Related Links:
NOTES:
- This work introduces SkillWrapper: a system for creating abstract representations of skills using foundation models.
- Unlike related work that assume predefined predicates for skill representations, our system solely relies on a foundation model to propose (or invent) predicates.
- The foundation model also proposes a set of execution traces from which our system actively collects data for deriving skill abstractions.
- David Paulius and Naman Shah are joint final authors (equal advising).
Citation:
Z. Yang, B. Hedegaard, A. Jaafar, Y. Wei, S. Thompson, S. S. Raman, H. Fu, S. Tellex, G. Konidaris, D. Paulius*, and N. Shah*. “SkillWrapper: Generative Predicate Invention for Skill Abstraction.”, arXiv preprint arXiv:2511.18203.