الباحثون

تياني لين

المنشورات 2

نسخة أولية وصول مفتوح

The Row Normalization Puzzle in Muon

Jiayu Zhang, تياني لين · 2026

This paper examines how row-wise renormalization affects Muon, focusing on the gap between NorMuon's worst-case guarantees and its practical performance (Li et al.). Despite its growing adoption and promising performance in large language model (LLM) pretraining, NorMuon's worst-case guarantees remain poorly understood …

نسخة أولية وصول مفتوح

A Zeroth-Order Paradigm for LLM Preference Alignment

بيتر تشن, شي تشين, ووتاو ين وآخرون · 2026

Direct preference alignment methods are widely used to align large language models (LLMs) with human preferences because of their computational and memory efficiency. However, likelihood displacement motivates alternative ways to extract information from preference pairs with small likelihood margins. In this paper, we …

المؤلفون المشاركون