Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling

Xiaokang Chen, Zhiyu Wu, Xingchao Liu, Zizheng Pan, Wen Liu +3 more
1/29/2025
cs.AIcs.CLcs.CV

No official code found. Best implementation: deepseek-ai/Janus (17,757 stars, MIT, last commit 2025-02-01).

Abstract

In this work, we introduce Janus-Pro, an advanced version of the previous work Janus. Specifically, Janus-Pro incorporates (1) an optimized training strategy, (2) expanded training data, and (3) scaling to larger model size. With these improvements, Janus-Pro achieves significant advancements in both multimodal understanding and text-to-image instruction-following capabilities, while also enhancing the stability of text-to-image generation. We hope this work will inspire further exploration in the field. Code and models are publicly available.

View on arXivView PDF

Code Implementations(1)

deepseek-ai/JanusCommunity100%
MIT

Janus-Series: Unified Multimodal Understanding and Generation Models

17,7572,233Python, MakefileOct 18, 20241 years agoMIT
any-to-anyfoundation-modelsllmmultimodalunified-model+1 more

Similar Papers

JanusCoder: Towards a Foundational Visual-Programmatic Interface for Code Intelligence

84% match2 repos782025

Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation

83% match1 repo17.8k2024

JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation

82% match1 repo17.8k2024

MANTIS: Interleaved Multi-Image Instruction Tuning

81% match1 repo2402024

ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation

81% match1 repo2832025

MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark

81% match2024

Cite this paper

@article{chen2025januspro,
  title  = {Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling},
  author = {Xiaokang Chen and Zhiyu Wu and Xingchao Liu and Zizheng Pan and Wen Liu and Zhenda Xie and Xingkai Yu and Chong Ruan},
  year   = {2025},
  eprint = {2501.17811},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2501.17811v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2501.17811.svg)](https://codeofpaper.com/paper/2501.17811)

Discussion