Model management

As explained in Delivering Machine Learning Projects: From design to deployment

Model management is the discipline of organizing, tracking, evaluating, and governing machine learning models throughout their development and deployment lifecycles.

Model management organizes machine learning models and their supporting artifacts so development remains iterative, controlled, accountable, and deployable. It connects experimentation with evaluation, qualification, deployment, and later investigation. section 5.5

A workable model-management system should:

• Commission a model repository and record every artifact used in machine learning pipelines. section 5.5
• Assign each model an identity linked to its binary or declarative specification for testing and production. section 5.5
• Track data, features, hyperparameters, algorithms, architectures, evaluation metrics, test results, and pipeline identities for every iteration. section 5.5
• Preserve model evolution and decision evidence to support investigations of stability, failures, blind spots, and governance. section 5.5

Figure 1.2 Drivers of complexity in an ML project: the domain, the data, and the models

Use the domain, data, and models as a complexity frame when deciding what model-management records and controls are necessary. section 1.2

How Delivering Machine Learning Projects: From design to deployment covers it

How deep it goes

Specifies repository contents and accountability rationale, but stops short of prescribing a particular repository technology.

Examples

Uses ticket S1.4 and records model identities, parameters, artifacts, pipeline status, evaluation results, and qualification tests.

What it assumes

Assumes familiarity with machine learning models and pipelines while emphasizing controlled delivery and accountability.

What you can do after

Operate a repository that compares model iterations and supports evidence-based selection, deployment, monitoring, and failure investigation.

How it connects

Connects model versioning with data pipelines, evaluation metrics, testing, monitoring, deployment, and governance.

liveBookmanning.com

Read Delivering Machine Learning Projects: From design to deployment on liveBook Get the book at manning.com

Related concepts

• MLflowused by · MLflow is an open-source platform for managing the machine-learning lifecycle, including experiment tracking, model and artifact management, and model deployment.
• authorizationhas part · Authorization is the determination and enforcement of rules governing which subjects may access resources or perform operations.
• model registryimplemented by · A model registry is a system for storing, organizing, versioning, and managing machine-learning models and their associated metadata throughout their lifecycle.
• language modeluses · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
• ollama listhas part · ollama list is the Ollama command that lists language models downloaded to the local model store.
• productionuses · A production environment is the live operational setting in which software or other systems serve real users, workloads, or business processes.

This explainer was generated by Manning's knowledge graph from the text of Delivering Machine Learning Projects: From design to deployment. The section links point to the parts of the book it draws on.