==========
Args:Namespace(dataset='market1501', batch_size=256, workers=4, height=256, width=128, num_instances=8, eps=0.6, eps_gap=0.02, k1=30, k2=6, arch='resnet_ibn152a', pretrained_path='/home/ma-user/work/Projects/ReIDNet_Finetune/TransReID/log/bs_exp/ResNet152_IBN_MSMT17/64bs_lr0.0004_ep120_warm20_seed0/resnet152_ibn_120.pth', features=0, dropout=0, momentum=0.2, drop_path_rate=0.3, hw_ratio=2, self_norm=True, conv_stem=False, lr=0.00035, weight_decay=0.0005, optimizer='SGD', epochs=50, iters=200, step_size=20, seed=1, print_freq=20, eval_step=10, temp=0.05, data_dir='/home/ma-user/work/Projects/ReIDNet_Finetune/CContrast/data', logs_dir='log/msmt2market/resnet152_ibn_cion', pooling_type='gem', use_hard=True)
==========
==> Load unlabeled dataset
=> Market1501 loaded
Dataset statistics:
  ----------------------------------------
  subset   | # ids | # images | # cameras
  ----------------------------------------
  train    |   751 |    12936 |         6
  query    |   750 |     3368 |         6
  gallery  |   751 |    15913 |         6
  ----------------------------------------
pooling_type: gem
checkpoint loaded!
optimizer: SGD
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.202 (0.612)	Data 0.000 (0.029)	
Computing jaccard distance...
Jaccard distance computing time cost: 21.420740604400635
Clustering criterion: eps: 0.600
==> Statistics for epoch 0: 546 clusters
Epoch: [0][20/200]	Time 0.670 (1.087)	Data 0.001 (0.112)	Loss 2.846 (3.486)
Epoch: [0][40/200]	Time 0.576 (0.871)	Data 0.001 (0.093)	Loss 2.275 (3.062)
Epoch: [0][60/200]	Time 0.565 (0.797)	Data 0.001 (0.085)	Loss 1.900 (2.667)
Epoch: [0][80/200]	Time 0.567 (0.761)	Data 0.000 (0.081)	Loss 1.844 (2.461)
Epoch: [0][100/200]	Time 0.558 (0.739)	Data 0.000 (0.080)	Loss 1.753 (2.299)
Epoch: [0][120/200]	Time 2.050 (0.735)	Data 1.400 (0.090)	Loss 1.579 (2.184)
Epoch: [0][140/200]	Time 0.568 (0.724)	Data 0.002 (0.088)	Loss 1.200 (2.093)
Epoch: [0][160/200]	Time 0.650 (0.716)	Data 0.001 (0.086)	Loss 1.466 (2.028)
Epoch: [0][180/200]	Time 0.560 (0.708)	Data 0.000 (0.085)	Loss 1.227 (1.961)
Epoch: [0][200/200]	Time 0.535 (0.702)	Data 0.000 (0.083)	Loss 1.359 (1.909)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.201 (0.255)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.549102544784546
==> Statistics for epoch 1: 607 clusters
Epoch: [1][20/200]	Time 0.648 (0.691)	Data 0.001 (0.103)	Loss 1.126 (0.341)
Epoch: [1][40/200]	Time 0.545 (0.667)	Data 0.003 (0.087)	Loss 1.864 (0.855)
Epoch: [1][60/200]	Time 0.624 (0.661)	Data 0.001 (0.081)	Loss 1.840 (1.028)
Epoch: [1][80/200]	Time 0.557 (0.657)	Data 0.001 (0.079)	Loss 1.526 (1.103)
Epoch: [1][100/200]	Time 0.575 (0.656)	Data 0.001 (0.077)	Loss 1.852 (1.145)
Epoch: [1][120/200]	Time 0.653 (0.656)	Data 0.000 (0.076)	Loss 1.075 (1.174)
Epoch: [1][140/200]	Time 0.554 (0.655)	Data 0.000 (0.076)	Loss 1.129 (1.191)
Epoch: [1][160/200]	Time 0.658 (0.655)	Data 0.000 (0.075)	Loss 1.559 (1.201)
Epoch: [1][180/200]	Time 0.545 (0.655)	Data 0.000 (0.075)	Loss 1.798 (1.207)
Epoch: [1][200/200]	Time 0.562 (0.662)	Data 0.001 (0.082)	Loss 1.080 (1.207)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.212 (0.243)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.107229948043823
==> Statistics for epoch 2: 619 clusters
Epoch: [2][20/200]	Time 2.164 (0.695)	Data 1.423 (0.109)	Loss 1.276 (0.243)
Epoch: [2][40/200]	Time 0.564 (0.674)	Data 0.001 (0.092)	Loss 1.046 (0.696)
Epoch: [2][60/200]	Time 0.564 (0.667)	Data 0.001 (0.086)	Loss 1.094 (0.863)
Epoch: [2][80/200]	Time 0.568 (0.665)	Data 0.001 (0.083)	Loss 1.556 (0.943)
Epoch: [2][100/200]	Time 0.559 (0.663)	Data 0.001 (0.081)	Loss 1.120 (0.989)
Epoch: [2][120/200]	Time 0.643 (0.662)	Data 0.001 (0.080)	Loss 1.423 (1.037)
Epoch: [2][140/200]	Time 0.565 (0.660)	Data 0.001 (0.078)	Loss 1.013 (1.056)
Epoch: [2][160/200]	Time 0.665 (0.659)	Data 0.001 (0.078)	Loss 1.486 (1.070)
Epoch: [2][180/200]	Time 0.564 (0.658)	Data 0.001 (0.078)	Loss 1.118 (1.080)
Epoch: [2][200/200]	Time 0.555 (0.658)	Data 0.001 (0.077)	Loss 1.278 (1.085)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.204 (0.242)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.091212034225464
==> Statistics for epoch 3: 622 clusters
Epoch: [3][20/200]	Time 2.044 (0.701)	Data 1.401 (0.108)	Loss 0.984 (0.223)
Epoch: [3][40/200]	Time 0.549 (0.679)	Data 0.001 (0.092)	Loss 0.791 (0.619)
Epoch: [3][60/200]	Time 0.661 (0.673)	Data 0.001 (0.086)	Loss 0.817 (0.750)
Epoch: [3][80/200]	Time 0.556 (0.667)	Data 0.003 (0.083)	Loss 1.035 (0.833)
Epoch: [3][100/200]	Time 0.663 (0.665)	Data 0.001 (0.080)	Loss 1.116 (0.874)
Epoch: [3][120/200]	Time 0.568 (0.663)	Data 0.001 (0.078)	Loss 0.915 (0.907)
Epoch: [3][140/200]	Time 0.668 (0.662)	Data 0.001 (0.078)	Loss 0.640 (0.911)
Epoch: [3][160/200]	Time 0.568 (0.661)	Data 0.001 (0.077)	Loss 0.844 (0.918)
Epoch: [3][180/200]	Time 0.541 (0.659)	Data 0.001 (0.076)	Loss 0.877 (0.926)
Epoch: [3][200/200]	Time 0.562 (0.659)	Data 0.001 (0.076)	Loss 1.238 (0.941)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.189 (0.239)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.46725034713745
==> Statistics for epoch 4: 626 clusters
Epoch: [4][20/200]	Time 2.071 (0.697)	Data 1.470 (0.111)	Loss 0.916 (0.187)
Epoch: [4][40/200]	Time 0.707 (0.679)	Data 0.001 (0.095)	Loss 1.134 (0.566)
Epoch: [4][60/200]	Time 0.561 (0.669)	Data 0.001 (0.086)	Loss 0.798 (0.705)
Epoch: [4][80/200]	Time 0.692 (0.666)	Data 0.001 (0.083)	Loss 1.139 (0.797)
Epoch: [4][100/200]	Time 0.546 (0.663)	Data 0.001 (0.080)	Loss 0.802 (0.827)
Epoch: [4][120/200]	Time 0.556 (0.661)	Data 0.001 (0.079)	Loss 1.002 (0.845)
Epoch: [4][140/200]	Time 0.561 (0.660)	Data 0.001 (0.078)	Loss 0.779 (0.864)
Epoch: [4][160/200]	Time 0.573 (0.659)	Data 0.001 (0.078)	Loss 0.768 (0.883)
Epoch: [4][180/200]	Time 0.668 (0.659)	Data 0.001 (0.077)	Loss 0.907 (0.887)
Epoch: [4][200/200]	Time 0.572 (0.659)	Data 0.001 (0.077)	Loss 1.060 (0.894)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.305 (0.250)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.61199188232422
==> Statistics for epoch 5: 623 clusters
Epoch: [5][20/200]	Time 2.035 (0.688)	Data 1.407 (0.106)	Loss 1.146 (0.195)
Epoch: [5][40/200]	Time 0.564 (0.674)	Data 0.001 (0.090)	Loss 1.061 (0.550)
Epoch: [5][60/200]	Time 0.581 (0.667)	Data 0.001 (0.085)	Loss 0.667 (0.678)
Epoch: [5][80/200]	Time 0.560 (0.663)	Data 0.001 (0.080)	Loss 0.678 (0.737)
Epoch: [5][100/200]	Time 0.539 (0.662)	Data 0.001 (0.078)	Loss 1.019 (0.765)
Epoch: [5][120/200]	Time 0.673 (0.660)	Data 0.001 (0.077)	Loss 0.963 (0.793)
Epoch: [5][140/200]	Time 0.565 (0.659)	Data 0.001 (0.076)	Loss 1.391 (0.823)
Epoch: [5][160/200]	Time 0.540 (0.658)	Data 0.001 (0.075)	Loss 0.651 (0.831)
Epoch: [5][180/200]	Time 0.541 (0.658)	Data 0.001 (0.074)	Loss 0.883 (0.835)
Epoch: [5][200/200]	Time 0.565 (0.657)	Data 0.001 (0.074)	Loss 0.552 (0.834)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.198 (0.248)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.728490591049194
==> Statistics for epoch 6: 625 clusters
Epoch: [6][20/200]	Time 2.023 (0.700)	Data 1.427 (0.109)	Loss 0.739 (0.157)
Epoch: [6][40/200]	Time 0.560 (0.673)	Data 0.001 (0.090)	Loss 0.704 (0.481)
Epoch: [6][60/200]	Time 0.695 (0.667)	Data 0.002 (0.085)	Loss 0.619 (0.609)
Epoch: [6][80/200]	Time 0.563 (0.662)	Data 0.002 (0.081)	Loss 0.756 (0.683)
Epoch: [6][100/200]	Time 0.642 (0.661)	Data 0.001 (0.078)	Loss 0.846 (0.721)
Epoch: [6][120/200]	Time 0.547 (0.660)	Data 0.001 (0.078)	Loss 0.836 (0.745)
Epoch: [6][140/200]	Time 0.676 (0.661)	Data 0.001 (0.077)	Loss 0.832 (0.755)
Epoch: [6][160/200]	Time 0.572 (0.661)	Data 0.001 (0.077)	Loss 0.705 (0.772)
Epoch: [6][180/200]	Time 0.561 (0.661)	Data 0.001 (0.077)	Loss 0.992 (0.777)
Epoch: [6][200/200]	Time 0.553 (0.662)	Data 0.001 (0.077)	Loss 0.798 (0.787)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.292 (0.250)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.068249464035034
==> Statistics for epoch 7: 622 clusters
Epoch: [7][20/200]	Time 2.033 (0.692)	Data 1.421 (0.112)	Loss 0.764 (0.159)
Epoch: [7][40/200]	Time 0.545 (0.672)	Data 0.001 (0.091)	Loss 0.624 (0.478)
Epoch: [7][60/200]	Time 0.539 (0.665)	Data 0.001 (0.085)	Loss 0.779 (0.584)
Epoch: [7][80/200]	Time 0.562 (0.663)	Data 0.002 (0.082)	Loss 0.649 (0.634)
Epoch: [7][100/200]	Time 0.546 (0.661)	Data 0.001 (0.079)	Loss 0.908 (0.679)
Epoch: [7][120/200]	Time 0.655 (0.661)	Data 0.001 (0.078)	Loss 0.796 (0.704)
Epoch: [7][140/200]	Time 0.566 (0.660)	Data 0.001 (0.077)	Loss 0.812 (0.722)
Epoch: [7][160/200]	Time 0.570 (0.661)	Data 0.001 (0.077)	Loss 0.594 (0.730)
Epoch: [7][180/200]	Time 0.569 (0.660)	Data 0.002 (0.075)	Loss 0.519 (0.735)
Epoch: [7][200/200]	Time 0.540 (0.660)	Data 0.001 (0.075)	Loss 0.625 (0.747)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.202 (0.246)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.70882487297058
==> Statistics for epoch 8: 625 clusters
Epoch: [8][20/200]	Time 2.090 (0.691)	Data 1.351 (0.104)	Loss 0.769 (0.152)
Epoch: [8][40/200]	Time 0.558 (0.669)	Data 0.001 (0.091)	Loss 0.879 (0.453)
Epoch: [8][60/200]	Time 0.655 (0.663)	Data 0.001 (0.084)	Loss 0.783 (0.567)
Epoch: [8][80/200]	Time 0.568 (0.661)	Data 0.001 (0.082)	Loss 0.627 (0.607)
Epoch: [8][100/200]	Time 0.545 (0.660)	Data 0.001 (0.081)	Loss 0.998 (0.646)
Epoch: [8][120/200]	Time 0.567 (0.661)	Data 0.001 (0.080)	Loss 0.671 (0.661)
Epoch: [8][140/200]	Time 0.566 (0.661)	Data 0.001 (0.079)	Loss 0.888 (0.674)
Epoch: [8][160/200]	Time 0.568 (0.661)	Data 0.001 (0.078)	Loss 0.677 (0.691)
Epoch: [8][180/200]	Time 0.559 (0.660)	Data 0.001 (0.077)	Loss 0.598 (0.698)
Epoch: [8][200/200]	Time 0.687 (0.660)	Data 0.001 (0.077)	Loss 0.709 (0.709)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.199 (0.245)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.97384810447693
==> Statistics for epoch 9: 621 clusters
Epoch: [9][20/200]	Time 2.025 (0.687)	Data 1.385 (0.103)	Loss 0.590 (0.134)
Epoch: [9][40/200]	Time 0.688 (0.669)	Data 0.001 (0.088)	Loss 0.606 (0.420)
Epoch: [9][60/200]	Time 0.548 (0.660)	Data 0.001 (0.082)	Loss 0.814 (0.524)
Epoch: [9][80/200]	Time 0.575 (0.659)	Data 0.002 (0.080)	Loss 0.843 (0.584)
Epoch: [9][100/200]	Time 0.565 (0.659)	Data 0.001 (0.078)	Loss 0.667 (0.606)
Epoch: [9][120/200]	Time 0.561 (0.658)	Data 0.001 (0.076)	Loss 0.747 (0.630)
Epoch: [9][140/200]	Time 0.560 (0.659)	Data 0.001 (0.075)	Loss 0.725 (0.649)
Epoch: [9][160/200]	Time 0.576 (0.659)	Data 0.002 (0.075)	Loss 0.699 (0.658)
Epoch: [9][180/200]	Time 0.659 (0.659)	Data 0.001 (0.075)	Loss 0.970 (0.670)
Epoch: [9][200/200]	Time 0.576 (0.658)	Data 0.001 (0.074)	Loss 0.807 (0.675)
Extract Features: [50/76]	Time 0.312 (0.246)	Data 0.000 (0.023)	
Mean AP: 93.6%

 * Finished epoch   9  model mAP: 93.6%  best: 93.6% *

==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.278 (0.248)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.567838430404663
==> Statistics for epoch 10: 616 clusters
Epoch: [10][20/200]	Time 2.089 (0.794)	Data 1.449 (0.110)	Loss 0.633 (0.128)
Epoch: [10][40/200]	Time 0.568 (0.728)	Data 0.001 (0.091)	Loss 0.584 (0.405)
Epoch: [10][60/200]	Time 0.579 (0.704)	Data 0.002 (0.085)	Loss 0.589 (0.501)
Epoch: [10][80/200]	Time 0.561 (0.693)	Data 0.002 (0.083)	Loss 0.962 (0.564)
Epoch: [10][100/200]	Time 0.567 (0.685)	Data 0.001 (0.081)	Loss 0.739 (0.584)
Epoch: [10][120/200]	Time 0.686 (0.683)	Data 0.001 (0.081)	Loss 0.832 (0.611)
Epoch: [10][140/200]	Time 0.568 (0.679)	Data 0.001 (0.080)	Loss 0.480 (0.626)
Epoch: [10][160/200]	Time 0.674 (0.677)	Data 0.001 (0.080)	Loss 0.572 (0.630)
Epoch: [10][180/200]	Time 0.558 (0.674)	Data 0.001 (0.079)	Loss 0.571 (0.632)
Epoch: [10][200/200]	Time 0.566 (0.673)	Data 0.001 (0.079)	Loss 0.833 (0.640)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.201 (0.247)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.570225954055786
==> Statistics for epoch 11: 617 clusters
Epoch: [11][20/200]	Time 2.104 (0.704)	Data 1.477 (0.118)	Loss 0.730 (0.131)
Epoch: [11][40/200]	Time 0.551 (0.676)	Data 0.001 (0.096)	Loss 0.600 (0.397)
Epoch: [11][60/200]	Time 0.567 (0.670)	Data 0.002 (0.088)	Loss 0.978 (0.500)
Epoch: [11][80/200]	Time 0.540 (0.669)	Data 0.003 (0.085)	Loss 0.773 (0.544)
Epoch: [11][100/200]	Time 0.646 (0.666)	Data 0.001 (0.084)	Loss 0.778 (0.578)
Epoch: [11][120/200]	Time 0.567 (0.662)	Data 0.001 (0.082)	Loss 0.584 (0.587)
Epoch: [11][140/200]	Time 0.561 (0.661)	Data 0.001 (0.082)	Loss 0.588 (0.609)
Epoch: [11][160/200]	Time 0.565 (0.661)	Data 0.001 (0.081)	Loss 0.670 (0.615)
