1.尝试能否通过一些评价指标在（10-50的范围内）去确定k，先统一降维幅度为原奇异值平方和的90%
    关于随机性的问题：在使用更细粒度的模型的时候效果不如使用较粗粒度的模型，但其聚类效果极其稳定，相反，较粗粒度的模型效果较好，但聚类效果略微不稳定。
    在vgg16、resnet20、fashion、svhn模型上，最终选出的k总是极为接近范围的下限。

    clusterSize lenet1 clusterSize lenet4 clusterSize lenet5 clusterSize vgg16 cluster resnet20 clusterSize fashion clusterSize svhn
    34 0.596347368 43 0.275915789 28 0.295094737 10 1.079747368 10 0.489557895 10 2.581673684 10 4.008105263
    39 0.831084211 48 0.723252632 33 0.281747368 10 2.154094737 11 0.504442105 10 2.347673684 10 3.558242105
    42 0.491568421 45 0.443326316 32 0.673052632 10 2.037242105 12 0.831168421 11 1.463705263 12 3.748715789
    47 0.851926316 41 0.303084211 32 0.227147368 11 1.712905263 11 1.926484211 10 1.821378947 11 3.416663158
    43 1.316884211 41 0.392652632 36 0.944810526 10 0.887610526 13 0.4406 10 0.888547368 10 5.249831579
    38 0.402515789 47 0.634336842 32 0.688115789 10 0.584115789 12 0.290042105 11 1.231684211 11 3.569031579
    42 0.680494737 42 0.282715789 32 0.324378947 10 0.9112 10 0.304231579 13 1.271210526 12 3.092273684
    47 0.6828 46 0.617042105 35 0.408389474 10 1.815663158 13 1.153189474 12 1.389368421 12 5.641
    42 1.016084211 47 0.313231579 30 0.428642105 12 2.368536842 12 0.994715789 11 2.934842105 10 2.897452632
    46 0.469873684 49 0.300694737 31 0.239326316 11 0.9234 12 1.672273684 12 2.4758 10 2.400252632
    40 0.627789474 47 0.843536842 37 0.532894737 11 1.596863158 15 0.293978947 10 2.485421053 11 3.277389474
    47 1.617926316 48 0.487989474 27 0.5108 10 2.538915789 11 0.654347368 13 0.908810526 10 1.664115789
    35 1.289705263 49 0.391747368 34 0.166926316 11 1.698147368 11 1.401389474 11 1.629042105 10 3.664726316
    41 0.923168421 46 0.269915789 39 0.348136842 10 0.993968421 11 0.777063158 11 2.053557895 10 2.682052632
    41 0.886368421 47 0.423884211 41 0.337442105 10 1.084010526 12 0.876336842 11 0.998778947 11 3.941294737
    47 0.954452632 42 0.724431579 29 0.516789474 10 1.861526316 12 0.528852632 24 0.882305263 11 4.216494737
    41 1.023768421 49 0.499294737 30 0.368052632 11 1.102252632 11 0.688736842 12 2.168231579 43 3.403947368
    35 0.454242105 49 1.179252632 37 0.610189474 10 2.054031579 11 1.813768421 11 2.169263158 14 5.164157895
    46 1.062431579 46 0.556536842 32 0.227494737 10 0.581726316 17 0.677031579 10 0.685347368 11 3.157547368
    41 1.592589474 45 0.501410526 34 0.463252632 10 1.537157895 13 0.741431579 11 1.972936842 13 1.938484211

