arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.01165cs.CLcs.AI
收藏

STAR:用于大语言模型数据高效微调的约束LoRA与动态主动学习

STAR: Constraint LoRA with Dynamic Active Learning for Data-Efficient Fine-Tuning of Large Language Models

School of Computer Science and Engineering, Southeast University(东南大学计算机科学与工程学院);SANY Group Co., Ltd.(三一集团有限公司)

Linhai Zhang, Jialong Wu, Deyu Zhou, Guoqiang Xu

AI总结:

本文提出STAR方法,通过动态不确定性度量与正则化及蒙特卡洛dropout,有效结合主动学习与LoRA,解决LLM数据高效微调中的不确定性差距和校准不佳问题,并在三个复杂推理任务上超越现有基线。

摘要:

尽管大语言模型(LLMs)通过提示方法展示了强大的少样本学习能力,但对于复杂推理任务,监督训练仍然是必要的。由于其庞大的参数量和内存消耗,针对LLMs已提出了参数高效微调(PEFT)方法和内存高效微调方法。然而,大规模标注数据消耗问题,即数据高效微调的目标,仍未得到探索。一个显而易见的途径是将PEFT方法与主动学习相结合。然而,实验结果表明,这种结合并非易事,且会产生较差的结果。通过探针实验,这一观察可能由两个主要原因解释:不确定性差距和模型校准不佳。因此,在本文中,我们提出了一种新方法,以有效整合基于不确定性的主动学习和LoRA。具体而言,针对不确定性差距,我们引入了一种动态不确定性度量,该度量在主动学习迭代过程中结合了基础模型的不确定性和完整模型的不确定性。针对模型校准不佳,我们在LoRA训练期间引入正则化方法以防止模型过度自信,并采用蒙特卡洛dropout机制来增强不确定性估计。实验结果表明,所提出的方法在三个复杂推理任务上优于现有基线模型。

英文摘要:

Though Large Language Models (LLMs) have demonstrated the powerful capabilities of few-shot learning through prompting methods, supervised training is still necessary for complex reasoning tasks. Because of their extensive parameters and memory consumption, both Parameter-Efficient Fine-Tuning (PEFT) methods and Memory-Efficient Fine-Tuning methods have been proposed for LLMs. Nevertheless, the issue of large annotated data consumption, the aim of Data-Efficient Fine-Tuning, remains unexplored. One obvious way is to combine the PEFT method with active learning. However, the experimental results show that such a combination is not trivial and yields inferior results. Through probe experiments, such observation might be explained by two main reasons: uncertainty gap and poor model calibration. Therefore, in this paper, we propose a novel approach to effectively integrate uncertainty-based active learning and LoRA. Specifically, for the uncertainty gap, we introduce a dynamic uncertainty measurement that combines the uncertainty of the base model and the uncertainty of the full model during the iteration of active learning. For poor model calibration, we incorporate the regularization method during LoRA training to keep the model from being over-confident, and the Monte-Carlo dropout mechanism is employed to enhance the uncertainty estimation. Experimental results show that the proposed approach outperforms existing baseline models on three complex reasoning tasks.

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。

↑