Generalized Latency Performance Estimation for Once-For-All Neural Architecture Search

2 code implementations • 4 Jan 2021 • Muhtadyuzzaman Syed, Arvind Akpuram Srinivasan

We find that the performance estimation strategy for OFA's search severely lacks generalizability of different hardware deployment platforms due to single hardware latency lookup tables that require significant amount of time and manual effort to build beforehand.

Neural Architecture Search