The LGC in you code can not drop the number of parameters during processing.
Closed this issue · 2 comments
Hi markdtw,
Thank you for the share!
I have tested your code and found that there was no drop in the number of parameters.
I set the epoch=4, and the parameter numbers are 1.43M after all of the four epochs.
So, it seems that the LGC in the code worded just like standard CONV.
It confused me.
Hi!
The "pruning" process in this implementation is just masking out parameters with another tensor. The actual re-grouping requires an additional step to shuffle the weights based on the mask tensor (In the author's Pytorch implementation, this is done at the post-processing step as well). So the number of parameters is still the same unless you specifically exclude those "masked-out" parameters.
As stated in the README, this phase is not implemented yet. Any help will be appreciated :)
Thank you!