多模态支持¶

本文档将引导您完成扩展基础模型的步骤，使其能够接收多模态输入。

1. 更新基础 vLLM 模型¶

假设您已经根据这些步骤在 vLLM 中实现了该模型。请按照以下步骤进一步更新模型：

实现 get_placeholder_str 以定义用于在文本提示词中表示多模态项的占位符字符串。这应该与模型的聊天模板（chat template）保持一致。

代码

class YourModelForImage2Seq(nn.Module):
    ...

    @classmethod
    def get_placeholder_str(cls, modality: str, i: int) -> str | None:
        if modality.startswith("image"):
            return "<image>"

        raise ValueError("Only image modality is supported")

在 __init__ 方法中，在 _mark_language_model 中初始化模型的语言组件，并在 _mark_tower_model 中初始化模型的多模态组件，例如：

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = "") -> None:
        super().__init__()

        config = vllm_config.model_config.hf_config

        with self._mark_tower_model(vllm_config, "image"):
            self.vision_encoder = ...
            self.multi_modal_projector = ...

        with self._mark_language_model(vllm_config):
            self.language_model = init_vllm_registered_model(
                vllm_config=vllm_config,
                hf_config=config.text_config,
                prefix=maybe_prefix(prefix, "language_model"),
            )

从 forward 方法中移除嵌入（embedding）部分

将多模态嵌入移动到 embed_multimodal。
文本嵌入和嵌入合并由 embed_input_ids 的默认实现自动处理。在大多数情况下，不需要重写它。

  def forward(
      self,
      input_ids: torch.Tensor | None,
-     pixel_values: torch.Tensor,
      positions: torch.Tensor,
      intermediate_tensors: IntermediateTensors | None = None,
      inputs_embeds: torch.Tensor | None = None,
  ) -> torch.Tensor:
-     if inputs_embeds is None:
-         inputs_embeds = self.get_input_embeddings()(input_ids)
-
-     if pixel_values is not None:
-         image_features = self.get_image_features(
-             pixel_values=pixel_values,
-         )
-         special_image_mask = self.get_placeholder_mask(
-             input_ids,
-             inputs_embeds=inputs_embeds,
-             image_features=image_features,
-         )
-         inputs_embeds = inputs_embeds.masked_scatter(
-             special_image_mask,
-             image_features,
-         )

       hidden_states = self.language_model(
           input_ids,
           positions,
           intermediate_tensors,
           inputs_embeds=inputs_embeds,
       )
     ...

+  def embed_multimodal(
+      self,
+      pixel_values: torch.Tensor,
+  ) -> MultiModalEmbeddings | None:
+      return self.get_image_features(
+          pixel_values=pixel_values,
+      )

下面我们提供了一个典型的 embed_multimodal 实现模式的样板代码，但您可以根据自己的需求进行调整。

def _process_image_input(self, image_input: YourModelImageInputs) -> torch.Tensor:
    image_features = self.vision_encoder(image_input)
    return self.multi_modal_projector(image_features)

def embed_multimodal(
    self,
    **kwargs: object,
) -> MultiModalEmbeddings | None:
    # Validate the multimodal input keyword arguments
    image_input = self._parse_and_validate_image_input(**kwargs)
    if image_input is None:
        return None

    # Run multimodal inputs through encoder and projector
    vision_embeddings = self._process_image_input(image_input)
    return vision_embeddings

重要

返回的 multimodal_embeddings 必须是形状为 (num_items, feature_size, hidden_size) 的 3D torch.Tensor，或者形状为 (feature_size, hidden_size) 的 2D torch.Tensor 的列表/元组，以便 multimodal_embeddings[i] 可以检索为请求中第 i 个多模态数据项（例如图像）生成的嵌入。

注意

默认情况下，vLLM 根据输入处理中 PlaceholderRange 所定义的位置信息，将多模态嵌入合并到文本嵌入中。此逻辑可以在 embed_input_ids 中找到。

如果您的模型在合并嵌入时需要额外的逻辑，您可以重写此方法。

完成上述步骤后，使用 SupportsMultiModal 接口更新模型类。

+ from vllm.model_executor.models.interfaces import SupportsMultiModal

- class YourModelForImage2Seq(nn.Module):
+ class YourModelForImage2Seq(nn.Module, SupportsMultiModal):