Epoch: [11][180/200]	Time 0.558 (0.661)	Data 0.001 (0.081)	Loss 0.584 (0.624)
Epoch: [11][200/200]	Time 0.575 (0.661)	Data 0.001 (0.080)	Loss 0.625 (0.628)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.198 (0.246)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.62064838409424
==> Statistics for epoch 12: 617 clusters
Epoch: [12][20/200]	Time 2.125 (0.699)	Data 1.531 (0.119)	Loss 0.690 (0.128)
Epoch: [12][40/200]	Time 0.687 (0.684)	Data 0.001 (0.098)	Loss 0.787 (0.380)
Epoch: [12][60/200]	Time 0.568 (0.674)	Data 0.001 (0.091)	Loss 0.469 (0.460)
Epoch: [12][80/200]	Time 0.564 (0.671)	Data 0.001 (0.088)	Loss 0.709 (0.516)
Epoch: [12][100/200]	Time 0.570 (0.668)	Data 0.002 (0.086)	Loss 0.556 (0.540)
Epoch: [12][120/200]	Time 0.558 (0.666)	Data 0.001 (0.084)	Loss 0.483 (0.557)
Epoch: [12][140/200]	Time 0.536 (0.664)	Data 0.001 (0.082)	Loss 0.587 (0.561)
Epoch: [12][160/200]	Time 0.559 (0.662)	Data 0.001 (0.081)	Loss 0.683 (0.579)
Epoch: [12][180/200]	Time 0.538 (0.661)	Data 0.001 (0.081)	Loss 0.698 (0.581)
Epoch: [12][200/200]	Time 0.539 (0.661)	Data 0.001 (0.082)	Loss 0.711 (0.588)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.199 (0.247)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.466551065444946
==> Statistics for epoch 13: 615 clusters
Epoch: [13][20/200]	Time 2.346 (0.717)	Data 1.590 (0.131)	Loss 0.462 (0.109)
Epoch: [13][40/200]	Time 0.568 (0.691)	Data 0.001 (0.106)	Loss 0.615 (0.361)
Epoch: [13][60/200]	Time 0.566 (0.680)	Data 0.001 (0.094)	Loss 0.587 (0.448)
Epoch: [13][80/200]	Time 0.568 (0.676)	Data 0.002 (0.089)	Loss 0.620 (0.506)
Epoch: [13][100/200]	Time 0.546 (0.671)	Data 0.001 (0.086)	Loss 0.782 (0.538)
Epoch: [13][120/200]	Time 0.557 (0.669)	Data 0.001 (0.085)	Loss 0.591 (0.564)
Epoch: [13][140/200]	Time 0.563 (0.667)	Data 0.001 (0.083)	Loss 0.479 (0.570)
Epoch: [13][160/200]	Time 0.667 (0.664)	Data 0.001 (0.081)	Loss 0.491 (0.576)
Epoch: [13][180/200]	Time 0.551 (0.664)	Data 0.001 (0.081)	Loss 0.888 (0.586)
Epoch: [13][200/200]	Time 0.545 (0.663)	Data 0.001 (0.080)	Loss 0.539 (0.590)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.194 (0.250)	Data 0.000 (0.032)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.752787828445435
==> Statistics for epoch 14: 613 clusters
Epoch: [14][20/200]	Time 2.111 (0.698)	Data 1.460 (0.110)	Loss 0.488 (0.107)
Epoch: [14][40/200]	Time 0.567 (0.678)	Data 0.001 (0.093)	Loss 0.608 (0.345)
Epoch: [14][60/200]	Time 0.563 (0.672)	Data 0.001 (0.085)	Loss 0.467 (0.425)
Epoch: [14][80/200]	Time 0.541 (0.670)	Data 0.001 (0.083)	Loss 0.783 (0.477)
Epoch: [14][100/200]	Time 0.551 (0.667)	Data 0.001 (0.082)	Loss 0.527 (0.505)
Epoch: [14][120/200]	Time 0.544 (0.666)	Data 0.002 (0.080)	Loss 0.719 (0.520)
Epoch: [14][140/200]	Time 0.663 (0.665)	Data 0.001 (0.080)	Loss 0.503 (0.539)
Epoch: [14][160/200]	Time 0.556 (0.664)	Data 0.001 (0.080)	Loss 0.568 (0.545)
Epoch: [14][180/200]	Time 0.557 (0.662)	Data 0.001 (0.080)	Loss 0.639 (0.555)
Epoch: [14][200/200]	Time 0.576 (0.662)	Data 0.001 (0.079)	Loss 0.512 (0.561)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.302 (0.247)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.19036889076233
==> Statistics for epoch 15: 613 clusters
Epoch: [15][20/200]	Time 2.134 (0.701)	Data 1.518 (0.120)	Loss 0.509 (0.102)
Epoch: [15][40/200]	Time 0.560 (0.683)	Data 0.001 (0.101)	Loss 0.584 (0.333)
Epoch: [15][60/200]	Time 0.562 (0.673)	Data 0.001 (0.091)	Loss 0.666 (0.416)
Epoch: [15][80/200]	Time 0.671 (0.671)	Data 0.002 (0.089)	Loss 0.594 (0.454)
Epoch: [15][100/200]	Time 0.565 (0.667)	Data 0.001 (0.086)	Loss 0.515 (0.478)
Epoch: [15][120/200]	Time 0.673 (0.667)	Data 0.001 (0.085)	Loss 0.604 (0.504)
Epoch: [15][140/200]	Time 0.567 (0.665)	Data 0.001 (0.083)	Loss 0.737 (0.515)
Epoch: [15][160/200]	Time 0.636 (0.664)	Data 0.001 (0.081)	Loss 0.823 (0.525)
Epoch: [15][180/200]	Time 0.569 (0.662)	Data 0.003 (0.080)	Loss 0.426 (0.528)
Epoch: [15][200/200]	Time 0.556 (0.662)	Data 0.001 (0.080)	Loss 0.628 (0.535)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.195 (0.251)	Data 0.000 (0.029)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.647536039352417
==> Statistics for epoch 16: 613 clusters
Epoch: [16][20/200]	Time 2.148 (0.700)	Data 1.525 (0.115)	Loss 0.513 (0.105)
Epoch: [16][40/200]	Time 0.558 (0.680)	Data 0.001 (0.097)	Loss 0.519 (0.326)
Epoch: [16][60/200]	Time 0.689 (0.676)	Data 0.001 (0.089)	Loss 0.397 (0.419)
Epoch: [16][80/200]	Time 0.566 (0.672)	Data 0.001 (0.085)	Loss 0.348 (0.455)
Epoch: [16][100/200]	Time 0.688 (0.669)	Data 0.001 (0.084)	Loss 0.547 (0.480)
Epoch: [16][120/200]	Time 0.545 (0.666)	Data 0.001 (0.082)	Loss 0.486 (0.488)
Epoch: [16][140/200]	Time 0.559 (0.667)	Data 0.001 (0.083)	Loss 0.554 (0.499)
Epoch: [16][160/200]	Time 0.577 (0.664)	Data 0.001 (0.082)	Loss 0.521 (0.512)
Epoch: [16][180/200]	Time 0.559 (0.666)	Data 0.001 (0.082)	Loss 0.327 (0.516)
Epoch: [16][200/200]	Time 0.574 (0.666)	Data 0.001 (0.082)	Loss 0.660 (0.525)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.321 (0.250)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.09821128845215
==> Statistics for epoch 17: 611 clusters
Epoch: [17][20/200]	Time 2.100 (0.697)	Data 1.468 (0.114)	Loss 0.558 (0.105)
Epoch: [17][40/200]	Time 0.583 (0.678)	Data 0.001 (0.096)	Loss 0.536 (0.320)
Epoch: [17][60/200]	Time 0.565 (0.670)	Data 0.002 (0.090)	Loss 0.521 (0.404)
Epoch: [17][80/200]	Time 0.539 (0.670)	Data 0.002 (0.089)	Loss 0.500 (0.450)
Epoch: [17][100/200]	Time 0.558 (0.667)	Data 0.001 (0.086)	Loss 0.510 (0.477)
Epoch: [17][120/200]	Time 0.627 (0.666)	Data 0.001 (0.083)	Loss 0.534 (0.492)
Epoch: [17][140/200]	Time 0.539 (0.665)	Data 0.001 (0.082)	Loss 0.653 (0.496)
Epoch: [17][160/200]	Time 0.561 (0.665)	Data 0.001 (0.082)	Loss 0.605 (0.505)
Epoch: [17][180/200]	Time 0.560 (0.664)	Data 0.001 (0.081)	Loss 0.654 (0.515)
Epoch: [17][200/200]	Time 0.558 (0.664)	Data 0.001 (0.081)	Loss 0.514 (0.517)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.190 (0.247)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.615832328796387
==> Statistics for epoch 18: 611 clusters
Epoch: [18][20/200]	Time 2.279 (0.712)	Data 1.544 (0.122)	Loss 0.770 (0.117)
Epoch: [18][40/200]	Time 0.589 (0.680)	Data 0.002 (0.097)	Loss 0.497 (0.304)
