Docs
Vectorizers
Multimodal Vectorizers

Multimodal Vectorizers

Unbody's Multi-Modal Vectorizers are designed to handle and integrate data from various content types, including text, images, videos, and audio.

This section covers how these vectorizers create a unified vector space for cross-modal interactions and the types of data they can process. We have also provided Information on choosing the right multi-modal vectorizer for your project's specific needs.

multi2vec-bind

  • Overview: This is a multimodal vectorizer capable of handling various data types.
  • About the Model: The model facilitates cross-modal search and retrieval, integrating different types of data into a unified vector space.
  • Third-Party Management: Not specified.
  • Strengths: The strength of the model is seamless interaction and search across various data types.
  • Limitations: The model has complexity in managing and configuring multimodal data.
  • Best For: It is best for projects requiring integrated analysis across different content types.
  • Production Status: In roadmap. You can upvote on our GitHub for earlier access.
  • Available Options: Supports text, images, videos, audio, IMU data, depth images, and thermal images.