QQQ: Quality Quattuor-Bit Quantization for Large Language Models

TL;DR

QQQ introduces a 4-bit weights and 8-bit activations quantization method for large language models that maintains performance while significantly accelerating inference speed through specialized kernel engineering.

Contribution

The paper proposes QQQ, a novel quantization technique with adaptive smoothing and Hessian-based compensation, plus optimized GEMM kernels, to improve speed and performance of LLM inference.

Findings

Achieves up to 2.24× speedup over FP16

Maintains performance comparable to state-of-the-art methods

Significantly accelerates W4A8 and W8A8 inference stages

Abstract

Quantization is a proven effective method for compressing large language models. Although popular techniques like W8A8 and W4A16 effectively maintain model performance, they often fail to concurrently speed up the prefill and decoding stages of inference. W4A8 is a promising strategy to accelerate both of them while usually leads to a significant performance degradation. To address these issues, we present QQQ, a Quality Quattuor-bit Quantization method with 4-bit weights and 8-bit activations. QQQ employs adaptive smoothing and Hessian-based compensation, significantly enhancing the performance of quantized models without extensive training. Furthermore, we meticulously engineer W4A8 GEMM kernels to increase inference speed. Our specialized per-channel W4A8 GEMM and per-group W4A8 GEMM achieve impressive speed increases of 3.67$\times$ and 3.29 $\times$ over FP16 GEMM. Our extensive experiments show that QQQ achieves performance on par with existing state-of-the-art LLM quantization methods while significantly accelerating inference, achieving speed boosts up to 2.24 $\times$, 2.10$\times$, and 1.25$\times$ compared to FP16, W8A8, and W4A16, respectively.