Binary Factuality Verification on HotpotQA
72.94AccuracyDiVA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DiVAModel=Qwen3-4B-Inst2026.01 | 72.94 | 72.46 | |
| DiVAModel=Qwen2.5-7B-Inst2026.01 | 70.98 | 70.88 | |
| FactScoreModel=GPT-42026.01 | 61.85 | 55.21 | |
| AG-VerifierModel=Qwen3-4B-Inst2026.01 | 60.39 | 60.51 | |
| AG-VerifierModel=Qwen2.5-7B-Inst2026.01 | 50.2 | 49.76 | |
| G-VerifierModel=Qwen3-4B-Inst2026.01 | 45.29 | 44.85 | |
| G-VerifierModel=Qwen2.5-7B-Inst2026.01 | 43.53 | 42.49 |