3 ms·
Scalable Power Sampling: Training-Free Reasoning for LLMs via Distrib Sharpening
- verdverm 10mo agoSubmission title matches the paper linked in the blog post: https://www.arxiv.org/abs/2601.21590 https://www.arxiv.org/abs/2601.21590 Submitted the blog post because it has the more digestible and intuitive explanations