| { | |
| "best_metric": 0.7818750739097595, | |
| "best_model_checkpoint": "./zhko_mbartLarge_100p_run1/checkpoint-208330", | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 416661, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.3855236632427764e-06, | |
| "loss": 2.8874, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.785446865700298e-06, | |
| "loss": 2.0531, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.185370068157819e-06, | |
| "loss": 1.9116, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.585293270615341e-06, | |
| "loss": 1.8191, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.1985216473072862e-05, | |
| "loss": 1.7481, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.4385139675530384e-05, | |
| "loss": 1.6872, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.6785062877987908e-05, | |
| "loss": 1.6479, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9184986080445428e-05, | |
| "loss": 1.6213, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.158490928290295e-05, | |
| "loss": 1.581, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.3980032638955555e-05, | |
| "loss": 1.56, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.637995584141308e-05, | |
| "loss": 1.5333, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.877507919746568e-05, | |
| "loss": 1.513, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.11750023999232e-05, | |
| "loss": 1.5147, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.3574925602380726e-05, | |
| "loss": 1.5006, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.597484880483824e-05, | |
| "loss": 1.4869, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8374772007295767e-05, | |
| "loss": 1.4786, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.077469520975329e-05, | |
| "loss": 1.4796, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.317461841221081e-05, | |
| "loss": 1.4451, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.557454161466833e-05, | |
| "loss": 1.4554, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7974464817125855e-05, | |
| "loss": 1.4498, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9996315072903376e-05, | |
| "loss": 1.4393, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.997212061736105e-05, | |
| "loss": 1.4227, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.994787767593588e-05, | |
| "loss": 1.427, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.99236347345107e-05, | |
| "loss": 1.4113, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.989939179308553e-05, | |
| "loss": 1.3914, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9875148851660354e-05, | |
| "loss": 1.3769, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.985095439611803e-05, | |
| "loss": 1.378, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9826759940575705e-05, | |
| "loss": 1.3477, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.980251699915053e-05, | |
| "loss": 1.3498, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9778274057725356e-05, | |
| "loss": 1.3406, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.975403111630018e-05, | |
| "loss": 1.3279, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.972983666075786e-05, | |
| "loss": 1.3323, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.970564220521553e-05, | |
| "loss": 1.3034, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.968139926379036e-05, | |
| "loss": 1.3118, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9657156322365184e-05, | |
| "loss": 1.2919, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.963296186682286e-05, | |
| "loss": 1.3006, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.9608718925397685e-05, | |
| "loss": 1.2771, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.958447598397251e-05, | |
| "loss": 1.2859, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.956023304254733e-05, | |
| "loss": 1.2668, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.953599010112216e-05, | |
| "loss": 1.2662, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.951179564557983e-05, | |
| "loss": 1.2618, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.948755270415466e-05, | |
| "loss": 1.2647, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.946330976272949e-05, | |
| "loss": 1.27, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9439066821304314e-05, | |
| "loss": 1.2298, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.941482387987913e-05, | |
| "loss": 1.2255, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9390580938453965e-05, | |
| "loss": 1.2062, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.936633799702879e-05, | |
| "loss": 1.2167, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.9342095055603616e-05, | |
| "loss": 1.2009, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.9317852114178435e-05, | |
| "loss": 1.2182, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.929360917275326e-05, | |
| "loss": 1.1898, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.926936623132809e-05, | |
| "loss": 1.2008, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.924512328990292e-05, | |
| "loss": 1.2015, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.922088034847774e-05, | |
| "loss": 1.1932, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.919663740705256e-05, | |
| "loss": 1.189, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.9172394465627395e-05, | |
| "loss": 1.1872, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.914815152420222e-05, | |
| "loss": 1.189, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.9124054040425596e-05, | |
| "loss": 1.1778, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.9099859584883265e-05, | |
| "loss": 1.1891, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.90756166434581e-05, | |
| "loss": 1.1887, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.905137370203292e-05, | |
| "loss": 1.1741, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.902713076060774e-05, | |
| "loss": 1.1698, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.900288781918257e-05, | |
| "loss": 1.1662, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.897864487775739e-05, | |
| "loss": 1.1631, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8954401936332225e-05, | |
| "loss": 1.1545, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.8930158994907044e-05, | |
| "loss": 1.1479, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8905964539364727e-05, | |
| "loss": 1.1533, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8881721597939545e-05, | |
| "loss": 1.1437, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.885747865651437e-05, | |
| "loss": 1.1414, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.8833235715089196e-05, | |
| "loss": 1.1423, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.880899277366402e-05, | |
| "loss": 1.1471, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.878474983223885e-05, | |
| "loss": 1.1345, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.876050689081367e-05, | |
| "loss": 1.1356, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.87362639493885e-05, | |
| "loss": 1.1395, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.8712021007963324e-05, | |
| "loss": 1.1184, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.868777806653815e-05, | |
| "loss": 1.1306, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.8663535125112975e-05, | |
| "loss": 1.1185, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.86392921836878e-05, | |
| "loss": 1.1229, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.8615146214028326e-05, | |
| "loss": 1.1162, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.859090327260315e-05, | |
| "loss": 1.1163, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.856666033117798e-05, | |
| "loss": 1.1314, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.85424173897528e-05, | |
| "loss": 1.1227, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.851817444832762e-05, | |
| "loss": 1.1236, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.8493931506902454e-05, | |
| "loss": 1.117, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.846968856547728e-05, | |
| "loss": 1.1063, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.8445445624052105e-05, | |
| "loss": 1.1144, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.8421202682626924e-05, | |
| "loss": 1.1052, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8396959741201756e-05, | |
| "loss": 1.1203, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8372765285659425e-05, | |
| "loss": 1.0938, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.834852234423425e-05, | |
| "loss": 1.0856, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.832427940280908e-05, | |
| "loss": 1.0768, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.830003646138391e-05, | |
| "loss": 1.0968, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.827579351995873e-05, | |
| "loss": 1.1018, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.825155057853355e-05, | |
| "loss": 1.0795, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.8227307637108385e-05, | |
| "loss": 1.0894, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.820306469568321e-05, | |
| "loss": 1.0946, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.817882175425803e-05, | |
| "loss": 1.0842, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.8154578812832855e-05, | |
| "loss": 1.0779, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.813033587140768e-05, | |
| "loss": 1.0849, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.810609292998251e-05, | |
| "loss": 1.0757, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.808184998855733e-05, | |
| "loss": 1.068, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.8057655533015014e-05, | |
| "loss": 1.0793, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.803350956335553e-05, | |
| "loss": 1.0865, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.8009315107813215e-05, | |
| "loss": 1.0698, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.7985072166388034e-05, | |
| "loss": 1.0703, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.796082922496286e-05, | |
| "loss": 1.0692, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.7936586283537685e-05, | |
| "loss": 1.0591, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.791234334211252e-05, | |
| "loss": 1.0642, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.7888100400687336e-05, | |
| "loss": 1.0771, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.786385745926216e-05, | |
