3 ms·
But not the training data.
by jacquesm 2mo ago
But not the training data.
- derekdahmer 2mo agoNot sure what your point is - the articles they trained on are copyrighted and not theirs to open source. The only they can legally open source is the weights, which they have done repeatedly. My point is both Aaron was trying to make the journals public and Meta made models that encapsulate data from the journals public.