conv.py 文件源码

def _compute_grad_weight(self, grad_output):
        input, weight, bias = self._get_saved_tensors()
        # TODO: no zero needed in the future
        grad_weight = weight.new().resize_as_(weight).zero_()
        grad_bias = bias.new().resize_as_(bias).zero_()
        if torch.typename(input) == 'torch.cuda.FloatTensor':
            args = self.additional_args[3:] + (1,)
            self._backend.VolumetricConvolution_accGradParameters(
                self._backend.library_state, input, grad_output, grad_weight,
                grad_bias, self.buffer1, self.buffer2,
                *args)
        else:
            self._backend.VolumetricConvolutionMM_accGradParameters(
                self._backend.library_state, input, grad_output, grad_weight,
                grad_bias, self.buffer1, 1)
        return grad_weight, grad_bias