Multimodal (CLIP) Embeddings

Weaviate's integration with the Hugging Face Transformers library allows you to access their CLIP models' capabilities directly from Weaviate.

Configure a Weaviate vector index to use the CLIP integration, and configure the Weaviate instance with a model image, and Weaviate will generate embeddings for various operations using the specified model in the CLIP inference container. This feature is called the vectorizer.

At import time, Weaviate generates multimodal object embeddings and saves them into the index. For vector and hybrid search operations, Weaviate converts queries of one or more modalities into embeddings. Multimodal search operations are also supported.

Your Weaviate instance must be configured with the CLIP multimodal vectorizer integration (multi2vec-clip) module.

• Check the cluster metadata to verify if the module is enabled.
• Follow the how-to configure modules guide to enable the module in Weaviate.

To use this integration, configure the container image of the CLIP model and the inference endpoint of the containerized model.

The following example shows how to configure the CLIP integration in Weaviate:

services:
  weaviate:
    # Other Weaviate configuration
    environment:
      CLIP_INFERENCE_API: http://multi2vec-clip:8080  # Set the inference API endpoint
  multi2vec-clip:  # Set the name of the inference container
    image: cr.weaviate.io/semitechnologies/multi2vec-clip:sentence-transformers-clip-ViT-B-32-multilingual-v1
    environment:
      ENABLE_CUDA: 0  # Set to 1 to enable

modules:

  multi2vec-clip:

    enabled: true
    tag: sentence-transformers-clip-ViT-B-32-multilingual-v1
    repo: semitechnologies/multi2vec-clip
    registry: cr.weaviate.io
    envconfig:
      enable_cuda: true

As this integration runs a local container with the CLIP model, no additional credentials (e.g. API key) are required. Connect to Weaviate as usual, such as in the examples below.

Configure a Weaviate index as follows to use a CLIP embedding model:

from weaviate.classes.config import Configure, DataType, Multi2VecField, Propertyclient.collections.create(    "DemoCollection",    properties=[        Property(name="title", data_type=DataType.TEXT),        Property(name="poster", data_type=DataType.BLOB),    ],    vector_config=[        Configure.Vectors.multi2vec_clip(            name="title_vector",            # Define the fields to be used for the vectorization - using image_fields, text_fields, video_fields            image_fields=[                Multi2VecField(name="poster", weight=0.9)            ],            text_fields=[                Multi2VecField(name="title", weight=0.1)            ]        )    ],    # Additional parameters not shown)

await client.collections.create({  name: 'DemoCollection',  properties: [    {      name: 'title',      dataType: 'text' as const,    },    {      name: 'poster',      dataType: 'blob' as const,    },  ],  vectorizers: [    weaviate.configure.vectors.multi2VecClip({      name: 'title_vector',      imageFields: [        {          name: 'poster',          weight: 0.9,        },      ],      textFields: [        {          name: 'title',          weight: 0.1,        },      ],    }),  ],  // Additional parameters not shown});

Optionally, if your stack includes multiple inference containers, specify the inference container(s) to use with a collection.

If no parameters are specified, the default inference URL from the Weaviate configuration is used.

Specify inferenceUrl for a single inference container.

from weaviate.classes.config import Configure, DataType, Multi2VecField, Propertyclient.collections.create(    "DemoCollection",    properties=[        Property(name="title", data_type=DataType.TEXT),        Property(name="poster", data_type=DataType.BLOB),    ],    vector_config=[        Configure.Vectors.multi2vec_clip(            name="title_vector",            # Define the fields to be used for the vectorization - using image_fields, text_fields, video_fields            image_fields=[                Multi2VecField(name="poster", weight=0.9)            ],            text_fields=[                Multi2VecField(name="title", weight=0.1)            ],            # inference_url="<custom_clip_url>"        )    ],    # Additional parameters not shown)

await client.collections.create({  name: 'DemoCollection',  properties: [    {      name: 'title',      dataType: 'text' as const,    },    {      name: 'poster',      dataType: 'blob' as const,    },  ],  vectorizers: [    weaviate.configure.vectors.multi2VecClip({      name: 'title_vector',      imageFields: [        {          name: 'poster',          weight: 0.9,        },      ],      textFields: [        {          name: 'title',          weight: 0.1,        },      ],      // inferenceUrl: '<custom_clip_url>'    }),  ],});

