Move embeddings directly to CPU

app.py

@@ -48,7 +48,7 @@ generator = pipeline("text-generation", model=model, tokenizer=tokenizer, max_ne
 
 def rag_pipeline(query):
     # Embed query
-    query_embedding = embedder.encode([query], convert_to_tensor=True).cpu().numpy()
+    query_embedding = embedder.encode([query], convert_to_tensor=True, device='cpu').numpy()
 
     # Retrieve top-k from FAISS
     D, I = index.search(query_embedding, k=5)