Hey guys! I'm new to Databricks and trying to figure out the best way to register my model in Databricks using Kedro. I don't want to serve my models as an API. I would just like to register them on the page below. Should I use MLflowModelTrackingDataset?

I don't really use this in Databricks, but seems right. What does the Databricks documentation says? If it says to use

log_model

, then that's right, since that's essentially what the tracking dataset does.

Yes this dataset is made for this. You can pass

save_args:
     registered_model_name: "my-model

To automatically register your dataset each time you save it.

After looking deeper, I think that the

MlflowArtifactDataset

+

PickleDataset

will be more appropriate for me, because my models are actually of a very unusual type. I'll carry out some tests and then leave some feedback in this thread. Thanks @Deepyaman Datta and @Yolan Honoré-Rougé