Daniel Rasmussen

Oh I should also mention that some of the suggestions here https://www.nengo.ai/nengo-dl/tips#speed-improvements should help with running the simulation in a loop (in particular if you can fit an unrolled model...

I think this is just an artifact of the way you are printing the profiling data. I'm guessing you've got something like ``` python def just_the_keras_call(): getattr(self.keras_model, func_type)(**func_args) ```` when...

I think the 16 vs 17 steps thing is just an environmental quirk on your machine (or at least, it doesn't show up on my machine). There is also a...

Just an update, we do plan to add some documentation to help users diagnose/fix memory issues to help resolve this, it'll just be a little while before we have a...

It's not currently possible to use an RNN cell inside a TensorNode (because you need some kind of mechanism to manage the state, fulfilling the role usually played by the...

Yeah actually thinking about it more I don't think I'd really want text wrapping, especially if we had nice auto-indenting on line breaks.

Hi Hugo, thanks for the report. We're in the midst of a significant rewrite of the Nengo GUI, so bugfixes and things like that will be fixed a bit more...

It looks like I made some changes that required later pytest versions without updating the requirements, sorry about that. Does it work with a pytest version >= 2.7?

Hi @stwrn, FYI we just did a new release that includes TF 2.10/11 support!

Merged, thank you for your contribution!