Scientific claim-citation verification on SCITANCE 3-class (test)
86.7Micro-F1DEEPSCIVERIFY
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DEEPSCIVERIFYEvidence=Full-text2026.05 | 86.7 | 81.5 | 88.9 | 4.5 | |
| Claude Sonnet 4.6Evidence=Full-text2026.05 | 82.4 | 76.2 | 85.7 | 0.2 | |
| GPT-4Evidence=Abstract2026.05 | 82.2 | 78.8 | 86.7 | — | |
| GPT-4Evidence=Full-text2026.05 | 81.3 | 76.7 | 84.6 | 0.9 | |
| Claude Sonnet 4.6Evidence=Abstract2026.05 | 81.1 | 73.8 | 85.6 | 1.1 | |
| GPT-5.4Evidence=Full-text2026.05 | 79.1 | 76.1 | 81.3 | 3.1 | |
| Gemini 2.5 FlashEvidence=Full-text2026.05 | 75.8 | 71 | 78 | 6.4 | |
| GPT-5.4Evidence=Abstract2026.05 | 74.4 | 72.4 | 81.1 | 7.8 | |
| Gemini 2.5 FlashEvidence=Abstract2026.05 | 74.4 | 70.8 | 75.6 | 7.8 |