search

KovenYu / MAR

Pytorch code for our CVPR'19 (oral) work: Unsupervised person re-identification by soft multilabel learning
https://kovenyu.com/publication/2019-cvpr-mar/
315 stars 83 forks source link

Issue in utils _update_centers function #22

Closed SkyJiashu closed 5 years ago

SkyJiashu commented 5 years ago

Thanks for your work and released code.

When I run the code by following your instruction, there is a special case.

In _update_centers function in utils.py, there is a if condition sentence: if len(ml_in_v) == 1: then continue.

In the special case, the univiews.shape is [1] and len(ml_in_v) is 1. Then the network will stop here because means will become empty.

I use the processed dataset provided by you. The ml_Market is computed by my machine.

Thanks for your help in advance.

KovenYu commented 5 years ago

Hi @SkyJiashu , thank you for your attention. "if len(ml_in_v) == 1: then continue" means that the examples in each view should be more than one, because std needs more than 1 examples. If you have a large batch size this should not be a problem (probabilistically).

SkyJiashu commented 5 years ago

Thanks for your reply. I just tried batch_szie 8. I will try batch_size 368 now.

SkyJiashu commented 5 years ago

Iter: [000/717] Freq 8.3 loss_total 9.580 loss_target 0.000 loss_ml 10086.398 loss_st 0.444 loss_source 0.066 [2019-09-09 07:58:39] Iter: [100/717] Freq 224.0 loss_total 7.650 loss_target 0.000 loss_ml 3460.829 loss_st 0.458 loss_source 0.049 [2019-09-09 08:00:34] Iter: [200/717] Freq 258.6 loss_total 7.306 loss_target 0.000 loss_ml 2744.283 loss_st 0.469 loss_source 0.042 [2019-09-09 08:02:27] Iter: [300/717] Freq 271.3 loss_total 7.281 loss_target 0.000 loss_ml 2461.099 loss_st 0.477 loss_source 0.041 [2019-09-09 08:04:22] Iter: [400/717] Freq 280.4 loss_total 7.227 loss_target 0.000 loss_ml 2292.613 loss_st 0.483 loss_source 0.040 [2019-09-09 08:06:13] Iter: [500/717] Freq 286.8 loss_total 7.223 loss_target 0.000 loss_ml 2191.325 loss_st 0.486 loss_source 0.039 [2019-09-09 08:08:02] Iter: [600/717] Freq 290.4 loss_total 7.233 loss_target 0.000 loss_ml 2109.329 loss_st 0.489 loss_source 0.039 [2019-09-09 08:09:54] Iter: [700/717] Freq 293.5 loss_total 7.233 loss_target 0.000 loss_ml 2044.922 loss_st 0.491 loss_source 0.039 [2019-09-09 08:11:44] Train loss_total 7.222 loss_target 0.000 loss_ml 2036.806 loss_st 0.492 loss_source 0.039

==>>[2019-09-09 08:12:07] [Epoch=001/020] Stage 1, [Need: 04:34:09] Iter: [000/717] Freq 87.2 loss_total 7.943 loss_target 0.687 loss_ml 1546.870 loss_st 0.482 loss_source 0.043 [2019-09-09 08:12:11] Iter: [100/717] Freq 204.2 loss_total 7.157 loss_target 0.617 loss_ml 1642.479 loss_st 0.495 loss_source 0.026 [2019-09-09 08:14:59] Iter: [200/717] Freq 207.9 loss_total 7.154 loss_target 0.622 loss_ml 1602.348 loss_st 0.493 loss_source 0.026 [2019-09-09 08:17:42] Iter: [300/717] Freq 208.5 loss_total 7.160 loss_target 0.624 loss_ml 1587.674 loss_st 0.493 loss_source 0.026 [2019-09-09 08:20:27] Iter: [400/717] Freq 208.9 loss_total 7.188 loss_target 0.623 loss_ml 1575.590 loss_st 0.492 loss_source 0.027 [2019-09-09 08:23:12] Iter: [500/717] Freq 209.5 loss_total 7.202 loss_target 0.624 loss_ml 1563.952 loss_st 0.493 loss_source 0.027 [2019-09-09 08:25:55] Iter: [600/717] Freq 209.7 loss_total 7.216 loss_target 0.624 loss_ml 1552.959 loss_st 0.492 loss_source 0.028 [2019-09-09 08:28:39] Iter: [700/717] Freq 210.2 loss_total 7.244 loss_target 0.624 loss_ml 1546.524 loss_st 0.492 loss_source 0.028 [2019-09-09 08:31:21] Train loss_total 7.257 loss_target 0.624 loss_ml 1547.583 loss_st 0.492 loss_source 0.029

