# unslothai/hyperlearn

2-2000x faster ML algos, 50% less memory usage, works on all hardware - new and old.

Repository: https://github.com/unslothai/hyperlearn
Canonical: https://ross.abutalabs.com/products/hyperlearn
Homepage: https://unsloth.ai
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: machine-learning, python, statistics, scikit-learn, statsmodels, pytorch, deep-learning, data-science, data-analysis, regression-models, econometrics, tensor, neural-network, gpu, research, optimization
Last push: 2024-11-19T02:09:54+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2928, "days_push": 653, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2505, forks 167 (observed 2026-08-28T04:06:57.121422+00:00)

## What it is
HyperLearn is a Python library providing optimized reimplementations of classical machine learning and statistical algorithms (SVD, eigendecomposition, regression, imputation) that are 2-2000x faster and use ~50% less memory than scikit-learn/NumPy equivalents. It works on both old and new hardware, including CPU and GPU, and its methods have been incorporated into PyTorch, SciPy, CuPy, and NVIDIA RAPIDS.

## Use cases
- speed up slow SVD and eigendecomposition in numpy or pytorch
- run scikit-learn style algorithms with less memory on large datasets
- fit ridge regression and linear models faster on CPU
- accelerate classical ML on old hardware without a modern GPU
- fast data imputation for missing values
- speed up t-SNE style embeddings used in RAPIDS

## When to choose
- you need faster linear algebra primitives (SVD, EIGH, GELS) than NumPy/PyTorch provide
- you are memory-constrained when running classical ML on big data
- you want scikit-learn-like APIs that work on older CPUs and GPUs

## When to avoid
- you need actively maintained, production-supported ML tooling
- you need deep learning model training rather than classical algorithms
- you need the latest scikit-learn features and estimators

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, math, data-science, gpu-computing
- domain: machine-learning, data-science, performance
- platform: python, cross-platform
- tags: scikit-learn, numpy, linear-algebra, svd, regression, econometrics, statistics, memory-optimization, algorithms, gpu

## Member repositories
- unslothai/hyperlearn (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:57.121422+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 2026-08-30T02:26:44.488419+00:00, confidence not recorded.
  - readme: https://github.com/unslothai/hyperlearn (fetched 2026-08-28T04:06:57.121422+00:00, sha bc1f7a702564)
  - homepage: https://unsloth.ai (fetched 2026-08-29T10:09:21.190808+00:00, sha d918d098f3c1)
  - site_page: https://unsloth.ai/docs/new/studio/chat (fetched 2026-08-29T10:09:21.211616+00:00, sha c011a1409bbb)
  - site_page: https://unsloth.ai/docs/new/studio (fetched 2026-08-29T10:09:21.213393+00:00, sha 422b7e13de96)
  - site_page: https://unsloth.ai/docs/basics/api (fetched 2026-08-29T10:09:21.215352+00:00, sha d815387918c1)
  - site_page: https://unsloth.ai/docs/models/tutorials (fetched 2026-08-29T10:09:21.200944+00:00, sha 184c61b12723)
  - site_page: https://unsloth.ai/docs/desktop (fetched 2026-08-29T10:09:21.203412+00:00, sha ebeb6fe03136)
  - site_page: https://unsloth.ai/docs (fetched 2026-08-29T10:09:21.205936+00:00, sha 8b17bfe94934)
  - site_page: https://unsloth.ai/docs/basics/diffusion-image (fetched 2026-08-29T10:09:21.207793+00:00, sha ffee3d1a978a)
  - site_page: https://unsloth.ai/docs/integrations/unsloth-start (fetched 2026-08-29T10:09:21.209646+00:00, sha a570a7214d8b)
- Data as of 2026-08-30T08:39:29.467469+00:00.