AMR Generation on LDC2015E86 (dev)
33.1BLEUGIGA-20M
Evaluation results
| Method | Links | |
|---|---|---|
| GIGA-20MPre-training Corpus=Gigaword, Training Size=20M sentences2017.04 | 33.1 | |
| GIGA-2MPre-training Corpus=Gigaword, Training Size=2M sentences2017.04 | 31.8 | |
| GIGA-200kPre-training Corpus=Gigaword, Training Size=200k sentences2017.04 | 27.2 | |
| PBMT*Training Corpus=LDC2014T122017.04 | 27.2 | |
| TREETOSTR2017.04 | 23 | |
| AMR-ONLYTraining Data=AMR corpus only2017.04 | 21.7 | |
| TSP2017.04 | 21.1 |