Update app.py

app.py

@@ -138,31 +138,16 @@ def predict(_chatbot, task_history) -> tuple:
     tokenized_inputs = tokenizer(inputs, return_tensors='pt')
     tokenized_inputs = tokenized_inputs.to(model.device)
     pred = model.generate(**tokenized_inputs)
-    response = tokenizer.decode(pred.cpu()[0], skip_special_tokens=False)
+    response = tokenizer.decode(pred.cpu()[0], skip_special_tokens=False)    
     if 'image' in query[0]:
         print("Model response:", response)
-    if 'image' in query[0]:
-        image = tokenizer.draw_bbox_on_latest_picture(response)
-        if image is not None:
-            image_path = save_image(image, uploaded_file_dir)
-            _chatbot[-1] = (chat_query, (image_path,))
-            if task_history:
-                task_history[-1] = (chat_query, image_path)
-            else:
-                task_history.append((chat_query, image_path))
-        else:
-            _chatbot[-1] = (chat_query, response)
-            if task_history:
-                task_history[-1] = (chat_query, response)
-            else:
-                task_history.append((chat_query, response))
+        image_path = response.strip()  # Ensure it's just a string path or text
+        _chatbot[-1] = (chat_query, image_path)
+        task_history[-1] = (chat_query, image_path)
     else:
-        text_response = response.strip()  # Ensure it's just a string
+        text_response = response.strip() 
         _chatbot[-1] = (chat_query, text_response)
-        if task_history:
-            task_history[-1] = (chat_query, text_response)
-        else:
-            task_history.append((chat_query, text_response))
+        task_history[-1] = (chat_query, text_response)
 
     return _chatbot, task_history