Data-to-Text Generation on WikiBio (test)
45.14BLEUhier
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| hierauxiliary_training_tasks=true2021.02 | 45.14 | 75.09 | 46.02 | 54.65 | 10.1 | 16.8 | 56.2 | |
| KGPT-GraphInput format=Graph, Pre-training=Yes2020.10 | 45.1 | — | — | — | — | — | — | |
| KGPT-SeqInput format=Sequence, Pre-training=Yes2020.10 | 45.06 | — | — | — | — | — | — | |
| Hierarchical+Auxiliary Loss2020.10 | 45.01 | — | — | — | — | — | — | |
| Field-Gating2020.10 | 44.71 | — | — | — | — | — | — | |
| Template-GPT-22020.10 | 44.67 | — | — | — | — | — | — | |
| KGPT-GraphInput format=Graph, Pre-training=No2020.10 | 44.64 | — | — | — | — | — | — | |
| KBAtt2020.10 | 44.59 | — | — | — | — | — | — | |
| KGPT-SeqInput format=Sequence, Pre-training=No2020.10 | 44.58 | — | — | — | — | — | — | |
| Order Planning2020.10 | 43.91 | — | — | — | — | — | — | |
| stndmode=standard2021.02 | 41.77 | 79.75 | 45.02 | 55.28 | 4.2 | 13.8 | 58.9 | |
| MBDsupervision=fine-grain hallucination annotations2021.02 | 41.56 | 79 | 46.4 | 56.16 | 1.43 | 14.6 | 58.8 | |
| Table2Seq2020.10 | 40.26 | — | — | — | — | — | — | |
| halwohallucination_modeling=fixed value for whole instance2021.02 | 36.5 | 79.5 | 40.5 | 51.7 | — | — | — | |
| hsmmarchitecture=multi-branch2021.02 | 35.17 | 71.72 | 39.84 | 48.32 | 7.98 | 14.8 | 58.6 | |
| Table NLM2020.10 | 34.7 | — | — | — | — | — | — | |
| stnd_filteredtraining_data=cleanest 20% of data2021.02 | 34.66 | 80.9 | 42.48 | 53.27 | 0.74 | 12 | 62.1 | |
| GoldDescription=gold standard reference texts2021.02 | — | — | — | — | 23.82 | 19.2 | 53.8 |