this post was submitted on 28 Oct 2024
11 points (92.3% liked)
Machine Learning
1765 readers
1 users here now
founded 4 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Yup. If you can't rebuild their model it's not open source.
Feels like open source institutions are making definitions to accommodate the desire for more open and available models while knowing that the entities making these very powerful models would never (or legally could never) release the data. Instead of just having a coherent definition of "open source" that only applies to some legitimately open source, but much more constrained, models, they want to tempt the big boys to release models by giving them a more amenable definition.