To me, "source available" implies that everything you need to reproduce the model is also available, and that doesn't appear to be the case. How is the resulting model more "free as in freedom" than a compiled binary?
- “open weights” for no training data and no restrictions on use,
- “weights available” for no training data and restrictions on use, like in this case.
I hope we can fix the legal landscape to enable publicly sharing training data but I can't really judge the companies keeping it a secret today.
This isn't a problem because OpenAI says, "training AI models using publicly available internet materials is fair use". /s
If they're going to get away with this collecting data and having a legal chain-of-custody so you can actually say it was only used to train models and no one else has access to it goes a long way.