Web-based Shopping Simulation on WebShop (test)
34.6Success RateMASA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MASABackbone=Qwen3-32B2026.05 | 34.6 | 59.9 | 7.3 | |
| MASABackbone=Qwen3-14B2026.05 | 29.2 | 54.4 | 8 | |
| MASABackbone=Qwen3-8B2026.05 | 28.6 | 60.1 | 4.7 | |
| MASABackbone=Qwen3-4B2026.05 | 26.4 | 49.1 | 8.4 | |
| No SkillBackbone=Qwen3-4B2026.05 | 23 | 42.2 | 9.5 | |
| Base SkillBackbone=Qwen3-4B2026.05 | 19.4 | 34.8 | 11 | |
| DS-AdapterBackbone=Qwen3-4B2026.05 | 19.2 | 25.7 | 12.4 | |
| Base SkillBackbone=Qwen3-32B2026.05 | 7.2 | 24.2 | 12 | |
| No SkillBackbone=Qwen3-32B2026.05 | 6.6 | 35.2 | 9.9 | |
| Base SkillBackbone=Qwen3-8B2026.05 | 6 | 32.6 | 9.3 | |
| No SkillBackbone=Qwen3-8B2026.05 | 4.6 | 32.7 | 10 | |
| DS-AdapterBackbone=Qwen3-8B2026.05 | 4.4 | 18.2 | 12.6 | |
| DS-AdapterBackbone=Qwen3-32B2026.05 | 3.6 | 14.3 | 13.3 | |
| No SkillBackbone=Qwen3-14B2026.05 | 2.8 | 19.9 | 12.7 | |
| DS-AdapterBackbone=Qwen3-14B2026.05 | 2 | 12.6 | 13.6 | |
| Base SkillBackbone=Qwen3-14B2026.05 | 1.6 | 14.8 | 13.5 |