Instructor

Структурированный вывод LLM с валидацией Pydantic.

Метаданные навыка

Справочник: полный SKILL.md

Instructor: структурированный вывод LLM

Когда использовать этот навык

Используйте Instructor, когда нужно:

• Надёжно извлекать структурированные данные из ответов LLM
• Валидировать вывод по схемам Pydantic автоматически
• Повторять неудачные попытки с автоматической обработкой ошибок
• Разбирать сложный JSON с контролем типов и валидацией
• Стримить частичные результаты для обработки в реальном времени
• Работать с несколькими провайдерами LLM через единый API

GitHub Stars: 15 000+ | Проверено в деле: 100 000+ разработчиков

Установка

# Base installation
pip install instructor

# With specific providers
pip install "instructor[anthropic]"  # Anthropic Claude
pip install "instructor[openai]"     # OpenAI
pip install "instructor[all]"        # All providers

Быстрый старт

Базовый пример: извлечение данных о пользователе

import instructor
from pydantic import BaseModel
from anthropic import Anthropic

# Define output structure
class User(BaseModel):
    name: str
    age: int
    email: str

# Create instructor client
client = instructor.from_anthropic(Anthropic())

# Extract structured data
user = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "John Doe is 30 years old. His email is john@example.com"
    }],
    response_model=User
)

print(user.name)   # "John Doe"
print(user.age)    # 30
print(user.email)  # "john@example.com"

С OpenAI

from openai import OpenAI

client = instructor.from_openai(OpenAI())

user = client.chat.completions.create(
    model="gpt-4o-mini",
    response_model=User,
    messages=[{"role": "user", "content": "Extract: Alice, 25, alice@email.com"}]
)

Основные понятия

1. Модели ответа (Pydantic)

Модели ответа задают структуру и правила валидации для вывода LLM.

Базовая модель

from pydantic import BaseModel, Field

class Article(BaseModel):
    title: str = Field(description="Article title")
    author: str = Field(description="Author name")
    word_count: int = Field(description="Number of words", gt=0)
    tags: list[str] = Field(description="List of relevant tags")

article = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Analyze this article: [article text]"
    }],
    response_model=Article
)

Что это даёт:

• Контроль типов через подсказки типов Python
• Автоматическую валидацию (word_count > 0)
• Самодокументирование за счёт описаний в Field
• Автодополнение в IDE

Вложенные модели

class Address(BaseModel):
    street: str
    city: str
    country: str

class Person(BaseModel):
    name: str
    age: int
    address: Address  # Nested model

person = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "John lives at 123 Main St, Boston, USA"
    }],
    response_model=Person
)

print(person.address.city)  # "Boston"

Необязательные поля

from typing import Optional

class Product(BaseModel):
    name: str
    price: float
    discount: Optional[float] = None  # Optional
    description: str = Field(default="No description")  # Default value

# LLM doesn't need to provide discount or description

Перечисления для ограничений

from enum import Enum

class Sentiment(str, Enum):
    POSITIVE = "positive"
    NEGATIVE = "negative"
    NEUTRAL = "neutral"

class Review(BaseModel):
    text: str
    sentiment: Sentiment  # Only these 3 values allowed

review = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "This product is amazing!"
    }],
    response_model=Review
)

print(review.sentiment)  # Sentiment.POSITIVE

2. Валидация

Pydantic проверяет вывод LLM автоматически. Если проверка не прошла, Instructor повторяет запрос.

Встроенные валидаторы

from pydantic import Field, EmailStr, HttpUrl

class Contact(BaseModel):
    name: str = Field(min_length=2, max_length=100)
    age: int = Field(ge=0, le=120)  # 0 <= age <= 120
    email: EmailStr  # Validates email format
    website: HttpUrl  # Validates URL format

# If LLM provides invalid data, Instructor retries automatically

Свои валидаторы

from pydantic import field_validator

class Event(BaseModel):
    name: str
    date: str
    attendees: int

    @field_validator('date')
    def validate_date(cls, v):
        """Ensure date is in YYYY-MM-DD format."""
        import re
        if not re.match(r'\d{4}-\d{2}-\d{2}', v):
            raise ValueError('Date must be YYYY-MM-DD format')
        return v

    @field_validator('attendees')
    def validate_attendees(cls, v):
        """Ensure positive attendees."""
        if v < 1:
            raise ValueError('Must have at least 1 attendee')
        return v

Валидация на уровне модели

from pydantic import model_validator

class DateRange(BaseModel):
    start_date: str
    end_date: str

    @model_validator(mode='after')
    def check_dates(self):
        """Ensure end_date is after start_date."""
        from datetime import datetime
        start = datetime.strptime(self.start_date, '%Y-%m-%d')
        end = datetime.strptime(self.end_date, '%Y-%m-%d')

        if end < start:
            raise ValueError('end_date must be after start_date')
        return self

3. Автоматический повтор

Instructor повторяет запрос автоматически, когда валидация не прошла, и передаёт LLM обратную связь об ошибке.

