Azure openai key dont work

[sixlycos]

After I referred to all the issues and tried all the modifications, I still got the prompt
openai.error.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.

The code is the same as the demo and has not been changed.
my OAI_CONFIG_LIST like :
`
[
{
"model": "mymodelname",
"api_key": "key",
"api_base": "endpoint",
"api_type": "azure",
"api_version": "2023-06-01-preview"
},
{
"model": "mymodelname",
"api_key": "key",
"api_base": "endpoint",
"api_type": "azure",
"api_version": "2023-06-01-preview"
}
]

`

This configuration works on openai.ChatCompletion.create，Indicates that my model is fully usable。

But I still don’t understand why it doesn’t work properly in autogen

[olivMertens]

Mine is fonctionning

[ { "model": "gpt-4", "api_key": "xxxxxxxx", "api_base": "xxxxxx", "api_type": "azure", "api_version": "2023-07-01-preview" }, { "model": "gpt-3.5-turbo", "api_key": "xxxxxxxxxxxxxxxxxxxxxxx", "api_base": "xxxxx", "api_type": "azure", "api_version": "2023-07-01-preview" } ]

[qingyun-wu]

@sixlycos, What are your model names? And are you using the same code as olivMertens? Thank you!

[afourney]

If it's complaining that there's no key, it might not even be reading the file. When you run the example, are any warnings printed to the console? For example you might see something like this:

"The specified config_list file '{config_list_path}' does not exist."

[instinctuallearning]

For Azure the model name refers to the OpenAI Studio deployment name.

[sonichi]

For Azure the model name refers to the OpenAI Studio deployment name.

This is worth being added to FAQ: https://microsoft.github.io/autogen/docs/FAQ#option-1-load-a-list-of-endpoints-from-json

[sixlycos]

For Azure the model name refers to the OpenAI Studio deployment name.

Yes, I mentioned that I read all the issues and I made sure that I used the deployment model name instead of you know directly copying gpt-4 in the example

This configuration works on openai.ChatCompletion.create

[sixlycos]

@sixlycos, What are your model names? And are you using the same code as olivMertens? Thank you!

just use this case
from autogen import AssistantAgent, UserProxyAgent, config_list_from_json config_list = config_list_from_json(env_or_file="OAI_CONFIG_LIST") assistant = AssistantAgent("assistant", llm_config={"config_list": config_list}) user_proxy = UserProxyAgent("user_proxy", code_execution_config={"work_dir": "coding"}) user_proxy.initiate_chat(assistant, message="Plot a chart of NVDA and TESLA stock price change YTD.")

[sixlycos]

If it's complaining that there's no key, it might not even be reading the file. When you run the example, are any warnings printed to the console? For example you might see something like this:

"The specified config_list file '{config_list_path}' does not exist."

I tried it, the configuration file exists and can be read

[sixlycos]

Could this be related to the autogen version? Do I need to roll back？

[sixlycos]

Something strange occurred: I attempted to redeploy a new model in a different location, and it works. However, the old model can no longer be invoked using auto-generation.

I'm not sure what the issue is, but I appreciate your enthusiasm. Your response was really prompt :D

[yhchang520]

Maybe the filter_dict configuration should contain the model in OAI_CONFIG_LIST, like this:

config_list = autogen.config_list_from_json(
     OAI_CONFIG_LIST,
     file_location=KEY_LOC,
     filter_dict={
         "model": {
             "gpt-3.5",
             "gpt-3.5-turbo",
             "gpt-4",
             "gpt-4-32k",
         },
     },
 )