[Model] Adding Support for Qwen2VL as an Embedding Model. Using MrLig…

…ht/dse-qwen2-2b-mrl-v1 (vllm-project#9944)

Signed-off-by: FurtherAI <[email protected]>
Co-authored-by: FurtherAI <[email protected]>
Signed-off-by: Tyler Michael Smith <[email protected]>

docs/source/models/supported_models.rst

@@ -584,6 +584,12 @@ Multimodal Embedding
     - :code:`TIGER-Lab/VLM2Vec-Full`
     - 🚧
     - ✅︎
+  * - :code:`Qwen2VLForConditionalGeneration`
+    - Qwen2-VL-based
+    - T + I
+    - :code:`MrLight/dse-qwen2-2b-mrl-v1`
+    - 
+    - ✅︎
 
 .. important::
   Some model architectures support both generation and embedding tasks.

docs/source/models/vlm.rst

@@ -310,4 +310,21 @@ Since the request schema is not defined by OpenAI client, we post a request to t
     response_json = response.json()
     print("Embedding output:", response_json["data"][0]["embedding"])
 
+Here is an example for serving the ``MrLight/dse-qwen2-2b-mrl-v1`` model.
+
+.. code-block:: bash
+
+    vllm serve MrLight/dse-qwen2-2b-mrl-v1 --task embedding \
+      --trust-remote-code --max-model-len 8192 --chat-template examples/template_dse_qwen2_vl.jinja
+
+.. important::
+
+    Like with VLM2Vec, we have to explicitly pass ``--task embedding``. Additionally, ``MrLight/dse-qwen2-2b-mrl-v1`` requires an EOS token for embeddings, 
+    which is handled by the jinja template.
+
+.. important::
+
+    Also important, ``MrLight/dse-qwen2-2b-mrl-v1`` requires a placeholder image of the minimum image size for text query embeddings. See the full code 
+    example below for details.
+
 A full code example can be found in `examples/openai_chat_embedding_client_for_multimodal.py <https://github.com/vllm-project/vllm/blob/main/examples/openai_chat_embedding_client_for_multimodal.py>`_.

examples/openai_chat_embedding_client_for_multimodal.py

@@ -1,33 +1,120 @@
+import argparse
+import base64
+import io
+
 import requests
+from PIL import Image
 
 image_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg"
 
-response = requests.post(
-    "http://localhost:8000/v1/embeddings",
-    json={
-        "model":
-        "TIGER-Lab/VLM2Vec-Full",
-        "messages": [{
+
+def vlm2vec():
+    response = requests.post(
+        "http://localhost:8000/v1/embeddings",
+        json={
+            "model":
+            "TIGER-Lab/VLM2Vec-Full",
+            "messages": [{
+                "role":
+                "user",
+                "content": [
+                    {
+                        "type": "image_url",
+                        "image_url": {
+                            "url": image_url
+                        }
+                    },
+                    {
+                        "type": "text",
+                        "text": "Represent the given image."
+                    },
+                ],
+            }],
+            "encoding_format":
+            "float",
+        },
+    )
+    response.raise_for_status()
+    response_json = response.json()
+
+    print("Embedding output:", response_json["data"][0]["embedding"])
+
+
+def dse_qwen2_vl(inp: dict):
+    # Embedding an Image
+    if inp["dtype"] == "image":
+        messages = [{
+            "role":
+            "user",
+            "content": [{
+                "type": "image_url",
+                "image_url": {
+                    "url": inp["image_url"],
+                }
+            }, {
+                "type": "text",
+                "text": "What is shown in this image?"
+            }]
+        }]
+    # Embedding a Text Query
+    else:
+        # MrLight/dse-qwen2-2b-mrl-v1 requires a placeholder image
+        # of the minimum input size
+        buffer = io.BytesIO()
+        image_placeholder = Image.new("RGB", (56, 56))
+        image_placeholder.save(buffer, "png")
+        buffer.seek(0)
+        image_placeholder = base64.b64encode(buffer.read()).decode('utf-8')
+        messages = [{
             "role":
             "user",
             "content": [
                 {
                     "type": "image_url",
                     "image_url": {
-                        "url": image_url
+                        "url": f"data:image/jpeg;base64,{image_placeholder}",
                     }
                 },
                 {
                     "type": "text",
-                    "text": "Represent the given image."
+                    "text": f"Query: {inp['content']}"
                 },
-            ],
-        }],
-        "encoding_format":
-        "float",
-    },
-)
-response.raise_for_status()
-response_json = response.json()
-
-print("Embedding output:", response_json["data"][0]["embedding"])
+            ]
+        }]
+
+    response = requests.post(
+        "http://localhost:8000/v1/embeddings",
+        json={
+            "model": "MrLight/dse-qwen2-2b-mrl-v1",
+            "messages": messages,
+            "encoding_format": "float",
+        },
+    )
+    response.raise_for_status()
+    response_json = response.json()
+
+    print("Embedding output:", response_json["data"][0]["embedding"])
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(
+        "Script to call a specified VLM through the API. Make sure to serve "
+        "the model with --task embedding before running this.")
+    parser.add_argument("model",
+                        type=str,
+                        choices=["vlm2vec", "dse_qwen2_vl"],
+                        required=True,
+                        help="Which model to call.")
+    args = parser.parse_args()
+
+    if args.model == "vlm2vec":
+        vlm2vec()
+    elif args.model == "dse_qwen2_vl":
+        dse_qwen2_vl({
+            "dtye": "image",
+            "image_url": image_url,
+        })
+        dse_qwen2_vl({
+            "dtype": "text",
+            "content": "What is the weather like today?",
+        })

examples/template_dse_qwen2_vl.jinja

@@ -0,0 +1,7 @@
+{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}{% raw %}<|im_start|>system
+You are a helpful assistant.<|im_end|>
+{% endraw %}{% endif %}<|im_start|>{{ message['role'] }}{% raw %}
+{% endraw %}{% if message['content'] is string %}{{ message['content'] }}<|im_end|>{% raw %}
+{% endraw %}{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>{% raw %}
+{% endraw %}{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant{% raw %}
+{% endraw %}{% endif %}<|endoftext|>

tests/conftest.py

@@ -243,6 +243,9 @@ def video_assets() -> _VideoAssets:
 class HfRunner:
 
     def wrap_device(self, x: _T, device: Optional[str] = None) -> _T:
+        if x is None or isinstance(x, (bool, )):
+            return x
+
         if device is None:
             device = "cpu" if current_platform.is_cpu() else "cuda"