Code for experiments

[SFrav]

Interesting article. Thanks for making the code and paper accessible.

Do you plan to make the code for the experiments available?

For me, I would think of incorporating this in computer vision tasks. This would be once the limitation of fixed point number limitation has been overcome.

[tk-rusch]

With this code, the experiments of the paper should be fully reproducible -- or are you talking about releasing the trained points?

[SFrav]

Thanks for the speedy reply. I see the code for the model, but not how to reproduce the examples in the SI (e.g. sections B, D.1, D.2 and D.3).
My ignorance showing here, but as a simple example of reproducibility is the result of C0 = 7.06574 estimated using the weights produced from the model as specified in the code then using latnetbuilder-t lattice-c ordinary-s N-d 32-e full-CBC-f CU:P2-q 2--weights file:/path/to/weights.txt?

[SFrav]

Just following up in case this got lost amongst other notifications.