pytorch

506eca24 - Revert D27752279: [nnc] Do not try to vectorize kernels that use float16

Revert D27752279: [nnc] Do not try to vectorize kernels that use float16 Test Plan: revert-hammer Differential Revision: D27752279 (https://github.com/pytorch/pytorch/commit/8df5e61fd6fc487134ba144bcb2d7a8abe9cfcf1) Original commit changeset: ac115080bf2a fbshipit-source-id: cbc0aa2dcb7691d9fc9d081c6169dea711cd9fac