arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2410.05085cs.CL
收藏

大语言模型随机性的解释敏感性:以新闻文本分类为例

Explanation sensitivity to the randomness of large language models: the case of journalistic text classification

UCLouvain(鲁汶大学);ICTEAM Institute(ICTEAM研究所);ILC Institute(ILC研究所)

Jeremie Bogaert, Marie-Catherine de Marneffe, Antonin Descampe, Louis Escouflaire, Cedrick Fairon, Francois-Xavier Standaert

AI总结:

本研究以法语新闻文本分类任务为例,发现不同随机种子训练的CamemBERT模型准确率相近但解释差异大,提出需表征解释的统计分布,并验证融合大模型解释特征可提升简单文本特征模型的准确率-可解释性权衡效果。

摘要:

大语言模型(LLMs)在多项自然语言处理任务中表现出色,但也带来了可解释性挑战。本文研究了大语言模型训练过程中的随机因素对其预测结果可解释性的影响。我们以法语观点性新闻文本分类任务为研究场景开展实验。通过微调后的CamemBERT模型与基于相关性传播的解释方法,我们发现使用不同随机种子训练出的模型准确率相近,但解释结果存在差异。因此我们提出,要实现大语言模型的可解释性,需要对解释的统计分布进行表征。随后我们探索了一种基于文本特征的更简单模型,该模型的解释结果稳定,但准确率较低,体现了准确率与可解释性之间的不同权衡。我们还证明,通过引入从CamemBERT的解释中提取的特征,可以提升该简单模型的性能。最后我们讨论了研究结果带来的新研究方向,尤其是训练随机性导致敏感性的成因问题。

英文摘要:

Large language models (LLMs) perform very well in several natural language processing tasks but raise explainability challenges. In this paper, we examine the effect of random elements in the training of LLMs on the explainability of their predictions. We do so on a task of opinionated journalistic text classification in French. Using a fine-tuned CamemBERT model and an explanation method based on relevance propagation, we find that training with different random seeds produces models with similar accuracy but variable explanations. We therefore claim that characterizing the explanations' statistical distribution is needed for the explainability of LLMs. We then explore a simpler model based on textual features which offers stable explanations but is less accurate. Hence, this simpler model corresponds to a different tradeoff between accuracy and explainability. We show that it can be improved by inserting features derived from CamemBERT's explanations. We finally discuss new research directions suggested by our results, in particular regarding the origin of the sensitivity observed in the training randomness.

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。

↑