VPTQ: Extreme Low-bit Vector Post-Training Quantization for Large Language Models
Yifei Liu, Jicheng Wen, Yang Wang, Shengyu Ye, Li Lyna Zhang, Ting Cao, Cheng Li, Mao Yang
TL;DR
This work tackles the challenge of deploying extremely low-bit weight quantization for large language models. It introduces Vector Post-Training Quantization (VPTQ), an optimization-guided vector quantization framework that quantizes weights column-by-column using independent codebooks, augmented by Residual Vector Quantization and Outlier Elimination to boost accuracy while keeping overhead low. By leveraging Second-Order Optimization and Hessian-weighted centroid initialization, VPTQ achieves state-of-the-art or near-state-of-the-art performance at 2-bit and maintains strong results at 3–4 bits across LLaMA-2, LLaMA-3, and Mistral-7B, with notable improvements in QA accuracy and perplexity, as well as 1.6–1.8x inference throughput gains. The proposed end-to-end algorithm demonstrates reduced quantization time (10.4–18.6%) and scalable deployment potential in resource-constrained settings.
Abstract
Scaling model size significantly challenges the deployment and inference of Large Language Models (LLMs). Due to the redundancy in LLM weights, recent research has focused on pushing weight-only quantization to extremely low-bit (even down to 2 bits). It reduces memory requirements, optimizes storage costs, and decreases memory bandwidth needs during inference. However, due to numerical representation limitations, traditional scalar-based weight quantization struggles to achieve such extreme low-bit. Recent research on Vector Quantization (VQ) for LLMs has demonstrated the potential for extremely low-bit model quantization by compressing vectors into indices using lookup tables. In this paper, we introduce Vector Post-Training Quantization (VPTQ) for extremely low-bit quantization of LLMs. We use Second-Order Optimization to formulate the LLM VQ problem and guide our quantization algorithm design by solving the optimization. We further refine the weights using Channel-Independent Second-Order Optimization for a granular VQ. In addition, by decomposing the optimization problem, we propose a brief and effective codebook initialization algorithm. We also extend VPTQ to support residual and outlier quantization, which enhances model accuracy and further compresses the model. Our experimental results show that VPTQ reduces model quantization perplexity by $0.01$-$0.34$ on LLaMA-2, $0.38$-$0.68$ on Mistral-7B, $4.41$-$7.34$ on LLaMA-3 over SOTA at 2-bit, with an average accuracy improvement of $0.79$-$1.5\%$ on LLaMA-2, $1\%$ on Mistral-7B, $11$-$22\%$ on LLaMA-3 on QA tasks on average. We only utilize $10.4$-$18.6\%$ of the quantization algorithm execution time, resulting in a $1.6$-$1.8\times$ increase in inference throughput compared to SOTA.