List your company

Web Action Generation Efficiency on Mind2Web Cross-Task

378.2Time to ProcedureDuSAR
249.7321,116.8911,984.052,851.209Dec 9, 2025

Evaluation results

MethodLinks
DuSARModel=Llama3.1-70B2025.12378.210.7
DuSARModel=Llama3.1-8B2025.12381.916.2
DuSARModel=Gemma3-12B2025.12393.411.3
DuSARModel=Llama2-7B2025.12539.580.5
ReActModel=Llama3.1-70B2025.121,482.7437.1
SynapseModel=Llama3.1-70B2025.121,545.818.5
ReActModel=Gemma3-12B2025.121,571.212.5
ReActModel=Llama3.1-8B2025.121,728626.3
SynapseModel=Gemma3-12B2025.121,777.218.5
SynapseModel=Llama3.1-8B2025.121,909.220.5
ReActModel=Llama2-7B2025.121,914.8111.1
SynapseModel=Llama2-7B2025.122,124.536.2
TRADModel=Llama3.1-70B2025.123,244.130
TRADModel=Llama3.1-8B2025.123,302657.8
TRADModel=Gemma3-12B2025.123,369.219.4
TRADModel=Llama2-7B2025.123,589.9109.3