Fact Verification on RAWFC
81.63PrecisionKG-CRAFT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KG-CRAFTBackbone=Llama 3.32026.01 | 81.63 | 81.53 | 81.58 | |
| KG-CRAFTBackbone=Claude 3.72026.01 | 69.37 | 68.59 | 68.98 | |
| DelphiAgentBackbone=gpt-4o-mini2026.01 | 68.53 | 63.95 | 64.68 | |
| DelphiAgentBackbone=gpt-4o2026.01 | 68.05 | 68.03 | 68.04 | |
| KG-CRAFTBackbone=Claude 3.52026.01 | 67.1 | 66.25 | 66.67 | |
| RAFTSbackbone=GPT-3.52024.06 | 62.8 | 52.6 | 57.3 | |
| RAFTSbackbone=Mistral 7B2024.06 | 62.6 | 51.6 | 56.6 | |
| L-DefenseBackbone=ChatGPT2026.01 | 61.72 | 61.01 | 61.2 | |
| L-DefenseBackbone=LLAMA22026.01 | 60.95 | 61 | 60.12 | |
| FactLLAMAknow2026.01 | 56.11 | 55.5 | 55.65 | |
| Claude 3.7 Sonnet2026.01 | 55.9 | 57.12 | 56.5 | |
| Claude 3.5 Sonnet2026.01 | 54.85 | 55.04 | 54.94 | |
| EExpFND2026.01 | 54.43 | 53.49 | 53.61 | |
| FactLLAMA2026.01 | 53.76 | 54 | 53.76 | |
| HiSS2024.06 | 53.4 | 54.4 | 53.9 | |
| Llama 3.3 70B2026.01 | 53.16 | 55.04 | 54.08 | |
| CofCED2024.06 | 53 | 51 | 52 | |
| CofCED2026.01 | 52.99 | 50.99 | 51.07 | |
| ReAct2024.06 | 51.2 | 48.5 | 49.8 | |
| SBERT-FC2024.06 | 51.1 | 46 | 48.4 | |
| SBERT-FC2026.01 | 51.06 | 45.92 | 45.51 | |
| GPT 3.52024.06 | 48.5 | 48.5 | 48.5 | |
| ChatGPT 3.5 Turbo2026.01 | 47.72 | 48.62 | 44.43 | |
| SentHAN2024.06 | 45.7 | 45.5 | 45.6 | |
| GenFE-MT2026.01 | 45.64 | 45.27 | 45.08 | |
| dEFEND2026.01 | 44.93 | 43.26 | 44.07 | |
| GenFE2026.01 | 44.92 | 44.74 | 44.43 | |
| dEFEND2024.06 | 44.9 | 43.2 | 44 | |
| CoT2024.06 | 42.4 | 46.6 | 44.4 | |
| Llama 2 7B2026.01 | 37.3 | 38.03 | 36.77 |