News for hackers. Stuff that matters.
Re:Tree Search Distillation for Language Models Using PPO by supermdguy on Sunday March 15, 2026 @12:20AM (#47384331)
> One might note that MCTS uses more inference compute on a per-sample basis than GRPO: of course it performs better

This part confused me, it sounded like they were only doing the MCTS at train time, and then using GRPO to distill the MCTS policy into the model weights. So wouldn’t the model still have the same inference cost?

Re:Tree Search Distillation for Language Models Using PPO by at2005 on Sunday March 15, 2026 @02:50AM (#47384918)
Ah, I meant that MCTS uses more inference-time compute (over GRPO) to produce a training sample