Paper
Who Speaks for the Pruned? Visual Token Pruning as Coverage Optimization
Visual token pruning reduces the inference cost of vision-language models (VLMs), but most methods only ask which tokens to keep. This retained-token view can keep redundant high-scoring tokens while leaving discarded evidence without a close representative. We propose CoverPruner, a training-free pruner that asks the complementary demand-side question: after a token is removed, which surviving original token represe · Authors: Qingchan Zhu, Weihang You, Hanqi Jiang, Changdi Yang, Tianming Liu, Geng Yuan · Published 2026-09-02 · arXiv:2609.03158
- Published
- 2026-09-02
- Authors
- 6
- arXiv
- 2609.03158
- Category
- cs.CV
Abstract
arXiv:2609.03158
Qingchan Zhu, Weihang You, Hanqi Jiang, Changdi Yang, Tianming Liu, Geng Yuan
Visual token pruning reduces the inference cost of vision-language models (VLMs), but most methods only ask which tokens to keep. This retained-token view can keep redundant high-scoring tokens while leaving discarded evidence without a close representative. We propose CoverPruner, a training-free pruner that asks the complementary demand-side question: after a token is removed, which surviving original token represents it for the target VLM? CoverPruner formulates pruning as Representational Coverage Maximization (RCM), covering the full projected visual-token set with query-weighted demand. It instantiates RCM with projector-space coverage and a lightweight first-layer attention probe. Across multiple VLM architectures and compression rates, CoverPruner achieves the best average accuracy among all compared methods, with the largest gains usually appearing under aggressive compression.