注意

模型类不一定非要命名为 *ForCausalLM。请查看 HuggingFace Transformers 文档以获取一些示例。

2. 指定处理信息¶

接下来，创建 BaseProcessingInfo 的子类，以提供与 HF 处理相关的基本信息。

最大输入项数¶

您需要重写抽象方法 get_supported_mm_limits，以返回模型支持的每个模态的最大输入项数。

例如，如果模型在每个提示词中支持任意数量的图像，但仅支持一个视频：

def get_supported_mm_limits(self) -> Mapping[str, int | None]:
    return {"image": None, "video": 1}

3. 指定 Dummy 输入¶

然后，继承 BaseDummyInputsBuilder 来构建用于 HF 处理的 Dummy 输入。处理后的输出也将用于内存分析（memory profiling）。

重写抽象方法 get_dummy_text 和 get_dummy_mm_data 以构建 Dummy 输入。这些 Dummy 输入应该导致模型处于最坏情况的内存使用状态，以便 vLLM 可以为其预留正确数量的内存。

假设内存使用量随 Token 数量的增加而增加，则可以构建 Dummy 输入以最大化输出嵌入的数量，该数量与占位符特征 Token 的数量相同。

基础示例：LLaVA无输入占位符：Fuyu

查看 HF 的 LlavaForConditionalGeneration 的代码

代码

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/llava/modeling_llava.py#L530-L544
n_image_tokens = (input_ids == self.config.image_token_index).sum().item()
n_image_features = image_features.shape[0] * image_features.shape[1]

if n_image_tokens != n_image_features:
    raise ValueError(
        f"Image features and image tokens do not match: tokens: {n_image_tokens}, features {n_image_features}"
    )
special_image_mask = (
    (input_ids == self.config.image_token_index)
    .unsqueeze(-1)
    .expand_as(inputs_embeds)
    .to(inputs_embeds.device)
)
image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)
inputs_embeds = inputs_embeds.masked_scatter(special_image_mask, image_features)

每张图像的占位符特征 Token 数量为 image_features.shape[1]。image_features 是在 get_image_features 方法内部计算的

代码

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/llava/modeling_llava.py#L290-L300
image_outputs = self.vision_tower(pixel_values, output_hidden_states=True)

selected_image_feature = image_outputs.hidden_states[vision_feature_layer]
if vision_feature_select_strategy == "default":
    selected_image_feature = selected_image_feature[:, 1:]
elif vision_feature_select_strategy == "full":
    selected_image_feature = selected_image_feature
else:
    raise ValueError(f"Unexpected select feature strategy: {self.config.vision_feature_select_strategy}")
image_features = self.multi_modal_projector(selected_image_feature)
return image_features

我们可以推断 image_features.shape[1] 是基于视觉塔（针对 llava-hf/llava-1.5-7b-hf 模型的 CLIPVisionModel）中的 image_outputs.hidden_states.shape[1]。此外，我们只需要序列长度（张量的第二维）来获取 image_features.shape[1]。序列长度由 CLIPVisionTransformer 中的初始隐藏状态决定，因为注意力机制不会改变输出隐藏状态的序列长度。

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/clip/modeling_clip.py#L1094-L1102
hidden_states = self.embeddings(pixel_values, interpolate_pos_encoding=interpolate_pos_encoding)
hidden_states = self.pre_layrnorm(hidden_states)

encoder_outputs = self.encoder(
    inputs_embeds=hidden_states,
    output_attentions=output_attentions,
    output_hidden_states=output_hidden_states,
    return_dict=return_dict,
)

为了找到序列长度，我们转到 CLIPVisionEmbeddings 的代码

代码

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/clip/modeling_clip.py#L247-L257
target_dtype = self.patch_embedding.weight.dtype
patch_embeds = self.patch_embedding(pixel_values.to(dtype=target_dtype))  # shape = [*, width, grid, grid]
patch_embeds = patch_embeds.flatten(2).transpose(1, 2)

class_embeds = self.class_embedding.expand(batch_size, 1, -1)
embeddings = torch.cat([class_embeds, patch_embeds], dim=1)
if interpolate_pos_encoding:
    embeddings = embeddings + self.interpolate_pos_encoding(embeddings, height, width)
else:
    embeddings = embeddings + self.position_embedding(self.position_ids)
return embeddings