| "loss": 1.0802, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.783961451783699e-05, | |
| "loss": 1.0765, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.781537157641181e-05, | |
| "loss": 1.058, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.779112863498664e-05, | |
| "loss": 1.0521, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.7766934179444314e-05, | |
| "loss": 1.053, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.774269123801914e-05, | |
| "loss": 1.0517, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.7718448296593965e-05, | |
| "loss": 1.0613, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.769420535516879e-05, | |
| "loss": 1.0551, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.7669962413743616e-05, | |
| "loss": 1.0462, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.764576795820129e-05, | |
| "loss": 1.0496, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.762152501677612e-05, | |
| "loss": 1.0407, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.759728207535094e-05, | |
| "loss": 1.06, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.757303913392577e-05, | |
| "loss": 1.042, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.7548796192500594e-05, | |
| "loss": 1.0401, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.752455325107542e-05, | |
| "loss": 1.036, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.7500310309650245e-05, | |
| "loss": 1.0461, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.747606736822507e-05, | |
| "loss": 1.0457, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.7451824426799896e-05, | |
| "loss": 1.0325, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.742762997125757e-05, | |
| "loss": 1.0353, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.74033870298324e-05, | |
| "loss": 1.0417, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.737914408840722e-05, | |
| "loss": 1.0486, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.735490114698204e-05, | |
| "loss": 1.0366, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.7330658205556874e-05, | |
| "loss": 1.0195, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.73064152641317e-05, | |
| "loss": 1.0377, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.7282172322706525e-05, | |
| "loss": 1.0373, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.72579778671642e-05, | |
| "loss": 1.0247, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.7233734925739026e-05, | |
| "loss": 1.021, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.7209491984313845e-05, | |
| "loss": 1.0328, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.718524904288868e-05, | |
| "loss": 1.0113, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.71610061014635e-05, | |
| "loss": 1.0208, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 41.5541, | |
| "eval_gen_len": 13.044, | |
| "eval_loss": 0.8814196586608887, | |
| "eval_runtime": 8677.1132, | |
| "eval_samples_per_second": 15.503, | |
| "eval_steps_per_second": 1.938, | |
| "step": 69443 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.713676316003833e-05, | |
| "loss": 1.0263, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.711252021861315e-05, | |
| "loss": 0.8753, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.708837424895367e-05, | |
| "loss": 0.883, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7064131307528505e-05, | |
| "loss": 0.867, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.703988836610333e-05, | |
| "loss": 0.8803, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.701564542467815e-05, | |
| "loss": 0.8856, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.6991402483252975e-05, | |
| "loss": 0.88, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.696720802771065e-05, | |
| "loss": 0.8847, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.6942965086285476e-05, | |
| "loss": 0.8794, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.691872214486031e-05, | |
| "loss": 0.8884, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.6894479203435134e-05, | |
| "loss": 0.877, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.687023626200995e-05, | |
| "loss": 0.8921, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.684599332058478e-05, | |
| "loss": 0.8823, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.6821750379159604e-05, | |
| "loss": 0.8741, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.6797507437734436e-05, | |
| "loss": 0.8902, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.6773264496309255e-05, | |
| "loss": 0.8735, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.674902155488408e-05, | |
| "loss": 0.889, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.6724778613458906e-05, | |
| "loss": 0.8863, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.670058415791658e-05, | |
| "loss": 0.8907, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.667634121649141e-05, | |
| "loss": 0.8783, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.665209827506623e-05, | |
| "loss": 0.8888, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.662790381952391e-05, | |
| "loss": 0.8844, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.6603660878098734e-05, | |
| "loss": 0.8711, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.657941793667356e-05, | |
| "loss": 0.8942, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.6555174995248385e-05, | |
| "loss": 0.888, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.653093205382321e-05, | |
| "loss": 0.8906, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.6506689112398036e-05, | |
| "loss": 0.8909, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.648244617097286e-05, | |
| "loss": 0.8837, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.645820322954769e-05, | |
| "loss": 0.8836, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.643396028812251e-05, | |
| "loss": 0.8819, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.640971734669734e-05, | |
| "loss": 0.8995, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.6385474405272164e-05, | |
| "loss": 0.8934, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.636123146384699e-05, | |
| "loss": 0.8836, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.6337037008304665e-05, | |
| "loss": 0.8941, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.631279406687949e-05, | |
| "loss": 0.8934, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.6288551125454316e-05, | |
| "loss": 0.8988, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.626430818402914e-05, | |
| "loss": 0.8805, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.624006524260397e-05, | |
| "loss": 0.8931, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.621582230117879e-05, | |
| "loss": 0.8956, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.619157935975362e-05, | |
| "loss": 0.8872, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.6167336418328444e-05, | |
| "loss": 0.8945, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.614314196278612e-05, | |
| "loss": 0.8919, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.6118899021360945e-05, | |
| "loss": 0.8801, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.6094656079935764e-05, | |
| "loss": 0.883, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.6070413138510596e-05, | |
| "loss": 0.888, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.6046218682968265e-05, | |
| "loss": 0.8878, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.60219757415431e-05, | |
| "loss": 0.8809, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.599773280011792e-05, | |
| "loss": 0.883, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.597348985869275e-05, | |
| "loss": 0.8769, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.594924691726757e-05, | |
| "loss": 0.8799, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.592500397584239e-05, | |
| "loss": 0.8907, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.5900761034417225e-05, | |
| "loss": 0.8813, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.587651809299205e-05, | |
| "loss": 0.8712, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.585227515156687e-05, | |
| "loss": 0.8903, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.5828032210141695e-05, | |
| "loss": 0.8948, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.580378926871653e-05, | |
| "loss": 0.8758, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.5779691784939896e-05, | |
| "loss": 0.8771, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.575544884351473e-05, | |
| "loss": 0.8867, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.5731205902089554e-05, | |
| "loss": 0.8853, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.570696296066437e-05, | |
| "loss": 0.8892, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.5682768505122055e-05, | |
| "loss": 0.8932, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.5658525563696874e-05, | |
| "loss": 0.8909, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.5634331108154556e-05, | |
| "loss": 0.9005, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.5610088166729375e-05, | |
| "loss": 0.8884, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.55858452253042e-05, | |
| "loss": 0.8807, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.5561602283879026e-05, | |
| "loss": 0.8842, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.553735934245386e-05, | |
| "loss": 0.8888, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.551311640102868e-05, | |
| "loss": 0.884, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.54888734596035e-05, | |
| "loss": 0.8828, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.546463051817833e-05, | |
| "loss": 0.893, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.5440387576753154e-05, | |
| "loss": 0.8732, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.541619312121083e-05, | |
| "loss": 0.8681, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.5391950179785655e-05, | |
| "loss": 0.874, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.536770723836048e-05, | |
| "loss": 0.8795, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.5343464296935306e-05, | |
| "loss": 0.8849, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.531922135551013e-05, | |
| "loss": 0.8806, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.529502689996781e-05, | |
| "loss": 0.8967, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.527078395854263e-05, | |
| "loss": 0.8782, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.524654101711746e-05, | |
| "loss": 0.8813, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.5222298075692284e-05, | |
