| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 22486, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.991105576803343e-07, | |
| "loss": 3.5693, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.981618192060246e-07, | |
| "loss": 1.3039, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.972130807317145e-07, | |
| "loss": 0.6178, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.962643422574046e-07, | |
| "loss": 0.2217, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.953156037830946e-07, | |
| "loss": 0.135, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.943668653087847e-07, | |
| "loss": 0.2547, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.934181268344748e-07, | |
| "loss": 0.2357, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.92513860476148e-07, | |
| "loss": 0.4032, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.915651220018381e-07, | |
| "loss": 0.2513, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.906163835275282e-07, | |
| "loss": 0.2058, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.89667645053218e-07, | |
| "loss": 0.2276, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.887189065789084e-07, | |
| "loss": 0.1798, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.877701681045983e-07, | |
| "loss": 0.2646, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.868214296302884e-07, | |
| "loss": 0.2434, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.858726911559785e-07, | |
| "loss": 0.3211, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.849239526816686e-07, | |
| "loss": 0.1601, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.839752142073586e-07, | |
| "loss": 0.2012, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.830264757330487e-07, | |
| "loss": 0.1635, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.820777372587388e-07, | |
| "loss": 0.2808, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.811289987844287e-07, | |
| "loss": 0.1622, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.801802603101188e-07, | |
| "loss": 0.23, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.792315218358089e-07, | |
| "loss": 0.1615, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.78282783361499e-07, | |
| "loss": 0.1424, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.77334044887189e-07, | |
| "loss": 0.2007, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.763853064128792e-07, | |
| "loss": 0.3649, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.75436567938569e-07, | |
| "loss": 0.1149, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.744878294642591e-07, | |
| "loss": 0.1882, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.735390909899492e-07, | |
| "loss": 0.2598, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.725903525156393e-07, | |
| "loss": 0.2198, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.716416140413294e-07, | |
| "loss": 0.1786, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.706928755670195e-07, | |
| "loss": 0.2713, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.697441370927096e-07, | |
| "loss": 0.2076, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.687953986183995e-07, | |
| "loss": 0.2343, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.678466601440898e-07, | |
| "loss": 0.3182, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.668979216697796e-07, | |
| "loss": 0.2102, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.659491831954697e-07, | |
| "loss": 0.2687, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.650004447211598e-07, | |
| "loss": 0.2186, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.6405170624685e-07, | |
| "loss": 0.1295, | |
| "step": 2432 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.631029677725398e-07, | |
| "loss": 0.1542, | |
| "step": 2496 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.621542292982299e-07, | |
| "loss": 0.1365, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.6120549082392e-07, | |
| "loss": 0.2106, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.6025675234961e-07, | |
| "loss": 0.2753, | |
| "step": 2688 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.593080138753002e-07, | |
| "loss": 0.2162, | |
| "step": 2752 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.583592754009902e-07, | |
| "loss": 0.2186, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.574105369266803e-07, | |
| "loss": 0.0991, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.564617984523702e-07, | |
| "loss": 0.1616, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.555130599780605e-07, | |
| "loss": 0.2029, | |
| "step": 3008 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.545643215037504e-07, | |
| "loss": 0.1659, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.536155830294405e-07, | |
| "loss": 0.1887, | |
| "step": 3136 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.526668445551306e-07, | |
| "loss": 0.2626, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.517181060808206e-07, | |
| "loss": 0.1325, | |
| "step": 3264 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.507693676065108e-07, | |
| "loss": 0.2125, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.498206291322007e-07, | |
| "loss": 0.1926, | |
| "step": 3392 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.488718906578907e-07, | |
| "loss": 0.2237, | |
| "step": 3456 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.479231521835809e-07, | |
| "loss": 0.1283, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.469744137092709e-07, | |