    2.为了进一步增强对pace的感知，需要在pace上获得大区间的数据。
    服务器命令：
    nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=lenet1 —select_layer_idx=-3 —dec_dim=8 —min_samples=4 —min_cluster_size=80 > lenet1.log 2>&1 &
    nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=lenet4 —select_layer_idx=-3 —min_samples=4 —min_cluster_size=80 > lenet4.log 2>&1 &
    🌟nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=lenet5 —select_layer_idx=-3 —min_samples=4 —min_cluster_size=80 > lenet5.log 2>&1 &
    nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=vgg16 —select_layer_idx=-3 —dec_dim=2 —min_samples=4 —min_cluster_size=80 > vgg16.log 2>&1 &
    nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=cifar10 —select_layer_idx=-2 —dec_dim=2 —min_samples=4 —min_cluster_size=80 > cifar10.log 2>&1 &
    nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=svhn —select_layer_idx=-3 —dec_dim=2 —min_samples=4 —min_cluster_size=80 > svhn.log 2>&1 &
    🌟nohup python -u -m mnist_cifar_imagenet_svhn.selection —exp_id=fashion —select_layer_idx=-3 —min_samples=4 —min_cluster_size=80 > fashion.log 2>&1 &
    代码的问题

    • image.pngmaxnum没有更新
    • 异常点过多，正常点簇过少时代码报错 |
      | 50 | 60 | 70 | 80 | 90 | 100 | 110 | 120 | 130 | 140 | 150 | 160 | 170 | 180 | 190 | 200 | 210 | 220 | 230 | 240 | 250 | 260 | 270 | 280 | 290 | 300 | 310 | 320 | 330 | 340 | 350 | 360 | 370 | 380 | 390 | 400 | 410 | 420 | 430 | 440 | 450 | 460 | 470 | 480 | 490 | 500 | 510 | 520 | 530 | 540 | 550 | 560 | 570 | 580 | 590 | 600 | 610 | 620 | 630 | 640 | 650 | 660 | 670 | 680 | 690 | 700 | 710 | 720 | 730 | 740 | 750 | 760 | 770 | 780 | 790 | 800 | 810 | 820 | 830 | 840 | 850 | 860 | 870 | 880 | 890 | 900 | 910 | 920 | 930 | 940 | 950 | 960 | 970 | 980 | 990 | mean | | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | —- | | mnist_lenet1 | 2.28 | 3.279 | 1.709 | 2.531 | 2.502 | 2.97 | 3.007 | 3.333 | 3.357 | 3.547 | 3.569 | 3.75 | 3.81 | 3.889 | 3.473 | 3.5 | 2.625 | 2.715 | 2.889 | 2.917 | 2.68 | 2.316 | 2.162 | 2.151 | 2.349 | 2.325 | 2.513 | 2.195 | 2.377 | 2.353 | 2.279 | 2.247 | 2.432 | 2.406 | 2.564 | 2.26 | 1.951 | 1.471 | 1.628 | 1.664 | 1.778 | 1.783 | 1.705 | 1.739 | 1.422 | 1.48 | 1.572 | 1.643 | 1.698 | 1.777 | 1.634 | 1.693 | 1.752 | 1.645 | 1.526 | 1.613 | 1.637 | 1.732 | 1.746 | 1.673 | 1.692 | 1.795 | 1.642 | 1.751 | 1.449 | 1.423 | 1.27 | 1.389 | 1.388 | 1.484 | 1.347 | 1.448 | 1.331 | 1.41 | 1.408 | 1.499 | 1.268 | 1.22 | 1.244 | 1.309 | 1.328 | 1.395 | 1.44 | 1.478 | 1.515 | 1.446 | 1.489 | 1.303 | 1.346 | 1.276 | 1.333 | 1.355 | 1.414 | 1.224 | 1.281 | 1.9854 | |
