Skip to content

Latest commit

 

History

History
44 lines (34 loc) · 1.63 KB

pipeline.md

File metadata and controls

44 lines (34 loc) · 1.63 KB
Raw

Pipeline

Overview

The Pipeline framework in MiniAutoGen is a flexible and modular system designed for data processing, particularly in chat applications. It allows for the sequential processing of data through a series of components, each handling specific tasks or transformations.

Core

class Pipeline

Description:

Represents a data processing pipeline, which is a sequence of PipelineComponent objects.

Attributes:

  • components (list of PipelineComponent): The list of components in the pipeline.

Constructor:

__init__(self, components=None)
  • Purpose: Initializes the pipeline with a list of components.
  • Parameters:
    • components (list of PipelineComponent, optional): The components to be included in the pipeline. Defaults to an empty list if not provided.

Methods:

add_component(self, component)
  • Purpose: Adds a new component to the pipeline.
  • Parameters:
    • component (PipelineComponent): The component to be added.
  • Exceptions: Raises TypeError if the component is not a subclass of PipelineComponent.
run(self, state)
  • Purpose: Executes the pipeline on the given state, passing the state through each component in sequence.
  • Parameters:
    • state (ChatPipelineState): The current state of the chat to be processed.
  • Returns: The modified ChatPipelineState after processing through all components.

Example Usage

Creating a Pipeline

my_pipeline = Pipeline(components=[component1, component2])

Adding a Component to the Pipeline

my_pipeline.add_component(new_component)