==>>[2019-09-09 08:31:51] [Epoch=002/020] Stage 1, [Need: 05:07:27] Iter: [000/717] Freq 83.6 loss_total 6.046 loss_target 0.622 loss_ml 1225.468 loss_st 0.482 loss_source 0.008 [2019-09-09 08:31:55] Iter: [100/717] Freq 207.7 loss_total 6.944 loss_target 0.626 loss_ml 1485.158 loss_st 0.487 loss_source 0.024 [2019-09-09 08:34:40] Iter: [200/717] Freq 209.6 loss_total 6.898 loss_target 0.624 loss_ml 1496.573 loss_st 0.486 loss_source 0.023 [2019-09-09 08:37:23] Iter: [300/717] Freq 209.2 loss_total 6.973 loss_target 0.623 loss_ml 1494.621 loss_st 0.485 loss_source 0.025 [2019-09-09 08:40:09] Iter: [400/717] Freq 210.0 loss_total 6.936 loss_target 0.623 loss_ml 1488.762 loss_st 0.484 loss_source 0.024 [2019-09-09 08:42:52] Iter: [500/717] Freq 210.9 loss_total 6.957 loss_target 0.623 loss_ml 1482.258 loss_st 0.483 loss_source 0.025 [2019-09-09 08:45:33] Iter: [600/717] Freq 210.7 loss_total 6.976 loss_target 0.623 loss_ml 1475.077 loss_st 0.483 loss_source 0.025 [2019-09-09 08:48:18] Iter: [700/717] Freq 211.3 loss_total 7.037 loss_target 0.623 loss_ml 1470.636 loss_st 0.483 loss_source 0.026 [2019-09-09 08:50:59] Train loss_total 7.045 loss_target 0.623 loss_ml 1470.176 loss_st 0.483 loss_source 0.027

==>>[2019-09-09 08:51:31] [Epoch=003/020] Stage 1, [Need: 05:05:02] Iter: [000/717] Freq 83.6 loss_total 6.677 loss_target 0.618 loss_ml 1272.430 loss_st 0.503 loss_source 0.016 [2019-09-09 08:51:36] Iter: [100/717] Freq 206.2 loss_total 6.693 loss_target 0.623 loss_ml 1455.757 loss_st 0.480 loss_source 0.020 [2019-09-09 08:54:21] Iter: [200/717] Freq 209.6 loss_total 6.660 loss_target 0.625 loss_ml 1441.812 loss_st 0.476 loss_source 0.020 [2019-09-09 08:57:03] Iter: [300/717] Freq 208.6 loss_total 6.660 loss_target 0.625 loss_ml 1435.751 loss_st 0.476 loss_source 0.020 [2019-09-09 08:59:51] Iter: [400/717] Freq 209.4 loss_total 6.735 loss_target 0.624 loss_ml 1439.949 loss_st 0.476 loss_source 0.022 [2019-09-09 09:02:34] Iter: [500/717] Freq 209.6 loss_total 6.787 loss_target 0.624 loss_ml 1435.292 loss_st 0.475 loss_source 0.023 [2019-09-09 09:05:19] Iter: [600/717] Freq 208.9 loss_total 6.800 loss_target 0.623 loss_ml 1432.070 loss_st 0.474 loss_source 0.024 [2019-09-09 09:08:07] Iter: [700/717] Freq 209.5 loss_total 6.836 loss_target 0.623 loss_ml 1431.110 loss_st 0.474 loss_source 0.024 [2019-09-09 09:10:49] Train loss_total 6.836 loss_target 0.623 loss_ml 1430.450 loss_st 0.474 loss_source 0.024

==>>[2019-09-09 09:11:22] [Epoch=004/020] Stage 1, [Need: 04:54:39] Iter: [000/717] Freq 86.0 loss_total 6.225 loss_target 0.641 loss_ml 1439.829 loss_st 0.483 loss_source 0.010 [2019-09-09 09:11:26] Iter: [100/717] Freq 210.5 loss_total 6.341 loss_target 0.622 loss_ml 1401.435 loss_st 0.470 loss_source 0.015 [2019-09-09 09:14:08] Iter: [200/717] Freq 208.8 loss_total 6.474 loss_target 0.622 loss_ml 1407.155 loss_st 0.469 loss_source 0.018 [2019-09-09 09:16:55] Iter: [300/717] Freq 207.5 loss_total 6.521 loss_target 0.620 loss_ml 1402.158 loss_st 0.468 loss_source 0.019 [2019-09-09 09:19:44] Iter: [400/717] Freq 206.9 loss_total 6.569 loss_target 0.620 loss_ml 1402.685 loss_st 0.467 loss_source 0.020 [2019-09-09 09:22:32] Iter: [500/717] Freq 207.3 loss_total 6.633 loss_target 0.620 loss_ml 1400.709 loss_st 0.467 loss_source 0.022 [2019-09-09 09:25:18] Iter: [600/717] Freq 206.4 loss_total 6.668 loss_target 0.620 loss_ml 1397.827 loss_st 0.467 loss_source 0.022 [2019-09-09 09:28:09] Iter: [700/717] Freq 206.3 loss_total 6.704 loss_target 0.620 loss_ml 1398.610 loss_st 0.467 loss_source 0.023 [2019-09-09 09:30:57] Train loss_total 6.710 loss_target 0.619 loss_ml 1398.605 loss_st 0.467 loss_source 0.023

