This is a patch release containing the following changes to v2.5.3:
- Improved performance for batch normalization for tbb/threadpool (421a2ce, 7b7b763)
- Fixed implicit conversion from double to float in examples (866b9ac)
- Fixed issue in int8 matmul primitive for specific shapes (035c2d4, 9a1bf19)
- Fixed performance regression for matmul primitive with binary post op and broadcast (dcd61ef, 31dec32)
- Fixed performance regression in binary primitive when using NHWC layout (228493c)