Karl Higley

@mkumari-ed Any chance you could share the model definition code and/or TF saved model file you're using to help me replicate the issue? (Slack would be fine if that works...

This is no longer the same (original) issue, which is why it was marked as done.

We do have some batch prediction functionality for models already, but it's not quite structured in a way that would make it a reasonable foundation for batch processing of graphs....

I think we should add an example for every new piece of significant functionality (i.e. almost all roadmap issues.)

It is up to date with the current state of our knowledge of the work

Excerpted from NVIDIA-Merlin/NVTabular#205, which is specifically focused on improving testing

Another related issue (potentially as follow-up work to this) is that we'd like to remove the dataloaders and framework utils from NVTabular since they've moved to Merlin Models, but the...

Sounds good to me! Go ahead and edit the list to make it reflect how you'd like to tackle the work? (I'm just the scribe here, trying to make sure...

@MarkMoTrin When you say "encoder-based embeddings as additional features," does that imply encoding an item attribute (like "text description") using an external model (like BERT) and supplying the resulting embeddings...

I have no idea (nor do I particularly care whether we run integration tests with real or synthetic data.) I've just written down what people kept telling me was supposed...