How does FP8 training work on Hopper GPUs, and how do you keep it numerically stable?
FP8 can nearly double training throughput over BF16, but with only a few mantissa bits the numerics leave no slack. Per-tensor scaling and a selective recipe are what get it to converge.
Updated Sep 2026 · Grounded in real GenAI, LLM, and AI/ML engineering interview loops and written to a senior-engineer editorial bar.
FP8 can nearly double training throughput over BF16, but with only a few mantissa bits the numerics leave no slack. Per-tensor scaling and a selective recipe are what get it to converge.
Lead with where the obvious approach breaks, because that is the judgment they are screening for — most candidates jump straight to the happy path and lose the room.
Then walk the failure back through the pipeline in order, naming the one metric the customer's exec sponsor actually cares about before you propose the fix.
