Igor Gustavo Hoelscher

Hey! This is something that has being a pain for us too. > Do you find decoding the hyperparameter from string to list hard? If so, what's the problem? Yes....

I believe the current data model also doesn't allow the natural use of `expire` given a Feature View TTL and #1802 will probably require custom `expire` strategies or custom logic...

@judahrand I just learned (the hard way) that the redis data model is not really being optimally used for retrieval and I saw you mentioned this in one of your...

Thanks for creating the issue @felixwang9817 > What about asking users to delete the feature view and then recreate the feature view? That's also an option, thanks! Since our FE...

@felixwang9817 if you want, I could walk you guys through our current setup, not sure if this would be of value for Feast/Tecton, but let me know. But as @klesouza...

@adchia it doesn't fully solve the issue. It's common to see that the underlying logic of a feature changed, but the base timestamp of the feature view didn't, so it...

@sihanwang41 will this come to gRPC also?

Thanks @akshay-anyscale , found all of them very useful! But I'm seeing some discrepancy between the times reported by ray-llm prometheus metrics and by `llmperf`. While evaluating a `llama2-7b` deployment...

Thank you very much @DvirDukhan (and others). From what I understand, changing the data model to `ProjectName:FeatureViewName:EntityId` (with or without Redis hashtags) means making it collocated by feature view. It...

@sudohainguyen Unfortunately, I have too much on my plate right now. We will, possibly, migrate to either DynamoDB or Cassandra (AWS Keyspaces) in the next weeks. The tradeoff between higher...