pytorch转onnx, onnx 12 中没有hardswish opt

在onnx opset 12下转以下模型时因不支持hardswish激活函数而报错

• GhostNet
• MobileNetv3Small
• EfficientNetLite0
• PP-LCNet
  解决方案是找到对应的nn.Hardswish层，将其替换为自己覆写的Hardswish实现:

class Hardswish(nn.Module):  # export-friendly version of nn.Hardswish()
    @staticmethod
    def forward(x):
        # return x * F.hardsigmoid(x)  # for torchscript and CoreML
        return x * F.hardtanh(x + 3, 0., 6.) / 6.  # for torchscript, CoreML and ONNX

以PP-LCNet为例，找到哪些层是Hardswish层，替换方法为

# 替换函数, 参考https://zhuanlan.zhihu.com/p/356273702
def _set_module(model, submodule_key, module):
    tokens = submodule_key.split('.')
    sub_tokens = tokens[:-1]
    cur_mod = model
    for s in sub_tokens:
        cur_mod = getattr(cur_mod, s)
    setattr(cur_mod, tokens[-1], module)
for k, m in model.named_modules():
	if 'dw_sp.2' in k or 'dw_sp.6' in k:
		_set_module(model, k, Hardswish())

当然也可以根据m来判断是否为nn.Hardswish的实例，

for k, m in model.named_modules():
	if isinstance(m, nn.Hardswish):
		_set_module(model, k, Hardswish())

YOLOv5-Multibackbone-Compression
Pytorch替换model对象任意层的方法