Uncertainty Quantification on MMLU OOD via Math Prompts
67.7AUROCMATU
Evaluation Results
| Method | Links | |
|---|---|---|
| MATUBase Model=Qwen2.5-7B, Protocol=strict distribution shift2026.04 | 67.7 | |
| Eigv(Agre)-finalBase Model=Qwen2.5-7B, Protocol=strict distribution shift2026.04 | 63.55 | |
| SAUP-MultipleBase Model=Qwen2.5-7B, Protocol=strict distribution shift2026.04 | 62.01 | |
| SAUP-SingleBase Model=Qwen2.5-7B, Protocol=strict distribution shift2026.04 | 59.3 |