| "loss": 0.8801, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.519805513426711e-05, | |
| "loss": 0.8816, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.5173812192841935e-05, | |
| "loss": 0.8842, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.514956925141676e-05, | |
| "loss": 0.8843, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.5125326309991586e-05, | |
| "loss": 0.896, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.510108336856641e-05, | |
| "loss": 0.8911, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.507684042714124e-05, | |
| "loss": 0.8879, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.505269445748176e-05, | |
| "loss": 0.8897, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.502845151605659e-05, | |
| "loss": 0.8814, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.5004257060514264e-05, | |
| "loss": 0.8737, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.498001411908909e-05, | |
| "loss": 0.8884, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.4955771177663915e-05, | |
| "loss": 0.8796, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.493152823623874e-05, | |
| "loss": 0.8753, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.4907285294813566e-05, | |
| "loss": 0.8796, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.4883042353388385e-05, | |
| "loss": 0.8846, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.485879941196322e-05, | |
| "loss": 0.8676, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.483455647053804e-05, | |
| "loss": 0.8861, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.481031352911287e-05, | |
| "loss": 0.8822, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.478607058768769e-05, | |
| "loss": 0.8665, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.476182764626252e-05, | |
| "loss": 0.8903, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.4737584704837345e-05, | |
| "loss": 0.8798, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.4713341763412164e-05, | |
| "loss": 0.8828, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.468909882198699e-05, | |
| "loss": 0.8782, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.4664855880561815e-05, | |
| "loss": 0.8732, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.464066142501949e-05, | |
| "loss": 0.8836, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.4616418483594316e-05, | |
| "loss": 0.8874, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.459217554216915e-05, | |
| "loss": 0.8798, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.456798108662682e-05, | |
| "loss": 0.8641, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.454373814520165e-05, | |
| "loss": 0.8652, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.451949520377647e-05, | |
| "loss": 0.8847, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.4495252262351294e-05, | |
| "loss": 0.8695, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.447100932092612e-05, | |
| "loss": 0.8761, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.4446766379500945e-05, | |
| "loss": 0.8881, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.442252343807577e-05, | |
| "loss": 0.8588, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.4398280496650596e-05, | |
| "loss": 0.8938, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.437403755522542e-05, | |
| "loss": 0.8678, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.434979461380025e-05, | |
| "loss": 0.8745, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.432555167237507e-05, | |
| "loss": 0.8833, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.43013087309499e-05, | |
| "loss": 0.8794, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.4277065789524724e-05, | |
| "loss": 0.8794, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.425282284809955e-05, | |
| "loss": 0.8747, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.4228579906674375e-05, | |
| "loss": 0.8746, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.42043369652492e-05, | |
| "loss": 0.8638, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.4180094023824026e-05, | |
| "loss": 0.8747, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.41558995682817e-05, | |
| "loss": 0.8934, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.413165662685653e-05, | |
| "loss": 0.8816, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.410741368543135e-05, | |
| "loss": 0.8681, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.408317074400618e-05, | |
| "loss": 0.871, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.4058927802581004e-05, | |
| "loss": 0.8615, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.403473334703867e-05, | |
| "loss": 0.8795, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.4010538891496355e-05, | |
| "loss": 0.8724, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.3986295950071174e-05, | |
| "loss": 0.8768, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.3962101494528856e-05, | |
| "loss": 0.8721, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.3937858553103675e-05, | |
| "loss": 0.8705, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.391361561167851e-05, | |
| "loss": 0.8741, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.388937267025333e-05, | |
| "loss": 0.8823, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.386527518647671e-05, | |
| "loss": 0.8669, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.3841032245051534e-05, | |
| "loss": 0.871, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.381678930362636e-05, | |
| "loss": 0.8676, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.379254636220118e-05, | |
| "loss": 0.8756, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 44.6906, | |
| "eval_gen_len": 13.1329, | |
| "eval_loss": 0.8033798336982727, | |
| "eval_runtime": 8678.5438, | |
| "eval_samples_per_second": 15.5, | |
| "eval_steps_per_second": 1.938, | |
| "step": 138887 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.376830342077601e-05, | |
| "loss": 0.8397, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.3744060479350836e-05, | |
| "loss": 0.7043, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.371981753792566e-05, | |
| "loss": 0.6911, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.369557459650048e-05, | |
| "loss": 0.6938, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.367138014095816e-05, | |
| "loss": 0.696, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.364718568541584e-05, | |
| "loss": 0.6957, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.3622942743990664e-05, | |
| "loss": 0.6974, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.359869980256548e-05, | |
| "loss": 0.7081, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.357445686114031e-05, | |
| "loss": 0.7022, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.355021391971514e-05, | |
| "loss": 0.7026, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.3525970978289967e-05, | |
| "loss": 0.698, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.3501728036864785e-05, | |
| "loss": 0.7091, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.347748509543961e-05, | |
| "loss": 0.6997, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.3453242154014436e-05, | |
| "loss": 0.7156, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.342899921258927e-05, | |
| "loss": 0.7017, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.340475627116409e-05, | |
| "loss": 0.703, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.338051332973891e-05, | |
| "loss": 0.6993, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.335627038831374e-05, | |
| "loss": 0.7043, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.3332075932771414e-05, | |
| "loss": 0.7203, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.330783299134624e-05, | |
| "loss": 0.7097, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.328359004992107e-05, | |
| "loss": 0.7115, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.325934710849589e-05, | |
| "loss": 0.6984, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.3235104167070716e-05, | |
| "loss": 0.7154, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.321090971152839e-05, | |
| "loss": 0.7102, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.318666677010322e-05, | |
| "loss": 0.7123, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.316242382867804e-05, | |
| "loss": 0.7167, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.313818088725287e-05, | |
| "loss": 0.7244, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.3113937945827694e-05, | |
| "loss": 0.7088, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.308969500440252e-05, | |
| "loss": 0.7123, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.3065452062977345e-05, | |
| "loss": 0.7179, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.304120912155217e-05, | |
| "loss": 0.7166, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.3017014666009846e-05, | |
| "loss": 0.7137, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.299277172458467e-05, | |
| "loss": 0.7256, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.29685287831595e-05, | |
| "loss": 0.7179, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.294428584173432e-05, | |
| "loss": 0.7123, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.292004290030915e-05, | |
| "loss": 0.7211, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.2895799958883974e-05, | |
| "loss": 0.7181, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.28715570174588e-05, | |
| "loss": 0.7225, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.2847314076033625e-05, | |
| "loss": 0.7156, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.28231196204913e-05, | |
| "loss": 0.7202, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.2798876679066126e-05, | |
| "loss": 0.7301, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.277463373764095e-05, | |
| "loss": 0.7135, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.275039079621578e-05, | |
| "loss": 0.7267, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.2726244826556296e-05, | |
| "loss": 0.7152, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.270200188513113e-05, | |
| "loss": 0.735, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.2677758943705954e-05, | |
| "loss": 0.7134, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.265351600228078e-05, | |
| "loss": 0.7326, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.26292730608556e-05, | |
| "loss": 0.7329, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.260503011943043e-05, | |
