Hi team We are migrating to latest

AAbhishek

Hi team, We are migrating to latest version of llama-index i.e. 0.8.36.
We have the following code to generate the embeddings, But seems like It doesn't work anymore with the latest version:

CODE:

Plain Text

embed_model = self.service_context.embed_model
for node in keyword_nodes:
    embed_model.queue_text_for_embedding(
        node.node.node_id,
        node.node.get_text(),
    )
_, text_embeddings = embed_model.get_queued_text_embeddings()
for idx in range(len(keyword_nodes)):
    keyword_nodes[idx].node.embedding = text_embeddings[idx]

Error: AttributeError: 'OpenAIEmbedding' object has no attribute 'queue_text_for_embedding'

/ / Any help here?

24 comments

LLogan M

Use text_embeddings = embed_mode.get_text_embedding_batch(texts, show_progress=False)

or if you need async, text_embeddings = await embed_mode.aget_text_embedding_batch(texts, show_progress=False)

LLogan M

We removed the stateful queue

AAbhishek

Okay, Got it! Thanks 🙏

AAbhishek

@Logan M It resulted in error:

Plain Text

embed_model = self.service_context.embed_model
        nodes_text = []
        for node in keyword_nodes:
            nodes_text.append(node.node.get_text())

        text_embeddings = embed_model.aget_text_embedding_batch(
            nodes_text, show_progress=False
        )
        for idx, node in enumerate(keyword_nodes):
            keyword_nodes[idx].node.embedding = text_embeddings[idx]

TypeError: 'coroutine' object is not subscriptable

LLogan M

you need to await

LLogan M

if you want async

LLogan M

I would only use async if you are already running inside an async function

LLogan M

otherwise use the sync version

AAbhishek

I tried the above using async and await It responded with

text_embeddings = await embed_model.aget_text_embedding_batch(
nodes_text, show_progress=False
)

   error_data = resp["error"]
KeyError: 'error'

 File "/home/jerry/miniconda3/envs/dobby/lib/python3.10/site-packages/openai/api_requestor.py", line 405, in handle_error_response
    raise error.APIError(
openai.error.APIError: Invalid response object from API: '{"status":"failure","message":"Portkey Error: API Key Not Found. Error Code:02"}' (HTTP response code was 401)

AAbhishek

Let me try with the sync version

AAbhishek

FYI, The above works with older version and doesn't resulted into key error for portkey.

LLogan M

Seems like an issue with portkey, not sure what changed there

LLogan M

I didn't know portkey supported embeddings tbh lol

AAbhishek

No problem, Will try with sync version for now. Because migration is important for us. Thanks though

AAbhishek

The sync version throws the same portkey api key error

LLogan M

How are you setting up portkey?

AAbhishek

The key is stored as env variable, apart from that
we set the portkey configs in headers for the openai models imported from from llama_index.embeddings.openai import OpenAIEmbedding,

and set the openai.api_base to portkey proxy url

AAbhishek

FYI, did a package update pip install -U portkey-ai but didn't work

LLogan M

So the key is only in an env variable, and you only modify the api_base ? Can you set the key directly in the headers as well?

LLogan M

Or pass they key as kwarg directly into the embeddings

OpenAIEmbedding(.., api_key=os.environ[..])

AAbhishek

Sure, giving it a try

AAbhishek

Thanks but didn't work, I guess they have introduced a new concept of virtual keys. That must be messing it up. I need to connect with them to resolve it,

Not sure on above though, Api key is something different, It must be always required

LLogan M

Yes they have switched the keys concept as well

AAbhishek

Hey Logan, Just sharing an update, It started working there was a misconfiguration with latest OpenAIEmbedding module from llama-index. Which now accepts a parameter called additional_kwargs instead of headers due to which the headers of portkey were not passed in request.

Thanks for your help 🙂

Add a reply

Find answers from the community

Hi team We are migrating to latest