Files
math-1b-sft-numinamath-bs51…/trainer_log.jsonl

169 lines
32 KiB
Plaintext
Raw Normal View History

{"current_steps": 20, "total_steps": 3358, "loss": 1.3592, "lr": 1.8811881188118813e-06, "epoch": 0.005956813104988831, "percentage": 0.6, "elapsed_time": "0:00:42", "remaining_time": "1:57:35"}
{"current_steps": 40, "total_steps": 3358, "loss": 1.0573, "lr": 3.861386138613862e-06, "epoch": 0.011913626209977662, "percentage": 1.19, "elapsed_time": "0:01:15", "remaining_time": "1:44:09"}
{"current_steps": 60, "total_steps": 3358, "loss": 0.8295, "lr": 5.841584158415842e-06, "epoch": 0.017870439314966492, "percentage": 1.79, "elapsed_time": "0:01:46", "remaining_time": "1:37:07"}
{"current_steps": 80, "total_steps": 3358, "loss": 0.7659, "lr": 7.821782178217822e-06, "epoch": 0.023827252419955324, "percentage": 2.38, "elapsed_time": "0:02:15", "remaining_time": "1:32:37"}
{"current_steps": 100, "total_steps": 3358, "loss": 0.7285, "lr": 9.801980198019802e-06, "epoch": 0.029784065524944156, "percentage": 2.98, "elapsed_time": "0:02:44", "remaining_time": "1:29:29"}
{"current_steps": 120, "total_steps": 3358, "loss": 0.7129, "lr": 9.99924640442892e-06, "epoch": 0.035740878629932984, "percentage": 3.57, "elapsed_time": "0:03:14", "remaining_time": "1:27:15"}
{"current_steps": 140, "total_steps": 3358, "loss": 0.7024, "lr": 9.99664167434275e-06, "epoch": 0.04169769173492182, "percentage": 4.17, "elapsed_time": "0:03:43", "remaining_time": "1:25:41"}
{"current_steps": 160, "total_steps": 3358, "loss": 0.6906, "lr": 9.99217747519571e-06, "epoch": 0.04765450483991065, "percentage": 4.76, "elapsed_time": "0:04:11", "remaining_time": "1:23:52"}
{"current_steps": 180, "total_steps": 3358, "loss": 0.6919, "lr": 9.98585546831169e-06, "epoch": 0.05361131794489948, "percentage": 5.36, "elapsed_time": "0:04:41", "remaining_time": "1:22:43"}
{"current_steps": 200, "total_steps": 3358, "loss": 0.6717, "lr": 9.977678006386153e-06, "epoch": 0.05956813104988831, "percentage": 5.96, "elapsed_time": "0:05:10", "remaining_time": "1:21:45"}
{"current_steps": 220, "total_steps": 3358, "loss": 0.6739, "lr": 9.967648132610596e-06, "epoch": 0.06552494415487714, "percentage": 6.55, "elapsed_time": "0:05:38", "remaining_time": "1:20:29"}
{"current_steps": 240, "total_steps": 3358, "loss": 0.6725, "lr": 9.95576957954004e-06, "epoch": 0.07148175725986597, "percentage": 7.15, "elapsed_time": "0:06:07", "remaining_time": "1:19:34"}
{"current_steps": 260, "total_steps": 3358, "loss": 0.6693, "lr": 9.942046767703982e-06, "epoch": 0.0774385703648548, "percentage": 7.74, "elapsed_time": "0:06:37", "remaining_time": "1:18:55"}
{"current_steps": 280, "total_steps": 3358, "loss": 0.6639, "lr": 9.926484803961332e-06, "epoch": 0.08339538346984364, "percentage": 8.34, "elapsed_time": "0:07:05", "remaining_time": "1:17:53"}
{"current_steps": 300, "total_steps": 3358, "loss": 0.666, "lr": 9.909089479599916e-06, "epoch": 0.08935219657483247, "percentage": 8.93, "elapsed_time": "0:07:31", "remaining_time": "1:16:41"}
{"current_steps": 320, "total_steps": 3358, "loss": 0.6663, "lr": 9.88986726818129e-06, "epoch": 0.0953090096798213, "percentage": 9.53, "elapsed_time": "0:07:58", "remaining_time": "1:15:40"}
{"current_steps": 340, "total_steps": 3358, "loss": 0.653, "lr": 9.868825323131646e-06, "epoch": 0.10126582278481013, "percentage": 10.13, "elapsed_time": "0:08:26", "remaining_time": "1:14:52"}
{"current_steps": 360, "total_steps": 3358, "loss": 0.6531, "lr": 9.8459714750797e-06, "epoch": 0.10722263588979895, "percentage": 10.72, "elapsed_time": "0:08:52", "remaining_time": "1:13:55"}
{"current_steps": 380, "total_steps": 3358, "loss": 0.6632, "lr": 9.821314228942587e-06, "epoch": 0.11317944899478778, "percentage": 11.32, "elapsed_time": "0:09:19", "remaining_time": "1:13:04"}
{"current_steps": 400, "total_steps": 3358, "loss": 0.652, "lr": 9.794862760760798e-06, "epoch": 0.11913626209977662, "percentage": 11.91, "elapsed_time": "0:09:46", "remaining_time": "1:12:16"}
{"current_steps": 420, "total_steps": 3358, "loss": 0.6572, "lr": 9.76662691428337e-06, "epoch": 0.12509307520476545, "percentage": 12.51, "elapsed_time": "0:10:12", "remaining_time": "1:11:27"}
