For inference and edge computing applications, where workloads need to run efficiently on CPUs rather than dedicated accelerators, a model …
http://dlvr.it/TVSckx