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 2203.00633cs.CL
收藏

Transformer Grammars:用句法归纳偏差扩充Transformer语言模型

Transformer Grammars: Augmenting Transformer Language Models with Syntactic Inductive Biases at Scale

DeepMind(深度思维);University of Oxford(牛津大学)

Laurent Sartran, Samuel Barrett, Adhiguna Kuncoro, Miloš Stanojević, Phil Blunsom, Chris Dyer

摘要:

我们介绍了Transformer语法(TG),一类新的Transformer语言模型,结合(i)的表达能力,可扩展性,和强大的性能的Transformers和(ii)递归语法组合,这在这里是通过一个特殊的注意掩码和确定性变换的线性树。我们发现,TG优于各种强大的基线上的高级语言建模的困惑,以及多个语法敏感的语言建模评估指标。此外,我们发现,递归的句法组合瓶颈,代表每个句子作为一个单一的向量损害困惑的文档级语言建模,提供证据表明,一种不同的记忆机制-一个是独立的组成的句法表示-在当前成功的模型中起着重要作用的长文本。

英文摘要:

We introduce Transformer Grammars (TGs), a novel class of Transformer language models that combine (i) the expressive power, scalability, and strong performance of Transformers and (ii) recursive syntactic compositions, which here are implemented through a special attention mask and deterministic transformation of the linearized tree. We find that TGs outperform various strong baselines on sentence-level language modeling perplexity, as well as on multiple syntax-sensitive language modeling evaluation metrics. Additionally, we find that the recursive syntactic composition bottleneck which represents each sentence as a single vector harms perplexity on document-level language modeling, providing evidence that a different kind of memory mechanism -- one that is independent of composed syntactic representations -- plays an important role in current successful models of long text.

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。

↑