I agree with 4x4High, something doesn't sound right with why this is happening.
My first guess would be, are you sure you don't have too long of pushrods? How is your rocker geometry? Did you check it as Zim stated?
Also, you probably are already doing this, I'll mention it just in case. Make sure you are rotating the rocker shaft with the countersink/flat up. It is very easy to install the rocker shaft with the rounded part up instead, and then the poly lock will sit higher and not engage the rocker shaft correctly, causing premature failure.
Dart was the first to make those adjustable guideplates if I recall, they are excellent quality and work great on the ones I have used, much easier to get perfect geometry. I have found it difficult to get all the rocker tips aligned without them. And if you try and bend the factory units the hardened surface will flake off, not good. The Dart units have locking nuts on the center bolts, that combined with the rocker studs, I think they are plenty secure. I don't know about the other brands though, I though Dart had them patented.
Ok, hearing that makes me feel a little safer with the guide plates.
Mine came with lock nuts as well.
I'm pretty sure that I had the flat part of the rocker shaft turned up,
but I'll definately double check that, thanks!
The rocker geometry is fine I believe, but as I said it's not done the usual way with these rockers.
Here's some info:
http://en.wikipedia.org/wiki/Mid-lift
But it really doesn't describe it so well.
I think this picture describes it best:
At half lift the "back" part of the rockers should be 90* to the pushrod,
and the "front" part should be 90* to the valve stem.
This makes it slightly more efficient, or so they say.
And since I'm pretty interrested in math and geometry, I thought it sounded like a pretty neat thing.
What I didn't think about, and the shop where I bought them failed to mention to me,
is that the rocker ends up higher on the stud, I realize that now.
As a help when installing them the rockers have a machined surface on the top of them, and when that is 90* to the rocker stud, the above is also true regarding the push rod and valve stem.
So all I had to do was set the cam lobe in question at half lobe lift with an indicator dial,
and then adjust my adjustable pushrods until the machined surface and the rocker stud were square (90*) against each other at the same time as I had half valve lift on the valve measured on the retainer.
This way the roller don't travel as much back and forth on the valve stem,
At mid lift it's actually the furthest away from the rocker stud, and at full lift it's back at the place it has when the valve is closed.
I didn't have mechanical lifters for the test as recomended though.
But instead I used really soft thin wire springs instead of valve springs, so that the lifters wouldn't bleed out.
Here's a link to the instructions for the rockers if anybody's interrested:
http://www.mid-lift.com/PUB/PUB050206PA.pdf