{"current_steps": 440, "total_steps": 3358, "loss": 0.6454, "lr": 9.736617197304615e-06, "epoch": 0.13104988830975428, "percentage": 13.1, "elapsed_time": "0:10:41", "remaining_time": "1:10:51"}
{"current_steps": 460, "total_steps": 3358, "loss": 0.6408, "lr": 9.704844777753686e-06, "epoch": 0.1370067014147431, "percentage": 13.7, "elapsed_time": "0:11:09", "remaining_time": "1:10:17"}
{"current_steps": 480, "total_steps": 3358, "loss": 0.6418, "lr": 9.671321479538522e-06, "epoch": 0.14296351451973194, "percentage": 14.29, "elapsed_time": "0:11:38", "remaining_time": "1:09:49"}
{"current_steps": 500, "total_steps": 3358, "loss": 0.6396, "lr": 9.63605977814564e-06, "epoch": 0.14892032762472077, "percentage": 14.89, "elapsed_time": "0:12:04", "remaining_time": "1:09:00"}
{"current_steps": 520, "total_steps": 3358, "loss": 0.6382, "lr": 9.59907279599747e-06, "epoch": 0.1548771407297096, "percentage": 15.49, "elapsed_time": "0:12:31", "remaining_time": "1:08:22"}
{"current_steps": 540, "total_steps": 3358, "loss": 0.6325, "lr": 9.56037429756892e-06, "epoch": 0.16083395383469842, "percentage": 16.08, "elapsed_time": "0:13:00", "remaining_time": "1:07:52"}
{"current_steps": 560, "total_steps": 3358, "loss": 0.645, "lr": 9.51997868426502e-06, "epoch": 0.16679076693968728, "percentage": 16.68, "elapsed_time": "0:13:28", "remaining_time": "1:07:21"}
{"current_steps": 580, "total_steps": 3358, "loss": 0.6403, "lr": 9.477900989061519e-06, "epoch": 0.1727475800446761, "percentage": 17.27, "elapsed_time": "0:13:54", "remaining_time": "1:06:38"}
{"current_steps": 600, "total_steps": 3358, "loss": 0.6362, "lr": 9.43415687091047e-06, "epoch": 0.17870439314966494, "percentage": 17.87, "elapsed_time": "0:14:21", "remaining_time": "1:06:01"}
{"current_steps": 620, "total_steps": 3358, "loss": 0.6321, "lr": 9.388762608912834e-06, "epoch": 0.18466120625465376, "percentage": 18.46, "elapsed_time": "0:14:48", "remaining_time": "1:05:23"}
{"current_steps": 640, "total_steps": 3358, "loss": 0.6386, "lr": 9.341735096260326e-06, "epoch": 0.1906180193596426, "percentage": 19.06, "elapsed_time": "0:15:16", "remaining_time": "1:04:50"}
{"current_steps": 660, "total_steps": 3358, "loss": 0.637, "lr": 9.293091833948709e-06, "epoch": 0.19657483246463142, "percentage": 19.65, "elapsed_time": "0:15:44", "remaining_time": "1:04:19"}
{"current_steps": 680, "total_steps": 3358, "loss": 0.6299, "lr": 9.242850924264918e-06, "epoch": 0.20253164556962025, "percentage": 20.25, "elapsed_time": "0:16:10", "remaining_time": "1:03:43"}
{"current_steps": 700, "total_steps": 3358, "loss": 0.6223, "lr": 9.191031064050396e-06, "epoch": 0.20848845867460908, "percentage": 20.85, "elapsed_time": "0:16:37", "remaining_time": "1:03:06"}
{"current_steps": 720, "total_steps": 3358, "loss": 0.6277, "lr": 9.137651537743193e-06, "epoch": 0.2144452717795979, "percentage": 21.44, "elapsed_time": "0:17:03", "remaining_time": "1:02:30"}
{"current_steps": 740, "total_steps": 3358, "loss": 0.6275, "lr": 9.082732210201371e-06, "epoch": 0.22040208488458674, "percentage": 22.04, "elapsed_time": "0:17:31", "remaining_time": "1:02:01"}
{"current_steps": 760, "total_steps": 3358, "loss": 0.6245, "lr": 9.026293519310426e-06, "epoch": 0.22635889798957556, "percentage": 22.63, "elapsed_time": "0:17:58", "remaining_time": "1:01:25"}
{"current_steps": 780, "total_steps": 3358, "loss": 0.6173, "lr": 8.968356468377454e-06, "epoch": 0.23231571109456442, "percentage": 23.23, "elapsed_time": "0:18:24", "remaining_time": "1:00:50"}
{"current_steps": 800, "total_steps": 3358, "loss": 0.6191, "lr": 8.908942618314912e-06, "epoch": 0.23827252419955325, "percentage": 23.82, "elapsed_time": "0:18:51", "remaining_time": "1:00:18"}
{"current_steps": 820, "total_steps": 3358, "loss": 0.619, "lr": 8.848074079616838e-06, "epoch": 0.24422933730454208, "percentage": 24.42, "elapsed_time": "0:19:18", "remaining_time": "0:59:45"}
{"current_steps": 840, "total_steps": 3358, "loss": 0.6167, "lr": 8.785773504130587e-06, "epoch": 0.2501861504095309, "percentage": 25.01, "elapsed_time": "0:19:45", "remaining_time": "0:59:13"}
