# state-spaces/mamba

Mamba SSM architecture

Repository: https://github.com/state-spaces/mamba
Canonical: https://ross.abutalabs.com/products/mamba
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-07-22T19:41:12+00:00

## Health v2 (maintenance only)
Score: 85/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 93, release rhythm 83, longevity 71
- inputs: {"age_days": 1007, "days_push": 42, "days_rel": 116, "gap_med": 7, "n_releases_24m": 12}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 18783, forks 1806 (observed 2026-08-28T04:11:27.124473+00:00)

## What it is
Mamba is a PyTorch library implementing the selective state space model (SSM) architecture for linear-time sequence modeling, with hardware-aware CUDA kernels. It serves as an efficient alternative to Transformers for language modeling and other information-dense sequence tasks.

## Use cases
- train a state space model for language modeling
- replace transformers with linear-time sequence models
- run efficient long-context sequence inference
- implement Mamba SSM layers in a PyTorch model
- experiment with subquadratic sequence modeling architectures

## When to choose
- you need linear-time scaling for very long sequences
- you want a Transformer alternative with competitive language modeling performance
- you have a CUDA GPU and want hardware-aware efficient SSM kernels

## When to avoid
- you need the broad ecosystem and tooling support of standard Transformers
- you have no GPU and cannot compile CUDA extensions
- you need off-the-shelf pretrained models for a quick application build

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-training, llm-inference
- domain: deep-learning, large-language-models, machine-learning
- platform: python, cross-platform
- tags: state-space-models, sequence-modeling, transformer-alternative, cuda-kernels, pytorch, natural-language-processing, gpu, linux

## Member repositories
- state-spaces/mamba (main) score 85

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:27.124473+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:01:17.917217+00:00, confidence not recorded.
  - readme: https://github.com/state-spaces/mamba (fetched 2026-08-28T04:11:27.124473+00:00, sha 3b95ed5011ef)
- Data as of 2026-08-30T08:39:29.467469+00:00.