embcli - CLI for Embedding¶

Overview¶

embcli is a command-line interface with a collection of plugins that provide access to various embedding models and vector stores, allowing users to test and compare their performance from the terminal easily. It provides a unified interface for various embedding models, making it easy to generate embeddings, calculate similarity scores, perform document indexing and search, and more!

embcli is designed to be extensible (in a similar way to llm utility), so you can add your own plugins for any embedding model or API you want to use.

Project Status¶

Alpha. embcli is in active development, and new features and plugins are being added regularly.

Quick Start¶

Installation¶

Install embcli-sbert plugin to use sentence-transformers models.

pip install embcli-sbert

Model Plugins and Multimodal Model Plugins shows the full list of available models.

emb command¶

A model plugin contains emb command, which provides a unified interface for embedding models.

emb --help shows the general usage of emb command.

emb --help

Usage: emb [OPTIONS] COMMAND [ARGS]...

Options:
  --help  Show this message and exit.

Commands:
  collections        List collections in the vector store.
  delete-collection  Delete a collection from the vector store.
  embed              Generate embeddings for the provided text or file...
  ingest             Ingest documents into the vector store.
  ingest-sample      Ingest example documents into the vector store.
  models             List available models.
  search             Search for documents in the vector store for the query.
  simscore           Calculate similarity score between two texts.
  vector-stores      List available vector stores.

Try it out¶

Get an embedding for an input text by a sentence-transformers model. all-MiniLM-L6-v2 is used as the default model.

emb embed -m sbert \
"Owls can rotate their necks 270 degrees without injury🦉"

Calculate similarity score between two text embeddings by all-MiniLM-L6-v2. The default metric is cosine similarity.

emb simscore -m sbert \
"The cat drifts toward sleep." "Sleep dances in the cat's eyes."

Index a sample corpus in a Chroma vector store collection. The default database path is ./chroma.db.

emb ingest-sample -m sbert -c catcafe --corpus cat-names-en

Search documents for a query in the indexed Chroma collection.

emb search -m sbert -c catcafe -q "Who's the most agile?"