Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
2
0
4
A staggering 63% of LLM-based techniques from last year can be outperformed by a single prompt on newer models, questioning the value of complex engineering solutions.
LLM-powered code refactoring is more dangerous than we thought: up to 35% of refactorings subtly change program behavior, and existing tests often miss these critical errors.