I am trying to find similar words to a list of words I have embedded with slight modifications. The goal in the end is to try and build a search tool that can generate a list of single words from an embedding. Reading through the literature though suggests that sentence transformers are not capable of performing this task as they are descriptive in nature and not generative.

One method recommended was to set a list of words to and encode all of them to find the similar words. While this works, it only will work on a predetermined set of words and this is limited in scope of what I would like to do.

I know I could call chatgpt or gemini or other LLMS and ask for similar words, but I would like find the new words based on the encoded vectors as I am interested in what the ludicrous transformations might output.

Is it possible to pull single words out of an embedding of a vector using a sentence transformer model?

code

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('paraphrase-MiniLM-L6-v2')

vector_1 = model.encode("mouse")
vector_2 = model.encode("cheese")
t_vector = vector_1 - vector_2
t2_vector = (vector_1 * vector_1) + vector_2

????