Model techniques map
Techniquesoptimizationtraining precision

implementation detail · filed under optimization

Two-dimensional block quantization

NVFP4 weights use two-dimensional block quantization.

source
1
model
1
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

NVFP4 layers use the E2M1 datatype with two-dimensional block quantization on weights

usedoptimizationin Nemotron 3 UltraNVIDIA

Filed alongside

Other methods under optimization :: training precision.