A historical-society desk is handed a running NVIDIA NIM container and a base URL. How should the volunteer call the model?
Select an answer to reveal the explanation.
Short Explanation
A running NIM container is an HTTPS JSON microservice. Send a chat/completions-style body with model, messages, and max tokens. Do not load the weights in the laptop, invent a binary protocol, or bypass the container with a kernel.
Full Explanation
NVIDIA NIM is consumed as a packaged HTTP microservice, typically with an OpenAI-compatible chat/completions body. The associate client sends HTTPS JSON that names the model, the messages, and a max-token limit. Loading weights in the desktop process throws away the running container. A custom binary protocol or a kernel is not how a NIM endpoint is called.