Insight Gained from Migrating a Machine Learning Model to Intelligence Processing Units

arXiv:2404.10730v1 Announce Type: new
Abstract: The discoveries in this paper show that Intelligence Processing Units (IPUs) offer a viable accelerator alternative to GPUs for machine learning (ML) applications within the fields of materials science and battery research. We investigate the process of migrating a model from GPU to IPU and explore several optimization techniques, including pipelining and gradient accumulation, aimed at enhancing the performance of IPU-based models. Furthermore, we have effectively migrated a specialized model to the IPU platform. This …

abstract accelerator applications arxiv battery cs.ai cs.lg discoveries fields gpu gpus insight intelligence machine machine learning machine learning model materials materials science paper process processing research science show type units