1 paper
Maolin Wang, Seyedramin Rasoulinezhad, Philip H. W. Leong +1
While integer arithmetic has been widely adopted for improved performance in deep quantized neural network inference, training remains a task primarily executed using floating poin…