search

wenwei202 / caffe

Caffe for Sparse and Low-rank Deep Neural Networks
Other
376 stars 134 forks source link

I have some problem about code #24

Open zhangguotai opened 6 years ago

zhangguotai commented 6 years ago

In the code, what are their meaning that LOWERED_CSRMM,LOWERED_CCNMM and DIRECT_SCONV? In the process of making sparsification, whether you use existing library, example mkl on CPU and CUDA on GPU

wenwei202 commented 6 years ago

Here is the explanation. DIRECT_SCONV is deprecated in this branch and moved to https://github.com/IntelLabs/SkimCaffe. We used mkl sparse BLAS and CUDA cuSPARSE for sparse computation.

zhangguotai commented 6 years ago

thank you,I have new query that how to set the number of number? what is reason? which is divided into different group that columns or rows??

wenwei202 commented 6 years ago

I don't quite understand your questions. The way you set group/block sizes depends on what kinds of structured sparsity you want to learn. After learning, some blocks/groups will be removed. Hope this answers your questions.