Update app.py

app.py

@@ -20,7 +20,7 @@ def predict(prompt, image_url, max_length):
     inputs = processor(prompts[0], return_tensors="pt").to(device)
     generated_ids = model.generate(**inputs, max_length=max_length)
     generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
-    return generated_text
+    return image, generated_text
 
 
 
@@ -40,7 +40,10 @@ io = gr.Interface(fn=predict,
                       gr.Textbox(label="image URL", placeholder="Insert the URL of the image to be described"),
                       gr.Slider(label="Max tokens", value=64, max=128, min=16, step=8)
                   ],
-                  outputs=gr.Textbox(label="IDEFICS Description"),
+                  outputs=[
+                      gr.Image(type='pil'),
+                      gr.Textbox(label="IDEFICS Description")
+                  ],
                   title=title, description=description, examples=examples,
                   allow_flagging=False, allow_screenshot=False)
 io.launch(debug=True)