Problem with initializing AzureOpenAI()

[toshnav]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

chat_llm = AzureChatOpenAI(
    deployment_name=os.environ["azure_deployment"],
    model_name=os.environ['azure_deployment'],
    api_version=os.environ['openai_api_version'],
    api_key=os.environ['api_key'],
    azure_endpoint=os.environ['azure_endpoint']
)

llm = AzureOpenAI(
    deployment_name=os.environ["azure_deployment"],
    model_name=os.environ['azure_deployment'],
    api_version=os.environ['openai_api_version'],
    api_key=os.environ['api_key'],
    azure_endpoint=os.environ['azure_endpoint']
)

Description

Hi All,

I am encountering an issue while trying to initialize the AzureOpenAI client. The error message states:

"As of openai>=1.0.0, Azure endpoints should be specified via the azure_endpoint parameter, not openai_api_base (or alias base_url). (type=value_error)"

Interestingly, I do not face this issue when initializing the AzureChatOpenAI client with similar parameters. It seems the two clients handle the endpoint configuration differently. Could someone please advise on why this discrepancy exists and how to resolve the issue for AzureOpenAI?

System Info

langchain==0.2.14
langchain-core==0.2.35
langchain-openai==0.1.22
langchain-text-splitters==0.2.2
platform : windows
pyhton version: 3

[dosubot[bot]]

Hey there, @toshnav! I'm here to help you with any bugs, questions, or to assist you in becoming a contributor.

The ValidationError when initializing the AzureOpenAI client is likely due to the validation logic in the validate_environment method of the AzureOpenAI class. Specifically, the error can occur if the openai_api_base (or its alias base_url) is specified incorrectly or conflicts with the deployment_name (or its alias azure_deployment) parameter.

From the provided code, the following conditions can trigger a ValidationError:

1. If openai_api_base is specified and does not contain /openai, it will raise a ValueError:

   if openai_api_base and values["validate_base_url"]:
       if "/openai" not in openai_api_base:
           values["openai_api_base"] = (
               values["openai_api_base"].rstrip("/") + "/openai"
           )
           raise ValueError(
               "As of openai>=1.0.0, Azure endpoints should be specified via "
               "the `azure_endpoint` param not `openai_api_base` "
               "(or alias `base_url`)."
           )

2. If both deployment_name and openai_api_base are specified, it will raise a ValueError:

   if values["deployment_name"]:
       raise ValueError(
           "As of openai>=1.0.0, if `deployment_name` (or alias "
           "`azure_deployment`) is specified then "
           "`openai_api_base` (or alias `base_url`) should not be. "
           "Instead use `deployment_name` (or alias `azure_deployment`) "
           "and `azure_endpoint`."
       )
       values["deployment_name"] = None

These validation checks ensure that the parameters are correctly set for the Azure OpenAI client. If these conditions are not met, a ValidationError will be raised.

To resolve this issue, ensure that:

• You are using azure_endpoint instead of openai_api_base.
• You do not specify both deployment_name and openai_api_base at the same time.

Here is an example of correct initialization:

from langchain_openai import AzureOpenAI

openai = AzureOpenAI(
    azure_deployment="your-deployment-name",
    azure_endpoint="https://<your-endpoint>.openai.azure.com/",
    api_key="your-api-key",
    openai_api_version="2023-05-15"
)

Ensure that the environment variables or parameters are set correctly to avoid validation errors ^[1][2][3].

To continue talking to Dosu, mention @dosu.

[toshnav]

@dosu Thanks for the help here but we cannot use both model_name and deployement_name both parameters while initializing the AzureOpenApi() function

[dosubot[bot]]

You are correct. You cannot use both model_name and deployment_name parameters while initializing the AzureOpenAI() function. The deployment_name parameter sets the base client URL to include /deployments/{azure_deployment}, which means you won't be able to use non-deployment endpoints. This is explicitly mentioned in the class documentation and enforced by validation checks to prevent conflicts between deployment-specific and non-deployment-specific configurations ^[1][2][3][4].

To resolve the issue, you should only use the deployment_name parameter and ensure that the azure_endpoint is correctly specified. Here's an example of correct initialization:

from langchain_openai import AzureOpenAI

llm = AzureOpenAI(
    deployment_name=os.environ["azure_deployment"],
    api_version=os.environ['openai_api_version'],
    api_key=os.environ['api_key'],
    azure_endpoint=os.environ['azure_endpoint']
)

Make sure to remove the model_name parameter to avoid conflicts.

To continue talking to Dosu, mention @dosu.