Stories by jxmorris12

Will Scaling Solve Robotics?

VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention

LLM Memory

What Problems to Solve (1966)

Test-Time Training

Thnickels

SFStreets: History of San Francisco place names

Muon Doesn't Clearly Grok Faster

René Girard and Mimetic Theory for Non-Philosophers

Becoming a Better Programmer by Tightening Feedback Loops

Approximating Language Model Training Data from Weights

Street photos of every building in New York in 1939/1940

The Launch of GPT-4

Superintelligence, from First Principles

How to Shuffle a Big Dataset

Q-learning is not yet scalable

Why are neural networks and cryptographic ciphers so similar?

So You Want to Work in Mechanistic Interpretability?

Bowling Alone: America's Declining Social Capital [pdf] (1995)

Introduction to Parallel Programming with CUDA

1
2
3
4
5
6
7
8
9
10