Canonical Boosts Machine Learning Development Support with Charmed MLFlow

Highlights:

• One of Charmed MLFlow’s core functions is to monitor experiments, keeping a record of parameters and results for comparison and analysis.
• Canonical says that when Charmed MLFlow and Charmed Kubeflow are used together, it can use extra features like hyper-parameter tuning, graphics processing unit scheduling, and model serving.

Recently, Canonical Ltd. ventured deeper into machine learning operations by officially releasing its Charmed MLFlow platform.

Charmed MLFlow, from Canonical, is their version of the well-known open-source MLFlow platform, designed to handle the complete machine learning model lifecycle. With seamless integration into Canonical’s software ecosystem, streamlined deployment, and ongoing security updates, Charmed MLFlow offers enhanced benefits.

Charmed MLFlow, according to the company, provides four primary functions in the development of machine learning, a subset of artificial intelligence that focuses on using data and algorithms to imitate approximately how humans learn, thereby progressively improving the accuracy of AI models.

The primary function of Charmed MLFlow is to monitor experiments, capture and compare parameters and outcomes. It also facilitates the packaging of machine learning code in a reusable, reproducible format, allowing it to be shared with other data scientists or transmitted to production.

Moreover, it serves as a tool to manage and deploy models sourced from diverse machine-learning libraries effectively. Lastly, it functions as a centralized model store, enabling teams to oversee the complete lifecycle of MLFlow models collaboratively. This includes critical steps like model versioning, stage transitions, and annotations.

Canonical said that the fact that Charmed MLFlow is easy to set up is one of its best features. Users can quickly set it up on a modest laptop within a few minutes, allowing for swift and efficient experimentation. While thoroughly tested on the Ubuntu operating system, it’s versatile enough for use on various platforms, including the Windows Subsystem for Linux.

Canonical said it is also very flexible because it can run in any environment, whether a public or private cloud, and works in multicloud scenarios. Also, it works with any Kubernetes distribution that meets the standards of the Cloud Native Computing Foundation, such as Charmed Kubernetes, MicroK8s, or Amazon EKS. When users need more computing power, they can move their models from the laptops where they were made to any cloud infrastructure.

Canonical said it has done much work to ensure that tools like Jupyter Notebook, Charmed Kubeflow, and KServe work well with Charmed MLFlow. Another benefit is that it works with the Canonical Observability Stack, which lets you monitor infrastructure.

Canonical says that when Charmed MLFlow and Charmed Kubeflow are used together, it can use extra features like hyper-parameter tuning, graphics processing unit scheduling, and model serving.

Charmed MLFlow is fully backed by Canonical, which says it can help with deployment, uptime monitoring, operations, and bug fixes.

Charmed MLFlow is the latest MLOps tool that Canonical is adding to its growing collection. It is available as part of the Canonical Ubuntu Pro subscription and is priced per node.

The Vice President of product management at Canonical, Cedric Gegout, said that the open-source version of MLFlow is one of the most popular AI frameworks for all stages of machine learning development. “Its popularity arises from its flexibility in facilitating modest local desktop experimentation and extensive cloud deployment, catering to both individual and enterprise needs,” he said. “This makes Charmed MLFlow a fitting addition to our Canonical MLOps suite, offering cost-effective solutions that enable developers to start small and scale up as their business grows.”