Create app.py

app.py

@@ -0,0 +1,24 @@
+import gradio as gr
+from transformers import ViltProcessor, ViltForVisualQuestionAnswering
+import torch
+
+processor = ViltProcessor.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
+model = ViltForVisualQuestionAnswering.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
+
+def answer_question(image, text):
+    encoding = processor(image, text, return_tensors="pt")
+    
+    # forward pass
+    with torch.no_grad():
+     outputs = model(**encoding)
+     
+   logits = outputs.logits
+   idx = logits.argmax(-1).item()
+   predicted_answer = model.config.id2label[idx])
+   
+   return predicted_answer
+   
+image = gr.inputs.Image(type="pil")
+question = gr.inputs.Textbox(label="Question")
+answer = gr.outputs.Textbox(label="Predicted answer")
+gr.Interface(fn=classify_image, inputs=[image, question], outputs=answer, enable_queue=True).launch(debug=True)