trace_synthesis/tasks/128.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

31 lines
745 B
JSON

{
"sites": [
"shopping_admin"
],
"task_id": 128,
"require_login": true,
"storage_state": "./.auth/shopping_admin_state.json",
"start_url": "http://localhost:28083/admin",
"geolocation": null,
"intent_template": "What's the total number of items sold in the most recent {{k}} orders?",
"instantiation_dict": {
"k": "2"
},
"intent": "What's the total number of items sold in the most recent 2 orders?",
"require_reset": false,
"eval": {
"eval_types": [
"string_match"
],
"reference_answers": {
"must_include": [
"9"
]
},
"reference_url": "",
"program_html": [],
"string_note": "",
"reference_answer_raw_annotation": "9"
},
"intent_template_id": 1002
}