arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.07470cs.CL
收藏

突破LLM在文本分类任务中的能力极限

Pushing The Limit of LLM Capacity for Text Classification

The Hong Kong Polytechnic University(香港理工大学);Zhengzhou University of Light Industry(郑州轻工业大学);Copenhagen University(哥本哈根大学);Halmstad University(哈尔姆斯塔德大学);The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳))

Yazhou Zhang, Mengyao Wang, Chenyu Ren, Qiuchi Li, Prayag Tiwari, Benyou Wang, Jing Qin

AI总结:

针对LLM时代文本分类研究价值受质疑的问题,提出自适应提升框架RGPT,通过循环集成基础学习器构建专用LLM,在四个基准上平均超越15个SOTA模型1.36%并超越人类分类水平。

摘要:

由于大型语言模型(LLMs)在众多下游NLP任务中展现出卓越的效能,文本分类未来研究的价值面临挑战与不确定性。在这个任务边界逐渐模糊的开放式语言建模时代,一个紧迫的问题浮现:在充分利用LLMs的优势下,我们在文本分类方面是否取得了重大进展?为回答此问题,我们提出RGPT,这是一种自适应提升框架,旨在通过循环集成一组强大的基础学习器,构建专门的文本分类LLM。这些基础学习器通过自适应调整训练样本分布并利用其迭代微调LLMs来构建。随后,通过循环整合来自先前学习器的历史预测,将这些基础学习器集成为专门的文本分类LLM。通过全面的实证比较,我们表明RGPT在四个基准上平均以1.36%的优势显著优于8个SOTA PLMs和7个SOTA LLMs。进一步的评估实验表明,RGPT明显超越了人类的分类水平。

英文摘要:

The value of text classification's future research has encountered challenges and uncertainties, due to the extraordinary efficacy demonstrated by large language models (LLMs) across numerous downstream NLP tasks. In this era of open-ended language modeling, where task boundaries are gradually fading, an urgent question emerges: have we made significant advances in text classification under the full benefit of LLMs? To answer this question, we propose RGPT, an adaptive boosting framework tailored to produce a specialized text classification LLM by recurrently ensembling a pool of strong base learners. The base learners are constructed by adaptively adjusting the distribution of training samples and iteratively fine-tuning LLMs with them. Such base learners are then ensembled to be a specialized text classification LLM, by recurrently incorporating the historical predictions from the previous learners. Through a comprehensive empirical comparison, we show that RGPT significantly outperforms 8 SOTA PLMs and 7 SOTA LLMs on four benchmarks by 1.36% on average. Further evaluation experiments show a clear surpassing of RGPT over human classification.

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。

↑