Proposal: Add support for a local embedding model

[tieckit]

Hi,

I’d like to contribute by adding support for serving a local embedding model.
The idea is to serve the embedding model via a vLLM container.

Specifically, I’m thinking of adding support for small Korean embedding models that can run on low-spec GPUs, such as:

• dragonkue/multilingual-e5-small-ko-v2

• dragonkue/BGE-m3-ko

Would this be a welcome contribution? Please let me know if there are any guidelines or preferences I should follow.

Thanks!

[tieckit]

I noticed that I can’t set any labels on this issue. Do I need special permissions to apply labels, or should I just leave it for a maintainer to label appropriately?

[teddylee777]

@mgpark-dl I will check on permissions!

Thx for the great idea.
I will be working on this feature soon.