search

gengala / pic

Official implementation of Probabilistic Integral Circuits
https://proceedings.mlr.press/v238/gala24a.html
MIT License
8 stars 0 forks source link

DLTM Sampling #1

Open mlnpapez opened 2 weeks ago

mlnpapez commented 2 weeks ago

I reproduced the results from your paper by running train_hclt.py on the MNIST dataset (achieving 1.22 bpd); however, after using backward with the DLTM model, I received the following output: samples

Are you going to add backward also to PIC?

Many thanks.

gengala commented 2 weeks ago

Hi, many thanks for playing with the code! The backward pass will not be implemented in PICs, as sampling from those energy-based-like neural nets is not straightforward and requires approximate techniques.