      | 0.076 | 1.667 | 0 | 1.112 | 1.111 | 1.72 | 0.909 | 0.644 | 0 | 0 | 0.213 | 0 | 0.339 | 0.556 | 0.332 | 0.5 | 0.232 | 0.455 | 0.715 | 0 | 0 | 0 | 0.28 | 0 | 0.28 | 0.333 | 0.603 | 0.625 | 0.906 | 0.885 | 0.854 | 0.835 | 1.081 | 1.055 | 1.279 | 1.03 | 0.976 | 0.994 | 1.16 | 0.941 | 0.668 | 0.736 | 0.851 | 0.905 | 0.818 | 0.898 | 0.982 | 1.05 | 1.132 | 1.206 | 1.273 | 1.351 | 1.399 | 1.502 | 1.018 | 1.098 | 1.148 | 1.233 | 1.269 | 1.218 | 1.229 | 1.327 | 1.196 | 1.309 | 1.016 | 1.137 | 1.138 | 1.25 | 1.252 | 1.349 | 1.213 | 1.185 | 1.047 | 1.154 | 1.155 | 1.249 | 1.144 | 1.098 | 1.123 | 1.19 | 1.222 | 1.279 | 1.325 | 1.364 | 1.403 | 1.331 | 1.389 | 1.307 | 1.356 | 1.276 | 1.333 | 1.355 | 1.423 | 1.326 | 1.382 | 0.940833677 | | mnist_lenet4 | 0.42 | 1.335 | 1.45 | 0.17 | 0.865 | 1 | 0 | 0.587 | 0.769 | 0 | 0 | 0 | 0 | 0.275 | 0.523 | 0.5 | 0.679 | 0.909 | 0.877 | 0.587 | 0.797 | 0.772 | 0.885 | 1.043 | 1.031 | 1.069 | 0.936 | 0.934 | 0.645 | 0.848 | 0.575 | 0.57 | 0.204 | 0 | 0.01 | 0.17 | 0.244 | 0 | 0.141 | 0.227 | 0.222 | 0.333 | 0.425 | 0.416 | 0.513 | 0.4 | 0.196 | 0.266 | 0.189 | 0.061 | 0 | 0.179 | 0.059 | 0.172 | 0.338 | 0.414 | 0.328 | 0.322 | 0.406 | 0.469 | 0.615 | 0.712 | 0.597 | 0.587 | 0.536 | 0.437 | 0.423 | 0.385 | 0.283 | 0.271 | 0.238 | 0.291 | 0.259 | 0.255 | 0.306 | 0.25 | 0.247 | 0.166 | 0.12 | 0.119 | 0.043 | 0 | 0 | 0.057 | 0 | 0 | 0 | 0.007 | 0.069 | 0 | 0.001 | 0 | 0 | 0 | 0.057 | 0.369326 | |
      | 0.42 | 1.335 | 1.45 | 0.17 | 0.865 | 1 | 0 | 0.587 | 0.769 | 0 | 0 | 0 | 0 | 0.275 | 0.523 | 0.5 | 0.679 | 0.909 | 0.877 | 0.587 | 0.797 | 0.772 | 0.885 | 1.043 | 1.031 | 1.069 | 0.936 | 0.934 | 0.645 | 0.848 | 0.575 | 0.57 | 0.204 | 0 | 0.01 | 0.17 | 0.244 | 0 | 0.141 | 0.227 | 0.222 | 0.333 | 0.425 | 0.416 | 0.513 | 0.4 | 0.196 | 0.266 | 0.189 | 0.061 | 0 | 0.179 | 0.059 | 0.172 | 0.338 | 0.414 | 0.328 | 0.322 | 0.406 | 0.469 | 0.615 | 0.712 | 0.597 | 0.587 | 0.536 | 0.437 | 0.423 | 0.385 | 0.283 | 0.271 | 0.238 | 0.291 | 0.259 | 0.255 | 0.306 | 0.25 | 0.247 | 0.166 | 0.12 | 0.119 | 0.043 | 0 | 0 | 0.057 | 0 | 0 | 0 | 0.007 | 0.069 | 0 | 0.001 | 0 | 0 | 0 | 0.057 | 0.369326 | | mnist_lenet5 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.06 | 0.27 | 0.417 | 0 | 0 | 0 | 0 | 0 | 0.047 | 0.179 | 0.298 | 0.395 | 0.417 | 0.398 | 0.386 | 0.386 | 0.369 | 0.051 | 