==>>[2019-09-09 09:31:29] [Epoch=005/020] Stage 1, [Need: 04:41:21] Iter: [000/717] Freq 86.0 loss_total 9.697 loss_target 0.607 loss_ml 1273.806 loss_st 0.443 loss_source 0.089 [2019-09-09 09:31:33] Iter: [100/717] Freq 206.2 loss_total 6.487 loss_target 0.620 loss_ml 1378.825 loss_st 0.466 loss_source 0.019 [2019-09-09 09:34:18] Iter: [200/717] Freq 205.3 loss_total 6.514 loss_target 0.619 loss_ml 1377.789 loss_st 0.466 loss_source 0.020 [2019-09-09 09:37:08] Iter: [300/717] Freq 203.8 loss_total 6.531 loss_target 0.619 loss_ml 1372.794 loss_st 0.464 loss_source 0.021 [2019-09-09 09:40:00] Iter: [400/717] Freq 204.9 loss_total 6.539 loss_target 0.618 loss_ml 1374.280 loss_st 0.463 loss_source 0.021 [2019-09-09 09:42:46] Iter: [500/717] Freq 205.2 loss_total 6.560 loss_target 0.619 loss_ml 1375.132 loss_st 0.463 loss_source 0.021 [2019-09-09 09:45:34] Iter: [600/717] Freq 205.1 loss_total 6.592 loss_target 0.618 loss_ml 1377.416 loss_st 0.462 loss_source 0.022 [2019-09-09 09:48:22] Iter: [700/717] Freq 205.7 loss_total 6.619 loss_target 0.618 loss_ml 1374.335 loss_st 0.462 loss_source 0.023 [2019-09-09 09:51:08] Train loss_total 6.625 loss_target 0.618 loss_ml 1372.910 loss_st 0.462 loss_source 0.023

==>>[2019-09-09 09:51:39] [Epoch=006/020] Stage 1, [Need: 04:25:54] Iter: [000/717] Freq 85.3 loss_total 5.598 loss_target 0.610 loss_ml 1239.213 loss_st 0.456 loss_source 0.004 [2019-09-09 09:51:43] Iter: [100/717] Freq 205.8 loss_total 6.331 loss_target 0.618 loss_ml 1375.195 loss_st 0.458 loss_source 0.018 [2019-09-09 09:54:29] Iter: [200/717] Freq 208.4 loss_total 6.395 loss_target 0.618 loss_ml 1359.088 loss_st 0.456 loss_source 0.019 [2019-09-09 09:57:13] Iter: [300/717] Freq 207.2 loss_total 6.399 loss_target 0.618 loss_ml 1352.542 loss_st 0.455 loss_source 0.020 [2019-09-09 10:00:02] Iter: [400/717] Freq 207.3 loss_total 6.414 loss_target 0.619 loss_ml 1354.975 loss_st 0.455 loss_source 0.020 [2019-09-09 10:02:49] Iter: [500/717] Freq 206.9 loss_total 6.454 loss_target 0.618 loss_ml 1353.630 loss_st 0.455 loss_source 0.021 [2019-09-09 10:05:37] Iter: [600/717] Freq 206.0 loss_total 6.479 loss_target 0.618 loss_ml 1353.888 loss_st 0.456 loss_source 0.021 [2019-09-09 10:08:28] Iter: [700/717] Freq 206.7 loss_total 6.495 loss_target 0.617 loss_ml 1349.789 loss_st 0.456 loss_source 0.022 [2019-09-09 10:11:13] Train loss_total 6.508 loss_target 0.617 loss_ml 1349.842 loss_st 0.456 loss_source 0.022

