vllm.multimodal.utils

_M `module-attribute` ¶

_M = TypeVar('_M')

MediaConnector ¶

Source code in vllm/multimodal/utils.py

class MediaConnector:

    def __init__(
        self,
        media_io_kwargs: Optional[dict[str, dict[str, Any]]] = None,
        connection: HTTPConnection = global_http_connection,
        *,
        allowed_local_media_path: str = "",
    ) -> None:
        """
        Args:
            media_io_kwargs: Additional args passed to process media 
                             inputs, keyed by modalities. For example, 
                             to set num_frames for video, set 
                             `--media-io-kwargs '{"video":{"num_frames":40}}'`
            connection: HTTP connection client to download media contents.
            allowed_local_media_path: A local directory to load media files
                                      from.
        """
        super().__init__()

        self.media_io_kwargs: dict[str, dict[
            str, Any]] = media_io_kwargs if media_io_kwargs else {}
        self.connection = connection

        if allowed_local_media_path:
            allowed_local_media_path_ = Path(allowed_local_media_path)

            if not allowed_local_media_path_.exists():
                raise ValueError(
                    "Invalid `--allowed-local-media-path`: The path "
                    f"{allowed_local_media_path_} does not exist.")
            if not allowed_local_media_path_.is_dir():
                raise ValueError(
                    "Invalid `--allowed-local-media-path`: The path "
                    f"{allowed_local_media_path_} must be a directory.")
        else:
            allowed_local_media_path_ = None

        self.allowed_local_media_path = allowed_local_media_path_

    def _load_data_url(
        self,
        url_spec: ParseResult,
        media_io: MediaIO[_M],
    ) -> _M:
        data_spec, data = url_spec.path.split(",", 1)
        media_type, data_type = data_spec.split(";", 1)

        if data_type != "base64":
            msg = "Only base64 data URLs are supported for now."
            raise NotImplementedError(msg)

        return media_io.load_base64(media_type, data)

    def _load_file_url(
        self,
        url_spec: ParseResult,
        media_io: MediaIO[_M],
    ) -> _M:
        allowed_local_media_path = self.allowed_local_media_path
        if allowed_local_media_path is None:
            raise RuntimeError("Cannot load local files without "
                               "`--allowed-local-media-path`.")

        filepath = Path(url_spec.path)
        if allowed_local_media_path not in filepath.resolve().parents:
            raise ValueError(
                f"The file path {filepath} must be a subpath "
                f"of `--allowed-local-media-path` {allowed_local_media_path}.")

        return media_io.load_file(filepath)

    def load_from_url(
        self,
        url: str,
        media_io: MediaIO[_M],
        *,
        fetch_timeout: Optional[int] = None,
    ) -> _M:
        url_spec = urlparse(url)

        if url_spec.scheme.startswith("http"):
            connection = self.connection
            data = connection.get_bytes(url, timeout=fetch_timeout)

            return media_io.load_bytes(data)

        if url_spec.scheme == "data":
            return self._load_data_url(url_spec, media_io)

        if url_spec.scheme == "file":
            return self._load_file_url(url_spec, media_io)

        msg = "The URL must be either a HTTP, data or file URL."
        raise ValueError(msg)

    async def load_from_url_async(
        self,
        url: str,
        media_io: MediaIO[_M],
        *,
        fetch_timeout: Optional[int] = None,
    ) -> _M:
        url_spec = urlparse(url)

        if url_spec.scheme.startswith("http"):
            connection = self.connection
            data = await connection.async_get_bytes(url, timeout=fetch_timeout)

            return media_io.load_bytes(data)

        if url_spec.scheme == "data":
            return self._load_data_url(url_spec, media_io)

        if url_spec.scheme == "file":
            return self._load_file_url(url_spec, media_io)

        msg = "The URL must be either a HTTP, data or file URL."
        raise ValueError(msg)

