Is using Rust bindings for the C++ api of PyTorch much faster than using Python?
Is using Rust bindings for the C++ api of PyTorch much faster than using Python?
github.com
GitHub - LaurentMazare/tch-rs: Rust bindings for the C++ api of PyTorch.
I tried to port a LSTM example in Python and the Rust version is much faster than the original. I wonder whether this is normal. 🤦‍