List your company

Safety Evaluation on AbstainQA (test)

14.7AccuracyPSO-Merging
-0.2763.6127.511.388May 28, 2026

Evaluation results

MethodLinks
PSO-MergingBackbone=Qwen2.5-1.5B2026.0514.7
EvoGMBackbone=Qwen2.5-1.5B2026.0512.1
Single BestBackbone=Qwen2.5-1.5B2026.0511.9
BaseBackbone=Qwen2.5-1.5B2026.0510.1
Model SwarmBackbone=Qwen2.5-1.5B2026.059.5
CMABackbone=Qwen2.5-1.5B2026.057.5
TABackbone=Qwen2.5-1.5B2026.056
Model SoupBackbone=Qwen2.5-1.5B2026.054.8
DAREBackbone=Qwen2.5-1.5B2026.054.3
TIESBackbone=Qwen2.5-1.5B2026.052.8
MTLBackbone=Qwen2.5-1.5B2026.050.3