mit-han-lab / once-for-all

[ICLR 2020] Once for All: Train One Network and Specialize it for Efficient Deployment
https://ofa.mit.edu/
MIT License
1.89k stars 333 forks source link

All memory on one card #68

Open Hukongtao opened 2 years ago

Hukongtao commented 2 years ago

When I train with two cards on my machine with horovodrun -np 2 -H localhost:2 python train_ofa_net.py, The code works fine. But I found that the memory of both processes is on the same card. Do you know why?

image