Compositional Question

Compositional question answering focuses on enabling artificial intelligence models to understand and answer questions that require combining multiple pieces of information or reasoning steps. Current research emphasizes improving the ability of large language models (LLMs) and vision-language models (VLMs) to handle complex, multi-hop reasoning through techniques like graph-based decoding, intermediate supervision, and novel benchmark creation to expose model weaknesses. This area is crucial for advancing AI's ability to perform complex reasoning tasks and has implications for applications ranging from question answering systems to knowledge base querying and visual reasoning.