Zigzag persistent homology on layer-wise hidden-state point clouds guides adaptive layer pruning of LVLMs and reportedly beats prior pruning methods across sparsity levels.
InInternational conference on ma- chine learning, pages 19730–19742
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
dataset 1
citation-polarity summary
fields
cs.CV 2verdicts
UNVERDICTED 2roles
dataset 1polarities
use dataset 1representative citing papers
LLaVA-OneVision is the first single open LMM to simultaneously achieve strong performance in single-image, multi-image, and video scenarios with cross-scenario transfer capabilities.
citing papers explorer
-
Topology-Aware Layer Pruning for Large Vision-Language Models
Zigzag persistent homology on layer-wise hidden-state point clouds guides adaptive layer pruning of LVLMs and reportedly beats prior pruning methods across sparsity levels.
-
LLaVA-OneVision: Easy Visual Task Transfer
LLaVA-OneVision is the first single open LMM to simultaneously achieve strong performance in single-image, multi-image, and video scenarios with cross-scenario transfer capabilities.