Non-Autoregressive Translation by Learning Target Categorical Codes

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.

PDF Abstract NAACL 2021 PDF NAACL 2021 Abstract


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Machine Translation IWSLT2014 German-English CNAT BLEU score 31.15 # 30
Machine Translation WMT2014 English-German CNAT BLEU score 26.6 # 57
Machine Translation WMT2014 German-English CNAT BLEU score 30.75 # 7