Epoch: [18][60/200]	Time 0.655 (0.675)	Data 0.001 (0.091)	Loss 0.380 (0.375)
Epoch: [18][80/200]	Time 0.564 (0.669)	Data 0.002 (0.087)	Loss 0.419 (0.431)
Epoch: [18][100/200]	Time 0.549 (0.666)	Data 0.001 (0.084)	Loss 0.528 (0.455)
Epoch: [18][120/200]	Time 0.564 (0.663)	Data 0.001 (0.082)	Loss 0.573 (0.473)
Epoch: [18][140/200]	Time 0.550 (0.661)	Data 0.001 (0.081)	Loss 0.419 (0.485)
Epoch: [18][160/200]	Time 0.661 (0.661)	Data 0.001 (0.081)	Loss 0.675 (0.485)
Epoch: [18][180/200]	Time 0.565 (0.662)	Data 0.002 (0.080)	Loss 0.681 (0.489)
Epoch: [18][200/200]	Time 0.578 (0.662)	Data 0.001 (0.080)	Loss 0.555 (0.496)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.197 (0.251)	Data 0.000 (0.031)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.739962100982666
==> Statistics for epoch 19: 612 clusters
Epoch: [19][20/200]	Time 2.043 (0.707)	Data 1.430 (0.113)	Loss 0.381 (0.088)
Epoch: [19][40/200]	Time 0.548 (0.688)	Data 0.001 (0.098)	Loss 0.508 (0.279)
Epoch: [19][60/200]	Time 0.556 (0.678)	Data 0.001 (0.092)	Loss 0.334 (0.352)
Epoch: [19][80/200]	Time 0.565 (0.674)	Data 0.002 (0.089)	Loss 0.517 (0.391)
Epoch: [19][100/200]	Time 0.658 (0.672)	Data 0.001 (0.087)	Loss 0.490 (0.418)
Epoch: [19][120/200]	Time 0.560 (0.670)	Data 0.001 (0.086)	Loss 0.633 (0.438)
Epoch: [19][140/200]	Time 0.553 (0.668)	Data 0.001 (0.085)	Loss 0.538 (0.449)
Epoch: [19][160/200]	Time 0.576 (0.667)	Data 0.001 (0.083)	Loss 0.619 (0.455)
Epoch: [19][180/200]	Time 0.545 (0.666)	Data 0.001 (0.082)	Loss 0.629 (0.458)
Epoch: [19][200/200]	Time 0.557 (0.665)	Data 0.001 (0.082)	Loss 0.668 (0.467)
Extract Features: [50/76]	Time 0.202 (0.244)	Data 0.000 (0.026)	
Mean AP: 93.7%

 * Finished epoch  19  model mAP: 93.7%  best: 93.7% *

==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.209 (0.245)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.87646245956421
==> Statistics for epoch 20: 614 clusters
Epoch: [20][20/200]	Time 2.259 (0.712)	Data 1.638 (0.126)	Loss 0.634 (0.105)
Epoch: [20][40/200]	Time 0.567 (0.683)	Data 0.001 (0.098)	Loss 0.387 (0.296)
Epoch: [20][60/200]	Time 0.577 (0.674)	Data 0.001 (0.088)	Loss 0.635 (0.363)
Epoch: [20][80/200]	Time 0.582 (0.673)	Data 0.001 (0.086)	Loss 0.537 (0.392)
Epoch: [20][100/200]	Time 0.566 (0.671)	Data 0.001 (0.083)	Loss 0.680 (0.404)
Epoch: [20][120/200]	Time 0.549 (0.668)	Data 0.001 (0.080)	Loss 0.398 (0.420)
Epoch: [20][140/200]	Time 0.567 (0.666)	Data 0.001 (0.079)	Loss 0.569 (0.428)
Epoch: [20][160/200]	Time 0.572 (0.664)	Data 0.001 (0.077)	Loss 0.401 (0.435)
Epoch: [20][180/200]	Time 0.564 (0.662)	Data 0.001 (0.076)	Loss 0.467 (0.440)
Epoch: [20][200/200]	Time 0.563 (0.661)	Data 0.001 (0.075)	Loss 0.502 (0.449)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.198 (0.246)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.817227602005005
==> Statistics for epoch 21: 612 clusters
Epoch: [21][20/200]	Time 2.173 (0.695)	Data 1.546 (0.115)	Loss 0.263 (0.078)
Epoch: [21][40/200]	Time 0.671 (0.676)	Data 0.001 (0.096)	Loss 0.750 (0.285)
Epoch: [21][60/200]	Time 0.570 (0.669)	Data 0.001 (0.088)	Loss 0.732 (0.346)
Epoch: [21][80/200]	Time 0.679 (0.667)	Data 0.002 (0.085)	Loss 0.369 (0.377)
Epoch: [21][100/200]	Time 0.550 (0.666)	Data 0.002 (0.084)	Loss 0.473 (0.412)
Epoch: [21][120/200]	Time 0.556 (0.665)	Data 0.002 (0.083)	Loss 0.442 (0.429)
Epoch: [21][140/200]	Time 0.568 (0.665)	Data 0.001 (0.083)	Loss 0.560 (0.438)
Epoch: [21][160/200]	Time 0.576 (0.667)	Data 0.001 (0.084)	Loss 0.386 (0.447)
Epoch: [21][180/200]	Time 0.634 (0.665)	Data 0.001 (0.083)	Loss 0.691 (0.452)
Epoch: [21][200/200]	Time 0.580 (0.664)	Data 0.001 (0.082)	Loss 0.348 (0.462)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.318 (0.254)	Data 0.000 (0.029)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.760512828826904
==> Statistics for epoch 22: 609 clusters
Epoch: [22][20/200]	Time 2.120 (0.693)	Data 1.509 (0.116)	Loss 0.458 (0.091)
Epoch: [22][40/200]	Time 0.567 (0.680)	Data 0.001 (0.095)	Loss 0.549 (0.279)
Epoch: [22][60/200]	Time 0.560 (0.668)	Data 0.001 (0.088)	Loss 0.316 (0.327)
Epoch: [22][80/200]	Time 0.566 (0.667)	Data 0.001 (0.086)	Loss 0.614 (0.373)
Epoch: [22][100/200]	Time 0.565 (0.665)	Data 0.001 (0.084)	Loss 0.371 (0.398)
Epoch: [22][120/200]	Time 0.667 (0.664)	Data 0.001 (0.082)	Loss 0.342 (0.416)
Epoch: [22][140/200]	Time 0.572 (0.663)	Data 0.001 (0.081)	Loss 0.290 (0.422)
Epoch: [22][160/200]	Time 0.564 (0.662)	Data 0.001 (0.081)	Loss 0.654 (0.438)
Epoch: [22][180/200]	Time 0.563 (0.663)	Data 0.001 (0.081)	Loss 0.417 (0.442)
Epoch: [22][200/200]	Time 0.543 (0.665)	Data 0.001 (0.082)	Loss 0.570 (0.449)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.204 (0.245)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.78816032409668
==> Statistics for epoch 23: 611 clusters
Epoch: [23][20/200]	Time 2.072 (0.702)	Data 1.452 (0.111)	Loss 0.467 (0.092)
Epoch: [23][40/200]	Time 0.548 (0.674)	Data 0.002 (0.088)	Loss 0.414 (0.286)
Epoch: [23][60/200]	Time 0.656 (0.669)	Data 0.001 (0.084)	Loss 0.623 (0.351)
Epoch: [23][80/200]	Time 0.549 (0.668)	Data 0.002 (0.083)	Loss 0.432 (0.385)
Epoch: [23][100/200]	Time 0.639 (0.668)	Data 0.001 (0.083)	Loss 0.351 (0.411)
Epoch: [23][120/200]	Time 0.570 (0.667)	Data 0.001 (0.082)	Loss 0.475 (0.422)
Epoch: [23][140/200]	Time 0.569 (0.665)	Data 0.001 (0.081)	Loss 0.597 (0.435)
Epoch: [23][160/200]	Time 0.566 (0.666)	Data 0.001 (0.081)	Loss 0.401 (0.439)
Epoch: [23][180/200]	Time 0.563 (0.664)	Data 0.001 (0.080)	Loss 0.358 (0.443)
Epoch: [23][200/200]	Time 0.560 (0.664)	Data 0.001 (0.080)	Loss 0.447 (0.448)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.208 (0.244)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.12023377418518
==> Statistics for epoch 24: 611 clusters
Epoch: [24][20/200]	Time 2.018 (0.696)	Data 1.384 (0.110)	Loss 0.415 (0.092)
Epoch: [24][40/200]	Time 0.668 (0.677)	Data 0.001 (0.093)	Loss 0.420 (0.296)
Epoch: [24][60/200]	Time 0.557 (0.669)	Data 0.001 (0.088)	Loss 0.288 (0.358)
Epoch: [24][80/200]	Time 0.670 (0.664)	Data 0.001 (0.084)	Loss 0.301 (0.394)
Epoch: [24][100/200]	Time 0.541 (0.662)	Data 0.001 (0.082)	Loss 0.646 (0.415)
Epoch: [24][120/200]	Time 0.643 (0.662)	Data 0.001 (0.081)	Loss 0.467 (0.423)
Epoch: [24][140/200]	Time 0.567 (0.664)	Data 0.001 (0.082)	Loss 0.354 (0.431)
