| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 40.0, | |
| "global_step": 3760, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.3297872340425533e-06, | |
| "loss": 29.6158, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "loss": 27.1775, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.98936170212766e-06, | |
| "loss": 24.7793, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.319148936170213e-06, | |
| "loss": 23.8977, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.648936170212766e-06, | |
| "loss": 20.9254, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.97872340425532e-06, | |
| "loss": 20.9818, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.308510638297872e-06, | |
| "loss": 16.5909, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0638297872340426e-05, | |
| "loss": 13.8147, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.196808510638298e-05, | |
| "loss": 10.0983, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.3297872340425532e-05, | |
| "loss": 6.1833, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.4627659574468085e-05, | |
| "loss": 3.6727, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.595744680851064e-05, | |
| "loss": 2.9481, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.7287234042553193e-05, | |
| "loss": 2.4666, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.8617021276595745e-05, | |
| "loss": 2.104, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.9946808510638297e-05, | |
| "loss": 1.8897, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1276595744680852e-05, | |
| "loss": 1.5951, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.2606382978723408e-05, | |
| "loss": 1.305, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.393617021276596e-05, | |
| "loss": 0.974, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.5265957446808515e-05, | |
| "loss": 0.7627, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.6595744680851064e-05, | |
| "loss": 0.6068, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.7925531914893616e-05, | |
| "loss": 0.4438, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.925531914893617e-05, | |
| "loss": 0.3635, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 3.0585106382978726e-05, | |
| "loss": 0.2884, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 3.191489361702128e-05, | |
| "loss": 0.2461, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.324468085106383e-05, | |
| "loss": 0.2003, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.4574468085106386e-05, | |
| "loss": 0.2406, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.590425531914894e-05, | |
| "loss": 0.1561, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.723404255319149e-05, | |
| "loss": 0.1476, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.8563829787234045e-05, | |
| "loss": 0.1365, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.9893617021276594e-05, | |
| "loss": 0.1181, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 4.1223404255319156e-05, | |
| "loss": 0.1274, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 4.2553191489361704e-05, | |
| "loss": 0.1283, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 4.388297872340425e-05, | |
| "loss": 0.1117, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.5212765957446815e-05, | |
| "loss": 0.1104, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 4.6542553191489364e-05, | |
| "loss": 0.1041, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 4.787234042553192e-05, | |
| "loss": 0.1026, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 4.920212765957447e-05, | |
| "loss": 0.1016, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.999982762728059e-05, | |
| "loss": 0.11, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 4.99978884614854e-05, | |
| "loss": 0.0975, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 4.999379483168148e-05, | |
| "loss": 0.0873, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 4.9987547090681766e-05, | |
| "loss": 0.0864, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 4.997914577695306e-05, | |
| "loss": 0.0955, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 4.996859161456965e-05, | |
| "loss": 0.0933, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.995588551315087e-05, | |
| "loss": 0.1315, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.994102856778274e-05, | |
| "loss": 0.0739, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 4.992402205892358e-05, | |
| "loss": 0.07, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.990486745229364e-05, | |
| "loss": 0.0766, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 4.9883566398748774e-05, | |
| "loss": 0.0718, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 4.986012073413816e-05, | |
| "loss": 0.0726, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 4.98345324791461e-05, | |
| "loss": 0.0725, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.9806803839117845e-05, | |
| "loss": 0.0813, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 4.97769372038695e-05, | |
| "loss": 0.074, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 4.974493514748213e-05, | |
| "loss": 0.0686, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 4.971080042807981e-05, | |
| "loss": 0.0754, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 4.9674535987592016e-05, | |
| "loss": 0.0781, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 4.96361449515e-05, | |
