Fetching the paper…

Model compression as constrained optimization, with application to neural nets. Part II: quantization · Around