AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is There a Way to Make It Faster?
The current implementation is a dynamic programming implementation that checks. It defines some penalties based on the characteristics of the whole paragraph. You cannot decide if something is best by, for example, checking the first line. So you have to check all the lines, recalculate all the penalties and decide on the best option. Last year we went over it again. We reduced that to a 0.5x performance to 1.1 or something like that. But there was actually a big request to enable it by default back. We might do it in the future once we see more in the real life usage.