@nwrickert this was already mentioned by @Rumraket and corrected if you read a few more comments down.
OK, one more try. From mechanical equations we can deduce what should be observed if the generalization of the behavior they represent is correct. From observations we can empirically verify that the equations are correct and tentatively confirm by inductive inference that the general behavior is such.
But to suggest that mechanical equations can be used to infer what underlies observable physical reality has no logical justification that I can see. The only logical means that I’m aware of to infer such things is through abductive reasoning, and the relevant evidence from which to infer such things would be physical observation, not abstract mechanical equations.
Assuming the above is correct, the antirealist would only be correct about theoretical concepts having no correspondence to reality by denying that abductive inference has no epistemological value, which is pretty much what seems to be the antirealist stance.
On the other hand, the realist would be wrong if he were to base his claim that they do correspond to reality if he were basing it on the assumption that its use as a conceptual framework for particular mechanical equations provides some epistemological value in some strange way. But it seems obvious, at least to me, that to justifiably claim correspondence with reality one would have to rely on abductive reasoning, i.e., following the relevant evidence where it leads.
So with that in mind, I again conclude that it seems to me that to be consistent one would have to take a position to either accept or reject the efficacy of abductive reasoning. And going back and forth when convenient would seem to be a completely arbitrary and logically unjustified move.
So for example, if one holds to the Copenhagen interpretation, it seems it would mean holding to anitrealism since its interpretation, as far as I can tell, is not attached to any observable reality and in that sense only instrumentally there to provide a conceptual framwork for the formulas. In other words, though a visualization that could in principle be real, it’s not a reality driven abductive inference from real and relevant observations.
But if one is committed to antirealism on theorectical concepts in QM, to be consistent it seems one would also be committed to antirealism on all theoretical concepts in science, including abiogenesis, relativity, big bang cosmology, evolution, etc.