Update app.py

app.py

@@ -8,7 +8,13 @@ models = {
     "BLIP": (AutoProcessor, BlipForQuestionAnswering, "Salesforce/blip-vqa-base"),
 }
 
-def get_format_response(selected_model,outputs):
+def get_format_response(image,question,selected_model):
+    # Load selected model and processor
+    processor, model_class, model_name = models[selected_model]
+    processor = processor.from_pretrained(model_name)
+    model = model_class.from_pretrained(model_name)
+    encoding = processor(image, question, return_tensors="pt")
+    outputs = model(**encoding)
     if selected_model=='ViLT':
         logits = outputs.logits
         idx = logits.argmax(-1).item()
@@ -25,11 +31,6 @@ st.subheader("A demo app showcasing VQA models. ViLT and BLIP model.")
 # Sidebar for model selection
 selected_model = st.sidebar.selectbox("Select Model", list(models.keys()))
 
-# Load selected model and processor
-processor, model_class, model_name = models[selected_model]
-processor = processor.from_pretrained(model_name)
-model = model_class.from_pretrained(model_name)
-
 # Image and question input
 uploaded_image = st.file_uploader("Upload Image")
 question = st.text_input("Ask a Question about the Image")
@@ -38,9 +39,7 @@ question = st.text_input("Ask a Question about the Image")
 if uploaded_image and question:
     image = Image.open(uploaded_image)
     st.image(image, caption="Uploaded Image")
-    encoding = processor(image, question, return_tensors="pt")
-    outputs = model(**encoding)
-    answer = get_format_response(selected_model,outputs)
+    answer = get_format_response(image,question,selected_model)
     # Display answer
     st.write(f"🤔 Model Answer: {answer} 🎉")
 # Disclaimer