Toward Small, Open-Source, Multi-Modal Language Model Agents for Science and Society

Abstract: Recent advances in large language models have shown impressive capabilities across scientific domains and societal applications, but their size and proprietary nature often limit accessibility and reproducibility. In this talk, I will present our work on developing small, open-source, multi-modal language model agents that can reason, plan and act in diverse scientific and societal contexts.

I will discuss methods for designing small but highly effective language models, integrating multi-modal inputs and coordinating multi-agent interactions to achieve complex tasks. I will also highlight real-world applications in science and society, emphasizing transparency, reproducibility and accessibility.

Bio: Xuan Wang is an assistant professor in the department of Computer Science at Virginia Tech and a faculty member of the Sanghani Center for Artificial Intelligence and Data Analytics. Her work has been recognized by the National Science Foundation CAREER Award (2025), NVIDIA Academic Grant (2025), Cisco Research Award (2025–2026) and the NAACL Best Demo Paper Award (2021). She received her Ph.D. in computer science, M.S. in statistics and M.S. in biochemistry from the University of Illinois at Urbana-Champaign, and her B.S. in biological science from Tsinghua University, China.