ScaledDotProductAttention

Swift

public init(
  scale: Float, isCausal: Bool = false, hasAttentionMask: Bool = false, upcast: Bool = false,
  multiHeadOutputProjectionFused: Bool = false, noBias: Bool = false, trainable: Bool? = nil,
  name: String = ""
)

Swift

public func callAsFunction<T: DynamicGraph.TensorGroup>(
  queries q: T, keys k: T, values v: T, attentionMask: T? = nil,
  streamContext: StreamContext? = nil
) -> T