baaivision/Emu
A series of multimodal foundation models that generate images and text through next-token prediction, developed by BAAI.
Not currently ranked — collecting fresh signals.
star history
Emu is a collection of generative multimodal models including Emu1 (generative pretraining in multimodality), Emu2 (in-context learners), and Emu3 (next-token prediction approach). These models function as multimodal generalists capable of understanding and generating both text and images in multimodal contexts. The project provides inference code, models, and demos for research and community use.
Frequently asked
- What is baaivision/Emu?
- A series of multimodal foundation models that generate images and text through next-token prediction, developed by BAAI.
- Is Emu open source?
- Yes — baaivision/Emu is open source, released under the Apache-2.0 license.
- What language is Emu written in?
- baaivision/Emu is primarily written in Python.
- How popular is Emu?
- baaivision/Emu has 1.8k stars on GitHub.
- Where can I find Emu?
- baaivision/Emu is on GitHub at https://github.com/baaivision/Emu.