高通压缩框架-剪枝-channel_pruning_auto_mode-（vgg16log）

最新推荐文章于 2023-10-09 19:09:35 发布
最新推荐文章于 2023-10-09 19:09:35 发布
阅读量545 收藏
点赞数
文章标签： 高通 aimet
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/m0_37400316/article/details/111697600
版权 

2020-12-25 04:41:29,977 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.9937539794921877  Accuracy: 25.07%)

2020-12-25 04:41:35,673 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.100000 ==> eval_score=25.070000
2020-12-25 04:41:35,673 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:41:36,653 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.302294848632813  Accuracy: 25.97%)

2020-12-25 04:41:41,922 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.200000 ==> eval_score=25.970000
2020-12-25 04:41:41,922 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:41:42,861 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.427123193359375  Accuracy: 26.27%)

2020-12-25 04:41:48,936 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.300000 ==> eval_score=26.270000
2020-12-25 04:41:48,937 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:41:49,904 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.454738647460937  Accuracy: 26.18%)

2020-12-25 04:41:55,963 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.400000 ==> eval_score=26.180000
2020-12-25 04:41:55,963 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:41:57,020 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.466765979003906  Accuracy: 26.32%)

2020-12-25 04:42:02,494 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.500000 ==> eval_score=26.320000
2020-12-25 04:42:02,495 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:42:03,435 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.470900659179687  Accuracy: 26.27%)

2020-12-25 04:42:08,722 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.600000 ==> eval_score=26.270000
2020-12-25 04:42:08,722 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:42:09,747 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.472457165527344  Accuracy: 26.34%)

2020-12-25 04:42:15,073 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.700000 ==> eval_score=26.340000
2020-12-25 04:42:15,073 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:42:16,082 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.473871508789062  Accuracy: 26.32%)

2020-12-25 04:42:21,475 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.800000 ==> eval_score=26.320000
2020-12-25 04:42:21,475 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:42:22,467 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.473880346679688  Accuracy: 26.32%)

2020-12-25 04:42:28,504 - CompRatioSelect - INFO - Layer features.3, comp_ratio 0.900000 ==> eval_score=26.320000
2020-12-25 04:42:28,505 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:42:29,555 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -2.657516162109375  Accuracy: 19.46%)

2020-12-25 04:42:34,095 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.100000 ==> eval_score=19.460000
2020-12-25 04:42:34,096 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:42:35,157 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.526272692871094  Accuracy: 23.48%)

2020-12-25 04:42:39,936 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.200000 ==> eval_score=23.480000
2020-12-25 04:42:39,936 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:42:41,059 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.122827709960937  Accuracy: 25.35%)

2020-12-25 04:42:45,778 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.300000 ==> eval_score=25.350000
2020-12-25 04:42:45,779 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:42:46,774 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.26903017578125  Accuracy: 25.84%)

2020-12-25 04:42:51,579 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.400000 ==> eval_score=25.840000
2020-12-25 04:42:51,580 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:42:52,546 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.369065234375  Accuracy: 26.15%)

2020-12-25 04:42:57,428 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.500000 ==> eval_score=26.150000
2020-12-25 04:42:57,429 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:42:58,634 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4147089477539065  Accuracy: 26.15%)

2020-12-25 04:43:03,727 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.600000 ==> eval_score=26.150000
2020-12-25 04:43:03,727 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:43:04,952 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.438264562988281  Accuracy: 26.21%)

2020-12-25 04:43:09,909 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.700000 ==> eval_score=26.210000
2020-12-25 04:43:09,910 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:43:11,044 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.466131091308593  Accuracy: 26.22%)

2020-12-25 04:43:16,220 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.800000 ==> eval_score=26.220000
2020-12-25 04:43:16,221 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:43:17,553 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.47099775390625  Accuracy: 26.31%)

2020-12-25 04:43:22,763 - CompRatioSelect - INFO - Layer features.7, comp_ratio 0.900000 ==> eval_score=26.310000
2020-12-25 04:43:22,763 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:43:23,941 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.4297052734375  Accuracy: 23.56%)

2020-12-25 04:43:28,419 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.100000 ==> eval_score=23.560000
2020-12-25 04:43:28,419 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:43:29,609 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.108662158203125  Accuracy: 25.64%)

2020-12-25 04:43:34,491 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.200000 ==> eval_score=25.640000
2020-12-25 04:43:34,491 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:43:35,614 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.286140185546875  Accuracy: 25.89%)

2020-12-25 04:43:40,367 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.300000 ==> eval_score=25.890000
2020-12-25 04:43:40,367 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:43:41,732 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.365822741699219  Accuracy: 26.12%)

2020-12-25 04:43:47,099 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.400000 ==> eval_score=26.120000
2020-12-25 04:43:47,100 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:43:48,462 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.408419006347656  Accuracy: 26.18%)

2020-12-25 04:43:53,643 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.500000 ==> eval_score=26.180000
2020-12-25 04:43:53,643 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:43:54,974 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.429431652832031  Accuracy: 26.22%)

