Selecting tokens that lie inside multiple anchor-centric semantic grain regions improves multimodal VQA/AVQA accuracy under latency and erasure constraints compared with pairwise attention-based selection.
Visual instruction tun ing,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SP 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Geometric Cross-Modal Token Selection for Latency-Constrained Multimodal Token Communication
Selecting tokens that lie inside multiple anchor-centric semantic grain regions improves multimodal VQA/AVQA accuracy under latency and erasure constraints compared with pairwise attention-based selection.