This repository is the official implementation of "Training for Multi-resolution Inference Using Reusable Quantization Terms" published in ASPLOS 2021.
In this work, we describe a novel training approach to support inference at multiple resolutions by reusing a single set of quantization terms. This training enables a single meta multi-resolution model to select from multiple resolutions to satisfy given runtime resource constraints. The proposed scheme relies on term quantization to enable flexible bit annihilation at any position for a value in a group of values. This is in contrast to uniform quantization which always truncates the lowest-order bits.