pytorch

68600fc7 - avoid extra copies in batchnorm inference by introducing a new op, _native_batch_norm_legit_no_training (#94946)

avoid extra copies in batchnorm inference by introducing a new op, _native_batch_norm_legit_no_training (#94946) Pull Request resolved: https://github.com/pytorch/pytorch/pull/94946 Approved by: https://github.com/ezyang