Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

◈ Vector Reads: 10-Minute LLM Paper Reads

One recent LLM paper a week, in about 10 minutes. Not a bullet-point recap of the abstract: each post walks through the real mechanism with a worked example, an analogy that actually made it click for me, the honest limitations the authors admit to themselves, and where it connects back to RAG.

Language Models Can Control Their Own Attention

A model that tells the inference engine which parts of the context it can skip, cutting attention cost by up to 52% for almost no accuracy loss.

Grounded AI: An Interactive RAG Tutorial Series
Chapter 10: The Capstone
Vector Reads
Language Models Can Control Their Own Attention