CVAIMay 10

Evading Visual Aphasia: Contrastive Adaptive Semantic Token Pruning for Vision-Language Models

arXiv:2605.0942961.0
AI Analysis

For practitioners deploying large vision-language models, COAST offers a robust, training-free pruning method that maintains high accuracy across diverse tasks and model families.

COAST introduces a training-free pruning method that reduces visual tokens by 77.8% and achieves 2.15x speedup while retaining 98.64% performance across seven benchmarks, addressing the problem of premature token pruning causing loss of visual grounding in vision-language models.

Are low-attention visual tokens truly redundant in vision-language reasoning? Existing pruning methods often assume so, ranking visual tokens by shallow text-to-image attention and discarding low-scoring patches to accelerate LVLM inference. We show that this scalar criterion is unreliable for compositional reasoning: tokens ignored in early layers can later become essential for resolving secondary objects, spatial relations, and contextual cues. Premature pruning can therefore induce Visual Aphasia, a failure mode in which the model loses visual grounding and falls back on language priors. We introduce COAST (COntrastive Adaptive Semantic Token Pruning), a training-free pruning framework that casts compression as adaptive semantic routing. COAST uses native cross-modal attention to identify query-specific anchors and estimate contextual dispersion via attention entropy, then adapts the retention trade-off between semantic evidence and spatial context. It further uses a contrastive routing score to preserve both anchor-aligned evidence and complementary spatial context. Across seven benchmarks, COAST reduces visual tokens by 77.8% and achieves a 2.15x latency speedup while retaining 98.64% of the original average performance. Beyond a single backbone or compression setting, COAST consistently outperforms strong pruning baselines across token budgets and generalizes across multiple LVLM families, showing that adaptive semantic routing is a robust alternative to one-shot scalar pruning

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes