2.495769500732422, + "step": 236020 + }, + { + "epoch": 0.6425479787668437, + "grad_norm": 2.31119441986084, + "learning_rate": 7.158838762436964e-06, + "loss": 2.467201042175293, + "step": 236040 + }, + { + "epoch": 0.6426024227575882, + "grad_norm": 2.0782761573791504, + "learning_rate": 7.1577483985280095e-06, + "loss": 2.475299263000488, + "step": 236060 + }, + { + "epoch": 0.6426568667483327, + "grad_norm": 2.191049098968506, + "learning_rate": 7.156658034619055e-06, + "loss": 2.4787994384765626, + "step": 236080 + }, + { + "epoch": 0.6427113107390772, + "grad_norm": 2.1344711780548096, + "learning_rate": 7.1555676707100995e-06, + "loss": 2.4714637756347657, + "step": 236100 + }, + { + "epoch": 0.6427657547298217, + "grad_norm": 2.2119526863098145, + "learning_rate": 7.154477306801145e-06, + "loss": 2.4750202178955076, + "step": 236120 + }, + { + "epoch": 0.6428201987205662, + "grad_norm": 2.7508461475372314, + "learning_rate": 7.153386942892191e-06, + "loss": 2.4726337432861327, + "step": 236140 + }, + { + "epoch": 0.6428746427113108, + "grad_norm": 2.143155574798584, + "learning_rate": 7.152296578983237e-06, + "loss": 2.4597402572631837, + "step": 236160 + }, + { + "epoch": 0.6429290867020553, + "grad_norm": 2.623805284500122, + "learning_rate": 7.151206215074281e-06, + "loss": 2.460310363769531, + "step": 236180 + }, + { + "epoch": 0.6429835306927998, + "grad_norm": 2.1588621139526367, + "learning_rate": 7.150115851165327e-06, + "loss": 2.4677499771118163, + "step": 236200 + }, + { + "epoch": 0.6430379746835443, + "grad_norm": 2.239147186279297, + "learning_rate": 7.149025487256373e-06, + "loss": 2.4797031402587892, + "step": 236220 + }, + { + "epoch": 0.6430924186742888, + "grad_norm": 2.177490711212158, + "learning_rate": 7.1479351233474184e-06, + "loss": 2.472820854187012, + "step": 236240 + }, + { + "epoch": 0.6431468626650333, + "grad_norm": 2.237399101257324, + "learning_rate": 7.1468447594384625e-06, + "loss": 2.478732872009277, + "step": 236260 + }, + { + "epoch": 0.6432013066557779, + "grad_norm": 2.208219528198242, + "learning_rate": 7.145754395529508e-06, + "loss": 2.466090774536133, + "step": 236280 + }, + { + "epoch": 0.6432557506465224, + "grad_norm": 2.2321712970733643, + "learning_rate": 7.144664031620554e-06, + "loss": 2.4968809127807616, + "step": 236300 + }, + { + "epoch": 0.6433101946372669, + "grad_norm": 2.2014617919921875, + "learning_rate": 7.1435736677116e-06, + "loss": 2.471793365478516, + "step": 236320 + }, + { + "epoch": 0.6433646386280114, + "grad_norm": 2.3961024284362793, + "learning_rate": 7.142483303802644e-06, + "loss": 2.4853281021118163, + "step": 236340 + }, + { + "epoch": 0.6434190826187559, + "grad_norm": 2.14431095123291, + "learning_rate": 7.14139293989369e-06, + "loss": 2.466749382019043, + "step": 236360 + }, + { + "epoch": 0.6434735266095005, + "grad_norm": 2.0929977893829346, + "learning_rate": 7.140302575984736e-06, + "loss": 2.474661445617676, + "step": 236380 + }, + { + "epoch": 0.643527970600245, + "grad_norm": 2.772372007369995, + "learning_rate": 7.1392122120757815e-06, + "loss": 2.4786865234375, + "step": 236400 + }, + { + "epoch": 0.6435824145909895, + "grad_norm": 2.1779580116271973, + "learning_rate": 7.138121848166826e-06, + "loss": 2.496051788330078, + "step": 236420 + }, + { + "epoch": 0.643636858581734, + "grad_norm": 2.28690242767334, + "learning_rate": 7.1370314842578714e-06, + "loss": 2.445780563354492, + "step": 236440 + }, + { + "epoch": 0.6436913025724785, + "grad_norm": 2.2411234378814697, + "learning_rate": 7.135941120348917e-06, + "loss": 2.4919460296630858, + "step": 236460 + }, + { + "epoch": 0.643745746563223, + "grad_norm": 2.192976713180542, + "learning_rate": 7.134850756439963e-06, + "loss": 2.492965316772461, + "step": 236480 + }, + { + "epoch": 0.6438001905539676, + "grad_norm": 2.3245372772216797, + "learning_rate": 7.133760392531007e-06, + "loss": 2.463351821899414, + "step": 236500 + }, + { + "epoch": 0.6438546345447121, + "grad_norm": 2.403543710708618, + "learning_rate": 7.132670028622053e-06, + "loss": 2.43532772064209, + "step": 236520 + }, + { + "epoch": 0.6439090785354566, + "grad_norm": 2.1702215671539307, + "learning_rate": 7.131579664713099e-06, + "loss": 2.4350889205932615, + "step": 236540 + }, + { + "epoch": 0.6439635225262011, + "grad_norm": 3.219754934310913, + "learning_rate": 7.130489300804145e-06, + "loss": 2.4729162216186524, + "step": 236560 + }, + { + "epoch": 0.6440179665169457, + "grad_norm": 2.2044270038604736, + "learning_rate": 7.129398936895189e-06, + "loss": 2.479174995422363, + "step": 236580 + }, + { + "epoch": 0.6440724105076903, + "grad_norm": 2.0941014289855957, + "learning_rate": 7.1283085729862345e-06, + "loss": 2.454443168640137, + "step": 236600 + }, + { + "epoch": 0.6441268544984348, + "grad_norm": 2.22419810295105, + "learning_rate": 7.12721820907728e-06, + "loss": 2.4771306991577147, + "step": 236620 + }, + { + "epoch": 0.6441812984891793, + "grad_norm": 2.2219364643096924, + "learning_rate": 7.126127845168326e-06, + "loss": 2.460701751708984, + "step": 236640 + }, + { + "epoch": 0.6442357424799238, + "grad_norm": 2.1707098484039307, + "learning_rate": 7.125037481259371e-06, + "loss": 2.483667755126953, + "step": 236660 + }, + { + "epoch": 0.6442901864706683, + "grad_norm": 2.320305109024048, + "learning_rate": 7.123947117350416e-06, + "loss": 2.4763254165649413, + "step": 236680 + }, + { + "epoch": 0.6443446304614128, + "grad_norm": 2.1811530590057373, + "learning_rate": 7.122856753441462e-06, + "loss": 2.476542663574219, + "step": 236700 + }, + { + "epoch": 0.6443990744521574, + "grad_norm": 2.135777235031128, + "learning_rate": 7.121766389532508e-06, + "loss": 2.4520055770874025, + "step": 236720 + }, + { + "epoch": 0.6444535184429019, + "grad_norm": 2.249192476272583, + "learning_rate": 7.120676025623553e-06, + "loss": 2.4844593048095702, + "step": 236740 + }, + { + "epoch": 0.6445079624336464, + "grad_norm": 2.269975423812866, + "learning_rate": 7.119585661714598e-06, + "loss": 2.45789794921875, + "step": 236760 + }, + { + "epoch": 0.6445624064243909, + "grad_norm": 2.2061538696289062, + "learning_rate": 7.118495297805643e-06, + "loss": 2.4579984664916994, + "step": 236780 + }, + { + "epoch": 0.6446168504151354, + "grad_norm": 2.0806567668914795, + "learning_rate": 7.117404933896689e-06, + "loss": 2.4598228454589846, + "step": 236800 + }, + { + "epoch": 0.64467129440588, + "grad_norm": 2.1056671142578125, + "learning_rate": 7.116314569987734e-06, + "loss": 2.486129570007324, + "step": 236820 + }, + { + "epoch": 0.6447257383966245, + "grad_norm": 2.3115804195404053, + "learning_rate": 7.115224206078779e-06, + "loss": 2.4650461196899416, + "step": 236840 + }, + { + "epoch": 0.644780182387369, + "grad_norm": 2.257445812225342, + "learning_rate": 7.114133842169825e-06, + "loss": 2.4433250427246094, + "step": 236860 + }, + { + "epoch": 0.6448346263781135, + "grad_norm": 2.103365659713745, + "learning_rate": 7.113043478260871e-06, + "loss": 2.4366117477416993, + "step": 236880 + }, + { + "epoch": 0.644889070368858, + "grad_norm": 2.2020480632781982, + "learning_rate": 7.111953114351916e-06, + "loss": 2.422313690185547, + "step": 236900 + }, + { + "epoch": 0.6449435143596025, + "grad_norm": 2.2962334156036377, + "learning_rate": 7.110862750442961e-06, + "loss": 2.4919673919677736, + "step": 236920 + }, + { + "epoch": 0.6449979583503471, + "grad_norm": 2.4552793502807617, + "learning_rate": 7.1097723865340065e-06, + "loss": 2.4430000305175783, + "step": 236940 + }, + { + "epoch": 0.6450524023410916, + "grad_norm": 2.1535394191741943, + "learning_rate": 7.108682022625052e-06, + "loss": 2.4650136947631838, + "step": 236960 + }, + { + "epoch": 0.6451068463318361, + "grad_norm": 2.1932201385498047, + "learning_rate": 7.107591658716097e-06, + "loss": 2.4706148147583007, + "step": 236980 + }, + { + "epoch": 0.6451612903225806, + "grad_norm": 2.166599988937378, + "learning_rate": 7.106501294807142e-06, + "loss": 2.4894477844238283, + "step": 237000 + }, + { + "epoch": 0.6452157343133251, + "grad_norm": 2.2872846126556396, + "learning_rate": 7.105410930898188e-06, + "loss": 2.464112091064453, + "step": 237020 + }, + { + "epoch": 0.6452701783040697, + "grad_norm": 2.2477681636810303, + "learning_rate": 7.104320566989234e-06, + "loss": 2.4725133895874025, + "step": 237040 + }, + { + "epoch": 0.6453246222948142, + "grad_norm": 2.073695182800293, + "learning_rate": 7.103230203080279e-06, + "loss": 2.457091522216797, + "step": 237060 + }, + { + "epoch": 0.6453790662855587, + "grad_norm": 2.1253108978271484, + "learning_rate": 7.102139839171324e-06, + "loss": 2.421588325500488, + "step": 237080 + }, + { + "epoch": 0.6454335102763032, + "grad_norm": 2.0885181427001953, + "learning_rate": 7.10104947526237e-06, + "loss": 2.496084976196289, + "step": 237100 + }, + { + "epoch": 0.6454879542670477, + "grad_norm": 2.1883482933044434, + "learning_rate": 7.099959111353415e-06, + "loss": 2.4533533096313476, + "step": 237120 + }, + { + "epoch": 0.6455423982577922, + "grad_norm": 2.209669351577759, + "learning_rate": 7.09886874744446e-06, + "loss": 2.4828428268432616, + "step": 237140 + }, + { + "epoch": 0.6455968422485369, + "grad_norm": 2.2220566272735596, + "learning_rate": 7.097778383535505e-06, + "loss": 2.4952741622924806, + "step": 237160 + }, + { + "epoch": 0.6456512862392814, + "grad_norm": 2.126969814300537, + "learning_rate": 7.096688019626551e-06, + "loss": 2.454747772216797, + "step": 237180 + }, + { + "epoch": 0.6457057302300259, + "grad_norm": 2.14159893989563, + "learning_rate": 7.095597655717596e-06, + "loss": 2.4765995025634764, + "step": 237200 + }, + { + "epoch": 0.6457601742207704, + "grad_norm": 2.1704351902008057, + "learning_rate": 7.094507291808642e-06, + "loss": 2.506072998046875, + "step": 237220 + }, + { + "epoch": 0.6458146182115149, + "grad_norm": 2.139110565185547, + "learning_rate": 7.093416927899687e-06, + "loss": 2.492441940307617, + "step": 237240 + }, + { + "epoch": 0.6458690622022595, + "grad_norm": 2.21885085105896, + "learning_rate": 7.092326563990733e-06, + "loss": 2.458908271789551, + "step": 237260 + }, + { + "epoch": 0.645923506193004, + "grad_norm": 2.15217924118042, + "learning_rate": 7.091236200081778e-06, + "loss": 2.4953052520751955, + "step": 237280 + }, + { + "epoch": 0.6459779501837485, + "grad_norm": 2.1649489402770996, + "learning_rate": 7.0901458361728234e-06, + "loss": 2.4874475479125975, + "step": 237300 + }, + { + "epoch": 0.646032394174493, + "grad_norm": 2.0975375175476074, + "learning_rate": 7.089055472263868e-06, + "loss": 2.5225288391113283, + "step": 237320 + }, + { + "epoch": 0.6460868381652375, + "grad_norm": 2.1826179027557373, + "learning_rate": 7.087965108354914e-06, + "loss": 2.4911882400512697, + "step": 237340 + }, + { + "epoch": 0.646141282155982, + "grad_norm": 2.6166839599609375, + "learning_rate": 7.086874744445959e-06, + "loss": 2.453167724609375, + "step": 237360 + }, + { + "epoch": 0.6461957261467266, + "grad_norm": 2.1865971088409424, + "learning_rate": 7.085784380537005e-06, + "loss": 2.484014701843262, + "step": 237380 + }, + { + "epoch": 0.6462501701374711, + "grad_norm": 2.298832416534424, + "learning_rate": 7.08469401662805e-06, + "loss": 2.4869058609008787, + "step": 237400 + }, + { + "epoch": 0.6463046141282156, + "grad_norm": 2.3014748096466064, + "learning_rate": 7.083603652719096e-06, + "loss": 2.4707735061645506, + "step": 237420 + }, + { + "epoch": 0.6463590581189601, + "grad_norm": 2.1587584018707275, + "learning_rate": 7.082513288810141e-06, + "loss": 2.4599540710449217, + "step": 237440 + }, + { + "epoch": 0.6464135021097046, + "grad_norm": 2.6340765953063965, + "learning_rate": 7.0814229249011865e-06, + "loss": 2.4618215560913086, + "step": 237460 + }, + { + "epoch": 0.6464679461004491, + "grad_norm": 2.539055109024048, + "learning_rate": 7.0803325609922315e-06, + "loss": 2.447477340698242, + "step": 237480 + }, + { + "epoch": 0.6465223900911937, + "grad_norm": 2.362637758255005, + "learning_rate": 7.079242197083277e-06, + "loss": 2.4696657180786135, + "step": 237500 + }, + { + "epoch": 0.6465768340819382, + "grad_norm": 2.2012526988983154, + "learning_rate": 7.078151833174322e-06, + "loss": 2.4506175994873045, + "step": 237520 + }, + { + "epoch": 0.6466312780726827, + "grad_norm": 2.036008358001709, + "learning_rate": 7.077061469265368e-06, + "loss": 2.442538261413574, + "step": 237540 + }, + { + "epoch": 0.6466857220634272, + "grad_norm": 2.3590192794799805, + "learning_rate": 7.075971105356413e-06, + "loss": 2.4941856384277346, + "step": 237560 + }, + { + "epoch": 0.6467401660541717, + "grad_norm": 2.149326801300049, + "learning_rate": 7.074880741447459e-06, + "loss": 2.4825956344604494, + "step": 237580 + }, + { + "epoch": 0.6467946100449163, + "grad_norm": 2.125791311264038, + "learning_rate": 7.073790377538504e-06, + "loss": 2.451371765136719, + "step": 237600 + }, + { + "epoch": 0.6468490540356608, + "grad_norm": 2.14355731010437, + "learning_rate": 7.07270001362955e-06, + "loss": 2.480946159362793, + "step": 237620 + }, + { + "epoch": 0.6469034980264053, + "grad_norm": 2.142333984375, + "learning_rate": 7.071609649720595e-06, + "loss": 2.487308883666992, + "step": 237640 + }, + { + "epoch": 0.6469579420171498, + "grad_norm": 2.2859060764312744, + "learning_rate": 7.07051928581164e-06, + "loss": 2.4643741607666017, + "step": 237660 + }, + { + "epoch": 0.6470123860078943, + "grad_norm": 2.1503283977508545, + "learning_rate": 7.069428921902685e-06, + "loss": 2.4351882934570312, + "step": 237680 + }, + { + "epoch": 0.6470668299986388, + "grad_norm": 2.152996063232422, + "learning_rate": 7.068338557993731e-06, + "loss": 2.492097091674805, + "step": 237700 + }, + { + "epoch": 0.6471212739893835, + "grad_norm": 2.2042765617370605, + "learning_rate": 7.067248194084776e-06, + "loss": 2.4718971252441406, + "step": 237720 + }, + { + "epoch": 0.647175717980128, + "grad_norm": 2.3627922534942627, + "learning_rate": 7.066157830175822e-06, + "loss": 2.462910270690918, + "step": 237740 + }, + { + "epoch": 0.6472301619708725, + "grad_norm": 2.092775821685791, + "learning_rate": 7.065067466266867e-06, + "loss": 2.4565935134887695, + "step": 237760 + }, + { + "epoch": 0.647284605961617, + "grad_norm": 2.090878963470459, + "learning_rate": 7.063977102357913e-06, + "loss": 2.467854309082031, + "step": 237780 + }, + { + "epoch": 0.6473390499523615, + "grad_norm": 2.062169075012207, + "learning_rate": 7.062886738448958e-06, + "loss": 2.4719003677368163, + "step": 237800 + }, + { + "epoch": 0.6473934939431061, + "grad_norm": 2.1365699768066406, + "learning_rate": 7.0617963745400035e-06, + "loss": 2.482225036621094, + "step": 237820 + }, + { + "epoch": 0.6474479379338506, + "grad_norm": 2.1645307540893555, + "learning_rate": 7.0607060106310484e-06, + "loss": 2.514569091796875, + "step": 237840 + }, + { + "epoch": 0.6475023819245951, + "grad_norm": 2.1269943714141846, + "learning_rate": 7.059615646722094e-06, + "loss": 2.444161796569824, + "step": 237860 + }, + { + "epoch": 0.6475568259153396, + "grad_norm": 2.3189051151275635, + "learning_rate": 7.058525282813139e-06, + "loss": 2.487582206726074, + "step": 237880 + }, + { + "epoch": 0.6476112699060841, + "grad_norm": 2.278582811355591, + "learning_rate": 7.057434918904185e-06, + "loss": 2.470074462890625, + "step": 237900 + }, + { + "epoch": 0.6476657138968286, + "grad_norm": 2.2447798252105713, + "learning_rate": 7.05634455499523e-06, + "loss": 2.470758819580078, + "step": 237920 + }, + { + "epoch": 0.6477201578875732, + "grad_norm": 2.2657742500305176, + "learning_rate": 7.055254191086276e-06, + "loss": 2.472309112548828, + "step": 237940 + }, + { + "epoch": 0.6477746018783177, + "grad_norm": 2.211719274520874, + "learning_rate": 7.054163827177321e-06, + "loss": 2.4962369918823244, + "step": 237960 + }, + { + "epoch": 0.6478290458690622, + "grad_norm": 2.131089925765991, + "learning_rate": 7.0530734632683666e-06, + "loss": 2.437000274658203, + "step": 237980 + }, + { + "epoch": 0.6478834898598067, + "grad_norm": 2.23991322517395, + "learning_rate": 7.0519830993594115e-06, + "loss": 2.48218994140625, + "step": 238000 + }, + { + "epoch": 0.6479379338505512, + "grad_norm": 2.147663116455078, + "learning_rate": 7.050892735450457e-06, + "loss": 2.452508544921875, + "step": 238020 + }, + { + "epoch": 0.6479923778412958, + "grad_norm": 2.1665446758270264, + "learning_rate": 7.049802371541502e-06, + "loss": 2.514125442504883, + "step": 238040 + }, + { + "epoch": 0.6480468218320403, + "grad_norm": 2.1201677322387695, + "learning_rate": 7.048712007632548e-06, + "loss": 2.4514793395996093, + "step": 238060 + }, + { + "epoch": 0.6481012658227848, + "grad_norm": 2.094632625579834, + "learning_rate": 7.047621643723593e-06, + "loss": 2.454210090637207, + "step": 238080 + }, + { + "epoch": 0.6481557098135293, + "grad_norm": 2.077122926712036, + "learning_rate": 7.046531279814639e-06, + "loss": 2.4902748107910155, + "step": 238100 + }, + { + "epoch": 0.6482101538042738, + "grad_norm": 2.0685627460479736, + "learning_rate": 7.045440915905684e-06, + "loss": 2.483229064941406, + "step": 238120 + }, + { + "epoch": 0.6482645977950183, + "grad_norm": 2.1734323501586914, + "learning_rate": 7.04435055199673e-06, + "loss": 2.484222412109375, + "step": 238140 + }, + { + "epoch": 0.6483190417857629, + "grad_norm": 2.2636666297912598, + "learning_rate": 7.043260188087775e-06, + "loss": 2.4637048721313475, + "step": 238160 + }, + { + "epoch": 0.6483734857765074, + "grad_norm": 2.1911330223083496, + "learning_rate": 7.04216982417882e-06, + "loss": 2.4620777130126954, + "step": 238180 + }, + { + "epoch": 0.6484279297672519, + "grad_norm": 2.140892267227173, + "learning_rate": 7.041079460269865e-06, + "loss": 2.471537399291992, + "step": 238200 + }, + { + "epoch": 0.6484823737579964, + "grad_norm": 2.2358856201171875, + "learning_rate": 7.039989096360911e-06, + "loss": 2.472271728515625, + "step": 238220 + }, + { + "epoch": 0.648536817748741, + "grad_norm": 2.111145257949829, + "learning_rate": 7.038898732451956e-06, + "loss": 2.466230773925781, + "step": 238240 + }, + { + "epoch": 0.6485912617394856, + "grad_norm": 2.199171543121338, + "learning_rate": 7.037808368543002e-06, + "loss": 2.4930553436279297, + "step": 238260 + }, + { + "epoch": 0.6486457057302301, + "grad_norm": 2.154628276824951, + "learning_rate": 7.036718004634047e-06, + "loss": 2.4788848876953127, + "step": 238280 + }, + { + "epoch": 0.6487001497209746, + "grad_norm": 2.370910406112671, + "learning_rate": 7.035627640725092e-06, + "loss": 2.5028371810913086, + "step": 238300 + }, + { + "epoch": 0.6487545937117191, + "grad_norm": 2.192122220993042, + "learning_rate": 7.034537276816138e-06, + "loss": 2.4830558776855467, + "step": 238320 + }, + { + "epoch": 0.6488090377024636, + "grad_norm": 2.4766409397125244, + "learning_rate": 7.0334469129071835e-06, + "loss": 2.463479995727539, + "step": 238340 + }, + { + "epoch": 0.6488634816932081, + "grad_norm": 2.353991746902466, + "learning_rate": 7.032356548998229e-06, + "loss": 2.4646766662597654, + "step": 238360 + }, + { + "epoch": 0.6489179256839527, + "grad_norm": 2.183943033218384, + "learning_rate": 7.0312661850892734e-06, + "loss": 2.4800626754760744, + "step": 238380 + }, + { + "epoch": 0.6489723696746972, + "grad_norm": 2.18300199508667, + "learning_rate": 7.030175821180319e-06, + "loss": 2.477071762084961, + "step": 238400 + }, + { + "epoch": 0.6490268136654417, + "grad_norm": 2.1089863777160645, + "learning_rate": 7.029085457271365e-06, + "loss": 2.4687068939208983, + "step": 238420 + }, + { + "epoch": 0.6490812576561862, + "grad_norm": 2.1761550903320312, + "learning_rate": 7.027995093362411e-06, + "loss": 2.4379837036132814, + "step": 238440 + }, + { + "epoch": 0.6491357016469307, + "grad_norm": 2.1544547080993652, + "learning_rate": 7.026904729453455e-06, + "loss": 2.4404760360717774, + "step": 238460 + }, + { + "epoch": 0.6491901456376752, + "grad_norm": 2.2950308322906494, + "learning_rate": 7.025814365544501e-06, + "loss": 2.527795600891113, + "step": 238480 + }, + { + "epoch": 0.6492445896284198, + "grad_norm": 2.1494925022125244, + "learning_rate": 7.024724001635547e-06, + "loss": 2.482629585266113, + "step": 238500 + }, + { + "epoch": 0.6492990336191643, + "grad_norm": 2.1288232803344727, + "learning_rate": 7.023633637726592e-06, + "loss": 2.4488025665283204, + "step": 238520 + }, + { + "epoch": 0.6493534776099088, + "grad_norm": 2.122493267059326, + "learning_rate": 7.0225432738176365e-06, + "loss": 2.4809186935424803, + "step": 238540 + }, + { + "epoch": 0.6494079216006533, + "grad_norm": 2.146115303039551, + "learning_rate": 7.021452909908682e-06, + "loss": 2.476763916015625, + "step": 238560 + }, + { + "epoch": 0.6494623655913978, + "grad_norm": 2.1787967681884766, + "learning_rate": 7.020362545999728e-06, + "loss": 2.451686668395996, + "step": 238580 + }, + { + "epoch": 0.6495168095821424, + "grad_norm": 2.141810894012451, + "learning_rate": 7.019272182090774e-06, + "loss": 2.4712249755859377, + "step": 238600 + }, + { + "epoch": 0.6495712535728869, + "grad_norm": 2.2904775142669678, + "learning_rate": 7.018181818181818e-06, + "loss": 2.4833492279052733, + "step": 238620 + }, + { + "epoch": 0.6496256975636314, + "grad_norm": 2.2092533111572266, + "learning_rate": 7.017091454272864e-06, + "loss": 2.4813209533691407, + "step": 238640 + }, + { + "epoch": 0.6496801415543759, + "grad_norm": 2.2464334964752197, + "learning_rate": 7.01600109036391e-06, + "loss": 2.452976608276367, + "step": 238660 + }, + { + "epoch": 0.6497345855451204, + "grad_norm": 2.3367581367492676, + "learning_rate": 7.0149107264549555e-06, + "loss": 2.487636947631836, + "step": 238680 + }, + { + "epoch": 0.6497890295358649, + "grad_norm": 2.110083818435669, + "learning_rate": 7.013820362546e-06, + "loss": 2.4737361907958983, + "step": 238700 + }, + { + "epoch": 0.6498434735266095, + "grad_norm": 2.2347259521484375, + "learning_rate": 7.012729998637045e-06, + "loss": 2.469451332092285, + "step": 238720 + }, + { + "epoch": 0.649897917517354, + "grad_norm": 2.3107426166534424, + "learning_rate": 7.011639634728091e-06, + "loss": 2.4588010787963865, + "step": 238740 + }, + { + "epoch": 0.6499523615080985, + "grad_norm": 2.1474082469940186, + "learning_rate": 7.010549270819137e-06, + "loss": 2.4827125549316404, + "step": 238760 + }, + { + "epoch": 0.650006805498843, + "grad_norm": 2.408618688583374, + "learning_rate": 7.009458906910181e-06, + "loss": 2.4465988159179686, + "step": 238780 + }, + { + "epoch": 0.6500612494895875, + "grad_norm": 2.28169846534729, + "learning_rate": 7.008368543001227e-06, + "loss": 2.4916288375854494, + "step": 238800 + }, + { + "epoch": 0.6501156934803322, + "grad_norm": 2.132213830947876, + "learning_rate": 7.007278179092273e-06, + "loss": 2.4809539794921873, + "step": 238820 + }, + { + "epoch": 0.6501701374710767, + "grad_norm": 2.225158452987671, + "learning_rate": 7.0061878151833186e-06, + "loss": 2.4592918395996093, + "step": 238840 + }, + { + "epoch": 0.6502245814618212, + "grad_norm": 2.1223466396331787, + "learning_rate": 7.005097451274363e-06, + "loss": 2.472447967529297, + "step": 238860 + }, + { + "epoch": 0.6502790254525657, + "grad_norm": 2.1398675441741943, + "learning_rate": 7.0040070873654085e-06, + "loss": 2.4842292785644533, + "step": 238880 + }, + { + "epoch": 0.6503334694433102, + "grad_norm": 2.2124195098876953, + "learning_rate": 7.002916723456454e-06, + "loss": 2.4613664627075194, + "step": 238900 + }, + { + "epoch": 0.6503879134340547, + "grad_norm": 2.184772491455078, + "learning_rate": 7.0018263595475e-06, + "loss": 2.4654701232910154, + "step": 238920 + }, + { + "epoch": 0.6504423574247993, + "grad_norm": 2.2386436462402344, + "learning_rate": 7.000735995638544e-06, + "loss": 2.448288154602051, + "step": 238940 + }, + { + "epoch": 0.6504968014155438, + "grad_norm": 2.237119674682617, + "learning_rate": 6.99964563172959e-06, + "loss": 2.4738449096679687, + "step": 238960 + }, + { + "epoch": 0.6505512454062883, + "grad_norm": 2.2008440494537354, + "learning_rate": 6.998555267820636e-06, + "loss": 2.457039642333984, + "step": 238980 + }, + { + "epoch": 0.6506056893970328, + "grad_norm": 2.113879919052124, + "learning_rate": 6.997464903911682e-06, + "loss": 2.4818187713623048, + "step": 239000 + }, + { + "epoch": 0.6506601333877773, + "grad_norm": 3.0045759677886963, + "learning_rate": 6.996374540002726e-06, + "loss": 2.4812604904174806, + "step": 239020 + }, + { + "epoch": 0.6507145773785219, + "grad_norm": 2.2761807441711426, + "learning_rate": 6.995284176093772e-06, + "loss": 2.4723997116088867, + "step": 239040 + }, + { + "epoch": 0.6507690213692664, + "grad_norm": 2.1320462226867676, + "learning_rate": 6.994193812184817e-06, + "loss": 2.4866838455200195, + "step": 239060 + }, + { + "epoch": 0.6508234653600109, + "grad_norm": 2.1448400020599365, + "learning_rate": 6.993103448275863e-06, + "loss": 2.4620227813720703, + "step": 239080 + }, + { + "epoch": 0.6508779093507554, + "grad_norm": 2.162111520767212, + "learning_rate": 6.992013084366907e-06, + "loss": 2.4746564865112304, + "step": 239100 + }, + { + "epoch": 0.6509323533414999, + "grad_norm": 2.510733127593994, + "learning_rate": 6.990922720457953e-06, + "loss": 2.457242012023926, + "step": 239120 + }, + { + "epoch": 0.6509867973322444, + "grad_norm": 2.0005404949188232, + "learning_rate": 6.989832356548999e-06, + "loss": 2.451953887939453, + "step": 239140 + }, + { + "epoch": 0.651041241322989, + "grad_norm": 2.3176302909851074, + "learning_rate": 6.988741992640045e-06, + "loss": 2.4541852951049803, + "step": 239160 + }, + { + "epoch": 0.6510956853137335, + "grad_norm": 2.254305839538574, + "learning_rate": 6.987651628731089e-06, + "loss": 2.4584587097167967, + "step": 239180 + }, + { + "epoch": 0.651150129304478, + "grad_norm": 2.2102859020233154, + "learning_rate": 6.986561264822135e-06, + "loss": 2.463004493713379, + "step": 239200 + }, + { + "epoch": 0.6512045732952225, + "grad_norm": 2.1406543254852295, + "learning_rate": 6.9854709009131805e-06, + "loss": 2.447849464416504, + "step": 239220 + }, + { + "epoch": 0.651259017285967, + "grad_norm": 2.1633718013763428, + "learning_rate": 6.984380537004226e-06, + "loss": 2.460264778137207, + "step": 239240 + }, + { + "epoch": 0.6513134612767116, + "grad_norm": 2.2655279636383057, + "learning_rate": 6.98329017309527e-06, + "loss": 2.472093200683594, + "step": 239260 + }, + { + "epoch": 0.6513679052674561, + "grad_norm": 2.21500825881958, + "learning_rate": 6.982199809186316e-06, + "loss": 2.453849220275879, + "step": 239280 + }, + { + "epoch": 0.6514223492582006, + "grad_norm": 2.4425556659698486, + "learning_rate": 6.981109445277362e-06, + "loss": 2.486118125915527, + "step": 239300 + }, + { + "epoch": 0.6514767932489451, + "grad_norm": 2.1969854831695557, + "learning_rate": 6.980019081368408e-06, + "loss": 2.446369171142578, + "step": 239320 + }, + { + "epoch": 0.6515312372396896, + "grad_norm": 2.182126998901367, + "learning_rate": 6.978928717459452e-06, + "loss": 2.4530101776123048, + "step": 239340 + }, + { + "epoch": 0.6515856812304341, + "grad_norm": 2.1087019443511963, + "learning_rate": 6.977838353550498e-06, + "loss": 2.4832761764526365, + "step": 239360 + }, + { + "epoch": 0.6516401252211788, + "grad_norm": 2.452056884765625, + "learning_rate": 6.9767479896415436e-06, + "loss": 2.4556564331054687, + "step": 239380 + }, + { + "epoch": 0.6516945692119233, + "grad_norm": 2.388214111328125, + "learning_rate": 6.975657625732589e-06, + "loss": 2.4667474746704103, + "step": 239400 + }, + { + "epoch": 0.6517490132026678, + "grad_norm": 2.2614846229553223, + "learning_rate": 6.9745672618236335e-06, + "loss": 2.4815765380859376, + "step": 239420 + }, + { + "epoch": 0.6518034571934123, + "grad_norm": 2.380319118499756, + "learning_rate": 6.973476897914679e-06, + "loss": 2.4742456436157227, + "step": 239440 + }, + { + "epoch": 0.6518579011841568, + "grad_norm": 2.240845203399658, + "learning_rate": 6.972386534005725e-06, + "loss": 2.459582710266113, + "step": 239460 + }, + { + "epoch": 0.6519123451749014, + "grad_norm": 2.155393123626709, + "learning_rate": 6.971296170096771e-06, + "loss": 2.4789146423339843, + "step": 239480 + }, + { + "epoch": 0.6519667891656459, + "grad_norm": 2.1733906269073486, + "learning_rate": 6.970205806187815e-06, + "loss": 2.4604496002197265, + "step": 239500 + }, + { + "epoch": 0.6520212331563904, + "grad_norm": 2.360888719558716, + "learning_rate": 6.969115442278861e-06, + "loss": 2.4791473388671874, + "step": 239520 + }, + { + "epoch": 0.6520756771471349, + "grad_norm": 2.17421293258667, + "learning_rate": 6.968025078369907e-06, + "loss": 2.498063850402832, + "step": 239540 + }, + { + "epoch": 0.6521301211378794, + "grad_norm": 2.1493215560913086, + "learning_rate": 6.9669347144609525e-06, + "loss": 2.4768436431884764, + "step": 239560 + }, + { + "epoch": 0.6521845651286239, + "grad_norm": 2.1364355087280273, + "learning_rate": 6.9658443505519966e-06, + "loss": 2.5000444412231446, + "step": 239580 + }, + { + "epoch": 0.6522390091193685, + "grad_norm": 2.1172709465026855, + "learning_rate": 6.964753986643042e-06, + "loss": 2.5091766357421874, + "step": 239600 + }, + { + "epoch": 0.652293453110113, + "grad_norm": 2.218360662460327, + "learning_rate": 6.963663622734088e-06, + "loss": 2.5053438186645507, + "step": 239620 + }, + { + "epoch": 0.6523478971008575, + "grad_norm": 2.245415210723877, + "learning_rate": 6.962573258825134e-06, + "loss": 2.4480722427368162, + "step": 239640 + }, + { + "epoch": 0.652402341091602, + "grad_norm": 2.163653612136841, + "learning_rate": 6.961482894916178e-06, + "loss": 2.449513626098633, + "step": 239660 + }, + { + "epoch": 0.6524567850823465, + "grad_norm": 2.316633462905884, + "learning_rate": 6.960392531007224e-06, + "loss": 2.4901754379272463, + "step": 239680 + }, + { + "epoch": 0.652511229073091, + "grad_norm": 2.166825532913208, + "learning_rate": 6.95930216709827e-06, + "loss": 2.46953125, + "step": 239700 + }, + { + "epoch": 0.6525656730638356, + "grad_norm": 2.1663055419921875, + "learning_rate": 6.9582118031893155e-06, + "loss": 2.456293296813965, + "step": 239720 + }, + { + "epoch": 0.6526201170545801, + "grad_norm": 2.3759005069732666, + "learning_rate": 6.95712143928036e-06, + "loss": 2.4555389404296877, + "step": 239740 + }, + { + "epoch": 0.6526745610453246, + "grad_norm": 2.141324996948242, + "learning_rate": 6.9560310753714055e-06, + "loss": 2.4544227600097654, + "step": 239760 + }, + { + "epoch": 0.6527290050360691, + "grad_norm": 2.2002766132354736, + "learning_rate": 6.954940711462451e-06, + "loss": 2.476362609863281, + "step": 239780 + }, + { + "epoch": 0.6527834490268136, + "grad_norm": 2.0872273445129395, + "learning_rate": 6.953850347553497e-06, + "loss": 2.474810981750488, + "step": 239800 + }, + { + "epoch": 0.6528378930175582, + "grad_norm": 2.303891897201538, + "learning_rate": 6.952759983644541e-06, + "loss": 2.484726142883301, + "step": 239820 + }, + { + "epoch": 0.6528923370083027, + "grad_norm": 2.112239122390747, + "learning_rate": 6.951669619735587e-06, + "loss": 2.461046600341797, + "step": 239840 + }, + { + "epoch": 0.6529467809990472, + "grad_norm": 2.0730679035186768, + "learning_rate": 6.950579255826633e-06, + "loss": 2.472232627868652, + "step": 239860 + }, + { + "epoch": 0.6530012249897917, + "grad_norm": 2.1205928325653076, + "learning_rate": 6.949488891917679e-06, + "loss": 2.4341655731201173, + "step": 239880 + }, + { + "epoch": 0.6530556689805362, + "grad_norm": 2.174835681915283, + "learning_rate": 6.948398528008723e-06, + "loss": 2.4469936370849608, + "step": 239900 + }, + { + "epoch": 0.6531101129712807, + "grad_norm": 2.118725538253784, + "learning_rate": 6.9473081640997686e-06, + "loss": 2.496298408508301, + "step": 239920 + }, + { + "epoch": 0.6531645569620254, + "grad_norm": 2.102863073348999, + "learning_rate": 6.946217800190814e-06, + "loss": 2.5084625244140626, + "step": 239940 + }, + { + "epoch": 0.6532190009527699, + "grad_norm": 2.248563289642334, + "learning_rate": 6.94512743628186e-06, + "loss": 2.491611289978027, + "step": 239960 + }, + { + "epoch": 0.6532734449435144, + "grad_norm": 2.1609957218170166, + "learning_rate": 6.944037072372904e-06, + "loss": 2.4173395156860353, + "step": 239980 + }, + { + "epoch": 0.6533278889342589, + "grad_norm": 2.174647331237793, + "learning_rate": 6.94294670846395e-06, + "loss": 2.486830139160156, + "step": 240000 + }, + { + "epoch": 0.6533823329250034, + "grad_norm": 2.1892762184143066, + "learning_rate": 6.941856344554996e-06, + "loss": 2.468126106262207, + "step": 240020 + }, + { + "epoch": 0.653436776915748, + "grad_norm": 2.177271604537964, + "learning_rate": 6.940765980646042e-06, + "loss": 2.495715522766113, + "step": 240040 + }, + { + "epoch": 0.6534912209064925, + "grad_norm": 2.1783573627471924, + "learning_rate": 6.9396756167370875e-06, + "loss": 2.477169418334961, + "step": 240060 + }, + { + "epoch": 0.653545664897237, + "grad_norm": 2.1642935276031494, + "learning_rate": 6.938585252828132e-06, + "loss": 2.4805036544799806, + "step": 240080 + }, + { + "epoch": 0.6536001088879815, + "grad_norm": 2.1904618740081787, + "learning_rate": 6.9374948889191774e-06, + "loss": 2.447653388977051, + "step": 240100 + }, + { + "epoch": 0.653654552878726, + "grad_norm": 2.162548303604126, + "learning_rate": 6.936404525010223e-06, + "loss": 2.4745487213134765, + "step": 240120 + }, + { + "epoch": 0.6537089968694705, + "grad_norm": 6.5801544189453125, + "learning_rate": 6.935314161101269e-06, + "loss": 2.490595245361328, + "step": 240140 + }, + { + "epoch": 0.6537634408602151, + "grad_norm": 2.141146183013916, + "learning_rate": 6.934223797192313e-06, + "loss": 2.483954429626465, + "step": 240160 + }, + { + "epoch": 0.6538178848509596, + "grad_norm": 2.1224873065948486, + "learning_rate": 6.933133433283359e-06, + "loss": 2.4578399658203125, + "step": 240180 + }, + { + "epoch": 0.6538723288417041, + "grad_norm": 2.102679967880249, + "learning_rate": 6.932043069374405e-06, + "loss": 2.5116764068603517, + "step": 240200 + }, + { + "epoch": 0.6539267728324486, + "grad_norm": 2.0934717655181885, + "learning_rate": 6.93095270546545e-06, + "loss": 2.458926773071289, + "step": 240220 + }, + { + "epoch": 0.6539812168231931, + "grad_norm": 2.3243019580841064, + "learning_rate": 6.929862341556495e-06, + "loss": 2.461593437194824, + "step": 240240 + }, + { + "epoch": 0.6540356608139377, + "grad_norm": 2.158876657485962, + "learning_rate": 6.9287719776475405e-06, + "loss": 2.447104644775391, + "step": 240260 + }, + { + "epoch": 0.6540901048046822, + "grad_norm": 2.410604476928711, + "learning_rate": 6.927681613738586e-06, + "loss": 2.483109474182129, + "step": 240280 + }, + { + "epoch": 0.6541445487954267, + "grad_norm": 2.117936134338379, + "learning_rate": 6.926591249829631e-06, + "loss": 2.481983757019043, + "step": 240300 + }, + { + "epoch": 0.6541989927861712, + "grad_norm": 2.1665267944335938, + "learning_rate": 6.925500885920676e-06, + "loss": 2.4697593688964843, + "step": 240320 + }, + { + "epoch": 0.6542534367769157, + "grad_norm": 2.185509443283081, + "learning_rate": 6.924410522011722e-06, + "loss": 2.4595947265625, + "step": 240340 + }, + { + "epoch": 0.6543078807676602, + "grad_norm": 2.1998047828674316, + "learning_rate": 6.923320158102768e-06, + "loss": 2.4702653884887695, + "step": 240360 + }, + { + "epoch": 0.6543623247584048, + "grad_norm": 2.0632050037384033, + "learning_rate": 6.922229794193813e-06, + "loss": 2.453224754333496, + "step": 240380 + }, + { + "epoch": 0.6544167687491493, + "grad_norm": 2.121499538421631, + "learning_rate": 6.921139430284858e-06, + "loss": 2.480272102355957, + "step": 240400 + }, + { + "epoch": 0.6544712127398938, + "grad_norm": 2.3409550189971924, + "learning_rate": 6.920049066375904e-06, + "loss": 2.470871925354004, + "step": 240420 + }, + { + "epoch": 0.6545256567306383, + "grad_norm": 2.2667696475982666, + "learning_rate": 6.9189587024669494e-06, + "loss": 2.4491199493408202, + "step": 240440 + }, + { + "epoch": 0.6545801007213828, + "grad_norm": 2.1978237628936768, + "learning_rate": 6.917868338557994e-06, + "loss": 2.4321529388427736, + "step": 240460 + }, + { + "epoch": 0.6546345447121275, + "grad_norm": 2.3125228881835938, + "learning_rate": 6.916777974649039e-06, + "loss": 2.43416805267334, + "step": 240480 + }, + { + "epoch": 0.654688988702872, + "grad_norm": 2.164189338684082, + "learning_rate": 6.915687610740085e-06, + "loss": 2.4739028930664064, + "step": 240500 + }, + { + "epoch": 0.6547434326936165, + "grad_norm": 4.728310585021973, + "learning_rate": 6.914597246831131e-06, + "loss": 2.4467830657958984, + "step": 240520 + }, + { + "epoch": 0.654797876684361, + "grad_norm": 2.2730064392089844, + "learning_rate": 6.913506882922176e-06, + "loss": 2.441886329650879, + "step": 240540 + }, + { + "epoch": 0.6548523206751055, + "grad_norm": 2.102600336074829, + "learning_rate": 6.912416519013221e-06, + "loss": 2.470627784729004, + "step": 240560 + }, + { + "epoch": 0.65490676466585, + "grad_norm": 2.333613395690918, + "learning_rate": 6.911326155104267e-06, + "loss": 2.473908233642578, + "step": 240580 + }, + { + "epoch": 0.6549612086565946, + "grad_norm": 2.210430860519409, + "learning_rate": 6.9102357911953125e-06, + "loss": 2.458387565612793, + "step": 240600 + }, + { + "epoch": 0.6550156526473391, + "grad_norm": 2.1911261081695557, + "learning_rate": 6.9091454272863575e-06, + "loss": 2.4300714492797852, + "step": 240620 + }, + { + "epoch": 0.6550700966380836, + "grad_norm": 2.1978559494018555, + "learning_rate": 6.9080550633774024e-06, + "loss": 2.4217657089233398, + "step": 240640 + }, + { + "epoch": 0.6551245406288281, + "grad_norm": 2.2211387157440186, + "learning_rate": 6.906964699468448e-06, + "loss": 2.467666816711426, + "step": 240660 + }, + { + "epoch": 0.6551789846195726, + "grad_norm": 2.134859800338745, + "learning_rate": 6.905874335559494e-06, + "loss": 2.4658123016357423, + "step": 240680 + }, + { + "epoch": 0.6552334286103171, + "grad_norm": 2.309422254562378, + "learning_rate": 6.904783971650539e-06, + "loss": 2.4763181686401365, + "step": 240700 + }, + { + "epoch": 0.6552878726010617, + "grad_norm": 2.1826772689819336, + "learning_rate": 6.903693607741584e-06, + "loss": 2.4394437789916994, + "step": 240720 + }, + { + "epoch": 0.6553423165918062, + "grad_norm": 2.202111005783081, + "learning_rate": 6.90260324383263e-06, + "loss": 2.4505239486694337, + "step": 240740 + }, + { + "epoch": 0.6553967605825507, + "grad_norm": 2.047269821166992, + "learning_rate": 6.901512879923676e-06, + "loss": 2.467425537109375, + "step": 240760 + }, + { + "epoch": 0.6554512045732952, + "grad_norm": 2.1336328983306885, + "learning_rate": 6.9004225160147206e-06, + "loss": 2.460774040222168, + "step": 240780 + }, + { + "epoch": 0.6555056485640397, + "grad_norm": 2.191910982131958, + "learning_rate": 6.8993321521057655e-06, + "loss": 2.4812679290771484, + "step": 240800 + }, + { + "epoch": 0.6555600925547843, + "grad_norm": 2.162796974182129, + "learning_rate": 6.898241788196811e-06, + "loss": 2.469528579711914, + "step": 240820 + }, + { + "epoch": 0.6556145365455288, + "grad_norm": 2.233105182647705, + "learning_rate": 6.897151424287857e-06, + "loss": 2.455393981933594, + "step": 240840 + }, + { + "epoch": 0.6556689805362733, + "grad_norm": 2.172408103942871, + "learning_rate": 6.896061060378902e-06, + "loss": 2.4635589599609373, + "step": 240860 + }, + { + "epoch": 0.6557234245270178, + "grad_norm": 2.061769962310791, + "learning_rate": 6.894970696469947e-06, + "loss": 2.4396121978759764, + "step": 240880 + }, + { + "epoch": 0.6557778685177623, + "grad_norm": 2.200164318084717, + "learning_rate": 6.893880332560993e-06, + "loss": 2.487082862854004, + "step": 240900 + }, + { + "epoch": 0.6558323125085068, + "grad_norm": 2.1580090522766113, + "learning_rate": 6.892789968652039e-06, + "loss": 2.4844892501831053, + "step": 240920 + }, + { + "epoch": 0.6558867564992514, + "grad_norm": 2.3567521572113037, + "learning_rate": 6.891699604743084e-06, + "loss": 2.4806732177734374, + "step": 240940 + }, + { + "epoch": 0.6559412004899959, + "grad_norm": 2.175100803375244, + "learning_rate": 6.890609240834129e-06, + "loss": 2.4837825775146483, + "step": 240960 + }, + { + "epoch": 0.6559956444807404, + "grad_norm": 2.1539981365203857, + "learning_rate": 6.889518876925174e-06, + "loss": 2.4715410232543946, + "step": 240980 + }, + { + "epoch": 0.6560500884714849, + "grad_norm": 2.1298553943634033, + "learning_rate": 6.88842851301622e-06, + "loss": 2.4640579223632812, + "step": 241000 + }, + { + "epoch": 0.6561045324622294, + "grad_norm": 2.1763741970062256, + "learning_rate": 6.887338149107265e-06, + "loss": 2.491659164428711, + "step": 241020 + }, + { + "epoch": 0.656158976452974, + "grad_norm": 2.171340227127075, + "learning_rate": 6.88624778519831e-06, + "loss": 2.4696109771728514, + "step": 241040 + }, + { + "epoch": 0.6562134204437186, + "grad_norm": 2.1714346408843994, + "learning_rate": 6.885157421289356e-06, + "loss": 2.454056167602539, + "step": 241060 + }, + { + "epoch": 0.656267864434463, + "grad_norm": 2.217531681060791, + "learning_rate": 6.884067057380402e-06, + "loss": 2.473500633239746, + "step": 241080 + }, + { + "epoch": 0.6563223084252076, + "grad_norm": 2.1185379028320312, + "learning_rate": 6.882976693471447e-06, + "loss": 2.4717153549194335, + "step": 241100 + }, + { + "epoch": 0.6563767524159521, + "grad_norm": 2.265713930130005, + "learning_rate": 6.881886329562492e-06, + "loss": 2.4510440826416016, + "step": 241120 + }, + { + "epoch": 0.6564311964066966, + "grad_norm": 2.712756633758545, + "learning_rate": 6.8807959656535375e-06, + "loss": 2.4555139541625977, + "step": 241140 + }, + { + "epoch": 0.6564856403974412, + "grad_norm": 2.1976749897003174, + "learning_rate": 6.879705601744583e-06, + "loss": 2.450712203979492, + "step": 241160 + }, + { + "epoch": 0.6565400843881857, + "grad_norm": 3.1020143032073975, + "learning_rate": 6.878615237835628e-06, + "loss": 2.476882553100586, + "step": 241180 + }, + { + "epoch": 0.6565945283789302, + "grad_norm": 2.1595213413238525, + "learning_rate": 6.877524873926673e-06, + "loss": 2.4664531707763673, + "step": 241200 + }, + { + "epoch": 0.6566489723696747, + "grad_norm": 2.0353050231933594, + "learning_rate": 6.876434510017719e-06, + "loss": 2.443642807006836, + "step": 241220 + }, + { + "epoch": 0.6567034163604192, + "grad_norm": 2.136225461959839, + "learning_rate": 6.875344146108765e-06, + "loss": 2.4614965438842775, + "step": 241240 + }, + { + "epoch": 0.6567578603511638, + "grad_norm": 2.19692325592041, + "learning_rate": 6.87425378219981e-06, + "loss": 2.494478225708008, + "step": 241260 + }, + { + "epoch": 0.6568123043419083, + "grad_norm": 2.2381484508514404, + "learning_rate": 6.873163418290855e-06, + "loss": 2.458331298828125, + "step": 241280 + }, + { + "epoch": 0.6568667483326528, + "grad_norm": 2.227457046508789, + "learning_rate": 6.872073054381901e-06, + "loss": 2.461412048339844, + "step": 241300 + }, + { + "epoch": 0.6569211923233973, + "grad_norm": 2.321389675140381, + "learning_rate": 6.8709826904729455e-06, + "loss": 2.475102424621582, + "step": 241320 + }, + { + "epoch": 0.6569756363141418, + "grad_norm": 2.1943325996398926, + "learning_rate": 6.869892326563991e-06, + "loss": 2.4728982925415037, + "step": 241340 + }, + { + "epoch": 0.6570300803048863, + "grad_norm": 2.1759212017059326, + "learning_rate": 6.868801962655036e-06, + "loss": 2.443863296508789, + "step": 241360 + }, + { + "epoch": 0.6570845242956309, + "grad_norm": 2.172431707382202, + "learning_rate": 6.867711598746082e-06, + "loss": 2.472945213317871, + "step": 241380 + }, + { + "epoch": 0.6571389682863754, + "grad_norm": 2.190725326538086, + "learning_rate": 6.866621234837127e-06, + "loss": 2.437623977661133, + "step": 241400 + }, + { + "epoch": 0.6571934122771199, + "grad_norm": 2.151733160018921, + "learning_rate": 6.865530870928173e-06, + "loss": 2.4413497924804686, + "step": 241420 + }, + { + "epoch": 0.6572478562678644, + "grad_norm": 2.180894136428833, + "learning_rate": 6.864440507019218e-06, + "loss": 2.473868179321289, + "step": 241440 + }, + { + "epoch": 0.6573023002586089, + "grad_norm": 2.2646396160125732, + "learning_rate": 6.863350143110264e-06, + "loss": 2.4493995666503907, + "step": 241460 + }, + { + "epoch": 0.6573567442493535, + "grad_norm": 2.2848074436187744, + "learning_rate": 6.862259779201309e-06, + "loss": 2.4655879974365233, + "step": 241480 + }, + { + "epoch": 0.657411188240098, + "grad_norm": 2.2235267162323, + "learning_rate": 6.8611694152923544e-06, + "loss": 2.4797271728515624, + "step": 241500 + }, + { + "epoch": 0.6574656322308425, + "grad_norm": 2.418712854385376, + "learning_rate": 6.860079051383399e-06, + "loss": 2.4799007415771483, + "step": 241520 + }, + { + "epoch": 0.657520076221587, + "grad_norm": 2.153155565261841, + "learning_rate": 6.858988687474445e-06, + "loss": 2.45664119720459, + "step": 241540 + }, + { + "epoch": 0.6575745202123315, + "grad_norm": 2.1212823390960693, + "learning_rate": 6.85789832356549e-06, + "loss": 2.453074645996094, + "step": 241560 + }, + { + "epoch": 0.657628964203076, + "grad_norm": 2.2990522384643555, + "learning_rate": 6.856807959656536e-06, + "loss": 2.4623159408569335, + "step": 241580 + }, + { + "epoch": 0.6576834081938207, + "grad_norm": 2.230309247970581, + "learning_rate": 6.855717595747581e-06, + "loss": 2.4828765869140623, + "step": 241600 + }, + { + "epoch": 0.6577378521845652, + "grad_norm": 2.23927640914917, + "learning_rate": 6.854627231838627e-06, + "loss": 2.5064006805419923, + "step": 241620 + }, + { + "epoch": 0.6577922961753097, + "grad_norm": 2.165876626968384, + "learning_rate": 6.853536867929672e-06, + "loss": 2.4598012924194337, + "step": 241640 + }, + { + "epoch": 0.6578467401660542, + "grad_norm": 2.1306872367858887, + "learning_rate": 6.8524465040207175e-06, + "loss": 2.4390077590942383, + "step": 241660 + }, + { + "epoch": 0.6579011841567987, + "grad_norm": 2.457263231277466, + "learning_rate": 6.8513561401117625e-06, + "loss": 2.476422119140625, + "step": 241680 + }, + { + "epoch": 0.6579556281475433, + "grad_norm": 2.2951159477233887, + "learning_rate": 6.850265776202808e-06, + "loss": 2.469248390197754, + "step": 241700 + }, + { + "epoch": 0.6580100721382878, + "grad_norm": 2.901761531829834, + "learning_rate": 6.849175412293853e-06, + "loss": 2.4690568923950194, + "step": 241720 + }, + { + "epoch": 0.6580645161290323, + "grad_norm": 2.0976152420043945, + "learning_rate": 6.848085048384899e-06, + "loss": 2.481166648864746, + "step": 241740 + }, + { + "epoch": 0.6581189601197768, + "grad_norm": 2.1749963760375977, + "learning_rate": 6.846994684475945e-06, + "loss": 2.483287239074707, + "step": 241760 + }, + { + "epoch": 0.6581734041105213, + "grad_norm": 2.1279077529907227, + "learning_rate": 6.84590432056699e-06, + "loss": 2.4843090057373045, + "step": 241780 + }, + { + "epoch": 0.6582278481012658, + "grad_norm": 2.175769090652466, + "learning_rate": 6.844813956658035e-06, + "loss": 2.4434368133544924, + "step": 241800 + }, + { + "epoch": 0.6582822920920104, + "grad_norm": 2.2021188735961914, + "learning_rate": 6.843723592749081e-06, + "loss": 2.4817441940307616, + "step": 241820 + }, + { + "epoch": 0.6583367360827549, + "grad_norm": 2.0868561267852783, + "learning_rate": 6.8426332288401264e-06, + "loss": 2.4578269958496093, + "step": 241840 + }, + { + "epoch": 0.6583911800734994, + "grad_norm": 2.769670009613037, + "learning_rate": 6.841542864931171e-06, + "loss": 2.4344778060913086, + "step": 241860 + }, + { + "epoch": 0.6584456240642439, + "grad_norm": 2.208425521850586, + "learning_rate": 6.840452501022216e-06, + "loss": 2.481387901306152, + "step": 241880 + }, + { + "epoch": 0.6585000680549884, + "grad_norm": 2.2205138206481934, + "learning_rate": 6.839362137113262e-06, + "loss": 2.500909996032715, + "step": 241900 + }, + { + "epoch": 0.6585545120457329, + "grad_norm": 2.221606969833374, + "learning_rate": 6.838271773204308e-06, + "loss": 2.4933395385742188, + "step": 241920 + }, + { + "epoch": 0.6586089560364775, + "grad_norm": 2.2678096294403076, + "learning_rate": 6.837181409295353e-06, + "loss": 2.487871551513672, + "step": 241940 + }, + { + "epoch": 0.658663400027222, + "grad_norm": 2.162005662918091, + "learning_rate": 6.836091045386398e-06, + "loss": 2.484480094909668, + "step": 241960 + }, + { + "epoch": 0.6587178440179665, + "grad_norm": 2.5096206665039062, + "learning_rate": 6.835000681477444e-06, + "loss": 2.4707416534423827, + "step": 241980 + }, + { + "epoch": 0.658772288008711, + "grad_norm": 2.1588528156280518, + "learning_rate": 6.8339103175684895e-06, + "loss": 2.4752500534057615, + "step": 242000 + }, + { + "epoch": 0.6588267319994555, + "grad_norm": 4.142319679260254, + "learning_rate": 6.8328199536595345e-06, + "loss": 2.4627561569213867, + "step": 242020 + }, + { + "epoch": 0.6588811759902001, + "grad_norm": 2.1238925457000732, + "learning_rate": 6.8317295897505794e-06, + "loss": 2.4729145050048826, + "step": 242040 + }, + { + "epoch": 0.6589356199809446, + "grad_norm": 2.2017171382904053, + "learning_rate": 6.830639225841625e-06, + "loss": 2.490605926513672, + "step": 242060 + }, + { + "epoch": 0.6589900639716891, + "grad_norm": 2.1041793823242188, + "learning_rate": 6.829548861932671e-06, + "loss": 2.4619752883911135, + "step": 242080 + }, + { + "epoch": 0.6590445079624336, + "grad_norm": 2.251431941986084, + "learning_rate": 6.828458498023716e-06, + "loss": 2.4669750213623045, + "step": 242100 + }, + { + "epoch": 0.6590989519531781, + "grad_norm": 2.143667459487915, + "learning_rate": 6.827368134114761e-06, + "loss": 2.4796533584594727, + "step": 242120 + }, + { + "epoch": 0.6591533959439226, + "grad_norm": 2.122243642807007, + "learning_rate": 6.826277770205807e-06, + "loss": 2.4484973907470704, + "step": 242140 + }, + { + "epoch": 0.6592078399346673, + "grad_norm": 2.179229974746704, + "learning_rate": 6.825187406296853e-06, + "loss": 2.5078241348266603, + "step": 242160 + }, + { + "epoch": 0.6592622839254118, + "grad_norm": 2.1220359802246094, + "learning_rate": 6.8240970423878976e-06, + "loss": 2.511673164367676, + "step": 242180 + }, + { + "epoch": 0.6593167279161563, + "grad_norm": 2.1943111419677734, + "learning_rate": 6.8230066784789425e-06, + "loss": 2.4561315536499024, + "step": 242200 + }, + { + "epoch": 0.6593711719069008, + "grad_norm": 2.1179678440093994, + "learning_rate": 6.821916314569988e-06, + "loss": 2.4628957748413085, + "step": 242220 + }, + { + "epoch": 0.6594256158976453, + "grad_norm": 2.449565887451172, + "learning_rate": 6.820825950661034e-06, + "loss": 2.4557525634765627, + "step": 242240 + }, + { + "epoch": 0.6594800598883899, + "grad_norm": 2.138389825820923, + "learning_rate": 6.819735586752079e-06, + "loss": 2.4661603927612306, + "step": 242260 + }, + { + "epoch": 0.6595345038791344, + "grad_norm": 2.378748893737793, + "learning_rate": 6.818645222843124e-06, + "loss": 2.429576301574707, + "step": 242280 + }, + { + "epoch": 0.6595889478698789, + "grad_norm": 2.132897138595581, + "learning_rate": 6.81755485893417e-06, + "loss": 2.4841480255126953, + "step": 242300 + }, + { + "epoch": 0.6596433918606234, + "grad_norm": 2.157775640487671, + "learning_rate": 6.816464495025216e-06, + "loss": 2.4948093414306642, + "step": 242320 + }, + { + "epoch": 0.6596978358513679, + "grad_norm": 2.1725285053253174, + "learning_rate": 6.815374131116261e-06, + "loss": 2.4691545486450197, + "step": 242340 + }, + { + "epoch": 0.6597522798421124, + "grad_norm": 2.1754038333892822, + "learning_rate": 6.814283767207306e-06, + "loss": 2.4873424530029298, + "step": 242360 + }, + { + "epoch": 0.659806723832857, + "grad_norm": 2.1424124240875244, + "learning_rate": 6.813193403298351e-06, + "loss": 2.4571201324462892, + "step": 242380 + }, + { + "epoch": 0.6598611678236015, + "grad_norm": 2.248645067214966, + "learning_rate": 6.812103039389397e-06, + "loss": 2.484387969970703, + "step": 242400 + }, + { + "epoch": 0.659915611814346, + "grad_norm": 2.0928702354431152, + "learning_rate": 6.811012675480441e-06, + "loss": 2.4515945434570314, + "step": 242420 + }, + { + "epoch": 0.6599700558050905, + "grad_norm": 2.2009925842285156, + "learning_rate": 6.809922311571487e-06, + "loss": 2.4519153594970704, + "step": 242440 + }, + { + "epoch": 0.660024499795835, + "grad_norm": 2.2387027740478516, + "learning_rate": 6.808831947662533e-06, + "loss": 2.455693817138672, + "step": 242460 + }, + { + "epoch": 0.6600789437865796, + "grad_norm": 2.2457213401794434, + "learning_rate": 6.807741583753579e-06, + "loss": 2.478462982177734, + "step": 242480 + }, + { + "epoch": 0.6601333877773241, + "grad_norm": 2.3305225372314453, + "learning_rate": 6.806651219844623e-06, + "loss": 2.4842811584472657, + "step": 242500 + }, + { + "epoch": 0.6601878317680686, + "grad_norm": 2.194786310195923, + "learning_rate": 6.805560855935669e-06, + "loss": 2.473526191711426, + "step": 242520 + }, + { + "epoch": 0.6602422757588131, + "grad_norm": 2.2579550743103027, + "learning_rate": 6.8044704920267145e-06, + "loss": 2.491230773925781, + "step": 242540 + }, + { + "epoch": 0.6602967197495576, + "grad_norm": 2.247117519378662, + "learning_rate": 6.80338012811776e-06, + "loss": 2.442332077026367, + "step": 242560 + }, + { + "epoch": 0.6603511637403021, + "grad_norm": 2.3671271800994873, + "learning_rate": 6.8022897642088044e-06, + "loss": 2.4679544448852537, + "step": 242580 + }, + { + "epoch": 0.6604056077310467, + "grad_norm": 2.2674777507781982, + "learning_rate": 6.80119940029985e-06, + "loss": 2.493411827087402, + "step": 242600 + }, + { + "epoch": 0.6604600517217912, + "grad_norm": 2.2747302055358887, + "learning_rate": 6.800109036390896e-06, + "loss": 2.4494182586669924, + "step": 242620 + }, + { + "epoch": 0.6605144957125357, + "grad_norm": 2.1999940872192383, + "learning_rate": 6.799018672481942e-06, + "loss": 2.4739107131958007, + "step": 242640 + }, + { + "epoch": 0.6605689397032802, + "grad_norm": 2.164489269256592, + "learning_rate": 6.797928308572986e-06, + "loss": 2.462371826171875, + "step": 242660 + }, + { + "epoch": 0.6606233836940247, + "grad_norm": 2.202578067779541, + "learning_rate": 6.796837944664032e-06, + "loss": 2.4772964477539063, + "step": 242680 + }, + { + "epoch": 0.6606778276847693, + "grad_norm": 2.205087184906006, + "learning_rate": 6.795747580755078e-06, + "loss": 2.490778923034668, + "step": 242700 + }, + { + "epoch": 0.6607322716755138, + "grad_norm": 2.8765923976898193, + "learning_rate": 6.794657216846123e-06, + "loss": 2.453616905212402, + "step": 242720 + }, + { + "epoch": 0.6607867156662584, + "grad_norm": 2.3052010536193848, + "learning_rate": 6.7935668529371675e-06, + "loss": 2.4785541534423827, + "step": 242740 + }, + { + "epoch": 0.6608411596570029, + "grad_norm": 2.2205471992492676, + "learning_rate": 6.792476489028213e-06, + "loss": 2.4625301361083984, + "step": 242760 + }, + { + "epoch": 0.6608956036477474, + "grad_norm": 2.0418946743011475, + "learning_rate": 6.791386125119259e-06, + "loss": 2.466537666320801, + "step": 242780 + }, + { + "epoch": 0.6609500476384919, + "grad_norm": 2.2656872272491455, + "learning_rate": 6.790295761210305e-06, + "loss": 2.4845008850097656, + "step": 242800 + }, + { + "epoch": 0.6610044916292365, + "grad_norm": 2.1869747638702393, + "learning_rate": 6.789205397301349e-06, + "loss": 2.485622787475586, + "step": 242820 + }, + { + "epoch": 0.661058935619981, + "grad_norm": 2.1907010078430176, + "learning_rate": 6.788115033392395e-06, + "loss": 2.4703609466552736, + "step": 242840 + }, + { + "epoch": 0.6611133796107255, + "grad_norm": 2.2134697437286377, + "learning_rate": 6.787024669483441e-06, + "loss": 2.4584125518798827, + "step": 242860 + }, + { + "epoch": 0.66116782360147, + "grad_norm": 2.2356154918670654, + "learning_rate": 6.7859343055744865e-06, + "loss": 2.5043224334716796, + "step": 242880 + }, + { + "epoch": 0.6612222675922145, + "grad_norm": 2.4844753742218018, + "learning_rate": 6.784843941665531e-06, + "loss": 2.451224136352539, + "step": 242900 + }, + { + "epoch": 0.661276711582959, + "grad_norm": 2.1445090770721436, + "learning_rate": 6.783753577756576e-06, + "loss": 2.4616756439208984, + "step": 242920 + }, + { + "epoch": 0.6613311555737036, + "grad_norm": 2.949293375015259, + "learning_rate": 6.782663213847622e-06, + "loss": 2.4740537643432616, + "step": 242940 + }, + { + "epoch": 0.6613855995644481, + "grad_norm": 2.2065465450286865, + "learning_rate": 6.781572849938668e-06, + "loss": 2.4683698654174804, + "step": 242960 + }, + { + "epoch": 0.6614400435551926, + "grad_norm": 2.181830883026123, + "learning_rate": 6.780482486029712e-06, + "loss": 2.439832878112793, + "step": 242980 + }, + { + "epoch": 0.6614944875459371, + "grad_norm": 2.143939733505249, + "learning_rate": 6.779392122120758e-06, + "loss": 2.482255554199219, + "step": 243000 + }, + { + "epoch": 0.6615489315366816, + "grad_norm": 2.261340379714966, + "learning_rate": 6.778301758211804e-06, + "loss": 2.477085494995117, + "step": 243020 + }, + { + "epoch": 0.6616033755274262, + "grad_norm": 2.1735801696777344, + "learning_rate": 6.7772113943028496e-06, + "loss": 2.5038543701171876, + "step": 243040 + }, + { + "epoch": 0.6616578195181707, + "grad_norm": 2.11099910736084, + "learning_rate": 6.776121030393894e-06, + "loss": 2.4536741256713865, + "step": 243060 + }, + { + "epoch": 0.6617122635089152, + "grad_norm": 2.2228877544403076, + "learning_rate": 6.7750306664849395e-06, + "loss": 2.4581375122070312, + "step": 243080 + }, + { + "epoch": 0.6617667074996597, + "grad_norm": 3.1836729049682617, + "learning_rate": 6.773940302575985e-06, + "loss": 2.437276840209961, + "step": 243100 + }, + { + "epoch": 0.6618211514904042, + "grad_norm": 2.14003586769104, + "learning_rate": 6.772849938667031e-06, + "loss": 2.476119613647461, + "step": 243120 + }, + { + "epoch": 0.6618755954811487, + "grad_norm": 2.104837417602539, + "learning_rate": 6.771759574758075e-06, + "loss": 2.4528247833251955, + "step": 243140 + }, + { + "epoch": 0.6619300394718933, + "grad_norm": 2.087573528289795, + "learning_rate": 6.770669210849121e-06, + "loss": 2.49672908782959, + "step": 243160 + }, + { + "epoch": 0.6619844834626378, + "grad_norm": 2.175617218017578, + "learning_rate": 6.769578846940167e-06, + "loss": 2.4977432250976563, + "step": 243180 + }, + { + "epoch": 0.6620389274533823, + "grad_norm": 2.075895309448242, + "learning_rate": 6.768488483031213e-06, + "loss": 2.4531320571899413, + "step": 243200 + }, + { + "epoch": 0.6620933714441268, + "grad_norm": 2.081122636795044, + "learning_rate": 6.767398119122257e-06, + "loss": 2.4371564865112303, + "step": 243220 + }, + { + "epoch": 0.6621478154348713, + "grad_norm": 2.1221818923950195, + "learning_rate": 6.7663077552133026e-06, + "loss": 2.465083694458008, + "step": 243240 + }, + { + "epoch": 0.662202259425616, + "grad_norm": 2.2840964794158936, + "learning_rate": 6.765217391304348e-06, + "loss": 2.481610107421875, + "step": 243260 + }, + { + "epoch": 0.6622567034163604, + "grad_norm": 2.1295156478881836, + "learning_rate": 6.764127027395394e-06, + "loss": 2.46563777923584, + "step": 243280 + }, + { + "epoch": 0.662311147407105, + "grad_norm": 2.180095672607422, + "learning_rate": 6.763036663486438e-06, + "loss": 2.4575874328613283, + "step": 243300 + }, + { + "epoch": 0.6623655913978495, + "grad_norm": 2.1245813369750977, + "learning_rate": 6.761946299577484e-06, + "loss": 2.4734052658081054, + "step": 243320 + }, + { + "epoch": 0.662420035388594, + "grad_norm": 2.1750454902648926, + "learning_rate": 6.76085593566853e-06, + "loss": 2.46466178894043, + "step": 243340 + }, + { + "epoch": 0.6624744793793385, + "grad_norm": 2.257809638977051, + "learning_rate": 6.759765571759576e-06, + "loss": 2.452814483642578, + "step": 243360 + }, + { + "epoch": 0.6625289233700831, + "grad_norm": 2.141688823699951, + "learning_rate": 6.75867520785062e-06, + "loss": 2.4503406524658202, + "step": 243380 + }, + { + "epoch": 0.6625833673608276, + "grad_norm": 2.1617536544799805, + "learning_rate": 6.757584843941666e-06, + "loss": 2.458536720275879, + "step": 243400 + }, + { + "epoch": 0.6626378113515721, + "grad_norm": 2.1492068767547607, + "learning_rate": 6.7564944800327115e-06, + "loss": 2.4518619537353517, + "step": 243420 + }, + { + "epoch": 0.6626922553423166, + "grad_norm": 2.1115236282348633, + "learning_rate": 6.755404116123757e-06, + "loss": 2.4403234481811524, + "step": 243440 + }, + { + "epoch": 0.6627466993330611, + "grad_norm": 2.211451530456543, + "learning_rate": 6.754313752214803e-06, + "loss": 2.474873352050781, + "step": 243460 + }, + { + "epoch": 0.6628011433238057, + "grad_norm": 2.2319536209106445, + "learning_rate": 6.753223388305847e-06, + "loss": 2.4872314453125, + "step": 243480 + }, + { + "epoch": 0.6628555873145502, + "grad_norm": 2.1500167846679688, + "learning_rate": 6.752133024396893e-06, + "loss": 2.483646011352539, + "step": 243500 + }, + { + "epoch": 0.6629100313052947, + "grad_norm": 2.1903293132781982, + "learning_rate": 6.751042660487939e-06, + "loss": 2.453011894226074, + "step": 243520 + }, + { + "epoch": 0.6629644752960392, + "grad_norm": 2.1616125106811523, + "learning_rate": 6.749952296578985e-06, + "loss": 2.444754981994629, + "step": 243540 + }, + { + "epoch": 0.6630189192867837, + "grad_norm": 3.0150630474090576, + "learning_rate": 6.748861932670029e-06, + "loss": 2.4598201751708983, + "step": 243560 + }, + { + "epoch": 0.6630733632775282, + "grad_norm": 2.316394090652466, + "learning_rate": 6.7477715687610746e-06, + "loss": 2.4597909927368162, + "step": 243580 + }, + { + "epoch": 0.6631278072682728, + "grad_norm": 2.133174419403076, + "learning_rate": 6.74668120485212e-06, + "loss": 2.4566858291625975, + "step": 243600 + }, + { + "epoch": 0.6631822512590173, + "grad_norm": 2.1321864128112793, + "learning_rate": 6.745590840943166e-06, + "loss": 2.4372507095336915, + "step": 243620 + }, + { + "epoch": 0.6632366952497618, + "grad_norm": 2.153871536254883, + "learning_rate": 6.74450047703421e-06, + "loss": 2.4546476364135743, + "step": 243640 + }, + { + "epoch": 0.6632911392405063, + "grad_norm": 2.139244318008423, + "learning_rate": 6.743410113125256e-06, + "loss": 2.5040304183959963, + "step": 243660 + }, + { + "epoch": 0.6633455832312508, + "grad_norm": 2.2373995780944824, + "learning_rate": 6.742319749216302e-06, + "loss": 2.471680450439453, + "step": 243680 + }, + { + "epoch": 0.6634000272219954, + "grad_norm": 2.1063809394836426, + "learning_rate": 6.741229385307348e-06, + "loss": 2.478608322143555, + "step": 243700 + }, + { + "epoch": 0.6634544712127399, + "grad_norm": 2.1870996952056885, + "learning_rate": 6.740139021398392e-06, + "loss": 2.4755908966064455, + "step": 243720 + }, + { + "epoch": 0.6635089152034844, + "grad_norm": 2.2262439727783203, + "learning_rate": 6.739048657489438e-06, + "loss": 2.462149810791016, + "step": 243740 + }, + { + "epoch": 0.6635633591942289, + "grad_norm": 2.2731645107269287, + "learning_rate": 6.7379582935804835e-06, + "loss": 2.433245849609375, + "step": 243760 + }, + { + "epoch": 0.6636178031849734, + "grad_norm": 2.117919683456421, + "learning_rate": 6.736867929671529e-06, + "loss": 2.467863655090332, + "step": 243780 + }, + { + "epoch": 0.6636722471757179, + "grad_norm": 2.601109743118286, + "learning_rate": 6.735777565762573e-06, + "loss": 2.4702239990234376, + "step": 243800 + }, + { + "epoch": 0.6637266911664625, + "grad_norm": 2.1683967113494873, + "learning_rate": 6.734687201853619e-06, + "loss": 2.458150291442871, + "step": 243820 + }, + { + "epoch": 0.663781135157207, + "grad_norm": 2.2595598697662354, + "learning_rate": 6.733596837944665e-06, + "loss": 2.488543701171875, + "step": 243840 + }, + { + "epoch": 0.6638355791479515, + "grad_norm": 2.229240655899048, + "learning_rate": 6.732506474035711e-06, + "loss": 2.465453338623047, + "step": 243860 + }, + { + "epoch": 0.663890023138696, + "grad_norm": 2.2275354862213135, + "learning_rate": 6.731416110126755e-06, + "loss": 2.469422149658203, + "step": 243880 + }, + { + "epoch": 0.6639444671294406, + "grad_norm": 2.1675283908843994, + "learning_rate": 6.730325746217801e-06, + "loss": 2.48233585357666, + "step": 243900 + }, + { + "epoch": 0.6639989111201852, + "grad_norm": 2.1970419883728027, + "learning_rate": 6.7292353823088465e-06, + "loss": 2.426266098022461, + "step": 243920 + }, + { + "epoch": 0.6640533551109297, + "grad_norm": 2.128718852996826, + "learning_rate": 6.728145018399892e-06, + "loss": 2.4682077407836913, + "step": 243940 + }, + { + "epoch": 0.6641077991016742, + "grad_norm": 2.271103858947754, + "learning_rate": 6.7270546544909365e-06, + "loss": 2.4885843276977537, + "step": 243960 + }, + { + "epoch": 0.6641622430924187, + "grad_norm": 2.098546266555786, + "learning_rate": 6.725964290581982e-06, + "loss": 2.454738807678223, + "step": 243980 + }, + { + "epoch": 0.6642166870831632, + "grad_norm": 2.15171480178833, + "learning_rate": 6.724873926673028e-06, + "loss": 2.488551139831543, + "step": 244000 + }, + { + "epoch": 0.6642711310739077, + "grad_norm": 2.3108761310577393, + "learning_rate": 6.723783562764074e-06, + "loss": 2.4336191177368165, + "step": 244020 + }, + { + "epoch": 0.6643255750646523, + "grad_norm": 2.04451322555542, + "learning_rate": 6.722693198855118e-06, + "loss": 2.4644689559936523, + "step": 244040 + }, + { + "epoch": 0.6643800190553968, + "grad_norm": 2.2314441204071045, + "learning_rate": 6.721602834946164e-06, + "loss": 2.506754684448242, + "step": 244060 + }, + { + "epoch": 0.6644344630461413, + "grad_norm": 2.1542837619781494, + "learning_rate": 6.72051247103721e-06, + "loss": 2.4415102005004883, + "step": 244080 + }, + { + "epoch": 0.6644889070368858, + "grad_norm": 2.1865155696868896, + "learning_rate": 6.7194221071282554e-06, + "loss": 2.497178649902344, + "step": 244100 + }, + { + "epoch": 0.6645433510276303, + "grad_norm": 2.082562208175659, + "learning_rate": 6.7183317432192995e-06, + "loss": 2.4946010589599608, + "step": 244120 + }, + { + "epoch": 0.6645977950183748, + "grad_norm": 2.1149075031280518, + "learning_rate": 6.717241379310345e-06, + "loss": 2.479905891418457, + "step": 244140 + }, + { + "epoch": 0.6646522390091194, + "grad_norm": 2.137515068054199, + "learning_rate": 6.716151015401391e-06, + "loss": 2.4570566177368165, + "step": 244160 + }, + { + "epoch": 0.6647066829998639, + "grad_norm": 2.3381426334381104, + "learning_rate": 6.715060651492437e-06, + "loss": 2.469348907470703, + "step": 244180 + }, + { + "epoch": 0.6647611269906084, + "grad_norm": 2.1230080127716064, + "learning_rate": 6.713970287583481e-06, + "loss": 2.4745548248291014, + "step": 244200 + }, + { + "epoch": 0.6648155709813529, + "grad_norm": 2.1494510173797607, + "learning_rate": 6.712879923674527e-06, + "loss": 2.485565948486328, + "step": 244220 + }, + { + "epoch": 0.6648700149720974, + "grad_norm": 2.3872056007385254, + "learning_rate": 6.711789559765573e-06, + "loss": 2.4712547302246093, + "step": 244240 + }, + { + "epoch": 0.664924458962842, + "grad_norm": 2.1440138816833496, + "learning_rate": 6.7106991958566185e-06, + "loss": 2.5077434539794923, + "step": 244260 + }, + { + "epoch": 0.6649789029535865, + "grad_norm": 2.0652401447296143, + "learning_rate": 6.709608831947663e-06, + "loss": 2.434395980834961, + "step": 244280 + }, + { + "epoch": 0.665033346944331, + "grad_norm": 2.352877140045166, + "learning_rate": 6.7085184680387084e-06, + "loss": 2.460641860961914, + "step": 244300 + }, + { + "epoch": 0.6650877909350755, + "grad_norm": 2.1421074867248535, + "learning_rate": 6.707428104129754e-06, + "loss": 2.4504112243652343, + "step": 244320 + }, + { + "epoch": 0.66514223492582, + "grad_norm": 2.1633493900299072, + "learning_rate": 6.7063377402208e-06, + "loss": 2.465115547180176, + "step": 244340 + }, + { + "epoch": 0.6651966789165645, + "grad_norm": 2.166914939880371, + "learning_rate": 6.705247376311844e-06, + "loss": 2.4700796127319338, + "step": 244360 + }, + { + "epoch": 0.6652511229073091, + "grad_norm": 2.2202062606811523, + "learning_rate": 6.70415701240289e-06, + "loss": 2.4702110290527344, + "step": 244380 + }, + { + "epoch": 0.6653055668980536, + "grad_norm": 2.1717278957366943, + "learning_rate": 6.703066648493936e-06, + "loss": 2.46016902923584, + "step": 244400 + }, + { + "epoch": 0.6653600108887981, + "grad_norm": 2.3167474269866943, + "learning_rate": 6.701976284584981e-06, + "loss": 2.4829875946044924, + "step": 244420 + }, + { + "epoch": 0.6654144548795426, + "grad_norm": 2.4131908416748047, + "learning_rate": 6.700885920676026e-06, + "loss": 2.4855457305908204, + "step": 244440 + }, + { + "epoch": 0.6654688988702872, + "grad_norm": 2.189479112625122, + "learning_rate": 6.6997955567670715e-06, + "loss": 2.492545318603516, + "step": 244460 + }, + { + "epoch": 0.6655233428610318, + "grad_norm": 2.3316383361816406, + "learning_rate": 6.698705192858117e-06, + "loss": 2.4605243682861326, + "step": 244480 + }, + { + "epoch": 0.6655777868517763, + "grad_norm": 2.1780989170074463, + "learning_rate": 6.697614828949162e-06, + "loss": 2.4722898483276365, + "step": 244500 + }, + { + "epoch": 0.6656322308425208, + "grad_norm": 2.0416297912597656, + "learning_rate": 6.696524465040207e-06, + "loss": 2.471494674682617, + "step": 244520 + }, + { + "epoch": 0.6656866748332653, + "grad_norm": 2.2285244464874268, + "learning_rate": 6.695434101131253e-06, + "loss": 2.476474952697754, + "step": 244540 + }, + { + "epoch": 0.6657411188240098, + "grad_norm": 2.158590078353882, + "learning_rate": 6.694343737222299e-06, + "loss": 2.4475101470947265, + "step": 244560 + }, + { + "epoch": 0.6657955628147543, + "grad_norm": 2.2639222145080566, + "learning_rate": 6.693253373313344e-06, + "loss": 2.4787853240966795, + "step": 244580 + }, + { + "epoch": 0.6658500068054989, + "grad_norm": 4.040594577789307, + "learning_rate": 6.692163009404389e-06, + "loss": 2.4744121551513674, + "step": 244600 + }, + { + "epoch": 0.6659044507962434, + "grad_norm": 2.241590976715088, + "learning_rate": 6.691072645495435e-06, + "loss": 2.4956792831420898, + "step": 244620 + }, + { + "epoch": 0.6659588947869879, + "grad_norm": 2.0865578651428223, + "learning_rate": 6.68998228158648e-06, + "loss": 2.474989128112793, + "step": 244640 + }, + { + "epoch": 0.6660133387777324, + "grad_norm": 2.884155511856079, + "learning_rate": 6.688891917677525e-06, + "loss": 2.4406957626342773, + "step": 244660 + }, + { + "epoch": 0.6660677827684769, + "grad_norm": 2.1851956844329834, + "learning_rate": 6.68780155376857e-06, + "loss": 2.433457374572754, + "step": 244680 + }, + { + "epoch": 0.6661222267592215, + "grad_norm": 2.2228002548217773, + "learning_rate": 6.686711189859616e-06, + "loss": 2.4574302673339843, + "step": 244700 + }, + { + "epoch": 0.666176670749966, + "grad_norm": 2.141082763671875, + "learning_rate": 6.685620825950662e-06, + "loss": 2.479276657104492, + "step": 244720 + }, + { + "epoch": 0.6662311147407105, + "grad_norm": 2.264219045639038, + "learning_rate": 6.684530462041707e-06, + "loss": 2.4822072982788086, + "step": 244740 + }, + { + "epoch": 0.666285558731455, + "grad_norm": 2.158721923828125, + "learning_rate": 6.683440098132752e-06, + "loss": 2.4480239868164064, + "step": 244760 + }, + { + "epoch": 0.6663400027221995, + "grad_norm": 2.0978219509124756, + "learning_rate": 6.682349734223798e-06, + "loss": 2.482442283630371, + "step": 244780 + }, + { + "epoch": 0.666394446712944, + "grad_norm": 2.379922389984131, + "learning_rate": 6.6812593703148435e-06, + "loss": 2.4762704849243162, + "step": 244800 + }, + { + "epoch": 0.6664488907036886, + "grad_norm": 2.3917412757873535, + "learning_rate": 6.6801690064058885e-06, + "loss": 2.4537675857543944, + "step": 244820 + }, + { + "epoch": 0.6665033346944331, + "grad_norm": 2.156283140182495, + "learning_rate": 6.6790786424969334e-06, + "loss": 2.4696985244750977, + "step": 244840 + }, + { + "epoch": 0.6665577786851776, + "grad_norm": 2.1164965629577637, + "learning_rate": 6.677988278587979e-06, + "loss": 2.4704557418823243, + "step": 244860 + }, + { + "epoch": 0.6666122226759221, + "grad_norm": 2.193084239959717, + "learning_rate": 6.676897914679025e-06, + "loss": 2.479136657714844, + "step": 244880 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5716147422790527, + "learning_rate": 6.67580755077007e-06, + "loss": 2.4812320709228515, + "step": 244900 + }, + { + "epoch": 0.6667211106574112, + "grad_norm": 2.1732497215270996, + "learning_rate": 6.674717186861115e-06, + "loss": 2.4668102264404297, + "step": 244920 + }, + { + "epoch": 0.6667755546481557, + "grad_norm": 2.099013328552246, + "learning_rate": 6.673626822952161e-06, + "loss": 2.4457860946655274, + "step": 244940 + }, + { + "epoch": 0.6668299986389002, + "grad_norm": 2.1250362396240234, + "learning_rate": 6.672536459043207e-06, + "loss": 2.4465465545654297, + "step": 244960 + }, + { + "epoch": 0.6668844426296447, + "grad_norm": 2.242255687713623, + "learning_rate": 6.6714460951342516e-06, + "loss": 2.4731136322021485, + "step": 244980 + }, + { + "epoch": 0.6669388866203892, + "grad_norm": 2.15645694732666, + "learning_rate": 6.6703557312252965e-06, + "loss": 2.4542091369628904, + "step": 245000 + }, + { + "epoch": 0.6669933306111338, + "grad_norm": 2.289368152618408, + "learning_rate": 6.669265367316342e-06, + "loss": 2.4836219787597655, + "step": 245020 + }, + { + "epoch": 0.6670477746018784, + "grad_norm": 2.2193825244903564, + "learning_rate": 6.668175003407388e-06, + "loss": 2.4699031829833986, + "step": 245040 + }, + { + "epoch": 0.6671022185926229, + "grad_norm": 2.170742988586426, + "learning_rate": 6.667084639498433e-06, + "loss": 2.4783098220825197, + "step": 245060 + }, + { + "epoch": 0.6671566625833674, + "grad_norm": 2.1100897789001465, + "learning_rate": 6.665994275589478e-06, + "loss": 2.4672021865844727, + "step": 245080 + }, + { + "epoch": 0.6672111065741119, + "grad_norm": 2.1760520935058594, + "learning_rate": 6.664903911680524e-06, + "loss": 2.484719657897949, + "step": 245100 + }, + { + "epoch": 0.6672655505648564, + "grad_norm": 2.7047791481018066, + "learning_rate": 6.66381354777157e-06, + "loss": 2.4813377380371096, + "step": 245120 + }, + { + "epoch": 0.6673199945556009, + "grad_norm": 2.1379830837249756, + "learning_rate": 6.662723183862615e-06, + "loss": 2.4601676940917967, + "step": 245140 + }, + { + "epoch": 0.6673744385463455, + "grad_norm": 2.200573682785034, + "learning_rate": 6.6616328199536605e-06, + "loss": 2.481377601623535, + "step": 245160 + }, + { + "epoch": 0.66742888253709, + "grad_norm": 2.300245523452759, + "learning_rate": 6.660542456044705e-06, + "loss": 2.47430477142334, + "step": 245180 + }, + { + "epoch": 0.6674833265278345, + "grad_norm": 2.1246964931488037, + "learning_rate": 6.659452092135751e-06, + "loss": 2.4268112182617188, + "step": 245200 + }, + { + "epoch": 0.667537770518579, + "grad_norm": 2.2560508251190186, + "learning_rate": 6.658361728226796e-06, + "loss": 2.4611185073852537, + "step": 245220 + }, + { + "epoch": 0.6675922145093235, + "grad_norm": 2.310204267501831, + "learning_rate": 6.657271364317842e-06, + "loss": 2.4695276260375976, + "step": 245240 + }, + { + "epoch": 0.6676466585000681, + "grad_norm": 2.083036422729492, + "learning_rate": 6.656181000408887e-06, + "loss": 2.4667057037353515, + "step": 245260 + }, + { + "epoch": 0.6677011024908126, + "grad_norm": 2.2104086875915527, + "learning_rate": 6.655090636499933e-06, + "loss": 2.455897331237793, + "step": 245280 + }, + { + "epoch": 0.6677555464815571, + "grad_norm": 2.075360059738159, + "learning_rate": 6.654000272590978e-06, + "loss": 2.456416130065918, + "step": 245300 + }, + { + "epoch": 0.6678099904723016, + "grad_norm": 2.2780745029449463, + "learning_rate": 6.6529099086820235e-06, + "loss": 2.4904550552368163, + "step": 245320 + }, + { + "epoch": 0.6678644344630461, + "grad_norm": 2.180687427520752, + "learning_rate": 6.6518195447730685e-06, + "loss": 2.4761606216430665, + "step": 245340 + }, + { + "epoch": 0.6679188784537906, + "grad_norm": 2.3201546669006348, + "learning_rate": 6.650729180864114e-06, + "loss": 2.451436996459961, + "step": 245360 + }, + { + "epoch": 0.6679733224445352, + "grad_norm": 2.3671371936798096, + "learning_rate": 6.649638816955159e-06, + "loss": 2.457667350769043, + "step": 245380 + }, + { + "epoch": 0.6680277664352797, + "grad_norm": 2.101468324661255, + "learning_rate": 6.648548453046205e-06, + "loss": 2.4639244079589844, + "step": 245400 + }, + { + "epoch": 0.6680822104260242, + "grad_norm": 2.1491074562072754, + "learning_rate": 6.64745808913725e-06, + "loss": 2.477705955505371, + "step": 245420 + }, + { + "epoch": 0.6681366544167687, + "grad_norm": 2.291306972503662, + "learning_rate": 6.646367725228296e-06, + "loss": 2.466878318786621, + "step": 245440 + }, + { + "epoch": 0.6681910984075132, + "grad_norm": 2.3472328186035156, + "learning_rate": 6.645277361319341e-06, + "loss": 2.458210563659668, + "step": 245460 + }, + { + "epoch": 0.6682455423982578, + "grad_norm": 2.24768328666687, + "learning_rate": 6.644186997410387e-06, + "loss": 2.4632131576538088, + "step": 245480 + }, + { + "epoch": 0.6682999863890023, + "grad_norm": 2.1114726066589355, + "learning_rate": 6.643096633501432e-06, + "loss": 2.481843185424805, + "step": 245500 + }, + { + "epoch": 0.6683544303797468, + "grad_norm": 2.1249144077301025, + "learning_rate": 6.6420062695924765e-06, + "loss": 2.4577075958251955, + "step": 245520 + }, + { + "epoch": 0.6684088743704913, + "grad_norm": 2.2598183155059814, + "learning_rate": 6.640915905683522e-06, + "loss": 2.456808662414551, + "step": 245540 + }, + { + "epoch": 0.6684633183612358, + "grad_norm": 2.2469024658203125, + "learning_rate": 6.639825541774568e-06, + "loss": 2.4947032928466797, + "step": 245560 + }, + { + "epoch": 0.6685177623519803, + "grad_norm": 2.1940510272979736, + "learning_rate": 6.638735177865613e-06, + "loss": 2.4525783538818358, + "step": 245580 + }, + { + "epoch": 0.668572206342725, + "grad_norm": 2.6276416778564453, + "learning_rate": 6.637644813956658e-06, + "loss": 2.4700841903686523, + "step": 245600 + }, + { + "epoch": 0.6686266503334695, + "grad_norm": 2.2426793575286865, + "learning_rate": 6.636554450047704e-06, + "loss": 2.4868139266967773, + "step": 245620 + }, + { + "epoch": 0.668681094324214, + "grad_norm": 2.3923449516296387, + "learning_rate": 6.63546408613875e-06, + "loss": 2.477323532104492, + "step": 245640 + }, + { + "epoch": 0.6687355383149585, + "grad_norm": 2.150285482406616, + "learning_rate": 6.634373722229795e-06, + "loss": 2.463764190673828, + "step": 245660 + }, + { + "epoch": 0.668789982305703, + "grad_norm": 2.2036917209625244, + "learning_rate": 6.63328335832084e-06, + "loss": 2.476217269897461, + "step": 245680 + }, + { + "epoch": 0.6688444262964476, + "grad_norm": 2.1631312370300293, + "learning_rate": 6.6321929944118854e-06, + "loss": 2.4684720993041993, + "step": 245700 + }, + { + "epoch": 0.6688988702871921, + "grad_norm": 2.2275869846343994, + "learning_rate": 6.631102630502931e-06, + "loss": 2.478368949890137, + "step": 245720 + }, + { + "epoch": 0.6689533142779366, + "grad_norm": 2.345223903656006, + "learning_rate": 6.630012266593976e-06, + "loss": 2.4809476852416994, + "step": 245740 + }, + { + "epoch": 0.6690077582686811, + "grad_norm": 2.193833589553833, + "learning_rate": 6.628921902685021e-06, + "loss": 2.4745073318481445, + "step": 245760 + }, + { + "epoch": 0.6690622022594256, + "grad_norm": 2.328796863555908, + "learning_rate": 6.627831538776067e-06, + "loss": 2.4902664184570313, + "step": 245780 + }, + { + "epoch": 0.6691166462501701, + "grad_norm": 2.1417744159698486, + "learning_rate": 6.626741174867113e-06, + "loss": 2.4991582870483398, + "step": 245800 + }, + { + "epoch": 0.6691710902409147, + "grad_norm": 2.160670518875122, + "learning_rate": 6.625650810958158e-06, + "loss": 2.4632850646972657, + "step": 245820 + }, + { + "epoch": 0.6692255342316592, + "grad_norm": 2.153722047805786, + "learning_rate": 6.624560447049203e-06, + "loss": 2.4837167739868162, + "step": 245840 + }, + { + "epoch": 0.6692799782224037, + "grad_norm": 2.05780029296875, + "learning_rate": 6.6234700831402485e-06, + "loss": 2.4831445693969725, + "step": 245860 + }, + { + "epoch": 0.6693344222131482, + "grad_norm": 2.0759525299072266, + "learning_rate": 6.622379719231294e-06, + "loss": 2.4316612243652345, + "step": 245880 + }, + { + "epoch": 0.6693888662038927, + "grad_norm": 2.2118260860443115, + "learning_rate": 6.621289355322339e-06, + "loss": 2.4745588302612305, + "step": 245900 + }, + { + "epoch": 0.6694433101946373, + "grad_norm": 2.1947782039642334, + "learning_rate": 6.620198991413384e-06, + "loss": 2.468369483947754, + "step": 245920 + }, + { + "epoch": 0.6694977541853818, + "grad_norm": 2.425834894180298, + "learning_rate": 6.61910862750443e-06, + "loss": 2.447012710571289, + "step": 245940 + }, + { + "epoch": 0.6695521981761263, + "grad_norm": 2.2459583282470703, + "learning_rate": 6.618018263595476e-06, + "loss": 2.482060432434082, + "step": 245960 + }, + { + "epoch": 0.6696066421668708, + "grad_norm": 2.13765811920166, + "learning_rate": 6.616927899686521e-06, + "loss": 2.479650115966797, + "step": 245980 + }, + { + "epoch": 0.6696610861576153, + "grad_norm": 2.110079288482666, + "learning_rate": 6.615837535777566e-06, + "loss": 2.472398567199707, + "step": 246000 + }, + { + "epoch": 0.6697155301483598, + "grad_norm": 2.172645330429077, + "learning_rate": 6.614747171868612e-06, + "loss": 2.467373275756836, + "step": 246020 + }, + { + "epoch": 0.6697699741391044, + "grad_norm": 2.3630356788635254, + "learning_rate": 6.613656807959657e-06, + "loss": 2.4895429611206055, + "step": 246040 + }, + { + "epoch": 0.6698244181298489, + "grad_norm": 2.236602783203125, + "learning_rate": 6.612566444050702e-06, + "loss": 2.464117240905762, + "step": 246060 + }, + { + "epoch": 0.6698788621205934, + "grad_norm": 2.1763601303100586, + "learning_rate": 6.611476080141747e-06, + "loss": 2.470602035522461, + "step": 246080 + }, + { + "epoch": 0.669933306111338, + "grad_norm": 2.590256452560425, + "learning_rate": 6.610385716232793e-06, + "loss": 2.4754430770874025, + "step": 246100 + }, + { + "epoch": 0.6699877501020824, + "grad_norm": 2.205484390258789, + "learning_rate": 6.609295352323839e-06, + "loss": 2.4731712341308594, + "step": 246120 + }, + { + "epoch": 0.6700421940928271, + "grad_norm": 2.1850080490112305, + "learning_rate": 6.608204988414884e-06, + "loss": 2.4506959915161133, + "step": 246140 + }, + { + "epoch": 0.6700966380835716, + "grad_norm": 2.2425804138183594, + "learning_rate": 6.607114624505929e-06, + "loss": 2.4607391357421875, + "step": 246160 + }, + { + "epoch": 0.6701510820743161, + "grad_norm": 2.0608904361724854, + "learning_rate": 6.606024260596975e-06, + "loss": 2.4719194412231444, + "step": 246180 + }, + { + "epoch": 0.6702055260650606, + "grad_norm": 2.5626614093780518, + "learning_rate": 6.6049338966880205e-06, + "loss": 2.4718242645263673, + "step": 246200 + }, + { + "epoch": 0.6702599700558051, + "grad_norm": 2.294684410095215, + "learning_rate": 6.6038435327790655e-06, + "loss": 2.4620742797851562, + "step": 246220 + }, + { + "epoch": 0.6703144140465496, + "grad_norm": 2.196855306625366, + "learning_rate": 6.6027531688701104e-06, + "loss": 2.474552345275879, + "step": 246240 + }, + { + "epoch": 0.6703688580372942, + "grad_norm": 2.0204176902770996, + "learning_rate": 6.601662804961156e-06, + "loss": 2.454560470581055, + "step": 246260 + }, + { + "epoch": 0.6704233020280387, + "grad_norm": 2.119367837905884, + "learning_rate": 6.600572441052202e-06, + "loss": 2.500835990905762, + "step": 246280 + }, + { + "epoch": 0.6704777460187832, + "grad_norm": 2.1311633586883545, + "learning_rate": 6.599482077143247e-06, + "loss": 2.492634582519531, + "step": 246300 + }, + { + "epoch": 0.6705321900095277, + "grad_norm": 2.2768476009368896, + "learning_rate": 6.598391713234292e-06, + "loss": 2.4620620727539064, + "step": 246320 + }, + { + "epoch": 0.6705866340002722, + "grad_norm": 2.1095869541168213, + "learning_rate": 6.597301349325338e-06, + "loss": 2.4929317474365233, + "step": 246340 + }, + { + "epoch": 0.6706410779910167, + "grad_norm": 2.202383041381836, + "learning_rate": 6.596210985416384e-06, + "loss": 2.453729248046875, + "step": 246360 + }, + { + "epoch": 0.6706955219817613, + "grad_norm": 2.11924147605896, + "learning_rate": 6.5951206215074286e-06, + "loss": 2.4717584609985352, + "step": 246380 + }, + { + "epoch": 0.6707499659725058, + "grad_norm": 2.25063157081604, + "learning_rate": 6.5940302575984735e-06, + "loss": 2.4737855911254885, + "step": 246400 + }, + { + "epoch": 0.6708044099632503, + "grad_norm": 2.2246158123016357, + "learning_rate": 6.592939893689519e-06, + "loss": 2.493770408630371, + "step": 246420 + }, + { + "epoch": 0.6708588539539948, + "grad_norm": 2.1483917236328125, + "learning_rate": 6.591849529780565e-06, + "loss": 2.420902442932129, + "step": 246440 + }, + { + "epoch": 0.6709132979447393, + "grad_norm": 2.191230058670044, + "learning_rate": 6.59075916587161e-06, + "loss": 2.4317949295043944, + "step": 246460 + }, + { + "epoch": 0.6709677419354839, + "grad_norm": 2.1281840801239014, + "learning_rate": 6.589668801962655e-06, + "loss": 2.4285770416259767, + "step": 246480 + }, + { + "epoch": 0.6710221859262284, + "grad_norm": 2.179835319519043, + "learning_rate": 6.588578438053701e-06, + "loss": 2.4743371963500977, + "step": 246500 + }, + { + "epoch": 0.6710766299169729, + "grad_norm": 2.1574313640594482, + "learning_rate": 6.587488074144747e-06, + "loss": 2.4796188354492186, + "step": 246520 + }, + { + "epoch": 0.6711310739077174, + "grad_norm": 2.059638023376465, + "learning_rate": 6.586397710235792e-06, + "loss": 2.4757871627807617, + "step": 246540 + }, + { + "epoch": 0.6711855178984619, + "grad_norm": 2.1052536964416504, + "learning_rate": 6.585307346326837e-06, + "loss": 2.453019142150879, + "step": 246560 + }, + { + "epoch": 0.6712399618892064, + "grad_norm": 2.1075947284698486, + "learning_rate": 6.584216982417882e-06, + "loss": 2.474720764160156, + "step": 246580 + }, + { + "epoch": 0.671294405879951, + "grad_norm": 2.295363664627075, + "learning_rate": 6.583126618508928e-06, + "loss": 2.4542221069335937, + "step": 246600 + }, + { + "epoch": 0.6713488498706955, + "grad_norm": 2.098989963531494, + "learning_rate": 6.582036254599972e-06, + "loss": 2.473710632324219, + "step": 246620 + }, + { + "epoch": 0.67140329386144, + "grad_norm": 2.1758885383605957, + "learning_rate": 6.580945890691018e-06, + "loss": 2.4745065689086916, + "step": 246640 + }, + { + "epoch": 0.6714577378521845, + "grad_norm": 2.701392412185669, + "learning_rate": 6.579855526782064e-06, + "loss": 2.4479320526123045, + "step": 246660 + }, + { + "epoch": 0.671512181842929, + "grad_norm": 2.209883689880371, + "learning_rate": 6.57876516287311e-06, + "loss": 2.460025978088379, + "step": 246680 + }, + { + "epoch": 0.6715666258336737, + "grad_norm": 2.1297640800476074, + "learning_rate": 6.577674798964154e-06, + "loss": 2.453420639038086, + "step": 246700 + }, + { + "epoch": 0.6716210698244182, + "grad_norm": 2.2069969177246094, + "learning_rate": 6.5765844350552e-06, + "loss": 2.4905864715576174, + "step": 246720 + }, + { + "epoch": 0.6716755138151627, + "grad_norm": 2.1963624954223633, + "learning_rate": 6.5754940711462455e-06, + "loss": 2.4866010665893556, + "step": 246740 + }, + { + "epoch": 0.6717299578059072, + "grad_norm": 2.1212916374206543, + "learning_rate": 6.574403707237291e-06, + "loss": 2.4429141998291017, + "step": 246760 + }, + { + "epoch": 0.6717844017966517, + "grad_norm": 2.2145090103149414, + "learning_rate": 6.573313343328335e-06, + "loss": 2.486524772644043, + "step": 246780 + }, + { + "epoch": 0.6718388457873962, + "grad_norm": 2.1687283515930176, + "learning_rate": 6.572222979419381e-06, + "loss": 2.45086727142334, + "step": 246800 + }, + { + "epoch": 0.6718932897781408, + "grad_norm": 2.2633299827575684, + "learning_rate": 6.571132615510427e-06, + "loss": 2.4606906890869142, + "step": 246820 + }, + { + "epoch": 0.6719477337688853, + "grad_norm": 2.1337409019470215, + "learning_rate": 6.570042251601473e-06, + "loss": 2.4790252685546874, + "step": 246840 + }, + { + "epoch": 0.6720021777596298, + "grad_norm": 2.116147041320801, + "learning_rate": 6.568951887692519e-06, + "loss": 2.474244499206543, + "step": 246860 + }, + { + "epoch": 0.6720566217503743, + "grad_norm": 2.1948800086975098, + "learning_rate": 6.567861523783563e-06, + "loss": 2.4749385833740236, + "step": 246880 + }, + { + "epoch": 0.6721110657411188, + "grad_norm": 2.074307680130005, + "learning_rate": 6.566771159874609e-06, + "loss": 2.4609466552734376, + "step": 246900 + }, + { + "epoch": 0.6721655097318634, + "grad_norm": 2.174121379852295, + "learning_rate": 6.565680795965654e-06, + "loss": 2.46236629486084, + "step": 246920 + }, + { + "epoch": 0.6722199537226079, + "grad_norm": 2.1777596473693848, + "learning_rate": 6.5645904320567e-06, + "loss": 2.4507125854492187, + "step": 246940 + }, + { + "epoch": 0.6722743977133524, + "grad_norm": 2.108762264251709, + "learning_rate": 6.563500068147744e-06, + "loss": 2.463070869445801, + "step": 246960 + }, + { + "epoch": 0.6723288417040969, + "grad_norm": 2.352750301361084, + "learning_rate": 6.56240970423879e-06, + "loss": 2.4949031829833985, + "step": 246980 + }, + { + "epoch": 0.6723832856948414, + "grad_norm": 2.100815534591675, + "learning_rate": 6.561319340329836e-06, + "loss": 2.497238349914551, + "step": 247000 + }, + { + "epoch": 0.6724377296855859, + "grad_norm": 2.1914308071136475, + "learning_rate": 6.560228976420882e-06, + "loss": 2.4869035720825194, + "step": 247020 + }, + { + "epoch": 0.6724921736763305, + "grad_norm": 2.1145694255828857, + "learning_rate": 6.559138612511926e-06, + "loss": 2.484343719482422, + "step": 247040 + }, + { + "epoch": 0.672546617667075, + "grad_norm": 2.2204477787017822, + "learning_rate": 6.558048248602972e-06, + "loss": 2.489060974121094, + "step": 247060 + }, + { + "epoch": 0.6726010616578195, + "grad_norm": 2.1168582439422607, + "learning_rate": 6.5569578846940175e-06, + "loss": 2.4855260848999023, + "step": 247080 + }, + { + "epoch": 0.672655505648564, + "grad_norm": 2.2230260372161865, + "learning_rate": 6.555867520785063e-06, + "loss": 2.4531810760498045, + "step": 247100 + }, + { + "epoch": 0.6727099496393085, + "grad_norm": 2.3837015628814697, + "learning_rate": 6.554777156876107e-06, + "loss": 2.458182525634766, + "step": 247120 + }, + { + "epoch": 0.6727643936300531, + "grad_norm": 2.1897366046905518, + "learning_rate": 6.553686792967153e-06, + "loss": 2.456781578063965, + "step": 247140 + }, + { + "epoch": 0.6728188376207976, + "grad_norm": 2.256089925765991, + "learning_rate": 6.552596429058199e-06, + "loss": 2.477829360961914, + "step": 247160 + }, + { + "epoch": 0.6728732816115421, + "grad_norm": 2.1695668697357178, + "learning_rate": 6.551506065149245e-06, + "loss": 2.463028907775879, + "step": 247180 + }, + { + "epoch": 0.6729277256022866, + "grad_norm": 2.117100238800049, + "learning_rate": 6.550415701240289e-06, + "loss": 2.4651519775390627, + "step": 247200 + }, + { + "epoch": 0.6729821695930311, + "grad_norm": 2.453723907470703, + "learning_rate": 6.549325337331335e-06, + "loss": 2.457353401184082, + "step": 247220 + }, + { + "epoch": 0.6730366135837756, + "grad_norm": 2.1294023990631104, + "learning_rate": 6.5482349734223806e-06, + "loss": 2.5092981338500975, + "step": 247240 + }, + { + "epoch": 0.6730910575745203, + "grad_norm": 2.1063239574432373, + "learning_rate": 6.547144609513426e-06, + "loss": 2.46433048248291, + "step": 247260 + }, + { + "epoch": 0.6731455015652648, + "grad_norm": 2.1859076023101807, + "learning_rate": 6.5460542456044705e-06, + "loss": 2.499747467041016, + "step": 247280 + }, + { + "epoch": 0.6731999455560093, + "grad_norm": 2.148057460784912, + "learning_rate": 6.544963881695516e-06, + "loss": 2.4621517181396486, + "step": 247300 + }, + { + "epoch": 0.6732543895467538, + "grad_norm": 2.473839282989502, + "learning_rate": 6.543873517786562e-06, + "loss": 2.506269073486328, + "step": 247320 + }, + { + "epoch": 0.6733088335374983, + "grad_norm": 2.1517043113708496, + "learning_rate": 6.542783153877608e-06, + "loss": 2.467769241333008, + "step": 247340 + }, + { + "epoch": 0.6733632775282428, + "grad_norm": 2.16170334815979, + "learning_rate": 6.541692789968652e-06, + "loss": 2.4744422912597654, + "step": 247360 + }, + { + "epoch": 0.6734177215189874, + "grad_norm": 2.106062650680542, + "learning_rate": 6.540602426059698e-06, + "loss": 2.4746599197387695, + "step": 247380 + }, + { + "epoch": 0.6734721655097319, + "grad_norm": 2.244438409805298, + "learning_rate": 6.539512062150744e-06, + "loss": 2.448276138305664, + "step": 247400 + }, + { + "epoch": 0.6735266095004764, + "grad_norm": 2.1198811531066895, + "learning_rate": 6.5384216982417895e-06, + "loss": 2.490996742248535, + "step": 247420 + }, + { + "epoch": 0.6735810534912209, + "grad_norm": 2.2101008892059326, + "learning_rate": 6.5373313343328336e-06, + "loss": 2.490976905822754, + "step": 247440 + }, + { + "epoch": 0.6736354974819654, + "grad_norm": 2.1271474361419678, + "learning_rate": 6.536240970423879e-06, + "loss": 2.467819023132324, + "step": 247460 + }, + { + "epoch": 0.67368994147271, + "grad_norm": 2.14493989944458, + "learning_rate": 6.535150606514925e-06, + "loss": 2.4805828094482423, + "step": 247480 + }, + { + "epoch": 0.6737443854634545, + "grad_norm": 2.2305121421813965, + "learning_rate": 6.534060242605971e-06, + "loss": 2.491692543029785, + "step": 247500 + }, + { + "epoch": 0.673798829454199, + "grad_norm": 2.128092050552368, + "learning_rate": 6.532969878697015e-06, + "loss": 2.4734214782714843, + "step": 247520 + }, + { + "epoch": 0.6738532734449435, + "grad_norm": 2.0679562091827393, + "learning_rate": 6.531879514788061e-06, + "loss": 2.4447933197021485, + "step": 247540 + }, + { + "epoch": 0.673907717435688, + "grad_norm": 2.218860149383545, + "learning_rate": 6.530789150879107e-06, + "loss": 2.4456514358520507, + "step": 247560 + }, + { + "epoch": 0.6739621614264325, + "grad_norm": 2.0984089374542236, + "learning_rate": 6.5296987869701525e-06, + "loss": 2.4880189895629883, + "step": 247580 + }, + { + "epoch": 0.6740166054171771, + "grad_norm": 2.285599946975708, + "learning_rate": 6.528608423061197e-06, + "loss": 2.4672895431518556, + "step": 247600 + }, + { + "epoch": 0.6740710494079216, + "grad_norm": 2.1478450298309326, + "learning_rate": 6.5275180591522425e-06, + "loss": 2.43823356628418, + "step": 247620 + }, + { + "epoch": 0.6741254933986661, + "grad_norm": 3.069329261779785, + "learning_rate": 6.526427695243288e-06, + "loss": 2.4556880950927735, + "step": 247640 + }, + { + "epoch": 0.6741799373894106, + "grad_norm": 2.1677441596984863, + "learning_rate": 6.525337331334334e-06, + "loss": 2.455697441101074, + "step": 247660 + }, + { + "epoch": 0.6742343813801551, + "grad_norm": 2.0922603607177734, + "learning_rate": 6.524246967425378e-06, + "loss": 2.4552093505859376, + "step": 247680 + }, + { + "epoch": 0.6742888253708997, + "grad_norm": 2.1295628547668457, + "learning_rate": 6.523156603516424e-06, + "loss": 2.457187271118164, + "step": 247700 + }, + { + "epoch": 0.6743432693616442, + "grad_norm": 2.274055004119873, + "learning_rate": 6.52206623960747e-06, + "loss": 2.445111846923828, + "step": 247720 + }, + { + "epoch": 0.6743977133523887, + "grad_norm": 2.2182297706604004, + "learning_rate": 6.520975875698516e-06, + "loss": 2.478449249267578, + "step": 247740 + }, + { + "epoch": 0.6744521573431332, + "grad_norm": 2.1287076473236084, + "learning_rate": 6.51988551178956e-06, + "loss": 2.477804756164551, + "step": 247760 + }, + { + "epoch": 0.6745066013338777, + "grad_norm": 2.2239232063293457, + "learning_rate": 6.5187951478806056e-06, + "loss": 2.5049484252929686, + "step": 247780 + }, + { + "epoch": 0.6745610453246222, + "grad_norm": 2.209949254989624, + "learning_rate": 6.517704783971651e-06, + "loss": 2.467711639404297, + "step": 247800 + }, + { + "epoch": 0.6746154893153669, + "grad_norm": 2.3118417263031006, + "learning_rate": 6.516614420062697e-06, + "loss": 2.533543014526367, + "step": 247820 + }, + { + "epoch": 0.6746699333061114, + "grad_norm": 2.248403549194336, + "learning_rate": 6.515524056153741e-06, + "loss": 2.4721399307250977, + "step": 247840 + }, + { + "epoch": 0.6747243772968559, + "grad_norm": 2.2601535320281982, + "learning_rate": 6.514433692244787e-06, + "loss": 2.456058311462402, + "step": 247860 + }, + { + "epoch": 0.6747788212876004, + "grad_norm": 2.0694589614868164, + "learning_rate": 6.513343328335833e-06, + "loss": 2.476514434814453, + "step": 247880 + }, + { + "epoch": 0.6748332652783449, + "grad_norm": 2.218540906906128, + "learning_rate": 6.512252964426879e-06, + "loss": 2.4975719451904297, + "step": 247900 + }, + { + "epoch": 0.6748877092690895, + "grad_norm": 2.215761184692383, + "learning_rate": 6.511162600517923e-06, + "loss": 2.4374208450317383, + "step": 247920 + }, + { + "epoch": 0.674942153259834, + "grad_norm": 2.263441562652588, + "learning_rate": 6.510072236608969e-06, + "loss": 2.4602514266967774, + "step": 247940 + }, + { + "epoch": 0.6749965972505785, + "grad_norm": 2.1265604496002197, + "learning_rate": 6.5089818727000144e-06, + "loss": 2.453156661987305, + "step": 247960 + }, + { + "epoch": 0.675051041241323, + "grad_norm": 2.2113757133483887, + "learning_rate": 6.50789150879106e-06, + "loss": 2.461354064941406, + "step": 247980 + }, + { + "epoch": 0.6751054852320675, + "grad_norm": 2.233546495437622, + "learning_rate": 6.506801144882104e-06, + "loss": 2.4689502716064453, + "step": 248000 + }, + { + "epoch": 0.675159929222812, + "grad_norm": 2.1174511909484863, + "learning_rate": 6.50571078097315e-06, + "loss": 2.4740388870239256, + "step": 248020 + }, + { + "epoch": 0.6752143732135566, + "grad_norm": 2.2638351917266846, + "learning_rate": 6.504620417064196e-06, + "loss": 2.4463407516479494, + "step": 248040 + }, + { + "epoch": 0.6752688172043011, + "grad_norm": 2.183871269226074, + "learning_rate": 6.503530053155242e-06, + "loss": 2.473862075805664, + "step": 248060 + }, + { + "epoch": 0.6753232611950456, + "grad_norm": 2.1344590187072754, + "learning_rate": 6.502439689246286e-06, + "loss": 2.4919208526611327, + "step": 248080 + }, + { + "epoch": 0.6753777051857901, + "grad_norm": 2.2078804969787598, + "learning_rate": 6.501349325337332e-06, + "loss": 2.4840524673461912, + "step": 248100 + }, + { + "epoch": 0.6754321491765346, + "grad_norm": 2.23453950881958, + "learning_rate": 6.5002589614283775e-06, + "loss": 2.460128593444824, + "step": 248120 + }, + { + "epoch": 0.6754865931672792, + "grad_norm": 2.1147143840789795, + "learning_rate": 6.499168597519423e-06, + "loss": 2.4579135894775392, + "step": 248140 + }, + { + "epoch": 0.6755410371580237, + "grad_norm": 2.1395623683929443, + "learning_rate": 6.4980782336104675e-06, + "loss": 2.495052528381348, + "step": 248160 + }, + { + "epoch": 0.6755954811487682, + "grad_norm": 2.1379427909851074, + "learning_rate": 6.496987869701513e-06, + "loss": 2.4891130447387697, + "step": 248180 + }, + { + "epoch": 0.6756499251395127, + "grad_norm": 2.3574657440185547, + "learning_rate": 6.495897505792559e-06, + "loss": 2.476913070678711, + "step": 248200 + }, + { + "epoch": 0.6757043691302572, + "grad_norm": 2.549781560897827, + "learning_rate": 6.494807141883605e-06, + "loss": 2.4871654510498047, + "step": 248220 + }, + { + "epoch": 0.6757588131210017, + "grad_norm": 2.2002015113830566, + "learning_rate": 6.493716777974649e-06, + "loss": 2.4566234588623046, + "step": 248240 + }, + { + "epoch": 0.6758132571117463, + "grad_norm": 2.1663143634796143, + "learning_rate": 6.492626414065695e-06, + "loss": 2.4509113311767576, + "step": 248260 + }, + { + "epoch": 0.6758677011024908, + "grad_norm": 2.766697645187378, + "learning_rate": 6.491536050156741e-06, + "loss": 2.4671449661254883, + "step": 248280 + }, + { + "epoch": 0.6759221450932353, + "grad_norm": 3.6629738807678223, + "learning_rate": 6.4904456862477864e-06, + "loss": 2.4841592788696287, + "step": 248300 + }, + { + "epoch": 0.6759765890839798, + "grad_norm": 2.1034131050109863, + "learning_rate": 6.4893553223388305e-06, + "loss": 2.4540523529052733, + "step": 248320 + }, + { + "epoch": 0.6760310330747243, + "grad_norm": 2.202651023864746, + "learning_rate": 6.488264958429876e-06, + "loss": 2.453190231323242, + "step": 248340 + }, + { + "epoch": 0.676085477065469, + "grad_norm": 2.145387649536133, + "learning_rate": 6.487174594520922e-06, + "loss": 2.497319793701172, + "step": 248360 + }, + { + "epoch": 0.6761399210562135, + "grad_norm": 2.1064798831939697, + "learning_rate": 6.486084230611968e-06, + "loss": 2.44128303527832, + "step": 248380 + }, + { + "epoch": 0.676194365046958, + "grad_norm": 2.2528278827667236, + "learning_rate": 6.484993866703012e-06, + "loss": 2.478895568847656, + "step": 248400 + }, + { + "epoch": 0.6762488090377025, + "grad_norm": 2.215542793273926, + "learning_rate": 6.483903502794058e-06, + "loss": 2.4731197357177734, + "step": 248420 + }, + { + "epoch": 0.676303253028447, + "grad_norm": 2.304497241973877, + "learning_rate": 6.482813138885104e-06, + "loss": 2.464215087890625, + "step": 248440 + }, + { + "epoch": 0.6763576970191915, + "grad_norm": 2.1842665672302246, + "learning_rate": 6.4817227749761495e-06, + "loss": 2.4672674179077148, + "step": 248460 + }, + { + "epoch": 0.6764121410099361, + "grad_norm": 2.1316299438476562, + "learning_rate": 6.480632411067194e-06, + "loss": 2.489968109130859, + "step": 248480 + }, + { + "epoch": 0.6764665850006806, + "grad_norm": 2.083235502243042, + "learning_rate": 6.4795420471582394e-06, + "loss": 2.4728910446166994, + "step": 248500 + }, + { + "epoch": 0.6765210289914251, + "grad_norm": 2.222794771194458, + "learning_rate": 6.478451683249285e-06, + "loss": 2.462653160095215, + "step": 248520 + }, + { + "epoch": 0.6765754729821696, + "grad_norm": 2.225860834121704, + "learning_rate": 6.477361319340331e-06, + "loss": 2.4573253631591796, + "step": 248540 + }, + { + "epoch": 0.6766299169729141, + "grad_norm": 2.424241065979004, + "learning_rate": 6.476270955431376e-06, + "loss": 2.502568817138672, + "step": 248560 + }, + { + "epoch": 0.6766843609636586, + "grad_norm": 2.388921022415161, + "learning_rate": 6.475180591522421e-06, + "loss": 2.452867126464844, + "step": 248580 + }, + { + "epoch": 0.6767388049544032, + "grad_norm": 2.2971386909484863, + "learning_rate": 6.474090227613467e-06, + "loss": 2.4358213424682615, + "step": 248600 + }, + { + "epoch": 0.6767932489451477, + "grad_norm": 2.2223715782165527, + "learning_rate": 6.472999863704512e-06, + "loss": 2.4779964447021485, + "step": 248620 + }, + { + "epoch": 0.6768476929358922, + "grad_norm": 2.1688034534454346, + "learning_rate": 6.4719094997955576e-06, + "loss": 2.4825695037841795, + "step": 248640 + }, + { + "epoch": 0.6769021369266367, + "grad_norm": 2.141879081726074, + "learning_rate": 6.4708191358866025e-06, + "loss": 2.4677085876464844, + "step": 248660 + }, + { + "epoch": 0.6769565809173812, + "grad_norm": 2.132951498031616, + "learning_rate": 6.469728771977648e-06, + "loss": 2.476546859741211, + "step": 248680 + }, + { + "epoch": 0.6770110249081258, + "grad_norm": 2.3943135738372803, + "learning_rate": 6.468638408068693e-06, + "loss": 2.454182243347168, + "step": 248700 + }, + { + "epoch": 0.6770654688988703, + "grad_norm": 2.1436874866485596, + "learning_rate": 6.467548044159739e-06, + "loss": 2.4777894973754884, + "step": 248720 + }, + { + "epoch": 0.6771199128896148, + "grad_norm": 2.159956693649292, + "learning_rate": 6.466457680250784e-06, + "loss": 2.487270736694336, + "step": 248740 + }, + { + "epoch": 0.6771743568803593, + "grad_norm": 2.217235803604126, + "learning_rate": 6.46536731634183e-06, + "loss": 2.4786306381225587, + "step": 248760 + }, + { + "epoch": 0.6772288008711038, + "grad_norm": 2.1984925270080566, + "learning_rate": 6.464276952432875e-06, + "loss": 2.478480911254883, + "step": 248780 + }, + { + "epoch": 0.6772832448618483, + "grad_norm": 2.106912612915039, + "learning_rate": 6.463186588523921e-06, + "loss": 2.4453205108642577, + "step": 248800 + }, + { + "epoch": 0.6773376888525929, + "grad_norm": 2.152909517288208, + "learning_rate": 6.462096224614966e-06, + "loss": 2.476936149597168, + "step": 248820 + }, + { + "epoch": 0.6773921328433374, + "grad_norm": 2.1647794246673584, + "learning_rate": 6.461005860706011e-06, + "loss": 2.4867551803588865, + "step": 248840 + }, + { + "epoch": 0.6774465768340819, + "grad_norm": 2.3641114234924316, + "learning_rate": 6.459915496797056e-06, + "loss": 2.481689453125, + "step": 248860 + }, + { + "epoch": 0.6775010208248264, + "grad_norm": 2.7228574752807617, + "learning_rate": 6.458825132888102e-06, + "loss": 2.4667074203491213, + "step": 248880 + }, + { + "epoch": 0.6775554648155709, + "grad_norm": 2.1373157501220703, + "learning_rate": 6.457734768979147e-06, + "loss": 2.4943288803100585, + "step": 248900 + }, + { + "epoch": 0.6776099088063156, + "grad_norm": 2.152343511581421, + "learning_rate": 6.456644405070193e-06, + "loss": 2.4621849060058594, + "step": 248920 + }, + { + "epoch": 0.67766435279706, + "grad_norm": 2.158562183380127, + "learning_rate": 6.455554041161238e-06, + "loss": 2.4909946441650392, + "step": 248940 + }, + { + "epoch": 0.6777187967878046, + "grad_norm": 2.1567466259002686, + "learning_rate": 6.454463677252284e-06, + "loss": 2.459807586669922, + "step": 248960 + }, + { + "epoch": 0.6777732407785491, + "grad_norm": 2.217625141143799, + "learning_rate": 6.453373313343329e-06, + "loss": 2.4892858505249023, + "step": 248980 + }, + { + "epoch": 0.6778276847692936, + "grad_norm": 2.251909017562866, + "learning_rate": 6.4522829494343745e-06, + "loss": 2.463770866394043, + "step": 249000 + }, + { + "epoch": 0.6778821287600381, + "grad_norm": 2.2134833335876465, + "learning_rate": 6.4511925855254195e-06, + "loss": 2.479544258117676, + "step": 249020 + }, + { + "epoch": 0.6779365727507827, + "grad_norm": 2.1839869022369385, + "learning_rate": 6.450102221616465e-06, + "loss": 2.486317253112793, + "step": 249040 + }, + { + "epoch": 0.6779910167415272, + "grad_norm": 2.1975324153900146, + "learning_rate": 6.44901185770751e-06, + "loss": 2.469026565551758, + "step": 249060 + }, + { + "epoch": 0.6780454607322717, + "grad_norm": 2.361891984939575, + "learning_rate": 6.447921493798556e-06, + "loss": 2.4621740341186524, + "step": 249080 + }, + { + "epoch": 0.6780999047230162, + "grad_norm": 2.1638543605804443, + "learning_rate": 6.446831129889601e-06, + "loss": 2.471809196472168, + "step": 249100 + }, + { + "epoch": 0.6781543487137607, + "grad_norm": 2.150155782699585, + "learning_rate": 6.445740765980647e-06, + "loss": 2.454885482788086, + "step": 249120 + }, + { + "epoch": 0.6782087927045053, + "grad_norm": 2.3044071197509766, + "learning_rate": 6.444650402071692e-06, + "loss": 2.507644844055176, + "step": 249140 + }, + { + "epoch": 0.6782632366952498, + "grad_norm": 2.038256883621216, + "learning_rate": 6.443560038162738e-06, + "loss": 2.461966705322266, + "step": 249160 + }, + { + "epoch": 0.6783176806859943, + "grad_norm": 2.1284232139587402, + "learning_rate": 6.4424696742537826e-06, + "loss": 2.4747383117675783, + "step": 249180 + }, + { + "epoch": 0.6783721246767388, + "grad_norm": 2.1395647525787354, + "learning_rate": 6.441379310344828e-06, + "loss": 2.456722068786621, + "step": 249200 + }, + { + "epoch": 0.6784265686674833, + "grad_norm": 2.04376220703125, + "learning_rate": 6.440288946435873e-06, + "loss": 2.4463258743286134, + "step": 249220 + }, + { + "epoch": 0.6784810126582278, + "grad_norm": 2.398928642272949, + "learning_rate": 6.439198582526919e-06, + "loss": 2.4659027099609374, + "step": 249240 + }, + { + "epoch": 0.6785354566489724, + "grad_norm": 2.2541377544403076, + "learning_rate": 6.438108218617964e-06, + "loss": 2.4515003204345702, + "step": 249260 + }, + { + "epoch": 0.6785899006397169, + "grad_norm": 2.280613422393799, + "learning_rate": 6.43701785470901e-06, + "loss": 2.4678119659423827, + "step": 249280 + }, + { + "epoch": 0.6786443446304614, + "grad_norm": 2.0979042053222656, + "learning_rate": 6.435927490800055e-06, + "loss": 2.4910327911376955, + "step": 249300 + }, + { + "epoch": 0.6786987886212059, + "grad_norm": 2.1266367435455322, + "learning_rate": 6.434837126891101e-06, + "loss": 2.452006721496582, + "step": 249320 + }, + { + "epoch": 0.6787532326119504, + "grad_norm": 2.1876206398010254, + "learning_rate": 6.433746762982146e-06, + "loss": 2.4466272354125977, + "step": 249340 + }, + { + "epoch": 0.678807676602695, + "grad_norm": 2.1275548934936523, + "learning_rate": 6.4326563990731914e-06, + "loss": 2.4728767395019533, + "step": 249360 + }, + { + "epoch": 0.6788621205934395, + "grad_norm": 2.23518443107605, + "learning_rate": 6.431566035164236e-06, + "loss": 2.4720458984375, + "step": 249380 + }, + { + "epoch": 0.678916564584184, + "grad_norm": 2.2662739753723145, + "learning_rate": 6.430475671255282e-06, + "loss": 2.4666265487670898, + "step": 249400 + }, + { + "epoch": 0.6789710085749285, + "grad_norm": 2.180981397628784, + "learning_rate": 6.429385307346327e-06, + "loss": 2.471550178527832, + "step": 249420 + }, + { + "epoch": 0.679025452565673, + "grad_norm": 2.208930253982544, + "learning_rate": 6.428294943437373e-06, + "loss": 2.485066223144531, + "step": 249440 + }, + { + "epoch": 0.6790798965564175, + "grad_norm": 2.2287235260009766, + "learning_rate": 6.427204579528418e-06, + "loss": 2.4941925048828124, + "step": 249460 + }, + { + "epoch": 0.6791343405471622, + "grad_norm": 2.174179792404175, + "learning_rate": 6.426114215619464e-06, + "loss": 2.4621145248413088, + "step": 249480 + }, + { + "epoch": 0.6791887845379067, + "grad_norm": 2.1378464698791504, + "learning_rate": 6.425023851710509e-06, + "loss": 2.494232177734375, + "step": 249500 + }, + { + "epoch": 0.6792432285286512, + "grad_norm": 2.191157817840576, + "learning_rate": 6.4239334878015545e-06, + "loss": 2.464225196838379, + "step": 249520 + }, + { + "epoch": 0.6792976725193957, + "grad_norm": 2.605945110321045, + "learning_rate": 6.4228431238925995e-06, + "loss": 2.4677318572998046, + "step": 249540 + }, + { + "epoch": 0.6793521165101402, + "grad_norm": 2.109851360321045, + "learning_rate": 6.421752759983645e-06, + "loss": 2.491108512878418, + "step": 249560 + }, + { + "epoch": 0.6794065605008847, + "grad_norm": 2.18222975730896, + "learning_rate": 6.42066239607469e-06, + "loss": 2.4675786972045897, + "step": 249580 + }, + { + "epoch": 0.6794610044916293, + "grad_norm": 2.6115305423736572, + "learning_rate": 6.419572032165736e-06, + "loss": 2.477887725830078, + "step": 249600 + }, + { + "epoch": 0.6795154484823738, + "grad_norm": 2.337655782699585, + "learning_rate": 6.418481668256781e-06, + "loss": 2.466161346435547, + "step": 249620 + }, + { + "epoch": 0.6795698924731183, + "grad_norm": 2.200374126434326, + "learning_rate": 6.417391304347827e-06, + "loss": 2.4789608001708983, + "step": 249640 + }, + { + "epoch": 0.6796243364638628, + "grad_norm": 2.2433855533599854, + "learning_rate": 6.416300940438872e-06, + "loss": 2.4949565887451173, + "step": 249660 + }, + { + "epoch": 0.6796787804546073, + "grad_norm": 2.340557098388672, + "learning_rate": 6.415210576529918e-06, + "loss": 2.473666954040527, + "step": 249680 + }, + { + "epoch": 0.6797332244453519, + "grad_norm": 2.1674556732177734, + "learning_rate": 6.414120212620963e-06, + "loss": 2.469489097595215, + "step": 249700 + }, + { + "epoch": 0.6797876684360964, + "grad_norm": 2.1322712898254395, + "learning_rate": 6.4130298487120075e-06, + "loss": 2.4547660827636717, + "step": 249720 + }, + { + "epoch": 0.6798421124268409, + "grad_norm": 2.197737455368042, + "learning_rate": 6.411939484803053e-06, + "loss": 2.4914913177490234, + "step": 249740 + }, + { + "epoch": 0.6798965564175854, + "grad_norm": 2.1806960105895996, + "learning_rate": 6.410849120894099e-06, + "loss": 2.4554464340209963, + "step": 249760 + }, + { + "epoch": 0.6799510004083299, + "grad_norm": 3.8166255950927734, + "learning_rate": 6.409758756985144e-06, + "loss": 2.4610706329345704, + "step": 249780 + }, + { + "epoch": 0.6800054443990744, + "grad_norm": 2.2967300415039062, + "learning_rate": 6.408668393076189e-06, + "loss": 2.4661062240600584, + "step": 249800 + }, + { + "epoch": 0.680059888389819, + "grad_norm": 2.220651626586914, + "learning_rate": 6.407578029167235e-06, + "loss": 2.491946792602539, + "step": 249820 + }, + { + "epoch": 0.6801143323805635, + "grad_norm": 8.966792106628418, + "learning_rate": 6.406487665258281e-06, + "loss": 2.45385684967041, + "step": 249840 + }, + { + "epoch": 0.680168776371308, + "grad_norm": 2.1521878242492676, + "learning_rate": 6.405397301349326e-06, + "loss": 2.4736480712890625, + "step": 249860 + }, + { + "epoch": 0.6802232203620525, + "grad_norm": 2.0047860145568848, + "learning_rate": 6.404306937440371e-06, + "loss": 2.471555709838867, + "step": 249880 + }, + { + "epoch": 0.680277664352797, + "grad_norm": 2.266540288925171, + "learning_rate": 6.4032165735314164e-06, + "loss": 2.4708955764770506, + "step": 249900 + }, + { + "epoch": 0.6803321083435416, + "grad_norm": 2.2753143310546875, + "learning_rate": 6.402126209622462e-06, + "loss": 2.4865385055541993, + "step": 249920 + }, + { + "epoch": 0.6803865523342861, + "grad_norm": 2.2157649993896484, + "learning_rate": 6.401035845713507e-06, + "loss": 2.4080102920532225, + "step": 249940 + }, + { + "epoch": 0.6804409963250306, + "grad_norm": 2.1873667240142822, + "learning_rate": 6.399945481804552e-06, + "loss": 2.4803293228149412, + "step": 249960 + }, + { + "epoch": 0.6804954403157751, + "grad_norm": 2.136596441268921, + "learning_rate": 6.398855117895598e-06, + "loss": 2.460279846191406, + "step": 249980 + }, + { + "epoch": 0.6805498843065196, + "grad_norm": 2.4504594802856445, + "learning_rate": 6.397764753986644e-06, + "loss": 2.5126474380493162, + "step": 250000 + }, + { + "epoch": 0.6806043282972641, + "grad_norm": 2.2837777137756348, + "learning_rate": 6.396674390077689e-06, + "loss": 2.47772274017334, + "step": 250020 + }, + { + "epoch": 0.6806587722880088, + "grad_norm": 2.1968884468078613, + "learning_rate": 6.395584026168734e-06, + "loss": 2.4439605712890624, + "step": 250040 + }, + { + "epoch": 0.6807132162787533, + "grad_norm": 2.1318604946136475, + "learning_rate": 6.3944936622597795e-06, + "loss": 2.4600650787353517, + "step": 250060 + }, + { + "epoch": 0.6807676602694978, + "grad_norm": 2.10019850730896, + "learning_rate": 6.393403298350825e-06, + "loss": 2.488929557800293, + "step": 250080 + }, + { + "epoch": 0.6808221042602423, + "grad_norm": 2.0963146686553955, + "learning_rate": 6.39231293444187e-06, + "loss": 2.4522581100463867, + "step": 250100 + }, + { + "epoch": 0.6808765482509868, + "grad_norm": 2.185973882675171, + "learning_rate": 6.391222570532915e-06, + "loss": 2.488820266723633, + "step": 250120 + }, + { + "epoch": 0.6809309922417314, + "grad_norm": 2.233618974685669, + "learning_rate": 6.390132206623961e-06, + "loss": 2.463215446472168, + "step": 250140 + }, + { + "epoch": 0.6809854362324759, + "grad_norm": 2.28733491897583, + "learning_rate": 6.389041842715007e-06, + "loss": 2.47155647277832, + "step": 250160 + }, + { + "epoch": 0.6810398802232204, + "grad_norm": 2.1224679946899414, + "learning_rate": 6.387951478806053e-06, + "loss": 2.4876731872558593, + "step": 250180 + }, + { + "epoch": 0.6810943242139649, + "grad_norm": 2.292346239089966, + "learning_rate": 6.386861114897097e-06, + "loss": 2.463095283508301, + "step": 250200 + }, + { + "epoch": 0.6811487682047094, + "grad_norm": 2.111330270767212, + "learning_rate": 6.385770750988143e-06, + "loss": 2.4873416900634764, + "step": 250220 + }, + { + "epoch": 0.6812032121954539, + "grad_norm": 2.187699317932129, + "learning_rate": 6.384680387079188e-06, + "loss": 2.4706844329833983, + "step": 250240 + }, + { + "epoch": 0.6812576561861985, + "grad_norm": 2.1378448009490967, + "learning_rate": 6.383590023170234e-06, + "loss": 2.4443405151367186, + "step": 250260 + }, + { + "epoch": 0.681312100176943, + "grad_norm": 2.3471877574920654, + "learning_rate": 6.382499659261278e-06, + "loss": 2.4353824615478517, + "step": 250280 + }, + { + "epoch": 0.6813665441676875, + "grad_norm": 2.16058349609375, + "learning_rate": 6.381409295352324e-06, + "loss": 2.4810308456420898, + "step": 250300 + }, + { + "epoch": 0.681420988158432, + "grad_norm": 2.0739665031433105, + "learning_rate": 6.38031893144337e-06, + "loss": 2.4722568511962892, + "step": 250320 + }, + { + "epoch": 0.6814754321491765, + "grad_norm": 2.1841909885406494, + "learning_rate": 6.379228567534416e-06, + "loss": 2.5031219482421876, + "step": 250340 + }, + { + "epoch": 0.6815298761399211, + "grad_norm": 2.137688398361206, + "learning_rate": 6.37813820362546e-06, + "loss": 2.4984567642211912, + "step": 250360 + }, + { + "epoch": 0.6815843201306656, + "grad_norm": 2.125009298324585, + "learning_rate": 6.377047839716506e-06, + "loss": 2.4661270141601563, + "step": 250380 + }, + { + "epoch": 0.6816387641214101, + "grad_norm": 2.1620917320251465, + "learning_rate": 6.3759574758075515e-06, + "loss": 2.475511360168457, + "step": 250400 + }, + { + "epoch": 0.6816932081121546, + "grad_norm": 2.1757750511169434, + "learning_rate": 6.374867111898597e-06, + "loss": 2.4793010711669923, + "step": 250420 + }, + { + "epoch": 0.6817476521028991, + "grad_norm": 2.158499002456665, + "learning_rate": 6.3737767479896414e-06, + "loss": 2.474176597595215, + "step": 250440 + }, + { + "epoch": 0.6818020960936436, + "grad_norm": 2.1387135982513428, + "learning_rate": 6.372686384080687e-06, + "loss": 2.4589990615844726, + "step": 250460 + }, + { + "epoch": 0.6818565400843882, + "grad_norm": 2.526066780090332, + "learning_rate": 6.371596020171733e-06, + "loss": 2.4737478256225587, + "step": 250480 + }, + { + "epoch": 0.6819109840751327, + "grad_norm": 2.1698787212371826, + "learning_rate": 6.370505656262779e-06, + "loss": 2.4427349090576174, + "step": 250500 + }, + { + "epoch": 0.6819654280658772, + "grad_norm": 2.2148373126983643, + "learning_rate": 6.369415292353823e-06, + "loss": 2.4847366333007814, + "step": 250520 + }, + { + "epoch": 0.6820198720566217, + "grad_norm": 2.1031761169433594, + "learning_rate": 6.368324928444869e-06, + "loss": 2.482541084289551, + "step": 250540 + }, + { + "epoch": 0.6820743160473662, + "grad_norm": 2.1433897018432617, + "learning_rate": 6.367234564535915e-06, + "loss": 2.467781829833984, + "step": 250560 + }, + { + "epoch": 0.6821287600381108, + "grad_norm": 2.2022578716278076, + "learning_rate": 6.36614420062696e-06, + "loss": 2.4716571807861327, + "step": 250580 + }, + { + "epoch": 0.6821832040288554, + "grad_norm": 2.1598381996154785, + "learning_rate": 6.3650538367180045e-06, + "loss": 2.4168025970458986, + "step": 250600 + }, + { + "epoch": 0.6822376480195999, + "grad_norm": 2.098571538925171, + "learning_rate": 6.36396347280905e-06, + "loss": 2.481513786315918, + "step": 250620 + }, + { + "epoch": 0.6822920920103444, + "grad_norm": 2.1646883487701416, + "learning_rate": 6.362873108900096e-06, + "loss": 2.511083793640137, + "step": 250640 + }, + { + "epoch": 0.6823465360010889, + "grad_norm": 2.207796812057495, + "learning_rate": 6.361782744991142e-06, + "loss": 2.459225082397461, + "step": 250660 + }, + { + "epoch": 0.6824009799918334, + "grad_norm": 2.232513904571533, + "learning_rate": 6.360692381082186e-06, + "loss": 2.4754194259643554, + "step": 250680 + }, + { + "epoch": 0.682455423982578, + "grad_norm": 2.0320496559143066, + "learning_rate": 6.359602017173232e-06, + "loss": 2.464939498901367, + "step": 250700 + }, + { + "epoch": 0.6825098679733225, + "grad_norm": 2.1732118129730225, + "learning_rate": 6.358511653264278e-06, + "loss": 2.4547815322875977, + "step": 250720 + }, + { + "epoch": 0.682564311964067, + "grad_norm": 2.271089792251587, + "learning_rate": 6.3574212893553235e-06, + "loss": 2.483118438720703, + "step": 250740 + }, + { + "epoch": 0.6826187559548115, + "grad_norm": 2.162426233291626, + "learning_rate": 6.356330925446368e-06, + "loss": 2.4667402267456056, + "step": 250760 + }, + { + "epoch": 0.682673199945556, + "grad_norm": 2.1747915744781494, + "learning_rate": 6.355240561537413e-06, + "loss": 2.491654396057129, + "step": 250780 + }, + { + "epoch": 0.6827276439363005, + "grad_norm": 2.1190762519836426, + "learning_rate": 6.354150197628459e-06, + "loss": 2.4795936584472655, + "step": 250800 + }, + { + "epoch": 0.6827820879270451, + "grad_norm": 2.1474361419677734, + "learning_rate": 6.353059833719505e-06, + "loss": 2.493179702758789, + "step": 250820 + }, + { + "epoch": 0.6828365319177896, + "grad_norm": 2.0500354766845703, + "learning_rate": 6.351969469810549e-06, + "loss": 2.462175750732422, + "step": 250840 + }, + { + "epoch": 0.6828909759085341, + "grad_norm": 2.60795521736145, + "learning_rate": 6.350879105901595e-06, + "loss": 2.5180959701538086, + "step": 250860 + }, + { + "epoch": 0.6829454198992786, + "grad_norm": 2.155686616897583, + "learning_rate": 6.349788741992641e-06, + "loss": 2.4376962661743162, + "step": 250880 + }, + { + "epoch": 0.6829998638900231, + "grad_norm": 2.358952045440674, + "learning_rate": 6.3486983780836866e-06, + "loss": 2.456951904296875, + "step": 250900 + }, + { + "epoch": 0.6830543078807677, + "grad_norm": 2.1419060230255127, + "learning_rate": 6.347608014174731e-06, + "loss": 2.4657135009765625, + "step": 250920 + }, + { + "epoch": 0.6831087518715122, + "grad_norm": 2.0734047889709473, + "learning_rate": 6.3465176502657765e-06, + "loss": 2.477741813659668, + "step": 250940 + }, + { + "epoch": 0.6831631958622567, + "grad_norm": 2.148495674133301, + "learning_rate": 6.345427286356822e-06, + "loss": 2.466075325012207, + "step": 250960 + }, + { + "epoch": 0.6832176398530012, + "grad_norm": 2.1331255435943604, + "learning_rate": 6.344336922447868e-06, + "loss": 2.470158004760742, + "step": 250980 + }, + { + "epoch": 0.6832720838437457, + "grad_norm": 2.4510140419006348, + "learning_rate": 6.343246558538912e-06, + "loss": 2.477840805053711, + "step": 251000 + }, + { + "epoch": 0.6833265278344902, + "grad_norm": 2.118058681488037, + "learning_rate": 6.342156194629958e-06, + "loss": 2.4481632232666017, + "step": 251020 + }, + { + "epoch": 0.6833809718252348, + "grad_norm": 2.151031017303467, + "learning_rate": 6.341065830721004e-06, + "loss": 2.4741540908813477, + "step": 251040 + }, + { + "epoch": 0.6834354158159793, + "grad_norm": 2.177277088165283, + "learning_rate": 6.33997546681205e-06, + "loss": 2.4660480499267576, + "step": 251060 + }, + { + "epoch": 0.6834898598067238, + "grad_norm": 2.082792043685913, + "learning_rate": 6.338885102903094e-06, + "loss": 2.474468231201172, + "step": 251080 + }, + { + "epoch": 0.6835443037974683, + "grad_norm": 2.3829774856567383, + "learning_rate": 6.33779473899414e-06, + "loss": 2.428599548339844, + "step": 251100 + }, + { + "epoch": 0.6835987477882128, + "grad_norm": 2.2313232421875, + "learning_rate": 6.336704375085185e-06, + "loss": 2.508144187927246, + "step": 251120 + }, + { + "epoch": 0.6836531917789574, + "grad_norm": 4.181398868560791, + "learning_rate": 6.335614011176231e-06, + "loss": 2.4587997436523437, + "step": 251140 + }, + { + "epoch": 0.683707635769702, + "grad_norm": 2.1274447441101074, + "learning_rate": 6.334523647267275e-06, + "loss": 2.4822492599487305, + "step": 251160 + }, + { + "epoch": 0.6837620797604465, + "grad_norm": 2.1177377700805664, + "learning_rate": 6.333433283358321e-06, + "loss": 2.4723175048828123, + "step": 251180 + }, + { + "epoch": 0.683816523751191, + "grad_norm": 2.1744043827056885, + "learning_rate": 6.332342919449367e-06, + "loss": 2.44156551361084, + "step": 251200 + }, + { + "epoch": 0.6838709677419355, + "grad_norm": 2.129460096359253, + "learning_rate": 6.331252555540413e-06, + "loss": 2.4561531066894533, + "step": 251220 + }, + { + "epoch": 0.68392541173268, + "grad_norm": 2.167262554168701, + "learning_rate": 6.330162191631457e-06, + "loss": 2.4712217330932615, + "step": 251240 + }, + { + "epoch": 0.6839798557234246, + "grad_norm": 2.1779401302337646, + "learning_rate": 6.329071827722503e-06, + "loss": 2.497138023376465, + "step": 251260 + }, + { + "epoch": 0.6840342997141691, + "grad_norm": 2.151113748550415, + "learning_rate": 6.3279814638135485e-06, + "loss": 2.477563667297363, + "step": 251280 + }, + { + "epoch": 0.6840887437049136, + "grad_norm": 2.1595828533172607, + "learning_rate": 6.326891099904594e-06, + "loss": 2.4552274703979493, + "step": 251300 + }, + { + "epoch": 0.6841431876956581, + "grad_norm": 2.1112780570983887, + "learning_rate": 6.325800735995638e-06, + "loss": 2.4578922271728514, + "step": 251320 + }, + { + "epoch": 0.6841976316864026, + "grad_norm": 2.141582727432251, + "learning_rate": 6.324710372086684e-06, + "loss": 2.484872055053711, + "step": 251340 + }, + { + "epoch": 0.6842520756771472, + "grad_norm": 2.1627919673919678, + "learning_rate": 6.32362000817773e-06, + "loss": 2.4694461822509766, + "step": 251360 + }, + { + "epoch": 0.6843065196678917, + "grad_norm": 2.146547555923462, + "learning_rate": 6.322529644268776e-06, + "loss": 2.4370952606201173, + "step": 251380 + }, + { + "epoch": 0.6843609636586362, + "grad_norm": 2.334885597229004, + "learning_rate": 6.32143928035982e-06, + "loss": 2.484930419921875, + "step": 251400 + }, + { + "epoch": 0.6844154076493807, + "grad_norm": 2.2789719104766846, + "learning_rate": 6.320348916450866e-06, + "loss": 2.451448440551758, + "step": 251420 + }, + { + "epoch": 0.6844698516401252, + "grad_norm": 2.1920602321624756, + "learning_rate": 6.3192585525419116e-06, + "loss": 2.479576301574707, + "step": 251440 + }, + { + "epoch": 0.6845242956308697, + "grad_norm": 2.3135902881622314, + "learning_rate": 6.318168188632957e-06, + "loss": 2.4637378692626952, + "step": 251460 + }, + { + "epoch": 0.6845787396216143, + "grad_norm": 2.1298930644989014, + "learning_rate": 6.3170778247240015e-06, + "loss": 2.458845520019531, + "step": 251480 + }, + { + "epoch": 0.6846331836123588, + "grad_norm": 2.1984362602233887, + "learning_rate": 6.315987460815047e-06, + "loss": 2.4550359725952147, + "step": 251500 + }, + { + "epoch": 0.6846876276031033, + "grad_norm": 2.2113845348358154, + "learning_rate": 6.314897096906093e-06, + "loss": 2.491580009460449, + "step": 251520 + }, + { + "epoch": 0.6847420715938478, + "grad_norm": 2.1597163677215576, + "learning_rate": 6.313806732997139e-06, + "loss": 2.4812877655029295, + "step": 251540 + }, + { + "epoch": 0.6847965155845923, + "grad_norm": 2.2979819774627686, + "learning_rate": 6.312716369088183e-06, + "loss": 2.4399370193481444, + "step": 251560 + }, + { + "epoch": 0.6848509595753369, + "grad_norm": 2.2016289234161377, + "learning_rate": 6.311626005179229e-06, + "loss": 2.4536884307861326, + "step": 251580 + }, + { + "epoch": 0.6849054035660814, + "grad_norm": 2.135798215866089, + "learning_rate": 6.310535641270275e-06, + "loss": 2.446124267578125, + "step": 251600 + }, + { + "epoch": 0.6849598475568259, + "grad_norm": 2.215487480163574, + "learning_rate": 6.3094452773613205e-06, + "loss": 2.473990631103516, + "step": 251620 + }, + { + "epoch": 0.6850142915475704, + "grad_norm": 2.178115129470825, + "learning_rate": 6.3083549134523646e-06, + "loss": 2.4706775665283205, + "step": 251640 + }, + { + "epoch": 0.6850687355383149, + "grad_norm": 2.1579692363739014, + "learning_rate": 6.30726454954341e-06, + "loss": 2.462403106689453, + "step": 251660 + }, + { + "epoch": 0.6851231795290594, + "grad_norm": 2.1621274948120117, + "learning_rate": 6.306174185634456e-06, + "loss": 2.46192626953125, + "step": 251680 + }, + { + "epoch": 0.685177623519804, + "grad_norm": 2.201956033706665, + "learning_rate": 6.305083821725502e-06, + "loss": 2.50306396484375, + "step": 251700 + }, + { + "epoch": 0.6852320675105485, + "grad_norm": 2.263481855392456, + "learning_rate": 6.303993457816546e-06, + "loss": 2.469553565979004, + "step": 251720 + }, + { + "epoch": 0.685286511501293, + "grad_norm": 2.1653170585632324, + "learning_rate": 6.302903093907592e-06, + "loss": 2.450550842285156, + "step": 251740 + }, + { + "epoch": 0.6853409554920376, + "grad_norm": 2.08866810798645, + "learning_rate": 6.301812729998638e-06, + "loss": 2.4512475967407226, + "step": 251760 + }, + { + "epoch": 0.685395399482782, + "grad_norm": 2.2102932929992676, + "learning_rate": 6.3007223660896835e-06, + "loss": 2.466568374633789, + "step": 251780 + }, + { + "epoch": 0.6854498434735266, + "grad_norm": 2.0468626022338867, + "learning_rate": 6.299632002180728e-06, + "loss": 2.4419788360595702, + "step": 251800 + }, + { + "epoch": 0.6855042874642712, + "grad_norm": 2.0800507068634033, + "learning_rate": 6.2985416382717735e-06, + "loss": 2.459552192687988, + "step": 251820 + }, + { + "epoch": 0.6855587314550157, + "grad_norm": 2.121936559677124, + "learning_rate": 6.297451274362819e-06, + "loss": 2.438599967956543, + "step": 251840 + }, + { + "epoch": 0.6856131754457602, + "grad_norm": 2.1802866458892822, + "learning_rate": 6.296360910453865e-06, + "loss": 2.491664695739746, + "step": 251860 + }, + { + "epoch": 0.6856676194365047, + "grad_norm": 2.2617766857147217, + "learning_rate": 6.29527054654491e-06, + "loss": 2.470338249206543, + "step": 251880 + }, + { + "epoch": 0.6857220634272492, + "grad_norm": 2.197308301925659, + "learning_rate": 6.294180182635955e-06, + "loss": 2.490169143676758, + "step": 251900 + }, + { + "epoch": 0.6857765074179938, + "grad_norm": 2.14023756980896, + "learning_rate": 6.293089818727001e-06, + "loss": 2.4594850540161133, + "step": 251920 + }, + { + "epoch": 0.6858309514087383, + "grad_norm": 2.1651523113250732, + "learning_rate": 6.291999454818047e-06, + "loss": 2.4564504623413086, + "step": 251940 + }, + { + "epoch": 0.6858853953994828, + "grad_norm": 2.234905481338501, + "learning_rate": 6.290909090909092e-06, + "loss": 2.4808767318725584, + "step": 251960 + }, + { + "epoch": 0.6859398393902273, + "grad_norm": 2.6179535388946533, + "learning_rate": 6.2898187270001365e-06, + "loss": 2.473373603820801, + "step": 251980 + }, + { + "epoch": 0.6859942833809718, + "grad_norm": 2.1871187686920166, + "learning_rate": 6.288728363091182e-06, + "loss": 2.476632308959961, + "step": 252000 + }, + { + "epoch": 0.6860487273717163, + "grad_norm": 2.2020814418792725, + "learning_rate": 6.287637999182228e-06, + "loss": 2.4673580169677733, + "step": 252020 + }, + { + "epoch": 0.6861031713624609, + "grad_norm": 2.172788381576538, + "learning_rate": 6.286547635273273e-06, + "loss": 2.4684125900268556, + "step": 252040 + }, + { + "epoch": 0.6861576153532054, + "grad_norm": 2.157499074935913, + "learning_rate": 6.285457271364318e-06, + "loss": 2.4722740173339846, + "step": 252060 + }, + { + "epoch": 0.6862120593439499, + "grad_norm": 2.1565210819244385, + "learning_rate": 6.284366907455364e-06, + "loss": 2.4820884704589843, + "step": 252080 + }, + { + "epoch": 0.6862665033346944, + "grad_norm": 2.1598315238952637, + "learning_rate": 6.28327654354641e-06, + "loss": 2.4557653427124024, + "step": 252100 + }, + { + "epoch": 0.6863209473254389, + "grad_norm": 2.195641040802002, + "learning_rate": 6.282186179637455e-06, + "loss": 2.481528472900391, + "step": 252120 + }, + { + "epoch": 0.6863753913161835, + "grad_norm": 2.3148555755615234, + "learning_rate": 6.2810958157285e-06, + "loss": 2.4607994079589846, + "step": 252140 + }, + { + "epoch": 0.686429835306928, + "grad_norm": 2.1681642532348633, + "learning_rate": 6.2800054518195454e-06, + "loss": 2.4526998519897463, + "step": 252160 + }, + { + "epoch": 0.6864842792976725, + "grad_norm": 2.216968536376953, + "learning_rate": 6.278915087910591e-06, + "loss": 2.464733695983887, + "step": 252180 + }, + { + "epoch": 0.686538723288417, + "grad_norm": 2.2089505195617676, + "learning_rate": 6.277824724001636e-06, + "loss": 2.4481784820556642, + "step": 252200 + }, + { + "epoch": 0.6865931672791615, + "grad_norm": 2.3285696506500244, + "learning_rate": 6.276734360092681e-06, + "loss": 2.4415964126586913, + "step": 252220 + }, + { + "epoch": 0.686647611269906, + "grad_norm": 2.056119441986084, + "learning_rate": 6.275643996183727e-06, + "loss": 2.4860980987548826, + "step": 252240 + }, + { + "epoch": 0.6867020552606506, + "grad_norm": 2.228698253631592, + "learning_rate": 6.274553632274773e-06, + "loss": 2.447612190246582, + "step": 252260 + }, + { + "epoch": 0.6867564992513951, + "grad_norm": 2.2180709838867188, + "learning_rate": 6.273463268365818e-06, + "loss": 2.4650373458862305, + "step": 252280 + }, + { + "epoch": 0.6868109432421396, + "grad_norm": 2.2037699222564697, + "learning_rate": 6.272372904456863e-06, + "loss": 2.4307775497436523, + "step": 252300 + }, + { + "epoch": 0.6868653872328842, + "grad_norm": 2.212880849838257, + "learning_rate": 6.2712825405479085e-06, + "loss": 2.4589052200317383, + "step": 252320 + }, + { + "epoch": 0.6869198312236287, + "grad_norm": 2.243600368499756, + "learning_rate": 6.270192176638954e-06, + "loss": 2.4458549499511717, + "step": 252340 + }, + { + "epoch": 0.6869742752143733, + "grad_norm": 2.180635929107666, + "learning_rate": 6.269101812729999e-06, + "loss": 2.4788673400878904, + "step": 252360 + }, + { + "epoch": 0.6870287192051178, + "grad_norm": 2.245570659637451, + "learning_rate": 6.268011448821044e-06, + "loss": 2.4748708724975588, + "step": 252380 + }, + { + "epoch": 0.6870831631958623, + "grad_norm": 2.1258769035339355, + "learning_rate": 6.26692108491209e-06, + "loss": 2.4864458084106444, + "step": 252400 + }, + { + "epoch": 0.6871376071866068, + "grad_norm": 2.1602303981781006, + "learning_rate": 6.265830721003136e-06, + "loss": 2.4738555908203126, + "step": 252420 + }, + { + "epoch": 0.6871920511773513, + "grad_norm": 2.3255879878997803, + "learning_rate": 6.264740357094181e-06, + "loss": 2.4828327178955076, + "step": 252440 + }, + { + "epoch": 0.6872464951680958, + "grad_norm": 2.2588019371032715, + "learning_rate": 6.263649993185226e-06, + "loss": 