2 upvotes
New method curbs cross-modal hallucinations in Omni LLMs
What is Omni-Woodpecker: Hallucination Correction via Evidence-Prior Rebalancing for Omni LLMs?
Omni-Woodpecker rebalances the weight of audio-visual evidence against the model’s internal priors, trimming hallucinated outputs. It can be applied to existing Omni LLMs as a lightweight post-training adjustment.
Who it is for
Anyone building multimodal AI agents can use it to improve grounding and avoid misleading answers.