OmniFusion
OmniFusion copied to clipboard
Add 4bit inference code
Hello, I'm really interested in your project. But it takes much vram to run. Could you please add inference code in 4bits using bitsandbytes or smth