{"current_steps": 860, "total_steps": 3358, "loss": 0.6248, "lr": 8.722064076627082e-06, "epoch": 0.2561429635145197, "percentage": 25.61, "elapsed_time": "0:20:11", "remaining_time": "0:58:38"}
{"current_steps": 880, "total_steps": 3358, "loss": 0.6176, "lr": 8.65696950617274e-06, "epoch": 0.26209977661950856, "percentage": 26.21, "elapsed_time": "0:20:37", "remaining_time": "0:58:04"}
{"current_steps": 900, "total_steps": 3358, "loss": 0.6251, "lr": 8.590514017306298e-06, "epoch": 0.2680565897244974, "percentage": 26.8, "elapsed_time": "0:21:04", "remaining_time": "0:57:33"}
{"current_steps": 920, "total_steps": 3358, "loss": 0.6138, "lr": 8.522722341023795e-06, "epoch": 0.2740134028294862, "percentage": 27.4, "elapsed_time": "0:21:31", "remaining_time": "0:57:02"}
{"current_steps": 940, "total_steps": 3358, "loss": 0.6099, "lr": 8.453619705575096e-06, "epoch": 0.2799702159344751, "percentage": 27.99, "elapsed_time": "0:21:57", "remaining_time": "0:56:29"}
{"current_steps": 960, "total_steps": 3358, "loss": 0.6222, "lr": 8.38323182707535e-06, "epoch": 0.2859270290394639, "percentage": 28.59, "elapsed_time": "0:22:26", "remaining_time": "0:56:03"}
{"current_steps": 980, "total_steps": 3358, "loss": 0.6091, "lr": 8.31158489993489e-06, "epoch": 0.29188384214445273, "percentage": 29.18, "elapsed_time": "0:22:54", "remaining_time": "0:55:34"}
{"current_steps": 1000, "total_steps": 3358, "loss": 0.6187, "lr": 8.238705587111162e-06, "epoch": 0.29784065524944153, "percentage": 29.78, "elapsed_time": "0:23:20", "remaining_time": "0:55:03"}
{"current_steps": 1020, "total_steps": 3358, "loss": 0.6127, "lr": 8.16462101018626e-06, "epoch": 0.3037974683544304, "percentage": 30.38, "elapsed_time": "0:23:47", "remaining_time": "0:54:32"}
{"current_steps": 1040, "total_steps": 3358, "loss": 0.6092, "lr": 8.089358739273806e-06, "epoch": 0.3097542814594192, "percentage": 30.97, "elapsed_time": "0:24:12", "remaining_time": "0:53:58"}
{"current_steps": 1060, "total_steps": 3358, "loss": 0.6083, "lr": 8.0129467827589e-06, "epoch": 0.31571109456440805, "percentage": 31.57, "elapsed_time": "0:24:40", "remaining_time": "0:53:28"}
{"current_steps": 1080, "total_steps": 3358, "loss": 0.6141, "lr": 7.935413576874966e-06, "epoch": 0.32166790766939685, "percentage": 32.16, "elapsed_time": "0:25:05", "remaining_time": "0:52:55"}
{"current_steps": 1100, "total_steps": 3358, "loss": 0.6145, "lr": 7.856787975121407e-06, "epoch": 0.3276247207743857, "percentage": 32.76, "elapsed_time": "0:25:33", "remaining_time": "0:52:27"}
{"current_steps": 1120, "total_steps": 3358, "loss": 0.602, "lr": 7.777099237525929e-06, "epoch": 0.33358153387937456, "percentage": 33.35, "elapsed_time": "0:26:00", "remaining_time": "0:51:57"}
{"current_steps": 1140, "total_steps": 3358, "loss": 0.6024, "lr": 7.696377019755624e-06, "epoch": 0.33953834698436336, "percentage": 33.95, "elapsed_time": "0:26:27", "remaining_time": "0:51:28"}
{"current_steps": 1160, "total_steps": 3358, "loss": 0.6061, "lr": 7.614651362080787e-06, "epoch": 0.3454951600893522, "percentage": 34.54, "elapsed_time": "0:26:54", "remaining_time": "0:50:58"}
{"current_steps": 1180, "total_steps": 3358, "loss": 0.6101, "lr": 7.531952678195621e-06, "epoch": 0.351451973194341, "percentage": 35.14, "elapsed_time": "0:27:20", "remaining_time": "0:50:28"}
{"current_steps": 1200, "total_steps": 3358, "loss": 0.617, "lr": 7.4483117438999685e-06, "epoch": 0.3574087862993299, "percentage": 35.74, "elapsed_time": "0:27:46", "remaining_time": "0:49:56"}
{"current_steps": 1220, "total_steps": 3358, "loss": 0.615, "lr": 7.3637596856462945e-06, "epoch": 0.3633655994043187, "percentage": 36.33, "elapsed_time": "0:28:12", "remaining_time": "0:49:25"}
{"current_steps": 1240, "total_steps": 3358, "loss": 0.602, "lr": 7.278327968956159e-06, "epoch": 0.36932241250930753, "percentage": 36.93, "elapsed_time": "0:28:39", "remaining_time": "0:48:56"}
{"current_steps": 1260, "total_steps": 3358, "loss": 0.6069, "lr": 7.192048386710537e-06, "epoch": 0.37527922561429633, "percentage": 37.52, "elapsed_time": "0:29:06", "remaining_time": "0:48:27"}
