trace_synthesis/tasks/529.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

40 lines
1.3 KiB
JSON

{
"sites": [
"shopping"
],
"task_id": 529,
"require_login": true,
"storage_state": "./.auth/shopping_state.json",
"start_url": "http://localhost:28082",
"geolocation": null,
"intent_template": "Draft a refund message via their \"contact us\" form for the {{product}} I bought {{time}}. It broke after three days of use. The shop requires the order id, the reason and the amount to refund in the message. Don't submit yet",
"instantiation_dict": {
"product": "bluetooth speaker",
"time": "Feb 2023"
},
"intent": "Draft a refund message via their \"contact us\" form for the bluetooth speaker I bought Feb 2023. It broke after three days of use. The shop requires the order id, the reason and the amount to refund in the message. Don't submit yet",
"require_reset": false,
"eval": {
"eval_types": [
"program_html"
],
"reference_answers": null,
"reference_url": "",
"program_html": [
{
"url": "last",
"locator": "document.querySelector('[title=\"What\u2019s on your mind?\"').value",
"required_contents": {
"must_include": [
"refund",
"broke",
"three days of use",
"000000148",
"169.95"
]
}
}
]
},
"intent_template_id": 154
}