Separando resposta no prompt

app.py

@@ -46,7 +46,10 @@ def ask_question(texto_extraido, question):
     vector_store = store_docs(docs_splitted)
     rag_chain = create_rag_chain(vector_store)
     
-    resposta = rag_chain.run(question)
+    # resposta = rag_chain.run(question)
+    response = rag_chain({"query": question})
+    resposta = response["result"]
+
     return resposta
 
 def launch_app():

rag_utils.py

@@ -98,7 +98,8 @@ def create_rag_chain(vectorstore):
         tokenizer=get_tokenizer(),
         max_new_tokens=512,
         temperature=0.1,
-        do_sample=False
+        do_sample=False,
+        return_full_text=False
     )
 
     # Adapta para LangChain
@@ -108,6 +109,7 @@ def create_rag_chain(vectorstore):
         llm=llm,
         retriever=vectorstore.as_retriever(),
         chain_type="stuff",
+        return_source_documents=True,
         chain_type_kwargs={"prompt": create_template()}
     )