Coverage for src/lilbee/providers/fleet/__init__.py: 100%

1"""The local inference engine: a managed ``llama-server`` fleet. 

2 

3Chat/embed/rerank/vision each run on a managed ``llama-server`` process, reached 

4over a thin httpx client. A single machine is a fleet-of-one; the same code 

5bin-packs models across N GPUs. This is the sole local engine (no in-process 

6binding); ``llm_provider=auto`` routes native GGUF refs here. 

7"""