我们可以推断出 embeddings.shape[1] == self.num_positions，其中

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/clip/modeling_clip.py#L195-L196
self.num_patches = (self.image_size // self.patch_size) ** 2
self.num_positions = self.num_patches + 1

总体而言，图像的占位符特征 Token 数量可以计算为

代码

def get_num_image_tokens(
    self,
    *,
    image_width: int,
    image_height: int,
) -> int:
    hf_config = self.get_hf_config()
    hf_processor = self.get_hf_processor()

    image_size = hf_config.vision_config.image_size
    patch_size = hf_config.vision_config.patch_size

    num_image_tokens = (image_size // patch_size) ** 2 + 1
    if hf_processor.vision_feature_select_strategy == "default":
        num_image_tokens -= 1

    return num_image_tokens

请注意，图像 Token 的数量并不取决于图像的宽度和高度。我们可以简单地使用一个 Dummy image_size 来计算多模态分析数据

代码

# NOTE: In actuality, this is usually implemented as part of the
# model's subclass of [`BaseProcessingInfo`][vllm.multimodal.processing.context.BaseProcessingInfo], but we show it as is
# here for simplicity.
def get_image_size_with_most_features(self) -> ImageSize:
    hf_config = self.get_hf_config()
    width = height = hf_config.image_size
    return ImageSize(width=width, height=height)

def get_dummy_mm_data(
    self,
    seq_len: int,
    mm_counts: Mapping[str, int],
    mm_options: Mapping[str, BaseDummyOptions],
) -> MultiModalDataDict:
    num_images = mm_counts.get("image", 0)

    target_width, target_height = \
        self.info.get_image_size_with_most_features()

    image_overrides = mm_options.get("image")

    return {
        "image": self._get_dummy_images(
            width=target_width,
            height=target_height,
            num_images=num_images,
            overrides=image_overrides,
        )
    }

对于文本，我们只需扩展模型配置中的多模态图像 Token，以匹配所需的图像数量。

def get_dummy_text(self, mm_counts: Mapping[str, int]) -> str:
    num_images = mm_counts.get("image", 0)

    processor = self.info.get_hf_processor()
    image_token = processor.image_token

    return image_token * num_images

查看 HF 的 FuyuForCausalLM 的代码

代码

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/modeling_fuyu.py#L311-L322
if image_patches is not None and past_key_values is None:
    patch_embeddings = [
        self.vision_embed_tokens(patch.to(self.vision_embed_tokens.weight.dtype))
        .squeeze(0)
        .to(inputs_embeds.device)
        for patch in image_patches
    ]
    inputs_embeds = self.gather_continuous_embeddings(
        word_embeddings=inputs_embeds,
        continuous_embeddings=patch_embeddings,
        image_patch_input_indices=image_patches_indices,
    )

批次中第 i 个项的占位符特征 Token 数量是 patch_embeddings[i].shape[0]，这与 image_patches[i].shape[0] 相同，即 num_total_patches。

与 LLaVA 不同，Fuyu 没有在建模文件中定义 Patch（图像块）的数量。我们从哪里可以获取更多信息？考虑到模型输入来自 FuyuProcessor 的输出，让我们看看预处理文件。

图像输出是在 FuyuProcessor 内部通过调用 FuyuImageProcessor.preprocess 然后调用 FuyuImageProcessor.preprocess_with_tokenizer_info 获取的。

在 FuyuImageProcessor.preprocess 中，图像被缩放并填充到目标 FuyuImageProcessor.size，并返回缩放后（但在填充前）的尺寸作为元数据。

代码

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/processing_fuyu.py#L541-L544
image_encoding = self.image_processor.preprocess(images, **output_kwargs["images_kwargs"])
batch_images = image_encoding["images"]
image_unpadded_heights = image_encoding["image_unpadded_heights"]
image_unpadded_widths = image_encoding["image_unpadded_widths"]

