cyyever

grad+jvp causes INTERNAL ASSERT FAILED

3

As the title says, I tried to combine grad and jvp and then pytorch said ``` catch exception:(new_grad_base.is_floating_point() || new_grad_base.is_complex()) && (self_base.is_floating_point() || self_base.is_complex()) INTERNAL ASSERT FAILED at "../torch/csrc/autograd/autograd_meta.cpp":132, please...

E_Y in equation 2.48 should use the font for expectation.

fix various pointer issues

2

Fix some issues found by static analyser cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10

Clean up more clang-tidy supression

1

1. remove unused NOLINTNEXTLINE(performance-move-const-arg)

More clang-tidy fixes

4

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10

Simplify OpenMP detection in CMake

4

We greatly simplify the handing of OpenMP in CMake by using caffe2::openmp target thoroughly. We follow the old behavior by defaulting to MKL OMP library and detecting OMP flags otherwise....

More tidy fixes

1

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10

Cxx17 header2

4

Fixes #ISSUE_NUMBER

[DeviceIndex][6/N] Use DeviceIndex in more places

2

This PR follows the series of patches beginning with #119142 and fixes various XPU and python related methods to use DeviceIndex.

[structural binding][5/N] Replace std::tie with structural binding

3

This PR follows https://github.com/pytorch/pytorch/pull/119774, it is a continued work to clean up std::tie. cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10