pytorch

144dacd8 - CUDA BFloat16 batched gemm (#45167)

CUDA BFloat16 batched gemm (#45167) Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/45167 Reviewed By: mruberry Differential Revision: D23860458 Pulled By: ngimel fbshipit-source-id: 698de424a046963a30017b58d227fa510f85bf3f