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/image_processing_fuyu.py#L480-L
if do_resize:
    batch_images = [
        [self.resize(image, size=size, input_data_format=input_data_format) for image in images]
        for images in batch_images
    ]

image_sizes = [get_image_size(images[0], channel_dim=input_data_format) for images in batch_images]
image_unpadded_heights = [[image_size[0]] for image_size in image_sizes]
image_unpadded_widths = [[image_size[1]] for image_size in image_sizes]

if do_pad:
    batch_images = [
        [
            self.pad_image(
                image,
                size=size,
                mode=padding_mode,
                constant_values=padding_value,
                input_data_format=input_data_format,
            )
            for image in images
        ]
        for images in batch_images
    ]

在 FuyuImageProcessor.preprocess_with_tokenizer_info 中，图像根据这些元数据被切分为 Patch（图像块）

代码

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/processing_fuyu.py#L417-L425
model_image_input = self.image_processor.preprocess_with_tokenizer_info(
    image_input=tensor_batch_images,
    image_present=image_present,
    image_unpadded_h=image_unpadded_heights,
    image_unpadded_w=image_unpadded_widths,
    image_placeholder_id=image_placeholder_id,
    image_newline_id=image_newline_id,
    variable_sized=True,
)

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/image_processing_fuyu.py#L638-L658
image_height, image_width = image.shape[1], image.shape[2]
if variable_sized:  # variable_sized=True
    new_h = min(
        image_height,
        math.ceil(image_unpadded_h[batch_index, subseq_index] / patch_height) * patch_height,
    )
    new_w = min(
        image_width,
        math.ceil(image_unpadded_w[batch_index, subseq_index] / patch_width) * patch_width,
    )
    image = image[:, :new_h, :new_w]
    image_height, image_width = new_h, new_w

num_patches = self.get_num_patches(image_height=image_height, image_width=image_width)
tensor_of_image_ids = torch.full(
    [num_patches], image_placeholder_id, dtype=torch.int32, device=image_input.device
)
patches = self.patchify_image(image=image.unsqueeze(0)).squeeze(0)
assert num_patches == patches.shape[0]

Patch 的数量进而由 FuyuImageProcessor.get_num_patches 定义

代码

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/image_processing_fuyu.py#L552-L562
patch_size = patch_size if patch_size is not None else self.patch_size
patch_height, patch_width = self.patch_size["height"], self.patch_size["width"]

if image_height % patch_height != 0:
    raise ValueError(f"{image_height=} must be divisible by {patch_height}")
if image_width % patch_width != 0:
    raise ValueError(f"{image_width=} must be divisible by {patch_width}")

num_patches_per_dim_h = image_height // patch_height
num_patches_per_dim_w = image_width // patch_width
num_patches = num_patches_per_dim_h * num_patches_per_dim_w

这些图像 Patch 对应于占位符 Token（|SPEAKER|）。因此，我们只需要最大化图像 Patch 的数量。由于输入图像首先会被调整大小以适应 image_processor.size，我们可以通过输入一个尺寸等于 image_processor.size 的图像来最大化图像 Patch 的数量。

def get_image_size_with_most_features(self) -> ImageSize:
    image_processor = self.get_image_processor()
    return ImageSize(
        width=image_processor.size["width"],
        height=image_processor.size["height"],
    )

Fuyu 在 HF 处理器的输入中不需要图像占位符，因此无论图像数量多少，Dummy 提示词文本都是空的。

def get_dummy_text(self, mm_counts: Mapping[str, int]) -> str:
    return ""

对于多模态图像分析数据，其逻辑与 LLaVA 非常相似

代码

def get_dummy_mm_data(
    self,
    seq_len: int,
    mm_counts: Mapping[str, int],
    mm_options: Mapping[str, BaseDummyOptions],
) -> MultiModalDataDict:
    target_width, target_height = \
        self.info.get_image_size_with_most_features()
    num_images = mm_counts.get("image", 0)

    image_overrides = mm_options.get("image")

    return {
        "image": self._get_dummy_images(
            width=target_width,
            height=target_height,
            num_images=num_images,
            overrides=image_overrides,
        )
    }

4. 指定处理细节¶

然后，创建 BaseMultiModalProcessor 的子类，以填写有关 HF 处理的缺失细节。

信息

多模态数据处理

重写 _get_mm_fields_config，以返回与输入多模态项相关的 HF 处理器所输出张量的模式（schema）。

基础示例：LLaVA带有后处理：Fuyu

CLIPImageProcessor 的输出是一个简单的张量，形状为 (num_images, num_channels, image_height, image_width)

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/clip/image_processing_clip.py#L339-L345
images = [
    to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
    for image in all_images
]

data = {"pixel_values": images}
return BatchFeature(data=data, tensor_type=return_tensors)

因此，我们按如下方式重写 _get_mm_fields_config

def _get_mm_fields_config(
    self,
    hf_inputs: BatchFeature,
    hf_processor_mm_kwargs: Mapping[str, object],
) -> Mapping[str, MultiModalFieldConfig]:
    return dict(
        pixel_values=MultiModalFieldConfig.batched("image"),
    )

注意

我们的实际代码还额外支持预计算的图像嵌入，可以通过 image_embeds 参数将其传递给模型。

FuyuImageProcessor.preprocess_with_tokenizer_info 的 image_patches 输出拼接了批次中属于某个项的每张图像的 Patch

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/image_processing_fuyu.py#L673-L679
        image_input_ids.append(tensor_of_image_ids)
        image_patches.append(patches)
    else:
        image_input_ids.append(torch.tensor([], dtype=torch.int32, device=image_input.device))

batch_image_input_ids.append(image_input_ids)
batch_image_patches.append(image_patches)

因此，FuyuImageProcessor 输出的 image_patches 形状为 (1, num_images, num_patches, patch_width * patch_height * num_channels)。

为了像在 LLaVA 中那样支持使用 MultiModalFieldConfig.batched，我们通过重写 BaseMultiModalProcessor._call_hf_processor 移除了额外的批次维度

代码

def _call_hf_processor(
    self,
    prompt: str,
    mm_data: Mapping[str, object],
    mm_kwargs: Mapping[str, object],
    tok_kwargs: Mapping[str, object],
) -> BatchFeature:
    processed_outputs = super()._call_hf_processor(
        prompt=prompt,
        mm_data=mm_data,
        mm_kwargs=mm_kwargs,
        tok_kwargs=tok_kwargs,
    )

    image_patches = processed_outputs.get("image_patches")
    if image_patches is not None:
        images = mm_data["images"]
        assert isinstance(images, list)

        # Original output: (1, num_images, Pn, Px * Py * C)
        # New output: (num_images, Pn, Px * Py * C)
        assert (isinstance(image_patches, list)
                and len(image_patches) == 1)
        assert (isinstance(image_patches[0], torch.Tensor)
                and len(image_patches[0]) == len(images))

        processed_outputs["image_patches"] = image_patches[0]

    return processed_outputs

注意

我们的实际代码对仅文本输入进行了特殊处理，以防止 HF 处理器发出不必要的警告。

注意

_call_hf_processor 方法指定了 mm_kwargs 和 tok_kwargs 用于处理。mm_kwargs 用于初始化和调用 huggingface 处理器，而 tok_kwargs 仅用于调用 huggingface 处理器。

这让我们可以像下面这样重写 _get_mm_fields_config

def _get_mm_fields_config(
    self,
    hf_inputs: BatchFeature,
    hf_processor_mm_kwargs: Mapping[str, object],
) -> Mapping[str, MultiModalFieldConfig]:
    return dict(image_patches=MultiModalFieldConfig.batched("image"))

提示词更新¶

重写 _get_prompt_updates 以返回 PromptUpdate 实例的列表。

每个 PromptUpdate 实例都指定了由 HF 处理器执行的更新操作（例如：插入、替换）。

基础示例：LLaVA处理附加 Token：Fuyu

查看 HF 的 LlavaProcessor

# https://github.com/huggingface/transformers/blob/v4.47.1/src/transformers/models/llava/processing_llava.py#L167-L170
prompt_strings = []
for sample in text:
    sample = sample.replace(self.image_token, self.image_token * num_image_tokens)
    prompt_strings.append(sample)

它只是将每个输入的 image_token 重复与占位符特征 Token 数量（num_image_tokens）相同的次数。基于此，我们按如下方式重写 _get_prompt_updates：

