Knowledge boosting during low-latency inference

Vidya Srinivas, Malek Itani, Tuochao Chen +3

Models for low-latency, streaming applications could benefit from the knowledge capacity of larger models, but edge devices cannot run these models due to resource constraints. A p…