A per-layer additive domain and task embedding trained with masked language modeling improves unsupervised domain adaptation for NMT and can be combined with back-translation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Unsupervised Domain Adaptation for Neural Machine Translation with Domain-Aware Feature Embeddings
A per-layer additive domain and task embedding trained with masked language modeling improves unsupervised domain adaptation for NMT and can be combined with back-translation.