2020-12-25 04:44:00,308 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.600000 ==> eval_score=26.220000
2020-12-25 04:44:00,309 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:44:01,692 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.454393225097657  Accuracy: 26.31%)

2020-12-25 04:44:07,095 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.700000 ==> eval_score=26.310000
2020-12-25 04:44:07,096 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:44:08,546 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.462876599121094  Accuracy: 26.29%)

2020-12-25 04:44:13,948 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.800000 ==> eval_score=26.290000
2020-12-25 04:44:13,949 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:44:15,545 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.470846752929687  Accuracy: 26.33%)

2020-12-25 04:44:21,429 - CompRatioSelect - INFO - Layer features.10, comp_ratio 0.900000 ==> eval_score=26.330000
2020-12-25 04:44:21,429 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:44:22,698 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -2.8578366088867186  Accuracy: 21.33%)

2020-12-25 04:44:27,727 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.100000 ==> eval_score=21.330000
2020-12-25 04:44:27,728 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:44:29,007 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.730611340332031  Accuracy: 24.51%)

2020-12-25 04:44:34,132 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.200000 ==> eval_score=24.510000
2020-12-25 04:44:34,133 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:44:35,510 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.042912268066407  Accuracy: 25.55%)

2020-12-25 04:44:40,732 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.300000 ==> eval_score=25.550000
2020-12-25 04:44:40,733 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:44:42,175 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.2132582763671875  Accuracy: 25.87%)

2020-12-25 04:44:47,420 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.400000 ==> eval_score=25.870000
2020-12-25 04:44:47,421 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:44:48,806 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.3038884643554685  Accuracy: 26.14%)

2020-12-25 04:44:54,095 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.500000 ==> eval_score=26.140000
2020-12-25 04:44:54,096 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:44:55,891 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.367148352050782  Accuracy: 26.17%)

2020-12-25 04:45:01,514 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.600000 ==> eval_score=26.170000
2020-12-25 04:45:01,514 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:45:02,951 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.404570227050781  Accuracy: 26.24%)

2020-12-25 04:45:09,305 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.700000 ==> eval_score=26.240000
2020-12-25 04:45:09,306 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:45:10,867 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.431726611328125  Accuracy: 26.35%)

2020-12-25 04:45:17,471 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.800000 ==> eval_score=26.350000
2020-12-25 04:45:17,472 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:45:19,117 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4540637084960935  Accuracy: 26.33%)

2020-12-25 04:45:24,975 - CompRatioSelect - INFO - Layer features.14, comp_ratio 0.900000 ==> eval_score=26.330000
2020-12-25 04:45:24,976 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:45:26,298 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -2.967351574707031  Accuracy: 22.61%)

2020-12-25 04:45:32,856 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.100000 ==> eval_score=22.610000
2020-12-25 04:45:32,856 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:45:34,400 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.7882016967773438  Accuracy: 24.8%)

2020-12-25 04:45:40,898 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.200000 ==> eval_score=24.800000
2020-12-25 04:45:40,898 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:45:42,567 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.065039904785157  Accuracy: 25.55%)

2020-12-25 04:45:49,479 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.300000 ==> eval_score=25.550000
2020-12-25 04:45:49,479 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:45:51,120 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.136180004882813  Accuracy: 25.47%)

2020-12-25 04:45:57,999 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.400000 ==> eval_score=25.470000
2020-12-25 04:45:58,000 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:45:59,694 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.2271967041015626  Accuracy: 21.34%)

2020-12-25 04:46:06,459 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.500000 ==> eval_score=21.340000
2020-12-25 04:46:06,460 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:46:08,337 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.259498095703125  Accuracy: 25.84%)

2020-12-25 04:46:15,259 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.600000 ==> eval_score=25.840000
2020-12-25 04:46:15,260 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:46:17,288 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.359782446289063  Accuracy: 26.36%)

2020-12-25 04:46:24,120 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.700000 ==> eval_score=26.360000
2020-12-25 04:46:24,120 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:46:26,007 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.373932788085938  Accuracy: 26.28%)

2020-12-25 04:46:31,722 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.800000 ==> eval_score=26.280000
2020-12-25 04:46:31,723 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:46:34,312 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.378542797851562  Accuracy: 26.13%)

2020-12-25 04:46:40,127 - CompRatioSelect - INFO - Layer features.17, comp_ratio 0.900000 ==> eval_score=26.130000
2020-12-25 04:46:40,128 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:46:41,466 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -2.8355881103515626  Accuracy: 22.04%)

2020-12-25 04:46:46,626 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.100000 ==> eval_score=22.040000
2020-12-25 04:46:46,627 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:46:48,043 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.7431603637695314  Accuracy: 24.77%)

2020-12-25 04:46:54,376 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.200000 ==> eval_score=24.770000
2020-12-25 04:46:54,377 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:46:56,046 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.9623151123046876  Accuracy: 25.43%)

2020-12-25 04:47:02,576 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.300000 ==> eval_score=25.430000
2020-12-25 04:47:02,577 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:47:04,264 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.016892028808594  Accuracy: 25.36%)

2020-12-25 04:47:10,927 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.400000 ==> eval_score=25.360000
2020-12-25 04:47:10,928 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:47:12,832 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -2.9982592163085937  Accuracy: 21.62%)

