Skip to main content
Backbone feature-extractor models (H0, H0-mini, H1, …). Model specifications are loaded from YAML config files in the configs/ directory. The Backbone factory wires them together with a pluggable client so that a single Backbone(...) call works for any backend (HTTP, AWS, GCP, …). Example: 
backbone = Backbone("h0-mini", base_url="http://localhost:8080")
backbone = Backbone("h0-mini", backend="aws", endpoint_name="prod")

​
Backbone

class Backbone()
Factory for creating backbone endpoint models. Model identity (spec, endpoint path) is read from YAML config files. The factory pairs each config with a pluggable client backend to produce a ready-to-use EndpointModel. Usage: 
>>> Backbone("h0-mini", base_url="http://localhost:8080")
>>> Backbone("h0-mini", backend="aws", endpoint_name="ep")
>>> Backbone.available_backbones()
['h0-large', 'h0-mini', 'h1', 'm-optimus', 'tissue-seg']

​
Backend

class Backend(str, Enum)
Supported inference backends.

​
available_backbones

@classmethod
def available_backbones(cls) -> List[str]
Return list of all registered backbone names.