CaVe-VLM-CoT: An Interpretable Vision-Language Model Framework

Sneha Rao, Shaina Raza, Dhanesh Ramachandram

Vision-Language Models (VLMs) remain prone to hallucinations, producing fluent but visually unfaithful outputs. Existing chain-of-thought and retrieval-augmented methods only parti…