NEWS.md

In einops: Flexible, Powerful, and Readable Tensor Operations

einops 0.2.1

Changes

• For CRAN check compatibility and issues with torch::torch_install(), torch-related tests auto skip on CRAN.
• Move lifecycle from imports to Suggests since it is a build time developer dependency instead of a user-required dependency.

einops 0.2.0

Additions

• Add backend for torch's torch_tensor objects for the three core functions rearrange(), reduce() and "repeat"()
• zeallot is now a suggested dependency for users
• lintr is now a suggested dependency for developers

Changes

• Minor updates in documentation and docker infrastructure in the sourcecode

einops 0.1.0

• Initial release. The only python einops function implementations are rearrange, reduce, repeat, and parse_shape, and the only backend supported is for base::array objects. The first einops tutorial has also been mostly re-implemented as a vignette

einops documentation built on Sept. 9, 2025, 5:29 p.m.