Hacker News Reader

Comment by alsosprachben | Hacker News Reader

Model weights are facts. Facts are not copyrightable.

> Model weights are facts.

Only in the way a math textbooks are facts. Model weights are bits on a medium,designed and derived from the combined efforts, judgement and expertise of many humans, before and after training.

Aren't model weights the output of human (?) authored software? The training software might have received creative input, but not the weights themselves. And if the training data itself counts, then the copyright would not be the labs, who stole the training data, but all the creatives who's works were used.

Aren't model weights the output of human (?) authored software? The training software might have received creative input, but not the weights themselves

How does this legal theory square with output of a compiler? Or pictures generated by photo-editing suites when humans tweak color levels?

Both your examples take human authored input and transform them in a pretty direct way.

I addressed this by saying if the training data counts as the human input, then the creative authorship of the weights are all the creatives that involuntarily supplied the training data -- not the labs training the models.

> the creative authorship of the weights are all the creatives that involuntarily supplied the training data -- not the labs training the models.

I agree with that copyright holders that provide training data were screwed over, and should be fairly compensated.

That said, a lot of work that goes into training a model is editorial: structure of the model, training regime, etc . Copyright applies not just to the literal elements, but to Sequence, Structure and Organization, which is how poetry Anthologies are under additional copyright to the editor, and not just divided among just the poets.

The words in a novel and the order in which try appear are so facts. But somehow that is clearly copyrightable.

It’s really not that simple

Reply on news.ycombinator.com