Reputation: 1
For the project I have 2 ML models. In the input stream, in stream analytics I get my data and the type of data. I have deployed an Azure Machine Learning Endpoint with ACI. This contains both my models. I followed this link for the multiple models deployment :
And I used this link to have a second input in my run function, and also my web service for that matter. Here I want to provide the type of data so I can choose the right model for prediction for that data.I used the offical Microsoft documentation for this.
When I try to run the Stream, I get this error:
The streaming job failed: Stream Analytics job has validation errors: Multiple input columns to the end point is not currently supported.
The last link also provides information as to how to generate swagger files automatically, which is something needed for my project as well.
If you know how to get around this issue, or if you know a way to switch between models easily, either in the run function, or in the Stream Analytics Query. Please let me know.
For the choosing a model in the Query, the issue is that only one ML function is allowed per query step.
Upvotes: 0
Views: 950
Reputation: 521
Stream Analytics today expects the swagger to be of a particular format (example). The link shows the version of dependencies that must be used in the yml file to generate the swagger in that specific format.
The team is very close to supporting the new swagger format (shown in the link you shared) and when this does become available, you will be able to choose the model from the dropdown in ASA and it will work.
Upvotes: 1