Using extra parameters for Azure API connections

[vswraith]

Is it possible to add extra openai parameters to connect to Azure endpoints?
Eg- our endpoint is deployed at
https://.azure-api.net/deployments/gpt-4/chat/completions?api-version=2023-05-01.

But by default, the code adds /openai to it.
Hence Autogen hits at
https://.azure-api.net/openai/deployments/gpt-4/chat/completions?api-version=2023-05-01 and this does not work.
In python, using the openAI lib, there is a way to override this by setting-

openai.ChatCompletion.azure_api_prefix=''
openai.Completion.azure_api_prefix=''

Can we do something similar for Autogen?

[victordibia]

Hi @vswraith ,

Thanks for mentioning this issue.
Can you give a brief example of how you are specifying your OAI_CONFIG?

[sonichi]

The same lines should work. Alternatively you can also pass them in the .create() method of autogen.Completion etc.

[gregnwosu]

how do you do this? please show me an example i need to set cache_prompt to false from initiate chat for ollama