Epoch: [24][160/200]	Time 0.574 (0.664)	Data 0.001 (0.081)	Loss 0.464 (0.434)
Epoch: [24][180/200]	Time 0.570 (0.663)	Data 0.001 (0.080)	Loss 0.604 (0.441)
Epoch: [24][200/200]	Time 0.566 (0.663)	Data 0.001 (0.079)	Loss 0.899 (0.446)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.200 (0.244)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.609506130218506
==> Statistics for epoch 25: 611 clusters
Epoch: [25][20/200]	Time 2.608 (0.728)	Data 1.864 (0.136)	Loss 0.660 (0.099)
Epoch: [25][40/200]	Time 0.562 (0.692)	Data 0.001 (0.109)	Loss 0.564 (0.295)
Epoch: [25][60/200]	Time 0.569 (0.680)	Data 0.001 (0.098)	Loss 0.442 (0.362)
Epoch: [25][80/200]	Time 0.571 (0.673)	Data 0.001 (0.091)	Loss 0.562 (0.389)
Epoch: [25][100/200]	Time 0.563 (0.670)	Data 0.001 (0.088)	Loss 0.390 (0.407)
Epoch: [25][120/200]	Time 0.572 (0.669)	Data 0.001 (0.086)	Loss 0.417 (0.420)
Epoch: [25][140/200]	Time 0.567 (0.667)	Data 0.001 (0.084)	Loss 0.521 (0.433)
Epoch: [25][160/200]	Time 0.655 (0.666)	Data 0.001 (0.082)	Loss 0.349 (0.435)
Epoch: [25][180/200]	Time 0.545 (0.666)	Data 0.001 (0.082)	Loss 0.465 (0.441)
Epoch: [25][200/200]	Time 0.562 (0.664)	Data 0.001 (0.081)	Loss 0.354 (0.443)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.200 (0.248)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.809378623962402
==> Statistics for epoch 26: 610 clusters
Epoch: [26][20/200]	Time 2.114 (0.702)	Data 1.508 (0.123)	Loss 0.252 (0.081)
Epoch: [26][40/200]	Time 0.683 (0.682)	Data 0.001 (0.100)	Loss 0.642 (0.270)
Epoch: [26][60/200]	Time 0.562 (0.678)	Data 0.001 (0.096)	Loss 0.433 (0.317)
Epoch: [26][80/200]	Time 0.555 (0.670)	Data 0.001 (0.090)	Loss 0.348 (0.351)
Epoch: [26][100/200]	Time 0.547 (0.669)	Data 0.001 (0.087)	Loss 0.557 (0.371)
Epoch: [26][120/200]	Time 0.545 (0.667)	Data 0.001 (0.086)	Loss 0.443 (0.385)
Epoch: [26][140/200]	Time 0.655 (0.667)	Data 0.001 (0.084)	Loss 0.470 (0.393)
Epoch: [26][160/200]	Time 0.561 (0.667)	Data 0.001 (0.083)	Loss 0.453 (0.400)
Epoch: [26][180/200]	Time 0.561 (0.666)	Data 0.001 (0.083)	Loss 0.340 (0.405)
Epoch: [26][200/200]	Time 0.574 (0.666)	Data 0.001 (0.082)	Loss 0.619 (0.414)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.198 (0.249)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.055524110794067
==> Statistics for epoch 27: 610 clusters
Epoch: [27][20/200]	Time 2.092 (0.705)	Data 1.435 (0.116)	Loss 0.479 (0.081)
Epoch: [27][40/200]	Time 0.697 (0.685)	Data 0.002 (0.095)	Loss 0.369 (0.263)
Epoch: [27][60/200]	Time 0.555 (0.676)	Data 0.001 (0.087)	Loss 0.347 (0.332)
Epoch: [27][80/200]	Time 0.540 (0.672)	Data 0.001 (0.083)	Loss 0.592 (0.371)
Epoch: [27][100/200]	Time 0.546 (0.671)	Data 0.001 (0.082)	Loss 0.342 (0.380)
Epoch: [27][120/200]	Time 0.565 (0.668)	Data 0.001 (0.081)	Loss 0.373 (0.398)
Epoch: [27][140/200]	Time 0.568 (0.666)	Data 0.001 (0.080)	Loss 0.593 (0.411)
Epoch: [27][160/200]	Time 0.566 (0.667)	Data 0.001 (0.082)	Loss 0.389 (0.416)
Epoch: [27][180/200]	Time 0.652 (0.667)	Data 0.001 (0.082)	Loss 0.601 (0.423)
Epoch: [27][200/200]	Time 0.540 (0.666)	Data 0.001 (0.082)	Loss 0.390 (0.424)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.295 (0.244)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.21742272377014
==> Statistics for epoch 28: 609 clusters
Epoch: [28][20/200]	Time 1.999 (0.698)	Data 1.380 (0.112)	Loss 0.480 (0.088)
Epoch: [28][40/200]	Time 0.595 (0.680)	Data 0.001 (0.095)	Loss 0.274 (0.271)
Epoch: [28][60/200]	Time 0.581 (0.672)	Data 0.001 (0.088)	Loss 0.472 (0.334)
Epoch: [28][80/200]	Time 0.561 (0.669)	Data 0.002 (0.085)	Loss 0.520 (0.374)
Epoch: [28][100/200]	Time 0.567 (0.665)	Data 0.001 (0.081)	Loss 0.463 (0.389)
Epoch: [28][120/200]	Time 0.546 (0.664)	Data 0.001 (0.080)	Loss 0.747 (0.404)
Epoch: [28][140/200]	Time 0.572 (0.664)	Data 0.001 (0.079)	Loss 0.445 (0.407)
Epoch: [28][160/200]	Time 0.580 (0.663)	Data 0.002 (0.078)	Loss 0.387 (0.414)
Epoch: [28][180/200]	Time 0.566 (0.664)	Data 0.001 (0.078)	Loss 0.341 (0.418)
Epoch: [28][200/200]	Time 0.563 (0.662)	Data 0.001 (0.077)	Loss 0.445 (0.423)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.199 (0.242)	Data 0.000 (0.022)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.65801763534546
==> Statistics for epoch 29: 609 clusters
Epoch: [29][20/200]	Time 2.143 (0.697)	Data 1.387 (0.107)	Loss 0.537 (0.090)
Epoch: [29][40/200]	Time 0.565 (0.676)	Data 0.001 (0.092)	Loss 0.433 (0.272)
Epoch: [29][60/200]	Time 0.598 (0.671)	Data 0.002 (0.086)	Loss 0.547 (0.332)
Epoch: [29][80/200]	Time 0.559 (0.672)	Data 0.001 (0.086)	Loss 0.582 (0.362)
Epoch: [29][100/200]	Time 0.557 (0.668)	Data 0.002 (0.084)	Loss 0.625 (0.381)
Epoch: [29][120/200]	Time 0.546 (0.666)	Data 0.002 (0.082)	Loss 0.642 (0.394)
Epoch: [29][140/200]	Time 0.553 (0.665)	Data 0.001 (0.082)	Loss 0.543 (0.404)
Epoch: [29][160/200]	Time 0.590 (0.665)	Data 0.001 (0.081)	Loss 0.506 (0.410)
Epoch: [29][180/200]	Time 0.563 (0.665)	Data 0.001 (0.081)	Loss 0.474 (0.418)
Epoch: [29][200/200]	Time 0.563 (0.665)	Data 0.001 (0.082)	Loss 0.552 (0.421)
Extract Features: [50/76]	Time 0.199 (0.246)	Data 0.000 (0.025)	
Mean AP: 94.0%

 * Finished epoch  29  model mAP: 94.0%  best: 94.0% *

==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.198 (0.245)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.728727102279663
==> Statistics for epoch 30: 610 clusters
Epoch: [30][20/200]	Time 2.007 (0.692)	Data 1.420 (0.106)	Loss 0.427 (0.074)
Epoch: [30][40/200]	Time 0.559 (0.682)	Data 0.001 (0.091)	Loss 0.495 (0.249)
Epoch: [30][60/200]	Time 0.564 (0.674)	Data 0.001 (0.086)	Loss 0.263 (0.309)
Epoch: [30][80/200]	Time 0.563 (0.672)	Data 0.002 (0.084)	Loss 0.573 (0.360)
Epoch: [30][100/200]	Time 0.560 (0.670)	Data 0.001 (0.083)	Loss 0.373 (0.381)
Epoch: [30][120/200]	Time 0.573 (0.668)	Data 0.001 (0.082)	Loss 0.499 (0.390)
Epoch: [30][140/200]	Time 0.686 (0.666)	Data 0.001 (0.080)	Loss 0.396 (0.400)
Epoch: [30][160/200]	Time 0.564 (0.666)	Data 0.001 (0.081)	Loss 0.536 (0.409)
Epoch: [30][180/200]	Time 0.621 (0.665)	Data 0.001 (0.080)	Loss 0.254 (0.411)
Epoch: [30][200/200]	Time 0.565 (0.663)	Data 0.001 (0.080)	Loss 0.196 (0.418)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.303 (0.249)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.3795907497406
