`select()`

Select one or more options from a given list or type representing a set of options.

API Reference

select()

Synchronously select one or more options from the given target.

Parameters:

Name	Type	Description	Default
`target`	`Any \| List[Any]`	The options to select from. Can be: - A list of values or types - A Literal[...] type - An Enum subclass - A Union[..., ...] type	required
`context`	`ContextType \| List[ContextType] \| None`	Optional additional context or conversation history. Defaults to None.	`None`
`multi_label`	`bool`	When True, return a list of selected objects; otherwise return a single selected object. Defaults to False.	`False`
`literal`	`bool`	When True and supported by the options, use Literal[...] for the selection field instead of indices. Defaults to True.	`True`
`include_reason`	`bool`	Reserved flag to include a textual reason field in the intermediate structured output (not exposed in the final result). Defaults to False.	`False`
`confidence`	`bool`	When True, enable log-probability based confidence scoring. Defaults to False.	`False`
`model`	`ModelParam`	The model to use for selection. Defaults to "openai:gpt-4o-mini".	`'openai:gpt-4o-mini'`
`model_settings`	`PydanticAIModelSettings \| None`	Model settings to pass to the operation (e.g., temperature). Defaults to None.	`None`
`instructions`	`PydanticAIInstructions \| None`	Additional instructions/hints for the model. Defaults to None.	`None`
`tools`	`ToolType \| List[ToolType] \| None`	List of tools available to the model. Defaults to None.	`None`
`deps`	`Deps \| None`	Optional dependencies (e.g., `pydantic_ai.RunContext`) for this operation. Defaults to None.	`None`
`usage_limits`	`PydanticAIUsageLimits \| None`	Usage limits (token/request) configuration. Defaults to None.	`None`
`stream`	`bool`	When True, return a stream wrapper that exposes the underlying model stream and remaps the final result to the selected object(s). Defaults to False.	`False`

Returns:

Type	Description
`Result[Output] \| _SelectionStreamWrapper`	Result[Output] \| _SelectionStreamWrapper: A `Result` whose `output` is either the selected object or a list of selected objects. When `include_reason` is True, the output is a dict with keys `selection` and `reason`. For streaming, the wrapper preserves the same final semantics.

aselect()

Asynchronously select one or more options from the given target.

Parameters:

Name	Type	Description	Default
`target`	`Any \| List[Any]`	The options to select from. Can be: - A list of values or types - A Literal[...] type - An Enum subclass - A Union[..., ...] type	required
`context`	`ContextType \| List[ContextType] \| None`	Optional additional context or conversation history. Defaults to None.	`None`
`multi_label`	`bool`	When True, return a list of selected objects; otherwise return a single selected object. Defaults to False.	`False`
`literal`	`bool`	When True and supported by the options, use Literal[...] for the selection field instead of indices. Defaults to True.	`True`
`include_reason`	`bool`	Reserved flag to include a textual reason field in the intermediate structured output (not exposed in the final result). Defaults to False.	`False`
`confidence`	`bool`	When True, enable log-probability based confidence scoring. Defaults to False.	`False`
`model`	`ModelParam`	The model to use for selection. Defaults to "openai:gpt-4o-mini".	`'openai:gpt-4o-mini'`
`model_settings`	`PydanticAIModelSettings \| None`	Model settings to pass to the operation (e.g., temperature). Defaults to None.	`None`
`instructions`	`PydanticAIInstructions \| None`	Additional instructions/hints for the model. Defaults to None.	`None`
`tools`	`ToolType \| List[ToolType] \| None`	List of tools available to the model. Defaults to None.	`None`
`deps`	`Deps \| None`	Optional dependencies (e.g., `pydantic_ai.RunContext`) for this operation. Defaults to None.	`None`
`usage_limits`	`PydanticAIUsageLimits \| None`	Usage limits (token/request) configuration. Defaults to None.	`None`
`stream`	`bool`	When True, return a stream wrapper that exposes the underlying model stream and remaps the final result to the selected object(s). Defaults to False.	`False`

Returns:

Type	Description
`Result[Output] \| _SelectionStreamWrapper`	Result[Output] \| _SelectionStreamWrapper: A `Result` whose `output` is either the selected object or a list of selected objects. When `include_reason` is True, the output is a dict with keys `selection` and `reason`. For streaming, the wrapper preserves the same final semantics.

Overview

The select() semantic operation is used to select one more options from a given list or type representing a set of options that a model can choose from.

SyncAsync

Single Classification

from zyx import select


result = select(
    target=["red", "green", "blue"], # (1)!
    context="What is the color of the sky?",
)


print(result.output)
"""
"blue"
"""

The target parameter when using select, can be a list of any compatible types (str, int, BaseModel, etc.), or a Literal[...], Enum, or Union[..., ...] type.

Single Classification

from zyx import aselect


async def main():
    result = await aselect(
        target=["red", "green", "blue"],
        context="What is the color of the sky?",
    )

Multi-Label Selection

When running this operation, you can choose between setting the multi_label parameter to True or False, which allows for single or multiple selection respectively.

Multi-Label Selection

from zyx import select


result = select(
    target=["red", "black", "blue", "yellow", "purple"],
    context="What colors make green?",
    multi_label=True,
)


print(result.output)
"""
["blue", "yellow"]
"""

Reasoning

Optionally, you can set the include_reason parameter to True, to allow the model to provide a textual reason for it's selection.

Including Reasoning

from zyx import select


result = select(
    ["positive", "negative"],
    context="This produce is okay.",
    include_reason=True,
)


print(result.output.selection)
print(result.output.reason)
"""
positive
The statement indicates that the produce is acceptable or satisfactory.
"""