[asgraham]

> They can determine how much each work contributed based on those weights, so it's dishonest for them to argue it isn't possible.

I don’t know about impossible but it’s definitely not a straightforward read from the post-training weights as you’re implying, unless you’re aware of some technique I’m not aware of.

The closest you could get would be the weight differential from training with a given work. But that’s massively dependent on training order, so that it’s certainly not at all a good measure of “contribution.”