AI RESEARCH
Power-SMC: Low-Latency Sequence-Level Power Sampling for Training-Free LLM Reasoning
arXiv CS.LG
•
ArXi:2602.10273v2 Announce Type: replace-cross Many recent reasoning gains in large language models can be explained as distribution sharpening: biasing generation toward high-likelihood trajectories already ed by the pretrained model, rather than modifying its weights. A natural formalization is the sequence-level power distribution $\pi_\alpha(y\mid x)\propto p_\theta(y\mid x)^\alpha$ ($\alpha>1$), which concentrates mass on whole sequences instead of adjusting token-level temperature.