Skip to main content

Module spec

Module spec 

Source
Expand description

The inference workload spec (ADR-101 section 1) and its boundary validation.

Structs§

InferenceSpec
An inference endpoint to adopt or run.
MemoryBudget
Memory the instance is expected to hold (placement input).
ServeArgs
Serve arguments. The adapter owns host and port; everything else is mapped to the runtime’s flags or passed through extra_args.

Enums§

InferFlavor
Which server software an adapter drives.
LatencyClass
interactive roles carry a latency bound; batch roles do not.

Constants§

KIND_INFERENCE
Workload kind string for inference endpoints.