Fetching the paper…

Training Compact Neural Networks with Binary Weights and Low Precision Activations · Around