Update project_model.py

project_model.py

@@ -204,12 +204,8 @@ def process_inputs(
 
     # Create the structured input for Gemma, including image, question, and visual context
     gemma_input = {
-        "role": "user",  # This indicates it's the user's turn
-        "content": [
-            {"type": "image", "image": session.annotated_image},  # Annotated image
-            {"type": "text", "text": question},  # The question the user asks
-            {"type": "text", "text": session.visual_context}  # Visual context generated from the image
-        ]
+        "image": session.annotated_image,  # Image context
+        "text": question + " " + session.visual_context  # Question and visual context combined as text
     }
 
     # Process the input through Gemma