Household Manipulation on ALFWorld (Seen)
84.65Average RewardSkillReranker
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SkillRerankerBackbone Model=DeepSeek-v4-Flash2026.07 | 84.65 | 15.48 | |
| SkillRouterBackbone Model=DeepSeek-v4-Flash2026.07 | 80.72 | 16.31 | |
| SkillRerankerBackbone Model=Qwen3.6-27B2026.07 | 78.57 | 12.06 | |
| LLM-as-selectorBackbone Model=DeepSeek-v4-Flash2026.07 | 75.36 | 17.21 | |
| LLM-as-selectorBackbone Model=Qwen3.6-27B2026.07 | 74.29 | 12.61 | |
| SkillRouterBackbone Model=Qwen3.6-27B2026.07 | 72.5 | 12.89 | |
| Graph of SkillsBackbone Model=Qwen3.6-27B2026.07 | 70.72 | 13.62 | |
| Graph of SkillsBackbone Model=DeepSeek-v4-Flash2026.07 | 70.36 | 17.55 | |
| SkillRerankerBackbone Model=GPT-5.4-Mini2026.07 | 67.5 | 17.55 | |
| SkillRouterBackbone Model=GPT-5.4-Mini2026.07 | 62.5 | 17.64 | |
| Graph of SkillsBackbone Model=GPT-5.4-Mini2026.07 | 58.57 | 20.04 | |
| LLM-as-selectorBackbone Model=GPT-5.4-Mini2026.07 | 48.57 | 19.62 |