You are using an Azure Databricks cluster for training your ML model.
In order to monitor and track the training process of the model, you want to set up MLFlow tracking.
By setting up MLFlow for tracking you can store logs and model artefacts ...
Click on the arrows to vote for the correct answer
A. B. C. D.Correct Answer: C.
Option A is incorrect because MLFlow outputs can also be stored in the Azure Databricks workspace.
Option B is incorrect because MLFlow outputs can also be directed to your Azure ML workspace.
Option C is CORRECT because by using MLFlow, the logs and any artefacts of model runs can be stored in both your Azure ML and Azure Databricks workspaces.
The two workspaces must be linked together.
Option D is incorrect because MLFlow logs actually can be stored in both of these types of workspaces.
References:
MLFlow is an open-source platform to manage the end-to-end Machine Learning lifecycle, including experiment tracking, packaging code into reproducible runs, and sharing and deploying models.
In this case, the question refers to using Azure Databricks to train an ML model and setting up MLFlow for monitoring and tracking the training process.
MLFlow allows storing the logs and model artifacts, such as trained models and visualizations, from the training process to enable easy tracking and management of the machine learning pipeline.
Now, regarding the options provided in the question:
Option A: "…only in your Azure ML Workspace" This option is not correct because MLFlow tracking cannot be set up only in the Azure ML Workspace. Even though Azure ML integrates with MLFlow, the two platforms are separate, and each has its own tracking system.
Option B: "…only in your Azure Databricks workspace" This option is not entirely accurate because MLFlow tracking can be set up solely in the Azure Databricks workspace. However, the logs and model artifacts will only be available in the Azure Databricks workspace.
Option C: "…in both of your Azure ML and Azure Databricks workspaces" This option is the correct answer. MLFlow tracking can be set up in both Azure ML and Azure Databricks workspaces. This enables users to store logs and model artifacts in both workspaces simultaneously, allowing for seamless collaboration between teams and easy access to model artifacts from multiple locations.
Option D: "…in neither of the Azure ML or Databricks workspaces." This option is incorrect as it contradicts the previous options, which indicate that MLFlow tracking can be set up in at least one of the workspaces.