    def fetch_audio(
        self,
        audio_url: str,
    ) -> tuple[np.ndarray, Union[int, float]]:
        """
        Load audio from a URL.
        """
        audio_io = AudioMediaIO(**self.media_io_kwargs.get("audio", {}))

        return self.load_from_url(
            audio_url,
            audio_io,
            fetch_timeout=envs.VLLM_AUDIO_FETCH_TIMEOUT,
        )

    async def fetch_audio_async(
        self,
        audio_url: str,
    ) -> tuple[np.ndarray, Union[int, float]]:
        """
        Asynchronously fetch audio from a URL.
        """
        audio_io = AudioMediaIO(**self.media_io_kwargs.get("audio", {}))

        return await self.load_from_url_async(
            audio_url,
            audio_io,
            fetch_timeout=envs.VLLM_AUDIO_FETCH_TIMEOUT,
        )

    def fetch_image(
        self,
        image_url: str,
        *,
        image_mode: str = "RGB",
    ) -> Image.Image:
        """
        Load a PIL image from a HTTP or base64 data URL.

        By default, the image is converted into RGB format.
        """
        image_io = ImageMediaIO(image_mode=image_mode,
                                **self.media_io_kwargs.get("image", {}))

        try:
            return self.load_from_url(
                image_url,
                image_io,
                fetch_timeout=envs.VLLM_IMAGE_FETCH_TIMEOUT,
            )
        except UnidentifiedImageError as e:
            # convert to ValueError to be properly caught upstream
            raise ValueError(str(e)) from e

    async def fetch_image_async(
        self,
        image_url: str,
        *,
        image_mode: str = "RGB",
    ) -> Image.Image:
        """
        Asynchronously load a PIL image from a HTTP or base64 data URL.

        By default, the image is converted into RGB format.
        """
        image_io = ImageMediaIO(image_mode=image_mode,
                                **self.media_io_kwargs.get("image", {}))

        try:
            return await self.load_from_url_async(
                image_url,
                image_io,
                fetch_timeout=envs.VLLM_IMAGE_FETCH_TIMEOUT,
            )
        except UnidentifiedImageError as e:
            # convert to ValueError to be properly caught upstream
            raise ValueError(str(e)) from e

    def fetch_video(
        self,
        video_url: str,
        *,
        image_mode: str = "RGB",
    ) -> tuple[npt.NDArray, dict[str, Any]]:
        """
        Load video from a HTTP or base64 data URL.
        """
        image_io = ImageMediaIO(image_mode=image_mode,
                                **self.media_io_kwargs.get("image", {}))
        video_io = VideoMediaIO(image_io,
                                **self.media_io_kwargs.get("video", {}))

        return self.load_from_url(
            video_url,
            video_io,
            fetch_timeout=envs.VLLM_VIDEO_FETCH_TIMEOUT,
        )

    async def fetch_video_async(
        self,
        video_url: str,
        *,
        image_mode: str = "RGB",
    ) -> tuple[npt.NDArray, dict[str, Any]]:
        """
        Asynchronously load video from a HTTP or base64 data URL.

        By default, the image is converted into RGB format.
        """
        image_io = ImageMediaIO(image_mode=image_mode,
                                **self.media_io_kwargs.get("image", {}))
        video_io = VideoMediaIO(image_io,
                                **self.media_io_kwargs.get("video", {}))

        return await self.load_from_url_async(
            video_url,
            video_io,
            fetch_timeout=envs.VLLM_VIDEO_FETCH_TIMEOUT,
        )

    def fetch_image_embedding(
        self,
        data: str,
    ) -> torch.Tensor:
        """
        Load image embedding from a URL.
        """
        image_embedding_io = ImageEmbeddingMediaIO()

        return image_embedding_io.load_base64("", data)

allowed_local_media_path `instance-attribute` ¶

allowed_local_media_path = allowed_local_media_path_

connection `instance-attribute` ¶

connection = connection

media_io_kwargs `instance-attribute` ¶

media_io_kwargs: dict[str, dict[str, Any]] = (
    media_io_kwargs if media_io_kwargs else {}
)

init ¶

__init__(
    media_io_kwargs: Optional[
        dict[str, dict[str, Any]]
    ] = None,
    connection: HTTPConnection = global_http_connection,
    *,
    allowed_local_media_path: str = "",
) -> None

Parameters:

Name	Type	Description	Default
`media_io_kwargs`	`Optional[dict[str, dict[str, Any]]]`	Additional args passed to process media inputs, keyed by modalities. For example, to set num_frames for video, set `--media-io-kwargs '{"video":{"num_frames":40}}'`	`None`
`connection`	`HTTPConnection`	HTTP connection client to download media contents.	`global_http_connection`
`allowed_local_media_path`	`str`	A local directory to load media files from.	`''`

Source code in vllm/multimodal/utils.py

def __init__(
    self,
    media_io_kwargs: Optional[dict[str, dict[str, Any]]] = None,
    connection: HTTPConnection = global_http_connection,
    *,
    allowed_local_media_path: str = "",
) -> None:
    """
    Args:
        media_io_kwargs: Additional args passed to process media 
                         inputs, keyed by modalities. For example, 
                         to set num_frames for video, set 
                         `--media-io-kwargs '{"video":{"num_frames":40}}'`
        connection: HTTP connection client to download media contents.
        allowed_local_media_path: A local directory to load media files
                                  from.
    """
    super().__init__()

    self.media_io_kwargs: dict[str, dict[
        str, Any]] = media_io_kwargs if media_io_kwargs else {}
    self.connection = connection

    if allowed_local_media_path:
        allowed_local_media_path_ = Path(allowed_local_media_path)

        if not allowed_local_media_path_.exists():
            raise ValueError(
                "Invalid `--allowed-local-media-path`: The path "
                f"{allowed_local_media_path_} does not exist.")
        if not allowed_local_media_path_.is_dir():
            raise ValueError(
                "Invalid `--allowed-local-media-path`: The path "
                f"{allowed_local_media_path_} must be a directory.")
    else:
        allowed_local_media_path_ = None

    self.allowed_local_media_path = allowed_local_media_path_

_load_data_url ¶

_load_data_url(
    url_spec: ParseResult, media_io: MediaIO[_M]
) -> _M

Source code in vllm/multimodal/utils.py

def _load_data_url(
    self,
    url_spec: ParseResult,
    media_io: MediaIO[_M],
) -> _M:
    data_spec, data = url_spec.path.split(",", 1)
    media_type, data_type = data_spec.split(";", 1)

    if data_type != "base64":
        msg = "Only base64 data URLs are supported for now."
        raise NotImplementedError(msg)

    return media_io.load_base64(media_type, data)

_load_file_url ¶

_load_file_url(
    url_spec: ParseResult, media_io: MediaIO[_M]
) -> _M

Source code in vllm/multimodal/utils.py

def _load_file_url(
    self,
    url_spec: ParseResult,
    media_io: MediaIO[_M],
) -> _M:
    allowed_local_media_path = self.allowed_local_media_path
    if allowed_local_media_path is None:
        raise RuntimeError("Cannot load local files without "
                           "`--allowed-local-media-path`.")

    filepath = Path(url_spec.path)
    if allowed_local_media_path not in filepath.resolve().parents:
        raise ValueError(
            f"The file path {filepath} must be a subpath "
            f"of `--allowed-local-media-path` {allowed_local_media_path}.")

    return media_io.load_file(filepath)

fetch_audio ¶

fetch_audio(
    audio_url: str,
) -> tuple[ndarray, Union[int, float]]

Load audio from a URL.

