LlamaIndex
Log in

Find answers from the community

Updated last month

When running query on primary and secondary separately

sshawtyisaten
·
when i run the query on primary and secondary separately i get,

Plain Text
>>> composable_memory.primary_memory.get("my cpa")
[ChatMessage(role=<MessageRole.SYSTEM: 'system'>, content='You are a helpful marketing assistant', additional_kwargs={}), ChatMessage(role=<MessageRole.USER: 'user'>, content='my cpa is $350', additional_kwargs={}), ChatMessage(role=<MessageRole.ASSISTANT: 'assistant'>, content='Are you looking for ways to reduce your CPA, or do you need help with analyzing or understanding this metric further?', additional_kwargs={})]

>>> composable_memory.secondary_memory_sources[0].get("what is my cpa")
[ChatMessage(role=<MessageRole.USER: 'user'>, content='my cpa is $350', additional_kwargs={}), ChatMessage(role=<MessageRole.ASSISTANT: 'assistant'>, content='Are you looking for ways to reduce your CPA, or do you need help with analyzing or understanding this metric further?', additional_kwargs={})]
s
1 comment
sshawtyisaten
actually it seems that this is happening bc secondary messages already exist in primary memory. another question i have is if i ask the agent "what is my cpa?"

it doesn't seem to answer based on what i told ai previously. it gives a response like To determine your CPA (Cost Per Acquisition), you need to divide the total cost of your marketing efforts by the number of acquisitions (or desired actions) achieved.

i wonder if there's a way refer to guide its response besides just changing the system prompt
Add a reply
Sign up and join the conversation on Discord