Uncertainty Estimation on HotpotQA
0.739AUROCCloseness Centrality
Evaluation Results
| Method | Links | |
|---|---|---|
| Closeness CentralityBackbone=Mistral-7B2026.04 | 0.739 | |
| TotalBackbone=Mistral-7B2026.04 | 0.736 | |
| SC ScoreBackbone=Mistral-7B2026.04 | 0.711 | |
| SC Based VCBackbone=Mistral-7B2026.04 | 0.677 | |
| Mean Token EntropyBackbone=Mistral-7B2026.04 | 0.559 | |
| Token EntropyBackbone=Mistral-7B2026.04 | 0.558 | |
| Max Sequence Prob.Backbone=Mistral-7B2026.04 | 0.556 | |
| Self CertaintyBackbone=Mistral-7B2026.04 | 0.553 | |
| AleatoricBackbone=Mistral-7B2026.04 | 0.548 | |
| PerplexityBackbone=Mistral-7B2026.04 | 0.548 | |
| Max Token Prob.Backbone=Mistral-7B2026.04 | 0.546 | |
| SelfCheckGPTBackbone=Mistral-7B2026.04 | 0.5262 | |
| SemanticEntropyBackbone=Mistral-7B2026.04 | 0.526 | |
| SC + VCBackbone=Mistral-7B2026.04 | 0.51 | |
| Kernel Lang. Ent.Backbone=Mistral-7B2026.04 | 0.491 | |
| PTrueBackbone=Mistral-7B2026.04 | 0.417 |