MCTS appears to be huge for LLMs, but why don't we try something like, idk, eXtreme Gradient Boosted Tree Search?

Source: x.com/marktenenholtz/status/1820867120064061651