==>>[2019-09-09 10:11:48] [Epoch=007/020] Stage 1, [Need: 04:09:02] Iter: [000/717] Freq 86.5 loss_total 5.765 loss_target 0.633 loss_ml 1209.728 loss_st 0.470 loss_source 0.004 [2019-09-09 10:11:52] Iter: [100/717] Freq 204.9 loss_total 6.280 loss_target 0.616 loss_ml 1337.178 loss_st 0.453 loss_source 0.018 [2019-09-09 10:14:38] Iter: [200/717] Freq 207.4 loss_total 6.375 loss_target 0.616 loss_ml 1338.346 loss_st 0.453 loss_source 0.020 [2019-09-09 10:17:23] Iter: [300/717] Freq 206.2 loss_total 6.379 loss_target 0.616 loss_ml 1340.454 loss_st 0.453 loss_source 0.020 [2019-09-09 10:20:13] Iter: [400/717] Freq 206.7 loss_total 6.357 loss_target 0.616 loss_ml 1339.220 loss_st 0.452 loss_source 0.020 [2019-09-09 10:22:59] Iter: [500/717] Freq 206.4 loss_total 6.346 loss_target 0.616 loss_ml 1340.832 loss_st 0.452 loss_source 0.019 [2019-09-09 10:25:48] Iter: [600/717] Freq 206.4 loss_total 6.365 loss_target 0.616 loss_ml 1341.040 loss_st 0.452 loss_source 0.020 [2019-09-09 10:28:35] Iter: [700/717] Freq 206.8 loss_total 6.393 loss_target 0.616 loss_ml 1338.094 loss_st 0.451 loss_source 0.021 [2019-09-09 10:31:20] Train loss_total 6.394 loss_target 0.616 loss_ml 1337.839 loss_st 0.451 loss_source 0.021

==>>[2019-09-09 10:31:52] [Epoch=008/020] Stage 1, [Need: 03:51:15] Iter: [000/717] Freq 87.5 loss_total 6.050 loss_target 0.611 loss_ml 1313.760 loss_st 0.453 loss_source 0.013 [2019-09-09 10:31:56] Iter: [100/717] Freq 204.9 loss_total 6.114 loss_target 0.610 loss_ml 1332.722 loss_st 0.446 loss_source 0.016 [2019-09-09 10:34:42] Iter: [200/717] Freq 205.8 loss_total 6.112 loss_target 0.614 loss_ml 1327.753 loss_st 0.446 loss_source 0.016 [2019-09-09 10:37:30] Iter: [300/717] Freq 204.9 loss_total 6.150 loss_target 0.615 loss_ml 1319.832 loss_st 0.446 loss_source 0.017 [2019-09-09 10:40:20] Iter: [400/717] Freq 205.3 loss_total 6.187 loss_target 0.615 loss_ml 1321.973 loss_st 0.446 loss_source 0.018 [2019-09-09 10:43:08] Iter: [500/717] Freq 205.9 loss_total 6.221 loss_target 0.615 loss_ml 1323.013 loss_st 0.446 loss_source 0.018 [2019-09-09 10:45:54] Iter: [600/717] Freq 205.4 loss_total 6.277 loss_target 0.614 loss_ml 1326.658 loss_st 0.446 loss_source 0.019 [2019-09-09 10:48:44] Iter: [700/717] Freq 205.9 loss_total 6.312 loss_target 0.614 loss_ml 1326.071 loss_st 0.446 loss_source 0.020 [2019-09-09 10:51:30] Train loss_total 6.316 loss_target 0.614 loss_ml 1328.874 loss_st 0.446 loss_source 0.020

==>>[2019-09-09 10:52:03] [Epoch=009/020] Stage 1, [Need: 03:33:06] Iter: [000/717] Freq 87.9 loss_total 5.304 loss_target 0.605 loss_ml 1478.565 loss_st 0.433 loss_source 0.002 [2019-09-09 10:52:07] Iter: [100/717] Freq 200.7 loss_total 6.127 loss_target 0.610 loss_ml 1338.082 loss_st 0.444 loss_source 0.017 [2019-09-09 10:54:57] Iter: [200/717] Freq 203.9 loss_total 6.084 loss_target 0.610 loss_ml 1322.129 loss_st 0.443 loss_source 0.016 [2019-09-09 10:57:44] Iter: [300/717] Freq 203.3 loss_total 6.103 loss_target 0.610 loss_ml 1321.324 loss_st 0.442 loss_source 0.017 [2019-09-09 11:00:35] Iter: [400/717] Freq 204.6 loss_total 6.133 loss_target 0.610 loss_ml 1324.399 loss_st 0.442 loss_source 0.017 [2019-09-09 11:03:21] Iter: [500/717] Freq 205.7 loss_total 6.174 loss_target 0.609 loss_ml 1319.936 loss_st 0.443 loss_source 0.018 [2019-09-09 11:06:06] Iter: [600/717] Freq 205.3 loss_total 6.227 loss_target 0.609 loss_ml 1323.732 loss_st 0.443 loss_source 0.019 [2019-09-09 11:08:56] Iter: [700/717] Freq 206.0 loss_total 6.243 loss_target 0.610 loss_ml 1321.482 loss_st 0.442 loss_source 0.019 [2019-09-09 11:11:40] Train loss_total 6.244 loss_target 0.610 loss_ml 1322.587 loss_st 0.442 loss_source 0.019