2.489181327819824, + "step": 252460 + }, + { + "epoch": 0.6873009391588404, + "grad_norm": 2.148300886154175, + "learning_rate": 6.262559629276272e-06, + "loss": 2.49456901550293, + "step": 252480 + }, + { + "epoch": 0.6873553831495849, + "grad_norm": 2.181274175643921, + "learning_rate": 6.2614692653673174e-06, + "loss": 2.456198310852051, + "step": 252500 + }, + { + "epoch": 0.6874098271403294, + "grad_norm": 2.201598644256592, + "learning_rate": 6.260378901458362e-06, + "loss": 2.470104217529297, + "step": 252520 + }, + { + "epoch": 0.6874642711310739, + "grad_norm": 2.0969159603118896, + "learning_rate": 6.259288537549407e-06, + "loss": 2.4461463928222655, + "step": 252540 + }, + { + "epoch": 0.6875187151218184, + "grad_norm": 2.207338809967041, + "learning_rate": 6.258198173640453e-06, + "loss": 2.471756362915039, + "step": 252560 + }, + { + "epoch": 0.687573159112563, + "grad_norm": 2.206662178039551, + "learning_rate": 6.257107809731499e-06, + "loss": 2.431795120239258, + "step": 252580 + }, + { + "epoch": 0.6876276031033075, + "grad_norm": 2.1674482822418213, + "learning_rate": 6.256017445822544e-06, + "loss": 2.4379804611206053, + "step": 252600 + }, + { + "epoch": 0.687682047094052, + "grad_norm": 2.1178836822509766, + "learning_rate": 6.254927081913589e-06, + "loss": 2.4770957946777346, + "step": 252620 + }, + { + "epoch": 0.6877364910847965, + "grad_norm": 2.1136932373046875, + "learning_rate": 6.253836718004635e-06, + "loss": 2.485418701171875, + "step": 252640 + }, + { + "epoch": 0.687790935075541, + "grad_norm": 2.17014741897583, + "learning_rate": 6.2527463540956805e-06, + "loss": 2.4748674392700196, + "step": 252660 + }, + { + "epoch": 0.6878453790662855, + "grad_norm": 2.095559597015381, + "learning_rate": 6.2516559901867255e-06, + "loss": 2.476960372924805, + "step": 252680 + }, + { + "epoch": 0.6878998230570301, + "grad_norm": 2.2020065784454346, + "learning_rate": 6.2505656262777704e-06, + "loss": 2.4505941390991213, + "step": 252700 + }, + { + "epoch": 0.6879542670477746, + "grad_norm": 2.1338465213775635, + "learning_rate": 6.249475262368816e-06, + "loss": 2.4480060577392577, + "step": 252720 + }, + { + "epoch": 0.6880087110385191, + "grad_norm": 4.742386817932129, + "learning_rate": 6.248384898459861e-06, + "loss": 2.472947883605957, + "step": 252740 + }, + { + "epoch": 0.6880631550292636, + "grad_norm": 2.180300235748291, + "learning_rate": 6.247294534550907e-06, + "loss": 2.4841083526611327, + "step": 252760 + }, + { + "epoch": 0.6881175990200081, + "grad_norm": 2.187344789505005, + "learning_rate": 6.246204170641952e-06, + "loss": 2.463958740234375, + "step": 252780 + }, + { + "epoch": 0.6881720430107527, + "grad_norm": 2.257136106491089, + "learning_rate": 6.245113806732998e-06, + "loss": 2.467138671875, + "step": 252800 + }, + { + "epoch": 0.6882264870014972, + "grad_norm": 2.067746877670288, + "learning_rate": 6.244023442824043e-06, + "loss": 2.4517375946044924, + "step": 252820 + }, + { + "epoch": 0.6882809309922417, + "grad_norm": 2.2156481742858887, + "learning_rate": 6.2429330789150886e-06, + "loss": 2.4771013259887695, + "step": 252840 + }, + { + "epoch": 0.6883353749829862, + "grad_norm": 2.1195600032806396, + "learning_rate": 6.2418427150061335e-06, + "loss": 2.459140396118164, + "step": 252860 + }, + { + "epoch": 0.6883898189737307, + "grad_norm": 2.1424081325531006, + "learning_rate": 6.240752351097179e-06, + "loss": 2.4638442993164062, + "step": 252880 + }, + { + "epoch": 0.6884442629644753, + "grad_norm": 2.0781562328338623, + "learning_rate": 6.239661987188224e-06, + "loss": 2.467732238769531, + "step": 252900 + }, + { + "epoch": 0.6884987069552199, + "grad_norm": 2.2697229385375977, + "learning_rate": 6.23857162327927e-06, + "loss": 2.4660614013671873, + "step": 252920 + }, + { + "epoch": 0.6885531509459644, + "grad_norm": 2.2103121280670166, + "learning_rate": 6.237481259370315e-06, + "loss": 2.4780605316162108, + "step": 252940 + }, + { + "epoch": 0.6886075949367089, + "grad_norm": 2.199937105178833, + "learning_rate": 6.236390895461361e-06, + "loss": 2.4800647735595702, + "step": 252960 + }, + { + "epoch": 0.6886620389274534, + "grad_norm": 2.1817007064819336, + "learning_rate": 6.235300531552406e-06, + "loss": 2.4863260269165037, + "step": 252980 + }, + { + "epoch": 0.6887164829181979, + "grad_norm": 2.211442470550537, + "learning_rate": 6.234210167643452e-06, + "loss": 2.457847785949707, + "step": 253000 + }, + { + "epoch": 0.6887709269089424, + "grad_norm": 2.162946939468384, + "learning_rate": 6.233119803734497e-06, + "loss": 2.440597343444824, + "step": 253020 + }, + { + "epoch": 0.688825370899687, + "grad_norm": 2.1716198921203613, + "learning_rate": 6.232029439825542e-06, + "loss": 2.446929931640625, + "step": 253040 + }, + { + "epoch": 0.6888798148904315, + "grad_norm": 2.3407020568847656, + "learning_rate": 6.230939075916587e-06, + "loss": 2.4730623245239256, + "step": 253060 + }, + { + "epoch": 0.688934258881176, + "grad_norm": 2.0629096031188965, + "learning_rate": 6.229848712007633e-06, + "loss": 2.4702785491943358, + "step": 253080 + }, + { + "epoch": 0.6889887028719205, + "grad_norm": 2.181076765060425, + "learning_rate": 6.228758348098678e-06, + "loss": 2.4720497131347656, + "step": 253100 + }, + { + "epoch": 0.689043146862665, + "grad_norm": 2.0681724548339844, + "learning_rate": 6.227667984189724e-06, + "loss": 2.4567455291748046, + "step": 253120 + }, + { + "epoch": 0.6890975908534096, + "grad_norm": 2.1969425678253174, + "learning_rate": 6.226577620280769e-06, + "loss": 2.4542083740234375, + "step": 253140 + }, + { + "epoch": 0.6891520348441541, + "grad_norm": 2.1300487518310547, + "learning_rate": 6.225487256371815e-06, + "loss": 2.4654626846313477, + "step": 253160 + }, + { + "epoch": 0.6892064788348986, + "grad_norm": 2.2577261924743652, + "learning_rate": 6.22439689246286e-06, + "loss": 2.4563690185546876, + "step": 253180 + }, + { + "epoch": 0.6892609228256431, + "grad_norm": 2.359915256500244, + "learning_rate": 6.2233065285539055e-06, + "loss": 2.473368263244629, + "step": 253200 + }, + { + "epoch": 0.6893153668163876, + "grad_norm": 2.179844856262207, + "learning_rate": 6.2222161646449505e-06, + "loss": 2.4696269989013673, + "step": 253220 + }, + { + "epoch": 0.6893698108071321, + "grad_norm": 2.195085287094116, + "learning_rate": 6.221125800735996e-06, + "loss": 2.4648439407348635, + "step": 253240 + }, + { + "epoch": 0.6894242547978767, + "grad_norm": 2.0510575771331787, + "learning_rate": 6.220035436827041e-06, + "loss": 2.4762636184692384, + "step": 253260 + }, + { + "epoch": 0.6894786987886212, + "grad_norm": 2.826329231262207, + "learning_rate": 6.218945072918087e-06, + "loss": 2.46229248046875, + "step": 253280 + }, + { + "epoch": 0.6895331427793657, + "grad_norm": 2.1272616386413574, + "learning_rate": 6.217854709009132e-06, + "loss": 2.458238983154297, + "step": 253300 + }, + { + "epoch": 0.6895875867701102, + "grad_norm": 2.201934337615967, + "learning_rate": 6.216764345100178e-06, + "loss": 2.4338802337646483, + "step": 253320 + }, + { + "epoch": 0.6896420307608547, + "grad_norm": 2.1581640243530273, + "learning_rate": 6.215673981191223e-06, + "loss": 2.465610694885254, + "step": 253340 + }, + { + "epoch": 0.6896964747515993, + "grad_norm": 2.2886099815368652, + "learning_rate": 6.214583617282269e-06, + "loss": 2.452173042297363, + "step": 253360 + }, + { + "epoch": 0.6897509187423438, + "grad_norm": 2.041555881500244, + "learning_rate": 6.2134932533733135e-06, + "loss": 2.4817035675048826, + "step": 253380 + }, + { + "epoch": 0.6898053627330883, + "grad_norm": 2.1098368167877197, + "learning_rate": 6.212402889464359e-06, + "loss": 2.503377151489258, + "step": 253400 + }, + { + "epoch": 0.6898598067238328, + "grad_norm": 2.1883552074432373, + "learning_rate": 6.211312525555404e-06, + "loss": 2.4419397354125976, + "step": 253420 + }, + { + "epoch": 0.6899142507145773, + "grad_norm": 2.107374906539917, + "learning_rate": 6.21022216164645e-06, + "loss": 2.479559898376465, + "step": 253440 + }, + { + "epoch": 0.6899686947053218, + "grad_norm": 2.6450817584991455, + "learning_rate": 6.209131797737495e-06, + "loss": 2.4912975311279295, + "step": 253460 + }, + { + "epoch": 0.6900231386960665, + "grad_norm": 2.309885263442993, + "learning_rate": 6.208041433828541e-06, + "loss": 2.440167045593262, + "step": 253480 + }, + { + "epoch": 0.690077582686811, + "grad_norm": 2.0616931915283203, + "learning_rate": 6.206951069919586e-06, + "loss": 2.4841541290283202, + "step": 253500 + }, + { + "epoch": 0.6901320266775555, + "grad_norm": 2.176485061645508, + "learning_rate": 6.205860706010632e-06, + "loss": 2.458058547973633, + "step": 253520 + }, + { + "epoch": 0.6901864706683, + "grad_norm": 2.1262593269348145, + "learning_rate": 6.204770342101677e-06, + "loss": 2.492025947570801, + "step": 253540 + }, + { + "epoch": 0.6902409146590445, + "grad_norm": 2.4170093536376953, + "learning_rate": 6.2036799781927224e-06, + "loss": 2.4422916412353515, + "step": 253560 + }, + { + "epoch": 0.6902953586497891, + "grad_norm": 2.1894140243530273, + "learning_rate": 6.202589614283768e-06, + "loss": 2.4637847900390626, + "step": 253580 + }, + { + "epoch": 0.6903498026405336, + "grad_norm": 2.281418800354004, + "learning_rate": 6.201499250374813e-06, + "loss": 2.4686044692993163, + "step": 253600 + }, + { + "epoch": 0.6904042466312781, + "grad_norm": 3.9151549339294434, + "learning_rate": 6.200408886465858e-06, + "loss": 2.4941930770874023, + "step": 253620 + }, + { + "epoch": 0.6904586906220226, + "grad_norm": 2.1414151191711426, + "learning_rate": 6.199318522556904e-06, + "loss": 2.4876325607299803, + "step": 253640 + }, + { + "epoch": 0.6905131346127671, + "grad_norm": 2.130913019180298, + "learning_rate": 6.19822815864795e-06, + "loss": 2.4566581726074217, + "step": 253660 + }, + { + "epoch": 0.6905675786035116, + "grad_norm": 2.1278862953186035, + "learning_rate": 6.197137794738995e-06, + "loss": 2.4839803695678713, + "step": 253680 + }, + { + "epoch": 0.6906220225942562, + "grad_norm": 2.505344867706299, + "learning_rate": 6.19604743083004e-06, + "loss": 2.475307846069336, + "step": 253700 + }, + { + "epoch": 0.6906764665850007, + "grad_norm": 2.1700665950775146, + "learning_rate": 6.1949570669210855e-06, + "loss": 2.4666860580444334, + "step": 253720 + }, + { + "epoch": 0.6907309105757452, + "grad_norm": 2.23818302154541, + "learning_rate": 6.193866703012131e-06, + "loss": 2.4776046752929686, + "step": 253740 + }, + { + "epoch": 0.6907853545664897, + "grad_norm": 2.4569883346557617, + "learning_rate": 6.192776339103176e-06, + "loss": 2.468303108215332, + "step": 253760 + }, + { + "epoch": 0.6908397985572342, + "grad_norm": 2.112875461578369, + "learning_rate": 6.191685975194221e-06, + "loss": 2.445184898376465, + "step": 253780 + }, + { + "epoch": 0.6908942425479788, + "grad_norm": 2.325356960296631, + "learning_rate": 6.190595611285267e-06, + "loss": 2.4733118057250976, + "step": 253800 + }, + { + "epoch": 0.6909486865387233, + "grad_norm": 2.242823600769043, + "learning_rate": 6.189505247376313e-06, + "loss": 2.466948318481445, + "step": 253820 + }, + { + "epoch": 0.6910031305294678, + "grad_norm": 2.2080578804016113, + "learning_rate": 6.188414883467357e-06, + "loss": 2.4478464126586914, + "step": 253840 + }, + { + "epoch": 0.6910575745202123, + "grad_norm": 2.1780107021331787, + "learning_rate": 6.187324519558403e-06, + "loss": 2.457563018798828, + "step": 253860 + }, + { + "epoch": 0.6911120185109568, + "grad_norm": 2.330148935317993, + "learning_rate": 6.186234155649449e-06, + "loss": 2.45691032409668, + "step": 253880 + }, + { + "epoch": 0.6911664625017013, + "grad_norm": 2.1304569244384766, + "learning_rate": 6.185143791740494e-06, + "loss": 2.462854194641113, + "step": 253900 + }, + { + "epoch": 0.6912209064924459, + "grad_norm": 2.261897563934326, + "learning_rate": 6.1840534278315385e-06, + "loss": 2.485154724121094, + "step": 253920 + }, + { + "epoch": 0.6912753504831904, + "grad_norm": 2.0626354217529297, + "learning_rate": 6.182963063922584e-06, + "loss": 2.5025619506835937, + "step": 253940 + }, + { + "epoch": 0.6913297944739349, + "grad_norm": 2.200496196746826, + "learning_rate": 6.18187270001363e-06, + "loss": 2.4579071044921874, + "step": 253960 + }, + { + "epoch": 0.6913842384646794, + "grad_norm": 2.14345383644104, + "learning_rate": 6.180782336104676e-06, + "loss": 2.4889631271362305, + "step": 253980 + }, + { + "epoch": 0.691438682455424, + "grad_norm": 2.2906219959259033, + "learning_rate": 6.17969197219572e-06, + "loss": 2.485426330566406, + "step": 254000 + }, + { + "epoch": 0.6914931264461684, + "grad_norm": 2.2245020866394043, + "learning_rate": 6.178601608286766e-06, + "loss": 2.481080436706543, + "step": 254020 + }, + { + "epoch": 0.6915475704369131, + "grad_norm": 2.0383646488189697, + "learning_rate": 6.177511244377812e-06, + "loss": 2.455729103088379, + "step": 254040 + }, + { + "epoch": 0.6916020144276576, + "grad_norm": 2.1060547828674316, + "learning_rate": 6.1764208804688575e-06, + "loss": 2.486200141906738, + "step": 254060 + }, + { + "epoch": 0.6916564584184021, + "grad_norm": 2.1536319255828857, + "learning_rate": 6.175330516559902e-06, + "loss": 2.4480764389038088, + "step": 254080 + }, + { + "epoch": 0.6917109024091466, + "grad_norm": 2.196636915206909, + "learning_rate": 6.1742401526509474e-06, + "loss": 2.4747743606567383, + "step": 254100 + }, + { + "epoch": 0.6917653463998911, + "grad_norm": 2.086031675338745, + "learning_rate": 6.173149788741993e-06, + "loss": 2.4416824340820313, + "step": 254120 + }, + { + "epoch": 0.6918197903906357, + "grad_norm": 2.098837375640869, + "learning_rate": 6.172059424833039e-06, + "loss": 2.453285980224609, + "step": 254140 + }, + { + "epoch": 0.6918742343813802, + "grad_norm": 2.215332508087158, + "learning_rate": 6.170969060924083e-06, + "loss": 2.4352981567382814, + "step": 254160 + }, + { + "epoch": 0.6919286783721247, + "grad_norm": 2.092223882675171, + "learning_rate": 6.169878697015129e-06, + "loss": 2.438361167907715, + "step": 254180 + }, + { + "epoch": 0.6919831223628692, + "grad_norm": 2.1720352172851562, + "learning_rate": 6.168788333106175e-06, + "loss": 2.480128860473633, + "step": 254200 + }, + { + "epoch": 0.6920375663536137, + "grad_norm": 2.176614999771118, + "learning_rate": 6.167697969197221e-06, + "loss": 2.48846492767334, + "step": 254220 + }, + { + "epoch": 0.6920920103443582, + "grad_norm": 2.1223227977752686, + "learning_rate": 6.166607605288265e-06, + "loss": 2.4735687255859373, + "step": 254240 + }, + { + "epoch": 0.6921464543351028, + "grad_norm": 2.057220458984375, + "learning_rate": 6.1655172413793105e-06, + "loss": 2.425617790222168, + "step": 254260 + }, + { + "epoch": 0.6922008983258473, + "grad_norm": 2.259631633758545, + "learning_rate": 6.164426877470356e-06, + "loss": 2.4570356369018556, + "step": 254280 + }, + { + "epoch": 0.6922553423165918, + "grad_norm": 2.207551956176758, + "learning_rate": 6.163336513561402e-06, + "loss": 2.4835603713989256, + "step": 254300 + }, + { + "epoch": 0.6923097863073363, + "grad_norm": 2.2746593952178955, + "learning_rate": 6.162246149652446e-06, + "loss": 2.439605140686035, + "step": 254320 + }, + { + "epoch": 0.6923642302980808, + "grad_norm": 2.1302993297576904, + "learning_rate": 6.161155785743492e-06, + "loss": 2.457821273803711, + "step": 254340 + }, + { + "epoch": 0.6924186742888254, + "grad_norm": 2.193956136703491, + "learning_rate": 6.160065421834538e-06, + "loss": 2.4644037246704102, + "step": 254360 + }, + { + "epoch": 0.6924731182795699, + "grad_norm": 2.2537055015563965, + "learning_rate": 6.158975057925584e-06, + "loss": 2.4698795318603515, + "step": 254380 + }, + { + "epoch": 0.6925275622703144, + "grad_norm": 2.1730575561523438, + "learning_rate": 6.157884694016628e-06, + "loss": 2.4791563034057615, + "step": 254400 + }, + { + "epoch": 0.6925820062610589, + "grad_norm": 2.4009928703308105, + "learning_rate": 6.156794330107674e-06, + "loss": 2.4544597625732423, + "step": 254420 + }, + { + "epoch": 0.6926364502518034, + "grad_norm": 2.053520441055298, + "learning_rate": 6.155703966198719e-06, + "loss": 2.4713342666625975, + "step": 254440 + }, + { + "epoch": 0.6926908942425479, + "grad_norm": 2.2778377532958984, + "learning_rate": 6.154613602289765e-06, + "loss": 2.477012825012207, + "step": 254460 + }, + { + "epoch": 0.6927453382332925, + "grad_norm": 2.0903141498565674, + "learning_rate": 6.153523238380809e-06, + "loss": 2.461705780029297, + "step": 254480 + }, + { + "epoch": 0.692799782224037, + "grad_norm": 2.1445775032043457, + "learning_rate": 6.152432874471855e-06, + "loss": 2.472817611694336, + "step": 254500 + }, + { + "epoch": 0.6928542262147815, + "grad_norm": 2.192462921142578, + "learning_rate": 6.151342510562901e-06, + "loss": 2.4522768020629884, + "step": 254520 + }, + { + "epoch": 0.692908670205526, + "grad_norm": 2.1591739654541016, + "learning_rate": 6.150252146653947e-06, + "loss": 2.4566890716552736, + "step": 254540 + }, + { + "epoch": 0.6929631141962705, + "grad_norm": 2.2585036754608154, + "learning_rate": 6.149161782744991e-06, + "loss": 2.469631385803223, + "step": 254560 + }, + { + "epoch": 0.6930175581870152, + "grad_norm": 2.293367862701416, + "learning_rate": 6.148071418836037e-06, + "loss": 2.468039894104004, + "step": 254580 + }, + { + "epoch": 0.6930720021777597, + "grad_norm": 2.0431978702545166, + "learning_rate": 6.1469810549270825e-06, + "loss": 2.4620447158813477, + "step": 254600 + }, + { + "epoch": 0.6931264461685042, + "grad_norm": 2.1714043617248535, + "learning_rate": 6.145890691018128e-06, + "loss": 2.4627246856689453, + "step": 254620 + }, + { + "epoch": 0.6931808901592487, + "grad_norm": 2.199202537536621, + "learning_rate": 6.144800327109172e-06, + "loss": 2.4830657958984377, + "step": 254640 + }, + { + "epoch": 0.6932353341499932, + "grad_norm": 2.135573387145996, + "learning_rate": 6.143709963200218e-06, + "loss": 2.4513132095336916, + "step": 254660 + }, + { + "epoch": 0.6932897781407377, + "grad_norm": 4.055387020111084, + "learning_rate": 6.142619599291264e-06, + "loss": 2.45244140625, + "step": 254680 + }, + { + "epoch": 0.6933442221314823, + "grad_norm": 2.2094016075134277, + "learning_rate": 6.14152923538231e-06, + "loss": 2.475177001953125, + "step": 254700 + }, + { + "epoch": 0.6933986661222268, + "grad_norm": 2.1441917419433594, + "learning_rate": 6.140438871473354e-06, + "loss": 2.432975006103516, + "step": 254720 + }, + { + "epoch": 0.6934531101129713, + "grad_norm": 2.1238245964050293, + "learning_rate": 6.1393485075644e-06, + "loss": 2.449440574645996, + "step": 254740 + }, + { + "epoch": 0.6935075541037158, + "grad_norm": 2.47666597366333, + "learning_rate": 6.138258143655446e-06, + "loss": 2.485744857788086, + "step": 254760 + }, + { + "epoch": 0.6935619980944603, + "grad_norm": 2.6938021183013916, + "learning_rate": 6.137167779746491e-06, + "loss": 2.4307119369506838, + "step": 254780 + }, + { + "epoch": 0.6936164420852049, + "grad_norm": 2.098723888397217, + "learning_rate": 6.1360774158375355e-06, + "loss": 2.4688596725463867, + "step": 254800 + }, + { + "epoch": 0.6936708860759494, + "grad_norm": 3.5833992958068848, + "learning_rate": 6.134987051928581e-06, + "loss": 2.4800142288208007, + "step": 254820 + }, + { + "epoch": 0.6937253300666939, + "grad_norm": 2.2784132957458496, + "learning_rate": 6.133896688019627e-06, + "loss": 2.4608762741088865, + "step": 254840 + }, + { + "epoch": 0.6937797740574384, + "grad_norm": 2.164108991622925, + "learning_rate": 6.132806324110673e-06, + "loss": 2.47775936126709, + "step": 254860 + }, + { + "epoch": 0.6938342180481829, + "grad_norm": 2.189681053161621, + "learning_rate": 6.131715960201717e-06, + "loss": 2.465667152404785, + "step": 254880 + }, + { + "epoch": 0.6938886620389274, + "grad_norm": 2.1649816036224365, + "learning_rate": 6.130625596292763e-06, + "loss": 2.4883445739746093, + "step": 254900 + }, + { + "epoch": 0.693943106029672, + "grad_norm": 2.3131773471832275, + "learning_rate": 6.129535232383809e-06, + "loss": 2.4653305053710937, + "step": 254920 + }, + { + "epoch": 0.6939975500204165, + "grad_norm": 2.2747416496276855, + "learning_rate": 6.1284448684748545e-06, + "loss": 2.4807708740234373, + "step": 254940 + }, + { + "epoch": 0.694051994011161, + "grad_norm": 2.210329055786133, + "learning_rate": 6.127354504565899e-06, + "loss": 2.4805355072021484, + "step": 254960 + }, + { + "epoch": 0.6941064380019055, + "grad_norm": 3.2022366523742676, + "learning_rate": 6.126264140656944e-06, + "loss": 2.462560272216797, + "step": 254980 + }, + { + "epoch": 0.69416088199265, + "grad_norm": 2.2246975898742676, + "learning_rate": 6.12517377674799e-06, + "loss": 2.4928335189819335, + "step": 255000 + }, + { + "epoch": 0.6942153259833946, + "grad_norm": 2.096008539199829, + "learning_rate": 6.124083412839036e-06, + "loss": 2.4792818069458007, + "step": 255020 + }, + { + "epoch": 0.6942697699741391, + "grad_norm": 2.1561779975891113, + "learning_rate": 6.12299304893008e-06, + "loss": 2.4771139144897463, + "step": 255040 + }, + { + "epoch": 0.6943242139648836, + "grad_norm": 2.1815927028656006, + "learning_rate": 6.121902685021126e-06, + "loss": 2.4288793563842774, + "step": 255060 + }, + { + "epoch": 0.6943786579556281, + "grad_norm": 2.196958303451538, + "learning_rate": 6.120812321112172e-06, + "loss": 2.4796268463134767, + "step": 255080 + }, + { + "epoch": 0.6944331019463726, + "grad_norm": 2.1370842456817627, + "learning_rate": 6.1197219572032176e-06, + "loss": 2.4720174789428713, + "step": 255100 + }, + { + "epoch": 0.6944875459371171, + "grad_norm": 2.2975523471832275, + "learning_rate": 6.118631593294262e-06, + "loss": 2.4848791122436524, + "step": 255120 + }, + { + "epoch": 0.6945419899278618, + "grad_norm": 2.77762770652771, + "learning_rate": 6.1175412293853075e-06, + "loss": 2.438604736328125, + "step": 255140 + }, + { + "epoch": 0.6945964339186063, + "grad_norm": 2.08358097076416, + "learning_rate": 6.116450865476353e-06, + "loss": 2.4724023818969725, + "step": 255160 + }, + { + "epoch": 0.6946508779093508, + "grad_norm": 2.117178440093994, + "learning_rate": 6.115360501567399e-06, + "loss": 2.4728023529052736, + "step": 255180 + }, + { + "epoch": 0.6947053219000953, + "grad_norm": 2.228181838989258, + "learning_rate": 6.114270137658443e-06, + "loss": 2.4633941650390625, + "step": 255200 + }, + { + "epoch": 0.6947597658908398, + "grad_norm": 2.2374918460845947, + "learning_rate": 6.113179773749489e-06, + "loss": 2.488844299316406, + "step": 255220 + }, + { + "epoch": 0.6948142098815843, + "grad_norm": 2.239380121231079, + "learning_rate": 6.112089409840535e-06, + "loss": 2.4530511856079102, + "step": 255240 + }, + { + "epoch": 0.6948686538723289, + "grad_norm": 2.119196891784668, + "learning_rate": 6.110999045931581e-06, + "loss": 2.474680519104004, + "step": 255260 + }, + { + "epoch": 0.6949230978630734, + "grad_norm": 2.2011029720306396, + "learning_rate": 6.1099086820226265e-06, + "loss": 2.4806503295898437, + "step": 255280 + }, + { + "epoch": 0.6949775418538179, + "grad_norm": 3.6665077209472656, + "learning_rate": 6.1088183181136706e-06, + "loss": 2.45779972076416, + "step": 255300 + }, + { + "epoch": 0.6950319858445624, + "grad_norm": 2.1476471424102783, + "learning_rate": 6.107727954204716e-06, + "loss": 2.497176170349121, + "step": 255320 + }, + { + "epoch": 0.6950864298353069, + "grad_norm": 2.3449113368988037, + "learning_rate": 6.106637590295762e-06, + "loss": 2.478171730041504, + "step": 255340 + }, + { + "epoch": 0.6951408738260515, + "grad_norm": 2.391514778137207, + "learning_rate": 6.105547226386808e-06, + "loss": 2.4498594284057615, + "step": 255360 + }, + { + "epoch": 0.695195317816796, + "grad_norm": 2.184908866882324, + "learning_rate": 6.104456862477852e-06, + "loss": 2.4873586654663087, + "step": 255380 + }, + { + "epoch": 0.6952497618075405, + "grad_norm": 2.2852182388305664, + "learning_rate": 6.103366498568898e-06, + "loss": 2.4417146682739257, + "step": 255400 + }, + { + "epoch": 0.695304205798285, + "grad_norm": 2.5787787437438965, + "learning_rate": 6.102276134659944e-06, + "loss": 2.440816307067871, + "step": 255420 + }, + { + "epoch": 0.6953586497890295, + "grad_norm": 2.1890759468078613, + "learning_rate": 6.1011857707509895e-06, + "loss": 2.4735828399658204, + "step": 255440 + }, + { + "epoch": 0.695413093779774, + "grad_norm": 2.3324105739593506, + "learning_rate": 6.100095406842034e-06, + "loss": 2.4643964767456055, + "step": 255460 + }, + { + "epoch": 0.6954675377705186, + "grad_norm": 2.6519999504089355, + "learning_rate": 6.0990050429330795e-06, + "loss": 2.490043067932129, + "step": 255480 + }, + { + "epoch": 0.6955219817612631, + "grad_norm": 2.114950656890869, + "learning_rate": 6.097914679024125e-06, + "loss": 2.4824594497680663, + "step": 255500 + }, + { + "epoch": 0.6955764257520076, + "grad_norm": 2.099684953689575, + "learning_rate": 6.096824315115171e-06, + "loss": 2.467336654663086, + "step": 255520 + }, + { + "epoch": 0.6956308697427521, + "grad_norm": 2.143617868423462, + "learning_rate": 6.095733951206215e-06, + "loss": 2.473716926574707, + "step": 255540 + }, + { + "epoch": 0.6956853137334966, + "grad_norm": 2.117720365524292, + "learning_rate": 6.094643587297261e-06, + "loss": 2.4417652130126952, + "step": 255560 + }, + { + "epoch": 0.6957397577242412, + "grad_norm": 2.2182695865631104, + "learning_rate": 6.093553223388307e-06, + "loss": 2.471840667724609, + "step": 255580 + }, + { + "epoch": 0.6957942017149857, + "grad_norm": 2.3886868953704834, + "learning_rate": 6.092462859479353e-06, + "loss": 2.489156723022461, + "step": 255600 + }, + { + "epoch": 0.6958486457057302, + "grad_norm": 2.152526378631592, + "learning_rate": 6.091372495570397e-06, + "loss": 2.4788169860839844, + "step": 255620 + }, + { + "epoch": 0.6959030896964747, + "grad_norm": 2.3160784244537354, + "learning_rate": 6.0902821316614426e-06, + "loss": 2.501434326171875, + "step": 255640 + }, + { + "epoch": 0.6959575336872192, + "grad_norm": 2.2462356090545654, + "learning_rate": 6.089191767752488e-06, + "loss": 2.4700592041015623, + "step": 255660 + }, + { + "epoch": 0.6960119776779637, + "grad_norm": 2.0768661499023438, + "learning_rate": 6.088101403843534e-06, + "loss": 2.476533317565918, + "step": 255680 + }, + { + "epoch": 0.6960664216687084, + "grad_norm": 2.270782709121704, + "learning_rate": 6.087011039934578e-06, + "loss": 2.449957084655762, + "step": 255700 + }, + { + "epoch": 0.6961208656594529, + "grad_norm": 2.088684320449829, + "learning_rate": 6.085920676025624e-06, + "loss": 2.456264114379883, + "step": 255720 + }, + { + "epoch": 0.6961753096501974, + "grad_norm": 2.118908405303955, + "learning_rate": 6.08483031211667e-06, + "loss": 2.4636634826660155, + "step": 255740 + }, + { + "epoch": 0.6962297536409419, + "grad_norm": 2.1346755027770996, + "learning_rate": 6.083739948207716e-06, + "loss": 2.456871223449707, + "step": 255760 + }, + { + "epoch": 0.6962841976316864, + "grad_norm": 2.1295080184936523, + "learning_rate": 6.08264958429876e-06, + "loss": 2.4868703842163087, + "step": 255780 + }, + { + "epoch": 0.696338641622431, + "grad_norm": 2.1304473876953125, + "learning_rate": 6.081559220389806e-06, + "loss": 2.465513229370117, + "step": 255800 + }, + { + "epoch": 0.6963930856131755, + "grad_norm": 2.1240365505218506, + "learning_rate": 6.0804688564808514e-06, + "loss": 2.5113813400268556, + "step": 255820 + }, + { + "epoch": 0.69644752960392, + "grad_norm": 2.248262882232666, + "learning_rate": 6.079378492571896e-06, + "loss": 2.4993703842163084, + "step": 255840 + }, + { + "epoch": 0.6965019735946645, + "grad_norm": 2.0902676582336426, + "learning_rate": 6.078288128662941e-06, + "loss": 2.467307281494141, + "step": 255860 + }, + { + "epoch": 0.696556417585409, + "grad_norm": 2.210524082183838, + "learning_rate": 6.077197764753987e-06, + "loss": 2.473190498352051, + "step": 255880 + }, + { + "epoch": 0.6966108615761535, + "grad_norm": 2.208458423614502, + "learning_rate": 6.076107400845033e-06, + "loss": 2.4393285751342773, + "step": 255900 + }, + { + "epoch": 0.6966653055668981, + "grad_norm": 2.1458845138549805, + "learning_rate": 6.075017036936078e-06, + "loss": 2.470427131652832, + "step": 255920 + }, + { + "epoch": 0.6967197495576426, + "grad_norm": 2.1877613067626953, + "learning_rate": 6.073926673027123e-06, + "loss": 2.486515426635742, + "step": 255940 + }, + { + "epoch": 0.6967741935483871, + "grad_norm": 2.1414692401885986, + "learning_rate": 6.072836309118169e-06, + "loss": 2.494556999206543, + "step": 255960 + }, + { + "epoch": 0.6968286375391316, + "grad_norm": 2.171194076538086, + "learning_rate": 6.0717459452092145e-06, + "loss": 2.4779653549194336, + "step": 255980 + }, + { + "epoch": 0.6968830815298761, + "grad_norm": 2.181830644607544, + "learning_rate": 6.0706555813002595e-06, + "loss": 2.5037868499755858, + "step": 256000 + }, + { + "epoch": 0.6969375255206207, + "grad_norm": 2.1566059589385986, + "learning_rate": 6.0695652173913045e-06, + "loss": 2.459769439697266, + "step": 256020 + }, + { + "epoch": 0.6969919695113652, + "grad_norm": 2.1743862628936768, + "learning_rate": 6.06847485348235e-06, + "loss": 2.493473243713379, + "step": 256040 + }, + { + "epoch": 0.6970464135021097, + "grad_norm": 2.1423451900482178, + "learning_rate": 6.067384489573396e-06, + "loss": 2.46411075592041, + "step": 256060 + }, + { + "epoch": 0.6971008574928542, + "grad_norm": 2.1467888355255127, + "learning_rate": 6.066294125664441e-06, + "loss": 2.4499488830566407, + "step": 256080 + }, + { + "epoch": 0.6971553014835987, + "grad_norm": 2.280094623565674, + "learning_rate": 6.065203761755486e-06, + "loss": 2.455854606628418, + "step": 256100 + }, + { + "epoch": 0.6972097454743432, + "grad_norm": 2.1347594261169434, + "learning_rate": 6.064113397846532e-06, + "loss": 2.476278877258301, + "step": 256120 + }, + { + "epoch": 0.6972641894650878, + "grad_norm": 2.2402877807617188, + "learning_rate": 6.063023033937578e-06, + "loss": 2.4484979629516603, + "step": 256140 + }, + { + "epoch": 0.6973186334558323, + "grad_norm": 2.4149107933044434, + "learning_rate": 6.061932670028623e-06, + "loss": 2.489321708679199, + "step": 256160 + }, + { + "epoch": 0.6973730774465768, + "grad_norm": 2.2702291011810303, + "learning_rate": 6.0608423061196675e-06, + "loss": 2.4467500686645507, + "step": 256180 + }, + { + "epoch": 0.6974275214373213, + "grad_norm": 2.176558256149292, + "learning_rate": 6.059751942210713e-06, + "loss": 2.457779121398926, + "step": 256200 + }, + { + "epoch": 0.6974819654280658, + "grad_norm": 2.1562492847442627, + "learning_rate": 6.058661578301759e-06, + "loss": 2.484665298461914, + "step": 256220 + }, + { + "epoch": 0.6975364094188103, + "grad_norm": 2.0961732864379883, + "learning_rate": 6.057571214392804e-06, + "loss": 2.4959085464477537, + "step": 256240 + }, + { + "epoch": 0.697590853409555, + "grad_norm": 2.4115102291107178, + "learning_rate": 6.056480850483849e-06, + "loss": 2.466206169128418, + "step": 256260 + }, + { + "epoch": 0.6976452974002995, + "grad_norm": 2.5792083740234375, + "learning_rate": 6.055390486574895e-06, + "loss": 2.4729084014892577, + "step": 256280 + }, + { + "epoch": 0.697699741391044, + "grad_norm": 2.212099552154541, + "learning_rate": 6.054300122665941e-06, + "loss": 2.4852567672729493, + "step": 256300 + }, + { + "epoch": 0.6977541853817885, + "grad_norm": 2.1105613708496094, + "learning_rate": 6.053209758756986e-06, + "loss": 2.447084426879883, + "step": 256320 + }, + { + "epoch": 0.697808629372533, + "grad_norm": 2.155029773712158, + "learning_rate": 6.052119394848031e-06, + "loss": 2.462441825866699, + "step": 256340 + }, + { + "epoch": 0.6978630733632776, + "grad_norm": 2.1956348419189453, + "learning_rate": 6.0510290309390764e-06, + "loss": 2.453537178039551, + "step": 256360 + }, + { + "epoch": 0.6979175173540221, + "grad_norm": 2.2084169387817383, + "learning_rate": 6.049938667030122e-06, + "loss": 2.472011375427246, + "step": 256380 + }, + { + "epoch": 0.6979719613447666, + "grad_norm": 3.0580947399139404, + "learning_rate": 6.048848303121167e-06, + "loss": 2.460765075683594, + "step": 256400 + }, + { + "epoch": 0.6980264053355111, + "grad_norm": 2.3281970024108887, + "learning_rate": 6.047757939212212e-06, + "loss": 2.4639495849609374, + "step": 256420 + }, + { + "epoch": 0.6980808493262556, + "grad_norm": 2.06786847114563, + "learning_rate": 6.046667575303258e-06, + "loss": 2.464935874938965, + "step": 256440 + }, + { + "epoch": 0.6981352933170001, + "grad_norm": 2.148314952850342, + "learning_rate": 6.045577211394304e-06, + "loss": 2.470310592651367, + "step": 256460 + }, + { + "epoch": 0.6981897373077447, + "grad_norm": 2.1258716583251953, + "learning_rate": 6.044486847485349e-06, + "loss": 2.4882862091064455, + "step": 256480 + }, + { + "epoch": 0.6982441812984892, + "grad_norm": 2.2564077377319336, + "learning_rate": 6.043396483576394e-06, + "loss": 2.4854047775268553, + "step": 256500 + }, + { + "epoch": 0.6982986252892337, + "grad_norm": 2.1919095516204834, + "learning_rate": 6.0423061196674395e-06, + "loss": 2.452340507507324, + "step": 256520 + }, + { + "epoch": 0.6983530692799782, + "grad_norm": 2.2714338302612305, + "learning_rate": 6.041215755758485e-06, + "loss": 2.482504463195801, + "step": 256540 + }, + { + "epoch": 0.6984075132707227, + "grad_norm": 2.1519861221313477, + "learning_rate": 6.04012539184953e-06, + "loss": 2.450212860107422, + "step": 256560 + }, + { + "epoch": 0.6984619572614673, + "grad_norm": 2.1949000358581543, + "learning_rate": 6.039035027940575e-06, + "loss": 2.4482137680053713, + "step": 256580 + }, + { + "epoch": 0.6985164012522118, + "grad_norm": 2.429910659790039, + "learning_rate": 6.037944664031621e-06, + "loss": 2.4613990783691406, + "step": 256600 + }, + { + "epoch": 0.6985708452429563, + "grad_norm": 2.23599910736084, + "learning_rate": 6.036854300122667e-06, + "loss": 2.481741523742676, + "step": 256620 + }, + { + "epoch": 0.6986252892337008, + "grad_norm": 2.1154394149780273, + "learning_rate": 6.035763936213712e-06, + "loss": 2.4844608306884766, + "step": 256640 + }, + { + "epoch": 0.6986797332244453, + "grad_norm": 2.152895212173462, + "learning_rate": 6.034673572304757e-06, + "loss": 2.4660665512084963, + "step": 256660 + }, + { + "epoch": 0.6987341772151898, + "grad_norm": 2.2032179832458496, + "learning_rate": 6.033583208395803e-06, + "loss": 2.4600431442260744, + "step": 256680 + }, + { + "epoch": 0.6987886212059344, + "grad_norm": 3.4070518016815186, + "learning_rate": 6.032492844486848e-06, + "loss": 2.495208168029785, + "step": 256700 + }, + { + "epoch": 0.6988430651966789, + "grad_norm": 2.2381319999694824, + "learning_rate": 6.031402480577893e-06, + "loss": 2.482936668395996, + "step": 256720 + }, + { + "epoch": 0.6988975091874234, + "grad_norm": 2.776573896408081, + "learning_rate": 6.030312116668938e-06, + "loss": 2.462519645690918, + "step": 256740 + }, + { + "epoch": 0.6989519531781679, + "grad_norm": 2.1695168018341064, + "learning_rate": 6.029221752759984e-06, + "loss": 2.479499626159668, + "step": 256760 + }, + { + "epoch": 0.6990063971689124, + "grad_norm": 2.100959539413452, + "learning_rate": 6.02813138885103e-06, + "loss": 2.4730310440063477, + "step": 256780 + }, + { + "epoch": 0.699060841159657, + "grad_norm": 2.1482186317443848, + "learning_rate": 6.027041024942075e-06, + "loss": 2.486612319946289, + "step": 256800 + }, + { + "epoch": 0.6991152851504016, + "grad_norm": 2.1615309715270996, + "learning_rate": 6.02595066103312e-06, + "loss": 2.469997787475586, + "step": 256820 + }, + { + "epoch": 0.6991697291411461, + "grad_norm": 2.2886452674865723, + "learning_rate": 6.024860297124166e-06, + "loss": 2.466026496887207, + "step": 256840 + }, + { + "epoch": 0.6992241731318906, + "grad_norm": 2.1601920127868652, + "learning_rate": 6.0237699332152115e-06, + "loss": 2.4629329681396483, + "step": 256860 + }, + { + "epoch": 0.6992786171226351, + "grad_norm": 2.133011817932129, + "learning_rate": 6.0226795693062565e-06, + "loss": 2.4661527633666993, + "step": 256880 + }, + { + "epoch": 0.6993330611133796, + "grad_norm": 2.086029291152954, + "learning_rate": 6.0215892053973014e-06, + "loss": 2.4603029251098634, + "step": 256900 + }, + { + "epoch": 0.6993875051041242, + "grad_norm": 2.213016986846924, + "learning_rate": 6.020498841488347e-06, + "loss": 2.4628345489501955, + "step": 256920 + }, + { + "epoch": 0.6994419490948687, + "grad_norm": 2.159109592437744, + "learning_rate": 6.019408477579392e-06, + "loss": 2.4881528854370116, + "step": 256940 + }, + { + "epoch": 0.6994963930856132, + "grad_norm": 2.343825578689575, + "learning_rate": 6.018318113670438e-06, + "loss": 2.472743797302246, + "step": 256960 + }, + { + "epoch": 0.6995508370763577, + "grad_norm": 2.243382215499878, + "learning_rate": 6.017227749761484e-06, + "loss": 2.455794906616211, + "step": 256980 + }, + { + "epoch": 0.6996052810671022, + "grad_norm": 2.1201553344726562, + "learning_rate": 6.016137385852529e-06, + "loss": 2.4707326889038086, + "step": 257000 + }, + { + "epoch": 0.6996597250578468, + "grad_norm": 2.4087655544281006, + "learning_rate": 6.015047021943574e-06, + "loss": 2.46091251373291, + "step": 257020 + }, + { + "epoch": 0.6997141690485913, + "grad_norm": 2.1816208362579346, + "learning_rate": 6.0139566580346196e-06, + "loss": 2.474874496459961, + "step": 257040 + }, + { + "epoch": 0.6997686130393358, + "grad_norm": 2.2182881832122803, + "learning_rate": 6.012866294125665e-06, + "loss": 2.477307891845703, + "step": 257060 + }, + { + "epoch": 0.6998230570300803, + "grad_norm": 2.1120665073394775, + "learning_rate": 6.01177593021671e-06, + "loss": 2.488067626953125, + "step": 257080 + }, + { + "epoch": 0.6998775010208248, + "grad_norm": 2.227956533432007, + "learning_rate": 6.010685566307755e-06, + "loss": 2.4249107360839846, + "step": 257100 + }, + { + "epoch": 0.6999319450115693, + "grad_norm": 2.1344261169433594, + "learning_rate": 6.009595202398801e-06, + "loss": 2.457261848449707, + "step": 257120 + }, + { + "epoch": 0.6999863890023139, + "grad_norm": 2.2431278228759766, + "learning_rate": 6.008504838489847e-06, + "loss": 2.4781789779663086, + "step": 257140 + }, + { + "epoch": 0.7000408329930584, + "grad_norm": 2.343228816986084, + "learning_rate": 6.007414474580892e-06, + "loss": 2.442851257324219, + "step": 257160 + }, + { + "epoch": 0.7000952769838029, + "grad_norm": 2.190061330795288, + "learning_rate": 6.006324110671937e-06, + "loss": 2.4885120391845703, + "step": 257180 + }, + { + "epoch": 0.7001497209745474, + "grad_norm": 2.286411762237549, + "learning_rate": 6.005233746762983e-06, + "loss": 2.4615127563476564, + "step": 257200 + }, + { + "epoch": 0.7002041649652919, + "grad_norm": 2.232841730117798, + "learning_rate": 6.0041433828540284e-06, + "loss": 2.4615810394287108, + "step": 257220 + }, + { + "epoch": 0.7002586089560365, + "grad_norm": 2.1477746963500977, + "learning_rate": 6.003053018945073e-06, + "loss": 2.4943944931030275, + "step": 257240 + }, + { + "epoch": 0.700313052946781, + "grad_norm": 2.132356643676758, + "learning_rate": 6.001962655036118e-06, + "loss": 2.4462678909301756, + "step": 257260 + }, + { + "epoch": 0.7003674969375255, + "grad_norm": 2.0758886337280273, + "learning_rate": 6.000872291127164e-06, + "loss": 2.4893869400024413, + "step": 257280 + }, + { + "epoch": 0.70042194092827, + "grad_norm": 2.173053503036499, + "learning_rate": 5.99978192721821e-06, + "loss": 2.4696849822998046, + "step": 257300 + }, + { + "epoch": 0.7004763849190145, + "grad_norm": 2.1403822898864746, + "learning_rate": 5.998691563309255e-06, + "loss": 2.4621400833129883, + "step": 257320 + }, + { + "epoch": 0.700530828909759, + "grad_norm": 2.2546005249023438, + "learning_rate": 5.9976011994003e-06, + "loss": 2.4827754974365233, + "step": 257340 + }, + { + "epoch": 0.7005852729005037, + "grad_norm": 2.135805368423462, + "learning_rate": 5.996510835491346e-06, + "loss": 2.4954591751098634, + "step": 257360 + }, + { + "epoch": 0.7006397168912482, + "grad_norm": 2.1861395835876465, + "learning_rate": 5.9954204715823915e-06, + "loss": 2.4826887130737303, + "step": 257380 + }, + { + "epoch": 0.7006941608819927, + "grad_norm": 2.0445919036865234, + "learning_rate": 5.9943301076734365e-06, + "loss": 2.4642871856689452, + "step": 257400 + }, + { + "epoch": 0.7007486048727372, + "grad_norm": 2.2422428131103516, + "learning_rate": 5.9932397437644815e-06, + "loss": 2.4225488662719727, + "step": 257420 + }, + { + "epoch": 0.7008030488634817, + "grad_norm": 2.1810951232910156, + "learning_rate": 5.992149379855527e-06, + "loss": 2.514790344238281, + "step": 257440 + }, + { + "epoch": 0.7008574928542262, + "grad_norm": 2.065518379211426, + "learning_rate": 5.991059015946573e-06, + "loss": 2.464278221130371, + "step": 257460 + }, + { + "epoch": 0.7009119368449708, + "grad_norm": 2.186593770980835, + "learning_rate": 5.989968652037618e-06, + "loss": 2.4636905670166014, + "step": 257480 + }, + { + "epoch": 0.7009663808357153, + "grad_norm": 2.200463056564331, + "learning_rate": 5.988878288128663e-06, + "loss": 2.4773208618164064, + "step": 257500 + }, + { + "epoch": 0.7010208248264598, + "grad_norm": 2.173471450805664, + "learning_rate": 5.987787924219709e-06, + "loss": 2.478508186340332, + "step": 257520 + }, + { + "epoch": 0.7010752688172043, + "grad_norm": 2.137352228164673, + "learning_rate": 5.986697560310755e-06, + "loss": 2.4668540954589844, + "step": 257540 + }, + { + "epoch": 0.7011297128079488, + "grad_norm": 2.149543285369873, + "learning_rate": 5.9856071964018e-06, + "loss": 2.4499000549316405, + "step": 257560 + }, + { + "epoch": 0.7011841567986934, + "grad_norm": 2.089416265487671, + "learning_rate": 5.9845168324928445e-06, + "loss": 2.4460243225097655, + "step": 257580 + }, + { + "epoch": 0.7012386007894379, + "grad_norm": 2.099186420440674, + "learning_rate": 5.98342646858389e-06, + "loss": 2.472059631347656, + "step": 257600 + }, + { + "epoch": 0.7012930447801824, + "grad_norm": 2.0933783054351807, + "learning_rate": 5.982336104674936e-06, + "loss": 2.482286834716797, + "step": 257620 + }, + { + "epoch": 0.7013474887709269, + "grad_norm": 2.2837870121002197, + "learning_rate": 5.981245740765981e-06, + "loss": 2.4706794738769533, + "step": 257640 + }, + { + "epoch": 0.7014019327616714, + "grad_norm": 2.561004877090454, + "learning_rate": 5.980155376857026e-06, + "loss": 2.4579166412353515, + "step": 257660 + }, + { + "epoch": 0.7014563767524159, + "grad_norm": 2.128649950027466, + "learning_rate": 5.979065012948072e-06, + "loss": 2.467303466796875, + "step": 257680 + }, + { + "epoch": 0.7015108207431605, + "grad_norm": 2.133514881134033, + "learning_rate": 5.977974649039118e-06, + "loss": 2.4408458709716796, + "step": 257700 + }, + { + "epoch": 0.701565264733905, + "grad_norm": 2.151991844177246, + "learning_rate": 5.976884285130163e-06, + "loss": 2.4507879257202148, + "step": 257720 + }, + { + "epoch": 0.7016197087246495, + "grad_norm": 2.1465229988098145, + "learning_rate": 5.975793921221208e-06, + "loss": 2.4675050735473634, + "step": 257740 + }, + { + "epoch": 0.701674152715394, + "grad_norm": 2.161174774169922, + "learning_rate": 5.9747035573122534e-06, + "loss": 2.458473014831543, + "step": 257760 + }, + { + "epoch": 0.7017285967061385, + "grad_norm": 2.1254384517669678, + "learning_rate": 5.973613193403299e-06, + "loss": 2.488755798339844, + "step": 257780 + }, + { + "epoch": 0.7017830406968831, + "grad_norm": 2.151353120803833, + "learning_rate": 5.972522829494344e-06, + "loss": 2.4728200912475584, + "step": 257800 + }, + { + "epoch": 0.7018374846876276, + "grad_norm": 2.1757683753967285, + "learning_rate": 5.971432465585389e-06, + "loss": 2.4543340682983397, + "step": 257820 + }, + { + "epoch": 0.7018919286783721, + "grad_norm": 4.161166667938232, + "learning_rate": 5.970342101676435e-06, + "loss": 2.4499305725097655, + "step": 257840 + }, + { + "epoch": 0.7019463726691166, + "grad_norm": 2.0805630683898926, + "learning_rate": 5.969251737767481e-06, + "loss": 2.464898681640625, + "step": 257860 + }, + { + "epoch": 0.7020008166598611, + "grad_norm": 2.2415411472320557, + "learning_rate": 5.968161373858526e-06, + "loss": 2.457888603210449, + "step": 257880 + }, + { + "epoch": 0.7020552606506056, + "grad_norm": 2.1869735717773438, + "learning_rate": 5.967071009949571e-06, + "loss": 2.4638818740844726, + "step": 257900 + }, + { + "epoch": 0.7021097046413503, + "grad_norm": 2.252115249633789, + "learning_rate": 5.9659806460406165e-06, + "loss": 2.4929922103881834, + "step": 257920 + }, + { + "epoch": 0.7021641486320948, + "grad_norm": 2.1578352451324463, + "learning_rate": 5.964890282131662e-06, + "loss": 2.4752313613891603, + "step": 257940 + }, + { + "epoch": 0.7022185926228393, + "grad_norm": 2.054483652114868, + "learning_rate": 5.963799918222707e-06, + "loss": 2.4661413192749024, + "step": 257960 + }, + { + "epoch": 0.7022730366135838, + "grad_norm": 2.1179802417755127, + "learning_rate": 5.962709554313752e-06, + "loss": 2.448129081726074, + "step": 257980 + }, + { + "epoch": 0.7023274806043283, + "grad_norm": 3.5459234714508057, + "learning_rate": 5.961619190404798e-06, + "loss": 2.491036033630371, + "step": 258000 + }, + { + "epoch": 0.7023819245950729, + "grad_norm": 2.223876714706421, + "learning_rate": 5.960528826495844e-06, + "loss": 2.4703598022460938, + "step": 258020 + }, + { + "epoch": 0.7024363685858174, + "grad_norm": 2.0676686763763428, + "learning_rate": 5.959438462586888e-06, + "loss": 2.4453784942626955, + "step": 258040 + }, + { + "epoch": 0.7024908125765619, + "grad_norm": 2.190012216567993, + "learning_rate": 5.958348098677934e-06, + "loss": 2.4696670532226563, + "step": 258060 + }, + { + "epoch": 0.7025452565673064, + "grad_norm": 2.210618734359741, + "learning_rate": 5.95725773476898e-06, + "loss": 2.4712263107299806, + "step": 258080 + }, + { + "epoch": 0.7025997005580509, + "grad_norm": 2.2329490184783936, + "learning_rate": 5.956167370860025e-06, + "loss": 2.456573486328125, + "step": 258100 + }, + { + "epoch": 0.7026541445487954, + "grad_norm": 2.154128074645996, + "learning_rate": 5.9550770069510695e-06, + "loss": 2.4838272094726563, + "step": 258120 + }, + { + "epoch": 0.70270858853954, + "grad_norm": 2.187608003616333, + "learning_rate": 5.953986643042115e-06, + "loss": 2.500945472717285, + "step": 258140 + }, + { + "epoch": 0.7027630325302845, + "grad_norm": 2.20569109916687, + "learning_rate": 5.952896279133161e-06, + "loss": 2.4716081619262695, + "step": 258160 + }, + { + "epoch": 0.702817476521029, + "grad_norm": 2.180884838104248, + "learning_rate": 5.951805915224207e-06, + "loss": 2.4629526138305664, + "step": 258180 + }, + { + "epoch": 0.7028719205117735, + "grad_norm": 2.099639415740967, + "learning_rate": 5.950715551315251e-06, + "loss": 2.453120231628418, + "step": 258200 + }, + { + "epoch": 0.702926364502518, + "grad_norm": 2.1496803760528564, + "learning_rate": 5.949625187406297e-06, + "loss": 2.481551933288574, + "step": 258220 + }, + { + "epoch": 0.7029808084932626, + "grad_norm": 2.0991036891937256, + "learning_rate": 5.948534823497343e-06, + "loss": 2.452920150756836, + "step": 258240 + }, + { + "epoch": 0.7030352524840071, + "grad_norm": 2.210585594177246, + "learning_rate": 5.9474444595883885e-06, + "loss": 2.4809871673583985, + "step": 258260 + }, + { + "epoch": 0.7030896964747516, + "grad_norm": 2.0899643898010254, + "learning_rate": 5.946354095679433e-06, + "loss": 2.448140525817871, + "step": 258280 + }, + { + "epoch": 0.7031441404654961, + "grad_norm": 2.2492828369140625, + "learning_rate": 5.9452637317704784e-06, + "loss": 2.451969337463379, + "step": 258300 + }, + { + "epoch": 0.7031985844562406, + "grad_norm": 2.8497700691223145, + "learning_rate": 5.944173367861524e-06, + "loss": 2.4649589538574217, + "step": 258320 + }, + { + "epoch": 0.7032530284469851, + "grad_norm": 2.255887985229492, + "learning_rate": 5.94308300395257e-06, + "loss": 2.460268974304199, + "step": 258340 + }, + { + "epoch": 0.7033074724377297, + "grad_norm": 2.282586097717285, + "learning_rate": 5.941992640043614e-06, + "loss": 2.468665885925293, + "step": 258360 + }, + { + "epoch": 0.7033619164284742, + "grad_norm": 2.0823276042938232, + "learning_rate": 5.94090227613466e-06, + "loss": 2.4843891143798826, + "step": 258380 + }, + { + "epoch": 0.7034163604192187, + "grad_norm": 2.2005791664123535, + "learning_rate": 5.939811912225706e-06, + "loss": 2.4865726470947265, + "step": 258400 + }, + { + "epoch": 0.7034708044099632, + "grad_norm": 2.195061683654785, + "learning_rate": 5.938721548316752e-06, + "loss": 2.4886093139648438, + "step": 258420 + }, + { + "epoch": 0.7035252484007077, + "grad_norm": 2.2442452907562256, + "learning_rate": 5.937631184407796e-06, + "loss": 2.463343048095703, + "step": 258440 + }, + { + "epoch": 0.7035796923914522, + "grad_norm": 2.0984084606170654, + "learning_rate": 5.9365408204988415e-06, + "loss": 2.452302932739258, + "step": 258460 + }, + { + "epoch": 0.7036341363821969, + "grad_norm": 2.4456088542938232, + "learning_rate": 5.935450456589887e-06, + "loss": 2.4741546630859377, + "step": 258480 + }, + { + "epoch": 0.7036885803729414, + "grad_norm": 2.151811122894287, + "learning_rate": 5.934360092680933e-06, + "loss": 2.4612659454345702, + "step": 258500 + }, + { + "epoch": 0.7037430243636859, + "grad_norm": 2.2172908782958984, + "learning_rate": 5.933269728771977e-06, + "loss": 2.4944988250732423, + "step": 258520 + }, + { + "epoch": 0.7037974683544304, + "grad_norm": 2.1294100284576416, + "learning_rate": 5.932179364863023e-06, + "loss": 2.4333065032958983, + "step": 258540 + }, + { + "epoch": 0.7038519123451749, + "grad_norm": 2.2531585693359375, + "learning_rate": 5.931089000954069e-06, + "loss": 2.462580108642578, + "step": 258560 + }, + { + "epoch": 0.7039063563359195, + "grad_norm": 2.0857152938842773, + "learning_rate": 5.929998637045115e-06, + "loss": 2.487356758117676, + "step": 258580 + }, + { + "epoch": 0.703960800326664, + "grad_norm": 2.2285351753234863, + "learning_rate": 5.928908273136159e-06, + "loss": 2.462744140625, + "step": 258600 + }, + { + "epoch": 0.7040152443174085, + "grad_norm": 2.1454508304595947, + "learning_rate": 5.927817909227205e-06, + "loss": 2.455343818664551, + "step": 258620 + }, + { + "epoch": 0.704069688308153, + "grad_norm": 2.267638683319092, + "learning_rate": 5.92672754531825e-06, + "loss": 2.469314956665039, + "step": 258640 + }, + { + "epoch": 0.7041241322988975, + "grad_norm": 2.32702374458313, + "learning_rate": 5.925637181409296e-06, + "loss": 2.4911399841308595, + "step": 258660 + }, + { + "epoch": 0.704178576289642, + "grad_norm": 2.1623659133911133, + "learning_rate": 5.924546817500342e-06, + "loss": 2.4560546875, + "step": 258680 + }, + { + "epoch": 0.7042330202803866, + "grad_norm": 2.276571035385132, + "learning_rate": 5.923456453591386e-06, + "loss": 2.4817943572998047, + "step": 258700 + }, + { + "epoch": 0.7042874642711311, + "grad_norm": 2.155763626098633, + "learning_rate": 5.922366089682432e-06, + "loss": 2.4338615417480467, + "step": 258720 + }, + { + "epoch": 0.7043419082618756, + "grad_norm": 2.0728578567504883, + "learning_rate": 5.921275725773478e-06, + "loss": 2.4466278076171877, + "step": 258740 + }, + { + "epoch": 0.7043963522526201, + "grad_norm": 2.3594069480895996, + "learning_rate": 5.9201853618645236e-06, + "loss": 2.4495893478393556, + "step": 258760 + }, + { + "epoch": 0.7044507962433646, + "grad_norm": 2.391331434249878, + "learning_rate": 5.919094997955568e-06, + "loss": 2.470225715637207, + "step": 258780 + }, + { + "epoch": 0.7045052402341092, + "grad_norm": 2.1794652938842773, + "learning_rate": 5.9180046340466135e-06, + "loss": 2.473577308654785, + "step": 258800 + }, + { + "epoch": 0.7045596842248537, + "grad_norm": 2.2235305309295654, + "learning_rate": 5.916914270137659e-06, + "loss": 2.4637413024902344, + "step": 258820 + }, + { + "epoch": 0.7046141282155982, + "grad_norm": 2.1430015563964844, + "learning_rate": 5.915823906228705e-06, + "loss": 2.478265571594238, + "step": 258840 + }, + { + "epoch": 0.7046685722063427, + "grad_norm": 2.196028709411621, + "learning_rate": 5.914733542319749e-06, + "loss": 2.478459930419922, + "step": 258860 + }, + { + "epoch": 0.7047230161970872, + "grad_norm": 2.8605496883392334, + "learning_rate": 5.913643178410795e-06, + "loss": 2.4447195053100588, + "step": 258880 + }, + { + "epoch": 0.7047774601878317, + "grad_norm": 2.1456618309020996, + "learning_rate": 5.912552814501841e-06, + "loss": 2.474671745300293, + "step": 258900 + }, + { + "epoch": 0.7048319041785763, + "grad_norm": 2.1725406646728516, + "learning_rate": 5.911462450592887e-06, + "loss": 2.449082946777344, + "step": 258920 + }, + { + "epoch": 0.7048863481693208, + "grad_norm": 2.0831620693206787, + "learning_rate": 5.910372086683931e-06, + "loss": 2.471668243408203, + "step": 258940 + }, + { + "epoch": 0.7049407921600653, + "grad_norm": 2.1107420921325684, + "learning_rate": 5.909281722774977e-06, + "loss": 2.45416259765625, + "step": 258960 + }, + { + "epoch": 0.7049952361508098, + "grad_norm": 2.1781086921691895, + "learning_rate": 5.908191358866022e-06, + "loss": 2.478516960144043, + "step": 258980 + }, + { + "epoch": 0.7050496801415543, + "grad_norm": 2.207207202911377, + "learning_rate": 5.907100994957068e-06, + "loss": 2.4880115509033205, + "step": 259000 + }, + { + "epoch": 0.705104124132299, + "grad_norm": 2.1120917797088623, + "learning_rate": 5.906010631048112e-06, + "loss": 2.467023468017578, + "step": 259020 + }, + { + "epoch": 0.7051585681230434, + "grad_norm": 2.306600570678711, + "learning_rate": 5.904920267139158e-06, + "loss": 2.447022819519043, + "step": 259040 + }, + { + "epoch": 0.705213012113788, + "grad_norm": 2.386176347732544, + "learning_rate": 5.903829903230204e-06, + "loss": 2.4944114685058594, + "step": 259060 + }, + { + "epoch": 0.7052674561045325, + "grad_norm": 2.227700710296631, + "learning_rate": 5.90273953932125e-06, + "loss": 2.4748722076416017, + "step": 259080 + }, + { + "epoch": 0.705321900095277, + "grad_norm": 2.2607715129852295, + "learning_rate": 5.901649175412294e-06, + "loss": 2.4727346420288088, + "step": 259100 + }, + { + "epoch": 0.7053763440860215, + "grad_norm": 2.1316006183624268, + "learning_rate": 5.90055881150334e-06, + "loss": 2.4837385177612306, + "step": 259120 + }, + { + "epoch": 0.7054307880767661, + "grad_norm": 2.1625866889953613, + "learning_rate": 5.8994684475943855e-06, + "loss": 2.4714096069335936, + "step": 259140 + }, + { + "epoch": 0.7054852320675106, + "grad_norm": 2.2403838634490967, + "learning_rate": 5.898378083685431e-06, + "loss": 2.461648368835449, + "step": 259160 + }, + { + "epoch": 0.7055396760582551, + "grad_norm": 2.240618944168091, + "learning_rate": 5.897287719776475e-06, + "loss": 2.473049354553223, + "step": 259180 + }, + { + "epoch": 0.7055941200489996, + "grad_norm": 2.268338203430176, + "learning_rate": 5.896197355867521e-06, + "loss": 2.4489973068237303, + "step": 259200 + }, + { + "epoch": 0.7056485640397441, + "grad_norm": 2.225579261779785, + "learning_rate": 5.895106991958567e-06, + "loss": 2.4539377212524416, + "step": 259220 + }, + { + "epoch": 0.7057030080304887, + "grad_norm": 2.291834592819214, + "learning_rate": 5.894016628049613e-06, + "loss": 2.484269142150879, + "step": 259240 + }, + { + "epoch": 0.7057574520212332, + "grad_norm": 2.1273231506347656, + "learning_rate": 5.892926264140657e-06, + "loss": 2.445290374755859, + "step": 259260 + }, + { + "epoch": 0.7058118960119777, + "grad_norm": 2.1226918697357178, + "learning_rate": 5.891835900231703e-06, + "loss": 2.479727554321289, + "step": 259280 + }, + { + "epoch": 0.7058663400027222, + "grad_norm": 2.1927530765533447, + "learning_rate": 5.8907455363227486e-06, + "loss": 2.4452320098876954, + "step": 259300 + }, + { + "epoch": 0.7059207839934667, + "grad_norm": 2.338886260986328, + "learning_rate": 5.889655172413794e-06, + "loss": 2.4603506088256837, + "step": 259320 + }, + { + "epoch": 0.7059752279842112, + "grad_norm": 2.1069395542144775, + "learning_rate": 5.8885648085048385e-06, + "loss": 2.4691043853759767, + "step": 259340 + }, + { + "epoch": 0.7060296719749558, + "grad_norm": 2.20413875579834, + "learning_rate": 5.887474444595884e-06, + "loss": 2.4830612182617187, + "step": 259360 + }, + { + "epoch": 0.7060841159657003, + "grad_norm": 2.065075159072876, + "learning_rate": 5.88638408068693e-06, + "loss": 2.482331466674805, + "step": 259380 + }, + { + "epoch": 0.7061385599564448, + "grad_norm": 2.1649842262268066, + "learning_rate": 5.885293716777976e-06, + "loss": 2.4839813232421877, + "step": 259400 + }, + { + "epoch": 0.7061930039471893, + "grad_norm": 2.2390120029449463, + "learning_rate": 5.88420335286902e-06, + "loss": 2.440964698791504, + "step": 259420 + }, + { + "epoch": 0.7062474479379338, + "grad_norm": 2.198901414871216, + "learning_rate": 5.883112988960066e-06, + "loss": 2.476190948486328, + "step": 259440 + }, + { + "epoch": 0.7063018919286784, + "grad_norm": 2.1599948406219482, + "learning_rate": 5.882022625051112e-06, + "loss": 2.468534469604492, + "step": 259460 + }, + { + "epoch": 0.7063563359194229, + "grad_norm": 2.1821885108947754, + "learning_rate": 5.8809322611421575e-06, + "loss": 2.428065872192383, + "step": 259480 + }, + { + "epoch": 0.7064107799101674, + "grad_norm": 2.155841112136841, + "learning_rate": 5.8798418972332016e-06, + "loss": 2.472729301452637, + "step": 259500 + }, + { + "epoch": 0.7064652239009119, + "grad_norm": 2.249006509780884, + "learning_rate": 5.878751533324247e-06, + "loss": 2.416828918457031, + "step": 259520 + }, + { + "epoch": 0.7065196678916564, + "grad_norm": 2.15010142326355, + "learning_rate": 5.877661169415293e-06, + "loss": 2.4580488204956055, + "step": 259540 + }, + { + "epoch": 0.7065741118824009, + "grad_norm": 2.108504295349121, + "learning_rate": 5.876570805506339e-06, + "loss": 2.474051856994629, + "step": 259560 + }, + { + "epoch": 0.7066285558731455, + "grad_norm": 2.297436475753784, + "learning_rate": 5.875480441597383e-06, + "loss": 2.4859375, + "step": 259580 + }, + { + "epoch": 0.70668299986389, + "grad_norm": 2.124868392944336, + "learning_rate": 5.874390077688429e-06, + "loss": 2.5033750534057617, + "step": 259600 + }, + { + "epoch": 0.7067374438546346, + "grad_norm": 2.1730945110321045, + "learning_rate": 5.873299713779475e-06, + "loss": 2.426584243774414, + "step": 259620 + }, + { + "epoch": 0.706791887845379, + "grad_norm": 2.1248135566711426, + "learning_rate": 5.8722093498705205e-06, + "loss": 2.4654071807861326, + "step": 259640 + }, + { + "epoch": 0.7068463318361236, + "grad_norm": 2.1456458568573, + "learning_rate": 5.871118985961565e-06, + "loss": 2.470267105102539, + "step": 259660 + }, + { + "epoch": 0.706900775826868, + "grad_norm": 2.1813626289367676, + "learning_rate": 5.8700286220526105e-06, + "loss": 2.487197685241699, + "step": 259680 + }, + { + "epoch": 0.7069552198176127, + "grad_norm": 2.1981570720672607, + "learning_rate": 5.868938258143656e-06, + "loss": 2.4603202819824217, + "step": 259700 + }, + { + "epoch": 0.7070096638083572, + "grad_norm": 2.129443883895874, + "learning_rate": 5.867847894234702e-06, + "loss": 2.4707914352416993, + "step": 259720 + }, + { + "epoch": 0.7070641077991017, + "grad_norm": 2.2023887634277344, + "learning_rate": 5.866757530325746e-06, + "loss": 2.498609924316406, + "step": 259740 + }, + { + "epoch": 0.7071185517898462, + "grad_norm": 2.117797613143921, + "learning_rate": 5.865667166416792e-06, + "loss": 2.4644615173339846, + "step": 259760 + }, + { + "epoch": 0.7071729957805907, + "grad_norm": 2.126894950866699, + "learning_rate": 5.864576802507838e-06, + "loss": 2.4534225463867188, + "step": 259780 + }, + { + "epoch": 0.7072274397713353, + "grad_norm": 2.2468771934509277, + "learning_rate": 5.863486438598884e-06, + "loss": 2.46060733795166, + "step": 259800 + }, + { + "epoch": 0.7072818837620798, + "grad_norm": 2.2013261318206787, + "learning_rate": 5.862396074689928e-06, + "loss": 2.4623056411743165, + "step": 259820 + }, + { + "epoch": 0.7073363277528243, + "grad_norm": 2.143829822540283, + "learning_rate": 5.8613057107809736e-06, + "loss": 2.472133445739746, + "step": 259840 + }, + { + "epoch": 0.7073907717435688, + "grad_norm": 2.1658506393432617, + "learning_rate": 5.860215346872019e-06, + "loss": 2.4505027770996093, + "step": 259860 + }, + { + "epoch": 0.7074452157343133, + "grad_norm": 2.1747286319732666, + "learning_rate": 5.859124982963065e-06, + "loss": 2.4724184036254884, + "step": 259880 + }, + { + "epoch": 0.7074996597250578, + "grad_norm": 2.3957011699676514, + "learning_rate": 5.858034619054109e-06, + "loss": 2.459834671020508, + "step": 259900 + }, + { + "epoch": 0.7075541037158024, + "grad_norm": 2.455833911895752, + "learning_rate": 5.856944255145155e-06, + "loss": 2.476182556152344, + "step": 259920 + }, + { + "epoch": 0.7076085477065469, + "grad_norm": 2.094125509262085, + "learning_rate": 5.855853891236201e-06, + "loss": 2.467231559753418, + "step": 259940 + }, + { + "epoch": 0.7076629916972914, + "grad_norm": 2.1444039344787598, + "learning_rate": 5.854763527327246e-06, + "loss": 2.494415283203125, + "step": 259960 + }, + { + "epoch": 0.7077174356880359, + "grad_norm": 2.164320468902588, + "learning_rate": 5.853673163418291e-06, + "loss": 2.4726966857910155, + "step": 259980 + }, + { + "epoch": 0.7077718796787804, + "grad_norm": 2.059440851211548, + "learning_rate": 5.852582799509337e-06, + "loss": 2.4458194732666017, + "step": 260000 + }, + { + "epoch": 0.707826323669525, + "grad_norm": 2.0772743225097656, + "learning_rate": 5.8514924356003824e-06, + "loss": 2.4667842864990233, + "step": 260020 + }, + { + "epoch": 0.7078807676602695, + "grad_norm": 2.115778923034668, + "learning_rate": 5.850402071691427e-06, + "loss": 2.4682504653930666, + "step": 260040 + }, + { + "epoch": 0.707935211651014, + "grad_norm": 2.1309237480163574, + "learning_rate": 5.849311707782472e-06, + "loss": 2.4909456253051756, + "step": 260060 + }, + { + "epoch": 0.7079896556417585, + "grad_norm": 2.214132785797119, + "learning_rate": 5.848221343873518e-06, + "loss": 2.449363136291504, + "step": 260080 + }, + { + "epoch": 0.708044099632503, + "grad_norm": 2.173006534576416, + "learning_rate": 5.847130979964564e-06, + "loss": 2.4676694869995117, + "step": 260100 + }, + { + "epoch": 0.7080985436232475, + "grad_norm": 2.241429328918457, + "learning_rate": 5.846040616055609e-06, + "loss": 2.4800722122192385, + "step": 260120 + }, + { + "epoch": 0.7081529876139921, + "grad_norm": 2.221987247467041, + "learning_rate": 5.844950252146654e-06, + "loss": 2.477657699584961, + "step": 260140 + }, + { + "epoch": 0.7082074316047366, + "grad_norm": 2.15274715423584, + "learning_rate": 5.8438598882377e-06, + "loss": 2.4937726974487306, + "step": 260160 + }, + { + "epoch": 0.7082618755954811, + "grad_norm": 2.1964166164398193, + "learning_rate": 5.8427695243287455e-06, + "loss": 2.457681083679199, + "step": 260180 + }, + { + "epoch": 0.7083163195862257, + "grad_norm": 2.242112636566162, + "learning_rate": 5.8416791604197905e-06, + "loss": 2.449971389770508, + "step": 260200 + }, + { + "epoch": 0.7083707635769702, + "grad_norm": 2.330554246902466, + "learning_rate": 5.8405887965108355e-06, + "loss": 2.4664379119873048, + "step": 260220 + }, + { + "epoch": 0.7084252075677148, + "grad_norm": 2.1240978240966797, + "learning_rate": 5.839498432601881e-06, + "loss": 2.497288703918457, + "step": 260240 + }, + { + "epoch": 0.7084796515584593, + "grad_norm": 2.1763830184936523, + "learning_rate": 5.838408068692927e-06, + "loss": 2.5127628326416014, + "step": 260260 + }, + { + "epoch": 0.7085340955492038, + "grad_norm": 2.3548598289489746, + "learning_rate": 5.837317704783972e-06, + "loss": 2.481984519958496, + "step": 260280 + }, + { + "epoch": 0.7085885395399483, + "grad_norm": 2.1829447746276855, + "learning_rate": 5.836227340875017e-06, + "loss": 2.474959373474121, + "step": 260300 + }, + { + "epoch": 0.7086429835306928, + "grad_norm": 2.247699737548828, + "learning_rate": 5.835136976966063e-06, + "loss": 2.486581039428711, + "step": 260320 + }, + { + "epoch": 0.7086974275214373, + "grad_norm": 2.2207374572753906, + "learning_rate": 5.834046613057109e-06, + "loss": 2.4323451995849608, + "step": 260340 + }, + { + "epoch": 0.7087518715121819, + "grad_norm": 2.2108638286590576, + "learning_rate": 5.832956249148154e-06, + "loss": 2.465822219848633, + "step": 260360 + }, + { + "epoch": 0.7088063155029264, + "grad_norm": 2.256023406982422, + "learning_rate": 5.831865885239199e-06, + "loss": 2.4457286834716796, + "step": 260380 + }, + { + "epoch": 0.7088607594936709, + "grad_norm": 2.21645450592041, + "learning_rate": 5.830775521330244e-06, + "loss": 2.4862756729125977, + "step": 260400 + }, + { + "epoch": 0.7089152034844154, + "grad_norm": 2.15846848487854, + "learning_rate": 5.82968515742129e-06, + "loss": 2.486338806152344, + "step": 260420 + }, + { + "epoch": 0.7089696474751599, + "grad_norm": 2.1892857551574707, + "learning_rate": 5.828594793512335e-06, + "loss": 2.4696609497070314, + "step": 260440 + }, + { + "epoch": 0.7090240914659045, + "grad_norm": 2.193631172180176, + "learning_rate": 5.827504429603381e-06, + "loss": 2.4873798370361326, + "step": 260460 + }, + { + "epoch": 0.709078535456649, + "grad_norm": 2.1522843837738037, + "learning_rate": 5.826414065694426e-06, + "loss": 2.4605268478393554, + "step": 260480 + }, + { + "epoch": 0.7091329794473935, + "grad_norm": 2.136629343032837, + "learning_rate": 5.825323701785472e-06, + "loss": 2.4697790145874023, + "step": 260500 + }, + { + "epoch": 0.709187423438138, + "grad_norm": 2.1475911140441895, + "learning_rate": 5.824233337876517e-06, + "loss": 2.4863548278808594, + "step": 260520 + }, + { + "epoch": 0.7092418674288825, + "grad_norm": 2.156848907470703, + "learning_rate": 5.8231429739675625e-06, + "loss": 2.476474380493164, + "step": 260540 + }, + { + "epoch": 0.709296311419627, + "grad_norm": 2.1377415657043457, + "learning_rate": 5.8220526100586074e-06, + "loss": 2.4404293060302735, + "step": 260560 + }, + { + "epoch": 0.7093507554103716, + "grad_norm": 3.1092188358306885, + "learning_rate": 5.820962246149653e-06, + "loss": 2.426186943054199, + "step": 260580 + }, + { + "epoch": 0.7094051994011161, + "grad_norm": 2.2002456188201904, + "learning_rate": 5.819871882240698e-06, + "loss": 2.482036018371582, + "step": 260600 + }, + { + "epoch": 0.7094596433918606, + "grad_norm": 2.8026955127716064, + "learning_rate": 5.818781518331744e-06, + "loss": 2.4742290496826174, + "step": 260620 + }, + { + "epoch": 0.7095140873826051, + "grad_norm": 2.1671576499938965, + "learning_rate": 5.817691154422789e-06, + "loss": 2.514863204956055, + "step": 260640 + }, + { + "epoch": 0.7095685313733496, + "grad_norm": 2.16829776763916, + "learning_rate": 5.816600790513835e-06, + "loss": 2.482589340209961, + "step": 260660 + }, + { + "epoch": 0.7096229753640941, + "grad_norm": 2.620131731033325, + "learning_rate": 5.81551042660488e-06, + "loss": 2.458735466003418, + "step": 260680 + }, + { + "epoch": 0.7096774193548387, + "grad_norm": 2.202598810195923, + "learning_rate": 5.8144200626959256e-06, + "loss": 2.4645151138305663, + "step": 260700 + }, + { + "epoch": 0.7097318633455832, + "grad_norm": 4.171568870544434, + "learning_rate": 5.8133296987869705e-06, + "loss": 2.4602533340454102, + "step": 260720 + }, + { + "epoch": 0.7097863073363277, + "grad_norm": 2.194974184036255, + "learning_rate": 5.812239334878016e-06, + "loss": 2.4543542861938477, + "step": 260740 + }, + { + "epoch": 0.7098407513270723, + "grad_norm": 2.105215311050415, + "learning_rate": 5.811148970969061e-06, + "loss": 2.4930961608886717, + "step": 260760 + }, + { + "epoch": 0.7098951953178168, + "grad_norm": 2.316056489944458, + "learning_rate": 5.810058607060107e-06, + "loss": 2.479433059692383, + "step": 260780 + }, + { + "epoch": 0.7099496393085614, + "grad_norm": 2.1207826137542725, + "learning_rate": 5.808968243151152e-06, + "loss": 2.473255729675293, + "step": 260800 + }, + { + "epoch": 0.7100040832993059, + "grad_norm": 2.189373731613159, + "learning_rate": 5.807877879242198e-06, + "loss": 2.44515323638916, + "step": 260820 + }, + { + "epoch": 0.7100585272900504, + "grad_norm": 2.1993603706359863, + "learning_rate": 5.806787515333243e-06, + "loss": 2.442978286743164, + "step": 260840 + }, + { + "epoch": 0.7101129712807949, + "grad_norm": 2.19616436958313, + "learning_rate": 5.805697151424289e-06, + "loss": 2.458598327636719, + "step": 260860 + }, + { + "epoch": 0.7101674152715394, + "grad_norm": 2.328848123550415, + "learning_rate": 5.804606787515334e-06, + "loss": 2.466115951538086, + "step": 260880 + }, + { + "epoch": 0.7102218592622839, + "grad_norm": 2.2811858654022217, + "learning_rate": 5.803516423606379e-06, + "loss": 2.5051403045654297, + "step": 260900 + }, + { + "epoch": 0.7102763032530285, + "grad_norm": 2.2591724395751953, + "learning_rate": 5.802426059697424e-06, + "loss": 2.477411079406738, + "step": 260920 + }, + { + "epoch": 0.710330747243773, + "grad_norm": 2.171213388442993, + "learning_rate": 5.80133569578847e-06, + "loss": 2.4402538299560548, + "step": 260940 + }, + { + "epoch": 0.7103851912345175, + "grad_norm": 2.332523822784424, + "learning_rate": 5.800245331879515e-06, + "loss": 2.438372993469238, + "step": 260960 + }, + { + "epoch": 0.710439635225262, + "grad_norm": 2.2557566165924072, + "learning_rate": 5.799154967970561e-06, + "loss": 2.469075584411621, + "step": 260980 + }, + { + "epoch": 0.7104940792160065, + "grad_norm": 2.159614086151123, + "learning_rate": 5.798064604061606e-06, + "loss": 2.4713203430175783, + "step": 261000 + }, + { + "epoch": 0.7105485232067511, + "grad_norm": 2.154496192932129, + "learning_rate": 5.796974240152652e-06, + "loss": 2.494599533081055, + "step": 261020 + }, + { + "epoch": 0.7106029671974956, + "grad_norm": 2.2937889099121094, + "learning_rate": 5.795883876243697e-06, + "loss": 2.47302360534668, + "step": 261040 + }, + { + "epoch": 0.7106574111882401, + "grad_norm": 2.1442298889160156, + "learning_rate": 5.794793512334742e-06, + "loss": 2.4546491622924806, + "step": 261060 + }, + { + "epoch": 0.7107118551789846, + "grad_norm": 2.116393566131592, + "learning_rate": 5.7937031484257875e-06, + "loss": 2.4793066024780273, + "step": 261080 + }, + { + "epoch": 0.7107662991697291, + "grad_norm": 2.101367712020874, + "learning_rate": 5.792612784516833e-06, + "loss": 2.479157066345215, + "step": 261100 + }, + { + "epoch": 0.7108207431604736, + "grad_norm": 2.544464349746704, + "learning_rate": 5.791522420607878e-06, + "loss": 2.472024345397949, + "step": 261120 + }, + { + "epoch": 0.7108751871512182, + "grad_norm": 2.2474679946899414, + "learning_rate": 5.790432056698923e-06, + "loss": 2.4805776596069338, + "step": 261140 + }, + { + "epoch": 0.7109296311419627, + "grad_norm": 2.1207096576690674, + "learning_rate": 5.789341692789969e-06, + "loss": 2.458091163635254, + "step": 261160 + }, + { + "epoch": 0.7109840751327072, + "grad_norm": 2.2154335975646973, + "learning_rate": 5.788251328881015e-06, + "loss": 2.452054023742676, + "step": 261180 + }, + { + "epoch": 0.7110385191234517, + "grad_norm": 3.208780527114868, + "learning_rate": 5.78716096497206e-06, + "loss": 2.4665082931518554, + "step": 261200 + }, + { + "epoch": 0.7110929631141962, + "grad_norm": 2.0601954460144043, + "learning_rate": 5.786070601063105e-06, + "loss": 2.4593574523925783, + "step": 261220 + }, + { + "epoch": 0.7111474071049408, + "grad_norm": 2.144284725189209, + "learning_rate": 5.7849802371541506e-06, + "loss": 2.4568153381347657, + "step": 261240 + }, + { + "epoch": 0.7112018510956853, + "grad_norm": 2.155583620071411, + "learning_rate": 5.783889873245196e-06, + "loss": 2.4554380416870116, + "step": 261260 + }, + { + "epoch": 0.7112562950864298, + "grad_norm": 2.091505289077759, + "learning_rate": 5.782799509336241e-06, + "loss": 2.44787483215332, + "step": 261280 + }, + { + "epoch": 0.7113107390771743, + "grad_norm": 2.0531396865844727, + "learning_rate": 5.781709145427286e-06, + "loss": 2.460442543029785, + "step": 261300 + }, + { + "epoch": 0.7113651830679188, + "grad_norm": 3.0039379596710205, + "learning_rate": 5.780618781518332e-06, + "loss": 2.47814884185791, + "step": 261320 + }, + { + "epoch": 0.7114196270586634, + "grad_norm": 2.275000810623169, + "learning_rate": 5.779528417609378e-06, + "loss": 2.477000427246094, + "step": 261340 + }, + { + "epoch": 0.711474071049408, + "grad_norm": 2.212613105773926, + "learning_rate": 5.778438053700423e-06, + "loss": 2.4654767990112303, + "step": 261360 + }, + { + "epoch": 0.7115285150401525, + "grad_norm": 2.2395377159118652, + "learning_rate": 5.777347689791468e-06, + "loss": 2.4936120986938475, + "step": 261380 + }, + { + "epoch": 0.711582959030897, + "grad_norm": 2.3306667804718018, + "learning_rate": 5.776257325882514e-06, + "loss": 2.4686378479003905, + "step": 261400 + }, + { + "epoch": 0.7116374030216415, + "grad_norm": 2.1608195304870605, + "learning_rate": 5.7751669619735594e-06, + "loss": 2.4642881393432616, + "step": 261420 + }, + { + "epoch": 0.711691847012386, + "grad_norm": 2.1943650245666504, + "learning_rate": 5.774076598064604e-06, + "loss": 2.4782217025756834, + "step": 261440 + }, + { + "epoch": 0.7117462910031306, + "grad_norm": 2.1363742351531982, + "learning_rate": 5.772986234155649e-06, + "loss": 2.457928276062012, + "step": 261460 + }, + { + "epoch": 0.7118007349938751, + "grad_norm": 2.138772487640381, + "learning_rate": 5.771895870246695e-06, + "loss": 2.4723209381103515, + "step": 261480 + }, + { + "epoch": 0.7118551789846196, + "grad_norm": 2.1944711208343506, + "learning_rate": 5.770805506337741e-06, + "loss": 2.425481986999512, + "step": 261500 + }, + { + "epoch": 0.7119096229753641, + "grad_norm": 2.0937626361846924, + "learning_rate": 5.769715142428786e-06, + "loss": 2.4888113021850584, + "step": 261520 + }, + { + "epoch": 0.7119640669661086, + "grad_norm": 2.6034438610076904, + "learning_rate": 5.768624778519831e-06, + "loss": 2.4658182144165037, + "step": 261540 + }, + { + "epoch": 0.7120185109568531, + "grad_norm": 2.1359851360321045, + "learning_rate": 5.767534414610877e-06, + "loss": 2.4596099853515625, + "step": 261560 + }, + { + "epoch": 0.7120729549475977, + "grad_norm": 2.1821956634521484, + "learning_rate": 5.7664440507019225e-06, + "loss": 2.46854305267334, + "step": 261580 + }, + { + "epoch": 0.7121273989383422, + "grad_norm": 2.3543426990509033, + "learning_rate": 5.7653536867929675e-06, + "loss": 2.428422546386719, + "step": 261600 + }, + { + "epoch": 0.7121818429290867, + "grad_norm": 2.2715957164764404, + "learning_rate": 5.7642633228840125e-06, + "loss": 2.483066749572754, + "step": 261620 + }, + { + "epoch": 0.7122362869198312, + "grad_norm": 2.3850948810577393, + "learning_rate": 5.763172958975058e-06, + "loss": 2.4888233184814452, + "step": 261640 + }, + { + "epoch": 0.7122907309105757, + "grad_norm": 2.116032600402832, + "learning_rate": 5.762082595066104e-06, + "loss": 2.4769733428955076, + "step": 261660 + }, + { + "epoch": 0.7123451749013203, + "grad_norm": 2.185786247253418, + "learning_rate": 5.760992231157149e-06, + "loss": 2.470112991333008, + "step": 261680 + }, + { + "epoch": 0.7123996188920648, + "grad_norm": 2.2831037044525146, + "learning_rate": 5.759901867248194e-06, + "loss": 2.470040512084961, + "step": 261700 + }, + { + "epoch": 0.7124540628828093, + "grad_norm": 2.2858829498291016, + "learning_rate": 5.75881150333924e-06, + "loss": 2.45028133392334, + "step": 261720 + }, + { + "epoch": 0.7125085068735538, + "grad_norm": 2.303095817565918, + "learning_rate": 5.757721139430286e-06, + "loss": 2.4592012405395507, + "step": 261740 + }, + { + "epoch": 0.7125629508642983, + "grad_norm": 2.194279909133911, + "learning_rate": 5.756630775521331e-06, + "loss": 2.484100914001465, + "step": 261760 + }, + { + "epoch": 0.7126173948550428, + "grad_norm": 2.1400327682495117, + "learning_rate": 5.7555404116123755e-06, + "loss": 2.4562700271606444, + "step": 261780 + }, + { + "epoch": 0.7126718388457874, + "grad_norm": 2.217771053314209, + "learning_rate": 5.754450047703421e-06, + "loss": 2.522180938720703, + "step": 261800 + }, + { + "epoch": 0.7127262828365319, + "grad_norm": 2.2323219776153564, + "learning_rate": 5.753359683794467e-06, + "loss": 2.46502799987793, + "step": 261820 + }, + { + "epoch": 0.7127807268272764, + "grad_norm": 2.184126853942871, + "learning_rate": 5.752269319885512e-06, + "loss": 2.4837570190429688, + "step": 261840 + }, + { + "epoch": 0.712835170818021, + "grad_norm": 2.779646396636963, + "learning_rate": 5.751178955976557e-06, + "loss": 2.47790412902832, + "step": 261860 + }, + { + "epoch": 0.7128896148087654, + "grad_norm": 2.3035993576049805, + "learning_rate": 5.750088592067603e-06, + "loss": 2.469633674621582, + "step": 261880 + }, + { + "epoch": 0.71294405879951, + "grad_norm": 2.1638271808624268, + "learning_rate": 5.748998228158649e-06, + "loss": 2.4709457397460937, + "step": 261900 + }, + { + "epoch": 0.7129985027902546, + "grad_norm": 2.2753665447235107, + "learning_rate": 5.747907864249694e-06, + "loss": 2.474020004272461, + "step": 261920 + }, + { + "epoch": 0.7130529467809991, + "grad_norm": 2.2227556705474854, + "learning_rate": 5.746817500340739e-06, + "loss": 2.4949853897094725, + "step": 261940 + }, + { + "epoch": 0.7131073907717436, + "grad_norm": 2.214385747909546, + "learning_rate": 5.7457271364317844e-06, + "loss": 2.473471832275391, + "step": 261960 + }, + { + "epoch": 0.7131618347624881, + "grad_norm": 2.1158273220062256, + "learning_rate": 5.74463677252283e-06, + "loss": 2.4687095642089845, + "step": 261980 + }, + { + "epoch": 0.7132162787532326, + "grad_norm": 2.128668785095215, + "learning_rate": 5.743546408613875e-06, + "loss": 2.474464988708496, + "step": 262000 + }, + { + "epoch": 0.7132707227439772, + "grad_norm": 2.199892520904541, + "learning_rate": 5.74245604470492e-06, + "loss": 2.4848701477050783, + "step": 262020 + }, + { + "epoch": 0.7133251667347217, + "grad_norm": 2.2153851985931396, + "learning_rate": 5.741365680795966e-06, + "loss": 2.4700462341308596, + "step": 262040 + }, + { + "epoch": 0.7133796107254662, + "grad_norm": 2.057861566543579, + "learning_rate": 5.740275316887012e-06, + "loss": 2.4549982070922853, + "step": 262060 + }, + { + "epoch": 0.7134340547162107, + "grad_norm": 2.251741409301758, + "learning_rate": 5.739184952978058e-06, + "loss": 2.4265167236328127, + "step": 262080 + }, + { + "epoch": 0.7134884987069552, + "grad_norm": 2.2766778469085693, + "learning_rate": 5.738094589069102e-06, + "loss": 2.4502017974853514, + "step": 262100 + }, + { + "epoch": 0.7135429426976997, + "grad_norm": 2.128197431564331, + "learning_rate": 5.7370042251601475e-06, + "loss": 2.5117406845092773, + "step": 262120 + }, + { + "epoch": 0.7135973866884443, + "grad_norm": 2.143838882446289, + "learning_rate": 5.735913861251193e-06, + "loss": 2.460011672973633, + "step": 262140 + }, + { + "epoch": 0.7136518306791888, + "grad_norm": 2.132913827896118, + "learning_rate": 5.734823497342239e-06, + "loss": 2.480647850036621, + "step": 262160 + }, + { + "epoch": 0.7137062746699333, + "grad_norm": 2.1485795974731445, + "learning_rate": 5.733733133433283e-06, + "loss": 2.4581146240234375, + "step": 262180 + }, + { + "epoch": 0.7137607186606778, + "grad_norm": 2.127619743347168, + "learning_rate": 5.732642769524329e-06, + "loss": 2.471807289123535, + "step": 262200 + }, + { + "epoch": 0.7138151626514223, + "grad_norm": 2.2604806423187256, + "learning_rate": 5.731552405615375e-06, + "loss": 2.461966323852539, + "step": 262220 + }, + { + "epoch": 0.7138696066421669, + "grad_norm": 2.165349245071411, + "learning_rate": 5.730462041706421e-06, + "loss": 2.477589797973633, + "step": 262240 + }, + { + "epoch": 0.7139240506329114, + "grad_norm": 2.159661293029785, + "learning_rate": 5.729371677797465e-06, + "loss": 2.465524673461914, + "step": 262260 + }, + { + "epoch": 0.7139784946236559, + "grad_norm": 2.1230103969573975, + "learning_rate": 5.728281313888511e-06, + "loss": 2.4420433044433594, + "step": 262280 + }, + { + "epoch": 0.7140329386144004, + "grad_norm": 2.0903406143188477, + "learning_rate": 5.727190949979556e-06, + "loss": 2.4680545806884764, + "step": 262300 + }, + { + "epoch": 0.7140873826051449, + "grad_norm": 2.0709407329559326, + "learning_rate": 5.726100586070602e-06, + "loss": 2.449498176574707, + "step": 262320 + }, + { + "epoch": 0.7141418265958894, + "grad_norm": 2.1162378787994385, + "learning_rate": 5.725010222161646e-06, + "loss": 2.430336570739746, + "step": 262340 + }, + { + "epoch": 0.714196270586634, + "grad_norm": 2.183098554611206, + "learning_rate": 5.723919858252692e-06, + "loss": 2.4646141052246096, + "step": 262360 + }, + { + "epoch": 0.7142507145773785, + "grad_norm": 2.119663715362549, + "learning_rate": 5.722829494343738e-06, + "loss": 2.459064483642578, + "step": 262380 + }, + { + "epoch": 0.714305158568123, + "grad_norm": 2.101686954498291, + "learning_rate": 5.721739130434784e-06, + "loss": 2.4800088882446287, + "step": 262400 + }, + { + "epoch": 0.7143596025588675, + "grad_norm": 2.287936210632324, + "learning_rate": 5.720648766525828e-06, + "loss": 2.4641611099243166, + "step": 262420 + }, + { + "epoch": 0.714414046549612, + "grad_norm": 2.0431644916534424, + "learning_rate": 5.719558402616874e-06, + "loss": 2.446469306945801, + "step": 262440 + }, + { + "epoch": 0.7144684905403567, + "grad_norm": 2.2064061164855957, + "learning_rate": 5.7184680387079195e-06, + "loss": 2.5142593383789062, + "step": 262460 + }, + { + "epoch": 0.7145229345311012, + "grad_norm": 2.177323579788208, + "learning_rate": 5.717377674798965e-06, + "loss": 2.457649993896484, + "step": 262480 + }, + { + "epoch": 0.7145773785218457, + "grad_norm": 2.141373634338379, + "learning_rate": 5.7162873108900094e-06, + "loss": 2.493073272705078, + "step": 262500 + }, + { + "epoch": 0.7146318225125902, + "grad_norm": 2.2091054916381836, + "learning_rate": 5.715196946981055e-06, + "loss": 2.47845516204834, + "step": 262520 + }, + { + "epoch": 0.7146862665033347, + "grad_norm": 2.1432502269744873, + "learning_rate": 5.714106583072101e-06, + "loss": 2.451591682434082, + "step": 262540 + }, + { + "epoch": 0.7147407104940792, + "grad_norm": 2.1953377723693848, + "learning_rate": 5.713016219163147e-06, + "loss": 2.460308074951172, + "step": 262560 + }, + { + "epoch": 0.7147951544848238, + "grad_norm": 2.2232532501220703, + "learning_rate": 5.711925855254191e-06, + "loss": 2.4707614898681642, + "step": 262580 + }, + { + "epoch": 0.7148495984755683, + "grad_norm": 2.1762378215789795, + "learning_rate": 5.710835491345237e-06, + "loss": 2.4842035293579103, + "step": 262600 + }, + { + "epoch": 0.7149040424663128, + "grad_norm": 2.1756441593170166, + "learning_rate": 5.709745127436283e-06, + "loss": 2.4714752197265626, + "step": 262620 + }, + { + "epoch": 0.7149584864570573, + "grad_norm": 2.619476079940796, + "learning_rate": 5.708654763527328e-06, + "loss": 2.441373825073242, + "step": 262640 + }, + { + "epoch": 0.7150129304478018, + "grad_norm": 2.3051106929779053, + "learning_rate": 5.7075643996183725e-06, + "loss": 2.4578418731689453, + "step": 262660 + }, + { + "epoch": 0.7150673744385464, + "grad_norm": 2.5024616718292236, + "learning_rate": 5.706474035709418e-06, + "loss": 2.486942672729492, + "step": 262680 + }, + { + "epoch": 0.7151218184292909, + "grad_norm": 2.328251600265503, + "learning_rate": 5.705383671800464e-06, + "loss": 2.4599485397338867, + "step": 262700 + }, + { + "epoch": 0.7151762624200354, + "grad_norm": 2.237823963165283, + "learning_rate": 5.70429330789151e-06, + "loss": 2.467120552062988, + "step": 262720 + }, + { + "epoch": 0.7152307064107799, + "grad_norm": 2.230823516845703, + "learning_rate": 5.703202943982554e-06, + "loss": 2.4324905395507814, + "step": 262740 + }, + { + "epoch": 0.7152851504015244, + "grad_norm": 2.265366792678833, + "learning_rate": 5.7021125800736e-06, + "loss": 2.463506507873535, + "step": 262760 + }, + { + "epoch": 0.7153395943922689, + "grad_norm": 2.141375780105591, + "learning_rate": 5.701022216164646e-06, + "loss": 2.483743667602539, + "step": 262780 + }, + { + "epoch": 0.7153940383830135, + "grad_norm": 2.2319753170013428, + "learning_rate": 5.6999318522556915e-06, + "loss": 2.518027687072754, + "step": 262800 + }, + { + "epoch": 0.715448482373758, + "grad_norm": 2.3209455013275146, + "learning_rate": 5.698841488346736e-06, + "loss": 2.477349853515625, + "step": 262820 + }, + { + "epoch": 0.7155029263645025, + "grad_norm": 2.2897324562072754, + "learning_rate": 5.697751124437781e-06, + "loss": 2.4480085372924805, + "step": 262840 + }, + { + "epoch": 0.715557370355247, + "grad_norm": 2.1937716007232666, + "learning_rate": 5.696660760528827e-06, + "loss": 2.4449302673339846, + "step": 262860 + }, + { + "epoch": 0.7156118143459915, + "grad_norm": 2.146775722503662, + "learning_rate": 5.695570396619873e-06, + "loss": 2.449961471557617, + "step": 262880 + }, + { + "epoch": 0.715666258336736, + "grad_norm": 2.252786636352539, + "learning_rate": 5.694480032710917e-06, + "loss": 2.4573450088500977, + "step": 262900 + }, + { + "epoch": 0.7157207023274806, + "grad_norm": 2.8124797344207764, + "learning_rate": 5.693389668801963e-06, + "loss": 2.4414466857910155, + "step": 262920 + }, + { + "epoch": 0.7157751463182251, + "grad_norm": 2.0960025787353516, + "learning_rate": 5.692299304893009e-06, + "loss": 2.479448127746582, + "step": 262940 + }, + { + "epoch": 0.7158295903089696, + "grad_norm": 2.326531171798706, + "learning_rate": 5.6912089409840546e-06, + "loss": 2.4869861602783203, + "step": 262960 + }, + { + "epoch": 0.7158840342997141, + "grad_norm": 2.1157126426696777, + "learning_rate": 5.690118577075099e-06, + "loss": 2.4771629333496095, + "step": 262980 + }, + { + "epoch": 0.7159384782904586, + "grad_norm": 2.4049320220947266, + "learning_rate": 5.6890282131661445e-06, + "loss": 2.4597745895385743, + "step": 263000 + }, + { + "epoch": 0.7159929222812033, + "grad_norm": 2.120011568069458, + "learning_rate": 5.68793784925719e-06, + "loss": 2.4772621154785157, + "step": 263020 + }, + { + "epoch": 0.7160473662719478, + "grad_norm": 2.1742992401123047, + "learning_rate": 5.686847485348236e-06, + "loss": 2.4596757888793945, + "step": 263040 + }, + { + "epoch": 0.7161018102626923, + "grad_norm": 2.6924285888671875, + "learning_rate": 5.68575712143928e-06, + "loss": 2.4510530471801757, + "step": 263060 + }, + { + "epoch": 0.7161562542534368, + "grad_norm": 2.33935809135437, + "learning_rate": 5.684666757530326e-06, + "loss": 2.484193229675293, + "step": 263080 + }, + { + "epoch": 0.7162106982441813, + "grad_norm": 2.1487789154052734, + "learning_rate": 5.683576393621372e-06, + "loss": 2.452068328857422, + "step": 263100 + }, + { + "epoch": 0.7162651422349258, + "grad_norm": 2.885380983352661, + "learning_rate": 5.682486029712418e-06, + "loss": 2.4682809829711916, + "step": 263120 + }, + { + "epoch": 0.7163195862256704, + "grad_norm": 2.6012156009674072, + "learning_rate": 5.681395665803462e-06, + "loss": 2.4676881790161134, + "step": 263140 + }, + { + "epoch": 0.7163740302164149, + "grad_norm": 2.3914132118225098, + "learning_rate": 5.680305301894508e-06, + "loss": 2.46326904296875, + "step": 263160 + }, + { + "epoch": 0.7164284742071594, + "grad_norm": 2.2529680728912354, + "learning_rate": 5.679214937985553e-06, + "loss": 2.4684528350830077, + "step": 263180 + }, + { + "epoch": 0.7164829181979039, + "grad_norm": 2.42870831489563, + "learning_rate": 5.678124574076599e-06, + "loss": 2.4881654739379884, + "step": 263200 + }, + { + "epoch": 0.7165373621886484, + "grad_norm": 2.2144927978515625, + "learning_rate": 5.677034210167643e-06, + "loss": 2.473519515991211, + "step": 263220 + }, + { + "epoch": 0.716591806179393, + "grad_norm": 2.2130026817321777, + "learning_rate": 5.675943846258689e-06, + "loss": 2.495940399169922, + "step": 263240 + }, + { + "epoch": 0.7166462501701375, + "grad_norm": 2.203342914581299, + "learning_rate": 5.674853482349735e-06, + "loss": 2.4694341659545898, + "step": 263260 + }, + { + "epoch": 0.716700694160882, + "grad_norm": 2.2311463356018066, + "learning_rate": 5.673763118440781e-06, + "loss": 2.4714309692382814, + "step": 263280 + }, + { + "epoch": 0.7167551381516265, + "grad_norm": 2.22001576423645, + "learning_rate": 5.672672754531825e-06, + "loss": 2.4780521392822266, + "step": 263300 + }, + { + "epoch": 0.716809582142371, + "grad_norm": 2.165001392364502, + "learning_rate": 5.671582390622871e-06, + "loss": 2.44540958404541, + "step": 263320 + }, + { + "epoch": 0.7168640261331155, + "grad_norm": 2.2865328788757324, + "learning_rate": 5.6704920267139165e-06, + "loss": 2.4621408462524412, + "step": 263340 + }, + { + "epoch": 0.7169184701238601, + "grad_norm": 2.3471486568450928, + "learning_rate": 5.669401662804962e-06, + "loss": 2.48131103515625, + "step": 263360 + }, + { + "epoch": 0.7169729141146046, + "grad_norm": 2.2150049209594727, + "learning_rate": 5.668311298896006e-06, + "loss": 2.4599782943725588, + "step": 263380 + }, + { + "epoch": 0.7170273581053491, + "grad_norm": 2.404400110244751, + "learning_rate": 5.667220934987052e-06, + "loss": 2.4483417510986327, + "step": 263400 + }, + { + "epoch": 0.7170818020960936, + "grad_norm": 2.06893253326416, + "learning_rate": 5.666130571078098e-06, + "loss": 2.450872230529785, + "step": 263420 + }, + { + "epoch": 0.7171362460868381, + "grad_norm": 2.1002814769744873, + "learning_rate": 5.665040207169144e-06, + "loss": 2.470054054260254, + "step": 263440 + }, + { + "epoch": 0.7171906900775827, + "grad_norm": 2.128734827041626, + "learning_rate": 5.663949843260188e-06, + "loss": 2.4660018920898437, + "step": 263460 + }, + { + "epoch": 0.7172451340683272, + "grad_norm": 2.111721992492676, + "learning_rate": 5.662859479351234e-06, + "loss": 2.4438554763793947, + "step": 263480 + }, + { + "epoch": 0.7172995780590717, + "grad_norm": 2.153696060180664, + "learning_rate": 5.6617691154422796e-06, + "loss": 2.480026435852051, + "step": 263500 + }, + { + "epoch": 0.7173540220498162, + "grad_norm": 2.21915602684021, + "learning_rate": 5.660678751533325e-06, + "loss": 2.4813411712646483, + "step": 263520 + }, + { + "epoch": 0.7174084660405607, + "grad_norm": 2.1150574684143066, + "learning_rate": 5.6595883876243695e-06, + "loss": 2.452133560180664, + "step": 263540 + }, + { + "epoch": 0.7174629100313052, + "grad_norm": 2.0960638523101807, + "learning_rate": 5.658498023715415e-06, + "loss": 2.452083969116211, + "step": 263560 + }, + { + "epoch": 0.7175173540220499, + "grad_norm": 2.2100911140441895, + "learning_rate": 5.657407659806461e-06, + "loss": 2.4779052734375, + "step": 263580 + }, + { + "epoch": 0.7175717980127944, + "grad_norm": 2.13100266456604, + "learning_rate": 5.656317295897507e-06, + "loss": 2.459911346435547, + "step": 263600 + }, + { + "epoch": 0.7176262420035389, + "grad_norm": 2.2048490047454834, + "learning_rate": 5.655226931988551e-06, + "loss": 2.4771343231201173, + "step": 263620 + }, + { + "epoch": 0.7176806859942834, + "grad_norm": 2.2623817920684814, + "learning_rate": 5.654136568079597e-06, + "loss": 2.508987236022949, + "step": 263640 + }, + { + "epoch": 0.7177351299850279, + "grad_norm": 2.1031241416931152, + "learning_rate": 5.653046204170643e-06, + "loss": 2.4692840576171875, + "step": 263660 + }, + { + "epoch": 0.7177895739757725, + "grad_norm": 2.178926944732666, + "learning_rate": 5.6519558402616885e-06, + "loss": 2.4727109909057616, + "step": 263680 + }, + { + "epoch": 0.717844017966517, + "grad_norm": 2.081800699234009, + "learning_rate": 5.6508654763527326e-06, + "loss": 2.4770477294921873, + "step": 263700 + }, + { + "epoch": 0.7178984619572615, + "grad_norm": 2.151404857635498, + "learning_rate": 5.649775112443778e-06, + "loss": 2.4734100341796874, + "step": 263720 + }, + { + "epoch": 0.717952905948006, + "grad_norm": 2.1243784427642822, + "learning_rate": 5.648684748534824e-06, + "loss": 2.450893020629883, + "step": 263740 + }, + { + "epoch": 0.7180073499387505, + "grad_norm": 2.1930503845214844, + "learning_rate": 5.64759438462587e-06, + "loss": 2.453316497802734, + "step": 263760 + }, + { + "epoch": 0.718061793929495, + "grad_norm": 2.2496531009674072, + "learning_rate": 5.646504020716915e-06, + "loss": 2.4933492660522463, + "step": 263780 + }, + { + "epoch": 0.7181162379202396, + "grad_norm": 2.1443302631378174, + "learning_rate": 5.64541365680796e-06, + "loss": 2.4530439376831055, + "step": 263800 + }, + { + "epoch": 0.7181706819109841, + "grad_norm": 2.1590912342071533, + "learning_rate": 5.644323292899006e-06, + "loss": 2.4751422882080076, + "step": 263820 + }, + { + "epoch": 0.7182251259017286, + "grad_norm": 2.175381898880005, + "learning_rate": 5.6432329289900515e-06, + "loss": 2.4598623275756837, + "step": 263840 + }, + { + "epoch": 0.7182795698924731, + "grad_norm": 2.135852575302124, + "learning_rate": 5.6421425650810965e-06, + "loss": 2.4709003448486326, + "step": 263860 + }, + { + "epoch": 0.7183340138832176, + "grad_norm": 2.2754054069519043, + "learning_rate": 5.6410522011721415e-06, + "loss": 2.474521255493164, + "step": 263880 + }, + { + "epoch": 0.7183884578739622, + "grad_norm": 2.2080681324005127, + "learning_rate": 5.639961837263187e-06, + "loss": 2.423064422607422, + "step": 263900 + }, + { + "epoch": 0.7184429018647067, + "grad_norm": 2.1890130043029785, + "learning_rate": 5.638871473354233e-06, + "loss": 2.485205078125, + "step": 263920 + }, + { + "epoch": 0.7184973458554512, + "grad_norm": 2.1245551109313965, + "learning_rate": 5.637781109445278e-06, + "loss": 2.460978889465332, + "step": 263940 + }, + { + "epoch": 0.7185517898461957, + "grad_norm": 2.1151981353759766, + "learning_rate": 5.636690745536323e-06, + "loss": 2.4644121170043944, + "step": 263960 + }, + { + "epoch": 0.7186062338369402, + "grad_norm": 2.243572473526001, + "learning_rate": 5.635600381627369e-06, + "loss": 2.477871894836426, + "step": 263980 + }, + { + "epoch": 0.7186606778276847, + "grad_norm": 2.197068691253662, + "learning_rate": 5.634510017718415e-06, + "loss": 2.4944173812866213, + "step": 264000 + }, + { + "epoch": 0.7187151218184293, + "grad_norm": 2.2727127075195312, + "learning_rate": 5.63341965380946e-06, + "loss": 2.447886276245117, + "step": 264020 + }, + { + "epoch": 0.7187695658091738, + "grad_norm": 2.3482601642608643, + "learning_rate": 5.6323292899005045e-06, + "loss": 2.488785743713379, + "step": 264040 + }, + { + "epoch": 0.7188240097999183, + "grad_norm": 2.2470357418060303, + "learning_rate": 5.63123892599155e-06, + "loss": 2.4342891693115236, + "step": 264060 + }, + { + "epoch": 0.7188784537906628, + "grad_norm": 2.162777900695801, + "learning_rate": 5.630148562082596e-06, + "loss": 2.4877119064331055, + "step": 264080 + }, + { + "epoch": 0.7189328977814073, + "grad_norm": 3.034318208694458, + "learning_rate": 5.629058198173641e-06, + "loss": 2.4609474182128905, + "step": 264100 + }, + { + "epoch": 0.7189873417721518, + "grad_norm": 2.259232759475708, + "learning_rate": 5.627967834264686e-06, + "loss": 2.4608390808105467, + "step": 264120 + }, + { + "epoch": 0.7190417857628965, + "grad_norm": 4.028125286102295, + "learning_rate": 5.626877470355732e-06, + "loss": 2.470561408996582, + "step": 264140 + }, + { + "epoch": 0.719096229753641, + "grad_norm": 2.1478919982910156, + "learning_rate": 5.625787106446777e-06, + "loss": 2.4755619049072264, + "step": 264160 + }, + { + "epoch": 0.7191506737443855, + "grad_norm": 2.1336212158203125, + "learning_rate": 5.624696742537823e-06, + "loss": 2.4792123794555665, + "step": 264180 + }, + { + "epoch": 0.71920511773513, + "grad_norm": 2.0944042205810547, + "learning_rate": 5.623606378628868e-06, + "loss": 2.479501724243164, + "step": 264200 + }, + { + "epoch": 0.7192595617258745, + "grad_norm": 2.161433458328247, + "learning_rate": 5.6225160147199134e-06, + "loss": 2.4827510833740236, + "step": 264220 + }, + { + "epoch": 0.7193140057166191, + "grad_norm": 2.17866849899292, + "learning_rate": 5.621425650810958e-06, + "loss": 2.471552276611328, + "step": 264240 + }, + { + "epoch": 0.7193684497073636, + "grad_norm": 2.1866061687469482, + "learning_rate": 5.620335286902004e-06, + "loss": 2.447291946411133, + "step": 264260 + }, + { + "epoch": 0.7194228936981081, + "grad_norm": 2.099342107772827, + "learning_rate": 5.619244922993049e-06, + "loss": 2.477312469482422, + "step": 264280 + }, + { + "epoch": 0.7194773376888526, + "grad_norm": 2.205947160720825, + "learning_rate": 5.618154559084095e-06, + "loss": 2.42673397064209, + "step": 264300 + }, + { + "epoch": 0.7195317816795971, + "grad_norm": 2.2081685066223145, + "learning_rate": 5.61706419517514e-06, + "loss": 2.4531681060791017, + "step": 264320 + }, + { + "epoch": 0.7195862256703416, + "grad_norm": 2.1996309757232666, + "learning_rate": 5.615973831266186e-06, + "loss": 2.4689125061035155, + "step": 264340 + }, + { + "epoch": 0.7196406696610862, + "grad_norm": 2.1109418869018555, + "learning_rate": 5.614883467357231e-06, + "loss": 2.4572622299194338, + "step": 264360 + }, + { + "epoch": 0.7196951136518307, + "grad_norm": 2.3862483501434326, + "learning_rate": 5.6137931034482765e-06, + "loss": 2.494122314453125, + "step": 264380 + }, + { + "epoch": 0.7197495576425752, + "grad_norm": 2.1388843059539795, + "learning_rate": 5.6127027395393215e-06, + "loss": 2.4636207580566407, + "step": 264400 + }, + { + "epoch": 0.7198040016333197, + "grad_norm": 2.1511642932891846, + "learning_rate": 5.611612375630367e-06, + "loss": 2.4992450714111327, + "step": 264420 + }, + { + "epoch": 0.7198584456240642, + "grad_norm": 2.012256145477295, + "learning_rate": 5.610522011721412e-06, + "loss": 2.4471109390258787, + "step": 264440 + }, + { + "epoch": 0.7199128896148088, + "grad_norm": 2.0803582668304443, + "learning_rate": 5.609431647812458e-06, + "loss": 2.4641092300415037, + "step": 264460 + }, + { + "epoch": 0.7199673336055533, + "grad_norm": 2.3427226543426514, + "learning_rate": 5.608341283903503e-06, + "loss": 2.465652847290039, + "step": 264480 + }, + { + "epoch": 0.7200217775962978, + "grad_norm": 2.1695613861083984, + "learning_rate": 5.607250919994549e-06, + "loss": 2.508738136291504, + "step": 264500 + }, + { + "epoch": 0.7200762215870423, + "grad_norm": 2.1816134452819824, + "learning_rate": 5.606160556085594e-06, + "loss": 2.4819223403930666, + "step": 264520 + }, + { + "epoch": 0.7201306655777868, + "grad_norm": 2.1256649494171143, + "learning_rate": 5.60507019217664e-06, + "loss": 2.4454139709472655, + "step": 264540 + }, + { + "epoch": 0.7201851095685313, + "grad_norm": 2.293055295944214, + "learning_rate": 5.6039798282676846e-06, + "loss": 2.488006591796875, + "step": 264560 + }, + { + "epoch": 0.7202395535592759, + "grad_norm": 2.313387632369995, + "learning_rate": 5.60288946435873e-06, + "loss": 2.4509220123291016, + "step": 264580 + }, + { + "epoch": 0.7202939975500204, + "grad_norm": 2.0822978019714355, + "learning_rate": 5.601799100449775e-06, + "loss": 2.4504438400268556, + "step": 264600 + }, + { + "epoch": 0.7203484415407649, + "grad_norm": 2.4722111225128174, + "learning_rate": 5.600708736540821e-06, + "loss": 2.4584794998168946, + "step": 264620 + }, + { + "epoch": 0.7204028855315094, + "grad_norm": 2.214951992034912, + "learning_rate": 5.599618372631866e-06, + "loss": 2.457798194885254, + "step": 264640 + }, + { + "epoch": 0.7204573295222539, + "grad_norm": 2.4348881244659424, + "learning_rate": 5.598528008722912e-06, + "loss": 2.4577560424804688, + "step": 264660 + }, + { + "epoch": 0.7205117735129986, + "grad_norm": 2.1402742862701416, + "learning_rate": 5.597437644813957e-06, + "loss": 2.4599130630493162, + "step": 264680 + }, + { + "epoch": 0.7205662175037431, + "grad_norm": 2.0933125019073486, + "learning_rate": 5.596347280905003e-06, + "loss": 2.5028587341308595, + "step": 264700 + }, + { + "epoch": 0.7206206614944876, + "grad_norm": 2.086519718170166, + "learning_rate": 5.595256916996048e-06, + "loss": 2.441948127746582, + "step": 264720 + }, + { + "epoch": 0.7206751054852321, + "grad_norm": 2.2566659450531006, + "learning_rate": 5.5941665530870935e-06, + "loss": 2.463831329345703, + "step": 264740 + }, + { + "epoch": 0.7207295494759766, + "grad_norm": 2.3134331703186035, + "learning_rate": 5.5930761891781384e-06, + "loss": 2.46997127532959, + "step": 264760 + }, + { + "epoch": 0.7207839934667211, + "grad_norm": 2.147765874862671, + "learning_rate": 5.591985825269184e-06, + "loss": 2.4461996078491213, + "step": 264780 + }, + { + "epoch": 0.7208384374574657, + "grad_norm": 2.086531162261963, + "learning_rate": 5.590895461360229e-06, + "loss": 2.4445709228515624, + "step": 264800 + }, + { + "epoch": 0.7208928814482102, + "grad_norm": 2.084160804748535, + "learning_rate": 5.589805097451275e-06, + "loss": 2.4473737716674804, + "step": 264820 + }, + { + "epoch": 0.7209473254389547, + "grad_norm": 2.0631186962127686, + "learning_rate": 5.58871473354232e-06, + "loss": 2.483388328552246, + "step": 264840 + }, + { + "epoch": 0.7210017694296992, + "grad_norm": 2.2807767391204834, + "learning_rate": 5.587624369633366e-06, + "loss": 2.46937255859375, + "step": 264860 + }, + { + "epoch": 0.7210562134204437, + "grad_norm": 2.110947370529175, + "learning_rate": 5.586534005724411e-06, + "loss": 2.4407121658325197, + "step": 264880 + }, + { + "epoch": 0.7211106574111883, + "grad_norm": 2.22296142578125, + "learning_rate": 5.5854436418154566e-06, + "loss": 2.445673179626465, + "step": 264900 + }, + { + "epoch": 0.7211651014019328, + "grad_norm": 2.3259482383728027, + "learning_rate": 5.5843532779065015e-06, + "loss": 2.4638671875, + "step": 264920 + }, + { + "epoch": 0.7212195453926773, + "grad_norm": 2.078463315963745, + "learning_rate": 5.583262913997547e-06, + "loss": 2.4615840911865234, + "step": 264940 + }, + { + "epoch": 0.7212739893834218, + "grad_norm": 2.1806745529174805, + "learning_rate": 5.582172550088592e-06, + "loss": 2.4830820083618166, + "step": 264960 + }, + { + "epoch": 0.7213284333741663, + "grad_norm": 2.2060539722442627, + "learning_rate": 5.581082186179638e-06, + "loss": 2.4928213119506837, + "step": 264980 + }, + { + "epoch": 0.7213828773649108, + "grad_norm": 2.1846117973327637, + "learning_rate": 5.579991822270683e-06, + "loss": 2.450822448730469, + "step": 265000 + }, + { + "epoch": 0.7214373213556554, + "grad_norm": 2.146695137023926, + "learning_rate": 5.578901458361729e-06, + "loss": 2.4744020462036134, + "step": 265020 + }, + { + "epoch": 0.7214917653463999, + "grad_norm": 2.2366044521331787, + "learning_rate": 5.577811094452774e-06, + "loss": 2.4916975021362306, + "step": 265040 + }, + { + "epoch": 0.7215462093371444, + "grad_norm": 2.2035412788391113, + "learning_rate": 5.57672073054382e-06, + "loss": 2.4595973968505858, + "step": 265060 + }, + { + "epoch": 0.7216006533278889, + "grad_norm": 2.0787057876586914, + "learning_rate": 5.575630366634865e-06, + "loss": 2.4564510345458985, + "step": 265080 + }, + { + "epoch": 0.7216550973186334, + "grad_norm": 2.1441903114318848, + "learning_rate": 5.57454000272591e-06, + "loss": 2.458936309814453, + "step": 265100 + }, + { + "epoch": 0.7217095413093779, + "grad_norm": 2.20794677734375, + "learning_rate": 5.573449638816955e-06, + "loss": 2.451192855834961, + "step": 265120 + }, + { + "epoch": 0.7217639853001225, + "grad_norm": 2.14927077293396, + "learning_rate": 5.572359274908001e-06, + "loss": 2.4751089096069334, + "step": 265140 + }, + { + "epoch": 0.721818429290867, + "grad_norm": 2.1229052543640137, + "learning_rate": 5.571268910999046e-06, + "loss": 2.4541696548461913, + "step": 265160 + }, + { + "epoch": 0.7218728732816115, + "grad_norm": 2.1467981338500977, + "learning_rate": 5.570178547090092e-06, + "loss": 2.4574663162231447, + "step": 265180 + }, + { + "epoch": 0.721927317272356, + "grad_norm": 2.328646421432495, + "learning_rate": 5.569088183181137e-06, + "loss": 2.4385761260986327, + "step": 265200 + }, + { + "epoch": 0.7219817612631005, + "grad_norm": 2.119055986404419, + "learning_rate": 5.567997819272183e-06, + "loss": 2.4586353302001953, + "step": 265220 + }, + { + "epoch": 0.7220362052538452, + "grad_norm": 2.177353858947754, + "learning_rate": 5.566907455363228e-06, + "loss": 2.446742630004883, + "step": 265240 + }, + { + "epoch": 0.7220906492445897, + "grad_norm": 2.2168996334075928, + "learning_rate": 5.565817091454273e-06, + "loss": 2.476353645324707, + "step": 265260 + }, + { + "epoch": 0.7221450932353342, + "grad_norm": 2.19124698638916, + "learning_rate": 5.5647267275453185e-06, + "loss": 2.5054012298583985, + "step": 265280 + }, + { + "epoch": 0.7221995372260787, + "grad_norm": 2.1054537296295166, + "learning_rate": 5.563636363636364e-06, + "loss": 2.4829133987426757, + "step": 265300 + }, + { + "epoch": 0.7222539812168232, + "grad_norm": 2.2770986557006836, + "learning_rate": 5.562545999727409e-06, + "loss": 2.4766759872436523, + "step": 265320 + }, + { + "epoch": 0.7223084252075677, + "grad_norm": 2.096186876296997, + "learning_rate": 5.561455635818454e-06, + "loss": 2.4704822540283202, + "step": 265340 + }, + { + "epoch": 0.7223628691983123, + "grad_norm": 2.1761746406555176, + "learning_rate": 5.5603652719095e-06, + "loss": 2.515234184265137, + "step": 265360 + }, + { + "epoch": 0.7224173131890568, + "grad_norm": 2.072622537612915, + "learning_rate": 5.559274908000546e-06, + "loss": 2.454246711730957, + "step": 265380 + }, + { + "epoch": 0.7224717571798013, + "grad_norm": 2.127833604812622, + "learning_rate": 5.558184544091591e-06, + "loss": 2.4365411758422852, + "step": 265400 + }, + { + "epoch": 0.7225262011705458, + "grad_norm": 2.2302515506744385, + "learning_rate": 5.557094180182636e-06, + "loss": 2.4235309600830077, + "step": 265420 + }, + { + "epoch": 0.7225806451612903, + "grad_norm": 2.213988780975342, + "learning_rate": 5.5560038162736815e-06, + "loss": 2.4521379470825195, + "step": 265440 + }, + { + "epoch": 0.7226350891520349, + "grad_norm": 2.1838083267211914, + "learning_rate": 5.554913452364727e-06, + "loss": 2.4465362548828127, + "step": 265460 + }, + { + "epoch": 0.7226895331427794, + "grad_norm": 2.11051344871521, + "learning_rate": 5.553823088455773e-06, + "loss": 2.46944522857666, + "step": 265480 + }, + { + "epoch": 0.7227439771335239, + "grad_norm": 2.21545672416687, + "learning_rate": 5.552732724546817e-06, + "loss": 2.4655542373657227, + "step": 265500 + }, + { + "epoch": 0.7227984211242684, + "grad_norm": 2.265408515930176, + "learning_rate": 5.551642360637863e-06, + "loss": 2.451296806335449, + "step": 265520 + }, + { + "epoch": 0.7228528651150129, + "grad_norm": 2.38443660736084, + "learning_rate": 5.550551996728909e-06, + "loss": 2.4660573959350587, + "step": 265540 + }, + { + "epoch": 0.7229073091057574, + "grad_norm": 2.1145758628845215, + "learning_rate": 5.549461632819955e-06, + "loss": 2.4890682220458986, + "step": 265560 + }, + { + "epoch": 0.722961753096502, + "grad_norm": 2.3577585220336914, + "learning_rate": 5.548371268910999e-06, + "loss": 2.4297399520874023, + "step": 265580 + }, + { + "epoch": 0.7230161970872465, + "grad_norm": 2.1312994956970215, + "learning_rate": 5.547280905002045e-06, + "loss": 2.4561676025390624, + "step": 265600 + }, + { + "epoch": 0.723070641077991, + "grad_norm": 2.167227268218994, + "learning_rate": 5.5461905410930904e-06, + "loss": 2.44588565826416, + "step": 265620 + }, + { + "epoch": 0.7231250850687355, + "grad_norm": 2.194406032562256, + "learning_rate": 5.545100177184136e-06, + "loss": 2.486087417602539, + "step": 265640 + }, + { + "epoch": 0.72317952905948, + "grad_norm": 2.0604915618896484, + "learning_rate": 5.54400981327518e-06, + "loss": 2.483621025085449, + "step": 265660 + }, + { + "epoch": 0.7232339730502246, + "grad_norm": 2.271528482437134, + "learning_rate": 5.542919449366226e-06, + "loss": 2.4630464553833007, + "step": 265680 + }, + { + "epoch": 0.7232884170409691, + "grad_norm": 2.1330769062042236, + "learning_rate": 5.541829085457272e-06, + "loss": 2.4784523010253907, + "step": 265700 + }, + { + "epoch": 0.7233428610317136, + "grad_norm": 2.1924948692321777, + "learning_rate": 5.540738721548318e-06, + "loss": 2.4840951919555665, + "step": 265720 + }, + { + "epoch": 0.7233973050224581, + "grad_norm": 2.093874216079712, + "learning_rate": 5.539648357639362e-06, + "loss": 2.4516231536865236, + "step": 265740 + }, + { + "epoch": 0.7234517490132026, + "grad_norm": 2.2507436275482178, + "learning_rate": 5.538557993730408e-06, + "loss": 2.4418708801269533, + "step": 265760 + }, + { + "epoch": 0.7235061930039471, + "grad_norm": 2.094421625137329, + "learning_rate": 5.5374676298214535e-06, + "loss": 2.4662303924560547, + "step": 265780 + }, + { + "epoch": 0.7235606369946918, + "grad_norm": 2.1718780994415283, + "learning_rate": 5.536377265912499e-06, + "loss": 2.490883445739746, + "step": 265800 + }, + { + "epoch": 0.7236150809854363, + "grad_norm": 2.243412733078003, + "learning_rate": 5.5352869020035435e-06, + "loss": 2.455329704284668, + "step": 265820 + }, + { + "epoch": 0.7236695249761808, + "grad_norm": 2.305760622024536, + "learning_rate": 5.534196538094589e-06, + "loss": 2.4875545501708984, + "step": 265840 + }, + { + "epoch": 0.7237239689669253, + "grad_norm": 3.1403608322143555, + "learning_rate": 5.533106174185635e-06, + "loss": 2.453879737854004, + "step": 265860 + }, + { + "epoch": 0.7237784129576698, + "grad_norm": 3.1854820251464844, + "learning_rate": 5.532015810276681e-06, + "loss": 2.500527191162109, + "step": 265880 + }, + { + "epoch": 0.7238328569484144, + "grad_norm": 2.099782943725586, + "learning_rate": 5.530925446367725e-06, + "loss": 2.477072906494141, + "step": 265900 + }, + { + "epoch": 0.7238873009391589, + "grad_norm": 2.1136114597320557, + "learning_rate": 5.529835082458771e-06, + "loss": 2.4924259185791016, + "step": 265920 + }, + { + "epoch": 0.7239417449299034, + "grad_norm": 2.147261381149292, + "learning_rate": 5.528744718549817e-06, + "loss": 2.480025291442871, + "step": 265940 + }, + { + "epoch": 0.7239961889206479, + "grad_norm": 2.3773951530456543, + "learning_rate": 5.527654354640862e-06, + "loss": 2.474856376647949, + "step": 265960 + }, + { + "epoch": 0.7240506329113924, + "grad_norm": 2.2485082149505615, + "learning_rate": 5.5265639907319065e-06, + "loss": 2.4740428924560547, + "step": 265980 + }, + { + "epoch": 0.7241050769021369, + "grad_norm": 2.1120736598968506, + "learning_rate": 5.525473626822952e-06, + "loss": 2.4375307083129885, + "step": 266000 + }, + { + "epoch": 0.7241595208928815, + "grad_norm": 2.159844398498535, + "learning_rate": 5.524383262913998e-06, + "loss": 2.4822982788085937, + "step": 266020 + }, + { + "epoch": 0.724213964883626, + "grad_norm": 2.145636796951294, + "learning_rate": 5.523292899005044e-06, + "loss": 2.49511604309082, + "step": 266040 + }, + { + "epoch": 0.7242684088743705, + "grad_norm": 2.311858654022217, + "learning_rate": 5.522202535096088e-06, + "loss": 2.47888069152832, + "step": 266060 + }, + { + "epoch": 0.724322852865115, + "grad_norm": 2.1886162757873535, + "learning_rate": 5.521112171187134e-06, + "loss": 2.4637256622314454, + "step": 266080 + }, + { + "epoch": 0.7243772968558595, + "grad_norm": 2.1129229068756104, + "learning_rate": 5.52002180727818e-06, + "loss": 2.4698770523071287, + "step": 266100 + }, + { + "epoch": 0.7244317408466041, + "grad_norm": 2.0940253734588623, + "learning_rate": 5.5189314433692255e-06, + "loss": 2.4497861862182617, + "step": 266120 + }, + { + "epoch": 0.7244861848373486, + "grad_norm": 2.1867313385009766, + "learning_rate": 5.51784107946027e-06, + "loss": 2.464603805541992, + "step": 266140 + }, + { + "epoch": 0.7245406288280931, + "grad_norm": 2.2110562324523926, + "learning_rate": 5.5167507155513154e-06, + "loss": 2.4643587112426757, + "step": 266160 + }, + { + "epoch": 0.7245950728188376, + "grad_norm": 2.1399741172790527, + "learning_rate": 5.515660351642361e-06, + "loss": 2.484614944458008, + "step": 266180 + }, + { + "epoch": 0.7246495168095821, + "grad_norm": 2.3024282455444336, + "learning_rate": 5.514569987733407e-06, + "loss": 2.4768301010131837, + "step": 266200 + }, + { + "epoch": 0.7247039608003266, + "grad_norm": 2.153799533843994, + "learning_rate": 5.513479623824451e-06, + "loss": 2.479395866394043, + "step": 266220 + }, + { + "epoch": 0.7247584047910712, + "grad_norm": 2.1224117279052734, + "learning_rate": 5.512389259915497e-06, + "loss": 2.481378936767578, + "step": 266240 + }, + { + "epoch": 0.7248128487818157, + "grad_norm": 2.189159631729126, + "learning_rate": 5.511298896006543e-06, + "loss": 2.4604610443115233, + "step": 266260 + }, + { + "epoch": 0.7248672927725602, + "grad_norm": 2.2090938091278076, + "learning_rate": 5.510208532097589e-06, + "loss": 2.4810359954833983, + "step": 266280 + }, + { + "epoch": 0.7249217367633047, + "grad_norm": 2.213879346847534, + "learning_rate": 5.509118168188633e-06, + "loss": 2.4650197982788087, + "step": 266300 + }, + { + "epoch": 0.7249761807540492, + "grad_norm": 2.270742654800415, + "learning_rate": 5.5080278042796785e-06, + "loss": 2.4487995147705077, + "step": 266320 + }, + { + "epoch": 0.7250306247447937, + "grad_norm": 2.2045326232910156, + "learning_rate": 5.506937440370724e-06, + "loss": 2.433880424499512, + "step": 266340 + }, + { + "epoch": 0.7250850687355384, + "grad_norm": 2.1851398944854736, + "learning_rate": 5.50584707646177e-06, + "loss": 2.4307003021240234, + "step": 266360 + }, + { + "epoch": 0.7251395127262829, + "grad_norm": 2.1917331218719482, + "learning_rate": 5.504756712552814e-06, + "loss": 2.480401802062988, + "step": 266380 + }, + { + "epoch": 0.7251939567170274, + "grad_norm": 2.219895362854004, + "learning_rate": 5.50366634864386e-06, + "loss": 2.467325973510742, + "step": 266400 + }, + { + "epoch": 0.7252484007077719, + "grad_norm": 2.3350889682769775, + "learning_rate": 5.502575984734906e-06, + "loss": 2.477588081359863, + "step": 266420 + }, + { + "epoch": 0.7253028446985164, + "grad_norm": 2.15301513671875, + "learning_rate": 5.501485620825952e-06, + "loss": 2.4530153274536133, + "step": 266440 + }, + { + "epoch": 0.725357288689261, + "grad_norm": 2.196981430053711, + "learning_rate": 5.500395256916996e-06, + "loss": 2.4572126388549806, + "step": 266460 + }, + { + "epoch": 0.7254117326800055, + "grad_norm": 2.1102330684661865, + "learning_rate": 5.499304893008042e-06, + "loss": 2.476506996154785, + "step": 266480 + }, + { + "epoch": 0.72546617667075, + "grad_norm": 2.204447031021118, + "learning_rate": 5.498214529099087e-06, + "loss": 2.4511226654052733, + "step": 266500 + }, + { + "epoch": 0.7255206206614945, + "grad_norm": 2.397404193878174, + "learning_rate": 5.497124165190133e-06, + "loss": 2.4370689392089844, + "step": 266520 + }, + { + "epoch": 0.725575064652239, + "grad_norm": 2.1465437412261963, + "learning_rate": 5.496033801281177e-06, + "loss": 2.495853805541992, + "step": 266540 + }, + { + "epoch": 0.7256295086429835, + "grad_norm": 2.21687650680542, + "learning_rate": 5.494943437372223e-06, + "loss": 2.491193389892578, + "step": 266560 + }, + { + "epoch": 0.7256839526337281, + "grad_norm": 2.0871243476867676, + "learning_rate": 5.493853073463269e-06, + "loss": 2.4468147277832033, + "step": 266580 + }, + { + "epoch": 0.7257383966244726, + "grad_norm": 2.245666265487671, + "learning_rate": 5.492762709554315e-06, + "loss": 2.479096794128418, + "step": 266600 + }, + { + "epoch": 0.7257928406152171, + "grad_norm": 2.2041831016540527, + "learning_rate": 5.491672345645359e-06, + "loss": 2.4681825637817383, + "step": 266620 + }, + { + "epoch": 0.7258472846059616, + "grad_norm": 2.1575121879577637, + "learning_rate": 5.490581981736405e-06, + "loss": 2.444543647766113, + "step": 266640 + }, + { + "epoch": 0.7259017285967061, + "grad_norm": 2.200998306274414, + "learning_rate": 5.4894916178274505e-06, + "loss": 2.4541784286499024, + "step": 266660 + }, + { + "epoch": 0.7259561725874507, + "grad_norm": 2.1086809635162354, + "learning_rate": 5.488401253918496e-06, + "loss": 2.447449493408203, + "step": 266680 + }, + { + "epoch": 0.7260106165781952, + "grad_norm": 2.161478281021118, + "learning_rate": 5.48731089000954e-06, + "loss": 2.482228469848633, + "step": 266700 + }, + { + "epoch": 0.7260650605689397, + "grad_norm": 2.1584970951080322, + "learning_rate": 5.486220526100586e-06, + "loss": 2.453478240966797, + "step": 266720 + }, + { + "epoch": 0.7261195045596842, + "grad_norm": 2.0951292514801025, + "learning_rate": 5.485130162191632e-06, + "loss": 2.5111162185668947, + "step": 266740 + }, + { + "epoch": 0.7261739485504287, + "grad_norm": 2.2468678951263428, + "learning_rate": 5.484039798282678e-06, + "loss": 2.4604412078857423, + "step": 266760 + }, + { + "epoch": 0.7262283925411732, + "grad_norm": 2.2226402759552, + "learning_rate": 5.482949434373722e-06, + "loss": 2.4465299606323243, + "step": 266780 + }, + { + "epoch": 0.7262828365319178, + "grad_norm": 2.2945756912231445, + "learning_rate": 5.481859070464768e-06, + "loss": 2.4677923202514647, + "step": 266800 + }, + { + "epoch": 0.7263372805226623, + "grad_norm": 2.1332345008850098, + "learning_rate": 5.480768706555814e-06, + "loss": 2.4644054412841796, + "step": 266820 + }, + { + "epoch": 0.7263917245134068, + "grad_norm": 2.2609496116638184, + "learning_rate": 5.479678342646859e-06, + "loss": 2.4383012771606447, + "step": 266840 + }, + { + "epoch": 0.7264461685041513, + "grad_norm": 2.1232750415802, + "learning_rate": 5.4785879787379035e-06, + "loss": 2.4703851699829102, + "step": 266860 + }, + { + "epoch": 0.7265006124948958, + "grad_norm": 2.1333258152008057, + "learning_rate": 5.477497614828949e-06, + "loss": 2.4914213180541993, + "step": 266880 + }, + { + "epoch": 0.7265550564856404, + "grad_norm": 2.2105448246002197, + "learning_rate": 5.476407250919995e-06, + "loss": 2.505196762084961, + "step": 266900 + }, + { + "epoch": 0.726609500476385, + "grad_norm": 2.2211384773254395, + "learning_rate": 5.475316887011041e-06, + "loss": 2.4717361450195314, + "step": 266920 + }, + { + "epoch": 0.7266639444671295, + "grad_norm": 2.112563133239746, + "learning_rate": 5.474226523102085e-06, + "loss": 2.4559982299804686, + "step": 266940 + }, + { + "epoch": 0.726718388457874, + "grad_norm": 2.2213406562805176, + "learning_rate": 5.473136159193131e-06, + "loss": 2.456025314331055, + "step": 266960 + }, + { + "epoch": 0.7267728324486185, + "grad_norm": 2.0672500133514404, + "learning_rate": 5.472045795284177e-06, + "loss": 2.4601066589355467, + "step": 266980 + }, + { + "epoch": 0.726827276439363, + "grad_norm": 2.1620819568634033, + "learning_rate": 5.4709554313752225e-06, + "loss": 2.451641082763672, + "step": 267000 + }, + { + "epoch": 0.7268817204301076, + "grad_norm": 2.0983364582061768, + "learning_rate": 5.469865067466267e-06, + "loss": 2.477038764953613, + "step": 267020 + }, + { + "epoch": 0.7269361644208521, + "grad_norm": 2.5098845958709717, + "learning_rate": 5.468774703557312e-06, + "loss": 2.4626977920532225, + "step": 267040 + }, + { + "epoch": 0.7269906084115966, + "grad_norm": 2.2397587299346924, + "learning_rate": 5.467684339648358e-06, + "loss": 2.420109748840332, + "step": 267060 + }, + { + "epoch": 0.7270450524023411, + "grad_norm": 2.1293694972991943, + "learning_rate": 5.466593975739404e-06, + "loss": 2.4635288238525392, + "step": 267080 + }, + { + "epoch": 0.7270994963930856, + "grad_norm": 2.3941991329193115, + "learning_rate": 5.465503611830448e-06, + "loss": 2.4644126892089844, + "step": 267100 + }, + { + "epoch": 0.7271539403838302, + "grad_norm": 2.2151575088500977, + "learning_rate": 5.464413247921494e-06, + "loss": 2.455942916870117, + "step": 267120 + }, + { + "epoch": 0.7272083843745747, + "grad_norm": 2.3567090034484863, + "learning_rate": 5.46332288401254e-06, + "loss": 2.4423496246337892, + "step": 267140 + }, + { + "epoch": 0.7272628283653192, + "grad_norm": 2.086655616760254, + "learning_rate": 5.4622325201035856e-06, + "loss": 2.4273349761962892, + "step": 267160 + }, + { + "epoch": 0.7273172723560637, + "grad_norm": 2.1158266067504883, + "learning_rate": 5.461142156194631e-06, + "loss": 2.4609569549560546, + "step": 267180 + }, + { + "epoch": 0.7273717163468082, + "grad_norm": 2.17856764793396, + "learning_rate": 5.4600517922856755e-06, + "loss": 2.464866828918457, + "step": 267200 + }, + { + "epoch": 0.7274261603375527, + "grad_norm": 28.404836654663086, + "learning_rate": 5.458961428376721e-06, + "loss": 2.471203422546387, + "step": 267220 + }, + { + "epoch": 0.7274806043282973, + "grad_norm": 2.2626986503601074, + "learning_rate": 5.457871064467767e-06, + "loss": 2.440241813659668, + "step": 267240 + }, + { + "epoch": 0.7275350483190418, + "grad_norm": 2.1807358264923096, + "learning_rate": 5.456780700558812e-06, + "loss": 2.4593645095825196, + "step": 267260 + }, + { + "epoch": 0.7275894923097863, + "grad_norm": 2.2340288162231445, + "learning_rate": 5.455690336649857e-06, + "loss": 2.469476509094238, + "step": 267280 + }, + { + "epoch": 0.7276439363005308, + "grad_norm": 2.1026105880737305, + "learning_rate": 5.454599972740903e-06, + "loss": 2.4250078201293945, + "step": 267300 + }, + { + "epoch": 0.7276983802912753, + "grad_norm": 2.1171445846557617, + "learning_rate": 5.453509608831949e-06, + "loss": 2.4371212005615233, + "step": 267320 + }, + { + "epoch": 0.7277528242820198, + "grad_norm": 2.2419426441192627, + "learning_rate": 5.452419244922994e-06, + "loss": 2.4728363037109373, + "step": 267340 + }, + { + "epoch": 0.7278072682727644, + "grad_norm": 2.2939720153808594, + "learning_rate": 5.4513288810140386e-06, + "loss": 2.468596267700195, + "step": 267360 + }, + { + "epoch": 0.7278617122635089, + "grad_norm": 2.2763826847076416, + "learning_rate": 5.450238517105084e-06, + "loss": 2.4904645919799804, + "step": 267380 + }, + { + "epoch": 0.7279161562542534, + "grad_norm": 2.1099705696105957, + "learning_rate": 5.44914815319613e-06, + "loss": 2.4591707229614257, + "step": 267400 + }, + { + "epoch": 0.7279706002449979, + "grad_norm": 2.4606926441192627, + "learning_rate": 5.448057789287175e-06, + "loss": 2.493545150756836, + "step": 267420 + }, + { + "epoch": 0.7280250442357424, + "grad_norm": 2.9431512355804443, + "learning_rate": 5.44696742537822e-06, + "loss": 2.467279815673828, + "step": 267440 + }, + { + "epoch": 0.728079488226487, + "grad_norm": 2.188955783843994, + "learning_rate": 5.445877061469266e-06, + "loss": 2.441498947143555, + "step": 267460 + }, + { + "epoch": 0.7281339322172315, + "grad_norm": 2.0671277046203613, + "learning_rate": 5.444786697560312e-06, + "loss": 2.445049285888672, + "step": 267480 + }, + { + "epoch": 0.728188376207976, + "grad_norm": 2.095878839492798, + "learning_rate": 5.443696333651357e-06, + "loss": 2.43599910736084, + "step": 267500 + }, + { + "epoch": 0.7282428201987206, + "grad_norm": 2.26240611076355, + "learning_rate": 5.442605969742402e-06, + "loss": 2.441500663757324, + "step": 267520 + }, + { + "epoch": 0.728297264189465, + "grad_norm": 2.203169822692871, + "learning_rate": 5.4415156058334475e-06, + "loss": 2.4579034805297852, + "step": 267540 + }, + { + "epoch": 0.7283517081802096, + "grad_norm": 2.1056180000305176, + "learning_rate": 5.440425241924493e-06, + "loss": 2.4785457611083985, + "step": 267560 + }, + { + "epoch": 0.7284061521709542, + "grad_norm": 2.08835506439209, + "learning_rate": 5.439334878015538e-06, + "loss": 2.469290924072266, + "step": 267580 + }, + { + "epoch": 0.7284605961616987, + "grad_norm": 2.4799678325653076, + "learning_rate": 5.438244514106583e-06, + "loss": 2.4697141647338867, + "step": 267600 + }, + { + "epoch": 0.7285150401524432, + "grad_norm": 2.157292366027832, + "learning_rate": 5.437154150197629e-06, + "loss": 2.465434265136719, + "step": 267620 + }, + { + "epoch": 0.7285694841431877, + "grad_norm": 2.1724777221679688, + "learning_rate": 5.436063786288675e-06, + "loss": 2.451911735534668, + "step": 267640 + }, + { + "epoch": 0.7286239281339322, + "grad_norm": 2.2526161670684814, + "learning_rate": 5.43497342237972e-06, + "loss": 2.451743316650391, + "step": 267660 + }, + { + "epoch": 0.7286783721246768, + "grad_norm": 2.200054168701172, + "learning_rate": 5.433883058470765e-06, + "loss": 2.468800354003906, + "step": 267680 + }, + { + "epoch": 0.7287328161154213, + "grad_norm": 2.2090656757354736, + "learning_rate": 5.4327926945618106e-06, + "loss": 2.47922420501709, + "step": 267700 + }, + { + "epoch": 0.7287872601061658, + "grad_norm": 2.2006707191467285, + "learning_rate": 5.431702330652856e-06, + "loss": 2.4830093383789062, + "step": 267720 + }, + { + "epoch": 0.7288417040969103, + "grad_norm": 2.2090470790863037, + "learning_rate": 5.430611966743901e-06, + "loss": 2.459152412414551, + "step": 267740 + }, + { + "epoch": 0.7288961480876548, + "grad_norm": 2.1440300941467285, + "learning_rate": 5.429521602834946e-06, + "loss": 2.4888660430908205, + "step": 267760 + }, + { + "epoch": 0.7289505920783993, + "grad_norm": 2.059429883956909, + "learning_rate": 5.428431238925992e-06, + "loss": 2.4487642288208007, + "step": 267780 + }, + { + "epoch": 0.7290050360691439, + "grad_norm": 2.049694061279297, + "learning_rate": 5.427340875017038e-06, + "loss": 2.499995231628418, + "step": 267800 + }, + { + "epoch": 0.7290594800598884, + "grad_norm": 2.188323736190796, + "learning_rate": 5.426250511108083e-06, + "loss": 2.4390586853027343, + "step": 267820 + }, + { + "epoch": 0.7291139240506329, + "grad_norm": 2.193204879760742, + "learning_rate": 5.425160147199128e-06, + "loss": 2.487156295776367, + "step": 267840 + }, + { + "epoch": 0.7291683680413774, + "grad_norm": 2.074232339859009, + "learning_rate": 5.424069783290174e-06, + "loss": 2.4560882568359377, + "step": 267860 + }, + { + "epoch": 0.7292228120321219, + "grad_norm": 2.1233599185943604, + "learning_rate": 5.4229794193812194e-06, + "loss": 2.4458663940429686, + "step": 267880 + }, + { + "epoch": 0.7292772560228665, + "grad_norm": 2.5833311080932617, + "learning_rate": 5.421889055472264e-06, + "loss": 2.473356246948242, + "step": 267900 + }, + { + "epoch": 0.729331700013611, + "grad_norm": 2.2485761642456055, + "learning_rate": 5.420798691563309e-06, + "loss": 2.4671131134033204, + "step": 267920 + }, + { + "epoch": 0.7293861440043555, + "grad_norm": 2.069770574569702, + "learning_rate": 5.419708327654355e-06, + "loss": 2.460224914550781, + "step": 267940 + }, + { + "epoch": 0.7294405879951, + "grad_norm": 2.2018792629241943, + "learning_rate": 5.418617963745401e-06, + "loss": 2.487644386291504, + "step": 267960 + }, + { + "epoch": 0.7294950319858445, + "grad_norm": 2.212275981903076, + "learning_rate": 5.417527599836446e-06, + "loss": 2.457563018798828, + "step": 267980 + }, + { + "epoch": 0.729549475976589, + "grad_norm": 2.5674076080322266, + "learning_rate": 5.416437235927491e-06, + "loss": 2.4740503311157225, + "step": 268000 + }, + { + "epoch": 0.7296039199673336, + "grad_norm": 2.287363290786743, + "learning_rate": 5.415346872018537e-06, + "loss": 2.4712154388427736, + "step": 268020 + }, + { + "epoch": 0.7296583639580781, + "grad_norm": 2.1761090755462646, + "learning_rate": 5.4142565081095825e-06, + "loss": 2.457046699523926, + "step": 268040 + }, + { + "epoch": 0.7297128079488227, + "grad_norm": 2.1455860137939453, + "learning_rate": 5.4131661442006275e-06, + "loss": 2.4689184188842774, + "step": 268060 + }, + { + "epoch": 0.7297672519395672, + "grad_norm": 2.2229578495025635, + "learning_rate": 5.4120757802916725e-06, + "loss": 2.454010581970215, + "step": 268080 + }, + { + "epoch": 0.7298216959303117, + "grad_norm": 2.201991081237793, + "learning_rate": 5.410985416382718e-06, + "loss": 2.442570686340332, + "step": 268100 + }, + { + "epoch": 0.7298761399210563, + "grad_norm": 2.1609489917755127, + "learning_rate": 5.409895052473764e-06, + "loss": 2.4777153015136717, + "step": 268120 + }, + { + "epoch": 0.7299305839118008, + "grad_norm": 2.1761937141418457, + "learning_rate": 5.408804688564809e-06, + "loss": 2.463014602661133, + "step": 268140 + }, + { + "epoch": 0.7299850279025453, + "grad_norm": 2.1101937294006348, + "learning_rate": 5.407714324655854e-06, + "loss": 2.4813205718994142, + "step": 268160 + }, + { + "epoch": 0.7300394718932898, + "grad_norm": 2.2905895709991455, + "learning_rate": 5.4066239607469e-06, + "loss": 2.4688417434692385, + "step": 268180 + }, + { + "epoch": 0.7300939158840343, + "grad_norm": 2.272813081741333, + "learning_rate": 5.405533596837946e-06, + "loss": 2.4683931350708006, + "step": 268200 + }, + { + "epoch": 0.7301483598747788, + "grad_norm": 2.1941072940826416, + "learning_rate": 5.404443232928991e-06, + "loss": 2.4565313339233397, + "step": 268220 + }, + { + "epoch": 0.7302028038655234, + "grad_norm": 2.321425199508667, + "learning_rate": 5.4033528690200355e-06, + "loss": 2.446472930908203, + "step": 268240 + }, + { + "epoch": 0.7302572478562679, + "grad_norm": 2.14164137840271, + "learning_rate": 5.402262505111081e-06, + "loss": 2.4629776000976564, + "step": 268260 + }, + { + "epoch": 0.7303116918470124, + "grad_norm": 3.344590902328491, + "learning_rate": 5.401172141202127e-06, + "loss": 2.479267692565918, + "step": 268280 + }, + { + "epoch": 0.7303661358377569, + "grad_norm": 2.278965473175049, + "learning_rate": 5.400081777293172e-06, + "loss": 2.4773675918579103, + "step": 268300 + }, + { + "epoch": 0.7304205798285014, + "grad_norm": 2.1823596954345703, + "learning_rate": 5.398991413384217e-06, + "loss": 2.489244270324707, + "step": 268320 + }, + { + "epoch": 0.730475023819246, + "grad_norm": 2.209670305252075, + "learning_rate": 5.397901049475263e-06, + "loss": 2.4585922241210936, + "step": 268340 + }, + { + "epoch": 0.7305294678099905, + "grad_norm": 2.2444140911102295, + "learning_rate": 5.396810685566308e-06, + "loss": 2.4622381210327147, + "step": 268360 + }, + { + "epoch": 0.730583911800735, + "grad_norm": 2.234121561050415, + "learning_rate": 5.395720321657354e-06, + "loss": 2.455783462524414, + "step": 268380 + }, + { + "epoch": 0.7306383557914795, + "grad_norm": 2.2339792251586914, + "learning_rate": 5.394629957748399e-06, + "loss": 2.4716529846191406, + "step": 268400 + }, + { + "epoch": 0.730692799782224, + "grad_norm": 2.207888603210449, + "learning_rate": 5.3935395938394444e-06, + "loss": 2.491077995300293, + "step": 268420 + }, + { + "epoch": 0.7307472437729685, + "grad_norm": 2.137082576751709, + "learning_rate": 5.392449229930489e-06, + "loss": 2.470447540283203, + "step": 268440 + }, + { + "epoch": 0.7308016877637131, + "grad_norm": 2.2178640365600586, + "learning_rate": 5.391358866021535e-06, + "loss": 2.4827640533447264, + "step": 268460 + }, + { + "epoch": 0.7308561317544576, + "grad_norm": 2.1356146335601807, + "learning_rate": 5.39026850211258e-06, + "loss": 2.484279251098633, + "step": 268480 + }, + { + "epoch": 0.7309105757452021, + "grad_norm": 2.2276248931884766, + "learning_rate": 5.389178138203626e-06, + "loss": 2.462516212463379, + "step": 268500 + }, + { + "epoch": 0.7309650197359466, + "grad_norm": 2.2068774700164795, + "learning_rate": 5.388087774294671e-06, + "loss": 2.4771432876586914, + "step": 268520 + }, + { + "epoch": 0.7310194637266911, + "grad_norm": 2.148254871368408, + "learning_rate": 5.386997410385717e-06, + "loss": 2.477598762512207, + "step": 268540 + }, + { + "epoch": 0.7310739077174356, + "grad_norm": 2.211470603942871, + "learning_rate": 5.385907046476762e-06, + "loss": 2.499307060241699, + "step": 268560 + }, + { + "epoch": 0.7311283517081802, + "grad_norm": 2.9788825511932373, + "learning_rate": 5.3848166825678075e-06, + "loss": 2.4563488006591796, + "step": 268580 + }, + { + "epoch": 0.7311827956989247, + "grad_norm": 2.2597358226776123, + "learning_rate": 5.3837263186588525e-06, + "loss": 2.4748043060302733, + "step": 268600 + }, + { + "epoch": 0.7312372396896692, + "grad_norm": 2.1962883472442627, + "learning_rate": 5.382635954749898e-06, + "loss": 2.4137094497680662, + "step": 268620 + }, + { + "epoch": 0.7312916836804138, + "grad_norm": 2.286665678024292, + "learning_rate": 5.381545590840943e-06, + "loss": 2.4921567916870115, + "step": 268640 + }, + { + "epoch": 0.7313461276711583, + "grad_norm": 2.1403114795684814, + "learning_rate": 5.380455226931989e-06, + "loss": 2.42291202545166, + "step": 268660 + }, + { + "epoch": 0.7314005716619029, + "grad_norm": 2.302122116088867, + "learning_rate": 5.379364863023034e-06, + "loss": 2.487557029724121, + "step": 268680 + }, + { + "epoch": 0.7314550156526474, + "grad_norm": 2.214755058288574, + "learning_rate": 5.37827449911408e-06, + "loss": 2.4617753982543946, + "step": 268700 + }, + { + "epoch": 0.7315094596433919, + "grad_norm": 2.0991718769073486, + "learning_rate": 5.377184135205125e-06, + "loss": 2.462372398376465, + "step": 268720 + }, + { + "epoch": 0.7315639036341364, + "grad_norm": 2.1451961994171143, + "learning_rate": 5.376093771296171e-06, + "loss": 2.4650690078735353, + "step": 268740 + }, + { + "epoch": 0.7316183476248809, + "grad_norm": 2.3882896900177, + "learning_rate": 5.3750034073872156e-06, + "loss": 2.489111137390137, + "step": 268760 + }, + { + "epoch": 0.7316727916156254, + "grad_norm": 2.175603151321411, + "learning_rate": 5.373913043478261e-06, + "loss": 2.4493995666503907, + "step": 268780 + }, + { + "epoch": 0.73172723560637, + "grad_norm": 2.2717444896698, + "learning_rate": 5.372822679569306e-06, + "loss": 2.461879539489746, + "step": 268800 + }, + { + "epoch": 0.7317816795971145, + "grad_norm": 2.1486599445343018, + "learning_rate": 5.371732315660352e-06, + "loss": 2.470279312133789, + "step": 268820 + }, + { + "epoch": 0.731836123587859, + "grad_norm": 2.2391252517700195, + "learning_rate": 5.370641951751397e-06, + "loss": 2.4876096725463865, + "step": 268840 + }, + { + "epoch": 0.7318905675786035, + "grad_norm": 2.2433433532714844, + "learning_rate": 5.369551587842443e-06, + "loss": 2.458558464050293, + "step": 268860 + }, + { + "epoch": 0.731945011569348, + "grad_norm": 2.3893043994903564, + "learning_rate": 5.368461223933489e-06, + "loss": 2.45668888092041, + "step": 268880 + }, + { + "epoch": 0.7319994555600926, + "grad_norm": 2.106616497039795, + "learning_rate": 5.367370860024534e-06, + "loss": 2.4839599609375, + "step": 268900 + }, + { + "epoch": 