{"current_steps": 1280, "total_steps": 3358, "loss": 0.6134, "lr": 7.10495304731828e-06, "epoch": 0.3812360387192852, "percentage": 38.12, "elapsed_time": "0:29:33", "remaining_time": "0:47:59"}
{"current_steps": 1300, "total_steps": 3358, "loss": 0.6049, "lr": 7.017074362767176e-06, "epoch": 0.387192851824274, "percentage": 38.71, "elapsed_time": "0:29:59", "remaining_time": "0:47:29"}
{"current_steps": 1320, "total_steps": 3358, "loss": 0.6024, "lr": 6.9284450365620385e-06, "epoch": 0.39314966492926284, "percentage": 39.31, "elapsed_time": "0:30:25", "remaining_time": "0:46:58"}
{"current_steps": 1340, "total_steps": 3358, "loss": 0.6133, "lr": 6.83909805155429e-06, "epoch": 0.3991064780342517, "percentage": 39.9, "elapsed_time": "0:30:52", "remaining_time": "0:46:29"}
{"current_steps": 1360, "total_steps": 3358, "loss": 0.6051, "lr": 6.749066657667609e-06, "epoch": 0.4050632911392405, "percentage": 40.5, "elapsed_time": "0:31:19", "remaining_time": "0:46:00"}
{"current_steps": 1380, "total_steps": 3358, "loss": 0.5988, "lr": 6.658384359524166e-06, "epoch": 0.41102010424422936, "percentage": 41.1, "elapsed_time": "0:31:46", "remaining_time": "0:45:31"}
{"current_steps": 1400, "total_steps": 3358, "loss": 0.6025, "lr": 6.5670849039761e-06, "epoch": 0.41697691734921816, "percentage": 41.69, "elapsed_time": "0:32:12", "remaining_time": "0:45:02"}
{"current_steps": 1420, "total_steps": 3358, "loss": 0.6027, "lr": 6.475202267546829e-06, "epoch": 0.422933730454207, "percentage": 42.29, "elapsed_time": "0:32:38", "remaining_time": "0:44:32"}
{"current_steps": 1440, "total_steps": 3358, "loss": 0.596, "lr": 6.382770643786907e-06, "epoch": 0.4288905435591958, "percentage": 42.88, "elapsed_time": "0:33:05", "remaining_time": "0:44:04"}
{"current_steps": 1460, "total_steps": 3358, "loss": 0.6054, "lr": 6.289824430549092e-06, "epoch": 0.43484735666418467, "percentage": 43.48, "elapsed_time": "0:33:31", "remaining_time": "0:43:34"}
{"current_steps": 1480, "total_steps": 3358, "loss": 0.6024, "lr": 6.196398217187418e-06, "epoch": 0.44080416976917347, "percentage": 44.07, "elapsed_time": "0:33:56", "remaining_time": "0:43:04"}
{"current_steps": 1500, "total_steps": 3358, "loss": 0.5967, "lr": 6.102526771684971e-06, "epoch": 0.4467609828741623, "percentage": 44.67, "elapsed_time": "0:34:23", "remaining_time": "0:42:36"}
{"current_steps": 1520, "total_steps": 3358, "loss": 0.6049, "lr": 6.008245027715205e-06, "epoch": 0.4527177959791511, "percentage": 45.27, "elapsed_time": "0:34:49", "remaining_time": "0:42:06"}
{"current_steps": 1540, "total_steps": 3358, "loss": 0.6101, "lr": 5.9135880716415885e-06, "epoch": 0.45867460908414, "percentage": 45.86, "elapsed_time": "0:35:16", "remaining_time": "0:41:39"}
{"current_steps": 1560, "total_steps": 3358, "loss": 0.5935, "lr": 5.818591129460437e-06, "epoch": 0.46463142218912884, "percentage": 46.46, "elapsed_time": "0:35:43", "remaining_time": "0:41:10"}
{"current_steps": 1580, "total_steps": 3358, "loss": 0.5953, "lr": 5.723289553691782e-06, "epoch": 0.47058823529411764, "percentage": 47.05, "elapsed_time": "0:36:10", "remaining_time": "0:40:42"}
{"current_steps": 1600, "total_steps": 3358, "loss": 0.605, "lr": 5.6277188102231396e-06, "epoch": 0.4765450483991065, "percentage": 47.65, "elapsed_time": "0:36:37", "remaining_time": "0:40:14"}
{"current_steps": 1620, "total_steps": 3358, "loss": 0.6118, "lr": 5.531914465111109e-06, "epoch": 0.4825018615040953, "percentage": 48.24, "elapsed_time": "0:37:04", "remaining_time": "0:39:46"}
{"current_steps": 1640, "total_steps": 3358, "loss": 0.598, "lr": 5.435912171345691e-06, "epoch": 0.48845867460908415, "percentage": 48.84, "elapsed_time": "0:37:31", "remaining_time": "0:39:18"}
{"current_steps": 1660, "total_steps": 3358, "loss": 0.6006, "lr": 5.3397476555822365e-06, "epoch": 0.49441548771407295, "percentage": 49.43, "elapsed_time": "0:37:57", "remaining_time": "0:38:49"}
{"current_steps": 1680, "total_steps": 3358, "loss": 0.5962, "lr": 5.243456704846011e-06, "epoch": 0.5003723008190618, "percentage": 50.03, "elapsed_time": "0:38:23", "remaining_time": "0:38:20"}
