October 2019
Intermediate to advanced
444 pages
10h 37m
English
Deep learning in Rust works—however, it comes with many strings attached. tch-rs (https://github.com/LaurentMazare/tch-rs) is a great framework if you already know some PyTorch and it lets you get into it right away. However, anyone who is new to the idea of machine learning should look at Python (and PyTorch) to get comfortable with the type of thinking that is required. tch-rs uses the C++ foundation of the Python version and provides a thin wrapper around the bindings it created. This means two things:
By using bindings, wrapped code is much more likely to leave some kind of memory unfreed thanks to the added layer ...