| "loss": 0.0683, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 4.959563062856744e-05, | |
| "loss": 0.0724, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 4.955299651055528e-05, | |
| "loss": 0.075, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 4.950824627192079e-05, | |
| "loss": 0.0618, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 4.946138376950087e-05, | |
| "loss": 0.0687, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 4.9412413042179615e-05, | |
| "loss": 0.0625, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 4.936133831054032e-05, | |
| "loss": 0.0635, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 4.930816397650159e-05, | |
| "loss": 0.0677, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 4.925289462293807e-05, | |
| "loss": 0.0571, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 4.9195535013285396e-05, | |
| "loss": 0.0683, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 4.9136090091129725e-05, | |
| "loss": 0.0701, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 4.907456497978156e-05, | |
| "loss": 0.0594, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 4.901096498183429e-05, | |
| "loss": 0.0546, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 4.894529557870716e-05, | |
| "loss": 0.0629, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 4.8877562430172815e-05, | |
| "loss": 0.0717, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 4.880777137386954e-05, | |
| "loss": 0.0611, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 4.873592842479813e-05, | |
| "loss": 0.0532, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 4.8662039774803476e-05, | |
| "loss": 0.0586, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 4.8586111792040935e-05, | |
| "loss": 0.0589, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 4.850815102042744e-05, | |
| "loss": 0.0553, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 4.842816417907759e-05, | |
| "loss": 0.0535, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 4.8346158161724454e-05, | |
| "loss": 0.0624, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 4.82621400361255e-05, | |
| "loss": 0.0536, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 4.817611704345344e-05, | |
| "loss": 0.0612, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 4.808809659767214e-05, | |
| "loss": 0.0535, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 4.7998086284897626e-05, | |
| "loss": 0.0511, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 4.7906093862744297e-05, | |
| "loss": 0.0542, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 4.781212725965629e-05, | |
| "loss": 0.0614, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 4.7716194574224214e-05, | |
| "loss": 0.0561, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 4.7618304074487124e-05, | |
| "loss": 0.0535, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 4.751846419721994e-05, | |
| "loss": 0.0506, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 4.7416683547206306e-05, | |
| "loss": 0.0522, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 4.731297089649703e-05, | |
| "loss": 0.0495, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 4.7207335183654e-05, | |
| "loss": 0.0538, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 4.709978551297984e-05, | |
| "loss": 0.049, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 4.699033115373322e-05, | |
| "loss": 0.0531, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 4.687898153933001e-05, | |
| "loss": 0.0556, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 4.6765746266530244e-05, | |
| "loss": 0.0573, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4.665063509461097e-05, | |
| "loss": 0.0494, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 4.653365794452521e-05, | |
| "loss": 0.0491, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 4.6414824898046896e-05, | |
| "loss": 0.0521, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 4.62941461969019e-05, | |
| "loss": 0.0518, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 4.617163224188544e-05, | |
| "loss": 0.0519, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 4.60472935919656e-05, | |
| "loss": 0.0537, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 4.5921140963373335e-05, | |
| "loss": 0.0481, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 4.579318522867886e-05, | |
| "loss": 0.0452, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 4.566343741585465e-05, | |
| "loss": 0.0484, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 4.553190870732488e-05, | |
| "loss": 0.0479, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 4.539861043900176e-05, | |
| "loss": 0.0446, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 4.526355409930847e-05, | |
| "loss": 0.0431, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 4.512675132818908e-05, | |
| "loss": 0.0486, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 4.498821391610527e-05, | |
| "loss": 0.0485, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 4.484795380302025e-05, | |
| "loss": 0.0421, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 11.6, | |
| "learning_rate": 4.470598307736964e-05, | |
| "loss": 0.0512, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 4.456231397501966e-05, | |
| "loss": 0.0479, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 4.441695887821253e-05, | |
