Query about Embedding using Falcon LLM and Typesense Support

TLDR gaurav wants to use Falcon LLM for embedding in their search database and asks if Typesense supports dynamic embedding. Kishore Nallan clarifies that LLMs generate text and can't produce embeddings.

Powered by Struct AI
Join the chat
Jul 11, 2023 (5 months ago)
Photo of md5-c793ac7faa870e19aa043d1f9b35abd1
12:47 PM
Hi, i was planning to try Falcon LLM as an embedding for my search based database. I had already used openai LLM which has static embedding. but Openai isnt working for me and i was planning to use Falcon LLM as embedding for my data. Falcon LLM has dynamic embedding, ie. the length of the embedding changes based on input strings.
Any idea if typesense can support dynamic embedding?
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
01:37 PM
You can't use an LLM for generating embedding. Embedding requires a model which accepts a text and returns a vector. LLMs generate text. Maybe I'm missing something here.