Perceptron XOR within std_neural_nets is broken


The perceptron xor example linked in does not work. I opened issue #1 a few weeks ago, and am wondering if it is better to see this fixed or seek other resources for my learning.

I think I could help fix it, but I’d want to keep a line of communication open with the nengo community if I do so.

Hi Jogama,

Thank you for pointing out the issue with that example!
There are quite a few Nengo examples located in various places and we are working on consolidating and updating these examples to the latest versions of Nengo (it will take some time, we are a small team :slightly_smiling_face:), so we apologize for not getting to your issue quickly. Regardless, it was great that you reported your issue here as well as these forums are much more closely monitored.

As for the issue you are experiencing with the example, I’ve posted an in-depth response on Github on how to fix the issues to get it to run with the latest versions of Nengo.

A short summary of the issue is that we’ve modified how the nengo.Sigmoid neuron behaves (this change was made in Nengo v2.4.0), and some parameters in the example notebook has to be updated to get it to work properly.

A great many thanks for the detailed response on github!

Regarding the repository itself, I’m happy to open a pull request with your changes, or with a caution on the about the changes happening to nengo examples, to close the issue.

I can also just close the issue, if that is what’s best.

If you are comfortable with creating a PR with the changes needed to make them functional again, that’ll be great! :smile:

If not, you can close the issue, as we do have it on our list of “to-dos” to consolidate the various Nengo examples we have floating around into one common repository. While we are doing that we’ll update the examples to ensure they work in the latest version of Nengo.

I opened up a pull request. Therein, all the computing functions run. I’ve not seen version controlled notebook files, so feel free to critique or merge, as seen fit.

1 Like

Thanks for doing that! :smiley:
When I get the chance to I’ll take a look at it and merge it in.

1 Like