ML Endpoints w/Fabric

[blull]

Our company is trying to utilize Fabric for our data science needs, including machine learning models and AutoML.   One feature that Azure Machine Learning has outside of Fabric, is the ability to deploy endpoints for models to make them available to outside applications.  How can we do this in Fabric?  I can't seem to find an option.  If we have to do some sort of hybrid configuration using Azure ML endpoints, and somehow make our Fabric models available, what is the process for that?

Thank you!

[blull]

I don't suppose it's possible to get maybe an unofficial example of using the Azureml python SDK to register a Fabric mlflow trained model?  I have no problem doing this for now, and utilizing both services.  I did a quick review and I'm not seeing anything that would give me an idea of how to do this.   

[v-gchenna-msft]

Unfortunately we don't have samples/docs for this yet. We have an internal team working on this but it is not out yet.

As i already mentioned, Batch & Realtime Model endpoints are in the internal Fabric roadmap.

Appreciate if you could share the feedback on our feedback channel. Which would be open for the user community to upvote & comment on. This allows our product teams to effectively prioritize your request against our existing feature backlog and gives insight into the potential impact of implementing the suggested feature.

[v-gchenna-msft]

Hi @blull ,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond back with the more details and we will try to help .

[blull]

I shared a some feedback in the feedback channel.  Do you have any recommendations on who we can reach out to, to accomplish what you mentioned (training in fabric and deploying in AML) without samples/documentation?  This is a bit of a hold up for us for really being able to use Fabric for our needs.  Do we just have to wait and check back in every few weeks?

[v-gchenna-msft]

Hi @blull ,

Currently we don't have any ETA when will these docs and feature is going to be available.
Appreciate your patience.

[v-gchenna-msft]

Hi @blull ,

We haven’t heard from you on the last response and was just checking back to see if you got some insights over your query.  Otherwise, will respond back with the more details and we will try to help .

[blull]

I still haven't figured out anything unfortunately.

[v-gchenna-msft]

Hi @blull ,

Apologies that your query is not resolved completely. 

This might require a deeper investigation from our engineering team and they can guide you better.

Please go ahead and raise a support ticket to reach our support team:

https://support.fabric.microsoft.com/support
Please provide the ticket number here as we can keep an eye on it.

[v-gchenna-msft]

Hi @blull ,

Thanks for using Fabric Community.

As I understand you are looking for ways to deploy endpoints for models developed in Fabric.

For your information - Batch & Realtime Model endpoints are in the internal roadmap.
Currently we don't have any ETA.

But you can train a model in Fabric and publish/register it in AML using the Azureml python SDK.
After a model is registered in AML it is like any azureml model and can be deployed.
Unfortunately we don't have samples/docs for this yet. We have an internal team working on this but it is not out yet.

Appreciate if you could share the feedback on our feedback channel. Which would be open for the user community to upvote & comment on. This allows our product teams to effectively prioritize your request against our existing feature backlog and gives insight into the potential impact of implementing the suggested feature.

For more insights you can refer the below links:
Machine learning model - Microsoft Fabric | Microsoft Learn
Register and work with models - Azure Machine Learning | Microsoft Learn

Hope this helps. Please let me know if you have any further queries.