Reproduce “G-CAST: Governed Continual Self-Training for E-Commerce Agents under Evolving Demands” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuechun Yu, Wenzheng Kong, Xiaofeng Mao, Jin Zhang
G-CAST adapts a B2B procurement agent as buyer demands and its tools, skills, and procedures change. It aligns reward rules with the capability contract—available information, actions, and required procedures—and tests candidates on new and retained tasks. It exceeds AutoTrainess in cumulative reward in Iterations 3 and 4, in one production application and model family.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.