Uncertainty Quantification on TriviaQA
22.4PRRLogTokU
Evaluation Results
| Method | Links | |
|---|---|---|
| LogTokUModel=Gemma2026.06 | 22.4 | |
| P(true)Model=Gemma2026.06 | 26.8 | |
| RAUQModel=Gemma2026.06 | 41.6 | |
| LogTokUModel=Fanar2026.06 | 44.5 | |
| P(true)Model=Fanar2026.06 | 44.8 | |
| RAUQModel=Qwen2026.06 | 58.1 | |
| P(true)Model=Qwen2026.06 | 59.8 | |
| LogTokUModel=Qwen2026.06 | 60.6 | |
| LogProbModel=Qwen2026.06 | 62.7 | |
| LogProbModel=Gemma2026.06 | 68.6 | |
| LogProbModel=Fanar2026.06 | 68.8 | |
| GLUModel=Gemma2026.06 | 69.8 | |
| GLU-Sα-AUModel=Gemma2026.06 | 70.6 | |
| RAUQModel=Fanar2026.06 | 71.2 | |
| GLUModel=Qwen2026.06 | 71.7 | |
| GLU-Sα-SPModel=Qwen2026.06 | 72.3 | |
| GLUModel=Fanar2026.06 | 73 | |
| GLU-S¯Model=Fanar2026.06 | 73.2 |