Mogao: An Omni Foundation Model for Interleaved Multi-Modal Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liao, Chao, Liu, Liyang, Wang, Xun, Luo, Zhengxiong, Zhang, Xinyu, Zhao, Wenliang, Wu, Jie, Li, Liang, Tian, Zhi, Huang, Weilin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913831289094144
author Liao, Chao
Liu, Liyang
Wang, Xun
Luo, Zhengxiong
Zhang, Xinyu
Zhao, Wenliang
Wu, Jie
Li, Liang
Tian, Zhi
Huang, Weilin
author_facet Liao, Chao
Liu, Liyang
Wang, Xun
Luo, Zhengxiong
Zhang, Xinyu
Zhao, Wenliang
Wu, Jie
Li, Liang
Tian, Zhi
Huang, Weilin
contents Recent progress in unified models for image understanding and generation has been impressive, yet most approaches remain limited to single-modal generation conditioned on multiple modalities. In this paper, we present Mogao, a unified framework that advances this paradigm by enabling interleaved multi-modal generation through a causal approach. Mogao integrates a set of key technical improvements in architecture design, including a deep-fusion design, dual vision encoders, interleaved rotary position embeddings, and multi-modal classifier-free guidance, which allow it to harness the strengths of both autoregressive models for text generation and diffusion models for high-quality image synthesis. These practical improvements also make Mogao particularly effective to process interleaved sequences of text and images arbitrarily. To further unlock the potential of unified models, we introduce an efficient training strategy on a large-scale, in-house dataset specifically curated for joint text and image generation. Extensive experiments show that Mogao not only achieves state-of-the-art performance in multi-modal understanding and text-to-image generation, but also excels in producing high-quality, coherent interleaved outputs. Its emergent capabilities in zero-shot image editing and compositional generation highlight Mogao as a practical omni-modal foundation model, paving the way for future development and scaling the unified multi-modal systems.
format Preprint
id arxiv_https___arxiv_org_abs_2505_05472
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mogao: An Omni Foundation Model for Interleaved Multi-Modal Generation
Liao, Chao
Liu, Liyang
Wang, Xun
Luo, Zhengxiong
Zhang, Xinyu
Zhao, Wenliang
Wu, Jie
Li, Liang
Tian, Zhi
Huang, Weilin
Computer Vision and Pattern Recognition
Recent progress in unified models for image understanding and generation has been impressive, yet most approaches remain limited to single-modal generation conditioned on multiple modalities. In this paper, we present Mogao, a unified framework that advances this paradigm by enabling interleaved multi-modal generation through a causal approach. Mogao integrates a set of key technical improvements in architecture design, including a deep-fusion design, dual vision encoders, interleaved rotary position embeddings, and multi-modal classifier-free guidance, which allow it to harness the strengths of both autoregressive models for text generation and diffusion models for high-quality image synthesis. These practical improvements also make Mogao particularly effective to process interleaved sequences of text and images arbitrarily. To further unlock the potential of unified models, we introduce an efficient training strategy on a large-scale, in-house dataset specifically curated for joint text and image generation. Extensive experiments show that Mogao not only achieves state-of-the-art performance in multi-modal understanding and text-to-image generation, but also excels in producing high-quality, coherent interleaved outputs. Its emergent capabilities in zero-shot image editing and compositional generation highlight Mogao as a practical omni-modal foundation model, paving the way for future development and scaling the unified multi-modal systems.
title Mogao: An Omni Foundation Model for Interleaved Multi-Modal Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.05472