Task

Subpackages

Submodules

llmsmith.task.base module

class llmsmith.task.base.Task(name: str)

Bases: Generic[T, U], ABC

Base class for all tasks. All subclasses of Task should implement the execute method. In case you are using some LLM (or DB for retrieval) which is still not supported by this library, you can write your own implementation easily by subclassing Task.

Parameters:: name (str) – The name of the task. The task name should be non-empty.
Raises:: ValueError – If task name fails validation.

abstract async execute(task_input: TaskInput[T]) → TaskOutput[U]

Abstract method for executing the task.

Parameters:: task_input (TaskInput[T]) – The input to the task.
Returns:: The output of the task.
Return type:: TaskOutput[U]

name() → str

Returns the name of the task.

Returns:: task name.
Return type:: str

llmsmith.task.models module

class llmsmith.task.models.ChatResponse(text: str, raw_output: Any, function_calls: dict[str, llmsmith.task.models.FunctionCall] = None)

Bases: object

function_calls: dict[str, FunctionCall] = None

raw_output: Any

text: str

class llmsmith.task.models.FunctionCall(id: str, name: str, args: dict[str, Any])

Bases: object

args: dict[str, Any]

id: str

name: str

class llmsmith.task.models.TaskInput(content: T)

Bases: Generic[T]

content: T

class llmsmith.task.models.TaskOutput(content: T, raw_output: Any)