Abstract
Transfer learning has been an important technique for low-resource neural machine translation. In this work, we build two systems to study how relatedness can benefit the translation performance. The primary system adopts machine translation model pre-trained on related language pair and the contrastive system adopts that pre-trained on unrelated language pair. We show that relatedness is not required for transfer learning to work but does benefit the performance.
Abstract (translated)
URL
https://arxiv.org/abs/2108.03739
