Find answers from the community

Updated 2 years ago

Embeddings

At a glance

The community member is exploring the idea of inheriting embeddings from a "Main_embedding" folder and combining it with user-specific "Private_embedding" folders. The goal is for each user (User_1, User_2) to have access to the "Main_embeddings" as well as their own private embeddings.

The comments suggest that the current llama index implementation does not directly support this functionality. The closest approach mentioned is to maintain separate indexes and use a "sub question engine" or "router retriever" to handle the combination of the main and private embeddings. However, there is no explicitly marked answer provided.

Hello Mates, just doing some brain storming and trying some new stuff on learning side.
I am wondering if there is a way we can inherit the embeddings.

Like we have a folder "Main_embedding" and it contains a the index that is saved on this disk
I am using this embedding to query the questions

Now I have a user and he is User_1
He has a folder "Private_embedding" and he has his own chat bot and has own embedded documents.

Same for "User_2"

Now can we inherit the the embedding from "Main_embeddings" along with user specific embedding for their chat bot.

Like User_1 gets "Main_embeddings + /user1/Private_embeddings"
and User_2 "Main_embeddings + /user2/Private_embeddings"
L
A
2 comments
Hmm, right now, I don't think llama index supports that.

The closest is keeping the two indexes separate, and using something like a sub question engine or router retriever
Okay lets see. I think sub question engine migth be something to fit in this scnario\
Add a reply
Sign up and join the conversation on Discord