Long-form Factuality Calibration on FactScore
0.076ECEDINCO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DINCOModel=Qwen3-8B2025.09 | 0.076 | 0.202 | 0.767 | 0.518 | 0.538 | |
| NVCModel=Gemma-3-4B-IT2025.09 | 0.123 | 0.23 | 0.726 | 0.695 | 0.704 | |
| SCModel=Qwen3-8B2025.09 | 0.162 | 0.226 | 0.771 | 0.468 | 0.494 | |
| DINCOModel=Gemma-3-4B-IT2025.09 | 0.172 | 0.21 | 0.793 | 0.724 | 0.712 | |
| NVCModel=Qwen3-8B2025.09 | 0.191 | 0.263 | 0.681 | 0.444 | 0.443 | |
| SCModel=Gemma-3-4B-IT2025.09 | 0.197 | 0.233 | 0.787 | 0.629 | 0.607 | |
| VCModel=Qwen3-8B2025.09 | 0.433 | 0.431 | 0.625 | 0.073 | 0.122 | |
| VCModel=Gemma-3-4B-IT2025.09 | 0.527 | 0.527 | 0.683 | -0.081 | -0.129 |