J-tau: A Fully Localized Japanese Benchmark for Tool-Agent-User Interaction
J-tau is presented, the first publicly available Japanese benchmark for evaluating tool-using LLM agents in interactive customer-support scenarios and carefully localizes tau-bench through machine translation, entity adaptation, and human review to enable reliable evaluation in Japanese.
Chihiro Yano, Jun Hirako, Hiroaki Yamagiwa et al.
· 0 citations