Web Action Generation Efficiency on Mind2Web (All)
363.6Time to Proposal StepsDuSAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DuSARModel=Llama3.1-70B2025.12 | 363.6 | 10.9 | |
| DuSARModel=Llama3.1-8B2025.12 | 375.9 | 12.5 | |
| DuSARModel=Gemma3-12B2025.12 | 392.5 | 13.4 | |
| DuSARModel=Llama2-7B2025.12 | 551.5 | 80.2 | |
| ReActModel=Llama3.1-70B2025.12 | 1,460.9 | 364.7 | |
| ReActModel=Gemma3-12B2025.12 | 1,543.4 | 12.7 | |
| ReActModel=Llama3.1-8B2025.12 | 1,687.2 | 634.3 | |
| SynapseModel=Gemma3-12B2025.12 | 1,764.8 | 19.8 | |
| SynapseModel=Llama3.1-70B2025.12 | 1,855.8 | 48.3 | |
| ReActModel=Llama2-7B2025.12 | 1,889.9 | 97.6 | |
| SynapseModel=Llama3.1-8B2025.12 | 1,976.9 | 48.5 | |
| SynapseModel=Llama2-7B2025.12 | 2,067.2 | 36.9 | |
| TRADModel=Llama3.1-70B2025.12 | 3,284.5 | 28.3 | |
| TRADModel=Llama3.1-8B2025.12 | 3,294.6 | 549.4 | |
| TRADModel=Gemma3-12B2025.12 | 3,370.1 | 19.4 | |
| TRADModel=Llama2-7B2025.12 | 3,598.6 | 162.7 |