ACG mitigates hallucinations in LVLMs via single-pass contrastive guidance in attention space that suppresses text-only biases through masking and orthogonal projection.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
VLMs exhibit only slight performance degradation on hallucination benchmarks when substantial image tokens are removed, with layer-wise analysis showing increased visual token similarity in deeper layers, suggesting current benchmarks inadequately test fine-grained visual grounding.
citing papers explorer
-
Attention-space Contrastive Guidance for Efficient Hallucination Mitigation in LVLMs
ACG mitigates hallucinations in LVLMs via single-pass contrastive guidance in attention space that suppresses text-only biases through masking and orthogonal projection.
-
Seeing without Looking: Do Vision-Language Benchmarks Really Test Vision?
VLMs exhibit only slight performance degradation on hallucination benchmarks when substantial image tokens are removed, with layer-wise analysis showing increased visual token similarity in deeper layers, suggesting current benchmarks inadequately test fine-grained visual grounding.