Commonsense Generation on CommonGen (dev)
47.67BLEU-3T5-3b + VAWI-SBS
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| T5-3b + VAWI-SBSBase Model=T5-3b, Augmentation=VAWI-SBS2022.12 | 47.67 | — | 37.54 | 33.41 | 59.94 | 18.34 | 34.67 | |
| T5-3b + VAWI-SBS + PETBase Model=T5-3b, Augmentation=VAWI-SBS + PET2022.12 | 47.4 | — | 37.36 | 33.71 | 59.78 | 18.18 | 34.17 | |
| T5-3b + NoneBase Model=T5-3b, Augmentation=None2022.12 | 45.92 | — | 35.92 | 33.02 | 58.57 | 17.71 | 33.51 | |
| T5-3b + ImagesBase Model=T5-3b, Augmentation=Images2022.12 | 45.69 | — | 35.5 | 33.55 | 58.94 | 17.51 | 32.91 | |
| BART-large + VAWI-SBSBase Model=BART-large, Augmentation=VAWI-SBS2022.12 | 44.56 | — | 34.17 | 32.47 | 58.46 | 17.23 | 33.67 | |
| BART-large + VAWI-SBS + PETBase Model=BART-large, Augmentation=VAWI-SBS + PET2022.12 | 43.12 | — | 33.76 | 32.2 | 58.12 | 16.91 | 33.17 | |
| BART-large + NoneBase Model=BART-large, Augmentation=None2022.12 | 42.8 | — | 32.42 | 31.36 | 57.57 | 16.56 | 32.94 | |
| BART-large + ImagesBase Model=BART-large, Augmentation=Images2022.12 | 42.67 | — | 32.67 | 32.12 | 57.46 | 16.78 | 32.81 | |
| Gehrmann et al.status=previous state-of-the-art2022.05 | — | 32.5 | — | — | — | — | — | |
| UL20Bmode=supervised finetuning2022.05 | — | 37.4 | — | — | — | — | — |