代码

def _get_prompt_updates(
    self,
    mm_items: MultiModalDataItems,
    hf_processor_mm_kwargs: Mapping[str, object],
    out_mm_kwargs: MultiModalKwargsItems,
) -> Sequence[PromptUpdate]:
    hf_config = self.info.get_hf_config()
    image_token_id = hf_config.image_token_index

    def get_replacement(item_idx: int):
        images = mm_items.get_items("image", ImageProcessorItems)

        image_size = images.get_image_size(item_idx)
        num_image_tokens = self.info.get_num_image_tokens(
            image_width=image_size.width,
            image_height=image_size.height,
        )

        return [image_token_id] * num_image_tokens

    return [
        PromptReplacement(
            modality="image",
            target=[image_token_id],
            replacement=get_replacement,
        ),
    ]

回顾步骤 2 中特征 Token 的布局

|SPEAKER||SPEAKER|...|SPEAKER||NEWLINE|
|SPEAKER||SPEAKER|...|SPEAKER||NEWLINE|
...
|SPEAKER||SPEAKER|...|SPEAKER||NEWLINE|

我们定义一个辅助函数来直接返回 ncols 和 nrows

代码

def get_image_feature_grid_size(
    self,
    *,
    image_width: int,
    image_height: int,
) -> tuple[int, int]:
    image_processor = self.get_image_processor()
    target_width = image_processor.size["width"]
    target_height = image_processor.size["height"]
    patch_width = image_processor.patch_size["width"]
    patch_height = image_processor.patch_size["height"]

    if not (image_width <= target_width and image_height <= target_height):
        height_scale_factor = target_height / image_height
        width_scale_factor = target_width / image_width
        optimal_scale_factor = min(height_scale_factor, width_scale_factor)

        image_height = int(image_height * optimal_scale_factor)
        image_width = int(image_width * optimal_scale_factor)

    ncols = math.ceil(image_width / patch_width)
    nrows = math.ceil(image_height / patch_height)
    return ncols, nrows

基于此，我们最初可以将替换 Token 定义为

代码

def get_replacement(item_idx: int):
    images = mm_items.get_items("image", ImageProcessorItems)
    image_size = images.get_image_size(item_idx)

    ncols, nrows = self.info.get_image_feature_grid_size(
        image_width=image_size.width,
        image_height=image_size.height,
    )

    # `_IMAGE_TOKEN_ID` corresponds to `|SPEAKER|`
    # `_NEWLINE_TOKEN_ID` corresponds to `|NEWLINE|`
    return ([_IMAGE_TOKEN_ID] * ncols + [_NEWLINE_TOKEN_ID]) * nrows

然而，这并不完全正确。调用 FuyuImageProcessor.preprocess_with_tokenizer_info 之后，还会在提示词中添加一个 BOS Token（<s>）

代码

# https://github.com/huggingface/transformers/blob/v4.48.3/src/transformers/models/fuyu/processing_fuyu.py#L417-L435
model_image_input = self.image_processor.preprocess_with_tokenizer_info(
    image_input=tensor_batch_images,
    image_present=image_present,
    image_unpadded_h=image_unpadded_heights,
    image_unpadded_w=image_unpadded_widths,
    image_placeholder_id=image_placeholder_id,
    image_newline_id=image_newline_id,
    variable_sized=True,
)
prompt_tokens, prompts_length = _tokenize_prompts_with_image_and_batch(
    tokenizer=self.tokenizer,
    prompts=prompts,
    scale_factors=scale_factors,
    max_tokens_to_generate=self.max_tokens_to_generate,
    max_position_embeddings=self.max_position_embeddings,
    add_BOS=True,
    add_beginning_of_answer_token=True,
)

