1 paper
Hanjun Luo, Qiushi Liu, Jingya Zhang +8
Large language models (LLMs) achieve strong reasoning performance, which depends critically on inference-time decisions. Yet these decisions are commonly handled by static, one-siz…