GRU: Efficient Memory for Sequences

Gated Recurrent Units (GRU) Implementation and Comparative Performance Analysis

This assignment requires students to implement a Gated Recurrent Unit (GRU) from scratch using a deep learning framework like PyTorch or NumPy. Students will then apply their implementation to a time-series forecasting or text generation task, comparing its performance in terms of training time, memory usage, and accuracy against a standard LSTM and a vanilla RNN. The project involves data preprocessing, model architectural design, hyperparameter tuning, and a detailed report on the efficiency benefits of the GRU architecture specifically regarding its reduced parameter count and memory footprint.