List your company

Question Answering on OfficeQA (test)

55.32AccuracyMetaSkill-Evolve
30.838437.194243.5549.9058Jul 6, 2026

Evaluation results

MethodLinks
MetaSkill-EvolveBackbone=Gemma-4 31B2026.0755.32
Single-LevelBackbone=Gemma-4 31B2026.0748.94
Static skillBackbone=Gemma-4 31B2026.0736.09
No-SkillBackbone=Gemma-4 31B2026.0731.78