Update app.py

app.py

@@ -288,7 +288,7 @@
 
 import torch
 import transformers
-from transformers import RagRetriever, RagSequenceForGeneration, AutoTokenizer, AutoModelForCausalLM
+from transformers import RagRetriever, RagSequenceForGeneration, AutoTokenizer, AutoModelForCausalLM, pipeline
 import gradio as gr
 
 device = 'cuda' if torch.cuda.is_available() else 'cpu'
@@ -304,11 +304,12 @@ retriever = RagRetriever.from_pretrained("facebook/rag-sequence-nq", index_name=
 rag_model = RagSequenceForGeneration.from_pretrained('facebook/rag-sequence-nq', retriever=retriever)
 rag_model.retriever.init_retrieval()
 rag_model.to(device)
+
 pipe = pipeline(
 "text-generation",
 model="google/gemma-2-2b-it",
 model_kwargs={"torch_dtype": torch.bfloat16},
-device=device,  # replace with "mps" to run on a Mac device
+device=device,  
 )
 
 def strip_title(title):