cases: Model Registries

[jorgeorpinel]

Per #2544

4.2 ML Model Registry
Model lifecycle (training, shadow, active, inactive)
Automated/Continuous training (remotely)
Discovery and reusability
Deploying models
Batch scoring example
+ Real-time inference

Review app page: https://dvcorg-casesmodelregistry.gtsb.io/doc/use-cases/model-registry

[jorgeorpinel]

+ update the image

@shcheklein do you mean properly designed? That may take some time. Meanwhile we could iterate on it (here's an idea) and publish a provisional one.

[shcheklein]

@jorgeorpinel let's iterate then, the one from the idea - you mentioned that it's more like a second image, it's bit too low level for the first one I think

[aguschin]

I still miss more clear indication that MLEM helps you use your model downstream. Pack it, deploy it. But since this is the page about MR, maybe that's ok.

[jorgeorpinel]

Yeah this paragraph improves with my understanding of MLEM. Reviewing mlem.ai/doc contents has helped 😋 . I rephrased it again, PTAL @aguschin .

[shcheklein]

@jorgeorpinel good, but looks too complicated now ... can we simplify it significantly (less arrows, may be less icons here and there, etc?)

[jorgeorpinel]

I rearranged things a little both with the figure (much simpler now) and benefits list + Added some links and admons (d6d01c6).

Please the latest https://dvcorg-casesmodelregistry.gtsb.io/doc/use-cases/model-registry

[shcheklein]

@jorgeorpinel it looks great! way simpler, I think we are ready to merge and then ask folks to re-draw it if needed.

[shcheklein]

Good stuff 🎉

[jorgeorpinel]

There is a "pretty" version of the image but not really improved in terms of the design, so I'm not publishing it yet. Just for the record:

[jorgeorpinel]

Also should we link to/from https://iterative.ai/model-registry ? I just found out about that page!