gated linear recurrent unit

Griffin is a variant of the gated linear recurrent unit (GLRU), a type of recurrent neural network architecture designed to improve sequence modeling by combining gating mechanisms and linear recurrent dynamics. It aims to enhance efficiency and gradient flow in tasks involving sequential data.