Natural Language Generation on WebNLG en
53.5ROUGE-2SOTA
Evaluation Results
| Method | Links | |
|---|---|---|
| SOTAEvaluation Protocol=Finetuning2022.08 | 53.5 | |
| PaLMEvaluation Protocol=Finetuning, Model Scale=540B2022.08 | 49.3 | |
| AlexaTM 20BEvaluation Protocol=Finetuning, Model Scale=20B2022.08 | 49 | |
| PaLMEvaluation Protocol=Finetuning, Model Scale=62B2022.08 | 48.6 | |
| PaLMEvaluation Protocol=1-shot, Model Scale=540B, Number of shots=12022.08 | 44.4 | |
| T5 XXLEvaluation Protocol=Few-shot, Model Scale=XXL2022.08 | 39.6 | |
| PaLMEvaluation Protocol=1-shot, Model Scale=62B, Number of shots=12022.08 | 38.6 | |
| AlexaTM 20BEvaluation Protocol=Few-shot, Model Scale=20B, Number of shots=32*2022.08 | 26.16 | |
| AlexaTM 20BEvaluation Protocol=1-shot, Model Scale=20B, Number of shots=12022.08 | 20 |