Azure ML Serving Stack: T5 Model Deployment

[oginskis]

Create an inference endpoint that serves T5 Model, this could be a single component or multiple components. The following capabilities should be provided:

• [ ] Storage Container with model files should be pre-created in our Azure account and the model must be uploaded there
• [ ] Pythorch inference service endpoint component
• [ ] Inference service endpoint autoscaling configuration
• [ ] Test, that component deploys on https://github.com/epam/hub-kubeflow-stacks/issues/11 stack

[oginskis]

Decided to use KServe example mnist model: https://github.com/epam/hub-kubeflow-stacks/commit/3337e9621354fc52f8d98db4c17c31a153738e26

[oginskis]

Will try to serve T5 when less busy with client projects

[oginskis]

Outdated