Model techniques map
Techniquesinference & servingdecoding strategy

implementation detail · filed under inference & serving

Same-checkpoint target and draft weights

Uses target and draft weights from the same checkpoint.

source
1
model
1
lab adopt it
1
strongest
default

How sources treat it

One count per evidence span, weakest treatment to strongest.

default 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

the target and draft weights therefore come from the same checkpoint.

defaultinference servingin DeepSeek-V4-Flash-Vision-ExpDeepSeek

Filed alongside

Other methods under inference & serving :: decoding strategy.