Track ML model experiments with new GitLab MLFlow integration

What this post added

This post introduces the Machine Learning Model Experiments feature in GitLab, which integrates with MLFlow to allow users to track ML model versions and their metadata directly within GitLab. This feature acts as an MLFlow backend, simplifying the process for data scientists by requiring minimal changes to their existing MLFlow client code. Key functionalities include logging experiments and runs (called 'candidates' in GitLab) without requiring users to set up or manage their own MLFlow servers. Access control is managed through GitLab projects, and users can manage and explore logged experiments directly in the GitLab UI. Data scientists can also download experiment data as CSV files. This feature is available to all GitLab users and is part of GitLab's broader effort to manage the entire ML model lifecycle.