0.154 | 0 | 0 | 0 | 0 | 0 | 0 | 0.06 | 0.121 | 0.191 | 0.252 | 0.25 | 0.244 | 0.238 | 0.233 | 0.282 | 0.338 | 0.39 | 0.425 | 0.415 | 0.204 | 0.2 | 0.212 | 0.055 | 0.103 | 0.148 | 0 | 0 | 0 | 0 | 0 | 0 | 0.164 | 0.162 | 0.297 | 0.257 | 0.209 | 0.163 | 0.3 | 0.294 | 0.289 | 0.285 | 0.253 | 0.357 | 0.321 | 0.278 | 0.267 | 0.263 | 0.26 | 0.256 | 0.225 | 0.183 | 0.156 | 0.126 | 0.121 | 0.119 | 0.117 | 0.112 | 0.083 | 0.054 | 0.022 | 0 | 0 | 0 | 0 | 0 | 0.071 | 0.047 | 0.017 | 0.102 | 0.147313 | |
      | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.06 | 0.27 | 0.417 | 0 | 0 | 0 | 0 | 0 | 0.047 | 0.179 | 0.298 | 0.395 | 0.417 | 0.398 | 0.386 | 0.386 | 0.369 | 0.051 | 0.154 | 0 | 0 | 0 | 0 | 0 | 0 | 0.06 | 0.121 | 0.191 | 0.252 | 0.25 | 0.244 | 0.238 | 0.233 | 0.282 | 0.338 | 0.39 | 0.425 | 0.415 | 0.204 | 0.2 | 0.212 | 0.055 | 0.103 | 0.148 | 0 | 0 | 0 | 0 | 0 | 0 | 0.164 | 0.162 | 0.297 | 0.257 | 0.209 | 0.163 | 0.3 | 0.294 | 0.289 | 0.285 | 0.253 | 0.357 | 0.321 | 0.278 | 0.267 | 0.263 | 0.26 | 0.256 | 0.225 | 0.183 | 0.156 | 0.126 | 0.121 | 0.119 | 0.117 | 0.112 | 0.083 | 0.054 | 0.022 | 0 | 0 | 0 | 0 | 0 | 0.071 | 0.047 | 0.017 | 0.102 | 0.147313 | | cifar10_vgg16 | 0 | 0.59 | 0 | 0 | 1.486 | 2 | 4.546 | 5.653 | 4.051 | 2.857 | 3.333 | 3.57 | 3.222 | 2.793 | 2.045 | 3.178 | 3.318 | 2.728 | 2.174 | 2.32 | 2.02 | 2.703 | 2.592 | 2.32 | 2.413 | 2.333 | 2.24 | 1.695 | 1.515 | 1.466 | 1.601 | 1.209 | 0.811 | 1.052 | 1.487 | 1.135 | 0.736 | 0.952 | 0.634 | 0.454 | 0.222 | 0.652 | 0.352 | 0.028 | 0.129 | 0.4 | 0.31 | 0.205 | 0.566 | 0.928 | 0.867 | 0.714 | 0.701 | 0.466 | 0.582 | 0.666 | 0.656 | 0.627 | 0.376 | 0.313 | 0.154 | 0.123 | 0.342 | 0.294 | 0.434 | 0.391 | 0.313 | 0.278 | 0.274 | 0.361 | 0.286 | 0.263 | 0.52 | 0.622 | 0.643 | 0.75 | 0.837 | 0.796 | 0.723 | 0.714 | 0.702 | 0.75 | 0.689 | 0.681 | 0.55 | 0.487 | 0.66 | 0.544 | 0.519 | 0.537 | 0.526 | 0.417 | 0.593 | 0.432 | 0.303 | 1.142105 | |
      | 0 | 0.59 | 0 | 0 | 1.486 | 2 | 4.546 | 5.653 | 4.051 | 2.857 | 3.333 | 3.57 | 3.222 | 2.793 | 2.045 | 3.178 | 3.318 | 2.728 | 2.174 | 2.32 | 2.02 | 2.703 | 2.592 | 2.32 | 2.413 | 2.333 | 2.24 | 1.695 | 1.515 | 1.466 | 1.601 | 1.209 | 0.811 | 1.052 | 1.487 | 1.135 | 0.736 | 0.952 | 0.634 | 0.454 | 0.222 | 0.652 | 0.352 | 0.028 | 0.129 | 0.4 | 0.31 | 0.205 | 0.566 | 0.928 | 0.867 | 0.714 | 0.701 | 0.466 | 0.582 | 0.666 | 0.656 | 0.627 | 0.376 | 0.313 | 0.154 | 0.123 | 0.342 | 0.294 | 0.434 | 0.391 | 0.313 | 0.278 | 0.274 | 0.361 | 0.286 | 0.263 | 0.52 | 0.622 | 0.643 | 0.75 | 0.837 | 0.796 | 0.723 | 0.714 | 0.702 | 0.75 | 0.689 | 0.681 | 0.55 | 0.487 | 0.66 | 0.544 | 0.519 | 0.537 | 0.526 | 0.417 | 0.593 | 0.432 | 0.303 | 1.142105 | | cifar10_resnet20 | 4.008 | 7.03 | 6.725 | 6.19 | 6.667 | 8 | 7.273 | 6.666 | 7.056 | 6.011 | 6.44 | 6.815 | 8.235 | 9.444 | 10 | 9.5 | 9.106 | 8.349 | 8.093 | 7.857 | 7.6 | 6.923 | 6.296 | 6.071 | 5.544 | 5.44 | 5.666 | 5.565 | 5.757 | 5.882 | 5.429 | 5.278 | 5.143 | 4.808 | 5.261 | 5.94 | 5.854 | 5.476 | 5.582 | 5.681 | 5.778 | 5.701 | 5.844 | 5.982 | 6.318 | 6.2 | 6.862 | 6.923 | 6.792 | 6.514 | 6.622 | 6.369 | 6.124 | 6.035 | 6.101 | 5.834 | 6.066 | 6.15 | 6.249 | 6.659 | 7.209 | 7.121 | 7.164 | 6.911 | 7.391 | 7.44 | 7.51 | 7.579 | 7.371 | 7.297 | 7.333 | 7.368 | 7.273 | 7.44 | 7.503 | 7.565 | 7.378 | 7.196 | 7.109 | 7.023 | 6.941 | 6.861 | 6.808 | 6.758 | 6.822 | 6.662 | 6.594 | 6.631 | 6.667 | 6.489 | 6.545 | 6.398 | 6.255 | 6.216 | 6.06 | 6.61122 | |
      | 9.891 | 10.309 | 11.011 | 8.69 | 6.667 | 8 | 7.273 | 5.833 | 5.517 | 6.725 | 6.44 | 6.19 | 7.058 | 6.667 | 6.316 | 5.5 | 5.297 | 5.167 | 5.919 | 5.773 | 5.2 | 5 | 5.185 | 5.357 | 4.854 | 5.107 | 5.021 | 4.94 | 4.545 | 4.117 | 3.714 | 3.611 | 4.062 | 3.756 | 4.235 | 4.19 | 4.391 | 4.285 | 4.186 | 4.318 | 4 | 4.397 | 4.567 | 4.523 | 4.481 | 4.4 | 4.509 | 4.808 | 4.906 | 5.033 | 4.985 | 4.94 | 5.247 | 5.173 | 5.084 | 5 | 4.755 | 4.86 | 4.821 | 4.94 | 4.901 | 5 | 4.925 | 4.852 | 4.927 | 5.011 | 5.116 | 4.94 | 4.906 | 4.865 | 4.8 | 4.737 | 4.675 | 4.748 | 4.718 | 4.69 | 4.662 | 4.757 | 4.699 | 4.642 | 4.588 | 5 | 5.314 | 5.508 | 5.923 | 5.885 | 5.934 | 6.087 | 6.129 | 5.957 | 6.019 | 5.877 | 5.739 | 5.808 | 5.656 | 5.289023309 | | svhn_lenet5 | 0 | 3.03 | 3.792 | 1.673 | 0.002 | 0.961 | 2.632 | 1.626 | 1.356 | 1.666 | 2.765 | 3.087 | 2.907 | 3.315 | 3.396 | 3.652 | 3.317 | 4.072 | 3.478 | 2.875 | 2.992 | 2.432 | 2.231 | 2.158 | 1.737 | 1.544 | 1.766 | 1.25 | 0.909 | 0.865 | 0.935 | 0.582 | 0.809 | 0.726 | 0.663 | 0.542 | 0.73 | 1.19 | 0.85 | 1.018 | 0.887 | 0.868 | 0.639 | 0.583 | 0.893 | 1 | 0.98 | 1.346 | 1.263 | 1.347 | 1.475 | 1.252 | 1.361 | 1.309 | 1.554 | 1.502 | 1.478 | 1.556 | 1.664 | 1.729 | 1.538 | 1.366 | 1.495 | 1.417 | 1.626 | 1.569 | 1.67 | 1.727 | 1.776 | 1.62 | 1.598 | 1.447 | 1.376 | 1.176 | 1.394 | 1.5 | 1.593 | 1.645 | 1.816 | 1.669 | 1.766 | 1.692 | 1.769 | 1.704 | 1.798 | 1.889 | 1.946 | 1.798 | 1.841 | 1.807 | 1.764 | 1.805 | 1.854 | 1.734 | 1.614 | 1.663326316 | |
      | 0 | 3.03 | 1.125 | 0.497 | 0.002 | 0 | 1.755 | 0.813 | 0.604 | 0 | 0 | 0.135 | 0.371 | 0.553 | 1.312 | 1.165 | 1.422 | 2.262 | 1.739 | 1.625 | 2.192 | 2.046 | 1.859 | 1.798 | 1.389 | 1.21 | 1.116 | 0.938 | 0.606 | 0.57 | 0.649 | 0.303 | 0.539 | 0.726 | 1.176 | 0.792 | 0.487 | 0.475 | 0.154 | 0.337 | 0.665 | 0.868 | 0.639 | 0.583 | 0.689 | 0.4 | 0.588 | 0.577 | 0.697 | 0.421 | 0.2 | 0.179 | 0.351 | 0.316 | 0.199 | 0 | 0.164 | 0.429 | 0.394 | 0.323 | 0.307 | 0.456 | 0.598 | 0.533 | 0.755 | 0.712 | 0.686 | 0.756 | 0.547 | 0.54 | 0.799 | 0.789 | 0.727 | 0.663 | 0.507 | 0.375 | 0.371 | 0.304 | 0.249 | 0.238 | 0.236 | 0.182 | 0.16 | 0.113 | 0.338 | 0.444 | 0.407 | 0.492 | 0.442 | 0.532 | 0.397 | 0.349 | 0.309 | 0.204 | 0.201 | 0.648031328 | | fashion_lenet5 | 0 | 1.667 | 1.429 | 2.5 | 3.333 | 3 | 2.727 | 1.667 | 1.538 | 0.714 | 1.333 | 1.875 | 1.765 | 1.667 | 1.579 | 1.5 | 1.429 | 1.818 | 2.174 | 1.667 | 1.6 | 1.538 | 1.852 | 1.071 | 0.69 | 0.333 | 0.645 | 0.938 | 0.909 | 0.294 | 0.571 | 0.556 | 0.811 | 1.053 | 1.026 | 1.25 | 1.22 | 0.954 | 0.937 | 0.922 | 1.129 | 0.893 | 0.878 | 1.074 | 1.056 | 1.04 | 1.22 | 1.394 | 1.372 | 1.536 | 1.513 | 1.49 | 1.293 | 1.447 | 1.596 | 1.406 | 1.387 | 1.531 | 1.51 | 1.646 | 1.778 | 1.906 | 1.882 | 1.858 | 1.979 | 1.811 | 1.648 | 1.768 | 1.747 | 1.727 | 1.707 | 1.687 | 1.798 | 1.522 | 1.505 | 1.49 | 1.352 | 1.338 | 1.325 | 1.309 | 1.412 | 1.279 | 1.379 | 1.478 | 1.348 | 1.333 | 1.429 | 1.305 | 1.291 | 1.383 | 1.473 | 1.562 | 1.649 | 1.531 | 1.515 | 1.411426 | |
      | 0 | 1.667 | 1.429 | 2.5 | 3.333 | 3 | 2.727 | 1.667 | 1.538 | 0.714 | 1.333 | 1.875 | 1.765 | 1.667 | 2.105 | 1.5 | 1.905 | 2.273 | 2.609 | 2.5 | 2.8 | 2.692 | 2.963 | 2.857 | 3.103 | 2.667 | 2.581 | 2.187 | 1.818 | 2.059 | 2.286 | 2.222 | 2.162 | 1.842 | 1.538 | 1.75 | 1.22 | 1.192 | 1.402 | 1.604 | 1.796 | 1.11 | 1.091 | 1.074 | 1.056 | 1.04 | 0.828 | 0.817 | 0.994 | 0.981 | 0.785 | 0.954 | 0.942 | 0.93 | 1.088 | 1.24 | 1.387 | 1.531 | 1.192 | 1.178 | 1.317 | 1.3 | 1.136 | 1.122 | 1.254 | 1.24 | 1.367 | 1.49 | 1.336 | 1.321 | 1.44 | 1.424 | 1.539 | 1.394 | 1.505 | 1.365 | 1.352 | 1.216 | 1.204 | 1.309 | 1.412 | 1.511 | 1.494 | 1.478 | 1.573 | 1.333 | 1.319 | 1.413 | 1.398 | 1.489 | 1.368 | 1.354 | 1.237 | 1.326 | 1.313 | 1.570157101 |

    pace的效果存在一定的随机性，但总体来看效果也非常好，与我们的方法对比，pace在lenet1、lenet4、lenet5、vgg16上的效果差不多，在renet20上的效果优于pace，在fashion和svhn上的效果不如pace。。。。。