KeyError: 'function_call'

需要结果字段有function_call

失败情况

gpt 3.5 /4/4.1和deepseek都试过 都有问题 成功率很低

感觉是experimental包不稳定

下面是原创的改写方法 舍弃了langchain_experimental包的create_openai_data_generator

import os
from typing import List

from langchain_core.prompts import PromptTemplate, FewShotPromptTemplate
from langchain_experimental.tabular_synthetic_data.prompts import SYNTHETIC_FEW_SHOT_PREFIX, SYNTHETIC_FEW_SHOT_SUFFIX
from langchain_openai import ChatOpenAI
from pydantic.v1 import Field, BaseModel

os.environ['http_proxy'] = '127.0.0.1:7777'
os.environ['https_proxy'] = '127.0.0.1:7777'

# os.environ["LANGCHAIN_TRACING_V2"] = "true"
os.environ["LANGSMITH_TRACING"] = "true"
os.environ["LANGCHAIN_PROJECT"] = "LangchainDemo"
os.environ["LANGCHAIN_API_KEY"] = ''

# 聊天机器人案例
# 创建模型
model = ChatOpenAI(model='deepseek-r1', api_key="sk-",
                   base_url="", temperature=1)


class MedicalBilling(BaseModel):
    patient_id: int = Field(..., description="患者ID，6位数字")
    patient_name: str = Field(..., description="患者姓名，中文姓名")
    diagnosis_code: str = Field(..., description="诊断代码，符合ICD-10格式")
    procedure_code: str = Field(..., description="医疗操作代码，5位数字")
    total_charge: float = Field(..., description="总费用（美元），保留1位小数")
    insurance_claim_amount: float = Field(..., description="保险理赔金额（美元），保留1位小数")


class MedicalBillingList(BaseModel):
    medical_billings: List[MedicalBilling]


examples = [
    {
        "example": "Patient ID: 123456, Patient Name: 张娜, Diagnosis Code: J20.9, Procedure Code: 99203, Total Charge: $500, Insurance Claim Amount: $350"
    },
    {
        "example": "Patient ID: 789012, Patient Name: 王兴鹏, Diagnosis Code: M54.5, Procedure Code: 99213, Total Charge: $150, Insurance Claim Amount: $120"
    },
    {
        "example": "Patient ID: 345678, Patient Name: 刘晓辉, Diagnosis Code: E11.9, Procedure Code: 99214, Total Charge: $300, Insurance Claim Amount: $250"
    },
]

# Prompt 组合模板
example_prompt = PromptTemplate(
    input_variables=['example'],
    template="{example}"
)

prompt_template = FewShotPromptTemplate(
    prefix=SYNTHETIC_FEW_SHOT_PREFIX,
    suffix=SYNTHETIC_FEW_SHOT_SUFFIX,
    examples=examples,
    example_prompt=example_prompt,
    input_variables=['subject', 'extra']
)

# 然后使用 with_structured_output   明确指定使用 function_calling 方法
chain = prompt_template | model.with_structured_output(MedicalBillingList, method="function_calling")

for i in range(2):
    result: MedicalBillingList = chain.invoke(
        {'extra': "仿照例子来生产模拟数据列表, 列表长度为10", 'subject': "医院账单表"})
    print(result)

核对输出