Jan

Hello @asarm Thank you for your interest:) For inference, you only need the `x` tensor, you provide it as input to your networn and you will get the predicted portfolio...

Heh, for some reason I thought we had `np.norm` in the code but it is clearly not the case. I will adjust the name of this issue.

Potentially we can run `sphinx-apidoc` on GitHub actions and then upload automatically to RTD.

See below a list of `dataset_id` that have the cutoff issue after synchronization (feel free to comment if you find others) * `717`

There is a duplicate issue #25

> How do we want to deal with slow tests? > > But I'd suggest to for sure delete `not slow` from `addopts` in tox because this way we always...

Anyway, the `todo` marker is stupid IMO.

This is a very relevant resource: https://web.archive.org/web/20170420055058/https://efferencecopy.net/allen-brain-fast-image-registration/ together with this https://github.com/charliehass/ecallen/blob/master/ecallen/images.py

This issue depends on #74

Here are the first results. We used `StratifiedKFold` with `n_splits=5`. One potential bias of our custom model (`textclassifier`) is that we used the the validations sets to do early stopping....