Agents & Tool Use
VisualWebArena
Released automatic and human success judgments of multimodal web-agent trajectories on VisualWebArena.
200items
5subjects
AgentRewardBench custom dataset terms: use must be lawful and respect third-party rights and applicable third-party conditions; the source describes research use under applicable fair-use provisions. Derivatives must retain those terms, captured in raw/README.md. The VisualWebArena code's MIT license is not asserted to license these separately released trajectories.license
agents_and_tool_usedomain
textmodality
gui_screenshotmodality
item-level responses released
Saturation status: Yes
Response matrix
Loading response matrix…
Correct (1)Incorrect (0)Unobserved
Scale: 1 = correct · 0 = incorrect