SliM-LLM: Salience-Driven Mixed-Precision Quantization for Large Language Models

TL;DR

SliM-LLM introduces a salience-driven mixed-precision quantization framework for large language models, improving accuracy and efficiency by adaptively allocating bit-widths based on weight importance, with demonstrated superior performance at low bit-widths.

Contribution

The paper presents a novel salience-driven mixed-precision quantization method that adaptively assigns bit-widths and calibrates quantizers based on weight importance, enhancing LLM compression without sacrificing speed.

Findings

2-bit LLaMA-7B reduces memory by 6x

Decreases perplexity by 48% over state-of-the-art PTQ

Maintains GPU inference speed

Abstract

Post-training quantization (PTQ) is an effective technique for compressing large language models (LLMs). However, while uniform-precision quantization is computationally efficient, it often compromises model performance. To address this, we propose SliM-LLM, a salience-driven mixed-precision quantization framework that allocates bit-widths at the group-wise. Our approach leverages the observation that important weights follow a structured distribution and introduces two key components: \textbf{1)} \textit{Salience-Determined Bit Allocation} adaptively assigns bit-widths to groups within each layer based on their salience; and \textbf{2)} \textit{Salience-Weighted Quantizer Calibration} optimizes quantizer parameters by incorporating element-level salience. With its structured partitioning, SliM-LLM provides a hardware-friendly solution that matches the efficiency of uniform quantization methods while improving accuracy. Experiments show that SliM-LLM achieves superior performance across various LLMs at low bit-widths. For example, a 2-bit quantized LLaMA-7B model reduces memory usage by nearly 6x compared to the floating-point baseline, decreases perplexity by 48\% compared to state-of-the-art gradient-free PTQ methods, and maintains GPU inference speed. Additionally, the extended version, SliM-LLM$^+$, which incorporates gradient-based quantization, further reduces perplexity by 35.1\%. Our code is available at https://github.com/Aaronhuang-778/SliM-LLM