Pranav Subramani

Benchmarking Memory Taken By Neural Network

3

Consider the following code taken from the JAX repository: https://github.com/google/jax/blob/master/examples/differentially_private_sgd.py I'd like to understand how to profile the memory taken by the neural network in an epoch. I looked into:...

Multivariate Gaussians

7

### Description Implement Multivariate leaves for SPNs #### Expected behavior: Multivariate gaussians work as expected and can be used in SPNs as leaves.

Feature Request: Prometheus structure learning algorithm

1

### Description To add to the base of the library, the structure learning algorithm, named prometheus (http://proceedings.mlr.press/v72/jaini18a.html). #### Expected behavior: Should be outputting a learned structure following the SPFlow data...

Work in Progress (*please don't merge yet*) - Added basic Edge/Node abstraction to compute the SPN layerwise - Added basic network architecture from TorchSPN to convert the spn structure to...

Review the function + comment if there's anything that can be fixed