Update app.py

app.py

@@ -118,7 +118,10 @@ def respond(
 
     messages.append({"role": "user", "content": message})
     input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt").to(accelerator.device) #.to('cuda')
-    input_str= str(input_ids)
+    
+    input_ids2 = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, return_tensors="pt").to(accelerator.device) #.to('cuda')
+
+    input_str= str(input_ids2)
     print('input str = ', input_str)
 ## <BOS_TOKEN><|START_OF_TURN_TOKEN|><|USER_TOKEN|>Hello, how are you?<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>
     # with autocast():