Search papers, labs, and topics across Lattice.
Kyushu University
2
0
4
Naively scaling LLMs isn't enough for multilingual log generation; language-specific logging idioms matter more than model size.
Judging AI coding agents solely by merged/rejected pull requests is misleading: over 60% of rejections aren't the agent's fault, and even "successful" merges often hide significant human intervention.
