with_cpu
Batchnorm2d backward implementation
Activation functions implementaions
Refactor tensor logic simd ops with map
Refactor simd algos to use map reduce
revert compiler compatibility to gcc12
Fix compilation
Fix small vector reduce
Fix training issues
Multidevice test utils
AvgPool2d implementation
Fix tensor filling and module registration
MaxPool2d cpu backend stub and test
Device checks and exchange
Refactor tensor factory to get rid of temp objects
Add matmul tests
Make OpenBLASS local dependency