| "loss": 0.7182, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.2580787178005256e-05, | |
| "loss": 0.7203, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.255654423658008e-05, | |
| "loss": 0.7227, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.25323012951549e-05, | |
| "loss": 0.7258, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.2508058353729726e-05, | |
| "loss": 0.7253, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.248381541230456e-05, | |
| "loss": 0.7289, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.245962095676223e-05, | |
| "loss": 0.7253, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.243537801533706e-05, | |
| "loss": 0.7309, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.2411135073911885e-05, | |
| "loss": 0.7304, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.2386892132486704e-05, | |
| "loss": 0.739, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.236264919106153e-05, | |
| "loss": 0.7291, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.233840624963636e-05, | |
| "loss": 0.7393, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.231416330821119e-05, | |
| "loss": 0.7237, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.2289920366786006e-05, | |
| "loss": 0.7326, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.226567742536083e-05, | |
| "loss": 0.7266, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.224153145570136e-05, | |
| "loss": 0.7212, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.221728851427619e-05, | |
| "loss": 0.7222, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.219304557285101e-05, | |
| "loss": 0.7362, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.216885111730869e-05, | |
| "loss": 0.7218, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.214460817588351e-05, | |
| "loss": 0.7372, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.2120365234458335e-05, | |
| "loss": 0.7184, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.209612229303316e-05, | |
| "loss": 0.7405, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.207187935160799e-05, | |
| "loss": 0.7246, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.204763641018281e-05, | |
| "loss": 0.7313, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.2023441954640494e-05, | |
| "loss": 0.7346, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.199919901321531e-05, | |
| "loss": 0.725, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.197495607179014e-05, | |
| "loss": 0.7358, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.1950713130364964e-05, | |
| "loss": 0.7369, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.192647018893979e-05, | |
| "loss": 0.7405, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.1902227247514615e-05, | |
| "loss": 0.7399, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.187803279197229e-05, | |
| "loss": 0.7273, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.1853789850547116e-05, | |
| "loss": 0.7332, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.182954690912194e-05, | |
| "loss": 0.7395, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.180535245357962e-05, | |
| "loss": 0.7434, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.178110951215444e-05, | |
| "loss": 0.7389, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.175686657072927e-05, | |
| "loss": 0.7344, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.1732623629304094e-05, | |
| "loss": 0.7314, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.170838068787892e-05, | |
| "loss": 0.7315, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.1684137746453745e-05, | |
| "loss": 0.732, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.165989480502857e-05, | |
| "loss": 0.7289, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.1635651863603396e-05, | |
| "loss": 0.7296, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.161140892217822e-05, | |
| "loss": 0.731, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.158716598075305e-05, | |
| "loss": 0.73, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.156292303932787e-05, | |
| "loss": 0.7343, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.153872858378555e-05, | |
| "loss": 0.7286, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.1514485642360374e-05, | |
| "loss": 0.7434, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.14902427009352e-05, | |
| "loss": 0.744, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.146599975951002e-05, | |
| "loss": 0.737, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.144175681808485e-05, | |
| "loss": 0.7445, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.1417513876659676e-05, | |
| "loss": 0.7396, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.13932709352345e-05, | |
| "loss": 0.7276, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.136902799380932e-05, | |
| "loss": 0.7468, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.134478505238415e-05, | |
| "loss": 0.7417, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.132054211095898e-05, | |
| "loss": 0.7424, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.1296299169533804e-05, | |
| "loss": 0.7314, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.127210471399148e-05, | |
| "loss": 0.7477, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.1247861772566305e-05, | |
| "loss": 0.7442, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.1223618831141124e-05, | |
| "loss": 0.7325, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.119937588971595e-05, | |
| "loss": 0.7442, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.117513294829078e-05, | |
| "loss": 0.7405, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.115089000686561e-05, | |
| "loss": 0.7345, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.1126647065440426e-05, | |
| "loss": 0.7328, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.110245260989811e-05, | |
| "loss": 0.7358, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.107820966847293e-05, | |
| "loss": 0.7534, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.105396672704775e-05, | |
| "loss": 0.7353, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.102972378562258e-05, | |
| "loss": 0.7441, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.100548084419741e-05, | |
| "loss": 0.7424, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.098123790277223e-05, | |
| "loss": 0.7383, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.0956994961347055e-05, | |
| "loss": 0.744, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.093275201992188e-05, | |
| "loss": 0.7392, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.090850907849671e-05, | |
| "loss": 0.731, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.088426613707153e-05, | |
| "loss": 0.7334, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.0860071681529214e-05, | |
| "loss": 0.7305, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.083582874010403e-05, | |
| "loss": 0.7339, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.081158579867886e-05, | |
| "loss": 0.7397, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.0787342857253684e-05, | |
| "loss": 0.7419, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.076309991582851e-05, | |
| "loss": 0.7392, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.0738856974403335e-05, | |
| "loss": 0.7375, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.071466251886101e-05, | |
| "loss": 0.7351, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.0690419577435836e-05, | |
| "loss": 0.753, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.066617663601066e-05, | |
| "loss": 0.7369, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.064193369458549e-05, | |
| "loss": 0.751, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.0617690753160306e-05, | |
| "loss": 0.7409, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.059349629761799e-05, | |
| "loss": 0.7454, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.056925335619281e-05, | |
| "loss": 0.7531, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.054501041476764e-05, | |
| "loss": 0.7482, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.0520767473342465e-05, | |
| "loss": 0.7299, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.049657301780014e-05, | |
| "loss": 0.7366, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.0472330076374966e-05, | |
| "loss": 0.7545, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.044808713494979e-05, | |
| "loss": 0.7475, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.042384419352461e-05, | |
| "loss": 0.7566, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 45.7383, | |
| "eval_gen_len": 13.0535, | |
| "eval_loss": 0.7818750739097595, | |
| "eval_runtime": 8588.0716, | |
| "eval_samples_per_second": 15.663, | |
| "eval_steps_per_second": 1.958, | |
| "step": 208330 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.039960125209944e-05, | |
| "loss": 0.6753, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.037535831067427e-05, | |
| "loss": 0.5634, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.035116385513194e-05, | |
| "loss": 0.5652, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.032692091370677e-05, | |
| "loss": 0.5603, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.0302677972281595e-05, | |
| "loss": 0.5516, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.027848351673927e-05, | |
| "loss": 0.5561, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.0254240575314096e-05, | |
| "loss": 0.5682, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.0229997633888915e-05, | |
| "loss": 0.5612, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.020575469246374e-05, | |
| "loss": 0.5686, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.0181560236921416e-05, | |
| "loss": 0.5705, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.015731729549624e-05, | |
| "loss": 0.5604, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.0133074354071074e-05, | |
| "loss": 0.5596, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.01088314126459e-05, | |
| "loss": 0.5693, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.008458847122072e-05, | |
| "loss": 0.5712, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.0060345529795544e-05, | |
| "loss": 0.5826, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.003610258837037e-05, | |
| "loss": 0.5733, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.00118596469452e-05, | |
| "loss": 0.571, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.998761670552002e-05, | |
