List your company

Reasoning on BoolQ

77.2Factual Accuracy (FA)SFT
64.267.57570.9574.325May 22, 2025

Evaluation results

MethodLinks
SFTBackbone=LLaMA2-7B2025.0577.233.2
SafeInstrBackbone=LLaMA2-7B2025.0576.829.4
SAPBackbone=LLaMA2-7B2025.0576.523
SAFTBackbone=LLaMA2-7B2025.057431.5
SaLoRABackbone=LLaMA2-7B2025.0573.527.4
LisaBackbone=LLaMA2-7B2025.0572.430.7
Base modelBackbone=LLaMA2-7B2025.0564.730.9