CUDA out of memory.

Open imperator-maximus opened this issue 3 years ago • 3 comments

Seems that 8GB is not supported yet:

RuntimeError: CUDA out of memory. Tried to allocate 1024.00 MiB (GPU 0; 8.00 GiB total capacity; 5.40 GiB already allocated; 0 bytes free; 6.53 GiB reserved in total by PyTorch)

Feb 17 '23 21:02 imperator-maximus

Have you found a fix for this? I'm getting the same issue.

Jul 19 '23 14:07 successor1

@imperator-maximus @successor1 Original study trained on 4-32GB Tesla V100 GPU.

Aug 15 '23 06:08 enkaranfiles

pipe.enable_model_cpu_offload() works well for CUDA 10G.

Jan 15 '24 10:01 delldu