ProjLens shows that backdoor parameters in MLLMs are encoded in low-rank subspaces of the projector and that embeddings shift toward the target direction with magnitude linear in input norm, activating only on poisoned samples.
Explainable and interpretable multimodal large language models: A comprehensive survey
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
roles
background 2representative citing papers
UHR-BAT is a budget-aware framework that uses text-guided multi-scale importance estimation plus region-wise preserve and merge strategies to compress visual tokens in ultra-high-resolution remote sensing vision-language models.
Introduces 3D-CBM framework mapping raw 3D inputs to multi-tiered interpretable concepts, achieving 88.8% concept accuracy and test-time intervention on PartNet and ShapeNet.
HONES ranks feed-forward neurons by their causal contributions from task-relevant attention heads and uses lightweight scaling to steer performance on multiple vision-language tasks.
Position paper claims multimodal LLMs can significantly advance scientific reasoning and proposes a four-stage roadmap plus challenges and suggestions.
citing papers explorer
-
ProjLens: Unveiling the Role of Projectors in Multimodal Model Safety
ProjLens shows that backdoor parameters in MLLMs are encoded in low-rank subspaces of the projector and that embeddings shift toward the target direction with magnitude linear in input norm, activating only on poisoned samples.