Autogen Keep Using OpenAI API, Not Local

[Kare-Udon]

I have followed the tutorial of https://microsoft.github.io/autogen/docs/reference/oai/completion/, start the FastAI servers.

My code is

from autogen import oai

response = oai.ChatCompletion.create(
    config_list=[
        {
            "model": "chatglm3-6b",
            "base_url": "http://localhost:8000/v1",
            "api_type": "open_ai",
            "api_key": "NULL", # just a placeholder
        }
    ],
    messages=[{"role": "user", "content": "Hi"}]
)

print(response)

And Autogen keeps requesting OpenAI API instead of localhost.

openai.error.Timeout: Request timed out: HTTPSConnectionPool(host='api.openai.com', port=443): Max retries exceeded with url: /v1/chat/completions

My Autogen version is 0.1.14

What is going wrong? Could someone help me please?

[rickyloynd-microsoft]

pyautogen==0.1.14 uses the old version of the openai package, openai==0.28.1. But most parts of the repo are for the newer version, pyautogen==0.2.0b6, and openai==1.x. base_url is for the newer versions. You can replace base_url with the old api_base. Hopefully that helps.

[Kare-Udon]

That is the right solution! Thank you for your warmly assist.