List your company

Hallucination Detection on MedicalQA

78.95AUROCCombined: PC + SE
52.9559.766.4573.2Mar 22, 2026

Evaluation results

MethodLinks
Combined: PC + SEBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0378.9594.42
Semantic EntropyBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0377.2490.12
Combined: PC + NLLBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0377.0593.42
PC ProbeBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0375.5592.13
Combined: PC + SE ProbeBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0374.9691.53
Semantic EntropyBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0370.1590.53
Sequence NLLBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0369.0585.34
Semantic EntropyBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0368.8586.63
Combined: PC + SEBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0368.1487.63
Combined: PC + SE ProbeBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0368.0488.23
PC ProbeBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0367.1488.43
Combined: PC + NLLBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0366.7586.14
Combined: PC + SEBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0366.6587.53
Combined: PC + SE ProbeBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0365.4587.03
Combined: PC + NLLBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0364.9587.73
Semantic EntropyBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0364.8585.03
SE ProbeBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0364.3585.24
PC ProbeBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0364.2586.83
Sequence NLLBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0363.5388.02
SE ProbeBase Model=Gemma 3 4B, Base Model Accuracy=17.3%2026.0363.4687.34
Combined: PC + SEBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0362.8485.93
SE ProbeBase Model=Llama 3.2 3B, Base Model Accuracy=22.0%2026.0362.6586.43
Sequence NLLBase Model=Qwen3 4B, Base Model Accuracy=23.3%2026.0361.3481.54
PC ProbeBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0356.9582.13
Combined: PC + SE ProbeBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0356.5581.83
Combined: PC + NLLBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0356.0581.43
Sequence NLLBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0355.1579.34
SE ProbeBase Model=Ministral 8B, Base Model Accuracy=23.3%2026.0353.9580.44