Fact Verification on ClearFacts
84Macro F1ClearCheck
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ClearCheckSize=8B, Output=binary2026.06 | 84 | — | |
| MiniCheckSize=7B, Output=binary2026.06 | 81.2 | — | |
| GPT-4o-miniOutput format=struct2026.06 | 69.8 | — | |
| GPT-4o-mini (0-shot)Output=struct2026.06 | 69.8 | 69.9 | |
| SEVA-GRPO (3B)Output format=struct, Size=3B2026.06 | 69 | — | |
| SEVA-GRPOSize=3B, Output=struct2026.06 | 69 | 69.6 | |
| SEVA-SFT (LoRA-128)Size=7B, Output=struct2026.06 | 68.5 | 68.6 | |
| MiniCheck-FT5Output format=binary2026.06 | 68.3 | — | |
| MiniCheck-Flan-T5Size=770M, Output=binary2026.06 | 68.3 | 68.3 | |
| Llama-3.1 (0-shot)Size=8B, Output=binary2026.06 | 67.2 | — | |
| SEVA-SFT (3B)Output format=struct, Size=3B2026.06 | 64.9 | — | |
| SEVA-SFTSize=3B, Output=struct2026.06 | 64.9 | 65.2 |