Update app.py

app.py

@@ -46,7 +46,7 @@ def stream_chat(
     message: str, 
     history: list, 
     temperature: float = 0.3, 
-    max_new_tokens: int = 1024, 
+    max_new_tokens: int = 256, 
     top_p: float = 1.0, 
     top_k: int = 20, 
     penalty: float = 1.2,
@@ -54,7 +54,7 @@ def stream_chat(
     print(f'message: {message}')
     print(f'history: {history}')
 
-    conversation = []
+    conversation = [{"role": "system", "content": "You are EXAONE model from LG AI Research, a helpful assistant."}]
     for prompt, answer in history:
         conversation.extend([
             {"role": "user", "content": prompt}, 
@@ -112,7 +112,7 @@ with gr.Blocks(css=CSS, theme="soft") as demo:
                 minimum=128,
                 maximum=4096,
                 step=1,
-                value=1024,
+                value=256,
                 label="Max new tokens",
                 render=False,
             ),