bn+conv融合，bn在前conv在后

先使用bn层，后使用conv层，如下图：

def _fuse_bn_conv_tensor(self, branch) -> Tuple[torch.Tensor, torch.Tensor]:
        """先bn,后conv

        :param branch:
        :return: Tuple of (kernel, bias) after fusing batchnorm.
        """
        kernel = branch.conv.weight
        running_mean = branch.bn.running_mean
        running_var = branch.bn.running_var
        gamma = branch.bn.weight
        beta = branch.bn.bias
        eps = branch.bn.eps
        std = (running_var + eps).sqrt()
        t = gamma / std
        t = torch.stack([t] * (kernel.shape[1]//t.shape[0]),dim=0).reshape(-1, self.in_channels // self.groups, 1, 1)
        t_beta = torch.stack([beta] * (kernel.shape[1]//beta.shape[0]),dim=0).reshape(-1, self.in_channels // self.groups, 1, 1)
        t_running_mean = torch.stack([running_mean] * (kernel.shape[1]//running_mean.shape[0]),dim=0).reshape(-1, self.in_channels // self.groups, 1, 1)
        return kernel * t, torch.sum(
            kernel
            * (
                t_beta - t_running_mean * t
            ),
            dim=(1, 2, 3),
        )+branch.conv.bias

这里还有个问题如果pad不为0的情况，那融合后的bias就不是Wconv*Bbn+Bconv，而是后面要接一个add，值为

    import torch
    import torch.nn.functional as F
    input_h, input_w=ir.get_shape(conv.input[0])[2],ir.get_shape(conv.input[0])[3]
    a=np.tile((t_beta - t_running_mean * t), (1, 1, input_h, input_w)) 
    a=torch.from_numpy(a)
    W1=torch.from_numpy(W)
    bc1 = F.conv2d(a, W1, stride=ir.get_attr(conv, 'strides'), padding=ir.get_attr(conv, 'pads')[0], dilation=1).numpy()#[0,...]
    bc1=np.float32(bc1)

就是后接个(t_beta - t_running_mean * t)过W的conv后的值