Table-to-text generation on WebNLG
65.4BLEU (Seen)HTLM
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HTLMEvaluation Protocol=Fine-tuning2021.07 | 65.4 | 48.4 | 55.6 | 0.46 | 0.39 | 0.42 | 0.33 | 0.51 | 0.4 | |
| FINE-TUNEBackbone=GPT-2LARGE2021.01 | 65.3 | 43.1 | 55.5 | 0.46 | 0.38 | 0.42 | 0.33 | 0.53 | 0.42 | |
| GPT-2LARGEEvaluation Protocol=Fine-tuning2021.07 | 65.3 | 43.1 | 55.5 | 0.46 | 0.38 | 0.42 | 0.33 | 0.53 | 0.42 | |
| HTLMEvaluation Protocol=Prefix (0.1%)2021.07 | 64.8 | 46.1 | 56.3 | 0.46 | 0.38 | 0.42 | 0.33 | 0.47 | 0.4 | |
| FINE-TUNEBackbone=GPT-2MEDIUM2021.01 | 64.2 | 27.7 | 46.5 | 0.45 | 0.3 | 0.38 | 0.33 | 0.76 | 0.53 | |
| GPT-2MEDIUMEvaluation Protocol=Fine-tuning2021.07 | 64.2 | 27.7 | 46.5 | 0.45 | 0.3 | 0.38 | 0.33 | 0.76 | 0.53 | |
| SOTA2021.01 | 63.9 | 52.8 | 57.1 | 0.46 | 0.41 | 0.44 | — | — | — | |
| PrefixBackbone=GPT-2LARGE2021.01 | 63.4 | 47.7 | 56.3 | 0.45 | 0.39 | 0.42 | 0.34 | 0.48 | 0.4 | |
| GPT-2LARGEEvaluation Protocol=Prefix (0.1%)2021.07 | 63.4 | 47.7 | 56.3 | 0.45 | 0.39 | 0.42 | 0.34 | 0.48 | 0.4 | |
| PREFIX (0.1%)Backbone=GPT-2MEDIUM2021.01 | 62.9 | 45.6 | 55.1 | 0.44 | 0.38 | 0.41 | 0.35 | 0.49 | 0.41 | |
| GPT-2MEDIUMEvaluation Protocol=Prefix (0.1%)2021.07 | 62.9 | 45.6 | 55.1 | 0.44 | 0.38 | 0.41 | 0.35 | 0.49 | 0.41 | |
| ADAPTER (3%)Backbone=GPT-2MEDIUM2021.01 | 60.4 | 48.3 | 54.9 | 0.43 | 0.38 | 0.41 | 0.35 | 0.45 | 0.39 | |
| ADAPTER (0.1%)Backbone=GPT-2MEDIUM2021.01 | 54.5 | 45.1 | 50.2 | 0.39 | 0.36 | 0.38 | 0.4 | 0.46 | 0.43 | |
| FT-TOP2Backbone=GPT-2MEDIUM2021.01 | 53.6 | 18.9 | 36 | 0.38 | 0.23 | 0.31 | 0.49 | 0.99 | 0.72 | |
| TILB-PipelineEvaluation Protocol=Base-lines2021.07 | 44.34 | 20.65 | 35.29 | 0.38 | 0.21 | 0.3 | 0.48 | 0.64 | 0.56 | |
| HTLMEvaluation Protocol=One-Shot2021.07 | 28.1 | 18.5 | 22.8 | 0.24 | 0.21 | 0.12 | 0.78 | 0.79 | 0.78 | |
| UIT-VNU-PipelineEvaluation Protocol=Base-lines2021.07 | 19.87 | 0.11 | 7.07 | 0.15 | 0.03 | 0.09 | 0.78 | 0.87 | 0.82 |