| "loss": 0.5771, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.9963373764094846e-05, | |
| "loss": 0.564, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.993913082266967e-05, | |
| "loss": 0.573, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.9914887881244504e-05, | |
| "loss": 0.5772, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.989069342570217e-05, | |
| "loss": 0.5759, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.9866450484277e-05, | |
| "loss": 0.578, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.9842207542851824e-05, | |
| "loss": 0.5694, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.981796460142665e-05, | |
| "loss": 0.5799, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.9793721660001475e-05, | |
| "loss": 0.5709, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.97694787185763e-05, | |
| "loss": 0.5824, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.9745235777151126e-05, | |
| "loss": 0.5866, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.972099283572595e-05, | |
| "loss": 0.5748, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.969679838018363e-05, | |
| "loss": 0.5772, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.967255543875845e-05, | |
| "loss": 0.59, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.964836098321613e-05, | |
| "loss": 0.5777, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.9624118041790954e-05, | |
| "loss": 0.5797, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.959987510036578e-05, | |
| "loss": 0.5866, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.9575632158940605e-05, | |
| "loss": 0.5912, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.955138921751543e-05, | |
| "loss": 0.5867, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.9527146276090256e-05, | |
| "loss": 0.5958, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.950290333466508e-05, | |
| "loss": 0.5864, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.947866039323991e-05, | |
| "loss": 0.5919, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.945446593769758e-05, | |
| "loss": 0.5968, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.943022299627241e-05, | |
| "loss": 0.6008, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.940598005484723e-05, | |
| "loss": 0.5913, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.938173711342206e-05, | |
| "loss": 0.5915, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.9357494171996885e-05, | |
| "loss": 0.5995, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.933329971645456e-05, | |
| "loss": 0.5883, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.9309056775029386e-05, | |
| "loss": 0.5973, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.928481383360421e-05, | |
| "loss": 0.5969, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.926057089217903e-05, | |
| "loss": 0.5864, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.923632795075386e-05, | |
| "loss": 0.6016, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.921208500932869e-05, | |
| "loss": 0.6012, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.9187842067903514e-05, | |
| "loss": 0.5976, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.916359912647833e-05, | |
| "loss": 0.5963, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.9139404670936015e-05, | |
| "loss": 0.6039, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.911521021539369e-05, | |
| "loss": 0.6036, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.9090967273968516e-05, | |
| "loss": 0.5949, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.9066724332543335e-05, | |
| "loss": 0.6058, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.904252987700102e-05, | |
| "loss": 0.5999, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.9018286935575836e-05, | |
| "loss": 0.6034, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.899404399415066e-05, | |
| "loss": 0.6161, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.8969801052725494e-05, | |
| "loss": 0.5963, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.894555811130032e-05, | |
| "loss": 0.6032, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.8921363655757995e-05, | |
| "loss": 0.6042, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.8897169200215664e-05, | |
| "loss": 0.6079, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.8872926258790496e-05, | |
| "loss": 0.6014, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.884868331736532e-05, | |
| "loss": 0.6053, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.882444037594014e-05, | |
| "loss": 0.6021, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.8800197434514966e-05, | |
| "loss": 0.6112, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.877595449308979e-05, | |
| "loss": 0.6094, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.8751711551664624e-05, | |
| "loss": 0.6032, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.872746861023944e-05, | |
| "loss": 0.5961, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.8703274154697125e-05, | |
| "loss": 0.6179, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.8679031213271944e-05, | |
| "loss": 0.612, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.865478827184677e-05, | |
| "loss": 0.6134, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.8630545330421595e-05, | |
| "loss": 0.6046, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.860630238899642e-05, | |
| "loss": 0.6184, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.8582059447571246e-05, | |
| "loss": 0.6116, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.855786499202892e-05, | |
| "loss": 0.6098, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.853362205060375e-05, | |
| "loss": 0.6051, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.850937910917857e-05, | |
| "loss": 0.6083, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.84851361677534e-05, | |
| "loss": 0.6211, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.8460893226328224e-05, | |
| "loss": 0.6174, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.843665028490305e-05, | |
| "loss": 0.623, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.8412407343477875e-05, | |
| "loss": 0.6225, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.83881644020527e-05, | |
| "loss": 0.6202, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.8363969946510376e-05, | |
| "loss": 0.6165, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.83397270050852e-05, | |
| "loss": 0.6138, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.831548406366003e-05, | |
| "loss": 0.6103, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.829124112223485e-05, | |
| "loss": 0.6202, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.826704666669253e-05, | |
| "loss": 0.615, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.8242803725267354e-05, | |
| "loss": 0.6095, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.821856078384218e-05, | |
| "loss": 0.6181, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.8194317842417005e-05, | |
| "loss": 0.6153, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.817007490099183e-05, | |
| "loss": 0.6153, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.814583195956665e-05, | |
| "loss": 0.6108, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.812158901814148e-05, | |
| "loss": 0.6162, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.809734607671631e-05, | |
| "loss": 0.6132, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.807310313529113e-05, | |
| "loss": 0.6124, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.804886019386595e-05, | |
| "loss": 0.627, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.8024665738323634e-05, | |
| "loss": 0.6221, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.800042279689845e-05, | |
| "loss": 0.6122, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.7976179855473285e-05, | |
| "loss": 0.6049, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.795193691404811e-05, | |
| "loss": 0.6228, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.7927742458505786e-05, | |
| "loss": 0.616, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.790349951708061e-05, | |
| "loss": 0.6167, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.787925657565544e-05, | |
| "loss": 0.6266, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.7855013634230256e-05, | |
| "loss": 0.6334, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.783077069280508e-05, | |
| "loss": 0.6152, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.7806527751379914e-05, | |
| "loss": 0.6214, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.778228480995474e-05, | |
| "loss": 0.6052, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.775804186852956e-05, | |
| "loss": 0.6202, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.773384741298724e-05, | |
| "loss": 0.6245, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.7709652957444916e-05, | |
| "loss": 0.6264, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.7685458501902585e-05, | |
| "loss": 0.6282, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.766121556047742e-05, | |
| "loss": 0.6225, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.763697261905224e-05, | |
| "loss": 0.6157, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.761277816350991e-05, | |
| "loss": 0.6173, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.7588535222084744e-05, | |
| "loss": 0.6224, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.756429228065956e-05, | |
| "loss": 0.6239, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.754004933923439e-05, | |
| "loss": 0.6242, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.7515806397809214e-05, | |
| "loss": 0.6284, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.7491563456384046e-05, | |
| "loss": 0.6279, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.7467320514958865e-05, | |
| "loss": 0.6289, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.744307757353369e-05, | |
| "loss": 0.6222, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.7418834632108516e-05, | |
| "loss": 0.6195, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.739459169068334e-05, | |
| "loss": 0.627, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.737034874925817e-05, | |
| "loss": 0.6222, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.734610580783299e-05, | |
| "loss": 0.6236, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.732186286640782e-05, | |
| "loss": 0.627, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.7297619924982644e-05, | |
