arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.09393cs.CLcs.AIcs.LG
收藏

利用动态Oracle改进自回归训练

Improving Autoregressive Training with Dynamic Oracles

Carnegie Mellon University(卡内基梅隆大学);University of Michigan(密歇根大学);Harvard University(哈佛大学)

Jianing Yang, Harshine Visvanathan, Yilin Wang, Xinyi Hu, Matthew Gormley

AI总结:

针对teacher forcing等方法的暴露偏差及训练推理指标不匹配问题,本文为基于片段的F1、ROUGE和BLEU构建动态oracle,并在NER与摘要任务上验证其优于基线。

摘要:

NLP中的许多任务都可以表述为序列决策问题,涵盖从序列标注到文本生成等任务。然而,对于许多任务而言,包括最大似然(teacher forcing)和计划采样在内的标准训练方法都存在暴露偏差,以及训练与推理阶段所采用指标不匹配的问题。DAgger提供了一种缓解这些问题的方案,但它需要一种面向特定指标的动态oracle算法,而对于许多常见指标(如基于片段的F1、ROUGE和BLEU)而言,这种算法并不存在。本文中,我们开发了这些新颖的动态oracle,并证明它们对于基于片段的F1等可分解指标能够保持DAgger的无悔保证。我们在命名实体识别(NER)、文本摘要和机器翻译(MT)上评估了该算法的性能。尽管带有动态oracle的DAgger在我们的MT实验中产生的结果不太理想,但它在NER和文本摘要任务上优于基线技术。

英文摘要:

Many tasks within NLP can be framed as sequential decision problems, ranging from sequence tagging to text generation. However, for many tasks, the standard training methods, including maximum likelihood (teacher forcing) and scheduled sampling, suffer from exposure bias and a mismatch between metrics employed during training and inference. DAgger provides a solution to mitigate these problems, yet it requires a metric-specific dynamic oracle algorithm, which does not exist for many common metrics like span-based F1, ROUGE, and BLEU. In this paper, we develop these novel dynamic oracles and show they maintain DAgger's no-regret guarantee for decomposable metrics like span-based F1. We evaluate the algorithm's performance on named entity recognition (NER), text summarization, and machine translation (MT). While DAgger with dynamic oracle yields less favorable results in our MT experiments, it outperforms the baseline techniques in NER and text summarization.

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。

↑