2020-12-25 04:47:18,439 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.500000 ==> eval_score=21.620000
2020-12-25 04:47:18,439 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:47:20,550 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.1691917724609375  Accuracy: 25.68%)

2020-12-25 04:47:27,538 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.600000 ==> eval_score=25.680000
2020-12-25 04:47:27,538 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:47:29,580 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.348636962890625  Accuracy: 26.0%)

2020-12-25 04:47:36,471 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.700000 ==> eval_score=26.000000
2020-12-25 04:47:36,471 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:47:38,537 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.365845581054687  Accuracy: 26.11%)

2020-12-25 04:47:45,273 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.800000 ==> eval_score=26.110000
2020-12-25 04:47:45,274 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:47:47,869 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.377284094238282  Accuracy: 26.21%)

2020-12-25 04:47:54,944 - CompRatioSelect - INFO - Layer features.20, comp_ratio 0.900000 ==> eval_score=26.210000
2020-12-25 04:47:54,944 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:47:56,500 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.0340448608398436  Accuracy: 21.96%)

2020-12-25 04:48:03,271 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.100000 ==> eval_score=21.960000
2020-12-25 04:48:03,272 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:48:04,924 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.7629205200195313  Accuracy: 24.75%)

2020-12-25 04:48:11,443 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.200000 ==> eval_score=24.750000
2020-12-25 04:48:11,444 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:48:13,242 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.9391884521484375  Accuracy: 25.29%)

2020-12-25 04:48:19,852 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.300000 ==> eval_score=25.290000
2020-12-25 04:48:19,853 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:48:21,681 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.3705059326171876  Accuracy: 24.66%)

2020-12-25 04:48:28,411 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.400000 ==> eval_score=24.660000
2020-12-25 04:48:28,412 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:48:30,648 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.092959033203125  Accuracy: 25.74%)

2020-12-25 04:48:37,312 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.500000 ==> eval_score=25.740000
2020-12-25 04:48:37,312 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:48:39,172 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.28777431640625  Accuracy: 26.1%)

2020-12-25 04:48:44,849 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.600000 ==> eval_score=26.100000
2020-12-25 04:48:44,850 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:48:46,773 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.3485037719726565  Accuracy: 26.15%)

2020-12-25 04:48:52,562 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.700000 ==> eval_score=26.150000
2020-12-25 04:48:52,563 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:48:54,458 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.346389123535157  Accuracy: 26.23%)

2020-12-25 04:49:00,556 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.800000 ==> eval_score=26.230000
2020-12-25 04:49:00,557 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:49:03,036 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.352585900878906  Accuracy: 26.15%)

2020-12-25 04:49:09,009 - CompRatioSelect - INFO - Layer features.24, comp_ratio 0.900000 ==> eval_score=26.150000
2020-12-25 04:49:09,010 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:49:10,548 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.7801083251953127  Accuracy: 25.25%)

2020-12-25 04:49:17,052 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.100000 ==> eval_score=25.250000
2020-12-25 04:49:17,053 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:49:19,019 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -3.7332941650390623  Accuracy: 25.15%)

2020-12-25 04:49:25,730 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.200000 ==> eval_score=25.150000
2020-12-25 04:49:25,730 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:49:27,792 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.2871577880859375  Accuracy: 26.15%)

2020-12-25 04:49:33,562 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.300000 ==> eval_score=26.150000
2020-12-25 04:49:33,562 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:49:35,739 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.318684301757813  Accuracy: 26.15%)

2020-12-25 04:49:41,820 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.400000 ==> eval_score=26.150000
2020-12-25 04:49:41,820 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:49:44,487 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.331072131347656  Accuracy: 26.23%)

2020-12-25 04:49:51,106 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.500000 ==> eval_score=26.230000
2020-12-25 04:49:51,107 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:49:53,532 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.3412935791015625  Accuracy: 26.06%)

2020-12-25 04:50:00,514 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.600000 ==> eval_score=26.060000
2020-12-25 04:50:00,514 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:50:02,973 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.335660168457031  Accuracy: 26.12%)

2020-12-25 04:50:09,331 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.700000 ==> eval_score=26.120000
2020-12-25 04:50:09,331 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:50:11,725 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.341002026367187  Accuracy: 26.2%)

2020-12-25 04:50:18,641 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.800000 ==> eval_score=26.200000
2020-12-25 04:50:18,641 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:50:21,077 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.321689184570313  Accuracy: 26.14%)

2020-12-25 04:50:28,207 - CompRatioSelect - INFO - Layer features.27, comp_ratio 0.900000 ==> eval_score=26.140000
2020-12-25 04:50:28,208 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:50:30,146 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.106214758300781  Accuracy: 26.06%)

2020-12-25 04:50:36,570 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.100000 ==> eval_score=26.060000
2020-12-25 04:50:36,570 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:50:38,711 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.141429211425781  Accuracy: 26.05%)

2020-12-25 04:50:45,328 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.200000 ==> eval_score=26.050000
2020-12-25 04:50:45,329 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:50:47,553 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.3922559204101566  Accuracy: 26.28%)