| "loss": 0.6254, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.727337698355747e-05, | |
| "loss": 0.627, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.7249134042132295e-05, | |
| "loss": 0.635, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.722489110070712e-05, | |
| "loss": 0.619, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.7200696645164796e-05, | |
| "loss": 0.6212, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.717645370373962e-05, | |
| "loss": 0.6289, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.715221076231445e-05, | |
| "loss": 0.6251, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.712796782088927e-05, | |
| "loss": 0.6262, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.71037248794641e-05, | |
| "loss": 0.6294, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.7079530423921774e-05, | |
| "loss": 0.6298, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.70552874824966e-05, | |
| "loss": 0.6239, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 46.5398, | |
| "eval_gen_len": 12.9925, | |
| "eval_loss": 0.7927529215812683, | |
| "eval_runtime": 8543.5664, | |
| "eval_samples_per_second": 15.745, | |
| "eval_steps_per_second": 1.968, | |
| "step": 277774 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.7031044541071425e-05, | |
| "loss": 0.5457, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.70068500855291e-05, | |
| "loss": 0.4447, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.6982607144103926e-05, | |
| "loss": 0.4393, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.695836420267875e-05, | |
| "loss": 0.4442, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.693416974713643e-05, | |
| "loss": 0.4464, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.690992680571125e-05, | |
| "loss": 0.4616, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.688568386428607e-05, | |
| "loss": 0.4539, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.6861440922860904e-05, | |
| "loss": 0.4506, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.683719798143573e-05, | |
| "loss": 0.4513, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.6812955040010555e-05, | |
| "loss": 0.4494, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.6788712098585374e-05, | |
| "loss": 0.4543, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.6764469157160206e-05, | |
| "loss": 0.4521, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.674022621573503e-05, | |
| "loss": 0.4621, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.671598327430986e-05, | |
| "loss": 0.4531, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.669178881876753e-05, | |
| "loss": 0.4547, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.666754587734236e-05, | |
| "loss": 0.4573, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.6643351421800034e-05, | |
| "loss": 0.4629, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.66191569662577e-05, | |
| "loss": 0.4655, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6594914024832535e-05, | |
| "loss": 0.4582, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.657067108340736e-05, | |
| "loss": 0.4728, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.654642814198218e-05, | |
| "loss": 0.4531, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.6522185200557005e-05, | |
| "loss": 0.4707, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.649794225913184e-05, | |
| "loss": 0.4699, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.647369931770666e-05, | |
| "loss": 0.4679, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.644945637628148e-05, | |
| "loss": 0.468, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.642521343485631e-05, | |
| "loss": 0.4692, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.640097049343113e-05, | |
| "loss": 0.466, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.637672755200596e-05, | |
| "loss": 0.464, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.6352484610580784e-05, | |
| "loss": 0.472, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.632829015503846e-05, | |
| "loss": 0.4728, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.6304047213613285e-05, | |
| "loss": 0.4765, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.627980427218811e-05, | |
| "loss": 0.4767, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.6255561330762936e-05, | |
| "loss": 0.4739, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.623131838933776e-05, | |
| "loss": 0.475, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.620707544791259e-05, | |
| "loss": 0.4735, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.618283250648741e-05, | |
| "loss": 0.4783, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.615858956506224e-05, | |
| "loss": 0.4768, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.6134346623637064e-05, | |
| "loss": 0.481, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.611015216809474e-05, | |
| "loss": 0.4861, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.6085909226669565e-05, | |
| "loss": 0.4804, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.606166628524439e-05, | |
| "loss": 0.483, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.6037423343819216e-05, | |
| "loss": 0.4827, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.601318040239404e-05, | |
| "loss": 0.4827, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.598893746096886e-05, | |
| "loss": 0.4861, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.596474300542654e-05, | |
| "loss": 0.4876, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.594054854988422e-05, | |
| "loss": 0.4849, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.5916305608459044e-05, | |
| "loss": 0.4825, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.589206266703386e-05, | |
| "loss": 0.4884, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.5867819725608695e-05, | |
| "loss": 0.4805, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.584357678418352e-05, | |
| "loss": 0.4856, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.5819333842758346e-05, | |
| "loss": 0.4856, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.5795090901333165e-05, | |
| "loss": 0.4799, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.5770847959908e-05, | |
| "loss": 0.4898, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.574660501848282e-05, | |
| "loss": 0.4885, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.572236207705765e-05, | |
| "loss": 0.4878, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.569811913563247e-05, | |
| "loss": 0.4832, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.567387619420729e-05, | |
| "loss": 0.4869, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.5649633252782125e-05, | |
| "loss": 0.4993, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 3.562539031135695e-05, | |
| "loss": 0.4983, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.5601195855814626e-05, | |
| "loss": 0.4918, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 3.557695291438945e-05, | |
| "loss": 0.4882, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 3.555270997296427e-05, | |
| "loss": 0.4985, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 3.5528467031539096e-05, | |
| "loss": 0.4898, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.550422409011392e-05, | |
| "loss": 0.4955, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.5479981148688754e-05, | |
| "loss": 0.4886, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.545573820726357e-05, | |
| "loss": 0.489, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 3.5431543751721255e-05, | |
| "loss": 0.5072, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.5407300810296074e-05, | |
| "loss": 0.5013, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.53830578688709e-05, | |
| "loss": 0.4902, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.5358814927445725e-05, | |
| "loss": 0.4912, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 3.53346204719034e-05, | |
| "loss": 0.5037, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 3.5310377530478226e-05, | |
| "loss": 0.4952, | |
| "step": 313500 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.528613458905306e-05, | |
| "loss": 0.4948, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.526189164762788e-05, | |
| "loss": 0.4971, | |
| "step": 314500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.523769719208555e-05, | |
| "loss": 0.4999, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.521350273654323e-05, | |
| "loss": 0.5081, | |
| "step": 315500 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 3.5189259795118054e-05, | |
| "loss": 0.498, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 3.516501685369288e-05, | |
| "loss": 0.5034, | |
| "step": 316500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 3.5140773912267705e-05, | |
| "loss": 0.4946, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.511653097084253e-05, | |
| "loss": 0.5005, | |
| "step": 317500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 3.5092288029417356e-05, | |
| "loss": 0.502, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.506804508799218e-05, | |
| "loss": 0.5057, | |
| "step": 318500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.504380214656701e-05, | |
| "loss": 0.5064, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.501960769102468e-05, | |
| "loss": 0.5086, | |
| "step": 319500 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.499536474959951e-05, | |
| "loss": 0.4985, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.4971121808174334e-05, | |
| "loss": 0.5065, | |
| "step": 320500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.494687886674916e-05, | |
| "loss": 0.4943, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.4922635925323985e-05, | |
| "loss": 0.5075, | |
| "step": 321500 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.489844146978166e-05, | |
| "loss": 0.5142, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.4874198528356486e-05, | |
| "loss": 0.5056, | |
| "step": 322500 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.484995558693131e-05, | |
| "loss": 0.5121, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.482576113138899e-05, | |
| "loss": 0.5106, | |
| "step": 323500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.480151818996381e-05, | |
| "loss": 0.5029, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.477727524853864e-05, | |
| "loss": 0.4978, | |
| "step": 324500 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.4753032307113464e-05, | |
