[PyCon Korea 2025] 고등학생이 만들어 본 LLM-as-a-judge 번역 벤치마크 리더보드 - 심기현
PyCon Korea · 14:19
A high-school student, Sim Gi-heon (Ki-Heon Shim), built Colti, an LLM-as-judge benchmark and leaderboard that scores Korean translation quality on hard-to-translate expressions, then compared 17 systems ranging from...