few codes not working with just Azure details

[SushmitaSahuAI]

Hi,
I am trying to execute : autogen/notebook
/agentchat_planning.ipynb . I ahve the OAI_CONFIG_LIST with all the azure details. What am i missing here.
but on running the code i am getting the following error:
187 "No API key provided. You can set your API key in code using 'openai.api_key = ', or you can set the environment variable OPENAI_API_KEY=). If your API key is stored in a file, you can point the openai module at it with 'openai.api_key_path = '. You can generate API keys in the OpenAI web interface. See https://platform.openai.com/account/api-keys for details."
188 )

AuthenticationError: No API key provided. You can set your API key in code using 'openai.api_key = ', or you can set the environment variable OPENAI_API_KEY=). If your API key is stored in a file, you can point the openai module at it with 'openai.api_key_path = '. You can generate API keys in the OpenAI web interface. See https://platform.openai.com/account/api-keys for details.

[nubgamerz]

This usually happens if it can't load your config.
In the assistant agent, set the

"config_list": autogen.config_list_openai_aoai(exclude="aoai"),

to

"config_list": config_list,

In your config_list var, just remove the references to openAI, to only load from your JSON

config_list = config_list_from_json(env_or_file="OAI_CONFIG_LIST")

[SushmitaSahuAI]

Hi,
I have done that only.
Sharing with you the code snippet:
config_list = autogen.config_list_from_json(
env_or_file="OAI_CONFIG_LIST",
filter_dict={
"model": {
"gpt-3.5-turbo",
"gpt-4",
},
},
)

[sonichi]

What's your OAI_CONFIG_LIST like?

[patricianist]

followed the above successfully got rid of the "No API key provided." error, but then I got this:
InvalidRequestError: Invalid URL (POST /v1/https:/xxxxxxxxxxxx.openai.azure.com/deployments/gpt-4/chat/completions)

My config_list looks like this
config_list = autogen.config_list_from_json( env_or_file="OAI_CONFIG_LIST" )

assistant agent init like this
assistant = autogen.AssistantAgent( name="assistant", llm_config={ "seed": 42, # seed for caching and reproducibility "config_list": config_list, # a list of OpenAI API configurations "temperature": 0, # temperature for sampling }, # configuration for autogen's enhanced inference API which is compatible with OpenAI API )

and OAI_CONFIG_LIST like this
[ { "model": "gpt-4", "api_key": "xxxxxxxxxxxxxxxxxxxxxxxx", "base_url": "https://xxxxxxxxxxxx.openai.azure.com/", "api_type": "azure", "api_version": "2023-07-01-preview" } ]

[aavetis]

i'm seeing the same. it shouldn't be hitting /v1/... for an azure oai endpoint, right?

[yigitkonur]

Indeed, this is the root of the issue. The addition of /v1/ as a suffix for Azure should not occur. It's quite strange to observe that it appears Microsoft has not conducted testing of their product on their own platform.

[sonichi]

I wonder whether there is a version mismatch. Could you try pyautogen==0.2.0b3?

[patricianist]

sorry on getting back late, it actually works on pyautogen==0.2.0b3 with Azure OpenAI