Technical piece in Pete Warden's blog. Nicely visualized. I was unaware of the term GEMM, but taking a look at the post, he explains it well. Mathematically it is how you multiply matrices. A pretty common thing in applied mathematics. But now you have to do it more efficiently, under varying conditions of matrix density. In our own experiments with Neural networks, we quickly discovered this. The basics have been around for a long time, but here is an example where their application for updating neural networks radically changes the needs. Again, I warn you , this is math-technical.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment