`make()`

Generate text and structured outputs based on a given target, with the ability to create synthetic content.

API Reference

make()

Generate ('make') a value of the provided target type using a model or Pydantic AI agent.

Parameters:

Name	Type	Description	Default
`target`	`TargetParam[Output]`	The target type or value to generate. Defaults to str.	`str`
`context`	`ContextType \| List[ContextType] \| None`	The context to use for the operation. Defaults to None.	`None`
`randomize`	`bool`	Injects a simple randomization instruction for more diverse outputs. This is automatically added if no context or instructions are provided. Defaults to False.	`False`
`confidence`	`bool`	Whether to include confidence scores in the result of the operation. This is currently only supported for OpenAI or OpenAI-like models. Defaults to False.	`False`
`model`	`ModelParam`	The model to use for the operation. This can be a string, Pydantic AI model, or Pydantic AI agent. Defaults to "openai:gpt-4o-mini".	`'openai:gpt-4o-mini'`
`model_settings`	`PydanticAIModelSettings \| None`	The model settings to use for the operation. Defaults to None.	`None`
`attachments`	`AttachmentType \| List[AttachmentType] \| None`	A single or list of attachment objects provided to the agent. An attachment is a piece of content that is provided to the agent in a 'persistent' fashion, where it is templated/placed specifically to avoid context rot or loss. Furthermore, attachments that are `Resources` provide the agent with an ability to interact with/modify them, like artifacts. Defaults to None.	`None`
`instructions`	`PydanticAIInstructions \| None`	The instructions to use for the operation. Defaults to None.	`None`
`tools`	`ToolType \| List[ToolType] \| None`	The tools to use for the operation. Defaults to None.	`None`
`deps`	`Deps \| None`	Reference to `deps` in `pydantic_ai.RunContext`, that can be passed to messages, tools and instructions. Defaults to None.	`None`
`usage_limits`	`PydanticAIUsageLimits \| None`	The usage limits to use for the operation. Defaults to None.	`None`
`observe`	`bool \| Observer \| None`	If True or provided, enables CLI observation output.	`None`
`stream`	`bool`	Whether to stream the output of the operation. Defaults to False.	`False`

Returns:

Type	Description
`Result[Output] \| Stream[Output]`	Result[Output] \| Stream[Output]: The result or stream of the operation.

amake()

Asynchronously generate ('make') a value of the provided target type using a model or Pydantic AI agent.

Parameters:

Name	Type	Description	Default
`target`	`TargetParam[Output]`	The target type or value to generate. Defaults to str.	`str`
`context`	`ContextType \| List[ContextType] \| None`	The context to use for the operation. Defaults to None.	`None`
`randomize`	`bool`	Injects a simple randomization instruction for more diverse outputs. This is automatically added if no context or instructions are provided. Defaults to False.	`False`
`confidence`	`bool`	Whether to include confidence scores in the result of the operation. This is currently only supported for OpenAI or OpenAI-like models. Defaults to False.	`False`
`model`	`ModelParam`	The model to use for the operation. This can be a string, Pydantic AI model, or Pydantic AI agent. Defaults to "openai:gpt-4o-mini".	`'openai:gpt-4o-mini'`
`model_settings`	`PydanticAIModelSettings \| None`	The model settings to use for the operation. Defaults to None.	`None`
`attachments`	`AttachmentType \| List[AttachmentType] \| None`	A single or list of attachment objects provided to the agent. An attachment is a piece of content that is provided to the agent in a 'persistent' fashion, where it is templated/placed specifically to avoid context rot or loss. Furthermore, attachments that are `Resources` provide the agent with an ability to interact with/modify them, like artifacts. Defaults to None.	`None`
`instructions`	`PydanticAIInstructions \| None`	The instructions to use for the operation. Defaults to None.	`None`
`tools`	`ToolType \| List[ToolType] \| None`	The tools to use for the operation. Defaults to None.	`None`
`deps`	`Deps \| None`	Reference to `deps` in `pydantic_ai.RunContext`, that can be passed to messages, tools and instructions. Defaults to None.	`None`
`usage_limits`	`PydanticAIUsageLimits \| None`	The usage limits to use for the operation. Defaults to None.	`None`
`observe`	`bool \| Observer \| None`	If True or provided, enables CLI observation output.	`None`
`stream`	`bool`	Whether to stream the output of the operation. Defaults to False.	`False`

Returns:

Type	Description
`Result[Output] \| Stream[Output]`	Result[Output] \| Stream[Output]: The result or stream of the operation.

Overview

The make() semantic operation is used to generate content and structured outputs based on a given target type or value.

SyncAsync

Generating a Structured Output

from zyx import make
from pydantic import BaseModel


class User(BaseModel):
    name : str
    age : int


result = make(
    target=User,
    context="John is 25 years old."
)


print(result.output)
"""
User(name="John", age=25)
"""

Generating a Structured Output

from zyx import amake
from pydantic import BaseModel


class User(BaseModel):
    name : str
    age : int


async def main():
    result = await amake(
        target=User,
        context="John is 25 years old."
    )

Did you know?

make() can be called with or without context allowing for model determined synthetic content generation.

result = make(User) # (1)!

print(result.output)
"""
name='Leona Frost' age=28
"""

In this case, only the target was provided.

Streaming

make() also supports streaming the output of the operation as it is generated.

Streaming the Output

from zyx import make


stream = make(
    target=list[str],
    context="Write a haiku about the weather.",
    stream=True # (1)!
)


for chunk in stream.partial(): # (2)!
    print(chunk)
"""
['Whispers of the breeze,', 'Clouds']
['Whispers of the breeze,', 'Clouds dance in the silver sky,']
['Whispers of the breeze,', 'Clouds dance in the silver sky,', 'Raindrops kiss the earth.']
['Whispers of the breeze,', 'Clouds dance in the silver sky,', 'Raindrops kiss the earth.']
['Whispers of the breeze,', 'Clouds dance in the silver sky,', 'Raindrops kiss the earth.']
"""

We set the stream parameter to True to enable streaming.
A stream can be iterated over using the text() or partial() methods. In this case, we are using a non-string structured output, so we use partial().