Radio: Rate-Distortion Optimization for Large Language Model Compression
Author:
Preprint
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.Keywords
All Related Versions
This publication has 10 references indexed in Scilit:
- OWQ: Outlier-Aware Weight Quantization for Efficient Fine-Tuning and Inference of Large Language ModelsProceedings of the AAAI Conference on Artificial Intelligence, 2024
- Loss aware post-training quantizationMachine Learning, 2021
- Transform Quantization for CNN CompressionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2021
- HAWQ: Hessian AWare Quantization of Neural Networks With Mixed-Precision2019
- HellaSwag: Can a Machine Really Finish Your Sentence?2019
- LQ-Nets: Learned Quantization for Highly Accurate and Compact Deep Neural Networks2018
- Distributed Optimization and Statistical Learning via the Alternating Direction Method of MultipliersFoundations and Trends in Machine Learning, 2011
- Vector Quantization and Signal Compression1992
- Least squares quantization in PCMIEEE Transactions on Information Theory, 1982
- Quantizing for minimum distortionIEEE Transactions on Information Theory, 1960