Skip to main content

Posts

2025

A minor point about precision
·1300 words
Things I learned digging into 5090 perf
·1080 words
Various approaches to parallelizing Muon
·2035 words

2024

Testing the 4090 48GB
·1211 words
Visualizing 6D Mesh Parallelism
·3383 words
Why reduction precision matters
·1153 words