List your company

Hallucination Detection on SciQ (AUC, F95, AUP)

88.99AUCSIVR
24.749241.427158.10574.7829Apr 17, 2026

Evaluation results

MethodLinks
SIVRModel=Qwen-3-4B-Instruct2026.0488.9952.5693.04
SIVRModel=Qwen-3-4B-Instruct2026.0488.9952.5693.04
Lookback LensModel=Qwen-3-14B2026.0485.3853.1988.06
SIVRModel=Qwen-3-14B2026.0484.749.4487.75
SATMD + MSPModel=Qwen-3-4B-Instruct2026.0484.473.1391.09
SATMD + MSPModel=Qwen-3-4B-Instruct2026.0484.473.1391.09
SAPLMAModel=Qwen-3-4B-Instruct2026.0483.860.2689.45
SAPLMAModel=Qwen-3-4B-Instruct2026.0483.860.2689.45
SAPLMAModel=Qwen-3-14B2026.0481.4659.7784.6
Lookback LensModel=Qwen-3-4B-Instruct2026.0481.2686.988.05
Lookback LensModel=Qwen-3-4B-Instruct2026.0481.2686.988.05
SATMD + MSPModel=Qwen-3-14B2026.0477.5278.3887.2
SARModel=Qwen-3-4B-Instruct2026.0473.6586.1179.62
SARModel=Qwen-3-4B-Instruct2026.0473.6586.1179.62
Temp-ScaledModel=Qwen-3-4B-Instruct2026.0472.6576.5375.95
Temp-ScaledModel=Qwen-3-4B-Instruct2026.0472.6576.5375.95
PerplexityModel=Qwen-3-4B-Instruct2026.0471.8877.376.11
PerplexityModel=Qwen-3-4B-Instruct2026.0471.8877.376.11
SEModel=Qwen-3-4B-Instruct2026.0471.7483.7580.02
SEModel=Qwen-3-4B-Instruct2026.0471.7483.7580.02
Max ProbModel=Qwen-3-4B-Instruct2026.0471.2977.375.51
Max ProbModel=Qwen-3-4B-Instruct2026.0471.2977.375.51
EntropyModel=Qwen-3-4B-Instruct2026.0470.3778.5775.91
EntropyModel=Qwen-3-4B-Instruct2026.0470.3778.5775.91
Temp-ScaledModel=Qwen-3-14B2026.0461.2886.0462.94
PerplexityModel=Qwen-3-14B2026.0461.1286.0463.83
EntropyModel=Qwen-3-14B2026.0460.9886.9665.64
SEModel=Qwen-3-14B2026.0460.7984.2764.37
Max ProbModel=Qwen-3-14B2026.0460.5886.0463.17
SARModel=Qwen-3-14B2026.0458.394.3861.75
CoE-CModel=Qwen-3-14B2026.0457.1194.2860.38
CoE-RModel=Qwen-3-14B2026.0451.6995.6555.43
EnergyModel=Qwen-3-14B2026.0450.0610061.74
P(True)Model=Qwen-3-14B2026.0449.798.460.6
P(True)Model=Qwen-3-4B-Instruct2026.0449.6199.7466.59
P(True)Model=Qwen-3-4B-Instruct2026.0449.6199.7466.59
CoE-CModel=Qwen-3-4B-Instruct2026.0446.1495.4155.53
CoE-CModel=Qwen-3-4B-Instruct2026.0446.1495.4155.53
EnergyModel=Qwen-3-4B-Instruct2026.0441.4810059.92
EnergyModel=Qwen-3-4B-Instruct2026.0441.4810059.92
CoE-RModel=Qwen-3-4B-Instruct2026.0427.2210047.25
CoE-RModel=Qwen-3-4B-Instruct2026.0427.2210047.25