A civic-center kiln desk has a refined classifier, and a colleague opens NVIDIA NeMo to serve it. Where should the finished artifact actually run?
Select an answer to reveal the explanation.
Short Explanation
NeMo is where you customize. The finished classifier runs on Triton Inference Server. cuDF will not serve a booking label, and an NCCL kernel is not a live server.
Full Explanation
Triton Inference Server is the cert-page product that hosts a finished model for live traffic. NeMo is the customize path, not the live inference process. RAPIDS cuDF joins tabular data and does not serve this classifier. An NCCL kernel is Professional plumbing and is not a serving stack.