2020-12-25 04:50:54,127 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.300000 ==> eval_score=26.280000
2020-12-25 04:50:54,128 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:50:56,457 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.41832626953125  Accuracy: 26.31%)

2020-12-25 04:51:03,088 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.400000 ==> eval_score=26.310000
2020-12-25 04:51:03,089 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:51:05,738 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.41874228515625  Accuracy: 26.29%)

2020-12-25 04:51:12,448 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.500000 ==> eval_score=26.290000
2020-12-25 04:51:12,449 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:51:14,971 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.428370166015625  Accuracy: 26.34%)

2020-12-25 04:51:20,535 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.600000 ==> eval_score=26.340000
2020-12-25 04:51:20,535 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:51:23,104 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.425316186523437  Accuracy: 26.32%)

2020-12-25 04:51:28,800 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.700000 ==> eval_score=26.320000
2020-12-25 04:51:28,801 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:51:31,313 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.424797448730469  Accuracy: 26.33%)

2020-12-25 04:51:38,298 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.800000 ==> eval_score=26.330000
2020-12-25 04:51:38,298 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:51:40,846 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.425589099121094  Accuracy: 26.3%)

2020-12-25 04:51:47,909 - CompRatioSelect - INFO - Layer features.30, comp_ratio 0.900000 ==> eval_score=26.300000
2020-12-25 04:51:47,909 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:51:49,787 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.202631921386719  Accuracy: 26.16%)

2020-12-25 04:51:55,078 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.100000 ==> eval_score=26.160000
2020-12-25 04:51:55,078 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:51:56,859 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.437729772949218  Accuracy: 26.33%)

2020-12-25 04:52:03,866 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.200000 ==> eval_score=26.330000
2020-12-25 04:52:03,866 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:52:05,984 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.447308581542969  Accuracy: 26.28%)

2020-12-25 04:52:12,854 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.300000 ==> eval_score=26.280000
2020-12-25 04:52:12,855 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:52:14,919 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.448629248046875  Accuracy: 26.34%)

2020-12-25 04:52:21,235 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.400000 ==> eval_score=26.340000
2020-12-25 04:52:21,235 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:52:23,858 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.453668115234375  Accuracy: 26.35%)

2020-12-25 04:52:30,910 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.500000 ==> eval_score=26.350000
2020-12-25 04:52:30,910 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:52:33,562 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.459936291503906  Accuracy: 26.32%)

2020-12-25 04:52:39,488 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.600000 ==> eval_score=26.320000
2020-12-25 04:52:39,489 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:52:41,953 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.454297351074219  Accuracy: 26.32%)

2020-12-25 04:52:47,902 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.700000 ==> eval_score=26.320000
2020-12-25 04:52:47,903 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:52:50,178 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4582498779296875  Accuracy: 26.34%)

2020-12-25 04:52:56,230 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.800000 ==> eval_score=26.340000
2020-12-25 04:52:56,231 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:52:58,577 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.457487329101562  Accuracy: 26.32%)

2020-12-25 04:53:05,134 - CompRatioSelect - INFO - Layer features.34, comp_ratio 0.900000 ==> eval_score=26.320000
2020-12-25 04:53:05,135 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:53:07,051 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4527604858398435  Accuracy: 26.35%)

2020-12-25 04:53:13,869 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.100000 ==> eval_score=26.350000
2020-12-25 04:53:13,870 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:53:15,929 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.473125329589844  Accuracy: 26.3%)

2020-12-25 04:53:22,774 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.200000 ==> eval_score=26.300000
2020-12-25 04:53:22,775 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:53:24,962 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.470657238769531  Accuracy: 26.33%)

2020-12-25 04:53:32,119 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.300000 ==> eval_score=26.330000
2020-12-25 04:53:32,120 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:53:34,558 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471886071777344  Accuracy: 26.33%)

2020-12-25 04:53:41,363 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.400000 ==> eval_score=26.330000
2020-12-25 04:53:41,363 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:53:44,200 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4731350463867185  Accuracy: 26.35%)

2020-12-25 04:53:51,487 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.500000 ==> eval_score=26.350000
2020-12-25 04:53:51,488 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:53:54,195 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.472640344238282  Accuracy: 26.35%)

2020-12-25 04:54:01,592 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.600000 ==> eval_score=26.350000
2020-12-25 04:54:01,593 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:54:03,982 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471326416015625  Accuracy: 26.34%)

2020-12-25 04:54:11,214 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.700000 ==> eval_score=26.340000
2020-12-25 04:54:11,214 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:54:13,659 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.470440930175781  Accuracy: 26.32%)

2020-12-25 04:54:19,516 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.800000 ==> eval_score=26.320000
2020-12-25 04:54:19,517 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:54:21,905 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.472367053222656  Accuracy: 26.34%)

2020-12-25 04:54:29,256 - CompRatioSelect - INFO - Layer features.37, comp_ratio 0.900000 ==> eval_score=26.340000
2020-12-25 04:54:29,257 - CompRatioSelect - INFO - Analyzing compression ratio: 0.1 =====================>
2020-12-25 04:54:31,205 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.45629208984375  Accuracy: 26.36%)

