Zhejiang University's natural language processing group (ZJU NLP) has released SkillX, a framework that automatically constructs reusable skill knowledge bases from the execution histories of LLM agents, publishing it on GitHub and arXiv. It abstracts an agent's past task trajectories into three tiers of skills that can be injected directly into other agents.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.