| "loss": 0.0467, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4.426993031449934e-05, | |
| "loss": 0.0439, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.4121240955660304e-05, | |
| "loss": 0.043, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.397090361661268e-05, | |
| "loss": 0.0464, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 12.23, | |
| "learning_rate": 4.381893125430628e-05, | |
| "loss": 0.0485, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 12.34, | |
| "learning_rate": 4.3665336966606774e-05, | |
| "loss": 0.0423, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 4.351013399116681e-05, | |
| "loss": 0.0429, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 12.55, | |
| "learning_rate": 4.335333570428516e-05, | |
| "loss": 0.0478, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 4.319495561975386e-05, | |
| "loss": 0.0424, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 12.77, | |
| "learning_rate": 4.303500738769348e-05, | |
| "loss": 0.0467, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 4.287350479337672e-05, | |
| "loss": 0.0472, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 12.98, | |
| "learning_rate": 4.271046175604029e-05, | |
| "loss": 0.0443, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 13.09, | |
| "learning_rate": 4.254589232768528e-05, | |
| "loss": 0.0392, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 4.2379810691866064e-05, | |
| "loss": 0.0407, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 4.2212231162467886e-05, | |
| "loss": 0.0409, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 4.2043168182473215e-05, | |
| "loss": 0.0435, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "learning_rate": 4.187263632271695e-05, | |
| "loss": 0.0434, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 4.170065028063062e-05, | |
| "loss": 0.0441, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 4.152722487897571e-05, | |
| "loss": 0.044, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 13.83, | |
| "learning_rate": 4.1352375064566054e-05, | |
| "loss": 0.0446, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 4.1176115906979774e-05, | |
| "loss": 0.0464, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 4.0998462597260346e-05, | |
| "loss": 0.0412, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 4.0819430446607465e-05, | |
| "loss": 0.0382, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 4.0639034885057376e-05, | |
| "loss": 0.0414, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "learning_rate": 4.045729146015303e-05, | |
| "loss": 0.0441, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 4.027421583560414e-05, | |
| "loss": 0.0435, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 4.0089823789937106e-05, | |
| "loss": 0.0424, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 3.990413121513526e-05, | |
| "loss": 0.0422, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 3.9717154115269076e-05, | |
| "loss": 0.0389, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 3.952890860511692e-05, | |
| "loss": 0.0423, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.933941090877615e-05, | |
| "loss": 0.0419, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 3.914867735826489e-05, | |
| "loss": 0.0425, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 15.21, | |
| "learning_rate": 3.895672439211431e-05, | |
| "loss": 0.0383, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 3.876356855395202e-05, | |
| "loss": 0.0388, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 15.43, | |
| "learning_rate": 3.856922649107613e-05, | |
| "loss": 0.0406, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "learning_rate": 3.8373714953020506e-05, | |
| "loss": 0.0415, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 3.817705079011125e-05, | |
| "loss": 0.0401, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 15.74, | |
| "learning_rate": 3.797925095201438e-05, | |
| "loss": 0.0358, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 3.778033248627506e-05, | |
| "loss": 0.0403, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 15.96, | |
| "learning_rate": 3.758031253684827e-05, | |
| "loss": 0.0431, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 16.06, | |
| "learning_rate": 3.737920834262134e-05, | |
| "loss": 0.0392, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 3.717703723592811e-05, | |
| "loss": 0.041, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "learning_rate": 3.69738166410552e-05, | |
| "loss": 0.0353, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 16.38, | |
| "learning_rate": 3.676956407274021e-05, | |
| "loss": 0.0388, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 16.49, | |
| "learning_rate": 3.6564297134662264e-05, | |
| "loss": 0.038, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 16.6, | |
| "learning_rate": 3.63580335179248e-05, | |
| "loss": 0.038, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 3.615079099953081e-05, | |
| "loss": 0.0352, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 3.59425874408508e-05, | |
| "loss": 0.0404, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 16.91, | |
| "learning_rate": 3.5733440786083304e-05, | |
| "loss": 0.0378, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 3.552336906070838e-05, | |
| "loss": 0.0435, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 17.13, | |
| "learning_rate": 3.531239036993409e-05, | |
