Tag
Introduces Entropy Constrained Adaptive Stochastic Quantization (ECASQ), a method that jointly optimizes quantization values to minimize mean squared error under an entropy budget and unbiasedness constraint for efficient compression in machine learning workloads.
This paper introduces inner product aware quantization methods that preserve inner products with unseen vectors, developing fast and adaptive algorithms with provable guarantees, achieving 2-10x speedup over prior ASQ methods.