local embedding in python without ollama running

[swaipdev]

Hi, for an offline app, I am trying to embed locally with just python (I don't want my user to have to install Ollama to do this).
Is this possible in mem0 ?
Thanks in advance

[swaipdev]

This would work (I imagine) by creating a new local embedding class in mem0/embeddings, that creates embeddings locally.
Then I would call it like this:

config = {
"embedder": {
"provider": "local",
"config": {
"model": "sentence-transformers/all-MiniLM-L6-v2", # Local model name
"embedding_dims": 384 # Set dimensions according to your model
}
}
}

Does this approach make any sense ?

[Dev-Khant]

Hi @swaipdev To run embedding locally you will have to install Ollama.