arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.05445cs.LGcs.CL
收藏

通过信息保留实现LLM的精确LoRA微调量化

Accurate LoRA-Finetuning Quantization of LLMs via Information Retention

ETH Zürich(苏黎世联邦理工学院);Beihang University(北京航空航天大学);Bytedance AI Lab(字节跳动人工智能实验室)

Haotong Qin, Xudong Ma, Xingyu Zheng, Xiaoyang Li, Yang Zhang, Shouda Liu, Jie Luo, Xianglong Liu, Michele Magno

AI总结:

针对现有方法导致量化LLM在LoRA微调中失效的问题,提出IR-QLoRA,通过信息校准量化与信息弹性连接保留信息,在2-4位宽下显著提升LLaMA系列精度,仅增加0.31%时间开销。

摘要:

LLM的LoRA微调量化已被广泛研究,以在资源受限的硬件上部署获得准确且紧凑的LLM。然而,现有方法会导致量化后的LLM严重退化,甚至无法从LoRA的微调中受益。本文提出了一种新颖的IR-QLoRA,通过信息保留推动带LoRA的量化LLM达到高精度。所提出的IR-QLoRA主要依赖于从统一信息视角推导出的两种技术:(1)基于统计的信息校准量化,使LLM的量化参数能够准确保留原始信息;(2)基于微调的信息弹性连接,使LoRA能够利用具有多样信息的弹性表示变换。综合实验表明,在2-4位宽下,IR-QLoRA能显著提高LLaMA和LLaMA2系列的精度,例如,与最先进的方法相比,4位LLaMA-7B在MMLU上实现了1.4%的提升。这一显著的性能提升仅需额外0.31%的时间消耗,揭示了我们的IR-QLoRA令人满意的效率。我们强调IR-QLoRA具有出色的通用性,兼容各种框架(如NormalFloat和Integer量化),并带来普遍的精度提升。代码可在https://github.com/htqin/ir-qlora获取。

英文摘要:

The LoRA-finetuning quantization of LLMs has been extensively studied to obtain accurate yet compact LLMs for deployment on resource-constrained hardware. However, existing methods cause the quantized LLM to severely degrade and even fail to benefit from the finetuning of LoRA. This paper proposes a novel IR-QLoRA for pushing quantized LLMs with LoRA to be highly accurate through information retention. The proposed IR-QLoRA mainly relies on two technologies derived from the perspective of unified information: (1) statistics-based Information Calibration Quantization allows the quantized parameters of LLM to retain original information accurately; (2) finetuning-based Information Elastic Connection makes LoRA utilizes elastic representation transformation with diverse information. Comprehensive experiments show that IR-QLoRA can significantly improve accuracy across LLaMA and LLaMA2 families under 2-4 bit-widths, e.g., 4- bit LLaMA-7B achieves 1.4% improvement on MMLU compared with the state-of-the-art methods. The significant performance gain requires only a tiny 0.31% additional time consumption, revealing the satisfactory efficiency of our IR-QLoRA. We highlight that IR-QLoRA enjoys excellent versatility, compatible with various frameworks (e.g., NormalFloat and Integer quantization) and brings general accuracy gains. The code is available at https://github.com/htqin/ir-qlora.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