__init__

async PDPrefillServer.__init__(llm_config: LLMConfig, *, engine_cls: Type[LLMEngine] | None = None, model_downloader: Type[LoraModelLoader] | None = None)

Asynchronous constructor that returns a fully started instance.

This is the default constructor used by Ray Serve deployments.

Parameters:

• llm_config (LLMConfig) – LLMConfig for the model.

• engine_cls (Type[LLMEngine] | None) – Dependency injection for the vllm engine class. Defaults to VLLMEngine.

• model_downloader (Type[LoraModelLoader] | None) – Dependency injection for the model downloader. Defaults to LoraModelLoader.