#judgment-failure

Judgment-focused Benchmark 도입으로 LLM 정확도 48.84%p 향상

I Built a Benchmark for the Failures Generic LLM Evaluations Miss