Source code in vllm/multimodal/utils.py

def fetch_audio(
    self,
    audio_url: str,
) -> tuple[np.ndarray, Union[int, float]]:
    """
    Load audio from a URL.
    """
    audio_io = AudioMediaIO(**self.media_io_kwargs.get("audio", {}))

    return self.load_from_url(
        audio_url,
        audio_io,
        fetch_timeout=envs.VLLM_AUDIO_FETCH_TIMEOUT,
    )

fetch_audio_async `async` ¶

fetch_audio_async(
    audio_url: str,
) -> tuple[ndarray, Union[int, float]]

Asynchronously fetch audio from a URL.

Source code in vllm/multimodal/utils.py

async def fetch_audio_async(
    self,
    audio_url: str,
) -> tuple[np.ndarray, Union[int, float]]:
    """
    Asynchronously fetch audio from a URL.
    """
    audio_io = AudioMediaIO(**self.media_io_kwargs.get("audio", {}))

    return await self.load_from_url_async(
        audio_url,
        audio_io,
        fetch_timeout=envs.VLLM_AUDIO_FETCH_TIMEOUT,
    )

fetch_image ¶

fetch_image(
    image_url: str, *, image_mode: str = "RGB"
) -> Image

Load a PIL image from a HTTP or base64 data URL.

By default, the image is converted into RGB format.

Source code in vllm/multimodal/utils.py

def fetch_image(
    self,
    image_url: str,
    *,
    image_mode: str = "RGB",
) -> Image.Image:
    """
    Load a PIL image from a HTTP or base64 data URL.

    By default, the image is converted into RGB format.
    """
    image_io = ImageMediaIO(image_mode=image_mode,
                            **self.media_io_kwargs.get("image", {}))

    try:
        return self.load_from_url(
            image_url,
            image_io,
            fetch_timeout=envs.VLLM_IMAGE_FETCH_TIMEOUT,
        )
    except UnidentifiedImageError as e:
        # convert to ValueError to be properly caught upstream
        raise ValueError(str(e)) from e

fetch_image_async `async` ¶

fetch_image_async(
    image_url: str, *, image_mode: str = "RGB"
) -> Image

Asynchronously load a PIL image from a HTTP or base64 data URL.

By default, the image is converted into RGB format.

Source code in vllm/multimodal/utils.py

async def fetch_image_async(
    self,
    image_url: str,
    *,
    image_mode: str = "RGB",
) -> Image.Image:
    """
    Asynchronously load a PIL image from a HTTP or base64 data URL.

    By default, the image is converted into RGB format.
    """
    image_io = ImageMediaIO(image_mode=image_mode,
                            **self.media_io_kwargs.get("image", {}))

    try:
        return await self.load_from_url_async(
            image_url,
            image_io,
            fetch_timeout=envs.VLLM_IMAGE_FETCH_TIMEOUT,
        )
    except UnidentifiedImageError as e:
        # convert to ValueError to be properly caught upstream
        raise ValueError(str(e)) from e

fetch_image_embedding ¶

fetch_image_embedding(data: str) -> Tensor

Load image embedding from a URL.

Source code in vllm/multimodal/utils.py

def fetch_image_embedding(
    self,
    data: str,
) -> torch.Tensor:
    """
    Load image embedding from a URL.
    """
    image_embedding_io = ImageEmbeddingMediaIO()

    return image_embedding_io.load_base64("", data)

fetch_video ¶

fetch_video(
    video_url: str, *, image_mode: str = "RGB"
) -> tuple[NDArray, dict[str, Any]]

Load video from a HTTP or base64 data URL.

Source code in vllm/multimodal/utils.py

def fetch_video(
    self,
    video_url: str,
    *,
    image_mode: str = "RGB",
) -> tuple[npt.NDArray, dict[str, Any]]:
    """
    Load video from a HTTP or base64 data URL.
    """
    image_io = ImageMediaIO(image_mode=image_mode,
                            **self.media_io_kwargs.get("image", {}))
    video_io = VideoMediaIO(image_io,
                            **self.media_io_kwargs.get("video", {}))

    return self.load_from_url(
        video_url,
        video_io,
        fetch_timeout=envs.VLLM_VIDEO_FETCH_TIMEOUT,
    )

fetch_video_async `async` ¶

fetch_video_async(
    video_url: str, *, image_mode: str = "RGB"
) -> tuple[NDArray, dict[str, Any]]

Asynchronously load video from a HTTP or base64 data URL.

