Replacing Language Model for Style Transfer

arXiv:2211.07343v2 Announce Type: replace-cross
Abstract: We introduce replacing language model (RLM), a sequence-to-sequence language modeling framework for text style transfer (TST). Our method autoregressively replaces each token of the source sentence with a text span that has a similar meaning but in the target style. The new span is generated via a non-autoregressive masked language model, which can better preserve the local-contextual meaning of the replaced token. This RLM generation scheme gathers the flexibility of autoregressive models and the accuracy …

arxiv cs.cl cs.lg language language model style style transfer transfer type