Pytorch not support half on cpu

[MHGL]

pytorch half op in python on cpu

Traceback (most recent call last):
  File "test.py", line 182, in <module>
    det()
  File "/opt/conda/lib/python3.8/site-packages/deepvac/syszux_deepvac.py", line 456, in __call__
    self.smokeTestForExport3rd(input)
  File "/opt/conda/lib/python3.8/site-packages/deepvac/syszux_annotation.py", line 6, in wrapper
    return f(*args, **kwargs)
  File "/opt/conda/lib/python3.8/site-packages/deepvac/syszux_deepvac.py", line 606, in smokeTestForExport3rd
    self.exportTorchViaScript()
  File "/opt/conda/lib/python3.8/site-packages/deepvac/syszux_deepvac.py", line 504, in exportTorchViaScript
    save_model.saveByInputOrNot()
  File "/opt/conda/lib/python3.8/site-packages/deepvac/syszux_deepvac.py", line 137, in saveByInputOrNot
    freeze_ts = torch.jit.freeze(self.ts)
  File "/opt/conda/lib/python3.8/site-packages/torch/jit/_freeze.py", line 106, in freeze
    optimize_frozen_module(out)
  File "/opt/conda/lib/python3.8/site-packages/torch/jit/_freeze.py", line 143, in optimize_frozen_module
    torch._C._jit_pass_optimize_frozen_graph(mod.graph)
RuntimeError: "rsqrt_cpu" not implemented for 'Half'

pytorch half op in c++ on cpu (torch.jit.script after model.half())

Traceback of TorchScript, original code (most recent call last):
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/container.py", line 119, in forward
    def forward(self, input):
        for module in self:
            input = module(input)
                    ~~~~~~ <--- HERE
        return input
  File "/home/liyang/Test/yolov5-1/modules/model.py", line 19, in forward
        xcpu = x.cpu()
        xnew = torch.cat([xcpu[..., ::2, ::2], xcpu[..., 1::2, ::2], xcpu[..., ::2, 1::2], xcpu[..., 1::2, 1::2]], 1)
        return self.conv(xnew.to(x.device))
               ~~~~~~~~~ <--- HERE
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/container.py", line 119, in forward
    def forward(self, input):
        for module in self:
            input = module(input)
                    ~~~~~~ <--- HERE
        return input
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 396, in forward
    def forward(self, input: Tensor) -> Tensor:
        return self._conv_forward(input, self.weight, self.bias)
               ~~~~~~~~~~~~~~~~~~ <--- HERE
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 392, in _conv_forward
                            weight, bias, self.stride,
                            _pair(0), self.dilation, self.groups)
        return F.conv2d(input, weight, bias, self.stride,
               ~~~~~~~~ <--- HERE
                        self.padding, self.dilation, self.groups)
RuntimeError: "unfolded2d_copy" not implemented for 'Half'

Aborted (core dumped)