By default, the image is converted into RGB format.

Source code in vllm/multimodal/utils.py

async def fetch_video_async(
    self,
    video_url: str,
    *,
    image_mode: str = "RGB",
) -> tuple[npt.NDArray, dict[str, Any]]:
    """
    Asynchronously load video from a HTTP or base64 data URL.

    By default, the image is converted into RGB format.
    """
    image_io = ImageMediaIO(image_mode=image_mode,
                            **self.media_io_kwargs.get("image", {}))
    video_io = VideoMediaIO(image_io,
                            **self.media_io_kwargs.get("video", {}))

    return await self.load_from_url_async(
        video_url,
        video_io,
        fetch_timeout=envs.VLLM_VIDEO_FETCH_TIMEOUT,
    )

load_from_url ¶

load_from_url(
    url: str,
    media_io: MediaIO[_M],
    *,
    fetch_timeout: Optional[int] = None,
) -> _M

Source code in vllm/multimodal/utils.py

def load_from_url(
    self,
    url: str,
    media_io: MediaIO[_M],
    *,
    fetch_timeout: Optional[int] = None,
) -> _M:
    url_spec = urlparse(url)

    if url_spec.scheme.startswith("http"):
        connection = self.connection
        data = connection.get_bytes(url, timeout=fetch_timeout)

        return media_io.load_bytes(data)

    if url_spec.scheme == "data":
        return self._load_data_url(url_spec, media_io)

    if url_spec.scheme == "file":
        return self._load_file_url(url_spec, media_io)

    msg = "The URL must be either a HTTP, data or file URL."
    raise ValueError(msg)

load_from_url_async `async` ¶

load_from_url_async(
    url: str,
    media_io: MediaIO[_M],
    *,
    fetch_timeout: Optional[int] = None,
) -> _M

Source code in vllm/multimodal/utils.py

async def load_from_url_async(
    self,
    url: str,
    media_io: MediaIO[_M],
    *,
    fetch_timeout: Optional[int] = None,
) -> _M:
    url_spec = urlparse(url)

    if url_spec.scheme.startswith("http"):
        connection = self.connection
        data = await connection.async_get_bytes(url, timeout=fetch_timeout)

        return media_io.load_bytes(data)

    if url_spec.scheme == "data":
        return self._load_data_url(url_spec, media_io)

    if url_spec.scheme == "file":
        return self._load_file_url(url_spec, media_io)

    msg = "The URL must be either a HTTP, data or file URL."
    raise ValueError(msg)

encode_audio_base64 ¶

encode_audio_base64(
    audio: ndarray, sampling_rate: float
) -> str

Encode audio as base64.

Source code in vllm/multimodal/utils.py

def encode_audio_base64(
    audio: np.ndarray,
    sampling_rate: float,
) -> str:
    """Encode audio as base64."""
    audio_io = AudioMediaIO()
    return audio_io.encode_base64((audio, sampling_rate))

encode_image_base64 ¶

encode_image_base64(
    image: Image,
    *,
    image_mode: str = "RGB",
    format: str = "JPEG",
) -> str

Encode a pillow image to base64 format.

By default, the image is converted into RGB format before being encoded.

Source code in vllm/multimodal/utils.py

def encode_image_base64(
    image: Image.Image,
    *,
    image_mode: str = "RGB",
    format: str = "JPEG",
) -> str:
    """
    Encode a pillow image to base64 format.

