Belinda Li

I am a third-year PhD student at MIT CSAIL, affiliated with the language & intelligence lab @ MIT. My advisor is Jacob Andreas. I am funded by an NDSEG Fellowship and Clare Boothe Luce Graduate Fellowship [Press].

Previously, I spent a year at Facebook AI Applied Research on the AI Integrity team. There I worked on applying state-of-the-art techniques from AI and NLP research to combat integrity problems (i.e. hate speech) on Facebook. Before that, I obtained my B.S. in Computer Science at the University of Washington, where I worked with Luke Zettlemoyer. You can view more in my CV here.

If you are an undergrad interested in NLP or machine learning research, please feel free to reach out to me! You can view my contact policy here.

Publications

2022

• Quantifying Adaptability in Pre-trained Language Models with 500 Tasks
  Belinda Z. Li, Jane Yu, Madian Khabsa, Luke Zettlemoyer, Alon Halevy, Jacob Andreas
  To appear in North American Chapter of the Association for Computational Linguistics (NAACL), 2022.
  [Paper] [Code] [Show Bib | Copy Bib]
  

2021

• Quantifying Adaptability in Pre-trained Language Models with 500 Tasks
  Belinda Z. Li, Jane Yu, Madian Khabsa, Luke Zettlemoyer, Alon Halevy, Jacob Andreas
  To appear in North American Chapter of the Association for Computational Linguistics (NAACL), 2022.
  [Paper] [Code] [Show Bib | Copy Bib]
  
• Implicit Representations of Meaning in Neural Language Models
  Belinda Z. Li, Maxwell Nye, Jacob Andreas
  In Association for Computational Linguistics (ACL), 2021.
  [Paper] [Code] [Show Bib | Copy Bib]
  
• On Unifying Misinformation Detection
  Nayeon Lee, Belinda Z. Li, Sinong Wang, Pascale Fung, Hao Ma, Wen-tau Yih, and Madian Khabsa
  In North American Association for Computational Linguistics (NAACL), 2021.
  [Paper] [Show Bib | Copy Bib]
  
• On the Influence of Masking Policies in Intermediate Pre-training
  Qinyuan Ye, Belinda Z. Li, Sinong Wang, Benjamin Bolte, Hao Ma, Wen-tau Yih, Xiang Ren, and Madian Khabsa
  Accepted to Empirical Methods for Natural Language Processing (EMNLP), 2021.
  [Paper] [Show Bib | Copy Bib]
  

2020

• Studying Strategically: Learning to Mask for Closed-book QA
  Qinyuan Ye, Belinda Z. Li, Sinong Wang, Benjamin Bolte, Hao Ma, Wen-tau Yih, Xiang Ren, and Madian Khabsa
  ArXiv Preprint.
  [Paper] [Show Bib | Copy Bib]
  
• Efficient One-Pass End-to-End Entity Linking for Questions
  Belinda Z. Li, Sewon Min, Srinivasan Iyer, Yashar Mehdad, and Wen-tau Yih
  In Empirical Methods in Natural Language Processing (EMNLP), 2020.
  [Paper] [Code] [Show Bib | Copy Bib]
  
• Linformer: Self-Attention with Linear Complexity
  Sinong Wang, Belinda Z. Li, Madian Khabsa, Han Fang, and Hao Ma
  ArXiv Preprint.
  [Paper] [Show Bib | Copy Bib]
  
• Language Models as Fact Checkers?
  Nayeon Lee, Belinda Z. Li, Sinong Wang, Wen-tau Yih, Hao Ma, and Madian Khabsa
  In ACL Workshop on Fact Extraction and VERification (FEVER@ACL), 2020.
  [Paper] [Press] [Show Bib | Copy Bib]
  
• Active Learning for Coreference Resolution using Discrete Annotation
  Belinda Z. Li, Gabriel Stanovsky, and Luke Zettlemoyer
  In Annual Meeting of the Association for Computational Linguistics (ACL), 2020.
  [Paper] [Code] [Demo*] [Show Bib | Copy Bib]
  
  *Timing experiments interface

Service

Program Committee (Reviewer): ACL 2023. COLING 2022, 2020. EMNLP 2022, 2021, 2020.

Student Volunteer: ACL 2020.

Misc

Outside of work, I dance ballet and Chinese classical dance at Hengda Dance Academy and Harvard Ballet Company.

I also play the piano (though I've unfortunately been out of practice for a few years).

I was fortunate to be part of the organizing effort to form the MIT Graduate Student Union, which MIT graduate students recently voted in favor of by a 2-to-1 margin!

Noam Chomsky is my great(x3)-grand-advisor. (Belinda Li → Jacob Andreas → Dan Klein → Chris Manning → Joan Bresnan → Noam Chomsky)

Under very generous definitions, my Erdős–Bacon number is 7. (My Erdős number is 4 through several different paths. My Bacon number is 3 through appearing in a video with Bernie Sanders, who happens to have a Bacon number of 2. Though dogmatic Bacon number purist will insist only movie appearances count.) I eagerly welcome any collaborators – academic or entertainment – willing to lower that number.