Epic Goals:

• Investigate work items needed and team dependencies in order to support this feature

Questions to answer:

• New blocks we need to ship or users writing?
• Does this exist in research?
  • This is already implemented in the branch and we cherry pick this
    • Where is it implemented? Where is the repo?
• Reconciliation w/upstream 
  • A lot of this works w/vllm. How will this effect llama.cpp? Hardware reqs?
  • Is this downstream-only?
• get repo, get demo/deck/Jupyter nb
• what’s the e2e workflow
• How does ATT want to use this? what’re they trying to solve?
• How will the hardware reqs translate to our work? Serving changes?

Feature Overview (mandatory - Complete while in New status)

• So far, teacher model has been used to generate QNA pairs. 
• Users also want to be able to prepare their data for different tasks such as classification/labelling/annotation.
• The idea is to use teacher model as an annotator.

Goals (mandatory - Complete while in New status)

• End users can use data labelling/classification as a SDG capability
• Enables end users to do more use cases with SDG capability