Expand description
Deployment and scheduling contract for inference-engine instances.
Engine isolation is deliberately separate from model architecture. A
TorchEngine keeps one contract whether an operator places it in the host
process, a tenant subprocess, or an isolated task/microVM.
Structs§
- Inference
Deployment Profile - Explicit inference deployment profile validated before scheduling.
- Inference
Instance Id - Tenant/model/replica identity used for placement and endpoint derivation.
- Inference
Resource Budget - Per-instance resource and lifecycle limits.
Enums§
- Inference
Compute - Accelerator class reserved for each engine instance.
- Inference
Isolation Profile - Address-space isolation selected for an inference engine.
- Inference
Tenancy - Tenancy posture of the deployment hosting inference.
Functions§
- inference_
iroh_ key_ purpose - Domain-separated carrier-key purpose for one opaque service instance.