{"current_steps": 1700, "total_steps": 3358, "loss": 0.6018, "lr": 5.147075153214256e-06, "epoch": 0.5063291139240507, "percentage": 50.63, "elapsed_time": "0:38:50", "remaining_time": "0:37:52"}
{"current_steps": 1720, "total_steps": 3358, "loss": 0.5993, "lr": 5.0506388684807805e-06, "epoch": 0.5122859270290394, "percentage": 51.22, "elapsed_time": "0:39:16", "remaining_time": "0:37:24"}
{"current_steps": 1740, "total_steps": 3358, "loss": 0.6042, "lr": 4.954183738807972e-06, "epoch": 0.5182427401340283, "percentage": 51.82, "elapsed_time": "0:39:43", "remaining_time": "0:36:56"}
{"current_steps": 1760, "total_steps": 3358, "loss": 0.5958, "lr": 4.857745659371249e-06, "epoch": 0.5241995532390171, "percentage": 52.41, "elapsed_time": "0:40:09", "remaining_time": "0:36:27"}
{"current_steps": 1780, "total_steps": 3358, "loss": 0.5927, "lr": 4.761360519000884e-06, "epoch": 0.530156366344006, "percentage": 53.01, "elapsed_time": "0:40:35", "remaining_time": "0:35:59"}
{"current_steps": 1800, "total_steps": 3358, "loss": 0.5931, "lr": 4.665064186826211e-06, "epoch": 0.5361131794489948, "percentage": 53.6, "elapsed_time": "0:41:02", "remaining_time": "0:35:31"}
{"current_steps": 1820, "total_steps": 3358, "loss": 0.5963, "lr": 4.568892498927148e-06, "epoch": 0.5420699925539836, "percentage": 54.2, "elapsed_time": "0:41:28", "remaining_time": "0:35:03"}
{"current_steps": 1840, "total_steps": 3358, "loss": 0.591, "lr": 4.47288124499802e-06, "epoch": 0.5480268056589724, "percentage": 54.79, "elapsed_time": "0:41:56", "remaining_time": "0:34:36"}
{"current_steps": 1860, "total_steps": 3358, "loss": 0.5891, "lr": 4.3770661550286485e-06, "epoch": 0.5539836187639613, "percentage": 55.39, "elapsed_time": "0:42:22", "remaining_time": "0:34:07"}
{"current_steps": 1880, "total_steps": 3358, "loss": 0.5981, "lr": 4.281482886007665e-06, "epoch": 0.5599404318689502, "percentage": 55.99, "elapsed_time": "0:42:49", "remaining_time": "0:33:39"}
{"current_steps": 1900, "total_steps": 3358, "loss": 0.5926, "lr": 4.186167008652975e-06, "epoch": 0.5658972449739389, "percentage": 56.58, "elapsed_time": "0:43:16", "remaining_time": "0:33:12"}
{"current_steps": 1920, "total_steps": 3358, "loss": 0.596, "lr": 4.091153994174343e-06, "epoch": 0.5718540580789278, "percentage": 57.18, "elapsed_time": "0:43:43", "remaining_time": "0:32:44"}
{"current_steps": 1940, "total_steps": 3358, "loss": 0.5881, "lr": 3.996479201073005e-06, "epoch": 0.5778108711839166, "percentage": 57.77, "elapsed_time": "0:44:09", "remaining_time": "0:32:16"}
{"current_steps": 1960, "total_steps": 3358, "loss": 0.5907, "lr": 3.902177861983213e-06, "epoch": 0.5837676842889055, "percentage": 58.37, "elapsed_time": "0:44:36", "remaining_time": "0:31:48"}
{"current_steps": 1980, "total_steps": 3358, "loss": 0.5956, "lr": 3.8082850705606387e-06, "epoch": 0.5897244973938943, "percentage": 58.96, "elapsed_time": "0:45:01", "remaining_time": "0:31:20"}
{"current_steps": 2000, "total_steps": 3358, "loss": 0.6, "lr": 3.7148357684224887e-06, "epoch": 0.5956813104988831, "percentage": 59.56, "elapsed_time": "0:45:27", "remaining_time": "0:30:51"}
{"current_steps": 2020, "total_steps": 3358, "loss": 0.5914, "lr": 3.6218647321441934e-06, "epoch": 0.6016381236038719, "percentage": 60.15, "elapsed_time": "0:45:53", "remaining_time": "0:30:23"}
{"current_steps": 2040, "total_steps": 3358, "loss": 0.5937, "lr": 3.529406560317521e-06, "epoch": 0.6075949367088608, "percentage": 60.75, "elapsed_time": "0:46:19", "remaining_time": "0:29:55"}
{"current_steps": 2060, "total_steps": 3358, "loss": 0.5899, "lr": 3.4374956606749337e-06, "epoch": 0.6135517498138496, "percentage": 61.35, "elapsed_time": "0:46:45", "remaining_time": "0:29:27"}
{"current_steps": 2080, "total_steps": 3358, "loss": 0.585, "lr": 3.3461662372849635e-06, "epoch": 0.6195085629188384, "percentage": 61.94, "elapsed_time": "0:47:11", "remaining_time": "0:28:59"}
{"current_steps": 2100, "total_steps": 3358, "loss": 0.597, "lr": 3.25545227782337e-06, "epoch": 0.6254653760238272, "percentage": 62.54, "elapsed_time": "0:47:37", "remaining_time": "0:28:31"}
