arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.07352cs.CLcs.AIcs.LG
收藏

用探测法预测微调性能

Predicting Fine-Tuning Performance with Probing

University of Toronto(多伦多大学);Vector Institute for Artificial Intelligence(向量人工智能研究所);Unity Health Toronto(多伦多联合健康中心)

Zining Zhu, Soroosh Shahtalebi, Frank Rudzicz

摘要:

大型NLP模型最近在语言理解任务中表现出令人印象深刻的性能,通常通过其微调性能进行评估。另外,探测作为一种解释大型NLP模型内在机制的轻量级方法越来越受到关注。在探测中,事后分类器在诊断特定能力的“域外”数据集上进行训练。虽然对语言模型的探索已经导致了有见地的发现,但它们似乎与模型的发展脱节。本文探讨了探测深度NLP模型以提取模型开发中广泛使用的代理信号的实用性-微调性能。我们发现,它是可能的,使用的准确性只有三个探测测试来预测微调性能的误差$40\%$ - $80\%$小于基线。我们进一步讨论了探测可以增强深度NLP模型开发的可能途径。

英文摘要:

Large NLP models have recently shown impressive performance in language understanding tasks, typically evaluated by their fine-tuned performance. Alternatively, probing has received increasing attention as being a lightweight method for interpreting the intrinsic mechanisms of large NLP models. In probing, post-hoc classifiers are trained on "out-of-domain" datasets that diagnose specific abilities. While probing the language models has led to insightful findings, they appear disjointed from the development of models. This paper explores the utility of probing deep NLP models to extract a proxy signal widely used in model development -- the fine-tuning performance. We find that it is possible to use the accuracies of only three probing tests to predict the fine-tuning performance with errors $40\%$ - $80\%$ smaller than baselines. We further discuss possible avenues where probing can empower the development of deep NLP models.

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。

↑