1 paper · 1 filter
Jungin Kim, Shinwoo Park, Yo-Sub Han
Identifying LLM-generated code through watermarking poses a challenge in preserving functional correctness. Previous methods rely on the assumption that watermarking high-entropy t…