Paper Detail

ReToken: One Token to Improve Vision-Language Models for Visual Retrieval

Yao Xiao, Reuben Tan, Zhen Zhu, Yuqun Wu, Jianfeng Gao, Derek Hoiem

arxiv Score 14.3

Published 2026-07-30 · First seen 2026-07-31

General AI

Abstract

Long visual context poses a challenge for vision-language models: performance degrades as the number of distractors grows, and processing all tokens at once is computationally infeasible under GPU memory constraints. We present ReToken, a single learnable embedding trained as an explicit retrieval target that selects a sparse set of query-relevant visual tokens from a pre-filled visual KV cache. Trained on only a small image-QA dataset, ReToken yields consistent gains across image and video benchmarks: on Visual Haystacks it improves Qwen3VL-8B by 13.4 points and InternVL3.5 by 12.4 points (>20% relative), and on LVBench it transfers zero-shot to long video for an 8.0-point gain with Qwen3VL-8B. Thanks to its lightweight design, both training and long-video inference fit on a single H100. Code is available at: https://github.com/avaxiao/ReToken

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@article{xiao2026retoken,
  title = {ReToken: One Token to Improve Vision-Language Models for Visual Retrieval},
  author = {Yao Xiao and Reuben Tan and Zhen Zhu and Yuqun Wu and Jianfeng Gao and Derek Hoiem},
  year = {2026},
  abstract = {Long visual context poses a challenge for vision-language models: performance degrades as the number of distractors grows, and processing all tokens at once is computationally infeasible under GPU memory constraints. We present ReToken, a single learnable embedding trained as an explicit retrieval target that selects a sparse set of query-relevant visual tokens from a pre-filled visual KV cache. Trained on only a small image-QA dataset, ReToken yields consistent gains across image and video benc},
  url = {https://arxiv.org/abs/2607.28627},
  keywords = {cs.CV, cs.AI, cs.LG, Computer science, Artificial intelligence, Embedding, Set (abstract data type), Context (archaeology)},
  eprint = {2607.28627},
  archiveprefix = {arXiv},
}

Metadata

{}