Question about OOM

Open jizaza opened this issue 1 year ago • 0 comments

Sorry to bother you again. When I use a single V100 (16384MiB) to train this model with patch size 389 and full size 1024, I find it can't work. And I also tried the methods you mentioned in Readme.md, it still has the error "out of memory". Do you have any more ideas about this? Thank you very much!!!!!

Oct 01 '24 06:10 jizaza