Any plans to support new voyage embeddings or jina v2 reranker?

General
1

What it says on the tin :slight_smile: Love these models but the lack of support means I’m tempted to go around weaviate which defeats a part of the promise of the product for me!

3 models on my wishlist:

2

hi @Baptiste_Cumin1 !!

Welcome to our community!

Is this voyage embedding the voyage-multilingual-2 one? If that’s the case, good news: It is already supported.

Here is how:

import weaviate
import os
from weaviate import classes as wvc

client = weaviate.connect_to_local(
    headers={
        "X-VoyageAI-Api-Key": os.getenv("VOYAGEAI_APIKEY"),
    }
)

client.collections.delete("Voyage")
collection = client.collections.create(
    "Voyage",
    vectorizer_config=wvc.config.Configure.Vectorizer.text2vec_voyageai(
        model="voyage-multilingual-2"
    ),
    properties=[
        wvc.config.Property(name="some_text", data_type=wvc.config.DataType.TEXT)
    ]
)
collection.data.insert(
    {
        "text": "This is an example"
    }
)

as for the Jina, indeed, we only have support for the text2vec:

I have raised this with our team and will keep you posted.

Thanks for using Weaviate!

3

Also, there is a PR pending review for the reranker from JinaAI:

4

Hi!

We have a follow up!

Jina Reranker was added in Release v1.26.1 - Hybrid search performance Fix, Tenants create API Fix, New JinaAI Reranker module · weaviate/weaviate · GitHub

Thanks for using Weaviate!