Pydantic

Data validation and structured output from LLMs using Pydantic models

Overview

Helps get structured output from LLMs.

Example

from pydantic import BaseModel, EmailStr
from typing import Literal, List

class CustomerQuery(BaseModel):
    name: str
    email: EmailStr
    query: str
    category: Literal['refund_request', 'information_request', 'other']
    is_complaint: bool
    tags: List[str]

Validation

valid_data = CustomerQuery.model_validate_json(json_output)

Common Errors

1. JSON ValidationError: Extra text or malformed JSON
2. Data ValidationError: Valid JSON but data doesn’t match model schema