Skip to content

Add embeddings recipe linking NVIDIA Nemotron embed finetune - #121

Open
mudithj wants to merge 1 commit into
mainfrom
add-embeddings-recipe
Open

Add embeddings recipe linking NVIDIA Nemotron embed finetune#121
mudithj wants to merge 1 commit into
mainfrom
add-embeddings-recipe

Conversation

@mudithj

@mudithj mudithj commented Jul 14, 2026

Copy link
Copy Markdown

What

Adds a new recipes/embeddings/ recipe that points to NVIDIA's Nemotron embedding fine-tuning recipe and frames how to run it on Baseten.

Why

Embedding fine-tuning is a common ask (domain-specific retrieval). NVIDIA maintains the full six-stage pipeline (SDG → data prep → fine-tune → eval → export → deploy), so rather than vendoring code NVIDIA owns, this is a pointer recipe. Because the pipeline is multi-stage and iterative, the recipe walks through running it interactively on a Baseten SSH-enabled GPU workstation (truss ssh setuptruss train workstation → SSH in → follow NVIDIA's recipe → truss train stop), which is how most people will actually run it.

Changes

  • recipes/embeddings/README.md — new pointer recipe with the interactive workstation workflow.
  • README.md — table of contents entry + recipe section.

🤖 Generated with Claude Code

…netune

Adds recipes/embeddings/README.md as a pointer recipe to NVIDIA's Nemotron
embedding fine-tuning pipeline, framed around running it interactively on a
Baseten SSH-enabled GPU workstation, and wires it into the root README table
of contents and recipe list.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

@rcano-baseten rcano-baseten left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

note that we could do something like embeddings/nemotron but I think this is good for now

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants