Berkeley AI Research

Starting December of 2021, I had the fortune of working with Ruiqi Zhong on problems in NLP related to scalable insights—using language models to help discover a ground truth where humans cannot.

I originally contributed to a project related to slant in news articles, where slant is defined by retweets from congresspeople. I fine-tuned various langauge models to better predict slant in each article.

Since May of 2022, I’ve worked on extending an existing research project about describing the differences between two distributions of text. Some contributions I have made include:

• Compiling an original benchmark of 50 interdisciplinary datasets, which has been presented to positive reception at conferences.
• Implementing multiple improvements to increase the computational efficiency of the system and improve the quality of the insights.