lxtGH/SFSegNets

Train on less than 8 GPUs

BabaVegato opened this issue · 1 comments

Hello, thanks for your detailed work.
I'd like to train your SFNet, but I don't have 8 GPUs available simultaineously in my lab, rather 2, 3, let's say 4. I'm a beginner in Apex, so I don't know if I can tweak the code to train on less than 8. Is it possible ?

Thanks.

lxtGH commented

You can train our model with 4 GPUs which can also achieve simular results.