Productmamba-ssm / Mamba
Mamba-SSM
A deep learning architecture implementation based on State Space Model (SSM) theory that processes long sequences with linear complexity through a selective scan mechanism, serving as an alternative to the quadratic complexity of Transformer architectures.
Timeline (last 90 days)
Sep 27
The Unsloth team released pre-compiled Python wheels targeting CUDA 13, covering three components: Flash-Attention 2, Causal-Conv1D, and Mamba-SSM.
Unverified50%