🐛 Bug

Hi, it looks like torch::nn::init::orthogonal_ calls a deprecated function internally, which prints the following warning:

[W BatchLinearAlgebra.cpp:1940] Warning: torch.qr is deprecated in favor of torch.linalg.qr and will be removed in a future PyTorch release.
The boolean parameter 'some' has been replaced with a string parameter 'mode'.
Q, R = torch.qr(A, some)
should be replaced with
Q, R = torch.linalg.qr(A, 'reduced' if some else 'complete') (function operator())

To Reproduce

Compile and run the following script:

#include <torch/torch.h>

int main() {
  auto x = torch::empty({3, 5});
  torch::nn::init::orthogonal_(x);
}

Expected behavior

No warning

Environment

• PyTorch Version (e.g., 1.0): LibTorch 1.9.0
• OS (e.g., Linux): macOS 11.4
• How you installed PyTorch (conda, pip, source): Downloaded from pytorch.org
• Build command you used (if compiling from source): n/a
• Python version: n/a
• CUDA/cuDNN version: None
• GPU models and configuration: n/a

cc @jianyuh @nikitaved @pearu @mruberry @heitorschueroff @walterddr @IvanYashchuk @xwang233 @lezcano @albanD @jbschlosser