==>>[2019-09-09 11:12:13] [Epoch=010/020] Stage 1, [Need: 03:14:31] Iter: [000/717] Freq 86.8 loss_total 7.268 loss_target 0.602 loss_ml 1314.978 loss_st 0.430 loss_source 0.043 [2019-09-09 11:12:17] Iter: [100/717] Freq 203.5 loss_total 5.923 loss_target 0.611 loss_ml 1322.328 loss_st 0.436 loss_source 0.014 [2019-09-09 11:15:05] Iter: [200/717] Freq 206.3 loss_total 5.992 loss_target 0.610 loss_ml 1318.604 loss_st 0.436 loss_source 0.016 [2019-09-09 11:17:50] Iter: [300/717] Freq 206.1 loss_total 6.052 loss_target 0.610 loss_ml 1320.901 loss_st 0.438 loss_source 0.017 [2019-09-09 11:20:38] Iter: [400/717] Freq 206.2 loss_total 6.079 loss_target 0.609 loss_ml 1315.209 loss_st 0.438 loss_source 0.017 [2019-09-09 11:23:26] Iter: [500/717] Freq 206.7 loss_total 6.118 loss_target 0.608 loss_ml 1315.599 loss_st 0.438 loss_source 0.018 [2019-09-09 11:26:12] Iter: [600/717] Freq 206.5 loss_total 6.142 loss_target 0.608 loss_ml 1317.638 loss_st 0.438 loss_source 0.018 [2019-09-09 11:29:00] Iter: [700/717] Freq 206.8 loss_total 6.182 loss_target 0.608 loss_ml 1311.943 loss_st 0.438 loss_source 0.019 [2019-09-09 11:31:45] Train loss_total 6.186 loss_target 0.608 loss_ml 1312.057 loss_st 0.438 loss_source 0.019

==>>[2019-09-09 11:32:16] [Epoch=011/020] Stage 1, [Need: 02:55:33] Iter: [000/717] Freq 87.7 loss_total 5.648 loss_target 0.589 loss_ml 1467.208 loss_st 0.452 loss_source 0.005 [2019-09-09 11:32:20] Iter: [100/717] Freq 205.7 loss_total 6.002 loss_target 0.606 loss_ml 1288.686 loss_st 0.438 loss_source 0.016 [2019-09-09 11:35:06] Iter: [200/717] Freq 204.7 loss_total 5.970 loss_target 0.606 loss_ml 1309.730 loss_st 0.437 loss_source 0.015 [2019-09-09 11:37:56] Iter: [300/717] Freq 204.3 loss_total 5.938 loss_target 0.607 loss_ml 1299.596 loss_st 0.436 loss_source 0.015 [2019-09-09 11:40:46] Iter: [400/717] Freq 204.7 loss_total 5.929 loss_target 0.607 loss_ml 1308.356 loss_st 0.436 loss_source 0.015 [2019-09-09 11:43:34] Iter: [500/717] Freq 204.3 loss_total 5.912 loss_target 0.606 loss_ml 1308.441 loss_st 0.435 loss_source 0.014 [2019-09-09 11:46:25] Iter: [600/717] Freq 204.1 loss_total 5.908 loss_target 0.607 loss_ml 1305.215 loss_st 0.435 loss_source 0.014 [2019-09-09 11:49:15] Iter: [700/717] Freq 204.9 loss_total 5.906 loss_target 0.606 loss_ml 1306.282 loss_st 0.434 loss_source 0.014 [2019-09-09 11:52:00] Train loss_total 5.903 loss_target 0.606 loss_ml 1306.698 loss_st 0.434 loss_source 0.014