{"current_steps": 2120, "total_steps": 3358, "loss": 0.5886, "lr": 3.1653875409248557e-06, "epoch": 0.6314221891288161, "percentage": 63.13, "elapsed_time": "0:48:03", "remaining_time": "0:28:03"}
{"current_steps": 2140, "total_steps": 3358, "loss": 0.589, "lr": 3.07600554361999e-06, "epoch": 0.637379002233805, "percentage": 63.73, "elapsed_time": "0:48:29", "remaining_time": "0:27:35"}
{"current_steps": 2160, "total_steps": 3358, "loss": 0.5937, "lr": 2.987339548862054e-06, "epoch": 0.6433358153387937, "percentage": 64.32, "elapsed_time": "0:48:55", "remaining_time": "0:27:08"}
{"current_steps": 2180, "total_steps": 3358, "loss": 0.5929, "lr": 2.8994225531484456e-06, "epoch": 0.6492926284437825, "percentage": 64.92, "elapsed_time": "0:49:22", "remaining_time": "0:26:40"}
{"current_steps": 2200, "total_steps": 3358, "loss": 0.588, "lr": 2.812287274241238e-06, "epoch": 0.6552494415487714, "percentage": 65.52, "elapsed_time": "0:49:48", "remaining_time": "0:26:13"}
{"current_steps": 2220, "total_steps": 3358, "loss": 0.5917, "lr": 2.7259661389914597e-06, "epoch": 0.6612062546537603, "percentage": 66.11, "elapsed_time": "0:50:15", "remaining_time": "0:25:45"}
{"current_steps": 2240, "total_steps": 3358, "loss": 0.5922, "lr": 2.6404912712716585e-06, "epoch": 0.6671630677587491, "percentage": 66.71, "elapsed_time": "0:50:41", "remaining_time": "0:25:18"}
{"current_steps": 2260, "total_steps": 3358, "loss": 0.5918, "lr": 2.5558944800211892e-06, "epoch": 0.6731198808637379, "percentage": 67.3, "elapsed_time": "0:51:07", "remaining_time": "0:24:50"}
{"current_steps": 2280, "total_steps": 3358, "loss": 0.5956, "lr": 2.472207247408727e-06, "epoch": 0.6790766939687267, "percentage": 67.9, "elapsed_time": "0:51:33", "remaining_time": "0:24:22"}
{"current_steps": 2300, "total_steps": 3358, "loss": 0.5879, "lr": 2.389460717116358e-06, "epoch": 0.6850335070737156, "percentage": 68.49, "elapsed_time": "0:51:59", "remaining_time": "0:23:55"}
{"current_steps": 2320, "total_steps": 3358, "loss": 0.5875, "lr": 2.30768568274967e-06, "epoch": 0.6909903201787044, "percentage": 69.09, "elapsed_time": "0:52:26", "remaining_time": "0:23:27"}
{"current_steps": 2340, "total_steps": 3358, "loss": 0.5828, "lr": 2.2269125763780974e-06, "epoch": 0.6969471332836932, "percentage": 69.68, "elapsed_time": "0:52:53", "remaining_time": "0:23:00"}
{"current_steps": 2360, "total_steps": 3358, "loss": 0.5838, "lr": 2.147171457209807e-06, "epoch": 0.702903946388682, "percentage": 70.28, "elapsed_time": "0:53:20", "remaining_time": "0:22:33"}
{"current_steps": 2380, "total_steps": 3358, "loss": 0.5831, "lr": 2.0684920004053666e-06, "epoch": 0.7088607594936709, "percentage": 70.88, "elapsed_time": "0:53:46", "remaining_time": "0:22:05"}
{"current_steps": 2400, "total_steps": 3358, "loss": 0.5904, "lr": 1.990903486034308e-06, "epoch": 0.7148175725986597, "percentage": 71.47, "elapsed_time": "0:54:13", "remaining_time": "0:21:38"}
{"current_steps": 2420, "total_steps": 3358, "loss": 0.589, "lr": 1.914434788178737e-06, "epoch": 0.7207743857036486, "percentage": 72.07, "elapsed_time": "0:54:38", "remaining_time": "0:21:10"}
{"current_steps": 2440, "total_steps": 3358, "loss": 0.5804, "lr": 1.8391143641880188e-06, "epoch": 0.7267311988086373, "percentage": 72.66, "elapsed_time": "0:55:05", "remaining_time": "0:20:43"}
{"current_steps": 2460, "total_steps": 3358, "loss": 0.5899, "lr": 1.7649702440885597e-06, "epoch": 0.7326880119136262, "percentage": 73.26, "elapsed_time": "0:55:31", "remaining_time": "0:20:16"}
{"current_steps": 2480, "total_steps": 3358, "loss": 0.592, "lr": 1.692030020152598e-06, "epoch": 0.7386448250186151, "percentage": 73.85, "elapsed_time": "0:55:57", "remaining_time": "0:19:48"}
{"current_steps": 2500, "total_steps": 3358, "loss": 0.5954, "lr": 1.6203208366299216e-06, "epoch": 0.7446016381236039, "percentage": 74.45, "elapsed_time": "0:56:24", "remaining_time": "0:19:21"}
{"current_steps": 2520, "total_steps": 3358, "loss": 0.5907, "lr": 1.5498693796462994e-06, "epoch": 0.7505584512285927, "percentage": 75.04, "elapsed_time": "0:56:50", "remaining_time": "0:18:53"}
