lllyasviel
Patching Lora
Is it possible to patch the Lora once instead of at each subsequent generation?
I think the point is many features are being patched in as they come, to give people access, but aren't necessarily finished, so some bugs will occur. It causes problems, but people get access to it sooner rather than having to wait a week or more for everything to be finished, while also giving a chance to have users discover problems to be fixed, otherwise you'd be waiting a month or more for the fully finished code and it might still be full of bugs because no one got to test it.
my hardware 64GB RAM 16GB VRAM i cant use 2 loras (100MB) ... booth RAM out of memory ... is it a bug ?
seems RAM is not purged well after patching
It feels like flux is and the features added to optimize it are slowly making all other models in forge slower. Constantly reloading every model each time no matter what, patching loras