| "loss": 0.5099, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.472878936568828e-05, | |
| "loss": 0.5093, | |
| "step": 325500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.4704546424263115e-05, | |
| "loss": 0.5085, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.468030348283794e-05, | |
| "loss": 0.5245, | |
| "step": 326500 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 3.4656060541412766e-05, | |
| "loss": 0.5107, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.4631817599987585e-05, | |
| "loss": 0.505, | |
| "step": 327500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.460757465856242e-05, | |
| "loss": 0.5029, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.458333171713724e-05, | |
| "loss": 0.5024, | |
| "step": 328500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 3.455908877571207e-05, | |
| "loss": 0.5099, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 3.453484583428689e-05, | |
| "loss": 0.517, | |
| "step": 329500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.451060289286171e-05, | |
| "loss": 0.5026, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 3.448640843731939e-05, | |
| "loss": 0.5098, | |
| "step": 330500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 3.4462165495894214e-05, | |
| "loss": 0.52, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 3.443797104035189e-05, | |
| "loss": 0.5084, | |
| "step": 331500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 3.4413728098926715e-05, | |
| "loss": 0.5184, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.438948515750155e-05, | |
| "loss": 0.5215, | |
| "step": 332500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 3.436524221607637e-05, | |
| "loss": 0.525, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 3.434099927465119e-05, | |
| "loss": 0.5016, | |
| "step": 333500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.431675633322602e-05, | |
| "loss": 0.5183, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.429256187768369e-05, | |
| "loss": 0.5177, | |
| "step": 334500 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.426831893625852e-05, | |
| "loss": 0.5171, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.4244075994833344e-05, | |
| "loss": 0.5143, | |
| "step": 335500 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 3.4219833053408176e-05, | |
| "loss": 0.5099, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.4195590111982995e-05, | |
| "loss": 0.5138, | |
| "step": 336500 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.417134717055782e-05, | |
| "loss": 0.5219, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 3.4147104229132646e-05, | |
| "loss": 0.5089, | |
| "step": 337500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.412286128770748e-05, | |
| "loss": 0.5252, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.40986183462823e-05, | |
| "loss": 0.5239, | |
| "step": 338500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 3.407437540485712e-05, | |
| "loss": 0.516, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 3.405013246343195e-05, | |
| "loss": 0.517, | |
| "step": 339500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.4025889522006773e-05, | |
| "loss": 0.5149, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.400169506646445e-05, | |
| "loss": 0.514, | |
| "step": 340500 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 3.3977452125039275e-05, | |
| "loss": 0.5215, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.39532091836141e-05, | |
| "loss": 0.5126, | |
| "step": 341500 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.3929014728071776e-05, | |
| "loss": 0.529, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.39047717866466e-05, | |
| "loss": 0.5155, | |
| "step": 342500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.388052884522143e-05, | |
| "loss": 0.5285, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.385628590379625e-05, | |
| "loss": 0.5231, | |
| "step": 343500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.383204296237108e-05, | |
| "loss": 0.5187, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 3.3807800020945903e-05, | |
| "loss": 0.5146, | |
| "step": 344500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.378355707952073e-05, | |
| "loss": 0.5305, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.3759314138095555e-05, | |
| "loss": 0.5251, | |
| "step": 345500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.373511968255323e-05, | |
| "loss": 0.5334, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 3.3710876741128056e-05, | |
| "loss": 0.5168, | |
| "step": 346500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.368663379970288e-05, | |
| "loss": 0.5245, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 46.9153, | |
| "eval_gen_len": 13.0368, | |
| "eval_loss": 0.8332552313804626, | |
| "eval_runtime": 8588.4827, | |
| "eval_samples_per_second": 15.662, | |
| "eval_steps_per_second": 1.958, | |
| "step": 347217 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.366239085827771e-05, | |
| "loss": 0.4247, | |
| "step": 347500 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.363814791685253e-05, | |
| "loss": 0.3521, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.361395346131021e-05, | |
| "loss": 0.3483, | |
| "step": 348500 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.3589759005767884e-05, | |
| "loss": 0.3564, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.356556455022556e-05, | |
| "loss": 0.3537, | |
| "step": 349500 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.3541321608800385e-05, | |
| "loss": 0.3593, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.3517078667375204e-05, | |
| "loss": 0.3542, | |
| "step": 350500 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.3492835725950036e-05, | |
| "loss": 0.3607, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.346859278452486e-05, | |
| "loss": 0.3539, | |
| "step": 351500 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.344439832898254e-05, | |
| "loss": 0.3558, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.342015538755736e-05, | |
| "loss": 0.3599, | |
| "step": 352500 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.339591244613219e-05, | |
| "loss": 0.3625, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.337166950470701e-05, | |
| "loss": 0.3583, | |
| "step": 353500 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.334747504916469e-05, | |
| "loss": 0.3628, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.332323210773951e-05, | |
| "loss": 0.3698, | |
| "step": 354500 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.3298989166314334e-05, | |
| "loss": 0.3605, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.3274746224889166e-05, | |
| "loss": 0.3647, | |
| "step": 355500 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 3.325050328346399e-05, | |
| "loss": 0.3602, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 3.322626034203881e-05, | |
| "loss": 0.3623, | |
| "step": 356500 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 3.3202017400613636e-05, | |
| "loss": 0.355, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.317777445918847e-05, | |
| "loss": 0.3662, | |
| "step": 357500 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.3153531517763294e-05, | |
| "loss": 0.3699, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.312928857633811e-05, | |
| "loss": 0.3638, | |
| "step": 358500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 3.310504563491294e-05, | |
| "loss": 0.3698, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 3.3080851179370614e-05, | |
| "loss": 0.3782, | |
| "step": 359500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 3.305660823794544e-05, | |
| "loss": 0.378, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.3032365296520265e-05, | |
| "loss": 0.3739, | |
| "step": 360500 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 3.30081223550951e-05, | |
| "loss": 0.3848, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 3.2983879413669916e-05, | |
| "loss": 0.3696, | |
| "step": 361500 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 3.295963647224474e-05, | |
| "loss": 0.3749, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.293539353081957e-05, | |
| "loss": 0.3749, | |
| "step": 362500 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.29111505893944e-05, | |
| "loss": 0.3758, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.288690764796922e-05, | |
| "loss": 0.38, | |
| "step": 363500 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 3.2862664706544043e-05, | |
| "loss": 0.3833, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 3.283842176511887e-05, | |
| "loss": 0.3792, | |
| "step": 364500 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.2814178823693695e-05, | |
| "loss": 0.3809, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.278998436815137e-05, | |
| "loss": 0.3887, | |
| "step": 365500 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 3.2765741426726196e-05, | |
| "loss": 0.3762, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.274149848530102e-05, | |
| "loss": 0.3816, | |
| "step": 366500 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.27173040297587e-05, | |
| "loss": 0.3895, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 3.269306108833352e-05, | |
| "loss": 0.391, | |
| "step": 367500 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 3.266881814690835e-05, | |
| "loss": 0.3844, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.2644575205483174e-05, | |
| "loss": 0.3836, | |
| "step": 368500 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.2620332264058e-05, | |
| "loss": 0.3894, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 3.2596089322632825e-05, | |
| "loss": 0.3864, | |
| "step": 369500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 3.257184638120765e-05, | |
| "loss": 0.3869, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 3.2547603439782476e-05, | |
| "loss": 0.388, | |
| "step": 370500 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 3.2523457470123e-05, | |
| "loss": 0.3891, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.249921452869783e-05, | |
| "loss": 0.391, | |
| "step": 371500 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 3.247497158727265e-05, | |
| "loss": 0.3888, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 3.245072864584748e-05, | |
| "loss": 0.3858, | |
| "step": 372500 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 3.2426485704422304e-05, | |
| "loss": 0.385, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 3.240229124887998e-05, | |