| "loss": 0.0352, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 3.5100522897136065e-05, | |
| "loss": 0.0416, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 3.4887784902290366e-05, | |
| "loss": 0.0336, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "learning_rate": 3.467419472039972e-05, | |
| "loss": 0.0318, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "learning_rate": 3.4459770759913314e-05, | |
| "loss": 0.035, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 17.66, | |
| "learning_rate": 3.424453150114023e-05, | |
| "loss": 0.0366, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 3.4028495494656746e-05, | |
| "loss": 0.0458, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 3.381168135970749e-05, | |
| "loss": 0.04, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 17.98, | |
| "learning_rate": 3.3594107782600753e-05, | |
| "loss": 0.0381, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 18.09, | |
| "learning_rate": 3.337579351509797e-05, | |
| "loss": 0.0345, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 18.19, | |
| "learning_rate": 3.315675737279763e-05, | |
| "loss": 0.0358, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 3.293701823351357e-05, | |
| "loss": 0.0346, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 3.271659503564804e-05, | |
| "loss": 0.0383, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 18.51, | |
| "learning_rate": 3.249550677655945e-05, | |
| "loss": 0.0332, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 18.62, | |
| "learning_rate": 3.227377251092507e-05, | |
| "loss": 0.0348, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 18.72, | |
| "learning_rate": 3.205141134909878e-05, | |
| "loss": 0.0382, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 18.83, | |
| "learning_rate": 3.182844245546403e-05, | |
| "loss": 0.0377, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 18.94, | |
| "learning_rate": 3.160488504678216e-05, | |
| "loss": 0.0366, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 3.138075839053615e-05, | |
| "loss": 0.0382, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 3.11560818032701e-05, | |
| "loss": 0.0332, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 3.093087464892434e-05, | |
| "loss": 0.0328, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 3.070515633716658e-05, | |
| "loss": 0.0382, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 19.47, | |
| "learning_rate": 3.047894632171907e-05, | |
| "loss": 0.0372, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 19.57, | |
| "learning_rate": 3.0252264098681947e-05, | |
| "loss": 0.0347, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 19.68, | |
| "learning_rate": 3.0025129204852964e-05, | |
| "loss": 0.0355, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "learning_rate": 2.979756121604366e-05, | |
| "loss": 0.0353, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 2.956957974539226e-05, | |
| "loss": 0.0352, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 2.9341204441673266e-05, | |
| "loss": 0.0362, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 20.11, | |
| "learning_rate": 2.911245498760398e-05, | |
| "loss": 0.0385, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 20.21, | |
| "learning_rate": 2.8883351098148216e-05, | |
| "loss": 0.0308, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 20.32, | |
| "learning_rate": 2.8653912518817055e-05, | |
| "loss": 0.0326, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 20.43, | |
| "learning_rate": 2.8424159023967135e-05, | |
| "loss": 0.0359, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 20.53, | |
| "learning_rate": 2.8194110415096335e-05, | |
| "loss": 0.0327, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 20.64, | |
| "learning_rate": 2.7963786519137203e-05, | |
| "loss": 0.0325, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 20.74, | |
| "learning_rate": 2.773320718674812e-05, | |
| "loss": 0.0358, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 20.85, | |
| "learning_rate": 2.7502392290602463e-05, | |
| "loss": 0.031, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 20.96, | |
| "learning_rate": 2.727136172367589e-05, | |
| "loss": 0.0357, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 21.06, | |
| "learning_rate": 2.7040135397531808e-05, | |
| "loss": 0.033, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 21.17, | |
| "learning_rate": 2.6808733240605323e-05, | |
| "loss": 0.0295, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "learning_rate": 2.657717519648562e-05, | |
| "loss": 0.0382, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 21.38, | |
| "learning_rate": 2.6345481222197205e-05, | |
| "loss": 0.0317, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 21.49, | |
| "learning_rate": 2.6113671286479812e-05, | |
| "loss": 0.0314, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 21.6, | |
| "learning_rate": 2.588176536806739e-05, | |
| "loss": 0.0328, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 2.5649783453966274e-05, | |
| "loss": 0.0337, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 21.81, | |
| "learning_rate": 2.5417745537732523e-05, | |
| "loss": 0.0347, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 21.91, | |
| "learning_rate": 2.5185671617748812e-05, | |
| "loss": 0.0319, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 2.4953581695500812e-05, | |
| "loss": 0.0343, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 22.13, | |
