Retrieving the right skill from a large library is essential for LLM agents, yet requests and skills are written for different purposes and in different words. Existing retrievers narrow this request-skill gap with one or two representation choices each, so it is unknown whether their gains hold in combination. No prio...
Tian-Yi Sun, Peng-Fei Yao, Yan-Ze Wang et al.· 0 citations
SkillCorpus is, to the authors' knowledge, the first end-to-end account of when a curated, retrieval-served community corpus improves real agent tasks, and where it does not, and is evaluated end-to-end across three benchmarks.
Yanze Wang, Pengfei Yao, Tian-Yi Sun et al.· arXiv.org· 1 citation
EvoAgentBench is introduced, a benchmark for agent self-evolution via Ability-guided transfer across four agentic domains: web research, algorithmic reasoning, software engineering, software engineering, and knowledge work, which shifts self-evolution evaluation from aggregate accuracy comparison to fine-grained diagno...
