Abductive Reasoning Generation on Abductive Commonsense Reasoning aNLG (test)
17.67BLEU-4ClarET
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ClarETSize=400M, Mode=Fine-tuned2022.03 | 17.67 | 41.04 | 57.31 | |
| ClarET2022.03 | 17.67 | 41.04 | 0.5731 | |
| BARTSize=400M, Mode=Fine-tuned2022.03 | 16.47 | 38.73 | 56.36 | |
| BART-largesize=large2022.03 | 16.47 | 38.73 | 0.5636 | |
| Fine-tuned GPT2-Ltraining=Fine-tuned, size=Large2022.03 | 13.52 | 18.01 | — | |
| GRFMode=Fine-tuned2022.03 | 11.62 | 34.62 | — | |
| GRF2022.03 | 11.62 | 34.62 | — | |
| GLMSize=335M, Mode=Fine-tuned2022.03 | 7.79 | 25.54 | 54.85 | |
| GLM-largesize=large2022.03 | 7.79 | 25.54 | 0.5485 | |
| COMET-E+GPT22022.03 | 3.03 | 22.93 | 0.4852 | |
| COMET-T+GPT22022.03 | 2.29 | 22.51 | 0.4846 | |
| GPT2-SSize=124M, Mode=Fine-tuned2022.03 | 2.23 | 22.83 | 48.74 | |
| O1-O2-Onlyinput=O1-O2 only2022.03 | 2.23 | 22.83 | 0.4874 | |
| GPT2-Fixedweights=Fixed2022.03 | 0 | 9.99 | 0.3669 |