# Retries up to 3 times if validation fails
user = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Extract user from: John, age unknown"
    }],
    response_model=User,
    max_retries=3  # Default is 3
)

# If age can't be extracted, Instructor tells the LLM:
# "Validation error: age - field required"
# LLM tries again with better extraction

Как это работает:

1. LLM формирует вывод
2. Pydantic проверяет его
3. Если невалидно: сообщение об ошибке отправляется обратно в LLM
4. LLM пробует снова, опираясь на обратную связь об ошибке
5. Повтор до max_retries

4. Стриминг

Стримьте частичные результаты для обработки в реальном времени.

Стриминг частичных объектов

from instructor import Partial

class Story(BaseModel):
    title: str
    content: str
    tags: list[str]

# Stream partial updates as LLM generates
for partial_story in client.messages.create_partial(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Write a short sci-fi story"
    }],
    response_model=Story
):
    print(f"Title: {partial_story.title}")
    print(f"Content so far: {partial_story.content[:100]}...")
    # Update UI in real-time

Стриминг итерируемых объектов

class Task(BaseModel):
    title: str
    priority: str

# Stream list items as they're generated
tasks = client.messages.create_iterable(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Generate 10 project tasks"
    }],
    response_model=Task
)

for task in tasks:
    print(f"- {task.title} ({task.priority})")
    # Process each task as it arrives

Настройка провайдера

Anthropic Claude

import instructor
from anthropic import Anthropic

client = instructor.from_anthropic(
    Anthropic(api_key="your-api-key")
)

# Use with Claude models
response = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[...],
    response_model=YourModel
)

OpenAI

from openai import OpenAI

client = instructor.from_openai(
    OpenAI(api_key="your-api-key")
)

response = client.chat.completions.create(
    model="gpt-4o-mini",
    response_model=YourModel,
    messages=[...]
)

Локальные модели (Ollama)

from openai import OpenAI

# Point to local Ollama server
client = instructor.from_openai(
    OpenAI(
        base_url="http://localhost:11434/v1",
        api_key="ollama"  # Required but ignored
    ),
    mode=instructor.Mode.JSON
)

response = client.chat.completions.create(
    model="llama3.1",
    response_model=YourModel,
    messages=[...]
)

Типовые паттерны

Паттерн 1: извлечение данных из текста

class CompanyInfo(BaseModel):
    name: str
    founded_year: int
    industry: str
    employees: int
    headquarters: str

text = """
Tesla, Inc. was founded in 2003. It operates in the automotive and energy
industry with approximately 140,000 employees. The company is headquartered
in Austin, Texas.
"""

company = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": f"Extract company information from: {text}"
    }],
    response_model=CompanyInfo
)

Паттерн 2: классификация

class Category(str, Enum):
    TECHNOLOGY = "technology"
    FINANCE = "finance"
    HEALTHCARE = "healthcare"
    EDUCATION = "education"
    OTHER = "other"

class ArticleClassification(BaseModel):
    category: Category
    confidence: float = Field(ge=0.0, le=1.0)
    keywords: list[str]

classification = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "Classify this article: [article text]"
    }],
    response_model=ArticleClassification
)

Паттерн 3: извлечение нескольких сущностей

class Person(BaseModel):
    name: str
    role: str

class Organization(BaseModel):
    name: str
    industry: str

class Entities(BaseModel):
    people: list[Person]
    organizations: list[Organization]
    locations: list[str]

text = "Tim Cook, CEO of Apple, announced at the event in Cupertino..."

entities = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": f"Extract all entities from: {text}"
    }],
    response_model=Entities
)

for person in entities.people:
    print(f"{person.name} - {person.role}")

Паттерн 4: структурированный анализ

class SentimentAnalysis(BaseModel):
    overall_sentiment: Sentiment
    positive_aspects: list[str]
    negative_aspects: list[str]
    suggestions: list[str]
    score: float = Field(ge=-1.0, le=1.0)

review = "The product works well but setup was confusing..."

analysis = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": f"Analyze this review: {review}"
    }],
    response_model=SentimentAnalysis
)

