Update app.py

app.py

@@ -9,7 +9,7 @@ from langchain.chains import ConversationalRetrievalChain
 from langchain.memory import ConversationBufferMemory
 
 # 🔹 Inicializa pipeline com modelo leve
-pipe = pipeline("text-generation", model="Qwen/Qwen2.5-1.8B-Instruct", max_new_tokens=256)
+pipe = pipeline("text2text-generation", model="google/flan-t5-small", max_new_tokens=256
 llm = HuggingFacePipeline(pipeline=pipe)
 
 # 🔹 Memória de conversação