| "learning_rate": 2.4721495773853366e-05, | |
| "loss": 0.0336, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 22.23, | |
| "learning_rate": 2.4489433855326518e-05, | |
| "loss": 0.031, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 22.34, | |
| "learning_rate": 2.4257415940371587e-05, | |
| "loss": 0.0349, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 22.45, | |
| "learning_rate": 2.4025462025647407e-05, | |
| "loss": 0.0319, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 22.55, | |
| "learning_rate": 2.3793592102296915e-05, | |
| "loss": 0.0311, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 22.66, | |
| "learning_rate": 2.3561826154224144e-05, | |
| "loss": 0.0331, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 22.77, | |
| "learning_rate": 2.333018415637196e-05, | |
| "loss": 0.0323, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 22.87, | |
| "learning_rate": 2.309868607300048e-05, | |
| "loss": 0.0349, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 22.98, | |
| "learning_rate": 2.2867351855966383e-05, | |
| "loss": 0.0331, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 23.09, | |
| "learning_rate": 2.263620144300345e-05, | |
| "loss": 0.0302, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 23.19, | |
| "learning_rate": 2.240525475600412e-05, | |
| "loss": 0.0353, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 23.3, | |
| "learning_rate": 2.2174531699302562e-05, | |
| "loss": 0.0321, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 23.4, | |
| "learning_rate": 2.1944052157959143e-05, | |
| "loss": 0.0336, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 23.51, | |
| "learning_rate": 2.171383599604668e-05, | |
| "loss": 0.0309, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 23.62, | |
| "learning_rate": 2.1483903054938426e-05, | |
| "loss": 0.0315, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 23.72, | |
| "learning_rate": 2.1254273151597966e-05, | |
| "loss": 0.0303, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 23.83, | |
| "learning_rate": 2.1024966076871364e-05, | |
| "loss": 0.0322, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 23.94, | |
| "learning_rate": 2.0796001593781417e-05, | |
| "loss": 0.034, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 24.04, | |
| "learning_rate": 2.0567399435824367e-05, | |
| "loss": 0.0313, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 24.15, | |
| "learning_rate": 2.033917930526916e-05, | |
| "loss": 0.0311, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 24.26, | |
| "learning_rate": 2.0111360871459388e-05, | |
| "loss": 0.0286, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 24.36, | |
| "learning_rate": 1.9883963769118097e-05, | |
| "loss": 0.0307, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 24.47, | |
| "learning_rate": 1.9657007596655477e-05, | |
| "loss": 0.0299, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 24.57, | |
| "learning_rate": 1.9430511914479853e-05, | |
| "loss": 0.0319, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 24.68, | |
| "learning_rate": 1.920449624331179e-05, | |
| "loss": 0.0328, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 24.79, | |
| "learning_rate": 1.8978980062501712e-05, | |
| "loss": 0.0333, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 24.89, | |
| "learning_rate": 1.8753982808351007e-05, | |
| "loss": 0.0299, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 1.852952387243698e-05, | |
| "loss": 0.0284, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 25.11, | |
| "learning_rate": 1.830562259994152e-05, | |
| "loss": 0.0286, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 25.21, | |
| "learning_rate": 1.808229828798379e-05, | |
| "loss": 0.0331, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 25.32, | |
| "learning_rate": 1.7859570183957174e-05, | |
| "loss": 0.0272, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 25.43, | |
| "learning_rate": 1.7637457483870333e-05, | |
| "loss": 0.0273, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 25.53, | |
| "learning_rate": 1.7415979330692857e-05, | |
| "loss": 0.0319, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 25.64, | |
| "learning_rate": 1.7195154812705343e-05, | |
| "loss": 0.0272, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 25.74, | |
| "learning_rate": 1.697500296185432e-05, | |
| "loss": 0.0351, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 25.85, | |
| "learning_rate": 1.6755542752111925e-05, | |
| "loss": 0.0303, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 25.96, | |
| "learning_rate": 1.6536793097840615e-05, | |
| "loss": 0.0332, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 26.06, | |
| "learning_rate": 1.631877285216304e-05, | |
| "loss": 0.0306, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 26.17, | |
| "learning_rate": 1.610150080533716e-05, | |
| "loss": 0.0311, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 26.28, | |
| "learning_rate": 1.5884995683136765e-05, | |
| "loss": 0.0321, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 26.38, | |
| "learning_rate": 1.5669276145237633e-05, | |
| "loss": 0.0351, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 26.49, | |
| "learning_rate": 1.5454360783609278e-05, | |
| "loss": 0.026, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 26.6, | |
| "learning_rate": 1.5240268120912631e-05, | |