0.7320538995508371, + "grad_norm": 2.2501325607299805, + "learning_rate": 5.366280496115579e-06, + "loss": 2.475172233581543, + "step": 268920 + }, + { + "epoch": 0.7321083435415816, + "grad_norm": 2.3773586750030518, + "learning_rate": 5.3651901322066245e-06, + "loss": 2.4204307556152345, + "step": 268940 + }, + { + "epoch": 0.7321627875323261, + "grad_norm": 2.1473288536071777, + "learning_rate": 5.36409976829767e-06, + "loss": 2.4575408935546874, + "step": 268960 + }, + { + "epoch": 0.7322172315230706, + "grad_norm": 2.2372865676879883, + "learning_rate": 5.363009404388715e-06, + "loss": 2.463383674621582, + "step": 268980 + }, + { + "epoch": 0.7322716755138151, + "grad_norm": 2.2512311935424805, + "learning_rate": 5.36191904047976e-06, + "loss": 2.450575256347656, + "step": 269000 + }, + { + "epoch": 0.7323261195045597, + "grad_norm": 2.104067802429199, + "learning_rate": 5.360828676570806e-06, + "loss": 2.4654741287231445, + "step": 269020 + }, + { + "epoch": 0.7323805634953042, + "grad_norm": 2.1965982913970947, + "learning_rate": 5.359738312661852e-06, + "loss": 2.469888687133789, + "step": 269040 + }, + { + "epoch": 0.7324350074860487, + "grad_norm": 2.2797813415527344, + "learning_rate": 5.358647948752897e-06, + "loss": 2.480851745605469, + "step": 269060 + }, + { + "epoch": 0.7324894514767932, + "grad_norm": 3.0857343673706055, + "learning_rate": 5.357557584843942e-06, + "loss": 2.405802536010742, + "step": 269080 + }, + { + "epoch": 0.7325438954675377, + "grad_norm": 2.1380300521850586, + "learning_rate": 5.3564672209349876e-06, + "loss": 2.446147346496582, + "step": 269100 + }, + { + "epoch": 0.7325983394582823, + "grad_norm": 2.127673864364624, + "learning_rate": 5.355376857026033e-06, + "loss": 2.4714561462402345, + "step": 269120 + }, + { + "epoch": 0.7326527834490268, + "grad_norm": 2.29109787940979, + "learning_rate": 5.354286493117078e-06, + "loss": 2.4665451049804688, + "step": 269140 + }, + { + "epoch": 0.7327072274397713, + "grad_norm": 2.306594133377075, + "learning_rate": 5.353196129208123e-06, + "loss": 2.4669322967529297, + "step": 269160 + }, + { + "epoch": 0.7327616714305158, + "grad_norm": 2.1967248916625977, + "learning_rate": 5.352105765299169e-06, + "loss": 2.464405822753906, + "step": 269180 + }, + { + "epoch": 0.7328161154212604, + "grad_norm": 2.397700309753418, + "learning_rate": 5.351015401390215e-06, + "loss": 2.4402626037597654, + "step": 269200 + }, + { + "epoch": 0.7328705594120049, + "grad_norm": 2.269468307495117, + "learning_rate": 5.34992503748126e-06, + "loss": 2.4666418075561523, + "step": 269220 + }, + { + "epoch": 0.7329250034027495, + "grad_norm": 2.155010461807251, + "learning_rate": 5.348834673572305e-06, + "loss": 2.48018798828125, + "step": 269240 + }, + { + "epoch": 0.732979447393494, + "grad_norm": 2.097353219985962, + "learning_rate": 5.347744309663351e-06, + "loss": 2.4844200134277346, + "step": 269260 + }, + { + "epoch": 0.7330338913842385, + "grad_norm": 2.182786226272583, + "learning_rate": 5.3466539457543964e-06, + "loss": 2.473532485961914, + "step": 269280 + }, + { + "epoch": 0.733088335374983, + "grad_norm": 2.2862884998321533, + "learning_rate": 5.345563581845441e-06, + "loss": 2.453318214416504, + "step": 269300 + }, + { + "epoch": 0.7331427793657275, + "grad_norm": 2.135607957839966, + "learning_rate": 5.344473217936486e-06, + "loss": 2.455883598327637, + "step": 269320 + }, + { + "epoch": 0.7331972233564721, + "grad_norm": 2.100304365158081, + "learning_rate": 5.343382854027532e-06, + "loss": 2.472870445251465, + "step": 269340 + }, + { + "epoch": 0.7332516673472166, + "grad_norm": 2.0965585708618164, + "learning_rate": 5.342292490118578e-06, + "loss": 2.4767168045043944, + "step": 269360 + }, + { + "epoch": 0.7333061113379611, + "grad_norm": 2.2040703296661377, + "learning_rate": 5.341202126209623e-06, + "loss": 2.444601631164551, + "step": 269380 + }, + { + "epoch": 0.7333605553287056, + "grad_norm": 2.212859869003296, + "learning_rate": 5.340111762300668e-06, + "loss": 2.472673797607422, + "step": 269400 + }, + { + "epoch": 0.7334149993194501, + "grad_norm": 2.274603843688965, + "learning_rate": 5.339021398391714e-06, + "loss": 2.463977241516113, + "step": 269420 + }, + { + "epoch": 0.7334694433101946, + "grad_norm": 2.250349283218384, + "learning_rate": 5.3379310344827595e-06, + "loss": 2.4695871353149412, + "step": 269440 + }, + { + "epoch": 0.7335238873009392, + "grad_norm": 2.1408848762512207, + "learning_rate": 5.336840670573804e-06, + "loss": 2.4850460052490235, + "step": 269460 + }, + { + "epoch": 0.7335783312916837, + "grad_norm": 2.115098714828491, + "learning_rate": 5.3357503066648495e-06, + "loss": 2.4836830139160155, + "step": 269480 + }, + { + "epoch": 0.7336327752824282, + "grad_norm": 2.2011351585388184, + "learning_rate": 5.334659942755895e-06, + "loss": 2.4329519271850586, + "step": 269500 + }, + { + "epoch": 0.7336872192731727, + "grad_norm": 2.114506721496582, + "learning_rate": 5.333569578846941e-06, + "loss": 2.4755300521850585, + "step": 269520 + }, + { + "epoch": 0.7337416632639172, + "grad_norm": 2.0764381885528564, + "learning_rate": 5.332479214937985e-06, + "loss": 2.470835304260254, + "step": 269540 + }, + { + "epoch": 0.7337961072546617, + "grad_norm": 2.155386209487915, + "learning_rate": 5.331388851029031e-06, + "loss": 2.4527721405029297, + "step": 269560 + }, + { + "epoch": 0.7338505512454063, + "grad_norm": 2.1580700874328613, + "learning_rate": 5.330298487120077e-06, + "loss": 2.461003875732422, + "step": 269580 + }, + { + "epoch": 0.7339049952361508, + "grad_norm": 2.199432611465454, + "learning_rate": 5.329208123211123e-06, + "loss": 2.452287292480469, + "step": 269600 + }, + { + "epoch": 0.7339594392268953, + "grad_norm": 2.2457804679870605, + "learning_rate": 5.328117759302167e-06, + "loss": 2.475724220275879, + "step": 269620 + }, + { + "epoch": 0.7340138832176398, + "grad_norm": 2.1367568969726562, + "learning_rate": 5.3270273953932125e-06, + "loss": 2.4925642013549805, + "step": 269640 + }, + { + "epoch": 0.7340683272083843, + "grad_norm": 2.2689907550811768, + "learning_rate": 5.325937031484258e-06, + "loss": 2.42697811126709, + "step": 269660 + }, + { + "epoch": 0.7341227711991289, + "grad_norm": 2.246185541152954, + "learning_rate": 5.324846667575304e-06, + "loss": 2.4528200149536135, + "step": 269680 + }, + { + "epoch": 0.7341772151898734, + "grad_norm": 2.2695627212524414, + "learning_rate": 5.323756303666348e-06, + "loss": 2.460063362121582, + "step": 269700 + }, + { + "epoch": 0.734231659180618, + "grad_norm": 2.217367172241211, + "learning_rate": 5.322665939757394e-06, + "loss": 2.4582395553588867, + "step": 269720 + }, + { + "epoch": 0.7342861031713624, + "grad_norm": 2.1605966091156006, + "learning_rate": 5.32157557584844e-06, + "loss": 2.4731157302856444, + "step": 269740 + }, + { + "epoch": 0.734340547162107, + "grad_norm": 2.1356136798858643, + "learning_rate": 5.320485211939486e-06, + "loss": 2.4519638061523437, + "step": 269760 + }, + { + "epoch": 0.7343949911528515, + "grad_norm": 2.4215049743652344, + "learning_rate": 5.31939484803053e-06, + "loss": 2.475921630859375, + "step": 269780 + }, + { + "epoch": 0.7344494351435961, + "grad_norm": 2.2009987831115723, + "learning_rate": 5.318304484121576e-06, + "loss": 2.4645198822021483, + "step": 269800 + }, + { + "epoch": 0.7345038791343406, + "grad_norm": 2.1296701431274414, + "learning_rate": 5.3172141202126214e-06, + "loss": 2.4627309799194337, + "step": 269820 + }, + { + "epoch": 0.7345583231250851, + "grad_norm": 2.1573753356933594, + "learning_rate": 5.316123756303667e-06, + "loss": 2.43763427734375, + "step": 269840 + }, + { + "epoch": 0.7346127671158296, + "grad_norm": 2.038119316101074, + "learning_rate": 5.315033392394711e-06, + "loss": 2.494472885131836, + "step": 269860 + }, + { + "epoch": 0.7346672111065741, + "grad_norm": 2.1562087535858154, + "learning_rate": 5.313943028485757e-06, + "loss": 2.4765722274780275, + "step": 269880 + }, + { + "epoch": 0.7347216550973187, + "grad_norm": 2.168764114379883, + "learning_rate": 5.312852664576803e-06, + "loss": 2.4947973251342774, + "step": 269900 + }, + { + "epoch": 0.7347760990880632, + "grad_norm": 2.4369254112243652, + "learning_rate": 5.311762300667849e-06, + "loss": 2.468943786621094, + "step": 269920 + }, + { + "epoch": 0.7348305430788077, + "grad_norm": 2.324160575866699, + "learning_rate": 5.310671936758893e-06, + "loss": 2.4516077041625977, + "step": 269940 + }, + { + "epoch": 0.7348849870695522, + "grad_norm": 2.180319309234619, + "learning_rate": 5.309581572849939e-06, + "loss": 2.456381988525391, + "step": 269960 + }, + { + "epoch": 0.7349394310602967, + "grad_norm": 2.075579881668091, + "learning_rate": 5.3084912089409845e-06, + "loss": 2.4510087966918945, + "step": 269980 + }, + { + "epoch": 0.7349938750510412, + "grad_norm": 2.139711618423462, + "learning_rate": 5.30740084503203e-06, + "loss": 2.47457218170166, + "step": 270000 + }, + { + "epoch": 0.7350483190417858, + "grad_norm": 2.1992135047912598, + "learning_rate": 5.3063104811230744e-06, + "loss": 2.474666404724121, + "step": 270020 + }, + { + "epoch": 0.7351027630325303, + "grad_norm": 2.2400400638580322, + "learning_rate": 5.30522011721412e-06, + "loss": 2.4499977111816404, + "step": 270040 + }, + { + "epoch": 0.7351572070232748, + "grad_norm": 2.1384739875793457, + "learning_rate": 5.304129753305166e-06, + "loss": 2.4582714080810546, + "step": 270060 + }, + { + "epoch": 0.7352116510140193, + "grad_norm": 2.280925989151001, + "learning_rate": 5.303039389396212e-06, + "loss": 2.463632583618164, + "step": 270080 + }, + { + "epoch": 0.7352660950047638, + "grad_norm": 2.1524875164031982, + "learning_rate": 5.301949025487256e-06, + "loss": 2.457538032531738, + "step": 270100 + }, + { + "epoch": 0.7353205389955084, + "grad_norm": 2.287384033203125, + "learning_rate": 5.300858661578302e-06, + "loss": 2.452723503112793, + "step": 270120 + }, + { + "epoch": 0.7353749829862529, + "grad_norm": 2.2501220703125, + "learning_rate": 5.299768297669348e-06, + "loss": 2.4814426422119142, + "step": 270140 + }, + { + "epoch": 0.7354294269769974, + "grad_norm": 2.2544567584991455, + "learning_rate": 5.298677933760393e-06, + "loss": 2.462166404724121, + "step": 270160 + }, + { + "epoch": 0.7354838709677419, + "grad_norm": 2.218193769454956, + "learning_rate": 5.2975875698514375e-06, + "loss": 2.4699405670166015, + "step": 270180 + }, + { + "epoch": 0.7355383149584864, + "grad_norm": 2.3099374771118164, + "learning_rate": 5.296497205942483e-06, + "loss": 2.469801139831543, + "step": 270200 + }, + { + "epoch": 0.7355927589492309, + "grad_norm": 2.2340657711029053, + "learning_rate": 5.295406842033529e-06, + "loss": 2.510662651062012, + "step": 270220 + }, + { + "epoch": 0.7356472029399755, + "grad_norm": 2.180129289627075, + "learning_rate": 5.294316478124575e-06, + "loss": 2.467178726196289, + "step": 270240 + }, + { + "epoch": 0.73570164693072, + "grad_norm": 2.1995396614074707, + "learning_rate": 5.293226114215619e-06, + "loss": 2.465924835205078, + "step": 270260 + }, + { + "epoch": 0.7357560909214645, + "grad_norm": 2.719085693359375, + "learning_rate": 5.292135750306665e-06, + "loss": 2.4701536178588865, + "step": 270280 + }, + { + "epoch": 0.735810534912209, + "grad_norm": 2.2511868476867676, + "learning_rate": 5.291045386397711e-06, + "loss": 2.4930538177490233, + "step": 270300 + }, + { + "epoch": 0.7358649789029535, + "grad_norm": 2.3187482357025146, + "learning_rate": 5.2899550224887565e-06, + "loss": 2.4413938522338867, + "step": 270320 + }, + { + "epoch": 0.7359194228936982, + "grad_norm": 2.1933071613311768, + "learning_rate": 5.288864658579801e-06, + "loss": 2.475570487976074, + "step": 270340 + }, + { + "epoch": 0.7359738668844427, + "grad_norm": 2.2070820331573486, + "learning_rate": 5.2877742946708464e-06, + "loss": 2.48094482421875, + "step": 270360 + }, + { + "epoch": 0.7360283108751872, + "grad_norm": 2.2093615531921387, + "learning_rate": 5.286683930761892e-06, + "loss": 2.479415512084961, + "step": 270380 + }, + { + "epoch": 0.7360827548659317, + "grad_norm": 2.1072683334350586, + "learning_rate": 5.285593566852938e-06, + "loss": 2.4466495513916016, + "step": 270400 + }, + { + "epoch": 0.7361371988566762, + "grad_norm": 2.2836976051330566, + "learning_rate": 5.284503202943982e-06, + "loss": 2.4879417419433594, + "step": 270420 + }, + { + "epoch": 0.7361916428474207, + "grad_norm": 2.250336170196533, + "learning_rate": 5.283412839035028e-06, + "loss": 2.4839550018310548, + "step": 270440 + }, + { + "epoch": 0.7362460868381653, + "grad_norm": 2.204437255859375, + "learning_rate": 5.282322475126074e-06, + "loss": 2.483613204956055, + "step": 270460 + }, + { + "epoch": 0.7363005308289098, + "grad_norm": 2.189387798309326, + "learning_rate": 5.28123211121712e-06, + "loss": 2.4599863052368165, + "step": 270480 + }, + { + "epoch": 0.7363549748196543, + "grad_norm": 2.1874759197235107, + "learning_rate": 5.280141747308164e-06, + "loss": 2.4530162811279297, + "step": 270500 + }, + { + "epoch": 0.7364094188103988, + "grad_norm": 2.216087579727173, + "learning_rate": 5.2790513833992095e-06, + "loss": 2.4791866302490235, + "step": 270520 + }, + { + "epoch": 0.7364638628011433, + "grad_norm": 1.9994758367538452, + "learning_rate": 5.277961019490255e-06, + "loss": 2.4438711166381837, + "step": 270540 + }, + { + "epoch": 0.7365183067918879, + "grad_norm": 2.174853801727295, + "learning_rate": 5.276870655581301e-06, + "loss": 2.5081804275512694, + "step": 270560 + }, + { + "epoch": 0.7365727507826324, + "grad_norm": 2.1600382328033447, + "learning_rate": 5.275780291672347e-06, + "loss": 2.4798784255981445, + "step": 270580 + }, + { + "epoch": 0.7366271947733769, + "grad_norm": 2.192089080810547, + "learning_rate": 5.274689927763391e-06, + "loss": 2.46124324798584, + "step": 270600 + }, + { + "epoch": 0.7366816387641214, + "grad_norm": 2.102381706237793, + "learning_rate": 5.273599563854437e-06, + "loss": 2.479006767272949, + "step": 270620 + }, + { + "epoch": 0.7367360827548659, + "grad_norm": 2.183133840560913, + "learning_rate": 5.272509199945483e-06, + "loss": 2.476656913757324, + "step": 270640 + }, + { + "epoch": 0.7367905267456104, + "grad_norm": 2.229678153991699, + "learning_rate": 5.2714188360365285e-06, + "loss": 2.469061279296875, + "step": 270660 + }, + { + "epoch": 0.736844970736355, + "grad_norm": 2.1074106693267822, + "learning_rate": 5.270328472127573e-06, + "loss": 2.4629968643188476, + "step": 270680 + }, + { + "epoch": 0.7368994147270995, + "grad_norm": 2.091398000717163, + "learning_rate": 5.269238108218618e-06, + "loss": 2.46102237701416, + "step": 270700 + }, + { + "epoch": 0.736953858717844, + "grad_norm": 2.151318311691284, + "learning_rate": 5.268147744309664e-06, + "loss": 2.4455791473388673, + "step": 270720 + }, + { + "epoch": 0.7370083027085885, + "grad_norm": 2.1828503608703613, + "learning_rate": 5.26705738040071e-06, + "loss": 2.4825765609741213, + "step": 270740 + }, + { + "epoch": 0.737062746699333, + "grad_norm": 2.161463975906372, + "learning_rate": 5.265967016491754e-06, + "loss": 2.46595458984375, + "step": 270760 + }, + { + "epoch": 0.7371171906900775, + "grad_norm": 2.307691812515259, + "learning_rate": 5.2648766525828e-06, + "loss": 2.4676982879638674, + "step": 270780 + }, + { + "epoch": 0.7371716346808221, + "grad_norm": 2.1403164863586426, + "learning_rate": 5.263786288673846e-06, + "loss": 2.4813642501831055, + "step": 270800 + }, + { + "epoch": 0.7372260786715666, + "grad_norm": 2.1807847023010254, + "learning_rate": 5.2626959247648916e-06, + "loss": 2.446313667297363, + "step": 270820 + }, + { + "epoch": 0.7372805226623111, + "grad_norm": 2.1905906200408936, + "learning_rate": 5.261605560855936e-06, + "loss": 2.4662059783935546, + "step": 270840 + }, + { + "epoch": 0.7373349666530556, + "grad_norm": 2.1430442333221436, + "learning_rate": 5.2605151969469815e-06, + "loss": 2.455078125, + "step": 270860 + }, + { + "epoch": 0.7373894106438001, + "grad_norm": 2.098170757293701, + "learning_rate": 5.259424833038027e-06, + "loss": 2.4888093948364256, + "step": 270880 + }, + { + "epoch": 0.7374438546345448, + "grad_norm": 2.2617294788360596, + "learning_rate": 5.258334469129073e-06, + "loss": 2.4939638137817384, + "step": 270900 + }, + { + "epoch": 0.7374982986252893, + "grad_norm": 2.2757046222686768, + "learning_rate": 5.257244105220117e-06, + "loss": 2.488413619995117, + "step": 270920 + }, + { + "epoch": 0.7375527426160338, + "grad_norm": 2.5682947635650635, + "learning_rate": 5.256153741311163e-06, + "loss": 2.468359375, + "step": 270940 + }, + { + "epoch": 0.7376071866067783, + "grad_norm": 2.4037811756134033, + "learning_rate": 5.255063377402209e-06, + "loss": 2.4666889190673826, + "step": 270960 + }, + { + "epoch": 0.7376616305975228, + "grad_norm": 2.2690770626068115, + "learning_rate": 5.253973013493255e-06, + "loss": 2.4665178298950194, + "step": 270980 + }, + { + "epoch": 0.7377160745882673, + "grad_norm": 2.097443103790283, + "learning_rate": 5.252882649584299e-06, + "loss": 2.4997703552246096, + "step": 271000 + }, + { + "epoch": 0.7377705185790119, + "grad_norm": 2.1246769428253174, + "learning_rate": 5.251792285675345e-06, + "loss": 2.4633310317993162, + "step": 271020 + }, + { + "epoch": 0.7378249625697564, + "grad_norm": 2.180536985397339, + "learning_rate": 5.25070192176639e-06, + "loss": 2.4282709121704102, + "step": 271040 + }, + { + "epoch": 0.7378794065605009, + "grad_norm": 2.239116668701172, + "learning_rate": 5.249611557857436e-06, + "loss": 2.470075798034668, + "step": 271060 + }, + { + "epoch": 0.7379338505512454, + "grad_norm": 2.0948426723480225, + "learning_rate": 5.24852119394848e-06, + "loss": 2.4942899703979493, + "step": 271080 + }, + { + "epoch": 0.7379882945419899, + "grad_norm": 2.115858793258667, + "learning_rate": 5.247430830039526e-06, + "loss": 2.4562917709350587, + "step": 271100 + }, + { + "epoch": 0.7380427385327345, + "grad_norm": 2.3896098136901855, + "learning_rate": 5.246340466130572e-06, + "loss": 2.4541006088256836, + "step": 271120 + }, + { + "epoch": 0.738097182523479, + "grad_norm": 2.1500403881073, + "learning_rate": 5.245250102221618e-06, + "loss": 2.480121040344238, + "step": 271140 + }, + { + "epoch": 0.7381516265142235, + "grad_norm": 2.2719764709472656, + "learning_rate": 5.244159738312662e-06, + "loss": 2.4833141326904298, + "step": 271160 + }, + { + "epoch": 0.738206070504968, + "grad_norm": 2.216062068939209, + "learning_rate": 5.243069374403708e-06, + "loss": 2.4808258056640624, + "step": 271180 + }, + { + "epoch": 0.7382605144957125, + "grad_norm": 2.158726692199707, + "learning_rate": 5.2419790104947535e-06, + "loss": 2.505080986022949, + "step": 271200 + }, + { + "epoch": 0.738314958486457, + "grad_norm": 2.176760673522949, + "learning_rate": 5.240888646585799e-06, + "loss": 2.5079322814941407, + "step": 271220 + }, + { + "epoch": 0.7383694024772016, + "grad_norm": 2.180323839187622, + "learning_rate": 5.239798282676843e-06, + "loss": 2.463527870178223, + "step": 271240 + }, + { + "epoch": 0.7384238464679461, + "grad_norm": 2.1219027042388916, + "learning_rate": 5.238707918767889e-06, + "loss": 2.455776405334473, + "step": 271260 + }, + { + "epoch": 0.7384782904586906, + "grad_norm": 2.8973028659820557, + "learning_rate": 5.237617554858935e-06, + "loss": 2.492385482788086, + "step": 271280 + }, + { + "epoch": 0.7385327344494351, + "grad_norm": 3.443429470062256, + "learning_rate": 5.236527190949981e-06, + "loss": 2.4568037033081054, + "step": 271300 + }, + { + "epoch": 0.7385871784401796, + "grad_norm": 3.0584938526153564, + "learning_rate": 5.235436827041025e-06, + "loss": 2.4591899871826173, + "step": 271320 + }, + { + "epoch": 0.7386416224309242, + "grad_norm": 2.099159002304077, + "learning_rate": 5.234346463132071e-06, + "loss": 2.4642927169799806, + "step": 271340 + }, + { + "epoch": 0.7386960664216687, + "grad_norm": 2.5302839279174805, + "learning_rate": 5.2332560992231166e-06, + "loss": 2.4590816497802734, + "step": 271360 + }, + { + "epoch": 0.7387505104124132, + "grad_norm": 2.16361141204834, + "learning_rate": 5.2321657353141615e-06, + "loss": 2.4790781021118162, + "step": 271380 + }, + { + "epoch": 0.7388049544031577, + "grad_norm": 2.126797676086426, + "learning_rate": 5.2310753714052065e-06, + "loss": 2.467802047729492, + "step": 271400 + }, + { + "epoch": 0.7388593983939022, + "grad_norm": 2.106839418411255, + "learning_rate": 5.229985007496252e-06, + "loss": 2.483515739440918, + "step": 271420 + }, + { + "epoch": 0.7389138423846467, + "grad_norm": 2.0836992263793945, + "learning_rate": 5.228894643587298e-06, + "loss": 2.4781782150268556, + "step": 271440 + }, + { + "epoch": 0.7389682863753914, + "grad_norm": 2.1482043266296387, + "learning_rate": 5.227804279678343e-06, + "loss": 2.5050586700439452, + "step": 271460 + }, + { + "epoch": 0.7390227303661359, + "grad_norm": 2.1598236560821533, + "learning_rate": 5.226713915769388e-06, + "loss": 2.4708572387695313, + "step": 271480 + }, + { + "epoch": 0.7390771743568804, + "grad_norm": 2.2254445552825928, + "learning_rate": 5.225623551860434e-06, + "loss": 2.465336799621582, + "step": 271500 + }, + { + "epoch": 0.7391316183476249, + "grad_norm": 2.1274125576019287, + "learning_rate": 5.22453318795148e-06, + "loss": 2.4927671432495115, + "step": 271520 + }, + { + "epoch": 0.7391860623383694, + "grad_norm": 2.133759021759033, + "learning_rate": 5.223442824042525e-06, + "loss": 2.4606817245483397, + "step": 271540 + }, + { + "epoch": 0.739240506329114, + "grad_norm": 2.220402717590332, + "learning_rate": 5.2223524601335696e-06, + "loss": 2.4547210693359376, + "step": 271560 + }, + { + "epoch": 0.7392949503198585, + "grad_norm": 2.221613645553589, + "learning_rate": 5.221262096224615e-06, + "loss": 2.4769989013671876, + "step": 271580 + }, + { + "epoch": 0.739349394310603, + "grad_norm": 2.664520740509033, + "learning_rate": 5.220171732315661e-06, + "loss": 2.4639692306518555, + "step": 271600 + }, + { + "epoch": 0.7394038383013475, + "grad_norm": 2.2107925415039062, + "learning_rate": 5.219081368406706e-06, + "loss": 2.4672243118286135, + "step": 271620 + }, + { + "epoch": 0.739458282292092, + "grad_norm": 2.1094608306884766, + "learning_rate": 5.217991004497751e-06, + "loss": 2.4958673477172852, + "step": 271640 + }, + { + "epoch": 0.7395127262828365, + "grad_norm": 2.231109619140625, + "learning_rate": 5.216900640588797e-06, + "loss": 2.4485029220581054, + "step": 271660 + }, + { + "epoch": 0.7395671702735811, + "grad_norm": 2.202028751373291, + "learning_rate": 5.215810276679843e-06, + "loss": 2.4780851364135743, + "step": 271680 + }, + { + "epoch": 0.7396216142643256, + "grad_norm": 2.121579170227051, + "learning_rate": 5.214719912770888e-06, + "loss": 2.472347450256348, + "step": 271700 + }, + { + "epoch": 0.7396760582550701, + "grad_norm": 2.1660776138305664, + "learning_rate": 5.213629548861933e-06, + "loss": 2.4744518280029295, + "step": 271720 + }, + { + "epoch": 0.7397305022458146, + "grad_norm": 2.035782814025879, + "learning_rate": 5.2125391849529785e-06, + "loss": 2.4473018646240234, + "step": 271740 + }, + { + "epoch": 0.7397849462365591, + "grad_norm": 2.1536667346954346, + "learning_rate": 5.211448821044024e-06, + "loss": 2.481658172607422, + "step": 271760 + }, + { + "epoch": 0.7398393902273036, + "grad_norm": 2.252565622329712, + "learning_rate": 5.210358457135069e-06, + "loss": 2.4656044006347657, + "step": 271780 + }, + { + "epoch": 0.7398938342180482, + "grad_norm": 2.227349281311035, + "learning_rate": 5.209268093226114e-06, + "loss": 2.488023567199707, + "step": 271800 + }, + { + "epoch": 0.7399482782087927, + "grad_norm": 2.149158239364624, + "learning_rate": 5.20817772931716e-06, + "loss": 2.4324317932128907, + "step": 271820 + }, + { + "epoch": 0.7400027221995372, + "grad_norm": 2.1932873725891113, + "learning_rate": 5.207087365408206e-06, + "loss": 2.4426977157592775, + "step": 271840 + }, + { + "epoch": 0.7400571661902817, + "grad_norm": 2.176774263381958, + "learning_rate": 5.205997001499251e-06, + "loss": 2.482851982116699, + "step": 271860 + }, + { + "epoch": 0.7401116101810262, + "grad_norm": 2.1792759895324707, + "learning_rate": 5.204906637590296e-06, + "loss": 2.461948013305664, + "step": 271880 + }, + { + "epoch": 0.7401660541717708, + "grad_norm": 2.386484146118164, + "learning_rate": 5.2038162736813416e-06, + "loss": 2.4841592788696287, + "step": 271900 + }, + { + "epoch": 0.7402204981625153, + "grad_norm": 2.427123546600342, + "learning_rate": 5.202725909772387e-06, + "loss": 2.5009042739868166, + "step": 271920 + }, + { + "epoch": 0.7402749421532598, + "grad_norm": 2.2267801761627197, + "learning_rate": 5.201635545863432e-06, + "loss": 2.4498868942260743, + "step": 271940 + }, + { + "epoch": 0.7403293861440043, + "grad_norm": 2.191328287124634, + "learning_rate": 5.200545181954477e-06, + "loss": 2.462368392944336, + "step": 271960 + }, + { + "epoch": 0.7403838301347488, + "grad_norm": 2.187793016433716, + "learning_rate": 5.199454818045523e-06, + "loss": 2.4400888442993165, + "step": 271980 + }, + { + "epoch": 0.7404382741254933, + "grad_norm": 2.064976215362549, + "learning_rate": 5.198364454136569e-06, + "loss": 2.4591087341308593, + "step": 272000 + }, + { + "epoch": 0.740492718116238, + "grad_norm": 2.2782578468322754, + "learning_rate": 5.197274090227614e-06, + "loss": 2.430394744873047, + "step": 272020 + }, + { + "epoch": 0.7405471621069825, + "grad_norm": 2.1525607109069824, + "learning_rate": 5.196183726318659e-06, + "loss": 2.4835453033447266, + "step": 272040 + }, + { + "epoch": 0.740601606097727, + "grad_norm": 2.8771519660949707, + "learning_rate": 5.195093362409705e-06, + "loss": 2.473150634765625, + "step": 272060 + }, + { + "epoch": 0.7406560500884715, + "grad_norm": 2.180631399154663, + "learning_rate": 5.1940029985007504e-06, + "loss": 2.4356529235839846, + "step": 272080 + }, + { + "epoch": 0.740710494079216, + "grad_norm": 2.16524338722229, + "learning_rate": 5.192912634591795e-06, + "loss": 2.488675117492676, + "step": 272100 + }, + { + "epoch": 0.7407649380699606, + "grad_norm": 2.1184656620025635, + "learning_rate": 5.19182227068284e-06, + "loss": 2.4448387145996096, + "step": 272120 + }, + { + "epoch": 0.7408193820607051, + "grad_norm": 2.1282382011413574, + "learning_rate": 5.190731906773886e-06, + "loss": 2.4745784759521485, + "step": 272140 + }, + { + "epoch": 0.7408738260514496, + "grad_norm": 2.3898820877075195, + "learning_rate": 5.189641542864932e-06, + "loss": 2.4485200881958007, + "step": 272160 + }, + { + "epoch": 0.7409282700421941, + "grad_norm": 2.142023801803589, + "learning_rate": 5.188551178955977e-06, + "loss": 2.443362808227539, + "step": 272180 + }, + { + "epoch": 0.7409827140329386, + "grad_norm": 2.1493947505950928, + "learning_rate": 5.187460815047022e-06, + "loss": 2.4549819946289064, + "step": 272200 + }, + { + "epoch": 0.7410371580236831, + "grad_norm": 2.146165370941162, + "learning_rate": 5.186370451138068e-06, + "loss": 2.4874191284179688, + "step": 272220 + }, + { + "epoch": 0.7410916020144277, + "grad_norm": 2.154920816421509, + "learning_rate": 5.1852800872291135e-06, + "loss": 2.460762596130371, + "step": 272240 + }, + { + "epoch": 0.7411460460051722, + "grad_norm": 2.2275540828704834, + "learning_rate": 5.1841897233201585e-06, + "loss": 2.4544960021972657, + "step": 272260 + }, + { + "epoch": 0.7412004899959167, + "grad_norm": 2.075645923614502, + "learning_rate": 5.183099359411204e-06, + "loss": 2.4672542572021485, + "step": 272280 + }, + { + "epoch": 0.7412549339866612, + "grad_norm": 2.174440622329712, + "learning_rate": 5.182008995502249e-06, + "loss": 2.482212448120117, + "step": 272300 + }, + { + "epoch": 0.7413093779774057, + "grad_norm": 2.1027204990386963, + "learning_rate": 5.180918631593295e-06, + "loss": 2.500848579406738, + "step": 272320 + }, + { + "epoch": 0.7413638219681503, + "grad_norm": 2.3034005165100098, + "learning_rate": 5.17982826768434e-06, + "loss": 2.4829904556274416, + "step": 272340 + }, + { + "epoch": 0.7414182659588948, + "grad_norm": 7.21920108795166, + "learning_rate": 5.178737903775386e-06, + "loss": 2.455769157409668, + "step": 272360 + }, + { + "epoch": 0.7414727099496393, + "grad_norm": 2.1678390502929688, + "learning_rate": 5.177647539866431e-06, + "loss": 2.4652477264404298, + "step": 272380 + }, + { + "epoch": 0.7415271539403838, + "grad_norm": 2.1571640968322754, + "learning_rate": 5.176557175957477e-06, + "loss": 2.473745918273926, + "step": 272400 + }, + { + "epoch": 0.7415815979311283, + "grad_norm": 2.094435453414917, + "learning_rate": 5.175466812048522e-06, + "loss": 2.457788848876953, + "step": 272420 + }, + { + "epoch": 0.7416360419218728, + "grad_norm": 2.1064326763153076, + "learning_rate": 5.174376448139567e-06, + "loss": 2.4629728317260744, + "step": 272440 + }, + { + "epoch": 0.7416904859126174, + "grad_norm": 2.220360517501831, + "learning_rate": 5.173286084230612e-06, + "loss": 2.4533370971679687, + "step": 272460 + }, + { + "epoch": 0.7417449299033619, + "grad_norm": 2.1740880012512207, + "learning_rate": 5.172195720321657e-06, + "loss": 2.462664031982422, + "step": 272480 + }, + { + "epoch": 0.7417993738941064, + "grad_norm": 2.09208607673645, + "learning_rate": 5.171105356412703e-06, + "loss": 2.472904586791992, + "step": 272500 + }, + { + "epoch": 0.7418538178848509, + "grad_norm": 2.2836949825286865, + "learning_rate": 5.170014992503749e-06, + "loss": 2.4832345962524416, + "step": 272520 + }, + { + "epoch": 0.7419082618755954, + "grad_norm": 2.1741549968719482, + "learning_rate": 5.168924628594794e-06, + "loss": 2.4832485198974608, + "step": 272540 + }, + { + "epoch": 0.74196270586634, + "grad_norm": 2.473696231842041, + "learning_rate": 5.167834264685839e-06, + "loss": 2.4645206451416017, + "step": 272560 + }, + { + "epoch": 0.7420171498570846, + "grad_norm": 2.096066474914551, + "learning_rate": 5.166743900776885e-06, + "loss": 2.453407859802246, + "step": 272580 + }, + { + "epoch": 0.7420715938478291, + "grad_norm": 2.591026544570923, + "learning_rate": 5.1656535368679305e-06, + "loss": 2.436084747314453, + "step": 272600 + }, + { + "epoch": 0.7421260378385736, + "grad_norm": 2.1125338077545166, + "learning_rate": 5.1645631729589754e-06, + "loss": 2.42541618347168, + "step": 272620 + }, + { + "epoch": 0.7421804818293181, + "grad_norm": 2.1456658840179443, + "learning_rate": 5.16347280905002e-06, + "loss": 2.4321975708007812, + "step": 272640 + }, + { + "epoch": 0.7422349258200626, + "grad_norm": 2.344741106033325, + "learning_rate": 5.162382445141066e-06, + "loss": 2.4447763442993162, + "step": 272660 + }, + { + "epoch": 0.7422893698108072, + "grad_norm": 2.283820867538452, + "learning_rate": 5.161292081232112e-06, + "loss": 2.491059494018555, + "step": 272680 + }, + { + "epoch": 0.7423438138015517, + "grad_norm": 2.1044936180114746, + "learning_rate": 5.160201717323157e-06, + "loss": 2.4795183181762694, + "step": 272700 + }, + { + "epoch": 0.7423982577922962, + "grad_norm": 2.107999801635742, + "learning_rate": 5.159111353414202e-06, + "loss": 2.455364990234375, + "step": 272720 + }, + { + "epoch": 0.7424527017830407, + "grad_norm": 3.8108022212982178, + "learning_rate": 5.158020989505248e-06, + "loss": 2.4328514099121095, + "step": 272740 + }, + { + "epoch": 0.7425071457737852, + "grad_norm": 2.104959487915039, + "learning_rate": 5.1569306255962936e-06, + "loss": 2.450847053527832, + "step": 272760 + }, + { + "epoch": 0.7425615897645298, + "grad_norm": 2.393596887588501, + "learning_rate": 5.1558402616873385e-06, + "loss": 2.475904655456543, + "step": 272780 + }, + { + "epoch": 0.7426160337552743, + "grad_norm": 2.1979222297668457, + "learning_rate": 5.1547498977783835e-06, + "loss": 2.4546985626220703, + "step": 272800 + }, + { + "epoch": 0.7426704777460188, + "grad_norm": 2.236992835998535, + "learning_rate": 5.153659533869429e-06, + "loss": 2.4827398300170898, + "step": 272820 + }, + { + "epoch": 0.7427249217367633, + "grad_norm": 2.3841302394866943, + "learning_rate": 5.152569169960475e-06, + "loss": 2.481201171875, + "step": 272840 + }, + { + "epoch": 0.7427793657275078, + "grad_norm": 2.130160331726074, + "learning_rate": 5.15147880605152e-06, + "loss": 2.464588737487793, + "step": 272860 + }, + { + "epoch": 0.7428338097182523, + "grad_norm": 2.1356897354125977, + "learning_rate": 5.150388442142565e-06, + "loss": 2.4580928802490236, + "step": 272880 + }, + { + "epoch": 0.7428882537089969, + "grad_norm": 2.083127021789551, + "learning_rate": 5.149298078233611e-06, + "loss": 2.465883255004883, + "step": 272900 + }, + { + "epoch": 0.7429426976997414, + "grad_norm": 2.132514238357544, + "learning_rate": 5.148207714324657e-06, + "loss": 2.5001955032348633, + "step": 272920 + }, + { + "epoch": 0.7429971416904859, + "grad_norm": 2.248494863510132, + "learning_rate": 5.147117350415702e-06, + "loss": 2.478958320617676, + "step": 272940 + }, + { + "epoch": 0.7430515856812304, + "grad_norm": 2.094892740249634, + "learning_rate": 5.1460269865067466e-06, + "loss": 2.471664810180664, + "step": 272960 + }, + { + "epoch": 0.7431060296719749, + "grad_norm": 2.251281499862671, + "learning_rate": 5.144936622597792e-06, + "loss": 2.4835229873657227, + "step": 272980 + }, + { + "epoch": 0.7431604736627194, + "grad_norm": 2.2147817611694336, + "learning_rate": 5.143846258688838e-06, + "loss": 2.4727317810058596, + "step": 273000 + }, + { + "epoch": 0.743214917653464, + "grad_norm": 2.2243144512176514, + "learning_rate": 5.142755894779883e-06, + "loss": 2.450891304016113, + "step": 273020 + }, + { + "epoch": 0.7432693616442085, + "grad_norm": 2.2616755962371826, + "learning_rate": 5.141665530870928e-06, + "loss": 2.4650348663330077, + "step": 273040 + }, + { + "epoch": 0.743323805634953, + "grad_norm": 2.2523257732391357, + "learning_rate": 5.140575166961974e-06, + "loss": 2.4818376541137694, + "step": 273060 + }, + { + "epoch": 0.7433782496256975, + "grad_norm": 2.2612485885620117, + "learning_rate": 5.13948480305302e-06, + "loss": 2.4818294525146483, + "step": 273080 + }, + { + "epoch": 0.743432693616442, + "grad_norm": 2.1930150985717773, + "learning_rate": 5.138394439144065e-06, + "loss": 2.4616193771362305, + "step": 273100 + }, + { + "epoch": 0.7434871376071867, + "grad_norm": 2.1506991386413574, + "learning_rate": 5.13730407523511e-06, + "loss": 2.471293258666992, + "step": 273120 + }, + { + "epoch": 0.7435415815979312, + "grad_norm": 2.0686168670654297, + "learning_rate": 5.1362137113261555e-06, + "loss": 2.4249523162841795, + "step": 273140 + }, + { + "epoch": 0.7435960255886757, + "grad_norm": 2.146376132965088, + "learning_rate": 5.135123347417201e-06, + "loss": 2.4420169830322265, + "step": 273160 + }, + { + "epoch": 0.7436504695794202, + "grad_norm": 2.425346851348877, + "learning_rate": 5.134032983508246e-06, + "loss": 2.454888343811035, + "step": 273180 + }, + { + "epoch": 0.7437049135701647, + "grad_norm": 2.096754550933838, + "learning_rate": 5.132942619599291e-06, + "loss": 2.4643558502197265, + "step": 273200 + }, + { + "epoch": 0.7437593575609092, + "grad_norm": 2.2840349674224854, + "learning_rate": 5.131852255690337e-06, + "loss": 2.477754592895508, + "step": 273220 + }, + { + "epoch": 0.7438138015516538, + "grad_norm": 2.1349599361419678, + "learning_rate": 5.130761891781383e-06, + "loss": 2.456025314331055, + "step": 273240 + }, + { + "epoch": 0.7438682455423983, + "grad_norm": 2.200472831726074, + "learning_rate": 5.129671527872428e-06, + "loss": 2.49227352142334, + "step": 273260 + }, + { + "epoch": 0.7439226895331428, + "grad_norm": 2.225447654724121, + "learning_rate": 5.128581163963473e-06, + "loss": 2.488237190246582, + "step": 273280 + }, + { + "epoch": 0.7439771335238873, + "grad_norm": 2.17026948928833, + "learning_rate": 5.1274908000545185e-06, + "loss": 2.476335334777832, + "step": 273300 + }, + { + "epoch": 0.7440315775146318, + "grad_norm": 2.1667704582214355, + "learning_rate": 5.126400436145564e-06, + "loss": 2.4557270050048827, + "step": 273320 + }, + { + "epoch": 0.7440860215053764, + "grad_norm": 2.138699769973755, + "learning_rate": 5.125310072236609e-06, + "loss": 2.465767478942871, + "step": 273340 + }, + { + "epoch": 0.7441404654961209, + "grad_norm": 2.237314224243164, + "learning_rate": 5.124219708327654e-06, + "loss": 2.4711286544799806, + "step": 273360 + }, + { + "epoch": 0.7441949094868654, + "grad_norm": 2.084148406982422, + "learning_rate": 5.1231293444187e-06, + "loss": 2.492319679260254, + "step": 273380 + }, + { + "epoch": 0.7442493534776099, + "grad_norm": 2.1522533893585205, + "learning_rate": 5.122038980509746e-06, + "loss": 2.447119140625, + "step": 273400 + }, + { + "epoch": 0.7443037974683544, + "grad_norm": 2.168116331100464, + "learning_rate": 5.120948616600791e-06, + "loss": 2.4909341812133787, + "step": 273420 + }, + { + "epoch": 0.7443582414590989, + "grad_norm": 2.283745527267456, + "learning_rate": 5.119858252691836e-06, + "loss": 2.467759895324707, + "step": 273440 + }, + { + "epoch": 0.7444126854498435, + "grad_norm": 2.2050559520721436, + "learning_rate": 5.118767888782882e-06, + "loss": 2.4677497863769533, + "step": 273460 + }, + { + "epoch": 0.744467129440588, + "grad_norm": 2.3656609058380127, + "learning_rate": 5.1176775248739274e-06, + "loss": 2.4814239501953126, + "step": 273480 + }, + { + "epoch": 0.7445215734313325, + "grad_norm": 2.0961878299713135, + "learning_rate": 5.116587160964972e-06, + "loss": 2.4561954498291017, + "step": 273500 + }, + { + "epoch": 0.744576017422077, + "grad_norm": 2.2870240211486816, + "learning_rate": 5.115496797056017e-06, + "loss": 2.4678436279296876, + "step": 273520 + }, + { + "epoch": 0.7446304614128215, + "grad_norm": 2.1921112537384033, + "learning_rate": 5.114406433147063e-06, + "loss": 2.488826942443848, + "step": 273540 + }, + { + "epoch": 0.7446849054035661, + "grad_norm": 2.1464271545410156, + "learning_rate": 5.113316069238109e-06, + "loss": 2.465192985534668, + "step": 273560 + }, + { + "epoch": 0.7447393493943106, + "grad_norm": 2.134878396987915, + "learning_rate": 5.112225705329153e-06, + "loss": 2.4575321197509767, + "step": 273580 + }, + { + "epoch": 0.7447937933850551, + "grad_norm": 2.3012611865997314, + "learning_rate": 5.111135341420199e-06, + "loss": 2.4714275360107423, + "step": 273600 + }, + { + "epoch": 0.7448482373757996, + "grad_norm": 2.253641366958618, + "learning_rate": 5.110044977511245e-06, + "loss": 2.4624677658081056, + "step": 273620 + }, + { + "epoch": 0.7449026813665441, + "grad_norm": 2.0799715518951416, + "learning_rate": 5.1089546136022905e-06, + "loss": 2.4932355880737305, + "step": 273640 + }, + { + "epoch": 0.7449571253572886, + "grad_norm": 2.165292739868164, + "learning_rate": 5.107864249693335e-06, + "loss": 2.473873329162598, + "step": 273660 + }, + { + "epoch": 0.7450115693480333, + "grad_norm": 2.1958940029144287, + "learning_rate": 5.1067738857843805e-06, + "loss": 2.470261573791504, + "step": 273680 + }, + { + "epoch": 0.7450660133387778, + "grad_norm": 2.189093589782715, + "learning_rate": 5.105683521875426e-06, + "loss": 2.4389524459838867, + "step": 273700 + }, + { + "epoch": 0.7451204573295223, + "grad_norm": 2.309645891189575, + "learning_rate": 5.104593157966472e-06, + "loss": 2.4202348709106447, + "step": 273720 + }, + { + "epoch": 0.7451749013202668, + "grad_norm": 2.2835640907287598, + "learning_rate": 5.103502794057516e-06, + "loss": 2.4303817749023438, + "step": 273740 + }, + { + "epoch": 0.7452293453110113, + "grad_norm": 2.2653443813323975, + "learning_rate": 5.102412430148562e-06, + "loss": 2.4675575256347657, + "step": 273760 + }, + { + "epoch": 0.7452837893017559, + "grad_norm": 2.3607850074768066, + "learning_rate": 5.101322066239608e-06, + "loss": 2.4610382080078126, + "step": 273780 + }, + { + "epoch": 0.7453382332925004, + "grad_norm": 2.133476734161377, + "learning_rate": 5.100231702330654e-06, + "loss": 2.455575942993164, + "step": 273800 + }, + { + "epoch": 0.7453926772832449, + "grad_norm": 2.3134918212890625, + "learning_rate": 5.099141338421698e-06, + "loss": 2.471114921569824, + "step": 273820 + }, + { + "epoch": 0.7454471212739894, + "grad_norm": 2.162212371826172, + "learning_rate": 5.0980509745127435e-06, + "loss": 2.4723257064819335, + "step": 273840 + }, + { + "epoch": 0.7455015652647339, + "grad_norm": 2.159773111343384, + "learning_rate": 5.096960610603789e-06, + "loss": 2.445766067504883, + "step": 273860 + }, + { + "epoch": 0.7455560092554784, + "grad_norm": 2.288069248199463, + "learning_rate": 5.095870246694835e-06, + "loss": 2.478403854370117, + "step": 273880 + }, + { + "epoch": 0.745610453246223, + "grad_norm": 2.0933969020843506, + "learning_rate": 5.094779882785879e-06, + "loss": 2.4586191177368164, + "step": 273900 + }, + { + "epoch": 0.7456648972369675, + "grad_norm": 2.1865057945251465, + "learning_rate": 5.093689518876925e-06, + "loss": 2.488653373718262, + "step": 273920 + }, + { + "epoch": 0.745719341227712, + "grad_norm": 2.651289701461792, + "learning_rate": 5.092599154967971e-06, + "loss": 2.485463333129883, + "step": 273940 + }, + { + "epoch": 0.7457737852184565, + "grad_norm": 2.1637678146362305, + "learning_rate": 5.091508791059017e-06, + "loss": 2.478460502624512, + "step": 273960 + }, + { + "epoch": 0.745828229209201, + "grad_norm": 2.1210896968841553, + "learning_rate": 5.0904184271500625e-06, + "loss": 2.433747482299805, + "step": 273980 + }, + { + "epoch": 0.7458826731999455, + "grad_norm": 2.200303792953491, + "learning_rate": 5.089328063241107e-06, + "loss": 2.5017051696777344, + "step": 274000 + }, + { + "epoch": 0.7459371171906901, + "grad_norm": 2.1911535263061523, + "learning_rate": 5.0882376993321524e-06, + "loss": 2.495553970336914, + "step": 274020 + }, + { + "epoch": 0.7459915611814346, + "grad_norm": 2.308957099914551, + "learning_rate": 5.087147335423198e-06, + "loss": 2.429429626464844, + "step": 274040 + }, + { + "epoch": 0.7460460051721791, + "grad_norm": 2.2195093631744385, + "learning_rate": 5.086056971514244e-06, + "loss": 2.4674692153930664, + "step": 274060 + }, + { + "epoch": 0.7461004491629236, + "grad_norm": 2.141336679458618, + "learning_rate": 5.084966607605288e-06, + "loss": 2.47595157623291, + "step": 274080 + }, + { + "epoch": 0.7461548931536681, + "grad_norm": 2.3319716453552246, + "learning_rate": 5.083876243696334e-06, + "loss": 2.4717784881591798, + "step": 274100 + }, + { + "epoch": 0.7462093371444127, + "grad_norm": 2.1204051971435547, + "learning_rate": 5.08278587978738e-06, + "loss": 2.5018720626831055, + "step": 274120 + }, + { + "epoch": 0.7462637811351572, + "grad_norm": 2.153726100921631, + "learning_rate": 5.081695515878426e-06, + "loss": 2.4768299102783202, + "step": 274140 + }, + { + "epoch": 0.7463182251259017, + "grad_norm": 2.1399731636047363, + "learning_rate": 5.08060515196947e-06, + "loss": 2.4806798934936523, + "step": 274160 + }, + { + "epoch": 0.7463726691166462, + "grad_norm": 2.185981273651123, + "learning_rate": 5.0795147880605155e-06, + "loss": 2.456767272949219, + "step": 274180 + }, + { + "epoch": 0.7464271131073907, + "grad_norm": 2.116811990737915, + "learning_rate": 5.078424424151561e-06, + "loss": 2.474702835083008, + "step": 274200 + }, + { + "epoch": 0.7464815570981352, + "grad_norm": 2.25966215133667, + "learning_rate": 5.077334060242607e-06, + "loss": 2.449764633178711, + "step": 274220 + }, + { + "epoch": 0.7465360010888799, + "grad_norm": 2.26845121383667, + "learning_rate": 5.076243696333651e-06, + "loss": 2.433452606201172, + "step": 274240 + }, + { + "epoch": 0.7465904450796244, + "grad_norm": 2.1932878494262695, + "learning_rate": 5.075153332424697e-06, + "loss": 2.4652055740356444, + "step": 274260 + }, + { + "epoch": 0.7466448890703689, + "grad_norm": 2.0873048305511475, + "learning_rate": 5.074062968515743e-06, + "loss": 2.45489616394043, + "step": 274280 + }, + { + "epoch": 0.7466993330611134, + "grad_norm": 2.191793203353882, + "learning_rate": 5.072972604606789e-06, + "loss": 2.4553903579711913, + "step": 274300 + }, + { + "epoch": 0.7467537770518579, + "grad_norm": 2.1102559566497803, + "learning_rate": 5.071882240697833e-06, + "loss": 2.4787113189697267, + "step": 274320 + }, + { + "epoch": 0.7468082210426025, + "grad_norm": 2.258791923522949, + "learning_rate": 5.070791876788879e-06, + "loss": 2.5038713455200194, + "step": 274340 + }, + { + "epoch": 0.746862665033347, + "grad_norm": 2.1625137329101562, + "learning_rate": 5.069701512879924e-06, + "loss": 2.4527578353881836, + "step": 274360 + }, + { + "epoch": 0.7469171090240915, + "grad_norm": 2.1102983951568604, + "learning_rate": 5.06861114897097e-06, + "loss": 2.4708925247192384, + "step": 274380 + }, + { + "epoch": 0.746971553014836, + "grad_norm": 2.1323366165161133, + "learning_rate": 5.067520785062014e-06, + "loss": 2.426454544067383, + "step": 274400 + }, + { + "epoch": 0.7470259970055805, + "grad_norm": 2.182142972946167, + "learning_rate": 5.06643042115306e-06, + "loss": 2.480246162414551, + "step": 274420 + }, + { + "epoch": 0.747080440996325, + "grad_norm": 2.2285614013671875, + "learning_rate": 5.065340057244106e-06, + "loss": 2.4904890060424805, + "step": 274440 + }, + { + "epoch": 0.7471348849870696, + "grad_norm": 2.1626532077789307, + "learning_rate": 5.064249693335152e-06, + "loss": 2.4334285736083983, + "step": 274460 + }, + { + "epoch": 0.7471893289778141, + "grad_norm": 2.5357253551483154, + "learning_rate": 5.063159329426196e-06, + "loss": 2.5121503829956056, + "step": 274480 + }, + { + "epoch": 0.7472437729685586, + "grad_norm": 2.2218449115753174, + "learning_rate": 5.062068965517242e-06, + "loss": 2.4821327209472654, + "step": 274500 + }, + { + "epoch": 0.7472982169593031, + "grad_norm": 2.187549114227295, + "learning_rate": 5.0609786016082875e-06, + "loss": 2.4728471755981447, + "step": 274520 + }, + { + "epoch": 0.7473526609500476, + "grad_norm": 2.1667847633361816, + "learning_rate": 5.059888237699333e-06, + "loss": 2.475331115722656, + "step": 274540 + }, + { + "epoch": 0.7474071049407922, + "grad_norm": 2.085934638977051, + "learning_rate": 5.0587978737903774e-06, + "loss": 2.4664794921875, + "step": 274560 + }, + { + "epoch": 0.7474615489315367, + "grad_norm": 2.2110819816589355, + "learning_rate": 5.057707509881423e-06, + "loss": 2.4468183517456055, + "step": 274580 + }, + { + "epoch": 0.7475159929222812, + "grad_norm": 2.1494123935699463, + "learning_rate": 5.056617145972469e-06, + "loss": 2.4680362701416017, + "step": 274600 + }, + { + "epoch": 0.7475704369130257, + "grad_norm": 2.0818519592285156, + "learning_rate": 5.055526782063515e-06, + "loss": 2.4827632904052734, + "step": 274620 + }, + { + "epoch": 0.7476248809037702, + "grad_norm": 2.1457855701446533, + "learning_rate": 5.054436418154559e-06, + "loss": 2.4554725646972657, + "step": 274640 + }, + { + "epoch": 0.7476793248945147, + "grad_norm": 2.208251714706421, + "learning_rate": 5.053346054245605e-06, + "loss": 2.4490425109863283, + "step": 274660 + }, + { + "epoch": 0.7477337688852593, + "grad_norm": 2.258096218109131, + "learning_rate": 5.052255690336651e-06, + "loss": 2.4689281463623045, + "step": 274680 + }, + { + "epoch": 0.7477882128760038, + "grad_norm": 2.2523436546325684, + "learning_rate": 5.051165326427696e-06, + "loss": 2.447797393798828, + "step": 274700 + }, + { + "epoch": 0.7478426568667483, + "grad_norm": 2.0918402671813965, + "learning_rate": 5.0500749625187405e-06, + "loss": 2.460179328918457, + "step": 274720 + }, + { + "epoch": 0.7478971008574928, + "grad_norm": 2.208775758743286, + "learning_rate": 5.048984598609786e-06, + "loss": 2.451076889038086, + "step": 274740 + }, + { + "epoch": 0.7479515448482373, + "grad_norm": 2.141573905944824, + "learning_rate": 5.047894234700832e-06, + "loss": 2.4833656311035157, + "step": 274760 + }, + { + "epoch": 0.748005988838982, + "grad_norm": 2.1759345531463623, + "learning_rate": 5.046803870791878e-06, + "loss": 2.427041244506836, + "step": 274780 + }, + { + "epoch": 0.7480604328297265, + "grad_norm": 3.0529751777648926, + "learning_rate": 5.045713506882922e-06, + "loss": 2.4668092727661133, + "step": 274800 + }, + { + "epoch": 0.748114876820471, + "grad_norm": 2.117006301879883, + "learning_rate": 5.044623142973968e-06, + "loss": 2.4692493438720704, + "step": 274820 + }, + { + "epoch": 0.7481693208112155, + "grad_norm": 2.1682281494140625, + "learning_rate": 5.043532779065014e-06, + "loss": 2.413998603820801, + "step": 274840 + }, + { + "epoch": 0.74822376480196, + "grad_norm": 2.685865640640259, + "learning_rate": 5.0424424151560595e-06, + "loss": 2.4384727478027344, + "step": 274860 + }, + { + "epoch": 0.7482782087927045, + "grad_norm": 2.123429536819458, + "learning_rate": 5.041352051247104e-06, + "loss": 2.4659923553466796, + "step": 274880 + }, + { + "epoch": 0.7483326527834491, + "grad_norm": 2.1761233806610107, + "learning_rate": 5.040261687338149e-06, + "loss": 2.4738479614257813, + "step": 274900 + }, + { + "epoch": 0.7483870967741936, + "grad_norm": 2.100001811981201, + "learning_rate": 5.039171323429195e-06, + "loss": 2.451877403259277, + "step": 274920 + }, + { + "epoch": 0.7484415407649381, + "grad_norm": 3.532482385635376, + "learning_rate": 5.038080959520241e-06, + "loss": 2.4820093154907226, + "step": 274940 + }, + { + "epoch": 0.7484959847556826, + "grad_norm": 2.342210054397583, + "learning_rate": 5.036990595611285e-06, + "loss": 2.460537147521973, + "step": 274960 + }, + { + "epoch": 0.7485504287464271, + "grad_norm": 2.4714431762695312, + "learning_rate": 5.035900231702331e-06, + "loss": 2.4391008377075196, + "step": 274980 + }, + { + "epoch": 0.7486048727371717, + "grad_norm": 2.169328451156616, + "learning_rate": 5.034809867793377e-06, + "loss": 2.46435661315918, + "step": 275000 + }, + { + "epoch": 0.7486593167279162, + "grad_norm": 2.081045150756836, + "learning_rate": 5.0337195038844226e-06, + "loss": 2.4633916854858398, + "step": 275020 + }, + { + "epoch": 0.7487137607186607, + "grad_norm": 2.2006142139434814, + "learning_rate": 5.032629139975467e-06, + "loss": 2.461091995239258, + "step": 275040 + }, + { + "epoch": 0.7487682047094052, + "grad_norm": 2.287360668182373, + "learning_rate": 5.0315387760665125e-06, + "loss": 2.457175636291504, + "step": 275060 + }, + { + "epoch": 0.7488226487001497, + "grad_norm": 2.233520030975342, + "learning_rate": 5.030448412157558e-06, + "loss": 2.4489585876464846, + "step": 275080 + }, + { + "epoch": 0.7488770926908942, + "grad_norm": 2.222695827484131, + "learning_rate": 5.029358048248604e-06, + "loss": 2.4580320358276366, + "step": 275100 + }, + { + "epoch": 0.7489315366816388, + "grad_norm": 2.168440580368042, + "learning_rate": 5.028267684339648e-06, + "loss": 2.4674917221069337, + "step": 275120 + }, + { + "epoch": 0.7489859806723833, + "grad_norm": 2.3796546459198, + "learning_rate": 5.027177320430694e-06, + "loss": 2.4916641235351564, + "step": 275140 + }, + { + "epoch": 0.7490404246631278, + "grad_norm": 2.206664800643921, + "learning_rate": 5.02608695652174e-06, + "loss": 2.480337142944336, + "step": 275160 + }, + { + "epoch": 0.7490948686538723, + "grad_norm": 2.2462332248687744, + "learning_rate": 5.024996592612786e-06, + "loss": 2.458471488952637, + "step": 275180 + }, + { + "epoch": 0.7491493126446168, + "grad_norm": 2.1912717819213867, + "learning_rate": 5.02390622870383e-06, + "loss": 2.463351249694824, + "step": 275200 + }, + { + "epoch": 0.7492037566353613, + "grad_norm": 2.1972498893737793, + "learning_rate": 5.0228158647948756e-06, + "loss": 2.4740245819091795, + "step": 275220 + }, + { + "epoch": 0.7492582006261059, + "grad_norm": 2.1755404472351074, + "learning_rate": 5.021725500885921e-06, + "loss": 2.4690486907958986, + "step": 275240 + }, + { + "epoch": 0.7493126446168504, + "grad_norm": 2.218330144882202, + "learning_rate": 5.020635136976967e-06, + "loss": 2.4484848022460937, + "step": 275260 + }, + { + "epoch": 0.7493670886075949, + "grad_norm": 2.065363883972168, + "learning_rate": 5.019544773068011e-06, + "loss": 2.4562843322753904, + "step": 275280 + }, + { + "epoch": 0.7494215325983394, + "grad_norm": 2.2038538455963135, + "learning_rate": 5.018454409159057e-06, + "loss": 2.454270362854004, + "step": 275300 + }, + { + "epoch": 0.7494759765890839, + "grad_norm": 2.1161651611328125, + "learning_rate": 5.017364045250103e-06, + "loss": 2.4626869201660155, + "step": 275320 + }, + { + "epoch": 0.7495304205798285, + "grad_norm": 2.1908183097839355, + "learning_rate": 5.016273681341149e-06, + "loss": 2.4440933227539063, + "step": 275340 + }, + { + "epoch": 0.749584864570573, + "grad_norm": 2.3076956272125244, + "learning_rate": 5.015183317432193e-06, + "loss": 2.4561704635620116, + "step": 275360 + }, + { + "epoch": 0.7496393085613176, + "grad_norm": 2.2423858642578125, + "learning_rate": 5.014092953523239e-06, + "loss": 2.4455867767333985, + "step": 275380 + }, + { + "epoch": 0.749693752552062, + "grad_norm": 2.1974334716796875, + "learning_rate": 5.0130025896142845e-06, + "loss": 2.426183891296387, + "step": 275400 + }, + { + "epoch": 0.7497481965428066, + "grad_norm": 2.1460890769958496, + "learning_rate": 5.01191222570533e-06, + "loss": 2.439506721496582, + "step": 275420 + }, + { + "epoch": 0.7498026405335511, + "grad_norm": 2.1728997230529785, + "learning_rate": 5.010821861796374e-06, + "loss": 2.475581741333008, + "step": 275440 + }, + { + "epoch": 0.7498570845242957, + "grad_norm": 2.2470622062683105, + "learning_rate": 5.00973149788742e-06, + "loss": 2.482396697998047, + "step": 275460 + }, + { + "epoch": 0.7499115285150402, + "grad_norm": 2.275479555130005, + "learning_rate": 5.008641133978466e-06, + "loss": 2.4803787231445313, + "step": 275480 + }, + { + "epoch": 0.7499659725057847, + "grad_norm": 2.1469709873199463, + "learning_rate": 5.007550770069512e-06, + "loss": 2.4378265380859374, + "step": 275500 + }, + { + "epoch": 0.7500204164965292, + "grad_norm": 2.26344633102417, + "learning_rate": 5.006460406160556e-06, + "loss": 2.4534687042236327, + "step": 275520 + }, + { + "epoch": 0.7500748604872737, + "grad_norm": 2.2210934162139893, + "learning_rate": 5.005370042251602e-06, + "loss": 2.4466279983520507, + "step": 275540 + }, + { + "epoch": 0.7501293044780183, + "grad_norm": 2.0994873046875, + "learning_rate": 5.0042796783426476e-06, + "loss": 2.4400733947753905, + "step": 275560 + }, + { + "epoch": 0.7501837484687628, + "grad_norm": 2.0978708267211914, + "learning_rate": 5.0031893144336925e-06, + "loss": 2.4621400833129883, + "step": 275580 + }, + { + "epoch": 0.7502381924595073, + "grad_norm": 2.1743197441101074, + "learning_rate": 5.002098950524738e-06, + "loss": 2.4644397735595702, + "step": 275600 + }, + { + "epoch": 0.7502926364502518, + "grad_norm": 2.1791951656341553, + "learning_rate": 5.001008586615783e-06, + "loss": 2.4739202499389648, + "step": 275620 + }, + { + "epoch": 0.7503470804409963, + "grad_norm": 2.2459678649902344, + "learning_rate": 4.999918222706829e-06, + "loss": 2.4656532287597654, + "step": 275640 + }, + { + "epoch": 0.7504015244317408, + "grad_norm": 2.8318066596984863, + "learning_rate": 4.998827858797874e-06, + "loss": 2.4404218673706053, + "step": 275660 + }, + { + "epoch": 0.7504559684224854, + "grad_norm": 2.1018459796905518, + "learning_rate": 4.99773749488892e-06, + "loss": 2.4755634307861327, + "step": 275680 + }, + { + "epoch": 0.7505104124132299, + "grad_norm": 2.1210854053497314, + "learning_rate": 4.996647130979965e-06, + "loss": 2.4677438735961914, + "step": 275700 + }, + { + "epoch": 0.7505648564039744, + "grad_norm": 2.232668399810791, + "learning_rate": 4.995556767071011e-06, + "loss": 2.4575407028198244, + "step": 275720 + }, + { + "epoch": 0.7506193003947189, + "grad_norm": 2.1250128746032715, + "learning_rate": 4.994466403162056e-06, + "loss": 2.450708770751953, + "step": 275740 + }, + { + "epoch": 0.7506737443854634, + "grad_norm": 2.19040584564209, + "learning_rate": 4.993376039253101e-06, + "loss": 2.479730415344238, + "step": 275760 + }, + { + "epoch": 0.750728188376208, + "grad_norm": 2.175441265106201, + "learning_rate": 4.992285675344146e-06, + "loss": 2.4745864868164062, + "step": 275780 + }, + { + "epoch": 0.7507826323669525, + "grad_norm": 2.173158884048462, + "learning_rate": 4.991195311435192e-06, + "loss": 2.5081371307373046, + "step": 275800 + }, + { + "epoch": 0.750837076357697, + "grad_norm": 2.0672762393951416, + "learning_rate": 4.990104947526237e-06, + "loss": 2.4759084701538088, + "step": 275820 + }, + { + "epoch": 0.7508915203484415, + "grad_norm": 2.1761903762817383, + "learning_rate": 4.989014583617283e-06, + "loss": 2.452461814880371, + "step": 275840 + }, + { + "epoch": 0.750945964339186, + "grad_norm": 2.1523494720458984, + "learning_rate": 4.987924219708328e-06, + "loss": 2.4577972412109377, + "step": 275860 + }, + { + "epoch": 0.7510004083299305, + "grad_norm": 2.1090853214263916, + "learning_rate": 4.986833855799374e-06, + "loss": 2.4590967178344725, + "step": 275880 + }, + { + "epoch": 0.7510548523206751, + "grad_norm": 2.1661036014556885, + "learning_rate": 4.985743491890419e-06, + "loss": 2.46838321685791, + "step": 275900 + }, + { + "epoch": 0.7511092963114196, + "grad_norm": 2.322887420654297, + "learning_rate": 4.9846531279814645e-06, + "loss": 2.4732028961181642, + "step": 275920 + }, + { + "epoch": 0.7511637403021642, + "grad_norm": 2.302712917327881, + "learning_rate": 4.9835627640725095e-06, + "loss": 2.4458292007446287, + "step": 275940 + }, + { + "epoch": 0.7512181842929087, + "grad_norm": 2.6426315307617188, + "learning_rate": 4.982472400163555e-06, + "loss": 2.4391143798828123, + "step": 275960 + }, + { + "epoch": 0.7512726282836532, + "grad_norm": 2.0597198009490967, + "learning_rate": 4.9813820362546e-06, + "loss": 2.4430978775024412, + "step": 275980 + }, + { + "epoch": 0.7513270722743978, + "grad_norm": 2.177609920501709, + "learning_rate": 4.980291672345646e-06, + "loss": 2.5009506225585936, + "step": 276000 + }, + { + "epoch": 0.7513815162651423, + "grad_norm": 2.7872629165649414, + "learning_rate": 4.979201308436691e-06, + "loss": 2.4726701736450196, + "step": 276020 + }, + { + "epoch": 0.7514359602558868, + "grad_norm": 2.1906511783599854, + "learning_rate": 4.978110944527737e-06, + "loss": 2.4686233520507814, + "step": 276040 + }, + { + "epoch": 0.7514904042466313, + "grad_norm": 2.210495710372925, + "learning_rate": 4.977020580618782e-06, + "loss": 2.476780128479004, + "step": 276060 + }, + { + "epoch": 0.7515448482373758, + "grad_norm": 2.117302656173706, + "learning_rate": 4.975930216709828e-06, + "loss": 2.4532299041748047, + "step": 276080 + }, + { + "epoch": 0.7515992922281203, + "grad_norm": 2.1818299293518066, + "learning_rate": 4.9748398528008725e-06, + "loss": 2.4759681701660154, + "step": 276100 + }, + { + "epoch": 0.7516537362188649, + "grad_norm": 2.113431692123413, + "learning_rate": 4.973749488891918e-06, + "loss": 2.470475196838379, + "step": 276120 + }, + { + "epoch": 0.7517081802096094, + "grad_norm": 2.1195833683013916, + "learning_rate": 4.972659124982963e-06, + "loss": 2.4622745513916016, + "step": 276140 + }, + { + "epoch": 0.7517626242003539, + "grad_norm": 2.119234561920166, + "learning_rate": 4.971568761074009e-06, + "loss": 2.4557861328125, + "step": 276160 + }, + { + "epoch": 0.7518170681910984, + "grad_norm": 2.6475789546966553, + "learning_rate": 4.970478397165054e-06, + "loss": 2.4362743377685545, + "step": 276180 + }, + { + "epoch": 0.7518715121818429, + "grad_norm": 2.0890913009643555, + "learning_rate": 4.9693880332561e-06, + "loss": 2.453235054016113, + "step": 276200 + }, + { + "epoch": 0.7519259561725874, + "grad_norm": 2.155729055404663, + "learning_rate": 4.968297669347145e-06, + "loss": 2.464823532104492, + "step": 276220 + }, + { + "epoch": 0.751980400163332, + "grad_norm": 3.1065783500671387, + "learning_rate": 4.967207305438191e-06, + "loss": 2.4853593826293947, + "step": 276240 + }, + { + "epoch": 0.7520348441540765, + "grad_norm": 2.500335693359375, + "learning_rate": 4.966116941529236e-06, + "loss": 2.4653884887695314, + "step": 276260 + }, + { + "epoch": 0.752089288144821, + "grad_norm": 2.1401584148406982, + "learning_rate": 4.9650265776202814e-06, + "loss": 2.4524362564086912, + "step": 276280 + }, + { + "epoch": 0.7521437321355655, + "grad_norm": 2.261495351791382, + "learning_rate": 4.963936213711326e-06, + "loss": 2.470339775085449, + "step": 276300 + }, + { + "epoch": 0.75219817612631, + "grad_norm": 2.216146469116211, + "learning_rate": 4.962845849802372e-06, + "loss": 2.4522850036621096, + "step": 276320 + }, + { + "epoch": 0.7522526201170546, + "grad_norm": 2.1389570236206055, + "learning_rate": 4.961755485893417e-06, + "loss": 2.491648292541504, + "step": 276340 + }, + { + "epoch": 0.7523070641077991, + "grad_norm": 2.1777257919311523, + "learning_rate": 4.960665121984463e-06, + "loss": 2.451585578918457, + "step": 276360 + }, + { + "epoch": 0.7523615080985436, + "grad_norm": 2.0807838439941406, + "learning_rate": 4.959574758075508e-06, + "loss": 2.4379413604736326, + "step": 276380 + }, + { + "epoch": 0.7524159520892881, + "grad_norm": 2.1903975009918213, + "learning_rate": 4.958484394166554e-06, + "loss": 2.4764509201049805, + "step": 276400 + }, + { + "epoch": 0.7524703960800326, + "grad_norm": 2.1037235260009766, + "learning_rate": 4.957394030257599e-06, + "loss": 2.4572010040283203, + "step": 276420 + }, + { + "epoch": 0.7525248400707771, + "grad_norm": 2.2091989517211914, + "learning_rate": 4.9563036663486445e-06, + "loss": 2.470936393737793, + "step": 276440 + }, + { + "epoch": 0.7525792840615217, + "grad_norm": 2.063936948776245, + "learning_rate": 4.9552133024396895e-06, + "loss": 2.4491559982299806, + "step": 276460 + }, + { + "epoch": 0.7526337280522662, + "grad_norm": 2.5272133350372314, + "learning_rate": 4.954122938530735e-06, + "loss": 2.4762651443481447, + "step": 276480 + }, + { + "epoch": 0.7526881720430108, + "grad_norm": 2.7942450046539307, + "learning_rate": 4.95303257462178e-06, + "loss": 2.4489665985107423, + "step": 276500 + }, + { + "epoch": 0.7527426160337553, + "grad_norm": 2.2394931316375732, + "learning_rate": 4.951942210712826e-06, + "loss": 2.4857091903686523, + "step": 276520 + }, + { + "epoch": 0.7527970600244998, + "grad_norm": 2.0834619998931885, + "learning_rate": 4.950851846803871e-06, + "loss": 2.4574001312255858, + "step": 276540 + }, + { + "epoch": 0.7528515040152444, + "grad_norm": 2.4167776107788086, + "learning_rate": 4.949761482894917e-06, + "loss": 2.4782739639282227, + "step": 276560 + }, + { + "epoch": 0.7529059480059889, + "grad_norm": 2.2919065952301025, + "learning_rate": 4.948671118985962e-06, + "loss": 2.4823787689208983, + "step": 276580 + }, + { + "epoch": 0.7529603919967334, + "grad_norm": 2.161365032196045, + "learning_rate": 4.947580755077008e-06, + "loss": 2.4654001235961913, + "step": 276600 + }, + { + "epoch": 0.7530148359874779, + "grad_norm": 2.215986728668213, + "learning_rate": 4.9464903911680526e-06, + "loss": 2.4477554321289063, + "step": 276620 + }, + { + "epoch": 0.7530692799782224, + "grad_norm": 2.2256240844726562, + "learning_rate": 4.9454000272590975e-06, + "loss": 2.4667858123779296, + "step": 276640 + }, + { + "epoch": 0.7531237239689669, + "grad_norm": 2.2651939392089844, + "learning_rate": 4.944309663350143e-06, + "loss": 2.442867088317871, + "step": 276660 + }, + { + "epoch": 0.7531781679597115, + "grad_norm": 2.2087929248809814, + "learning_rate": 4.943219299441188e-06, + "loss": 2.5122365951538086, + "step": 276680 + }, + { + "epoch": 0.753232611950456, + "grad_norm": 2.206308364868164, + "learning_rate": 4.942128935532234e-06, + "loss": 2.4356878280639647, + "step": 276700 + }, + { + "epoch": 0.7532870559412005, + "grad_norm": 2.117125988006592, + "learning_rate": 4.941038571623279e-06, + "loss": 2.494486999511719, + "step": 276720 + }, + { + "epoch": 0.753341499931945, + "grad_norm": 2.2824814319610596, + "learning_rate": 4.939948207714325e-06, + "loss": 2.464357376098633, + "step": 276740 + }, + { + "epoch": 0.7533959439226895, + "grad_norm": 2.2279064655303955, + "learning_rate": 4.93885784380537e-06, + "loss": 2.464679718017578, + "step": 276760 + }, + { + "epoch": 0.7534503879134341, + "grad_norm": 2.290912389755249, + "learning_rate": 4.937767479896416e-06, + "loss": 2.4942636489868164, + "step": 276780 + }, + { + "epoch": 0.7535048319041786, + "grad_norm": 3.7176473140716553, + "learning_rate": 4.936677115987461e-06, + "loss": 2.4398801803588865, + "step": 276800 + }, + { + "epoch": 0.7535592758949231, + "grad_norm": 2.2122068405151367, + "learning_rate": 4.9355867520785064e-06, + "loss": 2.4571582794189455, + "step": 276820 + }, + { + "epoch": 0.7536137198856676, + "grad_norm": 2.248304843902588, + "learning_rate": 4.934496388169551e-06, + "loss": 2.451504135131836, + "step": 276840 + }, + { + "epoch": 0.7536681638764121, + "grad_norm": 2.21036958694458, + "learning_rate": 4.933406024260597e-06, + "loss": 2.4693862915039064, + "step": 276860 + }, + { + "epoch": 0.7537226078671566, + "grad_norm": 2.220984935760498, + "learning_rate": 4.932315660351642e-06, + "loss": 2.463501739501953, + "step": 276880 + }, + { + "epoch": 0.7537770518579012, + "grad_norm": 2.1326262950897217, + "learning_rate": 4.931225296442688e-06, + "loss": 2.475307655334473, + "step": 276900 + }, + { + "epoch": 0.7538314958486457, + "grad_norm": 2.110729932785034, + "learning_rate": 4.930134932533733e-06, + "loss": 2.455893898010254, + "step": 276920 + }, + { + "epoch": 0.7538859398393902, + "grad_norm": 2.220388650894165, + "learning_rate": 4.929044568624779e-06, + "loss": 2.4698835372924806, + "step": 276940 + }, + { + "epoch": 0.7539403838301347, + "grad_norm": 2.126800537109375, + "learning_rate": 4.927954204715824e-06, + "loss": 2.4752729415893553, + "step": 276960 + }, + { + "epoch": 0.7539948278208792, + "grad_norm": 2.085675001144409, + "learning_rate": 4.9268638408068695e-06, + "loss": 2.4802114486694338, + "step": 276980 + }, + { + "epoch": 0.7540492718116238, + "grad_norm": 2.120007276535034, + "learning_rate": 4.9257734768979145e-06, + "loss": 2.476841926574707, + "step": 277000 + }, + { + "epoch": 0.7541037158023683, + "grad_norm": 2.106140613555908, + "learning_rate": 4.92468311298896e-06, + "loss": 2.452169418334961, + "step": 277020 + }, + { + "epoch": 0.7541581597931128, + "grad_norm": 2.1707751750946045, + "learning_rate": 4.923592749080005e-06, + "loss": 2.4662294387817383, + "step": 277040 + }, + { + "epoch": 0.7542126037838573, + "grad_norm": 2.1887106895446777, + "learning_rate": 4.922502385171051e-06, + "loss": 2.453616905212402, + "step": 277060 + }, + { + "epoch": 0.7542670477746019, + "grad_norm": 2.2378287315368652, + "learning_rate": 4.921412021262096e-06, + "loss": 2.4444881439208985, + "step": 277080 + }, + { + "epoch": 0.7543214917653464, + "grad_norm": 2.1355326175689697, + "learning_rate": 4.920321657353142e-06, + "loss": 2.4741601943969727, + "step": 277100 + }, + { + "epoch": 0.754375935756091, + "grad_norm": 2.1706390380859375, + "learning_rate": 4.919231293444187e-06, + "loss": 2.445605659484863, + "step": 277120 + }, + { + "epoch": 0.7544303797468355, + "grad_norm": 2.1296918392181396, + "learning_rate": 4.918140929535233e-06, + "loss": 2.484659767150879, + "step": 277140 + }, + { + "epoch": 0.75448482373758, + "grad_norm": 2.2692668437957764, + "learning_rate": 4.9170505656262776e-06, + "loss": 2.485629081726074, + "step": 277160 + }, + { + "epoch": 0.7545392677283245, + "grad_norm": 2.1796109676361084, + "learning_rate": 4.915960201717323e-06, + "loss": 2.484179878234863, + "step": 277180 + }, + { + "epoch": 0.754593711719069, + "grad_norm": 2.1644351482391357, + "learning_rate": 4.914869837808368e-06, + "loss": 2.4613716125488283, + "step": 277200 + }, + { + "epoch": 0.7546481557098136, + "grad_norm": 2.153338670730591, + "learning_rate": 4.913779473899414e-06, + "loss": 2.480179023742676, + "step": 277220 + }, + { + "epoch": 0.7547025997005581, + "grad_norm": 2.1167545318603516, + "learning_rate": 4.912689109990459e-06, + "loss": 2.460255813598633, + "step": 277240 + }, + { + "epoch": 0.7547570436913026, + "grad_norm": 2.1209120750427246, + "learning_rate": 4.911598746081505e-06, + "loss": 2.444293975830078, + "step": 277260 + }, + { + "epoch": 0.7548114876820471, + "grad_norm": 2.0995304584503174, + "learning_rate": 4.91050838217255e-06, + "loss": 2.4563404083251954, + "step": 277280 + }, + { + "epoch": 0.7548659316727916, + "grad_norm": 2.1231391429901123, + "learning_rate": 4.909418018263596e-06, + "loss": 2.473637008666992, + "step": 277300 + }, + { + "epoch": 0.7549203756635361, + "grad_norm": 2.2782652378082275, + "learning_rate": 4.9083276543546415e-06, + "loss": 2.463243865966797, + "step": 277320 + }, + { + "epoch": 0.7549748196542807, + "grad_norm": 2.105592966079712, + "learning_rate": 4.9072372904456865e-06, + "loss": 2.449611473083496, + "step": 277340 + }, + { + "epoch": 0.7550292636450252, + "grad_norm": 2.132791757583618, + "learning_rate": 4.906146926536732e-06, + "loss": 2.486115264892578, + "step": 277360 + }, + { + "epoch": 0.7550837076357697, + "grad_norm": 2.1532394886016846, + "learning_rate": 4.905056562627777e-06, + "loss": 2.480728340148926, + "step": 277380 + }, + { + "epoch": 0.7551381516265142, + "grad_norm": 2.15864634513855, + "learning_rate": 4.903966198718823e-06, + "loss": 2.464863586425781, + "step": 277400 + }, + { + "epoch": 0.7551925956172587, + "grad_norm": 2.226850748062134, + "learning_rate": 4.902875834809868e-06, + "loss": 2.4883306503295897, + "step": 277420 + }, + { + "epoch": 0.7552470396080032, + "grad_norm": 2.556842565536499, + "learning_rate": 4.901785470900914e-06, + "loss": 2.4786277770996095, + "step": 277440 + }, + { + "epoch": 0.7553014835987478, + "grad_norm": 2.1134657859802246, + "learning_rate": 4.900695106991959e-06, + "loss": 2.485093879699707, + "step": 277460 + }, + { + "epoch": 0.7553559275894923, + "grad_norm": 2.331197738647461, + "learning_rate": 4.899604743083005e-06, + "loss": 2.4520895004272463, + "step": 277480 + }, + { + "epoch": 0.7554103715802368, + "grad_norm": 2.224825859069824, + "learning_rate": 4.8985143791740495e-06, + "loss": 2.441752624511719, + "step": 277500 + }, + { + "epoch": 0.7554648155709813, + "grad_norm": 2.1298987865448, + "learning_rate": 4.897424015265095e-06, + "loss": 2.4684144973754885, + "step": 277520 + }, + { + "epoch": 0.7555192595617258, + "grad_norm": 2.320183038711548, + "learning_rate": 4.89633365135614e-06, + "loss": 2.461301803588867, + "step": 277540 + }, + { + "epoch": 0.7555737035524704, + "grad_norm": 2.149181365966797, + "learning_rate": 4.895243287447186e-06, + "loss": 2.4973447799682615, + "step": 277560 + }, + { + "epoch": 0.755628147543215, + "grad_norm": 2.1699209213256836, + "learning_rate": 4.894152923538231e-06, + "loss": 2.497310447692871, + "step": 277580 + }, + { + "epoch": 0.7556825915339594, + "grad_norm": 2.262857675552368, + "learning_rate": 4.893062559629277e-06, + "loss": 2.485859680175781, + "step": 277600 + }, + { + "epoch": 0.755737035524704, + "grad_norm": 2.147974967956543, + "learning_rate": 4.891972195720322e-06, + "loss": 2.464213562011719, + "step": 277620 + }, + { + "epoch": 0.7557914795154484, + "grad_norm": 2.354691982269287, + "learning_rate": 4.890881831811368e-06, + "loss": 2.4548294067382814, + "step": 277640 + }, + { + "epoch": 0.755845923506193, + "grad_norm": 2.3418726921081543, + "learning_rate": 4.889791467902413e-06, + "loss": 2.4674205780029297, + "step": 277660 + }, + { + "epoch": 0.7559003674969376, + "grad_norm": 2.142152786254883, + "learning_rate": 4.8887011039934584e-06, + "loss": 2.462133598327637, + "step": 277680 + }, + { + "epoch": 0.7559548114876821, + "grad_norm": 2.1148462295532227, + "learning_rate": 4.887610740084503e-06, + "loss": 2.471060562133789, + "step": 277700 + }, + { + "epoch": 0.7560092554784266, + "grad_norm": 2.2197248935699463, + "learning_rate": 4.886520376175549e-06, + "loss": 2.5037235260009765, + "step": 277720 + }, + { + "epoch": 0.7560636994691711, + "grad_norm": 2.224414348602295, + "learning_rate": 4.885430012266594e-06, + "loss": 2.483690643310547, + "step": 277740 + }, + { + "epoch": 0.7561181434599156, + "grad_norm": 2.836881637573242, + "learning_rate": 4.88433964835764e-06, + "loss": 2.4452632904052733, + "step": 277760 + }, + { + "epoch": 0.7561725874506602, + "grad_norm": 4.346047878265381, + "learning_rate": 4.883249284448685e-06, + "loss": 2.4678508758544924, + "step": 277780 + }, + { + "epoch": 0.7562270314414047, + "grad_norm": 4.7326483726501465, + "learning_rate": 4.882158920539731e-06, + "loss": 2.4840112686157227, + "step": 277800 + }, + { + "epoch": 0.7562814754321492, + "grad_norm": 2.215285301208496, + "learning_rate": 4.881068556630776e-06, + "loss": 2.466990280151367, + "step": 277820 + }, + { + "epoch": 0.7563359194228937, + "grad_norm": 2.100231409072876, + "learning_rate": 4.8799781927218215e-06, + "loss": 2.485938835144043, + "step": 277840 + }, + { + "epoch": 0.7563903634136382, + "grad_norm": 2.136993169784546, + "learning_rate": 4.8788878288128665e-06, + "loss": 2.467764472961426, + "step": 277860 + }, + { + "epoch": 0.7564448074043827, + "grad_norm": 2.1190736293792725, + "learning_rate": 4.877797464903912e-06, + "loss": 2.4658382415771483, + "step": 277880 + }, + { + "epoch": 0.7564992513951273, + "grad_norm": 2.2083606719970703, + "learning_rate": 4.876707100994957e-06, + "loss": 2.4445169448852537, + "step": 277900 + }, + { + "epoch": 0.7565536953858718, + "grad_norm": 2.184816598892212, + "learning_rate": 4.875616737086003e-06, + "loss": 2.471142578125, + "step": 277920 + }, + { + "epoch": 0.7566081393766163, + "grad_norm": 2.16937518119812, + "learning_rate": 4.874526373177048e-06, + "loss": 2.445456123352051, + "step": 277940 + }, + { + "epoch": 0.7566625833673608, + "grad_norm": 2.136695146560669, + "learning_rate": 4.873436009268094e-06, + "loss": 2.467037010192871, + "step": 277960 + }, + { + "epoch": 0.7567170273581053, + "grad_norm": 2.2296643257141113, + "learning_rate": 4.872345645359139e-06, + "loss": 2.468893623352051, + "step": 277980 + }, + { + "epoch": 0.7567714713488499, + "grad_norm": 2.2167508602142334, + "learning_rate": 4.871255281450185e-06, + "loss": 2.4509746551513674, + "step": 278000 + }, + { + "epoch": 0.7568259153395944, + "grad_norm": 2.2303829193115234, + "learning_rate": 4.8701649175412296e-06, + "loss": 2.491358184814453, + "step": 278020 + }, + { + "epoch": 0.7568803593303389, + "grad_norm": 2.130000591278076, + "learning_rate": 4.869074553632275e-06, + "loss": 2.4529678344726564, + "step": 278040 + }, + { + "epoch": 0.7569348033210834, + "grad_norm": 2.1323745250701904, + "learning_rate": 4.86798418972332e-06, + "loss": 2.453006935119629, + "step": 278060 + }, + { + "epoch": 0.7569892473118279, + "grad_norm": 2.2906014919281006, + "learning_rate": 4.866893825814366e-06, + "loss": 2.4511844635009767, + "step": 278080 + }, + { + "epoch": 0.7570436913025724, + "grad_norm": 2.2730114459991455, + "learning_rate": 4.865803461905411e-06, + "loss": 2.4895591735839844, + "step": 278100 + }, + { + "epoch": 0.757098135293317, + "grad_norm": 2.1914167404174805, + "learning_rate": 4.864713097996457e-06, + "loss": 2.4889495849609373, + "step": 278120 + }, + { + "epoch": 0.7571525792840615, + "grad_norm": 2.5366313457489014, + "learning_rate": 4.863622734087502e-06, + "loss": 2.4387451171875, + "step": 278140 + }, + { + "epoch": 0.757207023274806, + "grad_norm": 2.1389565467834473, + "learning_rate": 4.862532370178548e-06, + "loss": 2.486813545227051, + "step": 278160 + }, + { + "epoch": 0.7572614672655505, + "grad_norm": 2.1259765625, + "learning_rate": 4.861442006269593e-06, + "loss": 2.481136703491211, + "step": 278180 + }, + { + "epoch": 0.757315911256295, + "grad_norm": 2.1467013359069824, + "learning_rate": 4.8603516423606385e-06, + "loss": 2.446381378173828, + "step": 278200 + }, + { + "epoch": 0.7573703552470397, + "grad_norm": 2.752626895904541, + "learning_rate": 4.8592612784516834e-06, + "loss": 2.4802059173583983, + "step": 278220 + }, + { + "epoch": 0.7574247992377842, + "grad_norm": 2.206550359725952, + "learning_rate": 4.858170914542729e-06, + "loss": 2.4685541152954102, + "step": 278240 + }, + { + "epoch": 0.7574792432285287, + "grad_norm": 2.18062686920166, + "learning_rate": 4.857080550633774e-06, + "loss": 2.4476953506469727, + "step": 278260 + }, + { + "epoch": 0.7575336872192732, + "grad_norm": 2.3819401264190674, + "learning_rate": 4.85599018672482e-06, + "loss": 2.4603239059448243, + "step": 278280 + }, + { + "epoch": 0.7575881312100177, + "grad_norm": 2.0908312797546387, + "learning_rate": 4.854899822815865e-06, + "loss": 2.451718330383301, + "step": 278300 + }, + { + "epoch": 0.7576425752007622, + "grad_norm": 2.167080879211426, + "learning_rate": 4.853809458906911e-06, + "loss": 2.4774017333984375, + "step": 278320 + }, + { + "epoch": 0.7576970191915068, + "grad_norm": 2.0646426677703857, + "learning_rate": 4.852719094997956e-06, + "loss": 2.460525703430176, + "step": 278340 + }, + { + "epoch": 0.7577514631822513, + "grad_norm": 2.0944325923919678, + "learning_rate": 4.8516287310890016e-06, + "loss": 2.4729738235473633, + "step": 278360 + }, + { + "epoch": 0.7578059071729958, + "grad_norm": 2.285174608230591, + "learning_rate": 4.8505383671800465e-06, + "loss": 2.5087051391601562, + "step": 278380 + }, + { + "epoch": 0.7578603511637403, + "grad_norm": 2.111711025238037, + "learning_rate": 4.849448003271092e-06, + "loss": 2.4325834274291993, + "step": 278400 + }, + { + "epoch": 0.7579147951544848, + "grad_norm": 2.1817026138305664, + "learning_rate": 4.848357639362137e-06, + "loss": 2.4705698013305666, + "step": 278420 + }, + { + "epoch": 0.7579692391452293, + "grad_norm": 2.229261636734009, + "learning_rate": 4.847267275453183e-06, + "loss": 2.496653175354004, + "step": 278440 + }, + { + "epoch": 0.7580236831359739, + "grad_norm": 2.1661078929901123, + "learning_rate": 4.846176911544228e-06, + "loss": 2.440363121032715, + "step": 278460 + }, + { + "epoch": 0.7580781271267184, + "grad_norm": 2.1187305450439453, + "learning_rate": 4.845086547635274e-06, + "loss": 2.4614233016967773, + "step": 278480 + }, + { + "epoch": 0.7581325711174629, + "grad_norm": 2.309384346008301, + "learning_rate": 4.843996183726319e-06, + "loss": 2.4587400436401365, + "step": 278500 + }, + { + "epoch": 0.7581870151082074, + "grad_norm": 2.4566426277160645, + "learning_rate": 4.842905819817365e-06, + "loss": 2.458381462097168, + "step": 278520 + }, + { + "epoch": 0.7582414590989519, + "grad_norm": 2.1346027851104736, + "learning_rate": 4.84181545590841e-06, + "loss": 2.4447938919067385, + "step": 278540 + }, + { + "epoch": 0.7582959030896965, + "grad_norm": 2.307943105697632, + "learning_rate": 4.840725091999455e-06, + "loss": 2.4818441390991213, + "step": 278560 + }, + { + "epoch": 0.758350347080441, + "grad_norm": 2.139889717102051, + "learning_rate": 4.8396347280905e-06, + "loss": 2.4737987518310547, + "step": 278580 + }, + { + "epoch": 0.7584047910711855, + "grad_norm": 2.2688841819763184, + "learning_rate": 4.838544364181546e-06, + "loss": 2.471420478820801, + "step": 278600 + }, + { + "epoch": 0.75845923506193, + "grad_norm": 2.173308849334717, + "learning_rate": 4.837454000272591e-06, + "loss": 2.4733877182006836, + "step": 278620 + }, + { + "epoch": 0.7585136790526745, + "grad_norm": 2.2767629623413086, + "learning_rate": 4.836363636363637e-06, + "loss": 2.473252296447754, + "step": 278640 + }, + { + "epoch": 0.758568123043419, + "grad_norm": 2.0167033672332764, + "learning_rate": 4.835273272454682e-06, + "loss": 2.4344722747802736, + "step": 278660 + }, + { + "epoch": 0.7586225670341636, + "grad_norm": 2.3198676109313965, + "learning_rate": 4.834182908545728e-06, + "loss": 2.472797966003418, + "step": 278680 + }, + { + "epoch": 0.7586770110249081, + "grad_norm": 2.2437632083892822, + "learning_rate": 4.833092544636773e-06, + "loss": 2.464715576171875, + "step": 278700 + }, + { + "epoch": 0.7587314550156526, + "grad_norm": 2.294907808303833, + "learning_rate": 4.8320021807278185e-06, + "loss": 2.4517698287963867, + "step": 278720 + }, + { + "epoch": 0.7587858990063971, + "grad_norm": 2.106236457824707, + "learning_rate": 4.8309118168188635e-06, + "loss": 2.49122314453125, + "step": 278740 + }, + { + "epoch": 0.7588403429971416, + "grad_norm": 2.148080348968506, + "learning_rate": 4.829821452909909e-06, + "loss": 2.444244384765625, + "step": 278760 + }, + { + "epoch": 0.7588947869878863, + "grad_norm": 2.300325632095337, + "learning_rate": 4.828731089000954e-06, + "loss": 2.4555356979370115, + "step": 278780 + }, + { + "epoch": 0.7589492309786308, + "grad_norm": 2.1118762493133545, + "learning_rate": 4.827640725092e-06, + "loss": 2.4529727935791015, + "step": 278800 + }, + { + "epoch": 0.7590036749693753, + "grad_norm": 2.163386106491089, + "learning_rate": 4.826550361183045e-06, + "loss": 2.4587177276611327, + "step": 278820 + }, + { + "epoch": 0.7590581189601198, + "grad_norm": 2.150770425796509, + "learning_rate": 4.825459997274091e-06, + "loss": 2.4755647659301756, + "step": 278840 + }, + { + "epoch": 0.7591125629508643, + "grad_norm": 2.1764371395111084, + "learning_rate": 4.824369633365136e-06, + "loss": 2.4569381713867187, + "step": 278860 + }, + { + "epoch": 0.7591670069416088, + "grad_norm": 2.21684193611145, + "learning_rate": 4.823279269456182e-06, + "loss": 2.4616174697875977, + "step": 278880 + }, + { + "epoch": 0.7592214509323534, + "grad_norm": 2.1775946617126465, + "learning_rate": 4.8221889055472265e-06, + "loss": 2.4546709060668945, + "step": 278900 + }, + { + "epoch": 0.7592758949230979, + "grad_norm": 2.2678000926971436, + "learning_rate": 4.821098541638272e-06, + "loss": 2.4876211166381834, + "step": 278920 + }, + { + "epoch": 0.7593303389138424, + "grad_norm": 2.2121245861053467, + "learning_rate": 4.820008177729317e-06, + "loss": 2.516241264343262, + "step": 278940 + }, + { + "epoch": 0.7593847829045869, + "grad_norm": 2.139601230621338, + "learning_rate": 4.818917813820363e-06, + "loss": 2.4834442138671875, + "step": 278960 + }, + { + "epoch": 0.7594392268953314, + "grad_norm": 2.1791348457336426, + "learning_rate": 4.817827449911408e-06, + "loss": 2.470294189453125, + "step": 278980 + }, + { + "epoch": 0.759493670886076, + "grad_norm": 2.1645514965057373, + "learning_rate": 4.816737086002454e-06, + "loss": 2.476011276245117, + "step": 279000 + }, + { + "epoch": 0.7595481148768205, + "grad_norm": 2.144784688949585, + "learning_rate": 4.8156467220935e-06, + "loss": 2.451447677612305, + "step": 279020 + }, + { + "epoch": 0.759602558867565, + "grad_norm": 2.1311841011047363, + "learning_rate": 4.814556358184545e-06, + "loss": 2.4621454238891602, + "step": 279040 + }, + { + "epoch": 0.7596570028583095, + "grad_norm": 2.2006611824035645, + "learning_rate": 4.8134659942755905e-06, + "loss": 2.4595224380493166, + "step": 279060 + }, + { + "epoch": 0.759711446849054, + "grad_norm": 2.213090181350708, + "learning_rate": 4.8123756303666354e-06, + "loss": 2.496173286437988, + "step": 279080 + }, + { + "epoch": 0.7597658908397985, + "grad_norm": 3.097048044204712, + "learning_rate": 4.811285266457681e-06, + "loss": 2.4752052307128904, + "step": 279100 + }, + { + "epoch": 0.7598203348305431, + "grad_norm": 2.162982225418091, + "learning_rate": 4.810194902548726e-06, + "loss": 2.4322383880615233, + "step": 279120 + }, + { + "epoch": 0.7598747788212876, + "grad_norm": 2.203890085220337, + "learning_rate": 4.809104538639772e-06, + "loss": 2.444380187988281, + "step": 279140 + }, + { + "epoch": 0.7599292228120321, + "grad_norm": 2.1713032722473145, + "learning_rate": 4.808014174730817e-06, + "loss": 2.445944404602051, + "step": 279160 + }, + { + "epoch": 0.7599836668027766, + "grad_norm": 2.3677704334259033, + "learning_rate": 4.806923810821863e-06, + "loss": 2.460296630859375, + "step": 279180 + }, + { + "epoch": 0.7600381107935211, + "grad_norm": 2.1320600509643555, + "learning_rate": 4.805833446912908e-06, + "loss": 2.4632537841796873, + "step": 279200 + }, + { + "epoch": 0.7600925547842657, + "grad_norm": 2.1290218830108643, + "learning_rate": 4.8047430830039536e-06, + "loss": 2.4835166931152344, + "step": 279220 + }, + { + "epoch": 0.7601469987750102, + "grad_norm": 2.1746792793273926, + "learning_rate": 4.8036527190949985e-06, + "loss": 2.47896785736084, + "step": 279240 + }, + { + "epoch": 0.7602014427657547, + "grad_norm": 2.293684482574463, + "learning_rate": 4.802562355186044e-06, + "loss": 2.448287773132324, + "step": 279260 + }, + { + "epoch": 0.7602558867564992, + "grad_norm": 2.251598834991455, + "learning_rate": 4.801471991277089e-06, + "loss": 2.498388481140137, + "step": 279280 + }, + { + "epoch": 0.7603103307472437, + "grad_norm": 2.4515187740325928, + "learning_rate": 4.800381627368135e-06, + "loss": 2.4498384475708006, + "step": 279300 + }, + { + "epoch": 0.7603647747379882, + "grad_norm": 2.193697929382324, + "learning_rate": 4.79929126345918e-06, + "loss": 2.4820871353149414, + "step": 279320 + }, + { + "epoch": 0.7604192187287329, + "grad_norm": 2.209937572479248, + "learning_rate": 4.798200899550226e-06, + "loss": 2.4522523880004883, + "step": 279340 + }, + { + "epoch": 0.7604736627194774, + "grad_norm": 2.129160165786743, + "learning_rate": 4.797110535641271e-06, + "loss": 2.4492258071899413, + "step": 279360 + }, + { + "epoch": 0.7605281067102219, + "grad_norm": 2.1217377185821533, + "learning_rate": 4.796020171732317e-06, + "loss": 2.447457695007324, + "step": 279380 + }, + { + "epoch": 0.7605825507009664, + "grad_norm": 2.3440101146698, + "learning_rate": 4.794929807823362e-06, + "loss": 2.4704856872558594, + "step": 279400 + }, + { + "epoch": 0.7606369946917109, + "grad_norm": 2.1385364532470703, + "learning_rate": 4.793839443914407e-06, + "loss": 2.440283203125, + "step": 279420 + }, + { + "epoch": 0.7606914386824555, + "grad_norm": 2.0746777057647705, + "learning_rate": 4.792749080005452e-06, + "loss": 2.4571508407592773, + "step": 279440 + }, + { + "epoch": 0.7607458826732, + "grad_norm": 2.2053942680358887, + "learning_rate": 4.791658716096498e-06, + "loss": 2.4653947830200194, + "step": 279460 + }, + { + "epoch": 0.7608003266639445, + "grad_norm": 2.2224771976470947, + "learning_rate": 4.790568352187543e-06, + "loss": 2.453057289123535, + "step": 279480 + }, + { + "epoch": 0.760854770654689, + "grad_norm": 2.2670843601226807, + "learning_rate": 4.789477988278589e-06, + "loss": 2.459877777099609, + "step": 279500 + }, + { + "epoch": 0.7609092146454335, + "grad_norm": 2.148076057434082, + "learning_rate": 4.788387624369634e-06, + "loss": 2.4697956085205077, + "step": 279520 + }, + { + "epoch": 0.760963658636178, + "grad_norm": 2.2906978130340576, + "learning_rate": 4.78729726046068e-06, + "loss": 2.484098434448242, + "step": 279540 + }, + { + "epoch": 0.7610181026269226, + "grad_norm": 2.125762462615967, + "learning_rate": 4.786206896551725e-06, + "loss": 2.4614797592163087, + "step": 279560 + }, + { + "epoch": 0.7610725466176671, + "grad_norm": 2.1544559001922607, + "learning_rate": 4.7851165326427705e-06, + "loss": 2.472318649291992, + "step": 279580 + }, + { + "epoch": 0.7611269906084116, + "grad_norm": 2.132824420928955, + "learning_rate": 4.7840261687338155e-06, + "loss": 2.4595077514648436, + "step": 279600 + }, + { + "epoch": 0.7611814345991561, + "grad_norm": 2.181607961654663, + "learning_rate": 4.782935804824861e-06, + "loss": 2.4712934494018555, + "step": 279620 + }, + { + "epoch": 0.7612358785899006, + "grad_norm": 2.17961049079895, + "learning_rate": 4.781845440915906e-06, + "loss": 2.4742673873901366, + "step": 279640 + }, + { + "epoch": 0.7612903225806451, + "grad_norm": 2.1145215034484863, + "learning_rate": 4.780755077006952e-06, + "loss": 2.430704116821289, + "step": 279660 + }, + { + "epoch": 0.7613447665713897, + "grad_norm": 2.02986216545105, + "learning_rate": 4.779664713097997e-06, + "loss": 2.4560827255249023, + "step": 279680 + }, + { + "epoch": 0.7613992105621342, + "grad_norm": 2.175497531890869, + "learning_rate": 4.778574349189042e-06, + "loss": 2.443999099731445, + "step": 279700 + }, + { + "epoch": 0.7614536545528787, + "grad_norm": 2.0787062644958496, + "learning_rate": 4.777483985280088e-06, + "loss": 2.4648715972900392, + "step": 279720 + }, + { + "epoch": 0.7615080985436232, + "grad_norm": 2.425553560256958, + "learning_rate": 4.776393621371133e-06, + "loss": 2.462150764465332, + "step": 279740 + }, + { + "epoch": 0.7615625425343677, + "grad_norm": 2.2080910205841064, + "learning_rate": 4.7753032574621786e-06, + "loss": 2.4626186370849608, + "step": 279760 + }, + { + "epoch": 0.7616169865251123, + "grad_norm": 2.09403395652771, + "learning_rate": 4.7742128935532235e-06, + "loss": 2.4762948989868163, + "step": 279780 + }, + { + "epoch": 0.7616714305158568, + "grad_norm": 2.364335775375366, + "learning_rate": 4.773122529644269e-06, + "loss": 2.4375925064086914, + "step": 279800 + }, + { + "epoch": 0.7617258745066013, + "grad_norm": 2.160517692565918, + "learning_rate": 4.772032165735314e-06, + "loss": 2.4410451889038085, + "step": 279820 + }, + { + "epoch": 0.7617803184973458, + "grad_norm": 2.150026798248291, + "learning_rate": 4.77094180182636e-06, + "loss": 2.4820554733276365, + "step": 279840 + }, + { + "epoch": 0.7618347624880903, + "grad_norm": 2.2211196422576904, + "learning_rate": 4.769851437917405e-06, + "loss": 2.4394643783569334, + "step": 279860 + }, + { + "epoch": 0.7618892064788348, + "grad_norm": 2.1134414672851562, + "learning_rate": 4.768761074008451e-06, + "loss": 2.485949897766113, + "step": 279880 + }, + { + "epoch": 0.7619436504695795, + "grad_norm": 2.145073890686035, + "learning_rate": 4.767670710099496e-06, + "loss": 2.4854650497436523, + "step": 279900 + }, + { + "epoch": 0.761998094460324, + "grad_norm": 2.1784160137176514, + "learning_rate": 4.766580346190542e-06, + "loss": 2.464967155456543, + "step": 279920 + }, + { + "epoch": 0.7620525384510685, + "grad_norm": 2.0901315212249756, + "learning_rate": 4.765489982281587e-06, + "loss": 2.477215576171875, + "step": 279940 + }, + { + "epoch": 0.762106982441813, + "grad_norm": 2.2637805938720703, + "learning_rate": 4.764399618372632e-06, + "loss": 2.449182319641113, + "step": 279960 + }, + { + "epoch": 0.7621614264325575, + "grad_norm": 2.138047695159912, + "learning_rate": 4.763309254463677e-06, + "loss": 2.4999555587768554, + "step": 279980 + }, + { + "epoch": 0.7622158704233021, + "grad_norm": 2.141141176223755, + "learning_rate": 4.762218890554723e-06, + "loss": 2.449898910522461, + "step": 280000 + }, + { + "epoch": 0.7622703144140466, + "grad_norm": 2.3887035846710205, + "learning_rate": 4.761128526645768e-06, + "loss": 2.457413101196289, + "step": 280020 + }, + { + "epoch": 0.7623247584047911, + "grad_norm": 2.2186920642852783, + "learning_rate": 4.760038162736814e-06, + "loss": 2.4761985778808593, + "step": 280040 + }, + { + "epoch": 0.7623792023955356, + "grad_norm": 2.1367268562316895, + "learning_rate": 4.758947798827859e-06, + "loss": 2.445011329650879, + "step": 280060 + }, + { + "epoch": 0.7624336463862801, + "grad_norm": 2.1206183433532715, + "learning_rate": 4.757857434918905e-06, + "loss": 2.4606943130493164, + "step": 280080 + }, + { + "epoch": 0.7624880903770246, + "grad_norm": 2.125108242034912, + "learning_rate": 4.75676707100995e-06, + "loss": 2.451494598388672, + "step": 280100 + }, + { + "epoch": 0.7625425343677692, + "grad_norm": 2.3578083515167236, + "learning_rate": 4.7556767071009955e-06, + "loss": 2.467241668701172, + "step": 280120 + }, + { + "epoch": 0.7625969783585137, + "grad_norm": 2.1574673652648926, + "learning_rate": 4.7545863431920405e-06, + "loss": 2.447007942199707, + "step": 280140 + }, + { + "epoch": 0.7626514223492582, + "grad_norm": 2.1860668659210205, + "learning_rate": 4.753495979283086e-06, + "loss": 2.4805294036865235, + "step": 280160 + }, + { + "epoch": 0.7627058663400027, + "grad_norm": 2.280834913253784, + "learning_rate": 4.752405615374131e-06, + "loss": 2.4643157958984374, + "step": 280180 + }, + { + "epoch": 0.7627603103307472, + "grad_norm": 2.2119407653808594, + "learning_rate": 4.751315251465177e-06, + "loss": 2.4468828201293946, + "step": 280200 + }, + { + "epoch": 0.7628147543214918, + "grad_norm": 2.1656203269958496, + "learning_rate": 4.750224887556222e-06, + "loss": 2.4569883346557617, + "step": 280220 + }, + { + "epoch": 0.7628691983122363, + "grad_norm": 2.1388206481933594, + "learning_rate": 4.749134523647268e-06, + "loss": 2.4836143493652343, + "step": 280240 + }, + { + "epoch": 0.7629236423029808, + "grad_norm": 2.3478517532348633, + "learning_rate": 4.748044159738313e-06, + "loss": 2.4780252456665037, + "step": 280260 + }, + { + "epoch": 0.7629780862937253, + "grad_norm": 2.2146918773651123, + "learning_rate": 4.746953795829359e-06, + "loss": 2.4512767791748047, + "step": 280280 + }, + { + "epoch": 0.7630325302844698, + "grad_norm": 2.339329957962036, + "learning_rate": 4.7458634319204035e-06, + "loss": 2.4997880935668944, + "step": 280300 + }, + { + "epoch": 0.7630869742752143, + "grad_norm": 2.1319727897644043, + "learning_rate": 4.744773068011449e-06, + "loss": 2.4669239044189455, + "step": 280320 + }, + { + "epoch": 0.7631414182659589, + "grad_norm": 2.162532091140747, + "learning_rate": 4.743682704102494e-06, + "loss": 2.488203430175781, + "step": 280340 + }, + { + "epoch": 0.7631958622567034, + "grad_norm": 2.3063161373138428, + "learning_rate": 4.74259234019354e-06, + "loss": 2.4696922302246094, + "step": 280360 + }, + { + "epoch": 0.7632503062474479, + "grad_norm": 2.1431753635406494, + "learning_rate": 4.741501976284585e-06, + "loss": 2.4754756927490233, + "step": 280380 + }, + { + "epoch": 0.7633047502381924, + "grad_norm": 2.1768383979797363, + "learning_rate": 4.740411612375631e-06, + "loss": 2.460190010070801, + "step": 280400 + }, + { + "epoch": 0.7633591942289369, + "grad_norm": 2.194826126098633, + "learning_rate": 4.739321248466676e-06, + "loss": 2.4417871475219726, + "step": 280420 + }, + { + "epoch": 0.7634136382196816, + "grad_norm": 2.219580888748169, + "learning_rate": 4.738230884557722e-06, + "loss": 2.4346765518188476, + "step": 280440 + }, + { + "epoch": 0.7634680822104261, + "grad_norm": 2.071091651916504, + "learning_rate": 4.737140520648767e-06, + "loss": 2.427909088134766, + "step": 280460 + }, + { + "epoch": 0.7635225262011706, + "grad_norm": 2.208510637283325, + "learning_rate": 4.7360501567398124e-06, + "loss": 2.5001289367675783, + "step": 280480 + }, + { + "epoch": 0.7635769701919151, + "grad_norm": 2.1529784202575684, + "learning_rate": 4.734959792830857e-06, + "loss": 2.4525712966918944, + "step": 280500 + }, + { + "epoch": 0.7636314141826596, + "grad_norm": 2.157878875732422, + "learning_rate": 4.733869428921903e-06, + "loss": 2.4620119094848634, + "step": 280520 + }, + { + "epoch": 0.7636858581734041, + "grad_norm": 2.216573476791382, + "learning_rate": 4.732779065012948e-06, + "loss": 2.482649230957031, + "step": 280540 + }, + { + "epoch": 0.7637403021641487, + "grad_norm": 2.253061056137085, + "learning_rate": 4.731688701103994e-06, + "loss": 2.4510944366455076, + "step": 280560 + }, + { + "epoch": 0.7637947461548932, + "grad_norm": 2.1121771335601807, + "learning_rate": 4.730598337195039e-06, + "loss": 2.47554874420166, + "step": 280580 + }, + { + "epoch": 0.7638491901456377, + "grad_norm": 2.1211676597595215, + "learning_rate": 4.729507973286085e-06, + "loss": 2.4604137420654295, + "step": 280600 + }, + { + "epoch": 0.7639036341363822, + "grad_norm": 2.1932477951049805, + "learning_rate": 4.72841760937713e-06, + "loss": 2.446195602416992, + "step": 280620 + }, + { + "epoch": 0.7639580781271267, + "grad_norm": 2.1485071182250977, + "learning_rate": 4.7273272454681755e-06, + "loss": 2.4633499145507813, + "step": 280640 + }, + { + "epoch": 0.7640125221178712, + "grad_norm": 2.0707309246063232, + "learning_rate": 4.7262368815592205e-06, + "loss": 2.448940086364746, + "step": 280660 + }, + { + "epoch": 0.7640669661086158, + "grad_norm": 2.152212619781494, + "learning_rate": 4.725146517650266e-06, + "loss": 2.501931381225586, + "step": 280680 + }, + { + "epoch": 0.7641214100993603, + "grad_norm": 2.089617967605591, + "learning_rate": 4.724056153741311e-06, + "loss": 2.4386749267578125, + "step": 280700 + }, + { + "epoch": 0.7641758540901048, + "grad_norm": 2.172797441482544, + "learning_rate": 4.722965789832357e-06, + "loss": 2.4830953598022463, + "step": 280720 + }, + { + "epoch": 0.7642302980808493, + "grad_norm": 2.315295934677124, + "learning_rate": 4.721875425923402e-06, + "loss": 2.4913291931152344, + "step": 280740 + }, + { + "epoch": 0.7642847420715938, + "grad_norm": 2.4291439056396484, + "learning_rate": 4.720785062014448e-06, + "loss": 2.479191017150879, + "step": 280760 + }, + { + "epoch": 0.7643391860623384, + "grad_norm": 2.266786575317383, + "learning_rate": 4.719694698105493e-06, + "loss": 2.442205047607422, + "step": 280780 + }, + { + "epoch": 0.7643936300530829, + "grad_norm": 3.961745500564575, + "learning_rate": 4.718604334196539e-06, + "loss": 2.498158264160156, + "step": 280800 + }, + { + "epoch": 0.7644480740438274, + "grad_norm": 2.127507209777832, + "learning_rate": 4.7175139702875836e-06, + "loss": 2.4732187271118162, + "step": 280820 + }, + { + "epoch": 0.7645025180345719, + "grad_norm": 2.1041598320007324, + "learning_rate": 4.716423606378629e-06, + "loss": 2.4832284927368162, + "step": 280840 + }, + { + "epoch": 0.7645569620253164, + "grad_norm": 2.097285509109497, + "learning_rate": 4.715333242469674e-06, + "loss": 2.436824607849121, + "step": 280860 + }, + { + "epoch": 0.7646114060160609, + "grad_norm": 2.205538034439087, + "learning_rate": 4.71424287856072e-06, + "loss": 2.4696372985839843, + "step": 280880 + }, + { + "epoch": 0.7646658500068055, + "grad_norm": 2.1738173961639404, + "learning_rate": 4.713152514651765e-06, + "loss": 2.4615041732788088, + "step": 280900 + }, + { + "epoch": 0.76472029399755, + "grad_norm": 2.2091495990753174, + "learning_rate": 4.712062150742811e-06, + "loss": 2.481267738342285, + "step": 280920 + }, + { + "epoch": 0.7647747379882945, + "grad_norm": 2.1280229091644287, + "learning_rate": 4.710971786833856e-06, + "loss": 2.461243438720703, + "step": 280940 + }, + { + "epoch": 0.764829181979039, + "grad_norm": 2.15301513671875, + "learning_rate": 4.709881422924902e-06, + "loss": 2.474880027770996, + "step": 280960 + }, + { + "epoch": 0.7648836259697835, + "grad_norm": 2.031947374343872, + "learning_rate": 4.708791059015947e-06, + "loss": 2.4677356719970702, + "step": 280980 + }, + { + "epoch": 0.7649380699605282, + "grad_norm": 2.1933016777038574, + "learning_rate": 4.7077006951069925e-06, + "loss": 2.463370513916016, + "step": 281000 + }, + { + "epoch": 0.7649925139512727, + "grad_norm": 2.2107150554656982, + "learning_rate": 4.7066103311980374e-06, + "loss": 2.486993980407715, + "step": 281020 + }, + { + "epoch": 0.7650469579420172, + "grad_norm": 2.161745071411133, + "learning_rate": 4.705519967289083e-06, + "loss": 2.464832305908203, + "step": 281040 + }, + { + "epoch": 0.7651014019327617, + "grad_norm": 2.0948922634124756, + "learning_rate": 4.704429603380128e-06, + "loss": 2.4570398330688477, + "step": 281060 + }, + { + "epoch": 0.7651558459235062, + "grad_norm": 2.064138174057007, + "learning_rate": 4.703339239471174e-06, + "loss": 2.4667945861816407, + "step": 281080 + }, + { + "epoch": 0.7652102899142507, + "grad_norm": 2.2333340644836426, + "learning_rate": 4.702248875562219e-06, + "loss": 2.461087226867676, + "step": 281100 + }, + { + "epoch": 0.7652647339049953, + "grad_norm": 2.513615131378174, + "learning_rate": 4.701158511653265e-06, + "loss": 2.468945121765137, + "step": 281120 + }, + { + "epoch": 0.7653191778957398, + "grad_norm": 2.1165926456451416, + "learning_rate": 4.70006814774431e-06, + "loss": 2.497445297241211, + "step": 281140 + }, + { + "epoch": 0.7653736218864843, + "grad_norm": 2.259512424468994, + "learning_rate": 4.6989777838353556e-06, + "loss": 2.4676666259765625, + "step": 281160 + }, + { + "epoch": 0.7654280658772288, + "grad_norm": 2.1908693313598633, + "learning_rate": 4.6978874199264005e-06, + "loss": 2.4590429306030273, + "step": 281180 + }, + { + "epoch": 0.7654825098679733, + "grad_norm": 2.1949284076690674, + "learning_rate": 4.696797056017446e-06, + "loss": 2.483584976196289, + "step": 281200 + }, + { + "epoch": 0.7655369538587179, + "grad_norm": 3.182188034057617, + "learning_rate": 4.695706692108491e-06, + "loss": 2.463209533691406, + "step": 281220 + }, + { + "epoch": 0.7655913978494624, + "grad_norm": 2.1085078716278076, + "learning_rate": 4.694616328199537e-06, + "loss": 2.4750320434570314, + "step": 281240 + }, + { + "epoch": 0.7656458418402069, + "grad_norm": 2.24550724029541, + "learning_rate": 4.693525964290582e-06, + "loss": 2.458774948120117, + "step": 281260 + }, + { + "epoch": 0.7657002858309514, + "grad_norm": 2.1249430179595947, + "learning_rate": 4.692435600381628e-06, + "loss": 2.4879783630371093, + "step": 281280 + }, + { + "epoch": 0.7657547298216959, + "grad_norm": 2.1509246826171875, + "learning_rate": 4.691345236472673e-06, + "loss": 2.4646604537963865, + "step": 281300 + }, + { + "epoch": 0.7658091738124404, + "grad_norm": 2.1389899253845215, + "learning_rate": 4.690254872563719e-06, + "loss": 2.4837514877319338, + "step": 281320 + }, + { + "epoch": 0.765863617803185, + "grad_norm": 2.1405704021453857, + "learning_rate": 4.689164508654764e-06, + "loss": 2.483831787109375, + "step": 281340 + }, + { + "epoch": 0.7659180617939295, + "grad_norm": 2.177377462387085, + "learning_rate": 4.688074144745809e-06, + "loss": 2.4620237350463867, + "step": 281360 + }, + { + "epoch": 0.765972505784674, + "grad_norm": 2.115318536758423, + "learning_rate": 4.686983780836854e-06, + "loss": 2.4334953308105467, + "step": 281380 + }, + { + "epoch": 0.7660269497754185, + "grad_norm": 2.1251792907714844, + "learning_rate": 4.6858934169279e-06, + "loss": 2.4492733001708986, + "step": 281400 + }, + { + "epoch": 0.766081393766163, + "grad_norm": 2.265758514404297, + "learning_rate": 4.684803053018945e-06, + "loss": 2.463886260986328, + "step": 281420 + }, + { + "epoch": 0.7661358377569076, + "grad_norm": 2.2325632572174072, + "learning_rate": 4.683712689109991e-06, + "loss": 2.4682266235351564, + "step": 281440 + }, + { + "epoch": 0.7661902817476521, + "grad_norm": 2.3076956272125244, + "learning_rate": 4.682622325201036e-06, + "loss": 2.470530891418457, + "step": 281460 + }, + { + "epoch": 0.7662447257383966, + "grad_norm": 2.1916394233703613, + "learning_rate": 4.681531961292082e-06, + "loss": 2.438144874572754, + "step": 281480 + }, + { + "epoch": 0.7662991697291411, + "grad_norm": 2.368932008743286, + "learning_rate": 4.680441597383127e-06, + "loss": 2.4839366912841796, + "step": 281500 + }, + { + "epoch": 0.7663536137198856, + "grad_norm": 2.2261345386505127, + "learning_rate": 4.6793512334741725e-06, + "loss": 2.475635528564453, + "step": 281520 + }, + { + "epoch": 0.7664080577106301, + "grad_norm": 2.241443157196045, + "learning_rate": 4.6782608695652175e-06, + "loss": 2.455916976928711, + "step": 281540 + }, + { + "epoch": 0.7664625017013748, + "grad_norm": 2.344053030014038, + "learning_rate": 4.677170505656263e-06, + "loss": 2.479513931274414, + "step": 281560 + }, + { + "epoch": 0.7665169456921193, + "grad_norm": 2.166987180709839, + "learning_rate": 4.676080141747308e-06, + "loss": 2.476772689819336, + "step": 281580 + }, + { + "epoch": 0.7665713896828638, + "grad_norm": 2.167335271835327, + "learning_rate": 4.674989777838354e-06, + "loss": 2.4610116958618162, + "step": 281600 + }, + { + "epoch": 0.7666258336736083, + "grad_norm": 2.1527693271636963, + "learning_rate": 4.673899413929399e-06, + "loss": 2.445602607727051, + "step": 281620 + }, + { + "epoch": 0.7666802776643528, + "grad_norm": 2.0809805393218994, + "learning_rate": 4.672809050020445e-06, + "loss": 2.474208641052246, + "step": 281640 + }, + { + "epoch": 0.7667347216550974, + "grad_norm": 2.6392314434051514, + "learning_rate": 4.67171868611149e-06, + "loss": 2.461639404296875, + "step": 281660 + }, + { + "epoch": 0.7667891656458419, + "grad_norm": 2.3294193744659424, + "learning_rate": 4.670628322202536e-06, + "loss": 2.491948127746582, + "step": 281680 + }, + { + "epoch": 0.7668436096365864, + "grad_norm": 2.119185447692871, + "learning_rate": 4.6695379582935805e-06, + "loss": 2.465382766723633, + "step": 281700 + }, + { + "epoch": 0.7668980536273309, + "grad_norm": 2.1178841590881348, + "learning_rate": 4.668447594384626e-06, + "loss": 2.4594854354858398, + "step": 281720 + }, + { + "epoch": 0.7669524976180754, + "grad_norm": 2.267728328704834, + "learning_rate": 4.667357230475671e-06, + "loss": 2.4520709991455076, + "step": 281740 + }, + { + "epoch": 0.7670069416088199, + "grad_norm": 2.121157169342041, + "learning_rate": 4.666266866566717e-06, + "loss": 2.44702091217041, + "step": 281760 + }, + { + "epoch": 0.7670613855995645, + "grad_norm": 2.1744816303253174, + "learning_rate": 4.665176502657762e-06, + "loss": 2.449260139465332, + "step": 281780 + }, + { + "epoch": 0.767115829590309, + "grad_norm": 2.1115410327911377, + "learning_rate": 4.664086138748808e-06, + "loss": 2.4316577911376953, + "step": 281800 + }, + { + "epoch": 0.7671702735810535, + "grad_norm": 2.1960678100585938, + "learning_rate": 4.662995774839853e-06, + "loss": 2.4764108657836914, + "step": 281820 + }, + { + "epoch": 0.767224717571798, + "grad_norm": 2.139659881591797, + "learning_rate": 4.661905410930899e-06, + "loss": 2.4410755157470705, + "step": 281840 + }, + { + "epoch": 0.7672791615625425, + "grad_norm": 2.094282627105713, + "learning_rate": 4.660815047021944e-06, + "loss": 2.439167594909668, + "step": 281860 + }, + { + "epoch": 0.767333605553287, + "grad_norm": 2.2015140056610107, + "learning_rate": 4.6597246831129894e-06, + "loss": 2.4427799224853515, + "step": 281880 + }, + { + "epoch": 0.7673880495440316, + "grad_norm": 2.6152703762054443, + "learning_rate": 4.658634319204034e-06, + "loss": 2.4953351974487306, + "step": 281900 + }, + { + "epoch": 0.7674424935347761, + "grad_norm": 2.1795504093170166, + "learning_rate": 4.65754395529508e-06, + "loss": 2.447357749938965, + "step": 281920 + }, + { + "epoch": 0.7674969375255206, + "grad_norm": 2.3283400535583496, + "learning_rate": 4.656453591386125e-06, + "loss": 2.462725830078125, + "step": 281940 + }, + { + "epoch": 0.7675513815162651, + "grad_norm": 2.203075408935547, + "learning_rate": 4.655363227477171e-06, + "loss": 2.4339458465576174, + "step": 281960 + }, + { + "epoch": 0.7676058255070096, + "grad_norm": 2.1613621711730957, + "learning_rate": 4.654272863568216e-06, + "loss": 2.437388801574707, + "step": 281980 + }, + { + "epoch": 0.7676602694977542, + "grad_norm": 2.1498072147369385, + "learning_rate": 4.653182499659262e-06, + "loss": 2.465840148925781, + "step": 282000 + }, + { + "epoch": 0.7677147134884987, + "grad_norm": 2.3054447174072266, + "learning_rate": 4.652092135750307e-06, + "loss": 2.4432823181152346, + "step": 282020 + }, + { + "epoch": 0.7677691574792432, + "grad_norm": 2.5137717723846436, + "learning_rate": 4.6510017718413525e-06, + "loss": 2.4574247360229493, + "step": 282040 + }, + { + "epoch": 0.7678236014699877, + "grad_norm": 2.1157212257385254, + "learning_rate": 4.6499114079323975e-06, + "loss": 2.473183822631836, + "step": 282060 + }, + { + "epoch": 0.7678780454607322, + "grad_norm": 2.1736180782318115, + "learning_rate": 4.648821044023443e-06, + "loss": 2.4949787139892576, + "step": 282080 + }, + { + "epoch": 0.7679324894514767, + "grad_norm": 2.1210286617279053, + "learning_rate": 4.647730680114488e-06, + "loss": 2.4797574996948244, + "step": 282100 + }, + { + "epoch": 0.7679869334422214, + "grad_norm": 2.094676971435547, + "learning_rate": 4.646640316205534e-06, + "loss": 2.4882133483886717, + "step": 282120 + }, + { + "epoch": 0.7680413774329659, + "grad_norm": 2.239926815032959, + "learning_rate": 4.645549952296579e-06, + "loss": 2.4488162994384766, + "step": 282140 + }, + { + "epoch": 0.7680958214237104, + "grad_norm": 2.218538284301758, + "learning_rate": 4.644459588387625e-06, + "loss": 2.4781837463378906, + "step": 282160 + }, + { + "epoch": 0.7681502654144549, + "grad_norm": 2.1770269870758057, + "learning_rate": 4.64336922447867e-06, + "loss": 2.4666488647460936, + "step": 282180 + }, + { + "epoch": 0.7682047094051994, + "grad_norm": 2.197014093399048, + "learning_rate": 4.642278860569716e-06, + "loss": 2.4785224914550783, + "step": 282200 + }, + { + "epoch": 0.768259153395944, + "grad_norm": 3.144704818725586, + "learning_rate": 4.6411884966607606e-06, + "loss": 2.4945390701293944, + "step": 282220 + }, + { + "epoch": 0.7683135973866885, + "grad_norm": 2.1292295455932617, + "learning_rate": 4.640098132751806e-06, + "loss": 2.459884834289551, + "step": 282240 + }, + { + "epoch": 0.768368041377433, + "grad_norm": 2.1822943687438965, + "learning_rate": 4.639007768842851e-06, + "loss": 2.447785186767578, + "step": 282260 + }, + { + "epoch": 0.7684224853681775, + "grad_norm": 2.0709519386291504, + "learning_rate": 4.637917404933897e-06, + "loss": 2.444061279296875, + "step": 282280 + }, + { + "epoch": 0.768476929358922, + "grad_norm": 2.1859190464019775, + "learning_rate": 4.636827041024942e-06, + "loss": 2.4689849853515624, + "step": 282300 + }, + { + "epoch": 0.7685313733496665, + "grad_norm": 2.1661570072174072, + "learning_rate": 4.635736677115988e-06, + "loss": 2.4971696853637697, + "step": 282320 + }, + { + "epoch": 0.7685858173404111, + "grad_norm": 2.137336254119873, + "learning_rate": 4.634646313207033e-06, + "loss": 2.4617630004882813, + "step": 282340 + }, + { + "epoch": 0.7686402613311556, + "grad_norm": 3.0321402549743652, + "learning_rate": 4.633555949298079e-06, + "loss": 2.460610198974609, + "step": 282360 + }, + { + "epoch": 0.7686947053219001, + "grad_norm": 2.19132137298584, + "learning_rate": 4.6324655853891245e-06, + "loss": 2.457204246520996, + "step": 282380 + }, + { + "epoch": 0.7687491493126446, + "grad_norm": 2.168740749359131, + "learning_rate": 4.6313752214801695e-06, + "loss": 2.446546936035156, + "step": 282400 + }, + { + "epoch": 0.7688035933033891, + "grad_norm": 2.151362657546997, + "learning_rate": 4.630284857571215e-06, + "loss": 2.4932277679443358, + "step": 282420 + }, + { + "epoch": 0.7688580372941337, + "grad_norm": 2.271771192550659, + "learning_rate": 4.62919449366226e-06, + "loss": 2.463292694091797, + "step": 282440 + }, + { + "epoch": 0.7689124812848782, + "grad_norm": 2.1677887439727783, + "learning_rate": 4.628104129753306e-06, + "loss": 2.474192237854004, + "step": 282460 + }, + { + "epoch": 0.7689669252756227, + "grad_norm": 2.3133456707000732, + "learning_rate": 4.627013765844351e-06, + "loss": 2.4697093963623047, + "step": 282480 + }, + { + "epoch": 0.7690213692663672, + "grad_norm": 2.1100728511810303, + "learning_rate": 4.625923401935397e-06, + "loss": 2.469565010070801, + "step": 282500 + }, + { + "epoch": 0.7690758132571117, + "grad_norm": 2.174938678741455, + "learning_rate": 4.624833038026442e-06, + "loss": 2.447895812988281, + "step": 282520 + }, + { + "epoch": 0.7691302572478562, + "grad_norm": 2.246105432510376, + "learning_rate": 4.623742674117488e-06, + "loss": 2.4564056396484375, + "step": 282540 + }, + { + "epoch": 0.7691847012386008, + "grad_norm": 2.111150026321411, + "learning_rate": 4.6226523102085325e-06, + "loss": 2.456229019165039, + "step": 282560 + }, + { + "epoch": 0.7692391452293453, + "grad_norm": 2.1775224208831787, + "learning_rate": 4.621561946299578e-06, + "loss": 2.468149185180664, + "step": 282580 + }, + { + "epoch": 0.7692935892200898, + "grad_norm": 2.134185791015625, + "learning_rate": 4.620471582390623e-06, + "loss": 2.445658874511719, + "step": 282600 + }, + { + "epoch": 0.7693480332108343, + "grad_norm": 2.205016613006592, + "learning_rate": 4.619381218481669e-06, + "loss": 2.4573898315429688, + "step": 282620 + }, + { + "epoch": 0.7694024772015788, + "grad_norm": 2.210902214050293, + "learning_rate": 4.618290854572714e-06, + "loss": 2.4590421676635743, + "step": 282640 + }, + { + "epoch": 0.7694569211923235, + "grad_norm": 2.1576075553894043, + "learning_rate": 4.61720049066376e-06, + "loss": 2.437337303161621, + "step": 282660 + }, + { + "epoch": 0.769511365183068, + "grad_norm": 2.0420708656311035, + "learning_rate": 4.616110126754805e-06, + "loss": 2.464455413818359, + "step": 282680 + }, + { + "epoch": 0.7695658091738125, + "grad_norm": 2.1328446865081787, + "learning_rate": 4.615019762845851e-06, + "loss": 2.4452423095703124, + "step": 282700 + }, + { + "epoch": 0.769620253164557, + "grad_norm": 2.250861644744873, + "learning_rate": 4.613929398936896e-06, + "loss": 2.457489776611328, + "step": 282720 + }, + { + "epoch": 0.7696746971553015, + "grad_norm": 2.123886823654175, + "learning_rate": 4.6128390350279414e-06, + "loss": 2.4611190795898437, + "step": 282740 + }, + { + "epoch": 0.769729141146046, + "grad_norm": 2.238128423690796, + "learning_rate": 4.611748671118986e-06, + "loss": 2.4479196548461912, + "step": 282760 + }, + { + "epoch": 0.7697835851367906, + "grad_norm": 2.129063844680786, + "learning_rate": 4.610658307210032e-06, + "loss": 2.4202688217163084, + "step": 282780 + }, + { + "epoch": 0.7698380291275351, + "grad_norm": 2.1620326042175293, + "learning_rate": 4.609567943301077e-06, + "loss": 2.5086578369140624, + "step": 282800 + }, + { + "epoch": 0.7698924731182796, + "grad_norm": 3.6240622997283936, + "learning_rate": 4.608477579392123e-06, + "loss": 2.4570112228393555, + "step": 282820 + }, + { + "epoch": 0.7699469171090241, + "grad_norm": 2.1130707263946533, + "learning_rate": 4.607387215483168e-06, + "loss": 2.4468074798583985, + "step": 282840 + }, + { + "epoch": 0.7700013610997686, + "grad_norm": 2.1368141174316406, + "learning_rate": 4.606296851574214e-06, + "loss": 2.4216424942016603, + "step": 282860 + }, + { + "epoch": 0.7700558050905131, + "grad_norm": 2.199218988418579, + "learning_rate": 4.605206487665259e-06, + "loss": 2.4713382720947266, + "step": 282880 + }, + { + "epoch": 0.7701102490812577, + "grad_norm": 2.2204387187957764, + "learning_rate": 4.6041161237563045e-06, + "loss": 2.4887453079223634, + "step": 282900 + }, + { + "epoch": 0.7701646930720022, + "grad_norm": 2.257039785385132, + "learning_rate": 4.6030257598473495e-06, + "loss": 2.4726823806762694, + "step": 282920 + }, + { + "epoch": 0.7702191370627467, + "grad_norm": 2.217129707336426, + "learning_rate": 4.601935395938395e-06, + "loss": 2.4787221908569337, + "step": 282940 + }, + { + "epoch": 0.7702735810534912, + "grad_norm": 2.4784653186798096, + "learning_rate": 4.60084503202944e-06, + "loss": 2.4654619216918947, + "step": 282960 + }, + { + "epoch": 0.7703280250442357, + "grad_norm": 2.1293234825134277, + "learning_rate": 4.599754668120486e-06, + "loss": 2.459638977050781, + "step": 282980 + }, + { + "epoch": 0.7703824690349803, + "grad_norm": 2.2138659954071045, + "learning_rate": 4.598664304211531e-06, + "loss": 2.4405281066894533, + "step": 283000 + }, + { + "epoch": 0.7704369130257248, + "grad_norm": 2.1300740242004395, + "learning_rate": 4.597573940302577e-06, + "loss": 2.451405715942383, + "step": 283020 + }, + { + "epoch": 0.7704913570164693, + "grad_norm": 2.1096365451812744, + "learning_rate": 4.596483576393622e-06, + "loss": 2.4692798614501954, + "step": 283040 + }, + { + "epoch": 0.7705458010072138, + "grad_norm": 2.163487672805786, + "learning_rate": 4.595393212484668e-06, + "loss": 2.4695308685302733, + "step": 283060 + }, + { + "epoch": 0.7706002449979583, + "grad_norm": 2.2446517944335938, + "learning_rate": 4.594302848575713e-06, + "loss": 2.471483612060547, + "step": 283080 + }, + { + "epoch": 0.7706546889887028, + "grad_norm": 2.1514997482299805, + "learning_rate": 4.593212484666758e-06, + "loss": 2.4520387649536133, + "step": 283100 + }, + { + "epoch": 0.7707091329794474, + "grad_norm": 2.1753835678100586, + "learning_rate": 4.592122120757803e-06, + "loss": 2.4516412734985353, + "step": 283120 + }, + { + "epoch": 0.7707635769701919, + "grad_norm": 2.213170289993286, + "learning_rate": 4.591031756848849e-06, + "loss": 2.458201599121094, + "step": 283140 + }, + { + "epoch": 0.7708180209609364, + "grad_norm": 2.2815370559692383, + "learning_rate": 4.589941392939894e-06, + "loss": 2.466690254211426, + "step": 283160 + }, + { + "epoch": 0.7708724649516809, + "grad_norm": 2.1740493774414062, + "learning_rate": 4.58885102903094e-06, + "loss": 2.4755971908569334, + "step": 283180 + }, + { + "epoch": 0.7709269089424254, + "grad_norm": 2.2384815216064453, + "learning_rate": 4.587760665121985e-06, + "loss": 2.487929916381836, + "step": 283200 + }, + { + "epoch": 0.77098135293317, + "grad_norm": 2.19755482673645, + "learning_rate": 4.586670301213031e-06, + "loss": 2.4821619033813476, + "step": 283220 + }, + { + "epoch": 0.7710357969239146, + "grad_norm": 2.1489346027374268, + "learning_rate": 4.585579937304076e-06, + "loss": 2.461418342590332, + "step": 283240 + }, + { + "epoch": 0.771090240914659, + "grad_norm": 2.1275506019592285, + "learning_rate": 4.5844895733951215e-06, + "loss": 2.455596351623535, + "step": 283260 + }, + { + "epoch": 0.7711446849054036, + "grad_norm": 2.212744951248169, + "learning_rate": 4.5833992094861664e-06, + "loss": 2.469028091430664, + "step": 283280 + }, + { + "epoch": 0.7711991288961481, + "grad_norm": 2.2427759170532227, + "learning_rate": 4.582308845577212e-06, + "loss": 2.4811344146728516, + "step": 283300 + }, + { + "epoch": 0.7712535728868926, + "grad_norm": 2.060178518295288, + "learning_rate": 4.581218481668257e-06, + "loss": 2.4662437438964844, + "step": 283320 + }, + { + "epoch": 0.7713080168776372, + "grad_norm": 2.205477714538574, + "learning_rate": 4.580128117759303e-06, + "loss": 2.446382522583008, + "step": 283340 + }, + { + "epoch": 0.7713624608683817, + "grad_norm": 2.1269779205322266, + "learning_rate": 4.579037753850348e-06, + "loss": 2.4748157501220702, + "step": 283360 + }, + { + "epoch": 0.7714169048591262, + "grad_norm": 2.235485076904297, + "learning_rate": 4.577947389941394e-06, + "loss": 2.477804946899414, + "step": 283380 + }, + { + "epoch": 0.7714713488498707, + "grad_norm": 2.1925079822540283, + "learning_rate": 4.576857026032439e-06, + "loss": 2.4502349853515626, + "step": 283400 + }, + { + "epoch": 0.7715257928406152, + "grad_norm": 2.11830997467041, + "learning_rate": 4.5757666621234846e-06, + "loss": 2.494005584716797, + "step": 283420 + }, + { + "epoch": 0.7715802368313598, + "grad_norm": 2.116966962814331, + "learning_rate": 4.5746762982145295e-06, + "loss": 2.472487449645996, + "step": 283440 + }, + { + "epoch": 0.7716346808221043, + "grad_norm": 2.2576191425323486, + "learning_rate": 4.573585934305575e-06, + "loss": 2.480820655822754, + "step": 283460 + }, + { + "epoch": 0.7716891248128488, + "grad_norm": 2.1085896492004395, + "learning_rate": 4.57249557039662e-06, + "loss": 2.46026725769043, + "step": 283480 + }, + { + "epoch": 0.7717435688035933, + "grad_norm": 2.3103349208831787, + "learning_rate": 4.571405206487666e-06, + "loss": 2.443019485473633, + "step": 283500 + }, + { + "epoch": 0.7717980127943378, + "grad_norm": 2.398590564727783, + "learning_rate": 4.570314842578711e-06, + "loss": 2.459520721435547, + "step": 283520 + }, + { + "epoch": 0.7718524567850823, + "grad_norm": 2.1002025604248047, + "learning_rate": 4.569224478669757e-06, + "loss": 2.479033088684082, + "step": 283540 + }, + { + "epoch": 0.7719069007758269, + "grad_norm": 2.3850486278533936, + "learning_rate": 4.568134114760802e-06, + "loss": 2.464238929748535, + "step": 283560 + }, + { + "epoch": 0.7719613447665714, + "grad_norm": 2.1096079349517822, + "learning_rate": 4.567043750851848e-06, + "loss": 2.4772924423217773, + "step": 283580 + }, + { + "epoch": 0.7720157887573159, + "grad_norm": 2.2006170749664307, + "learning_rate": 4.565953386942893e-06, + "loss": 2.4620691299438477, + "step": 283600 + }, + { + "epoch": 0.7720702327480604, + "grad_norm": 2.1989474296569824, + "learning_rate": 4.564863023033938e-06, + "loss": 2.454282760620117, + "step": 283620 + }, + { + "epoch": 0.7721246767388049, + "grad_norm": 2.2941792011260986, + "learning_rate": 4.563772659124983e-06, + "loss": 2.4555946350097657, + "step": 283640 + }, + { + "epoch": 0.7721791207295495, + "grad_norm": 2.149545431137085, + "learning_rate": 4.562682295216029e-06, + "loss": 2.4946287155151365, + "step": 283660 + }, + { + "epoch": 0.772233564720294, + "grad_norm": 2.1509361267089844, + "learning_rate": 4.561591931307074e-06, + "loss": 2.4703330993652344, + "step": 283680 + }, + { + "epoch": 0.7722880087110385, + "grad_norm": 2.181654930114746, + "learning_rate": 4.56050156739812e-06, + "loss": 2.4852596282958985, + "step": 283700 + }, + { + "epoch": 0.772342452701783, + "grad_norm": 2.2343101501464844, + "learning_rate": 4.559411203489165e-06, + "loss": 2.4668962478637697, + "step": 283720 + }, + { + "epoch": 0.7723968966925275, + "grad_norm": 2.1737945079803467, + "learning_rate": 4.558320839580211e-06, + "loss": 2.471978950500488, + "step": 283740 + }, + { + "epoch": 0.772451340683272, + "grad_norm": 2.2176880836486816, + "learning_rate": 4.557230475671256e-06, + "loss": 2.472610092163086, + "step": 283760 + }, + { + "epoch": 0.7725057846740166, + "grad_norm": 2.3687024116516113, + "learning_rate": 4.5561401117623015e-06, + "loss": 2.4525964736938475, + "step": 283780 + }, + { + "epoch": 0.7725602286647612, + "grad_norm": 2.218982458114624, + "learning_rate": 4.5550497478533465e-06, + "loss": 2.4921865463256836, + "step": 283800 + }, + { + "epoch": 0.7726146726555057, + "grad_norm": 2.15779447555542, + "learning_rate": 4.553959383944392e-06, + "loss": 2.454944038391113, + "step": 283820 + }, + { + "epoch": 0.7726691166462502, + "grad_norm": 2.1838507652282715, + "learning_rate": 4.552869020035437e-06, + "loss": 2.4768198013305662, + "step": 283840 + }, + { + "epoch": 0.7727235606369947, + "grad_norm": 2.109570026397705, + "learning_rate": 4.551778656126482e-06, + "loss": 2.463595962524414, + "step": 283860 + }, + { + "epoch": 0.7727780046277393, + "grad_norm": 2.1934289932250977, + "learning_rate": 4.550688292217528e-06, + "loss": 2.454434394836426, + "step": 283880 + }, + { + "epoch": 0.7728324486184838, + "grad_norm": 2.2083942890167236, + "learning_rate": 4.549597928308573e-06, + "loss": 2.4600500106811523, + "step": 283900 + }, + { + "epoch": 0.7728868926092283, + "grad_norm": 2.219890594482422, + "learning_rate": 4.548507564399619e-06, + "loss": 2.4652532577514648, + "step": 283920 + }, + { + "epoch": 0.7729413365999728, + "grad_norm": 2.01924204826355, + "learning_rate": 4.547417200490664e-06, + "loss": 2.464713478088379, + "step": 283940 + }, + { + "epoch": 0.7729957805907173, + "grad_norm": 2.0885231494903564, + "learning_rate": 4.5463268365817095e-06, + "loss": 2.4948862075805662, + "step": 283960 + }, + { + "epoch": 0.7730502245814618, + "grad_norm": 2.2329721450805664, + "learning_rate": 4.5452364726727545e-06, + "loss": 2.4740068435668947, + "step": 283980 + }, + { + "epoch": 0.7731046685722064, + "grad_norm": 2.1606695652008057, + "learning_rate": 4.5441461087638e-06, + "loss": 2.4534868240356444, + "step": 284000 + }, + { + "epoch": 0.7731591125629509, + "grad_norm": 2.1396732330322266, + "learning_rate": 4.543055744854845e-06, + "loss": 2.4733314514160156, + "step": 284020 + }, + { + "epoch": 0.7732135565536954, + "grad_norm": 2.1304166316986084, + "learning_rate": 4.541965380945891e-06, + "loss": 2.4757896423339845, + "step": 284040 + }, + { + "epoch": 0.7732680005444399, + "grad_norm": 2.0729384422302246, + "learning_rate": 4.540875017036936e-06, + "loss": 2.4936756134033202, + "step": 284060 + }, + { + "epoch": 0.7733224445351844, + "grad_norm": 2.137153387069702, + "learning_rate": 4.539784653127982e-06, + "loss": 2.4772193908691404, + "step": 284080 + }, + { + "epoch": 0.7733768885259289, + "grad_norm": 2.2052948474884033, + "learning_rate": 4.538694289219027e-06, + "loss": 2.4647192001342773, + "step": 284100 + }, + { + "epoch": 0.7734313325166735, + "grad_norm": 2.269282817840576, + "learning_rate": 4.537603925310073e-06, + "loss": 2.461333084106445, + "step": 284120 + }, + { + "epoch": 0.773485776507418, + "grad_norm": 2.15901517868042, + "learning_rate": 4.536513561401118e-06, + "loss": 2.4589847564697265, + "step": 284140 + }, + { + "epoch": 0.7735402204981625, + "grad_norm": 2.1604244709014893, + "learning_rate": 4.535423197492163e-06, + "loss": 2.446041297912598, + "step": 284160 + }, + { + "epoch": 0.773594664488907, + "grad_norm": 2.156071186065674, + "learning_rate": 4.534332833583208e-06, + "loss": 2.4840667724609373, + "step": 284180 + }, + { + "epoch": 0.7736491084796515, + "grad_norm": 2.174039840698242, + "learning_rate": 4.533242469674254e-06, + "loss": 2.4687747955322266, + "step": 284200 + }, + { + "epoch": 0.7737035524703961, + "grad_norm": 2.156130313873291, + "learning_rate": 4.532152105765299e-06, + "loss": 2.462873840332031, + "step": 284220 + }, + { + "epoch": 0.7737579964611406, + "grad_norm": 2.1584928035736084, + "learning_rate": 4.531061741856345e-06, + "loss": 2.484590530395508, + "step": 284240 + }, + { + "epoch": 0.7738124404518851, + "grad_norm": 2.2571816444396973, + "learning_rate": 4.52997137794739e-06, + "loss": 2.4498313903808593, + "step": 284260 + }, + { + "epoch": 0.7738668844426296, + "grad_norm": 2.6696348190307617, + "learning_rate": 4.528881014038436e-06, + "loss": 2.4872756958007813, + "step": 284280 + }, + { + "epoch": 0.7739213284333741, + "grad_norm": 2.8204805850982666, + "learning_rate": 4.527790650129481e-06, + "loss": 2.4604475021362306, + "step": 284300 + }, + { + "epoch": 0.7739757724241186, + "grad_norm": 2.134139060974121, + "learning_rate": 4.5267002862205265e-06, + "loss": 2.494354248046875, + "step": 284320 + }, + { + "epoch": 0.7740302164148632, + "grad_norm": 2.199530839920044, + "learning_rate": 4.5256099223115715e-06, + "loss": 2.4462589263916015, + "step": 284340 + }, + { + "epoch": 0.7740846604056077, + "grad_norm": 2.2617228031158447, + "learning_rate": 4.524519558402617e-06, + "loss": 2.4887218475341797, + "step": 284360 + }, + { + "epoch": 0.7741391043963523, + "grad_norm": 2.2628915309906006, + "learning_rate": 4.523429194493662e-06, + "loss": 2.458592987060547, + "step": 284380 + }, + { + "epoch": 0.7741935483870968, + "grad_norm": 2.198223352432251, + "learning_rate": 4.522338830584708e-06, + "loss": 2.4835142135620116, + "step": 284400 + }, + { + "epoch": 0.7742479923778413, + "grad_norm": 2.19942307472229, + "learning_rate": 4.521248466675753e-06, + "loss": 2.4462738037109375, + "step": 284420 + }, + { + "epoch": 0.7743024363685859, + "grad_norm": 2.0504376888275146, + "learning_rate": 4.520158102766799e-06, + "loss": 2.453719711303711, + "step": 284440 + }, + { + "epoch": 0.7743568803593304, + "grad_norm": 2.171734571456909, + "learning_rate": 4.519067738857844e-06, + "loss": 2.4741764068603516, + "step": 284460 + }, + { + "epoch": 0.7744113243500749, + "grad_norm": 2.2319180965423584, + "learning_rate": 4.51797737494889e-06, + "loss": 2.4688846588134767, + "step": 284480 + }, + { + "epoch": 0.7744657683408194, + "grad_norm": 2.120608329772949, + "learning_rate": 4.5168870110399345e-06, + "loss": 2.4865747451782227, + "step": 284500 + }, + { + "epoch": 0.7745202123315639, + "grad_norm": 2.140383720397949, + "learning_rate": 4.51579664713098e-06, + "loss": 2.444554328918457, + "step": 284520 + }, + { + "epoch": 0.7745746563223084, + "grad_norm": 2.153264045715332, + "learning_rate": 4.514706283222025e-06, + "loss": 2.442038345336914, + "step": 284540 + }, + { + "epoch": 0.774629100313053, + "grad_norm": 2.1150782108306885, + "learning_rate": 4.513615919313071e-06, + "loss": 2.4715700149536133, + "step": 284560 + }, + { + "epoch": 0.7746835443037975, + "grad_norm": 2.1022276878356934, + "learning_rate": 4.512525555404116e-06, + "loss": 2.452084541320801, + "step": 284580 + }, + { + "epoch": 0.774737988294542, + "grad_norm": 2.1662633419036865, + "learning_rate": 4.511435191495162e-06, + "loss": 2.4813510894775392, + "step": 284600 + }, + { + "epoch": 0.7747924322852865, + "grad_norm": 2.110895872116089, + "learning_rate": 4.510344827586207e-06, + "loss": 2.4642995834350585, + "step": 284620 + }, + { + "epoch": 0.774846876276031, + "grad_norm": 2.1170003414154053, + "learning_rate": 4.509254463677253e-06, + "loss": 2.459609794616699, + "step": 284640 + }, + { + "epoch": 0.7749013202667756, + "grad_norm": 2.1796488761901855, + "learning_rate": 4.508164099768298e-06, + "loss": 2.481125259399414, + "step": 284660 + }, + { + "epoch": 0.7749557642575201, + "grad_norm": 2.1672346591949463, + "learning_rate": 4.5070737358593434e-06, + "loss": 2.4724409103393556, + "step": 284680 + }, + { + "epoch": 0.7750102082482646, + "grad_norm": 2.1933817863464355, + "learning_rate": 4.505983371950388e-06, + "loss": 2.444260025024414, + "step": 284700 + }, + { + "epoch": 0.7750646522390091, + "grad_norm": 2.191450357437134, + "learning_rate": 4.504893008041434e-06, + "loss": 2.463542175292969, + "step": 284720 + }, + { + "epoch": 0.7751190962297536, + "grad_norm": 2.8311431407928467, + "learning_rate": 4.503802644132479e-06, + "loss": 2.446404457092285, + "step": 284740 + }, + { + "epoch": 0.7751735402204981, + "grad_norm": 2.2836246490478516, + "learning_rate": 4.502712280223525e-06, + "loss": 2.4702375411987303, + "step": 284760 + }, + { + "epoch": 0.7752279842112427, + "grad_norm": 2.208383083343506, + "learning_rate": 4.50162191631457e-06, + "loss": 2.4880929946899415, + "step": 284780 + }, + { + "epoch": 0.7752824282019872, + "grad_norm": 2.1621389389038086, + "learning_rate": 4.500531552405616e-06, + "loss": 2.460036849975586, + "step": 284800 + }, + { + "epoch": 0.7753368721927317, + "grad_norm": 2.255011558532715, + "learning_rate": 4.499441188496661e-06, + "loss": 2.4436622619628907, + "step": 284820 + }, + { + "epoch": 0.7753913161834762, + "grad_norm": 2.218003034591675, + "learning_rate": 4.4983508245877065e-06, + "loss": 2.4430444717407225, + "step": 284840 + }, + { + "epoch": 0.7754457601742207, + "grad_norm": 2.2330832481384277, + "learning_rate": 4.4972604606787515e-06, + "loss": 2.463235282897949, + "step": 284860 + }, + { + "epoch": 0.7755002041649653, + "grad_norm": 2.335919141769409, + "learning_rate": 4.496170096769797e-06, + "loss": 2.4705663681030274, + "step": 284880 + }, + { + "epoch": 0.7755546481557098, + "grad_norm": 2.1986470222473145, + "learning_rate": 4.495079732860842e-06, + "loss": 2.4770336151123047, + "step": 284900 + }, + { + "epoch": 0.7756090921464543, + "grad_norm": 2.6839194297790527, + "learning_rate": 4.493989368951888e-06, + "loss": 2.444274139404297, + "step": 284920 + }, + { + "epoch": 0.7756635361371989, + "grad_norm": 2.1298367977142334, + "learning_rate": 4.492899005042933e-06, + "loss": 2.4707878112792967, + "step": 284940 + }, + { + "epoch": 0.7757179801279434, + "grad_norm": 2.1023967266082764, + "learning_rate": 4.491808641133979e-06, + "loss": 2.5064611434936523, + "step": 284960 + }, + { + "epoch": 0.7757724241186879, + "grad_norm": 2.2253644466400146, + "learning_rate": 4.490718277225024e-06, + "loss": 2.4552234649658202, + "step": 284980 + }, + { + "epoch": 0.7758268681094325, + "grad_norm": 2.374208450317383, + "learning_rate": 4.48962791331607e-06, + "loss": 2.4746334075927736, + "step": 285000 + }, + { + "epoch": 0.775881312100177, + "grad_norm": 2.217982769012451, + "learning_rate": 4.4885375494071146e-06, + "loss": 2.464767074584961, + "step": 285020 + }, + { + "epoch": 0.7759357560909215, + "grad_norm": 2.295271158218384, + "learning_rate": 4.48744718549816e-06, + "loss": 2.4844924926757814, + "step": 285040 + }, + { + "epoch": 0.775990200081666, + "grad_norm": 2.1357293128967285, + "learning_rate": 4.486356821589205e-06, + "loss": 2.4551374435424806, + "step": 285060 + }, + { + "epoch": 0.7760446440724105, + "grad_norm": 4.148131847381592, + "learning_rate": 4.485266457680251e-06, + "loss": 2.479273796081543, + "step": 285080 + }, + { + "epoch": 0.776099088063155, + "grad_norm": 2.150317668914795, + "learning_rate": 4.484176093771296e-06, + "loss": 2.493754005432129, + "step": 285100 + }, + { + "epoch": 0.7761535320538996, + "grad_norm": 2.3400988578796387, + "learning_rate": 4.483085729862342e-06, + "loss": 2.46259765625, + "step": 285120 + }, + { + "epoch": 0.7762079760446441, + "grad_norm": 2.0643274784088135, + "learning_rate": 4.481995365953387e-06, + "loss": 2.475621795654297, + "step": 285140 + }, + { + "epoch": 0.7762624200353886, + "grad_norm": 2.156977415084839, + "learning_rate": 4.480905002044433e-06, + "loss": 2.468291473388672, + "step": 285160 + }, + { + "epoch": 0.7763168640261331, + "grad_norm": 2.045701265335083, + "learning_rate": 4.479814638135478e-06, + "loss": 2.4644535064697264, + "step": 285180 + }, + { + "epoch": 0.7763713080168776, + "grad_norm": 2.165806293487549, + "learning_rate": 4.4787242742265235e-06, + "loss": 2.4762630462646484, + "step": 285200 + }, + { + "epoch": 0.7764257520076222, + "grad_norm": 2.238816022872925, + "learning_rate": 4.4776339103175684e-06, + "loss": 2.4481443405151366, + "step": 285220 + }, + { + "epoch": 0.7764801959983667, + "grad_norm": 2.150869607925415, + "learning_rate": 4.476543546408614e-06, + "loss": 2.4776906967163086, + "step": 285240 + }, + { + "epoch": 0.7765346399891112, + "grad_norm": 2.4545085430145264, + "learning_rate": 4.475453182499659e-06, + "loss": 2.484516906738281, + "step": 285260 + }, + { + "epoch": 0.7765890839798557, + "grad_norm": 2.1447365283966064, + "learning_rate": 4.474362818590705e-06, + "loss": 2.4454360961914063, + "step": 285280 + }, + { + "epoch": 0.7766435279706002, + "grad_norm": 2.1091272830963135, + "learning_rate": 4.47327245468175e-06, + "loss": 2.4650032043457033, + "step": 285300 + }, + { + "epoch": 0.7766979719613447, + "grad_norm": 2.377136707305908, + "learning_rate": 4.472182090772796e-06, + "loss": 2.485867500305176, + "step": 285320 + }, + { + "epoch": 0.7767524159520893, + "grad_norm": 2.1221718788146973, + "learning_rate": 4.471091726863841e-06, + "loss": 2.4502532958984373, + "step": 285340 + }, + { + "epoch": 