Skip to content

Commit d8de2ba

Browse files
pytorchbotswolchok
andauthored
Fix ATen mode op_upsample_nearest2d_test
Was broken, now it's not. Differential Revision: [D68930165](https://our.internmc.facebook.com/intern/diff/D68930165/) ghstack-source-id: 263965167 Pull Request resolved: #8084 Co-authored-by: Scott Wolchok <[email protected]>
1 parent 8c82000 commit d8de2ba

File tree

1 file changed

+6
-0
lines changed

1 file changed

+6
-0
lines changed

kernels/test/op_upsample_nearest2d_test.cpp

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -45,6 +45,12 @@ class OpUpsampleNearest2dTest : public OperatorTest {
4545
void test_upsample_nearest2d_dtype() {
4646
TensorFactory<DTYPE> tf;
4747

48+
if (torch::executor::testing::SupportedFeatures::get()->is_aten &&
49+
(DTYPE == ScalarType::Char || DTYPE == ScalarType::Short ||
50+
DTYPE == ScalarType::Int || DTYPE == ScalarType::Long)) {
51+
// not supported.
52+
return;
53+
}
4854
const auto input = tf.make({1, 1, 2, 2}, {1, 2, 3, 4});
4955
std::array<int64_t, 2> output_size = {4, 4};
5056
auto out = tf.zeros({1, 1, 4, 4});

0 commit comments

Comments
 (0)