I was about to suggest using state-change as a way to dump heat, but that means more humidity in the atmosphere which gets us closer to unlivable wet-bulb temperatures.

Then there's a transmission 'window' in our atmosphere (8-13 micrometers), meaning IR in that range can just radiate out to space. So far, the applications have been around passive cooling. I haven't dug into the numbers, but that might be a more practicable way to dump heat - and then, of course, there are other reactor designs that don't have the same need for cooling per GW output.