Don't have Telegram yet? Try it now!
https://hgpu.org/?p=23722
Efficient Inference For Neural Machine Translation