Torchsort Versions Save

What's Changed

• Fix incorrect results when running on non-default GPU by @teddykoker in https://github.com/teddykoker/torchsort/pull/49

Full Changelog: https://github.com/teddykoker/torchsort/compare/v0.1.8...v0.1.9

What's Changed

• Update README.md by @teddykoker in https://github.com/teddykoker/torchsort/pull/36
• Better error message if CUDA extension is not available by @teddykoker in https://github.com/teddykoker/torchsort/pull/43

Full Changelog: https://github.com/teddykoker/torchsort/compare/v0.1.7...v0.1.8

Add half-precision (fp16) support.

Ensure cuda code is added to source release on PyPi.

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

Google colab installation fix.

• 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.

• Support python 3.7