==> Statistics for epoch 31: 609 clusters
Epoch: [31][20/200]	Time 2.207 (0.699)	Data 1.449 (0.116)	Loss 0.377 (0.078)
Epoch: [31][40/200]	Time 0.564 (0.677)	Data 0.001 (0.094)	Loss 0.388 (0.245)
Epoch: [31][60/200]	Time 0.565 (0.671)	Data 0.001 (0.088)	Loss 0.636 (0.317)
Epoch: [31][80/200]	Time 0.572 (0.670)	Data 0.002 (0.085)	Loss 0.381 (0.355)
Epoch: [31][100/200]	Time 0.560 (0.671)	Data 0.001 (0.084)	Loss 0.543 (0.372)
Epoch: [31][120/200]	Time 0.668 (0.667)	Data 0.001 (0.082)	Loss 0.707 (0.390)
Epoch: [31][140/200]	Time 0.569 (0.666)	Data 0.001 (0.081)	Loss 0.505 (0.403)
Epoch: [31][160/200]	Time 0.635 (0.665)	Data 0.001 (0.080)	Loss 0.522 (0.411)
Epoch: [31][180/200]	Time 0.565 (0.665)	Data 0.001 (0.080)	Loss 0.505 (0.419)
Epoch: [31][200/200]	Time 0.567 (0.664)	Data 0.001 (0.079)	Loss 0.504 (0.418)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.197 (0.252)	Data 0.000 (0.030)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.90362000465393
==> Statistics for epoch 32: 610 clusters
Epoch: [32][20/200]	Time 2.131 (0.696)	Data 1.524 (0.122)	Loss 0.404 (0.078)
Epoch: [32][40/200]	Time 0.569 (0.677)	Data 0.001 (0.099)	Loss 0.481 (0.260)
Epoch: [32][60/200]	Time 0.560 (0.669)	Data 0.001 (0.091)	Loss 0.360 (0.316)
Epoch: [32][80/200]	Time 0.564 (0.668)	Data 0.001 (0.089)	Loss 0.329 (0.347)
Epoch: [32][100/200]	Time 0.675 (0.665)	Data 0.001 (0.086)	Loss 0.487 (0.367)
Epoch: [32][120/200]	Time 0.557 (0.666)	Data 0.001 (0.087)	Loss 0.489 (0.378)
Epoch: [32][140/200]	Time 0.569 (0.663)	Data 0.002 (0.085)	Loss 0.369 (0.385)
Epoch: [32][160/200]	Time 0.578 (0.664)	Data 0.001 (0.085)	Loss 0.401 (0.390)
Epoch: [32][180/200]	Time 0.561 (0.663)	Data 0.001 (0.084)	Loss 0.349 (0.396)
Epoch: [32][200/200]	Time 0.542 (0.662)	Data 0.001 (0.083)	Loss 0.575 (0.402)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.201 (0.245)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.2304630279541
==> Statistics for epoch 33: 609 clusters
Epoch: [33][20/200]	Time 2.083 (0.696)	Data 1.475 (0.115)	Loss 0.373 (0.074)
Epoch: [33][40/200]	Time 0.697 (0.683)	Data 0.001 (0.097)	Loss 0.552 (0.272)
Epoch: [33][60/200]	Time 0.551 (0.674)	Data 0.001 (0.091)	Loss 0.399 (0.350)
Epoch: [33][80/200]	Time 0.562 (0.671)	Data 0.002 (0.087)	Loss 0.410 (0.386)
Epoch: [33][100/200]	Time 0.565 (0.672)	Data 0.001 (0.086)	Loss 0.452 (0.399)
Epoch: [33][120/200]	Time 0.560 (0.669)	Data 0.001 (0.083)	Loss 0.406 (0.409)
Epoch: [33][140/200]	Time 0.567 (0.667)	Data 0.001 (0.083)	Loss 0.571 (0.417)
Epoch: [33][160/200]	Time 0.565 (0.665)	Data 0.001 (0.082)	Loss 0.470 (0.423)
Epoch: [33][180/200]	Time 0.662 (0.664)	Data 0.008 (0.082)	Loss 0.544 (0.426)
Epoch: [33][200/200]	Time 0.564 (0.662)	Data 0.001 (0.081)	Loss 0.401 (0.429)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.298 (0.246)	Data 0.000 (0.025)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.685116052627563
==> Statistics for epoch 34: 608 clusters
Epoch: [34][20/200]	Time 2.145 (0.697)	Data 1.500 (0.111)	Loss 0.329 (0.067)
Epoch: [34][40/200]	Time 0.572 (0.677)	Data 0.001 (0.094)	Loss 0.401 (0.227)
Epoch: [34][60/200]	Time 0.538 (0.675)	Data 0.001 (0.092)	Loss 0.663 (0.306)
Epoch: [34][80/200]	Time 0.567 (0.671)	Data 0.002 (0.087)	Loss 0.590 (0.341)
Epoch: [34][100/200]	Time 0.542 (0.669)	Data 0.001 (0.086)	Loss 0.569 (0.370)
Epoch: [34][120/200]	Time 0.645 (0.669)	Data 0.001 (0.085)	Loss 0.366 (0.386)
Epoch: [34][140/200]	Time 0.566 (0.666)	Data 0.001 (0.083)	Loss 0.470 (0.399)
Epoch: [34][160/200]	Time 0.706 (0.665)	Data 0.001 (0.083)	Loss 0.281 (0.407)
Epoch: [34][180/200]	Time 0.561 (0.665)	Data 0.001 (0.082)	Loss 0.482 (0.415)
Epoch: [34][200/200]	Time 0.565 (0.664)	Data 0.001 (0.082)	Loss 0.621 (0.418)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.201 (0.249)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.82594871520996
==> Statistics for epoch 35: 608 clusters
Epoch: [35][20/200]	Time 2.271 (0.717)	Data 1.511 (0.119)	Loss 0.621 (0.101)
Epoch: [35][40/200]	Time 0.568 (0.692)	Data 0.001 (0.099)	Loss 0.332 (0.279)
Epoch: [35][60/200]	Time 0.672 (0.679)	Data 0.001 (0.088)	Loss 0.452 (0.338)
Epoch: [35][80/200]	Time 0.570 (0.670)	Data 0.002 (0.083)	Loss 0.598 (0.370)
Epoch: [35][100/200]	Time 0.655 (0.667)	Data 0.001 (0.080)	Loss 0.266 (0.378)
Epoch: [35][120/200]	Time 0.585 (0.664)	Data 0.001 (0.078)	Loss 0.377 (0.389)
Epoch: [35][140/200]	Time 0.565 (0.661)	Data 0.001 (0.076)	Loss 0.360 (0.404)
Epoch: [35][160/200]	Time 0.567 (0.660)	Data 0.001 (0.074)	Loss 0.371 (0.407)
Epoch: [35][180/200]	Time 0.565 (0.659)	Data 0.001 (0.074)	Loss 0.560 (0.411)
Epoch: [35][200/200]	Time 0.658 (0.658)	Data 0.001 (0.073)	Loss 0.523 (0.414)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.201 (0.241)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.8279709815979
==> Statistics for epoch 36: 609 clusters
Epoch: [36][20/200]	Time 1.892 (0.702)	Data 1.247 (0.106)	Loss 0.288 (0.081)
Epoch: [36][40/200]	Time 0.546 (0.675)	Data 0.001 (0.087)	Loss 0.377 (0.263)
Epoch: [36][60/200]	Time 0.670 (0.667)	Data 0.001 (0.080)	Loss 0.446 (0.327)
Epoch: [36][80/200]	Time 0.567 (0.661)	Data 0.001 (0.076)	Loss 0.480 (0.350)
Epoch: [36][100/200]	Time 0.589 (0.659)	Data 0.001 (0.074)	Loss 0.512 (0.369)
Epoch: [36][120/200]	Time 0.582 (0.659)	Data 0.004 (0.073)	Loss 0.573 (0.388)
Epoch: [36][140/200]	Time 0.568 (0.658)	Data 0.001 (0.072)	Loss 0.449 (0.399)
Epoch: [36][160/200]	Time 0.565 (0.657)	Data 0.001 (0.071)	Loss 0.430 (0.404)
Epoch: [36][180/200]	Time 0.567 (0.656)	Data 0.001 (0.071)	Loss 0.495 (0.412)
Epoch: [36][200/200]	Time 0.571 (0.656)	Data 0.001 (0.071)	Loss 0.390 (0.420)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.200 (0.242)	Data 0.000 (0.022)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.79559898376465
==> Statistics for epoch 37: 607 clusters
Epoch: [37][20/200]	Time 0.572 (0.696)	Data 0.002 (0.117)	Loss 0.361 (0.111)
Epoch: [37][40/200]	Time 0.660 (0.679)	Data 0.002 (0.094)	Loss 0.559 (0.298)
Epoch: [37][60/200]	Time 0.552 (0.672)	Data 0.001 (0.089)	Loss 0.396 (0.351)
Epoch: [37][80/200]	Time 0.570 (0.669)	Data 0.001 (0.085)	Loss 0.279 (0.378)
Epoch: [37][100/200]	Time 0.562 (0.666)	Data 0.001 (0.084)	Loss 0.510 (0.397)
Epoch: [37][120/200]	Time 0.559 (0.665)	Data 0.000 (0.081)	Loss 0.566 (0.409)
