"slow_conv2d_cpu" not implemented for 'Half'

I am trying to run some software I found called CLIP Interrogator in CPU mode, as I'd like to preserver VRAM for other operations, and if i run on CUDA I get out of memory errors due to my main program holding ontop VRAM for it's diffusion.

I specifically get the error:

"slow_conv2d_cpu" not implemented for 'Half' from /usr/local/lib/python3.7/dist-packages/torch/nn/modules/conv.py on the function:

    def _conv_forward(self, input: Tensor, weight: Tensor, bias: Optional[Tensor]):
        if self.padding_mode != 'zeros':
            return F.conv2d(F.pad(input, self._reversed_padding_repeated_twice, mode=self.padding_mode),
                            weight, bias, self.stride,
                            _pair(0), self.dilation, self.groups)
        return F.conv2d(input, weight, bias, self.stride,
                        self.padding, self.dilation, self.groups) #<-- Error on this line

Is there a way to force this to full float, or any other work around?