BadRequestError: Unsupported data type when using AzureOpenAIEmbeddings

[elianna-wang]

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

from langchain_openai import AzureOpenAIEmbeddings

embeddings = AzureOpenAIEmbeddings(
    model="gpt-4-turbo-2024-04-09",
    azure_endpoint=AZURE_OPENAI_ENDPOINT,
    openai_api_key=AZURE_OPENAI_API_KEY,
    azure_deployment=AZURE_OPENAI_DEPLOYMENT
)

input_text = "Who is 2+2?"

embedded_query = embeddings.embed_query(input_text)

Description

I'm trying to use AzureOpenAIEmbeddings but I kept getting BadRequestError: Unsupported data type

Please see the example code.

I ensured the azure_endpoint is in format of https://.openai.azure.com/openai/deployments//chat/

Same environment variables work really well when I'm not embedding but calling model invoke. I'm not sure why I kept getting the same result.

Please see full error message below:

BadRequestError Traceback (most recent call last)
Cell In[6], line 12
3 embeddings = AzureOpenAIEmbeddings(
4 model="gpt-4-turbo-2024-04-09",
5 azure_endpoint=AZURE_OPENAI_ENDPOINT,
6 openai_api_key=AZURE_OPENAI_API_KEY,
7 azure_deployment=AZURE_OPENAI_DEPLOYMENT
8 )
10 input_text = "Who is 2+2?"
---> 12 embedded_query = embeddings.embed_query(input_text)
14 # print(f"Embedding length: {len(embedded_query)}")
15 # print(embedded_query[:10])

File ~/Documents/Project/LLM_Local_Test/.venv3.9.13_new/lib/python3.9/site-packages/langchain_openai/embeddings/base.py:629, in OpenAIEmbeddings.embed_query(self, text)
620 def embed_query(self, text: str) -> List[float]:
621 """Call out to OpenAI's embedding endpoint for embedding query text.
622
623 Args:
(...)
627 Embedding for the text.
628 """
--> 629 return self.embed_documents([text])[0]

File ~/Documents/Project/LLM_Local_Test/.venv3.9.13_new/lib/python3.9/site-packages/langchain_openai/embeddings/base.py:588, in OpenAIEmbeddings.embed_documents(self, texts, chunk_size)
...
(...)
1066 retries_taken=retries_taken,
1067 )

BadRequestError: Unsupported data type

System Info

langchain 0.3.3
langchain-community 0.3.2
langchain-core 0.3.10
langchain-openai 0.2.2
langchain-pinecone 0.1.3
langchain-text-splitters 0.3.0

Python Version: 3.9.13

[Yuya-hub-hash]

I've struggled for hours to figure out this same issue, and it was because of the wrong AZURE_OPENAI_ENDPOINT.
Nothing else worked for me at all, and I hope this helps yours too.

[deepak-hl]

@elianna-wang , try this
embedding = AzureOpenAIEmbeddings(
openai_api_version="2024-04-01-preview",
openai_api_type="azure",
openai_api_key=os.getenv("AZURE_OPENAI_API_KEY"),
azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT"),
deployment=os.getenv("DEPLOYMENT_NAME"),
model="text-embedding-ada-002",
validate_base_url=True,
)

[elianna-wang]

Thank you guys! @deepak-hl helped me get the problem resolved! The Endpoint was also a good callout.

[koushik173]

from langchain_openai import AzureOpenAIEmbeddings

embeddings = AzureOpenAIEmbeddings(
model="text-embedding-3-large",
# dimensions: Optional[int] = None, # Can specify dimensions with new text-embedding-3 models
azure_endpoint= "https://entimeoai.openai.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2024-02-15-preview",
api_key="fvg", # Can provide an API key directly. If missing read env variable AZURE_OPENAI_API_KEY
openai_api_version="2024-02-15-preview" # If not provided, will read env variable AZURE_OPENAI_API_VERSION
)

Use embed_query for a single string

text = "LangChain is the framework for building context-aware reasoning applications"
single_vector = embeddings.embed_query(text)

print(single_vector)

its show
BadRequestError: Unsupported data type