Problem
This project is only developed to be able to ingest text data, but there are many AI workflows out there.
Most of the ground work for this is already done, but new embedding models specifically designed for those types of workflows would need to be incorporated, along with file type handling which is missing in the main workflow. (As in, if you pass anything other than text into the main or batch embedding function, they break.)
Solution
In order to embed multi-modal media, the function would need to be able to recognize the incoming media type and implement a different embedding model based on it. A future iteration of this project could include text2img, img2img, text2video, img2video, etc.
Additional information
No response
👨👧👦 Contributing
Problem
This project is only developed to be able to ingest text data, but there are many AI workflows out there.
Most of the ground work for this is already done, but new embedding models specifically designed for those types of workflows would need to be incorporated, along with file type handling which is missing in the main workflow. (As in, if you pass anything other than text into the main or batch embedding function, they break.)
Solution
In order to embed multi-modal media, the function would need to be able to recognize the incoming media type and implement a different embedding model based on it. A future iteration of this project could include text2img, img2img, text2video, img2video, etc.
Additional information
No response
👨👧👦 Contributing