Search papers, labs, and topics across Lattice.
1
0
2
Current AI coding agents struggle with large-scale refactoring tasks, achieving only a 41.2% success rate on a newly curated benchmark designed to challenge their capabilities.