These are the results that are verified. The math
What we don't know about is if our model (and not the math or the results) has anything to do with "reality" (and that may not even be possible)
Take the particle/wave duality for example. We're talking about an electron being 'a particle' but sometimes 'a wave'. Of course, we can predict things about the behaviour of this electron, but the story behind it, of particles, waves, may not be accurate, and this may be replaced by something that completely does away with electrons (difficult to imagine since electrons have mass) but comes with similar predictions.
Edit: hnhg: You're right, rephrased properly :)
Personally, I hate this view. There are actual facts, how things work, and there are our best educated guesses about what we're observing. One of these things is what I would call a scientific fact, the other is our best educated guess (and I don't mean that as negative as it sounds; we can do no better and may never be able to).
I'm fine with saying "this is the best we can do so we have to go with it". I'm not fine with saying "this is the best we can do so this is simply the fact itself" as this leads to ridiculous situations like saying "X is a fact" when in reality there are more questions than answers about X.
To move on to some philosophical background behind your assertion: the idea that there is some 'real', 'objective', 'true' reality that is modeled by these models, is a pre-Kantian point of view. A post-Kantian point of view is that these models of how the universe works are our reality. They are the only thing we can possibly have knowledge of: the models we use to categorize and structure our observations.
Most of them only exist in our brains. They got there by nature and nurture, without us usually spelling them out. They govern our daily behavior. We've tried to formalize the more complex ones in language that is independent of individual humans.
Sometimes observations are not consistent with a model, so we revise the model. Still we only know the new version of the model, which describes what we expect to observe.