[Core] Support image processor (#4197)

docs/source/conf.py

@@ -90,6 +90,7 @@ autodoc_mock_imports = [
     "sentencepiece",
     "vllm.cuda_utils",
     "vllm._C",
+    "PIL",
     "numpy",
     "tqdm",
     "tensorizer",
@@ -116,12 +117,13 @@ class MockedClassDocumenter(autodoc.ClassDocumenter):
 autodoc.ClassDocumenter = MockedClassDocumenter
 
 intersphinx_mapping = {
-    'python': ('https://docs.python.org/3', None),
-    'typing_extensions':
-    ('https://typing-extensions.readthedocs.io/en/latest', None),
-    'numpy': ('https://numpy.org/doc/stable', None),
-    'torch': ('https://pytorch.org/docs/stable', None),
-    'psutil': ('https://psutil.readthedocs.io/en/stable', None),
+    "python": ("https://docs.python.org/3", None),
+    "typing_extensions":
+    ("https://typing-extensions.readthedocs.io/en/latest", None),
+    "pillow": ("https://pillow.readthedocs.io/en/stable", None),
+    "numpy": ("https://numpy.org/doc/stable", None),
+    "torch": ("https://pytorch.org/docs/stable", None),
+    "psutil": ("https://psutil.readthedocs.io/en/stable", None),
 }
 
 autodoc_preserve_defaults = True

docs/source/dev/multimodal/multimodal_index.rst

@@ -0,0 +1,51 @@
+Multi-Modality
+==============
+
+.. currentmodule:: vllm.multimodal
+    
+vLLM provides experimental support for multi-modal models through the :mod:`vllm.multimodal` package.
+
+:class:`vllm.inputs.PromptStrictInputs` accepts an additional attribute ``multi_modal_data``
+which allows you to pass in multi-modal input alongside text and token prompts.
+
+By default, vLLM models do not support multi-modal inputs. To enable multi-modal support for a model,
+you must decorate the model class with :meth:`MULTIMODAL_REGISTRY.register_dummy_data <MultiModalRegistry.register_dummy_data>`,
+as well as :meth:`MULTIMODAL_REGISTRY.register_input <MultiModalRegistry.register_input>` for each modality type to support.
+
+.. contents::
+   :local:
+   :backlinks: none
+
+Module Contents
++++++++++++++++
+
+.. automodule:: vllm.multimodal
+
+Registry
+--------
+
+.. data:: vllm.multimodal.MULTIMODAL_REGISTRY
+
+    The global :class:`MultiModalRegistry` which is used by model runners.
+
+.. autoclass:: vllm.multimodal.MultiModalRegistry
+    :members:
+    :show-inheritance:
+
+Base Classes
+------------
+
+.. autoclass:: vllm.multimodal.MultiModalData
+    :members:
+    :show-inheritance:
+
+.. autoclass:: vllm.multimodal.MultiModalPlugin
+    :members:
+    :show-inheritance:
+
+Image Classes
+-------------
+
+.. automodule:: vllm.multimodal.image
+    :members:
+    :show-inheritance:

docs/source/index.rst

@@ -88,6 +88,7 @@ Documentation
    models/adding_model
    models/engine_args
    models/lora
+   models/vlm
    models/performance
 
 .. toctree::
@@ -99,17 +100,18 @@ Documentation
    quantization/fp8_e4m3_kvcache
 
 .. toctree::
-   :maxdepth: 2
+   :maxdepth: 1
    :caption: Developer Documentation
    
    dev/sampling_params
    dev/offline_inference/offline_index
    dev/engine/engine_index
    dev/kernel/paged_attention
+   dev/multimodal/multimodal_index
    dev/dockerfile/dockerfile
 
 .. toctree::
-   :maxdepth: 2
+   :maxdepth: 1
    :caption: Community
 
    community/meetups

docs/source/models/supported_models.rst

@@ -87,6 +87,10 @@ Alongside each architecture, we include some popular models that use it.
     - LLaMA, Llama 2, Meta Llama 3, Vicuna, Alpaca, Yi
     - :code:`meta-llama/Meta-Llama-3-8B-Instruct`, :code:`meta-llama/Meta-Llama-3-70B-Instruct`, :code:`meta-llama/Llama-2-13b-hf`, :code:`meta-llama/Llama-2-70b-hf`, :code:`openlm-research/open_llama_13b`, :code:`lmsys/vicuna-13b-v1.3`, :code:`01-ai/Yi-6B`, :code:`01-ai/Yi-34B`, etc.
     - ✅︎
+  * - :code:`LlavaForConditionalGeneration`
+    - LLaVA-1.5
+    - :code:`llava-hf/llava-1.5-7b-hf`\*, :code:`llava-hf/llava-1.5-13b-hf`\*, etc.
+    -
   * - :code:`MiniCPMForCausalLM`
     - MiniCPM
     - :code:`openbmb/MiniCPM-2B-sft-bf16`, :code:`openbmb/MiniCPM-2B-dpo-bf16`, etc.

docs/source/models/vlm.rst

@@ -0,0 +1,56 @@
+.. _vlm:
+
+Using VLMs
+==========
+
+This document shows you how to run and serve Vision Language Models (VLMs) using vLLM.
+
+Engine Arguments
+----------------
+
+The following :ref:`engine arguments <engine_args>` are specific to VLMs:
+
+.. argparse::
+    :module: vllm.engine.arg_utils
+    :func: _vlm_engine_args_parser
+    :prog: -m vllm.entrypoints.openai.api_server
+    :nodefaultconst:
+
+Offline Batched Inference
+-------------------------
+
+To initialize a VLM, the aforementioned arguments must be passed to the ``LLM`` class for instantiating the engine.
+
+.. code-block:: python
+
+    llm = LLM(
+        model="llava-hf/llava-1.5-7b-hf",
+        image_input_type="pixel_values",
+        image_token_id=32000,
+        image_input_shape="1,3,336,336",
+        image_feature_size=576,
+    )
+
+For now, we only support a single image per text prompt. To pass an image to the model, note the following in :class:`vllm.inputs.PromptStrictInputs`:
+
+* ``prompt``: The prompt should have a number of ``<image>`` tokens equal to ``image_feature_size``.
+* ``multi_modal_data``: This should be an instance of :class:`~vllm.multimodal.image.ImagePixelData` or :class:`~vllm.multimodal.image.ImageFeatureData`.
+
+.. code-block:: python
+
+    prompt = "<image>" * 576 + (
+        "\nUSER: What is the content of this image?\nASSISTANT:")
+
+    # Load the image using PIL.Image
+    image = ...
+
+    outputs = llm.generate({
+        "prompt": prompt,
+        "multi_modal_data": ImagePixelData(image),
+    })
+
+    for o in outputs:
+        generated_text = o.outputs[0].text
+        print(generated_text)
+
+A code example can be found in `examples/llava_example.py <https://github.com/vllm-project/vllm/blob/main/examples/llava_example.py>`_.