<ul><li>Reasoning is central to human intelligence, enabling structured problem-solving across diverse tasks.</li><li>Recent advances in large language models (LLMs) have enhanced reasoning abilities in arithmetic, commonsense, and symbolic domains.</li><li>Extending these capabilities to multimodal contexts, where models integrate visual and textual inputs, remains a challenge.</li><li>This paper provides an overview of reasoning techniques in textual and multimodal LLMs, highlighting challenges, opportunities, and directions for future research.</li></ul>

Why Reasoning Matters? A Survey of Advancements in Multimodal Reasoning (v1)

Discover more