add additional feature

app.py

@@ -11,6 +11,8 @@ config = ViltConfig.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
 processor = ViltProcessor.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
 model = ViltForQuestionAnswering.from_pretrained("Minqin/carets_vqa_finetuned")
 
+orig_model = ViltForQuestionAnswering("dandelin/vilt-b32-finetuned-vqa")
+
 uploaded_file = st.file_uploader("Please upload one image (jpg)", type="jpg")
 
 question = st.text_input("Type here one question on the image")
@@ -29,4 +31,9 @@ if uploaded_file is not None:
     idx = logits.argmax(-1).item()
     pred = model.config.id2label[idx]
 
-    st.text(f"Answer: {pred}")
+    orig_outputs = orig_model(**encoding)
+    orig_logits = orig_outputs.logits
+    idx = orig_logits.argmax(-1).item()
+    orig_pred = orig_model.config.id2label[idx]
+    st.text(f"Answer of ViLT: {orig_pred}")
+    st.text(f"Answer after fine-tuning: {pred}")