Skip to content

Nemotron customize vs Nemotron retrieval recipes

Nemotron customize is an open-source agent skill for ai models and llm tooling. Nemotron retrieval recipes is an open-source agent skill for knowledge and memory.

  • Nemotron customize is best for plan a data prep and supervised fine-tuning pipeline for a nemotron model.
  • Nemotron retrieval recipes is best for diagnose why relevant passages rank low after a first-stage retrieval step.
What it is
Nemotron customize is an open-source agent skill for ai models and llm tooling.
Nemotron retrieval recipes is an open-source agent skill for knowledge and memory.
Best for
Plan a data prep and supervised fine-tuning pipeline for a Nemotron model
Diagnose why relevant passages rank low after a first-stage retrieval step
Kind
Skill
Skill
Pricing
Open source
Open source
Platforms
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
GitHub stars
4K
4K
Last commit
Nov 2026
Nov 2026
License
Apache-2.0
Apache-2.0
Open source
Good for
Plan a data prep and supervised fine-tuning pipeline for a Nemotron model; Run a LoRA fine-tune of a Hugging Face model on two GPUs; Set up DPO or GRPO alignment after an SFT run
Diagnose why relevant passages rank low after a first-stage retrieval step; Plan a Nemotron reranker fine-tuning run with a dry-run preview first; Choose between tuning an embedding model or a reranker for a retrieval stack

What Nemotron customize does

  • Pipeline planning

    Lays out the order of customization stages and the artifacts passed between them before any files change.

  • Step routing

    Matches a request such as LoRA fine-tuning or endpoint evaluation to the correct existing Nemotron step.

  • Config authoring

    Writes new YAML configs in step config folders based on the shipped defaults, leaving checked-in files untouched.

  • Artifact validation

    Checks that pack size, sequence length, tokenizer and checkpoint paths line up across chained steps.

  • Hardware guidance

    Suggests backend and training approach based on model size and available GPU count.

  • Secret-safe handling

    Refers to auth variables by name only and redacts token or key values when reading env files.

What Nemotron retrieval recipes does

  • Recipe family routing

    Maps a retrieval failure mode to either the embedding or the reranking recipe family.

  • Current-checkout priority

    Treats the active repository's recipe, CLI and config files as authoritative over bundled notes.

  • Dry-run first workflow

    Validates commands with help and dry-run invocations before any long-running stage starts.

  • Model profile selection

    Chooses the embed profile for the Nemotron 3 Embed or Llama Embed model and keeps artifacts separate.

  • Stage-by-stage execution

    Runs multi-stage pipelines from prep through eval, with export and deploy as opt-in steps.

  • Remote run configuration

    Covers env.toml profiles, GPU scoping and polling for remote or batch execution.

Nemotron customize vs Nemotron retrieval recipes: questions

Is Nemotron customize or Nemotron retrieval recipes cheaper?

Both are open source. Compare current plans on each official site.

What is the difference between Nemotron customize and Nemotron retrieval recipes?

Nemotron customize is an open-source agent skill for ai models and llm tooling. Nemotron retrieval recipes is an open-source agent skill for knowledge and memory.

Keep comparing