einops

A new flavour of deep learning operations

einops Permissive license MIT Active ★ 9,573 v0.8.2 released 2026-01-26

Install

einops on PyPI

pip

pip install einops

uv

uv add einops

poetry

poetry add einops

Package facts

License	MIT (permissive)
Python support	supports the current Python release (>=3.9)
Install friction	low — pure-Python wheel
Runtime dependencies	none
Maintenance	actively maintained — 199 days since the last release
Last repo commit	2026-07-05
First released	2018-10-30
Popularity	one of the top 1,000 most-downloaded packages on PyPI (30-day window, as of 2026-08-13)
Known vulnerabilities	none known (OSV.dev, checked 2026-08-13)

Evidence: einops-0.8.2-py3-none-any.whl

Keywords: deep learning, einops, machine learning, neural networks, scientific computations, tensor manipulation

Intended Audience :: Science/ResearchLicense :: OSI Approved :: MIT LicenseProgramming Language :: Python :: 3

About einops

from the package's own PyPI description — quoted content, verbatim

https://user-images.githubusercontent.com/6318811/177030658-66f0eb5d-e136-44d8-99c9-86ae298ead5b.mp4

einops

Run tests (image) PyPI version (image) Documentation (image) Supported python versions (image)

Flexible and powerful tensor operations for readable and reliable code. <br /> Supports numpy, pytorch, tensorflow, jax, and...

Read as markdown · JSON record · Source repository · Homepage

AI interpretation — verify before relying

AI-generated interpretation of the package facts above; every digit, version, license, or vulnerability id it cites is grounded in the facts already shown on this page

Einops provides readable tensor operations (rearrange, reduce, repeat, pack, unpack, einsum) that work across numpy, PyTorch, TensorFlow, JAX, and other frameworks, making multidimensional array manipulation explicit and maintainable.

Installation is straightforward with no runtime dependencies; the package is actively maintained with a recent release (199 days ago) and strong community adoption (9573 GitHub stars).

MIT license permits unrestricted use, modification, and distribution in both open-source and proprietary projects with minimal obligations.

Usage

pip install einops

from einops import rearrange, reduce, repeat

# Rearrange tensor dimensions
output = rearrange(input_tensor, 'b c h w -> b (c h w)')

# Reduce with operation
output = reduce(input_tensor, 'b c (h h2) (w w2) -> b h w c', 'mean', h2=2, w2=2)

# Repeat along new axis
output = repeat(input_tensor, 'h w -> h w c', c=3)

Requires Python 3.9 or later; no compiled dependencies.

Verdict: Einops is a mature, well-maintained library with zero security vulnerabilities and permissive MIT licensing. Its zero runtime dependencies and low install friction make it a low-risk addition to any project. Active development and broad framework support make it a reliable choice for readable tensor operations.

Needs verification

Performance overhead of einops operations compared to native framework equivalents.
Compatibility guarantees with specific versions of PyTorch, TensorFlow, or JAX.
Production-readiness of torch.compile support and its performance implications.

tensor reshape operationseinsum notation for deep learningreadable array manipulationframework-agnostic tensor operationsmultidimensional array packingeinstein summation operationstensor rearrange and reduce