oa::FnMatrix::flashAttentionCausal

flashAttentionCausal: IO-aware causal scaled dot-product attention. Q/K/V and output use contiguous [batchHeads, sequence, headDim] storage. Compatibility alias for the explicit Flash provider of the shared SDPA op.

FnC++ public headers

Function Documentation

Matrix oa::FnMatrix::flashAttentionCausal(
	const Matrix & inQ,
	const Matrix & inK,
	const Matrix & inV,
	oa::F32 inScale
)

Parameters

inQ

const Matrix &

—

inK

const Matrix &

—

inV

const Matrix &

—

inScale

oa::F32

—

Returns

Matrix

The declared return value.