usr_proxy.initiate_chat not working

[SahanaKaluve412]

code that we are using:
user_question="what are the type of customer accounts available?"
text=f'User Question: {user_question}'
result=user_proxy.initiate_chat(manager,
message=user_question,)

error that we are facing :
TypeError: Missing required arguments; Expected either ('messages' and 'model') or ('messages', 'model' and 'stream') arguments to be given

documentation that we referred to:
autogen/notebook/agentchat_oai_assistant_groupchat.ipynb at 0.2 · microsoft/autogen
can anybody help

[ekzhu]

Can you show the complete code?

it looks like your llm configuration may be incorrect or using a non-compatible endpoint.

[thrishathi02]

haha, @SahanaKaluve412 and I are trying to fix the same code. We've tried all possible LLM configuration combinations, but nothing has worked yet.

[ekzhu]

Oh I see. Can't test code at the moment. But one thing we know for sure is that user proxy and initiate chat has not changed in autogen-agentchat.

[thrishathi02]

Thanks for the clarification. But we have two more questions :

1. Could you think of any other wrong configurations in the code that could lead to this error?

2. We're using pyautogen==0.2.28, instead of autogen-agentchat. Because using autogen-agentchat gave us the same error.

[ekzhu]

The type error indicates the create request is missing model and messages. So I think you do need to provide model in the config.

[thrishathi02]

Does it make a difference if we're using pyautogen instead of autogen-agentchat. Which would you recommend?

[ekzhu]

We lost admin access to pyautogen so if you want to use package we are currently uploading fixes to, use autogen-agentchat.

Note that the autogen-agentchat package will be upgraded to v0.4 version this week, with breaking changes. So going forward we recommend upgrading. This may also solve your problem, plus a much better API. To keep using the v0.2 version pin the version to ~=0.2

[SahanaKaluve412]

Got it thanks a lot, we think using model instead of azure_deployment is solving some of our errors because we are getting a different error now so going forward, we will switch to autogen-agentchat.
again, thanks a lot for all the inputs