| "loss": 0.2052, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.46025675234961e-07, | |
| "loss": 0.1983, | |
| "step": 3648 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.45076936760651e-07, | |
| "loss": 0.1976, | |
| "step": 3712 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.441281982863411e-07, | |
| "loss": 0.1999, | |
| "step": 3776 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.431794598120312e-07, | |
| "loss": 0.2231, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.422307213377212e-07, | |
| "loss": 0.2057, | |
| "step": 3904 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.412819828634113e-07, | |
| "loss": 0.2376, | |
| "step": 3968 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.403332443891013e-07, | |
| "loss": 0.1325, | |
| "step": 4032 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.393845059147913e-07, | |
| "loss": 0.2595, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.384357674404815e-07, | |
| "loss": 0.1817, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.374870289661715e-07, | |
| "loss": 0.2224, | |
| "step": 4224 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.365382904918616e-07, | |
| "loss": 0.1444, | |
| "step": 4288 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.355895520175517e-07, | |
| "loss": 0.2695, | |
| "step": 4352 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.346408135432417e-07, | |
| "loss": 0.1859, | |
| "step": 4416 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.336920750689318e-07, | |
| "loss": 0.1551, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.327433365946217e-07, | |
| "loss": 0.2093, | |
| "step": 4544 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.317945981203118e-07, | |
| "loss": 0.1922, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.308458596460019e-07, | |
| "loss": 0.2056, | |
| "step": 4672 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.298971211716919e-07, | |
| "loss": 0.2087, | |
| "step": 4736 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.289483826973821e-07, | |
| "loss": 0.1552, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.279996442230721e-07, | |
| "loss": 0.1619, | |
| "step": 4864 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.270509057487621e-07, | |
| "loss": 0.2634, | |
| "step": 4928 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.261021672744523e-07, | |
| "loss": 0.1459, | |
| "step": 4992 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.251534288001423e-07, | |
| "loss": 0.225, | |
| "step": 5056 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.242046903258323e-07, | |
| "loss": 0.1878, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.232559518515224e-07, | |
| "loss": 0.1286, | |
| "step": 5184 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.223072133772124e-07, | |
| "loss": 0.1059, | |
| "step": 5248 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.213584749029025e-07, | |
| "loss": 0.1541, | |
| "step": 5312 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.204097364285925e-07, | |
| "loss": 0.2427, | |
| "step": 5376 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.194609979542827e-07, | |
| "loss": 0.1368, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.185122594799727e-07, | |
| "loss": 0.2477, | |
| "step": 5504 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.175635210056627e-07, | |
| "loss": 0.1385, | |
| "step": 5568 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.166147825313529e-07, | |
| "loss": 0.2537, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.156660440570428e-07, | |
| "loss": 0.2374, | |
| "step": 5696 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.147173055827329e-07, | |
| "loss": 0.1942, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.13768567108423e-07, | |
| "loss": 0.1416, | |
| "step": 5824 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.12819828634113e-07, | |
| "loss": 0.2205, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.118710901598031e-07, | |
| "loss": 0.1265, | |
| "step": 5952 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.109223516854932e-07, | |
| "loss": 0.1972, | |
| "step": 6016 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.099736132111833e-07, | |
| "loss": 0.2119, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.090248747368733e-07, | |
| "loss": 0.1284, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.080761362625632e-07, | |
| "loss": 0.1665, | |
| "step": 6208 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.071273977882534e-07, | |
| "loss": 0.1545, | |
| "step": 6272 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.061786593139434e-07, | |
| "loss": 0.1505, | |
| "step": 6336 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.052299208396334e-07, | |
| "loss": 0.1871, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.042811823653236e-07, | |
| "loss": 0.2302, | |
| "step": 6464 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.033324438910136e-07, | |
| "loss": 0.1395, | |
| "step": 6528 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.023837054167037e-07, | |
| "loss": 0.2195, | |
| "step": 6592 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.014349669423938e-07, | |
| "loss": 0.2419, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.004862284680838e-07, | |
| "loss": 0.2173, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.995374899937739e-07, | |
| "loss": 0.2522, | |
