trace_synthesis/tasks/692.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

38 lines
1017 B
JSON

{
"sites": [
"shopping"
],
"task_id": 692,
"require_login": true,
"storage_state": "./.auth/shopping_state.json",
"start_url": "http://localhost:28082",
"geolocation": null,
"intent_template": "Draft an email to the shop owner via their contact us function for a coupon as {{reason}}",
"instantiation_dict": {
"reason": "I am a student"
},
"intent": "Draft an email to the shop owner via their contact us function for a coupon as I am a student",
"require_reset": false,
"eval": {
"eval_types": [
"url_match",
"program_html"
],
"reference_answers": null,
"reference_url": "http://localhost:28082/contact",
"program_html": [
{
"url": "last",
"locator": "document.querySelector('[title=\"What\u2019s on your mind?\"').value",
"required_contents": {
"must_include": [
"coupon",
"student"
]
}
}
],
"url_note": "GOLD in PRED"
},
"intent_template_id": 163
}