II.
Benchmark JSON
Structured · livebenchmark:visualwebarena
VisualWebArena json
Inspect the normalized record payload exactly as the atlas UI reads it.
{
"id": "benchmark:visualwebarena",
"_kind": "Benchmark",
"_file": "benchmarks/benchmarks/benchmarks-browser.yaml",
"_cluster": "benchmarks",
"attributes": {
"displayName": "VisualWebArena",
"homepageUrl": "https://jykoh.com/vwa",
"kind": "web-agent",
"targetsKind": "AgentVersion",
"description": "Multimodal extension of WebArena adding visually grounded web tasks\n(image search, classifieds, shopping with screenshots) to test\nvision-language web agents.\n"
},
"outgoingEdges": [
{
"from": "benchmark:visualwebarena",
"to": "skill-area:browser-automation",
"kind": "covers",
"attributes": {
"attributes": {
"coverage": "partial",
"weight": 0.3
}
}
},
{
"from": "benchmark:visualwebarena",
"to": "skill-area:vision-extraction",
"kind": "covers",
"attributes": {
"attributes": {
"coverage": "partial",
"weight": 0.3
}
}
},
{
"from": "benchmark:visualwebarena",
"to": "domain:web-development",
"kind": "applies_to",
"attributes": {}
}
],
"incomingEdges": []
}