List your company

Hallucination Detection on HaluEvalQA

89ROC-AUCSinkProbe
-3.06641620.83544244.737368.639158Apr 12, 2026Apr 13, 2026Apr 14, 2026Apr 15, 2026Apr 16, 2026Apr 17, 2026

Evaluation results

MethodLinks
SinkProbeLLM=Llama3.1-8B2026.0489
MTopDivLLM=Llama3.1-8B2026.0488.1
LookbackLensLLM=Llama3.1-8B2026.0487.9
LapEigvalLLM=Llama3.1-8B2026.0487.8
AttnEigvalLLM=Llama3.1-8B2026.0485
AttnLogDetLLM=Llama3.1-8B2026.0484.9
SinkProbeLLM=Mistral-Nemo2026.0484.9
SinkProbeLLM=Llama3.2-3B2026.0484.7
MTopDivLLM=Llama3.2-3B2026.0484.6
SinkProbeLLM=Phi3.52026.0484.6
LapEigvalLLM=Llama3.2-3B2026.0483.7
LookbackLensLLM=Mistral-Nemo2026.0483.7
MTopDivLLM=Mistral-Nemo2026.0483.7
LookbackLensLLM=Llama3.2-3B2026.0483.6
MTopDivLLM=Phi3.52026.0483.5
LapEigvalLLM=Mistral-Nemo2026.0483.1
LookbackLensLLM=Phi3.52026.0482.8
LapEigvalLLM=Phi3.52026.0482.2
AttnLogDetLLM=Llama3.2-3B2026.0481.7
AttnEigvalLLM=Llama3.2-3B2026.0481.7
AttnLogDetLLM=Phi3.52026.0480.5
AttnEigvalLLM=Phi3.52026.0480.3
AttnLogDetLLM=Mistral-Nemo2026.0479.7
AttnEigvalLLM=Mistral-Nemo2026.0479.1
AttnScoreLLM=Llama3.1-8B2026.0477
AttnScoreLLM=Llama3.2-3B2026.0473.7
AttnScoreLLM=Phi3.52026.0471.2
AttnScoreLLM=Mistral-Nemo2026.0468.8
HallucinationProbesDetector category=White-box, Probe=LoRA-KL probe on LLaMA-3.1-8B (lambda_KL = 0.05)2026.040.9493
RAGognizerDetector category=White-box, Proxy generator=Qwen3-4B-Instruct-25072026.040.9038
HDM-2-3BDetector category=Black-box2026.040.8385
SelfCheckGPT (NLI)Detector category=Black-box2026.040.8343
MiniCheck-7BDetector category=Black-box2026.040.8138
DeBERTa-v3 (Ent.)Detector category=Black-box2026.040.7832
LettuceDetect-LDetector category=Black-box2026.040.7702
HHEM-2.1-OpenDetector category=Black-box2026.040.77
Semantic EntropyDetector category=White-box2026.040.7407
DeBERTa-v3 (Con.)Detector category=Black-box2026.040.681
INSIDE (EigenScore)Detector category=White-box, Proxy generator=Qwen3-4B-Instruct-25072026.040.4746