You have your terminology a little mixed up. "High dynamic range" refers to an image with a large distance between the darkest shades and the brightest highlights. The human vision system is better at coping with this than film or camera sensors, so in order to approximate that (or to go beyond!), one might take an image with a dark exposure (clipping the highlights but preserving shadow detail) and one with a higher exposure (blacking out the shadows but preserving the highlights) and merge the two. A merged file that contains all of this blended dynamic range is a HDR file, but, generally, there's no way to view this given the limitations of real hardware and printing technology. (Can you image a print glowing as brightly as the sun?)
So, instead, that HDR image is merged in some way down to a single image with a representable range of tones. This is often called an HDR image but really it is a regular-dynamic-range image which was made with HDR processing.
Even if the merge wasn't extreme, it can't be done because nothing tags the data as coming from one of the originals. If a pixel in the image is medium gray, is it because it was very bright in the dark exposure, or because it was dark in the bright exposure? There's no way to tell. It's like you have a glass of water with yellow dye and a glass of water with blue dye. Mix it together and you get green water, and you can't go back.
If you think you might want to, save the originals.