trace_synthesis/tasks/150.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

33 lines
800 B
JSON

{
"sites": [
"shopping"
],
"task_id": 150,
"require_login": true,
"storage_state": "./.auth/shopping_state.json",
"start_url": "http://localhost:28082",
"geolocation": null,
"intent_template": "What is the {{option}} configuration of the {{product}} I bought {{time}}",
"instantiation_dict": {
"option": "price",
"product": "fake tree",
"time": "Jan 2023"
},
"intent": "What is the price configuration of the fake tree I bought Jan 2023",
"require_reset": false,
"eval": {
"eval_types": [
"string_match"
],
"reference_answers": {
"must_include": [
"260.69"
]
},
"reference_url": "",
"program_html": [],
"string_note": "",
"reference_answer_raw_annotation": "260.69"
},
"intent_template_id": 155
}