Reproduce “SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving” 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
Zhilong Ge, Yuting Shao, Yutao Yang, Yuxuan Cai, Jie Zhou, Kai Chen, Bo Zhang, Qin Chen and 1 more
SkillGym turns human-written agent skills into executable tasks with code-based verifiers, then uses successful action-and-observation trajectories to fine-tune models. Its suite contains 2,756 accepted environments, including 1,081 classified as skill-dependent under a reference configuration. On SkillsBench v1.1, SkillGym-Agent scores 26.81% without skills versus 23.34% for its base with skills under Claude Code, while skills still improve trained agents.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.