Web Navigation Task Completion on Mind2Web Cross-domain
68.3Success RatePolySkill (+Update)
Evaluation Results
| Method | Links | |
|---|---|---|
| PolySkill (+Update)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 68.3 | |
| ASI (+Update)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 67.3 | |
| ASI (All skills)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 66.9 | |
| PolySkill (All Skills)Backbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 66.4 | |
| BaselineBackbone=Claude-3.7-Sonnet, Training Tasks=1009 tasks2025.10 | 66.2 | |
| PolySkill (+Update)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 63.4 | |
| BaselineBackbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 62.3 | |
| ASI (+Update)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 62.1 | |
| PolySkill (All Skills)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 60.1 | |
| ASI (All skills)Backbone=GPT-4.1, Training Tasks=1009 tasks2025.10 | 57.3 |