Statistician Shalizi Examines Attention and LLMs via Markov Chains and Kernel Smoothing

CMU statistician Cosma Shalizi published notes examining Attention and LLMs through Markov chains and kernel smoothing. Commentators note that dismissing effective methods as 'naive special cases' often means missing their non-obvious strengths.

2026-10-13 ~ 2026-10-13 · 2 related posts

• Cosma Shalizi's notebooks read Attention and Transformers as kernel smoothing and Markov models — tak3sh8 · 2026-10-13
• When you dismiss a working method as 'naive,' you may be missing something non-obvious — tak3sh8 · 2026-10-13