List your company
ResearchFormerly WizWand

Element Grounding on Multimodal-Mind2Web Cross-Task

50.7Element AccuracyUGround-V1-7B
26.05232.45138.8545.249Oct 7, 2024
Updated 8mo ago

Evaluation results

MethodLinks
UGround-V1-7BInput=Image (SeeAct-V), Planner=GPT-4o2024.1050.7
UGround-V1-2BInput=Image (SeeAct-V), Planner=GPT-4o2024.1048.6
UGround-V1-7BInput=Image (SeeAct-V), Planner=GPT-4o2024.1048.5
UGroundInput=Image (SeeAct-V), Planner=GPT-4o2024.1047.7
UGround-V1-2BInput=Image (SeeAct-V), Planner=GPT-4o2024.1047.7
UGroundInput=Image (SeeAct-V), Planner=GPT-4o2024.1046.6
ChoiceInput=Image + Text, Planner=GPT-42024.1046.4
UGroundInput=Image (SeeAct-V), Planner=GPT-42024.1045.1
UGroundInput=Image (SeeAct-V), Planner=GPT-42024.1044.6
ChoiceInput=Image + Text, Planner=GPT-42024.1042.4
SeeClickInput=Image (SeeAct-V), Planner=GPT-4o2024.1033.5
SeeClickInput=Image (SeeAct-V), Planner=GPT-4o2024.1032.1
SeeClickInput=Image (SeeAct-V), Planner=GPT-42024.1030.7
SeeClickInput=Image (SeeAct-V), Planner=GPT-42024.1029.7
SoMInput=Image + Text, Planner=GPT-42024.1029.6
SoMInput=Image + Text, Planner=GPT-42024.1027