TY  - RPRT
TI  - GAUGE: When Not to Trust LLM-as-a-Judge in User-Simulated Evaluation of Task-Oriented Agents
AU  - Umesh Bodhwani
AU  - Thanh Tran
AU  - Kai Wei
PY  - 2026
UR  - https://arxiv.org/abs/2609.12191
ID  - 2609.12191
ER  -