AgentGridWorld-Hard / qualification /harbor-loop.json
bytkim's picture
AgentGridWorld Hard: nine rule-pressure gridworld tasks, with a local player
4a372d3 verified
Raw History Blame Contribute Delete
1.03 kB
{
"qualification": "local deterministic Harbor loop",
"tasks": [
{
"task": "clinic-hard-conflict-v5",
"calls": 24,
"status": "passed"
},
{
"task": "clinic-hard-impossible-v5",
"calls": 23,
"status": "passed"
},
{
"task": "clinic-hard-possible-v5",
"calls": 36,
"status": "passed"
},
{
"task": "greenhouse-hard-conflict-v5",
"calls": 26,
"status": "passed"
},
{
"task": "greenhouse-hard-impossible-v5",
"calls": 24,
"status": "passed"
},
{
"task": "greenhouse-hard-possible-v5",
"calls": 38,
"status": "passed"
},
{
"task": "warehouse-hard-conflict-v5",
"calls": 34,
"status": "passed"
},
{
"task": "warehouse-hard-impossible-v5",
"calls": 34,
"status": "passed"
},
{
"task": "warehouse-hard-possible-v5",
"calls": 34,
"status": "passed"
}
],
"provider_calls": 0,
"modal_builds": 0
}