Radio: Rate-Distortion Optimization for Large Language Model Compression

TL;DR

This paper introduces a rate-distortion optimization approach for compressing large language models, enabling flexible post-training quantization to reduce size and resource usage while maintaining accuracy.

Contribution

It establishes a theoretical foundation for LLM quantization using rate-distortion theory and proposes a scalable, flexible quantization method for very large models.

Findings

Scales to models with hundreds of billions of parameters

Allows user-specified trade-offs between model size and accuracy

Provides a theoretically grounded quantization technique

Abstract

In recent years, the compression of large language models (LLMs) has emerged as a key problem in facilitating LLM deployment on resource-limited devices, reducing compute costs, and mitigating the environmental footprint due to large-scale AI infrastructure. Here, we establish the foundations of LLM quantization from a rate-distortion theory perspective and propose a quantization technique based on simple rate-distortion optimization. Our technique scales to models containing hundreds of billions of weight parameters and offers users the flexibility to compress models, post-training, to a model size or accuracy specified by the user.