Here are the outputs of all 5 models in two directions:
- Positive -> Negative: all_model_outputs.0
- Negative -> Positive: all_model_outputs.1 (Note: "1" refers to formal, and "0" informal.)
There are some abbreviations in the above output files, which are explained here:
| Abbrev. | meaning | example |
|---|---|---|
| src | source sentence (from the test set) | she was not happy being there . |
| ca | Cross Alignment (Shen et al., 2017) | she was not happy to everyone . |
| multi_decoder | Multi-Decoder (Fu et al., 2018) | she was not happy of there . |
| style_emb | Style-Embedding (Fu et al., 2018) | she was not sucks cooked go there . |
| dar | Delete-And-Retrieve (Li et al., 2018) | being very very nice being there . |
| ours | IMaT (Jin et al., 2019) | she was very happy with there . |