Text Generation on WritingPrompts (WP) (test)
0.224BLEU-1HINT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| HINT2021.05 | 0.224 | 32.73 | 0.084 | 0.567 | 0.805 | 0.313 | |
| HINTablation=without discourse-level representation learning2021.05 | 0.223 | 33.18 | 0.083 | 0.588 | 0.818 | 0.307 | |
| HINTablation=without sentence-level representation learning2021.05 | 0.216 | 33.08 | 0.08 | 0.598 | 0.823 | 0.303 | |
| HINTablation=without sentence and discourse-level representation learning2021.05 | 0.207 | 33.71 | 0.076 | 0.61 | 0.845 | 0.28 | |
| BART2021.05 | 0.205 | 34.42 | 0.075 | 0.62 | 0.854 | 0.291 | |
| BART-Posttraining_mode=post-training2021.05 | 0.205 | 35.11 | 0.076 | 0.671 | 0.862 | 0.271 | |
| GPT-22021.05 | 0.2 | — | 0.073 | 0.655 | 0.883 | 0.287 | |
| Plan & Write2021.05 | 0.199 | — | 0.07 | 0.524 | 0.851 | 0.272 | |
| BART-MTLtraining_mode=multi-task learning2021.05 | 0.198 | 35.71 | 0.076 | 0.654 | 0.846 | 0.305 | |
| Seq2Seq2021.05 | 0.165 | 129.51 | 0.07 | 0.623 | 0.819 | 0.283 |