Did the newest F+ TPU model increase overhead? My Coral inference speed went from 6.1 ms to 7.4 ms.
void phoenix
Did the newest F+ TPU model increase overhead? My Coral inference speed went from 6.1 ms to 7.4 ms.
drowsy cradle
yes, the new labels added a slight increase in inference speed, still much faster than the default coral model
void phoenix
OK ... not an issue just looking for an explanation