upvote
But it usually makes sense new models probably use somewhat same training pipelines and data.
reply