Yann Dubois

Thanks for digging into it. What exactly do you mean by shuffling ? We do no shuffle the test loader ( https://github.com/YannDubs/disentangling-vae/blob/a54b794dcf816a3892a1960b2aa1c9900cb09a16/main.py#L235 ) if that's your point. BTW : I'm...

Hey Lennart, The `compression` function was simply meant to show how to use the model. As you can see in the code it's extremely simple to compress a single batch...

Yes unfortunately removing batch compression will make the compressor very slow. But actually it’s not really needed, you can compress in batch and still save images separately. Concerning the decompressor...

I think having a single operation would be a nice addition actually because (as Miles said): + the separation between the three functions is not always clear + `reduce` and...

I agree it's essentially unusable for medium / large datasets. E.g. for 100k examples it takes 3 min for me: ```python # pip install scikit-multilearn from sklearn.datasets import make_multilabel_classification X,Y...

It seems that `make_multilabel_classification` doesn't scale well with the number of positive labels actually. ```python from sklearn.datasets import make_multilabel_classification X,Y = make_multilabel_classification(n_samples=100000, n_classes=100, n_labels=20) from skmultilearn.model_selection import iterative_train_test_split X_train, y_train,...

I also agree! And would be glad to help to maintain it: it's a shame to let this project die...

The circleci failure comes from another script (`Running pytorch example in pytorch_fairseq_roberta.md`), not this one.

@Jason3S @jmcker thanks for these snippet. I'm really bad at regex and have been trying to remove spell checking in inline code (between backticks) in comments. Any idea on how...

I see, no worries. Nothing crucial but I put a [new pretrained compressor](https://arxiv.org/abs/2106.10800) on [torchhub](https://pytorch.org/hub/), and they give the choice of downloading the model in a scripted version and mine...