Revert the mixin changes

vllm/model_executor/models/qwen2_vl.py

@@ -820,34 +820,7 @@ class Qwen2VLMultiModalDataParser(MultiModalDataParser):
         return super()._parse_video_data(data)
 
 
-class QwenVLSeriesProcessingInfoMixin:
+class Qwen2VLProcessingInfo(BaseProcessingInfo):
-    """
-    Mixin that provides get_num_mm_encoder_tokens()
-    and get_num_mm_connector_tokens() methods for
-    QwenVL series models without affecting other multi-modal models.
-    """
-
-    def get_num_mm_encoder_tokens(
-        self,
-        num_image_tokens: int,
-    ) -> int:
-        hf_config = self.get_hf_config()
-        vision_config = hf_config.vision_config
-        merge_size = vision_config.spatial_merge_size
-
-        return num_image_tokens * merge_size**2
-
-    def get_num_mm_connector_tokens(
-        self,
-        num_vision_tokens: int,
-    ) -> int:
-        hf_config = self.get_hf_config()
-        vision_config = hf_config.vision_config
-        merge_size = vision_config.spatial_merge_size
-        return num_vision_tokens // merge_size**2
-
-
-class Qwen2VLProcessingInfo(QwenVLSeriesProcessingInfoMixin, BaseProcessingInfo):
     def get_hf_config(self):
         return self.ctx.get_hf_config(Qwen2VLConfig)
 
@@ -1131,6 +1104,25 @@ class Qwen2VLMultiModalProcessor(BaseMultiModalProcessor[Qwen2VLProcessingInfo])
             for modality in ("image", "video")
         ]
 
+    def get_num_mm_encoder_tokens(
+        self,
+        num_image_tokens: int,
+    ) -> int:
+        hf_config = self.get_hf_config()
+        vision_config = hf_config.vision_config
+        merge_size = vision_config.spatial_merge_size
+
+        return num_image_tokens * merge_size**2
+
+    def get_num_mm_connector_tokens(
+        self,
+        num_vision_tokens: int,
+    ) -> int:
+        hf_config = self.get_hf_config()
+        vision_config = hf_config.vision_config
+        merge_size = vision_config.spatial_merge_size
+        return num_vision_tokens // merge_size**2
+
     def _get_mm_fields_config(
         self,
         hf_inputs: BatchFeature,

vllm/multimodal/processing.py

@@ -1412,6 +1412,28 @@ class BaseMultiModalProcessor(ABC, Generic[_I]):
         """
         raise NotImplementedError
 
+    def get_num_mm_encoder_tokens(
+        self,
+        num_image_tokens: int,
+    ) -> int:
+        """
+        Implement this function to enable LoRA support 
+        for the tower module of the multi-modal model
+        Given the number of image tokens, output the number of multi-modal encoder tokens
+        """
+        raise NotImplementedError
+
+    def get_num_mm_connector_tokens(
+        self,
+        num_vision_tokens: int,
+    ) -> int:
+        """
+        Implement this function to enable LoRA support
+        for the connector module of the multi-modal model
+        Given the number of vision tokens, output the number of multi-modal connector tokens
+        """
+        raise NotImplementedError
+
     def _bind_and_group_updates(
         self,
         prompt_updates: Sequence[PromptUpdate],