Open model is such a misnomer: it's like calling an ELF an "open executable".
Distributing things for free doesn't make them "open". The reality is that free (as in free beer) weights are closer to freeware than anything open source. In fact, since all these models are build using pirated media a more appropriate term could be plain old warez.
Getting weights without the training set and training scripts still gives you a form that's modifiable by end users, a single person can fine-tune a model. Getting the training scripts and dataset gives you nothing useful unless you have millions to burn. "Open weights" are closer to the spirit of open source than training scripts and datasets.
That's true although I think it not quite the same as freeware in the sense that we can realistically (say) fine tune an open model in ways that don't quite apply to (say) someone sending you a binary
Comments
Open model is such a misnomer: it's like calling an ELF an "open executable".
Distributing things for free doesn't make them "open". The reality is that free (as in free beer) weights are closer to freeware than anything open source. In fact, since all these models are build using pirated media a more appropriate term could be plain old warez.
Getting weights without the training set and training scripts still gives you a form that's modifiable by end users, a single person can fine-tune a model. Getting the training scripts and dataset gives you nothing useful unless you have millions to burn. "Open weights" are closer to the spirit of open source than training scripts and datasets.
That's true although I think it not quite the same as freeware in the sense that we can realistically (say) fine tune an open model in ways that don't quite apply to (say) someone sending you a binary
You can do interesting things with executable binaries as well, see WINE :)