updated with gpu

app.py

@@ -7,6 +7,7 @@ from transformers import AutoConfig, AutoTokenizer, pipeline, AutoModelForCausal
 import torch
 import re
 import transformers
+import spaces
 
 # Initialize embeddings and ChromaDB
 model_name = "sentence-transformers/all-mpnet-base-v2"
@@ -62,6 +63,7 @@ books_db_client_retriever = RetrievalQA.from_chain_type(
 )
 
 # Function to retrieve answer using the RAG system
+@spaces.GPU(duration=120)
 def test_rag(query):
     books_retriever = books_db_client_retriever.run(query)