I have some use cases where I need to do some large scale data cleanup, but using an LLM is overkill and I already get good results with smaller embeddings.
I want to try using this model and taking advantage of the Matryoshka dimension reduction to manage the progressively more complex use cases.
Not really sure I understand how these work, do you just feed it a large textual document like a transcript or something, and it turns it into a more machine readable vector format or something?
Or is it just a much smaller LLM that’s more optimized for reading than generating?
I’ve been looking forward to trying this one.
I have some use cases where I need to do some large scale data cleanup, but using an LLM is overkill and I already get good results with smaller embeddings.
I want to try using this model and taking advantage of the Matryoshka dimension reduction to manage the progressively more complex use cases.
Not really sure I understand how these work, do you just feed it a large textual document like a transcript or something, and it turns it into a more machine readable vector format or something?
Or is it just a much smaller LLM that’s more optimized for reading than generating?
Basically yes
I’ve only built my own systems that use sentence transformers
You pass in a list of strings, it generates a list of vectors, those vectors can be used for all sorts of similarity analysis and retrieval.