代码中的几个疑问

[duchenghannuli]

感谢您的贡献和开源！但我在阅读代码中有几个疑问
Thank you for your contribution and open access! But I have a few questions when reading the code

请问为什么在#Encoder and Bottleneck的forward_features中进行了多次的维度调整？并且在循环每个层后还对x_downsample进行了一次添加元素操作
Why are there multiple dimension adjustments in "forward_features" of #Encoder and Bottleneck? And an additional element addition operation is performed on x_downsample after looping each layer?

在进行VGG的嵌套做跳跃连接时，x2_1为什么使用x_downsample[4]的my_up值而不是[3]？
When doing VGG skip connections, why does x2_1 use the my_up value of x_downsample[4] instead of [3]?

如能解答 不胜感激
I would be grateful if you can answer