Update app.py

app.py

@@ -102,7 +102,7 @@ async def generate_sql(request: QuestionRequest):
     """
 
     # Generate SQL query
-    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
     generated_ids = model.generate(
         **inputs,
         num_return_sequences=1,