List your company

Hallucination Detection on HUB (test)

76.8AlgorithmicHALT-L
22.137636.328850.5264.7112Feb 2, 2026

Evaluation results

MethodLinks
HALT-LBase Model=Llama 3.1 8B2026.0276.860.1747.6756.6772.8972.7174.0766.9365.476.9267.0163.03
HALT-QBase Model=Llama 3.1 8B2026.0232.6858.639.7141.327362.968.7870.7549.7858.4562.7455.6
∆HdecBase Model=Llama 3.1 8B2026.0226.4437.4162.0335.1742.1866.9549.4245.0324.9244.5148.0343.41
HaltsBase Model=Llama 3.1 8B2026.0224.4836.4766.6734.2739.1561.6343.432.9924.5944.5142.940.81
PPLBase Model=Llama 3.1 8B2026.0224.2435.0943.0733.2539.1544.1428.3124.4924.5944.5133.9534.08
HoverallBase Model=Llama 3.1 8B2026.0224.2434.5538.1431.9739.154225.7324.4924.5944.5132.8632.94
LettuceBase Model=Llama 3.1 8B2026.0224.2441.538.1441.0683.3841.877.359.7133.3644.516448.5