Enrico Bothmann

In principle, yes. However, if there are currently issues with that (and waiting for more input for the scoring, which I am not familiar with), it's maybe a good idea to ignore them for the time being ...

But regarding the negativity of d_sigma, yes, that can happen, even though we would usually regard this value as a probability for the event to happen. But it's a bit more complicated than this in practice. The negative values here come from the so-called parton density function, which can become negative (at higher orders in perturbation theory). Note that this is only a statement about individual points in the sample. Overall, with enough statistics, you will always get a positive sigma for any meaningful physics observable.

This might be a question for @grossim (?)

Initially the idea was to provide data for different final-state particle multiplicities. In that case, each file would have a different n (but it would be constant in each data file). So the statement about the differing dimensionality is a bit misleading, given that you are in the final form of the project only provided a single file for a single multiplicity. Sorry about that.

Yes. With the cross sections predicted by your model, which should only be trained using events.hdf5.

Hi @jag, the events.hdf5 file is the training data set. The events-validation.hdf5 is data that should not be used for training. It can be used to check how well the trained model predicts unknown points.