pydantic定义llm response数据模型

pydantic基于BaseModel定义的数据模型，能通过model_json_schema输出json解读。

示例如下

from pydantic import BaseModelclass semantic_group(BaseModel):semantic_unit:strentities:list[str]relationships:list[str]class text_decomposition(BaseModel):Output:list[semantic_group]class relationship_reconstraction(BaseModel):source:strrelationship:strtarget:strclass elements(BaseModel):title:strdescription:strtext_decomposition.model_json_schema()

输出如下

{'$defs': {'semantic_group': {'properties': {'semantic_unit': {'title': 'Semantic Unit',
     'type': 'string'},
    'entities': {'items': {'type': 'string'},
     'title': 'Entities',
     'type': 'array'},
    'relationships': {'items': {'type': 'string'},
     'title': 'Relationships',
     'type': 'array'}},
   'required': ['semantic_unit', 'entities', 'relationships'],
   'title': 'semantic_group',
   'type': 'object'}},
 'properties': {'Output': {'items': {'$ref': '#/$defs/semantic_group'},
   'title': 'Output',
   'type': 'array'}},
 'required': ['Output'],
 'title': 'text_decomposition',
 'type': 'object'}

reference

---

Pydantic 模型上的配置

https://docs.pydantic.org.cn/latest/concepts/config/