Local Embeddings

Generate embeddings locally using @huggingface/transformers

Local Embeddings

AI-generated

Summary

Generate normalized vector embeddings locally from text using selected Hugging Face transformer models.

Inputs

model (required) — Select the embedding model to generate vector embeddings; options include multilingual-e5-small, multilingual-e5-base, and all-MiniLM-L6-v2.

Output shape

a list of normalized embedding vectors corresponding to input texts

The model is downloaded at first use and cached locally. Input texts are transformed as 'query: ' or 'passage: ' depending on the model. This node outputs embeddings to be consumed by a connected AI Vector Store node.

Discussion