I don't shrink grain size while forging. I wait until after I'm done and am ready to heat treat. By heat treating you can reduce grain size by one of two ways, both involve cycling the steel through crystal phase changes multiple times. One method is multiple quenches, which I don't recommend because quenching can also promote cracking, from microscopic cracks to outright breakage, by multiple routes. To me there is no reason to expose the steel to those forces more often than is required. The other is multiple normalizations, which is austinizing the steel and then letting it air cool and produces less stress in the steel.
Either way that you choose, you have to do it right. First of all, all you want to do is the get the steel to change back and forth from a body centered cube to a face centered cube (austinization). Your are not concerned with getting the carbon in the steel to go back into solution so there's no real reason to do a long soak. All you have to do is to make sure that the steel is austinized throughout before cooling the steel again. The way that it works is that when the new crystals form to change the phase of the crystals they will form on the boundries of the old crystals in a smaller size. Cycling the crystals through these changes will produce progressively smaller crystals-but only if it's done right. Expose the steel to too high a temperature for too long will undue all this and actually grow the grain. You will probably do just about all the grain refinement that can be done in three cycles.
Stress relief also occures by reforming the grains with these methods, especially normalizations. There are also other things, such as alloying and what is used to deoxidize with, that have an impact on these processes but lets not get into that at this time. What I gave you is pretty much what I do.
Doug