v0.1.10

What's Changed

• Add cuda 12.1, PyTorch 2.1 releases by @teddykoker in #79
• Add pyproject.toml for PEP 517 builds and project metadata by @LouisJalouzot in #90
• New releases by @teddykoker in #91
• Ensure cuda compilation in CI build by @teddykoker in #94

New Contributors

• @LouisJalouzot made their first contribution in #90

Full Changelog: v.1.9.0...v0.1.10

v0.1.9

What's Changed

• Fix incorrect results when running on non-default GPU by @teddykoker in #49

Full Changelog: v0.1.8...v0.1.9

v0.1.8

What's Changed

• Update README.md by @teddykoker in #36
• Better error message if CUDA extension is not available by @teddykoker in #43

Full Changelog: v0.1.7...v0.1.8

v0.1.7

Add half-precision (fp16) support.

v0.1.6

Ensure cuda code is added to source release on PyPi.

v0.1.5

Remove -arch="compute_50" from extra args (affected building when TORCH_CUDA_ARCH_LIST is used).

v0.1.4

Google colab installation fix.

v0.1.3

• Fixes CUDA memory leak with torchsort.soft_rank with regularization="kl".

• Adds validation for correct user input for torchsort.soft_rank and torchsort.soft_sort.

v0.1.2

• Support python 3.7

v0.1.1

Merge pull request #11 from teddykoker/docker-fix

docker fix