Update app.py

app.py

@@ -61,7 +61,7 @@ def generate_response(user_query, relevant_segment):
         input_ids = tokenizer.encode(user_message, return_tensors='pt')
         
         # Create attention mask (1 for real tokens, 0 for padding tokens)
-        attention_mask = (input_ids != tokenizer.pad_token_id).to(torch.long)
+        attention_mask = (input_ids != tokenizer.pad_token_id).long()
 
         # Generate the response using the model
         output = model.generate(