More of this is welcome, they should live up their name and original purpose and share other models (code, weights, dataset) in the open source community as well.