Maura Pintor
Maura Pintor
Thank you, it works now on the docker container from `utils/docker/Dockerfile` (it is not updated yet on DockerHub).
Hi @danif93, thank you for letting us know! We identified the problem, and we will provide a fix soon. The error is caused by an update from the foolbox library,...
The problem should be solved with the new version. Please, update to v0.15.2!
Hello @rikonaka, thank you for maintaining this library. Do you have any update on this PR? Is there something that we can do to speed up the merge?
> > Hello @rikonaka, thank you for maintaining this library. Do you have any update on this PR? Is there something that we can do to speed up the merge?...
It's more complicated than that, we should pass the preprocessing items here, as the quantization has to be done directly in the input space. I will come up with a...
> Note that quantization can also be non-uniform. Just saying that you may consider giving a less generic name to the class. Agreed. Thanks! We can create a generic superclass...
Opening new related issue (#103) on creating a constraint interface that uses preprocessing. This is useful also for the quantization.
Implementation plan, with input space constraint, and custom levels: ```python class QuantizationConstraint(InputSpaceConstraint): def __init__(self, preprocessing=None, levels=255) -> None: if isinstance(levels, int): if levels < 2: raise ValueError("Number of levels must...
Closed with #104.