Uncertainty Quantification on TQA 16 additional samples (test)
94.1AUROCClustered Self-Assessment
Evaluation Results
| Method | Links | |
|---|---|---|
| Clustered Self-AssessmentBackbone=Qwen2.5-32B2026.06 | 94.1 | |
| Clustered Self-AssessmentBackbone=Qwen2.5-14B2026.06 | 93.9 | |
| Clustered Self-AssessmentBackbone=Qwen3-14B2026.06 | 93.8 | |
| Clustered Self-AssessmentBackbone=Qwen3-30B-A3B2026.06 | 93.5 | |
| Clustered Self-AssessmentBackbone=Qwen2.5-7B2026.06 | 93.1 | |
| Clustered Self-AssessmentBackbone=Gemma-3-12B2026.06 | 93.1 | |
| Clustered Self-AssessmentBackbone=Gemma-3-27B2026.06 | 92.6 | |
| DegBackbone=Gemma-3-12B2026.06 | 90.6 | |
| sentenceSARBackbone=Qwen3-14B2026.06 | 89.9 | |
| LN EntropyBackbone=Qwen3-14B2026.06 | 89.8 | |
| SARBackbone=Qwen3-14B2026.06 | 89.8 | |
| ProbabilityBackbone=Qwen3-14B2026.06 | 89.7 | |
| tokenSARBackbone=Gemma-3-12B2026.06 | 89.7 | |
| PerplexityBackbone=Qwen3-14B2026.06 | 89.5 | |
| PerplexityBackbone=Gemma-3-12B2026.06 | 89.5 | |
| Predictive EntropyBackbone=Qwen3-14B2026.06 | 89.5 | |
| PerplexityBackbone=Gemma-3-27B2026.06 | 89.4 | |
| DegBackbone=Gemma-3-27B2026.06 | 89.3 | |
| tokenSARBackbone=Gemma-3-27B2026.06 | 89.1 | |
| ProbabilityBackbone=Qwen2.5-7B2026.06 | 89 | |
| P(True)Backbone=Qwen2.5-32B2026.06 | 89 | |
| sentenceSARBackbone=Qwen2.5-7B2026.06 | 88.9 | |
| PerplexityBackbone=Qwen2.5-7B2026.06 | 88.8 | |
| LN EntropyBackbone=Qwen2.5-7B2026.06 | 88.7 | |
| EigVBackbone=Qwen3-14B2026.06 | 88.7 | |
| Predictive EntropyBackbone=Qwen2.5-7B2026.06 | 88.6 | |
| P(True)Backbone=Qwen3-14B2026.06 | 88.6 | |
| SARBackbone=Qwen2.5-7B2026.06 | 88.6 | |
| sentenceSARBackbone=Qwen2.5-32B2026.06 | 88.5 | |
| SARBackbone=Qwen2.5-32B2026.06 | 88.4 | |
| EigVBackbone=Gemma-3-12B2026.06 | 88.4 | |
| ProbabilityBackbone=Qwen2.5-32B2026.06 | 88.3 | |
| ProbabilityBackbone=Gemma-3-27B2026.06 | 88.3 | |
| sentenceSARBackbone=Qwen2.5-14B2026.06 | 88.3 | |
| SARBackbone=Qwen2.5-14B2026.06 | 88.3 | |
| SARBackbone=Qwen3-30B-A3B2026.06 | 88.3 | |
| ProbabilityBackbone=Qwen2.5-14B2026.06 | 88.2 | |
| LN EntropyBackbone=Qwen2.5-32B2026.06 | 88.2 | |
| EigVBackbone=Gemma-3-27B2026.06 | 88.2 | |
| Predictive EntropyBackbone=Qwen2.5-32B2026.06 | 88.1 | |
| PerplexityBackbone=Qwen2.5-14B2026.06 | 88 | |
| LN EntropyBackbone=Qwen2.5-14B2026.06 | 88 | |
| PerplexityBackbone=Qwen2.5-32B2026.06 | 87.9 | |
| Predictive EntropyBackbone=Qwen2.5-14B2026.06 | 87.9 | |
| EigVBackbone=Qwen2.5-7B2026.06 | 87.9 | |
| ProbabilityBackbone=Gemma-3-12B2026.06 | 87.8 | |
