Skip to content

Decide on which API should be used for embedding chunks of text #2992

Description

@murphp15

What is the feature request? What problem does it solve?
To build a RAG pipeline we need to be able to create embeddings from text chunks.
The idea of building these pipelines is that they can be used as part of a private AI solution, because of this we can't just use huggingface (public) or openai's apis to generate the embeddings. We need to use a privately hosted model in VMWare to generate these embeddings.
Even though we are targetting a private AI solution embedding apis should be reasonably interchangeable.

It is important not to reinvent the wheel here and look at what langchain are doing.

Definition of done
There should be an example data job that runs and embeds chunks of text

Metadata

Metadata

Assignees

Labels

enhancementNew feature or requestinitiative: VDK for Private AIInitiative including the effort to support Private AI usecases of VMWare with VDK

Type

No type

Fields

No fields configured for issues without a type.

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions