In this paper, we re-examine the Markov property in the context of neural machine translation. We design a Markov Autoregressive Transformer~(MAT) and undertake a comprehensive assessment of its performance across four WMT benchmarks. Our findings indicate that MAT with an order larger than 4 can generate translations with quality on par with that of conventional autoregressive transformers. In addition, counter-intuitively, we also find that the advantages of utilizing a higher-order MAT do not specifically contribute to the translation of longer sentences.
@article{arxiv.2402.02084,
title = {Revisiting the Markov Property for Machine Translation},
author = {Cunxiao Du and Hao Zhou and Zhaopeng Tu and Jing Jiang},
journal= {arXiv preprint arXiv:2402.02084},
year = {2024}
}