2020-12-25 04:54:38,147 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.100000 ==> eval_score=26.360000
2020-12-25 04:54:38,147 - CompRatioSelect - INFO - Analyzing compression ratio: 0.2 =====================>
2020-12-25 04:54:40,219 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.468901538085937  Accuracy: 26.34%)

2020-12-25 04:54:47,188 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.200000 ==> eval_score=26.340000
2020-12-25 04:54:47,188 - CompRatioSelect - INFO - Analyzing compression ratio: 0.3 =====================>
2020-12-25 04:54:49,283 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471708813476562  Accuracy: 26.34%)

2020-12-25 04:54:56,266 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.300000 ==> eval_score=26.340000
2020-12-25 04:54:56,267 - CompRatioSelect - INFO - Analyzing compression ratio: 0.4 =====================>
2020-12-25 04:54:58,486 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.4703867065429685  Accuracy: 26.32%)

2020-12-25 04:55:05,295 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.400000 ==> eval_score=26.320000
2020-12-25 04:55:05,295 - CompRatioSelect - INFO - Analyzing compression ratio: 0.5 =====================>
2020-12-25 04:55:08,158 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.470028295898437  Accuracy: 26.33%)

2020-12-25 04:55:15,340 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.500000 ==> eval_score=26.330000
2020-12-25 04:55:15,341 - CompRatioSelect - INFO - Analyzing compression ratio: 0.6 =====================>
2020-12-25 04:55:18,115 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471591943359375  Accuracy: 26.33%)

2020-12-25 04:55:25,419 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.600000 ==> eval_score=26.330000
2020-12-25 04:55:25,419 - CompRatioSelect - INFO - Analyzing compression ratio: 0.7 =====================>
2020-12-25 04:55:27,797 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.472099169921875  Accuracy: 26.32%)

2020-12-25 04:55:33,704 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.700000 ==> eval_score=26.320000
2020-12-25 04:55:33,704 - CompRatioSelect - INFO - Analyzing compression ratio: 0.8 =====================>
2020-12-25 04:55:36,037 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471060083007813  Accuracy: 26.33%)

2020-12-25 04:55:42,865 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.800000 ==> eval_score=26.330000
2020-12-25 04:55:42,865 - CompRatioSelect - INFO - Analyzing compression ratio: 0.9 =====================>
2020-12-25 04:55:45,208 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.471531066894531  Accuracy: 26.33%)

2020-12-25 04:55:52,249 - CompRatioSelect - INFO - Layer features.40, comp_ratio 0.900000 ==> eval_score=26.330000
2020-12-25 04:55:52,250 - CompRatioSelect - INFO - Greedy selection: Saved eval dict to ./data/greedy_selection_eval_scores_dict.pkl
2020-12-25 04:55:52,255 - CompRatioSelect - INFO - Greedy selection: overall_min_score=19.460000, overall_max_score=26.360000
2020-12-25 04:55:52,255 - CompRatioSelect - INFO - Greedy selection: Original model cost=(Cost: memory=14715584, mac=313201664)
2020-12-25 04:56:42,651 - CompRatioSelect - INFO - Greedy selection: final choice - comp_ratio=0.749572, score=26.319991
2020-12-25 04:56:44,120 - ChannelPruning - INFO - finished linear regression fit 
2020-12-25 04:56:45,965 - ChannelPruning - INFO - finished linear regression fit 
2020-12-25 04:56:47,749 - ChannelPruning - INFO - finished linear regression fit 
2020-12-25 04:56:50,293 - ChannelPruning - INFO - finished linear regression fit 
2020-12-25 04:56:52,415 - ChannelPruning - INFO - finished linear regression fit 
2020-12-25 04:56:54,523 - ChannelPruning - INFO - finished linear regression fit 
--------------------testing----------------------------
Computing :Loss: -4.47374072265625  Accuracy: 26.32%)

--------------------testing----------------------------
Computing :Loss: -3.9216089111328123  Accuracy: 26.15%)

VGG01(
  (features): Sequential(
    (0): Conv2d(3, 44, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (1): BatchNorm2d(44, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (2): ReLU(inplace=True)
    (3): Conv2d(44, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (5): ReLU(inplace=True)
    (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (7): Conv2d(64, 115, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (8): BatchNorm2d(115, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (9): ReLU(inplace=True)
    (10): Conv2d(115, 102, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): BatchNorm2d(102, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (12): ReLU(inplace=True)
    (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (14): Conv2d(102, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (15): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (16): ReLU(inplace=True)
    (17): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (18): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (19): ReLU(inplace=True)
    (20): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (21): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (22): ReLU(inplace=True)
    (23): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (24): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (25): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (26): ReLU(inplace=True)
    (27): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (28): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (29): ReLU(inplace=True)
    (30): Conv2d(512, 204, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (31): BatchNorm2d(204, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (32): ReLU(inplace=True)
    (33): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (34): Conv2d(204, 51, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (35): BatchNorm2d(51, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (36): ReLU(inplace=True)
    (37): Conv2d(51, 51, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (38): BatchNorm2d(51, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (39): ReLU(inplace=True)
    (40): Conv2d(51, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (41): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (42): ReLU(inplace=True)
    (43): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (44): AvgPool2d(kernel_size=1, stride=1, padding=0)
  )
  (classifier): Linear(in_features=512, out_features=10, bias=True)
)
**********************************************************************************************
Compressed Model Statistics
Baseline model accuracy: 26.320000, Compressed model accuracy: 26.150000
Compression ratio for memory=0.438253, mac=0.749572

