⚡️ What is FastEmbed?

FastEmbed is a lightweight, fast, Python library built for embedding generation. We support popular text models. Please open a Github issue if you want us to add a new model.

  1. Light & Fast

    • Quantized model weights
    • ONNX Runtime for inference

  2. Accuracy/Recall

    • Better than OpenAI Ada-002
    • Default is Flag Embedding, which has shown good results on the MTEB leaderboard
    • List of supported models - including multilingual models

Here is an example for Retrieval Embedding Generation and how to use FastEmbed with Qdrant.

🚀 Installation

To install the FastEmbed library, pip works:

  1. pip install fastembed

📖 Usage

  1. from fastembed import TextEmbedding
  3. documents: list[str] = [
  4.     "passage: Hello, World!",
  5.     "query: Hello, World!",
  6.     "passage: This is an example passage.",
  7.     "fastembed is supported by and maintained by Qdrant."
  8. ]
  9. embedding_model = TextEmbedding()
  10. embeddings: list[np.ndarray] = embedding_model.embed(documents)

Usage with Qdrant

Installation with Qdrant Client in Python:

  1. pip install qdrant-client[fastembed]

Might have to use pip install 'qdrant-client[fastembed]' on zsh.

  1. from qdrant_client import QdrantClient
  3. # Initialize the client
  4. client = QdrantClient(":memory:")  # Using an in-process Qdrant
  6. # Prepare your documents, metadata, and IDs
  7. docs = ["Qdrant has Langchain integrations", "Qdrant also has Llama Index integrations"]
  8. metadata = [
  9.     {"source": "Langchain-docs"},
  10.     {"source": "Llama-index-docs"},
  11. ]
  12. ids = [42, 2]
  14. client.add(
  15.     collection_name="demo_collection",
  16.     documents=docs,
  17.     metadata=metadata,
  18.     ids=ids
  19. )
  21. search_result = client.query(
  22.     collection_name="demo_collection",
  23.     query_text="This is a query document"
  24. )
  25. print(search_result)