Simsimd

[2.0.4](https://github.com/ashvardanian/SimSIMD/compare/v2.0.3...v2.0.4) (2023-10-05)


Make

* Avoid uncommon Python version ([c290513](https://github.com/ashvardanian/SimSIMD/commit/c290513e0807a1c028cb30e28958673e21d1e386))

[2.0.3](https://github.com/ashvardanian/SimSIMD/compare/v2.0.2...v2.0.3) (2023-10-05)


Make

* Temporarily disable SVE checks ([8ad641f](https://github.com/ashvardanian/SimSIMD/commit/8ad641f778b85dc5272685282afecbef654ad4cc))

[2.0.2](https://github.com/ashvardanian/SimSIMD/compare/v2.0.1...v2.0.2) (2023-10-05)


Make

* Use `arch=armv8.2-a+dotprod` flag ([1f2dc65](https://github.com/ashvardanian/SimSIMD/commit/1f2dc65ff2fb4dffff6b555bc6e5ab7cbb54fe11))

[2.0.1](https://github.com/ashvardanian/SimSIMD/compare/v2.0.0...v2.0.1) (2023-10-05)


Fix

* Revert to older `neon_i8` implementation ([3bb25d5](https://github.com/ashvardanian/SimSIMD/commit/3bb25d5ffd45ecdb0b2dfcd2515f6eb43a665fe6))

The second release of SimSIMD introduces several enhancements, including design optimizations in both SIMD-accelerated implementations and high-level wrappers.

- The Python interface has been refined to closely resemble the `spatial.distance` module from SciPy.
- The newly included Python Linux wheels leverage OpenMP for parallelization, boosting performance.
- Enhanced flexibility: all distance functions can now handle arrays of arbitrary lengths.
- Arm capabilities introspection has been transitioned to Assembly for improved efficiency.

The result: you can now replace `scipy.spatial.distance.cosine`, `scipy.spatial.distance.sqeuclidean`, `numpy.inner`, as well as `scipy.spatial.distance.cdist` with the __3-200x faster SIMD-accelerated SimSIMD__ 🥳.

[1.5.2](https://github.com/ashvardanian/SimSIMD/compare/v1.5.1...v1.5.2) (2023-10-05)


Fix

* Higher error tolerance ([5c3b5b4](https://github.com/ashvardanian/SimSIMD/commit/5c3b5b49b305a21e5053b8b35771bfb999b8f77b))