    By default, the image is converted into RGB format before being encoded.
    """
    image_io = ImageMediaIO(image_mode=image_mode)
    return image_io.encode_base64(image, image_format=format)

encode_video_base64 ¶

encode_video_base64(frames: NDArray) -> str

Source code in vllm/multimodal/utils.py

def encode_video_base64(frames: npt.NDArray) -> str:
    image_io = ImageMediaIO()
    video_io = VideoMediaIO(image_io)
    return video_io.encode_base64(frames)

fetch_audio ¶

fetch_audio(
    audio_url: str,
    audio_io_kwargs: Optional[dict[str, Any]] = None,
) -> tuple[ndarray, Union[int, float]]

Parameters:

Name	Type	Description	Default
`audio_url`	`str`	URL of the audio file to fetch.	required
`audio_io_kwargs`	`Optional[dict[str, Any]]`	Additional kwargs passed to handle audio IO.	`None`

Source code in vllm/multimodal/utils.py

def fetch_audio(
    audio_url: str,
    audio_io_kwargs: Optional[dict[str, Any]] = None,
) -> tuple[np.ndarray, Union[int, float]]:
    """
    Args:
        audio_url: URL of the audio file to fetch.
        audio_io_kwargs: Additional kwargs passed to handle audio IO.
    """
    media_io_kwargs = None if not audio_io_kwargs else {
        "audio": audio_io_kwargs
    }
    media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
    return media_connector.fetch_audio(audio_url)

fetch_image ¶

fetch_image(
    image_url: str,
    image_io_kwargs: Optional[dict[str, Any]] = None,
) -> Image

Parameters:

Name	Type	Description	Default
`image_url`	`str`	URL of the image file to fetch.	required
`image_io_kwargs`	`Optional[dict[str, Any]]`	Additional kwargs passed to handle image IO.	`None`

Source code in vllm/multimodal/utils.py

def fetch_image(
    image_url: str,
    image_io_kwargs: Optional[dict[str, Any]] = None,
) -> Image.Image:
    """
    Args:
        image_url: URL of the image file to fetch.
        image_io_kwargs: Additional kwargs passed to handle image IO.
    """
    media_io_kwargs = None if not image_io_kwargs else {
        "image": image_io_kwargs
    }
    media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
    return media_connector.fetch_image(image_url)

fetch_video ¶

fetch_video(
    video_url: str,
    video_io_kwargs: Optional[dict[str, Any]] = None,
) -> tuple[NDArray, dict[str, Any]]

Parameters:

Name	Type	Description	Default
`video_url`	`str`	URL of the video file to fetch.	required
`video_io_kwargs`	`Optional[dict[str, Any]]`	Additional kwargs passed to handle video IO.	`None`

Source code in vllm/multimodal/utils.py

def fetch_video(
    video_url: str,
    video_io_kwargs: Optional[dict[str, Any]] = None,
) -> tuple[npt.NDArray, dict[str, Any]]:
    """
    Args:
        video_url: URL of the video file to fetch.
        video_io_kwargs: Additional kwargs passed to handle video IO.
    """
    media_io_kwargs = None if not video_io_kwargs else {
        "video": video_io_kwargs
    }
    media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
    return media_connector.fetch_video(video_url)

group_mm_inputs_by_modality ¶

group_mm_inputs_by_modality(
    mm_inputs: list[MultiModalKwargs],
) -> list[list[MultiModalKwargs]]

Group consecutive MultiModalKwargs from mm_inputs with the same modality together into the same list for batching purpose. For MultiModalKwargs with multiple modalities, put them into their own list.

Parameters:

Name	Type	Description	Default
`mm_inputs`	`list[MultiModalKwargs]`	List of MultiModalKwargs.	required

Returns:

Type	Description
`list[list[MultiModalKwargs]]`	list[list[vllm.multimodal.MultiModalKwargs]]: List of list of
`list[list[MultiModalKwargs]]`	`MultiModalKwargs`, each inner list contains consecutive
`list[list[MultiModalKwargs]]`	`MultiModalKwargs` with same modality.

