Profile
Akira
Kawabata
Ph.D. student at SOKENDAI, affiliated with the National Institute of Informatics (NII). Advised by Saku Sugawara.
News
Recent
- Apr 2026 Our paper on scalable rubric-augmented reward modeling is accepted to ACL 2026.
- Mar 2025 Received the INMA 30 Under 30 Award.
- Sep 2024 Our paper on LLM-based rationale verification is accepted to EMNLP 2024.
Publications
Papers
C2: Scalable Rubric-Augmented Reward Modeling from Binary Preferences
Abstract — ACL 2026
We turn each binary preference pair into an automatically-generated rubric, so a single label scales into many fine-grained supervision signals. The resulting reward models align language models more reliably with less labeled data.
Rationale-Aware Answer Verification by Pairwise Self-Evaluation
Abstract — EMNLP 2024
A verifier that compares pairs of rationales — not just final answers — catches the cases where an LLM is right for the wrong reason. Consistent gains across reasoning benchmarks where surface answers under-discriminate.
Evaluating the Rationale Understanding of Critical Reasoning in Logical Reading Comprehension
Abstract — EMNLP 2023
A benchmark that probes whether models actually understand the rationales behind critical-reasoning answers, rather than memorising answer patterns. A persistent rationale-level gap shows up across modern LLMs.
Experience
Experience & Education
Work
Apr 2023 — Mar 2026
Research Engineer
The Asahi Shimbun Company
Education
Apr 2026 — present
Ph.D. student
SOKENDAI / NII — Department of Informatics. Supervisor: Saku Sugawara.
Apr 2021 — Mar 2023
M.Eng.
Nara Institute of Science and Technology — Supervisor: Taro Watanabe.
Apr 2016 — Mar 2021
B.A.
Hitotsubashi University
Awards
Honors & Scholarships
- 2026 SOKENDAI Special Researcher Program, supported by JST BOOST.
- 2025 International News Media Association (INMA) 30 Under 30 Awards.
- 2023 Young Researcher's Encouragement Award, 29th Annual Meeting of the Association of Natural Language Processing.