CUDA error: out of memory

YudeWang / SEAM

Self-supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation, CVPR 2020 (Oral)

MIT License

539 stars 97 forks source link

CUDA error: out of memory #19

Closed Eli-YiLi closed 3 years ago

Eli-YiLi commented 3 years ago

I CANNOT RUN IT EVEN USE 8 GPUs (ONE IMG PER GPU)

Eli-YiLi commented 3 years ago

THE MATTER IS torch.nn.DataParallel

Eli-YiLi commented 3 years ago

--ntasks=1 IN SRUN

hyalvin commented 2 years ago

Hi, I have got the same problem when running the code with an 8*GPU(a100) server, the process just stuck on this line model = torch.nn.DataParallel(model).cuda(), so what should I do to solve this?