I am an MRes student supervised by Dr. Edoardo Maria Ponti in University of Edinburgh. I was a research assistant in WestlakeNLP supervised by Prof. Yue Zhang. Before that, I was a MSc student in Computer Science at the University of Edinburgh, and my thesis is supervised by Prof. Frank Keller.
My current research interest is about alignment and grounding current Large Language Models across languages and cultures. My long term research goal is to make contribution to build general-purpose artificial intelligence to interact with people from different cultural backgrounds better. Specifically, the generation of language models should be controlled, and the knowledge of the language models should be grounded to real world in diverse cultures and values. I have research experience about controlled text generation, multi-modal learning, and cross-lingual transfer. You can view my CV.
My name in Chinese is 胡涵旭.
- MRes in ILCC, University of Edinburgh, 2023.9 - 2024.9 (expected)
- MSc in Computer Science, University of Edinburgh, 2021.9 - 2022.11, Distinction Degree
- Bachelor in Measurement, Control Technology, and Instrument, Nanjing University of Science and Technology, 2017.9 - 2021.6, rank 2/50
Honors & Awards
- (2022) IBM Third Place Prize for Best Project in MLP Course (Top%2)
- (2019) First Class School Scholarship (awarded to 5%)
- Research Assistant at WestlakeNLP, supervised by Prof. Yue Zhang, 2022 Oct - Present.
- Research Internship at PCA Lab, supervised by Dr. Xiushen Wei, 2021 May - 2021 Sept.
Publication and Pre-prints
Chain-of-Symbol Prompting Elicits Planning in Large Language Models
Hanxu Hu*, Hongyuan Lu*, Huajian Zhang, Wai Lam, Yue Zhang
Preprint, under review.
Meta-Learning For Vision-and-Language Cross-Lingual Transfer
Hanxu Hu and Frank Keller
3rd MRL Workshop @ EMNLP 2023
Best Paper Honorable Mention
Improving User Controlled Table-To-Text Generation Robustness
Hanxu Hu, Yunqing Liu, Zhongyi Yu and Laura Perez-Beltrachini
EACL 2023 (Findings).
A Channel Mix Method For Fine-grained Cross-modal Retreval
Yang Shen, Xuhao Sun, Hanxu Hu, Zhipeng Chen, Xiushen Wei
IEEE ICME 2022