Simon Yu
You can also call me U Chi Lok (余知樂) or Simão (in Portuguese)
I am an incoming PhD student at Northeastern University, advised by Weiyan Shi. I was a M.Res. and BSc student at the University of Edinburgh, also part of the EdinburghNLP, supervised by Jeff Pan. I am also fortuantely work closely with Pasquale Minervini and Jacques Fleuriot. I am part of the Cohere For AI community with working with Marzieh Fadaee. I have done research about alignment, instruction tuning and safety in retrieval.
My research interests lie in two main directions: i) LM Alignment: Enhancing alignment of language models for general performance and safety purposes, as well improving the alignment of language models for other aspects, such as multilingual capabilities and code generation. ii) Efficiency: making language models more efficient in both training and inference.
One of the most influential lesson to me is from The Bitter Lesson. The idea is not just limited to AI but can be applied to any choice in life. Always choose the path with greater long-term benefits, even if it seems hard or impossible.
news
Nov 16, 2024 | New! I will be presenting our paper related to Is Multilingualism solved? at EMNLP 2024. |
---|---|
Oct 07, 2024 | New! I will be presenting at the COLM-2024, see you in UPenn! |
Sep 01, 2024 | New! Started as a PhD student at Northeastern University, advised by Prof. Weiyan Shi. |
Jul 10, 2024 | New! Our paper on Retrieval Safety is accepted by COLM-2024! |
Nov 10, 2023 | New! Started Colloboration with Liangyu Chen@NTU, Marzieh Fadaee@Cohere and Sara Ahmadian@Google in Efficient Data Selection. |
selected publications
- Preprint
- EMNLPIn The Conference on Empirical Methods in Natural Language Processing, 2024
- Arxiv
- COLMIn Proceedings of the Conference on Language Modeling (COLM), 2024
Acknowledgement
Since I began my research, I have met many intelligent, disciplined, and wonderful peers to work with, including (but not limited to) Andrej Jovanovic@Cambridge, Hanxu Hu@UZH, Chenmien Tan@Edinburgh, Pinzhen Chen@Edinburgh, Yijun Yang@Edinburgh and Liangyu Chen@Stanford. I have truly learned a lot from them, and I enjoyed all the discussion we had.