trace_synthesis/tasks/680.json
yuyr a84d51a101 1. 增加r1生成综合策略代码和输出;
2. 增加tasks;
3. 增加analysis部分,对策略进行归纳分类,然后进行评测。
2025-04-17 17:40:15 +08:00

37 lines
904 B
JSON

{
"sites": [
"shopping_admin"
],
"task_id": 680,
"require_login": true,
"storage_state": "./.auth/shopping_admin_state.json",
"start_url": "http://localhost:28083/admin",
"geolocation": null,
"intent_template": "Lookup orders that are {{status}}",
"instantiation_dict": {
"status": "on hold"
},
"intent": "Lookup orders that are on hold",
"require_reset": false,
"eval": {
"eval_types": [
"url_match",
"program_html"
],
"reference_answers": null,
"reference_url": "http://localhost:28083/admin/sales/order/",
"program_html": [
{
"url": "last",
"locator": "document.querySelector(\"div.admin__data-grid-filters-current\").outerText",
"required_contents": {
"must_include": [
"On Hold"
]
}
}
],
"url_note": "GOLD in PRED"
},
"intent_template_id": 253
}