RESOURCES
Open-weight models, adapters trained on site, and what runs where on a Czar or a Grid.
Articles in Models
- Models
Vision-language-action models served from the rack
Vision-language-action models connect what a robot sees and is told to what it does next. They are heavy. Substrate serves them from the rack, with short budgets and clean fallback, so the robot need not carry them.
· 5 min read
- Models
Who owns an adapter trained on site
An adapter produced by the Adaptation service is trained on the customer's data, on the customer's appliance. It belongs to the customer, stays on site, and is theirs to serve, copy, or delete.
· 4 min read
- Models
Adapting a model on the site's own data
The Adaptation service fine-tunes an open model on the site's own captures and documents, entirely on the appliance. The adapter it produces stays on site and is served behind the same API.
· 4 min read
- Models
Which model runs where on a Grid
A Czar Grid holds many models across several nodes. Where each one is loaded decides how well it batches, how close it sits to its urgent callers, and how the pool behaves when a node leaves.
· 4 min read
- Models
Open-weight models on the appliance
Substrate serves open-weight models: language, vision, speech, embedding, and vision-language-action. Open weights are what let a model run entirely inside the building and be adapted there.
· 4 min read