[Model][VLM] Support R-4B Model (#23246)

Signed-off-by: yannqi <yannqi@qq.com>
Signed-off-by: 杨奇(yann qi) <51905299+yannqi@users.noreply.github.com>
Signed-off-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: yannqiyang <yannqiyang@tencent.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

examples/offline_inference/vision_language_multi_image.py

@@ -992,6 +992,39 @@ def load_qwen2_5_vl(question: str, image_urls: list[str]) -> ModelRequestData:
     )
 
 
+def load_r_vl(question: str, image_urls: list[str]) -> ModelRequestData:
+    model_name = "YannQi/R-4B"
+    engine_args = EngineArgs(
+        model=model_name,
+        max_model_len=16384,
+        max_num_seqs=16,
+        limit_mm_per_prompt={"image": len(image_urls)},
+    )
+
+    placeholders = [{"type": "image", "image": url} for url in image_urls]
+    messages = [
+        {
+            "role": "user",
+            "content": [
+                *placeholders,
+                {"type": "text", "text": question},
+            ],
+        }
+    ]
+
+    processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
+
+    prompt = processor.apply_chat_template(
+        messages, tokenize=False, add_generation_prompt=True
+    )
+
+    return ModelRequestData(
+        engine_args=engine_args,
+        prompt=prompt,
+        image_data=[fetch_image(url) for url in image_urls],
+    )
+
+
 def load_smolvlm(question: str, image_urls: list[str]) -> ModelRequestData:
     model_name = "HuggingFaceTB/SmolVLM2-2.2B-Instruct"
 
@@ -1193,6 +1226,7 @@ model_example_map = {
     "qwen_vl_chat": load_qwen_vl_chat,
     "qwen2_vl": load_qwen2_vl,
     "qwen2_5_vl": load_qwen2_5_vl,
+    "rvl": load_r_vl,
     "smolvlm": load_smolvlm,
     "step3": load_step3,
     "tarsier": load_tarsier,