hmkx/HiNeRV

A problem about multi-GPU training

Opened this issue · 0 comments

When I use multi-GPU training, the result is lower than that of single one. What is the reason for this? does it have anything to do with accelerate?