Embeddings "/v1/embeddings"endpoint support.

[sjnho]

Describe the feature or improvement you're requesting

Embeddings use api url as "/engines/text-embedding-ada-002/embeddings", can we support the "/v1/embeddings" with model in Request body?
I 'm new to openai api and only found new entrypoint described in https://platform.openai.com/docs/api-reference/embeddings/create

Additional context

No response

[lewiswilliam]

Hi @sjnho

Yes you can create embeddings by passing the model in request body in the following way:
curl https://api.openai.com/v1/embeddings \ -H "Authorization: Bearer $OPENAI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "input": "The food was delicious and the waiter...", "model": "text-embedding-ada-002" }'

Copying the image here for your reference as well:

You can get more information on models for embedding here:
https://platform.openai.com/docs/guides/embeddings/what-are-embeddings

text-embedding-ada-002 is the suggested one though. Let know if you have any further queries.

[sjnho]

@lewiswilliam Thanks for reply. Sorry, I may not have explained it very clearly. I mean when I use

import openai
# choose text to embed
text_string = "sample text"

# choose an embedding
model_id = "text-similarity-davinci-001"
embedding = openai.Embedding.create(input=text_string, model=model_id)

This library will request to https://api.openai.com/engine/text-similarity-davinci-001/embeddings
Could it request to https://api.openai.com/v1/embeddings ?

[lewiswilliam]

@sjnho thats a first generation model which is no more recommended, please check here https://platform.openai.com/docs/guides/embeddings/what-are-embeddings

Why dont you try text-embedding-ada-002, its cheaper and probably more better, also recommended 😀

[sjnho]

@sjnho thats a first generation model which is no more recommended, please check here https://platform.openai.com/docs/guides/embeddings/what-are-embeddings

Why dont you try text-embedding-ada-002, its cheaper and probably more better, also recommended 😀

@lewiswilliam It's just sample code from readme. But when I changed to text-embedding-ada-002 , this libray still request to /engines/text-embedding-ada-002/embeddings , not request to /v1/embeddings, this is why I open this issue.

[lewiswilliam]

@sjnho my guess is this: since we are using the python sdk, they have managed the routing in the sdk itself. If you go via /v1/embeddings then 1st step in that process will be to route it to the correct model which in this case will be /engines/text-embedding-ada-002/embeddings

Moreover the python sdk also goes to v1 as per following:

[japanvik]

Hi @lewiswilliam

Could you point me to a reference? I don't see it anywhere in the openai documentation, which indicates that the end point should be formatted as /engines/text-embedding-ada-002/embeddings. The official doc indicates that the endpoint is https://api.openai.com/v1/embeddings and the model is passed in as a request parameter upon POST.

Also, what is the python sdk you are referring to here?

Thanks
V

[lewiswilliam]

Hi @japanvik

You are correct, the end point to be used is https://api.openai.com/v1/embeddings

As per the above discussion, the openai python sdk internally goes to /engines/text-embedding-ada-002/embeddings

The python sdk is the official openai python sdk which can be found here - https://github.com/openai/openai-python

[japanvik]

Hi @lewiswilliam

According to the official Open AI documentation, the engines endpoint is deprecated.
I believe it still works for backward compatibility reasons, but it should be calling /v1/embeddings with a json payload indicating the model to be used.

@hallacy any thoughts?

[MalteHB]

Hi all
I would also really appreciate an update on this.
@hallacy are you going to update the SDK to support the newest API documentation?

[RobertCraigie]

The SDK was updated to support embeddings & the v1 SDK also includes support for embeddings!