ReToken: One Token to Improve Vision–Language Models for Visual Retrieval
ReToken introduces one learnable retrieval token that selects sparse visual tokens from long contexts, improving vision-language models by up to 13.4 points on visual retrieval while fitting on a single GPU.
Published 2026Atlanta Poster Session 6 · Fri, Dec 11, 4:30 PM–7:30 PM local time · Hall C1▲ 8 on Hugging FaceCode ★ 18arXiv ↗OpenReview ↗
Only vote on papers you've read. Sign in with GitHub to vote.
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