Averaging in AdaGVAE - Githubissues

nmichlo / disent

🧶 Modular VAE disentanglement framework for python built with PyTorch Lightning ▸ Including metrics and datasets ▸ With strongly supervised, weakly supervised and unsupervised methods ▸ Easily configured and run with Hydra config ▸ Inspired by disentanglement_lib

https://disent.michlo.dev

MIT License

122 stars 18 forks source link

Averaging in AdaGVAE #32

Closed Confusezius closed 2 years ago

Confusezius commented 2 years ago

Great work with this repository!

A quick note: In the GVAE averaging operation, I assume that

https://github.com/nmichlo/disent/blob/d3aa2d30194735d13fb700ad9844e60d742ab914/disent/frameworks/vae/_weaklysupervised__adavae.py#L238

is supposed to be

ave_mean = 0.5 * (d0_posterior.mean + d1_posterior.mean)

right?

nmichlo commented 2 years ago

Oh goodness, yes. I don’t know how I let this regression slip through.

Would you like to submit a PR with the change, I’ll happily merge it?

Thank you for the kind words and submitting this issue.

nmichlo commented 2 years ago

Fixed: https://github.com/nmichlo/disent/releases/tag/v0.5.1 (have also released v0.6.0 with MPI3D fixes)

Found an extra error in the minimal version and the function above:

Same error as you found https://github.com/nmichlo/disent/blob/d3aa2d30194735d13fb700ad9844e60d742ab914/disent/frameworks/vae/_weaklysupervised__adavae.py#L221
Should be torch.where(share_mask, <ave_posterior>, <orig_posterior>) https://github.com/nmichlo/disent/blob/d3aa2d30194735d13fb700ad9844e60d742ab914/disent/frameworks/vae/_weaklysupervised__adavae.py#L326-L329