pith. sign in

arxiv: 2302.08007 · v2 · pith:6OZUPSJXnew · submitted 2023-02-16 · 💻 cs.LG · cs.AI· cs.AR

With Shared Microexponents, A Little Shifting Goes a Long Way

classification 💻 cs.LG cs.AIcs.AR
keywords microexponentsquantizationsharedblockfloating-pointformatsincludingnarrow-precision
0
0 comments X
read the original abstract

This paper introduces Block Data Representations (BDR), a framework for exploring and evaluating a wide spectrum of narrow-precision formats for deep learning. It enables comparison of popular quantization standards, and through BDR, new formats based on shared microexponents (MX) are identified, which outperform other state-of-the-art quantization approaches, including narrow-precision floating-point and block floating-point. MX utilizes multiple levels of quantization scaling with ultra-fine scaling factors based on shared microexponents in the hardware. The effectiveness of MX is demonstrated on real-world models including large-scale generative pretraining and inferencing, and production-scale recommendation systems.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.