Hallucination Detection on MedMCQA
75.57AUCLookback Lens
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Lookback LensModel=Qwen-3-14B2026.04 | 75.57 | 65.71 | 59.93 | |
| EnergyModel=Qwen-3-14B2026.04 | 69.84 | 71.7 | 50.27 | |
| SAPLMAModel=Qwen-3-14B2026.04 | 69.59 | 81.13 | 54.39 | |
| Lookback LensModel=Qwen-3-4B-Instruct2026.04 | 69.01 | 69.53 | 47.78 | |
| Lookback LensModel=Qwen-3-4B-Instruct2026.04 | 69.01 | 69.53 | 47.78 | |
| SATMD + MSPModel=Qwen-3-4B-Instruct2026.04 | 68.75 | 85.5 | 50.8 | |
| SATMD + MSPModel=Qwen-3-4B-Instruct2026.04 | 68.75 | 85.5 | 50.8 | |
| CoE-CModel=Qwen-3-4B-Instruct2026.04 | 67.95 | 80.9 | 54.26 | |
| CoE-CModel=Qwen-3-4B-Instruct2026.04 | 67.95 | 80.9 | 54.26 | |
| EntropyModel=Qwen-3-14B2026.04 | 67.33 | 80.38 | 48.26 | |
| EntropyModel=Qwen-3-4B-Instruct2026.04 | 66.86 | 83.79 | 51.06 | |
| EntropyModel=Qwen-3-4B-Instruct2026.04 | 66.86 | 83.79 | 51.06 | |
| PerplexityModel=Qwen-3-14B2026.04 | 66.46 | 80.75 | 46.57 | |
| Max ProbModel=Qwen-3-14B2026.04 | 66.38 | 82.64 | 47.27 | |
| Max ProbModel=Qwen-3-4B-Instruct2026.04 | 66.37 | 83.63 | 50.67 | |
| Max ProbModel=Qwen-3-4B-Instruct2026.04 | 66.37 | 83.63 | 50.67 | |
| PerplexityModel=Qwen-3-4B-Instruct2026.04 | 66.35 | 84.11 | 50.39 | |
| PerplexityModel=Qwen-3-4B-Instruct2026.04 | 66.35 | 84.11 | 50.39 | |
| SEModel=Qwen-3-14B2026.04 | 66.24 | 82.54 | 50.49 | |
| Temp-ScaledModel=Qwen-3-4B-Instruct2026.04 | 65.96 | 82.99 | 50.14 | |
| Temp-ScaledModel=Qwen-3-4B-Instruct2026.04 | 65.96 | 82.99 | 50.14 | |
| Temp-ScaledModel=Qwen-3-14B2026.04 | 65.75 | 83.02 | 45.58 | |
| SIVRModel=Qwen-3-14B2026.04 | 64.71 | 82.35 | 49.13 | |
| SIVRModel=Qwen-3-4B-Instruct2026.04 | 63.62 | 72.44 | 49.3 | |
| SIVRModel=Qwen-3-4B-Instruct2026.04 | 63.62 | 72.44 | 49.3 | |
| SATMD + MSPModel=Qwen-3-14B2026.04 | 63.54 | 79.59 | 47.32 | |
| EnergyModel=Qwen-3-4B-Instruct2026.04 | 63.48 | 84.59 | 47.89 | |
| EnergyModel=Qwen-3-4B-Instruct2026.04 | 63.48 | 84.59 | 47.89 | |
| SARModel=Qwen-3-14B2026.04 | 62.75 | 83.05 | 54.98 | |
| CoE-RModel=Qwen-3-4B-Instruct2026.04 | 62.74 | 88.92 | 48.21 | |
| CoE-RModel=Qwen-3-4B-Instruct2026.04 | 62.74 | 88.92 | 48.21 | |
| SEModel=Qwen-3-4B-Instruct2026.04 | 62.31 | 87.2 | 48.35 | |
| SEModel=Qwen-3-4B-Instruct2026.04 | 62.31 | 87.2 | 48.35 | |
| SAPLMAModel=Qwen-3-4B-Instruct2026.04 | 62.28 | 82.4 | 48.94 | |
| SAPLMAModel=Qwen-3-4B-Instruct2026.04 | 62.28 | 82.4 | 48.94 | |
| SARModel=Qwen-3-4B-Instruct2026.04 | 58.87 | 92 | 42.18 | |
| SARModel=Qwen-3-4B-Instruct2026.04 | 58.87 | 92 | 42.18 | |
| P(True)Model=Qwen-3-14B2026.04 | 57.23 | 86.8 | 37.11 | |
| P(True)Model=Qwen-3-4B-Instruct2026.04 | 55.78 | 91.97 | 39.61 | |
| P(True)Model=Qwen-3-4B-Instruct2026.04 | 55.78 | 91.97 | 39.61 | |
| CoE-CModel=Qwen-3-14B2026.04 | 52.33 | 96.23 | 37.19 | |
| CoE-RModel=Qwen-3-14B2026.04 | 51.67 | 94.72 | 37.57 |