commit 5b8b2382d162c00b7667a628286994427d659885
author Jiewen Tan <jwtan@fb.com> Thu Oct 28 11:57:28 2021 -0700
committer Facebook GitHub Bot <facebook-github-bot@users.noreply.github.com> Thu Oct 28 11:59:00 2021 -0700
tree ab86aabb3e34b1dfd639adcf69d959711e786766
parent f3aae62942a786b3d1e6568a5a9e3250834c0ba3

Mark mv as CompositeExplicitAutograd (#67373)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/67373

From the implementation of mv, it's decomposed into addmv. So it should
be a CompositeExplicitAutograd op.

Test Plan: It shouldn't change any behaviors. So, CI.

Reviewed By: bdhirsh

Differential Revision: D31973265

Pulled By: alanwaketan

fbshipit-source-id: 3b6850f08e6f671b908a177f148cc6194baa8a13

aten/src/ATen/native/native_functions.yaml

diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index e8a280e..d59d440 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -3127,7 +3127,7 @@
 - func: mv(Tensor self, Tensor vec) -> Tensor
   variants: function, method
   dispatch:
-    CPU, CUDA, SparseCsrCUDA: mv
+    CompositeExplicitAutograd: mv
     SparseCPU, SparseCUDA, SparseCsrCPU: mv_sparse
 
 - func: mv.out(Tensor self, Tensor vec, *, Tensor(a!) out) -> Tensor(a!)