Bfloat16

Jump to bottom Edit New page

Alex Merced edited this page May 21, 2026 · 1 revision

Bfloat16

Brain Floating Point 16-bit format, a truncated version of the 32-bit float format designed to accelerate machine learning computations and save memory.

Part of the Data & AI Terms glossary.

Add a custom sidebar

Clone this wiki locally