.

utils.py

@@ -79,7 +79,7 @@ class Image2Text:
         seq = re.sub(r"<.*?>", "", seq, count=1).strip()  
         seq = self.processor.token2json(seq)
         contents = seq['content'].split('[newline]')
-        return seq, contents
+        return contents
 
     def get_text(self, img_path, num_beams=4):
         pixel_values = self.load_img(img_path)