Model

A learnware is a well-performed trained model with a specification, where the model is an indispensable component of the learnware.

In this section, we will introduce the model module implemented within learnware package. We will first introduce the BaseModel, which defines the standard format for models in the learnware package. Following that, we will introduce the ModelContainer, which implements model deployment in conda virtual environments and Docker containers.

BaseModel

The BaseModel class is a fundamental component of the learnware package which provides standardized interface for model training, prediction and fine-tuning. This class is created to make it easier for users to submit learnwares to the market. It helps ensure that submitted models follow a clear set of rules and requirements.

All user models should inherit the BaseModel class. Here’s a more detailed explanation of key components:

• input_shape: Specify the shape of the input features your model expects.

• output_shape: Define the shape of the output predictions generated by your model.

• predict: Implement the predict method to make predictions using your model.

• fit (optional): Use the fit method for training a model with input data and labels.

• finetune (optional): Utilize the finetune method for further adjusting pre-existing models sourced from the market.

By adhering to these standards, the compatibility and quality of submitted learnwares in the market are ensured.

ModelContainer

The ModelContainer class is an essential component of the learnware package, designed to facilitate the management, deployment, and execution of machine learning models within a containerized environment. It inherits from the BaseModel class and extends its functionality to encapsulate model deployment and execution.

ModelCondaContainer

The ModelCondaContainer class is an extension of the ModelContainer class within the learnware package. Its primary purpose is to enable the management, deployment, and execution of machine learning models in a containerized environment, with a specific focus on using Conda virtual environments. This class inherits functionality from ModelContainer while providing additional capabilities related to Conda-based model execution.

Specifically, the ModelCondaContainer supports the automatic creation of new Conda virtual environments based on the requirements.txt file (for pip installation) or environment.yaml file (for Conda installation) included within the learnware itself. It also installs the environment dependencies of the learnware, enabling it to run.

ModelDockerContainer

The ModelDockerContainer class is a specialized extension of the ModelContainer class within the learnware package. It is designed to manage, deploy, and execute machine learning models within a containerized environment, specifically using Docker containers. This class inherits functionality from ModelContainer and enhances it with features related to Docker-based model execution.

Compared to ModelCondaContainer, ModelDockerContainer confines the model’s execution within a Docker container. It installs the learnware’s virtual environment inside the Docker container, isolating the learnware’s execution from the host machine, thus enhancing the security of the learnware.

Similar to the ModelCondaContainer class, the ModelDockerContainer class also supports both types of environment dependency files for learnware: requirements.txt for pip-based installation and environment.yaml for conda-based installation. It automates the creation of Docker containers and the installation of learnware’s environment dependencies within the container, enabling the learnware to run.