Language AI Researcher
My work bridges theory and deployment for large language models, with a focus on evaluation, safety, and multilingual adaptation.
Graduate School of Informatics, Example University
Taro Yamada, Alice Smith, Kenji Sato
We analyze failure modes of tool-augmented LLM agents in multilingual settings and introduce a reliability benchmark.
Mika Tanaka, Taro Yamada, Robert Lee
We reduce factual hallucinations by explicitly training evidence-grounded reasoning.
Taro Yamada, Emi Kobayashi
We propose a safe response optimization method combining human feedback and constrained decoding.