Incribo

In this article, we explore Incribo, a specialized package for managing AI embeddings. Building upon established principles from state management libraries like Redux, Incribo offers a structured approach to handling embedding data through components such as Branch Manager and Consistency Manager. The article guides readers through the package's installation process, demonstrates how to create and work with embedding objects, and highlights key features like version tracking and branch management. Through detailed explanations and practical examples, we demonstrate how Incribo transforms raw embedding data into a manageable, version-controlled resource for AI development projects.

Package Components

The Incribo package provides several key components for managing AI embeddings. These include Branch Manager, Consistency Manager, and Dynamic Embeddings, which work together to maintain and manipulate embedding data.

At its core, Incribo functions similarly to Redux, which is a state management library originally developed for React applications. By adapting this pattern for embeddings, Incribo helps developers manage the state of their embedding data efficiently.

To utilize the package, users must set up a virtual environment and install Incribo using pipx install incribo. Once installed, developers can create stateful embeddings by importing the Incribo package and creating an Embedding object initialized with existing embeddings. This structure allows for clear separation between different aspects of embedding management, making the data more organized and easier to work with.

The package includes several useful methods for working with Embedding objects. These allow users to retrieve the vector and model name associated with each embedding, as well as print the complete Embedding object for debugging or logging purposes. Additionally, Incribo includes functionality for tracking changes over time through its History Tracker and managing version control for different states of embedding data.

Installation and Setup

Incribo requires users to set up a virtual environment before installation. This can be achieved by creating a new virtual environment using the standard Python tools or by activating an existing one. Once the virtual environment is ready, users can install Incribo using pipx, which manages isolated installations of Python tools.

The installation command is pipx install incribo, which installs the package and ensures that all dependencies are properly configured. After installation, users can verify the setup by importing the Incribo package and creating an Embedding object with existing embeddings. This initial configuration establishes the foundation for managing embeddings using Incribo's stateful approach.

Creating Stateful Embeddings

To create meaningful embeddings using Incribo, you first need to set up a virtual environment and install the package using pipx install incribo. Once installed, you can begin working with embeddings by importing the package and creating an Embedding object initialized with existing data.

When working with Embedding objects, several key methods are available for managing and retrieving embedding data. The package provides access to the vector representation of the embedding through the embedding.vector property, which returns the numerical array representing the embedding. Additionally, you can retrieve the model name used to generate the embedding via the embedding.model_name property.

For detailed debugging or logging purposes, Incribo includes a built-in print method that outputs the entire Embedding object to the console. This functionality makes it easier to inspect and verify the contents of embedding objects directly within your development environment.

Package Features

The Incribo package extends its functionality through several specialized components designed for embedding management. These include a History Tracker for version control and a Branch Manager for organizing different states of embedding data.

The History Tracker maintains a record of changes across different versions of embedding data, allowing users to track modifications and revert to previous states if necessary. This functionality is particularly useful for managing complex embedding workflows where multiple iterations of the same data may be required.

The Branch Manager, inspired by Git's branching model, enables users to work with multiple versions of embedding data simultaneously. This feature is especially valuable in collaborative environments or when different parts of a project require separate embedding versions.

Additional features include a Consistency Manager to ensure that embedding data maintains its integrity across different operations and a Dynamic Embeddings component that allows for flexible embedding generation and manipulation. Together, these components form a comprehensive toolkit for managing embedding data in AI applications.