| "loss": 0.0273, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 26.7, | |
| "learning_rate": 1.5027016608903594e-05, | |
| "loss": 0.0325, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 26.81, | |
| "learning_rate": 1.4814624626842798e-05, | |
| "loss": 0.029, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 26.91, | |
| "learning_rate": 1.4603110479911592e-05, | |
| "loss": 0.0304, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 1.4392492397634321e-05, | |
| "loss": 0.0297, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 27.13, | |
| "learning_rate": 1.418278853230729e-05, | |
| "loss": 0.0271, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 27.23, | |
| "learning_rate": 1.3974016957434208e-05, | |
| "loss": 0.0327, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 27.34, | |
| "learning_rate": 1.3766195666168603e-05, | |
| "loss": 0.0275, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 27.45, | |
| "learning_rate": 1.3559342569762933e-05, | |
| "loss": 0.0319, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 27.55, | |
| "learning_rate": 1.3353475496025047e-05, | |
| "loss": 0.0286, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 27.66, | |
| "learning_rate": 1.314861218778156e-05, | |
| "loss": 0.0326, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 27.77, | |
| "learning_rate": 1.2944770301348727e-05, | |
| "loss": 0.0289, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 27.87, | |
| "learning_rate": 1.2741967405010698e-05, | |
| "loss": 0.0294, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 27.98, | |
| "learning_rate": 1.254022097750539e-05, | |
| "loss": 0.0308, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 28.09, | |
| "learning_rate": 1.23395484065181e-05, | |
| "loss": 0.0334, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 28.19, | |
| "learning_rate": 1.213996698718283e-05, | |
| "loss": 0.0279, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 28.3, | |
| "learning_rate": 1.1941493920591828e-05, | |
| "loss": 0.0293, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 28.4, | |
| "learning_rate": 1.1744146312313012e-05, | |
| "loss": 0.0274, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 28.51, | |
| "learning_rate": 1.1547941170915686e-05, | |
| "loss": 0.0303, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 28.62, | |
| "learning_rate": 1.1352895406504751e-05, | |
| "loss": 0.0302, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 28.72, | |
| "learning_rate": 1.1159025829263183e-05, | |
| "loss": 0.0283, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 28.83, | |
| "learning_rate": 1.0966349148003325e-05, | |
| "loss": 0.0313, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 28.94, | |
| "learning_rate": 1.0774881968726722e-05, | |
| "loss": 0.0269, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 29.04, | |
| "learning_rate": 1.0584640793192995e-05, | |
| "loss": 0.0336, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 29.15, | |
| "learning_rate": 1.0395642017497648e-05, | |
| "loss": 0.0293, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 29.26, | |
| "learning_rate": 1.020790193065883e-05, | |
| "loss": 0.027, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 29.36, | |
| "learning_rate": 1.0021436713213606e-05, | |
| "loss": 0.0281, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 29.47, | |
| "learning_rate": 9.836262435823316e-06, | |
| "loss": 0.031, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 29.57, | |
| "learning_rate": 9.652395057888555e-06, | |
| "loss": 0.0284, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 29.68, | |
| "learning_rate": 9.469850426173688e-06, | |
| "loss": 0.0292, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 29.79, | |
| "learning_rate": 9.288644273441083e-06, | |
| "loss": 0.0284, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 29.89, | |
| "learning_rate": 9.108792217095203e-06, | |
| "loss": 0.0285, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 8.930309757836517e-06, | |
| "loss": 0.0257, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 30.11, | |
| "learning_rate": 8.753212278325684e-06, | |
| "loss": 0.0309, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 30.21, | |
| "learning_rate": 8.577515041857681e-06, | |
| "loss": 0.0274, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 30.32, | |
| "learning_rate": 8.40323319104637e-06, | |
| "loss": 0.0309, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 30.43, | |
| "learning_rate": 8.23038174651942e-06, | |
| "loss": 0.0274, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 30.53, | |
| "learning_rate": 8.05897560562371e-06, | |
| "loss": 0.0263, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 30.64, | |
| "learning_rate": 7.889029541141465e-06, | |
| "loss": 0.0293, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 30.74, | |
| "learning_rate": 7.720558200016922e-06, | |
| "loss": 0.0294, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 30.85, | |
| "learning_rate": 7.553576102094115e-06, | |
| "loss": 0.0281, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 30.96, | |
| "learning_rate": 7.3880976388653625e-06, | |
| "loss": 0.0303, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 31.06, | |
| "learning_rate": 7.224137072230982e-06, | |
| "loss": 0.028, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 31.17, | |
| "learning_rate": 7.061708533270089e-06, | |