{"current_steps": 2540, "total_steps": 3358, "loss": 0.5824, "lr": 1.4807018672724066e-06, "epoch": 0.7565152643335815, "percentage": 75.64, "elapsed_time": "0:57:16", "remaining_time": "0:18:26"}
{"current_steps": 2560, "total_steps": 3358, "loss": 0.5831, "lr": 1.4128440397669301e-06, "epoch": 0.7624720774385704, "percentage": 76.24, "elapsed_time": "0:57:44", "remaining_time": "0:17:59"}
{"current_steps": 2580, "total_steps": 3358, "loss": 0.5857, "lr": 1.3463211499974993e-06, "epoch": 0.7684288905435592, "percentage": 76.83, "elapsed_time": "0:58:10", "remaining_time": "0:17:32"}
{"current_steps": 2600, "total_steps": 3358, "loss": 0.59, "lr": 1.2811579540429836e-06, "epoch": 0.774385703648548, "percentage": 77.43, "elapsed_time": "0:58:36", "remaining_time": "0:17:05"}
{"current_steps": 2620, "total_steps": 3358, "loss": 0.585, "lr": 1.2173787019806688e-06, "epoch": 0.7803425167535368, "percentage": 78.02, "elapsed_time": "0:59:03", "remaining_time": "0:16:38"}
{"current_steps": 2640, "total_steps": 3358, "loss": 0.5866, "lr": 1.1550071288617486e-06, "epoch": 0.7862993298585257, "percentage": 78.62, "elapsed_time": "0:59:29", "remaining_time": "0:16:10"}
{"current_steps": 2660, "total_steps": 3358, "loss": 0.584, "lr": 1.0940664458784733e-06, "epoch": 0.7922561429635145, "percentage": 79.21, "elapsed_time": "0:59:55", "remaining_time": "0:15:43"}
{"current_steps": 2680, "total_steps": 3358, "loss": 0.5825, "lr": 1.0345793317262387e-06, "epoch": 0.7982129560685034, "percentage": 79.81, "elapsed_time": "1:00:22", "remaining_time": "0:15:16"}
{"current_steps": 2700, "total_steps": 3358, "loss": 0.5936, "lr": 9.765679241638649e-07, "epoch": 0.8041697691734921, "percentage": 80.41, "elapsed_time": "1:00:48", "remaining_time": "0:14:49"}
{"current_steps": 2720, "total_steps": 3358, "loss": 0.5944, "lr": 9.200538117751595e-07, "epoch": 0.810126582278481, "percentage": 81.0, "elapsed_time": "1:01:15", "remaining_time": "0:14:22"}
{"current_steps": 2740, "total_steps": 3358, "loss": 0.5942, "lr": 8.650580259348595e-07, "epoch": 0.8160833953834699, "percentage": 81.6, "elapsed_time": "1:01:40", "remaining_time": "0:13:54"}
{"current_steps": 2760, "total_steps": 3358, "loss": 0.5903, "lr": 8.116010329819418e-07, "epoch": 0.8220402084884587, "percentage": 82.19, "elapsed_time": "1:02:07", "remaining_time": "0:13:27"}
{"current_steps": 2780, "total_steps": 3358, "loss": 0.5855, "lr": 7.597027266031992e-07, "epoch": 0.8279970215934475, "percentage": 82.79, "elapsed_time": "1:02:33", "remaining_time": "0:13:00"}
{"current_steps": 2800, "total_steps": 3358, "loss": 0.5859, "lr": 7.093824204299299e-07, "epoch": 0.8339538346984363, "percentage": 83.38, "elapsed_time": "1:03:00", "remaining_time": "0:12:33"}
{"current_steps": 2820, "total_steps": 3358, "loss": 0.5997, "lr": 6.60658840850486e-07, "epoch": 0.8399106478034252, "percentage": 83.98, "elapsed_time": "1:03:27", "remaining_time": "0:12:06"}
{"current_steps": 2840, "total_steps": 3358, "loss": 0.5847, "lr": 6.135501200413724e-07, "epoch": 0.845867460908414, "percentage": 84.57, "elapsed_time": "1:03:53", "remaining_time": "0:11:39"}
{"current_steps": 2860, "total_steps": 3358, "loss": 0.5893, "lr": 5.680737892194654e-07, "epoch": 0.8518242740134029, "percentage": 85.17, "elapsed_time": "1:04:20", "remaining_time": "0:11:12"}
{"current_steps": 2880, "total_steps": 3358, "loss": 0.5892, "lr": 5.242467721178774e-07, "epoch": 0.8577810871183916, "percentage": 85.77, "elapsed_time": "1:04:45", "remaining_time": "0:10:44"}
{"current_steps": 2900, "total_steps": 3358, "loss": 0.5821, "lr": 4.82085378687901e-07, "epoch": 0.8637379002233805, "percentage": 86.36, "elapsed_time": "1:05:13", "remaining_time": "0:10:18"}
{"current_steps": 2920, "total_steps": 3358, "loss": 0.5824, "lr": 4.416052990293596e-07, "epoch": 0.8696947133283693, "percentage": 86.96, "elapsed_time": "1:05:41", "remaining_time": "0:09:51"}
{"current_steps": 2940, "total_steps": 3358, "loss": 0.5886, "lr": 4.028215975516314e-07, "epoch": 0.8756515264333582, "percentage": 87.55, "elapsed_time": "1:06:07", "remaining_time": "0:09:24"}
