Filippo Vicentini

>@PhilipVinc I noticed that this binarisation works well for states with local states 0, 1, 2, ..., but fails for spins (with states -1 and 1). Is there any way...

>@PhilipVinc @femtobit shall I delete the NDM stuff from this PR and make a new one with NDMMultiVal using either one hot or binary encoding? The same for the RBM?...

this would be good to go if you fix the build failure.

No probs, hope you're better. > As mentioned in my earlier comment, I think it's best not to binarise non-binary variables. I don't actually understand your earlier comment. What is...

My rationale is that if you/we want to add something to `netket.nn` it must be usable by anyone in as many problems as possible. I don't mind forcing users to...

One problem that would arise with checkpointing is how to handle MPI-training: our variational states right now store a per-rank sampler state with a per-rank key. If we are running...

Let's see if I can ~harass them into submission to the NetKet empire~ convince them to merge the PR...

So, just by asking, the PR was approved and I guess in due time it will show up in a release. So the next step for netket would be to...

Hi @rbktech, Nobody is working on this at the moment, and It would be great if you decided to tackle it! We are in the final phases of finalising the...

Good! I'd write them using pytest-benchmark, as this is already integrated in our testing environment. High-level things to test that come to my mind: - SR solving. See for example...