You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
the NDE and SCANDAL methods currently rely on our own MAF implementation, which is not really state-of-the-art in neural density estimation anymore. I propose we switch to neural spline flows and use the implementation in the excellent nflows library.
This would break backwards compatibility with trained NDE / SCANDAL models, but I am not aware of anyone using these methods currently.
Cheers,
Johann
The text was updated successfully, but these errors were encountered:
Hi,
the NDE and SCANDAL methods currently rely on our own MAF implementation, which is not really state-of-the-art in neural density estimation anymore. I propose we switch to neural spline flows and use the implementation in the excellent nflows library.
This would break backwards compatibility with trained NDE / SCANDAL models, but I am not aware of anyone using these methods currently.
Cheers,
Johann
The text was updated successfully, but these errors were encountered: