titleOut of Sight, Still in Mind: Token Compression for Omni-LLMs
authorsSuho Yoo, Youngjoon Jang, Hyebin Cho, Joon Son Chung
date2026-11-17
venuearXiv (preprint, not peer reviewed)
idarXiv:2607.21179
affiliationKAIST AI
upvotes3 (Hugging Face Daily Papers, 2026-11-21)
linksabs pdf hf code
cite
relatedOmniShowcase: Omnidirectional Subject-Consistent Video Generation from Multi-View Images
Look Only When Needed: Certificate-Guided Visual Routing for Efficient Diffusion VLMs
Language Models Can Control Their Own Attention
OmniRetrieval: Unified Retrieval across Heterogeneous Knowledge Sources
GRACE: Generation-aware latent compression for efficient video generation
abstractThe goal of this paper is to reduce the input token cost of Omni-modal large language models (Omni-LLMs) at inference time. Omni-LLMs reason jointly over audio, video and text, but the cost of the three streams is highly unbalanced: visual tokens account for the vast majority of the input, and are highly redundant. In this paper, we propose ReMo, a training-free framework that compresses visual tokens by redistributing their information across modalities: a visual token is kept only if its information appears nowhere else. ReMo achieves this in two ways: (i) it aligns audio and video in a shared space via closed-form calibration, and removes visual tokens already explained by the audio or by other visual tokens; and (ii) it replaces object-level visual tokens, identified by a lightweight object detector, with compact text proxies, short descriptions of each object and its location, conveying the same content in far fewer tokens. On Qwen2.5-Omni at two model scales, ReMo removes 54% of the input tokens with comparable accuracy. It performs comparably to the full-token model, reaching 101.2% and 101.4% of its average accuracy over five audio-visual benchmarks.

bibtex

@misc{yoo2026out,
  title={Out of Sight, Still in Mind: Token Compression for Omni-LLMs},
  author={Suho Yoo and Youngjoon Jang and Hyebin Cho and Joon Son Chung},
  year={2026},
  eprint={2607.21179},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2607.21179}
}