Update llm.py

llm.py

@@ -1,10 +1,9 @@
 from transformers import pipeline
 import torch
 
-device = 0 if torch.cuda.is_available() else -1
 qa_pipeline = pipeline(
     "text2text-generation",
-    model="google/flan-t5-large",
+    model="google/flan-t5-base",
     device=device
 )