Minimal and efficient JAX implementation of the Mamba State Space Model in JAX/Flax. Inspired by 'Mamba: Linear-Time Sequence Modeling with Selective State Spaces,' this repo provides fast, scalable, and well-documented state-of-the-art sequence modeling tools.

Attention-Driven Transformers replace most attention layers with simple linear–activation blocks, keeping a single top attention layer to drive global representation learning. This design improves forecasting accuracy and speed over standard transformer models.

Linear-time sequence modeling that replaces attention's O(n²d) complexity with O(nd) summation-based aggregation. Demonstrates constraint-driven emergence: how functional representations can develop from optimization pressure and architectural constraints alone, without explicit pairwise interactions.

Counting Sort O(n) implementation based on CLRS

A linear-time string preprocessing algorithm. Can be used for linear-time pattern matching O(n+m)