The ScienceOne session, produced by the Institute of Automation, Chinese Academy of Sciences, is organized to promote the co-building of an open, shared, and platform-driven AI for Science ecosystem with global partners.

In this session, Prof. ZHANG Qichao will introduce the concepts of reinforcement learning and large language model (LLM) reasoning. It first presents the basic preliminaries of reinforcement learning, then demonstrates common techniques for LLM reasoning. The talk also covers the ScienceOne post-training methodology and its application to astronomy tasks.

About the ANSO Science Lecture Series

The ANSO Science Lecture Series will feature distinguished scientists, industry leaders, and science policymakers who will present the latest scientific achievements, innovative technologies, and practical methodologies. It provides a platform for knowledge sharing and exchange, enabling member institutions and early-career researchers to gain insights into global scientific trends, foster collaborative innovation, and work together to address common global challenges.