Update README.md

Merge pull request #10 from fdraxler/main

Update toy example to include 8 gaussians

Simplify dependencies for installing fff as a module

Revert commit "add embedding normalization to GraphAutoencoder"

add embedding normalization to GraphAutoencoder

Currently just copying loss.py does not work. Reflect this in the README.md

Also fix order of train and val split in labels

Prevent leaking train data into valset

device handling in energy_toy_distribution and move fix_device() to utils.utils

update toy example

Merge branch 'main' into manifold-fff

added M-FFF comments

improved docstrings

rewirte loss.py: added mfff_loss. move content of m_loss to loss.py, Restructure some stuff, add a lot of docs, add exact nll in other losses