==>>[2019-09-09 11:52:30] [Epoch=012/020] Stage 1, [Need: 02:36:32] Iter: [000/717] Freq 84.4 loss_total 5.492 loss_target 0.605 loss_ml 1231.723 loss_st 0.429 loss_source 0.008 [2019-09-09 11:52:34] Iter: [100/717] Freq 205.2 loss_total 5.692 loss_target 0.606 loss_ml 1295.873 loss_st 0.429 loss_source 0.011 [2019-09-09 11:55:20] Iter: [200/717] Freq 207.6 loss_total 5.755 loss_target 0.605 loss_ml 1301.396 loss_st 0.430 loss_source 0.012 [2019-09-09 11:58:05] Iter: [300/717] Freq 208.3 loss_total 5.741 loss_target 0.605 loss_ml 1294.338 loss_st 0.430 loss_source 0.012 [2019-09-09 12:00:50] Iter: [400/717] Freq 209.8 loss_total 5.769 loss_target 0.605 loss_ml 1292.147 loss_st 0.431 loss_source 0.013 [2019-09-09 12:03:31] Iter: [500/717] Freq 210.3 loss_total 5.767 loss_target 0.605 loss_ml 1296.726 loss_st 0.430 loss_source 0.013 [2019-09-09 12:06:15] Iter: [600/717] Freq 210.2 loss_total 5.772 loss_target 0.605 loss_ml 1294.503 loss_st 0.430 loss_source 0.013 [2019-09-09 12:08:59] Iter: [700/717] Freq 210.5 loss_total 5.776 loss_target 0.605 loss_ml 1292.155 loss_st 0.430 loss_source 0.013 [2019-09-09 12:11:42] Train loss_total 5.780 loss_target 0.605 loss_ml 1292.226 loss_st 0.431 loss_source 0.013

==>>[2019-09-09 12:12:13] [Epoch=013/020] Stage 1, [Need: 02:17:03] Iter: [000/717] Freq 83.7 loss_total 6.647 loss_target 0.620 loss_ml 1138.104 loss_st 0.444 loss_source 0.028 [2019-09-09 12:12:17] Iter: [100/717] Freq 208.3 loss_total 5.696 loss_target 0.603 loss_ml 1284.658 loss_st 0.429 loss_source 0.011 [2019-09-09 12:15:01] Iter: [200/717] Freq 208.6 loss_total 5.698 loss_target 0.603 loss_ml 1292.698 loss_st 0.429 loss_source 0.011 [2019-09-09 12:17:46] Iter: [300/717] Freq 207.2 loss_total 5.717 loss_target 0.603 loss_ml 1289.368 loss_st 0.429 loss_source 0.012 [2019-09-09 12:20:35] Iter: [400/717] Freq 206.3 loss_total 5.712 loss_target 0.604 loss_ml 1287.088 loss_st 0.429 loss_source 0.012 [2019-09-09 12:23:25] Iter: [500/717] Freq 205.9 loss_total 5.723 loss_target 0.604 loss_ml 1290.669 loss_st 0.429 loss_source 0.012 [2019-09-09 12:26:15] Iter: [600/717] Freq 205.2 loss_total 5.714 loss_target 0.605 loss_ml 1290.317 loss_st 0.429 loss_source 0.012 [2019-09-09 12:29:06] Iter: [700/717] Freq 205.4 loss_total 5.727 loss_target 0.605 loss_ml 1296.197 loss_st 0.429 loss_source 0.012 [2019-09-09 12:31:54] Train loss_total 5.734 loss_target 0.605 loss_ml 1296.825 loss_st 0.429 loss_source 0.012

==>>[2019-09-09 12:32:27] [Epoch=014/020] Stage 1, [Need: 01:57:45] Iter: [000/717] Freq 83.3 loss_total 5.398 loss_target 0.611 loss_ml 1078.433 loss_st 0.424 loss_source 0.007 [2019-09-09 12:32:31] Iter: [100/717] Freq 202.6 loss_total 5.653 loss_target 0.606 loss_ml 1300.023 loss_st 0.428 loss_source 0.011 [2019-09-09 12:35:19] Iter: [200/717] Freq 202.1 loss_total 5.700 loss_target 0.604 loss_ml 1295.994 loss_st 0.429 loss_source 0.011 [2019-09-09 12:38:11] Iter: [300/717] Freq 201.5 loss_total 5.724 loss_target 0.603 loss_ml 1297.736 loss_st 0.428 loss_source 0.012 [2019-09-09 12:41:04] Iter: [400/717] Freq 202.5 loss_total 5.700 loss_target 0.604 loss_ml 1297.518 loss_st 0.428 loss_source 0.012 [2019-09-09 12:43:52] Iter: [500/717] Freq 203.3 loss_total 5.712 loss_target 0.604 loss_ml 1300.217 loss_st 0.428 loss_source 0.012 [2019-09-09 12:46:40] Iter: [600/717] Freq 203.1 loss_total 5.713 loss_target 0.603 loss_ml 1296.051 loss_st 0.428 loss_source 0.012 [2019-09-09 12:49:31] Iter: [700/717] Freq 203.9 loss_total 5.714 loss_target 0.603 loss_ml 1295.927 loss_st 0.428 loss_source 0.012 [2019-09-09 12:52:17] Train loss_total 5.712 loss_target 0.603 loss_ml 1295.506 loss_st 0.428 loss_source 0.012

