Fusing Part-of-Speech Information in Low-Resource Neural Paraphrase Generation

Figure 1

The “base” encoder.