0.7768068599428338, + "grad_norm": 2.262449264526367, + "learning_rate": 4.4700013629548865e-06, + "loss": 2.4456851959228514, + "step": 285360 + }, + { + "epoch": 0.7768613039335783, + "grad_norm": 2.1195671558380127, + "learning_rate": 4.4689109990459315e-06, + "loss": 2.439801788330078, + "step": 285380 + }, + { + "epoch": 0.7769157479243228, + "grad_norm": 2.1293113231658936, + "learning_rate": 4.467820635136977e-06, + "loss": 2.451975440979004, + "step": 285400 + }, + { + "epoch": 0.7769701919150673, + "grad_norm": 2.092716932296753, + "learning_rate": 4.466730271228022e-06, + "loss": 2.484000015258789, + "step": 285420 + }, + { + "epoch": 0.777024635905812, + "grad_norm": 2.2179112434387207, + "learning_rate": 4.465639907319068e-06, + "loss": 2.449155807495117, + "step": 285440 + }, + { + "epoch": 0.7770790798965564, + "grad_norm": 2.138514757156372, + "learning_rate": 4.464549543410113e-06, + "loss": 2.5010488510131834, + "step": 285460 + }, + { + "epoch": 0.777133523887301, + "grad_norm": 2.117729663848877, + "learning_rate": 4.463459179501159e-06, + "loss": 2.4485816955566406, + "step": 285480 + }, + { + "epoch": 0.7771879678780454, + "grad_norm": 2.0893516540527344, + "learning_rate": 4.462368815592204e-06, + "loss": 2.4542724609375, + "step": 285500 + }, + { + "epoch": 0.77724241186879, + "grad_norm": 2.197559356689453, + "learning_rate": 4.46127845168325e-06, + "loss": 2.443451499938965, + "step": 285520 + }, + { + "epoch": 0.7772968558595345, + "grad_norm": 2.160212516784668, + "learning_rate": 4.460188087774295e-06, + "loss": 2.466639518737793, + "step": 285540 + }, + { + "epoch": 0.7773512998502791, + "grad_norm": 2.22768497467041, + "learning_rate": 4.45909772386534e-06, + "loss": 2.4879520416259764, + "step": 285560 + }, + { + "epoch": 0.7774057438410236, + "grad_norm": 2.2278568744659424, + "learning_rate": 4.458007359956385e-06, + "loss": 2.4573272705078124, + "step": 285580 + }, + { + "epoch": 0.7774601878317681, + "grad_norm": 2.258925676345825, + "learning_rate": 4.456916996047431e-06, + "loss": 2.449568748474121, + "step": 285600 + }, + { + "epoch": 0.7775146318225126, + "grad_norm": 2.197702407836914, + "learning_rate": 4.455826632138476e-06, + "loss": 2.488836669921875, + "step": 285620 + }, + { + "epoch": 0.7775690758132571, + "grad_norm": 2.6492552757263184, + "learning_rate": 4.454736268229522e-06, + "loss": 2.453011894226074, + "step": 285640 + }, + { + "epoch": 0.7776235198040017, + "grad_norm": 2.1715946197509766, + "learning_rate": 4.453645904320567e-06, + "loss": 2.46215877532959, + "step": 285660 + }, + { + "epoch": 0.7776779637947462, + "grad_norm": 2.198124647140503, + "learning_rate": 4.452555540411613e-06, + "loss": 2.4993499755859374, + "step": 285680 + }, + { + "epoch": 0.7777324077854907, + "grad_norm": 2.108022928237915, + "learning_rate": 4.451465176502658e-06, + "loss": 2.4507728576660157, + "step": 285700 + }, + { + "epoch": 0.7777868517762352, + "grad_norm": 2.2955150604248047, + "learning_rate": 4.4503748125937035e-06, + "loss": 2.4509159088134767, + "step": 285720 + }, + { + "epoch": 0.7778412957669797, + "grad_norm": 2.277409076690674, + "learning_rate": 4.4492844486847485e-06, + "loss": 2.468332290649414, + "step": 285740 + }, + { + "epoch": 0.7778957397577242, + "grad_norm": 2.1237711906433105, + "learning_rate": 4.448194084775794e-06, + "loss": 2.474851608276367, + "step": 285760 + }, + { + "epoch": 0.7779501837484688, + "grad_norm": 2.1429097652435303, + "learning_rate": 4.44710372086684e-06, + "loss": 2.4747507095336916, + "step": 285780 + }, + { + "epoch": 0.7780046277392133, + "grad_norm": 2.288626194000244, + "learning_rate": 4.446013356957885e-06, + "loss": 2.4249263763427735, + "step": 285800 + }, + { + "epoch": 0.7780590717299578, + "grad_norm": 2.418513536453247, + "learning_rate": 4.444922993048931e-06, + "loss": 2.473281478881836, + "step": 285820 + }, + { + "epoch": 0.7781135157207023, + "grad_norm": 2.2944438457489014, + "learning_rate": 4.443832629139976e-06, + "loss": 2.4833820343017576, + "step": 285840 + }, + { + "epoch": 0.7781679597114468, + "grad_norm": 2.2087690830230713, + "learning_rate": 4.442742265231022e-06, + "loss": 2.4498008728027343, + "step": 285860 + }, + { + "epoch": 0.7782224037021914, + "grad_norm": 2.1407604217529297, + "learning_rate": 4.4416519013220666e-06, + "loss": 2.4472982406616213, + "step": 285880 + }, + { + "epoch": 0.7782768476929359, + "grad_norm": 2.2454564571380615, + "learning_rate": 4.440561537413112e-06, + "loss": 2.461353302001953, + "step": 285900 + }, + { + "epoch": 0.7783312916836804, + "grad_norm": 2.2689621448516846, + "learning_rate": 4.439471173504157e-06, + "loss": 2.4336275100708007, + "step": 285920 + }, + { + "epoch": 0.7783857356744249, + "grad_norm": 2.1662397384643555, + "learning_rate": 4.438380809595203e-06, + "loss": 2.4590688705444337, + "step": 285940 + }, + { + "epoch": 0.7784401796651694, + "grad_norm": 2.2416574954986572, + "learning_rate": 4.437290445686248e-06, + "loss": 2.456320571899414, + "step": 285960 + }, + { + "epoch": 0.7784946236559139, + "grad_norm": 2.165201187133789, + "learning_rate": 4.436200081777294e-06, + "loss": 2.462750244140625, + "step": 285980 + }, + { + "epoch": 0.7785490676466585, + "grad_norm": 2.1235947608947754, + "learning_rate": 4.435109717868339e-06, + "loss": 2.461822509765625, + "step": 286000 + }, + { + "epoch": 0.778603511637403, + "grad_norm": 2.2134900093078613, + "learning_rate": 4.434019353959385e-06, + "loss": 2.496587562561035, + "step": 286020 + }, + { + "epoch": 0.7786579556281475, + "grad_norm": 2.1569104194641113, + "learning_rate": 4.43292899005043e-06, + "loss": 2.4426177978515624, + "step": 286040 + }, + { + "epoch": 0.778712399618892, + "grad_norm": 2.0916640758514404, + "learning_rate": 4.4318386261414755e-06, + "loss": 2.477397155761719, + "step": 286060 + }, + { + "epoch": 0.7787668436096365, + "grad_norm": 2.163520574569702, + "learning_rate": 4.4307482622325204e-06, + "loss": 2.4762041091918947, + "step": 286080 + }, + { + "epoch": 0.7788212876003812, + "grad_norm": 2.2038886547088623, + "learning_rate": 4.429657898323566e-06, + "loss": 2.467477226257324, + "step": 286100 + }, + { + "epoch": 0.7788757315911257, + "grad_norm": 2.01656436920166, + "learning_rate": 4.428567534414611e-06, + "loss": 2.4900569915771484, + "step": 286120 + }, + { + "epoch": 0.7789301755818702, + "grad_norm": 2.163644552230835, + "learning_rate": 4.427477170505657e-06, + "loss": 2.451234245300293, + "step": 286140 + }, + { + "epoch": 0.7789846195726147, + "grad_norm": 2.284881114959717, + "learning_rate": 4.426386806596702e-06, + "loss": 2.4619531631469727, + "step": 286160 + }, + { + "epoch": 0.7790390635633592, + "grad_norm": 2.167201519012451, + "learning_rate": 4.425296442687748e-06, + "loss": 2.4546607971191405, + "step": 286180 + }, + { + "epoch": 0.7790935075541037, + "grad_norm": 2.0606961250305176, + "learning_rate": 4.424206078778793e-06, + "loss": 2.4656116485595705, + "step": 286200 + }, + { + "epoch": 0.7791479515448483, + "grad_norm": 2.251702308654785, + "learning_rate": 4.4231157148698386e-06, + "loss": 2.4525802612304686, + "step": 286220 + }, + { + "epoch": 0.7792023955355928, + "grad_norm": 2.345322847366333, + "learning_rate": 4.4220253509608835e-06, + "loss": 2.4421928405761717, + "step": 286240 + }, + { + "epoch": 0.7792568395263373, + "grad_norm": 2.816236734390259, + "learning_rate": 4.420934987051929e-06, + "loss": 2.4570098876953126, + "step": 286260 + }, + { + "epoch": 0.7793112835170818, + "grad_norm": 2.166260242462158, + "learning_rate": 4.419844623142974e-06, + "loss": 2.45389404296875, + "step": 286280 + }, + { + "epoch": 0.7793657275078263, + "grad_norm": 2.1784636974334717, + "learning_rate": 4.41875425923402e-06, + "loss": 2.457464027404785, + "step": 286300 + }, + { + "epoch": 0.7794201714985708, + "grad_norm": 2.109041452407837, + "learning_rate": 4.417663895325065e-06, + "loss": 2.468497467041016, + "step": 286320 + }, + { + "epoch": 0.7794746154893154, + "grad_norm": 2.6839282512664795, + "learning_rate": 4.416573531416111e-06, + "loss": 2.4434127807617188, + "step": 286340 + }, + { + "epoch": 0.7795290594800599, + "grad_norm": 2.129361152648926, + "learning_rate": 4.415483167507156e-06, + "loss": 2.4537673950195313, + "step": 286360 + }, + { + "epoch": 0.7795835034708044, + "grad_norm": 2.220675230026245, + "learning_rate": 4.414392803598202e-06, + "loss": 2.456338310241699, + "step": 286380 + }, + { + "epoch": 0.7796379474615489, + "grad_norm": 2.118924379348755, + "learning_rate": 4.413302439689247e-06, + "loss": 2.4722705841064454, + "step": 286400 + }, + { + "epoch": 0.7796923914522934, + "grad_norm": 2.229717254638672, + "learning_rate": 4.412212075780292e-06, + "loss": 2.4731042861938475, + "step": 286420 + }, + { + "epoch": 0.779746835443038, + "grad_norm": 2.1074347496032715, + "learning_rate": 4.411121711871337e-06, + "loss": 2.4503511428833007, + "step": 286440 + }, + { + "epoch": 0.7798012794337825, + "grad_norm": 2.190943479537964, + "learning_rate": 4.410031347962383e-06, + "loss": 2.448336601257324, + "step": 286460 + }, + { + "epoch": 0.779855723424527, + "grad_norm": 2.7572481632232666, + "learning_rate": 4.408940984053428e-06, + "loss": 2.4662452697753907, + "step": 286480 + }, + { + "epoch": 0.7799101674152715, + "grad_norm": 2.4195592403411865, + "learning_rate": 4.407850620144474e-06, + "loss": 2.4524866104125977, + "step": 286500 + }, + { + "epoch": 0.779964611406016, + "grad_norm": 2.1407554149627686, + "learning_rate": 4.406760256235519e-06, + "loss": 2.445451354980469, + "step": 286520 + }, + { + "epoch": 0.7800190553967605, + "grad_norm": 2.117828607559204, + "learning_rate": 4.405669892326565e-06, + "loss": 2.467841911315918, + "step": 286540 + }, + { + "epoch": 0.7800734993875051, + "grad_norm": 2.242096424102783, + "learning_rate": 4.40457952841761e-06, + "loss": 2.4896726608276367, + "step": 286560 + }, + { + "epoch": 0.7801279433782496, + "grad_norm": 2.1333060264587402, + "learning_rate": 4.4034891645086555e-06, + "loss": 2.493269920349121, + "step": 286580 + }, + { + "epoch": 0.7801823873689941, + "grad_norm": 2.1451168060302734, + "learning_rate": 4.4023988005997005e-06, + "loss": 2.4760015487670897, + "step": 286600 + }, + { + "epoch": 0.7802368313597386, + "grad_norm": 2.2029147148132324, + "learning_rate": 4.401308436690746e-06, + "loss": 2.447871208190918, + "step": 286620 + }, + { + "epoch": 0.7802912753504831, + "grad_norm": 2.2636232376098633, + "learning_rate": 4.400218072781791e-06, + "loss": 2.460087776184082, + "step": 286640 + }, + { + "epoch": 0.7803457193412278, + "grad_norm": 2.2529022693634033, + "learning_rate": 4.399127708872837e-06, + "loss": 2.4660778045654297, + "step": 286660 + }, + { + "epoch": 0.7804001633319723, + "grad_norm": 2.1101512908935547, + "learning_rate": 4.398037344963882e-06, + "loss": 2.4516944885253906, + "step": 286680 + }, + { + "epoch": 0.7804546073227168, + "grad_norm": 2.135556221008301, + "learning_rate": 4.396946981054928e-06, + "loss": 2.4619878768920898, + "step": 286700 + }, + { + "epoch": 0.7805090513134613, + "grad_norm": 2.039868116378784, + "learning_rate": 4.395856617145973e-06, + "loss": 2.466480255126953, + "step": 286720 + }, + { + "epoch": 0.7805634953042058, + "grad_norm": 2.237130880355835, + "learning_rate": 4.394766253237019e-06, + "loss": 2.470405387878418, + "step": 286740 + }, + { + "epoch": 0.7806179392949503, + "grad_norm": 2.1146881580352783, + "learning_rate": 4.3936758893280635e-06, + "loss": 2.4406002044677733, + "step": 286760 + }, + { + "epoch": 0.7806723832856949, + "grad_norm": 2.1979496479034424, + "learning_rate": 4.392585525419109e-06, + "loss": 2.4687946319580076, + "step": 286780 + }, + { + "epoch": 0.7807268272764394, + "grad_norm": 2.1798722743988037, + "learning_rate": 4.391495161510154e-06, + "loss": 2.4710103988647463, + "step": 286800 + }, + { + "epoch": 0.7807812712671839, + "grad_norm": 2.3195581436157227, + "learning_rate": 4.3904047976012e-06, + "loss": 2.4685361862182615, + "step": 286820 + }, + { + "epoch": 0.7808357152579284, + "grad_norm": 2.1855132579803467, + "learning_rate": 4.389314433692245e-06, + "loss": 2.5025978088378906, + "step": 286840 + }, + { + "epoch": 0.7808901592486729, + "grad_norm": 2.2681238651275635, + "learning_rate": 4.388224069783291e-06, + "loss": 2.4802244186401365, + "step": 286860 + }, + { + "epoch": 0.7809446032394175, + "grad_norm": 2.208434820175171, + "learning_rate": 4.387133705874336e-06, + "loss": 2.4536205291748048, + "step": 286880 + }, + { + "epoch": 0.780999047230162, + "grad_norm": 2.171483039855957, + "learning_rate": 4.386043341965382e-06, + "loss": 2.4746444702148436, + "step": 286900 + }, + { + "epoch": 0.7810534912209065, + "grad_norm": 2.3027191162109375, + "learning_rate": 4.384952978056427e-06, + "loss": 2.470471000671387, + "step": 286920 + }, + { + "epoch": 0.781107935211651, + "grad_norm": 2.2104856967926025, + "learning_rate": 4.3838626141474724e-06, + "loss": 2.441546630859375, + "step": 286940 + }, + { + "epoch": 0.7811623792023955, + "grad_norm": 2.25447416305542, + "learning_rate": 4.382772250238517e-06, + "loss": 2.439841461181641, + "step": 286960 + }, + { + "epoch": 0.78121682319314, + "grad_norm": 2.2409846782684326, + "learning_rate": 4.381681886329563e-06, + "loss": 2.477861213684082, + "step": 286980 + }, + { + "epoch": 0.7812712671838846, + "grad_norm": 2.2881057262420654, + "learning_rate": 4.380591522420608e-06, + "loss": 2.4687267303466798, + "step": 287000 + }, + { + "epoch": 0.7813257111746291, + "grad_norm": 2.1605608463287354, + "learning_rate": 4.379501158511654e-06, + "loss": 2.517867088317871, + "step": 287020 + }, + { + "epoch": 0.7813801551653736, + "grad_norm": 2.142336845397949, + "learning_rate": 4.378410794602699e-06, + "loss": 2.43564510345459, + "step": 287040 + }, + { + "epoch": 0.7814345991561181, + "grad_norm": 2.095888137817383, + "learning_rate": 4.377320430693745e-06, + "loss": 2.471352958679199, + "step": 287060 + }, + { + "epoch": 0.7814890431468626, + "grad_norm": 2.210846424102783, + "learning_rate": 4.37623006678479e-06, + "loss": 2.477796745300293, + "step": 287080 + }, + { + "epoch": 0.7815434871376072, + "grad_norm": 2.163764476776123, + "learning_rate": 4.3751397028758355e-06, + "loss": 2.481192398071289, + "step": 287100 + }, + { + "epoch": 0.7815979311283517, + "grad_norm": 2.132892370223999, + "learning_rate": 4.3740493389668805e-06, + "loss": 2.440433120727539, + "step": 287120 + }, + { + "epoch": 0.7816523751190962, + "grad_norm": 2.1271071434020996, + "learning_rate": 4.372958975057926e-06, + "loss": 2.48265380859375, + "step": 287140 + }, + { + "epoch": 0.7817068191098407, + "grad_norm": 2.498258590698242, + "learning_rate": 4.371868611148971e-06, + "loss": 2.4674726486206056, + "step": 287160 + }, + { + "epoch": 0.7817612631005852, + "grad_norm": 2.284119129180908, + "learning_rate": 4.370778247240017e-06, + "loss": 2.4724746704101563, + "step": 287180 + }, + { + "epoch": 0.7818157070913297, + "grad_norm": 2.256667375564575, + "learning_rate": 4.369687883331062e-06, + "loss": 2.500177574157715, + "step": 287200 + }, + { + "epoch": 0.7818701510820744, + "grad_norm": 2.145491600036621, + "learning_rate": 4.368597519422108e-06, + "loss": 2.418486785888672, + "step": 287220 + }, + { + "epoch": 0.7819245950728189, + "grad_norm": 2.2851336002349854, + "learning_rate": 4.367507155513153e-06, + "loss": 2.4700971603393556, + "step": 287240 + }, + { + "epoch": 0.7819790390635634, + "grad_norm": 2.157041072845459, + "learning_rate": 4.366416791604199e-06, + "loss": 2.437362861633301, + "step": 287260 + }, + { + "epoch": 0.7820334830543079, + "grad_norm": 2.136422872543335, + "learning_rate": 4.3653264276952436e-06, + "loss": 2.416800880432129, + "step": 287280 + }, + { + "epoch": 0.7820879270450524, + "grad_norm": 2.189206600189209, + "learning_rate": 4.364236063786289e-06, + "loss": 2.4864322662353517, + "step": 287300 + }, + { + "epoch": 0.782142371035797, + "grad_norm": 2.154348373413086, + "learning_rate": 4.363145699877334e-06, + "loss": 2.4623687744140623, + "step": 287320 + }, + { + "epoch": 0.7821968150265415, + "grad_norm": 2.9358983039855957, + "learning_rate": 4.36205533596838e-06, + "loss": 2.4627973556518556, + "step": 287340 + }, + { + "epoch": 0.782251259017286, + "grad_norm": 2.356473207473755, + "learning_rate": 4.360964972059425e-06, + "loss": 2.4526845932006838, + "step": 287360 + }, + { + "epoch": 0.7823057030080305, + "grad_norm": 2.2350666522979736, + "learning_rate": 4.359874608150471e-06, + "loss": 2.446418571472168, + "step": 287380 + }, + { + "epoch": 0.782360146998775, + "grad_norm": 2.2753472328186035, + "learning_rate": 4.358784244241516e-06, + "loss": 2.4743932723999023, + "step": 287400 + }, + { + "epoch": 0.7824145909895195, + "grad_norm": 2.1851770877838135, + "learning_rate": 4.357693880332562e-06, + "loss": 2.4364667892456056, + "step": 287420 + }, + { + "epoch": 0.7824690349802641, + "grad_norm": 2.2460360527038574, + "learning_rate": 4.356603516423607e-06, + "loss": 2.420404052734375, + "step": 287440 + }, + { + "epoch": 0.7825234789710086, + "grad_norm": 2.2071821689605713, + "learning_rate": 4.3555131525146525e-06, + "loss": 2.468459129333496, + "step": 287460 + }, + { + "epoch": 0.7825779229617531, + "grad_norm": 2.2507333755493164, + "learning_rate": 4.3544227886056974e-06, + "loss": 2.465528678894043, + "step": 287480 + }, + { + "epoch": 0.7826323669524976, + "grad_norm": 2.2468459606170654, + "learning_rate": 4.353332424696743e-06, + "loss": 2.4656883239746095, + "step": 287500 + }, + { + "epoch": 0.7826868109432421, + "grad_norm": 2.185741424560547, + "learning_rate": 4.352242060787788e-06, + "loss": 2.486988067626953, + "step": 287520 + }, + { + "epoch": 0.7827412549339866, + "grad_norm": 2.6842269897460938, + "learning_rate": 4.351151696878834e-06, + "loss": 2.445750045776367, + "step": 287540 + }, + { + "epoch": 0.7827956989247312, + "grad_norm": 2.306513786315918, + "learning_rate": 4.350061332969879e-06, + "loss": 2.4821002960205076, + "step": 287560 + }, + { + "epoch": 0.7828501429154757, + "grad_norm": 2.280068874359131, + "learning_rate": 4.348970969060925e-06, + "loss": 2.4913812637329102, + "step": 287580 + }, + { + "epoch": 0.7829045869062202, + "grad_norm": 2.1636643409729004, + "learning_rate": 4.34788060515197e-06, + "loss": 2.4630136489868164, + "step": 287600 + }, + { + "epoch": 0.7829590308969647, + "grad_norm": 2.350600004196167, + "learning_rate": 4.3467902412430156e-06, + "loss": 2.4703418731689455, + "step": 287620 + }, + { + "epoch": 0.7830134748877092, + "grad_norm": 2.170149326324463, + "learning_rate": 4.3456998773340605e-06, + "loss": 2.4781492233276365, + "step": 287640 + }, + { + "epoch": 0.7830679188784538, + "grad_norm": 2.099271774291992, + "learning_rate": 4.344609513425106e-06, + "loss": 2.4320383071899414, + "step": 287660 + }, + { + "epoch": 0.7831223628691983, + "grad_norm": 2.13574481010437, + "learning_rate": 4.343519149516151e-06, + "loss": 2.447528076171875, + "step": 287680 + }, + { + "epoch": 0.7831768068599428, + "grad_norm": 2.235912799835205, + "learning_rate": 4.342428785607197e-06, + "loss": 2.464247703552246, + "step": 287700 + }, + { + "epoch": 0.7832312508506873, + "grad_norm": 2.2158801555633545, + "learning_rate": 4.341338421698242e-06, + "loss": 2.4624691009521484, + "step": 287720 + }, + { + "epoch": 0.7832856948414318, + "grad_norm": 2.259174108505249, + "learning_rate": 4.340248057789288e-06, + "loss": 2.48431282043457, + "step": 287740 + }, + { + "epoch": 0.7833401388321763, + "grad_norm": 2.1590936183929443, + "learning_rate": 4.339157693880333e-06, + "loss": 2.485108184814453, + "step": 287760 + }, + { + "epoch": 0.783394582822921, + "grad_norm": 2.1895670890808105, + "learning_rate": 4.338067329971379e-06, + "loss": 2.4735538482666017, + "step": 287780 + }, + { + "epoch": 0.7834490268136655, + "grad_norm": 2.1212868690490723, + "learning_rate": 4.336976966062424e-06, + "loss": 2.47717399597168, + "step": 287800 + }, + { + "epoch": 0.78350347080441, + "grad_norm": 2.177624225616455, + "learning_rate": 4.335886602153469e-06, + "loss": 2.476218414306641, + "step": 287820 + }, + { + "epoch": 0.7835579147951545, + "grad_norm": 2.11885404586792, + "learning_rate": 4.334796238244514e-06, + "loss": 2.4827764511108397, + "step": 287840 + }, + { + "epoch": 0.783612358785899, + "grad_norm": 2.1155765056610107, + "learning_rate": 4.33370587433556e-06, + "loss": 2.4866300582885743, + "step": 287860 + }, + { + "epoch": 0.7836668027766436, + "grad_norm": 2.1459217071533203, + "learning_rate": 4.332615510426605e-06, + "loss": 2.4702049255371095, + "step": 287880 + }, + { + "epoch": 0.7837212467673881, + "grad_norm": 2.12821364402771, + "learning_rate": 4.331525146517651e-06, + "loss": 2.4577653884887694, + "step": 287900 + }, + { + "epoch": 0.7837756907581326, + "grad_norm": 2.1399343013763428, + "learning_rate": 4.330434782608696e-06, + "loss": 2.499367332458496, + "step": 287920 + }, + { + "epoch": 0.7838301347488771, + "grad_norm": 2.242231845855713, + "learning_rate": 4.329344418699742e-06, + "loss": 2.497626304626465, + "step": 287940 + }, + { + "epoch": 0.7838845787396216, + "grad_norm": 2.187558650970459, + "learning_rate": 4.328254054790787e-06, + "loss": 2.4612100601196287, + "step": 287960 + }, + { + "epoch": 0.7839390227303661, + "grad_norm": 2.1769368648529053, + "learning_rate": 4.3271636908818325e-06, + "loss": 2.4376338958740233, + "step": 287980 + }, + { + "epoch": 0.7839934667211107, + "grad_norm": 2.306795597076416, + "learning_rate": 4.3260733269728775e-06, + "loss": 2.4736745834350584, + "step": 288000 + }, + { + "epoch": 0.7840479107118552, + "grad_norm": 2.052888870239258, + "learning_rate": 4.324982963063923e-06, + "loss": 2.4298656463623045, + "step": 288020 + }, + { + "epoch": 0.7841023547025997, + "grad_norm": 2.2157487869262695, + "learning_rate": 4.323892599154968e-06, + "loss": 2.4771013259887695, + "step": 288040 + }, + { + "epoch": 0.7841567986933442, + "grad_norm": 3.0107498168945312, + "learning_rate": 4.322802235246013e-06, + "loss": 2.4747089385986327, + "step": 288060 + }, + { + "epoch": 0.7842112426840887, + "grad_norm": 2.3985681533813477, + "learning_rate": 4.321711871337059e-06, + "loss": 2.477919006347656, + "step": 288080 + }, + { + "epoch": 0.7842656866748333, + "grad_norm": 2.2422921657562256, + "learning_rate": 4.320621507428104e-06, + "loss": 2.4745712280273438, + "step": 288100 + }, + { + "epoch": 0.7843201306655778, + "grad_norm": 2.1118884086608887, + "learning_rate": 4.31953114351915e-06, + "loss": 2.469796371459961, + "step": 288120 + }, + { + "epoch": 0.7843745746563223, + "grad_norm": 2.1495361328125, + "learning_rate": 4.318440779610195e-06, + "loss": 2.461039924621582, + "step": 288140 + }, + { + "epoch": 0.7844290186470668, + "grad_norm": 2.1948723793029785, + "learning_rate": 4.3173504157012405e-06, + "loss": 2.4642383575439455, + "step": 288160 + }, + { + "epoch": 0.7844834626378113, + "grad_norm": 2.1283600330352783, + "learning_rate": 4.3162600517922855e-06, + "loss": 2.440486717224121, + "step": 288180 + }, + { + "epoch": 0.7845379066285558, + "grad_norm": 2.0856919288635254, + "learning_rate": 4.315169687883331e-06, + "loss": 2.4565505981445312, + "step": 288200 + }, + { + "epoch": 0.7845923506193004, + "grad_norm": 2.098461389541626, + "learning_rate": 4.314079323974376e-06, + "loss": 2.4830684661865234, + "step": 288220 + }, + { + "epoch": 0.7846467946100449, + "grad_norm": 2.086543083190918, + "learning_rate": 4.312988960065422e-06, + "loss": 2.4734159469604493, + "step": 288240 + }, + { + "epoch": 0.7847012386007894, + "grad_norm": 2.4266936779022217, + "learning_rate": 4.311898596156467e-06, + "loss": 2.4725074768066406, + "step": 288260 + }, + { + "epoch": 0.7847556825915339, + "grad_norm": 2.150494337081909, + "learning_rate": 4.310808232247513e-06, + "loss": 2.4631925582885743, + "step": 288280 + }, + { + "epoch": 0.7848101265822784, + "grad_norm": 2.087376356124878, + "learning_rate": 4.309717868338558e-06, + "loss": 2.4570878982543944, + "step": 288300 + }, + { + "epoch": 0.7848645705730231, + "grad_norm": 2.135113000869751, + "learning_rate": 4.308627504429604e-06, + "loss": 2.4616573333740233, + "step": 288320 + }, + { + "epoch": 0.7849190145637676, + "grad_norm": 2.0858161449432373, + "learning_rate": 4.307537140520649e-06, + "loss": 2.475284194946289, + "step": 288340 + }, + { + "epoch": 0.7849734585545121, + "grad_norm": 2.2079410552978516, + "learning_rate": 4.306446776611694e-06, + "loss": 2.44442081451416, + "step": 288360 + }, + { + "epoch": 0.7850279025452566, + "grad_norm": 2.080451488494873, + "learning_rate": 4.305356412702739e-06, + "loss": 2.4530256271362303, + "step": 288380 + }, + { + "epoch": 0.7850823465360011, + "grad_norm": 2.1450936794281006, + "learning_rate": 4.304266048793785e-06, + "loss": 2.469155120849609, + "step": 288400 + }, + { + "epoch": 0.7851367905267456, + "grad_norm": 2.2154150009155273, + "learning_rate": 4.30317568488483e-06, + "loss": 2.4815155029296876, + "step": 288420 + }, + { + "epoch": 0.7851912345174902, + "grad_norm": 2.176356315612793, + "learning_rate": 4.302085320975876e-06, + "loss": 2.4730443954467773, + "step": 288440 + }, + { + "epoch": 0.7852456785082347, + "grad_norm": 2.340212821960449, + "learning_rate": 4.300994957066921e-06, + "loss": 2.4547027587890624, + "step": 288460 + }, + { + "epoch": 0.7853001224989792, + "grad_norm": 2.216135263442993, + "learning_rate": 4.299904593157967e-06, + "loss": 2.470402145385742, + "step": 288480 + }, + { + "epoch": 0.7853545664897237, + "grad_norm": 2.0822672843933105, + "learning_rate": 4.298814229249012e-06, + "loss": 2.444148635864258, + "step": 288500 + }, + { + "epoch": 0.7854090104804682, + "grad_norm": 2.1940605640411377, + "learning_rate": 4.2977238653400575e-06, + "loss": 2.4589662551879883, + "step": 288520 + }, + { + "epoch": 0.7854634544712127, + "grad_norm": 2.2052814960479736, + "learning_rate": 4.2966335014311024e-06, + "loss": 2.49671630859375, + "step": 288540 + }, + { + "epoch": 0.7855178984619573, + "grad_norm": 2.134781837463379, + "learning_rate": 4.295543137522148e-06, + "loss": 2.4544628143310545, + "step": 288560 + }, + { + "epoch": 0.7855723424527018, + "grad_norm": 2.2015862464904785, + "learning_rate": 4.294452773613193e-06, + "loss": 2.473946189880371, + "step": 288580 + }, + { + "epoch": 0.7856267864434463, + "grad_norm": 2.179047107696533, + "learning_rate": 4.293362409704239e-06, + "loss": 2.4770391464233397, + "step": 288600 + }, + { + "epoch": 0.7856812304341908, + "grad_norm": 2.125819444656372, + "learning_rate": 4.292272045795284e-06, + "loss": 2.476253128051758, + "step": 288620 + }, + { + "epoch": 0.7857356744249353, + "grad_norm": 2.3314883708953857, + "learning_rate": 4.29118168188633e-06, + "loss": 2.4673295974731446, + "step": 288640 + }, + { + "epoch": 0.7857901184156799, + "grad_norm": 2.2181999683380127, + "learning_rate": 4.290091317977375e-06, + "loss": 2.453556251525879, + "step": 288660 + }, + { + "epoch": 0.7858445624064244, + "grad_norm": 2.4071455001831055, + "learning_rate": 4.2890009540684206e-06, + "loss": 2.483364295959473, + "step": 288680 + }, + { + "epoch": 0.7858990063971689, + "grad_norm": 2.249704360961914, + "learning_rate": 4.2879105901594655e-06, + "loss": 2.461797523498535, + "step": 288700 + }, + { + "epoch": 0.7859534503879134, + "grad_norm": 2.187087297439575, + "learning_rate": 4.286820226250511e-06, + "loss": 2.424793815612793, + "step": 288720 + }, + { + "epoch": 0.7860078943786579, + "grad_norm": 2.1235697269439697, + "learning_rate": 4.285729862341556e-06, + "loss": 2.461003875732422, + "step": 288740 + }, + { + "epoch": 0.7860623383694024, + "grad_norm": 2.129319906234741, + "learning_rate": 4.284639498432602e-06, + "loss": 2.4631072998046877, + "step": 288760 + }, + { + "epoch": 0.786116782360147, + "grad_norm": 2.1540164947509766, + "learning_rate": 4.283549134523647e-06, + "loss": 2.468946838378906, + "step": 288780 + }, + { + "epoch": 0.7861712263508915, + "grad_norm": 2.5318377017974854, + "learning_rate": 4.282458770614693e-06, + "loss": 2.461598587036133, + "step": 288800 + }, + { + "epoch": 0.786225670341636, + "grad_norm": 2.208415985107422, + "learning_rate": 4.281368406705738e-06, + "loss": 2.4574661254882812, + "step": 288820 + }, + { + "epoch": 0.7862801143323805, + "grad_norm": 2.100050926208496, + "learning_rate": 4.280278042796784e-06, + "loss": 2.4428329467773438, + "step": 288840 + }, + { + "epoch": 0.786334558323125, + "grad_norm": 2.12251353263855, + "learning_rate": 4.279187678887829e-06, + "loss": 2.444896697998047, + "step": 288860 + }, + { + "epoch": 0.7863890023138697, + "grad_norm": 2.2451395988464355, + "learning_rate": 4.2780973149788744e-06, + "loss": 2.447504234313965, + "step": 288880 + }, + { + "epoch": 0.7864434463046142, + "grad_norm": 2.1492695808410645, + "learning_rate": 4.277006951069919e-06, + "loss": 2.4596269607543944, + "step": 288900 + }, + { + "epoch": 0.7864978902953587, + "grad_norm": 2.1455562114715576, + "learning_rate": 4.275916587160965e-06, + "loss": 2.4502973556518555, + "step": 288920 + }, + { + "epoch": 0.7865523342861032, + "grad_norm": 2.135951280593872, + "learning_rate": 4.27482622325201e-06, + "loss": 2.442689323425293, + "step": 288940 + }, + { + "epoch": 0.7866067782768477, + "grad_norm": 2.5663936138153076, + "learning_rate": 4.273735859343056e-06, + "loss": 2.4576095581054687, + "step": 288960 + }, + { + "epoch": 0.7866612222675922, + "grad_norm": 2.144374370574951, + "learning_rate": 4.272645495434101e-06, + "loss": 2.4656076431274414, + "step": 288980 + }, + { + "epoch": 0.7867156662583368, + "grad_norm": 2.2602474689483643, + "learning_rate": 4.271555131525147e-06, + "loss": 2.480480766296387, + "step": 289000 + }, + { + "epoch": 0.7867701102490813, + "grad_norm": 2.24920916557312, + "learning_rate": 4.270464767616192e-06, + "loss": 2.490527534484863, + "step": 289020 + }, + { + "epoch": 0.7868245542398258, + "grad_norm": 2.2324578762054443, + "learning_rate": 4.2693744037072375e-06, + "loss": 2.470096778869629, + "step": 289040 + }, + { + "epoch": 0.7868789982305703, + "grad_norm": 2.249910831451416, + "learning_rate": 4.2682840397982825e-06, + "loss": 2.4818906784057617, + "step": 289060 + }, + { + "epoch": 0.7869334422213148, + "grad_norm": 2.1250617504119873, + "learning_rate": 4.267193675889328e-06, + "loss": 2.4293323516845704, + "step": 289080 + }, + { + "epoch": 0.7869878862120594, + "grad_norm": 2.25526762008667, + "learning_rate": 4.266103311980373e-06, + "loss": 2.464258003234863, + "step": 289100 + }, + { + "epoch": 0.7870423302028039, + "grad_norm": 2.2244832515716553, + "learning_rate": 4.265012948071419e-06, + "loss": 2.4305118560791015, + "step": 289120 + }, + { + "epoch": 0.7870967741935484, + "grad_norm": 2.166741371154785, + "learning_rate": 4.263922584162464e-06, + "loss": 2.476080894470215, + "step": 289140 + }, + { + "epoch": 0.7871512181842929, + "grad_norm": 2.1581625938415527, + "learning_rate": 4.26283222025351e-06, + "loss": 2.4425785064697267, + "step": 289160 + }, + { + "epoch": 0.7872056621750374, + "grad_norm": 2.204691171646118, + "learning_rate": 4.261741856344556e-06, + "loss": 2.4472278594970702, + "step": 289180 + }, + { + "epoch": 0.7872601061657819, + "grad_norm": 2.1692874431610107, + "learning_rate": 4.260651492435601e-06, + "loss": 2.456013298034668, + "step": 289200 + }, + { + "epoch": 0.7873145501565265, + "grad_norm": 2.2954509258270264, + "learning_rate": 4.259561128526646e-06, + "loss": 2.467819404602051, + "step": 289220 + }, + { + "epoch": 0.787368994147271, + "grad_norm": 2.3056535720825195, + "learning_rate": 4.258470764617691e-06, + "loss": 2.4008251190185548, + "step": 289240 + }, + { + "epoch": 0.7874234381380155, + "grad_norm": 2.193065643310547, + "learning_rate": 4.257380400708737e-06, + "loss": 2.453276824951172, + "step": 289260 + }, + { + "epoch": 0.78747788212876, + "grad_norm": 2.2055776119232178, + "learning_rate": 4.256290036799782e-06, + "loss": 2.479031562805176, + "step": 289280 + }, + { + "epoch": 0.7875323261195045, + "grad_norm": 2.1379172801971436, + "learning_rate": 4.255199672890828e-06, + "loss": 2.4431522369384764, + "step": 289300 + }, + { + "epoch": 0.7875867701102491, + "grad_norm": 2.3557932376861572, + "learning_rate": 4.254109308981873e-06, + "loss": 2.4878969192504883, + "step": 289320 + }, + { + "epoch": 0.7876412141009936, + "grad_norm": 2.1239371299743652, + "learning_rate": 4.253018945072919e-06, + "loss": 2.4702953338623046, + "step": 289340 + }, + { + "epoch": 0.7876956580917381, + "grad_norm": 2.266371011734009, + "learning_rate": 4.251928581163964e-06, + "loss": 2.4736963272094727, + "step": 289360 + }, + { + "epoch": 0.7877501020824826, + "grad_norm": 2.183803081512451, + "learning_rate": 4.2508382172550095e-06, + "loss": 2.457631301879883, + "step": 289380 + }, + { + "epoch": 0.7878045460732271, + "grad_norm": 2.827709674835205, + "learning_rate": 4.2497478533460545e-06, + "loss": 2.441872215270996, + "step": 289400 + }, + { + "epoch": 0.7878589900639716, + "grad_norm": 2.0962958335876465, + "learning_rate": 4.2486574894371e-06, + "loss": 2.4745105743408202, + "step": 289420 + }, + { + "epoch": 0.7879134340547163, + "grad_norm": 2.162069797515869, + "learning_rate": 4.247567125528145e-06, + "loss": 2.4376598358154298, + "step": 289440 + }, + { + "epoch": 0.7879678780454608, + "grad_norm": 2.0713446140289307, + "learning_rate": 4.246476761619191e-06, + "loss": 2.4754966735839843, + "step": 289460 + }, + { + "epoch": 0.7880223220362053, + "grad_norm": 2.1300582885742188, + "learning_rate": 4.245386397710236e-06, + "loss": 2.457859420776367, + "step": 289480 + }, + { + "epoch": 0.7880767660269498, + "grad_norm": 2.149726152420044, + "learning_rate": 4.244296033801282e-06, + "loss": 2.446318435668945, + "step": 289500 + }, + { + "epoch": 0.7881312100176943, + "grad_norm": 2.207035779953003, + "learning_rate": 4.243205669892327e-06, + "loss": 2.455927848815918, + "step": 289520 + }, + { + "epoch": 0.7881856540084389, + "grad_norm": 2.1020734310150146, + "learning_rate": 4.242115305983373e-06, + "loss": 2.430756378173828, + "step": 289540 + }, + { + "epoch": 0.7882400979991834, + "grad_norm": 2.175586462020874, + "learning_rate": 4.2410249420744175e-06, + "loss": 2.4614309310913085, + "step": 289560 + }, + { + "epoch": 0.7882945419899279, + "grad_norm": 2.237579822540283, + "learning_rate": 4.239934578165463e-06, + "loss": 2.473292350769043, + "step": 289580 + }, + { + "epoch": 0.7883489859806724, + "grad_norm": 2.1517715454101562, + "learning_rate": 4.238844214256508e-06, + "loss": 2.445669174194336, + "step": 289600 + }, + { + "epoch": 0.7884034299714169, + "grad_norm": 2.220510244369507, + "learning_rate": 4.237753850347554e-06, + "loss": 2.445952224731445, + "step": 289620 + }, + { + "epoch": 0.7884578739621614, + "grad_norm": 2.195098400115967, + "learning_rate": 4.236663486438599e-06, + "loss": 2.464571952819824, + "step": 289640 + }, + { + "epoch": 0.788512317952906, + "grad_norm": 2.2446138858795166, + "learning_rate": 4.235573122529645e-06, + "loss": 2.4695058822631837, + "step": 289660 + }, + { + "epoch": 0.7885667619436505, + "grad_norm": 2.1308343410491943, + "learning_rate": 4.23448275862069e-06, + "loss": 2.457079315185547, + "step": 289680 + }, + { + "epoch": 0.788621205934395, + "grad_norm": 2.3382840156555176, + "learning_rate": 4.233392394711736e-06, + "loss": 2.461593246459961, + "step": 289700 + }, + { + "epoch": 0.7886756499251395, + "grad_norm": 2.480173110961914, + "learning_rate": 4.232302030802781e-06, + "loss": 2.483359718322754, + "step": 289720 + }, + { + "epoch": 0.788730093915884, + "grad_norm": 2.397277355194092, + "learning_rate": 4.2312116668938264e-06, + "loss": 2.438639831542969, + "step": 289740 + }, + { + "epoch": 0.7887845379066285, + "grad_norm": 3.3599236011505127, + "learning_rate": 4.230121302984871e-06, + "loss": 2.4711273193359373, + "step": 289760 + }, + { + "epoch": 0.7888389818973731, + "grad_norm": 2.1182308197021484, + "learning_rate": 4.229030939075917e-06, + "loss": 2.4552587509155273, + "step": 289780 + }, + { + "epoch": 0.7888934258881176, + "grad_norm": 2.22459077835083, + "learning_rate": 4.227940575166962e-06, + "loss": 2.463431167602539, + "step": 289800 + }, + { + "epoch": 0.7889478698788621, + "grad_norm": 2.142108678817749, + "learning_rate": 4.226850211258008e-06, + "loss": 2.4378377914428713, + "step": 289820 + }, + { + "epoch": 0.7890023138696066, + "grad_norm": 2.161928653717041, + "learning_rate": 4.225759847349053e-06, + "loss": 2.4804079055786135, + "step": 289840 + }, + { + "epoch": 0.7890567578603511, + "grad_norm": 2.0864615440368652, + "learning_rate": 4.224669483440099e-06, + "loss": 2.4621322631835936, + "step": 289860 + }, + { + "epoch": 0.7891112018510957, + "grad_norm": 2.336743116378784, + "learning_rate": 4.223579119531144e-06, + "loss": 2.4660377502441406, + "step": 289880 + }, + { + "epoch": 0.7891656458418402, + "grad_norm": 2.1887640953063965, + "learning_rate": 4.2224887556221895e-06, + "loss": 2.464976692199707, + "step": 289900 + }, + { + "epoch": 0.7892200898325847, + "grad_norm": 2.1793923377990723, + "learning_rate": 4.2213983917132345e-06, + "loss": 2.487997627258301, + "step": 289920 + }, + { + "epoch": 0.7892745338233292, + "grad_norm": 2.1346356868743896, + "learning_rate": 4.22030802780428e-06, + "loss": 2.460707092285156, + "step": 289940 + }, + { + "epoch": 0.7893289778140737, + "grad_norm": 2.217749834060669, + "learning_rate": 4.219217663895325e-06, + "loss": 2.49768180847168, + "step": 289960 + }, + { + "epoch": 0.7893834218048182, + "grad_norm": 3.1246728897094727, + "learning_rate": 4.218127299986371e-06, + "loss": 2.4727781295776365, + "step": 289980 + }, + { + "epoch": 0.7894378657955629, + "grad_norm": 2.2628250122070312, + "learning_rate": 4.217036936077416e-06, + "loss": 2.4815317153930665, + "step": 290000 + }, + { + "epoch": 0.7894923097863074, + "grad_norm": 2.3779571056365967, + "learning_rate": 4.215946572168462e-06, + "loss": 2.450430488586426, + "step": 290020 + }, + { + "epoch": 0.7895467537770519, + "grad_norm": 2.2127602100372314, + "learning_rate": 4.214856208259507e-06, + "loss": 2.4486413955688477, + "step": 290040 + }, + { + "epoch": 0.7896011977677964, + "grad_norm": 2.0957555770874023, + "learning_rate": 4.213765844350553e-06, + "loss": 2.4675003051757813, + "step": 290060 + }, + { + "epoch": 0.7896556417585409, + "grad_norm": 2.1266486644744873, + "learning_rate": 4.2126754804415976e-06, + "loss": 2.448954391479492, + "step": 290080 + }, + { + "epoch": 0.7897100857492855, + "grad_norm": 2.4497978687286377, + "learning_rate": 4.211585116532643e-06, + "loss": 2.4851053237915037, + "step": 290100 + }, + { + "epoch": 0.78976452974003, + "grad_norm": 2.200936794281006, + "learning_rate": 4.210494752623688e-06, + "loss": 2.464680862426758, + "step": 290120 + }, + { + "epoch": 0.7898189737307745, + "grad_norm": 2.1603312492370605, + "learning_rate": 4.209404388714734e-06, + "loss": 2.4504159927368163, + "step": 290140 + }, + { + "epoch": 0.789873417721519, + "grad_norm": 2.223205327987671, + "learning_rate": 4.208314024805779e-06, + "loss": 2.4677297592163088, + "step": 290160 + }, + { + "epoch": 0.7899278617122635, + "grad_norm": 2.158630132675171, + "learning_rate": 4.207223660896825e-06, + "loss": 2.4566623687744142, + "step": 290180 + }, + { + "epoch": 0.789982305703008, + "grad_norm": 2.116391897201538, + "learning_rate": 4.20613329698787e-06, + "loss": 2.4438304901123047, + "step": 290200 + }, + { + "epoch": 0.7900367496937526, + "grad_norm": 2.162806987762451, + "learning_rate": 4.205042933078916e-06, + "loss": 2.461451530456543, + "step": 290220 + }, + { + "epoch": 0.7900911936844971, + "grad_norm": 2.1446075439453125, + "learning_rate": 4.203952569169961e-06, + "loss": 2.4479291915893553, + "step": 290240 + }, + { + "epoch": 0.7901456376752416, + "grad_norm": 2.320834159851074, + "learning_rate": 4.2028622052610065e-06, + "loss": 2.4744815826416016, + "step": 290260 + }, + { + "epoch": 0.7902000816659861, + "grad_norm": 2.200120210647583, + "learning_rate": 4.2017718413520514e-06, + "loss": 2.4591121673583984, + "step": 290280 + }, + { + "epoch": 0.7902545256567306, + "grad_norm": 2.1224615573883057, + "learning_rate": 4.200681477443097e-06, + "loss": 2.441903305053711, + "step": 290300 + }, + { + "epoch": 0.7903089696474752, + "grad_norm": 2.2348198890686035, + "learning_rate": 4.199591113534142e-06, + "loss": 2.432171630859375, + "step": 290320 + }, + { + "epoch": 0.7903634136382197, + "grad_norm": 2.0629091262817383, + "learning_rate": 4.198500749625188e-06, + "loss": 2.457505226135254, + "step": 290340 + }, + { + "epoch": 0.7904178576289642, + "grad_norm": 2.1317312717437744, + "learning_rate": 4.197410385716233e-06, + "loss": 2.474913215637207, + "step": 290360 + }, + { + "epoch": 0.7904723016197087, + "grad_norm": 2.246678113937378, + "learning_rate": 4.196320021807279e-06, + "loss": 2.4649942398071287, + "step": 290380 + }, + { + "epoch": 0.7905267456104532, + "grad_norm": 2.119601011276245, + "learning_rate": 4.195229657898324e-06, + "loss": 2.433853340148926, + "step": 290400 + }, + { + "epoch": 0.7905811896011977, + "grad_norm": 2.1849992275238037, + "learning_rate": 4.1941392939893696e-06, + "loss": 2.4513654708862305, + "step": 290420 + }, + { + "epoch": 0.7906356335919423, + "grad_norm": 2.3773128986358643, + "learning_rate": 4.1930489300804145e-06, + "loss": 2.485912895202637, + "step": 290440 + }, + { + "epoch": 0.7906900775826868, + "grad_norm": 2.1679937839508057, + "learning_rate": 4.19195856617146e-06, + "loss": 2.4514957427978517, + "step": 290460 + }, + { + "epoch": 0.7907445215734313, + "grad_norm": 2.337738275527954, + "learning_rate": 4.190868202262505e-06, + "loss": 2.433082389831543, + "step": 290480 + }, + { + "epoch": 0.7907989655641758, + "grad_norm": 2.4414570331573486, + "learning_rate": 4.189777838353551e-06, + "loss": 2.446895790100098, + "step": 290500 + }, + { + "epoch": 0.7908534095549203, + "grad_norm": 2.1687052249908447, + "learning_rate": 4.188687474444596e-06, + "loss": 2.45517578125, + "step": 290520 + }, + { + "epoch": 0.790907853545665, + "grad_norm": 2.1205496788024902, + "learning_rate": 4.187597110535642e-06, + "loss": 2.474724769592285, + "step": 290540 + }, + { + "epoch": 0.7909622975364095, + "grad_norm": 2.141146421432495, + "learning_rate": 4.186506746626687e-06, + "loss": 2.4790006637573243, + "step": 290560 + }, + { + "epoch": 0.791016741527154, + "grad_norm": 2.1402337551116943, + "learning_rate": 4.185416382717733e-06, + "loss": 2.488852691650391, + "step": 290580 + }, + { + "epoch": 0.7910711855178985, + "grad_norm": 2.231748104095459, + "learning_rate": 4.184326018808778e-06, + "loss": 2.464628982543945, + "step": 290600 + }, + { + "epoch": 0.791125629508643, + "grad_norm": 2.1848788261413574, + "learning_rate": 4.183235654899823e-06, + "loss": 2.496651458740234, + "step": 290620 + }, + { + "epoch": 0.7911800734993875, + "grad_norm": 2.2127468585968018, + "learning_rate": 4.182145290990868e-06, + "loss": 2.4969676971435546, + "step": 290640 + }, + { + "epoch": 0.7912345174901321, + "grad_norm": 2.22038197517395, + "learning_rate": 4.181054927081914e-06, + "loss": 2.4542848587036135, + "step": 290660 + }, + { + "epoch": 0.7912889614808766, + "grad_norm": 2.144343852996826, + "learning_rate": 4.179964563172959e-06, + "loss": 2.486754608154297, + "step": 290680 + }, + { + "epoch": 0.7913434054716211, + "grad_norm": 2.1880080699920654, + "learning_rate": 4.178874199264005e-06, + "loss": 2.4903032302856447, + "step": 290700 + }, + { + "epoch": 0.7913978494623656, + "grad_norm": 2.1980888843536377, + "learning_rate": 4.17778383535505e-06, + "loss": 2.4909347534179687, + "step": 290720 + }, + { + "epoch": 0.7914522934531101, + "grad_norm": 2.117363929748535, + "learning_rate": 4.176693471446096e-06, + "loss": 2.4705026626586912, + "step": 290740 + }, + { + "epoch": 0.7915067374438546, + "grad_norm": 2.107501745223999, + "learning_rate": 4.175603107537141e-06, + "loss": 2.4718439102172853, + "step": 290760 + }, + { + "epoch": 0.7915611814345992, + "grad_norm": 2.120208501815796, + "learning_rate": 4.1745127436281865e-06, + "loss": 2.456391716003418, + "step": 290780 + }, + { + "epoch": 0.7916156254253437, + "grad_norm": 2.12628436088562, + "learning_rate": 4.1734223797192315e-06, + "loss": 2.4662479400634765, + "step": 290800 + }, + { + "epoch": 0.7916700694160882, + "grad_norm": 2.135129451751709, + "learning_rate": 4.172332015810277e-06, + "loss": 2.4685468673706055, + "step": 290820 + }, + { + "epoch": 0.7917245134068327, + "grad_norm": 2.124415159225464, + "learning_rate": 4.171241651901322e-06, + "loss": 2.4755861282348635, + "step": 290840 + }, + { + "epoch": 0.7917789573975772, + "grad_norm": 2.320451021194458, + "learning_rate": 4.170151287992368e-06, + "loss": 2.463625907897949, + "step": 290860 + }, + { + "epoch": 0.7918334013883218, + "grad_norm": 2.156308889389038, + "learning_rate": 4.169060924083414e-06, + "loss": 2.500005531311035, + "step": 290880 + }, + { + "epoch": 0.7918878453790663, + "grad_norm": 2.2084357738494873, + "learning_rate": 4.167970560174459e-06, + "loss": 2.478459930419922, + "step": 290900 + }, + { + "epoch": 0.7919422893698108, + "grad_norm": 2.126823663711548, + "learning_rate": 4.166880196265505e-06, + "loss": 2.442005920410156, + "step": 290920 + }, + { + "epoch": 0.7919967333605553, + "grad_norm": 2.1993343830108643, + "learning_rate": 4.16578983235655e-06, + "loss": 2.470698356628418, + "step": 290940 + }, + { + "epoch": 0.7920511773512998, + "grad_norm": 2.1667656898498535, + "learning_rate": 4.164699468447595e-06, + "loss": 2.4633508682250977, + "step": 290960 + }, + { + "epoch": 0.7921056213420443, + "grad_norm": 2.255162239074707, + "learning_rate": 4.16360910453864e-06, + "loss": 2.423664665222168, + "step": 290980 + }, + { + "epoch": 0.7921600653327889, + "grad_norm": 2.262725591659546, + "learning_rate": 4.162518740629686e-06, + "loss": 2.493786430358887, + "step": 291000 + }, + { + "epoch": 0.7922145093235334, + "grad_norm": 2.1144485473632812, + "learning_rate": 4.161428376720731e-06, + "loss": 2.485045051574707, + "step": 291020 + }, + { + "epoch": 0.7922689533142779, + "grad_norm": 2.1678740978240967, + "learning_rate": 4.160338012811777e-06, + "loss": 2.4654470443725587, + "step": 291040 + }, + { + "epoch": 0.7923233973050224, + "grad_norm": 2.2538797855377197, + "learning_rate": 4.159247648902822e-06, + "loss": 2.4645002365112303, + "step": 291060 + }, + { + "epoch": 0.7923778412957669, + "grad_norm": 2.2561211585998535, + "learning_rate": 4.158157284993868e-06, + "loss": 2.449733924865723, + "step": 291080 + }, + { + "epoch": 0.7924322852865116, + "grad_norm": 2.2146027088165283, + "learning_rate": 4.157066921084913e-06, + "loss": 2.4443733215332033, + "step": 291100 + }, + { + "epoch": 0.792486729277256, + "grad_norm": 2.186262607574463, + "learning_rate": 4.155976557175958e-06, + "loss": 2.469771385192871, + "step": 291120 + }, + { + "epoch": 0.7925411732680006, + "grad_norm": 2.6239378452301025, + "learning_rate": 4.1548861932670034e-06, + "loss": 2.4618097305297852, + "step": 291140 + }, + { + "epoch": 0.792595617258745, + "grad_norm": 2.1417229175567627, + "learning_rate": 4.153795829358048e-06, + "loss": 2.432249641418457, + "step": 291160 + }, + { + "epoch": 0.7926500612494896, + "grad_norm": 2.415273666381836, + "learning_rate": 4.152705465449094e-06, + "loss": 2.4592306137084963, + "step": 291180 + }, + { + "epoch": 0.7927045052402341, + "grad_norm": 2.1870250701904297, + "learning_rate": 4.151615101540139e-06, + "loss": 2.480809783935547, + "step": 291200 + }, + { + "epoch": 0.7927589492309787, + "grad_norm": 2.2239458560943604, + "learning_rate": 4.150524737631185e-06, + "loss": 2.456791114807129, + "step": 291220 + }, + { + "epoch": 0.7928133932217232, + "grad_norm": 2.2156665325164795, + "learning_rate": 4.14943437372223e-06, + "loss": 2.4546173095703123, + "step": 291240 + }, + { + "epoch": 0.7928678372124677, + "grad_norm": 2.0639054775238037, + "learning_rate": 4.148344009813276e-06, + "loss": 2.4591236114501953, + "step": 291260 + }, + { + "epoch": 0.7929222812032122, + "grad_norm": 2.2693910598754883, + "learning_rate": 4.147253645904321e-06, + "loss": 2.4808748245239256, + "step": 291280 + }, + { + "epoch": 0.7929767251939567, + "grad_norm": 2.631901264190674, + "learning_rate": 4.1461632819953665e-06, + "loss": 2.464629554748535, + "step": 291300 + }, + { + "epoch": 0.7930311691847013, + "grad_norm": 2.1455063819885254, + "learning_rate": 4.1450729180864115e-06, + "loss": 2.49014949798584, + "step": 291320 + }, + { + "epoch": 0.7930856131754458, + "grad_norm": 2.228254556655884, + "learning_rate": 4.143982554177457e-06, + "loss": 2.4729814529418945, + "step": 291340 + }, + { + "epoch": 0.7931400571661903, + "grad_norm": 2.2674190998077393, + "learning_rate": 4.142892190268502e-06, + "loss": 2.475399971008301, + "step": 291360 + }, + { + "epoch": 0.7931945011569348, + "grad_norm": 2.394425392150879, + "learning_rate": 4.141801826359548e-06, + "loss": 2.467416000366211, + "step": 291380 + }, + { + "epoch": 0.7932489451476793, + "grad_norm": 2.1740293502807617, + "learning_rate": 4.140711462450593e-06, + "loss": 2.4689958572387694, + "step": 291400 + }, + { + "epoch": 0.7933033891384238, + "grad_norm": 2.473489761352539, + "learning_rate": 4.139621098541639e-06, + "loss": 2.4361114501953125, + "step": 291420 + }, + { + "epoch": 0.7933578331291684, + "grad_norm": 2.1477394104003906, + "learning_rate": 4.138530734632684e-06, + "loss": 2.4523784637451174, + "step": 291440 + }, + { + "epoch": 0.7934122771199129, + "grad_norm": 2.178603172302246, + "learning_rate": 4.13744037072373e-06, + "loss": 2.478835105895996, + "step": 291460 + }, + { + "epoch": 0.7934667211106574, + "grad_norm": 2.1575794219970703, + "learning_rate": 4.1363500068147746e-06, + "loss": 2.4400863647460938, + "step": 291480 + }, + { + "epoch": 0.7935211651014019, + "grad_norm": 2.2141802310943604, + "learning_rate": 4.13525964290582e-06, + "loss": 2.453192138671875, + "step": 291500 + }, + { + "epoch": 0.7935756090921464, + "grad_norm": 2.1997032165527344, + "learning_rate": 4.134169278996865e-06, + "loss": 2.451388359069824, + "step": 291520 + }, + { + "epoch": 0.793630053082891, + "grad_norm": 2.318195343017578, + "learning_rate": 4.133078915087911e-06, + "loss": 2.4415021896362306, + "step": 291540 + }, + { + "epoch": 0.7936844970736355, + "grad_norm": 2.157681703567505, + "learning_rate": 4.131988551178956e-06, + "loss": 2.4635644912719727, + "step": 291560 + }, + { + "epoch": 0.79373894106438, + "grad_norm": 2.1090586185455322, + "learning_rate": 4.130898187270002e-06, + "loss": 2.4901695251464844, + "step": 291580 + }, + { + "epoch": 0.7937933850551245, + "grad_norm": 2.1825926303863525, + "learning_rate": 4.129807823361047e-06, + "loss": 2.4509302139282227, + "step": 291600 + }, + { + "epoch": 0.793847829045869, + "grad_norm": 2.2207539081573486, + "learning_rate": 4.128717459452093e-06, + "loss": 2.463504981994629, + "step": 291620 + }, + { + "epoch": 0.7939022730366135, + "grad_norm": 2.243436813354492, + "learning_rate": 4.127627095543138e-06, + "loss": 2.4565439224243164, + "step": 291640 + }, + { + "epoch": 0.7939567170273581, + "grad_norm": 2.169614791870117, + "learning_rate": 4.1265367316341835e-06, + "loss": 2.44311466217041, + "step": 291660 + }, + { + "epoch": 0.7940111610181027, + "grad_norm": 2.1080241203308105, + "learning_rate": 4.1254463677252284e-06, + "loss": 2.475821685791016, + "step": 291680 + }, + { + "epoch": 0.7940656050088472, + "grad_norm": 2.0632426738739014, + "learning_rate": 4.124356003816274e-06, + "loss": 2.484261894226074, + "step": 291700 + }, + { + "epoch": 0.7941200489995917, + "grad_norm": 2.136152982711792, + "learning_rate": 4.123265639907319e-06, + "loss": 2.4601055145263673, + "step": 291720 + }, + { + "epoch": 0.7941744929903362, + "grad_norm": 2.110703229904175, + "learning_rate": 4.122175275998365e-06, + "loss": 2.4761091232299806, + "step": 291740 + }, + { + "epoch": 0.7942289369810808, + "grad_norm": 2.1717569828033447, + "learning_rate": 4.12108491208941e-06, + "loss": 2.4281606674194336, + "step": 291760 + }, + { + "epoch": 0.7942833809718253, + "grad_norm": 2.0860254764556885, + "learning_rate": 4.119994548180456e-06, + "loss": 2.446787643432617, + "step": 291780 + }, + { + "epoch": 0.7943378249625698, + "grad_norm": 2.2015106678009033, + "learning_rate": 4.118904184271501e-06, + "loss": 2.4922985076904296, + "step": 291800 + }, + { + "epoch": 0.7943922689533143, + "grad_norm": 2.2417256832122803, + "learning_rate": 4.1178138203625466e-06, + "loss": 2.478451728820801, + "step": 291820 + }, + { + "epoch": 0.7944467129440588, + "grad_norm": 2.2481842041015625, + "learning_rate": 4.1167234564535915e-06, + "loss": 2.4720037460327147, + "step": 291840 + }, + { + "epoch": 0.7945011569348033, + "grad_norm": 2.410611152648926, + "learning_rate": 4.115633092544637e-06, + "loss": 2.455064392089844, + "step": 291860 + }, + { + "epoch": 0.7945556009255479, + "grad_norm": 2.1114752292633057, + "learning_rate": 4.114542728635682e-06, + "loss": 2.47774658203125, + "step": 291880 + }, + { + "epoch": 0.7946100449162924, + "grad_norm": 2.4267354011535645, + "learning_rate": 4.113452364726728e-06, + "loss": 2.4729076385498048, + "step": 291900 + }, + { + "epoch": 0.7946644889070369, + "grad_norm": 2.1413445472717285, + "learning_rate": 4.112362000817773e-06, + "loss": 2.485400390625, + "step": 291920 + }, + { + "epoch": 0.7947189328977814, + "grad_norm": 2.418945550918579, + "learning_rate": 4.111271636908819e-06, + "loss": 2.464427947998047, + "step": 291940 + }, + { + "epoch": 0.7947733768885259, + "grad_norm": 2.176713466644287, + "learning_rate": 4.110181272999864e-06, + "loss": 2.463326263427734, + "step": 291960 + }, + { + "epoch": 0.7948278208792704, + "grad_norm": 2.3068368434906006, + "learning_rate": 4.10909090909091e-06, + "loss": 2.462550163269043, + "step": 291980 + }, + { + "epoch": 0.794882264870015, + "grad_norm": 2.2858195304870605, + "learning_rate": 4.108000545181955e-06, + "loss": 2.4875308990478517, + "step": 292000 + }, + { + "epoch": 0.7949367088607595, + "grad_norm": 2.2661712169647217, + "learning_rate": 4.106910181273e-06, + "loss": 2.4245321273803713, + "step": 292020 + }, + { + "epoch": 0.794991152851504, + "grad_norm": 2.149343967437744, + "learning_rate": 4.105819817364045e-06, + "loss": 2.4605667114257814, + "step": 292040 + }, + { + "epoch": 0.7950455968422485, + "grad_norm": 2.202749729156494, + "learning_rate": 4.104729453455091e-06, + "loss": 2.4846319198608398, + "step": 292060 + }, + { + "epoch": 0.795100040832993, + "grad_norm": 2.2026705741882324, + "learning_rate": 4.103639089546136e-06, + "loss": 2.447983741760254, + "step": 292080 + }, + { + "epoch": 0.7951544848237376, + "grad_norm": 2.129042863845825, + "learning_rate": 4.102548725637182e-06, + "loss": 2.4663890838623046, + "step": 292100 + }, + { + "epoch": 0.7952089288144821, + "grad_norm": 3.6468539237976074, + "learning_rate": 4.101458361728227e-06, + "loss": 2.4564224243164063, + "step": 292120 + }, + { + "epoch": 0.7952633728052266, + "grad_norm": 2.1521477699279785, + "learning_rate": 4.100367997819273e-06, + "loss": 2.4709423065185545, + "step": 292140 + }, + { + "epoch": 0.7953178167959711, + "grad_norm": 2.1951026916503906, + "learning_rate": 4.099277633910318e-06, + "loss": 2.4383672714233398, + "step": 292160 + }, + { + "epoch": 0.7953722607867156, + "grad_norm": 2.314944267272949, + "learning_rate": 4.0981872700013635e-06, + "loss": 2.458671760559082, + "step": 292180 + }, + { + "epoch": 0.7954267047774601, + "grad_norm": 2.046083450317383, + "learning_rate": 4.0970969060924085e-06, + "loss": 2.451187515258789, + "step": 292200 + }, + { + "epoch": 0.7954811487682047, + "grad_norm": 2.1672534942626953, + "learning_rate": 4.096006542183453e-06, + "loss": 2.4764156341552734, + "step": 292220 + }, + { + "epoch": 0.7955355927589493, + "grad_norm": 2.149850845336914, + "learning_rate": 4.094916178274499e-06, + "loss": 2.4301523208618163, + "step": 292240 + }, + { + "epoch": 0.7955900367496938, + "grad_norm": 2.1349387168884277, + "learning_rate": 4.093825814365544e-06, + "loss": 2.468537139892578, + "step": 292260 + }, + { + "epoch": 0.7956444807404383, + "grad_norm": 2.1087028980255127, + "learning_rate": 4.09273545045659e-06, + "loss": 2.4565834045410155, + "step": 292280 + }, + { + "epoch": 0.7956989247311828, + "grad_norm": 2.1500425338745117, + "learning_rate": 4.091645086547635e-06, + "loss": 2.45511531829834, + "step": 292300 + }, + { + "epoch": 0.7957533687219274, + "grad_norm": 2.2011947631835938, + "learning_rate": 4.090554722638681e-06, + "loss": 2.4139028549194337, + "step": 292320 + }, + { + "epoch": 0.7958078127126719, + "grad_norm": 2.0936877727508545, + "learning_rate": 4.089464358729726e-06, + "loss": 2.4418399810791014, + "step": 292340 + }, + { + "epoch": 0.7958622567034164, + "grad_norm": 2.236600160598755, + "learning_rate": 4.0883739948207715e-06, + "loss": 2.454728889465332, + "step": 292360 + }, + { + "epoch": 0.7959167006941609, + "grad_norm": 2.2250149250030518, + "learning_rate": 4.0872836309118165e-06, + "loss": 2.4259855270385744, + "step": 292380 + }, + { + "epoch": 0.7959711446849054, + "grad_norm": 2.169023036956787, + "learning_rate": 4.086193267002862e-06, + "loss": 2.477433776855469, + "step": 292400 + }, + { + "epoch": 0.7960255886756499, + "grad_norm": 2.259873867034912, + "learning_rate": 4.085102903093907e-06, + "loss": 2.4749425888061523, + "step": 292420 + }, + { + "epoch": 0.7960800326663945, + "grad_norm": 2.1898703575134277, + "learning_rate": 4.084012539184953e-06, + "loss": 2.440983772277832, + "step": 292440 + }, + { + "epoch": 0.796134476657139, + "grad_norm": 2.2674596309661865, + "learning_rate": 4.082922175275998e-06, + "loss": 2.504379463195801, + "step": 292460 + }, + { + "epoch": 0.7961889206478835, + "grad_norm": 2.0332305431365967, + "learning_rate": 4.081831811367044e-06, + "loss": 2.475302314758301, + "step": 292480 + }, + { + "epoch": 0.796243364638628, + "grad_norm": 2.2693300247192383, + "learning_rate": 4.080741447458089e-06, + "loss": 2.451036262512207, + "step": 292500 + }, + { + "epoch": 0.7962978086293725, + "grad_norm": 2.119685173034668, + "learning_rate": 4.079651083549135e-06, + "loss": 2.4231546401977537, + "step": 292520 + }, + { + "epoch": 0.7963522526201171, + "grad_norm": 2.1036953926086426, + "learning_rate": 4.07856071964018e-06, + "loss": 2.4598968505859373, + "step": 292540 + }, + { + "epoch": 0.7964066966108616, + "grad_norm": 2.2257955074310303, + "learning_rate": 4.077470355731225e-06, + "loss": 2.4421924591064452, + "step": 292560 + }, + { + "epoch": 0.7964611406016061, + "grad_norm": 2.1626994609832764, + "learning_rate": 4.076379991822271e-06, + "loss": 2.444980812072754, + "step": 292580 + }, + { + "epoch": 0.7965155845923506, + "grad_norm": 2.179117202758789, + "learning_rate": 4.075289627913316e-06, + "loss": 2.4522682189941407, + "step": 292600 + }, + { + "epoch": 0.7965700285830951, + "grad_norm": 2.213405132293701, + "learning_rate": 4.074199264004362e-06, + "loss": 2.463826370239258, + "step": 292620 + }, + { + "epoch": 0.7966244725738396, + "grad_norm": 2.0751020908355713, + "learning_rate": 4.073108900095407e-06, + "loss": 2.4451812744140624, + "step": 292640 + }, + { + "epoch": 0.7966789165645842, + "grad_norm": 2.197491407394409, + "learning_rate": 4.072018536186453e-06, + "loss": 2.442708969116211, + "step": 292660 + }, + { + "epoch": 0.7967333605553287, + "grad_norm": 2.208310604095459, + "learning_rate": 4.070928172277498e-06, + "loss": 2.484300231933594, + "step": 292680 + }, + { + "epoch": 0.7967878045460732, + "grad_norm": 2.2378063201904297, + "learning_rate": 4.0698378083685435e-06, + "loss": 2.452728271484375, + "step": 292700 + }, + { + "epoch": 0.7968422485368177, + "grad_norm": 2.122612476348877, + "learning_rate": 4.0687474444595885e-06, + "loss": 2.4920669555664063, + "step": 292720 + }, + { + "epoch": 0.7968966925275622, + "grad_norm": 2.283987283706665, + "learning_rate": 4.067657080550634e-06, + "loss": 2.451206588745117, + "step": 292740 + }, + { + "epoch": 0.7969511365183068, + "grad_norm": 2.210667848587036, + "learning_rate": 4.066566716641679e-06, + "loss": 2.4572582244873047, + "step": 292760 + }, + { + "epoch": 0.7970055805090513, + "grad_norm": 2.1378660202026367, + "learning_rate": 4.065476352732725e-06, + "loss": 2.4567886352539063, + "step": 292780 + }, + { + "epoch": 0.7970600244997958, + "grad_norm": 2.2209157943725586, + "learning_rate": 4.06438598882377e-06, + "loss": 2.4547876358032226, + "step": 292800 + }, + { + "epoch": 0.7971144684905404, + "grad_norm": 2.065418243408203, + "learning_rate": 4.063295624914816e-06, + "loss": 2.4284658432006836, + "step": 292820 + }, + { + "epoch": 0.7971689124812849, + "grad_norm": 2.15999436378479, + "learning_rate": 4.062205261005861e-06, + "loss": 2.4735002517700195, + "step": 292840 + }, + { + "epoch": 0.7972233564720294, + "grad_norm": 2.3901920318603516, + "learning_rate": 4.061114897096907e-06, + "loss": 2.461644744873047, + "step": 292860 + }, + { + "epoch": 0.797277800462774, + "grad_norm": 2.1584787368774414, + "learning_rate": 4.0600245331879516e-06, + "loss": 2.420149803161621, + "step": 292880 + }, + { + "epoch": 0.7973322444535185, + "grad_norm": 2.0616936683654785, + "learning_rate": 4.058934169278997e-06, + "loss": 2.461374282836914, + "step": 292900 + }, + { + "epoch": 0.797386688444263, + "grad_norm": 2.1320106983184814, + "learning_rate": 4.057843805370042e-06, + "loss": 2.4743984222412108, + "step": 292920 + }, + { + "epoch": 0.7974411324350075, + "grad_norm": 2.216703414916992, + "learning_rate": 4.056753441461088e-06, + "loss": 2.445353317260742, + "step": 292940 + }, + { + "epoch": 0.797495576425752, + "grad_norm": 2.1751937866210938, + "learning_rate": 4.055663077552133e-06, + "loss": 2.4246593475341798, + "step": 292960 + }, + { + "epoch": 0.7975500204164965, + "grad_norm": 2.2108535766601562, + "learning_rate": 4.054572713643179e-06, + "loss": 2.472290802001953, + "step": 292980 + }, + { + "epoch": 0.7976044644072411, + "grad_norm": 2.17155122756958, + "learning_rate": 4.053482349734224e-06, + "loss": 2.449789619445801, + "step": 293000 + }, + { + "epoch": 0.7976589083979856, + "grad_norm": 2.1606805324554443, + "learning_rate": 4.05239198582527e-06, + "loss": 2.445292854309082, + "step": 293020 + }, + { + "epoch": 0.7977133523887301, + "grad_norm": 2.1644225120544434, + "learning_rate": 4.051301621916315e-06, + "loss": 2.4919214248657227, + "step": 293040 + }, + { + "epoch": 0.7977677963794746, + "grad_norm": 2.1725523471832275, + "learning_rate": 4.0502112580073605e-06, + "loss": 2.462664031982422, + "step": 293060 + }, + { + "epoch": 0.7978222403702191, + "grad_norm": 2.2682363986968994, + "learning_rate": 4.0491208940984054e-06, + "loss": 2.464804840087891, + "step": 293080 + }, + { + "epoch": 0.7978766843609637, + "grad_norm": 2.0125224590301514, + "learning_rate": 4.048030530189451e-06, + "loss": 2.462942886352539, + "step": 293100 + }, + { + "epoch": 0.7979311283517082, + "grad_norm": 2.201266288757324, + "learning_rate": 4.046940166280496e-06, + "loss": 2.4545610427856444, + "step": 293120 + }, + { + "epoch": 0.7979855723424527, + "grad_norm": 2.185912847518921, + "learning_rate": 4.045849802371542e-06, + "loss": 2.471666145324707, + "step": 293140 + }, + { + "epoch": 0.7980400163331972, + "grad_norm": 2.194899797439575, + "learning_rate": 4.044759438462587e-06, + "loss": 2.446695899963379, + "step": 293160 + }, + { + "epoch": 0.7980944603239417, + "grad_norm": 2.252870798110962, + "learning_rate": 4.043669074553633e-06, + "loss": 2.4532497406005858, + "step": 293180 + }, + { + "epoch": 0.7981489043146862, + "grad_norm": 2.1346256732940674, + "learning_rate": 4.042578710644678e-06, + "loss": 2.4411394119262697, + "step": 293200 + }, + { + "epoch": 0.7982033483054308, + "grad_norm": 2.11370587348938, + "learning_rate": 4.0414883467357235e-06, + "loss": 2.4846357345581054, + "step": 293220 + }, + { + "epoch": 0.7982577922961753, + "grad_norm": 2.1689820289611816, + "learning_rate": 4.0403979828267685e-06, + "loss": 2.472296142578125, + "step": 293240 + }, + { + "epoch": 0.7983122362869198, + "grad_norm": 2.1028499603271484, + "learning_rate": 4.039307618917814e-06, + "loss": 2.4816890716552735, + "step": 293260 + }, + { + "epoch": 0.7983666802776643, + "grad_norm": 2.147871732711792, + "learning_rate": 4.038217255008859e-06, + "loss": 2.47789306640625, + "step": 293280 + }, + { + "epoch": 0.7984211242684088, + "grad_norm": 2.222787618637085, + "learning_rate": 4.037126891099905e-06, + "loss": 2.4712329864501954, + "step": 293300 + }, + { + "epoch": 0.7984755682591534, + "grad_norm": 2.1607954502105713, + "learning_rate": 4.03603652719095e-06, + "loss": 2.464852714538574, + "step": 293320 + }, + { + "epoch": 0.798530012249898, + "grad_norm": 2.2115366458892822, + "learning_rate": 4.034946163281996e-06, + "loss": 2.4428375244140623, + "step": 293340 + }, + { + "epoch": 0.7985844562406424, + "grad_norm": 2.2411270141601562, + "learning_rate": 4.033855799373041e-06, + "loss": 2.4801351547241213, + "step": 293360 + }, + { + "epoch": 0.798638900231387, + "grad_norm": 2.333876848220825, + "learning_rate": 4.032765435464087e-06, + "loss": 2.4581750869750976, + "step": 293380 + }, + { + "epoch": 0.7986933442221315, + "grad_norm": 2.1285274028778076, + "learning_rate": 4.031675071555132e-06, + "loss": 2.4345060348510743, + "step": 293400 + }, + { + "epoch": 0.798747788212876, + "grad_norm": 2.1848762035369873, + "learning_rate": 4.030584707646177e-06, + "loss": 2.441354179382324, + "step": 293420 + }, + { + "epoch": 0.7988022322036206, + "grad_norm": 2.0916690826416016, + "learning_rate": 4.029494343737222e-06, + "loss": 2.4337364196777345, + "step": 293440 + }, + { + "epoch": 0.7988566761943651, + "grad_norm": 2.065307855606079, + "learning_rate": 4.028403979828268e-06, + "loss": 2.455360984802246, + "step": 293460 + }, + { + "epoch": 0.7989111201851096, + "grad_norm": 2.5005624294281006, + "learning_rate": 4.027313615919313e-06, + "loss": 2.4875207901000977, + "step": 293480 + }, + { + "epoch": 0.7989655641758541, + "grad_norm": 2.210836887359619, + "learning_rate": 4.026223252010359e-06, + "loss": 2.486106109619141, + "step": 293500 + }, + { + "epoch": 0.7990200081665986, + "grad_norm": 2.1803205013275146, + "learning_rate": 4.025132888101404e-06, + "loss": 2.4262937545776366, + "step": 293520 + }, + { + "epoch": 0.7990744521573432, + "grad_norm": 2.2083895206451416, + "learning_rate": 4.02404252419245e-06, + "loss": 2.4545572280883787, + "step": 293540 + }, + { + "epoch": 0.7991288961480877, + "grad_norm": 2.222836494445801, + "learning_rate": 4.022952160283495e-06, + "loss": 2.4767938613891602, + "step": 293560 + }, + { + "epoch": 0.7991833401388322, + "grad_norm": 2.1217362880706787, + "learning_rate": 4.0218617963745405e-06, + "loss": 2.46533317565918, + "step": 293580 + }, + { + "epoch": 0.7992377841295767, + "grad_norm": 3.098860502243042, + "learning_rate": 4.0207714324655855e-06, + "loss": 2.463370704650879, + "step": 293600 + }, + { + "epoch": 0.7992922281203212, + "grad_norm": 2.1433537006378174, + "learning_rate": 4.019681068556631e-06, + "loss": 2.481301689147949, + "step": 293620 + }, + { + "epoch": 0.7993466721110657, + "grad_norm": 2.119786500930786, + "learning_rate": 4.018590704647676e-06, + "loss": 2.473905563354492, + "step": 293640 + }, + { + "epoch": 0.7994011161018103, + "grad_norm": 2.161344528198242, + "learning_rate": 4.017500340738722e-06, + "loss": 2.4656139373779298, + "step": 293660 + }, + { + "epoch": 0.7994555600925548, + "grad_norm": 2.738086462020874, + "learning_rate": 4.016409976829767e-06, + "loss": 2.4769962310791014, + "step": 293680 + }, + { + "epoch": 0.7995100040832993, + "grad_norm": 2.2285499572753906, + "learning_rate": 4.015319612920813e-06, + "loss": 2.4883113861083985, + "step": 293700 + }, + { + "epoch": 0.7995644480740438, + "grad_norm": 2.100851535797119, + "learning_rate": 4.014229249011858e-06, + "loss": 2.463359832763672, + "step": 293720 + }, + { + "epoch": 0.7996188920647883, + "grad_norm": 2.145179033279419, + "learning_rate": 4.013138885102904e-06, + "loss": 2.4316307067871095, + "step": 293740 + }, + { + "epoch": 0.7996733360555329, + "grad_norm": 2.2465016841888428, + "learning_rate": 4.0120485211939485e-06, + "loss": 2.4956100463867186, + "step": 293760 + }, + { + "epoch": 0.7997277800462774, + "grad_norm": 2.2786035537719727, + "learning_rate": 4.010958157284994e-06, + "loss": 2.4441448211669923, + "step": 293780 + }, + { + "epoch": 0.7997822240370219, + "grad_norm": 2.1506996154785156, + "learning_rate": 4.009867793376039e-06, + "loss": 2.4852920532226563, + "step": 293800 + }, + { + "epoch": 0.7998366680277664, + "grad_norm": 2.2275047302246094, + "learning_rate": 4.008777429467085e-06, + "loss": 2.4339548110961915, + "step": 293820 + }, + { + "epoch": 0.7998911120185109, + "grad_norm": 2.262197494506836, + "learning_rate": 4.00768706555813e-06, + "loss": 2.4556875228881836, + "step": 293840 + }, + { + "epoch": 0.7999455560092554, + "grad_norm": 2.13226580619812, + "learning_rate": 4.006596701649176e-06, + "loss": 2.4593429565429688, + "step": 293860 + }, + { + "epoch": 0.8, + "grad_norm": 2.204904794692993, + "learning_rate": 4.005506337740221e-06, + "loss": 2.459553909301758, + "step": 293880 + }, + { + "epoch": 0.8000544439907445, + "grad_norm": 2.0763227939605713, + "learning_rate": 4.004415973831267e-06, + "loss": 2.4714195251464846, + "step": 293900 + }, + { + "epoch": 0.800108887981489, + "grad_norm": 2.2150168418884277, + "learning_rate": 4.003325609922312e-06, + "loss": 2.47412166595459, + "step": 293920 + }, + { + "epoch": 0.8001633319722335, + "grad_norm": 2.3156332969665527, + "learning_rate": 4.0022352460133574e-06, + "loss": 2.452288055419922, + "step": 293940 + }, + { + "epoch": 0.800217775962978, + "grad_norm": 2.2086613178253174, + "learning_rate": 4.001144882104402e-06, + "loss": 2.4448570251464843, + "step": 293960 + }, + { + "epoch": 0.8002722199537227, + "grad_norm": 2.136932134628296, + "learning_rate": 4.000054518195448e-06, + "loss": 2.4508237838745117, + "step": 293980 + }, + { + "epoch": 0.8003266639444672, + "grad_norm": 2.290703296661377, + "learning_rate": 3.998964154286493e-06, + "loss": 2.4576141357421877, + "step": 294000 + }, + { + "epoch": 0.8003811079352117, + "grad_norm": 2.147207260131836, + "learning_rate": 3.997873790377539e-06, + "loss": 2.4498767852783203, + "step": 294020 + }, + { + "epoch": 0.8004355519259562, + "grad_norm": 2.0824429988861084, + "learning_rate": 3.996783426468584e-06, + "loss": 2.469449996948242, + "step": 294040 + }, + { + "epoch": 0.8004899959167007, + "grad_norm": 2.439044713973999, + "learning_rate": 3.99569306255963e-06, + "loss": 2.479676628112793, + "step": 294060 + }, + { + "epoch": 0.8005444399074452, + "grad_norm": 2.1798768043518066, + "learning_rate": 3.994602698650675e-06, + "loss": 2.4251735687255858, + "step": 294080 + }, + { + "epoch": 0.8005988838981898, + "grad_norm": 2.1771984100341797, + "learning_rate": 3.9935123347417205e-06, + "loss": 2.459973907470703, + "step": 294100 + }, + { + "epoch": 0.8006533278889343, + "grad_norm": 2.1887855529785156, + "learning_rate": 3.9924219708327655e-06, + "loss": 2.455340576171875, + "step": 294120 + }, + { + "epoch": 0.8007077718796788, + "grad_norm": 2.146658182144165, + "learning_rate": 3.991331606923811e-06, + "loss": 2.4665550231933593, + "step": 294140 + }, + { + "epoch": 0.8007622158704233, + "grad_norm": 2.194070339202881, + "learning_rate": 3.990241243014856e-06, + "loss": 2.4419734954833983, + "step": 294160 + }, + { + "epoch": 0.8008166598611678, + "grad_norm": 2.312904119491577, + "learning_rate": 3.989150879105902e-06, + "loss": 2.448651123046875, + "step": 294180 + }, + { + "epoch": 0.8008711038519123, + "grad_norm": 2.1421823501586914, + "learning_rate": 3.988060515196947e-06, + "loss": 2.4955718994140623, + "step": 294200 + }, + { + "epoch": 0.8009255478426569, + "grad_norm": 2.29695463180542, + "learning_rate": 3.986970151287993e-06, + "loss": 2.442241096496582, + "step": 294220 + }, + { + "epoch": 0.8009799918334014, + "grad_norm": 2.249741315841675, + "learning_rate": 3.985879787379039e-06, + "loss": 2.50671443939209, + "step": 294240 + }, + { + "epoch": 0.8010344358241459, + "grad_norm": 2.198955774307251, + "learning_rate": 3.984789423470084e-06, + "loss": 2.4876216888427733, + "step": 294260 + }, + { + "epoch": 0.8010888798148904, + "grad_norm": 2.2340052127838135, + "learning_rate": 3.983699059561129e-06, + "loss": 2.4545804977416994, + "step": 294280 + }, + { + "epoch": 0.8011433238056349, + "grad_norm": 2.1195871829986572, + "learning_rate": 3.982608695652174e-06, + "loss": 2.4683433532714845, + "step": 294300 + }, + { + "epoch": 0.8011977677963795, + "grad_norm": 2.139873743057251, + "learning_rate": 3.98151833174322e-06, + "loss": 2.454142761230469, + "step": 294320 + }, + { + "epoch": 0.801252211787124, + "grad_norm": 2.1782801151275635, + "learning_rate": 3.980427967834265e-06, + "loss": 2.4721961975097657, + "step": 294340 + }, + { + "epoch": 0.8013066557778685, + "grad_norm": 2.1184732913970947, + "learning_rate": 3.979337603925311e-06, + "loss": 2.441437339782715, + "step": 294360 + }, + { + "epoch": 0.801361099768613, + "grad_norm": 2.1061534881591797, + "learning_rate": 3.978247240016356e-06, + "loss": 2.477140998840332, + "step": 294380 + }, + { + "epoch": 0.8014155437593575, + "grad_norm": 2.0968782901763916, + "learning_rate": 3.977156876107402e-06, + "loss": 2.4475269317626953, + "step": 294400 + }, + { + "epoch": 0.801469987750102, + "grad_norm": 2.1756861209869385, + "learning_rate": 3.976066512198447e-06, + "loss": 2.4887262344360352, + "step": 294420 + }, + { + "epoch": 0.8015244317408466, + "grad_norm": 2.149304151535034, + "learning_rate": 3.9749761482894925e-06, + "loss": 2.4924072265625, + "step": 294440 + }, + { + "epoch": 0.8015788757315911, + "grad_norm": 2.2509360313415527, + "learning_rate": 3.9738857843805375e-06, + "loss": 2.4511459350585936, + "step": 294460 + }, + { + "epoch": 0.8016333197223356, + "grad_norm": 2.152989625930786, + "learning_rate": 3.972795420471583e-06, + "loss": 2.4646448135375976, + "step": 294480 + }, + { + "epoch": 0.8016877637130801, + "grad_norm": 2.1163268089294434, + "learning_rate": 3.971705056562628e-06, + "loss": 2.4880340576171873, + "step": 294500 + }, + { + "epoch": 0.8017422077038246, + "grad_norm": 2.1361312866210938, + "learning_rate": 3.970614692653674e-06, + "loss": 2.495440673828125, + "step": 294520 + }, + { + "epoch": 0.8017966516945693, + "grad_norm": 2.2109036445617676, + "learning_rate": 3.969524328744719e-06, + "loss": 2.4847156524658205, + "step": 294540 + }, + { + "epoch": 0.8018510956853138, + "grad_norm": 2.1171600818634033, + "learning_rate": 3.968433964835765e-06, + "loss": 2.4884365081787108, + "step": 294560 + }, + { + "epoch": 0.8019055396760583, + "grad_norm": 2.192814826965332, + "learning_rate": 3.96734360092681e-06, + "loss": 2.4621679306030275, + "step": 294580 + }, + { + "epoch": 0.8019599836668028, + "grad_norm": 2.195127248764038, + "learning_rate": 3.966253237017856e-06, + "loss": 2.466383171081543, + "step": 294600 + }, + { + "epoch": 0.8020144276575473, + "grad_norm": 2.026430130004883, + "learning_rate": 3.9651628731089005e-06, + "loss": 2.4326854705810548, + "step": 294620 + }, + { + "epoch": 0.8020688716482918, + "grad_norm": 2.1961731910705566, + "learning_rate": 3.964072509199946e-06, + "loss": 2.457748603820801, + "step": 294640 + }, + { + "epoch": 0.8021233156390364, + "grad_norm": 2.1502904891967773, + "learning_rate": 3.962982145290991e-06, + "loss": 2.4424617767333983, + "step": 294660 + }, + { + "epoch": 0.8021777596297809, + "grad_norm": 2.259658098220825, + "learning_rate": 3.961891781382037e-06, + "loss": 2.434810256958008, + "step": 294680 + }, + { + "epoch": 0.8022322036205254, + "grad_norm": 2.1996278762817383, + "learning_rate": 3.960801417473082e-06, + "loss": 2.4471330642700195, + "step": 294700 + }, + { + "epoch": 0.8022866476112699, + "grad_norm": 2.2168166637420654, + "learning_rate": 3.959711053564128e-06, + "loss": 2.485653877258301, + "step": 294720 + }, + { + "epoch": 0.8023410916020144, + "grad_norm": 2.130722761154175, + "learning_rate": 3.958620689655173e-06, + "loss": 2.493173599243164, + "step": 294740 + }, + { + "epoch": 0.802395535592759, + "grad_norm": 2.147111654281616, + "learning_rate": 3.957530325746219e-06, + "loss": 2.4543970108032225, + "step": 294760 + }, + { + "epoch": 0.8024499795835035, + "grad_norm": 2.1184353828430176, + "learning_rate": 3.956439961837264e-06, + "loss": 2.4645788192749025, + "step": 294780 + }, + { + "epoch": 0.802504423574248, + "grad_norm": 2.402848958969116, + "learning_rate": 3.9553495979283094e-06, + "loss": 2.491629981994629, + "step": 294800 + }, + { + "epoch": 0.8025588675649925, + "grad_norm": 2.2152931690216064, + "learning_rate": 3.954259234019354e-06, + "loss": 2.4593694686889647, + "step": 294820 + }, + { + "epoch": 0.802613311555737, + "grad_norm": 2.209665536880493, + "learning_rate": 3.9531688701104e-06, + "loss": 2.483359527587891, + "step": 294840 + }, + { + "epoch": 0.8026677555464815, + "grad_norm": 2.194638252258301, + "learning_rate": 3.952078506201445e-06, + "loss": 2.4707000732421873, + "step": 294860 + }, + { + "epoch": 0.8027221995372261, + "grad_norm": 2.315463066101074, + "learning_rate": 3.950988142292491e-06, + "loss": 2.4395519256591798, + "step": 294880 + }, + { + "epoch": 0.8027766435279706, + "grad_norm": 2.183377265930176, + "learning_rate": 3.949897778383536e-06, + "loss": 2.4717243194580076, + "step": 294900 + }, + { + "epoch": 0.8028310875187151, + "grad_norm": 2.4767556190490723, + "learning_rate": 3.948807414474582e-06, + "loss": 2.43975887298584, + "step": 294920 + }, + { + "epoch": 0.8028855315094596, + "grad_norm": 2.466430187225342, + "learning_rate": 3.947717050565627e-06, + "loss": 2.4431943893432617, + "step": 294940 + }, + { + "epoch": 0.8029399755002041, + "grad_norm": 2.1566250324249268, + "learning_rate": 3.9466266866566725e-06, + "loss": 2.454777145385742, + "step": 294960 + }, + { + "epoch": 0.8029944194909487, + "grad_norm": 2.1844234466552734, + "learning_rate": 3.9455363227477175e-06, + "loss": 2.458016014099121, + "step": 294980 + }, + { + "epoch": 0.8030488634816932, + "grad_norm": 2.3793530464172363, + "learning_rate": 3.944445958838763e-06, + "loss": 2.4823015213012694, + "step": 295000 + }, + { + "epoch": 0.8031033074724377, + "grad_norm": 2.1413180828094482, + "learning_rate": 3.943355594929808e-06, + "loss": 2.463953971862793, + "step": 295020 + }, + { + "epoch": 0.8031577514631822, + "grad_norm": 2.27536940574646, + "learning_rate": 3.942265231020854e-06, + "loss": 2.4992908477783202, + "step": 295040 + }, + { + "epoch": 0.8032121954539267, + "grad_norm": 2.2564940452575684, + "learning_rate": 3.941174867111899e-06, + "loss": 2.451290512084961, + "step": 295060 + }, + { + "epoch": 0.8032666394446712, + "grad_norm": 2.176689624786377, + "learning_rate": 3.940084503202945e-06, + "loss": 2.4313940048217773, + "step": 295080 + }, + { + "epoch": 0.8033210834354159, + "grad_norm": 2.17814040184021, + "learning_rate": 3.93899413929399e-06, + "loss": 2.446146011352539, + "step": 295100 + }, + { + "epoch": 0.8033755274261604, + "grad_norm": 2.2621283531188965, + "learning_rate": 3.937903775385036e-06, + "loss": 2.458536911010742, + "step": 295120 + }, + { + "epoch": 0.8034299714169049, + "grad_norm": 2.151034116744995, + "learning_rate": 3.936813411476081e-06, + "loss": 2.4708478927612303, + "step": 295140 + }, + { + "epoch": 0.8034844154076494, + "grad_norm": 2.229539632797241, + "learning_rate": 3.935723047567126e-06, + "loss": 2.4510358810424804, + "step": 295160 + }, + { + "epoch": 0.8035388593983939, + "grad_norm": 2.1541709899902344, + "learning_rate": 3.934632683658171e-06, + "loss": 2.467909049987793, + "step": 295180 + }, + { + "epoch": 0.8035933033891384, + "grad_norm": 2.2268004417419434, + "learning_rate": 3.933542319749217e-06, + "loss": 2.4737627029418947, + "step": 295200 + }, + { + "epoch": 0.803647747379883, + "grad_norm": 2.2851996421813965, + "learning_rate": 3.932451955840262e-06, + "loss": 2.455154228210449, + "step": 295220 + }, + { + "epoch": 0.8037021913706275, + "grad_norm": 2.244736433029175, + "learning_rate": 3.931361591931308e-06, + "loss": 2.483885383605957, + "step": 295240 + }, + { + "epoch": 0.803756635361372, + "grad_norm": 2.1964926719665527, + "learning_rate": 3.930271228022353e-06, + "loss": 2.464681625366211, + "step": 295260 + }, + { + "epoch": 0.8038110793521165, + "grad_norm": 2.2964000701904297, + "learning_rate": 3.929180864113398e-06, + "loss": 2.442915916442871, + "step": 295280 + }, + { + "epoch": 0.803865523342861, + "grad_norm": 2.2568349838256836, + "learning_rate": 3.928090500204444e-06, + "loss": 2.4489826202392577, + "step": 295300 + }, + { + "epoch": 0.8039199673336056, + "grad_norm": 2.127201795578003, + "learning_rate": 3.927000136295489e-06, + "loss": 2.4569934844970702, + "step": 295320 + }, + { + "epoch": 0.8039744113243501, + "grad_norm": 2.0984621047973633, + "learning_rate": 3.9259097723865344e-06, + "loss": 2.4649057388305664, + "step": 295340 + }, + { + "epoch": 0.8040288553150946, + "grad_norm": 2.1042258739471436, + "learning_rate": 3.924819408477579e-06, + "loss": 2.4376953125, + "step": 295360 + }, + { + "epoch": 0.8040832993058391, + "grad_norm": 2.0707197189331055, + "learning_rate": 3.923729044568625e-06, + "loss": 2.4756677627563475, + "step": 295380 + }, + { + "epoch": 0.8041377432965836, + "grad_norm": 2.2689695358276367, + "learning_rate": 3.92263868065967e-06, + "loss": 2.485275650024414, + "step": 295400 + }, + { + "epoch": 0.8041921872873281, + "grad_norm": 2.103736400604248, + "learning_rate": 3.921548316750716e-06, + "loss": 2.464661407470703, + "step": 295420 + }, + { + "epoch": 0.8042466312780727, + "grad_norm": 2.1631619930267334, + "learning_rate": 3.920457952841761e-06, + "loss": 2.488402557373047, + "step": 295440 + }, + { + "epoch": 0.8043010752688172, + "grad_norm": 2.042888641357422, + "learning_rate": 3.919367588932807e-06, + "loss": 2.4662113189697266, + "step": 295460 + }, + { + "epoch": 0.8043555192595617, + "grad_norm": 2.1936557292938232, + "learning_rate": 3.918277225023852e-06, + "loss": 2.4768836975097654, + "step": 295480 + }, + { + "epoch": 0.8044099632503062, + "grad_norm": 2.498051881790161, + "learning_rate": 3.9171868611148975e-06, + "loss": 2.439731979370117, + "step": 295500 + }, + { + "epoch": 0.8044644072410507, + "grad_norm": 2.180732250213623, + "learning_rate": 3.9160964972059425e-06, + "loss": 2.426557922363281, + "step": 295520 + }, + { + "epoch": 0.8045188512317953, + "grad_norm": 2.1765708923339844, + "learning_rate": 3.915006133296988e-06, + "loss": 2.4373991012573244, + "step": 295540 + }, + { + "epoch": 0.8045732952225398, + "grad_norm": 2.1897573471069336, + "learning_rate": 3.913915769388033e-06, + "loss": 2.452822685241699, + "step": 295560 + }, + { + "epoch": 0.8046277392132843, + "grad_norm": 2.1377227306365967, + "learning_rate": 3.912825405479079e-06, + "loss": 2.453508758544922, + "step": 295580 + }, + { + "epoch": 0.8046821832040288, + "grad_norm": 2.1434614658355713, + "learning_rate": 3.911735041570124e-06, + "loss": 2.474075126647949, + "step": 295600 + }, + { + "epoch": 0.8047366271947733, + "grad_norm": 2.144958257675171, + "learning_rate": 3.91064467766117e-06, + "loss": 2.4377296447753904, + "step": 295620 + }, + { + "epoch": 0.8047910711855178, + "grad_norm": 2.220808506011963, + "learning_rate": 3.909554313752215e-06, + "loss": 2.4829261779785154, + "step": 295640 + }, + { + "epoch": 0.8048455151762625, + "grad_norm": 2.376037359237671, + "learning_rate": 3.908463949843261e-06, + "loss": 2.472855567932129, + "step": 295660 + }, + { + "epoch": 0.804899959167007, + "grad_norm": 2.185885190963745, + "learning_rate": 3.9073735859343056e-06, + "loss": 2.4496498107910156, + "step": 295680 + }, + { + "epoch": 0.8049544031577515, + "grad_norm": 2.3660337924957275, + "learning_rate": 3.906283222025351e-06, + "loss": 2.446150016784668, + "step": 295700 + }, + { + "epoch": 0.805008847148496, + "grad_norm": 2.0951664447784424, + "learning_rate": 3.905192858116396e-06, + "loss": 2.4751655578613283, + "step": 295720 + }, + { + "epoch": 0.8050632911392405, + "grad_norm": 3.093782663345337, + "learning_rate": 3.904102494207442e-06, + "loss": 2.4553110122680666, + "step": 295740 + }, + { + "epoch": 0.8051177351299851, + "grad_norm": 2.1869900226593018, + "learning_rate": 3.903012130298487e-06, + "loss": 2.452750587463379, + "step": 295760 + }, + { + "epoch": 0.8051721791207296, + "grad_norm": 2.153716564178467, + "learning_rate": 3.901921766389533e-06, + "loss": 2.4648015975952147, + "step": 295780 + }, + { + "epoch": 0.8052266231114741, + "grad_norm": 2.1507582664489746, + "learning_rate": 3.900831402480578e-06, + "loss": 2.475617027282715, + "step": 295800 + }, + { + "epoch": 0.8052810671022186, + "grad_norm": 2.2351925373077393, + "learning_rate": 3.899741038571624e-06, + "loss": 2.44527587890625, + "step": 295820 + }, + { + "epoch": 0.8053355110929631, + "grad_norm": 2.199040174484253, + "learning_rate": 3.898650674662669e-06, + "loss": 2.451358413696289, + "step": 295840 + }, + { + "epoch": 0.8053899550837076, + "grad_norm": 2.2451560497283936, + "learning_rate": 3.8975603107537145e-06, + "loss": 2.437290573120117, + "step": 295860 + }, + { + "epoch": 0.8054443990744522, + "grad_norm": 2.1306192874908447, + "learning_rate": 3.8964699468447594e-06, + "loss": 2.487727737426758, + "step": 295880 + }, + { + "epoch": 0.8054988430651967, + "grad_norm": 2.2561566829681396, + "learning_rate": 3.895379582935805e-06, + "loss": 2.464097595214844, + "step": 295900 + }, + { + "epoch": 0.8055532870559412, + "grad_norm": 2.1089894771575928, + "learning_rate": 3.89428921902685e-06, + "loss": 2.4894603729248046, + "step": 295920 + }, + { + "epoch": 0.8056077310466857, + "grad_norm": 2.229572296142578, + "learning_rate": 3.893198855117896e-06, + "loss": 2.4343711853027346, + "step": 295940 + }, + { + "epoch": 0.8056621750374302, + "grad_norm": 2.1809253692626953, + "learning_rate": 3.892108491208941e-06, + "loss": 2.463169479370117, + "step": 295960 + }, + { + "epoch": 0.8057166190281748, + "grad_norm": 2.219625473022461, + "learning_rate": 3.891018127299987e-06, + "loss": 2.4648508071899413, + "step": 295980 + }, + { + "epoch": 0.8057710630189193, + "grad_norm": 2.1357421875, + "learning_rate": 3.889927763391032e-06, + "loss": 2.4974788665771483, + "step": 296000 + }, + { + "epoch": 0.8058255070096638, + "grad_norm": 2.384037733078003, + "learning_rate": 3.8888373994820775e-06, + "loss": 2.442979621887207, + "step": 296020 + }, + { + "epoch": 0.8058799510004083, + "grad_norm": 2.1886911392211914, + "learning_rate": 3.8877470355731225e-06, + "loss": 2.4669826507568358, + "step": 296040 + }, + { + "epoch": 0.8059343949911528, + "grad_norm": 2.532080888748169, + "learning_rate": 3.886656671664168e-06, + "loss": 2.4429862976074217, + "step": 296060 + }, + { + "epoch": 0.8059888389818973, + "grad_norm": 2.201733112335205, + "learning_rate": 3.885566307755213e-06, + "loss": 2.4742204666137697, + "step": 296080 + }, + { + "epoch": 0.8060432829726419, + "grad_norm": 2.159026861190796, + "learning_rate": 3.884475943846259e-06, + "loss": 2.4517641067504883, + "step": 296100 + }, + { + "epoch": 0.8060977269633864, + "grad_norm": 2.190856695175171, + "learning_rate": 3.883385579937304e-06, + "loss": 2.4650495529174803, + "step": 296120 + }, + { + "epoch": 0.8061521709541309, + "grad_norm": 2.1939613819122314, + "learning_rate": 3.88229521602835e-06, + "loss": 2.4323703765869142, + "step": 296140 + }, + { + "epoch": 0.8062066149448754, + "grad_norm": 2.1688497066497803, + "learning_rate": 3.881204852119395e-06, + "loss": 2.4639989852905275, + "step": 296160 + }, + { + "epoch": 0.80626105893562, + "grad_norm": 2.2544772624969482, + "learning_rate": 3.880114488210441e-06, + "loss": 2.4570926666259765, + "step": 296180 + }, + { + "epoch": 0.8063155029263646, + "grad_norm": 2.1263327598571777, + "learning_rate": 3.879024124301486e-06, + "loss": 2.446055030822754, + "step": 296200 + }, + { + "epoch": 0.8063699469171091, + "grad_norm": 2.044301986694336, + "learning_rate": 3.877933760392531e-06, + "loss": 2.4724105834960937, + "step": 296220 + }, + { + "epoch": 0.8064243909078536, + "grad_norm": 2.2533047199249268, + "learning_rate": 3.876843396483576e-06, + "loss": 2.469990921020508, + "step": 296240 + }, + { + "epoch": 0.8064788348985981, + "grad_norm": 2.1997597217559814, + "learning_rate": 3.875753032574622e-06, + "loss": 2.4379270553588865, + "step": 296260 + }, + { + "epoch": 0.8065332788893426, + "grad_norm": 2.2672224044799805, + "learning_rate": 3.874662668665667e-06, + "loss": 2.4781639099121096, + "step": 296280 + }, + { + "epoch": 0.8065877228800871, + "grad_norm": 2.092061758041382, + "learning_rate": 3.873572304756713e-06, + "loss": 2.4720956802368166, + "step": 296300 + }, + { + "epoch": 0.8066421668708317, + "grad_norm": 2.214001178741455, + "learning_rate": 3.872481940847758e-06, + "loss": 2.439854621887207, + "step": 296320 + }, + { + "epoch": 0.8066966108615762, + "grad_norm": 2.1752734184265137, + "learning_rate": 3.871391576938804e-06, + "loss": 2.4809778213500975, + "step": 296340 + }, + { + "epoch": 0.8067510548523207, + "grad_norm": 2.1934688091278076, + "learning_rate": 3.870301213029849e-06, + "loss": 2.4609712600708007, + "step": 296360 + }, + { + "epoch": 0.8068054988430652, + "grad_norm": 2.1610517501831055, + "learning_rate": 3.8692108491208945e-06, + "loss": 2.439749908447266, + "step": 296380 + }, + { + "epoch": 0.8068599428338097, + "grad_norm": 2.186983108520508, + "learning_rate": 3.8681204852119395e-06, + "loss": 2.4643838882446287, + "step": 296400 + }, + { + "epoch": 0.8069143868245542, + "grad_norm": 2.1866321563720703, + "learning_rate": 3.867030121302985e-06, + "loss": 2.4708763122558595, + "step": 296420 + }, + { + "epoch": 0.8069688308152988, + "grad_norm": 2.1953318119049072, + "learning_rate": 3.86593975739403e-06, + "loss": 2.489318084716797, + "step": 296440 + }, + { + "epoch": 0.8070232748060433, + "grad_norm": 2.123492956161499, + "learning_rate": 3.864849393485076e-06, + "loss": 2.4351499557495115, + "step": 296460 + }, + { + "epoch": 0.8070777187967878, + "grad_norm": 2.432128667831421, + "learning_rate": 3.863759029576121e-06, + "loss": 2.4574163436889647, + "step": 296480 + }, + { + "epoch": 0.8071321627875323, + "grad_norm": 2.4030299186706543, + "learning_rate": 3.862668665667167e-06, + "loss": 2.4559562683105467, + "step": 296500 + }, + { + "epoch": 0.8071866067782768, + "grad_norm": 2.1592798233032227, + "learning_rate": 3.861578301758212e-06, + "loss": 2.44708137512207, + "step": 296520 + }, + { + "epoch": 0.8072410507690214, + "grad_norm": 2.1663978099823, + "learning_rate": 3.8604879378492576e-06, + "loss": 2.4807357788085938, + "step": 296540 + }, + { + "epoch": 0.8072954947597659, + "grad_norm": 2.0887606143951416, + "learning_rate": 3.8593975739403025e-06, + "loss": 2.4407039642333985, + "step": 296560 + }, + { + "epoch": 0.8073499387505104, + "grad_norm": 2.112077474594116, + "learning_rate": 3.858307210031348e-06, + "loss": 2.455720329284668, + "step": 296580 + }, + { + "epoch": 0.8074043827412549, + "grad_norm": 2.1429779529571533, + "learning_rate": 3.857216846122393e-06, + "loss": 2.498796272277832, + "step": 296600 + }, + { + "epoch": 0.8074588267319994, + "grad_norm": 2.185004234313965, + "learning_rate": 3.856126482213439e-06, + "loss": 2.4758579254150392, + "step": 296620 + }, + { + "epoch": 0.8075132707227439, + "grad_norm": 2.2579610347747803, + "learning_rate": 3.855036118304484e-06, + "loss": 2.461013412475586, + "step": 296640 + }, + { + "epoch": 0.8075677147134885, + "grad_norm": 2.182508707046509, + "learning_rate": 3.85394575439553e-06, + "loss": 2.4440893173217773, + "step": 296660 + }, + { + "epoch": 0.807622158704233, + "grad_norm": 2.10155987739563, + "learning_rate": 3.852855390486575e-06, + "loss": 2.4487159729003904, + "step": 296680 + }, + { + "epoch": 0.8076766026949775, + "grad_norm": 2.2047104835510254, + "learning_rate": 3.851765026577621e-06, + "loss": 2.4889347076416017, + "step": 296700 + }, + { + "epoch": 0.807731046685722, + "grad_norm": 2.241265296936035, + "learning_rate": 3.850674662668666e-06, + "loss": 2.4950492858886717, + "step": 296720 + }, + { + "epoch": 0.8077854906764665, + "grad_norm": 2.281447172164917, + "learning_rate": 3.8495842987597114e-06, + "loss": 2.4773021697998048, + "step": 296740 + }, + { + "epoch": 0.8078399346672112, + "grad_norm": 2.166266679763794, + "learning_rate": 3.848493934850756e-06, + "loss": 2.4972904205322264, + "step": 296760 + }, + { + "epoch": 0.8078943786579557, + "grad_norm": 2.6207504272460938, + "learning_rate": 3.847403570941802e-06, + "loss": 2.4712791442871094, + "step": 296780 + }, + { + "epoch": 0.8079488226487002, + "grad_norm": 2.213988780975342, + "learning_rate": 3.846313207032847e-06, + "loss": 2.4659976959228516, + "step": 296800 + }, + { + "epoch": 0.8080032666394447, + "grad_norm": 2.202003240585327, + "learning_rate": 3.845222843123893e-06, + "loss": 2.4511337280273438, + "step": 296820 + }, + { + "epoch": 0.8080577106301892, + "grad_norm": 2.203089714050293, + "learning_rate": 3.844132479214938e-06, + "loss": 2.435023880004883, + "step": 296840 + }, + { + "epoch": 0.8081121546209337, + "grad_norm": 2.214477062225342, + "learning_rate": 3.843042115305984e-06, + "loss": 2.4600414276123046, + "step": 296860 + }, + { + "epoch": 0.8081665986116783, + "grad_norm": 2.2245423793792725, + "learning_rate": 3.841951751397029e-06, + "loss": 2.484920692443848, + "step": 296880 + }, + { + "epoch": 0.8082210426024228, + "grad_norm": 2.163802146911621, + "learning_rate": 3.8408613874880745e-06, + "loss": 2.4745731353759766, + "step": 296900 + }, + { + "epoch": 0.8082754865931673, + "grad_norm": 2.161470651626587, + "learning_rate": 3.8397710235791195e-06, + "loss": 2.474041748046875, + "step": 296920 + }, + { + "epoch": 0.8083299305839118, + "grad_norm": 2.171327590942383, + "learning_rate": 3.838680659670165e-06, + "loss": 2.443278694152832, + "step": 296940 + }, + { + "epoch": 0.8083843745746563, + "grad_norm": 2.1576926708221436, + "learning_rate": 3.83759029576121e-06, + "loss": 2.477937698364258, + "step": 296960 + }, + { + "epoch": 0.8084388185654009, + "grad_norm": 2.1688075065612793, + "learning_rate": 3.836499931852256e-06, + "loss": 2.449773979187012, + "step": 296980 + }, + { + "epoch": 0.8084932625561454, + "grad_norm": 2.1049094200134277, + "learning_rate": 3.835409567943301e-06, + "loss": 2.4749053955078124, + "step": 297000 + }, + { + "epoch": 0.8085477065468899, + "grad_norm": 2.2321629524230957, + "learning_rate": 3.834319204034347e-06, + "loss": 2.520363426208496, + "step": 297020 + }, + { + "epoch": 0.8086021505376344, + "grad_norm": 2.1068620681762695, + "learning_rate": 3.833228840125392e-06, + "loss": 2.451420211791992, + "step": 297040 + }, + { + "epoch": 0.8086565945283789, + "grad_norm": 2.14387583732605, + "learning_rate": 3.832138476216438e-06, + "loss": 2.490779685974121, + "step": 297060 + }, + { + "epoch": 0.8087110385191234, + "grad_norm": 2.17561674118042, + "learning_rate": 3.8310481123074826e-06, + "loss": 2.48048095703125, + "step": 297080 + }, + { + "epoch": 0.808765482509868, + "grad_norm": 2.222903251647949, + "learning_rate": 3.829957748398528e-06, + "loss": 2.4659711837768556, + "step": 297100 + }, + { + "epoch": 0.8088199265006125, + "grad_norm": 2.181581974029541, + "learning_rate": 3.828867384489573e-06, + "loss": 2.4384105682373045, + "step": 297120 + }, + { + "epoch": 0.808874370491357, + "grad_norm": 3.156251907348633, + "learning_rate": 3.827777020580619e-06, + "loss": 2.479769706726074, + "step": 297140 + }, + { + "epoch": 0.8089288144821015, + "grad_norm": 2.070291757583618, + "learning_rate": 3.826686656671664e-06, + "loss": 2.45157413482666, + "step": 297160 + }, + { + "epoch": 0.808983258472846, + "grad_norm": 2.2920308113098145, + "learning_rate": 3.82559629276271e-06, + "loss": 2.448743438720703, + "step": 297180 + }, + { + "epoch": 0.8090377024635906, + "grad_norm": 2.169417142868042, + "learning_rate": 3.824505928853755e-06, + "loss": 2.468251609802246, + "step": 297200 + }, + { + "epoch": 0.8090921464543351, + "grad_norm": 2.1502604484558105, + "learning_rate": 3.823415564944801e-06, + "loss": 2.423567008972168, + "step": 297220 + }, + { + "epoch": 0.8091465904450796, + "grad_norm": 2.137403726577759, + "learning_rate": 3.822325201035846e-06, + "loss": 2.446335220336914, + "step": 297240 + }, + { + "epoch": 0.8092010344358241, + "grad_norm": 2.1782524585723877, + "learning_rate": 3.8212348371268915e-06, + "loss": 2.4599941253662108, + "step": 297260 + }, + { + "epoch": 0.8092554784265686, + "grad_norm": 2.145338296890259, + "learning_rate": 3.820144473217936e-06, + "loss": 2.4493579864501953, + "step": 297280 + }, + { + "epoch": 0.8093099224173131, + "grad_norm": 2.4886932373046875, + "learning_rate": 3.819054109308982e-06, + "loss": 2.4692859649658203, + "step": 297300 + }, + { + "epoch": 0.8093643664080578, + "grad_norm": 2.1334633827209473, + "learning_rate": 3.817963745400027e-06, + "loss": 2.444925308227539, + "step": 297320 + }, + { + "epoch": 0.8094188103988023, + "grad_norm": 2.141392946243286, + "learning_rate": 3.816873381491073e-06, + "loss": 2.4726032257080077, + "step": 297340 + }, + { + "epoch": 0.8094732543895468, + "grad_norm": 2.164663314819336, + "learning_rate": 3.815783017582118e-06, + "loss": 2.4751548767089844, + "step": 297360 + }, + { + "epoch": 0.8095276983802913, + "grad_norm": 2.1797056198120117, + "learning_rate": 3.8146926536731638e-06, + "loss": 2.4679567337036135, + "step": 297380 + }, + { + "epoch": 0.8095821423710358, + "grad_norm": 2.192995071411133, + "learning_rate": 3.8136022897642087e-06, + "loss": 2.4292461395263674, + "step": 297400 + }, + { + "epoch": 0.8096365863617803, + "grad_norm": 2.3680834770202637, + "learning_rate": 3.8125119258552545e-06, + "loss": 2.4623249053955076, + "step": 297420 + }, + { + "epoch": 0.8096910303525249, + "grad_norm": 2.445474624633789, + "learning_rate": 3.8114215619462995e-06, + "loss": 2.4846385955810546, + "step": 297440 + }, + { + "epoch": 0.8097454743432694, + "grad_norm": 2.195136070251465, + "learning_rate": 3.8103311980373453e-06, + "loss": 2.464842414855957, + "step": 297460 + }, + { + "epoch": 0.8097999183340139, + "grad_norm": 2.1019320487976074, + "learning_rate": 3.8092408341283903e-06, + "loss": 2.47759895324707, + "step": 297480 + }, + { + "epoch": 0.8098543623247584, + "grad_norm": 2.161297082901001, + "learning_rate": 3.808150470219436e-06, + "loss": 2.4550941467285154, + "step": 297500 + }, + { + "epoch": 0.8099088063155029, + "grad_norm": 2.2438225746154785, + "learning_rate": 3.807060106310481e-06, + "loss": 2.4856164932250975, + "step": 297520 + }, + { + "epoch": 0.8099632503062475, + "grad_norm": 2.1351375579833984, + "learning_rate": 3.805969742401527e-06, + "loss": 2.481258201599121, + "step": 297540 + }, + { + "epoch": 0.810017694296992, + "grad_norm": 2.1194469928741455, + "learning_rate": 3.804879378492572e-06, + "loss": 2.4461648941040037, + "step": 297560 + }, + { + "epoch": 0.8100721382877365, + "grad_norm": 2.1498281955718994, + "learning_rate": 3.8037890145836176e-06, + "loss": 2.4727762222290037, + "step": 297580 + }, + { + "epoch": 0.810126582278481, + "grad_norm": 2.325529098510742, + "learning_rate": 3.8026986506746626e-06, + "loss": 2.451437759399414, + "step": 297600 + }, + { + "epoch": 0.8101810262692255, + "grad_norm": 2.1596603393554688, + "learning_rate": 3.8016082867657084e-06, + "loss": 2.4683780670166016, + "step": 297620 + }, + { + "epoch": 0.81023547025997, + "grad_norm": 2.160184383392334, + "learning_rate": 3.8005179228567542e-06, + "loss": 2.463332748413086, + "step": 297640 + }, + { + "epoch": 0.8102899142507146, + "grad_norm": 2.224862813949585, + "learning_rate": 3.799427558947799e-06, + "loss": 2.454520034790039, + "step": 297660 + }, + { + "epoch": 0.8103443582414591, + "grad_norm": 2.2068638801574707, + "learning_rate": 3.798337195038845e-06, + "loss": 2.4574050903320312, + "step": 297680 + }, + { + "epoch": 0.8103988022322036, + "grad_norm": 2.226670026779175, + "learning_rate": 3.79724683112989e-06, + "loss": 2.4779409408569335, + "step": 297700 + }, + { + "epoch": 0.8104532462229481, + "grad_norm": 2.0995607376098633, + "learning_rate": 3.7961564672209358e-06, + "loss": 2.433201026916504, + "step": 297720 + }, + { + "epoch": 0.8105076902136926, + "grad_norm": 2.1023662090301514, + "learning_rate": 3.7950661033119807e-06, + "loss": 2.4719709396362304, + "step": 297740 + }, + { + "epoch": 0.8105621342044372, + "grad_norm": 2.20090389251709, + "learning_rate": 3.7939757394030265e-06, + "loss": 2.467014122009277, + "step": 297760 + }, + { + "epoch": 0.8106165781951817, + "grad_norm": 2.344078302383423, + "learning_rate": 3.7928853754940715e-06, + "loss": 2.4548055648803713, + "step": 297780 + }, + { + "epoch": 0.8106710221859262, + "grad_norm": 2.106318950653076, + "learning_rate": 3.7917950115851173e-06, + "loss": 2.455253791809082, + "step": 297800 + }, + { + "epoch": 0.8107254661766707, + "grad_norm": 2.1507859230041504, + "learning_rate": 3.7907046476761623e-06, + "loss": 2.4733997344970704, + "step": 297820 + }, + { + "epoch": 0.8107799101674152, + "grad_norm": 2.1257190704345703, + "learning_rate": 3.789614283767208e-06, + "loss": 2.4338682174682615, + "step": 297840 + }, + { + "epoch": 0.8108343541581597, + "grad_norm": 2.1358652114868164, + "learning_rate": 3.788523919858253e-06, + "loss": 2.4544496536254883, + "step": 297860 + }, + { + "epoch": 0.8108887981489044, + "grad_norm": 2.2798640727996826, + "learning_rate": 3.7874335559492984e-06, + "loss": 2.460374641418457, + "step": 297880 + }, + { + "epoch": 0.8109432421396489, + "grad_norm": 2.3622922897338867, + "learning_rate": 3.786343192040344e-06, + "loss": 2.404472541809082, + "step": 297900 + }, + { + "epoch": 0.8109976861303934, + "grad_norm": 2.1730027198791504, + "learning_rate": 3.785252828131389e-06, + "loss": 2.467706298828125, + "step": 297920 + }, + { + "epoch": 0.8110521301211379, + "grad_norm": 2.2454237937927246, + "learning_rate": 3.7841624642224346e-06, + "loss": 2.4547731399536135, + "step": 297940 + }, + { + "epoch": 0.8111065741118824, + "grad_norm": 2.1778998374938965, + "learning_rate": 3.78307210031348e-06, + "loss": 2.470322608947754, + "step": 297960 + }, + { + "epoch": 0.811161018102627, + "grad_norm": 2.132307767868042, + "learning_rate": 3.7819817364045253e-06, + "loss": 2.4770822525024414, + "step": 297980 + }, + { + "epoch": 0.8112154620933715, + "grad_norm": 2.303610324859619, + "learning_rate": 3.7808913724955707e-06, + "loss": 2.4606792449951174, + "step": 298000 + }, + { + "epoch": 0.811269906084116, + "grad_norm": 2.144117593765259, + "learning_rate": 3.779801008586616e-06, + "loss": 2.442562484741211, + "step": 298020 + }, + { + "epoch": 0.8113243500748605, + "grad_norm": 2.1027474403381348, + "learning_rate": 3.7787106446776615e-06, + "loss": 2.4699771881103514, + "step": 298040 + }, + { + "epoch": 0.811378794065605, + "grad_norm": 2.7756054401397705, + "learning_rate": 3.777620280768707e-06, + "loss": 2.4559940338134765, + "step": 298060 + }, + { + "epoch": 0.8114332380563495, + "grad_norm": 2.2659964561462402, + "learning_rate": 3.7765299168597523e-06, + "loss": 2.465328598022461, + "step": 298080 + }, + { + "epoch": 0.8114876820470941, + "grad_norm": 2.1369566917419434, + "learning_rate": 3.7754395529507977e-06, + "loss": 2.444474792480469, + "step": 298100 + }, + { + "epoch": 0.8115421260378386, + "grad_norm": 2.2033867835998535, + "learning_rate": 3.774349189041843e-06, + "loss": 2.454736328125, + "step": 298120 + }, + { + "epoch": 0.8115965700285831, + "grad_norm": 2.2487542629241943, + "learning_rate": 3.7732588251328884e-06, + "loss": 2.4633819580078127, + "step": 298140 + }, + { + "epoch": 0.8116510140193276, + "grad_norm": 2.2700040340423584, + "learning_rate": 3.772168461223934e-06, + "loss": 2.4600915908813477, + "step": 298160 + }, + { + "epoch": 0.8117054580100721, + "grad_norm": 2.2345311641693115, + "learning_rate": 3.771078097314979e-06, + "loss": 2.461016845703125, + "step": 298180 + }, + { + "epoch": 0.8117599020008167, + "grad_norm": 2.212611436843872, + "learning_rate": 3.7699877334060246e-06, + "loss": 2.4499324798583983, + "step": 298200 + }, + { + "epoch": 0.8118143459915612, + "grad_norm": 2.2193989753723145, + "learning_rate": 3.76889736949707e-06, + "loss": 2.4664638519287108, + "step": 298220 + }, + { + "epoch": 0.8118687899823057, + "grad_norm": 2.30907940864563, + "learning_rate": 3.7678070055881154e-06, + "loss": 2.472958564758301, + "step": 298240 + }, + { + "epoch": 0.8119232339730502, + "grad_norm": 2.141667127609253, + "learning_rate": 3.7667166416791607e-06, + "loss": 2.4410560607910154, + "step": 298260 + }, + { + "epoch": 0.8119776779637947, + "grad_norm": 2.4059324264526367, + "learning_rate": 3.765626277770206e-06, + "loss": 2.5109106063842774, + "step": 298280 + }, + { + "epoch": 0.8120321219545392, + "grad_norm": 2.1276609897613525, + "learning_rate": 3.7645359138612515e-06, + "loss": 2.453466033935547, + "step": 298300 + }, + { + "epoch": 0.8120865659452838, + "grad_norm": 2.4381103515625, + "learning_rate": 3.763445549952297e-06, + "loss": 2.443012809753418, + "step": 298320 + }, + { + "epoch": 0.8121410099360283, + "grad_norm": 2.1815366744995117, + "learning_rate": 3.7623551860433423e-06, + "loss": 2.5026546478271485, + "step": 298340 + }, + { + "epoch": 0.8121954539267728, + "grad_norm": 2.059875249862671, + "learning_rate": 3.7612648221343877e-06, + "loss": 2.4476261138916016, + "step": 298360 + }, + { + "epoch": 0.8122498979175173, + "grad_norm": 2.180669069290161, + "learning_rate": 3.760174458225433e-06, + "loss": 2.4641254425048826, + "step": 298380 + }, + { + "epoch": 0.8123043419082618, + "grad_norm": 2.1335744857788086, + "learning_rate": 3.7590840943164784e-06, + "loss": 2.4775920867919923, + "step": 298400 + }, + { + "epoch": 0.8123587858990065, + "grad_norm": 2.16050124168396, + "learning_rate": 3.757993730407524e-06, + "loss": 2.4806894302368163, + "step": 298420 + }, + { + "epoch": 0.812413229889751, + "grad_norm": 2.222740411758423, + "learning_rate": 3.7569033664985692e-06, + "loss": 2.483866882324219, + "step": 298440 + }, + { + "epoch": 0.8124676738804955, + "grad_norm": 2.222707509994507, + "learning_rate": 3.7558130025896146e-06, + "loss": 2.437549591064453, + "step": 298460 + }, + { + "epoch": 0.81252211787124, + "grad_norm": 2.2256946563720703, + "learning_rate": 3.75472263868066e-06, + "loss": 2.4658681869506838, + "step": 298480 + }, + { + "epoch": 0.8125765618619845, + "grad_norm": 2.256791353225708, + "learning_rate": 3.7536322747717054e-06, + "loss": 2.473845863342285, + "step": 298500 + }, + { + "epoch": 0.812631005852729, + "grad_norm": 2.1736643314361572, + "learning_rate": 3.7525419108627508e-06, + "loss": 2.4609813690185547, + "step": 298520 + }, + { + "epoch": 0.8126854498434736, + "grad_norm": 2.3442392349243164, + "learning_rate": 3.751451546953796e-06, + "loss": 2.4539278030395506, + "step": 298540 + }, + { + "epoch": 0.8127398938342181, + "grad_norm": 2.2768027782440186, + "learning_rate": 3.7503611830448415e-06, + "loss": 2.483498382568359, + "step": 298560 + }, + { + "epoch": 0.8127943378249626, + "grad_norm": 2.116518020629883, + "learning_rate": 3.749270819135887e-06, + "loss": 2.4744890213012694, + "step": 298580 + }, + { + "epoch": 0.8128487818157071, + "grad_norm": 2.1599388122558594, + "learning_rate": 3.7481804552269323e-06, + "loss": 2.473450469970703, + "step": 298600 + }, + { + "epoch": 0.8129032258064516, + "grad_norm": 2.0987560749053955, + "learning_rate": 3.7470900913179777e-06, + "loss": 2.4669708251953124, + "step": 298620 + }, + { + "epoch": 0.8129576697971961, + "grad_norm": 2.0947070121765137, + "learning_rate": 3.745999727409023e-06, + "loss": 2.4720090866088866, + "step": 298640 + }, + { + "epoch": 0.8130121137879407, + "grad_norm": 2.134885787963867, + "learning_rate": 3.7449093635000685e-06, + "loss": 2.452412796020508, + "step": 298660 + }, + { + "epoch": 0.8130665577786852, + "grad_norm": 2.187945604324341, + "learning_rate": 3.743818999591114e-06, + "loss": 2.4695526123046876, + "step": 298680 + }, + { + "epoch": 0.8131210017694297, + "grad_norm": 2.201375961303711, + "learning_rate": 3.7427286356821592e-06, + "loss": 2.465734100341797, + "step": 298700 + }, + { + "epoch": 0.8131754457601742, + "grad_norm": 2.228065252304077, + "learning_rate": 3.7416382717732046e-06, + "loss": 2.4624786376953125, + "step": 298720 + }, + { + "epoch": 0.8132298897509187, + "grad_norm": 2.1949334144592285, + "learning_rate": 3.74054790786425e-06, + "loss": 2.490309715270996, + "step": 298740 + }, + { + "epoch": 0.8132843337416633, + "grad_norm": 2.176713705062866, + "learning_rate": 3.7394575439552954e-06, + "loss": 2.463164520263672, + "step": 298760 + }, + { + "epoch": 0.8133387777324078, + "grad_norm": 2.275866985321045, + "learning_rate": 3.7383671800463408e-06, + "loss": 2.4533130645751955, + "step": 298780 + }, + { + "epoch": 0.8133932217231523, + "grad_norm": 2.323287010192871, + "learning_rate": 3.737276816137386e-06, + "loss": 2.4466554641723635, + "step": 298800 + }, + { + "epoch": 0.8134476657138968, + "grad_norm": 2.278930425643921, + "learning_rate": 3.7361864522284315e-06, + "loss": 2.478664016723633, + "step": 298820 + }, + { + "epoch": 0.8135021097046413, + "grad_norm": 2.122955322265625, + "learning_rate": 3.735096088319477e-06, + "loss": 2.4682638168334963, + "step": 298840 + }, + { + "epoch": 0.8135565536953858, + "grad_norm": 2.1678519248962402, + "learning_rate": 3.7340057244105223e-06, + "loss": 2.453213691711426, + "step": 298860 + }, + { + "epoch": 0.8136109976861304, + "grad_norm": 2.145918607711792, + "learning_rate": 3.7329153605015677e-06, + "loss": 2.477849769592285, + "step": 298880 + }, + { + "epoch": 0.8136654416768749, + "grad_norm": 2.1139469146728516, + "learning_rate": 3.731824996592613e-06, + "loss": 2.4747005462646485, + "step": 298900 + }, + { + "epoch": 0.8137198856676194, + "grad_norm": 2.1349833011627197, + "learning_rate": 3.7307346326836585e-06, + "loss": 2.458802032470703, + "step": 298920 + }, + { + "epoch": 0.8137743296583639, + "grad_norm": 2.1198112964630127, + "learning_rate": 3.7296442687747034e-06, + "loss": 2.4676538467407227, + "step": 298940 + }, + { + "epoch": 0.8138287736491084, + "grad_norm": 2.272472381591797, + "learning_rate": 3.7285539048657492e-06, + "loss": 2.486889457702637, + "step": 298960 + }, + { + "epoch": 0.813883217639853, + "grad_norm": 2.138103723526001, + "learning_rate": 3.727463540956794e-06, + "loss": 2.4688371658325194, + "step": 298980 + }, + { + "epoch": 0.8139376616305976, + "grad_norm": 2.6504461765289307, + "learning_rate": 3.72637317704784e-06, + "loss": 2.4716302871704103, + "step": 299000 + }, + { + "epoch": 0.813992105621342, + "grad_norm": 2.0980947017669678, + "learning_rate": 3.725282813138885e-06, + "loss": 2.458892250061035, + "step": 299020 + }, + { + "epoch": 0.8140465496120866, + "grad_norm": 2.109550952911377, + "learning_rate": 3.7241924492299308e-06, + "loss": 2.4781112670898438, + "step": 299040 + }, + { + "epoch": 0.8141009936028311, + "grad_norm": 2.489163398742676, + "learning_rate": 3.7231020853209757e-06, + "loss": 2.4887531280517576, + "step": 299060 + }, + { + "epoch": 0.8141554375935756, + "grad_norm": 2.4280388355255127, + "learning_rate": 3.7220117214120216e-06, + "loss": 2.4745216369628906, + "step": 299080 + }, + { + "epoch": 0.8142098815843202, + "grad_norm": 2.390979290008545, + "learning_rate": 3.7209213575030665e-06, + "loss": 2.4326427459716795, + "step": 299100 + }, + { + "epoch": 0.8142643255750647, + "grad_norm": 2.229527473449707, + "learning_rate": 3.7198309935941123e-06, + "loss": 2.4565383911132814, + "step": 299120 + }, + { + "epoch": 0.8143187695658092, + "grad_norm": 2.100252866744995, + "learning_rate": 3.7187406296851573e-06, + "loss": 2.436206817626953, + "step": 299140 + }, + { + "epoch": 0.8143732135565537, + "grad_norm": 2.1371002197265625, + "learning_rate": 3.717650265776203e-06, + "loss": 2.4776357650756835, + "step": 299160 + }, + { + "epoch": 0.8144276575472982, + "grad_norm": 2.1501667499542236, + "learning_rate": 3.716559901867248e-06, + "loss": 2.4643491744995116, + "step": 299180 + }, + { + "epoch": 0.8144821015380428, + "grad_norm": 2.194366693496704, + "learning_rate": 3.715469537958294e-06, + "loss": 2.4874851226806642, + "step": 299200 + }, + { + "epoch": 0.8145365455287873, + "grad_norm": 2.103853225708008, + "learning_rate": 3.714379174049339e-06, + "loss": 2.4634571075439453, + "step": 299220 + }, + { + "epoch": 0.8145909895195318, + "grad_norm": 2.4109482765197754, + "learning_rate": 3.7132888101403846e-06, + "loss": 2.445675277709961, + "step": 299240 + }, + { + "epoch": 0.8146454335102763, + "grad_norm": 2.1088929176330566, + "learning_rate": 3.7121984462314296e-06, + "loss": 2.4563358306884764, + "step": 299260 + }, + { + "epoch": 0.8146998775010208, + "grad_norm": 2.1413209438323975, + "learning_rate": 3.7111080823224754e-06, + "loss": 2.4884855270385744, + "step": 299280 + }, + { + "epoch": 0.8147543214917653, + "grad_norm": 2.1255102157592773, + "learning_rate": 3.7100177184135204e-06, + "loss": 2.434785270690918, + "step": 299300 + }, + { + "epoch": 0.8148087654825099, + "grad_norm": 2.1795217990875244, + "learning_rate": 3.708927354504566e-06, + "loss": 2.4579692840576173, + "step": 299320 + }, + { + "epoch": 0.8148632094732544, + "grad_norm": 2.1460742950439453, + "learning_rate": 3.707836990595612e-06, + "loss": 2.482725143432617, + "step": 299340 + }, + { + "epoch": 0.8149176534639989, + "grad_norm": 2.1482269763946533, + "learning_rate": 3.706746626686657e-06, + "loss": 2.4589942932128905, + "step": 299360 + }, + { + "epoch": 0.8149720974547434, + "grad_norm": 2.2359604835510254, + "learning_rate": 3.7056562627777028e-06, + "loss": 2.4457910537719725, + "step": 299380 + }, + { + "epoch": 0.8150265414454879, + "grad_norm": 2.183096170425415, + "learning_rate": 3.7045658988687477e-06, + "loss": 2.4631183624267576, + "step": 299400 + }, + { + "epoch": 0.8150809854362325, + "grad_norm": 2.201312780380249, + "learning_rate": 3.7034755349597935e-06, + "loss": 2.46835994720459, + "step": 299420 + }, + { + "epoch": 0.815135429426977, + "grad_norm": 2.176942825317383, + "learning_rate": 3.7023851710508385e-06, + "loss": 2.4535104751586916, + "step": 299440 + }, + { + "epoch": 0.8151898734177215, + "grad_norm": 2.849346399307251, + "learning_rate": 3.7012948071418843e-06, + "loss": 2.4600706100463867, + "step": 299460 + }, + { + "epoch": 0.815244317408466, + "grad_norm": 2.3461968898773193, + "learning_rate": 3.7002044432329293e-06, + "loss": 2.4436660766601563, + "step": 299480 + }, + { + "epoch": 0.8152987613992105, + "grad_norm": 2.157839059829712, + "learning_rate": 3.699114079323975e-06, + "loss": 2.468160629272461, + "step": 299500 + }, + { + "epoch": 0.815353205389955, + "grad_norm": 2.2654154300689697, + "learning_rate": 3.69802371541502e-06, + "loss": 2.4714096069335936, + "step": 299520 + }, + { + "epoch": 0.8154076493806997, + "grad_norm": 2.1017491817474365, + "learning_rate": 3.696933351506066e-06, + "loss": 2.476171875, + "step": 299540 + }, + { + "epoch": 0.8154620933714442, + "grad_norm": 2.190779209136963, + "learning_rate": 3.695842987597111e-06, + "loss": 2.4543071746826173, + "step": 299560 + }, + { + "epoch": 0.8155165373621887, + "grad_norm": 2.214170455932617, + "learning_rate": 3.6947526236881566e-06, + "loss": 2.4661380767822267, + "step": 299580 + }, + { + "epoch": 0.8155709813529332, + "grad_norm": 2.124026298522949, + "learning_rate": 3.6936622597792016e-06, + "loss": 2.4591747283935548, + "step": 299600 + }, + { + "epoch": 0.8156254253436777, + "grad_norm": 2.1244888305664062, + "learning_rate": 3.6925718958702474e-06, + "loss": 2.457121467590332, + "step": 299620 + }, + { + "epoch": 0.8156798693344222, + "grad_norm": 2.2698333263397217, + "learning_rate": 3.6914815319612924e-06, + "loss": 2.4538150787353517, + "step": 299640 + }, + { + "epoch": 0.8157343133251668, + "grad_norm": 2.1628236770629883, + "learning_rate": 3.690391168052338e-06, + "loss": 2.4579050064086916, + "step": 299660 + }, + { + "epoch": 0.8157887573159113, + "grad_norm": 2.161128282546997, + "learning_rate": 3.689300804143383e-06, + "loss": 2.458216094970703, + "step": 299680 + }, + { + "epoch": 0.8158432013066558, + "grad_norm": 3.4942879676818848, + "learning_rate": 3.688210440234429e-06, + "loss": 2.4835771560668944, + "step": 299700 + }, + { + "epoch": 0.8158976452974003, + "grad_norm": 2.356295347213745, + "learning_rate": 3.687120076325474e-06, + "loss": 2.470211410522461, + "step": 299720 + }, + { + "epoch": 0.8159520892881448, + "grad_norm": 2.314574956893921, + "learning_rate": 3.6860297124165197e-06, + "loss": 2.463446617126465, + "step": 299740 + }, + { + "epoch": 0.8160065332788894, + "grad_norm": 2.1365389823913574, + "learning_rate": 3.6849393485075647e-06, + "loss": 2.462197113037109, + "step": 299760 + }, + { + "epoch": 0.8160609772696339, + "grad_norm": 2.2204513549804688, + "learning_rate": 3.6838489845986105e-06, + "loss": 2.4682998657226562, + "step": 299780 + }, + { + "epoch": 0.8161154212603784, + "grad_norm": 2.686986207962036, + "learning_rate": 3.6827586206896554e-06, + "loss": 2.4325435638427733, + "step": 299800 + }, + { + "epoch": 0.8161698652511229, + "grad_norm": 2.1529455184936523, + "learning_rate": 3.6816682567807013e-06, + "loss": 2.4921655654907227, + "step": 299820 + }, + { + "epoch": 0.8162243092418674, + "grad_norm": 2.132070302963257, + "learning_rate": 3.6805778928717462e-06, + "loss": 2.4670316696166994, + "step": 299840 + }, + { + "epoch": 0.8162787532326119, + "grad_norm": 2.1405553817749023, + "learning_rate": 3.679487528962792e-06, + "loss": 2.478412055969238, + "step": 299860 + }, + { + "epoch": 0.8163331972233565, + "grad_norm": 2.1254360675811768, + "learning_rate": 3.678397165053837e-06, + "loss": 2.472757911682129, + "step": 299880 + }, + { + "epoch": 0.816387641214101, + "grad_norm": 2.1437201499938965, + "learning_rate": 3.677306801144883e-06, + "loss": 2.4874700546264648, + "step": 299900 + }, + { + "epoch": 0.8164420852048455, + "grad_norm": 2.2395505905151367, + "learning_rate": 3.6762164372359278e-06, + "loss": 2.461580276489258, + "step": 299920 + }, + { + "epoch": 0.81649652919559, + "grad_norm": 2.323289394378662, + "learning_rate": 3.675126073326973e-06, + "loss": 2.4941925048828124, + "step": 299940 + }, + { + "epoch": 0.8165509731863345, + "grad_norm": 2.1083450317382812, + "learning_rate": 3.6740357094180185e-06, + "loss": 2.46127872467041, + "step": 299960 + }, + { + "epoch": 0.8166054171770791, + "grad_norm": 2.1525232791900635, + "learning_rate": 3.672945345509064e-06, + "loss": 2.44946174621582, + "step": 299980 + }, + { + "epoch": 0.8166598611678236, + "grad_norm": 2.3575551509857178, + "learning_rate": 3.6718549816001093e-06, + "loss": 2.467741012573242, + "step": 300000 + }, + { + "epoch": 0.8167143051585681, + "grad_norm": 2.186439037322998, + "learning_rate": 3.6707646176911547e-06, + "loss": 2.470595932006836, + "step": 300020 + }, + { + "epoch": 0.8167687491493126, + "grad_norm": 2.351013422012329, + "learning_rate": 3.6696742537822e-06, + "loss": 2.4494606018066407, + "step": 300040 + }, + { + "epoch": 0.8168231931400571, + "grad_norm": 2.284839391708374, + "learning_rate": 3.6685838898732455e-06, + "loss": 2.4449329376220703, + "step": 300060 + }, + { + "epoch": 0.8168776371308016, + "grad_norm": 2.1047613620758057, + "learning_rate": 3.667493525964291e-06, + "loss": 2.4626562118530275, + "step": 300080 + }, + { + "epoch": 0.8169320811215462, + "grad_norm": 2.4999895095825195, + "learning_rate": 3.6664031620553362e-06, + "loss": 2.4535491943359373, + "step": 300100 + }, + { + "epoch": 0.8169865251122908, + "grad_norm": 2.2643401622772217, + "learning_rate": 3.6653127981463816e-06, + "loss": 2.444451332092285, + "step": 300120 + }, + { + "epoch": 0.8170409691030353, + "grad_norm": 2.2017855644226074, + "learning_rate": 3.664222434237427e-06, + "loss": 2.440277671813965, + "step": 300140 + }, + { + "epoch": 0.8170954130937798, + "grad_norm": 2.098442316055298, + "learning_rate": 3.6631320703284724e-06, + "loss": 2.4861143112182615, + "step": 300160 + }, + { + "epoch": 0.8171498570845243, + "grad_norm": 2.1925971508026123, + "learning_rate": 3.6620417064195178e-06, + "loss": 2.4656932830810545, + "step": 300180 + }, + { + "epoch": 0.8172043010752689, + "grad_norm": 2.2584314346313477, + "learning_rate": 3.660951342510563e-06, + "loss": 2.484150695800781, + "step": 300200 + }, + { + "epoch": 0.8172587450660134, + "grad_norm": 2.305443286895752, + "learning_rate": 3.6598609786016085e-06, + "loss": 2.4619672775268553, + "step": 300220 + }, + { + "epoch": 0.8173131890567579, + "grad_norm": 2.3271658420562744, + "learning_rate": 3.658770614692654e-06, + "loss": 2.475362205505371, + "step": 300240 + }, + { + "epoch": 0.8173676330475024, + "grad_norm": 2.202609062194824, + "learning_rate": 3.6576802507836993e-06, + "loss": 2.4547040939331053, + "step": 300260 + }, + { + "epoch": 0.8174220770382469, + "grad_norm": 2.246244192123413, + "learning_rate": 3.6565898868747447e-06, + "loss": 2.4803787231445313, + "step": 300280 + }, + { + "epoch": 0.8174765210289914, + "grad_norm": 2.205280303955078, + "learning_rate": 3.65549952296579e-06, + "loss": 2.443652534484863, + "step": 300300 + }, + { + "epoch": 0.817530965019736, + "grad_norm": 2.4960615634918213, + "learning_rate": 3.6544091590568355e-06, + "loss": 2.4829010009765624, + "step": 300320 + }, + { + "epoch": 0.8175854090104805, + "grad_norm": 2.1484577655792236, + "learning_rate": 3.653318795147881e-06, + "loss": 2.4660640716552735, + "step": 300340 + }, + { + "epoch": 0.817639853001225, + "grad_norm": 2.159511089324951, + "learning_rate": 3.6522284312389262e-06, + "loss": 2.458489990234375, + "step": 300360 + }, + { + "epoch": 0.8176942969919695, + "grad_norm": 2.273815155029297, + "learning_rate": 3.6511380673299716e-06, + "loss": 2.4384027481079102, + "step": 300380 + }, + { + "epoch": 0.817748740982714, + "grad_norm": 2.120718479156494, + "learning_rate": 3.650047703421017e-06, + "loss": 2.458690071105957, + "step": 300400 + }, + { + "epoch": 0.8178031849734586, + "grad_norm": 2.2683167457580566, + "learning_rate": 3.6489573395120624e-06, + "loss": 2.4338212966918946, + "step": 300420 + }, + { + "epoch": 0.8178576289642031, + "grad_norm": 2.0868120193481445, + "learning_rate": 3.6478669756031078e-06, + "loss": 2.443487358093262, + "step": 300440 + }, + { + "epoch": 0.8179120729549476, + "grad_norm": 2.105907917022705, + "learning_rate": 3.646776611694153e-06, + "loss": 2.435785102844238, + "step": 300460 + }, + { + "epoch": 0.8179665169456921, + "grad_norm": 2.2885236740112305, + "learning_rate": 3.6456862477851986e-06, + "loss": 2.4430437088012695, + "step": 300480 + }, + { + "epoch": 0.8180209609364366, + "grad_norm": 2.059382677078247, + "learning_rate": 3.644595883876244e-06, + "loss": 2.432861328125, + "step": 300500 + }, + { + "epoch": 0.8180754049271811, + "grad_norm": 2.23630428314209, + "learning_rate": 3.6435055199672893e-06, + "loss": 2.416617012023926, + "step": 300520 + }, + { + "epoch": 0.8181298489179257, + "grad_norm": 2.076650619506836, + "learning_rate": 3.6424151560583347e-06, + "loss": 2.44552001953125, + "step": 300540 + }, + { + "epoch": 0.8181842929086702, + "grad_norm": 2.262747049331665, + "learning_rate": 3.64132479214938e-06, + "loss": 2.4773759841918945, + "step": 300560 + }, + { + "epoch": 0.8182387368994147, + "grad_norm": 2.4681968688964844, + "learning_rate": 3.6402344282404255e-06, + "loss": 2.4489425659179687, + "step": 300580 + }, + { + "epoch": 0.8182931808901592, + "grad_norm": 2.125530958175659, + "learning_rate": 3.639144064331471e-06, + "loss": 2.4431028366088867, + "step": 300600 + }, + { + "epoch": 0.8183476248809037, + "grad_norm": 2.296302556991577, + "learning_rate": 3.6380537004225163e-06, + "loss": 2.4234848022460938, + "step": 300620 + }, + { + "epoch": 0.8184020688716483, + "grad_norm": 2.174379587173462, + "learning_rate": 3.6369633365135616e-06, + "loss": 2.515105438232422, + "step": 300640 + }, + { + "epoch": 0.8184565128623928, + "grad_norm": 2.1770784854888916, + "learning_rate": 3.635872972604607e-06, + "loss": 2.4380674362182617, + "step": 300660 + }, + { + "epoch": 0.8185109568531374, + "grad_norm": 2.0992431640625, + "learning_rate": 3.6347826086956524e-06, + "loss": 2.4423303604125977, + "step": 300680 + }, + { + "epoch": 0.8185654008438819, + "grad_norm": 2.3225157260894775, + "learning_rate": 3.633692244786698e-06, + "loss": 2.466267776489258, + "step": 300700 + }, + { + "epoch": 0.8186198448346264, + "grad_norm": 2.202219247817993, + "learning_rate": 3.632601880877743e-06, + "loss": 2.457095146179199, + "step": 300720 + }, + { + "epoch": 0.8186742888253709, + "grad_norm": 2.1124320030212402, + "learning_rate": 3.6315115169687886e-06, + "loss": 2.491261291503906, + "step": 300740 + }, + { + "epoch": 0.8187287328161155, + "grad_norm": 2.2372684478759766, + "learning_rate": 3.630421153059834e-06, + "loss": 2.4625127792358397, + "step": 300760 + }, + { + "epoch": 0.81878317680686, + "grad_norm": 2.077552080154419, + "learning_rate": 3.6293307891508793e-06, + "loss": 2.4862367630004885, + "step": 300780 + }, + { + "epoch": 0.8188376207976045, + "grad_norm": 2.186021089553833, + "learning_rate": 3.6282404252419247e-06, + "loss": 2.4662197113037108, + "step": 300800 + }, + { + "epoch": 0.818892064788349, + "grad_norm": 2.157745122909546, + "learning_rate": 3.62715006133297e-06, + "loss": 2.444980239868164, + "step": 300820 + }, + { + "epoch": 0.8189465087790935, + "grad_norm": 2.18925142288208, + "learning_rate": 3.6260596974240155e-06, + "loss": 2.4384637832641602, + "step": 300840 + }, + { + "epoch": 0.819000952769838, + "grad_norm": 2.3268773555755615, + "learning_rate": 3.624969333515061e-06, + "loss": 2.463589668273926, + "step": 300860 + }, + { + "epoch": 0.8190553967605826, + "grad_norm": 2.143731117248535, + "learning_rate": 3.6238789696061063e-06, + "loss": 2.4633277893066405, + "step": 300880 + }, + { + "epoch": 0.8191098407513271, + "grad_norm": 2.2504870891571045, + "learning_rate": 3.6227886056971517e-06, + "loss": 2.492100143432617, + "step": 300900 + }, + { + "epoch": 0.8191642847420716, + "grad_norm": 2.2307281494140625, + "learning_rate": 3.621698241788197e-06, + "loss": 2.4814521789550783, + "step": 300920 + }, + { + "epoch": 0.8192187287328161, + "grad_norm": 3.3059091567993164, + "learning_rate": 3.6206078778792424e-06, + "loss": 2.4769941329956056, + "step": 300940 + }, + { + "epoch": 0.8192731727235606, + "grad_norm": 2.2471022605895996, + "learning_rate": 3.619517513970288e-06, + "loss": 2.448422431945801, + "step": 300960 + }, + { + "epoch": 0.8193276167143052, + "grad_norm": 2.1575896739959717, + "learning_rate": 3.618427150061333e-06, + "loss": 2.461749267578125, + "step": 300980 + }, + { + "epoch": 0.8193820607050497, + "grad_norm": 2.2807440757751465, + "learning_rate": 3.6173367861523786e-06, + "loss": 2.432668113708496, + "step": 301000 + }, + { + "epoch": 0.8194365046957942, + "grad_norm": 2.2984378337860107, + "learning_rate": 3.616246422243424e-06, + "loss": 2.4826868057250975, + "step": 301020 + }, + { + "epoch": 0.8194909486865387, + "grad_norm": 2.1498947143554688, + "learning_rate": 3.6151560583344698e-06, + "loss": 2.475354766845703, + "step": 301040 + }, + { + "epoch": 0.8195453926772832, + "grad_norm": 2.2785165309906006, + "learning_rate": 3.6140656944255147e-06, + "loss": 2.485830307006836, + "step": 301060 + }, + { + "epoch": 0.8195998366680277, + "grad_norm": 2.099302053451538, + "learning_rate": 3.6129753305165606e-06, + "loss": 2.4566484451293946, + "step": 301080 + }, + { + "epoch": 0.8196542806587723, + "grad_norm": 2.182417392730713, + "learning_rate": 3.6118849666076055e-06, + "loss": 2.425431823730469, + "step": 301100 + }, + { + "epoch": 0.8197087246495168, + "grad_norm": 2.3451201915740967, + "learning_rate": 3.6107946026986513e-06, + "loss": 2.478995132446289, + "step": 301120 + }, + { + "epoch": 0.8197631686402613, + "grad_norm": 2.1424639225006104, + "learning_rate": 3.6097042387896963e-06, + "loss": 2.4703546524047852, + "step": 301140 + }, + { + "epoch": 0.8198176126310058, + "grad_norm": 2.1617043018341064, + "learning_rate": 3.608613874880742e-06, + "loss": 2.45684871673584, + "step": 301160 + }, + { + "epoch": 0.8198720566217503, + "grad_norm": 2.227137804031372, + "learning_rate": 3.607523510971787e-06, + "loss": 2.4513803482055665, + "step": 301180 + }, + { + "epoch": 0.819926500612495, + "grad_norm": 2.3452508449554443, + "learning_rate": 3.606433147062833e-06, + "loss": 2.4336755752563475, + "step": 301200 + }, + { + "epoch": 0.8199809446032394, + "grad_norm": 2.1630964279174805, + "learning_rate": 3.605342783153878e-06, + "loss": 2.4403617858886717, + "step": 301220 + }, + { + "epoch": 0.820035388593984, + "grad_norm": 4.05525016784668, + "learning_rate": 3.6042524192449236e-06, + "loss": 2.4739233016967774, + "step": 301240 + }, + { + "epoch": 0.8200898325847285, + "grad_norm": 2.18023419380188, + "learning_rate": 3.6031620553359686e-06, + "loss": 2.463140869140625, + "step": 301260 + }, + { + "epoch": 0.820144276575473, + "grad_norm": 2.1238059997558594, + "learning_rate": 3.6020716914270144e-06, + "loss": 2.453287887573242, + "step": 301280 + }, + { + "epoch": 0.8201987205662175, + "grad_norm": 2.253984212875366, + "learning_rate": 3.6009813275180594e-06, + "loss": 2.4723268508911134, + "step": 301300 + }, + { + "epoch": 0.8202531645569621, + "grad_norm": 2.28885555267334, + "learning_rate": 3.599890963609105e-06, + "loss": 2.485281562805176, + "step": 301320 + }, + { + "epoch": 0.8203076085477066, + "grad_norm": 2.1878128051757812, + "learning_rate": 3.59880059970015e-06, + "loss": 2.4334178924560548, + "step": 301340 + }, + { + "epoch": 0.8203620525384511, + "grad_norm": 2.187490224838257, + "learning_rate": 3.597710235791196e-06, + "loss": 2.4772472381591797, + "step": 301360 + }, + { + "epoch": 0.8204164965291956, + "grad_norm": 2.401315450668335, + "learning_rate": 3.596619871882241e-06, + "loss": 2.4474382400512695, + "step": 301380 + }, + { + "epoch": 0.8204709405199401, + "grad_norm": 3.346252679824829, + "learning_rate": 3.5955295079732867e-06, + "loss": 2.4622289657592775, + "step": 301400 + }, + { + "epoch": 0.8205253845106847, + "grad_norm": 2.320021152496338, + "learning_rate": 3.5944391440643317e-06, + "loss": 2.4835371017456054, + "step": 301420 + }, + { + "epoch": 0.8205798285014292, + "grad_norm": 2.110395669937134, + "learning_rate": 3.5933487801553775e-06, + "loss": 2.4697784423828124, + "step": 301440 + }, + { + "epoch": 0.8206342724921737, + "grad_norm": 2.16194486618042, + "learning_rate": 3.5922584162464225e-06, + "loss": 2.4561281204223633, + "step": 301460 + }, + { + "epoch": 0.8206887164829182, + "grad_norm": 2.215428113937378, + "learning_rate": 3.5911680523374683e-06, + "loss": 2.464041900634766, + "step": 301480 + }, + { + "epoch": 0.8207431604736627, + "grad_norm": 2.458461284637451, + "learning_rate": 3.5900776884285132e-06, + "loss": 2.461879348754883, + "step": 301500 + }, + { + "epoch": 0.8207976044644072, + "grad_norm": 2.246894359588623, + "learning_rate": 3.588987324519559e-06, + "loss": 2.461435317993164, + "step": 301520 + }, + { + "epoch": 0.8208520484551518, + "grad_norm": 2.1893210411071777, + "learning_rate": 3.587896960610604e-06, + "loss": 2.425850486755371, + "step": 301540 + }, + { + "epoch": 0.8209064924458963, + "grad_norm": 2.191983461380005, + "learning_rate": 3.58680659670165e-06, + "loss": 2.4578598022460936, + "step": 301560 + }, + { + "epoch": 0.8209609364366408, + "grad_norm": 2.3467326164245605, + "learning_rate": 3.5857162327926948e-06, + "loss": 2.4638734817504884, + "step": 301580 + }, + { + "epoch": 0.8210153804273853, + "grad_norm": 2.0654962062835693, + "learning_rate": 3.5846258688837406e-06, + "loss": 2.462871551513672, + "step": 301600 + }, + { + "epoch": 0.8210698244181298, + "grad_norm": 2.090588331222534, + "learning_rate": 3.5835355049747855e-06, + "loss": 2.438115882873535, + "step": 301620 + }, + { + "epoch": 0.8211242684088744, + "grad_norm": 2.116077423095703, + "learning_rate": 3.5824451410658314e-06, + "loss": 2.421620178222656, + "step": 301640 + }, + { + "epoch": 0.8211787123996189, + "grad_norm": 2.202277898788452, + "learning_rate": 3.5813547771568763e-06, + "loss": 2.4679065704345704, + "step": 301660 + }, + { + "epoch": 0.8212331563903634, + "grad_norm": 2.1463005542755127, + "learning_rate": 3.580264413247922e-06, + "loss": 2.461629295349121, + "step": 301680 + }, + { + "epoch": 0.8212876003811079, + "grad_norm": 2.1728341579437256, + "learning_rate": 3.579174049338967e-06, + "loss": 2.485525894165039, + "step": 301700 + }, + { + "epoch": 0.8213420443718524, + "grad_norm": 2.1367878913879395, + "learning_rate": 3.578083685430013e-06, + "loss": 2.450217628479004, + "step": 301720 + }, + { + "epoch": 0.8213964883625969, + "grad_norm": 2.2226760387420654, + "learning_rate": 3.576993321521058e-06, + "loss": 2.460504913330078, + "step": 301740 + }, + { + "epoch": 0.8214509323533415, + "grad_norm": 2.3879356384277344, + "learning_rate": 3.5759029576121037e-06, + "loss": 2.4654273986816406, + "step": 301760 + }, + { + "epoch": 0.821505376344086, + "grad_norm": 2.186469078063965, + "learning_rate": 3.5748125937031486e-06, + "loss": 