Delopa on Nostr: when you talk about opening the models, do you include the datasets? Or just ...

when you talk about opening the models, do you include the datasets? Or just opensourcing models' codes and giving out weights? Because it could give me a slightly better opinion only in the second case.