QA-LoRA: Quantization-Aware Low-Rank Adaptation of Large Language Models
QA-LoRA proposes quantization-aware low-rank adaptation that quantizes LLM weights during fine-tuning and merges adapters into quantized models without accuracy loss.
Published Sep 26, 202321 citations▲ 46 on Hugging FaceCode ★ 148arXiv ↗
Only vote on papers you've read. Sign in with GitHub to vote.
QA-LoRA's slick group-wise split delivers real INT4 fine-tuning and lossless merge, but it lacks error bars, ablation, released code, and rigorous reasoning benchmarks to fully back its deploy claims.
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.