"Unhardening" hardened punches to deform them

I assume that if you had any equipment to do it with, you wouldn't be asking the question, so this is the low tech answer:
Propane torch, heat red hot (verging on orange), and drop the part into a bucket of ashes, cover the part with ashes so there will be at least and inch between the part and air, or the walls of the container. Let it cool slowly. That should anneal it well enough to allow you to cold work it a bit.
To re-harden: propane torch again, orange hot, quench in cooking oil. Temper twice in a kitchen oven at 400 degrees, quench in oil after each temper heat.
Adam
If you know what the steel is the answer might change, so post the information if you have it.

Adam,
Your assumptions are correct. I owe you a few beers. Thank You
John

One does not quench after the draw (soak @ 400 degrees). Simply let it air cool. Once should be adequate, assuming you allow a short soak period. (1 hour/inch thickness).
Harold

You are most welcome. If it doesn't work out, let us know. There are slightly more involved ways of doing the annealing that still don't require a full blown heat treat setup. I'm betting that what you are working with will not be highly alloyed, in which case what I said will work fine.
Adam

Second draw is to temper the untempered martensite converted by the first draw, but you are probably right that there won't be any if it is low alloy steel. I was allowing for the possibility that it might be medium alloy, it can't hurt. I'm used to quenching after the temper, but that is when flame tempering. In that case the temperature distribution in the part is uneven, and I don't want the thin sections drawing heat from thick sections, so best just to cool the whole thing down. You are completely correct that after a soak temper the quench won't do anything useful.
Adam