This example demonstrates how to use Azure ML Python SDK v1 to train a basic model on an AML dataset in a compute cluster.
What does this sample demonstrate:
- The concept of the script that orchestrate a training job in Azure ML vs. the training script that isn't aware it's running in Azure ML.
- Mount a dataset that references files in Azure Blob Storage to the compute cluster for training.
- Use a curated Azure ML training environment, which is different from the environment required to run the orchestration script.
- Track training experimentation using MLFlow.
- Register the model in MLFlow model format.
What doesn't this sample demonstrate:
- Azure ML pipeline
- Comprehensive unit and integration tests
- Model inferencing