Long-form generation on FActScore (test)
0.8581AUROCSeSE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SeSEModel=Gemini-3-Flash2025.11 | 0.8581 | 0.818 | |
| DSEModel=Gemini-3-Flash2025.11 | 0.8315 | 0.8057 | |
| SeSEModel=DeepSeek-V3.12025.11 | 0.8105 | 0.7801 | |
| DSEModel=DeepSeek-V3.12025.11 | 0.7842 | 0.7684 | |
| P(True)Model=Gemini-3-Flash2025.11 | 0.7658 | 0.78 | |
| P(True)Model=DeepSeek-V3.12025.11 | 0.7216 | 0.7371 | |
| PH-VUModel=Gemini-3-Flash2025.11 | 0.6713 | 0.721 | |
| PH-VUModel=DeepSeek-V3.12025.11 | 0.6366 | 0.706 | |
| SCModel=Gemini-3-Flash2025.11 | 0.6226 | 0.6874 | |
| SCModel=DeepSeek-V3.12025.11 | 0.6066 | 0.6864 | |
| IL-VUModel=DeepSeek-V3.12025.11 | 0.538 | 0.6281 | |
| IL-VUModel=Gemini-3-Flash2025.11 | 0.526 | 0.6599 |