comfyanonymous
Can't apply loras on playground 2.5
Applying lora onto playground 2.5 model doesn't seem to work, it works in diffusers, they seem to have made changes to support it: https://github.com/huggingface/diffusers/pull/7126
Are you using loras that were specifically trained for playground 2.5 and that used the specific EDM training method? PG2.5 is its own custom trained model. It shares the same SDXL architecture, but they use their completely own dataset. Therefore, loras that were trained for regular SDXL models will not work with PG2.5 because the data in their network is inherently completely different.
Are you using loras that were specifically trained for playground 2.5 and that used the specific EDM training method? PG2.5 is its own custom trained model. It shares the same SDXL architecture, but they use their completely own dataset. Therefore, loras that were trained for regular SDXL models will not work with PG2.5 because the data in their network is inherently completely different.
Yeah I kno, I trained them myself, as stated they work using diffusers, I know about edm
Post a simple comfyui workflow of it not working
You just apply lora onto the model using the EDM scheduler and you can see it, its really just the base workflow for lora with EDM scheduler.