Паттерн 5: пакетная обработка

def extract_person(text: str) -> Person:
    return client.messages.create(
        model="claude-sonnet-4-5-20250929",
        max_tokens=1024,
        messages=[{
            "role": "user",
            "content": f"Extract person from: {text}"
        }],
        response_model=Person
    )

texts = [
    "John Doe is a 30-year-old engineer",
    "Jane Smith, 25, works in marketing",
    "Bob Johnson, age 40, software developer"
]

people = [extract_person(text) for text in texts]

Продвинутые возможности

Union-типы

from typing import Union

class TextContent(BaseModel):
    type: str = "text"
    content: str

class ImageContent(BaseModel):
    type: str = "image"
    url: HttpUrl
    caption: str

class Post(BaseModel):
    title: str
    content: Union[TextContent, ImageContent]  # Either type

# LLM chooses appropriate type based on content

Динамические модели

from pydantic import create_model

# Create model at runtime
DynamicUser = create_model(
    'User',
    name=(str, ...),
    age=(int, Field(ge=0)),
    email=(EmailStr, ...)
)

user = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[...],
    response_model=DynamicUser
)

Свои режимы

# For providers without native structured outputs
client = instructor.from_anthropic(
    Anthropic(),
    mode=instructor.Mode.JSON  # JSON mode
)

# Available modes:
# - Mode.ANTHROPIC_TOOLS (recommended for Claude)
# - Mode.JSON (fallback)
# - Mode.TOOLS (OpenAI tools)

Управление контекстом

# Single-use client
with instructor.from_anthropic(Anthropic()) as client:
    result = client.messages.create(
        model="claude-sonnet-4-5-20250929",
        max_tokens=1024,
        messages=[...],
        response_model=YourModel
    )
    # Client closed automatically

Обработка ошибок

Обработка ошибок валидации

from pydantic import ValidationError

try:
    user = client.messages.create(
        model="claude-sonnet-4-5-20250929",
        max_tokens=1024,
        messages=[...],
        response_model=User,
        max_retries=3
    )
except ValidationError as e:
    print(f"Failed after retries: {e}")
    # Handle gracefully

except Exception as e:
    print(f"API error: {e}")

Свои сообщения об ошибках

class ValidatedUser(BaseModel):
    name: str = Field(description="Full name, 2-100 characters")
    age: int = Field(description="Age between 0 and 120", ge=0, le=120)
    email: EmailStr = Field(description="Valid email address")

    class Config:
        # Custom error messages
        json_schema_extra = {
            "examples": [
                {
                    "name": "John Doe",
                    "age": 30,
                    "email": "john@example.com"
                }
            ]
        }

Рекомендации

1. Понятные описания полей

# ❌ Bad: Vague
class Product(BaseModel):
    name: str
    price: float

# ✅ Good: Descriptive
class Product(BaseModel):
    name: str = Field(description="Product name from the text")
    price: float = Field(description="Price in USD, without currency symbol")

2. Подходящая валидация

# ✅ Good: Constrain values
class Rating(BaseModel):
    score: int = Field(ge=1, le=5, description="Rating from 1 to 5 stars")
    review: str = Field(min_length=10, description="Review text, at least 10 chars")

3. Примеры прямо в промпте

messages = [{
    "role": "user",
    "content": """Extract person info from: "John, 30, engineer"

Example format:
{
  "name": "John Doe",
  "age": 30,
  "occupation": "engineer"
}"""
}]

4. Перечисления для фиксированных категорий

# ✅ Good: Enum ensures valid values
class Status(str, Enum):
    PENDING = "pending"
    APPROVED = "approved"
    REJECTED = "rejected"

class Application(BaseModel):
    status: Status  # LLM must choose from enum

5. Аккуратная обработка отсутствующих данных

class PartialData(BaseModel):
    required_field: str
    optional_field: Optional[str] = None
    default_field: str = "default_value"

# LLM only needs to provide required_field

Сравнение с альтернативами

Когда выбрать Instructor:

• Нужен структурированный, проверенный вывод
• Важны контроль типов и поддержка в IDE
• Требуются автоматические повторы
• Строите системы извлечения данных

Когда выбрать альтернативу:

• DSPy: нужна оптимизация промптов
• LangChain: строите сложные цепочки
• Вручную: простые разовые извлечения

Ресурсы

• Документация: https://python.useinstructor.com
• GitHub: https://github.com/jxnl/instructor (15k+ звёзд)
• Сборник примеров: https://python.useinstructor.com/examples
• Discord: доступна поддержка сообщества

Смотрите также

• references/validation.md — продвинутые паттерны валидации
• references/providers.md — настройка под конкретного провайдера
• references/examples.md — реальные сценарии использования