It's always open weights.
It's never open data
Well, it is, it's your data to begin with after all but admitting that would create some problems.
This model is sort of interesting since it seems to be using a lot of synthetic training data – but your point stands
So it's a rip off of a rip off, is that whats interesting?
