GRU (Gated Recurrent Unit)

GRUs use gating mechanisms to control how information is stored, updated, and passed through a neural network. They primarily use an update gate and a reset gate to determine which previous information should be retained or discarded. Compared with LSTMs, GRUs have a simpler architecture with fewer parameters while still addressing problems such as vanishing gradients in traditional recurrent neural networks.

Many AI tasks require models to understand relationships across sequences, such as words in a sentence or values in a time series. GRUs can capture these dependencies while often requiring less computation than more complex recurrent architectures.