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

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


Abstract
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.
Anthology ID:
2022.tacl-1.81
Volume:
Transactions of the Association for Computational Linguistics, Volume 10
Month:
Year:
2022
Address:
Cambridge, MA
Editors:
Brian Roark, Ani Nenkova
Venue:
TACL
SIG:
Publisher:
MIT Press
Note:
Pages:
1423–1439
Language:
URL:
https://aclanthology.org/2022.tacl-1.81
DOI:
10.1162/tacl_a_00526
Bibkey:
Cite (ACL):
Laurent Sartran, Samuel Barrett, Adhiguna Kuncoro, Miloš Stanojević, Phil Blunsom, and Chris Dyer. 2022. Transformer Grammars: Augmenting Transformer Language Models with Syntactic Inductive Biases at Scale. Transactions of the Association for Computational Linguistics, 10:1423–1439.
Cite (Informal):
Transformer Grammars: Augmenting Transformer Language Models with Syntactic Inductive Biases at Scale (Sartran et al., TACL 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-5/2022.tacl-1.81.pdf