Chain-of-Spot introduces Interactive Reasoning to enhance feature extraction and improve LVLM performance.
Enhancing large vision-language models through interactive reasoning with Chain-of-Spot.