Add support for Embedding Model finetuning

Open kmattoo1 opened this issue 2 months ago • 4 comments

Embedding model training support for llama 3.2-1b nvembedqa is missing from any recipes and model configuration

Oct 01 '25 16:10 kmattoo1

Still discussing if these should go to the bridge or automodel.

Oct 02 '25 16:10 ericharper

This has been decided to go to automodel. cc @akoumpa

Oct 15 '25 18:10 bernardwin

@meatybobby what is the status ? can we cross post the Automodel PR ?

Oct 15 '25 19:10 euronymous-aithal

@meatybobby what is the status ? can we cross post the Automodel PR ? We just init the process with nemo retriever team. We will need to add new dataloader in AutoModel for the embedding model finetuning.

Oct 16 '25 00:10 meatybobby