pyknos
pyknos copied to clipboard
sbi-dev
conditional density estimation with neural networks
In line with the planned license change in [sbi](https://github.com/sbi-dev/sbi/pull/997), I suggest to change the license of pyknos to Apache 2.0 as well. Please see https://github.com/sbi-dev/sbi/pull/997 for details. Please indicate your...
The MDN class has several inline comments and TODOs that are unclear. We should refactor it and improve documentation.
After having implemented non-atomic SNPE-C, I am writing this issue to keep track of things that would have been desirable to exist in `mdn`. ### Get mixture components The only...
get_models from nflows doesn't scale, is annoying Instead, as a minimum, we want a pluggable system (=extensible) for defining models that is based on templating. * A designated directory, e.g....
See [nbstripout: stripping metadata](https://github.com/kynan/nbstripout#stripping-metadata)
We want a sample-based comparison of a ground-truth and an estimated density, or two estimated densities. See [Jan-Matthis's classifier-based 2-sample test](https://github.com/mackelab/sbibm/blob/master/sbibm/metrics/c2st.py#L9).
Building on Marcel's work, add an MDN (Mixture Density Network) implementation that can play well with e.g. SNPE-A proposal posterior corrections. This is the wishlist - [ ] basic MDN...
- [ ] Add a data generator for the branching Gaussian - [ ] Infer density using existing flows, in a notebook - [ ] Add an issue for metrics...