HorizonWind2004/reconstruction-alignment
A self-supervised alignment method that enhances unified multimodal models' zero-shot performance across image generation and editing tasks.
Not currently ranked — collecting fresh signals.
star history
The paper introduces Reconstruction Alignment (RecA), a self-supervised training technique that improves unified multimodal models by aligning visual and textual representations through a reconstruction objective. It has been validated on architectures including BAGEL, Janus, Show-o, and others, requiring only 6×80GB A100 GPUs for 4.5 hours. The method claims to boost BAGEL’s image editing capabilities beyond FLUX-Kontext while remaining fully open-source and reproducible.
Frequently asked
- What is HorizonWind2004/reconstruction-alignment?
- A self-supervised alignment method that enhances unified multimodal models' zero-shot performance across image generation and editing tasks.
- Is reconstruction-alignment open source?
- Yes — HorizonWind2004/reconstruction-alignment is open source, released under the Apache-2.0 license.
- What language is reconstruction-alignment written in?
- HorizonWind2004/reconstruction-alignment is primarily written in Python.
- How popular is reconstruction-alignment?
- HorizonWind2004/reconstruction-alignment has 406 stars on GitHub.
- Where can I find reconstruction-alignment?
- HorizonWind2004/reconstruction-alignment is on GitHub at https://github.com/HorizonWind2004/reconstruction-alignment.