Atsuki Yamaguchi

English | 日本語

About

I am a third-year PhD candidate in the Sheffield NLP Group, focusing on cross-lingual transfer and efficient language modeling. I have 7+ years of combined academic and industry experience in NLP and hold an MSc in Computer Science with Speech and Language Processing (Distinction) from the University of Sheffield (2020). From 2021 to 2023, I was a researcher in the R&D Group at Hitachi, Ltd. (Japan), working on information extraction and efficient language model development. My research has been published at top-tier NLP/ML conferences and journals, including ACL, EMNLP, TMLR, and Computational Linguistics. On top of the research activities, I also contribute to the community as a member of the ACL Rolling Review (ARR) support team.

Download CV CV of Failures

I am planning to submit my thesis in March 2027 and am officially entering the job market!

I am primarily seeking Faculty positions starting April 2027. I am also open to Research Fellow, Post-doc, or Research Scientist roles as secondary options. If you know of any opportunities or have suggestions, please contact me at [email protected].

• Impact: Published in EMNLP/ACL/EACL/CL/TMLR (first-author); ICML/NeurIPS/COLING (co-author)
• Industry-experience: ~3 years of full-time R&D @ Hitachi, Ltd.
• Leadership: EACL 2026 SRW Chair & ARR Support Team
• Award-Winning: Multi-time ARR Outstanding Reviewer & Nominated for University of Sheffield Teaching Support Award
• Research: Cross-lingual Transfer & Efficient Pre-training

News

• Oct 24, 2025
  Our paper on extremely low-resource vocabulary expansion has been accepted to Computational Linguistics! See here for the preprint.
• Sep 15, 2025
  Our paper on vocabulary expansion of chat models has been accepted to Transactions on Machine Learning Research (TMLR)! See here.
• Sep 1, 2025
  I will be co-chairing the EACL 2026 Student Research Workshop. Call for papers will be out soon!
• Sep 1, 2025
  Updated CV. See here.

Research Interests

Cross-lingual Transfer, Language Modelling, Natural Language Understanding, Natural Language Processing, Machine Learning

Recent Work

Towards efficient and fair natural language processing across languages

This research helps large language models (LLMs) work well in languages other than English so more people worldwide can use AI. My research particularly focuses on efficient ways to adapt models to new language(s) and on expanding their vocabularies so the models perform better across many languages.

• Atsuki Yamaguchi, Terufumi Morishita, Aline Villavicencio and Nikolaos Aletras, Mitigating Catastrophic Forgetting in Target Language Adaptation of LLMs via Source-Shielded Updates, arXiv preprint, December 2025.
• Atsuki Yamaguchi, Terufumi Morishita, Aline Villavicencio and Nikolaos Aletras, Adapting Chat Language Models Using Only Target Unlabeled Language Data, Transactions on Machine Learning Research (TMLR), September 2025.
• Atsuki Yamaguchi, Aline Villavicencio and Nikolaos Aletras, How Can We Effectively Expand the Vocabulary of LLMs with 0.01GB of Target Language Text?, Computational Linguistics, November 2025.
• Atsuki Yamaguchi, Aline Villavicencio and Nikolaos Aletras, An Empirical Study on Cross-lingual Vocabulary Adaptation for Efficient Language Model Inference, Findings of the Association for Computational Linguistics: EMNLP 2024, November 2024.

Explore efficient pretraining methods for language models

This research explores efficient pretraining methods for language models, seeking ways to acquire high-quality language representations with fewer resources. We compare simple and efficient pretraining objectives for language models and investigate which settings lead to the acquisition of useful language representations.

• Atsuki Yamaguchi, Maggie Mi and Nikolaos Aletras, Enhancing Linguistic Competence of Language Models through Pre-training with Language Learning Tasks, arXiv preprint, January 2026.
• Atsuki Yamaguchi, Hiroaki Ozaki, Terufumi Morishita, Gaku Morio and Yasuhiro Sogawa, How does the task complexity of masked pretraining objectives affect downstream performance?, Findings of the Association for Computational Linguistics: ACL 2023, July 2023.
• Atsuki Yamaguchi, George Chrysostomou, Katerina Margatina and Nikolaos Aletras, Frustratingly Simple Pretraining Alternatives to Masked Language Modeling, The 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021), November 2021.

Links

• Google Scholar
• Semantic Scholar
• ACL Anthology
• DBLP

Contact

You can send me messages via this contact form.

Updated on April 3, 2026