Module linear 

Linear attention using random feature approximation (Performer-style)

Complexity: O(n * k * d) where k = number of random features

Structs

LinearAttention

Linear attention with random feature maps

Enums

KernelType

Kernel type for linear attention