List your company

Response and Tool-Call Quality Evaluation on E-commerce Assistant Evaluation Set (Offline)

61.4TCRManualSkill
46.11250.08154.0558.019Jun 11, 2026

Evaluation results

MethodLinks
ManualSkillAdded Component=Human-designed skill baseline2026.0661.456.170.965.364.9
SkillChain (Full)Added Component=+ Dual-path evaluation & Body refinement2026.0661.372.282.362.772.2
SkillChainS1+S2Added Component=+ Routing failure mining & Description update2026.0655.363.476.554.167.2
NoSkillAdded Component=Production baseline (no Skill constraint)2026.0654.449.169.652.859.1
SkillChainS1Added Component=Auto Skill creation + human reflection2026.0646.758.976.653.762.5