Non-Autoregressive Translation by Learning Target Categorical Codes
arXiv:2103.11405
Abstract
Non-autoregressive Transformer is a promising text generation model. However, current non-autoregressive models still fall behind their autoregressive counterparts in translation quality. We attribute this accuracy gap to the lack of dependency modeling among decoder inputs. In this paper, we propose CNAT, which learns implicitly categorical codes as latent variables into the non-autoregressive decoding. The interaction among these categorical codes remedies the missing dependencies and improves the model capacity. Experiment results show that our model achieves comparable or better performance in machine translation tasks, compared with several strong baselines.
arXiv comment11 pages, 3 figures, 7 tables. Accepted by NAACL-2021
References in corpus (6)
- Distilling the Knowledge in a Neural Network
- Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
- Fast Decoding in Sequence Models using Discrete Latent Variables
- Syntactically Supervised Transformers for Faster Neural Machine Translation
- Non-autoregressive Transformer by Position Learning
- Glancing Transformer for Non-Autoregressive Neural Machine Translation