**********************************************************************************************

Per-layer Stats
    Name:features.3, compression-ratio: 0.7
    Name:features.7, compression-ratio: None
    Name:features.10, compression-ratio: 0.9
    Name:features.14, compression-ratio: 0.8
    Name:features.17, compression-ratio: None
    Name:features.20, compression-ratio: None
    Name:features.24, compression-ratio: None
    Name:features.27, compression-ratio: None
    Name:features.30, compression-ratio: None
    Name:features.34, compression-ratio: 0.4
    Name:features.37, compression-ratio: 0.1
    Name:features.40, compression-ratio: 0.1

**********************************************************************************************

Greedy Eval Dict
    Layer: features.3
        Ratio=0.1, Eval score=25.07
        Ratio=0.2, Eval score=25.97
        Ratio=0.3, Eval score=26.27
        Ratio=0.4, Eval score=26.18
        Ratio=0.5, Eval score=26.32
        Ratio=0.6, Eval score=26.27
        Ratio=0.7, Eval score=26.34
        Ratio=0.8, Eval score=26.32
        Ratio=0.9, Eval score=26.32
    Layer: features.7
        Ratio=0.1, Eval score=19.46
        Ratio=0.2, Eval score=23.48
        Ratio=0.3, Eval score=25.35
        Ratio=0.4, Eval score=25.84
        Ratio=0.5, Eval score=26.15
        Ratio=0.6, Eval score=26.15
        Ratio=0.7, Eval score=26.21
        Ratio=0.8, Eval score=26.22
        Ratio=0.9, Eval score=26.31
    Layer: features.10
        Ratio=0.1, Eval score=23.56
        Ratio=0.2, Eval score=25.64
        Ratio=0.3, Eval score=25.89
        Ratio=0.4, Eval score=26.12
        Ratio=0.5, Eval score=26.18
        Ratio=0.6, Eval score=26.22
        Ratio=0.7, Eval score=26.31
        Ratio=0.8, Eval score=26.29
        Ratio=0.9, Eval score=26.33
    Layer: features.14
        Ratio=0.1, Eval score=21.33
        Ratio=0.2, Eval score=24.51
        Ratio=0.3, Eval score=25.55
        Ratio=0.4, Eval score=25.87
        Ratio=0.5, Eval score=26.14
        Ratio=0.6, Eval score=26.17
        Ratio=0.7, Eval score=26.24
        Ratio=0.8, Eval score=26.35
        Ratio=0.9, Eval score=26.33
    Layer: features.17
        Ratio=0.1, Eval score=22.61
        Ratio=0.2, Eval score=24.8
        Ratio=0.3, Eval score=25.55
        Ratio=0.4, Eval score=25.47
        Ratio=0.5, Eval score=21.34
        Ratio=0.6, Eval score=25.84
        Ratio=0.7, Eval score=26.36
        Ratio=0.8, Eval score=26.28
        Ratio=0.9, Eval score=26.13
    Layer: features.20
        Ratio=0.1, Eval score=22.04
        Ratio=0.2, Eval score=24.77
        Ratio=0.3, Eval score=25.43
        Ratio=0.4, Eval score=25.36
        Ratio=0.5, Eval score=21.62
        Ratio=0.6, Eval score=25.68
        Ratio=0.7, Eval score=26.0
        Ratio=0.8, Eval score=26.11
        Ratio=0.9, Eval score=26.21
    Layer: features.24
        Ratio=0.1, Eval score=21.96
        Ratio=0.2, Eval score=24.75
        Ratio=0.3, Eval score=25.29
        Ratio=0.4, Eval score=24.66
        Ratio=0.5, Eval score=25.74
        Ratio=0.6, Eval score=26.1
        Ratio=0.7, Eval score=26.15
        Ratio=0.8, Eval score=26.23
        Ratio=0.9, Eval score=26.15
    Layer: features.27
        Ratio=0.1, Eval score=25.25
        Ratio=0.2, Eval score=25.15
        Ratio=0.3, Eval score=26.15
        Ratio=0.4, Eval score=26.15
        Ratio=0.5, Eval score=26.23
        Ratio=0.6, Eval score=26.06
        Ratio=0.7, Eval score=26.12
        Ratio=0.8, Eval score=26.2
        Ratio=0.9, Eval score=26.14
    Layer: features.30
        Ratio=0.1, Eval score=26.06
        Ratio=0.2, Eval score=26.05
        Ratio=0.3, Eval score=26.28
        Ratio=0.4, Eval score=26.31
        Ratio=0.5, Eval score=26.29
        Ratio=0.6, Eval score=26.34
        Ratio=0.7, Eval score=26.32
        Ratio=0.8, Eval score=26.33
        Ratio=0.9, Eval score=26.3
    Layer: features.34
        Ratio=0.1, Eval score=26.16
        Ratio=0.2, Eval score=26.33
        Ratio=0.3, Eval score=26.28
        Ratio=0.4, Eval score=26.34
        Ratio=0.5, Eval score=26.35
        Ratio=0.6, Eval score=26.32
        Ratio=0.7, Eval score=26.32
        Ratio=0.8, Eval score=26.34
        Ratio=0.9, Eval score=26.32
    Layer: features.37
        Ratio=0.1, Eval score=26.35
        Ratio=0.2, Eval score=26.3
        Ratio=0.3, Eval score=26.33
        Ratio=0.4, Eval score=26.33
        Ratio=0.5, Eval score=26.35
        Ratio=0.6, Eval score=26.35
        Ratio=0.7, Eval score=26.34
        Ratio=0.8, Eval score=26.32
        Ratio=0.9, Eval score=26.34
    Layer: features.40
        Ratio=0.1, Eval score=26.36
        Ratio=0.2, Eval score=26.34
        Ratio=0.3, Eval score=26.34
        Ratio=0.4, Eval score=26.32
        Ratio=0.5, Eval score=26.33
        Ratio=0.6, Eval score=26.33
        Ratio=0.7, Eval score=26.32
        Ratio=0.8, Eval score=26.33
        Ratio=0.9, Eval score=26.33