| "step": 6784 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.985887515194639e-07, | |
| "loss": 0.2317, | |
| "step": 6848 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.97640013045154e-07, | |
| "loss": 0.1959, | |
| "step": 6912 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.96691274570844e-07, | |
| "loss": 0.1889, | |
| "step": 6976 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.95742536096534e-07, | |
| "loss": 0.2131, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.947937976222242e-07, | |
| "loss": 0.1321, | |
| "step": 7104 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.938450591479142e-07, | |
| "loss": 0.1679, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.928963206736043e-07, | |
| "loss": 0.1826, | |
| "step": 7232 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.919475821992944e-07, | |
| "loss": 0.2144, | |
| "step": 7296 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.909988437249843e-07, | |
| "loss": 0.1614, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.900501052506744e-07, | |
| "loss": 0.241, | |
| "step": 7424 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.891013667763645e-07, | |
| "loss": 0.1825, | |
| "step": 7488 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.881526283020546e-07, | |
| "loss": 0.17, | |
| "step": 7552 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.872038898277446e-07, | |
| "loss": 0.1562, | |
| "step": 7616 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.862551513534347e-07, | |
| "loss": 0.2264, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.853064128791248e-07, | |
| "loss": 0.1325, | |
| "step": 7744 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.843576744048148e-07, | |
| "loss": 0.1601, | |
| "step": 7808 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.83408935930505e-07, | |
| "loss": 0.1784, | |
| "step": 7872 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.82460197456195e-07, | |
| "loss": 0.1447, | |
| "step": 7936 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.815114589818849e-07, | |
| "loss": 0.166, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.80562720507575e-07, | |
| "loss": 0.1554, | |
| "step": 8064 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.796139820332651e-07, | |
| "loss": 0.1097, | |
| "step": 8128 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.786652435589552e-07, | |
| "loss": 0.1322, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.777165050846452e-07, | |
| "loss": 0.262, | |
| "step": 8256 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.767677666103353e-07, | |
| "loss": 0.1755, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.758190281360254e-07, | |
| "loss": 0.1646, | |
| "step": 8384 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.748702896617154e-07, | |
| "loss": 0.1481, | |
| "step": 8448 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.739215511874055e-07, | |
| "loss": 0.0985, | |
| "step": 8512 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.729728127130955e-07, | |
| "loss": 0.1401, | |
| "step": 8576 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.720240742387855e-07, | |
| "loss": 0.2057, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.710753357644757e-07, | |
| "loss": 0.1677, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.701265972901657e-07, | |
| "loss": 0.1398, | |
| "step": 8768 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.691778588158557e-07, | |
| "loss": 0.1665, | |
| "step": 8832 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.682439443802069e-07, | |
| "loss": 0.1737, | |
| "step": 8896 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.67295205905897e-07, | |
| "loss": 0.1642, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.66346467431587e-07, | |
| "loss": 0.1487, | |
| "step": 9024 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.653977289572771e-07, | |
| "loss": 0.186, | |
| "step": 9088 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.644489904829672e-07, | |
| "loss": 0.1902, | |
| "step": 9152 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.635002520086571e-07, | |
| "loss": 0.1977, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.625515135343473e-07, | |
| "loss": 0.1853, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.616027750600373e-07, | |
| "loss": 0.1156, | |
| "step": 9344 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.606540365857273e-07, | |
| "loss": 0.179, | |
| "step": 9408 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.597052981114175e-07, | |
| "loss": 0.1978, | |
| "step": 9472 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.587565596371075e-07, | |
| "loss": 0.1735, | |
| "step": 9536 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.578078211627976e-07, | |
| "loss": 0.1579, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.568590826884876e-07, | |
| "loss": 0.1444, | |
| "step": 9664 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.559103442141777e-07, | |
| "loss": 0.1664, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.549616057398677e-07, | |
| "loss": 0.1715, | |
| "step": 9792 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.540128672655577e-07, | |
| "loss": 0.1189, | |
| "step": 9856 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.530641287912479e-07, | |
| "loss": 0.1255, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.521153903169379e-07, | |
| "loss": 0.2314, | |
| "step": 9984 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.511814758812891e-07, | |
