PyTorch implementation of a Selective State Space Model (SSM) using the Mamba architecture. Designed as a diagnostic framework, it features a custom signal-to-noise synthetic dataset generator to evaluate latent state memory trajectories.