Prompts used in memory_agent.py do not match the prompts written in the paper

[f-rower]

These are the prompts used in memory_agent.py for memory update

DEFAULT_MEMORY_UPDATE_PROMPT = """Use the entire history of the groupchat (presented before this message) to populate and update the current memory json with factual imformation.

*** OUTPUT SHOULD ONLY BE VALID JSON.  
Be very careful to not include anything that renders the output not directly loadable with json.loads(). *** 

For context, current memory: {memory}. 

Newest response by yourself: {new_response}

Updated memory in JSON format: """

DEFAULT_MEMORY_REPLY_PROMPT = """Using the above instruction, group chat history, and memory json with important information from the chat history, to solve the conversation delegation agent's task. 

Respond according to the ***last instruction from Conversation delegation agent***. 

For cuntext, current memory from chat based on your own outputs: {memory}. 

Newest instruction: {instruction}. 

Output: """