Core Velocity Lab

Machine-Learning

FlashAttention-2 in Vulkan with Tensor Cores support

Gradient of the attention op