Paper Presentation

Data Selection via Optimal Control for Language Models

- By Yuxian Gu, PhD student, Tsinghua University

Read the Paper

In the ever-evolving landscape of natural language processing, the performance of language models heavily relies on the quality and relevance of the training data. Join us for an engaging webinar featuring Yuxian Gu, who will present his groundbreaking research titled 'Data Selection via Optimal Control for Language Models'. This session will delve into innovative methods that enhance the efficiency and effectiveness of language models by optimizing data selection processes.

In this presentation, Yuxian Gu explores the concept of data selection through optimal control, proposing a framework that improves model performance and reduces training time and resource consumption. This work aims to bridge the gap between theoretical foundations and practical applications in the field, offering insights into how optimal control can be effectively implemented for data selection in language models.


Meet our Speaker:

Yuxian Gu

Yuxian Gu is a 4th-year Ph.D. student in the Conversational AI Group at the Department of Computer Science and Technology, Tsinghua University, under the supervision of Prof. Minlie Huang. Previously, he was an intern at Microsoft Research Asia. His research interests focus on developing efficient methods for the entire life cycle of language models, including pre-training, downstream adaptation, and inference.