Search Relevance Evaluation on Taobao AI Search All Queries
55.5Hit RateLEAPS-HR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LEAPS-HRBase Model=Qwen3-14B, Optimization Objective=HR, Training Strategy=RL2026.01 | 55.5 | 67 | 58 | |
| LEAPS-ERBase Model=Qwen3-14B, Optimization Objective=ER, Training Strategy=RL2026.01 | 54 | 72 | 62 | |
| LEAPS-GRBase Model=Qwen3-14B, Optimization Objective=GR, Training Strategy=RL2026.01 | 52 | 73.5 | 60 | |
| SFT (Qwen3-14B)Base Model=Qwen3-14B, Training Strategy=SFT2026.01 | 43.2 | 57.5 | 49.5 | |
| SFT (Tbstar-16A2B)Base Model=Tbstar-16A2B, Training Strategy=SFT2026.01 | 43 | 57.2 | 49.4 | |
| Production Baseline2026.01 | 41.5 | 55.2 | 47.6 |