DeP mitigates MLLM hallucinations by dynamically perturbing text prompts to identify and reinforce stable visual evidence regions while counteracting language prior biases using attention variance and logit statistics.
arXiv preprint arXiv:2505.10634 (2025)
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
baseline 1
citation-polarity summary
verdicts
UNVERDICTED 2roles
baseline 1polarities
baseline 1representative citing papers
VGA constructs precise visual grounding from token semantics to guide MLLM attention toward relevant regions, dynamically suppressing described areas in captioning, and achieves SOTA dehallucination with negligible overhead.
citing papers explorer
-
Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation
DeP mitigates MLLM hallucinations by dynamically perturbing text prompts to identify and reinforce stable visual evidence regions while counteracting language prior biases using attention variance and logit statistics.
-
Tell Model Where to Look: Mitigating Hallucinations in MLLMs by Vision-Guided Attention
VGA constructs precise visual grounding from token semantics to guide MLLM attention toward relevant regions, dynamically suppressing described areas in captioning, and achieves SOTA dehallucination with negligible overhead.