Open Models Almanac
Methodsinference & servinginference quantization

implementation detail · filed under inference & serving

MXFP8 activation quantization

Intermediate activations are represented in microscaled 8-bit floating point.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See inference quantization for the methods filed beside it.

Also called Microscaling FP8 (MXFP8) activations, MXFP8 activations.

sources
2
model
1
lab adopt it
1
strongest
used

Documented in

Moonshot AI

Evidence

2 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
used 2

Kimi K3

Moonshot AIadopted2 sources

MXFP8 activations: Intermediate activations use 8-bit floating point, providing higher precision where it matters most

usedmodel architecturein Kimi K3Moonshot AI
1 more from this model
↑