Epoch: [37][140/200]	Time 0.542 (0.662)	Data 0.000 (0.079)	Loss 0.629 (0.412)
Epoch: [37][160/200]	Time 0.565 (0.662)	Data 0.000 (0.079)	Loss 0.672 (0.425)
Epoch: [37][180/200]	Time 0.649 (0.661)	Data 0.000 (0.078)	Loss 0.381 (0.423)
Epoch: [37][200/200]	Time 0.550 (0.669)	Data 0.001 (0.085)	Loss 0.471 (0.423)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.205 (0.247)	Data 0.000 (0.033)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.61056089401245
==> Statistics for epoch 38: 608 clusters
Epoch: [38][20/200]	Time 2.015 (0.699)	Data 1.382 (0.116)	Loss 0.431 (0.080)
Epoch: [38][40/200]	Time 0.564 (0.677)	Data 0.001 (0.096)	Loss 0.343 (0.257)
Epoch: [38][60/200]	Time 0.536 (0.667)	Data 0.001 (0.089)	Loss 0.446 (0.326)
Epoch: [38][80/200]	Time 0.578 (0.665)	Data 0.001 (0.086)	Loss 0.449 (0.358)
Epoch: [38][100/200]	Time 0.566 (0.662)	Data 0.001 (0.083)	Loss 0.483 (0.376)
Epoch: [38][120/200]	Time 0.649 (0.661)	Data 0.001 (0.081)	Loss 0.440 (0.393)
Epoch: [38][140/200]	Time 0.549 (0.661)	Data 0.002 (0.080)	Loss 0.404 (0.398)
Epoch: [38][160/200]	Time 0.657 (0.662)	Data 0.002 (0.079)	Loss 0.412 (0.407)
Epoch: [38][180/200]	Time 0.542 (0.661)	Data 0.002 (0.079)	Loss 0.403 (0.406)
Epoch: [38][200/200]	Time 0.562 (0.661)	Data 0.001 (0.079)	Loss 0.409 (0.411)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.217 (0.251)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.228002071380615
==> Statistics for epoch 39: 607 clusters
Epoch: [39][20/200]	Time 0.547 (0.695)	Data 0.001 (0.113)	Loss 0.445 (0.095)
Epoch: [39][40/200]	Time 0.654 (0.675)	Data 0.001 (0.091)	Loss 0.456 (0.239)
Epoch: [39][60/200]	Time 0.574 (0.672)	Data 0.001 (0.087)	Loss 0.534 (0.323)
Epoch: [39][80/200]	Time 0.555 (0.667)	Data 0.001 (0.082)	Loss 0.359 (0.358)
Epoch: [39][100/200]	Time 0.542 (0.665)	Data 0.001 (0.079)	Loss 0.531 (0.379)
Epoch: [39][120/200]	Time 0.566 (0.662)	Data 0.000 (0.076)	Loss 0.464 (0.389)
Epoch: [39][140/200]	Time 0.545 (0.662)	Data 0.000 (0.076)	Loss 0.420 (0.394)
Epoch: [39][160/200]	Time 0.564 (0.661)	Data 0.000 (0.075)	Loss 0.486 (0.400)
Epoch: [39][180/200]	Time 0.637 (0.659)	Data 0.000 (0.074)	Loss 0.322 (0.405)
Epoch: [39][200/200]	Time 0.550 (0.665)	Data 0.001 (0.080)	Loss 0.527 (0.409)
Extract Features: [50/76]	Time 0.213 (0.243)	Data 0.000 (0.024)	
Mean AP: 94.0%

 * Finished epoch  39  model mAP: 94.0%  best: 94.0%

==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.200 (0.242)	Data 0.000 (0.023)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.997610330581665
==> Statistics for epoch 40: 608 clusters
Epoch: [40][20/200]	Time 2.281 (0.705)	Data 1.509 (0.113)	Loss 0.406 (0.082)
Epoch: [40][40/200]	Time 0.544 (0.682)	Data 0.001 (0.096)	Loss 0.561 (0.262)
Epoch: [40][60/200]	Time 0.540 (0.674)	Data 0.001 (0.090)	Loss 0.440 (0.327)
Epoch: [40][80/200]	Time 0.560 (0.671)	Data 0.004 (0.087)	Loss 0.409 (0.349)
Epoch: [40][100/200]	Time 0.540 (0.668)	Data 0.001 (0.084)	Loss 0.334 (0.362)
Epoch: [40][120/200]	Time 0.564 (0.665)	Data 0.001 (0.081)	Loss 0.393 (0.381)
Epoch: [40][140/200]	Time 0.565 (0.663)	Data 0.002 (0.080)	Loss 0.313 (0.391)
Epoch: [40][160/200]	Time 0.562 (0.662)	Data 0.001 (0.079)	Loss 0.326 (0.395)
Epoch: [40][180/200]	Time 0.544 (0.661)	Data 0.001 (0.079)	Loss 0.373 (0.404)
Epoch: [40][200/200]	Time 0.655 (0.660)	Data 0.002 (0.078)	Loss 0.466 (0.409)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.188 (0.240)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.230844497680664
==> Statistics for epoch 41: 608 clusters
Epoch: [41][20/200]	Time 1.949 (0.696)	Data 1.314 (0.110)	Loss 0.564 (0.088)
Epoch: [41][40/200]	Time 0.575 (0.670)	Data 0.002 (0.088)	Loss 0.308 (0.261)
Epoch: [41][60/200]	Time 0.563 (0.666)	Data 0.001 (0.082)	Loss 0.365 (0.330)
Epoch: [41][80/200]	Time 0.563 (0.660)	Data 0.002 (0.078)	Loss 0.428 (0.354)
Epoch: [41][100/200]	Time 0.564 (0.656)	Data 0.001 (0.075)	Loss 0.309 (0.366)
Epoch: [41][120/200]	Time 0.544 (0.656)	Data 0.001 (0.074)	Loss 0.577 (0.388)
Epoch: [41][140/200]	Time 0.544 (0.656)	Data 0.001 (0.074)	Loss 0.302 (0.392)
Epoch: [41][160/200]	Time 0.570 (0.657)	Data 0.001 (0.074)	Loss 0.365 (0.403)
Epoch: [41][180/200]	Time 0.561 (0.658)	Data 0.001 (0.074)	Loss 0.385 (0.405)
Epoch: [41][200/200]	Time 0.553 (0.661)	Data 0.002 (0.076)	Loss 0.502 (0.406)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.187 (0.240)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.244791269302368
==> Statistics for epoch 42: 608 clusters
Epoch: [42][20/200]	Time 2.069 (0.699)	Data 1.451 (0.113)	Loss 0.494 (0.084)
Epoch: [42][40/200]	Time 0.700 (0.681)	Data 0.000 (0.094)	Loss 0.626 (0.264)
Epoch: [42][60/200]	Time 0.577 (0.671)	Data 0.000 (0.084)	Loss 0.516 (0.327)
Epoch: [42][80/200]	Time 0.544 (0.668)	Data 0.002 (0.084)	Loss 0.469 (0.354)
Epoch: [42][100/200]	Time 0.549 (0.664)	Data 0.001 (0.080)	Loss 0.348 (0.368)
Epoch: [42][120/200]	Time 0.550 (0.662)	Data 0.002 (0.078)	Loss 0.612 (0.377)
Epoch: [42][140/200]	Time 0.571 (0.660)	Data 0.002 (0.076)	Loss 0.404 (0.383)
Epoch: [42][160/200]	Time 0.572 (0.660)	Data 0.001 (0.076)	Loss 0.419 (0.392)
Epoch: [42][180/200]	Time 0.645 (0.659)	Data 0.002 (0.075)	Loss 0.421 (0.398)
Epoch: [42][200/200]	Time 0.559 (0.659)	Data 0.001 (0.076)	Loss 0.440 (0.402)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.299 (0.245)	Data 0.000 (0.023)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.469496726989746
==> Statistics for epoch 43: 607 clusters
Epoch: [43][20/200]	Time 0.548 (0.687)	Data 0.001 (0.107)	Loss 0.479 (0.110)
Epoch: [43][40/200]	Time 0.678 (0.667)	Data 0.002 (0.088)	Loss 0.516 (0.260)
Epoch: [43][60/200]	Time 0.568 (0.662)	Data 0.001 (0.081)	Loss 0.387 (0.318)
Epoch: [43][80/200]	Time 0.546 (0.657)	Data 0.001 (0.077)	Loss 0.335 (0.353)
Epoch: [43][100/200]	Time 0.570 (0.658)	Data 0.001 (0.075)	Loss 0.451 (0.367)
Epoch: [43][120/200]	Time 0.561 (0.656)	Data 0.000 (0.073)	Loss 0.393 (0.375)
Epoch: [43][140/200]	Time 0.644 (0.656)	Data 0.000 (0.073)	Loss 0.504 (0.386)
Epoch: [43][160/200]	Time 0.563 (0.654)	Data 0.000 (0.072)	Loss 0.219 (0.394)
Epoch: [43][180/200]	Time 0.551 (0.655)	Data 0.000 (0.073)	Loss 0.469 (0.400)
