Table of Content

Solution Overview

The Orca solution contains three distinct modules:

• OrcaDB, a server for managing relational data and tracking memory usage during model inference.
• OrcaLib, a client-side framework that makes it easy to:
  • Create, update, and edit datasets for use with memory-augmented models
  • Add memory-augmentation to new or existing models
• OrcaCurate, a UI toolchain for real-time viewing and editing the memories (i.e., data) your models are using during inference.

Getting started with OrcaLib

First steps

Install OrcaLib

OrcaLib is available on PyPI at **https://pypi.org/project/orcalib/**. To install OrcaLib, run the following command:

pip install orcalib

Import OrcaLib

After installing OrcaLib in your Python environment, you’ll need to import it in your scripts. We recommend importing with the alias orca. Inside OrcaLib, you’ll find global settings, tools for managing data, and the classes you need to memory augment your pytorch models — or build new ones.

import orcalib as orca