Inference

NeMo Platform inference provider registration lifecycle through the platform SDK (secret -> create temp provider -> verify -> delete -> create final provider).

NVIDIA NeMo Updated 2.3k repo stars

File contents

Inference provider tasks

  • For provider registration tasks, use this sequence:
    1. Create API key secret
    2. Create temporary provider
    3. List providers
    4. Get provider details
    5. Delete temporary provider
    6. Create final verification provider
  • Use nemo_api with resource secrets for the API key secret, passing workspace="<active request workspace>".
  • Use nemo_api with resource inference.providers for provider CRUD, passing workspace="<active request workspace>" on every call.
  • Ensure final provider uses exactly the requested host URL and description.
  • Retrieve or list the final provider and compare its host URL and description with the requested values before reporting success.

nvidia-nemo/nemo-platform commit 01dda599ff

Frequently asked questions

npx get-skill add nvidia-nemo/inference-2