Multiple Choice Question Answering on MMLU-Pro Physics
0.973Calibration Threshold (q-hat)Conformal Social Choice
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Conformal Social ChoiceDebate Round=3, Alpha=0.05, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.973 | 94.2 | 1.27 | 87.2 | 100 | |
| Conformal Social ChoiceDebate Round=2, Alpha=0.05, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.957 | 94 | 1.27 | 86 | 88.9 | |
| Conformal Social ChoiceDebate Round=0, Alpha=0.05, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.95 | 95.1 | 2.09 | 47.8 | 98.4 | |
| Conformal Social ChoiceDebate Round=1, Alpha=0.05, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.923 | 93.5 | 1.26 | 81.8 | 92.8 | |
| Conformal Social ChoiceDebate Round=0, Alpha=0.10, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.769 | 89.4 | 1.33 | 69.2 | 92.2 | |
| Conformal Social ChoiceDebate Round=1, Alpha=0.10, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.651 | 88.5 | 1.02 | 97.5 | 81 | |
| Conformal Social ChoiceDebate Round=3, Alpha=0.10, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.537 | 88.3 | 1 | 99.8 | 100 | |
| Conformal Social ChoiceDebate Round=2, Alpha=0.10, Model Ensemble=Haiku + DeepSeek-R1 + Qwen-3 32B2026.04 | 0.492 | 87.8 | 1 | 99.5 | 53.9 |