Fact-checking on FEVER (F1Mac, AUC, Acc, ∆Acc)
94.3F1 MacroNone (Original Claims)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| None (Original Claims)Input Setting=Gold-Evidence2026.01 | 94.3 | 97.9 | 94.3 | — | |
| ParaphrasingInput Setting=Gold-Evidence2026.01 | 91.2 | 95.3 | 91.2 | -0.031 | |
| Word SwapInput Setting=Gold-Evidence2026.01 | 89.7 | 95.7 | 89.7 | -0.046 | |
| Character PerturbationsInput Setting=Gold-Evidence2026.01 | 86.8 | 93.8 | 86.9 | -0.074 | |
| Synonym SubstitutionInput Setting=Gold-Evidence2026.01 | 84.3 | 92.7 | 84.5 | -0.098 | |
| None (Original Claims)Input Setting=Claim-Only2026.01 | 81.4 | 89.3 | 81.4 | — | |
| Synonym SubstitutionInput Setting=Claim-Only2026.01 | 76.4 | 84.8 | 76.4 | -0.05 | |
| Persuasion (Blind)Input Setting=Gold-Evidence2026.01 | 76.4 | 86.2 | 76.4 | -0.179 | |
| Character PerturbationsInput Setting=Claim-Only2026.01 | 76.2 | 84.6 | 76.2 | -0.052 | |
| Word SwapInput Setting=Claim-Only2026.01 | 75.5 | 84.7 | 75.7 | -0.057 | |
| ParaphrasingInput Setting=Claim-Only2026.01 | 75.5 | 83.7 | 75.7 | -0.057 | |
| Persuasion (Blind)Input Setting=Claim-Only2026.01 | 67.8 | 74.5 | 68.2 | -0.132 | |
| Persuasion (Oracle)Input Setting=Gold-Evidence2026.01 | 14.2 | 2.9 | 16.4 | -0.779 | |
| Persuasion (Oracle)Input Setting=Claim-Only2026.01 | 4.2 | 0.4 | 4.3 | -0.771 |