2.490578269958496, + "step": 301780 + }, + { + "epoch": 0.8215598203348305, + "grad_norm": 2.225651979446411, + "learning_rate": 3.5737222297941944e-06, + "loss": 2.460983467102051, + "step": 301800 + }, + { + "epoch": 0.821614264325575, + "grad_norm": 2.138284683227539, + "learning_rate": 3.5726318658852394e-06, + "loss": 2.457489585876465, + "step": 301820 + }, + { + "epoch": 0.8216687083163196, + "grad_norm": 2.273923397064209, + "learning_rate": 3.571541501976285e-06, + "loss": 2.4648014068603517, + "step": 301840 + }, + { + "epoch": 0.821723152307064, + "grad_norm": 2.5834641456604004, + "learning_rate": 3.57045113806733e-06, + "loss": 2.4832059860229494, + "step": 301860 + }, + { + "epoch": 0.8217775962978087, + "grad_norm": 2.1525306701660156, + "learning_rate": 3.569360774158376e-06, + "loss": 2.472956085205078, + "step": 301880 + }, + { + "epoch": 0.8218320402885532, + "grad_norm": 2.196913957595825, + "learning_rate": 3.568270410249421e-06, + "loss": 2.4529394149780273, + "step": 301900 + }, + { + "epoch": 0.8218864842792977, + "grad_norm": 2.170774221420288, + "learning_rate": 3.5671800463404668e-06, + "loss": 2.4726398468017576, + "step": 301920 + }, + { + "epoch": 0.8219409282700422, + "grad_norm": 2.194836139678955, + "learning_rate": 3.5660896824315117e-06, + "loss": 2.4745391845703124, + "step": 301940 + }, + { + "epoch": 0.8219953722607867, + "grad_norm": 2.1247637271881104, + "learning_rate": 3.5649993185225575e-06, + "loss": 2.459978485107422, + "step": 301960 + }, + { + "epoch": 0.8220498162515313, + "grad_norm": 2.1346747875213623, + "learning_rate": 3.5639089546136025e-06, + "loss": 2.4647247314453127, + "step": 301980 + }, + { + "epoch": 0.8221042602422758, + "grad_norm": 2.2257437705993652, + "learning_rate": 3.5628185907046483e-06, + "loss": 2.4730960845947267, + "step": 302000 + }, + { + "epoch": 0.8221587042330203, + "grad_norm": 2.3170316219329834, + "learning_rate": 3.5617282267956933e-06, + "loss": 2.4572534561157227, + "step": 302020 + }, + { + "epoch": 0.8222131482237648, + "grad_norm": 2.2570478916168213, + "learning_rate": 3.5606378628867386e-06, + "loss": 2.444306182861328, + "step": 302040 + }, + { + "epoch": 0.8222675922145093, + "grad_norm": 2.3662314414978027, + "learning_rate": 3.559547498977784e-06, + "loss": 2.480535888671875, + "step": 302060 + }, + { + "epoch": 0.8223220362052538, + "grad_norm": 2.397458791732788, + "learning_rate": 3.5584571350688294e-06, + "loss": 2.4747041702270507, + "step": 302080 + }, + { + "epoch": 0.8223764801959984, + "grad_norm": 2.096287727355957, + "learning_rate": 3.557366771159875e-06, + "loss": 2.46569766998291, + "step": 302100 + }, + { + "epoch": 0.8224309241867429, + "grad_norm": 2.188339948654175, + "learning_rate": 3.55627640725092e-06, + "loss": 2.4662540435791014, + "step": 302120 + }, + { + "epoch": 0.8224853681774874, + "grad_norm": 2.218679904937744, + "learning_rate": 3.5551860433419656e-06, + "loss": 2.4802831649780273, + "step": 302140 + }, + { + "epoch": 0.8225398121682319, + "grad_norm": 2.2083375453948975, + "learning_rate": 3.554095679433011e-06, + "loss": 2.4571609497070312, + "step": 302160 + }, + { + "epoch": 0.8225942561589764, + "grad_norm": 2.129500150680542, + "learning_rate": 3.5530053155240563e-06, + "loss": 2.4272289276123047, + "step": 302180 + }, + { + "epoch": 0.822648700149721, + "grad_norm": 2.1329505443573, + "learning_rate": 3.5519149516151017e-06, + "loss": 2.4496625900268554, + "step": 302200 + }, + { + "epoch": 0.8227031441404655, + "grad_norm": 2.243518829345703, + "learning_rate": 3.550824587706147e-06, + "loss": 2.4430118560791017, + "step": 302220 + }, + { + "epoch": 0.82275758813121, + "grad_norm": 2.199380874633789, + "learning_rate": 3.5497342237971925e-06, + "loss": 2.469982719421387, + "step": 302240 + }, + { + "epoch": 0.8228120321219545, + "grad_norm": 2.181675434112549, + "learning_rate": 3.548643859888238e-06, + "loss": 2.470357894897461, + "step": 302260 + }, + { + "epoch": 0.822866476112699, + "grad_norm": 2.6076014041900635, + "learning_rate": 3.5475534959792833e-06, + "loss": 2.461256217956543, + "step": 302280 + }, + { + "epoch": 0.8229209201034435, + "grad_norm": 2.171091318130493, + "learning_rate": 3.5464631320703287e-06, + "loss": 2.4648614883422852, + "step": 302300 + }, + { + "epoch": 0.8229753640941881, + "grad_norm": 2.4233222007751465, + "learning_rate": 3.545372768161374e-06, + "loss": 2.4717483520507812, + "step": 302320 + }, + { + "epoch": 0.8230298080849326, + "grad_norm": 2.0853271484375, + "learning_rate": 3.5442824042524194e-06, + "loss": 2.45626220703125, + "step": 302340 + }, + { + "epoch": 0.8230842520756771, + "grad_norm": 2.1991589069366455, + "learning_rate": 3.543192040343465e-06, + "loss": 2.4505619049072265, + "step": 302360 + }, + { + "epoch": 0.8231386960664216, + "grad_norm": 2.3718814849853516, + "learning_rate": 3.54210167643451e-06, + "loss": 2.4633115768432616, + "step": 302380 + }, + { + "epoch": 0.8231931400571662, + "grad_norm": 7.819444179534912, + "learning_rate": 3.5410113125255556e-06, + "loss": 2.4633480072021485, + "step": 302400 + }, + { + "epoch": 0.8232475840479108, + "grad_norm": 2.1843624114990234, + "learning_rate": 3.539920948616601e-06, + "loss": 2.456752586364746, + "step": 302420 + }, + { + "epoch": 0.8233020280386553, + "grad_norm": 2.095085382461548, + "learning_rate": 3.5388305847076464e-06, + "loss": 2.483867645263672, + "step": 302440 + }, + { + "epoch": 0.8233564720293998, + "grad_norm": 2.25152325630188, + "learning_rate": 3.5377402207986917e-06, + "loss": 2.477570152282715, + "step": 302460 + }, + { + "epoch": 0.8234109160201443, + "grad_norm": 2.064549446105957, + "learning_rate": 3.536649856889737e-06, + "loss": 2.461251449584961, + "step": 302480 + }, + { + "epoch": 0.8234653600108888, + "grad_norm": 2.129208564758301, + "learning_rate": 3.5355594929807825e-06, + "loss": 2.459293746948242, + "step": 302500 + }, + { + "epoch": 0.8235198040016333, + "grad_norm": 2.2110705375671387, + "learning_rate": 3.534469129071828e-06, + "loss": 2.438674736022949, + "step": 302520 + }, + { + "epoch": 0.8235742479923779, + "grad_norm": 2.1629059314727783, + "learning_rate": 3.5333787651628733e-06, + "loss": 2.442635726928711, + "step": 302540 + }, + { + "epoch": 0.8236286919831224, + "grad_norm": 3.6725549697875977, + "learning_rate": 3.5322884012539187e-06, + "loss": 2.4729251861572266, + "step": 302560 + }, + { + "epoch": 0.8236831359738669, + "grad_norm": 2.290517807006836, + "learning_rate": 3.531198037344964e-06, + "loss": 2.426327133178711, + "step": 302580 + }, + { + "epoch": 0.8237375799646114, + "grad_norm": 2.256222724914551, + "learning_rate": 3.5301076734360094e-06, + "loss": 2.457784652709961, + "step": 302600 + }, + { + "epoch": 0.8237920239553559, + "grad_norm": 2.084169864654541, + "learning_rate": 3.529017309527055e-06, + "loss": 2.457149696350098, + "step": 302620 + }, + { + "epoch": 0.8238464679461005, + "grad_norm": 2.165405511856079, + "learning_rate": 3.5279269456181002e-06, + "loss": 2.424334716796875, + "step": 302640 + }, + { + "epoch": 0.823900911936845, + "grad_norm": 2.141383409500122, + "learning_rate": 3.5268365817091456e-06, + "loss": 2.454950141906738, + "step": 302660 + }, + { + "epoch": 0.8239553559275895, + "grad_norm": 2.2005698680877686, + "learning_rate": 3.525746217800191e-06, + "loss": 2.457020568847656, + "step": 302680 + }, + { + "epoch": 0.824009799918334, + "grad_norm": 2.326054811477661, + "learning_rate": 3.5246558538912364e-06, + "loss": 2.479656982421875, + "step": 302700 + }, + { + "epoch": 0.8240642439090785, + "grad_norm": 2.1698944568634033, + "learning_rate": 3.5235654899822818e-06, + "loss": 2.478354644775391, + "step": 302720 + }, + { + "epoch": 0.824118687899823, + "grad_norm": 2.132683515548706, + "learning_rate": 3.5224751260733276e-06, + "loss": 2.466364288330078, + "step": 302740 + }, + { + "epoch": 0.8241731318905676, + "grad_norm": 2.1564080715179443, + "learning_rate": 3.5213847621643725e-06, + "loss": 2.44052791595459, + "step": 302760 + }, + { + "epoch": 0.8242275758813121, + "grad_norm": 2.057793378829956, + "learning_rate": 3.5202943982554183e-06, + "loss": 2.4887601852416994, + "step": 302780 + }, + { + "epoch": 0.8242820198720566, + "grad_norm": 2.1729156970977783, + "learning_rate": 3.5192040343464633e-06, + "loss": 2.464190864562988, + "step": 302800 + }, + { + "epoch": 0.8243364638628011, + "grad_norm": 2.2111196517944336, + "learning_rate": 3.518113670437509e-06, + "loss": 2.4439889907836916, + "step": 302820 + }, + { + "epoch": 0.8243909078535456, + "grad_norm": 2.36781644821167, + "learning_rate": 3.517023306528554e-06, + "loss": 2.4715608596801757, + "step": 302840 + }, + { + "epoch": 0.8244453518442902, + "grad_norm": 2.2010722160339355, + "learning_rate": 3.5159329426196e-06, + "loss": 2.46173095703125, + "step": 302860 + }, + { + "epoch": 0.8244997958350347, + "grad_norm": 2.1244001388549805, + "learning_rate": 3.514842578710645e-06, + "loss": 2.4591766357421876, + "step": 302880 + }, + { + "epoch": 0.8245542398257792, + "grad_norm": 2.891005516052246, + "learning_rate": 3.5137522148016907e-06, + "loss": 2.467193031311035, + "step": 302900 + }, + { + "epoch": 0.8246086838165237, + "grad_norm": 2.0688743591308594, + "learning_rate": 3.5126618508927356e-06, + "loss": 2.4668792724609374, + "step": 302920 + }, + { + "epoch": 0.8246631278072682, + "grad_norm": 2.155921697616577, + "learning_rate": 3.5115714869837814e-06, + "loss": 2.4480073928833006, + "step": 302940 + }, + { + "epoch": 0.8247175717980127, + "grad_norm": 2.108665704727173, + "learning_rate": 3.5104811230748264e-06, + "loss": 2.489157867431641, + "step": 302960 + }, + { + "epoch": 0.8247720157887574, + "grad_norm": 2.1359245777130127, + "learning_rate": 3.509390759165872e-06, + "loss": 2.465936851501465, + "step": 302980 + }, + { + "epoch": 0.8248264597795019, + "grad_norm": 2.198276996612549, + "learning_rate": 3.508300395256917e-06, + "loss": 2.4669275283813477, + "step": 303000 + }, + { + "epoch": 0.8248809037702464, + "grad_norm": 2.126007318496704, + "learning_rate": 3.507210031347963e-06, + "loss": 2.4447851181030273, + "step": 303020 + }, + { + "epoch": 0.8249353477609909, + "grad_norm": 2.2416675090789795, + "learning_rate": 3.506119667439008e-06, + "loss": 2.446781349182129, + "step": 303040 + }, + { + "epoch": 0.8249897917517354, + "grad_norm": 2.234290361404419, + "learning_rate": 3.5050293035300537e-06, + "loss": 2.4881052017211913, + "step": 303060 + }, + { + "epoch": 0.8250442357424799, + "grad_norm": 2.406229019165039, + "learning_rate": 3.5039389396210987e-06, + "loss": 2.473225212097168, + "step": 303080 + }, + { + "epoch": 0.8250986797332245, + "grad_norm": 2.1646032333374023, + "learning_rate": 3.5028485757121445e-06, + "loss": 2.4481821060180664, + "step": 303100 + }, + { + "epoch": 0.825153123723969, + "grad_norm": 2.2156152725219727, + "learning_rate": 3.5017582118031895e-06, + "loss": 2.4538528442382814, + "step": 303120 + }, + { + "epoch": 0.8252075677147135, + "grad_norm": 2.1430575847625732, + "learning_rate": 3.5006678478942353e-06, + "loss": 2.474178695678711, + "step": 303140 + }, + { + "epoch": 0.825262011705458, + "grad_norm": 2.2356202602386475, + "learning_rate": 3.4995774839852802e-06, + "loss": 2.451570510864258, + "step": 303160 + }, + { + "epoch": 0.8253164556962025, + "grad_norm": 2.2684667110443115, + "learning_rate": 3.498487120076326e-06, + "loss": 2.4384416580200194, + "step": 303180 + }, + { + "epoch": 0.8253708996869471, + "grad_norm": 2.37821364402771, + "learning_rate": 3.497396756167371e-06, + "loss": 2.499533271789551, + "step": 303200 + }, + { + "epoch": 0.8254253436776916, + "grad_norm": 2.2254068851470947, + "learning_rate": 3.496306392258417e-06, + "loss": 2.451074409484863, + "step": 303220 + }, + { + "epoch": 0.8254797876684361, + "grad_norm": 2.1676197052001953, + "learning_rate": 3.4952160283494618e-06, + "loss": 2.4489372253417967, + "step": 303240 + }, + { + "epoch": 0.8255342316591806, + "grad_norm": 2.238978147506714, + "learning_rate": 3.4941256644405076e-06, + "loss": 2.4542184829711915, + "step": 303260 + }, + { + "epoch": 0.8255886756499251, + "grad_norm": 2.2517733573913574, + "learning_rate": 3.4930353005315526e-06, + "loss": 2.470111846923828, + "step": 303280 + }, + { + "epoch": 0.8256431196406696, + "grad_norm": 2.2556567192077637, + "learning_rate": 3.4919449366225984e-06, + "loss": 2.468317413330078, + "step": 303300 + }, + { + "epoch": 0.8256975636314142, + "grad_norm": 2.126821994781494, + "learning_rate": 3.4908545727136433e-06, + "loss": 2.484976577758789, + "step": 303320 + }, + { + "epoch": 0.8257520076221587, + "grad_norm": 2.2194573879241943, + "learning_rate": 3.489764208804689e-06, + "loss": 2.4519269943237303, + "step": 303340 + }, + { + "epoch": 0.8258064516129032, + "grad_norm": 2.1715569496154785, + "learning_rate": 3.488673844895734e-06, + "loss": 2.418045234680176, + "step": 303360 + }, + { + "epoch": 0.8258608956036477, + "grad_norm": 2.17651104927063, + "learning_rate": 3.48758348098678e-06, + "loss": 2.4249862670898437, + "step": 303380 + }, + { + "epoch": 0.8259153395943922, + "grad_norm": 2.234375, + "learning_rate": 3.486493117077825e-06, + "loss": 2.4570289611816407, + "step": 303400 + }, + { + "epoch": 0.8259697835851368, + "grad_norm": 2.1385297775268555, + "learning_rate": 3.4854027531688707e-06, + "loss": 2.4635007858276365, + "step": 303420 + }, + { + "epoch": 0.8260242275758813, + "grad_norm": 2.245492696762085, + "learning_rate": 3.4843123892599156e-06, + "loss": 2.464923858642578, + "step": 303440 + }, + { + "epoch": 0.8260786715666258, + "grad_norm": 2.1330294609069824, + "learning_rate": 3.4832220253509614e-06, + "loss": 2.4547292709350588, + "step": 303460 + }, + { + "epoch": 0.8261331155573703, + "grad_norm": 2.2431280612945557, + "learning_rate": 3.4821316614420064e-06, + "loss": 2.452731132507324, + "step": 303480 + }, + { + "epoch": 0.8261875595481148, + "grad_norm": 2.585092306137085, + "learning_rate": 3.4810412975330522e-06, + "loss": 2.4158920288085937, + "step": 303500 + }, + { + "epoch": 0.8262420035388593, + "grad_norm": 3.5562376976013184, + "learning_rate": 3.479950933624097e-06, + "loss": 2.457168769836426, + "step": 303520 + }, + { + "epoch": 0.826296447529604, + "grad_norm": 2.2438759803771973, + "learning_rate": 3.478860569715143e-06, + "loss": 2.4527721405029297, + "step": 303540 + }, + { + "epoch": 0.8263508915203485, + "grad_norm": 2.159149408340454, + "learning_rate": 3.477770205806188e-06, + "loss": 2.462723159790039, + "step": 303560 + }, + { + "epoch": 0.826405335511093, + "grad_norm": 2.1955020427703857, + "learning_rate": 3.4766798418972338e-06, + "loss": 2.474930191040039, + "step": 303580 + }, + { + "epoch": 0.8264597795018375, + "grad_norm": 2.169370651245117, + "learning_rate": 3.4755894779882787e-06, + "loss": 2.4579986572265624, + "step": 303600 + }, + { + "epoch": 0.826514223492582, + "grad_norm": 2.319645643234253, + "learning_rate": 3.4744991140793245e-06, + "loss": 2.435913848876953, + "step": 303620 + }, + { + "epoch": 0.8265686674833266, + "grad_norm": 2.2405714988708496, + "learning_rate": 3.4734087501703695e-06, + "loss": 2.456711196899414, + "step": 303640 + }, + { + "epoch": 0.8266231114740711, + "grad_norm": 2.2317144870758057, + "learning_rate": 3.4723183862614153e-06, + "loss": 2.434950256347656, + "step": 303660 + }, + { + "epoch": 0.8266775554648156, + "grad_norm": 4.043149471282959, + "learning_rate": 3.4712280223524603e-06, + "loss": 2.465501403808594, + "step": 303680 + }, + { + "epoch": 0.8267319994555601, + "grad_norm": 2.220015048980713, + "learning_rate": 3.470137658443506e-06, + "loss": 2.4560596466064455, + "step": 303700 + }, + { + "epoch": 0.8267864434463046, + "grad_norm": 6.734568119049072, + "learning_rate": 3.469047294534551e-06, + "loss": 2.4381597518920897, + "step": 303720 + }, + { + "epoch": 0.8268408874370491, + "grad_norm": 2.1345372200012207, + "learning_rate": 3.467956930625597e-06, + "loss": 2.439009094238281, + "step": 303740 + }, + { + "epoch": 0.8268953314277937, + "grad_norm": 2.132018566131592, + "learning_rate": 3.466866566716642e-06, + "loss": 2.4364952087402343, + "step": 303760 + }, + { + "epoch": 0.8269497754185382, + "grad_norm": 2.130510091781616, + "learning_rate": 3.4657762028076876e-06, + "loss": 2.4791135787963867, + "step": 303780 + }, + { + "epoch": 0.8270042194092827, + "grad_norm": 2.181938409805298, + "learning_rate": 3.4646858388987326e-06, + "loss": 2.4785715103149415, + "step": 303800 + }, + { + "epoch": 0.8270586634000272, + "grad_norm": 2.081627130508423, + "learning_rate": 3.4635954749897784e-06, + "loss": 2.453839874267578, + "step": 303820 + }, + { + "epoch": 0.8271131073907717, + "grad_norm": 2.1091253757476807, + "learning_rate": 3.4625051110808234e-06, + "loss": 2.4793113708496093, + "step": 303840 + }, + { + "epoch": 0.8271675513815163, + "grad_norm": 2.1317038536071777, + "learning_rate": 3.461414747171869e-06, + "loss": 2.4551507949829103, + "step": 303860 + }, + { + "epoch": 0.8272219953722608, + "grad_norm": 2.2395122051239014, + "learning_rate": 3.460324383262914e-06, + "loss": 2.4431270599365233, + "step": 303880 + }, + { + "epoch": 0.8272764393630053, + "grad_norm": 2.215568780899048, + "learning_rate": 3.45923401935396e-06, + "loss": 2.4626943588256838, + "step": 303900 + }, + { + "epoch": 0.8273308833537498, + "grad_norm": 2.517679214477539, + "learning_rate": 3.458143655445005e-06, + "loss": 2.4765651702880858, + "step": 303920 + }, + { + "epoch": 0.8273853273444943, + "grad_norm": 2.1714556217193604, + "learning_rate": 3.4570532915360507e-06, + "loss": 2.514416313171387, + "step": 303940 + }, + { + "epoch": 0.8274397713352388, + "grad_norm": 2.1364002227783203, + "learning_rate": 3.4559629276270957e-06, + "loss": 2.4649469375610353, + "step": 303960 + }, + { + "epoch": 0.8274942153259834, + "grad_norm": 2.1137218475341797, + "learning_rate": 3.4548725637181415e-06, + "loss": 2.483974075317383, + "step": 303980 + }, + { + "epoch": 0.8275486593167279, + "grad_norm": 2.137150764465332, + "learning_rate": 3.4537821998091864e-06, + "loss": 2.449506378173828, + "step": 304000 + }, + { + "epoch": 0.8276031033074724, + "grad_norm": 2.2873423099517822, + "learning_rate": 3.4526918359002322e-06, + "loss": 2.482419013977051, + "step": 304020 + }, + { + "epoch": 0.827657547298217, + "grad_norm": 2.080299139022827, + "learning_rate": 3.451601471991277e-06, + "loss": 2.4754478454589846, + "step": 304040 + }, + { + "epoch": 0.8277119912889614, + "grad_norm": 2.205444097518921, + "learning_rate": 3.450511108082323e-06, + "loss": 2.4689659118652343, + "step": 304060 + }, + { + "epoch": 0.827766435279706, + "grad_norm": 2.173759937286377, + "learning_rate": 3.449420744173368e-06, + "loss": 2.528788185119629, + "step": 304080 + }, + { + "epoch": 0.8278208792704506, + "grad_norm": 2.2527551651000977, + "learning_rate": 3.4483303802644134e-06, + "loss": 2.5066537857055664, + "step": 304100 + }, + { + "epoch": 0.8278753232611951, + "grad_norm": 2.396599769592285, + "learning_rate": 3.4472400163554588e-06, + "loss": 2.4505678176879884, + "step": 304120 + }, + { + "epoch": 0.8279297672519396, + "grad_norm": 2.326996326446533, + "learning_rate": 3.446149652446504e-06, + "loss": 2.454693412780762, + "step": 304140 + }, + { + "epoch": 0.8279842112426841, + "grad_norm": 2.126673460006714, + "learning_rate": 3.4450592885375495e-06, + "loss": 2.4324804306030274, + "step": 304160 + }, + { + "epoch": 0.8280386552334286, + "grad_norm": 2.150111198425293, + "learning_rate": 3.443968924628595e-06, + "loss": 2.434731292724609, + "step": 304180 + }, + { + "epoch": 0.8280930992241732, + "grad_norm": 2.476923942565918, + "learning_rate": 3.4428785607196403e-06, + "loss": 2.491319274902344, + "step": 304200 + }, + { + "epoch": 0.8281475432149177, + "grad_norm": 2.250732183456421, + "learning_rate": 3.4417881968106857e-06, + "loss": 2.4576852798461912, + "step": 304220 + }, + { + "epoch": 0.8282019872056622, + "grad_norm": 2.1262454986572266, + "learning_rate": 3.440697832901731e-06, + "loss": 2.4279590606689454, + "step": 304240 + }, + { + "epoch": 0.8282564311964067, + "grad_norm": 2.1616344451904297, + "learning_rate": 3.4396074689927765e-06, + "loss": 2.4744192123413087, + "step": 304260 + }, + { + "epoch": 0.8283108751871512, + "grad_norm": 2.136009931564331, + "learning_rate": 3.438517105083822e-06, + "loss": 2.43536262512207, + "step": 304280 + }, + { + "epoch": 0.8283653191778957, + "grad_norm": 2.064000129699707, + "learning_rate": 3.4374267411748672e-06, + "loss": 2.458211326599121, + "step": 304300 + }, + { + "epoch": 0.8284197631686403, + "grad_norm": 2.0602800846099854, + "learning_rate": 3.4363363772659126e-06, + "loss": 2.4744543075561523, + "step": 304320 + }, + { + "epoch": 0.8284742071593848, + "grad_norm": 2.138176202774048, + "learning_rate": 3.435246013356958e-06, + "loss": 2.485465431213379, + "step": 304340 + }, + { + "epoch": 0.8285286511501293, + "grad_norm": 2.072434425354004, + "learning_rate": 3.4341556494480034e-06, + "loss": 2.448414421081543, + "step": 304360 + }, + { + "epoch": 0.8285830951408738, + "grad_norm": 2.3158395290374756, + "learning_rate": 3.4330652855390488e-06, + "loss": 2.4504552841186524, + "step": 304380 + }, + { + "epoch": 0.8286375391316183, + "grad_norm": 2.2786309719085693, + "learning_rate": 3.431974921630094e-06, + "loss": 2.4226226806640625, + "step": 304400 + }, + { + "epoch": 0.8286919831223629, + "grad_norm": 2.1324851512908936, + "learning_rate": 3.4308845577211395e-06, + "loss": 2.466865348815918, + "step": 304420 + }, + { + "epoch": 0.8287464271131074, + "grad_norm": 2.1028289794921875, + "learning_rate": 3.4297941938121853e-06, + "loss": 2.4625816345214844, + "step": 304440 + }, + { + "epoch": 0.8288008711038519, + "grad_norm": 2.1210079193115234, + "learning_rate": 3.4287038299032303e-06, + "loss": 2.466093635559082, + "step": 304460 + }, + { + "epoch": 0.8288553150945964, + "grad_norm": 2.404226779937744, + "learning_rate": 3.427613465994276e-06, + "loss": 2.4702241897583006, + "step": 304480 + }, + { + "epoch": 0.8289097590853409, + "grad_norm": 2.169057607650757, + "learning_rate": 3.426523102085321e-06, + "loss": 2.438840866088867, + "step": 304500 + }, + { + "epoch": 0.8289642030760854, + "grad_norm": 2.261399984359741, + "learning_rate": 3.425432738176367e-06, + "loss": 2.470210838317871, + "step": 304520 + }, + { + "epoch": 0.82901864706683, + "grad_norm": 2.2825393676757812, + "learning_rate": 3.424342374267412e-06, + "loss": 2.4406408309936523, + "step": 304540 + }, + { + "epoch": 0.8290730910575745, + "grad_norm": 2.21160626411438, + "learning_rate": 3.4232520103584577e-06, + "loss": 2.471915626525879, + "step": 304560 + }, + { + "epoch": 0.829127535048319, + "grad_norm": 2.1331307888031006, + "learning_rate": 3.4221616464495026e-06, + "loss": 2.459783935546875, + "step": 304580 + }, + { + "epoch": 0.8291819790390635, + "grad_norm": 2.1816294193267822, + "learning_rate": 3.4210712825405484e-06, + "loss": 2.474011993408203, + "step": 304600 + }, + { + "epoch": 0.829236423029808, + "grad_norm": 2.0943515300750732, + "learning_rate": 3.4199809186315934e-06, + "loss": 2.4841779708862304, + "step": 304620 + }, + { + "epoch": 0.8292908670205527, + "grad_norm": 2.1625888347625732, + "learning_rate": 3.418890554722639e-06, + "loss": 2.4229211807250977, + "step": 304640 + }, + { + "epoch": 0.8293453110112972, + "grad_norm": 2.2244925498962402, + "learning_rate": 3.417800190813684e-06, + "loss": 2.4559759140014648, + "step": 304660 + }, + { + "epoch": 0.8293997550020417, + "grad_norm": 2.1603527069091797, + "learning_rate": 3.41670982690473e-06, + "loss": 2.476026153564453, + "step": 304680 + }, + { + "epoch": 0.8294541989927862, + "grad_norm": 2.2206783294677734, + "learning_rate": 3.415619462995775e-06, + "loss": 2.481175994873047, + "step": 304700 + }, + { + "epoch": 0.8295086429835307, + "grad_norm": 2.0821151733398438, + "learning_rate": 3.4145290990868207e-06, + "loss": 2.4666858673095704, + "step": 304720 + }, + { + "epoch": 0.8295630869742752, + "grad_norm": 2.184412956237793, + "learning_rate": 3.4134387351778657e-06, + "loss": 2.478183937072754, + "step": 304740 + }, + { + "epoch": 0.8296175309650198, + "grad_norm": 2.1884169578552246, + "learning_rate": 3.4123483712689115e-06, + "loss": 2.451799201965332, + "step": 304760 + }, + { + "epoch": 0.8296719749557643, + "grad_norm": 2.138315200805664, + "learning_rate": 3.4112580073599565e-06, + "loss": 2.437017250061035, + "step": 304780 + }, + { + "epoch": 0.8297264189465088, + "grad_norm": 2.1130635738372803, + "learning_rate": 3.4101676434510023e-06, + "loss": 2.4789316177368166, + "step": 304800 + }, + { + "epoch": 0.8297808629372533, + "grad_norm": 2.3222131729125977, + "learning_rate": 3.4090772795420473e-06, + "loss": 2.4699621200561523, + "step": 304820 + }, + { + "epoch": 0.8298353069279978, + "grad_norm": 2.1092984676361084, + "learning_rate": 3.407986915633093e-06, + "loss": 2.4589284896850585, + "step": 304840 + }, + { + "epoch": 0.8298897509187424, + "grad_norm": 2.2522168159484863, + "learning_rate": 3.406896551724138e-06, + "loss": 2.4529470443725585, + "step": 304860 + }, + { + "epoch": 0.8299441949094869, + "grad_norm": 2.262317180633545, + "learning_rate": 3.405806187815184e-06, + "loss": 2.483204650878906, + "step": 304880 + }, + { + "epoch": 0.8299986389002314, + "grad_norm": 2.19528865814209, + "learning_rate": 3.404715823906229e-06, + "loss": 2.4607521057128907, + "step": 304900 + }, + { + "epoch": 0.8300530828909759, + "grad_norm": 2.1073129177093506, + "learning_rate": 3.4036254599972746e-06, + "loss": 2.443794059753418, + "step": 304920 + }, + { + "epoch": 0.8301075268817204, + "grad_norm": 2.2438693046569824, + "learning_rate": 3.4025350960883196e-06, + "loss": 2.4781803131103515, + "step": 304940 + }, + { + "epoch": 0.8301619708724649, + "grad_norm": 2.215294122695923, + "learning_rate": 3.4014447321793654e-06, + "loss": 2.4631086349487306, + "step": 304960 + }, + { + "epoch": 0.8302164148632095, + "grad_norm": 2.2409868240356445, + "learning_rate": 3.4003543682704103e-06, + "loss": 2.440281295776367, + "step": 304980 + }, + { + "epoch": 0.830270858853954, + "grad_norm": 2.2904529571533203, + "learning_rate": 3.399264004361456e-06, + "loss": 2.453475761413574, + "step": 305000 + }, + { + "epoch": 0.8303253028446985, + "grad_norm": 2.194836378097534, + "learning_rate": 3.398173640452501e-06, + "loss": 2.4709291458129883, + "step": 305020 + }, + { + "epoch": 0.830379746835443, + "grad_norm": 2.137301445007324, + "learning_rate": 3.397083276543547e-06, + "loss": 2.432232666015625, + "step": 305040 + }, + { + "epoch": 0.8304341908261875, + "grad_norm": 2.1885693073272705, + "learning_rate": 3.395992912634592e-06, + "loss": 2.469437599182129, + "step": 305060 + }, + { + "epoch": 0.8304886348169321, + "grad_norm": 2.1311028003692627, + "learning_rate": 3.3949025487256377e-06, + "loss": 2.4260221481323243, + "step": 305080 + }, + { + "epoch": 0.8305430788076766, + "grad_norm": 2.166962146759033, + "learning_rate": 3.3938121848166827e-06, + "loss": 2.469914436340332, + "step": 305100 + }, + { + "epoch": 0.8305975227984211, + "grad_norm": 2.2135472297668457, + "learning_rate": 3.3927218209077285e-06, + "loss": 2.448972702026367, + "step": 305120 + }, + { + "epoch": 0.8306519667891656, + "grad_norm": 2.1579091548919678, + "learning_rate": 3.3916314569987734e-06, + "loss": 2.4612554550170898, + "step": 305140 + }, + { + "epoch": 0.8307064107799101, + "grad_norm": 2.1054890155792236, + "learning_rate": 3.3905410930898192e-06, + "loss": 2.4480777740478517, + "step": 305160 + }, + { + "epoch": 0.8307608547706546, + "grad_norm": 2.936447858810425, + "learning_rate": 3.389450729180864e-06, + "loss": 2.4553022384643555, + "step": 305180 + }, + { + "epoch": 0.8308152987613993, + "grad_norm": 2.1139767169952393, + "learning_rate": 3.38836036527191e-06, + "loss": 2.4449798583984377, + "step": 305200 + }, + { + "epoch": 0.8308697427521438, + "grad_norm": 2.099935293197632, + "learning_rate": 3.387270001362955e-06, + "loss": 2.4577661514282227, + "step": 305220 + }, + { + "epoch": 0.8309241867428883, + "grad_norm": 2.3010106086730957, + "learning_rate": 3.3861796374540008e-06, + "loss": 2.4471851348876954, + "step": 305240 + }, + { + "epoch": 0.8309786307336328, + "grad_norm": 2.1728289127349854, + "learning_rate": 3.3850892735450457e-06, + "loss": 2.467456245422363, + "step": 305260 + }, + { + "epoch": 0.8310330747243773, + "grad_norm": 2.338601589202881, + "learning_rate": 3.3839989096360915e-06, + "loss": 2.446323013305664, + "step": 305280 + }, + { + "epoch": 0.8310875187151218, + "grad_norm": 2.09867787361145, + "learning_rate": 3.3829085457271365e-06, + "loss": 2.4301563262939454, + "step": 305300 + }, + { + "epoch": 0.8311419627058664, + "grad_norm": 2.1550145149230957, + "learning_rate": 3.3818181818181823e-06, + "loss": 2.4777408599853517, + "step": 305320 + }, + { + "epoch": 0.8311964066966109, + "grad_norm": 2.2742793560028076, + "learning_rate": 3.3807278179092273e-06, + "loss": 2.4639570236206056, + "step": 305340 + }, + { + "epoch": 0.8312508506873554, + "grad_norm": 2.1722490787506104, + "learning_rate": 3.379637454000273e-06, + "loss": 2.4987360000610352, + "step": 305360 + }, + { + "epoch": 0.8313052946780999, + "grad_norm": 2.314215898513794, + "learning_rate": 3.378547090091318e-06, + "loss": 2.4696218490600588, + "step": 305380 + }, + { + "epoch": 0.8313597386688444, + "grad_norm": 2.205272912979126, + "learning_rate": 3.377456726182364e-06, + "loss": 2.4688861846923826, + "step": 305400 + }, + { + "epoch": 0.831414182659589, + "grad_norm": 2.2261950969696045, + "learning_rate": 3.376366362273409e-06, + "loss": 2.436928558349609, + "step": 305420 + }, + { + "epoch": 0.8314686266503335, + "grad_norm": 2.079754114151001, + "learning_rate": 3.3752759983644546e-06, + "loss": 2.4528135299682616, + "step": 305440 + }, + { + "epoch": 0.831523070641078, + "grad_norm": 2.35622239112854, + "learning_rate": 3.3741856344554996e-06, + "loss": 2.4476247787475587, + "step": 305460 + }, + { + "epoch": 0.8315775146318225, + "grad_norm": 2.1065196990966797, + "learning_rate": 3.3730952705465454e-06, + "loss": 2.4592763900756838, + "step": 305480 + }, + { + "epoch": 0.831631958622567, + "grad_norm": 2.1652631759643555, + "learning_rate": 3.3720049066375904e-06, + "loss": 2.474535369873047, + "step": 305500 + }, + { + "epoch": 0.8316864026133115, + "grad_norm": 2.149209976196289, + "learning_rate": 3.370914542728636e-06, + "loss": 2.4678491592407226, + "step": 305520 + }, + { + "epoch": 0.8317408466040561, + "grad_norm": 2.0592284202575684, + "learning_rate": 3.369824178819681e-06, + "loss": 2.4561588287353517, + "step": 305540 + }, + { + "epoch": 0.8317952905948006, + "grad_norm": 2.2069895267486572, + "learning_rate": 3.368733814910727e-06, + "loss": 2.450443077087402, + "step": 305560 + }, + { + "epoch": 0.8318497345855451, + "grad_norm": 2.3089308738708496, + "learning_rate": 3.367643451001772e-06, + "loss": 2.472702407836914, + "step": 305580 + }, + { + "epoch": 0.8319041785762896, + "grad_norm": 2.282118320465088, + "learning_rate": 3.3665530870928177e-06, + "loss": 2.4667415618896484, + "step": 305600 + }, + { + "epoch": 0.8319586225670341, + "grad_norm": 2.171278953552246, + "learning_rate": 3.3654627231838627e-06, + "loss": 2.458933448791504, + "step": 305620 + }, + { + "epoch": 0.8320130665577787, + "grad_norm": 2.0920236110687256, + "learning_rate": 3.3643723592749085e-06, + "loss": 2.4879417419433594, + "step": 305640 + }, + { + "epoch": 0.8320675105485232, + "grad_norm": 2.228982448577881, + "learning_rate": 3.3632819953659535e-06, + "loss": 2.456717681884766, + "step": 305660 + }, + { + "epoch": 0.8321219545392677, + "grad_norm": 2.106105089187622, + "learning_rate": 3.3621916314569993e-06, + "loss": 2.4318241119384765, + "step": 305680 + }, + { + "epoch": 0.8321763985300122, + "grad_norm": 2.2385945320129395, + "learning_rate": 3.3611012675480442e-06, + "loss": 2.504447174072266, + "step": 305700 + }, + { + "epoch": 0.8322308425207567, + "grad_norm": 2.229700803756714, + "learning_rate": 3.36001090363909e-06, + "loss": 2.4711822509765624, + "step": 305720 + }, + { + "epoch": 0.8322852865115012, + "grad_norm": 2.33524489402771, + "learning_rate": 3.358920539730135e-06, + "loss": 2.447879600524902, + "step": 305740 + }, + { + "epoch": 0.8323397305022459, + "grad_norm": 2.184288501739502, + "learning_rate": 3.357830175821181e-06, + "loss": 2.4701210021972657, + "step": 305760 + }, + { + "epoch": 0.8323941744929904, + "grad_norm": 2.1591382026672363, + "learning_rate": 3.3567398119122258e-06, + "loss": 2.4626352310180666, + "step": 305780 + }, + { + "epoch": 0.8324486184837349, + "grad_norm": 2.1898555755615234, + "learning_rate": 3.3556494480032716e-06, + "loss": 2.459513854980469, + "step": 305800 + }, + { + "epoch": 0.8325030624744794, + "grad_norm": 2.183820962905884, + "learning_rate": 3.3545590840943165e-06, + "loss": 2.445861053466797, + "step": 305820 + }, + { + "epoch": 0.8325575064652239, + "grad_norm": 2.142435312271118, + "learning_rate": 3.3534687201853623e-06, + "loss": 2.463715744018555, + "step": 305840 + }, + { + "epoch": 0.8326119504559685, + "grad_norm": 2.2108993530273438, + "learning_rate": 3.3523783562764073e-06, + "loss": 2.4520872116088865, + "step": 305860 + }, + { + "epoch": 0.832666394446713, + "grad_norm": 2.2044286727905273, + "learning_rate": 3.351287992367453e-06, + "loss": 2.457185173034668, + "step": 305880 + }, + { + "epoch": 0.8327208384374575, + "grad_norm": 2.1928582191467285, + "learning_rate": 3.350197628458498e-06, + "loss": 2.436082458496094, + "step": 305900 + }, + { + "epoch": 0.832775282428202, + "grad_norm": 2.155156373977661, + "learning_rate": 3.349107264549544e-06, + "loss": 2.4777639389038084, + "step": 305920 + }, + { + "epoch": 0.8328297264189465, + "grad_norm": 2.221428394317627, + "learning_rate": 3.348016900640589e-06, + "loss": 2.45751895904541, + "step": 305940 + }, + { + "epoch": 0.832884170409691, + "grad_norm": 2.5801000595092773, + "learning_rate": 3.3469265367316347e-06, + "loss": 2.4696624755859373, + "step": 305960 + }, + { + "epoch": 0.8329386144004356, + "grad_norm": 2.1570472717285156, + "learning_rate": 3.3458361728226796e-06, + "loss": 2.4578651428222655, + "step": 305980 + }, + { + "epoch": 0.8329930583911801, + "grad_norm": 2.8447203636169434, + "learning_rate": 3.3447458089137254e-06, + "loss": 2.4532352447509767, + "step": 306000 + }, + { + "epoch": 0.8330475023819246, + "grad_norm": 2.1433255672454834, + "learning_rate": 3.3436554450047704e-06, + "loss": 2.4533361434936523, + "step": 306020 + }, + { + "epoch": 0.8331019463726691, + "grad_norm": 2.0880520343780518, + "learning_rate": 3.342565081095816e-06, + "loss": 2.487934875488281, + "step": 306040 + }, + { + "epoch": 0.8331563903634136, + "grad_norm": 2.215543270111084, + "learning_rate": 3.341474717186861e-06, + "loss": 2.426515579223633, + "step": 306060 + }, + { + "epoch": 0.8332108343541582, + "grad_norm": 2.171102285385132, + "learning_rate": 3.340384353277907e-06, + "loss": 2.4825836181640626, + "step": 306080 + }, + { + "epoch": 0.8332652783449027, + "grad_norm": 2.091744899749756, + "learning_rate": 3.3392939893689524e-06, + "loss": 2.453047180175781, + "step": 306100 + }, + { + "epoch": 0.8333197223356472, + "grad_norm": 2.510458469390869, + "learning_rate": 3.3382036254599977e-06, + "loss": 2.430638885498047, + "step": 306120 + }, + { + "epoch": 0.8333741663263917, + "grad_norm": 2.454195499420166, + "learning_rate": 3.337113261551043e-06, + "loss": 2.461172866821289, + "step": 306140 + }, + { + "epoch": 0.8334286103171362, + "grad_norm": 2.1810710430145264, + "learning_rate": 3.3360228976420885e-06, + "loss": 2.462154197692871, + "step": 306160 + }, + { + "epoch": 0.8334830543078807, + "grad_norm": 2.242767572402954, + "learning_rate": 3.334932533733134e-06, + "loss": 2.4794254302978516, + "step": 306180 + }, + { + "epoch": 0.8335374982986253, + "grad_norm": 2.227656602859497, + "learning_rate": 3.333842169824179e-06, + "loss": 2.4449626922607424, + "step": 306200 + }, + { + "epoch": 0.8335919422893698, + "grad_norm": 2.3047127723693848, + "learning_rate": 3.3327518059152247e-06, + "loss": 2.4233760833740234, + "step": 306220 + }, + { + "epoch": 0.8336463862801143, + "grad_norm": 2.130230665206909, + "learning_rate": 3.3316614420062696e-06, + "loss": 2.463527297973633, + "step": 306240 + }, + { + "epoch": 0.8337008302708588, + "grad_norm": 2.103426933288574, + "learning_rate": 3.3305710780973154e-06, + "loss": 2.4518814086914062, + "step": 306260 + }, + { + "epoch": 0.8337552742616033, + "grad_norm": 2.3424320220947266, + "learning_rate": 3.3294807141883604e-06, + "loss": 2.45180606842041, + "step": 306280 + }, + { + "epoch": 0.8338097182523478, + "grad_norm": 2.2596070766448975, + "learning_rate": 3.3283903502794062e-06, + "loss": 2.450198745727539, + "step": 306300 + }, + { + "epoch": 0.8338641622430925, + "grad_norm": 2.213658571243286, + "learning_rate": 3.327299986370451e-06, + "loss": 2.454375076293945, + "step": 306320 + }, + { + "epoch": 0.833918606233837, + "grad_norm": 2.2333219051361084, + "learning_rate": 3.326209622461497e-06, + "loss": 2.46614990234375, + "step": 306340 + }, + { + "epoch": 0.8339730502245815, + "grad_norm": 2.3217053413391113, + "learning_rate": 3.325119258552542e-06, + "loss": 2.44808292388916, + "step": 306360 + }, + { + "epoch": 0.834027494215326, + "grad_norm": 2.2530975341796875, + "learning_rate": 3.3240288946435878e-06, + "loss": 2.443306541442871, + "step": 306380 + }, + { + "epoch": 0.8340819382060705, + "grad_norm": 2.3423874378204346, + "learning_rate": 3.3229385307346327e-06, + "loss": 2.4908849716186525, + "step": 306400 + }, + { + "epoch": 0.8341363821968151, + "grad_norm": 2.2345807552337646, + "learning_rate": 3.3218481668256785e-06, + "loss": 2.4711187362670897, + "step": 306420 + }, + { + "epoch": 0.8341908261875596, + "grad_norm": 2.1731886863708496, + "learning_rate": 3.3207578029167235e-06, + "loss": 2.5028377532958985, + "step": 306440 + }, + { + "epoch": 0.8342452701783041, + "grad_norm": 2.364593982696533, + "learning_rate": 3.3196674390077693e-06, + "loss": 2.4448341369628905, + "step": 306460 + }, + { + "epoch": 0.8342997141690486, + "grad_norm": 2.153322219848633, + "learning_rate": 3.3185770750988143e-06, + "loss": 2.4743524551391602, + "step": 306480 + }, + { + "epoch": 0.8343541581597931, + "grad_norm": 2.099527597427368, + "learning_rate": 3.31748671118986e-06, + "loss": 2.448191452026367, + "step": 306500 + }, + { + "epoch": 0.8344086021505376, + "grad_norm": 2.144587993621826, + "learning_rate": 3.316396347280905e-06, + "loss": 2.4751529693603516, + "step": 306520 + }, + { + "epoch": 0.8344630461412822, + "grad_norm": 2.1567635536193848, + "learning_rate": 3.315305983371951e-06, + "loss": 2.473349952697754, + "step": 306540 + }, + { + "epoch": 0.8345174901320267, + "grad_norm": 2.186215400695801, + "learning_rate": 3.314215619462996e-06, + "loss": 2.4429426193237305, + "step": 306560 + }, + { + "epoch": 0.8345719341227712, + "grad_norm": 2.4325578212738037, + "learning_rate": 3.3131252555540416e-06, + "loss": 2.491769790649414, + "step": 306580 + }, + { + "epoch": 0.8346263781135157, + "grad_norm": 2.1586172580718994, + "learning_rate": 3.3120348916450866e-06, + "loss": 2.463186836242676, + "step": 306600 + }, + { + "epoch": 0.8346808221042602, + "grad_norm": 2.142259359359741, + "learning_rate": 3.3109445277361324e-06, + "loss": 2.490799140930176, + "step": 306620 + }, + { + "epoch": 0.8347352660950048, + "grad_norm": 2.2034833431243896, + "learning_rate": 3.3098541638271774e-06, + "loss": 2.4626752853393556, + "step": 306640 + }, + { + "epoch": 0.8347897100857493, + "grad_norm": 2.3412461280822754, + "learning_rate": 3.308763799918223e-06, + "loss": 2.438037872314453, + "step": 306660 + }, + { + "epoch": 0.8348441540764938, + "grad_norm": 2.191511631011963, + "learning_rate": 3.307673436009268e-06, + "loss": 2.4667476654052733, + "step": 306680 + }, + { + "epoch": 0.8348985980672383, + "grad_norm": 2.2047643661499023, + "learning_rate": 3.306583072100314e-06, + "loss": 2.468806266784668, + "step": 306700 + }, + { + "epoch": 0.8349530420579828, + "grad_norm": 2.345555543899536, + "learning_rate": 3.305492708191359e-06, + "loss": 2.443142127990723, + "step": 306720 + }, + { + "epoch": 0.8350074860487273, + "grad_norm": 2.0673000812530518, + "learning_rate": 3.3044023442824047e-06, + "loss": 2.4126710891723633, + "step": 306740 + }, + { + "epoch": 0.8350619300394719, + "grad_norm": 2.0811080932617188, + "learning_rate": 3.3033119803734497e-06, + "loss": 2.4237014770507814, + "step": 306760 + }, + { + "epoch": 0.8351163740302164, + "grad_norm": 2.115900754928589, + "learning_rate": 3.3022216164644955e-06, + "loss": 2.4823816299438475, + "step": 306780 + }, + { + "epoch": 0.8351708180209609, + "grad_norm": 2.0779476165771484, + "learning_rate": 3.3011312525555404e-06, + "loss": 2.474812698364258, + "step": 306800 + }, + { + "epoch": 0.8352252620117054, + "grad_norm": 2.0526955127716064, + "learning_rate": 3.3000408886465862e-06, + "loss": 2.4742767333984377, + "step": 306820 + }, + { + "epoch": 0.8352797060024499, + "grad_norm": 2.154292106628418, + "learning_rate": 3.298950524737631e-06, + "loss": 2.4939483642578124, + "step": 306840 + }, + { + "epoch": 0.8353341499931946, + "grad_norm": 2.1807925701141357, + "learning_rate": 3.297860160828677e-06, + "loss": 2.452671241760254, + "step": 306860 + }, + { + "epoch": 0.835388593983939, + "grad_norm": 2.1901395320892334, + "learning_rate": 3.296769796919722e-06, + "loss": 2.471615028381348, + "step": 306880 + }, + { + "epoch": 0.8354430379746836, + "grad_norm": 2.2629878520965576, + "learning_rate": 3.295679433010768e-06, + "loss": 2.467458724975586, + "step": 306900 + }, + { + "epoch": 0.8354974819654281, + "grad_norm": 2.2491719722747803, + "learning_rate": 3.2945890691018128e-06, + "loss": 2.4517927169799805, + "step": 306920 + }, + { + "epoch": 0.8355519259561726, + "grad_norm": 2.181966781616211, + "learning_rate": 3.2934987051928586e-06, + "loss": 2.4385091781616213, + "step": 306940 + }, + { + "epoch": 0.8356063699469171, + "grad_norm": 2.3194544315338135, + "learning_rate": 3.2924083412839035e-06, + "loss": 2.4748538970947265, + "step": 306960 + }, + { + "epoch": 0.8356608139376617, + "grad_norm": 2.1350953578948975, + "learning_rate": 3.2913179773749493e-06, + "loss": 2.4870220184326173, + "step": 306980 + }, + { + "epoch": 0.8357152579284062, + "grad_norm": 2.234923839569092, + "learning_rate": 3.2902276134659943e-06, + "loss": 2.4488277435302734, + "step": 307000 + }, + { + "epoch": 0.8357697019191507, + "grad_norm": 2.260326623916626, + "learning_rate": 3.28913724955704e-06, + "loss": 2.4741668701171875, + "step": 307020 + }, + { + "epoch": 0.8358241459098952, + "grad_norm": 2.07431697845459, + "learning_rate": 3.288046885648085e-06, + "loss": 2.44393310546875, + "step": 307040 + }, + { + "epoch": 0.8358785899006397, + "grad_norm": 2.091407537460327, + "learning_rate": 3.286956521739131e-06, + "loss": 2.4662498474121093, + "step": 307060 + }, + { + "epoch": 0.8359330338913843, + "grad_norm": 2.348522186279297, + "learning_rate": 3.285866157830176e-06, + "loss": 2.4657304763793944, + "step": 307080 + }, + { + "epoch": 0.8359874778821288, + "grad_norm": 2.2778127193450928, + "learning_rate": 3.2847757939212216e-06, + "loss": 2.482528305053711, + "step": 307100 + }, + { + "epoch": 0.8360419218728733, + "grad_norm": 2.277686834335327, + "learning_rate": 3.2836854300122666e-06, + "loss": 2.4629640579223633, + "step": 307120 + }, + { + "epoch": 0.8360963658636178, + "grad_norm": 2.1943185329437256, + "learning_rate": 3.2825950661033124e-06, + "loss": 2.4438602447509767, + "step": 307140 + }, + { + "epoch": 0.8361508098543623, + "grad_norm": 2.140704870223999, + "learning_rate": 3.2815047021943574e-06, + "loss": 2.4675453186035154, + "step": 307160 + }, + { + "epoch": 0.8362052538451068, + "grad_norm": 2.099386692047119, + "learning_rate": 3.280414338285403e-06, + "loss": 2.477709197998047, + "step": 307180 + }, + { + "epoch": 0.8362596978358514, + "grad_norm": 2.120615005493164, + "learning_rate": 3.279323974376448e-06, + "loss": 2.47421875, + "step": 307200 + }, + { + "epoch": 0.8363141418265959, + "grad_norm": 2.106839418411255, + "learning_rate": 3.278233610467494e-06, + "loss": 2.497504234313965, + "step": 307220 + }, + { + "epoch": 0.8363685858173404, + "grad_norm": 3.071495771408081, + "learning_rate": 3.277143246558539e-06, + "loss": 2.492755126953125, + "step": 307240 + }, + { + "epoch": 0.8364230298080849, + "grad_norm": 2.117466688156128, + "learning_rate": 3.2760528826495847e-06, + "loss": 2.4446643829345702, + "step": 307260 + }, + { + "epoch": 0.8364774737988294, + "grad_norm": 2.357306957244873, + "learning_rate": 3.2749625187406297e-06, + "loss": 2.481509780883789, + "step": 307280 + }, + { + "epoch": 0.836531917789574, + "grad_norm": 2.124091863632202, + "learning_rate": 3.2738721548316755e-06, + "loss": 2.4621959686279298, + "step": 307300 + }, + { + "epoch": 0.8365863617803185, + "grad_norm": 2.16664719581604, + "learning_rate": 3.2727817909227205e-06, + "loss": 2.4497377395629885, + "step": 307320 + }, + { + "epoch": 0.836640805771063, + "grad_norm": 2.139960765838623, + "learning_rate": 3.2716914270137663e-06, + "loss": 2.469817543029785, + "step": 307340 + }, + { + "epoch": 0.8366952497618075, + "grad_norm": 2.2828681468963623, + "learning_rate": 3.2706010631048112e-06, + "loss": 2.4656824111938476, + "step": 307360 + }, + { + "epoch": 0.836749693752552, + "grad_norm": 2.1250717639923096, + "learning_rate": 3.269510699195857e-06, + "loss": 2.4710506439208983, + "step": 307380 + }, + { + "epoch": 0.8368041377432965, + "grad_norm": 2.168257474899292, + "learning_rate": 3.268420335286902e-06, + "loss": 2.456283950805664, + "step": 307400 + }, + { + "epoch": 0.8368585817340412, + "grad_norm": 2.175391912460327, + "learning_rate": 3.267329971377948e-06, + "loss": 2.4775516510009767, + "step": 307420 + }, + { + "epoch": 0.8369130257247857, + "grad_norm": 2.1983492374420166, + "learning_rate": 3.2662396074689928e-06, + "loss": 2.436379241943359, + "step": 307440 + }, + { + "epoch": 0.8369674697155302, + "grad_norm": 2.2596688270568848, + "learning_rate": 3.2651492435600386e-06, + "loss": 2.434809684753418, + "step": 307460 + }, + { + "epoch": 0.8370219137062747, + "grad_norm": 2.192554473876953, + "learning_rate": 3.2640588796510836e-06, + "loss": 2.4614233016967773, + "step": 307480 + }, + { + "epoch": 0.8370763576970192, + "grad_norm": 4.387274742126465, + "learning_rate": 3.2629685157421294e-06, + "loss": 2.4623022079467773, + "step": 307500 + }, + { + "epoch": 0.8371308016877637, + "grad_norm": 2.153883934020996, + "learning_rate": 3.2618781518331743e-06, + "loss": 2.469378852844238, + "step": 307520 + }, + { + "epoch": 0.8371852456785083, + "grad_norm": 2.2211825847625732, + "learning_rate": 3.26078778792422e-06, + "loss": 2.451165199279785, + "step": 307540 + }, + { + "epoch": 0.8372396896692528, + "grad_norm": 2.2093913555145264, + "learning_rate": 3.259697424015265e-06, + "loss": 2.469611167907715, + "step": 307560 + }, + { + "epoch": 0.8372941336599973, + "grad_norm": 2.195302724838257, + "learning_rate": 3.258607060106311e-06, + "loss": 2.463166618347168, + "step": 307580 + }, + { + "epoch": 0.8373485776507418, + "grad_norm": 2.2551400661468506, + "learning_rate": 3.257516696197356e-06, + "loss": 2.4667667388916015, + "step": 307600 + }, + { + "epoch": 0.8374030216414863, + "grad_norm": 2.56738018989563, + "learning_rate": 3.2564263322884017e-06, + "loss": 2.4727859497070312, + "step": 307620 + }, + { + "epoch": 0.8374574656322309, + "grad_norm": 2.2551567554473877, + "learning_rate": 3.2553359683794466e-06, + "loss": 2.4804828643798826, + "step": 307640 + }, + { + "epoch": 0.8375119096229754, + "grad_norm": 2.2403995990753174, + "learning_rate": 3.2542456044704924e-06, + "loss": 2.458493232727051, + "step": 307660 + }, + { + "epoch": 0.8375663536137199, + "grad_norm": 2.1772665977478027, + "learning_rate": 3.2531552405615374e-06, + "loss": 2.4669952392578125, + "step": 307680 + }, + { + "epoch": 0.8376207976044644, + "grad_norm": 2.199685573577881, + "learning_rate": 3.2520648766525832e-06, + "loss": 2.4398591995239256, + "step": 307700 + }, + { + "epoch": 0.8376752415952089, + "grad_norm": 2.2502574920654297, + "learning_rate": 3.250974512743628e-06, + "loss": 2.4430959701538084, + "step": 307720 + }, + { + "epoch": 0.8377296855859534, + "grad_norm": 2.7023680210113525, + "learning_rate": 3.249884148834674e-06, + "loss": 2.465561294555664, + "step": 307740 + }, + { + "epoch": 0.837784129576698, + "grad_norm": 2.1979455947875977, + "learning_rate": 3.248793784925719e-06, + "loss": 2.450435447692871, + "step": 307760 + }, + { + "epoch": 0.8378385735674425, + "grad_norm": 2.3690378665924072, + "learning_rate": 3.2477034210167648e-06, + "loss": 2.4459112167358397, + "step": 307780 + }, + { + "epoch": 0.837893017558187, + "grad_norm": 2.2798914909362793, + "learning_rate": 3.24661305710781e-06, + "loss": 2.467792510986328, + "step": 307800 + }, + { + "epoch": 0.8379474615489315, + "grad_norm": 2.162426471710205, + "learning_rate": 3.2455226931988555e-06, + "loss": 2.4317975997924806, + "step": 307820 + }, + { + "epoch": 0.838001905539676, + "grad_norm": 2.291078567504883, + "learning_rate": 3.244432329289901e-06, + "loss": 2.4770843505859377, + "step": 307840 + }, + { + "epoch": 0.8380563495304206, + "grad_norm": 2.3787829875946045, + "learning_rate": 3.2433419653809463e-06, + "loss": 2.4673583984375, + "step": 307860 + }, + { + "epoch": 0.8381107935211651, + "grad_norm": 2.1357309818267822, + "learning_rate": 3.2422516014719917e-06, + "loss": 2.4770427703857423, + "step": 307880 + }, + { + "epoch": 0.8381652375119096, + "grad_norm": 2.179305076599121, + "learning_rate": 3.241161237563037e-06, + "loss": 2.4540855407714846, + "step": 307900 + }, + { + "epoch": 0.8382196815026541, + "grad_norm": 2.0844051837921143, + "learning_rate": 3.2400708736540825e-06, + "loss": 2.4479501724243162, + "step": 307920 + }, + { + "epoch": 0.8382741254933986, + "grad_norm": 2.2862114906311035, + "learning_rate": 3.238980509745128e-06, + "loss": 2.4481124877929688, + "step": 307940 + }, + { + "epoch": 0.8383285694841431, + "grad_norm": 2.206423282623291, + "learning_rate": 3.2378901458361732e-06, + "loss": 2.4525238037109376, + "step": 307960 + }, + { + "epoch": 0.8383830134748878, + "grad_norm": 2.1338324546813965, + "learning_rate": 3.2367997819272186e-06, + "loss": 2.4569583892822267, + "step": 307980 + }, + { + "epoch": 0.8384374574656323, + "grad_norm": 2.1609838008880615, + "learning_rate": 3.235709418018264e-06, + "loss": 2.491690254211426, + "step": 308000 + }, + { + "epoch": 0.8384919014563768, + "grad_norm": 2.182049036026001, + "learning_rate": 3.2346190541093094e-06, + "loss": 2.465208053588867, + "step": 308020 + }, + { + "epoch": 0.8385463454471213, + "grad_norm": 2.2969179153442383, + "learning_rate": 3.2335286902003548e-06, + "loss": 2.4595788955688476, + "step": 308040 + }, + { + "epoch": 0.8386007894378658, + "grad_norm": 2.124053716659546, + "learning_rate": 3.2324383262914e-06, + "loss": 2.4935861587524415, + "step": 308060 + }, + { + "epoch": 0.8386552334286104, + "grad_norm": 2.6006970405578613, + "learning_rate": 3.2313479623824455e-06, + "loss": 2.449694061279297, + "step": 308080 + }, + { + "epoch": 0.8387096774193549, + "grad_norm": 2.399310827255249, + "learning_rate": 3.230257598473491e-06, + "loss": 2.4420749664306642, + "step": 308100 + }, + { + "epoch": 0.8387641214100994, + "grad_norm": 2.1344404220581055, + "learning_rate": 3.2291672345645363e-06, + "loss": 2.4593692779541017, + "step": 308120 + }, + { + "epoch": 0.8388185654008439, + "grad_norm": 2.118391752243042, + "learning_rate": 3.2280768706555817e-06, + "loss": 2.4979209899902344, + "step": 308140 + }, + { + "epoch": 0.8388730093915884, + "grad_norm": 2.401531219482422, + "learning_rate": 3.226986506746627e-06, + "loss": 2.4497575759887695, + "step": 308160 + }, + { + "epoch": 0.8389274533823329, + "grad_norm": 2.131190061569214, + "learning_rate": 3.2258961428376725e-06, + "loss": 2.4703144073486327, + "step": 308180 + }, + { + "epoch": 0.8389818973730775, + "grad_norm": 2.3478176593780518, + "learning_rate": 3.224805778928718e-06, + "loss": 2.478018569946289, + "step": 308200 + }, + { + "epoch": 0.839036341363822, + "grad_norm": 2.199016571044922, + "learning_rate": 3.2237154150197632e-06, + "loss": 2.441293144226074, + "step": 308220 + }, + { + "epoch": 0.8390907853545665, + "grad_norm": 2.2084851264953613, + "learning_rate": 3.2226250511108086e-06, + "loss": 2.439841461181641, + "step": 308240 + }, + { + "epoch": 0.839145229345311, + "grad_norm": 2.184558629989624, + "learning_rate": 3.221534687201854e-06, + "loss": 2.4741369247436524, + "step": 308260 + }, + { + "epoch": 0.8391996733360555, + "grad_norm": 2.135152816772461, + "learning_rate": 3.2204443232928994e-06, + "loss": 2.4547874450683596, + "step": 308280 + }, + { + "epoch": 0.8392541173268001, + "grad_norm": 2.832486629486084, + "learning_rate": 3.2193539593839444e-06, + "loss": 2.480840301513672, + "step": 308300 + }, + { + "epoch": 0.8393085613175446, + "grad_norm": 2.1988751888275146, + "learning_rate": 3.21826359547499e-06, + "loss": 2.451776313781738, + "step": 308320 + }, + { + "epoch": 0.8393630053082891, + "grad_norm": 2.226414442062378, + "learning_rate": 3.217173231566035e-06, + "loss": 2.4504871368408203, + "step": 308340 + }, + { + "epoch": 0.8394174492990336, + "grad_norm": 2.1974356174468994, + "learning_rate": 3.216082867657081e-06, + "loss": 2.510205841064453, + "step": 308360 + }, + { + "epoch": 0.8394718932897781, + "grad_norm": 2.211817741394043, + "learning_rate": 3.214992503748126e-06, + "loss": 2.489768218994141, + "step": 308380 + }, + { + "epoch": 0.8395263372805226, + "grad_norm": 2.0252442359924316, + "learning_rate": 3.2139021398391717e-06, + "loss": 2.437493324279785, + "step": 308400 + }, + { + "epoch": 0.8395807812712672, + "grad_norm": 2.157477855682373, + "learning_rate": 3.2128117759302167e-06, + "loss": 2.5009416580200194, + "step": 308420 + }, + { + "epoch": 0.8396352252620117, + "grad_norm": 2.1731045246124268, + "learning_rate": 3.2117214120212625e-06, + "loss": 2.4655662536621095, + "step": 308440 + }, + { + "epoch": 0.8396896692527562, + "grad_norm": 2.0461862087249756, + "learning_rate": 3.2106310481123075e-06, + "loss": 2.4596837997436523, + "step": 308460 + }, + { + "epoch": 0.8397441132435007, + "grad_norm": 2.1798086166381836, + "learning_rate": 3.2095406842033533e-06, + "loss": 2.481204795837402, + "step": 308480 + }, + { + "epoch": 0.8397985572342452, + "grad_norm": 2.2229645252227783, + "learning_rate": 3.2084503202943982e-06, + "loss": 2.441004753112793, + "step": 308500 + }, + { + "epoch": 0.8398530012249897, + "grad_norm": 2.1401727199554443, + "learning_rate": 3.207359956385444e-06, + "loss": 2.4668437957763674, + "step": 308520 + }, + { + "epoch": 0.8399074452157343, + "grad_norm": 2.2710511684417725, + "learning_rate": 3.206269592476489e-06, + "loss": 2.444001579284668, + "step": 308540 + }, + { + "epoch": 0.8399618892064789, + "grad_norm": 2.147351026535034, + "learning_rate": 3.205179228567535e-06, + "loss": 2.4423906326293947, + "step": 308560 + }, + { + "epoch": 0.8400163331972234, + "grad_norm": 2.1347243785858154, + "learning_rate": 3.2040888646585798e-06, + "loss": 2.4435535430908204, + "step": 308580 + }, + { + "epoch": 0.8400707771879679, + "grad_norm": 2.1332650184631348, + "learning_rate": 3.2029985007496256e-06, + "loss": 2.4618438720703124, + "step": 308600 + }, + { + "epoch": 0.8401252211787124, + "grad_norm": 2.2204747200012207, + "learning_rate": 3.2019081368406705e-06, + "loss": 2.4501169204711912, + "step": 308620 + }, + { + "epoch": 0.840179665169457, + "grad_norm": 2.1979949474334717, + "learning_rate": 3.2008177729317163e-06, + "loss": 2.466474151611328, + "step": 308640 + }, + { + "epoch": 0.8402341091602015, + "grad_norm": 2.1740763187408447, + "learning_rate": 3.1997274090227613e-06, + "loss": 2.45048828125, + "step": 308660 + }, + { + "epoch": 0.840288553150946, + "grad_norm": 2.022533893585205, + "learning_rate": 3.198637045113807e-06, + "loss": 2.43920955657959, + "step": 308680 + }, + { + "epoch": 0.8403429971416905, + "grad_norm": 2.2332942485809326, + "learning_rate": 3.197546681204852e-06, + "loss": 2.469182014465332, + "step": 308700 + }, + { + "epoch": 0.840397441132435, + "grad_norm": 2.238438844680786, + "learning_rate": 3.196456317295898e-06, + "loss": 2.4308319091796875, + "step": 308720 + }, + { + "epoch": 0.8404518851231795, + "grad_norm": 2.1369869709014893, + "learning_rate": 3.195365953386943e-06, + "loss": 2.4654010772705077, + "step": 308740 + }, + { + "epoch": 0.8405063291139241, + "grad_norm": 2.203949451446533, + "learning_rate": 3.1942755894779887e-06, + "loss": 2.471665954589844, + "step": 308760 + }, + { + "epoch": 0.8405607731046686, + "grad_norm": 2.074122667312622, + "learning_rate": 3.1931852255690336e-06, + "loss": 2.4593175888061523, + "step": 308780 + }, + { + "epoch": 0.8406152170954131, + "grad_norm": 3.463984727859497, + "learning_rate": 3.1920948616600794e-06, + "loss": 2.4657482147216796, + "step": 308800 + }, + { + "epoch": 0.8406696610861576, + "grad_norm": 2.1912009716033936, + "learning_rate": 3.1910044977511244e-06, + "loss": 2.45584716796875, + "step": 308820 + }, + { + "epoch": 0.8407241050769021, + "grad_norm": 2.2051563262939453, + "learning_rate": 3.18991413384217e-06, + "loss": 2.4403923034667967, + "step": 308840 + }, + { + "epoch": 0.8407785490676467, + "grad_norm": 2.166795015335083, + "learning_rate": 3.188823769933215e-06, + "loss": 2.456346130371094, + "step": 308860 + }, + { + "epoch": 0.8408329930583912, + "grad_norm": 2.2892048358917236, + "learning_rate": 3.187733406024261e-06, + "loss": 2.4413774490356444, + "step": 308880 + }, + { + "epoch": 0.8408874370491357, + "grad_norm": 2.0891530513763428, + "learning_rate": 3.186643042115306e-06, + "loss": 2.4469839096069337, + "step": 308900 + }, + { + "epoch": 0.8409418810398802, + "grad_norm": 2.2333033084869385, + "learning_rate": 3.1855526782063517e-06, + "loss": 2.47973690032959, + "step": 308920 + }, + { + "epoch": 0.8409963250306247, + "grad_norm": 2.081514358520508, + "learning_rate": 3.1844623142973967e-06, + "loss": 2.467250633239746, + "step": 308940 + }, + { + "epoch": 0.8410507690213692, + "grad_norm": 2.1149840354919434, + "learning_rate": 3.1833719503884425e-06, + "loss": 2.4762441635131838, + "step": 308960 + }, + { + "epoch": 0.8411052130121138, + "grad_norm": 2.167435884475708, + "learning_rate": 3.1822815864794875e-06, + "loss": 2.4474470138549806, + "step": 308980 + }, + { + "epoch": 0.8411596570028583, + "grad_norm": 2.245513677597046, + "learning_rate": 3.1811912225705333e-06, + "loss": 2.4769720077514648, + "step": 309000 + }, + { + "epoch": 0.8412141009936028, + "grad_norm": 2.1406311988830566, + "learning_rate": 3.1801008586615783e-06, + "loss": 2.4452798843383787, + "step": 309020 + }, + { + "epoch": 0.8412685449843473, + "grad_norm": 2.1548690795898438, + "learning_rate": 3.179010494752624e-06, + "loss": 2.454709053039551, + "step": 309040 + }, + { + "epoch": 0.8413229889750918, + "grad_norm": 2.2218971252441406, + "learning_rate": 3.177920130843669e-06, + "loss": 2.4389549255371095, + "step": 309060 + }, + { + "epoch": 0.8413774329658364, + "grad_norm": 2.144990921020508, + "learning_rate": 3.176829766934715e-06, + "loss": 2.45540771484375, + "step": 309080 + }, + { + "epoch": 0.841431876956581, + "grad_norm": 3.1193935871124268, + "learning_rate": 3.17573940302576e-06, + "loss": 2.4554370880126952, + "step": 309100 + }, + { + "epoch": 0.8414863209473255, + "grad_norm": 2.2006635665893555, + "learning_rate": 3.1746490391168056e-06, + "loss": 2.490598678588867, + "step": 309120 + }, + { + "epoch": 0.84154076493807, + "grad_norm": 2.156881093978882, + "learning_rate": 3.1735586752078506e-06, + "loss": 2.4520936965942384, + "step": 309140 + }, + { + "epoch": 0.8415952089288145, + "grad_norm": 2.106018304824829, + "learning_rate": 3.1724683112988964e-06, + "loss": 2.465687370300293, + "step": 309160 + }, + { + "epoch": 0.841649652919559, + "grad_norm": 2.329751968383789, + "learning_rate": 3.1713779473899413e-06, + "loss": 2.483169746398926, + "step": 309180 + }, + { + "epoch": 0.8417040969103036, + "grad_norm": 2.1639564037323, + "learning_rate": 3.170287583480987e-06, + "loss": 2.470989799499512, + "step": 309200 + }, + { + "epoch": 0.8417585409010481, + "grad_norm": 2.2471752166748047, + "learning_rate": 3.169197219572032e-06, + "loss": 2.4441638946533204, + "step": 309220 + }, + { + "epoch": 0.8418129848917926, + "grad_norm": 2.2868995666503906, + "learning_rate": 3.168106855663078e-06, + "loss": 2.4437467575073244, + "step": 309240 + }, + { + "epoch": 0.8418674288825371, + "grad_norm": 2.259819746017456, + "learning_rate": 3.167016491754123e-06, + "loss": 2.4590686798095702, + "step": 309260 + }, + { + "epoch": 0.8419218728732816, + "grad_norm": 2.1640777587890625, + "learning_rate": 3.1659261278451687e-06, + "loss": 2.4802913665771484, + "step": 309280 + }, + { + "epoch": 0.8419763168640262, + "grad_norm": 2.096191167831421, + "learning_rate": 3.1648357639362136e-06, + "loss": 2.467943572998047, + "step": 309300 + }, + { + "epoch": 0.8420307608547707, + "grad_norm": 2.1629092693328857, + "learning_rate": 3.1637454000272595e-06, + "loss": 2.467557907104492, + "step": 309320 + }, + { + "epoch": 0.8420852048455152, + "grad_norm": 2.1450462341308594, + "learning_rate": 3.1626550361183044e-06, + "loss": 2.456623840332031, + "step": 309340 + }, + { + "epoch": 0.8421396488362597, + "grad_norm": 2.14745831489563, + "learning_rate": 3.1615646722093502e-06, + "loss": 2.4816347122192384, + "step": 309360 + }, + { + "epoch": 0.8421940928270042, + "grad_norm": 2.1236493587493896, + "learning_rate": 3.160474308300395e-06, + "loss": 2.4731550216674805, + "step": 309380 + }, + { + "epoch": 0.8422485368177487, + "grad_norm": 2.198995590209961, + "learning_rate": 3.159383944391441e-06, + "loss": 2.472142791748047, + "step": 309400 + }, + { + "epoch": 0.8423029808084933, + "grad_norm": 2.1560885906219482, + "learning_rate": 3.158293580482486e-06, + "loss": 2.4653173446655274, + "step": 309420 + }, + { + "epoch": 0.8423574247992378, + "grad_norm": 2.3287594318389893, + "learning_rate": 3.1572032165735318e-06, + "loss": 2.4480167388916017, + "step": 309440 + }, + { + "epoch": 0.8424118687899823, + "grad_norm": 2.0864930152893066, + "learning_rate": 3.1561128526645767e-06, + "loss": 2.4548372268676757, + "step": 309460 + }, + { + "epoch": 0.8424663127807268, + "grad_norm": 2.2125988006591797, + "learning_rate": 3.1550224887556225e-06, + "loss": 2.431964111328125, + "step": 309480 + }, + { + "epoch": 0.8425207567714713, + "grad_norm": 2.256221294403076, + "learning_rate": 3.153932124846668e-06, + "loss": 2.4596519470214844, + "step": 309500 + }, + { + "epoch": 0.8425752007622159, + "grad_norm": 2.472249984741211, + "learning_rate": 3.1528417609377133e-06, + "loss": 2.4403114318847656, + "step": 309520 + }, + { + "epoch": 0.8426296447529604, + "grad_norm": 2.1664326190948486, + "learning_rate": 3.1517513970287587e-06, + "loss": 2.4748682022094726, + "step": 309540 + }, + { + "epoch": 0.8426840887437049, + "grad_norm": 2.201761245727539, + "learning_rate": 3.150661033119804e-06, + "loss": 2.4727888107299805, + "step": 309560 + }, + { + "epoch": 0.8427385327344494, + "grad_norm": 2.2437429428100586, + "learning_rate": 3.1495706692108495e-06, + "loss": 2.4400197982788088, + "step": 309580 + }, + { + "epoch": 0.8427929767251939, + "grad_norm": 2.110671043395996, + "learning_rate": 3.148480305301895e-06, + "loss": 2.453703689575195, + "step": 309600 + }, + { + "epoch": 0.8428474207159384, + "grad_norm": 2.215358257293701, + "learning_rate": 3.1473899413929402e-06, + "loss": 2.414950942993164, + "step": 309620 + }, + { + "epoch": 0.842901864706683, + "grad_norm": 2.174013376235962, + "learning_rate": 3.1462995774839856e-06, + "loss": 2.4633533477783205, + "step": 309640 + }, + { + "epoch": 0.8429563086974275, + "grad_norm": 2.222120523452759, + "learning_rate": 3.145209213575031e-06, + "loss": 2.470562744140625, + "step": 309660 + }, + { + "epoch": 0.843010752688172, + "grad_norm": 2.1421520709991455, + "learning_rate": 3.1441188496660764e-06, + "loss": 2.485869789123535, + "step": 309680 + }, + { + "epoch": 0.8430651966789166, + "grad_norm": 2.2204506397247314, + "learning_rate": 3.1430284857571218e-06, + "loss": 2.4544654846191407, + "step": 309700 + }, + { + "epoch": 0.843119640669661, + "grad_norm": 2.1837158203125, + "learning_rate": 3.141938121848167e-06, + "loss": 2.4636825561523437, + "step": 309720 + }, + { + "epoch": 0.8431740846604056, + "grad_norm": 2.300999879837036, + "learning_rate": 3.1408477579392126e-06, + "loss": 2.455809783935547, + "step": 309740 + }, + { + "epoch": 0.8432285286511502, + "grad_norm": 2.101208448410034, + "learning_rate": 3.139757394030258e-06, + "loss": 2.475550651550293, + "step": 309760 + }, + { + "epoch": 0.8432829726418947, + "grad_norm": 2.1911165714263916, + "learning_rate": 3.1386670301213033e-06, + "loss": 2.453682518005371, + "step": 309780 + }, + { + "epoch": 0.8433374166326392, + "grad_norm": 2.096806764602661, + "learning_rate": 3.1375766662123487e-06, + "loss": 2.430566596984863, + "step": 309800 + }, + { + "epoch": 0.8433918606233837, + "grad_norm": 2.239189386367798, + "learning_rate": 3.136486302303394e-06, + "loss": 2.4551723480224608, + "step": 309820 + }, + { + "epoch": 0.8434463046141282, + "grad_norm": 2.1849849224090576, + "learning_rate": 3.1353959383944395e-06, + "loss": 2.476499557495117, + "step": 309840 + }, + { + "epoch": 0.8435007486048728, + "grad_norm": 2.1732072830200195, + "learning_rate": 3.134305574485485e-06, + "loss": 2.479359436035156, + "step": 309860 + }, + { + "epoch": 0.8435551925956173, + "grad_norm": 2.235578775405884, + "learning_rate": 3.1332152105765303e-06, + "loss": 2.4921512603759766, + "step": 309880 + }, + { + "epoch": 0.8436096365863618, + "grad_norm": 2.9108479022979736, + "learning_rate": 3.1321248466675756e-06, + "loss": 2.442944145202637, + "step": 309900 + }, + { + "epoch": 0.8436640805771063, + "grad_norm": 2.211273431777954, + "learning_rate": 3.131034482758621e-06, + "loss": 2.472854423522949, + "step": 309920 + }, + { + "epoch": 0.8437185245678508, + "grad_norm": 2.1648435592651367, + "learning_rate": 3.1299441188496664e-06, + "loss": 2.468985939025879, + "step": 309940 + }, + { + "epoch": 0.8437729685585953, + "grad_norm": 2.226635217666626, + "learning_rate": 3.128853754940712e-06, + "loss": 2.4656463623046876, + "step": 309960 + }, + { + "epoch": 0.8438274125493399, + "grad_norm": 2.1820578575134277, + "learning_rate": 3.127763391031757e-06, + "loss": 2.4493398666381836, + "step": 309980 + }, + { + "epoch": 0.8438818565400844, + "grad_norm": 2.378418207168579, + "learning_rate": 3.1266730271228026e-06, + "loss": 2.4672130584716796, + "step": 310000 + }, + { + "epoch": 0.8439363005308289, + "grad_norm": 2.217729330062866, + "learning_rate": 3.125582663213848e-06, + "loss": 2.504856491088867, + "step": 310020 + }, + { + "epoch": 0.8439907445215734, + "grad_norm": 2.31693434715271, + "learning_rate": 3.1244922993048933e-06, + "loss": 2.473748779296875, + "step": 310040 + }, + { + "epoch": 0.8440451885123179, + "grad_norm": 2.1995913982391357, + "learning_rate": 3.1234019353959387e-06, + "loss": 2.443367767333984, + "step": 310060 + }, + { + "epoch": 0.8440996325030625, + "grad_norm": 2.181100845336914, + "learning_rate": 3.122311571486984e-06, + "loss": 2.46557559967041, + "step": 310080 + }, + { + "epoch": 0.844154076493807, + "grad_norm": 2.1769731044769287, + "learning_rate": 3.1212212075780295e-06, + "loss": 2.4504669189453123, + "step": 310100 + }, + { + "epoch": 0.8442085204845515, + "grad_norm": 2.225799083709717, + "learning_rate": 3.120130843669075e-06, + "loss": 2.4738958358764647, + "step": 310120 + }, + { + "epoch": 0.844262964475296, + "grad_norm": 2.2209222316741943, + "learning_rate": 3.1190404797601203e-06, + "loss": 2.4654361724853517, + "step": 310140 + }, + { + "epoch": 0.8443174084660405, + "grad_norm": 2.2344748973846436, + "learning_rate": 3.1179501158511657e-06, + "loss": 2.4355276107788084, + "step": 310160 + }, + { + "epoch": 0.844371852456785, + "grad_norm": 2.129765272140503, + "learning_rate": 3.116859751942211e-06, + "loss": 2.4611783981323243, + "step": 310180 + }, + { + "epoch": 0.8444262964475296, + "grad_norm": 2.1837918758392334, + "learning_rate": 3.1157693880332564e-06, + "loss": 2.4528223037719727, + "step": 310200 + }, + { + "epoch": 0.8444807404382741, + "grad_norm": 2.1150293350219727, + "learning_rate": 3.114679024124302e-06, + "loss": 2.4635856628417967, + "step": 310220 + }, + { + "epoch": 0.8445351844290186, + "grad_norm": 2.218179702758789, + "learning_rate": 3.113588660215347e-06, + "loss": 2.463981056213379, + "step": 310240 + }, + { + "epoch": 0.8445896284197631, + "grad_norm": 2.1360957622528076, + "learning_rate": 3.1124982963063926e-06, + "loss": 2.44417667388916, + "step": 310260 + }, + { + "epoch": 0.8446440724105077, + "grad_norm": 2.1049959659576416, + "learning_rate": 3.111407932397438e-06, + "loss": 2.4476646423339843, + "step": 310280 + }, + { + "epoch": 0.8446985164012523, + "grad_norm": 2.180095672607422, + "learning_rate": 3.1103175684884834e-06, + "loss": 2.4437734603881838, + "step": 310300 + }, + { + "epoch": 0.8447529603919968, + "grad_norm": 2.2476446628570557, + "learning_rate": 3.1092272045795287e-06, + "loss": 2.4764430999755858, + "step": 310320 + }, + { + "epoch": 0.8448074043827413, + "grad_norm": 2.2436254024505615, + "learning_rate": 3.108136840670574e-06, + "loss": 2.489554595947266, + "step": 310340 + }, + { + "epoch": 0.8448618483734858, + "grad_norm": 2.2435123920440674, + "learning_rate": 3.107046476761619e-06, + "loss": 2.4656843185424804, + "step": 310360 + }, + { + "epoch": 0.8449162923642303, + "grad_norm": 2.2124686241149902, + "learning_rate": 3.105956112852665e-06, + "loss": 2.430064010620117, + "step": 310380 + }, + { + "epoch": 0.8449707363549748, + "grad_norm": 2.156125545501709, + "learning_rate": 3.10486574894371e-06, + "loss": 2.461280632019043, + "step": 310400 + }, + { + "epoch": 0.8450251803457194, + "grad_norm": 2.0570731163024902, + "learning_rate": 3.1037753850347557e-06, + "loss": 2.444899559020996, + "step": 310420 + }, + { + "epoch": 0.8450796243364639, + "grad_norm": 2.2333874702453613, + "learning_rate": 3.1026850211258006e-06, + "loss": 2.4553504943847657, + "step": 310440 + }, + { + "epoch": 0.8451340683272084, + "grad_norm": 2.116737127304077, + "learning_rate": 3.1015946572168464e-06, + "loss": 2.4766124725341796, + "step": 310460 + }, + { + "epoch": 0.8451885123179529, + "grad_norm": 2.114071846008301, + "learning_rate": 3.1005042933078914e-06, + "loss": 2.459075164794922, + "step": 310480 + }, + { + "epoch": 0.8452429563086974, + "grad_norm": 2.2859296798706055, + "learning_rate": 3.0994139293989372e-06, + "loss": 2.4622320175170898, + "step": 310500 + }, + { + "epoch": 0.845297400299442, + "grad_norm": 2.125248670578003, + "learning_rate": 3.098323565489982e-06, + "loss": 2.4391254425048827, + "step": 310520 + }, + { + "epoch": 0.8453518442901865, + "grad_norm": 2.2317965030670166, + "learning_rate": 3.097233201581028e-06, + "loss": 2.4691919326782226, + "step": 310540 + }, + { + "epoch": 0.845406288280931, + "grad_norm": 2.238107204437256, + "learning_rate": 3.096142837672073e-06, + "loss": 2.420863914489746, + "step": 310560 + }, + { + "epoch": 0.8454607322716755, + "grad_norm": 2.1435651779174805, + "learning_rate": 3.0950524737631188e-06, + "loss": 2.451018524169922, + "step": 310580 + }, + { + "epoch": 0.84551517626242, + "grad_norm": 2.0620193481445312, + "learning_rate": 3.0939621098541637e-06, + "loss": 2.491366958618164, + "step": 310600 + }, + { + "epoch": 0.8455696202531645, + "grad_norm": 2.1586880683898926, + "learning_rate": 3.0928717459452095e-06, + "loss": 2.4331167221069334, + "step": 310620 + }, + { + "epoch": 0.8456240642439091, + "grad_norm": 2.2584939002990723, + "learning_rate": 3.0917813820362545e-06, + "loss": 2.4798641204833984, + "step": 310640 + }, + { + "epoch": 0.8456785082346536, + "grad_norm": 2.229311466217041, + "learning_rate": 3.0906910181273003e-06, + "loss": 2.4631128311157227, + "step": 310660 + }, + { + "epoch": 0.8457329522253981, + "grad_norm": 2.1333389282226562, + "learning_rate": 3.0896006542183453e-06, + "loss": 2.4353992462158205, + "step": 310680 + }, + { + "epoch": 0.8457873962161426, + "grad_norm": 2.1510539054870605, + "learning_rate": 3.088510290309391e-06, + "loss": 2.4496702194213866, + "step": 310700 + }, + { + "epoch": 0.8458418402068871, + "grad_norm": 2.132209062576294, + "learning_rate": 3.087419926400436e-06, + "loss": 2.4353302001953123, + "step": 310720 + }, + { + "epoch": 0.8458962841976316, + "grad_norm": 2.159945011138916, + "learning_rate": 3.086329562491482e-06, + "loss": 2.4422992706298827, + "step": 310740 + }, + { + "epoch": 0.8459507281883762, + "grad_norm": 2.111948251724243, + "learning_rate": 3.085239198582527e-06, + "loss": 2.4420732498168944, + "step": 310760 + }, + { + "epoch": 0.8460051721791207, + "grad_norm": 2.2069480419158936, + "learning_rate": 3.0841488346735726e-06, + "loss": 2.421742630004883, + "step": 310780 + }, + { + "epoch": 0.8460596161698652, + "grad_norm": 2.156297445297241, + "learning_rate": 3.0830584707646176e-06, + "loss": 2.480872917175293, + "step": 310800 + }, + { + "epoch": 0.8461140601606097, + "grad_norm": 2.1168315410614014, + "learning_rate": 3.0819681068556634e-06, + "loss": 2.461793899536133, + "step": 310820 + }, + { + "epoch": 0.8461685041513543, + "grad_norm": 3.215242385864258, + "learning_rate": 3.0808777429467083e-06, + "loss": 2.4384700775146486, + "step": 310840 + }, + { + "epoch": 0.8462229481420989, + "grad_norm": 2.118260145187378, + "learning_rate": 3.079787379037754e-06, + "loss": 2.485757255554199, + "step": 310860 + }, + { + "epoch": 0.8462773921328434, + "grad_norm": 2.192326068878174, + "learning_rate": 3.078697015128799e-06, + "loss": 2.5215330123901367, + "step": 310880 + }, + { + "epoch": 0.8463318361235879, + "grad_norm": 2.2348098754882812, + "learning_rate": 3.077606651219845e-06, + "loss": 2.4695817947387697, + "step": 310900 + }, + { + "epoch": 0.8463862801143324, + "grad_norm": 3.775548219680786, + "learning_rate": 3.07651628731089e-06, + "loss": 2.4428714752197265, + "step": 310920 + }, + { + "epoch": 0.8464407241050769, + "grad_norm": 2.165458917617798, + "learning_rate": 3.0754259234019357e-06, + "loss": 2.472163200378418, + "step": 310940 + }, + { + "epoch": 0.8464951680958214, + "grad_norm": 2.0664992332458496, + "learning_rate": 3.0743355594929807e-06, + "loss": 2.4557931900024412, + "step": 310960 + }, + { + "epoch": 0.846549612086566, + "grad_norm": 2.1996443271636963, + "learning_rate": 3.0732451955840265e-06, + "loss": 2.447515678405762, + "step": 310980 + }, + { + "epoch": 0.8466040560773105, + "grad_norm": 2.243807554244995, + "learning_rate": 3.0721548316750714e-06, + "loss": 2.452012062072754, + "step": 311000 + }, + { + "epoch": 0.846658500068055, + "grad_norm": 2.2488694190979004, + "learning_rate": 3.0710644677661172e-06, + "loss": 2.458872413635254, + "step": 311020 + }, + { + "epoch": 0.8467129440587995, + "grad_norm": 2.1734695434570312, + "learning_rate": 3.069974103857162e-06, + "loss": 2.4785943984985352, + "step": 311040 + }, + { + "epoch": 0.846767388049544, + "grad_norm": 2.1721315383911133, + "learning_rate": 3.068883739948208e-06, + "loss": 2.439459228515625, + "step": 311060 + }, + { + "epoch": 0.8468218320402886, + "grad_norm": 2.222952127456665, + "learning_rate": 3.067793376039253e-06, + "loss": 2.4794538497924803, + "step": 311080 + }, + { + "epoch": 0.8468762760310331, + "grad_norm": 2.207152843475342, + "learning_rate": 3.0667030121302988e-06, + "loss": 2.4767932891845703, + "step": 311100 + }, + { + "epoch": 0.8469307200217776, + "grad_norm": 2.0972325801849365, + "learning_rate": 3.0656126482213437e-06, + "loss": 2.456637382507324, + "step": 311120 + }, + { + "epoch": 0.8469851640125221, + "grad_norm": 2.136852741241455, + "learning_rate": 3.0645222843123896e-06, + "loss": 2.482778549194336, + "step": 311140 + }, + { + "epoch": 0.8470396080032666, + "grad_norm": 2.122149705886841, + "learning_rate": 3.0634319204034345e-06, + "loss": 2.456224250793457, + "step": 311160 + }, + { + "epoch": 0.8470940519940111, + "grad_norm": 2.1973140239715576, + "learning_rate": 3.0623415564944803e-06, + "loss": 2.4916913986206053, + "step": 311180 + }, + { + "epoch": 0.8471484959847557, + "grad_norm": 2.132355213165283, + "learning_rate": 3.061251192585526e-06, + "loss": 2.4724979400634766, + "step": 311200 + }, + { + "epoch": 0.8472029399755002, + "grad_norm": 2.3389580249786377, + "learning_rate": 3.060160828676571e-06, + "loss": 2.478450393676758, + "step": 311220 + }, + { + "epoch": 0.8472573839662447, + "grad_norm": 2.1700496673583984, + "learning_rate": 3.059070464767617e-06, + "loss": 2.4786474227905275, + "step": 311240 + }, + { + "epoch": 0.8473118279569892, + "grad_norm": 2.1844656467437744, + "learning_rate": 3.057980100858662e-06, + "loss": 2.4490318298339844, + "step": 311260 + }, + { + "epoch": 0.8473662719477337, + "grad_norm": 2.412119150161743, + "learning_rate": 3.0568897369497077e-06, + "loss": 2.453999137878418, + "step": 311280 + }, + { + "epoch": 0.8474207159384783, + "grad_norm": 2.1303515434265137, + "learning_rate": 3.0557993730407526e-06, + "loss": 2.4369441986083986, + "step": 311300 + }, + { + "epoch": 0.8474751599292228, + "grad_norm": 2.117560625076294, + "learning_rate": 3.0547090091317985e-06, + "loss": 2.4603261947631836, + "step": 311320 + }, + { + "epoch": 0.8475296039199673, + "grad_norm": 2.2011656761169434, + "learning_rate": 3.0536186452228434e-06, + "loss": 2.465738296508789, + "step": 311340 + }, + { + "epoch": 0.8475840479107118, + "grad_norm": 2.158132314682007, + "learning_rate": 3.052528281313889e-06, + "loss": 2.46982479095459, + "step": 311360 + }, + { + "epoch": 0.8476384919014563, + "grad_norm": 2.1420557498931885, + "learning_rate": 3.051437917404934e-06, + "loss": 2.4548923492431642, + "step": 311380 + }, + { + "epoch": 0.8476929358922008, + "grad_norm": 2.165956735610962, + "learning_rate": 3.0503475534959796e-06, + "loss": 2.4535507202148437, + "step": 311400 + }, + { + "epoch": 0.8477473798829455, + "grad_norm": 2.223855495452881, + "learning_rate": 3.049257189587025e-06, + "loss": 2.483488082885742, + "step": 311420 + }, + { + "epoch": 0.84780182387369, + "grad_norm": 2.146869659423828, + "learning_rate": 3.0481668256780703e-06, + "loss": 2.4464149475097656, + "step": 311440 + }, + { + "epoch": 0.8478562678644345, + "grad_norm": 2.4267091751098633, + "learning_rate": 3.0470764617691157e-06, + "loss": 2.47824764251709, + "step": 311460 + }, + { + "epoch": 0.847910711855179, + "grad_norm": 2.0248334407806396, + "learning_rate": 3.045986097860161e-06, + "loss": 2.465378189086914, + "step": 311480 + }, + { + "epoch": 0.8479651558459235, + "grad_norm": 3.264507532119751, + "learning_rate": 3.0448957339512065e-06, + "loss": 2.4490400314331056, + "step": 311500 + }, + { + "epoch": 0.8480195998366681, + "grad_norm": 2.206585168838501, + "learning_rate": 3.043805370042252e-06, + "loss": 2.454502487182617, + "step": 311520 + }, + { + "epoch": 0.8480740438274126, + "grad_norm": 2.113612651824951, + "learning_rate": 3.0427150061332973e-06, + "loss": 2.4719757080078124, + "step": 311540 + }, + { + "epoch": 0.8481284878181571, + "grad_norm": 2.311467170715332, + "learning_rate": 3.0416246422243427e-06, + "loss": 2.4628274917602537, + "step": 311560 + }, + { + "epoch": 0.8481829318089016, + "grad_norm": 2.3968546390533447, + "learning_rate": 3.040534278315388e-06, + "loss": 2.4713993072509766, + "step": 311580 + }, + { + "epoch": 0.8482373757996461, + "grad_norm": 2.154656410217285, + "learning_rate": 3.0394439144064334e-06, + "loss": 2.474111557006836, + "step": 311600 + }, + { + "epoch": 0.8482918197903906, + "grad_norm": 2.2190473079681396, + "learning_rate": 3.038353550497479e-06, + "loss": 2.442338562011719, + "step": 311620 + }, + { + "epoch": 0.8483462637811352, + "grad_norm": 2.577594041824341, + "learning_rate": 3.037263186588524e-06, + "loss": 2.455849838256836, + "step": 311640 + }, + { + "epoch": 0.8484007077718797, + "grad_norm": 2.192030906677246, + "learning_rate": 3.0361728226795696e-06, + "loss": 2.450428771972656, + "step": 311660 + }, + { + "epoch": 0.8484551517626242, + "grad_norm": 2.084263324737549, + "learning_rate": 3.035082458770615e-06, + "loss": 2.468338394165039, + "step": 311680 + }, + { + "epoch": 0.8485095957533687, + "grad_norm": 2.386582136154175, + "learning_rate": 3.0339920948616604e-06, + "loss": 2.4649726867675783, + "step": 311700 + }, + { + "epoch": 0.8485640397441132, + "grad_norm": 2.243454694747925, + "learning_rate": 3.0329017309527057e-06, + "loss": 2.4524551391601563, + "step": 311720 + }, + { + "epoch": 0.8486184837348578, + "grad_norm": 2.232956647872925, + "learning_rate": 3.031811367043751e-06, + "loss": 2.471108245849609, + "step": 311740 + }, + { + "epoch": 0.8486729277256023, + "grad_norm": 2.1780307292938232, + "learning_rate": 3.0307210031347965e-06, + "loss": 2.4462631225585936, + "step": 311760 + }, + { + "epoch": 0.8487273717163468, + "grad_norm": 2.2346699237823486, + "learning_rate": 3.029630639225842e-06, + "loss": 2.481822967529297, + "step": 311780 + }, + { + "epoch": 0.8487818157070913, + "grad_norm": 2.2764525413513184, + "learning_rate": 3.0285402753168873e-06, + "loss": 2.457745361328125, + "step": 311800 + }, + { + "epoch": 0.8488362596978358, + "grad_norm": 2.119126796722412, + "learning_rate": 3.0274499114079327e-06, + "loss": 2.4637990951538087, + "step": 311820 + }, + { + "epoch": 0.8488907036885803, + "grad_norm": 2.165076732635498, + "learning_rate": 3.026359547498978e-06, + "loss": 2.433483695983887, + "step": 311840 + }, + { + "epoch": 0.8489451476793249, + "grad_norm": 2.1468279361724854, + "learning_rate": 3.0252691835900234e-06, + "loss": 2.476772689819336, + "step": 311860 + }, + { + "epoch": 0.8489995916700694, + "grad_norm": 2.0896809101104736, + "learning_rate": 3.024178819681069e-06, + "loss": 2.4741897583007812, + "step": 311880 + }, + { + "epoch": 0.8490540356608139, + "grad_norm": 2.3422603607177734, + "learning_rate": 3.0230884557721142e-06, + "loss": 2.4779266357421874, + "step": 311900 + }, + { + "epoch": 0.8491084796515584, + "grad_norm": 2.092068910598755, + "learning_rate": 3.0219980918631596e-06, + "loss": 2.438765525817871, + "step": 311920 + }, + { + "epoch": 0.849162923642303, + "grad_norm": 2.221945285797119, + "learning_rate": 3.020907727954205e-06, + "loss": 2.477712059020996, + "step": 311940 + }, + { + "epoch": 0.8492173676330474, + "grad_norm": 2.1419646739959717, + "learning_rate": 3.0198173640452504e-06, + "loss": 2.4532838821411134, + "step": 311960 + }, + { + "epoch": 0.8492718116237921, + "grad_norm": 2.2233800888061523, + "learning_rate": 3.0187270001362958e-06, + "loss": 2.464187812805176, + "step": 311980 + }, + { + "epoch": 0.8493262556145366, + "grad_norm": 2.1723859310150146, + "learning_rate": 3.017636636227341e-06, + "loss": 2.4788333892822267, + "step": 312000 + }, + { + "epoch": 0.8493806996052811, + "grad_norm": 2.1697726249694824, + "learning_rate": 3.0165462723183865e-06, + "loss": 2.4459667205810547, + "step": 312020 + }, + { + "epoch": 0.8494351435960256, + "grad_norm": 2.373213291168213, + "learning_rate": 3.015455908409432e-06, + "loss": 2.447671890258789, + "step": 312040 + }, + { + "epoch": 0.8494895875867701, + "grad_norm": 2.2334887981414795, + "learning_rate": 3.0143655445004773e-06, + "loss": 2.42330322265625, + "step": 312060 + }, + { + "epoch": 0.8495440315775147, + "grad_norm": 2.128275156021118, + "learning_rate": 3.0132751805915227e-06, + "loss": 2.4710784912109376, + "step": 312080 + }, + { + "epoch": 0.8495984755682592, + "grad_norm": 2.186189889907837, + "learning_rate": 3.012184816682568e-06, + "loss": 2.485123634338379, + "step": 312100 + }, + { + "epoch": 0.8496529195590037, + "grad_norm": 2.116152048110962, + "learning_rate": 3.0110944527736135e-06, + "loss": 2.489293670654297, + "step": 312120 + }, + { + "epoch": 0.8497073635497482, + "grad_norm": 2.1966161727905273, + "learning_rate": 3.010004088864659e-06, + "loss": 2.447100067138672, + "step": 312140 + }, + { + "epoch": 0.8497618075404927, + "grad_norm": 2.204806089401245, + "learning_rate": 3.0089137249557042e-06, + "loss": 2.4693721771240233, + "step": 312160 + }, + { + "epoch": 0.8498162515312372, + "grad_norm": 2.166883707046509, + "learning_rate": 3.0078233610467496e-06, + "loss": 2.4389617919921873, + "step": 312180 + }, + { + "epoch": 0.8498706955219818, + "grad_norm": 3.7196943759918213, + "learning_rate": 3.006732997137795e-06, + "loss": 2.4771921157836916, + "step": 312200 + }, + { + "epoch": 0.8499251395127263, + "grad_norm": 2.1885650157928467, + "learning_rate": 3.0056426332288404e-06, + "loss": 2.4898956298828123, + "step": 312220 + }, + { + "epoch": 0.8499795835034708, + "grad_norm": 2.114668130874634, + "learning_rate": 3.0045522693198858e-06, + "loss": 2.4502231597900392, + "step": 312240 + }, + { + "epoch": 0.8500340274942153, + "grad_norm": 2.237694263458252, + "learning_rate": 3.003461905410931e-06, + "loss": 2.449079132080078, + "step": 312260 + }, + { + "epoch": 0.8500884714849598, + "grad_norm": 2.208566188812256, + "learning_rate": 3.0023715415019765e-06, + "loss": 2.4795345306396483, + "step": 312280 + }, + { + "epoch": 0.8501429154757044, + "grad_norm": 2.17065691947937, + "learning_rate": 3.001281177593022e-06, + "loss": 2.4425567626953124, + "step": 312300 + }, + { + "epoch": 0.8501973594664489, + "grad_norm": 2.085301637649536, + "learning_rate": 3.0001908136840673e-06, + "loss": 2.4249292373657227, + "step": 312320 + }, + { + "epoch": 0.8502518034571934, + "grad_norm": 2.174900770187378, + "learning_rate": 2.9991004497751127e-06, + "loss": 2.4660923004150392, + "step": 312340 + }, + { + "epoch": 0.8503062474479379, + "grad_norm": 2.131312608718872, + "learning_rate": 2.998010085866158e-06, + "loss": 2.4573396682739257, + "step": 312360 + }, + { + "epoch": 0.8503606914386824, + "grad_norm": 2.1253867149353027, + "learning_rate": 2.9969197219572035e-06, + "loss": 2.4496240615844727, + "step": 312380 + }, + { + "epoch": 0.8504151354294269, + "grad_norm": 2.098051071166992, + "learning_rate": 2.995829358048249e-06, + "loss": 2.455041694641113, + "step": 312400 + }, + { + "epoch": 0.8504695794201715, + "grad_norm": 2.2032623291015625, + "learning_rate": 2.9947389941392942e-06, + "loss": 2.4521865844726562, + "step": 312420 + }, + { + "epoch": 0.850524023410916, + "grad_norm": 2.157118797302246, + "learning_rate": 2.9936486302303396e-06, + "loss": 2.463439178466797, + "step": 312440 + }, + { + "epoch": 0.8505784674016605, + "grad_norm": 2.118434190750122, + "learning_rate": 2.9925582663213846e-06, + "loss": 2.4536359786987303, + "step": 312460 + }, + { + "epoch": 0.850632911392405, + "grad_norm": 2.217179298400879, + "learning_rate": 2.9914679024124304e-06, + "loss": 2.4708698272705076, + "step": 312480 + }, + { + "epoch": 0.8506873553831495, + "grad_norm": 2.191066026687622, + "learning_rate": 2.9903775385034754e-06, + "loss": 2.4567893981933593, + "step": 312500 + }, + { + "epoch": 0.8507417993738942, + "grad_norm": 2.297384262084961, + "learning_rate": 2.989287174594521e-06, + "loss": 2.462121772766113, + "step": 312520 + }, + { + "epoch": 0.8507962433646387, + "grad_norm": 2.1335995197296143, + "learning_rate": 2.988196810685566e-06, + "loss": 2.4782993316650392, + "step": 312540 + }, + { + "epoch": 0.8508506873553832, + "grad_norm": 2.265035629272461, + "learning_rate": 2.987106446776612e-06, + "loss": 2.428846740722656, + "step": 312560 + }, + { + "epoch": 0.8509051313461277, + "grad_norm": 2.235579013824463, + "learning_rate": 2.986016082867657e-06, + "loss": 2.43856258392334, + "step": 312580 + }, + { + "epoch": 0.8509595753368722, + "grad_norm": 2.0919816493988037, + "learning_rate": 2.9849257189587027e-06, + "loss": 2.483297920227051, + "step": 312600 + }, + { + "epoch": 0.8510140193276167, + "grad_norm": 2.2607624530792236, + "learning_rate": 2.9838353550497477e-06, + "loss": 2.4483970642089843, + "step": 312620 + }, + { + "epoch": 0.8510684633183613, + "grad_norm": 2.1429803371429443, + "learning_rate": 2.9827449911407935e-06, + "loss": 2.4640199661254885, + "step": 312640 + }, + { + "epoch": 0.8511229073091058, + "grad_norm": 2.1433587074279785, + "learning_rate": 2.9816546272318384e-06, + "loss": 2.4890325546264647, + "step": 312660 + }, + { + "epoch": 0.8511773512998503, + "grad_norm": 2.165120840072632, + "learning_rate": 2.9805642633228843e-06, + "loss": 2.4635576248168944, + "step": 312680 + }, + { + "epoch": 0.8512317952905948, + "grad_norm": 2.12033748626709, + "learning_rate": 2.9794738994139292e-06, + "loss": 2.4638511657714846, + "step": 312700 + }, + { + "epoch": 0.8512862392813393, + "grad_norm": 2.5107262134552, + "learning_rate": 2.978383535504975e-06, + "loss": 2.4628143310546875, + "step": 312720 + }, + { + "epoch": 0.8513406832720839, + "grad_norm": 2.094013214111328, + "learning_rate": 2.97729317159602e-06, + "loss": 2.465468406677246, + "step": 312740 + }, + { + "epoch": 0.8513951272628284, + "grad_norm": 2.248645782470703, + "learning_rate": 2.976202807687066e-06, + "loss": 2.4368419647216797, + "step": 312760 + }, + { + "epoch": 0.8514495712535729, + "grad_norm": 2.2685508728027344, + "learning_rate": 2.9751124437781108e-06, + "loss": 2.432665252685547, + "step": 312780 + }, + { + "epoch": 0.8515040152443174, + "grad_norm": 2.1230390071868896, + "learning_rate": 2.9740220798691566e-06, + "loss": 2.4783514022827147, + "step": 312800 + }, + { + "epoch": 0.8515584592350619, + "grad_norm": 2.0705764293670654, + "learning_rate": 2.9729317159602015e-06, + "loss": 2.4803308486938476, + "step": 312820 + }, + { + "epoch": 0.8516129032258064, + "grad_norm": 2.266949415206909, + "learning_rate": 2.9718413520512473e-06, + "loss": 2.4376943588256834, + "step": 312840 + }, + { + "epoch": 0.851667347216551, + "grad_norm": 2.1635968685150146, + "learning_rate": 2.9707509881422923e-06, + "loss": 2.458274269104004, + "step": 312860 + }, + { + "epoch": 0.8517217912072955, + "grad_norm": 2.2012786865234375, + "learning_rate": 2.969660624233338e-06, + "loss": 2.452184867858887, + "step": 312880 + }, + { + "epoch": 0.85177623519804, + "grad_norm": 2.1234445571899414, + "learning_rate": 2.968570260324384e-06, + "loss": 2.451400375366211, + "step": 312900 + }, + { + "epoch": 0.8518306791887845, + "grad_norm": 2.125652551651001, + "learning_rate": 2.967479896415429e-06, + "loss": 2.453238868713379, + "step": 312920 + }, + { + "epoch": 0.851885123179529, + "grad_norm": 2.1790122985839844, + "learning_rate": 2.9663895325064747e-06, + "loss": 2.489188003540039, + "step": 312940 + }, + { + "epoch": 0.8519395671702735, + "grad_norm": 2.156851291656494, + "learning_rate": 2.9652991685975197e-06, + "loss": 2.45584659576416, + "step": 312960 + }, + { + "epoch": 0.8519940111610181, + "grad_norm": 2.1010098457336426, + "learning_rate": 2.9642088046885655e-06, + "loss": 2.46651611328125, + "step": 312980 + }, + { + "epoch": 0.8520484551517626, + "grad_norm": 2.150221824645996, + "learning_rate": 2.9631184407796104e-06, + "loss": 2.424361228942871, + "step": 313000 + }, + { + "epoch": 0.8521028991425071, + "grad_norm": 2.159344434738159, + "learning_rate": 2.9620280768706562e-06, + "loss": 2.440184783935547, + "step": 313020 + }, + { + "epoch": 0.8521573431332516, + "grad_norm": 2.0853285789489746, + "learning_rate": 2.960937712961701e-06, + "loss": 2.4356382369995115, + "step": 313040 + }, + { + "epoch": 0.8522117871239961, + "grad_norm": 2.111982583999634, + "learning_rate": 2.959847349052747e-06, + "loss": 2.4935455322265625, + "step": 313060 + }, + { + "epoch": 0.8522662311147408, + "grad_norm": 2.175476312637329, + "learning_rate": 2.958756985143792e-06, + "loss": 2.437669372558594, + "step": 313080 + }, + { + "epoch": 0.8523206751054853, + "grad_norm": 2.1962475776672363, + "learning_rate": 2.9576666212348378e-06, + "loss": 2.452184295654297, + "step": 313100 + }, + { + "epoch": 0.8523751190962298, + "grad_norm": 2.2436532974243164, + "learning_rate": 2.9565762573258827e-06, + "loss": 2.464410400390625, + "step": 313120 + }, + { + "epoch": 0.8524295630869743, + "grad_norm": 2.4105091094970703, + "learning_rate": 2.9554858934169286e-06, + "loss": 2.448053741455078, + "step": 313140 + }, + { + "epoch": 0.8524840070777188, + "grad_norm": 2.4628937244415283, + "learning_rate": 2.9543955295079735e-06, + "loss": 2.4197988510131836, + "step": 313160 + }, + { + "epoch": 0.8525384510684633, + "grad_norm": 2.223945379257202, + "learning_rate": 2.9533051655990193e-06, + "loss": 2.454151725769043, + "step": 313180 + }, + { + "epoch": 0.8525928950592079, + "grad_norm": 2.175999402999878, + "learning_rate": 2.9522148016900643e-06, + "loss": 2.4506868362426757, + "step": 313200 + }, + { + "epoch": 0.8526473390499524, + "grad_norm": 2.182126045227051, + "learning_rate": 2.95112443778111e-06, + "loss": 2.435779571533203, + "step": 313220 + }, + { + "epoch": 0.8527017830406969, + "grad_norm": 2.1175220012664795, + "learning_rate": 2.950034073872155e-06, + "loss": 2.453575325012207, + "step": 313240 + }, + { + "epoch": 0.8527562270314414, + "grad_norm": 2.152127265930176, + "learning_rate": 2.948943709963201e-06, + "loss": 2.4625885009765627, + "step": 313260 + }, + { + "epoch": 0.8528106710221859, + "grad_norm": 2.191624164581299, + "learning_rate": 2.947853346054246e-06, + "loss": 2.4504024505615236, + "step": 313280 + }, + { + "epoch": 0.8528651150129305, + "grad_norm": 2.089552402496338, + "learning_rate": 2.9467629821452916e-06, + "loss": 2.505786323547363, + "step": 313300 + }, + { + "epoch": 0.852919559003675, + "grad_norm": 2.0808167457580566, + "learning_rate": 2.9456726182363366e-06, + "loss": 2.468325614929199, + "step": 313320 + }, + { + "epoch": 0.8529740029944195, + "grad_norm": 2.074127435684204, + "learning_rate": 2.9445822543273824e-06, + "loss": 2.4779598236083986, + "step": 313340 + }, + { + "epoch": 0.853028446985164, + "grad_norm": 2.244351387023926, + "learning_rate": 2.9434918904184274e-06, + "loss": 2.4696163177490233, + "step": 313360 + }, + { + "epoch": 0.8530828909759085, + "grad_norm": 2.187955856323242, + "learning_rate": 2.942401526509473e-06, + "loss": 2.4289934158325197, + "step": 313380 + }, + { + "epoch": 0.853137334966653, + "grad_norm": 2.1177315711975098, + "learning_rate": 2.941311162600518e-06, + "loss": 2.4567195892333986, + "step": 313400 + }, + { + "epoch": 0.8531917789573976, + "grad_norm": 2.1679399013519287, + "learning_rate": 2.9402207986915635e-06, + "loss": 2.445880126953125, + "step": 313420 + }, + { + "epoch": 0.8532462229481421, + "grad_norm": 2.274639129638672, + "learning_rate": 2.939130434782609e-06, + "loss": 2.465170478820801, + "step": 313440 + }, + { + "epoch": 0.8533006669388866, + "grad_norm": 2.182011604309082, + "learning_rate": 2.9380400708736543e-06, + "loss": 2.4756174087524414, + "step": 313460 + }, + { + "epoch": 0.8533551109296311, + "grad_norm": 2.0997474193573, + "learning_rate": 2.9369497069646997e-06, + "loss": 2.4675994873046876, + "step": 313480 + }, + { + "epoch": 0.8534095549203756, + "grad_norm": 2.17667555809021, + "learning_rate": 2.935859343055745e-06, + "loss": 2.471206855773926, + "step": 313500 + }, + { + "epoch": 0.8534639989111202, + "grad_norm": 2.287196636199951, + "learning_rate": 2.9347689791467905e-06, + "loss": 2.428497314453125, + "step": 313520 + }, + { + "epoch": 0.8535184429018647, + "grad_norm": 2.1779942512512207, + "learning_rate": 2.933678615237836e-06, + "loss": 2.4763700485229494, + "step": 313540 + }, + { + "epoch": 0.8535728868926092, + "grad_norm": 2.2456493377685547, + "learning_rate": 2.9325882513288812e-06, + "loss": 2.4888637542724608, + "step": 313560 + }, + { + "epoch": 0.8536273308833537, + "grad_norm": 2.079812526702881, + "learning_rate": 2.9314978874199266e-06, + "loss": 2.461044502258301, + "step": 313580 + }, + { + "epoch": 0.8536817748740982, + "grad_norm": 2.1968953609466553, + "learning_rate": 2.930407523510972e-06, + "loss": 2.4862302780151366, + "step": 313600 + }, + { + "epoch": 0.8537362188648427, + "grad_norm": 2.2133123874664307, + "learning_rate": 2.9293171596020174e-06, + "loss": 2.4511032104492188, + "step": 313620 + }, + { + "epoch": 0.8537906628555874, + "grad_norm": 2.418487548828125, + "learning_rate": 2.9282267956930628e-06, + "loss": 2.4415374755859376, + "step": 313640 + }, + { + "epoch": 0.8538451068463319, + "grad_norm": 2.161839723587036, + "learning_rate": 2.927136431784108e-06, + "loss": 2.4695642471313475, + "step": 313660 + }, + { + "epoch": 0.8538995508370764, + "grad_norm": 2.151266574859619, + "learning_rate": 2.9260460678751535e-06, + "loss": 2.455911064147949, + "step": 313680 + }, + { + "epoch": 0.8539539948278209, + "grad_norm": 2.2023932933807373, + "learning_rate": 2.924955703966199e-06, + "loss": 2.458487701416016, + "step": 313700 + }, + { + "epoch": 0.8540084388185654, + "grad_norm": 2.1852660179138184, + "learning_rate": 2.9238653400572443e-06, + "loss": 2.4496883392333983, + "step": 313720 + }, + { + "epoch": 0.85406288280931, + "grad_norm": 2.2092957496643066, + "learning_rate": 2.9227749761482897e-06, + "loss": 2.4879255294799805, + "step": 313740 + }, + { + "epoch": 0.8541173268000545, + "grad_norm": 2.15661358833313, + "learning_rate": 2.921684612239335e-06, + "loss": 2.4676891326904298, + "step": 313760 + }, + { + "epoch": 0.854171770790799, + "grad_norm": 2.309650421142578, + "learning_rate": 2.9205942483303805e-06, + "loss": 2.4408327102661134, + "step": 313780 + }, + { + "epoch": 0.8542262147815435, + "grad_norm": 2.1586554050445557, + "learning_rate": 2.919503884421426e-06, + "loss": 2.4433414459228517, + "step": 313800 + }, + { + "epoch": 0.854280658772288, + "grad_norm": 2.163823127746582, + "learning_rate": 2.9184135205124712e-06, + "loss": 2.424875259399414, + "step": 313820 + }, + { + "epoch": 0.8543351027630325, + "grad_norm": 2.490560531616211, + "learning_rate": 2.9173231566035166e-06, + "loss": 2.489766502380371, + "step": 313840 + }, + { + "epoch": 0.8543895467537771, + "grad_norm": 2.1345858573913574, + "learning_rate": 2.916232792694562e-06, + "loss": 2.470442771911621, + "step": 313860 + }, + { + "epoch": 0.8544439907445216, + "grad_norm": 2.180687427520752, + "learning_rate": 2.9151424287856074e-06, + "loss": 2.4233575820922852, + "step": 313880 + }, + { + "epoch": 0.8544984347352661, + "grad_norm": 2.2253172397613525, + "learning_rate": 2.9140520648766528e-06, + "loss": 2.4766157150268553, + "step": 313900 + }, + { + "epoch": 0.8545528787260106, + "grad_norm": 2.272223711013794, + "learning_rate": 2.912961700967698e-06, + "loss": 2.440506172180176, + "step": 313920 + }, + { + "epoch": 0.8546073227167551, + "grad_norm": 2.100292444229126, + "learning_rate": 2.9118713370587436e-06, + "loss": 2.4347101211547852, + "step": 313940 + }, + { + "epoch": 0.8546617667074997, + "grad_norm": 2.113360643386841, + "learning_rate": 2.910780973149789e-06, + "loss": 2.459528350830078, + "step": 313960 + }, + { + "epoch": 0.8547162106982442, + "grad_norm": 2.242711067199707, + "learning_rate": 2.9096906092408343e-06, + "loss": 2.4784507751464844, + "step": 313980 + }, + { + "epoch": 0.8547706546889887, + "grad_norm": 2.2203409671783447, + "learning_rate": 2.9086002453318797e-06, + "loss": 2.482291030883789, + "step": 314000 + }, + { + "epoch": 0.8548250986797332, + "grad_norm": 2.3263683319091797, + "learning_rate": 2.907509881422925e-06, + "loss": 2.4698577880859376, + "step": 314020 + }, + { + "epoch": 0.8548795426704777, + "grad_norm": 2.1202330589294434, + "learning_rate": 2.9064195175139705e-06, + "loss": 2.4537139892578126, + "step": 314040 + }, + { + "epoch": 0.8549339866612222, + "grad_norm": 2.131345748901367, + "learning_rate": 2.905329153605016e-06, + "loss": 2.4566423416137697, + "step": 314060 + }, + { + "epoch": 0.8549884306519668, + "grad_norm": 2.169746160507202, + "learning_rate": 2.9042387896960613e-06, + "loss": 2.4738073348999023, + "step": 314080 + }, + { + "epoch": 0.8550428746427113, + "grad_norm": 2.1355130672454834, + "learning_rate": 2.9031484257871066e-06, + "loss": 2.462982177734375, + "step": 314100 + }, + { + "epoch": 0.8550973186334558, + "grad_norm": 2.4392802715301514, + "learning_rate": 2.902058061878152e-06, + "loss": 2.4260065078735353, + "step": 314120 + }, + { + "epoch": 0.8551517626242003, + "grad_norm": 2.3389041423797607, + "learning_rate": 2.9009676979691974e-06, + "loss": 2.456889343261719, + "step": 314140 + }, + { + "epoch": 0.8552062066149448, + "grad_norm": 2.1617913246154785, + "learning_rate": 2.899877334060243e-06, + "loss": 2.445348358154297, + "step": 314160 + }, + { + "epoch": 0.8552606506056893, + "grad_norm": 2.1322696208953857, + "learning_rate": 2.898786970151288e-06, + "loss": 2.4621551513671873, + "step": 314180 + }, + { + "epoch": 0.855315094596434, + "grad_norm": 2.1604058742523193, + "learning_rate": 2.8976966062423336e-06, + "loss": 2.4686344146728514, + "step": 314200 + }, + { + "epoch": 0.8553695385871785, + "grad_norm": 2.1149425506591797, + "learning_rate": 2.896606242333379e-06, + "loss": 2.4668529510498045, + "step": 314220 + }, + { + "epoch": 0.855423982577923, + "grad_norm": 2.0053272247314453, + "learning_rate": 2.8955158784244243e-06, + "loss": 2.4634937286376952, + "step": 314240 + }, + { + "epoch": 0.8554784265686675, + "grad_norm": 2.4383883476257324, + "learning_rate": 2.8944255145154697e-06, + "loss": 2.4810325622558596, + "step": 314260 + }, + { + "epoch": 0.855532870559412, + "grad_norm": 2.2955479621887207, + "learning_rate": 2.893335150606515e-06, + "loss": 2.459117317199707, + "step": 314280 + }, + { + "epoch": 0.8555873145501566, + "grad_norm": 2.160229206085205, + "learning_rate": 2.8922447866975605e-06, + "loss": 2.4658939361572267, + "step": 314300 + }, + { + "epoch": 0.8556417585409011, + "grad_norm": 2.23634672164917, + "learning_rate": 2.891154422788606e-06, + "loss": 2.443364143371582, + "step": 314320 + }, + { + "epoch": 0.8556962025316456, + "grad_norm": 2.1372077465057373, + "learning_rate": 2.8900640588796513e-06, + "loss": 2.4552736282348633, + "step": 314340 + }, + { + "epoch": 0.8557506465223901, + "grad_norm": 2.2052273750305176, + "learning_rate": 2.8889736949706967e-06, + "loss": 2.455745887756348, + "step": 314360 + }, + { + "epoch": 0.8558050905131346, + "grad_norm": 2.146361827850342, + "learning_rate": 2.887883331061742e-06, + "loss": 2.4420372009277345, + "step": 314380 + }, + { + "epoch": 0.8558595345038791, + "grad_norm": 2.189875364303589, + "learning_rate": 2.8867929671527874e-06, + "loss": 2.4601823806762697, + "step": 314400 + }, + { + "epoch": 0.8559139784946237, + "grad_norm": 2.966458320617676, + "learning_rate": 2.885702603243833e-06, + "loss": 2.463932228088379, + "step": 314420 + }, + { + "epoch": 0.8559684224853682, + "grad_norm": 2.117102861404419, + "learning_rate": 2.884612239334878e-06, + "loss": 2.455828857421875, + "step": 314440 + }, + { + "epoch": 0.8560228664761127, + "grad_norm": 2.306384801864624, + "learning_rate": 2.8835218754259236e-06, + "loss": 2.472800636291504, + "step": 314460 + }, + { + "epoch": 0.8560773104668572, + "grad_norm": 2.2022109031677246, + "learning_rate": 2.882431511516969e-06, + "loss": 2.4456369400024416, + "step": 314480 + }, + { + "epoch": 0.8561317544576017, + "grad_norm": 2.194019317626953, + "learning_rate": 2.8813411476080144e-06, + "loss": 2.4708547592163086, + "step": 314500 + }, + { + "epoch": 0.8561861984483463, + "grad_norm": 2.1579995155334473, + "learning_rate": 2.8802507836990593e-06, + "loss": 2.4702693939208986, + "step": 314520 + }, + { + "epoch": 0.8562406424390908, + "grad_norm": 2.13850998878479, + "learning_rate": 2.879160419790105e-06, + "loss": 2.469405174255371, + "step": 314540 + }, + { + "epoch": 0.8562950864298353, + "grad_norm": 2.133603096008301, + "learning_rate": 2.87807005588115e-06, + "loss": 2.5000865936279295, + "step": 314560 + }, + { + "epoch": 0.8563495304205798, + "grad_norm": 2.120267868041992, + "learning_rate": 2.876979691972196e-06, + "loss": 2.463787078857422, + "step": 314580 + }, + { + "epoch": 0.8564039744113243, + "grad_norm": 2.1717519760131836, + "learning_rate": 2.8758893280632417e-06, + "loss": 2.4453916549682617, + "step": 314600 + }, + { + "epoch": 0.8564584184020688, + "grad_norm": 2.2347097396850586, + "learning_rate": 2.8747989641542867e-06, + "loss": 2.482539176940918, + "step": 314620 + }, + { + "epoch": 0.8565128623928134, + "grad_norm": 2.192258834838867, + "learning_rate": 2.8737086002453325e-06, + "loss": 2.461356353759766, + "step": 314640 + }, + { + "epoch": 0.8565673063835579, + "grad_norm": 2.4132943153381348, + "learning_rate": 2.8726182363363774e-06, + "loss": 2.455046844482422, + "step": 314660 + }, + { + "epoch": 0.8566217503743024, + "grad_norm": 2.345547914505005, + "learning_rate": 2.8715278724274232e-06, + "loss": 2.451700210571289, + "step": 314680 + }, + { + "epoch": 0.8566761943650469, + "grad_norm": 2.183666944503784, + "learning_rate": 2.870437508518468e-06, + "loss": 2.4666284561157226, + "step": 314700 + }, + { + "epoch": 0.8567306383557914, + "grad_norm": 2.0895209312438965, + "learning_rate": 2.869347144609514e-06, + "loss": 2.464884567260742, + "step": 314720 + }, + { + "epoch": 0.856785082346536, + "grad_norm": 2.2616360187530518, + "learning_rate": 2.868256780700559e-06, + "loss": 2.4519601821899415, + "step": 314740 + }, + { + "epoch": 0.8568395263372806, + "grad_norm": 2.1335296630859375, + "learning_rate": 2.867166416791605e-06, + "loss": 2.4564794540405273, + "step": 314760 + }, + { + "epoch": 0.8568939703280251, + "grad_norm": 2.1187024116516113, + "learning_rate": 2.8660760528826498e-06, + "loss": 2.4792758941650392, + "step": 314780 + }, + { + "epoch": 0.8569484143187696, + "grad_norm": 2.2238245010375977, + "learning_rate": 2.8649856889736956e-06, + "loss": 2.444593048095703, + "step": 314800 + }, + { + "epoch": 0.8570028583095141, + "grad_norm": 2.136206865310669, + "learning_rate": 2.8638953250647405e-06, + "loss": 2.4445522308349608, + "step": 314820 + }, + { + "epoch": 0.8570573023002586, + "grad_norm": 2.093987464904785, + "learning_rate": 2.8628049611557863e-06, + "loss": 2.432907295227051, + "step": 314840 + }, + { + "epoch": 0.8571117462910032, + "grad_norm": 2.2317311763763428, + "learning_rate": 2.8617145972468313e-06, + "loss": 2.459058380126953, + "step": 314860 + }, + { + "epoch": 0.8571661902817477, + "grad_norm": 2.167358875274658, + "learning_rate": 2.860624233337877e-06, + "loss": 2.4518054962158202, + "step": 314880 + }, + { + "epoch": 0.8572206342724922, + "grad_norm": 2.127546548843384, + "learning_rate": 2.859533869428922e-06, + "loss": 2.45472412109375, + "step": 314900 + }, + { + "epoch": 0.8572750782632367, + "grad_norm": 2.0742101669311523, + "learning_rate": 2.858443505519968e-06, + "loss": 2.468873977661133, + "step": 314920 + }, + { + "epoch": 0.8573295222539812, + "grad_norm": 2.106065273284912, + "learning_rate": 2.857353141611013e-06, + "loss": 2.4661237716674806, + "step": 314940 + }, + { + "epoch": 0.8573839662447258, + "grad_norm": 2.0866763591766357, + "learning_rate": 2.8562627777020586e-06, + "loss": 2.442196846008301, + "step": 314960 + }, + { + "epoch": 0.8574384102354703, + "grad_norm": 2.1613104343414307, + "learning_rate": 2.8551724137931036e-06, + "loss": 2.4724517822265626, + "step": 314980 + }, + { + "epoch": 0.8574928542262148, + "grad_norm": 2.283703565597534, + "learning_rate": 2.8540820498841494e-06, + "loss": 2.4754661560058593, + "step": 315000 + }, + { + "epoch": 0.8575472982169593, + "grad_norm": 2.355896234512329, + "learning_rate": 2.8529916859751944e-06, + "loss": 2.443320465087891, + "step": 315020 + }, + { + "epoch": 0.8576017422077038, + "grad_norm": 2.2025864124298096, + "learning_rate": 2.85190132206624e-06, + "loss": 2.4716346740722654, + "step": 315040 + }, + { + "epoch": 0.8576561861984483, + "grad_norm": 2.317406415939331, + "learning_rate": 2.850810958157285e-06, + "loss": 2.4581674575805663, + "step": 315060 + }, + { + "epoch": 0.8577106301891929, + "grad_norm": 2.3515498638153076, + "learning_rate": 2.849720594248331e-06, + "loss": 2.480997848510742, + "step": 315080 + }, + { + "epoch": 0.8577650741799374, + "grad_norm": 2.4185051918029785, + "learning_rate": 2.848630230339376e-06, + "loss": 2.441114807128906, + "step": 315100 + }, + { + "epoch": 0.8578195181706819, + "grad_norm": 2.355015754699707, + "learning_rate": 2.8475398664304217e-06, + "loss": 2.4462055206298827, + "step": 315120 + }, + { + "epoch": 0.8578739621614264, + "grad_norm": 2.1470675468444824, + "learning_rate": 2.8464495025214667e-06, + "loss": 2.4414350509643556, + "step": 315140 + }, + { + "epoch": 0.8579284061521709, + "grad_norm": 2.1830623149871826, + "learning_rate": 2.8453591386125125e-06, + "loss": 2.4422021865844727, + "step": 315160 + }, + { + "epoch": 0.8579828501429154, + "grad_norm": 2.1261329650878906, + "learning_rate": 2.8442687747035575e-06, + "loss": 2.4646543502807616, + "step": 315180 + }, + { + "epoch": 0.85803729413366, + "grad_norm": 2.122422456741333, + "learning_rate": 2.8431784107946033e-06, + "loss": 2.4853355407714846, + "step": 315200 + }, + { + "epoch": 0.8580917381244045, + "grad_norm": 4.6821160316467285, + "learning_rate": 2.8420880468856482e-06, + "loss": 2.4645999908447265, + "step": 315220 + }, + { + "epoch": 0.858146182115149, + "grad_norm": 2.1483492851257324, + "learning_rate": 2.840997682976694e-06, + "loss": 2.4357940673828127, + "step": 315240 + }, + { + "epoch": 0.8582006261058935, + "grad_norm": 2.2937214374542236, + "learning_rate": 2.839907319067739e-06, + "loss": 2.4544307708740236, + "step": 315260 + }, + { + "epoch": 0.858255070096638, + "grad_norm": 3.812721014022827, + "learning_rate": 2.838816955158785e-06, + "loss": 2.469078254699707, + "step": 315280 + }, + { + "epoch": 0.8583095140873827, + "grad_norm": 2.155679702758789, + "learning_rate": 2.8377265912498298e-06, + "loss": 2.488087844848633, + "step": 315300 + }, + { + "epoch": 0.8583639580781272, + "grad_norm": 2.1472084522247314, + "learning_rate": 2.8366362273408756e-06, + "loss": 2.4945741653442384, + "step": 315320 + }, + { + "epoch": 0.8584184020688717, + "grad_norm": 2.2458574771881104, + "learning_rate": 2.8355458634319206e-06, + "loss": 2.4340587615966798, + "step": 315340 + }, + { + "epoch": 0.8584728460596162, + "grad_norm": 2.212789535522461, + "learning_rate": 2.8344554995229664e-06, + "loss": 2.4741937637329103, + "step": 315360 + }, + { + "epoch": 0.8585272900503607, + "grad_norm": 2.7119526863098145, + "learning_rate": 2.8333651356140113e-06, + "loss": 2.4605424880981444, + "step": 315380 + }, + { + "epoch": 0.8585817340411052, + "grad_norm": 2.38373064994812, + "learning_rate": 2.832274771705057e-06, + "loss": 2.494608497619629, + "step": 315400 + }, + { + "epoch": 0.8586361780318498, + "grad_norm": 2.212939500808716, + "learning_rate": 2.831184407796102e-06, + "loss": 2.480524253845215, + "step": 315420 + }, + { + "epoch": 0.8586906220225943, + "grad_norm": 2.2446377277374268, + "learning_rate": 2.830094043887148e-06, + "loss": 2.4666072845458986, + "step": 315440 + }, + { + "epoch": 0.8587450660133388, + "grad_norm": 3.8683605194091797, + "learning_rate": 2.829003679978193e-06, + "loss": 2.476669502258301, + "step": 315460 + }, + { + "epoch": 0.8587995100040833, + "grad_norm": 2.076054811477661, + "learning_rate": 2.8279133160692387e-06, + "loss": 2.4251270294189453, + "step": 315480 + }, + { + "epoch": 0.8588539539948278, + "grad_norm": 2.2326104640960693, + "learning_rate": 2.8268229521602836e-06, + "loss": 2.460312843322754, + "step": 315500 + }, + { + "epoch": 0.8589083979855724, + "grad_norm": 2.2316179275512695, + "learning_rate": 2.825732588251329e-06, + "loss": 2.4887868881225588, + "step": 315520 + }, + { + "epoch": 0.8589628419763169, + "grad_norm": 2.1640610694885254, + "learning_rate": 2.8246422243423744e-06, + "loss": 2.452700614929199, + "step": 315540 + }, + { + "epoch": 0.8590172859670614, + "grad_norm": 2.1593658924102783, + "learning_rate": 2.82355186043342e-06, + "loss": 2.4505756378173826, + "step": 315560 + }, + { + "epoch": 0.8590717299578059, + "grad_norm": 2.2356374263763428, + "learning_rate": 2.822461496524465e-06, + "loss": 2.466439056396484, + "step": 315580 + }, + { + "epoch": 0.8591261739485504, + "grad_norm": 2.285173177719116, + "learning_rate": 2.8213711326155106e-06, + "loss": 2.4850297927856446, + "step": 315600 + }, + { + "epoch": 0.8591806179392949, + "grad_norm": 2.3045027256011963, + "learning_rate": 2.820280768706556e-06, + "loss": 2.4387325286865233, + "step": 315620 + }, + { + "epoch": 0.8592350619300395, + "grad_norm": 2.130070447921753, + "learning_rate": 2.8191904047976013e-06, + "loss": 2.4564445495605467, + "step": 315640 + }, + { + "epoch": 0.859289505920784, + "grad_norm": 2.1794095039367676, + "learning_rate": 2.8181000408886467e-06, + "loss": 2.4776792526245117, + "step": 315660 + }, + { + "epoch": 0.8593439499115285, + "grad_norm": 2.2312660217285156, + "learning_rate": 2.817009676979692e-06, + "loss": 2.4293777465820314, + "step": 315680 + }, + { + "epoch": 0.859398393902273, + "grad_norm": 2.2068910598754883, + "learning_rate": 2.8159193130707375e-06, + "loss": 2.4262811660766603, + "step": 315700 + }, + { + "epoch": 0.8594528378930175, + "grad_norm": 2.3709161281585693, + "learning_rate": 2.814828949161783e-06, + "loss": 2.470372200012207, + "step": 315720 + }, + { + "epoch": 0.8595072818837621, + "grad_norm": 2.1338841915130615, + "learning_rate": 2.8137385852528283e-06, + "loss": 2.4552404403686525, + "step": 315740 + }, + { + "epoch": 0.8595617258745066, + "grad_norm": 2.6588621139526367, + "learning_rate": 2.8126482213438737e-06, + "loss": 2.4564456939697266, + "step": 315760 + }, + { + "epoch": 0.8596161698652511, + "grad_norm": 2.1289894580841064, + "learning_rate": 2.811557857434919e-06, + "loss": 2.4552743911743162, + "step": 315780 + }, + { + "epoch": 0.8596706138559956, + "grad_norm": 2.3611180782318115, + "learning_rate": 2.8104674935259644e-06, + "loss": 2.4678955078125, + "step": 315800 + }, + { + "epoch": 0.8597250578467401, + "grad_norm": 2.205160140991211, + "learning_rate": 2.80937712961701e-06, + "loss": 2.423232650756836, + "step": 315820 + }, + { + "epoch": 0.8597795018374846, + "grad_norm": 2.228558301925659, + "learning_rate": 2.808286765708055e-06, + "loss": 2.440159034729004, + "step": 315840 + }, + { + "epoch": 0.8598339458282293, + "grad_norm": 2.1228387355804443, + "learning_rate": 2.8071964017991006e-06, + "loss": 2.464446449279785, + "step": 315860 + }, + { + "epoch": 0.8598883898189738, + "grad_norm": 2.2251474857330322, + "learning_rate": 2.806106037890146e-06, + "loss": 2.4616062164306642, + "step": 315880 + }, + { + "epoch": 0.8599428338097183, + "grad_norm": 2.2319910526275635, + "learning_rate": 2.8050156739811914e-06, + "loss": 2.4586009979248047, + "step": 315900 + }, + { + "epoch": 0.8599972778004628, + "grad_norm": 2.2260935306549072, + "learning_rate": 2.8039253100722367e-06, + "loss": 2.4669395446777345, + "step": 315920 + }, + { + "epoch": 0.8600517217912073, + "grad_norm": 2.2417943477630615, + "learning_rate": 2.802834946163282e-06, + "loss": 2.460641288757324, + "step": 315940 + }, + { + "epoch": 0.8601061657819519, + "grad_norm": 3.7827606201171875, + "learning_rate": 2.8017445822543275e-06, + "loss": 2.4558876037597654, + "step": 315960 + }, + { + "epoch": 0.8601606097726964, + "grad_norm": 2.432931423187256, + "learning_rate": 2.800654218345373e-06, + "loss": 2.4828632354736326, + "step": 315980 + }, + { + "epoch": 0.8602150537634409, + "grad_norm": 2.1722850799560547, + "learning_rate": 2.7995638544364183e-06, + "loss": 2.439310836791992, + "step": 316000 + }, + { + "epoch": 0.8602694977541854, + "grad_norm": 2.2823050022125244, + "learning_rate": 2.7984734905274637e-06, + "loss": 2.436667251586914, + "step": 316020 + }, + { + "epoch": 0.8603239417449299, + "grad_norm": 2.6879191398620605, + "learning_rate": 2.797383126618509e-06, + "loss": 2.459165573120117, + "step": 316040 + }, + { + "epoch": 0.8603783857356744, + "grad_norm": 2.129837989807129, + "learning_rate": 2.7962927627095544e-06, + "loss": 2.445771408081055, + "step": 316060 + }, + { + "epoch": 0.860432829726419, + "grad_norm": 2.1141867637634277, + "learning_rate": 2.7952023988006e-06, + "loss": 2.4753068923950194, + "step": 316080 + }, + { + "epoch": 0.8604872737171635, + "grad_norm": 2.0484259128570557, + "learning_rate": 2.794112034891645e-06, + "loss": 2.459038734436035, + "step": 316100 + }, + { + "epoch": 0.860541717707908, + "grad_norm": 2.1293158531188965, + "learning_rate": 2.7930216709826906e-06, + "loss": 2.464015579223633, + "step": 316120 + }, + { + "epoch": 0.8605961616986525, + "grad_norm": 2.19146728515625, + "learning_rate": 2.791931307073736e-06, + "loss": 2.4632339477539062, + "step": 316140 + }, + { + "epoch": 0.860650605689397, + "grad_norm": 2.329801321029663, + "learning_rate": 2.7908409431647814e-06, + "loss": 2.485910415649414, + "step": 316160 + }, + { + "epoch": 0.8607050496801416, + "grad_norm": 2.222123384475708, + "learning_rate": 2.7897505792558268e-06, + "loss": 2.4476650238037108, + "step": 316180 + }, + { + "epoch": 0.8607594936708861, + "grad_norm": 2.2425615787506104, + "learning_rate": 2.788660215346872e-06, + "loss": 2.479160499572754, + "step": 316200 + }, + { + "epoch": 0.8608139376616306, + "grad_norm": 2.2324461936950684, + "learning_rate": 2.7875698514379175e-06, + "loss": 2.46849365234375, + "step": 316220 + }, + { + "epoch": 0.8608683816523751, + "grad_norm": 2.259443998336792, + "learning_rate": 2.786479487528963e-06, + "loss": 2.4620378494262694, + "step": 316240 + }, + { + "epoch": 0.8609228256431196, + "grad_norm": 2.2003464698791504, + "learning_rate": 2.7853891236200083e-06, + "loss": 2.4630914688110352, + "step": 316260 + }, + { + "epoch": 0.8609772696338641, + "grad_norm": 2.1095261573791504, + "learning_rate": 2.7842987597110537e-06, + "loss": 2.465712547302246, + "step": 316280 + }, + { + "epoch": 0.8610317136246087, + "grad_norm": 2.2044215202331543, + "learning_rate": 2.7832083958020995e-06, + "loss": 2.4579387664794923, + "step": 316300 + }, + { + "epoch": 0.8610861576153532, + "grad_norm": 2.179614782333374, + "learning_rate": 2.7821180318931445e-06, + "loss": 2.439103126525879, + "step": 316320 + }, + { + "epoch": 0.8611406016060977, + "grad_norm": 2.172931671142578, + "learning_rate": 2.7810276679841903e-06, + "loss": 2.470634651184082, + "step": 316340 + }, + { + "epoch": 0.8611950455968422, + "grad_norm": 2.1731863021850586, + "learning_rate": 2.7799373040752352e-06, + "loss": 2.4600292205810548, + "step": 316360 + }, + { + "epoch": 0.8612494895875867, + "grad_norm": 2.182324171066284, + "learning_rate": 2.778846940166281e-06, + "loss": 2.4803096771240236, + "step": 316380 + }, + { + "epoch": 0.8613039335783312, + "grad_norm": 2.2456870079040527, + "learning_rate": 2.777756576257326e-06, + "loss": 2.4599666595458984, + "step": 316400 + }, + { + "epoch": 0.8613583775690759, + "grad_norm": 2.2460856437683105, + "learning_rate": 2.776666212348372e-06, + "loss": 2.4553062438964846, + "step": 316420 + }, + { + "epoch": 0.8614128215598204, + "grad_norm": 2.252270460128784, + "learning_rate": 2.7755758484394168e-06, + "loss": 2.441234588623047, + "step": 316440 + }, + { + "epoch": 0.8614672655505649, + "grad_norm": 2.2030909061431885, + "learning_rate": 2.7744854845304626e-06, + "loss": 2.4674564361572267, + "step": 316460 + }, + { + "epoch": 0.8615217095413094, + "grad_norm": 2.271202564239502, + "learning_rate": 2.7733951206215075e-06, + "loss": 2.471293067932129, + "step": 316480 + }, + { + "epoch": 0.8615761535320539, + "grad_norm": 2.27593994140625, + "learning_rate": 2.7723047567125533e-06, + "loss": 2.452200698852539, + "step": 316500 + }, + { + "epoch": 0.8616305975227985, + "grad_norm": 2.3145995140075684, + "learning_rate": 2.7712143928035983e-06, + "loss": 2.4588979721069335, + "step": 316520 + }, + { + "epoch": 0.861685041513543, + "grad_norm": 2.1603031158447266, + "learning_rate": 2.770124028894644e-06, + "loss": 2.472795104980469, + "step": 316540 + }, + { + "epoch": 0.8617394855042875, + "grad_norm": 2.3418569564819336, + "learning_rate": 2.769033664985689e-06, + "loss": 2.422628402709961, + "step": 316560 + }, + { + "epoch": 0.861793929495032, + "grad_norm": 2.156092643737793, + "learning_rate": 2.767943301076735e-06, + "loss": 2.4620513916015625, + "step": 316580 + }, + { + "epoch": 0.8618483734857765, + "grad_norm": 2.1473989486694336, + "learning_rate": 2.76685293716778e-06, + "loss": 2.4955520629882812, + "step": 316600 + }, + { + "epoch": 0.861902817476521, + "grad_norm": 2.2103424072265625, + "learning_rate": 2.7657625732588257e-06, + "loss": 2.454710578918457, + "step": 316620 + }, + { + "epoch": 0.8619572614672656, + "grad_norm": 2.2555296421051025, + "learning_rate": 2.7646722093498706e-06, + "loss": 2.454916000366211, + "step": 316640 + }, + { + "epoch": 0.8620117054580101, + "grad_norm": 2.120286464691162, + "learning_rate": 2.7635818454409164e-06, + "loss": 2.4586326599121096, + "step": 316660 + }, + { + "epoch": 0.8620661494487546, + "grad_norm": 2.1429765224456787, + "learning_rate": 2.7624914815319614e-06, + "loss": 2.4394552230834963, + "step": 316680 + }, + { + "epoch": 0.8621205934394991, + "grad_norm": 2.2177276611328125, + "learning_rate": 2.761401117623007e-06, + "loss": 2.4559724807739256, + "step": 316700 + }, + { + "epoch": 0.8621750374302436, + "grad_norm": 2.139721155166626, + "learning_rate": 2.760310753714052e-06, + "loss": 2.4804254531860352, + "step": 316720 + }, + { + "epoch": 0.8622294814209882, + "grad_norm": 2.2098708152770996, + "learning_rate": 2.759220389805098e-06, + "loss": 2.459665298461914, + "step": 316740 + }, + { + "epoch": 0.8622839254117327, + "grad_norm": 2.2570269107818604, + "learning_rate": 2.758130025896143e-06, + "loss": 2.4575336456298826, + "step": 316760 + }, + { + "epoch": 0.8623383694024772, + "grad_norm": 2.490790367126465, + "learning_rate": 2.7570396619871887e-06, + "loss": 2.4563737869262696, + "step": 316780 + }, + { + "epoch": 0.8623928133932217, + "grad_norm": 2.1358633041381836, + "learning_rate": 2.7559492980782337e-06, + "loss": 2.4284034729003907, + "step": 316800 + }, + { + "epoch": 0.8624472573839662, + "grad_norm": 2.2782411575317383, + "learning_rate": 2.7548589341692795e-06, + "loss": 2.466057014465332, + "step": 316820 + }, + { + "epoch": 0.8625017013747107, + "grad_norm": 2.147547483444214, + "learning_rate": 2.7537685702603245e-06, + "loss": 2.454916763305664, + "step": 316840 + }, + { + "epoch": 0.8625561453654553, + "grad_norm": 2.258552312850952, + "learning_rate": 2.7526782063513703e-06, + "loss": 2.4221523284912108, + "step": 316860 + }, + { + "epoch": 0.8626105893561998, + "grad_norm": 2.049060106277466, + "learning_rate": 2.7515878424424153e-06, + "loss": 2.466473197937012, + "step": 316880 + }, + { + "epoch": 0.8626650333469443, + "grad_norm": 2.1104891300201416, + "learning_rate": 2.750497478533461e-06, + "loss": 2.459389877319336, + "step": 316900 + }, + { + "epoch": 0.8627194773376888, + "grad_norm": 2.3037946224212646, + "learning_rate": 2.749407114624506e-06, + "loss": 2.460658073425293, + "step": 316920 + }, + { + "epoch": 0.8627739213284333, + "grad_norm": 2.325394868850708, + "learning_rate": 2.748316750715552e-06, + "loss": 2.4418523788452147, + "step": 316940 + }, + { + "epoch": 0.862828365319178, + "grad_norm": 2.1673147678375244, + "learning_rate": 2.747226386806597e-06, + "loss": 2.487133026123047, + "step": 316960 + }, + { + "epoch": 0.8628828093099224, + "grad_norm": 2.2462282180786133, + "learning_rate": 2.7461360228976426e-06, + "loss": 2.4557876586914062, + "step": 316980 + }, + { + "epoch": 0.862937253300667, + "grad_norm": 2.1917858123779297, + "learning_rate": 2.7450456589886876e-06, + "loss": 2.4553466796875, + "step": 317000 + }, + { + "epoch": 0.8629916972914115, + "grad_norm": 2.42779278755188, + "learning_rate": 2.7439552950797334e-06, + "loss": 2.4610620498657227, + "step": 317020 + }, + { + "epoch": 0.863046141282156, + "grad_norm": 2.1274349689483643, + "learning_rate": 2.7428649311707783e-06, + "loss": 2.438853073120117, + "step": 317040 + }, + { + "epoch": 0.8631005852729005, + "grad_norm": 2.0987510681152344, + "learning_rate": 2.741774567261824e-06, + "loss": 2.4363418579101563, + "step": 317060 + }, + { + "epoch": 0.8631550292636451, + "grad_norm": 2.3670623302459717, + "learning_rate": 2.740684203352869e-06, + "loss": 2.438869285583496, + "step": 317080 + }, + { + "epoch": 0.8632094732543896, + "grad_norm": 2.1208033561706543, + "learning_rate": 2.739593839443915e-06, + "loss": 2.456068992614746, + "step": 317100 + }, + { + "epoch": 0.8632639172451341, + "grad_norm": 2.1700873374938965, + "learning_rate": 2.73850347553496e-06, + "loss": 2.4443937301635743, + "step": 317120 + }, + { + "epoch": 0.8633183612358786, + "grad_norm": 2.2524287700653076, + "learning_rate": 2.7374131116260057e-06, + "loss": 2.4974273681640624, + "step": 317140 + }, + { + "epoch": 0.8633728052266231, + "grad_norm": 2.1819956302642822, + "learning_rate": 2.7363227477170507e-06, + "loss": 2.448808288574219, + "step": 317160 + }, + { + "epoch": 0.8634272492173677, + "grad_norm": 2.158203601837158, + "learning_rate": 2.7352323838080965e-06, + "loss": 2.468483734130859, + "step": 317180 + }, + { + "epoch": 0.8634816932081122, + "grad_norm": 2.2085769176483154, + "learning_rate": 2.7341420198991414e-06, + "loss": 2.4618621826171876, + "step": 317200 + }, + { + "epoch": 0.8635361371988567, + "grad_norm": 2.353342294692993, + "learning_rate": 2.7330516559901872e-06, + "loss": 2.4817359924316404, + "step": 317220 + }, + { + "epoch": 0.8635905811896012, + "grad_norm": 2.1743667125701904, + "learning_rate": 2.731961292081232e-06, + "loss": 2.4480922698974608, + "step": 317240 + }, + { + "epoch": 0.8636450251803457, + "grad_norm": 2.132906675338745, + "learning_rate": 2.730870928172278e-06, + "loss": 2.4462833404541016, + "step": 317260 + }, + { + "epoch": 0.8636994691710902, + "grad_norm": 2.090588092803955, + "learning_rate": 2.729780564263323e-06, + "loss": 2.4272228240966798, + "step": 317280 + }, + { + "epoch": 0.8637539131618348, + "grad_norm": 2.1014764308929443, + "learning_rate": 2.7286902003543688e-06, + "loss": 2.4603912353515627, + "step": 317300 + }, + { + "epoch": 0.8638083571525793, + "grad_norm": 2.0813469886779785, + "learning_rate": 2.7275998364454137e-06, + "loss": 2.4494001388549806, + "step": 317320 + }, + { + "epoch": 0.8638628011433238, + "grad_norm": 2.141667127609253, + "learning_rate": 2.7265094725364595e-06, + "loss": 2.4491195678710938, + "step": 317340 + }, + { + "epoch": 0.8639172451340683, + "grad_norm": 2.161940336227417, + "learning_rate": 2.7254191086275045e-06, + "loss": 2.4836044311523438, + "step": 317360 + }, + { + "epoch": 0.8639716891248128, + "grad_norm": 2.172940254211426, + "learning_rate": 2.7243287447185503e-06, + "loss": 2.447826957702637, + "step": 317380 + }, + { + "epoch": 0.8640261331155573, + "grad_norm": 2.192715644836426, + "learning_rate": 2.7232383808095953e-06, + "loss": 2.4728633880615236, + "step": 317400 + }, + { + "epoch": 0.8640805771063019, + "grad_norm": 2.2029500007629395, + "learning_rate": 2.722148016900641e-06, + "loss": 2.474819374084473, + "step": 317420 + }, + { + "epoch": 0.8641350210970464, + "grad_norm": 2.163341999053955, + "learning_rate": 2.721057652991686e-06, + "loss": 2.44796085357666, + "step": 317440 + }, + { + "epoch": 0.8641894650877909, + "grad_norm": 2.109614133834839, + "learning_rate": 2.719967289082732e-06, + "loss": 2.481281280517578, + "step": 317460 + }, + { + "epoch": 0.8642439090785354, + "grad_norm": 2.1650638580322266, + "learning_rate": 2.718876925173777e-06, + "loss": 2.4335987091064455, + "step": 317480 + }, + { + "epoch": 0.8642983530692799, + "grad_norm": 2.107105255126953, + "learning_rate": 2.7177865612648226e-06, + "loss": 2.455829048156738, + "step": 317500 + }, + { + "epoch": 0.8643527970600245, + "grad_norm": 2.415745496749878, + "learning_rate": 2.7166961973558676e-06, + "loss": 2.4634246826171875, + "step": 317520 + }, + { + "epoch": 0.864407241050769, + "grad_norm": 2.047970771789551, + "learning_rate": 2.7156058334469134e-06, + "loss": 2.4474233627319335, + "step": 317540 + }, + { + "epoch": 0.8644616850415136, + "grad_norm": 2.27022123336792, + "learning_rate": 2.7145154695379584e-06, + "loss": 2.453880500793457, + "step": 317560 + }, + { + "epoch": 0.864516129032258, + "grad_norm": 2.2654829025268555, + "learning_rate": 2.713425105629004e-06, + "loss": 2.447931098937988, + "step": 317580 + }, + { + "epoch": 0.8645705730230026, + "grad_norm": 2.1292412281036377, + "learning_rate": 2.712334741720049e-06, + "loss": 2.4374505996704103, + "step": 317600 + }, + { + "epoch": 0.864625017013747, + "grad_norm": 2.168879985809326, + "learning_rate": 2.7112443778110945e-06, + "loss": 2.463462257385254, + "step": 317620 + }, + { + "epoch": 0.8646794610044917, + "grad_norm": 2.1398301124572754, + "learning_rate": 2.71015401390214e-06, + "loss": 2.463790702819824, + "step": 317640 + }, + { + "epoch": 0.8647339049952362, + "grad_norm": 2.4346275329589844, + "learning_rate": 2.7090636499931853e-06, + "loss": 2.453133964538574, + "step": 317660 + }, + { + "epoch": 0.8647883489859807, + "grad_norm": 2.168562650680542, + "learning_rate": 2.7079732860842307e-06, + "loss": 2.446704864501953, + "step": 317680 + }, + { + "epoch": 0.8648427929767252, + "grad_norm": 2.1586813926696777, + "learning_rate": 2.706882922175276e-06, + "loss": 2.453260803222656, + "step": 317700 + }, + { + "epoch": 0.8648972369674697, + "grad_norm": 2.2760493755340576, + "learning_rate": 2.7057925582663215e-06, + "loss": 2.441365051269531, + "step": 317720 + }, + { + "epoch": 0.8649516809582143, + "grad_norm": 2.35034441947937, + "learning_rate": 2.704702194357367e-06, + "loss": 2.4823928833007813, + "step": 317740 + }, + { + "epoch": 0.8650061249489588, + "grad_norm": 2.1748926639556885, + "learning_rate": 2.7036118304484122e-06, + "loss": 2.4353824615478517, + "step": 317760 + }, + { + "epoch": 0.8650605689397033, + "grad_norm": 2.2169904708862305, + "learning_rate": 2.7025214665394576e-06, + "loss": 2.496116828918457, + "step": 317780 + }, + { + "epoch": 0.8651150129304478, + "grad_norm": 2.203223466873169, + "learning_rate": 2.701431102630503e-06, + "loss": 2.4848514556884767, + "step": 317800 + }, + { + "epoch": 0.8651694569211923, + "grad_norm": 2.2092409133911133, + "learning_rate": 2.7003407387215484e-06, + "loss": 2.465234565734863, + "step": 317820 + }, + { + "epoch": 0.8652239009119368, + "grad_norm": 2.213635206222534, + "learning_rate": 2.6992503748125938e-06, + "loss": 2.4467632293701174, + "step": 317840 + }, + { + "epoch": 0.8652783449026814, + "grad_norm": 2.1957881450653076, + "learning_rate": 2.698160010903639e-06, + "loss": 2.455009651184082, + "step": 317860 + }, + { + "epoch": 0.8653327888934259, + "grad_norm": 2.207207679748535, + "learning_rate": 2.6970696469946845e-06, + "loss": 2.4255144119262697, + "step": 317880 + }, + { + "epoch": 0.8653872328841704, + "grad_norm": 2.203946113586426, + "learning_rate": 2.69597928308573e-06, + "loss": 2.4340124130249023, + "step": 317900 + }, + { + "epoch": 0.8654416768749149, + "grad_norm": 2.181990385055542, + "learning_rate": 2.6948889191767753e-06, + "loss": 2.478020668029785, + "step": 317920 + }, + { + "epoch": 0.8654961208656594, + "grad_norm": 2.1900687217712402, + "learning_rate": 2.6937985552678207e-06, + "loss": 2.4551837921142576, + "step": 317940 + }, + { + "epoch": 0.865550564856404, + "grad_norm": 2.3041746616363525, + "learning_rate": 2.6927081913588665e-06, + "loss": 2.4924665451049806, + "step": 317960 + }, + { + "epoch": 0.8656050088471485, + "grad_norm": 2.189619779586792, + "learning_rate": 2.6916178274499115e-06, + "loss": 2.4987585067749025, + "step": 317980 + }, + { + "epoch": 0.865659452837893, + "grad_norm": 2.1081714630126953, + "learning_rate": 2.6905274635409573e-06, + "loss": 2.475162887573242, + "step": 318000 + }, + { + "epoch": 0.8657138968286375, + "grad_norm": 2.2206947803497314, + "learning_rate": 2.6894370996320022e-06, + "loss": 2.4431806564331056, + "step": 318020 + }, + { + "epoch": 0.865768340819382, + "grad_norm": 2.1160707473754883, + "learning_rate": 2.688346735723048e-06, + "loss": 2.471898651123047, + "step": 318040 + }, + { + "epoch": 0.8658227848101265, + "grad_norm": 2.2372469902038574, + "learning_rate": 2.687256371814093e-06, + "loss": 2.4603376388549805, + "step": 318060 + }, + { + "epoch": 0.8658772288008711, + "grad_norm": 2.2727763652801514, + "learning_rate": 2.686166007905139e-06, + "loss": 2.4548404693603514, + "step": 318080 + }, + { + "epoch": 0.8659316727916156, + "grad_norm": 2.0985169410705566, + "learning_rate": 2.6850756439961838e-06, + "loss": 2.4972387313842774, + "step": 318100 + }, + { + "epoch": 0.8659861167823601, + "grad_norm": 2.175276756286621, + "learning_rate": 2.6839852800872296e-06, + "loss": 2.438019561767578, + "step": 318120 + }, + { + "epoch": 0.8660405607731047, + "grad_norm": 2.258202075958252, + "learning_rate": 2.6828949161782746e-06, + "loss": 2.4226310729980467, + "step": 318140 + }, + { + "epoch": 0.8660950047638492, + "grad_norm": 2.1645212173461914, + "learning_rate": 2.6818045522693204e-06, + "loss": 2.4062191009521485, + "step": 318160 + }, + { + "epoch": 0.8661494487545938, + "grad_norm": 2.2126200199127197, + "learning_rate": 