**********************************************************************************************


Process finished with exit code 0
CV/NLP大虾
关注
  • 0
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 2
    评论
Channel Pruning：用于加速深度神经网络的通道剪枝方法
librahfacebook的博客
07-27 5250
我们介绍了一种新的通道剪枝方法，用来加速非常深的卷积神经网络。给定一个训练过的CNN模型，我们提出来一种迭代式的两步骤算法，通过基于通道选择的LASSO回归和最小二乘法重建来有效地修剪每一层。我们进一步将这个算法扩展到多层、多分支的情况。该方法减少了积累误差，提高了与各种体系架构的兼容性。
高通压缩框架-剪枝-SVD_auto_mode-（vgg16log）
m0_37400316的博客
12-26 395
a72ac4f2f51e:/usr/local/bin/python -u /opt/project/torch_model.py 2020-12-25 05:33:13,332 - root - INFO - AIMET WARNING:param.Dimension: Use method 'get_param_values' via param namespace WARNING:param.Dimension: Use method 'get_param_values' via param nam
通道剪枝Channel Pruning
年轻即出发，
10-12 8242
你感觉不舒服的时候，就是成长的时候…… 论文：Channel Pruning for Accelerating Very Deep Neural Networks Github: https://github.com/yihui-he/channel-pruning ICCV2017 旷视科技 论文提出了新的通道剪枝方法，基于2步迭代的剪枝方法。首先基于LASSO回...
模型通道剪枝汇总（channel pruning）
qq_38109843的博客
07-20 3746
目前所看到的通道剪枝的方法不是特别多，先总结一下现有的 论文：https://openaccess.thecvf.com/content_iccv_2017/html/Liu_Learning_Efficient_Convolutional_ICCV_2017_paper.html 解读：https://blog.csdn.net/h__ang/article/details/89376079 关于这篇论文和相关的yolov3项目，我做了一个总结，传送门在这里,有完整代码，复现很方便：https://bl
Channel Pruning for Accelerating Very Deep Neural Networks
jwy2014的专栏
12-05 412
https://github.com/yihui-he/channel-pruning   ICCV 2017, by Yihui He, Xiangyu Zhang and Jian Sun Please have a look at AMC: AutoML for Model Compression and Acceleration on Mobile Devices ECCV'18, ...
yiziqi.rar_α-?剪枝_一字棋_棋 人机对战
09-21
剪枝在一字棋人机对战中的应用详解》 一字棋，又称井字游戏或Noughts and Crosses，是一种简单却深具策略性的两人对弈游戏。在计算机科学领域，尤其是人工智能（AI）研究中，一字棋常常被用作教学和实验的工具，...
8_num.rar_8-puzzle_puzzle 8_人工智能
09-24
压缩包子文件的文件名称“8_num”可能指的是用于存储八数码问题实例的数据或者程序的主入口。 现在我们深入探讨A*算法的细节： 1. **A*算法的基本步骤：** - 初始化：设置初始节点，通常是问题的初始状态，计算其...
DT.rar_decision tree_dt_dt-cwt_决策树_决策树dt
07-15
5. **剪枝**：为了防止过拟合，可能需要对生成的树进行简化，例如通过预剪枝或后剪枝。 6. **模型评估**：使用交叉验证、准确率、召回率、F1分数等指标评估模型性能。 在"DT"文件中，很可能包含了实现这些步骤的...
Torch-Pruning:pytorch修剪工具包，用于结构化神经网络修剪和自动层依赖维护
04-30
当通过DependencyGraph.get_pruning_plan将修剪功能（例如torch_pruning.prune_conv）应用于特定层时，此程序包将遍历整个图以修复不一致的模块（例如BN）。 如果模型中有torch.split或torch.cat ，则修剪索引将...
java--fivechessgame.rar_FiveChessGame_Java fivechessgame_五子棋java
09-24
为了增加趣味性，还可以设计AI（人工智能）对战模式，利用简单的搜索算法如Minimax或Alpha-Beta剪枝，让计算机自动决策棋步，与玩家进行对弈。 此外，为了让游戏更适应女性玩家，开发者可能关注了界面的美观性和...
模型压缩之通道剪枝channel-pruning-master.zip
08-01