| "loss": 0.18, | |
| "step": 10048 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.502327374069791e-07, | |
| "loss": 0.1948, | |
| "step": 10112 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.492839989326692e-07, | |
| "loss": 0.1763, | |
| "step": 10176 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.483352604583593e-07, | |
| "loss": 0.1984, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.473865219840493e-07, | |
| "loss": 0.169, | |
| "step": 10304 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.464377835097394e-07, | |
| "loss": 0.1268, | |
| "step": 10368 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.454890450354294e-07, | |
| "loss": 0.2261, | |
| "step": 10432 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.445403065611195e-07, | |
| "loss": 0.1751, | |
| "step": 10496 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.435915680868095e-07, | |
| "loss": 0.1758, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.426428296124995e-07, | |
| "loss": 0.1565, | |
| "step": 10624 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.416940911381897e-07, | |
| "loss": 0.0861, | |
| "step": 10688 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.407453526638797e-07, | |
| "loss": 0.1382, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.397966141895698e-07, | |
| "loss": 0.1268, | |
| "step": 10816 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.388478757152599e-07, | |
| "loss": 0.1418, | |
| "step": 10880 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.378991372409498e-07, | |
| "loss": 0.2417, | |
| "step": 10944 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.369503987666399e-07, | |
| "loss": 0.1285, | |
| "step": 11008 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.3600166029233e-07, | |
| "loss": 0.1719, | |
| "step": 11072 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.350529218180201e-07, | |
| "loss": 0.1432, | |
| "step": 11136 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.341041833437101e-07, | |
| "loss": 0.2443, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.331554448694002e-07, | |
| "loss": 0.1348, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.322067063950903e-07, | |
| "loss": 0.1251, | |
| "step": 11328 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.312579679207803e-07, | |
| "loss": 0.156, | |
| "step": 11392 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.303092294464703e-07, | |
| "loss": 0.3104, | |
| "step": 11456 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.293604909721605e-07, | |
| "loss": 0.1834, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.284117524978504e-07, | |
| "loss": 0.1312, | |
| "step": 11584 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.274630140235405e-07, | |
| "loss": 0.1026, | |
| "step": 11648 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.265142755492306e-07, | |
| "loss": 0.1805, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.255655370749206e-07, | |
| "loss": 0.2334, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.246167986006107e-07, | |
| "loss": 0.1606, | |
| "step": 11840 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.236680601263008e-07, | |
| "loss": 0.1009, | |
| "step": 11904 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.227193216519909e-07, | |
| "loss": 0.1337, | |
| "step": 11968 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.217705831776809e-07, | |
| "loss": 0.2247, | |
| "step": 12032 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.20821844703371e-07, | |
| "loss": 0.163, | |
| "step": 12096 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.19873106229061e-07, | |
| "loss": 0.1729, | |
| "step": 12160 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.18924367754751e-07, | |
| "loss": 0.2133, | |
| "step": 12224 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.179756292804412e-07, | |
| "loss": 0.2887, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.170268908061312e-07, | |
| "loss": 0.128, | |
| "step": 12352 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.160781523318212e-07, | |
| "loss": 0.2019, | |
| "step": 12416 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.151294138575113e-07, | |
| "loss": 0.128, | |
| "step": 12480 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.141806753832014e-07, | |
| "loss": 0.1888, | |
| "step": 12544 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.132319369088915e-07, | |
| "loss": 0.2176, | |
| "step": 12608 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.122831984345814e-07, | |
| "loss": 0.1555, | |
| "step": 12672 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.113344599602715e-07, | |
| "loss": 0.1948, | |
| "step": 12736 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.103857214859616e-07, | |
| "loss": 0.233, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.094369830116516e-07, | |
| "loss": 0.1574, | |
| "step": 12864 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.084882445373418e-07, | |
| "loss": 0.1377, | |
| "step": 12928 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.075395060630318e-07, | |
| "loss": 0.1563, | |
| "step": 12992 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.065907675887218e-07, | |
| "loss": 0.1345, | |
| "step": 13056 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.05642029114412e-07, | |
| "loss": 0.1489, | |
| "step": 13120 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.047081146787631e-07, | |
| "loss": 0.2524, | |
