Gijs van den Oord

Hi Jaro, 'forcing' is indeed not a good name for this 'modelwrapper', forcing is usually reserved for external source terms in the PDE's of some other model. So I would...

Unfortunately the BMI-tester only works for python implementations, I would expect a testing library for native implementations.

Yes these rotations are triggered by the 'data_transformation_tr', which is used if you represent the experimental data in a covariance-diagonal basis I guess. I'm not sure when this is actually...

The mask in LossInvCovmat is not used for masking training/validation I think.

> @goord Why does the experimental output have `rotation=None` while the others have `rotation=obsrot`, around [here](https://github.com/NNPDF/nnpdf/blob/bc64d5506721cff8190c73c14b268256560864f2/n3fit/src/n3fit/model_gen.py#L282)? Is that intentional? It interferes a bit with how I thought the observable computation...

> Hm I don't fully understand, but is it ok to uniformize this? I now calculate all observables without any mask once, so using the same settings, and then mask...

Hi @graziano-giuliani thanks for the feedback. If the med22 case is representative (in terms of physics and transport schemes etc.) for higher-resolution test cases, then I propose we profile the...

Hi @Radonirinaunimi I'm afraid most of our GPU budget is burnt and I'm not sure the folks at SURF are willing to give us more again... About the penalties in...

Well we can at least run a 5-day job, there is enough budget for that. In the meantime we can explore our options for more compute (Leonardo or new pilot...

Are you running on GPU or CPU @scarlehoff ?