对训练好的模型进行通道剪枝（channel pruning），分为两步进行：第一步是channel selection，采用LASSO regression来做，其实就是添加了一个L1范数来约束权重，因为L1范数可以使得权重更加稀疏，这样就可以把那些稀疏的channel剪掉；第二步是reconstruction，基于linear least squares来约束剪枝后输出的feature map要尽可能和减枝前的输出feature map相等，也就是最小二乘值越小越好。
pytorch-pruning 网络剪枝数据集的训练集2
04-21
https://download.csdn.net/download/weixin_32759777/12346888这个代码的训练集2
论文阅读：Channel Pruning for Accelerating Very Deep Neural Networks
Crayonshin的博客
09-22 226
论文链接 代码链接(caffe) Motivation & Abstract 作者指出当前的CNN加速工作分为三种： optimized implementation （直接优化？） 量化（这个还没接触过，不过以后一定会钻研） structured simplification 结构简化 作者的方法就属于第三种结构简化，这种方法又可以细分为三种，如下图： 图c是张量分解,即将...
通道剪枝channel pruning
最新发布
qq_45655136的博客
10-09 1213
论文：Channel Pruning for Accelerating Very Deep Neural Networks论文链接：https://arxiv.org/abs/1707.06168代码地址：https://github.com/yihui-he/channel-pruning这是一篇ICCV2017的文章，关于用通道剪枝channel pruning来做模型加速，通道剪枝是模型压缩和加速领域的一个重要分支。
Network Slimming——有效的通道剪枝方法（Channel Pruning）
热门推荐
AI Flash
06-11 1万+
"Learning Efficient Convolutional Networks through Network Slimming"这篇文章提出了一种有效的结构性剪枝方法，即规整的通道剪枝策略：在训练期间，通过对网络BN层的gamma系数施加L1正则约束，使得模型朝着结构性稀疏的方向调整参数。此时BN层的Gamma系数所起的作用，类似于信息流通道的开关系数，控制着信息流通道的开关闭合。完成稀疏...
【蒸馏+剪枝】（一）Cascaded channel pruning using hierarchical self-distillation
梁瑛平的博客
08-19 611
本文提出了一种基于教师模型、助教模型和学生模型的层次知识蒸馏的过滤级剪枝方法，助教模型与目标学生共享相同的架构和权重。作者建议使用来自相应教师模型的梯度信息独立地修剪每个模型，通过考虑每个师生模型对的相对大小，该公式提供了知识蒸馏的能力差距和滤波器显著性更新的偏差之间的自然权衡
《DMCP：Differentiable Markov Channel Pruning for Neural Networks》论文笔记
m_buddy的博客
01-03 489
参考代码：dmcp 1. 概述 导读：在网络剪枝领域中已经有一些工作将结构搜索的概念引入到剪枝方法中，如AMC使用强化学习的方式使控制器输出每一层的裁剪比例。但是正如NAS的发展过程一样，这些基于强化学习的搜索方法需要大量的训练和验证过程，而最直接的便是使用类似DARTS的直接梯度优化方法。而这篇文章中将网络的channel剪枝建模为一个可微分的马尔可夫过程（Differentiable Markov Channel Pruning，DMCP），通过直接产生的梯度信息进行优化，因而更加高效，整个剪枝的过程
论文解读《Channel pruning based on convolutional neural network sensitivity》
qq_46031969的博客
01-01 635
论文解读《Channel pruning based on convolutional neural network sensitivity》
Residual Squeeze VGG16
qq_35800608的博客
05-31 1205
本人在学习深度压缩方面的知识，刚开始多久，自己建了一个QQ群希望能和大家一起讨论学习深度压缩加速的知识。 QQ群：621888628 1.本文的贡献： 使用SQUEEZENET改进的Fire Module即（ResSquVGG16）来压缩VGG16模型，在MIT Places365标准场景数据集上训练。在精度变化不大的情况下将训练时间减少了23.86％，模型大小缩小了88.4％。   2
Discrimination-aware channel pruning论文解读
liguiyuan的博客
05-11 927
1. 算法介绍 paper：https://arxiv.org/abs/1810.11809 Discrimination-aware channel pruning (DCP, Zhuang et al., 2018)鉴别力感知的通道剪枝。这个称为鉴别力感知，就是在卷积层通道中找出那些真正具有贡献、有鉴别感知力的通道，然后保留下来，而那些对没有鉴别力（往往指那些通道只包含全零权重）的通道就剪...

“相关推荐”对你有帮助么？

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论 2
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值