| "loss": 0.0284, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 31.28, | |
| "learning_rate": 6.900826021022708e-06, | |
| "loss": 0.0294, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 31.38, | |
| "learning_rate": 6.741503401283272e-06, | |
| "loss": 0.0293, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 31.49, | |
| "learning_rate": 6.583754405405529e-06, | |
| "loss": 0.0247, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 31.6, | |
| "learning_rate": 6.427592629119172e-06, | |
| "loss": 0.0257, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 31.7, | |
| "learning_rate": 6.273031531358034e-06, | |
| "loss": 0.0275, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 31.81, | |
| "learning_rate": 6.120084433100084e-06, | |
| "loss": 0.0302, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 31.91, | |
| "learning_rate": 5.968764516219452e-06, | |
| "loss": 0.0291, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 5.819084822350224e-06, | |
| "loss": 0.0302, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 32.13, | |
| "learning_rate": 5.671058251762529e-06, | |
| "loss": 0.03, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 32.23, | |
| "learning_rate": 5.5246975622506354e-06, | |
| "loss": 0.0304, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 32.34, | |
| "learning_rate": 5.380015368033476e-06, | |
| "loss": 0.0243, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 32.45, | |
| "learning_rate": 5.2370241386674705e-06, | |
| "loss": 0.0292, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 32.55, | |
| "learning_rate": 5.0957361979717764e-06, | |
| "loss": 0.0295, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 32.66, | |
| "learning_rate": 4.956163722966223e-06, | |
| "loss": 0.0266, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 32.77, | |
| "learning_rate": 4.818318742821765e-06, | |
| "loss": 0.0271, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 32.87, | |
| "learning_rate": 4.682213137823774e-06, | |
| "loss": 0.0287, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 32.98, | |
| "learning_rate": 4.547858638348107e-06, | |
| "loss": 0.0282, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 33.09, | |
| "learning_rate": 4.415266823850117e-06, | |
| "loss": 0.029, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 33.19, | |
| "learning_rate": 4.284449121866699e-06, | |
| "loss": 0.0273, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 33.3, | |
| "learning_rate": 4.1554168070313255e-06, | |
| "loss": 0.0315, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 33.4, | |
| "learning_rate": 4.028181000102424e-06, | |
| "loss": 0.0242, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 33.51, | |
| "learning_rate": 3.90275266700485e-06, | |
| "loss": 0.0247, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 33.62, | |
| "learning_rate": 3.779142617884823e-06, | |
| "loss": 0.0271, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 33.72, | |
| "learning_rate": 3.65736150617822e-06, | |
| "loss": 0.0247, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 33.83, | |
| "learning_rate": 3.5374198276924175e-06, | |
| "loss": 0.0273, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 33.94, | |
| "learning_rate": 3.4193279197017094e-06, | |
| "loss": 0.0288, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 34.04, | |
| "learning_rate": 3.3030959600563323e-06, | |
| "loss": 0.0313, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 34.15, | |
| "learning_rate": 3.1887339663053463e-06, | |
| "loss": 0.0266, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 34.26, | |
| "learning_rate": 3.0762517948332127e-06, | |
| "loss": 0.0283, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 34.36, | |
| "learning_rate": 2.965659140010335e-06, | |
| "loss": 0.0263, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 34.47, | |
| "learning_rate": 2.856965533357539e-06, | |
| "loss": 0.0282, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 34.57, | |
| "learning_rate": 2.7501803427245816e-06, | |
| "loss": 0.0257, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 34.68, | |
| "learning_rate": 2.645312771482805e-06, | |
| "loss": 0.0293, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 34.79, | |
| "learning_rate": 2.542371857731879e-06, | |
| "loss": 0.0295, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 34.89, | |
| "learning_rate": 2.441366473520909e-06, | |
| "loss": 0.026, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 2.3423053240837515e-06, | |
| "loss": 0.0295, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 35.11, | |
| "learning_rate": 2.245196947088754e-06, | |
| "loss": 0.0311, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 35.21, | |
| "learning_rate": 2.1500497119029322e-06, | |
| "loss": 0.0279, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 35.32, | |
| "learning_rate": 2.056871818870648e-06, | |
| "loss": 0.0261, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 35.43, | |
| "learning_rate": 1.9656712986068664e-06, | |
| "loss": 0.0285, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 35.53, | |
| "learning_rate": 1.8764560113050027e-06, | |
| "loss": 0.0282, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 35.64, | |
| "learning_rate": 1.7892336460595043e-06, | |
| "loss": 0.0258, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 35.74, | |
| "learning_rate": 1.7040117202031774e-06, | |
