LangChain AzureOpenAIEmbeddings由于模型试图访问微软而无法正常工作,

flvtvl50  于 6个月前  发布在  其他
关注(0)|答案(1)|浏览(47)

检查其他资源

  • 我为这个问题添加了一个非常描述性的标题。
  • 我使用集成搜索在LangChain文档中进行了搜索。
  • 我使用GitHub搜索查找类似的问题,但没有找到。
  • 我确信这是LangChain中的一个bug,而不是我的代码。
  • 通过更新到LangChain的最新稳定版本(或特定集成包)无法解决此错误。

示例代码

from langchain_openai import AzureOpenAIEmbeddings
os.environ["AZURE_OPENAI_API_KEY"] = get_auth_token()
os.environ["OPENAI_API_KEY"] = get_auth_token()
os.environ["AZURE_OPENAI_ENDPOINT"] = 'https://workspace.openai.azure.com/'
os.environ["OPENAI_ENDPOINT"] = 'https://workspace.openai.azure.com/'
os.environ['OPENAI_API_TYPE'] = "azure"
os.environ['OPENAI_API_VERSION']='2023-07-01-preview'

embeddings = AzureOpenAIEmbeddings(
    model='text-embedding-3-small',
    chunk_size=1
)

embeddings.embed_documents(['text'])

错误信息和堆栈跟踪(如果适用)

---------------------------------------------------------------------------
SSLEOFError                               Traceback (most recent call last)
File /anaconda/envs/nlp_min/lib/python3.10/site-packages/urllib3/connectionpool.py:670, in HTTPConnectionPool.urlopen(self, method, url, body, headers, retries, redirect, assert_same_host, timeout, pool_timeout, release_conn, chunked, body_pos, **response_kw)
    [669](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/urllib3/connectionpool.py:669) # Make the request on the httplib connection object.

......

    [706](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/sessions.py:706) elapsed = preferred_clock() - start

File /anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:517, in HTTPAdapter.send(self, request, stream, timeout, verify, cert, proxies)
    [513](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:513)         raise ProxyError(e, request=request)
    [515](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:515)     if isinstance(e.reason, _SSLError):
    [516](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:516)         # This branch is for urllib3 v1.22 and later.
--> [517](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:517)         raise SSLError(e, request=request)
    [519](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:519)     raise ConnectionError(e, request=request)
    [521](https://vscode-remote+amlext-002b2f737562736372697074696f6e732f32316164386262372d633338382d343161352d613931612d6362336539323161356439612f7265736f7572636547726f7570732f676932756f6b79757439356e6c6c392d636f6d6d6f6e2f70726f7669646572732f4d6963726f736f66742e4d616368696e654c6561726e696e6753657276696365732f776f726b7370616365732f676932756f6b79757439356e6c6c392d616d6c2f636f6d70757465732f6465762d41313030.vscode-resource.vscode-cdn.net/anaconda/envs/nlp_min/lib/python3.10/site-packages/requests/adapters.py:521) except ClosedPoolError as e:

SSLError: HTTPSConnectionPool(host='openaipublic.blob.core.windows.net', port=443): Max retries exceeded with url: /encodings/cl100k_base.tiktoken (Caused by SSLError(SSLEOFError(8, '[SSL: UNEXPECTED_EOF_WHILE_READING] EOF occurred in violation of protocol (_ssl.c:1007)')))

描述

我尝试了这段代码片段以及许多变体,都没有成功,问题在于底层函数试图访问openaipublic.blob.core.windows.net,这是不允许的。为什么它要访问外部链接,而它只需要连接到我们的Azure OpenAI端点?

系统信息

langchain==0.2.7
langchain-chroma==0.1.2
langchain-community==0.0.8
langchain-core==0.2.18
langchain-openai==0.1.16
langchain-text-splitters==0.2.2

2sbarzqh

2sbarzqh1#

你好 @hassanzadeh ,
你可以尝试将 azure_deployment 作为嵌入模型的名称传递,如下所示:

from langchain_openai import AzureOpenAIEmbeddings

embeddings = AzureOpenAIEmbeddings(
    azure_deployment="<your-embeddings-deployment-name>",
    openai_api_version="2023-05-15",
)

参考文档:https://python.langchain.com/v0.2/docs/integrations/text_embedding/azureopenai/

相关问题