Accel-NASBench: Sustainable Benchmarking for Accelerator-Aware NAS

arXiv:2404.08005v1 Announce Type: new
Abstract: One of the primary challenges impeding the progress of Neural Architecture Search (NAS) is its extensive reliance on exorbitant computational resources. NAS benchmarks aim to simulate runs of NAS experiments at zero cost, remediating the need for extensive compute. However, existing NAS benchmarks use synthetic datasets and model proxies that make simplified assumptions about the characteristics of these datasets and models, leading to unrealistic evaluations. We present a technique that allows searching for training proxies …

abstract accel accelerator aim architecture arxiv benchmarking benchmarks challenges computational compute cost cs.lg datasets eess.iv however nas neural architecture search progress proxies reliance resources search sustainable synthetic type