BitDistiller: Unleashing the Potential of Sub-4-Bit LLMs via Self-Distillation

TL;DR

BitDistiller is a novel framework combining quantization-aware training and self-distillation to enhance ultra-low precision (sub-4-bit) large language models, achieving superior performance with reduced resource requirements.

Contribution

It introduces a tailored asymmetric quantization, clipping technique, and a confidence-aware distillation objective for improved ultra-low precision LLMs.

Findings

Outperforms existing methods in 3-bit and 2-bit settings

Achieves better general language understanding and reasoning

Requires fewer data and training resources

Abstract

The upscaling of Large Language Models (LLMs) has yielded impressive advances in natural language processing, yet it also poses significant deployment challenges. Weight quantization has emerged as a widely embraced solution to reduce memory and computational demands. This paper introduces BitDistiller, a framework that synergizes Quantization-Aware Training (QAT) with Knowledge Distillation (KD) to boost the performance of LLMs at ultra-low precisions (sub-4-bit). Specifically, BitDistiller first incorporates a tailored asymmetric quantization and clipping technique to maximally preserve the fidelity of quantized weights, and then proposes a novel Confidence-Aware Kullback-Leibler Divergence (CAKLD) objective, which is employed in a self-distillation manner to enable faster convergence and superior model performance. Empirical evaluations demonstrate that BitDistiller significantly surpasses existing methods in both 3-bit and 2-bit configurations on general language understanding and complex reasoning benchmarks. Notably, BitDistiller is shown to be more cost-effective, demanding fewer data and training resources. The code is available at https://github.com/DD-DuDa/BitDistiller.