Hiroto Kurita
Hi there 👋 I am a 2nd-year M.S. student at TohokuNLP Group, Tohoku University, working with Kentaro Inui and Sho Yokoi.
🧪 Working on theoretical understanding of text embeddings and large language models, also enjoys training large models! 🤖
Blogs2023/10/25 6:342023/10/25 12:17
Education
October 2022-
Present, Sendai
Tohoku University
M.S. in Information Science
Advisor: Kentaro Inui, Sho Yokoi
April 2019-
Sept. 2022, Sendai
Tohoku University
B.E. in Computer Science
Advisor: Kentaro Inui, Sho Yokoi
Early Graduation
Work Experiences
October 2023-
Present, Tokyo
Research & Machine Learning Engineer Intern
Kotoba Technologies, Inc.
October 2023-
Present, Sendai
Research Assistant
Tohoku University
October 2023-
Present, Sendai
Research Assistant
Tohoku University
April 2023-
September 2023, Sendai
Teaching Assistant
Tohoku University
Information and data literacy
July 2022-
April 2023, Tokyo
Machine Learning Engineer Intern
ELYZA, Inc.
April 2019-
September 2022, Sendai
Machine Learning Software Engineer Intern
Adansons, Inc.
Working on data-centric AI platform products
Mentored two student interns from India
August 2022, Tokyo
Summer Intern
Nikkei
September 2020-
Mar. 2021,
Atlanta Online
Research Intern
Georgia Institute of Technology
Advisor: Biagio Mandracchia
Part of Nakatani RIES program
Projects
Jan. 2023-
Present
Training Japanese Large Language Models
with Massive Scale CPUs on Fugaku (GPT-Fugaku)
Main collaborators: Keisuke Sakaguchi, Rio Yokota,
Noriyuki Kojima, Jungo Kasai, Shota Sasaki, Kazuki Fujii, Shukai Nakamura, Taishi Nakamura, and more.
May 2023
Three-Points Scale Up of
Japanese Large Language Models Training
2023 1st ABCI Grand Challenge
Collaborators: Keisuke Sakaguchi, Rio Yokota, Koji Nishiguchi, Noriyuki Kojima, Jungo Kasai, Shota Sasaki, Kazuto Ando and Shukai Nakamura
Publications (Refereed)
- Hiroto Kurita*, Ikumi Ito*, Hiroaki Funayama, Shota Sasaki, Shoji Moriya, Ye Mengyu, Kazuma Kokuta, Ryujin Hatakeyama, Shusaku Sone and Kentaro Inui (*equal contributions). ”TohokuNLP at SemEval-2023 Task 5: Clickbait Spoiling via Simple Seq2seq Generation and Ensembling”. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval 2023), July 2023. [paper]
Publications (Non-refereed)
- 栗田 宙人, 小林 悟郎, 横井 祥, 乾 健太郎. 対照損失に基づく文エンコーダーは各単語をその情報利得で重み付ける. 第26回情報論的学習理論ワークショップ (IBIS), October–November 2023
- 栗田 宙人, 小林 悟郎, 横井 祥, 乾 健太郎. 対照学習に基づく文埋込は各単語をその情報利得で重み付ける. NLP若手の会 第18回シンポジウム (YANS), August 2023.
- 栗田 宙人, 小林 悟郎, 横井 祥, 乾 健太郎. BERTを用いた文埋め込みモデルによる単語の暗黙的な重み付け. 言語処理学会第29回年次大会, March 2023.
- 栗田宙人, 小林悟郎, 横井祥, 乾健太郎. BERTを用いた文埋め込みモデルによる単語の暗黙的な重み付け. 第17回NLP若手の会 シンポジウム (YANS), August 2022.
Awards
- Best Paper Nominee at SemEval, 2023
- 第17回NLP若手の会 シンポジウム (YANS) PKSHA Technology賞 / Sponsorship Award in the 17th YANS Symposium, 2020. (1/68=1.5%)
- 第17回NLP若手の会 シンポジウム (YANS) 奨励賞 / Encouragement Award in the 17th YANS Symposium, 2020 (10/68=14.7%)
- Nakatani RIES JP Fellow, 2020. (fully funded scholarship for research internship in the U.S.)
Talks
- Hiroto Kurita. Recent Trends of Large Language Models. WBA Tohoku Summer Meetup (Invited Talk). July 2023.
Misc.
- TOEFL iBT 98/120pts 2021
- TOEIC L&R 960/990pts 2021