vllm.entrypoints.pooling.pooling.protocol ¶

class IOProcessorRequest(PoolingBasicRequestMixin, EncodingRequestMixin, Generic[T]):
    data: T
    task: PoolingTask = "plugin"

    def to_pooling_params(self):
        return PoolingParams()

class IOProcessorResponse(OpenAIBaseModel, Generic[T]):
    request_id: str | None = None
    """
    The request_id associated with this response
    """
    created_at: int = Field(default_factory=lambda: int(time.time()))

    data: T
    """
    When using plugins IOProcessor plugins, the actual output is generated
    by the plugin itself. Hence, we use a generic type for the response data
    """

class PoolingBytesResponse(OpenAIBaseModel):
    content: list[bytes]
    headers: dict[str, str] | None = None
    media_type: str = "application/octet-stream"

class PoolingChatRequest(
    PoolingBasicRequestMixin, ChatRequestMixin, EmbedRequestMixin, ClassifyRequestMixin
):
    task: PoolingTask | None = None

    mm_processor_kwargs: dict[str, Any] | None = Field(
        default=None,
        description=("Additional kwargs to pass to the HF processor."),
    )

    def to_pooling_params(self):
        return PoolingParams(
            truncate_prompt_tokens=self.truncate_prompt_tokens,
            dimensions=self.dimensions,
            use_activation=get_use_activation(self),
        )

class PoolingCompletionRequest(
    PoolingBasicRequestMixin,
    CompletionRequestMixin,
    EmbedRequestMixin,
    ClassifyRequestMixin,
):
    task: PoolingTask | None = None

    def to_pooling_params(self):
        return PoolingParams(
            truncate_prompt_tokens=self.truncate_prompt_tokens,
            dimensions=self.dimensions,
            use_activation=get_use_activation(self),
        )

class PoolingResponse(OpenAIBaseModel):
    id: str = Field(default_factory=lambda: f"pool-{random_uuid()}")
    object: str = "list"
    created: int = Field(default_factory=lambda: int(time.time()))
    model: str
    data: list[PoolingResponseData]
    usage: UsageInfo

class PoolingResponseData(OpenAIBaseModel):
    index: int
    object: str = "pooling"
    data: list[list[float]] | list[float] | str