1 paper · 1 filter
Jiajie Li, Huayi Zhang, Peng Lin +2
We present a novel framework that improves the reliability of LLM judges by selectively augmenting LLM with auxiliary evaluation dimensions. Existing LLM judges often miss crucial…