Fact Verification on FEVER-S
54AccuracyCPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPOBackbone=Mistral-7B2024.06 | 54 | 53 | |
| ToTBackbone=LLAMA2-13B2024.06 | 51.3 | 3,433.2 | |
| CPOBackbone=LLAMA2-13B2024.06 | 50.7 | 68.2 | |
| ToTBackbone=Mistral-7B2024.06 | 50.5 | 5,537.8 | |
| CoTBackbone=LLAMA2-13B2024.06 | 50 | 61.1 | |
| TS-SFTBackbone=Mistral-7B2024.06 | 49.7 | 49.5 | |
| CPOBackbone=LLAMA2-7B2024.06 | 49 | 41.2 | |
| TS-SFTBackbone=LLAMA2-13B2024.06 | 48.8 | 58 | |
| CoTBackbone=Mistral-7B2024.06 | 48 | 51 | |
| ToTBackbone=LLAMA2-7B2024.06 | 47.5 | 2,539.5 | |
| TS-SFTBackbone=LLAMA2-7B2024.06 | 46 | 40.4 | |
| CoTBackbone=LLAMA2-7B2024.06 | 44.3 | 40.6 |