| "loss": 0.3935, | |
| "step": 373500 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.2378048307454805e-05, | |
| "loss": 0.3886, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.2353805366029624e-05, | |
| "loss": 0.3928, | |
| "step": 374500 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 3.2329562424604456e-05, | |
| "loss": 0.3903, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 3.230531948317928e-05, | |
| "loss": 0.4016, | |
| "step": 375500 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 3.22810765417541e-05, | |
| "loss": 0.3944, | |
| "step": 376000 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 3.2256833600328926e-05, | |
| "loss": 0.4017, | |
| "step": 376500 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.223259065890376e-05, | |
| "loss": 0.3953, | |
| "step": 377000 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 3.2208347717478584e-05, | |
| "loss": 0.3985, | |
| "step": 377500 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 3.21842017478191e-05, | |
| "loss": 0.3957, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 3.215995880639393e-05, | |
| "loss": 0.3971, | |
| "step": 378500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 3.213571586496876e-05, | |
| "loss": 0.3952, | |
| "step": 379000 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 3.211152140942643e-05, | |
| "loss": 0.3943, | |
| "step": 379500 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.2087326953884105e-05, | |
| "loss": 0.3969, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 3.206308401245893e-05, | |
| "loss": 0.3996, | |
| "step": 380500 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 3.2038841071033756e-05, | |
| "loss": 0.3963, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 3.201459812960859e-05, | |
| "loss": 0.404, | |
| "step": 381500 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 3.199035518818341e-05, | |
| "loss": 0.3968, | |
| "step": 382000 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.196611224675823e-05, | |
| "loss": 0.3917, | |
| "step": 382500 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 3.194186930533306e-05, | |
| "loss": 0.4022, | |
| "step": 383000 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 3.191762636390789e-05, | |
| "loss": 0.4027, | |
| "step": 383500 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.189338342248271e-05, | |
| "loss": 0.405, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.1869140481057535e-05, | |
| "loss": 0.405, | |
| "step": 384500 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.184494602551521e-05, | |
| "loss": 0.4053, | |
| "step": 385000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 3.1820703084090036e-05, | |
| "loss": 0.4063, | |
| "step": 385500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 3.179646014266486e-05, | |
| "loss": 0.4076, | |
| "step": 386000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 3.177221720123969e-05, | |
| "loss": 0.4099, | |
| "step": 386500 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 3.174797425981451e-05, | |
| "loss": 0.3982, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 3.172373131838934e-05, | |
| "loss": 0.4, | |
| "step": 387500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.1699488376964164e-05, | |
| "loss": 0.407, | |
| "step": 388000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.167529392142184e-05, | |
| "loss": 0.4046, | |
| "step": 388500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 3.1651050979996665e-05, | |
| "loss": 0.4076, | |
| "step": 389000 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 3.162680803857149e-05, | |
| "loss": 0.4031, | |
| "step": 389500 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 3.1602565097146316e-05, | |
| "loss": 0.4142, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 3.157832215572114e-05, | |
| "loss": 0.4085, | |
| "step": 390500 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.155407921429597e-05, | |
| "loss": 0.4037, | |
| "step": 391000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 3.152983627287079e-05, | |
| "loss": 0.405, | |
| "step": 391500 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 3.150559333144562e-05, | |
| "loss": 0.3994, | |
| "step": 392000 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 3.1481350390020444e-05, | |
| "loss": 0.4124, | |
| "step": 392500 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 3.145715593447812e-05, | |
| "loss": 0.4099, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.1432912993052945e-05, | |
| "loss": 0.4144, | |
| "step": 393500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.140867005162777e-05, | |
| "loss": 0.4116, | |
| "step": 394000 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 3.1384427110202596e-05, | |
| "loss": 0.4118, | |
| "step": 394500 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 3.1360184168777415e-05, | |
| "loss": 0.4041, | |
| "step": 395000 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 3.133594122735225e-05, | |
| "loss": 0.4075, | |
| "step": 395500 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 3.1311746771809916e-05, | |
| "loss": 0.4136, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.128750383038475e-05, | |
| "loss": 0.4119, | |
| "step": 396500 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 3.126330937484242e-05, | |
| "loss": 0.4129, | |
| "step": 397000 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 3.123906643341725e-05, | |
| "loss": 0.4128, | |
| "step": 397500 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 3.1214823491992075e-05, | |
| "loss": 0.4187, | |
| "step": 398000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.11905805505669e-05, | |
| "loss": 0.4133, | |
| "step": 398500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 3.116633760914172e-05, | |
| "loss": 0.4204, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 3.114209466771655e-05, | |
| "loss": 0.4102, | |
| "step": 399500 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 3.111785172629138e-05, | |
| "loss": 0.4191, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 3.10936087848662e-05, | |
| "loss": 0.4137, | |
| "step": 400500 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 3.106936584344102e-05, | |
| "loss": 0.4156, | |
| "step": 401000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.104512290201585e-05, | |
| "loss": 0.4229, | |
| "step": 401500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 3.102087996059068e-05, | |
| "loss": 0.4145, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 3.0996637019165505e-05, | |
| "loss": 0.4152, | |
| "step": 402500 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 3.097244256362318e-05, | |
| "loss": 0.4161, | |
| "step": 403000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 3.0948199622198006e-05, | |
| "loss": 0.4131, | |
| "step": 403500 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 3.0923956680772825e-05, | |
| "loss": 0.4234, | |
| "step": 404000 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 3.089976222523051e-05, | |
| "loss": 0.422, | |
| "step": 404500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 3.0875519283805326e-05, | |
| "loss": 0.4174, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 3.085127634238015e-05, | |
| "loss": 0.4186, | |
| "step": 405500 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 3.082708188683783e-05, | |
| "loss": 0.4215, | |
| "step": 406000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 3.080283894541265e-05, | |
| "loss": 0.4247, | |
| "step": 406500 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 3.077859600398748e-05, | |
| "loss": 0.4287, | |
| "step": 407000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 3.075435306256231e-05, | |
| "loss": 0.4263, | |
| "step": 407500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 3.073011012113713e-05, | |
| "loss": 0.4208, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 3.0705867179711955e-05, | |
| "loss": 0.4225, | |
| "step": 408500 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 3.068167272416963e-05, | |
| "loss": 0.4225, | |
| "step": 409000 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 3.0657429782744456e-05, | |
| "loss": 0.4158, | |
| "step": 409500 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 3.063318684131928e-05, | |
| "loss": 0.4212, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 3.060894389989411e-05, | |
| "loss": 0.4246, | |
| "step": 410500 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 3.058470095846893e-05, | |
| "loss": 0.4238, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 3.056045801704376e-05, | |
| "loss": 0.4321, | |
| "step": 411500 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 3.0536215075618584e-05, | |
| "loss": 0.4271, | |
| "step": 412000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 3.0511972134193413e-05, | |
| "loss": 0.4302, | |
| "step": 412500 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 3.048772919276823e-05, | |
| "loss": 0.422, | |
| "step": 413000 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 3.046348625134306e-05, | |
| "loss": 0.4267, | |
| "step": 413500 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 3.0439243309917886e-05, | |
| "loss": 0.4291, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 3.0415097340258415e-05, | |
| "loss": 0.4291, | |
| "step": 414500 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 3.0390854398833234e-05, | |
| "loss": 0.4251, | |
| "step": 415000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 3.0366611457408063e-05, | |
| "loss": 0.4233, | |
| "step": 415500 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 3.0342417001865735e-05, | |
| "loss": 0.4266, | |
| "step": 416000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3.0318174060440564e-05, | |
| "loss": 0.4333, | |
| "step": 416500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 46.5833, | |
| "eval_gen_len": 12.9999, | |
| "eval_loss": 0.8992630243301392, | |
| "eval_runtime": 8567.0337, | |
| "eval_samples_per_second": 15.702, | |
| "eval_steps_per_second": 1.963, | |
| "step": 416661 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 416661, | |
| "total_flos": 1.4447335300747231e+19, | |
| "train_loss": 0.7209485457693631, | |
| "train_runtime": 374884.8401, | |
| "train_samples_per_second": 44.457, | |
| "train_steps_per_second": 2.779 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1041645, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "total_flos": 1.4447335300747231e+19, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |