Claim Verification on SCIFACT
94.32AccuracyInfoRE + CoT
Evaluation Results
| Method | Links | |
|---|---|---|
| InfoRE + CoTBackbone=GPT-4, Zero-shot=true2024.04 | 94.32 | |
| InfoREBackbone=GPT-4, Zero-shot=true2024.04 | 93.67 | |
| CoTBackbone=GPT-4, Zero-shot=true2024.04 | 92.47 | |
| StandardBackbone=GPT-4, Zero-shot=true2024.04 | 91.4 | |
| InfoRE + CoTBackbone=GPT-3.5, Zero-shot=true2024.04 | 82.26 | |
| InfoREBackbone=GPT-3.5, Zero-shot=true2024.04 | 81.54 | |
| CoTBackbone=GPT-3.5, Zero-shot=true2024.04 | 78.49 | |
| StandardBackbone=GPT-3.5, Zero-shot=true2024.04 | 77.42 | |
| InfoRE + CoTBackbone=LLAMA2-70B, Zero-shot=true2024.04 | 64.02 | |
| InfoREBackbone=LLAMA2-70B, Zero-shot=true2024.04 | 63.81 | |
| CoTBackbone=LLAMA2-70B, Zero-shot=true2024.04 | 61.24 | |
| StandardBackbone=LLAMA2-70B, Zero-shot=true2024.04 | 60.7 |