Uncertainty Estimation on JudgeBench (test)
71.53AUROCEnergy
Evaluation Results
| Method | Links | |
|---|---|---|
| EnergyBackbone Model=GPT-OSS OSS 20B2025.12 | 71.53 | |
| ProbeBackbone Model=QWEN 32B2025.12 | 71.17 | |
| VerbalisedBackbone Model=QWEN 32B2025.12 | 69.2 | |
| CoERBackbone Model=GPT-OSS OSS 20B2025.12 | 69.12 | |
| ProbeBackbone Model=J1 QWEN 32B2025.12 | 69.05 | |
| VerbalisedBackbone Model=J1 QWEN 32B2025.12 | 68.86 | |
| CoECBackbone Model=GPT-OSS OSS 20B2025.12 | 68.46 | |
| EnergyBackbone Model=QWEN 32B2025.12 | 67.79 | |
| Sem. EntropyBackbone Model=J1 QWEN 32B2025.12 | 66.87 | |
| Sem. EntropyBackbone Model=QWEN 32B2025.12 | 65.63 | |
| VerbalisedBackbone Model=GPT-OSS OSS 20B2025.12 | 64.47 | |
| Sem. EntropyBackbone Model=GPT-OSS OSS 20B2025.12 | 64.42 | |
| EntropyBackbone Model=J1 QWEN 32B2025.12 | 64.33 | |
| MaxProbBackbone Model=J1 QWEN 32B2025.12 | 64.27 | |
| PerplexityBackbone Model=J1 QWEN 32B2025.12 | 63.88 | |
| TempSclBackbone Model=J1 QWEN 32B2025.12 | 63.74 | |
| CoTKBackbone Model=GPT-OSS OSS 20B2025.12 | 63.72 | |
| ProbeBackbone Model=J1 LLaMA 70B2025.12 | 63.17 | |
| CoTKBackbone Model=J1 QWEN 32B2025.12 | 62.9 | |
| VerbalisedBackbone Model=J1 LLaMA 70B2025.12 | 62.04 | |
| ProbeBackbone Model=GPT-OSS OSS 20B2025.12 | 62.03 | |
| Sem. EntropyBackbone Model=J1 LLaMA 70B2025.12 | 60.63 | |
| CoERBackbone Model=J1 QWEN 32B2025.12 | 60.36 | |
| CoECBackbone Model=J1 QWEN 32B2025.12 | 60.26 | |
| CoTKBackbone Model=J1 LLaMA 70B2025.12 | 59.98 | |
| PerplexityBackbone Model=J1 LLaMA 70B2025.12 | 59.62 | |
| MaxProbBackbone Model=J1 LLaMA 70B2025.12 | 59.61 | |
| EntropyBackbone Model=J1 LLaMA 70B2025.12 | 59.6 | |
| TempSclBackbone Model=J1 LLaMA 70B2025.12 | 59.6 | |
| CoECBackbone Model=J1 LLaMA 70B2025.12 | 59.13 | |
| ProbeBackbone Model=LLaMA 70B2025.12 | 58.14 | |
| Sem. EntropyBackbone Model=LLaMA 70B2025.12 | 58.13 | |
| VerbalisedBackbone Model=LLaMA 70B2025.12 | 57.95 | |
| ProbeBackbone Model=J1 LLaMA 8B2025.12 | 55.36 | |
| Sem. EntropyBackbone Model=LLaMA 8B2025.12 | 54.71 | |
| Sem. EntropyBackbone Model=J1 LLaMA 8B2025.12 | 54.19 | |
| VerbalisedBackbone Model=J1 LLaMA 8B2025.12 | 53.5 | |
| EnergyBackbone Model=J1 LLaMA 8B2025.12 | 53.34 | |
| ProbeBackbone Model=LLaMA 8B2025.12 | 52.38 | |
| CoECBackbone Model=LLaMA 70B2025.12 | 52.24 | |
| CoERBackbone Model=LLaMA 8B2025.12 | 52.06 | |
| TempSclBackbone Model=LLaMA 70B2025.12 | 51.55 | |
| TempSclBackbone Model=QWEN 32B2025.12 | 51.53 | |
| MaxProbBackbone Model=LLaMA 70B2025.12 | 51.48 | |
| MaxProbBackbone Model=QWEN 32B2025.12 | 51.47 | |
| PerplexityBackbone Model=LLaMA 70B2025.12 | 51.31 | |
| PerplexityBackbone Model=QWEN 32B2025.12 | 51.2 | |
| CoTKBackbone Model=LLaMA 70B2025.12 | 51.11 | |
| EntropyBackbone Model=LLaMA 70B2025.12 | 51.08 | |
| CoTKBackbone Model=J1 LLaMA 8B2025.12 | 50.87 | |
| EntropyBackbone Model=J1 LLaMA 8B2025.12 | 50.85 | |
| PerplexityBackbone Model=J1 LLaMA 8B2025.12 | 50.81 | |
| EntropyBackbone Model=QWEN 32B2025.12 | 50.79 | |
| TempSclBackbone Model=J1 LLaMA 8B2025.12 | 50.72 | |
| MaxProbBackbone Model=J1 LLaMA 8B2025.12 | 50.67 | |
| CoECBackbone Model=J1 LLaMA 8B2025.12 | 50.56 | |
| EnergyBackbone Model=LLaMA 70B2025.12 | 50.07 | |
| EnergyBackbone Model=J1 LLaMA 70B2025.12 | 50 | |
| CoECBackbone Model=LLaMA 8B2025.12 | 49.99 | |
| CoERBackbone Model=J1 LLaMA 8B2025.12 | 49.98 | |
| CoERBackbone Model=J1 LLaMA 70B2025.12 | 49.82 | |
| EnergyBackbone Model=J1 QWEN 32B2025.12 | 49.34 | |
| CoTKBackbone Model=QWEN 32B2025.12 | 48.8 | |
| EnergyBackbone Model=LLaMA 8B2025.12 | 47.99 | |
| CoERBackbone Model=LLaMA 70B2025.12 | 47.81 | |
| VerbalisedBackbone Model=LLaMA 8B2025.12 | 46.83 | |
| EntropyBackbone Model=GPT-OSS OSS 20B2025.12 | 46.82 | |
| EntropyBackbone Model=LLaMA 8B2025.12 | 46.81 | |
| MaxProbBackbone Model=LLaMA 8B2025.12 | 46.77 | |
| CoTKBackbone Model=LLaMA 8B2025.12 | 46.69 | |
| TempSclBackbone Model=LLaMA 8B2025.12 | 46.54 | |
| PerplexityBackbone Model=LLaMA 8B2025.12 | 46.44 | |
| CoERBackbone Model=QWEN 32B2025.12 | 46.25 | |
| CoECBackbone Model=QWEN 32B2025.12 | 45.99 | |
| MaxProbBackbone Model=GPT-OSS OSS 20B2025.12 | 43.54 | |
| PerplexityBackbone Model=GPT-OSS OSS 20B2025.12 | 43.15 | |
| TempSclBackbone Model=GPT-OSS OSS 20B2025.12 | 42.11 |