ljleb/comfy-mecha

Can you help optimize it

Closed this issue · 0 comments

  I made a workflows myself to test the effect of each layer change in the model merge. However, halfway through the run, there was an error exit, indicating that the model cache had run out of memory. Can it be set to automatically delete the cache and release memory?

SDXL分层的各层测试(完成)new new.json
Thank you for your reply!