Explore a selective State Space Model that addresses transformers' quadratic bottleneck with linear scaling and significantly faster inference speeds.