==>>[2019-09-09 12:52:48] [Epoch=015/020] Stage 1, [Need: 01:38:22] Iter: [000/717] Freq 85.8 loss_total 5.220 loss_target 0.641 loss_ml 1132.827 loss_st 0.410 loss_source 0.005 [2019-09-09 12:52:52] Iter: [100/717] Freq 201.3 loss_total 5.577 loss_target 0.607 loss_ml 1299.206 loss_st 0.427 loss_source 0.009 [2019-09-09 12:55:42] Iter: [200/717] Freq 203.5 loss_total 5.603 loss_target 0.605 loss_ml 1292.792 loss_st 0.428 loss_source 0.010 [2019-09-09 12:58:30] Iter: [300/717] Freq 203.4 loss_total 5.637 loss_target 0.605 loss_ml 1289.240 loss_st 0.428 loss_source 0.010 [2019-09-09 13:01:20] Iter: [400/717] Freq 203.4 loss_total 5.638 loss_target 0.604 loss_ml 1287.391 loss_st 0.428 loss_source 0.010 [2019-09-09 13:04:10] Iter: [500/717] Freq 203.9 loss_total 5.652 loss_target 0.604 loss_ml 1286.517 loss_st 0.428 loss_source 0.011 [2019-09-09 13:06:58] Iter: [600/717] Freq 203.4 loss_total 5.652 loss_target 0.604 loss_ml 1287.383 loss_st 0.428 loss_source 0.011 [2019-09-09 13:09:51] Iter: [700/717] Freq 203.0 loss_total 5.665 loss_target 0.604 loss_ml 1287.661 loss_st 0.428 loss_source 0.011 [2019-09-09 13:12:43] Train loss_total 5.670 loss_target 0.604 loss_ml 1286.469 loss_st 0.428 loss_source 0.011

==>>[2019-09-09 13:13:14] [Epoch=016/020] Stage 1, [Need: 01:18:53] Iter: [000/717] Freq 88.3 loss_total 6.062 loss_target 0.579 loss_ml 1237.043 loss_st 0.419 loss_source 0.021 [2019-09-09 13:13:18] Iter: [100/717] Freq 200.2 loss_total 5.664 loss_target 0.603 loss_ml 1277.227 loss_st 0.428 loss_source 0.011 [2019-09-09 13:16:08] Iter: [200/717] Freq 204.7 loss_total 5.615 loss_target 0.603 loss_ml 1285.817 loss_st 0.427 loss_source 0.010 [2019-09-09 13:18:53] Iter: [300/717] Freq 203.7 loss_total 5.634 loss_target 0.603 loss_ml 1290.623 loss_st 0.427 loss_source 0.010 [2019-09-09 13:21:45] Iter: [400/717] Freq 205.5 loss_total 5.656 loss_target 0.602 loss_ml 1290.798 loss_st 0.427 loss_source 0.011 [2019-09-09 13:24:29] Iter: [500/717] Freq 206.3 loss_total 5.645 loss_target 0.602 loss_ml 1289.957 loss_st 0.428 loss_source 0.011 [2019-09-09 13:27:14] Iter: [600/717] Freq 205.6 loss_total 5.640 loss_target 0.602 loss_ml 1289.632 loss_st 0.427 loss_source 0.011 [2019-09-09 13:30:05] Iter: [700/717] Freq 206.0 loss_total 5.633 loss_target 0.602 loss_ml 1289.877 loss_st 0.427 loss_source 0.011 [2019-09-09 13:32:51] Train loss_total 5.633 loss_target 0.602 loss_ml 1289.996 loss_st 0.427 loss_source 0.011

