Machine Learning

• Explaining generative language models to (almost) anyone
• Partial Dependence Plots
• Partial Dependence Plots
• LIME plots
• ICE plots
• Interpretable Machine Learning
• A Visual Exploration of Gaussian Processes
• Efficient Machine Learning Inference
• Responsible AI: The Role of Data and Model Cards
• Rules of Machine Learning
• Minimizing real-time prediction serving latency in machine learning
• Introduction to the Azure ML-Ops Project Accelerator
• ML Ops Github List
• Phase Zero

• MLOps Stack Canvas

• The 2019 Accelerate State of DevOps: Elite performance, productivity, and scaling
• Introduction to Azure Storage

• Azure GitHub List

• Machine Learning operations
• Network isolation with Azure Machine Learning registries
• Share models, components, and environments across workspaces with registries
• Set up MLOps with Azure DevOps
• Network isolation with managed online endpoints
• Manage and increase quotas and limits for resources with Azure Machine Learning
• Autoscaling

Limitations:

• Inability to deploy multiple registered models to a deployment
• Inability to get the image build status when building a custom environment in a CICD pipeline
• Inability to see the managed endpoint quota limits in a UI
• Inability to downgrade SKU without causing downtime
• Scaling out instances is rather slow, therefore auto-scaling rules need to be preemptive. Downtime is to expected in certain outage scenarios or in unexpected surges of demand.
• When the outbound access is secured, can we access a storage account?