Go Kamoda

Master's Student
@TohokuNLP

Publications

Weight-based Analysis of Detokenization in Language Models: Understanding the First Stage of Inference Without Inference

Go Kamoda, Benjamin Heinzerling, Tatsuro Inaba, Keito Kudo, Keisuke Sakaguchi, Kentaro Inui

In Findings of the Association for Computational Linguistics: NAACL 2025

SoftMatcha: A Soft and Fast Pattern Matcher for Billion-Scale Corpus Searches

Hiroyuki Deguchi, Go Kamoda, Yusuke Matsushita, Chihiro Taguchi, Kohei Suenaga, Masaki Waga, Sho Yokoi

In The Thirteenth International Conference on Learning Representations (ICLR 2025)

OpenReview Project Page

@inproceedings{ deguchi-iclr-2025-softmatcha,
    author = { Hiroyuki Deguchi and Go Kamoda and Yusuke Matsushita and Chihiro Taguchi and Kohei Suenaga and Masaki Waga and Sho Yokoi },
    title = { SoftMatcha: A Soft and Fast Pattern Matcher for Billion-Scale Corpus Searches },
    booktitle = { The Thirteenth International Conference on Learning Representations (ICLR 2025) },
    year = { 2025 },
    url = { https://openreview.net/forum?id=Q6PAnqYVpo }
}

Quantifying the Influence of Evaluation Aspects on Long-Form Response Assessment

Go Kamoda, Akari Asai, Ana Brassard, Keisuke Sakaguchi

In Proceedings of the 31st International Conference on Computational Linguistics (COLING 2025)

@inproceedings{ kamoda-coling-2025-quantifying,
    author = { Go Kamoda and Akari Asai and Ana Brassard and Keisuke Sakaguchi },
    title = { Quantifying the Influence of Evaluation Aspects on Long-Form Response Assessment },
    booktitle = { Proceedings of the 31st International Conference on Computational Linguistics (COLING 2025) },
    year = { 2025 },
    url = { https://aclanthology.org/2025.coling-main.588/ }
}

The Curse of Popularity: Popular Entities have Catastrophic Side Effects when Deleting Knowledge from Language Models

Ryosuke Takahashi, Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

In arXiv [cs.CL]

@article{ takahashi-arxiv-2024-curse,
    author = { Ryosuke Takahashi and Go Kamoda and Benjamin Heinzerling and Keisuke Sakaguchi and Kentaro Inui },
    title = { The Curse of Popularity: Popular Entities have Catastrophic Side Effects when Deleting Knowledge from Language Models },
    journal = { arXiv [cs.CL] },
    year = { 2024 },
    url = { https://arxiv.org/abs/2406.06032 }
}

Test-time Augmentation for Factual Probing

Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

In Findings of the Association for Computational Linguistics: EMNLP 2023

@inproceedings{ kamoda-emnlp-2023-test,
    author = { Go Kamoda and Benjamin Heinzerling and Keisuke Sakaguchi and Kentaro Inui },
    title = { Test-time Augmentation for Factual Probing },
    booktitle = { Findings of the Association for Computational Linguistics: EMNLP 2023 },
    year = { 2023 },
    url = { https://aclanthology.org/2023.findings-emnlp.236/ }
}

Domestic Conferences

言語モデルのパラメータから探るDetokenizationメカニズム

鴨田豪, Benjamin Heinzerling, 稲葉達郎, 工藤慧音, 坂口慶祐, 乾健太郎

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper

言語モデルの内部表現における文法情報の局所性について

佐藤宏亮, 鴨田豪, Benjamin Heinzerling, 坂口慶祐

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper

LM は日本の時系列構造をどうエンコードするか

佐々木睦史, 高橋良允, 鴨田豪, Benjamin Heinzerling, 坂口慶祐, 乾健太郎

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper Sponsor Award (Nikkei CDIO)

日本語バイト符号化マスク言語モデルの開発と分析

工藤慧音, 鴨田豪, 塩野大輝, 鈴木潤

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper

層の冗長性と層同士の独立性に基づく言語モデルの層交換の成否の特徴づけ

小林春斗, 原知正, 鴨田豪, 横井祥

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper Researcher Encouragement Award (20/487)

SoftMatcha: 大規模コーパス検索のための柔らかくも高速なパターンマッチャー

出口祥之, 鴨田豪, 松下祐介, 田口智大, 末永幸平, 和賀正樹, 横井祥

In Proceedings of the Thirty-first Annual Meeting of the Association for Natural Language Processing (NLP 2025)

Conference Paper

事前学習–文脈内学習パラダイムで生じる頻度バイアスの較正

*伊藤郁海, *鴨田豪, 熊谷雄介, 横井祥 (* Equal contribution)

In The 19th Symposium of Young Researcher Association for NLP Studies (YANS 2024)

層同士の接続可能性と各層が影響を与える部分空間の重なり度合いの関係性

小林春斗, 原知正, 鴨田豪, 横井祥

In The 19th Symposium of Young Researcher Association for NLP Studies (YANS 2024)

Conference Researcher Encouragement Award (23/187)

柔らかいgrep/KWICに向けて：高速単語列マッチングの埋め込み表現による連続化

出口祥之, 鴨田豪, 松下祐介, 慶田開, 和賀正樹, 横井祥

In The 19th Symposium of Young Researcher Association for NLP Studies (YANS 2024)

Conference Project Page Demo Award (1/15)Sponsor Award (Recruit Co., Ltd.)

大規模言語モデルの情報推薦バイアスの較正

熊谷雄介, *伊藤郁海, *鴨田豪, 横井祥 (* Equal contribution)

In The 38th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2024)

Abstract Conference

長文生成の多面的評価:人手評価と自動評価の向上を目指して

鴨田豪, 浅井明里, Ana Brassard, 坂口慶祐

In Proceedings of the Thirtieth Annual Meeting of the Association for Natural Language Processing (NLP 2024)

Abstract Conference Outstanding Paper Award (12/599)

言語モデルからの知識削除：頻出実体の知識は副作用が破滅的

高橋良允, 鴨田豪, Benjamin Heinzerling, 坂口慶祐, 乾健太郎

In Proceedings of the Thirtieth Annual Meeting of the Association for Natural Language Processing (NLP 2024)

Abstract Conference Researcher Encouragement Award (18/427)

Test-time Augmentation for Factual Probing

Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

In Proceedings of the Twenty-ninth Annual Meeting of the Association for Natural Language Processing (NLP 2023)

Abstract Conference

Affiliation

2023.04 - 2025.03	Master's Student TohokuNLP, Graduate School of Information Sciences, Tohoku University
2020.04 - 2023.03	Bachelor's Student Department of Electrical, Information and Physics Engineering, Tohoku University

Experiences

2023.09 Yokohama, Japan	NS Solutions R&D Internship
2021.03 - 2023.08 Tokyo, Japan	Programming Mentor, School Managemenent Member Infratop (DMM WebCamp)