2.6807141883603653e-06, + "loss": 2.4818769454956056, + "step": 318180 + }, + { + "epoch": 0.8662038927453383, + "grad_norm": 2.1044812202453613, + "learning_rate": 2.679623824451411e-06, + "loss": 2.4421054840087892, + "step": 318200 + }, + { + "epoch": 0.8662583367360828, + "grad_norm": 2.213916301727295, + "learning_rate": 2.678533460542456e-06, + "loss": 2.464706230163574, + "step": 318220 + }, + { + "epoch": 0.8663127807268273, + "grad_norm": 2.205289602279663, + "learning_rate": 2.677443096633502e-06, + "loss": 2.4245473861694338, + "step": 318240 + }, + { + "epoch": 0.8663672247175718, + "grad_norm": 2.175266981124878, + "learning_rate": 2.676352732724547e-06, + "loss": 2.4262554168701174, + "step": 318260 + }, + { + "epoch": 0.8664216687083163, + "grad_norm": 2.214271068572998, + "learning_rate": 2.6752623688155927e-06, + "loss": 2.4618682861328125, + "step": 318280 + }, + { + "epoch": 0.8664761126990609, + "grad_norm": 2.2175071239471436, + "learning_rate": 2.6741720049066376e-06, + "loss": 2.444815444946289, + "step": 318300 + }, + { + "epoch": 0.8665305566898054, + "grad_norm": 2.1730635166168213, + "learning_rate": 2.6730816409976834e-06, + "loss": 2.460593414306641, + "step": 318320 + }, + { + "epoch": 0.8665850006805499, + "grad_norm": 2.2270874977111816, + "learning_rate": 2.6719912770887284e-06, + "loss": 2.4811782836914062, + "step": 318340 + }, + { + "epoch": 0.8666394446712944, + "grad_norm": 2.3167428970336914, + "learning_rate": 2.6709009131797742e-06, + "loss": 2.442706489562988, + "step": 318360 + }, + { + "epoch": 0.8666938886620389, + "grad_norm": 2.0947484970092773, + "learning_rate": 2.669810549270819e-06, + "loss": 2.455239105224609, + "step": 318380 + }, + { + "epoch": 0.8667483326527835, + "grad_norm": 2.128800630569458, + "learning_rate": 2.668720185361865e-06, + "loss": 2.446066665649414, + "step": 318400 + }, + { + "epoch": 0.866802776643528, + "grad_norm": 2.1431026458740234, + "learning_rate": 2.66762982145291e-06, + "loss": 2.4870458602905274, + "step": 318420 + }, + { + "epoch": 0.8668572206342725, + "grad_norm": 2.187032461166382, + "learning_rate": 2.6665394575439558e-06, + "loss": 2.4685300827026366, + "step": 318440 + }, + { + "epoch": 0.866911664625017, + "grad_norm": 2.19012713432312, + "learning_rate": 2.6654490936350007e-06, + "loss": 2.4369287490844727, + "step": 318460 + }, + { + "epoch": 0.8669661086157615, + "grad_norm": 2.166083574295044, + "learning_rate": 2.6643587297260465e-06, + "loss": 2.4519979476928713, + "step": 318480 + }, + { + "epoch": 0.867020552606506, + "grad_norm": 2.203056812286377, + "learning_rate": 2.6632683658170915e-06, + "loss": 2.4584049224853515, + "step": 318500 + }, + { + "epoch": 0.8670749965972506, + "grad_norm": 2.155975341796875, + "learning_rate": 2.6621780019081373e-06, + "loss": 2.459667778015137, + "step": 318520 + }, + { + "epoch": 0.8671294405879951, + "grad_norm": 2.2190935611724854, + "learning_rate": 2.6610876379991823e-06, + "loss": 2.4416034698486326, + "step": 318540 + }, + { + "epoch": 0.8671838845787396, + "grad_norm": 2.258786201477051, + "learning_rate": 2.659997274090228e-06, + "loss": 2.4469015121459963, + "step": 318560 + }, + { + "epoch": 0.8672383285694841, + "grad_norm": 2.1631369590759277, + "learning_rate": 2.658906910181273e-06, + "loss": 2.47725772857666, + "step": 318580 + }, + { + "epoch": 0.8672927725602286, + "grad_norm": 2.1546998023986816, + "learning_rate": 2.657816546272319e-06, + "loss": 2.4615123748779295, + "step": 318600 + }, + { + "epoch": 0.8673472165509731, + "grad_norm": 2.4838504791259766, + "learning_rate": 2.656726182363364e-06, + "loss": 2.470859718322754, + "step": 318620 + }, + { + "epoch": 0.8674016605417177, + "grad_norm": 2.127459764480591, + "learning_rate": 2.6556358184544096e-06, + "loss": 2.444490432739258, + "step": 318640 + }, + { + "epoch": 0.8674561045324622, + "grad_norm": 2.11016845703125, + "learning_rate": 2.6545454545454546e-06, + "loss": 2.4933135986328123, + "step": 318660 + }, + { + "epoch": 0.8675105485232067, + "grad_norm": 2.1605942249298096, + "learning_rate": 2.6534550906365004e-06, + "loss": 2.4703496932983398, + "step": 318680 + }, + { + "epoch": 0.8675649925139512, + "grad_norm": 2.364733934402466, + "learning_rate": 2.6523647267275454e-06, + "loss": 2.4809553146362306, + "step": 318700 + }, + { + "epoch": 0.8676194365046958, + "grad_norm": 2.1368205547332764, + "learning_rate": 2.651274362818591e-06, + "loss": 2.477342414855957, + "step": 318720 + }, + { + "epoch": 0.8676738804954404, + "grad_norm": 2.166203737258911, + "learning_rate": 2.650183998909636e-06, + "loss": 2.4753559112548826, + "step": 318740 + }, + { + "epoch": 0.8677283244861849, + "grad_norm": 2.205613374710083, + "learning_rate": 2.649093635000682e-06, + "loss": 2.4831687927246096, + "step": 318760 + }, + { + "epoch": 0.8677827684769294, + "grad_norm": 2.481257915496826, + "learning_rate": 2.648003271091727e-06, + "loss": 2.465372848510742, + "step": 318780 + }, + { + "epoch": 0.8678372124676739, + "grad_norm": 2.437617301940918, + "learning_rate": 2.6469129071827727e-06, + "loss": 2.4627912521362303, + "step": 318800 + }, + { + "epoch": 0.8678916564584184, + "grad_norm": 2.0482065677642822, + "learning_rate": 2.6458225432738177e-06, + "loss": 2.4619359970092773, + "step": 318820 + }, + { + "epoch": 0.8679461004491629, + "grad_norm": 2.1022045612335205, + "learning_rate": 2.6447321793648635e-06, + "loss": 2.4721139907836913, + "step": 318840 + }, + { + "epoch": 0.8680005444399075, + "grad_norm": 2.155766248703003, + "learning_rate": 2.6436418154559084e-06, + "loss": 2.4308958053588867, + "step": 318860 + }, + { + "epoch": 0.868054988430652, + "grad_norm": 2.869142532348633, + "learning_rate": 2.6425514515469542e-06, + "loss": 2.4569770812988283, + "step": 318880 + }, + { + "epoch": 0.8681094324213965, + "grad_norm": 2.1325230598449707, + "learning_rate": 2.641461087637999e-06, + "loss": 2.424960136413574, + "step": 318900 + }, + { + "epoch": 0.868163876412141, + "grad_norm": 2.242865562438965, + "learning_rate": 2.640370723729045e-06, + "loss": 2.506148910522461, + "step": 318920 + }, + { + "epoch": 0.8682183204028855, + "grad_norm": 2.2118823528289795, + "learning_rate": 2.63928035982009e-06, + "loss": 2.483169746398926, + "step": 318940 + }, + { + "epoch": 0.8682727643936301, + "grad_norm": 2.2993829250335693, + "learning_rate": 2.638189995911136e-06, + "loss": 2.4725765228271483, + "step": 318960 + }, + { + "epoch": 0.8683272083843746, + "grad_norm": 2.6732115745544434, + "learning_rate": 2.6370996320021808e-06, + "loss": 2.481172561645508, + "step": 318980 + }, + { + "epoch": 0.8683816523751191, + "grad_norm": 2.1176998615264893, + "learning_rate": 2.6360092680932266e-06, + "loss": 2.440003204345703, + "step": 319000 + }, + { + "epoch": 0.8684360963658636, + "grad_norm": 2.2190604209899902, + "learning_rate": 2.6349189041842715e-06, + "loss": 2.438130187988281, + "step": 319020 + }, + { + "epoch": 0.8684905403566081, + "grad_norm": 2.2526912689208984, + "learning_rate": 2.6338285402753173e-06, + "loss": 2.4719823837280273, + "step": 319040 + }, + { + "epoch": 0.8685449843473526, + "grad_norm": 2.2288925647735596, + "learning_rate": 2.6327381763663623e-06, + "loss": 2.4432985305786135, + "step": 319060 + }, + { + "epoch": 0.8685994283380972, + "grad_norm": 2.2308170795440674, + "learning_rate": 2.631647812457408e-06, + "loss": 2.4556901931762694, + "step": 319080 + }, + { + "epoch": 0.8686538723288417, + "grad_norm": 2.3089616298675537, + "learning_rate": 2.630557448548453e-06, + "loss": 2.4372346878051756, + "step": 319100 + }, + { + "epoch": 0.8687083163195862, + "grad_norm": 2.1103999614715576, + "learning_rate": 2.629467084639499e-06, + "loss": 2.4255678176879885, + "step": 319120 + }, + { + "epoch": 0.8687627603103307, + "grad_norm": 2.43021821975708, + "learning_rate": 2.628376720730544e-06, + "loss": 2.49244384765625, + "step": 319140 + }, + { + "epoch": 0.8688172043010752, + "grad_norm": 2.2582156658172607, + "learning_rate": 2.6272863568215896e-06, + "loss": 2.4969738006591795, + "step": 319160 + }, + { + "epoch": 0.8688716482918198, + "grad_norm": 2.3622233867645264, + "learning_rate": 2.6261959929126346e-06, + "loss": 2.4455434799194338, + "step": 319180 + }, + { + "epoch": 0.8689260922825643, + "grad_norm": 2.051647901535034, + "learning_rate": 2.6251056290036804e-06, + "loss": 2.456389808654785, + "step": 319200 + }, + { + "epoch": 0.8689805362733088, + "grad_norm": 2.1226680278778076, + "learning_rate": 2.6240152650947254e-06, + "loss": 2.4857635498046875, + "step": 319220 + }, + { + "epoch": 0.8690349802640533, + "grad_norm": 2.312274694442749, + "learning_rate": 2.622924901185771e-06, + "loss": 2.4677501678466798, + "step": 319240 + }, + { + "epoch": 0.8690894242547978, + "grad_norm": 2.1705071926116943, + "learning_rate": 2.621834537276816e-06, + "loss": 2.4667900085449217, + "step": 319260 + }, + { + "epoch": 0.8691438682455424, + "grad_norm": 2.0225484371185303, + "learning_rate": 2.620744173367862e-06, + "loss": 2.440914535522461, + "step": 319280 + }, + { + "epoch": 0.869198312236287, + "grad_norm": 2.2116568088531494, + "learning_rate": 2.619653809458907e-06, + "loss": 2.4795095443725588, + "step": 319300 + }, + { + "epoch": 0.8692527562270315, + "grad_norm": 2.195183277130127, + "learning_rate": 2.6185634455499527e-06, + "loss": 2.4850141525268556, + "step": 319320 + }, + { + "epoch": 0.869307200217776, + "grad_norm": 2.2990729808807373, + "learning_rate": 2.6174730816409977e-06, + "loss": 2.470770263671875, + "step": 319340 + }, + { + "epoch": 0.8693616442085205, + "grad_norm": 5.261287689208984, + "learning_rate": 2.6163827177320435e-06, + "loss": 2.442361068725586, + "step": 319360 + }, + { + "epoch": 0.869416088199265, + "grad_norm": 2.1815719604492188, + "learning_rate": 2.6152923538230885e-06, + "loss": 2.456659507751465, + "step": 319380 + }, + { + "epoch": 0.8694705321900096, + "grad_norm": 2.288893461227417, + "learning_rate": 2.6142019899141343e-06, + "loss": 2.4615617752075196, + "step": 319400 + }, + { + "epoch": 0.8695249761807541, + "grad_norm": 2.2840704917907715, + "learning_rate": 2.6131116260051792e-06, + "loss": 2.4554229736328126, + "step": 319420 + }, + { + "epoch": 0.8695794201714986, + "grad_norm": 2.757744312286377, + "learning_rate": 2.612021262096225e-06, + "loss": 2.473519706726074, + "step": 319440 + }, + { + "epoch": 0.8696338641622431, + "grad_norm": 2.0916666984558105, + "learning_rate": 2.61093089818727e-06, + "loss": 2.461538314819336, + "step": 319460 + }, + { + "epoch": 0.8696883081529876, + "grad_norm": 2.141080856323242, + "learning_rate": 2.609840534278316e-06, + "loss": 2.468178367614746, + "step": 319480 + }, + { + "epoch": 0.8697427521437321, + "grad_norm": 2.1753344535827637, + "learning_rate": 2.6087501703693608e-06, + "loss": 2.470345115661621, + "step": 319500 + }, + { + "epoch": 0.8697971961344767, + "grad_norm": 2.1796178817749023, + "learning_rate": 2.6076598064604066e-06, + "loss": 2.46854305267334, + "step": 319520 + }, + { + "epoch": 0.8698516401252212, + "grad_norm": 2.1909584999084473, + "learning_rate": 2.6065694425514515e-06, + "loss": 2.4488252639770507, + "step": 319540 + }, + { + "epoch": 0.8699060841159657, + "grad_norm": 2.20074200630188, + "learning_rate": 2.6054790786424974e-06, + "loss": 2.4666574478149412, + "step": 319560 + }, + { + "epoch": 0.8699605281067102, + "grad_norm": 2.120511054992676, + "learning_rate": 2.6043887147335423e-06, + "loss": 2.429676055908203, + "step": 319580 + }, + { + "epoch": 0.8700149720974547, + "grad_norm": 2.212317943572998, + "learning_rate": 2.603298350824588e-06, + "loss": 2.4456233978271484, + "step": 319600 + }, + { + "epoch": 0.8700694160881992, + "grad_norm": 11.029317855834961, + "learning_rate": 2.602207986915633e-06, + "loss": 2.4370040893554688, + "step": 319620 + }, + { + "epoch": 0.8701238600789438, + "grad_norm": 2.1355998516082764, + "learning_rate": 2.601117623006679e-06, + "loss": 2.4603004455566406, + "step": 319640 + }, + { + "epoch": 0.8701783040696883, + "grad_norm": 2.1988704204559326, + "learning_rate": 2.6000272590977243e-06, + "loss": 2.459480857849121, + "step": 319660 + }, + { + "epoch": 0.8702327480604328, + "grad_norm": 2.074908494949341, + "learning_rate": 2.5989368951887692e-06, + "loss": 2.478759765625, + "step": 319680 + }, + { + "epoch": 0.8702871920511773, + "grad_norm": 2.170436382293701, + "learning_rate": 2.597846531279815e-06, + "loss": 2.468490982055664, + "step": 319700 + }, + { + "epoch": 0.8703416360419218, + "grad_norm": 2.235748291015625, + "learning_rate": 2.59675616737086e-06, + "loss": 2.453093719482422, + "step": 319720 + }, + { + "epoch": 0.8703960800326664, + "grad_norm": 2.3135056495666504, + "learning_rate": 2.595665803461906e-06, + "loss": 2.4692989349365235, + "step": 319740 + }, + { + "epoch": 0.8704505240234109, + "grad_norm": 2.201694965362549, + "learning_rate": 2.594575439552951e-06, + "loss": 2.466811943054199, + "step": 319760 + }, + { + "epoch": 0.8705049680141554, + "grad_norm": 2.136842966079712, + "learning_rate": 2.5934850756439966e-06, + "loss": 2.471079444885254, + "step": 319780 + }, + { + "epoch": 0.8705594120049, + "grad_norm": 2.177124261856079, + "learning_rate": 2.5923947117350416e-06, + "loss": 2.4552486419677733, + "step": 319800 + }, + { + "epoch": 0.8706138559956444, + "grad_norm": 2.2532622814178467, + "learning_rate": 2.5913043478260874e-06, + "loss": 2.492083740234375, + "step": 319820 + }, + { + "epoch": 0.870668299986389, + "grad_norm": 2.1887669563293457, + "learning_rate": 2.5902139839171323e-06, + "loss": 2.4545787811279296, + "step": 319840 + }, + { + "epoch": 0.8707227439771336, + "grad_norm": 2.2314183712005615, + "learning_rate": 2.589123620008178e-06, + "loss": 2.4541526794433595, + "step": 319860 + }, + { + "epoch": 0.8707771879678781, + "grad_norm": 2.1119751930236816, + "learning_rate": 2.588033256099223e-06, + "loss": 2.454593467712402, + "step": 319880 + }, + { + "epoch": 0.8708316319586226, + "grad_norm": 2.208070755004883, + "learning_rate": 2.586942892190269e-06, + "loss": 2.4347488403320314, + "step": 319900 + }, + { + "epoch": 0.8708860759493671, + "grad_norm": 2.2693936824798584, + "learning_rate": 2.585852528281314e-06, + "loss": 2.445761871337891, + "step": 319920 + }, + { + "epoch": 0.8709405199401116, + "grad_norm": 2.191817283630371, + "learning_rate": 2.5847621643723597e-06, + "loss": 2.4913766860961912, + "step": 319940 + }, + { + "epoch": 0.8709949639308562, + "grad_norm": 2.137413740158081, + "learning_rate": 2.5836718004634046e-06, + "loss": 2.4609432220458984, + "step": 319960 + }, + { + "epoch": 0.8710494079216007, + "grad_norm": 2.164623737335205, + "learning_rate": 2.5825814365544505e-06, + "loss": 2.479058265686035, + "step": 319980 + }, + { + "epoch": 0.8711038519123452, + "grad_norm": 2.210383176803589, + "learning_rate": 2.5814910726454954e-06, + "loss": 2.451313591003418, + "step": 320000 + }, + { + "epoch": 0.8711582959030897, + "grad_norm": 2.137683153152466, + "learning_rate": 2.5804007087365412e-06, + "loss": 2.446742820739746, + "step": 320020 + }, + { + "epoch": 0.8712127398938342, + "grad_norm": 2.1826791763305664, + "learning_rate": 2.579310344827586e-06, + "loss": 2.4694475173950194, + "step": 320040 + }, + { + "epoch": 0.8712671838845787, + "grad_norm": 2.119666814804077, + "learning_rate": 2.578219980918632e-06, + "loss": 2.4661483764648438, + "step": 320060 + }, + { + "epoch": 0.8713216278753233, + "grad_norm": 2.100310802459717, + "learning_rate": 2.577129617009677e-06, + "loss": 2.458609199523926, + "step": 320080 + }, + { + "epoch": 0.8713760718660678, + "grad_norm": 2.131622314453125, + "learning_rate": 2.5760392531007228e-06, + "loss": 2.4416973114013674, + "step": 320100 + }, + { + "epoch": 0.8714305158568123, + "grad_norm": 2.1463074684143066, + "learning_rate": 2.5749488891917677e-06, + "loss": 2.4685043334960937, + "step": 320120 + }, + { + "epoch": 0.8714849598475568, + "grad_norm": 2.214644432067871, + "learning_rate": 2.5738585252828135e-06, + "loss": 2.444478988647461, + "step": 320140 + }, + { + "epoch": 0.8715394038383013, + "grad_norm": 2.1749470233917236, + "learning_rate": 2.5727681613738585e-06, + "loss": 2.4491796493530273, + "step": 320160 + }, + { + "epoch": 0.8715938478290459, + "grad_norm": 2.1746439933776855, + "learning_rate": 2.5716777974649043e-06, + "loss": 2.4476694107055663, + "step": 320180 + }, + { + "epoch": 0.8716482918197904, + "grad_norm": 2.8787195682525635, + "learning_rate": 2.5705874335559493e-06, + "loss": 2.4521404266357423, + "step": 320200 + }, + { + "epoch": 0.8717027358105349, + "grad_norm": 2.2030088901519775, + "learning_rate": 2.569497069646995e-06, + "loss": 2.4609264373779296, + "step": 320220 + }, + { + "epoch": 0.8717571798012794, + "grad_norm": 2.1293492317199707, + "learning_rate": 2.56840670573804e-06, + "loss": 2.4534000396728515, + "step": 320240 + }, + { + "epoch": 0.8718116237920239, + "grad_norm": 2.19291615486145, + "learning_rate": 2.567316341829086e-06, + "loss": 2.4869573593139647, + "step": 320260 + }, + { + "epoch": 0.8718660677827684, + "grad_norm": 2.114595890045166, + "learning_rate": 2.566225977920131e-06, + "loss": 2.4716691970825195, + "step": 320280 + }, + { + "epoch": 0.871920511773513, + "grad_norm": 2.1440727710723877, + "learning_rate": 2.5651356140111766e-06, + "loss": 2.4827945709228514, + "step": 320300 + }, + { + "epoch": 0.8719749557642575, + "grad_norm": 2.2415177822113037, + "learning_rate": 2.5640452501022216e-06, + "loss": 2.4600868225097656, + "step": 320320 + }, + { + "epoch": 0.872029399755002, + "grad_norm": 2.1313674449920654, + "learning_rate": 2.5629548861932674e-06, + "loss": 2.4535985946655274, + "step": 320340 + }, + { + "epoch": 0.8720838437457465, + "grad_norm": 2.247192144393921, + "learning_rate": 2.5618645222843124e-06, + "loss": 2.477400016784668, + "step": 320360 + }, + { + "epoch": 0.872138287736491, + "grad_norm": 2.4423558712005615, + "learning_rate": 2.560774158375358e-06, + "loss": 2.456755828857422, + "step": 320380 + }, + { + "epoch": 0.8721927317272357, + "grad_norm": 2.2673416137695312, + "learning_rate": 2.559683794466403e-06, + "loss": 2.4788671493530274, + "step": 320400 + }, + { + "epoch": 0.8722471757179802, + "grad_norm": 2.187716484069824, + "learning_rate": 2.558593430557449e-06, + "loss": 2.4334503173828126, + "step": 320420 + }, + { + "epoch": 0.8723016197087247, + "grad_norm": 2.167587995529175, + "learning_rate": 2.557503066648494e-06, + "loss": 2.4547889709472654, + "step": 320440 + }, + { + "epoch": 0.8723560636994692, + "grad_norm": 2.166264295578003, + "learning_rate": 2.5564127027395397e-06, + "loss": 2.4495893478393556, + "step": 320460 + }, + { + "epoch": 0.8724105076902137, + "grad_norm": 2.082613706588745, + "learning_rate": 2.5553223388305847e-06, + "loss": 2.45922908782959, + "step": 320480 + }, + { + "epoch": 0.8724649516809582, + "grad_norm": 2.1116509437561035, + "learning_rate": 2.5542319749216305e-06, + "loss": 2.499855804443359, + "step": 320500 + }, + { + "epoch": 0.8725193956717028, + "grad_norm": 2.490018606185913, + "learning_rate": 2.5531416110126754e-06, + "loss": 2.4666570663452148, + "step": 320520 + }, + { + "epoch": 0.8725738396624473, + "grad_norm": 2.534747362136841, + "learning_rate": 2.5520512471037213e-06, + "loss": 2.4755151748657225, + "step": 320540 + }, + { + "epoch": 0.8726282836531918, + "grad_norm": 2.3313326835632324, + "learning_rate": 2.5509608831947662e-06, + "loss": 2.4814964294433595, + "step": 320560 + }, + { + "epoch": 0.8726827276439363, + "grad_norm": 2.217115879058838, + "learning_rate": 2.549870519285812e-06, + "loss": 2.4587818145751954, + "step": 320580 + }, + { + "epoch": 0.8727371716346808, + "grad_norm": 2.404435634613037, + "learning_rate": 2.548780155376857e-06, + "loss": 2.476334571838379, + "step": 320600 + }, + { + "epoch": 0.8727916156254254, + "grad_norm": 2.1115238666534424, + "learning_rate": 2.547689791467903e-06, + "loss": 2.466839599609375, + "step": 320620 + }, + { + "epoch": 0.8728460596161699, + "grad_norm": 2.1623847484588623, + "learning_rate": 2.5465994275589478e-06, + "loss": 2.4478891372680662, + "step": 320640 + }, + { + "epoch": 0.8729005036069144, + "grad_norm": 2.2394986152648926, + "learning_rate": 2.5455090636499936e-06, + "loss": 2.4556407928466797, + "step": 320660 + }, + { + "epoch": 0.8729549475976589, + "grad_norm": 2.1535098552703857, + "learning_rate": 2.5444186997410385e-06, + "loss": 2.4550735473632814, + "step": 320680 + }, + { + "epoch": 0.8730093915884034, + "grad_norm": 2.5894696712493896, + "learning_rate": 2.5433283358320843e-06, + "loss": 2.469426727294922, + "step": 320700 + }, + { + "epoch": 0.8730638355791479, + "grad_norm": 2.08901309967041, + "learning_rate": 2.5422379719231293e-06, + "loss": 2.4921422958374024, + "step": 320720 + }, + { + "epoch": 0.8731182795698925, + "grad_norm": 2.1630859375, + "learning_rate": 2.541147608014175e-06, + "loss": 2.440824508666992, + "step": 320740 + }, + { + "epoch": 0.873172723560637, + "grad_norm": 2.1495425701141357, + "learning_rate": 2.54005724410522e-06, + "loss": 2.474867248535156, + "step": 320760 + }, + { + "epoch": 0.8732271675513815, + "grad_norm": 2.1484854221343994, + "learning_rate": 2.538966880196266e-06, + "loss": 2.447332572937012, + "step": 320780 + }, + { + "epoch": 0.873281611542126, + "grad_norm": 2.218339681625366, + "learning_rate": 2.537876516287311e-06, + "loss": 2.479343223571777, + "step": 320800 + }, + { + "epoch": 0.8733360555328705, + "grad_norm": 2.1762819290161133, + "learning_rate": 2.5367861523783567e-06, + "loss": 2.496250534057617, + "step": 320820 + }, + { + "epoch": 0.873390499523615, + "grad_norm": 2.4103870391845703, + "learning_rate": 2.5356957884694016e-06, + "loss": 2.4753545761108398, + "step": 320840 + }, + { + "epoch": 0.8734449435143596, + "grad_norm": 2.2035136222839355, + "learning_rate": 2.5346054245604474e-06, + "loss": 2.4614892959594727, + "step": 320860 + }, + { + "epoch": 0.8734993875051041, + "grad_norm": 2.2194957733154297, + "learning_rate": 2.5335150606514924e-06, + "loss": 2.456675910949707, + "step": 320880 + }, + { + "epoch": 0.8735538314958486, + "grad_norm": 2.1149497032165527, + "learning_rate": 2.532424696742538e-06, + "loss": 2.459543991088867, + "step": 320900 + }, + { + "epoch": 0.8736082754865931, + "grad_norm": 2.0445141792297363, + "learning_rate": 2.531334332833583e-06, + "loss": 2.4643476486206053, + "step": 320920 + }, + { + "epoch": 0.8736627194773376, + "grad_norm": 2.6882293224334717, + "learning_rate": 2.530243968924629e-06, + "loss": 2.4538516998291016, + "step": 320940 + }, + { + "epoch": 0.8737171634680823, + "grad_norm": 2.170360803604126, + "learning_rate": 2.529153605015674e-06, + "loss": 2.4737085342407226, + "step": 320960 + }, + { + "epoch": 0.8737716074588268, + "grad_norm": 2.13942289352417, + "learning_rate": 2.5280632411067197e-06, + "loss": 2.4636425018310546, + "step": 320980 + }, + { + "epoch": 0.8738260514495713, + "grad_norm": 3.6018831729888916, + "learning_rate": 2.5269728771977647e-06, + "loss": 2.4492332458496096, + "step": 321000 + }, + { + "epoch": 0.8738804954403158, + "grad_norm": 2.1624085903167725, + "learning_rate": 2.5258825132888105e-06, + "loss": 2.451654052734375, + "step": 321020 + }, + { + "epoch": 0.8739349394310603, + "grad_norm": 2.2677958011627197, + "learning_rate": 2.5247921493798555e-06, + "loss": 2.470798110961914, + "step": 321040 + }, + { + "epoch": 0.8739893834218048, + "grad_norm": 2.139615297317505, + "learning_rate": 2.5237017854709013e-06, + "loss": 2.4945669174194336, + "step": 321060 + }, + { + "epoch": 0.8740438274125494, + "grad_norm": 2.129537582397461, + "learning_rate": 2.5226114215619462e-06, + "loss": 2.488241195678711, + "step": 321080 + }, + { + "epoch": 0.8740982714032939, + "grad_norm": 2.2301442623138428, + "learning_rate": 2.521521057652992e-06, + "loss": 2.4724662780761717, + "step": 321100 + }, + { + "epoch": 0.8741527153940384, + "grad_norm": 2.1581332683563232, + "learning_rate": 2.520430693744037e-06, + "loss": 2.4640106201171874, + "step": 321120 + }, + { + "epoch": 0.8742071593847829, + "grad_norm": 2.27298641204834, + "learning_rate": 2.519340329835083e-06, + "loss": 2.47103271484375, + "step": 321140 + }, + { + "epoch": 0.8742616033755274, + "grad_norm": 2.12192440032959, + "learning_rate": 2.518249965926128e-06, + "loss": 2.458933448791504, + "step": 321160 + }, + { + "epoch": 0.874316047366272, + "grad_norm": 2.0920608043670654, + "learning_rate": 2.5171596020171736e-06, + "loss": 2.4272762298583985, + "step": 321180 + }, + { + "epoch": 0.8743704913570165, + "grad_norm": 2.1720476150512695, + "learning_rate": 2.5160692381082186e-06, + "loss": 2.449008750915527, + "step": 321200 + }, + { + "epoch": 0.874424935347761, + "grad_norm": 2.2086589336395264, + "learning_rate": 2.5149788741992644e-06, + "loss": 2.4531494140625, + "step": 321220 + }, + { + "epoch": 0.8744793793385055, + "grad_norm": 2.1874592304229736, + "learning_rate": 2.5138885102903093e-06, + "loss": 2.4472532272338867, + "step": 321240 + }, + { + "epoch": 0.87453382332925, + "grad_norm": 2.138909339904785, + "learning_rate": 2.512798146381355e-06, + "loss": 2.4637825012207033, + "step": 321260 + }, + { + "epoch": 0.8745882673199945, + "grad_norm": 3.5947835445404053, + "learning_rate": 2.5117077824724e-06, + "loss": 2.4411203384399416, + "step": 321280 + }, + { + "epoch": 0.8746427113107391, + "grad_norm": 2.1330714225769043, + "learning_rate": 2.510617418563446e-06, + "loss": 2.457366943359375, + "step": 321300 + }, + { + "epoch": 0.8746971553014836, + "grad_norm": 2.1202995777130127, + "learning_rate": 2.509527054654491e-06, + "loss": 2.4343755722045897, + "step": 321320 + }, + { + "epoch": 0.8747515992922281, + "grad_norm": 2.220067024230957, + "learning_rate": 2.5084366907455367e-06, + "loss": 2.4348175048828127, + "step": 321340 + }, + { + "epoch": 0.8748060432829726, + "grad_norm": 2.082794427871704, + "learning_rate": 2.507346326836582e-06, + "loss": 2.457904052734375, + "step": 321360 + }, + { + "epoch": 0.8748604872737171, + "grad_norm": 2.031527280807495, + "learning_rate": 2.5062559629276275e-06, + "loss": 2.4490179061889648, + "step": 321380 + }, + { + "epoch": 0.8749149312644617, + "grad_norm": 2.0931403636932373, + "learning_rate": 2.505165599018673e-06, + "loss": 2.4690881729125977, + "step": 321400 + }, + { + "epoch": 0.8749693752552062, + "grad_norm": 2.715630054473877, + "learning_rate": 2.5040752351097182e-06, + "loss": 2.4694047927856446, + "step": 321420 + }, + { + "epoch": 0.8750238192459507, + "grad_norm": 2.1244864463806152, + "learning_rate": 2.5029848712007636e-06, + "loss": 2.4652061462402344, + "step": 321440 + }, + { + "epoch": 0.8750782632366952, + "grad_norm": 2.129422903060913, + "learning_rate": 2.501894507291809e-06, + "loss": 2.4727542877197264, + "step": 321460 + }, + { + "epoch": 0.8751327072274397, + "grad_norm": 2.110976457595825, + "learning_rate": 2.5008041433828544e-06, + "loss": 2.4779247283935546, + "step": 321480 + }, + { + "epoch": 0.8751871512181842, + "grad_norm": 2.2779688835144043, + "learning_rate": 2.4997137794738998e-06, + "loss": 2.4891971588134765, + "step": 321500 + }, + { + "epoch": 0.8752415952089289, + "grad_norm": 2.1201157569885254, + "learning_rate": 2.498623415564945e-06, + "loss": 2.459792137145996, + "step": 321520 + }, + { + "epoch": 0.8752960391996734, + "grad_norm": 2.229426383972168, + "learning_rate": 2.4975330516559905e-06, + "loss": 2.444591522216797, + "step": 321540 + }, + { + "epoch": 0.8753504831904179, + "grad_norm": 2.1696815490722656, + "learning_rate": 2.496442687747036e-06, + "loss": 2.4513582229614257, + "step": 321560 + }, + { + "epoch": 0.8754049271811624, + "grad_norm": 2.2117137908935547, + "learning_rate": 2.4953523238380813e-06, + "loss": 2.4384716033935545, + "step": 321580 + }, + { + "epoch": 0.8754593711719069, + "grad_norm": 2.096604347229004, + "learning_rate": 2.4942619599291267e-06, + "loss": 2.4653041839599608, + "step": 321600 + }, + { + "epoch": 0.8755138151626515, + "grad_norm": 2.2053327560424805, + "learning_rate": 2.493171596020172e-06, + "loss": 2.431104278564453, + "step": 321620 + }, + { + "epoch": 0.875568259153396, + "grad_norm": 2.16892671585083, + "learning_rate": 2.4920812321112175e-06, + "loss": 2.4449111938476564, + "step": 321640 + }, + { + "epoch": 0.8756227031441405, + "grad_norm": 2.2235560417175293, + "learning_rate": 2.490990868202263e-06, + "loss": 2.449441146850586, + "step": 321660 + }, + { + "epoch": 0.875677147134885, + "grad_norm": 2.1608340740203857, + "learning_rate": 2.4899005042933082e-06, + "loss": 2.4674762725830077, + "step": 321680 + }, + { + "epoch": 0.8757315911256295, + "grad_norm": 2.199240207672119, + "learning_rate": 2.4888101403843536e-06, + "loss": 2.4822210311889648, + "step": 321700 + }, + { + "epoch": 0.875786035116374, + "grad_norm": 2.0966737270355225, + "learning_rate": 2.487719776475399e-06, + "loss": 2.4617433547973633, + "step": 321720 + }, + { + "epoch": 0.8758404791071186, + "grad_norm": 2.164076805114746, + "learning_rate": 2.4866294125664444e-06, + "loss": 2.459352493286133, + "step": 321740 + }, + { + "epoch": 0.8758949230978631, + "grad_norm": 2.1622490882873535, + "learning_rate": 2.4855390486574894e-06, + "loss": 2.45355224609375, + "step": 321760 + }, + { + "epoch": 0.8759493670886076, + "grad_norm": 2.1697726249694824, + "learning_rate": 2.4844486847485347e-06, + "loss": 2.4879894256591797, + "step": 321780 + }, + { + "epoch": 0.8760038110793521, + "grad_norm": 2.4260761737823486, + "learning_rate": 2.48335832083958e-06, + "loss": 2.466286849975586, + "step": 321800 + }, + { + "epoch": 0.8760582550700966, + "grad_norm": 2.0040760040283203, + "learning_rate": 2.4822679569306255e-06, + "loss": 2.4770185470581056, + "step": 321820 + }, + { + "epoch": 0.8761126990608411, + "grad_norm": 2.1759517192840576, + "learning_rate": 2.481177593021671e-06, + "loss": 2.4516633987426757, + "step": 321840 + }, + { + "epoch": 0.8761671430515857, + "grad_norm": 2.610215902328491, + "learning_rate": 2.4800872291127163e-06, + "loss": 2.4427791595458985, + "step": 321860 + }, + { + "epoch": 0.8762215870423302, + "grad_norm": 2.248239755630493, + "learning_rate": 2.4789968652037617e-06, + "loss": 2.447276306152344, + "step": 321880 + }, + { + "epoch": 0.8762760310330747, + "grad_norm": 2.1456756591796875, + "learning_rate": 2.477906501294807e-06, + "loss": 2.4331478118896483, + "step": 321900 + }, + { + "epoch": 0.8763304750238192, + "grad_norm": 2.1978750228881836, + "learning_rate": 2.4768161373858524e-06, + "loss": 2.4640134811401366, + "step": 321920 + }, + { + "epoch": 0.8763849190145637, + "grad_norm": 2.1807498931884766, + "learning_rate": 2.475725773476898e-06, + "loss": 2.453565788269043, + "step": 321940 + }, + { + "epoch": 0.8764393630053083, + "grad_norm": 2.192903757095337, + "learning_rate": 2.4746354095679432e-06, + "loss": 2.4598697662353515, + "step": 321960 + }, + { + "epoch": 0.8764938069960528, + "grad_norm": 2.2244319915771484, + "learning_rate": 2.4735450456589886e-06, + "loss": 2.458848571777344, + "step": 321980 + }, + { + "epoch": 0.8765482509867973, + "grad_norm": 2.1875908374786377, + "learning_rate": 2.472454681750034e-06, + "loss": 2.456049919128418, + "step": 322000 + }, + { + "epoch": 0.8766026949775418, + "grad_norm": 2.230710029602051, + "learning_rate": 2.4713643178410794e-06, + "loss": 2.4222639083862303, + "step": 322020 + }, + { + "epoch": 0.8766571389682863, + "grad_norm": 2.183117151260376, + "learning_rate": 2.4702739539321248e-06, + "loss": 2.4212368011474608, + "step": 322040 + }, + { + "epoch": 0.8767115829590308, + "grad_norm": 2.1483168601989746, + "learning_rate": 2.46918359002317e-06, + "loss": 2.4577974319458007, + "step": 322060 + }, + { + "epoch": 0.8767660269497755, + "grad_norm": 2.212827444076538, + "learning_rate": 2.4680932261142155e-06, + "loss": 2.4256612777709963, + "step": 322080 + }, + { + "epoch": 0.87682047094052, + "grad_norm": 2.0950570106506348, + "learning_rate": 2.467002862205261e-06, + "loss": 2.4881301879882813, + "step": 322100 + }, + { + "epoch": 0.8768749149312645, + "grad_norm": 2.198049783706665, + "learning_rate": 2.4659124982963063e-06, + "loss": 2.453268051147461, + "step": 322120 + }, + { + "epoch": 0.876929358922009, + "grad_norm": 2.127964496612549, + "learning_rate": 2.4648221343873517e-06, + "loss": 2.4443115234375, + "step": 322140 + }, + { + "epoch": 0.8769838029127535, + "grad_norm": 2.1617937088012695, + "learning_rate": 2.463731770478397e-06, + "loss": 2.442508888244629, + "step": 322160 + }, + { + "epoch": 0.8770382469034981, + "grad_norm": 2.291287899017334, + "learning_rate": 2.4626414065694425e-06, + "loss": 2.4699859619140625, + "step": 322180 + }, + { + "epoch": 0.8770926908942426, + "grad_norm": 2.1561689376831055, + "learning_rate": 2.4615510426604883e-06, + "loss": 2.467154121398926, + "step": 322200 + }, + { + "epoch": 0.8771471348849871, + "grad_norm": 2.132920503616333, + "learning_rate": 2.4604606787515337e-06, + "loss": 2.431305694580078, + "step": 322220 + }, + { + "epoch": 0.8772015788757316, + "grad_norm": 2.2062056064605713, + "learning_rate": 2.459370314842579e-06, + "loss": 2.5050069808959963, + "step": 322240 + }, + { + "epoch": 0.8772560228664761, + "grad_norm": 2.166963815689087, + "learning_rate": 2.4582799509336244e-06, + "loss": 2.4637718200683594, + "step": 322260 + }, + { + "epoch": 0.8773104668572206, + "grad_norm": 2.1860358715057373, + "learning_rate": 2.45718958702467e-06, + "loss": 2.4619794845581056, + "step": 322280 + }, + { + "epoch": 0.8773649108479652, + "grad_norm": 2.753422975540161, + "learning_rate": 2.456099223115715e-06, + "loss": 2.448541831970215, + "step": 322300 + }, + { + "epoch": 0.8774193548387097, + "grad_norm": 2.167470693588257, + "learning_rate": 2.4550088592067606e-06, + "loss": 2.440627098083496, + "step": 322320 + }, + { + "epoch": 0.8774737988294542, + "grad_norm": 2.22129487991333, + "learning_rate": 2.453918495297806e-06, + "loss": 2.4659481048583984, + "step": 322340 + }, + { + "epoch": 0.8775282428201987, + "grad_norm": 2.2394399642944336, + "learning_rate": 2.4528281313888514e-06, + "loss": 2.4500137329101563, + "step": 322360 + }, + { + "epoch": 0.8775826868109432, + "grad_norm": 2.0760092735290527, + "learning_rate": 2.4517377674798967e-06, + "loss": 2.4999788284301756, + "step": 322380 + }, + { + "epoch": 0.8776371308016878, + "grad_norm": 2.25998592376709, + "learning_rate": 2.450647403570942e-06, + "loss": 2.4624881744384766, + "step": 322400 + }, + { + "epoch": 0.8776915747924323, + "grad_norm": 2.1480987071990967, + "learning_rate": 2.4495570396619875e-06, + "loss": 2.454069900512695, + "step": 322420 + }, + { + "epoch": 0.8777460187831768, + "grad_norm": 2.2643675804138184, + "learning_rate": 2.448466675753033e-06, + "loss": 2.4577741622924805, + "step": 322440 + }, + { + "epoch": 0.8778004627739213, + "grad_norm": 2.0937881469726562, + "learning_rate": 2.4473763118440783e-06, + "loss": 2.4619707107543944, + "step": 322460 + }, + { + "epoch": 0.8778549067646658, + "grad_norm": 2.153798818588257, + "learning_rate": 2.4462859479351237e-06, + "loss": 2.4716232299804686, + "step": 322480 + }, + { + "epoch": 0.8779093507554103, + "grad_norm": 2.336376428604126, + "learning_rate": 2.445195584026169e-06, + "loss": 2.491668128967285, + "step": 322500 + }, + { + "epoch": 0.8779637947461549, + "grad_norm": 2.201444387435913, + "learning_rate": 2.4441052201172144e-06, + "loss": 2.463787841796875, + "step": 322520 + }, + { + "epoch": 0.8780182387368994, + "grad_norm": 2.081448793411255, + "learning_rate": 2.44301485620826e-06, + "loss": 2.458613967895508, + "step": 322540 + }, + { + "epoch": 0.8780726827276439, + "grad_norm": 2.2402002811431885, + "learning_rate": 2.4419244922993052e-06, + "loss": 2.4614112854003904, + "step": 322560 + }, + { + "epoch": 0.8781271267183884, + "grad_norm": 2.1613693237304688, + "learning_rate": 2.4408341283903506e-06, + "loss": 2.434016990661621, + "step": 322580 + }, + { + "epoch": 0.8781815707091329, + "grad_norm": 2.0910956859588623, + "learning_rate": 2.439743764481396e-06, + "loss": 2.4476558685302736, + "step": 322600 + }, + { + "epoch": 0.8782360146998776, + "grad_norm": 2.073072671890259, + "learning_rate": 2.4386534005724414e-06, + "loss": 2.444916343688965, + "step": 322620 + }, + { + "epoch": 0.8782904586906221, + "grad_norm": 2.1247386932373047, + "learning_rate": 2.4375630366634868e-06, + "loss": 2.468027877807617, + "step": 322640 + }, + { + "epoch": 0.8783449026813666, + "grad_norm": 2.12941837310791, + "learning_rate": 2.436472672754532e-06, + "loss": 2.444265365600586, + "step": 322660 + }, + { + "epoch": 0.8783993466721111, + "grad_norm": 2.2010715007781982, + "learning_rate": 2.4353823088455775e-06, + "loss": 2.4497398376464843, + "step": 322680 + }, + { + "epoch": 0.8784537906628556, + "grad_norm": 2.246645450592041, + "learning_rate": 2.434291944936623e-06, + "loss": 2.450438690185547, + "step": 322700 + }, + { + "epoch": 0.8785082346536001, + "grad_norm": 2.1494455337524414, + "learning_rate": 2.4332015810276683e-06, + "loss": 2.4813331604003905, + "step": 322720 + }, + { + "epoch": 0.8785626786443447, + "grad_norm": 2.199671983718872, + "learning_rate": 2.4321112171187137e-06, + "loss": 2.4752643585205076, + "step": 322740 + }, + { + "epoch": 0.8786171226350892, + "grad_norm": 2.1476240158081055, + "learning_rate": 2.431020853209759e-06, + "loss": 2.451142501831055, + "step": 322760 + }, + { + "epoch": 0.8786715666258337, + "grad_norm": 3.158512830734253, + "learning_rate": 2.4299304893008045e-06, + "loss": 2.4881851196289064, + "step": 322780 + }, + { + "epoch": 0.8787260106165782, + "grad_norm": 2.238128185272217, + "learning_rate": 2.42884012539185e-06, + "loss": 2.453720474243164, + "step": 322800 + }, + { + "epoch": 0.8787804546073227, + "grad_norm": 2.2253220081329346, + "learning_rate": 2.4277497614828952e-06, + "loss": 2.44596004486084, + "step": 322820 + }, + { + "epoch": 0.8788348985980673, + "grad_norm": 2.1095383167266846, + "learning_rate": 2.4266593975739406e-06, + "loss": 2.470063591003418, + "step": 322840 + }, + { + "epoch": 0.8788893425888118, + "grad_norm": 2.3220183849334717, + "learning_rate": 2.425569033664986e-06, + "loss": 2.465946388244629, + "step": 322860 + }, + { + "epoch": 0.8789437865795563, + "grad_norm": 2.109091281890869, + "learning_rate": 2.4244786697560314e-06, + "loss": 2.4816781997680666, + "step": 322880 + }, + { + "epoch": 0.8789982305703008, + "grad_norm": 2.6274538040161133, + "learning_rate": 2.4233883058470768e-06, + "loss": 2.447765350341797, + "step": 322900 + }, + { + "epoch": 0.8790526745610453, + "grad_norm": 2.171804189682007, + "learning_rate": 2.422297941938122e-06, + "loss": 2.4674345016479493, + "step": 322920 + }, + { + "epoch": 0.8791071185517898, + "grad_norm": 2.3325610160827637, + "learning_rate": 2.4212075780291675e-06, + "loss": 2.472167205810547, + "step": 322940 + }, + { + "epoch": 0.8791615625425344, + "grad_norm": 2.2876663208007812, + "learning_rate": 2.420117214120213e-06, + "loss": 2.4693424224853517, + "step": 322960 + }, + { + "epoch": 0.8792160065332789, + "grad_norm": 2.1001176834106445, + "learning_rate": 2.4190268502112583e-06, + "loss": 2.4719409942626953, + "step": 322980 + }, + { + "epoch": 0.8792704505240234, + "grad_norm": 2.0810232162475586, + "learning_rate": 2.4179364863023037e-06, + "loss": 2.4321033477783205, + "step": 323000 + }, + { + "epoch": 0.8793248945147679, + "grad_norm": 2.182774543762207, + "learning_rate": 2.416846122393349e-06, + "loss": 2.456620788574219, + "step": 323020 + }, + { + "epoch": 0.8793793385055124, + "grad_norm": 2.2379026412963867, + "learning_rate": 2.4157557584843945e-06, + "loss": 2.448881912231445, + "step": 323040 + }, + { + "epoch": 0.8794337824962569, + "grad_norm": 2.1698083877563477, + "learning_rate": 2.41466539457544e-06, + "loss": 2.4798198699951173, + "step": 323060 + }, + { + "epoch": 0.8794882264870015, + "grad_norm": 2.1424560546875, + "learning_rate": 2.4135750306664852e-06, + "loss": 2.4924348831176757, + "step": 323080 + }, + { + "epoch": 0.879542670477746, + "grad_norm": 2.154073476791382, + "learning_rate": 2.4124846667575306e-06, + "loss": 2.45074405670166, + "step": 323100 + }, + { + "epoch": 0.8795971144684905, + "grad_norm": 2.27166748046875, + "learning_rate": 2.411394302848576e-06, + "loss": 2.4695980072021486, + "step": 323120 + }, + { + "epoch": 0.879651558459235, + "grad_norm": 2.1454834938049316, + "learning_rate": 2.4103039389396214e-06, + "loss": 2.427112007141113, + "step": 323140 + }, + { + "epoch": 0.8797060024499795, + "grad_norm": 2.24627685546875, + "learning_rate": 2.4092135750306668e-06, + "loss": 2.4564985275268554, + "step": 323160 + }, + { + "epoch": 0.8797604464407242, + "grad_norm": 2.1917433738708496, + "learning_rate": 2.408123211121712e-06, + "loss": 2.4648929595947267, + "step": 323180 + }, + { + "epoch": 0.8798148904314687, + "grad_norm": 2.9672632217407227, + "learning_rate": 2.4070328472127576e-06, + "loss": 2.4410072326660157, + "step": 323200 + }, + { + "epoch": 0.8798693344222132, + "grad_norm": 2.240434408187866, + "learning_rate": 2.405942483303803e-06, + "loss": 2.467042922973633, + "step": 323220 + }, + { + "epoch": 0.8799237784129577, + "grad_norm": 2.193566083908081, + "learning_rate": 2.4048521193948483e-06, + "loss": 2.463426399230957, + "step": 323240 + }, + { + "epoch": 0.8799782224037022, + "grad_norm": 2.2357118129730225, + "learning_rate": 2.4037617554858937e-06, + "loss": 2.4623077392578123, + "step": 323260 + }, + { + "epoch": 0.8800326663944467, + "grad_norm": 2.336883544921875, + "learning_rate": 2.402671391576939e-06, + "loss": 2.451679801940918, + "step": 323280 + }, + { + "epoch": 0.8800871103851913, + "grad_norm": 2.38135027885437, + "learning_rate": 2.4015810276679845e-06, + "loss": 2.4557180404663086, + "step": 323300 + }, + { + "epoch": 0.8801415543759358, + "grad_norm": 2.2507526874542236, + "learning_rate": 2.40049066375903e-06, + "loss": 2.454335594177246, + "step": 323320 + }, + { + "epoch": 0.8801959983666803, + "grad_norm": 2.157151222229004, + "learning_rate": 2.3994002998500753e-06, + "loss": 2.4430313110351562, + "step": 323340 + }, + { + "epoch": 0.8802504423574248, + "grad_norm": 2.4105398654937744, + "learning_rate": 2.3983099359411206e-06, + "loss": 2.427517890930176, + "step": 323360 + }, + { + "epoch": 0.8803048863481693, + "grad_norm": 2.184047222137451, + "learning_rate": 2.397219572032166e-06, + "loss": 2.452433776855469, + "step": 323380 + }, + { + "epoch": 0.8803593303389139, + "grad_norm": 2.2554783821105957, + "learning_rate": 2.3961292081232114e-06, + "loss": 2.4711484909057617, + "step": 323400 + }, + { + "epoch": 0.8804137743296584, + "grad_norm": 2.180468797683716, + "learning_rate": 2.395038844214257e-06, + "loss": 2.467798042297363, + "step": 323420 + }, + { + "epoch": 0.8804682183204029, + "grad_norm": 2.101155996322632, + "learning_rate": 2.393948480305302e-06, + "loss": 2.467201805114746, + "step": 323440 + }, + { + "epoch": 0.8805226623111474, + "grad_norm": 2.2330095767974854, + "learning_rate": 2.3928581163963476e-06, + "loss": 2.468464660644531, + "step": 323460 + }, + { + "epoch": 0.8805771063018919, + "grad_norm": 2.3592376708984375, + "learning_rate": 2.391767752487393e-06, + "loss": 2.452862358093262, + "step": 323480 + }, + { + "epoch": 0.8806315502926364, + "grad_norm": 2.1717617511749268, + "learning_rate": 2.3906773885784383e-06, + "loss": 2.470799446105957, + "step": 323500 + }, + { + "epoch": 0.880685994283381, + "grad_norm": 2.2639286518096924, + "learning_rate": 2.3895870246694837e-06, + "loss": 2.4674678802490235, + "step": 323520 + }, + { + "epoch": 0.8807404382741255, + "grad_norm": 2.429429769515991, + "learning_rate": 2.388496660760529e-06, + "loss": 2.4802600860595705, + "step": 323540 + }, + { + "epoch": 0.88079488226487, + "grad_norm": 2.1535685062408447, + "learning_rate": 2.3874062968515745e-06, + "loss": 2.4773881912231444, + "step": 323560 + }, + { + "epoch": 0.8808493262556145, + "grad_norm": 2.199794054031372, + "learning_rate": 2.38631593294262e-06, + "loss": 2.462181282043457, + "step": 323580 + }, + { + "epoch": 0.880903770246359, + "grad_norm": 2.382798194885254, + "learning_rate": 2.3852255690336653e-06, + "loss": 2.45593376159668, + "step": 323600 + }, + { + "epoch": 0.8809582142371036, + "grad_norm": 2.1126339435577393, + "learning_rate": 2.3841352051247107e-06, + "loss": 2.467656135559082, + "step": 323620 + }, + { + "epoch": 0.8810126582278481, + "grad_norm": 2.1427807807922363, + "learning_rate": 2.383044841215756e-06, + "loss": 2.443381690979004, + "step": 323640 + }, + { + "epoch": 0.8810671022185926, + "grad_norm": 2.2180228233337402, + "learning_rate": 2.3819544773068014e-06, + "loss": 2.4852472305297852, + "step": 323660 + }, + { + "epoch": 0.8811215462093371, + "grad_norm": 2.2706899642944336, + "learning_rate": 2.380864113397847e-06, + "loss": 2.430097770690918, + "step": 323680 + }, + { + "epoch": 0.8811759902000816, + "grad_norm": 2.4781417846679688, + "learning_rate": 2.379773749488892e-06, + "loss": 2.4413784027099608, + "step": 323700 + }, + { + "epoch": 0.8812304341908261, + "grad_norm": 2.24735951423645, + "learning_rate": 2.3786833855799376e-06, + "loss": 2.4435848236083983, + "step": 323720 + }, + { + "epoch": 0.8812848781815708, + "grad_norm": 2.290060043334961, + "learning_rate": 2.377593021670983e-06, + "loss": 2.446522521972656, + "step": 323740 + }, + { + "epoch": 0.8813393221723153, + "grad_norm": 2.1080873012542725, + "learning_rate": 2.3765026577620284e-06, + "loss": 2.4693120956420898, + "step": 323760 + }, + { + "epoch": 0.8813937661630598, + "grad_norm": 2.107759952545166, + "learning_rate": 2.3754122938530737e-06, + "loss": 2.4757102966308593, + "step": 323780 + }, + { + "epoch": 0.8814482101538043, + "grad_norm": 2.187382221221924, + "learning_rate": 2.374321929944119e-06, + "loss": 2.5023931503295898, + "step": 323800 + }, + { + "epoch": 0.8815026541445488, + "grad_norm": 2.114976167678833, + "learning_rate": 2.3732315660351645e-06, + "loss": 2.4558128356933593, + "step": 323820 + }, + { + "epoch": 0.8815570981352934, + "grad_norm": 2.2265777587890625, + "learning_rate": 2.3721412021262095e-06, + "loss": 2.439564895629883, + "step": 323840 + }, + { + "epoch": 0.8816115421260379, + "grad_norm": 2.139777898788452, + "learning_rate": 2.371050838217255e-06, + "loss": 2.4651233673095705, + "step": 323860 + }, + { + "epoch": 0.8816659861167824, + "grad_norm": 2.2123544216156006, + "learning_rate": 2.3699604743083002e-06, + "loss": 2.47039737701416, + "step": 323880 + }, + { + "epoch": 0.8817204301075269, + "grad_norm": 2.1963865756988525, + "learning_rate": 2.368870110399346e-06, + "loss": 2.451356887817383, + "step": 323900 + }, + { + "epoch": 0.8817748740982714, + "grad_norm": 2.315502882003784, + "learning_rate": 2.3677797464903914e-06, + "loss": 2.453390693664551, + "step": 323920 + }, + { + "epoch": 0.8818293180890159, + "grad_norm": 2.1952223777770996, + "learning_rate": 2.366689382581437e-06, + "loss": 2.461086654663086, + "step": 323940 + }, + { + "epoch": 0.8818837620797605, + "grad_norm": 4.66331672668457, + "learning_rate": 2.3655990186724822e-06, + "loss": 2.452017593383789, + "step": 323960 + }, + { + "epoch": 0.881938206070505, + "grad_norm": 2.7359912395477295, + "learning_rate": 2.3645086547635276e-06, + "loss": 2.4404600143432615, + "step": 323980 + }, + { + "epoch": 0.8819926500612495, + "grad_norm": 2.1504061222076416, + "learning_rate": 2.363418290854573e-06, + "loss": 2.511570930480957, + "step": 324000 + }, + { + "epoch": 0.882047094051994, + "grad_norm": 2.165630578994751, + "learning_rate": 2.3623279269456184e-06, + "loss": 2.460332489013672, + "step": 324020 + }, + { + "epoch": 0.8821015380427385, + "grad_norm": 2.133615493774414, + "learning_rate": 2.3612375630366638e-06, + "loss": 2.473195457458496, + "step": 324040 + }, + { + "epoch": 0.882155982033483, + "grad_norm": 2.2750158309936523, + "learning_rate": 2.360147199127709e-06, + "loss": 2.4614564895629885, + "step": 324060 + }, + { + "epoch": 0.8822104260242276, + "grad_norm": 2.225663185119629, + "learning_rate": 2.3590568352187545e-06, + "loss": 2.451143264770508, + "step": 324080 + }, + { + "epoch": 0.8822648700149721, + "grad_norm": 2.1059539318084717, + "learning_rate": 2.3579664713098e-06, + "loss": 2.471072196960449, + "step": 324100 + }, + { + "epoch": 0.8823193140057166, + "grad_norm": 2.262608528137207, + "learning_rate": 2.3568761074008453e-06, + "loss": 2.4766382217407226, + "step": 324120 + }, + { + "epoch": 0.8823737579964611, + "grad_norm": 2.1754062175750732, + "learning_rate": 2.3557857434918907e-06, + "loss": 2.473529815673828, + "step": 324140 + }, + { + "epoch": 0.8824282019872056, + "grad_norm": 2.279654026031494, + "learning_rate": 2.354695379582936e-06, + "loss": 2.4745929718017576, + "step": 324160 + }, + { + "epoch": 0.8824826459779502, + "grad_norm": 2.0842666625976562, + "learning_rate": 2.3536050156739815e-06, + "loss": 2.4565921783447267, + "step": 324180 + }, + { + "epoch": 0.8825370899686947, + "grad_norm": 2.113913059234619, + "learning_rate": 2.352514651765027e-06, + "loss": 2.4581367492675783, + "step": 324200 + }, + { + "epoch": 0.8825915339594392, + "grad_norm": 2.1819100379943848, + "learning_rate": 2.3514242878560722e-06, + "loss": 2.443955421447754, + "step": 324220 + }, + { + "epoch": 0.8826459779501837, + "grad_norm": 2.1607940196990967, + "learning_rate": 2.3503339239471176e-06, + "loss": 2.4627811431884767, + "step": 324240 + }, + { + "epoch": 0.8827004219409282, + "grad_norm": 2.1496033668518066, + "learning_rate": 2.349243560038163e-06, + "loss": 2.4944021224975588, + "step": 324260 + }, + { + "epoch": 0.8827548659316727, + "grad_norm": 2.155200481414795, + "learning_rate": 2.3481531961292084e-06, + "loss": 2.4804924011230467, + "step": 324280 + }, + { + "epoch": 0.8828093099224174, + "grad_norm": 2.2159488201141357, + "learning_rate": 2.3470628322202538e-06, + "loss": 2.455026054382324, + "step": 324300 + }, + { + "epoch": 0.8828637539131619, + "grad_norm": 2.2851674556732178, + "learning_rate": 2.345972468311299e-06, + "loss": 2.447259521484375, + "step": 324320 + }, + { + "epoch": 0.8829181979039064, + "grad_norm": 2.2640554904937744, + "learning_rate": 2.3448821044023445e-06, + "loss": 2.4361013412475585, + "step": 324340 + }, + { + "epoch": 0.8829726418946509, + "grad_norm": 2.1804401874542236, + "learning_rate": 2.34379174049339e-06, + "loss": 2.482321929931641, + "step": 324360 + }, + { + "epoch": 0.8830270858853954, + "grad_norm": 2.1622447967529297, + "learning_rate": 2.3427013765844353e-06, + "loss": 2.435403060913086, + "step": 324380 + }, + { + "epoch": 0.88308152987614, + "grad_norm": 2.1521878242492676, + "learning_rate": 2.3416110126754807e-06, + "loss": 2.4333147048950194, + "step": 324400 + }, + { + "epoch": 0.8831359738668845, + "grad_norm": 2.247019052505493, + "learning_rate": 2.340520648766526e-06, + "loss": 2.459889602661133, + "step": 324420 + }, + { + "epoch": 0.883190417857629, + "grad_norm": 2.135283946990967, + "learning_rate": 2.3394302848575715e-06, + "loss": 2.4996891021728516, + "step": 324440 + }, + { + "epoch": 0.8832448618483735, + "grad_norm": 2.2846384048461914, + "learning_rate": 2.338339920948617e-06, + "loss": 2.4952880859375, + "step": 324460 + }, + { + "epoch": 0.883299305839118, + "grad_norm": 2.149585723876953, + "learning_rate": 2.3372495570396622e-06, + "loss": 2.470632553100586, + "step": 324480 + }, + { + "epoch": 0.8833537498298625, + "grad_norm": 3.012005567550659, + "learning_rate": 2.3361591931307076e-06, + "loss": 2.4837099075317384, + "step": 324500 + }, + { + "epoch": 0.8834081938206071, + "grad_norm": 2.1833274364471436, + "learning_rate": 2.335068829221753e-06, + "loss": 2.441239929199219, + "step": 324520 + }, + { + "epoch": 0.8834626378113516, + "grad_norm": 2.3050332069396973, + "learning_rate": 2.3339784653127984e-06, + "loss": 2.437725639343262, + "step": 324540 + }, + { + "epoch": 0.8835170818020961, + "grad_norm": 2.235551357269287, + "learning_rate": 2.3328881014038438e-06, + "loss": 2.4559186935424804, + "step": 324560 + }, + { + "epoch": 0.8835715257928406, + "grad_norm": 2.4584310054779053, + "learning_rate": 2.331797737494889e-06, + "loss": 2.430055618286133, + "step": 324580 + }, + { + "epoch": 0.8836259697835851, + "grad_norm": 2.0946969985961914, + "learning_rate": 2.3307073735859346e-06, + "loss": 2.4221002578735353, + "step": 324600 + }, + { + "epoch": 0.8836804137743297, + "grad_norm": 2.1545581817626953, + "learning_rate": 2.32961700967698e-06, + "loss": 2.4535850524902343, + "step": 324620 + }, + { + "epoch": 0.8837348577650742, + "grad_norm": 2.226158618927002, + "learning_rate": 2.3285266457680253e-06, + "loss": 2.4756664276123046, + "step": 324640 + }, + { + "epoch": 0.8837893017558187, + "grad_norm": 2.125767230987549, + "learning_rate": 2.3274362818590707e-06, + "loss": 2.4641498565673827, + "step": 324660 + }, + { + "epoch": 0.8838437457465632, + "grad_norm": 2.281601667404175, + "learning_rate": 2.326345917950116e-06, + "loss": 2.4456525802612306, + "step": 324680 + }, + { + "epoch": 0.8838981897373077, + "grad_norm": 2.577674627304077, + "learning_rate": 2.3252555540411615e-06, + "loss": 2.439946746826172, + "step": 324700 + }, + { + "epoch": 0.8839526337280522, + "grad_norm": 2.2634429931640625, + "learning_rate": 2.324165190132207e-06, + "loss": 2.4899290084838865, + "step": 324720 + }, + { + "epoch": 0.8840070777187968, + "grad_norm": 2.1492772102355957, + "learning_rate": 2.3230748262232523e-06, + "loss": 2.456701469421387, + "step": 324740 + }, + { + "epoch": 0.8840615217095413, + "grad_norm": 2.153764247894287, + "learning_rate": 2.3219844623142976e-06, + "loss": 2.4083484649658202, + "step": 324760 + }, + { + "epoch": 0.8841159657002858, + "grad_norm": 2.0797324180603027, + "learning_rate": 2.320894098405343e-06, + "loss": 2.45068359375, + "step": 324780 + }, + { + "epoch": 0.8841704096910303, + "grad_norm": 2.1968448162078857, + "learning_rate": 2.3198037344963884e-06, + "loss": 2.4803287506103517, + "step": 324800 + }, + { + "epoch": 0.8842248536817748, + "grad_norm": 2.1723508834838867, + "learning_rate": 2.318713370587434e-06, + "loss": 2.5078866958618162, + "step": 324820 + }, + { + "epoch": 0.8842792976725194, + "grad_norm": 2.1927988529205322, + "learning_rate": 2.317623006678479e-06, + "loss": 2.447830009460449, + "step": 324840 + }, + { + "epoch": 0.884333741663264, + "grad_norm": 2.220576763153076, + "learning_rate": 2.3165326427695246e-06, + "loss": 2.464419364929199, + "step": 324860 + }, + { + "epoch": 0.8843881856540085, + "grad_norm": 2.1689345836639404, + "learning_rate": 2.31544227886057e-06, + "loss": 2.442691612243652, + "step": 324880 + }, + { + "epoch": 0.884442629644753, + "grad_norm": 2.173433303833008, + "learning_rate": 2.3143519149516153e-06, + "loss": 2.437782096862793, + "step": 324900 + }, + { + "epoch": 0.8844970736354975, + "grad_norm": 2.251504421234131, + "learning_rate": 2.3132615510426607e-06, + "loss": 2.4397586822509765, + "step": 324920 + }, + { + "epoch": 0.884551517626242, + "grad_norm": 2.157464027404785, + "learning_rate": 2.312171187133706e-06, + "loss": 2.461907386779785, + "step": 324940 + }, + { + "epoch": 0.8846059616169866, + "grad_norm": 2.3001246452331543, + "learning_rate": 2.3110808232247515e-06, + "loss": 2.4798784255981445, + "step": 324960 + }, + { + "epoch": 0.8846604056077311, + "grad_norm": 2.162721633911133, + "learning_rate": 2.309990459315797e-06, + "loss": 2.4786176681518555, + "step": 324980 + }, + { + "epoch": 0.8847148495984756, + "grad_norm": 2.6160638332366943, + "learning_rate": 2.3089000954068423e-06, + "loss": 2.489892578125, + "step": 325000 + }, + { + "epoch": 0.8847692935892201, + "grad_norm": 2.831645965576172, + "learning_rate": 2.3078097314978877e-06, + "loss": 2.4486413955688477, + "step": 325020 + }, + { + "epoch": 0.8848237375799646, + "grad_norm": 2.1448822021484375, + "learning_rate": 2.306719367588933e-06, + "loss": 2.4212671279907227, + "step": 325040 + }, + { + "epoch": 0.8848781815707092, + "grad_norm": 2.0966603755950928, + "learning_rate": 2.3056290036799784e-06, + "loss": 2.4564746856689452, + "step": 325060 + }, + { + "epoch": 0.8849326255614537, + "grad_norm": 2.2314352989196777, + "learning_rate": 2.304538639771024e-06, + "loss": 2.490030288696289, + "step": 325080 + }, + { + "epoch": 0.8849870695521982, + "grad_norm": 2.1307060718536377, + "learning_rate": 2.303448275862069e-06, + "loss": 2.4641918182373046, + "step": 325100 + }, + { + "epoch": 0.8850415135429427, + "grad_norm": 2.136586904525757, + "learning_rate": 2.3023579119531146e-06, + "loss": 2.4649890899658202, + "step": 325120 + }, + { + "epoch": 0.8850959575336872, + "grad_norm": 2.1127889156341553, + "learning_rate": 2.30126754804416e-06, + "loss": 2.4646167755126953, + "step": 325140 + }, + { + "epoch": 0.8851504015244317, + "grad_norm": 2.1242835521698, + "learning_rate": 2.3001771841352054e-06, + "loss": 2.4168699264526365, + "step": 325160 + }, + { + "epoch": 0.8852048455151763, + "grad_norm": 2.183683395385742, + "learning_rate": 2.2990868202262507e-06, + "loss": 2.415896987915039, + "step": 325180 + }, + { + "epoch": 0.8852592895059208, + "grad_norm": 2.2130799293518066, + "learning_rate": 2.297996456317296e-06, + "loss": 2.4806549072265627, + "step": 325200 + }, + { + "epoch": 0.8853137334966653, + "grad_norm": 2.1153604984283447, + "learning_rate": 2.2969060924083415e-06, + "loss": 2.4702924728393554, + "step": 325220 + }, + { + "epoch": 0.8853681774874098, + "grad_norm": 2.2372095584869385, + "learning_rate": 2.295815728499387e-06, + "loss": 2.4391719818115236, + "step": 325240 + }, + { + "epoch": 0.8854226214781543, + "grad_norm": 2.2086737155914307, + "learning_rate": 2.2947253645904323e-06, + "loss": 2.438358497619629, + "step": 325260 + }, + { + "epoch": 0.8854770654688988, + "grad_norm": 2.130305051803589, + "learning_rate": 2.2936350006814777e-06, + "loss": 2.4690786361694337, + "step": 325280 + }, + { + "epoch": 0.8855315094596434, + "grad_norm": 2.1767454147338867, + "learning_rate": 2.292544636772523e-06, + "loss": 2.4491371154785155, + "step": 325300 + }, + { + "epoch": 0.8855859534503879, + "grad_norm": 2.159379005432129, + "learning_rate": 2.2914542728635684e-06, + "loss": 2.42294979095459, + "step": 325320 + }, + { + "epoch": 0.8856403974411324, + "grad_norm": 2.207521677017212, + "learning_rate": 2.290363908954614e-06, + "loss": 2.462306785583496, + "step": 325340 + }, + { + "epoch": 0.8856948414318769, + "grad_norm": 2.170405387878418, + "learning_rate": 2.289273545045659e-06, + "loss": 2.4582056045532226, + "step": 325360 + }, + { + "epoch": 0.8857492854226214, + "grad_norm": 2.2439193725585938, + "learning_rate": 2.2881831811367046e-06, + "loss": 2.4632951736450197, + "step": 325380 + }, + { + "epoch": 0.885803729413366, + "grad_norm": 2.210641384124756, + "learning_rate": 2.28709281722775e-06, + "loss": 2.4428462982177734, + "step": 325400 + }, + { + "epoch": 0.8858581734041105, + "grad_norm": 2.273066282272339, + "learning_rate": 2.2860024533187954e-06, + "loss": 2.485540771484375, + "step": 325420 + }, + { + "epoch": 0.885912617394855, + "grad_norm": 2.2893950939178467, + "learning_rate": 2.2849120894098408e-06, + "loss": 2.4455135345458983, + "step": 325440 + }, + { + "epoch": 0.8859670613855996, + "grad_norm": 2.2064599990844727, + "learning_rate": 2.283821725500886e-06, + "loss": 2.4382041931152343, + "step": 325460 + }, + { + "epoch": 0.886021505376344, + "grad_norm": 2.2054362297058105, + "learning_rate": 2.2827313615919315e-06, + "loss": 2.477309799194336, + "step": 325480 + }, + { + "epoch": 0.8860759493670886, + "grad_norm": 2.1558034420013428, + "learning_rate": 2.281640997682977e-06, + "loss": 2.4531862258911135, + "step": 325500 + }, + { + "epoch": 0.8861303933578332, + "grad_norm": 2.156755208969116, + "learning_rate": 2.2805506337740223e-06, + "loss": 2.454728126525879, + "step": 325520 + }, + { + "epoch": 0.8861848373485777, + "grad_norm": 2.157195568084717, + "learning_rate": 2.2794602698650677e-06, + "loss": 2.4950841903686523, + "step": 325540 + }, + { + "epoch": 0.8862392813393222, + "grad_norm": 2.134199380874634, + "learning_rate": 2.278369905956113e-06, + "loss": 2.446722412109375, + "step": 325560 + }, + { + "epoch": 0.8862937253300667, + "grad_norm": 2.1620376110076904, + "learning_rate": 2.2772795420471585e-06, + "loss": 2.4529043197631837, + "step": 325580 + }, + { + "epoch": 0.8863481693208112, + "grad_norm": 2.2310423851013184, + "learning_rate": 2.276189178138204e-06, + "loss": 2.445751953125, + "step": 325600 + }, + { + "epoch": 0.8864026133115558, + "grad_norm": 2.0911643505096436, + "learning_rate": 2.2750988142292492e-06, + "loss": 2.4460403442382814, + "step": 325620 + }, + { + "epoch": 0.8864570573023003, + "grad_norm": 2.078727960586548, + "learning_rate": 2.2740084503202946e-06, + "loss": 2.4713298797607424, + "step": 325640 + }, + { + "epoch": 0.8865115012930448, + "grad_norm": 2.1519548892974854, + "learning_rate": 2.27291808641134e-06, + "loss": 2.4836698532104493, + "step": 325660 + }, + { + "epoch": 0.8865659452837893, + "grad_norm": 2.183272123336792, + "learning_rate": 2.2718277225023854e-06, + "loss": 2.455546188354492, + "step": 325680 + }, + { + "epoch": 0.8866203892745338, + "grad_norm": 2.2373921871185303, + "learning_rate": 2.2707373585934308e-06, + "loss": 2.4514001846313476, + "step": 325700 + }, + { + "epoch": 0.8866748332652783, + "grad_norm": 2.1948513984680176, + "learning_rate": 2.269646994684476e-06, + "loss": 2.421228218078613, + "step": 325720 + }, + { + "epoch": 0.8867292772560229, + "grad_norm": 2.148890256881714, + "learning_rate": 2.2685566307755215e-06, + "loss": 2.4318559646606444, + "step": 325740 + }, + { + "epoch": 0.8867837212467674, + "grad_norm": 2.192594289779663, + "learning_rate": 2.267466266866567e-06, + "loss": 2.4723093032836916, + "step": 325760 + }, + { + "epoch": 0.8868381652375119, + "grad_norm": 2.318793773651123, + "learning_rate": 2.2663759029576123e-06, + "loss": 2.4485809326171877, + "step": 325780 + }, + { + "epoch": 0.8868926092282564, + "grad_norm": 2.0846686363220215, + "learning_rate": 2.2652855390486577e-06, + "loss": 2.472014617919922, + "step": 325800 + }, + { + "epoch": 0.8869470532190009, + "grad_norm": 2.1979448795318604, + "learning_rate": 2.264195175139703e-06, + "loss": 2.4664024353027343, + "step": 325820 + }, + { + "epoch": 0.8870014972097455, + "grad_norm": 2.2608368396759033, + "learning_rate": 2.2631048112307485e-06, + "loss": 2.4339920043945313, + "step": 325840 + }, + { + "epoch": 0.88705594120049, + "grad_norm": 2.1925435066223145, + "learning_rate": 2.262014447321794e-06, + "loss": 2.479869842529297, + "step": 325860 + }, + { + "epoch": 0.8871103851912345, + "grad_norm": 2.138810634613037, + "learning_rate": 2.2609240834128392e-06, + "loss": 2.4754446029663084, + "step": 325880 + }, + { + "epoch": 0.887164829181979, + "grad_norm": 2.1305999755859375, + "learning_rate": 2.2598337195038846e-06, + "loss": 2.4617694854736327, + "step": 325900 + }, + { + "epoch": 0.8872192731727235, + "grad_norm": 2.098050594329834, + "learning_rate": 2.25874335559493e-06, + "loss": 2.466966438293457, + "step": 325920 + }, + { + "epoch": 0.887273717163468, + "grad_norm": 2.205371141433716, + "learning_rate": 2.2576529916859754e-06, + "loss": 2.4695390701293944, + "step": 325940 + }, + { + "epoch": 0.8873281611542126, + "grad_norm": 2.158071279525757, + "learning_rate": 2.2565626277770208e-06, + "loss": 2.454389190673828, + "step": 325960 + }, + { + "epoch": 0.8873826051449571, + "grad_norm": 2.35225510597229, + "learning_rate": 2.255472263868066e-06, + "loss": 2.4403507232666017, + "step": 325980 + }, + { + "epoch": 0.8874370491357016, + "grad_norm": 2.0836904048919678, + "learning_rate": 2.2543818999591116e-06, + "loss": 2.46983642578125, + "step": 326000 + }, + { + "epoch": 0.8874914931264462, + "grad_norm": 2.1617813110351562, + "learning_rate": 2.253291536050157e-06, + "loss": 2.4441232681274414, + "step": 326020 + }, + { + "epoch": 0.8875459371171907, + "grad_norm": 2.204765796661377, + "learning_rate": 2.2522011721412023e-06, + "loss": 2.4258148193359377, + "step": 326040 + }, + { + "epoch": 0.8876003811079353, + "grad_norm": 2.339221239089966, + "learning_rate": 2.2511108082322477e-06, + "loss": 2.4731239318847655, + "step": 326060 + }, + { + "epoch": 0.8876548250986798, + "grad_norm": 2.1568081378936768, + "learning_rate": 2.250020444323293e-06, + "loss": 2.451967239379883, + "step": 326080 + }, + { + "epoch": 0.8877092690894243, + "grad_norm": 2.1101224422454834, + "learning_rate": 2.2489300804143385e-06, + "loss": 2.464212417602539, + "step": 326100 + }, + { + "epoch": 0.8877637130801688, + "grad_norm": 2.1915194988250732, + "learning_rate": 2.247839716505384e-06, + "loss": 2.471063995361328, + "step": 326120 + }, + { + "epoch": 0.8878181570709133, + "grad_norm": 2.157841444015503, + "learning_rate": 2.2467493525964293e-06, + "loss": 2.4403003692626952, + "step": 326140 + }, + { + "epoch": 0.8878726010616578, + "grad_norm": 2.2618489265441895, + "learning_rate": 2.2456589886874746e-06, + "loss": 2.451036834716797, + "step": 326160 + }, + { + "epoch": 0.8879270450524024, + "grad_norm": 2.112720012664795, + "learning_rate": 2.24456862477852e-06, + "loss": 2.475986099243164, + "step": 326180 + }, + { + "epoch": 0.8879814890431469, + "grad_norm": 2.4844753742218018, + "learning_rate": 2.2434782608695654e-06, + "loss": 2.4602405548095705, + "step": 326200 + }, + { + "epoch": 0.8880359330338914, + "grad_norm": 2.1540462970733643, + "learning_rate": 2.242387896960611e-06, + "loss": 2.486110305786133, + "step": 326220 + }, + { + "epoch": 0.8880903770246359, + "grad_norm": 2.0979206562042236, + "learning_rate": 2.241297533051656e-06, + "loss": 2.47955265045166, + "step": 326240 + }, + { + "epoch": 0.8881448210153804, + "grad_norm": 2.2762277126312256, + "learning_rate": 2.2402071691427016e-06, + "loss": 2.4623991012573243, + "step": 326260 + }, + { + "epoch": 0.8881992650061249, + "grad_norm": 2.2479088306427, + "learning_rate": 2.239116805233747e-06, + "loss": 2.4531164169311523, + "step": 326280 + }, + { + "epoch": 0.8882537089968695, + "grad_norm": 2.179372787475586, + "learning_rate": 2.2380264413247923e-06, + "loss": 2.461906623840332, + "step": 326300 + }, + { + "epoch": 0.888308152987614, + "grad_norm": 2.186039924621582, + "learning_rate": 2.2369360774158377e-06, + "loss": 2.4689075469970705, + "step": 326320 + }, + { + "epoch": 0.8883625969783585, + "grad_norm": 2.2049942016601562, + "learning_rate": 2.235845713506883e-06, + "loss": 2.453384590148926, + "step": 326340 + }, + { + "epoch": 0.888417040969103, + "grad_norm": 2.270679473876953, + "learning_rate": 2.2347553495979285e-06, + "loss": 2.465496635437012, + "step": 326360 + }, + { + "epoch": 0.8884714849598475, + "grad_norm": 2.229444980621338, + "learning_rate": 2.233664985688974e-06, + "loss": 2.4147838592529296, + "step": 326380 + }, + { + "epoch": 0.8885259289505921, + "grad_norm": 2.147355079650879, + "learning_rate": 2.2325746217800193e-06, + "loss": 2.467232322692871, + "step": 326400 + }, + { + "epoch": 0.8885803729413366, + "grad_norm": 2.167787551879883, + "learning_rate": 2.2314842578710647e-06, + "loss": 2.4597145080566407, + "step": 326420 + }, + { + "epoch": 0.8886348169320811, + "grad_norm": 2.179722309112549, + "learning_rate": 2.23039389396211e-06, + "loss": 2.4616157531738283, + "step": 326440 + }, + { + "epoch": 0.8886892609228256, + "grad_norm": 2.6410906314849854, + "learning_rate": 2.2293035300531554e-06, + "loss": 2.4431488037109377, + "step": 326460 + }, + { + "epoch": 0.8887437049135701, + "grad_norm": 2.126833915710449, + "learning_rate": 2.228213166144201e-06, + "loss": 2.4664274215698243, + "step": 326480 + }, + { + "epoch": 0.8887981489043146, + "grad_norm": 2.2664296627044678, + "learning_rate": 2.227122802235246e-06, + "loss": 2.4497825622558596, + "step": 326500 + }, + { + "epoch": 0.8888525928950592, + "grad_norm": 2.593071460723877, + "learning_rate": 2.2260324383262916e-06, + "loss": 2.4712514877319336, + "step": 326520 + }, + { + "epoch": 0.8889070368858037, + "grad_norm": 2.21513295173645, + "learning_rate": 2.224942074417337e-06, + "loss": 2.4323667526245116, + "step": 326540 + }, + { + "epoch": 0.8889614808765482, + "grad_norm": 2.256413698196411, + "learning_rate": 2.2238517105083824e-06, + "loss": 2.4600563049316406, + "step": 326560 + }, + { + "epoch": 0.8890159248672928, + "grad_norm": 2.1840531826019287, + "learning_rate": 2.2227613465994277e-06, + "loss": 2.468799018859863, + "step": 326580 + }, + { + "epoch": 0.8890703688580373, + "grad_norm": 2.3308308124542236, + "learning_rate": 2.221670982690473e-06, + "loss": 2.43118896484375, + "step": 326600 + }, + { + "epoch": 0.8891248128487819, + "grad_norm": 2.1046080589294434, + "learning_rate": 2.2205806187815185e-06, + "loss": 2.456951904296875, + "step": 326620 + }, + { + "epoch": 0.8891792568395264, + "grad_norm": 2.104435920715332, + "learning_rate": 2.219490254872564e-06, + "loss": 2.4222490310668947, + "step": 326640 + }, + { + "epoch": 0.8892337008302709, + "grad_norm": 2.159008741378784, + "learning_rate": 2.2183998909636093e-06, + "loss": 2.4383007049560548, + "step": 326660 + }, + { + "epoch": 0.8892881448210154, + "grad_norm": 2.2859177589416504, + "learning_rate": 2.2173095270546547e-06, + "loss": 2.450821876525879, + "step": 326680 + }, + { + "epoch": 0.8893425888117599, + "grad_norm": 2.18314790725708, + "learning_rate": 2.2162191631457e-06, + "loss": 2.4660839080810546, + "step": 326700 + }, + { + "epoch": 0.8893970328025044, + "grad_norm": 2.0946857929229736, + "learning_rate": 2.2151287992367454e-06, + "loss": 2.436962127685547, + "step": 326720 + }, + { + "epoch": 0.889451476793249, + "grad_norm": 2.268111228942871, + "learning_rate": 2.214038435327791e-06, + "loss": 2.46466007232666, + "step": 326740 + }, + { + "epoch": 0.8895059207839935, + "grad_norm": 2.259904623031616, + "learning_rate": 2.212948071418836e-06, + "loss": 2.4408321380615234, + "step": 326760 + }, + { + "epoch": 0.889560364774738, + "grad_norm": 2.170428514480591, + "learning_rate": 2.2118577075098816e-06, + "loss": 2.4558618545532225, + "step": 326780 + }, + { + "epoch": 0.8896148087654825, + "grad_norm": 2.1729626655578613, + "learning_rate": 2.210767343600927e-06, + "loss": 2.440708541870117, + "step": 326800 + }, + { + "epoch": 0.889669252756227, + "grad_norm": 2.895301580429077, + "learning_rate": 2.2096769796919724e-06, + "loss": 2.437582778930664, + "step": 326820 + }, + { + "epoch": 0.8897236967469716, + "grad_norm": 2.2671921253204346, + "learning_rate": 2.2085866157830178e-06, + "loss": 2.4519346237182615, + "step": 326840 + }, + { + "epoch": 0.8897781407377161, + "grad_norm": 2.318237543106079, + "learning_rate": 2.207496251874063e-06, + "loss": 2.4808107376098634, + "step": 326860 + }, + { + "epoch": 0.8898325847284606, + "grad_norm": 2.2401578426361084, + "learning_rate": 2.2064058879651085e-06, + "loss": 2.428093910217285, + "step": 326880 + }, + { + "epoch": 0.8898870287192051, + "grad_norm": 2.1490190029144287, + "learning_rate": 2.205315524056154e-06, + "loss": 2.419697570800781, + "step": 326900 + }, + { + "epoch": 0.8899414727099496, + "grad_norm": 2.2244091033935547, + "learning_rate": 2.2042251601471993e-06, + "loss": 2.4449262619018555, + "step": 326920 + }, + { + "epoch": 0.8899959167006941, + "grad_norm": 2.093366861343384, + "learning_rate": 2.2031347962382447e-06, + "loss": 2.4718229293823244, + "step": 326940 + }, + { + "epoch": 0.8900503606914387, + "grad_norm": 2.090256452560425, + "learning_rate": 2.20204443232929e-06, + "loss": 2.4513877868652343, + "step": 326960 + }, + { + "epoch": 0.8901048046821832, + "grad_norm": 2.1963460445404053, + "learning_rate": 2.2009540684203355e-06, + "loss": 2.476681137084961, + "step": 326980 + }, + { + "epoch": 0.8901592486729277, + "grad_norm": 2.2833924293518066, + "learning_rate": 2.199863704511381e-06, + "loss": 2.435977554321289, + "step": 327000 + }, + { + "epoch": 0.8902136926636722, + "grad_norm": 2.1224594116210938, + "learning_rate": 2.1987733406024262e-06, + "loss": 2.476375389099121, + "step": 327020 + }, + { + "epoch": 0.8902681366544167, + "grad_norm": 2.187490463256836, + "learning_rate": 2.1976829766934716e-06, + "loss": 2.4715063095092775, + "step": 327040 + }, + { + "epoch": 0.8903225806451613, + "grad_norm": 2.5064189434051514, + "learning_rate": 2.196592612784517e-06, + "loss": 2.4472545623779296, + "step": 327060 + }, + { + "epoch": 0.8903770246359058, + "grad_norm": 2.4884536266326904, + "learning_rate": 2.1955022488755624e-06, + "loss": 2.4448556900024414, + "step": 327080 + }, + { + "epoch": 0.8904314686266503, + "grad_norm": 2.096653461456299, + "learning_rate": 2.1944118849666078e-06, + "loss": 2.4751211166381837, + "step": 327100 + }, + { + "epoch": 0.8904859126173948, + "grad_norm": 2.189302682876587, + "learning_rate": 2.193321521057653e-06, + "loss": 2.4439538955688476, + "step": 327120 + }, + { + "epoch": 0.8905403566081393, + "grad_norm": 2.659914493560791, + "learning_rate": 2.1922311571486985e-06, + "loss": 2.4703420639038085, + "step": 327140 + }, + { + "epoch": 0.8905948005988839, + "grad_norm": 2.1153769493103027, + "learning_rate": 2.191140793239744e-06, + "loss": 2.486583709716797, + "step": 327160 + }, + { + "epoch": 0.8906492445896285, + "grad_norm": 10.15243911743164, + "learning_rate": 2.1900504293307893e-06, + "loss": 2.4814115524291993, + "step": 327180 + }, + { + "epoch": 0.890703688580373, + "grad_norm": 2.191159725189209, + "learning_rate": 2.1889600654218347e-06, + "loss": 2.4798622131347656, + "step": 327200 + }, + { + "epoch": 0.8907581325711175, + "grad_norm": 2.121642589569092, + "learning_rate": 2.18786970151288e-06, + "loss": 2.4564849853515627, + "step": 327220 + }, + { + "epoch": 0.890812576561862, + "grad_norm": 2.1470465660095215, + "learning_rate": 2.1867793376039255e-06, + "loss": 2.456532859802246, + "step": 327240 + }, + { + "epoch": 0.8908670205526065, + "grad_norm": 2.111433744430542, + "learning_rate": 2.185688973694971e-06, + "loss": 2.4491193771362303, + "step": 327260 + }, + { + "epoch": 0.8909214645433511, + "grad_norm": 2.2333922386169434, + "learning_rate": 2.1845986097860162e-06, + "loss": 2.4479751586914062, + "step": 327280 + }, + { + "epoch": 0.8909759085340956, + "grad_norm": 2.233583450317383, + "learning_rate": 2.1835082458770616e-06, + "loss": 2.5038000106811524, + "step": 327300 + }, + { + "epoch": 0.8910303525248401, + "grad_norm": 2.1661758422851562, + "learning_rate": 2.182417881968107e-06, + "loss": 2.4367002487182616, + "step": 327320 + }, + { + "epoch": 0.8910847965155846, + "grad_norm": 2.209606409072876, + "learning_rate": 2.1813275180591524e-06, + "loss": 2.4749500274658205, + "step": 327340 + }, + { + "epoch": 0.8911392405063291, + "grad_norm": 2.0881009101867676, + "learning_rate": 2.1802371541501978e-06, + "loss": 2.423758888244629, + "step": 327360 + }, + { + "epoch": 0.8911936844970736, + "grad_norm": 2.8374483585357666, + "learning_rate": 2.179146790241243e-06, + "loss": 2.4605327606201173, + "step": 327380 + }, + { + "epoch": 0.8912481284878182, + "grad_norm": 2.327759265899658, + "learning_rate": 2.1780564263322886e-06, + "loss": 2.46280460357666, + "step": 327400 + }, + { + "epoch": 0.8913025724785627, + "grad_norm": 2.2130420207977295, + "learning_rate": 2.176966062423334e-06, + "loss": 2.4597213745117186, + "step": 327420 + }, + { + "epoch": 0.8913570164693072, + "grad_norm": 2.6320314407348633, + "learning_rate": 2.1758756985143793e-06, + "loss": 2.4579082489013673, + "step": 327440 + }, + { + "epoch": 0.8914114604600517, + "grad_norm": 2.149237871170044, + "learning_rate": 2.1747853346054247e-06, + "loss": 2.442211723327637, + "step": 327460 + }, + { + "epoch": 0.8914659044507962, + "grad_norm": 3.8841047286987305, + "learning_rate": 2.17369497069647e-06, + "loss": 2.4737497329711915, + "step": 327480 + }, + { + "epoch": 0.8915203484415407, + "grad_norm": 2.1160809993743896, + "learning_rate": 2.1726046067875155e-06, + "loss": 2.4738491058349608, + "step": 327500 + }, + { + "epoch": 0.8915747924322853, + "grad_norm": 2.12872576713562, + "learning_rate": 2.171514242878561e-06, + "loss": 2.457752227783203, + "step": 327520 + }, + { + "epoch": 0.8916292364230298, + "grad_norm": 2.12471079826355, + "learning_rate": 2.1704238789696063e-06, + "loss": 2.451456069946289, + "step": 327540 + }, + { + "epoch": 0.8916836804137743, + "grad_norm": 2.1802477836608887, + "learning_rate": 2.1693335150606516e-06, + "loss": 2.462839889526367, + "step": 327560 + }, + { + "epoch": 0.8917381244045188, + "grad_norm": 2.2571818828582764, + "learning_rate": 2.168243151151697e-06, + "loss": 2.462339973449707, + "step": 327580 + }, + { + "epoch": 0.8917925683952633, + "grad_norm": 2.1728668212890625, + "learning_rate": 2.1671527872427424e-06, + "loss": 2.4534372329711913, + "step": 327600 + }, + { + "epoch": 0.8918470123860079, + "grad_norm": 2.1001832485198975, + "learning_rate": 2.166062423333788e-06, + "loss": 2.435433578491211, + "step": 327620 + }, + { + "epoch": 0.8919014563767524, + "grad_norm": 2.134958267211914, + "learning_rate": 2.164972059424833e-06, + "loss": 2.4433116912841797, + "step": 327640 + }, + { + "epoch": 0.891955900367497, + "grad_norm": 2.1220574378967285, + "learning_rate": 2.1638816955158786e-06, + "loss": 2.4482406616210937, + "step": 327660 + }, + { + "epoch": 0.8920103443582414, + "grad_norm": 2.12331223487854, + "learning_rate": 2.162791331606924e-06, + "loss": 2.4630029678344725, + "step": 327680 + }, + { + "epoch": 0.892064788348986, + "grad_norm": 2.1716599464416504, + "learning_rate": 2.1617009676979693e-06, + "loss": 2.4352439880371093, + "step": 327700 + }, + { + "epoch": 0.8921192323397305, + "grad_norm": 2.2277138233184814, + "learning_rate": 2.1606106037890147e-06, + "loss": 2.459020805358887, + "step": 327720 + }, + { + "epoch": 0.8921736763304751, + "grad_norm": 2.312821388244629, + "learning_rate": 2.15952023988006e-06, + "loss": 2.4279573440551756, + "step": 327740 + }, + { + "epoch": 0.8922281203212196, + "grad_norm": 2.133018970489502, + "learning_rate": 2.1584298759711055e-06, + "loss": 2.483248710632324, + "step": 327760 + }, + { + "epoch": 0.8922825643119641, + "grad_norm": 2.142096757888794, + "learning_rate": 2.157339512062151e-06, + "loss": 2.4670106887817385, + "step": 327780 + }, + { + "epoch": 0.8923370083027086, + "grad_norm": 2.187185525894165, + "learning_rate": 2.1562491481531963e-06, + "loss": 2.453533935546875, + "step": 327800 + }, + { + "epoch": 0.8923914522934531, + "grad_norm": 2.2186710834503174, + "learning_rate": 2.1551587842442417e-06, + "loss": 2.4448394775390625, + "step": 327820 + }, + { + "epoch": 0.8924458962841977, + "grad_norm": 3.6118063926696777, + "learning_rate": 2.154068420335287e-06, + "loss": 2.446792221069336, + "step": 327840 + }, + { + "epoch": 0.8925003402749422, + "grad_norm": 2.075739860534668, + "learning_rate": 2.1529780564263324e-06, + "loss": 2.480371856689453, + "step": 327860 + }, + { + "epoch": 0.8925547842656867, + "grad_norm": 2.186591386795044, + "learning_rate": 2.151887692517378e-06, + "loss": 2.4684642791748046, + "step": 327880 + }, + { + "epoch": 0.8926092282564312, + "grad_norm": 2.3207411766052246, + "learning_rate": 2.150797328608423e-06, + "loss": 2.4283111572265623, + "step": 327900 + }, + { + "epoch": 0.8926636722471757, + "grad_norm": 2.6488873958587646, + "learning_rate": 2.1497069646994686e-06, + "loss": 2.4653417587280275, + "step": 327920 + }, + { + "epoch": 0.8927181162379202, + "grad_norm": 2.4355528354644775, + "learning_rate": 2.148616600790514e-06, + "loss": 2.4537380218505858, + "step": 327940 + }, + { + "epoch": 0.8927725602286648, + "grad_norm": 2.0949954986572266, + "learning_rate": 2.1475262368815594e-06, + "loss": 2.433603858947754, + "step": 327960 + }, + { + "epoch": 0.8928270042194093, + "grad_norm": 2.169740676879883, + "learning_rate": 2.1464358729726047e-06, + "loss": 2.446173667907715, + "step": 327980 + }, + { + "epoch": 0.8928814482101538, + "grad_norm": 2.9796645641326904, + "learning_rate": 2.14534550906365e-06, + "loss": 2.4760921478271483, + "step": 328000 + }, + { + "epoch": 0.8929358922008983, + "grad_norm": 2.26124906539917, + "learning_rate": 2.1442551451546955e-06, + "loss": 2.459187316894531, + "step": 328020 + }, + { + "epoch": 0.8929903361916428, + "grad_norm": 2.271294355392456, + "learning_rate": 2.143164781245741e-06, + "loss": 2.489370346069336, + "step": 328040 + }, + { + "epoch": 0.8930447801823874, + "grad_norm": 2.1100823879241943, + "learning_rate": 2.1420744173367863e-06, + "loss": 2.4836606979370117, + "step": 328060 + }, + { + "epoch": 0.8930992241731319, + "grad_norm": 2.318607807159424, + "learning_rate": 2.1409840534278317e-06, + "loss": 2.4447395324707033, + "step": 328080 + }, + { + "epoch": 0.8931536681638764, + "grad_norm": 2.255452871322632, + "learning_rate": 2.139893689518877e-06, + "loss": 2.469495964050293, + "step": 328100 + }, + { + "epoch": 0.8932081121546209, + "grad_norm": 2.5907342433929443, + "learning_rate": 2.1388033256099224e-06, + "loss": 2.4514373779296874, + "step": 328120 + }, + { + "epoch": 0.8932625561453654, + "grad_norm": 2.1962056159973145, + "learning_rate": 2.137712961700968e-06, + "loss": 2.468372917175293, + "step": 328140 + }, + { + "epoch": 0.8933170001361099, + "grad_norm": 2.066338539123535, + "learning_rate": 2.136622597792013e-06, + "loss": 2.4159875869750977, + "step": 328160 + }, + { + "epoch": 0.8933714441268545, + "grad_norm": 2.1663572788238525, + "learning_rate": 2.1355322338830586e-06, + "loss": 2.444746208190918, + "step": 328180 + }, + { + "epoch": 0.893425888117599, + "grad_norm": 2.1537985801696777, + "learning_rate": 2.134441869974104e-06, + "loss": 2.4763341903686524, + "step": 328200 + }, + { + "epoch": 0.8934803321083435, + "grad_norm": 2.121487617492676, + "learning_rate": 2.1333515060651494e-06, + "loss": 2.448764610290527, + "step": 328220 + }, + { + "epoch": 0.893534776099088, + "grad_norm": 2.7046568393707275, + "learning_rate": 2.1322611421561948e-06, + "loss": 2.4493614196777345, + "step": 328240 + }, + { + "epoch": 0.8935892200898325, + "grad_norm": 2.153538942337036, + "learning_rate": 2.13117077824724e-06, + "loss": 2.4577203750610352, + "step": 328260 + }, + { + "epoch": 0.8936436640805772, + "grad_norm": 2.3617115020751953, + "learning_rate": 2.1300804143382855e-06, + "loss": 2.4688627243041994, + "step": 328280 + }, + { + "epoch": 0.8936981080713217, + "grad_norm": 2.0757875442504883, + "learning_rate": 2.128990050429331e-06, + "loss": 2.439555549621582, + "step": 328300 + }, + { + "epoch": 0.8937525520620662, + "grad_norm": 2.173870801925659, + "learning_rate": 2.1278996865203763e-06, + "loss": 2.4615453720092773, + "step": 328320 + }, + { + "epoch": 0.8938069960528107, + "grad_norm": 2.1054177284240723, + "learning_rate": 2.1268093226114217e-06, + "loss": 2.479058265686035, + "step": 328340 + }, + { + "epoch": 0.8938614400435552, + "grad_norm": 2.121140480041504, + "learning_rate": 2.125718958702467e-06, + "loss": 2.4782693862915037, + "step": 328360 + }, + { + "epoch": 0.8939158840342997, + "grad_norm": 2.254413366317749, + "learning_rate": 2.1246285947935125e-06, + "loss": 2.456187438964844, + "step": 328380 + }, + { + "epoch": 0.8939703280250443, + "grad_norm": 2.174229383468628, + "learning_rate": 2.123538230884558e-06, + "loss": 2.4552087783813477, + "step": 328400 + }, + { + "epoch": 0.8940247720157888, + "grad_norm": 2.2403762340545654, + "learning_rate": 2.1224478669756032e-06, + "loss": 2.451803779602051, + "step": 328420 + }, + { + "epoch": 0.8940792160065333, + "grad_norm": 2.1561341285705566, + "learning_rate": 2.1213575030666486e-06, + "loss": 2.391636276245117, + "step": 328440 + }, + { + "epoch": 0.8941336599972778, + "grad_norm": 2.2656147480010986, + "learning_rate": 2.120267139157694e-06, + "loss": 2.464099884033203, + "step": 328460 + }, + { + "epoch": 0.8941881039880223, + "grad_norm": 2.286954641342163, + "learning_rate": 2.1191767752487394e-06, + "loss": 2.462222671508789, + "step": 328480 + }, + { + "epoch": 0.8942425479787668, + "grad_norm": 2.168168306350708, + "learning_rate": 2.1180864113397848e-06, + "loss": 2.4468122482299806, + "step": 328500 + }, + { + "epoch": 0.8942969919695114, + "grad_norm": 2.150639772415161, + "learning_rate": 2.11699604743083e-06, + "loss": 2.453501319885254, + "step": 328520 + }, + { + "epoch": 0.8943514359602559, + "grad_norm": 2.2113378047943115, + "learning_rate": 2.1159056835218755e-06, + "loss": 2.458517837524414, + "step": 328540 + }, + { + "epoch": 0.8944058799510004, + "grad_norm": 2.157959461212158, + "learning_rate": 2.114815319612921e-06, + "loss": 2.4670503616333006, + "step": 328560 + }, + { + "epoch": 0.8944603239417449, + "grad_norm": 2.1595773696899414, + "learning_rate": 2.1137249557039663e-06, + "loss": 2.4636466979980467, + "step": 328580 + }, + { + "epoch": 0.8945147679324894, + "grad_norm": 2.2701826095581055, + "learning_rate": 2.1126345917950117e-06, + "loss": 2.4667530059814453, + "step": 328600 + }, + { + "epoch": 0.894569211923234, + "grad_norm": 2.1594526767730713, + "learning_rate": 2.111544227886057e-06, + "loss": 2.473455047607422, + "step": 328620 + }, + { + "epoch": 0.8946236559139785, + "grad_norm": 2.184509515762329, + "learning_rate": 2.1104538639771025e-06, + "loss": 2.4877292633056642, + "step": 328640 + }, + { + "epoch": 0.894678099904723, + "grad_norm": 2.2458579540252686, + "learning_rate": 2.109363500068148e-06, + "loss": 2.4522579193115233, + "step": 328660 + }, + { + "epoch": 0.8947325438954675, + "grad_norm": 2.1560022830963135, + "learning_rate": 2.1082731361591932e-06, + "loss": 2.44058837890625, + "step": 328680 + }, + { + "epoch": 0.894786987886212, + "grad_norm": 2.1861393451690674, + "learning_rate": 2.1071827722502386e-06, + "loss": 2.4965063095092774, + "step": 328700 + }, + { + "epoch": 0.8948414318769565, + "grad_norm": 2.3138515949249268, + "learning_rate": 2.106092408341284e-06, + "loss": 2.478641700744629, + "step": 328720 + }, + { + "epoch": 0.8948958758677011, + "grad_norm": 2.3478214740753174, + "learning_rate": 2.1050020444323294e-06, + "loss": 2.502108573913574, + "step": 328740 + }, + { + "epoch": 0.8949503198584456, + "grad_norm": 2.221078395843506, + "learning_rate": 2.1039116805233748e-06, + "loss": 2.479274559020996, + "step": 328760 + }, + { + "epoch": 0.8950047638491901, + "grad_norm": 2.2801473140716553, + "learning_rate": 2.10282131661442e-06, + "loss": 2.4440586090087892, + "step": 328780 + }, + { + "epoch": 0.8950592078399346, + "grad_norm": 2.259838104248047, + "learning_rate": 2.1017309527054656e-06, + "loss": 2.445511054992676, + "step": 328800 + }, + { + "epoch": 0.8951136518306791, + "grad_norm": 2.361400842666626, + "learning_rate": 2.100640588796511e-06, + "loss": 2.471160125732422, + "step": 328820 + }, + { + "epoch": 0.8951680958214238, + "grad_norm": 2.2793660163879395, + "learning_rate": 2.0995502248875563e-06, + "loss": 2.467310905456543, + "step": 328840 + }, + { + "epoch": 0.8952225398121683, + "grad_norm": 2.2522008419036865, + "learning_rate": 2.0984598609786017e-06, + "loss": 2.4705543518066406, + "step": 328860 + }, + { + "epoch": 0.8952769838029128, + "grad_norm": 2.244114875793457, + "learning_rate": 2.097369497069647e-06, + "loss": 2.4512989044189455, + "step": 328880 + }, + { + "epoch": 0.8953314277936573, + "grad_norm": 2.2827000617980957, + "learning_rate": 2.0962791331606925e-06, + "loss": 2.434099006652832, + "step": 328900 + }, + { + "epoch": 0.8953858717844018, + "grad_norm": 2.2772910594940186, + "learning_rate": 2.095188769251738e-06, + "loss": 2.4562620162963866, + "step": 328920 + }, + { + "epoch": 0.8954403157751463, + "grad_norm": 2.410257339477539, + "learning_rate": 2.0940984053427833e-06, + "loss": 2.4953004837036135, + "step": 328940 + }, + { + "epoch": 0.8954947597658909, + "grad_norm": 2.1401312351226807, + "learning_rate": 2.0930080414338286e-06, + "loss": 2.4472558975219725, + "step": 328960 + }, + { + "epoch": 0.8955492037566354, + "grad_norm": 2.201925039291382, + "learning_rate": 2.0919176775248744e-06, + "loss": 2.4440439224243162, + "step": 328980 + }, + { + "epoch": 0.8956036477473799, + "grad_norm": 2.203538417816162, + "learning_rate": 2.0908273136159194e-06, + "loss": 2.4568580627441405, + "step": 329000 + }, + { + "epoch": 0.8956580917381244, + "grad_norm": 2.257260799407959, + "learning_rate": 2.089736949706965e-06, + "loss": 2.4767383575439452, + "step": 329020 + }, + { + "epoch": 0.8957125357288689, + "grad_norm": 2.1139142513275146, + "learning_rate": 2.08864658579801e-06, + "loss": 2.471129035949707, + "step": 329040 + }, + { + "epoch": 0.8957669797196135, + "grad_norm": 2.17061710357666, + "learning_rate": 2.0875562218890556e-06, + "loss": 2.455923652648926, + "step": 329060 + }, + { + "epoch": 0.895821423710358, + "grad_norm": 2.1180357933044434, + "learning_rate": 2.086465857980101e-06, + "loss": 2.466852569580078, + "step": 329080 + }, + { + "epoch": 0.8958758677011025, + "grad_norm": 2.284637928009033, + "learning_rate": 2.0853754940711463e-06, + "loss": 2.439505386352539, + "step": 329100 + }, + { + "epoch": 0.895930311691847, + "grad_norm": 2.1442067623138428, + "learning_rate": 2.0842851301621917e-06, + "loss": 2.446559715270996, + "step": 329120 + }, + { + "epoch": 0.8959847556825915, + "grad_norm": 2.117913007736206, + "learning_rate": 2.083194766253237e-06, + "loss": 2.454502296447754, + "step": 329140 + }, + { + "epoch": 0.896039199673336, + "grad_norm": 2.2071533203125, + "learning_rate": 2.0821044023442825e-06, + "loss": 2.474490165710449, + "step": 329160 + }, + { + "epoch": 0.8960936436640806, + "grad_norm": 2.1685447692871094, + "learning_rate": 2.081014038435328e-06, + "loss": 2.4550312042236326, + "step": 329180 + }, + { + "epoch": 0.8961480876548251, + "grad_norm": 2.1409270763397217, + "learning_rate": 2.0799236745263733e-06, + "loss": 2.4487390518188477, + "step": 329200 + }, + { + "epoch": 0.8962025316455696, + "grad_norm": 5.355565547943115, + "learning_rate": 2.0788333106174187e-06, + "loss": 2.4344154357910157, + "step": 329220 + }, + { + "epoch": 0.8962569756363141, + "grad_norm": 2.180236577987671, + "learning_rate": 2.077742946708464e-06, + "loss": 2.4919292449951174, + "step": 329240 + }, + { + "epoch": 0.8963114196270586, + "grad_norm": 2.396418809890747, + "learning_rate": 2.0766525827995094e-06, + "loss": 2.4415876388549806, + "step": 329260 + }, + { + "epoch": 0.8963658636178032, + "grad_norm": 2.2048263549804688, + "learning_rate": 2.075562218890555e-06, + "loss": 2.4293792724609373, + "step": 329280 + }, + { + "epoch": 0.8964203076085477, + "grad_norm": 2.213222026824951, + "learning_rate": 2.0744718549816e-06, + "loss": 2.469288635253906, + "step": 329300 + }, + { + "epoch": 0.8964747515992922, + "grad_norm": 2.527961492538452, + "learning_rate": 2.0733814910726456e-06, + "loss": 2.4741350173950196, + "step": 329320 + }, + { + "epoch": 0.8965291955900367, + "grad_norm": 2.145200252532959, + "learning_rate": 2.072291127163691e-06, + "loss": 2.425510597229004, + "step": 329340 + }, + { + "epoch": 0.8965836395807812, + "grad_norm": 2.203207492828369, + "learning_rate": 2.0712007632547364e-06, + "loss": 2.450243377685547, + "step": 329360 + }, + { + "epoch": 0.8966380835715257, + "grad_norm": 2.2271502017974854, + "learning_rate": 2.0701103993457817e-06, + "loss": 2.456206703186035, + "step": 329380 + }, + { + "epoch": 0.8966925275622704, + "grad_norm": 2.211130142211914, + "learning_rate": 2.069020035436827e-06, + "loss": 2.4651527404785156, + "step": 329400 + }, + { + "epoch": 0.8967469715530149, + "grad_norm": 2.1141140460968018, + "learning_rate": 2.0679296715278725e-06, + "loss": 2.490657424926758, + "step": 329420 + }, + { + "epoch": 0.8968014155437594, + "grad_norm": 2.3052380084991455, + "learning_rate": 2.066839307618918e-06, + "loss": 2.430864143371582, + "step": 329440 + }, + { + "epoch": 0.8968558595345039, + "grad_norm": 2.2147233486175537, + "learning_rate": 2.0657489437099633e-06, + "loss": 2.4301889419555662, + "step": 329460 + }, + { + "epoch": 0.8969103035252484, + "grad_norm": 2.2228548526763916, + "learning_rate": 2.0646585798010087e-06, + "loss": 2.455720329284668, + "step": 329480 + }, + { + "epoch": 0.896964747515993, + "grad_norm": 2.179882764816284, + "learning_rate": 2.063568215892054e-06, + "loss": 2.4930084228515623, + "step": 329500 + }, + { + "epoch": 0.8970191915067375, + "grad_norm": 2.2933924198150635, + "learning_rate": 2.0624778519830994e-06, + "loss": 2.4599536895751952, + "step": 329520 + }, + { + "epoch": 0.897073635497482, + "grad_norm": 2.111830472946167, + "learning_rate": 2.061387488074145e-06, + "loss": 2.4728736877441406, + "step": 329540 + }, + { + "epoch": 0.8971280794882265, + "grad_norm": 2.163421869277954, + "learning_rate": 2.06029712416519e-06, + "loss": 2.4607397079467774, + "step": 329560 + }, + { + "epoch": 0.897182523478971, + "grad_norm": 3.402698278427124, + "learning_rate": 2.0592067602562356e-06, + "loss": 2.4455204010009766, + "step": 329580 + }, + { + "epoch": 0.8972369674697155, + "grad_norm": 2.150768756866455, + "learning_rate": 2.058116396347281e-06, + "loss": 2.4672740936279296, + "step": 329600 + }, + { + "epoch": 0.8972914114604601, + "grad_norm": 2.1370198726654053, + "learning_rate": 2.0570260324383264e-06, + "loss": 2.478487014770508, + "step": 329620 + }, + { + "epoch": 0.8973458554512046, + "grad_norm": 2.1955723762512207, + "learning_rate": 2.0559356685293718e-06, + "loss": 2.4376958847045898, + "step": 329640 + }, + { + "epoch": 0.8974002994419491, + "grad_norm": 2.2506935596466064, + "learning_rate": 2.054845304620417e-06, + "loss": 2.472940444946289, + "step": 329660 + }, + { + "epoch": 0.8974547434326936, + "grad_norm": 2.184413194656372, + "learning_rate": 2.0537549407114625e-06, + "loss": 2.470047187805176, + "step": 329680 + }, + { + "epoch": 0.8975091874234381, + "grad_norm": 2.089059352874756, + "learning_rate": 2.052664576802508e-06, + "loss": 2.45018310546875, + "step": 329700 + }, + { + "epoch": 0.8975636314141826, + "grad_norm": 2.1449482440948486, + "learning_rate": 2.0515742128935533e-06, + "loss": 2.4708505630493165, + "step": 329720 + }, + { + "epoch": 0.8976180754049272, + "grad_norm": 2.2827937602996826, + "learning_rate": 2.0504838489845987e-06, + "loss": 2.433520698547363, + "step": 329740 + }, + { + "epoch": 0.8976725193956717, + "grad_norm": 2.0888662338256836, + "learning_rate": 2.049393485075644e-06, + "loss": 2.4451309204101563, + "step": 329760 + }, + { + "epoch": 0.8977269633864162, + "grad_norm": 2.183907985687256, + "learning_rate": 2.0483031211666895e-06, + "loss": 2.4709407806396486, + "step": 329780 + }, + { + "epoch": 0.8977814073771607, + "grad_norm": 2.2843761444091797, + "learning_rate": 2.047212757257735e-06, + "loss": 2.430804634094238, + "step": 329800 + }, + { + "epoch": 0.8978358513679052, + "grad_norm": 2.2664315700531006, + "learning_rate": 2.0461223933487802e-06, + "loss": 2.4462749481201174, + "step": 329820 + }, + { + "epoch": 0.8978902953586498, + "grad_norm": 2.195239543914795, + "learning_rate": 2.0450320294398256e-06, + "loss": 2.466603088378906, + "step": 329840 + }, + { + "epoch": 0.8979447393493943, + "grad_norm": 2.118802070617676, + "learning_rate": 2.043941665530871e-06, + "loss": 2.432032585144043, + "step": 329860 + }, + { + "epoch": 0.8979991833401388, + "grad_norm": 2.163774013519287, + "learning_rate": 2.0428513016219164e-06, + "loss": 2.45437068939209, + "step": 329880 + }, + { + "epoch": 0.8980536273308833, + "grad_norm": 2.2430953979492188, + "learning_rate": 2.0417609377129618e-06, + "loss": 2.4560226440429687, + "step": 329900 + }, + { + "epoch": 0.8981080713216278, + "grad_norm": 3.206691026687622, + "learning_rate": 2.040670573804007e-06, + "loss": 2.4644847869873048, + "step": 329920 + }, + { + "epoch": 0.8981625153123723, + "grad_norm": 2.169719696044922, + "learning_rate": 2.0395802098950525e-06, + "loss": 2.476120376586914, + "step": 329940 + }, + { + "epoch": 0.898216959303117, + "grad_norm": 2.1826367378234863, + "learning_rate": 2.038489845986098e-06, + "loss": 2.461088943481445, + "step": 329960 + }, + { + "epoch": 0.8982714032938615, + "grad_norm": 2.1751413345336914, + "learning_rate": 2.0373994820771433e-06, + "loss": 2.4554164886474608, + "step": 329980 + }, + { + "epoch": 0.898325847284606, + "grad_norm": 2.23036527633667, + "learning_rate": 2.0363091181681887e-06, + "loss": 2.4796865463256834, + "step": 330000 + }, + { + "epoch": 0.8983802912753505, + "grad_norm": 2.087171792984009, + "learning_rate": 2.035218754259234e-06, + "loss": 2.4714481353759767, + "step": 330020 + }, + { + "epoch": 0.898434735266095, + "grad_norm": 2.303896903991699, + "learning_rate": 2.0341283903502795e-06, + "loss": 2.497036933898926, + "step": 330040 + }, + { + "epoch": 0.8984891792568396, + "grad_norm": 2.1322097778320312, + "learning_rate": 2.033038026441325e-06, + "loss": 2.409867286682129, + "step": 330060 + }, + { + "epoch": 0.8985436232475841, + "grad_norm": 2.1339032649993896, + "learning_rate": 2.0319476625323702e-06, + "loss": 2.4752361297607424, + "step": 330080 + }, + { + "epoch": 0.8985980672383286, + "grad_norm": 2.0759530067443848, + "learning_rate": 2.0308572986234156e-06, + "loss": 2.4385913848876952, + "step": 330100 + }, + { + "epoch": 0.8986525112290731, + "grad_norm": 2.1507723331451416, + "learning_rate": 2.029766934714461e-06, + "loss": 2.4515640258789064, + "step": 330120 + }, + { + "epoch": 0.8987069552198176, + "grad_norm": 2.1848721504211426, + "learning_rate": 2.0286765708055064e-06, + "loss": 2.496630096435547, + "step": 330140 + }, + { + "epoch": 0.8987613992105621, + "grad_norm": 2.21720814704895, + "learning_rate": 2.0275862068965518e-06, + "loss": 2.455435371398926, + "step": 330160 + }, + { + "epoch": 0.8988158432013067, + "grad_norm": 2.163804292678833, + "learning_rate": 2.026495842987597e-06, + "loss": 2.4357473373413088, + "step": 330180 + }, + { + "epoch": 0.8988702871920512, + "grad_norm": 2.185798168182373, + "learning_rate": 2.0254054790786425e-06, + "loss": 2.4362606048583983, + "step": 330200 + }, + { + "epoch": 0.8989247311827957, + "grad_norm": 2.110543966293335, + "learning_rate": 2.024315115169688e-06, + "loss": 2.4508697509765627, + "step": 330220 + }, + { + "epoch": 0.8989791751735402, + "grad_norm": 2.129146099090576, + "learning_rate": 2.0232247512607333e-06, + "loss": 2.4588762283325196, + "step": 330240 + }, + { + "epoch": 0.8990336191642847, + "grad_norm": 2.1068027019500732, + "learning_rate": 2.0221343873517787e-06, + "loss": 2.4583761215209963, + "step": 330260 + }, + { + "epoch": 0.8990880631550293, + "grad_norm": 2.2405924797058105, + "learning_rate": 2.021044023442824e-06, + "loss": 2.453594970703125, + "step": 330280 + }, + { + "epoch": 0.8991425071457738, + "grad_norm": 2.1539371013641357, + "learning_rate": 2.0199536595338695e-06, + "loss": 2.4596668243408204, + "step": 330300 + }, + { + "epoch": 0.8991969511365183, + "grad_norm": 2.1238749027252197, + "learning_rate": 2.018863295624915e-06, + "loss": 2.438271713256836, + "step": 330320 + }, + { + "epoch": 0.8992513951272628, + "grad_norm": 2.093621253967285, + "learning_rate": 2.0177729317159602e-06, + "loss": 2.4283538818359376, + "step": 330340 + }, + { + "epoch": 0.8993058391180073, + "grad_norm": 2.3187358379364014, + "learning_rate": 2.0166825678070056e-06, + "loss": 2.4419839859008787, + "step": 330360 + }, + { + "epoch": 0.8993602831087518, + "grad_norm": 2.134178876876831, + "learning_rate": 2.015592203898051e-06, + "loss": 2.4609748840332033, + "step": 330380 + }, + { + "epoch": 0.8994147270994964, + "grad_norm": 2.1032662391662598, + "learning_rate": 2.0145018399890964e-06, + "loss": 2.457990837097168, + "step": 330400 + }, + { + "epoch": 0.8994691710902409, + "grad_norm": 2.1324687004089355, + "learning_rate": 2.013411476080142e-06, + "loss": 2.443449592590332, + "step": 330420 + }, + { + "epoch": 0.8995236150809854, + "grad_norm": 2.1835715770721436, + "learning_rate": 2.012321112171187e-06, + "loss": 2.431312370300293, + "step": 330440 + }, + { + "epoch": 0.8995780590717299, + "grad_norm": 2.1693875789642334, + "learning_rate": 2.0112307482622326e-06, + "loss": 2.488117980957031, + "step": 330460 + }, + { + "epoch": 0.8996325030624744, + "grad_norm": 2.140078067779541, + "learning_rate": 2.010140384353278e-06, + "loss": 2.453039360046387, + "step": 330480 + }, + { + "epoch": 0.899686947053219, + "grad_norm": 2.2173173427581787, + "learning_rate": 2.0090500204443233e-06, + "loss": 2.4572065353393553, + "step": 330500 + }, + { + "epoch": 0.8997413910439636, + "grad_norm": 2.2009987831115723, + "learning_rate": 2.0079596565353687e-06, + "loss": 2.4411647796630858, + "step": 330520 + }, + { + "epoch": 0.8997958350347081, + "grad_norm": 2.0698282718658447, + "learning_rate": 2.006869292626414e-06, + "loss": 2.454006004333496, + "step": 330540 + }, + { + "epoch": 0.8998502790254526, + "grad_norm": 2.261396884918213, + "learning_rate": 2.0057789287174595e-06, + "loss": 2.442704772949219, + "step": 330560 + }, + { + "epoch": 0.8999047230161971, + "grad_norm": 2.1621651649475098, + "learning_rate": 2.004688564808505e-06, + "loss": 2.4438549041748048, + "step": 330580 + }, + { + "epoch": 0.8999591670069416, + "grad_norm": 2.1978821754455566, + "learning_rate": 2.0035982008995503e-06, + "loss": 2.4696006774902344, + "step": 330600 + }, + { + "epoch": 0.9000136109976862, + "grad_norm": 2.1591339111328125, + "learning_rate": 2.0025078369905956e-06, + "loss": 2.4672040939331055, + "step": 330620 + }, + { + "epoch": 0.9000680549884307, + "grad_norm": 2.3608877658843994, + "learning_rate": 2.001417473081641e-06, + "loss": 2.4717487335205077, + "step": 330640 + }, + { + "epoch": 0.9001224989791752, + "grad_norm": 2.1664814949035645, + "learning_rate": 2.0003271091726864e-06, + "loss": 2.451778602600098, + "step": 330660 + }, + { + "epoch": 0.9001769429699197, + "grad_norm": 2.1732840538024902, + "learning_rate": 1.9992367452637322e-06, + "loss": 2.476238250732422, + "step": 330680 + }, + { + "epoch": 0.9002313869606642, + "grad_norm": 2.150529623031616, + "learning_rate": 1.9981463813547776e-06, + "loss": 2.4756444931030273, + "step": 330700 + }, + { + "epoch": 0.9002858309514087, + "grad_norm": 2.4073116779327393, + "learning_rate": 1.997056017445823e-06, + "loss": 2.458508110046387, + "step": 330720 + }, + { + "epoch": 0.9003402749421533, + "grad_norm": 3.5760293006896973, + "learning_rate": 1.9959656535368684e-06, + "loss": 2.5016735076904295, + "step": 330740 + }, + { + "epoch": 0.9003947189328978, + "grad_norm": 2.1870193481445312, + "learning_rate": 1.9948752896279138e-06, + "loss": 2.5000797271728517, + "step": 330760 + }, + { + "epoch": 0.9004491629236423, + "grad_norm": 2.1526401042938232, + "learning_rate": 1.993784925718959e-06, + "loss": 2.4813474655151366, + "step": 330780 + }, + { + "epoch": 0.9005036069143868, + "grad_norm": 2.279395580291748, + "learning_rate": 1.9926945618100045e-06, + "loss": 2.4572990417480467, + "step": 330800 + }, + { + "epoch": 0.9005580509051313, + "grad_norm": 2.341886043548584, + "learning_rate": 1.99160419790105e-06, + "loss": 2.409333610534668, + "step": 330820 + }, + { + "epoch": 0.9006124948958759, + "grad_norm": 2.092716693878174, + "learning_rate": 1.9905138339920953e-06, + "loss": 2.496010971069336, + "step": 330840 + }, + { + "epoch": 0.9006669388866204, + "grad_norm": 2.173316240310669, + "learning_rate": 1.9894234700831407e-06, + "loss": 2.466060447692871, + "step": 330860 + }, + { + "epoch": 0.9007213828773649, + "grad_norm": 2.13812255859375, + "learning_rate": 1.988333106174186e-06, + "loss": 2.442301559448242, + "step": 330880 + }, + { + "epoch": 0.9007758268681094, + "grad_norm": 2.2394986152648926, + "learning_rate": 1.9872427422652315e-06, + "loss": 2.4524425506591796, + "step": 330900 + }, + { + "epoch": 0.9008302708588539, + "grad_norm": 2.1985373497009277, + "learning_rate": 1.986152378356277e-06, + "loss": 2.463021659851074, + "step": 330920 + }, + { + "epoch": 0.9008847148495984, + "grad_norm": 2.1169755458831787, + "learning_rate": 1.9850620144473222e-06, + "loss": 2.4470804214477537, + "step": 330940 + }, + { + "epoch": 0.900939158840343, + "grad_norm": 2.1475605964660645, + "learning_rate": 1.9839716505383676e-06, + "loss": 2.473802185058594, + "step": 330960 + }, + { + "epoch": 0.9009936028310875, + "grad_norm": 2.0610954761505127, + "learning_rate": 1.982881286629413e-06, + "loss": 2.4487436294555662, + "step": 330980 + }, + { + "epoch": 0.901048046821832, + "grad_norm": 2.4808833599090576, + "learning_rate": 1.9817909227204584e-06, + "loss": 2.445720100402832, + "step": 331000 + }, + { + "epoch": 0.9011024908125765, + "grad_norm": 2.1634347438812256, + "learning_rate": 1.9807005588115038e-06, + "loss": 2.436562728881836, + "step": 331020 + }, + { + "epoch": 0.901156934803321, + "grad_norm": 2.1833274364471436, + "learning_rate": 1.979610194902549e-06, + "loss": 2.444841003417969, + "step": 331040 + }, + { + "epoch": 0.9012113787940657, + "grad_norm": 2.1543476581573486, + "learning_rate": 1.9785198309935946e-06, + "loss": 2.4567163467407225, + "step": 331060 + }, + { + "epoch": 0.9012658227848102, + "grad_norm": 2.192274570465088, + "learning_rate": 1.9774294670846395e-06, + "loss": 2.466863250732422, + "step": 331080 + }, + { + "epoch": 0.9013202667755547, + "grad_norm": 2.194086790084839, + "learning_rate": 1.976339103175685e-06, + "loss": 2.4406661987304688, + "step": 331100 + }, + { + "epoch": 0.9013747107662992, + "grad_norm": 2.2964327335357666, + "learning_rate": 1.9752487392667303e-06, + "loss": 2.4665624618530275, + "step": 331120 + }, + { + "epoch": 0.9014291547570437, + "grad_norm": 2.1335268020629883, + "learning_rate": 1.9741583753577757e-06, + "loss": 2.445698356628418, + "step": 331140 + }, + { + "epoch": 0.9014835987477882, + "grad_norm": 2.27303409576416, + "learning_rate": 1.973068011448821e-06, + "loss": 2.441417694091797, + "step": 331160 + }, + { + "epoch": 0.9015380427385328, + "grad_norm": 2.143566131591797, + "learning_rate": 1.9719776475398664e-06, + "loss": 2.477713203430176, + "step": 331180 + }, + { + "epoch": 0.9015924867292773, + "grad_norm": 2.1592562198638916, + "learning_rate": 1.970887283630912e-06, + "loss": 2.454041862487793, + "step": 331200 + }, + { + "epoch": 0.9016469307200218, + "grad_norm": 2.1802406311035156, + "learning_rate": 1.9697969197219572e-06, + "loss": 2.469167709350586, + "step": 331220 + }, + { + "epoch": 0.9017013747107663, + "grad_norm": 2.192686080932617, + "learning_rate": 1.9687065558130026e-06, + "loss": 2.4853343963623047, + "step": 331240 + }, + { + "epoch": 0.9017558187015108, + "grad_norm": 2.271202802658081, + "learning_rate": 1.967616191904048e-06, + "loss": 2.476897430419922, + "step": 331260 + }, + { + "epoch": 0.9018102626922554, + "grad_norm": 2.135983943939209, + "learning_rate": 1.9665258279950934e-06, + "loss": 2.4928489685058595, + "step": 331280 + }, + { + "epoch": 0.9018647066829999, + "grad_norm": 2.1463308334350586, + "learning_rate": 1.9654354640861388e-06, + "loss": 2.4573545455932617, + "step": 331300 + }, + { + "epoch": 0.9019191506737444, + "grad_norm": 2.2418951988220215, + "learning_rate": 1.964345100177184e-06, + "loss": 2.4385374069213865, + "step": 331320 + }, + { + "epoch": 0.9019735946644889, + "grad_norm": 2.153505325317383, + "learning_rate": 1.9632547362682295e-06, + "loss": 2.453179359436035, + "step": 331340 + }, + { + "epoch": 0.9020280386552334, + "grad_norm": 2.2217185497283936, + "learning_rate": 1.962164372359275e-06, + "loss": 2.452877426147461, + "step": 331360 + }, + { + "epoch": 0.9020824826459779, + "grad_norm": 2.449097156524658, + "learning_rate": 1.9610740084503203e-06, + "loss": 2.492290496826172, + "step": 331380 + }, + { + "epoch": 0.9021369266367225, + "grad_norm": 2.272094249725342, + "learning_rate": 1.9599836445413657e-06, + "loss": 2.4774911880493162, + "step": 331400 + }, + { + "epoch": 0.902191370627467, + "grad_norm": 2.1536941528320312, + "learning_rate": 1.958893280632411e-06, + "loss": 2.4473173141479494, + "step": 331420 + }, + { + "epoch": 0.9022458146182115, + "grad_norm": 2.29388165473938, + "learning_rate": 1.9578029167234565e-06, + "loss": 2.4642553329467773, + "step": 331440 + }, + { + "epoch": 0.902300258608956, + "grad_norm": 2.1739327907562256, + "learning_rate": 1.956712552814502e-06, + "loss": 2.454051399230957, + "step": 331460 + }, + { + "epoch": 0.9023547025997005, + "grad_norm": 2.0676116943359375, + "learning_rate": 1.9556221889055472e-06, + "loss": 2.4871967315673826, + "step": 331480 + }, + { + "epoch": 0.9024091465904451, + "grad_norm": 2.259885311126709, + "learning_rate": 1.9545318249965926e-06, + "loss": 2.471974182128906, + "step": 331500 + }, + { + "epoch": 0.9024635905811896, + "grad_norm": 2.146824836730957, + "learning_rate": 1.953441461087638e-06, + "loss": 2.4732837677001953, + "step": 331520 + }, + { + "epoch": 0.9025180345719341, + "grad_norm": 2.0696358680725098, + "learning_rate": 1.9523510971786834e-06, + "loss": 2.4322040557861326, + "step": 331540 + }, + { + "epoch": 0.9025724785626786, + "grad_norm": 2.2198688983917236, + "learning_rate": 1.9512607332697288e-06, + "loss": 2.4562231063842774, + "step": 331560 + }, + { + "epoch": 0.9026269225534231, + "grad_norm": 2.1938304901123047, + "learning_rate": 1.950170369360774e-06, + "loss": 2.4532737731933594, + "step": 331580 + }, + { + "epoch": 0.9026813665441676, + "grad_norm": 2.1536591053009033, + "learning_rate": 1.9490800054518195e-06, + "loss": 2.4584781646728517, + "step": 331600 + }, + { + "epoch": 0.9027358105349123, + "grad_norm": 2.726436138153076, + "learning_rate": 1.947989641542865e-06, + "loss": 2.4339271545410157, + "step": 331620 + }, + { + "epoch": 0.9027902545256568, + "grad_norm": 2.1133177280426025, + "learning_rate": 1.9468992776339103e-06, + "loss": 2.4641098022460937, + "step": 331640 + }, + { + "epoch": 0.9028446985164013, + "grad_norm": 2.221067428588867, + "learning_rate": 1.9458089137249557e-06, + "loss": 2.449214744567871, + "step": 331660 + }, + { + "epoch": 0.9028991425071458, + "grad_norm": 2.128244161605835, + "learning_rate": 1.944718549816001e-06, + "loss": 2.4381608963012695, + "step": 331680 + }, + { + "epoch": 0.9029535864978903, + "grad_norm": 3.1549432277679443, + "learning_rate": 1.9436281859070465e-06, + "loss": 2.443782997131348, + "step": 331700 + }, + { + "epoch": 0.9030080304886349, + "grad_norm": 2.1713755130767822, + "learning_rate": 1.942537821998092e-06, + "loss": 2.4503211975097656, + "step": 331720 + }, + { + "epoch": 0.9030624744793794, + "grad_norm": 2.582700252532959, + "learning_rate": 1.9414474580891372e-06, + "loss": 2.4577783584594726, + "step": 331740 + }, + { + "epoch": 0.9031169184701239, + "grad_norm": 2.2149064540863037, + "learning_rate": 1.9403570941801826e-06, + "loss": 2.442597198486328, + "step": 331760 + }, + { + "epoch": 0.9031713624608684, + "grad_norm": 2.312009572982788, + "learning_rate": 1.939266730271228e-06, + "loss": 2.4337236404418947, + "step": 331780 + }, + { + "epoch": 0.9032258064516129, + "grad_norm": 2.1859793663024902, + "learning_rate": 1.9381763663622734e-06, + "loss": 2.4686981201171876, + "step": 331800 + }, + { + "epoch": 0.9032802504423574, + "grad_norm": 2.067499876022339, + "learning_rate": 1.937086002453319e-06, + "loss": 2.4437026977539062, + "step": 331820 + }, + { + "epoch": 0.903334694433102, + "grad_norm": 2.123396158218384, + "learning_rate": 1.935995638544364e-06, + "loss": 2.4745195388793944, + "step": 331840 + }, + { + "epoch": 0.9033891384238465, + "grad_norm": 2.2228472232818604, + "learning_rate": 1.9349052746354096e-06, + "loss": 2.456745910644531, + "step": 331860 + }, + { + "epoch": 0.903443582414591, + "grad_norm": 2.165954828262329, + "learning_rate": 1.933814910726455e-06, + "loss": 2.4707929611206056, + "step": 331880 + }, + { + "epoch": 0.9034980264053355, + "grad_norm": 2.273448944091797, + "learning_rate": 1.9327245468175003e-06, + "loss": 2.4755218505859373, + "step": 331900 + }, + { + "epoch": 0.90355247039608, + "grad_norm": 2.27935791015625, + "learning_rate": 1.9316341829085457e-06, + "loss": 2.4665048599243162, + "step": 331920 + }, + { + "epoch": 0.9036069143868245, + "grad_norm": 2.1653268337249756, + "learning_rate": 1.930543818999591e-06, + "loss": 2.4467357635498046, + "step": 331940 + }, + { + "epoch": 0.9036613583775691, + "grad_norm": 2.143155813217163, + "learning_rate": 1.9294534550906365e-06, + "loss": 2.468916130065918, + "step": 331960 + }, + { + "epoch": 0.9037158023683136, + "grad_norm": 2.1899538040161133, + "learning_rate": 1.928363091181682e-06, + "loss": 2.435365676879883, + "step": 331980 + }, + { + "epoch": 0.9037702463590581, + "grad_norm": 2.1274666786193848, + "learning_rate": 1.9272727272727273e-06, + "loss": 2.4540205001831055, + "step": 332000 + }, + { + "epoch": 0.9038246903498026, + "grad_norm": 2.142781972885132, + "learning_rate": 1.9261823633637726e-06, + "loss": 2.4887722015380858, + "step": 332020 + }, + { + "epoch": 0.9038791343405471, + "grad_norm": 2.05829119682312, + "learning_rate": 1.925091999454818e-06, + "loss": 2.4718761444091797, + "step": 332040 + }, + { + "epoch": 0.9039335783312917, + "grad_norm": 2.2141566276550293, + "learning_rate": 1.9240016355458634e-06, + "loss": 2.4719465255737303, + "step": 332060 + }, + { + "epoch": 0.9039880223220362, + "grad_norm": 2.230133056640625, + "learning_rate": 1.922911271636909e-06, + "loss": 2.458184814453125, + "step": 332080 + }, + { + "epoch": 0.9040424663127807, + "grad_norm": 2.1259331703186035, + "learning_rate": 1.921820907727954e-06, + "loss": 2.4830547332763673, + "step": 332100 + }, + { + "epoch": 0.9040969103035252, + "grad_norm": 2.198929786682129, + "learning_rate": 1.9207305438189996e-06, + "loss": 2.4660823822021483, + "step": 332120 + }, + { + "epoch": 0.9041513542942697, + "grad_norm": 2.1411333084106445, + "learning_rate": 1.919640179910045e-06, + "loss": 2.469764518737793, + "step": 332140 + }, + { + "epoch": 0.9042057982850142, + "grad_norm": 2.2336833477020264, + "learning_rate": 1.9185498160010903e-06, + "loss": 2.4464126586914063, + "step": 332160 + }, + { + "epoch": 0.9042602422757589, + "grad_norm": 2.5602715015411377, + "learning_rate": 1.9174594520921357e-06, + "loss": 2.4524269104003906, + "step": 332180 + }, + { + "epoch": 0.9043146862665034, + "grad_norm": 2.137615919113159, + "learning_rate": 1.916369088183181e-06, + "loss": 2.4397300720214843, + "step": 332200 + }, + { + "epoch": 0.9043691302572479, + "grad_norm": 2.2787652015686035, + "learning_rate": 1.9152787242742265e-06, + "loss": 2.44146614074707, + "step": 332220 + }, + { + "epoch": 0.9044235742479924, + "grad_norm": 2.2002406120300293, + "learning_rate": 1.914188360365272e-06, + "loss": 2.475433921813965, + "step": 332240 + }, + { + "epoch": 0.9044780182387369, + "grad_norm": 2.253138303756714, + "learning_rate": 1.9130979964563173e-06, + "loss": 2.449189376831055, + "step": 332260 + }, + { + "epoch": 0.9045324622294815, + "grad_norm": 2.5371463298797607, + "learning_rate": 1.9120076325473627e-06, + "loss": 2.4864730834960938, + "step": 332280 + }, + { + "epoch": 0.904586906220226, + "grad_norm": 2.239511013031006, + "learning_rate": 1.910917268638408e-06, + "loss": 2.482571029663086, + "step": 332300 + }, + { + "epoch": 0.9046413502109705, + "grad_norm": 2.1209256649017334, + "learning_rate": 1.9098269047294534e-06, + "loss": 2.469852638244629, + "step": 332320 + }, + { + "epoch": 0.904695794201715, + "grad_norm": 2.2425832748413086, + "learning_rate": 1.908736540820499e-06, + "loss": 2.4677257537841797, + "step": 332340 + }, + { + "epoch": 0.9047502381924595, + "grad_norm": 2.252819299697876, + "learning_rate": 1.907646176911544e-06, + "loss": 2.459505081176758, + "step": 332360 + }, + { + "epoch": 0.904804682183204, + "grad_norm": 2.1322691440582275, + "learning_rate": 1.9065558130025898e-06, + "loss": 2.4640830993652343, + "step": 332380 + }, + { + "epoch": 0.9048591261739486, + "grad_norm": 2.298055410385132, + "learning_rate": 1.9054654490936352e-06, + "loss": 2.4359413146972657, + "step": 332400 + }, + { + "epoch": 0.9049135701646931, + "grad_norm": 2.1989707946777344, + "learning_rate": 1.9043750851846806e-06, + "loss": 2.4790679931640627, + "step": 332420 + }, + { + "epoch": 0.9049680141554376, + "grad_norm": 2.2557716369628906, + "learning_rate": 1.903284721275726e-06, + "loss": 2.4693120956420898, + "step": 332440 + }, + { + "epoch": 0.9050224581461821, + "grad_norm": 2.141300916671753, + "learning_rate": 1.9021943573667713e-06, + "loss": 2.486277770996094, + "step": 332460 + }, + { + "epoch": 0.9050769021369266, + "grad_norm": 2.129547595977783, + "learning_rate": 1.9011039934578167e-06, + "loss": 2.4718626022338865, + "step": 332480 + }, + { + "epoch": 0.9051313461276712, + "grad_norm": 2.1975855827331543, + "learning_rate": 1.9000136295488621e-06, + "loss": 2.4831871032714843, + "step": 332500 + }, + { + "epoch": 0.9051857901184157, + "grad_norm": 2.236990213394165, + "learning_rate": 1.8989232656399075e-06, + "loss": 2.4260026931762697, + "step": 332520 + }, + { + "epoch": 0.9052402341091602, + "grad_norm": 2.341106414794922, + "learning_rate": 1.8978329017309529e-06, + "loss": 2.450288200378418, + "step": 332540 + }, + { + "epoch": 0.9052946780999047, + "grad_norm": 2.275076150894165, + "learning_rate": 1.8967425378219983e-06, + "loss": 2.467268371582031, + "step": 332560 + }, + { + "epoch": 0.9053491220906492, + "grad_norm": 2.1105153560638428, + "learning_rate": 1.8956521739130437e-06, + "loss": 2.4555118560791014, + "step": 332580 + }, + { + "epoch": 0.9054035660813937, + "grad_norm": 2.1529548168182373, + "learning_rate": 1.894561810004089e-06, + "loss": 2.4719587326049806, + "step": 332600 + }, + { + "epoch": 0.9054580100721383, + "grad_norm": 2.153388261795044, + "learning_rate": 1.8934714460951344e-06, + "loss": 2.4490839004516602, + "step": 332620 + }, + { + "epoch": 0.9055124540628828, + "grad_norm": 2.233729839324951, + "learning_rate": 1.8923810821861798e-06, + "loss": 2.442694664001465, + "step": 332640 + }, + { + "epoch": 0.9055668980536273, + "grad_norm": 2.2343568801879883, + "learning_rate": 1.8912907182772252e-06, + "loss": 2.454148864746094, + "step": 332660 + }, + { + "epoch": 0.9056213420443718, + "grad_norm": 2.09012508392334, + "learning_rate": 1.8902003543682706e-06, + "loss": 2.465760040283203, + "step": 332680 + }, + { + "epoch": 0.9056757860351163, + "grad_norm": 2.1177492141723633, + "learning_rate": 1.889109990459316e-06, + "loss": 2.4480993270874025, + "step": 332700 + }, + { + "epoch": 0.905730230025861, + "grad_norm": 2.5954995155334473, + "learning_rate": 1.8880196265503614e-06, + "loss": 2.468134307861328, + "step": 332720 + }, + { + "epoch": 0.9057846740166055, + "grad_norm": 2.05617618560791, + "learning_rate": 1.8869292626414067e-06, + "loss": 2.456990051269531, + "step": 332740 + }, + { + "epoch": 0.90583911800735, + "grad_norm": 2.3073983192443848, + "learning_rate": 1.8858388987324521e-06, + "loss": 2.457769012451172, + "step": 332760 + }, + { + "epoch": 0.9058935619980945, + "grad_norm": 2.173879384994507, + "learning_rate": 1.8847485348234975e-06, + "loss": 2.437226676940918, + "step": 332780 + }, + { + "epoch": 0.905948005988839, + "grad_norm": 2.152513027191162, + "learning_rate": 1.883658170914543e-06, + "loss": 2.475214958190918, + "step": 332800 + }, + { + "epoch": 0.9060024499795835, + "grad_norm": 2.2217776775360107, + "learning_rate": 1.8825678070055883e-06, + "loss": 2.447332000732422, + "step": 332820 + }, + { + "epoch": 0.9060568939703281, + "grad_norm": 2.1704680919647217, + "learning_rate": 1.8814774430966337e-06, + "loss": 2.4424320220947267, + "step": 332840 + }, + { + "epoch": 0.9061113379610726, + "grad_norm": 2.0976336002349854, + "learning_rate": 1.880387079187679e-06, + "loss": 2.4608869552612305, + "step": 332860 + }, + { + "epoch": 0.9061657819518171, + "grad_norm": 2.126817464828491, + "learning_rate": 1.8792967152787244e-06, + "loss": 2.472446060180664, + "step": 332880 + }, + { + "epoch": 0.9062202259425616, + "grad_norm": 2.3979480266571045, + "learning_rate": 1.8782063513697698e-06, + "loss": 2.4569087982177735, + "step": 332900 + }, + { + "epoch": 0.9062746699333061, + "grad_norm": 2.2034740447998047, + "learning_rate": 1.8771159874608152e-06, + "loss": 2.4857030868530274, + "step": 332920 + }, + { + "epoch": 0.9063291139240506, + "grad_norm": 2.1592860221862793, + "learning_rate": 1.8760256235518606e-06, + "loss": 2.447429656982422, + "step": 332940 + }, + { + "epoch": 0.9063835579147952, + "grad_norm": 2.1601812839508057, + "learning_rate": 1.874935259642906e-06, + "loss": 2.448238563537598, + "step": 332960 + }, + { + "epoch": 0.9064380019055397, + "grad_norm": 2.138329029083252, + "learning_rate": 1.8738448957339514e-06, + "loss": 2.4505550384521486, + "step": 332980 + }, + { + "epoch": 0.9064924458962842, + "grad_norm": 2.1313817501068115, + "learning_rate": 1.8727545318249968e-06, + "loss": 2.4678167343139648, + "step": 333000 + }, + { + "epoch": 0.9065468898870287, + "grad_norm": 2.223676919937134, + "learning_rate": 1.8716641679160421e-06, + "loss": 2.425686264038086, + "step": 333020 + }, + { + "epoch": 0.9066013338777732, + "grad_norm": 5.220577239990234, + "learning_rate": 1.8705738040070875e-06, + "loss": 2.4740007400512694, + "step": 333040 + }, + { + "epoch": 0.9066557778685178, + "grad_norm": 2.0834224224090576, + "learning_rate": 1.869483440098133e-06, + "loss": 2.502302551269531, + "step": 333060 + }, + { + "epoch": 0.9067102218592623, + "grad_norm": 2.1159985065460205, + "learning_rate": 1.8683930761891783e-06, + "loss": 2.429169464111328, + "step": 333080 + }, + { + "epoch": 0.9067646658500068, + "grad_norm": 2.1951119899749756, + "learning_rate": 1.8673027122802237e-06, + "loss": 2.47696590423584, + "step": 333100 + }, + { + "epoch": 0.9068191098407513, + "grad_norm": 2.3631744384765625, + "learning_rate": 1.866212348371269e-06, + "loss": 2.431589126586914, + "step": 333120 + }, + { + "epoch": 0.9068735538314958, + "grad_norm": 2.2201485633850098, + "learning_rate": 1.8651219844623145e-06, + "loss": 2.4554067611694337, + "step": 333140 + }, + { + "epoch": 0.9069279978222403, + "grad_norm": 2.139641523361206, + "learning_rate": 1.8640316205533598e-06, + "loss": 2.4626752853393556, + "step": 333160 + }, + { + "epoch": 0.9069824418129849, + "grad_norm": 2.072629690170288, + "learning_rate": 1.8629412566444052e-06, + "loss": 2.4733041763305663, + "step": 333180 + }, + { + "epoch": 0.9070368858037294, + "grad_norm": 2.268573045730591, + "learning_rate": 1.8618508927354506e-06, + "loss": 2.455702209472656, + "step": 333200 + }, + { + "epoch": 0.9070913297944739, + "grad_norm": 2.1323978900909424, + "learning_rate": 1.860760528826496e-06, + "loss": 2.423478698730469, + "step": 333220 + }, + { + "epoch": 0.9071457737852184, + "grad_norm": 2.093439817428589, + "learning_rate": 1.8596701649175414e-06, + "loss": 2.446327972412109, + "step": 333240 + }, + { + "epoch": 0.9072002177759629, + "grad_norm": 2.241854429244995, + "learning_rate": 1.8585798010085868e-06, + "loss": 2.4653446197509767, + "step": 333260 + }, + { + "epoch": 0.9072546617667075, + "grad_norm": 2.085636615753174, + "learning_rate": 1.8574894370996322e-06, + "loss": 2.456900787353516, + "step": 333280 + }, + { + "epoch": 0.907309105757452, + "grad_norm": 2.1375081539154053, + "learning_rate": 1.8563990731906775e-06, + "loss": 2.4505245208740236, + "step": 333300 + }, + { + "epoch": 0.9073635497481966, + "grad_norm": 2.183929204940796, + "learning_rate": 1.855308709281723e-06, + "loss": 2.452897834777832, + "step": 333320 + }, + { + "epoch": 0.907417993738941, + "grad_norm": 2.2387242317199707, + "learning_rate": 1.8542183453727683e-06, + "loss": 2.484584617614746, + "step": 333340 + }, + { + "epoch": 0.9074724377296856, + "grad_norm": 6.4515581130981445, + "learning_rate": 1.8531279814638137e-06, + "loss": 2.4419929504394533, + "step": 333360 + }, + { + "epoch": 0.9075268817204301, + "grad_norm": 2.2806408405303955, + "learning_rate": 1.852037617554859e-06, + "loss": 2.4687849044799806, + "step": 333380 + }, + { + "epoch": 0.9075813257111747, + "grad_norm": 2.8805716037750244, + "learning_rate": 1.8509472536459045e-06, + "loss": 2.4972410202026367, + "step": 333400 + }, + { + "epoch": 0.9076357697019192, + "grad_norm": 2.2150352001190186, + "learning_rate": 1.8498568897369499e-06, + "loss": 2.45491886138916, + "step": 333420 + }, + { + "epoch": 0.9076902136926637, + "grad_norm": 2.255824565887451, + "learning_rate": 1.8487665258279952e-06, + "loss": 2.4827064514160155, + "step": 333440 + }, + { + "epoch": 0.9077446576834082, + "grad_norm": 2.2454047203063965, + "learning_rate": 1.8476761619190406e-06, + "loss": 2.4448831558227537, + "step": 333460 + }, + { + "epoch": 0.9077991016741527, + "grad_norm": 2.155466318130493, + "learning_rate": 1.846585798010086e-06, + "loss": 2.4482233047485353, + "step": 333480 + }, + { + "epoch": 0.9078535456648973, + "grad_norm": 2.165902614593506, + "learning_rate": 1.8454954341011314e-06, + "loss": 2.4657752990722654, + "step": 333500 + }, + { + "epoch": 0.9079079896556418, + "grad_norm": 2.28450608253479, + "learning_rate": 1.8444050701921768e-06, + "loss": 2.4800071716308594, + "step": 333520 + }, + { + "epoch": 0.9079624336463863, + "grad_norm": 2.1438379287719727, + "learning_rate": 1.8433147062832222e-06, + "loss": 2.4551315307617188, + "step": 333540 + }, + { + "epoch": 0.9080168776371308, + "grad_norm": 2.1935713291168213, + "learning_rate": 1.8422243423742676e-06, + "loss": 2.4808799743652346, + "step": 333560 + }, + { + "epoch": 0.9080713216278753, + "grad_norm": 2.3453986644744873, + "learning_rate": 1.841133978465313e-06, + "loss": 2.46637020111084, + "step": 333580 + }, + { + "epoch": 0.9081257656186198, + "grad_norm": 2.2263715267181396, + "learning_rate": 1.8400436145563583e-06, + "loss": 2.4449462890625, + "step": 333600 + }, + { + "epoch": 0.9081802096093644, + "grad_norm": 2.3277323246002197, + "learning_rate": 1.8389532506474037e-06, + "loss": 2.471623420715332, + "step": 333620 + }, + { + "epoch": 0.9082346536001089, + "grad_norm": 2.1241753101348877, + "learning_rate": 1.837862886738449e-06, + "loss": 2.4276796340942384, + "step": 333640 + }, + { + "epoch": 0.9082890975908534, + "grad_norm": 2.1811795234680176, + "learning_rate": 1.8367725228294945e-06, + "loss": 2.4688125610351563, + "step": 333660 + }, + { + "epoch": 0.9083435415815979, + "grad_norm": 2.172529935836792, + "learning_rate": 1.8356821589205399e-06, + "loss": 2.4564016342163084, + "step": 333680 + }, + { + "epoch": 0.9083979855723424, + "grad_norm": 2.1953272819519043, + "learning_rate": 1.8345917950115853e-06, + "loss": 2.4962364196777345, + "step": 333700 + }, + { + "epoch": 0.908452429563087, + "grad_norm": 2.2233126163482666, + "learning_rate": 1.8335014311026306e-06, + "loss": 2.4409343719482424, + "step": 333720 + }, + { + "epoch": 0.9085068735538315, + "grad_norm": 2.240703582763672, + "learning_rate": 1.832411067193676e-06, + "loss": 2.490866470336914, + "step": 333740 + }, + { + "epoch": 0.908561317544576, + "grad_norm": 2.354243040084839, + "learning_rate": 1.8313207032847214e-06, + "loss": 2.4711469650268554, + "step": 333760 + }, + { + "epoch": 0.9086157615353205, + "grad_norm": 2.1735687255859375, + "learning_rate": 1.8302303393757668e-06, + "loss": 2.4487068176269533, + "step": 333780 + }, + { + "epoch": 0.908670205526065, + "grad_norm": 2.2095255851745605, + "learning_rate": 1.8291399754668122e-06, + "loss": 2.4604244232177734, + "step": 333800 + }, + { + "epoch": 0.9087246495168095, + "grad_norm": 2.1687793731689453, + "learning_rate": 1.8280496115578576e-06, + "loss": 2.444449234008789, + "step": 333820 + }, + { + "epoch": 0.9087790935075541, + "grad_norm": 2.165102243423462, + "learning_rate": 1.826959247648903e-06, + "loss": 2.454157066345215, + "step": 333840 + }, + { + "epoch": 0.9088335374982986, + "grad_norm": 2.3202672004699707, + "learning_rate": 1.8258688837399483e-06, + "loss": 2.4176507949829102, + "step": 333860 + }, + { + "epoch": 0.9088879814890432, + "grad_norm": 2.3592045307159424, + "learning_rate": 1.8247785198309937e-06, + "loss": 2.4641191482543947, + "step": 333880 + }, + { + "epoch": 0.9089424254797877, + "grad_norm": 2.2302441596984863, + "learning_rate": 1.8236881559220391e-06, + "loss": 2.4508634567260743, + "step": 333900 + }, + { + "epoch": 0.9089968694705322, + "grad_norm": 2.2514164447784424, + "learning_rate": 1.8225977920130845e-06, + "loss": 2.43680419921875, + "step": 333920 + }, + { + "epoch": 0.9090513134612768, + "grad_norm": 2.2052202224731445, + "learning_rate": 1.8215074281041299e-06, + "loss": 2.430042839050293, + "step": 333940 + }, + { + "epoch": 0.9091057574520213, + "grad_norm": 2.2038745880126953, + "learning_rate": 1.820417064195175e-06, + "loss": 2.4549074172973633, + "step": 333960 + }, + { + "epoch": 0.9091602014427658, + "grad_norm": 2.151944160461426, + "learning_rate": 1.8193267002862204e-06, + "loss": 2.4252092361450197, + "step": 333980 + }, + { + "epoch": 0.9092146454335103, + "grad_norm": 2.2194111347198486, + "learning_rate": 1.8182363363772658e-06, + "loss": 2.443617248535156, + "step": 334000 + }, + { + "epoch": 0.9092690894242548, + "grad_norm": 2.0970571041107178, + "learning_rate": 1.8171459724683112e-06, + "loss": 2.490275573730469, + "step": 334020 + }, + { + "epoch": 0.9093235334149993, + "grad_norm": 2.1085596084594727, + "learning_rate": 1.8160556085593566e-06, + "loss": 2.468169593811035, + "step": 334040 + }, + { + "epoch": 0.9093779774057439, + "grad_norm": 2.2098658084869385, + "learning_rate": 1.8149652446504024e-06, + "loss": 2.4721824645996096, + "step": 334060 + }, + { + "epoch": 0.9094324213964884, + "grad_norm": 2.1216561794281006, + "learning_rate": 1.8138748807414478e-06, + "loss": 2.446278762817383, + "step": 334080 + }, + { + "epoch": 0.9094868653872329, + "grad_norm": 2.1477789878845215, + "learning_rate": 1.8127845168324932e-06, + "loss": 2.4348876953125, + "step": 334100 + }, + { + "epoch": 0.9095413093779774, + "grad_norm": 2.2955551147460938, + "learning_rate": 1.8116941529235386e-06, + "loss": 2.4606948852539063, + "step": 334120 + }, + { + "epoch": 0.9095957533687219, + "grad_norm": 2.365230083465576, + "learning_rate": 1.810603789014584e-06, + "loss": 2.4732250213623046, + "step": 334140 + }, + { + "epoch": 0.9096501973594664, + "grad_norm": 2.3431622982025146, + "learning_rate": 1.8095134251056293e-06, + "loss": 2.453443717956543, + "step": 334160 + }, + { + "epoch": 0.909704641350211, + "grad_norm": 2.1674253940582275, + "learning_rate": 1.8084230611966747e-06, + "loss": 2.460926628112793, + "step": 334180 + }, + { + "epoch": 0.9097590853409555, + "grad_norm": 2.2029523849487305, + "learning_rate": 1.8073326972877201e-06, + "loss": 2.449582099914551, + "step": 334200 + }, + { + "epoch": 0.9098135293317, + "grad_norm": 2.1778547763824463, + "learning_rate": 1.8062423333787655e-06, + "loss": 2.4529903411865233, + "step": 334220 + }, + { + "epoch": 0.9098679733224445, + "grad_norm": 2.230541229248047, + "learning_rate": 1.8051519694698109e-06, + "loss": 2.458131217956543, + "step": 334240 + }, + { + "epoch": 0.909922417313189, + "grad_norm": 2.1065304279327393, + "learning_rate": 1.8040616055608563e-06, + "loss": 2.4432580947875975, + "step": 334260 + }, + { + "epoch": 0.9099768613039336, + "grad_norm": 2.0811972618103027, + "learning_rate": 1.8029712416519017e-06, + "loss": 2.4724748611450194, + "step": 334280 + }, + { + "epoch": 0.9100313052946781, + "grad_norm": 2.1467208862304688, + "learning_rate": 1.801880877742947e-06, + "loss": 2.4871295928955077, + "step": 334300 + }, + { + "epoch": 0.9100857492854226, + "grad_norm": 2.1430327892303467, + "learning_rate": 1.8007905138339924e-06, + "loss": 2.4593006134033204, + "step": 334320 + }, + { + "epoch": 0.9101401932761671, + "grad_norm": 2.1182682514190674, + "learning_rate": 1.7997001499250378e-06, + "loss": 2.45487003326416, + "step": 334340 + }, + { + "epoch": 0.9101946372669116, + "grad_norm": 2.1337356567382812, + "learning_rate": 1.7986097860160832e-06, + "loss": 2.4544647216796873, + "step": 334360 + }, + { + "epoch": 0.9102490812576561, + "grad_norm": 2.1268763542175293, + "learning_rate": 1.7975194221071286e-06, + "loss": 2.4883609771728517, + "step": 334380 + }, + { + "epoch": 0.9103035252484007, + "grad_norm": 2.1316232681274414, + "learning_rate": 1.796429058198174e-06, + "loss": 2.449487495422363, + "step": 334400 + }, + { + "epoch": 0.9103579692391452, + "grad_norm": 2.4661030769348145, + "learning_rate": 1.7953386942892194e-06, + "loss": 2.442811965942383, + "step": 334420 + }, + { + "epoch": 0.9104124132298897, + "grad_norm": 2.0950379371643066, + "learning_rate": 1.7942483303802647e-06, + "loss": 2.441197967529297, + "step": 334440 + }, + { + "epoch": 0.9104668572206343, + "grad_norm": 2.1568591594696045, + "learning_rate": 1.79315796647131e-06, + "loss": 2.459012031555176, + "step": 334460 + }, + { + "epoch": 0.9105213012113788, + "grad_norm": 2.300830125808716, + "learning_rate": 1.7920676025623553e-06, + "loss": 2.4553356170654297, + "step": 334480 + }, + { + "epoch": 0.9105757452021234, + "grad_norm": 2.2195639610290527, + "learning_rate": 1.7909772386534007e-06, + "loss": 2.4453025817871095, + "step": 334500 + }, + { + "epoch": 0.9106301891928679, + "grad_norm": 2.144901990890503, + "learning_rate": 1.789886874744446e-06, + "loss": 2.4370595932006838, + "step": 334520 + }, + { + "epoch": 0.9106846331836124, + "grad_norm": 2.1521902084350586, + "learning_rate": 1.7887965108354915e-06, + "loss": 2.4610113143920898, + "step": 334540 + }, + { + "epoch": 0.9107390771743569, + "grad_norm": 2.0901644229888916, + "learning_rate": 1.7877061469265368e-06, + "loss": 2.4709360122680666, + "step": 334560 + }, + { + "epoch": 