Reliability Estimation on TriviaQA
83.4AUROCGLU
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GLUModel=Qwen2026.06 | 83.4 | — | 71.7 | |
| RAUQModel=Fanar2026.06 | 82.1 | — | 71.2 | |
| GLUModel=Fanar2026.06 | 82.1 | — | 73 | |
| GLUModel=Gemma2026.06 | 81.7 | — | 69.8 | |
| RAUQModel=Qwen2026.06 | 78.9 | — | 58.1 | |
| LogTokUModel=Qwen2026.06 | 78.6 | — | 60.6 | |
| LogTokUTarget Model=Qwen3-32B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 75.22 | 86.78 | — | |
| LogTokUModel=Fanar2026.06 | 70.6 | — | 44.5 | |
| RAUQModel=Gemma2026.06 | 70 | — | 41.6 | |
| DisAADTarget Model=Qwen3-32B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 69.35 | 81.25 | — | |
| DisAADTarget Model=GPT-4, Setting=Black-box, O(1) Complexity=single-sample2026.05 | 68.93 | 95.8 | — | |
| EntropyTarget Model=Qwen3-32B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 66.58 | 80.57 | — | |
| SETarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 65.8 | 82.91 | — | |
| DSETarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 65.28 | 81.62 | — | |
| DisAADTarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=single-sample2026.05 | 65.03 | 88.52 | — | |
| DisAADTarget Model=LLaMA2-70B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 65.03 | 88.52 | — | |
| LogTokUTarget Model=LLaMA2-70B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 64.4 | 86.62 | — | |
| LENTarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 64.24 | 84.21 | — | |
| ProbabilityTarget Model=Qwen3-32B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 63.78 | 79.12 | — | |
| ProbabilityTarget Model=LLaMA2-70B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 60.06 | 85.89 | — | |
| LogTokUModel=Gemma2026.06 | 59.7 | — | 22.4 | |
| EntropyTarget Model=LLaMA2-70B, Setting=White-box, O(1) Complexity=single-sample2026.05 | 59.32 | 85.56 | — | |
| SETarget Model=GPT-4, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 57.56 | 91.54 | — | |
| EigVTarget Model=GPT-4, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 55.91 | 90.15 | — | |
| EigVTarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 55.88 | 86.95 | — | |
| DSETarget Model=GPT-4, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 54.89 | 90.98 | — | |
| LENTarget Model=GPT-4, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 50.89 | 89.16 | — | |
| LESTarget Model=GPT-4, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 47.33 | 91.1 | — | |
| LESTarget Model=LLaMA2-70B, Setting=Black-box, O(1) Complexity=multi-sample2026.05 | 46.22 | 90.26 | — |