2 ms·
I'm not seeing the connection. This work is about low-level optimization of matrix multiplication. The repo you linked seems to be about replacing back-propagat
by yorwba 3y ago
I'm not seeing the connection. This work is about low-level optimization of matrix multiplication. The repo you linked seems to be about replacing back-propagated gradients with a cheaper estimate. What's the similarity you see between these two?
- ixaxaar 3y agoCorrect, I think I mistook it as "use a small neural net to approximate matrix multiplication" instead it seems as "use cheaper replacements of matrix mul without much acc loss". Wellll that means I can give dni another try :D