| "step": 13184 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.037742002431142e-07, | |
| "loss": 0.1534, | |
| "step": 13248 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.028254617688043e-07, | |
| "loss": 0.136, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.018767232944942e-07, | |
| "loss": 0.151, | |
| "step": 13376 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.009279848201844e-07, | |
| "loss": 0.1423, | |
| "step": 13440 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.999792463458744e-07, | |
| "loss": 0.2284, | |
| "step": 13504 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.990305078715644e-07, | |
| "loss": 0.1515, | |
| "step": 13568 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.980817693972546e-07, | |
| "loss": 0.1431, | |
| "step": 13632 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.971330309229446e-07, | |
| "loss": 0.1759, | |
| "step": 13696 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.961842924486347e-07, | |
| "loss": 0.2942, | |
| "step": 13760 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.952355539743248e-07, | |
| "loss": 0.1382, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.942868155000148e-07, | |
| "loss": 0.181, | |
| "step": 13888 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.933380770257049e-07, | |
| "loss": 0.2471, | |
| "step": 13952 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.923893385513949e-07, | |
| "loss": 0.1487, | |
| "step": 14016 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.91440600077085e-07, | |
| "loss": 0.1653, | |
| "step": 14080 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.90491861602775e-07, | |
| "loss": 0.193, | |
| "step": 14144 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.89543123128465e-07, | |
| "loss": 0.115, | |
| "step": 14208 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.885943846541552e-07, | |
| "loss": 0.1413, | |
| "step": 14272 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.876456461798452e-07, | |
| "loss": 0.1508, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.866969077055352e-07, | |
| "loss": 0.1752, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.857481692312254e-07, | |
| "loss": 0.2432, | |
| "step": 14464 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.847994307569153e-07, | |
| "loss": 0.1978, | |
| "step": 14528 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.838506922826054e-07, | |
| "loss": 0.1445, | |
| "step": 14592 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.829019538082955e-07, | |
| "loss": 0.1484, | |
| "step": 14656 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.819532153339855e-07, | |
| "loss": 0.1887, | |
| "step": 14720 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.810044768596756e-07, | |
| "loss": 0.216, | |
| "step": 14784 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.800557383853657e-07, | |
| "loss": 0.1803, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.791069999110558e-07, | |
| "loss": 0.1332, | |
| "step": 14912 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.781582614367458e-07, | |
| "loss": 0.2439, | |
| "step": 14976 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.772095229624358e-07, | |
| "loss": 0.1689, | |
| "step": 15040 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.76260784488126e-07, | |
| "loss": 0.1823, | |
| "step": 15104 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.753120460138159e-07, | |
| "loss": 0.1905, | |
| "step": 15168 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.74363307539506e-07, | |
| "loss": 0.2558, | |
| "step": 15232 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.734145690651961e-07, | |
| "loss": 0.1531, | |
| "step": 15296 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.724658305908861e-07, | |
| "loss": 0.1849, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.715170921165762e-07, | |
| "loss": 0.1317, | |
| "step": 15424 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.705683536422663e-07, | |
| "loss": 0.1096, | |
| "step": 15488 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.696196151679564e-07, | |
| "loss": 0.2193, | |
| "step": 15552 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.686708766936464e-07, | |
| "loss": 0.1658, | |
| "step": 15616 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.677221382193365e-07, | |
| "loss": 0.1553, | |
| "step": 15680 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.667733997450265e-07, | |
| "loss": 0.1772, | |
| "step": 15744 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.658246612707165e-07, | |
| "loss": 0.2147, | |
| "step": 15808 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.648759227964067e-07, | |
| "loss": 0.1096, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.639271843220967e-07, | |
| "loss": 0.1613, | |
| "step": 15936 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.629784458477867e-07, | |
| "loss": 0.1488, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.620297073734768e-07, | |
| "loss": 0.2256, | |
| "step": 16064 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.610809688991669e-07, | |
| "loss": 0.2512, | |
| "step": 16128 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.60132230424857e-07, | |
| "loss": 0.1264, | |
| "step": 16192 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.59183491950547e-07, | |
| "loss": 0.1162, | |
| "step": 16256 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.58234753476237e-07, | |
| "loss": 0.1401, | |
