Fact Verification on FEVER (Accuracy and Latency)
61.4AccuracyToT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ToTBackbone=Mistral-7B2024.06 | 61.4 | 3,291.3 | |
| TS-SFTBackbone=Mistral-7B2024.06 | 60.2 | 41.7 | |
| CPOBackbone=Mistral-7B2024.06 | 59.9 | 40.6 | |
| CoTBackbone=Mistral-7B2024.06 | 57.9 | 41.8 | |
| CPOBackbone=LLAMA2-7B2024.06 | 53.2 | 36.8 | |
| CPOBackbone=LLAMA2-13B2024.06 | 49.2 | 43.9 | |
| ToTBackbone=LLAMA2-13B2024.06 | 48.8 | 3,139.8 | |
| TS-SFTBackbone=LLAMA2-13B2024.06 | 48.8 | 44 | |
| CoTBackbone=LLAMA2-13B2024.06 | 48.2 | 45.4 | |
| TS-SFTBackbone=LLAMA2-7B2024.06 | 47.5 | 34 | |
| ToTBackbone=LLAMA2-7B2024.06 | 47.3 | 2,087.1 | |
| CoTBackbone=LLAMA2-7B2024.06 | 45.8 | 33.8 |