Balancing Efficiency and Flexibility for DNN Acceleration

NeurIPS 2019