CodeQuant: Unified Clustering and Quantization for Enhanced Outlier Smoothing in Low-Precision Mixture-of-Experts

TL;DR

CodeQuant introduces a unified clustering and quantization method that effectively reduces outlier-induced errors in low-precision MoE models, leading to significant speedups and improved accuracy.

Contribution

It proposes a novel scheme combining learnable rotation and clustering to smooth outliers, enhancing low-precision deployment of large language models.

Findings

Achieves up to 4.15x speedup on hardware.

Delivers higher accuracy than existing quantization methods.

Effectively reduces quantization errors in MoE models.

Abstract

Outliers have emerged as a fundamental bottleneck in preserving accuracy for low-precision large models, particularly within Mixture-of-Experts (MoE) architectures that are increasingly central to large-scale language modeling. Under post-training quantization (PTQ), these outliers induce substantial quantization errors, leading to severe accuracy degradation. While recent rotation-based smoothing techniques alleviate the problem by redistributing outlier magnitudes, residual errors remain and continue to impede reliable low-precision deployment. In this work, we tackle this challenge by introducing \textit{CodeQuant}, a unified quantization-and-clustering scheme that contains smoothing activation outliers via learnable rotation and absorbing weight outliers into fine-tuned cluster centroids for MoE. This design reduces the influence of extreme values by fitting them within cluster centroids, thereby lowering quantization error while maintaining expressive capacity. Coupled with a dedicated kernel design for GPU and CPU, CodeQuant achieves up to $4.15\times$ speedup while delivering significantly higher accuracy than state-of-the-art quantization approaches across diverse MoE models. Our results highlight CodeQuant as a promising direction for efficient and accurate deployment of MoE-based large language models under low-precision constraints. Our code is available at https://github.com/SAI-Lab-NYU/CodeQuant.