Don't have Telegram yet? Try it now!
https://hgpu.org/?p=16730
How to scale distributed deep learning?