Skip to main content

QdrantDocumentStore

Use the Qdrant database as your document store.

Basic Information​

  • Used with the following retrievers:
    • QdrantEmbeddingRetriever
    • QdrantSparseEmbeddingRetriever
    • QdrantHybridRetriever
  • Type: haystack_integrations.document_stores.qdrant.document_store.QdrantDocumentStore
Hosting

You must host the Qdrant database locally to make it work with deepset AI Platform. For instructions, see Qdrant documentation.

Overview​

For details, see QdrantDocumentStore in Haystack documentation.

Authorization​

To use an API key for your Qdrant database, create a secret for it and then pass it its name in the api_key parameter of the document store. To learn more about secrets, see Add Secrets to Connect to Third Party Providers.

Usage

For instructions on how to create a secret to connect to the document store, write documents, and then retrieve them, see Connect to external database.

Example Configuration​

In an Index​

To write the preprocessed files into the document store:

  1. Add DocumentWriter to your pipeline.
  2. Add QdrantDocumentStore and configure it on the component card.
  3. Connect the two components.

In a Query Pipeline​

To retrieve files from the document store:

  1. Add a Qdrant retriever to your pipeline.
  2. Add QdrantDocumentStore and configure it on the component card.
  3. Connect the document store and the retriever.

Example​

You connect the components using their document_store connection point:

The Configure button under the document_store parameter on a component card

When you switch to YAML, you can see that the document store is an argument of DocumentWriter. It's the same for retrievers:

components:
  DocumentWriter:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      policy: NONE
      document_store:
        type: haystack_integrations.document_stores.qdrant.document_store.QdrantDocumentStore # document store configuration
        init_parameters:
          port: 6333
          grpc_port: 6334
          prefer_grpc: false
          force_disable_check_same_thread: false
          index: Document
          embedding_dim: 768
          on_disk: false
          use_sparse_embeddings: false
          sparse_idf: false
          similarity: cosine
          return_embedding: false
          progress_bar: true
          recreate_index: false
          wait_result_from_api: true
          write_batch_size: 100
          scroll_size: 10000

Init Parameters​

To check the parameters you can customize for this document store, see QdrantDocumentStore API reference in Haystack documentation.