draft support image query

modeling.py

@@ -183,10 +183,24 @@ class JinaVLForRanking(Qwen2VLForConditionalGeneration):
                 batch_inputs.append(formatting_prompts_func(q, d, query_type=query_type, doc_type=doc_type))
 
             batch_images = None
+            # if doc_type == 'image':
+            #     batch_images = load_images([d for (q, d) in mini_batch])
+            # elif query_type == 'image':
+            #     batch_images = load_images([q for (q, d) in mini_batch])
+
+            doc_images = []
+            query_images = []
             if doc_type == 'image':
-                batch_images = load_images([d for (q, d) in mini_batch])
-            elif query_type == 'image':
-                batch_images = load_images([q for (q, d) in mini_batch])
+                doc_images = load_images([d for (q, d) in mini_batch])
+            if query_type == 'image':
+                query_images = load_images([q for (q, d) in mini_batch])
+
+            if len(doc_images) == len(query_images) and len(doc_images) > 0:
+                batch_images = [[d, q] for q, d in zip(query_images, doc_images)]
+            elif len(doc_images) > 0:
+                batch_images = doc_images
+            elif len(query_images) > 0:
+                batch_images = query_images
 
             batch = self._processor(
                 text=batch_inputs,