Google's MUSE: muse-model.github.io reproduced ✅ Why token-base image generation❓ - very under explored research - needed for true multi-modal models - better at style transfer 👉 huggingface.co/spaces/amused/…