Table 1:
ROUGE scores achieved by the standard BART model and our proposed multi-headed architecture on the dev and test sets.
| Model | ROUGE-L (dev) | ROUGE-L(test) |
|---|---|---|
| BART | 20.4 | 19.7 |
| Multi-head | 19.9 | 19.3 |
ROUGE scores achieved by the standard BART model and our proposed multi-headed architecture on the dev and test sets.
| Model | ROUGE-L (dev) | ROUGE-L(test) |
|---|---|---|
| BART | 20.4 | 19.7 |
| Multi-head | 19.9 | 19.3 |