| LN EntropyBackbone=Qwen3-30B-A3B2026.06 | 87.7 | |
| EigVBackbone=Qwen2.5-32B2026.06 | 87.7 | |
| PerplexityBackbone=Qwen3-30B-A3B2026.06 | 87.6 | |
| P(True)Backbone=Qwen3-30B-A3B2026.06 | 87.6 | |
| ProbabilityBackbone=Qwen3-30B-A3B2026.06 | 87.5 | |
| sentenceSARBackbone=Qwen3-30B-A3B2026.06 | 87.5 | |
| Semantic EntropyBackbone=Qwen3-14B2026.06 | 87.5 | |
| EccBackbone=Gemma-3-12B2026.06 | 87.5 | |
| SARBackbone=Gemma-3-12B2026.06 | 87.4 | |
| Eigen ScoreBackbone=Qwen3-14B2026.06 | 87.1 | |
| Semantic EntropyBackbone=Qwen2.5-32B2026.06 | 87.1 | |
| EccBackbone=Gemma-3-27B2026.06 | 87.1 | |
| DegBackbone=Qwen3-14B2026.06 | 87.1 | |
| Predictive EntropyBackbone=Qwen3-30B-A3B2026.06 | 87 | |
| EigVBackbone=Qwen3-30B-A3B2026.06 | 87 | |
| SARBackbone=Gemma-3-27B2026.06 | 86.9 | |
| P(True)Backbone=Qwen2.5-14B2026.06 | 86.8 | |
| tokenSARBackbone=Qwen3-14B2026.06 | 86.8 | |
| Semantic EntropyBackbone=Qwen2.5-7B2026.06 | 86.8 | |
| EigVBackbone=Qwen2.5-14B2026.06 | 86.7 | |
| Eigen ScoreBackbone=Qwen2.5-7B2026.06 | 86.5 | |
| DegBackbone=Qwen2.5-7B2026.06 | 86.2 | |
| Semantic EntropyBackbone=Gemma-3-12B2026.06 | 86 | |
| Semantic EntropyBackbone=Qwen3-30B-A3B2026.06 | 85.9 | |
| DegBackbone=Qwen2.5-32B2026.06 | 85.9 | |
| tokenSARBackbone=Qwen2.5-7B2026.06 | 85.7 | |
| LN EntropyBackbone=Gemma-3-12B2026.06 | 85.6 | |
| Eigen ScoreBackbone=Qwen2.5-32B2026.06 | 85.6 | |
| sentenceSARBackbone=Gemma-3-12B2026.06 | 85.6 | |
| Semantic EntropyBackbone=Qwen2.5-14B2026.06 | 85.5 | |
| tokenSARBackbone=Qwen2.5-14B2026.06 | 85.3 | |
| sentenceSARBackbone=Gemma-3-27B2026.06 | 85.3 | |
| tokenSARBackbone=Qwen3-30B-A3B2026.06 | 85.1 | |
| Eigen ScoreBackbone=Qwen2.5-14B2026.06 | 85 | |
| EccBackbone=Qwen3-14B2026.06 | 85 | |
| DegBackbone=Qwen3-30B-A3B2026.06 | 84.9 | |
| Predictive EntropyBackbone=Gemma-3-12B2026.06 | 84.8 | |
| Eigen ScoreBackbone=Qwen3-30B-A3B2026.06 | 84.8 | |
| DegBackbone=Qwen2.5-14B2026.06 | 84.8 | |
| LN EntropyBackbone=Gemma-3-27B2026.06 | 84.6 | |
| tokenSARBackbone=Qwen2.5-32B2026.06 | 84.6 | |
| Predictive EntropyBackbone=Gemma-3-27B2026.06 | 84.5 | |
| Semantic EntropyBackbone=Gemma-3-27B2026.06 | 84.5 | |
| EccBackbone=Qwen2.5-7B2026.06 | 84.1 | |
| EccBackbone=Qwen2.5-32B2026.06 | 83.5 | |
| P(True)Backbone=Qwen2.5-7B2026.06 | 83.2 | |
| Eigen ScoreBackbone=Gemma-3-12B2026.06 | 82.6 | |
| EccBackbone=Qwen2.5-14B2026.06 | 82.5 | |
| Eigen ScoreBackbone=Gemma-3-27B2026.06 | 82.4 | |
| EccBackbone=Qwen3-30B-A3B2026.06 | 82.4 | |
| P(True)Backbone=Gemma-3-27B2026.06 | 78.9 | |
| P(True)Backbone=Gemma-3-12B2026.06 | 73.7 |