RESOURCES
Engineering notes, definitions, and direct answers on compute by the hour for AI jobs. Nothing here has shipped.
Topics
Training, batch inference, footage, datasets, simulation: the jobs the network is built for.
BrowseWhat never comes to the network, and the one test that sorts a workload.
BrowseStaging, isolation, the output coming back, and what is accounted for when a job ends.
BrowsePaying by the hour for a need that is heavy some of the time. No rates are published yet.
Browse
LATEST
Recently published.
- Questions
Can we join the preview?
Not yet. Design-partner intake is not open, and there is no form. The program describes the kind of team it would want first, and the best preparation is to describe a real job precisely.
· 3 min read
- Program
How the preview decides what to publish
A technical preview can say almost nothing that is not a plan or a guess. The program's rule is to publish what has been decided, name what has not, and refuse to fill the gap with claims.
· 4 min read
- Workloads
Large-model inference that can wait
The same large model can answer a person now or work through a pile of documents overnight. Only the second kind is a network job. How to tell them apart, and how the asynchronous pass is described.
· 4 min read
- Program
What is not yet open, and what that means for you
No jobs can be submitted, no rates are published, and design-partner intake is closed. A plain list of what E31 Network does not yet offer, and what a team can usefully do in the meantime.
· 4 min read
- Data and output
What we have not yet published about retention
E31 Network has not published how long staged data or returned output is kept, or how it is deleted. This article says exactly that, and explains what is known instead.
· 4 min read
- Glossary
Design partner
One of a small number of teams with heavy, bursty AI jobs who would work with the E31 Network program at the engineering level before it opens. Intake is not yet open.
· 2 min read
- Program
What a design partner would be asked to do
Intake is not yet open. When it is, a design partner would be asked to describe a real job precisely, engage at the engineering level, run it for real, and report what did not work.
· 4 min read
- Questions
How is E31 Network different from Czar?
Czar is a sealed appliance you run in your own rack on its own subscription. E31 Network is compute rented by the hour for batch jobs and released when they finish. Owned versus rented, steady versus bursty.
· 3 min read
- Workloads
Evaluation jobs and regression sweeps
Every model should be measured before it ships and against the version before. How an evaluation pass, and a full matrix of candidates against suites, would run as jobs on E31 Network.
· 4 min read
58 articles across 8 topics.