[Model] Add PaddleOCR-VL Model Support (#12953)

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>

python/sglang/srt/multimodal/processors/paddleocr_vlm.py

@@ -0,0 +1,30 @@
+# Reference: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/paddleocr-genai-vllm-server:latest
+# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from sglang.srt.models.paddleocr_vl import PaddleOCRVLForConditionalGeneration
+from sglang.srt.multimodal.processors.base_processor import MultimodalSpecialTokens
+from sglang.srt.multimodal.processors.qwen_vl import QwenVLImageProcessor
+
+
+class PaddleOCRVLImageProcessor(QwenVLImageProcessor):
+    models = [PaddleOCRVLForConditionalGeneration]
+
+    def __init__(self, hf_config, server_args, _processor, *args, **kwargs):
+        super().__init__(hf_config, server_args, _processor, *args, **kwargs)
+
+        self.mm_tokens = MultimodalSpecialTokens(
+            image_token="<|IMAGE_START|><|IMAGE_PLACEHOLDER|><|IMAGE_END|>",
+            image_token_id=hf_config.image_token_id,
+            video_token_id=hf_config.video_token_id,
+        ).build(_processor)

python/sglang/srt/multimodal/processors/qwen_vl.py

@@ -235,10 +235,8 @@ class QwenVLImageProcessor(SGLangBaseProcessor):
 
         super().__init__(hf_config, server_args, _processor, *args, **kwargs)
 
-        self.IM_START_TOKEN_ID = hf_config.vision_start_token_id
-        self.IM_END_TOKEN_ID = hf_config.vision_end_token_id
         self.vision_start_token_id = hf_config.vision_start_token_id
-        self.vision_end_token_id = hf_config.vision_end_token_id
+        self.vision_end_token_id = getattr(hf_config, "vision_end_token_id", None)
 
         self.audio_start_token_id = getattr(hf_config, "audio_start_token_id", None)
         self.audio_token_id = getattr(hf_config, "audio_token_id", None)
@@ -351,8 +349,8 @@ class QwenVLImageProcessor(SGLangBaseProcessor):
         return {
             "input_ids": input_ids.tolist(),
             "mm_items": mm_items,
-            "im_start_id": self.IM_START_TOKEN_ID,
-            "im_end_id": self.IM_END_TOKEN_ID,
+            "im_start_id": self.vision_start_token_id,
+            "im_end_id": self.vision_end_token_id,
             "im_token_id": self.mm_tokens.image_token_id,
             "video_token_id": self.mm_tokens.video_token_id,
             "audio_token_id": self.mm_tokens.audio_token_id,