List your company

Open-domain Question Answering on PopQA (test)

51.26EM AccuracyOriginal (Unsafe)
3.669616.024828.3840.7352Mar 18, 2026Mar 28, 2026Apr 8, 2026Apr 19, 2026Apr 29, 2026May 10, 2026May 21, 2026

Evaluation results

MethodLinks
Original (Unsafe)Backbone=Llama-3.2-3B2026.0351.26
SEAL-TAGBackbone=Llama-3.2-3B2026.0351.07
Few-Shot PromptingBackbone=Llama-3.2-3B2026.0348.57
Eraser4RAGBackbone=Llama-3.2-3B2026.0343.61
P3DeferBackbone=Llama-3.2-3B2026.0331.72
EVE-Agent2026.0530
PrivacyMindBackbone=Llama-3.2-3B2026.0328.58
DPVoteRAGBackbone=Llama-3.2-3B2026.0326.19
Dr. Zero2026.0513.4
Initialsearch=disabled2026.058.1
Initialsearch=enabled2026.055.5