Source code in vllm/multimodal/utils.py

def group_mm_inputs_by_modality(
        mm_inputs: list[MultiModalKwargs]) -> list[list[MultiModalKwargs]]:
    """Group consecutive MultiModalKwargs from mm_inputs with the same modality
    together into the same list for batching purpose. For MultiModalKwargs with
    multiple modalities, put them into their own list.

    Args:
        mm_inputs: List of MultiModalKwargs.

    Returns:
        list[list[vllm.multimodal.MultiModalKwargs]]: List of list of
        `MultiModalKwargs`, each inner list contains consecutive
        `MultiModalKwargs` with same modality.
    """
    if not mm_inputs:
        return []

    def modality_group_func(mm_input: MultiModalKwargs) -> Union[str, int]:
        # If the input has multiple modalities, return a id as the unique key
        # for the mm_input input.
        if len(mm_input.modalities) > 1:
            return id(mm_input)

        elif len(mm_input.modalities) == 1:
            return list(mm_input.modalities)[0]

        # FIXME(Isotr0py): Modality of mm_input from legacy pipeline is empty,
        # this is used to make InternVL with legacy pipeline still work with v1.
        else:
            return ""

    return [
        list(group) for _, group in groupby(mm_inputs, key=modality_group_func)
    ]

merge_and_sort_multimodal_metadata ¶

merge_and_sort_multimodal_metadata(
    mm_positions: MultiModalPlaceholderDict,
    mm_hashes: Optional[MultiModalHashDict],
) -> tuple[
    list[str], list[PlaceholderRange], Optional[list[str]]
]

Given a MultiModalPlaceholderDict, merge all PlaceholderRange objects from all available modalities into a single list of PlaceholderRange, sorted by their offset (starting index in the input sequence) in the ascending order.

Optionally if a MultiModalHashDict is given, same operation will be applied to the object and the sorted list of hashes will be returned.

Returns:

Type	Description
`list[str]`	list[str]: List of item modalities in order of their positions in the
`list[PlaceholderRange]`	input sequence.
`Optional[list[str]]`	list[PlaceholderRange]: Sorted list of all PlaceholderRanges from
`tuple[list[str], list[PlaceholderRange], Optional[list[str]]]`	mm_positions.
`tuple[list[str], list[PlaceholderRange], Optional[list[str]]]`	Optional[list[str]]: Sorted list of all hashes from mm_hashes if given,
`tuple[list[str], list[PlaceholderRange], Optional[list[str]]]`	None otherwise.

Source code in vllm/multimodal/utils.py

def merge_and_sort_multimodal_metadata(
    mm_positions: MultiModalPlaceholderDict,
    mm_hashes: Optional[MultiModalHashDict],
) -> tuple[list[str], list[PlaceholderRange], Optional[list[str]]]:
    """Given a MultiModalPlaceholderDict, merge all PlaceholderRange
    objects from all available modalities into a single list of 
    PlaceholderRange, sorted by their offset (starting index in the input
    sequence) in the ascending order.

    Optionally if a `MultiModalHashDict` is given, same operation will be
    applied to the object and the sorted list of hashes will be returned.

    Returns:
        list[str]: List of item modalities in order of their positions in the
        input sequence.
        list[PlaceholderRange]: Sorted list of all PlaceholderRanges from
        mm_positions.
        Optional[list[str]]: Sorted list of all hashes from mm_hashes if given,
        None otherwise.
    """

    modalities = list(mm_positions.keys())

    assert len(modalities) > 0, "No modalities found in the mm_positions."

