FP8 & Precision

The Hidden Cost of FP8 in Production AI Factories

If you’re running FP8 fine-tuning in production and not measuring TruthfulQA, you’re not measuring the thing that breaks. The cost isn’t compute. It’s the quality you can’t see until it’s too late.

In progress
Full write-up coming soon
Real numbers from production runs. Publishing with the full cost model.
Built on this research

Everything in this article is measured, not promised. See it running:

Talk to NOESIS liveSee the 33x benchmarkBack the research