Yu Zhao
PhD Student · School of Informatics, University of Edinburgh
I am a PhD student at the University of Edinburgh and a member of EdinburghNLP, supervised by Pasquale Minervini, Mirella Lapata, and Fazl Barez.
My research focuses on large language models: how pre-training data shapes them, how they use knowledge and long contexts, and how to make them more efficient and controllable. I am currently working on efficient and faithful vision-language model agents, and on building interpretability agents that automatically discover the mechanisms of LLMs.
2026 News
- One paper accepted at EMNLP 2026 on multimodal retrieval heads.
- Joined Tencent London as a Research Intern.
- Three papers accepted at ICML 2026 on multimodal reasoning, diffusion language models, and multi-token prediction.
Industry Experience
-
Research Intern · Tencent LondonJun 2026 – PresentLondon, England, United Kingdom · On-siteVLM Agents, Unified VLMs, Multimodal Reasoning -
Research Intern · Microsoft Research CambridgeMay 2025 – Aug 2025Cambridge, England, United Kingdom · On-siteReinforcement Learning for Multimodal Reasoning & GUI Agents -
Research Engineer · Xiaomi · Full-timeMar 2023 – Aug 2023Beijing, China · On-siteLLMs Pre-Training at Xiaomi AI Lab
Publications
* equal contribution. Citation counts from Google Scholar.
- 2026
- ICML
- ICML
- ICML
- EMNLP
- 2025
- NAACL
- NeurIPS
- NAACL
- AAAI
- COLM
- Workshop
- 2024
- ACL
- EMNLP
- arXiv
- 2022 and earlier
- EMNLP
- KDD
- NLPCC