| "step": 16320 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.572860150019271e-07, | |
| "loss": 0.1336, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.563372765276171e-07, | |
| "loss": 0.1234, | |
| "step": 16448 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.553885380533072e-07, | |
| "loss": 0.1195, | |
| "step": 16512 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.544397995789973e-07, | |
| "loss": 0.2435, | |
| "step": 16576 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.534910611046873e-07, | |
| "loss": 0.1109, | |
| "step": 16640 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.525423226303775e-07, | |
| "loss": 0.2088, | |
| "step": 16704 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.515935841560675e-07, | |
| "loss": 0.141, | |
| "step": 16768 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.506448456817574e-07, | |
| "loss": 0.1428, | |
| "step": 16832 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.496961072074475e-07, | |
| "loss": 0.1505, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.487473687331376e-07, | |
| "loss": 0.2152, | |
| "step": 16960 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.477986302588277e-07, | |
| "loss": 0.2008, | |
| "step": 17024 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.468498917845177e-07, | |
| "loss": 0.1872, | |
| "step": 17088 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.459011533102078e-07, | |
| "loss": 0.1313, | |
| "step": 17152 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.449524148358979e-07, | |
| "loss": 0.1099, | |
| "step": 17216 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.440036763615879e-07, | |
| "loss": 0.138, | |
| "step": 17280 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.430549378872781e-07, | |
| "loss": 0.1871, | |
| "step": 17344 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.42106199412968e-07, | |
| "loss": 0.18, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.41157460938658e-07, | |
| "loss": 0.1337, | |
| "step": 17472 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.402087224643482e-07, | |
| "loss": 0.1222, | |
| "step": 17536 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.392599839900382e-07, | |
| "loss": 0.1434, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.383112455157283e-07, | |
| "loss": 0.1538, | |
| "step": 17664 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.373625070414183e-07, | |
| "loss": 0.1908, | |
| "step": 17728 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.364137685671084e-07, | |
| "loss": 0.1244, | |
| "step": 17792 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.354650300927985e-07, | |
| "loss": 0.1593, | |
| "step": 17856 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.345162916184885e-07, | |
| "loss": 0.1588, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.335675531441787e-07, | |
| "loss": 0.1639, | |
| "step": 17984 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.326188146698686e-07, | |
| "loss": 0.1431, | |
| "step": 18048 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.316700761955586e-07, | |
| "loss": 0.2002, | |
| "step": 18112 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.307213377212488e-07, | |
| "loss": 0.1761, | |
| "step": 18176 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.297725992469388e-07, | |
| "loss": 0.1597, | |
| "step": 18240 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.288238607726288e-07, | |
| "loss": 0.1952, | |
| "step": 18304 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.27875122298319e-07, | |
| "loss": 0.1843, | |
| "step": 18368 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.26926383824009e-07, | |
| "loss": 0.1032, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.259776453496991e-07, | |
| "loss": 0.1952, | |
| "step": 18496 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.25028906875389e-07, | |
| "loss": 0.193, | |
| "step": 18560 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.240801684010791e-07, | |
| "loss": 0.137, | |
| "step": 18624 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.231314299267692e-07, | |
| "loss": 0.1992, | |
| "step": 18688 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.221826914524592e-07, | |
| "loss": 0.138, | |
| "step": 18752 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.212339529781494e-07, | |
| "loss": 0.2263, | |
| "step": 18816 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.202852145038394e-07, | |
| "loss": 0.2101, | |
| "step": 18880 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.193364760295294e-07, | |
| "loss": 0.1731, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.183877375552196e-07, | |
| "loss": 0.1523, | |
| "step": 19008 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.174389990809096e-07, | |
| "loss": 0.1671, | |
| "step": 19072 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.164902606065996e-07, | |
| "loss": 0.1549, | |
| "step": 19136 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.155415221322897e-07, | |
| "loss": 0.1346, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.145927836579797e-07, | |
| "loss": 0.2403, | |
| "step": 19264 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.136588692223308e-07, | |
| "loss": 0.1909, | |
| "step": 19328 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.12710130748021e-07, | |
| "loss": 0.1801, | |
| "step": 19392 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.11761392273711e-07, | |