==>>[2019-09-09 13:33:23] [Epoch=017/020] Stage 1, [Need: 00:59:14] Iter: [000/717] Freq 85.5 loss_total 5.415 loss_target 0.599 loss_ml 1179.304 loss_st 0.446 loss_source 0.003 [2019-09-09 13:33:27] Iter: [100/717] Freq 206.8 loss_total 5.613 loss_target 0.601 loss_ml 1306.293 loss_st 0.426 loss_source 0.010 [2019-09-09 13:36:12] Iter: [200/717] Freq 210.1 loss_total 5.653 loss_target 0.603 loss_ml 1308.942 loss_st 0.427 loss_source 0.011 [2019-09-09 13:38:54] Iter: [300/717] Freq 208.5 loss_total 5.650 loss_target 0.604 loss_ml 1300.655 loss_st 0.427 loss_source 0.011 [2019-09-09 13:41:42] Iter: [400/717] Freq 209.1 loss_total 5.641 loss_target 0.603 loss_ml 1293.910 loss_st 0.427 loss_source 0.011 [2019-09-09 13:44:26] Iter: [500/717] Freq 208.8 loss_total 5.626 loss_target 0.603 loss_ml 1290.932 loss_st 0.426 loss_source 0.011 [2019-09-09 13:47:13] Iter: [600/717] Freq 207.5 loss_total 5.616 loss_target 0.603 loss_ml 1294.961 loss_st 0.426 loss_source 0.010 [2019-09-09 13:50:05] Iter: [700/717] Freq 207.4 loss_total 5.613 loss_target 0.603 loss_ml 1293.459 loss_st 0.426 loss_source 0.010 [2019-09-09 13:52:52] Train loss_total 5.615 loss_target 0.603 loss_ml 1294.656 loss_st 0.426 loss_source 0.010

==>>[2019-09-09 13:53:24] [Epoch=018/020] Stage 1, [Need: 00:39:31] Iter: [000/717] Freq 84.7 loss_total 5.297 loss_target 0.629 loss_ml 1322.734 loss_st 0.431 loss_source 0.002 [2019-09-09 13:53:28] Iter: [100/717] Freq 204.3 loss_total 5.650 loss_target 0.601 loss_ml 1266.729 loss_st 0.427 loss_source 0.011 [2019-09-09 13:56:15] Iter: [200/717] Freq 207.6 loss_total 5.618 loss_target 0.602 loss_ml 1279.130 loss_st 0.426 loss_source 0.011 [2019-09-09 13:58:59] Iter: [300/717] Freq 205.8 loss_total 5.637 loss_target 0.602 loss_ml 1287.542 loss_st 0.426 loss_source 0.011 [2019-09-09 14:01:50] Iter: [400/717] Freq 206.0 loss_total 5.623 loss_target 0.602 loss_ml 1286.949 loss_st 0.426 loss_source 0.011 [2019-09-09 14:04:38] Iter: [500/717] Freq 205.9 loss_total 5.639 loss_target 0.603 loss_ml 1290.681 loss_st 0.426 loss_source 0.011 [2019-09-09 14:07:26] Iter: [600/717] Freq 204.4 loss_total 5.636 loss_target 0.603 loss_ml 1287.162 loss_st 0.426 loss_source 0.011 [2019-09-09 14:10:21] Iter: [700/717] Freq 204.4 loss_total 5.629 loss_target 0.603 loss_ml 1287.624 loss_st 0.426 loss_source 0.011 [2019-09-09 14:13:11] Train loss_total 5.627 loss_target 0.603 loss_ml 1286.365 loss_st 0.426 loss_source 0.011

==>>[2019-09-09 14:13:42] [Epoch=019/020] Stage 1, [Need: 00:19:47] Iter: [000/717] Freq 85.6 loss_total 5.260 loss_target 0.592 loss_ml 1382.883 loss_st 0.410 loss_source 0.006 [2019-09-09 14:13:46] Iter: [100/717] Freq 201.0 loss_total 5.562 loss_target 0.602 loss_ml 1283.190 loss_st 0.424 loss_source 0.010 [2019-09-09 14:16:36] Iter: [200/717] Freq 205.1 loss_total 5.558 loss_target 0.603 loss_ml 1291.237 loss_st 0.424 loss_source 0.010 [2019-09-09 14:19:21] Iter: [300/717] Freq 205.5 loss_total 5.574 loss_target 0.602 loss_ml 1288.985 loss_st 0.425 loss_source 0.010 [2019-09-09 14:22:09] Iter: [400/717] Freq 206.0 loss_total 5.590 loss_target 0.602 loss_ml 1288.623 loss_st 0.426 loss_source 0.010 [2019-09-09 14:24:55] Iter: [500/717] Freq 206.3 loss_total 5.603 loss_target 0.602 loss_ml 1287.619 loss_st 0.426 loss_source 0.010 [2019-09-09 14:27:42] Iter: [600/717] Freq 206.3 loss_total 5.604 loss_target 0.602 loss_ml 1289.316 loss_st 0.426 loss_source 0.010 [2019-09-09 14:30:30] Iter: [700/717] Freq 206.6 loss_total 5.606 loss_target 0.602 loss_ml 1288.093 loss_st 0.426 loss_source 0.010 [2019-09-09 14:33:16] Train loss_total 5.607 loss_target 0.602 loss_ml 1289.568 loss_st 0.426 loss_source 0.010
Test r1 66.835 MAP 40.664 r5 82.868 r10 88.183

This is my result. Thanks.