List your company

Hallucination Detection on WebQSP

82F1 ScoreKG-Guard
34.6846.96559.2571.535May 29, 2026

Evaluation results

MethodLinks
KG-GuardGraph encoder=GraphTransformer2026.0582————
KG-GuardGraph Architecture=GraphTransformer2026.058284.981.385.991.4
KG-GuardGraph encoder=GAT2026.0579.8————
KG-GuardGraph Architecture=GAT2026.0579.882.179.883.990.7
Most Frequent2026.0565.9————
Most Frequent2026.0565.949.110049.149.1
GPT 5 MiniInput modalities=G+RS+RT2026.0559.2————
GPT 5 MiniJudge Inputs=G+RS+RT2026.0559.285.345.369.365.5
GPT 5 MiniInput modalities=G2026.0555.9————
GPT 5 MiniJudge Inputs=G2026.0555.985.541.667.964.2
Random2026.0549.6————
Random2026.0549.649.449.950.349.2
SelfCheckNumber of samples=102026.0548.7————
SelfCheckN=102026.0548.7683860.856.3
SelfCheckNumber of samples=52026.0542.1————
SelfCheckN=52026.0542.168.730.45955
Llama 4 ScoutInput modalities=G+RS+RT2026.0541————
Llama 4 ScoutJudge Inputs=G+RS+RT2026.054179.627.66157.5
Llama 4 ScoutInput modalities=G2026.0536.7————
Llama 4 ScoutJudge Inputs=G2026.0536.766.725.357.153.5
Llama 4 ScoutInput modalities=G+RS2026.0536.5————
Llama 4 ScoutJudge Inputs=G+RS2026.0536.573.424.358.555