    # For single modality, placeholder ranges and hashes are already sorted
    # so we can return the list directly.
    if len(modalities) == 1:
        modality = modalities[0]
        placeholder_list = list(mm_positions[modality])

        return [modality] * len(
            placeholder_list
        ), placeholder_list, None if not mm_hashes else mm_hashes[modality]

    # Create a list of (modality, placeholder, hash) tuples for all placeholders
    all_items = []
    for modality in modalities:
        placeholder_list = list(mm_positions[modality])
        hash_list: list[Optional[str]] = list(
            mm_hashes[modality]) if mm_hashes and modality in mm_hashes else [
                None
            ] * len(placeholder_list)

        for placeholder, hash_value in zip(placeholder_list, hash_list):
            all_items.append((modality, placeholder, hash_value))

    # Sort all items by offset
    all_items.sort(key=lambda x: x[1].offset)

    # Split into separate lists
    sorted_modalities = [item[0] for item in all_items]
    merged_placeholders = [item[1] for item in all_items]
    merged_hashes = [str(item[2])
                     for item in all_items] if mm_hashes is not None else None

    return sorted_modalities, merged_placeholders, merged_hashes

run_dp_sharded_vision_model ¶

run_dp_sharded_vision_model(
    image_input: Tensor, vision_model: Module
) -> Tensor

Run a vision model with data parallelism (DP) sharding. The function will shard the input image tensor on the first dimension and run the vision model

Parameters:

Name	Type	Description	Default
`image_input`	`Tensor`	Image input tensor.	required
`vision_model`	`Module`	Vision model.	required

Returns:

Type	Description
`Tensor`	torch.Tensor: Output image embeddings

Source code in vllm/multimodal/utils.py

def run_dp_sharded_vision_model(image_input: torch.Tensor,
                                vision_model: torch.nn.Module) -> torch.Tensor:
    """Run a vision model with data parallelism (DP) sharding. The function 
    will shard the input image tensor on the first dimension and run the vision
    model

    Args:
        image_input (torch.Tensor): Image input tensor.
        vision_model (torch.nn.Module): Vision model.

    Returns:
        torch.Tensor: Output image embeddings
    """

    num_chunks = image_input.shape[0]
    mp_world_size = get_tensor_model_parallel_world_size()
    num_chunks_per_rank = (num_chunks + mp_world_size - 1) // mp_world_size
    num_padded_chunks = num_chunks_per_rank * mp_world_size - num_chunks
    pad = (0, ) * (2 * (image_input.dim() - 1)) + (0, num_padded_chunks)
    image_input_padded = torch.nn.functional.pad(image_input, pad)
    rank = get_tensor_model_parallel_rank()
    image_input_per_rank = image_input_padded[rank *
                                              num_chunks_per_rank:(rank + 1) *
                                              num_chunks_per_rank, ...]

    vision_embeddings = vision_model(image_input_per_rank)
    vision_embeddings = tensor_model_parallel_all_gather(vision_embeddings,
                                                         dim=0)
    vision_embeddings = vision_embeddings[:num_chunks, ...]
    return vision_embeddings

vllm.multimodal.utils

_M module-attribute ¶

MediaConnector ¶

allowed_local_media_path instance-attribute ¶

connection instance-attribute ¶

media_io_kwargs instance-attribute ¶

__init__ ¶

_load_data_url ¶

_load_file_url ¶

fetch_audio ¶

fetch_audio_async async ¶

fetch_image ¶

fetch_image_async async ¶

fetch_image_embedding ¶

fetch_video ¶

fetch_video_async async ¶

load_from_url ¶

load_from_url_async async ¶

encode_audio_base64 ¶

encode_image_base64 ¶

encode_video_base64 ¶

fetch_audio ¶

fetch_image ¶

fetch_video ¶

group_mm_inputs_by_modality ¶

merge_and_sort_multimodal_metadata ¶

run_dp_sharded_vision_model ¶

_M `module-attribute` ¶

allowed_local_media_path `instance-attribute` ¶

connection `instance-attribute` ¶

media_io_kwargs `instance-attribute` ¶

init ¶

fetch_audio_async `async` ¶

fetch_image_async `async` ¶

fetch_video_async `async` ¶

load_from_url_async `async` ¶