Regular biography
Beidi Chen is an Assistant Professor, Affiliated Faculty, in the Computer Science Department at Carnegie Mellon University. Her research focuses on Systems. She is affiliated with the CIT - Electrical and Computer Engineering department. Her profile page indicates she advises Zhuoming Chen.
Scholar-generated biography
Beidi Chen is a researcher at Carnegie Mellon University specializing in Machine Learning. Her work focuses on improving the efficiency and scalability of large language models (LLMs) through innovative techniques such as streaming inference, memory-efficient training, and speculative decoding. Chen's research addresses challenges in generative inference, including reducing computational overhead and enhancing throughput. Her publications explore methods like attention sinks, gradient low-rank projection, and structured matrix representations to optimize LLM performance. She also investigates decentralized training and hardware-agnostic algorithms for large-scale deep learning systems. Her contributions aim to advance the practical deployment of LLMs in real-world applications.