Quiz 31 Question 14 of 20

Your organization is planning to train a massive, multi-billion parameter large language model (LLM) from scratch in your new data center. You need to select a GPU infrastructure that provides the highest memory capacity per node, optimal training throughput, and the most efficient floating-point operations for transformer workloads. Which architecture represents the state-of-the-art solution for this level of enterprise AI training?

Select an answer to reveal the explanation.

Motivation