Model techniques map
Techniquesinference & servinginference kernel

specific method · filed under inference & serving

FlashAttention 4

The fourth version of FlashAttention, identified as an attention backend.

Also called fa4.

source
1
model
1
lab adopt it
1
strongest
optional

How sources treat it

One count per evidence span, weakest treatment to strongest.

optional 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

--mm-attention-backend fa4

optionalsoftware implementationin Step 3.7 FlashStepFun

Filed alongside

Other methods under inference & serving :: inference kernel.