[Bug]: 疑似未调用 azure openai api

[mingregister]

这个bug是否已存在现有issue了？

• [X] 我确认没有已有issue，且已阅读常见问题。

错误表现

不支持Azure

复现操作

config.json里面填写如下配置：

    "openai_api_type": "azure", 
    "azure_openai_api_key": "xxxxxxxx", 
    "azure_openai_api_base_url": "xxxxxxxxxxxxxx.openai.azure.com",
    "azure_openai_api_version": "2023-07-01-preview", 
    "azure_deployment_name": "gpt-35-turbo-0613", 
    "azure_embedding_deployment_name": "text-embedding-ada-002", 
    "azure_embedding_model_name": "text-embedding-ada-002", 

调用的还是openai。

查看源代码：

# modules/models/base_model.py
class ModelType(Enum):
    @classmethod
    def get_type(cls, model_name: str):
        model_type = None
        model_name_lower = model_name.lower()
         ....
        elif "azure" in model_name_lower or "api" in model_name_lower:
            model_type = ModelType.LangchainChat
        return model_type

# modules/models/models.py
def get_model() -> BaseLLMModel:
    # ...
    # model_name是没有azure或者api选项的。
    model_type = ModelType.get_type(model_name)
    # ...
    try:
        # ...
        elif model_type == ModelType.LangchainChat:
            from .Azure import Azure_OpenAI_Client
            model = Azure_OpenAI_Client(model_name, user_name=user_name)
        # ...

我尝试手动在ONLINE_MODELS中增加 Azure选项， 页面运行后报错：

2023-11-14 11:53:40,957 [INFO] [base_model.py:457] 用户的输入为：hello?
Exception in thread Thread-6:
Traceback (most recent call last):
  File "/usr/lib64/python3.9/threading.py", line 980, in _bootstrap_inner
    self.run()
  File "/usr/lib64/python3.9/threading.py", line 917, in run
    self._target(*self._args, **self._kwargs)
  File "/opt/ChuanhuChatGPT/modules/models/base_model.py", line 875, in thread_func
    self.model(messages=history, callbacks=[
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/base.py", line 551, in __call__
    generation = self.generate(
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/base.py", line 309, in generate
    raise e
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/base.py", line 299, in generate
    self._generate_with_cache(
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/base.py", line 446, in _generate_with_cache
    return self._generate(
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/openai.py", line 333, in _generate
    for chunk in self._stream(
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/openai.py", line 305, in _stream
    for chunk in self.completion_with_retry(
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/openai.py", line 272, in completion_with_retry
    retry_decorator = _create_retry_decorator(self, run_manager=run_manager)
  File "/opt/ChuanhuChatGPT/.venv/lib64/python3.9/site-packages/langchain/chat_models/openai.py", line 68, in _create_retry_decorator
    openai.error.Timeout,
AttributeError: module 'openai' has no attribute 'error'

运行环境

• OS: Rocket Linux 9.2
• Browser: Chrome
• Gradio version: gradio==3.43.2
• Python version: 3.9.6

帮助解决

• [ ] 我愿意协助解决！

补充说明

No response

[GaiZhenbiao]

emm 主要是我没有Azure，所以很难测试这个…… 待我去搞个Azure账号

[daidong78]

我也发现这个问题

[vaxin]

这个bug修了么？

[FreedomXD]

我在本地解决了，index_func.py中 改成

from langchain.embeddings import AzureOpenAIEmbeddings, OpenAIEmbeddings
        if os.environ.get("OPENAI_API_TYPE", "openai") == "openai":
            embeddings = OpenAIEmbeddings(openai_api_base=os.environ.get(
                "OPENAI_API_BASE", None), openai_api_key=os.environ.get("OPENAI_EMBEDDING_API_KEY", api_key))
        else:
            embeddings = AzureOpenAIEmbeddings(deployment=os.environ["AZURE_EMBEDDING_DEPLOYMENT_NAME"], openai_api_key=os.environ["AZURE_OPENAI_API_KEY"],
                                          model=os.environ["AZURE_EMBEDDING_MODEL_NAME"], azure_endpoint=os.environ["AZURE_OPENAI_API_BASE_URL"], openai_api_type="azure")

你试试