Tatsuki Kuribayashi
@ttk_kuribayashi
Incoming Assistant Professor at @mbzuai. Previously, Postdoc at @mbzuai, PhD at @NLPTohoku, and co-founder of @langsmith_nlp.
Starting in August, I’ll start an Assistant Professor (NLP) position in @mbzuai. I’d continue to work on interdisciplinary topics bridging NLP to fundamental linguistic/cogsci questions. I'll have a small team and look for one postdoc and many visitors! 👉 kuribayashi4.github.io
We welcome Tatsuya Hiraoka, @7a7hi, as an Affiliate Assistant Professor from July 2025! 🎉🎉 Very nice to meet you, and please help us from Abu Dhabi 😂🇦🇪
7月から客員助教として，NAIST渡辺研のお手伝いをさせていただきます🙇 日本・アブダビを繋ぐパイプのひとつとして，いろいろとコラボできると嬉しいです！ また一年ほど前に入籍してまして，妻が7月からアブダビに来てくれています． 人生楽しみつつ頑張りますので，引き続きよろしくお願いします！
8月から @mbzuai にて助教を務めることになりました。 引き続き（NLPと言語学を橋渡しできるような）興味深い仕事ができればと思います。 小さなチームも持ち、ポスドク・ビジター探しております。日本との共同研究も強固にしたく、今後ともよろしくお願いいたします！ 👉 kuribayashi4.github.io
📝 Our #ACL2025 paper is now on arXiv! "Information Locality as an Inductive Bias for Neural Language Models" We quantify how local predictability of a language affects the learnability by neural LMs using our metric, m-local entropy. paper: arxiv.org/abs/2506.05136
Excited to share a new preprint w/ @Michael_Lepori & @meanwhileina! A dominant approach in AI/cogsci uses *outputs* from AI models (eg logprobs) to predict human behavior. But how does model *processing* (across a forward pass) relate to human real-time processing? 👇 (1/12)
🚨ATTN Natural Stories users!🚨 We found a misalignment in the self-paced reading times. Everything is off by one position. In the released dataset, the SPR RTs for the word at index t are actually for index t+1. If you are using the dataset, please use the realigned data.
We’re flying to Osaka for #EXPO2025. Our first workshop is "AI for Arts and Culture: Unleashing creativity with an AI tool" 🎶 We will showcase how Audiomatic, a startup born at MBZUAI, is changing how music and Al sound together. Join us for a live demo and interactive music…
I'll also help with a demo presentation of the AI safety work today (April 30th, 11:00-12:30), led by my colleague @haonanlp!
We are excited to announce Libra-Leaderboard: The first LLM leaderboard dedicated to balancing safety and capability in LLMs. As AI advances, ensuring its safety becomes more critical than ever. By prioritizing safety measurement, we aim to inspire the AI community to make safety…
I'm attending #NAACL2025 to organize the CMCL workshop in person (May 3rd, Santa Ana room) w/ @byungdoh! We received many insightful cognitively motivated NLP works this year, too. Stay tuned! Program: cmclorg.github.io/program
Now I humbly request your participation May 3 in Santa Ana @naaclmeeting; @ttk_kuribayashi and I will be there
Two papers at NAACL 2025. See you in ABQ🙌 1) Repetition Neurons: How Do Language Models Produce Repetitions? aclanthology.org/2025.naacl-sho… 2) The Geometry of Numerical Reasoning: Language Models Compare Numeric Properties in Linear Subspaces aclanthology.org/2025.naacl-sho…
Big news!📢 We've launched our first undergraduate program: Bachelor of Science in AI. This multi-disciplinary program combines core AI expertise with entrepreneurship and real-world industry experience—empowering our students to #BuildWhatsNext. Applications for Fall 2025 are…
MBZUAIでインターン的なものありませんかね、という雑談から実現しました。私自身が滞在費を提供するのは厳しいのですが、乾さん、Timさんをはじめ、私から周りを巻き込むことには協力できるので、声をかけていただければ幸いです（また学振など予算があれば数ヶ月の滞在はいつでも歓迎だと思います）
はてなブログに投稿しました MBZUAI（アブダビ）に滞在しました - 草茫茫 weeeeddie.hatenablog.com/entry/2025/03/… #はてなブログ
Abstract submissions for the 1st Computational Psycholinguistics Meeting is now open! cpl2025.sites.uu.nl
We are looking for emergency reviewers for the CMCL workshop! Send me a message if you would like to volunteer! @CMCL_workshop @naaclmeeting
今週末から約3週間日本にいます。言語処理学会@長崎にも行きます！（For MBZUAI people, I'll be in Japan to attend a conference etc. and will return in early April. I'll still be very active in Slack/email/Zoom）
🔬Our new preprint in cognitive modeling & LM interpretability: arxiv.org/abs/2502.01615 Why does LLM surprisal seemingly deviate from some human measures, e.g., reading times? Because fast human-like sentence processing is captured in LLMs' early layers, not in the final layer.
このチームはなんというか青春のようでした。おめでとうございます！
大変光栄なことに育志賞をいただきました。 約6年間、惜しみなくご指導くださった横井さん@sho_yokoi、栗林さん@ttk_kuribayashi、乾さん@inuikentaro、Tohoku NLP @tohoku_nlpの皆さんに心よりお礼申し上げます。 言語処理学会、YANSなどで議論・コメントしてくださった皆様にも深く感謝いたします。
The deadline for CMCL 2025 is approaching (Feb 16th)! We're looking forward to seeing your insightful NLP×cogsci work! Web: cmclorg.github.io Previous edition: aclanthology.org/volumes/2024.c…
📢Time is ticking! Less than ONE month left to submit your work to CMCL 2025 This year, we’re excited to be co-located with NAACL in New Mexico on May 3 or 4. Don’t miss out! 📝Deadline: 𝐅𝐞𝐛𝐫𝐮𝐚𝐫𝐲 𝟏𝟔 👉Full details here:cmclorg.github.io/CfP @naaclmeeting @naacl