Skip to content

vllm_omni.entrypoints.omni ¶

logger `module-attribute` ¶

logger = init_logger(__name__)

Omni ¶

Bases: OmniBase

Synchronous entrypoint for offline generation.

abort ¶

abort(request_id: str | Iterable[str]) -> None

generate ¶

generate(
    prompts: OmniPromptType | Sequence[OmniPromptType],
    sampling_params_list: OmniSamplingParams
    | Sequence[OmniSamplingParams]
    | None = None,
    *,
    py_generator: Literal[True],
    use_tqdm: bool | Callable[..., tqdm] = True,
) -> Generator[OmniRequestOutput, None, None]

generate(
    prompts: OmniPromptType | Sequence[OmniPromptType],
    sampling_params_list: OmniSamplingParams
    | Sequence[OmniSamplingParams]
    | None = None,
    *,
    py_generator: Literal[False] = False,
    use_tqdm: bool | Callable[..., tqdm] = True,
) -> list[OmniRequestOutput]

generate(
    prompts: OmniPromptType | Sequence[OmniPromptType],
    sampling_params_list: OmniSamplingParams
    | Sequence[OmniSamplingParams]
    | None = None,
    *,
    py_generator: bool = False,
    use_tqdm: bool | Callable[..., tqdm] = True,
) -> (
    Generator[OmniRequestOutput, None, None]
    | list[OmniRequestOutput]
)