Kazuki Yano

Ph.D. student @ Tohoku NLP Group.

I am a second-year Ph.D. student in Information Science at Fundamental AI lab (member of Tohoku NLP Group) at Tohoku University, advised by Prof. Jun Suzuki. I am also a Research Fellow in the JST BOOST Program.

My research focuses on pre-training large language models and building efficient multimodal LLMs, with an interest in how training recipes shape model capability and scalability. In summer 2026, I will work as an Amazon Japan Science Fellow on AI agents.

yano.kazuki [at] dc.tohoku.ac.jp

News

June 2026 I will start as an Amazon Japan Science Fellow, working on AI agents.
June 2026 My first author paper "Multimodal Depth Up-scaling" accepted at INTERSPEECH 2026.
Jan 2026 My first author paper "WSO" accepted at ICLR 2026.
Jan 2026 Two papers accepted at EACL 2026.
Oct 2025 Two papers including my first author paper accepted at COLM 2025.
Apr 2025 My first author paper "STEP" accepted at NAACL 2025.