Viktor Gal
Viktor Gal
@bmurauer ok so a very quick fix is that you change the labeling to 0 base 💃 but yeah this shoudl work with any arbitrary labeling... i'm working on it...
cloud is killed i'll look into the showroom thnx for reportint
ok i've just realised that the second time it is just wrapped by a unique_ptr (i read it for some reason `std::make_unique`), still i'm not so sure why do we...
@theartful why do we seed there twice the same object?
@theartful ok thnx for confirming! @lambday seems we need to update unit tests for xval mmd
@lambday this change currently breaks some mmd unit tests :( would just simply updating to new values be sufficient? only prng changes here...
yep it's the same prng all over
@lambday note that if i would just add there 2 dummy calls for the prng, before creating the folds objects, things would pass...
@lambday in fact i've just tested it locally and the hack i've written actually makes all the unit tests pass that are currently failing... so this is just the problem...
Not so sure I understand the rational for this separate prng...? And why not just to update the seed or expected value? I mean it does not use one global...