Announcing our improved experiment tracking tools in Azure Machine Learning

We are thrilled to announce the public preview of our enhanced experiment tracking tools in Azure Machine Learning, specifically designed to provide you with a more comprehensive understanding, management, and visualization of your training jobs. These improvements comprise a dynamic and customizable dashboard view, an upgraded jobs list view, and a wide range of features aimed at simplifying the organization and analysis of your experimentation results.

Key Features:

• New dynamic and flexible dashboard view
• Improved jobs list view with customizable columns and filters
• Easily compare metrics and images across training jobs
• Add markdown for notes and insights
• Create and save custom views for easy collaboration

Explore the New Dashboard View:

Our new dashboard view combines various tiles – including chart visualizations and markdown – to provide a comprehensive and customizable overview of your experimentation results. By default, the dashboard generates visualizations of your resource utilization, evaluation, and logged metrics for the selected jobs.

Customize Your Dashboard:

• Change job colors with the color picker
• Visualize and hide jobs as needed
• Add and edit charts for custom visualizations
• Compare metrics and images across training jobs
• Add markdown for notes and insights
• Create and save custom views for collaboration

Improved Jobs List View:

The jobs list view now allows you to easily track and organize information with features such as:

• Customizable job display names
• Select, pin, and reorder columns
• Filter jobs based on various criteria
• Perform batch actions on jobs

Get Started:

To try out these new features, simply enable the preview via the preview panel and navigate to the Jobs tab in the studio.

To learn more, read our documentation to:

• Visualize training results in studio (preview) - Azure Machine Learning | Microsoft Learn
• Organize & track training jobs (preview) - Azure Machine Learning | Microsoft Learn

Watch Microsoft Build 2023 sessions to get familiar with other Azure Machine Learning announcements:

• Breakout: Build and maintain your company Copilot with Azure Machine Learning and GPT-4  
• Breakout: Practical deep-dive into machine learning techniques and MLOps  
• Breakout: Building and using AI models responsibly