tuned prompt

llama.py

@@ -13,7 +13,7 @@ register_dummy()
 def initialize_config(inference: Inference) -> Terminal:
 
     # systemmessage at the very begin of the chat. Will be concatenated with the automatic tool usage descriptions
-    system_prompt = "Hold a casual conversation with the user. Keep responses short at max 5 sentences and on point. Answer using markdown to the user. When providing code examples, avoid comments which provide no additional information."
+    system_prompt = "Hold a casual conversation with the user. Keep responses short at max 5 sentences and on point. Answer using markdown to the user. When providing code examples, avoid comments which provide no additional information. Do not summarize."
     current_date_and_time = datetime.datetime.now().strftime("Current date is %Y-%m-%d and its %H:%M %p right now.")
     append_toolcalls = False
     if append_toolcalls: