Text Generation on Story
63.6Coherence Win RateContrastive Decoding
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Contrastive DecodingCD Model=GPT-2 XL, Baseline Decoding Strategy=nucleus, Baseline Model=GPT-2 XL2022.10 | 63.6 | 4.5 | 31.8 | 40.4 | 10.6 | 49 | |
| Contrastive DecodingCD Model=OPT-13B, Baseline Decoding Strategy=typical, Baseline Model=OPT-13B2022.10 | 62.6 | 20.2 | 17.2 | 52 | 21.2 | 26.8 | |
| Contrastive DecodingCD Model=OPT-13B, Baseline Decoding Strategy=nucleus, Baseline Model=OPT-13B2022.10 | 61.6 | 10.1 | 28.3 | 44.9 | 29.3 | 25.8 | |
| Contrastive DecodingCD Model=GPT-2 XL, Baseline Decoding Strategy=typical, Baseline Model=GPT-2 XL2022.10 | 50.6 | 25.6 | 23.8 | 38.7 | 36.3 | 25 |