Web Navigation Task Completion on Mind2Web Cross-task
64.6Success RatePolySkill (+Update)
Evaluation Results
| Method | Links | |
|---|---|---|
| PolySkill (+Update)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 64.6 | |
| PolySkill (+Update)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 63.2 | |
| PolySkill (Same sub-dom.)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 62.3 | |
| ASI (+Update)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 62.1 | |
| PolySkill (Same domain)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 62 | |
| PolySkill (All Skills)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 61.3 | |
| ASI (Same sub-domain)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 61.2 | |
| ASI (Same domain)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 60.9 | |
| ASI (All skills)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 60.3 | |
| ASI (+Update)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 59.4 | |
| BaselineBackbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 59.1 | |
| PolySkill (Same sub-dom.)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 58.6 | |
| PolySkill (Same domain)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 58.3 | |
| ASI (Same sub-domain)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 56.3 | |
| PolySkill (All Skills)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 55.4 | |
| ASI (Same domain)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 55.2 | |
| BaselineBackbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 53.8 | |
| ASI (All skills)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 52.3 |