Response Model¶

Defining LLM output schemas in Pydantic is done via pydantic.BaseModel. To learn more about models in Pydantic, check out their documentation.

After defining a Pydantic model, we can use it as the response_model in your client create calls to OpenAI or any other supported model. The job of the response_model parameter is to:

Define the schema and prompts for the language model
Validate the response from the API
Return a Pydantic model instance.

Prompting¶

When defining a response model, we can use docstrings and field annotations to define the prompt that will be used to generate the response.

from pydantic import BaseModel, Field


class User(BaseModel):
    """
    This is the prompt that will be used to generate the response.
    Any instructions here will be passed to the language model.
    """

    name: str = Field(description="The name of the user.")
    age: int = Field(description="The age of the user.")

Here all docstrings, types, and field annotations will be used to generate the prompt. The prompt will be generated by the create method of the client and will be used to generate the response.

Optional Values¶

If we use Optional and default, they will be considered not required when sent to the language model

from pydantic import BaseModel, Field
from typing import Optional


class User(BaseModel):
    name: str = Field(description="The name of the user.")
    age: int = Field(description="The age of the user.")
    email: Optional[str] = Field(description="The email of the user.", default=None)

Dynamic model creation¶

There are some occasions where it is desirable to create a model using runtime information to specify the fields. For this, Pydantic provides the create_model function to allow models to be created on the fly:

from pydantic import BaseModel, create_model


class FooModel(BaseModel):
    foo: str
    bar: int = 123


BarModel = create_model(
    'BarModel',
    apple=(str, 'russet'),
    banana=(str, 'yellow'),
    __base__=FooModel,
)
print(BarModel)
#> <class '__main__.BarModel'>
print(BarModel.model_fields.keys())
#> dict_keys(['foo', 'bar', 'apple', 'banana'])

When would I use this?

Consider a situation where the model is dynamically defined, based on some configuration or database. For example, we could have a database table that stores the properties of a model for some model name or id. We could then query the database for the properties of the model and use that to create the model.

SELECT property_name, property_type, description
FROM prompt
WHERE model_name = {model_name}

We can then use this information to create the model.

from pydantic import BaseModel, create_model
from typing import List

types = {
    'string': str,
    'integer': int,
    'boolean': bool,
    'number': float,
    'List[str]': List[str],
}

# Mocked cursor.fetchall()
cursor = [
    ('name', 'string', 'The name of the user.'),
    ('age', 'integer', 'The age of the user.'),
    ('email', 'string', 'The email of the user.'),
]

BarModel = create_model(
    'User',
    **{
        property_name: (types[property_type], description)
        for property_name, property_type, description in cursor
    },
    __base__=BaseModel,
)

print(BarModel.model_json_schema())
"""
{
    'properties': {
        'name': {'default': 'The name of the user.', 'title': 'Name', 'type': 'string'},
        'age': {'default': 'The age of the user.', 'title': 'Age', 'type': 'integer'},
        'email': {
            'default': 'The email of the user.',
            'title': 'Email',
            'type': 'string',
        },
    },
    'title': 'User',
    'type': 'object',
}
"""

This would be useful when different users have different descriptions for the same model. We can use the same model but have different prompts for each user.

Adding Behavior¶

We can add methods to our Pydantic models, just as any plain Python class. We might want to do this to add some custom logic to our models.

from pydantic import BaseModel
from typing import Literal

from openai import OpenAI

import instructor

client = instructor.patch(OpenAI())


class SearchQuery(BaseModel):
    query: str
    query_type: Literal["web", "image", "video"]

    def execute(self):
        print(f"Searching for {self.query} of type {self.query_type}")
        #> Searching for cat of type image
        return "Results for cat"


query = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[{"role": "user", "content": "Search for a picture of a cat"}],
    response_model=SearchQuery,
)

results = query.execute()
print(results)
#> Results for cat

Now we can call execute on our model instance after extracting it from a language model. If you want to see more examples of this checkout our post on RAG is more than embeddings