| "loss": 0.1196, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.10812653799401e-07, | |
| "loss": 0.0749, | |
| "step": 19520 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.098639153250912e-07, | |
| "loss": 0.1386, | |
| "step": 19584 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.089151768507812e-07, | |
| "loss": 0.219, | |
| "step": 19648 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.079664383764713e-07, | |
| "loss": 0.1572, | |
| "step": 19712 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.070176999021614e-07, | |
| "loss": 0.191, | |
| "step": 19776 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.060689614278513e-07, | |
| "loss": 0.1882, | |
| "step": 19840 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.051202229535414e-07, | |
| "loss": 0.1654, | |
| "step": 19904 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.041714844792315e-07, | |
| "loss": 0.1397, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.032227460049216e-07, | |
| "loss": 0.1948, | |
| "step": 20032 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.022740075306116e-07, | |
| "loss": 0.2171, | |
| "step": 20096 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.013252690563016e-07, | |
| "loss": 0.2474, | |
| "step": 20160 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.003765305819918e-07, | |
| "loss": 0.2014, | |
| "step": 20224 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.994277921076818e-07, | |
| "loss": 0.1256, | |
| "step": 20288 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.984790536333718e-07, | |
| "loss": 0.1634, | |
| "step": 20352 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.975303151590619e-07, | |
| "loss": 0.1672, | |
| "step": 20416 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.965815766847519e-07, | |
| "loss": 0.1773, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.95632838210442e-07, | |
| "loss": 0.1157, | |
| "step": 20544 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.946840997361321e-07, | |
| "loss": 0.2241, | |
| "step": 20608 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.937353612618222e-07, | |
| "loss": 0.1108, | |
| "step": 20672 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.927866227875122e-07, | |
| "loss": 0.1821, | |
| "step": 20736 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.918378843132023e-07, | |
| "loss": 0.1459, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.908891458388924e-07, | |
| "loss": 0.2022, | |
| "step": 20864 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.899404073645823e-07, | |
| "loss": 0.1864, | |
| "step": 20928 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.889916688902723e-07, | |
| "loss": 0.1436, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.880429304159625e-07, | |
| "loss": 0.1771, | |
| "step": 21056 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.870941919416525e-07, | |
| "loss": 0.1782, | |
| "step": 21120 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.861454534673426e-07, | |
| "loss": 0.1754, | |
| "step": 21184 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.851967149930327e-07, | |
| "loss": 0.1483, | |
| "step": 21248 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.842479765187227e-07, | |
| "loss": 0.1373, | |
| "step": 21312 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.833140620830739e-07, | |
| "loss": 0.219, | |
| "step": 21376 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.82365323608764e-07, | |
| "loss": 0.1474, | |
| "step": 21440 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.81416585134454e-07, | |
| "loss": 0.1713, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.80467846660144e-07, | |
| "loss": 0.1034, | |
| "step": 21568 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.795191081858341e-07, | |
| "loss": 0.185, | |
| "step": 21632 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.785703697115241e-07, | |
| "loss": 0.284, | |
| "step": 21696 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.776216312372143e-07, | |
| "loss": 0.1953, | |
| "step": 21760 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.766728927629043e-07, | |
| "loss": 0.168, | |
| "step": 21824 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.757241542885943e-07, | |
| "loss": 0.1852, | |
| "step": 21888 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.747754158142845e-07, | |
| "loss": 0.1358, | |
| "step": 21952 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.738266773399745e-07, | |
| "loss": 0.1885, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.728779388656646e-07, | |
| "loss": 0.22, | |
| "step": 22080 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.719292003913545e-07, | |
| "loss": 0.214, | |
| "step": 22144 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.709804619170446e-07, | |
| "loss": 0.1198, | |
| "step": 22208 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.700317234427347e-07, | |
| "loss": 0.1458, | |
| "step": 22272 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.690829849684247e-07, | |
| "loss": 0.2405, | |
| "step": 22336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.681342464941149e-07, | |
| "loss": 0.1162, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.671855080198049e-07, | |
| "loss": 0.1508, | |
| "step": 22464 | |
| } | |
| ], | |
| "logging_steps": 64, | |
| "max_steps": 67458, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 22486, | |
| "total_flos": 4.77416235737088e+19, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |