An 8.9–71.3 TOPS/W Deep Learning Accelerator for Arbitrarily Quantized Neural Networks | IEEE Journals & Magazine | IEEE Xplore