My previouse research focused on the development of a retrieval-augmented language model (RAG) for the task of open-domain question answering, from self trained state space model (Mamba-2 130M & 1.3B).
Publications
Single-Pass Document Scanning for Question Answering
arXiv Preprint, 2025
TL;DR: We trained State Space Models (Mamba-2) for long-context Q&A, achieving performance comparable to GPT-4o on extremely long documents while being more computationally efficient.
@misc{cao2025singlepassdocumentscanningquestion, title={Single-Pass Document Scanning for Question Answering}, author={Weili Cao and Jianyou Wang and Qirui Zheng and Longtian Bao and Qirui Zheng and Taylor Berg-Kirkpatrick and Ramamohan Paturi and Leon Bergen}, year={2025}, eprint={2504.03101}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2504.03101}, }How Novices Use Program Visualizations to Understand Code that Manipulates Data Tables
Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1 (SIGCSE TS 25), ACM, 2025
TL;DR: A case study on how novices use program visualizations to understand code manipulating data tables, revealing the importance of visualizations in learning programming.
@inproceedings{wu2025novices, title={How Novices Use Program Visualizations to Understand Code that Manipulates Data Tables}, author={Wu, Ylesia and Zheng, Qirui and Lau, Sam}, booktitle={Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1}, pages={1267--1273}, year={2025} }