multi2vec-bind

Overview

Added in v1.21

The multi2vec-bind module enables Weaviate to use the ImageBind model to vectorize data at import time.

Key notes:

• This module is not available on Weaviate Cloud Services (WCS).
• Enabling this module will enable multiple near<Media> search operators.
• Model encapsulated in Docker container.
• This module is not compatible with Auto-schema. You must define your classes manually as shown below.

multi2vec-bind allows Weaviate to generate vectors data containing any number of the following modalities:

• text
• images
• videos
• audio
• inertial measurement unit (IMU, i.e. accelerometer and gyroscope data)
• single channel depth images, and
• single channel thermal images.

Weaviate instance configuration

Not applicable to WCS

This module is not available on Weaviate Cloud Services.

Memory requirements

The multi2vec-bind module requires a significant amount of memory to run. You may need to increase the memory limit for the multi2vec-bind container to 12 GB or more, such as through Docker Desktop's settings. You can additionally set a limit on your Docker Compose file as shown below, however your Docker Desktop memory limit must be equal to or higher than the limit set in the Docker Compose file.

Docker Compose file

To use multi2vec-bind, you must enable it in your Docker Compose file (e.g. docker-compose.yml).

Use the configuration tool

While you can do so manually, we recommend using the Weaviate configuration tool to generate the Docker Compose file.

Parameters

Weaviate:

• ENABLE_MODULES (Required): The modules to enable. Include multi2vec-bind to enable the module.
• DEFAULT_VECTORIZER_MODULE (Optional): The default vectorizer module. You can set this to multi2vec-bind to make it the default for all classes.
• BIND_INFERENCE_API (Required): The URL of the inference container.

Inference container:

• image (Required): The image name of the inference container.
• mem_limit (Optional): The memory limit for the inference container. Suggest setting to 12G or higher. (Also review the memory limit in Docker Desktop settings.)
• ENABLE_CUDA (Optional): Set to 1 to enable GPU usage. Default is 0 (CPU only).

---
version: '3.4'
services:
  weaviate:
    command:
    - --host
    - 0.0.0.0
    - --port
    - '8080'
    - --scheme
    - http
    image: cr.weaviate.io/semitechnologies/weaviate:1.24.10
    ports:
    - 8080:8080
    - 50051:50051
    restart: on-failure:0
    environment:
      QUERY_DEFAULTS_LIMIT: 25
      AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: 'true'
      PERSISTENCE_DATA_PATH: '/var/lib/weaviate'
      DEFAULT_VECTORIZER_MODULE: 'multi2vec-bind'
      ENABLE_MODULES: 'multi2vec-bind'
      BIND_INFERENCE_API: 'http://multi2vec-bind:8080'
      CLUSTER_HOSTNAME: 'node1'
  multi2vec-bind:
    mem_limit: 12g
    image: cr.weaviate.io/semitechnologies/multi2vec-bind:imagebind
    environment:
      ENABLE_CUDA: '0'
...

Have you enabled CUDA?

This module will benefit greatly from GPU usage. Make sure to enable CUDA if you have a compatible GPU available (ENABLE_CUDA=1).

Class configuration

You can configure how the module will behave in each class through the Weaviate schema.

Vectorization settings

You can set vectorizer behavior using the moduleConfig section under each class and property:

Class-level

• vectorizer - what module to use to vectorize the data.
• vectorizeClassName – whether to vectorize the class name. Default: true.
• <media>Fields - property names to map for different modalities (under moduleConfig.multi2vec-bind).
  • i.e. one or more of [textFields, imageFields, audioFields, videoFields, depthFields, thermalFields, IMUFields]
• weights - optional parameter to weigh the different modalities in producing the final vector.

Property-level

• skip – whether to skip vectorizing the property altogether. Default: false
• vectorizePropertyName – whether to vectorize the property name. Default: false
• dataType - the data type of the property. For use in the appropriate <media>Fields, must be set to text or blob as appropriate.

Example

The following example class definition sets the multi2vec-bind module as the vectorizer for the class BindExample. It also sets:

• name property as a text datatype and as the text field,
• image property as a blob datatype and as the image field,
• audio property as a blob datatype and as the audio field, and
• video property as a blob datatype and as the video field.

{
  "classes": [
    {
      "class": "BindExample",
      "vectorizer": "multi2vec-bind",
      "moduleConfig": {
        "multi2vec-bind": {
          "textFields": ["name"],
          "imageFields": ["image"],
          "audioFields": ["audio"],
          "videoFields": ["video"],
        }
      },
      "properties": [
        {
          "dataType": ["text"],
          "name": "name"
        },
        {
          "dataType": ["blob"],
          "name": "image"
        },
        {
          "dataType": ["blob"],
          "name": "audio"
        },
        {
          "dataType": ["blob"],
          "name": "video"
        }
      ]
    }
  ]
}

Example with weights

The following example adds weights for various properties, with the textFields at 0.4, and the imageFields, audioFields, and videoFields at 0.2 each.

{
  "classes": [
    {
      "class": "BindExample",
      "moduleConfig": {
        "multi2vec-bind": {
          ...
          "weights": {
            "textFields": [0.4],
            "imageFields": [0.2],
            "audioFields": [0.2],
            "videoFields": [0.2],
          }
        }
      }
    }
  ]
}

All blob properties must be in base64-encoded data.

Adding blob data objects

Any blob property type data must be base64 encoded. To obtain the base64-encoded value of an image for example, you can use the helper methods in the Weaviate clients or run the following command:

cat my_image.png | base64

Additional search operators

The multi2vec-bind vectorizer module will enable the following search operators: nearText, nearImage, nearAudio, nearVideo, nearDepth, nearThermal, and nearIMU.

These operators can be used to perform cross-modal search and retrieval.

This means that when using the multi2vec-bind module any query using one modality (e.g. text) will include results in all available modalities, as all objects will be encoded into a single vector space.

Model license(s)

The multi2vec-bind module uses the ImageBind model. ImageBind code and model weights are released under the CC-BY-NC 4.0 license. See the LICENSE for additional details.

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

Questions and feedback

If you have any questions or feedback, let us know in our user forum.

• Technical questions
• Documentation feedback