| "loss": 0.0256, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 35.85, | |
| "learning_rate": 1.620797578659261e-06, | |
| "loss": 0.0287, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 35.96, | |
| "learning_rate": 1.5395983933084196e-06, | |
| "loss": 0.0281, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 36.06, | |
| "learning_rate": 1.4604211623706315e-06, | |
| "loss": 0.0263, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 36.17, | |
| "learning_rate": 1.3832727098020332e-06, | |
| "loss": 0.0282, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 36.28, | |
| "learning_rate": 1.308159684706789e-06, | |
| "loss": 0.029, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 36.38, | |
| "learning_rate": 1.2350885607640445e-06, | |
| "loss": 0.0242, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 36.49, | |
| "learning_rate": 1.1640656356699808e-06, | |
| "loss": 0.0262, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 36.6, | |
| "learning_rate": 1.0950970305950353e-06, | |
| "loss": 0.0247, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 36.7, | |
| "learning_rate": 1.028188689656351e-06, | |
| "loss": 0.0285, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 36.81, | |
| "learning_rate": 9.633463794054776e-07, | |
| "loss": 0.0274, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 36.91, | |
| "learning_rate": 9.005756883313759e-07, | |
| "loss": 0.0301, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 37.02, | |
| "learning_rate": 8.39882026378766e-07, | |
| "loss": 0.0277, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 37.13, | |
| "learning_rate": 7.812706244818669e-07, | |
| "loss": 0.0295, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 37.23, | |
| "learning_rate": 7.247465341135795e-07, | |
| "loss": 0.0279, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 37.34, | |
| "learning_rate": 6.703146268500827e-07, | |
| "loss": 0.0255, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 37.45, | |
| "learning_rate": 6.179795939510263e-07, | |
| "loss": 0.0286, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 37.55, | |
| "learning_rate": 5.677459459551587e-07, | |
| "loss": 0.024, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 37.66, | |
| "learning_rate": 5.1961801229162e-07, | |
| "loss": 0.0267, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 37.77, | |
| "learning_rate": 4.7359994090678816e-07, | |
| "loss": 0.0289, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 37.87, | |
| "learning_rate": 4.2969569790678456e-07, | |
| "loss": 0.0278, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 37.98, | |
| "learning_rate": 3.8790906721566935e-07, | |
| "loss": 0.0259, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 38.09, | |
| "learning_rate": 3.482436502492858e-07, | |
| "loss": 0.03, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 38.19, | |
| "learning_rate": 3.1070286560491436e-07, | |
| "loss": 0.0289, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 38.3, | |
| "learning_rate": 2.7528994876659663e-07, | |
| "loss": 0.0271, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 38.4, | |
| "learning_rate": 2.420079518263119e-07, | |
| "loss": 0.0278, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 38.51, | |
| "learning_rate": 2.1085974322091473e-07, | |
| "loss": 0.0258, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 38.62, | |
| "learning_rate": 1.8184800748491937e-07, | |
| "loss": 0.0283, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 38.72, | |
| "learning_rate": 1.5497524501913163e-07, | |
| "loss": 0.0249, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 38.83, | |
| "learning_rate": 1.3024377187515203e-07, | |
| "loss": 0.0275, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 38.94, | |
| "learning_rate": 1.076557195557576e-07, | |
| "loss": 0.0293, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 39.04, | |
| "learning_rate": 8.721303483121002e-08, | |
| "loss": 0.0278, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 39.15, | |
| "learning_rate": 6.891747957145922e-08, | |
| "loss": 0.0286, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 39.26, | |
| "learning_rate": 5.277063059429543e-08, | |
| "loss": 0.0295, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 39.36, | |
| "learning_rate": 3.877387952945788e-08, | |
| "loss": 0.0272, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 39.47, | |
| "learning_rate": 2.692843269869183e-08, | |
| "loss": 0.0277, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 39.57, | |
| "learning_rate": 1.7235311011778997e-08, | |
| "loss": 0.0259, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 39.68, | |
| "learning_rate": 9.695349878549586e-09, | |
| "loss": 0.0247, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 39.79, | |
| "learning_rate": 4.309199136889874e-09, | |
| "loss": 0.0317, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 39.89, | |
| "learning_rate": 1.077322996717589e-09, | |
| "loss": 0.0285, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0255, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "step": 3760, | |
| "total_flos": 4.579924393259827e+16, | |
| "train_loss": 0.608423255047107, | |
| "train_runtime": 971.1571, | |
| "train_samples_per_second": 247.457, | |
| "train_steps_per_second": 3.872 | |
| } | |
| ], | |
| "max_steps": 3760, | |
| "num_train_epochs": 40, | |
| "total_flos": 4.579924393259827e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |