Efficient implementations of state-of-the-art linear attention models in Pytorch and Triton - View it on GitHub