List your company

False Success Detection on tau2-bench task-disjoint

0.545Judge AUROCGPT-4o
0.517750.5313750.5450.558625Jun 1, 2026

Evaluation results

MethodLinks
GPT-4oDomain=Cust. service, Agent=Conversational2026.060.545—
TF-IDF + XGBoostDomain=Cust. service, Agent=Conversational2026.06—0.825