CBQ: Cross-Block Quantization for Large Language Models

TL;DR

CBQ introduces a novel cross-block quantization method for large language models that effectively handles outliers and dependencies across multiple blocks, significantly improving low-bit quantization accuracy and efficiency.

Contribution

The paper proposes CBQ, a cross-block reconstruction-based PTQ method with a coarse-to-fine preprocessing and adaptive LoRA-Rounding, enabling better low-bit quantization of LLMs.

Findings

CBQ outperforms existing methods on various LLMs and datasets.

CBQ quantizes the LLAMA1-65B model to 4 bits in 4.3 hours on a single GPU.

CBQ achieves a favorable balance between model performance and quantization efficiency.

Abstract

Post-training quantization (PTQ) has played a key role in compressing large language models (LLMs) with ultra-low costs. However, existing PTQ methods only focus on handling the outliers within one layer or one block, which ignores the dependency of blocks and leads to severe performance degradation in low-bit settings. In this paper, we propose CBQ, a cross-block reconstruction-based PTQ method for LLMs. CBQ employs a cross-block dependency using a homologous reconstruction scheme, establishing long-range dependencies across multiple blocks to minimize error accumulation. Furthermore, CBQ incorporates a coarse-to-fine preprocessing (CFP) strategy for suppressing weight and activation outliers, coupled with an adaptive LoRA-Rounding technique for precise weight quantization. These innovations enable CBQ to not only handle extreme outliers effectively but also improve overall quantization accuracy. Extensive experiments show that CBQ achieves superior low-bit quantization (W4A4, W4A8, W2A16) and outperforms existing state-of-the-art methods across various LLMs and datasets. Notably, CBQ quantizes the 4-bit LLAMA1-65B model within only 4.3 hours on a single GPU, achieving a commendable tradeoff between performance and quantization efficiency.