Web Navigation / Skill Induction on Mind2Web Cross-Domain GPT-4.1
63.4Success Rate (SR)PolySkill
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PolySkillLibrary Regime=+Update, Base LLM=GPT-4.12026.06 | 63.4 | 6.9 | 56 | 31 | |
| SKILLMIGRATORLibrary Regime=+Update, Base LLM=GPT-4.12026.06 | 63 | 6.2 | 49 | 35.4 | |
| ReActLibrary Regime=Static, Base LLM=GPT-4.12026.06 | 62.3 | 8 | — | — | |
| ASILibrary Regime=+Update, Base LLM=GPT-4.12026.06 | 62.1 | 7.4 | 66 | 23.6 | |
| PolySkillLibrary Regime=Static, Base LLM=GPT-4.12026.06 | 60.1 | 7.2 | 36 | 22.7 | |
| SKILLMIGRATORLibrary Regime=Static, Base LLM=GPT-4.12026.06 | 59.4 | 6.6 | 31 | 28.1 | |
| ASILibrary Regime=Static, Base LLM=GPT-4.12026.06 | 57.3 | 7.6 | 33 | 17.4 |