DIG is Ling Yang's research group at Peking University. We study LLM / Agent Post-Training, RL Systems / Infrastructure, Recursive Self-Improvement, and AI for Discovery.
Learning, systems, self-improvement, and discovery for next-generation models and agents.
Post-training for reasoning, agentic behavior, reward / verifier design, and long-horizon learning.
Explore → 02Scalable infrastructure for asynchronous, long-horizon, and multi-environment reinforcement learning.
Explore → 03Agents that improve through experience, memory, skills, environments, and self-evolving harnesses.
Explore → 04Models that discover problems, form hypotheses, design experiments, and revise against evidence.
Explore →DIG develops reusable research systems and open-source projects with the Gen-Verse community.
Incoming Assistant Professor and PhD advisor at Peking University; currently a Postdoctoral Researcher in ECE / Princeton AI Lab at Princeton University.
We welcome Master's students, PhD students, postdocs, and research interns, as well as academic and industrial collaborations.