Yulu Qin（秦瑜璐）

Hi, I'm currently a second-year PhD Student at the Boston University advised by Prof. Najoung Kim. Previously, I worked as an assistant research scientist in the Human & Machine Intelligence Lab at NYU Center for Data Science, advised by Prof. Brenden Lake, where my research focuses on the intersection of cognitive science, linguistics and machine learning.

I received my Master's in Computer Science from NYU Courant. For my undergrad, I spent two years at Nankai University (Tianjin, China) and completed my B.A. in Sociology and Computer Information Science at The Ohio State University (Columbus, Ohio).

Email  /  scholar  /  cv  /  Github

Research

I am interested in language acquisition and its relevance to deep learning models, particularly in how human language acquisition can inform these models. My past projects include interpreting the inner syntactic and semantic representations of Large Language Models (LLMs).

We designed a taxonomy-informed GQA benchmark and found that vision-language models outperform text-only models; vision training improves how models use taxonomic information without fundamentally changing the underlying knowledge.

* and † indicate equal contribution

We designed a benchmark to test whether coding agents can autonomously implement extensions to existing AI research.

* and † indicate equal contribution

We trained LSTM and Transformer-based models on single-child-directed speech to examine what syntactic and semantic knowledge they can learn and how well they generalize.

We conducted a comprehensive analysis to examine BERT's inner workings and learned semantic/syntactic representations via probing, visualization and subject-verb agreement tests.

Big thank you to Jon Barron for the source code of this website.