Uncertainty Estimation on MMLU AutoGen (test)
0.7315AUROCMATU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MATUBase LLM=Qwen2.5-7B2026.04 | 0.7315 | 0.8833 | |
| SAUP-MultipleBase LLM=Qwen2.5-7B2026.04 | 0.7193 | 0.8589 | |
| Eigv(Agre)-finalBase LLM=Qwen2.5-7B2026.04 | 0.7105 | 0.8617 | |
| Eigv(Agre)-WholeBase LLM=Qwen2.5-7B2026.04 | 0.6867 | 0.8516 | |
| P(true)Base LLM=Qwen2.5-7B2026.04 | 0.6556 | 0.8363 | |
| SAUP-SingleBase LLM=Qwen2.5-7B2026.04 | 0.6484 | 0.8552 | |
| MATUBase LLM=GPT-4o2026.04 | 0.6277 | 0.5841 | |
| Eigv(Agre)-finalBase LLM=GPT-4o2026.04 | 0.5981 | 0.5649 | |
| MATUBase LLM=Llama3.1-8B2026.04 | 0.5954 | 0.4745 | |
| P(true)Base LLM=GPT-4o2026.04 | 0.5802 | 0.5528 | |
| P(true)Base LLM=Llama3.1-8B2026.04 | 0.5775 | 0.4368 | |
| Eigv(Agre)-WholeBase LLM=GPT-4o2026.04 | 0.5759 | 0.5438 | |
| Eigv(Agre)-finalBase LLM=Llama3.1-8B2026.04 | 0.5521 | 0.4288 | |
| Eigv(Agre)-WholeBase LLM=Llama3.1-8B2026.04 | 0.5316 | 0.3762 | |
| SAUP-SingleBase LLM=Llama3.1-8B2026.04 | 0.5138 | 0.3031 | |
| SAUP-MultipleBase LLM=Llama3.1-8B2026.04 | 0.5018 | 0.2973 |