Problem:
While Model Hub is introduced conceptually, it is not very clear how it fits into a real application architecture.
For example:
- When should developers call Model Hub directly vs through SDK?
- How does inference fit into a full agent pipeline?
- How is state or memory expected to be handled alongside Model Hub?
Suggestion:
Add a “Typical Architecture” section showing:
- user input → model inference → post-processing → output
- where Model Hub sits in the pipeline
- how it interacts with agents or external systems
This would help developers think in systems instead of isolated API calls.
Problem:
While Model Hub is introduced conceptually, it is not very clear how it fits into a real application architecture.
For example:
Suggestion:
Add a “Typical Architecture” section showing:
This would help developers think in systems instead of isolated API calls.