Add `provider` to `create_embeddings` function

[MendeBadra]

So, I've been exploring ways to use OpenAI.jl package's provider to have TogetherAI as mentioned in the README.md.

provider = OpenAI.OpenAIProvider(
    api_key = ENV["TOGETHER_API_KEY"],
    base_url = "https://api.together.xyz/v1"  # Must override OpenAI base
)

I can use create_chat function easily to chat with this model

response = create_chat(
    provider,
    "meta-llama/Llama-3.3-70B-Instruct-Turbo-Free",  # Use Together-compatible model name
    [Dict("role" => "user", "content" => "Write some ancient Mongolian poetry")]
)

println(response.response["choices"][begin]["message"]["content"])

However, there is no provider interface to the create_embeddings function, so I created my own:

import OpenAI: create_embeddings

function create_embeddings(
    provider::OpenAI.AbstractOpenAIProvider,
    input,
    model_id::String = DEFAULT_TOGETHERAI_EMBEDDING_MODEL_ID;
    http_kwargs::NamedTuple = NamedTuple(),
    kwargs...)
    return OpenAI.openai_request("embeddings",
        provider;
        method = "POST",
        http_kwargs = http_kwargs,
        model = model_id,
        input, 
        kwargs...)
end

resp = create_embeddings(
    provider,
    # ENV["TOGETHER_API_KEY"],
    horror_movies_tidier.overview[1];
    #model_id = DEFAULT_TOGETHERAI_EMBEDDING_MODEL_ID
)

Maybe we should add this function into the library itself?

[roryl23]

I like this idea, open up a PR and we can get it into a release!

[MendeBadra]

So, turns out I was using an older version of OpenAI.jl package: "0.10.1". There is already a function in the src/OpenAI.jl directory. So, it's no need to add this function. But I think the documentation could explain this better (for both create_chat and create_embeddings.