Search papers, labs, and topics across Lattice.
Ben-Gurion University of the Negev
1
0
3
Multi-agent code editing with structured failure feedback boosts task success by 17%, suggesting a promising path to more reliable LLM-driven code manipulation.