Quantized Neural Networks for Low-Precision Accumulation with Guaranteed Overflow Avoidance

Open in new window