Join us for an insightful session on Entity Resolution (ER) with Zeyu Zhang, a third-year Ph.D. student at the University of Amsterdam’s! He will dig into techniques to solve one of the long-standing problem of data integration by using pre-trained language models.

Title: Practices of Using Small and Large Language Models for Entity Resolution

Time: 04/28(Mon) 13:00-14:00 (UTC+01:00)London

Location: Online

Speaker: Zeyu Zhang | University of Amsterdam

External: https://app.huawei.com/wmeeting/join/95886293/MipIlMdSMCMSGacGvUoquknvJnJSqHfS0

Meeting ID: 95886293

Passcode: 505536

Registration: https://www.smartsurvey.co.uk/s/3N8U7J/

Abstract:

While Large Language Models (LLMs) excel at knowledge reasoning tasks, they face significant challenges in formal mathematical theorem proving due to data scarcity and strict logical precision requirements. This talk introduces the DeepSeek-Prover series, highlighting how automated dataset construction and reasoning annotations have effectively transferred knowledge from data-rich to data-scarce domains, achieving state-of-the-art results in formal proofs. Additionally, formal theorem proving will be discussed as an ideal benchmark for evaluating rigorous reasoning capabilities of language models.

Bio:

Zeyu Zhang is a third-year Ph.D. student at the INDElab, University of Amsterdam. He finished his bachelor and master study from the Harbin Institute of Technology (HIT) and the Eindhoven University of Technology (TU/e), respectively. Zeyu’s research focuses on tabular data understanding, spanning from conventional machine learning models to large language models.

【BAYES COFFEE HOUSE TECH TALK SERIES】Practices of Using Small and Large Language Models for Entity Resolution / Huawei-Edinburgh Joint Lab by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0