Transformer-based models have brought a radical change to neural machine translation. A key feature of the Transformer architecture is the so-called multi-head attention mechanism, which allows the model to focus simultaneously on different parts of the input. However, recent works have shown that most attention heads learn simple, and often redundant, positional patterns. In this paper, we propose to replace all but one attention head of each encoder layer with simple fixed – non-learnable – attentive patterns that are solely based on position and do not require any external knowledge. Our experiments with different data sizes and multiple language pairs show that fixing the attention heads on the encoder side of the Transformer at training time does not impact the translation quality and even increases BLEU scores by up to 3 points in low-resource scenarios.

Raganato, A., Scherrer, Y., Tiedemann, J. (2020). Fixed encoder self-attention patterns in transformer-based machine translation. In Findings of the Association for Computational Linguistics: EMNLP 2020 (pp.556-568). Association for Computational Linguistics (ACL) [10.18653/v1/2020.findings-emnlp.49].

Fixed encoder self-attention patterns in transformer-based machine translation

Raganato, A
;
2020

Abstract

Transformer-based models have brought a radical change to neural machine translation. A key feature of the Transformer architecture is the so-called multi-head attention mechanism, which allows the model to focus simultaneously on different parts of the input. However, recent works have shown that most attention heads learn simple, and often redundant, positional patterns. In this paper, we propose to replace all but one attention head of each encoder layer with simple fixed – non-learnable – attentive patterns that are solely based on position and do not require any external knowledge. Our experiments with different data sizes and multiple language pairs show that fixing the attention heads on the encoder side of the Transformer at training time does not impact the translation quality and even increases BLEU scores by up to 3 points in low-resource scenarios.
paper
transformer; machine translation; attention
English
Findings of the Association for Computational Linguistics: EMNLP 2020
2020
Findings of the Association for Computational Linguistics: EMNLP 2020
978-1-952148-90-3
2020
556
568
reserved
Raganato, A., Scherrer, Y., Tiedemann, J. (2020). Fixed encoder self-attention patterns in transformer-based machine translation. In Findings of the Association for Computational Linguistics: EMNLP 2020 (pp.556-568). Association for Computational Linguistics (ACL) [10.18653/v1/2020.findings-emnlp.49].
File in questo prodotto:
File Dimensione Formato  
2020.findings-emnlp.49.pdf

Solo gestori archivio

Dimensione 1.2 MB
Formato Adobe PDF
1.2 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/361583
Citazioni
  • Scopus 32
  • ???jsp.display-item.citation.isi??? 23
Social impact