03
Did you ever wonder how an LLM generates one token?
A distilled understanding of what happens when an LLM generates one token, with an interactive explainer.
POSTS / MEASURED FIELD NOTES
The notebook
Claims with their setup, uncertainty, and caveats attached. One substantial measurement each month, with shorter laboratory notes between.
→
02
When the benchmark waits for the server
Why a closed-loop load test can understate the queueing real arrivals experience, and how that changes my next AI-serving experiment.
01
Who is your benchmark for?
“Two defensible answers to 'how fast is this box?', and whose question each answer serves.”
Get the next measurement
I publish one measured post a month: get it by email.
No noise. Unsubscribe at any time.