The proposed hardware-aware latency pruning (HALP) paradigm. Considering both performance and latency contributions, HALP formulates global structural pruning as a global resource allocation problem.
BibTex:
Before browse our site, please accept our cookies policy
