RL for LLM Reasoning Is Sparse Policy Selection, Not Capability Learning

 (arxiv.org)

1 points | by BlackGlory 2 hours ago

0 comments