arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.04714cs.CLcs.LGstat.ML
收藏

基于预训练语言模型的不确定性量化:大规模实证分析

Uncertainty Quantification with Pre-trained Language Models: A Large-Scale Empirical Analysis

Massachusetts Institute of Technology(麻省理工学院);Carnegie Mellon University(卡内基梅隆大学);University of Cambridge(剑桥大学);Stanford University(斯坦福大学)

Yuxin Xiao, Paul Pu Liang, Umang Bhatt, Willie Neiswanger, Ruslan Salakhutdinov, Louis-Philippe Morency

摘要:

预训练的语言模型(PLM)由于其在各种自然语言处理(NLP)任务中引人注目的预测性能而越来越受欢迎。在为NLP任务制定基于PLM的预测管道时,管道最大限度地减少校准误差也至关重要,特别是在安全关键型应用中。也就是说,管道应该可靠地指示我们何时可以信任它的预测。特别是,管道背后有各种考虑因素:(1)PLM的选择和(2)PLM的大小,(3)不确定性量化器的选择,(4)微调损耗的选择等等。虽然先前的工作已经研究了其中的一些考虑因素,他们通常得出结论的基础上有限的范围内的实证研究。对于如何构建一个校准良好的基于PLM的预测管道,仍然缺乏整体分析。为了填补这一空白,我们根据三个流行的NLP分类任务和域转移的设置,比较了各种流行的选项。作为回应,我们建议如下:(1)使用ELECTRA进行PLM编码,(2)如果可能,使用更大的PLM,(3)使用Temp Scaling作为不确定性量化器,以及(4)使用Focal Loss进行微调。

英文摘要:

Pre-trained language models (PLMs) have gained increasing popularity due to their compelling prediction performance in diverse natural language processing (NLP) tasks. When formulating a PLM-based prediction pipeline for NLP tasks, it is also crucial for the pipeline to minimize the calibration error, especially in safety-critical applications. That is, the pipeline should reliably indicate when we can trust its predictions. In particular, there are various considerations behind the pipeline: (1) the choice and (2) the size of PLM, (3) the choice of uncertainty quantifier, (4) the choice of fine-tuning loss, and many more. Although prior work has looked into some of these considerations, they usually draw conclusions based on a limited scope of empirical studies. There still lacks a holistic analysis on how to compose a well-calibrated PLM-based prediction pipeline. To fill this void, we compare a wide range of popular options for each consideration based on three prevalent NLP classification tasks and the setting of domain shift. In response, we recommend the following: (1) use ELECTRA for PLM encoding, (2) use larger PLMs if possible, (3) use Temp Scaling as the uncertainty quantifier, and (4) use Focal Loss for fine-tuning.

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。

↑