The University of Illinois at Urbana-Champaign
Year in School
REU Faculty Mentor
Research Area Interest
Biography & Research Abstract
Training large machine learning models with limited memory GPUs can be impossible. Currently most training relies on time-consuming learning-based approaches. Last semester our group worked on a system called Baechi, which adopts an algorithmic approach to the placement problem for running machine learning training graphs on a small cluster of memory-constrained devices. This algorithmic approach has already been implemented on the Tensorflow platform, and is much faster than learning-based approaches, with a limited increase in step time. My goal during the summer is to implement this algorithmic approach on the Pytorch platform and perform experiments.
Hello this is Yitao Meng, I'm a junior student in computer science (with a dual degree in math). I'm interested in big data analysis, AI, distributed system, and I'm planing to go to grad school in related area.