After configuring the vectorizer, import data into Weaviate. Weaviate generates embeddings for the objects using the specified model.

collection = client.collections.use("DemoCollection")with collection.batch.fixed_size(batch_size=200) as batch:    for src_obj in source_objects:        poster_b64 = url_to_base64(src_obj["poster_path"])        weaviate_obj = {            "title": src_obj["title"],            "poster": poster_b64  # Add the image in base64 encoding        }        # The model provider integration will automatically vectorize the object        batch.add_object(            properties=weaviate_obj,            # vector=vector  # Optionally provide a pre-obtained vector        )

const collectionName = 'DemoCollection'
const myCollection = client.collections.use(collectionName)

Once the vectorizer is configured, Weaviate will perform vector and hybrid search operations using the specified CLIP model.

When you perform a vector search, Weaviate converts the text query into an embedding using the specified model and returns the most similar objects from the database.

The query below returns the n most similar objects from the database, set by limit.

collection = client.collections.use("DemoCollection")response = collection.query.near_text(    query="A holiday film",  # The model provider integration will automatically vectorize the query    limit=2)for obj in response.objects:    print(obj.properties["title"])

const collectionName = 'DemoCollection'
const myCollection = client.collections.use(collectionName)

When you perform a hybrid search, Weaviate converts the text query into an embedding using the specified model and returns the best scoring objects from the database.

The query below returns the n best scoring objects from the database, set by limit.

collection = client.collections.use("DemoCollection")response = collection.query.hybrid(    query="A holiday film",  # The model provider integration will automatically vectorize the query    limit=2)for obj in response.objects:    print(obj.properties["title"])

const collectionName = 'DemoCollection'
const myCollection = client.collections.use(collectionName)

When you perform a media search such as a near image search, Weaviate converts the query into an embedding using the specified model and returns the most similar objects from the database.

To perform a near media search such as near image search, convert the media query into a base64 string and pass it to the search query.

The query below returns the n most similar objects to the input image from the database, set by limit.

def url_to_base64(url):
    import requests
    import base64

    image_response = requests.get(url)
    content = image_response.content
    return base64.b64encode(content).decode("utf-8")

const base64String = 'SOME_BASE_64_REPRESENTATION';

result = await myCollection.query.nearImage(
  base64String,  // The model provider integration will automatically vectorize the query
  {
    limit: 2,
  }
)

console.log(JSON.stringify(result.objects, null, 2));

Lists of pre-built Docker images for this integration are below.

We add new model support over time. For a complete list of available models, see the Docker Hub tags for the multi2vec-clip container.

As an alternative, you can run the inference container independently from Weaviate. To do so, follow these steps:

• Enable multi2vec-clip and omit multi2vec-clip container parameters in your Weaviate configuration
• Run the inference container separately, e.g. using Docker, and
• Use CLIP_INFERENCE_API or inferenceUrl to set the URL of the inference container.

For example, run the container with Docker:

docker run -itp "8000:8080" semitechnologies/multi2vec-clip:sentence-transformers-clip-ViT-B-32-multilingual-v1

Then, set CLIP_INFERENCE_API="http://localhost:8000". If Weaviate is part of the same Docker network, as a part of the same docker-compose.yml file, you can use the Docker networking/DNS, such as CLIP_INFERENCE_API=http://multi2vec-clip:8080.

• Transformers text embedding models + Weaviate.
• Transformers reranker models + Weaviate.

Once the integrations are configured at the collection, the data management and search operations in Weaviate work identically to any other collection. See the following model-agnostic examples:

• The How-to: Manage collections and How-to: Manage objects guides show how to perform data operations (i.e. create, read, update, delete collections and objects within them).
• The How-to: Query & Search guides show how to perform search operations (i.e. vector, keyword, hybrid) as well as retrieval augmented generation.

Each of the compatible models has its own license. For detailed information, review the license for the model you are using in the Hugging Face Model Hub.

It is your responsibility to evaluate whether the terms of its license(s), if any, are appropriate for your intended use.

• Hugging Face Model Hub

Have a question or feedback? Here's how to reach us.