要将视觉嵌入仅分配给图像 Token，您可以返回 PromptUpdateDetails 的实例，而不是字符串

代码

hf_config = self.info.get_hf_config()
bos_token_id = hf_config.bos_token_id  # `<s>`
assert isinstance(bos_token_id, int)

def get_replacement_fuyu(item_idx: int):
    images = mm_items.get_items("image", ImageProcessorItems)
    image_size = images.get_image_size(item_idx)

    ncols, nrows = self.info.get_image_feature_grid_size(
        image_width=image_size.width,
        image_height=image_size.height,
    )
    image_tokens = ([_IMAGE_TOKEN_ID] * ncols + [_NEWLINE_TOKEN_ID]) * nrows

    return PromptUpdateDetails.select_token_id(
        image_tokens + [bos_token_id],
        embed_token_id=_IMAGE_TOKEN_ID,
    )

最后，注意到 HF 处理器会从分词后的提示词中移除 |ENDOFTEXT| Token，我们可以搜索它以在字符串的开头进行替换

代码

def _get_prompt_updates(
    self,
    mm_items: MultiModalDataItems,
    hf_processor_mm_kwargs: Mapping[str, object],
    out_mm_kwargs: MultiModalKwargsItems,
) -> Sequence[PromptUpdate]:
    hf_config = self.info.get_hf_config()
    bos_token_id = hf_config.bos_token_id
    assert isinstance(bos_token_id, int)

    tokenizer = self.info.get_tokenizer()
    eot_token_id = tokenizer.bos_token_id
    assert isinstance(eot_token_id, int)

    def get_replacement_fuyu(item_idx: int):
        images = mm_items.get_items("image", ImageProcessorItems)
        image_size = images.get_image_size(item_idx)

        ncols, nrows = self.info.get_image_feature_grid_size(
            image_width=image_size.width,
            image_height=image_size.height,
        )
        image_tokens = ([_IMAGE_TOKEN_ID] * ncols + [_NEWLINE_TOKEN_ID]) * nrows

        return PromptUpdateDetails.select_token_id(
            image_tokens + [bos_token_id],
            embed_token_id=_IMAGE_TOKEN_ID,
        )

    return [
        PromptReplacement(
            modality="image",
            target=[eot_token_id],
            replacement=get_replacement_fuyu,
        )
    ]

定义完 BaseProcessingInfo（步骤 2）、BaseDummyInputsBuilder（步骤 3）和 BaseMultiModalProcessor（步骤 4）后，使用 MULTIMODAL_REGISTRY.register_processor 装饰模型类，以将它们注册到多模态注册表中

  from vllm.model_executor.models.interfaces import SupportsMultiModal
+ from vllm.multimodal import MULTIMODAL_REGISTRY

+ @MULTIMODAL_REGISTRY.register_processor(
+     YourMultiModalProcessor,
+     info=YourProcessingInfo,
+     dummy_inputs=YourDummyInputsBuilder,
+ )
  class YourModelForImage2Seq(nn.Module, SupportsMultiModal):

注意事项¶

插入特征 Token（不进行替换）¶

某些 HF 处理器直接插入特征 Token，而不替换原始提示词中的任何内容。在这种情况下，您可以在 _get_prompt_updates 中使用 PromptInsertion 代替 PromptReplacement。

示例

BLIP-2（在提示词开头插入）： vllm/model_executor/models/blip2.py
Molmo（在 <|endoftext|> Token 之后插入）： vllm/model_executor/models/molmo.py

_get_prompt_updates 假设每次应用提示词更新时都对应一个多模态项。如果无论有多少个多模态项，HF 处理器都会进行额外的处理，您应该重写 _apply_hf_processor_tokens_only，以使处理后的 Token 输入与在文本输入上应用 HF 处理器的结果保持一致。这是因为根据我们的设计，Token 输入会绕过 HF 处理器。

示例

Chameleon（追加 sep_token）： vllm/model_executor/models/chameleon.py
Fuyu（追加 boa_token）： vllm/model_executor/models/fuyu.py
Molmo（应用未在其他地方定义的聊天模板）： vllm/model_executor/models/molmo.py

自定义 HF 处理器¶

某些模型并未在 HF Hub 上定义 HF 处理器类。在这种情况下，您可以定义一个与 HF 处理器具有相同调用签名的自定义 HF 处理器，并将其传递给 _call_hf_processor。

示例

DeepSeek-VL2： vllm/model_executor/models/deepseek_vl2.py
InternVL： vllm/model_executor/models/internvl.py
Qwen-VL： vllm/model_executor/models/qwen_vl.py