Update README

README.md

@@ -182,6 +182,7 @@ gets more complicated. Sorting by Euclidean distance can be up-to 10x faster.
 ## Downstream projects using x86-simd-sort
 
 - NumPy uses this as a [submodule](https://github.com/numpy/numpy/pull/22315) to accelerate `np.sort, np.argsort, np.partition and np.argpartition`.
+- PyTorch uses this as a [submodule](https://github.com/pytorch/pytorch/pull/127936) to accelerate `torch.sort, torch.argsort`.
 - A slightly modifed version this library has been integrated into [openJDK](https://github.com/openjdk/jdk/pull/14227).
 - [GRAPE](https://github.com/alibaba/libgrape-lite.git): C++ library for parallel graph processing.
 - AVX-512 version of the key-value sort has been submitted to [Oceanbase](https://github.com/oceanbase/oceanbase/pull/1325).

meson.build

@@ -1,5 +1,5 @@
 project('x86-simd-sort', 'cpp',
-        version : '5.0.x',
+        version : '6.0.x',
         license : 'BSD 3-clause',
         default_options : ['cpp_std=c++17'])
 fs = import('fs')