Update app.py

app.py

@@ -49,11 +49,11 @@ def generate(
     
 
     messages = [
-        {"role": "user", "content": prompt}
+        {"role": "user", "content": f"[INST] {prompt} [/INST]"}
     ]
 
     inputs = tokenizer.apply_chat_template(messages, return_tensors="pt")
-    stream = model.generate(inputs, **generate_kwargs, stream=True, details=True, return_full_text=False)
+    stream = model.generate(inputs, **generate_kwargs)
     output = ""
 
     for response in stream: