Pick out the text.

app.py

@@ -7,7 +7,9 @@ from transformers import pipeline
 pipe = pipeline("image-to-text", model="microsoft/git-base")
 
 def classify(image):
-    return pipe(image)
+    output = pipe(image)
+    # example: [{'generated_text': 'front wheels of a truck'}]
+    return output[0]['generated_text']
     
 demo = gr.Interface(fn=classify, inputs=gr.Image(type="filepath"), outputs="text", title="Image Captioning with git-base")