FP8 & Precision

Numerical Stability in Low-Precision LLM Training

The papers show you the loss curves that converge. We collected the ones that don’t. Low-precision training breaks in ways that never make it into publications — we ran them on purpose to find out why.

⏲
In progress
Full write-up coming soon
Failure taxonomy is done. Write-up in progress.
Built on this research

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

Talk to NOESIS liveSee the 33x benchmarkBack the research