0.9107935211651014, + "grad_norm": 2.0955288410186768, + "learning_rate": 1.7866157830175822e-06, + "loss": 2.439345932006836, + "step": 334580 + }, + { + "epoch": 0.9108479651558459, + "grad_norm": 2.086885452270508, + "learning_rate": 1.7855254191086276e-06, + "loss": 2.455860137939453, + "step": 334600 + }, + { + "epoch": 0.9109024091465905, + "grad_norm": 2.2319180965423584, + "learning_rate": 1.784435055199673e-06, + "loss": 2.459773635864258, + "step": 334620 + }, + { + "epoch": 0.910956853137335, + "grad_norm": 2.1392571926116943, + "learning_rate": 1.7833446912907184e-06, + "loss": 2.4389392852783205, + "step": 334640 + }, + { + "epoch": 0.9110112971280795, + "grad_norm": 2.18758225440979, + "learning_rate": 1.7822543273817638e-06, + "loss": 2.45858211517334, + "step": 334660 + }, + { + "epoch": 0.911065741118824, + "grad_norm": 2.133136510848999, + "learning_rate": 1.7811639634728092e-06, + "loss": 2.452266502380371, + "step": 334680 + }, + { + "epoch": 0.9111201851095685, + "grad_norm": 2.1828269958496094, + "learning_rate": 1.7800735995638545e-06, + "loss": 2.462586784362793, + "step": 334700 + }, + { + "epoch": 0.9111746291003131, + "grad_norm": 2.2029590606689453, + "learning_rate": 1.7789832356549e-06, + "loss": 2.4626180648803713, + "step": 334720 + }, + { + "epoch": 0.9112290730910576, + "grad_norm": 2.234471321105957, + "learning_rate": 1.7778928717459453e-06, + "loss": 2.4450788497924805, + "step": 334740 + }, + { + "epoch": 0.9112835170818021, + "grad_norm": 3.312516450881958, + "learning_rate": 1.7768025078369907e-06, + "loss": 2.4520030975341798, + "step": 334760 + }, + { + "epoch": 0.9113379610725466, + "grad_norm": 2.1630280017852783, + "learning_rate": 1.775712143928036e-06, + "loss": 2.424354362487793, + "step": 334780 + }, + { + "epoch": 0.9113924050632911, + "grad_norm": 2.1262924671173096, + "learning_rate": 1.7746217800190815e-06, + "loss": 2.4532730102539064, + "step": 334800 + }, + { + "epoch": 0.9114468490540356, + "grad_norm": 2.1464054584503174, + "learning_rate": 1.7735314161101269e-06, + "loss": 2.4681833267211912, + "step": 334820 + }, + { + "epoch": 0.9115012930447802, + "grad_norm": 2.2425081729888916, + "learning_rate": 1.7724410522011722e-06, + "loss": 2.4405887603759764, + "step": 334840 + }, + { + "epoch": 0.9115557370355247, + "grad_norm": 2.1912620067596436, + "learning_rate": 1.7713506882922176e-06, + "loss": 2.468632125854492, + "step": 334860 + }, + { + "epoch": 0.9116101810262692, + "grad_norm": 2.2793290615081787, + "learning_rate": 1.770260324383263e-06, + "loss": 2.4472373962402343, + "step": 334880 + }, + { + "epoch": 0.9116646250170137, + "grad_norm": 2.2198102474212646, + "learning_rate": 1.7691699604743084e-06, + "loss": 2.4499563217163085, + "step": 334900 + }, + { + "epoch": 0.9117190690077582, + "grad_norm": 2.561133861541748, + "learning_rate": 1.7680795965653538e-06, + "loss": 2.4647634506225584, + "step": 334920 + }, + { + "epoch": 0.9117735129985028, + "grad_norm": 2.112095355987549, + "learning_rate": 1.7669892326563992e-06, + "loss": 2.471176528930664, + "step": 334940 + }, + { + "epoch": 0.9118279569892473, + "grad_norm": 2.403343439102173, + "learning_rate": 1.7658988687474446e-06, + "loss": 2.4457218170166017, + "step": 334960 + }, + { + "epoch": 0.9118824009799918, + "grad_norm": 2.0601296424865723, + "learning_rate": 1.76480850483849e-06, + "loss": 2.4394289016723634, + "step": 334980 + }, + { + "epoch": 0.9119368449707363, + "grad_norm": 2.12188458442688, + "learning_rate": 1.7637181409295353e-06, + "loss": 2.4350276947021485, + "step": 335000 + }, + { + "epoch": 0.9119912889614809, + "grad_norm": 2.1264967918395996, + "learning_rate": 1.7626277770205807e-06, + "loss": 2.456133270263672, + "step": 335020 + }, + { + "epoch": 0.9120457329522254, + "grad_norm": 2.225029230117798, + "learning_rate": 1.761537413111626e-06, + "loss": 2.462057876586914, + "step": 335040 + }, + { + "epoch": 0.91210017694297, + "grad_norm": 2.2979536056518555, + "learning_rate": 1.7604470492026715e-06, + "loss": 2.4491458892822267, + "step": 335060 + }, + { + "epoch": 0.9121546209337145, + "grad_norm": 2.2459778785705566, + "learning_rate": 1.7593566852937169e-06, + "loss": 2.4575832366943358, + "step": 335080 + }, + { + "epoch": 0.912209064924459, + "grad_norm": 2.2969372272491455, + "learning_rate": 1.7582663213847623e-06, + "loss": 2.465738868713379, + "step": 335100 + }, + { + "epoch": 0.9122635089152035, + "grad_norm": 2.275634288787842, + "learning_rate": 1.7571759574758076e-06, + "loss": 2.4526010513305665, + "step": 335120 + }, + { + "epoch": 0.912317952905948, + "grad_norm": 2.1071438789367676, + "learning_rate": 1.756085593566853e-06, + "loss": 2.4190557479858397, + "step": 335140 + }, + { + "epoch": 0.9123723968966925, + "grad_norm": 2.7492001056671143, + "learning_rate": 1.7549952296578984e-06, + "loss": 2.417451858520508, + "step": 335160 + }, + { + "epoch": 0.9124268408874371, + "grad_norm": 2.1312596797943115, + "learning_rate": 1.7539048657489438e-06, + "loss": 2.4554956436157225, + "step": 335180 + }, + { + "epoch": 0.9124812848781816, + "grad_norm": 2.1107003688812256, + "learning_rate": 1.7528145018399892e-06, + "loss": 2.4277580261230467, + "step": 335200 + }, + { + "epoch": 0.9125357288689261, + "grad_norm": 2.3441689014434814, + "learning_rate": 1.7517241379310346e-06, + "loss": 2.4468318939208986, + "step": 335220 + }, + { + "epoch": 0.9125901728596706, + "grad_norm": 2.1417486667633057, + "learning_rate": 1.75063377402208e-06, + "loss": 2.457892417907715, + "step": 335240 + }, + { + "epoch": 0.9126446168504151, + "grad_norm": 2.124021053314209, + "learning_rate": 1.7495434101131253e-06, + "loss": 2.4578872680664063, + "step": 335260 + }, + { + "epoch": 0.9126990608411597, + "grad_norm": 2.194331645965576, + "learning_rate": 1.7484530462041707e-06, + "loss": 2.491381072998047, + "step": 335280 + }, + { + "epoch": 0.9127535048319042, + "grad_norm": 2.2496020793914795, + "learning_rate": 1.7473626822952161e-06, + "loss": 2.468398094177246, + "step": 335300 + }, + { + "epoch": 0.9128079488226487, + "grad_norm": 2.6966893672943115, + "learning_rate": 1.7462723183862615e-06, + "loss": 2.4654006958007812, + "step": 335320 + }, + { + "epoch": 0.9128623928133932, + "grad_norm": 2.2077677249908447, + "learning_rate": 1.7451819544773069e-06, + "loss": 2.46605224609375, + "step": 335340 + }, + { + "epoch": 0.9129168368041377, + "grad_norm": 2.2261455059051514, + "learning_rate": 1.7440915905683523e-06, + "loss": 2.452287483215332, + "step": 335360 + }, + { + "epoch": 0.9129712807948822, + "grad_norm": 2.1998114585876465, + "learning_rate": 1.7430012266593977e-06, + "loss": 2.458436393737793, + "step": 335380 + }, + { + "epoch": 0.9130257247856268, + "grad_norm": 2.1295130252838135, + "learning_rate": 1.741910862750443e-06, + "loss": 2.4351200103759765, + "step": 335400 + }, + { + "epoch": 0.9130801687763713, + "grad_norm": 2.1594645977020264, + "learning_rate": 1.7408204988414884e-06, + "loss": 2.451560211181641, + "step": 335420 + }, + { + "epoch": 0.9131346127671158, + "grad_norm": 2.145472526550293, + "learning_rate": 1.7397301349325338e-06, + "loss": 2.473901557922363, + "step": 335440 + }, + { + "epoch": 0.9131890567578603, + "grad_norm": 2.056753396987915, + "learning_rate": 1.7386397710235792e-06, + "loss": 2.433148002624512, + "step": 335460 + }, + { + "epoch": 0.9132435007486048, + "grad_norm": 2.2501957416534424, + "learning_rate": 1.7375494071146246e-06, + "loss": 2.4478248596191405, + "step": 335480 + }, + { + "epoch": 0.9132979447393494, + "grad_norm": 2.1550145149230957, + "learning_rate": 1.73645904320567e-06, + "loss": 2.4520517349243165, + "step": 335500 + }, + { + "epoch": 0.913352388730094, + "grad_norm": 2.314356565475464, + "learning_rate": 1.7353686792967154e-06, + "loss": 2.455545425415039, + "step": 335520 + }, + { + "epoch": 0.9134068327208384, + "grad_norm": 2.165449857711792, + "learning_rate": 1.7342783153877607e-06, + "loss": 2.5151037216186523, + "step": 335540 + }, + { + "epoch": 0.913461276711583, + "grad_norm": 2.182640790939331, + "learning_rate": 1.7331879514788061e-06, + "loss": 2.46872615814209, + "step": 335560 + }, + { + "epoch": 0.9135157207023274, + "grad_norm": 2.4434866905212402, + "learning_rate": 1.7320975875698515e-06, + "loss": 2.4325899124145507, + "step": 335580 + }, + { + "epoch": 0.913570164693072, + "grad_norm": 2.287102699279785, + "learning_rate": 1.731007223660897e-06, + "loss": 2.4564067840576174, + "step": 335600 + }, + { + "epoch": 0.9136246086838166, + "grad_norm": 2.3679471015930176, + "learning_rate": 1.7299168597519423e-06, + "loss": 2.459584617614746, + "step": 335620 + }, + { + "epoch": 0.9136790526745611, + "grad_norm": 2.332838535308838, + "learning_rate": 1.7288264958429877e-06, + "loss": 2.447663116455078, + "step": 335640 + }, + { + "epoch": 0.9137334966653056, + "grad_norm": 2.122788906097412, + "learning_rate": 1.727736131934033e-06, + "loss": 2.4639562606811523, + "step": 335660 + }, + { + "epoch": 0.9137879406560501, + "grad_norm": 2.1790831089019775, + "learning_rate": 1.7266457680250784e-06, + "loss": 2.4498010635375977, + "step": 335680 + }, + { + "epoch": 0.9138423846467946, + "grad_norm": 2.247821807861328, + "learning_rate": 1.7255554041161238e-06, + "loss": 2.4386871337890623, + "step": 335700 + }, + { + "epoch": 0.9138968286375392, + "grad_norm": 2.242020606994629, + "learning_rate": 1.7244650402071692e-06, + "loss": 2.456431007385254, + "step": 335720 + }, + { + "epoch": 0.9139512726282837, + "grad_norm": 2.2385942935943604, + "learning_rate": 1.7233746762982146e-06, + "loss": 2.4482770919799806, + "step": 335740 + }, + { + "epoch": 0.9140057166190282, + "grad_norm": 2.1703929901123047, + "learning_rate": 1.7222843123892602e-06, + "loss": 2.476211738586426, + "step": 335760 + }, + { + "epoch": 0.9140601606097727, + "grad_norm": 2.172481060028076, + "learning_rate": 1.7211939484803056e-06, + "loss": 2.4695926666259767, + "step": 335780 + }, + { + "epoch": 0.9141146046005172, + "grad_norm": 2.216040849685669, + "learning_rate": 1.720103584571351e-06, + "loss": 2.4778491973876955, + "step": 335800 + }, + { + "epoch": 0.9141690485912617, + "grad_norm": 2.091249704360962, + "learning_rate": 1.7190132206623964e-06, + "loss": 2.4227996826171876, + "step": 335820 + }, + { + "epoch": 0.9142234925820063, + "grad_norm": 5.150817394256592, + "learning_rate": 1.7179228567534417e-06, + "loss": 2.461182975769043, + "step": 335840 + }, + { + "epoch": 0.9142779365727508, + "grad_norm": 2.1013214588165283, + "learning_rate": 1.7168324928444871e-06, + "loss": 2.4221975326538088, + "step": 335860 + }, + { + "epoch": 0.9143323805634953, + "grad_norm": 2.108919143676758, + "learning_rate": 1.7157421289355325e-06, + "loss": 2.460637092590332, + "step": 335880 + }, + { + "epoch": 0.9143868245542398, + "grad_norm": 2.2599775791168213, + "learning_rate": 1.714651765026578e-06, + "loss": 2.461884880065918, + "step": 335900 + }, + { + "epoch": 0.9144412685449843, + "grad_norm": 2.13987398147583, + "learning_rate": 1.7135614011176233e-06, + "loss": 2.4535593032836913, + "step": 335920 + }, + { + "epoch": 0.9144957125357289, + "grad_norm": 2.246454954147339, + "learning_rate": 1.7124710372086687e-06, + "loss": 2.427115821838379, + "step": 335940 + }, + { + "epoch": 0.9145501565264734, + "grad_norm": 2.1290109157562256, + "learning_rate": 1.711380673299714e-06, + "loss": 2.4663524627685547, + "step": 335960 + }, + { + "epoch": 0.9146046005172179, + "grad_norm": 2.062758684158325, + "learning_rate": 1.7102903093907594e-06, + "loss": 2.4452667236328125, + "step": 335980 + }, + { + "epoch": 0.9146590445079624, + "grad_norm": 2.1132614612579346, + "learning_rate": 1.7091999454818048e-06, + "loss": 2.4352190017700197, + "step": 336000 + }, + { + "epoch": 0.9147134884987069, + "grad_norm": 2.156541109085083, + "learning_rate": 1.7081095815728502e-06, + "loss": 2.4719593048095705, + "step": 336020 + }, + { + "epoch": 0.9147679324894514, + "grad_norm": 2.134988307952881, + "learning_rate": 1.7070192176638956e-06, + "loss": 2.436207389831543, + "step": 336040 + }, + { + "epoch": 0.914822376480196, + "grad_norm": 2.2282094955444336, + "learning_rate": 1.705928853754941e-06, + "loss": 2.486520004272461, + "step": 336060 + }, + { + "epoch": 0.9148768204709405, + "grad_norm": 2.148207187652588, + "learning_rate": 1.7048384898459864e-06, + "loss": 2.4434350967407226, + "step": 336080 + }, + { + "epoch": 0.914931264461685, + "grad_norm": 2.0428106784820557, + "learning_rate": 1.7037481259370318e-06, + "loss": 2.4649003982543944, + "step": 336100 + }, + { + "epoch": 0.9149857084524295, + "grad_norm": 2.1738317012786865, + "learning_rate": 1.7026577620280771e-06, + "loss": 2.4318891525268556, + "step": 336120 + }, + { + "epoch": 0.915040152443174, + "grad_norm": 2.222489833831787, + "learning_rate": 1.7015673981191225e-06, + "loss": 2.49435977935791, + "step": 336140 + }, + { + "epoch": 0.9150945964339187, + "grad_norm": 2.227182149887085, + "learning_rate": 1.700477034210168e-06, + "loss": 2.447536087036133, + "step": 336160 + }, + { + "epoch": 0.9151490404246632, + "grad_norm": 2.3605544567108154, + "learning_rate": 1.6993866703012133e-06, + "loss": 2.4473575592041015, + "step": 336180 + }, + { + "epoch": 0.9152034844154077, + "grad_norm": 2.2255971431732178, + "learning_rate": 1.6982963063922587e-06, + "loss": 2.4673919677734375, + "step": 336200 + }, + { + "epoch": 0.9152579284061522, + "grad_norm": 2.2757151126861572, + "learning_rate": 1.697205942483304e-06, + "loss": 2.468960189819336, + "step": 336220 + }, + { + "epoch": 0.9153123723968967, + "grad_norm": 2.1684393882751465, + "learning_rate": 1.6961155785743495e-06, + "loss": 2.5087329864501955, + "step": 336240 + }, + { + "epoch": 0.9153668163876412, + "grad_norm": 2.2489612102508545, + "learning_rate": 1.6950252146653948e-06, + "loss": 2.4374631881713866, + "step": 336260 + }, + { + "epoch": 0.9154212603783858, + "grad_norm": 2.2535359859466553, + "learning_rate": 1.6939348507564402e-06, + "loss": 2.4392261505126953, + "step": 336280 + }, + { + "epoch": 0.9154757043691303, + "grad_norm": 2.1463496685028076, + "learning_rate": 1.6928444868474856e-06, + "loss": 2.4670392990112306, + "step": 336300 + }, + { + "epoch": 0.9155301483598748, + "grad_norm": 2.0885913372039795, + "learning_rate": 1.691754122938531e-06, + "loss": 2.4672401428222654, + "step": 336320 + }, + { + "epoch": 0.9155845923506193, + "grad_norm": 2.5777266025543213, + "learning_rate": 1.6906637590295764e-06, + "loss": 2.4532211303710936, + "step": 336340 + }, + { + "epoch": 0.9156390363413638, + "grad_norm": 2.209062337875366, + "learning_rate": 1.6895733951206218e-06, + "loss": 2.473771858215332, + "step": 336360 + }, + { + "epoch": 0.9156934803321083, + "grad_norm": 2.118757724761963, + "learning_rate": 1.6884830312116672e-06, + "loss": 2.465814971923828, + "step": 336380 + }, + { + "epoch": 0.9157479243228529, + "grad_norm": 2.2169461250305176, + "learning_rate": 1.6873926673027125e-06, + "loss": 2.4601722717285157, + "step": 336400 + }, + { + "epoch": 0.9158023683135974, + "grad_norm": 2.1855597496032715, + "learning_rate": 1.686302303393758e-06, + "loss": 2.457229804992676, + "step": 336420 + }, + { + "epoch": 0.9158568123043419, + "grad_norm": 2.2431647777557373, + "learning_rate": 1.6852119394848033e-06, + "loss": 2.433224105834961, + "step": 336440 + }, + { + "epoch": 0.9159112562950864, + "grad_norm": 2.262021064758301, + "learning_rate": 1.6841215755758487e-06, + "loss": 2.466252326965332, + "step": 336460 + }, + { + "epoch": 0.9159657002858309, + "grad_norm": 2.3619656562805176, + "learning_rate": 1.683031211666894e-06, + "loss": 2.443251609802246, + "step": 336480 + }, + { + "epoch": 0.9160201442765755, + "grad_norm": 2.1097500324249268, + "learning_rate": 1.6819408477579395e-06, + "loss": 2.435862922668457, + "step": 336500 + }, + { + "epoch": 0.91607458826732, + "grad_norm": 2.196214437484741, + "learning_rate": 1.6808504838489849e-06, + "loss": 2.4652593612670897, + "step": 336520 + }, + { + "epoch": 0.9161290322580645, + "grad_norm": 2.7603416442871094, + "learning_rate": 1.67976011994003e-06, + "loss": 2.4364675521850585, + "step": 336540 + }, + { + "epoch": 0.916183476248809, + "grad_norm": 2.183602809906006, + "learning_rate": 1.6786697560310754e-06, + "loss": 2.481303024291992, + "step": 336560 + }, + { + "epoch": 0.9162379202395535, + "grad_norm": 2.145059823989868, + "learning_rate": 1.6775793921221208e-06, + "loss": 2.422413444519043, + "step": 336580 + }, + { + "epoch": 0.916292364230298, + "grad_norm": 2.287376880645752, + "learning_rate": 1.6764890282131662e-06, + "loss": 2.4514181137084963, + "step": 336600 + }, + { + "epoch": 0.9163468082210426, + "grad_norm": 2.218745470046997, + "learning_rate": 1.6753986643042116e-06, + "loss": 2.417209434509277, + "step": 336620 + }, + { + "epoch": 0.9164012522117871, + "grad_norm": 2.1787610054016113, + "learning_rate": 1.674308300395257e-06, + "loss": 2.429497718811035, + "step": 336640 + }, + { + "epoch": 0.9164556962025316, + "grad_norm": 2.152329921722412, + "learning_rate": 1.6732179364863023e-06, + "loss": 2.4363765716552734, + "step": 336660 + }, + { + "epoch": 0.9165101401932761, + "grad_norm": 2.295255184173584, + "learning_rate": 1.6721275725773477e-06, + "loss": 2.4418020248413086, + "step": 336680 + }, + { + "epoch": 0.9165645841840206, + "grad_norm": 2.189162492752075, + "learning_rate": 1.6710372086683931e-06, + "loss": 2.4427528381347656, + "step": 336700 + }, + { + "epoch": 0.9166190281747653, + "grad_norm": 2.1997015476226807, + "learning_rate": 1.6699468447594385e-06, + "loss": 2.470888137817383, + "step": 336720 + }, + { + "epoch": 0.9166734721655098, + "grad_norm": 2.1269376277923584, + "learning_rate": 1.6688564808504839e-06, + "loss": 2.4527332305908205, + "step": 336740 + }, + { + "epoch": 0.9167279161562543, + "grad_norm": 2.172860622406006, + "learning_rate": 1.6677661169415293e-06, + "loss": 2.460366439819336, + "step": 336760 + }, + { + "epoch": 0.9167823601469988, + "grad_norm": 2.6224825382232666, + "learning_rate": 1.6666757530325747e-06, + "loss": 2.4547164916992186, + "step": 336780 + }, + { + "epoch": 0.9168368041377433, + "grad_norm": 2.1935031414031982, + "learning_rate": 1.66558538912362e-06, + "loss": 2.4432554244995117, + "step": 336800 + }, + { + "epoch": 0.9168912481284878, + "grad_norm": 2.6544458866119385, + "learning_rate": 1.6644950252146654e-06, + "loss": 2.4262012481689452, + "step": 336820 + }, + { + "epoch": 0.9169456921192324, + "grad_norm": 2.2295706272125244, + "learning_rate": 1.6634046613057108e-06, + "loss": 2.4265535354614256, + "step": 336840 + }, + { + "epoch": 0.9170001361099769, + "grad_norm": 2.1299149990081787, + "learning_rate": 1.6623142973967562e-06, + "loss": 2.429402160644531, + "step": 336860 + }, + { + "epoch": 0.9170545801007214, + "grad_norm": 2.1549675464630127, + "learning_rate": 1.6612239334878016e-06, + "loss": 2.4509164810180666, + "step": 336880 + }, + { + "epoch": 0.9171090240914659, + "grad_norm": 2.1572675704956055, + "learning_rate": 1.660133569578847e-06, + "loss": 2.461958885192871, + "step": 336900 + }, + { + "epoch": 0.9171634680822104, + "grad_norm": 2.22823429107666, + "learning_rate": 1.6590432056698924e-06, + "loss": 2.4643814086914064, + "step": 336920 + }, + { + "epoch": 0.917217912072955, + "grad_norm": 1.9993964433670044, + "learning_rate": 1.6579528417609377e-06, + "loss": 2.4329042434692383, + "step": 336940 + }, + { + "epoch": 0.9172723560636995, + "grad_norm": 2.1942334175109863, + "learning_rate": 1.6568624778519831e-06, + "loss": 2.435124969482422, + "step": 336960 + }, + { + "epoch": 0.917326800054444, + "grad_norm": 2.2449324131011963, + "learning_rate": 1.6557721139430285e-06, + "loss": 2.4448030471801756, + "step": 336980 + }, + { + "epoch": 0.9173812440451885, + "grad_norm": 2.0833263397216797, + "learning_rate": 1.654681750034074e-06, + "loss": 2.4503787994384765, + "step": 337000 + }, + { + "epoch": 0.917435688035933, + "grad_norm": 2.091477155685425, + "learning_rate": 1.6535913861251193e-06, + "loss": 2.447056007385254, + "step": 337020 + }, + { + "epoch": 0.9174901320266775, + "grad_norm": 2.320767641067505, + "learning_rate": 1.6525010222161647e-06, + "loss": 2.4573646545410157, + "step": 337040 + }, + { + "epoch": 0.9175445760174221, + "grad_norm": 2.2191689014434814, + "learning_rate": 1.65141065830721e-06, + "loss": 2.4654413223266602, + "step": 337060 + }, + { + "epoch": 0.9175990200081666, + "grad_norm": 2.21181583404541, + "learning_rate": 1.6503202943982554e-06, + "loss": 2.448031425476074, + "step": 337080 + }, + { + "epoch": 0.9176534639989111, + "grad_norm": 2.439760208129883, + "learning_rate": 1.6492299304893008e-06, + "loss": 2.4751033782958984, + "step": 337100 + }, + { + "epoch": 0.9177079079896556, + "grad_norm": 2.1833550930023193, + "learning_rate": 1.6481395665803462e-06, + "loss": 2.4330814361572264, + "step": 337120 + }, + { + "epoch": 0.9177623519804001, + "grad_norm": 2.2249765396118164, + "learning_rate": 1.6470492026713916e-06, + "loss": 2.4188772201538087, + "step": 337140 + }, + { + "epoch": 0.9178167959711447, + "grad_norm": 2.3139779567718506, + "learning_rate": 1.645958838762437e-06, + "loss": 2.449759101867676, + "step": 337160 + }, + { + "epoch": 0.9178712399618892, + "grad_norm": 2.1808221340179443, + "learning_rate": 1.6448684748534824e-06, + "loss": 2.464913749694824, + "step": 337180 + }, + { + "epoch": 0.9179256839526337, + "grad_norm": 2.2651515007019043, + "learning_rate": 1.6437781109445278e-06, + "loss": 2.4744462966918945, + "step": 337200 + }, + { + "epoch": 0.9179801279433782, + "grad_norm": 2.2426040172576904, + "learning_rate": 1.6426877470355731e-06, + "loss": 2.4485713958740236, + "step": 337220 + }, + { + "epoch": 0.9180345719341227, + "grad_norm": 2.209425687789917, + "learning_rate": 1.6415973831266185e-06, + "loss": 2.4378496170043946, + "step": 337240 + }, + { + "epoch": 0.9180890159248672, + "grad_norm": 2.2196767330169678, + "learning_rate": 1.640507019217664e-06, + "loss": 2.4446054458618165, + "step": 337260 + }, + { + "epoch": 0.9181434599156119, + "grad_norm": 2.3348283767700195, + "learning_rate": 1.6394166553087093e-06, + "loss": 2.4687000274658204, + "step": 337280 + }, + { + "epoch": 0.9181979039063564, + "grad_norm": 2.1176483631134033, + "learning_rate": 1.6383262913997547e-06, + "loss": 2.4604118347167967, + "step": 337300 + }, + { + "epoch": 0.9182523478971009, + "grad_norm": 2.444187879562378, + "learning_rate": 1.6372359274908e-06, + "loss": 2.4323278427124024, + "step": 337320 + }, + { + "epoch": 0.9183067918878454, + "grad_norm": 2.1245009899139404, + "learning_rate": 1.6361455635818455e-06, + "loss": 2.465196990966797, + "step": 337340 + }, + { + "epoch": 0.9183612358785899, + "grad_norm": 2.191099166870117, + "learning_rate": 1.6350551996728908e-06, + "loss": 2.4795434951782225, + "step": 337360 + }, + { + "epoch": 0.9184156798693344, + "grad_norm": 3.3995678424835205, + "learning_rate": 1.6339648357639362e-06, + "loss": 2.42464542388916, + "step": 337380 + }, + { + "epoch": 0.918470123860079, + "grad_norm": 2.165790557861328, + "learning_rate": 1.6328744718549816e-06, + "loss": 2.4505462646484375, + "step": 337400 + }, + { + "epoch": 0.9185245678508235, + "grad_norm": 2.146594524383545, + "learning_rate": 1.631784107946027e-06, + "loss": 2.4311676025390625, + "step": 337420 + }, + { + "epoch": 0.918579011841568, + "grad_norm": 2.199666738510132, + "learning_rate": 1.6306937440370724e-06, + "loss": 2.4768321990966795, + "step": 337440 + }, + { + "epoch": 0.9186334558323125, + "grad_norm": 2.1685662269592285, + "learning_rate": 1.629603380128118e-06, + "loss": 2.423782730102539, + "step": 337460 + }, + { + "epoch": 0.918687899823057, + "grad_norm": 2.163461446762085, + "learning_rate": 1.6285130162191634e-06, + "loss": 2.4851417541503906, + "step": 337480 + }, + { + "epoch": 0.9187423438138016, + "grad_norm": 2.160773277282715, + "learning_rate": 1.6274226523102088e-06, + "loss": 2.4459815979003907, + "step": 337500 + }, + { + "epoch": 0.9187967878045461, + "grad_norm": 2.107217311859131, + "learning_rate": 1.6263322884012541e-06, + "loss": 2.461981010437012, + "step": 337520 + }, + { + "epoch": 0.9188512317952906, + "grad_norm": 2.0579066276550293, + "learning_rate": 1.6252419244922995e-06, + "loss": 2.4724365234375, + "step": 337540 + }, + { + "epoch": 0.9189056757860351, + "grad_norm": 2.1660962104797363, + "learning_rate": 1.624151560583345e-06, + "loss": 2.458013343811035, + "step": 337560 + }, + { + "epoch": 0.9189601197767796, + "grad_norm": 2.1978025436401367, + "learning_rate": 1.6230611966743903e-06, + "loss": 2.4405778884887694, + "step": 337580 + }, + { + "epoch": 0.9190145637675241, + "grad_norm": 2.3108174800872803, + "learning_rate": 1.6219708327654357e-06, + "loss": 2.4570232391357423, + "step": 337600 + }, + { + "epoch": 0.9190690077582687, + "grad_norm": 2.13663911819458, + "learning_rate": 1.620880468856481e-06, + "loss": 2.452432060241699, + "step": 337620 + }, + { + "epoch": 0.9191234517490132, + "grad_norm": 2.16794753074646, + "learning_rate": 1.6197901049475265e-06, + "loss": 2.4189426422119142, + "step": 337640 + }, + { + "epoch": 0.9191778957397577, + "grad_norm": 2.2958502769470215, + "learning_rate": 1.6186997410385718e-06, + "loss": 2.4406787872314455, + "step": 337660 + }, + { + "epoch": 0.9192323397305022, + "grad_norm": 2.1163077354431152, + "learning_rate": 1.6176093771296172e-06, + "loss": 2.434347724914551, + "step": 337680 + }, + { + "epoch": 0.9192867837212467, + "grad_norm": 2.111454963684082, + "learning_rate": 1.6165190132206626e-06, + "loss": 2.467413902282715, + "step": 337700 + }, + { + "epoch": 0.9193412277119913, + "grad_norm": 2.5992085933685303, + "learning_rate": 1.615428649311708e-06, + "loss": 2.488700103759766, + "step": 337720 + }, + { + "epoch": 0.9193956717027358, + "grad_norm": 2.249971628189087, + "learning_rate": 1.6143382854027534e-06, + "loss": 2.4657426834106446, + "step": 337740 + }, + { + "epoch": 0.9194501156934803, + "grad_norm": 2.145705223083496, + "learning_rate": 1.6132479214937988e-06, + "loss": 2.4641801834106447, + "step": 337760 + }, + { + "epoch": 0.9195045596842248, + "grad_norm": 2.1588447093963623, + "learning_rate": 1.6121575575848442e-06, + "loss": 2.467062759399414, + "step": 337780 + }, + { + "epoch": 0.9195590036749693, + "grad_norm": 2.2640633583068848, + "learning_rate": 1.6110671936758895e-06, + "loss": 2.486491012573242, + "step": 337800 + }, + { + "epoch": 0.9196134476657138, + "grad_norm": 2.6902365684509277, + "learning_rate": 1.609976829766935e-06, + "loss": 2.453923988342285, + "step": 337820 + }, + { + "epoch": 0.9196678916564585, + "grad_norm": 2.3422763347625732, + "learning_rate": 1.6088864658579803e-06, + "loss": 2.451644515991211, + "step": 337840 + }, + { + "epoch": 0.919722335647203, + "grad_norm": 2.1171867847442627, + "learning_rate": 1.6077961019490257e-06, + "loss": 2.473830986022949, + "step": 337860 + }, + { + "epoch": 0.9197767796379475, + "grad_norm": 2.22298526763916, + "learning_rate": 1.606705738040071e-06, + "loss": 2.485067367553711, + "step": 337880 + }, + { + "epoch": 0.919831223628692, + "grad_norm": 2.1786348819732666, + "learning_rate": 1.6056153741311165e-06, + "loss": 2.46085205078125, + "step": 337900 + }, + { + "epoch": 0.9198856676194365, + "grad_norm": 2.0339272022247314, + "learning_rate": 1.6045250102221619e-06, + "loss": 2.436570739746094, + "step": 337920 + }, + { + "epoch": 0.9199401116101811, + "grad_norm": 2.0123538970947266, + "learning_rate": 1.6034346463132072e-06, + "loss": 2.4544816970825196, + "step": 337940 + }, + { + "epoch": 0.9199945556009256, + "grad_norm": 2.1765880584716797, + "learning_rate": 1.6023442824042526e-06, + "loss": 2.466366195678711, + "step": 337960 + }, + { + "epoch": 0.9200489995916701, + "grad_norm": 2.199058771133423, + "learning_rate": 1.601253918495298e-06, + "loss": 2.456272506713867, + "step": 337980 + }, + { + "epoch": 0.9201034435824146, + "grad_norm": 3.5001332759857178, + "learning_rate": 1.6001635545863434e-06, + "loss": 2.4365009307861327, + "step": 338000 + }, + { + "epoch": 0.9201578875731591, + "grad_norm": 2.1556665897369385, + "learning_rate": 1.5990731906773888e-06, + "loss": 2.4543533325195312, + "step": 338020 + }, + { + "epoch": 0.9202123315639036, + "grad_norm": 2.201931953430176, + "learning_rate": 1.5979828267684342e-06, + "loss": 2.4617918014526365, + "step": 338040 + }, + { + "epoch": 0.9202667755546482, + "grad_norm": 2.2244300842285156, + "learning_rate": 1.5968924628594796e-06, + "loss": 2.481087303161621, + "step": 338060 + }, + { + "epoch": 0.9203212195453927, + "grad_norm": 2.2155749797821045, + "learning_rate": 1.595802098950525e-06, + "loss": 2.465167427062988, + "step": 338080 + }, + { + "epoch": 0.9203756635361372, + "grad_norm": 2.2548444271087646, + "learning_rate": 1.5947117350415703e-06, + "loss": 2.4696430206298827, + "step": 338100 + }, + { + "epoch": 0.9204301075268817, + "grad_norm": 2.1879987716674805, + "learning_rate": 1.5936213711326157e-06, + "loss": 2.4579925537109375, + "step": 338120 + }, + { + "epoch": 0.9204845515176262, + "grad_norm": 2.1385741233825684, + "learning_rate": 1.592531007223661e-06, + "loss": 2.4422937393188477, + "step": 338140 + }, + { + "epoch": 0.9205389955083708, + "grad_norm": 2.1758029460906982, + "learning_rate": 1.5914406433147065e-06, + "loss": 2.4600173950195314, + "step": 338160 + }, + { + "epoch": 0.9205934394991153, + "grad_norm": 2.2345213890075684, + "learning_rate": 1.5903502794057519e-06, + "loss": 2.4381645202636717, + "step": 338180 + }, + { + "epoch": 0.9206478834898598, + "grad_norm": 2.2245121002197266, + "learning_rate": 1.5892599154967973e-06, + "loss": 2.4732532501220703, + "step": 338200 + }, + { + "epoch": 0.9207023274806043, + "grad_norm": 2.1458301544189453, + "learning_rate": 1.5881695515878426e-06, + "loss": 2.4701066970825196, + "step": 338220 + }, + { + "epoch": 0.9207567714713488, + "grad_norm": 2.199620246887207, + "learning_rate": 1.587079187678888e-06, + "loss": 2.46347599029541, + "step": 338240 + }, + { + "epoch": 0.9208112154620933, + "grad_norm": 2.184943199157715, + "learning_rate": 1.5859888237699334e-06, + "loss": 2.46417293548584, + "step": 338260 + }, + { + "epoch": 0.9208656594528379, + "grad_norm": 2.259795904159546, + "learning_rate": 1.5848984598609788e-06, + "loss": 2.443757247924805, + "step": 338280 + }, + { + "epoch": 0.9209201034435824, + "grad_norm": 2.1802632808685303, + "learning_rate": 1.5838080959520242e-06, + "loss": 2.465754508972168, + "step": 338300 + }, + { + "epoch": 0.9209745474343269, + "grad_norm": 2.102527379989624, + "learning_rate": 1.5827177320430696e-06, + "loss": 2.442578125, + "step": 338320 + }, + { + "epoch": 0.9210289914250714, + "grad_norm": 2.130995750427246, + "learning_rate": 1.581627368134115e-06, + "loss": 2.4569782257080077, + "step": 338340 + }, + { + "epoch": 0.9210834354158159, + "grad_norm": 2.19991397857666, + "learning_rate": 1.5805370042251603e-06, + "loss": 2.4611913681030275, + "step": 338360 + }, + { + "epoch": 0.9211378794065606, + "grad_norm": 2.2011778354644775, + "learning_rate": 1.5794466403162057e-06, + "loss": 2.4514497756958007, + "step": 338380 + }, + { + "epoch": 0.9211923233973051, + "grad_norm": 2.2067956924438477, + "learning_rate": 1.5783562764072511e-06, + "loss": 2.463110160827637, + "step": 338400 + }, + { + "epoch": 0.9212467673880496, + "grad_norm": 2.1454761028289795, + "learning_rate": 1.5772659124982965e-06, + "loss": 2.434858512878418, + "step": 338420 + }, + { + "epoch": 0.9213012113787941, + "grad_norm": 2.2948925495147705, + "learning_rate": 1.5761755485893419e-06, + "loss": 2.479843521118164, + "step": 338440 + }, + { + "epoch": 0.9213556553695386, + "grad_norm": 2.1936464309692383, + "learning_rate": 1.5750851846803873e-06, + "loss": 2.435364532470703, + "step": 338460 + }, + { + "epoch": 0.9214100993602831, + "grad_norm": 2.1329429149627686, + "learning_rate": 1.5739948207714327e-06, + "loss": 2.464430236816406, + "step": 338480 + }, + { + "epoch": 0.9214645433510277, + "grad_norm": 2.1219427585601807, + "learning_rate": 1.572904456862478e-06, + "loss": 2.443027114868164, + "step": 338500 + }, + { + "epoch": 0.9215189873417722, + "grad_norm": 2.3751220703125, + "learning_rate": 1.5718140929535234e-06, + "loss": 2.458546447753906, + "step": 338520 + }, + { + "epoch": 0.9215734313325167, + "grad_norm": 2.2009761333465576, + "learning_rate": 1.5707237290445688e-06, + "loss": 2.455882453918457, + "step": 338540 + }, + { + "epoch": 0.9216278753232612, + "grad_norm": 2.177001476287842, + "learning_rate": 1.5696333651356142e-06, + "loss": 2.468668556213379, + "step": 338560 + }, + { + "epoch": 0.9216823193140057, + "grad_norm": 2.2398009300231934, + "learning_rate": 1.5685430012266596e-06, + "loss": 2.459335517883301, + "step": 338580 + }, + { + "epoch": 0.9217367633047502, + "grad_norm": 2.062504529953003, + "learning_rate": 1.567452637317705e-06, + "loss": 2.4425697326660156, + "step": 338600 + }, + { + "epoch": 0.9217912072954948, + "grad_norm": 2.3271360397338867, + "learning_rate": 1.5663622734087501e-06, + "loss": 2.473727798461914, + "step": 338620 + }, + { + "epoch": 0.9218456512862393, + "grad_norm": 2.168670177459717, + "learning_rate": 1.5652719094997955e-06, + "loss": 2.417561912536621, + "step": 338640 + }, + { + "epoch": 0.9219000952769838, + "grad_norm": 2.1322367191314697, + "learning_rate": 1.564181545590841e-06, + "loss": 2.466874122619629, + "step": 338660 + }, + { + "epoch": 0.9219545392677283, + "grad_norm": 2.2299602031707764, + "learning_rate": 1.5630911816818863e-06, + "loss": 2.457337188720703, + "step": 338680 + }, + { + "epoch": 0.9220089832584728, + "grad_norm": 2.1421844959259033, + "learning_rate": 1.5620008177729317e-06, + "loss": 2.466720962524414, + "step": 338700 + }, + { + "epoch": 0.9220634272492174, + "grad_norm": 2.0999069213867188, + "learning_rate": 1.560910453863977e-06, + "loss": 2.448742485046387, + "step": 338720 + }, + { + "epoch": 0.9221178712399619, + "grad_norm": 2.1080574989318848, + "learning_rate": 1.5598200899550225e-06, + "loss": 2.4518714904785157, + "step": 338740 + }, + { + "epoch": 0.9221723152307064, + "grad_norm": 2.300995349884033, + "learning_rate": 1.5587297260460678e-06, + "loss": 2.4715227127075194, + "step": 338760 + }, + { + "epoch": 0.9222267592214509, + "grad_norm": 2.1940932273864746, + "learning_rate": 1.5576393621371132e-06, + "loss": 2.468620491027832, + "step": 338780 + }, + { + "epoch": 0.9222812032121954, + "grad_norm": 2.1421306133270264, + "learning_rate": 1.5565489982281586e-06, + "loss": 2.4735490798950197, + "step": 338800 + }, + { + "epoch": 0.9223356472029399, + "grad_norm": 2.346722364425659, + "learning_rate": 1.555458634319204e-06, + "loss": 2.4406509399414062, + "step": 338820 + }, + { + "epoch": 0.9223900911936845, + "grad_norm": 2.1967780590057373, + "learning_rate": 1.5543682704102494e-06, + "loss": 2.4459484100341795, + "step": 338840 + }, + { + "epoch": 0.922444535184429, + "grad_norm": 2.1212968826293945, + "learning_rate": 1.5532779065012948e-06, + "loss": 2.458100509643555, + "step": 338860 + }, + { + "epoch": 0.9224989791751735, + "grad_norm": 2.16797137260437, + "learning_rate": 1.5521875425923402e-06, + "loss": 2.462246894836426, + "step": 338880 + }, + { + "epoch": 0.922553423165918, + "grad_norm": 2.101433753967285, + "learning_rate": 1.5510971786833855e-06, + "loss": 2.4614696502685547, + "step": 338900 + }, + { + "epoch": 0.9226078671566625, + "grad_norm": 2.1748547554016113, + "learning_rate": 1.550006814774431e-06, + "loss": 2.456348991394043, + "step": 338920 + }, + { + "epoch": 0.9226623111474072, + "grad_norm": 2.15600323677063, + "learning_rate": 1.5489164508654763e-06, + "loss": 2.4677146911621093, + "step": 338940 + }, + { + "epoch": 0.9227167551381517, + "grad_norm": 2.2045531272888184, + "learning_rate": 1.5478260869565217e-06, + "loss": 2.4852231979370116, + "step": 338960 + }, + { + "epoch": 0.9227711991288962, + "grad_norm": 2.2241735458374023, + "learning_rate": 1.546735723047567e-06, + "loss": 2.4295291900634766, + "step": 338980 + }, + { + "epoch": 0.9228256431196407, + "grad_norm": 2.256232976913452, + "learning_rate": 1.5456453591386125e-06, + "loss": 2.4844919204711915, + "step": 339000 + }, + { + "epoch": 0.9228800871103852, + "grad_norm": 2.145331382751465, + "learning_rate": 1.5445549952296579e-06, + "loss": 2.446238708496094, + "step": 339020 + }, + { + "epoch": 0.9229345311011297, + "grad_norm": 2.2027831077575684, + "learning_rate": 1.5434646313207032e-06, + "loss": 2.4755325317382812, + "step": 339040 + }, + { + "epoch": 0.9229889750918743, + "grad_norm": 2.214470624923706, + "learning_rate": 1.5423742674117486e-06, + "loss": 2.4488155364990236, + "step": 339060 + }, + { + "epoch": 0.9230434190826188, + "grad_norm": 2.199159622192383, + "learning_rate": 1.541283903502794e-06, + "loss": 2.484781837463379, + "step": 339080 + }, + { + "epoch": 0.9230978630733633, + "grad_norm": 2.209658622741699, + "learning_rate": 1.5401935395938394e-06, + "loss": 2.4462419509887696, + "step": 339100 + }, + { + "epoch": 0.9231523070641078, + "grad_norm": 2.2336723804473877, + "learning_rate": 1.5391031756848848e-06, + "loss": 2.4504150390625, + "step": 339120 + }, + { + "epoch": 0.9232067510548523, + "grad_norm": 2.183842897415161, + "learning_rate": 1.5380128117759304e-06, + "loss": 2.4722673416137697, + "step": 339140 + }, + { + "epoch": 0.9232611950455969, + "grad_norm": 2.291206121444702, + "learning_rate": 1.5369224478669758e-06, + "loss": 2.4689586639404295, + "step": 339160 + }, + { + "epoch": 0.9233156390363414, + "grad_norm": 2.1948912143707275, + "learning_rate": 1.5358320839580212e-06, + "loss": 2.455231475830078, + "step": 339180 + }, + { + "epoch": 0.9233700830270859, + "grad_norm": 2.135784149169922, + "learning_rate": 1.5347417200490665e-06, + "loss": 2.4701372146606446, + "step": 339200 + }, + { + "epoch": 0.9234245270178304, + "grad_norm": 2.143982410430908, + "learning_rate": 1.533651356140112e-06, + "loss": 2.4507673263549803, + "step": 339220 + }, + { + "epoch": 0.9234789710085749, + "grad_norm": 2.097395181655884, + "learning_rate": 1.5325609922311573e-06, + "loss": 2.4546092987060546, + "step": 339240 + }, + { + "epoch": 0.9235334149993194, + "grad_norm": 2.14581561088562, + "learning_rate": 1.5314706283222027e-06, + "loss": 2.433406448364258, + "step": 339260 + }, + { + "epoch": 0.923587858990064, + "grad_norm": 2.140432357788086, + "learning_rate": 1.530380264413248e-06, + "loss": 2.4386444091796875, + "step": 339280 + }, + { + "epoch": 0.9236423029808085, + "grad_norm": 2.1307320594787598, + "learning_rate": 1.5292899005042935e-06, + "loss": 2.462544631958008, + "step": 339300 + }, + { + "epoch": 0.923696746971553, + "grad_norm": 2.1548166275024414, + "learning_rate": 1.5281995365953389e-06, + "loss": 2.442224884033203, + "step": 339320 + }, + { + "epoch": 0.9237511909622975, + "grad_norm": 2.1973536014556885, + "learning_rate": 1.5271091726863842e-06, + "loss": 2.4481227874755858, + "step": 339340 + }, + { + "epoch": 0.923805634953042, + "grad_norm": 2.467590570449829, + "learning_rate": 1.5260188087774296e-06, + "loss": 2.433904457092285, + "step": 339360 + }, + { + "epoch": 0.9238600789437866, + "grad_norm": 2.194406270980835, + "learning_rate": 1.524928444868475e-06, + "loss": 2.4339929580688477, + "step": 339380 + }, + { + "epoch": 0.9239145229345311, + "grad_norm": 2.2190911769866943, + "learning_rate": 1.5238380809595204e-06, + "loss": 2.446406936645508, + "step": 339400 + }, + { + "epoch": 0.9239689669252756, + "grad_norm": 2.265545606613159, + "learning_rate": 1.5227477170505658e-06, + "loss": 2.440068817138672, + "step": 339420 + }, + { + "epoch": 0.9240234109160201, + "grad_norm": 2.1126859188079834, + "learning_rate": 1.5216573531416112e-06, + "loss": 2.4467596054077148, + "step": 339440 + }, + { + "epoch": 0.9240778549067646, + "grad_norm": 2.1272165775299072, + "learning_rate": 1.5205669892326566e-06, + "loss": 2.476199150085449, + "step": 339460 + }, + { + "epoch": 0.9241322988975091, + "grad_norm": 2.167335271835327, + "learning_rate": 1.519476625323702e-06, + "loss": 2.485526275634766, + "step": 339480 + }, + { + "epoch": 0.9241867428882538, + "grad_norm": 2.3754587173461914, + "learning_rate": 1.5183862614147473e-06, + "loss": 2.4407224655151367, + "step": 339500 + }, + { + "epoch": 0.9242411868789983, + "grad_norm": 2.166564702987671, + "learning_rate": 1.5172958975057927e-06, + "loss": 2.455112648010254, + "step": 339520 + }, + { + "epoch": 0.9242956308697428, + "grad_norm": 2.209061861038208, + "learning_rate": 1.516205533596838e-06, + "loss": 2.4538200378417967, + "step": 339540 + }, + { + "epoch": 0.9243500748604873, + "grad_norm": 2.369776964187622, + "learning_rate": 1.5151151696878835e-06, + "loss": 2.4814983367919923, + "step": 339560 + }, + { + "epoch": 0.9244045188512318, + "grad_norm": 2.15116810798645, + "learning_rate": 1.5140248057789289e-06, + "loss": 2.4656757354736327, + "step": 339580 + }, + { + "epoch": 0.9244589628419763, + "grad_norm": 2.173788547515869, + "learning_rate": 1.5129344418699743e-06, + "loss": 2.444101333618164, + "step": 339600 + }, + { + "epoch": 0.9245134068327209, + "grad_norm": 2.1688244342803955, + "learning_rate": 1.5118440779610196e-06, + "loss": 2.464558792114258, + "step": 339620 + }, + { + "epoch": 0.9245678508234654, + "grad_norm": 2.2068042755126953, + "learning_rate": 1.510753714052065e-06, + "loss": 2.4602827072143554, + "step": 339640 + }, + { + "epoch": 0.9246222948142099, + "grad_norm": 2.395569324493408, + "learning_rate": 1.5096633501431104e-06, + "loss": 2.5221309661865234, + "step": 339660 + }, + { + "epoch": 0.9246767388049544, + "grad_norm": 2.176290512084961, + "learning_rate": 1.5085729862341558e-06, + "loss": 2.415729522705078, + "step": 339680 + }, + { + "epoch": 0.9247311827956989, + "grad_norm": 2.2747342586517334, + "learning_rate": 1.5074826223252012e-06, + "loss": 2.451853942871094, + "step": 339700 + }, + { + "epoch": 0.9247856267864435, + "grad_norm": 2.1483020782470703, + "learning_rate": 1.5063922584162466e-06, + "loss": 2.443844795227051, + "step": 339720 + }, + { + "epoch": 0.924840070777188, + "grad_norm": 2.123997211456299, + "learning_rate": 1.505301894507292e-06, + "loss": 2.4752851486206056, + "step": 339740 + }, + { + "epoch": 0.9248945147679325, + "grad_norm": 2.0813069343566895, + "learning_rate": 1.5042115305983373e-06, + "loss": 2.4453750610351563, + "step": 339760 + }, + { + "epoch": 0.924948958758677, + "grad_norm": 2.1622066497802734, + "learning_rate": 1.5031211666893827e-06, + "loss": 2.459107208251953, + "step": 339780 + }, + { + "epoch": 0.9250034027494215, + "grad_norm": 2.166455030441284, + "learning_rate": 1.5020308027804281e-06, + "loss": 2.446603775024414, + "step": 339800 + }, + { + "epoch": 0.925057846740166, + "grad_norm": 2.2248940467834473, + "learning_rate": 1.5009404388714735e-06, + "loss": 2.5084768295288087, + "step": 339820 + }, + { + "epoch": 0.9251122907309106, + "grad_norm": 2.1591670513153076, + "learning_rate": 1.4998500749625189e-06, + "loss": 2.465211868286133, + "step": 339840 + }, + { + "epoch": 0.9251667347216551, + "grad_norm": 2.390361785888672, + "learning_rate": 1.4987597110535643e-06, + "loss": 2.476884460449219, + "step": 339860 + }, + { + "epoch": 0.9252211787123996, + "grad_norm": 2.158626079559326, + "learning_rate": 1.4976693471446097e-06, + "loss": 2.4529850006103517, + "step": 339880 + }, + { + "epoch": 0.9252756227031441, + "grad_norm": 2.104466676712036, + "learning_rate": 1.496578983235655e-06, + "loss": 2.4655340194702147, + "step": 339900 + }, + { + "epoch": 0.9253300666938886, + "grad_norm": 2.4188270568847656, + "learning_rate": 1.4954886193267004e-06, + "loss": 2.470931053161621, + "step": 339920 + }, + { + "epoch": 0.9253845106846332, + "grad_norm": 2.1072123050689697, + "learning_rate": 1.4943982554177458e-06, + "loss": 2.4520648956298827, + "step": 339940 + }, + { + "epoch": 0.9254389546753777, + "grad_norm": 2.2052106857299805, + "learning_rate": 1.4933078915087912e-06, + "loss": 2.441959762573242, + "step": 339960 + }, + { + "epoch": 0.9254933986661222, + "grad_norm": 2.1398446559906006, + "learning_rate": 1.4922175275998366e-06, + "loss": 2.4567129135131838, + "step": 339980 + }, + { + "epoch": 0.9255478426568667, + "grad_norm": 2.1127209663391113, + "learning_rate": 1.491127163690882e-06, + "loss": 2.467386245727539, + "step": 340000 + }, + { + "epoch": 0.9256022866476112, + "grad_norm": 2.1345930099487305, + "learning_rate": 1.4900367997819274e-06, + "loss": 2.464441680908203, + "step": 340020 + }, + { + "epoch": 0.9256567306383557, + "grad_norm": 2.209714412689209, + "learning_rate": 1.4889464358729727e-06, + "loss": 2.503788375854492, + "step": 340040 + }, + { + "epoch": 0.9257111746291004, + "grad_norm": 2.206772804260254, + "learning_rate": 1.4878560719640181e-06, + "loss": 2.460630416870117, + "step": 340060 + }, + { + "epoch": 0.9257656186198449, + "grad_norm": 2.176943302154541, + "learning_rate": 1.4867657080550635e-06, + "loss": 2.47645263671875, + "step": 340080 + }, + { + "epoch": 0.9258200626105894, + "grad_norm": 2.573927640914917, + "learning_rate": 1.4856753441461089e-06, + "loss": 2.47085018157959, + "step": 340100 + }, + { + "epoch": 0.9258745066013339, + "grad_norm": 2.188420057296753, + "learning_rate": 1.4845849802371543e-06, + "loss": 2.461654281616211, + "step": 340120 + }, + { + "epoch": 0.9259289505920784, + "grad_norm": 2.1682281494140625, + "learning_rate": 1.4834946163281997e-06, + "loss": 2.4954717636108397, + "step": 340140 + }, + { + "epoch": 0.925983394582823, + "grad_norm": 2.1639723777770996, + "learning_rate": 1.482404252419245e-06, + "loss": 2.4737653732299805, + "step": 340160 + }, + { + "epoch": 0.9260378385735675, + "grad_norm": 2.1521623134613037, + "learning_rate": 1.4813138885102904e-06, + "loss": 2.4414752960205077, + "step": 340180 + }, + { + "epoch": 0.926092282564312, + "grad_norm": 2.1840877532958984, + "learning_rate": 1.4802235246013358e-06, + "loss": 2.4586519241333007, + "step": 340200 + }, + { + "epoch": 0.9261467265550565, + "grad_norm": 2.157975912094116, + "learning_rate": 1.4791331606923812e-06, + "loss": 2.4268939971923826, + "step": 340220 + }, + { + "epoch": 0.926201170545801, + "grad_norm": 2.036367893218994, + "learning_rate": 1.4780427967834266e-06, + "loss": 2.4820070266723633, + "step": 340240 + }, + { + "epoch": 0.9262556145365455, + "grad_norm": 2.179063081741333, + "learning_rate": 1.476952432874472e-06, + "loss": 2.4701324462890626, + "step": 340260 + }, + { + "epoch": 0.9263100585272901, + "grad_norm": 2.254706859588623, + "learning_rate": 1.4758620689655174e-06, + "loss": 2.448742485046387, + "step": 340280 + }, + { + "epoch": 0.9263645025180346, + "grad_norm": 2.1360433101654053, + "learning_rate": 1.4747717050565628e-06, + "loss": 2.468243408203125, + "step": 340300 + }, + { + "epoch": 0.9264189465087791, + "grad_norm": 2.110529899597168, + "learning_rate": 1.4736813411476081e-06, + "loss": 2.4585891723632813, + "step": 340320 + }, + { + "epoch": 0.9264733904995236, + "grad_norm": 2.1308422088623047, + "learning_rate": 1.4725909772386535e-06, + "loss": 2.476977729797363, + "step": 340340 + }, + { + "epoch": 0.9265278344902681, + "grad_norm": 2.2052788734436035, + "learning_rate": 1.471500613329699e-06, + "loss": 2.4763059616088867, + "step": 340360 + }, + { + "epoch": 0.9265822784810127, + "grad_norm": 2.390261173248291, + "learning_rate": 1.4704102494207443e-06, + "loss": 2.4467035293579102, + "step": 340380 + }, + { + "epoch": 0.9266367224717572, + "grad_norm": 2.081829071044922, + "learning_rate": 1.4693198855117897e-06, + "loss": 2.4625934600830077, + "step": 340400 + }, + { + "epoch": 0.9266911664625017, + "grad_norm": 2.2762486934661865, + "learning_rate": 1.468229521602835e-06, + "loss": 2.4346750259399412, + "step": 340420 + }, + { + "epoch": 0.9267456104532462, + "grad_norm": 2.2781736850738525, + "learning_rate": 1.4671391576938805e-06, + "loss": 2.455393600463867, + "step": 340440 + }, + { + "epoch": 0.9268000544439907, + "grad_norm": 2.1809351444244385, + "learning_rate": 1.4660487937849258e-06, + "loss": 2.445695686340332, + "step": 340460 + }, + { + "epoch": 0.9268544984347352, + "grad_norm": 2.267754316329956, + "learning_rate": 1.4649584298759712e-06, + "loss": 2.441493034362793, + "step": 340480 + }, + { + "epoch": 0.9269089424254798, + "grad_norm": 2.265521764755249, + "learning_rate": 1.4638680659670166e-06, + "loss": 2.4516727447509767, + "step": 340500 + }, + { + "epoch": 0.9269633864162243, + "grad_norm": 2.1760826110839844, + "learning_rate": 1.462777702058062e-06, + "loss": 2.4807918548583983, + "step": 340520 + }, + { + "epoch": 0.9270178304069688, + "grad_norm": 2.1960248947143555, + "learning_rate": 1.4616873381491074e-06, + "loss": 2.4582509994506836, + "step": 340540 + }, + { + "epoch": 0.9270722743977133, + "grad_norm": 2.174368381500244, + "learning_rate": 1.4605969742401528e-06, + "loss": 2.43404541015625, + "step": 340560 + }, + { + "epoch": 0.9271267183884578, + "grad_norm": 2.1577186584472656, + "learning_rate": 1.4595066103311981e-06, + "loss": 2.4541908264160157, + "step": 340580 + }, + { + "epoch": 0.9271811623792025, + "grad_norm": 2.105989456176758, + "learning_rate": 1.4584162464222435e-06, + "loss": 2.4802597045898436, + "step": 340600 + }, + { + "epoch": 0.927235606369947, + "grad_norm": 2.0946247577667236, + "learning_rate": 1.457325882513289e-06, + "loss": 2.4726085662841797, + "step": 340620 + }, + { + "epoch": 0.9272900503606915, + "grad_norm": 2.1130263805389404, + "learning_rate": 1.4562355186043343e-06, + "loss": 2.426925277709961, + "step": 340640 + }, + { + "epoch": 0.927344494351436, + "grad_norm": 2.2358462810516357, + "learning_rate": 1.4551451546953797e-06, + "loss": 2.4500709533691407, + "step": 340660 + }, + { + "epoch": 0.9273989383421805, + "grad_norm": 6.233712673187256, + "learning_rate": 1.454054790786425e-06, + "loss": 2.4254032135009767, + "step": 340680 + }, + { + "epoch": 0.927453382332925, + "grad_norm": 2.1217727661132812, + "learning_rate": 1.4529644268774703e-06, + "loss": 2.4893938064575196, + "step": 340700 + }, + { + "epoch": 0.9275078263236696, + "grad_norm": 2.1334147453308105, + "learning_rate": 1.4518740629685156e-06, + "loss": 2.4410003662109374, + "step": 340720 + }, + { + "epoch": 0.9275622703144141, + "grad_norm": 2.144564151763916, + "learning_rate": 1.450783699059561e-06, + "loss": 2.444795608520508, + "step": 340740 + }, + { + "epoch": 0.9276167143051586, + "grad_norm": 2.1862661838531494, + "learning_rate": 1.4496933351506064e-06, + "loss": 2.442826843261719, + "step": 340760 + }, + { + "epoch": 0.9276711582959031, + "grad_norm": 2.2825682163238525, + "learning_rate": 1.4486029712416518e-06, + "loss": 2.4603530883789064, + "step": 340780 + }, + { + "epoch": 0.9277256022866476, + "grad_norm": 2.1008358001708984, + "learning_rate": 1.4475126073326972e-06, + "loss": 2.487883949279785, + "step": 340800 + }, + { + "epoch": 0.9277800462773921, + "grad_norm": 2.284494161605835, + "learning_rate": 1.4464222434237426e-06, + "loss": 2.470454216003418, + "step": 340820 + }, + { + "epoch": 0.9278344902681367, + "grad_norm": 2.1609480381011963, + "learning_rate": 1.4453318795147884e-06, + "loss": 2.478993797302246, + "step": 340840 + }, + { + "epoch": 0.9278889342588812, + "grad_norm": 2.130582332611084, + "learning_rate": 1.4442415156058338e-06, + "loss": 2.4587926864624023, + "step": 340860 + }, + { + "epoch": 0.9279433782496257, + "grad_norm": 2.301842451095581, + "learning_rate": 1.4431511516968791e-06, + "loss": 2.44119930267334, + "step": 340880 + }, + { + "epoch": 0.9279978222403702, + "grad_norm": 3.5482168197631836, + "learning_rate": 1.4420607877879245e-06, + "loss": 2.4343278884887694, + "step": 340900 + }, + { + "epoch": 0.9280522662311147, + "grad_norm": 2.2406370639801025, + "learning_rate": 1.44097042387897e-06, + "loss": 2.4350612640380858, + "step": 340920 + }, + { + "epoch": 0.9281067102218593, + "grad_norm": 2.183027505874634, + "learning_rate": 1.4398800599700153e-06, + "loss": 2.4604394912719725, + "step": 340940 + }, + { + "epoch": 0.9281611542126038, + "grad_norm": 2.2380869388580322, + "learning_rate": 1.4387896960610607e-06, + "loss": 2.4865850448608398, + "step": 340960 + }, + { + "epoch": 0.9282155982033483, + "grad_norm": 2.2287888526916504, + "learning_rate": 1.437699332152106e-06, + "loss": 2.4514083862304688, + "step": 340980 + }, + { + "epoch": 0.9282700421940928, + "grad_norm": 2.1001384258270264, + "learning_rate": 1.4366089682431515e-06, + "loss": 2.469633674621582, + "step": 341000 + }, + { + "epoch": 0.9283244861848373, + "grad_norm": 2.4362218379974365, + "learning_rate": 1.4355186043341968e-06, + "loss": 2.4460771560668944, + "step": 341020 + }, + { + "epoch": 0.9283789301755818, + "grad_norm": 3.4742653369903564, + "learning_rate": 1.4344282404252422e-06, + "loss": 2.4253110885620117, + "step": 341040 + }, + { + "epoch": 0.9284333741663264, + "grad_norm": 2.166907548904419, + "learning_rate": 1.4333378765162876e-06, + "loss": 2.4409725189208986, + "step": 341060 + }, + { + "epoch": 0.9284878181570709, + "grad_norm": 2.2715089321136475, + "learning_rate": 1.432247512607333e-06, + "loss": 2.4185375213623046, + "step": 341080 + }, + { + "epoch": 0.9285422621478154, + "grad_norm": 2.1284003257751465, + "learning_rate": 1.4311571486983784e-06, + "loss": 2.439889335632324, + "step": 341100 + }, + { + "epoch": 0.9285967061385599, + "grad_norm": 2.273660182952881, + "learning_rate": 1.4300667847894238e-06, + "loss": 2.4592708587646483, + "step": 341120 + }, + { + "epoch": 0.9286511501293044, + "grad_norm": 2.1124088764190674, + "learning_rate": 1.4289764208804692e-06, + "loss": 2.4664011001586914, + "step": 341140 + }, + { + "epoch": 0.928705594120049, + "grad_norm": 2.189420461654663, + "learning_rate": 1.4278860569715145e-06, + "loss": 2.4342464447021483, + "step": 341160 + }, + { + "epoch": 0.9287600381107936, + "grad_norm": 2.2691924571990967, + "learning_rate": 1.42679569306256e-06, + "loss": 2.491094779968262, + "step": 341180 + }, + { + "epoch": 0.928814482101538, + "grad_norm": 2.2183549404144287, + "learning_rate": 1.4257053291536051e-06, + "loss": 2.468614387512207, + "step": 341200 + }, + { + "epoch": 0.9288689260922826, + "grad_norm": 2.176668882369995, + "learning_rate": 1.4246149652446505e-06, + "loss": 2.4698118209838866, + "step": 341220 + }, + { + "epoch": 0.9289233700830271, + "grad_norm": 2.20194935798645, + "learning_rate": 1.4235246013356959e-06, + "loss": 2.453536605834961, + "step": 341240 + }, + { + "epoch": 0.9289778140737716, + "grad_norm": 2.2404983043670654, + "learning_rate": 1.4224342374267413e-06, + "loss": 2.4580205917358398, + "step": 341260 + }, + { + "epoch": 0.9290322580645162, + "grad_norm": 2.1642956733703613, + "learning_rate": 1.4213438735177866e-06, + "loss": 2.4597600936889648, + "step": 341280 + }, + { + "epoch": 0.9290867020552607, + "grad_norm": 2.2903618812561035, + "learning_rate": 1.420253509608832e-06, + "loss": 2.4721090316772463, + "step": 341300 + }, + { + "epoch": 0.9291411460460052, + "grad_norm": 2.403428554534912, + "learning_rate": 1.4191631456998774e-06, + "loss": 2.452099609375, + "step": 341320 + }, + { + "epoch": 0.9291955900367497, + "grad_norm": 2.281306266784668, + "learning_rate": 1.4180727817909228e-06, + "loss": 2.453789710998535, + "step": 341340 + }, + { + "epoch": 0.9292500340274942, + "grad_norm": 2.1389503479003906, + "learning_rate": 1.4169824178819682e-06, + "loss": 2.4505851745605467, + "step": 341360 + }, + { + "epoch": 0.9293044780182388, + "grad_norm": 2.1326169967651367, + "learning_rate": 1.4158920539730136e-06, + "loss": 2.4660131454467775, + "step": 341380 + }, + { + "epoch": 0.9293589220089833, + "grad_norm": 2.1423449516296387, + "learning_rate": 1.414801690064059e-06, + "loss": 2.4646984100341798, + "step": 341400 + }, + { + "epoch": 0.9294133659997278, + "grad_norm": 2.125492811203003, + "learning_rate": 1.4137113261551043e-06, + "loss": 2.4756820678710936, + "step": 341420 + }, + { + "epoch": 0.9294678099904723, + "grad_norm": 2.859484910964966, + "learning_rate": 1.4126209622461497e-06, + "loss": 2.4799795150756836, + "step": 341440 + }, + { + "epoch": 0.9295222539812168, + "grad_norm": 2.185234308242798, + "learning_rate": 1.4115305983371951e-06, + "loss": 2.4698919296264648, + "step": 341460 + }, + { + "epoch": 0.9295766979719613, + "grad_norm": 2.1653053760528564, + "learning_rate": 1.4104402344282405e-06, + "loss": 2.4760881423950196, + "step": 341480 + }, + { + "epoch": 0.9296311419627059, + "grad_norm": 2.1691231727600098, + "learning_rate": 1.4093498705192859e-06, + "loss": 2.447560501098633, + "step": 341500 + }, + { + "epoch": 0.9296855859534504, + "grad_norm": 2.2282285690307617, + "learning_rate": 1.4082595066103313e-06, + "loss": 2.4287675857543944, + "step": 341520 + }, + { + "epoch": 0.9297400299441949, + "grad_norm": 2.137817859649658, + "learning_rate": 1.4071691427013767e-06, + "loss": 2.4388233184814454, + "step": 341540 + }, + { + "epoch": 0.9297944739349394, + "grad_norm": 4.627491474151611, + "learning_rate": 1.406078778792422e-06, + "loss": 2.458663749694824, + "step": 341560 + }, + { + "epoch": 0.9298489179256839, + "grad_norm": 2.1662917137145996, + "learning_rate": 1.4049884148834674e-06, + "loss": 2.459979438781738, + "step": 341580 + }, + { + "epoch": 0.9299033619164285, + "grad_norm": 2.17667555809021, + "learning_rate": 1.4038980509745128e-06, + "loss": 2.439846992492676, + "step": 341600 + }, + { + "epoch": 0.929957805907173, + "grad_norm": 2.138317108154297, + "learning_rate": 1.4028076870655582e-06, + "loss": 2.4621049880981447, + "step": 341620 + }, + { + "epoch": 0.9300122498979175, + "grad_norm": 2.1789186000823975, + "learning_rate": 1.4017173231566036e-06, + "loss": 2.462358856201172, + "step": 341640 + }, + { + "epoch": 0.930066693888662, + "grad_norm": 2.3215651512145996, + "learning_rate": 1.400626959247649e-06, + "loss": 2.4696866989135744, + "step": 341660 + }, + { + "epoch": 0.9301211378794065, + "grad_norm": 2.088531970977783, + "learning_rate": 1.3995365953386944e-06, + "loss": 2.4657135009765625, + "step": 341680 + }, + { + "epoch": 0.930175581870151, + "grad_norm": 2.197232484817505, + "learning_rate": 1.3984462314297397e-06, + "loss": 2.4641319274902345, + "step": 341700 + }, + { + "epoch": 0.9302300258608956, + "grad_norm": 2.1443381309509277, + "learning_rate": 1.3973558675207851e-06, + "loss": 2.475508117675781, + "step": 341720 + }, + { + "epoch": 0.9302844698516402, + "grad_norm": 2.1426751613616943, + "learning_rate": 1.3962655036118305e-06, + "loss": 2.504039192199707, + "step": 341740 + }, + { + "epoch": 0.9303389138423847, + "grad_norm": 2.2633216381073, + "learning_rate": 1.395175139702876e-06, + "loss": 2.4705268859863283, + "step": 341760 + }, + { + "epoch": 0.9303933578331292, + "grad_norm": 2.336160898208618, + "learning_rate": 1.3940847757939213e-06, + "loss": 2.4427162170410157, + "step": 341780 + }, + { + "epoch": 0.9304478018238737, + "grad_norm": 2.2199273109436035, + "learning_rate": 1.3929944118849667e-06, + "loss": 2.4377079010009766, + "step": 341800 + }, + { + "epoch": 0.9305022458146182, + "grad_norm": 2.1122055053710938, + "learning_rate": 1.391904047976012e-06, + "loss": 2.4515121459960936, + "step": 341820 + }, + { + "epoch": 0.9305566898053628, + "grad_norm": 2.145378351211548, + "learning_rate": 1.3908136840670574e-06, + "loss": 2.4779863357543945, + "step": 341840 + }, + { + "epoch": 0.9306111337961073, + "grad_norm": 2.2197844982147217, + "learning_rate": 1.3897233201581028e-06, + "loss": 2.4432987213134765, + "step": 341860 + }, + { + "epoch": 0.9306655777868518, + "grad_norm": 2.298818349838257, + "learning_rate": 1.3886329562491482e-06, + "loss": 2.456409454345703, + "step": 341880 + }, + { + "epoch": 0.9307200217775963, + "grad_norm": 2.2002949714660645, + "learning_rate": 1.3875425923401936e-06, + "loss": 2.455148696899414, + "step": 341900 + }, + { + "epoch": 0.9307744657683408, + "grad_norm": 2.081707715988159, + "learning_rate": 1.386452228431239e-06, + "loss": 2.437255859375, + "step": 341920 + }, + { + "epoch": 0.9308289097590854, + "grad_norm": 2.09818434715271, + "learning_rate": 1.3853618645222844e-06, + "loss": 2.457993507385254, + "step": 341940 + }, + { + "epoch": 0.9308833537498299, + "grad_norm": 2.168928861618042, + "learning_rate": 1.3842715006133298e-06, + "loss": 2.4704185485839845, + "step": 341960 + }, + { + "epoch": 0.9309377977405744, + "grad_norm": 2.0536599159240723, + "learning_rate": 1.3831811367043751e-06, + "loss": 2.4523441314697267, + "step": 341980 + }, + { + "epoch": 0.9309922417313189, + "grad_norm": 2.6187996864318848, + "learning_rate": 1.3820907727954205e-06, + "loss": 2.4575239181518556, + "step": 342000 + }, + { + "epoch": 0.9310466857220634, + "grad_norm": 2.0899922847747803, + "learning_rate": 1.381000408886466e-06, + "loss": 2.483549690246582, + "step": 342020 + }, + { + "epoch": 0.9311011297128079, + "grad_norm": 2.09486985206604, + "learning_rate": 1.3799100449775113e-06, + "loss": 2.4631284713745116, + "step": 342040 + }, + { + "epoch": 0.9311555737035525, + "grad_norm": 2.2238335609436035, + "learning_rate": 1.3788196810685567e-06, + "loss": 2.4808059692382813, + "step": 342060 + }, + { + "epoch": 0.931210017694297, + "grad_norm": 2.1026010513305664, + "learning_rate": 1.377729317159602e-06, + "loss": 2.453264617919922, + "step": 342080 + }, + { + "epoch": 0.9312644616850415, + "grad_norm": 2.1753125190734863, + "learning_rate": 1.3766389532506475e-06, + "loss": 2.4459115982055666, + "step": 342100 + }, + { + "epoch": 0.931318905675786, + "grad_norm": 2.208554744720459, + "learning_rate": 1.3755485893416928e-06, + "loss": 2.453504180908203, + "step": 342120 + }, + { + "epoch": 0.9313733496665305, + "grad_norm": 2.2375950813293457, + "learning_rate": 1.3744582254327382e-06, + "loss": 2.4396995544433593, + "step": 342140 + }, + { + "epoch": 0.9314277936572751, + "grad_norm": 2.198831796646118, + "learning_rate": 1.3733678615237836e-06, + "loss": 2.4582889556884764, + "step": 342160 + }, + { + "epoch": 0.9314822376480196, + "grad_norm": 2.113739013671875, + "learning_rate": 1.372277497614829e-06, + "loss": 2.4263818740844725, + "step": 342180 + }, + { + "epoch": 0.9315366816387641, + "grad_norm": 2.1690053939819336, + "learning_rate": 1.3711871337058744e-06, + "loss": 2.447779083251953, + "step": 342200 + }, + { + "epoch": 0.9315911256295086, + "grad_norm": 2.172701358795166, + "learning_rate": 1.3700967697969198e-06, + "loss": 2.4355140686035157, + "step": 342220 + }, + { + "epoch": 0.9316455696202531, + "grad_norm": 2.152858257293701, + "learning_rate": 1.3690064058879652e-06, + "loss": 2.442831039428711, + "step": 342240 + }, + { + "epoch": 0.9317000136109976, + "grad_norm": 2.5476717948913574, + "learning_rate": 1.3679160419790105e-06, + "loss": 2.4518495559692384, + "step": 342260 + }, + { + "epoch": 0.9317544576017422, + "grad_norm": 2.153571367263794, + "learning_rate": 1.366825678070056e-06, + "loss": 2.480322265625, + "step": 342280 + }, + { + "epoch": 0.9318089015924867, + "grad_norm": 2.224252462387085, + "learning_rate": 1.3657353141611013e-06, + "loss": 2.46029052734375, + "step": 342300 + }, + { + "epoch": 0.9318633455832313, + "grad_norm": 2.217400550842285, + "learning_rate": 1.3646449502521467e-06, + "loss": 2.4922296524047853, + "step": 342320 + }, + { + "epoch": 0.9319177895739758, + "grad_norm": 2.8142333030700684, + "learning_rate": 1.363554586343192e-06, + "loss": 2.439060592651367, + "step": 342340 + }, + { + "epoch": 0.9319722335647203, + "grad_norm": 2.708239793777466, + "learning_rate": 1.3624642224342375e-06, + "loss": 2.430702972412109, + "step": 342360 + }, + { + "epoch": 0.9320266775554649, + "grad_norm": 2.182708740234375, + "learning_rate": 1.3613738585252829e-06, + "loss": 2.4552080154418947, + "step": 342380 + }, + { + "epoch": 0.9320811215462094, + "grad_norm": 2.276909828186035, + "learning_rate": 1.3602834946163282e-06, + "loss": 2.446647071838379, + "step": 342400 + }, + { + "epoch": 0.9321355655369539, + "grad_norm": 2.178365468978882, + "learning_rate": 1.3591931307073736e-06, + "loss": 2.4784912109375, + "step": 342420 + }, + { + "epoch": 0.9321900095276984, + "grad_norm": 2.217862367630005, + "learning_rate": 1.358102766798419e-06, + "loss": 2.434463691711426, + "step": 342440 + }, + { + "epoch": 0.9322444535184429, + "grad_norm": 2.2956490516662598, + "learning_rate": 1.3570124028894644e-06, + "loss": 2.448841094970703, + "step": 342460 + }, + { + "epoch": 0.9322988975091874, + "grad_norm": 2.073270320892334, + "learning_rate": 1.3559220389805098e-06, + "loss": 2.4754878997802736, + "step": 342480 + }, + { + "epoch": 0.932353341499932, + "grad_norm": 2.1287484169006348, + "learning_rate": 1.3548316750715552e-06, + "loss": 2.4664255142211915, + "step": 342500 + }, + { + "epoch": 0.9324077854906765, + "grad_norm": 2.3127801418304443, + "learning_rate": 1.3537413111626006e-06, + "loss": 2.450522804260254, + "step": 342520 + }, + { + "epoch": 0.932462229481421, + "grad_norm": 2.108224391937256, + "learning_rate": 1.3526509472536462e-06, + "loss": 2.4534112930297853, + "step": 342540 + }, + { + "epoch": 0.9325166734721655, + "grad_norm": 2.2208034992218018, + "learning_rate": 1.3515605833446915e-06, + "loss": 2.469833564758301, + "step": 342560 + }, + { + "epoch": 0.93257111746291, + "grad_norm": 2.284727096557617, + "learning_rate": 1.350470219435737e-06, + "loss": 2.459909439086914, + "step": 342580 + }, + { + "epoch": 0.9326255614536546, + "grad_norm": 2.1443850994110107, + "learning_rate": 1.3493798555267823e-06, + "loss": 2.4441709518432617, + "step": 342600 + }, + { + "epoch": 0.9326800054443991, + "grad_norm": 2.0933496952056885, + "learning_rate": 1.3482894916178277e-06, + "loss": 2.444174385070801, + "step": 342620 + }, + { + "epoch": 0.9327344494351436, + "grad_norm": 2.278067111968994, + "learning_rate": 1.347199127708873e-06, + "loss": 2.463516426086426, + "step": 342640 + }, + { + "epoch": 0.9327888934258881, + "grad_norm": 2.20754075050354, + "learning_rate": 1.3461087637999185e-06, + "loss": 2.486113739013672, + "step": 342660 + }, + { + "epoch": 0.9328433374166326, + "grad_norm": 2.117919445037842, + "learning_rate": 1.3450183998909639e-06, + "loss": 2.4752010345458983, + "step": 342680 + }, + { + "epoch": 0.9328977814073771, + "grad_norm": 2.3775010108947754, + "learning_rate": 1.3439280359820092e-06, + "loss": 2.4377649307250975, + "step": 342700 + }, + { + "epoch": 0.9329522253981217, + "grad_norm": 2.1739439964294434, + "learning_rate": 1.3428376720730546e-06, + "loss": 2.4587493896484376, + "step": 342720 + }, + { + "epoch": 0.9330066693888662, + "grad_norm": 2.2226226329803467, + "learning_rate": 1.3417473081641e-06, + "loss": 2.4871118545532225, + "step": 342740 + }, + { + "epoch": 0.9330611133796107, + "grad_norm": 5.996043682098389, + "learning_rate": 1.3406569442551454e-06, + "loss": 2.4757850646972654, + "step": 342760 + }, + { + "epoch": 0.9331155573703552, + "grad_norm": 2.251312732696533, + "learning_rate": 1.3395665803461908e-06, + "loss": 2.4837913513183594, + "step": 342780 + }, + { + "epoch": 0.9331700013610997, + "grad_norm": 2.1781914234161377, + "learning_rate": 1.3384762164372362e-06, + "loss": 2.4710935592651366, + "step": 342800 + }, + { + "epoch": 0.9332244453518443, + "grad_norm": 2.227832317352295, + "learning_rate": 1.3373858525282816e-06, + "loss": 2.481161880493164, + "step": 342820 + }, + { + "epoch": 0.9332788893425888, + "grad_norm": 2.2576444149017334, + "learning_rate": 1.336295488619327e-06, + "loss": 2.436482620239258, + "step": 342840 + }, + { + "epoch": 0.9333333333333333, + "grad_norm": 2.529205322265625, + "learning_rate": 1.3352051247103723e-06, + "loss": 2.4735437393188477, + "step": 342860 + }, + { + "epoch": 0.9333877773240778, + "grad_norm": 2.121447801589966, + "learning_rate": 1.3341147608014177e-06, + "loss": 2.4387182235717773, + "step": 342880 + }, + { + "epoch": 0.9334422213148224, + "grad_norm": 2.2326574325561523, + "learning_rate": 1.333024396892463e-06, + "loss": 2.4642547607421874, + "step": 342900 + }, + { + "epoch": 0.9334966653055669, + "grad_norm": 2.2030563354492188, + "learning_rate": 1.3319340329835085e-06, + "loss": 2.4641347885131837, + "step": 342920 + }, + { + "epoch": 0.9335511092963115, + "grad_norm": 2.190335750579834, + "learning_rate": 1.3308436690745539e-06, + "loss": 2.427920341491699, + "step": 342940 + }, + { + "epoch": 0.933605553287056, + "grad_norm": 2.5443215370178223, + "learning_rate": 1.3297533051655993e-06, + "loss": 2.4787954330444335, + "step": 342960 + }, + { + "epoch": 0.9336599972778005, + "grad_norm": 2.1065433025360107, + "learning_rate": 1.3286629412566446e-06, + "loss": 2.4735973358154295, + "step": 342980 + }, + { + "epoch": 0.933714441268545, + "grad_norm": 2.180603265762329, + "learning_rate": 1.32757257734769e-06, + "loss": 2.4619470596313477, + "step": 343000 + }, + { + "epoch": 0.9337688852592895, + "grad_norm": 2.1807124614715576, + "learning_rate": 1.3264822134387354e-06, + "loss": 2.4561819076538085, + "step": 343020 + }, + { + "epoch": 0.933823329250034, + "grad_norm": 2.1088178157806396, + "learning_rate": 1.3253918495297808e-06, + "loss": 2.43560848236084, + "step": 343040 + }, + { + "epoch": 0.9338777732407786, + "grad_norm": 2.1902482509613037, + "learning_rate": 1.3243014856208262e-06, + "loss": 2.449108695983887, + "step": 343060 + }, + { + "epoch": 0.9339322172315231, + "grad_norm": 2.5462357997894287, + "learning_rate": 1.3232111217118716e-06, + "loss": 2.43713264465332, + "step": 343080 + }, + { + "epoch": 0.9339866612222676, + "grad_norm": 2.1566970348358154, + "learning_rate": 1.322120757802917e-06, + "loss": 2.4647029876708983, + "step": 343100 + }, + { + "epoch": 0.9340411052130121, + "grad_norm": 2.257763385772705, + "learning_rate": 1.3210303938939623e-06, + "loss": 2.451059913635254, + "step": 343120 + }, + { + "epoch": 0.9340955492037566, + "grad_norm": 2.2009012699127197, + "learning_rate": 1.3199400299850077e-06, + "loss": 2.3991188049316405, + "step": 343140 + }, + { + "epoch": 0.9341499931945012, + "grad_norm": 2.2078137397766113, + "learning_rate": 1.3188496660760531e-06, + "loss": 2.4456764221191407, + "step": 343160 + }, + { + "epoch": 0.9342044371852457, + "grad_norm": 2.1588664054870605, + "learning_rate": 1.3177593021670985e-06, + "loss": 2.434499168395996, + "step": 343180 + }, + { + "epoch": 0.9342588811759902, + "grad_norm": 2.1553986072540283, + "learning_rate": 1.3166689382581439e-06, + "loss": 2.4574462890625, + "step": 343200 + }, + { + "epoch": 0.9343133251667347, + "grad_norm": 2.2498819828033447, + "learning_rate": 1.3155785743491893e-06, + "loss": 2.4773319244384764, + "step": 343220 + }, + { + "epoch": 0.9343677691574792, + "grad_norm": 2.2151217460632324, + "learning_rate": 1.3144882104402347e-06, + "loss": 2.4481788635253907, + "step": 343240 + }, + { + "epoch": 0.9344222131482237, + "grad_norm": 2.176738977432251, + "learning_rate": 1.31339784653128e-06, + "loss": 2.491876220703125, + "step": 343260 + }, + { + "epoch": 0.9344766571389683, + "grad_norm": 2.1665151119232178, + "learning_rate": 1.3123074826223252e-06, + "loss": 2.452737045288086, + "step": 343280 + }, + { + "epoch": 0.9345311011297128, + "grad_norm": 2.2809641361236572, + "learning_rate": 1.3112171187133706e-06, + "loss": 2.461943817138672, + "step": 343300 + }, + { + "epoch": 0.9345855451204573, + "grad_norm": 2.1285746097564697, + "learning_rate": 1.310126754804416e-06, + "loss": 2.461170959472656, + "step": 343320 + }, + { + "epoch": 0.9346399891112018, + "grad_norm": 2.188777208328247, + "learning_rate": 1.3090363908954614e-06, + "loss": 2.4349996566772463, + "step": 343340 + }, + { + "epoch": 0.9346944331019463, + "grad_norm": 2.1882646083831787, + "learning_rate": 1.3079460269865068e-06, + "loss": 2.427676200866699, + "step": 343360 + }, + { + "epoch": 0.9347488770926909, + "grad_norm": 2.228269577026367, + "learning_rate": 1.3068556630775521e-06, + "loss": 2.4666282653808596, + "step": 343380 + }, + { + "epoch": 0.9348033210834354, + "grad_norm": 2.061398983001709, + "learning_rate": 1.3057652991685975e-06, + "loss": 2.453468132019043, + "step": 343400 + }, + { + "epoch": 0.93485776507418, + "grad_norm": 2.12908935546875, + "learning_rate": 1.304674935259643e-06, + "loss": 2.463906097412109, + "step": 343420 + }, + { + "epoch": 0.9349122090649244, + "grad_norm": 2.141536235809326, + "learning_rate": 1.3035845713506883e-06, + "loss": 2.4679946899414062, + "step": 343440 + }, + { + "epoch": 0.934966653055669, + "grad_norm": 2.207362413406372, + "learning_rate": 1.3024942074417337e-06, + "loss": 2.474183464050293, + "step": 343460 + }, + { + "epoch": 0.9350210970464135, + "grad_norm": 2.190946102142334, + "learning_rate": 1.301403843532779e-06, + "loss": 2.4291542053222654, + "step": 343480 + }, + { + "epoch": 0.9350755410371581, + "grad_norm": 2.1704976558685303, + "learning_rate": 1.3003134796238245e-06, + "loss": 2.4494983673095705, + "step": 343500 + }, + { + "epoch": 0.9351299850279026, + "grad_norm": 2.0921363830566406, + "learning_rate": 1.2992231157148698e-06, + "loss": 2.4180477142333983, + "step": 343520 + }, + { + "epoch": 0.9351844290186471, + "grad_norm": 2.144568681716919, + "learning_rate": 1.2981327518059152e-06, + "loss": 2.446521759033203, + "step": 343540 + }, + { + "epoch": 0.9352388730093916, + "grad_norm": 2.1012067794799805, + "learning_rate": 1.2970423878969606e-06, + "loss": 2.4447982788085936, + "step": 343560 + }, + { + "epoch": 0.9352933170001361, + "grad_norm": 2.168358087539673, + "learning_rate": 1.295952023988006e-06, + "loss": 2.4782783508300783, + "step": 343580 + }, + { + "epoch": 0.9353477609908807, + "grad_norm": 2.3277971744537354, + "learning_rate": 1.2948616600790514e-06, + "loss": 2.4914213180541993, + "step": 343600 + }, + { + "epoch": 0.9354022049816252, + "grad_norm": 2.18902850151062, + "learning_rate": 1.2937712961700968e-06, + "loss": 2.4553810119628907, + "step": 343620 + }, + { + "epoch": 0.9354566489723697, + "grad_norm": 2.152571201324463, + "learning_rate": 1.2926809322611422e-06, + "loss": 2.462171936035156, + "step": 343640 + }, + { + "epoch": 0.9355110929631142, + "grad_norm": 2.2065412998199463, + "learning_rate": 1.2915905683521875e-06, + "loss": 2.467154121398926, + "step": 343660 + }, + { + "epoch": 0.9355655369538587, + "grad_norm": 2.1296133995056152, + "learning_rate": 1.290500204443233e-06, + "loss": 2.434794044494629, + "step": 343680 + }, + { + "epoch": 0.9356199809446032, + "grad_norm": 2.203429937362671, + "learning_rate": 1.2894098405342783e-06, + "loss": 2.4492937088012696, + "step": 343700 + }, + { + "epoch": 0.9356744249353478, + "grad_norm": 2.088610887527466, + "learning_rate": 1.2883194766253237e-06, + "loss": 2.4719058990478517, + "step": 343720 + }, + { + "epoch": 0.9357288689260923, + "grad_norm": 2.1218676567077637, + "learning_rate": 1.287229112716369e-06, + "loss": 2.4672170639038087, + "step": 343740 + }, + { + "epoch": 0.9357833129168368, + "grad_norm": 2.1642096042633057, + "learning_rate": 1.2861387488074145e-06, + "loss": 2.4610233306884766, + "step": 343760 + }, + { + "epoch": 0.9358377569075813, + "grad_norm": 3.6804428100585938, + "learning_rate": 1.2850483848984599e-06, + "loss": 2.4839107513427736, + "step": 343780 + }, + { + "epoch": 0.9358922008983258, + "grad_norm": 2.2167434692382812, + "learning_rate": 1.2839580209895052e-06, + "loss": 2.479197883605957, + "step": 343800 + }, + { + "epoch": 0.9359466448890704, + "grad_norm": 2.1340579986572266, + "learning_rate": 1.2828676570805506e-06, + "loss": 2.434906578063965, + "step": 343820 + }, + { + "epoch": 0.9360010888798149, + "grad_norm": 2.2040634155273438, + "learning_rate": 1.281777293171596e-06, + "loss": 2.4583534240722655, + "step": 343840 + }, + { + "epoch": 0.9360555328705594, + "grad_norm": 2.234349250793457, + "learning_rate": 1.2806869292626414e-06, + "loss": 2.492068290710449, + "step": 343860 + }, + { + "epoch": 0.9361099768613039, + "grad_norm": 2.069369316101074, + "learning_rate": 1.2795965653536868e-06, + "loss": 2.4608890533447267, + "step": 343880 + }, + { + "epoch": 0.9361644208520484, + "grad_norm": 2.18218731880188, + "learning_rate": 1.2785062014447322e-06, + "loss": 2.467756652832031, + "step": 343900 + }, + { + "epoch": 0.9362188648427929, + "grad_norm": 2.1967415809631348, + "learning_rate": 1.2774158375357776e-06, + "loss": 2.4851770401000977, + "step": 343920 + }, + { + "epoch": 0.9362733088335375, + "grad_norm": 2.414201259613037, + "learning_rate": 1.276325473626823e-06, + "loss": 2.476579475402832, + "step": 343940 + }, + { + "epoch": 0.936327752824282, + "grad_norm": 2.1698625087738037, + "learning_rate": 1.2752351097178683e-06, + "loss": 2.445139503479004, + "step": 343960 + }, + { + "epoch": 0.9363821968150265, + "grad_norm": 2.17692494392395, + "learning_rate": 1.2741447458089137e-06, + "loss": 2.455683708190918, + "step": 343980 + }, + { + "epoch": 0.936436640805771, + "grad_norm": 2.1161346435546875, + "learning_rate": 1.273054381899959e-06, + "loss": 2.4720706939697266, + "step": 344000 + }, + { + "epoch": 0.9364910847965155, + "grad_norm": 2.1330740451812744, + "learning_rate": 1.2719640179910045e-06, + "loss": 2.4900434494018553, + "step": 344020 + }, + { + "epoch": 0.93654552878726, + "grad_norm": 2.1474430561065674, + "learning_rate": 1.2708736540820499e-06, + "loss": 2.4527111053466797, + "step": 344040 + }, + { + "epoch": 0.9365999727780047, + "grad_norm": 2.173952102661133, + "learning_rate": 1.2697832901730953e-06, + "loss": 2.4243753433227537, + "step": 344060 + }, + { + "epoch": 0.9366544167687492, + "grad_norm": 2.2686386108398438, + "learning_rate": 1.2686929262641406e-06, + "loss": 2.4339366912841798, + "step": 344080 + }, + { + "epoch": 0.9367088607594937, + "grad_norm": 2.1642374992370605, + "learning_rate": 1.267602562355186e-06, + "loss": 2.4671257019042967, + "step": 344100 + }, + { + "epoch": 0.9367633047502382, + "grad_norm": 2.678501605987549, + "learning_rate": 1.2665121984462314e-06, + "loss": 2.4715431213378904, + "step": 344120 + }, + { + "epoch": 0.9368177487409827, + "grad_norm": 2.194873809814453, + "learning_rate": 1.2654218345372768e-06, + "loss": 2.47326774597168, + "step": 344140 + }, + { + "epoch": 0.9368721927317273, + "grad_norm": 2.1630804538726807, + "learning_rate": 1.2643314706283222e-06, + "loss": 2.4564937591552733, + "step": 344160 + }, + { + "epoch": 0.9369266367224718, + "grad_norm": 2.4993245601654053, + "learning_rate": 1.2632411067193676e-06, + "loss": 2.4522838592529297, + "step": 344180 + }, + { + "epoch": 0.9369810807132163, + "grad_norm": 2.2683157920837402, + "learning_rate": 1.262150742810413e-06, + "loss": 2.4621303558349608, + "step": 344200 + }, + { + "epoch": 0.9370355247039608, + "grad_norm": 2.2215051651000977, + "learning_rate": 1.2610603789014583e-06, + "loss": 2.471462440490723, + "step": 344220 + }, + { + "epoch": 0.9370899686947053, + "grad_norm": 2.1685760021209717, + "learning_rate": 1.259970014992504e-06, + "loss": 2.4661300659179686, + "step": 344240 + }, + { + "epoch": 0.9371444126854498, + "grad_norm": 2.130620241165161, + "learning_rate": 1.2588796510835493e-06, + "loss": 2.4871912002563477, + "step": 344260 + }, + { + "epoch": 0.9371988566761944, + "grad_norm": 2.1927294731140137, + "learning_rate": 1.2577892871745947e-06, + "loss": 2.4533449172973634, + "step": 344280 + }, + { + "epoch": 0.9372533006669389, + "grad_norm": 2.124845504760742, + "learning_rate": 1.25669892326564e-06, + "loss": 2.431418800354004, + "step": 344300 + }, + { + "epoch": 0.9373077446576834, + "grad_norm": 2.132362127304077, + "learning_rate": 1.2556085593566855e-06, + "loss": 2.457034683227539, + "step": 344320 + }, + { + "epoch": 0.9373621886484279, + "grad_norm": 2.1526737213134766, + "learning_rate": 1.2545181954477309e-06, + "loss": 2.483063888549805, + "step": 344340 + }, + { + "epoch": 0.9374166326391724, + "grad_norm": 2.277092218399048, + "learning_rate": 1.2534278315387763e-06, + "loss": 2.4560516357421873, + "step": 344360 + }, + { + "epoch": 0.937471076629917, + "grad_norm": 2.211355447769165, + "learning_rate": 1.2523374676298216e-06, + "loss": 2.4695837020874025, + "step": 344380 + }, + { + "epoch": 0.9375255206206615, + "grad_norm": 2.224087715148926, + "learning_rate": 1.251247103720867e-06, + "loss": 2.4557321548461912, + "step": 344400 + }, + { + "epoch": 0.937579964611406, + "grad_norm": 2.1891841888427734, + "learning_rate": 1.2501567398119124e-06, + "loss": 2.457280158996582, + "step": 344420 + }, + { + "epoch": 0.9376344086021505, + "grad_norm": 2.120859146118164, + "learning_rate": 1.2490663759029576e-06, + "loss": 2.442308044433594, + "step": 344440 + }, + { + "epoch": 0.937688852592895, + "grad_norm": 2.2266252040863037, + "learning_rate": 1.247976011994003e-06, + "loss": 2.423489570617676, + "step": 344460 + }, + { + "epoch": 0.9377432965836395, + "grad_norm": 2.1464478969573975, + "learning_rate": 1.2468856480850484e-06, + "loss": 2.457006645202637, + "step": 344480 + }, + { + "epoch": 0.9377977405743841, + "grad_norm": 2.152918577194214, + "learning_rate": 1.2457952841760937e-06, + "loss": 2.462800407409668, + "step": 344500 + }, + { + "epoch": 0.9378521845651286, + "grad_norm": 2.3266663551330566, + "learning_rate": 1.2447049202671391e-06, + "loss": 2.4735599517822267, + "step": 344520 + }, + { + "epoch": 0.9379066285558731, + "grad_norm": 2.273192882537842, + "learning_rate": 1.2436145563581845e-06, + "loss": 2.438454437255859, + "step": 344540 + }, + { + "epoch": 0.9379610725466176, + "grad_norm": 2.0911941528320312, + "learning_rate": 1.24252419244923e-06, + "loss": 2.4607839584350586, + "step": 344560 + }, + { + "epoch": 0.9380155165373621, + "grad_norm": 2.090615749359131, + "learning_rate": 1.2414338285402753e-06, + "loss": 2.4813514709472657, + "step": 344580 + }, + { + "epoch": 0.9380699605281068, + "grad_norm": 2.2211499214172363, + "learning_rate": 1.2403434646313207e-06, + "loss": 2.448238182067871, + "step": 344600 + }, + { + "epoch": 0.9381244045188513, + "grad_norm": 2.2093217372894287, + "learning_rate": 1.239253100722366e-06, + "loss": 2.4357841491699217, + "step": 344620 + }, + { + "epoch": 0.9381788485095958, + "grad_norm": 2.0926709175109863, + "learning_rate": 1.2381627368134114e-06, + "loss": 2.4502870559692385, + "step": 344640 + }, + { + "epoch": 0.9382332925003403, + "grad_norm": 2.2086288928985596, + "learning_rate": 1.237072372904457e-06, + "loss": 2.4887216567993162, + "step": 344660 + }, + { + "epoch": 0.9382877364910848, + "grad_norm": 2.1046032905578613, + "learning_rate": 1.2359820089955024e-06, + "loss": 2.4607866287231444, + "step": 344680 + }, + { + "epoch": 0.9383421804818293, + "grad_norm": 2.262145757675171, + "learning_rate": 1.2348916450865478e-06, + "loss": 2.4603689193725584, + "step": 344700 + }, + { + "epoch": 0.9383966244725739, + "grad_norm": 2.373107671737671, + "learning_rate": 1.2338012811775932e-06, + "loss": 2.4526138305664062, + "step": 344720 + }, + { + "epoch": 0.9384510684633184, + "grad_norm": 2.1612746715545654, + "learning_rate": 1.2327109172686386e-06, + "loss": 2.476826858520508, + "step": 344740 + }, + { + "epoch": 0.9385055124540629, + "grad_norm": 2.076749324798584, + "learning_rate": 1.231620553359684e-06, + "loss": 2.4627754211425783, + "step": 344760 + }, + { + "epoch": 0.9385599564448074, + "grad_norm": 2.191399335861206, + "learning_rate": 1.2305301894507294e-06, + "loss": 2.488047409057617, + "step": 344780 + }, + { + "epoch": 0.9386144004355519, + "grad_norm": 2.2482168674468994, + "learning_rate": 1.2294398255417747e-06, + "loss": 2.4778057098388673, + "step": 344800 + }, + { + "epoch": 0.9386688444262965, + "grad_norm": 2.2710530757904053, + "learning_rate": 1.2283494616328201e-06, + "loss": 2.4447868347167967, + "step": 344820 + }, + { + "epoch": 0.938723288417041, + "grad_norm": 2.178103446960449, + "learning_rate": 1.2272590977238655e-06, + "loss": 2.4533245086669924, + "step": 344840 + }, + { + "epoch": 0.9387777324077855, + "grad_norm": 2.0966060161590576, + "learning_rate": 1.226168733814911e-06, + "loss": 2.4646764755249024, + "step": 344860 + }, + { + "epoch": 0.93883217639853, + "grad_norm": 2.1212003231048584, + "learning_rate": 1.2250783699059563e-06, + "loss": 2.46298770904541, + "step": 344880 + }, + { + "epoch": 0.9388866203892745, + "grad_norm": 2.1867926120758057, + "learning_rate": 1.2239880059970017e-06, + "loss": 2.46209774017334, + "step": 344900 + }, + { + "epoch": 0.938941064380019, + "grad_norm": 2.248117208480835, + "learning_rate": 1.222897642088047e-06, + "loss": 2.4626441955566407, + "step": 344920 + }, + { + "epoch": 0.9389955083707636, + "grad_norm": 2.184436559677124, + "learning_rate": 1.2218072781790924e-06, + "loss": 2.4778186798095705, + "step": 344940 + }, + { + "epoch": 0.9390499523615081, + "grad_norm": 2.2736752033233643, + "learning_rate": 1.2207169142701378e-06, + "loss": 2.4485685348510744, + "step": 344960 + }, + { + "epoch": 0.9391043963522526, + "grad_norm": 2.1083080768585205, + "learning_rate": 1.2196265503611832e-06, + "loss": 2.4394790649414064, + "step": 344980 + }, + { + "epoch": 0.9391588403429971, + "grad_norm": 2.1874475479125977, + "learning_rate": 1.2185361864522286e-06, + "loss": 2.4566659927368164, + "step": 345000 + }, + { + "epoch": 0.9392132843337416, + "grad_norm": 6.710339546203613, + "learning_rate": 1.217445822543274e-06, + "loss": 2.466725540161133, + "step": 345020 + }, + { + "epoch": 0.9392677283244862, + "grad_norm": 3.1446025371551514, + "learning_rate": 1.2163554586343194e-06, + "loss": 2.4566057205200194, + "step": 345040 + }, + { + "epoch": 0.9393221723152307, + "grad_norm": 2.12314510345459, + "learning_rate": 1.2152650947253648e-06, + "loss": 2.4469301223754885, + "step": 345060 + }, + { + "epoch": 0.9393766163059752, + "grad_norm": 2.1680855751037598, + "learning_rate": 1.2141747308164101e-06, + "loss": 2.47009220123291, + "step": 345080 + }, + { + "epoch": 0.9394310602967197, + "grad_norm": 2.1910133361816406, + "learning_rate": 1.2130843669074555e-06, + "loss": 2.4426767349243166, + "step": 345100 + }, + { + "epoch": 0.9394855042874642, + "grad_norm": 2.3509128093719482, + "learning_rate": 1.211994002998501e-06, + "loss": 2.4492609024047853, + "step": 345120 + }, + { + "epoch": 0.9395399482782087, + "grad_norm": 2.1884639263153076, + "learning_rate": 1.2109036390895463e-06, + "loss": 2.443819999694824, + "step": 345140 + }, + { + "epoch": 0.9395943922689534, + "grad_norm": 2.24027681350708, + "learning_rate": 1.2098132751805917e-06, + "loss": 2.4358903884887697, + "step": 345160 + }, + { + "epoch": 0.9396488362596979, + "grad_norm": 2.2217867374420166, + "learning_rate": 1.208722911271637e-06, + "loss": 2.4645605087280273, + "step": 345180 + }, + { + "epoch": 0.9397032802504424, + "grad_norm": 2.665565252304077, + "learning_rate": 1.2076325473626825e-06, + "loss": 2.4809513092041016, + "step": 345200 + }, + { + "epoch": 0.9397577242411869, + "grad_norm": 2.643838405609131, + "learning_rate": 1.2065421834537278e-06, + "loss": 2.4580883026123046, + "step": 345220 + }, + { + "epoch": 0.9398121682319314, + "grad_norm": 2.170154571533203, + "learning_rate": 1.2054518195447732e-06, + "loss": 2.4648662567138673, + "step": 345240 + }, + { + "epoch": 0.9398666122226759, + "grad_norm": 2.1078450679779053, + "learning_rate": 1.2043614556358186e-06, + "loss": 2.466602897644043, + "step": 345260 + }, + { + "epoch": 0.9399210562134205, + "grad_norm": 2.1701254844665527, + "learning_rate": 1.203271091726864e-06, + "loss": 2.458883285522461, + "step": 345280 + }, + { + "epoch": 0.939975500204165, + "grad_norm": 2.2308807373046875, + "learning_rate": 1.2021807278179094e-06, + "loss": 2.452902984619141, + "step": 345300 + }, + { + "epoch": 0.9400299441949095, + "grad_norm": 2.1191961765289307, + "learning_rate": 1.2010903639089548e-06, + "loss": 2.4749107360839844, + "step": 345320 + }, + { + "epoch": 0.940084388185654, + "grad_norm": 2.118650197982788, + "learning_rate": 1.2000000000000002e-06, + "loss": 2.4736547470092773, + "step": 345340 + }, + { + "epoch": 0.9401388321763985, + "grad_norm": 2.0823090076446533, + "learning_rate": 1.1989096360910453e-06, + "loss": 2.475556755065918, + "step": 345360 + }, + { + "epoch": 0.9401932761671431, + "grad_norm": 2.237790107727051, + "learning_rate": 1.1978192721820907e-06, + "loss": 2.457291603088379, + "step": 345380 + }, + { + "epoch": 0.9402477201578876, + "grad_norm": 2.2164762020111084, + "learning_rate": 1.196728908273136e-06, + "loss": 2.464055824279785, + "step": 345400 + }, + { + "epoch": 0.9403021641486321, + "grad_norm": 2.1018855571746826, + "learning_rate": 1.1956385443641815e-06, + "loss": 2.4670026779174803, + "step": 345420 + }, + { + "epoch": 0.9403566081393766, + "grad_norm": 2.780339479446411, + "learning_rate": 1.1945481804552269e-06, + "loss": 2.463861656188965, + "step": 345440 + }, + { + "epoch": 0.9404110521301211, + "grad_norm": 2.2605621814727783, + "learning_rate": 1.1934578165462723e-06, + "loss": 2.4506711959838867, + "step": 345460 + }, + { + "epoch": 0.9404654961208656, + "grad_norm": 2.171783924102783, + "learning_rate": 1.1923674526373176e-06, + "loss": 2.481921577453613, + "step": 345480 + }, + { + "epoch": 0.9405199401116102, + "grad_norm": 2.100375175476074, + "learning_rate": 1.1912770887283632e-06, + "loss": 2.4565765380859377, + "step": 345500 + }, + { + "epoch": 0.9405743841023547, + "grad_norm": 2.263181686401367, + "learning_rate": 1.1901867248194086e-06, + "loss": 2.4952970504760743, + "step": 345520 + }, + { + "epoch": 0.9406288280930992, + "grad_norm": 2.2502732276916504, + "learning_rate": 1.189096360910454e-06, + "loss": 2.4628658294677734, + "step": 345540 + }, + { + "epoch": 0.9406832720838437, + "grad_norm": 2.153986692428589, + "learning_rate": 1.1880059970014994e-06, + "loss": 2.483845329284668, + "step": 345560 + }, + { + "epoch": 0.9407377160745882, + "grad_norm": 2.1607284545898438, + "learning_rate": 1.1869156330925448e-06, + "loss": 2.469967079162598, + "step": 345580 + }, + { + "epoch": 0.9407921600653328, + "grad_norm": 2.2047464847564697, + "learning_rate": 1.1858252691835902e-06, + "loss": 2.4637874603271483, + "step": 345600 + }, + { + "epoch": 0.9408466040560773, + "grad_norm": 2.185331106185913, + "learning_rate": 1.1847349052746356e-06, + "loss": 2.4754199981689453, + "step": 345620 + }, + { + "epoch": 0.9409010480468218, + "grad_norm": 2.248919725418091, + "learning_rate": 1.183644541365681e-06, + "loss": 2.4598527908325196, + "step": 345640 + }, + { + "epoch": 0.9409554920375663, + "grad_norm": 2.2887990474700928, + "learning_rate": 1.1825541774567263e-06, + "loss": 2.473756790161133, + "step": 345660 + }, + { + "epoch": 0.9410099360283108, + "grad_norm": 2.2508010864257812, + "learning_rate": 1.1814638135477717e-06, + "loss": 2.4472694396972656, + "step": 345680 + }, + { + "epoch": 0.9410643800190553, + "grad_norm": 2.2437150478363037, + "learning_rate": 1.180373449638817e-06, + "loss": 2.485594940185547, + "step": 345700 + }, + { + "epoch": 0.9411188240098, + "grad_norm": 2.1702771186828613, + "learning_rate": 1.1792830857298625e-06, + "loss": 2.4759859085083007, + "step": 345720 + }, + { + "epoch": 0.9411732680005445, + "grad_norm": 2.2668957710266113, + "learning_rate": 1.1781927218209079e-06, + "loss": 2.4420330047607424, + "step": 345740 + }, + { + "epoch": 0.941227711991289, + "grad_norm": 2.9041240215301514, + "learning_rate": 1.1771023579119533e-06, + "loss": 2.4147504806518554, + "step": 345760 + }, + { + "epoch": 0.9412821559820335, + "grad_norm": 2.293510913848877, + "learning_rate": 1.1760119940029986e-06, + "loss": 2.4802669525146483, + "step": 345780 + }, + { + "epoch": 0.941336599972778, + "grad_norm": 2.1670844554901123, + "learning_rate": 1.174921630094044e-06, + "loss": 2.458706855773926, + "step": 345800 + }, + { + "epoch": 0.9413910439635226, + "grad_norm": 2.3027310371398926, + "learning_rate": 1.1738312661850894e-06, + "loss": 2.43499641418457, + "step": 345820 + }, + { + "epoch": 0.9414454879542671, + "grad_norm": 2.231506109237671, + "learning_rate": 1.1727409022761348e-06, + "loss": 2.48581485748291, + "step": 345840 + }, + { + "epoch": 0.9414999319450116, + "grad_norm": 2.27426815032959, + "learning_rate": 1.1716505383671802e-06, + "loss": 2.4422607421875, + "step": 345860 + }, + { + "epoch": 0.9415543759357561, + "grad_norm": 2.2245981693267822, + "learning_rate": 1.1705601744582256e-06, + "loss": 2.480513000488281, + "step": 345880 + }, + { + "epoch": 0.9416088199265006, + "grad_norm": 2.132448196411133, + "learning_rate": 1.169469810549271e-06, + "loss": 2.4543455123901365, + "step": 345900 + }, + { + "epoch": 0.9416632639172451, + "grad_norm": 2.1815907955169678, + "learning_rate": 1.1683794466403163e-06, + "loss": 2.4622032165527346, + "step": 345920 + }, + { + "epoch": 0.9417177079079897, + "grad_norm": 2.1375091075897217, + "learning_rate": 1.1672890827313617e-06, + "loss": 2.44509334564209, + "step": 345940 + }, + { + "epoch": 0.9417721518987342, + "grad_norm": 2.1726322174072266, + "learning_rate": 1.1661987188224071e-06, + "loss": 2.433189582824707, + "step": 345960 + }, + { + "epoch": 0.9418265958894787, + "grad_norm": 2.387911558151245, + "learning_rate": 1.1651083549134525e-06, + "loss": 2.435739517211914, + "step": 345980 + }, + { + "epoch": 0.9418810398802232, + "grad_norm": 2.1682467460632324, + "learning_rate": 1.1640179910044979e-06, + "loss": 2.4525680541992188, + "step": 346000 + }, + { + "epoch": 0.9419354838709677, + "grad_norm": 2.161318302154541, + "learning_rate": 1.1629276270955433e-06, + "loss": 2.4614776611328124, + "step": 346020 + }, + { + "epoch": 0.9419899278617123, + "grad_norm": 2.1962220668792725, + "learning_rate": 1.1618372631865887e-06, + "loss": 2.4572200775146484, + "step": 346040 + }, + { + "epoch": 0.9420443718524568, + "grad_norm": 2.185791015625, + "learning_rate": 1.160746899277634e-06, + "loss": 2.4474843978881835, + "step": 346060 + }, + { + "epoch": 0.9420988158432013, + "grad_norm": 2.1916229724884033, + "learning_rate": 1.1596565353686794e-06, + "loss": 2.4838815689086915, + "step": 346080 + }, + { + "epoch": 0.9421532598339458, + "grad_norm": 2.1470093727111816, + "learning_rate": 1.1585661714597248e-06, + "loss": 2.4617992401123048, + "step": 346100 + }, + { + "epoch": 0.9422077038246903, + "grad_norm": 2.332794666290283, + "learning_rate": 1.1574758075507702e-06, + "loss": 2.463155746459961, + "step": 346120 + }, + { + "epoch": 0.9422621478154348, + "grad_norm": 2.075380325317383, + "learning_rate": 1.1563854436418156e-06, + "loss": 2.4615947723388674, + "step": 346140 + }, + { + "epoch": 0.9423165918061794, + "grad_norm": 2.1197383403778076, + "learning_rate": 1.155295079732861e-06, + "loss": 2.4857294082641603, + "step": 346160 + }, + { + "epoch": 0.9423710357969239, + "grad_norm": 2.177988052368164, + "learning_rate": 1.1542047158239064e-06, + "loss": 2.473561668395996, + "step": 346180 + }, + { + "epoch": 0.9424254797876684, + "grad_norm": 2.218895196914673, + "learning_rate": 1.1531143519149517e-06, + "loss": 2.4777368545532226, + "step": 346200 + }, + { + "epoch": 0.9424799237784129, + "grad_norm": 2.2554004192352295, + "learning_rate": 1.1520239880059971e-06, + "loss": 2.4789169311523436, + "step": 346220 + }, + { + "epoch": 0.9425343677691574, + "grad_norm": 2.1827542781829834, + "learning_rate": 1.1509336240970425e-06, + "loss": 2.4520849227905273, + "step": 346240 + }, + { + "epoch": 0.942588811759902, + "grad_norm": 2.172567367553711, + "learning_rate": 1.149843260188088e-06, + "loss": 2.4619651794433595, + "step": 346260 + }, + { + "epoch": 0.9426432557506466, + "grad_norm": 2.2869832515716553, + "learning_rate": 1.1487528962791333e-06, + "loss": 2.470001220703125, + "step": 346280 + }, + { + "epoch": 0.9426976997413911, + "grad_norm": 2.2100164890289307, + "learning_rate": 1.1476625323701787e-06, + "loss": 2.4788434982299803, + "step": 346300 + }, + { + "epoch": 0.9427521437321356, + "grad_norm": 2.2419087886810303, + "learning_rate": 1.146572168461224e-06, + "loss": 2.4590673446655273, + "step": 346320 + }, + { + "epoch": 0.9428065877228801, + "grad_norm": 2.165937900543213, + "learning_rate": 1.1454818045522694e-06, + "loss": 2.446994590759277, + "step": 346340 + }, + { + "epoch": 0.9428610317136246, + "grad_norm": 2.1816928386688232, + "learning_rate": 1.1443914406433148e-06, + "loss": 2.4818937301635744, + "step": 346360 + }, + { + "epoch": 0.9429154757043692, + "grad_norm": 2.2108042240142822, + "learning_rate": 1.1433010767343602e-06, + "loss": 2.442349433898926, + "step": 346380 + }, + { + "epoch": 0.9429699196951137, + "grad_norm": 2.122537851333618, + "learning_rate": 1.1422107128254056e-06, + "loss": 2.470324897766113, + "step": 346400 + }, + { + "epoch": 0.9430243636858582, + "grad_norm": 2.167867660522461, + "learning_rate": 1.141120348916451e-06, + "loss": 2.4521839141845705, + "step": 346420 + }, + { + "epoch": 0.9430788076766027, + "grad_norm": 2.1194252967834473, + "learning_rate": 1.1400299850074964e-06, + "loss": 2.4671567916870116, + "step": 346440 + }, + { + "epoch": 0.9431332516673472, + "grad_norm": 2.130103588104248, + "learning_rate": 1.1389396210985418e-06, + "loss": 2.459711456298828, + "step": 346460 + }, + { + "epoch": 0.9431876956580917, + "grad_norm": 2.3557987213134766, + "learning_rate": 1.1378492571895871e-06, + "loss": 2.4549707412719726, + "step": 346480 + }, + { + "epoch": 0.9432421396488363, + "grad_norm": 2.157651901245117, + "learning_rate": 1.1367588932806325e-06, + "loss": 2.4283193588256835, + "step": 346500 + }, + { + "epoch": 0.9432965836395808, + "grad_norm": 2.181702136993408, + "learning_rate": 1.135668529371678e-06, + "loss": 2.4598434448242186, + "step": 346520 + }, + { + "epoch": 0.9433510276303253, + "grad_norm": 2.194549322128296, + "learning_rate": 1.1345781654627233e-06, + "loss": 2.470107078552246, + "step": 346540 + }, + { + "epoch": 0.9434054716210698, + "grad_norm": 2.1019251346588135, + "learning_rate": 1.1334878015537687e-06, + "loss": 2.4571044921875, + "step": 346560 + }, + { + "epoch": 0.9434599156118143, + "grad_norm": 2.1805858612060547, + "learning_rate": 1.132397437644814e-06, + "loss": 2.4628885269165037, + "step": 346580 + }, + { + "epoch": 0.9435143596025589, + "grad_norm": 2.2278995513916016, + "learning_rate": 1.1313070737358595e-06, + "loss": 2.443345832824707, + "step": 346600 + }, + { + "epoch": 0.9435688035933034, + "grad_norm": 2.156080484390259, + "learning_rate": 1.1302167098269048e-06, + "loss": 2.4612400054931642, + "step": 346620 + }, + { + "epoch": 0.9436232475840479, + "grad_norm": 2.1043272018432617, + "learning_rate": 1.1291263459179502e-06, + "loss": 2.4432750701904298, + "step": 346640 + }, + { + "epoch": 0.9436776915747924, + "grad_norm": 2.1828219890594482, + "learning_rate": 1.1280359820089956e-06, + "loss": 2.4587080001831056, + "step": 346660 + }, + { + "epoch": 0.9437321355655369, + "grad_norm": 2.0990633964538574, + "learning_rate": 1.126945618100041e-06, + "loss": 2.4514747619628907, + "step": 346680 + }, + { + "epoch": 0.9437865795562814, + "grad_norm": 2.301168441772461, + "learning_rate": 1.1258552541910864e-06, + "loss": 2.4843143463134765, + "step": 346700 + }, + { + "epoch": 0.943841023547026, + "grad_norm": 2.1995644569396973, + "learning_rate": 1.1247648902821318e-06, + "loss": 2.451220703125, + "step": 346720 + }, + { + "epoch": 0.9438954675377705, + "grad_norm": 2.3319852352142334, + "learning_rate": 1.1236745263731772e-06, + "loss": 2.4365917205810548, + "step": 346740 + }, + { + "epoch": 0.943949911528515, + "grad_norm": 2.48228120803833, + "learning_rate": 1.1225841624642225e-06, + "loss": 2.4868820190429686, + "step": 346760 + }, + { + "epoch": 0.9440043555192595, + "grad_norm": 2.19393253326416, + "learning_rate": 1.121493798555268e-06, + "loss": 2.454001235961914, + "step": 346780 + }, + { + "epoch": 0.944058799510004, + "grad_norm": 2.0969202518463135, + "learning_rate": 1.1204034346463133e-06, + "loss": 2.45998592376709, + "step": 346800 + }, + { + "epoch": 0.9441132435007487, + "grad_norm": 2.27797794342041, + "learning_rate": 1.1193130707373587e-06, + "loss": 2.467386245727539, + "step": 346820 + }, + { + "epoch": 0.9441676874914932, + "grad_norm": 2.094670534133911, + "learning_rate": 1.118222706828404e-06, + "loss": 2.459659767150879, + "step": 346840 + }, + { + "epoch": 0.9442221314822377, + "grad_norm": 2.406909704208374, + "learning_rate": 1.1171323429194495e-06, + "loss": 2.444355583190918, + "step": 346860 + }, + { + "epoch": 0.9442765754729822, + "grad_norm": 2.174123525619507, + "learning_rate": 1.1160419790104949e-06, + "loss": 2.451638412475586, + "step": 346880 + }, + { + "epoch": 0.9443310194637267, + "grad_norm": 2.2312002182006836, + "learning_rate": 1.1149516151015402e-06, + "loss": 2.4423540115356444, + "step": 346900 + }, + { + "epoch": 0.9443854634544712, + "grad_norm": 2.093635082244873, + "learning_rate": 1.1138612511925856e-06, + "loss": 2.4666162490844727, + "step": 346920 + }, + { + "epoch": 0.9444399074452158, + "grad_norm": 2.286031723022461, + "learning_rate": 1.112770887283631e-06, + "loss": 2.4564123153686523, + "step": 346940 + }, + { + "epoch": 0.9444943514359603, + "grad_norm": 2.2294270992279053, + "learning_rate": 1.1116805233746764e-06, + "loss": 2.4569370269775392, + "step": 346960 + }, + { + "epoch": 0.9445487954267048, + "grad_norm": 2.1733169555664062, + "learning_rate": 1.1105901594657218e-06, + "loss": 2.458172416687012, + "step": 346980 + }, + { + "epoch": 0.9446032394174493, + "grad_norm": 2.1557884216308594, + "learning_rate": 1.1094997955567672e-06, + "loss": 2.485264015197754, + "step": 347000 + }, + { + "epoch": 0.9446576834081938, + "grad_norm": 2.1603899002075195, + "learning_rate": 1.1084094316478126e-06, + "loss": 2.467695617675781, + "step": 347020 + }, + { + "epoch": 0.9447121273989384, + "grad_norm": 2.1326539516448975, + "learning_rate": 1.107319067738858e-06, + "loss": 2.4834011077880858, + "step": 347040 + }, + { + "epoch": 0.9447665713896829, + "grad_norm": 2.182849645614624, + "learning_rate": 1.1062287038299033e-06, + "loss": 2.4588409423828126, + "step": 347060 + }, + { + "epoch": 0.9448210153804274, + "grad_norm": 2.151972770690918, + "learning_rate": 1.1051383399209487e-06, + "loss": 2.4966623306274416, + "step": 347080 + }, + { + "epoch": 0.9448754593711719, + "grad_norm": 2.1138532161712646, + "learning_rate": 1.104047976011994e-06, + "loss": 2.454730224609375, + "step": 347100 + }, + { + "epoch": 0.9449299033619164, + "grad_norm": 2.1142492294311523, + "learning_rate": 1.1029576121030395e-06, + "loss": 2.442974090576172, + "step": 347120 + }, + { + "epoch": 0.9449843473526609, + "grad_norm": 2.310373544692993, + "learning_rate": 1.1018672481940849e-06, + "loss": 2.4322015762329103, + "step": 347140 + }, + { + "epoch": 0.9450387913434055, + "grad_norm": 2.1184589862823486, + "learning_rate": 1.1007768842851303e-06, + "loss": 2.4965993881225588, + "step": 347160 + }, + { + "epoch": 0.94509323533415, + "grad_norm": 2.112152338027954, + "learning_rate": 1.0996865203761756e-06, + "loss": 2.434868812561035, + "step": 347180 + }, + { + "epoch": 0.9451476793248945, + "grad_norm": 2.138551950454712, + "learning_rate": 1.098596156467221e-06, + "loss": 2.4623691558837892, + "step": 347200 + }, + { + "epoch": 0.945202123315639, + "grad_norm": 2.1513001918792725, + "learning_rate": 1.0975057925582664e-06, + "loss": 2.4723968505859375, + "step": 347220 + }, + { + "epoch": 0.9452565673063835, + "grad_norm": 2.186530351638794, + "learning_rate": 1.0964154286493118e-06, + "loss": 2.4392715454101563, + "step": 347240 + }, + { + "epoch": 0.9453110112971281, + "grad_norm": 2.2269389629364014, + "learning_rate": 1.0953250647403572e-06, + "loss": 2.4503814697265627, + "step": 347260 + }, + { + "epoch": 0.9453654552878726, + "grad_norm": 2.2512199878692627, + "learning_rate": 1.0942347008314026e-06, + "loss": 2.470828628540039, + "step": 347280 + }, + { + "epoch": 0.9454198992786171, + "grad_norm": 2.099820137023926, + "learning_rate": 1.093144336922448e-06, + "loss": 2.43841438293457, + "step": 347300 + }, + { + "epoch": 0.9454743432693616, + "grad_norm": 2.157053232192993, + "learning_rate": 1.0920539730134933e-06, + "loss": 2.4463754653930665, + "step": 347320 + }, + { + "epoch": 0.9455287872601061, + "grad_norm": 2.1518771648406982, + "learning_rate": 1.0909636091045387e-06, + "loss": 2.4637060165405273, + "step": 347340 + }, + { + "epoch": 0.9455832312508506, + "grad_norm": 2.1437125205993652, + "learning_rate": 1.0898732451955841e-06, + "loss": 2.4759469985961915, + "step": 347360 + }, + { + "epoch": 0.9456376752415953, + "grad_norm": 2.158287286758423, + "learning_rate": 1.0887828812866295e-06, + "loss": 2.47210750579834, + "step": 347380 + }, + { + "epoch": 0.9456921192323398, + "grad_norm": 2.1536219120025635, + "learning_rate": 1.0876925173776749e-06, + "loss": 2.4587934494018553, + "step": 347400 + }, + { + "epoch": 0.9457465632230843, + "grad_norm": 2.1611475944519043, + "learning_rate": 1.0866021534687203e-06, + "loss": 2.476515769958496, + "step": 347420 + }, + { + "epoch": 0.9458010072138288, + "grad_norm": 2.3170738220214844, + "learning_rate": 1.0855117895597657e-06, + "loss": 2.4453786849975585, + "step": 347440 + }, + { + "epoch": 0.9458554512045733, + "grad_norm": 2.1883950233459473, + "learning_rate": 1.084421425650811e-06, + "loss": 2.4346569061279295, + "step": 347460 + }, + { + "epoch": 0.9459098951953178, + "grad_norm": 2.184891700744629, + "learning_rate": 1.0833310617418564e-06, + "loss": 2.474034309387207, + "step": 347480 + }, + { + "epoch": 0.9459643391860624, + "grad_norm": 2.201502799987793, + "learning_rate": 1.0822406978329018e-06, + "loss": 2.449844741821289, + "step": 347500 + }, + { + "epoch": 0.9460187831768069, + "grad_norm": 2.235699415206909, + "learning_rate": 1.0811503339239472e-06, + "loss": 2.472549629211426, + "step": 347520 + }, + { + "epoch": 0.9460732271675514, + "grad_norm": 2.28009295463562, + "learning_rate": 1.0800599700149926e-06, + "loss": 2.4479074478149414, + "step": 347540 + }, + { + "epoch": 0.9461276711582959, + "grad_norm": 2.206451416015625, + "learning_rate": 1.078969606106038e-06, + "loss": 2.474107360839844, + "step": 347560 + }, + { + "epoch": 0.9461821151490404, + "grad_norm": 2.1893486976623535, + "learning_rate": 1.0778792421970834e-06, + "loss": 2.469002532958984, + "step": 347580 + }, + { + "epoch": 0.946236559139785, + "grad_norm": 2.226454019546509, + "learning_rate": 1.0767888782881287e-06, + "loss": 2.479373550415039, + "step": 347600 + }, + { + "epoch": 0.9462910031305295, + "grad_norm": 2.3094496726989746, + "learning_rate": 1.0756985143791741e-06, + "loss": 2.479780578613281, + "step": 347620 + }, + { + "epoch": 0.946345447121274, + "grad_norm": 2.14473295211792, + "learning_rate": 1.0746081504702195e-06, + "loss": 2.4458490371704102, + "step": 347640 + }, + { + "epoch": 0.9463998911120185, + "grad_norm": 2.1382315158843994, + "learning_rate": 1.073517786561265e-06, + "loss": 2.4259443283081055, + "step": 347660 + }, + { + "epoch": 0.946454335102763, + "grad_norm": 2.072338104248047, + "learning_rate": 1.0724274226523103e-06, + "loss": 2.4421113967895507, + "step": 347680 + }, + { + "epoch": 0.9465087790935075, + "grad_norm": 2.2163214683532715, + "learning_rate": 1.0713370587433557e-06, + "loss": 2.432292175292969, + "step": 347700 + }, + { + "epoch": 0.9465632230842521, + "grad_norm": 2.2184159755706787, + "learning_rate": 1.070246694834401e-06, + "loss": 2.462876892089844, + "step": 347720 + }, + { + "epoch": 0.9466176670749966, + "grad_norm": 2.197134017944336, + "learning_rate": 1.0691563309254464e-06, + "loss": 2.441412353515625, + "step": 347740 + }, + { + "epoch": 0.9466721110657411, + "grad_norm": 2.2022979259490967, + "learning_rate": 1.0680659670164918e-06, + "loss": 2.426915740966797, + "step": 347760 + }, + { + "epoch": 0.9467265550564856, + "grad_norm": 2.1955678462982178, + "learning_rate": 1.0669756031075372e-06, + "loss": 2.4463674545288088, + "step": 347780 + }, + { + "epoch": 0.9467809990472301, + "grad_norm": 2.3761043548583984, + "learning_rate": 1.0658852391985826e-06, + "loss": 2.4702253341674805, + "step": 347800 + }, + { + "epoch": 0.9468354430379747, + "grad_norm": 3.042038679122925, + "learning_rate": 1.064794875289628e-06, + "loss": 2.4605010986328124, + "step": 347820 + }, + { + "epoch": 0.9468898870287192, + "grad_norm": 2.29543137550354, + "learning_rate": 1.0637045113806734e-06, + "loss": 2.4634374618530273, + "step": 347840 + }, + { + "epoch": 0.9469443310194637, + "grad_norm": 2.141103506088257, + "learning_rate": 1.0626141474717188e-06, + "loss": 2.4494848251342773, + "step": 347860 + }, + { + "epoch": 0.9469987750102082, + "grad_norm": 2.163409471511841, + "learning_rate": 1.0615237835627641e-06, + "loss": 2.452511024475098, + "step": 347880 + }, + { + "epoch": 0.9470532190009527, + "grad_norm": 2.1277740001678467, + "learning_rate": 1.0604334196538095e-06, + "loss": 2.459602928161621, + "step": 347900 + }, + { + "epoch": 0.9471076629916972, + "grad_norm": 2.120027780532837, + "learning_rate": 1.059343055744855e-06, + "loss": 2.4281150817871096, + "step": 347920 + }, + { + "epoch": 0.9471621069824419, + "grad_norm": 2.185415029525757, + "learning_rate": 1.0582526918359003e-06, + "loss": 2.441940498352051, + "step": 347940 + }, + { + "epoch": 0.9472165509731864, + "grad_norm": 2.105086088180542, + "learning_rate": 1.0571623279269457e-06, + "loss": 2.463619041442871, + "step": 347960 + }, + { + "epoch": 0.9472709949639309, + "grad_norm": 2.1518239974975586, + "learning_rate": 1.056071964017991e-06, + "loss": 2.4528751373291016, + "step": 347980 + }, + { + "epoch": 0.9473254389546754, + "grad_norm": 2.1778645515441895, + "learning_rate": 1.0549816001090365e-06, + "loss": 2.4876476287841798, + "step": 348000 + }, + { + "epoch": 0.9473798829454199, + "grad_norm": 2.420905590057373, + "learning_rate": 1.0538912362000818e-06, + "loss": 2.4654998779296875, + "step": 348020 + }, + { + "epoch": 0.9474343269361645, + "grad_norm": 2.198805332183838, + "learning_rate": 1.0528008722911272e-06, + "loss": 2.4352540969848633, + "step": 348040 + }, + { + "epoch": 0.947488770926909, + "grad_norm": 2.2065000534057617, + "learning_rate": 1.0517105083821726e-06, + "loss": 2.462965202331543, + "step": 348060 + }, + { + "epoch": 0.9475432149176535, + "grad_norm": 2.3342344760894775, + "learning_rate": 1.050620144473218e-06, + "loss": 2.446828842163086, + "step": 348080 + }, + { + "epoch": 0.947597658908398, + "grad_norm": 2.1624879837036133, + "learning_rate": 1.0495297805642634e-06, + "loss": 2.473956489562988, + "step": 348100 + }, + { + "epoch": 0.9476521028991425, + "grad_norm": 2.3151261806488037, + "learning_rate": 1.0484394166553088e-06, + "loss": 2.4563793182373046, + "step": 348120 + }, + { + "epoch": 0.947706546889887, + "grad_norm": 2.182800054550171, + "learning_rate": 1.0473490527463542e-06, + "loss": 2.4801307678222657, + "step": 348140 + }, + { + "epoch": 0.9477609908806316, + "grad_norm": 5.580105781555176, + "learning_rate": 1.0462586888373995e-06, + "loss": 2.464143180847168, + "step": 348160 + }, + { + "epoch": 0.9478154348713761, + "grad_norm": 2.1767685413360596, + "learning_rate": 1.045168324928445e-06, + "loss": 2.461241340637207, + "step": 348180 + }, + { + "epoch": 0.9478698788621206, + "grad_norm": 2.113574743270874, + "learning_rate": 1.0440779610194903e-06, + "loss": 2.469601058959961, + "step": 348200 + }, + { + "epoch": 0.9479243228528651, + "grad_norm": 2.182797431945801, + "learning_rate": 1.0429875971105357e-06, + "loss": 2.442951202392578, + "step": 348220 + }, + { + "epoch": 0.9479787668436096, + "grad_norm": 2.2206103801727295, + "learning_rate": 1.041897233201581e-06, + "loss": 2.449660873413086, + "step": 348240 + }, + { + "epoch": 0.9480332108343542, + "grad_norm": 2.0698585510253906, + "learning_rate": 1.0408068692926265e-06, + "loss": 2.4455324172973634, + "step": 348260 + }, + { + "epoch": 0.9480876548250987, + "grad_norm": 2.43375301361084, + "learning_rate": 1.0397165053836719e-06, + "loss": 2.4811145782470705, + "step": 348280 + }, + { + "epoch": 0.9481420988158432, + "grad_norm": 2.1625864505767822, + "learning_rate": 1.0386261414747172e-06, + "loss": 2.4611568450927734, + "step": 348300 + }, + { + "epoch": 0.9481965428065877, + "grad_norm": 2.1159162521362305, + "learning_rate": 1.0375357775657626e-06, + "loss": 2.460637092590332, + "step": 348320 + }, + { + "epoch": 0.9482509867973322, + "grad_norm": 2.1364922523498535, + "learning_rate": 1.036445413656808e-06, + "loss": 2.470447540283203, + "step": 348340 + }, + { + "epoch": 0.9483054307880767, + "grad_norm": 2.1520748138427734, + "learning_rate": 1.0353550497478534e-06, + "loss": 2.456960678100586, + "step": 348360 + }, + { + "epoch": 0.9483598747788213, + "grad_norm": 2.1244449615478516, + "learning_rate": 1.0342646858388988e-06, + "loss": 2.448678207397461, + "step": 348380 + }, + { + "epoch": 0.9484143187695658, + "grad_norm": 2.2696993350982666, + "learning_rate": 1.0331743219299442e-06, + "loss": 2.500416564941406, + "step": 348400 + }, + { + "epoch": 0.9484687627603103, + "grad_norm": 2.221996784210205, + "learning_rate": 1.0320839580209896e-06, + "loss": 2.4821727752685545, + "step": 348420 + }, + { + "epoch": 0.9485232067510548, + "grad_norm": 2.2216289043426514, + "learning_rate": 1.030993594112035e-06, + "loss": 2.4621225357055665, + "step": 348440 + }, + { + "epoch": 0.9485776507417993, + "grad_norm": 2.176767587661743, + "learning_rate": 1.0299032302030803e-06, + "loss": 2.4801836013793945, + "step": 348460 + }, + { + "epoch": 0.9486320947325438, + "grad_norm": 2.169757604598999, + "learning_rate": 1.0288128662941257e-06, + "loss": 2.4387107849121095, + "step": 348480 + }, + { + "epoch": 0.9486865387232885, + "grad_norm": 2.2131385803222656, + "learning_rate": 1.027722502385171e-06, + "loss": 2.4658021926879883, + "step": 348500 + }, + { + "epoch": 0.948740982714033, + "grad_norm": 2.3038294315338135, + "learning_rate": 1.0266321384762165e-06, + "loss": 2.4447969436645507, + "step": 348520 + }, + { + "epoch": 0.9487954267047775, + "grad_norm": 2.0933821201324463, + "learning_rate": 1.0255417745672619e-06, + "loss": 2.458466720581055, + "step": 348540 + }, + { + "epoch": 0.948849870695522, + "grad_norm": 2.2070155143737793, + "learning_rate": 1.0244514106583073e-06, + "loss": 2.4386186599731445, + "step": 348560 + }, + { + "epoch": 0.9489043146862665, + "grad_norm": 2.1967320442199707, + "learning_rate": 1.0233610467493526e-06, + "loss": 2.4722597122192385, + "step": 348580 + }, + { + "epoch": 0.9489587586770111, + "grad_norm": 2.1928300857543945, + "learning_rate": 1.022270682840398e-06, + "loss": 2.4628190994262695, + "step": 348600 + }, + { + "epoch": 0.9490132026677556, + "grad_norm": 2.2201905250549316, + "learning_rate": 1.0211803189314434e-06, + "loss": 2.4700077056884764, + "step": 348620 + }, + { + "epoch": 0.9490676466585001, + "grad_norm": 2.1143107414245605, + "learning_rate": 1.0200899550224888e-06, + "loss": 2.469452667236328, + "step": 348640 + }, + { + "epoch": 0.9491220906492446, + "grad_norm": 2.1108322143554688, + "learning_rate": 1.0189995911135342e-06, + "loss": 2.4722400665283204, + "step": 348660 + }, + { + "epoch": 0.9491765346399891, + "grad_norm": 2.1696901321411133, + "learning_rate": 1.0179092272045796e-06, + "loss": 2.453065299987793, + "step": 348680 + }, + { + "epoch": 0.9492309786307336, + "grad_norm": 2.233374834060669, + "learning_rate": 1.016818863295625e-06, + "loss": 2.455297088623047, + "step": 348700 + }, + { + "epoch": 0.9492854226214782, + "grad_norm": 2.115967273712158, + "learning_rate": 1.0157284993866703e-06, + "loss": 2.460116958618164, + "step": 348720 + }, + { + "epoch": 0.9493398666122227, + "grad_norm": 2.1517903804779053, + "learning_rate": 1.0146381354777157e-06, + "loss": 2.4552846908569337, + "step": 348740 + }, + { + "epoch": 0.9493943106029672, + "grad_norm": 2.2939510345458984, + "learning_rate": 1.0135477715687611e-06, + "loss": 2.460941696166992, + "step": 348760 + }, + { + "epoch": 0.9494487545937117, + "grad_norm": 2.2782464027404785, + "learning_rate": 1.0124574076598065e-06, + "loss": 2.477347564697266, + "step": 348780 + }, + { + "epoch": 0.9495031985844562, + "grad_norm": 2.1469194889068604, + "learning_rate": 1.0113670437508519e-06, + "loss": 2.430673027038574, + "step": 348800 + }, + { + "epoch": 0.9495576425752008, + "grad_norm": 2.1342668533325195, + "learning_rate": 1.0102766798418973e-06, + "loss": 2.467954635620117, + "step": 348820 + }, + { + "epoch": 0.9496120865659453, + "grad_norm": 2.213771343231201, + "learning_rate": 1.0091863159329427e-06, + "loss": 2.4429861068725587, + "step": 348840 + }, + { + "epoch": 0.9496665305566898, + "grad_norm": 3.7458655834198, + "learning_rate": 1.008095952023988e-06, + "loss": 2.469932746887207, + "step": 348860 + }, + { + "epoch": 0.9497209745474343, + "grad_norm": 2.182314872741699, + "learning_rate": 1.0070055881150334e-06, + "loss": 2.4781314849853517, + "step": 348880 + }, + { + "epoch": 0.9497754185381788, + "grad_norm": 2.2490551471710205, + "learning_rate": 1.005915224206079e-06, + "loss": 2.4573957443237306, + "step": 348900 + }, + { + "epoch": 0.9498298625289233, + "grad_norm": 2.1670799255371094, + "learning_rate": 1.0048248602971244e-06, + "loss": 2.446543312072754, + "step": 348920 + }, + { + "epoch": 0.9498843065196679, + "grad_norm": 2.1925296783447266, + "learning_rate": 1.0037344963881698e-06, + "loss": 2.497975540161133, + "step": 348940 + }, + { + "epoch": 0.9499387505104124, + "grad_norm": 2.1613376140594482, + "learning_rate": 1.0026441324792152e-06, + "loss": 2.4399925231933595, + "step": 348960 + }, + { + "epoch": 0.9499931945011569, + "grad_norm": 2.215346097946167, + "learning_rate": 1.0015537685702604e-06, + "loss": 2.4319747924804687, + "step": 348980 + }, + { + "epoch": 0.9500476384919014, + "grad_norm": 2.3224079608917236, + "learning_rate": 1.0004634046613057e-06, + "loss": 2.4657524108886717, + "step": 349000 + }, + { + "epoch": 0.9501020824826459, + "grad_norm": 2.258064031600952, + "learning_rate": 9.993730407523511e-07, + "loss": 2.454776954650879, + "step": 349020 + }, + { + "epoch": 0.9501565264733906, + "grad_norm": 2.0916192531585693, + "learning_rate": 9.982826768433965e-07, + "loss": 2.4415040969848634, + "step": 349040 + }, + { + "epoch": 0.950210970464135, + "grad_norm": 2.260622024536133, + "learning_rate": 9.97192312934442e-07, + "loss": 2.443537712097168, + "step": 349060 + }, + { + "epoch": 0.9502654144548796, + "grad_norm": 2.368648052215576, + "learning_rate": 9.961019490254873e-07, + "loss": 2.4492650985717774, + "step": 349080 + }, + { + "epoch": 0.950319858445624, + "grad_norm": 2.1703953742980957, + "learning_rate": 9.950115851165327e-07, + "loss": 2.484700584411621, + "step": 349100 + }, + { + "epoch": 0.9503743024363686, + "grad_norm": 2.2104005813598633, + "learning_rate": 9.93921221207578e-07, + "loss": 2.4761541366577147, + "step": 349120 + }, + { + "epoch": 0.9504287464271131, + "grad_norm": 2.671311855316162, + "learning_rate": 9.928308572986234e-07, + "loss": 2.4823904037475586, + "step": 349140 + }, + { + "epoch": 0.9504831904178577, + "grad_norm": 2.1826212406158447, + "learning_rate": 9.917404933896688e-07, + "loss": 2.449553680419922, + "step": 349160 + }, + { + "epoch": 0.9505376344086022, + "grad_norm": 2.2294540405273438, + "learning_rate": 9.906501294807142e-07, + "loss": 2.403702735900879, + "step": 349180 + }, + { + "epoch": 0.9505920783993467, + "grad_norm": 2.2729125022888184, + "learning_rate": 9.895597655717596e-07, + "loss": 2.442325210571289, + "step": 349200 + }, + { + "epoch": 0.9506465223900912, + "grad_norm": 2.211282253265381, + "learning_rate": 9.88469401662805e-07, + "loss": 2.4541717529296876, + "step": 349220 + }, + { + "epoch": 0.9507009663808357, + "grad_norm": 2.1192119121551514, + "learning_rate": 9.873790377538504e-07, + "loss": 2.46596736907959, + "step": 349240 + }, + { + "epoch": 0.9507554103715803, + "grad_norm": 2.557798147201538, + "learning_rate": 9.862886738448958e-07, + "loss": 2.4465137481689454, + "step": 349260 + }, + { + "epoch": 0.9508098543623248, + "grad_norm": 3.1250312328338623, + "learning_rate": 9.851983099359411e-07, + "loss": 2.4326759338378907, + "step": 349280 + }, + { + "epoch": 0.9508642983530693, + "grad_norm": 2.137293815612793, + "learning_rate": 9.841079460269865e-07, + "loss": 2.479672431945801, + "step": 349300 + }, + { + "epoch": 0.9509187423438138, + "grad_norm": 2.1201775074005127, + "learning_rate": 9.83017582118032e-07, + "loss": 2.4315420150756837, + "step": 349320 + }, + { + "epoch": 0.9509731863345583, + "grad_norm": 2.6218276023864746, + "learning_rate": 9.819272182090773e-07, + "loss": 2.464273452758789, + "step": 349340 + }, + { + "epoch": 0.9510276303253028, + "grad_norm": 2.2363743782043457, + "learning_rate": 9.808368543001227e-07, + "loss": 2.431739616394043, + "step": 349360 + }, + { + "epoch": 0.9510820743160474, + "grad_norm": 2.1641361713409424, + "learning_rate": 9.79746490391168e-07, + "loss": 2.4290708541870116, + "step": 349380 + }, + { + "epoch": 0.9511365183067919, + "grad_norm": 2.3524351119995117, + "learning_rate": 9.786561264822135e-07, + "loss": 2.4584583282470702, + "step": 349400 + }, + { + "epoch": 0.9511909622975364, + "grad_norm": 2.169772148132324, + "learning_rate": 9.775657625732588e-07, + "loss": 2.476827621459961, + "step": 349420 + }, + { + "epoch": 0.9512454062882809, + "grad_norm": 2.419675588607788, + "learning_rate": 9.764753986643042e-07, + "loss": 2.4421653747558594, + "step": 349440 + }, + { + "epoch": 0.9512998502790254, + "grad_norm": 2.1895177364349365, + "learning_rate": 9.753850347553496e-07, + "loss": 2.408854675292969, + "step": 349460 + }, + { + "epoch": 0.95135429426977, + "grad_norm": 2.1639528274536133, + "learning_rate": 9.74294670846395e-07, + "loss": 2.4588014602661135, + "step": 349480 + }, + { + "epoch": 0.9514087382605145, + "grad_norm": 2.2448673248291016, + "learning_rate": 9.732043069374404e-07, + "loss": 2.4490671157836914, + "step": 349500 + }, + { + "epoch": 0.951463182251259, + "grad_norm": 2.105717182159424, + "learning_rate": 9.721139430284858e-07, + "loss": 2.4479911804199217, + "step": 349520 + }, + { + "epoch": 0.9515176262420035, + "grad_norm": 2.2200679779052734, + "learning_rate": 9.710235791195312e-07, + "loss": 2.4486928939819337, + "step": 349540 + }, + { + "epoch": 0.951572070232748, + "grad_norm": 2.1835031509399414, + "learning_rate": 9.699332152105765e-07, + "loss": 2.498482513427734, + "step": 349560 + }, + { + "epoch": 0.9516265142234925, + "grad_norm": 2.0934314727783203, + "learning_rate": 9.68842851301622e-07, + "loss": 2.4802661895751954, + "step": 349580 + }, + { + "epoch": 0.9516809582142371, + "grad_norm": 2.2028753757476807, + "learning_rate": 9.677524873926673e-07, + "loss": 2.478076934814453, + "step": 349600 + }, + { + "epoch": 0.9517354022049817, + "grad_norm": 2.1442654132843018, + "learning_rate": 9.666621234837127e-07, + "loss": 2.4634765625, + "step": 349620 + }, + { + "epoch": 0.9517898461957262, + "grad_norm": 3.923370838165283, + "learning_rate": 9.65571759574758e-07, + "loss": 2.4543054580688475, + "step": 349640 + }, + { + "epoch": 0.9518442901864707, + "grad_norm": 2.088456869125366, + "learning_rate": 9.644813956658035e-07, + "loss": 2.49084529876709, + "step": 349660 + }, + { + "epoch": 0.9518987341772152, + "grad_norm": 2.2405731678009033, + "learning_rate": 9.633910317568489e-07, + "loss": 2.4717260360717774, + "step": 349680 + }, + { + "epoch": 0.9519531781679597, + "grad_norm": 2.1789281368255615, + "learning_rate": 9.623006678478942e-07, + "loss": 2.455335807800293, + "step": 349700 + }, + { + "epoch": 0.9520076221587043, + "grad_norm": 2.123392105102539, + "learning_rate": 9.612103039389396e-07, + "loss": 2.4661672592163084, + "step": 349720 + }, + { + "epoch": 0.9520620661494488, + "grad_norm": 2.226212501525879, + "learning_rate": 9.601199400299852e-07, + "loss": 2.4498970031738283, + "step": 349740 + }, + { + "epoch": 0.9521165101401933, + "grad_norm": 2.1282148361206055, + "learning_rate": 9.590295761210306e-07, + "loss": 2.4428947448730467, + "step": 349760 + }, + { + "epoch": 0.9521709541309378, + "grad_norm": 2.1468710899353027, + "learning_rate": 9.57939212212076e-07, + "loss": 2.44675235748291, + "step": 349780 + }, + { + "epoch": 0.9522253981216823, + "grad_norm": 2.2774598598480225, + "learning_rate": 9.568488483031214e-07, + "loss": 2.4856945037841798, + "step": 349800 + }, + { + "epoch": 0.9522798421124269, + "grad_norm": 2.1981735229492188, + "learning_rate": 9.557584843941668e-07, + "loss": 2.426077461242676, + "step": 349820 + }, + { + "epoch": 0.9523342861031714, + "grad_norm": 2.2341744899749756, + "learning_rate": 9.546681204852122e-07, + "loss": 2.4568613052368162, + "step": 349840 + }, + { + "epoch": 0.9523887300939159, + "grad_norm": 2.1310670375823975, + "learning_rate": 9.535777565762574e-07, + "loss": 2.4751279830932615, + "step": 349860 + }, + { + "epoch": 0.9524431740846604, + "grad_norm": 2.142225742340088, + "learning_rate": 9.524873926673028e-07, + "loss": 2.4279502868652343, + "step": 349880 + }, + { + "epoch": 0.9524976180754049, + "grad_norm": 2.1103739738464355, + "learning_rate": 9.513970287583482e-07, + "loss": 2.429928016662598, + "step": 349900 + }, + { + "epoch": 0.9525520620661494, + "grad_norm": 2.0862600803375244, + "learning_rate": 9.503066648493936e-07, + "loss": 2.4645622253417967, + "step": 349920 + }, + { + "epoch": 0.952606506056894, + "grad_norm": 2.173398494720459, + "learning_rate": 9.49216300940439e-07, + "loss": 2.4583547592163084, + "step": 349940 + }, + { + "epoch": 0.9526609500476385, + "grad_norm": 2.181708812713623, + "learning_rate": 9.481259370314844e-07, + "loss": 2.4540842056274412, + "step": 349960 + }, + { + "epoch": 0.952715394038383, + "grad_norm": 2.6310372352600098, + "learning_rate": 9.470355731225297e-07, + "loss": 2.465944290161133, + "step": 349980 + }, + { + "epoch": 0.9527698380291275, + "grad_norm": 2.1320247650146484, + "learning_rate": 9.459452092135751e-07, + "loss": 2.458542251586914, + "step": 350000 + }, + { + "epoch": 0.952824282019872, + "grad_norm": 2.1443850994110107, + "learning_rate": 9.448548453046205e-07, + "loss": 2.4601415634155273, + "step": 350020 + }, + { + "epoch": 0.9528787260106166, + "grad_norm": 2.2022721767425537, + "learning_rate": 9.437644813956659e-07, + "loss": 2.4611722946166994, + "step": 350040 + }, + { + "epoch": 0.9529331700013611, + "grad_norm": 2.207942247390747, + "learning_rate": 9.426741174867113e-07, + "loss": 2.460970878601074, + "step": 350060 + }, + { + "epoch": 0.9529876139921056, + "grad_norm": 2.1953065395355225, + "learning_rate": 9.415837535777567e-07, + "loss": 2.464069938659668, + "step": 350080 + }, + { + "epoch": 0.9530420579828501, + "grad_norm": 2.332519292831421, + "learning_rate": 9.404933896688021e-07, + "loss": 2.4426416397094726, + "step": 350100 + }, + { + "epoch": 0.9530965019735946, + "grad_norm": 2.1901330947875977, + "learning_rate": 9.394030257598474e-07, + "loss": 2.461216354370117, + "step": 350120 + }, + { + "epoch": 0.9531509459643391, + "grad_norm": 2.138965129852295, + "learning_rate": 9.383126618508928e-07, + "loss": 2.4433853149414064, + "step": 350140 + }, + { + "epoch": 0.9532053899550837, + "grad_norm": 2.1171207427978516, + "learning_rate": 9.372222979419382e-07, + "loss": 2.4474262237548827, + "step": 350160 + }, + { + "epoch": 0.9532598339458282, + "grad_norm": 2.1741437911987305, + "learning_rate": 9.361319340329836e-07, + "loss": 2.454656791687012, + "step": 350180 + }, + { + "epoch": 0.9533142779365728, + "grad_norm": 2.241067886352539, + "learning_rate": 9.35041570124029e-07, + "loss": 2.4615720748901366, + "step": 350200 + }, + { + "epoch": 0.9533687219273173, + "grad_norm": 2.177783966064453, + "learning_rate": 9.339512062150744e-07, + "loss": 2.464455223083496, + "step": 350220 + }, + { + "epoch": 0.9534231659180618, + "grad_norm": 2.1024065017700195, + "learning_rate": 9.328608423061198e-07, + "loss": 2.482996940612793, + "step": 350240 + }, + { + "epoch": 0.9534776099088064, + "grad_norm": 2.125339984893799, + "learning_rate": 9.317704783971651e-07, + "loss": 2.43640022277832, + "step": 350260 + }, + { + "epoch": 0.9535320538995509, + "grad_norm": 2.208615303039551, + "learning_rate": 9.306801144882105e-07, + "loss": 2.450023651123047, + "step": 350280 + }, + { + "epoch": 0.9535864978902954, + "grad_norm": 2.150506019592285, + "learning_rate": 9.295897505792559e-07, + "loss": 2.445269012451172, + "step": 350300 + }, + { + "epoch": 0.9536409418810399, + "grad_norm": 2.2032339572906494, + "learning_rate": 9.284993866703013e-07, + "loss": 2.4765142440795898, + "step": 350320 + }, + { + "epoch": 0.9536953858717844, + "grad_norm": 2.2777957916259766, + "learning_rate": 9.274090227613467e-07, + "loss": 2.4425302505493165, + "step": 350340 + }, + { + "epoch": 0.9537498298625289, + "grad_norm": 2.262288808822632, + "learning_rate": 9.263186588523921e-07, + "loss": 2.466840934753418, + "step": 350360 + }, + { + "epoch": 0.9538042738532735, + "grad_norm": 2.475059747695923, + "learning_rate": 9.252282949434375e-07, + "loss": 2.460484504699707, + "step": 350380 + }, + { + "epoch": 0.953858717844018, + "grad_norm": 3.5004336833953857, + "learning_rate": 9.241379310344828e-07, + "loss": 2.4407276153564452, + "step": 350400 + }, + { + "epoch": 0.9539131618347625, + "grad_norm": 2.1978743076324463, + "learning_rate": 9.230475671255281e-07, + "loss": 2.4713211059570312, + "step": 350420 + }, + { + "epoch": 0.953967605825507, + "grad_norm": 2.1636962890625, + "learning_rate": 9.219572032165735e-07, + "loss": 2.4507204055786134, + "step": 350440 + }, + { + "epoch": 0.9540220498162515, + "grad_norm": 2.11268949508667, + "learning_rate": 9.208668393076189e-07, + "loss": 2.4629789352416993, + "step": 350460 + }, + { + "epoch": 0.9540764938069961, + "grad_norm": 2.1738908290863037, + "learning_rate": 9.197764753986643e-07, + "loss": 2.4390384674072267, + "step": 350480 + }, + { + "epoch": 0.9541309377977406, + "grad_norm": 2.1225526332855225, + "learning_rate": 9.186861114897097e-07, + "loss": 2.4720270156860353, + "step": 350500 + }, + { + "epoch": 0.9541853817884851, + "grad_norm": 2.152305841445923, + "learning_rate": 9.17595747580755e-07, + "loss": 2.4499832153320313, + "step": 350520 + }, + { + "epoch": 0.9542398257792296, + "grad_norm": 6.752519130706787, + "learning_rate": 9.165053836718004e-07, + "loss": 2.452586364746094, + "step": 350540 + }, + { + "epoch": 0.9542942697699741, + "grad_norm": 2.084258794784546, + "learning_rate": 9.154150197628458e-07, + "loss": 2.4288967132568358, + "step": 350560 + }, + { + "epoch": 0.9543487137607186, + "grad_norm": 2.9600656032562256, + "learning_rate": 9.143246558538914e-07, + "loss": 2.4094013214111327, + "step": 350580 + }, + { + "epoch": 0.9544031577514632, + "grad_norm": 2.176217794418335, + "learning_rate": 9.132342919449368e-07, + "loss": 2.4496791839599608, + "step": 350600 + }, + { + "epoch": 0.9544576017422077, + "grad_norm": 2.245309352874756, + "learning_rate": 9.121439280359822e-07, + "loss": 2.436925506591797, + "step": 350620 + }, + { + "epoch": 0.9545120457329522, + "grad_norm": 2.2988486289978027, + "learning_rate": 9.110535641270276e-07, + "loss": 2.4575849533081056, + "step": 350640 + }, + { + "epoch": 0.9545664897236967, + "grad_norm": 2.1399054527282715, + "learning_rate": 9.09963200218073e-07, + "loss": 2.4404693603515626, + "step": 350660 + }, + { + "epoch": 0.9546209337144412, + "grad_norm": 2.134059429168701, + "learning_rate": 9.088728363091182e-07, + "loss": 2.449141502380371, + "step": 350680 + }, + { + "epoch": 0.9546753777051857, + "grad_norm": 2.161797285079956, + "learning_rate": 9.077824724001636e-07, + "loss": 2.443853569030762, + "step": 350700 + }, + { + "epoch": 0.9547298216959303, + "grad_norm": 2.0841124057769775, + "learning_rate": 9.06692108491209e-07, + "loss": 2.464777946472168, + "step": 350720 + }, + { + "epoch": 0.9547842656866748, + "grad_norm": 2.293334722518921, + "learning_rate": 9.056017445822544e-07, + "loss": 2.45369873046875, + "step": 350740 + }, + { + "epoch": 0.9548387096774194, + "grad_norm": 2.1737890243530273, + "learning_rate": 9.045113806732998e-07, + "loss": 2.455328941345215, + "step": 350760 + }, + { + "epoch": 0.9548931536681639, + "grad_norm": 2.262268543243408, + "learning_rate": 9.034210167643452e-07, + "loss": 2.4590484619140627, + "step": 350780 + }, + { + "epoch": 0.9549475976589084, + "grad_norm": 2.0699963569641113, + "learning_rate": 9.023306528553906e-07, + "loss": 2.4414525985717774, + "step": 350800 + }, + { + "epoch": 0.955002041649653, + "grad_norm": 2.1615419387817383, + "learning_rate": 9.012402889464359e-07, + "loss": 2.4875715255737303, + "step": 350820 + }, + { + "epoch": 0.9550564856403975, + "grad_norm": 2.127291679382324, + "learning_rate": 9.001499250374813e-07, + "loss": 2.494236183166504, + "step": 350840 + }, + { + "epoch": 0.955110929631142, + "grad_norm": 2.2805683612823486, + "learning_rate": 8.990595611285267e-07, + "loss": 2.472647476196289, + "step": 350860 + }, + { + "epoch": 0.9551653736218865, + "grad_norm": 2.159484624862671, + "learning_rate": 8.979691972195721e-07, + "loss": 2.4210527420043944, + "step": 350880 + }, + { + "epoch": 0.955219817612631, + "grad_norm": 2.2381184101104736, + "learning_rate": 8.968788333106175e-07, + "loss": 2.457685089111328, + "step": 350900 + }, + { + "epoch": 0.9552742616033755, + "grad_norm": 2.2212400436401367, + "learning_rate": 8.957884694016629e-07, + "loss": 2.458315849304199, + "step": 350920 + }, + { + "epoch": 0.9553287055941201, + "grad_norm": 2.127018928527832, + "learning_rate": 8.946981054927083e-07, + "loss": 2.450121307373047, + "step": 350940 + }, + { + "epoch": 