Masanari Oi

NOTE: Some of my earlier publications appear under the name Masanari Ohi, which was a previous romanization of my name. Both refer to myself, Masanari Oi.

International Conference (peer-reviewed)

• Youmi Ma, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Masanari Ohi, Hinari Shimada, Taihei Shiotani, Koshiro Saito, Koki Maeda, Kakeru Hattori, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki. Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models. COLM2025.
  [paper]

• Yuto Nishimura, Takumi Hirose, Masanari Ohi, Hideki Nakayama, Nakamasa Inoue. HALL-E: Hierarchical Neural Codec Language Model for Minute-Long Zero-Shot Text-to-Speech Synthesis. ICLR2025.
  [paper]

• Nakamasa Inoue, Shinta Otake, Takumi Hirose, Masanari Ohi, Rei Kawakami. ELP-Adapters: Parameter Efficient Adapter Tuning for Various Speech Processing Tasks. IEEE/ACM TASLP.
  [paper]

• Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Hiroki Iida, Masanari Ohi, Kakeru Hattori, Shota Hirai, Sakae Mizuki, Rio Yokota, Naoaki Okazaki. Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities. COLM2024.
  [paper]

• Naoaki Okazaki, Kakeru Hattori, Shota Hirai, Hiroki Iida, Masanari Ohi, Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Rio Yokota, Sakae Mizuki. Building a Large Japanese Web Corpus for Large Language Models. COLM2024.
  [paper] 

• Masanari Ohi, Masahiro Kaneko, Ryuto Koike, Mengsay Loem, Naoaki Okazaki. Likelihood-based Mitigation of Evaluation Bias in Large Language Models. ACL2024 (short, Findings).
  [paper][code]

Journal (peer-reviewed)

• Masanari Ohi, Masahiro Kaneko, Ryuto Koike, Mengsay Loem, Naoaki Okazaki. 大規模言語モデルにおける評価バイアスの尤度に基づく緩和. (Likelihood-based Mitigation of Evaluation Bias in Large Language Models.)  自然言語処理, 32(2):480–496, Jun 2025.
  [paper]

Workshop

• Koshiro Saito, Sakae Mizuki, Masanari Ohi, Taishi Nakamura, Taihei Shiotani, Koki Maeda, Youmi Ma, Kakeru Hattori, Kazuki Fujii, Takumi Okamoto, Shigeki Ishida, Hiroya Takamura, Rio Yokota, Naoaki Okazaki. Why We Build Local Large Language Models: An Observational Analysis from 35 Japanese and Multilingual LLMs. Multilingual and Equitable Language Technologies (MELT), held in conjunction with COLM 2025.
  [paper]

Preprint (non-peer-reviewed)

• Masanari Ohi, Masahiro Kaneko, Naoki Okazaki, Nakamasa Inoue. Multi-modal, Multi-task, Multi-criteria Automatic Evaluation Using a Vision Language Model. arXiv.
  [paper]

Domestic Conference and Symposium (non-peer-reviewed)

• Taihei Shiotani, Masahiro Kaneko, Ayana Niwa, Yuki Maruyama, Daisuke Oba, Masanari Oi, Naoaki Okazaki. JUBAKU: 日本文化における偏見評価のための敵対的ベンチマーク.(Adversarial Benchmark for Evaluating Stereotypes in Japanese Culture.) The 39th Annual Conference of the Japanese Society for Artificial Intelligence, 2025, Osaka, Japan. (in Japanese). Annual Conference Award.

• Masanari Oi, Masahiro Kaneko, Naoki Okazaki, Nakamasa Inoue. 複数タスク・複数項目に跨ったマルチモーダル自動評価手法. (Multi-modal, Multi-task, Multi-criteria Automatic Evaluation Using a Vision Language Model.) The 31st Annual Meeting of the Association of Natural Language Processing (NLP2025), Nagasaki, Japan. (in Japanese).  Committee Special Award.

• Youmi Ma, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Masanari Ohi, Hinari Shimada, Taihei Shiotani, Koshiro Saito, Koki Maeda, Kakeru Hattori, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki. 模倣学習による大規模言語モデルの指示チューニング. The 31st Annual Meeting of the Association of Natural Language Processing (NLP2025), Nagasaki, Japan. (in Japanese). 

• Kakeru Hattori, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Taihei Shiotani, Kokoro Ueki, Takuro Niitsuma, Akira Kawabata, Hideaki Tamori, Youmi Ma, Koki Maeda, Masanari Ohi, Koshiro Saito, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki. 新聞記事からつくる 時事と社会に強い日本語LLM.  The 31st Annual Meeting of the Association of Natural Language Processing (NLP2025), Nagasaki, Japan. (in Japanese). 

• Kakeru Hattori, Naoaki Okazaki, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Masanari Ohi, Taihei Shiotani, Koshiro Saito, Youmi Ma, Koki Maeda, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura. Swallowコーパスv2: 教育的な日本語ウェブコーパスの構築.  The 31st Annual Meeting of the Association of Natural Language Processing (NLP2025), Nagasaki, Japan. (in Japanese). 

• Koshiro Saito, Sakae Mizuki, Masanari Ohi, Takashi Nakamura, Taihei Shiotani, Koki Maeda, Ma Youmi, Kakeru Hattori, Kazuki Fujii, Takumi Okamoto, Shigeki Ishida, Hiroya Takamura, Rio Yokota, Naoaki Okazaki . LLMに日本語テキストを学習させる意義. (Advantages of Training LLMs on Japanese Text.) The 261st NL Research Presentation. Best Research Award.

• Masanari Ohi, Masahiro Kaneko, Ryuto Koike, Mengsay Loem, Naoaki Okazaki. 大規模言語モデルにおける評価バイアスの尤度に基づく緩和. (Likelihood-based Mitigation of Evaluation Bias in Large Language Models.) The 30th Annual Meeting of the Association of Natural Language Processing (NLP2024), Hyogo, Japan. (in Japanese). Young Researcher Award. 

• Naoaki Okazaki, Kakeru Hattori, Shota Hirai, Hiroki Iida, Masanari Ohi, Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Rio Yokota, Sakae Mizuki. Swallowコーパス: 日本語大規模ウェブコーパス. (Building a Large Japanese Web Corpus for Large Language Models.) The 30th Annual Meeting of the Association of Natural Language Processing (NLP2024), Hyogo, Japan. (in Japanese). Best Paper Award.

• Sakae Mizuki, Hiroki Iida, Kazuki Fujii Taishi Nakamura, Mengsay Loem, Masanari Ohi, Kakeru Hattori, Shota Hirai, Rio Yokota, Naoaki Okazaki. 大規模言語モデルの日本語能力の効率的な強化: 継続事前学習における語彙拡張と対訳コーパスの活用. The 30th Annual Meeting of the Association of Natural Language Processing (NLP2024), Hyogo, Japan. (in Japanese)

• Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Hiroki Iida, Masanari Ohi, Kakeru Hattori, Shota Hirai, Sakae Mizuki, Rio Yokota, Naoaki Okazaki. 継続事前学習による日本語に強い大規模言語モデルの構築. The 30th Annual Meeting of the Association of Natural Language Processing (NLP2024), Hyogo, Japan. (in Japanese). Best Paper Award.

Reviewer

Apr 2024 - Present: Master of Engineering (Computer Science) Institute of Science Tokyo (Formerly known as Tokyo Institute of Technology)