3 ms·
Agreed! Flux & other Julia Ml packages are awesome and have best in class API. Performance and memory usage aren’t yet on par with TF/PyTorch (or at least when
by tbenst 5y ago
Agreed! Flux & other Julia Ml packages are awesome and have best in class API. Performance and memory usage aren’t yet on par with TF/PyTorch (or at least when I last checked last year), but with more contributors and time I could see this closing and would love to use Julia for ML work
- amkkma 5y agoI believe we are currently at pytorch parity (and sometimes better) for speed. Memory usage depends....And this is without the extensive upcoming compiler improvements. The reason for the lag is that Julia has been focusing on general composable compiler, codegen and metaprogramming infrastructure which isn't domain specific, whereas pytorch and friends has been putting lots of dev money into c++ ML focused optimizers. Once the new compiler stuff is in place, it would be relatively trivial to write such optimizations, in user space, in pure Julia. Then exceeding that would be fairly simple also, plus things like static analysis of array shapes
- Royi 5y agoDo you have updated comparison with some data of performance and memory vs. PyTorch / TF?
- amkkma 5y agoI don't have a comprehensive suite of numbers at the moment (which is why I qualified it with "I believe"). My tentative conclusion is based on experience by a flux maintainer benchmarking forwards and backwards passes compared to pytorch, along with reports like this: https://discourse.julialang.org/t/flux-came-a-long-way/62288 https://discourse.julialang.org/t/flux-came-a-long-way/62288