QA-LoRA: Quantization-Aware Low-Rank Adaptation of Large Language Models

TL;DR

QA-LoRA introduces a quantization-aware low-rank adaptation method that enables efficient fine-tuning of large language models with reduced computational resources while maintaining accuracy.

Contribution

It presents a novel algorithm combining quantization and low-rank adaptation, improving efficiency and ease of implementation for large language model fine-tuning.

Findings

Effective in reducing memory and time during fine-tuning

Maintains accuracy after quantization and adaptation

Applicable to LLaMA and LLaMA2 models across various tasks

Abstract

Recently years have witnessed a rapid development of large language models (LLMs). Despite the strong ability in many language-understanding tasks, the heavy computational burden largely restricts the application of LLMs especially when one needs to deploy them onto edge devices. In this paper, we propose a quantization-aware low-rank adaptation (QA-LoRA) algorithm. The motivation lies in the imbalanced degrees of freedom of quantization and adaptation, and the solution is to use group-wise operators which increase the degree of freedom of quantization meanwhile decreasing that of adaptation. QA-LoRA is easily implemented with a few lines of code, and it equips the original LoRA with two-fold abilities: (i) during fine-tuning, the LLM's weights are quantized (e.g., into INT4) to reduce time and memory usage; (ii) after fine-tuning, the LLM and auxiliary weights are naturally integrated into a quantized model without loss of accuracy. We apply QA-LoRA to the LLaMA and LLaMA2 model families and validate its effectiveness in different fine-tuning datasets and downstream scenarios. Code will be made available at https://github.com/yuhuixu1993/qa-lora.