List your company

Truthfulness Evaluation on TruthfulQA (NLL)

1.62NLLPPT (Distribution) + TS
1.49242.35373.2154.0763May 26, 2026

Evaluation results

MethodLinks
PPT (Distribution) + TSBackbone=LLAMA3-8B, Temperature Scaling=true2026.051.62
Base + TSBackbone=LLAMA3-8B, Temperature Scaling=true2026.051.76
PPT (Distribution)Backbone=LLAMA3-8B, Temperature Scaling=false2026.051.78
PPT (Mean)Backbone=LLAMA3-8B, Temperature Scaling=false2026.052.36
PPT (Distribution)Backbone=QWEN2-7B, Temperature Scaling=false2026.053.04
PPT (Distribution) + TSBackbone=QWEN2-7B, Temperature Scaling=true2026.053.16
BaseBackbone=LLAMA3-8B, Temperature Scaling=false2026.053.49
PPT (Mean)Backbone=QWEN2-7B, Temperature Scaling=false2026.054.48
BaseBackbone=QWEN2-7B, Temperature Scaling=false2026.054.81
Base + TSBackbone=QWEN2-7B, Temperature Scaling=true2026.054.81