Update app.py

app.py

@@ -27,7 +27,7 @@ tokenizer = AutoTokenizer.from_pretrained("openai-community/gpt2")
 model_response = AutoModelForCausalLM.from_pretrained("openai-community/gpt2").to(device)
 
 # Load the NLU model for intent detection
-nlu_model = AutoModelForSequenceClassification.from_pretrained("distilbert/distilbert-base-uncased-finetuned-sst-2-english")
+nlu_model = AutoModelForSequenceClassification.from_pretrained("distilbert/distilbert-base-uncased-finetuned-sst-2-english").to(device)
 
 # Define the function to find the most relevant document
 @spaces.GPU(duration=120)
@@ -38,7 +38,6 @@ def retrieve_relevant_doc(query):
     return df.iloc[best_match_idx]['Abstract']
 
 # Define the function to detect intent
-@spaces.GPU(duration=120)
 def detect_intent(query):
     inputs = tokenizer(query, return_tensors="pt").to(device)
     outputs = nlu_model(inputs["input_ids"], attention_mask=inputs["attention_mask"])