move type inference for arange into c++ (#27629)

Summary:
Fixes https://github.com/pytorch/pytorch/issues/17662

I'm not sure if `arange` needs to be in python_arg_parser at all, given the schemas in native_functions.yaml. In any case this at least fixes the dytpe mismatch.

In follow up PRs I will try to handle some of the other ops that do type inference at the python level, like randint.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/27629

Differential Revision: D17885939

Pulled By: eellison

fbshipit-source-id: f97a8bc722b7ab77de1c42a992e49a4a3175ad60
This commit is contained in:
eellison
2019-11-11 11:24:43 -08:00
committed by Facebook Github Bot
parent 9de0b63554
commit e01fc56ecb
9 changed files with 105 additions and 33 deletions
+5 -5
View File
@@ -929,9 +929,9 @@ TEST_F(FunctionalTest, LocalResponseNorm) {
TEST_F(FunctionalTest, Linear) {
{
const auto x = torch::arange(100, 118).resize_({3, 3, 2});
const auto w = torch::arange(200, 206).resize_({3, 2});
const auto b = torch::arange(300, 303);
const auto x = torch::arange(100., 118).resize_({3, 3, 2});
const auto w = torch::arange(200., 206).resize_({3, 2});
const auto b = torch::arange(300., 303);
const auto y = F::linear(x, w, b);
ASSERT_EQ(y.ndimension(), 3);
ASSERT_EQ(y.sizes(), torch::IntArrayRef({3, 3, 3}));
@@ -950,8 +950,8 @@ TEST_F(FunctionalTest, Linear) {
ASSERT_TRUE(torch::allclose(y, y_exp));
}
{
const auto x = torch::arange(100, 118).resize_({3, 3, 2});
const auto w = torch::arange(200, 206).resize_({3, 2});
const auto x = torch::arange(100., 118).resize_({3, 3, 2});
const auto w = torch::arange(200., 206).resize_({3, 2});
const auto y = F::linear(x, w);
ASSERT_EQ(y.ndimension(), 3);
ASSERT_EQ(y.sizes(), torch::IntArrayRef({3, 3, 3}));