Currently an exercise should remember the data from your last set. From there you can adjust higher or lower for your current set. Is this good enough or would you like to see a rolling list? If it's the latter, I'd have to see how we can implement that within the current interface.