Haokun Liu 刘昊坤
“be like earth”
For a more formal tone, please check out my CV.
I’m Haokun, a PhD student at the University of Toronto (and Vector Institute), advised by Colin Raffel.
I do Machine Learning / Natural Language Processing research for a living. In the good old days, it was just alchemy. Now, we have slipped into sorcery and cultism. At this pace, we will soon achieve Superintelligence, even if we make no progress on AI at all.
My research interests revolve around:
- Modularity (Like MoE, with some transfer / federated learning flavor),
- Adapting pretrained models for downstream tasks (Finetuning, Model Compression, Few-shot Learning),
- Understanding the models (Probing, Evaluation, BERTology).
Most of my previous years were in academic labs.
- Before Colin moved to Toronto, I spent two years in UNC Chapel Hill (2021 - 2023), the only place I could afford to live in a studio. A Chinese proverb says: “Easy it is to go from frugality to extravagance, but hard to go the other way around.” That is so true!
- I got my M.S. from New York Univerity (2018 - 2020), where I had a fantastic time in ML2 Lab, working closely with Sam Bowman and other group members. I am forever grateful for the opportunity, knowledge, and camaraderie. Fun fact: In 2020, among all the chaos, I became the first “Junior Research Scientist” at NYU ML2.
- I did my B.S. at Peking University (2013 - 2018) while researching in Zhi-hong Deng 's lab. I also worked with Sanja Fidler in a research internship at the University of Toronto (2017). Back then, as a particularly stubborn and inexperienced student, I completely bought into the “Believe in yourself. Don’t listen to your advisor. Never give up your arbitrary ideas.” mantra. And I managed to learn a million ways to get my project into trouble in just four years.
I occasionally have short excursions to the industry world.
- Together with some great collaborators, I had a few fun and bold projects at MIT-IBM (2023).
- I worked as a student researcher on scaling law analysis at Google Translate (2022).
- Some summarization stuff at AllenAI Semantic Scholar Team as a Predoctoral Young Investigator (2021).
You can find my list of publications at Semantic Scholar or Google Scholar.
Welcome to contact me via haokunliu412@gmail.com , if anything above interests you.