0.9553831495848646, + "grad_norm": 2.1192126274108887, + "learning_rate": 8.936077415837536e-07, + "loss": 2.454562759399414, + "step": 350960 + }, + { + "epoch": 0.9554375935756091, + "grad_norm": 2.1585328578948975, + "learning_rate": 8.92517377674799e-07, + "loss": 2.4698001861572267, + "step": 350980 + }, + { + "epoch": 0.9554920375663536, + "grad_norm": 2.1634483337402344, + "learning_rate": 8.914270137658444e-07, + "loss": 2.452590751647949, + "step": 351000 + }, + { + "epoch": 0.9555464815570981, + "grad_norm": 2.2044484615325928, + "learning_rate": 8.903366498568898e-07, + "loss": 2.4406429290771485, + "step": 351020 + }, + { + "epoch": 0.9556009255478427, + "grad_norm": 2.154435396194458, + "learning_rate": 8.892462859479352e-07, + "loss": 2.438435745239258, + "step": 351040 + }, + { + "epoch": 0.9556553695385872, + "grad_norm": 2.0984957218170166, + "learning_rate": 8.881559220389806e-07, + "loss": 2.4279506683349608, + "step": 351060 + }, + { + "epoch": 0.9557098135293317, + "grad_norm": 2.1607143878936768, + "learning_rate": 8.87065558130026e-07, + "loss": 2.437275695800781, + "step": 351080 + }, + { + "epoch": 0.9557642575200762, + "grad_norm": 2.154597759246826, + "learning_rate": 8.859751942210713e-07, + "loss": 2.468624305725098, + "step": 351100 + }, + { + "epoch": 0.9558187015108207, + "grad_norm": 2.301464080810547, + "learning_rate": 8.848848303121167e-07, + "loss": 2.4613454818725584, + "step": 351120 + }, + { + "epoch": 0.9558731455015652, + "grad_norm": 2.1226024627685547, + "learning_rate": 8.837944664031621e-07, + "loss": 2.463974380493164, + "step": 351140 + }, + { + "epoch": 0.9559275894923098, + "grad_norm": 2.15545916557312, + "learning_rate": 8.827041024942075e-07, + "loss": 2.4565324783325195, + "step": 351160 + }, + { + "epoch": 0.9559820334830543, + "grad_norm": 2.1316187381744385, + "learning_rate": 8.816137385852529e-07, + "loss": 2.482953643798828, + "step": 351180 + }, + { + "epoch": 0.9560364774737988, + "grad_norm": 2.1259605884552, + "learning_rate": 8.805233746762983e-07, + "loss": 2.46197624206543, + "step": 351200 + }, + { + "epoch": 0.9560909214645433, + "grad_norm": 2.7574687004089355, + "learning_rate": 8.794330107673437e-07, + "loss": 2.466006851196289, + "step": 351220 + }, + { + "epoch": 0.9561453654552878, + "grad_norm": 2.091499090194702, + "learning_rate": 8.78342646858389e-07, + "loss": 2.4627567291259767, + "step": 351240 + }, + { + "epoch": 0.9561998094460324, + "grad_norm": 2.213979959487915, + "learning_rate": 8.772522829494344e-07, + "loss": 2.4486846923828125, + "step": 351260 + }, + { + "epoch": 0.956254253436777, + "grad_norm": 2.1283302307128906, + "learning_rate": 8.761619190404798e-07, + "loss": 2.4512044906616213, + "step": 351280 + }, + { + "epoch": 0.9563086974275214, + "grad_norm": 2.372837543487549, + "learning_rate": 8.750715551315252e-07, + "loss": 2.4444660186767577, + "step": 351300 + }, + { + "epoch": 0.956363141418266, + "grad_norm": 2.13641095161438, + "learning_rate": 8.739811912225706e-07, + "loss": 2.468976593017578, + "step": 351320 + }, + { + "epoch": 0.9564175854090105, + "grad_norm": 2.1498003005981445, + "learning_rate": 8.72890827313616e-07, + "loss": 2.436046028137207, + "step": 351340 + }, + { + "epoch": 0.956472029399755, + "grad_norm": 2.1050145626068115, + "learning_rate": 8.718004634046614e-07, + "loss": 2.444528007507324, + "step": 351360 + }, + { + "epoch": 0.9565264733904996, + "grad_norm": 2.2523539066314697, + "learning_rate": 8.707100994957067e-07, + "loss": 2.4715179443359374, + "step": 351380 + }, + { + "epoch": 0.9565809173812441, + "grad_norm": 2.0873212814331055, + "learning_rate": 8.696197355867521e-07, + "loss": 2.4376810073852537, + "step": 351400 + }, + { + "epoch": 0.9566353613719886, + "grad_norm": 2.2225489616394043, + "learning_rate": 8.685293716777975e-07, + "loss": 2.462472152709961, + "step": 351420 + }, + { + "epoch": 0.9566898053627331, + "grad_norm": 2.1134891510009766, + "learning_rate": 8.67439007768843e-07, + "loss": 2.467983055114746, + "step": 351440 + }, + { + "epoch": 0.9567442493534776, + "grad_norm": 2.1232335567474365, + "learning_rate": 8.663486438598884e-07, + "loss": 2.4523603439331056, + "step": 351460 + }, + { + "epoch": 0.9567986933442222, + "grad_norm": 2.1784257888793945, + "learning_rate": 8.652582799509338e-07, + "loss": 2.4494585037231444, + "step": 351480 + }, + { + "epoch": 0.9568531373349667, + "grad_norm": 2.1116507053375244, + "learning_rate": 8.641679160419792e-07, + "loss": 2.488919258117676, + "step": 351500 + }, + { + "epoch": 0.9569075813257112, + "grad_norm": 2.2714030742645264, + "learning_rate": 8.630775521330245e-07, + "loss": 2.4606159210205076, + "step": 351520 + }, + { + "epoch": 0.9569620253164557, + "grad_norm": 2.143476724624634, + "learning_rate": 8.619871882240699e-07, + "loss": 2.4468147277832033, + "step": 351540 + }, + { + "epoch": 0.9570164693072002, + "grad_norm": 2.1834545135498047, + "learning_rate": 8.608968243151153e-07, + "loss": 2.451340675354004, + "step": 351560 + }, + { + "epoch": 0.9570709132979447, + "grad_norm": 2.1419901847839355, + "learning_rate": 8.598064604061607e-07, + "loss": 2.4634634017944337, + "step": 351580 + }, + { + "epoch": 0.9571253572886893, + "grad_norm": 2.1790173053741455, + "learning_rate": 8.587160964972061e-07, + "loss": 2.4241817474365233, + "step": 351600 + }, + { + "epoch": 0.9571798012794338, + "grad_norm": 2.4928083419799805, + "learning_rate": 8.576257325882515e-07, + "loss": 2.5003528594970703, + "step": 351620 + }, + { + "epoch": 0.9572342452701783, + "grad_norm": 2.2665185928344727, + "learning_rate": 8.565353686792969e-07, + "loss": 2.461861991882324, + "step": 351640 + }, + { + "epoch": 0.9572886892609228, + "grad_norm": 2.1072378158569336, + "learning_rate": 8.554450047703422e-07, + "loss": 2.463109016418457, + "step": 351660 + }, + { + "epoch": 0.9573431332516673, + "grad_norm": 2.1879892349243164, + "learning_rate": 8.543546408613876e-07, + "loss": 2.46834774017334, + "step": 351680 + }, + { + "epoch": 0.9573975772424119, + "grad_norm": 2.254565477371216, + "learning_rate": 8.53264276952433e-07, + "loss": 2.4611227035522463, + "step": 351700 + }, + { + "epoch": 0.9574520212331564, + "grad_norm": 2.138934373855591, + "learning_rate": 8.521739130434783e-07, + "loss": 2.4483922958374023, + "step": 351720 + }, + { + "epoch": 0.9575064652239009, + "grad_norm": 2.2174978256225586, + "learning_rate": 8.510835491345237e-07, + "loss": 2.484824371337891, + "step": 351740 + }, + { + "epoch": 0.9575609092146454, + "grad_norm": 2.1678531169891357, + "learning_rate": 8.499931852255691e-07, + "loss": 2.4706136703491213, + "step": 351760 + }, + { + "epoch": 0.9576153532053899, + "grad_norm": 2.0551838874816895, + "learning_rate": 8.489028213166145e-07, + "loss": 2.4602582931518553, + "step": 351780 + }, + { + "epoch": 0.9576697971961344, + "grad_norm": 2.199105739593506, + "learning_rate": 8.478124574076598e-07, + "loss": 2.453785514831543, + "step": 351800 + }, + { + "epoch": 0.957724241186879, + "grad_norm": 2.220552444458008, + "learning_rate": 8.467220934987052e-07, + "loss": 2.4452133178710938, + "step": 351820 + }, + { + "epoch": 0.9577786851776235, + "grad_norm": 2.420825242996216, + "learning_rate": 8.456317295897506e-07, + "loss": 2.4449539184570312, + "step": 351840 + }, + { + "epoch": 0.957833129168368, + "grad_norm": 2.19104266166687, + "learning_rate": 8.44541365680796e-07, + "loss": 2.4812341690063477, + "step": 351860 + }, + { + "epoch": 0.9578875731591125, + "grad_norm": 2.1258785724639893, + "learning_rate": 8.434510017718414e-07, + "loss": 2.467973327636719, + "step": 351880 + }, + { + "epoch": 0.957942017149857, + "grad_norm": 2.1807706356048584, + "learning_rate": 8.423606378628868e-07, + "loss": 2.468144989013672, + "step": 351900 + }, + { + "epoch": 0.9579964611406016, + "grad_norm": 2.202852964401245, + "learning_rate": 8.412702739539322e-07, + "loss": 2.4403575897216796, + "step": 351920 + }, + { + "epoch": 0.9580509051313462, + "grad_norm": 2.2319600582122803, + "learning_rate": 8.401799100449775e-07, + "loss": 2.475321578979492, + "step": 351940 + }, + { + "epoch": 0.9581053491220907, + "grad_norm": 2.1004388332366943, + "learning_rate": 8.390895461360229e-07, + "loss": 2.477943229675293, + "step": 351960 + }, + { + "epoch": 0.9581597931128352, + "grad_norm": 2.4209847450256348, + "learning_rate": 8.379991822270683e-07, + "loss": 2.473113441467285, + "step": 351980 + }, + { + "epoch": 0.9582142371035797, + "grad_norm": 2.431953191757202, + "learning_rate": 8.369088183181137e-07, + "loss": 2.4645664215087892, + "step": 352000 + }, + { + "epoch": 0.9582686810943242, + "grad_norm": 2.120767116546631, + "learning_rate": 8.358184544091591e-07, + "loss": 2.4831172943115236, + "step": 352020 + }, + { + "epoch": 0.9583231250850688, + "grad_norm": 2.115478277206421, + "learning_rate": 8.347280905002045e-07, + "loss": 2.469455337524414, + "step": 352040 + }, + { + "epoch": 0.9583775690758133, + "grad_norm": 2.131136178970337, + "learning_rate": 8.336377265912499e-07, + "loss": 2.4597545623779298, + "step": 352060 + }, + { + "epoch": 0.9584320130665578, + "grad_norm": 2.146509885787964, + "learning_rate": 8.325473626822952e-07, + "loss": 2.4606128692626954, + "step": 352080 + }, + { + "epoch": 0.9584864570573023, + "grad_norm": 2.2756874561309814, + "learning_rate": 8.314569987733406e-07, + "loss": 2.4419200897216795, + "step": 352100 + }, + { + "epoch": 0.9585409010480468, + "grad_norm": 2.232490301132202, + "learning_rate": 8.30366634864386e-07, + "loss": 2.4681751251220705, + "step": 352120 + }, + { + "epoch": 0.9585953450387913, + "grad_norm": 2.2153053283691406, + "learning_rate": 8.292762709554314e-07, + "loss": 2.4639596939086914, + "step": 352140 + }, + { + "epoch": 0.9586497890295359, + "grad_norm": 2.158090591430664, + "learning_rate": 8.281859070464768e-07, + "loss": 2.4572683334350587, + "step": 352160 + }, + { + "epoch": 0.9587042330202804, + "grad_norm": 2.2867164611816406, + "learning_rate": 8.270955431375222e-07, + "loss": 2.4371538162231445, + "step": 352180 + }, + { + "epoch": 0.9587586770110249, + "grad_norm": 2.1477389335632324, + "learning_rate": 8.260051792285676e-07, + "loss": 2.459969711303711, + "step": 352200 + }, + { + "epoch": 0.9588131210017694, + "grad_norm": 2.3798036575317383, + "learning_rate": 8.249148153196129e-07, + "loss": 2.481698989868164, + "step": 352220 + }, + { + "epoch": 0.9588675649925139, + "grad_norm": 2.2026138305664062, + "learning_rate": 8.238244514106583e-07, + "loss": 2.4406328201293945, + "step": 352240 + }, + { + "epoch": 0.9589220089832585, + "grad_norm": 2.1953368186950684, + "learning_rate": 8.227340875017037e-07, + "loss": 2.46319637298584, + "step": 352260 + }, + { + "epoch": 0.958976452974003, + "grad_norm": 2.1357736587524414, + "learning_rate": 8.216437235927492e-07, + "loss": 2.450445556640625, + "step": 352280 + }, + { + "epoch": 0.9590308969647475, + "grad_norm": 2.236117362976074, + "learning_rate": 8.205533596837946e-07, + "loss": 2.47236328125, + "step": 352300 + }, + { + "epoch": 0.959085340955492, + "grad_norm": 2.1029276847839355, + "learning_rate": 8.1946299577484e-07, + "loss": 2.4322158813476564, + "step": 352320 + }, + { + "epoch": 0.9591397849462365, + "grad_norm": 2.272139310836792, + "learning_rate": 8.183726318658854e-07, + "loss": 2.440318489074707, + "step": 352340 + }, + { + "epoch": 0.959194228936981, + "grad_norm": 2.053541898727417, + "learning_rate": 8.172822679569307e-07, + "loss": 2.4430294036865234, + "step": 352360 + }, + { + "epoch": 0.9592486729277256, + "grad_norm": 2.2569239139556885, + "learning_rate": 8.161919040479761e-07, + "loss": 2.4913116455078126, + "step": 352380 + }, + { + "epoch": 0.9593031169184701, + "grad_norm": 2.338249683380127, + "learning_rate": 8.151015401390215e-07, + "loss": 2.461634063720703, + "step": 352400 + }, + { + "epoch": 0.9593575609092146, + "grad_norm": 2.3179736137390137, + "learning_rate": 8.140111762300669e-07, + "loss": 2.454416275024414, + "step": 352420 + }, + { + "epoch": 0.9594120048999591, + "grad_norm": 2.0910685062408447, + "learning_rate": 8.129208123211123e-07, + "loss": 2.4379383087158204, + "step": 352440 + }, + { + "epoch": 0.9594664488907036, + "grad_norm": 2.1747353076934814, + "learning_rate": 8.118304484121577e-07, + "loss": 2.486322212219238, + "step": 352460 + }, + { + "epoch": 0.9595208928814483, + "grad_norm": 2.3492398262023926, + "learning_rate": 8.107400845032031e-07, + "loss": 2.4716352462768554, + "step": 352480 + }, + { + "epoch": 0.9595753368721928, + "grad_norm": 2.18989634513855, + "learning_rate": 8.096497205942484e-07, + "loss": 2.460053825378418, + "step": 352500 + }, + { + "epoch": 0.9596297808629373, + "grad_norm": 2.2295823097229004, + "learning_rate": 8.085593566852938e-07, + "loss": 2.448078727722168, + "step": 352520 + }, + { + "epoch": 0.9596842248536818, + "grad_norm": 2.6193790435791016, + "learning_rate": 8.074689927763392e-07, + "loss": 2.488469123840332, + "step": 352540 + }, + { + "epoch": 0.9597386688444263, + "grad_norm": 2.3702096939086914, + "learning_rate": 8.063786288673846e-07, + "loss": 2.4446361541748045, + "step": 352560 + }, + { + "epoch": 0.9597931128351708, + "grad_norm": 2.132111072540283, + "learning_rate": 8.0528826495843e-07, + "loss": 2.413182258605957, + "step": 352580 + }, + { + "epoch": 0.9598475568259154, + "grad_norm": 2.155590295791626, + "learning_rate": 8.041979010494754e-07, + "loss": 2.488195037841797, + "step": 352600 + }, + { + "epoch": 0.9599020008166599, + "grad_norm": 2.1425342559814453, + "learning_rate": 8.031075371405208e-07, + "loss": 2.483505439758301, + "step": 352620 + }, + { + "epoch": 0.9599564448074044, + "grad_norm": 2.6822383403778076, + "learning_rate": 8.020171732315661e-07, + "loss": 2.463607406616211, + "step": 352640 + }, + { + "epoch": 0.9600108887981489, + "grad_norm": 2.1760952472686768, + "learning_rate": 8.009268093226115e-07, + "loss": 2.423660087585449, + "step": 352660 + }, + { + "epoch": 0.9600653327888934, + "grad_norm": 2.1650636196136475, + "learning_rate": 7.998364454136569e-07, + "loss": 2.46704044342041, + "step": 352680 + }, + { + "epoch": 0.960119776779638, + "grad_norm": 2.155306100845337, + "learning_rate": 7.987460815047023e-07, + "loss": 2.445388412475586, + "step": 352700 + }, + { + "epoch": 0.9601742207703825, + "grad_norm": 2.395766258239746, + "learning_rate": 7.976557175957477e-07, + "loss": 2.4186122894287108, + "step": 352720 + }, + { + "epoch": 0.960228664761127, + "grad_norm": 2.1494476795196533, + "learning_rate": 7.965653536867931e-07, + "loss": 2.470869445800781, + "step": 352740 + }, + { + "epoch": 0.9602831087518715, + "grad_norm": 2.0840859413146973, + "learning_rate": 7.954749897778384e-07, + "loss": 2.450109672546387, + "step": 352760 + }, + { + "epoch": 0.960337552742616, + "grad_norm": 2.271646022796631, + "learning_rate": 7.943846258688837e-07, + "loss": 2.477682113647461, + "step": 352780 + }, + { + "epoch": 0.9603919967333605, + "grad_norm": 2.3301806449890137, + "learning_rate": 7.932942619599291e-07, + "loss": 2.451028251647949, + "step": 352800 + }, + { + "epoch": 0.9604464407241051, + "grad_norm": 2.1739721298217773, + "learning_rate": 7.922038980509745e-07, + "loss": 2.4685747146606447, + "step": 352820 + }, + { + "epoch": 0.9605008847148496, + "grad_norm": 2.1898064613342285, + "learning_rate": 7.911135341420199e-07, + "loss": 2.4605262756347654, + "step": 352840 + }, + { + "epoch": 0.9605553287055941, + "grad_norm": 2.1690192222595215, + "learning_rate": 7.900231702330653e-07, + "loss": 2.465250778198242, + "step": 352860 + }, + { + "epoch": 0.9606097726963386, + "grad_norm": 2.112919569015503, + "learning_rate": 7.889328063241107e-07, + "loss": 2.466477966308594, + "step": 352880 + }, + { + "epoch": 0.9606642166870831, + "grad_norm": 2.222947835922241, + "learning_rate": 7.878424424151561e-07, + "loss": 2.4624019622802735, + "step": 352900 + }, + { + "epoch": 0.9607186606778276, + "grad_norm": 2.161391496658325, + "learning_rate": 7.867520785062014e-07, + "loss": 2.465654182434082, + "step": 352920 + }, + { + "epoch": 0.9607731046685722, + "grad_norm": 2.1493184566497803, + "learning_rate": 7.856617145972468e-07, + "loss": 2.459163475036621, + "step": 352940 + }, + { + "epoch": 0.9608275486593167, + "grad_norm": 2.3179988861083984, + "learning_rate": 7.845713506882922e-07, + "loss": 2.4590175628662108, + "step": 352960 + }, + { + "epoch": 0.9608819926500612, + "grad_norm": 2.182685375213623, + "learning_rate": 7.834809867793376e-07, + "loss": 2.4378330230712892, + "step": 352980 + }, + { + "epoch": 0.9609364366408057, + "grad_norm": 2.3444736003875732, + "learning_rate": 7.82390622870383e-07, + "loss": 2.443465805053711, + "step": 353000 + }, + { + "epoch": 0.9609908806315502, + "grad_norm": 2.192911148071289, + "learning_rate": 7.813002589614284e-07, + "loss": 2.4605390548706056, + "step": 353020 + }, + { + "epoch": 0.9610453246222949, + "grad_norm": 2.330294609069824, + "learning_rate": 7.802098950524738e-07, + "loss": 2.4562198638916017, + "step": 353040 + }, + { + "epoch": 0.9610997686130394, + "grad_norm": 2.185724973678589, + "learning_rate": 7.791195311435191e-07, + "loss": 2.452997398376465, + "step": 353060 + }, + { + "epoch": 0.9611542126037839, + "grad_norm": 2.058422327041626, + "learning_rate": 7.780291672345645e-07, + "loss": 2.4292282104492187, + "step": 353080 + }, + { + "epoch": 0.9612086565945284, + "grad_norm": 2.1516292095184326, + "learning_rate": 7.769388033256099e-07, + "loss": 2.473719024658203, + "step": 353100 + }, + { + "epoch": 0.9612631005852729, + "grad_norm": 3.1816790103912354, + "learning_rate": 7.758484394166554e-07, + "loss": 2.468141555786133, + "step": 353120 + }, + { + "epoch": 0.9613175445760174, + "grad_norm": 2.171504497528076, + "learning_rate": 7.747580755077008e-07, + "loss": 2.4414241790771483, + "step": 353140 + }, + { + "epoch": 0.961371988566762, + "grad_norm": 2.093100070953369, + "learning_rate": 7.736677115987462e-07, + "loss": 2.4816152572631838, + "step": 353160 + }, + { + "epoch": 0.9614264325575065, + "grad_norm": 2.1459953784942627, + "learning_rate": 7.725773476897916e-07, + "loss": 2.4762451171875, + "step": 353180 + }, + { + "epoch": 0.961480876548251, + "grad_norm": 2.5813095569610596, + "learning_rate": 7.71486983780837e-07, + "loss": 2.4308115005493165, + "step": 353200 + }, + { + "epoch": 0.9615353205389955, + "grad_norm": 2.4648256301879883, + "learning_rate": 7.703966198718823e-07, + "loss": 2.4616487503051756, + "step": 353220 + }, + { + "epoch": 0.96158976452974, + "grad_norm": 2.1870036125183105, + "learning_rate": 7.693062559629277e-07, + "loss": 2.4443126678466798, + "step": 353240 + }, + { + "epoch": 0.9616442085204846, + "grad_norm": 2.3824517726898193, + "learning_rate": 7.682158920539731e-07, + "loss": 2.4823175430297852, + "step": 353260 + }, + { + "epoch": 0.9616986525112291, + "grad_norm": 2.275909423828125, + "learning_rate": 7.671255281450185e-07, + "loss": 2.448736572265625, + "step": 353280 + }, + { + "epoch": 0.9617530965019736, + "grad_norm": 2.2332496643066406, + "learning_rate": 7.660351642360639e-07, + "loss": 2.449398231506348, + "step": 353300 + }, + { + "epoch": 0.9618075404927181, + "grad_norm": 2.137622356414795, + "learning_rate": 7.649448003271093e-07, + "loss": 2.457689666748047, + "step": 353320 + }, + { + "epoch": 0.9618619844834626, + "grad_norm": 2.1331429481506348, + "learning_rate": 7.638544364181546e-07, + "loss": 2.452582359313965, + "step": 353340 + }, + { + "epoch": 0.9619164284742071, + "grad_norm": 2.254743814468384, + "learning_rate": 7.627640725092e-07, + "loss": 2.461313819885254, + "step": 353360 + }, + { + "epoch": 0.9619708724649517, + "grad_norm": 2.257249116897583, + "learning_rate": 7.616737086002454e-07, + "loss": 2.4935981750488283, + "step": 353380 + }, + { + "epoch": 0.9620253164556962, + "grad_norm": 2.1010098457336426, + "learning_rate": 7.605833446912908e-07, + "loss": 2.4859947204589843, + "step": 353400 + }, + { + "epoch": 0.9620797604464407, + "grad_norm": 2.1425087451934814, + "learning_rate": 7.594929807823362e-07, + "loss": 2.460502052307129, + "step": 353420 + }, + { + "epoch": 0.9621342044371852, + "grad_norm": 2.181656837463379, + "learning_rate": 7.584026168733816e-07, + "loss": 2.4710865020751953, + "step": 353440 + }, + { + "epoch": 0.9621886484279297, + "grad_norm": 2.215878486633301, + "learning_rate": 7.57312252964427e-07, + "loss": 2.4615097045898438, + "step": 353460 + }, + { + "epoch": 0.9622430924186743, + "grad_norm": 2.1386680603027344, + "learning_rate": 7.562218890554723e-07, + "loss": 2.4264623641967775, + "step": 353480 + }, + { + "epoch": 0.9622975364094188, + "grad_norm": 2.213007926940918, + "learning_rate": 7.551315251465177e-07, + "loss": 2.4641895294189453, + "step": 353500 + }, + { + "epoch": 0.9623519804001633, + "grad_norm": 2.2335100173950195, + "learning_rate": 7.540411612375631e-07, + "loss": 2.439118766784668, + "step": 353520 + }, + { + "epoch": 0.9624064243909078, + "grad_norm": 2.1340391635894775, + "learning_rate": 7.529507973286085e-07, + "loss": 2.4409780502319336, + "step": 353540 + }, + { + "epoch": 0.9624608683816523, + "grad_norm": 2.2744088172912598, + "learning_rate": 7.518604334196539e-07, + "loss": 2.4582691192626953, + "step": 353560 + }, + { + "epoch": 0.9625153123723968, + "grad_norm": 2.1523611545562744, + "learning_rate": 7.507700695106993e-07, + "loss": 2.4422603607177735, + "step": 353580 + }, + { + "epoch": 0.9625697563631415, + "grad_norm": 2.14823579788208, + "learning_rate": 7.496797056017447e-07, + "loss": 2.433456802368164, + "step": 353600 + }, + { + "epoch": 0.962624200353886, + "grad_norm": 2.1634929180145264, + "learning_rate": 7.4858934169279e-07, + "loss": 2.449726676940918, + "step": 353620 + }, + { + "epoch": 0.9626786443446305, + "grad_norm": 2.0774660110473633, + "learning_rate": 7.474989777838354e-07, + "loss": 2.453421401977539, + "step": 353640 + }, + { + "epoch": 0.962733088335375, + "grad_norm": 2.1634018421173096, + "learning_rate": 7.464086138748808e-07, + "loss": 2.452396774291992, + "step": 353660 + }, + { + "epoch": 0.9627875323261195, + "grad_norm": 2.3875086307525635, + "learning_rate": 7.453182499659262e-07, + "loss": 2.4424903869628904, + "step": 353680 + }, + { + "epoch": 0.9628419763168641, + "grad_norm": 2.116349697113037, + "learning_rate": 7.442278860569716e-07, + "loss": 2.478017044067383, + "step": 353700 + }, + { + "epoch": 0.9628964203076086, + "grad_norm": 2.230351448059082, + "learning_rate": 7.43137522148017e-07, + "loss": 2.4662267684936525, + "step": 353720 + }, + { + "epoch": 0.9629508642983531, + "grad_norm": 2.445967435836792, + "learning_rate": 7.420471582390624e-07, + "loss": 2.4887619018554688, + "step": 353740 + }, + { + "epoch": 0.9630053082890976, + "grad_norm": 2.343846082687378, + "learning_rate": 7.409567943301077e-07, + "loss": 2.4928310394287108, + "step": 353760 + }, + { + "epoch": 0.9630597522798421, + "grad_norm": 2.2456817626953125, + "learning_rate": 7.398664304211531e-07, + "loss": 2.453443908691406, + "step": 353780 + }, + { + "epoch": 0.9631141962705866, + "grad_norm": 2.1765506267547607, + "learning_rate": 7.387760665121984e-07, + "loss": 2.4398120880126952, + "step": 353800 + }, + { + "epoch": 0.9631686402613312, + "grad_norm": 2.263152599334717, + "learning_rate": 7.376857026032438e-07, + "loss": 2.4781612396240233, + "step": 353820 + }, + { + "epoch": 0.9632230842520757, + "grad_norm": 2.3345863819122314, + "learning_rate": 7.365953386942892e-07, + "loss": 2.4740758895874024, + "step": 353840 + }, + { + "epoch": 0.9632775282428202, + "grad_norm": 2.1671392917633057, + "learning_rate": 7.355049747853346e-07, + "loss": 2.445692253112793, + "step": 353860 + }, + { + "epoch": 0.9633319722335647, + "grad_norm": 2.2017102241516113, + "learning_rate": 7.3441461087638e-07, + "loss": 2.4465688705444335, + "step": 353880 + }, + { + "epoch": 0.9633864162243092, + "grad_norm": 2.0672338008880615, + "learning_rate": 7.333242469674253e-07, + "loss": 2.463761901855469, + "step": 353900 + }, + { + "epoch": 0.9634408602150538, + "grad_norm": 2.1550374031066895, + "learning_rate": 7.322338830584707e-07, + "loss": 2.456124114990234, + "step": 353920 + }, + { + "epoch": 0.9634953042057983, + "grad_norm": 2.1951048374176025, + "learning_rate": 7.311435191495161e-07, + "loss": 2.4248367309570313, + "step": 353940 + }, + { + "epoch": 0.9635497481965428, + "grad_norm": 2.086315155029297, + "learning_rate": 7.300531552405615e-07, + "loss": 2.4597122192382814, + "step": 353960 + }, + { + "epoch": 0.9636041921872873, + "grad_norm": 2.2695906162261963, + "learning_rate": 7.289627913316071e-07, + "loss": 2.451654815673828, + "step": 353980 + }, + { + "epoch": 0.9636586361780318, + "grad_norm": 2.836993455886841, + "learning_rate": 7.278724274226525e-07, + "loss": 2.451350784301758, + "step": 354000 + }, + { + "epoch": 0.9637130801687763, + "grad_norm": 2.137652635574341, + "learning_rate": 7.267820635136979e-07, + "loss": 2.4426952362060548, + "step": 354020 + }, + { + "epoch": 0.9637675241595209, + "grad_norm": 2.1147217750549316, + "learning_rate": 7.256916996047431e-07, + "loss": 2.4687055587768554, + "step": 354040 + }, + { + "epoch": 0.9638219681502654, + "grad_norm": 2.2887580394744873, + "learning_rate": 7.246013356957885e-07, + "loss": 2.4684516906738283, + "step": 354060 + }, + { + "epoch": 0.9638764121410099, + "grad_norm": 2.0641908645629883, + "learning_rate": 7.235109717868339e-07, + "loss": 2.4756103515625, + "step": 354080 + }, + { + "epoch": 0.9639308561317544, + "grad_norm": 2.288494110107422, + "learning_rate": 7.224206078778793e-07, + "loss": 2.449272346496582, + "step": 354100 + }, + { + "epoch": 0.963985300122499, + "grad_norm": 2.1489057540893555, + "learning_rate": 7.213302439689247e-07, + "loss": 2.460859680175781, + "step": 354120 + }, + { + "epoch": 0.9640397441132434, + "grad_norm": 2.1022109985351562, + "learning_rate": 7.202398800599701e-07, + "loss": 2.439683532714844, + "step": 354140 + }, + { + "epoch": 0.9640941881039881, + "grad_norm": 2.3684892654418945, + "learning_rate": 7.191495161510155e-07, + "loss": 2.4501068115234377, + "step": 354160 + }, + { + "epoch": 0.9641486320947326, + "grad_norm": 2.1483960151672363, + "learning_rate": 7.180591522420608e-07, + "loss": 2.4538219451904295, + "step": 354180 + }, + { + "epoch": 0.9642030760854771, + "grad_norm": 2.1323893070220947, + "learning_rate": 7.169687883331062e-07, + "loss": 2.4431764602661135, + "step": 354200 + }, + { + "epoch": 0.9642575200762216, + "grad_norm": 2.1490440368652344, + "learning_rate": 7.158784244241516e-07, + "loss": 2.4558109283447265, + "step": 354220 + }, + { + "epoch": 0.9643119640669661, + "grad_norm": 2.282989740371704, + "learning_rate": 7.14788060515197e-07, + "loss": 2.448965072631836, + "step": 354240 + }, + { + "epoch": 0.9643664080577107, + "grad_norm": 2.1539273262023926, + "learning_rate": 7.136976966062424e-07, + "loss": 2.412673759460449, + "step": 354260 + }, + { + "epoch": 0.9644208520484552, + "grad_norm": 2.6866695880889893, + "learning_rate": 7.126073326972878e-07, + "loss": 2.4514389038085938, + "step": 354280 + }, + { + "epoch": 0.9644752960391997, + "grad_norm": 2.1607766151428223, + "learning_rate": 7.115169687883332e-07, + "loss": 2.4499113082885744, + "step": 354300 + }, + { + "epoch": 0.9645297400299442, + "grad_norm": 2.2501423358917236, + "learning_rate": 7.104266048793785e-07, + "loss": 2.458065986633301, + "step": 354320 + }, + { + "epoch": 0.9645841840206887, + "grad_norm": 2.1605682373046875, + "learning_rate": 7.093362409704239e-07, + "loss": 2.4440120697021483, + "step": 354340 + }, + { + "epoch": 0.9646386280114332, + "grad_norm": 2.0754597187042236, + "learning_rate": 7.082458770614693e-07, + "loss": 2.4696687698364257, + "step": 354360 + }, + { + "epoch": 0.9646930720021778, + "grad_norm": 2.100167751312256, + "learning_rate": 7.071555131525147e-07, + "loss": 2.471893882751465, + "step": 354380 + }, + { + "epoch": 0.9647475159929223, + "grad_norm": 2.180392026901245, + "learning_rate": 7.060651492435601e-07, + "loss": 2.4474496841430664, + "step": 354400 + }, + { + "epoch": 0.9648019599836668, + "grad_norm": 2.235208749771118, + "learning_rate": 7.049747853346055e-07, + "loss": 2.443048286437988, + "step": 354420 + }, + { + "epoch": 0.9648564039744113, + "grad_norm": 2.0801007747650146, + "learning_rate": 7.038844214256509e-07, + "loss": 2.4530813217163088, + "step": 354440 + }, + { + "epoch": 0.9649108479651558, + "grad_norm": 2.1536800861358643, + "learning_rate": 7.027940575166962e-07, + "loss": 2.422709274291992, + "step": 354460 + }, + { + "epoch": 0.9649652919559004, + "grad_norm": 2.1735401153564453, + "learning_rate": 7.017036936077416e-07, + "loss": 2.4660417556762697, + "step": 354480 + }, + { + "epoch": 0.9650197359466449, + "grad_norm": 2.067909002304077, + "learning_rate": 7.00613329698787e-07, + "loss": 2.4579673767089845, + "step": 354500 + }, + { + "epoch": 0.9650741799373894, + "grad_norm": 2.137151002883911, + "learning_rate": 6.995229657898324e-07, + "loss": 2.4394811630249023, + "step": 354520 + }, + { + "epoch": 0.9651286239281339, + "grad_norm": 3.337752103805542, + "learning_rate": 6.984326018808778e-07, + "loss": 2.468097114562988, + "step": 354540 + }, + { + "epoch": 0.9651830679188784, + "grad_norm": 2.1254682540893555, + "learning_rate": 6.973422379719232e-07, + "loss": 2.4582815170288086, + "step": 354560 + }, + { + "epoch": 0.9652375119096229, + "grad_norm": 2.1368825435638428, + "learning_rate": 6.962518740629686e-07, + "loss": 2.450556182861328, + "step": 354580 + }, + { + "epoch": 0.9652919559003675, + "grad_norm": 2.2623465061187744, + "learning_rate": 6.951615101540139e-07, + "loss": 2.4482492446899413, + "step": 354600 + }, + { + "epoch": 0.965346399891112, + "grad_norm": 2.2279839515686035, + "learning_rate": 6.940711462450593e-07, + "loss": 2.4437488555908202, + "step": 354620 + }, + { + "epoch": 0.9654008438818565, + "grad_norm": 3.2083494663238525, + "learning_rate": 6.929807823361047e-07, + "loss": 2.47625617980957, + "step": 354640 + }, + { + "epoch": 0.965455287872601, + "grad_norm": 2.185157060623169, + "learning_rate": 6.918904184271501e-07, + "loss": 2.430605888366699, + "step": 354660 + }, + { + "epoch": 0.9655097318633455, + "grad_norm": 2.1106128692626953, + "learning_rate": 6.908000545181955e-07, + "loss": 2.4579116821289064, + "step": 354680 + }, + { + "epoch": 0.9655641758540902, + "grad_norm": 2.177098035812378, + "learning_rate": 6.897096906092409e-07, + "loss": 2.4263839721679688, + "step": 354700 + }, + { + "epoch": 0.9656186198448347, + "grad_norm": 3.1250569820404053, + "learning_rate": 6.886193267002863e-07, + "loss": 2.4516300201416015, + "step": 354720 + }, + { + "epoch": 0.9656730638355792, + "grad_norm": 2.1082265377044678, + "learning_rate": 6.875289627913316e-07, + "loss": 2.462030029296875, + "step": 354740 + }, + { + "epoch": 0.9657275078263237, + "grad_norm": 2.1881730556488037, + "learning_rate": 6.86438598882377e-07, + "loss": 2.4678024291992187, + "step": 354760 + }, + { + "epoch": 0.9657819518170682, + "grad_norm": 2.103811740875244, + "learning_rate": 6.853482349734224e-07, + "loss": 2.4602521896362304, + "step": 354780 + }, + { + "epoch": 0.9658363958078127, + "grad_norm": 3.0588839054107666, + "learning_rate": 6.842578710644678e-07, + "loss": 2.4475637435913087, + "step": 354800 + }, + { + "epoch": 0.9658908397985573, + "grad_norm": 2.423189401626587, + "learning_rate": 6.831675071555133e-07, + "loss": 2.472620964050293, + "step": 354820 + }, + { + "epoch": 0.9659452837893018, + "grad_norm": 2.244135856628418, + "learning_rate": 6.820771432465587e-07, + "loss": 2.4543596267700196, + "step": 354840 + }, + { + "epoch": 0.9659997277800463, + "grad_norm": 2.2889292240142822, + "learning_rate": 6.809867793376041e-07, + "loss": 2.47003059387207, + "step": 354860 + }, + { + "epoch": 0.9660541717707908, + "grad_norm": 2.1377761363983154, + "learning_rate": 6.798964154286495e-07, + "loss": 2.482548141479492, + "step": 354880 + }, + { + "epoch": 0.9661086157615353, + "grad_norm": 2.1677558422088623, + "learning_rate": 6.788060515196948e-07, + "loss": 2.4512706756591798, + "step": 354900 + }, + { + "epoch": 0.9661630597522799, + "grad_norm": 2.1754379272460938, + "learning_rate": 6.777156876107402e-07, + "loss": 2.465412902832031, + "step": 354920 + }, + { + "epoch": 0.9662175037430244, + "grad_norm": 3.6920583248138428, + "learning_rate": 6.766253237017856e-07, + "loss": 2.4666187286376955, + "step": 354940 + }, + { + "epoch": 0.9662719477337689, + "grad_norm": 2.0789384841918945, + "learning_rate": 6.75534959792831e-07, + "loss": 2.4458742141723633, + "step": 354960 + }, + { + "epoch": 0.9663263917245134, + "grad_norm": 2.215470552444458, + "learning_rate": 6.744445958838764e-07, + "loss": 2.4857803344726563, + "step": 354980 + }, + { + "epoch": 0.9663808357152579, + "grad_norm": 2.257788896560669, + "learning_rate": 6.733542319749218e-07, + "loss": 2.4741865158081056, + "step": 355000 + }, + { + "epoch": 0.9664352797060024, + "grad_norm": 2.2058963775634766, + "learning_rate": 6.722638680659672e-07, + "loss": 2.4388832092285155, + "step": 355020 + }, + { + "epoch": 0.966489723696747, + "grad_norm": 2.2132089138031006, + "learning_rate": 6.711735041570125e-07, + "loss": 2.4315158843994142, + "step": 355040 + }, + { + "epoch": 0.9665441676874915, + "grad_norm": 2.0533480644226074, + "learning_rate": 6.700831402480579e-07, + "loss": 2.462642860412598, + "step": 355060 + }, + { + "epoch": 0.966598611678236, + "grad_norm": 2.0493929386138916, + "learning_rate": 6.689927763391032e-07, + "loss": 2.4278356552124025, + "step": 355080 + }, + { + "epoch": 0.9666530556689805, + "grad_norm": 2.133028030395508, + "learning_rate": 6.679024124301486e-07, + "loss": 2.4538110733032226, + "step": 355100 + }, + { + "epoch": 0.966707499659725, + "grad_norm": 2.1965548992156982, + "learning_rate": 6.66812048521194e-07, + "loss": 2.448040008544922, + "step": 355120 + }, + { + "epoch": 0.9667619436504695, + "grad_norm": 2.231102228164673, + "learning_rate": 6.657216846122394e-07, + "loss": 2.4852338790893556, + "step": 355140 + }, + { + "epoch": 0.9668163876412141, + "grad_norm": 2.399357318878174, + "learning_rate": 6.646313207032847e-07, + "loss": 2.4414005279541016, + "step": 355160 + }, + { + "epoch": 0.9668708316319586, + "grad_norm": 2.077402353286743, + "learning_rate": 6.635409567943301e-07, + "loss": 2.440708541870117, + "step": 355180 + }, + { + "epoch": 0.9669252756227031, + "grad_norm": 2.2121729850769043, + "learning_rate": 6.624505928853755e-07, + "loss": 2.4457929611206053, + "step": 355200 + }, + { + "epoch": 0.9669797196134476, + "grad_norm": 2.2691009044647217, + "learning_rate": 6.613602289764209e-07, + "loss": 2.4497329711914064, + "step": 355220 + }, + { + "epoch": 0.9670341636041921, + "grad_norm": 2.171236276626587, + "learning_rate": 6.602698650674663e-07, + "loss": 2.468514251708984, + "step": 355240 + }, + { + "epoch": 0.9670886075949368, + "grad_norm": 2.2697832584381104, + "learning_rate": 6.591795011585117e-07, + "loss": 2.46898193359375, + "step": 355260 + }, + { + "epoch": 0.9671430515856813, + "grad_norm": 2.2986907958984375, + "learning_rate": 6.580891372495571e-07, + "loss": 2.4710081100463865, + "step": 355280 + }, + { + "epoch": 0.9671974955764258, + "grad_norm": 2.1459946632385254, + "learning_rate": 6.569987733406024e-07, + "loss": 2.4595577239990236, + "step": 355300 + }, + { + "epoch": 0.9672519395671703, + "grad_norm": 2.1833105087280273, + "learning_rate": 6.559084094316478e-07, + "loss": 2.4495607376098634, + "step": 355320 + }, + { + "epoch": 0.9673063835579148, + "grad_norm": 2.233766555786133, + "learning_rate": 6.548180455226932e-07, + "loss": 2.4770845413208007, + "step": 355340 + }, + { + "epoch": 0.9673608275486593, + "grad_norm": 2.126459836959839, + "learning_rate": 6.537276816137386e-07, + "loss": 2.4705514907836914, + "step": 355360 + }, + { + "epoch": 0.9674152715394039, + "grad_norm": 2.075371265411377, + "learning_rate": 6.52637317704784e-07, + "loss": 2.443764877319336, + "step": 355380 + }, + { + "epoch": 0.9674697155301484, + "grad_norm": 2.2895195484161377, + "learning_rate": 6.515469537958294e-07, + "loss": 2.459269332885742, + "step": 355400 + }, + { + "epoch": 0.9675241595208929, + "grad_norm": 2.200188159942627, + "learning_rate": 6.504565898868748e-07, + "loss": 2.4322673797607424, + "step": 355420 + }, + { + "epoch": 0.9675786035116374, + "grad_norm": 2.226827383041382, + "learning_rate": 6.493662259779201e-07, + "loss": 2.456978416442871, + "step": 355440 + }, + { + "epoch": 0.9676330475023819, + "grad_norm": 2.0550661087036133, + "learning_rate": 6.482758620689655e-07, + "loss": 2.4513153076171874, + "step": 355460 + }, + { + "epoch": 0.9676874914931265, + "grad_norm": 2.2662875652313232, + "learning_rate": 6.471854981600109e-07, + "loss": 2.473147964477539, + "step": 355480 + }, + { + "epoch": 0.967741935483871, + "grad_norm": 3.7084054946899414, + "learning_rate": 6.460951342510563e-07, + "loss": 2.471365547180176, + "step": 355500 + }, + { + "epoch": 0.9677963794746155, + "grad_norm": 3.64436936378479, + "learning_rate": 6.450047703421017e-07, + "loss": 2.436781883239746, + "step": 355520 + }, + { + "epoch": 0.96785082346536, + "grad_norm": 2.18603253364563, + "learning_rate": 6.439144064331471e-07, + "loss": 2.443876838684082, + "step": 355540 + }, + { + "epoch": 0.9679052674561045, + "grad_norm": 2.0344130992889404, + "learning_rate": 6.428240425241925e-07, + "loss": 2.4541143417358398, + "step": 355560 + }, + { + "epoch": 0.967959711446849, + "grad_norm": 2.129730463027954, + "learning_rate": 6.417336786152378e-07, + "loss": 2.4714681625366213, + "step": 355580 + }, + { + "epoch": 0.9680141554375936, + "grad_norm": 2.2492361068725586, + "learning_rate": 6.406433147062832e-07, + "loss": 2.455005073547363, + "step": 355600 + }, + { + "epoch": 0.9680685994283381, + "grad_norm": 2.2053751945495605, + "learning_rate": 6.395529507973286e-07, + "loss": 2.4613534927368166, + "step": 355620 + }, + { + "epoch": 0.9681230434190826, + "grad_norm": 2.1858623027801514, + "learning_rate": 6.38462586888374e-07, + "loss": 2.438990592956543, + "step": 355640 + }, + { + "epoch": 0.9681774874098271, + "grad_norm": 2.2448339462280273, + "learning_rate": 6.373722229794195e-07, + "loss": 2.4672496795654295, + "step": 355660 + }, + { + "epoch": 0.9682319314005716, + "grad_norm": 2.1669881343841553, + "learning_rate": 6.362818590704649e-07, + "loss": 2.4233028411865236, + "step": 355680 + }, + { + "epoch": 0.9682863753913162, + "grad_norm": 2.1316723823547363, + "learning_rate": 6.351914951615103e-07, + "loss": 2.4432256698608397, + "step": 355700 + }, + { + "epoch": 0.9683408193820607, + "grad_norm": 2.281198501586914, + "learning_rate": 6.341011312525557e-07, + "loss": 2.4616350173950194, + "step": 355720 + }, + { + "epoch": 0.9683952633728052, + "grad_norm": 2.182692527770996, + "learning_rate": 6.33010767343601e-07, + "loss": 2.460408020019531, + "step": 355740 + }, + { + "epoch": 0.9684497073635497, + "grad_norm": 2.2684566974639893, + "learning_rate": 6.319204034346464e-07, + "loss": 2.472786521911621, + "step": 355760 + }, + { + "epoch": 0.9685041513542942, + "grad_norm": 2.163668394088745, + "learning_rate": 6.308300395256918e-07, + "loss": 2.4471473693847656, + "step": 355780 + }, + { + "epoch": 0.9685585953450387, + "grad_norm": 2.0751216411590576, + "learning_rate": 6.297396756167372e-07, + "loss": 2.437038803100586, + "step": 355800 + }, + { + "epoch": 0.9686130393357834, + "grad_norm": 2.1758365631103516, + "learning_rate": 6.286493117077826e-07, + "loss": 2.4806751251220702, + "step": 355820 + }, + { + "epoch": 0.9686674833265279, + "grad_norm": 2.319887399673462, + "learning_rate": 6.27558947798828e-07, + "loss": 2.4672327041625977, + "step": 355840 + }, + { + "epoch": 0.9687219273172724, + "grad_norm": 2.043480634689331, + "learning_rate": 6.264685838898734e-07, + "loss": 2.5067678451538087, + "step": 355860 + }, + { + "epoch": 0.9687763713080169, + "grad_norm": 2.1499578952789307, + "learning_rate": 6.253782199809187e-07, + "loss": 2.467677879333496, + "step": 355880 + }, + { + "epoch": 0.9688308152987614, + "grad_norm": 2.2252426147460938, + "learning_rate": 6.242878560719641e-07, + "loss": 2.4290950775146483, + "step": 355900 + }, + { + "epoch": 0.968885259289506, + "grad_norm": 2.1736433506011963, + "learning_rate": 6.231974921630095e-07, + "loss": 2.4955820083618163, + "step": 355920 + }, + { + "epoch": 0.9689397032802505, + "grad_norm": 2.1305975914001465, + "learning_rate": 6.221071282540549e-07, + "loss": 2.454598236083984, + "step": 355940 + }, + { + "epoch": 0.968994147270995, + "grad_norm": 2.2314162254333496, + "learning_rate": 6.210167643451003e-07, + "loss": 2.4418764114379883, + "step": 355960 + }, + { + "epoch": 0.9690485912617395, + "grad_norm": 2.2170839309692383, + "learning_rate": 6.199264004361457e-07, + "loss": 2.462321090698242, + "step": 355980 + }, + { + "epoch": 0.969103035252484, + "grad_norm": 2.1673240661621094, + "learning_rate": 6.18836036527191e-07, + "loss": 2.4584930419921873, + "step": 356000 + }, + { + "epoch": 0.9691574792432285, + "grad_norm": 2.2498505115509033, + "learning_rate": 6.177456726182364e-07, + "loss": 2.432104301452637, + "step": 356020 + }, + { + "epoch": 0.9692119232339731, + "grad_norm": 2.115211009979248, + "learning_rate": 6.166553087092818e-07, + "loss": 2.4783308029174806, + "step": 356040 + }, + { + "epoch": 0.9692663672247176, + "grad_norm": 2.291454315185547, + "learning_rate": 6.155649448003272e-07, + "loss": 2.4678586959838866, + "step": 356060 + }, + { + "epoch": 0.9693208112154621, + "grad_norm": 2.138159990310669, + "learning_rate": 6.144745808913726e-07, + "loss": 2.453738784790039, + "step": 356080 + }, + { + "epoch": 0.9693752552062066, + "grad_norm": 3.1030848026275635, + "learning_rate": 6.13384216982418e-07, + "loss": 2.457876777648926, + "step": 356100 + }, + { + "epoch": 0.9694296991969511, + "grad_norm": 2.240900993347168, + "learning_rate": 6.122938530734633e-07, + "loss": 2.4421009063720702, + "step": 356120 + }, + { + "epoch": 0.9694841431876957, + "grad_norm": 2.315671682357788, + "learning_rate": 6.112034891645086e-07, + "loss": 2.4299182891845703, + "step": 356140 + }, + { + "epoch": 0.9695385871784402, + "grad_norm": 2.1435673236846924, + "learning_rate": 6.10113125255554e-07, + "loss": 2.4917892456054687, + "step": 356160 + }, + { + "epoch": 0.9695930311691847, + "grad_norm": 2.2232253551483154, + "learning_rate": 6.090227613465994e-07, + "loss": 2.461570167541504, + "step": 356180 + }, + { + "epoch": 0.9696474751599292, + "grad_norm": 2.279165506362915, + "learning_rate": 6.079323974376448e-07, + "loss": 2.4357393264770506, + "step": 356200 + }, + { + "epoch": 0.9697019191506737, + "grad_norm": 2.1355388164520264, + "learning_rate": 6.068420335286902e-07, + "loss": 2.4666444778442385, + "step": 356220 + }, + { + "epoch": 0.9697563631414182, + "grad_norm": 2.1918466091156006, + "learning_rate": 6.057516696197356e-07, + "loss": 2.458469772338867, + "step": 356240 + }, + { + "epoch": 0.9698108071321628, + "grad_norm": 2.205862522125244, + "learning_rate": 6.04661305710781e-07, + "loss": 2.461903953552246, + "step": 356260 + }, + { + "epoch": 0.9698652511229073, + "grad_norm": 2.141672372817993, + "learning_rate": 6.035709418018263e-07, + "loss": 2.4279296875, + "step": 356280 + }, + { + "epoch": 0.9699196951136518, + "grad_norm": 2.549272298812866, + "learning_rate": 6.024805778928718e-07, + "loss": 2.4633001327514648, + "step": 356300 + }, + { + "epoch": 0.9699741391043963, + "grad_norm": 2.1843152046203613, + "learning_rate": 6.013902139839172e-07, + "loss": 2.45450382232666, + "step": 356320 + }, + { + "epoch": 0.9700285830951408, + "grad_norm": 2.1639788150787354, + "learning_rate": 6.002998500749626e-07, + "loss": 2.447886848449707, + "step": 356340 + }, + { + "epoch": 0.9700830270858853, + "grad_norm": 2.2002601623535156, + "learning_rate": 5.99209486166008e-07, + "loss": 2.460004425048828, + "step": 356360 + }, + { + "epoch": 0.97013747107663, + "grad_norm": 2.340700149536133, + "learning_rate": 5.981191222570534e-07, + "loss": 2.441849136352539, + "step": 356380 + }, + { + "epoch": 0.9701919150673745, + "grad_norm": 2.187084913253784, + "learning_rate": 5.970287583480988e-07, + "loss": 2.449142837524414, + "step": 356400 + }, + { + "epoch": 0.970246359058119, + "grad_norm": 2.1820785999298096, + "learning_rate": 5.959383944391442e-07, + "loss": 2.505253028869629, + "step": 356420 + }, + { + "epoch": 0.9703008030488635, + "grad_norm": 2.1730904579162598, + "learning_rate": 5.948480305301895e-07, + "loss": 2.466230773925781, + "step": 356440 + }, + { + "epoch": 0.970355247039608, + "grad_norm": 2.3417327404022217, + "learning_rate": 5.937576666212349e-07, + "loss": 2.468271827697754, + "step": 356460 + }, + { + "epoch": 0.9704096910303526, + "grad_norm": 2.130441904067993, + "learning_rate": 5.926673027122803e-07, + "loss": 2.4451242446899415, + "step": 356480 + }, + { + "epoch": 0.9704641350210971, + "grad_norm": 2.290736675262451, + "learning_rate": 5.915769388033257e-07, + "loss": 2.4661401748657226, + "step": 356500 + }, + { + "epoch": 0.9705185790118416, + "grad_norm": 2.056166172027588, + "learning_rate": 5.904865748943711e-07, + "loss": 2.4570247650146486, + "step": 356520 + }, + { + "epoch": 0.9705730230025861, + "grad_norm": 2.2695395946502686, + "learning_rate": 5.893962109854165e-07, + "loss": 2.453468132019043, + "step": 356540 + }, + { + "epoch": 0.9706274669933306, + "grad_norm": 2.29316782951355, + "learning_rate": 5.883058470764619e-07, + "loss": 2.44427433013916, + "step": 356560 + }, + { + "epoch": 0.9706819109840751, + "grad_norm": 2.134075403213501, + "learning_rate": 5.872154831675072e-07, + "loss": 2.4574838638305665, + "step": 356580 + }, + { + "epoch": 0.9707363549748197, + "grad_norm": 2.1597251892089844, + "learning_rate": 5.861251192585526e-07, + "loss": 2.490538787841797, + "step": 356600 + }, + { + "epoch": 0.9707907989655642, + "grad_norm": 2.1170785427093506, + "learning_rate": 5.85034755349598e-07, + "loss": 2.462362289428711, + "step": 356620 + }, + { + "epoch": 0.9708452429563087, + "grad_norm": 2.464540481567383, + "learning_rate": 5.839443914406433e-07, + "loss": 2.4730146408081053, + "step": 356640 + }, + { + "epoch": 0.9708996869470532, + "grad_norm": 2.1522557735443115, + "learning_rate": 5.828540275316887e-07, + "loss": 2.4417613983154296, + "step": 356660 + }, + { + "epoch": 0.9709541309377977, + "grad_norm": 2.3833107948303223, + "learning_rate": 5.817636636227341e-07, + "loss": 2.448527145385742, + "step": 356680 + }, + { + "epoch": 0.9710085749285423, + "grad_norm": 2.110703229904175, + "learning_rate": 5.806732997137794e-07, + "loss": 2.4712596893310548, + "step": 356700 + }, + { + "epoch": 0.9710630189192868, + "grad_norm": 2.108290910720825, + "learning_rate": 5.795829358048249e-07, + "loss": 2.477106475830078, + "step": 356720 + }, + { + "epoch": 0.9711174629100313, + "grad_norm": 2.1146483421325684, + "learning_rate": 5.784925718958703e-07, + "loss": 2.447034454345703, + "step": 356740 + }, + { + "epoch": 0.9711719069007758, + "grad_norm": 2.2952182292938232, + "learning_rate": 5.774022079869157e-07, + "loss": 2.4668485641479494, + "step": 356760 + }, + { + "epoch": 0.9712263508915203, + "grad_norm": 2.1516199111938477, + "learning_rate": 5.763118440779611e-07, + "loss": 2.4507495880126955, + "step": 356780 + }, + { + "epoch": 0.9712807948822648, + "grad_norm": 2.185699701309204, + "learning_rate": 5.752214801690065e-07, + "loss": 2.4396106719970705, + "step": 356800 + }, + { + "epoch": 0.9713352388730094, + "grad_norm": 2.1417396068573, + "learning_rate": 5.741311162600519e-07, + "loss": 2.4449640274047852, + "step": 356820 + }, + { + "epoch": 0.9713896828637539, + "grad_norm": 2.2143561840057373, + "learning_rate": 5.730407523510973e-07, + "loss": 2.4542171478271486, + "step": 356840 + }, + { + "epoch": 0.9714441268544984, + "grad_norm": 2.28153133392334, + "learning_rate": 5.719503884421426e-07, + "loss": 2.4805288314819336, + "step": 356860 + }, + { + "epoch": 0.9714985708452429, + "grad_norm": 2.1301302909851074, + "learning_rate": 5.70860024533188e-07, + "loss": 2.4729598999023437, + "step": 356880 + }, + { + "epoch": 0.9715530148359874, + "grad_norm": 2.5408318042755127, + "learning_rate": 5.697696606242334e-07, + "loss": 2.4528526306152343, + "step": 356900 + }, + { + "epoch": 0.971607458826732, + "grad_norm": 2.1063148975372314, + "learning_rate": 5.686792967152788e-07, + "loss": 2.4494348526000977, + "step": 356920 + }, + { + "epoch": 0.9716619028174766, + "grad_norm": 2.152219533920288, + "learning_rate": 5.675889328063242e-07, + "loss": 2.4598438262939455, + "step": 356940 + }, + { + "epoch": 0.971716346808221, + "grad_norm": 2.0778489112854004, + "learning_rate": 5.664985688973696e-07, + "loss": 2.4700578689575194, + "step": 356960 + }, + { + "epoch": 0.9717707907989656, + "grad_norm": 2.439974069595337, + "learning_rate": 5.65408204988415e-07, + "loss": 2.4737564086914063, + "step": 356980 + }, + { + "epoch": 0.9718252347897101, + "grad_norm": 2.242602825164795, + "learning_rate": 5.643178410794603e-07, + "loss": 2.4497562408447267, + "step": 357000 + }, + { + "epoch": 0.9718796787804546, + "grad_norm": 2.1270222663879395, + "learning_rate": 5.632274771705057e-07, + "loss": 2.4631858825683595, + "step": 357020 + }, + { + "epoch": 0.9719341227711992, + "grad_norm": 2.198601484298706, + "learning_rate": 5.621371132615511e-07, + "loss": 2.4698442459106444, + "step": 357040 + }, + { + "epoch": 0.9719885667619437, + "grad_norm": 2.1722872257232666, + "learning_rate": 5.610467493525965e-07, + "loss": 2.4532970428466796, + "step": 357060 + }, + { + "epoch": 0.9720430107526882, + "grad_norm": 2.180332660675049, + "learning_rate": 5.599563854436419e-07, + "loss": 2.463650703430176, + "step": 357080 + }, + { + "epoch": 0.9720974547434327, + "grad_norm": 2.1554293632507324, + "learning_rate": 5.588660215346873e-07, + "loss": 2.4254829406738283, + "step": 357100 + }, + { + "epoch": 0.9721518987341772, + "grad_norm": 2.3567147254943848, + "learning_rate": 5.577756576257327e-07, + "loss": 2.436672592163086, + "step": 357120 + }, + { + "epoch": 0.9722063427249218, + "grad_norm": 2.1819753646850586, + "learning_rate": 5.56685293716778e-07, + "loss": 2.4493207931518555, + "step": 357140 + }, + { + "epoch": 0.9722607867156663, + "grad_norm": 2.1864402294158936, + "learning_rate": 5.555949298078234e-07, + "loss": 2.47387809753418, + "step": 357160 + }, + { + "epoch": 0.9723152307064108, + "grad_norm": 2.256673812866211, + "learning_rate": 5.545045658988688e-07, + "loss": 2.4744983673095704, + "step": 357180 + }, + { + "epoch": 0.9723696746971553, + "grad_norm": 2.5285634994506836, + "learning_rate": 5.534142019899142e-07, + "loss": 2.4639459609985352, + "step": 357200 + }, + { + "epoch": 0.9724241186878998, + "grad_norm": 2.1975111961364746, + "learning_rate": 5.523238380809596e-07, + "loss": 2.458737373352051, + "step": 357220 + }, + { + "epoch": 0.9724785626786443, + "grad_norm": 2.3361613750457764, + "learning_rate": 5.51233474172005e-07, + "loss": 2.4598020553588866, + "step": 357240 + }, + { + "epoch": 0.9725330066693889, + "grad_norm": 2.2669317722320557, + "learning_rate": 5.501431102630504e-07, + "loss": 2.429587554931641, + "step": 357260 + }, + { + "epoch": 0.9725874506601334, + "grad_norm": 2.2108235359191895, + "learning_rate": 5.490527463540957e-07, + "loss": 2.448151206970215, + "step": 357280 + }, + { + "epoch": 0.9726418946508779, + "grad_norm": 2.1740987300872803, + "learning_rate": 5.479623824451411e-07, + "loss": 2.480845260620117, + "step": 357300 + }, + { + "epoch": 0.9726963386416224, + "grad_norm": 2.1203806400299072, + "learning_rate": 5.468720185361865e-07, + "loss": 2.4376155853271486, + "step": 357320 + }, + { + "epoch": 0.9727507826323669, + "grad_norm": 2.217273473739624, + "learning_rate": 5.457816546272319e-07, + "loss": 2.462362289428711, + "step": 357340 + }, + { + "epoch": 0.9728052266231114, + "grad_norm": 2.1671149730682373, + "learning_rate": 5.446912907182773e-07, + "loss": 2.4337329864501953, + "step": 357360 + }, + { + "epoch": 0.972859670613856, + "grad_norm": 2.5431411266326904, + "learning_rate": 5.436009268093227e-07, + "loss": 2.4415151596069338, + "step": 357380 + }, + { + "epoch": 0.9729141146046005, + "grad_norm": 2.185333251953125, + "learning_rate": 5.42510562900368e-07, + "loss": 2.4461032867431642, + "step": 357400 + }, + { + "epoch": 0.972968558595345, + "grad_norm": 2.246903657913208, + "learning_rate": 5.414201989914134e-07, + "loss": 2.4843820571899413, + "step": 357420 + }, + { + "epoch": 0.9730230025860895, + "grad_norm": 2.1278090476989746, + "learning_rate": 5.403298350824588e-07, + "loss": 2.4687280654907227, + "step": 357440 + }, + { + "epoch": 0.973077446576834, + "grad_norm": 2.1445422172546387, + "learning_rate": 5.392394711735042e-07, + "loss": 2.459486389160156, + "step": 357460 + }, + { + "epoch": 0.9731318905675787, + "grad_norm": 2.1546335220336914, + "learning_rate": 5.381491072645496e-07, + "loss": 2.4583599090576174, + "step": 357480 + }, + { + "epoch": 0.9731863345583232, + "grad_norm": 2.1592321395874023, + "learning_rate": 5.37058743355595e-07, + "loss": 2.457550048828125, + "step": 357500 + }, + { + "epoch": 0.9732407785490677, + "grad_norm": 2.2074122428894043, + "learning_rate": 5.359683794466404e-07, + "loss": 2.4641082763671873, + "step": 357520 + }, + { + "epoch": 0.9732952225398122, + "grad_norm": 3.183244228363037, + "learning_rate": 5.348780155376858e-07, + "loss": 2.4507755279541015, + "step": 357540 + }, + { + "epoch": 0.9733496665305567, + "grad_norm": 2.1412460803985596, + "learning_rate": 5.337876516287311e-07, + "loss": 2.4558507919311525, + "step": 357560 + }, + { + "epoch": 0.9734041105213012, + "grad_norm": 2.0986242294311523, + "learning_rate": 5.326972877197765e-07, + "loss": 2.443571853637695, + "step": 357580 + }, + { + "epoch": 0.9734585545120458, + "grad_norm": 4.201929569244385, + "learning_rate": 5.316069238108219e-07, + "loss": 2.482063102722168, + "step": 357600 + }, + { + "epoch": 0.9735129985027903, + "grad_norm": 2.1630947589874268, + "learning_rate": 5.305165599018673e-07, + "loss": 2.4688005447387695, + "step": 357620 + }, + { + "epoch": 0.9735674424935348, + "grad_norm": 2.267396926879883, + "learning_rate": 5.294261959929127e-07, + "loss": 2.4790279388427736, + "step": 357640 + }, + { + "epoch": 0.9736218864842793, + "grad_norm": 2.1449787616729736, + "learning_rate": 5.283358320839581e-07, + "loss": 2.503507614135742, + "step": 357660 + }, + { + "epoch": 0.9736763304750238, + "grad_norm": 2.168884515762329, + "learning_rate": 5.272454681750034e-07, + "loss": 2.450724220275879, + "step": 357680 + }, + { + "epoch": 0.9737307744657684, + "grad_norm": 2.1433613300323486, + "learning_rate": 5.261551042660488e-07, + "loss": 2.4514881134033204, + "step": 357700 + }, + { + "epoch": 0.9737852184565129, + "grad_norm": 2.249432325363159, + "learning_rate": 5.250647403570942e-07, + "loss": 2.463713264465332, + "step": 357720 + }, + { + "epoch": 0.9738396624472574, + "grad_norm": 2.4081313610076904, + "learning_rate": 5.239743764481396e-07, + "loss": 2.428228187561035, + "step": 357740 + }, + { + "epoch": 0.9738941064380019, + "grad_norm": 2.2130777835845947, + "learning_rate": 5.22884012539185e-07, + "loss": 2.451729965209961, + "step": 357760 + }, + { + "epoch": 0.9739485504287464, + "grad_norm": 2.0816729068756104, + "learning_rate": 5.217936486302304e-07, + "loss": 2.441775894165039, + "step": 357780 + }, + { + "epoch": 0.9740029944194909, + "grad_norm": 2.0976688861846924, + "learning_rate": 5.207032847212758e-07, + "loss": 2.473161506652832, + "step": 357800 + }, + { + "epoch": 0.9740574384102355, + "grad_norm": 3.170086622238159, + "learning_rate": 5.196129208123211e-07, + "loss": 2.4613388061523436, + "step": 357820 + }, + { + "epoch": 0.97411188240098, + "grad_norm": 2.1362955570220947, + "learning_rate": 5.185225569033665e-07, + "loss": 2.448390007019043, + "step": 357840 + }, + { + "epoch": 0.9741663263917245, + "grad_norm": 2.103715419769287, + "learning_rate": 5.174321929944119e-07, + "loss": 2.440205192565918, + "step": 357860 + }, + { + "epoch": 0.974220770382469, + "grad_norm": 2.204022169113159, + "learning_rate": 5.163418290854573e-07, + "loss": 2.4625080108642576, + "step": 357880 + }, + { + "epoch": 0.9742752143732135, + "grad_norm": 2.3117198944091797, + "learning_rate": 5.152514651765027e-07, + "loss": 2.4408123016357424, + "step": 357900 + }, + { + "epoch": 0.9743296583639581, + "grad_norm": 2.247004508972168, + "learning_rate": 5.141611012675481e-07, + "loss": 2.458891677856445, + "step": 357920 + }, + { + "epoch": 0.9743841023547026, + "grad_norm": 2.1544477939605713, + "learning_rate": 5.130707373585935e-07, + "loss": 2.451349067687988, + "step": 357940 + }, + { + "epoch": 0.9744385463454471, + "grad_norm": 2.2141802310943604, + "learning_rate": 5.119803734496388e-07, + "loss": 2.453470993041992, + "step": 357960 + }, + { + "epoch": 0.9744929903361916, + "grad_norm": 2.2887821197509766, + "learning_rate": 5.108900095406842e-07, + "loss": 2.450872611999512, + "step": 357980 + }, + { + "epoch": 0.9745474343269361, + "grad_norm": 2.1586248874664307, + "learning_rate": 5.097996456317296e-07, + "loss": 2.4421743392944335, + "step": 358000 + }, + { + "epoch": 0.9746018783176806, + "grad_norm": 2.078795909881592, + "learning_rate": 5.08709281722775e-07, + "loss": 2.449357604980469, + "step": 358020 + }, + { + "epoch": 0.9746563223084252, + "grad_norm": 2.2410686016082764, + "learning_rate": 5.076189178138204e-07, + "loss": 2.4678623199462892, + "step": 358040 + }, + { + "epoch": 0.9747107662991698, + "grad_norm": 2.1001665592193604, + "learning_rate": 5.065285539048658e-07, + "loss": 2.448874092102051, + "step": 358060 + }, + { + "epoch": 0.9747652102899143, + "grad_norm": 2.1777467727661133, + "learning_rate": 5.054381899959112e-07, + "loss": 2.429946517944336, + "step": 358080 + }, + { + "epoch": 0.9748196542806588, + "grad_norm": 2.1532554626464844, + "learning_rate": 5.043478260869565e-07, + "loss": 2.4370975494384766, + "step": 358100 + }, + { + "epoch": 0.9748740982714033, + "grad_norm": 2.1681199073791504, + "learning_rate": 5.032574621780019e-07, + "loss": 2.473641586303711, + "step": 358120 + }, + { + "epoch": 0.9749285422621479, + "grad_norm": 2.168701410293579, + "learning_rate": 5.021670982690473e-07, + "loss": 2.4705495834350586, + "step": 358140 + }, + { + "epoch": 0.9749829862528924, + "grad_norm": 2.1716551780700684, + "learning_rate": 5.010767343600927e-07, + "loss": 2.462473487854004, + "step": 358160 + }, + { + "epoch": 0.9750374302436369, + "grad_norm": 2.1217851638793945, + "learning_rate": 4.999863704511381e-07, + "loss": 2.42730712890625, + "step": 358180 + }, + { + "epoch": 0.9750918742343814, + "grad_norm": 3.0729050636291504, + "learning_rate": 4.988960065421835e-07, + "loss": 2.468169403076172, + "step": 358200 + }, + { + "epoch": 0.9751463182251259, + "grad_norm": 2.1727712154388428, + "learning_rate": 4.978056426332289e-07, + "loss": 2.4449750900268556, + "step": 358220 + }, + { + "epoch": 0.9752007622158704, + "grad_norm": 2.402588367462158, + "learning_rate": 4.967152787242742e-07, + "loss": 2.4821496963500977, + "step": 358240 + }, + { + "epoch": 0.975255206206615, + "grad_norm": 2.3174655437469482, + "learning_rate": 4.956249148153196e-07, + "loss": 2.4333290100097655, + "step": 358260 + }, + { + "epoch": 0.9753096501973595, + "grad_norm": 2.2194719314575195, + "learning_rate": 4.94534550906365e-07, + "loss": 2.4629817962646485, + "step": 358280 + }, + { + "epoch": 0.975364094188104, + "grad_norm": 2.0742764472961426, + "learning_rate": 4.934441869974104e-07, + "loss": 2.4592376708984376, + "step": 358300 + }, + { + "epoch": 0.9754185381788485, + "grad_norm": 2.1605000495910645, + "learning_rate": 4.923538230884558e-07, + "loss": 2.4557306289672853, + "step": 358320 + }, + { + "epoch": 0.975472982169593, + "grad_norm": 2.3464415073394775, + "learning_rate": 4.912634591795012e-07, + "loss": 2.447266960144043, + "step": 358340 + }, + { + "epoch": 0.9755274261603376, + "grad_norm": 2.056499481201172, + "learning_rate": 4.901730952705466e-07, + "loss": 2.4173460006713867, + "step": 358360 + }, + { + "epoch": 0.9755818701510821, + "grad_norm": 2.16349458694458, + "learning_rate": 4.89082731361592e-07, + "loss": 2.488057327270508, + "step": 358380 + }, + { + "epoch": 0.9756363141418266, + "grad_norm": 2.5946080684661865, + "learning_rate": 4.879923674526373e-07, + "loss": 2.461350440979004, + "step": 358400 + }, + { + "epoch": 0.9756907581325711, + "grad_norm": 2.2558648586273193, + "learning_rate": 4.869020035436828e-07, + "loss": 2.4833324432373045, + "step": 358420 + }, + { + "epoch": 0.9757452021233156, + "grad_norm": 2.291002035140991, + "learning_rate": 4.858116396347282e-07, + "loss": 2.458721733093262, + "step": 358440 + }, + { + "epoch": 0.9757996461140601, + "grad_norm": 2.737513542175293, + "learning_rate": 4.847212757257735e-07, + "loss": 2.435151481628418, + "step": 358460 + }, + { + "epoch": 0.9758540901048047, + "grad_norm": 2.135676145553589, + "learning_rate": 4.836309118168189e-07, + "loss": 2.4661432266235352, + "step": 358480 + }, + { + "epoch": 0.9759085340955492, + "grad_norm": 2.26497483253479, + "learning_rate": 4.825405479078643e-07, + "loss": 2.4459041595458983, + "step": 358500 + }, + { + "epoch": 0.9759629780862937, + "grad_norm": 53.07352828979492, + "learning_rate": 4.814501839989096e-07, + "loss": 2.4612064361572266, + "step": 358520 + }, + { + "epoch": 0.9760174220770382, + "grad_norm": 2.195158004760742, + "learning_rate": 4.80359820089955e-07, + "loss": 2.4910457611083983, + "step": 358540 + }, + { + "epoch": 0.9760718660677827, + "grad_norm": 2.0785439014434814, + "learning_rate": 4.792694561810004e-07, + "loss": 2.46887264251709, + "step": 358560 + }, + { + "epoch": 0.9761263100585272, + "grad_norm": 2.1175436973571777, + "learning_rate": 4.781790922720458e-07, + "loss": 2.449156379699707, + "step": 358580 + }, + { + "epoch": 0.9761807540492718, + "grad_norm": 2.2298765182495117, + "learning_rate": 4.770887283630912e-07, + "loss": 2.479303741455078, + "step": 358600 + }, + { + "epoch": 0.9762351980400163, + "grad_norm": 3.8485336303710938, + "learning_rate": 4.7599836445413663e-07, + "loss": 2.4645708084106444, + "step": 358620 + }, + { + "epoch": 0.9762896420307609, + "grad_norm": 2.11474871635437, + "learning_rate": 4.74908000545182e-07, + "loss": 2.4551572799682617, + "step": 358640 + }, + { + "epoch": 0.9763440860215054, + "grad_norm": 2.1827778816223145, + "learning_rate": 4.7381763663622735e-07, + "loss": 2.4103420257568358, + "step": 358660 + }, + { + "epoch": 0.9763985300122499, + "grad_norm": 2.1685943603515625, + "learning_rate": 4.7272727272727273e-07, + "loss": 2.4362503051757813, + "step": 358680 + }, + { + "epoch": 0.9764529740029945, + "grad_norm": 2.122023105621338, + "learning_rate": 4.716369088183181e-07, + "loss": 2.4647539138793944, + "step": 358700 + }, + { + "epoch": 0.976507417993739, + "grad_norm": 2.2977051734924316, + "learning_rate": 4.705465449093635e-07, + "loss": 2.4725009918212892, + "step": 358720 + }, + { + "epoch": 0.9765618619844835, + "grad_norm": 2.146434783935547, + "learning_rate": 4.694561810004089e-07, + "loss": 2.4666730880737306, + "step": 358740 + }, + { + "epoch": 0.976616305975228, + "grad_norm": 2.1242411136627197, + "learning_rate": 4.683658170914543e-07, + "loss": 2.4467844009399413, + "step": 358760 + }, + { + "epoch": 0.9766707499659725, + "grad_norm": 2.1364171504974365, + "learning_rate": 4.6727545318249966e-07, + "loss": 2.46685733795166, + "step": 358780 + }, + { + "epoch": 0.976725193956717, + "grad_norm": 2.5764644145965576, + "learning_rate": 4.6618508927354505e-07, + "loss": 2.4572851181030275, + "step": 358800 + }, + { + "epoch": 0.9767796379474616, + "grad_norm": 2.306605577468872, + "learning_rate": 4.6509472536459043e-07, + "loss": 2.464299201965332, + "step": 358820 + }, + { + "epoch": 0.9768340819382061, + "grad_norm": 2.158967971801758, + "learning_rate": 4.6400436145563587e-07, + "loss": 2.432646942138672, + "step": 358840 + }, + { + "epoch": 0.9768885259289506, + "grad_norm": 2.1629104614257812, + "learning_rate": 4.6291399754668126e-07, + "loss": 2.4761966705322265, + "step": 358860 + }, + { + "epoch": 0.9769429699196951, + "grad_norm": 2.188626527786255, + "learning_rate": 4.6182363363772664e-07, + "loss": 2.4671537399291994, + "step": 358880 + }, + { + "epoch": 0.9769974139104396, + "grad_norm": 2.204017162322998, + "learning_rate": 4.6073326972877203e-07, + "loss": 2.429484176635742, + "step": 358900 + }, + { + "epoch": 0.9770518579011842, + "grad_norm": 2.1063272953033447, + "learning_rate": 4.596429058198174e-07, + "loss": 2.4534950256347656, + "step": 358920 + }, + { + "epoch": 0.9771063018919287, + "grad_norm": 2.2024924755096436, + "learning_rate": 4.585525419108628e-07, + "loss": 2.4341098785400392, + "step": 358940 + }, + { + "epoch": 0.9771607458826732, + "grad_norm": 2.2131869792938232, + "learning_rate": 4.574621780019082e-07, + "loss": 2.459109306335449, + "step": 358960 + }, + { + "epoch": 0.9772151898734177, + "grad_norm": 2.163788318634033, + "learning_rate": 4.5637181409295357e-07, + "loss": 2.451397705078125, + "step": 358980 + }, + { + "epoch": 0.9772696338641622, + "grad_norm": 2.178220510482788, + "learning_rate": 4.5528145018399896e-07, + "loss": 2.4281475067138674, + "step": 359000 + }, + { + "epoch": 0.9773240778549067, + "grad_norm": 2.2471466064453125, + "learning_rate": 4.5419108627504434e-07, + "loss": 2.446029472351074, + "step": 359020 + }, + { + "epoch": 0.9773785218456513, + "grad_norm": 2.296626091003418, + "learning_rate": 4.5310072236608973e-07, + "loss": 2.4648571014404297, + "step": 359040 + }, + { + "epoch": 0.9774329658363958, + "grad_norm": 2.191359758377075, + "learning_rate": 4.520103584571351e-07, + "loss": 2.4631978988647463, + "step": 359060 + }, + { + "epoch": 0.9774874098271403, + "grad_norm": 2.1741628646850586, + "learning_rate": 4.509199945481805e-07, + "loss": 2.4452829360961914, + "step": 359080 + }, + { + "epoch": 0.9775418538178848, + "grad_norm": 2.166220188140869, + "learning_rate": 4.498296306392259e-07, + "loss": 2.441759490966797, + "step": 359100 + }, + { + "epoch": 0.9775962978086293, + "grad_norm": 2.277773380279541, + "learning_rate": 4.4873926673027127e-07, + "loss": 2.4668760299682617, + "step": 359120 + }, + { + "epoch": 0.977650741799374, + "grad_norm": 2.116697311401367, + "learning_rate": 4.4764890282131666e-07, + "loss": 2.453720474243164, + "step": 359140 + }, + { + "epoch": 0.9777051857901184, + "grad_norm": 2.10990571975708, + "learning_rate": 4.4655853891236204e-07, + "loss": 2.4539207458496093, + "step": 359160 + }, + { + "epoch": 0.977759629780863, + "grad_norm": 2.1643357276916504, + "learning_rate": 4.454681750034074e-07, + "loss": 2.450424385070801, + "step": 359180 + }, + { + "epoch": 0.9778140737716075, + "grad_norm": 2.308854818344116, + "learning_rate": 4.4437781109445276e-07, + "loss": 2.4782806396484376, + "step": 359200 + }, + { + "epoch": 0.977868517762352, + "grad_norm": 13.869656562805176, + "learning_rate": 4.4328744718549815e-07, + "loss": 2.458262062072754, + "step": 359220 + }, + { + "epoch": 0.9779229617530965, + "grad_norm": 2.1729166507720947, + "learning_rate": 4.4219708327654353e-07, + "loss": 2.4604429244995116, + "step": 359240 + }, + { + "epoch": 0.9779774057438411, + "grad_norm": 4.2177228927612305, + "learning_rate": 4.41106719367589e-07, + "loss": 2.472533416748047, + "step": 359260 + }, + { + "epoch": 0.9780318497345856, + "grad_norm": 2.128896713256836, + "learning_rate": 4.400163554586344e-07, + "loss": 2.5000160217285154, + "step": 359280 + }, + { + "epoch": 0.9780862937253301, + "grad_norm": 2.2095425128936768, + "learning_rate": 4.3892599154967974e-07, + "loss": 2.4558042526245116, + "step": 359300 + }, + { + "epoch": 0.9781407377160746, + "grad_norm": 2.1539924144744873, + "learning_rate": 4.3783562764072513e-07, + "loss": 2.4559669494628906, + "step": 359320 + }, + { + "epoch": 0.9781951817068191, + "grad_norm": 2.1923210620880127, + "learning_rate": 4.367452637317705e-07, + "loss": 2.464494323730469, + "step": 359340 + }, + { + "epoch": 0.9782496256975637, + "grad_norm": 2.114518404006958, + "learning_rate": 4.356548998228159e-07, + "loss": 2.4803014755249024, + "step": 359360 + }, + { + "epoch": 0.9783040696883082, + "grad_norm": 2.1999285221099854, + "learning_rate": 4.345645359138613e-07, + "loss": 2.471239471435547, + "step": 359380 + }, + { + "epoch": 0.9783585136790527, + "grad_norm": 2.1499016284942627, + "learning_rate": 4.3347417200490667e-07, + "loss": 2.442822074890137, + "step": 359400 + }, + { + "epoch": 0.9784129576697972, + "grad_norm": 2.192291736602783, + "learning_rate": 4.3238380809595206e-07, + "loss": 2.4823276519775392, + "step": 359420 + }, + { + "epoch": 0.9784674016605417, + "grad_norm": 2.116377830505371, + "learning_rate": 4.3129344418699744e-07, + "loss": 2.443181037902832, + "step": 359440 + }, + { + "epoch": 0.9785218456512862, + "grad_norm": 2.2063732147216797, + "learning_rate": 4.3020308027804283e-07, + "loss": 2.4683584213256835, + "step": 359460 + }, + { + "epoch": 0.9785762896420308, + "grad_norm": 2.070991039276123, + "learning_rate": 4.291127163690882e-07, + "loss": 2.442973518371582, + "step": 359480 + }, + { + "epoch": 0.9786307336327753, + "grad_norm": 2.1392102241516113, + "learning_rate": 4.280223524601336e-07, + "loss": 2.4554386138916016, + "step": 359500 + }, + { + "epoch": 0.9786851776235198, + "grad_norm": 2.2245359420776367, + "learning_rate": 4.26931988551179e-07, + "loss": 2.4455371856689454, + "step": 359520 + }, + { + "epoch": 0.9787396216142643, + "grad_norm": 2.247687816619873, + "learning_rate": 4.2584162464222437e-07, + "loss": 2.454142761230469, + "step": 359540 + }, + { + "epoch": 0.9787940656050088, + "grad_norm": 2.1004281044006348, + "learning_rate": 4.2475126073326976e-07, + "loss": 2.4463871002197264, + "step": 359560 + }, + { + "epoch": 0.9788485095957533, + "grad_norm": 2.168161153793335, + "learning_rate": 4.2366089682431514e-07, + "loss": 2.458595848083496, + "step": 359580 + }, + { + "epoch": 0.9789029535864979, + "grad_norm": 2.24660325050354, + "learning_rate": 4.2257053291536053e-07, + "loss": 2.4726655960083006, + "step": 359600 + }, + { + "epoch": 0.9789573975772424, + "grad_norm": 2.1467182636260986, + "learning_rate": 4.214801690064059e-07, + "loss": 2.442031669616699, + "step": 359620 + }, + { + "epoch": 0.9790118415679869, + "grad_norm": 2.1545662879943848, + "learning_rate": 4.203898050974513e-07, + "loss": 2.437221717834473, + "step": 359640 + }, + { + "epoch": 0.9790662855587314, + "grad_norm": 2.1790852546691895, + "learning_rate": 4.192994411884967e-07, + "loss": 2.4233119964599608, + "step": 359660 + }, + { + "epoch": 0.9791207295494759, + "grad_norm": 2.1340458393096924, + "learning_rate": 4.1820907727954207e-07, + "loss": 2.461550712585449, + "step": 359680 + }, + { + "epoch": 0.9791751735402205, + "grad_norm": 2.0527961254119873, + "learning_rate": 4.171187133705875e-07, + "loss": 2.4332469940185546, + "step": 359700 + }, + { + "epoch": 0.979229617530965, + "grad_norm": 2.1213810443878174, + "learning_rate": 4.160283494616329e-07, + "loss": 2.420339012145996, + "step": 359720 + }, + { + "epoch": 0.9792840615217095, + "grad_norm": 2.198991060256958, + "learning_rate": 4.149379855526783e-07, + "loss": 2.4704511642456053, + "step": 359740 + }, + { + "epoch": 0.979338505512454, + "grad_norm": 2.1300888061523438, + "learning_rate": 4.1384762164372367e-07, + "loss": 2.441942596435547, + "step": 359760 + }, + { + "epoch": 0.9793929495031986, + "grad_norm": 2.6531031131744385, + "learning_rate": 4.1275725773476905e-07, + "loss": 2.44848747253418, + "step": 359780 + }, + { + "epoch": 0.979447393493943, + "grad_norm": 2.3376712799072266, + "learning_rate": 4.1166689382581444e-07, + "loss": 2.4593585968017577, + "step": 359800 + }, + { + "epoch": 0.9795018374846877, + "grad_norm": 2.0439462661743164, + "learning_rate": 4.1057652991685977e-07, + "loss": 2.4565988540649415, + "step": 359820 + }, + { + "epoch": 0.9795562814754322, + "grad_norm": 2.2605247497558594, + "learning_rate": 4.0948616600790516e-07, + "loss": 2.451874542236328, + "step": 359840 + }, + { + "epoch": 0.9796107254661767, + "grad_norm": 2.1228878498077393, + "learning_rate": 4.0839580209895054e-07, + "loss": 2.4145442962646486, + "step": 359860 + }, + { + "epoch": 0.9796651694569212, + "grad_norm": 2.1607918739318848, + "learning_rate": 4.0730543818999593e-07, + "loss": 2.4531883239746093, + "step": 359880 + }, + { + "epoch": 0.9797196134476657, + "grad_norm": 2.0993494987487793, + "learning_rate": 4.062150742810413e-07, + "loss": 2.4505165100097654, + "step": 359900 + }, + { + "epoch": 0.9797740574384103, + "grad_norm": 2.1879234313964844, + "learning_rate": 4.051247103720867e-07, + "loss": 2.4514432907104493, + "step": 359920 + }, + { + "epoch": 0.9798285014291548, + "grad_norm": 2.130733013153076, + "learning_rate": 4.040343464631321e-07, + "loss": 2.445279121398926, + "step": 359940 + }, + { + "epoch": 0.9798829454198993, + "grad_norm": 2.2462358474731445, + "learning_rate": 4.0294398255417747e-07, + "loss": 2.4517873764038085, + "step": 359960 + }, + { + "epoch": 0.9799373894106438, + "grad_norm": 2.224194049835205, + "learning_rate": 4.0185361864522286e-07, + "loss": 2.418988037109375, + "step": 359980 + }, + { + "epoch": 0.9799918334013883, + "grad_norm": 2.183591842651367, + "learning_rate": 4.0076325473626824e-07, + "loss": 2.46230411529541, + "step": 360000 + }, + { + "epoch": 0.9800462773921328, + "grad_norm": 2.1962344646453857, + "learning_rate": 3.9967289082731363e-07, + "loss": 2.4389415740966798, + "step": 360020 + }, + { + "epoch": 0.9801007213828774, + "grad_norm": 2.09198260307312, + "learning_rate": 3.98582526918359e-07, + "loss": 2.4697813034057616, + "step": 360040 + }, + { + "epoch": 0.9801551653736219, + "grad_norm": 2.1793437004089355, + "learning_rate": 3.974921630094044e-07, + "loss": 2.4696928024291993, + "step": 360060 + }, + { + "epoch": 0.9802096093643664, + "grad_norm": 2.086306095123291, + "learning_rate": 3.964017991004498e-07, + "loss": 2.4366273880004883, + "step": 360080 + }, + { + "epoch": 0.9802640533551109, + "grad_norm": 2.242799758911133, + "learning_rate": 3.9531143519149517e-07, + "loss": 2.4470493316650392, + "step": 360100 + }, + { + "epoch": 0.9803184973458554, + "grad_norm": 2.1397476196289062, + "learning_rate": 3.942210712825406e-07, + "loss": 2.5003917694091795, + "step": 360120 + }, + { + "epoch": 0.9803729413366, + "grad_norm": 2.1810085773468018, + "learning_rate": 3.93130707373586e-07, + "loss": 2.4501310348510743, + "step": 360140 + }, + { + "epoch": 0.9804273853273445, + "grad_norm": 2.2488131523132324, + "learning_rate": 3.920403434646314e-07, + "loss": 2.4846439361572266, + "step": 360160 + }, + { + "epoch": 0.980481829318089, + "grad_norm": 2.194228172302246, + "learning_rate": 3.9094997955567677e-07, + "loss": 2.4778709411621094, + "step": 360180 + }, + { + "epoch": 0.9805362733088335, + "grad_norm": 2.402918815612793, + "learning_rate": 3.8985961564672215e-07, + "loss": 2.4173646926879884, + "step": 360200 + }, + { + "epoch": 0.980590717299578, + "grad_norm": 2.2141168117523193, + "learning_rate": 3.8876925173776754e-07, + "loss": 2.483641815185547, + "step": 360220 + }, + { + "epoch": 0.9806451612903225, + "grad_norm": 2.184047222137451, + "learning_rate": 3.876788878288129e-07, + "loss": 2.4633033752441404, + "step": 360240 + }, + { + "epoch": 0.9806996052810671, + "grad_norm": 2.1033291816711426, + "learning_rate": 3.865885239198583e-07, + "loss": 2.446396064758301, + "step": 360260 + }, + { + "epoch": 0.9807540492718116, + "grad_norm": 2.1800947189331055, + "learning_rate": 3.854981600109037e-07, + "loss": 2.45935001373291, + "step": 360280 + }, + { + "epoch": 0.9808084932625561, + "grad_norm": 2.2013349533081055, + "learning_rate": 3.844077961019491e-07, + "loss": 2.4628280639648437, + "step": 360300 + }, + { + "epoch": 0.9808629372533006, + "grad_norm": 2.1007399559020996, + "learning_rate": 3.8331743219299447e-07, + "loss": 2.4627311706542967, + "step": 360320 + }, + { + "epoch": 0.9809173812440452, + "grad_norm": 2.1575310230255127, + "learning_rate": 3.822270682840398e-07, + "loss": 2.451046371459961, + "step": 360340 + }, + { + "epoch": 0.9809718252347898, + "grad_norm": 2.2835872173309326, + "learning_rate": 3.811367043750852e-07, + "loss": 2.45285701751709, + "step": 360360 + }, + { + "epoch": 0.9810262692255343, + "grad_norm": 2.184190034866333, + "learning_rate": 3.8004634046613057e-07, + "loss": 2.4714405059814455, + "step": 360380 + }, + { + "epoch": 0.9810807132162788, + "grad_norm": 2.2557573318481445, + "learning_rate": 3.7895597655717596e-07, + "loss": 2.4562448501586913, + "step": 360400 + }, + { + "epoch": 0.9811351572070233, + "grad_norm": 2.1765177249908447, + "learning_rate": 3.7786561264822134e-07, + "loss": 2.4812185287475588, + "step": 360420 + }, + { + "epoch": 0.9811896011977678, + "grad_norm": 2.120683431625366, + "learning_rate": 3.7677524873926673e-07, + "loss": 2.4710504531860353, + "step": 360440 + }, + { + "epoch": 0.9812440451885123, + "grad_norm": 2.5224361419677734, + "learning_rate": 3.756848848303121e-07, + "loss": 2.475469779968262, + "step": 360460 + }, + { + "epoch": 0.9812984891792569, + "grad_norm": 2.1395647525787354, + "learning_rate": 3.745945209213575e-07, + "loss": 2.4238344192504884, + "step": 360480 + }, + { + "epoch": 0.9813529331700014, + "grad_norm": 2.334937334060669, + "learning_rate": 3.735041570124029e-07, + "loss": 2.4540992736816407, + "step": 360500 + }, + { + "epoch": 0.9814073771607459, + "grad_norm": 2.4232254028320312, + "learning_rate": 3.7241379310344827e-07, + "loss": 2.4601682662963866, + "step": 360520 + }, + { + "epoch": 0.9814618211514904, + "grad_norm": 2.174778699874878, + "learning_rate": 3.713234291944937e-07, + "loss": 2.44627685546875, + "step": 360540 + }, + { + "epoch": 0.9815162651422349, + "grad_norm": 2.207078218460083, + "learning_rate": 3.702330652855391e-07, + "loss": 2.448503875732422, + "step": 360560 + }, + { + "epoch": 0.9815707091329795, + "grad_norm": 2.0825188159942627, + "learning_rate": 3.691427013765845e-07, + "loss": 2.4378084182739257, + "step": 360580 + }, + { + "epoch": 0.981625153123724, + "grad_norm": 2.095630645751953, + "learning_rate": 3.6805233746762987e-07, + "loss": 2.4525304794311524, + "step": 360600 + }, + { + "epoch": 0.9816795971144685, + "grad_norm": 2.174567222595215, + "learning_rate": 3.6696197355867525e-07, + "loss": 2.4600635528564454, + "step": 360620 + }, + { + "epoch": 0.981734041105213, + "grad_norm": 2.1065518856048584, + "learning_rate": 3.6587160964972064e-07, + "loss": 2.431296157836914, + "step": 360640 + }, + { + "epoch": 0.9817884850959575, + "grad_norm": 2.1467444896698, + "learning_rate": 3.64781245740766e-07, + "loss": 2.4378759384155275, + "step": 360660 + }, + { + "epoch": 0.981842929086702, + "grad_norm": 2.1361844539642334, + "learning_rate": 3.636908818318114e-07, + "loss": 2.4292972564697264, + "step": 360680 + }, + { + "epoch": 0.9818973730774466, + "grad_norm": 2.24653697013855, + "learning_rate": 3.626005179228568e-07, + "loss": 2.4979347229003905, + "step": 360700 + }, + { + "epoch": 0.9819518170681911, + "grad_norm": 2.1428415775299072, + "learning_rate": 3.615101540139022e-07, + "loss": 2.449814796447754, + "step": 360720 + }, + { + "epoch": 0.9820062610589356, + "grad_norm": 2.193626880645752, + "learning_rate": 3.6041979010494757e-07, + "loss": 2.466954803466797, + "step": 360740 + }, + { + "epoch": 0.9820607050496801, + "grad_norm": 2.082791805267334, + "learning_rate": 3.5932942619599295e-07, + "loss": 2.473979187011719, + "step": 360760 + }, + { + "epoch": 0.9821151490404246, + "grad_norm": 2.267942190170288, + "learning_rate": 3.5823906228703834e-07, + "loss": 2.4452560424804686, + "step": 360780 + }, + { + "epoch": 0.9821695930311691, + "grad_norm": 2.216172218322754, + "learning_rate": 3.571486983780837e-07, + "loss": 2.451679992675781, + "step": 360800 + }, + { + "epoch": 0.9822240370219137, + "grad_norm": 2.202177047729492, + "learning_rate": 3.560583344691291e-07, + "loss": 2.454060935974121, + "step": 360820 + }, + { + "epoch": 0.9822784810126582, + "grad_norm": 2.1059627532958984, + "learning_rate": 3.549679705601745e-07, + "loss": 2.4209369659423827, + "step": 360840 + }, + { + "epoch": 0.9823329250034027, + "grad_norm": 2.1168289184570312, + "learning_rate": 3.5387760665121983e-07, + "loss": 2.430647087097168, + "step": 360860 + }, + { + "epoch": 0.9823873689941472, + "grad_norm": 2.161983013153076, + "learning_rate": 3.527872427422652e-07, + "loss": 2.4524627685546876, + "step": 360880 + }, + { + "epoch": 0.9824418129848917, + "grad_norm": 2.107415199279785, + "learning_rate": 3.516968788333106e-07, + "loss": 2.471023750305176, + "step": 360900 + }, + { + "epoch": 0.9824962569756364, + "grad_norm": 2.0823962688446045, + "learning_rate": 3.50606514924356e-07, + "loss": 2.433024597167969, + "step": 360920 + }, + { + "epoch": 0.9825507009663809, + "grad_norm": 2.165266513824463, + "learning_rate": 3.4951615101540137e-07, + "loss": 2.462089729309082, + "step": 360940 + }, + { + "epoch": 0.9826051449571254, + "grad_norm": 2.2003135681152344, + "learning_rate": 3.4842578710644686e-07, + "loss": 2.481658363342285, + "step": 360960 + }, + { + "epoch": 0.9826595889478699, + "grad_norm": 2.1367366313934326, + "learning_rate": 3.473354231974922e-07, + "loss": 2.449643135070801, + "step": 360980 + }, + { + "epoch": 0.9827140329386144, + "grad_norm": 2.223567485809326, + "learning_rate": 3.462450592885376e-07, + "loss": 2.4826974868774414, + "step": 361000 + }, + { + "epoch": 0.9827684769293589, + "grad_norm": 2.2377803325653076, + "learning_rate": 3.4515469537958297e-07, + "loss": 2.441967010498047, + "step": 361020 + }, + { + "epoch": 0.9828229209201035, + "grad_norm": 2.6126720905303955, + "learning_rate": 3.4406433147062835e-07, + "loss": 2.4558107376098635, + "step": 361040 + }, + { + "epoch": 0.982877364910848, + "grad_norm": 2.0896339416503906, + "learning_rate": 3.4297396756167374e-07, + "loss": 2.45910587310791, + "step": 361060 + }, + { + "epoch": 0.9829318089015925, + "grad_norm": 2.181469202041626, + "learning_rate": 3.418836036527191e-07, + "loss": 2.4607025146484376, + "step": 361080 + }, + { + "epoch": 0.982986252892337, + "grad_norm": 2.195068120956421, + "learning_rate": 3.407932397437645e-07, + "loss": 2.4466650009155275, + "step": 361100 + }, + { + "epoch": 0.9830406968830815, + "grad_norm": 2.1025805473327637, + "learning_rate": 3.397028758348099e-07, + "loss": 2.446424674987793, + "step": 361120 + }, + { + "epoch": 0.9830951408738261, + "grad_norm": 2.167210340499878, + "learning_rate": 3.386125119258553e-07, + "loss": 2.44610538482666, + "step": 361140 + }, + { + "epoch": 0.9831495848645706, + "grad_norm": 2.2592132091522217, + "learning_rate": 3.3752214801690067e-07, + "loss": 2.479727363586426, + "step": 361160 + }, + { + "epoch": 0.9832040288553151, + "grad_norm": 2.2407047748565674, + "learning_rate": 3.3643178410794605e-07, + "loss": 2.4444601058959963, + "step": 361180 + }, + { + "epoch": 0.9832584728460596, + "grad_norm": 2.1251957416534424, + "learning_rate": 3.3534142019899144e-07, + "loss": 2.4400829315185546, + "step": 361200 + }, + { + "epoch": 0.9833129168368041, + "grad_norm": 2.1988744735717773, + "learning_rate": 3.342510562900368e-07, + "loss": 2.4556882858276365, + "step": 361220 + }, + { + "epoch": 0.9833673608275486, + "grad_norm": 2.0824806690216064, + "learning_rate": 3.331606923810822e-07, + "loss": 2.4650272369384765, + "step": 361240 + }, + { + "epoch": 0.9834218048182932, + "grad_norm": 2.212235927581787, + "learning_rate": 3.320703284721276e-07, + "loss": 2.4629377365112304, + "step": 361260 + }, + { + "epoch": 0.9834762488090377, + "grad_norm": 2.2599263191223145, + "learning_rate": 3.30979964563173e-07, + "loss": 2.4574365615844727, + "step": 361280 + }, + { + "epoch": 0.9835306927997822, + "grad_norm": 2.0393004417419434, + "learning_rate": 3.2988960065421837e-07, + "loss": 2.4415523529052736, + "step": 361300 + }, + { + "epoch": 0.9835851367905267, + "grad_norm": 2.1830687522888184, + "learning_rate": 3.2879923674526375e-07, + "loss": 2.45001163482666, + "step": 361320 + }, + { + "epoch": 0.9836395807812712, + "grad_norm": 2.139017343521118, + "learning_rate": 3.2770887283630914e-07, + "loss": 2.4352670669555665, + "step": 361340 + }, + { + "epoch": 0.9836940247720158, + "grad_norm": 2.189970016479492, + "learning_rate": 3.266185089273545e-07, + "loss": 2.4352018356323244, + "step": 361360 + }, + { + "epoch": 0.9837484687627603, + "grad_norm": 2.114441394805908, + "learning_rate": 3.2552814501839996e-07, + "loss": 2.4806806564331056, + "step": 361380 + }, + { + "epoch": 0.9838029127535048, + "grad_norm": 2.1327874660491943, + "learning_rate": 3.2443778110944535e-07, + "loss": 2.4773006439208984, + "step": 361400 + }, + { + "epoch": 0.9838573567442493, + "grad_norm": 2.209829807281494, + "learning_rate": 3.2334741720049073e-07, + "loss": 2.487761116027832, + "step": 361420 + }, + { + "epoch": 0.9839118007349938, + "grad_norm": 2.1762938499450684, + "learning_rate": 3.222570532915361e-07, + "loss": 2.42545108795166, + "step": 361440 + }, + { + "epoch": 0.9839662447257383, + "grad_norm": 2.1476731300354004, + "learning_rate": 3.211666893825815e-07, + "loss": 2.4500444412231444, + "step": 361460 + }, + { + "epoch": 0.984020688716483, + "grad_norm": 2.0920557975769043, + "learning_rate": 3.200763254736269e-07, + "loss": 2.4466442108154296, + "step": 361480 + }, + { + "epoch": 0.9840751327072275, + "grad_norm": 2.06368088722229, + "learning_rate": 3.189859615646722e-07, + "loss": 2.4590099334716795, + "step": 361500 + }, + { + "epoch": 0.984129576697972, + "grad_norm": 2.2314493656158447, + "learning_rate": 3.178955976557176e-07, + "loss": 2.4707111358642577, + "step": 361520 + }, + { + "epoch": 0.9841840206887165, + "grad_norm": 2.1209826469421387, + "learning_rate": 3.16805233746763e-07, + "loss": 2.4281116485595704, + "step": 361540 + }, + { + "epoch": 0.984238464679461, + "grad_norm": 2.198383092880249, + "learning_rate": 3.157148698378084e-07, + "loss": 2.481128692626953, + "step": 361560 + }, + { + "epoch": 0.9842929086702056, + "grad_norm": 2.131300687789917, + "learning_rate": 3.1462450592885377e-07, + "loss": 2.4615242004394533, + "step": 361580 + }, + { + "epoch": 0.9843473526609501, + "grad_norm": 2.2052650451660156, + "learning_rate": 3.1353414201989915e-07, + "loss": 2.4379838943481444, + "step": 361600 + }, + { + "epoch": 0.9844017966516946, + "grad_norm": 2.1823506355285645, + "learning_rate": 3.1244377811094454e-07, + "loss": 2.4456146240234373, + "step": 361620 + }, + { + "epoch": 0.9844562406424391, + "grad_norm": 2.1381640434265137, + "learning_rate": 3.113534142019899e-07, + "loss": 2.4736087799072264, + "step": 361640 + }, + { + "epoch": 0.9845106846331836, + "grad_norm": 2.205202579498291, + "learning_rate": 3.102630502930353e-07, + "loss": 2.4635122299194334, + "step": 361660 + }, + { + "epoch": 0.9845651286239281, + "grad_norm": 2.17990779876709, + "learning_rate": 3.091726863840807e-07, + "loss": 2.4574821472167967, + "step": 361680 + }, + { + "epoch": 0.9846195726146727, + "grad_norm": 2.4499075412750244, + "learning_rate": 3.0808232247512613e-07, + "loss": 2.4641960144042967, + "step": 361700 + }, + { + "epoch": 0.9846740166054172, + "grad_norm": 2.154808521270752, + "learning_rate": 3.069919585661715e-07, + "loss": 2.4614587783813477, + "step": 361720 + }, + { + "epoch": 0.9847284605961617, + "grad_norm": 2.549010992050171, + "learning_rate": 3.059015946572169e-07, + "loss": 2.4776548385620116, + "step": 361740 + }, + { + "epoch": 0.9847829045869062, + "grad_norm": 2.1409409046173096, + "learning_rate": 3.0481123074826224e-07, + "loss": 2.462144660949707, + "step": 361760 + }, + { + "epoch": 0.9848373485776507, + "grad_norm": 2.8285555839538574, + "learning_rate": 3.037208668393076e-07, + "loss": 2.4031656265258787, + "step": 361780 + }, + { + "epoch": 0.9848917925683952, + "grad_norm": 2.0988407135009766, + "learning_rate": 3.02630502930353e-07, + "loss": 2.458088493347168, + "step": 361800 + }, + { + "epoch": 0.9849462365591398, + "grad_norm": 2.1343929767608643, + "learning_rate": 3.015401390213984e-07, + "loss": 2.4928258895874023, + "step": 361820 + }, + { + "epoch": 0.9850006805498843, + "grad_norm": 5.957095146179199, + "learning_rate": 3.004497751124438e-07, + "loss": 2.453632354736328, + "step": 361840 + }, + { + "epoch": 0.9850551245406288, + "grad_norm": 2.2684600353240967, + "learning_rate": 2.9935941120348917e-07, + "loss": 2.441920852661133, + "step": 361860 + }, + { + "epoch": 0.9851095685313733, + "grad_norm": 2.142714262008667, + "learning_rate": 2.9826904729453455e-07, + "loss": 2.4393341064453127, + "step": 361880 + }, + { + "epoch": 0.9851640125221178, + "grad_norm": 2.127443552017212, + "learning_rate": 2.9717868338557994e-07, + "loss": 2.477780342102051, + "step": 361900 + }, + { + "epoch": 0.9852184565128624, + "grad_norm": 2.3186042308807373, + "learning_rate": 2.960883194766254e-07, + "loss": 2.425144577026367, + "step": 361920 + }, + { + "epoch": 0.9852729005036069, + "grad_norm": 2.495447874069214, + "learning_rate": 2.9499795556767076e-07, + "loss": 2.453400421142578, + "step": 361940 + }, + { + "epoch": 0.9853273444943514, + "grad_norm": 2.051461935043335, + "learning_rate": 2.9390759165871615e-07, + "loss": 2.447189140319824, + "step": 361960 + }, + { + "epoch": 0.9853817884850959, + "grad_norm": 2.1269068717956543, + "learning_rate": 2.9281722774976153e-07, + "loss": 2.433441925048828, + "step": 361980 + }, + { + "epoch": 0.9854362324758404, + "grad_norm": 2.085367202758789, + "learning_rate": 2.917268638408069e-07, + "loss": 2.4238718032836912, + "step": 362000 + }, + { + "epoch": 0.985490676466585, + "grad_norm": 2.160243272781372, + "learning_rate": 2.9063649993185225e-07, + "loss": 2.4477073669433596, + "step": 362020 + }, + { + "epoch": 0.9855451204573296, + "grad_norm": 2.9799773693084717, + "learning_rate": 2.8954613602289764e-07, + "loss": 2.4813552856445313, + "step": 362040 + }, + { + "epoch": 0.9855995644480741, + "grad_norm": 2.2103636264801025, + "learning_rate": 2.88455772113943e-07, + "loss": 2.468206787109375, + "step": 362060 + }, + { + "epoch": 0.9856540084388186, + "grad_norm": 5.429610729217529, + "learning_rate": 2.873654082049884e-07, + "loss": 2.4128662109375, + "step": 362080 + }, + { + "epoch": 0.9857084524295631, + "grad_norm": 2.1227238178253174, + "learning_rate": 2.862750442960338e-07, + "loss": 2.455354118347168, + "step": 362100 + }, + { + "epoch": 0.9857628964203076, + "grad_norm": 2.261260509490967, + "learning_rate": 2.8518468038707923e-07, + "loss": 2.4415252685546873, + "step": 362120 + }, + { + "epoch": 0.9858173404110522, + "grad_norm": 2.1834099292755127, + "learning_rate": 2.840943164781246e-07, + "loss": 2.444552993774414, + "step": 362140 + }, + { + "epoch": 0.9858717844017967, + "grad_norm": 2.194148302078247, + "learning_rate": 2.8300395256917e-07, + "loss": 2.444590377807617, + "step": 362160 + }, + { + "epoch": 0.9859262283925412, + "grad_norm": 2.466644048690796, + "learning_rate": 2.819135886602154e-07, + "loss": 2.4691444396972657, + "step": 362180 + }, + { + "epoch": 0.9859806723832857, + "grad_norm": 2.2181172370910645, + "learning_rate": 2.808232247512608e-07, + "loss": 2.4607780456542967, + "step": 362200 + }, + { + "epoch": 0.9860351163740302, + "grad_norm": 2.080512523651123, + "learning_rate": 2.7973286084230616e-07, + "loss": 2.4504459381103514, + "step": 362220 + }, + { + "epoch": 0.9860895603647747, + "grad_norm": 2.20058536529541, + "learning_rate": 2.7864249693335155e-07, + "loss": 2.4612972259521486, + "step": 362240 + }, + { + "epoch": 0.9861440043555193, + "grad_norm": 2.111318588256836, + "learning_rate": 2.7755213302439693e-07, + "loss": 2.468363380432129, + "step": 362260 + }, + { + "epoch": 0.9861984483462638, + "grad_norm": 2.2539381980895996, + "learning_rate": 2.764617691154423e-07, + "loss": 2.424253463745117, + "step": 362280 + }, + { + "epoch": 0.9862528923370083, + "grad_norm": 2.2019195556640625, + "learning_rate": 2.7537140520648765e-07, + "loss": 2.4472774505615233, + "step": 362300 + }, + { + "epoch": 0.9863073363277528, + "grad_norm": 2.3747050762176514, + "learning_rate": 2.7428104129753304e-07, + "loss": 2.430218505859375, + "step": 362320 + }, + { + "epoch": 0.9863617803184973, + "grad_norm": 2.239980936050415, + "learning_rate": 2.731906773885785e-07, + "loss": 2.4300865173339843, + "step": 362340 + }, + { + "epoch": 0.9864162243092419, + "grad_norm": 2.152071952819824, + "learning_rate": 2.7210031347962386e-07, + "loss": 2.436962890625, + "step": 362360 + }, + { + "epoch": 0.9864706682999864, + "grad_norm": 2.167032241821289, + "learning_rate": 2.7100994957066925e-07, + "loss": 2.457579803466797, + "step": 362380 + }, + { + "epoch": 0.9865251122907309, + "grad_norm": 2.174330711364746, + "learning_rate": 2.6991958566171463e-07, + "loss": 2.4491477966308595, + "step": 362400 + }, + { + "epoch": 0.9865795562814754, + "grad_norm": 2.161189079284668, + "learning_rate": 2.6882922175276e-07, + "loss": 2.440556526184082, + "step": 362420 + }, + { + "epoch": 0.9866340002722199, + "grad_norm": 2.219719648361206, + "learning_rate": 2.677388578438054e-07, + "loss": 2.436641502380371, + "step": 362440 + }, + { + "epoch": 0.9866884442629644, + "grad_norm": 2.2485716342926025, + "learning_rate": 2.666484939348508e-07, + "loss": 2.452747917175293, + "step": 362460 + }, + { + "epoch": 0.986742888253709, + "grad_norm": 2.1482746601104736, + "learning_rate": 2.655581300258962e-07, + "loss": 2.49993953704834, + "step": 362480 + }, + { + "epoch": 0.9867973322444535, + "grad_norm": 2.1141774654388428, + "learning_rate": 2.6446776611694156e-07, + "loss": 2.471950149536133, + "step": 362500 + }, + { + "epoch": 0.986851776235198, + "grad_norm": 2.371985912322998, + "learning_rate": 2.6337740220798695e-07, + "loss": 2.4419017791748048, + "step": 362520 + }, + { + "epoch": 0.9869062202259425, + "grad_norm": 2.049701452255249, + "learning_rate": 2.6228703829903233e-07, + "loss": 2.4648513793945312, + "step": 362540 + }, + { + "epoch": 0.986960664216687, + "grad_norm": 2.107646942138672, + "learning_rate": 2.611966743900777e-07, + "loss": 2.455722999572754, + "step": 362560 + }, + { + "epoch": 0.9870151082074317, + "grad_norm": 2.1843998432159424, + "learning_rate": 2.601063104811231e-07, + "loss": 2.460293006896973, + "step": 362580 + }, + { + "epoch": 0.9870695521981762, + "grad_norm": 2.221411943435669, + "learning_rate": 2.590159465721685e-07, + "loss": 2.471713066101074, + "step": 362600 + }, + { + "epoch": 0.9871239961889207, + "grad_norm": 2.2164220809936523, + "learning_rate": 2.579255826632139e-07, + "loss": 2.4645946502685545, + "step": 362620 + }, + { + "epoch": 0.9871784401796652, + "grad_norm": 2.169617176055908, + "learning_rate": 2.5683521875425926e-07, + "loss": 2.4566152572631834, + "step": 362640 + }, + { + "epoch": 0.9872328841704097, + "grad_norm": 2.2048745155334473, + "learning_rate": 2.5574485484530465e-07, + "loss": 2.432381820678711, + "step": 362660 + }, + { + "epoch": 0.9872873281611542, + "grad_norm": 2.1693851947784424, + "learning_rate": 2.5465449093635003e-07, + "loss": 2.4213396072387696, + "step": 362680 + }, + { + "epoch": 0.9873417721518988, + "grad_norm": 2.1834557056427, + "learning_rate": 2.535641270273954e-07, + "loss": 2.463570213317871, + "step": 362700 + }, + { + "epoch": 0.9873962161426433, + "grad_norm": 2.158716917037964, + "learning_rate": 2.524737631184408e-07, + "loss": 2.45739803314209, + "step": 362720 + }, + { + "epoch": 0.9874506601333878, + "grad_norm": 2.2310080528259277, + "learning_rate": 2.513833992094862e-07, + "loss": 2.4303932189941406, + "step": 362740 + }, + { + "epoch": 0.9875051041241323, + "grad_norm": 2.2416741847991943, + "learning_rate": 2.502930353005316e-07, + "loss": 2.4548971176147463, + "step": 362760 + }, + { + "epoch": 0.9875595481148768, + "grad_norm": 2.0872623920440674, + "learning_rate": 2.4920267139157696e-07, + "loss": 2.4581935882568358, + "step": 362780 + }, + { + "epoch": 0.9876139921056214, + "grad_norm": 2.1524229049682617, + "learning_rate": 2.4811230748262235e-07, + "loss": 2.4833492279052733, + "step": 362800 + }, + { + "epoch": 0.9876684360963659, + "grad_norm": 2.1677496433258057, + "learning_rate": 2.4702194357366773e-07, + "loss": 2.449978065490723, + "step": 362820 + }, + { + "epoch": 0.9877228800871104, + "grad_norm": 2.109353542327881, + "learning_rate": 2.459315796647131e-07, + "loss": 2.450301742553711, + "step": 362840 + }, + { + "epoch": 0.9877773240778549, + "grad_norm": 2.195887327194214, + "learning_rate": 2.448412157557585e-07, + "loss": 2.4347639083862305, + "step": 362860 + }, + { + "epoch": 0.9878317680685994, + "grad_norm": 2.168722152709961, + "learning_rate": 2.437508518468039e-07, + "loss": 2.474567413330078, + "step": 362880 + }, + { + "epoch": 0.9878862120593439, + "grad_norm": 2.201847791671753, + "learning_rate": 2.426604879378493e-07, + "loss": 2.4708837509155273, + "step": 362900 + }, + { + "epoch": 0.9879406560500885, + "grad_norm": 2.7058029174804688, + "learning_rate": 2.4157012402889466e-07, + "loss": 2.456010437011719, + "step": 362920 + }, + { + "epoch": 0.987995100040833, + "grad_norm": 2.1978657245635986, + "learning_rate": 2.4047976011994005e-07, + "loss": 2.446823310852051, + "step": 362940 + }, + { + "epoch": 0.9880495440315775, + "grad_norm": 2.0904297828674316, + "learning_rate": 2.3938939621098543e-07, + "loss": 2.4630767822265627, + "step": 362960 + }, + { + "epoch": 0.988103988022322, + "grad_norm": 2.176776170730591, + "learning_rate": 2.3829903230203084e-07, + "loss": 2.479173278808594, + "step": 362980 + }, + { + "epoch": 0.9881584320130665, + "grad_norm": 2.5487165451049805, + "learning_rate": 2.3720866839307623e-07, + "loss": 2.4719635009765626, + "step": 363000 + }, + { + "epoch": 0.988212876003811, + "grad_norm": 2.1050028800964355, + "learning_rate": 2.3611830448412162e-07, + "loss": 2.462274932861328, + "step": 363020 + }, + { + "epoch": 0.9882673199945556, + "grad_norm": 2.1576013565063477, + "learning_rate": 2.3502794057516698e-07, + "loss": 2.480904960632324, + "step": 363040 + }, + { + "epoch": 0.9883217639853001, + "grad_norm": 2.331570863723755, + "learning_rate": 2.3393757666621236e-07, + "loss": 2.443215751647949, + "step": 363060 + }, + { + "epoch": 0.9883762079760446, + "grad_norm": 2.1149933338165283, + "learning_rate": 2.3284721275725775e-07, + "loss": 2.4521352767944338, + "step": 363080 + }, + { + "epoch": 0.9884306519667891, + "grad_norm": 2.2924132347106934, + "learning_rate": 2.3175684884830313e-07, + "loss": 2.459880065917969, + "step": 363100 + }, + { + "epoch": 0.9884850959575336, + "grad_norm": 2.129495620727539, + "learning_rate": 2.3066648493934852e-07, + "loss": 2.439177322387695, + "step": 363120 + }, + { + "epoch": 0.9885395399482783, + "grad_norm": 2.126955032348633, + "learning_rate": 2.295761210303939e-07, + "loss": 2.4525897979736326, + "step": 363140 + }, + { + "epoch": 0.9885939839390228, + "grad_norm": 2.1451313495635986, + "learning_rate": 2.284857571214393e-07, + "loss": 2.4200876235961912, + "step": 363160 + }, + { + "epoch": 0.9886484279297673, + "grad_norm": 2.201991081237793, + "learning_rate": 2.273953932124847e-07, + "loss": 2.4774478912353515, + "step": 363180 + }, + { + "epoch": 0.9887028719205118, + "grad_norm": 2.205603837966919, + "learning_rate": 2.263050293035301e-07, + "loss": 2.488859939575195, + "step": 363200 + }, + { + "epoch": 0.9887573159112563, + "grad_norm": 2.1690399646759033, + "learning_rate": 2.2521466539457547e-07, + "loss": 2.4940690994262695, + "step": 363220 + }, + { + "epoch": 0.9888117599020008, + "grad_norm": 2.1235084533691406, + "learning_rate": 2.2412430148562086e-07, + "loss": 2.4645969390869142, + "step": 363240 + }, + { + "epoch": 0.9888662038927454, + "grad_norm": 2.2475898265838623, + "learning_rate": 2.2303393757666624e-07, + "loss": 2.4418498992919924, + "step": 363260 + }, + { + "epoch": 0.9889206478834899, + "grad_norm": 2.4904842376708984, + "learning_rate": 2.2194357366771163e-07, + "loss": 2.465376091003418, + "step": 363280 + }, + { + "epoch": 0.9889750918742344, + "grad_norm": 2.198563575744629, + "learning_rate": 2.20853209758757e-07, + "loss": 2.462053680419922, + "step": 363300 + }, + { + "epoch": 0.9890295358649789, + "grad_norm": 2.1002845764160156, + "learning_rate": 2.1976284584980238e-07, + "loss": 2.471263122558594, + "step": 363320 + }, + { + "epoch": 0.9890839798557234, + "grad_norm": 2.3214242458343506, + "learning_rate": 2.1867248194084776e-07, + "loss": 2.4574506759643553, + "step": 363340 + }, + { + "epoch": 0.989138423846468, + "grad_norm": 2.0877866744995117, + "learning_rate": 2.1758211803189315e-07, + "loss": 2.4362251281738283, + "step": 363360 + }, + { + "epoch": 0.9891928678372125, + "grad_norm": 2.1853668689727783, + "learning_rate": 2.1649175412293853e-07, + "loss": 2.4306562423706053, + "step": 363380 + }, + { + "epoch": 0.989247311827957, + "grad_norm": 2.1516170501708984, + "learning_rate": 2.1540139021398394e-07, + "loss": 2.4907743453979494, + "step": 363400 + }, + { + "epoch": 0.9893017558187015, + "grad_norm": 2.189951181411743, + "learning_rate": 2.1431102630502933e-07, + "loss": 2.455769920349121, + "step": 363420 + }, + { + "epoch": 0.989356199809446, + "grad_norm": 2.1947526931762695, + "learning_rate": 2.1322066239607472e-07, + "loss": 2.462779235839844, + "step": 363440 + }, + { + "epoch": 0.9894106438001905, + "grad_norm": 2.1450953483581543, + "learning_rate": 2.121302984871201e-07, + "loss": 2.461948013305664, + "step": 363460 + }, + { + "epoch": 0.9894650877909351, + "grad_norm": 2.209822177886963, + "learning_rate": 2.110399345781655e-07, + "loss": 2.4546865463256835, + "step": 363480 + }, + { + "epoch": 0.9895195317816796, + "grad_norm": 2.355290412902832, + "learning_rate": 2.0994957066921087e-07, + "loss": 2.435725784301758, + "step": 363500 + }, + { + "epoch": 0.9895739757724241, + "grad_norm": 2.1586430072784424, + "learning_rate": 2.0885920676025626e-07, + "loss": 2.4822906494140624, + "step": 363520 + }, + { + "epoch": 0.9896284197631686, + "grad_norm": 2.2661166191101074, + "learning_rate": 2.0776884285130164e-07, + "loss": 2.4649024963378907, + "step": 363540 + }, + { + "epoch": 0.9896828637539131, + "grad_norm": 2.226486921310425, + "learning_rate": 2.06678478942347e-07, + "loss": 2.4366477966308593, + "step": 363560 + }, + { + "epoch": 0.9897373077446577, + "grad_norm": 2.1074299812316895, + "learning_rate": 2.055881150333924e-07, + "loss": 2.501089668273926, + "step": 363580 + }, + { + "epoch": 0.9897917517354022, + "grad_norm": 2.386596202850342, + "learning_rate": 2.0449775112443777e-07, + "loss": 2.43350772857666, + "step": 363600 + }, + { + "epoch": 0.9898461957261467, + "grad_norm": 2.1090338230133057, + "learning_rate": 2.034073872154832e-07, + "loss": 2.4592540740966795, + "step": 363620 + }, + { + "epoch": 0.9899006397168912, + "grad_norm": 2.195037603378296, + "learning_rate": 2.0231702330652857e-07, + "loss": 2.457616996765137, + "step": 363640 + }, + { + "epoch": 0.9899550837076357, + "grad_norm": 2.375734806060791, + "learning_rate": 2.0122665939757396e-07, + "loss": 2.4448766708374023, + "step": 363660 + }, + { + "epoch": 0.9900095276983802, + "grad_norm": 2.1556003093719482, + "learning_rate": 2.0013629548861934e-07, + "loss": 2.462374687194824, + "step": 363680 + }, + { + "epoch": 0.9900639716891249, + "grad_norm": 2.1984000205993652, + "learning_rate": 1.9904593157966473e-07, + "loss": 2.4787086486816405, + "step": 363700 + }, + { + "epoch": 0.9901184156798694, + "grad_norm": 2.176408052444458, + "learning_rate": 1.9795556767071012e-07, + "loss": 2.475101280212402, + "step": 363720 + }, + { + "epoch": 0.9901728596706139, + "grad_norm": 2.224597930908203, + "learning_rate": 1.968652037617555e-07, + "loss": 2.4881732940673826, + "step": 363740 + }, + { + "epoch": 0.9902273036613584, + "grad_norm": 2.2673652172088623, + "learning_rate": 1.957748398528009e-07, + "loss": 2.4714927673339844, + "step": 363760 + }, + { + "epoch": 0.9902817476521029, + "grad_norm": 2.178597927093506, + "learning_rate": 1.9468447594384627e-07, + "loss": 2.4653366088867186, + "step": 363780 + }, + { + "epoch": 0.9903361916428475, + "grad_norm": 2.13218355178833, + "learning_rate": 1.9359411203489166e-07, + "loss": 2.4593738555908202, + "step": 363800 + }, + { + "epoch": 0.990390635633592, + "grad_norm": 2.2090060710906982, + "learning_rate": 1.9250374812593707e-07, + "loss": 2.4410242080688476, + "step": 363820 + }, + { + "epoch": 0.9904450796243365, + "grad_norm": 2.2214133739471436, + "learning_rate": 1.9141338421698246e-07, + "loss": 2.4599639892578127, + "step": 363840 + }, + { + "epoch": 0.990499523615081, + "grad_norm": 2.2819550037384033, + "learning_rate": 1.9032302030802784e-07, + "loss": 2.4789257049560547, + "step": 363860 + }, + { + "epoch": 0.9905539676058255, + "grad_norm": 2.2625484466552734, + "learning_rate": 1.892326563990732e-07, + "loss": 2.464994430541992, + "step": 363880 + }, + { + "epoch": 0.99060841159657, + "grad_norm": 2.156785488128662, + "learning_rate": 1.881422924901186e-07, + "loss": 2.4843776702880858, + "step": 363900 + }, + { + "epoch": 0.9906628555873146, + "grad_norm": 2.0964508056640625, + "learning_rate": 1.8705192858116397e-07, + "loss": 2.471826934814453, + "step": 363920 + }, + { + "epoch": 0.9907172995780591, + "grad_norm": 2.1957035064697266, + "learning_rate": 1.8596156467220936e-07, + "loss": 2.442118263244629, + "step": 363940 + }, + { + "epoch": 0.9907717435688036, + "grad_norm": 2.1377766132354736, + "learning_rate": 1.8487120076325474e-07, + "loss": 2.4688699722290037, + "step": 363960 + }, + { + "epoch": 0.9908261875595481, + "grad_norm": 2.1110570430755615, + "learning_rate": 1.8378083685430013e-07, + "loss": 2.4530269622802736, + "step": 363980 + }, + { + "epoch": 0.9908806315502926, + "grad_norm": 2.326707124710083, + "learning_rate": 1.8269047294534552e-07, + "loss": 2.461065101623535, + "step": 364000 + }, + { + "epoch": 0.9909350755410371, + "grad_norm": 2.165027141571045, + "learning_rate": 1.816001090363909e-07, + "loss": 2.4547828674316405, + "step": 364020 + }, + { + "epoch": 0.9909895195317817, + "grad_norm": 2.3165056705474854, + "learning_rate": 1.805097451274363e-07, + "loss": 2.453180694580078, + "step": 364040 + }, + { + "epoch": 0.9910439635225262, + "grad_norm": 2.1959939002990723, + "learning_rate": 1.794193812184817e-07, + "loss": 2.4645946502685545, + "step": 364060 + }, + { + "epoch": 0.9910984075132707, + "grad_norm": 2.1697638034820557, + "learning_rate": 1.7832901730952708e-07, + "loss": 2.467992401123047, + "step": 364080 + }, + { + "epoch": 0.9911528515040152, + "grad_norm": 2.25614595413208, + "learning_rate": 1.7723865340057247e-07, + "loss": 2.4541465759277346, + "step": 364100 + }, + { + "epoch": 0.9912072954947597, + "grad_norm": 2.1753904819488525, + "learning_rate": 1.7614828949161786e-07, + "loss": 2.4436330795288086, + "step": 364120 + }, + { + "epoch": 0.9912617394855043, + "grad_norm": 3.946932554244995, + "learning_rate": 1.7505792558266322e-07, + "loss": 2.429534149169922, + "step": 364140 + }, + { + "epoch": 0.9913161834762488, + "grad_norm": 2.1996350288391113, + "learning_rate": 1.739675616737086e-07, + "loss": 2.4434362411499024, + "step": 364160 + }, + { + "epoch": 0.9913706274669933, + "grad_norm": 2.35652756690979, + "learning_rate": 1.7287719776475399e-07, + "loss": 2.435232162475586, + "step": 364180 + }, + { + "epoch": 0.9914250714577378, + "grad_norm": 2.1430583000183105, + "learning_rate": 1.7178683385579937e-07, + "loss": 2.457654571533203, + "step": 364200 + }, + { + "epoch": 0.9914795154484823, + "grad_norm": 2.116114616394043, + "learning_rate": 1.7069646994684476e-07, + "loss": 2.4800134658813477, + "step": 364220 + }, + { + "epoch": 0.9915339594392268, + "grad_norm": 2.272937774658203, + "learning_rate": 1.6960610603789017e-07, + "loss": 2.4514396667480467, + "step": 364240 + }, + { + "epoch": 0.9915884034299715, + "grad_norm": 2.191537618637085, + "learning_rate": 1.6851574212893556e-07, + "loss": 2.4469320297241213, + "step": 364260 + }, + { + "epoch": 0.991642847420716, + "grad_norm": 2.148358106613159, + "learning_rate": 1.6742537821998094e-07, + "loss": 2.4527824401855467, + "step": 364280 + }, + { + "epoch": 0.9916972914114605, + "grad_norm": 2.192064046859741, + "learning_rate": 1.6633501431102633e-07, + "loss": 2.4444276809692385, + "step": 364300 + }, + { + "epoch": 0.991751735402205, + "grad_norm": 2.3016605377197266, + "learning_rate": 1.652446504020717e-07, + "loss": 2.4957948684692384, + "step": 364320 + }, + { + "epoch": 0.9918061793929495, + "grad_norm": 2.215818405151367, + "learning_rate": 1.641542864931171e-07, + "loss": 2.467081832885742, + "step": 364340 + }, + { + "epoch": 0.9918606233836941, + "grad_norm": 2.1799206733703613, + "learning_rate": 1.6306392258416248e-07, + "loss": 2.4846263885498048, + "step": 364360 + }, + { + "epoch": 0.9919150673744386, + "grad_norm": 2.0504744052886963, + "learning_rate": 1.6197355867520787e-07, + "loss": 2.4829118728637694, + "step": 364380 + }, + { + "epoch": 0.9919695113651831, + "grad_norm": 2.11826491355896, + "learning_rate": 1.6088319476625323e-07, + "loss": 2.4585437774658203, + "step": 364400 + }, + { + "epoch": 0.9920239553559276, + "grad_norm": 2.0923283100128174, + "learning_rate": 1.5979283085729861e-07, + "loss": 2.4083654403686525, + "step": 364420 + }, + { + "epoch": 0.9920783993466721, + "grad_norm": 2.241306781768799, + "learning_rate": 1.58702466948344e-07, + "loss": 2.4217060089111326, + "step": 364440 + }, + { + "epoch": 0.9921328433374166, + "grad_norm": 2.199127435684204, + "learning_rate": 1.576121030393894e-07, + "loss": 2.451068305969238, + "step": 364460 + }, + { + "epoch": 0.9921872873281612, + "grad_norm": 2.16464900970459, + "learning_rate": 1.565217391304348e-07, + "loss": 2.4406463623046877, + "step": 364480 + }, + { + "epoch": 0.9922417313189057, + "grad_norm": 2.1838955879211426, + "learning_rate": 1.5543137522148018e-07, + "loss": 2.434691047668457, + "step": 364500 + }, + { + "epoch": 0.9922961753096502, + "grad_norm": 2.687420606613159, + "learning_rate": 1.5434101131252557e-07, + "loss": 2.4723751068115236, + "step": 364520 + }, + { + "epoch": 0.9923506193003947, + "grad_norm": 2.1229300498962402, + "learning_rate": 1.5325064740357096e-07, + "loss": 2.4690013885498048, + "step": 364540 + }, + { + "epoch": 0.9924050632911392, + "grad_norm": 2.112302541732788, + "learning_rate": 1.5216028349461634e-07, + "loss": 2.4629371643066404, + "step": 364560 + }, + { + "epoch": 0.9924595072818838, + "grad_norm": 2.1795809268951416, + "learning_rate": 1.5106991958566173e-07, + "loss": 2.4782100677490235, + "step": 364580 + }, + { + "epoch": 0.9925139512726283, + "grad_norm": 2.24991512298584, + "learning_rate": 1.499795556767071e-07, + "loss": 2.407380485534668, + "step": 364600 + }, + { + "epoch": 0.9925683952633728, + "grad_norm": 2.1269068717956543, + "learning_rate": 1.488891917677525e-07, + "loss": 2.475540351867676, + "step": 364620 + }, + { + "epoch": 0.9926228392541173, + "grad_norm": 2.1775527000427246, + "learning_rate": 1.4779882785879788e-07, + "loss": 2.4510284423828126, + "step": 364640 + }, + { + "epoch": 0.9926772832448618, + "grad_norm": 2.0374951362609863, + "learning_rate": 1.4670846394984327e-07, + "loss": 2.452041816711426, + "step": 364660 + }, + { + "epoch": 0.9927317272356063, + "grad_norm": 2.208798408508301, + "learning_rate": 1.4561810004088866e-07, + "loss": 2.4778499603271484, + "step": 364680 + }, + { + "epoch": 0.9927861712263509, + "grad_norm": 2.207481622695923, + "learning_rate": 1.4452773613193404e-07, + "loss": 2.4833412170410156, + "step": 364700 + }, + { + "epoch": 0.9928406152170954, + "grad_norm": 2.131452798843384, + "learning_rate": 1.4343737222297943e-07, + "loss": 2.447982406616211, + "step": 364720 + }, + { + "epoch": 0.9928950592078399, + "grad_norm": 2.232306718826294, + "learning_rate": 1.423470083140248e-07, + "loss": 2.473301315307617, + "step": 364740 + }, + { + "epoch": 0.9929495031985844, + "grad_norm": 88.0784683227539, + "learning_rate": 1.412566444050702e-07, + "loss": 2.4254568099975584, + "step": 364760 + }, + { + "epoch": 0.9930039471893289, + "grad_norm": 2.096451997756958, + "learning_rate": 1.4016628049611558e-07, + "loss": 2.4140697479248048, + "step": 364780 + }, + { + "epoch": 0.9930583911800736, + "grad_norm": 2.1030869483947754, + "learning_rate": 1.3907591658716097e-07, + "loss": 2.452566719055176, + "step": 364800 + }, + { + "epoch": 0.993112835170818, + "grad_norm": 2.159454107284546, + "learning_rate": 1.3798555267820638e-07, + "loss": 2.4829524993896483, + "step": 364820 + }, + { + "epoch": 0.9931672791615626, + "grad_norm": 2.348741054534912, + "learning_rate": 1.3689518876925177e-07, + "loss": 2.462098503112793, + "step": 364840 + }, + { + "epoch": 0.9932217231523071, + "grad_norm": 2.124546766281128, + "learning_rate": 1.3580482486029713e-07, + "loss": 2.4341196060180663, + "step": 364860 + }, + { + "epoch": 0.9932761671430516, + "grad_norm": 2.1360039710998535, + "learning_rate": 1.347144609513425e-07, + "loss": 2.4592430114746096, + "step": 364880 + }, + { + "epoch": 0.9933306111337961, + "grad_norm": 2.723626136779785, + "learning_rate": 1.336240970423879e-07, + "loss": 2.4850358963012695, + "step": 364900 + }, + { + "epoch": 0.9933850551245407, + "grad_norm": 2.2250959873199463, + "learning_rate": 1.3253373313343328e-07, + "loss": 2.4375560760498045, + "step": 364920 + }, + { + "epoch": 0.9934394991152852, + "grad_norm": 2.2131619453430176, + "learning_rate": 1.314433692244787e-07, + "loss": 2.4541648864746093, + "step": 364940 + }, + { + "epoch": 0.9934939431060297, + "grad_norm": 2.1506786346435547, + "learning_rate": 1.3035300531552408e-07, + "loss": 2.4524824142456056, + "step": 364960 + }, + { + "epoch": 0.9935483870967742, + "grad_norm": 2.2849161624908447, + "learning_rate": 1.2926264140656944e-07, + "loss": 2.4719839096069336, + "step": 364980 + }, + { + "epoch": 0.9936028310875187, + "grad_norm": 2.1741859912872314, + "learning_rate": 1.2817227749761483e-07, + "loss": 2.470775604248047, + "step": 365000 + }, + { + "epoch": 0.9936572750782633, + "grad_norm": 2.1022417545318604, + "learning_rate": 1.270819135886602e-07, + "loss": 2.454205322265625, + "step": 365020 + }, + { + "epoch": 0.9937117190690078, + "grad_norm": 2.1153719425201416, + "learning_rate": 1.2599154967970562e-07, + "loss": 2.458204650878906, + "step": 365040 + }, + { + "epoch": 0.9937661630597523, + "grad_norm": 2.098997116088867, + "learning_rate": 1.24901185770751e-07, + "loss": 2.486542320251465, + "step": 365060 + }, + { + "epoch": 0.9938206070504968, + "grad_norm": 2.1750540733337402, + "learning_rate": 1.238108218617964e-07, + "loss": 2.438352584838867, + "step": 365080 + }, + { + "epoch": 0.9938750510412413, + "grad_norm": 2.147740364074707, + "learning_rate": 1.2272045795284178e-07, + "loss": 2.46875114440918, + "step": 365100 + }, + { + "epoch": 0.9939294950319858, + "grad_norm": 2.2745516300201416, + "learning_rate": 1.2163009404388714e-07, + "loss": 2.462163734436035, + "step": 365120 + }, + { + "epoch": 0.9939839390227304, + "grad_norm": 2.243086576461792, + "learning_rate": 1.2053973013493255e-07, + "loss": 2.4919364929199217, + "step": 365140 + }, + { + "epoch": 0.9940383830134749, + "grad_norm": 2.139028549194336, + "learning_rate": 1.1944936622597794e-07, + "loss": 2.4344526290893556, + "step": 365160 + }, + { + "epoch": 0.9940928270042194, + "grad_norm": 2.126824140548706, + "learning_rate": 1.1835900231702332e-07, + "loss": 2.474148750305176, + "step": 365180 + }, + { + "epoch": 0.9941472709949639, + "grad_norm": 2.1860156059265137, + "learning_rate": 1.172686384080687e-07, + "loss": 2.4815868377685546, + "step": 365200 + }, + { + "epoch": 0.9942017149857084, + "grad_norm": 2.1810052394866943, + "learning_rate": 1.1617827449911408e-07, + "loss": 2.445050811767578, + "step": 365220 + }, + { + "epoch": 0.9942561589764529, + "grad_norm": 2.1426234245300293, + "learning_rate": 1.1508791059015947e-07, + "loss": 2.4538755416870117, + "step": 365240 + }, + { + "epoch": 0.9943106029671975, + "grad_norm": 2.1954293251037598, + "learning_rate": 1.1399754668120487e-07, + "loss": 2.4272409439086915, + "step": 365260 + }, + { + "epoch": 0.994365046957942, + "grad_norm": 2.2856571674346924, + "learning_rate": 1.1290718277225025e-07, + "loss": 2.4523603439331056, + "step": 365280 + }, + { + "epoch": 0.9944194909486865, + "grad_norm": 2.2770864963531494, + "learning_rate": 1.1181681886329564e-07, + "loss": 2.4637950897216796, + "step": 365300 + }, + { + "epoch": 0.994473934939431, + "grad_norm": 2.227137804031372, + "learning_rate": 1.1072645495434101e-07, + "loss": 2.470928192138672, + "step": 365320 + }, + { + "epoch": 0.9945283789301755, + "grad_norm": 2.1262824535369873, + "learning_rate": 1.096360910453864e-07, + "loss": 2.445475769042969, + "step": 365340 + }, + { + "epoch": 0.9945828229209202, + "grad_norm": 2.1349616050720215, + "learning_rate": 1.085457271364318e-07, + "loss": 2.452885055541992, + "step": 365360 + }, + { + "epoch": 0.9946372669116647, + "grad_norm": 2.177222728729248, + "learning_rate": 1.0745536322747718e-07, + "loss": 2.4430728912353517, + "step": 365380 + }, + { + "epoch": 0.9946917109024092, + "grad_norm": 2.1828742027282715, + "learning_rate": 1.0636499931852257e-07, + "loss": 2.4605913162231445, + "step": 365400 + }, + { + "epoch": 0.9947461548931537, + "grad_norm": 2.2228052616119385, + "learning_rate": 1.0527463540956795e-07, + "loss": 2.4657121658325196, + "step": 365420 + }, + { + "epoch": 0.9948005988838982, + "grad_norm": 2.2068285942077637, + "learning_rate": 1.0418427150061334e-07, + "loss": 2.4785530090332033, + "step": 365440 + }, + { + "epoch": 0.9948550428746427, + "grad_norm": 2.2207846641540527, + "learning_rate": 1.0309390759165874e-07, + "loss": 2.428748893737793, + "step": 365460 + }, + { + "epoch": 0.9949094868653873, + "grad_norm": 2.1533732414245605, + "learning_rate": 1.0200354368270411e-07, + "loss": 2.443098258972168, + "step": 365480 + }, + { + "epoch": 0.9949639308561318, + "grad_norm": 2.46291184425354, + "learning_rate": 1.009131797737495e-07, + "loss": 2.4262939453125, + "step": 365500 + }, + { + "epoch": 0.9950183748468763, + "grad_norm": 2.22602915763855, + "learning_rate": 9.982281586479488e-08, + "loss": 2.4271020889282227, + "step": 365520 + }, + { + "epoch": 0.9950728188376208, + "grad_norm": 2.194065809249878, + "learning_rate": 9.873245195584027e-08, + "loss": 2.4558143615722656, + "step": 365540 + }, + { + "epoch": 0.9951272628283653, + "grad_norm": 2.175860643386841, + "learning_rate": 9.764208804688565e-08, + "loss": 2.4641395568847657, + "step": 365560 + }, + { + "epoch": 0.9951817068191099, + "grad_norm": 2.2931931018829346, + "learning_rate": 9.655172413793105e-08, + "loss": 2.4529563903808596, + "step": 365580 + }, + { + "epoch": 0.9952361508098544, + "grad_norm": 2.079444646835327, + "learning_rate": 9.546136022897644e-08, + "loss": 2.449985694885254, + "step": 365600 + }, + { + "epoch": 0.9952905948005989, + "grad_norm": 2.2587215900421143, + "learning_rate": 9.437099632002181e-08, + "loss": 2.4634382247924806, + "step": 365620 + }, + { + "epoch": 0.9953450387913434, + "grad_norm": 2.1760475635528564, + "learning_rate": 9.32806324110672e-08, + "loss": 2.4164443969726563, + "step": 365640 + }, + { + "epoch": 0.9953994827820879, + "grad_norm": 2.2857325077056885, + "learning_rate": 9.219026850211258e-08, + "loss": 2.4513336181640626, + "step": 365660 + }, + { + "epoch": 0.9954539267728324, + "grad_norm": 2.1579558849334717, + "learning_rate": 9.109990459315798e-08, + "loss": 2.471873474121094, + "step": 365680 + }, + { + "epoch": 0.995508370763577, + "grad_norm": 2.1473495960235596, + "learning_rate": 9.000954068420337e-08, + "loss": 2.4818016052246095, + "step": 365700 + }, + { + "epoch": 0.9955628147543215, + "grad_norm": 2.1204895973205566, + "learning_rate": 8.891917677524875e-08, + "loss": 2.445977783203125, + "step": 365720 + }, + { + "epoch": 0.995617258745066, + "grad_norm": 2.1034657955169678, + "learning_rate": 8.782881286629412e-08, + "loss": 2.4588890075683594, + "step": 365740 + }, + { + "epoch": 0.9956717027358105, + "grad_norm": 2.148012638092041, + "learning_rate": 8.673844895733951e-08, + "loss": 2.473000335693359, + "step": 365760 + }, + { + "epoch": 0.995726146726555, + "grad_norm": 2.364865779876709, + "learning_rate": 8.564808504838491e-08, + "loss": 2.4934055328369142, + "step": 365780 + }, + { + "epoch": 0.9957805907172996, + "grad_norm": 2.164809226989746, + "learning_rate": 8.45577211394303e-08, + "loss": 2.453487014770508, + "step": 365800 + }, + { + "epoch": 0.9958350347080441, + "grad_norm": 2.1759886741638184, + "learning_rate": 8.346735723047568e-08, + "loss": 2.474957275390625, + "step": 365820 + }, + { + "epoch": 0.9958894786987886, + "grad_norm": 2.2473535537719727, + "learning_rate": 8.237699332152107e-08, + "loss": 2.4374422073364257, + "step": 365840 + }, + { + "epoch": 0.9959439226895331, + "grad_norm": 2.143202781677246, + "learning_rate": 8.128662941256645e-08, + "loss": 2.452170753479004, + "step": 365860 + }, + { + "epoch": 0.9959983666802776, + "grad_norm": 2.075052261352539, + "learning_rate": 8.019626550361185e-08, + "loss": 2.4775741577148436, + "step": 365880 + }, + { + "epoch": 0.9960528106710221, + "grad_norm": 2.2354018688201904, + "learning_rate": 7.910590159465722e-08, + "loss": 2.452326011657715, + "step": 365900 + }, + { + "epoch": 0.9961072546617668, + "grad_norm": 2.1701338291168213, + "learning_rate": 7.801553768570261e-08, + "loss": 2.500111198425293, + "step": 365920 + }, + { + "epoch": 0.9961616986525113, + "grad_norm": 2.4978444576263428, + "learning_rate": 7.6925173776748e-08, + "loss": 2.4659257888793946, + "step": 365940 + }, + { + "epoch": 0.9962161426432558, + "grad_norm": 2.1342382431030273, + "learning_rate": 7.583480986779338e-08, + "loss": 2.4521915435791017, + "step": 365960 + }, + { + "epoch": 0.9962705866340003, + "grad_norm": 2.1204605102539062, + "learning_rate": 7.474444595883877e-08, + "loss": 2.4313995361328127, + "step": 365980 + }, + { + "epoch": 0.9963250306247448, + "grad_norm": 2.2083804607391357, + "learning_rate": 7.365408204988415e-08, + "loss": 2.4920976638793944, + "step": 366000 + }, + { + "epoch": 0.9963794746154894, + "grad_norm": 2.2830324172973633, + "learning_rate": 7.256371814092955e-08, + "loss": 2.444645118713379, + "step": 366020 + }, + { + "epoch": 0.9964339186062339, + "grad_norm": 2.1408944129943848, + "learning_rate": 7.147335423197492e-08, + "loss": 2.4677156448364257, + "step": 366040 + }, + { + "epoch": 0.9964883625969784, + "grad_norm": 2.12532377243042, + "learning_rate": 7.038299032302031e-08, + "loss": 2.454502296447754, + "step": 366060 + }, + { + "epoch": 0.9965428065877229, + "grad_norm": 2.088575839996338, + "learning_rate": 6.929262641406571e-08, + "loss": 2.468748664855957, + "step": 366080 + }, + { + "epoch": 0.9965972505784674, + "grad_norm": 2.3880836963653564, + "learning_rate": 6.820226250511108e-08, + "loss": 2.4690088272094726, + "step": 366100 + }, + { + "epoch": 0.9966516945692119, + "grad_norm": 2.266122579574585, + "learning_rate": 6.711189859615648e-08, + "loss": 2.4579524993896484, + "step": 366120 + }, + { + "epoch": 0.9967061385599565, + "grad_norm": 2.3249363899230957, + "learning_rate": 6.602153468720186e-08, + "loss": 2.4453800201416014, + "step": 366140 + }, + { + "epoch": 0.996760582550701, + "grad_norm": 2.085144519805908, + "learning_rate": 6.493117077824724e-08, + "loss": 2.4239988327026367, + "step": 366160 + }, + { + "epoch": 0.9968150265414455, + "grad_norm": 2.186183452606201, + "learning_rate": 6.384080686929264e-08, + "loss": 2.461771011352539, + "step": 366180 + }, + { + "epoch": 0.99686947053219, + "grad_norm": 2.059595823287964, + "learning_rate": 6.275044296033802e-08, + "loss": 2.4650852203369142, + "step": 366200 + }, + { + "epoch": 0.9969239145229345, + "grad_norm": 2.287527084350586, + "learning_rate": 6.16600790513834e-08, + "loss": 2.4275840759277343, + "step": 366220 + }, + { + "epoch": 0.996978358513679, + "grad_norm": 2.3866219520568848, + "learning_rate": 6.056971514242879e-08, + "loss": 2.463980484008789, + "step": 366240 + }, + { + "epoch": 0.9970328025044236, + "grad_norm": 2.332535743713379, + "learning_rate": 5.947935123347417e-08, + "loss": 2.493862533569336, + "step": 366260 + }, + { + "epoch": 0.9970872464951681, + "grad_norm": 2.2634506225585938, + "learning_rate": 5.8388987324519564e-08, + "loss": 2.440521240234375, + "step": 366280 + }, + { + "epoch": 0.9971416904859126, + "grad_norm": 2.266477346420288, + "learning_rate": 5.729862341556495e-08, + "loss": 2.4788761138916016, + "step": 366300 + }, + { + "epoch": 0.9971961344766571, + "grad_norm": 2.285344123840332, + "learning_rate": 5.6208259506610336e-08, + "loss": 2.4679887771606444, + "step": 366320 + }, + { + "epoch": 0.9972505784674016, + "grad_norm": 2.1425325870513916, + "learning_rate": 5.511789559765573e-08, + "loss": 2.470718765258789, + "step": 366340 + }, + { + "epoch": 0.9973050224581462, + "grad_norm": 2.6866848468780518, + "learning_rate": 5.402753168870111e-08, + "loss": 2.4899971008300783, + "step": 366360 + }, + { + "epoch": 0.9973594664488907, + "grad_norm": 2.2529032230377197, + "learning_rate": 5.293716777974649e-08, + "loss": 2.4333894729614256, + "step": 366380 + }, + { + "epoch": 0.9974139104396352, + "grad_norm": 2.313429355621338, + "learning_rate": 5.1846803870791885e-08, + "loss": 2.463421440124512, + "step": 366400 + }, + { + "epoch": 0.9974683544303797, + "grad_norm": 2.127286672592163, + "learning_rate": 5.0756439961837264e-08, + "loss": 2.4513002395629884, + "step": 366420 + }, + { + "epoch": 0.9975227984211242, + "grad_norm": 2.110673666000366, + "learning_rate": 4.9666076052882656e-08, + "loss": 2.4390161514282225, + "step": 366440 + }, + { + "epoch": 0.9975772424118687, + "grad_norm": 2.115696907043457, + "learning_rate": 4.857571214392804e-08, + "loss": 2.4351654052734375, + "step": 366460 + }, + { + "epoch": 0.9976316864026133, + "grad_norm": 2.1633505821228027, + "learning_rate": 4.748534823497342e-08, + "loss": 2.4633563995361327, + "step": 366480 + }, + { + "epoch": 0.9976861303933579, + "grad_norm": 2.216682195663452, + "learning_rate": 4.6394984326018813e-08, + "loss": 2.4417179107666014, + "step": 366500 + }, + { + "epoch": 0.9977405743841024, + "grad_norm": 2.2078378200531006, + "learning_rate": 4.53046204170642e-08, + "loss": 2.4614328384399413, + "step": 366520 + }, + { + "epoch": 0.9977950183748469, + "grad_norm": 2.359995126724243, + "learning_rate": 4.421425650810958e-08, + "loss": 2.475500297546387, + "step": 366540 + }, + { + "epoch": 0.9978494623655914, + "grad_norm": 2.6683597564697266, + "learning_rate": 4.312389259915497e-08, + "loss": 2.4909339904785157, + "step": 366560 + }, + { + "epoch": 0.997903906356336, + "grad_norm": 2.3164265155792236, + "learning_rate": 4.2033528690200356e-08, + "loss": 2.460844039916992, + "step": 366580 + }, + { + "epoch": 0.9979583503470805, + "grad_norm": 2.1423497200012207, + "learning_rate": 4.094316478124575e-08, + "loss": 2.4365633010864256, + "step": 366600 + }, + { + "epoch": 0.998012794337825, + "grad_norm": 2.1628050804138184, + "learning_rate": 3.985280087229113e-08, + "loss": 2.4333621978759767, + "step": 366620 + }, + { + "epoch": 0.9980672383285695, + "grad_norm": 2.6077239513397217, + "learning_rate": 3.876243696333652e-08, + "loss": 2.441092109680176, + "step": 366640 + }, + { + "epoch": 0.998121682319314, + "grad_norm": 2.281550407409668, + "learning_rate": 3.7672073054381906e-08, + "loss": 2.466069221496582, + "step": 366660 + }, + { + "epoch": 0.9981761263100585, + "grad_norm": 2.1351895332336426, + "learning_rate": 3.658170914542729e-08, + "loss": 2.452229881286621, + "step": 366680 + }, + { + "epoch": 0.9982305703008031, + "grad_norm": 2.190704584121704, + "learning_rate": 3.549134523647268e-08, + "loss": 2.4520875930786135, + "step": 366700 + }, + { + "epoch": 0.9982850142915476, + "grad_norm": 2.121429443359375, + "learning_rate": 3.440098132751806e-08, + "loss": 2.448509407043457, + "step": 366720 + }, + { + "epoch": 0.9983394582822921, + "grad_norm": 2.223041534423828, + "learning_rate": 3.331061741856345e-08, + "loss": 2.4608989715576173, + "step": 366740 + }, + { + "epoch": 0.9983939022730366, + "grad_norm": 4.288431644439697, + "learning_rate": 3.2220253509608834e-08, + "loss": 2.471683883666992, + "step": 366760 + }, + { + "epoch": 0.9984483462637811, + "grad_norm": 2.189678430557251, + "learning_rate": 3.112988960065422e-08, + "loss": 2.447933578491211, + "step": 366780 + }, + { + "epoch": 0.9985027902545257, + "grad_norm": 2.1892969608306885, + "learning_rate": 3.0039525691699606e-08, + "loss": 2.4980489730834963, + "step": 366800 + }, + { + "epoch": 0.9985572342452702, + "grad_norm": 2.167409896850586, + "learning_rate": 2.8949161782744995e-08, + "loss": 2.465646743774414, + "step": 366820 + }, + { + "epoch": 0.9986116782360147, + "grad_norm": 2.061704635620117, + "learning_rate": 2.7858797873790377e-08, + "loss": 2.4428876876831054, + "step": 366840 + }, + { + "epoch": 0.9986661222267592, + "grad_norm": 2.1553328037261963, + "learning_rate": 2.6768433964835766e-08, + "loss": 2.4525075912475587, + "step": 366860 + }, + { + "epoch": 0.9987205662175037, + "grad_norm": 2.1027133464813232, + "learning_rate": 2.567807005588115e-08, + "loss": 2.4485452651977537, + "step": 366880 + }, + { + "epoch": 0.9987750102082482, + "grad_norm": 2.152679681777954, + "learning_rate": 2.458770614692654e-08, + "loss": 2.455280303955078, + "step": 366900 + }, + { + "epoch": 0.9988294541989928, + "grad_norm": 2.214081048965454, + "learning_rate": 2.3497342237971923e-08, + "loss": 2.449759674072266, + "step": 366920 + }, + { + "epoch": 0.9988838981897373, + "grad_norm": 2.225510835647583, + "learning_rate": 2.2406978329017312e-08, + "loss": 2.474194145202637, + "step": 366940 + }, + { + "epoch": 0.9989383421804818, + "grad_norm": 2.3249118328094482, + "learning_rate": 2.1316614420062698e-08, + "loss": 2.450154113769531, + "step": 366960 + }, + { + "epoch": 0.9989927861712263, + "grad_norm": 2.4043965339660645, + "learning_rate": 2.0226250511108087e-08, + "loss": 2.474380111694336, + "step": 366980 + }, + { + "epoch": 0.9990472301619708, + "grad_norm": 2.1438581943511963, + "learning_rate": 1.9135886602153472e-08, + "loss": 2.4504020690917967, + "step": 367000 + }, + { + "epoch": 0.9991016741527154, + "grad_norm": 2.2675251960754395, + "learning_rate": 1.8045522693198855e-08, + "loss": 2.4582799911499023, + "step": 367020 + }, + { + "epoch": 0.99915611814346, + "grad_norm": 2.1915926933288574, + "learning_rate": 1.6955158784244244e-08, + "loss": 2.4259408950805663, + "step": 367040 + }, + { + "epoch": 0.9992105621342044, + "grad_norm": 2.5689408779144287, + "learning_rate": 1.586479487528963e-08, + "loss": 2.4405086517333983, + "step": 367060 + }, + { + "epoch": 0.999265006124949, + "grad_norm": 3.2477705478668213, + "learning_rate": 1.4774430966335017e-08, + "loss": 2.464736557006836, + "step": 367080 + }, + { + "epoch": 0.9993194501156935, + "grad_norm": 2.3395988941192627, + "learning_rate": 1.3684067057380401e-08, + "loss": 2.435229682922363, + "step": 367100 + }, + { + "epoch": 0.999373894106438, + "grad_norm": 2.5150749683380127, + "learning_rate": 1.2593703148425788e-08, + "loss": 2.4993825912475587, + "step": 367120 + }, + { + "epoch": 0.9994283380971826, + "grad_norm": 2.7006707191467285, + "learning_rate": 1.1503339239471174e-08, + "loss": 2.438041877746582, + "step": 367140 + }, + { + "epoch": 0.9994827820879271, + "grad_norm": 2.1441943645477295, + "learning_rate": 1.0412975330516561e-08, + "loss": 2.4709169387817385, + "step": 367160 + }, + { + "epoch": 0.9995372260786716, + "grad_norm": 2.209688901901245, + "learning_rate": 9.322611421561947e-09, + "loss": 2.4540994644165037, + "step": 367180 + }, + { + "epoch": 0.9995916700694161, + "grad_norm": 2.0987143516540527, + "learning_rate": 8.232247512607334e-09, + "loss": 2.456298065185547, + "step": 367200 + }, + { + "epoch": 0.9996461140601606, + "grad_norm": 2.1015682220458984, + "learning_rate": 7.14188360365272e-09, + "loss": 2.4620651245117187, + "step": 367220 + }, + { + "epoch": 0.9997005580509052, + "grad_norm": 2.3255767822265625, + "learning_rate": 6.0515196946981066e-09, + "loss": 2.4439517974853517, + "step": 367240 + }, + { + "epoch": 0.9997550020416497, + "grad_norm": 2.220130205154419, + "learning_rate": 4.961155785743492e-09, + "loss": 2.456905746459961, + "step": 367260 + }, + { + "epoch": 0.9998094460323942, + "grad_norm": 2.148595094680786, + "learning_rate": 3.870791876788879e-09, + "loss": 2.4264347076416017, + "step": 367280 + }, + { + "epoch": 0.9998638900231387, + "grad_norm": 2.3798465728759766, + "learning_rate": 2.780427967834265e-09, + "loss": 2.4460451126098635, + "step": 367300 + }, + { + "epoch": 0.9999183340138832, + "grad_norm": 2.078782320022583, + "learning_rate": 1.6900640588796512e-09, + "loss": 2.436610984802246, + "step": 367320 + }, + { + "epoch": 0.9999727780046277, + "grad_norm": 2.0716283321380615, + "learning_rate": 5.997001499250375e-10, + "loss": 2.4097354888916014, + "step": 367340 + } + ], + "logging_steps": 20, + "max_steps": 367350, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 200, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 256, + "trial_name": null, + "trial_params": null +}