List your company

Test-Time Personalization on Truthful QA

99.6Creative Win RateT-POP
82.33686.81891.395.782Sep 29, 2025

Evaluation results

MethodLinks
T-POPModel=Mistral-7B2025.0999.695.472.396.5
AmuletModel=Llama-3.1-8B2025.0999.593.285.592.8
T-POPModel=Qwen2-7B2025.0998.5967999.1
T-POPModel=Llama-3.1-8B2025.0997.395.468.893.5
LAModel=Llama-3.1-8B2025.0997.296.574.795.3
LAModel=Mistral-7B2025.0994.590.170.391.9
PrefModel=Llama-3.1-8B2025.0994.187.371.982.7
AmuletModel=Mistral-7B2025.0993.378.770.888.2
BS16Model=Llama-3.1-8B2025.0992.386.776.982.6
PrefModel=Qwen2-7B2025.0990.178.288.981.9
LAModel=Qwen2-7B2025.0988.284.392.280.8
PrefModel=Mistral-7B2025.0985.479.177.485.8
BS16Model=Qwen2-7B2025.0985.479.590.181.2
AmuletModel=Qwen2-7B2025.0983.978.991.293
BS16Model=Mistral-7B2025.098377.580.582.9