Open VoVAllen opened 8 months ago
Thanks for your suggestion! We will consider this in the next version.
Looking forward to it! It's super useful in the vector database, so that user can use a subvector to build the index, with much less resources needed but got similar results.
Any Update ? This feature is very important for us.
Is there any plan to support Matryoshka embedding like OpenAI text-embedding-3?
Ref: https://arxiv.org/abs/2205.13147