Epoch: [43][200/200]	Time 0.584 (0.661)	Data 0.001 (0.078)	Loss 0.481 (0.408)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.294 (0.244)	Data 0.000 (0.024)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.51201367378235
==> Statistics for epoch 44: 608 clusters
Epoch: [44][20/200]	Time 2.142 (0.705)	Data 1.552 (0.122)	Loss 0.508 (0.094)
Epoch: [44][40/200]	Time 0.536 (0.683)	Data 0.001 (0.102)	Loss 0.372 (0.272)
Epoch: [44][60/200]	Time 0.540 (0.671)	Data 0.001 (0.093)	Loss 0.408 (0.332)
Epoch: [44][80/200]	Time 0.686 (0.670)	Data 0.001 (0.089)	Loss 0.443 (0.359)
Epoch: [44][100/200]	Time 0.564 (0.667)	Data 0.001 (0.087)	Loss 0.432 (0.377)
Epoch: [44][120/200]	Time 0.567 (0.665)	Data 0.001 (0.085)	Loss 0.560 (0.387)
Epoch: [44][140/200]	Time 0.559 (0.664)	Data 0.001 (0.084)	Loss 0.386 (0.397)
Epoch: [44][160/200]	Time 0.553 (0.662)	Data 0.001 (0.083)	Loss 0.417 (0.401)
Epoch: [44][180/200]	Time 0.629 (0.663)	Data 0.001 (0.083)	Loss 0.335 (0.409)
Epoch: [44][200/200]	Time 0.569 (0.662)	Data 0.001 (0.082)	Loss 0.265 (0.412)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.197 (0.238)	Data 0.000 (0.026)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.559412002563477
==> Statistics for epoch 45: 607 clusters
Epoch: [45][20/200]	Time 0.559 (0.708)	Data 0.001 (0.132)	Loss 0.425 (0.123)
Epoch: [45][40/200]	Time 0.573 (0.681)	Data 0.002 (0.099)	Loss 0.451 (0.294)
Epoch: [45][60/200]	Time 0.548 (0.674)	Data 0.001 (0.093)	Loss 0.413 (0.334)
Epoch: [45][80/200]	Time 0.542 (0.670)	Data 0.001 (0.089)	Loss 0.507 (0.364)
Epoch: [45][100/200]	Time 0.571 (0.669)	Data 0.002 (0.089)	Loss 0.438 (0.381)
Epoch: [45][120/200]	Time 0.668 (0.668)	Data 0.000 (0.087)	Loss 0.323 (0.391)
Epoch: [45][140/200]	Time 0.558 (0.668)	Data 0.000 (0.086)	Loss 0.646 (0.395)
Epoch: [45][160/200]	Time 0.568 (0.665)	Data 0.000 (0.084)	Loss 0.426 (0.397)
Epoch: [45][180/200]	Time 0.561 (0.665)	Data 0.000 (0.084)	Loss 0.489 (0.404)
Epoch: [45][200/200]	Time 0.549 (0.671)	Data 0.001 (0.090)	Loss 0.623 (0.406)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.199 (0.251)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.659570693969727
==> Statistics for epoch 46: 608 clusters
Epoch: [46][20/200]	Time 2.126 (0.714)	Data 1.516 (0.121)	Loss 0.523 (0.091)
Epoch: [46][40/200]	Time 0.577 (0.688)	Data 0.001 (0.102)	Loss 0.436 (0.269)
Epoch: [46][60/200]	Time 0.674 (0.680)	Data 0.001 (0.092)	Loss 0.361 (0.327)
Epoch: [46][80/200]	Time 0.565 (0.674)	Data 0.001 (0.089)	Loss 0.538 (0.345)
Epoch: [46][100/200]	Time 0.561 (0.671)	Data 0.001 (0.086)	Loss 0.321 (0.365)
Epoch: [46][120/200]	Time 0.573 (0.670)	Data 0.001 (0.084)	Loss 0.391 (0.380)
Epoch: [46][140/200]	Time 0.566 (0.668)	Data 0.001 (0.083)	Loss 0.463 (0.390)
Epoch: [46][160/200]	Time 0.565 (0.667)	Data 0.001 (0.082)	Loss 0.307 (0.394)
Epoch: [46][180/200]	Time 0.576 (0.666)	Data 0.001 (0.082)	Loss 0.526 (0.404)
Epoch: [46][200/200]	Time 0.566 (0.665)	Data 0.001 (0.081)	Loss 0.316 (0.409)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.195 (0.246)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.05201029777527
==> Statistics for epoch 47: 608 clusters
Epoch: [47][20/200]	Time 2.095 (0.698)	Data 1.462 (0.114)	Loss 0.612 (0.088)
Epoch: [47][40/200]	Time 0.565 (0.674)	Data 0.001 (0.096)	Loss 0.289 (0.235)
Epoch: [47][60/200]	Time 0.668 (0.667)	Data 0.001 (0.089)	Loss 0.433 (0.298)
Epoch: [47][80/200]	Time 0.570 (0.668)	Data 0.002 (0.087)	Loss 0.488 (0.329)
Epoch: [47][100/200]	Time 0.733 (0.669)	Data 0.002 (0.085)	Loss 0.492 (0.358)
Epoch: [47][120/200]	Time 0.556 (0.665)	Data 0.001 (0.083)	Loss 0.469 (0.371)
Epoch: [47][140/200]	Time 0.575 (0.664)	Data 0.001 (0.082)	Loss 0.431 (0.380)
Epoch: [47][160/200]	Time 0.566 (0.663)	Data 0.001 (0.081)	Loss 0.518 (0.391)
Epoch: [47][180/200]	Time 0.566 (0.662)	Data 0.001 (0.081)	Loss 0.607 (0.393)
Epoch: [47][200/200]	Time 0.662 (0.662)	Data 0.001 (0.080)	Loss 0.373 (0.400)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.202 (0.247)	Data 0.000 (0.028)	
Computing jaccard distance...
Jaccard distance computing time cost: 20.01339554786682
==> Statistics for epoch 48: 608 clusters
Epoch: [48][20/200]	Time 2.102 (0.701)	Data 1.455 (0.117)	Loss 0.436 (0.083)
Epoch: [48][40/200]	Time 0.568 (0.678)	Data 0.001 (0.096)	Loss 0.302 (0.253)
Epoch: [48][60/200]	Time 0.566 (0.669)	Data 0.000 (0.089)	Loss 0.509 (0.306)
Epoch: [48][80/200]	Time 0.554 (0.664)	Data 0.001 (0.085)	Loss 0.446 (0.334)
Epoch: [48][100/200]	Time 0.537 (0.663)	Data 0.002 (0.084)	Loss 0.511 (0.346)
Epoch: [48][120/200]	Time 0.565 (0.660)	Data 0.001 (0.082)	Loss 0.442 (0.366)
Epoch: [48][140/200]	Time 0.651 (0.661)	Data 0.001 (0.081)	Loss 0.388 (0.376)
Epoch: [48][160/200]	Time 0.569 (0.659)	Data 0.001 (0.080)	Loss 0.619 (0.386)
Epoch: [48][180/200]	Time 0.632 (0.659)	Data 0.001 (0.079)	Loss 0.496 (0.395)
Epoch: [48][200/200]	Time 0.559 (0.659)	Data 0.001 (0.079)	Loss 0.465 (0.399)
==> Create pseudo labels for unlabeled data
Extract Features: [50/51]	Time 0.297 (0.249)	Data 0.000 (0.027)	
Computing jaccard distance...
Jaccard distance computing time cost: 19.603390216827393
==> Statistics for epoch 49: 607 clusters
Epoch: [49][20/200]	Time 0.687 (0.703)	Data 0.001 (0.117)	Loss 0.434 (0.090)
Epoch: [49][40/200]	Time 0.563 (0.684)	Data 0.001 (0.100)	Loss 0.377 (0.239)
Epoch: [49][60/200]	Time 0.669 (0.675)	Data 0.001 (0.093)	Loss 0.465 (0.302)
Epoch: [49][80/200]	Time 0.565 (0.671)	Data 0.001 (0.089)	Loss 0.316 (0.334)
Epoch: [49][100/200]	Time 0.567 (0.667)	Data 0.001 (0.087)	Loss 0.384 (0.349)
Epoch: [49][120/200]	Time 0.552 (0.665)	Data 0.000 (0.084)	Loss 0.401 (0.363)
Epoch: [49][140/200]	Time 0.537 (0.664)	Data 0.000 (0.083)	Loss 0.370 (0.375)
Epoch: [49][160/200]	Time 0.669 (0.665)	Data 0.000 (0.082)	Loss 0.405 (0.379)
Epoch: [49][180/200]	Time 0.560 (0.663)	Data 0.000 (0.081)	Loss 0.446 (0.387)
Epoch: [49][200/200]	Time 0.596 (0.671)	Data 0.001 (0.088)	Loss 0.475 (0.393)
Extract Features: [50/76]	Time 0.189 (0.252)	Data 0.000 (0.031)	
Mean AP: 94.0%

 * Finished epoch  49  model mAP: 94.0%  best: 94.0% *

==> Test with the best model:
=> Loaded checkpoint 'log/msmt2market/resnet152_ibn_cion/model_best.pth.tar'
Extract Features: [50/76]	Time 0.191 (0.243)	Data 0.000 (0.024)	
Mean AP: 94.0%
CMC Scores:
  top-1          97.1%
  top-5          98.9%
  top-10         99.5%
Total running time:  2:23:34.595233