{"current_steps": 2960, "total_steps": 3358, "loss": 0.5974, "lr": 3.657487073675281e-07, "epoch": 0.8816083395383469, "percentage": 88.15, "elapsed_time": "1:06:33", "remaining_time": "0:08:56"}
{"current_steps": 2980, "total_steps": 3358, "loss": 0.5817, "lr": 3.3040042492210103e-07, "epoch": 0.8875651526433358, "percentage": 88.74, "elapsed_time": "1:06:58", "remaining_time": "0:08:29"}
{"current_steps": 3000, "total_steps": 3358, "loss": 0.5987, "lr": 2.9678990485837553e-07, "epoch": 0.8935219657483247, "percentage": 89.34, "elapsed_time": "1:07:24", "remaining_time": "0:08:02"}
{"current_steps": 3020, "total_steps": 3358, "loss": 0.5919, "lr": 2.6492965512193934e-07, "epoch": 0.8994787788533135, "percentage": 89.93, "elapsed_time": "1:07:51", "remaining_time": "0:07:35"}
{"current_steps": 3040, "total_steps": 3358, "loss": 0.6026, "lr": 2.348315323061845e-07, "epoch": 0.9054355919583023, "percentage": 90.53, "elapsed_time": "1:08:18", "remaining_time": "0:07:08"}
{"current_steps": 3060, "total_steps": 3358, "loss": 0.5784, "lr": 2.0650673723995097e-07, "epoch": 0.9113924050632911, "percentage": 91.13, "elapsed_time": "1:08:44", "remaining_time": "0:06:41"}
{"current_steps": 3080, "total_steps": 3358, "loss": 0.5834, "lr": 1.7996581081920483e-07, "epoch": 0.91734921816828, "percentage": 91.72, "elapsed_time": "1:09:10", "remaining_time": "0:06:14"}
{"current_steps": 3100, "total_steps": 3358, "loss": 0.5887, "lr": 1.5521863008431725e-07, "epoch": 0.9233060312732688, "percentage": 92.32, "elapsed_time": "1:09:37", "remaining_time": "0:05:47"}
{"current_steps": 3120, "total_steps": 3358, "loss": 0.5851, "lr": 1.3227440454437745e-07, "epoch": 0.9292628443782577, "percentage": 92.91, "elapsed_time": "1:10:03", "remaining_time": "0:05:20"}
{"current_steps": 3140, "total_steps": 3358, "loss": 0.6009, "lr": 1.1114167274993537e-07, "epoch": 0.9352196574832464, "percentage": 93.51, "elapsed_time": "1:10:30", "remaining_time": "0:04:53"}
{"current_steps": 3160, "total_steps": 3358, "loss": 0.5902, "lr": 9.1828299115434e-08, "epoch": 0.9411764705882353, "percentage": 94.1, "elapsed_time": "1:10:55", "remaining_time": "0:04:26"}
{"current_steps": 3180, "total_steps": 3358, "loss": 0.5862, "lr": 7.434147099251165e-08, "epoch": 0.9471332836932241, "percentage": 94.7, "elapsed_time": "1:11:21", "remaining_time": "0:03:59"}
{"current_steps": 3200, "total_steps": 3358, "loss": 0.5837, "lr": 5.86876959952809e-08, "epoch": 0.953090096798213, "percentage": 95.29, "elapsed_time": "1:11:48", "remaining_time": "0:03:32"}
{"current_steps": 3220, "total_steps": 3358, "loss": 0.587, "lr": 4.4872799578554724e-08, "epoch": 0.9590469099032017, "percentage": 95.89, "elapsed_time": "1:12:14", "remaining_time": "0:03:05"}
{"current_steps": 3240, "total_steps": 3358, "loss": 0.591, "lr": 3.290192286994287e-08, "epoch": 0.9650037230081906, "percentage": 96.49, "elapsed_time": "1:12:40", "remaining_time": "0:02:38"}
{"current_steps": 3260, "total_steps": 3358, "loss": 0.5793, "lr": 2.2779520756611163e-08, "epoch": 0.9709605361131795, "percentage": 97.08, "elapsed_time": "1:13:06", "remaining_time": "0:02:11"}
{"current_steps": 3280, "total_steps": 3358, "loss": 0.5977, "lr": 1.450936022742544e-08, "epoch": 0.9769173492181683, "percentage": 97.68, "elapsed_time": "1:13:32", "remaining_time": "0:01:44"}
{"current_steps": 3300, "total_steps": 3358, "loss": 0.5908, "lr": 8.094518971088484e-09, "epoch": 0.9828741623231572, "percentage": 98.27, "elapsed_time": "1:13:57", "remaining_time": "0:01:17"}
{"current_steps": 3320, "total_steps": 3358, "loss": 0.5854, "lr": 3.537384230797303e-09, "epoch": 0.9888309754281459, "percentage": 98.87, "elapsed_time": "1:14:24", "remaining_time": "0:00:51"}
{"current_steps": 3340, "total_steps": 3358, "loss": 0.5844, "lr": 8.396519158465621e-10, "epoch": 0.9947877885331348, "percentage": 99.46, "elapsed_time": "1:14:51", "remaining_time": "0:00:24"}
{"current_steps": 3358, "total_steps": 3358, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "1:15:21", "remaining_time": "0:00:00"}