kohya_ss
kohya_ss copied to clipboard
all LORAs produce "modules.devices.NansException: A tensor with all NaNs was produced in Unet" in StableDiffusion
Was using very old Kohya version, updated with fresh install to be able to train SDXL and all new LORA become broken. Backup copy of old version also stopped working after installing new kohya and getting fatal error on train.
Pretty much same scenario happened to me. Did anyone manage to fix it?