Update app.py

app.py

@@ -127,6 +127,7 @@ def retrieval_pipeline(query, question):
 
     return retrieved_contents, websearch_time, webcrawl_time, embedding_time, retrieval_time, links
 
+
 async def predict(message, history):
     # message is in format: "Search: <query>; Question: <question>"
     # we need to parse both parts into variables
@@ -141,12 +142,12 @@ async def predict(message, history):
         context = ""
         for title, content in retrieved_contents.items():
             context += f'Artigo "{title}"\nConteúdo:\n{content}\n\n'
-        prompt = f'{context.strip()}\n\nBaseado nos conteúdos dos artigos, responda: "{message}"\n\nPor favor, mencione a fonte da sua resposta.'
+        prompt = f'{context.strip()}\n\nBaseado nos conteúdos dos artigos, responda: "{question}"\n\nPor favor, mencione a fonte da sua resposta.'
     else:
         context = ""
         for title, content in retrieved_contents.items():
             context += f'Article "{title}"\nContent:\n{content}\n\n'
-        prompt = f'{context.strip()}\n\nBased on the article\'s contents, answer: "{message}"\n\nPlease, mention the source of your answer.'
+        prompt = f'{context.strip()}\n\nBased on the article\'s contents, answer: "{question}"\n\nPlease, mention the source of your answer.'
 
     print(prompt)
 
@@ -156,7 +157,8 @@ async def predict(message, history):
     body = { "stream": True,
              "models": [
                     "mistralai/mistral-7b-instruct:free",
-                    "openchat/openchat-7b:free"
+                    "openchat/openchat-7b:free",
+                    "huggingfaceh4/zephyr-7b-beta:free"
              ],
              "route": "fallback",
              "max_tokens": 1024,