Research and Publications
My research in the past has focused on both interpretability in language models as well as multilingual capabilities of information retrieval systems. Currently I am interested in the interplay between semantics and contexts and how they manifest themselves in embeddings.
|
|
What does the Knowledge Neuron Thesis have to do with Knowledge?
Jingcheng Niu,
Andrew Liu,
Zining Zhu,
Gerald Penn
The Twelfth International Conference on Learning Representations (ICLR), 2024 (Spotlight)
Current interpretability techniques identify patterns that are interpretable both syntactically and semantically; however, these patterns do not constitute knowledge.
|
|
The Impact of Incidental Multilingual Text on Cross-Lingual Transferring in Monolingual Retrieval
Andrew Liu,
Edward Xu,
Xinyu Zhang,
Jimmy Lin
The 47th European Conference on Information Retrieval (ECIR), 2025
Despite the large amounts of incidental multilingual texts in training data, they play a very minor role in cross-lingual transfer capabilities.
|
|
Similarity, Transformation and the Newly Found Invariance of Influence Functions
Andrew Liu,
Gerald Penn
The Eight Meeting of the Society for Computation in Linguistics (SCIL), 2025
Influence functions have emerged as a feasible way to assign blame to datapoints. Our findings show influence scores may encode semantics in the same strange way that Sentence-BERT vectors do, and possibly exhibit a higher degree of robustness to semantics-preserving variance.
|
Teaching Experience
(Fall 2024) TA for CSC110: Foundations of Computer Science @ University of Toronto
(Winter 2025) TA for CSC384: Introduction to Artificial Intelligence @ University of Toronto
|
|