Isaac Godfried

Yeah I think the newer version by default uses PostgreSQL to save scrapping times as shelving is unstable. So I think you need to set the following `db = Database(os.environ['db'],...

At the moment Postgres is not part of the Dockerfile as it would take up too much memory. The way I was using is as part of a docker-compose with...

I think you just need the app id and secret which you should be able to get just in development mode. You should be able to get an app in...

Temporarily reverted back to old JSON message in order to keep Kafka running. As soon as we have a proper schema will use this. Any help is appreciated.

That is not currently supported in our framework. I think there are a couple of other projects that allow you to scrape profiles. If you would like to add that...

Thanks for your follow-up. I unfortunately don't think that this will work. My current problem is that I need to serve ONNX models originally exported from PyTorch inside a [Flink...

Thanks for the suggestion, but I'm primarily looking at serving PyTorch models and not Tensorflow ones. That is why I'm exporting to ONNX.

Honestly surprised that is working at all. The only model at the moment that supports using TemporalEmbeddings is the `Informer` model which takes it as a separate parameter. Could you...

Yeah you probably want to use the decoded predictions, specifically https://github.com/AIStream-Peelout/flow-forecast/blob/5ebd89dcb758ea5d3fea9e3bd0ece94566ba3026/flood_forecast/temporal_decoding.py however that is Informer specific. The `generate_predictions_non_decoded` hasn't been tested in a while and might be dropped entirely in...

This issue should be pretty easy for first timers to fix