nemotron-retrieval-recipes

Plan Nemotron retrieval recipes by applying safety checks and dry-run recommendations within your repository checkout before executing long-running GPU work. The skill enforces prerequisites, references current SKILL.md guidance, and selects the correct recipe family to validate embed and rerank flows.

Debug Nemotron retrieval failures by diagnosing embed and rerank flow errors, comparing base versus fine-tuned model metrics, and reconciling public recipe documentation with your current repository checkout to identify configuration mismatches or outdated guidance.

You can tune hyperparameters for Nemotron embed and rerank flows. The workflow assists with interpreting evaluation metrics, selecting the correct recipe family, and adjusting parameters safely within the current repository checkout before deploying results.

The Nemotron retrieval workflow handles export and deployment handoffs. It assists with remote execution planning while preserving secrets, minimizing risk, and ensuring configurations are safely transitioned from tuning to deployment across embed and rerank flows.

Run dry-runs before Nemotron retrieval tuning to enforce prerequisites and prevent costly errors during long-running GPU work. The workflow recommends dry-runs to validate the current checkout and ensure safe, auditable steps before heavy experimentation.