On the Information Redundancy in Non-Autoregressive Translation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909189377359872 |
|---|---|
| author | Wang, Zhihao Wang, Longyue Su, Jinsong Yao, Junfeng Tu, Zhaopeng |
| author_facet | Wang, Zhihao Wang, Longyue Su, Jinsong Yao, Junfeng Tu, Zhaopeng |
| contents | Token repetition is a typical form of multi-modal problem in fully non-autoregressive translation (NAT). In this work, we revisit the multi-modal problem in recently proposed NAT models. Our study reveals that these advanced models have introduced other types of information redundancy errors, which cannot be measured by the conventional metric - the continuous repetition ratio. By manually annotating the NAT outputs, we identify two types of information redundancy errors that correspond well to lexical and reordering multi-modality problems. Since human annotation is time-consuming and labor-intensive, we propose automatic metrics to evaluate the two types of redundant errors. Our metrics allow future studies to evaluate new methods and gain a more comprehensive understanding of their effectiveness. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_02673 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | On the Information Redundancy in Non-Autoregressive Translation Wang, Zhihao Wang, Longyue Su, Jinsong Yao, Junfeng Tu, Zhaopeng Computation and Language Token repetition is a typical form of multi-modal problem in fully non-autoregressive translation (NAT). In this work, we revisit the multi-modal problem in recently proposed NAT models. Our study reveals that these advanced models have introduced other types of information redundancy errors, which cannot be measured by the conventional metric - the continuous repetition ratio. By manually annotating the NAT outputs, we identify two types of information redundancy errors that correspond well to lexical and reordering multi-modality problems. Since human annotation is time-consuming and labor-intensive, we propose automatic metrics to evaluate the two types of redundant errors. Our metrics allow future studies to evaluate new methods and gain a more comprehensive understanding of their effectiveness. |
| title | On the Information Redundancy in Non-Autoregressive Translation |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2405.02673 |