Question on top-down layer in NAS-FPN

[ggjy]

Hi, it seems that there is no top-down path for NAS-FPN (https://github.com/tensorflow/tpu/blob/master/models/official/detection/modeling/architecture/nasfpn.py) in the official implement? I am not sure about this.

这里的实现是把NAS-FPN接在top-down之后的feature map上么？官方代码好像是没有top-down那步？我自己在pytorch上的实现发现没有top-down的话NAS-FPN会掉点，不知道是不是我哪理解错了？