search

Eclipsess / CHIP_NeurIPS2021

Code for CHIP: CHannel Independence-based Pruning for Compact Neural Networks (NeruIPS 2021).
33 stars 6 forks source link

Hello, how do you calculate your Params and Flops? #12

Closed Zzz-zcy closed 1 year ago

Zzz-zcy commented 2 years ago

Hello, how do you calculate your Params and Flops?

Eclipsess commented 2 years ago

By thop

Zzz-zcy commented 2 years ago

您好，还有关于特征图conv_feature_map_tensor().npy里面的数据是四维的，比如conv_feature_map_tensor(1).npy是(128, 16, 32, 32)，其中32×32是特征图大小，16是通道数目（也就是16张特征图），128代表一个batch_size中128张不同的输入。conv_feature_map_tensor().npy有275个数据文件，是因为repeat=5，取了5次batch_size为128的数据得出的四维特征图张量，所以得到（resnet56有55层的输出）55层*repeat=275个文件，我这样理解的对么？

Eclipsess commented 1 year ago

exactly right!