pytorch

0a53c962 - Back out "Add _int_mm to expose cuBLAS int8@int8 -> int32 matmul (#94339)" (#96885)

Back out "Add _int_mm to expose cuBLAS int8@int8 -> int32 matmul (#94339)" (#96885) Summary: Backing out _int_mm to expose cuBLAS int8@int8 -> int32 matmul (#94339) Test Plan: CI Pull Request resolved: https://github.com/pytorch/pytorch/pull/96885 Approved by: https://github.com/drisspg